Nitrogen, oxides of (EHC 188, 1997, 2nd edition)

UNITED NATIONS ENVIRONMENT PROGRAMME
INTERNATIONAL LABOUR ORGANISATION
WORLD HEALTH ORGANIZATION

INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY

ENVIRONMENTAL HEALTH CRITERIA 188

Nitrogen Oxides

(Second Edition)

This report contains the collective views of an international group of
experts and does not necessarily represent the decisions or the stated
policy of the United Nations Environment Programme, the International
Labour Organisation, or the World Health Organization.

First draft prepared by Drs J.A. Graham, L.D. Grant, L.J. Folinsbee,
D.J. Kotchmar and J.H.B. Garner, US Environmental Protection Agency

Published under the joint sponsorship of the United Nations
Environment Programme, the International Labour Organisation, and the
World Health Organization, and produced within the framework of the
Inter-Organization Programme for the Sound Management of Chemicals.

World Health Organization
Geneva, 1997

The International Programme on Chemical Safety (IPCS) is a joint
venture of the United Nations Environment Programme, the International
Labour Organisation, and the World Health Organization. The main
objective of the IPCS is to carry out and disseminate evaluations of
the effects of chemicals on human health and the quality of the
environment. Supporting activities include the development of
epidemiological, experimental laboratory, and risk-assessment methods
that could produce internationally comparable results, and the
development of manpower in the field of toxicology. Other activities
carried out by the IPCS include the development of know-how for coping
with chemical accidents, coordination of laboratory testing and
epidemiological studies, and promotion of research on the mechanisms
of the biological action of chemicals.

The Inter-Organization Programme for the Sound Management of
Chemicals (IOMC) was established in 1995 by UNEP, ILO, the Food and
Agriculture Organization of the United Nations, WHO, the United
Nations Industrial Development Organization and the Organisation for
Economic Co-operation and Development (Participating Organizations),
following recommendations made by the 1992 UN Conference on
Environment and Development to strengthen cooperation and increase
coordination in the field of chemical safety. The purpose of the IOMC
is to promote coordination of the policies and activities pursued by
the Participating Organizations, jointly or separately, to achieve the
sound management of chemicals in relation to human health and the
environment.

WHO Library Cataloguing in Publication Data

Nitrogen oxides - 2nd ed.

(Environmental health criteria ; 188)

1.Nitrogen dioxide 2.Nitrogen oxides
I.Series

ISBN 92 4 157188 8 (NLM Classification: WA 754)
ISSN 0250-863X

The World Health Organization welcomes requests for permission to
reproduce or translate its publications, in part or in full.
Applications and enquiries should be addressed to the Office of
Publications, World Health Organization, Geneva, Switzerland, which
will be glad to provide the latest information on any changes made to
the text, plans for new editions, and reprints and translations
already available.

^(c) World Health Organization 1997

Publications of the World Health Organization enjoy copyright
protection in accordance with the provisions of Protocol 2 of the
Universal Copyright Convention. All rights reserved.

The designations employed and the presentation of the material
in this publication do not imply the expression of any opinion
whatsoever on the part of the Secretariat of the World Health
Organization concerning the legal status of any country, territory,
city or area or of its authorities, or concerning the delimitation of
its frontiers or boundaries.

The mention of specific companies or of certain manufacturers'
products does not imply that they are endorsed or recommended by the
World Health Organization in preference to others of a similar nature
that are not mentioned. Errors and omissions excepted, the names of
proprietary products are distinguished by initial capital letters.

The Federal Ministry for the Environment, Nature Conservation and
Nuclear Safety, Germany, provided financial support for, and
undertook the printing of, this publication

CONTENTS

ENVIRONMENTAL HEALTH CRITERIA FOR NITROGEN OXIDES

Preamble

1. SUMMARY

1.1. Nitrogen oxides and related compounds
1.1.1. Atmospheric transport
1.1.2. Measurement
1.1.3. Exposure
1.2. Effects of atmospheric nitrogen species, particularly
nitrogen oxides, on vegetation
1.3. Health effects of exposures to nitrogen dioxide
1.3.1. Studies of the effects of nitrogen compounds on
experimental animals
1.3.1.1 Biochemical and cellular mechanisms of
action of nitrogen oxides
1.3.1.2 Effects on host defence
1.3.1.3 Effects of chronic exposure on the
development of chronic lung disease
1.3.1.4 Potential carcinogenic or co-carcinogenic
effects
1.3.1.5 Age susceptibility
1.3.1.6 Influence of exposure patterns
1.3.2. Controlled human exposure studies on nitrogen
oxides
1.3.3. Epidemiology studies on nitrogen dioxide
1.3.4. Health-based guidance values for nitrogen dioxide

2. PHYSICAL AND CHEMICAL PROPERTIES, AIR SAMPLING AND ANALYSIS,
TRANSFORMATIONS AND TRANSPORT IN THE ATMOSPHERE

2.1. Introduction
2.1.1. The nomenclature and measurement of atmospheric
nitrogen species
2.2. Nitrogen species and their physical and chemical properties
2.2.1. Nitrogen oxides
2.2.1.1 Nitric oxide
2.2.1.2 Nitrogen dioxide
2.2.1.3 Nitrous oxide
2.2.1.4 Other nitrogen oxides
2.2.2. Nitrogen acids
2.2.2.1 Nitric acid
2.2.2.2 Nitrous acid
2.2.3. Ammonia
2.2.4. Ammonium nitrate
2.2.5. Peroxyacetyl nitrate
2.2.6. Organic nitrites and nitrates

2.3. Sampling and analysis methods
2.3.1. Nitric oxide
2.3.1.1 Nitric oxide continuous methods
2.3.1.2 Passive samplers for NO
2.3.1.3 Calibration of NO analysis methods
2.3.1.4 Sampling considerations for NO
2.3.2. Nitrogen dioxide
2.3.2.1 Chemiluminescence (NO + O₃)
2.3.2.2 Chemiluminescence (luminol)
2.3.2.3 Laser-induced fluorescence and tuneable
diode laser absorption spectrometry
2.3.2.4 Wet chemical methods
2.3.2.5 Other methods
2.3.2.6 Passive samplers
2.3.2.7 Calibration
2.3.3. Total reactive odd nitrogen
2.3.4. Peroxyacetyl nitrate
2.3.5. Other organic nitrates
2.3.6. Nitric acid
2.3.7. Nitrous acid
2.3.8. Dinitrogen pentoxide and nitrate radicals
2.3.9. Particulate nitrate
2.3.10. Nitrous oxide
2.3.11. Summary
2.4. Transport and transformation of nitrogen oxides in the air
2.4.1. Introduction
2.4.2. Chemical transformations of oxides of nitrogen
2.4.2.1 Nitric oxide, nitrogen dioxide and ozone
2.4.2.2 Transformations in indoor air
2.4.2.3 Formation of other oxidized nitrogen
species
2.4.3. Advection and dispersion of atmospheric nitrogen
species
2.4.3.1 Transport of reactive nitrogen species
in urban plumes
2.4.3.2 Air quality models
2.4.3.3 Regional transport
2.5. Conversion factor for nitrogen dioxide
2.6. Summary

3. SOURCES, EMISSIONS AND AIR CONCENTRATIONS

3.1. Introduction
3.2. Sources of nitrogen oxides
3.2.1. Sources of NO_x emission
3.2.1.1 Fuel combustion
3.2.1.2 Biomass burning
3.2.1.3 Lightning
3.2.1.4 Soils
3.2.1.5 Oceans

3.2.2. Removal from the ambient environment
3.2.3. Summary of global budgets for nitrogen oxides
3.3. Ambient concentrations of nitrogen oxides
3.3.1. International comparison studies of NO_x
concentrations
3.3.2. Example case studies of NO_x and NO₂
concentrations
3.4. Occurrence of nitrogen oxides indoors
3.4.1. Indoor sources
3.4.1.1 Gas-fuelled cooking stoves
3.4.1.2 Unvented gas space heaters and water
heaters
3.4.1.3 Kerosene space heaters
3.4.1.4 Wood stoves
3.4.1.5 Tobacco products
3.4.2. Removal of nitrogen oxides from indoor environments
3.5. Indoor concentrations of nitrogen oxides
3.5.1. Homes without indoor combustion sources
3.5.2. Homes with combustion appliances
3.5.3. Homes with combustion space heaters
3.5.4. Indoor nitrous acid concentrations
3.5.5. Predictive models for indoor NO₂ concentration
3.6. Human exposure
3.7. Exposure of plants and ecosystems

4. EFFECTS OF ATMOSPHERIC NITROGEN COMPOUNDS (PARTICULARLY NITROGEN
OXIDES) ON PLANTS

4.1. Properties of NO_x and NH_y
4.1.1. Adsorption and uptake
4.1.2. Toxicity, detoxification and assimilation
4.1.3. Physiology and growth aspects
4.1.4. Interactions with climatic conditions
4.1.5. Interactions with the habitat
4.1.6. Increasing pest incidence
4.1.7. Conclusions for various atmospheric nitrogen
species and mixtures
4.1.7.1 NO₂
4.1.7.2 NO
4.1.7.3 NH₃
4.1.7.4 NH₄⁺ and NO₃^- in wet and occult
deposition
4.1.7.5 Mixtures
4.1.8. Appraisal
4.1.8.1 Representativity of the data
4.1.9. General conclusions
4.2. Effects on natural and semi-natural ecosystems
4.2.1. Effects on freshwater and intertidal ecosystems
4.2.1.1 Effects of nitrogen deposition on
shallow softwater lakes

4.2.1.2 Effects of nitrogen deposition on lakes
and streams
4.2.2. Effects on ombrotrophic bogs and wetlands
4.2.2.1 Effects on ombrotrophic (raised) bogs
4.2.2.2 Effects on mesotrophic fens
4.2.2.3 Effects on fresh- and saltwater marshes
4.2.3. Effects on species-rich grasslands
4.2.3.1 Effects of nitrogen on calcareous
grasslands
4.2.3.2 Critical loads for nitrogen in
calcareous grasslands
4.2.3.3 Comparison with other semi-natural
grasslands
4.2.4. Effects on heathlands
4.2.4.1 Effects on inland dry heathlands
4.2.4.2 Effects of nitrogen on inland wet
heathlands
4.2.4.3 Effects of nitrogen on arctic and alpine
healthlands
4.2.4.4 Effects on herbs of matgrass swards
4.2.5. Effects of nitrogen deposition on forests
4.2.5.1 Effects on forest tree species
4.2.5.2 Effects on tree epiphytes, ground
vegetation and ground fauna of forests
4.2.6. Effects on estuarine and marine ecosystems
4.2.7. Appraisal and conclusions

5. STUDIES OF THE EFFECTS OF NITROGEN OXIDES ON EXPERIMENTAL ANIMALS

5.1. Introduction
5.2. Nitrogen dioxide
5.2.1. Dosimetry
5.2.1.1 Respiratory tract dosimetry
5.2.1.2 Systemic dosimetry
5.2.2. Respiratory tract effects
5.2.2.1 Host defence mechanisms
5.2.2.2 Lung biochemistry
5.2.2.3 Pulmonary function
5.2.2.4 Morphological studies
5.2.3. Genotoxicity, potential carcinogenic or
co-carcinogenic effects
5.2.4. Extrapulmonary effects
5.3. Effects of mixtures containing nitrogen dioxide
5.4. Effects of other nitrogen oxide compounds
5.4.1. Nitric oxide
5.4.1.1 Endogenous formation of NO
5.4.1.2 Absorption of NO
5.4.1.3 Effects of NO on pulmonary function,
morphology and host lung defence
function

5.4.1.4 Metabolic effects
5.4.1.5 Haematological changes
5.4.1.6 Biochemical mechanisms for nitric oxide
effects: reaction with iron and effects
on enzymes and nucleic acids
5.4.2. Nitric acid
5.4.3. Nitrates
5.5. Summary of studies of the effects of nitrogen compounds on
experimental animals

6. CONTROLLED HUMAN EXPOSURE STUDIES OF NITROGEN OXIDES

6.1. Introduction
6.2. Effects of nitrogen dioxide
6.2.1. Nitrogen dioxide effects on pulmonary function and
airway responsiveness to bronchoconstrictive agents
6.2.1.1 Nitrogen dioxide effects in healthy
subjects
6.2.1.2 Nitrogen dioxide effects on asthmatics
6.2.1.3 Nitrogen dioxide effects on patients
with chronic obstructive pulmonary
disease
6.2.1.4 Age-related differential susceptibility
6.2.2. Nitrogen dioxide effects on pulmonary host defences
and bronchoalveolar lavage fluid biomarkers
6.2.3. Other classes of nitrogen dioxide effects
6.3. Effects of other nitrogen oxide compounds
6.4. Effects of nitrogen dioxide/gas or gas/aerosol mixtures on
lung function
6.5. Summary of controlled human exposure studies of oxides of
nitrogen

7. EPIDEMIOLOGICAL STUDIES OF NITROGEN OXIDES

7.1. Introduction
7.2. Methodological considerations
7.2.1. Measurement error
7.2.2. Misclassification of the health outcome
7.2.3. Adjustment for covariates
7.2.4. Selection bias
7.2.5. Internal consistency
7.2.6. Plausibility of the effect
7.3. Studies of respiratory illness
7.3.1. Indoor air studies
7.3.1.1 St Thomas' Hospital Medical School
Studies (United Kingdom)
7.3.1.2 Harvard University - Six Cities Studies
(USA)
7.3.1.3 University of Iowa Study (USA)

7.3.1.4 Agricultural University of Wageningen
(The Netherlands)
7.3.1.5 Ohio State University Study (USA)
7.3.1.6 University of Dundee (United Kingdom)
7.3.1.7 Harvard University - Chestnut Ridge
Study (USA)
7.3.1.8 University of New Mexico Study (USA)
7.3.1.9 University of Basel Study (Switzerland)
7.3.1.10 Yale University Study (USA)
7.3.1.11 Freiburg University Study (Germany)
7.3.1.12 McGill University Study (Canada)
7.3.1.13 Health and Welfare Canada Study (Canada)
7.3.1.14 University of North Carolina Study (USA)
7.3.1.15 University of Tucson Study (USA)
7.3.1.16 Hong Kong Anti-Cancer Society Study
(Hong Kong)
7.3.1.17 Recent studies
7.3.2. Outdoor studies
7.3.2.1 Harvard University - Six City Studies
(USA)
7.3.2.2 University of Basel Study (Switzerland)
7.3.2.3 University of Wuppertal Studies
(Germany)
7.3.2.4 University of Tubigen (Germany)
7.3.2.5 Harvard University - Chestnut Ridge
Study (USA)
7.3.2.6 University of Helsinki Studies (Finland)
7.3.2.7 Helsinki City Health Department Study
(Finland)
7.3.2.8 Oulu University Study (Finland)
7.3.2.9 Seth GS Medical College Study (India)
7.4. Pulmonary function studies
7.4.1. Harvard University - Six City Studies (USA)
7.4.2. National Health and Nutrition Examination Survey
Study (USA)
7.4.3. Harvard University - Chestnut Ridge Study (USA)
7.4.4. Other pulmonary function studies
7.5. Other exposure settings
7.5.1. Skating rink exposures
7.6. Occupational exposures
7.7. Synthesis of the evidence for school-age children
7.7.1. Health outcome measures
7.7.2. Biologically plausible hypothesis
7.7.3. Publication bias
7.7.4. Selection of studies
7.7.4.1 Brief description of selected studies
7.7.4.2 Studies not selected for quantitative
analysis
7.7.5. Quantitative analysis

7.8. Synthesis of the evidence for young children
7.9. Summary

8. EVALUATION OF HEALTH AND ENVIRONMENT RISKS ASSOCIATED WITH
NITROGEN OXIDES

8.1. Sources and exposure
8.2. Evaluation of the effects of atmospheric nitrogen species
on the environment
8.2.1. Guidance values - critical levels for air
concentrations of nitrogen oxides
8.2.2. Environment-based guidance values - critical loads
for total nitrogen deposition
8.3. Evaluation of health risks associated with nitrogen oxides
8.3.1. Concentration-response relationships
8.3.2. Subpopulations potentially at risk
8.3.3. Derivation of health-based guidance values

9. CONCLUSIONS AND RECOMMENDATIONS FOR PROTECTION OF HUMAN HEALTH
AND THE ENVIRONMENT

10. FURTHER RESEARCH

REFERENCES

RESUME

RESUMEN

NOTE TO READERS OF THE CRITERIA MONOGRAPHS

Every effort has been made to present information in the criteria
monographs as accurately as possible without unduly delaying their
publication. In the interest of all users of the Environmental Health
Criteria monographs, readers are requested to communicate any errors
that may have occurred to the Director of the International Programme
on Chemical Safety, World Health Organization, Geneva, Switzerland, in
order that they may be included in corrigenda.

* * *

A detailed data profile and a legal file can be obtained from the
International Register of Potentially Toxic Chemicals, Case postale
356, 1219 Châtelaine, Geneva, Switzerland (Telephone No. 9799111).

Environmental Health Criteria

PREAMBLE

Objectives

In 1973 the WHO Environmental Health Criteria Programme was
initiated with the following objectives:

(i) to assess information on the relationship between exposure to
environmental pollutants and human health, and to provide
guidelines for setting exposure limits;

(ii) to identify new or potential pollutants;

(iii) to identify gaps in knowledge concerning the health effects of
pollutants;

(iv) to promote the harmonization of toxicological and
epidemiological methods in order to have internationally
comparable results.

The first Environmental Health Criteria (EHC) monograph, on
mercury, was published in 1976 and since that time an ever-increasing
number of assessments of chemicals and of physical effects have been
produced. In addition, many EHC monographs have been devoted to
evaluating toxicological methodology, e.g., for genetic, neurotoxic,
teratogenic and nephrotoxic effects. Other publications have been
concerned with epidemiological guidelines, evaluation of short-term
tests for carcinogens, biomarkers, effects on the elderly and so
forth.

Since its inauguration the EHC Programme has widened its scope,
and the importance of environmental effects, in addition to health
effects, has been increasingly emphasized in the total evaluation of
chemicals.

The original impetus for the Programme came from World Health
Assembly resolutions and the recommendations of the 1972 UN Conference
on the Human Environment. Subsequently the work became an integral
part of the International Programme on Chemical Safety (IPCS), a
cooperative programme of UNEP, ILO and WHO. In this manner, with the
strong support of the new partners, the importance of occupational
health and environmental effects was fully recognized. The EHC
monographs have become widely established, used and recognized
throughout the world.

The recommendations of the 1992 UN Conference on Environment and
Development and the subsequent establishment of the Intergovernmental
Forum on Chemical Safety with the priorities for action in the six
programme areas of Chapter 19, Agenda 21, all lend further weight to
the need for EHC assessments of the risks of chemicals.

Scope

The criteria monographs are intended to provide critical reviews
on the effect on human health and the environment of chemicals and of
combinations of chemicals and physical and biological agents. As
such, they include and review studies that are of direct relevance for
the evaluation. However, they do not describe every study carried
out. Worldwide data are used and are quoted from original studies,
not from abstracts or reviews. Both published and unpublished reports
are considered and it is incumbent on the authors to assess all the
articles cited in the references. Preference is always given to
published data. Unpublished data are only used when relevant
published data are absent or when they are pivotal to the risk
assessment. A detailed policy statement is available that describes
the procedures used for unpublished proprietary data so that this
information can be used in the evaluation without compromising its
confidential nature (WHO (1990) Revised Guidelines for the Preparation
of Environmental Health Criteria Monographs. PCS/90.69, Geneva, World
Health Organization).

In the evaluation of human health risks, sound human data,
whenever available, are preferred to animal data. Animal and
in vitro studies provide support and are used mainly to supply
evidence missing from human studies. It is mandatory that research on
human subjects is conducted in full accord with ethical principles,
including the provisions of the Helsinki Declaration.

The EHC monographs are intended to assist national and
international authorities in making risk assessments and subsequent
risk management decisions. They represent a thorough evaluation of
risks and are not, in any sense, recommendations for regulation or
standard setting. These latter are the exclusive purview of national
and regional governments.

Content

The layout of EHC monographs for chemicals is outlined below.

* Summary - a review of the salient facts and the risk evaluation
of the chemical
* Identity - physical and chemical properties, analytical methods
* Sources of exposure
* Environmental transport, distribution and transformation
* Environmental levels and human exposure
* Kinetics and metabolism in laboratory animals and humans

* Effects on laboratory mammals and in vitro test systems
* Effects on humans
* Effects on other organisms in the laboratory and field
* Evaluation of human health risks and effects on the environment
* Conclusions and recommendations for protection of human health
and the environment
* Further research
* Previous evaluations by international bodies, e.g., IARC, JECFA,
JMPR

Selection of chemicals

Since the inception of the EHC Programme, the IPCS has organized
meetings of scientists to establish lists of priority chemicals for
subsequent evaluation. Such meetings have been held in: Ispra, Italy,
1980; Oxford, United Kingdom, 1984; Berlin, Germany, 1987; and North
Carolina, USA, 1995. The selection of chemicals has been based on the
following criteria: the existence of scientific evidence that the
substance presents a hazard to human health and/or the environment;
the possible use, persistence, accumulation or degradation of the
substance shows that there may be significant human or environmental
exposure; the size and nature of populations at risk (both human and
other species) and risks for environment; international concern, i.e.
the substance is of major interest to several countries; adequate data
on the hazards are available.

If an EHC monograph is proposed for a chemical not on the
priority list, the IPCS Secretariat consults with the Cooperating
Organizations and all the Participating Institutions before embarking
on the preparation of the monograph.

Procedures

The order of procedures that result in the publication of an EHC
monograph is shown in the flow chart. A designated staff member of
IPCS, responsible for the scientific quality of the document, serves
as Responsible Officer (RO). The IPCS Editor is responsible for
layout and language. The first draft, prepared by consultants or,
more usually, staff from an IPCS Participating Institution, is based
initially on data provided from the International Register of
Potentially Toxic Chemicals, and reference data bases such as Medline
and Toxline.

The draft document, when received by the RO, may require an
initial review by a small panel of experts to determine its scientific
quality and objectivity. Once the RO finds the document acceptable as
a first draft, it is distributed, in its unedited form, to well over
150 EHC contact points throughout the world who are asked to comment
on its completeness and accuracy and, where necessary, provide
additional material. The contact points, usually designated by
governments, may be Participating Institutions, IPCS Focal Points, or

individual scientists known for their particular expertise. Generally
some four months are allowed before the comments are considered by the
RO and author(s). A second draft incorporating comments received and
approved by the Director, IPCS, is then distributed to Task Group
members, who carry out the peer review, at least six weeks before
their meeting.

The Task Group members serve as individual scientists, not as
representatives of any organization, government or industry. Their
function is to evaluate the accuracy, significance and relevance of
the information in the document and to assess the health and
environmental risks from exposure to the chemical. A summary and
recommendations for further research and improved safety aspects are
also required. The composition of the Task Group is dictated by the
range of expertise required for the subject of the meeting and by the
need for a balanced geographical distribution.

The three cooperating organizations of the IPCS recognize
the important role played by nongovernmental organizations.
Representatives from relevant national and international associations
may be invited to join the Task Group as observers. While observers
may provide a valuable contribution to the process, they can only
speak at the invitation of the Chairperson. Observers do not
participate in the final evaluation of the chemical; this is the sole
responsibility of the Task Group members. When the Task Group
considers it to be appropriate, it may meet in camera.

All individuals who as authors, consultants or advisers
participate in the preparation of the EHC monograph must, in addition
to serving in their personal capacity as scientists, inform the RO if
at any time a conflict of interest, whether actual or potential, could
be perceived in their work. They are required to sign a conflict of
interest statement. Such a procedure ensures the transparency and
probity of the process.

When the Task Group has completed its review and the RO is
satisfied as to the scientific correctness and completeness of the
document, it then goes for language editing, reference checking, and
preparation of camera-ready copy. After approval by the Director,
IPCS, the monograph is submitted to the WHO Office of Publications for
printing. At this time a copy of the final draft is sent to the
Chairperson and Rapporteur of the Task Group to check for any errors.

It is accepted that the following criteria should initiate the
updating of an EHC monograph: new data are available that would
substantially change the evaluation; there is public concern for
health or environmental effects of the agent because of greater
exposure; an appreciable time period has elapsed since the last
evaluation.

All Participating Institutions are informed, through the EHC
progress report, of the authors and institutions proposed for the
drafting of the documents. A comprehensive file of all comments
received on drafts of each EHC monograph is maintained and is
available on request. The Chairpersons of Task Groups are briefed
before each meeting on their role and responsibility in ensuring that
these rules are followed.

WHO TASK GROUP ON ENVIRONMENTAL HEALTH CRITERIA FOR NITROGEN OXIDES

Members

Dr K. Bentley^*, Health and Environment Policy Section, Department
of Community Services and Health, Canberra ACT, Australia

Dr S. Dobson, Institute of Terrestrial Ecology, Monks Wood
Experimental Station, Abbots Ripton, Huntingdon, Cambridgeshire,
United Kingdom

Dr L. van der Eerden, Centre "De Bom" Wageningen, The Netherlands

Dr L. Folinsbee, Health Effects Research Laboratory, US Environmental
Protection Agency, Research Triangle Park, North Carolina, USA
(Rapporteur)

Dr L. Grant^*, National Center for Environmental Assessment, US
Environmental Protection Agency, Research Triangle Park, North
Carolina, USA

Mr L. Heiskanen, Health and Environment Policy Section, Department of
Community Services and Health, Canberra ACT, Australia

Mr G.M. Johnson, CSIRO, Division of Coal and Energy Technology, Centre
for Pollution Assessment and Control, North Ryde, NSW, Australia

Dr J. Kagawa, Professor of Hygiene and Public Health, Tokyo Women's
Medical College, Shinjuku-ku, Tokyo, Japan

Dr R.R. Khan, Ministry of Environment and Forests, Paryavaran Bhawan,
New Delhi, India

Dr D.B. Menzel, University of California, Department of Community &
Environment and Medicine, California, USA

Dr L. Neas, Department of Environmental Health, Environmental
Epidemiology Program, Harvard School of Public Health, Boston,
Massachusetts, USA

Dr S.E. Paulson, Department of Atmospheric Sciences, University of
California, Los Angeles, California, USA

Dr P.J.A. Rombout, Department for Inhalation Toxicology, National
Institute of Public Health and Environmental Hygiene, Bilthoven,
The Netherlands (Chairman)

^* Invited, but unable to attend

Dr W. Tyler, Veterinary Anatomy and Cell Biology, University of
California, California, USA

Dr K. Victorin, Karolinska Institute, Institute of Environmental
Medicine, Stockholm, Sweden

Dr A. Woodward, Department of Community Medicine, University of
Adelaide, Adelaide, Australia

Dr R. Ye, Deputy Director, National Environmental Protection Agency,
Xizhimennei Nanziaojie, Beijing, People's Republic of China

Observers

Professor M. Moore, National Research Centre for Environmental
Toxicology, Nathan, Australia

Dr M. Pain, Department of Thoracic Medicine, Royal Melbourne Hospital,
Melbourne VIC, Australia

Dr P. Psaila-Savona, WA Department of Health, Perth WA, Australia

Mr B. Taylor, Policy and Planning Group, Public and Planning Group,
Public Health Commission, Wellington, New Zealand

Mr B. Saxby, AGL Gas Companies, North Sydney NSW, New Zealand

Secretariat

Dr B.H. Chen, International Programme on Chemical Safety, World Health
Organization, Geneva, Switzerland (Secretary)

Dr M. Younes, WHO European Centre for Environment & Health, Bilthoven,
The Netherlands

ENVIRONMENTAL HEALTH CRITERIA FOR NITROGEN OXIDES

A WHO Task Group on Environmental Health Criteria for Nitrogen
Oxides met in Melbourne, Australia from 14 to 18 November 1994. The
meeting was hosted by the Clean Air Society of Australia and New
Zealand and the Victorian Departments of Health and Environment,
Australia. Dr B.H. Chen, IPCS, opened the meeting and welcomed the
participants on behalf of the Director, IPCS, and the three IPCS
cooperating organizations (UNEP/ILO/WHO). The Task Group reviewed and
revised the draft criteria monograph and made an evaluation of the
risks for human health and the environment from exposure to nitrogen
oxides.

The first draft of this monograph was prepared by Drs J.A.
Graham, L.D. Grant, L.J. Folinsbee, D.J. Kotchmar and J.H.B. Garner,
US EPA. Drs W.G. Ewald, T.B. McMullen and B.E. Tilton, US EPA,
contributed to the preparation of the first draft. The second draft
was prepared by Dr L.D. Grant incorporating comments received
following the circulation of the first draft to the IPCS Contact
Points for Environmental Health Criteria. Drs R. Bobbink, L. Van der
Eerden and S. Dobson prepared the final text of the environmental
section. Mr G.M. Johnson contributed to the final text of the
chemistry section.

Dr B.H. Chen and Dr P.G. Jenkins, both members of the IPCS
Central Unit, were responsible for the overall scientific content and
technical editing, respectively.

The efforts of all who helped in the preparation and finalization
of the document are gratefully acknowledged.

Financial support for this Task Group meeting was provided by the
Department of Community Services and Health, Australia, Victorian
Departments of Health and Environment, Australia, and the Clean Air
Society of Australia and New Zealand.

ABBREVIATIONS

ADP adenosine diphosphate
AM alveolar macrophages
AQG Air Quality Guidelines
BAL bronchoalveolar lavage
BHPN N-bis (2-hydroxypropyl) nitrosamine
CI confidence interval
CLM chemiluminescence method
COPD chronic obstructive pulmonary disease
ECD electron capture detection
FEF forced expiratory flow
FEV forced expiratory volume
FTIR Fourier transformed infrared
FVC forced vital capacity
GC gas chromatography
GDH glutamate dehydrogenase
(c)GMP (cyclic) guanosine monophosphate
GS glutamine synthetase
HNO₂ nitrous acid
HNO₃ nitric acid
LIF laser-induced fluorescence
MS mass spectrometry
N₂ nitrogen (elemental)
NH₃ ammonia
NH₄⁺ ammonium ion
NH_y the sum of NH₃ and NH₄⁺
NiR nitrate reductase
NK natural killer
NO nitric oxide
NO₂ nitrogen dioxide
NO₂^- nitrite ion
NO₃^- nitrate ion
N₂O nitrous oxide
N₂O₅ nitrogen pentoxide
NO_x nitric oxide plus nitrogen dioxide
NO_y gas-phase oxidized nitrogen species (except nitrous oxide)
NPSH non-protein sulfhydryl
NR nitrate reductase
O₃ ozone
PAN peroxyacetyl nitrate
PBzN peroxybenzoyl nitrate
PEF peak expiratory flow
PFC plaque-forming cell
PMN polymorphonuclear leukocyte
ppb parts per billion (10^-9)
ppm parts per million (10^-6)
ppt parts per trillion (10^-12)
pptv parts per trillion (by volume)
PSD passive sampling device

R_aw airway resistance
ROC reactive organic carbon
RUBISCO ribulose 1,5-biphosphate carboxylase
SD standard deviation
SES socioeconomic status
SG_aw specific airway conductance
SO₂ sulfur dioxide
SO_y sulfur oxides
SPM suspended particulate matter
SR_aw specific airway resistance
TDLAS tuneable diode laser absorption spectrometry
TSP total suspended particulate
VOC volatile organic carbon

1. SUMMARY

1.1 Nitrogen oxides and related compounds

Nitrogen oxides can be present at significant concentrations in
ambient air and in indoor air. The types and concentrations of
nitrogenous compounds present can vary greatly from location to
location, with time of day, and with season. The main sources of
nitrogen oxide emissions are combustion processes. Fossil fuel power
stations, motor vehicles and domestic combustion appliances emit
nitrogen oxides, mostly in the form of nitric oxide (NO) and some
(usually less than about 10%) in the form of nitrogen dioxide (NO₂).
In the air, chemical reactions occur that oxidize NO to NO₂ and other
products. There are also biological processes that liberate nitrogen
species from soils, including nitrous oxide (N₂O). Emissions of N₂O
can cause perturbation of the stratospheric ozone layer.

Human health may be affected when significant concentrations of
NO₂ or other nitrogenous species, such as peroxyacetyl nitrate (PAN),
nitric acid (HNO₃), nitrous acid (HNO₂), and nitrated organic
compounds, are present. In addition, nitrates and HNO₃ may cause
health effects and significant effects on ecosystems when deposited on
the ground.

The sum of NO and NO₂ is generally referred to as NO_x. Once
released into the air, NO is oxidized to NO₂ by available oxidants
(particularly ozone, O₃). This happens rapidly under some conditions
in outdoor air; in indoor air, it is generally a much slower process.
Nitrogen oxides are a controlling precursor of photochemical oxidant
air pollution resulting in ozone and smog formation; interactions of
nitrogen oxides (except N₂O) with reactive organic compounds and
sunlight form ozone in the troposphere and smog in urban areas.

NO and NO₂ may also undergo reactions to form a range of other
oxides of nitrogen, both in indoor and outdoor air, including HNO₂,
HNO₃, nitrogen trioxide (NO₃), dinitrogen pentoxide (N₂O₅), PAN
and other organic nitrates. The complex range of gas-phase nitrogen
oxides is referred to as NO_y. The partitioning of oxides of nitrogen
among these compounds is strongly dependent on the concentrations of
other oxidants and on the meteorological history of the air.

HNO₃ is formed from the reaction of OH^- and NO₂. It is a
major sink for active nitrogen and also a contributor to acidic
deposition. Potential physical and chemical sinks for HNO₃ include
wet and dry deposition, photolysis, reaction with OH radicals, and
reaction with gaseous ammonia to form ammonium nitrate aerosol.

PANs are formed from the combination of organic peroxy radicals
with NO₂. PAN is the most abundant organic nitrate in the
troposphere and can serve as a temporary reservoir for reactive
nitrogen, which may be regionally transported.

The NO₃ radical, a short-lived NO_y species that is formed in
the troposphere primarily by the reaction of NO₂ with O₃, undergoes
rapid photolysis in daylight or reaction with NO. Appreciable
concentrations are observed during the night.

N₂O₅ is primarily a night-time constituent of ambient air as it
is formed from the reaction of NO₃ and NO₂. In ambient air, N₂O₅
reacts heterogeneously with water to form HNO₃, which in turn is
deposited.

N₂O is ubiquitous because it is a product of natural biological
processes in soil. It is not known, however, to be involved in any
reactions in the troposphere. N₂O participates in upper atmospheric
reactions contributing to stratospheric ozone (O₃) depletion and is
also a relatively potent greenhouse gas that contributes to global
warming.

1.1.1 Atmospheric transport

The transport and dispersion of the various nitrogenous
species in the lower troposphere is dependent on both meteorological
and chemical parameters. Advection, diffusion and chemical
transformations combine to dictate the atmospheric residence times.
In turn, atmospheric residence times help determine the geographic
extent of transport of given species. Surface emissions are dispersed
vertically and horizontally through the atmosphere by turbulent mixing
processes that are dependent to a large extent on the vertical
temperature structure and wind speed.

As the result of meteorological processes, NO_x emitted in the
early morning hours in an urban area typically disperses vertically
and moves downwind as the day progresses. On sunny summer days, most
of the NO_x will have been converted to HNO₃ and PAN by sunset, with
concomitant formation of ozone. Much of the HNO₃ is removed by
deposition as the air mass is transported, but HNO₃ and PAN carried
in layers aloft (above the nighttime inversion layer but below a
higher subsidence inversion) can potentially be transported long
distances in oxidant-laden air masses.

1.1.2 Measurement

There are a number of methods available to measure airborne
nitrogen-containing species. This document briefly covers
methodologies currently available or in general use for in situ

monitoring of airborne concentrations in both ambient and indoor
environments. The species considered are NO, NO₂, NO_x, total
reactive odd nitrogen (NO_y), PAN and other organic nitrates, HNO₃,
HNO₂, N₂O₅, the nitrate radical, NO₃^-, and N₂O.

Measuring concentrations of nitrogen oxides is not trivial.
While a straightforward, widely available method exists for measuring
NO (the chemiluminescent reaction with ozone), this is an exception
for nitrogen oxides. Chemiluminescence is also the most common
technique used for NO₂; NO₂ is first reduced to NO. Unfortunately,
the catalyst typically used for the reduction is not specific, and has
various conversion efficiencies for other oxidized nitrogen compounds.
For this reason, great care must be taken in interpreting the results
of the common chemiluminescence analyser in terms of NO₂, as the
signal may include many other compounds. Additional difficulties
arise from nitrogen oxides that may partition between the gaseous and
particulate phases both in the atmosphere and in the sampling
procedure.

1.1.3 Exposure

Human and environmental exposure to nitrogen oxides varies
greatly from indoors to outdoors, from cities to the countryside, and
with time of day and season. The concentrations of NO and NO₂
typically present outdoors in a range of urban situations are
relatively well established. The concentrations encountered indoors
depend on the specific details of the nature of combustion appliances,
chimneys and ventilation. When unvented combustion appliances are
used for cooking or heating, indoor concentrations of nitrogen oxides
typically greatly exceed those existing outside. Recent research has
shown in these circumstances that HNO₂ can reach significant
concentrations. One report showed that HNO₂ can represent over 10% of
the concentrations usually reported as NO₂.

1.2 Effects of atmospheric nitrogen species, particularly nitrogen
oxides, on vegetation

Most of earth's biodiversity is found in (semi-)natural
ecosystems, both in aquatic and terrestrial habitats. Nitrogen is the
limiting nutrient for plant growth in many (semi-)natural ecosystems.
Most of the plant species from these habitats are adapted to nutrient-
poor conditions, and can only compete successfully on soils with low
nitrogen levels.

Human activities, both industrial and agricultural, have greatly
increased the amount of biologically available nitrogen compounds,
thereby disturbing the natural nitrogen cycle. Various forms of
nitrogen pollute the air: mainly NO, NO₂ and ammonia (NH₃) as dry
deposition; and nitrate (NO₃^-) and ammonium (NH₄⁺) as wet
deposition. NH_y refers to the sum of NH₃ and NH₄⁺. Another

contribution is from occult deposition (fog and clouds). There are
many more nitrogen-containing air pollutants (e.g., N₂O₅, PAN, N₂O,
amines), but these are neglected here, either because their
contribution to the total nitrogen deposition is believed to be small,
or because their concentrations are probably far below effect
thresholds.

Nitrogen-containing air pollutants can affect vegetation
indirectly, via photochemical reaction products, or directly after
being deposited on vegetation, soil or water surface. The indirect
pathway is largely neglected here although it includes very relevant
processes, and should be taken into account when evaluating the entire
impact of nitrogen-containing air pollutants: NO₂ is a precursor for
tropospheric O₃, which acts both as a phytotoxin and a greenhouse
gas.

The impacts of increased nitrogen deposition upon biological
systems can be the result of direct uptake by foliage or uptake via
the soil. At the level of individual plants, the most relevant
effects are injury to the tissue, changes in biomass production and
increased susceptibility to secondary stress factors. At the
vegetation level, deposited nitrogen acts as a nutrient; this results
in changes in competitive relationships between species and loss of
biodiversity. The critical loads for nitrogen depend on (i) the type
of ecosystem; (ii) the land use and management in the past and
present; and (iii) the abiotic conditions (especially those that
influence the nitrification potential and immobilization rate in the
soil).

Adsorption on the outer surface of the leaves takes place and may
damage wax layers of the cuticle, but the quantitative relevance for
the field situation has not yet been proved. Uptake of NO_x and NH₃
is driven by the concentration gradient between atmosphere and
mesophyll. It generally, but not always, is directly determined by
stomatal conductance and thus depends on factors influencing stomatal
aperture. There is increasing evidence that foliar uptake of nitrogen
reduces the uptake of nitrogen by the roots. Uptake and exchange of
ions through the leaf surface is a relatively slow process, and thus
is only relevant if the surface remains wet for longer periods.

NO is only slightly soluble in water, but the presence of other
substances can alter the solubility. NO₂ has a higher solubility,
while that of NH₃ is much higher. NO₂^- (the primary reaction
product of NO_x), NH₃ and NH₄⁺ are all highly phytotoxic, and could
well be the cause of adverse effects of nitrogen-containing air
pollutants. The free radical *N=O may play a role in the phytotoxicity
of NO.

More-than-additive effects (synergism) have been found in nearly
all studies concerning SO₂ plus NO₂. With other NO₂ mixtures (NO,
O₃ and CO₂), interactive effects are the exception rather than the
rule.

When climatic conditions and supply of other nutrients allow
biomass production, both NO_x and NH_y result in growth stimulation at
low concentrations and growth reduction at higher concentrations.
However, the exposure level at which growth stimulation turns into
growth inhibition is much lower for NO_x than for NH_y.

Evidence exists that plants are more sensitive at low light
intensity (e.g., at night and in winter) and at low temperatures (just
above 0°C). NO_x and NH_y can increase the sensitivity of plants to
frost, drought, wind and insect damage.

An interaction exists between soil chemistry and sensitivity of
vegetation to nitrogen deposition; this is related to pH and nitrogen
availability.

The relative contribution of NO and NO₂ to the NO_x effect on
plants is unclear. The vast majority of information is on effects of
NO₂ but available information on NO suggests that NO and NO₂ have
comparable phytotoxic effects.

Air quality guidelines refer to thresholds for adverse effects.
Two different types of effect thresholds exist: critical levels (CLEs)
and critical loads (CLOs). The critical level is defined as the
concentration in the atmosphere above which direct adverse effects on
receptors, such as plants, ecosystems or materials, may occur
according to present knowledge. The critical load is defined as a
quantitative estimate of an exposure (deposition) to one or more
pollutants below which significant harmful effects on specified
sensitive elements of the environment do not occur according to
present knowledge.

According to current practice, critical levels have been derived
from assessment of the lowest exposure concentrations causing adverse
effects on physiology or growth of plants (biochemical effects were
excluded), using a graphical method.

To include the impact of NO, a critical level for NO_x is
proposed instead of one for NO₂; for this purpose it has been assumed
that NO and NO₂ act in an additive manner. A strong case can be made
for the provision of critical levels for short-term exposure. However,
currently there are insufficient data to provide these with sufficient
confidence. Current evidence suggests a critical level of about
75 µg/m³ for NO_x as a 24-h mean.

The critical level for NO_x (NO and NO₂ added in ppb and
expressed as NO₂ in µg/m³) is considered to be 30 µg/m³ as an
annual mean.

Information on organisms in the environment is almost exclusively
restricted to plants, with minimum data on soil fauna. This
evaluation and guidance values are, therefore, expressed in terms of
nitrogen species effects on vegetation. However, it is expected that
plants will form the most sensitive component of natural systems and
that the effect on biodiversity of plant communities is a sensitive
indicator of effects on the whole ecosystem.

Critical loads are derived from empirical data and steady-state
soil models. Estimated critical loads for total nitrogen deposition
in a variety of natural aquatic and terrestrial ecosystems are given.
Possible differential effects of deposited nitrogen species (NO_x and
NH_y) are insufficiently known to differentiate between nitrogen
species for critical load estimation.

The great majority of ecosystems for which there is sufficient
information to estimate critical loads are from temperate climates.
The few arctic and montane ecosystems included, which might be
expected to be representative of higher latitudes, have the least
reliable basis. There is no information on tropical ecosystems and
little on estuarine or marine ecosystems in any climatic zone.
Nutrient-poor tropical ecosystems such as rain forests and mangrove
swamps are likely to be adversely affected by nitrogen deposition.
The lack of both deposition data and effect thresholds make it
impossible to make risk assessments for these climatic regions.

The most sensitive ecosystems (ombrotrophic bogs, shallow soft-
water lakes and arctic and alpine heaths) for which effects thresholds
can be estimated show critical loads of 5-10 kg N.ha^-1.year^-1 based
on decreased biological diversity in plant communities. A more
average value for the limited range of ecosystems studied is 15-20 kg
N.ha^-1.year^-1, which applies to forest trees.

The atmospheric chemistry of nitrogen oxides includes the
capacity for ozone generation in the troposphere, ozone depletion in
the stratosphere, and contribution to global warming as greenhouse
gases. Nitrogen oxides and ammonia contribute to soil acidification
(along with sulfur oxides) and thereby to increased bioavailability of
aluminium.

The phytotoxic effects of nitrogen oxides on plants have little
direct relevance to crop plants when concentrations marginally exceed
the critical level. However, the role of NO_x in the generation of
ozone and other phytotoxic substances, e.g., organic nitrates leads to

crop loss. Nitrogen deposited on growing crops will represent a very
small increase in total available nitrogen compared to that added as
fertilizer.

1.3 Health effects of exposures to nitrogen dioxide

A large number of studies designed to evaluate the health effects
of NO_x have been conducted. Of the NO_x compounds, NO₂ has been
most studied. The discussion in this section focuses on NO₂, NO,
HNO₂ and HNO₃, while nitrates are mentioned briefly.

1.3.1 Studies of the effects of nitrogen compounds on experimental
animals

Extrapolating animal data to humans has both qualitative and
quantitative components. As summarized below, NO₂ causes a
constellation of effects in several animal species; most notably,
effects on host defence against infectious pulmonary disease, lung
metabolism/biochemistry, lung function and lung structure. Because of
basic physiological, metabolic and structural similarities in all
mammals (laboratory animals and humans), the commonality of the
observations in several animal species leads to a reasonable
conclusion that NO₂ could cause similar types of effects in humans.
However, because of the differences between mammalian species, exactly
what exposures would actually cause these effects in humans is not yet
known. That is the topic of quantitative extrapolation. Limited
modelling research on the dosimetric aspect (i.e., the dose to the
target tissue/cell that actually causes toxicity) of quantitative
extrapolation suggests that the distribution of the deposition of NO₂
within the respiratory tract of animals and humans is similar,
without yet providing adequate values to use for animal-to-human
extrapolation. Unfortunately, very little information is available on
the other key aspect of extrapolation, species sensitivity (i.e., the
response of the tissues of different species to a given dose). Thus,
from currently available animal studies, we know which human health
effects NO₂ may cause. We are unable to assert with great confidence
the effects that are actually caused by a given inhaled dose of
NO₂.

With the above issues in mind, the animal toxicology database
for NO₂ is summarized below according to major classes of effects
and topics of special interest. Although it is clear that the
effects of NO₂ exposure extend beyond the confines of the lung, the
interpretation of these systemic effects relative to potential human
risk is not clear. Therefore they are not summarized further here,
but are discussed in later chapters. Although interactions of NO₂
and other co-occurring pollutants, such as O₃ and sulfuric acid
(H₂SO₄), can be quite important, especially if synergism occurs, the
database does not yet allow conclusions that enable assessment of
real-world potential interactions.

1.3.1.1 Biochemical and cellular mechanisms of action of nitrogen
oxides

NO₂ acts as a strong oxidant. Unsaturated lipids are readily
oxidized with peroxides as the dominant product. Both ascorbic acid
(vitamin C) and alpha-tocopherol (vitamin E) inhibit the peroxidation
of unsaturated lipids. When ascorbic acid is sealed within bilayer
liposomes, NO₂ rapidly oxidizes the sealed ascorbic acid. The
protective effects of alpha-tocopherol and ascorbic acid in animals
and humans are due to the inhibition of NO₂ oxidation. NO₂ also
oxidizes membrane proteins. The oxidation of either membrane lipids
or proteins results in the loss of cell permeability control. The
lungs of NO₂-exposed humans and experimental animals have larger
amounts of protein within the lumen. The recruitment of inflammatory
cells and the changes in the lung are due to these events.

The oxidant properties of NO₂ also induce the peroxide
detoxification pathway of glutathione peroxidase, glutathione
reductase and glucose-6-phosphate dehydrogenase. Following NO₂
exposure the increase in the peroxide detoxification pathway in
animals follows an exposure-response relationship.

The mechanism of action of NO is less clear. NO is readily
oxidized to NO₂ and peroxidation then occurs. Because of the
concurrent exposure to some NO₂ in NO exposures, it is difficult to
discriminate NO effects from NO₂. NO functions as an intracellular
second messenger modulating a wide variety of essential enzymes, and
it inhibits its own production (e.g., negative feedback). NO
activates guanylate cyclase which in turn increases intracellular cGMP
levels. A possible mechanism of action of nitrates may be through the
release of histamine from mast cell granules. Acidic nitrogenous air
pollutants, particularly HNO₃, may act by alteration of intracellular
pH.

PAN decomposes in water, generating hydrogen peroxide. Little is
known of the mechanism of action, but oxidative stress is likely for
PAN and its congeners.

Inorganic nitrates may act through alterations in intracellular
pH. Nitrate ion is transported into alveolar type 2 cells acidifying
the cell. Nitrate also mobilizes histamine from mast cells. HNO₂
could also act to alter intracellular pH, but this mechanism is
unclear.

The mechanisms of action of the other nitrogen oxides are
unknown.

Acute exposure to NO₂ at a concentration of 750 µg/m³ (0.4 ppm)
can result in lipid peroxidation. NO₂ can oxidize polyunsaturated

fatty acids in cell membranes as well as functional groups of proteins
(either soluble proteins in the cell, such as enzymes, or structural
proteins, such as components of cell membranes). Such oxidation
reactions (mediated by free radicals) are a mechanism by which NO₂
exerts direct toxicity on lung cells. This mechanism of action is
supported by animal studies showing the importance of lung antioxidant
defences, both endogenous (e.g., maintenance of lung glutathione
levels) and exogenous (e.g., dietary vitamins C and E), in protecting
against the effects of NO₂. Many studies have suggested that various
enzymes in the lung, including glutathione peroxidase, superoxide
dismutase and catalase, may also serve to defend the lung against
oxidant attack.

1.3.1.2 Effects on host defence

Although the primary function of the respiratory tract is to
ensure an efficient exchange of gases, this organ system also provides
the body with a first line of defence against inhaled viable and non-
viable airborne agents. An extensive database clearly shows that
exposure to NO₂ can result in the dysfunction of these host defences,
increasing susceptibility to infectious respiratory disease. The
host-defence parameters affected by NO₂ include the functional and
biochemical activity of cells in lungs, alveolar macrophages (AMs),
immunological competence, susceptibility to experimentally induced
respiratory infections, and the rate of mucociliary clearance.

Alveolar macrophages are affected by NO₂. These cells
are responsible for maintaining the sterility of the pulmonary
region, clearing particles from this region, and participating in
immunological functions. Functional changes that have been reported
include the following: the suppression of phagocytic ability and
stimulation of lung clearance at 560 µg/m³ (0.3 ppm) 2 h/day for
13 days; a decrease in bactericidal activity at 4320 µg/m³ (2.3 ppm)
for 17 h; and a decreased response to migration inhibition factor at
3760 µg/m³ (2.0 ppm) 8 h/day, 5 days/week for 6 months. The
morphological appearance of these defence cells changes after chronic
exposure to NO₂.

The importance of host defences becomes evident when animals have
to cope with laboratory-induced pulmonary infections. Animals exposed
to NO₂ succumb to bacterial or viral infection in a concentration-
dependent manner. Mortality also increases with increased NO₂
concentration or duration of exposure. After acute exposure,
effects are observed at concentrations as low as 3760 µg/m³ (2 ppm).
Exposure to concentrations as low as 940 µg/m³ (0.5 ppm) will cause
effects in the infectivity model after 6 months.

Both humoral and cell-mediated defence systems are changed by
NO₂ exposure. In the cases in which the immune system has been
investigated, effects have been observed after short-term exposure to

concentrations > 9400 µg/m³ (5 ppm). The effects are complex
since the direction of the change (i.e., increase or decrease) is
dependent upon NO₂ concentration and the length of exposure.

1.3.1.3 Effects of chronic exposure on the development of chronic
lung disease

Humans are chronically exposed to NO₂. Therefore, such
exposures in animals have been studied rather extensively, typically
using morphological and/or morphometric methods. This research has
generally shown that a variety of pulmonary structural and correlated
functional alterations occur. Some of these changes may be reversible
when exposure ceases.

Pulmonary function may be altered following chronic NO₂ exposure
of experimental animals. Impaired gas exchange occurred following
exposure to 7520 µg/m³ (4.0 ppm) NO₂ for four months and this was
reflected in decreased arterial O₂ tension, impaired physical
performance and increased anaerobic metabolism.

Although NO₂ produces morphological changes in the respiratory
tract, the database is sometimes confusing due to quantitative and
qualitative variability in responsiveness between, and even within,
species. The rat, the most commonly used experimental animal in
morphological assessments of exposure, appears to be relatively
resistant to NO₂. Short-term exposures to concentrations of
9400 µg/m³ (5.0 ppm) or less generally have little effect in the
rat, where similar exposures in the guinea-pig may result in some
centriacinar epithelial damage.

Longer-term exposures result in lesions in some species with
concentrations as low as 560 to 940 µg/m³ (0.3 to 0.5 ppm). These
are characterized by epithelial remodelling similar to that described
above, but with the involvement of more proximal airways and
thickening of the interstitium. Many of these changes, however, will
resolve even with continued exposure, and long-term exposures to
levels above about 3760 µg/m³ (2.0 ppm) are required for more
extensive and permanent changes in the lungs. Some effects are
relatively persistent (e.g., bronchiolitis), whereas others tend to be
reversible and limited even with continued exposure. In any case, it
seems that for either short- or long-term exposure, the response is
more dependent upon concentration than duration of exposure.

There is substantial evidence that long-term exposure of several
species of laboratory animals to high concentrations of NO₂ results
in morphological lung lesions. Destruction of alveolar walls, an
essential additional criterion for human emphysema, has been reliably
reported in lungs from animals in a limited number of studies. The

lowest NO₂ concentration for the shortest exposure duration that will
result in emphysematous lung lesions cannot be determined from these
published studies.

1.3.1.4 Potential carcinogenic or co-carcinogenic effects

NO₂ has been shown to be mutagenic in Salmonella bacteria, but
was not mutagenic in one study with a mammalian cell culture. Other
studies using cell cultures have demonstrated sister chromatid
exchanges (SCE) and DNA single strand breaks. No genotoxic effects
have been demonstrated in vivo concerning lymphocytes, spermatocytes
or bone marrow cells, but two inhalation studies with high
concentrations (50 760 and 56 400 µg/m³, 27 and 30 ppm) for 3 h and
16 h, respectively, have demonstrated such effects in lung cells.

Literature searches revealed no published reports of NO₂ studies
using classical whole-animal chronic bioassays for carcinogenesis.
Research with mice having spontaneously high tumour rates was
equivocal. In one study, NO₂ at 18 800 µg/m³ (10 ppm) slightly
enhanced the incidence of lung adenomas in a sensitive strain of mice
(A/J). Although several co-carcinogenesis investigations have been
undertaken, conclusions are precluded because of problems with
methodology and interpretation. Reports on whether NO₂ facilitates
the metastasis of tumours to the lung are also inadequate to form
conclusions. Other investigations have centred on whether NO₂ could
produce nitrates and nitrites that, by reacting with amines in the
body, could produce nitrosamines. A few studies suggest that
nitrosamines are formed in animals treated with high doses of amines
and exposed to NO₂, but other studies have indicated that nitrosamine
formation is unlikely.

1.3.1.5 Age susceptibility

Investigations into age dependency are inadequate and results so
far are equivocal.

1.3.1.6 Influence of exposure patterns

Several animal toxicological studies have elucidated the
relationships between concentration (C) and duration (T) of exposure,
indicating that the relationship is complex. Most of this research
has used the infectivity model. Early C × T studies demonstrated that
concentration had more impact on mortality than did duration of
exposure. An evaluation of the toxicity of NO₂ exposures cannot be
delineated by C × T relationships.

1.3.2 Controlled human exposure studies on nitrogen oxides

Human responses to a variety of oxidized nitrogen compounds have
been evaluated. By far, the largest database and the one most

suitable for risk assessment is that available for controlled
exposures to NO₂. The database on human responses to NO, HNO₃
vapour, HNO₂ vapour and inorganic nitrate aerosols is not as
extensive. A number of sensitive or potentially sensitive subgroups
have been examined, including adolescent and adult asthmatics, older
adults, and patients with chronic obstructive pulmonary disease (COPD)
and pulmonary hypertension. Exercise during exposure increases the
total uptake and alters the distribution of the deposited inhaled
material within the lung. The relative proportion of NO₂ deposited in
the lower respiratory tract is also increased by exercise. This may
increase the effects of the above compounds in people who exercise
during exposure.

As is typical with human biological response to inhaled particles
and gases, there is variability in the biological response to NO₂.
Healthy individuals tend to be less responsive to the effects of NO₂
than individuals with lung disease. Asthmatics are clearly the most
responsive group to NO₂ that has been studied to date. Individuals
with COPD may be more responsive than healthy individuals, but they
have limited capacity to respond to NO₂ and thus quantitative
differences between COPD patients and others are difficult to assess.
Sufficient information is not available at present to evaluate whether
age and sex play a role in the response to NO₂.

Healthy subjects can detect the odour of NO₂, in some cases at
concentrations below 188 µg/m³ (0.1 ppm). Generally, NO₂ exposure
did not increase respiratory symptoms in any of the subject groups
tested.

NO₂ causes decrements in lung function, particularly increased
airway resistance in resting healthy subjects at 2-h concentrations as
low as 4700 µg/m³ (approx.2.5 ppm). Available data are insufficient
to determine the nature of the concentration-response relationship.

Exposure to NO₂ results in increased airway responsiveness to
bronchoconstrictive agents in exercising healthy, non-smoking subjects
exposed to concentrations as low as 2800 µg/m³ (approx.1.5 ppm) for
1 h or longer.

Exposure of asthmatics to NO₂ causes, in some subjects,
increased airway responsiveness to a variety of provocative mediators,
including cholinergic and histaminergic chemicals, SO₂ and cold air.
The presence of these responses appears to be influenced by the
exposure protocol, particularly whether or not the exposure includes
exercise. These responses may begin at concentrations as low as
380 µg/m³ (0.2 ppm). A meta-analysis suggests that effects may occur
at even lower concentrations. However, an unambiguous concentration-
response relationship is observed between 350 to 1150 µg/m³
(approx.0.2 to 0.6 ppm).

The implications of this overall trend are unclear, but increased
airway responsiveness could potentially lead to increased response to
aeroallergens or temporary exacerbation of asthma, possibly leading to
increased medication usage or even increased hospital admissions.

Modest increases in airway resistance may occur in COPD patients
from brief exposure (15-60 min) to concentrations of NO₂ as low as
2800 µg/m³ (approx.1.5 ppm), and decrements in spirometric measures
of lung function (3 to 8% change in FEV₁ (forced expiratory volume
in 1 second)) may also be observed with longer exposures (3 h) to
concentrations as low as 600 µg/m³ (approx.0.3 ppm).

Exposure to NO₂ at levels above 2800 µg/m³ (approx.1.5 ppm) may
alter the numbers and types of inflammatory cells in the distal
airways or alveoli. NO₂ may alter the functioning of cells within
the lungs and production of mediators that may be important in lung
host defences. The constellation of changes in host defences,
alterations in lung cells and their activities, and changes in
biochemical mediators is consistent with the epidemiological findings
of increased host susceptibility associated with NO₂ exposure.

In studies on mixtures of NO₂ with other pollutants, NO₂ has
not been observed to increase responses to other co-occurring
pollutant(s) beyond that which would be observed for the other
pollutant(s) alone. A notable exception is the observation that
pre-exposure to NO₂ enhanced the ozone-induced change in airway
responsiveness in healthy exercising subjects during a subsequent
ozone exposure. This observation suggests the possibility of delayed
or persistent responses to NO₂.

Within an NO₂ concentration range that may be of interest with
regard to risk evaluation (i.e., 100-600 µg/m³), the characteristics
of the concentration-response relationship for acute changes in lung
function, airway responsiveness to bronchoconstricting agents or
symptoms cannot be determined from the available data.

On the basis of an effect at 400 µg/m³ and the possibility of
effects at lower levels, based on a meta analysis, a one-hour average
daily maximum NO₂ concentration of 200 µg/m³ (approx.0.11 ppm) is
recommended as a short-term guideline.

NO is acknowledged as an important endogenous second messenger
within several organ systems. Inhaled NO concentrations above
6000 µg/m³ (approx.5 ppm) can cause vasodilation in the pulmonary
circulation without affecting the systemic circulation. The lowest
effective concentration has not been established. Information on
pulmonary function and lung host defences consequent to NO exposure
are too limited for any conclusions to be drawn at this time.

Relatively high concentrations (> 40 000 µg/m³) have been used in
clinical applications for brief periods (< 1 h) without reported
adverse reactions.

Nitric acid levels in the range of 250-500 µg/m³ (97-194 ppb)
may cause some pulmonary function responses in adolescent asthmatics,
but not in healthy adults.

Limited information on HNO₂ suggests that it may cause eye
inflammation at 760 µg/m³ (0.40 ppm). There are currently no
published data on human pulmonary responses to HNO₂.

Limited data on inorganic nitrates suggest that there are no lung
function effects of nitrate aerosols at concentrations of 7000 µg/m³
or less.

1.3.3 Epidemiology studies on nitrogen dioxide

Epidemiological studies on the health effects of nitrogen oxides
have mainly focused on NO₂. Many indoor and outdoor epidemiological
studies designed to evaluate the health effects of NO₂ have been
conducted. Two health outcome measurements of NO₂ exposure are
generally considered: lung function measurements and respiratory
symptoms and diseases.

The evidence from individual studies of the effect of NO₂ on
lower respiratory symptoms and disease in school-aged children is
somewhat mixed. The consistency of these studies was examined and
the evidence synthesized in a combined quantitative analysis
(meta-analysis) of the subject studies. Most of the indoor studies
showed increased lower respiratory morbidity in children associated
with long-term exposure to NO₂. Mean weekly NO₂ concentrations
in bedrooms in studies reporting NO₂ levels were predominantly
between 15 and 122 µg/m³ (0.008 and 0.065 ppm). Combining the
indoor studies as if the end-points were similar gives an estimated
odds ratio of 1.2 (95% confidence limits of 1.1 and 1.3) for the effect
per 28.3 µg/m³ (0.015 ppm) increase of NO₂ on lower respiratory
morbidity. This suggests that, subject to assumptions made for the
combined analysis, an increase of about 20% in the odds of lower
respiratory symptoms and disease corresponds to each increase of
28.3 µg/m³ (0.015 ppm) in estimated 2-week average NO₂
exposure. Thus, the combined evidence is supportive for the effects
of estimated exposure to NO₂ on lower respiratory symptoms and
disease in children aged 5 to 12 years.

In individual indoor studies of infants 2 years of age or younger,
no consistent relationship was found between estimates of NO₂
exposure and the prevalence of respiratory symptoms and disease. Based

on a meta-analysis of these indoor infant studies, subject to the
assumptions made for the meta-analysis, the combined odds ratio for the
increase in respiratory disease per increase of 28.2 µg/m³ (0.015 ppm)
NO₂ was 1.09 with a 95% confidence interval of 0.95 to 1.26, where
mean weekly NO₂ concentrations in bedrooms were predominantly between
9.4 and 94 µg/m³ (0.005 and 0.050 ppm) in studies reporting levels.
The increase in risk was very small and was not reported consistently
by all studies. We cannot conclude that the evidence suggests an effect
in infants comparable to that seen in older children. The reasons for
these age-related differences are not clear.

The measured NO₂ studies gave a higher estimated odds ratio than
the surrogate estimates, which is consistent with a measurement error
effect. The effect of having adjusted for covariates such as
socioeconomic status, smoking and sex was that those studies that
adjusted for a particular covariate found larger odds ratios than
those that did not.

Although many of the epidemiological studies that involved
measured NO₂ levels used measurements over only 1 or 2 weeks, these
levels were used to characterize children's exposures over a much
longer period. The standard respiratory symptom questionnaire used by
most of these studies summarizes information on health status over an
entire year. The 28.2 µg/m³ (0.015 ppm) difference in NO₂ levels
used in the meta-analyses relates to a difference in the household
annual average exposure between gas and electric cooking stoves.
Some studies measured NO₂ levels only in the winter and may have
overestimated annual average exposures. This would tend to have
underestimated the health effect of a 28.2 µg/m³ (0.015 ppm)
difference in the annual NO₂ exposure. A study based on a household
annual average exposure measured in both the winter and summer found a
stronger health effect than many of the other studies. The true
biologically relevant exposure period is unknown, but these exposures
extended over a lengthy period up to the entire lifetime of the child.

The association between outdoor NO₂ and respiratory health is
not clear from current research. There is some evidence that the
duration of respiratory illness may be increased at higher ambient
NO₂ levels. A major difficulty in the analysis of outdoor studies is
distinguishing possible effects of NO₂ from those of other associated
pollutants.

Several uncertainties need to be considered in interpreting the
above studies and meta-analysis. Error in measuring exposure is
potentially one of the most important methodological problems in
epidemiological studies of NO₂. Although there is evidence that
symptoms are associated with indicators of NO₂ exposure, the quality
of these exposure estimates may be inadequate to determine a
quantitative relationship between exposure and symptoms. Most of the
studies that measured NO₂ exposure did so only for periods of 1 to

2 weeks and reported the values as averages. Few of the studies
attempted to relate the observed effects to the pattern of exposure
(e.g., transient NO₂ peaks). Furthermore, measured NO₂ concentration
may not be the biologically relevant dose; estimating actual exposure
requires knowledge of pollutant species, levels and related human
activity patterns. However, only very limited activity and aerometric
data are available that examine such factors. The extrapolation to
possible patterns of ambient exposure is difficult. In addition,
although the level of similarity and common elements between the
outcome measures in the NO₂ studies provide some confidence in their
use in the quantitative analysis, the symptoms and illnesses combined
are to some extent different and could indeed reflect different
underlying processes. Thus, caution is necessary in interpreting the
meta-analysis results.

Other epidemiological studies have attempted to relate some
measure of indoor and/or outdoor NO₂ exposure to changes in pulmonary
function. These changes were marginally significant. Most studies
did not find any effects, which is consistent with controlled human
exposure study data. However, there is insufficient epidemiological
evidence to draw any conclusions about the long- or short-term effects
of NO₂ on pulmonary function.

On the basis of a background level of 15 µg/m³ (0.008 ppm) and
the fact that significant adverse health effects occur with an
additional level of 28.2 µg/m³ (0.015 ppm) or more, an annual
guideline value of 40 µg/m³ (0.023 ppm) is proposed. This value will
avoid the most severe exposures. The fact that a no-effect level for
subchronic or chronic NO₂ exposure concentrations has not yet been
determined should be emphasized.

1.3.4 Health-based guidance values for nitrogen dioxide

On the basis of human controlled exposure studies, the
recommended short-term guidance value is for a one-hour average NO₂
daily maximum concentration of 200 µg/m³ (0.11 ppm). The recommended
long-term guidance value, based on epidemiological studies of
increased risk of respiratory illness in children, is 40 µg/m³
(0.023 ppm) annual average.

2. PHYSICAL AND CHEMICAL PROPERTIES, AIR SAMPLING AND ANALYSIS,
TRANSFORMATIONS AND TRANSPORT IN THE ATMOSPHERE

2.1 Introduction

Nitrogen oxides are produced by combustion processes and are
emitted to the air mainly as NO together with some NO₂. Natural
biological processes and lightning also emit NO and N₂O. In the
atmosphere nitrogen oxides undergo complex chemical and photochemical
reactions; NO is oxidized to NO₂ and other products and eventually to
HNO₃ and nitrates. Nitrogenous species are removed from the air to
the ground by wet and dry deposition processes. Oxidized nitrogen
compounds can have impacts on human health and the environment, and
are important to the formation of photochemical smog and tropospheric
ozone.

In this chapter the properties of nitrogen compounds are briefly
described and techniques for their sampling and analysis outlined.
Atmospheric chemical reactions that cause the oxidation of NO to NO₂
and the production of ozone, organic nitrates and HNO₃ are described.
The differences between night-time and day-time chemistry and the
composition of the atmosphere are discussed. The nature of the
nitrogen species and their chemical reactions in urban regions, in
chimney plumes such as those from power stations, in air advected away
from urban regions and in rural and remote areas are described. The
role of nitrogen oxides in photochemical smog production and the
effects of nitrous oxide on stratospheric ozone are briefly discussed.

2.1.1 The nomenclature and measurement of atmospheric nitrogen
species

There are several methods available for determining nitrogen
species, but many of these techniques are nonspecific.

To denote various mixtures of nitrogen species, the terms NO_x,
NO_y and NO_z are often employed. It is customary to refer to the sum
of NO and NO₂ emitted from a source as NO_x, the unit of measure for
NO_x being the NO₂ mass equivalent of the NO plus NO₂.

The term NO_y is frequently used to denote the sum of the gas
phase oxidized nitrogen species (except N₂O) and NO_z to denote the
sum of NO_y plus the oxidized nitrogen present as particulate matter.
Measurement of NO_z requires a combination of particulate and gas
phase sampling and analysis.

A confusion arises because one of the most commonly used methods
for determining NO₂ in ambient air (thermal conversion of NO₂ to NO
and measurement of the resultant NO by chemiluminescent reaction with
O₃) is nonspecific and responds to several gaseous species in
addition to NO₂. These include organic nitrogen compounds and,

depending on the converter, HNO₃, although HNO₃ can be readily lost
to the sampling system. Therefore, depending on the composition of
the air being sampled, the results from this type of instrument can be
representative of NO_y rather than NO_x (or NO₂) concentrations.
This technique is used in most routine determinations of ambient NO_x
and NO₂ concentrations but the discrepancy between these values and
true NO_x and NO₂ can be considerable for air in which the pollutant
emissions have undergone substantial exposure to sunlight.

Nitrous oxide is ubiquitous in the atmosphere because it is a
product of biological processes in soil as well as anthropogenic
activities. It is not involved to any appreciable extent in chemical
reactions in the lower atmosphere, but it is an active "greenhouse"
gas. In the stratosphere N₂O forms NO by reaction with excited
oxygen atoms, and this NO then acts to deplete the stratospheric O₃
concentration.

Although NO₃, dinitrogen trioxide (N₂O₃), dinitrogen tetroxide
(N₂O₄), and N₂O₅ may play a role in atmospheric chemical reactions
leading to the transformation, transport, and ultimate removal of
nitrogen compounds from ambient air, they are present in very low
concentrations, even in polluted environments.

NH₃ is generated during decomposition of nitrogenous matter in
natural ecosystems and may be locally produced in high concentrations
by human activities such as intensive animal husbandry and feedlots.
Under suitable conditions NH₃ can react with oxidized nitrogen
species to form ammonium nitrate aerosol.

2.2 Nitrogen species and their physical and chemical properties

There are seven oxides of nitrogen that may be present in ambient
air, namely: NO, NO₂, N₂O, NO₃, N₂O₃, N₂O₄ and N₂O₅. In
addition these can be present as HNO₂, HNO₃ and various organic
nitrogen species, such as PAN, other organic nitrates and particles
containing oxidized nitrogen compounds (particularly adsorbed nitric
acid). Of these species, NO and NO₂ are the ones most often measured
and are present in the greatest concentrations in urban and industrial
air.

The chemical and physical properties of individual nitrogen
species are given below and are summarized in Table 1.

Table 1. Some physical and thermodynamic properties of oxides of nitrogen and other nitrogen compounds^a

Oxide Relative Melting point Boiling point Solubility in water Thermodynamic functions
molecular (°C)^b,c,d (°C)^b,c at 0°C (cm³ per 100 g)^b (Ideal gas, 1 atm, 25°C)
mass (g/mol)
Enthalpy of Entropy
formation (cal/mol-deg)
(kcal/mol)

NO 30.01 -163.6 -151.8 7.34 21.58 50.35

NO₂ 46.01 -11.2 21.2 Reacts with H₂O forming 7.91 57.34
HNO₂ and HNO₃

N₂O 44.01 -90.8 -88.5 130.52 19.61 52.55

N₂O₃ 76.01 -102 47 Reacts with H₂O forming 19.80 73.91
(decomposes) HNO₂

N₂O₄ 92.02 -11.3 21.2 Reacts with H₂O forming 2.17 72.72
HNO₂ and HNO₃

N₂O₅ 108.01 30 3.24 Reacts with H₂O forming 2.7 82.8
(decomposes) HNO₂

HNO₂ 47.01 - - - - -

HNO₃ 63.01 -42 83 -32.1 63.7

Table 1. (Con't)

PAN 121.06 - - - - -
(CH₃COOONO₂)

NH₄NO₃ 80.04 169.6 210 at 118.3 g/100 cm³ -87.37 36.11
11 torr H₂O at 0°C

^a Adopted from: US EPA (1993)
^b Matheson Gas Data Book (Matheson Company, 1966)
^c Handbook of Chemistry and Physics (Weast et al., 1986)
^d At 0°C and 1 atm pressure
2.2.1 Nitrogen oxides

2.2.1.1 Nitric oxide

NO is a colourless, odourless gas that is only slightly soluble
in water. It is a by-product of combustion processes, arising from
(i) high temperature oxidation of molecular nitrogen from the
combustion air, and (ii) from oxidation of nitrogen present in certain
fuels such as coal and heavy oil.

2.2.1.2 Nitrogen dioxide

NO₂ is a reddish-orange-brown gas with a characteristic pungent
odour. The boiling point is 21.1°C, but the low partial pressure of
NO₂ in the atmosphere prevents condensation. NO₂ is corrosive and
highly oxidizing. About 5 to 10% by volume of the total emissions of
NO_x from combustion sources is usually in the form of NO₂, although
substantial variations from one source type to another have been
observed.

In the atmosphere, photochemical reactions involving ozone
and organic compounds convert NO to NO₂. NO₂ is an efficient
absorber of light over a broad range of ultraviolet (UV) and visible
wavelengths. Because of its brown colour, NO₂ can contribute to
discoloration and reduced visibility of polluted air. Photolysis of
NO₂ by sunlight produces NO and an oxygen atom, which usually adds to
an oxygen molecule to produce ozone.

2.2.1.3 Nitrous oxide

N₂O is a colourless gas with a slight odour at high
concentrations. It is emitted to the atmosphere as a trace component
from some combustion sources and from the consumption of nitrate by
an ubiquitous group of denitrification bacteria that use nitrate as
their terminal electron acceptor in the absence of oxygen (Delwiche,
1970; Brezonik, 1972; Keeney, 1973; Focht & Verstraete, 1977). At
atmospheric concentrations N₂O has no significant physiological
effects in humans, although at higher concentrations it is employed as
an anaesthetic.

N₂O does not play a significant role in atmospheric reactions in
the lower troposphere. In the stratosphere it reacts with singlet
oxygen to produce NO, which participates in O₃ decomposition in
the stratosphere. These reactions are of concern because of the
possibility that increasing N₂O concentrations resulting from fossil
fuel use, and also from denitrification of excess fertilizer, may
contribute to a decrease in stratospheric O₃ (Council for
Agricultural Science and Technology, 1976; Crutzen, 1976) with
consequent potential for adverse impacts on ecosystems and human

health. Also of concern is the fact that N₂O absorbs long-wave
radiation, and therefore serves as a radiatively important greenhouse
gas that may contribute to global warming.

2.2.1.4 Other nitrogen oxides

Other nitrogen oxides can be present in trace quantities in the
air. NO₃ has been identified in laboratory systems containing
NO₂/O₃, NO₂/O and N₂O₅ as an important reactive transient
(Johnston, 1966). It is likely to be present in photochemical smog.
In the presence of sunlight, NO₃ is rapidly converted to either NO or
NO₂ (Wayne et al., 1991). Nitrogen trioxide is highly reactive
towards both NO and NO₂. Its expected concentration in polluted air
is very low (about 10^-6 µg/m³). However, traces of NO₃ may play an
important role in atmospheric chemistry, especially at night when it
may serve as a reservoir for NO_x (Wayne et al., 1991). In the
atmosphere N₂O₃ is in equilibrium with NO and NO₂. It reacts with
water to form HNO₂. N₂O₄ is the dimer of NO₂, formed in
equilibrium with NO₂ molecules, and it readily dissociates to NO₂.
N₂O₅ can be a trace night-time component of the air because it is
formed by a reaction between NO₂ and NO₃. Since NO₃ can exist in
appreciable quantities only in the absence of sunlight, N₂O₅ is only
important at night, when its reaction with water can be a significant
source of nitric acid.

2.2.2 Nitrogen acids

2.2.2.1 Nitric acid

HNO₃ is the most oxidized form of nitrogen. In the gaseous
state it is colourless. It is photochemically stable in the
troposphere. HNO₃ is volatile, so that at typical concentrations and
temperatures in the atmosphere the vapour does not coalesce into
aerosol and is not retained on particles unless the aerosol contains
reactants such as sodium chloride or ammonium salts to react with the
acid, when it produces particulate nitrates (Wolff, 1984).

In the aqueous phase (e.g., rain drops), HNO₃ dissociates to
form the nitrate ion (NO₃^-). Because nitrate is chemically
unreactive in dilute aqueous solution, nearly all of the
transformations involving nitrate in natural waters result from
biochemical pathways. The nitrate salts of all common metals are
quite soluble.

2.2.2.2 Nitrous acid

HNO₂ is formed when NO and NO₂ are present in the atmosphere,
as a result of their reaction with water. In sunlight, the dominant

pathway for HNO₂ formation is the reaction of NO with hydroxyl
radicals. During the daytime, atmospheric concentrations of HNO₂ are
limited by the photolysis of HNO₂ to produce NO and hydroxyl radical.

Nitrous acid is a weak reducing agent and is oxidized to nitrate
only by strong chemical oxidants and by nitrifying bacteria.

2.2.3 Ammonia

NH₃ is the completely reduced form of nitrogen. It is a
colourless gas with a pungent odour. It is extremely soluble in
water, forming ammonium (NH_y⁺) and hydroxyl (OH^-) ions. In the
atmosphere, NH₃ has been reported to be converted into NO_x by
reaction with hydroxyl radicals (Soederlund & Svensson, 1976). In the
stratosphere, NH₃ can be dissociated by irradiation with sunlight at
wavelengths below 230 nm (McConnell, 1973).

2.2.4 Ammonium nitrate

Gas-phase ammonia reacts with nitric acid to form ammonium
nitrate (NH₄NO₃). Ammonium nitrate is a solid at room temperature.
Like ammonia, it is very soluble in water and hence will be absorbed
by any water droplets present. Thus it readily forms an aerosol in
the atmosphere. Pathways to aerosol formation include nucleation and
condensation on existing particles. The presence of NH₄NO₃
particles can result in a visible haze.

2.2.5 Peroxyacetyl nitrate

Of the various peroxy nitrates found in ambient air, peroxyacetyl
nitrate (CH₃COOONO₂), or PAN, is found at the highest concentrations.
PAN undergoes a temperature-dependent decomposition to its precursors,
NO₂ and acetyl peroxy radicals. At low ambient temperatures PAN
can have a substantial lifetime in the atmosphere (Cox & Roffey, 1977).
In polluted air PAN concentrations can reach several parts per billion.

2.2.6 Organic nitrites and nitrates

A wide variety of organic nitrites (RNO₂) and nitrates (RNO₃),
where R denotes CH₃, CH₂CH₃, benzyl, etc., may be found in ambient
air. Some of these are emitted directly while others are formed by
photochemical reactions in the atmosphere.

2.3 Sampling and analysis methods

This section outlines methods for measuring nitrogen-containing
species in the atmosphere. The main focus is on methodologies
currently available and in general use for monitoring concentrations
in both ambient and indoor air.

Table 2 summarizes sampling and analytical methods for selected
species and addresses relevant characteristics, including the type of
method (i.e., in situ, remote, active, passive, continuous or
integrative), the stage of development of the method, sampling
duration, precision, accuracy and detection limits.

2.3.1 Nitric oxide

2.3.1.1 Nitric oxide continuous methods

Nitric oxide reacts rapidly with O₃ to give NO₂ in an excited
electronic stage. The transition of excited NO to the grand state can
be accompanied by the emission of light in the red-infrared spectral
range. When this chemiluminescent reaction occurs under controlled
conditions, the intensity of the emitted light is proportional to the
concentration of the NO reactant. This provides the basis of the
chemiluminescence method (CLM) for analysis of NO. This method is a
continuous technique and is the most commonly used method for
measuring NO in ambient air. Commercial instruments for measuring NO
and NO₂ are available with detection limits of approximately 5 ppb
and response times of the order of minutes. CLM measurement of NO₂
can also be accomplished by firstly converting the NO₂ of the sample
to NO. This is discussed in section 2.3.2.1.

Other NO analytical methods include laser-induced fluorescence
(LIF) (Bradshaw et al., 1985), absorption spectroscopy (e.g., tuneable
diode laser absorption spectroscopy, TDLAS) and passive samplers.

2.3.1.2 Passive samplers for NO

Passive samplers are used for air with higher-than-typical
ambient concentrations, which may be found indoors or in the
workplace. They are often used to obtain data at a large number of
sites. Sampling typically lasts a few hours.

The Palmes tube is a passive sampler that relies on diffusion of
an analyte molecule through a quiescent diffusion path of known length
and cross-sectional area to a reactive surface where the molecule is
captured by chemical reaction (Palmes et al., 1976). The Palmes tube
does not measure NO directly. Two tubes are required; the first one
has reactive grids coated with triethanolamine (TEA) to collect NO₂,
the second tube is similar but has an additional reactive surface
coated with chromic acid to convert NO to NO₂, which is in turn
collected by the TEA-coated grids. The NO concentration of the air is
determined from the difference in the results from the two tubes. The
data is corrected for the effects of the different diffusivities of NO
and NO₂ molecules. To ensure reliable results, contact between the
chromic-acid-coated surface and the TEA-coated grids for longer than
24 h must be avoided. Analysis of the material contained in the TEA

Table 2. Selected instruments and methods for determining oxides of nitrogen in ambient air (from: Sickles, 1992)

Species Methods^a Type^b Development Sample Performance Comments References
stage^c duration
Precision Accuracy MDL^d

NO CLM I, A, C C 5 min < 10% < 20% < 9 ppb - Finlayson-Pitts &
(NO + O₃) Pitts (1986)

TP-LIF I, A, C R 30 sec - 16% 10 ppt - Bradshaw et al. (1985);
Davis et al. (1987)

TDLAS I, A, C R, C 60 sec - - 0.5 ppb 40-m path length NASA (1983)

PSD I, P, IN C 24 h - - 70 ppb-h^e

NO₂ CLM I, A, C C 5 min 10% 20% 9 ppb Commonly used Finlayson-Pitts &
(NO + O₃) method; many Pitts (1986)
interferences

CLM I, A, C R < 100 sec 20 ppt 30% 10-25 ppt Uses thermal or Helas et al. (1987);
(NO + O₃) photolytic Fehsenfeld et al.
converters (1987)

CLM I, A, C C 100 sec 0.6 ppb - 10 ppt Interferences:
(Luminol) PAN, HNO₂, O₃

TP-LIF I, A, C R 2 min 20 ppt 16% 12 ppt - Davis (1988)

TDLAS I, A, C R, C 60 sec - 15% 100 ppt 150-m path length NASA (1983)

DOAS R, A, C R, C 12 min - 10% 4 ppb 800-m path length Platt & Perner (1983)

Bubbler I, A, IN RM 24 h 6 ppb 10% 8 ppb^e Purdue & Hauser (1980)

Table 2. (Con't)

Species Methods^a Type^b Development Sample Performance Comments References
stage^c duration
Precision Accuracy MDL^d

TEA I, A, IN L 24 h 15% 10% 0.2 ppb^e Interferences: Sickles et al. (1990)
filter PAN and HNO_2^f

Guaiacol I, A, IN L 1 h 4% - 0.1 ppb^e Stability of Buttini et al. (1987)
Denuder extract uncertain

DPA I, A, IN L 8 h 8% - 0.1 ppb^e DPA may volatilize; Lipari (1984)
Cartridge interferences:
HNO₂ and PAN

TEA PSD I, P, IN L 24 h 30% - 30 ppb-h^e Similar to Palmes
Tube; interferences
as above^f

NO_y CLM I, A, C R 10 sec - 15% 10 ppt CO with Au Fahey et al. (1986)
(NO + O₃) reducing catalyst

PAN GC-ECD I, A, IN R, RM 15 min - 30% 10 ppt^e Sensitivity can be Vierkorn-Rudolph
enhanced by using et al. (1985)
cryogenic sampling
and capillary
columns

GC-CLM I, A, IN L - - - - CLM (NO + O₃) and
(Luminol) reported

Other organic GC-ECD/MS I, A, C R 24 h - - 1 ppt^e Sample collected Atlas (1988)
Nitrates on charcoal

Table 2. (Con't)

Species Methods^a Type^b Development Sample Performance Comments References
stage^c duration
Precision Accuracy MDL^d

NHO₃ Filter I, A, IN R, RM 24 h 10% 20% 8 ppt^e May be nylon or Finlayson-Pitts &
calcium chloride Pitts (1986)
impregnated filter;
subject to
artifacts^f

Denuder I, A, IN R, RM 24 h 8% - 8 ppt^e Not subject to Sickles (1987);
above artifacts^f Sickles et al. (1989)

TDLAS I, A, C R, C 5 min - 20% 100 ppt 150-m path length NASA (1983)

HNO₂ Denuder I, A, IN R, RM 24 h 15% - 10 ppt^e Annular denuder Sickles et al. (1989);

preferred^f Vossler et al. (1988)

LIF I, A, C R 15 min - - 20 ppt OH detected
following photo-
fragmentation

DOAS R, A, C R, C 12 min - 30% 600 ppt 800-m path length Biermann et al. (1988)

Table 2. (Con't)

Species Methods^a Type^b Development Sample Performance Comments References
stage^c duration
Precision Accuracy MDL^d

NO₃ DOAS R, A, C R, C 12 min - 15% 20 ppt 800-m path length Platt & Perner (1983)

Particulate Denuder/ I, A, IN R, RM 24 h 10% - 40 ng/m^3e Use of denuders Vossler et al. (1988)
NO₃ Filter(s) avoids artifacts;
denuders collect
HNO₃ and NH₃;
teflon and nylon
filters used

N₂O GC-ECD I, A, IN R, RM 15 min 3% - 20 ppb^e -

^a CLM (NO + O₃) = Chemiluminescent using NO + O₃ reaction ^b I = In situ
TP-LIF = Two-photon laser-induced A = Active
TDLAS = Tuneable diode laser absorption spectroscopy C = Continuous
TTFMS = Two-tone frequency modulated spectroscopy P = Passive
PSD = Passive sampling device IN = Integrative
CLM (Luminol) = Chemiluminescent using reaction with Luminol R = Remote
DOAS = Differential optical absorption spectroscopy
DIAL = Differential absorption lidar ^c C = Commercially available
TEA = Triethanolamine R = Research tool
DPA = Diphenylamine L = Laboratory prototype
GC-ECD = Gas chromatography with electron capture detector RM = Routine method
CG-CLM = Gas chromatography with CLM detector
LIF = Laser-induced fluorescence ^d MDL = Minimum detection limit
GC-MS = gas chromatography with mass spectrometer ^e Depends on the sampled air volume (i.e., flow rate and sampling
duration)
^f Uses ion chromatographic or colorimetric analytical finish
is accomplished by extracting the grids into solution and analysing
the extract for NO₂^- by the use of the spectrophotometric or ion
chromatographic method (Miller, 1984). The colorimetric analysis is
calibrated by dilution of gravimetrically prepared nitrite solutions.
The Palmes Tube method was proposed for sampling occupational
exposures where the dosage does not exceed 25 ppm for 8 h (i.e.,
200 ppm-h). The reliability of this method for measuring NO in the
field at the parts-per-billion or parts-per-million level remains to
be demonstrated.

A badge-type sampler similar to the Palmes tube has been devised
by Yanagisawa & Nishimura (1982). This device uses a series of
12 layers of chromium-trioxide-impregnated glass fibre to oxidize NO
to NO₂. This technique is claimed to be more sensitive by
approximately a factor of 10 than the Palmes tube and to have a lower
limit dosage of 0.07 ppm-h.

2.3.1.3 Calibration of NO analysis methods

Calibration of CLM, TP-LIF and TDLAS measurement systems for NO
all rely on compressed gas mixtures of known concentration being
available. Typically compressed gas mixtures are supplied in
passivated aluminium/stainless steel gas bottles certified by the
manufacturer and with NO diluted with N₂ concentration in the rage of
1 to 50 ppm (Schiff et al., 1983; Carroll et al., 1985; Bradshaw et
al., 1985). Calibrations are performed by dynamic dilution of the
reference NO/N₂ mixture with air to give NO concentrations within the
range of 0.1 to 5 ppm.

For passive NO samplers, only the analysis portion of the
procedure is routinely calibrated (using gravimetrically prepared
nitrite solution).

2.3.1.4 Sampling considerations for NO

Oxides of nitrogen are reactive species and exhibit various
solubilities (Table 1). The most inert materials (i.e. glass and
Teflon^TM) are recommended for use in sampling trains. Since ambient
air contains water vapour that may be sorbed on sampling lines,
surface effects may influence the integrity of air samples containing
the more reactive and more soluble NO_y species. In hot, humid
conditions condensation in the sample lines of liquid water from the
air can cause difficulties when analysis equipment is installed in an
air-conditioned environment. To minimize contamination of the system
by dust and foreign matter, it is common practice to sample through an
inert (teflon) sample inlet filter. Of the NO_y species, NO is
probably the least susceptible to surface effects, whereas surface
effects are very important in the sampling of HNO₃.

Nitric oxide reacts rapidly with O₃ to form NO₂. In the
presence of sunlight NO₂ in air photolyses to yield NO and O₃. Thus
in daylight NO, O₃ and NO₂ can exist simultaneously in ambient air
in a condition known as a "photostationary state". The relative
amounts of the three species at any time are influenced by the
intensity of the sunlight present at that moment. Photolysis ceases
when a sample is drawn into a dark sampling line, but NO and O₃ can
continue to react to form NO₂. Therefore residence times in sampling
lines must be minimized to maintain the intensity of the NO/NO₂ ratio
of the sample.

2.3.2 Nitrogen dioxide

Airborne concentrations of NO₂ can be determined by several
methods including CLM, LIF, absorption spectroscopy, including
differential optical absorption spectroscopy (DOAS) and TDLAS, bubbler
and passive collection with subsequent wet chemical analysis. The
most common techniques are chemiluminescence and passive sampling.

2.3.2.1 Chemiluminescence (NO + O₃)

Instruments discussed in this section do not detect NO₂
directly. They sample continuously and rely on the conversion of some
or all of the NO₂ in the air sample to NO, followed by the CLM
reaction of NO and O₃. The NO₂ concentration is calculated from the
difference in the signal given by the sample after passing through the
converter compared to that when the converter is by-passed.

Several methods have been employed to reduce NO₂ to NO (Kelly,
1986). They include catalytic reduction using heated molybdenum or
stainless steel, reaction with carbon monoxide over a gold catalyst
surface, reaction with iron sulfate at room temperature, reaction with
carbon at 200°C, and photolysis of NO₂ to NO by light in the
wavelength range of 320 to 400 nm.

CLM instruments for the determination of NO₂ are readily
available commercially. Field evaluation of nine instruments showed
that the minimum detection limits (MDLs) ranged from 5 to 13 ppb
(Michie et al., 1983; Holland & McElroy, 1986).

Converters may be non-specific for NO₂ and may convert
several other nitrogen-containing compounds to NO, giving rise to
overestimates for NO₂ concentrations. Using commercial instruments,
Winer et al. (1974) found over 90% conversion of PAN, ethyl nitrate
and ethyl nitrite to NO with a molybdenum converter, and similar
responses to PAN and n-propyl nitrate with a carbon converter. With
a stainless steel converter at 650°C, Matthews et al. (1977) reported
100% conversion for NO₂, 86% for NH₃, 82% for CH₃NH₂, 68% for HCN,
1% for N₂O and 0% for N₂. Using a commercial instrument, Joseph &
Spicer (1978) found quantitative conversion of HNO₃ to NO with a

molybdenum converter at 350°C. Similar responses to PAN, methyl
nitrate, n-propyl nitrate, n-butyl nitrate and HNO₃, substantial
response to nitrocresol, and no response to peroxybenzoyl nitrate
(PBzN) were reported with a commercial instrument using a molybdenum
converter at 450°C (Grosjean & Harrison, 1985). These results were
confirmed for PAN and HNO₃ by Rickman & Wright (1986) using
commercial instruments with a molybdenum converter at 375°C and a
carbon converter at 285°C.

Interference from species that do not contain nitrogen have also
been reported. Joshi & Bufalini (1978), using a commercial instrument
with a carbon converter, found significant apparent NO₂ responses
to phosgene, trichloroacetyl chloride, chloroform, chlorine (Cl₂),
hydrogen chloride, and photochemical reaction products of a
perchloroethylene-NO_x mixture. Grosjean & Harrison (1985) reported
substantial responses to photochemical reaction products of Cl₂-NO_x
and Cl₂-methanethiol mixtures and small negative responses to
methanethiol, methyl sulfide, and ethyl sulfide. Sickles & Wright
(1979), using a commercial instrument with a molybdenum converter at
450°C, found small negative responses to 3-methylthiophene,
methanethiol, ethanethiol, ethyl sulfide, ethyl disulfide, methyl
disulfide, hydrogen sulfide, 2,5-dimethylthiophene, methyl sulfide
and methyl ethyl sulfide, and negligible responses to thiophene,
2-methylthiophene, carbonyl sulfide and carbon disulfide.

Methods of sample trapping followed by batch measurement of NO
and NO₂ in the desorbed sample using a chemiluminescence instrument
have been reported. Gallagher et al. (1985) used cryosampling of
stratospheric whole-air samples, and Braman et al. (1986) used
copper(I) iodide coated denuder tubes to sample NO₂ in ambient air.

2.3.2.2 Chemiluminescence (luminol)

A method for the direct chemiluminescence determination of NO₂
was reported by Maeda et al. (1980) and is based on the CLM reaction
of gaseous NO₂ with a surface wetted with an alkaline solution of
luminol (5-amino-2,3-dihydro-1,4-phthalazinedione). The light
emission is strong at wavelengths between 380 and 520 nm. The
intensity of the light can be proportional to the NO₂ concentration
in the sampled air, and the NO₂ concentration can be determined by
calibration of the instrument with air of known NO₂ concentration.

Since the introduction of the luminol method by Maeda et al.
(1980), improvements have been made to develop an instrument
suitable for use in the field (Wendel et al., 1983), and additional
modifications have been made recently to produce a continuous
commercial instrument (Schiff et al., 1986). Detection limits of 5 to
30 ppt and a response time of seconds have been claimed, based on
laboratory tests (Wendel et al., 1983; Schiff et al., 1986). Recent
laboratory evaluation of two instruments has revealed a detection

limit (i.e., twice the standard deviation of the clean air response)
of 5 ppt, and 95% rise and fall times of 110 and 15 seconds (Rickman
et al., 1988). Field tests of the same instruments have shown an
operating precision of ± 0.6 ppb.

2.3.2.3 Laser-induced fluorescence and tuneable diode laser
absorption spectrometry

Two newer techniques that show considerable promise for measuring
NO₂ specifically are photofragmentation/2-photon LIF and TDLAS. The
LIF and TDLAS techniques provide specific spectroscopic methods to
measure NO₂ directly and compare favourably to the sample photolysis-
chemiluminescence technique (Fehsenfeld et al., 1990; Gregory et al,
1990b). For NO₂ concentrations above 0.2 ppb, no interferences were
found for TDLAS (Fehsenfeld et al., 1990).

2.3.2.4 Wet chemical methods

Most wet chemical methods for measuring NO₂ involve the
collection of NO₂ in solution, followed by a colorimetric finish
using an azo dye. Many variations of this method exist, including
both manual and automated versions. These include the Griess-Saltzman
method, the continuous Saltzman method, the alkaline guiacol
method, the sodium arsenite method (manual or continuous), the
triethanolamine-guaiacol-sulfite (TGS) method and the TEA method.
These methods have been reviewed by Purdue & Hauser (1980).

2.3.2.5 Other methods

Several other methods for the determination of NO₂ have been
reported. Atmospheric pressure ionization mass spectrometry has been
investigated for the continuous measurement of NO₂ and SO₂ in
ambient air (Benoit, 1983). Methods employing photothermal detection
of NO₂ have been reported (Poizat & Atkinson, 1982; Higashi et al.,
1983; Adams et al., 1986).

A portable, battery-powered analyser specific to NO₂, which uses
an electrochemical cell as the detector, is commercially available.
By careful selection and design of the cell, levels down to
approximately 0.1 ppm (v/v) can be detected, although with
uncertainties of approximately 20-50%. The detection cell has a
finite life, dependent on the time integral of the NO₂ concentrations
measured. When the cell deteriorates, the instrument typically
develops a gradual drift.

2.3.2.6 Passive samplers

Passive samplers are frequently used in industrial hygiene,
indoor air and personal exposure studies and are less frequently used
for ambient air analysis. Namiesnik et al. (1984) have provided an
overview of passive samplers.

One type of passive NO₂ sampler for ambient application is the
nitration plate. It is essentially an open petri dish containing
TEA-impregnated filter paper. Mulik & Williams (1986) have adapted
the nitration plate concept by adding diffusion barriers in their
design of a passive sampling device (PSD) for NO₂ in ambient and
personal exposure applications. The device employs a TEA-coated
cellulose filter paper, two 200-mesh stainless steel diffusion screens
and two stainless steel perforated plates on each side of the coated
filter to act as diffusion barriers and permit NO₂ collection on both
faces of the filter paper. After sampling, the paper is removed
from the PSD, extracted in water, and analysed for NO_2^- by
ion chromatography. A sensitivity of 0.03 ppm-h and a rate of
2.6 cm³/second were claimed. Comparison of PSD results with
chemiluminescence determinations of NO₂ in laboratory tests at
concentrations between 10 and 250 ppb showed a linear relation and
high correlation (i.e., r = 0.996) (Mulik & Williams, 1987).
Interference from PAN and HNO₂ would be expected (Sickles, 1987).
Results of TDLAS and triplicate daily PSD NO₂ measurements in a
13-day field study showed good agreement between the study average
values but a correlation coefficient for daily results of only 0.47
(Mulik & Williams, 1987; Sickles et al., 1990). The Palmes tube
described in section 2.3.1.2 has been used to sample air in the
workplace and indoor environments to assess personal exposure to NO₂
(Palmes et al., 1976; Wallace & Ott, 1982).

2.3.2.7 Calibration

Calibration methods for NO₂ use permeation tubes or gas-phase
titration (GPT) to generate known concentrations of NO₂.
Calibrations are performed dynamically using dilution with purified
air.

GPT employs the rapid, quantitative gas-phase reaction between
NO, usually supplied as a known concentration from a gas cylinder, and
O₃ supplied from a stable O₃ generator, to produce one NO₂ molecule
for each NO molecule consumed by reaction. When O₃ is added to
excess NO in a titration system, the decrease in NO concentration
(and O₃) is equivalent to the increase in NO₂ produced (US EPA,
1987b).

Use of cylinders of compressed gas containing NO₂ for
calibration purposes (Fehsenfeld et al., 1987; Davis, 1988) is unwise
because of the uncertain stability of the NO₂ concentrations
delivered; this is a consequence of its relatively high boiling point.

2.3.3 Total reactive odd nitrogen

In this monograph, gas-phase total reactive odd nitrogen is
represented by NO_y. Individual components comprising NO_y are gas

phase NO, NO₂, NO₃, N₂O₅, HNO₂, HNO₃, peroxynitric acid
(HO₂NO₂), PAN, and other organic nitrates. NH₃ and N₂O are not
components of NO_y.

Researchers have successfully combined highly sensitive research-
grade CLM NO detectors with catalytic converters that are sufficiently
active to reduce most of the important gas phase NO_y species to NO
for subsequent detection (Helas et al., 1981; Dickerson, 1984; Fahey
et al., 1986; Fehsenfeld et al., 1987).

2.3.4 Peroxyacetyl nitrate

Several methods have been used to measure the concentration of
PAN in ambient air. Roberts (1990) has provided an overview of many
of these methods. A well-developed method is gas chromatography using
electron capture detection (GC-ECD) (Darley et al., 1963; Smith et
al., 1972; Stephens & Price, 1973; Singh & Salas, 1983).

2.3.5 Other organic nitrates

Other organic nitrates (e.g., alkyl nitrates, peroxypropionyl
nitrate and PBzN) can also be present in the atmosphere, but usually
at lower concentrations than PAN (Fahey et al., 1986). In general,
similar methods for sampling, analysis and calibration may be used for
other organic nitrates as are used for PAN (Stephens, 1969). FTIR,
GC-ECD and GC-MS may be used to measure these compounds.

2.3.6 Nitric acid

Several methods are available for the determination of HNO₃
concentrations in the atmosphere. These include filtration (Okita et
al., 1976; Spicer et al., 1978a), denuder tubes (Forrest et al., 1982;
De Santis et al., 1985; Ferm, 1986), CLM (Joseph and Spicer, 1978) and
absorption spectroscopy (Tuazon et al., 1978; Schiff et al., 1983;
Biermann et al., 1988). Many of these techniques carry significant
uncertainties, which have been compared by Hering et al. (1988).

2.3.7 Nitrous acid

Available techniques for the measurement of HNO₂ in ambient
atmospheres employ denuders (Ferm & Sjodin, 1985), annular denuders
(De Santis et al., 1985), CLM (Braman et al., 1986), PF/LIF (Rodgers &
Davis, 1989), absorption spectroscopy (Tuazon et al., 1978; Biermann
et al., 1988) and FTIR (Finlayson-Pitts & Pitts, 1986).

2.3.8 Dinitrogen pentoxide and nitrate radicals

N₂O₅ is readily reduced to NO at temperatures above 200°C and
may be measured nonspecifically as NO₂ with CLM NO₂ analysers
(Bollinger et al., 1983; Fahey et al., 1986).

Ambient concentrations of the NO₃ radical have been measured
using DOAS; concentrations between 1 and 430 ppt have been observed
(Atkinson et al., 1986).

2.3.9 Particulate nitrate

Many methods are available for sampling ambient aerosols,
including impactors, filtration, and filtration coupled with devices
to remove particles larger than a specified size (e.g., elutriators,
impactors and cyclones).

Particulate nitrate samples are generally collected by
filtration, extracted, and analysed directly or indirectly for nitrate
by ion chromatography or colorimetry.

2.3.10 Nitrous oxide

The most commonly used analytical method for N₂O employs GC-ECD.
It has a detection limit of 20 ppb (Thijsse, 1978) and a precision of
± 3% at the background level of 330 ppb (Cicerone et al., 1978).

2.3.11 Summary

Gas-phase CLM instruments have replaced manual (wet) methods
to a large extent in air quality monitoring network applications.
Gas-phase CLM measurement technology permits the determination of NO,
NO₂ and NO_y in the low ppt range. Although CLM NO detectors coupled
with catalytic NO₂ to NO converters are still not specific for NO₂,
they have proved to be useful for measuring NO_y. CLM NO detectors
coupled with photolytic NO₂ to NO converters have shown improved
specificity for NO₂. Most ambient NO₂ monitoring data reported are
from the nonspecific thermal conversing technique.

Passive samplers for NO₂ have been used primarily for workplace
and indoor applications, but hold promise for averaged ambient
measurements as well. GC-ECD is useful in the determination of PAN,
other organic nitrates and N₂O.

2.4 Transport and transformation of nitrogen oxides in the air

2.4.1 Introduction

Oxides of nitrogen are transformed by and removed from the
atmosphere by a complex web of reactions that are fundamental to the
formation and destruction of ozone and other oxidants. The
predominant form of oxidized nitrogen (NO, NO₂, HNO₃, etc.)
in the lower atmosphere varies, depending upon sunlight intensity,
temperature, pollutant emissions, period of time since these emissions
occurred and the meteorological history of an airmass.

2.4.2 Chemical transformations of oxides of nitrogen

2.4.2.1 Nitric oxide, nitrogen dioxide and ozone

The dominant source of nitrogen oxides in the air is combustion
processes (see chapter 3); 90-95% of these nitrogen oxides are usually
emitted as NO and 5-10% as NO₂. NO may be oxidized to NO₂ by
atmospheric oxygen according to reaction 2-1:

NO + NO + O₂ -> 2 NO₂ (2-1)

However at low NO concentrations this reaction is slow and is
important only when NO > 1 ppm (Boström C, 1993). NO concentrations
greater than 1 ppm are not frequently found in ambient air, but they
may possibly occur in indoor air and in plumes from industrial sources
(see Chapter 3). When concentrations are below 1 ppm, NO is oxidized
to NO₂ by two types of reaction. The first type of reaction is given
in equations 2-2 to 2-4. NO can react with O₃:

NO + O₃ -> NO₂ + O₂ (2-2)

Also O₃ is formed when NO₂ is photolysed, forming NO plus an O atom

NO₂ + hnu -> O + NO (2-3)

and O atoms react rapidly with O₂ to form ozone:

M
O + O₂ -> O₃ (2-4)

Thus reactions 2-2, 2-3 and 2-4 recycle O₃ rather than producing a
net increase in O₃ concentrations, where the "M" represents a third
molecule such as N₂, O₂, etc., that absorbs excess vibrational
energy from the newly formed O₃ molecules. However, a second
oxidation path involving the reaction of organic species can lead to
increases in O₃ concentrations and in the conversion rate of NO to
NO₂ (2-9 and 2-10). Organic compounds in the air are commonly
referred to as VOC (volatile organic carbon), ROC (reactive organic
carbon) and non-methane hydrocarbons (NmHC). Urban areas are usually
characterized by significant sources of both nitrogen oxides and ROC
emissions. With suitable atmospheric conditions this can lead to the
formation of photochemical smog. The smog-forming reactions are
initiated by photolytic reactions which produce free radicals, for
example:

(i) the photolysis of O₃

O₃ + hnu -> O₂ + O* (2-5)

O* is an excited form of atomic oxygen, which can react with water to
produce the hydroxyl radical (OH):

O* + H₂O -> 2OH (2-6)

(ii) the photolysis of aldehydes, which also results in the production
of OH. Aldehydes are emitted in motor vehicle exhaust and are
produced in the air by reaction of ROC species with OH. OH is the
most important oxidizing agent in the lower atmosphere; it can react
with all organic compounds, usually forming water and producing an
organic radical.

For a generalized organic compound, R-H (R = CH₃, CHO, CH₂CH₃,
etc.), the principal elements of the reaction sequence are:

R-H + OH -> H₂O + R (2-7)

M
R + O₂ -> RO₂ (fast) (2-8)

RO₂ provides a pathway to oxidize NO to NO₂ without destroying O₃
(unlike reaction 2-2):

RO₂ + NO -> NO₂ + RO (2-9)

RO can undergo reactions that form additional HO₂ or RO₂. HO₂
reacts with NO to form NO₂ and regenerate OH:

HO₂ + NO -> NO₂ + OH (2-10)

In the case of photochemical smog episodes, the quantity of NO_x
emitted into the air determines the ultimate quantity of O₃ that may
be produced. The ROC concentration and sunlight intensity are the
major determinates of the rates at which NO will be oxidized to
produce net increases in NO₂ and O₃ concentrations. Ozone
production is terminated when NO and NO₂ are consumed by reaction to
form products such as HNO₃ (see below), resulting in insufficient NO
concentration for reactions 2-9 and 2-10 to proceed at significant
rates.

In large cities with sunny climates and poor dispersion of
emissions (e.g., Los Angeles and Mexico City), O₃ concentrations in
excess of 200 ppb are not uncommon.

2.4.2.2 Transformations in indoor air

Oxides of nitrogen in indoor air arise from two sources: a)
outdoor air; and b) indoor sources, such as combustion appliances and
heaters. Photochemical reactions do not take place under artificial
lighting, so chemical transformations are limited by the amounts of
oxidizing species (HO₂, O₃, etc.) that arrive in outdoor air, or are
generated by combustion sources.

2.4.2.3 Formation of other oxidized nitrogen species

Oxidation products of NO_x arising from tropospheric
photochemical reactions include HNO₃, HO₂NO₂, HNO₂,
peroxyacylnitrates (RC(O)O₂NO₂), N₂O₅, nitrate radical (NO₃) and
organic nitrates (RNO₃).

Fig. 1 shows a summary for the interconversion pathways for
oxides of nitrogen. These pathways govern urban and indoor air, as
well as "clean" air, but the partitioning between the nitrogen oxide
species varies according to the specific conditions characteristic of
each type of airmass.

a) Nitric acid

Nitric acid is a strong mineral acid that contributes to acidic
deposition from the air. In terms of atmospheric chemistry, HNO₃ is
a major sink for active nitrogen. In daylight, HNO₃ is formed by the
reaction of NO₂ with the OH radical:

M
NO₂ + OH -> HNO₃ (2-11)

This reaction is a chain-terminating step in the free radical
chemistry that produces urban photochemical smog and it removes
reactive nitrogen as well as the hydroxyl radical. Reaction 2-11 is a
relatively fast reaction that can produce significant amounts of HNO₃
over a period of a few hours. At night, in polluted air containing
significant ozone concentrations, the heterogeneous reaction between
gaseous N₂O₅ and liquid water is thought to be a source of HNO₃
(N₂O₅ is produced from NO₃ (see section 2.4.3.5) and NO₂). This
pathway to HNO₃ production is negligible during daytime, because the
NO₃ radical photolyses rapidly and is not present in sufficient
quantities to react with NO₂. The NO₃ radical can also abstract a
hydrogen atom from certain organic compounds (such as aldehydes,
dicarbonyls and cresols) to provide another night-time source of
HNO₃.

Logan (1983) has estimated a lifetime of 1 to 10 days for
HNO₃ in the lower troposphere. The primary removal mechanism is
deposition. The loss of HNO₃ by rain-out is subject to precipitation
frequency while the loss rate by dry deposition varies with the nature

of the ground and vegetation and atmospheric mixing characteristics of
the boundary layer. Chemical destruction mechanisms for HNO₃ also
exist, but their importance is not well understood and is suspected to
be minor for the lower troposphere.

In the presence of NH₃, HNO₃ may form the salt, ammonium
nitrate:

HNO₃(g) + NH₃(g) -> NH₄NO₃ (2-12)

Ammonium nitrate gas readily condenses to the particulate phase.
Ammonium nitrate aerosol can be responsible for significant visibility
reduction and particulate pollution, e.g., where nitric acid is
produced in air from urban areas and this interacts with NH₃ emitted
from agricultural operations.

b) Nitrous acid

HNO₂ is produced from the reaction of NO and OH:

M
NO + OH -> HNO₂ (2-13)

In indoor air other reactions (particularly surface reactions)
may be important sources of nitrous acid.

There have been a few measurements of nitrous acid in urban
environments (Harris et al., 1982; Winer et al., l987). Daytime
levels of nitrous acid are expected to be low because it photolyses
rapidly, yielding NO and ·OH. This reaction probably serves as a
source of OH radicals during the morning in urban regions, where
nitrous acid may form (from NO, NO₂ and H₂O) and accumulate during
the night-time hours. Reaction 2-13 may lead to a build up of nitrous
acid in urban air only during the late afternoon and evening hours
when sunlight intensities are low but some OH radicals are still
present.

c) Peroxynitric acid

While peroxynitric acid (HO₂NO₂) has never been measured in the
atmosphere, it is expected to be present in the upper troposphere.
Models suggest concentrations in the 10 to 100 ppt range at altitudes
above 6 kilometres (Logan, 1983; Singh, 1987). HO₂NO₂ is thermally
unstable, so that boundary layer concentrations are expected to be
extremely low (< 1 ppt). Peroxynitric acid is formed through the
combination of a hydroperoxy (HO₂) radical with NO₂. In the upper
troposphere, HO₂NO₂ is destroyed by photolysis or by reaction with
OH radicals.

d) Peroxyacyl nitrates

Peroxyacetyl nitrate (PAN) is the most abundant of this family of
organic nitrates. The second most abundant homologue, peroxypropionyl
nitrate (PPN), is generally less than 10% of the PAN concentration,
and species with higher relative molecular mass, such as PBzN, are
expected to have even lower concentrations. PAN is a strong oxidant
and is known to be phytotoxic; it is formed from the reaction of
acetylperoxy radical with NO:

CH₃C(O)OO + NO₂ +M -> CH₃C(O)O₂NO₂ +M (2-14)

PAN is thermally unstable and so its lifetime is very dependent
on ambient temperature. For example, PAN lifetimes of about 5 and
20 h have been calculated for 20°C and 10°C, respectively.

In cold conditions PAN can serve as a reservoir for reactive
nitrogen, which is liberated when the temperature of the air is
increased. PAN can be lost from the atmosphere by dry deposition over
land, but it is very likely that a significant fraction of PAN
produced in urban plumes can be transported into the regional
environment.

e) Nitrate radical

The nitrate (NO₃) radical is a short-lived species formed mainly
by the reaction of NO₂ with O₃, although other sources of NO₃
radicals exist (Wayne et al., 1991).

NO₂ + O₃ -> NO₃ + O₂ (2-15)

NO₃ also reacts with NO₂ to form N₂O₅

M
NO₂ + NO₃ -> N₂O₅ (2-16)

Nitrate radicals rapidly photolyse, resulting in a lifetime of
about 5 seconds at midday. They also react rapidly with NO, which
limits their lifetime both during the day- and night-time hours. At
night if atmospheric NO concentrations are approximately 320 pptv,
then the lifetime of NO₃ radicals is similar to that at midday (about
5 seconds).

At night, NO₃ concentrations range from about 0.3 ppt in clean
tropospheric air to 70 ppt in urban areas (Biermann et al., 1988). In
clean background environments, it has been reported that measured NO₃
radical levels are significantly less than those predicted by the
above reactions. Several loss mechanisms have been suggested (Noxon
et al., 1980; Platt et al., 1981): (i) NO₃ radical reaction with

organic compounds; (ii) heterogeneous losses of NO₃ radicals and/or
N₂O₅ on particle surfaces; (iii) reactions of NO₃ radicals with
H₂O vapour; and (iv) reaction of NO₃ radicals with NO.

f) Dinitrogen pentoxide

N₂O₅ is formed from NO₃ and NO₂ (reaction 2-15). Since NO₃
is present only at night, N₂O₅ is also primarily a night-time
species. N₂O₅ is thermally unstable, decomposing to NO₃ and NO₂
(reaction 2-15). At high altitudes in the troposphere, where
temperatures are low, N₂O₅ can act as a temporary reservoir for
NO₃. Dinitrogen pentoxide photolyses at wavelengths less than 330 nm
to give NO₃ and NO₂.

Dinitrogen pentoxide reacts heterogeneously with water to form
HNO₃. This serves as the main night-time production mechanism for
HNO₃ and it provides an important route for removal of oxidized
nitrogen from the atmosphere, since HNO₃ is readily removed by dry
and wet deposition. Other atmospheric reactions of N₂O₅ include its
reaction with gas-phase water to form HNO₃ and possible reactions
with aromatic VOCs such as naphthalene and pyrene (Pitts et al., 1985;
Atkinson et al., 1986). Nitroarenes appear to be the product of
N₂O₅-aromatic reactions.

2.4.3 Advection and dispersion of atmospheric nitrogen species

The transport and dispersion of the various nitrogen species is
dependent on both meteorological and chemical parameters. Advection,
diffusion and chemical transformations dictate the atmospheric
residence time of a particular trace gas. Nitrogenous species that
undergo slow chemical changes in the troposphere and are not readily
removed by depositional processes can have atmospheric lifetimes of
several months. Gases with lifetimes of the order of months can be
dispersed over continental scales and possibly even over an entire
hemisphere. At the other extreme are gases that undergo rapid
chemical transformation and/or depositional losses limiting their
atmospheric residence times to a few hours or less. Dispersion of
these short-lived species may be limited to only a few kilometres from
their point of emission.

Surface emissions are dispersed vertically and horizontally
through the atmosphere by turbulent mixing processes. These processes
are dependent to a large extent on the vertical temperature structure
of the boundary layers and on wind speed. In the vertical dimension,
transport occurs as follows (see also Fig. 2.):

a) the daytime and/or night-time mixed layer; this layer can extend
from the surface up to a few hundred metres at night or to
several thousand metres during the daytime;

b) a layer that can exist during the night-time above a low level
surface inversion and below the daytime mixing height; this layer
generally is situated between 200 and 2000 m altitude;

c) the free troposphere; this transport zone is above the boundary
layer mixing region.

During the warm, summertime period, vertical mixing follows a
fairly predictable diurnal cycle. A surface inversion normally
develops during the evening hours and persists throughout the night-
time and morning period until broken by sunlight heating the surface
of the earth. While the inversion is in place, surface NO_x emissions
can lead to relatively high local concentrations because of restricted
vertical dispersion. Following the break-up of the night-time surface
inversion, vertical mixing will increase and surface-based emissions
will disperse to higher altitudes. The depth of the vertical mixing
during the daytime is often controlled by synoptic weather features.
Temperature inversions aloft, associated with high pressure systems,
are common in many parts of the world.

The dispersion processes described above, coupled with the
chemical transformations of reactive nitrogen compounds, determine the
distances oxidized nitrogen will be transported in the troposphere. A
reasonable understanding exists concerning the short-term (daylight
hours) fate of NO_x emitted in urban areas during the morning hours.
As described above, NO_x emitted in the early morning hours in an
urban area will disperse vertically and move downwind as the day
progresses. On sunny summer days, most of the NO_x will have been
converted to HNO₃ and PAN by sunset. Much of the HNO₃ will be
removed by depositional processes as the air mass moves along. After
dusk, an upper portion of the daytime mixed layer will be decoupled
from the surface because of formation of a low-level radiation
inversion. Transport will continue in this upper level during the
night-time hours and, although photochemical processes will cease,
dark-phase chemical reactions can proceed. Peroxyacetyl nitrate and
HNO₃, if carried along in this layer, can be transported long
distances.

2.4.3.1 Transport of reactive nitrogen species in urban plumes

Overall removal rates for reactive nitrogen species during
daytime at mid-latitudes have been measured or calculated for a few
areas. For example, in the plume from Boston, USA, after correction
for dilution, removal rates ranged from 0.14 to 0.24 h^-1 on 4 days
(Spicer, 1982, Altshuller, 1986). In Los Angeles and Detroit, the
removal rate has been estimated to be 0.04-0.1 h^-1 (Calvert, 1976;
Chang et al., 1979; Kelly, 1987). Formation and removal of HNO₃ is
thought to be the rate-controlling step for removal of reactive
nitrogen.

2.4.3.2 Air quality models

Air quality models are mathematical descriptions of pollutant
emissions, atmospheric transport, diffusion and chemical reactions
of pollutants. However, air quality models are very complex and
difficult to test for validity. Inputs include emissions, topography
and meteorology of a region. Air quality models represent an
integration of knowledge for the chemistry and physics of the
atmospheric system; they offer some predictive capability for the
effectiveness of pollution control strategies. Models have also been
developed for indoor air.

2.4.3.3 Regional transport

Transport of reactive nitrogen species in regional air masses can
involve several mechanisms. Mesoscale phenomena, such as land-sea
breeze circulations or mountain-valley wind flows, will transport
pollutants over distances of ten to hundreds of kilometres. On a
larger scale, synoptic weather systems such as the migratory highs
that cross the eastern USA and other areas of the world in the
summertime influence air quality over many hundreds of kilometres.
The accumulation and fate of nitrogen compounds will differ somewhat
between the mesoscale and synoptic systems. Mountain-valley and land-
water transport mechanisms have dual temporal scales because of their
dependence on solar heating. However, in the larger-scale synoptic
systems, reactive nitrogen species can build up over multiday periods.
The residence time of air parcels within a slow-moving high pressure
system can be as long as 6 days (Vukovich et al., 1977).

In many cases, the transport mechanisms mentioned above are
interrelated. Mountain-valley or land-water breezes can dictate
pollutant transport in the immediate vicinity of sources, but the
eventual fate of reactive nitrogen species will be distribution into
the synoptic system.

2.5 Conversion factor for nitrogen dioxide

1 ppm = 1.88 mg/m³
1 mg/m³ = 0.53 ppm

2.6 Summary

Combustion provides the major source of oxides of nitrogen in
both indoor and outdoor air, producing mostly NO with some NO₂. The
sum of NO and NO₂ is generally referred to as NO_x. Once released
into the air, NO is oxidized to NO₂ by available oxidants,
particularly O₃, and by photochemical reactions involving reactive
organic compounds. This happens rapidly under some conditions in
outdoor air; for indoor air, it is generally a much slower process.
Nitrogen oxides are a controlling precursor of ozone and smog

formation; interactions of nitrogen oxides (except N₂O) with reactive
organic compounds and sunlight form ozone in the troposphere and smog
in urban areas.

In both indoor and outdoor air, NO and NO₂ may undergo reactions
to form a suite of other nitrogenous species including HNO₂, HNO₃,
NO₃, N₂O₅, PAN and other organic nitrates. The complete suite of
gas-phase nitrogen oxides is referred to as NO_y. The partitioning
of nitrogen among these compounds is strongly dependent on the
concentrations of other oxidants, sunlight exposure, the presence of
reactive organic compounds and the meteorological history of the air.

A sensitive, specific and reliable analytical method exists for
measuring NO (by the chemiluminescent reaction with ozone), but this
is an exception for NO_y species. Chemiluminescence is also the
most common technique used for NO₂, which is first reduced to NO.
Unfortunately, the method of reduction usually used is not specific
for NO₂, and it has various conversion efficiencies for other
oxidized nitrogen compounds that may also be present in the air
sample. For this reason, care must be taken in interpreting the NO₂
values given by the common chemiluminescence analyser, as the signal
may include responses from interfering compounds. Additional
difficulties arise from nitrogen species such as HNO₃ that may
partition between the gas and particulate phases both in the
atmosphere and in the sampling procedure.

3. SOURCES, EMISSIONS AND AIR CONCENTRATIONS

3.1 Introduction

Oxides of nitrogen can have significant concentrations in ambient
air and in indoor air. The types and concentrations of nitrogenous
compounds present can vary greatly from location to location, with
time of day, and with the season. The main sources of nitrogen oxides
emissions are combustion processes. Fossil fuel power stations, motor
vehicles and domestic combustion appliances emit nitrogen oxides,
mostly in the form of NO but with some (usually less than about 10%)
in the form of NO₂. In the air chemical reactions occur which
oxidize NO to NO₂ and other products (chapter 2). Also, there are
biological processes in soils which liberate nitrogen species,
including N₂O. Emissions of N₂O can cause perturbation of the
stratospheric ozone layer.

Human health may be affected when significant concentrations of
NO₂ or other nitrogenous species, such as PAN, HNO₃, HNO₂ and
nitrated organic compounds, are present. In addition, nitrates and
nitric acid can cause significant effects on ecosystems when deposited
on the ground.

Indoors, the use of combustion appliances for cooking and heating
can give rise to greater NO and NO₂ concentrations than are present
outdoors, especially when the appliance is not vented to the outside.
Recent research has shown that in these circumstances nitrous acid can
reach significant concentrations (Brauer et al., 1993).

This chapter discusses both ambient and indoor sources
of nitrogenous compounds, their emissions, and the resulting
concentrations that may directly affect human health or participate
in atmospheric chemical pathways leading to effects on human health
and welfare. Nitrogen-containing compounds are also of particular
interest because of their secondary impacts. For example, production
of photochemical smog and ozone pollution depends on emissions to the
air of nitrogen oxides together with volatile organic compounds.
Nitric acid, which is produced in the air by the reaction of hydroxyl
radicals (OH*) with NO₂, is one of the major components of acidic
precipitation. As well as being present in the gas phase, oxidized
nitrogen can, by reaction and adsorption, become incorporated into
aerosol particles. Graedel et al. (1986) identified 20 inorganic
nitrogen-containing species detectable in the atmosphere. Near
cities and urban regions the species usually present in greatest
concentrations are NO and NO₂, and these are the most reliably
measured and frequently monitored nitrogen oxide species.

Knowledge of emission patterns and concentrations of nitrogenous
compounds is critically important for air quality planning and human
health and environment risk assessments. Because nitrogen oxides and

their reaction products have lifetimes of several days in the
atmosphere, they can be transported long distances by the wind and
give rise to environmental impacts far from their source of emission.

3.2 Sources of nitrogen oxides

Combustion systems emit NO and NO₂ and together these species
are usually denoted as NO_x.

When NO_x emissions are expressed in mass units, the mass is
expressed as if all the NO had been converted to NO₂. Another
convention adopted in some of the following sections is to report the
emissions on a mass basis in terms of the nitrogen content.

3.2.1 Sources of NO_x emission

3.2.1.1 Fuel combustion

Annual production of NO_x from combustion of fossil fuels is
typically estimated from emission factors for various combustion
processes, combined with worldwide consumption data for coal, oil and
natural gas. Logan (1983) provided a tabular summary of emission
factors, which has been updated by the US National Acid Precipitation
Assessment Program (Placet et al., 1991). Owing to variations in
process operating conditions, the emission factors must be considered
to be uncertain by about ± 30%. Table 3 provides a summary of global
emission estimates for NO_x according to fuel type. The estimates of
Logan (1983) are slightly higher than those of Ehhalt & Drummond
(1982), the largest discrepancies being in emission estimates for the
transportation sector. The differences arise because Logan (1983)
based estimates of emissions on fuel usage, while Ehhalt & Drummond
(1982) scaled the totals somewhat indirectly by using world automobile
population numbers.

Dignon (1992) has assembled a database for mapping (with a
resolution of one degree in latitude and longitude) and estimated
global NO_x and sulfur oxides emissions from their common principal
anthropogenic source, i.e. fossil fuel combustion. For 1980, the
global total was estimated to be 22 million tonnes, as nitrogen.
Countries heading the list (in millions of tonnes of nitrogen per
year) were: USA, 6.4; USSR, 4.4; China, 1.7; Japan, 0.80; and Federal
Republic of Germany, 0.66. An estimated 95% of NO_x emissions from
fossil fuel combustion originates in the northern hemisphere.

For oceanic regions, shipping is a source of NO_x emissions.
Aircraft also emit nitrogen oxides and this may be significant for the
upper troposphere and stratosphere.

Table 3. Estimates of global emissions of nitrogen oxides (NO_x) from combustion of fossil fuels and biomass (from: US EPA, 1993)^a

Source type Annual consumption Emission factors^b Global source strength
(10⁶ tonnes, unless indicated otherwise) (10⁶ tonnes nitrogen/year)

(E & D) (L) (C et al.) (E & D) (L) (E & D) (L) (C et al.)

Fossil fuels^c

Hard coal 2150 2696 - 1.0-2.8 2.7 3.9 (1.9-5.8) 6.4 -
Lignite 810 - 0.9-2.7 1.6 (0.8-2.3) -
Light fuel oil 300 1.39 - 1.5-3.0 2.2^d 0.7 (0.5-0.9) 3.1 -
Heavy fuel oil 470 1.5-3.1 1.1 (0.7-1.5) -
Natural gas 1.04 1.2 × 10⁹ m³ - 0.6-3.0 1.9^d 1.9 (0.6-3.1) 2.3 -
Industrial sources - - - 1.2 -
Automobiles (4.1-5.4) 1.0 × 10⁹ m³ - 0.9-1.2^e 8.0^d 4.3 (3.7-6.4) 8.0
× 10¹² km

Total 13.5 (8.2-18.5) 19.9

Biomass burning^f

Savanna (6-14) × 10³ 2000 1200 1.0 1.7 3.1 (1.8-4.3) 3.4 2.1
Forest clearings (2.7-6.7) × 10³ 4100 2700 1.0-1.6 2.0 2.1 (0.8-3.4) 8.2 4.7
Fuel wood - 850 1100 - 0.5 2.0 (1-3) 0.4 0.5
Agricultural waste - 15 1900 - 1.6 4.0 (2-6) 0.02 3.3

Total 11.2 (5.6-16.4) 12.0 10.6

^a Estimates according to Ehhalt & Drummond (1982) (E & D) and Logan (1983) (L). Ranges are given in parentheses.
^b Emission factors refer to grams of nitrogen per kg of fuel consumed, unless indicated otherwise
^c Petroleum refining and manufacture of nitric acid and cement; global emissions were obtained by scaling USA emissions for each
industrial process
^d Grams of nitrogen per m³ of fuel consumed
^e Grams of nitrogen per km
^f For biomass-burning, Crutzen et al. (1979) (C et al.) have given annual consumption rates differing somewhat from those of the other
authors. The data of Crutzen et al. (1979) and the resulting nitrogen oxides production rates are included for comparison
3.2.1.2 Biomass burning

Table 3 includes a breakdown of estimates for release of NO_x
from burning of biomass. In natural fires and the burning of wood,
temperatures are rarely high enough to cause oxidation of nitrogen
molecules of the air. The emissions are thereby more closely related
to the fixed nitrogen content of the fuel. Logan (1983) reviewed a
number of experimental determinations of nitrogen emission factors
that indicate yields are highest for grass and agricultural refuse
fires (1.3 g nitrogen/kg fuel), less for prescribed forest fires
(0.6 g nitrogen/kg fuel), and still lower for burning of fuel wood in
stoves and fireplaces (0.4 g nitrogen/kg fuel). The values roughly
reflect differences in nitrogen content of the materials burned.
Biomass burning is mainly associated with agricultural practices in
the tropics, which include plant, slash, and shift practices as well
as natural or intentional burning of savanna vegetation at the end of
the dry season. Forest wildfires and use of wood as fuel make a
lesser contribution.

3.2.1.3 Lightning

Thunderstorm activity has been viewed as a major NO_x source
since 1827, when Von Liebig proposed it as a natural mechanism for
fixation of atmospheric nitrogen. Electrical discharges in air
generate NO_x by thermal dissociation of nitrogen molecules due to
ohmic heating inside the discharge channel and shockwave heating of
the surroundings. Laboratory studies by Chameides et al. (1977) and
Levine et al. (1981) indicate an NO_x yield of 6 × 10¹⁶ molecules per
joule of spent energy. Great uncertainties exist, however, about the
total energy generated by lightning in the atmosphere. Noxon (1976,
1978) first studied the increase of NO_x in the air during a
thunderstorm. His results provide the basis for many of the estimates
shown in Table 4. Reviews by Kowalczyk & Bauer (1981) Borucki &
Chameides (1984) and Albritton et al. (1984) provide a best estimate
of annual generation by lightning: 1 million tonnes of NO_x in North
America and 13 million tonnes globally (Placet et al., 1991).

3.2.1.4 Soils

The biochemical release of NO_x from soils is poorly understood,
and the flux estimates must be viewed with caution. Both rely on the
observations by Galbally & Roy (1978), who used the flux box method in
conjunction with chemiluminescence detection of NO_x. They found
average fluxes of 5.7 and 12.6 µg nitrogen/m²*h on ungrazed and
grazed pastures, respectively, where NO was the main product. More
recent measurements of Slemr & Seiler (1984) indicate that the release
of NO_x from soils depends critically on the temperature and moisture

content of the soil, which in turn complicates the estimate of the
global emissions. Slemr & Seiler (1984) also found an average release
rate of 20 µg nitrogen/m² per h for uncovered natural soils, evenly
divided between NO and NO₂. Grass coverage reduced the escape flux,
whereas fertilization enhanced it. Ammonium fertilizers were about
five times more effective than nitrate fertilizers. This suggests
that nitrification as a source of NO_x is more important than
denitrification. According to Slemr & Seiler (1984), an annual global
flux of 10 million tonnes of nitrogen represents an upper limit to the
release of NO_x from soils. Galbally et al. (1985) presented more
detailed estimates for arid lands, and Table 4 provides a compilation
of current literature used to develop the global budgets. Soil is
also a source of N₂O and NH₃ emissions.

In the presence of low concentrations, plants can emit NH₃,
rather than absorb it. This is especially true with scenescing and
with highly fertilized plants (Grünhage et al., 1992; Holtan-Hartwig &
Bockman 1994; Fangmeijer et al., 1994). Release to the atmosphere of
N₂ and NO by plants has also been reported. In some cases this was
part of the response following exposure to nitrogen-containing
pollutants, but other mechanisms are involved (Wellburn, 1990). NO and
N₂O are emitted in significant quantities by the soil. The reason
why the deposition velocity of NO is relatively low see (see Table 5)
is partly due to the fact that the downward flux (and uptake by the
canopy) is "mathematically" compensated by soil emissions. In other
words: a low deposition velocity does not always mean that the uptake
by the vegetation is low. In the case of N₂O, soil emissions are
mostly larger than deposition; this emission is the result of
denitrification and is positively related to the nitrogen and water
content and the temperature of the soil. This is why the release of
nitrogen from the ecosystem in the form of N₂O is dependent on the
ecosystem type, climate and land use (fertilization and water table
height). Skiba et al. (1992) estimated for the United Kingdom the NO
and N₂O emissions from agricultural land to be 2-6% of the nationwide
NO_x emissions and 16-64% of the N₂O emissions, respectively.

Table 4. Global and North America natural emissions (average and range) of nitrogen oxides (NO_x)
from lightning, soils and oceans

Global North America Reference
(10⁶ tonnes/year) (10⁶ tonnes/year)

Lightning 8.6 (2.6-26) Borucki & Chameides (1984)
18 1.7 Albritton et al. (1984)
13 (7-26) 1 (0.3-2) Kowalczyk & Bauer (1981); Placet et al. (1991)

Soils 50 (as NO₂) Lipschultz et al. (1981)
30 (as NO) Levine et al. (1984); Galbally & Roy (1978)
36 Slemr & Seiler (1984)
2 Placet et al. (1991)

Oceans 0.35 Zafiriou & McFarland (1981); Logan (1983)

Table 5. Deposition velocity of nitrogen-containing gases and
aerosols

Deposition velocity Reference
(mm/second)

NO₂ 0.1-10 Grennfelt et al. (1983);
Anonymous (1991)

NO 0.2-1 Prinz (1982)

NH₃ 12 (-5 - +40) Grünhage et al. (1992);
Sutton et al. (1993);
Fangmeijer et al. (1994);
Holtan-Hartwig & Bockman (1994)

NH₄⁺ 1.4 (0.03-15) Fangmeijer et al. (1994)

Estimates of global emissions of N₂O and ammonia are summarized
in Table 6.

Table 6. Annual global estimates (average and range) of N₂O and
NH₃ emissions to the troposphere (10⁶ tonnes
of nitrogen)

Source N₂O NH₃ Reference

Soils 10 (2-20) 15 Dawson (1977);
Boettger et al. (1978)

Ocean 26 (12-38) Hahn (1981)

Biomass burning 2 2-8 Crutzen et al. (1979);
Crutzen (1983)

Fossil fuels 1.6 0.2 Weiss & Craig (1976);
Boettger et al. (1978)

Fertilizer 0.1 3 Boettger et al. (1978);
Crutzen et al. (1979);
Crutzen (1983);
Stedman & Shetter (1983)

Domestic animals 22 Soederlund & Svensson (1976)
Boettger et al. (1978);
Crutzen et al. (1979);
Crutzen (1983);
Stedman & Shetter (1983)

3.2.1.5 Oceans

There have been few measurements of NO_x, N₂O or NH₃ fluxes
over the ocean, and current literature suggests that the sea is a
negligible source of NO. Zafiriou & McFarland (1981) observed a
supersaturation of seawater with regard to NO in regions of relatively
high concentrations of nitrite, owing to upwelling conditions. The

excess NO must, in this case, arise from photochemical decomposition
of nitrite by sunlight. Logan (1983) estimated a local source
strength of 1.3 × 10¹² molecules/m² per second under these
conditions. Linear extrapolation results in an annual global flux
estimate of 350 000 tonnes of nitrogen.

3.2.2 Removal from the ambient environment

Wet precipitation and dry deposition provide two of the major
mechanisms for removal of NO_x from the atmosphere. The addition to
the plant soil ecosystem of nitrate (and ammonium) by rainwater
constitutes an important source of fixed nitrogen to the terrestrial
biosphere, and until 1930 practically all studies of nitrate in
rainwater were concerned with the input of fixed nitrogen into
agricultural soils. Eriksson (1952) and Boettger et al. (1978) have
compiled many of the available data. Despite the wealth of
information, it remains difficult to derive a global average for the
deposition of nitrate, because of an uneven global coverage of the
data, unfavourably short measurement periods at many locations, and
inadequate collection and handling techniques for rainwater samples.
In addition, the concentration of nitrate in rainwater has increased
in those parts of the world where the utilization of fossil fuels has
led to a rise in the emissions of NO_x, i.e. primarily western Europe
and the USA.

Dry deposition is important as a sink for those gases that are
readily absorbed by materials covering the earth surface. In the
budget of NO_x, the gases affected most by dry deposition are NO₂ and
HNO₃. The deposition velocity of NO is too small and the
concentration of peroxyacetyl nitrates is not high enough for a
significant contribution.

According to Grennfelt et al. (1983) and Wellburn (1990), NO_3^-
and HNO₃ have a higher deposition velocity then NH₃, but this was
not quantified. HNO₂ is assumed to have a deposition velocity equal
to SO₂: 1-30 mm/second (Table 5).

There are several other nitrogen-containing air pollutants with
relatively high deposition velocities. These generally add only small
amounts to the total nitrogen deposition, because most of the time
their ambient concentrations are relatively low.

Atmospheric nitrogen deposition can significantly change the
chemical composition of the soil. In the rooting zone these changes
have an impact on vegetation. The changes in deeper soil layers
are particularly relevant if groundwater is used as a source of
drinking-water. Groundwater under fertilized agricultural land can be
heavily polluted with nitrate (and aluminium), but this is beyond the
scope of this chapter. Due to atmospheric nitrogen deposition, the

groundwater under forests and other non-fertilized vegetation can
become polluted with nitrate. For instance, in 20% of the forested
area of the Netherlands, the nitrate concentration in phreatic
groundwater is higher than 50 mg/litre (the EC drinking-water
standard); in 37% it is higher than 25 mg/litre (Boumans & Beltman,
1991). The average annual nitrogen deposition in the Netherlands is
45 kg/ha; approximately 10 kg/ha is from dry deposition of NO_x. The
nitrate concentration in groundwater is strongly related to the soil
type. With the same atmospheric deposition, the nitrate concentration
increases as follows: peaty soils < moderately drained sandy soils
< well-drained rich sandy soils (Boumans, 1994). A distinct relation
also exists concerning the age of the trees: tree stands in Wales
showed nitrate leaching (measured in the stream water draining the
catchments), but only with stands older than 30 years. Younger trees
used the nitrogen as nutrient, but the nitrogen demand of the older
trees was lower. The annual nitrogen deposition in that region was
estimated to be 20 kg/ha (Emmett et al., 1993).

3.2.3 Summary of global budgets for nitrogen oxides

The principal routes to the production of NO_x are combustion
processes, nitrification and denitrification in soils, and lightning
discharges. The major removal mechanism is oxidation to HNO₃,
followed by wet and dry deposition. In developing Table 7, the dry
deposition velocities for NO₂ over bare soil, grass and agricultural
crops were assumed to fall in the range of 3 to 8 mm/second. However,
over water the velocities are significantly smaller, so that losses of
NO₂ by deposition onto the ocean surface can be ignored. The
absorption of nitric acid by soil, grass and water is rapid, and dry
deposition correspondingly important, but the global flux is difficult
to estimate because information on HNO₃ mixing ratios is still
sparse. Logan (1983) adopted NO mixing ratios of 50 pptv over the
oceans and 100 pptv over the continents. The mixing ratios assumed
for NO₂ were 100 and 400 pptv, respectively. Allowance was made for
higher mixing ratios in industrialized areas affected by pollution.
Logan (1983) included the deposition of particulate nitrate over the
oceans, using a settling velocity of 3 mm/second. This process
contributes 2 million tonnes nitrogen/year to a total dry deposition
rate of 12 to 22 million tonnes nitrogen/year.

Efforts by Boettger et al. (1978), Ehhalt & Drummond (1982),
Galbally et al. (1985) and Warneck (1988) to quantify the sources and
sinks have led to an improved understanding of the global budget of
NO_x, in which the flux of NO_x into the troposphere and the rate of
nitrate deposition are approximately balanced. Ehhalt & Drummond
(1982) relied on the detailed evaluation of data by Boettger et al.
(1978). Their analysis emphasized measurements from the period 1950
to 1977, and they prepared a world map for nitrate deposition rates,

which were then integrated along 5° latitude belts. Logan (1983)
considered recent network data from North America and Europe; Galloway
et al. (1982) reported measurements of nitrate in precipitation at
remote locations in Alaska, South America, Australia and the Indian
Ocean. Both estimates gave wet nitrate deposition rates in the range
of 2 to 14 million tonnes nitrogen/year for the marine environment and
8 to 30 million tonnes nitrogen/year on the continents. An earlier
appraisal by Soederlund & Svensson (1976) led to rather similar
values, i.e. 5 to 16 and 13 to 30 million tonnes nitrogen/year,
respectively, although it was primarily based on Eriksson's (1952)
compilation of data from the period 1880 to 1930.

Table 7. Global budget (average and range) of nitrogen oxides
in the troposphere (from US EPA, 1993)^a

Type of source or sink Global flux
(10⁶ tonnes nitrogen/year)

Ehhalt & Logan (1983)
Drummond (1982)

Production

Fossil-fuel combustion 13.5 (8.2-18.5) 21 (14-28)
Biomass burning 11.2 (5.6-16.4) 12 (4-24)
Release from soils 5.5 (1-10) 8 (4-16)
Lightning discharges 5.0 (2-8) 8 (2-20)
NH₃ oxidation 3.1 (1.2-4.9) uncertain (1-10)
Ocean surface (biologic) - < 1
High-flying aircraft 0.3 (0.2-0.4) -
Stratosphere 0.6 (0.3-0.9) approx. 0.5

Total production 39 (19-59) 50 (25-99)

Losses

Wet deposition of NO₃^-, land 17 (10-24) 19 (8-30)
Wet deposition of NO₃^-, oceans 8 (2-14) 8 (4-12)
Wet deposition, combined 24 (15-33) 27 (12-42)
Dry deposition of NO_x - 16 (12-22)

Total loss 24 (15-40) 43 (24-64)

^a Derived from estimates according to Ehhalt & Drummond (1982)
and Logan (1983)

On continents, one should also consider the interception of
aerosol particulates by high growing vegetation. The interception of
nitrate is expected to be particularly effective. Hoefken &
Gravenhorst (1982) studied the enrichment of nitrate in rainwater
collected underneath forest canopies compared to that collected in
open areas outside forests. The effect is caused by the wash-off of
dry-deposited material from foliage. Hoefken & Gravenhorst (1982)
found that, in a beech forest, nitrate was enhanced by a factor of
1.4, whereas in a spruce forest enhancement by a factor of 4.1
occurred. Unfortunately, they were unable to differentiate between
contributions of particulate nitrate versus gaseous nitrate to the
total dry deposition.

If losses of NO₂ and HNO₃ by dry deposition are included in the
total budget of NO_x, one obtains a reasonable balance between the
sources and sinks, as Table 7 shows. Ehhalt & Drummond (1982) noted
that an appreciable part of their dry deposition is already included
in their wet deposition rates, because rain gauges frequently are left
open continuously, so that the collection of nitrate occurs during
both wet and dry periods. For NO₂, they estimated a dry deposition
rate of 7 million tonnes nitrogen/year. Because of the uncertainty,
they chose to include it in the error bounds and not in the mean value
of total NO_x-derived nitrogen deposition. Clearly, the total budget
of NO_x is far from being well defined. Moreover, in view of the
relatively short residence times of chemical species involved in the
NO_x cycle, it is questionable whether a global budget gives an
adequate description of the tropospheric behaviour of NO_x and its
reaction products. Supplemental regional budgets could be more
appropriate.

3.3 Ambient concentrations of nitrogen oxides

Because cities usually have an aggregation of emissions sources
ambient concentrations of NO and NO₂ tend to be greatest in cities.
High concentrations of NO are common in street canyons, owing to motor
vehicle emissions. In rural areas the emissions may have spent
considerable time in the atmosphere and have undergone reactions to
produce significant concentrations of other species, such as HNO₃ and
PAN.

3.3.1 International comparison studies of NO_x concentrations

Data for monthly average concentrations of NO_x collected by the
World Meteorological Organization at five background locations in
Europe for the period 1983 to 1985 are summarized in Fig. 3 (WMO,
1988, 1989). Fig. 4 presents published monthly averages of NO₂ in
1987 for 12 stations in a cooperative network under the Organisation
for Economic Co-operation and Development (OECD) (Grennfelt et al.,
1989). These two figures show that concentrations of both NO_x and
NO₂ tend to be higher during winter months.

Measurements of NO₂ in several countries during the late 1970s
and early 1980s are summarized in "Assessment of Urban Air Quality"
(WHO, 1988). The trends in composite annual averages for urban NO₂
monitoring stations in five countries are portrayed in Fig. 5 for the
period 1975 to 1985. The trend in the Canadian data appears to have
been downward, but essentially stable trends were evident for data
from the other countries. Annual averages in the 1980-1984 period for
42 cities around the world are summarized in the same report (WHO,
1988). During that period, only one city, Sao Paulo, reported an
annual average greater than 0.053 ppm (100 µg/m³).

Short-term peak values (1-h or 30-min maxima, or 98th or 95th
percentile values) have been reported for 18 cities during the
1980-1984 period (WHO, 1988). Ten of these cities (Amsterdam,
Brussels, Hamilton, Hong Kong, Jerusalem, Montreal, Munich, Rotterdam,
Tel Aviv and Toronto) reported values above the WHO 1-h guideline
level of 400 µg/m³ (0.21 ppm) for at least one year during that
5-year period. For eleven cities in the WHO report, both the annual
average and a "1-hour" peak statistic were reported for the 1980-1984
period. Fig. 6 compares these two statistics. It shows that three
cities, Amsterdam, Jerusalem and Tel Aviv, reported an average peak
value above the WHO 1-hour guideline value of 400 µg/m³ (0.21 ppm).
It should be kept in mind that the peak-value statistic is more
susceptible to undetected spurious measurements than is the annual
average. Data from the remaining eight cities place them in the
quadrant below the target levels for both the annual average and the
1-hour peak. A similar situation is seen in the majority of cities in
the USA and is discussed in the next section.

More recent data on NO₂ trends in the world's largest cities
have been reported by WHO/UNEP (1992) in the monograph "Urban Air
Pollution in Megacities of the World". Such trends for six selected
cities from various regions of the world are illustrated in Fig. 7, a
composite of figures extracted directly from the WHO/UNEP (1992)
report. In general, the overall trends appeared to be relatively
stable for most of the cities (and/or specific neighbourhoods).
However, there were a few exceptions, e.g., an apparent decrease in
the late 1980s for Bombay and an apparent increase during the same
period for some areas of Moscow. There are substantial differences in
the concentrations reported for different cities.

Table 8 summarizes emissions of nitrogen oxides and ambient
monitoring data from the WHO/UNEP (1992) report for the years
indicated. Included are estimates for total emissions and percentages
attributed to mobile sources, primarily private motor vehicles and
public land transport systems. However, the quality and type of
information contained in the report is mixed, reflecting a variety of
monitoring methods and reporting policies in different countries.
Ambient data in some cities was reported as NO_x, and in others as
NO₂; reporting periods varied from one hour to one year.

Table 8. Estimated mobile and stationary source emissions of nitrogen oxides in
megacities (from: WHO/UNEP, 1992)^a

City Total emissions of Mobile source Ambient concentration
nitrogen oxides contribution (µg/m³)
(tonnes/year) (%)

Bangkok 60 000 (1990) 30 max 1 h NO_x (as NO₂)
270 at one site; < 320 at
three stations (1987)

Beijing na

Bombay 56 000 (1990) 52 NO₂ 70-85 (annual 98th
percentile, 1990)

Buenos Aires 27 000 (1989) 48 na

Cairo 24 700 (1989) 23 NO_x 380-1400 (1979,
monthly means; single
study)

Calcutta 36 550 (1990) 29

Delhi 73 000 (1990) 20 NO₂ 500 (1990, 8 h)
(mostly diesel)

Jakarta 20 500 (1989) 75 NO_x 28 (1990, annual mean)

Karachi 50 000 (1989) 38 38-544 (12-13 June 1988;
single study)

Table 8. (Con't)

City Total emissions of Mobile source Ambient concentration
nitrogen oxides contribution (µg/m³)
(tonnes/year) (%)

London 79 000 (1983) 75 (1984) NO₂ max 1 h 867; > 600
for 8 h; > 205 for 72 h
(episode 12-15 Dec. 1991);
98th percentile > 135;
50th percentile > 50 (1989);
NO recorded but not
reported

Los Angeles 440 000 (1987) 76 NO₂ max 1 h 526; > 400
at 8 out of 24 stations (1990)

Manila 119 000 (1990 - 90 na
dubious accuracy)

Mexico City 177 300 (1991) 75 NO₂ hourly maxima
301-714 (1986-91)

Moscow 210 000 (1990) 19 NO₂ max daily means
100-150

New York 120 000 New York na NO₂ 1 h max 402; daily
City; 513 000 New max 160; annual mean 87
York metropolitan (1990)
area (1985)

Rio de Janeiro 63 000 (1978) 92 na

Sao Paulo 245 000 (1988) 82 NO₂ max 1 h
600-1500 (1988)

Table 8. (Con't)

City Total emissions of Mobile source Ambient concentration
nitrogen oxides contribution (µg/m³)
(tonnes/year) (%)

Seoul 270 000 (1990) 78 NO₂ annual means only

Shanghai 127 000 (1983); na NO_x annual mean 50;
1991 emissions indoor level 90
assumed 50%
higher, i.e.
approx. 190 000

Tokyo 52 700 (1985) 67% from motor daily mean 98th percentile
vehicles; 5% from > 115 tolerable level at
ship and aircraft 25% of stations

^a na = not available

As shown in Table 8, of importance for air quality management is
the large contribution of NO_x from motor vehicles reported for some
cities and the continuing growth in this contribution. For example,
emissions from vehicles in Bombay (about 29 000 tonnes per year in
1990) are expected to increase by an additional 14 600 tonnes/year by
the year 2000 (WHO/UNEP, 1992).

Estimates for Jakarta attribute some three-quarters of NO_x
emissions to motor vehicles, which is comparable with London, Los
Angeles and Mexico City. Data from Manila indicate that some 90% of
NO_x originates from motor vehicles.

3.3.2 Example case studies of NO_x and NO₂ concentrations

Data from a range of countries and locations are given in Table 9
(Agra, India) and Tables 10 and 11 (various cities in China).

Table 9. Concentrations of NO₂ measured in the vicinity of the
Taj Mahal, Agra India^a

Year Mean monthly concentration range (µg/m³)

1987 5.5 to 41.9
1988 6.3 to 33.1
1989 4.2 to 15.2

^a Highest concentrations tend to occur in winter
Personal communication from R.R. Khan, Ministry of Environment and
Forests, New Delhi, India (1994)

In urban areas in the USA, hourly patterns at fixed-site ambient
air monitors often follow a bimodal pattern of morning and evening
peaks, related to motor vehicular traffic patterns. Sites affected by
large stationary sources of NO₂ (or NO that reacts to produce NO₂)
are often characterized by short episodes at relatively high
concentrations, as the plume moves to downwind areas.

Since 1980, the annual average level among NO₂-reporting
stations in the USA has been below 0.03 ppm, with no significant
trend evident. This is exemplified in Fig. 8 (US EPA, 1991) by
annual averages for the period 1980 to 1989 for 60 metropolitan
areas subdivided into three population categories: 16 areas with a
population of 250 000 to 500 000, 14 with 500 000 to one million, and

Table 10. Annual average NO_X concentration (µg/m³) in China from 1981 to 1990^a

Year Cities all over China Southern cities Northern cities

Concentration Annual Concentration Annual Concentration Annual
range average range average range average

1981 10-90 50 10-80 40 20-90 60
1982 10-110 45 10-90 40 30-110 50
1983 9-94 46 9-79 36 29-94 55
1984 10-95 42 13-75 37 10-95 46
1985 13-49 50 13-84 41 22-49 59
1986 14-108 48 14-98 41 18-108 55
1987 17-199 56 17-60 43 30-199 69
1988 9-110 45 9-110 42 8-120 48
1989 10-140 47 10-133 43 12-140 51
1990 7-130 43 12-71 38 7-130 47

^a General Environmental Monitoring Station of China (1991)

Table 11. Statistical data for the percentiles of ambient annual average NO_x concentrations (µg/m³) for Chinese cities (1986-1990)^a

Year Number Minimum Percentile Maximum Arithmetic Geometric
of cities value value
5 10 25 50 75 90 95 Average Standard Average Standard
deviation deviation

1986 71 14 17 20 30 43 60 81 88 108 48 22 43 488

1987 71 13 16 21 33 46 60 74 80 105 48 20 44 478

1988 73 8 11 18 30 43 58 67 84 120 45 22 40 547

1989 63 10 14 19 30 44 58 64 87 140 47 26 41 546

1990 59 7 13 17 27 38 51 71 86 130 43 23 37 554

^a General Environmental Monitoring Station of China (1991)

30 with over one million. No group exhibited a time trend, but the
areas with more than one million people clearly reported levels higher
than the smaller metropolitan areas. For 103 Metropolitan Statistical
Areas (MSA) reporting a valid year's data for at least one station in
1988 and/or 1989, peak annual averages ranged from 0.007 to 0.061 ppm
(Fig. 9). The only recently measured concentrations exceeding the USA
annual average standard (0.053 ppm) have occurred at stations in
southern California.

The seasonal patterns at stations in California are usually quite
marked and reach their highest levels through the autumn and winter
months. Stations elsewhere in the USA usually have less prominent
seasonal patterns and may peak in the winter or summer, or may contain
little discernable variation (Fig. 10) (US EPA, 1991).

One-hour NO₂ values at stations in the USA can exceed 0.2 ppm,
but in 1988 only 16 stations (12 of which are in California) reported
an apparently credible second high 1-h value above 0.2 ppm (Fig. 11).
Because at least 98% of 1-h values at most stations are below 0.1 ppm,
these values above 0.2 ppm are quite rare excursions whose validity
should be verified (US EPA, 1991).

3.4 Occurrence of nitrogen oxides indoors

This section summarizes emissions of NO_x from sources that
affect indoor air quality and are commonly found in residential
environments. There are several reasons for considering these
emissions. Firstly, examining emissions from several types of sources
and source categories can help identify the relative impact of each
source on indoor air quality and thus its influence on human exposure.
Secondly, such information is needed to understand the fundamental
physical and chemical processes influencing emissions. This
understanding can be used to help develop strategies for reducing
emissions. Finally, studying emissions from indoor sources can
provide source strength input data needed for indoor air quality
modelling. Knowledge of indoor concentrations is an important
component in estimating the total exposure of individuals to nitrogen
oxides.

An important factor for indoor air quality is how (or if) the
combustion products from appliances are vented outside the building.
It should be noted that several common types of vented appliances
usually emit NO_x to the outdoors; examples include gas-fired
furnaces, water heaters and clothes dryers, as well as stoves and
furnaces using wood, coal and other fuels. Under some circumstances
even these vented emissions may filter back inside and contribute to
elevated NO_x levels indoors. For example, Hollowell et al. (1977)
reported high NO and NO₂ concentrations in a house where a vented
forced-air gas-fired heating system was used. Elevated concentrations
may also be a problem with malfunctioning vented appliances. Other

data (e.g., Fortmann et al., 1984), however, suggest that fugitive
emissions of NO_x from vented appliances are small. The importance of
unvented appliances to indoor NO_x levels is well documented; this
section focuses on emissions from such appliances.

3.4.1 Indoor sources

3.4.1.1 Gas-fuelled cooking stoves

Several research programmes have investigated NO_x emissions
from stoves fuelled with natural and liquid petroleum gas (Himmel &
DeWerth, 1974; Cote et al., 1974; Massachusetts Institute of
Technology, 1976; Yamanaka et al., 1979; Traynor et al., 1982b; Cole
et al., 1983; Caceres et al., 1983; Fortmann et al., 1984;
Moschandreas et al., 1985; Cole & Zawacki, 1985; Tikalsky et al.,
1987; Borrazzo et al., 1987a). Most of these studies have included
investigations of several other pollutants, including CO, aldehydes
and unburned hydrocarbons. Table 12 lists average emission factors
for range-top burners and for oven and broiler burners operated at
maximum heat input rate. Data are shown for both well-adjusted blue
flames and for poorly adjusted yellow flames. Each of the averages is
based on the total number of stoves tested for that category, using
data from the above studies. For top burners with blue flames, a
total of 27 values are represented; for yellow flames, there are 23
values (24 for NO_x). Averages for the oven and broiler burners
represent 20 blue flame and 16 yellow flame values. Values are
generally very similar for emissions from these two types of burners
on the same stove. Overall, the results show that well-adjusted blue
flames emit more NO but less NO₂ than poorly adjusted yellow flames.
Emission factors from range-top burners are comparable to those from
oven and broiler burners.

Table 12. Average emission factors for nitric oxide (NO),
nitrogen dioxide (NO₂) and nitrogen oxides (NO_x)
from burners on gas stoves

Flame Factor for Factor for Factor for
type NO (µg/kJ) NO₂ (µg/kJ) NO_x (µg/kJ)

Top burners blue 20.0 ± 4.5 10.2 ± 3.1 41.0 ± 8.2
Top burners yellow 16.9 ± 4.5 15.0 ± 4.8 42.0 ± 9.1
Ovens and broilers blue 21.9 ± 6.3 7.23 ± 3.01 40.9 ± 8.6
Ovens and broilers yellow 19.8 ± 9.6 11.4 ± 5.7 39.0 ± 10.8

3.4.1.2 Unvented gas space heaters and water heaters

The findings of several investigators (Thrasher & DeWerth, 1979;
Traynor et al., 1983a, 1984b; Zawacki et al., 1986) are summarized in
Table 13. The most significant result is the markedly lower emissions
from heaters equipped with catalytic burners, radiant ceramic tile
burners and improved-design steel burners (radiant and Bunsen),
compared to emissions from simpler convection designs using
conventional cast-iron Bunsen burners. Equipping convective heaters
with radiant tiles does not make much difference to emission levels,
nor does the choice of natural gas or liquid petroleum gas fuel.
Other studies by Billick et al. (1984), Zawacki et al. (1984) and
Moschandreas et al. (1985) produced similar results.

3.4.1.3 Kerosene space heaters

The data presented in Table 14 show that emission factors of NO
and NO₂ for radiant kerosene heaters are generally much smaller than
those for convective kerosene heaters. Emissions of NO from two-stage
heaters are only slightly greater than those from radiant heaters,
whereas emissions of NO₂ are the lowest of the three heater types.
Most of the emissions from radiant heaters are in the form of NO₂;
for convective heaters that are two-stage heaters, the emissions of NO
and NO₂ are of comparable magnitude. There are insufficient data
to evaluate changes in emissions as kerosene heaters age. Other
products, including particles, present in these emissions may also be
of concern for their possible health effects.

3.4.1.4 Wood stoves

A number of studies have examined pollutant emissions from wood
stoves. Some of these studies have developed emission factors based
on concentrations in the flue gases; such information would be useful
for assessing the contribution of wood stove emissions to ambient air
quality. Very little information is available, however, on fugitive
emissions from wood stoves into the indoor living space.

In a detailed literature survey, Smith (1987) reported that
emissions of pollutants from wood stoves are highly variable,
depending on the type of wood used, stove design, the way the stove is
used and other factors. He reported emission factors for NO_x and
other pollutants for wood stoves used in developing countries. Many
of these stoves are unvented, which results in excessive indoor
concentrations as the combustion products are exhausted into the room.
The major health concerns for wood fires without chimneys arise from
pollutants other than NO₂, such as particulate matter.

Table 13. Summary of studies with unvented convective (C) and infrared (I) space heaters

Type of Number Heat input NO emission NO₂ emission NO_x emission Reference
heater (kJ/min) (µg/kJ) (µg/kJ) (µg/kJ)

Convective 5 86-661 24-47 2.2-7.3 39-77 Thrasher & DeWerth (1979)

Convective 8 188-830 9.5-22 9.5-20 34-47 Traynor et al. (1983a)

Infrared 5 245-352 0.1-1 4.1-6.2 4.9-6.2 Traynor et al. (1984b)
Convective 4 335-626 17.8-28.7 10-18.3 40.1-57.5

Infrared 5 264-334 0.005-1.7 1.6-4.8 2.7-5.7 Zawacki et al. (1986)
Convective 5 176-703 5.3-44.4 7.6-23.3 27.1-76.4

Table 14. Average emission factors for nitric oxide (NO), nitrogen dioxide (NO₂) and nitrogen oxides (NO_x) from kerosene heaters

Type of heater Heat input rate Emission factor Emission factor Emission factor Reference
(kJ/min) for NO (µg/kJ) for NO₂ (µg/kJ) for NO_x (µg/kJ)

Radiant, new 144 0.45 ± 0.05 4.4 ± 0.2 5.1 ± 0.2 Leaderer (1982)
Radiant, new 113 0.08 ± 0.05 5.0 ± 0.2 5.1 ± 0.2
Radiant, new 84.4 0 5.9 ± 0.3 5.9 ± 0.3

Convective, new 158 17 ± 0.3 7.0 ± 0.4 33 ± 0.6
Convective, new 97.9 12 ± 0.6 15 ± 0.3 33 ± 1.0
Convective, new 37.3 11 ± 0.9 17 ± 1.0 34 ± 1.7

Radiant, new 137 1.3 ± 0.7 4.6 ± 0.8 6.6 ± 1.3 Traynor et al. (1983b)

Radiant, 1 year old 111 2.1 5.1 8.3

Convective, new 131 25 ± 0.7 13 ± 0.8 51 ± 1.3

Convective, 5 years old 94.8 11 ± 0.1 32 ± 2.8 49 ± 2.8

Radiant 110-200 - - 13 ± 1.8 Yamanaka et al. (1979)

Convective 110-200 - - 70 ± 6.8

Traynor et al. (1984a) have studied wood stoves (three airtight
and one non-airtight) used in a house. For each experiment, airborne
concentrations of several pollutants were measured inside and outside
the house during operation of one of the stoves. The results showed
that all indoor and outdoor concentrations of NO and NO₂ were
below 0.02 ppm. Moreover, indoor air concentrations of some other
pollutants were high during use of the non-airtight stove. The
airtight stoves had little influence on indoor concentrations of any
pollutants. In another study, Traynor et al. (1982a) found elevated
airborne concentrations of NO and NO₂ in three occupied houses during
operation of wood stoves and a wood furnace. The concentrations were
highly variable.

Because of the limited data, it is difficult to reach
quantitative conclusions regarding the importance of wood stoves.
However, the limited information available suggests that wood stoves
are not a major contributor to indoor nitrogen oxide exposures. This
is consistent with the small NO emission rates expected from the low
temperature combustion processes characteristic of wood stoves.

3.4.1.5 Tobacco products

A number of studies have compared concentrations of NO_x and
other pollutants in houses with smokers and houses without smokers.
In general, these studies have shown that concentrations are somewhat
greater in the homes of smokers.

A few studies have reported emissions of NO_x from cigarettes
while sampling both sidestream and mainstream smoke together.
Woods (1983) reported 0.079 mg NO_x/cigarette, while Moschandreas
et al. (1985) listed emissions of 2.78 mg/cigarette for NO and
0.73 mg/cigarette for NO₂. The National Research Council (1986)
reported total NO_x emissions of 100 to 600 µg/cigarette for
mainstream smoke, with values 4 to 10 times greater for sidestream
smoke. According to the report, virtually all of the emitted NO_x is
in the form of NO; once emitted, the NO is gradually oxidized to NO₂.
Thus environments containing cigarette smoke may have higher
concentrations of both NO and NO₂ than environments without such
smoke. The NO₂ concentration on trains travelling between Changchun
and Harbin, China, was found to be related to the amount of cigarette
smoking, which was greater on daytime trains than on night-time ones.
On a one-way daytime train the average NO₂ concentration was 54 ppb
(range, 37-84 ppb), whereas on a two-way night-time train it was
40.6 ppb (range, 30-59 ppb) (Du et al., 1992).

3.4.2 Removal of nitrogen oxides from indoor environments

A number of field studies of NO₂ levels in residences have
reported that NO₂ is removed more rapidly than can be accounted for
by infiltration alone (Wade et al., 1975; Macriss & Elkins, 1977;

Oezkaynak et al., 1982; Traynor et al., 1982a; Ryan et al.,
1983; Leaderer et al., 1986). Indoors, NO₂ is removed by
infiltration/ventilation and by interior surfaces and furnishings.
The removal of NO₂ by interior surfaces and furnishings and reactions
occurring in air is often referred to as the reactive decay rate of
NO₂, and it can be a significant factor in the actual NO₂ levels
measured in residences. Failure to account for the reactive decay
rate can lead to a serious underestimation of emission rate
measurements in chamber and test house studies and a serious
overestimation of indoor concentrations when using emission rates to
model indoor levels. The NO₂ reactive decay rate is typically
determined by subtracting the decay of NO₂, after a source is shut
off, from that of a relatively non-reactive gas (e.g., CO, CO₂, SF₆,
NO), which can be related to ventilation rates, expressed in room
air changes per hour. The measured reactive decay rates in the
above-mentioned field studies ranged from 0.1 to 1.6 air change
times/hour. All studies noted that the reactive decay of NO₂ is as
important and in some cases more important than infiltration in
removing NO₂ indoors. Leaderer et al. (1986) monitored NO₂, NO, CO
and CO₂ continuously in seven houses over periods ranging from 2 to
8 days. They reported that the NO₂ decay rate was always greater
than that due to infiltration alone and was highly variable among
houses and among time periods within a house.

In an effort to identify the factors that control the NO₂
reactive decay rate, a number of small chamber (Miyazaki, 1984; Spicer
et al., 1986), large chamber (Moschandreas et al., 1985; Leaderer et
al., 1986) and test house studies (Yamanaka, 1984; Borrazzo et al.,
1987b; Fortmann et al., 1987) have been conducted. The most extensive
small chamber work was reported by Spicer et al. (1986), where 35
residential materials were screened for NO₂ reactivity in a 1.64-m³
chamber and a limited number of the materials were tested for the
impact of relative humidity on the reactivity rate. Fig. 12 shows the
relative rates of NO₂ removal for the materials screened. The figure
indicates that many of the materials used for building construction
and furnishings are significant sinks for NO₂ and that their removal
rate is highly variable. Many of the materials were found to reduce a
significant proportion of the removed NO₂ to NO. In no cases was
NO₂ re-emitted, although some materials emitted NO. The authors
noted that the materials that removed NO₂ most rapidly fall in two
categories: (1) porous mineral materials of high surface area; and (2)
cellulosic material derived from plant matter. Higher relative
humidities were found to enhance the removal rate for some materials
(e.g., wool carpet), reduce the removal rate for some (e.g., cement
block), and have little effect on others (e.g., wallboard). In a
series of small (0.69 m³) chamber studies (Miyazaki, 1984) reactive
decay rates for NO₂ were found to vary as a function of material type
and to increase with increasing surface area of the material, degree
of stirring in the chamber, temperature and relative humidity. A
saturation effect was noted on some of the carpets tested.

In a series of large chamber studies (34-m³ chamber), Leaderer
et al. (1986) evaluated the reactive decay rate of NO₂ as a function
of material type, surface area of material, relative humidity and air
mixing. The reactive decay rate was found to vary as a function of
material surface roughness and surface area. Carpeting was found to
be most effective in removing NO₂, whereas painted wallboard was
least effective. Increases in relative humidity were associated with
increases in removal rates for all materials tested, but the slope was
a shallow one. Of particular interest is the finding in this study
that the degree of air mixing and turbulence was a dominant variable
in determining the reactive decay rate for NO₂. Moschandreas et al.
(1985) evaluated six materials in a 14.5-m³ chamber and found
variations in decay rates according to material types and a positive
impact of relative humidity on NO₂ decay rates in an empty chamber.

Yamanaka (1984), in assessing NO₂ reactive decay rates in a
Japanese living room, found the decay to consist of both homogeneous
and heterogeneous processes. The rates were found to vary as a
function of surface property and sharply as a function of relative
humidity. NO production during the decay was noted. In a test house
study, Fortmann et al. (1987) noted that the NO₂ decay rate tends to
decrease as the concentration increases. It is not clear whether this
is due to surface saturation or second-order kinetics. This study
also noted a sharp increase in NO levels during the NO₂ decay,
indicating NO production as a result of the NO₂ decay. In a test
house study conducted over a 7-month period, Borrazzo et al. (1987b)
found that reaction rates for NO₂ in the test house were sensitive to
the location in the house where they were measured. This indicates
that reaction losses during transport of NO₂ from room to room in a
house may be important.

Reactive decay of NO₂ associated with interior surface materials
and furnishings is an important mechanism for removing NO₂ from the
air within homes. Reactive decay rates for NO₂ vary as a function of
the type and surface area of the material. The impact of relative
humidity on the decay rate is unclear, with some studies showing a
pronounced impact (Yamanaka, 1984), while others show only moderate or
little impact (e.g., Spicer et al., 1986; Leaderer et al., 1986). The
degree of air mixing or turbulence can have an important effect on the
reactive decay rate. A by-product of NO₂ removal by materials may be
NO production, and a saturation effect may occur for some materials.
Reactive decay of NO₂ in residences is highly variable between
residences, within rooms in a residence, and on a temporal basis
within a residence. The large number of variables controlling the
reactive decay rate make it very difficult to assess in large field
studies through questionnaire or integrated air sampling.

3.5 Indoor concentrations of nitrogen oxides

Indoor concentrations of NO₂ are a function of outdoor
concentrations, indoor sources (source type, condition of source,
source use, etc.), infiltration/ventilation, air mixing within and
between rooms, reactive decay by interior surfaces, and air cleaning
or source venting.

3.5.1 Homes without indoor combustion sources

Typical studies in homes without indoor sources of NO₂,
summarized in Table 15, have reported concentrations lower than
outdoor levels due to removal from the air of NO_x by the building
envelope and interior surfaces. Thus indoor/outdoor concentration
ratios are consistently less than unity. These homes provide some
degree of protection from outdoor concentrations. Indoor/outdoor
ratios vary considerably according to the season of the year, the
lowest ratios occurring in the winter and highest occurring during the
summer. Although urban concentrations are often highest in winter,
this pattern in the indoor/outdoor ratio, attributed to seasonal
differences in infiltration rates, NO₂ reactivity rates, the
penetration factor and outdoor concentrations, can result in higher
indoor concentrations in summer than in winter. The indoor-to-outdoor
ratio for these homes does not appear to depend on geographical area,
housing type or outdoor concentration. Results of monitoring in
Portage, Wisconsin, USA, show that the presence of a gas stove
contributes dramatically to the indoor NO₂ levels. Table 16, taken
from the report of Quackenboss et al. (1986) and based on data
collected in 1981 and 1982, clearly shows that gas stoves increase not
only indoor concentrations but also the personal exposure of children.

3.5.2 Homes with combustion appliances

It is estimated that gas (natural gas and liquid propane) is used
for cooking, heating water or drying clothes in about 45% of all homes
in the USA (US Bureau of the Census, 1982) and in nearly 100% of homes
in some other countries (e.g., the Netherlands). Gas appliances
(gas cooker/oven, water heater, etc.) are the major indoor source
category for indoor residential NO₂ by virtue of the number of homes
with such sources. NO₂ concentrations in homes with gas appliances
are higher than those without such appliances. Within this category,
the gas cooker/oven and unvented heaters are by far the major
contributors. Cookers and ovens are especially important sources when
used inappropriately as a supplementary room heater. Average indoor
concentrations (based on a 1- to 2-week measurement period) in excess
of 100 µg/m³ have been measured in some homes with gas cookers
(Table 17). Homes where gas cookers with pilot lights are used have

higher NO₂ levels than homes that have gas cookers without pilot
lights. Average NO₂ concentrations in homes with gas cookers/ovens
exhibit a spatial gradient within and between rooms. Kitchen
concentrations of NO₂ are higher than other rooms and a steep
vertical concentration gradient in the kitchen has been observed in
some homes, concentrations being highest nearest the ceiling. Average
NO₂ concentrations are highest during the winter months and lowest
during the summer months. This seasonal temporal gradient is
attributed to differences in infiltration, appliance use, NO₂
reactivity rates and indoors and outdoor concentrations. The impact
of gas appliance use on indoor NO₂ levels may be superimposed upon
the background level resulting from outdoor concentrations. Only very
limited data exist on short-term average (3 h or less) indoor
concentrations of NO₂ associated with gas appliance use. These data
suggest that short-term average concentrations of NO₂ are several
times the longer-term average concentrations measured.

A wide variety of fuel types can be used for cooking and heating
in different localities. These can produce various effects on indoor
air quality. As an example, Table 18 gives data for indoor NO_x
concentrations measured at Lanzhou City, China, where coal and
liquified gas were used in apartments and houses (Duan et al., 1992).

Table 15. Average outdoor concentrations of nitrogen dioxide (NO₂) and average indoor/outdoor ratios in homes without gas appliances or
unvented space heaters^a

Location Housing Averaging Seasons Number Average NO₂ Indoor/outdoor ratios Reference
type^b time of outdoor
homes concentration
(µg/m³) Kitchen Bedroom

Southern California Mixed 7 days Summer 70 71.9 0.80 0.75 Southern California
Spring 100 43.5 0.72 0.60 Gas Company (1986)
Winter 69 91.2 0.56 0.47

New Haven, CT Single family 14 days Winter 60 13.2 0.56 0.55 Leaderer et al. (1986)
unattached

Albuquerque, NM Mixed 14 days Winter 1 60 14.1 - 0.50 Marbury et al. (1988)
Winter 2 56 19.6 - 0.32

California Mobile homes 7 days Summer 46 25.9 0.61 0.54 Petreas et al. (1988)
Winter 23 44.6 0.27 0.26

Portage, WI Mixed 7 days Summer 47 15.2 0.91 0.72 Quackenboss et al. (1986)
Winter 47 17.2 0.65 0.45

Tucson, AZ Mixed 14 days Summer 56 19.9 0.86 0.76 Quackenboss et al. (1986)
Spring/Autumn 41 25.6 0.71 0.55
Winter 23 36.8 0.64 0.52

Boston, MA Mixed 14 days Summer 117 31.7 0.76 0.75 Ryan et al. (1988)
Autumn 117 37.8 0.43 0.40
Winter/Spring 124 33.5 0.53 0.47

Table 15. (Con't)

Location Housing Averaging Seasons Number Average NO₂ Indoor/outdoor ratios Reference
type^b time of outdoor
homes concentration
(µg/m³) Kitchen Bedroom

Northern Central Single family 5 days Winter 9 53.8 Koontz et al. (1986)
Texas unattached

Suffolk County, Single family 7 days Winter 49 35.5 0.47 - Research Triangle
NY unattached Institute (1990)

Onondago County, Single family 7 days Winter 66 21.7 0.70 -
NY unattached

Portage, WI Single family 7 days Average over 25 12.8 0.65 0.51 Spengler et al. (1983)
unattached all seasons

Watertown, MA Not given 3-4 days November 18 37.0 0.65 0.51 Clausing et al. (1984)
December 10 46.0 0.39 0.30

Middlesbrough, UK Not given 7 days Winter 87 35.0 0.97 0.75 Goldstein et al. (1979)

Middlesbrough, UK Not given 7 days Winter 15 34.7 - 0.75 Melia et al. (1982a,b)

^a Data from field studies of private residences in the USA and United Kingdom
^b "Mixed" indicates a single family in an attached or unattached dwelling, condominium or apartment
Table 16. Nitrogen dioxide concentrations (ppm) according to season and
stove type in Portage, Wisconsin, USA^a

Season Stove Indoor Outdoor Personal

Mean SD Mean SD Mean SD

Summer Gas 0.016 0.006 0.006 0.003 0.014 0.004
Electric 0.007 0.003 0.008 0.003 0.009 0.003

Winter Gas 0.027 0.013 0.008 0.003 0.023 0.009
Electric 0.005 0.003 0.009 0.003 0.008 0.003

^a From: Quackenboss et al. (1986); SD = standard deviation
Table 17. Indoor and outdoor concentrations of nitrogen dioxide (NO₂) in homes with gas appliances, and the calculated average
contribution of those appliances to indoor residential NO₂ levels

Location Housing Averaging Type of Season No. of Average measured NO₂ Indoor NO₂ due to source Reference
type^a time appliance homes (µg/m³) (µg/m³)
(days)
Outdoors Kitchen Bedroom Other Kitchen Bedroom Other Comment^b

USA

Southern Mixed 7 Oven/range, Summer 147 75.3 91.6 68.4 - 31 12 - 1,2 Southern
California ± pilot Spring 202 49.2 79.2 51.3 - 35 22 - 1,2 California
Winter 141 104 101.5 69 - 48 20 - 1,2 Gas Company
(1986)
Oven/range, Winter 98 107 113 76 - 53 26 - 1,2
pilot

Oven/range, Winter 38 97 74 53 - 20 7 - 1,2
no pilot

Water heater Winter 21 92 59 50 - 11 11 - 1,2,3
in home

Wall furnace Winter 90 121 161 113 - 49 36 - 1,4

Floor Summer 42 119 177 126 - 66 52 - 1,4
furnace

Table 17. (Con't)

New Haven, Single 14 Oven/range, Winter 42 14.8 44.7 27.6 30.4 36 20 22 1,5 Leaderer
CT family, ± pilot et al.
unattached (1986)

Albuquerque, Mixed 14 Oven/range, Winter 82 19.1 - 33.1 41.9 - 24 31 1,5,6 Marbury et
NM ± pilot Winter 75 20.3 - 30.9 39.3 - 24 32 al. (1988)

California Mobile 7 Oven/range, Summer 265 21.1 43.1 30.2 - 30 19 - Petreas et
homes ± pilot Winter 231 42.1 53.7 37.5 - 42 27 - 1,7 al. (1988)

Portage, Mixed 7 Oven/range, Summer 36 11.5 38.9 21.1 29.6 29 13 20 Quackenboss
WI ± pilot Winter 34 15.4 69.6 31.2 50.7 60 15 42 1,8 et al.
(1986)

Tucson, Mixed 14 Oven/range, Summer 13 23.1 39.1 26.3 30.7 19 8 11 Quackenboss
AZ ± pilot Spring/ 11 36.3 45.8 31.9 42.4 20 12 17 et al.
Autumn (1986)
Winter 10 45.2 60.6 43.4 50.7 32 20 25 1,9

Boston, Mixed 14 Oven/range, Summer 301 41.6 65.9 45.6 50.9 33 15 19 Ryan et al.
MA ± pilot Autumn 277 43.7 74.3 47.5 52.8 56 30 34 (1988)
Winter/ 298 40.5 73.5 48.6 55.1 52 30 34 1,9
Spring

Table 17. (Con't)

Central Single 5 Oven/range, Winter 22 34.6 - - 54.1 - - 37 1,10 Koontz et
Texas family, ± pilot al. (1986)
unattached

Suffolk Co., Single 7 Oven/range, Winter 42 37.6 77.5 - 52.4 60 - 37 Research
NY family, ± pilot Triangle
unattached Institute
(1990)

Onondago Single 7 Oven/range, Winter 56 30.6 62.6 0 50 41 - 27 1,9
Co., NY family, ± pilot
unattached

New York, Apartments 2 Oven/range Summer 14 109 122 98 106 30 6 13 Goldstein
NY Autumn 1 15 61 96 65 71 53 22 18 et al.
Autumn 2 9 73 108 66 76 45 15 25 (1985)
± pilot Winter 1 8 100 121 76 95 61 16 35
Winter 2 18 75 126 63 82 81 18 37 9,11,12
Spring 13 95 121 82 99 55 16 33

Portage, WI Single 7 Natural gas All 36 15.8 65.5 36.7 - 55 29 - Spengler et
family, Oven/range, seasons al. (1983)
unattached no pilot

Liquified All 76 11.6 65.6 37.6 - 58 31 - 1,13
petroleum seasons
gas
Oven/range,
no pilot

Table 17. (Con't)

Watertown, Not given 3-4 Gas cooking Novemb. 60 37 74 45 51 50 26 33 1,9,14 Clausing et
MA Decemb. 51 46 86 46 60 68 32 44 al. (1984)

Netherlands

Arnet Not given 7 Gas cooking Autumn/ 294 35 118 - 97 97 - 37 Noy et al.
Enschede no pilot Winter (1984)
Water
heaters

Ede Not given 7 Gas cooking Autumn/ 173 44 113 43 54 89 17 28 Noy et al.
no pilot Winter (1984)
Water
heaters

Vlagttwedde Rural area 7 Gas cooking Autumn/ 162 28 107 24 51 90 7 34
no pilot Winter Water
heaters

Rotterdam I, Inner city 7 Gas cooking Autumn/ 228 45 144 51 80 117 24 53
no pilot Winter
Water
heaters

Rotterdam II, Inner city 7 Gas cooking Autumn/ 102 45 143 64 73 117 37 46 9,17
no pilot Winter Water
heaters

Table 17. (Con't)

United Kingdom

Middlesbrough Not given 7 Gas cooking Winter 428 35 213 58 - 179 24 - 1,15 Goldstein
no pilot et al.
(1979)

Middlesbrough Not given 7 Gas cooking Winter 183 34.7 - 60 82.7 - 39 61 1,16 Melia et
al.
(1982a,b)

^a "Mixed" indicates a single family in an attached or unattached dwelling, condominium or apartment
^b 1. Background correction determined by multiplying: (a) the indoor/outdoor ratio for homes in the study with no indoor NO₂ sources
for a given season; by (b) the outdoor NO₂ concentration measured for the home with sources; and subtracting the product from
the indoor level measured in the house.
2. Homes containing forced air gas furnace. These homes are thought not to contribute significantly to indoor levels for this
sample.
3. Homes with electric cooker/oven, forced air gas furnace, and gas water heater in home. Comparison is made with electric
cooker/oven, forced air gas furnace, and gas water heater located outside home.
4. Homes have gas cooker/oven with source contribution calculated after correction of a gas cooker/oven. Values are background
corrected with gas stove.
5. Living room or activity room.
6. Sampling was done over two different periods for the same houses within the same winter period.
7. Outdoor values were obtained from five locations, housing type, mobile home.
8. Other location in home; bedroom refers to average of levels in one or more bedrooms in house.
9. Other location in the main living room.
10. Other location is point nearest centre of home.

Table 17 (Con't)

11. 48-h samples over 30 consecutive days.
12. Indoor/outdoor (I/O) ratio is assessed to be 0.6, 0.7, and 0.85 for the Winter, Spring/Autumn and Summer periods,
respectively, for all locations, because no control home (no gas appliances) mean measurements were available. Using these
I/O ratios, the impact of sources was calculated as footnote 1.
13. Each home was sampled six times over a 1-year period.
14. Outdoor levels are average for homes with or without gas appliances.
15. Outdoor levels were recorded at 75 locations in the general sampling area and were not home-specific. Bedroom levels were
obtained for 107 of the 428 homes.
16. Outdoor levels were recorded at 82 locations in the general sampling areas and were not home-specific. Outdoor levels were
recorded at the beginning and end of the study.
17. Indoor/outdoor (I/O) ratio is assumed to be 0.6 for all locations, because no control home (no gas appliances) measurements
were available. Using I/O ratio of 0.6, the impact of sources was calculated as in footnote 1.
Table 18. Indoor concentration of NO_x in Lanzhou city, China^a

Type of residence Average NO_x
concentration (mg/m³)

Winter Summer

Apartment building with central 0.141 0.059
heating, liquified gas for cooking

Apartment building without central 0.136 0.059
heating, coal for cooking and heating

One-storey house, coal for cooking 0.106 0.046
and heating

^a From: Duan et al. (1992)

3.5.3 Homes with combustion space heaters

Unvented kerosene and gas space heaters are important sources of
NO and NO₂ in homes, both because of the NO and NO₂ production rates
of the heaters and the long periods of time that they are in use. The
concentrations of NO emitted are usually several times higher than
those of NO₂. However, in the literature, indoor air concentrations
of NO are frequently not reported.

Field studies indicate that average residential concentrations
(1- or 2-week average levels) exhibit a wide variation, depending
primarily on the amount of heater use and the type of heater
(Table 19). Under similar operating conditions, unvented gas space
heaters appear to be associated with higher indoor NO₂ concentrations
than kerosene heaters. Average concentrations in homes using unvented
kerosene heaters have been found to be well in excess of 100 µg/m³.
In one study (Leaderer et al., 1986), calculations of NO₂
concentrations in residences during kerosene heater use (in homes
without gas appliances) indicate that approximately 50% of the homes
have NO₂ concentrations above 100 µg/m³ and 8% above 480 µg/m³. A
peak NO₂ concentration of 847 µg/m³ was measured over a 1-h period
in a home during use of a kerosene heater.

Table 19. Two-week average nitrogen dioxide (NO₂) levels for homes
in New Haven, Connecticut, USA, during winter, 1983^a

Source category; NO₂ (µg/m³)
location

n Mean SD^b % above
100 µg/m³

No kerosene heater
or gas stove
Outdoors 144 13.2 5.3 0
House average 145 7.4 4.2 0
Kitchen 147 7.6 3.7 0
Living room 146 7.3 3.4 0
Bedroom 145 7.3 8.6 0

One kerosene heater,
no gas stove
Outdoors 95 12.9 4.6 0
House average 95 36.8 32.8 2.1
Kitchen 96 39.1 35.5 4.2
Living room 96 38.5 36.6 5.2
Bedroom 95 31.9 30.8 5.3

No kerosene heater,
one gas stove
Outdoors 42 14.8 4.2 0
House average 42 34.3 26.2 4.8
Kitchen 42 44.7 31.4 4.8
Living room 42 30.4 24.8 4.8
Bedroom 42 27.8 25.1 4.8

One kerosene heater,
one gas stove
Outdoors 18 14.5 5.2 0
House average 18 66.8 43.9 16.7
Kitchen 18 74.5 52.0 22.2
Living room 18 57.4 38.6 11.1
Bedroom 18 68.5 56.5 16.7

Two kerosene heaters,
no gas stove
Outdoors 13 16.5 9.4 0
House average 13 69.5 38.0 23.0
Kitchen 13 73.0 31.7 23.0
Living room 13 73.6 44.3 38.5
Bedroom 13 67.8 44.9 23.1

Table 19. (Con't)

Source category; NO₂ (µg/m³)
location

n Mean SD^b % above
100 µg/m³

Two kerosene heaters,
one gas stove
Outdoors 3 22.1 6.2 0
House average 3 85.8 24.4 33.3
Kitchen 3 94.0 22.7 66.6
Living room 3 77.6 38.4 33.3
Bedroom 3 85.8 19.5 33.3

^a From: Leaderer et al. (1986); repeat monitoring data (n = 19)
are included
^b SD = standard deviation

A large field study (Koontz et al., 1986) of indoor NO₂
concentrations in Texas homes using unvented gas space heaters (most
also had gas cookers) found that approximately 70% of the homes had
average concentrations in excess of 100 µg/m³ and 20% had average
concentrations in excess of 480 µg/m³. This study found that the
indoor/outdoor temperature difference was the best indicator of
average indoor NO₂ levels during the colder winter periods when
heating demands are greatest.

Only limited data have so far been published on short-term peak
indoor concentrations for homes with unvented gas space heaters, and
no data are available on spatial variations or concentrations solely
during the hours of heater operation.

No spatial gradient of NO₂ was found in homes with unvented
kerosene space heaters, contrary to the strong spatial gradient noted
for homes with gas appliances. This is probably due to the strong
convective heat output and the long operating hours of the heaters,
which result in rapid mixing within the homes.

Ferrari et al. (1988) conducted a study of air quality in
homes with unvented space heaters in Sydney, Australia, over
two winters. NO₂ concentrations were measured by both continuous
(chemiluminescence with O₃ method) and passive monitors (badges and

Palmes tubes). Concentrations of NO₂ exceeded 0.10 ppm (average
concentration) in 85% of homes tested, and 0.16 ppm in 44% of homes.
More than 10% of homes had average NO₂ concentrations exceeding
0.32 ppm, and the maximum recorded was greater than 0.5 ppm. Unvented
gas space heaters are common in Sydney, and average use is about 3 h
per night during the winter. As a result, an estimated 0.5 million
residents are exposed to NO₂ concentrations exceeding 0.16 ppm for
several hours per night during the colder months of the year.

Improper use of gas appliances (e.g., using a gas oven or
stove to heat a living space) and improperly operating gas appliances
or vented heating systems (e.g., out-of-repair gas cooker or
improper operation of a gas wall or floor furnace) can be important
contributors to indoor NO₂ concentrations, but few data are available
to assess the magnitude of that contribution. Little or no field data
exist that would allow for an assessment of the contributions of wood-
or coal-burning stoves or fireplaces to indoor NO₂ concentrations,
but such a contribution would be expected to be small. Cigarette
smoking is expected to add relatively small amounts of NO₂ to homes
(see also Tables 15 and 18).

In developing countries, biomass fuels (e.g., wood, biogas,
animal dung, etc.) are much more widely used for home heating and/or
cooking than in developed countries, these fuels often being burnt in
open hearth fires or poorly vented appliances within indoor spaces of
residential dwellings (WHO, 1992). As noted by Sims & Kjellström
(1991), a very conservatively estimated 400 million people are
affected by biomass smoke problems worldwide, mostly in rural areas of
developing countries. A disproportionate number of women and young
children are exposed, owing to the greater numbers of hours typically
spent by them indoors and their involvement in cooking and other
household chores. Increased NO_x concentrations, as well as greater
concentrations of carbon monoxide, suspended particulate matter (SPM)
and volatile organic compounds (VOCs) are normally found in biomass
smoke (Chen et al., 1990). Reviews of field studies in rural areas of
developing countries indicate that exposure levels to biomass smoke
components are usually rather high. Indoor concentrations for NO₂,
for example, were found to fall within the range of 0.1 to 0.3 mg/m³
in India, Nepal, Nigeria, Kenya, Guatemala and Papua New Guinea, as
reported in reviews by WHO (1984) and Smith (1986, 1987). Similarly,
Hong (1991) reported NO₂ concentrations in the range of 0.01 to
0.22 mg/m³ resulting from indoor combustion of biogas in homes in
Chengdu, Szechuan Province, China. Hong (1991) also reported NO_x
concentrations in the range of 0.02 to 0.16 mg/m³ in other houses in
Gansu Province, China, where dried cow dung was used as a fuel. The
above NO₂ indoor air concentrations from biomass smoke should be
compared with the WHO Air Quality Guidelines recommendation of
0.15 mg/m³ for daily exposures to NO₂ (WHO, 1987).

3.5.4 Indoor nitrous acid concentrations

Recent studies have demonstrated that substantial concentrations
of HNO₂ can be present inside residential buildings, especially when
unvented combustion sources are used. HNO₂ is formed by the reaction
of NO₂ with water on surfaces and the reaction is promoted by high
humidity. HNO₂ may also be produced by other mechanisms, and this is
the subject of active research. Brauer et al. (1993) found that HNO₂
can represent over 10% of the concentrations usually reported as NO₂.

3.5.5 Predictive models for indoor NO₂ concentration

Efforts to model indoor NO₂ levels have employed two distinct
approaches: physical/chemical and empirical/statistical models.

The physical/chemical modelling approach has been used by
numerous investigators in chamber, test house and small field studies
(involving a small number of homes) to estimate emission rates of NO₂
from combustion sources (e.g., Traynor et al., 1982a; Leaderer, 1982;
Moschandreas et al., 1984), to estimate reactive decay rates (e.g.,
Oezkaynak et al., 1982; Yamanaka, 1984; Leaderer et al., 1986; Spicer
et al., 1986; Borrazzo et al., 1987a), to estimate the impact of
ventilation and mixing on the spatial and temporal distribution of
NO₂ (e.g., Oezkaynak et al., 1982; Traynor et al., 1982b; Borrazzo
et al., 1987a), and to evaluate the applicability of emission
rates determined under controlled conditions in estimating indoor
concentrations of NO₂ (e.g., Traynor et al., 1982b; Borrazzo et al.,
1987a). More recently, studies have reported the use of distributions
of the input variables to the mass balance equation (emission rates,
source use, decay rates, ventilation rates, etc.), determined from the
published literature, to estimate distributions of indoor NO₂ levels
for specific sources and combinations of sources (Traynor et al.,
1987; Hemphill et al., 1987).

Prediction of indoor concentrations or concentration
distributions of NO₂ in homes with combustion sources using
physical/chemical (mass-balance) models requires accurate information
for input parameters (e.g., emission rates). Although data are
available for some of the input parameters under controlled
experimental conditions, there are very limited data available
concerning either the variability of such input parameters in actual
homes or the factors that control variability (e.g., variability of
emission or decay rates). Obtaining field measurements or estimates
of the inputs in large numbers of homes would be expensive and
time-consuming. Such modelling efforts do, however, help to identify
the potential range of indoor NO₂ concentrations, factors that may
result in high levels, and the potential effectiveness of mitigation
efforts.

Empirical/statistical models have been developed from large field
studies that have measured NO₂ concentrations in residences and
associated outdoor levels for time periods of a week or more. These
have typically used questionnaires to obtain information on sources in
the residences, source use, building characteristics (house volume,
number of rooms, etc.), building use, and meteorological conditions.
In some cases, additional measurements, including temperature have
been recorded. Several investigators have attempted to fit simple
regression models to their field study databases in an effort to
determine whether the variations in NO₂ levels seen among houses can
be explained by variations in questionnaire responses. The goal has
been to see how well questionnaire information or easily available
information (meteorological data) can predict indoor NO₂ levels. In
most cases a linear model has been used, but several investigators
have used log transformations of variables. These employ
questionnaire responses and measured physical data (house volume,
etc.) as independent variables and have met with moderate success.
Linear regression models, with the exception of the Petreas et al.
(1988) model, explain from 40 to 70% of the variations in residential
NO₂ levels and typically have large standard errors associated with
their estimates. Although log transformations of variables have
always produced a higher percentage of explained variation due to the
skewed distribution of the original variables, interpretation of the
coefficients in a nonlinear model can require special attention.

Regression models developed from field studies employing
questionnaires to explain variations in indoor levels of NO₂ have met
with only moderate success.

Better information, through additional measurements and better
questionnaire design, is needed on a range of factors if the
statistical/empirical models are to be used to estimate indoor
concentrations of NO₂ in homes without measurements. Factors include
source type and condition, source use, contaminant removal
(infiltration and reactive decay) and between and within room mixing.

3.6 Human exposure

To assess the health impact of exposure to nitrogen oxides, it is
essential to conduct an accurate exposure assessment. Such data are
of paramount importance for the definition of dose-effect and
dose-response relationships. In fact, the risk to human health is not
simply determined by indoor and outdoor concentrations of nitrogen
oxides, but rather by the personal exposure of every individual. The
integrated exposure is the sum of the individual exposures to oxides
of nitrogen over all possible time intervals for all settings or
environments. It requires, thus, the consideration of long-term

average concentration level, variations and short-term exposures, as
well as the activity patterns and personal and home characteristics of
individuals (Berglund et al., 1994).

Exposure is a function of concentration and time. People spend
various periods in different types of micro-environments with various
concentration levels. On average, people spend about 90% of their
time indoors (at home, work, school, etc.), about 5% in transit
(Chapin, 1974), and 7% (range 3-12%) near smokers (Quackenboss et al.,
1982). These values vary with the season, day of the week, age,
occupation and other factors but it is decidedly important to predict
indoor pollutant levels when total exposure is being estimated.

Adequate exposure assessment for NO₂ is particularly critical in
conducting and evaluating epidemiological studies. Failure to measure
or estimate exposures adequately and address the uncertainty in the
measured exposures can lead to misclassification errors (Shy et al.,
1978; Gladen & Rogan, 1979; Oezkaynak et al., 1986; Willett, 1989;
Dosemeci et al., 1990; Lebret, 1990). Early studies comparing the
incidence of respiratory illness in children living in homes with and
without gas stoves used a simple categorical variable of NO₂
exposure; the presence or absence of a gas cooker. Such a dichotomous
grouping can result in a severe non-differential misclassification
error in assigning exposure categories. This type of error is likely
to underestimate the true relationship and could possibly result in a
null finding.

In assessing human exposure to NO₂ (and other oxides of
nitrogen), averaging times chosen should account for the type of
effect to be expected. With regard to NO₂, the principal biological
responses include (a) relatively transient effects on respiratory
function associated with acute, short-term (< 1 h) exposures, and (b)
the likelihood of increased risk for respiratory disease in children
associated with frequently repeated short-term peak exposures and/or
lower level long-term exposures.

Indirect and direct methods for personal exposure assessment are
available. Indirect methods combine measures of concentrations at
fixed sites in various types of micro-environments with information
on where people have spent their time (time-activity patterns).
Time-weighted average (TWA) exposure models have been developed to
estimate total personal exposure (Fugas, 1975; Duan, 1982; Duan,
1991). The NO₂ exposure levels predicted from TWA exposure models
have been shown to correlate closely with the exposure levels obtained
by direct measurements of personal exposure (Nitta & Maeda, 1982;
Quackenboss et al., 1986; Sega & Fugas, 1991). However, the large
variation in NO₂ concentrations (distribution) within each type of

micro-environment (because of variability in, for example, stove use,
emission rates, ventilation frequencies, and the day-to-day and
person-to-person variations in the use of time) decreases the accuracy
of the predicted exposure and increases the risk for misclassification
of the exposure.

Direct measurements of concentrations in the breathing zone
of a person using personal passive exposure monitors provide
time-integrated measurements of exposure for a certain period across
the various micro-environments where a person spends time. It is
important to collect exposure data over time intervals consistent with
the expected effects. Effects from long-term, low-level exposure may
be different from effects from short periods of high concentration
(intermittent peak exposure). Intermittent peak exposure, which
occurs during cooking on a gas stove, may be significant to total
exposure and adverse health effects. If effects from peak exposure
are to be considered in the exposure assessment, the sampling time
must be short enough to detect these peak exposures. Such a short
sampling time is possible with the more sensitive passive samplers and
with conventional air monitors, such as chemiluminescence NO_x
monitors. However, direct methods of measuring personal exposure
are relatively costly and time-consuming. Within-person and
between-person variability, both in personal exposure and personal
use of time, can be large.

Hence a sufficient number of personal exposure measurements must
be collected for each person (repeated measurements), and a sufficient
number of individuals must be sampled before the measurements can be
considered to be representative. Personal daily exposures have been
shown to vary between individuals on the same day by a factor of up to
about 15 in the urban area of Stockholm and between days for the same
individual by a factor of up to 10 (Berglund et al., 1993).

A comparison of personal NO₂ exposures, as measured by Palmes
diffusion tubes, and NO₂ exposures measured in residences had a
correlation of 0.94 for a subsample of 23 individuals (Leaderer et
al., 1986). Results of this comparison are depicted in Fig. 13 and
show an excellent correlation between average household exposure and
measured personal exposure.

It is important to note that indoor concentrations are strong
predictors of personal exposure. In the case of homes with gas or
electric stoves, personal exposure has been shown to be closely
related to the household indoor average concentrations (Quackenboss
et al., 1986; Harlos et al., 1987a).

Results of monitoring in Portage, Wisconsin, verify that the
presence of a gas stove contributes dramatically to personal NO₂
exposure levels. Table 16, derived from the reports of Quackenboss et
al. (1986) and based on data collected in 1981 and 1982, clearly shows
that gas stoves increase not only indoor concentrations but also the
personal exposure of children.

On the other hand, outdoor concentrations, even if measured
outside each residence, have been found to be relatively poor
predictors of personal exposure (Quackenboss et al., 1986; Leaderer et
al., 1986). The association between personal exposure and outdoor
levels of NO₂ is weakest during the winter for both gas and electric
stove groups.

The only route of NO₂ exposure is inhalation. The dose is
dependent on the inhalation volume and thus on physical activity, age,
etc. Lung absorption of NO₂ is about 80-90% during rest and over 90%
during physical activity (WHO, 1987).

Efforts have been made to find a sufficient biological marker for
NO₂ exposure and dose. Increased urinary excretion of collagen and
elastin (pulmonary connective tissue) breakdown products (including
hydroxyproline, hydroxylysine and desmosine) has been suggested as a
marker of diffuse pulmonary injury related to inhaled NO₂. A
significant relationship between urinary hydroxyproline excretion and
daily NO₂ exposure was found among housewives in Japan, but the
hydroxyproline excretion fell within the normal distribution for
healthy people (Yanagisawa et al., 1986). The majority of the
housewives were exposed to active or passive cigarette smoke, and this
exposure was independently related to the excretion of hydroxyproline.
Other investigators have not been able to substantiate the
relationship between urinary hydroxyproline excretion and NO₂
exposure (Muelenaer et al., 1987; Adgate et al., 1992).

Measurements of the NO-haem protein complex in bronchoalveolar
lavage (Maples et al., 1991) and of 3-nitrotyrosine in urine (Oshima
et al., 1990) have been suggested as biological markers for NO₂
exposure. The work in progress to find a suitable biological marker
for NO₂ exposure at levels found in the general environment is
promising; however, no metabolite has yet proved to be sensitive or
specific enough.

Personal exposure to air pollutants can be assessed by direct or
indirect measures. Direct measures include biomarkers and use of
personal monitors. No validated biomarkers for exposure are presently
available for NO₂.

Studies using passive monitors to measure NO₂ exposures lasting
one day to one week have been conducted in the USA (Dockery et al.,
1981; Clausing et al., 1986; Leaderer et al., 1986; Quackenboss et
al., 1986; Harlos et al., 1987; Schwab et al., 1990), in the
Netherlands (Hoek et al., 1984), in Japan (Nitta & Maeda, 1982;
Yanagisawa et al., 1984), and in Hong Kong (Koo et al., 1990). These
studies generally indicate that outdoor levels of NO₂, although
related to both personal levels and indoor concentrations, are poor
predictors of personal exposures for most populations. Average indoor
air residential concentrations (e.g., whole-house average or bedroom
level) tend to be the best predictor of personal exposure, typically
explaining 50 to 80% of the variation in personal exposures.

Indirect measures of personal exposure to NO₂ employ various
degrees of micro-environmental monitoring and questionnaires to
estimate an individual's or population's total exposure. One such
model (Billick et al., 1991), developed from an extensive monitoring
and questionnaire database, can estimate population exposure
distributions from easily obtained data on outdoor NO₂ concentrations,
housing characteristics and time-activity patterns. This model is
proposed for use in evaluating the impact of various NO₂ mitigation
measures. The model is promising, but has not yet been validated nor
has associated uncertainty been characterized.

3.7 Exposure of plants and ecosystems

The sensitivity of plants to nitrogen oxides is determined both
by their genetic characteristics and by environmental conditions.

The relation between exposure and uptake by plants depends on
aerodynamic and stomatal resistance and thus increases with increasing
light intensity, wind velocity and air humidity. After uptake, the
response of a plant depends on its metabolic activity, and thus
increases with poorer nutritional supply and lower temperature.

Moreover, the sensitivity of plants depends on the general air
pollution situation. Emission of SO₂ is often combined with NO_x,
and these compounds act synergistically. Therefore, the impact of
NO_x may be higher in regions with elevated SO₂ concentrations. NO_x
forms part of photochemical smog. Although ozone is the most
phytotoxic, the contribution of NO_x to adverse effects on plants is
not negligible.

For vegetation and ecosystems the impact of NO_x is through its
contribution to total nitrogen disposition rather than its direct
toxicity. Thus, other nitrogen-containing pollutants have to be taken
into consideration.

The dependencies of sensitivity, as summarized above, mean that
wide variation exists in the vulnerability of different regions of the
world.

4. EFFECTS OF ATMOSPHERIC NITROGEN COMPOUNDS (PARTICULARLY NITROGEN
OXIDES) ON PLANTS

Effects of nitrogen on ecosystems are caused through deposition
onto soil and foliar uptake of nitrogen in various forms. Total
effects are often difficult to separate into component effects. This
section, therefore, covers nitrogen inputs in all forms to ecosystems.
Much of the research focuses on European ecosystems, where the
majority of the research has been conducted. Here NH_y deposition
either dominates or is a major constituent of total nitrogen input.
However, this is not true for other parts of the world. All effects
of atmospheric nitrogen input, in whatever form, are included as
indicators of more globally relevant effects on ecosystems but the
reader should bear in mind local conditions of nitrogen input when
assessing likely local consequences.

NO_x, as used in this chapter, refers to the total nitrogen
measured by chemiluminescence detectors; this is NO₂ following
conversion to NO, and NO itself. Other nitrogen oxides may interfere
to some extent in this method.

Elemental nitrogen (N₂) forms 80% of the atmosphere of the
earth. This is equivalent to about 75 × 10⁶ kg above each hectare of
the earth's surface. In unpolluted conditions a small fraction
(1-15 kg nitrogen per ha per year) is converted by nitrogen-fixing
microorganisms to biologically more active forms of nitrogen: NH₄⁺
and NO₃^-. The natural deposition of nitrogen-containing atmospheric
compounds other than N₂ is much less. The soil contains 5 times more
nitrogen than the atmosphere, but weathering of rock is a negligible
source of biologically active nitrogen. By denitrification (reduction
of NO₃^- to N₂ and to a lesser extent N₂O, NO and NH₃), 1-30 kg
nitrogen per ha per year is recycled from the earth to the atmosphere.

Human activities, both industrial and agricultural, have greatly
increased the amount of biologically active nitrogen compounds,
thereby disturbing the natural nitrogen cycle. Various forms of
nitrogen pollute the air, mainly NO, NO₂ and NH₃ as dry deposition
and NO₃^- and NH₄⁺ as wet deposition. Another contribution
is from occult deposition (fog and clouds). There are many more
nitrogen-containing air pollutants (e.g., N₂O₅, PAN, N₂O, amines)
but these have not been considered in this chapter, either because
their contribution to the total nitrogen deposition is considered to
be small or because their concentrations are probably far below the
effect thresholds.

Transformations of nitrogen, as it moves from the atmosphere,
through ecosystems and back to the atmosphere, form the nitrogen
cycle. This is illustrated, together with anthropogenic sources of
nitrogen, in Fig. 14. The component processes affected by chronic
nitrogen deposition are indicated in Fig. 15.

Nitrogen-containing air pollutants can affect vegetation
indirectly, via chemical reactions in the atmosphere, or directly
after being deposited on vegetation, soil or water surfaces. The
indirect pathway is largely neglected in this chapter, although it
includes very relevant processes, and should be taken into account
when evaluating the entire impact of nitrogen-containing air
pollutants: NO and NO₂ are precursors for tropospheric ozone (O₃),
which acts both as a phytotoxin and a greenhouse gas. The effects of
O₃ will be discussed in a forthcoming Environmental Health Criteria
monograph. N₂O contributes to the depletion of stratospheric O₃,
resulting in increasing ultraviolet radiation. This and other aspects
of global climate change will be evaluated in a WHO/WMO/UNEP document
entitled "Climate and Health: potential impacts of climate change".
The direct impact of airborne nitrogen is due to toxic effects,
eutrophication and soil acidification. One effect of soil
acidification is that aluminum enters into solution, hence increasing
its bioavailability. This result in root damage. Aluminum toxicity
will be discussed in a further Environmental Health Criteria
monograph.

Most biodiversity is found in (semi-)natural ecosystems, both
aquatic and terrestrial. Nitrogen is the limiting nutrient for plant
growth in many (semi-)natural ecosystems. Most of the plant species
from these (semi-)natural habitats are adapted to nutrient-poor
conditions, and can only compete successfully in soils with low
nitrogen levels (Chapin, 1980; Tamm, 1991). Ellenberg (1988b) surveyed
the nitrogen requirements of 1805 plant species from Germany and
concluded that 50% can compete successfully only in habitats that are
deficient in nitrogen. Furthermore, of the plants threatened by
increased nitrogen deposition, 75-80% are indicator species for
low-nitrogen habitats. When stratified by ecosystem type, it is also
clear that the trend of rare species occurring with greater frequency
in nitrogen-poor habitats is a common phenomenon across many
ecosystems (Fig. 16 and Fig. 17). Plant species threatened by high
nitrogen deposition are common across many ecosystem types (Ellenberg,
1988b). The critical loads for nitrogen depend on (i) the type of
ecosystem; (ii) the land use and management in the past and present;
and (iii) the abiotic conditions (especially those which influence the
nitrification potential and immobilization rate in the soil). The
impact of increased nitrogen deposition upon biological systems can be
the result of direct uptake by the foliage or uptake via the soil.
The most relevant effects at the level of individual plants are injury
to the tissue, changes in biomass production and increased
susceptibility to secondary stress factors. At the vegetation level,
this results in changes in competitive relationships between species
and loss of biodiversity.

Effects on individual plants are discussed in section 4.1. The
following natural ecosystems are treated in detail in section 4.2:
freshwater ecosystems (shallow soft-water bodies, lakes and streams)
and terrestrial ecosystems (wetlands and bogs, species-rich
grasslands, heathlands and forests). Estuarine and marine systems
are also considered.

Air quality guidelines refer to thresholds for adverse effects.
Two different types of effect thresholds exist: critical levels and
critical loads.

The critical level is defined as:

the concentration in the atmosphere above which direct adverse
effects on receptors, such as plants, ecosystems or materials,
may occur according to present knowledge.

The critical load is defined as:

a quantitative estimate of an exposure (deposition) to one or
more pollutants below which significant harmful effects on
specified sensitive elements of the environment do not occur
according to present knowledge.

Generally, critical levels for nitrogen-containing air pollutants
are expressed in terms of exposure (µg/m³ and exposure duration),
while critical loads are expressed in terms of deposition (kg nitrogen
per ha per year). Both critical level and load are intended to be
set so as to protect vegetation, and can be converted into each
other knowing the deposition velocity. Thus, it might seem to be
superfluous to assess both critical levels and loads. However, with
the currently accepted approach, critical levels and loads are more or
less complementary: critical levels focus on effect thresholds for
short-term exposure (1 year or less), while critical loads focus on
safe deposition quantities for long-term exposure (10-100 years):
critical levels are not aimed to protect plants completely against
adverse effects. No-observed-effect concentrations (NOECs) are
usually lower than critical levels. For instance, a critical level
can be set at 5% yield reduction. Thus, owing simply to differences in
definition, the critical level is generally higher than the critical
load (Fig. 18b).

In current practice there are other differences between critical
levels and loads: critical levels give details on individual compounds
and focus on responses on plant level, while critical loads cover all
nitrogen-containing compounds and focus on the vegetation or ecosystem
level. In other words: critical loads focus on functioning of the
ecosystem, while critical levels focus on protection of the relatively
sensitive plant species.

In the critical level concept, the different nitrogen-containing
compounds are evaluated separately, because of their differences in
phytotoxic properties, even when their load in terms of kg nitrogen
per ha per year is the same (Ashenden et al., 1993). Another
difference between critical level and critical load is that critical
level considers the possibility of more- or less-than-additive effects
(Wellburn, 1990), while in the critical load concept additivity of
nitrogen-containing or acidifying compounds is presumed. Moreover,
nitrogen-containing air pollutants have their impact not only because
of their contribution to the nitrogen supply. Sometimes other effects
seem to dominate. For instance, although occult deposition is
generally small in terms of nitrogen deposition, it may be of great
significance because of its ability to affect plant surfaces.

It was concluded for these reasons that both critical levels and
loads are necessary within the scope of air quality guidelines for
nitrogen-containing compounds.

Assessing effect thresholds is relatively simple in the case of
toxic compounds with an exposure/response relationship which follows
the usual sigmoid curve: the lowest exposure level that results in a
response that is significantly different from the control treatment is
the effect threshold. However, this approach is essentially invalid
for exposure of nitrogen-limited vegetation to nitrogen-containing air
pollutants. Nitrogen is a macro-nutrient and so each addition of
nitrogen can result in a physiological response: growth stimulation
gradually increases with higher exposure levels and changes in growth
inhibition at higher levels (Fig. 18a). Moreover, depending on the
definition of adverse effects, the status of the vegetation may not be
optimal at background levels (Fig. 18b). These features complicate
the assessment of effect thresholds for nitrogen-containing compounds.
Nevertheless, in this chapter effect thresholds are presented,
according to current practice.

4.1 Properties of NO_x and NH_y

In this section general information is initially presented on
uptake, detoxification, metabolism and growth aspects. In the
following subsections the data determining the critical levels for
individual compounds and mixtures are discussed. The relevance of this
information and possibilities for generalization are discussed in
sections 4.1.8 and 4.1.9, where the critical levels are estimated.
Deposition on and emission from soils and vegetation is discussed in
chapter 3.

4.1.1 Adsorption and uptake

The impact of a pollutant on plants is determined by its
adsorption, rate of uptake (flux) and the reaction of the plants.
Foliar uptake is probably dominant for NO, NO₂ (Wellburn, 1990) and
NH₃ (Pérez-Soba & Van der Eerden, 1993), while the pathway via soil
and roots is the major route for nitrogen-containing pollutants in wet
deposition.

The flux of the compounds from the atmosphere into the plant is a
complicated process, which is highly dependent on the properties of
both plant and compound and on environmental conditions. This is why
deposition velocities proved to be highly variable (chapter 3).

The flux from the atmosphere to the leaf surface (and soil)
depends on the aerodynamic and boundary layer resistances, which
are determined by meteorological conditions and plant and leaf
architecture. The flux from the leaf surface to the final site of
reaction in the cell is determined by stomatal, cuticular and
mesophyll resistance. The reaction of the plant to the nitrogen that
arrives at the target side is dependent on the intrinsic properties of
the plant and on its nutritional status, and again on environmental
conditions.

The flux of atmospheric nitrogen through the soil is conditioned
by properties of soil and vegetation and by meteorological conditions.
The chemical composition of soil water, the rate of nitrification
(NH₄⁺ -> NO₃^-), the preference of the plant for either NH₄⁺ or
NO₃^-, the root architecture and the metabolic activity of the plants
play major roles in this uptake (Schulze et al., 1989).

Adsorption on the outer surface of leaves certainly takes
place. Exposure to relatively high concentrations of gaseous NH₃
(180 µg/m³) or NH₄⁺ in rainwater (5 mmol/litre) damages the
crystalline structure of the epicuticular wax layer of the needles of
Pseudotsuga menziesii (Van der Eerden & Pérez-Soba, 1992). NO₂
(Fowler et al., 1980) and NH₄⁺ and NO₃^- in wet and occult
deposition can disturb leaf surfaces in several ways (Jacobson, 1991).
The quantitative relevance of this effect for the field situations has
not yet been shown in detail.

Uptake of NH₃ and NO_x is driven by the concentration gradient
between atmosphere and mesophyll. It is generally directly determined
by stomatal conductance and thus depends on factors influencing
stomatal aperture. Although in higher plants uptake through the
stomata strongly dominates, there are indications that penetration
through the cuticle is not completely negligible. This has been
demonstrated for NO and NO₂ (Wellburn, 1990). While stomata greatly
influence the foliar uptake of aerial nitrogen compounds, many of
these compounds subsequently alter stomatal aperture and the extent of

further uptake. The nitrogen status of plants is also known to affect
stomatal behaviour towards other environmental conditions such as
drought (Ghashghaie & Saugier, 1989).

The flux of NH₃ into a plant appeared to be linearly related to
the atmospheric concentration (Van der Eerden et al., 1991), there
being no mesophyll resistance (Van Hove et al., 1989). This relation
can become less then linear with high concentrations, e.g., some
hundreds of µg/m³ (Wollenheber & Raven, 1993). Mesophyll resistance
is, however, probably more significant for NO and NO₂ (Capron et al.,
1994).

There is increasing evidence that foliar uptake of nitrogen
reduces the uptake of nitrogen by the roots (Srivastava & Ormrod,
1986; Pérez-Soba & van der Eerden, 1993), although the driving
mechanism is not yet clear.

In the presence of low concentrations plants can emit NH₃,
rather than absorb it (chapter 3). NO and N₂O are emitted in
significant quantities by the soil (chapter 3).

Rain, clouds, fog and aerosols always contain significant amounts
of ions including NH₄⁺ and NO₃^-. In the past, foliar uptake of
nitrogen from wet deposition was considered to be negligible, but
recent research using ¹⁵N and throughfall analysis shows that this
path can contribute a high proportion of the total plant uptake (see
Pearson & Stewart, 1993, and section 2.4). In general, cations (e.g.,
NH₄⁺) are more easily taken up through the cuticle than anions
(e.g., NO₃^-). A substantial foliar uptake of NH₄⁺ from rainwater
has been measured in several tree species (Garten & Hanson, 1989).
Lower plants, such as bryophytes and lichens do not have stomata and a
waxy waterproof cuticle, and thus the potential for direct uptake of
pollutants in the form of wet or dry deposition is greater. Epiphytic
lichens are active absorbers of both NH₄⁺ and NO₃^- (Reiners &
Olson, 1984). Uptake and exchange of ions through the leaf surface is
a relatively slow process, and thus is only relevant if the surface
remains wet for long periods.

4.1.2 Toxicity, detoxification and assimilation

One would expect a positive relationship between the solubility
of a compound and its biological impact. NO is only slightly soluble
in water, but the presence of other substances can alter its
solubility. NO₂ has a higher solubility, while that of NH₃ is much
higher.

Much information exists on mechanisms of toxicity, although it is
sometimes confusing. NO₂, NO, HNO₂ and HNO₃ can be incorporated
into nitrogen metabolism using the pathway: NO₃^- -> NO₂^- ->

(NH₃ <--> NH₄⁺) <--> glutamate -> glutamine -> other amino
acids, amides, proteins, polyamines, etc. The enzymes involved
include nitrate reductase (NR), nitrite reductase (NiR) and glutamine
synthetase (GS). Glutamate dehydrogenase (GDH) plays a role in the
internal cycling of NH₄⁺.

After exposure to NO₂, nitrate can accumulate for some weeks;
accumulation of nitrite is rarely reported, although it is certainly
an intermediate. Nitrite levels can be elevated for some hours due to
the fact that NR activity is induced faster than that of NiR. In many
cases storage of excess nitrogen has been found to be in the form of
arginine (Van Dijk & Roelofs, 1988), which could last months or
longer.

NO₂^-, NH₃ and NH₄⁺ are highly phytotoxic, and could well be
the cause of adverse effects of nitrogen-containing air pollutants.
Wellburn (1990) suggested that the free radical *N=O plays a role in
the phytotoxicity of NO_x.

High concentrations can cause visible injury via lipid breakdown
and cellular plasmolysis. At lower concentrations inhibition of lipid
biosynthesis may dominate, rather than damage of existing lipids
(Wellburn, 1990).

Raven (1988) assumed that the adverse effects of nitrogen-
containing compounds are due to their interference with cellular
acid/base regulation. They can influence cellular pH both before
and after assimilation. Assimilation of most air pollutants,
including NH₃, has been shown to result in production of protons
(Wollenheber & Raven, 1993). Assimilation of nitrate and a high
buffer capacity can prevent the plant from being damaged by this
acidification (Pearson & Stewart, 1993). If these adverse effects can
effectively be counteracted, assimilation of nitrogen-containing
compounds will result in growth stimulation.

Synergistic effects have been found in nearly all studies
concerning SO₂ and NO₂ (Wellburn et al., 1981). Inhibition of NiR
by SO₂, resulting in the inability of the plant to detoxify nitrite,
might be the cause of this interaction.

4.1.3 Physiology and growth aspects

When climatic conditions and nutrient supply allow biomass
production, both NO_x and NH_y result in growth stimulation at low
concentrations and growth reduction at higher concentrations.
However, the exposure level at which growth stimulation turns into
growth inhibition is much lower for NO_x than for NH_y (see Fig. 18a).

Foliar uptake of NH₃ generally results in an increase in
photosynthesis and dark respiration, and in the amount of RUBISCO
(ribulose 1,5-biphosphate carboxylase oxygenase) and chlorophyll.
Some authors have shown that stomatal conductance increases and the
stomata remain open in the dark, resulting in enhanced transpiration
and drought sensitivity (Van der Eerden & Pérez-Soba, 1992). Most
experiments with NO and NO₂ have been conducted with relatively high
concentration levels (> 200 µg/m³). These experiments show
inhibition of photosynthesis by both NO and NO₂, possibly additively
(Capron & Mansfield, 1976). Inhibition by NO may be stronger than
that of NO₂ (Saxe, 1986). The threshold for this response is well
below the threshold for visible injury (Wellburn, 1990) and
transpiration (Saxe, 1986). With lower (nearer to ambient) NO_x
concentrations, stimulation of photosynthesis may well occur. Both
NO_x and NH_y generally cause an increase in shoot/root ratio. The
specific root length and the amount of mycorrhizal infection can be
reduced by both compounds. However, these alterations in root
properties resemble general responses to increased nitrogen nutrient
supply.

4.1.4 Interactions with climatic conditions

Evidence suggests that exposure of vegetation to NH₃ and to
mixtures of NO₂ and SO₂ can influence the subsequent response to
drought and frost stress. There is also evidence that environmental
conditions can affect the response to NO_x and to NH₃.

The foliar uptake of nitrogenous compounds in the form of wet and
occult deposition is largely via the cuticle. Uptake and exchange of
ions through the leaf surface is a relatively slow process, and thus
is especially relevant if the surface remains wet for longer periods,
e.g., in regions where exposure to mist and clouds is frequent.

The solubility of most gases, including NO, NO₂ and NH₃, rises
as temperature falls, while the metabolic activity of plants and thus
their detoxification capacity is lower. On the other hand, stomatal
conductivity and thus the influx of gases generally falls as
temperature falls.

Guderian (1988) proposed a lower critical level in winter than
for the whole year, in acknowledgement of several results that
indicate greater toxicity of NO₂ during winter conditions. For
example, exposure of Poa pratensis in outdoor chambers to 120 µg/m³
inhibited growth during winter but had little effect or even
stimulated growth in summer and autumn (Whitmore & Freer-Smith, 1982).
Mortensen (1986) found that low light and non-injurious low
temperature conditions can enhance the toxicity of NO_x. Capron et
al. (1991) found that the depression relative to the control of net
photosynthesis by 1250 µg NO/m³ plus 575 µg NO₂/m³ at 10°C was
three times, and at 5°C was almost five times, that recorded at 20°C.

An interaction between NO_x and temperature may also occur at lower
realistic concentrations. This is suggested by the observation of
nitrite accumulation at low temperatures during fumigation of
Picea rubra with 38 µg NO₂/m³ plus 54 µg SO₂/m³ (Wolfenden et
al., 1991). This temperature effect may play a role in combination
with elevated concentrations of CO₂ as well: the stimulating effect
of CO₂ on net photosynthesis was inhibited by NO_x to a larger extent
when applied at lower temperature (Capron et al., 1994). Observation
of NH₃ injury to plants also indicates that this is greatest in
winter (Van der Eerden, 1982).

In contrast with the view that NO_x (and NH₃) injury is greater
at low temperatures, Srivastava et al. (1975) found that inhibition by
NO_x of photosynthesis was greatest under optimal temperature and high
light conditions, when stomatal conductance to the gas would be
highest.

The exposure of plants to NO_x and NH₃ may reduce their ability
to withstand drought stress, owing to loss of control of transpiration
by stomata and to an increase in the shoot/root ratio (Lucas, 1990;
Atkinson et al., 1991; Fangmeijer et al., 1994).

4.1.5 Interactions with the habitat

Whether the atmospheric input of nitrogen has a positive or
negative impact depends on the plant species and habitat. Based on
experimental evidence, Pearson & Stewart (1993) hypothesized that
species which are part of a climax vegetation on nutrient-poor acidic
soils are often relatively sensitive to NO_x and NH_y. Morgan et al.
(1992) found that NO_x disrupted the NR activity to a greater extent
in calcifuge than calcicole moss species. Ombrotrophic mires and
other strongly nitrogen-limited systems may be especially prone to
detrimental effects from input of nitrogen-containing air pollutants.

The assimilation of low concentrations of NO₂ and the
incorporation into amino acids by NR (Morgan et al., 1992; Table 20)
are indicators that this pollutant can contribute to the nitrogen
budget of plants (Pérez-Soba et al., 1994). The contribution of NO_x
to the nitrogen supply increases as root-available nitrogen is lowered
(Okano & Totsuka, 1986; Rowland et al., 1987). Srivastava & Ormrod
(1986) observed reduced ability to respond to a supply of nitrate to
the roots when Hordeum vulgare was fumigated with NO₂. Similarly,
Pérez-Soba & Van der Eerden (1993) found reduced uptake of NH_4⁺ from
the soil when Pinus sylvestris was fumigated with NH₃. Although
there is much evidence that nitrogen-containing air pollutants play a
role in the nitrogen demand and nitrogen metabolism of the plant,
Ashenden et al. (1993) found no obvious relationship between
sensitivity to NO₂ and nitrogen preference, as indicated by Ellenberg
(1985).

4.1.6 Increasing pest incidence

Any change in chemical composition of plants due to the uptake of
nitrogenous air pollutants could alter the behaviour of pests and
pathogens. Evidence indicates that, in general, NO_x and NH_y
increase the growth rate of herbivorous insects (Dohmen et al., 1984;
Flückiger & Braun, 1986; Houlden et al., 1990; Van der Eerden et al.,
1991). This may also apply to fungal pathogens (van Dijk et al.,
1992).

4.1.7 Conclusions for various atmospheric nitrogen species and
mixtures

4.1.7.1 NO₂

In Table 20 the lowest effective exposure levels for NO₂ are
given. Three different types of effects are considered:

* (bio)chemical: e.g., enzyme activity, consumption quality
* physiological: e.g., CO₂ assimilation, stomatal conductivity
* growth aspects: e.g., biomass, reproduction, stress sensitivity

Four exposure durations are used in this table. These are
(including an indication of the exposure durations and the margins):

* short term (hours): < 8 h
* air pollution episodes (days): 8 h to 2 weeks
* growing season or winter season (months): 2 weeks to 6 months
* long term (years): > 6 months

To avoid the information being too selective, in each cell in
this table a species is used only once. For each cell the three
lowest effective concentrations and exposure durations are given;
species and references are mentioned in footnotes. Exposure levels
far higher than current levels measured in the field situation have
not been considered.

The fact that not all cells in Table 20 are filled with
information is because many of the experiments have been conducted
with unrealistically high concentrations. The majority of observations
mentioned in the table are on crops; several of these show growth
stimulation. Most of the responses on a biochemical level deal with
enhanced NR activity, which shows that the plants are capable of
assimilating the NO₂. A general effect threshold as derived from
Table 20 would be substantially higher if enhanced NR and biomass
production of crops were not assumed to be an adverse effect.
However, growth stimulation is often considered an adverse effect in
most types of natural vegetation. Moreover, Pearson & Stewart (1993)

assumed detoxification of NH_y and NO_x to be a potentially adverse
effect, because it contributes to cellular acidification, which can
not always be counteracted.

4.1.7.2 NO

In Table 21 the lowest effective exposure levels for NO are
given.

Most research into the effects of nitric oxide has been based on
glasshouse crops, particularly the tomato (Lycopersicon esculentum).
Virtually all experiments deal with photosynthesis or enzymatic
reactions and a few growth aspects were measured. The existing data
are rather difficult to interpret since controlled fumigation with NO
inevitably results in some oxidation to NO₂. Thus atmospheres will
contain a mixture of the oxides. There is growing interest in the
distinct properties and effects of NO and NO₂, and the mechanisms of
their cellular action probably differ (Wellburn, 1990). The exchange
properties of NO and NO₂ over vegetation (personal communication by
D. Fowler to the IPCS) and single plants (Saxe, 1986) appear quite
different. Their effects are also contrasting, and there is clearly
some dispute over which oxide is the most toxic. Earlier studies of
the inhibition of photosynthesis found NO to act more rapidly than
NO₂ (at several ppm) but to cause less overall depression of the
photosynthetic rate (Hill & Bennet, 1970). More recent photosynthetic
studies by Saxe (1986), using similar concentrations, found NO to be
considerably more toxic than NO₂. There is very little information
on contrasting effects of the two oxides at low concentrations, but
this also adds weight to the suggestion that NO is biologically more
toxic. In her studies of NR in bryophytes, Morgan et al. (1992)
discovered that exposure to NO initially inhibited NR while NO₂
induced activity. At present, however, there is insufficient
knowledge across a range of species to establish separate critical
levels for NO and NO₂, and studies using a wider variety of
vegetation are urgently required.

4.1.7.3 NH₃

The lowest effective exposure levels for NH₃ are given in Table
22.

The toxicity of NH₃ during very short exposure periods has been
tested for the purpose of evaluating accidental releases during
transport or industrial processes. The estimated critical level for
10 min is (100 ppm) (personal communication by Lee & Davison to the
IPCS). This type of exposure is out of the context of this monograph.

Table 20. Lowest exposure levels (in µg/m³) and durations at which NO₂
caused significant effects^a

(Bio)chemical Physiological Growth aspects

Long term 200 (130); 104 h/week;
7 months^r
120-500; 9.5 months^s
122; 37 weeks^t

Growing season 50; 39 days^b 120; 22 days^j 10-43; 130 days^u
or winter 125; 140 days^c 190 (65); 105 h 55-75; 62 days^v
940; 19 days^d in 15 days^k 150-190 (28-33);
120 h in 40 days^w

Air pollution 140; 1 day^e 375 (165); 35 h in 375; 2 weeks^x
episodes 160; 7 days^f 5 days^l 190; 3 days^m 100 (25);
65; 1 day^g 375 (165); 35 h 20 h in 5 days^y
in 5 daysⁿ

Short term 7500, 6 h^h 940; 1 h^o 2000-3000; 3.5 h^z
7500; 4 hⁱ 850; 7 h^p
1100; 1.5 h^q

^a If the fumigation was not continuous an average value has been estimated
and quoted in parentheses (calculated assuming 10 µg/m³ during the periods
in which the fumigation was switched off).
^b Pinus sylvestris; changes in amino acid composition, with no physiological
changes (Näsholm et al., 1991)
^c Lolium perenne; increase in GDH activity (Wellburn et al., 1981)
^d Lycopersicum esculentum; decrease in nitrate content of the leaves (Taylor
& Eaton, 1966)
^e Picea rubens, increase in NR activity (Norby et al., 1989)

Table 20 (Con't)

^f Pinus sylvestris, increase in NR activity (Wingsle et al., 1987)
^g Several bryophyte species; increase in NR activity (Morgan et al., 1992)
^h Zea mais; increase in NiR activity (Yoneyama et al., 1979)
ⁱ Vicia faba; change in amino acid composition (Ito et al., 1984)
^j Betula sp; increased water loss (Neighbour et al., 1988)
^k Phaseolus vulgaris; reversible increase in dark respiration (Sandhu & Gupta, 1989)
^l Glycine max; increase in photosynthesis (Sabarathnam et al., 1988a,b)
^m Phaseolus vulgaris; increase in transpiration (Ashenden, 1979)
ⁿ Glycine max; enhanced dark respiration (Sabarathnam et al., 1988b)
^o Vicia faba; reversible structural damage on cellular level (Wellburn et al., 1972)
^p Pisum sativum; emission of stress ethylene (Mehlhorn & Wellburn, 1987)
^q Medicago sativa, Avena sativa; inhibition of photosynthesis (Hill & Bennet, 1970)
^r Several grass species; reduction in shoot growth (Whitmore & Mansfield, 1983)
^s Citrus sinensis; increased fruit drop (Thompson et al., 1970)
^t Polytrichum formosum and 3 fern species; injury and changes in growth (Ashenden
et al., 1990; Bell et al., 1992)
^u Brassica napus and Hordeum vulgare; growth stimulation (resp.: Adaros et al.,
1991a,b)
^v Phaseolus vulgaris; increase in total dry matter, not in yield (Bender et al.,
1991)
^w Raphanus sativus; growth stimulation (Runeckles & Palmer, 1987)
^x Helianthus annuus; reduction in net assimilation rate (Okano et al., 1985b)
^y Pinus strobus; slight needle necrosis in 2 of 8 clones (Yang et al., 1983)
^z Nicotiana tabacum; leaf necrosis (Bush et al., 1962)

Table 21. Lowest exposure levels (in µg/m³) at which NO caused
significant effects^a

(Bio)chemical Physiological Growth aspects

Growing season 44; 21 days^b 625; 16 daysⁿ
500; 28 days^c 500;^o

Air pollution 375; 8 days^d 1250; 4 daysⁱ 1250; 5 days^p
episodes 44; 8-24 h^e 125; 20 h^j
1875; 18 h^f

Short term 188; 7 h^g 750; 1 h^k
500; 3 h^h 2500; 10 min^l
1875; 20 min^m

^a If the fumigation was not continuous an average value has been
estimated and quoted in parentheses (calculated assuming 10 µg/m³
during the periods in which the fumigation was switched off).
^b Four bryophyte species; inhibition of nitrate-induction of NR
(Morgan et al., 1992)
^c Lycopersicon esculentum; induction of NiR (Wellburn et al., 1980)
^d Lactuca sativa; induction of NiR (Besford & Hand, 1989)
^e Ctenidium molluscum (bryophyte); inhibition of NR (Morgan et al., 1992)
^f Capsicum annum; reduction in NiR activity (Murray & Wellburn, 1980)
^g Pisum sativum; increase in ethylene release (Mehlhorn & Wellburn, 1987)
^h Lycopersicon esculentum; induction of NiR (Wellburn et al., 1980)
ⁱ Eight indoor ornamental species; inhibition of photosynthesis
(Saxe, 1986)
^j Lycopersicon esculentum; inhibition of photosynthesis (Capron &
Mansfield, 1989)
^k Avena sativa & Medicago sativa; inhibition of photosynthesis (Hill &
Bennet, 1970)
^l Lactuca sativa; inhibition of photosynthesis (Capron, 1989)
^m Lycopersicon esculentum; inhibition of photosynthesis (Mortensen, 1986)
ⁿ Lactuca sativa; reduction in plant mass (Capron et al., 1991)
^o Lycopersicon esculentum; reduction in plant mass (Anderson &
Mansfield, 1979)
^p Lycopersicon esculentum; reduction in plant mass (Bruggink et al., 1988)

Table 22. Lowest exposure levels (in µg/m³) at which NH₃ caused
significant effects^a

(Bio)chemical Physiological Growth aspects

Long term 50; 8 months^b 53; 9 months^h 25; 1 year^k
53; 8 months^l
35; 16 months^m

Growing season 100; 6 weeks^c 50; 6 weeksⁱ 60; 2 monthsⁿ
or winter 60; 14 weeks^d 20; 90 days^o
180; 13 weeks^e 30; 23 days^p

Air pollution 2000; 24 h^f 213; 5 days^j 120; 11 days^q
episodes 213; 5 days^g 1000; 2 weeks^r
300; 3 days^s

Short term 30 000; 1 h^t
2000 2 h^u
2000 6 h^v

^a If the fumigation was not continuous an average value has been
estimated and quoted in parentheses (calculated assuming
10 µg/m³ during the periods in which the fumigation was
switched off).
^b Species of Violion caninea alliance; imbalanced nutrient
status (Dueck & Elderson, 1992)
^c Deschampsia flexuosa; change in amino acid composition (Van
der Eerden et al., 1990)
^d Pinus sylvestris; increased GS activity (Pérez-Soba et al.,
1990)
^e Pseudotsuga menziesii; imbalanced nutrient status (Van der
Eerden et al., 1992)
^f Lycopersicum esculentum; increase of NH_4⁺ in the dark
(Van der Eerden, 1982)
^g Lolium perenne; 30% of N in the plant is derived from
foliar uptake (Wollenheber & Raven, 1993)
^h Pinus sylvestris; increased loss of water after two weeks
of desiccation (Dueck et al., 1990)
ⁱ Populus sp.; increase in stomatal conductance in leaves;
increase in mesophyll conductance and maximum photosynthetic
rate at a slightly higher exposure level (Van Hove et al., 1989)
^j Lolium perenne; significant impact acid/base regulation and
nutrients status

Table 22 (Con't)

^k Pseudotsuga menziesii; erosion of wax layer (Thijse & Baas,
1990; the authors have some doubts about the causality of this
effect (personal communication)
^l Calluna vulgaris; reduction in survival rate after winter
(Dueck, 1990)
^m Arnica montana; reduced survival after winter and flowering
(Van der Eerden et al., 1991)
ⁿ Field exposure during winter; median concentration; severe
injury of several conifer species (Van der Eerden, 1982)
^o Viola canina, Agrostis capillaris; 50% growth stimulation
of the shoot (not of the roots) (Van der Eerden et al., 1991)
^p Racomitrium lanuginosum; chlorosis (Van der Eerden et al.,
1991)
^q Hypnum jutlandicum; chlorosis (Van der Eerden et al., 1991)
^r Lepidium sativum; reduction in dry weight (Van Haut &
Prinz, 1979)
^s Several horticultural crops; leaf injury
^t Various deciduous trees; leaf injury (Ewert, 1979)
^u Brassica sp., Helianthus sp.; leaf injury (Benedict & Breen,
1955)
^v Rosa sp.; leaf injury rose (Garber, 1935)

Several cells in Table 22 could not be filled; the majority of quoted
effects are on biomass production and tissue injury. It is clear that
the data in this table are not random; nearly all of the information
originating from one Dutch research group. Only a few pollution
climates were considered. The results may be representative for
mild oceanic climates, but probably not for cold climates with dark
winters: toxicity of NH₃ increases with lower temperature and lower
light intensity. The effects of NH₃ need to be studied with more
plant species and under more climatic conditions in order to obtain
critical levels with sufficient potential for generalization.

4.1.7.4 NH₄⁺ and NO₃^- in wet and occult deposition

NH₄⁺, NO₃^- and H⁺ make up about half of the ionic
composition of rain, clouds, fog and aerosols. The impact of the
acidity of rain and clouds has received much attention in recent years
(Jacobson, 1991). This is not the case with other compounds in wet
deposition, although their relevance is recognized. At the same pH,
Cape et al. (1991) found a much greater effect of sulfuric acid than
of nitric acid, indicating that the impact of acid rain is not only
through protons, but also through anions.

There is an abundance of information on the effects of NH₄⁺ in
soil solution. However, threshold concentrations for NH₄⁺ in the
soil (e.g. Schenk & Wehrman, 1979) can not be considered a critical
level for rain water quality, because the type of exposure and
response is completely different.

Wet deposition containing NH₄⁺ can reduce frost tolerance (Cape
et al., 1990) and induce leaching of K⁺ and other cations (Roelofs
et al., 1985). It is not yet clear whether this type of ion exchange
can have deleterious effects on its own in the field situation.

Currently, too few quantitative data on the effects of nitrogen-
containing wet and occult deposition are available for critical levels
for this group of compounds to be derived.

4.1.7.5 Mixtures

A polluted atmosphere generally consists of a cocktail of
compounds, but certain combinations are more frequent. Because of
their role in the formation of tropospheric O₃, simultaneous
co-occurrence of relatively high levels of O₃ and NO are rarely
observed, while sequential co-occurrences are much more frequent
(Kosta-Rick & Manning, 1993). If burning of fossil fuels results in
emission of SO₂, this is often combined with emission of NO_x.

a) SO₂ plus NO₂

Synergism has been found in nearly all studies concerning this
combination, with only few exceptions (Kuppers & Klump 1988; Murray et
al., 1992). Based on data presented by Whitmore (1985), for Poa
pratensis the effect threshold for combinations of SO₂ and NO₂, in
equal concentrations when expressed in ppm, is in the range of 1.2-2.0
ppm.days (Fig. 19). This threshold applies to effects by combinations
of SO₂ and NO₂; the effects of single exposures were not assessed in
this study. However, it is reasonable from other references to expect
synergism, and thus to include this threshold in Table 23, in which
combined effects are summarized. Another threshold for combinations
of SO₂ and NO₂ was defined by Van der Eerden & Duym (1988) (Fig.
20; Table 23).

b) SO₂ plus NH₃

Adsorption of either NH₃ or SO₂ on leaf surfaces is enhanced by
the presence of the other compound (Van Hove et al., 1989).
Interactive physiological effects have been found as well (Dueck,
1990; Dueck et al., 1990; Dueck & Elderson, 1992). Currently, there
is far too little information on this combination to quantify this
interaction.

Table 23. Lowest exposure levels at which NO₂ increases the
effect of SO₂, O₃, or SO₂ plus O₃

(Bio)chemical Physiological Growth aspects

Long term 150-190; 9 months^f
220; 60 weeks^g
19; 10-41 weeks^h

Growing season 55-75; 34 days^b 135; 28 days^d 30; 38 daysⁱ
or winter 135; 28 days^c 10-43; 130 days^j
30; 43 days^k

Air pollution 80; 2 weeks^l
episodes 75; 1 day^m

Short term 153; 1 h^e 325; 1 h^m
400; 1 hⁿ

^a If the fumigation was not continuous an average value has
been estimated and quoted in parentheses (calculated
assuming 10 µg/m³ during the periods in which the
fumigation was switched off).
^b Phaseolus vulgaris; inhibition of parts of nitrogen
metabolism, when exposed sequentially with O₃
(100-120 µg/m³; 8 h/day)
^c Lolium perenne; decrease in proline content during winter
hardening when applied in combination with SO₂ at
188 µg/m³ (Davison et al., 1987)
^d Lolium perenne; less negative osmotic potential during
winter hardening when applied in combination with SO₂ at
188 µg/m³ (Davison et al., 1987)
^e Phaseolus vulgaris; Inhibition of photosynthesis when in
combination with SO₂ (215 µg/m³); without SO₂
inhibition at 760 µg/m³ (Bennet et al., 1990)
^f Several crops; growth stimulation by NO₂ turns into a
reduction in synergism with sequential treatment with O₃
(160-200 µg/m³; 6 h/day) (Runeckles & Palmer, 1987)
^g Six tree species; reduced plant growth in combination with
SO₂ (280 µg/m³), both antagonism and synergism
(Freer-Smith, 1984)
^h 10 grass species were tested in combination with SO₂
(27 µg/m³). Three species showed growth stimulation.
Reduced growth was found at higher concentrations.
Interactions with acidic mist and with O₃ were found
(Ashenden et al., 1993).

Table 23 (Con't)

ⁱ Poa pratensis; inhibition of biomass production; in
combination with SO₂ (42 µg/m³) for 38 days; the longest
exposure period used in the experiments. Calculated from
data from Whitmore (1985), assuming synergism and a critical
level for SO₂ plus NO₂ of 1.2 ppm.days (Whitmore,
1985).
^j Brassica napus and Hordeum vulgare; antagonism (and
rarely synergism) with O₃ (6-44 µg/m³; 8 h/day) and
SO₂ (9-33 µg/m³, continuously): enhanced yield turns into
reduction (Adaros et al., 1991a,b)
^k Plantago mayor; reduced shoot dry weight synergism with
SO₂ (60 µg/m³) and O₃ (60 µg/m³, 8 h/day)
(Mooi, 1984)
^l Poa pratensis; inhibition of biomass production; in
combination with SO₂ (110 µg/m³) for 2 weeks (the upper
margin of the exposure period of this cell in the table; the
shortest fumigation in this survey was 20 days. Calculated
from data from Whitmore (1985), assuming synergism and a
critical level for SO₂ plus NO₂ of 1.2 ppm.days
(Whitmore, 1985).
^m Critical level for NO₂ assuming SO₂ to be present at
70 µg/m³; considered to be a critical level for a 24-h mean
(UNECE, 1994) (Van der Eerden & Duym, 1988)
ⁿ Lycopersicon esculentum; reduction in plant mass if in
combination or preceded by O₃ (160 µg/m³; 1 h)
(Goodyear & Ormrod, 1988).

c) NO plus NO₂

When activated charcoal has been used as filter material in NO₂
fumigation experiments, NO must have been present as well, because
activated charcoal has virtually no capacity to absorb NO. In those
studies, responses must have been due to NO₂ plus NO. Although the
toxicity of NO was often considered to be much less than that of NO₂,
currently the two compounds are assumed to be equally toxic and to
act additively. However, Wellburn (1990) and others have stated
that NO is more toxic, and Saxe (1994) showed that the variation in
sensitivity amongst species is different for the two compounds. This
supports the suggestion of Wellburn that the mechanism of toxicity is
different.

For the purpose of deriving critical levels, the assumption of
additivity may result in an underestimation. However, there are not
enough data to quantify this.

d) Mixtures with O₃

The combination NH₃ plus O₃ has rarely been studied. No
statistically significant interactions have been found as yet, but in
one study the threshold for leaf injury was higher in the presence
of NH₃ (Van der Eerden et al., 1994). The combination NO₂ plus O₃
has been studied more frequently, but the responses differed
considerably between experiments and species. Additivity or
antagonism was found by Runeckles & Palmer (1987), Adaros et al.
(1991a,b), and Bender et al. (1991). Synergism was reported by Ito et
al. (1984), Runeckles & Palmer (1987) and Kosta-Rick & Manning (1993).

The combination of SO₂ plus O₃ plus NO₂ has also been studied.
Again the responses varied between plant species and experiment.
Antagonism, additivity and synergism have all been found (Kosta-Rick &
Manning, 1993).

e) Mixtures with elevated CO₂

Generally, an increased supply of CO₂ to crops results in an
enhanced biomass production. The responses of native species are more
variable but are also frequently positive. This growth stimulation is
limited by a deficiency of other nutrients. Nitrogen can be one such
limiting factor, and for this reason a nitrogen fertilizer such as
NH_y and possibly low NO_x concentrations could be hypothesized to
have a more-than-additive relationship with CO₂. However, as yet
there is no experimental evidence for this. Van der Eerden
et al. (1994) and Pérez-Soba et al. (1994) found stimulation of
photosynthesis and growth by both NH₃ and CO₂, but not by a
combination of these two compounds.

Effects of the combination of NO_x and CO₂ have not yet been
studied within the scope of global climate change. But some relevant
information could be gained from the literature dealing with CO₂
enrichment in glasshouses. When the flue gases of the heating system
are used as a CO₂ source, NO_x (in which NO is dominant) becomes a
major contaminant. The fertilizing effect of elevated CO₂ can
largely disappear in the presence of NO_x (Anderson & Mansfield, 1979;
Saxe & Voight Christensen, 1984; Mortensen, 1985; Bruggink et al.,
1988; Capron et al., 1994).

The CO₂, NH₃ and NO_x concentrations used in combination in
these experiments were relatively high and therefore cannot be used
in the critical level assessment. More experiments with lower
concentrations are required.

Table 23 indicates, surprisingly, that the effective long-term
exposures are generally higher than those of shorter duration.
However, long-term responses (more than half a year) have rarely been
studied. Therefore, the information on effects over growing season
periods may be much more representative of long-term effects.

A study included in a report by UNECE (1994) used 21 µg SO₂/m³
and 11 µg NO₂/m³, over the entire growing season and found synergism
in reducing biomass production of Pisum sativum and Spinacea
oleracea. Similar results were found for Hordeum vulgare and
Brassica oleracea, when fumigation was conducted for 120-190 days
with 30-40 µg SO₂/m³ and 30-50 µg NO₂/m³. This study cannot be
used for the assessment of critical levels because it has not yet been
published, but it indicates that lower levels of the two pollutants
than those quoted in Table 23 can influence plant responses.

4.1.8 Appraisal

Table 24 shows the former air quality guidelines for NO₂ and
some other critical levels assessed in the same period. Fig. 21
summarizes the results given in Tables 20 to 23. In this figure
curves are drawn to estimate critical levels according to current
practice, known as the "envelope" approach. After having plotted all
effective exposure levels in a graph of concentration and exposure
time, a curve is drawn just below the lowest effective exposures.
Critical levels can be derived from this curve. Fig. 21 shows that
more experiments with exposure periods of 0.5 to 5 days are required
to give a more solid basis for the estimation of critical levels of
24 h.

Table 24. Critical levels for NO₂

Concentration Exposure time Reference
(µg/m³)

95 4 h WHO (1987)
30^a annual mean WHO (1987)
800 1 h Guderian (1988)
60 growing season Guderian (1988)
40 winter Guderian (1988)

^a SO₂ and O₃ not higher than 30 µg/m³ and 60 µg/m³, respectively

A second approach to arrive at critical levels is the statistical
model of Kooijman (1987). Based on the variation in sensitivity
between species, critical levels are calculated taking into account
the number of tested species and the level of uncertainty (Van der
Eerden et al., 1991). The second approach is better, but only part of
the available data is suitable for this approach.

Tables 20 to 23 show that some new relevant information has
appeared. Comparing the data of Table 20 with those of Table 21
(Fig. 21a and 21b), it appears that NO₂ has slightly higher effect
thresholds than NO. However, this probably reflects the separate
attention paid to these compounds, rather than any difference in
toxicity. It is now obvious that the toxicity of NO cannot be
ignored, and it should be included in the guidance values. The
consideration of NO and NO₂ together (leading to a guidance value for
NO_x) seems the best way of evaluating the impact of NO. However,
future research should evaluate the specific phytotoxic properties of
the individual compounds and their combinations.

It is not yet possible to discriminate in the critical level
assessment between separate types of vegetation, such as crops,
plantation forests, natural forests and other natural vegetation. A
1-h average for NO₂ of 800 µg/m³ to prevent acute damage
(Table 24) is probably too high. A critical level for NO_x of around
300 µg/m³ would be better. A critical level of 95 µg/m³ as a 4-h
mean, as proposed in the previous WHO guidelines (WHO, 1987), is still
realistic, but not very practical. If critical levels for short
periods (e.g., 1 or 8 h) should be defined, it is probably necessary
to separate day- and night-time exposures. A critical level for a
24-h mean is more practical, as this is relevant for both peak
concentrations of several hours and air pollution episodes of several
days.

For the growing season and winter half year, Guderian (1988)
suggested critical levels of 60 and 40 µg/m³, respectively. From
Table 20 it can be seen that the critical level of 60 µg/m³ can cause
substantial growth stimulation rather than reduction. Within the
context of air quality guidelines, this type of response must be
regarded as potentially adverse because, for instance, of its
influence on competition within the natural vegetation. From current
knowledge it is evident that 60 µg/m³ is too high to prevent growth
stimulation. In addition, the critical level of 30 µg/m³ for an
annual mean, given in the 1987 WHO guidelines, will almost certainly
not protect all plant species. However, for crops, where growth
stimulation is rarely an adverse effect, this could be acceptable if
secondary effects are negligible. The experimental basis for the WHO
air quality guidelines of 1987 was relatively poor, but evidence is
increasing that these are certainly not unrealistically low. Not even

all direct adverse effects are eliminated by these levels (Adaros et
al., 1991a,b; Bender et al., 1991; Ashenden et al., 1993). Thus, the
updated guidelines/guidance values should be lower than the ones of
1987.

A long-term critical level for NO₂ of 10 µg/m³, especially to
avoid eutrophication of nutrient-poor vegetation, was proposed by
Guderian (1988) and Zierock et al. (1986). The basis for this
proposal was the work of Lee et al. (1985) and Press et al. (1986),
who found reduced growth of Sphagnum cuspidatum in regions with an
annual mean concentration of 38 and 11 µg/m³, respectively, compared
to the growth in another region with 4 µg/m³ after 140 days of
exposure. However, Lee et al. (1985) also showed that the poor growth
of Sphagnum was more closely related to the excessively high
concentrations of nitrate and ammonium ions in bog water rather than
to the concentration of NO₂ alone. Thus, this information could well
be used to assess water quality guidelines, but is not very useful as
a basis for air quality guidelines.

4.1.8.1 Representativity of the data

Critical levels for adverse effects of NH₃ on plants were
estimated using the model of Kooijman (Van der Eerden et al., 1991).
To protect 95% of the species at P < 0.05, a 24-h critical level of
270 and an annual mean critical level of 8 µg/m³ were estimated.
With the graphical approach the 24-h average was a little lower and
the annual mean somewhat higher (13 and 200 µg/m³, respectively;
Fig. 21).

On the basis of a review by Cape (1994), critical levels for H⁺
and NH₄⁺ were adopted for locations where ground-level cloud is
present for more than 10% of the time and where calcium and magnesium
concentrations in rain or cloud do not exceed H⁺ and NH₄⁺
concentrations (mainly high elevation areas in cold climate zones):
300 µmol NH₄⁺/litre as an annual mean (UNECE, 1994).

There remains considerable deficiency in the amount and scope of
experimentally derived information on which to base air quality
guidelines. This results from the fact that most experiments have
been performed under conditions that cannot directly be compared to
outdoor circumstances. In most experiments, only primary effects such
as photosynthesis and biomass production were evaluated, and rarely
secondary effects such as alteration of stress tolerance or
competitive ability. The plant species chosen in most experiments
were crops, although evidence suggests that some native species are
relatively more sensitive. For instance, lower plants such as
bryophytes and lichens are not protected by a waxy waterproof cuticle
and do not have the potential to close stomata. Furthermore, Pearson
& Stewart (1993) suggested that plants species from nutrient-poor
acidic soils are more sensitive.

Further work, employing low concentrations of NH_y and NO_x
(especially NO) in different climates, is urgently required. It is
not realistic to screen for all likely growth and physico-chemical
effects in the majority of species in order to arrive at general
effect thresholds. Selections must be made on the basis of an
understanding of differences in sensitivity between species. However,
an obvious mechanistic explanation for sensitivity differences is not
yet available. For instance, there appears to be no relationship
between the sensitivity to NO₂ and the nitrogen preference
(Ellenberg, 1985; Ashenden et al., 1993). Sensitivity classifications
for some tens of species have been made for NO₂ and NH₃ (e.g. US
EPA, 1978; Taylor et al., 1987), but it appears difficult to extend
predicitions very far beyond those examined. The hypotheses of Raven
(1988) and Pearson & Stewart (1993) should be studied in more detail
in laboratory experiments and field studies, as they could result in
an efficient selection criterium for further screening.

An attempt to determine the global situation regarding
nitrogen-containing compounds is being made. The assumption that all
deposited nitrogen-containing compounds (which is part of the critical
load concept) act additionally in their impact on vegetation is poorly
based on experimental results and is probably not valid for the short
term.

Generalizations and simplifications have to be made to arrive at
conclusions that are applicable in environmental policy making, but
this should be done with great care. Mechanistic simulation models
can become a powerful tool for making general predictions on the basis
of various air pollution experiments (Van de Geijn et al., 1993).
However, sufficient knowledge of biochemical and physiological
mechanisms to incorporate the impact of air pollution on vegetation
into these models is still lacking. This applies especially to
natural vegetation where stress sensitivity and competition are key
factors.

Many gaps in understanding the impact of nitrogen-containing air
pollution on vegetation still exist, and this is a good reason to use
a safety factor in determining critical levels and loads. However,
currently there is still no broadly accepted approach to quantify such
a safety factor.

4.1.9 General conclusions

The sum of information on gaseous NH₃ and on NH₄⁺ in wet and
occult deposition is still too limited to arrive at air quality
guidelines, as they should have a broad applicability. The critical
levels for NH₃ and NH₄⁺ are probably only valid for temperate
oceanic climatic zones (see sections 4.1.7.3, 4.1.7.4 and 4.1.8).

In most studies with NO and NO₂, no significant effects were
found at levels below 100 µg/m³, but several relevant exceptions
exist. NO₂ altered the response to O₃ generally with a
less-than-additive interaction. In combination with SO₂, NO₂ acted
more-than-additively in most cases. With CO₂ and with NO, no
interaction and thus additivity were generally found. The lowest
effective concentration levels of NO₂ are about equal for NO₂ alone
and in combination with O₃ or SO₂, although, generally, at
concentrations near to its effect threshold NO₂ causes growth
stimulation if it is the only pollutant, while in combination with
SO₂ and/or O₃ it results in growth inhibition.

To include the impact of NO, a critical level for NO_x instead of
one for NO₂ is proposed, assuming that NO and NO₂ act in an additive
manner. A strong case can be made for the provision of critical
levels for short-term exposures, but currently there are insufficient
data to provide these with sufficient confidence. Current evidence
exists for a critical level of around 75 µg/m³ for NO_x as a 24-h
mean.

The critical level for NO_x (NO and NO₂, added in ppb and
expressed as NO₂ in µg/m³) is 30 µg/m³ as an annual mean. At
concentrations slightly above this critical level, growth stimulation
will be the dominant effect if the ambient conditions allow growth and
NO_x is the only pollutant. This stimulation may be combined with a
minor increase in sensitivity to biotic and abiotic stresses. In
cases where biomass production is a positive effect, e.g., in
agriculture and plantation forests, the critical level can be higher.
Current knowledge is insufficient to arrive at critical levels for
these systems.

The critical level can be converted into deposition quantities.
With deposition velocities of 3 and 0.3 mm/second for NO₂ and NO,
respectively (see section 3.2.2 and Table 5), the annual deposition
corresponding to a NO_x concentration of 30 µg/m³ is 4.8 kg/ha when
half of the NO_x is NO₂ and 8.3 kg/ha when all is NO₂. This
indicates that at a concentration near to its critical level the
contribution of NO_x to the nitrogen demand is negligible for
fertilized crops but not for natural vegetation (see section 4.2).

4.2 Effects on natural and semi-natural ecosystems

4.2.1 Effects on freshwater and intertidal ecosystems

In this section the effects of atmospheric nitrogen deposition
on freshwater and intertidal ecosystems are evaluated. The
effects of increased emissions of nitrogen compounds with respect to
eutrophication are examined in order to establish ecosystem guidelines

for nitrogen deposition. The ecological effects of nitrogen
deposition are reviewed for (i) shallow softwater lakes and (ii) lakes
and streams.

4.2.1.1 Effects of nitrogen deposition on shallow softwater lakes

In the lowlands of western Europe, soft water is often found on
sandy soil which is poor in calcium carbonate or almost devoid of it.
The water is poorly buffered and the concentrations of calcium in the
water layer are very low. The water bodies are shallow and fully
mixed, with periodically fluctuating water levels. They are mainly
fed by rain water and thus are oligotrophic. These softwater
ecosystems are characterized by plant communities from the
phytosociological alliance Littorellion (Schoof-van Pelt, 1973;
Wittig, 1982; Roelofs, 1986; Vöge, 1988; Arts, 1990). The stands of
these communities are characterized by the presence of rare and
endangered isoetids, such as Littorella uniflora, Lobelia dortmanna,
Isoetes lacustris, I. echinospora, Echinodorus species, Luronium
natans and many other softwater macrophytes. These softwater bodies
are now almost all within nature reserves and have become very rare in
western Europe. A strong decline in amphibians has also been observed
in these water bodies (Leuven et al., 1986).

The effects of nitrogen pollutants on these softwater bodies have
been intensively studied in the Netherlands both in field surveys and
experimental studies. Field observations on about 70 softwater bodies
(with well-developed isoetid vegetation in the 1950s) showed that the
water, in which these macrophytes were still abundant in the early
1980s, was poorly buffered (alkalinity of 50-500 µeq/litre), slightly
acidic (pH=5-6) and very poor in nitrogen (Roelofs, 1983; Arts et al.,
1990). The softwater sites where these plant species had disappeared
could be divided into two groups. In 12 of the 53 softwater sites,
eutrophication, resulting from nutrient-enriched water, seemed to be
the cause of the decline. In this group of non-acidified water
bodies, plant species, such as Myriophyllum alterniflorum, Lemna
minor or Riccia fluitans had become dominant. High concentrations
of phosphate and ammonium ions were measured in the sediment. In some
of the larger water bodies no macrophytes were found, as a result of
dense plankton bloom. In the second group of lakes and pools (41 out
of 53) another development had taken place: the isoetid species were
replaced by dense stands of Juncus bulbosus or aquatic mosses such
as Sphagnum cuspidatum or Drepanocladus fluitans. This clearly
indicates acidification of the water in recent decades, probably
caused by enhanced atmospheric deposition. In the same field study it
was shown that the nitrogen levels in the water were higher in
ecosystems where the natural vegetation had disappeared, compared with
ecosystems where the Littorellion stands were still present (Roelofs,
1983). This strongly suggests the detrimental effects of atmospheric
nitrogen deposition in these softwater lakes.

Several ecophysiological studies have revealed the importance of
(i) inorganic carbon status of the water as a result of intermediate
levels of alkalinity, and (ii) low nitrogen concentrations for the
growth of the endangered isoetid macrophytes. Furthermore, almost all
of the typical softwater plants had a relatively low potential growth
rate. Increased acidity and higher concentrations of ammonium ion in
the water clearly stimulated the development of Juncus bulbosus and
submerged mosses such as Sphagnum and Drepanocladus species
(Roelofs et al., 1984; Den Hartog, 1986). Cultivation experiments
confirmed that the nitrogen species involved (ammonium or nitrate
ions) differentially influenced the growth of the studied species of
water plants. Almost all of the characteristic softwater isoetids
developed better when nitrate was added instead of ammonium, whereas
Juncus bulbosus and aquatic mosses (Sphagnum & Drepanocladus) were
clearly stimulated by ammonium (Schuurkes et al., 1986). The
importance of ammonium for the growth of these aquatic mosses was also
reported by Glime (1992).

The effects of atmospheric deposition have been studied in
small-scale softwater systems during a 2-year treatment with different
artificial rainwaters. Acidification, without airborne nitrogen input
(using sulfuric acid), did not result in a mass growth of Juncus
bulbosus, and a diverse isoetid vegetation remained present.
However, after increasing the nitrogen concentration in the
precipitation (as ammonium sulfate), similar changes to those seen in
field conditions were observed, i.e. a dramatic increase in the
dominance of Juncus bulbosus, of submerged aquatic mosses and of
Agrostic canina (Schuurkes et al., 1987). It became obvious that
the observed changes occurred because of the effects of ammonium
sulfate deposition, leading to both eutrophication and acidification.
The increased levels of ammonium in the system directly stimulated the
growth of plants such as Juncus bulbosus, whereas the surplus
ammonium would be nitrified in this water (pH > 4.0). During this
nitrification process, H⁺ ions are produced, which increases the
acidity of the system. The results of this study clearly demonstrated
that the changes in composition of the vegetation had already occurred
after a 2-year treatment with > 19 kg nitrogen per ha per year. A
reliable critical load for nitrogen deposition in these shallow
softwater lakes is thus most likely to be below 19 kg nitrogen per ha
per year and probably between 5 to 10 kg nitrogen per ha per year.
This value is supported by the observation that the greatest decline
in the species composition of the Dutch Litorellion communities has
coincided with nitrogen loads of around 10-13 kg nitrogen per ha per
year (Arts, 1990).

4.2.1.2 Effects of nitrogen deposition on lakes and streams

There is ample evidence that an increase of acidic and
acidifying compounds in atmospheric deposition had resulted in recent
acidification of lakes and streams in geologically sensitive regions

of Scandinavia, western Europe, Canada and the USA (Hultberg, 1988;
Muniz, 1991). This acidification is characterized by a decrease in pH
and acid neutralizing capacity and by increases in concentrations of
sulfate, aluminium, and sometimes nitrate and ammonium. It has been
shown since the 1970s, using various approaches (field surveys,
laboratory studies, whole-lake experiments), that these changes have
had major consequences for plant and animal species (macrofauna,
fishes) and for the functioning of these aquatic ecosystems.

The critical loads of acidity (from SO_y and NO_y) for aquatic
ecosystems, based on steady-state water chemistry models, were
published by the UN Economic Commission for Europe (UNECE) in 1988 and
1992. These models incorporate both sulfur and nitrogen acidity, and
critical loads are calculated on the basis of: (i) base cation
deposition; (ii) internal alkalinity production or base cation
concentrations; and (iii) nitrate leaching from the water system. The
calculated critical loads are thus site-specific (sensitive areas or
not) and also depend on the local hydrology and precipitation (for
full details, see Hultberg (1988), Henriksen (1988) and Kämäri et al.
(1992)). The critical loads of nitrogen acidity (kg nitrogen per ha
per year) for the most sensitive lakes and streams are:

Scandinavian 1.4-4.2 (Hultberg, 1988; Henriksen,
waters 1988; Kämäri et al., 1992)

Alpine lakes 3.5-6.1 (Marchetto et al., 1994)

Humic moorland 3.5-4.5 (Schuurkes et al., 1987;
pools van Dam & Buskens, 1993)

In many areas with high water alkalinity and/or high base cation
deposition, the values of the critical load for nitrogen acidity are
much higher than those for sensitive freshwaters. At present, the
possible effects of nitrogen eutrophication by ammonia/ammonium or
nitrate deposition are not incorporated in the establishment of
critical loads for nitrogen, except for shallow softwater lakes (see
section 4.2.1.1). This is because primary production in almost all
aquatic ecosystems is limited by phosphorus availability, and thus
nitrogen enrichment has been considered unimportant in this respect
(Moss, 1988). This certainly holds for those aquatic ecosystems
considered above, where the critical load with respect to acidifying
effects are certainly more relevant than the effects of nitrogen
eutrophication. It is, however, to be expected that the following
aquatic ecosystems are sensitive to nitrogen enrichment: (i) alpine
lakes; (ii) water with low background nitrogen; and (iii) humic lakes
(Kämäri et al., 1992). The effects of nitrogen eutrophication
(including ammonia/ammonium) in these water bodies need further
research and should be taken into account in future critical loads
determinations for nitrogen. At present it is not possible to present

reliable critical loads for nitrogen eutrophication in these aquatic
ecosystems. An overview of critical loads for nitrogen in aquatic
ecosystems is given in section 8.2.2.

4.2.2 Effects on ombrotrophic bogs and wetlands

In this section the effects of atmospheric nitrogen deposition in
(semi-)natural wetlands are evaluated. The effects of enhanced
nitrogen inputs are considered for: (i) ombrotrophic (raised) bogs;
(ii) fens; and (iii) intertidal fresh- and saltwater marshes. A
common feature of all these systems is the anaerobic nature of their
waterlogged soils, characterized by low redox potential, high
concentrations of toxic reduced substances and high rates of
denitrification (Gambrell & Patrick, 1978; Schlesinger, 1991).

4.2.2.1 Effects on ombrotrophic (raised) bogs

Ombrotrophic ("rain-nourished") bogs, which receive all their
nutrients from the atmosphere, are particularly sensitive to airborne
nitrogen loads. These bogs are systems of acidic wet areas and are
very common in the boreal and temperate parts of Europe. Because of
the anaerobic conditions, decomposition rates are slow, favouring the
development of peat. In western Europe and high northern latitudes,
typical plant species include bog-mosses ( Sphagnum species), sedges
(Carex; Eriophorum) and heathers ( Andromeda, Calluna and Erica).
Insectivorous plant species (e.g., Drosera) are especially
characteristic of these bogs; they compensate for low nitrogen
concentrations by trapping and digesting insects (Ellenberg, 1988b).

Clear effects of nitrogen eutrophication have been observed in
Dutch ombrotrophic bogs. The composition of the moss layer in the
small remnants of the formerly large bog areas has markedly changed in
recent decades as nitrogen loads have increased to 20-40 kg nitrogen
per ha per year (especially as NH_4⁺/NH₃). The most characteristic
species (Sphagnum) are replaced by the more nitrophilous mosses
(Greven, 1992). A national survey of Danish ombrotrophic bogs has
shown a decline of the original bog vegetation together with an
increase of more nitrogen-dependent species in areas with high ammonia
deposition (> 25 kg ammonium nitrogen per ha per year (Aaby, 1990).

The effects of atmospheric nitrogen deposition on ombrotrophic
bogs have also been intensively studied in the United Kingdom (Lee et
al., 1989; Lee & Studholme, 1992). Many characteristic Sphagnum
species are now largely absent from affected ombrotrophic bog areas
in the United Kingdom, such as the southern Pennines in England.
Atmospheric nitrogen deposition has more than doubled in these areas
to around 30 kg nitrogen per ha per year, compared with areas of
healthy Sphagnum growth. In contrast to the situation in
continental western Europe, most of the nitrogen deposition in the
United Kingdom is of nitrogen oxides, although the importance of

ammonia/ammonium deposition is also increasing in the United Kingdom
(Fowler et al., 1980; Sutton et al., 1993). Several studies on bogs
in the United Kingdom have shown that increased supplies of nitrogen
are rapidly absorbed and utilized by bog-mosses (Sphagnum),
reflecting the importance of nitrogen as a nutrient and its scarcity
in unpolluted regions (Woodin et al., 1985; Woodin & Lee, 1987). The
high nitrogen loadings are, however, supraoptimal for the growth of
many characteristic Sphagnum species, as demonstrated by restricted
development in growth experiments and transplantation studies between
clean and polluted locations. In areas with high nitrogen loads, such
as the Pennines, the growth of Sphagnum is in general less than in
unpolluted areas (Lee & Studholme, 1992). After transplantation of
Sphagnum from an unpolluted site to a bog in the southern Pennines,
a rapid increase in plant nitrogen content from around 12 to 20 mg/g
dry weight was observed (Press et al., 1986). A large increase in
arginine in the shoots of these bog-mosses was also found after
application of nitrogen. In field experiments in northern and
southern parts of Sweden, nitrogen was found to be the limiting factor
for the growth of Sphagnum. However, other nutrients, especially
phosphorus, may become secondarily limiting to plant growth when
nitrogen inputs reach a threshold (Aerts et al., 1992).

A further possible effect of the increased nitrogen content of
Sphagnum is an increased decay rate of the peat, as nitrogen content
strongly influences decomposition rates (Swift et al., 1979). The
decay rate of Sphagnum peat in Swedish ombrotrophic bogs has been
studied along a gradient of nitrogen deposition (Hogg et al., 1994).
The results of this short-term decay experiment indicated that the
decomposition rate of Sphagnum peat is more influenced by the
phosphorus content of the material than by its nitrogen content,
although some relation with nitrogen supply has been observed.
Further evidence is necessary to evaluate the long-term effects of
enhanced nitrogen supply on the decay of peat.

All these studies strongly indicate the detrimental effects of
atmospheric nitrogen on the development of the bog-forming Sphagnum
species. However, enhanced nitrogen deposition can influence the
competitive relationships in nutrient-deficient vegetation such as
bogs. The effects of the supply of extra nitrogen on the population
ecology of Drosera rotundifolia has been recently studied in a
4-year experiment in Swedish ombrotrophic bogs (Redbo-Torstensson,
1994). It was demonstrated that experimental applications of more
than 10 kg nitrogen (as NH₄NO₃) per ha per year clearly affected the
population of this insectivorous species: the establishment of new
individuals and the survival of the plants significantly decreased in
the vegetation treated with extra nitrogen. This decrease in the
total population density of the characteristic bog species Drosera
was not caused by toxic effects of nitrogen, but by enhanced

competition for light with tall species such as Eriophorum and
Andromeda, which responded positively to the increased nitrogen
inputs.

On the basis of the United Kingdom and Scandinavian studies, it
has become clear that increased nitrogen loads strongly affect
ombrotrophic bog ecosystems, especially because of the high nitrogen
retention capacity and closed nitrogen cycling. The growth of
bog-mosses is reduced, directly by nitrogen and indirectly by a
changed competitive relationship between the prostrate dominants
(e.g. Eriophorum) and the subordinate plant species. A reliable
critical load for nitrogen in these ombrotrophic bogs is 5-10 kg
nitrogen per ha per year, although additional long-term studies with
enhanced nitrogen (both nitrogen oxides and ammonia/ammonium) are
necessary to validate this figure.

4.2.2.2 Effects on mesotrophic fens

Fens are wetland ecosystems that are typical of alkaline to
slightly acidic habitats in many countries. The alkalinity is due to
groundwater draining from surrounding rocks or sediments that are
relatively rich in calcium carbonate. Most of these fen ecosystems
are characterized by rare and endangered plants species. The effects
of nitrogen enrichment upon mesotrophic fens have been intensively
studied in the Netherlands (Verhoeven & Schmitz 1991; Koerselman &
Verhoeven, 1992). These fen ecosystems are characterised by many
Carex species and are rich in forbs (e.g., Pedicularis palustris;
orchids). Most of these Dutch fen ecosystems are managed as hay
meadows, with removal of the plant material further restricting
nutrient levels, and are now nature reserves.

A considerable increase of tall graminoids (grass or Carex
species), with a somewhat higher potential growth rate, was observed
after experimentally adding nitrogen to three Dutch fen ecosystems
(Vermeer, 1986; Verhoeven & Schmitz, 1991). This increase caused a
significant decrease in the diversity of subordinate plant species.
In one of the Dutch fen sites investigated, which had a long history
of hay making, it has been shown that phosphorus deficiency was also a
major factor in the productivity of the system, since there was a high
output of phosphorus from the ecosystem with the hay (Verhoeven &
Schmitz, 1991; Koerselman & Verhoeven, 1992). Using the results of
fertilization trials and nutrient budget studies in these fen
ecosystems (Koerselman et al., 1990; Koerselman & Verhoeven, 1992),
with their relatively closed nitrogen cycle, it seems reasonable to
establish a critical load of 20-35 kg nitrogen per ha per year, based
upon the output of the nitrogen from the fen system via normal
management. In some fen ecosystems, the critical nitrogen load based
on the change in diversity may be substantially higher, because of the
limitation of productivity by phosphorus (Egloff, 1987; Verhoeven &
Schmitz, 1991). In this situation, however, the risks of nitrogen

losses to surface water or groundwater will increase because of
phosphorus limitation, and this effect should be taken into account in
critical load evaluation. High rates of denitrification could also
influence the establishment of critical loads for these fen
ecosystems, and this aspect needs further investigation.

4.2.2.3 Effects on fresh- and saltwater marshes

In the wetland ecosystems discussed above, the nitrogen cycle is
more closed than that of intertidal marshes. The data on atmospheric
nitrogen inputs to the nitrogen cycling in intertidal fresh- and
saltwater marshes (with large prostrate graminoids as species of
Spartina, Typha and Carex) have been reviewed by Morris (1991).
It has become evident that nitrogen inputs to these marsh ecosystems
via surface water (well above 100 kg nitrogen per ha per year) are
much higher than the atmospheric loading. In non-tidal freshwater
marshes, it has been demonstrated in many studies that denitrification
is very high with a very large output of nitrogen from the ecosystem
(Morris, 1991). Because of the combined effect of these processes,
atmospheric nitrogen deposition is of only minor importance for these
marshes, and it is not useful to establish a critical load for
airborne nitrogen to these systems. In his review Morris (1991)
formulated a critical load for atmospheric nitrogen in wetland
ecosystems of around 20 kg nitrogen per ha per year. It is more
appropriate to make a distinction for different types of wetlands, as
shown above. An overview of the critical loads for wetlands is given
in chapter 8.

4.2.3 Effects on species-rich grasslands

Almost all of the research on the effects of atmospheric
deposition on terrestrial vegetation has focused on ecosystems
(e.g., forest, heathland and bogs) involving poorly buffered acidic
soils. Semi-natural grasslands with traditional agricultural use have
also been an important part of the landscape in western and central
Europe, and contain, or used to contain, many rare and endangered
plant and animal species. A number of these grasslands have been set
aside as nature reserves in several European countries (Ellenberg,
1988b; Woodin & Farmer, 1993). These semi-natural grasslands, which
are of conservation interest, are generally nutrient-poor because of
long agricultural use with low levels of manure and the removal
of plant growth by grazing or hay making. The vegetation is
characterized by many low growing species and is of nutrient-poor soil
status (Ellenberg, 1988b). Although these grasslands are nowadays
rare, the proportion of endangered plant and animal species in these
communities is very high (Van Dijk, 1992). Many experiments have
shown that application of artificial fertilizer (nitrogen, phosphorus
and potassium) changes these grasslands into tall, species-poor
stands, dominated by a few highly productive crop grasses (Van Den
Bergh, 1979; Willems, 1980; Van Hecke et al., 1981). To maintain a

large diversity of species, addition of fertilizer has to be avoided.
It is thus to be expected that these species-rich grasslands will be
affected by increased atmospheric nitrogen input (Wellburn, 1988;
Liljelund & Torstensson, 1988; Ellenberg, 1988b).

Many semi-natural grassland types are present in western and
central Europe. Most of these grasslands belong to the so-called
neutral grasslands (Molinio-Arrhenateretea; moist to moderately dry),
to the dry calcareous grasslands (Festuca-Brometea) or to the acid
grasslands on very poor soils (Nardetalia). Detailed descriptions
have been given by Ellenberg (1988b) and Van Dijk (1992). To obtain
critical loads for nitrogen for all these grasslands, it would be
essential to study the effects of nitrogen eutrophication in a
representative range within these communities. Detailed data are,
however, scarce. Therefore, the results of an integrated research
programme on nitrogen eutrophication in Dutch calcareous grasslands
are used as a target study in this chapter to obtain (i) information
on observed changes in this type of grassland caused by enhanced
nitrogen input, and (ii) a reliable estimation of the critical load
for nitrogen in these well-buffered non-acidic grasslands. The
results of this calcareous grassland study will be discussed in
respect to other semi-natural grasslands.

4.2.3.1 Effects of nitrogen on calcareous grasslands

Calcareous grasslands are communities on limestone. The subsoils
consist of various kinds of limestone with high contents of calcium
carbonate (> 90%), covered by shallow well-buffered rendzina soils
(A/C-profiles; pH of the top soil is 7-8 with a calcium carbonate
content of around 10%). The depth of the soil varies between 10 and
50 cm and the availability of nitrogen and phosphorus is low. The
grasslands are generally found on slopes with limestone in the subsoil
and a deep groundwater table. Plant productivity is low, and the peak
standing crop is in general between 150 and 400 g/m². The canopy of
the vegetation is open and low (10-20 cm). Calcareous grasslands are
among the most species-rich plant communities in Europe and contain a
large number of rare and endangered species. The area of these
semi-natural grasslands has decreased substantially in Europe during
the second half of this century (Wolkinger & Plank, 1981; Ratcliffe,
1984). Some remnants have become nature reserves in several European
countries. To maintain the characteristic calcareous vegetation a
specific management is needed to prevent their natural succession
towards woodland (Wells, 1974; Dierschke, 1985). The calcareous
grasslands in the Netherlands are mown in autumn with removal of the
hay (Bobbink & Willems, 1987).

a) Nitrogen enrichment and vegetation composition

The effects of nitrogen enrichment in Dutch calcareous grasslands
on vegetation composition have been investigated in two field

experiments (Bobbink et al., 1988; Bobbink, 1991). Either potassium
(100 kg per ha per year), phosphorus (30 kg per ha per year) or
nitrogen (100 kg per ha per year), as well as a complete fertilization
(nitrogen, phosphorus and potassium), were applied for 3 years to
study the long-term effects on vegetation composition. Nitrogen was
given as ammonium nitrate and was applied to two nature reserves with
opposite aspects (north and south). Total above-ground biomass
increased considerably, as expected, after three years of nitrogen,
phosphorus and potassium fertilization. In the experiments where
the nutrients were applied individually, a moderate increase in
above-ground dry weight was only seen with nitrogen addition: (about
330 g/m² compared with about 210 g/m² in the untreated plots). The
dry weight distribution of the species was significantly affected by
nutrient treatments. In the nitrogen-treated vegetation, the dry
weight of the grass species Brachypodium pinnatum was about 3 times
higher than in the control plots. Nitrogen application also resulted
in a drastic reduction of the biomass of forb species (including
several Dutch Red List species) and of the total number of species.
The observed decrease in species diversity in the nitrogen-treated
vegetation was not caused by nitrogen toxicity, but by the change in
vertical structure of the grassland vegetation. The growth of
Brachypodium was strongly stimulated and its overtopping leaves
reduced the light within the vegetation. It overshadowed the other
characteristic species and growth of these species declined rapidly
(Bobbink et al., 1988; Bobbink, 1991). Stimulation of Brachypodium
growth and a substantial reduction in species diversity were observed
following application of nitrogen to a 5-year permanent plot study
using a factorial design (Willems et al., 1993).

Many characteristic lichens and mosses have also disappeared in
recent years from European calcareous grasslands (During & Willen,
1986). This has been caused partly by the indirect effects of extra
nitrogen inputs, as shown experimentally by Van Tooren et al. (1990).
Data on the effects of nitrogen eutrophication on the species-rich
fauna of calcareous grassland are not available. However, it is very
likely that the diversity of animals, especially of insects, will also
be reduced when tall grasses are strongly dominating the vegetation,
because of the decreasing abundance of many herbaceous flowering
species which act as host or forage plants.

b) Nitrogen enrichment and nutrient storage in calcareous grasslands

The seasonal distribution of nutrients after nitrogen
fertilization in spring (120 kg nitrogen as ammonium nitrate) has been
studied with the repeated harvest approach (Bobbink et al., 1989).
It resulted in a significantly increased peak standing crop of
Brachypodium . This grass proves to have very efficient nitrogen
uptake and very efficient withdrawal from its senescent shoots into
its well-developed rhizome system. Brachypodium can benefit from the
extra nitrogen redistributed to the below-ground rhizomes by enhanced

growth in the next spring. The distribution of nitrogen has also been
quantified in 3-year fertilization experiments. Brachypodium
greatly monopolized (> 75%) the nitrogen storage in both the
above-ground and below-ground compartments of the vegetation with
increasing nitrogen availability (Bobbink et al., 1988; Bobbink,
1991).

Nitrogen cycling and accumulation in calcareous grassland can be
significantly influenced by two major outputs from the system:
(i) leaching from the soil; and (ii) removal with management regimes.
Nitrogen losses by denitrification in dry calcareous grasslands are
low (< 1 kg nitrogen per ha per year), owing to the well-drained soil
conditions (Mosier et al., 1981). Ammonium and nitrate leaching has
been studied in Dutch calcareous grasslands by Van Dam et al. (1992).
Both the water fluxes and the solute fluxes at different soil depths
have been measured over 2 years in untreated vegetation and in
calcareous grassland vegetation sprayed at 2-weekly intervals with
ammonium sulfate (50 kg nitrogen per ha per year). The nitrogen
leaching from the untreated vegetation was very low (0.7 kg nitrogen
per ha per year) and amounted to only 2% of the total atmospheric
deposition of nitrogen. After the spraying with ammonium sulfate,
nitrogen leaching increased significantly to 3.5 kg nitrogen per ha
per year, although this figure was also a very small proportion (4%)
of the nitrogen input in this vegetation (Van Dam et al., 1992). It
is thus evident that calcareous grassland ecosystems retain nitrogen
almost completely in the system. This is caused by a combination of
enhanced plant uptake (Bobbink et al., 1988; Bobbink, 1991) and
increased immobilization in the soil organic matter (Van Dam et al.,
1992).

4.2.3.2 Critical loads for nitrogen in calcareous grasslands

The most important output of nitrogen from calcareous grassland
is via exploitation or management. The annual nitrogen removal in the
hay varies slightly between years and sites, but in general between
17 and 22 kg nitrogen per ha is removed from the system under normal
management conditions in the Netherlands (Bobbink, 1991; Bobbink &
Willems, 1991). The ambient nitrogen deposition in Dutch calcareous
grasslands, as determined by Van Dam (1990), is high (35-40 kg
nitrogen per ha per year) and is nowadays the major nitrogen input to
the system. Legume species (Leguminosae) also occur in calcareous
vegetation, and form an additional nitrogen input owing to the
nitrogen-fixing microorganisms in their root nodules (about 5 kg
nitrogen per ha per year).

The nitrogen mass balance of Dutch calcareous grasslands is
summarized in Table 25. It is obvious that calcareous grasslands now
significantly accumulate nitrogen (16-26 kg per ha per year) in the
Netherlands. A critical nitrogen load has been determined with a mass
balance model, because of the lack of long-term addition experiments

with low nitrogen loads. Assuming a critical long-term immobilization
rate for nitrogen of 0-6 kg nitrogen per ha per year, the critical
nitrogen load can be derived by adding the nitrogen fluxes due to net
uptake, denitrification and leaching, corrected for the nitrogen input
by fixation. In this way, 15-25 kg nitrogen per ha per year is
considered as nitrogen critical load for this ecosystem. Nitrogen
cycling within the system (between plants and soil) is not used for
this calculation.

Table 25. Nitrogen mass balance (kg nitrogen per ha per year)
for dry calcareous grassland in the Netherlands

Input Output

Atmospheric deposition 35-40 Harvest 17-22
Nitrogen fixation 5 Denitrification 1
Soil leaching 1

Total 40-45 Total 19-24

In calcareous grassland in England, addition of nitrogen
stimulated the dominance of grasses in most cases (Smith et al., 1971;
Jeffrey & Pigott, 1973). In these studies, the application of
50-100 kg nitrogen per ha per year resulted in a strong dominance of
the grasses Festuca rubra, F. ovina or Agrostis stolonifera.
However, Brachypodium and Bromus erectus, the most frequent
species in calcareous grassland in continental Europe, were absent
from these sites. Following a survey of data from a number of
conservation sites in southern England, Pitcairn et al. (1991)
concluded that Brachypodium had expanded in the United Kingdom
during the last 100 years. They considered that much of the early
spread could be attributed to a decline in grazing pressure but that
the more recent spread had, in some cases, taken place despite grazing
or mowing, and could be related to nitrogen inputs. However, a study
of chalk grassland at Parsonage Downs (United Kingdom) showed no
substantial change in species composition over the twenty years
between 1970 and 1990, a period when nitrogen deposition is thought to
have increased significantly (Wells et al., 1993). Brachypodium was
present in the sward but had not expanded as in the Dutch grasslands.
In a linked experimental study, applications of nitrogen to eight
forbs and one grass (Brachypodium) at levels of 20, 40 and 80 kg
nitrogen per ha per year for two years did not result in
Brachypodium becoming dominant.

Apart from the Dutch studies, the effects of enhanced nitrogen
inputs have been little investigated in continental European
calcareous grasslands. Some data from a recent fertilization

experiment at the alvar grasslands, a thin-soiled vegetation over flat
limestone, on the Swedish island Öland, suggest that the vegetation
hardly responds to applications of nitrogen or phosphorus (Sykes & Van
der Maarel, 1991; personal communication by Van der Maarel). Only
irrigation in combination with nutrients has caused an increase in
grasses. This is probably due to the low annual precipitation in this
area (400-500 mm).

Increased nitrogen availability is probably of major importance
in many European calcareous grasslands. An increased availability of
nitrogen is indicated by enhanced growth of some tall grasses,
especially stress-tolerant species, which have a slightly higher
potential growth rate and efficient nitrogen utilization. It clearly
depends on the original species composition, as to which of the
grass species will increase following enhanced nitrogen inputs.
Furthermore, the difference between the Dutch and United Kingdom
results may reflect differences in management; the impacts of grazing
in the United Kingdom grasslands could offset any competitive
advantage the grasses may have obtained from additional nitrogen
inputs. The critical load for nitrogen in these calcareous grasslands
could be influenced by management; long-term studies involving
additional nitrogen input with various management schemes are needed
to quantify these aspects.

4.2.3.3 Comparison with other semi-natural grasslands

Productivity in grasslands is strongly influenced by nutrients,
as shown in many agricultural studies (e.g. Chapin, 1980). It is also
well-known that large amounts of fertilizer (nitrogen, phosphorus and
potassium) alter almost all grassland types into tall, species-poor
swards dominated by a few highly productive crop grasses (e.g. Bakelaar
& Odum, 1978; Van Den Bergh, 1979; Willems, 1980; Van Hecke et al.,
1981). Most of these species-rich grasslands are deficient in
nitrogen or phosphorous, and thus characterized by plant species of
nutrient-poor habitats. It is thus likely that these grasslands are
sensitive to nitrogen eutrophication from the atmosphere (Wellburn,
1988; Ellenberg, 1988b). Thus, it is also important to establish
critical loads for nitrogen in the species-rich grasslands, although
data from experiments with nitrogen application in these semi-natural
grasslands are scarce.

Increased nitrogen availability can also affect other
semi-natural grasslands, although experimental evidence is quite
scarce. A classical study into the effects of nutrients on neutral
grasslands is the Park Grass experiment at Rothamsted, England, which
has been running since 1856 (Williams, 1978). After application of
nitrogen as ammonium sulfate or sodium nitrate (48 kg nitrogen per ha
per year), the vegetation became very poor in species and dominated by
grasses such as Holcus lanatus or Agrostis sp. The effects of
nutrients in dry and wet dune grasslands (1% calcium carbonate) on

sandy soils have been studied at Braunton Burrows (Devon, England) by
Willis (1963). Nutrients were applied over 2 years (6 × 40 kg
nitrogen per ha per year) using a factorial design for nitrogen and
phosphorus. Nitrogen proved to be the most important nutrient in
stimulating the growth of some grass species ( Festuca rubra and Poa
pratensis). This enhanced growth reduced significantly the abundance
of many small plants such as prostrate phanerogamic species, mosses
and lichens (Willis, 1963). In this coastal area with low nitrogen
deposition (currently about 10 kg nitrogen per ha per year) the
vegetation of dune grasslands is at present still species-rich,
whereas in many Dutch dune grasslands with higher nitrogen loading
(20-30 kg nitrogen per ha per year) certain grasses have increased and
it has become a problem to maintain diversity. Recent studies of the
response of mesothrophic grasslands in the United Kingdom have shown
that additions as small as 25 kg per ha per year can lead to changes
in species diversity after several years of fertilizer additions and
that changes take place more rapidly at higher rates of addition
(Mountford et al., 1994). This indicates that many of these
semi-natural grasslands are also sensitive to nitrogen eutrophication
and that the critical loads are likely to be of the same magnitude or
slightly higher (20-30 kg nitrogen per ha per year) than in calcareous
grasslands.

Many other semi-natural grassland types occur, especially in the
montane-subalpine regions, containing a large proportion of the
biodiversity of the area. However, data are too scarce to establish
reliable load for these grasslands, although it may be expected that:
(i) most of these grassland are sensitive to nitrogen; and (ii) the
critical load for nitrogen is probably lower than for lowland
(calcareous) grasslands. The presented critical loads for
species-rich grasslands are summarized in section 8.2.2.

4.2.4 Effects on heathlands

Various types of plant communities have been described as heath,
but the term is applied here to plant communities where the dominant
vegetation is small-leaved dwarf-shrubs forming a canopy of 1 m or
less above soil surface. Grasses and forbs may form discontinuous
strata, and there is frequently a ground layer of mosses or lichens
(Gimingham et al., 1979; De Smidt, 1979). Dwarf-shrub heathlands
occur in various parts of the world, especially in montane habitats,
but are widespread in the atlantic and sub-atlantic parts of Europe.
In these parts of the European continent, natural heathland is limited
to a narrow coastal zone. Inland lowland heathlands are man-made
(semi-natural), although they have existed for several centuries.
Lowland healths are widely dominated by the Ericaceae, especially
Calluna vulgaris in the dry heathlands and Erica tetralix in the
wet heathlands (Gimingham et al., 1979). In these heaths, development
towards woodland has been prevented by mowing, burning, sheep grazing
and sod removal.

Until the beginning of this century, the balance of nutrient
input and output was in equilibrium in the lowland heathlands of
western Europe (De Smidt, 1979; Gimingham & De Smidt, 1983). The
original land use implied a regular, periodic removal of nutrients
from the ecosystems via grazing and sod removal (Heil & Aerts, 1993).
Sod removal was practised less systematically in many Scandinavian and
Scottish heathlands (Gimingham & De Smidt, 1983). Here Calluna has
been renewed by burning at regular intervals, varying from 4-6 years
in southern Sweden to 15-20 years in western Norway (Nilsson, 1978;
Skogen, 1979). The original land use of the lowland heathland ceased
in the early 1900s and the area occupied by this community decreased
markedly all over its distribution area (Gimingham, 1972; De Smidt,
1979; Ellenberg, 1988b). Dwarf-shrub heathlands may be divided into
four categories according to broad differences in habitat: (1) dry
heathlands; (2) wet heathlands; (3) montane and (4) arctic-alpine
heathlands.

4.2.4.1 Effects on inland dry heathlands

During recent decades many lowland heathlands in western Europe
have become dominated by grass species. An evaluation, using aerial
photographs, has shown that more than 35% of Dutch heathland has been
altered into grassland (Van Kootwijk & Van der Voet, 1989). In recent
years, similar changes have been observed in SW Norway, which has the
largest local emission of ammonia as well as the heaviest nitrogen
input through long-range deposition in Norway (Anonymous, 1991). It
has been suggested that nitrogen eutrophication might be a significant
factor in this transition to grasslands. Field and laboratory
experiments confirm the importance of nutrients, especially in the
early phase of heathland development (Heil & Diemont, 1983; Roelofs
1986; Heil & Bruggink, 1987; Aerts et al., 1990). However, Calluna
can compete successfully with the grasses, even at high nitrogen
loading, if its canopy remains closed (Aerts et al., 1990). Apart
from the changes in competitive interactions between Calluna and the
grasses, heather beetle plagues and nitrogen accumulation in the soil
are important factors in the changing lowland heaths. Furthermore,
evidence is growing that frost sensitivity of the dominant
dwarf-shrubs may also be affected by increasing nitrogen inputs.

Heathland canopies have a strong filtering effect on air
pollutants, a varying canopy structure being an important factor. For
sulfur and nitrogen it has been shown that bulk deposition accounts
for only about 35-40% of total atmospheric input (Heil et al., 1987;
Bobbink et al., 1992b). Total atmospheric deposition of nitrogen is
30-45 kg nitrogen per ha per year in the heathland sites in the
eastern part of the Netherlands. More than 70% of the total nitrogen
input is deposited as ammonium or ammonia (Bobbink et al., 1992b;
Bobbink & Heil, 1993). Although data for nitrogen inputs in other

European lowland heathlands are missing, it is very likely that in
many European agricultural regions nitrogen deposition has increased
in recent years (Asman, 1987; Buijsman et al., 1987).

In Calluna heathland, outbreaks of the chrysomelid heather
beetle (Lochmaea suturalis) occur frequently. These beetles feed
exclusively on the green parts of Calluna. The closed Calluna
canopy is opened over large areas and the interception of light by
Calluna decreases strongly (Berdowski, 1987, 1993). Thus the growth
of the under-storey grasses ( Deschampsia or Molinia) is enhanced
significantly. In general insect grazing is affected by the nutritive
value of the plant material, and the nitrogen content is especially
important in this respect (Crawley, 1983). Experimental applications
of nitrogen to heathland vegetation cause the concentrations of this
element in the green parts of Calluna to increase (Heil & Bruggink,
1987; Bobbink & Heil, 1993). It is, therefore, very likely that the
frequency and intensity of heather beetle outbreaks are stimulated by
increased atmospheric nitrogen input in Dutch heathland. This
hypothesis is supported by the observations of Blankwaardt (1977), who
reported that from 1915 onwards heather beetle outbreaks were observed
in the Netherlands with an interval of about 20 years, whereas in the
last 15 years the outbreaks have occurred with a periodicity of less
than 8 years. It has also been observed that during a heather beetle
outbreak Calluna plants are more severely damaged in nitrogen-
fertilized vegetation (Heil & Diemont, 1983). In a rearing experiment
with larvae of the heather beetle, Brunsting & Heil (1985)
demonstrated that the growth of the larvae was increased by higher
nitrogen concentrations in the leaves of Calluna. Van der Eerden
et al. (1990) studied the effects of ammonium sulfate on the growth of
heather beetle after a outbreak of the beetle in vegetation
artificially sprayed under a cover. They found no significant effect
of the treatments on total number or on biomass of the first stage
larvae. However, the development of subsequent larval stages was
accelerated by the application of ammonium sulfate in the artificial
rain: the percentage of third stage larvae increased by 20%, compared
with larvae in the control treatment. Furthermore, heather beetle
larvae were put on Calluna shoots taken from plants which had been
fumigated with ammonia in open-top chambers (12 months; 4 to
105 µg/m³) (Van der Eerden et al., 1991). After 7 days the larvae
were counted and weighed. Both the mass and the development rate of
the larvae clearly increased with increasing concentrations of
ammonia. The heather beetle has recently been found in SW Norway and
it is expanding its territory. It is probably an important cause of
Calluna death in this region (Hansen, 1991). It can be concluded
that nitrogen inputs influence outbreaks of heather beetle, although
the exact relationship between both processes needs further research.

In the past Dutch inland heathlands were grazed by flocks of
sheep and sods were generally removed at intervals of 25-50 years
(De Smidt, 1979). Nowadays these heathlands are mostly managed by

mechanical sod removal, which can restore the heathland vegetation if
a seed bank of the original heathland species is still present
(Bruggink, 1993). The increase in organic matter and in the amounts
of nitrogen in the system during secondary succession is well known
(Begon et al., 1990). It was shown in the 1970s that during secondary
heathland succession the above-ground and below-ground biomass and the
amount of litter increase (Chapman et al., 1975; Gimingham et al.,
1979). It is likely that changes in nitrogen accumulation will have
occurred due to the increase in atmospheric deposition.

Berendse (1990) performed a detailed study on the accumulation of
organic matter and of nitrogen during the secondary succession after
sod removal in the Netherlands. He found a large increase in plant
biomass, soil organic matter and total nitrogen storage in the first
20 to 30 years after sod removal. Furthermore, it was demonstrated
that nitrogen mineralization was low during the first 10 years (about
10 kg nitrogen per ha per year), but increased considerably over the
next 20 years to 50-110 kg nitrogen per ha per year. Regression
analysis of the total nitrogen storage versus the years after sod
removal revealed an annual nitrogen increase in the system of about
33 kg nitrogen per ha per year (probably somewhat lower in the early
years and higher in later years) (Berendse, 1990). These values are
in good agreement with measured nitrogen deposition in Dutch
heathlands in the late 1980s (Bobbink et al., 1992b).

Thus, the organic matter in the soil increases rapidly after sod
removal, which removes almost all of the soil organic matter.
However, this process is accelerated by the enhanced dry matter
production and litter production of the dwarf shrubs caused by the
extra nitrogen inputs. Nitrogen accumulation in the system also
increases. Hardly any nitrogen disappears from the system because
nitrate leaching to deeper layers is only of minor importance in Dutch
heathlands, as shown by De Boer (1989) and Van Der Maas (1990).
Nitrogen availability from atmospheric inputs, in addition to
mineralization, is within a relatively short period (about 10 years)
high enough to stimulate the transition of heathland to grassland,
especially after the opening of the heather canopy by secondary
causes.

It has been demonstrated that frost sensitivity of some tree
species increases with increasing concentrations of air pollutants
(e.g. Aronsson, 1980; Dueck et al., 1991). This increase in frost
sensitivity is sometimes correlated with enhanced nitrogen
concentrations in the foliage of the trees. Long-term effects of air
pollutants on the frost sensitivity of Calluna and Erica are to be
expected because of (i) the evergreen growth form of these species and
(ii) the increasing content of nitrogen in the leaves of Calluna,
associated with increased nitrogen deposition in the Netherlands and
Norway (Heil & Bruggink, 1987; Hansen, 1991). It has been suggested
that damage to Calluna shoots in the successive severe winters of
the mid-1980s was at least partly caused by the increased frost

sensitivity. Investigations into the effects of air pollutants on the
frost sensitivity of heathland species outside the Netherlands started
in the early 1990s (Hansen, 1991; Uren, 1992).

The effects of sulfur dioxide, ammonium sulfate and ammonia upon
frost sensitivity in Calluna have been studied by Van der Eerden
et al. (1990). After fumigation with sulfur dioxide (90 µg/m³ for
3 months), increased frost injury in Calluna was only found at
temperatures that seldom occur in the Netherlands (lower than -20°C).
Fumigation with ammonia of Calluna plants in open-top chambers over
a 4-7 month period (100 µg/m³) revealed that frost sensitivity was
not affected in autumn (September or November), whereas in February,
just before growth started, frost injury increased significantly at
-12°C (Van der Eerden et al., 1991). These authors also studied the
frost sensitivity of Calluna vegetation sprayed with six different
levels of ammonium sulfate (3-91 kg nitrogen per ha per year). The
frost sensitivity increased slightly, although significantly, after
5 months in vegetation treated with the highest level of ammonium
sulfate (400 µmol/litre; 91 kg nitrogen per ha per year), compared
with the control treatments. However, frost sensitivity of Calluna
decreased again two months later and no significant effects of the
ammonium sulfate application upon frost hardiness were seen at that
time. Thus, high levels of ammonia or ammonium sulfate seem to
increase the frost sensitivity of Calluna plants, although the
significance of this phenomenon is still uncertain at ambient nitrogen
inputs. The relation between frost sensitivity and nitrogen input has
not yet been sufficiently quantified to use it for a precise
assessment of critical loads in this respect.

It has been shown above that atmospheric nitrogen is the trigger
for changes of lowland dry heathlands into grass swards in the
Netherlands. Lowland dry heathlands in the United Kingdom do not show
consistent patterns over the past 10 to 40 years. Pitcairn et al.
(1991) assessed changes in abundance of Calluna in three heaths in
East Anglia (eastern England) over recent decades. All three heaths
showed a decline in Calluna and an increase in grasses. The authors
concluded that increases in nitrogen deposition was at least partly
responsible for the changes, but also noted that the management had
changed. A wider assessment of heathlands in SE England showed that
in some cases Calluna had declined and subsequently been invaded by
grasses while other areas were still dominated by dwarf shrubs (Marrs,
1993). This clearly stresses the importance of management for the
maintenance of dwarf shrubs in heathlands. A simulation model, which
integrates processes such as atmospheric nitrogen input, heather
beetle outbreak, soil nitrogen accumulation, sod removal and
competition between species, has been used to establish the critical
loads of nitrogen deposition in lowland dry heathlands (Heil &
Bobbink, 1993a,b). The model has been calibrated with data from field
and laboratory experiments in the Netherlands. As an indicator of the
effects of atmospheric nitrogen, the proportion and increase of

grasses in the heathland system are used. Atmospheric nitrogen
deposition has varied between 5 and 75 kg nitrogen per ha per year in
steps of 5-10 kg nitrogen during different simulations. From these
simulations, the value for the critical load of nitrogen for the
changes from dwarf shrubs to grasses was 15-20 kg nitrogen per ha per
year.

4.2.4.2 Effects of nitrogen on inland wet heathlands

The western European lowland heathlands of wet habitats are
dominated by the dwarf shrub Erica tetralix (Ellenberg, 1988b) and
are generally richer in plant species than the dry heathlands. In
recent decades a drastic change in species composition of Dutch wet
heathlands has been observed. Nowadays, many wet heathlands that were
originally dominated by Erica have become monospecific stands of the
grass Molinia. Together with Erica almost all of the rare plant
species have disappeared from the system. It has been hypothesized
that this change has been caused by atmospheric nitrogen
eutrophication.

Competition experiments using micro-ecosystems have clearly shown
that Molinia is a better competitor than Erica at high nitrogen
availability. After 2 years of application of nitrogen (150 kg per ha
per year), the relative competitive strength of Molinia compared
with Erica doubled (Berendse & Aerts, 1984). A 3-year field
experiment with nitrogen application in Dutch lowland wet heathland
(around 160 kg nitrogen per ha per year) also indicated that Molinia
is able to outdo Erica at high nitrogen availability (Aerts &
Berendse, 1988). In contrast to the competitive relations between
Calluna and the grasses, Molinia can outdo Erica without opening
of the dwarf shrub canopy. This difference is caused by the lower
canopy of Erica (25-35 cm), compared with Calluna, and the tall
growth form of Molinia, which can overgrow and shade Erica if
enough nitrogen is available. It is in this respect also important
that heather beetle plagues do not occur in wet heathlands and that no
frost damage has been observed in this community.

It has been demonstrated that in many Dutch wet heathlands the
accumulation of litter and humus has led to increased nitrogen
mineralization (100-130 kg nitrogen per ha per year) (Berendse et al.,
1987). In the first 10 years after sod removal the annual nitrogen
mineralization is very low, but afterwards it increases rapidly. The
leaching of accumulated nitrogen from wet heathlands is extremely low
(Berendse, 1990). The observed nitrogen availabilities are high
enough to change Erica -dominated wet heathlands into monostands of
Molinia.

Berendse (1988) developed a wet heathland model to simulate
carbon and nitrogen dynamics during secondary succession. He
incorporated in this model the competitive relationships between

Erica and Molinia, the litter production from both species, soil
nitrogen accumulation and mineralization, leaching, atmospheric
nitrogen deposition and sheep grazing. He simulated the development
of lowland wet heathland after sod removal, because almost all of the
Dutch communities are already strongly dominated by Molinia and it
is impossible to expect changes in this situation without drastic
management. Using the biomass of Molinia with respect to Erica as
an indicator, his results suggested 17-22 kg nitrogen per ha per year
as the critical load for the transition of lowland wet heathland into
a grass-dominated sward (Berendse, 1988). The decrease in endangered
wet heathland forbs is partly caused by the overshading by Molinia,
but some species had already disappeared from wet heathlands before
the increase of Molinia started. The critical load for this decline
is probably lower than the given values and is discussed in section
4.2.4.4.

4.2.4.3 Effects of nitrogen on arctic and alpine heathlands

Semi-natural Calluna heathlands are found in the lowlands along
the Norwegian coast to 68°N and show distinct plant gradients in the
south-north direction, from coast to inland and from lowland to upland
areas (Fremstad et al., 1991). In central parts of western Norway the
plant composition changes at an altitude of about 400 m, above which
alpine species occur regularly in the heaths. At this altitude
oceanic upland Calluna and Erica heaths merge into alpine heaths,
which are naturally occurring, non-anthropogenic communities. Some
oligotrophic alpine heaths also contain Calluna, but most heaths in
Fennoscandia and in European parts of Russia are dominated by other
ericoid species ( Vaccinium spp., Empetrum nigrum s. lat.,
Arctostaphylos spp., Loiseleuria procumbens, Phyllodoce caerulea,
Betula nana, Juniperus communis and Salix spp.). Many heath types
have a more or less continuous layer of mosses and lichens. Related
heaths are found in alpine regions in the British Isles, in Iceland,
in southernmost Greenland, in northern Russia, and on siliceous rocks
in the Alps (Grabherr, 1979; Elvebakk, 1985; Ellenberg, 1988b).

Alpine and arctic habitats have many ecological characteristics
in common, although the climatic conditions are more severe in the
arctic regions than in most alpine regions. The growing season is
short (3-3.5 months in the low arctic zone), air and soil temperatures
are low, winds are frequent and strong, and the distribution of plant
communities depends on the distribution of snow during winter and
spring. Most alpine and all arctic zones are influenced by frost
activity or solifluction, except for soils in the low alpine and
hemiarctic zones, where podzolic soils are found. Decomposition of
organic matter and nutrient cycling are slow, and a large amount of
the nitrogen input is stored in the soil in forms which can not be
used by plants (Chapin, 1980). The low nutrient availability limits

primary production. Most species are adapted to a strict nitrogen
economy and their nitrogen indicator values are generally low
(Ellenberg, 1979).

Barsdate & Alexander (1975) investigated the nitrogen balance of
an arctic area in Alaska. The most important sources of nitrogen were
nitrogen fixation (75%) and ammonia in precipitation (22%). Most of
the nitrogen input is retained in living biomass, and very little is
leached from the soil. Denitrification is also low, partly due to
nutrient deficiency. Nitrogen metabolism as such does not seem to be
inhibited by low soil temperatures (Haag, 1974). Nitrogen fixation in
arctic habitats has been studied in bacteria, soil algae, lichens and
legume species (Leguminosae) (Novichkova-Ivanova, 1971). Blue-green
algae (cyanobacteria) are especially important in this respect, either
as free-living species, species associated with mosses or phycobionts
in lichens (e.g. Peltigera, Nephroma and Stereocaulon). The rate
of nitrogen fixation depends on temperature and moisture, and thus
varies through the year (Alexander & Schnell, 1973).

It is to be expected that arctic and alpine communities are
sensitive to increased atmospheric nitrogen input, because nitrogen
retention is very efficient, although primary production is also
strongly regulated by factors other than nitrogen (temperature,
moisture) (Tamm, 1991). The effects of increased nitrogen
availability on alpine/tundra vegetation have been studied in several
fertilizer experiments. In most experiments full nitrogen, phosphorus
and potassium fertilizer was used, although sometimes nitrogen was
applied separately. The following effects of nitrogen addition have
been observed:

* In alpine and arctic vegetation, nitrogen is quickly absorbed by
phanerogamic species and incorporated into their tissues. The
increase in nitrogen contents differs for graminoids, deciduous
and evergreen species (Summers, 1978; Shaver & Chapin, 1980;
Lechowicz & Shaver, 1982; Karlsson, 1987).

* Phanerogamic plant species respond to nitrogen application in
different ways: increased growth and biomass, enhanced number of
tillers, more flowers and changes in phenology (Henry et al.,
1986).

* Some phanerogamic plant species are damaged or even killed at
high doses of nitrogen fertilizer (Henry et al., 1986).

* Changes in species cover and composition are likely when nitrogen
is applied for a longer period of time (5-10 years).

All these studies concentrated on effects on phanerogamic plant
species; little information is available on the effects of nitrogen on
cryptogams. Many authors, however, stress that nitrogen fixation
probably will decrease when atmospheric deposition increases in arctic

and alpine ecosystems. In forest studies it has already been shown
that Cladonia spp. and some mosses are very sensitive to nitrogen
addition. The suggested critical load for arctic and alpine heaths
(5-15 kg nitrogen per ha per year) is lower than that for lowland
heathland (15-20 kg nitrogen per ha per year).

4.2.4.4 Effects on herbs of matgrass swards

In recent decades, in addition to the transition from
dwarf-shrub-dominated to grass-dominated heathlands, a reduced species
diversity in these ecosystems has been observed. Species of the
acidic Nardetalia grasslands and related dry and wet heathlands seem
to be especially sensitive. Many of these herbaceous species (e.g.,
Arnica montana, Antennaria dioica, Dactylorhiza maculata, Gentiana
pneumonanthe, Genista pilosa, Genista tinctoria, Lycopodium inundatum,
Narthecium ossifragum, Pedicularis sylvatica, Polygala serpyllifolia
and Thymus serpyllum) are declining or have even become locally extinct
in the Netherlands. The distribution of these species is related to
small-scale, spatial variability of the heathland soils. It has been
suggested that atmospheric deposition has caused such changes (Van Dam
et al., 1986). Dwarf shrubs as well as grass species are nowadays
dominant in the former habitats of these endangered species.

Enhanced nitrogen fluxes into nutrient-poor heathland soil
leads to an increased nitrogen availability in the soil. However,
most of the deposited nitrogen in western Europe originates from
ammonia/ammonium deposition and may also cause acidification as a
result of nitrification. Whether eutrophication or acidification or a
combination of both processes is important depends on pH, buffer
capacity and nitrification rates of the soil. Roelofs et al. (1985)
found that, in dwarf-shrub-dominated heathland soils, nitrification is
inhibited at pH 4.0-4.2 and that ammonium accumulates while nitrate
decreases to almost zero at these or lower pH values. Furthermore,
nitrification has been observed in the soils from the habitats of the
endangered species, owing to its somewhat higher pH and higher buffer
capacity. In soils within the pH rage of 4.1-5.9, the acidity
produced is buffered by cation exchange processes (Ulrich, 1983). The
pH will drop when calcium is depleted, and this may cause the decline
of those species that are generally found on soils with somewhat
higher pH. To study the pH effects on root growth and survival rate,
hydroculture experiments have been conducted over 4-week periods with
several of the endangered species ( Arnica, Antennaria, Viola,
Hieracium pilosella and Gentiana) and with the dominant species
( Molinia and Deschampsia) (Van Dobben, 1991). The dominant
species indeed have a lower pH optimum (3.5 and 4.0, respectively)
than the endangered species (4.2-6.0). However, the endangered
species could survive very low pH without visible injuries during this
short experimental period.

The pH decrease may indirectly result in an increased leaching of
base cations, increased aluminium mobilization and thus enhanced
aluminium/calcium (Al/Ca) ratios of the soil (Van Breemen et al.,
1982). Furthermore, the reduction of the soil pH may inhibit
nitrification and result in ammonium accumulation and consequently
increased NH₄/NO₃ ratios. In a recent field study the
characteristics of the soil of several of these threatened heathland
species have been compared with the soil characteristics of the
dominant species ( Calluna vulgaris, Erica tetralix and Molinia
caerulea) (Houdijk et al., 1993). Generally the endangered species
grow on soil with higher pH, lower nitrogen content, and lower Al/Ca
ratios than the dominant species. The NH₄⁺/NO₃ ratios were higher
in the dwarf-shrub-dominated soils than in the soil of the endangered
species. Fennema (1990, 1992) has demonstrated that soil from
locations where Arnica is still present had a higher pH and lower
Al/Ca ratio than soil of former Arnica stands. However, he found no
differences in total soil nitrogen or NH₄/NO₃ ratios. Both these
studies indicate that high Al/Ca ratios or even increased NH₄/NO₃
ratios are associated with the decline of these species. However, no
significant effects of Al and Al/Ca on growth rates have been observed
in hydroculture experiments in which the effects of Al and Al/Ca
ratios on root growth and survival rate were studied (Van Dobben,
1991). Comparable experiments of Pegtel (1987) with Arnica and
Deschampsia and Kroeze et al. (1989) with Antennaria, Viola,
Filago minima, and Deschampsia showed similar results. However,
results of a hydroculture experiment with Arnica showed that this
species is very sensitive to enhanced Al/Ca ratios at intermediate or
low nutrient levels (De Graaf, 1994). Pot experiments have indicated
that increased NH₄/NO₃ ratios lead to decreased health of Thymus.
Hydroculture experiments with this plant species confirmed that
increased NH₄/NO₃ ratios affected the cation uptake (Houdijk, 1993).
In a pot experiment Thymus, planted on acid heathland soil and on
artificially buffered heathland soil, was sprayed with 0, 15 and
150 kg nitrogen (as ammonium) per ha per year during 6 months (Houdijk
et al., 1993). In this relatively short period, a deposition of 15 kg
nitrogen (as ammonium) per ha per year on the acid soil did not lead
to ammonium accumulation in the soil. As a result of nitrification,
soil pH decreased faster than in the absence of ammonium deposition.
At the highest deposition (150 kg nitrogen (as ammonium) per ha per
year), nitrification rates in the acid heathland soils were too low to
prevent ammonium accumulation, and increased NH₄/NO₃ ratios probably
caused the decline of Thymus. Only in the artificially buffered
soils with higher pH were nitrification rates high enough to balance
ammonium and nitrate. Thymus plants on these soils were healthy
despite very high total nitrogen contents.

At present, however, there is too little information available on
these rare heathland and acidic grassland species to formulate a
critical load for nitrogen. The observation that these heathland

species generally disappear before dwarf shrubs are replaced by
grasses leads to the assumption that their critical load is lower than
the critical load for the transition to grasses (< 15-20 kg nitrogen
per ha per year) and probably between 10 and 15 kg nitrogen per ha per
year. An overview of the critical loads in heathlands is given in
section 8.2.2.

4.2.5 Effects of nitrogen deposition on forests

4.2.5.1 Effects on forest tree species

The growth of the vast majority of the forest tree species in the
Northern hemisphere was until recently limited by nitrogen. In
forestry, nitrogen fertilizers were used to increase wood production
(Tamm, 1991). An increase in the supply of an essential nutrient,
including nitrogen, will stimulate tree growth; the initial impact of
enhanced nitrogen deposition will, therefore, be a fertilizer effect.
However, continued high inputs of nitrogen produces negative effects
on tree growth (Chapin, 1980). Until the mid-1980s, almost all of the
research on forest decline focused on acidification, but it has now
become evident that enhanced nitrogen deposition may also be important
in recent forest decline.

The effects of high atmospheric nitrogen input are very complex
(Wellburn, 1988; Pitelka & Raynal, 1989; Heij et la., 1991; Pearson &
Stewart, 1993). Chronic nitrogen deposition may result in nitrogen
saturation, when enhanced nitrogen inputs no longer stimulate tree
growth, but start to disrupt ecosystem structure and function, and
increased amounts of nitrogen are lost from the ecosystem in leachate
(Agren, 1983; Aber et al., 1989; Tamm, 1991). The nitrogen input at
which saturation occurs depends on a number of factors including the
amount of deposition, vegetation type and age (see chapter 3), soil
type and management history. The following indirect processes,
besides the direct effect of gaseous pollutants on the shoots, are
important:

* Soil acidification, due to nitrification of ammonium. This
process leads to accelerating leaching of base cations and, in
poorly buffered soils, to increased dissolution of aluminium,
which can damage fine roots development and mycorrhizas, and thus
reduce nutrient uptake (Ulrich, 1983; Ritter, 1990).

* Eutrophication. Whether ammonium will accumulate in soil or
not is strongly dependent upon the nitrification rate and the
deposition levels (Boxman et al., 1988). In addition to an
initial growth stimulation and changes in root/shoot ratio,
ammonium accumulation will lead to an imbalance of the
nutritional state of the soil and concomitantly of the trees
(Roelofs et al., 1985; Van Dijk & Roelofs, 1988; Schulze et al.,
1989; Boxman et al., 1991). Accumulation of nitrates in the

ecosystem may also lead to eutrophication. As a consequence of
all these processes, the health of the trees declines and their
sensitivity to drought, frost, insect pests and to pathogens can
increase markedly (Wellburn, 1988). These phenomena may also
play a secondary, but certainly not unimportant, role in the
dieback of forest trees and have also been reviewed.

Although many tree species occur in natural forest ecosystems,
almost all studies on air pollution have concentrated on a few
forestry tree species from acidic, nutrient-poor soils. Most of these
species are conifers ( Picea, Pinus and Pseudotsuga spp.) and the
following section concentrates on the long-term soil-mediated effects
on these trees. Available data on broad-leaved species ( Fagus,
Quercus) are also considered. Long-term effects of nitrogen
eutrophication on the composition of the tree layer in natural forests
may be expected but have not yet been quantified. Soil acidification
per se has only been briefly reviewed, because the critical load for
acidity and tree growth is well established (Nilsson & Grennfelt,
1988; Downing et al., 1993).

a) Soil-mediated changes in nutritional status of forest tree species

It has been shown that in areas with high ammonia/ammonium
deposition, ammonium accumulates in acid forest soils with little or
no nitrification. Van Dijk & Roelofs (1988) found ammonium ion
accumulation in damaged Pinus and Pseudotsuga stands receiving
60-100 kg nitrogen per ha per year, although the pH of the soil was
the same as that in healthy stands. This build-up of ammonium ion
leads to increased ratios of ammonium to base cations (Roelofs et al.,
1985; Boxman et al., 1988), a reduction of base cation uptake and,
eventually, nutritional problems. Using soil columns with different
ammonium sulfate spraying treatments, critical ratios of excess
ammonium to base cations have been determined (Boxman et al., 1988).
The nutritional problems of the coniferous species studied have been
found above values of 5, 10 and 1, respectively, for the NH₄/K,
NH₄/Mg and Al/Ca ratios in soil solution. In soil with zero or a low
nitrification rate, 10-15 kg nitrogen per ha per year is a reliable
critical load to prevent critical ammonium to cation ratios, whereas
in base-cation-rich soil with moderate to high nitrification rates the
critical loads obtained are higher (20-30 kg nitrogen per ha per
year).

The nutritional status of the coniferous trees studied, after
enhanced nitrogen inputs, is affected by both ammonium accumulation
and soil acidification. Base cation concentrations in the soil are
reduced by leaching, whereas base cation uptake by plants is reduced
by excess of ammonium and of aluminium. Furthermore, root growth is
decreased (see later). Laboratory, greenhouse and field measurements
in the Netherlands, Germany and southern Sweden (Van Dijk & Roelofs,
1988; Van Dijk et al., 1989, 1990, 1992a; Hofmann et al., 1990;

Schulze & Freer-Smith, 1991; Boxman et al., 1991, 1994; Ericsson et
al., 1993) have shown that the complex of factors just noted produce
severe deficiencies of magnesium and potassium in coniferous trees.
Most of these studies were in areas, or involved experiments, with
large inputs (> 40-100 kg nitrogen per ha per year).

The magnesium and phosphorus concentrations in leaves of oak
trees (Fagus sylvatica), a common deciduous tree in Europe,
decreased significantly from 1984 to 1992 in permanent plots in NW
Switzerland. Furthermore, the magnesium concentrations in the leaves
of young Fagus sylvatica decreased significantly within a 4-year
period of fertilizer application at > 25 kg nitrogen per ha per
year (Flückiger & Braun, 1994). In Sweden, suboptimal concentrations
of magnesium and potassium in Fagus leaves were found in areas with
the highest nitrogen deposition (Balsberg-Pählsson, 1989) and addition
of nitrogen enhanced nutritional imbalance in a 120-year-old Fagus
stand (Balsberg-Pählsson, 1992). It is thus clear that this deciduous
tree species is also sensitive to nutritional imbalance induced by
enhanced nitrogen supply.

Base cations are also lost from the canopy by increased leaching,
linked to high amounts of atmospheric deposition (Wood & Bormann,
1975; Roelofs et al., 1985; Bobbink et al., 1992b). As a result of
high nitrogen inputs, the organic nitrogen concentration in the
needles of conifers has increased significantly to supra-optimal
levels (Van Dijk & Roelofs, 1988; De Kam et al., 1991). Concentrations
of nitrogen-rich free amino acids, especially arginine, have
significantly increased in the needles with high nitrogen concentration
(> 1.5% nitrogen in Picea abies) (Hällgren & Näsholm, 1988;
Pietila et al., 1991; Van Dijk et al., 1992) and in Fagus leaves
(Balsberg-Pählsson, 1992).

Although there is clear evidence that high NH₃/NH₄ loads
produce adverse changes in the nutritional status and the growth of
the investigated coniferous and broad-leaved trees, it is difficult to
obtain a critical load for nitrogen from these studies, because of the
complexity of the ecosystem. A quite reliable critical load for
nitrogen deposition on beech tree health is around 15-20 kg nitrogen
per ha per year, as demonstrated in the Swiss studies (Flückiger &
Braun, 1994).

The results of the EC nitrogen saturation study (NITREX), which
incorporates long-term experiments in both clean and nitrogen-polluted
areas and whole ecosystem manipulation of nitrogen inputs, are
providing important evidence on the effects of nitrogen deposition
on tree health and ecosystem health. Atmospheric deposition of
nitrogen was reduced from 40 to 2 kg nitrogen per ha per year in a
nitrogen-saturated Pinus sylvestris stand in the Netherlands (Boxman
et al., 1994, 1995). Throughfall water was intercepted with a roof
and replaced by clean throughfall water from 1989 onwards. In the

clean plot a quick response of the soil solution chemistry was
observed. The nitrogen concentrations in the upper soil and the
fluxes of this element through the soil profile decreased. As a
result, base cation leaching and the ratios of ammonium to various
cations also decreased; potassium and magnesium concentrations in the
needles increased significantly. The needle nitrogen concentrations
were only slightly reduced in the "clean" situation, but they were
significantly lower than in the needles of the control plots. The
concentration of arginine decreased significantly in the needles of
the trees from the clean throughfall plot. Furthermore, tree growth
became higher after 4 years of clean throughfall than in control plots
with high nitrogen deposition. No changes in the mycorrhizal status or
in the undergrowth have so far been observed (Boxman et al., 1994,
1995). This study clearly demonstrates the detrimental effects of
enhanced atmospheric nitrogen deposition on the nutritional balance of
coniferous trees.

b) Nitrogen deposition and tree susceptibility to frost, drought and
pathogens

It has been suggested by several authors that sensitivity of
trees to secondary stress factors is increased by high nitrogen
loading (Wellburn, 1988; Pitelka & Raynal, 1989). In field fertilizer
applications it is often observed that tree growth starts earlier in
the season, which may increase damage by late frost. Furthermore, it
has been shown, after nutrient applications, that frost damage to
Pinus sylvestris increases considerably at needle nitrogen
concentrations above 1.8% (Aronsson, 1980), although other fertilizer
studies have demonstrated reverse effects, i.e. improved nitrogen
status of the plants diminishes frost damage (De Hayes et al., 1989;
Klein et al., 1989; Cape et al., 1991).

Only few data are available with respect to frost damage in
direct relation to airborne nitrogen deposition. After exposure to
NH₃ and SO₂, Pinus sylvestris saplings became more frost sensitive
(< -10°C) than control plants (Dueck et al., 1990). Dueck et al.
(1990) also determined the frost sensitivity of Pinus sylvestris
growing in areas with low ammonia/ammonium pollution (approximately
4 µg NH₃/m³) and in highly polluted areas (40 µg NH₃/m³).
Surprisingly, the frost sensitivity was not higher in the polluted
area than in the other investigated sites, and was sometimes even
lower. After experimental treatment with ammonia (53 µg NH₃/m³) the
growth of the trees had increased, indicating that the observed change
in frost sensitivity might have occurred as a result of changes in
physiology and nutrient imbalance.

The effects of simulated acid mist containing sulfate, ammonium,
nitrate and H⁺ on the frost sensitivity of Picea rubens has been
studied (Sheppard et al., 1993; Sheppard, 1994). There was a strong

correlation between the application of sulfate-containing mist and an
increase in frost sensitivity, but no such correlation was seen after
treatment with ammonium or nitrate ions. Sulfur compounds clearly
affect the frost sensitivity of coniferous trees, but this effect may
be a consequence of the nutritional status (nitrogen, base cations) of
the trees (Sheppard, 1994). It is concluded that the effects of
increased nitrogen inputs on frost sensitivity remain uncertain.
Insufficient research has been carried out to use the results for
assessment of a critical load.

The water uptake of coniferous trees species may be affected by
increasing nitrogen deposition, owing to an increase in shoot-to-root
ratio and a reduction in fine-root length. Indeed, the health of many
tree species in the regions of the Netherlands with high nitrogen
deposition was particularly poor in the dry years in the mid-1980s,
but improved again during the subsequent normal years (Heij et al.,
1991). Many authors have mentioned a negative impact of high nitrogen
supply on the development of fine roots and mycorrhiza, although
positive effects have also been described (Persson & Ahlstrom, 1991).

Van Dijk et al. (1990) applied 0, 48, 480 kg nitrogen (as
ammonium sulfate) per ha per year to young Pinus sylvestris, Pinus
nigra and Pseudotsuga menziesii in a pot experiment. After seven
months the coarse root biomass had not changed, but the fine root
biomass decreased by 36% at the highest nitrogen application. In
parallel, a 63% decrease in mycorrhizal infection at the highest
nitrogen application was found. In the Dutch EC nitrogen saturation
study, the fine root biomass and the number of root tips of Pinus
sylvestris increased after reduction of the current nitrogen
deposition to pre-industrial levels, indicating restricted root growth
and nutrient uptake capacity at the ambient nitrogen load of about
40 kg nitrogen per ha per year (Boxman et al., 1994, 1995).

In a hydroculture experiment with Pinus nigra at pH=4.0, Boxman
et al. (1991) found an increase in coarse/fine root ratio after
increasing the ammonium concentration to 5000 µM. Furthermore, a
clear relation was found between the nitrogen content of the fine
roots and mycorrhizal infection (as measured as the number of
dichotomously branched roots). In a hydroculture experiment Jentschke
et al. (1991) found, however, that 2700 µM nitrate had hardly any
effect on the mycorrhizal development of Picea abies seedlings
inoculated with Lactarius rufus. Ammonium at 2700 µM only had a
slight negative effect on mycorrhizal development, whereas a reduction
in root growth was recorded. In a pot experiment with Picea abies,
Meyer (1988) found optimal mycorrhizal development when the mineral
nitrogen content of the soil was 40 mg nitrogen/kg dry soil, while at
350 mg nitrogen/kg dry soil a 95% reduction in mycorrhizal development
was found. In this study no correlation was found with the soil pH.
Alexander & Fairly (1983) found, after fertilizer application to a
35-year-old Picea sitchensis stand with 300 kg nitrogen (as ammonium

sulfate) per ha, a 15% reduction in mycorrhizal development in the
second year after application. Termorshuizen (1990) applied 0 to
400 kg nitrogen ha per year either as ammonium or nitrate to young
Pinus sylvestris inoculated with Paxillus involutus in a pot
experiment. Above application rates of 10 kg nitrogen per ha per year
there was a decrease in the amount of mycorrhizal root tips and the
number of sclerotia.

In addition to the above-mentioned data for coniferous trees, it
had been shown that the shoot-to-root ratios of young Fagus
sylvatica trees, grown in containers with acid forest soil,
increased significantly from about 1 to between 2 and 3 after a 4-year
experimental application of nitrogen (25 kg nitrogen per ha per year
or more) (Flückiger & Braun, 1994).

It is thus likely that enhanced nitrogen inputs affect drought
sensitivity through changes in shoot to root ratios, number of fine
roots and the ectomycorrhizal infection of the roots. However, the
data are too few to use for the assessment of a critical load of
nitrogen, based upon this aspect of reduced tree health.

There may also be significant effects of fungal pathogens or
insect pests associated with increasing nitrogen deposition. The
foliar concentrations of nitrogen increased markedly in tree needles
or leaves in experiments with nitrogen additions, and also in forest
sites with high atmospheric nitrogen loading (Roelofs et al., 1985;
Van Dijk & Roelofs, 1988; Balsberg-Pählsson, 1992). Animal grazing
generally increases with increasing palatability of the leaves or
shoots. Nitrogen is of major importance for the palatability of plant
material, and this certainly holds for insect grazing (Crawley, 1983).
Secondary plant chemicals, e.g., phenolics, are important for
increased resistance of plants. The total amount of phenolics in
Fagus leaves in a 120-year stand decreased by more than 30% after
fertilizer application of about 45 kg nitrogen per ha per year,
compared with the control treatment (Balsberg-Pählsson, 1992). An
ecologically important relation between nitrogen enrichment and insect
pests has been quantified for lowland heathland (Brunsting & Heil,
1985; Berdowski, 1993, see section 4.1) but not, so far, for forest
ecosystems.

From 1982 to 1985 an epidemic outbreak of the pathogenic fungus
Sphaeropsis sapinea was observed in coniferous forest (mainly
Pinus nigra) in the Netherlands. This greatly affected whole
stands, and was especially severe in the south-east part of the
Netherlands, where there was high airborne nitrogen deposition
(Roelofs et al., 1985). Van Dijk et al. (1992) showed that there was
a significantly higher foliar nitrogen concentration in the infected
stands, together with higher soil ammonium levels, than in the
uninfected stands. Most of the additional nitrogen in the needles of
the affected stands was stored as nitrogen-rich free amino acids,

especially arginine. Proline concentrations were also higher in the
infected trees, indicting a relation with water stress (Van Dijk
et al., 1992).

The effects of Sphaeropsis have also been studied by De Kam et
al. (1991). Two-year-old plants of Pinus nigra were grown for
3 years in pots and given five treatments of ammonium sulfate (very
low to about 300 kg nitrogen per ha per year), in combination with two
levels of potassium sulfate. The 5-year-old plants were then
inoculated with Sphaeropsis. The bark necroses were much more
frequent in the plants treated with ammonium sulfate than in the
controls. Effects of ammonium sulfate upon fungal damage were even
observed at an addition of 75 kg nitrogen per ha per year, but were
very significant in the plants treated with 150 kg nitrogen per ha per
year. After potassium addition the number of necroses caused by the
fungus was greatly reduced (De Kam et al., 1991).

In beech forests in NW Switzerland, a significant positive
correlation has been found between the nitrogen/potassium ratios in
the leaves and necroses caused by the beech cancer Nectria ditissima
(Flückiger & Braun, 1994). These authors also experimentally
inoculated Fagus sylvatica trees at different applications of
nitrogen with this beech cancer and observed increased dieback of new
leaves and shoots. Furthermore, the infestation of Fagus sylvatica
with beech aphids (Phyllaphis fagi) was also affected by the
nitrogen availabilities. The degree of infestation with the aphid
increased significantly with enhanced leaf nitrogen/potassium ratios
(Flückiger & Braun, 1994). Although evidence for nitrogen-mediated
changes in susceptibility to fungal pests and insect attacks has until
now been based upon observations of only few species, it is obvious
that trees became more susceptible to these attacks with increasing
nitrogen enrichment and this may play a crucial role in the dieback of
some forest stands.

A critical load for nitrogen had been established at 10-15 kg
nitrogen (at no or low nitrification) to 20-30 kg nitrogen per ha per
year in highly nitrifying soils, based upon nutritional imbalance of
coniferous species (Boxman et al., 1988). Recent evidence of Fagus
sylvatica tree health in acidic forests indicated a critical load
of 15-20 kg nitrogen per ha per year, based upon both field and
experimental observations. Elevated nitrogen deposition can
seriously affect tree healthy via a complex web of interactions (e.g.
susceptibility to frost and drought). Pathogens may play an important
role in tree decline, but at this moment it is not possible to combine
the observed processes and effects to an overall value for a critical
load of nitrogen for tree health.

4.2.5.2 Effects on tree epiphytes, ground vegetation and ground fauna
of forests

a) Effects on ground-living and epiphytic lichens and algae

The effects of SO_y as an acidifier on epiphytic lichens have
been extensively studied (Insarova et al., 1992; Van Dobben, 1993).
SO_y was previously the dominant airborne pollutant, and it has been
shown that most (epiphytic) lichens are more negatively affected by
acidity than by nitrogen compounds (except NO_y). Most lichens have
green algae as photobionts and are affected by acidity but not by
nitrogen. Some of them even react positively to nitrogen (Insarova et
al., 1992). However, 10% of all lichen species in the world have
cyanobacteria (blue-green algae) as the photobiont. These
cyanobacterial lichens are negatively affected by acidity, and also by
nitrogen. Most of the NW European lichens with cyanobacteria live on
the soil surface or are tree epiphytes. The most pollution-sensitive
lichens are among them and they are threatened by extinction in NW
Europe. This is probably the result of increased nitrogen deposition,
which inhibits the functioning of the cyanobacteria. In the
Netherlands, for example, all cyanobacterial lichens that were present
at the end of the 19th century are now absent. In Denmark, 96% of the
lichens with cyanobacteria are extinct or threatened. Furthermore,
the cyanobacterial lichens appear frequently on the Red List of the
European Union countries (Hallingbäck, 1991).

Very few data exist to establish a critical load for nitrogen for
these lichens with blue-green algae. Nohrstedt et al. (1988)
investigated the effects of nitrogen application (as ammonium nitrate
or calcium nitrate) on ground-living lichens ( Peltigera aphtosa and
Nephroma arcticum) with blue-green algae as photobionts. The plots
were treated once or three or four times with 120, 240 or 360 kg
nitrogen per ha. After a short period all Peltigera and Nephroma
lichens were eliminated and even 19 years later no recolonization had
occurred. However, it is impossible to transform these very high
doses to critical loads. The effects of air pollutants on lichens are
usually related to concentrations in the air or in the precipitation.
It is probably more relevant to relate the effects of nitrogen on
cyanobacterial lichens to deposition than to concentrations. For tree
epiphytes stemflow is most relevant, whereas for ground-living lichens
throughfall will be more important. Although much research is still
needed, it has been suggested that a load of 5-15 kg nitrogen per ha
per year is already critical for the growth of these cyanobacterial
lichens (Hallingbäck, 1991). These lichens may be the most sensitive
components of some forest ecosystems and thus determine the critical
load for these systems.

Free-living green algae, especially of the genus Pleurococcus
( Protococcus and Demococcus are synonyms), are strongly stimulated
by enhanced nitrogen deposition. They cover practically all outdoor

surfaces which are not subject to frequent desiccation in regions with
high nitrogen deposition, such as in the Netherlands and in Denmark.
The thickness and the colonization rate of spruce needles by green
algae has been investigated in the Swedish Environmental Monitoring
Programme (Brakenhielm, 1991). The Swedish data show that these algae
do not colonize spruce needles in regions with a total deposition
(throughfall) lower than about 5 kg nitrogen per ha per year. In
areas with deposition above 20 kg nitrogen per ha per year, the green
algal cover of the needles is so thick and the algae colonize so early
that they may impede the photosynthesis of the spruce trees.

b) Effects on forest ground vegetation

In the Netherlands the forest vegetation of a site in the central
part of the country was investigated in 1958 (with about 20 kg
nitrogen per ha per year) and in 1981 (with about 40 kg nitrogen per
ha per year). All lichens had disappeared during this period and a
considerable increase in Deschampsia flexuosa and Corydalis
claviculata was found. A large representative sample test (n=2000),
covering about 90% of the Dutch forests, revealed in the mid-1980s
that among the 40 most common forest plants were: Galeopsis
tetrahit, Rubus species, Deschampsia flexuosa, Dryoptesis
cathusiana, Molinia caerulea, Poa trivialis, and Urtica dioica
(Dirkse & Van Dobben, 1989; Dirkse, 1993). In Sweden, Quercus robur
stands in two geographical areas with different nitrogen deposition
were compared with special emphasis on nitrogen indicator species
(Tyler, 1987). The stands were quite comparable except for the
nitrogen inputs: 6-8 kg nitrogen per ha per year and 12-15 kg nitrogen
per ha per year, respectively. In the stand with the highest
deposition, the soil solution was more acidic, probably due to acidic
deposition as well (± 10 kg sulfur per ha per year), and it was
estimated that acidification of the soil has accelerated during the
last 30 to 50 years. The following species were more common in the
most polluted site: Urtica dioca, Epilobium augustifolium, Rubus
idaeus, Stellaria media, Galium aparine, Aegopodium podagraria and
Sambucus spp. Thus, both in Sweden and the Netherlands, species
indicative of nitrogen enrichment became common (Ellenberg, 1988b).

Comparable observations were reported by Falkengren-Grerup (1986)
and by Falkengren-Grerup & Eriksson (1990), who examined the changes
in soil and vegetation in Quercus and Fagus stands in southern
Sweden. They concluded that the exchangeable base cations were reduced
and that aluminium had doubled over the past 35 years. They also
found a decrease in soil pH, with a disappearance of several species
when pH dropped below a threshold. In spite of soil acidification
some species had increased in cover, and the most plausible
explanation seemed to be increased nitrogen deposition, which was
about 15-20 kg nitrogen per ha per year in southern Sweden and which
had doubled since 1955. A marked increase in cover was found for

Lactuca muralis, Dryopteris filix-max, Epilobium augustifolium,
Rubus idaeus, Melica uniflora, Aegopodium podagraria, Stellaria
holostea and S. nemorum, some of these species being nitrogen
indicators. Despite soil acidification, acid-tolerant species
( Deschampsia flexuosa, Maianthemum bifolium and Luzula pilosa) did
not increase. A distinct decrease was observed for Dentaria
bulbifera, Pulmonaria officinalis and Polygonatum multiflorum.
Furthermore, Rosen et al. (1992) found a significant positive
correlation between the increase of Deschampsia flexuosa cover in
the last 20 years in the Swedish forests and the pattern of nitrogen
deposition.

In a large semi-natural Fagus-Quercus forest in NE France,
about 50 permanent vegetation plots were investigated in 1972 and
1991. The changes in species composition on calcareous soils and in
moderately acidic habitats were followed. During the study period a
significant increase in nitrophilous ground flora was observed in the
high-pH (6.9) stands. This indicated that at this location (with
ambient deposition of 15-20 kg nitrogen per ha per year) there was a
distinct effect of increasing nitrogen availability (Thimonier et al.,
1994).

From 1968 to 1985, three sites in a 30-year-old Pinus
sylvestris forest in Lisselbo (central Sweden) were annually
fertilized with 0, 20, 40 and 60 kg nitrogen per ha per year (as
NH₄NO₃ plus ambient deposition of 10 kg nitrogen per ha per year).
The original ground vegetation consisted of Calluna vulgaris,
Vaccinium vitis-idea, V. myrtillus, Cladonia spp., Cladina spp.,
and the mosses Dicranum spp., Pleurozium spp. and Hylocomium
spp. The first changes were observed within 8 to 15 years and after
about 20 years the experimental plots were compared and statistically
analysed. The original species disappeared at nitrogen applications
above 20 kg (plus ambient deposition) nitrogen per ha per year and
were replaced by Epilobium augustifolium, Rubus idaeus, Deschampsia
flexuosa, Dryopteris carthusiana and the moss Brachythecium
oedipodium (Dirkse et al., 1991; Van Dobben, 1993). In another
experiment at Lisselbo the combined effects of acidification (addition
of H₂SO₄, pH=2.0) and nitrogen addition (0 and 40 kg nitrogen per ha
per year) were investigated. The increased nitrogen level seemed to
be the more important factor. Acidification was the next most
discriminating factor: all species disappeared, except for the moss
Pohlia nutans at high additions of acidity (Dirkse & Van Dobben,
1989; Dirkse et al., 1991).

In southern Sweden, Tyler et al. (1992) studied the effects of
the application of ammonium nitrate (60-180 kg nitrogen per ha per
year) over a 5-year period on stands of Fagus sylvatica. They
observed a large reduction in biomass of the ground vegetation with
the application of nitrogen, and the frequency of most herb layer
species declined significantly. Soil measurements revealed that, in

addition to eutrophication effects, the acidification of the soil
solution was also important for the decline of the original ground
vegetation. In an experiment on the effects of nitrogen fertilizer
application on bryophytes, it appeared that Brachythecium
oedipodium, B. reflexum and B. starkei increased significantly at
levels up to 60 kg nitrogen per ha per year. At higher doses these
species tended to decline, however. Hylocomium splendens and
Pleurozium schreberi declined considerably at doses of 30 to 60 kg
nitrogen per ha per year (Dirkse & Martaki, 1992).

c) Effects on macrofungi and mycorrhizas

During the last two decades many reports have described a
decrease in species diversity and abundance of macrofungi. These
changes can probably be attributed to indirect effects of air
pollution, in particular to increases in the amount of available
nitrogen (possibly in combination with acidification), and/or to
decreased health of trees with concomitant reduction of transport to
the roots (Arnolds, 1991).

When comparing sites over time, the number of fruiting bodies of
macrofungi showed marked differences. Most studies in western Europe,
however, have revealed that the number of ectomycorrhizal fungi
species has declined (Arnolds, 1991). In the Netherlands the average
number of ectomycorrhizal species per foray declined significantly
from 71 in 1912-1954 to 38 in 1973-1982. Similar changes have been
observed in Germany: 94 ectomycorrhizal species found in 1950-1979 in
the Völklinger area (Saarland) have not been recorded recently. From
the 236 species found in 1918-1942 in the Darmstadt area (Germany),
only 137 were recorded in the early 1970s, a loss of 99 species,
including many mycorrhizal fungi (Arnolds, 1991). In contrast to the
decline in mycorrhizal fungi, the number of saprotrophic species
remained practically unchanged, while the number of lignocolous
species increased. This may be related to soil acidification with a
increase in aluminium, since the proportion of forest areas in western
Europe with a soil pH below 4.2 increased from less than 1% in 1960 to
15% in 1988 (Schneider & Bresser, 1988).

Arnolds (1988, 1991) concluded that acidification has very little
effect on the diversity of ectomycorrhizal fungi, but rather triggers
changes in species composition. He regarded the increased nitrogen
flux to the forest floor as the most important factor in the decline
of mycorrhizal fungi. Termorshuizen & Schaffers (1987) found a
negative correlation between the total nitrogen input in mature
Pinus sylvestris stands and the abundance of fruit bodies of
ectomycorrhizal fungi. Similar results were obtained by Schlechte
(1986) who compared two sites with Picea abies in the Göttingen area
of Germany. An obvious negative relation was found between nitrogen
input (23 versus 42 kg nitrogen per ha per year) and ectomycorrhizal
species: 85 basidiomycetes including 21 ectomycorrhizas (25%) at the

less polluted site compared with 55 basidiomycetes including
3 ectomycorrhizas (5%) at the most polluted site. Environmental
factors other than nitrogen did not differ significantly. The
negative impact of nitrogen seems only to hold true for mature forests
(Termorshuizen & Schaffers, 1987). Jansen & de Vries (1988) found a
maximum in fruit-body production in > 20-year-old Pseudotsuga
menziesii stands at about 25 kg nitrogen per ha per year. Meyer
(1988) found a similar optimum when Picea abies was planted in soil
mixed with different amounts of sawdust having a high carbon/nitrogen
ratio.

Experiments with nitrogen fertilizer have produced similar
results. In a fertilizer trial with simulated nitrogen deposition in
a Fagus forest in southern Sweden (ambient deposition 15-20 kg
nitrogen per ha per year), Ruhling & Tyler (1991) found, after
applying NH₄NO₃ (60 and 180 kg nitrogen per ha per year), that
within 3 to 4 years almost all mycorrhizal species ceased fruit-body
production. In contrast, several decomposer species increased
fruit-body production. Wood decomposers showed no obvious reaction to
the treatment. No fruit-bodies were recovered when 300 kg nitrogen
per ha was applied to Pinus sylvestris stands as liquid manure
(Ritter & Tölle, 1978). The mycorrhizal frequency of the roots,
however, was still 55% as compared to 87% in the controls.
Application of 112 kg nitrogen (as NH₄NO₃) per ha to 11-year-old
Pinus taeda stands revealed an 88% reduction in the number of
fruit-bodies and a 14% decrease in the number of mycorrhizas per unit
of soil volume (Menge & Grand, 1978). In the Lisselbo study the
number of fruit-bodies decreased considerably at each nitrogen
fertilizer dose (Wasterlund, 1982). Termorshuizen (1990) applied
0, 30 and 60 kg nitrogen (as ammonium sulfate or nitrate) per ha per
year to young Pinus sylvestris stands. In general fruit-body
production was more negatively influenced by the higher ammonium
levels than nitrate levels. The mycorrhizal frequency and the number
of mycorrhizas per unit of soil volume were not influenced. It was
concluded by Termorshuizen (1990) that fruit-body production is much
more sensitive to nitrogen enrichment that mycorrhizal formation.
Branderud (1995) found after only 1.5 year a decrease in fruit-body
production of mycorrhizal species at a nitrogen application of 35 kg
nitrogen (as NH₄NO₃) per ha in a Picea abies stand at the Swedish
Nitrex stand.

In contrast, some studies have shown an increase in the number of
fruit-bodies of insensitive mycorrhizal fungi after nitrogen
fertilizer application, e.g., Paxillus involutes (Hora, 1959),
Laccaria bicolor (Ohenoja, 1988) and Lactarius rufus (Hora, 1959).

d) Effects on soil fauna of forests

Almost all studies of changes in faunal species composition due
to nitrogen enrichment have been conducted in arable fields or

agricultural grasslands using complete fertilization and thus cannot
be used to substantiate critical loads for semi-natural forest
ecosystems (Marshall, 1977). The relationship between acidity and
soil fauna has also been studied in northern coniferous forests, but
only very few studies have incorporated the effects of nitrogenous
compounds (Gärdenfors, 1987). The abundance of Nematoda,
Oligochaeta and microarthropods (especially Collembola) had
increased in some studies, but decreased in others, after application
of high doses of nitrogen fertilizers (> 150 kg nitrogen per ha per
year) (Abrahamsen & Thompson, 1979; Huhta et al., 1983; Vilkamaa &
Huhta, 1986). A reduction in the nitrogen deposition in a Pinus
sylvestris stand (Nitrex site Ysselstein) to pre-industrial levels
increased the species diversity of microarthropods due to a decreased
dominance of some species (Boxman et al., 1995). However, it is not
possible to use these few data to formulate a critical load for
changes in forest soil fauna due to increased nitrogen deposition.

On the basis of the results presented in this overview, the
critical load for changes in the ground vegetation of both coniferous
and deciduous acidic forest may be 15 to 20 kg nitrogen per ha per
year. The critical load for changes in the fruit-body production of
ectomycorrhizal fungi is probably about 30 kg nitrogen per ha per
year, while the critical load for changes in mycorrhizal frequency of
tree roots is hard to estimate, but certainly considerably higher.
There is insufficient data on the effects of enhanced nitrogen
deposition on faunal components of forest ecosystems to allow critical
loads to be set. Epiphytic or ground-living lichens with
cyanobacteria as the photobiont probably form a sensitive part of
forest ecosystems and have an estimated critical load of 10-15 kg
nitrogen per ha per year. A summary of the critical loads for forests
is given in chapter 8.

4.2.6 Effects on estuarine and marine ecosystems

Few topics in aquatic biology have received as much attention
over the past decade as the debate over whether estuarine and coastal
ecosystems are limited by nitrogen, phosphorus or some other factor
(Hecky & Kilham, 1988). Numerous geochemical and experimental studies
have suggested that nitrogen limitation is much more common in
estuarine and coastal waters than in freshwater systems. Taken as a
whole, the productivity of estuarine waters in the USA correlates more
closely with supply rates of nitrogen than with those of other
nutrients (Nixon & Pilson, 1983).

Estimation of the contribution of nitrogen deposition to the
eutrophication of estuarine and coastal waters is made difficult by
the multiple direct anthropogenic sources (e.g., from agriculture and
sewage) of nitrogen against which the importance of atmospheric
sources must be weighed. Estuaries and coastal areas are common
locations for cities and ports. The crux of any assessment of the

importance of nitrogen deposition to estuarine eutrophication lies in
establishing the relative importance of direct anthropogenic exposure
(e.g., sewage and agricultural run-off) and indirect effects
(e.g., atmospheric deposition).

The effects of nitrogen deposition in certain estuarine systems
have been investigated. Complete nitrogen budgets, as well as
information on nutrient limitation and seasonal nutrient dynamics,
have been compiled for two large "estuaries", the Baltic Sea
(Scandinavia) and the Chesapeake Bay (USA), and for the Mediterranean
Sea. In the case of the Mediterranean, Loye-Pilot et al. (1990)
suggest that 50% of the nitrogen load originates as deposition falling
directly on the water surface. In the case of the Baltic and
Chesapeake, deposition of atmospheric nitrogen has been suggested as a
major contributor to eutrophication. Data for other coastal and
estuarine systems are less complete, but similarities between these
two systems and other estuarine systems suggest that their results may
be more widely applicable. Discussion in this monograph is limited to
these two case studies, with some speculation about how other
estuaries may be related.

The Baltic Sea is perhaps the best-documented case study of the
effects of nitrogen additions in causing estuarine eutrophication.
Like many other coastal waters, the Baltic Sea has experienced a
rapidly increasing anthropogenic nutrient load. It has been estimated
that the supply of nitrogen has increased by a factor of 4, and
phosphorus by a factor of 8, since the beginning of the 20th century
(Larsson et al., 1985). The first observable changes attributable to
eutrophication of the Baltic were declines in the concentration of
dissolved oxygen in the 1960s (Rosenberg et al., 1990). Decreased
dissolved oxygen concentrations result when decomposition in deeper
waters is enhanced by the increased supply of sedimenting algal cells
from the surface water layers to the sediment. In the case of the
Baltic, the spring algal blooms that now result from nutrient
enrichment consist of large, rapidly sedimenting algal cells, which
supply large amounts of organic matter to the sediment for
decomposition (Enoksson et al., 1990). Since the 1960s, researchers
in the Baltic have documented increases in algal productivity,
increased incidence of nuisance algal blooms, and periodic failures
and unpredictability in fish and Norway Lobster catches (Fleischer &
Stibe, 1989; Rosenberg et al., 1990). It has now been shown by a
number of methods that algal productivity in nearly all areas of the
Baltic Sea is limited by nitrogen. Nitrogen-to-phosphorus ratios
range from 6:1 to 60:1 (Rosenberg et al., 1990), but the higher ratios
are only found in the remote and relatively unaffected area of the
Bothnian Bay (between Sweden and Finland). Productivity in the spring
(the season of highest algal biomass) is fuelled by nutrients supplied
from deeper waters during spring overturn (Graneli et al., 1990); deep
waters are low in nitrogen and high in phosphorus, resulting in
nitrogen-to-phosphorus ratios near 5 (Rosenberg et al., 1990),

suggesting potential nitrogen limitation when deep waters are mixed
with surface waters. Low nitrogen-to-phosphorus ratios in deep water
result from denitrification in the deep sediments (Shaffer & Rönner,
1984). Primary productivity measurements in the Kattegat (the portion
of the Baltic between Denmark and Sweden) correlate closely with
uptake of NO₃^-, but not of PO₄^3- (Rydberg et al., 1990). Level II
and III nutrient enrichment experiments conducted in coastal areas of
the Baltic, as well as in the Kattegat, indicate nitrogen limitation
at most seasons of the year (Graneli et al., 1990). Growth
stimulation of algae has also been produced by addition of rain water
to experimental enclosures, in amounts as small as 10% of the total
volume (Graneli et al., 1990); rain water in the Baltic is rich in
nitrogen but poor in phosphorus. In portions of the Baltic where
freshwater inputs keep the salinity low, blooms of the nitrogen-fixing
cyanobacterium Aphanizomenon flos-aquae are common (Graneli et al.,
1990); cyanobacterial blooms are common features of nitrogen-limited
freshwater lakes but are usually absent from marine waters.

Nitrogen budget estimates indicate that the Baltic Sea as a whole
receives 7.6 × 10¹⁰ eq of nitrogen per year, of which 2.8 × 10¹⁰ eq
per year (37%) comes directly from atmospheric deposition (Rosenberg
et al., 1990). Fleischer & Stibe (1989) reported that the nitrogen
flux from agricultural watersheds feeding the Baltic has been
decreasing since about 1980 but that the nitrogen contribution from
forested watersheds is increasing. They cite both increases in
nitrogen deposition and the spread of modern forestry practices as
causes for the increase. It should be noted, however, that the Baltic
also experiences a substantial phosphorus load from agricultural and
urban lands, and that phosphorus inputs may help to maintain
nitrogen-limited conditions (Graneli et al., 1990). If the Baltic had
received consistent nitrogen additions (e.g., from the atmosphere or
from agricultural run-off) in the absence of phosphorus additions, it
might well have evolved into a phosphorus-limited system some time
ago.

The physical structure of the Baltic Sea, with a shallow sill
limiting exchange of water with the North Sea contributes to the
eutrophication of the basin, by trapping nutrients in the basin once
they reach the deeper waters. Because the larger algal cells that
result from nutrient enrichment in the basin provide more nutrients to
the deep water through sedimentation, and because only shallow waters
have the ability to exchange with the North Sea, it is estimated that
less than 10% of nutrients added to the Baltic are exported over the
sill to the North Sea (Wulff et al., 1990). Throughout much of the
year (i.e., especially during the dry months) productivity in the
Baltic is maintained by nutrients recycled within the water column
(Enoksson et al., 1990). The trapping of nutrients within the basin
and recycling of nutrients from deeper water by circulation patterns

suggest that eutrophication of the Baltic is a self-accelerating
process (Enoksson et al., 1990) and has a long time-lag between
reductions of inputs and improvements in water quality.

In the USA, a large effort has been made to establish the
relative importance of sources of nitrogen to Chesapeake Bay (D'Elia
et al., 1982; Smullen et al., 1982; Fisher et al., 1988; Tyler, 1988).
Estimates of the contribution of nitrogen to Chesapeake Bay from each
individual source are very uncertain; estimating the proportion of
nitrogen deposition exported from forested watersheds is especially
problematic but critical to the analysis, because about 80% of the
Chesapeake Bay basin is forested. Nonetheless, three attempts at
determining the proportion of the total nitrate load to the Bay
attributable to nitrogen deposition all produce estimates in the range
of 18 to 31%. Supplies of nitrogen from deposition exceed supplies
from all other non-point sources to the Bay (e.g., agricultural
run-off, pastureland run-off, urban run-off), and only point source
inputs represent a greater input than deposition.

It is considered that the data from these studies are indicators
of the impact of anthropogenic nitrogen. Nevertheless, they are
insufficient to estimate critical loads for estuarine/marine systems.
It may well by that critical loads for these systems differ for
different climatic regions.

4.2.7 Appraisal and conclusions

Atmospheric deposition of nitrogen-containing and acidifying
compounds have an impact on soil and groundwater quality and on the
health and species composition of vegetation. Critical loads for
these effects are given in Table 26. Critical loads have been derived
using empirical data that relate loads directly to effects and
steady-state soil models that calculate critical loads from critical
chemical values for ion concentrations or ratios in foliage, soil
solution and groundwater (De Vries, 1993). Information on the effects
which occur when critical loads are exceeded is given in Table 27.
The values given in Tables 26 and 27 apply to forest vegetation in a
temperate climate. Whether they are representative of other climates
is uncertain. An overview of the critical loads for atmospheric
nitrogen deposition in a range of natural and semi-natural ecosystems
is given in chapter 8.

Effects of nitrogen and acidifying deposition on soil and
groundwater chemistry are most evident. Field studies showed that
deposited nitrogen is partly retained in the forest soil. Even at
high nitrogen deposition rates, as in the Netherlands, soil
acidification (which is mainly manifested by leaching of aluminium and
nitrate) is mainly caused by sulfur deposition. A relatively small
contribution of nitrogen to acidification does not imply that sulfur
has a larger impact on the health of forests, since the relationship

between soil acidification and forest health is not very clear. The
eutrophying impact of nitrogen is probably more important than the
acidifying impact at present.

There is substantial evidence from field surveys in several
countries of Europe that exceeding critical loads does not imply
dieback of the forest trees in the short term (one or two decades).
However, it does increase the risk of damage due to secondary stress
factors and it affects the long-term sustainability of forests. These
risks increase with the extent to which present loads exceed critical
loads and with the duration.

Table 26. Critical loads for acidity and nitrogen for forest ecosystems in temperate climates
(From: De Vries, 1993)

Effects Criteria^a Critical loads
(kg per ha per year)
(H for acidity;
N for eutrophication)

Coniferous Deciduous
forests forests

Acidity root damage; Al < 0.2 mol/m³ 1.1^b 1.4^b
inhibition of uptake; Al/Ca < 1.0 mol/mol 1.4^b 1.1^b
Al depletion; delta Al(OH)₃ = 0 mmol/m³ 1.2^b 1.3^b
Al pollution Al < 0.02 mol/m³ 0.5^b 0.3^b

Eutrophication inhibition of uptake of K; NH₄/K < 5 mol/mol 17-70^c
increased susceptibility; N < 1.8% 21-42^d
vegetation changes; NO₃ < 0.1 mol/m³ 7-20^e 11-20^e
nitrate pollution NO₃ < 0.4-0.8 mol/m³ 13-21^f 24-41^f

^a Background information on the various criteria is given in De Vries (1993). Critical Al and NO₃^-
concentrations and critical Al/Ca and NH₄/K ratios related to root damage, inhibition of nutrient
uptake and vegetation changes refer to the soil solution. Critical Al and NO₃^- concentrations
related to pollution refer to phreatic groundwater. Critical nitrogen contents related to an
increased risk for frost damage and diseases refer to the foliage.
^b Derived by a steady-state model. Al pollution refers to phreatic groundwater. For groundwater
used for the preparation of drinking-water, a critical acid load of 1600 mol/ha per year was derived
(De Vries, 1993).
^c Derived by a steady-state model assuming 50% nitrification in the mineral topsoil (second value).
^d Empirical data on the relation between nitrogen deposition and foliar nitrogen contents.

Table 26 (Con't)

^e The first value is derived by a steady-state model (worst case) and the second value is based
on empirical data.
^f Derived by a steady-state model using critical NO₃^- concentrations of 0.4 and 0.8 mol/m³,
respectively. NO₃^- pollution refers to phreatric groundwater. For deep groundwater, the
critical load will be higher because of denitrification.

Table 27. Possible and observed effects when critical loads are exceeded

Possible effects Average critical load Observed effects in the field
(kg per ha per year)^a

Root damage 1.1-1.4 H critical Al concentrations
exceeded greatly

Inhibition of 1.1-1.4 H critical Al/Ca ratios
uptake exceeded greatly

17-70 N critical NH₄/K ratios
exceeded slightly

Aluminium depletion 1.2-1.3 H depletion of secondary Al compounds

Groundwater 0.3-0.5 H critical Al concentrations
pollution exceeded greatly

13-21 N critical NO₃ concentrations
exceeded substantially

Increased 21-42 N critical N contents exceeded
susceptibility substantially; nutrient imbalances;
increased shoot/root ratios

Vegetation changes 7-20 N strong increase in nitrophilous species

^a H = acidity; N = total nitrogen

5. STUDIES OF THE EFFECTS OF NITROGEN OXIDES ON EXPERIMENTAL ANIMALS

5.1 Introduction

Most of the data reviewed in this chapter concerns the effects of
NO₂, since the bulk of the NO_x literature is on NO₂. The results
of the few comparative NO_x studies suggest that NO₂ is the most
toxic species studied so far. Most of the reports describe the
effects of NO₂ on the respiratory tract, but extrapulmonary effects
are also briefly discussed. A broad range of NO₂ concentrations has
been evaluated, but emphasis has been placed primarily on those
studies with exposure concentrations of 9400 µg/m³ (5.0 ppm) or less,
with the exception of studies on dosimetry and emphysema. Discussions
of available literature on the effects of other nitrogen compounds,
e.g., NO, HNO₃, and mixtures containing NO₂, also are included. WHO
(1987), Berglund et al. (1993) and US EPA (1993) comprise other
reviews of the animal toxicological literature concerning NO_x
effects.

5.2 Nitrogen dioxide

5.2.1 Dosimetry

It is generally agreed that effects of NO₂ observed in several
laboratory animal species can be qualitatively extrapolated to humans.
However, to extrapolate animal data quantitatively to humans,
knowledge of both dosimetry and species sensitivity must be
considered. Dosimetry refers to estimating the quantity of NO₂
absorbed by target sites within the respiratory tract. Even when two
species receive an identical local tissue/cellular dose, cellular
sensitivity to that dose is likely to show interspecies variability
due to differences in defence and repair mechanisms and other
physiological/metabolic parameters. Current knowledge of dosimetry is
more advanced than that of species sensitivity, impeding quantitative
animal-to-human extrapolation of effective NO₂ concentrations.
Nevertheless, information on dosimetry alone can be crucial to
interpretation of the data base. Both theoretical (modelling) and
experimental dosimetry studies are discussed below.

5.2.1.1 Respiratory tract dosimetry

The uptake of NO₂ in the upper respiratory tract (above the
larynx) has been experimentally studied in dogs, rats and rabbits.
The upper airways of dogs and rabbits exposed to 7520 to 77 080 µg/m³
(4.0 to 41.0 ppm) NO₂ removed 42.1% of the NO₂ drawn through
the nose (Yokoyama, 1968). The uptake of NO₂ by isolated
upper respiratory tracts of naive and previously exposed rats
(76 000 µg/m³, 40.4 ppm NO₂) was 28% and 25%, respectively (Cavanagh
& Morris, 1987). Kleinman & Mautz (1987) exposed dogs to 1880 or

9400 µg/m³ (1.0 or 5.0 ppm) NO₂ and found that more NO₂ was
absorbed in the upper respiratory tract with nasal breathing than with
oral breathing. In addition, the percentage uptake of NO₂ by the
upper respiratory tract decreased with increasing ventilation rates.
As ventilation increased up to four times resting values, NO₂ uptake
during nasal breathing decreased from approximately 85% to less than
80% and during oral breathing decreased from about 60% to approximately
45%. At rest, about 85% of the inhaled NO₂ entering the lungs was
absorbed by the lower respiratory tract; this increased to 100% with
high ventilation rates.

Miller et al. (1982) and Overton (1984) modelled NO₂ uptake in
the lower respiratory tract using the same dosimetry model described
by Miller et al. (1978) for ozone (O₃), but with the diffusion
coefficient and Henry's law constant appropriate to NO₂; however,
values of the latter constant and reaction chemistry were considered
uncertain. For all species modelled (i.e., rat, guinea-pig, rabbit
and humans), the results indicate that NO₂ is absorbed throughout the
lower respiratory tract, but the major dose to tissue is delivered in
the centriacinar region (i.e., junction between the conducting and
respiratory airways), findings consistent with the site of
morphological effects (see section 5.2.2.4).

Total respiratory tract uptake has been measured in healthy and
diseased humans. In healthy humans exposed to an NO/NO₂ mixture
containing 545 to 13 500 µg/m³ (0.29 to 7.2 ppm) NO₂ for brief (but
unspecified) periods, 81 to 90% of the NO₂ was absorbed during normal
respiration; this increased to 91 to 92% with maximal ventilation
(Wagner, 1970). Bauer et al. (1986) exposed adult asthmatics to
564 µg/m³ (0.3 ppm) NO₂ via a mouthpiece for 30 min, including
10 min of exercise (30 litres/min) and measured inspired and expired
NO₂ concentrations. At rest, the average uptake was 72%; during
exercise, the average uptake was 87%, a statistically significant
increase. Because of the large increase in minute ventilation, the
deposition was 3.1 µg/min at rest and 14.8 µg/min during exercise.

As discussed above, increased ventilation increases the quantity
of NO₂ delivered to the respiratory tract and shifts the site of
deposition. Typically, the percentage uptake of NO₂ in the upper
respiratory tract decreases, with a consequent increase in uptake by
the lower respiratory tract owing to the deeper penetration of the
inspired gas with increased tidal volume. These experimental results
are qualitatively similar to conclusions for the modelled effects of
ventilation on O₃ dosimetry (Miller et al., 1985; Overton et al.,
1987a,b).

5.2.1.2 Systemic dosimetry

Once deposited, NO₂ dissolves in lung fluids and various
chemical reactions occur, giving rise to products that are found in
the blood and other body fluids. Labelled ¹³NO₂ (564 to 1710 µg/m³
(0.3 to 0.91 ppm)) inhaled for 7 to 9 min by rhesus monkeys was
distributed throughout the lungs (Goldstein et al., 1977b). These
investigators also concluded that NO₂ probably reacts with water in
the fluids of the respiratory tract to form nitrous and nitric acids.
Saul & Archer (1983) provided support for this pathway using rats
inhaling NO₂. This study subsequently led to the discovery of
endogenous NO (Moncada et al., 1988, 1991).

The current database indicates that once NO₂ is absorbed in lung
fluids, the subsequent reaction products are rapidly taken up and then
translocated via the bloodstream. For example, Oda et al. (1981)
reported a concentration-dependent increase in both NO₂^- and NO3^-
levels in the blood of mice during 1-h exposures to 9400 to
75 200 µg/m³ (5.0 to 40.0 ppm) NO₂. The blood levels of NO₂^- and
NO₃^- declined rapidly after exposures ended, with decay half-times
of a few minutes for NO₂^- and about 1 h for NO₃^-.

5.2.2 Respiratory tract effects

5.2.2.1 Host defence mechanisms

Respiratory tract defences encompass many interrelated responses;
however, for simplicity, they can be divided into physical and
cellular defence mechanisms. Physical defence mechanisms include the
mucociliary system of the conducting airways. Ciliary action moves
particles and dissolved gases within the mucous layer towards the
pharynx, where the mucus is swallowed or expectorated. Both nasal
and tracheobronchial regions are immunologically active (e.g.,
nasal-associated lymphoid tissue and bronchial-associated lymphoid
tissue), but this function has not been studied following NO₂
exposure. Cellular defence mechanisms (phagocytic and immunological
reactions) operate in the pulmonary region of the lung. Alveolar
macrophages (AMs) are the first line of cellular defence. The AMs
perform such activities as detoxifying and/or removing inhaled
particles, maintaining sterility against inhaled microorganisms,
interacting with lymphoid cells in a variety of immunological
reactions, and removing damaged or dying cells from the alveoli
through phagocytosis. Polymorphonuclear leukocytes (PMNs), another
group of phagocytic cells, are present in relatively small numbers
(i.e., a small percentage of cells obtained from bronchoalveolar
lavage (BAL) fluid) from normal lungs, but in response to a variety of
insults, there can be an influx of PMNs from blood into the lung
tissues and onto the surface of the airways. Once recruited to the
lung, PMNs then ingest and kill opsonized microbes and other foreign
substances by mechanisms similar to those for AMs.

The responses of PMNs and AMs are frequently studied using BAL,
the washing of the airways and alveolar spaces with saline. Cells and
fluid obtained from this procedure can be used in a variety of ways to
assess immune responses.

Humoral and cell-mediated immunity are also active in the
respiratory tract. The humoral part of this system primarily involves
the B cells that function in the synthesis and secretion of antibodies
into the blood and body fluids. The cell-mediated component primarily
involves T lymphocytes, which are involved in delayed hypersensitivity
and defences against viral, fungal, bacterial and neoplastic disease.

a) Mucociliary clearance

Exposure to NO₂ can cause loss of cilia and ciliated epithelial
cells, as discussed in section 5.2.2.4 on morphological changes. Such
changes are reflected in the functional impairment of mucociliary
clearance at high levels of NO₂ (> 9400 µg/m³, 5.0 ppm) (Giordano
& Morrow, 1972; Kita & Omichi, 1974). At lower exposures (2 h/day for
2, 7 and 14 days to 564 and 1880 µg/m³, 0.3 and 1.0 ppm NO₂), the
mucociliary clearance of inhaled tracer particles deposited in the
tracheobronchial tree of rabbits was not altered (Schlesinger et al.,
1987).

b) Alveolar macrophages

Structural, biochemical, and functional changes in AMs observed
in experimental animal studies to be caused by NO₂ exposure are
summarized in Table 28. The adversity of these effects is not clearly
understood at present, but they are taken as hallmarks of adverse
reactions. Studies of AMs in humans are discussed in chapter 6.

Alveolar macrophages isolated from mice continuously exposed to
3760 µg/m³ (2.0 ppm) NO₂ or to 940 µg/m³ (0.5 ppm) NO₂
continuously with a 1-h peak to 3760 µg/m³ (2.0 ppm) for 5 days/week
showed distinctive morphological changes after 21 weeks of exposure,
compared to controls (Aranyi et al., 1976). Structural changes
included the loss of surface processes, appearance of fenestrae, bleb
formation and denuded surface areas. Continuous exposure to a lower
NO₂ level did not result in any significant morphological changes.
Numerous morphological studies have shown that NO₂ exposure increases
the number of AMs (see section 5.2.2.4).

BAL methods have also been used to study AMs. Mochitate et al.
(1986) reported a significant increase in the total number of AMs
isolated from rats during 10 days of exposure to 7520 µg/m³ (4.0 ppm)
NO₂, but the number of PMNs did not increase. The AMs from exposed
animals also exhibited increased metabolic activity, as measured by
the activities of glucose-6-phosphate dehydrogenase, glutathione
peroxidase and pyruvate kinase. The AMs also showed an increase in

Table 28. Effects of nitrogen dioxide (NO₂) on alveolar macrophages^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

564 0.3 2 h/day, Rabbits Increase in alveolar clearance. Schlesinger &
1880 1.0 14 days Gearhart (1987)

564 0.3 2 h/day, 13 days Rabbit Decreased AM phagocytic capacity at 564 µg/m³; increase Schlesinger
1880 1.0 at 1880 µg/m³ after 2 days of exposure. No effect on cell (1987a,b)
number or viability; random mobility reduced at 564 µg/m³
only. No effects from 6 days of exposure on.

564 0.3 2 h/day, 1 or Rabbit Acceleration in alveolar clearance at < 1880 µg/m³. Vollmuth et
1880 1.0 14 days al. (1986)
5640 3.0

940 or 0.5 or Continuous base Mouse No observable effects on AM morphology. Aranyi et al.
188 base; 0.1 base; with 2-h/day peak (1976)
1880 peak 1.0 peak (5 days/week),
24 weeks

3760 or 2.0 or 0.5 Continuous base Mouse Morphological changes, such as loss of surface processes, Aranyi et al.
940 base; base; with 7 h/day peak appearance of fenestrae, bleb formation, and denuded (1976)
3760 peak 2.0 peak (5 days/week), surface areas.
21 weeks

1880 1.0 17 h Mouse Bactericidal activity significantly decreased by 6 and Goldstein et al.
4320 2.3 35% at 4320 and 12 400 µg/m³, respectively; no effect at (1974)
12 400 6.6 1880 µg/m³.

Table 28. (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

1880 base; 1.0 base; 7 h/day, 5 days Rat Accumulation of AMs. Superimposed spikes produced Gregory et al.
9400 peak 5.0 peak per week base with changes that may persist with continued exposures. (1983)
one 1.5-h peak/day,
15 weeks

2444-31 960 1.3-17.0 - Rat Decreased production of superoxide anion radical. Amoruso et al.
(1981)

3760 2.0 8 h/day, Baboon Impaired AM responsiveness to migration inhibitory Greene &
5 days/week, factor. Schneider (1978)
6 months

5640 3.0 3 h Rabbit Increased swelling of AMs. Dowell et al.
(1971)

6768 3.6 2 h Rat Enhanced AM agglutination with concanavalin A. Goldstein et al.
(1977a)

7520 4.0 6 h/day, 7, 14, Rat Changes in AM morphology; no change in numbers of AMs Hooftman et al.
or 21 days or phagocytic capacity. (1988)

7520 4.0 10 days Rat Increase in number of AMs; no increase in PMNs; increased Mochitate et al.
metabolic activity, protein and DNA synthesis; all (1986)
responses peaked on day 4 and returned to normal on
day 10.

Table 28. (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

7520 4.0 Up to 10 days Rat Increase in number of AMs, reaching a peak on days 3 and Suzuki et al.
5; no increase in number of PMNs; decrease in AM viability (1986)
throughout exposure period. Suppression of phagocytic
activity on day 7 that returned to normal value at
day 10. Decrease in superoxide radical production on
days 3, 5 and 10.

9400 5.0 7 days Mouse No effect on phagocytic activity. Lefkowitz et al.
(1986)

9400 5.0 3 h Rabbit No change in AM resistance to pox virus. Acton & Myrvik
(1972)

^a Modified from US EPA (1993)
^b AM = alveolar macrophage; PMN = polymorphonuclear leukocyte
the rate of synthesis of protein and DNA. All responses peaked on day
4 and returned to control levels by the tenth day. Suzuki et al.
(1986) made similar observations and, in addition, found that the
viability of AMs was decreased on day 1 and remained depressed for the
remainder of the exposure period. Increased numbers and metabolic
activity of AMs would be expected to have a positive influence on host
defences. However, AMs are rich in proteolytic enzymes, and increased
numbers could result in some tissue destruction when the enzymes are
released. Furthermore, as discussed below, although more AMs may be
present, they often have a decreased phagocytic ability.

Schlesinger (1987a,b) found no significant changes in the number
or the viability of AMs in BAL from rabbits exposed to 564 or
1880 µg/m³ (0.3 or 1.0 ppm) NO₂, 2 h/day, for 13 days. Although
there were no effects on the numbers of AMs that phagocytosed latex
spheres, 2 days of exposure to 564 µg/m³ (0.3 ppm) decreased the
phagocytic capacity (i.e., number of spheres per cell); the higher
level of NO₂ increased phagocytosis. Longer exposures had no effect.
The phagocytic activity of rat AMs was significantly depressed after
7 days of exposure to 7520 µg/m³ (4.0 ppm) but returned to the
control value at 10 days of exposure (Suzuki et al., 1986). There may
be a species difference in responsiveness because Lefkowitz et al.
(1986) did not observe a depression in phagocytosis in mice exposed
for 7 days to 9400 µg/m³ (5.0 ppm) NO₂. Suzuki et al. (1986)
proposed that the inhibition of phagocytosis might be due to NO₂
effects on membrane lipid peroxidation. Studies by Dowell et al.
(1971) and Goldstein et al. (1977a) add support to this hypothesis.
Acute exposure to 5640-7520 µg/m³ (3.0-4.0 ppm) caused swelling of
AMs (Dowell et al., 1971) and increased AM agglutination with
concanavalin A (Goldstein et al., 1977a), suggesting damage to the
membrane function.

Two independent studies have shown that NO₂ exposure decreases
the ability of rat AMs to produce superoxide anion involved in
antibacterial activity. Amoruso et al. (1981) presented evidence
of such an effect at NO₂ concentrations ranging from 2440 to
32 000 µg/m³ (1.3 to 17.0 ppm). The duration of the NO₂ exposure
was not given; all exposures were expressed in terms of parts per
million × hours. A 50% decrease of superoxide anion production began
after exposure to 54 700 µg/m³ × h (29.1 ppm × h) NO₂. Suzuki et
al. (1986) reported a marked decrease in the ability of rat AMs to
produce superoxide anion following a 10-day exposure to either 7520 or
15 000 µg/m³ (4.0 or 8.0 ppm) NO₂. At the highest concentration,
the effect was significant each day, but at the lower concentration,
the depression was significant only on exposure days 3, 5 and 10.

Alveolar macrophages obtained by BAL from baboons exposed to
3760 µg/m³ (2.0 ppm) NO₂ for 8 h/day, 5 days/week, for 6 months had
impaired responsiveness to migration inhibitory factor produced by
sensitized lymphocytes (Greene & Schneider, 1978). This substance

affects the behaviour of AMs by inhibiting free migration, which, in
turn, interferes with the functional capacity of these defence cells.
In addition, the random mobility of AMs was significantly depressed in
rabbits following a 2 h/day exposure for 13 days to 564 µg/m³
(0.3 ppm), but not to 1880 µg/m³ (1.0 ppm) (Schlesinger, 1987b).

Vollmuth et al. (1986) studied the clearance of strontium-
85-tagged polystyrene latex spheres from the lungs of rabbits
following a single 2-h exposure to 564, 1880, 5640 or 18 800 µg/m³
(0.3, 1.0, 3.0 or 10.0 ppm) NO₂. An acceleration in clearance
occurred immediately after exposure to the two lowest NO₂
concentrations; a similar effect was found by Schlesinger & Gearhart
(1987). At the higher levels of NO₂, an acceleration in clearance
was not evident until midway through the 14-day post-exposure period.
Repeated exposure for 14 days (2 h/day) to 1880 or 18 800 µg/m³
(1.0 or 10.0 ppm) NO₂ produced a response similar to a single
exposure at the same concentration.

c) Humoral and cell-mediated immunity

Various humoral and cell-mediated effects are summarized in Table
29.

Exposing sheep to 9400 µg/m³ (5.0 ppm) NO₂, 1.5 h/day for 10 to
11 days showed that intermittent short-term exposure may temporarily
alter the pulmonary immune responsiveness (Joel et al., 1982). One
technique commonly used in determining the production of specific
antibody-forming cells is to measure the number of plaque-forming
cells (PFCs) in the blood or tissues of immunized animals. In this
study, the authors assessed immunological response by monitoring the
daily output of PFCs in the efferent lymph of caudal mediastinal lymph
nodes of sheep immunized with horse erythrocytes (a T-cell dependent
antigen). Although the number of animals used was small and the data
were not analysed statistically, it would appear that, in the animals
that were immunized 2 days (but not 4 days) after NO₂ exposure
started, the output of PFC was below control values. Blastogenic
responses of T cells from the efferent pulmonary lymph and venous
blood also appeared to be decreased.

Hillam et al. (1983) examined the effects of a 24-h exposure to
9400, 18 800 and 48 900 µg/m³ (5.0, 10.0 and 26.0 ppm) NO₂ on
cellular immunity in rats after intratracheal immunization with sheep
erythrocytes (SRBCs). Cellular immunity was evaluated by antigen-
specific lymphocyte stimulation assays of pooled lymphoid cell
suspensions from either the thoracic lymph nodes or the spleen.
Concentration-related elevation of cellular immunity in thoracic lymph
nodes and spleen were reported after immunizing the lung with SRBCs.

Table 29. Effects of nitrogen dioxide (NO₂) on the immune system^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

188 base; 0.1 base; Continuous base Mouse Suppression of splenic T and B cell responsiveness to Maigetter et al.
470, 940, 0.25, 0.5, with 3-h/day peak mitogens variable and not related to concentration or (1978)
or 1880 or 1.0 peak (5 days/week), 1, 3, duration, except for the 940 µg/m³ continuous group,
peak 6, 9, 12 months which had a linear decrease in PHA-induced mitogenesis
with NO₂ duration.
940 0.5 Continuous

470 0.25 7 h/day, Mouse Reduced percentage of total T-cell population and trend Richters & Damji
5 days/week, (AKR/cum) towards reduced percentage of certain T-cell (1988)
7 weeks subpopulations; no reduction of mature T cells or natural
killer cells.

470 0.25 7 h/day, Mouse Reduced percentage of total T-cell population and Richters & Damji
5 days/week, (AKR/cum) percentages of T helper/inducer cells on days 37 and 181. (1990)
36 weeks

658 0.35 7 h/day, Mouse Trend towards suppression in total percentage of T-cells. Richters & Damji
5 days/week, (C57BL/6J) No effects on percentages of other T-cell subpopulations. (1988)
12 weeks

752 0.4 24 h/day Mouse Decrease in primary PFC response at >752 µg/m³. Fujimaki et al.
3010 1.6 4 weeks Increase in secondary PFC response at 3010 µg/m³. (1982)

Table 29. (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

940 base; 0.5 base; 22-h/day base Rat No effect on splenic or circulatory B or T cell response Selgrade et al.
2820 peak 1.5 peak (7 days/week); to mitogens. After 3 weeks of exposure only, decrease in (1991)
6-h ramped peak splenic natural killer cell activity. No histological
(5 days/week) changes in lymphoid tissues
1, 3, 13, 52,
78 weeks

940 base, 0.5 base, Continuous base Mouse Vaccination with influenza A2/Taiwan virus after exposure. Ehrlich et al.
3760 peak 2.0 peak with 1 h/day Decrease in serum neutralizing antibody; haemagglutination (1975)
(5 days/week) inhibition titres unchanged. Before virus challenge, NO₂
3760 2.0 peak, 3 months exposure decreased serum IgA and increased IgG₁, IgM, and
IgG₂; after virus, serum IgA unchanged and IgM increased.

1880 1.0 493 days Monkey Monkeys challenged five times with monkey-adapted Fenters et al.
influenza virus during NO₂ exposure. Haemagglutination (1973)
inhibition antibody titres not altered. Compared to
controls, NO₂ caused an earlier and greater increase in
serum neutralization antibody titres to the virus.

1880 1.0 6 months Guinea-pig Intranasal challenge with K. pneumoniae after exposure. Kosmider et al.
Decreased haemolytic activity of complement; decrease in (1973)
all immunoelectrophoretic fractions.

2820 1.5 24 h/day, 6, Mouse Reduction in number of splenic PFCs; lowering Lefkowitz et al.
9400 5.0 14, or 21 days concentration to 2820 µg/m³ and extending the length to (1986)
14 or 21 days decreased PFCs by 33 and 50%, respectively;
no effect on cell-mediated immune system or
haemagglutination titres.

Table 29. (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9400 5.0 1.5 h/day, Sheep Reduction in PFCs from pulmonary lymph and in mitogenesis Joel et al.
10-11 days of T cells from pulmonary lymph and blood. (1982)

9400 5.0 4 h/day, Guinea-pig Serum antibodies against lung tissue increased with Balchum et al.
28 200 15.0 5 days/week, concentration and duration of exposure. (1965)
5.52 months

9400 5.0 Continuous, 169 Monkey Initial depression in serum neutralization titres with Fenters et al.
days, challenged return to normal by day 133; no effect on secondary (1971)
4 x with mouse- response on haemmagglutin inhibition titre.
adapted influenza
virus

9400 5.0 3-7 days Mouse No effect on serum interferon levels. Lefkowitz et al.
47 000 25.0 (1983, 1984)

9400 5.0 24 h Rat Concentration-related elevation of cellular immunity in Hillam et al.
18 800 10.0 thoracic lymph nodes and spleen after immunizing the lung (1983)
48 900 26.0 with sheep RBCs.

Table 29. (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9400 5.0 Continuous, Monkey Depressed postvaccination serum neutralizing antibody Ehrlich &
6 months formation. Fenters (1973)

9400 5.0 12 h Mouse No effect on primary and secondary splenic PFC response. Fujimaki &
Shimizu (1981);
Fujimaki et al.
(1981)

^a Source: Modified from US EPA (1993)
^b PFC = plaque-forming cell; PHA = phytohaemagglutinin; Ig = immunoglobulin; RBCs = red blood cells
Fujimaki et al. (1982) investigated the effect of a 4-week
exposure to 752 and 3000 µg/m³ (0.4 and 1.6 ppm) NO₂ in mice (i.e.,
primary and secondary antibody response to SRBCs, using the splenic
PFC response as the end-point). The primary PFC response was
decreased by both NO₂ concentrations. Secondary antibody response
was not affected at 752 µg/m³ (0.4 ppm), but was slightly enhanced at
3000 µg/m³ NO₂. Acute exposure (12 h) of mice to 9400 µg/m³
(5.0 ppm) NO₂ caused no such effects (Fujimaki & Shimizu, 1981;
Fujimaki et al., 1981).

The effect of exposure pattern was examined by Maigetter et al.
(1978) by exposing mice for up to 1 year to 940 µg NO₂/m³ (0.5 ppm)
continuously or to three regimens having a continuous baseline of
188 µg/m³ (0.1 ppm) with 3-h peaks (5 days/week) of either 470, 940
or 1880 µg/m³ (0.25, 0.5 or 1.0 ppm). General mitogenic responses of
splenic lymphocytes to phytohaemagglutinin (PHA) (a T cell dependent
mitogen) and lipopolysaccharide (a B-cell dependent mitogen)
decreased, but this was not related to the concentration or duration
of exposure, with a single exception. The decrease in PHA-induced
mitogenesis was linearly related to the increased duration of NO₂
exposure to 940 µg/m³ (0.5 ppm).

Shorter exposure (6 days) to 9400 µg/m³ (5.0 ppm) NO₂ did not
affect mitogenesis of T cells (Lefkowitz et al., 1986). Although NO₂
did not affect haemagglutination antibody titres, it did reduce the
number of splenic PFCs to SRBCs. The authors stated (data were not
shown) that mice exposed to 2820 µg/m³ (1.5 ppm) NO₂ for 14 or
21 days showed a 33 and 50% decrease, respectively, in the number of
PFCs.

Kosmider et al. (1973) exposed guinea-pigs to 1880 µg/m³
(1.0 ppm) NO₂ for 6 months and reported a significant reduction in
all serum immunoglobulin (Ig) fractions and complement. Decreased
levels of these substances may lead to an increase in the frequency,
duration and severity of an infectious disease. Mice exposed to NO₂
had decreased serum levels of IgA and exhibited nonspecific increases
in serum IgM, IgG and IgG₂ (Ehrlich et al., 1975).

These effects on lymphocyte function may reflect changes in
lymphocyte populations. Richters & Damji (1988) found that the
percentage of the total T lymphocyte population was reduced in the
spleens of AKR/cum mice exposed for 7 weeks (7 h/day, 5 days/week) to
470 µg/m³ (0.25 ppm) NO₂. The percentages of mature helper/inducer
T and T cytotoxic/suppressor lymphocytes were also lower in the
spleens of exposed animals. There were no changes in the percentages
of natural killer cells or mature T cells. Upon a longer (36-week)
exposure, Richters & Damji (1990) found that the numbers of T-helper/
inducer (CD4⁺) lymphocytes (spleen) were reduced, but no effects
were observed on T cytotoxic/suppressor cells. Spontaneously
developing lymphomas in NO₂-exposed animals progressed more

slowly than those in control animals. This was attributed to
the NO₂-induced reduction in the T-helper/inducer lymphocytes.
C57BL/6J mice exposed to 658 µg/m³ (0.35 ppm) for 7 h/day,
5 days/week for 12 weeks, also showed a suppression in the percentage
of total matured T lymphocytes, but no effect on any specific
subpopulation upon longer exposure (36 weeks) to 470 µg/m³ (0.25 ppm)
(Richters & Damji, 1988). Selgrade et al. (1991) found that chronic
exposure (up to 78 weeks) to an urban pattern of NO₂ (baseline of
940 µg/m³ (0.5 ppm) with a ramped 6-h peak to 2820 µg/m³ (1.5 ppm))
had no effect on splenic or circulating B or T cell mitogenic
response. However, there was a transient decrease in splenic natural
killer cell activity (at 3 weeks only).

Few studies have been undertaken to assess the effects of NO₂ on
interferon production. Mice exposed to either 9400 or 47 000 µg/m³
(5.0 or 25.0 ppm) NO₂ for 3 to 7 days had serum levels of interferon
similar to those of controls (Lefkowitz et al., 1983, 1984).

Induction of autoimmunity was suggested by the work of Balchum
et al. (1965). Guinea-pigs exposed to 9400 µg/m³ (5.0 ppm) and
28 200 µg/m³ (15.0 ppm) NO₂ had an increase in the titre of serum
antibodies against lung tissue, starting after 160 h of NO₂ exposure.
These antibody titres continued to increase with NO₂ concentration
and duration of exposure.

The impact of NO₂ on the humoral immune response of squirrel
monkeys to intratracheally delivered influenza vaccine was studied by
Fenters et al. (1971, 1973) and Ehrlich & Fenters (1973). In monkeys
exposed for 493 days to 1880 µg/m³ (1.0 ppm) NO₂ and immunized with
monkey-adapted virus (A/PR/8/34), the serum neutralizing antibody
titres were significantly increased earlier and to a greater degree
than those of controls (Fenters et al., 1973; Ehrlich & Fenters,
1973). In monkeys exposed to 9400 µg/m³ (5.0 ppm) NO₂ for a total
of 169 days and immunized with mouse-adapted influenza virus (A/PR/8),
serum neutralization titres were lower than controls initially; no
significant difference was observed by 133 days of exposure (Fenters
et al., 1971; Ehrlich & Fenters, 1973). In all of these studies, the
haemagglutination inhibition antibody titres were not affected.
Differences between studies might be due to the difference in the
virus used for immunization, along with exposure differences. Also,
exposure to 1880 µg/m³ (1.0 ppm) NO₂ may have increased the
establishment of infection and the survival of the monkey-adapted
virus within the respiratory tract, resulting in an increase in
antibody production.

Mice that were vaccinated with influenza virus (A-2/Taiwan/ 1/64)
after 3 months of continuous exposure to 3760 µg/m³ (2.0 ppm) or to
940 µg/m³ (0.5 ppm) NO₂ with a 1-h daily (5 days/week) spike
exposure to 3760 µg/m³ (2.0 ppm) had mean serum neutralizing antibody
titres that were four-fold lower than those of clean air controls
(Ehrlich et al., 1975). The haemagglutination inhibition antibody
titres in these animals were unchanged. This agrees with the Fenters
et al. (1973) findings in monkeys exposed to 1880 µg/m³ (1.0 ppm) for
over 1 year.

d) Interaction with infectious agents

Various experimental approaches have been employed using animals
in an effort to determine the overall functional efficiency of the
host's pulmonary defences following NO₂ exposure. In the most
commonly used infectivity model, animals are exposed to either NO₂
or filtered air. After NO₂ exposure, the treatment groups are
combined and exposed briefly to an aerosol of a viable agent, such as
Streptococcus sp., Klebsiella pneumoniae, Diplococcus pneumoniae
or influenza virus. The animals are then returned to clean air
for a holding period (usually 15 days), and the mortality in the
NO₂-exposed and the control groups are compared. If host defences are
compromised by the NO₂ exposure, mortality rates will be higher
(Ehrlich, 1966; Henry et al., 1970; Coffin & Gardner, 1972; Ehrlich et
al., 1979; Gardner, 1982). Although the end-point is mortality, it is
a sensitive indicator of the depression of the defence mechanisms used
to control infection. Because these specific defence mechanisms are
common to laboratory animals and humans, the increased susceptibility
to infection can be qualitatively extrapolated to humans, even though
mortality would not be an expected outcome in humans receiving
appropriate medical treatment. However, different exposure levels of
NO₂ and infectious agents may be required to produce changes in human
host defences. Effects of NO₂ on pulmonary infectious disease in
humans are discussed in chapters 6 and 7. Table 30 summarizes effects
of exposure to NO₂ and infectious agents observed in animals.

An enhancement in mortality following exposure to NO₂ in
combination with a pathogenic microorganism could be due to several
factors. Goldstein et al. (1973) showed decreases in pulmonary
bactericidal activity following NO₂ exposure. In their first
experiments, mice breathed aerosols of Staphylococcus aureus
(S. aureus) labelled with radioactive phosphorus and were then
exposed to NO₂ for 4 h. Physical removal of the bacteria was not
affected by any of the NO₂ concentrations used up to 27 800 µg/m³
(14.8 ppm). Concentrations > 13 200 µg/m³ (7.0 ppm) NO₂ lowered
the bactericidal activity by > 7%. Lower concentrations (3570 and
7140 µg/m³ (1.9 and 3.8 ppm)) had no significant effect. In another
experiment (Goldstein et al., 1974), mice breathed 1800, 4320 and
12 400 (1.0, 2.3 and 6.6 ppm) NO₂ for 17 h and then were exposed
to an aerosol of S. aureus. Four hours later, the animals were
examined for the number of organisms present in the lungs. No
difference in the number of bacteria inhaled was found in the
NO₂-exposed animals. Concentrations of 4320 and 12 400 µg/m³
(2.3 and 6.6 ppm) NO₂ decreased pulmonary bactericidal activity by
6 and 35%, respectively, compared to controls. Exposure to 1880 µg/m³
(1.0 ppm) NO₂ had no significant effect. Goldstein et al. (1974)
hypothesized that the decreased bactericidal activity was due to
defects in AM function. Jakab (1987) confirmed these findings and
found that the concentration of NO₂ required to suppress pulmonary
bactericidal activity in mice depended on the specific organism. For
example, exposure to > 7520 µg/m³ (> 4.0 ppm) NO₂ for 4 h
after bacterial challenge depressed bactericidal activity
against S. aureus, but it required a concentration of 18 800 to
37 600 µg/m³ (10.0 to 20.0 ppm) before the lung's ability to kill
deposited Pasteurella and Proteus was impaired. Parker et al.
(1989) made similar observations in mice exposed for 4 h to 9400 or
18 800 µg/m³ (5.0 or 10.0 ppm) NO₂ and infected with Mycoplasma
pulmonis. The higher concentration of NO₂ increased mortality.
Both concentrations: (1) reduced lung bactericidal activity and
increased bacterial growth, without affecting deposition or physical
clearance; and (2) increased the incidence of lung lesions as well as
their severity. Davis et al. (1991) found no effects of lower NO₂
concentrations on bactericidal activity using the same model system.

Table 30. Interaction of nitrogen dioxide (NO₂) with infectious agents^a

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

100 base, 0.05 base, Continuous, with Mouse Streptococcus No effect Gardner (1980);
188 peak 0.1 peak 1 h peak, twice/day sp. Gardner et al.
(5 days/week), (1982); Graham
15 days et al. (1987)

940 + 0.5 + Increased mortality
1880 peak 1.0 peak

2260 + 1.2 + Increased mortality
4700 peak 2.5 peak

376 base, 0.2 base, Continuous base Mouse Streptococcus Spike plus baseline caused Miller et al.
1500 peak 0.8 peak with 1-h peak sp. significantly greater mortality (1987)
twice/day than baseline.
(5 days/week),
1 year

564-940 0.3-0.5 Continuous, Mouse A/PR/8 High incidence of adenomatous Motomiya et al.
3 months virus proliferation of peripheral and (1973)
bronchial epithelial cells; NO₂
alone and virus alone caused less
severe alterations.

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

Continuous, No enhancement of effect of NO₂
6 months and virus.

940 0.5 3 h/day, Mouse Streptococcus Increase in mortality with Ehrlich et al.
3 months sp. reduction in mean survival time. (1979)

940 0.5 Intermittent, Mouse Klebsiella Increased mortality after 6 months Ehrlich &
6 or 18 h/day, pneumoniae intermittent exposure or after Henry (1968)
up to 12 months 3, 6, 9 or 12 months continuous
exposure; following 12 months
exposure, increased mortality was
Continuous, significant only in continuously
24 h/day up to exposed mice.
12 months

940-1880 0.5-1.0 Continuous, Mouse A/PR/8 Increased susceptibility to Ito (1971)
39 days (female) virus infection
18 800 10.0 2 h/day, 1, 3,
and 5 days

940-52 600 0.5-28.0 Varied Mouse Streptococcus Increased mortality with increased Gardner et al.
sp. time and concentration; (1977a,b); Coffin
concentration is more important et al. (1977)
than time.

940 0.5 24 h/day, Mouse K. pneumoniae Significant increase in mortality McGrath &
1880 1.0 7 days/week, after 3-day exposure to 9400 µg/m³; Oyervides (1985)
2820 1.5 3 months no effect at other concentrations,

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

9400 5.0 3 days but control mortality was very
high.

1880 1.0 17 h Mouse Staphylococcus No difference in number of bacteria Goldstein et al.
4320 2.3 aureus after deposited, but at 4320 and (1974)
12 400 6.6 NO₂ exposure 12 400 µg/m³, there was a decrease
in pulmonary bactericidal activity
of 6 and 35%, respectively;
no effect at 1880 µg/m³.

1880-4700 1.0-2.5 4 h Mouse S. aureus Impaired bactericidal activity Jakab (1988)
between 1800 and 4700 µg/m³ in
animals injected with
corticosteroids

4320 2.3 6% decrease in bactericidal
activity

12 400 6.6 35% decrease in bactericidal
activity

1880 1.0 48 h Mouse Streptococcus Increase proliferation of Sherwood et al.
sp.; S. aureus Streptococcus sp., but not (1981)
S. aureus, in lung

1880 1.0 3 h Mouse Streptococcus Exercise on continuously moving Illing et al.
5640 3.0 sp. wheels during exposure; increased (1980)
mortality at 5640 µg/m³

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

2820 1.5 Continuous or Mouse Streptococcus After 1 week, mortality with Gardner et al.
intermittent sp. continuous exposure was greater (1979)
(7 h/day), 7 days than that for intermittent; after
per week, 2 weeks 2 weeks, no significant difference
between continuous and intermittent
exposure.

6580 3.5 Increased mortality with increased
duration of exposure; no
significant difference between
continuous and intermittent
exposure; with data adjusted for
total difference in the production
of concentration and time,
mortality essentially the same.

2820 base, 1.5 base, Continuous 60 h Mouse Streptococcus Mortality increased with 3.5- and Gardner (1980);
8460 peak 4.5 peak then peak for 1, sp. 7-h single spike when bacterial Garnder et al.
3.5 or 7 h, then challenge was immediate, and (1982); Graham
continuous 18 h 18 h after the spike et al. (1987)

8460 4.5 1, 3.5, or 7 h Mortality proportional to duration
when bacterial challenge was
immediate, but not 18 h
post-exposure.

2820 1.5 7 h/day, 4, 5, Mouse Streptococcus Elevated temperature (32°C) Gardner et al.
and 7 days sp. increased mortality after 7 days. (1982)

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

2820 1.5 2 h Mouse K. pneumoniae Increased mortality only at Purvis &
4700 2.5 > 6580 µg/m³. Increase in Ehrlich (1966);
mortality

6580 3.5 K. pneumoniae challenge 1 and 6 h Ehrlich (1979)
9400 5.0 after 9400 or 18 800 µg/m³;
18 800 10.0 when K. pneumoniae challenge 27 h
28 200 15.0 following NO₂ exposure, effect
only at 28 200 µg/m³.

3570 1.9 4 h Mouse S. aureus Physical removal of bacteria Goldstein et al.
7140 3.8 prior to NO₂ unchanged at 3570 to 27 800 µg/m³. (1973)
exposure

13 160 7.0 7% lower bactericidal activity
17 300 9.2 14% lower bactericidal activity
27 800 14.8 50% lower bactericidal activity

3760 2.0 3 h Mouse Streptococcus Increased mortality Ehrlich et al.
sp. (1977);
Ehrlich (1980)

4700 2.5-30.0 4 h Mouse S. aureus, Concentration-related decrease Jakab (1987)
56 400 Pasteurella and in bactericidal activity, starting
Proteus at > 7500 µg/m³ with S. aureus when
NO₂ exposure was after bacterial
challenge; when NO₂ was before
bacterial challenge, effect at
18 800 µg/m³. Higher concentration
required to affect other organisms.

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

6580 3.5 2 h Mouse K. pneumoniae Increased mortality of all species Ehrlich (1975)

65 830 35.0 2 h Hamster
94 050 50.0 2 h Squirrel
monkey

9400 5.0 6 h/day, Mouse Cytomegalovirus Increase in virus susceptibility Rose et al.
6 days (1988)

9400 5.0 Continuous, Squirrel K. pneumoniae Increased viral-induced mortality Henry et al.
2 months monkey or A/PR/8 (1/3). Increase in Klebsiella- (1970)
influenza virus induced mortality (2/7); no
deaths. control

19 000 10.0 Continuous, Increased virus-induced mortality
1 month (6/6) within 2-3 days after
infection; no control deaths.
Increase in Klebsiella-induced
mortality (1/4); no control deaths.

9400 5.0 4 h Mouse Mycoplasma NO₂ increased incidence and Parker et al.
19 000 10.0 pulmonis severity of pneumonia lesions and (1989)
decreased the number of organisms
needed to induce pneumonia; no
effect on physical clearance,
decreased mycoplasmal killing
and increased growth; no effect on
specific IgM in serum;

Table 30 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Infective agent Effects Reference

C57Bl/6N mice generally more
sensitive than C3H/HeN mice.
At 19 000 µg/m³, one strain
(C57BL/6N) of mice had increased
mortality.

9400 5.0 2 months Squirrel K. pneumoniae Mortality 2/7; bacteria present Henry et al.
monkey in lung of survivors at autopsy. (1969)

65 800 35.0 1 month Mortality 1/4; bacteria present
in lungs of survivors at autopsy.

94 000 50.0 2 h Mortality 3/3

^a Modified from US EPA (1993)
Differences in species susceptibility to NO₂ or to a pathogen
may play a role in the enhancement of mortality seen in experimental
animals. An enhancement in mortality was noted in mice, hamsters and
monkeys acutely exposed to NO₂ for 2 h followed by a challenge
of K. pneumonia (Ehrlich, 1975). However, differences in
susceptibility were noted between the species. Ehrlich found
increased mortality occurred in monkeys only at 94 000 µg/m³
(50.0 ppm), whereas, lower NO₂ levels increased mortality in mice
(6580 µg/m³, 3.5 ppm) and hamsters (65 800 µg/m³, 35.0 ppm). The
mouse model was the most sensitive to NO₂ exposure, as shown by
enhanced mortality from K. pneumoniae following exposure to
6580 µg/m³ (3.5 ppm) but not to 2820-4700 µg/m³ (1.5-2.5 ppm) NO₂
for 2 h (Purvis & Ehrlich, 1963; Ehrlich, 1975). With prolonged
(2 month) exposure, Henry et al. (1969) found that lower levels of
NO₂ (9400 µg/m³, 5.0 ppm) increased susceptibility to bacterial
infections in monkeys than the 50.0 ppm concentration found to be
effective by Ehrlich (1975) with acute (2 h) exposure. The
sensitivity is also affected by the test organism. For example, when
Streptococcus sp. was the infectious agent, a 3-h exposure to
3760 µg/m³ (2.0 ppm) NO₂ caused an increased in mortality in mice
(Ehrlich et al., 1977). Sherwood et al. (1981) illustrated that
exposure to 1880 µg/m³ (1.0 ppm) NO₂ for 48 h increased the
propensity of virulent group-C streptococci, but not S. aureus, to
proliferate within mouse lungs and cause earlier mortality.

Additional factors can influence the interaction of NO₂ and
infectious agents. Mice placed on continuously moving exercise wheels
during exposure to 5640 µg/m³ (3.0 ppm) NO₂, but not 1880 µg/m³
(1.0 ppm), for 3 h showed enhanced mortality over non-exercised
NO₂-exposed mice using the streptococcal infectivity model (Illing et
al., 1980). The presence of other environmental factors, such as O₃
(Ehrlich et al., 1977; Gardner, 1980; Gardner et al., 1982; Graham et
al., 1987) or elevated temperatures (Gardner et al., 1982), also
exacerbated the effects of NO₂.

The influence of a wide variety of exposure regimens has been
evaluated using the infectivity model. For example, Gardner et al.
(1977b) examined the effect of varying durations of continuous
exposure on the mortality of mice exposed to six concentrations of
NO₂ (940 to 52 600 µg/m³ (0.5 to 28.0 ppm)) for durations ranging
from 15 min to 1 year. Streptococcus sp. was used for all
concentrations, except 940 µg/m³, in which case K. pneumoniae was
used. Mortality increased linearly with increasing duration of
exposure to a given concentration of NO₂. Mortality also increased
with increasing concentration of NO₂ as indicated by the steeper
slopes with higher concentrations. When the product of concentration
and time (C × T) was held constant, the relationship between
concentration and time produced significantly different mortality
responses. At a constant C × T of approximately 21 ppm-h, a 14-h
exposure to 2820 µg/m³ (1.5 ppm) NO₂ increased mortality by 12.5%,
whereas a 1.5-h exposure to 27 300 µg/m³ (14.0 ppm) NO₂ enhanced
mortality by 58.5%. These findings demonstrate that concentration is
more important than time in determining the degree of injury induced
by NO₂ in this model, and they were confirmed at additional C × T
values (Gardner et al., 1977a,b, 1982; Coffin et al., 1977).

Gardner et al. (1979) also compared the effect of continuous
versus intermittent exposure to NO₂ followed by bacterial challenge
with Streptococcus sp. Mice were exposed either continuously or
intermittently (7 h/day, 7 days/week) to 2820 or 6580 µg/m³ (1.5 or
3.5 ppm) NO₂. The continuous exposure of mice to 2820 µg/m³ NO₂
increased mortality after 24 h of exposure. During the first week of
exposure, the mortality was significantly higher in mice exposed
continuously to NO₂ than in those exposed intermittently. By the
14th day of exposure, the difference between intermittent and
continuous exposure became indistinguishable. At the higher
concentration, there was essentially no difference between continuous
and intermittent regimens. This suggests that fluctuating levels of
NO₂ may ultimately be as toxic as sustained high levels (Gardner et
al., 1979).

Mice were exposed continuously or intermittently (6 or 18 h/day)
to 940 µg/m³ (0.5 ppm) NO₂ for up to 12 months (Ehrlich & Henry,
1968). None of the exposure regimens affected resistance to
K. pneumoniae infection during the first month. Those exposed
continuously exhibited decreased resistance to the infectious agent,
as demonstrated by a significant enhancement in mortality at 3, 6, 9
and 12 months. In another experiment, a significant enhancement did

not occur at 3 months, but was observed after 6 months of exposure.
After 6 months, mice exposed intermittently (6 or 18 h/day) to NO₂
showed significant increases in mortality over controls (18%). Only
the continuously exposed animals showed increased mortality (23%) over
controls following 12 months of exposure. After 12 months of
exposure, mice in the three experimental groups showed a reduced
capacity to clear viable bacteria from their lungs. This was first
observed after 6 months in the continuously exposed group and after
9 months in the intermittently exposed groups. These changes,
however, were not statistically tested for significance. Although it
is not possible to compare directly the results of the studies using
Streptococcus sp. to those using K. pneumoniae, the data suggest
that, as the concentration of NO₂ is decreased, a longer exposure
time is necessary for the intermittent exposure regimen to produce a
level of effect equivalent to that of a continuous exposure. McGrath
& Oyervides (1985) did not confirm these findings in mice exposed to
940, 1880 and 2820 µg/m³ (0.5, 1.0 and 1.5 ppm) NO₂ for 3 months.
The inconsistency may be attributed to the fact that the McGrath &
Oyervides (1985) study had 95% mortality in the control groups, making
it virtually impossible to detect a further enhancement in mortality
due to NO₂.

Gardner (1980), Gardner et al. (1982) and Graham et al. (1987)
reported extensive investigations on the response to airborne
infections in mice breathing NO₂ spike exposures superimposed on a
lower continuous background level of NO₂, which simulated the pattern
(although not the NO₂ concentrations) of exposure in the urban
environment in the USA. Mice were exposed to spikes of 8460 µg/m³
(4.5 ppm) for 1, 3.5 or 7 h and then were challenged with
Streptococcus sp. either immediately or 18 h after exposure.
Mortality was proportional to the duration of the spike when the mice
were challenged with bacteria immediately after exposure, but mice had
recovered from the exposure by 18 h. Similar findings were reported
by Purvis & Ehrlich (1963) using K. pneumoniae. When a spike of
8460 µg/m³ (4.5 ppm) was superimposed on a continuous background of
2820 µg/m³ (1.5 ppm) for 62 h preceding and 18 h following the spike,
mortality was significantly enhanced by a spike lasting 3.5 or 7 h
when the infectious agent was administered 18 h after the spike
(Gardner, 1980; Gardner et al., 1982; Graham et al., 1987). Possible
explanations for these differences due to the presence or absence of a
background exposure are that mice continuously exposed are not capable
of recovery or that new AMs or PMNs recruited to the site of infection
are impaired by the continuous exposure to NO₂. The effect of
multiple spikes was examined by exposing mice for 2 weeks to two
daily 1-h spikes (morning and afternoon, 5 days/week) of 8460 µg/m³
(4.5 ppm) superimposed on a continuous background of 2820 µg/m³
(1.5 ppm) NO₂. Mice were challenged with the infectious agent either
immediately before or after the morning spike. When the infectious
agent was given before the morning spike, the increase in mortality

did not closely approach that of a continuous exposure to 2820 µg/m³
(1.5 ppm) NO₂. However, in mice challenged after the morning spike,
by 2 weeks of exposure, the increased mortality over controls
approached that equivalent to continuous exposure to 2820 µg/m³
(1.5 ppm) NO₂. Thus, the magnitude of the effect of the base-plus-
spike group, which had a higher C × T than the continuous groups, did
not exceed the effect of the continuous group. These findings
demonstrate that the pattern of exposure determines the response and
that the response is not predictable based on a simple C × T
relationship.

Further investigations into the effects of chronic exposure to
NO₂ spikes on murine antibacterial lung defences have been conducted
using a spike-to-baseline ratio of 4:1, which is not uncommon in the
urban environment in the USA (Miller et al., 1987). For 1 year, mice
were exposed 23 h/day, 7 days/week, to a baseline of 376 µg/m³
(0.2 ppm) or to this baseline level on which was superimposed a 1-h
spike of 1500 µg/m³ (0.8 ppm) NO₂, twice a day, 5 days/week. The
animals exposed to the baseline level did not exhibit any significant
effects; however, the streptococcal-induced mortality of the mice
exposed to the baseline plus spike regimen was significantly greater
than that of either the NO₂-background-exposed mice or the control
mice. Human epidemiological studies in chapter 7 indicate increased
risk of respiratory infection. Data from experimental animals support
the epidemiological responses in humans.

Antiviral defences are also compromised by NO₂. Squirrel
monkeys exposed to 9400 or 18 800 µg/m³ (5.0 or 10.0 ppm) NO₂ for
2 or 1 month, respectively, had an increased susceptibility to a
laboratory-induced viral influenza infection (Henry et al., 1970).
All six animals exposed to the highest concentration died within 2
to 3 days of infection with the influenza virus; at the lower
concentration, one out of three monkeys died.

Mice exposed continuously for 3 months to 564-940 µg/m³
(0.3-0.5 ppm) NO₂ followed by a challenge with A/PR/8 influenza virus
exhibited significant pulmonary pathological responses (Motomiya et
al., 1973). A greater incidence of adenomatous proliferation of
bronchial epithelial cells resulted from the combined exposures of
virus plus NO₂ than with either the viral or NO₂ exposures alone.
Continuous NO₂ exposure for an additional 3 months did not enhance
the effect of NO₂ or the subsequent virus challenge.

Ito (1971) challenged mice with influenza A/PR/8 virus after
continuous exposure to 940 to 1880 µg/m³ (0.5 to 1.0 ppm) NO₂ for
39 days and to 18 800 µg/m³ (10.0 ppm) NO₂, 2 h daily for 1, 3 and
5 days. Acute and intermittent exposure to 18 800 µg/m³ (10.0 ppm)
NO₂ as well as continuous exposure to 940 to 1880 µg/m³ (0.5 to
1.0 ppm) NO₂ increased the susceptibility of mice to influenza virus
as demonstrated by increased mortality.

The lower respiratory tract of mice became significantly more
susceptible to murine cytomegalovirus infection after 6-h exposures
for 6 days to 9400 µg/m³ (5.0 ppm) NO₂ (Rose et al., 1988). No
effects occurred at levels < 4700 µg/m³ (2.5 ppm). Exposure to
9400 µg/m³ (5.0 ppm) NO₂ did not significantly alter the course of
a parainfluenza (murine sendai virus) infection in mice as measured by
the infectious pulmonary virus titres in the lungs. However, this
concentration of NO₂, when combined with the virus exposure, did
increase the severity of the pulmonary disease process (viral
pneumonitis) (Jakab, 1988).

5.2.2.2 Lung biochemistry

Studies of lung biochemistry in animals exposed to NO₂ have
focused on either the putative mechanisms of toxic action of NO₂ or
on detection of indicators of tissue and cell damage. One theory of
the mechanism underlying NO₂ toxicity is that NO₂ initiates lipid
peroxidation in unsaturated fatty acids in membranes of target cells,
thereby causing cell injury or death (Menzel, 1976). Another theory
is that NO₂ oxidizes water-soluble, low molecular weight reducing
substances and proteins, resulting in a metabolic dysfunction that
manifests itself in toxicity (Freeman & Mudd, 1981). It is likely
that NO₂ acts by both means. Several potential biochemical mechanisms
related to detoxification of NO₂ or to responses to NO₂ intoxication
have been proposed and summarized below according to impacts on
lipids, proteins, and antioxidant metabolism and antioxidants. The
following discussion focuses on inhalation studies because they are
more interpretable for risk assessment purposes; in vitro exposure
studies have been reviewed elsewhere (US EPA, 1993).

a) Lipid peroxidation

Animal toxicology studies evaluating effects of NO₂ on lipid
peroxidation are summarized in Table 31.

Lipid peroxidation induced by NO₂ exposure has been detected at
exposure levels as low as 75 µg/m³ (0.04 ppm). Lipid peroxidation,
measured as ethane exhalation, was detected after 9 months of exposure
of rats to 75-750 µg/m³ (0.04-0.4 ppm) (Sagai et al., 1984). Lipid
peroxidation has also been evaluated by measuring the content of lipid
peroxides or substances reactive to thiobarbituric acid in alveolar
lavage fluid and lung tissue after exposure to similar NO₂
concentrations (Ichinose & Sagai, 1982; Ichinose et al., 1983). Acute
or subacute exposure to higher concentrations of NO₂ has also been
shown to cause a rapid increase in lung peroxide levels in several
species.

Table 31. Effects of nitrogen dioxide (NO₂) on lung lipid metabolism^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

75 0.04 Continuous, 9, Rat Increased TBA products at 7520 µg/m³ after 9 months and Sagai et al.
752 0.4 18 or 27 months at > 752 µg/m³ after 18 months; increased ethane (1984)
7520 4.0 exhalation at all levels. No changes in total lipid,
phospholipid, total cholesterol or triglyceride contents.

75 0.04 Continuous, 6, Increased ethane exhalation after 9 and 18 months.
225 0.12 9 and 18 months
752 0.4

752 0.4 2 weeks Rat Changes in TBA-reactive substances, exhaled ethane and Ichinose et
2260 1.2 1-16 weeks enzyme activities in lung homogenates, dependent on al. (1983)
7520 4.0 concentration and duration of exposure.
18 800 10.0

75 0.04 9, 18,
752 0.4 27 months
7520 4.0

752 0.4 4 months Rat Duration-dependent increase in ethane exhalation and Ichinose &
2260 1.2 TBA-reactive substances; peak increase in early weeks of Sagai (l982)
7520 4.0 exposure, return towards control in mid-exposure, and
increase late in exposure.

752 0.4 72 h Guinea-pig No effect at 752 µg/m³; increase in lung lipid content in Selgrade et
1880 1.0 BAL of vitamin C-depleted, but not normal, animals at al. (1981)
5640 3.0 1880 µg/m³ or more.
9400 5.0

Table 31 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9400 5.0 3 h Increased lung lipid content in vitamin C-depleted
guinea-pigs 18-h after exposure.

752 0.4 1 week No effects in normal or vitamin C-depleted animals.

1880 1.0 Continuous, Rabbit Decrease in lecithin synthesis after 1 week; less marked Seto et al.
2 weeks depression after 2 weeks. (1975)

1880 1.0 4 h/day, Rat Vitamin E supplement reduced the lipid peroxidation. Thomas et al.
6 days (1967)

5450 2.9 Continuous, Rat Increase in lung wet weight (l2.7%) and decrease in total Arner &
5 days/week lipid (8.7%); decrease in saturated fatty acid content of Rhoades (1973)
9 months lung lavage fluid and tissue; increase in surface tension
of lung lavage fluid; and decrease in lung compliance.

1880 1.0 2 h Rabbit 1800 µg/m³: elevated thromboxane B₂. 5640 µg/m³: Schlesinger
5640 3.0 depressed thromboxane B₂. 18 800 µg/m³: depressed et al. (1990)
18 800 10.0 6-keto-prostaglandin F_1alpha and thromboxane B₂.

5640 3.0 Continuous, Rat Decrease in linoleic and linolenic acid content of BAL. Menzel et al.
17 days (1972)

5640 3.0 7 days Rat Increased TBA reactants with vitamin E deficiency. Sevanian et al.
(1982)

^a Modified from US EPA (1993)
^b TBA = Thiobarbituric acid; BAL = Bronchoalveolar lavage
Lipid peroxidation results in an alteration in phospholipid
composition. Exposure of either mice or guinea-pigs to an NO₂
level of 750 µg/m³ (0.4 ppm) for a week resulted in a decreased
concentration of phosphatidyl ethanolamine and a relative increase in
the phosphatidyl choline concentration (Sagai et al., 1987).

Several investigators have also demonstrated NO₂-induced lipid
peroxidation in in vitro systems. The cell type most commonly used
is the endothelial cell from either pig arteries or aorta. Studies
using these cell types have recently attempted to relate the effect on
lipid metabolism to functional parameters such as membrane fluidity
and enzyme activation or inactivation.

Membrane fluidity changes are related to lipid peroxidation.
NO₂-induced changes in membrane fluidity have been demonstrated in
alveolar macrophages and endothelial cells in culture. Endothelial
cells exposed to a NO₂ level of 9400 µg/m³ (5 ppm), for instance,
exhibit decreased membrane fluidity after 3 h. Thus, NO₂ changes the
physical state of the membrane lipids, perhaps through initiating
lipid peroxidation, and hence impairs membrane functions (Patel et
al., 1988).

Lipid peroxidation can also activate phospholipase activities.
Activation of phospholipase A₁ in cultured endothelial cells by NO₂
has been demonstrated. This activation, which is specific for
phospholipase A₁ occurs at an NO₂ concentration of 9400 µg/m³
(5 ppm) after 40 h of exposure and is speculated to depend on a
specific NO₂-induced increase in phosphatidyl serine in the plasma
membranes (Sekharam et al., 1991).

One function of phospholipases is the release of arachidonic
acid. The effect of NO₂ on the release and metabolism of arachidonic
acid has been studied both in vivo and in vitro. Both an increase
and a decrease in the metabolism of arachidonic acid has been observed
in several species. In vivo exposure of rats to 18 800 µg/m³
(10 ppm) for 2 h resulted in decreased levels of prostaglandins E₂
and F₂alpha, as well as thromboxane B₂, in lavage fluid. On the
other hand, at an exposure level of 1880 µg/m³ (1 ppm), the
concentrations of thromboxane B₂ were increased (Schlesinger et al.,
1990).

b) Effects on lung proteins and enzymes

Nitrogen dioxide can cause lung inflammation (associated with
concomitant infiltration of serum protein, enzymes and inflammatory
cells) and hyperplasia of Type 2 cells. Thus, some changes in lung
enzyme activity and protein content may reflect inflammation and/or
changes in cell types, rather than direct effects of NO₂ on lung cell
enzymes. Some direct effects of NO₂ on enzymes are possible because

NO₂ can oxidize various reducible amino acids or side chains of
proteins in aqueous solution (Freeman & Mudd, 1981). These effects
are summarized in Table 32.

Nitrogen dioxide can increase the protein content of BAL in
vitamin-C-deficient guinea-pigs (Sherwin & Carlson, 1973; Selgrade et
al., 1981; Hatch et al., 1986; Slade et al., 1989). Selgrade et al.
(1981) found effects at NO₂ levels as low as 1880 µg/m³ (1.0 ppm)
after a 72-h exposure, but a 1-week exposure to 752 µg/m³ (0.4 ppm)
did not increase protein levels. The results of the 1-week exposure
apparently conflict with those of Sherwin & Carlson (1973), who found
increased protein content of BAL from vitamin-C-deficient guinea-pigs
exposed to 752 µg/m³ (0.4 ppm) NO₂ for 1 week. Differences in
exposure techniques, protein measurement methods, and/or degree of
vitamin C deficiencies may explain the difference between the two
studies. Hatch et al. (1986) found that the NO₂-induced increase in
BAL protein in vitamin-C-deficient guinea-pigs was accompanied by an
increase in lung content of non-protein sulfhydryls and ascorbic acid
and a decrease in vitamin E content. The increased susceptibility to
NO₂ was observed when lung vitamin C was reduced (by diet) to levels
below 50% of normal. A depletion of lung non-protein sulfhydryls also
enhances susceptibility to a high level (18 800 µg/m³, 10.0 ppm) of
NO₂ (Slade et al., 1989).

The effects of NO₂ on structural proteins of the lungs has been
of major interest because elastic recoil is lost after exposure
(section 5.2.2.3). Last et al. (1983) examined collagen synthesis
rates by lung minces from animals exposed to NO₂. In rats
continuously exposed to 9400 to 47 000 µg/m³ (5.0 to 25.0 ppm) NO₂
for 7 days, there was a linear concentration-related increase in
collagen synthesis rate. In a subsequent paper, Last & Warren (1987)
confirmed that 9400 µg/m³ (5.0 ppm) increased collagen synthesis.
Such biochemical changes are typically interpreted as reflecting
increases in total lung collagen, which, if sufficient, could result
in pulmonary fibrosis after longer periods of exposure. However, such
correlations have not been made directly after NO₂ exposure.

Table 32. Effects of nitrogen dioxide (NO₂) on lung proteins and enzymes^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

75 0.04 Continuous, Rat NPSHs increased at the 2 higher NO₂ levels after 9 or 18 Sagai et al.
752 0.4 9 and 18 months months; GSH peroxidase activity decreased at 752 µg/m³ (1984)
7520 4.0 after 18 months and at 7520 µg/m³ after 9 or 18 months;
GSH reductase activity increased after a 9-month exposure
to 7520 µg/m³; G-6-PD was increased after a 9- or 18-month
exposure to 7520 µg/m³; no effects on 6-phosphogluconate
dehydrogenase, superoxide dismutase, or disulfide
reductase; some GSH S-transferases had decreased
activities after an 18-month exposure to 752 or
7520 µg/m³.

752 0.4 72 h Guinea-pig No effect at 752 µg/m³; increase in BAL protein in Selgrade et
1880 1.0 vitamin-C-depleted but not normal animals at > 1880 µg/m³. al. (1981)
5640 3.0
9400 5.0

9400 5.0 3 h Increased BAL protein in vitamin-C-depleted guinea-pigs
15-h post-exposure.

752 0.4 Continuous, No effect on BAL protein in vitamin-C-depleted guinea-pigs.
1 week

752 0.4 Continuous, Guinea-pig Increase in BAL protein content of guinea-pigs with an Sherwin &
1 week unquantified vitamin C deficiency. Carlson (1973)

Table 32 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

752 0.4 1 to 14 weeks Rat Complex concentration and duration dependence of effects. Takahashi et
2260 1.2 Example: at 752 µg/m³, cytochrome P-450 levels decreased al. (1986)
7520 4.0 at 2 weeks, returned to control level by 5 weeks. At
2260 µg/m³, cytochrome P-450 levels decreased initially,
increased at 5 weeks, and decreased at 10 weeks. Effects
on succinate-cytochrome c reductase also.

752 0.4 4 months Rat Duration-dependent pattern for increase in activities of Ichinose &
2260 1.2 antioxidant enzymes; increase, peaking at week 4, and Sagai (1982)
7520 4.0 then decreasing; concentration-dependent effects.

752 0.4 2 weeks Rat No effect on TBA reactants, antioxidants or antioxidant Ichinose &
Guinea-pig enzyme activities. Sagai (1989)

752 0.4 7 days Rat Decrease in cytochrome P-450 level at > 2260 µg/m³. Mochitate et
2260 1.2 al. (1984)
7520 4.0

846 0.45 7 h/day Mouse No changes in lung serotonin levels. Sherwin et
4 weeks al. (1986)

884 0.47 Continuous, 10, Mouse Increased content of serum proteins in homogenized whole Sherwin &
12, 14 days lung tissue. Layfield (1974)

940 0.5 Continuous, Mouse Decrease in lung GSH peroxidase activity at 1880 µg/m³ Ayaz &
1880 1.0 17 months in vitamin-E-deficient mice. Increased activity in Csallany (1978)
vitamin-E-supplemented mice at > 940 µg/m³.

Table 32 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

1880 1.0 Continuous, Rat Activities of GSH reductase and G-6-PD increased at Chow et al.
4320 2.3 4 days 11 700 µg/m³ proportional to duration of exposure; no (1974)
11 700 6.2 effect on GSH peroxidase. No effects at < 4320 µg/m³.

1880 1.0 15 weeks Rat Changes in BAL fluid and lung tissue levels of enzymes Gregory et al.
9400 5.0 early in exposure; resolved by 15 weeks. (1983)

3760 2.0 3 days Rat Decreased superoxide dismutase activity. Azoulay-Dupuis
18 800 10.0 Guinea-pig et al. (1983)

3760 2.0 Continuous, Rat Increased activities of several glycolytic enzymes. Mochitate et
7520 4.0 7, 10, 14 days At < 7520 µg/m³, pyruvate kinase increased on days al. (1985)
4 and 7; recovery occurred by day 14. G-6-PD increased
at all levels; at 3760 µg/m³, 14 days of exposure needed.

3760 2.0 1-7 days Rat Increased lung protein content; increase in microsomal Mochitate et
7520 4.0 succinate cytochrome c reductase activity. al. (1984)
18 800 10.0

5640 3.0 7 days Rat Various changes in lung homogenate protein and DNA Elsayed &
content and enzyme activities; changes more severe in Mustafa (1982)
vitamin-E-deficient rats.

5640 3.0 7 days Rat No effects on antioxidant metabolism or oxygen Mustafa et al.
9400 5.0 4 days consumption enzymes at < 9400 µg/m³. (1979)

7520 4.0 7, 14 and Rat Increased gamma-glutamyl transferase on days 14 and 21; Hooftman et al.
18 800 10.0 21 days no consistent effect on alkaline phosphatase, lactate (1988)
47 000 25.0 dehydrogenase or total protein.

Table 32 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9020 4.8 3 h Guinea-pig Increased BAL protein content in vitamin-C-deficient Hatch et al.
guinea-pigs. (1986)

8460 4.5 16 h Increased lung wet weight, alterations in lung
antioxidant levels in vitamin-C-deficient guinea-pigs.

9020 4.8 7 days Mouse No significant changes in lung homogenate parameters. Mustafa et al.
(1984)

9400 5.0 14-72 h Mouse Increase in lung protein (14 to 58 h) by radioactive Csallany (1975)
label incorporation.

9400-47 000 5.0-25.0 Continuous, Rat Concentration-related increase in rate of collagen Last et al.
7 days synthesis; 125% increase at 9400 µg/m³. (1983)

9400 5.0 3 h Rabbit Benzo[a]pyrene hydroxylase activity of tracheal mucosa Palmer et al.
37 600 20.0 not affected. (1972)
94 000 50.0

^a Modified from US EPA (1993)
^b NPSHs = Non-protein sulfhydryls; GSH = Glutathione; G-6-PD = Glucose-6-phosphate dehydrogenase; BAL = Bronchoalveolar lavage
Alterations in lung xenobiotic metabolism follow a complex
duration of exposure pattern in rats exposed to 752, 2260 and
7520 µg/m³ (0.4, 1.2 and 4.0 ppm) NO₂ (Takahashi et al., 1986). At
the lowest NO₂ concentration tested, cytochrome P-450 levels
decreased initially (at 2 weeks) and then returned to control levels
by 5 weeks, where they remained throughout exposure. At 2260 µg/m³
(1.2 ppm), cytochrome P-450 levels decreased initially, then increased
after 5 weeks of exposure and decreased again by 10 weeks. A similar
pattern of response occurred at the highest concentration. Only
7520 µg/m³ (4.0 ppm) NO₂ affected other microsomal electron-
transport systems. The activity of succinate-cytochrome c reductase
was decreased by 14 weeks of exposure to 752 µg/m³ (0.4 ppm), but at
the higher NO₂ levels, the activity was decreased sooner. In
contrast, Mochitate et al. (1984) also found a decrease in levels of
cytochrome P-450 at > 2260 µg/m³ (1.2 ppm) in rats exposed for
7 days.

Glycolytic pathways are also increased by NO₂ exposure,
apparently due to a concurrent increase in Type 2 cells (Mochitate et
al., 1985). The most sensitive enzyme was pyruvate kinase, exhibiting
an increased activity after a 14-day exposure to 3760 µg/m³ (2.0 ppm)
NO₂. At higher NO₂ concentrations (e.g., 7520 µg/m³, 4.0 ppm),
pyruvate kinase activity increased sooner (4 and 7 days) and then
decreased to control levels by 14 days.

c) Antioxidant defence systems

Since NO₂ is an oxidant and lipid peroxidation is believed to be
a major molecular event responsible for the toxic effects of NO₂,
much attention has been focused on the effect of the antioxidant
defence system in the epithelial lining fluid and in pulmonary cells.
Investigations with subacute and chronic NO₂ exposure levels of 75
to 62 040 µg/m³ (0.04-33 ppm) have been performed both in vivo
and in vitro and focussed on effects on low molecular weight
antioxidants such as glutathione, vitamin E and vitamin C, as well as
on some enzymes involved in the synthesis and catabolism of
glutathione. Experiments made in vitro using human plasma have
shown a rapid depletion of vitamin C and glutathione and a loss of
vitamin E. This result was achieved with a concentration of
26 320 µg/m³ (14 ppm) (Halliwel et al., 1992).

Menzel (1970) proposed that antioxidants might protect the lung
from NO₂ damage by inhibiting lipid peroxidation. Data related to
this hypothesis have been reported (Thomas et al., 1968; Menzel et
al., 1972; Fletcher & Tappel, 1973; Csallany, 1975; Ayaz & Csallany,
1978; Slade et al., 1989). Several laboratories have observed changes
in the activity of enzymes in the lungs of NO₂-exposed animals that
regulate levels of glutathione (GSH), the major water-soluble
reductant in the lung. Chow et al. (1974) exposed rats to 1880, 4320
or 11 700 µg/m³ (1.0, 2.3 or 6.2 ppm) NO₂ continuously for 4 days to
examine the effect on activities of GSH reductase, glucose-6-phosphate
dehydrogenase and GSH peroxidase in the soluble fraction of exposed
rat lungs. Linear regression analysis of the correlation between the
NO₂ concentration and enzymatic activity showed a significant
positive correlation coefficient of 0.63 for GSH reductase and of 0.84
for glucose-6-phosphate dehydrogenase. No correlation was found
between the GSH peroxidase activity and the NO₂ concentration. The
activities of GSH reductase and glucose-6-phosphate dehydrogenase were
significantly increased during exposure to 11 700 µg/m³ (6.2 ppm)
NO₂; GSH peroxidase activity was not affected. The possible role of
oedema and cellular inflammation in these findings was not examined.
These researchers concluded that after a slightly longer exposure
(14 days), 3760 µg/m³ (2.0 ppm) NO₂ increased the activity of
glucose-6-phosphate dehydrogenase in rats (Mochitate et al., 1985).
There is evidence from recent studies that glutathione and vitamins C
and E are all involved in normal protection of the lung from NO₂
(Rietjens et al., 1986; Hatch et al., 1986; Slade et al., 1989).

Sagai et al. (1984) studied the effects of prolonged (9 and 18
months) exposure to 75, 752 and 7520 µg/m³ (0.04, 0.4 and 4.0 ppm)
NO₂ on rats. After both exposure durations, non-protein sulfhydryl
levels were increased at > 752 µg/m³; exposure to 7520 µg/m³
(4.0 ppm) decreased the activity of GSH peroxidase and increased
glucose-6-phosphate dehydrogenase activity. Glutathione peroxidase
activity was also decreased in rats exposed to 752 µg/m³ NO₂ for
18 months. Three GSH S-transferases were also studied, two of which
(aryl S-transferase and aralkyl S-transferase) exhibited decreased
activities after 18 months of exposure to > 752 µg/m³ NO₂. No
effects were observed on the activities of 6-phosphogluconate
dehydrogenase, superoxide dismutase or disulfide reductase. When
effects were observed, they followed a concentration and exposure-
duration response function. The decreases in antioxidant metabolism

were inversely related to the apparent formation of lipid peroxides
(see lipid peroxidation subsection). Shorter exposures (4 months) to
NO₂ between 752 and 7520 µg/m³ (0.4 and 4.0 ppm) also caused
concentration- and duration-dependent effects on antioxidant enzyme
activities (Ichinose & Sagai, 1982). For example, glucose-6-phosphate
dehydrogenase increased, reaching a peak at 1 month, and then
decreased towards the control value. Briefer (2-week) exposures to
752 µg/m³ (0.4 ppm) NO₂ caused no such effects in rats or
guinea-pigs (Ichinose & Sagai, 1989).

Ayaz & Csallany (1978) exposed vitamin-E-deficient and vitamin-E-
supplemented mice continuously for 17 months to 940 or 1880 µg/m³
(0.5 or 1.0 ppm) NO₂ and assayed them for GSH peroxidase activity.
Exposure to 1880 µg/m³ (1.0 ppm) NO₂ decreased enzyme activity in
the vitamin-E-deficient mice. However, in vitamin-E-supplemented
mice, GSH peroxidase activity increased at 940 µg/m³ (0.5 ppm) NO₂.

5.2.2.3 Pulmonary function

Animal studies of NO₂ effects on pulmonary function are
summarized in Table 33. NO₂ concentrations in many urban areas of
the USA and elsewhere consist of spikes superimposed on a relatively
constant background level. Miller et al. (1987) evaluated this urban
pattern of NO₂ exposure in mice using continuous 7 days/week,
23 h/day exposures to 376 µg/m³ (0.2 ppm) NO₂ with twice daily
(5 days/week) 1-h spike exposures to 1500 µg/m³ (0.8 ppm) NO₂ for
32 and 52 weeks. Mice exposed to clean air and to the constant
background concentration of 376 µg/m³ (0.2 ppm) served as controls.
Vital capacity tended to be lower (p = 0.054) in mice exposed to NO₂
with diurnal spikes than in mice exposed to air. Lung distensibility,
measured as respiratory system compliance, also tended to be lower in
mice exposed to diurnal spikes of NO₂ compared with constant NO₂
exposure or air exposure. These changes suggest that up to 52 weeks
of low-level NO₂ exposure with diurnal spikes may produce a subtle
decrease in lung distensibility, although part of this loss in
compliance may be a reflection of the reduced vital capacity. Vital
capacity appeared to remain suppressed for at least 30 days after
exposure. Lung morphology in these mice was evaluated only by light
microscopy (a relatively insensitive method) and showed no exposure-
related lesions. The decrease in lung distensibility suggested by
this study is consistent with the thickening of collagen fibrils in
monkeys (Bils, 1976) and the increase in lung collagen synthesis rates
of rats (Last et al., 1983) after exposure to higher levels of NO₂.

Tepper et al. (1993) exposed 60-day-old rats to 940 µg/m³
0.5 ppm) NO₂, 22 h/day, 7 days/week, with a 2-h spike of 2820 µg/m³
(1.5 ppm) NO₂, 5 days/week for up to 78 weeks. There were no effects
on pulmonary function between 1 and 52 weeks of exposure. Following
78 weeks of exposure, flow at 25% forced vital capacity was decreased,
perhaps indicating airway obstruction. A significant decrease in the
frequency of breathing was also observed at 78 weeks that was
paralleled by a trend toward increased expiratory resistance and
expiratory time. Taken together, these results suggest that few, if
any, significant effects were seen that suggest incipient lung
degeneration.

The age sensitivity to exposure to diurnal spikes of NO₂ was
studied by Stevens et al. (1988), who exposed 1-day- and 7-week-old
rats to continuous baselines of 940, 1880 and 3760 µg/m³ (0.5, 1.0
and 2.0 ppm) NO₂ with twice daily 1-h spikes at 3 times these
baseline concentrations for 1, 3 and 7 weeks. In neonatal rats, vital
capacity and respiratory system compliance increased following 3
weeks, but not 6 weeks, of exposure to the 1880 and 3760 µg/m³ NO₂
baselines with spikes. In young adult rats, respiratory system
compliance decreased following 6 weeks of exposure, and body weight
decreased following 3 and 6 weeks of exposure to the 3760 µg/m³
baseline with spike. In the young adult rats, pulmonary function
changes returned to normal values 3 weeks after exposure ceased. A
correlated morphometric study (Chang et al., 1986) is summarized in
section 5.2.2.4.

Lafuma et al. (1987) exposed 12-week-old hamsters with and
without laboratory-induced (elastase) emphysema to 3760 µg/m³
(2.0 ppm) NO₂, 8 h/day, 5 days/week for 8 weeks. Vital capacity
and pulmonary compliance were not affected by NO₂ exposure.

5.2.2.4 Morphological studies

Inhalation of NO₂ produces morphological alterations in the
respiratory tract, as summarized in Tables 34 and 35. This
discussion is generally limited to those studies using NO₂ levels
< 9400 µg/m³ (5.0 ppm), but results of studies of emphysema
conducted at higher concentrations are also discussed. Examination of
the tables shows variability in responses at similar exposure levels
in different studies. This may be due to differences in animal
species or strain, age, diet, microbiological status of the animals,

or aspects of experimental protocol. The latter includes the
methodology used to evaluate the morphological response. For example,
simple light microscopic examination may reveal no effect, whereas
other techniques, such as quantitative morphological (morphometric)
procedures with electron microscopy, can detect more subtle structural
changes.

There is a large degree of interspecies variability in
responsiveness to NO₂; this is clearly evident from those few studies
where different species were exposed under identical conditions
(Wagner et al., 1965; Furiosi et al., 1973; Azoulay-Dupuis et al.,
1983). Variability in response may be due to differences in effective
dose of NO₂ reaching target sites, but other species differences are
likely to play a role. Guinea-pigs, hamsters and monkeys all appear
to be more severely affected morphologically by equivalent exposure to
NO₂ than are rats, the most commonly used experimental animal.
However, in most cases, similar types of histological lesions are
produced when similar effective concentrations are used.

a) Sites affected and time course of effects

The anatomic region most sensitive to NO₂ and within which
injury is first noted is the centriacinar region. This region
includes the terminal conducting airways (terminal bronchioles),
respiratory bronchioles, and adjacent alveolar ducts and alveoli.
Within this region, those cells that are most sensitive to
NO₂-induced injury are the ciliated cells of the bronchiolar
epithelium and the Type 1 cells of the alveolar epithelium, which are
then replaced with non-ciliated bronchiolar (Clara) cells and Type
2 cells, respectively. In addition to these dynamic changes,
permanent alterations resembling emphysema-like disease may result
from chronic exposure.

The temporal progression of early events due to NO₂ exposure has
been described best in the rat (e.g., Freeman et al., 1966, 1968c,
1972; Stephens et al., 1971a, 1972; Evans et al., 1972, 1973a,b, 1974,
1975, 1976, 1977; Cabral-Anderson et al., 1977; Rombout et al., 1986)
and guinea-pig (Sherwin et al., 1973). The earliest alterations
resulting from exposure to concentrations of > 3760 µg/m³
(2.0 ppm) are seen within 24 to 72 h of exposure and include increased
AM aggregation, desquamation of Type 1 cells and ciliated bronchiolar
cells, and accumulation of fibrin in small airways. However, repair
of injured tissue and replacement of destroyed cells can begin within
24 to 48 h of continuous exposure. Hyperplasia of nonciliated
bronchiolar (Clara) cells occurs in the bronchioli, whereas in the
alveoli, the damaged Type 1 cells are replaced with Type 2 cells.
These new cells are relatively resistant to effects of continued NO₂
exposure.

Table 33. Effects of nitrogen dioxide (NO₂) on pulmonary function^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

376 0.2 23 h/day base Mouse Decreased vital capacity following base + spike Miller et
(7 days/week), 1-h NO₂ exposures compared with control and base NO₂ al. (1987)
376 base, 0.2 base, peaks twice/day, exposures. Tendency toward decreased respiratory
1500 peak 0.8 peak 32 and 52 weeks system compliance following spike NO₂ exposures
compared and control and base NO₂ exposures.

940 base, 0.5 base, 23 h/day Rat (1-day Increased lung volume and compliance in neonates Stevens et
2820 peak 1.5 peak (7 days/week) base, and following 3-week, but not 6-week, exposure to the al. (1988)
1-h peaks twice/day 7-weeks two higher exposure levels. Decreased body weight
1880 base, 1.0 base, (5 days/week); old) and lung compliance in adult rats following 6-week
5640 peak 3.0 peak 1, 3 and 6 weeks exposure to 3760 µg/m³ + spike. Adults recovered
3 weeks after exposure.
3760 base, 2.0 base,
11 300 peak 6.0 peak

940 base, 0.5 base, 22 h/day (7 days per Rat Decreased delta FEF₂₅ and frequency of breathing Tepper et al.
2820 peak 1.5 peak week), 2-h peak following 78-week NO₂ exposure. (1993)
(5 days/week); 1,
3, 12, 52 and
78 weeks

3760 2.0 8 h/day, Hamster No change in vital capacity or lung compliance Lafuma et al.
5 days/week, following NO₂ exposures in both normal and (1987)
8 weeks elastase-treated animals.

10 200 5.4 3 h/day for 7, 14 Rat Tendency toward increased lung volume at low Yokoyama et
or 30 days inflation pressures. al. (1980)

^a Modified from: US EPA (1993)
^b PaO₂ = Arterial oxygen tension; delta FEF₂₅ = Change in forced expiratory flow at 25% of forced vital capacity;
PaCO₂ = Arterial carbon dioxide tension

Table 34. Effects of acute and subchronic exposure to nitrogen dioxide (NO₂) on lung morphology^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

207 0.11 Continuous, Rat (1, Various morphometric changes, depending on age Kyono & Kawai
865 0.46 1 month 3, 12, and exposure level. Multiphasic pattern (e.g., (1982)
5260 2.8 21 months decrease in air-blood barrier thickness from 1 to
16 500 8.8 old) 12 months of age, and increase in 21-month-old
rats).

639 0.34 6 h/day, 5 days Mouse Type 2 cell hypertrophy and hyperplasia; increase Sherwin &
per week, 6 weeks in mean linear intercept and amount of alveolar Richters (1982)
wall area.

940 0.5 4 h Rat Loss of cytoplasic granules in and rupture of Thomas et al.
mast cells. (1967)

940 0.5 Continuous, up Rat Increased number of mast cells in trachea as exposure Hayashi et al.
to 6 days duration increased. (1987)

Table 34 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

940 base, 0.5 base, 23 h/day (7 days Rat (1 day In proximal alveolar region: base (940 µg/m³) + peak Crapo et al.
2820 peak 1.5 peak per week) base, 1-h and caused Type 2 cells to become spread over more (1984); Chang
peaks twice/day 6 weeks surface area in neonates and adults; Type 2 cell et al. (1986,
3760 base, 2.0 base, (5 days/week); old) hypertrophy and increase in number of AMs in adults; 1988)
11 280 peak 6.0 peak 6 weeks Type 2 cells thinner in neonates. Base (3760 µg/m³)
+ peak (only adults studied) caused similar changes
plus an increase in numbers of Type 1 cells, which
were smaller than normal Type 1 cells.
In terminal bronchiolar region: base (940 µg/m³) +
peak caused no effects on percentage distribution of
ciliated cells and Clara cells in neonates or adults,
but neonates (only) had a increase in ciliated cell
surface area and mean luminal surface area of Clara
cells. Base (3760 µg/m³) + peak (only adults studied)
had fewer ciliated cells with a reduced surface area
and alterations in the shape of Clara cells.

1000 0.53 Continuous Rat At < 2500 µg/m³: no pathology. At 5000 µg/m³: focal Rombout et al.
2500 1.33 (24 h/day) thickening of centriacinar septa by 2 days; progressive (1986)
5000 2.66 28 days loss of cilia and abnormal cilia in trachea and main
bronchi at > 4 days; hypertrophy of bronchiolar
epithelium at > 8 days.
At days 16 and 28, all epithelial cells hypertrophied.

Table 34 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

1000 0.53 24 h/day, Guinea- No pathology Steadman et al.
90 days pig, (1966)
rabbit,
dog,
monkey,
rat

1320-1500 0.7-0.8 Continuous, Mouse Mucous hypersecretion; focal degeneration and Nakajima et al.
1 month desquamation of mucous membrane; terminal (1980)
bronchiolar epithelial hyperplasia; some alveolar
enlargement; shortening of cilia.

1880 1-1.5 Continuous, Mouse Terminal bronchiolar epithelial hyperplasia; some Nakajima et al.
2820 1 month alveolar enlargement. (1980)

1880 1.0 1 h Rat Degranulation and decreased number of mast cells. Thomas et al.
(1967)

3760 2.0 3 days Rat No historical changes Azoulay-Dupuis
et al. (1983)

3760 2.0 3 days Guinea-pig Thickening of alveolar walls; oedema; increase in Azoulay-Dupuis
AM numbers; loss of bronchiolar cilia; inflammation. et al. (1983)

Table 34 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

3760 2.0 8 h/day, Hamster Moderate alveolar enlargement, primarily at Lafuma et al.
5 days/week, bronchiolar-alveolar duct junction; increase in mean (1987)
8 weeks linear intercept; decrease internal surface area of
lung; no lesions in bronchial, bronchiolar, alveolar
duct, or alveolar epithelium; no change in
macrophage number.

3760 2.0 Continuous, Guinea-pig Type 2 cell hypertrophy at 7 or 21 days. Sherwin et al.
7-21 days (1973)

3760 2.0 Continuous, Guinea-pig Increase in number of LDH-positive cells with time Sherwin et al.
1-3 weeks of exposure. Correlated to increase in Type 2 cells (1973)
(LDH positive).

3760 2.0 Continuous, Rat Minimal effect: some cilia loss in terminal bronchioles; Azoulay et al.
6 weeks some distended or disrupted alveolar walls. (1978)

9400 5.0 Continuous, Cynomolgus Bronchiolar epithelia hyperplasia; some focal Busey et al.
18 800 10.0 90 days monkey pulmonary odema. (1974)

^a Modified from US EPA (1993)
^b AMs = Alveolar macrophages; LDH = Lactate dehydrogenase

Table 35. Effects of chronic exposure to nitrogen dioxide (NO₂) on lung morphology^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

75 0.04 Continuous, Rat At 75 µg/m³: no significant change, but some tendency Kubota et al.
752 0.4 9-27 months towards increase in arithmetic mean thickness of air-blood (1987)
7520 4.0 barrier. At 752 µg/m³: slight increase in arithmetic
mean thickness of air-blood barrier by 18 months, becoming
significant by 27 months; some interstitial oedema and
slight change in bronchiolar and alveolar epithelium by
27 months. At 7520 µg/m³: hypertrophy and hyperplasia of
bronchiolar epithelium and increase in arithmetic mean
thickness of air-blood barrier by 9 months, which became
significant at 18 months and decreased slightly by
27 months; Clara cell hyperplasia. By 27 months:
interstitial fibrosis and hypertrophy of Type 1 and
Type 2 cells.

188 base; 0.1 base; Continuous Mouse Dilated airspaces and aveolar wall destruction (small Port et al.
1880 peak 1.0 peak baseline; 2-h sample size). (1977)
daily peak;
6 months

Table 35 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

940 0.5 Continuous, Rat At 940 µg/m³: swelling of terminal bronchiolar cilia and Yamamoto &
1880 1.0 7 months hyperplasia of Type 2 cells. Takahashi
7520 4.0 At 1880 µg/m³: cilia loss in terminal bronchioles; (1984)
hyperplasia of Type 2 cells; and interstitial oedema.
At 7520 µg/m³: cilia loss in terminal bronchioles;
hyperplasia of Type 2 cells, interstitial oedema; decrease
in number of lamellar bodies in Type 2 cells; lysosomes
with osmiophilic lamellar structure in ciliated cells of
terminal bronchioles.

940 0.5 Continuous, up Rat Type 2 cell hypertrophy and interstitial oedema by Hayashi et al.
to 19 months 4 months; increased thickness of alveolar septa by (1987)
6 months; fibrous pleural thickening by 19 months.

940 0.5 6-24 h/day, Mouse 3 months: pneumonitis and alveolar size increase; loss of Blair et al.
3-12 months cilia in respiratory bronchioles and bronchiolar (1969)
inflammation with 24 h/day.
6-12 months: pneumonitis; cilia loss; bronchial and
bronchiolar inflammation; alveolar size increase.

1500 0.8 Continuous, Rat Minimal changes: slight enlargement of alveoli and Freeman et
lifetime (up alveolar ducts; some rounding of bronchial and bronchiolar al. (1966)
to 33 months) epithelial cells; increase in elastic fibers around
alveolar ducts.

1880 1.0 Continuous, Squirrel No pathology Fenters et
16 months monkey al. (1973)

Table 35 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

1880 1.0 6 h/day, Dog At 1880 µg/m³ - 6 months: no pathology; 12 months: Wagner et
5 days/week, up dilated alveoli and alveolar ducts; 18 months: al. (1965)
to 18 months dilated alveoli, oedema, thickening alveolar septa
due to inflammation.
9400 5.0 At 9400 µg/m³ - 6 months: no pathology; 12 months:
dilated alveolar ducts; 18 months: oedema, congestion,
and thickened alveolar septa due to inflammatory cells.

1880 1.0 6 h/day Guinea-pig Mild thickening of alveolar septa due to inflammation; Wagner et
5 days/week, some alveolar dilatation. al. (1965)
18 months

1880 1.0 7 h/day, Rat No pathology Gregory et
5 days/week, al. (1983)
15 weeks

3760 2.0 Continuous, Rat Loss of cilia in terminal bronchioles; abnormal Stephens et
2 years ciliogenesis; crystalloid inclusions in bronchiolar al. (1971a,b)
epithelial cells; increased thickness of collagen fibrils
and basement membrane in terminal bronchioles.

3760 2.0 Continuous, up Rat Hypertrophy of ciliated cells and cilia loss by 72 h; Stephens et
to 12 months decreased number of ciliated cells by 7 days; normal al. (1972)
ciliated cells from 21 days-12 months.

3760 2.0 Continuous, up Rat No change in turnover of terminal bronchiolar epithelial Evans et al.
to 360 days cells; increase in turnover of Type 2 cells in peripheral (1972)
alveoli by 1 day, but normal by 7 days.

Table 35 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

3760 2.0 Continuous, Monkey Bronchiolar epithelial hypertrophy, especially adjacent Furiosi et al.
14 months (Macaca to alveolar ducts; change to cuboidal cells in proximal (1973)
peciosa) bronchiolar epithelium.

3760 2.0 Continuous, Rat Minimal effect: some terminal bronchiolar epithelial Furiosi et al.
14 months hypertrophy. (1973)

3760 2.0 Continuous, Rat Alveolar distension, especially near alveolar duct level; Freeman et
lifetime (up to increased variability in alveolar size; loss of cilia and al. (1968b)
763 days); 1500 hypertrophy in terminal bronchiolar cells; no
µg/m³ for 1st inflammation.
69 days, then
3760 µg/m³

7520 4.0 Continuous, Rat Bronchial epithelial hyperplasia Haydon et al.
16 months (1965)

9400 5.0 6 h/day, Mouse No pathology Wagner et al.
5 days/week, (1965)
14 months

9400 5.0 4-7.5 h/day, Guinea-pig Some dilatation of terminal bronchioles; tracheal Balchum et al.
5 days/week, inflammation; pneumonitis. (1965)
5.5 months

Table 35 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9400 5.0 7 h/day, Rat Focal hyperinflation and areas of subpleural accumulation Gregory et al.
5 days/week, of macrophages. (1983)
15 weeks

^a Modified from US EPA (1993)
The time course of alveolar lesions over a chronic exposure was
examined by Kubota et al. (1987) in small groups of rats exposed to
7520 µg/m³ (4.0 ppm) NO₂, 24 h/day for up to 27 months. One phase,
which lasted for 9 to 18 months of exposure, consisted of a decrease
in number and an increase in cell volume of Type 1 epithelium, an
increase in the relative ratio of Type 2 to Type 1 cells, and an
increase in the number and volume of Type 2 cells. A second phase,
between 18 to 27 months of exposure, showed some recovery of the
alveolar epithelium, but the total volume of interstitial tissue
decreased and collagen fibres in the interstitium increased. Thus,
some lesions resolved with continued exposure, whereas others
progressed. At 752 µg/m³ (0.4 ppm), Kubota et al. (1987) found that
the lesion typically was milder and its initiation delayed, compared
to the higher concentration. In general, most NO₂-induced lesions
were resolved following a recovery period. This period may be as
short as 30 days for exposures at < 9400 µg/m³ (5.0 ppm). With
continuous exposure, early morphological damage may also be resolved.
For example, in rats exposed continuously for 7 months to 940 µg/m³
(0.5 ppm) NO₂, resolution of epithelial lesions occurred by 4 to
6 months of exposure (Yamamoto & Takahashi, 1984).

b) Effects of nitrogen dioxide as a function of exposure pattern

Several morphological studies were designed to evaluate ambient
NO₂ patterns consisting of a low baseline level with transient spikes
of NO₂. However, in some cases, there was no group at the baseline
exposure only, preventing evaluation of the contribution of peaks to
the responses. Gregory et al. (1983) exposed rats (14 to 16 weeks
old) for 7 h/day, 5 days/week for up to 15 weeks to atmospheres
consisting of the following concentrations of NO₂: (1) 1880 µg/m³
(1.0 ppm), (2) 9400 µg/m³ (5.0 ppm), or (3) 1880 µg/m³ (1.0 ppm)
with two 1.5-h spikes of 9400 µg/m³ (5.0 ppm) per day. After
15 weeks of exposure, histopathology was minimal, with focal
hyperinflation and areas of subpleural accumulation of macrophages
found in some of the animals exposed either to the baseline of
9400 µg/m³ (5.0 ppm) or to 1880 µg/m³ (1.0 ppm) with the 9400 µg/m³
(5.0 ppm) spikes.

Port et al. (1977) observed dilated respiratory bronchioles and
alveolar ducts in mice exposed to 188 µg/m³ (0.1 ppm) NO₂ with daily
2-h peaks to 1880 µg/m³ (1.0 ppm), for 6 months. Miller et al.
(1987) found no morphological effects in mice exposed for 1 year,
although host defence and functional changes were noted (see sections
5.2.2.1 and 5.2.2.3).

Crapo et al. (1984) and Chang et al. (1986) used quantitative
morphometric analyses to examine the proximal alveolar and terminal
bronchiolar regions of rats exposed for 6 weeks to a baseline
concentration of 940 or 3760 µg/m³ (0.5 or 2.0 ppm) NO₂, 23 h/day
for 7 days/week, onto which were superimposed two daily 30 min spikes
of 3 times the baseline concentration for 5 days/week. At the lower
exposure level, the volumes of the Type 2 epithelium, interstitial
matrix, and AMs increased, whereas the volume of the fibroblasts
decreased. The surface area of Type 2 cells increased. Most of these
changes also occurred at the higher exposure level, and in some cases
the change was greater than that at the lower level (i.e., increase in
Type 1 and Type 2 epithelial volume). At both levels of exposure, the
volume of Type 2 cells and interstitial fibroblasts increased, with no
significant changes in their numbers, and the number of AMs decreased.
The number of Type 1 cells decreased and their average surface area
increased in the highest exposure group. Generally, there was a
spreading and hypertrophy of Type 2 cells. A correlation between
decreased compliance (Stevens et al., 1988) and thickening of the
alveolar interstitium was found (see section 5.2.2.3 for details of
the pulmonary function portion of the study). Examination of the
terminal bronchiolar region revealed no effects at the lower exposure
level. At the higher level, there was a 19% decrease in ciliated
cells per unit area of the epithelial basement membrane and a
reduction in the mean ciliated surface area. The size of the dome
protrusions of non-ciliated bronchiolar (Clara) cells was decreased,
giving the bronchial epithelium a flattened appearance, but there was
no change in the number of cells.

c) Factors affecting susceptibility to morphological changes

Age-related responsiveness to an urban pattern of NO₂ was
evaluated by Chang et al. (1986, 1988) using 1-day- or 6-week-old rats
exposed for 6 weeks to a baseline of 940 µg/m³ (0.5 ppm) NO₂ for
23 h/day, 7 days/week, with two 1-h spikes (given in the morning and
afternoon) of 2820 µg/m³ (1.5 ppm) 5 days/week. Electron microscopic
morphometric procedures were used. In the proximal alveolar region,
only the older animals showed an increase in the surface density of
the alveolar basement membrane. The increase in the mean cellular
volume of Type 2 cells was greater in the young adult animals,
although the neonates also exhibited this effect. Although there was
no qualitative evidence of morphological injury in the terminal
bronchioles of the neonatal rats, there was a 19% increase in the
average ciliated cell surface and a 13% increase of the mean luminal

surface area of non-ciliated bronchiolar (Clara) cells that was not
evident in the young adult rats. Generally, the neonatal rats were as
sensitive or more susceptible than young adults, depending upon the
end-point. However, the terminal bronchioles of the neonatal rats
were more susceptible than those of young adults (Chang et al.,
1988). For example, the lower exposure altered ciliated cells and
non-ciliated bronchiolar (Clara) cells in the neonates but not the
young adults. Other indices were unaffected. Pulmonary function was
also altered in similarly exposed rats (Stevens et al., 1988) (see
section 5.2.2.3). Interpretation of the neonatal effects is
difficult. Assuming that rats prior to weaning are more resistant to
NO₂ (Stephens et al., 1978) (see below), effects observed after a
6-week exposure from birth may have resulted from the last 3 weeks of
exposure, as the first 3 weeks may constitute a more resistant period.
In contrast, effects observed in young adults probably reflect the
impact of the entire 6-week exposure.

In one of the more extensive studies, Kyono & Kawai (1982)
exposed rats at 1, 3, 12, and 21 months of age continuously for
1 month to 207 µg/m³, 865 µg/m³, 5260 µg/m³ or 16 500 µg/m³
(0.11, 0.46, 2.8 or 8.8 ppm) NO₂. Various morphometric parameters
were assessed, including arithmetic mean thickness of the air-blood
barrier and the volume density of various alveolar wall components.
Quantitative estimations deliberately excluded the site of main damage
(i.e., the peripheral alveolar wall was examined). Analysis of
individual results was complex, but depending upon the animal's age
and the specified end-point, exposure levels as low as 207 µg/m³
(0.11 ppm) changed specific morphometric parameters. There was a
trend towards a concentration-dependent increase in air-blood barrier
thickness in all age groups, with evidence of age-related differences
in response. At any concentration, the response of this end-point
decreased in rats from 1 to 12 months old, but increased again in
21-month-old animals. Type 1 and 2 cells showed various degrees of
response, depending on both age at onset of exposure and exposure
concentration. The response of each lung component did not always
show a simple concentration-dependent increase or decrease, but
suggested a multiphasic reaction pattern.

The above studies with rats may not have used the most
susceptible animal model, as demonstrated by Azoulay-Dupuis et al.
(1983), who exposed both rats and guinea-pigs aged 5 to > 60 days
old to 3760 (2.0 ppm) for 3 days. Rats at all ages and guinea-pigs
< 45 days old were not affected. The 45-day-old guinea-pigs showed
thickening of alveolar walls, alveolar oedema, and inflammation,
whereas animals older than 45 days showed similar, but more frequent,
alterations that seemed to increase with age. Adults also had focal
loss of cilia in bronchioli.

In general, it appears that neonates, prior to weaning, are
relatively resistant to NO₂, and that responsiveness then increases
(Stephens et al., 1978). Furthermore, the responsiveness of mature
animals appears to decline somewhat with age, until an increase in
responsiveness occurs at some point in senescence. However, the
morphological response to NO₂ in animals of different ages involves
similarities in the cell types affected and in the nature of the
damage incurred. Age-related differences occur in the extent of
damage and in the time required for repair, the latter taking longer
in older animals. The reasons for age differences in susceptibility
are not known, but may involve differences in doses to the target
cells and variable sensitivity of target cells during different growth
phases.

The database regarding the effects of levels of NO₂
< 9400 µg/m³ (5.0 ppm) on animals with pre-existing respiratory
disease is very limited and only includes animals with laboratory-
induced emphysema or infections. Lafuma et al. (1987) exposed both
normal and elastase-induced emphysematous hamsters (2 months old) to
3760 µg/m³ (2.0 ppm) NO₂ for 8 h/day, 5 days/week, for 8 weeks.
Morphometric analyses indicated that emphysematous lesions were
exacerbated by NO₂ (i.e., NO₂ increased pulmonary volume and
decreased internal alveolar surface area). The investigators
suggested that these results may imply a role for NO₂ in enhancing
pre-existing emphysema. Acute infectious (influenza) lung disease
enhanced the morphological effects of NO₂ in squirrel monkeys
exposed continuously to 1880 µg/m³ (1.0 ppm) NO₂ for 16 months
(Fenters et al., 1973).

d) Emphysema following nitrogen dioxide exposure

Numerous investigators have observed morphological lesions that
led them to the diagnosis of NO₂-induced emphysema. However, to
evaluate these reports independently, it is necessary to apply the
current definition of emphysema, especially because the definition
changed after several of the reports were published. Such an
evaluation is described in detail by the US EPA (1993), based upon the
most recent definition of emphysema from the report of the US National
Heart, Lung and Blood Institute (NHLBI), Division of Lung Diseases
Workshop (National Institutes of Health, 1985). According to this
document, in human lungs: "Emphysema is defined as a condition of the
lung characterized by abnormal, permanent enlargement of airspaces
distal to the terminal bronchiole, accompanied by destruction of their
walls, and without obvious fibrosis". Destruction in emphysema is
further defined as "non-uniformity in the pattern of respiratory
airspace enlargement so that the orderly appearance of the acinus and
its components is disturbed and may be lost". The report further
indicates: "Destruction...may be recognized by subgross examination of
an inflation-fixed lung slice...". However, emphysema in animal

models was defined differently. An animal model of emphysema is
defined as "an abnormal state of the lungs in which there is
enlargement of the airspaces distal to the terminal bronchiole.
Airspace enlargement should be determined qualitatively in appropriate
specimens and quantitatively by stereologic methods". Thus, in animal
models of emphysema, airspace wall destruction need not be present.
"Appropriate specimens" presumably refers to lungs fixed in the
inflated state. When reports of emphysema following NO₂ exposures
of animals are to be extrapolated to potential hazards for humans, the
definition of human emphysema, rather than that for emphysema in
experimental animals, should be used. The presence of airspace wall
destruction, critical to the definition of human emphysema, can only
be determined independently in published reports by careful review of
the authors' description of the lesions or by examining the
micrographs that the author selected for publication. Because
descriptions in some reports are inadequate for independent
evaluation, more evidence may exist for emphysema than is summarized
here. All reports reviewed are summarized in Table 36, but only those
showing emphysema of the type seen in human lungs are discussed in the
text that follows.

Haydon et al. (1967) reported emphysema in rabbits exposed
continuously (presumably 24 h/day) for 3 to 4 months to 15 000 or
22 600 µg/m³ (8.0 or 12.0 ppm) NO₂. They reported enlarged lungs
that failed to collapse when the thorax was opened. The lungs were
fixed in an expanded state via the trachea. In 100-µm thick sections
from formaldehyde-fixed dried lungs they reported "dilated" airspaces
with "distorted architecture." In those and other tissue
preparations, they reported that the airspaces appeared "grossly
enlarged and irregular, which appears to be due to disrupted alveoli
... and the absence of adjacent alveolar collapse." Thus, in
appropriately fixed lungs, they reported evidence of enlarged
airspaces with destructive changes in alveolar walls. Although no
stereology was performed, this appears to be emphysema of the type
seen in human lungs.

Freeman et al. (1972) exposed rats to 37 600 µg/m³ (20.0 ppm)
NO₂, which was reduced during the exposure to 28 200 µg/m³
(15.0 ppm) or to 18 800 µg/m³ (10.0 ppm), for varying periods up to
33 months. Following removal at necropsy, the lungs were fixed via the
trachea at 25 cm of fixative pressure. Morphometry of lung and
alveolar size was performed in a suitable, although unconventional,
manner. The morphometry indicated enlargement of alveoli and reduction
in alveolar surface area. The authors also both reported alveolar
destruction and illustrated alveolar destruction in their figures.
They correctly concluded that they had demonstrated emphysema in their
NO₂-exposed rats. However, it is not entirely clear whether both
experimental groups or only the group exposed to 28 200 µg/m³
(15.0 ppm) had emphysema.

Table 36. Effects of nitrogen dioxide (NO₂) on the development of emphysema^a

NO₂ concentration

µg/m³ ppm Exposure Species Emphysema^b Reference

188 with 2-h peaks to 1880 0.1 with Daily, 6 months Mouse ± Port et al. (1977)
peaks to 1.0

263 plus 2050 µg/m³ NO 0.14 16 h/day, 68 months Beagle dog - Hyde et al. (1978)
1200 plus 310 µg/m³ NO 0.64 +

940 0.5 6, 18 or 24 h/day, 1-12 months Mouse - Blair et al. (1969)

1500 0.8 51-813 days Rat - Haydon et al. (1965)
7520 4.0

1880 (with and without viral 1.0 16 months Squirrel ± Ehrlich & Fenters (1973)
challenge) monkey

3760 2.0 Continuous, 112-763 days Rat - Freeman et al. (1968c)

3760 2.0 8 h/day, 5 days/week Hamster - Lafuma et al. (1987)
for 8 weeks

9400 5.0 3 months Squirrel ± Ehrlich & Fenters (1973)
18 800 10.0 monkey

Table 36 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Emphysema^b Reference

9400 5.0 Up to 18 months Dog, - Wagner et al. (1965)
rabbit,
guinea-pig,
rat, hamster,
mouse

15 000 8.0 3-4 months (presumably Rabbit + Haydon et al. (1967)
22 560 12.0 24 h/day)

28 200 15.0 3-5 months Rat - Stephens et al. (1976)

28 200 15.0 Continuously for 35 days then Rat ± Port et al. (1977)
intermittently for at least
2 years

33 800 18.0 24 h/day for 1-6 days or Rat ± Freeman et al. (1968a)
4 weeks

37 600 reduced to either 20.0 reduced to Up to 33 months Rat + Freeman et al. (1972)
28 200 or 18 800 15.0 or 10.0

47 000 25.0 32-65 days Rat - Freeman & Haydon (1964)

56 400 30.0 22 h/day, 12 months Hamster - Kleinerman et al. (1985)

56 400 30.0 Continuous, up to 140 days Rat ± Glasgow et al. (1987)

56 400 30.0 Continuous, up to 8 weeks Rat - Blank et al. (1978)

Table 36 (Con't)

NO₂ concentration

µg/m³ ppm Exposure Species Emphysema^b Reference

56 400 to 65 800 30.0-35.0 23 h/day for 7 days Hamster - Lam et al. (1983)

65 800 35.0 6 h/day for 25 days Rat - Stavert et al. (1986)

75 200 40.0 6 or 8 weeks Mouse - Buckley & Loosli (1969)

94 000 to 169 200 for 50-90 reduced 2 h/day, 5 days/week, Hamster, ± Gross et al. (1968)
4 weeks, reduced to 56 400 to 30-50 12 months guinea-pig
to 94 000

84 600 to 103 400 45-55 22-23 h/day, 10 weeks Hamster - Kleinerman & Cowdrey
(1968)

^a Modified from US EPA (1993)
^b + = emphysema; - = no emphysema; ± = equivocal
Emphysema is defined according to the 1985 US National Heart, Lung, and Blood Institute Workshop criteria for human emphysema.
Although many of the papers reviewed (US EPA, 1993) reported finding emphysema, some of these studies were reported according
to previous, different criteria; some reports did not fully describe the methods used; and/or the results obtained were not in
sufficient detail to allow independent confirmation of the presence of emphysema. Thus, a "-" (i.e. no emphysema) should only be
interpreted as lack of proof of emphysema, because it is conceivable that if the study were repeated with current methods and
the current criteria applied, it might be judged to be positive.
Hyde et al. (1978) studied beagle dogs that had been exposed 16 h
daily for 68 months to either filtered air or to 1200 µg/m³
(0.64 ppm) NO₂ with 310 µg/m³ (0.25 ppm) NO or to 263 µg/m³
(0.14 ppm) NO₂ with 2050 µg/m³ (1.67 ppm) NO. The dogs then
breathed clean air during a 32- to 36-month post-exposure period. The
right lungs were fixed via the trachea at 30-cm fixative pressure
in a distended state. Semiautomated image analysis was used for
morphometry of air spaces. The dogs exposed to 1200 µg/m³ NO₂ with
310 µg/m³ NO had significantly larger lungs with enlarged air spaces
and evidence of destruction of alveolar walls. These effects were not
observed in dogs exposed to 270 µg/m³ NO₂ with 2050 µg/m³ NO,
implying a significant role of the NO₂ in the production of the
lesions. The lesions in dogs exposed to the higher NO₂ concentration
meet the criteria of the 1985 NHLBI workshop for emphysema of the type
seen in human lungs.

5.2.3 Genotoxicity, potential carcinogenic or co-carcinogenic effects

NO₂ forms nitrous and nitric acids in aqueous solutions, which
are in equilibrium with the nitrite (NO₂^-) and nitrate (NO₃^-) ions
that constitute the main metabolites of NO₂. Nitrous acid/NO₂^- is
mutagenic in vitro, causing deamination of bases in DNA. The
formation of N-nitroso compounds from secondary amines and amides is
another mechanism for indirect mutagenic activity (Zimmermann, 1977).

In vitro studies with NO₂ have demonstrated mutations in
bacteria (Salmonella strain TA100) (Isomura et al., 1984; Victorin &
Stahlberg, 1988) but not in a mammalian cell culture (Isomura et al.,
1984). Other experiments using cell cultures were positive concerning
chromatid-type chromosome abberations, sister chromatid exchanges
(SCE) and DNA single strand breaks (Tsuda et al., 1981; Shiraishi &
Bandow, 1985; Gorsdorf et al., 1990).

NO₂ did not induce recessive lethal mutations or somatic
mutations in Drosophila (Inoue et al., 1981; Victorin et al., 1990)
and was negative in in vivo studies with mice concerning chromosome
abberations in peripheral lymphocytes or spermatocytes (Gooch et al.,
1977) and micronuclei in bone marrow cells in mice (Victorin et al.,
1990).

Two studies have dealt with genotoxic effects in the relevant
target organ, i.e. the lung, and both were positive at high
concentrations. In the first one, Isomura et al. (1984) demonstrated
the induction of mutations and chromosome abberations in lung cells of
rats exposed to 27 ppm (50 000 µg/m³) for 3 h. In the other (Walles
et al., 1995), DNA single strand breaks were induced in lung cells of
mice exposed to 54 000 µg/m³ (30 ppm) for 16 h or 94 000 µg/m³
(50 ppm) for 5 h.

Several studies have evaluated the issue of carcinogenesis and
co-carcinogenesis, but results are often unclear or conflicting
(Table 37). However, there do not appear to be any published reports
on studies using classical carcinogenesis whole-animal bioassays. An
excellent critical review and discussion of some of the important
theoretical issues in interpreting these types of studies has been
published (Witschi, 1988). Although lung epithelial hyperplasia
(section 5.2.2.4) and enhancement of endogenous retrovirus expression
(Roy-Burman et al., 1982) have been thought by some to suggest
increased carcinogenic potential, such findings are not conclusive
(see US EPA, 1993).

Wagner et al. (1965) suggested that NO₂ may accelerate the
production of tumours in CAF₁/Jax mice (a strain that has
spontaneously high pulmonary tumour rates) after continuous exposure
to 9400 µg/m³ (5.0 ppm) NO₂. After 12 months of exposure, 7 out of
10 mice in the exposed group had tumours, compared to 4 of 10 in the
controls. No differences in tumour production were observed after 14
and 16 months of exposure. A statistical evaluation of the data was
not presented. The frequency and incidence of spontaneously
occurring pulmonary adenomas was increased in strain A/J mice (with
spontaneously high tumour rates) after exposure to 18 800 µg/m³
(10.0 ppm) NO₂ for 6 h/day, 5 days/week, for 6 months (Adkins et al.,
1986). These small, but statistically significant, increases were
only detectable when the control response from nine groups (n = 400)
were pooled. Exposure to 1880 and 9400 µg/m³ (1.0 and 5.0 ppm) NO₂
had no effect. In contrast, Richters & Damji (1990) found that an
intermittent exposure to 470 µg/m³ (0.25 ppm) NO₂ for up to 26 weeks
decreased the progression of a spontaneous T cell lymphoma in
AKR/ cum mice and increased survival rates. The investigators
attribute this effect to an NO₂-induced decrease in the proliferation
of T cell subpopulation (especially T-helper/inducer lymphocytes) that
produce growth factors for the lymphoma.

Whether NO₂ facilitates metastases has been the subject of
several experiments by Richters & Kuraitis (1981, 1983), Richters &
Richters (1983) and Richters et al. (1985). Mice were exposed to
several concentrations and durations of NO₂ and were injected
intravenously with a cultured-derived melanoma cell line (B16) after
exposure; subsequent tumours in the lung were counted. Although
some of the experiments showed an increased number of lung
tumours, statistical methods were inappropriate. Furthermore, the
experimental technique used in these studies probably did not evaluate
metastases formation, as the term is generally understood, but more
correctly, colonization of the lung by tumour cells.

Table 37. Effects of nitrogen dioxide (NO₂) on carcinogenesis or co-carcinogenesis^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

188-18 800 0.1-10.0 0.5-4 h Mouse Mice exposed to DMA had whole-body concentration- Iqbal et al.
related increase in DMN. (1981)

470 0.25 7 h/day, Mouse NO₂ slowed progression of spontaneous T cell Richters & Damji
5 days/week, lymphomas in AKR/cum mice, increased survival, and (1990)
up to 26 weeks decreased number of splenic CD4⁺ T cells.

752 0.4 7-8 h/day, Mouse Increased lung tumors and mortality in mice injected Richters &
940 0.5 5 days/week, with melanoma cells after NO₂ exposure. Kuraitis (1981,
1500 0.8 12 weeks 1983); Richters
et al. (1985)

940-1500 0.5-0.8 Continuous, Mouse Hyperplastic foci identical to that observed after Nakajima et
30 days exposure to known carcinogens. al. (1972)

1500 0.8 8 h/day, Mouse Enhanced retrovirus expression in two strains of Roy-Burman
5 days/week, mice. et al. (1982)
18 weeks

1880 1.0 6 h/day, Mouse No effect at 1880 or 9400 µg/m³. At 18 800 µg/m³, Adkins et al.
9400 5.0 5 days/week, spontaneous adenomas in strain A/J mice increased (1986)
18 800 10.0 6 weeks only when compared to pooled control group.

2000 1.1 Continuous, Rat DMA plus NO₂ did not produce tumors. Design and Benemansky
3010 1.6 lifetime statistical analyses not appropriate; exposure et al. (1981)
methods not described.

Table 37 (Con't)

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

9400-18 800 5.0-10.0 2 h/day, Mouse Mice given 4-nitroquinoline-1-oxide during NO₂ Ide & Otsu
5 days/week, exposure; NO₂ had no effect on tumor production. (1973)
50 weeks

18 800 10.0 2 h/day, Mouse Mice given 4-nitroquinoline-1-oxide and NO₂. Otsu & Ide
5 days/week, NO₂ decreased incidence of lung tumors. (1975)
50 weeks

28 200-94 000 15.0-50.0 1-4 h Mouse Mice gavaged with morpholine had concentration- Iqbal et al.
dependent increase in whole-body content of NMOR. (1980)

31 020-38 500 16.5-20.5 5-6 h/day, Mouse In vivo production of NMOR when 1 g/kg of morpholine Van Stee et
4 days; plus was administered each day prior to exposure. al. (1983)
3 h on 5th day

84 600 45.0 2 h Mouse Mice gavaged with morpholine had an in vivo increase Norkus et al.
in NMOR production. (1984)

199 000 106.0 0.5-4 h Rat Rats given morpholine in their diets or by gavage had Mirvish et al.
Mouse no NMOR detected in their bodies. (1981)
In mice morpholine, by gavage, yielded no significant
in vivo NMOR production.

^a Modified from US EPA (1993)
^b DMA = Dimethylamine; DMN = Dimethylnitrosamine; NMOR = N-nitrosomorpholine
Ide & Otsu (1973) did not find that chronic exposure to high
concentrations of NO₂ (somewhere between 9400 and 18 800 µg/m³,
5.0 and 10.0 ppm) enhanced tumour production in conventional mice
receiving five weekly injections of 0.25 mg 4-nitroquinoline-1-oxide
(a lung-tumour-specific carcinogen). Benemansky et al. (1981) used a
known carcinogen, nitrosodimethylamine or its precursor dimethylamine
(DMA) to test for interactions with a chronic exposure to NO₂.
However, appropriate statistical techniques and control groups were
not employed and the methods of exposure and monitoring of NO₂ were
not reported, thus precluding accurate evaluation. In another study,
rats were injected with N-bis (2-hydroxy-propyl)nitrosamine (BHPN)
and continuously exposed to 75, 750 or 7500 µg/m³ (0.04, 0.4 or
4.0 ppm) NO₂ for 17 months. Although the data indicated five times
as many lung adenomas or adenocarcinomas in the rats injected with
BHPN and exposed to 7500 µg/m³ NO₂ (5/40 compared to 1/10), the
results failed to achieve statistical significance (Ichinose et al.,
1991).

Because of evidence that NO₂ could produce NO₂^- and NO₃^- in
the blood and the fact that NO₂^- is known to react with amines to
produce animal carcinogens (nitrosamines), the possibility that NO₂
could produce cancer via nitrosamine formation has been investigated.
Iqbal et al. (1980) was the first to demonstrate a linear time-
dependent and concentration-dependent relationship between the amount
of N-nitrosomorpholine (NMOR) (an animal carcinogen) found in
whole-mouse homogenates after the mice were gavaged with 2 mg of
morpholine (an exogenous amine that is rapidly nitrosated) and
exposure to 28 200 to 94 000 µg/m³ (15.0 to 50.0 ppm) NO₂ for
between 1 and 4 h. In a follow-up study, Iqbal et al. (1981) used
DMA, an amine that is slowly nitrosated to dimethylnitrosamine (DMN).
They reported a concentration-related increase in biosynthesis of DMN
at NO₂ concentrations as low as 188 µg/m³ (0.1 ppm); however, the
rate was significantly greater at concentrations above 18 800 µg/m³
(10.0 ppm) NO₂. Increased length of exposure also increased DMN
formation between 0.5 and 2 h, but synthesis of DMN was less after 3
or 4 h of exposure than after 0.5 h.

Mirvish et al. (1981) conducted analogous research and concluded
that the results of Iqbal et al. (1980) were technically flawed, but
found that in vivo exposure to NO₂ could produce a nitrosating
agent (NSA) that would nitrosate morpholine only when morpholine was
added in vitro. Further experiments showed that the NSA was
localized in the skin (Mirvish et al., 1983) and that mouse skin
cholesterol was a likely NSA (Mirvish et al., 1986). It has also been
reported that only very lipid-soluble amines, which can penetrate the
skin, would be available to the NSA. Compounds such as morpholine,
which are not lipid-soluble, could only react with NO₂ when it was
painted directly on the skin (Mirvish et al., 1988). Iqbal (1984),
responding to the Mirvish et al. (1981) criticisms, verified their
earlier studies (Iqbal et al., 1980). In vivo nitrosation was also
demonstrated by Norkus et al. (1984) after morpholine administration
and a 2-h exposure to 84 600 µg/m³ (45 ppm) NO₂.

Another study (Van Stee et al., 1983) reported that mice gavaged
with 1 g of morpholine/kg body weight per day and then exposed (5-6 h
daily for 5 days) to 31 000 to 38 500 µg/m³ (16.5 to 20.5 ppm) NO₂
revealed that NMOR could be produced in vivo. The single site
containing the greatest amount of NMOR was the gastrointestinal tract.

Shoaf et al. (1989) studied the uptake and nitrosation of primary
amines by NO₂ in isolated ventilated rat lungs. The rate of
nitrosation was very low because the nitrosation of primary amines is
a general acid/base catalysed reaction that would be at a minimum at
pH 7. The authors could not replicate the previous nitrosation
studies. At a maximum, only 0.0001% of an amine would be nitrosated.
Such a rate is at or below the detection limit for nitrosamine. The
studies reporting nitrosation may be seriously in error. Nitrosation
may be a very minimal reaction and of little consequence.

Victorin (1994) reviewed the genotoxicity of nitrogen oxides and
concluded that there is no clear evidence of a carcinogenic potential
of NO₂. Victorin (1994) also directed attention to the possibility
that NO_x compounds in photochemical smog may contribute secondarily
to formation of other genotoxic compounds. For example, it was noted
that strongly mutagenic nitro-PAH compounds are easily formed and
mutagenic reaction products may be formed from alkenes through
photochemical reactions.

Overall, the above critical evaluation indicates that there is no
evidence establishing that tumours can be directly induced by NO₂
exposure alone. Also, the available evidence for NO₂ promoting or
enhancing the production or growth of tumours caused by other agents
is quite limited and conflicting. It must therefore be concluded that
the evidence for carcinogenicity of nitrogen oxides is at present
inadequate, but the issue should be addressed by further research.

5.2.4 Extrapulmonary effects

Exposure to NO₂ produces a wide array of health effects beyond
the confines of the lung. Thus, NO₂ and/or some of its reactive
products penetrate the lung or nasal epithelial and endothelial layers
to enter the blood and produce alterations in blood and various other
organs (Shoaf et al., 1989). Effects on the systemic immune system
are discussed under section 5.2.2.1. Information regarding the
effects of NO₂ on animal behaviour and brain enzymes is limited to a
few studies that cannot be readily interpreted in terms of human risks
and will not be discussed. The summary of other systemic effects is
quite brief because the database suggests that effects on the
respiratory tract are of more concern. A more detailed discussion of
extrapulmonary responses can be found in US EPA (1993).

Results of research on the number of erythrocytes and leukocytes,
haemoglobin concentration, and contents of erythrocyte membranes are
inconsistent. In the only such study conducted below 9400 µg/m³
(5.0 ppm) NO₂, Nakajima & Kusumoto (1968) found that the amount of
methaemoglobin was not increased when mice were exposed to 1500 µg/m³
(0.8 ppm) NO₂ for 5 days. This topic was of interest because some
(but not all) in vitro studies and high concentration in vivo NO₂
studies found methaemoglobin effects (US EPA, 1993).

Several studies have examined hepatic function either directly or
indirectly after NO₂ exposure. Changes in serum chemistry (e.g.,
plasma cholinesterase, Drozdz et al., 1976; Menzel et al., 1977)
suggest that NO₂ exposure may affect the liver. Xenobiotic
metabolism appears to be affected by NO₂. A 3-h exposure to NO₂
concentrations as low as 470 µg/m³ (0.25 ppm) increased
pentobarbital-induced sleeping times in female, but not male, mice
(Miller et al., 1980; Graham et al., 1982). Higher exposures
(9400 µg/m³, 5.0 ppm; 3 h) did not affect the level of hepatic
cytochrome P-450 or the activities of several mixed-function oxidases
in mice (Graham et al., 1982). Other authors found mixed effects
(i.e. increase or decrease depending on exposures) on liver cytochrome
P-450 levels in rats (Takano & Miyazaki, 1984; Takahashi et al.,
1986). Significant decreases in cytochrome P-450 from rat liver
microsomes were also found after 7 days of exposure to 752 or
7520 µg/m³ (0.4 or 4.0 ppm) NO₂, but not after exposure to
2260 µg/m³ (1.2 ppm) NO₂ (Mochitate et al., 1984). NADPH-cytochrome
C reductase was reduced with 5 days of exposure to 7520 and
18 800 µg/m³ (4.0 and 10.0 ppm) NO₂. Drozdz et al. (1976) found
decreased total liver protein and sialic acid, but increased protein-
bound hexoses in guinea-pigs exposed to 2000 µg/m³ (1.05 ppm) NO₂,
8 h/day for 180 days. Liver alanine and aspartate aminotransferase
activity was increased in the mitochondrial fraction but decreased in
the cytoplasmic fraction of the liver. Electron micrographs of the
liver showed intracellular oedema and inflammatory and parenchymal
degenerative changes.

Takahashi et al. (1986) found that continuous exposure to 2260
and 7520 µg/m³ (1.2 and 4.0 ppm) NO₂ increased the amount of
cytochrome P-450 and cytochrome b₅ in the kidney after 8 weeks of
exposure. Continued exposure for 12 weeks resulted in less
substantial increases in the amount and activity of the microsomal
electron-transport enzymes. This is in contrast to the decreased
activity in the liver.

Increases in urinary protein and specific gravity of the urine
were reported by Sherwin & Layfield (1974) in guinea-pigs exposed
continuously to 940 µg/m³ (0.5 ppm) NO₂ for 14 days. Proteinuria
was detected in another group of animals when the exposure was reduced
to 752 µg/m³ (0.4 ppm) NO₂ for 4 h/day. Disc electrophoresis of the

urinary proteins demonstrated the presence of albumin and alpha-,
beta-, and gamma-globulins. The presence of high molecular weight
proteins in urine is characteristic of the nephrotic syndrome.
Differences in water consumption or in the histology of the kidney
were not found.

Few studies have examined the effects of NO₂ on reproduction and
development or the heritable mutagenic potential of NO₂. Exposure to
1800 µg/m³ (1.0 ppm) NO₂ for 7 h/day (5 days/week for 21 days)
resulted in no alterations in spermatogenesis, germinal cells or
interstitial cells of the testes of six rats (Kripke & Sherwin, 1984).
Similarly, breeding studies by Shalamberidze & Tsereteli (1971) found
that long-term NO₂ exposure had no effect on fertility. However,
there was a statistically significant decrease in litter size and
neonatal weight when male and female rats exposed to 2440 µg/m³
(1.3 ppm) NO₂, 12 h/day for 3 months were bred. In utero death due
to NO₂ exposure resulted in smaller litter sizes, but no direct
teratogenic effects were observed in the offspring. In fact, after
several weeks, NO₂-exposed litters approached weights similar to
those of controls.

Inhalation exposure of pregnant Wistar rats to NO₂
concentrations of 1000 and 10 000 µg/m³ for 6 h/day throughout
gestation (21 days) was found to have maternal toxic effects and to
induce developmental disturbances in the progeny (Tabacova et al.,
1984; Balabaeva & Tabacova, 1985; Tabacova & Balabaeva, 1988). The
maternal weight gain during gestation was significantly reduced at
10 000 µg/m³ (5.3 ppm). Pathomorphological changes, manifested at
the higher exposure level, were found in maternal organs, e.g.,
desquamative bronchitis and bronchiolitis in the lung, mild
parenchymal dystrophy and reduction of glycogen in the liver, and
blood stasis and inflammatory reaction in the placenta. At gross
examination, the placentae of the dams exposed to 10 000 µg/m³ were
smaller in size than those of control rats. A marked increase of
lipid peroxides was found in maternal lungs and particularly in the
placenta at both exposure levels by the end of gestation (Balabaeva &
Tabacova, 1985). Disturbances in the prenatal development of the
progeny were registered, such as two- to four-fold increase in late
post-implantation lethality at 1000 and 10 000 µg/m³ (0.5 and
5.3 ppm), respectively, as well as reduced fetal weight at term and
stunted growth at 10 000 µg/m³ (Tabacova et al., 1984). These
effects were significantly related to the content of lipid peroxides
in the placenta, which was suggestive of a pathogenetic role of
placental damage (Tabacova & Balabaeva, 1988). Teratogenic effects
were not observed, but dose-dependent morphological signs of
embryotoxicity and retarded intrauterine development, such as
generalized oedema, subcutaneous haematoma, retarded ossification
and skeletal aberrations, were found at both exposure levels.

In the only study that has examined postnatal development, a
significant delay in eye opening and incisor eruption was observed in
the progeny of maternally exposed Wistar rats (Tabacova et al., 1985).
The dams were exposed to 50, 100, 1000 or 10 000 µg/m³ (0.03, 0.05,
0.53 or 5.3 ppm) NO₂ for 6 h/day, 7 days/week throughout gestation,
and the offspring were studied for 2-month post-exposure. Significant
deficits in the onset of normal neuromotor development and reduced
open field activity were detected in the offspring of dams exposed to
1000 and 10 000 µg/m³ NO₂.

5.3 Effects of mixtures containing nitrogen dioxide

Humans are exposed to pollutant mixtures in the ambient air, and,
because pollutant interactions do occur, it is difficult to predict
the effects of NO₂ in a mixture based upon the effects of NO₂ alone.
Epidemiological studies (chapter 7), by their very nature, evaluate
ambient air mixtures, but the presence of confounding variables makes
it difficult to demonstrate a cause-effect relationship. In contrast,
controlled animal and human clinical studies can often demonstrate the
cause of a response, but are typically limited to binary or tertiary
mixtures, which do not truly reflect ambient air exposures. When
combinations of air pollutants are studied, there are a number of
possible outcomes on human or animal responses. The result of
exposure to two or more pollutants may be simply the sum of the
responses to individual pollutants; this is referred to as additive.
Another possibility is that the resultant response may be greater than
the sum of the individual responses, suggesting some type of
interaction or augmentation of the response; this is referred to as
synergism. Finally, responses may be less than additive; this is
often called antagonism. Generally, such human clinical studies, which
focused on pulmonary function, have not found that acute exposures to
NO₂ has any impact on the response to other co-occurring pollutants
(e.g., O₃) or that additive effects occur. Animal toxicological
studies, with a wider array of designs and end-points, have shown an
array of interactions, including no interaction, additivity and
synergism. Because no clear understanding of NO₂ interactions has
yet emerged from this database, only a brief overview is provided
here. A more substantive review can be found in US EPA (1993). Other
animal studies sought to understand the effects of ambient air
mixtures containing NO₂ or vehicular combustion exhausts containing
NO_x. Generally these studies provide useful information on the
mixtures, but lack NO₂-only groups, making it impossible to discern
the influence of NO₂. Therefore, this class of research is not
described here, but is reviewed elsewhere (US EPA, 1993).

The vast majority of interaction studies have involved NO₂ and
O₃. For lung morphology end-points, NO₂ had no interaction with O₃
(Freeman et al., 1974) or with sulfur dioxide (SO₂) (Azouley et al.,
1980) after a subchronic exposure. Some biochemical responses to NO₂

plus O₃ display no positive interaction or synergism. For example,
Mustafa et al. (1984) found synergism for some end-points (e.g.,
increased activities of O₂ consumption and antioxidant enzymes),
but no interaction for others (e.g., DNA or protein content)
in rats exposed for 7 days. Ichinose & Sagai (1989) observed a
species-dependence in regard to the interaction of O₃ (752 µg/m³,
0.4 ppm) and NO₂ (752 µg/m³, 0.4 ppm) after 2 weeks of exposure.
Guinea-pigs, but not rats, had a synergistic increase in lung lipid
peroxides. Rats, but not guinea-pigs, had synergistic increases in
antioxidant factors (e.g., non-protein thiols, vitamin C, glucose-6-
phosphate dehydrogenase, GSH peroxidase). Schlesinger et al. (1990)
observed a synergistic increase in prostaglandin E₂ and F_{2 alpha} in the
lung lavage of acutely exposed rabbits; the response appeared to have
been driven by O₃. However, with 7 or 14 days of repeated 2-h
exposures, only prostaglandin E₂ was decreased and appeared to have
been driven by NO₂; there was no synergism (Schlesinger et al.,
1991).

The infectivity model has been frequently used to study NO₂-O₃
mixtures. In this model, mice are exposed to O₃ and NO₂ alone or in
mixtures for various durations. The mice are then challenged with an
aerosol of viable bacteria. An increase in mortality indicates
detrimental effects on lung host-defence mechanisms. Ehrlich et al.
(1977) found additivity after acute exposure to mixtures of NO₂ and
O₃. They reported synergism after subchronic exposures. Exposure
scenarios involving NO₂ and O₃ have also been performed using a
continuous baseline exposure to one concentration or mixture, with
superimposed short-term peaks to a higher level. This body of work
(Ehrlich et al., 1979; Gardner, 1980; Gardner et al., 1982; Graham et
al., 1987) shows that differences in the pattern and concentrations of
the exposure are responsible for the increased susceptibility to
pulmonary infection, without indicating clearly the mechanism
controlling the interaction.

Some aerosols may potentiate response to NO₂ by producing local
changes in the lungs that enhance the toxic action of co-inhaled NO₂.
The impacts of NO₂ and H₂SO₄ on lung host defences have been
examined by Schlesinger & Gearhart (1987) and Schlesinger (1987a). In
the former study, rabbits were exposed for 2 h/day for 14 days to
either 564 µg/m³ (0.3 ppm) or 1880 µg/m³ (1.0 ppm) NO₂, or
500 µg/m³ H₂SO₄ alone, or to mixtures of the low and high NO₂
concentrations with H₂SO₄. Exposure to either concentration of NO₂
accelerated alveolar clearance, whereas H₂SO₄ alone retarded
clearance. Exposure to either concentration of NO₂ with H₂SO₄
resulted in retardation of clearance in a similar manner to that seen
with H₂SO₄ alone.

Schlesinger (1987a) used a similar exposure design, but different
end-points. Exposure to 1800 µg/m³ (1.0 ppm) NO₂ with acid resulted
in an increase in the numbers of PMNs in lavage fluid at all time
points (not seen with either pollutant alone), and an increase in
phagocytic capacity of AMs after two or six exposures. In contrast,
exposure to 564 µg/m³ (0.3 ppm) NO₂ with acid resulted in depressed
phagocytic capacity and mobility. The NO₂/H₂SO₄ mixture was
generally either additive or synergistic, depending on the specific
cellular end-point being examined.

Last et al. (1983) and Last & Warren (1987) found that exposure
to high levels of NO₂ (< 9400 µg/m³, 5.0 ppm) with very high
concentrations of H₂SO₄ (1 mg/m³) caused a synergistic increase in
collagen synthesis rate and protein content of the lavage fluid of
rats.

Dogs were exposed for 68 months (16 h/day) to raw or
photochemically reactive vehicle exhaust which included mixtures of
NO_x œ one with a high NO₂ level and a low NO level (1200 µg/m³,
0.64 ppm, NO₂; 310 µg/m³, 0.25 ppm, NO), and one with a low NO₂
level and a high NO level (270 µg/m³, 0.14 ppm, NO₂; 2050 µg/m³,
1.67 ppm, NO) (Stara et al., 1980). Following the end of exposure,
the animals were maintained for about 3 years in normal indoor air.
Numerous pulmonary function, haematological and histological
end-points were examined at various times during and after exposure.
The lack of an NO₂-only or NO-only group precludes determination of
the nature of the interaction. Even so, the main findings are of
interest. Pulmonary function changes appeared to progress after
exposure ceased. Dogs in the high NO₂ group had morphological
changes considered to be analogous to human centrilobular emphysema
(see section 2.2.2.4). Because these morphological measurements were
made after a 2.5- to 3-year holding period in clean air, it cannot be
determined with certainty whether these disease processes abated or
progressed during this time. This study suggests progression of
damage after exposure ends.

5.4 Effects of other nitrogen oxide compounds

5.4.1 Nitric oxide

The toxicological database for NO is small, relative to NO₂. It
is often difficult to obtain pure NO in air without some contamination
with NO₂. An excellent review on the effects of NO on animals and
humans has been prepared by Gustafsson (1993) for the Swedish
Environmental Protection Agency. The following sections are based on
the information in this review.

5.4.1.1 Endogenous formation of NO

Endogenous NO synthesis occurs by NO formation from physiological
substrate (the amino acid L-arginine) in cells of many of the organ
systems, such as nerve tissue, blood vessels and the immune system.
NO has been found to be produced by at least three different
oxygen-utilizing NO synthases, for purposes such as signalling in the
nervous system, mediating vasodilation in both systemic and pulmonary
circulation, and mediating cytotoxicity and host defence reactions in
the immune system (Garthwaite, 1991; Barinaga, 1991; Moncada et al.,
1991; McCall & Valance, 1992; Snyder & Bredt, 1992; Moncada, 1992).
The impact of these findings for an understanding of the toxicological
effects of NO is still difficult to assess.

The actions of endogenous NO can be divided into two main groups.
The first group involves low concentrations of NO (nano- to picomolar)
formed by constitutive enzymes in nerve and endothelial cells. Nitric
oxide has local discrete actions exerted via activation of an enzyme,
guanylate cyclase, in the target cell (Ignarro, 1989). The second
group involves high concentrations of NO (micro- to nanomolar) formed
by enzymes that can increase in amount through the induction of these
enzymes upon exposure to bacterial toxins or to growth-regulating
factors (cytokinins). The inducible NO formation occurs especially in
macrophages and neutrophil leukocytes and is important for the killing
of bacteria and parasites, and possibly also for cytostasis in
antitumour reactions (Hibbs et al., 1988; Ignarro, 1989; Moncada et
al., 1991; Moncada, 1992).

For effects of inhaled NO it is important to consider that
endogenous NO regulates pulmonary vascular resistance; it is found in
small amounts in exhaled air and has been suggested to be necessary
for normal oxygenation of the blood (Persson et al., 1990; Gustafsson
et al., 1991).

5.4.1.2 Absorption of NO

Yoshida et al. (1981) found that < 10% of the NO "inhaled" by
isolated perfused lungs of rabbits was absorbed. In normally
breathing humans, 85 to 92% of NO was absorbed at concentrations
ranging from 400 to 6100 µg/m³ (0.33 to 5.0 ppm) (Wagner, 1970;
Yoshida & Kasama, 1987); values for NO₂ were 81 to 90% (Wagner,
1970). Absorption of NO with exercise was 91 to 93% in humans
(Wagner, 1970). Yoshida et al. (1980) found the percentage of
absorption of NO in rats acutely exposed to 169 300 µg/m³ (138 ppm),
331 300 µg/m³ (270 ppm) and 1 079 800 µg/m³ (880 ppm) to be 90%, 60%
and 20%, respectively. The progressive decrease in absorption was
ascribed to an exposure-induced decrease in ventilation. In dogs
exposed to vehicle exhaust mixtures, 73% of the constituent NO was
removed by the nasopharyngeal region; this compared to 90% removal for

NO₂ (Vaughan et al., 1969). Thus, respiratory tract absorption of NO
has some similarities to that for NO₂, in spite of solubility
differences. The lower solubility of NO may, however, result in
greater amounts reaching the pulmonary region, where it may then
diffuse into blood and react with haemoglobin (Yoshida & Kasama,
1987). In vivo exposures seem to indicate that NO has a faster rate
of diffusion through tissue than NO₂ (Chiodi & Mohler, 1985).

5.4.1.3 Effects of NO on pulmonary function, morphology and host lung
defence function

No change in respiratory function was found in guinea-pigs
exposed to NO at 19 600 µg/m³ (16 ppm) or 61 300 µg/m³ (50 ppm) for
4 h (Murphy et al., 1964). Increased airway responsiveness to
acetylcholine was observed in guinea-pigs exposed to 6130 µg/m³
(5 ppm) NO for 30 min, twice a week for 7 weeks. In sheep,
significant reversal of vasoconstriction to an infused thromboxane
analogue was seen with acute exposure to 6130 µg/m³ NO (Fratacci et
al., 1991). At the same exposure level, hypoxic vasoconstriction was
significantly diminished and was nearly abolished at 49 000 µg/m³
(40 ppm) NO in inhaled air (Frostell et al., 1991).

Reversal of methacholine-induced bronchoconstriction by NO has
been reported in guinea-pigs at 6130 µg/m³ (5 ppm) (Dupuy et
al., 1992), while in rabbits full reversal of methacholine
bronchoconstriction was seen at 98 100 µg/m³ (80 ppm) (Högman et
al., 1993). Relaxation of bronchial smooth muscle can be exerted
in vitro by mechanisms dependent on an intact airway epithelium. An
endogenous muscle-relaxing factor released by the epithelium has been
suggested, but it is not clear whether it is endogenous NO (Barnes,
1993).

The few studies that have examined histological response to
non-lethal levels of NO are outlined in Table 38. With chronic
exposure, the morphological changes seen are similar to those with
NO₂ (see section 5.2.2.4 on morphological effects of NO₂), except
that NO levels needed to produce them are higher. Additionally, Hugod
(1979) noted that the absence of NO-induced alterations in the
alveolar epithelium suggested that the observed responses occurred
after absorption of NO; that is, they were not caused by direct action
of deposited NO. Perhaps higher exposure concentrations of NO are
needed for direct toxic action (e.g., results of Holt et al., 1979).
Some of the effects seen by Oda et al. (1976) with 12 270 µg/m³
(10.0 ppm) NO may have been due to the presence of 1880 to 2820 µg/m³
(1.0 to 1.5 ppm) NO₂ in the exposure atmosphere.

It is important to note that in all existing studies of NO
toxicity in the lungs, histological evaluation of the lungs was
rudimentary and no quantitative measurements were carried out to test
for airspace enlargement or destruction.

Table 38. Effects of nitric oxide (NO) on respiratory tract morphology^a

NO₂ Concentration

µg/m³ ppm Exposure Species Effects^b Reference

2460 2 Continuous, Rat Slight emphysema-like alterations of alveoli. Azoulay et
(NO₂ = 0.08 ppm)^b 6 weeks al. (1977)

2950 2.4 Continuous, Mouse No difference from control. Oda et al.
(NO₂ = 0.01-0.04 ppm)^b for lifetime (1980b)
(23-29 months)

6150 5 Continuous, Rabbit Oedema; thickening of alveolo-capillary Hugod (1979)
(NO₂ = < 0.1 ppm)^b 14 days membrane due to fluid in interstitial space;
fluid-filled vacuoles seen in arteriolar
endothelial cells and at junctions of
endothelial cells; no changes in alveolar
epithelium; no inflammation.

12 300 10 2 h/day, 5 days Mouse Enlarged air spaces in lung periphery; Holt et al.
per week, up to paraseptal emphysema; some haemorrhage; (1979)
30 weeks some congestion in alveolar septa; increased
concentration of goblet cells in bronchi.

12 300 10 Continuous, Mouse Bronchiolar epithelial hyperplasia; hyperaemia; Oda et al.
(NO₂ = 1-1.5 ppm)^b 6.5 months congestion; enlargement of alveolar septum; (1976)
increase in ratio of lung to body weight.

^a Modified from US EPA (1993)
^b This represents reported nitrogen dioxide (NO₂) levels measured during exposure
A recent study (Mercer et al., 1995) suggests that NO may be more
potent than NO₂ in introducing certain changes in lung morphology.
More specifically, male rats were exposed to either NO or NO₂ at
0.5 ppm with twice daily 1-h spikes of 1.5 ppm for 9 weeks. The
number of pores of Kohn and detached alveolar septa were evaluated by
electron microscopy, using stereological procedures for the study of
lung structure that involved morphometric analyses of electron
micrographs. The average number of pores per lung for the NO group
exceeded by approx. 2.5 times the mean number for the NO₂ groups,
which was more than 10 times that for controls. Analogously, the
average number of detached septa per lung was significantly higher for
the NO group (X = 117) than the NO₂ group (X = 20) or the controls
(X = 4). There was also a statistically significant 30% reduction in
interstitial cells in the NO group, but no significant differences in
the other parenchymal cell types between the controls and the NO- or
NO₂-exposed groups. Lastly, the thickness of the interstitial space
was reduced for the NO group (X = 0.24 µm versus 0.32 µm for controls)
but not for the NO₂ group (X = 0.29 µm), and epithelial cell
thickness did not differ between the groups.

The effects of NO on host defence function of the lungs has been
examined in two studies. Holt et al. (1979) found immunological
alterations in mice exposed to 12 270 µg/m³ (10 ppm) NO for 2 h/day
(5 days/week for 30 weeks). However, interpretation is complicated by
the duration dependence of some of the responses (e.g., an enhancement
of the humoral immune response to SRBCs was seen at 10 weeks, but this
was not evident at the end of the exposure series). The effects of NO
on bacterial defences were examined by Azoulay et al. (1981). Male
and female mice were exposed continuously to 3760 µg/m³ (2.0 ppm) NO
for 6 h to 4 weeks to assess the effect on resistance to infection
induced by a bacterial aerosol administered after each NO exposure.
There were no statistically significant effects for either sex at any
of the time points studied.

5.4.1.4 Metabolic effects

Mice exposed to NO concentrations of 12 300 to 25 800 µg/m³
(10 to 21 ppm) for 3 h daily for 7 days showed no change in the
levels of reduced glutathione in their lungs (Watanabe et al., 1980).
In vitro data indicate that NO stimulates guanylate cyclase and
therefore leads to smooth muscle relaxation and vasodilation and
functional effects on the nervous system (Katsuki et al., 1977;
Ignarro, 1989; Garthwaite, 1991; Moncada et al., 1991). These effects
are probably responsible for vasodilation in the pulmonary circulation
and an acute bronchodilator effect of inhaled NO. However, it is
unclear whether other effects might be exerted from ambient NO via
this pathway. Due to the rapid inactivation of NO in haemoglobin,
internal organs other than the lungs are unlikely to be affected
directly by cyclic GMP-mediated vasodilator influence from ambient
concentrations of NO.

Methaemoglobin formation, via the formation of nitrosylhaemoglobin
(Oda et al., 1975, 1979, 1980a,b; Case et al., 1979; Nakajima et al.,
1980) and subsequent oxidation with oxygen, is well known (Kon et al.,
1977; Chiodi & Mohler, 1985). During NO exposure of mice to 24 500 to
98 100 µg/m³ (20-80 ppm), the levels of methaemoglobin were found
to increase exponentially with the NO concentration (Oda et al.,
1980b). After the cessation of NO exposure, methaemoglobin decreased
rapidly, with a half-time of only a few minutes. In humans the ability
to reduce methaemoglobin varies genetically and is lower in infants.
Of the NO reaction products with haemoglobin, methaemoglobin attains
higher levels than nitrosylhaemoglobin (Maeda et al., 1987). Exposure
of mice to 2940 µg/m³ (2.4 ppm) NO for 23-29 months resulted in
nitrosylhaemoglobin levels at 0.01%, while the maximal methaemoglobin
level was 0.3% (Oda et al., 1980b). At 12 300 µg/m³ for 6.5
months the nitrosylhaemoglobin level was 0.13% and the level of
methaemoglobin was 0.2% (Oda et al., 1976). Rats exposed to
2450 µg/m³ (2 ppm) continuously for six weeks showed no detectable
methaemoglobin (Azoulay et al., 1977).

5.4.1.5 Haematological changes

Mice exposed to NO at 11 070 µg/m³ (9 ppm) for 16 h had
decreased iron transferrin (Case et al., 1979), and when exposed to
12 300 µg/m³ (10 ppm) for 6.5 months had increased leukocyte count
and proportion of polymorphonuclear cells (Oda et al., 1976). Red
blood cell morphology, spleen weight and bilirubin were also affected.
A slight increase in haemolysis was seen in mice exposed to
2940 µg/m³ (2.4 ppm) of NO (Oda et al., 1980a).

5.4.1.6 Biochemical mechanisms for nitric oxide effects: reaction
with iron and effects on enzymes and nucleic acids

NO has an affinity for haem-bound iron which is two times higher
than that of carbon monoxide. This affinity leads to the formation of
methaemoglobin and the stimulation of guanylate cyclase. Furthermore,
NO reacts with thiol-associated iron in enzymes and eventually
displaces the iron. This is a possible mechanism for the cytotoxic
effects of NO (Hibbs et al., 1988; Weinberg, 1992). In vitro, the
NO donor sodium nitroprusside has been shown to mobilize iron from
ferritin (Reif & Simmons, 1990). NO might possibly modulate
arachidonic acid metabolism via interference with iron (Kanner et al.,
1991a,b).

NO inhibits aconitase, an enzyme in the Krebs cycle, and also
complex 1 and 2 of the respiratory chain (Hibbs et al., 1988; Persson
et al., 1990; Stadler et al., 1991). Permanent modification of
haemoglobin has been found; possibly via deamination (Moriguchi
et al., 1992). NO can also deaminate DNA, evoke DNA chain breaks,

and inhibit DNA polymerase and ribonucleotide reductase (Wink et al.,
1991; Lepoivre et al., 1991; Kwon et al., 1991; Nguya et al., 1992).
NO might be antimitogenic and inhibit T cell proliferation in rat
spleen cells (Fu & Blankenhorn, 1992), and NO donors inhibit DNA
synthesis, cell proliferation, and mitogenesis in vascular tissue
(Garg & Hassid, 1989; Nakaki et al., 1990). ADP (adenosine
diphosphate) ribosylation is stimulated by NO-generating agents
(Nakaki et al., 1990).

Substantial in vitro evidence has recently been published
describing other effects of NO in tissues. These include: inhibition
of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) via ADP
ribosylation (Alheid et al., 1987; Dimmeler et al., 1992); macrophage
mediated-nitric oxide dependent mechanisms which include inhibition of
the electron transport chain (Nathan, 1992); inhibition of DNA
synthesis (Hibbs et al., 1988); inhibition of protein synthesis
(Curran et al., 1991) and decrease in cytosolic free calcium by a
cGMP-independent mechanism (Garg & Hassid, 1991).

5.4.2 Nitric acid

There have been only a few toxicological studies of HNO₃, which
exists in ambient air generally as a highly water-soluble vapour. A
few investigators have examined the histological response to instilled
HNO₃ (usually 1%), a procedure used in developing models of
bronchiolitis obliterans in various animals, namely dogs, rabbits and
rats (Totten & Moran, 1961; Greenberg et al., 1971; Gardiner &
Schanker, 1976; Mink et al., 1984). However, the relevance of such
instillation studies is questionable, except to provide information
for the design of inhalation studies.

Only two studies have been designed specifically to examine the
pulmonary response to pure HNO₃ vapour. Abraham et al. (1982)
exposed both normal sheep and allergic sheep (i.e., having airway
responses similar to those occurring in humans with allergic airway
disease) for 4 h to 4120 µg/m³ (1.6 ppm) HNO₃ vapour. The exposure,
using a "head-only" chamber, decreased specific pulmonary flow
resistance in both groups of sheep; this indicated the absence of any
bronchoconstriction. Allergic, but not normal, sheep showed increased
airway reactivity to carbachol, both immediately and 24 h after HNO₃
exposure. In another study, rats exposed for 4 h to 1000 µg/m³
(0.38 ppm) HNO₃ vapour or for 4 h/day for 4 days to 250 µg/m³
(0.1 ppm) HNO₃ showed a decrease in stimulated or unstimulated
respiratory burst activity of alveolar macrophages (AMs) obtained by
lavage, as well as an increase in elastase inhibitory capacity of BAL
(Nadziejko et al., 1992).

5.4.3 Nitrates

Only one inhalation study conducted at levels < 1 mg/m³
NO₃^- has been reported. Busch et al. (1986) exposed rats and
guinea-pigs with either normal lungs or elastase-induced emphysema to
ammonium nitrate aerosols at 1 mg/m³ for 6 h/day, 5 days/week for
4 weeks. Using both light and electron microscopy, the investigators
concluded that there were no significant effects of exposure on lung
structure.

5.5 Summary of studies of the effects of nitrogen compounds on
experimental animals

Responses to NO₂ exposure have been observed in several
laboratory animal species, resulting in the conclusion that these
effects could occur in humans. In addition, mathematical dosimetry
models suggest that the greatest dose of NO₂ is delivered to the same
region in both animal and human lungs (i.e. the centriacinar region
which is the junction of the conducting airway with the gas exchange
area). Thus, the responses of laboratory animals can be qualitatively
extrapolated to humans.

NO₂ exposure causes lung structural alterations. Exposure to
3760 µg/m³ (2.0 ppm) for 3 days has resulted in centriacinar damage,
including damaged cilia and alveolar wall oedema. Prolonged exposures
produce changes in the cells lining the centriacinar region, and the
tissue in this region (i.e., alveolar interstitium) becomes thicker.
These effects were seen in rats exposed to 940 µg/m³ (0.5 ppm)
baseline with brief peaks of 2800 µg/m³ (1.5 ppm) for 6 weeks or
exposures to 940 µg/m³ (0.5 ppm) NO₂ for 4 to 6 months.

Several animal studies clearly demonstrate that chronic exposure
to concentrations of NO₂ > 9400 µg/m³ (> 5.0 ppm) can
cause emphysema of the type seen in human lungs. Increased lung
distensibility was reported in mice exposed to 375 µg/m³ (0.2 ppm)
with peaks of 1500 µg/m³ (0.8 ppm) after 1 year of exposure.

NO₂ increases susceptibility to bacterial and viral pulmonary
infections in animals. Reduced phagocytic activity and reduced
mobility were observed in AMs from rabbits exposed for 13 days to
500 µg/m³ (0.3 ppm). The lowest observed concentration that
increases lung susceptibility to bacterial infections after acute
exposure is 3750 µg/m³ (2.0 ppm) NO₂ (a 3-h exposure study in mice).
Acute (17 h) exposures to > 4250 µg/m³ (> 2.3 ppm) NO₂ also
decrease pulmonary bactericidal activity in mice. After long-term
exposures (e.g., 3 to 6 months) to 940 µg/m³ (0.5 ppm) NO₂, mice
have decreased resistance to lung bacterial infections. Exposure of
mice for 1 year to 375 µg/m³ (0.2 ppm/week) with 1480 µg/m³
(0.8 ppm) spike followed by infection with streptococcus resulted in

increased mortality. NO₂ also increases lung susceptibility to viral
infections in mice. Subchronic (7-week) exposures to concentrations
as low as 470 µg/m³ (0.25 ppm) NO₂ can alter the systemic immune
system in mice.

NO₂ exposure has been shown to cause a clear dose-related
decrease in pulmonary antibacterial defences. Decreases in pulmonary
antibacterial defences occurred at concentrations ranging from
7520 µg/m³ (4 ppm) for Staphylococcus aureus to 37 500 µg/m³
(20 ppm) for Proteus mirabilis. Dose-response increases in
bacterial-induced mortality in mice was demonstrated with continuous
exposure to 940 µg/m³ (0.5 ppm) after 3 months.

When the relationship of NO₂ exposure concentration and duration
was studied, concentration had more influence than duration on the
outcome. This conclusion is primarily based on investigations of lung
antibacterial defences of mice, which also indicate that the exposure
pattern (e.g., baseline level with daily peaks of NO₂ or exposure
24 h/day versus 6 to 7 h/day) has an impact on the study results.

Structural changes in the lung become more severe as exposure
progresses from weeks to months at a given NO₂ concentration. Longer
exposures resulted in effects at lower concentrations.

NO₂ showed positive effects in some studies with Salmonella
strain TA100 and caused DNA strand breaks in a mammalian cell culture.
NO seems to be less active. High concentrations of NO₂ have induced
mutations in lung cells in vivo, but not in other organs. There are
no classical chronic bioassays for carcinogenicity. Studies concerning
enhancement of spontaneous tumours, co-carcinogenic effects, or
facilitation of the metastases of tumours to the lung are inadequate
to form conclusions. Possible secondary effects concern the
in vivo formation of nitrite and nitrosamines and atmospherically
formed mutagenic reaction products from NO_x and hydrocarbons.

The effects of exposure to mixtures of NO₂ and other pollutants
are dependent on the exposure regimen, species and end-point measured.
Most mixture research involves NO₂ and O₃ and shows that additivity
and synergism can occur. A similar conclusion can be drawn from the
more limited research with NO₂ and sulfuric acid. Findings of either
additivity or synergism are of concern because of the ubiquitous
co-occurrence of NO₂ and O₃. Extrapolation of these findings is not
currently possible.

NO is a potent vasodilator and effects can be demonstrated with
inhaled concentrations of approximately 6130 µg/m³ (5 ppm) in sheep
and guinea-pigs. NO also reduces resistance to bacterial infection
via the inhalation route in female mice exposed to 2452 µg/m³
(2 ppm). Morphological alterations in the alveoli and thickening of

the alveolocapillary membrane are seen in rabbits at 6130 µg/m³.
Methaemoglobin formation is seen at concentrations above 12 260 µg/m³
(10 ppm).

NO₂ acts as a strong oxidant. Unsaturated lipids are readily
oxidized with peroxides as the dominant product. Both ascorbic acid
and alpha-tocopherol inhibit the peroxidation of unsaturated lipids.
When ascorbic acid is sealed within bi-layer liposomes, NO₂ rapidly
oxidizes the sealed ascorbic acid. The protective effects of
alpha-tocophernol (vitamin E) and ascorbic acid (vitamin C) in animals
and humans are due to the inhibition of NO₂ oxidation. NO₂ also
oxidizes membrane proteins. The oxidation of either membrane lipids or
proteins results in the loss of cell permeability control. The lungs
of NO₂-exposed humans and experimental animals have larger amounts of
protein within the lumen. The recruitment of inflammatory cells and
the remodelling of the lung are a consequence of these events.

The oxidant properties of NO₂ also induce the peroxide
detoxification pathway of glutathione peroxidase, glutathione
reductase, and glucose-6-phosphate dehydrogenase. Increases in the
peroxide detoxification pathway occur in animals in a roughly
dose-response relationship following NO₂ exposure.

The mechanism of action of NO is less clear. NO is readily
oxidized to NO₂ and then peroxidation occurs. Because of concomitant
exposure to some NO₂ in NO exposures, it is difficult to discriminate
NO effects from those of NO₂. NO is, however, a potent second
messenger modulating a wide variety of essential cellular functions.

Peroxyacetyl nitrate (PAN) decomposes in water generating
hydrogen peroxide. Little is known of the mechanism of action, but
oxidative stress is likely for PAN and its congeners.

Inorganic nitrates may act by alterations in intracellular pH.
Nitrate ion is transported into Type 2 cells, acidifying the cell.
Nitrate also mobilizes histamine from mast cells. Nitrous acid could
also act to alter intracellular pH, but this mechanism is unclear.

The mechanisms of action of the other nitrogen oxides are unknown
at present.

6. CONTROLLED HUMAN EXPOSURE STUDIES OF NITROGEN OXIDES

6.1 Introduction

The effects of nitrogen oxides (NO_x) on human volunteers exposed
under controlled exposure conditions are evaluated in this chapter.
Of the NO_x species typically found in the ambient air, NO₂ has been
the most extensively studied. Nitric oxide (NO), nitrates, nitrous
acid and nitric acid also have been evaluated and are discussed here,
as are investigations of mixtures of NO_y and other co-occurring
pollutants. A more extensive detailed review of this literature can
be found in US EPA (1993).

Most volunteers for human clinical studies are young, healthy
adult males, but other potentially susceptible subpopulations,
especially asthmatics, patients with chronic obstructive pulmonary
disease (COPD), children and the elderly have also been studied. Many
exposures are conducted while the volunteer performs some form of
controlled exercise. The exercise increases ventilation, which
increases the mass of pollutant inhaled per unit time and may alter
the distribution of the dose within the lung. More information on
NO₂ dosimetry is presented in chapter 5. Important methodological and
experimental design considerations for controlled human studies have
been discussed in greater detail by Folinsbee (1988).

In many human clinical studies of NO₂ exposure, both pulmonary
function and airway responsiveness to bronchoconstrictors have been
measured. Spirometric measurements of lung volume, as well as
measurements of airway resistance, ventilation volume, breathing
pattern, and other tests provide information about some of the basic
physiological functions of the lung. Dynamic spirometry tests (forced
expiratory tests such as forced expiratory volume in 1 second (FEV₁),
maximal and partial flow-volume curves (including those using gases of
different densities such as helium), peak flow measurements, etc.),
and measurements of specific airway resistance/conductance (SR_aw,
SG_aw) are also used. Most of these tests evaluate large airway
function. However, since NO₂ deposition occurs primarily around
the junction of the tracheobronchial and pulmonary regions (section
5.2.1), many of these tests may not provide the necessary information
to evaluate fully the effects of NO₂. Other tests that may evaluate
small airway function (e.g., multiple breath nitrogen washout tests,
closing volume tests, aerosol deposition/distribution tests, density
dependence of flow-volume curves, and frequency dependence of dynamic
compliance) are less frequently used, and the extent to which they
indicate small airways function is not clearly established. As
discussed below, NO₂ can increase airway responsiveness to chemicals
that cause bronchoconstriction, such as histamine or cholinergic
agonists (i.e., acetylcholine, carbachol or methacholine). Other
challenge tests use allergens, exercise, hypertonic saline or cold-dry

air. Responses are usually measured by evaluating changes in airway
resistance (R_aw) or spirometry (e.g. FEV₁) after each dose of the
challenge is administered. Generally, asthmatics are significantly
more responsive than healthy normal subjects to these types of airway
challenge (O'Connor et al., 1987). However, there is some overlap
between the most responsive healthy subjects and the least responsive
(to histamine) asthmatics (Pattemore et al., 1990).

In the following sections, the changes in pulmonary function and
airway responsiveness after NO₂ exposure in healthy subjects are
discussed. Responses of asthmatics and patients with chronic
obstructive pulmonary disease (COPD) are then evaluated. A brief note
regarding age-related susceptibility is followed by a review of the
effects of NO₂ on pulmonary host defences and on biochemical markers
in lung lavage fluid or in the blood. The effects of two other
oxidized nitrogen compounds, NO and nitric acid vapour are also
discussed. Finally, the effects of mixtures of oxidized nitrogen
compounds (NO₂, NO, HNO₃) with other gaseous or particulate
pollutants are considered. An overall summary is presented at the end
of the chapter.

6.2 Effects of nitrogen dioxide

6.2.1 Nitrogen dioxide effects on pulmonary function and airway
responsiveness to bronchoconstrictive agents

Much research has focused on NO₂-induced changes in pulmonary
function and airway responsiveness to bronchoconstrictive agents.
Healthy adults do not typically respond to low levels of NO₂
(< 1880 µg/m³, 1 ppm). However, asthmatics appear to be the most
susceptible members of the population (section 6.2.1.2). Asthmatics
are generally much more sensitive to inhaled bronchoconstrictors. The
potential addition of an NO₂-induced increase in airway response to
the already heightened responsiveness to other substances raises the
possibility of exacerbation of asthma by NO₂. Another potentially
susceptible group includes patients with COPD (section 6.2.1.3). A
major concern with COPD patients is the absence of an adequate
pulmonary reserve, so that even a relatively small alteration in lung
function in these individuals could potentially cause serious
problems. In addition, both adolescents and the elderly have
been evaluated, to determine whether differential age-related
susceptibility exists (section 6.2.1.4).

Table 39. Effects of nitrogen dioxide (NO₂) on lung function and airway responsiveness of healthy subjects^a

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

188 0.1 60 15 M 23-29 years, No symptoms; no odour Hazucha et al.
NS detection; no effect (1982, 1983)
on SR_aw.

188 0.1 240 6 Normal adults No effects of NO₂ Sackner et al.
564 0.3 (1980)
940 0.5
1880 1.0

226 0.12 60 4 M/6 F 13-18 years No effects on lung Koenig et al.
function. (1985)

226 0.12 40 10 32.5 3 M/7 F 14-19 years No effects on R_tau or Koenig et al.
338 0.18 40 4 M/6 F 15-19 years spirometry. (1987a,b)

230 0.12 20 5 M/4 F 20-36 years, Suggestion of change Bylin et al.
460 0.24 NS in SR_aw in normals: (1985)
910 0.48 SR_aw tended to increase
at 476 µg/m³ and
tended to decrease at
910 µg/m³. Analysis of
variance indicates no
significance. No effects
on bronchial reactivity.
Median odour threshold
75 µg/m³.

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

282 0.15 120 60 50 W 6 M 19-24 years No symptoms; no Kagawa & Tsuru
pulmonary function (1979); Johnson
effects. Suggested et al. (1990)
individual changes
in SG_aw.

338 0.18 30 10 (L) L approx. 25 9 M 18-23 years, No change in lung Kim et al.
564 0.3 16 (H) H approx. 72 "collegiate function. (1991)
athletes"

508 0.27 60 Healthy, Possible small Rehn et al.
1993 1.06 young M increase in R_aw at (1982)
508 µg/m³ (0.27 ppm).

564 0.3 120 60 50 W 6 19-25 years No effect on SG_aw. Kagawa (1986)

564 0.3 225 30 approx. 40 10 M/10 F 20-48 years No symptom, lung Morrow & Utell
(3 × 10) (FEV₁/FVC function or airway (1989)
76-95%) reactivity responses
to carbachol for either
of the 20-48 year or
the 49-69 year age
groups.

564 0.3 225 21 30-40 10 M/10 F 49-69 years,
(3 × 7) (FEV₁/FVC
72-84%)

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

940 0.5 120 15 Light/ 10 Healthy, three Decreased quasistatic Kerr et al.
moderate ex-smokers in compliance. Non-random (1979)
group exposure sequence air-
NO₂. No change in
spirometry or resistance.
Apparent compliance
change may be due to
exposure order.

940 0.5 120 15 10 Normal adults Decreased static lung Kulle (1982)
compliance.

940 0.5 240 30 55 10 M 26.4 years No significant effects Stacy et al.
on spirometry or R_aw. (1983)

1128 0.6 120 60 25 8 M/8 F 51-76 years No statistically Drechsler-Parks
significant changes et al. (1987)
in lung function due
to NO₂ exposure in
either age group.

8 M/8 F 18-26 years,
NS

1128 0.6 180 60 approx. 40 7 M/2 F Healthy, NS No change in Frampton et al.
(6 × 10) spirometry, R_aw or (1989a)
carbachol reactivity.

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

94 with 0.05 with 135 60 11 M/4 F Non-reactive
3760 2.0 spikes 3 × 15 (6 × 10) (carbachol)
spikes

(1) 1128 (1) 0.6 180 60 39 6 M/2 F 30.3 ± 1.4 There were no changes Frampton et al.
years, NS in airway mechanics (1991)
(FVC, FEV₁, SG_aw).
Responsiveness to
(2) Var. (2) Var. 180 60 43 11 M/4 F 25.3 ± 1.2 carbachol was
(94 (0.05 years, NS significantly increased
background background after 2820 µg/m³ NO₂
with with 180 60 approx. 40 5 M/3 F 32.6 ± 1.6 (Group 3) but not after
3 × 15 min 3 × 15 years, NS the other exposures
at 3760) min at (Groups 1 and 2). Degree
2.0 ppm) of baseline
responsiveness to
carbachol was not
related to response after
2820 µg/m³.

(3) 2820 (3) 1.5 180 60 39 12 M/3 F 23.5 ± 0.7
years, NS

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

1128 0.6 120/day 60 approx. 30-40 4 M/1 F NS, 21-36 No effects of repeated Boushey et al.
for 4 days years, FEV₁/ NO₂ exposure on (1988) (Part 2)
FVC% range respiratory function
73-83%, (SR_aw, FVC, FEV₁) or
"normal" symptoms.
methacholine
responsiveness

1128 0.6 60 60 70 20 M Healthy No effect of NO₂ on Adams et al.
50 20 F spirometry or airway (1987)
resistance.

1166 0.62 120 15 33 5 M Healthy No significant Folinsbee et al.
30 33 5 M pulmonary function (1978)
responses attributed
to NO₂ exposure.

1316-3760 0.7-2.0 10 10 Increased resistance Suzuki &
10 min after exposure. Ishikawa (1965)

1316 0.7 60 5 19-22 years, No effects on airway Toyama et al.
3 of 5 were conductance. (1981)
investigators

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

1880 1.0 120 (2 60 Light 16 Healthy Air-NO₂-NO₂ fixed Hackney et al.
consecutive exposure sequence. (1978)
days) 1.5% decrease in FVC
after second day of
NO₂. Not clear that
the decreased FVC is
an NO₂ effect or an
order effect. No other
effects.

1880 1.0 120/day, 22 Healthy, NS, Overall trend for a Goings et al.
3760 2.0 3 days 21, 22 seronegative slight decrement in (1989)
5640 3.0 22 FEV₁ with NO₂ exposure
(< 1%). No change in
methacholine
responsiveness as a
result of NO₂ exposure
or viral infection
status.

1880 1.0 120 16 11 S After 14 100 µg/m³ Beil & Ulmer
4700 2.5 120 16 5 NS (120 min) and (1976)
9400 5.0 120 16 9400 µg/m³ (14 h),
14 100 7.5 120 16 8 S responsiveness to
9400 5.0 840 8 acetylcholine increased.
Resistance increased
after all but the
1880 µg/m³ exposure.

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

3760 2.0 60 8 M/3 F 18-36 years, Vitamin C blocked Mohsenin (1987b)
NS NO₂-induced increase
in airway reactivity
to methacholine.

3760 2.0 120 13 M/5 F Normal, NS, No symptoms; no lung Mohsenin (1988)
18-33 years function changes.
Increased methacholine
reactivity.

7520-9400 4.0-5.0 10 Bag exposure Abe (1967)
technique. Airway
resistance increased
30 min after end of
exposure. No change in
spirometry.

7520 4.0 75 15 (L) L 20-29 16 M/ 9 F 18-45 years, No change in SR_aw Linn & Hackney
15 (H) H 44-57 NS associated with NO₂. (1983); Linn
Small but significant et al. (1985b)
decrease in blood
pressure; some mild
increase in symptoms.

9400 5.0 15 16 Healthy Decreased D_LCO 18%. Von Nieding et
al. (1973a)

Table 39 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristice
(min) (min) (litres/min) gender
µg/m³ ppm

9400 5.0 120 Intermittent Light 11 M Healthy Increased resistance Von Nieding et
60%. Remained elevated al. (1977)
for 60 min. Possible
decrease in P_aO₂.

9400 5.0 120 60 220 11 M Healthy Resistance increased Von Nieding et
(4 × 15) 60%. Remained elevated al. (1979)
60 min after exposure.
Possible decrease in
earlobe PO₂.

^a Modified from US EPA (1993)
Abbreviations:
M = Male; F = Female; S = Active smoker; NS = Non-smoker; FEV₁ = Forced expiratory volume in 1 second; FVC = Forced vital capacity;
SR_aw = Specific airway resistance; Var = Variable; R_aw = Airway resistance; SG_aw = Specific airway conductance; W = Watts; L = Light;
H = Heavy; R_T = Total respiratory resistance; D_LCO = Diffusing capacity for carbon monoxide; P_aO₂ = Arterial partial pressure of oxygen;
PO₂ = Partial pressure of oxygen
6.2.1.1 Nitrogen dioxide effects in healthy subjects

The effects of NO₂ levels greater than 1880 µg/m³ (1.0 ppm) on
respiratory function in healthy subjects have been examined in several
studies (Table 39). Early work indicated that NO₂ increased R_aw
or total respiratory resistance (R_T) at concentrations above
2820 µg/m³ (1.5 ppm) in healthy volunteers (Abe, 1967; Von Nieding et
al., 1970, 1973a, 1979; Von Nieding & Wagner, 1977). Although Beil &
Ulmer (1976) found a small but statistically significant increase in
R_T after a 2-h exposure to > 4700 µg/m³ (> 2.5 ppm) NO₂, the
response was not appreciably increased by raising the NO₂
concentration to 9400 or 14 100 µg/m³ (5.0 or 7.5 ppm). Also, airway
responsiveness to acetylcholine was increased after exposure to
14 100 µg/m³ for 2 h or to 9400 µg/m³ for 14 h, but not after the
2-h exposures to < 9400 µg/m³.

In contrast, some investigators found no effects at high
concentrations. For example, a 75-min exposure with light and heavy
exercise to 7520 µg/m³ (4.0 ppm) NO₂ did not affect R_aw (Linn et
al., 1985b), and a 1-h resting exposure to 3760 µg/m³ (2 ppm) did not
cause a change in lung volume, flow-volume characteristics on either
full or partial expiratory flow-volume (PEFV) curves, or SG_aw
(Mohsenin, 1987b, 1988). However, NO₂ did increase airway
responsiveness to methacholine (Mohsenin, 1987b, 1988).

Goings et al. (1989) found no effects of exposure to NO₂ at
1880, 3760 or 5640 µg/m³ (1, 2 or 3 ppm; for 2 h/day on 3 consecutive
days) on respiratory symptoms, lung function or airway reactivity to
methacholine. Laboratory-induced influenza virus infection did not
alter airway responsiveness in either sham (clean air) or NO₂
exposure groups. The infectivity portion of this study is discussed
in section 6.2.2.

The influence of exposure pattern was examined by Frampton et al.
(1991), using healthy subjects exposed for 3 h to either 1128 µg/m³
(0.60 ppm), 2820 µg/m³ (1.5 ppm) or a variable concentration protocol
where three 15 min peaks of 3760 µg/m³ (2.0 ppm) were added to a
background level of 94 µg/m³ (0.05 ppm). Nitrogen dioxide did not
affect airway mechanics (forced vital capacity (FVC), FEV₁, SG_aw).
However, after exposure to 2820 µg/m³, but not to the other
concentrations, there was a small but statistically significant
increase in airway responsiveness to carbachol. This study supported
the earlier observations by Mohsenin (1987b, 1988) of increased airway
responsiveness after a 1-h exposure to 3760 µg/m³. Mohsenin (1987b)
further observed that the NO₂-induced increase in airway
responsiveness could be blocked by elevation of serum ascorbate level
through pretreatment with the antioxidant ascorbic acid (vitamin C).

At concentrations below 1880 µg/m³ (1.0 ppm) NO₂, pulmonary
function and airway responsiveness have generally not been found to be
affected in healthy adult subjects (Beil & Ulmer, 1976; Folinsbee et
al., 1978; Hackney et al., 1978; Kerr et al., 1979; Sackner et al.,
1980; Toyama et al., 1981; Kulle, 1982; Hazucha et al., 1982, 1983;
Stacy et al., 1983; Kagawa, 1986; Adams et al., 1987; Drechsler-Parks
et al., 1987; Drechsler-Parks, 1987; Boushey et al., 1988; Morrow &
Utell, 1989; Frampton et al., 1989a, 1991; Kim et al., 1991).
Although some investigators have at times reported statistically
significant effects, there does not appear to be a consistent pattern
of acute responses in healthy subjects at these low NO₂
concentrations.

Kagawa & Tsuru (1979) reported the lowest NO₂ exposure
concentration that appeared to cause an effect. Healthy men were
exposed to 282 µg/m³ (0.15 ppm) NO₂ for 2 h while performing light,
intermittent exercise. The authors suggested that NO₂ caused some
statistically significant changes, i.e. a 0.5% decrease in vital
capacity (VC) and a 16% decrease in an index of small airway function
(i.e. FEF75_HeO2: FEF75_AIR; the ratio of forced expiratory flow at 75%
FVC expired while breathing a helium-oxygen mixture compared to FEF75
while breathing air). These findings should be interpreted with the
consideration that multiple t-tests were used in the statistical
analysis of these data. Rehn et al. (1982) reported a small (17%)
increase in SR_aw in men exposed to 500 µg/m³ (0.27 ppm) for 1 h, but
a higher concentration (2000 µg/m³, 1.06 ppm) did not cause an
effect.

Bylin et al. (1985) reported that the SR_aw of normal subjects
exposed to 230, 460 and 910 µg/m³ (0.12, 0.24 and 0.48 ppm) for
20 min was unaffected. Specific comparisons revealed a significant 11%
increase in SR_aw at 460 µg/m³ (0.24 ppm) and a 9% decrease in SR_aw
at 910 µg/m³. Bronchial responsiveness to histamine was increased by
910 µg/m³ NO₂.

Symptomatic responses of subjects exposed to NO₂ were evaluated
in several of the above studies. None of these studies, including
exposures for as long as 75 min to 7520 µg/m³ (4.0 ppm) NO₂ (Linn &
Hackney, 1983; Linn et al., 1985b), resulted in a significant increase
in respiratory symptoms. In studies of sensory effects, subjects were
unable to detect the odour of 188 µg/m³ (0.1 ppm) NO₂ (Hazucha et
al., 1983), but Bylin et al. (1985) observed an odour threshold of
75 µg/m³ (0.04 ppm) for normal subjects and 150 µg/m³ (0.08 ppm) for
asthmatics.

6.2.1.2 Nitrogen dioxide effects on asthmatics

Studies of the effects of exposures to NO₂ on respiratory
function and airway responsiveness of asthmatics are summarized in
Table 40. Asthmatics are generally more responsive than healthy
subjects to NO₂. However, as can be seen in Table 40, there is
substantial variability in observed responses between and even within
laboratories. This variability is illustrated in Fig. 22 and 23, in
which changes in airway resistance and FEV₁ are related to the
"exposure dose" of NO₂ (calculated as ppm × litres of air breathed
over the duration of exposure) (US EPA, 1993). The individual
investigations that yielded the data used to develop these
illustrations will be discussed in more detail below. Other studies,
not discussed separately, are also summarized in Table 40. The review
by the US EPA (1993) provides more detail on many of these studies.
Although differences in exposure protocols may explain some of the
differences between studies, the explanation most often invoked is
that there may be differences in the severity of asthma among the
subject groups tested. There are numerous definitions of "asthma
severity" (see, for example, National Institutes of Health, 1991).
Those applied to the key asthma studies discussed here (based on the
data available) are: (1) mild: controlled by bronchodilators and
avoidance of known precipitating factors, does not interfere with
normal activities; and (2) moderate: often requires periodic use of
inhaled steroids in treatment and may interfere with work or school
activities. Those with severe asthma are seldom used as subjects for
NO₂ studies because their disease can include life-threatening
episodes. Typical volunteers for the studies described here had mild
allergic asthma.

Avol et al. (1988) studied a group of moderate-to-severe
asthmatics exposed to 564 and 1128 µg/m³ (0.3 and 0.6 ppm) NO₂ for
2 h with moderate intermittent exercise. NO₂ did not cause
significant changes in SR_aw or FEV₁. Results of tests of airway
responsiveness to cold air suggested a slightly increased response
after exposure to 564 µg/m³, but not after 1128 µg/m³. A post hoc
analysis of a subgroup of subjects with the most abnormal lung
function (i.e., FEV₁/FVC ratios < 0.65) did not find enhanced
susceptibility. In a subsequent study using 564 µg/m³ NO₂, Avol et
al. (1989) found decreases in FEV₁, FVC and peak expiratory flow rate
(PEFR), but no change in responsiveness to cold air challenge.

Table 40. Effects of nitrogen dioxide (NO₂) on lung function and airway responsiveness of asthmatics^a

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

188 0.1 60 9 20-51 years, No effect of NO₂ on Ahmed et al.
"history of FEV₁, SG_aw or on (1983a)
bronchial asthma" ronchial reactivity to
ragweed antigen, either
immediately or 24 h
after exposure.

188 0.1 60 20 M/34 F 18-39 years No significant effect Ahmed et al.
on SG_aw, FEV₁, V_ISOV; (1983b)
variable effect on
carbachol reactivity.
No information on
controlled exposure.

188 0.1 60 15 M 21-46 years, No significant Hazucha et al.
mild or inactive changes in R_T or (1982, 1983)
disease responsiveness to
methacholine associated
with NO₂ exposure.

207 0.11 60 6 M/1 F 1 Smoker, No change in SR_aw or Orehek et al.
(132-301) (0.07-0.16) 3 asthmatic, in responsiveness to (1981)
4 allergic grass pollen in 3
allergic asthmatics
and 4 allergic
subjects.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

210 0.11 60 13 M/7 F 15-44 years, 13/20 subjects had Orehek et al.
(169-244) (0.09-0.13) 13 mild/7 mod enhanced responses to (1976)
(n = 20) asthmatics; carbachol after
210 µg/m³ NO₂. Post
hoc statistical
analysis questionable.

489 0.26 65 years 1/4 subjects had Orehek et al.
(n = 4) enhanced responses (1976)
to carbachol after
489 µg/m³ NO₂.

226 0.12 60 4 M/6 F 12-18 years, No significant effects Koenig et al.
asympt., on pulmonary function (1985)
extrinsic due to NO₂. Increased
allergic symptoms after NO₂
asthmatics exposures.

226 0.12 60 4 M/6 F 12-18 years No change in FEV₁, Koenig et al.
226 0.12 40 10 33 4 M/6 F 11-19 years R_T increased 10.4% (1987a,b)
338 0.18 40 10 39 7 M/3 F 12-18 years, (NS), 3% decrease
asympt., in FEV₁ (p < 0.06).
extrinsic
allergic
asthmatics

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

230 0.12 20 6 M/2 F 17-45 years, No significant change Bylin et al.
460 0.24 20 very mild in SR_aw at any NO₂ (1985)
910 0.48 20 asympt. levels. Histamine
reactivity tended to
increase.

260 0.14 30 8 M/12 F 17-56 years, Overall trend for SR_aw Bylin et al.
510 0.27 very mild to decline during (1988)
1000 0.53 asympt. exposure period, not
related to NO₂
concentration.
Histamine bronchial
reactivity tended to
increase after 260 and
510 µg/m³ NO₂ exposure.

376 0.2 120 60 approx. 20 12 M/19 F 18-55 years, No effects on Kleinman et al.
wide range of spirometry or airway (1983)
asthma severity resistance. Airway
reactivity to
methacholine results
variable-tended to
increase with exposure.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

470 0.25 30 10 30 9 M/2 F 18-55 years, Mouthpiece exposure Joerres &
mild asympt. system. No changes in Magnussen
methacholine (1991)
responsiveness were
observed after NO₂
exposure.

470 0.25 30 10 M/4 F 20-55 years, After NO₂ exposure, Joerres &
mild asthma, responsiveness to Magnussen
most asympt. inhaled SO₂ was (1990)
increased. No effect
of NO₂ alone on SR_aw.

564 0.3 30 20 approx. 30 5 M/4 F 23-34 years No changes in SR_aw, Rubinstein et
FVC, FEV₁, SBN₂ or al. (1990)
symptoms after NO₂
exposure. NO₂ exposure
did not increase airway
responsiveness to SO₂.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

564 0.3 30 10 30 15 20-45 years, Resting 20 min Bauer et al.
mild asympt. exposures produced no (1986)
effects. Slight excess
decrease in FEV₁ and
PEFR in NO₂ plus
exercise above that
caused by exercise
alone. PEFR, -16% (air).
-28% (NO₂); FEV₁ -5.5%
(air), -9.3% (NO₂).
Significantly increased
response to cold air
after NO₂ exposure.

564 0.3 225 30 30-40 10 M/10 F 19-54 years Group findings Morrow & Utell
(3 × 10) indicated no (1989)
significant responses.
No change in lung
function, symptoms,
carbachol reactivity.
Subjects studied
previously (Bauer et
al., 1986) showed
possible responses to
NO₂. New subject
subgroup showed
significantly greater
response in air
exposures.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

564 A. 03 110 60 42 A. 13 M 19-35 years, FEV₁ decreased 11% Roger et al.
mild asthmatics in NO₂ but only 7% in (1990)
air, after first 10 min
of exercise. Smaller
changes later in
exposure.

282 B. 0.15 75 30 42 B. 21 No increase in airway
564 0.3 reactivity to
1128 0.6 methacholine 2 h after
exposure. Nochange in
FEV₁ or SR_aw as a
result at NO₂
exposure.

564 0.3 180 90 30 24 M/10 F 10-16 years After 60 min of Avol et al.
exposure, FEV₁, FVC (1989)
and PEFR (-3.4, -4.0
and -5.6%,
respectively) were
significantly reduced.
No change in airways
responsiveness to cold
air challenge. SR_aw
increased 17% after NO₂
exposure. After 180 min
of exposure, the
responses had returned
to baseline levels.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

564 0.3 120 60 40 27 M/32 F 18-50 years, Exercise-related Avol et al.
some moderate increases in symptoms. (1988)
asthmatics Possible NO₂-related
decrease in FEV₁,
PEFR. Increased cold
air response after
564 µg/m³.

1128 0.6 120 60 41 More consistent
increases in SR_aw at
1128 µg/m³ but not
significantly different
from air and 564 µg/m³.

564 0.3 60 30 41 15 M/6 F 20-34 years, No effect of NO₂. Linn et al.
1880 1.0 60 30 41 mild asthmatics Exercise-related (1986)
5640 3.0 60 30 41 increase in SR_aw under
all conditions.

940 0.5 120 15 9 M/4 F 19-50 years, Increased respiratory Kulle (1982)
3 Smokers symptoms in 4/13
subjects. Also,
increased static lung
compliance. Impossible
to determine amount of
effect due to NO₂.

Table 40 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

940 0.5 60 10 22-44 years, No change in symptoms. Mohsenin (1987b)
mild asthmatics Significant group mean
increase in
responsiveness to
methacholine after
NO₂ exposure. No other
function changes.

940 + 0.5 + 120 60 approx. 20 6 M/12 F 33 years, No significant effect Linn et al.
857 0.3 ppm 6 ex-smokers physician- on spirometry, RT. (1980a)
SO₂ SO₂ asthma diagnosed

7520 4.0 75 a. 15 a. 25 12 M/11 F 18-34 years, No NO₂ effects on Linn & Hackney
b. 15 b. 49 physician- SR_aw, symptoms, heart (1984); Linn et
diagnosed rate, skin al. (1985b)
asthma conductance. Small
decrease in
systolic blood
pressure.

^a Modified from US EPA (1993)
M = male; F = female; SG_aw = specific airway conductance; FEV₁ = forced expiratory volume in 1 second; V_ISOV = volume of isoflow;
PEFR = peak expiratory flow; SR_aw = specific airway resistance; FVC = forced vital capacity; Asympt. = asymptomatic;
R_T = total respiratory resistance; NS = not significant; SO₂ = sulfur dioxide; SBN₂ = single breath nitrogen washout

Roger et al. (1990) reported the effects of NO₂ exposure on mild
asthmatics. Their first study was a pilot study of 12 mild asthmatics
exposed to 564 µg/m³ (0.3 ppm) for 110 min, including three 10-min
periods of exercise. After the first 10 min of exercise in NO₂,
there was a decrease in FEV₁ that persisted for the remainder of the
exposure period, although the overall responses were progressively
less with successive periods of exercise, as is common with
exercise-induced asthma when the exercise is intermittent. Their
subsequent concentration-response study of twenty-one subjects
included six responsive subjects from the pilot study; volunteers were
exposed to 282, 564 and 1128 µg/m³ (0.15, 0.30 and 0.60 ppm) NO₂ for
75 min, with three 10-min exercise periods. In contrast to the pilot
study, there were no effects of NO₂ on pulmonary function or airway
responsiveness to methacholine, tested 2 h after exposure ceased. The
authors suggested that the differences between the pilot and the main
study may have been due to more reactive airways in the pilot study
asthmatics. Because the studies were conducted during different
seasons, seasonal differences in temperature, air pollution, ambient
aeroallergens or other factors may have contributed to some of the
variability in response.

Asthmatics exposed to 230, 460 and 910 µg/m³ (0.12, 0.24 and
0.48 ppm) NO₂ for 20 min were studied by Bylin et al. (1985).
Changes in SR_aw during the four exposures averaged +3% after air and
+9%, -2% and -14% after the three levels of NO₂, respectively; these
changes were not significantly different. There was a tendency for an
increase in thoracic gas volume (TGV) after NO₂ exposures (9 to
10%), but differences in pre-exposure values for TGV were probably
responsible, rather than NO₂. There were no significant changes in
tidal volume or respiratory rate. At the highest concentration tested
(910 µg/m³, 0.48 ppm), histamine bronchial responsiveness was
increased.

In mild asthmatics exposed for 30 min to 260, 510 and 1000 µg/m³
(0.14, 0.27 and 0.53 ppm), there were no significant changes in SR_aw,
although there was a general trend for SR_aw to fall throughout the
period of exposure at all NO₂ concentrations (Bylin et al., 1988).
There was, however, a significant increase (p = 0.03) in airway
responsiveness to histamine after 30 min of exposure to 510 µg/m³
(0.27 ppm) only. The absence of a concentration-related increase in
responsiveness is not inconsistent with other studies. This
observation contrasts with earlier results (Bylin et al., 1985) that

suggested a possible increased responsiveness after exposure to
910 µg/m³ (0.48 ppm). Because of the use of a non-parametric pair
comparison test that was not adjusted for multiple comparisons, the
raw data presented in the paper were subjected to reanalysis (US EPA,
1993) using a Friedman non-parametric analogue of an F test, which
is probably more appropriate for these data than a series of
Wilcoxon matched pairs signed rank tests. This analysis showed no
statistically significant change in histamine responsiveness due to
NO₂ exposure.

Asthmatics exposed to 564 µg/m³ (0.3 ppm) NO₂ by mouthpiece
for 20 min at rest followed by 10 min of exercise (30 litres/min)
experienced a statistically significant spirometric response to NO₂
(Bauer et al., 1986). After NO₂ exposure, 9 out of 15 asthmatics had
a decrease in FEV₁; both the pre-post exposure difference on the NO₂
day (10.1%) and the pre-post NO₂ minus the pre-post air (i.e.,
delta-delta) differences (6%) were significant using a paired t-test.
Maximum expiratory flow at 60% total lung capacity (PEFV curve) was
also decreased, but FVC and SG_aw were not altered. Nine out of
twelve subjects experienced an increase in airway responsiveness to
cold air. The mouthpiece exposure system used in this study contained
relatively dry air (relative humidity, RH, of 9 to 14% at 20°C) and
airway drying may have interacted with NO₂ to cause greater
responses. However, Bauer et al. (1986) controlled for the airway
drying effect by exposing subjects to clean air at the same
temperature and RH. Nevertheless, air temperature and humidity
effects may be an important consideration for NO₂ effects in winter
in the temperate regions of the world.

Linn et al. (1985b) and Linn & Hackney (1984) exposed mild
asthmatics to 7520 µg/m³ (4.0 ppm) NO₂ for 75 min, with two 15-min
exercise periods. There was no significant difference in lung
function that could be attributed to NO₂; if anything, SR_aw tended
to be slightly lower with the NO₂ exposures.

The reasons for the differences between the group of asthmatics
exposed to 7520 µg/m³ (4 ppm) for 75 min (with exercise) (Linn et
al., 1985b) and the group exposed to 564 µg/m³ (0.30 ppm) for 30 min
with exercise studied by Bauer et al. (1986) are not clear. The
subjects of Bauer et al. were exposed to NO₂ in dry air through a
mouthpiece which could have caused some drying of the upper airways;
Linn et al. (1985b) used a chamber exposure. Second, the subjects in
the Linn et al. (1985b) study tended to have milder asthma than the
subjects in the Bauer et al. (1986) study. There were differences in
the season in which the two studies were conducted, and there may have
been a difference in background exposure to NO₂ (outdoors and/or
indoors). In addition, increased bronchial reactivity to cold air was
an important finding in the Bauer et al. (1986) study, but it was not
measured by Linn et al. (1985b).

Further research was conducted by Linn et al. (1986) on mild
asthmatics exposed to 564, 1880 and 5640 µg/m³ (0.30, 1.0 and
3.0 ppm) NO₂ for 1 h. The exposures included intermittent, moderate
exercise. As in the previous study with 7520 µg/m³ (4.0 ppm) NO₂,
there were no significant effects of NO₂ on spirometry, SR_aw or
symptoms. Furthermore, there was no significant effect on airway
responsiveness to cold air. In order to examine the suggestion that
the severity of response to NO₂ may be related to the clinical
severity of asthma, the authors selected three subjects characterized
as having more severe illness. Although they experienced markedly
larger changes in resistance than other milder asthmatics under all
exposure conditions, there was no indication that the responses of
these subjects were related to NO₂ exposure.

Mohsenin (1987a) found no changes in symptoms, spirometry, or
plethysmography in mild asthmatics exposed to 940 µg/m³ (0.5 ppm)
NO₂ for 1 h at rest. However, airway responsiveness to methacholine
increased after the NO₂ exposure.

The effects of previous NO₂ exposure on SO₂-induced
bronchoconstriction has been examined by Joerres & Magnussen (1990)
and Rubinstein et al. (1990). Neither study found changes in
pulmonary function after NO₂ exposure. Joerres & Magnussen (1990)
exposed mild-to-moderate asthmatic subjects to 470 µg/m³ (0.25 ppm)
NO₂ for 30 min while breathing through a mouthpiece at rest. After
the NO₂ exposure, airway responsiveness to 1965 µg/m³ (0.75 ppm)
SO₂ was increased. Rubinstein et al. (1990) exposed asthmatics to
564 µg/m³ (0.30 ppm) NO₂ for 30 min (including 20 min light
exercise). No mean change in responsiveness to SO₂ occurred, but one
subject showed a tendency toward increased responsiveness. The
reasons for the different findings in these two studies is not clear,
especially as the subjects of Rubinstein et al. (1990) were exposed to
a higher NO₂ concentration and exercised during exposure. However,
Joerres & Magnussen's subjects appeared to have had slightly more
severe asthma and were somewhat older. The modest increase in SR_aw
caused by exercise in the Rubinstein et al. (1990) study may have
induced a refractory state to SO₂. Finally, the different method of
administering the SO₂ bronchoprovocation test may have had an
influence. Joerres & Magnussen (1990) increased minute ventilation
(V._E) at a constant SO₂ concentration, whereas Rubinstein et al.
(1990) increased SO₂ concentration at constant V_E.

A number of studies of the effects of NO₂ exposure in asthmatics
on changes in airway responsiveness to bronchoconstrictors have been
presented in Table 40, but not evaluated in the text. Various types
of inhalation challenge tests have been used (methacholine, histamine,
cold air, etc.). Some exposures were conducted at rest and others
while performing some exercise. For twenty studies for which
individual data were available, a meta analysis (Folinsbee, 1992) was
performed to assess the changes in airway responsiveness in asthmatics
exposed to NO₂. The aim of the meta analysis was to examine the
diversity of response seen in the various studies and to examine
factors such as NO₂ concentration, exercise, and airway challenge
method that could help explain some of the variability in response.
Such questions could not be adequately addressed using individual
studies. The analysis provides only a qualitative examination of
concentration-response relationships. For this analysis, the
directional change (i.e., increased or decreased) in airway
responsiveness after NO₂ exposure was determined for each subject.
The data were then organized by exposure concentration range and
whether or not exposures included exercise. Within each exposure
category the fraction of subjects with increased airway responsiveness
was determined (see Table 41). For the total of 355 individual NO₂
exposures, 59% of the asthmatics had increased responsiveness. If the
response was not associated with NO₂ exposure, the fraction would be
expected to approach 50%. The excess increase in responsiveness can
be attributed primarily to the NO₂ exposures conducted at rest
(fraction was 69%). There was a larger fraction of increased
responsiveness during the resting exposures in all three concentration
ranges (see Table 41). In the exercising studies, however, there was
no effect because only 51% had an increase in airway responsiveness.
There was a trend for a slightly larger percentage (approx. 75%) of
subjects to have increased airway responsiveness after NO₂ exposures
above 376 µg/m³ (0.20 ppm) and under resting conditions. Of those
six studies independently reporting a statistically significant
response (Kleinman et al., 1983; Bylin et al., 1985, 1988; Bauer et
al., 1986; Mohsenin, 1987a; Joerres & Magnussen, 1990), four were
resting exposures, and in four the exposure duration was 30 min or
less. Although the authors offered various hypotheses for this

apparent effect of low-level NO₂ resting exposures, the mechanisms
are unknown. Changes in responsiveness were seen with relatively
brief exposures. One possible explanation for the absence of
response in the exercising exposures is that exercise-induced
bronchoconstriction may interfere with the NO₂-induced response or
that prior exercise may cause the airways to become refractory to the
effects of NO₂. Possible confounding influences of nitric oxide, not
measured in most studies, cannot be determined.

Table 41. Fraction of nitrogen dioxide-exposed subjects with
increased airway responsiveness^a

Nitrogen dioxide All Exposures Exposure
concentration exposures with exercise at rest
(ppm)

Asthmatics

0.05-0.20 0.64 (105)^b 0.59 (17) 0.65 (88)^b
0.20-0.30 0.57 (169) 0.52 (136) 0.76 (33)^b
> 0.30 0.59 (81) 0.49 (48) 0.73 (33)^c
All NO₂ 0.59 (355)^b 0.51 (202) 0.69 (154)^b
concentrations

Healthy

< 1.0 0.47 (36) 0.47 (36)
< 1.0 0.79 (29)^b 0.73 (15) 0.86 (14)^c

^a Data are fraction of subjects with an increase in airways
responsiveness above the value for clean air. Numbers in
parenthesis indicate actual number of subjects in each category.
Total number = 355. Ties (i.e. no change) were excluded.
^b p < 0.01 two-tailed sign test
^c p < 0.05 two-tailed sign test

A similar meta analysis for healthy subjects indicated increased
airway responsiveness after exposure to NO₂ concentrations greater
than 1880 µg/m³ (1 ppm). Exercise during exposure did not appear to
influence the responses as much in the healthy subjects as in the
asthmatics, but a similar trend was evident.

6.2.1.3 Nitrogen dioxide effects on patients with chronic obstructive
pulmonary disease

Patients with COPD represent an important potentially sensitive
population group. Studies evaluating NO₂ effects on respiratory
function in COPD subjects are summarized in Table 42. The results of
two NO₂ exposure studies (9400 to 15 040 µg/m³, 5 to 8 ppm NO₂ for
up to 5 min) were discussed by Von Nieding et al. (1980), who found
that the responses of bronchitics were generally similar to those of
healthy subjects. There was a tendency for the response to NO₂ to be
greater in the subjects with the highest baseline R_aw. Percentage
changes ranged from approximately 25 to 50%. In a review of their
studies, Von Nieding & Wagner (1979) showed that R_aw increased in
chronic bronchitics exposed to > 3760 µg/m³ (2.0 ppm) NO₂.

The responses of COPD patients were affected by exposure (with
mild exercise) to 564 µg/m³ (0.3 ppm) NO₂ for 3.75 h (Morrow &
Utell, 1989). Forced vital capacity showed progressive and
significant decreases during and following NO₂ exposure, the largest
change of -9.6% occurring after 3.75 h of exposure. Smaller
decrements in FEV₁ (-5.2%) occurred at the end of exposure. There was
no effect of NO₂ on SG_aw or diffusing capacity. The severity of
disease (based on impairment of lung function: FEV₁ < 60% predicted
vs. > 60% predicted) generally did not influence the magnitude of
response to NO₂. The COPD patients showed a decrement in FEV₁
compared to the healthy, elderly non-smokers who experienced an
improvement in FEV₁. In contrast, Linn et al. (1985a) found no
effects from a 1-h exposure (with exercise) to 940, 1880 and
3760 µg/m³ (0.5, 1.0 and 2.0 ppm) NO₂ in a diverse group of COPD
patients. The reasons for the marked difference in responses between
the two studies are not known. Ambient exposure to air pollution in
general and NO₂ in particular was probably much higher for the
subjects in the Linn et al. (1985a) study. Thus, attenuation of
physiological responses may have been a factor.

Hackney et al. (1992) studied effects of field exposure to
ambient air and chamber exposure to 564 µg/m³ (0.3 ppm) NO₂ in
older adults with evidence of COPD and a history of heavy smoking.
They reported only slight adverse effects of NO₂. The study did not
strongly confirm the findings of Morrow & Utell (1989) and Morrow et
al. (1992), and the authors speculated that ambient exposure history
may have been responsible for differences between these studies.

6.2.1.4 Age-related differential susceptibility

Studies evaluating possible age-related differences in
susceptibility to NO₂ effects on respiratory function in healthy
subjects are summarized in Table 39.

Research on asthmatics is summarized in Table 40. Spirometry
measurements of young (18 to 26 years old) and older (51 to 76 years
old) men and women were not affected by exposure to 1128 µg/m³
(0.6 ppm) NO₂ with light intermittent exercise (Drechsler-Parks et
al., 1987; Drechsler-Parks, 1987). In addition, Morrow & Utell (1989)
did not observe any pulmonary function or airway responsiveness
effects due to a lower level of NO₂ (564 µg/m³, 0.3 ppm) in young or
elderly healthy subjects.

Koenig et al. (1985) found no "consistent significant changes in
pulmonary functional parameters" after 1-h resting exposures of
asthmatic adolescents to 226 µg/m³ (0.12 ppm) NO₂. Subsequent
mouthpiece exposures to 226 µg/m³ NO₂, with exercise, caused
increases in R_T and decreases in FEV₁ after both air and NO₂
exposure, which were apparently due to exercise alone (Koenig et al.,
1987a,b). When subjects were exposed to a higher level of NO₂
(338 µg/m³, 0.18 ppm), no differences in R_T occurred. Decreases in
FEV₁ were -1.3 and -3.3% for air and NO₂, respectively; this
difference (p = 0.06) may indicate a possible response trend.

6.2.2 Nitrogen dioxide effects on pulmonary host defences and
bronchoalveolar lavage fluid biomarkers

Nitrogen dioxide can enhance susceptibility to infectious
pulmonary disease, as clearly demonstrated in the animal toxicological
literature (chapter 5). Epidemiological studies (chapter 7) suggest
similar effects. Human clinical studies of NO₂ effects on host
defences are summarized in Table 43.

Kulle & Clements (1988) and Goings et al. (1989) (two reports of
the same study) examined the effect of NO₂ exposure on susceptibility
to attenuated influenza virus. Healthy adults were exposed for
2 h/day for 3 days to either clean air or 1880, 3760 or 5640 µg/m³
(0, 1.0, 2.0 or 3.0 ppm) NO₂. The virus was administered
intranasally after the second day of exposure, and infectivity was
defined as the presence of virus in nasal washes, a rise in either
nasal wash or serum antibody titres to the virus, or both. Although
the rates of infection were elevated after NO₂ exposure in some of
the NO₂-exposed groups (91% of subjects exposed to 1880 or
3760 µg/m³ (1 or 2 ppm) infected vs. 71% of controls), the changes
were not significant. The investigators concluded that the results of
the study were inconclusive, rather than negative, because the
experimental design had a low power to detect a 20% difference in
infection rate, decreasing the possibility of statistical
significance.

Table 42. Effects of nitrogen dioxide on lung function and airway responsiveness of chronic obstructive pulmonary disease patients^a

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

564 0.3 225 21 25 13 M/7 F 47-70 years, Total NO₂ inhaled Morrow &
(3 × 7) 8 mild, dose 1.215 mg. Utell (1989)
12 moderate Decrease in FVC after
exposure-9.6%. 5.2%
decline in FEV₁
significant after
approx. 4-h exposure.

564 0.3 240 28 25 15 M/11 F 47-69 No significant change Hackney et al.
(4 × 7) in FVC or FEV₁ with (1992)
NO₂ exposure

940 0.5 120 15 25 7 24-53 years, No effects in Kerr et al.
daily cough bronchitics alone. (1979)
for 3 months Possible decrease in
quasistatic compliance.

940 0.5 60 30 16 13 M/9 F 48-69 years, No change in FVC, Linn et al.
some with FEV₁, etc. at any NO₂ (1985a)
1880 1.0 emphysema, level. SR_aw tended to
some with increase after first
3760 2.0 chronic exercise period.
bronchitis Possible decrease in
peak flow at
3760 µg/m³. No symptom
changes. No change
in S_aO₂.

Table 42 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

940-9400 0.5-5 15 88 Decrease in earlobe Von Nieding et
blood PO₂ at al. (1971, 1970)
> 7520 µg/m³.
Increased R_aw at
> 3008 µg/m³.

1880-9400 1-5 30 breaths 84 M 30-72 years, Increase in R_aw Von Nieding et
(15 min) chronic non- related to NO₂ al. (1973a)
specific disease concentration. No
effect on R_aw below
2820 µg/m³.

9400 5 60 Changes in PO₂ of
earlobe capillary
blood. Change occurred
in first 15 min,
effect did not
increase with further
exposure.

Table 42 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

1880-15 040 1-8 ppm 5-60 116 25-74 years At 7520-9400 µg/m³ Von Nieding &
for 15 min, P_aO₂ Wagner (1979)
decreased
(arterialized capillary
blood). R_aw increased
with exposure to
> 3008 µg/m³.

^a Modified from US EPA (1993)
Abbreviations: FVC = Forced vital capacity; FEV₁ = Forced expiratory volume in 1 second; P_aO₂ = Arterial partial pressure of oxygen;
PO₂ = Partial pressure of oxygen; R_aw = Airway resistance; SR_aw = Specific airway resistance; S_aO₂ = Arterial oxygen saturation

Table 43. Effects of nitrogen dioxide on host defences of humans^a

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

508 0.27 60 M Healthy, young No change in nasal or Rehn et al.
1993 1.06 tracheobronchial (1982)
clearance.

(1) 1128 (1) 0.6 180 60 39 6 M/2 F 30.3 ± 1.4 Total NO₂ uptake (1) Frampton et al.
years, healthy, 3.4 mg (2) 5.6 mg, (3) (1989b)
NS approx.3.3 mg (4)
8.1 mg. BAL fluid
analysis showed no
significant effect on
total protein or
albumin
(2) Var (2) Var 180 60 43 11 M/4 F 25.3 ± 1.2 Apparent increase in
(94 (0.05 years, healthy, alpha-2-macro-globulin
background background NS 3.5 h after exposure
with 3 × with 3 × 15 to 0.6 ppm (Group 1)
15 min at min at but not after the
3760) 2.0 ppm) other protocols. No
changes in percentage
of lymphocytes or
neutrophils. Concluded
that NO₂ at these
concentrations neither
(3) 1128 (3) 0.6 180 60 approx. 40 5 M/3 F 32.6 ± 1.6 altered epithelial
years, healthy, permeability nor
NS caused inflammatory
cell influx.

Table 43 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

(4) 2820 (4) 1.5 180 60 39 12 M/3 F 23.5 ± 0.7
years, healthy,
NS

1128 0.6 120/day 60 approx. 30-40 4 M/1 F 21-36 years, Slight increase in Boushey et al.
for 4 days Healthy, NS. circulating (venous) (1988) (Part 2)
FEV₁/FVC% lymphocytes:
range 73-83%, 1792 ± 544 per mm³
"normal" (post-NO₂) vs.
methacholine 1598 ± 549 per mm³
responsiveness (baseline). No change
in BAL lymphocytes
except an increase in
natural killer cells:
7.2 ± 3.1% (post-NO₂)
vs. 4.2 ± 2.4%
(baseline). No change
observed in IL-1
or TNF.

1128 0.6 180 60 approx. 40 7 M/2 F Healthy, NS No change in cell Frampton et al.
(6 × 10) recovery or (1989a)
differential counts.
Possible decrease in
macrophage
inactivation of virus

Table 43 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

94 with 0.05 with 135 60 11 M/4 F Nonreactive in vitro. Possible
3760 2.0 spikes 3 × 15 (6 × 10) (carbachol), no sensitive subgroup.
spikes recent upper
resp. infection

1880 1.0 180 Intermittent 3 M/5 F Healthy No responses. Jorres et al.
(1992)

1880 1.0 120/day 22 Healthy, NS, Study conducted over Goings et al.
3760 2.0 3 days 21, 22 seronegative 3-year period. NO₂ did (1989)
5640 3.0 22 not significantly
increase viral
infectivity, although
a trend was observed.
This study had a low
power to detect small
differences in
infection rate.

3760 2.0 240 120 50 10 Healthy, NS Increased bronchial Devlin et al.
PMN's and decreased (1992); Becker
macrophage phagocytosiset al. (1993)

3760 2.0 360 Intermittent 12 Healthy, NS Immediate and 18-h Frampton et al.
post-BAL increase (1992)
in PMN.

Table 43 (Con't)

NO₂ concentration Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

4230 2.25 20 20 approx. 35 8 Healthy, NS Increased levels of Sandstroem et
7520 4.0 8 mast cells in BAL al. (1989)
10 340 5.5 8 fluid at all
Total n = 18 concentrations.
Increased numbers of
lymphocytes at
> 7520 µg/m³
(BAL 24-h
post-exposure).

7520 4.0 20 min- 20 approx. 35 8 Healthy, NS Total cell counts Sandstroem et
alternate were reduced. Alveolar al. (1990a)
days for macrophages had
12 days enhanced phagocytic
activity but fewer
were present.
Decreased numbers
of mast cells, T and
B lymphocytes, and
natural killer cells
(BAL 24-h
post-exposure).

^a Modified from US EPA (1991)

Abbreviations: M = Male; F = Female; NS = Non-smoker; FEV₁ = Forced expiratory volume in 1 second; FVC = Forced vital capacity;
BAL = Bronchoalveolar lavage; IL-1 = Interleukin-1; TNF = Tumour necrosis factor; VAR = Variable
Others investigated the effects of NO₂ on cells and fluids in
bronchoalveolar lavage (BAL) of healthy adults. Frampton et al.
(1989a) used two different exposure protocols that had the same
concentration × time product. One group was exposed for 3 h to
1128 µg/m³ (0.6 ppm), whereas the other was exposed to a background
level of 94 µg/m³ (0.05 ppm) with three 15-min spikes of 3760 µg/m³
(2.0 ppm). Both exposures included exercise. Pulmonary function and
airway responsiveness were not affected. Alveolar macrophages (AM)
obtained by BAL after exposure to 1128 µg/m³ NO₂ tended to
inactivate virus less effectively than AM collected after air
exposure. The AMs that showed the impairment of virus inactivation
also showed an increase in interleukin-1 production, not seen in the
AMs from other subjects. Interleukin-1 is a proinflammatory protein
produced by AMs, which performs a number of immunoregulatory
functions, including induction of fibroblast proliferation, activation
of lymphocytes, and chemotaxis for monocytes. The study had
relatively low statistical power to detect an effect. Becker et al.
(1993) reported no change in virus inactivation properties of alveolar
macrophages lavaged from subjects exposed to 3760 µg/m³ (2 ppm) for
4 h.

Using exposures similar to the above, with the addition of two
groups exposed to 2820 µg/m³ (1.5 ppm) NO₂ for 3 h, one with BAL at
3.5 h post-exposure and the other with BAL at 18 h post-exposure,
Frampton et al. (1989b) examined changes in protein in BAL fluid. The
total protein and albumin content of BAL fluid obtained at either 3.5-
or 18-h post-exposure was not changed. In BAL fluid obtained 3.5 h
after exposure to 1128 µg/m³ (0.60 ppm) there was an increase in
alpha-2-macroglobulin, a regulatory protein that has antiprotease
activity and immunoregulatory effects. This response was not seen in
the group lavaged at 18 h post-exposure and no such effect occurred at
a higher NO₂ concentration (2820 µg/m³).

Sandstroem et al. (1989) exposed healthy subjects to 4230, 7520
and 10 340 µg/m³ (2.25, 4.0 and 5.5 ppm) for 20 min (with moderate
exercise) and performed BAL 24 h after exposure. Increased numbers of
mast cells were observed at all NO₂ concentrations; numbers of
lymphocytes were increased only at > 7520 µg/m³. In order to
determine the time course of this response, Sandstroem et al. (1990a)
exposed four groups of healthy subjects to 7520 µg/m³ NO₂ for 20 min
(mild exercise) and then performed BAL 4, 8, 24 or 72 h after
exposure. Increased numbers of mast cells and lymphocytes were
observed at 4, 8 and 24 h but not at 72 h. There was no change in the
numbers of AMs, eosinophils, polymorphonuclear leukocytes, T cells or
epithelial cells, or in the albumin concentration of lavage fluid.
The authors interpreted the increased numbers of mast cells and
lymphocytes as a nonspecific inflammatory response.

Sandstroem et al. (1990b) also evaluated responses to repeated
NO₂ exposures. Healthy subjects were exposed to 7520 µg/m³
(4.0 ppm) NO₂ for 20 min/day (with moderate exercise) on alternate
days over a 12-day period (seven exposures in all); BAL was performed

24 h after the last exposure. The first 20 ml of BAL fluid was
treated separately and presumed to represent primarily bronchial cells
and secretions; subsequent fractions presumably were from the alveolar
region. In the first fraction, there was a reduction in the numbers
of mast cells and AMs; AM phagocytic activity (on a per cell
basis) was increased. In addition, there were reduced numbers of
T-suppressor cells, B cells and natural killer (NK) cells in the
alveolar portion of the BAL. This pattern of cellular response
contrasts with that after single NO₂ exposure (Sandstroem et al.,
1990a).

Rubinstein et al. (1991) studied five healthy volunteers exposed
for 2 h/day for 4 days to 1128 µg/m³ (0.60 ppm) NO₂ with
intermittent exercise. A slight increase in circulating (venous
blood) lymphocytes was observed. The only change observed in BAL
cells was a modest increase in the percentage of NK cells, suggesting
a possible increase in immune surveillance.

Three recent studies examined the effects of longer exposures to
1880 or 3760 µg/m³ (1.0 to 2.0 ppm) NO₂ on lavaged cells and
mediators. Devlin et al. (1992) (also Becker et al., 1993) studied
healthy subjects exposed to 3760 µg/m³ NO₂ for 4 h with alternating
15-min periods of rest and moderate exercise. One of the main
findings after NO₂ exposure was that there was a three-fold increase
in PMNs in the first lavage sample, representing predominantly
bronchial cells and fluid. In addition, macrophages recovered from
the predominantly alveolar fraction showed a 42% decrease in ability
to phagocytose Candida albicans and a 72% decrease in release of
superoxide anion. In another study, Frampton et al. (1992) exposed
exercising subjects to 3760 µg/m³ NO₂ for 6 h. Bronchoalveolar
lavage was performed either immediately or 18 h after exposure.
There was a modest increase in lavage fluid PMN levels (< two-fold
increase) but no change in lymphocytes. Alveolar macrophage
production of superoxide anion was not altered in these subjects.
These two studies suggest that NO₂ exposure may induce a mild
bronchial inflammation and may also lead to impaired macrophage
function. On the other hand, Joerres et al. (1992) examined both
healthy and asthmatic subjects exposed to 1880 µg/m³ NO₂ for 3 h,
but observed no changes in cells or mediators in BAL fluid or in the
appearance of bronchial mucosal biopsies after this exposure. Neither
macrophage function nor a specific bronchial washing were examined in
this study.

Rehn et al. (1982) reported that a 1-h exposure to either 500 or
2000 µg/m³ (0.27 or 1.06 ppm) NO₂ did not alter nasal or
tracheobronchial mucociliary clearance rates.

6.2.3 Other classes of nitrogen dioxide effects

There have been isolated reports that higher levels of NO₂
(> 7520 µg/m³, 4.0 ppm) can decrease arterial oxygen partial
pressure (PaO₂) (Von Nieding & Wagner, 1977; Von Nieding et al.,
1979) and cause a small decrease in systemic blood pressure (Linn et
al., 1985b). However, the impact of such changes is not clear,
especially considering the high concentrations of NO₂ required.

The effects of NO₂ on the constituents of BAL fluid, blood and
urine have been examined in very few studies and are reviewed in more
detail elsewhere (US EPA, 1993). The general purpose of this research
was to examine mechanisms of pulmonary effects or to determine whether
NO₂ exposure could result in systemic effects. Investigations of the
effects of NO₂ on levels of serum enzymes and antioxidants have been
conducted, but few effects were found and they cannot be interpreted
(Posin et al., 1978; Chaney et al., 1981). For example, Chaney et al.
(1981) found an increase in glutathione levels, but Posin et al.
(1978), using a higher NO₂ concentration, did not find such an
effect. Studies of exposure to NO₂ concentrations between 2820 and
7520 µg/m³ (1.5 and 4.0 ppm) found either slight or no changes in BAL
levels of alpha-1-antitrypsin, which inhibits protease activity
(Mohsenin & Gee, 1987; Johnson et al., 1990; Mohsenin, 1991). Healthy
subjects exposed to 7520 µg/m³ NO₂ (Mohsenin, 1991) at rest for 3 h
showed increased lipid peroxidation products in BAL fluid obtained
immediately after exposure. In addition, the activity or the elastase
inhibitory capacity (EIC) of alpha-1-protease inhibitor (alpha-1-PI)
was decreased after NO₂ exposure. However, vitamin C supplementation
for 4 weeks prior to NO₂ exposure markedly attenuated the EIC
response and resulted in a lower level of lipid peroxidation products.
The author suggested that the reduced activity of alpha-1-PI may have
implications for the pathogenesis of emphysema, especially in smokers.
At a lower NO₂ concentration (3760 µg/m³, 2.0 ppm, for 4 h), Becker
et al. (1993) reported no change in alpha-1-antitrypsin. Potential
effects of NO₂ on collagen metabolism have been investigated by
examining urinary excretion of collagen metabolites after a 3-day
(4 h/day) exposure to 1128 µg/m³ (0.6 ppm) NO₂, but no effects were
found (Muelenaer et al., 1987).

6.3 Effects of other nitrogen oxide compounds

Relatively few controlled human exposure studies have been
conducted that evaluate NO_x species other than NO₂. Such studies
are summarized in Table 44 and concisely discussed here.

Table 44. Effects of other nitrogen oxide (NO_x) compounds on humans^a

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

HNO₂ 0.004 210 15 Healthy A dose-dependent Kjaergaard et
0.077 (11 M/4 F) 22-57 years vasodilation in al. (1993)
0.395 bulbar conjunctiva.
Significant increase
of polymorphonuclear
neutrophils, cuboidal
and squamous epithelium
cell counts in the
tear fluid

HNO₃

129 0.050 40 10 approx. 25-30 5 M/4 F 12-17 years, FEV₁ decreased -4.4% Koenig et al.
asthmatic after HNO₃ and -1.7% (1989a)
after HNO₃ plus air
exposure. R_T increased
+22.5% after HNO₃ and
+7.4% after air
exposure.

200 0.078 120 100 Mod. 4 M/1 F Healthy In BAL, increase in Becker et al.
AM phagocytosis and (1991)
AM infection
resistance.

500 0.194 240 240 40 10 Healthy No effect on FEV₁, Aris et al.
FVC, SR_aw or BAL (1991)
cells.

Table 44 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

1230 1.0 120 60 50 W 8 M 19-24 years Suggested change in Kagawa (1982)
density dependance of
expired flow.

12 300- 10-39 15 191 Healthy, Increase in total Von Nieding et
47 970 20-50 years respiratory resistance al. (1973b)
at > 24 600 µg/m³ and
a decrease in P_aO₂ at
> 18 450 µg/m³.

NH₄NO₃

200 (1.1 MMAD) 120 60 approx. 20 20 Normal No significant changes Kleinman et al.
19 Asthmatic due to NH₄NO₃ in (1980)
normals or asthmatics
except possible
decrease in RT.
No symptoms and
effects.

80 + 940 (0.55 MMAD) 240 30 55 12 Normal No effects. Stacy et al.
µg/m³ +0.5 ppm (1983)
NO₂ NO₂

Table 44 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

NaNO₃

10, 100, (0.2 MMAD) 10 5 Normal No effects. Sackner et al.
1000 5 Asthmatic (1979)

1000 6 Normal
6 Asthmatic

7000 (0.46 16 (× 2) 10 Normal No effects. Utell et al.
MMAD) 32 (total) 11 Mild asthmatics (1979)

7000 (0.49 16 (× 2) 11 Influenza No symptoms. SG_aw Utell et al.
MMAD) 32 (total) patients decrease 17% and V_E (1980)
max 40% TLC decreased
by 12% after nitrate,
within 2 days of onset
of illness. Similar
effect 1 week later
but not 3 weeks later.

^a Modified from US EPA (1993)
Abbreviations:

W = Watt; M = Male; P_aO₂ = Arterial partial pressure of oxygen; HNO₃ = Nitric acid; FEV₁ = Forced expiratory volume in 1 second;
FVC = Forced vital capacity; SR_aw = Specific airway resistance; BAL = Bronchoalveolar lavage; AM = Alveolar macrophage; F = Female;
R_T = Total respiratory resistance; NS = Not significant; MMAD = Mass median aerodynamic diameter; SG_aw = Specific airway conductance;
V_E max 40% TLC = Maximum expiratory flow at 40% of total lung capacity on a partial expiratory flow-volume curve
Von Nieding et al. (1973b) exposed healthy subjects and smokers
to 12 300 to 47 970 µg/m³ (10 to 39 ppm) NO for 15 min. Total
respiratory resistance increased significantly (approx. 10-12%) after
exposure to > 24 600 µg/m³ (> 20 ppm) NO. Diffusing capacity
was not changed, but a small decrease (7 to 8 torr) in PaO₂ was noted
between 18 450 and 36 900 µg/m³ (15 to 30 ppm). Kagawa (1982)
examined the effects of a 1230 µg/m³ (1 ppm) NO exposure for 2 h in
normal subjects. A few individuals had increases in SG_aw, and a few
had decreases. Analysis of the group mean data produced only one
apparently statistically significant change: an 11% decrease in flow
at 50% FVC in a helium-air mixture compared to this flow in air.
However, because the data were analysed by multiple t-tests the
results should be interpreted with this in mind.

NO is naturally formed in the body from the amino acid L-arginine
and performs a second messenger function in several organ systems. It
has been measured in expired air (Gustafsson et al., 1991) and causes
vasodilation in the pulmonary circulation. Recently, NO has been used
clinically to treat pulmonary hypertension in COPD patients and in
infants with persistent pulmonary hypertension of the newborn (Zapol
et al., 1994).

In healthy volunteers made hypoxic by breathing 12% oxygen in
nitrogen, the inhalation of 49 403 µg/m³ (40 ppm) NO prevented the
hypoxia-induced increase in pulmonary artery pressure (Frostell et
al., 1993). Systemic arterial pressure was not changed. No
evaluation of effects on lung function were performed. Adnot et al.
(1993) studied a group of COPD patients who had pulmonary artery
pressures averaging 32 mmHg. They breathed 6130 to 49 403 µg/m³
(5 to 40 ppm) NO for successive 10-min periods. There was a
dose-dependant decrease in pulmonary artery pressure during NO
inhalation and no alteration of systemic arterial pressure. Moinard
et al. (1994) observed a 20% drop in pulmonary artery pressure in COPD
patients after breathing 18 391 µg/m³ (15 ppm) NO for 10 min. Based
on an improvement in alveolar ventilation in some segments of the
lung, the authors postulated that NO may also act as a bronchodilator.
Hoegman et al. (1993) suggested a modest bronchodilator effect of
98 080 µg/m³ (80 ppm) NO. Based on findings in animals, which are
summarized in chapter 5, NO does cause bronchodilation at similar
concentrations (Barnes, 1993).

Nitrous acid and nitric acid may be formed from the reaction of
NO₂ with water. Nitrous acid may also be produced directly in the
combustion process.

Koenig et al. (1989a) examined the responses of adolescent
asthmatics to a 40-min exposure to 129 µg/m³ (0.05 ppm) HNO₃ vapour
via a mouthpiece exposure system. After 30 min of rest and 10 min of
exercise while breathing HNO₃, there was a 4.4% decrease in FEV₁
compared to a 1.7% decrease after air breathing. A 22.5% increase in
total respiratory resistance was also observed after HNO₃ exposure,
compared to a 7.4% increase after air breathing.

The effects of HNO₃ on BAL endpoints have been reported. Becker
et al. (1992) exposed healthy subjects to 200 µg/m³ (0.078 ppm) HNO₃
for 120 min, including 100 min of moderate exercise. Bronchoalveolar
lavage performed 18-h after exposure indicated increased phagocytic
activity of AMs and increased resistance to respiratory syncytial
virus infection. There were no changes in markers of tissue damage.
Aris et al. (1991) exposed healthy subjects to 500 µg/m³ (0.194 ppm)
HNO₃ for 4 h, including moderate exercise. No change in lactate
dehydrogenase levels, lavage fluid protein or differential cell counts
in the BAL were observed. Pulmonary function (FEV₁, FVC and SR_aw)
was not significantly affected.

Kjaergaard et al. (1993) studied the effects of nitrous acid on
the eyes of 15 healthy non-smokers exposed to 8, 148 or 758 µg/m³
(4, 77 or 395 ppb) for 3.5 h. There was an increase in trigeminal
sensitivity (CO₂ induced eye irritation) related to the concentration
of nitrous acid. Eye inflammation was increased, as indicated by
increased PMNs and epithelial cells in tear fluid.

Neither sodium nitrate (NaNO₃) nor ammonium nitrate caused
effects on pulmonary function of normal or asthmatic subjects (Sackner
et al., 1979; Utell et al., 1979; Kleinman et al., 1980; Stacy et al.,
1983). However, there was a decrease in airway conductance and in
PEFV curves in normal subjects with acute influenza exposed to
7 mg/m³ of NaNO₃ aerosol (Utell et al., 1980). This is several
orders of magnitude above the nitrate concentrations found in most
ambient air.

6.4 Effects of nitrogen dioxide/gas or gas/aerosol mixtures on lung
function

Table 45 summarizes studies of human subjects exposed to
NO₂-containing pollutant mixtures. Most of the studies have been
limited primarily to spirometry and plethysmography. More extensive
discussion can be found in US EPA (1993).

With a few exceptions (to be discussed below), most research on
interactions either showed no effects of the individual pollutants or
the mixture, or it indicated that NO₂ did not enhance the effects of
the other pollutant(s) in the mixture (Table 45). Most attention has
focussed on NO₂ mixtures with ozone (O₃), although combinations with
SO₂, NO, particles, and a mixture of SO₂ plus O₃ have also been
tested. Due to the varied exposure protocols in the database, no
consistent physiological trends are evident. The generally negative
responses could either reflect a true lack of interaction or other
important design considerations. For example, asthmatics were not
studied. Because pulmonary function studies of NO₂ alone cause
variable effects with no clear concentration-responses, detecting
interactions would be expected to be difficult unless there was
significant synergism.

Table 45. Effects of nitrogen dioxide mixtures on healthy subjects^a

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

75 NO₂ 0.04 NO₂ 60 60 56 42 M/8 F Healthy No apparent effect Avol et al.
(Amb) over and above that (1983)
of O₃ alone.

75 NO₂ 0.04 NO₂ 60 60 22.4 33 M/33 F Children, No effects of ambient Avol et al.
(Amb) 8-11 years air exposures. (1985a, 1987)

103 NO₂ 0.055 NO₂ 60 60 32 46 M/13 F Adolescents, Ambient air exposures Avol et al.
(Amb) 12-15 years effect attributed (1985b)
to O₃.

132 NO₂ 0.07 NO₂ 120 60 approx. 20 14 M/20 F 29 years Small decreases in Linn et al.
(Amb) FVC, FEV₁, in ambient (1980b)
air mostly attributable
to O₃. No association
of NO₂ levels with lung
function change.

545 NO₂ (a) 0.29 NO₂ 240 (2 120 approx. 20 4 Healthy With each group, Hackney et al.
+980 O₃ +0.50 O₃ consecutive minimal alterations (1975b)
days of in pulmonary function
545 NO₂ (b) 0.29 NO₂ exposure caused by O₃ exposure.
+980 O₃ +0.50 O₃ to each Effects were not
+34 350 +30.0 CO mixture) increased by addition
CO of NO₂ or NO₂ plus CO
to test atmospheres.

Table 45 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

545 NO₂ (a) 0.29 NO₂ 120 (2 60 approx. 20 7 Healthy Little or no change Hackney et al.
+490 O₃ +0.25 O₃ consecutive in pulmonary function (1975b)
days of found with O₃ alone.
545 NO₂ (b) 0.29 NO₂ exposure) Addition of NO₂ or of
+490 O₃ +0.25 O₃ NO₂ plus CO did not
+34 350 +30.0 CO noticeably increase
CO the effect. Seven
subjects included;
some believed to be
unusually reactive
to respiratory
irritants.

940 NO₂ 0.50 NO₂ 120 30 40 10 M Young adults, FEV₁, decreased 8-14%. Folinsbee
+980 O₃ +0.5 O₃ NS No differences between et al. (1981)
O₃ plus NO₂ and O₃
alone.

1128 NO₂ 0.60 NO₂ 120 60 25 8 M/8 F 18-26 years, No significant Drechsler-Parks
+882 O₃ +0.45 O₃ NS changes attributable (1987)
to NO₂.

8 M/8 F 51-76 years Tendency (p > 0.05)
8 M/8 F 51-76 years for NO₂ plus O₃ to be
greater than O₃ alone.

1128 NO₂ 0.60 NO₂ 60 60 70 20 M Healthy No additional effect Adams et al.
+588 O₃ + 0.30 O₃ 50 20 F of NO₂ over and above (1987)
effect of O₃.

Table 45 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

1128 0.60 ppm NO₂ 120 60 40 21 F Healthy, NS NO₂ exposure increased Hazucha et al.
NO₂ airway responses to (1994)
0.3 ppm O₃ 120 60 40 methacholine after a
(3 h later) subsequent O₃ exposure.

282 NO₂ 0.15 NO₂ 120 60 approx. 25 6 M Some Possible small Kagawa (1986)
+294 O₃ + 0.15 O₃ smokers decrease in SG_aw.
+200 + H₂SO₄
H₂SO₄

282 NO₂ 0.15 NO₂ 120 60 approx. 25 3 M Some Possible small
+294 O₃ + 0.15 O₃ smokers decrease in FEV₁.
+393 SO₂ + 0.15 SO₂
+200 + H₂SO₄
H₂SO₄

564 NO₂ 0.30 NO₂ 120 20 approx. 25 6 M Some Possible small
+588 O₃ +0.30 O₃ smokers decrease in SG_aw.
+200 + H₂SO₄
H₂SO₄

282 NO₂ 0.15 NO₂ 120 60 approx. 25 7 M 19-23 years No significant Kagawa
+294 O₃ +0.15 O₃ enhancement of the (1983a,b)
+393 SO₂ +0.15 SO₂ effects of O₃ and/or
SO₂ by presence of
NO₂.

Table 45 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

301 NO₂ 0.16 NO₂ 480 0 15 16-26 years No change in FVC, Islam &
+157 O₃ +0.08 O₃ acetylcholine airway Ulmer (1979b)
+891 SO₂ +0.34 SO₂ reactivity.

564 NO₂ 0.3 NO₂ 120 60 6 F 19-25 years No significant effects Kagawa (1990)
+738 NO +0.6 NO NS on pulmonary function
or airway
responsiveness to
acetylcholine.

940 NO₂ 0.50 NO₂ 135 60 approx. 20 11 M/9 F 20-53 years No effects on Kleinman
1310 SO₂ + 0.5 SO₂ function; possible et al. (1985)
+26 + symptom responses.
Zn(NH₄)₂ Zn(NH₄)₂ NO₂ effects not
(SO₄)₂ (SO₄)₂ discernible from
+330 NaCl + NaCl mixture.

940 NO₂ 0.50 NO₂ 120 60 approx. 20 10 M/14 F 26 ± 4 No significant effect Linn et al.
1310 SO₂ + 0.50 SO₂ years, 21 NS, on lung function in (1980a)
3 S normals. Trend for a
slight decrease in
FVC after combined
exposure.

Table 45 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

7520-9400 4-5 NO₂ 10 5 M 21-40 years, Time course of Abe (1967)
NO₂ +4-5 SO₂ 4 NS, 1 S response different.
+4920-6150 SO₂ alone had immediate
SO₂ increase in resistance;
NO₂ had delayed
increase. Mixture had
intermediate effects
on resistance.

9400 NO₂ 5.0 NO₂ 120 60 ? 8 M < 30 years FVC (-5%), FEV_1.0 Islam &
+1960 O₃ +0.1 O₃ 8 M 30-40 years (-11.7%), decreased Ulmer (1979a)
+13 100 SO₂ +5.0 SO₂ 8 M > 49 years with exercise exposure
to this mixture
in < 30 years group.

9400 NO₂ 5.0 NO₂ 120 intermittent 9 M Healthy, No interaction on P_aO₂ Von Nieding
+196 O₃ +0.1 O₃ 20-38 years or R_T et al. (1977)
+13 100 SO₂ +5.0 SO₂

9400 NO₂ 5.0 NO₂ 120 intermittent 11 M Healthy, No interaction on P_aO₂
+196 O₃ +0.1 O₃ 20-38 years, or Rt
25 S, 9 NS

9400 NO₂ 5.0 NO₂ 120 60 approx. 20 23-38 years, R_T increased from 1.5 Von Nieding
+196 O₃ + 0.1 O₃ (70 W) two atopic to 2.4 (p < 0.01); et al. (1979)
+13 100 + 5.0 SO₂ questionable decrease
SO₂ in P_aO₂ (8 torr).

Table 45 (Con't)

Concentrations Exposure Exercise Exercise Number of Subject Effects Reference
duration duration ventilation subjects/ characteristics
(min) (min) (litres/min) gender
µg/m³ ppm

188 NO₂ 0.1 NO₂ 120 60 approx. 20 23-38 years, No effects.
+786 SO₂ +0.3 SO₂ two atopic

^a Modified from US EPA (1993)
Abbreviations:

Amb = Ambient air; CO = Carbon monoxide; F = Female; FEV₁ = Forced expiratory volume in 1 second; FEV_1.0 = Forced expiratory volume
in 1 second; FVC = Forced vital capacity; H₂SO₄ = Sulfuric acid; M = Male; NaCl = Sodium chloride; (NH₄)₂SO₄ = Ammonium sulfate;
NO = Nitric oxide; NS = Non-smoker; O₃ = Ozone; P_aO₂ = Arterial partial pressure of oxygen; R_T = Total respiratory resistance;
S = Active smoker; SG_aw = Specific airway conductance; SO₂ = Sulfur dioxide; W = Watts; ZnSO₄ = Zinc sulfate
Abe (1967) studied brief exposures to NO₂-SO₂ mixtures. Both
gases were at 4 to 5 ppm (i.e., 7520 to 9400 µg/m³ NO₂ and 4920 to
6150 µg/m³ SO₂). The effects were additive, with both gases causing
bronchoconstriction. Independently, the effect of SO₂ was immediate
and short-lasting, whereas the effect of NO₂ was delayed and more
persistent. The effect of the mixed gases was intermediate between
the two independent responses. Kagawa (1983a,b) reported that the
interaction of 282 µg/m³ (0.15 ppm) NO₂ plus 393 µg/m³ (0.15 ppm)
SO₂ in normal subjects exposed for 2 h with light intermittent
exercise caused an increase in SG_aw. However, because a large number
of repeated t-tests with an alpha level of 0.05 were used, it is
possible that the responses were due to chance.

The Rancho Los Amigos group (Linn et al., 1980b; Linn & Hackney,
1983; Avol et al., 1983, 1985a, 1987) conducted several studies of
NO₂-containing ambient air mixtures. The mean NO₂ level in the
ambient air (from the Los Angeles Air Basin) ranged from 75 to
132 µg/m³ (0.04 to 0.07 ppm). Normal and asthmatic adults,
adolescents and children were exposed for approximately 2 h during the
summer smog seasons of 1978 to 1984. The various pulmonary function
effects observed (see Table 45) were attributed to O₃. However, in
another study, Hazucha et al. (1994) found that ozone-induced
increases in airway responsiveness to methacholine were enhanced by
prior (3 h earlier) exposure to 1128 µg/m³ (0.60 ppm) NO₂. There
was also a slightly greater FEV₁ decrement after the NO₂-O₃
sequence.

There has been one study on the effects of HNO₃ vapour in
combination with O₃ (Aris et al., 1991). Ten healthy men were
exposed (with moderate exercise) to 430 µg/m³ HNO₃ for 2 h and then,
after 1 h, to 392 µg/m³ (0.20 ppm) O₃ for 3 h. No changes were
observed in FVC, FEV₁ or SR_aw after HNO₃ exposure. Ozone exposure
caused increased SR_aw and decreased FVC and FEV₁. Prior exposure to
HNO₃ vapour rather than air resulted in somewhat smaller changes in
lung function after ozone exposure. Clearly HNO₃ did not potentiate
responses to ozone.

6.5 Summary of controlled human exposure studies of oxides of
nitrogen

Human responses to a variety of oxidized nitrogen compounds have
been evaluated. By far, the largest database and the one most
suitable for risk assessment is that available for controlled
exposures to NO₂. The database on human responses to NO, nitric acid
vapour, nitrous acid vapour and inorganic nitrate aerosols is not as
extensive. A number of sensitive or potentially sensitive subgroups
have been examined, including adolescent and adult asthmatics, older
adults, and patients with chronic obstructive pulmonary disease and
pulmonary hypertension. Exercise increases the total uptake and

alters the distribution of the inhaled material within the lung. The
proportion of NO₂ deposited in the lower respiratory tract is also
increased by exercise. This may increase the effects of the above
compounds in people who exercise during exposure.

As is typical with human biological response to inhaled particles
and gases, there is variability in the biological response to NO₂.
Healthy individuals tend to be less responsive to the effects of NO₂
than individuals with lung disease. Asthmatics are clearly the most
responsive group to NO₂ that has been studied to date. Individuals
with chronic obstructive pulmonary disease may be more responsive than
healthy individuals, but they have limited capacity to respond to NO₂
and thus quantitative differences between COPD patients and others are
difficult to assess. There is not sufficient information available at
present to evaluate whether age or gender should be considered in the
risk evaluation.

NO₂ causes decrements in lung function, particularly increased
airway resistance in resting healthy subjects at 2-h concentrations as
low as 4700 µg/m³ (approx. 2.5 ppm). Available data are insufficient
to determine the nature of the concentration-response relationship.

NO₂ exposure results in increased airway responsiveness to
broncoconstrictive agents in exercising healthy, non-smoking subjects
exposed to concentrations as low as 2800 µg/m³ (approx. 1.5 ppm) for
exposure durations of 1 h or longer.

Exposure of asthmatics to NO₂ causes, in some subjects,
increased airway responsiveness to a variety of provocative mediators,
including cholinergic and histaminergic chemicals, SO₂ and cold air.
The presence of these responses appears to be influenced by the
exposure protocol, particularly whether or not the exposure includes
exercise. These responses may begin at concentrations as low as
380 µg/m³ (0.20 ppm). A meta analysis suggests that effects may
occur at even lower concentrations. However, no concentration-
response relationship is observed between 350 and 1150 µg/m³
(approx. 0.2 and 0.6 ppm).

Modest increases in airway resistance may occur in patients with
COPD from brief exposure (15-60 min) to concentrations of NO₂ as low
as 2800 µg/m³ (approx. 1.5 ppm) and decrements in spirometric
measures of lung function (3 to 8%) change in FEV₁ may also be
observed with longer exposures (3 h) to concentrations as low as
600 µg/m³ (approx. 0.3 ppm).

Exposure to NO₂ at levels above 2800 µg/m³ (approx. 1.5 ppm)
may alter numbers and types of inflammatory cells in the distal
airways or alveoli. NO₂ may alter the function of cells within the
lung and production of mediators that may be important in lung host

defences. The constellation of changes in host defences, alterations
in lung cells and their activities, and changes in biochemical
mediators is consistent with the epidemiological findings of increased
host susceptibility associated with NO₂ exposure.

In studies of mixtures of NO₂ with other pollutants, NO₂ has
not been observed to increase responses to other co-occurring
pollutant(s) beyond what would be observed for the other pollutant(s)
alone. A notable exception is the observation that pre-exposure to
NO₂ enhances the ozone-induced change in airway-responsiveness in
healthy, exercising subjects during a subsequent ozone exposure. This
observation suggests the possibility of delayed or persistent
responses to NO₂.

Within an NO₂ concentration range that may be of interest with
regard to risk evaluation (i.e., 100-600 µg/m³), the characteristics
of the concentration-response relationship for acute changes in lung
function, airway responsiveness to bronchoconstricting agents, or
symptoms cannot be determined from the available data.

NO is acknowledged as an important endogenous second messenger
within several organ systems. Inhaled NO concentrations above
6000 µg/m³ (approx. 5 ppm) can cause vasodilation in the pulmonary
circulation without affecting the systemic circulation. The lowest
effective concentration is not established. Information on pulmonary
function and lung host defences consequent to NO exposure are too
limited for any conclusions to be drawn at this time. Relatively high
concentrations (> 40 000 µg/m³) have been used in clinical
applications for brief periods (< 1 h) without reported adverse
reactions.

Nitric acid levels in the range of 250-500 µg/m³ (100-200 ppb)
may cause some pulmonary function responses in adolescent asthmatics,
but not in healthy adults.

Limited information on nitrous acid suggests that it may cause
eye inflammation at 760 µg/m³ (0.40 ppm). There are currently no
published data on human pulmonary responses to nitrous acid.

Limited data on inorganic nitrates suggest that there are no
lung function effects of nitrate aerosols with concentrations of
7000 µg/m³ or less.

7. EPIDEMIOLOGICAL STUDIES OF NITROGEN OXIDES

7.1 Introduction

This chapter discusses epidemiological evidence regarding effects
of NO_x on human health. Primary emphasis is placed on assessment of
the effects of NO₂ because it is the oxide of nitrogen measured in
most epidemiological studies and the one of greatest concern from a
public health perspective. Human health effects associated with
exposure to NO₂ have been the subject of several literature reviews
since 1970 (National Research Council, 1971, 1977; US EPA, 1982a,
1993; Samet et al., 1987, 1988). Oxides of nitrogen have also been
reviewed previously by the World Health Organization (WHO, 1977),
which presented a comprehensive review of studies conducted up to
1977. This chapter focuses on studies conducted since 1977, while
also using some key information from earlier literature, as reviewed
in more detail by US EPA (1993).

The studies discussed in this chapter are those that provide
useful quantitative information on exposure-effect relationships for
health effects associated with levels of NO₂ likely to be encountered
in the ambient air. In addition, some studies that do not provide
quantitative information are briefly discussed in the text in order to
help elucidate particular points concerning the health effects of
NO₂.

7.2 Methodological considerations

Key epidemiological studies on NO₂ health effects are evaluated
below for several factors of importance for interpreting their results
(US EPA, 1982a,c). Such factors include: (1) exposure measurement
error; (2) misclassification of the health outcome; (3) adjustment
for covariates; (4) selection bias; (5) internal consistency; and
(6) plausibility of the effect based on other evidence.

7.2.1 Measurement error

Measurement error regarding exposure may be a major problem in
epidemiological studies of NO₂. Ideally, personal monitors should be
placed on all subjects for the entire period of a study, but this is
often not feasible. Moreover, personal monitoring may not overcome
measurement error altogether. For example, the monitors that are
presently available do not accurately measure short-term peaks or
long-term chronic exposures. Other means of estimating NO₂ exposure
include source description, in-home monitors and fixed-site outdoor
monitors. These approaches are generally cheaper than personal
monitors but may be subject to greater measurement error, both random
(non-systematic) and systematic.

In general, a measurement error in estimation of exposure that is
independent of the health outcome will result in underestimation of
associations between exposure and dichotomous health outcomes (Samet &
Utell, 1990). Whittlemore & Keller (1988) examined the data of Melia
et al. (1980) and showed that a 20% misclassification rate of the
exposure category could result in an underestimate of the logistic
regression coefficient by as much as 50%. Even when exposure
measurement error is not independent of the outcome, measures of
association are biased towards the null, unless the probability of the
health outcome is very close to 0 or 1 (Stefanski & Carroll, 1985).

At present, there is little information on the relative
importance of peak and average NO₂ levels as causes of respiratory
effects in humans. In most homes and outdoor settings, peak values may
be related to average values, and reduction of peaks may lower
time-weighted averages. However, if health effects are largely
associated with the peak levels of NO₂, then the use of averages as
the sole guide to exposures will increase measurement error.

NO₂ may act as a precursor for other biologically active
substances (such as nitrous acid). If these agents are responsible
for some or all of the observed respiratory effects, then measurement
of NO₂ will provide an imprecise estimate of the effective dose.

7.2.2 Misclassification of the health outcome

Misclassification of the health outcome can occur whether the
outcome is continuous, (such as a measure of pulmonary function) or
dichotomous (such as the presence or absence of respiratory symptoms).
Lung function is typically measured with spirometry, a well-
standardized technique (Ferris, 1978). The measurement errors of the
instruments collecting the data have also been carefully estimated,
and random errors will simply add to the error variance. On the other
hand, respiratory symptoms and health status are usually measured by a
questionnaire. Responses to symptom questions will be correlated and
will depend on the interpretation of the respondent. As noted below,
a specific respiratory disease is likely to be reflected by a
constellation of symptoms. Therefore, it is appropriate to consider
aggregate, as well as single, specific symptom reports. Obviously,
questionnaire measurements involving recent recall are better than
those based on recall of events occurring several years earlier.
Questionnaires for cough and phlegm production have been standardized,
e.g., the British Medical Research Council (BMRC) questionnaire
(American Thoracic Society, 1969) and revisions of that questionnaire
(Ferris, 1978; Samet, 1978). These questionnaires and modifications
of them have been used extensively.

7.2.3 Adjustment for covariates

It is common when analysing a data set to discover that one or
more key covariates for the analysis were not measured. Schenker et
al. (1983) discussed socioeconomic status, passive smoking and gender
as important covariates in childhood respiratory disease studies.
Other covariates often of importance are age, humidity and other
co-occurring pollutants (e.g., particulate matter). The concern is
that, had missing covariates been measured, the estimate of the
regression coefficient of a variable of interest would have been
significantly different. Although the problem is faced by most
investigators, literature on the subject is sparse. For example,
Kupper (1984) showed that high correlations between the variables just
described will result in "unreliable parameter estimates with large
variances". Gail (1986) considered the special case of omitting a
balanced covariate from the analysis of a cohort study and concluded
that: "In principle, the bias may be either toward or away from zero,
though in more important examples - the bias is toward zero. In
important applications with additive or multiplicative regression,
there is no bias". Neither report provided information on how to
attempt to correct for the bias or on approaches for investigating the
possible bias in a given situation.

Most studies of respiratory disease and NO₂ exposure discussed
here measured important covariates such as age, socioeconomic level of
the parents, gender and parental smoking habits. The estimated effect
(regression coefficient of disease on NO₂ exposure) will be
overestimated if a missing covariate is positively or negatively
correlated with both exposure and health outcome. The estimated
effect will be underestimated if positively correlated with exposure
or outcome and negatively correlated with the other. Ware et al.
(1984) found that parents with some college education were more likely
to report respiratory symptoms and less likely to use a gas stove,
leading to an underestimate of the health effect, if education were
omitted from the analysis.

7.2.4 Selection bias

The possibility of selection bias, although a concern of every
study, seems very low for NO₂ epidemiological studies. Selection
bias would require selection of participants based on exposure
(e.g., use of gas stove) and also health outcome. Because most
epidemiological studies of these exposures are population based, there
is little possibility of selection based on health end-points.
Nevertheless, the loss of subjects by attrition associated with both
exposure and health studies must be considered.

7.2.5 Internal consistency

Internal consistency is also a useful check on the validity of a
study, but authors often do not report sufficient detail to check for
such consistency. For example, in the case of known risk factors for
respiratory effects, a study should find the anticipated associations
(e.g., passive smoking with increased respiratory illness or with more
wheeze in asthmatic children), and certain patterns of age or gender
effects should be observed. Consistency between studies also provides
an indication of the overall strength of the database.

7.2.6 Plausibility of the effect

Health outcomes should be ones for which there are plausible
bases to suspect that NO₂ exposure could contribute to such effects.
Two health outcome measures have been most extensively considered
in the epidemiological studies: lung function measurements and
respiratory illness occurrence. Human clinical and animal
toxicological studies have not indicated a demonstrated effect on lung
function at ambient levels in normal subjects. On the other hand,
human clinical and animal toxicological studies have shown that NO₂
exposure can impair components of the respiratory host defence system,
resulting in increased susceptibility of the host to respiratory
infection. Thus, reported increases in respiratory symptoms and
disease among children in epidemiological studies of NO₂ exposure can
be plausibly hypothesized to reflect an increase in respiratory
infection.

Each study is subsequently reviewed with special attention given
to the above factors. Those studies that address these factors most
appropriately provide a stronger basis for the conclusions that they
draw. Consistency between studies indicates the level of the strength
of the whole database.

7.3 Studies of respiratory illness

Respiratory illness and factors determining its occurrence and
severity are important public health concerns. The possible
association of NO₂ exposure with respiratory illness is of public
health importance because both the potential for exposure to NO₂ and
childhood respiratory illness are common (Samet et al., 1983; Samet &
Utell, 1990). This takes on added importance because recurrent
childhood respiratory illness (independent of NO₂) may be a risk
factor for later susceptibility to lung damage (Samet et al., 1983;
Glezen, 1989; Gold et al., 1989). The epidemiological studies
relating NO₂ exposure to respiratory illness are discussed in
sections 7.3.1 and 7.3.2.

7.3.1 Indoor air studies

In this section, studies that meet criteria for use in a
quantitative analysis are presented. Firstly, studies conducted by
Melia and colleagues in the United Kingdom are discussed. This is
followed by an evaluation of two large studies conducted in six cities
in the USA. Several other quantitative studies conducted by different
authors in various countries and cities are then presented. These are
followed by discussion of some additional recent large-scale studies
that yield useful quantitative information, e.g., a study of NO₂
relationship to respiratory disease in young children in Albuquerque,
New Mexico, USA. Lastly, other studies that provide information
concerning respiratory illness are also discussed.

7.3.1.1 St Thomas' Hospital Medical School Studies (United Kingdom)

Results of several British studies have been reported by Melia
et al. (1977, 1978, 1979, 1980, 1982a,b, 1985, 1988), Goldstein et al.
(1979, 1981), and Florey et al. (1979, 1982). Parts of these studies
were reviewed previously (US EPA, 1982a), but their importance
requires a more complete discussion of them.

The initial study (Melia et al., 1977) was based on a survey of
5658 children (excluding asthmatics, thus 100 less than the number
reported), aged 6 to 11 years, with sufficient questionnaire
information in 28 randomly selected areas of England and Scotland. A
self-administered questionnaire was completed by a parent to obtain
information on the presence of morning cough, day or night cough,
colds going to chest, chest sounds of wheezing or whistling, and
attacks of bronchitis. The questionnaire, distributed in 1973, asked
about symptoms during the previous 12 months. Colds going to the
chest accounted for the majority of symptoms reported. Information
about cooking fuel (gas or electric), age, gender and social class
(manual versus non-manual labour) was obtained, but there were no
questions about parental smoking. Melia et al. (1977) noted that
although they could not include family smoking habits in the analysis,
the known relation between smoking and social class (Tobacco Research
Council, 1976) allowed them to avoid at least some of the potential
bias from this source. It seemed unlikely that, within the social
class groups studied, there was a higher prevalence of smoking in
homes where gas was used for cooking. No measurements of NO₂, either
indoors or outdoors, were given.

The authors presented their results in the form of a contingency
table for non-asthmatics with complete covariate information. Table 46
is a summary of that data for non-asthmatic children. The authors
indicated that there was a trend for increased symptoms in homes with
gas stoves, but the increase was only significant for girls in urban
areas. The authors gave no measures of increased risk. The data in

Table 46 have been reanalysed using a multiple logistic model as shown
in Table 47. Because it had been suggested that gender had an effect
on the relationship with "gas cooker", interaction terms for gender
were included in the original model. None of these proved to be
significant, and they were subsequently dropped from the model. When
separate terms for each gender were used for the effect of "gas
cooker", an estimated odds ratio of 1.25 was obtained for boys and an
odds ratio of 1.39 was obtained for girls. The combined odds ratio
for both genders was 1.31 (95% confidence limits of 1.16 and 1.48) and
was statistically significant (p < 0.0001). The other main effects
of gender, SES and age were all statistically significant. This
reanalysis suggests that gas stove use was associated with an
estimated 31% increase in the odds of children having respiratory
illness symptoms.

Melia et al. (1979) reported further results of a national survey
covering a new cohort of 4827 boys and girls, aged 5 to 10 years, from
27 randomly selected areas that were examined in 1977. The study
collected information on the number of smokers in the home. In the
1977 cross-sectional study, only prevalence of day or night cough in
boys (p approx. or equal 0.02) and colds going to the chest in girls
(p < 0.05) were found to be significantly higher in children from
homes where gas was used for cooking compared with children from homes
where electricity was used. As shown in Table 48, grouping responses
according to the six respiratory questions into (1) none or (2) one or
more symptoms or diseases yielded a prevalence higher in children
from homes where gas was used for cooking than in those from homes
where electricity was used (p approx. or equal 0.01 in boys,
p = 0.07 in girls). The effects of gender, social class, use of pilot
lights and number of smokers in the house were examined.

The reanalysis of the data in Table 48, applying a multiple
logistic model, is given in Table 49. This model contained the same
terms as the analysis in Table 47. As in the previous analysis, none
of the interaction terms proved to be significant, and they were
subsequently dropped from the model. When separate terms for each
gender were used for the effect of "gas cooker", an estimated odds
ratio of 1.29 was obtained for boys and an odds ratio of 1.19 was
obtained for girls. The combined odds ratio for both genders was
1.24 (95% confidence limits of 1.09 and 1.42). This effect was
statistically significant (p < 0.0002). The other main effects of
gender, SES and age were all statistically significant. This
reanalysis suggests that gas stove use in this study is associated
with an estimated 24% increase in the odds of having symptoms.

Table 46. Symptom rates of United Kingdom children by age, gender,
social class and type of cooker^a

Social classes I-IIIa Social classes IIIb-V
(non-manual) (manual)
Electric Gas Electric Gas

Age < 8 years

Boys 25.6% 26.1% 29.9% 37.5%
(203) (88) (375) (309)

Girls 22.2% 30.4% 31.8% 33.5%
(171) (112) (393) (337)

Age 8 to 11 years

Boys 20.8% 23.3% 25.0% 29.0%
(365) (189) (675) (654)

Girls 18.1% 19.2% 17.8% 27.8%
(303) (187) (674) (623)

^a Numbers in parentheses refer to number of subjects; source:
Melia et al. (1977)

Table 47. Multiple logistic analysis of data from the study of Melia
et al. (1977)

Factor^a Odds ratio 95% Confidence p value
interval

SES and age by gender
interactions (2 d.f.) 0.2922

Gas by gender
interaction (1 d.f.) 0.3953

Gas cooker 1.31 1.16-1.48 < 0.0001

Gender (female) 0.86 0.76-0.97 0.0121

SES (manual) 1.31 1.14-1.51 0.0001

Age (< 8 years) 1.47 1.30-1.66 < 0.0001

^a SES = Socioeconomic status; d.f. = Degrees of freedom

Table 48. Unadjusted rates of one or more symptoms among United
Kingdom children by age, gender, social class and type of
cooker^a

Social classes I-IIIa Social classes IIIb-V
(non-manual) (manual)
Electric Gas Electric Gas

Age < 8 years
Boys 27.4% 31.7% 32.8% 36.7%
(277) (145) (485) (313)

Girls 24.4% 27.6% 27.8% 36.3%
(291) (134) (497) (336)

Age 8 to 11 years
Boys 19.2% 28.3% 23.6% 26.9%
(286) (113) (501) (338)

Girls 14.8% 18.6% 21.5% 18.5%
(243) (118) (437) (313)

^a Numbers in parentheses refer to number of subjects; source:
Melia et al. (1979)

Table 49. Multiple logistic analysis of data from the study of Melia
et al. (1979)

Factor^a Odds ratio 95% Confidence p value
interval

SES and age by gender
interactions (2 d.f.) 0.5749

Gas by gender
interaction (1 d.f.) 0.5566

Gas cooker 1.24 1.09-1.42 < 0.0001

Gender (female) 0.82 0.72-0.94 0.0030

SES (manual) 1.25 1.08-1.45 0.0034

Age (< 8 years) 1.69 1.48-1.93 < 0.0001

^a SES = Socioeconomic status; d.f. = Degrees of freedom

In 1978, 808 schoolchildren (Melia et al., 1980), aged 6 to
7 years, were studied in Middlesborough, an urban area of northern
England. Respiratory illness was defined as in the previous study.
Weekly indoor NO₂ measurements were collected from 66% of the homes,
the remaining 34% refusing to participate. NO₂ was measured weekly
by triethanolamine diffusion tubes (Palmes tubes) attached to walls in
the kitchen area and in the children's bedrooms. In homes with gas
stoves, weekly levels of NO₂ in kitchens ranged from 10 to 596 µg/m³
(0.005 to 0.317 ppm) with a mean of 211 µg/m³ (0.112 ppm) and levels
in bedrooms ranged from 8 to 318 µg/m³ (0.004 to 0.169 ppm) with a
mean of 56 µg/m³ (0.031 ppm). In homes with electric stoves, weekly
levels of NO₂ in kitchens ranged from 11 to 353 µg/m³ (0.006 to
0.188 ppm) with a mean of 34 µg/m³ (0.018 ppm), and levels in
bedrooms ranged from 6 to 70 µg/m³ (0.003 to 0.037 ppm) with a mean
of 26 µg/m³ (0.014 ppm). Outdoor levels of NO₂ were determined
using diffusion tubes systematically located throughout the area; the
weekly average ranged from 26 to 45 µg/m³ (0.014 to 0.024 ppm). One
analysis by the authors was restricted to those 103 children in homes
where gas stoves were present and where bedroom NO₂ exposure was
measured; the data are shown in Table 50. A linear regression model
was fit to the logistic transformation of the rates. Cooking fuel was
found to be associated with respiratory illness, independent of social
class, age, gender or presence of a smoker in the house (p = 0.06).
However, when social class was excluded from the regression, the
association was weaker (p = 0.11). For the 6- and 7-year-old children
living in homes with gas stoves, there appeared to be an increase in
respiratory illness with increasing levels of NO₂ in their bedrooms
(p = 0.10), but no significant relationship was found between
respiratory symptoms in those children, their siblings or parents and
levels of NO₂ in kitchens.

Because no concentration-response estimates were given by the
authors, a multiple logistic model was fitted to the data in Table 50
with a linear slope for NO₂ and separate intercepts for boys and
girls. NO₂ levels for the groups were estimated by fitting a
log-normal distribution to the grouped NO₂ data, and the average
exposures within each interval were estimated (see Hasselblad et al.,
1980). The estimated logistic regression coefficient for NO₂
(in µg/m³) was 0.015 with a standard error of 0.007. The likelihood
ratio test for NO₂ gave a chi-square of 4.94 with one degree of
freedom, with a corresponding p value of 0.03.

The study was repeated in January to March of 1980 by Melia et
al. (1982a,b). This time, children aged 5 to 6 years were sampled
from the same neighbourhood as the previous study, but only families
with gas stoves were recruited. Environmental measurements were made
and covariate data were collected in a manner similar to the previous
study (Melia et al., 1980). Measurements of NO₂ were available for
54% of the homes. The unadjusted rates of one or more symptoms by

Table 50. Unadjusted rates of one or more symptoms among United
Kingdom boys and girls according to bedroom levels of
nitrogen dioxide^a

Bedroom levels of NO₂ (ppm)

< 0.020 0.020-0.039 > 0.039 Total

Boys 43.5% 57.9% 69.2% 54.5%
(23) (19) (13) (55)

Girls 44.0% 60.0% 75.0% 54.2%
(25) (15) (8) (48)

TOTAL 43.7% 58.8% 71.4% 54.4%
(48) (34) (21) (103)

^a Numbers in parentheses refer to number of subjects
(from: Melia et al., 1980)

gender and exposure level are shown in Table 51. The authors
concluded that "... no relation was found between the prevalence of
respiratory illness and levels of NO₂". A reanalysis by Hasselblad
et al. (1992) of the data in Table 51 was made using a multiple
logistic model similar to the one used for the previous study (Melia
et al., 1980). The model included a linear slope for NO₂ and
separate intercepts for boys and girls. Nitrogen dioxide levels for
the groups were estimated by fitting a log-normal distribution to the
grouped bedroom NO₂ data. The estimated logistic regression
coefficient for NO₂ (in µg/m³) was 0.0037 with a standard error of
0.0052. The likelihood ratio test for the effect of NO₂ gave a
chi-square of 0.51 with one degree of freedom (p = 0.48).

Melia et al. (1983) investigated the association between gas
cooking in the home and respiratory illness in a study of 390 infants
born between 1975 and 1978. When the child reached 1 year of age, the
mother was interviewed by a trained field worker to complete a
questionnaire. The mother was asked whether the child usually
experienced morning cough, day or night cough, wheeze or colds going
to the chest, and whether the child had experienced bronchitis, asthma
or pneumonia during the past 12 months. No relation was found between
type of fuel used for cooking at home and the prevalence of
respiratory symptoms and diseases recalled by the mother after
allowing for the effects of gender, social class and parental
smoking. The authors gave prevalence rates of children having at

least one symptom, according to gas stove use and gender. The
combined odds ratio for presence of symptoms according to gas stove
use was 0.63 with 95% confidence interval of 0.36 to 1.10.

Table 51. Unadjusted rates of one or more symptoms among United
Kingdom boys and girls according to bedroom levels of
nitrogen dioxide^a

Bedroom levels of NO₂ (ppm)

< 0.020 0.020-0.039 > 0.039 Total

Boys 56.4% 67.6% 72.0% 64.4%
(39) (37) (25) (101)

Girls 60.0% 41.0% 52.2% 49.4%
(25) (39) (23) (87)

Total 57.8% 53.9% 62.5% 57.5%
(64) (76) (48) (188)

^a Numbers in parentheses refer to number of subjects; source:
Melia et al. (1982a,b)

Melia et al. (1988) studied factors affecting respiratory
morbidity in 1964 primary school children living in 20 inner city
areas of England in 1983 as part of a national study of health and
growth. Data on age, gender, respiratory illness, cooking fuels,
mother's education and size of family were obtained by questionnaire.
Smoking was not studied. The same respiratory questions were asked as
in previous studies. Melia et al. (1990) reported indoor levels of
NO₂ associated with gas stoves in inner city areas of England in
1987. The mean weekly NO₂ level measured in 22 bedrooms of homes
with gas stoves was 45 ± 25 µg/m³ (24.1 ± 13.2 ppb). The mean weekly
NO₂ level measured in four bedrooms of homes without gas stoves was
40 ± 22 µg/m³ (20.7 ± 11.8 ppb). Melia et al. (1988) reported a
relative risk of 1.06 (95% confidence interval of 0.94 to 1.17) for
one or more respiratory conditions associated with exposure to gas or
kerosene fuel used in the home after adjustment for ethnic group,
gender, age group, mother's education, family size and single parent
family status.

7.3.1.2 Harvard University - Six Cities Studies (USA)

Several authors (Spengler et al., 1979, 1986; Speizer et al.,
1980; Ferris et al., 1983; Ware et al., 1984; Berkey et al., 1986;
Quackenboss et al., 1986; Dockery et al., 1989a; Neas et al., 1990,
1991) have reported on two cohorts of children studied in six
different cities in the USA. The six cities were selected to
represent a range of air quality based on their historic levels of
outdoor pollution. They included: Watertown, Massachusetts; Kingston
and Harriman, Tennessee; southeast St. Louis, Missouri; Steubenville,
Ohio; Portage, Wisconsin; and Topeka, Kansas. In each community
during 1974-1977, approximately 1000 first- and second-grade
schoolchildren were enrolled in the first year and an additional
500 first-graders were enrolled in the next year (Ferris et al.,
1979). Families reported the number of people living in the home and
their smoking habits, parental occupation and educational background,
and fuels used for cooking and heating. Outdoor pollution was
measured at fixed sites in the communities as well as at selected
households. Indoor pollution including NO₂ was measured in several
rooms of selected households.

Speizer et al. (1980) reported results from the six cities
studies based on 8120 children, aged 6 to 10 years, who had been
followed for 1 to 3 years. Health end-points were measured by a
standard respiratory questionnaire completed by the parents of the
children. The authors used log-linear models to estimate the effect
of current use of gas stoves versus electric stoves on the rates of
serious respiratory illness before age 2, yielding an odds ratio of
1.12 (95% confidence limits of 1.00 and 1.26) for gas stove use. The
results were adjusted for presence of adult smokers, presence of air
conditioning, and family SES.

Ware et al. (1984) reported results for a larger cohort of
10 160 white children, aged 6 to 9 years, in the same six cities over
a longer period (1974-1979). Directly standardized rates of reported
illnesses and symptoms did not show any consistent pattern of
increased risk for children from homes with gas stoves. Logistic
regression analyses controlling for age, gender, city and maternal
smoking level gave estimated odds ratios for the effect of gas stoves
ranging from 0.93 to 1.07 for bronchitis, chronic cough, persistent
wheeze, lower respiratory illness index, and illness for the last
year. The lower respiratory illness index indicated the presence of
bronchitis, restriction of activity due to lower respiratory illness,
or chronic cough during the past year. The 95% confidence bounds
around all of these symptom-specific odds ratios included 1. Only two
odds ratios approached statistical significance: (1) history of
bronchitis (odds ratio = 0.86, 95% confidence interval 0.74 to 1.00)
and (2) respiratory illness before age 2 (odds ratio = 1.13, 95%
confidence interval 0.99 to 1.28). When the odds ratio for

respiratory illness before age 2 was adjusted for parental education,
the odds ratio was 1.11 with 95% confidence limits of 0.97 and 1.27
(p = 0.14). Thus, the study suggests an increase of about 11% in
respiratory illness before the age of 2 years, which is about the same
as that reported by Speizer et al. (1980), although the increase was
not statistically significant at the 0.05 level. The end-point in the
Ware et al. (1984) study most similar to that of the Melia studies was
the lower respiratory illness index. The authors gave the unadjusted
prevalence, and from those data, an estimated odds ratio of 1.08 with
95% confidence limits of 0.97 and 1.19 was calculated. Although this
odds ratio was not adjusted for other covariates, such adjustments
minimally affected other end-points in this study. Analyses by Ware
et al. (1984) on the other end-points found that effects of adjustment
for covariates was minimal.

During the period from 1983 to 1986, a new cohort of about
1000 second- to fifth-grade schoolchildren in each community was
enrolled and given an initial symptom questionnaire (Dockery et al.,
1989a). The authors studied reported respiratory symptoms on a
subsequent symptom questionnaire (second annual) for 5338 white
children aged 7 to 11 years at the time of enrolment. The end-points
of chronic cough, bronchitis, restriction of activity due to chest
illness, and persistent wheeze were not associated with gas stove use
in the home, but the health end-point of doctor-diagnosed respiratory
illness prior to age 2 yielded an odds ratio of 1.15 with 95%
confidence limits of 0.96 to 1.37. The odds ratio for chronic cough
was 1.15 with 95% confidence limits of 0.89 and 1.91. The odds ratio
was adjusted for age, sex, parental education, city of residence, and
use of unvented kerosene heaters.

Neas et al. (1990, 1991) studied the effects of measured NO₂
among a stratified one-third random sample of the children that were
part of the Dockery et al. (1989a) analysis. The sample was
restricted to 1286 white children 7 to 11 years of age at enrolment
with complete covariate information and at least one valid indoor
measurement of both NO₂ and respirable particles. Methods for
measuring indoor pollutants were described by Spengler et al. (1986).
Indoor pollutants were measured in each child's home for 2 weeks
during the heating season and 2 weeks during the cooling season. The
two 2-week measurements were averaged to estimate each child's annual
average NO₂ exposure. NO₂ was measured by Palmes passive diffusion
tubes at three locations: kitchen, activity room and the child's
bedroom. The three locations were averaged to create a household
annual average NO₂ exposure.

The analysis of the Neas et al. (1990, 1991) study was based on
the final symptom questionnaire (third annual), completed by parents
following the indoor measurements. The questionnaire reported
symptoms during the previous year, including attacks of shortness of
breath with wheeze, persistent wheeze, chronic cough, chronic phlegm

and bronchitis. The authors used a multiple logistic model with
separate city intercepts, indicator variables for gender and age,
parental history of chronic obstructive pulmonary disease, parental
history of asthma, parental education and single parent family status.
Increases in symptoms were estimated for an additional NO₂ exposure
of 28.3 µg/m³ (0.015 ppm). Table 52 shows the odds ratios for the
five separate symptoms associated with the increase in NO₂ exposure.

Table 52. Odds ratios and 95% confidence intervals for the effect of
an additional load of 0.015 ppm NO₂ on the symptom
prevalence (from: Neas et al., 1991)

Symptom Odds ratio 95% Confidence interval

Shortness of breath 1.23 0.93 to 1.61
Persistent wheeze 1.16 0.89 to 1.52
Chronic cough 1.18 0.87 to 1.60
Chronic phlegm 1.25 0.94 to 1.66
Bronchitis 1.05 0.75 to 1.47

Neas et al. (1990, 1991) defined a combined symptom as the
presence of any of the symptoms just reported. A multiple logistic
regression of this combined lower respiratory symptom, equivalent to
the single response regression, gave an estimated odds ratio of 1.40
with a 95% confidence interval of 1.14 to 1.72. The odds ratio for
the combined symptom score was slightly higher than in other studies,
but was not inconsistent with those results. The reference category
for each of the symptom-specific odds ratios included some children
with the other lower respiratory symptoms, whereas the children in the
reference category for combined lower respiratory symptoms were free
of any of these symptoms. When split by gender, the odds ratio was
higher in girls, a result similar to the gender modification reported
by Melia et al. (1979). When separate logistic analyses were
performed for each community, the adjusted odds ratios ranged from
1.26 for Topeka, Kansas, to 1.86 for Portage, Wisconsin. When the
cohort was restricted to the 495 children in homes with a gas stove,
the adjusted odds ratio was 1.37 with a 95% confidence interval of
1.02 to 1.84. Table 53 provides the adjusted odds ratios for combined
lower respiratory symptoms across ordered NO₂ exposure categories.
The association is statistically significant for the upper exposure
category and the overall results are consistent with a linear
dose-response relationship between NO₂ and lower respiratory symptoms
in children.

Table 53. Odds ratios and 95% confidence intervals for the effect of
ordered NO₂ exposures on the prevalence of lower
respiratory symptoms (from: Neas et al., 1991)

NO₂ level (ppm) Number of Odds 95% Confidence
children ratio interval

Range Mean

0 to 0.0049 0.0037 263 1.00
0.005 to 0.0099 0.0073 360 1.06 0.71 to 1.58
0.010 to 0.0199 0.0144 317 1.36 0.89 to 2.08
0.020 to 0.0782 0.0310 346 1.65 1.03 to 2.63

Neas et al. (1992) reported that the estimated effect of an
additional load of 28.3 µg NO₂/m³ (0.015 ppm) on lower respiratory
symptoms was consistent across the seasons and sampling locations.
Table 54 provides the odds ratios and 95% confidence intervals for
this association by season and sampler location. The NO₂ levels
measured by the activity room and bedroom sampler were more strongly
associated with lower respiratory symptoms than those in the kitchen.
The NO₂ measurements in the kitchen were influenced more by transient
peak levels associated with meal preparation on gas stoves, whereas
the other sampling locations were more reflective of the child's
long-term average exposures to NO₂ in the home. Spengler et al.
(1992) suggested that children spend relatively little time (0.5 h per
day) in the kitchen when the range is operating.

7.3.1.3 University of Iowa Study (USA)

Ekwo et al. (1983) surveyed 1355 children 6 to 12 years of age
for respiratory symptoms and lung function in the Iowa City School
District. Parents of the children completed a questionnaire that was
a modification of one developed by the American Thoracic Society. The
children were a random sample from those families whose parents had
completed the questionnaire. Eight measures of respiratory illness
were reported by the authors, but only two were similar to the
end-points used in the United Kingdom studies (section 7.3.1.1) and
the Harvard Six City studies (section 7.3.1.2). Parental smoking was
also measured and used as a covariate in the analyses. Results of the
analyses, based on 1138 children, are presented in Table 55. No
measurements of NO₂ exposure, either inside or outside the homes,
were reported.

Table 54. Odds ratios and 95% confidence intervals for the effect of
an additional 0.015 ppm NO₂ on the prevalence of lower
respiratory symptoms according to sampling location and
season (from: Neas et al., 1992)

Sampler location and Mean difference Odds 95% Confidence
season gas vs. electric ratio interval
(ppm)

Household annual average 0.016 1.40 1.14 to 1.72
Household winter average 0.018 1.16 1.04 to 1.29
Household summer average 0.014 1.46 1.13 to 1.89
Kitchen annual average 0.022 1.23 1.05 to 1.44
Activity room annual
average 0.014 1.50 1.20 to 1.87
Bedroom annual average 0.013 1.47 1.17 to 1.85

Table 55. Analysis of Iowa city school children respiratory symptoms
according to gas stove type and parental smoking
(from: Ekwo et al., 1983)

Factor Hospitalization for Chest congestion and
chest illness phlegm with colds
before age two

Odds ratio SE^a Odds ratio SE^a

Gas stove use 2.4^b 0.684 1.1 0.188
Smoking effects
Father alone smokes 2.3^b 0.856 1.0 0.213
Mother alone smokes 2.9^b 1.239 1.3 0.363
Both smoke 1.6 0.859 1.2 0.383

^a SE = Standard error of the odds ratio
^b Indicates statistical significance at the 0.05 probability level

7.3.1.4 Agricultural University of Wageningen (The Netherlands)

Houthuijs et al. (1987), Brunekreef et al. (1987), and Dijkstra
et al. (1990) studied the effect of indoor factors on respiratory
health in 6- to 9-year-old children from 10 primary schools in five

non-industrial communities in the southeast region of the Netherlands.
Personal exposure to NO₂ and home concentrations were measured. An
important NO₂ emission and exposure source in these homes are
geysers, which are unvented, gas-fired, hot water sources at the water
tap. Exposure to tobacco smoke was assessed by a questionnaire that
also reported symptom information. The study used Palmes diffusion
tubes to measure a single weekly average personal NO₂ exposure. In
January and February 1985, NO₂ in the homes of 593 children who had
not moved in the last 4 years was measured for 1 week. Personal
exposure was also estimated from time budgets and room monitoring.
Estimated and measured exposures to NO₂ are given in Table 56.

Table 56. Estimated and measured personal NO₂ exposure (µg/m³)
for a single weekly average (from: Houthuijs et al., 1987)

NO₂ Source Estimated Measured

Number Arithmetic Standard Arithmetic Standard
mean deviation mean deviation

No geyser 370 22 7 22 9
Vented geyser 112 29 9 31 12
Unvented geyser 111 40 9 42 11

Three health measures were obtained from the questionnaire, a
modified form of the WHO questionnaire. The different items were
combined to create three categories: cough, wheeze and asthma. Asthma
was defined as attacks of shortness of breath with wheezing in the
past year. The presence of any of the three symptoms was used as a
combination variable. The results are presented in Table 57. A
logistic regression model was used to fit the combination variable.
Exposure was estimated by fitting a log-normal distribution to the
grouped data, and the mean exposure values for each group were
estimated by a maximum likelihood technique (Hasselblad et al.,
1980). The estimated logistic regression coefficient was œ0.002,
corresponding to an odds ratio of 0.94 for an increase of 28.3 µg/m³
(0.015 ppm) in NO₂, with 95% confidence interval of 0.70 to 1.27.
Thus, these studies did not demonstrate an increase in respiratory
disease with increasing NO₂ exposure, but the range of uncertainty is
quite large and the rates were not adjusted for covariates such as
parental smoking and age of the child. One potential explanation
offered by the authors for the negative findings with respect to NO₂
exposure was the smaller sample size of the measured NO₂ data
compared to the categorical data (i.e., gas stove versus electric

stove use). They could not estimate whether more precision was gained
by use of measured NO₂ than was lost by the reduction in the sample
size. Houthuijs et al. (1987) reported earlier that the presence of
an unvented geyser in the kitchen is associated with a higher
prevalence of respiratory symptoms and that the NO₂ difference
between no geyser present and an unvented geyser is about 0.01 ppm.

7.3.1.5 Ohio State University Study (USA)

Mitchell et al. (1975) and Keller et al. (1979a) conducted a
12-month study of respiratory illness and pulmonary function in
families in Columbus, Ohio, prior to 1978. The sample included 441
families divided into two groups using either gas or electric cooking.
Participating households were given diaries to record respiratory
illnesses for 2-week periods. Respiratory illnesses included colds,
sore throat, hoarseness, earache, phlegm and cough. Only one incident
of illness per person per 2-week period was recorded. The study
measured NO₂ exposure, by both the Jacobs-Hochheiser and continuous
chemiluminescence methods. The electric stove users averaged
38 µg/m³ (0.02 ppm) NO₂ exposure, whereas the gas stove users
averaged 94 µg/m³ (0.05 ppm). The report did not indicate which
rooms were measured in order to obtain this average.

No differences were found in any of the illness rates for
fathers, mothers or children. No analyses were carried out using
multiple logistic regression or Poisson regression (these methods were
relatively new at the time). No estimates were made that can be
considered comparable to the odds ratios reported in the other
studies. However, the authors did show a bar graph of all respiratory
illness for children under 12. The rates were 389 (per 100 person-
years) for electric stove use and 377 for gas stove use. These rates
were not significantly different even after adjustment for covariates,
including family size, age, gender, length of residence and father's
education. No mention was made of adjustments for smoking status or
smoking exposure for the children.

In a second, related study (Keller et al., 1979b), 580 people
drawn from households that participated in the earlier study were
examined to confirm the reports and to determine the frequency
distribution of reported symptoms among parents and children in gas or
electric cooking homes. A nurse-epidemiologist examined selected
subjects who reported ill and obtained throat cultures. The
percentage of children having respiratory illnesses in homes with a
gas stove was 85.1% (n = 87) versus 88.8% (n = 89) in homes with
electric stoves. The unadjusted proportions permit the calculation of
an estimated odds ratio of 0.71 with 95% confidence interval of 0.30
to 1.74. Unfortunately the adjusted rates were not reported.

Neas et al. (1991) commented that Keller's model controls for a
series of variables that specify the child's prior illness history and
that if chronic exposure to NO₂ is a risk factor for prior illnesses,
controlling for the child's illness history would substantially reduce
the estimated effect of current NO₂ exposure.

7.3.1.6 University of Dundee (United Kingdom)

Ogston et al. (1985) studied infant mortality and morbidity in
the Tayside region of northern Scotland. The subjects were 1565
infants born to mothers who were living in Tayside in 1980. Episodes
of respiratory illness were recorded during the first year of life.
The information was supplemented by observations made by a health
visitor and scrutinized by a paediatrician who checked diagnostic
criteria and validity. One health end-point assessed was defined as
the presence of any respiratory disease during the year. The use of
gas cooking fuel was associated with increase respiratory illness
(odds ratio = 1.14, 95% confidence interval 0.86 to 1.50) after
adjustment for parental smoking, mother's age and type of home heating
(Table 58). The study did not give measured NO₂ exposure values, but
referenced the other studies conducted elsewhere in the United Kingdom
for exposure estimates.

7.3.1.7 Harvard University - Chestnut Ridge Study (USA)

Schenker et al. (1983) reported a large respiratory disease study
of 4071 children aged 5 to 14 in the Chestnut Ridge region of western
Pennsylvania. The region is predominately rural, with numerous
underground coal mines and four large coal-fired electricity-
generating plants in the area. A standardized children's
questionnaire (Ferris, 1978) was sent to parents of all children in
grades 1 to 6 in targeted schools. An SES scale derived from the
parent's occupation and education was divided into quintiles to
provide SES strata. Important confounding factors considered in the
analysis were gender, SES and maternal smoking. In the multiple
logistic model, no significant association was found between gas stove
use and any of the respiratory or illness variables after adjusting
for SES. No odds ratios or other numerical data were reported.

Table 57. Frequency and prevalence of reported respiratory symptoms with respect to different
categories of mean indoor NO₂ concentrations in a population of 775 children
aged 6 to 12 old (from: Dijkstra et al., 1990)

Frequency and prevalence in category of indoor NO₂

Symptom 0-20 µg/m³ 21-40 µg/m³ 41-60 µg/m³ > 60 µg/m³
(n = 336) (n = 267) (n = 93) (n = 79)

Cough 16 4.8% 12 4.5% 7 7.5% 3 3.8%

Wheeze 30 8.9% 18 6.7% 3 3.2% 7 8.9%

Asthma 22 6.6% 12 4.5% 2 2.2% 3 3.8%

One or more symptoms 36 10.7% 24 9.0% 8 8.6% 8 10.1%

Table 58. Regression coefficients for multiple logistic analyses of
respiratory illness in Tayside children (from: Ogston et
al., 1985)

Factor Regression Odds ratio 95% Confidence
coefficient limits

Parental smoking 0.429 1.54

Age of mother -0.094 not available
(in 5-year groups)

Presence of gas stove 0.130 1.14 0.86, 1.50

7.3.1.8 University of New Mexico Study (USA)

Samet et al. (1993) conducted a prospective cohort study between
January 1988 and June 1990 to test the hypothesis that exposure to
NO₂ increases the incidence and severity of respiratory illness
during the first 18 months of life. A total of 1315 infants were
enrolled into the study at birth in Albuquerque, New Mexico. The
subjects were healthy infants from homes without smokers and who spent
less than 20 h/week in day care. Illness experience was monitored by
a daily diary of symptoms completed by the mother and a telephone
interview conducted every two weeks. For a sample of the ill
children, a nurse practitioner made a home visit to conduct a
standardized history and physical assessment. Exposure to NO₂ was
estimated by a 2-week average concentration measured in the subjects'
bedrooms with passive samplers. Estimates of exposure based on
bedroom concentration were tightly correlated with estimates of
exposures calculated as time-weighted averages of the concentrations
in the kitchen, bedroom and activity room. The authors defined
illness events as the occurrence on at least two consecutive days of
any of the following: runny or stuffy nose, wet cough, dry cough,
wheezing or trouble with breathing. Wheezing was defined as a

high-pitched musical sound audible during breathing, and trouble with
breathing as the parent's perception of rapid or laboured breathing.
Illness events ended with two consecutive symptom-free days.

The analysis was limited to the 1205 subjects completing at least
1 month of observation; of these, 823 completed the full protocol.
Multivariate methods were used to control for potential confounding
factors and to test for effect modification. In analyses of
determinants of incident illnesses, the outcome variable was the
occurrence of illness during 2-week intervals of days at risk. The
independent variables considered in the multivariate analyses included
the fixed factors of birth order, gender, ethnicity, parental asthma
and atopic status, household income, and maternal education. Other
variables considered were the temporally varying factors of age,
calendar month, day-care attendance and breast-feeding. Potential
confounding and effect modification by cigarette smoking was
controlled by excluding subjects from households with smokers.

Lambert et al. (1993) reported that in this prospective cohort
study during the winter, bedroom concentrations in homes with gas
stoves averaged 0.021 ppm (SD = 0.022 ppm). In bedrooms of homes with
electric stoves, concentrations averaged 0.007 ppm (SD = 0.006 ppm).
Approximately 77% of the bedroom NO₂ observations were less than
0.02 ppm; only 5% were greater than 0.04 ppm. The 90th percentile of
the weekly measured concentrations was 0.05 ppm NO₂ in bedrooms.

Samet et al. (1993) performed the analysis using the generalized
estimated equations described by Zeger & Liang (1986). This takes into
account the correlation structure when estimating regression
coefficients and their standard errors. The multivariate models
examined the effects of the unlagged NO₂ exposures, lagged NO₂
exposures and stove type (Table 59). None of the odds ratios was
significantly different from unity, the value for the reference
category of 0 to 0.02 ppm. Additionally, the odds ratios did not tend
to increase consistently from the middle category of exposure to the
highest category. Furthermore, exposure to NO₂ and the durations of
the four illness categories were not associated. The authors added
NO₂ exposure to the model as a continuous variable, while controlling

for the same covariates included in Table 59. For each of the five
illness variables, the estimated multiplier of the odds ratio per
0.001 ppm increment of NO₂ was 0.999, with confidence limits
extending from approximately 0.995 to 1.002.

7.3.1.9 University of Basel Study (Switzerland)

Braun-Fahrlaender et al. (1989, 1992) and Rutishauser et al.
(1990a,b) studied the incidence and duration of common airway symptoms
in children up to 5 years old over a 1-year period in a rural, a
suburban and two urban areas of Switzerland. Parents were asked to
record daily their child's respiratory symptoms (from a list) over a
6-week period. Additionally, covariates, including family size,
parental education, living conditions, health status of the child,
parents' respiratory health, and smoking habits of the family, were
assessed by questionnaire. During the same 6-week period NO₂ was
measured weekly using Palmes tubes, both inside and outside the home
of the participants. Meteorological data were obtained from local
monitoring stations, but additional air quality data from fixed
monitoring sites were only available for the two urban study areas.
NO₂ concentrations inside the home were on average lower than in the
outside air (Fig. 24). Indoor levels for Basel, Zurich, Wetzikon and
Rufzerfeld were 33.8, 28.4, 20.5 and 11.2 µg/m³ (0.018, 0.015, 0.011
and 0.006 ppm), respectively. The indoor NO₂ concentration depended
to some extent on the concentration of the outside air.

The analysis was restricted to 1063 Swiss nationals (from a total
of 1225 participating families). For all four study areas, regional
mean incidence rates of upper respiratory illness, cough, breathing
difficulties and total respiratory illness, adjusted for individual
covariates and weather data, were regressed (using Poisson regression)
against regional differences in annual mean NO₂ concentrations. All
the relative risks were computed for a 20-µg/m³ (0.011-ppm) increase
in pollution concentration. The NO₂ concentration measured by indoor
passive sampler was associated with the duration of any episode
(relative duration of 1.16, 95% confidence interval of 1.12 to 1.21),
upper respiratory episodes (relative duration of 1.18, 95% confidence
interval of 1.01 to 1.38), and coughing episodes (relative duration of
1.15, 95% confidence interval of 1.03 to 1.29). A discussion of
associations with outdoor levels is presented in section 7.3.2.

Table 59. Odds ratios^a for effect of nitrogen dioxide exposure on incidence of respiratory illness
(from: Samet et al., 1993)

NO₂ exposure All illnesses All lower Lower, with Lower, with
wet cough wheezing

Odds ratio 95% CI^b Odds ratio 95% CI^b Odds ratio 95% CI^b Odds ratio 95% CI^b

Unlagged^c 1.04 0.96-1.12 0.98 0.89-1.09 1.00 0.89-1.12 0.92 0.73-1.15
0.02-0.06 ppm 0.94 0.81-1.08 0.93 0.76-1.13 0.94 0.77-1.16 0.88 0.56-1.37
> 0.04 ppm

Lagged^c 1.01 0.93-1.10 0.97 0.87-1.08 0.97 0.87-1.09 0.95 0.75-1.19
0.02-0.06 ppm 0.92 0.77-1.10 0.91 0.72-1.15 0.89 0.68-1.16 0.98 0.66-1.48
> 0.04 ppm

Gas Stove^d 0.98 0.90-1.07 0.91 0.81-1.04 0.94 0.82-1.07 0.84 0.64-1.09

^a Obtained by generalized estimating equation method. Adjusted for season, age, gender, ethnicity, birth order, day care, income,
maternal education, breast feeding, parental atopy and asthma, and maternal history of respiratory symptoms.
^b CI = Confidence interval
^c Reference category is 0-0.02 ppm NO₂
^d Reference category is electric stove
7.3.1.10 Yale University Study (USA)

Berwick et al. (1984, 1987, 1989), Leaderer et al. (1986) and
Berwick (1987) reported on a 12-week study (six 2-week time periods)
of lower and upper respiratory symptoms in 159 women and 121 children
(aged 12 or less) living in Connecticut. Levels of NO₂ were measured
in 91% of the homes, 57 of which had kerosene heaters and 62 of which
did not. Ambient NO₂ levels ranged from 9 to 19 µg/m³ (0.005 to
0.01 ppm) for the six 2-week time periods. Two-week average indoor
NO₂ levels in homes of monitored children were highest for homes with
kerosene heaters and gas stoves (91 µg/m³, 0.05 ppm; n = 8),
second highest for kerosene only (36 µg/m³, 0.02 ppm; n = 45), third
highest for gas stoves only (32 µg/m³, 0.02 ppm; n = 13), and lowest
for no sources (6 µg/m³, 0.003 ppm; n = 43). Indoor levels did not
fluctuate greatly over time, as indicated by the 2-week averages. A
comparison of personal NO₂ exposures, as measured by Palmes diffusion
tubes, and NO₂ exposures measured in residences had a correlation of
0.94 for a subsample of 23 individuals. Results of this comparison
show an excellent correlation between average household exposure and
measured personal exposure (see section 3.6 and Fig. 13).

The study defined lower respiratory illness as the presence of at
least two of the following: fever, chest pain, productive cough,
wheeze, chest cold, physician-diagnosed bronchitis, physician-
diagnosed pneumonia and asthma. Information on many potential
covariates (e.g., SES, age, gender and exposure to environmental
tobacco smoke) was obtained. The covariates having the largest effect
were age of child, family SES and history of respiratory illness, as
shown by multiple logistic analysis. When controlling for SES and
history of respiratory illness, children under 7 years of age exposed
to 30 µg NO₂/m³ (0.016 ppm) or more were found to have a risk of
lower respiratory symptoms 2.25 times higher than that of unexposed
children (95% confidence limits of 1.69 and 4.79). Older children and
adults showed no increased risk.

Although the Berwick study had relatively extensive information
on exposure, several problems are evident. Unvented kerosene
space-heaters also release volatile organic compounds and combustion
particles. The 4-year age-specific relative risks for lower
respiratory disease are very variable, and it is not clear why these
3-year strata were collapsed into 2 strata at 7 years of age. The

analyses may be sensitive to the adjustment for SES, which can be
correlated with exposure. This is less of a problem in studies with
larger sample sizes (e.g., Melia et al. 1977, 1979), but may be
critical in the Berwick study. Furthermore, Neas et al. (1991) noted
that the Berwick study controlled for prior illnesses, as did the
Keller study, which would reduce the estimated effect of current NO₂
exposure.

7.3.1.11 Freiburg University Study (Germany)

Kuehr et al. (1991) conducted a cross-sectional study on the
prevalence of asthma in childhood in relation to NO₂ levels in the
city of Freiburg and two Black Forest communities. A study group of
704 children (with 41 asthmatic) aged 7 to 16 years took part in a
standardized interview and medical examination. Indoor and outdoor
exposure information was taken into account. Passive smoking
exposures were assessed. Stoves used as heating devices carried a
4.8-fold relative risk for asthma compared to other types of heating
(95% CI 1.95-11.8).

7.3.1.12 McGill University Study (Canada)

In a case-control study carried out in Montreal, Quebec, Canada,
between 1988 and 1990, NO₂ levels measured by passive NO₂ monitoring
badge were studied in relation to the incidence of asthma among 3- and
4-year-old children (Infante-Rivard, 1993). Multivariate
unconditional logistic regression was carried out for the 140 subjects
who had NO₂ measurements; the analysis included NO₂ and the
variables retained in the final conditional model that includes SES
and parental smoking. The author reported an increase in asthma
incidence associated with NO₂ exposure levels. However, the Task
Group noted the exceptionally large effect estimates given the
exposure levels.

7.3.1.13 Health and Welfare Canada Study (Canada)

Dekker et al. (1991) studied asthma and wheezing syndromes as
part of a questionnaire-based study of 17 962 Canadian school
children. The questionnaire was developed from the 1978 American
Thoracic Society questionnaire, which was the same as that used in the
Harvard Six Cities Study. For analysis, the sample was restricted to
children aged 5 to 8 years and excluded those children with cystic
fibrosis as well as those living in mobile homes, tents, vans,
trailers and boats. The authors calculated odds ratios adjusted for

age, race, gender, parental education, gender of the respondent,
region of residence, crowding, dampness and environmental tobacco
smoke. The adjusted odds ratio of asthma as a function of gas cooking
was 1.95 with 95% confidence limits of 1.41 and 2.68. The adjusted
odds ratio of wheezing as a function of gas cooking was 1.04 with 95%
confidence limits of 0.77 and 1.42. The authors noted that this
finding needed to be treated with caution, however, because of the few
subjects with asthma in the study who were exposed to gas cooking
(n = 60).

7.3.1.14 University of North Carolina Study (USA)

Margolis et al. (1992) studied the prevalence of persistent
respiratory symptoms in 393 infants of different SES by analysing data
from a community-based cohort study of respiratory illness in the
first year of life in central North Carolina between 1986 and 1988.
Infants were limited to those weighing more than 2000 g and who did
not require neonatal care outside the normal newborn nursery. Of
those eligible, 47% were enrolled and, of these, 77% completed the
study and were included in the analysis. Compared with the 1241
infants from families refusing enrolment, the 1091 eligible study
infants were more likely to be of high SES and were more often black.
Study infants were less likely to have mothers who smoked.

The presence of persistent respiratory symptoms was measured at
the 12-month home interview using an American Thoracic Society
children questionnaire (modified for infants) for studies of
respiratory illness. Infants who were reported to "usually cough" or
"occasionally wheeze" were classified as having persistent respiratory
symptoms.

Of the 393 infants that Margolis et al. (1992) included in their
study, approximately 41 lived in homes with gas cooking. The relative
risk of persistent respiratory symptoms among infants exposed to gas
cooking unadjusted for any covariates was 1.12 (95% confidence
interval of 0.63 to 2.04).

7.3.1.15 University of Tucson Study (USA)

The study by Dodge (1982) was based on a cohort of 676 children
in the third and fourth grades (about 90% aged 8-10 years) of schools
in three Arizona communities. Gas cooking stoves were associated with
increased symptoms: asthma odds ratio = 1.47, wheeze odds ratio
= 1.24, sputum odds ratio = 2.28, and cough odds ratio = 2.21.
However, only 79 children (19%) had electric heat, so the numbers were
small and only cough was significant at the 0.05 level. After
controlling for height and age, gas stoves were not associated with a
decline in the growth of FEV₁.

7.3.1.16 Hong Kong Anti-Cancer Society Study (Hong Kong)

In 1985, 362 primary school children (age 7-13 years) were
included in a study of NO₂ exposure and respiratory illness in Hong
Kong (Koo et al., 1990). Exposures to NO₂ were estimated by use of
personal badge monitors, worn for a single period of 24 h, and
supplemented by monitors placed in classrooms. NO₂ exposures were
estimated in the same manner for the mothers of the study children.
Mothers and children completed respiratory illness questionnaires. No
association was found between respiratory symptoms and NO₂ exposures
for children (mean 19 ppb). Among the mothers (mean exposure 19 ppb)
allergic rhinitis and chronic cough were associated with NO₂.

7.3.1.17 Recent studies

This section includes studies that have reported preliminary
results only or have appeared recently in the scientific literature.

Spengler et al. (1993) reported results for evaluation of more
than 15 000 schoolchildren in various sites in the USA and Canada, but
found no statistically significant increases in respiratory symptoms
to be associated with use of gas heaters or cookers.

Goren et al. (1993) reported no association between gas heating
and respiratory health effects among 8000 schoolchildren in Israel.

Preliminary results reported by Peat et al. (1990) indicated no
relationship between relatively high NO₂ in Australian homes with
gas use in Sydney and respiratory symptoms or bronchial hyper-
responsiveness.

Pilotto (1994) reported a prospective study of health effects of
unflued gas heater emissions on 425 Australian schoolchildren aged
6-11 years. Short-term indoor monitoring by means of passive
diffusion badge monitors placed in classrooms or worn at home was
carried out to determine daily 6-h averages. Children exposed to a
level of 0.08 ppm or more, compared with a background level of
0.02 ppm, had increased rates of respiratory illnesses and school
absences.

7.3.2 Outdoor studies

Several studies have examined the relationship of estimated
ambient NO₂ levels to respiratory health outcome measures, including
various respiratory symptomatologies. Those that provide a
quantitative estimate of effect are indicated in Table 60.

Table 60. Effects of outdoor NO₂ exposure on respiratory disease

Study Health end-point NO₂ levels (ppm)/period Odds ratio or 95% CI
estimate

Dockery et al. (1989b) Bronchitis 0.007-0.023 annual average 1.7 0.5 to 5.5
Chronic cough 1.6 0.3 to 10.5
Chest illness 1.2 0.3 to 4.8
Wheeze 0.8 0.4 to 1.6
Asthma 0.6 0.3 to 0.9

Braun-Fahrlaender et al. (1992) Duration of respiratory Change of 0.011 6-week 1.11 1.07 to 1.16
episodes average

Schwartz et al. (1991) Croup 0.005-0.037 daily 1.28 1.07 to 1.54

Jaakkola et al. (1991) Upper respiratory Contrasted polluted versus 1.6 1.1 to 2.1
infection less polluted areas by
comparison of annual levels

7.3.2.1 Harvard University - Six City Studies (USA)

As part of the US Six City Studies, Dockery et al. (1989b)
obtained respiratory illness and symptom data from questionnaires
distributed from September 1980 to April 1981. Indoor air aspects
of this study (Dockery et al., 1989a) were described in the section
on indoor studies. The questionnaires obtained information on
bronchitis, cough, chest illness, wheeze and asthma. A centrally
located air monitoring station was established in 1974 where ambient
sulfur dioxide, NO₂, ozone, total suspended particulate matter and
meteorological variables were measured. The authors used multiple
logistic regression analysis in order to adjust for covariates of
gender, age, maternal smoking, gas stove use and separate intercepts
for each city. Although the strongest associations were found between
respiratory symptoms and particulate matter, there were increased odds
ratios of respiratory symptoms with ambient NO₂. These were not
statistically significant, but the direction for bronchitis, chronic
cough and chest illness was consistent with the studies of indoor
exposure. The odds ratios for various health end-points for an
increase in NO₂ from the lowest-exposure city to the highest-exposure
city 12 to 43 µg/m³ (0.0065 to 0.0226 ppm) are shown in Table 60.

7.3.2.2 University of Basel Study (Switzerland)

Braun-Fahrlaender et al. (1992) studied the incidence and
duration of common airway symptoms in children up to 5 years old.
This study, also discussed in section 7.3.1.9, was conducted over
a 1-year period in a rural, a suburban and two urban areas of
Switzerland. Parents were asked to record their child's respiratory
symptoms (from a list) daily over a 6-week period. Additionally,
covariates including family size, parental education, living
conditions, health status of the child, parents' respiratory health
and smoking habits of the family were assessed by questionnaire.
Weekly NO₂ measurements were made during the same 6-week period using
Palmes tubes, both inside and outside the home of the participants.
Meteorological data were obtained from local monitoring stations, but
additional air quality data from fixed monitoring sites were only
available for the two urban study areas. The analysis was restricted
to 1063 Swiss nationals (from a total of 1225 participating families).
For all four study areas, regional mean incidence rates of upper
respiratory illness, cough, breathing difficulties and total
respiratory illness, adjusted for individual covariates and weather
data, were regressed (using Poisson regression) against regional
differences in annual mean NO₂ concentrations. There was no
association between long-term differences in NO₂ levels by region and
mean annual rates of respiratory incidence.

The adjusted annual mean symptom duration by region and the
corresponding NO₂ levels (measured by passive samplers) are shown in
Table 61. A second-stage regression of the adjusted natural logarithm
of regional mean duration on NO₂ levels yields significant
associations between outdoor NO₂ levels and the average duration of
any respiratory episode (relative duration of 1.11, 95% confidence
interval of 1.07 to 1.16) and upper respiratory episodes (relative
duration of 1.14, 95% confidence interval of 1.03 to 1.25). A
positive trend for the duration of coughing episodes was also seen
(relative duration of 1.09, 95% confidence interval of 0.97 to 1.22).
No association was seen with the duration of breathing difficulties.
All the relative risks are computed for a 20-µg/m³ (0.011-ppm)
increase in pollution concentration. In the suburban and rural areas,
NO₂ was the only air pollutant measured. Correlation between
outdoor passive NO₂ sampler and total suspended particulate (TSP)
measurements in the two urban study areas was quite high (0.52). The
high correlation between NO₂ and TSP suggests that this NO₂
association may reflect confounding with TSP. The lack of TSP data
for the other two regions precludes eliminating TSP as a possible
confounder in this analysis. But the consistency of the NO₂ findings
are evident and, although the association with symptom duration in
Zurich and Basel may well be due to confounding with TSP, the
cross-sectional association across the four regions supports a
possible NO₂ role.

7.3.2.3 University of Wuppertal Studies (Germany)

Schwartz et al. (1991) evaluated respiratory illness in five
German communities. Children's hospitals, paediatric departments of
general hospitals, and paediatricians reported daily the numbers of
cases of croup. A diagnosis of croup was based on symptoms of
hoarseness and barking cough, inspiratory stridor, dyspnoea, and a
sudden onset. The counts were modelled using Poisson regression
with adjustments for weather, season, temperature, humidity and
autoregressive errors. Statistically significant effects of both
ambient particulate matter and NO₂ were found on the counts of
respiratory illnesses. A relationship between short-term fluctuations
in air pollution and short-term fluctuations in medical visits for
croup symptoms was found in this study. The estimated relative risk
was 1.28 with 95% confidence limits of 1.07 and 1.54 for an increase
from 10 to 70 µg NO₂/m³ (0.005 to 0.037 ppm).

7.3.2.4 University of Tubigen (Germany)

Rebmann et al. (1991) studied 875 cases of croup in Baden-
Württemberg in relation to ambient NO₂ levels over a 2-year period.
Monthly NO₂ means varied from 23 to 78 µg/m³. Statistical
regression methods indicated weak but statistically significant
influences of the daily ambient NO₂ mean on the occurrence of croup.

Table 61. Adjusted annual symptom duration (days) and NO₂ levels in four regions of Switzerland
(from: Braun-Fahrlaender et al., 1992)

Region Any symptom URI duration^a Cough Breathing difficulty Indoor NO₂ Outdoor NO₂
duration duration duration concentration concentration
(ppm) (ppm)

Basel 4.50 1.99 2.32 1.55 0.0166 0.0272

Zurich 4.21 1.85 2.01 1.72 0.0118 0.0248

Wetzikon 4.00 1.62 2.10 3.47 0.0103 0.0173

Rafzerfeld 3.88 1.72 2.02 1.25 0.0059 0.0133

^a URI = Upper respiratory illness
7.3.2.5 Harvard University - Chestnut Ridge Study (USA)

In the autumn of 1980, Vedal et al. (1987) conducted a panel
study on 351 children selected from the 1979 Chestnut Ridge study.
Parents and children were instructed at the beginning of the school
year in completing daily diaries of respiratory symptoms. Lower
respiratory illness was defined as wheeze, pain on breathing, or
phlegm production. Of the 351 subjects selected for the 8 month of
follow-up, 128 participated in the completion of diaries. Three
subgroups were established: one without respiratory symptoms, one with
symptoms of persistent wheeze, and one with cough or phlegm production
but without persistent wheeze. Maximum hourly NO₂ levels, measured
at a single monitoring site in the study region, for each 24-h period
were used to reflect the daily pollutant level. During September 1980
to April 1981, the mean NO₂ maximum daily level was 40.5 µg/m³
(0.021 ppm) with a range of 12 to 79 µg/m³ (0.006 to 0.042 ppm).
Regression models could not be fit for asymptomatic subjects; thus 55
subjects were included in the analysis of lower respiratory illness,
but NO₂ levels were not predictive of any symptom outcome.

7.3.2.6 University of Helsinki Studies (Finland)

Jaakkola et al. (1991) studied the effects of low-level air
pollution in three cities by comparing the frequency of upper
respiratory infections over a 12-month period in 1982 as reported by
parents of children aged 14 to 18 months (n = 679) and 6 years
(n = 759). Pollutants studied included ambient levels of NO₂, the
annual mean of which was 15 µg/m³ (0.008 ppm). Other pollutants
monitored were sulfur dioxide, hydrogen sulfide and particles.
Passive smoking and SES were taken into account. The authors reported
a significant association between the occurrence of upper respiratory
infections and living in an air-polluted area for both age groups
studied, both between and within cities. The adjusted odds ratio was
1.6 (95% confidence interval of 1.1 to 2.1) in the 6-year-old age
group. The authors concluded that the combined effect of sulfur
dioxide, particulates, NO₂, hydrogen sulfide and other pollutants may
be a contributing factor in the study results.

7.3.2.7 Helsinki City Health Department Study (Finland)

Pönkä (1991) studied effects of ambient air pollution and minimum
temperature on the number of patients admitted to hospital for asthma
attacks in Helsinki from 1987 to 1989. During the 3-year period,
4209 hospitalizations for asthma occurred. The temperature ranged from
œ37 to +26°C, with a 3-year mean of 5°C, and the number of admissions
increased during cold weather. After standardization for minimum
temperature, the multiple-regression analysis indicated that NO₂ and
carbon monoxide levels were significantly related to asthma admission.
The NO₂ levels averaged 38.6 µg/m³ (0.02 ppm) for the 3-year period,
ranging from 4.0 to 169.6 µg/m³ (0.002 to 0.09 ppm). During the

period of high NO₂ (mean 45.8 µg/m³, 0.024 ppm) levels, the mean
number of all admissions was 29% greater than during the lower
pollution period (28.1 µg/m³, 0.015 ppm). Indoor NO₂ levels and
cooking fuel use were not reported.

7.3.2.8 Oulu University Study (Finland)

The number of daily attendances for asthma at the emergency room
of the Oulu University Central Hospital, Finland, was recorded for one
year, along with daily measures of air pollutants at four points
around the city (Rossi et al., 1993). Daily mean levels of NO₂
ranged up to 69 µg/m³ (maxima 0-154 µg/m³). Asthma visits were
reported to be significantly associated with NO₂, SO₂, H₂S and TSP
levels. After adjustment for daily temperature, only NO₂ was
significantly correlated with attendances. The association of NO₂
and asthma attacks was stronger in winter months than during the
summer.

7.3.2.9 Seth GS Medical College Study (India)

A survey of air pollution and health was carried out in Bombay,
India, in 1978 (Kamat et al., 1980). The study included 4129 adults
in three urban areas and one rural area. A single monthly mean NO₂
level was reported for each study area - annual averages were 4 µg/m³
in the rural area, and 14-16 µg/m³ in the city. Winter levels in
the city study were higher than at other times of the year (up to
40 µg/m³). It was reported that chronic cough with sputum, frequent
colds and exertional dyspnoea were significantly associated with NO₂
levels. These symptoms were also associated with atmospheric levels
of SO₂ and suspended particulate matter, and it was not possible to
identify a separate influence of NO₂ alone.

7.4 Pulmonary function studies

Pulmonary function studies are part of any comprehensive
investigation of the possible effects of any air pollutant.
Measurements can be made in the field, they are non-invasive, and
their reproducibility has been well documented. Age, height, gender
and presence of respiratory symptoms are important determinants of
lung function. Furthermore, changes in pulmonary function have been
associated with exposure to tobacco smoke, particulate matter and
other factors. The studies reviewed below evaluate pulmonary function
changes in relation to indoor or outdoor NO₂ exposures. Several of
the respiratory disease studies described earlier also included
information on pulmonary function.

7.4.1 Harvard University - Six City Studies (USA)

Ware et al. (1984) described analysis of lung function values
using multiple linear regression on the logarithm of the lung function
measures. Covariates included sex, height, age, weight, smoking
status of each parent, and educational attainment of the parents.
Exposure to gas stoves was associated with reductions of 0.7% in mean
FEV₁ (forced expiratory volume in 1 second) and 0.6% in mean forced
vital capacity (FVC) at the first examination (p < 0.01), and
reductions of 0.3% at the second examination (not significant).
The estimated effect of exposure to gas stoves was reduced by
approximately 30% after adjustment for parental education. The
authors stated that the adjustment for parental education may be an
over-adjustment, and may partially represent gas stove use because of
association between parental education and type of stove.

Berkey et al. (1986) used the data from children seen at two to
five annual visits to study factors affecting pulmonary function
growth. Children whose mothers smoked one pack of cigarettes per day
had FEV₁ growth rates approximately 0.17% per year lower (p = 0.05).
The same data provided no evidence for an effect of gas stove exposure
on growth rate.

Dockery et al. (1989b) obtained pulmonary function data
during the 1980 and 1981 school year. Only TSP concentration was
consistently associated with estimated lower levels of pulmonary
function. There was little evidence for an association between lower
pulmonary function levels and the annual mean concentration of NO₂ or
any other pollutant.

Neas et al. (1991) also reported that indoor NO₂ levels were not
significantly associated with a deficit in children's pulmonary
function levels in either of two examinations (FEV₁ and FVC).

7.4.2 National Health and Nutrition Examination Survey Study (USA)

Schwartz (1989) studied air pollution effects on lung function in
children and youths aged 6 to 24 years. FVC, FEV₁, and peak flow
measurements taken as part of the National Health and Nutrition
Examination Survey II (NHANES II) were examined after controlling
for age, height, race, gender, body mass, cigarette smoking and
respiratory symptoms. Air pollution measurements were taken from all
population-oriented monitors in the US EPA database. Each person was
assigned the average value of each air pollutant from the nearest
monitor for the 365 days preceding the spirogram. Highly significant
negative regression coefficients were found for three pollutants
(TSP, NO₂ and O₃) with the three lung function measurements. For an
increase of NO₂ exposure of 28.3 µg/m³ (0.015 ppm), an estimated
decrease of about 0.045 litres was seen in both FVC and FEV₁.

7.4.3 Harvard University - Chestnut Ridge Study (USA)

Vedal et al. (1987) conducted a panel study on 351 children
selected from the 1979 Chestnut Ridge cross-sectional study of
elementary school-aged children (mean age = 9.5 years). Peak
expiratory flow (PEF) was measured daily in 144 children for
9 consecutive weeks and was regressed against daily maximum hourly
ambient concentrations of NO₂, SO₂ and coefficient of haze. No air
pollutant was strongly associated with PEF. All pollutant levels were
relatively low; NO₂ levels ranged from 12 to 79 µg/m³ (0.006 to
0.042 ppm). No indoor measurements were made, nor were any surrogates
for indoor pollution included in the analysis.

7.4.4 Other pulmonary function studies

Ekwo et al. (1983) obtained pulmonary function measurements from
89 children whose parents did not smoke and 94 children whose parents
smoked, and reported no differences in lung function associated with
gas stove use in a cohort of children 6 to 12 years of age.

Dijkstra et al. (1990) examined pulmonary function in Dutch
children; lung function was measured at the schools. There was a weak
negative association between FEF_25-75% (25 and 75% of FVC) and exposure
to NO₂. FEV₁, PEF and FEF_25-75% were all negatively associated with
exposure to tobacco smoke. The authors concluded that the study
failed to document clear associations between indoor exposure to NO₂
and lung function changes in 6- to 12-year-old Dutch children.

Lebowitz et al. (1985) studied a cluster sample of 117 middle-
class households in Tucson, Arizona, USA. Symptom diaries and peak
flows were obtained over a 2-year period. Outdoor sampling of O₃,
TSP, CO and NO₂ was done in or near the clusters. Indoor sampling of
O₃, TSP, respirable suspended particles and CO was done in a
subsample of the homes. Information such as the presence of a gas
stove or smoking was also obtained. The presence of a gas stove was
used as a surrogate for indoor NO_x exposure. Children's peak flow
was associated with gas stove use (p = 0.066) for an analysis
excluding TSP. In adult asthmatics, gas stove use was significantly
associated with peak flow decrements (p < 0.001). This was true
across smoking groups, but the difference was greatest for smokers.

Lung function studies were conducted in a prospective survey
undertaken by Kamat et al. (1980) on 4129 subjects in three urban
areas of Bombay and a rural control area during February to July,
1977. The survey revealed that the population in low polluted areas
had higher lung function for FEF_25-75% and PEF. Thus, there was
suggestive evidence that the higher values obtained from lung function
tests in rural subjects as compared to urban subjects could be due to
increased levels of NO₂.

7.5 Other exposure settings

Certain recreational settings have been shown to result in NO₂
exposures that greatly exceed the chronic, low-level exposures
described in the previous epidemiological studies.

7.5.1 Skating rink exposures

Hedberg et al. (1989) reported cough, shortness of breath, and
other symptoms among players and spectators of two high school hockey
games played at an indoor ice arena in Minnesota, USA. These symptoms
were related to emissions from a malfunctioning engine of the ice
resurfacer. Although the exact levels of NO₂ were not known at the
time of the hockey game, levels of 7500 µg/m³ (4 ppm) were detected
2 days later with the ventilation system working, suggesting that
levels during the games were higher. Hedberg et al. (1989, 1990)
reported that pulmonary function testing performed on members of one
hockey team with a single exposure demonstrated no decrease in lung
function parameters at either 10 days or 2 months after exposure.
Dewailly et al. (1988) reported another incident at a skating rink in
Quebec, Canada, in 1988 involving referees and employees with
respiratory symptoms such as coughing, dyspnoea and a suffocating
feeling. Five days after the incident, NO₂ levels had come down to
5600 µg/m³ (3 ppm), suggesting much higher levels during the
incident.

In another skating rink study, Smith et al. (1992) reported the
outcome of a questionnaire administered to all students from two high
schools on 25 February, 1992, 3 days after 11 students participating
in a Wisconsin indoor ice hockey tournament had been treated in
emergency rooms for acute respiratory symptoms (i.e., cough,
haemoptysis, chest pain and dyspnoea). The game had been attended by
131 students, 57 of whom reported symptoms. A simulation test on
24 February yielded NO₂ levels of 2800 µg/m³ (1.5 ppm) in the air
over the rink after use of the ice resurfacing machine. Higher levels
may have been reached on the night of the game.

Brauer & Spengler (1994) measured indoor air NO₂ concentrations
at 20 skating rinks (most of all the operating ones) in the New
England area of the USA. Palmes tubes were used to measure NO₂ over
a 7-day sampling period at each rink, the samplers being placed on
the main resurfacer used in the rink, at the score keepers' bench
around a breathing height, at the opposite side of the rink