(PDF) Effects of Pesticides in the Field - EPiF - ResearchGate

EU & SETAC EUROPE WorkshopOctober 2003Le Croisic, France

Editors:Matthias Liess

Colin BrownPeter Dohmen

Sabine DuquesneAndy Hart

Fred Heimbach

Jenny KreugerLaurent LagadicSteve MaundWolfgang ReinertMartin StrelokeJosé V. Tarazona

EFFECTS OF

PESTICIDESFIELDIN THE

SETA

CEf

fect

s o

f Pe

stic

ide

s in

th

e F

ield

EPIF

About the editors

Matthias Liess leads the Effect Propagation Group,Aquatic Ecotoxicology, UFZ-Centre for EnvironmentalResearch, Leipzig, Germany.

Colin Brown, Professor of Environmental Sciencewithin the Environment Departmentof the Universityof York, UK.

G. Peter Dohmen senior scientist at the AgriculturalResearch Center, BASF-AG, Germany.

Sabine Duquesne, research scientist at the Effectpropagation group, Aquatic Ecotoxicology, UFZ-Centrefor Environmental Research, Leipzig, Germany.

Andy Hart leads the Risk Analysis Team at theCentral Science Laboratory (CSL), York, UK.

Fred Heimbach, senior scientist at the Institute ofEnviromental Biology in the Crop Protection Divisionof Bayer CropScience, Monchheim, Germany.

Jenny Kreuger, Assistant Professor at the SwedishUniversity of Agricultural Sciences, Division of WaterQuality Management.

Laurent Lagadic, research director at the INRA(National Institute for Agronomic Research) inRennes, France.

Steve Maund, Global Lead for aquatic ecology,Syngenta, Basel, Switzerland

Wolfgang Reinert, European Commission, Healthand Consumer Protection - Directorate General, UnitChemicals, Contaminants and Pesticides, Brussels,Belgium.

Martin Streloke, senior scientist, Federal Office ofConsumer Protection and Food Safety (BVL),Braunschweig, Germany.

Jose_ V. Tarazona, Director of the Department of theEnvironment. Spanish National Institute forAgricultural and Food Research and Technology (INIA),Madrid, Spain.

The EU Uniform Principles for the assessment of plantprotection products (PPP’s) require that if the prelimi-nary risk characterization indicates potential con-cerns, it has to be granted that "... under field condi-tions no unacceptable impact on the viability ofexposed organisms ..." occurs. To date, such assess-ments have been made by conducting higher-tierstudies. The aims of the workshop, were to: (i) reviewavailable field monitoring studies addressing theenvironmental effects of PPP’s due to agriculture, (ii)compare observed effects of PPP’s under field condi-tions with the impact predicted on the basis of thecurrent risk assessment guidance, and (iii) identifyrequirements for future monitoring studies and thedesign for higher tier tests.

EEffffeeccttss ooff PPeessttiicciiddeess iinn tthhee FFiieelldd –– EEPPIIFF is one of many SETAC publications that offer timelyreviews and new perspectives on current topics relat-ing to broad environmental toxicology and chemistryissues. SETAC publications often are based on the col-laborative efforts of top scientists and policymakers,and they undergo extensive prepublication reviews.SETAC assumes an active leadership in the develop-ment of educational programs and publishes thepeer-reviewed, international journals, EnvironmentalToxicology and Chemistry and Integrated Environ-mental Assessment and Management.For more information, contact the SETAC Office near-est you; details about the Society and its activitiescan be found at www.setac.org.

Administrative Offices

SETAC Office1010 North 12th AvenuePensacola, Florida 32501-3367T 850 469 1500 F 850 469 9778E [email protected]

SETAC OfficeAvenue de la Toison d’Or 67Brussels, BelgiumT 322 772 72 81 F 32 2 770 53 83E [email protected]

E n v i r o n m e n ta l S c i e n c e

"Effects of Pesticides in the Field" - EPIF

Editors:

Matthias Liess Colin Brown Peter DohmenSabine Duquesne Andy HartFred Heimbach

Jenny Kreuger Laurent Lagadic Steve MaundWolfgang ReinertMartin Streloke José V. Tarazona

Ecological Assessment ofAquatic Resources: Linking Scienceto Decision Making

Barbour, Norton, Preston, Thornton, editors,2004

Amphibian Decline: An Integrated Analysisof Multiple Stressor Effects

Greg Linder, Sherry K. Krest, Donald W.Sparling, 2003

Metals in Aquatic Systems:A Review of Exposure, Bioaccumulation, and Toxicity Models

Paquin, Farley, Santore, Kavvadas, Mooney, Winfield, Wu, Di Toro, 2003

Silver: Environmental Transport,Fate, Effects, and Models:Papers from Environmental Toxicologyand Chemistry, 1983 to 2002

Gorusch, Kramer, La Point, 2003

Contaminated Soils: From Soil-ChemicalInteractions to Ecosystem Management

Lanno, editor, 2003

Environmental Impacts of Pulpand Paper Waste Streams

Stuthridge, van den Heuvel, Marvin, Slade, Gifford, editors, 2003

Porewater Toxicity Testing: Biological,Chemical, and Ecological Considerations

Carr and Nipper, editors, 2003

Reevaluation of the State of the Sciencefor Water-Quality Criteria Development

Reiley, Stubblefield, Adams, Di Toro, Erickson,Hodson, Keating Jr, editors, 2003

Ecological Risk Assessmentof Contaminated Sediments

Ingersoll, Dillon, Biddinger, editors, 1997

For information about SETAC publications, includ-ing SETAC’s international journals,Environmental Toxicology and Chemistry and Inte-grated Environmental Assessment and Management,contact the SETAC Administrative Office nearest you:

SETAC Office 1010 North 12th Avenue Pensacola, FL 32501-3367 USAPhone 850 469 1500Fax 850 469 9778 Mail: [email protected]

SETAC Office Avenue de la Toison d’Or 67B-1060 Brussels, BelgiumPhone 32 27 72 72 81Fax 32 27 70 53 86Mail: [email protected]

w w w. s e tac .o r gE n v i r o n m e n ta l Q ua l i t y Th r o u g h S c i e n c e ®

ot h e r t i t l e s

3

OTHER TITLESFROM THE SOCIETY OF ENVIRONMENTAL

TOXICOLOGY AND CHEMISTRY (SETAC)

www.setac.org

EU & SETAC Europe Workshop (2003 : Le Croisic, France)

Effects of Pesticides in the Field editors, Matthias Liess … [et al.].p. cm.Includes bibliographical references.ISBN 1-880611-81-3 (alk. paper)1. Pesticides – Environmental aspects – Congresses. I.Liess, Matthias, 1960 – II. Title.QH545.P4E9 2003577.27’9–cd22 2005051627

© 2005 Society of Environmental Toxicologyand Chemistry (SETAC)

SETAC Press is an imprint of the Society of EnvironmentalToxicology and Chemistry.No claim is made to original U.S. Government works.

International Standard Book Number 1-880611-81-3Printed in Berlin, Germany12 11 10 09 08 07 06 05 10 9 8 7 6 5 4 3 2 1

The paper used in this publication meets the minimumrequirements of the American National Standard forInformation Sciences – Permanence of Paper for PrintedLibrary Materials, ANSI Z39.48-1984.

Reference listing: Liess M, Brown C, Dohmen P,Duquesne S, Hart A, Heimbach F, Kreuger J, Lagadic L,Maund S, Reinert W, Streloke M, Tarazona JV. 2005. Effectsof pesticides in the field. Brussels (BE): Society of Environ-mental Toxicology and Chemistry (SETAC). 136 p.

Design & LayoutDarius Samek2medien, BerlinJohanna MichelAlexa Sabarth

PrintingDruckerei Hermstein, Berlin

Information in this book was obtained from indi-vidual experts and highly regarded sources. It isthe publisher’s intent to print accurate and reliableinformation, and numerous references are cited;however, the authors, editors, and publisher can-not be responsible for the validity of all informa-tion presented here or for the consequences of itsuse. Information contained herein does not neces-sarily reflect the policy or views of the Society ofEnvironmental Toxicology and Chemistry(SETAC). Mention of commercial or noncommer-cial products and services does not imply endorse-ment or affiliation by the author or SETAC.

No part of this publication may be reproduced,stored in a retrieval system, or transmitted in anyform or by any means, electronic, electrostatic,magnetic tape, mechanical, photocopying, record-ing, or otherwise, without permission in writingfrom the copyright holder.

All rights reserved. Authorization to photocopyitems for internal or personal use, or for the per-sonal or internal use of specific clients, may begranted by the Society of Environmental Toxicolo-gy and Chemistry (SETAC), provided that the ap-propriate fee is paid directly to the CopyrightClearance Center, Inc., 222 Rosewood Drive, Dan-vers, MA 01923 USA (Telephone 978 750 8400) orto SETAC. Before photocopying items for educa-tional classroom use, please contact the CopyrightClearance Center (http://www.copyright.com) orthe SETAC Office in North America (Telephone850 469 1500, Fax 850 469 9778, [email protected]).

SETAC’s consent does not extend to copying forgeneral distribution, for promotion, for creatingnew works, or for resale. Specific permission mustbe obtained in writing from SETAC for such copy-ing. Direct inquiries to the Society of Environmen-tal Toxicology and Chemistry (SETAC), 1010 North12th Avenue, Pensacola, FL 32501-3367, USA.

4

LIBRARY OF CONGRESSCATALOGING-IN-PUBLICATION DATA

l i b r a ry o f c o n g r e s s

http://www.copyright.com

Books published by the Society of EnvironmentalToxicology and Chemistry (SETAC) provide in-depth reviews and critical appraisals on scientificsubjects relevant to understanding the impacts ofchemicals and technology on the environment.The books explore topics reviewed and recom-mended by the Publications Advisory Council andapproved by the SETAC North America Board ofDirectors, SETAC Europe Council, or the SETACWorld Council for their importance, timeliness,and contribution to multidisciplinary approachesto solving environmental problems. The diversityand breadth of subjects covered in the publicationsreflect the wide range of disciplines encompassedby environmental toxicology, environmentalchemistry, hazard and risk assessment, and life-cycle assessment. SETAC books attempt to presentthe reader with authoritative coverage of the liter-ature, as well as paradigms, methodologies, andcontroversies; research needs; and new develop-ments specific to the featured topics. The booksare generally peer reviewed for SETAC by ac-knowledged experts.

SETAC publications, which include Technical IssuePapers (TIPs), workshop summaries, newsletter(SETAC Globe), and journals (Environmental Toxi-cology and Chemistry and Integrated EnvironmentalAssessment and Management), are useful to envi-ronmental scientists in research, research manage-ment, chemical manufacturing and regulation,risk assessment, life-cycle assessment, and educa-tion, as well as to students considering or prepar-ing for careers in these areas. The publicationsprovide information for keeping abreast of recentdevelopments in familiar subject areas and forrapid introduction to principles and approaches innew subject areas.

SETAC recognizes and thanks the pastcoordinating editors of SETAC books:

C.G. Ingersoll,Columbia Environmental Research CenterU.S. Geological Survey, Columbia, MO, USA

T.W. La Point,Institute of Applied SciencesUniversity of North Texas, Denton, TX, USA

B.T. Walton,U.S. Environmental Protection AgencyResearch Triangle Park, NC, USA

C.H. Ward,Department of EnvironmentalSciences and EngineeringRice University, Houston, TX, USA

5

SETACPUBLICATIONS

s e ta c p u b l i c at i o n s

This book presents the proceedings of a SETACWorkshop convened by the Society of Environmen-tal Toxicology and Chemistry (SETAC) and theEuropean Union in Le Croisic, France, in October2003. The 77 scientists involved in this workshoprepresented 17 countries from Europe, the UnitedStates and Canada and offered expertise in ecolo-gy, ecotoxicology, environmental regulation, andrisk assessment.

The workshop was made possible by the gener-ous support of many organizations, including

European CommissionBASFBayer CropScienceBVL, Germany DGAL, FranceDow AgroSciencesDuPontINRA, FranceMakhteshim-AganSyngentaUFZ, Germany

The workshop also was supported by the verycapable management of the local organising com-mittee – Anne Alix, Thierry Caquet, Agnès Girard,Mark Hanson, Laurent Lagadic, Patricia Marhin,and Maria-José Servia-Garcia. And finally, theproduction of this book was made possible withthe restless efforts of Sabine Duquesne and MimiMeredith.

Matthias LiessColin BrownG. Peter DohmenSabine DuquesneAndy HartFred HeimbachJenny KreugerLaurent LagadicSteve MaundMartin StrelokeJosé V. Tarazona

6

ACKNOWLEDGMENTS

a c k n o w l e d g m e n t s

ta b l e o f c o n t e n t s

7

List of Participants and Breakout Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Part AExecutive Summary

1 Workshop Focus and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2 Reasons for the Workshop. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3 Workshop structure and approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4 Field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.2 Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244.3 Examples of field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5 Methodological considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

6 Linking tiers and extrapolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

7 Implications for risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

8 Main outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Part BWW oo rr kk ss hh oo pp SS yy nn tt hh ee ss ii ss

1 Development of workshop conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2 Field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312.2 Studies on effects of pesticides in the field. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.2.1 Monitoring studies for aquatic invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.2.2 Monitoring and experimental studies for terrestrial

invertebrates and plants in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.2.3 Monitoring and experimental studies for birds and mammals in Europe . . . . . . . . 36

3 Essentials of field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.1 Exposure assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.1.1 Chemical use patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.1.2 Good Agricultural Practices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.1.3 Modelling exposure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.1.4 Routes of exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.2 Biological effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2.1 Defining reference and control sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2.2 Detecting effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423.2.3 Recovery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.2.4 Bioassays and biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.3 Relevance of the study situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3.1 Relevance of the landscape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3.2 Relevance of the pollution scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3.3 Ecological relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3.4 Spatial and temporal relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

8


4 Linking Tiers and extrapolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.1 Examples of comparison between observed and predicted effects . . . . . . . . . . . . . . . . . . . . 48

4.1.1 Effects on aquatic organisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.1.2 Effects on soil invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.1.3 Effects on birds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

4.2 Limitations in comparing observed and predicted effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.2.1 Limitations of field studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.2.2 Limitations of mesoscosm experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5 Research requirements and further strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1 Research requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.1.1 Improved exposure assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.2 Defining target images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.3 Assessment of direct effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.4 Assessment of indirect effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.5 Additional biological and ecological knowledge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.6 Environmental parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.1.7 Mixture toxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.1.8 Magnitude of recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.1.9 Benchmark cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

5.2 Further strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.2.1 Defining acceptability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.2.2 Spatial unit of assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525.2.3 Harmonization within Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535.2.4 Good Agricultural Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Implications for Regulatory Risk Assessment and Management . . . . . . . . . . . . . . . . . . . . . . . . . . 536.1 Current value and scope of field studies in risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6.1.1 Informing and evaluating the risk assessment process . . . . . . . . . . . . . . . . . . . . . . . . . 536.1.2 Verifying risk mitigation measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536.1.3 Risk communication strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.1.4 Broader sustainability issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

6.2 Implications for future regulatory risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55


9

Appendix AAA bb ss tt rr aa cc tt ss oo ff PP ll aa tt ff oo rr mm PP rr ee ss ee nn tt aa tt ii oo nn ss

Plenary keynoteStress and disturbance in nature (R. Sibly) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Observation of effects in the fieldLinking terrestrial vertebrate risk assessment to effects in the field(M. Clook, M. Fletcher). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66Field studies of pesticide effects on terrestrial invertebrates(J. M. Holland, J. A. Ewald, N. J. Aebischer) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Field studies of herbicide effects on terrestrial plants(F. M.W. De Jong, G. R. De Snoo) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Landscape-level pesticide risk assessment using freshwater macroinvertebratecommunity structure in the orchard agriculture of Altes Land, Germany(C. Schäfers, U. Hommen, M. Dembinski, J. Gonzalez-Valero) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75The Lourens River, South Africa: A case study for aMediterranean agricultural catchment (R. Schulz) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Ability to predict effects: linking tiers and extrapolationExposure monitoring for aquatic risk assessment (J. Kreuger) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80Exposure modelling for aquatic risk assessment (C. Brown) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82Predictability and acceptability of effects of

pesticides in freshwater ecosystems (T. C. M Brock) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84Predicting effects of pesticides in streams (M. Liess) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87From laboratory to field: extrapolation of pesticide risks

to non-target invertebrates communities (J. Römbke, G. Frampton) . . . . . . . . . . . . . . . . . . . . . . . 90Ability to predict effects in birds: linking tiers and extrapolation

(P. Mineau, M. Whiteside) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Implications for risk assessmentEcosystem dynamics and stability: Are the effects

of pesticides ecologically acceptable? (H. T. Ratte, F. Lennartz, M. Roß-Nickoll) . . . . . . . . . . . . . 98The role of data from monitoring studies in the regulatory framework

(M. Streloke, J. V. Tarazona, W. Reinert). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101Geographical differences in the evaluation and prediction

of the effects of pesticides (J. V. Tarazona) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

Appendix BLL ii ss tt oo ff PP oo ss tt ee rr PP rr ee ss ee nn tt aa tt ii oo nn ss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

Appendix CRR ee pp oo rr tt ss oo ff tt hh ee BB rr ee aa kk oo uu tt GG rr oo uu pp ss

Aquatic ecosystems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Terrestrial Invertebratesand Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124Birds and Mammals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

10

E d i to r s

CONTENTS

Matthias Liess obtained his PhD and Habilitation from the University of Braun-schweig, Germany, where his studies focused on ecology and ecotoxicology.

Since 1985 he participated and lead laboratory and field studies to analyseprocesses in freshwater and marine ecosystems and how they are disturbedby natural and anthropogenic stressors. He is an author on more than eightyscientific publications.

Matthias Liess leads the Effect propagation group, Aquatic Ecotoxicology,UFZ-Centre for Environmental Research, Leipzig, Germany.

Colin Brown is Professor of Environmental Science within the EnvironmentDepartment of the University of York.

He jointly leads the EcoChemistry research team that is a joint venture ofthe Central Science Laboratory and the University of York. Colin graduated fromLeeds University with a BSc in Agricultural Chemistry and gained his PhD atNewcastle University, researching the transport of pesticides from drainedland.

He worked for Cranfield University from 1993 to 2004. Colin’s research fo-cuses on fate and effects of pesticides in the environment, particularly man-agement of aquatic exposure to pesticides and use of models within riskassessment. He recently chaired the DG-Sanco FOCUS work group on landscapeand mitigation factors in ecological risk assessment.

G. Peter Dohmen studied biology and ecology in Aachen and London andobtained his PhD in Munich working on effects of air pollutants on plants andplant/insect interactions.

Following a post-doc research project he joined BASF-AG, to work as eco-toxicologist at the Agricultural Research Center. In this position he has workedon terrestrial and aquatic ecotoxicological studies from standard laboratorytests to field studies.

He has supervised a number of mesocosm studies and performed fieldworkin paddy rice and in tropical systems. Peter Dohmen has been involved in arange of method development initiatives for ecotoxicological test systems andparticipated in several national and international working groups on standardand higher risk assessment methodologies and strategies.

E d i to r s

11

CONTENTS

Andy Hart leads the Risk Analysis Team at the Central Science Laboratory (CSL),York, UK, a research agency of the UK Department for Environment, Food andRural Affairs.

He has a BSc in Environmental Biology and a DPhil in Behavioural Ecology,and joined the UK agriculture ministry in 1982 to research effects of pesticideson birds. He expanded his work first to other types of pesticide effects and thento the use of probabilistic methods of risk assessment.

Andy is a member of the European Food Safety Authority (EFSA) ScientificPanel on plant health, plant protection products and their residues, and co-ordinator of EUFRAM, an EU project developing a framework for probabilisticassessment of environmental impacts of pesticides (www.eufram.com).

Sabine Duquesne obtained her PhD in 1992 at the University of Lille, France,in marine ecotoxicology.

She then worked as a researcher at the University of Queensland until 2000and conducted various projects in the field, in Australia and Antarctica, to studythe suborganismal and organismal responses of marine species to environmen-tal changes. Her current work focuses as well on freshwater ecosystems and onchanges at the population level.

She has expertise in various research fields including bioindicators, biologi-cal markers, higher tier test systems, risk assessment. Sabine Duquesne is nowbased at the UFZ-Centre for Environmental Research in Germany.

Fred Heimbach is a research scientist at the Institute of Enviromental Biologyin the Crop Protection Division of Bayer CropScience in Monchheim, Germany.

He optained his MSc degree and PhD in conducted research on marineinsects at the Instutute of Zoology, Physiological Ecology at the Universityof Cologne.

Since 1979 he has worked at Bayer CropScience on the side-effects of pesti-cides on nontarget organisms. In addition to his work, he gives lectures on eco-foxicolgy at the University of Cologne.

Dr. Heimbach has researched the development of single-species toxicitytests for both terrestrial and aquatic organisms and has worked with micro-cosms and mescocosms on development of multispecies tests for aquatic or-ganisms. As an active member of European and international working groups,he participated in the development of suitable methods for testing pesticidesand other chemicals for their potential side-effects on nontarget organisms.

www.eufram.com

12

E d i to r s

CONTENTS

Jenny Kreuger is an assistant professor at the Swedish University of Agri-cultural Sciences, Division of Water Quality Management.

Her research work has been focused on the environmental impact of agri-cultural activities on the quality of surface water and groundwater bodies.Special emphasis has been on studies of pesticide transport in soil and waterat different scales, including also work on the effects of pesticide managementpractices on water quality at the catchment scale and studies of atmosphericdeposition of pesticides.

She is responsible for the Swedish national environmental monitoring pro-gram for pesticides in stream water, groundwater, and atmospheric deposition.

Laurent Lagadic is a Research Director at the INRA (National Institute for Agro-nomic Research) in Rennes, France.

He obtained his MSc Degree in Population Biology and PhD in FundamentalToxicology from the University of Paris XI, Orsay. He has been involved for about20 years in the study of the effects of pesticides in invertebrates, both targetand non-target organisms. His work has been focused on the physiologicalchanges (defence mechanisms, hormone and energy metabolism) induced bypesticides, and their consequences on individual reproductive performancesand population dynamics. He currently leads a research group in aquatic eco-toxicology, where various approaches (biochemistry, immunology, populationdynamics & genetics, community ecology) are applied to the study of multi-scale effects of pesticides in aquatic environments. Current research of thegroup includes endocrine disrupting effects of pesticides and interactive effectsof toxicants in mixtures.

Dr. Lagadic is involved in the coordination of national research programmesin ecotoxicology, and he has participated to several international initiatives inenvironmental toxicology. He also provides external expertise in aquatic eco-toxicology, either for field assessment of effects of pesticides, or for the designand use of micro- and mesocosms in regulatory risk assessment of pesticides.

Steve Maund graduated from the University of York with a BSc in Biology andfrom the University of Wales Institute of Science and Technology with an MScin Applied Hydrobiology in 1988.

After working on EU funded research at the University of Cardiff, he joinedICI (subsequently to become Zeneca, and now Syngenta) in 1991, working firstin the US, then the UK, and since 2002 he has been based in Basel, Switzerland.Steve has been involved in many of the recent techno-regulatory developmentsin aquatic risk assessment of plant protection products in Europe includingHARAP, CLASSIC, and FOCUS.

He is an author on more than fifty scientific publications.

E d i to r s

13

CONTENTS

José V. Tarazona. DVM, Ph.D. Director of the Department of the Environmentat the Spanish National Institute for Agricultural and Food Research and Tech-nology (INIA).

His scientific activity covers aquatic and terrestrial ecotoxicology, ecologicalhazard and risk assessment, environmental impact assessment, environmentalindicators and the environmental dimension of Sustainable Development.He is Involved in regulatory aspects on hazard and risk assessment pro-grammes for industrial chemicals, pesticides, biocides, and pharmaceuticalswithin the EU.

Tarazona is a member of the scientific advisory board of the EU since 1992;was member of the CSTE and CSTEE, chair of the WG on Environmental Risk As-sessment at the Task Force Harmonisation of Risk Assessment Procedures, andcurrently is vice-president of the Scientific Committee on Human and Environ-mental Risks.

In addition he has been adviser/consultant for other EU bodies includingEFSA and the EEA, OECD, WHO and UN, and several European and Latin Ameri-can countries.

Martin Streloke is an aquatic ecotoxicologist with special interests in regu-latory risk assessment and management. In 1991, he received a PhD in Zoologyfrom Hanover University, and joined the Biologische Bundesanstalt für Land-und Forstwirtschaft (BBA) in Braunschweig, Germany, in 1992, where he workedin the field of regulatory risk assessment and management on aquatic organ-isms. In 2002 he joined the German Federal Office of Consumer Protection andFood Safety (BVL), the responsible authority for the authorization of plant pro-tection products, where he is responsible for risk management issues (e.g. riskmitigation measures to protect the environment). He was a member of theorganizing committees of the Higher-tier Aquatic Risk Assessment for Pesti-cides (HARAP), Community-Level Aquatic System Studies-Interpretation Criteria(CLASSIC), and Workshop on Risk Assessment and Risk Mitigation Measures inthe Context of the Authorization of Plant Protection Products (WORMM) work-shops dealing with the development of improved test methods, higher-tier riskassessments, and realistic risk management tools. He assists in the develop-ment of European Union (EU) documents on regulation for plant protectionproducts (e.g. , Guidance Document on Aquatic Organisms).

Dr. Wolfgang Reinert – European Commission, Health and Consumer Protec-tion – Directorate General, Unit Chemicals, Contaminants and Pesticides

Wo r ks h o p Pa r t i c i p i a n t s

Participants of the EU & SETAC Europe Workshop“Effect of Pesticides in the Field – EPiF”

CONTENTS

e f f e c t s o f p e st i c i d e s i n t h e f i e l d

CONTENTS

LIST OFPARTICIPANTS

CONTENTS

Aagaard, AlfPesticide Division, Danish EPADenmark

Alix, Anne 2

Structure Scientifique Mixte INRA-DGAL France

Arts, GertieAlterra Green World ResearchWageningen University and Research CentreThe Netherlands

Bassères, AnneTOTALGroupement de Recherches de Lacq France

Biggs, JeremyThe Ponds Conservation Trust:Policy and ResearchUnited Kingdom

Bischof, GabrielaFederal Biological Research Centerfor Agriculture and ForestryGermany

Blanck, HansBotanical Institute, Göteborg UniversitySweden

Brock, Théo 3

Alterra Green World Research,Wageningen University and Research Centre,The Netherlands

Brown, Colin 1Central Science LaboratoryUnited Kingdom

Brown, KevinEcotox LimitedUnited Kingdom

Campbell, Peter 3

Syngenta AG, Ecological SciencesUnited Kingdom

Candolfi, Marco 6

RCC Ltd.Switzerland

Capri, EttoreUniversita Cattolica del Sacro CuoreItaly

Caquet, Thierry 2

INRAFrance

Clook, Mark 4

Pesticide Safety DirectorateUnited Kingdom

Crane, Mark 6

Crane ConsultantsUnited Kingdom

De Jong, FrankNational Institute for Public Healthand the Environment (RIVM)The Netherlands

Dinter, AxelDuPont de NemoursGermany

Dohmen, Peter 1

BASF AGGermany

Douglas, Mark 5

Dow AgroSciencesUnited Kingdom

Duquesne, SabineUFZ Centre for Environmental ResearchGermany

Ericson, GunillaSwedish National Chemicals Inspectorate, Swe-den

Farrelly, Eamonn 6

Braddan Scientific Ltd.United Kingdom

Forbes, ValeryRoskilde UniversityDenmark

Giddings, Jeffrey 4

Parametrix, Inc.USA

workshop participants a – g

17

CONTENTS

h – p

18

workshop participants

CONTENTS

Hanson, Mark 2

INRAFrance

Hart, Andy 1

Central Science LaboratoryUnited Kingdom

Heimbach, Fred 1

Bayer CropScienceGermany

Hofkens, SofieService Public Fédéral Santé Publique, Sécuritéde la Chaîne Alimentaire et Environnement,Belgium

Holland, JohnThe Game Conservancy TrustUnited Kingdom

Hubert, AgnèsStructure Scientifique Mixte INRA-DGALFrance

Joermann, Gerd 6

Federal Office for Consumer Protectionand Food Safety (BVL)Germany

Kreuger, Jenny 1

Swedish University of Agricultural SciencesSweden

Kubiak, RolandSLFAGermany

Kula, ChristineFederal Office for Consumer Protection andFood Safety (BVL)Germany

Kyriazi, KaterinaHellenic Ministry of AgricultureGreece

Lagadic, Laurent 1, 2

INRAFrance

Lawlor, PeterPesticide Control ServiceIreland

Lennartz, FredGAIAC and PRO TERRAGermany

Lewis, GavinJSC International Ltd.United Kingdom

Liess, Matthias 1

UFZ Centre for Environmental ResearchGermany

Maltby, Lorraine 5

The University of SheffieldUnited Kingdom

Maund, Steve 1

Syngenta Crop Protection AGSwitzerland

Michalski, BrittaUmweltbundesamt (UBA) –Federal Environment AgencyGermany

Migula, PawelUniversity of SilesiaPoland

Miles, MarkDow AgroSciencesUnited Kingdom

Mineau, PierreNational Wildlife Research Centre,Canada

Norman, Steve 4

Makhteshim-AganBelgium

Pestanudo, Susana LuisDireccao-General de Proteccao das Culturas,Portugal

Pilling, EdSyngenta AGUnited Kingdom

workshop participants r – w

19

CONTENTS

Ratte, Hans ToniAachen UniversityGermany

Redolfi, ElenaInternational Centre for Pesticidesand Health Risk Prevention (ICPS)Italy

Römbke, JörgECT ÖkotoxikologieGermany

Romijn, KeesBayer CropScienceGermany

Schäfers, ChristophFraunhofer Institut für Molekularbiologieund Angewandte ÖkologieGermany

Schmitz, SusanneUmweltbundesamt (UBA) –Federal Environment AgencyGermany

Schulte, ChristophUmweltbundesamt (UBA) –Federal Environment AgencyGermany

Schulz, RalfSyngenta AG, Ecological SciencesUnited Kingdom

Schüürmann, GerritUFZ Centre for Environmental ResearchGermany

Servia-Garcia, Maria-José 2

INRAFrance

Sibly, Richard 5

The University of ReadingUnited Kingdom

Solomon, Keith 3

University of GuelphCanada

Steininger, AngelikaAustria Agency for Health and Food SafetyAustria

Streloke, Martin 1

Federal Office for Consumer Protectionand Food Safety (BVL)Germany

Tarazona, Jose 1

INIASpain

van Straalen, Nico 5

Vrie Universiteit AmsterdamThe Netherlands

van Vliet, PeterBoard for the Autorisation of PesticidesThe Netherlands

Virtanen, VirpiFinnish Environment InstituteFinland

Wogram, JörnUmweltbundesamt (UBA) –Federal Environment AgencyGermany

Woin, PerLund UniversitySweden

Wolf, ChristianBayer CropScienceGermany

1) Member of the Organising Committee2) Member of the Local Organising Committee3) Session Chairman4) Session Rapporteur5) Breakout Group Chairman6) Breakout Group Rapporteur

CONTENTS


CONTENTS

aEXECUTIVESUMMARY

CONTENTS

1 WORKSHOP FOCUSAND OBJECTIVES

The overall aims of the workshop were� to consider to what extent current data

demonstrate exposure and effects of pesti-cides under field conditions, and

� to examine whether these effects (or lackthereof) would be predicted on the basisof current risk assessment procedures.

Five specific topics were addressed to meetthese aims:1) review of available field studies for which

the environmental effects of pesticides inagricultural landscapes, in both terrestrialand aquatic compartments;

2) examination of the essentials of fieldstudies (exposure and biological effectsassessment);

3) linking of tiers and extrapolation (com-paring observed effects and predicted ef-fects and limitations of such compar-isons);

4) consideration of research requirementsand further strategies; and

5) determination of the implications for reg-ulatory risk assessment and management.

2 REASONS FOR THE WORKSHOP

Procedures for first-tier and higher-tier riskcharacterisation of plant protection products(PPPs, or pesticides) for the aquatic and ter-restrial environment are now reasonablywell defined in the European Union (EU).The EU Uniform Principles for the assess-ment of plant protection products requirethat if the preliminary risk characterisationindicates potential concerns, registrationcannot be granted unless it can be demon-strated that “… under field conditions no un-acceptable impact on the viability of exposedorganisms…” occurs (Annex VI, Directive91/414/EEC). To date, such assessments havebeen made by conducting higher-tier studies,which have included a range of laboratory,semi-field, and field experiments. In theHigher-tier Aquatic Risk Assessment for Pes-ticides (HARAP) workshop (Campbell et al.1999), field studies were one of the ap-

proaches proposed to further characterisepotential risks identified in a first-tier assess-ment. Over recent years, a number of fieldstudies have been conducted to address ex-posure and effects of pesticides in the agri-cultural environment (“the field”). The cur-rent workshop focused on the “Effects of Pes-ticides in the Field”. Results of field studieswere discussed with respect to (1) the occur-rence and magnitude of observed effects and(2) the links between observed and predict-ed effects (derived from the first-tier andhigher-tier risk assessment schemes). Thisdiscussion was carried out to assess whetherthe risk assessment scheme is under- or over-protective with respect to effects observed inthe environment.

3 WORKSHOP STRUCTUREAND APPROACH

The workshop comprised 75 scientists fromEurope and North America, representinggovernment, industry, and academia and ex-perienced in risk assessment and field effectsof pesticides.

The state of the art of the current knowl-edge and information available about fieldeffects of pesticides was reviewed via a seriesof platform and poster presentations (seeAppendixes A and B). These presentationswere separated into 3 main topics: � observation of effects in the field;� ability to predict effects: linking tiers and

extrapolation; and� implications for risk assessment.Following the platform presentations, thetopics were discussed further in differentbreakout groups (aquatic organisms, terres-trial invertebrates and plants, and birds andmammals). Rapporteurs presented the out-comes from their breakout groups, whichwere further discussed in plenary sessionsamong all participants.

e x e c u t i v e s u m m a ry 1 · wo rk s h o p f o c u s a n d o b j e c t i ve s

23

CONTENTS

4 FIELD STUDIES

4.1 DEFINITIONS

In the context of this workshop, investiga-tions into the effects of pesticides in the fieldregrouped under the name “field studies” in-cluded 2 types of studies, which were de-fined as follows:

Monitoring study: an investigation intothe overall impact of pesticide use on a spe-cific ecosystem through surveying or moni-toring that consists of characterisation of ex-posure (chemical monitoring, exposure mod-elling) and observations of effects (biologi-cal monitoring) occurring in the field ortreated area as a consequence of use and/ormisuse of pesticides.

Experimental study: an experiment intothe impact of a specific product or active sub-stance applied under controlled conditionsin the field. Such studies are performed inthe natural environment within an agricul-tural context (and thus contrast with meso-cosm studies).

Monitoring studies were the main focus ofthe workshop.

4.2 OVERVIEW

Over recent years, some studies of environ-mental exposure to pesticides in the agricul-tural environment aimed at quantifying anyeffects on non-target species in the differentecosystems (aquatic organisms, terrestrialinvertebrates and plants, vertebrates). Ef-fects were frequently observed and de-scribed, but a firm link to pesticide usageoften was not established. Indeed, demon-strating causality between exposure and ef-fects is difficult, especially in studies inwhich long-term and indirect effects are re-sponsible for biological impairments. Rea-sons for this difficulty include problems inaccurately characterising exposure, the pres-ence of confounding factors, and naturalvariability in the system.

The workshop revealed that direct effectsdue to the use of pesticides have beendemonstrated in some cases (e.g., aquatic in-sect larvae and non-target plants). Indirect

effects due to changes of predator–prey bal-ance and competitive interactions have beenobserved in certain groups of organisms.However, these direct and indirect effectswere often, but not always, decreased oreven annulled through long-term recoveryprocesses.

4.3 EXAMPLES OF FIELD STUDIES

The investigations summarised below con-sidered monitoring studies a priority becausethis was the main focus of the workshop.However, some experimental studies werealso reviewed, especially when the numberof monitoring studies was too limited.

Aquatic organismsThe monitoring studies presented at theworkshop dealt mainly with the effects of in-secticides on macroinvertebrates. They wereperformed mostly in streams of intensivelycultivated areas. Effects of pesticides wereidentified in several of the field studies(Altes Land and the Braunschweig area inGermany, the Lourens River in South Africa).When recovery was studied, it was shownthat in some cases the invertebrate commu-nity composition did not recover within 1year. In the studies that established a clearlink between exposure and effects, it wasoften difficult to assign a specific level of ex-posure to the observed effects. This was dueto fundamental problems and uncertaintiesin accurately characterising exposure.

Terrestrial invertebratesand plants

For terrestrial invertebrates, the number ofavailable monitoring studies is limited, andthose available mostly studied effects onhoneybees. The UK Wildlife Incident Investi-gation Scheme (WIIS) is an example of sucha study; one of its significant findings wasthat the current risk assessment scheme forbees appears to be protective.

Monitoring programmes for other speciesof terrestrial invertebrates (non-targetarthropods and below-ground invertebrates)are not available. However, studies focusedon the ecology of individual species showed

4 · f i e l d s t u d i e s

24

pa rt a

CONTENTS

long-term effects of pesticides associatedwith a decline in abundance of some inverte-brate families or species (e.g., Sussex study).In some cases, long-lasting effects were alsoshown. For example, in the SCARAB (Seek-ing Confirmation About Results at Box-worth) project, the Collembola was thearthropod group most affected by conven-tional pesticide use. Long-term effects (>4y), reflected by a lack of recovery of somespecies, occurred at a location with repeateduse of organophosphate insecticides.

Some field studies are available for plants,but there are no monitoring studies. In alarge-scale experimental study, clear short-term effects were shown, with recovery oc-curring within a year after application.

Birds and mammalsThe most significant monitoring scheme per-formed in the UK (WIIS) investigated directeffects of pesticides on wildlife, pets, andsome livestock and showed evidence of pesti-cide poisoning.

In 1968 the Sussex study (also called the“partridge survival project”) began to inves-tigate the reasons for the decline in the greypartridge population, and it identified indi-rect effects of pesticides (reduction in chickfood availability) as the main cause. More re-cent projects have also had a similar hypoth-esis and findings about the relevance of indi-rect and/or chronic effects (e.g., the cornbunting study).

5 METHODOLOGICALCONSIDERATIONS

One outcome of the workshop is that moreguidance and research are needed to assistwith the design and conducting of field stud-ies (monitoring and experimental studies).The following requirements were identified:� Exposure: Data on pesticide use in the

study area (chemical use pattern) and ap-propriate characterisation of exposure(e.g., detection of peak concentrations,number of applications) are part of theconsiderations. Geographic informationsystem (GIS) technologies may have ap-plications in analysing landscape factors

that influence exposure, such as proximi-ty of habitats to treated fields and connec-tivity of habitats.

� Reference or control sites: These must beestablished so that studies are suitablyrepresentative. They could then be extrap-olated to different regions or used to de-termine whether risk mitigation works.Both reference and impacted sites shouldnot be subject to major stressors otherthan the one tested.

� Cause and effect: To establish the cause ofobserved effects, the observations shouldbe designed so that, wherever possible,the mechanisms causing effects can be re-vealed. Direct effects should be assessed,but the potential consequences of indirecteffects as well as chronic (long-term) anddelayed effects should also be considered,depending on the problem formulation.The use of additional measurement or ex-perimentation (e.g., biomarkers, bioas-says, mesocosms) can provide useful com-plementary tools to better link exposureand effects.

� Confounding factors: Current and/or pastexposure to multiple substances and stres-sors (e.g., weather, water quality, speciesinteractions) interferes with or decreasesthe power of detection of field effects ofpesticides. The studies should be designedso that these problems are minimised.

� Natural variability: Field studies are con-ducted against a backdrop of natural vari-ability that includes the normal operatingrange of the system under consideration(intersite, interreplicate, temporal, andspatial heterogeneity). This variabilityleads to uncertainties and/or limitationsthat need to be considered during evalua-tion of the study.

� Biological and ecological information:Such data are needed for many taxa tobetter evaluate their potential sensitivityto pesticides and also their recovery po-tential. Information that is lacking or in-complete can include life-history charac-teristics, dispersal ability, and behaviouralecology, physiology, population genetics,and ecotoxicity data.

� Environmental parameters: These can betypical for specific regions and can influ-

e x e c u t i v e s u m m a ry 5 · m e t h o d o l o g i c a l c o n s i d e ra t i o n s

25

CONTENTS

ence the fate of the pesticide, the ecologi-cal performance of organisms, and theirresponse to toxicants (e.g., drying, varia-tions of temperatures). Environmental pa-rameters should be considered becausethey may alter the sensitivity of organismsand influence the fate of pesticides (e.g.,disappearance rate, bioavailability).

� Representativeness and regional under-standing: This will assist our ability to ex-trapolate results to another region, anoth-er ecosystem, and different scales of as-sessment. The system monitored shouldbe characterised with respect to the agri-cultural landscape, pesticide usage, eco-logical and biological properties, and spa-tial and temporal scales. Influence of cli-matic conditions and geographic positionshould also be understood (i.e., the envi-ronmental parameters mentioned above).Appropriate landscape characterisationwill eventually support the developmentof representative, region-specific sce-narios.

� Defining acceptability: Criteria for defin-ing the acceptability of observed effectsshould include both scientific recommen-dations and societal considerations. Forexample, the selection of reference sitesand target images depends partly on whatsociety considers to be acceptable. It isnecessary both to establish the targetimage and to decide how much deviationfrom that target image is acceptable.

6 LINKING TIERS ANDEXTRAPOLATION

Effects of pesticides were identified in sever-al of the field studies that were presented.These include direct effects (i.e., aquaticstudies: Liess, platform presentation, andSchäfers, platform presentation; soil inverte-brates: Römbke, platform presentation;plant study: De Jong, platform presentation;studies on birds and mammals: WIIS fromthe UK and indirect effects [observed in vari-ous groups of organisms]). These studiesshould offer the opportunity to calibrate therelationship between biological effects ob-served in the field and effects predicted in

test systems based on the current risk assess-ment schemes. However, there are variouslimitations in undertaking such calibration.For example, on the basis of the knowledgeusually available, it is often difficult to deter-mine whether (1) effects are due to normaluse or to misuse of pesticides and (2) expo-sure in the field is accurately characterised.Furthermore, effects of pesticides in the fieldare mostly related to multisubstance contam-ination, but current risk assessment does notconsider mixture toxicity. Indirect effects canhave important ecological consequences, butthey can be difficult to quantify and they areoften taken into account within current riskassessment schemes by applying safety fac-tors to direct effects measured in mono-species tests. For these reasons, it is difficultto verify whether current risk assessmentprocedures provide suitable environmentalprotection (i.e., whether they are overprotec-tive, underprotective, or appropriate).

7 IMPLICATIONS FORRISK ASSESSMENT

Field studies can provide useful informationand knowledge for the risk assessment that

� can be used as a “reality check” for therisk assessment process and may identifyareas where the present process may beover- or underprotective; hence, theycould be used to calibrate current risk as-sessment;

� may provide a means to evaluate the ef-fectiveness of risk mitigation measures inthe field;

� may give information on the temporaltrends of environmental impacts associat-ed with use of pesticides at a local/region-al scale when performed over a few years;or

� are relatively easy to communicate to thepublic and other stakeholders, as real situ-ations are represented (rather than surro-gates of the reality).

However, it seems difficult to incorporatesuch studies into decision making with re-spect to registration for a single active sub-

6 · l i n k i n g t i e r s a n d e x t ra p o l a t i o n

26

pa rt a

CONTENTS

stance because the field situation is oftencharacterised by multiple exposure and mix-ture of substances and hence differs fromhigher-tier tests performed as part of the riskassessment. Both types of field studies (mon-itoring and experimental studies) can be use-ful in regulatory risk assessment for the post-authorisation phase or for exploring moregeneral trends in the environment. Regulatory risk assessment and field studiesprovide complementary information. Riskassessment seeks to provide a generalised,protective framework on which to base deci-sion making. Results from field studies high-light areas that could be considered withinadvanced risk assessment. These include abetter characterisation of chemical exposure,further developments in incorporating inter-nal and external recovery, and characterisa-tion of indirect effects.

Risk assessment should eventually considerlandscape characteristics. Use of environ-mental databases containing various types ofinformation (e.g., biological, ecological, andlandscape characteristics; farming practises;pesticide usage) will help us to evaluate fieldstudies, particularly monitoring studies, andwill improve risk assessment by increasingrealism.

8 MAIN OUTCOMES

Concerning the effects of pesticides in thefield, the following conclusions werereached within the workshop:� Effects of pesticides were identified in

several of the field studies presented. Di-rect and indirect effects have been ob-served, as well as recovery processes thatoften decrease or annul these effects to agreater or lesser extent.

� The influence of natural variability, con-founding factors, etc., in the establish-ment of cause-and-effect relationshipscould be reduced by strategies such as cat-egorising species according to ecologicaltraits, identification of suitable controlsites, and more effective sampling strate-gies.

� An improved assessment of exposure, ap-propriate for each ecosystem, will enableus to establish a better correlation be-tween the magnitude of observed effectsand the level of exposure.

� The risk associated with pesticide use canbe predicted in a more realistic way by in-cluding parameters into risk assessmentstrategies that are environmentally rele-vant at the landscape level (i.e., recoverythrough recolonisation, life-history traits,etc.).

� Monitoring studies are recognised as avaluable retrospective tool to verify thefield relevance of risk assessmentschemes. They can also be used to demon-strate the effectiveness of risk mitigationmeasures.

It was generally accepted within the work-shop that it is challenging to determine pesti-cide exposure as the cause of effects in thefield. This difficulty is due to problems ofnatural variability, multiple substances andmultiple stressors, confounding factors, andinsufficient statistical power.

Concerning the current regulatory risk as-sessment procedure, many participants feltthat current approaches are reasonable toensure the protection of non-target organ-isms and, in some cases, may be consideredtoo conservative. However, effects of pesti-cides were identified in several of the fieldstudies presented. It could not be deter-mined whether good agricultural practise ormisuse was responsible for the observed ef-fects in many of these studies; therefore,some uncertainty remains about the actuallevel of protection. It was stated that furtherresearch is needed to evaluate accurately thedegree of protection, so that risk assessmentprocedures can be adjusted if necessary.

e x e c u t i v e s u m m a ry 8 · m a i n o u t c o m e s

27

CONTENTS

CONTENTS


CONTENTS

bWORKSHOPSYNTHESIS

CONTENTS

1 DEVELOPMENT OFWORKSHOP CONCLUSIONS

The workshop comprised 75 scientists repre-senting government, industry, and academiaand experienced in risk assessment and fieldeffects of pesticides. Participants were fromEU member states, the US, and Canada andthus a broad range of expertise, experience,and viewpoints were available.

The workshop analysed the state of theart of the current knowledge and informa-tion available about field effects of pesticidesassessed by performing field or monitoringstudies. This was reviewed via a series of ple-nary sessions, platform presentations (seeabstracts in Appendix A), and poster presen-tations (see titles in Appendix B). These wereseparated into 3 main topics: (1) observa-tions of effects in the field, (2) ability to pre-dict effects, and (3) implications for risk as-sessment.

Following the platform presentations, thetopics were discussed further in differentbreakout groups. The discussions were basedon data made available at the workshop.However, additional data were considered ifrelevant and scientifically sound. Partici-pants were divided according to their exper-tise into 4 breakout groups: (1) aquatic or-ganisms (2 groups), (2) terrestrial inverte-brates and plants, and (3) birds and mam-mals. Each breakout group produced aworking paper that reflected the content oftheir discussions. Rapporteurs presented theoutcomes from their breakout groups, andthese were further discussed among all par-ticipants in plenary sessions. The final ple-nary of the workshop discussed in detail thefuture role of field and monitoring studies ofpesticides in the EU and summarised the out-comes.

After the workshop, the steering commit-tee decided future actions and schedules andcommented on and revised the drafts of theworkshop report. The report coordinatormerged the working papers of the 2 aquaticbreakout groups into 1 report, which wasthen reviewed by the rapporteurs for the 2aquatic groups. The 3 breakout group reports(aquatic invertebrates, terrestrial inverte-brates and plants, and birds and mammals)

were then distributed to the participants,who were invited to comment in 2 rounds.The first and second sets of comments wereintegrated by the rapporteurs and the reportcoordinator, respectively. The workshopsteering committee decided that the 3 break-out group reports should be used as a basisto write the “workshop synthesis” and thatthe breakout group reports should be provid-ed as an appendix (Appendix D) to the finalreport because they were very different interms of substance, structure, and amountsof information provided. Once the workshopsteering committee agreed on the content ofthe workshop synthesis (Part B), this wascomplemented by an executive summary(Part A).

The draft report was then distributed tothe workshop participants, who were invitedto comment. Their comments were used bythe workshop report coordinator to make re-visions to the report, which was then subject-ed to final review by the workshop steeringcommittee.

The report (Parts A and B) was reviewedby 2 external reviewers. Appendixes Athrough D had to be kept as original materialand therefore were not externally reviewed.

2 FIELD STUDIES

2.1 DEFINITIONS

Investigations into the effects of pesticides inthe field generally fall into 2 main categoriesthat can be defined as follows:1) Monitoring study: an investigation into

the overall impact of pesticide use on aspecific ecosystem through surveying ormonitoring that consists of characterisa-tion of exposure (chemical monitoring,exposure modelling) and observations ofeffects (biological monitoring) occurringin the field or treated area as a conse-quence of use and/or misuse of pesticides.

2) Experimental study: an experimentanalysing the impact of a specific productor active substance applied under con-trolled conditions in the field. Such stud-ies are performed in the natural environ-ment within an agricultural context and

w o r ks h o p sy n t h e s i s 1 · d e ve l o p m e n t o f wo rk s h o p c o n c l u s i o n s

31

CONTENTS

thus contrast with mesocosm studies. Ex-perimental studies are usually conductedwith untreated controls and sometimeswith reference treatments in replicatedplots.

The main focus of the workshop was onmonitoring studies. However, some experi-mental studies were also reviewed becausethey can provide valuable information to aidinterpretation of monitoring data.

2.2 STUDIES ON EFFECTSOF PESTICIDES IN THE FIELD

A number of field studies were reviewed dur-ing the course of the workshop. For aquaticinvertebrates, sufficient monitoring studieswere available to discuss them in detail(Table 1). For terrestrial invertebrates andplants (Table 2), as well as for birds andmammals (Table 3), experimental studieswere also discussed due to the limited num-ber of monitoring studies available.

2.2.1 MONITORING STUDIES FORAQUATIC INVERTEBRATES

Biological effects of pesticides were identi-fied in several of the monitoring studies thatwere presented at the workshop. It should benoted that the studies reviewed focusedmainly on aquatic macroinvertebrates andinsecticides; thus, their broader applicabilityis limited. Furthermore, the studies weregenerally carried out in intensively cultivat-ed areas. Altered communities likely due topesticides were observed in the Altes Land(near Hamburg, Germany), a special case ofintensive agriculture for apple orchardswhere buffer zones are not appropriate(Schäfers, platform presentation). Effectswere also shown following runoff events inthe Liess studies (Liess, platform presenta-tion; (Liess 1994, 1998; Liess and Schulz1999; Liess et al. 1999; Schulz and Liess1999a, 1999b). These investigations tookplace at sites with a risk of surface runoff(Braunschweig area, Germany) ranging from


32

pa rt b

CONTENTS

Exposure Endpoint Species Reference

Substance Quantification Duration

Cypermethrin 2.25 mg/kg >150 days Abundance,emergence

Dipteran(Chironomidae)

Kedwards et al. 1999

Parathion-ethyl 6 µg/L 1 h Communitycomposition

11 invertebrate taxa Liess and Schulz 1999

Parathion-ethyl 6 µg/L 1 h Abundance, drift,mortality

Trichoptera, otherinvertebrates

Schulz and Liess 1999a

FenvalerateParathion-ethyl

0.85–6.2 µg/L 1 h Mortality abundance Amphipods Trichoptera Schulz and Liess 1999b

Endosulfan 1.3–10 µg/kgSPMDa

Unclear Abundance Ephemeroptera,Trichoptera

Leonard et al. 1999

Azinphos-methylChlorpyrifos Parathion-methyl

0.82 µg/L344 mg/kg1-550 µg/L

1–3 h Communitycomposition

Ephemeroptera,other insects

Schulz et al. 2002

Endosulfan 10–318 µg/kg Few h Abundance, drift Various species ofinvertebrates

Jergentz et al. 2004

Azinphos-methylChlorpyrifosEndosulfan

4±2 µg/kg29±19 µg/kg54±19 µg/kg

Severalweeks

Communitycomposition

Various invertebratetaxa

Thiere and Schulz 2004

5 fungicides 4 herbicides 1 insecticide

Various concentrations 1 h Communitycomposition


Berenzen et al. 2005

7 insecticides,6 fungicides,8 herbicides

Various concentrations 1 h Communitycomposition


Liess and Von der Ohe,2005

Table 1: Monitoring studies from the literature showing a clear relationship between exposure to pesticide (mostly due tosurface runoff) and biological effects on field aquatic organisms (modified after Schulz, 2004; see also review of Hom-men et al. 2004).

a semipermeable device

“very low” to “high” on a scale defined forthe German agricultural area (Bach et al.2000). The Schulz studies from South Africaalso took place in an intensively cultivatedarea exposed to pesticides via surface runoffand spray drifts (Schulz, platform presenta-tion; Schulz et al. 2001, 2002). Only a fewmonitoring studies exist that focus on biolog-ical effects and include appropriate exposuremonitoring to allow the establishment ofcause and effect.

Indeed, only 11 studies that included theinformation necessary to establish causalitybetween insecticide contamination of surfacewaters due to usual agricultural practisesand effects on aquatic fauna (i.e., exposurequantified, effects on field populations –thus excluding in situ bioassays – measured,control situations without effects included)were identified (Table 1); most of them arereported in a recent review (Schulz 2004). Itis important to note that for many of thesestudies, the pesticide concentrations mea-sured in the field were not large enough tosupport an explanation of the observed ef-fects based on acute toxicity data alone(Schulz 2004). Based on current knowledge,it is not known (1) whether measured con-centrations in the field regularly underesti-mate the real exposure, (2) whether relevantlong-term effects are considered in the fieldbut not under controlled conditions, and (3)whether there are differences in sensitivitybetween laboratory and field organisms.

2.2.2 MONITORING AND EXPERIMENTALSTUDIES FOR TERRESTRIALINVERTEBRATES AND PLANTSIN EUROPE

MONITORING STUDIESFOR TERRESTRIAL INVERTEBRATES

MONITORING SCHEMES

Honeybees Monitoring of honeybees is mostly per-formed for food-safety purposes and to pro-vide centralised data for incident reportingand for appraisal of wider-scale effects byregulatory bodies. The UK Wildlife Incident

Investigation Scheme (WIIS) (Barnett et al.2002) investigates suspected incidents ofhoneybee poisoning by pesticides, most ofwhich are reported by beekeepers. Previousanalyses of the results showed that no pesti-cides classified as having “low risk” were im-plicated in poisoning incidents (Aldridge andHart 1993), while, for example, pyrethroidsclassified as “high risk” are of low hazard inthe field (Inglesfield 1989). The current riskassessment scheme for honeybees thusseems to be successful, at least in that whilethere are false positives (e.g., withpyrethroids), there do not seem to be falsenegatives. However, potential side effects ofpesticides on honeybees are intensively dis-cussed in the public domain (e.g., in France;Römbke, platform presentation). Monitoringstudies have also been conducted in othercountries. For example, a honeybee monitor-ing system is operational in Germany, butthis system is more irregular than that in theUK and there are large regional differencesin efficiency. The foraging activity and be-haviour of the bees on the crop and at thehive entrance are recorded as part of fieldtest guidelines (EPPO 1992).

Non-target arthropodsand below-ground invertebrates

The number of monitoring studies on terres-trial invertebrates other than honeybees isvery limited. Indeed, there is no system likethe UK WIIS that is effective for non-targetarthropods (NTAs) or below-ground organ-isms. The possible application of the honey-bee scheme to other terrestrial invertebrateswas considered difficult because developedand accepted schemes are lacking. However,the monitoring of soil invertebrates is cur-rently facilitated by the standardisation ofsampling methods by the International Orga-nization for Standardization (e.g., tests withearthworms). When focusing on non-targetarthropods and below-ground invertebrates,species from in-crop areas and off-crop areas(which are habitats for many differentspecies) should be differentiated.

Ecological studies Ecological studies are not usually designedto examine pesticide effects but rather to

w o r ks h o p sy n t h e s i s 2 · f i e l d s t u d i e s

33

CONTENTS

study the ecology of individual species. How-ever, pesticide effects can be considered andtherefore detected in such investigations. Itwas noted that long-term data and land-scape-scale studies are lacking, even thoughpesticides can be applied over extensiveareas (Holland, platform presentation).However, several projects are in progressthat aim to identify the “normal” speciescomposition of soil invertebrates at referencesites (e.g., Ruf et al. 2003; Schouten et al.2004).

Examples:� The overall impact of pesticides was ex-

amined in an extensive, long-term studyconducted by The Game ConservancyTrust, referred to as the “Sussex study”(or the partridge survival project). Theextensive dataset includes information onthe abundance of many invertebrate fami-lies or species within each field and con-sequently allows their range across thestudy area to be investigated (Aebischer1991). In the late 1990s, the Sussex datawere reanalysed to look at the impact ofpesticides (Holland, platform presenta-tion; Ewald and Aebischer 1999, 2000).Although this has yet to be fully explored,it is evident that some of the invertebrateshave declined in abundance or sometimeshave disappeared as a result of pesticideuse (Holland 2002). Outcomes have led tomanagement measures, including recom-mendations with regard to adoption ofconservation headlands.

� Relatively few other studies have investi-gated long-term or large-scale effects.Holland (Holland, platform presentation)cited 2 examples of 2-year studies that in-vestigated effects of insecticides and con-sidered recolonisation (Wick and Freier2000; Kennedy et al. 2001). The insecti-cide reduced numbers of invertebrateswithin a season, but there were no effects1 year afterwards on open plots with nobarrier to recolonisation.

Another type of ecological study covers localor regional projects that are usually support-ed by state or federal agencies that do not ex-plicitly aim at monitoring effects of pesti-

cides. They try to (1) determine the biologi-cal soil quality (Germany: Ruf et al. 2003),(2) assess the habitat function of the soil (in-dependently, whether this function is affect-ed by pesticides, soil compaction, other fac-tors, or a combination of all), (3) identify thereference sites, or (4) assess effects of soilpollution (e.g., heavy metals in the UK: Spur-geon et al. 1996), including an increasing in-tensity of (agricultural) land use (France:Ponge et al. 2003).

EXPERIMENTAL STUDIESFOR TERRESTRIAL INVERTEBRATESAs described earlier, experimental studies in-vestigate the impacts of a specific substanceintroduced into the field under controlledconditions for the specific purpose of the ex-periment.

SMALL-SCALE FIELD STUDIESVan Straalen and van Rijn (1998) reviewed aseries of studies performed since 1964 thatinvestigated the effects of different pesti-cides (lindane, dimethoate, parathion, chlor-pyrifos, carbofuran, carbaryl, benomyl, andatrazine) on various species of terrestrial in-vertebrates. Thompson (2003) gives anoverview of a wide variety of semi-field andfield studies assessing bee behaviour follow-ing field applications, indicating what typeof field observation is needed for predictinglong-term consequences of pesticides. Guide-lines are available on test methods for evalu-ating the side effects of pesticides on the for-aging activity of honeybees and their behav-iour on the crop and at the hive entrance(EPPO 1992). Some studies have also beenperformed by companies at agricultural sitessprayed with specific pesticides, but these re-sults are not usually published (e.g., studieson earthworms are based on methods de-scribed in the Earthworm Field Test, ISO1999). Vogt (1994, 2000) also compiled re-sults from semi-field and field studies withterrestrial arthropods and compared themwith results from laboratory tests based onthe “worst-case assumption”. Results showthat (1) using laboratory tests to screenharmless products is reliable and (2) predict-ing the magnitude of field effects for prod-


34

pa rt b

CONTENTS

ucts found to be harmful in the laboratory isnot possible.

FARM-SCALE STUDIES Farm-scale studies are large-scale experi-ments for which large blocks of farmlandcomprising multiple fields are subjected tocontrasting treatments, and populations orcommunities of terrestrial invertebrates arethen monitored. Farm-scale studies are notroutinely performed because of their cost,and only a few examples exist.

Examples:� A well-known example is the Boxworth

Project from the UK, which broke newground during the 1980s by being the firstlong-term, farm-scale project to investi-gate the ecological impacts of intensivearable farming practises (Greig-Smith etal. 1992). Three treatment regimes wereapplied to different areas of the samefarm over 5 years. Overall, 1 regime (“fullinsurance”) involved more than twice thenumber of pesticide applications andabout 6 × the number of insecticide appli-cations used in the other 2 regimes (so-called “supervised” and “integrated”treatments, similar to each other and totypical UK farm practises at the time).Over the 5 years of application, the resultsof full insurance treatment compared tothe other 2 treatments showed that (1) ef-

fects on soil invertebrates such as spring-tails were not consistent, with somespecies showing decreases and others in-creases; and (2) herbivorous, predatory,and parasitic invertebrates in the crop haddensities that declined by about 50%,while detritivores were little affected. Theresponses of individual species variedwithin each group, and this was attrib-uted to complex differences in exposure,capacity for recolonisation, and changesin prey populations. Changes in somespecies were clear and consistent, butmost showed considerable variation fromyear to year.

� The TALISMAN (Towards a Lower InputSystem Minimising Agrochemicals and Ni-trogen) and SCARAB projects evolvedsubsequently as follow-on studies to theBoxworth Project and were designed toexamine in greater detail many of the is-sues raised by the Boxworth Project.These 2 projects complemented eachother in their aims and objectives. TALIS-MAN focused primarily on the economicand agronomic issues of reducing pesti-cide and nitrogen fertilizer use. TheSCARAB project examined ecological sideeffects of pesticides on non-target organ-isms including insects, spiders, earth-worms, and soil microbes and is thus ofmore interest in the context of this work-shop. The numbers of certain species of


35

Study Type Endpoint Species Location Reference

WIIS Monitoring(monitoring scheme)

Lethality Beneficial insects (bees) UK Barnett et al. 2002

The GameConservancy Trust

Monitoring(ecological studies)

Abundance Various speciesof invertebrates

Sussex study,UK

Holland et al. 2002

Long-term study Long-term effects(over 2 seasons)recolonization

Nontarget arthropods(Carabidae,Staphylinidae,Linyphiidae)

Germany,UK

Wick and Freier 2000Kennedy et al. 2001

Benomyl Experimental(small scale study)

Long-term effects Various species ofgrassland species

UK Bembridge et al. 1998

Boxworth Project Experimental(farm-scale studies)

Long-term effects,abundance andrecolonization

Crop invertebrates,predators of pests, soilinvertebrates, plants infield margin

UK Greig-Smith et al.1992

SCARAB Long-term effects,abundance

Invertebrates (Collem-bola, earthworms, etc),soil microflora

UK Frampton 1997aTarrant et al. 1997

Table 2: Examples of monitoring and experimental studies from the literature showing effects of pesticides on non-tar-get terrestrial invertebrates

CONTENTS

non-target insects and spiders declinedafter the annual application of somebroad-spectrum insecticides, but recoveryusually occurred within the same season.The results showed considerable inter-species variability of both initial Collem-bola vulnerability to the pesticide regimes(Frampton 1997a; Frampton 2000) andsubsequent recovery rates among Collem-bola (Frampton 1997b, 1997c, 2000).Some species of Entomobryidae wereeliminated from the field (but not the sur-rounding margins) by repeated applica-tions of organophosphorus (OP) insecti-cides (Frampton 2002). Recovery of 1species took more than 4 years after theinputs of OP insecticides had stopped, de-spite the presence of potential recolonistsat the field margins throughout the period(Department for Environment, Food, andRural Affairs [DEFRA] Project PN0934). Alimitation of the SCARAB and Boxworthprojects was the lack of spatial replica-tion. However, this drawback was over-come in the SCARAB project by temporalmanipulation of the pesticide regimes(Frampton 2000, 2001, 2002).

FIELD STUDIES ON PLANTSThe use of plants in monitoring studies canbe more straightforward than for othergroups of organisms because general infor-mation on life-history strategies, phenology,and habitat requirements of wild plants arereadily available. Furthermore, terrestrialplants are sessile organisms, and thus theirpresence in samples may be clearly related tolocal environmental conditions. However,this area has not been considered very thor-oughly so far with respect to regulation inEurope.

One study examined the effects of herbi-cide use in arable fields on ditchbank vegeta-tion adjacent to sprayed fields (De Snoo andvan der Poll 1999). The number of dicotyle-donous species adjacent to an unsprayedbuffer zone was about 50% higher comparedto the number of species directly adjacent tothe parcel treated with a herbicide againstdicotyledonous weeds. The buffer zone re-duced drift deposition from about 5% to less

than 0.1% of the field dose (De Jong, plat-form presentation).

In a large-scale field experiment, clearshort-term effects on plant growth and phy-totoxic symptoms on non-target vegetation ofditchbanks and road verges occurred at lowdosages (<5% of the field rate) of a broadspectrum of herbicides at several metresfrom a treated plot. Effects within the sameseason were generally observed at higherdosages. The inclusion of unsprayed bufferzones and other drift-reducing measurescould prevent these effects. Full recoverywas observed in the next year for all dosages(up to 64% of the maximum field rate) andmeasures of effect (De Jong, platform pre-sentation).

Studies on non-target plants in field mar-gins during the Boxworth project suggestedthat they were more influenced by habitatstructure than by pesticide drift or treat-ments with pesticides.

2.2.3 MONITORING AND EXPERIMENTALSTUDIES FOR BIRDS AND MAMMALSIN EUROPE

MONITORING STUDIES

MONITORING SCHEMESA reactive monitoring scheme (or incidentinvestigation scheme) is defined as one thatconsiders whether a deleterious event (e.g.,death of a bird or mammal) results from theuse of a pesticide and, when evidence isgathered, whether it is attributable to thecorrect use, misuse, or abuse of a compound.The organisation of reactive monitoringschemes in Europe seems to be underdevel-oped, as it exists in only 7 out of 18 countries(De Snoo et al. 1999). Such schemes havesome limitations, including � low probability of an incident being

reported, � use of retrospective data (measurement

of lethality), � bias towards large species or species of

conservation interest, and � no detection of long-term changes due to

chronic and indirect effects.


36

pa rt b

CONTENTS

However, they also have some advantages:they are a safety net for the regulatory sys-tems in terms of acute direct effects, they arecountrywide, and they highlight the need forappropriate risk management and issues re-garding misuse and abuse.

Examples:� The UK WIIS investigates the deaths of

wildlife (i.e., birds, mammals, and benefi-cial insects), pets, and some livestockthroughout the UK where there is evi-dence that pesticide poisoning may be in-volved. Annual reports of findings arepublished (e.g., Barnett et al. 2002).

� Another successful scheme, established inFrance, is operated by a national network(SAGIR, a sanitary surveillance networkfor wildlife in France) and aims at assess-ing wildlife health rather than just pesti-cide poisonings. Initially, it focused main-ly on game but is now extended to otherspecies.

ECOLOGICAL STUDIES These long-term, species-specific monitoringstudies are not usually designed to examinepesticide effects but rather to study the ecol-ogy of individual species. However, pesticideeffects may be detected when there are, forexample, variations with regard to pesticideuse or any obvious hints of causal effects.

Examples:� Partridge survival project: This project

(simply known as the “Sussex study” andconducted by The Game ConservancyTrust) started in 1968 and was triggeredby the decline of the grey partridge popu-lation. It studied general effects of farm-ing practise and indirect effects of pesti-cides but excluded direct toxic effects ofpesticides. The cause identified as respon-sible for the decline was a reduction in theavailability of chick food. A chain ofcausal links from pesticide use to insectabundance to chick survival to populationsize was demonstrated (Potts 1986). Re-sults have not been used for the assess-ment of individual pesticides but led tomanagement measures (i.e., recommen-

dations with regard to the development ofconservation headlands).

� Corn bunting study: This study had thesame hypothesis of cause (decrease inavailability of chick-food invertebrates)and chronic effects (chick mortality) as inthe partridge project. This hypothesis wasalso demonstrated, and it was concludedthat these effects may reduce breedingsuccess (Brickle et al. 2000).

� Turtle dove study: This study comparedecological parameters in the 1990s withthose in the 1960s. A change in diet, an in-crease in foraging distance, and a de-crease in the number of nesting attemptswere demonstrated, but a link to the useof pesticides was not shown (Browne andAebischer 2004).

� Sparrowhawk study: This study demon-strated direct impacts of cyclodiene insec-ticides (such as dieldrin) on populationsof sparrowhawk (Newton 1986; Newtonand Wyllie 1992; Sibly et al. 2000).

� Yellowhammer study: This study showedevidence of some indirect effects of pesti-cides on breeding of yellowhammer(DEFRA Project PN0925: (Anonymous2004).

� Studies commissioned by the Danish Envi-ronmental Protection Agency: these stud-ies were undertaken to analyse and modelthe distribution of birds in Danish farm-land in relation to various parameters in-cluding pesticide use (Petersen 1996;Petersen and Jacobsen 1997).

EXPERIMENTAL STUDIESExamples of farm-scale studies are includedin the following:� The effects of pesticides on avian end-

points were studied in the Boxworth Pro-ject (Greig-Smith et al. 1992), a 5-year,government-funded investigation intosustainable methods of farming. Avianendpoints included abundance, reproduc-tive performance, frequency of nest visitsby birds, feeding of young, diet composi-tion (from faecal samples), andcholinesterase inhibition in nestlings. Theproject also measured the abundance ofwood mice by using capture-marking-re-capture methods. The project had a num-


37

CONTENTS

ber of limitations, the most important ofwhich was the lack of replication becauseof cost limitations. This weakness made itdifficult to assess whether differences be-tween the blocks were due to the treat-ments or to confounding factors. In addi-tion, despite the relatively large size of theblocks, the number of active bird nestsavailable for study at the time of any par-ticular pesticide application was small.The only vertebrate endpoint that showeda clear pesticide effect was the abundanceof wood mice. Applications of methiocarbpellets in the autumn had a dramaticshort-term effect on their populations.However, immigration of juvenile micefrom untreated areas led to a rapid recov-ery in numbers. It was suggested that ap-plications at other times of the year, or inareas without adjacent woodland to allowrecolonisation, might produce longer-last-ing effects.

� A large-scale field experimental study hasrecently been completed in the UK to as-sess the relative importance of food avail-

ability on the demography of farmlandbird populations as part of the “IndirectEffects of Pesticides Project” (DEFRA Pro-ject PN0925: Anonymous 2004). Threetreatments and a control were applied in afactorial design to 1-km2 blocks of farm-land at 3 sites. One treatment involved in-creased insecticide application in summerto decrease invertebrate abundance; an-other also had increased insecticide inputsin summer along with provision of extraseed in winter. Where the extra insecti-cides were applied, there was a reductionin the reproductive output of yellowham-mers at all 3 sites. The third treatment in-volved providing extra seed in winter toincrease the food supply: there was someevidence that this increased overwintersurvival of yellowhammers, but the effectwas not consistent across all sites. Resultsfor other bird species were inconclusive.

3 · e s s e n t i a l s o f f i e l d s t u d i e s

38

pa rt b

CONTENTS

Name Study type Endpoint Species Location Reference

WIIS Monitoring(monitoring schemes)

Lethality Wildlife (birds mam-mals), pets, livestock

UK Barnett et al. 2002

SAGIR LethalityHealth

Game birds, otherwildlife

France Berny et al. 1997

Partridge survival pro-ject (Sussex study)

Monitoring(ecological studies)

Lethality (throughindirect effects)

Grey partridge Sussex, UK Potts 1986Aebischer 1991Ewald and Aebischer1999; Ewald andAebischer 2000

Corn bunting project Lethality (throughindirect effects)

Corn bunting UK Brickle et al. 2000

Turtle dove project Indirect effects Turtle dove UK Browne and Aebischer2004

Sparrowhawk study Abundance Sparrowhawk UK Newton 1986;Newton and Wyllie1992Sibly et al. 2000.

Boxworth project Experimental(farm-scale studies)

Abundance, breedingperformance,behaviour, nestlinggrowth, ChE inhibi-tion in nestlings

Birds UK Greig-Smith et al.1992

Abundance, Recovery Small mammals

Indirect effectsproject

Abundance,Reproductive perfor-mances Overwintersurvival.

Birds UK Boatman et al. 2004Anonymous 2004

Table 3: Examples of monitoring and experimental studies from the literature showing effects of pesticide on birds andmammals

3 ESSENTIALS OF FIELD STUDIES

A review of monitoring and experimentalstudies during the workshop clearly showedthat these studies should include both as-sessment of exposure and monitoring of bio-logical effects in order to attempt to establishcausality. Criteria or concepts that are impor-tant for these studies (especially for monitor-ing studies) were further discussed in break-out groups, and the main outcomes are re-ported in this chapter.

3.1 EXPOSURE ASSESSMENT

One issue in interpreting field data is that ex-posure may be poorly characterised and thevalidity of field studies without exposure as-sessment is questionable. This is particularlyrelevant for monitoring studies. Further-more, studies can be used to confirm mitiga-tion measures or to validate exposure mod-els. Therefore, quantification of exposureconcentrations should be as complete as pos-sible.

For aquatic organisms, monitoring expo-sure focuses mainly on water concentrations.For terrestrial organisms, exposure datacomprising residues in food items are ex-tremely important (Anonymous 2002). Forbirds or mammals, monitoring exposure (ex-pressed as dose or body burden) is extremelydifficult because it not only is a function ofthe environmental concentration but also de-pends on the animal’s behaviour.

The following issues are relevant for anadequate assessment of exposure of organ-isms to toxicants in the field, although theyare mainly illustrated by examples fromaquatic ecosystems.

3.1.1 CHEMICAL USE PATTERNS

Chemical use patterns include qualitative,quantitative, spatial, and temporal informa-tion on the use of pesticides within a moni-tored area. Such knowledge is needed be-cause it provides valuable information for (1)evaluating potential effects of pesticides onnon-target organisms and (2) designing, for

example, appropriate exposure monitoringfor mixtures of chemicals.

An adequate evaluation of pesticide expo-sure, for example, in catchments, should in-clude information associated with landscapecharacteristics (e.g., river basin structure,catchment size, soil type, rainfall, waterflow) and agricultural practises (e.g., fertil-izer application, land use practises, amountand type of pesticides used, as well as spray-ing season). However, there are often prob-lems in obtaining such data, especially thoserelated to pesticide use (Kreuger, platformpresentation).

Studies involving field measurementsshould use an event-triggered sampling de-sign. For example, it is necessary to sampleat the time of application for drift inputs inaquatic systems. In contrast, it is importantto sample at the start of the hydrologicalevent for rainfall-driven inputs (runoff anddrain flow) and later to capture both thepeak pesticide concentration and the patternover time. The problem is enhanced by thefact that time of pesticide application is moreor less unpredictable. Problems linked to res-olution in space and time could be partly re-duced by using GIS. Indeed, GIS may haveapplications in analysing factors influencingexposure, such as landscape characteristics,proximity of habitats to treated fields, andtemporal use patterns.

3.1.2 GOOD AGRICULTURAL PRACTISES

Exposure to pesticides results from field ap-plications but may additionally result frommisuse and/or point sources (e.g., failure toobserve appropriate buffer strips, waste-water treatment works, or improper cleaningof equipment in the farmyard). Regulatoryrisk assessment of pesticides does not usuallyconsider misuse or point sources because it isassumed that pesticides are used in accor-dance with the principles of Good Agricul-tural Practise (GAP). Predicted environmen-tal concentrations do not currently considerexposure from poor agricultural practise. Along-term programme in Sweden that aimedat informing farmers about GAP led to a 90%reduction in pesticide contamination in

w o r ks h o p sy n t h e s i s 3 · e s s e n t i a l s o f f i e l d s t u d i e s

39

CONTENTS

streams, whereas the amount of pesticide ap-plied did not decrease (Kreuger, platformpresentation).

The problem of point source and misuseneeds to be kept in perspective because, forexample, chemical monitoring data from theNetherlands indicate that chemicals that ex-ceed maximum permissible concentrationsderived from lower- and/or higher-tier tests(see www.bestrijdingsmiddelenatlas.nl) maybe associated with unknown sources. In ad-dition, experience from Germany with moni-toring in groundwater indicates that severalcases of exceeding maximum permissibleconcentrations (drinking water standards)originated from point-source contaminationand/or misuse. However, these cases mainlyconcerned water-soluble herbicides becausethey are easily detectable with multiresiduemethods (Bach et al. 2000, 2001; Muller etal. 2002). In the case of insecticides, only afew studies used event-controlled samplingstrategies in order to detect maximum short-term concentrations (Kreuger 1998). Hence,it should be pointed out that especially in thecase of short-term pulses of insecticides thatresult from diffuse sources, measured con-centrations are generally underestimatedcompared to the real concentrations becauseof the sampling and/or analytical protocols.

3.1.3 MODELLING EXPOSURE

Models are used extensively to predict pesti-cide exposure for regulatory risk assessmentand are a possible option to reduce the prob-lems related to field measurements. As de-scribed by Brown (platform presentation),the models are used to estimate field behav-iour, mainly on the basis of laboratory mea-surements, and offer several advantages overdirect determination of exposure in the field:modelling is faster and cheaper, it is possibleto assess fate and exposure under the fullrange of possible use conditions, and theanalysis can be repeated to account forchanges in use or impact of mitigation mea-sures. Nevertheless, it is critical to establishthe extent to which model predictions for en-vironmental concentrations of pesticidesmatch real concentrations in the field. How-

ever, the predictive power of exposure mod-els is questionable because extensive valida-tion studies are missing. Problems with re-liance on modelling alone include, in addi-tion, the following: (1) pesticides and unde-tected, yet toxicologically important,substances may be missed either in chemicalanalyses or in models, (2) exposure process-es are not always well understood, and (3)environmental data available to perform thecalculation are not always sufficient (e.g.,presence of buffer strips, pesticide usage).Nevertheless, it is possible to combine mod-elling with monitoring data to provide fur-ther insights into quantifying exposure. Forexample, catchment studies can be used tocalibrate models that include product usedata and landscape information, which canthen be used for generic exposure assess-ment (e.g., the PESTSURF development inDenmark), or modelling and monitoringdata can be coupled within GIS (e.g.,Padovani et al., poster presentation). Cur-rently, the validation of exposure models bycomparing measured and modelled loadscan be problematic because of the lack ofmonitoring data. For example, for avian riskassessment, the estimates of exposure havehad almost no verification or validationagainst actual field data (Mineau, platformpresentation).

According to the issues mentioned above,the pesticide concentrations measured inchemical monitoring programmes can differfrom those predicted by regulatory models(e.g., Crane et al. 2003). This may be due tovarious reasons and may lead to the follow-ing� Differences between predicted and mea-

sured concentrations because of� sampling from different types of water

bodies (e.g., flowing rather than staticwater bodies),

� inappropriate model or missing or in-accurate input data, or

� differences in the dissipation of thepesticide in water, between observa-tions in the field, and in predictions onthe basis of laboratory water–sedimentstudies.

� Lower concentrations measured than pre-dicted because of


40

pa rt b

CONTENTS

www.bestrijdingsmiddelenatlas.nl

� sampling during a low vulnerability sit-uation (e.g., in relation to timing ofrainfall),

� missing of peak concentrations, or� conservatism in the initial tiers of the

risk assessment process (e.g., worst-case 90th percentile modelled environ-mental exposure).

� Higher concentrations measured thanpredicted because of� point-source contamination (which is

not included in models),� poor agricultural practise (e.g., buffer

zone violation), or� addition of concentrations as a result

of simultaneous uses in the area, whilepredicted concentrations focus on onesingle-use situation.

3.1.4 ROUTES OF EXPOSURE

Appropriate monitoring requires a good un-derstanding of the routes of exposure. Rou-tine monitoring for pesticides and other sub-stances in surface waters is usually per-formed by water analyses. Indeed for aquaticorganisms such as macroinvertebrates, themain route of exposure is usually via thewater and only to a lesser extent throughsediment or food. However, for organismssuch as sediment dwellers, exposure via in-terstitial water associated with sediment maybe of greater relevance. Routine monitoringis mostly carried out by authorities and regu-latory bodies for legislative reasons that areusually unconnected with pesticide authori-sation and control. Many of these pro-grammes have poor temporal resolution(e.g., monthly or quarterly) in relation topeak pesticide concentrations. In addition,sampling locations often concentrate on larg-er water bodies (e.g., at the inlet of drinkingwater plants), and analytes are frequently bi-ased towards water-soluble herbicides andare not related to use patterns. Routinechemical monitoring is therefore often oflimited value for the prediction of ecologicaleffects, although it is acknowledged thatunder the Water Framework Directive thisshould be augmented by monitoring for bio-logical impacts.

The published data from case studies onexposure to agricultural insecticides in sur-face waters have recently been reviewed(Schulz platform presentation; Hommen etal. 2004; Schulz 2004).

For terrestrial organisms, the preciseroute of exposure is also a critical factor de-termining the risk of insecticides (Akkerhuis1993; Wiles and Jepson 1994). Emphasis isclearly on oral exposure via food for verte-brates, for example. Therefore, exposuredata comprising residues in food items areextremely important to refine the risk assess-ment (Anonymous 2002). There is evidencethat the dermal route may also be importantfor certain compounds (Mineau, platformpresentation; Mineau 2002), and exposurevia this route is likely to be underestimatedwithin current risk assessment schemes. Oraluptake via preening and inhalation may alsoneed to be considered.

For birds or mammals, exposure (ex-pressed as dose or body burden) not only is afunction of the environmental concentrationbut also depends on the animal’s behaviour –such as time spent in treated areas, feedingpreferences, or feeding rates – which is high-ly variable and quite often unknown.

3.2 BIOLOGICAL EFFECTS

3.2.1 DEFINING REFERENCEAND CONTROL SITES

A key issue for performing field studies thatare suitably representative and that can beextrapolated across different regions is to de-fine control or reference sites. Such sitesneed to be selected with a view to protectiongoals.

Often there is no “control” site because se-lecting a truly uncontaminated site is un-achievable in regions with historically modi-fied ecosystems (i.e., most developed coun-tries). Nevertheless, ecosystems modified byanthropogenic activities may have many ofthe properties of truly natural systems (e.g.,certain types of ditches may simulate theproperties of natural water bodies). The in-tensity of such activities (i.e., the degree of


41

CONTENTS

disturbance) could be used as an indicatorfor selection of appropriate reference sites.

Reference sites should fulfil the followingcharacteristics:� ensuring that the pesticide exposure

regime is such that no effect is induced;� incorporating the natural variability char-

acteristics of the community under con-sideration;

� minimising the importance of environ-mental parameters and confounding fac-tors by avoiding differences other thanpesticides between reference and testedsites (e.g., for aquatic systems: minimaldifference in neighbouring cultures,stream or pond morphology, nutrient situ-ation, hydrodynamics, soil and sedimenttype).

� excluding any major anthropogenic dis-turbance but including whatever distur-bance induced by natural stressors (e.g.,drying) is typical for the area (apart fromexposure to pesticides).

To fulfil these characteristics, reference sitesare often selected in smaller upstream sec-tions, while impacted sites are often in largerwater bodies located downstream. Effects indownstream sections can be detected, al-though not always properly quantified, when“internal” reference sites (comparisonsthrough time at the same site) or “relative”reference sites (sites along a gradient of con-tamination) are used. However, in contami-nated and reference sites, background conta-mination (current or past) should be avoidedas far as possible because of the issue ofadaptive processes that may have occurred inresident populations and that may lead to re-placement of sensitive species or individualsby more tolerant ones and consequently toan increase in population and communitytolerance (e.g., pollution-induced communi-ty tolerance [PICT]) (Blanck 2002).

Reference sites may be more useful whenselected at a more integrated level (e.g.,community structure according to ecologicaltraits as opposed to species abundance).

Definition of target images would supportthe development of such reference condi-tions. There are some parallels with the defi-nition of target images for the EU WaterFramework Directive (developed for pesti-

cides and other stressors) that could proveuseful in implementing the concept into reg-ulatory risk assessment (EC 2000 cited inRömbke and Breure 2005).

In conclusion, the definition of the refer-ence condition may be simply a pragmaticdecision based upon a combination of scien-tific judgment, “public acceptability”, andevidence-based determination of “good qual-ity” in agricultural landscapes. Examples ofsuch definitions for aquatic communities areavailable (Nijboer 2000; Nijboer et al. 2004).

3.2.2 DETECTING EFFECTS

CONFOUNDING PARAMETERSAND NATURAL VARIABILITYConfounding parameters and natural vari-ability can make it difficult to identifyand/or discern effects of pesticides evenwhen a proper reference site has been select-ed. It is important to characterise and under-stand them properly in order to (1) accountfor them in the study design and interpreta-tion of results and (2) reduce the uncertain-ties and increase the power of detection ofpesticide effects in monitoring and fieldstudies.

The importance of confounding parame-ters can be reduced by careful selection, butthey cannot be completely excluded becausethey are part of the natural system. They canbe site-specific, context-dependent, and eco-logically relevant parameters with an impor-tant influence on the habitat. Factors can bephysical (e.g., habitat structure and flowregime), chemical (e.g., water chemistry, nu-trients, pesticide metabolites and degrada-tion products with known toxicological sig-nificance, other pollutants), or biological(e.g., community composition). As an illus-tration, if data on long-term changes of com-munity composition at a regional scale (e.g.,species checklists) were available, they couldallow us to detect changes in regional biodi-versity in a particular area. However, with-out additional regional information on con-founding factors (e.g., changes in land useand agricultural practises, water manage-ment), elucidation of causality and robust in-terpretation may be compromised.


42

pa rt b

CONTENTS

The natural variability of a community isdue to variations in different environmentalparameters combined with the normal oper-ating range of the biological characteristics,as determined by phenotypic plasticity andgenetic variability. Some of the natural vari-ability may be accounted for by categorisingspecies according to ecological traits (e.g.,sensitivity to pesticides, life-history charac-teristics, dispersal ability) and then compar-ing effects in these groups across differentexposure concentrations, as in the concept ofspecies at risk (SPEAR) (Liess, platform pre-sentation; Liess and Von der Ohe 2005). Sim-ilarly, categorising habitats with reference tothe organisms within them (e.g., organismsassociated with oxygen-rich microhabitats)may help to elucidate the effects of otherstressors or focus on specific risks and thusisolate possible effects of pesticides (e.g., theRiver Invertebrate Prediction and Classifica-tion System [RIVPACS] for the UK; theAQEM consortium for the EU, which has de-veloped an assessment system for streams in8 European countries based on benthicmacroinvertebrates [www.aqem.de]; and astudy on the composition of plant and inver-tebrate communities in field margins [Ratte,platform presentation]).

ENDPOINTS Measuring the correct endpoints (e.g., acute,chronic, structure, function) is essential todetect effects. Endpoints commonly used anddiscussed in monitoring and field studies in-clude the following:� Individual level: survival, growth, repro-

ductive performance, histological parame-ters, physiological parameters, biochemi-cal parameters (e.g., biomarkers, enzymeassays, fat content, energy metabolism,respiratory rates, subcellular endpointsfor all species, eggshell thickness forbirds), behavioural responses (e.g., avoid-ance, feeding rates, repellency, foragingplot choice, and courtship behaviour forterrestrial vertebrates and invertebrates),and pesticide biotransformation pathwaysand tissue residues.

� Population level: abundance (includingspatial distribution), population dynam-ics, population growth and extinction

rates, recolonisation and recovery, popu-lation stage and genetic structures, in-breeding coefficient, and biomass.

� Community level: species diversity and/orrichness, shift in species dominance, bio-mass (plants, animals, potentially mi-crobes), and functionality (communitymetabolism, nutrient cycles, coarse organ-ic matter breakdown).

The most ecologically relevant and thor-oughly studied effects are usually investigat-ed at population and community levels be-cause they are the ultimate changes of con-cern in the field. However, there are excep-tions. For example, effects are usuallystudied on the individual level for endan-gered species and species of special public in-terest such as birds.

The power of monitoring schemes can begreatly increased when they take whole com-munities into account, because whole-com-munity schemes maximise the likelihood ofdetecting effects on sensitive species andalso make it possible to detect overall com-munity changes. On the other hand, theseschemes can obscure effects on some keyspecies if these effects are compensated orhidden by other changes such as the pres-ence of confounding parameters. In addition,the level of taxonomic identification influ-ences the power of monitoring schemes. Theability to detect effects will be decreasedwhen only a higher level of phylogenetic taxais concentrated on. In such cases, the impactmay not be detected because of replacementof affected species within the tested taxa.

Population- and community-level end-points (e.g., abundance) are used to charac-terise populations in field studies for aquaticand terrestrial invertebrates. Those end-points are challenging to investigate for birdsand mammals because of the large homeranges of the animals and their long genera-tion times (Kendall and Lacher 1994). Never-theless, the best evidence for effects of pesti-cides on birds is at the ecosystem level, as isshown in the studies of the grey partridgesurvival project conducted by The GameConservancy Trust (Potts 1986).

In addition to structural endpoints, func-tional endpoints (e.g., O2 consumption, O2

production, leaf litter decomposition, soil pa-


43

CONTENTS

www.aqem.de

rameters influenced by earthworms) shouldalso be considered for detecting effects be-cause they are also ecologically important(e.g., herbicides often inhibit oxygen produc-tion). However, functional endpoints areusually less sensitive than structural end-points due to possible replacement of sensi-tive species (i.e., functional redundancy),and they are rarely considered in field trials,which are usually performed at insufficientspatial and temporal scale to detect effects(Holland, platform presentation). Neverthe-less, there are a few examples of the success-ful investigation of functional endpoints suchas coarse organic matter breakdown (for lit-ter-dwelling and below-ground inverte-brates). Recently, the effects of pesticides onorganic matter breakdown have been as-sessed as part of the EU pesticide registrationprocess using the litter-bag method (EPFES2003).

DIRECT AND INDIRECT EFFECTSPesticides can be deleterious to individuals,populations, and communities by exertingdirect and indirect effects. Monitoring stud-ies have the potential to detect both directand indirect effects, such as the significantincrease of Collembola abundance and thedecrease of Linyphiidae abundance followinguse of synthetic pyrethroid insecticides(Frampton 1999). However, it can be chal-lenging to hypothesise the causal influencesbehind an observed effect and to identifywhether an effect is direct or indirect. Theuse of additional experimentation (e.g.,bioassays, micro- or mesocosms) could pro-vide useful complementary information tobetter distinguish between direct and indi-rect effects.

Direct effects of pesticides are related tothe sensitivity of the species towards pesti-cides and are illustrated by concepts such asthe species sensitivity distribution (SSD;Brock, platform presentation) and the rela-tive sensitivity distribution (RSD; Wogramand Liess 2001; Von Der Ohe and Liess 2004).Indirect effects result mainly from changes inbiological interactions (e.g., altered preda-tor–prey relationship, competition alter-ation, disappearance of food or habitat re-sources).

Grouping species according to ecologicaltraits (e.g., sensitivity towards toxicants,long generation time, lower ability for re-colonisation) could be a promising approachto assess direct and indirect effects. For ex-ample, in the SPEAR concept (Liess, platformpresentation; Liess and Von der Ohe 2005),aquatic species are grouped according totheir (1) life-cycle traits to account for differ-ences in recovery or recolonisation potentialand (2) sensitivity to particular pesticidemodes of action. Using this concept, it wasshown that the invertebrate communities ex-posed to toxic pressure are characterised by areduced number and abundance of species atrisk. This leads to an increase in abundanceof species not at risk due to reduced competi-tion with or predation by species at risk,which is a typical example of an indirect ef-fect.

Recent findings for plants suggest thatcrop species sensitivity to herbicides is ade-quate to represent the response of non-cropspecies regardless of chemical class or expo-sure (McKelvey et al. 2002). However, thereare some non-crop species that have noclosely related crop species. Thus, these non-crop species might not be represented suffi-ciently by tests with common crop species.Depending on the question underlying fieldtests, it might be necessary to consider non-crop species as well.

Although the respective importance of di-rect and indirect effects of pesticides remainsunclear for birds and mammals, indirect ef-fects (through disturbance of the food chain)can be more important than direct effects forspecies such as insectivorous birds (Sother-ton and Holland 2002).

Further evidence concerning indirect ef-fects of pesticides will be gained from the In-direct Effects of Pesticides Project funded bythe Pesticides Safety Directorate of DEFRA inthe UK (Holland et al. 2002).

TOLERANCEIt was discussed that the absence of observedacute effects is not necessarily an indicationthat pesticides do not exert any effects. Rea-sons can include monitoring with too littlepower to detect effects. Acute effects couldalso be absent due to the development of tol-


44

pa rt b

CONTENTS

erance or resistance at different levels of bio-logical organisation (individuals, population[common problem in pests], or community[e.g., PICT]). Thus, chronic effects need tobe taken into account within the context ofthe development of tolerance on various lev-els of biological organisation.

3.2.3. RECOVERY

CHARACTERISTICS OF RECOVERYRecovery is an integral attribute of how bio-logical systems (individuals, populations,communities) respond to a stressor over thelong term. Therefore, it should be consideredin the interpretation of effects.

The term “recovery” must be properlycharacterised. In the context of the work-shop and at the population and communitylevels, it includes both internal recovery (re-covery from within the affected system) andexternal recovery (recolonisation by disper-sal from external sources). Field studiesshould attempt to distinguish and quantifythese 2 aspects. This endeavour may be verydifficult because it requires incorporation ofsite-specific population dynamics, life-histo-ry information for affected populations, andinformation on metapopulation dynamics inthe landscape. Evaluation of the degree ofconnectivity of ecosystems at local or land-scape levels may also be very useful. Assess-ing recovery requires considerable knowl-edge about different landscape factors andhow they interact with organism life history.Thus, time and space are important parame-ters to consider in the recovery process.Many questions therefore remain in this areaof ecology:� How does recovery potential differ across

systems and landscapes?� Is the possibility or probability of recovery

reduced if a large part of the landscape isadversely affected by pesticide exposure(Liess, platform presentation)?

� How large should the protected part ofthe landscape be (e.g., what percentage ofan interconnected ditch system) to facili-tate recolonisation?

� How long does it take for populations in astream or pond to recover through inter-nal recovery versus recolonisation?

Recovery might take longer than expected insome cases. For example, removal of key-stone species could lead to alternative stablestates of the community (e.g., macrophyte-dominated versus algal-dominated aquaticsystems; Scheffer 1998). Also, Matthews etal. (1996) have suggested that ecosystemspossess a “memory” of stressful events andthat there might be a lack of return to origi-nal conditions after chemical exposure.Mechanisms of resistance are good examplesof such historical changes (PICT concept,Blanck 2002).

Whether populations are r- or K-selectedmay influence the speed and capacity for sys-tem recovery. Adaptation to stress may leadto a change not only towards less sensitivespecies but also towards species that havebetter recovery or recolonisation potential(e.g., shorter generation time and/or aeriallife stage). For example, in an aquatic fieldstudy (Liess, platform presentation), themain recolonisation process observed was byorganisms drifting from the small and non-contaminated areas upstream towards thecontaminated sites downstream. This resultsuggests that the area required as a source ofrecolonisation may not need to be large (de-pending on the species) to have a significanteffect on downstream sections.

For aquatic ecosystems, it was suspectedthat recovery might be faster in flowing sys-tems than in isolated ponds and lakes. It ispossible that recovery in interconnectedwater bodies may occur within 1 year of evensubstantial disturbances due to pesticides,provided the stressor is removed and de-pending on the species under consideration(i.e., length of generation cycle; Liess andVon der Ohe 2005). It is much harder tomake predictions for isolated water bodiesbecause the recovery processes in these arepoorly understood. A recent study in pondmesocosms treated with a non-persistent in-secticide showed that invertebrate communi-ties did not recover in physically isolated sys-tems as their composition diverged from thatof control (untreated) ponds, whereas fullrecovery was achieved after about 2 months


45

CONTENTS

in ponds for which recolonisation was possi-ble (Lagadic et al. 2004). However, becausethere are often gradients of pesticide concen-tration within larger water bodies, there arefrequently refugia and thus potential for in-ternal recovery. In addition, it is argued thatwhen there is frequent occurrence of stres-sors other than pesticides, the community isr-selected and able to recover faster from adisturbance than a community that is rarelyexposed to stressors. For example, a study ofephemeral ditches in northern Germanyfound no differences in communities be-tween ditches within arable areas comparedto those alongside meadows (Sönnichsen2002). This result was attributed to the highrecovery or recolonisation potential of thesystems, possibly linked to the ability of thespecific community to recover quickly be-cause it consisted of species that recoverfrom periodic drought. This hypothesis wasfurther supported by the absence of specieswith a low recovery or recolonisation poten-tial. It was also proposed by Lagadic et al.(2002) to explain the absence of effects of in-secticides on invertebrate communities incoastal wetlands subjected to chemical mos-quito control .

For terrestrial invertebrates, field data onrecovery from applications of specific chemi-cals have been summarised in a comprehen-sive study (van Straalen and van Rijn 1998).Some of the outcomes, also mentioned inRömbke’s platform presentation, are as fol-lows:� For Collembola, the potential for recovery

from adjacent, untreated field margins orunsprayed areas of crop would have beenoverestimated (e.g., Frampton 2002). Amanipulative study carried out withinDEFRA Project PN0934 also suggests thatlimited recovery of Collembola wouldoccur from field edges if OP insecticideswere used, despite high abundance anddiversity of Collembola at the field edge.However, Collembola immigration fromfield edges did occur in a study withoutpesticide use (Alvarez et al. 2000).

� For earthworms, recovery strongly de-pends on the persistence of the pesticides(Jones and Hart 1998), and the role of re-

colonisation is not clear (Heimbach 1997;Mather and Christensen 1998).

For terrestrial vertebrates such as birds, visi-ble occurrence of poisoning (loss of a few in-dividuals) may be considered unacceptableby management authorities. Regulatory ac-tion should thus be taken well below a levelat which the population might be directly af-fected. However, indirect effects of pesticideshave been demonstrated for farmland birdspecies, and these could reduce the chancesof achieving an overall recovery of the popu-lations. Therefore, the issue of recovery andrecolonisation is relevant for small mam-malian species that are sedentary in thetreated area (e.g., small rodents in pasturesor orchards) as well as for birds and largermammals.

MODELLING RECOVERYThe incorporation of information about re-covery in recolonisation and populationmodels would be very useful for effects pre-diction and has the potential to assist in hy-pothesis testing. It would be desirable to usegeneralised information on recolonisationstatistics for different species to generate hy-potheses and to design experiments.

So far, recolonisation data suitable forsuch models are limited in availability andmainly relate to aquatic ecosystems. Theydeal with reproduction, drift, and dispersal.For example, data on the recovery potentialof organisms are available at the family levelon the PondFX website(www.ent3.orst.edu/PondFX) and partly atthe species level in Liess and Von der Ohe(2005). An extensive database at the specieslevel is currently being collected in a cooper-ative project among Alterra (G. Arts), ThePonds Conservation Trust (J. Biggs), and theUFZ Centre for Environmental Research (M.Liess).

3.2.4 BIOASSAYS AND BIOMARKERS

To help establish causality between exposureand effects, a combined approach using ex-perimentation (e.g., in situ bioassays, bio-markers, mesocosms) and field observationmay be useful (Liess and Schulz 1999; Sibly


46

pa rt b

CONTENTS

www.ent3.orst.edu/PondFX

et al. 2000; Schulz et al. 2002; De Coen andJanssen 2003a, 2003b; Hanson and Lagadic2003; Jergentz, Passacq et al. 2004). It is dif-ficult to determine whether in situ bioassaysare more or less sensitive than field trials be-cause the sensitivity depends on the bioassaymethodology (Wiles and Frampton 1996).On the one hand, it was suggested that theymight be more sensitive than field trials be-cause of increased exposure in cages andlack of escape opportunities (e.g., with Gam-marus pulex; Schulz and Liess 1999a. On theother hand, other studies showed that in situbioassays and field trials have similar sensi-tivity (Frampton 1999; Liess and Schulz1999). Nevertheless, it was concluded that insitu bioassays could be useful systems, par-ticularly when natural populations showvariable demographics and when direct neg-ative effects of pesticides are concerned.

It was also concluded that biomarker se-lection on the basis of the mode of action ofpesticides is essential in order to discrimi-nate between various stressors (Lagadic etal. 2002). Indeed, interpretation of resultscan be difficult because of problems in at-tributing a response to a single toxicant. Se-lection of biomarkers should also be basedon their physiological role and, more specifi-cally, on their metabolic implications in indi-vidual performance (growth, reproduction,etc.). Their biological relevance therefore de-pends on the possibility of extrapolatingfrom biomarker responses to population-level endpoints (Migula 2000; Scott-Fords-mand and Weeks 2000). Nevertheless, be-cause exposure is difficult to quantify, espe-cially for vertebrates, the biomarker ap-proach is very useful.

3.3 RELEVANCE OF THE STUDY SITUATION

3.3.1 RELEVANCE OF THE LANDSCAPE

The system that is monitored should be rep-resentative of the agricultural landscape,and the challenge is in defining this. Forwater bodies of the UK, there are examplesin which classification has been systematical-ly achieved (e.g., Anonymous 2003). In thisclassification, 12 agricultural landscape types

were defined using soil, hydrogeological,and cropping descriptors; then, the morpho-logical, physicochemical, and biologicalproperties of water bodies in the differentlandscapes were analysed by compiling dis-parate datasets, and the results were enteredinto a database to facilitate access to the in-formation. A step in the same direction wasmade by identifying the so-called EURO-Soils that are considered representative forsoils of the EU (Gawlik et al. 1996).

3.3.2 RELEVANCE OF THEPOLLUTION SCENARIO

A study can be directed towards worst-caseor typical scenarios or designed to representthe full range of real scenarios. The choice ofan appropriate scenario will depend on thepurpose and the needs of the study. Howev-er, the system investigated should have thecapacity to respond to the range of contami-nation expected.

3.3.3 ECOLOGICAL RELEVANCE

Representatives of the main taxa and of thedifferent life-cycle strategies (multi- to semi-voltine species, r- and K-species) should bepresent in sufficient numbers for adequatestatistical analysis. The influence of otherstressors should be minimised unless theyare typical for the region investigated. Forexample, if drying or dredging is a typicalenvironmental stressor in a specific ecosys-tem and the community is therefore domi-nated by r-strategist species, then this systemcan be considered as representative. Howev-er, excessive maintenance in non-targetareas (e.g., very frequent dredging of streambeds) should not be considered representa-tive.

3.3.4 SPATIAL AND TEMPORAL RELEVANCE

The study must be spatially (environmentalvariables) and temporally (climate, usagepatterns) representative. At a smaller scale(e.g., small water bodies) there may be high


47

CONTENTS

variability; therefore, the number of sites ortimes of sampling must be sufficient to re-duce the uncertainties. Similarly, at a largerscale, localised impacts or peak concentra-tions may not be detected or captured. Spa-tial and temporal scales are particularly im-portant for the understanding of the recov-ery from pesticide applications.

4 LINKING TIERS ANDEXTRAPOLATION

The EU Uniform Principles for the assess-ment of PPPs require that if the preliminaryrisk characterisation indicates potential con-cerns, registration cannot be granted unlessit can be demonstrated that “… under fieldconditions no unacceptable impact on the vi-ability of exposed organisms…” occurs. Todate, such assessments have been made byconducting higher-tier studies (e.g., meso-cosms), and thus the relevance of these as-sessments to field situations may have cer-tain limitations. Examples of comparisonsbetween observed and predicted effects arepresented below. Some potential problems incomparing effects observed in the field andeffects occurring under controlled conditionsare also listed.

4.1 EXAMPLES OF COMPARISONBETWEEN OBSERVED ANDPREDICTED EFFECTS

4.1.1 EFFECTS ON AQUATIC ORGANISMS

When observed and predicted effects arecompared, various endpoints can be consid-ered. The following endpoints were dis-cussed during the workshop.

Short-term effectsObserved effects were in line with predictedeffects. Indeed, measured pesticide concen-trations of around 1:10 of the acute 48-hlethal concentration (LC50) of Daphniamagna led to a short-term reduction in theabundance and number of sensitive species.Below 1:100 of the acute 48-h LC50 of Daph-nia magna, no such effect was observed

(Liess, platform presentation; Liess and Vonder Ohe 2005). Hence, an assessment factorof 100 on the acute LC50 would prevent pos-sible short-term effects. Another investiga-tion showed a decreased effect of spray drifton the invertebrate community in the pres-ence of a buffer zone (Schäfers, platformpresentation). However, this positive effectcould not be quantified because pesticideconcentrations were not measured.

Long-term effectsEffects on community composition were ob-served at concentrations for which no effectwas predicted. Measured pesticide concen-trations of around 1:100 of the acute 48-hLC50 of Daphnia magna led to a long-termreduction of a proportion of sensitivespecies. However, because the levels of cont-amination may have been quantified insuffi-ciently, it remains uncertain at which con-centration these changes occur (Liess, plat-form presentation; Liess and Von der Ohe2005). In addition, it was stated that the ef-fect of pesticides on long-term alterations ofcommunity structure depends not only onthe sensitivity of the species but also on thereproduction rate of the exposed organisms.The importance of such life-history charac-teristics in governing recovery was shown bySherratt (1999) and Liess and Von der Ohe(2005).

RecolonisationThe proportion of species potentially affect-ed depends not only on the contamination atthe site but also on the presence of undis-turbed stream sections upstream. The levelsof biological impairment observed at siteswith high concentrations of pesticides andgood habitat quality (indexed as undisturbedupstream sections) were similar to those atsites where pesticide concentrations werelow but habitat quality was poor (Liess, plat-form presentation; Liess and Von der Ohe2005). Thus, landscape and land use infor-mation are important to consider in the pre-diction of effects but are not included in riskassessment.


48

pa rt b

CONTENTS

4.1.2 EFFECTS ON SOIL INVERTEBRATES

Van Straalen and van Rijn (1998) comparedpredicted and observed effects and targetedthe study on recovery times. They showedthat in 25 out of 32 comparisons, recovery inthe field takes place within a time periodnecessary for degradation of the pesticide toa no-effect level. In some cases, recovery wasslower than expected, a possible result of in-direct effects rather than of direct toxicity tothe organisms considered.

4.1.3 EFFECTS ON BIRDS

Mineau (2002) used regression methods toanalyse the relationship between the occur-rence of avian mortality in pesticide fieldstudies (mostly conducted in North Americafor regulatory purposes) and a number ofpotential predictor variables, including acuteoral and dermal toxicity, Henry’s Law con-stant (reflecting potential for inhalation ex-posure), and application rate. The bestmodel was able to classify “safe” and “lethal”applications in the study sample with betterthan 80% success.

In a follow-up study (Mineau, platformpresentation), predictions of this model werecompared with acute toxicity–exposure ra-tios (TERacute) calculated according to theEU Guidance Document on risk assessmentfor mammals and birds (Anonymous 2002).They also compared the TERacute results di-rectly with the occurrence of mortality in theindividual field studies. The results showedthat a substantial proportion of pesticideswere misclassified, therefore suggesting thatthe predictive power of the current Tier I riskassessment scheme is not satisfactory. Im-proving the parameters for the current riskassessment model (e.g., residue per unitdose [RUD]) is desirable but not sufficientbecause the model itself seems to have cer-tain deficits. For example, this study showedthat dermal exposure is not as negligible asthe current risk assessment scheme suppos-es, at least for organophosphate pesticides.Improving the model structure may thus bethe best way to enhance the discriminatorypower of the assessment procedure.

4.2 LIMITATIONS IN COMPARINGOBSERVED AND PREDICTED EFFECTS

4.2.1 LIMITATIONS OF FIELD STUDIES

CHARACTERISATION OF EXPOSURE Characterisation of exposure can be insuffi-cient for the following reasons:� Sampling problems (e.g., missed peak of

contamination, sample deterioration) re-sulting in an underestimation of pesticideconcentrations. In the case of birds andmammals, exposure is not quantified inthe same way as for other species becauseit is a function of environmental concen-trations, time spent in the area, feedingpreferences of the species, etc.

� Analytical problems such as matrix inter-ference and presence of undetected conta-minants or of those exerting biological ef-fects below analytical detection levels.

ROUTES OF EXPOSURE Whatever the study type (laboratory, micro-cosm, mesocosm, monitoring, or fieldstudy), the routes of exposure must be wellunderstood so that relationships betweenboth types of observations can be interpret-ed. This may be easier to achieve for short-term effects than for long-term effects be-cause exposure to stressors (including pesti-cides) and the resulting effects are more dif-ficult to characterise over longer timeperiods.

DETECTABILITY OF EFFECTS The statistical power to detect effects of pes-ticides in monitoring and field studies is gen-erally less than for studies performed in lab-oratory test systems or in indoor micro- ormesocosms because of greater natural vari-ability and more confounding factors. How-ever, the combination of an elevated numberof investigated sites with the appropriate sta-tistics may enable us to discriminate the im-pacts of individual stressors and to reveallinks between exposure and effects as shownabove.

SIGNIFICANCE OF ENDPOINTSTo detect field effects of pesticides, relevantendpoints have to be monitored. For exam-

w o r ks h o p sy n t h e s i s 4 · l i n k i n g t i e r s a n d e x t ra p o l a t i o n

49

CONTENTS

ple, monitoring studies on birds and mam-mals are focused mostly on lethality. Thus,one should be aware that the absence of ob-served effects does not prove that no effectoccurs. Indeed, the long-term use of pesti-cides is more likely to result in sublethal ef-fects (e.g., direct effects such as eggshellthinning, indirect effects such as decline ofpopulation following removal of food-chaincomponents) than in the death of a bird bypoisoning. In light of this situation, monitor-ing studies revealing longer-term changes incommunity structure would be more appro-priate.

LONG-TERM VERSUS SHORT-TERM EFFECTS,DIRECT AND INDIRECT EFFECTSMost effects shown in monitoring and fieldstudies of aquatic invertebrates and birdsand mammals are chronic, long-term, andindirect effects. Therefore, it is insufficient toassess only acute effects (e.g., lethality),even though this is currently done, particu-larly in reactive monitoring schemes for birdsand mammals. Indeed, if observed and pre-dicted effects are to be compared, the sameendpoints must be considered.

RELEVANCE OF THE STUDY SITUATIONThe conditions selected must be representa-tive and appropriate for the needs of thestudy performed. These include the charac-teristics in terms of landscape, pollution sce-nario, ecological characteristics, and spatialand temporal scales.

4.2.2 LIMITATIONS OF MESOCOSMEXPERIMENTS

Although small-scale and large-scale meso-cosm studies can play an important role inestablishing causality, testing hypotheses,and interpreting field data, some limitationsremain.

EXPOSURE TO MULTIPLESUBSTANCES OR STRESSORSSingle-substance tests are not necessarilypredictive of field situations, where exposureis usually more complex (multiple sub-stances or stressors). Although controlled ex-

periments may provide fairly good effectspredictions for known exposures (e.g., thosecompounds or other stressors that are exper-imentally applied), these predictions are lessrobust when exposure is more complicated(e.g., unknown compounds). However, con-trolled experiments such as mesocosm stud-ies remain the best way to approach multipleexposures (Brock, platform presentation), ascompared to simplistic laboratory test sys-tems, particularly because the former in-clude the effects of any biologically activedegradation products.

DIRECT AND INDIRECT EFFECTSControlled experiments are probably predic-tive of direct effects, but they may not beprotective in all cases. Indeed, indirect ef-fects and long-term effects may depend onfactors that may not be realistically incorpo-rated into test systems. One relevant exam-ple is the problem of communities in test sys-tems, which may have a higher percentage ofspecies with a short generation time com-pared to natural communities in permanentstreams and ponds in the field. Test systemsare usually established for a limited period oftime, and therefore r-strategists tend to dom-inate the communities if colonisation of thetest systems is only natural.

RECOVERYControlled experiments are predictive ofthreshold effects concentrations when com-pared to single-species tests. However, recov-ery through internal recovery and recoloni-sation is often context dependent (e.g., vol-tinism in the particular assemblage, hydro-logical connectivity, latitude). Recovery datafrom mesocosm studies should be carefullyconsidered because they are not directly rep-resentative of natural conditions. Indeed,compared to the field situation, the recoveryprocesses may be (1) slower for some species(e.g., those migrating inside of the waterbody) or (2) faster for others (e.g., those re-covering through aerial stages), especiallywhen control and exposed mesocosms areclose to each other (Lagadic et al. 2004).


50

pa rt b

CONTENTS

TIME SCALEConsideration of time scale in relation to thegeneration time of the investigated speciesshift is essential in test systems. Effects varyaccording to duration of exposure, and thusthe time scale with respect to persistence ofeffects (e.g., indirect effects, recovery) isalso important in the assessment of environ-mental risks. Nevertheless, with the excep-tion of some studies in outdoor mesocosms,controlled experiments are usually shorterthan 1 year. There is thus a gap betweenwhat the studies can and/or set out todemonstrate and the appropriate test designto achieve this goal. The time scale in rela-tion to the generation time of the investigat-ed species is also of importance, especiallywhen recovery is being studied.

5 RESEARCH REQUIREMENTSAND FURTHER STRATEGIES

The main research requirements and furtherstrategies identified in this workshop aresummarised below. Outcomes of field stud-ies (field monitoring and field experimentalstudies) performed while addressing thesevarious issues will facilitate further interpre-tation and characterisation of pesticide ef-fects in the field.

5.1 RESEARCH REQUIREMENTS

5.1.1 IMPROVED EXPOSURE ASSESSMENT

Exposure assessment should be improved inthe different ecosystems considered withinrisk assessment. There is a need to improvesampling techniques and strategies availableto support field studies. Incorporation oftime-varying exposure into risk assessment isan important requirement. Mathematicalmodels for the aquatic compartment willoften predict the variation in exposure con-centrations with time, but this information isnot used within the risk assessment proce-dure. Ecological aspects of exposure (e.g.,timing and duration of presence of sensitivelife stages) also require further development.

5.1.2 DEFINING TARGET IMAGES

Target images need to be defined to evaluatemonitoring studies. Gathering extensiveknowledge about the variability of ecosys-tems, the effects of pesticides, and the subse-quent recovery of communities in the fieldwill enable us to define principles in settingtarget images.

5.1.3 ASSESSMENT OF DIRECT EFFECTS

Focusing on sensitive species is very usefulfor detecting direct effects of pesticides, pro-vided that the spectra of sensitive species arethe same for both active substances and theirmetabolites or degradation products. Inorder to strengthen field assessments, itwould be desirable to link mode of actionand species-specific metabolic pathways toobserved effects.

5.1.4 ASSESSMENT OF INDIRECT EFFECTS

One important outcome of the workshop isthe understanding that indirect effects are ofhigh importance when effects of pesticides inthe field are studied. Nevertheless, generali-ties in defining ecological traits that indicatespecific effects are missing (i.e., species withhigh recovery potential may benefit frompesticides). In addition, it was stated that in-direct effects are often difficult to demon-strate. Targeted laboratory investigations,micro- or mesocosm studies, and in situbioassays can help to elucidate underlyingmechanisms and thus enable a better obser-vation of indirect effects. Information on in-direct effects (especially on their origin)could also be gained from a survey of ecolog-ical literature, especially in the field of func-tional ecology (e.g., food web dynamics,temporal and/or spatial competition be-tween species).

w o r ks h o p sy n t h e s i s 5 · re s e a rc h re qu i re m e n t s a n d f u r t h e r st ra te g i e s

51

CONTENTS

5.1.5 ADDITIONAL BIOLOGICAL ANDECOLOGICAL KNOWLEDGE

Additional biological and ecological knowl-edge is needed, for example, to better assessdirect and indirect effects or the potential forrecovery and recolonisation. Further infor-mation is needed regarding the following:� Species diversity (e.g., through well-edu-

cated taxonomists and better taxonomictools such as computer-aided identifica-tion keys). This is especially relevant forterrestrial invertebrates.

� Ecological characteristics of key species inthe different agro-ecosystems (e.g., foodsources for terrestrial invertebrates, lifehistory, ecophysiology, sensitivity to pesti-cides, life cycle, dispersal, behaviouraldata, interactions or competition or pre-dation between species and trophic net-works).

5.1.6 ENVIRONMENTAL PARAMETERS

The potential of environmental parameters(e.g., UV radiation, drought) to alter sensi-tivity of the different levels of biological or-ganisation (individual, population, commu-nity) needs to be better defined. This issueleads to further questions such as the differ-ent sensitivity of species or communities ac-cording to their geographical origin (i.e.,northern and southern Europe), populationand community genetics (biodiversity, resis-tance and fitness, extinction risks), or stabili-ty and continuity of ecosystems (drying ver-sus permanent water bodies) in relation tothe variations of biological response to pesti-cides.

5.1.7 MIXTURE TOXICITY

It is necessary to reduce the uncertaintyabout whether or not risk assessment thatconsiders only single compounds is also con-servative for multiple compounds or stressesthat may occur in the field (Brock, platformpresentation). An investigation on aquaticecosystems presented during the workshop(Liess, platform presentation) supports the

view that single-chemical risk assessment ofthe most toxic compound could be sufficientfor the mixture of pesticides in the scenariothat was presented (small agriculturalstreams). This will especially be the case forhigh-risk compounds entering small catch-ments. A multiple exposure with severalcompounds at comparable toxicity is unlikelyunder such scenarios. Other experiments,however, showed that even compounds atconcentrations below the no-observed-effectconcentration (NOEC) can contribute to thetoxicity of mixtures (Rajapakse et al. 2002;Walter et al. 2002). Also, reduction in thetoxicity of active ingredients by chemicalspresent in the mixture (i.e., antagonism) ortoxicity of spray-tank adjuvants cannot beruled out (Jumel et al. 2002; Caquet et al.2005). This finding underlines the need forfurther information.

5.1.8 MAGNITUDE OF RECOVERY

Several examples during the workshopshowed the great importance of recovery forpopulation and community dynamics. Moreinformation on quantifying recovery wouldhelp us to better assess the overall effects ofpesticides. Empirical information and fur-ther development of appropriate modellingapproaches would be very useful for this pur-pose.

5.1.9 BENCHMARK CASES

The establishment of well-developed genericbenchmark cases could provide referencesfor other compounds and be used for valida-tion of laboratory-based risk assessmentschemes.

5.2 FURTHER STRATEGIES

5.2.1 DEFINING ACCEPTABILITY

Criteria for defining the acceptability of ob-served effects need to be further establishedbecause they should include both scientificand societal considerations. The role of sci-

5 · re s e a rc h re qu i re m e n t s a n d f u r t h e r st ra te g i e s

52

pa rt b

CONTENTS

ence is to identify the options and conse-quences so that society can make decisions.For example, the selection of reference sitesand target images depends partly on what isconsidered acceptable by society. It is neces-sary to establish the target image and then todecide how much deviation from this targetimage is acceptable.

5.2.2 SPATIAL UNIT OF ASSESSMENT

Modelling of landscape-scale impacts helpsto address the question of what magnitude ofimpact is significant and/or acceptable. Spa-tially explicit modelling using GIS combinedwith biogeography and dispersal data for thetarget or non-target species could be an ef-fective way of achieving this goal. Referencedata on the undisturbed habitat or communi-ty are necessary to provide a benchmark forassessing changes and measuring recovery.For example, recovery of impacted aquaticcommunities may be facilitated by recoloni-sation from undisturbed water bodies.

5.2.3 HARMONISATION WITHIN EUROPE

The quality and comparability of future fieldstudies from different locations in Europeneed to be improved. This improvementshould be facilitated by the strengthening ofa European network of expertise in this re-search area as a result of the workshop.

5.2.4 GOOD AGRICULTURAL PRACTISES

The studies presented at the workshopdemonstrated that in many cases it is diffi-cult to distinguish between observed effectsattributable to pesticide use according toGAP and those arising from misuse. Futurestudies designed to resolve this problem areworthwhile applications of monitoring andexperimental studies and are highly recom-mended.

6 IMPLICATIONS FORREGULATORY RISK ASSESSMENTAND MANAGEMENT

6.1 CURRENT VALUE AND SCOPE OFFIELD STUDIES IN RISK ASSESSMENT

The general consensus was that it is difficultto envisage field studies (monitoring studiesand experimental studies) being incorporat-ed into decision making with respect to reg-istration for a single active substance, be-cause the field situation is more complex andhence differs from the higher-tier tests per-formed as part of the risk assessment. How-ever, they provide useful information, as de-scribed below.

6.1.1 INFORMING AND EVALUATING THERISK ASSESSMENT PROCESS

Field studies are useful to monitor pesticideuse after approval in the most realistic way,as a check on pre-authorisation predictions.Therefore, they can be used to verifywhether the current risk assessment providessuitable environmental protection by gener-ally validating (or invalidating) the ecologi-cally acceptable concentration derived fromhigher-tier test systems.

Field studies are potentially useful forreregistration purposes. They can also helpin the verification and improvement of riskassessments because the information gath-ered can be used to formulate questions andguide the selection and design of studies at apre-registration stage (e.g., better genericmesocosm studies).

6.1.2 VERIFYING RISK MITIGATIONMEASURES

Monitoring studies provide a potential toolfor checking the effectiveness of mitigationmeasures applied to use of a pesticide wherepotential risks are identified in the registra-tion procedure. Results may help to improverisk mitigation measures where necessary.The studies may provide a possible means foridentifying where further enforcement of or

w o r ks h o p sy n t h e s i s 6 · i m p l i c a t i o n s f o r re g u l a to r y r i s k a s s e s s m e n ta n d m a n a ge m e n t

53

CONTENTS

training for such measures is needed(though it is acknowledged that the estab-lishment of monitoring sites may stimulatebetter practise than is found more generally;see Kreuger, platform presentation). Demon-stration of the efficacy of mitigation mea-sures to farmers and other stakeholders mayalso be an important function.

Monitoring to evaluate mitigation mea-sures may involve measurement of exposureand/or effects, depending upon the measureand risk under consideration.

Examples: � Impacts can differ when spray direction is

changed and can decrease significantlywhen a reasonable buffer zone is ob-served. For example, the potential bene-fits of effective risk mitigation wereshown in the Altes Land study when thebuffer zone was 3 m to 5 m (Schäfers,platform presentation).

� Vegetated areas such as constructed wet-lands serve as tools to mitigate exposureand related risk (Rodgers and Dunn1992).

� Relatively small areas in streams that donot suffer impacts (e.g., a forested area)can significantly mitigate impacts in con-nected water bodies (Liess and Von derOhe 2005).

6.1.3 RISK COMMUNICATION STRATEGIES

Monitoring studies offer the potential forbetter communication of risk assessment de-cisions to policy makers, the general public,and other stakeholders because they can in-form on the magnitude or absence of ecolog-ical effects occurring on a “real” scale. Cou-pled with GIS technology, these studies mayprove a useful tool for visualising and com-municating results.

6.1.4 BROADER SUSTAINABILITY ISSUES

Monitoring studies may provide a means forlooking at the overlap between the variousenvironmental protection regulations, ensur-ing that management practises developed for

one area do not have an impact on another.For example, there will be applications in en-suring complementarities between pesticideregulation and specific actions to support na-ture conservation and protection of Red Listspecies or implementation of the WaterFramework Directive. Issues in which differ-ent environmental regulations overlap maybe highly complex, and it is essential to spec-ify the appropriate questions before design-ing the study and to frame problem formula-tion depending on the regulatory applicationso that the different requirements are met.Ultimately, monitoring studies may assistwith checking that economic and social de-velopment proceeds in a way that is environ-mentally sustainable.

6.2 IMPLICATIONS FOR FUTUREREGULATORY RISK ASSESSMENT

Relevant issues resulting from currentknowledge and future research develop-ments should eventually be implemented inregulatory risk assessment to enable appro-priate calibration and verification of theschemes.

One important conclusion of the workshop isthat exposure should be defined accuratelyso that critical effect concentrations detectedin the field and those predicted from esti-mated exposure under Directive 91/414/EECcan be compared. In addition, results frommonitoring studies highlighted issues thatwould be relevant for increasing realism infuture risk assessment. For example, the fol-lowing parameters or issues should be con-sidered and/or implemented because theyare or could be important to assess the oc-currence of effects under field conditions.� Chemical monitoring: This can be effec-

tively used, for example, in surface watersfor regulatory purposes to show that riskmitigation measures work properly. How-ever, to date the prediction of effects onthe basis of such data is only approximate,because many processes can influence bi-ological effects and their recovery.

� Route of exposure: The relative impor-tance of different routes of exposure

6 · i m p l i c a t i o n s f o r re g u l a to r y r i s k a s s e s s m e n ta n d m a n a ge m e n t

54

pa rt b

CONTENTS

should be appropriately implemented inregulatory risk assessment. For example,studies on birds pointed out that the roleof dermal exposure is likely to be underes-timated in current regulatory risk assess-ment schemes.

� Mixture toxicity: Some results showedthat risk prediction based on the mosttoxic product could lead to reasonableregulatory decisions. However, further re-search is needed to determine whetherrisk assessment that considers only singlecompounds is generally conservative formixture toxicity.

� Target images: Most of the issues citedabove should help to define target imagesthat are needed to evaluate monitoringstudies. Close contacts between monitor-ing activities under Directive 91/414/EECand the Water Framework Directive areneeded so that overlaps can be avoided.

� Environmental parameters: These may in-crease or decrease the sensitivity of or-ganisms towards pesticides. Such parame-ters should therefore be included in riskassessment.

� Biological and ecological data: More dataare needed on the different biological andecological traits of non-target species. Forexample, these data could be added to theinformation on landscape and analysedusing GIS systems to facilitate use for reg-ulatory purposes and to improve the real-ism in risk assessment.

� Indirect effects: These have been ob-served for all groups of organisms, andtheir relevance with regard to the long-term repercussions on populations andcommunities should be investigated. Thecausal relationship to direct effects couldbe better investigated by, for example, theuse of biomarkers.

� Recovery and recolonisation: Theseprocesses have important implications forassessing the occurrence of effects. But itwill be challenging to include these con-cepts into a risk assessment in a field-rele-vant manner. Indeed, repeated exposureor other stressors might disturb theseprocesses. Furthermore, species with alow recovery potential (i.e., univoltinespecies and/or species with low dispersal

potential) should be considered morecarefully than species with high recoverypotential (i.e., short generation timeand/or high dispersal ability).

� Incidence schemes: The organisation ofsuch schemes should be established or im-proved when monitoring studies are diffi-cult to conduct (i.e., for vertebrates suchas birds). The schemes are a safety net forthe regulatory system and can highlightthe need for appropriate risk managementand issues regarding misuse and abuse.

� Landscape analysis: Landscape character-istics can modify the importance of fieldeffects of pesticides and recovery (e.g.,conservation headlands support a diversecommunity of terrestrial organisms inagricultural landscapes; forested streamsections facilitate the diversification ofaquatic communities through colonisa-tion). Classifying landscape characteris-tics and including habitat quality in riskassessment may put the risks of contami-nation into context with respect to otherstressors.

7 REFERENCES

Aebischer NJ 1991. Twenty years of monitoringinvertebrates and weeds in cereal fields in Sus-sex. In: Firbank LG CN, Darbyshire JF, PottsGR, editor. The Ecology of Temperate CerealFields. Oxford: Blackwell Scientific Publica-tions. p 305–331.

Akkerhuis G. 1993. Walking behavior and popu-lation density of adult Linyphiid spiders in rela-tion to minimizing the plot size in short- termpesticide studies with pyrethroid insecticides.Environ Pollut 80:163–171.

Aldridge CA, Hart ADM. 1993. Validation of theEPPO/CoE risk assessment scheme for honey-bees. October 26–28, 1993; Wageningen, TheNetherlands, p 37–41.

Alvarez T, Frampton GK, Goulson D. 2000.The role of hedgerows in the recolonisation ofarable fields by epigeal Collembola. Pedobiolo-gia 44:516–526.

Anonymous. 2002. Guidance Document on RiskAssessment for Birds and Mammals. UnderCouncil Directive 91/414/EEC, DocumentSANCO/4145/2000. Brussels, Belgium. 44 p.

w o r ks h o p sy n t h e s i s 7 · re f e re n c e s a – b

55

CONTENTS

Anonymous. 2003. Aquatic ecosystems in the UKagricultural landscape. Department for Envi-ronment, Food, and Rural Affairs (DEFRA), UK.Project nr PN0931.

Anonymous. 2004. Assessing the Indirect Effectsof Pesticides on Birds. Unpublished report toDepartment for Environment, Food, and RuralAffairs (DEFRA), UK. Project nr PN0925.

Bach M, Huber A, Frede HG. 2001. Input path-ways and river load of pesticides in Germany –a national scale modeling assessment. WaterSci Technol 43:261–268.

Bach M, Huber A, Frede HG, Mohaupt V,Zullei-Seibert N. 2000. Schätzung der Ein-träge von Pflanzenschutzmitteln aus der Land-wirtschaft in die OberflächengewässerDeutschlands. Berlin: Schmidt-Verlag. 274 p.

Barnett EA, Fletcher MR, Hunter K, SharpEA. 2002. Pesticide Poisoning of Animals 2001:Investigations of suspected Incidents in theUnited Kingdom. Department for Environment,Food and Rural Affairs, London (UK), 43p.

Bembridge J, Kedwards TJ, Edwards PJ. 1998.Variation in earthworm populations and meth-ods for assessing responses to perturbations.In: Sheppard SC, Bembridge JD, Holmstrup M,Posthuma L, editors. Advances in earthwormecotoxicology. Pensacola (FL): SETAC. p341–352

Berenzen N, Kumke T, Schulz HK, Schulz R.2005. Macroinvertebrate community structurein agricultural streams: Impact of runoff-relat-ed pesticide contamination. Ecotoxicol EnvironSaf 60:37–46.

Berny PJ, Buronfosse T, Buronfosse F,Lamarque F, Lorgue G. 1997. Field evidenceof secondary poisoning of foxes (Vulpes vulpes)and buzzards (Buteo buteo) by bromadiolone, a4-year survey. Chemosphere 35:1817–1829.

Blanck H. 2002. A critical review of proceduresand approaches used for assessing pollution-in-duced community tolerance (PICT) in bioticcommunities. Human Ecolog Risk Assess8:1003–1034.

Boatman ND, Brickle NW, Hart JD, MilsomTP, Morris AJ, Murray AWA, Murray KA,Robertson PA. 2004. Evidence for the indirecteffects of pesticides on farmland birds. Ibis146:131–143.

Brickle NW, Harper DGC, Aebischer NJ,Cockayne SH. 2000. Effects of agricultural in-tensification on the breeding success of cornbuntings Miliaria calandra. J Appl Ecol37:742–755.

Browne SJ, Aebischer NJ. 2004. Temporalchanges in the breeding ecology of Europeanturtle doves Streptopelia turtur in Britain, andimplications for conservation. Ibis 146:125–137.

Campbell PJ, Arnold DJS, Brock TCM,Grandy NJ, Heger W, Heimbach F, MaundSJ, Streloke M. 1999. Guidance document onhigher-tier aquatic risk assessment for pesti-cides (HARAP). Brussels: SETAC Europe. 178 p.

Caquet T, Deydier-Stephan L, Lacroix G, LeRouzic B, Lescher-Moutoué F. (2005) Ef-fects of fomesafen, alone and in combinationwith an adjuvant, on planktonic communitiesin freshwater outdoor pond mesocosms. Envi-ron Toxicol Chem. Forthcoming.

Crane M, Whitehouse P, Comber S, Watts C,Giddings J, Moore DRJ, Grist E. 2003. Eval-uation of probabilistic risk assessment of pesti-cides in the UK: chlorpyrifos use on top fruit.Pest Manag Sci 59:512–526.

De Coen WM, Janssen CR. 2003a. The missingbiomarker link: Relationships between effectson the cellular energy allocation biomarker oftoxicant-stressed Daphnia magna and corre-sponding population characteristics. EnvironToxicol Chem 22:1632–1641.

De Coen WM, Janssen CR. 2003b. A multivariatebiomarker-based model predicting population-level responses of Daphnia magna. Environ Tox-icol Chem 22:2195–2201.

De Snoo GR, Scheidegger NMI, De Jong FMW.1999. Vertebrate wildlife incidents with pesti-cides: a European survey. Pestic Sci 55:47–54.

De Snoo GR, van der Poll RJ. 1999. Effect ofherbicide drift on adjacent boundary vegeta-tion. Agric Ecosys Environ 73:1–6.

EPFES. 2003. Guidance Document: Effects of PlantProtection Products on Functional Endpoints inSoil (EPFES). Römbke JHF, Hoy S, Kula C,Scott-Fordsmand J, Sousa P, Stephenson G,Weeks J, editors. Pensacola (FL): SETAC. 92 p.

EPPO. 1992. Guideline on test methods for evalu-ating the side effects of plant protection prod-ucts on honeybees. OEPP/EPPO Bull.22:203–216.

7 · re f e re n c e s b – f

56

pa rt b

CONTENTS

Ewald JA, Aebischer NJ. 1999. Pesticide use,avian food resources and bird densities in Sus-sex. Peterborough, UK: Joint Nature Conserva-tion Committee JNCC. Report nr 296. 148 p

Ewald JA, Aebischer NJ. 2000. Trends in pesti-cide use and efficacy during 26 years of chang-ing agriculture in southern England. EnvironMonit Assess 64:493–529.

Frampton GK. 1997a. The potential of Collembo-la as indicators of pesticide usage: Evidenceand methods from the UK arable ecosystem.Pedobiologia 41:179–184.

Frampton GK. 1997b. Species spectrum, severityand persistence of pesticide side-effects on U.K.arable springtail populations; 6–8 January1997; Montpellier, France. p 129–136.

Frampton GK. 1997c. Off-target effects of pesti-cides – are we targeting the right indicatorspecies for risk assessment? In: Haskell PT MP,editor. New Studies In Ecotoxicology, WelshPest Management Forum. Cardiff: Welsh PestManagement Forum. p 23–25.

Frampton GK. 1999. Spatial variation in non-tar-get effects of the insecticides chlorpyrifos,cypermethrin and pirimicarb on Collembola inwinter wheat. Pesticide Science 55:875–886.

Frampton GK. 2000. Recovery responses of soilsurface Collembola after spatial and temporalchanges in long-term regimes of pesticide use.Pedobiologia 44:489–501.

Frampton GK. 2001. Large-scale monitoring ofnon-target pesticide effects on farmlandarthropods in England. In: Johnston JJ, editor.Pesticides and Wildlife. Washington DC: Ameri-can Chemical Society (ACS Symposium Series771). p 54–67.

Frampton GK. 2002. Long-term impacts of anorganophosphate-based regime of pesticides onfield and field-edge Collembola communities.Pest Manag. Sci. 58:991–1001.

Gawlik BM, Sotiriou N, Kuhnt G, Karcher W,Kettrup A, Muntau H. 1996. European refer-ence soils as a common basis for soil testing ofenvironmental chemicals in the EU. FreseniusEnviron Bull 5:610–618.

Greig-Smith PW, Frampton GK, Hardy AR.1992. Pesticides, cereal farming and the envi-ronment: the Boxworth Project. London, UK.288 p.

Hanson ML, Lagadic L. 2005. Chitobiase activityas an indicator of aquatic ecosystem health.Aquat Ecosys Health Manag. Forthcoming.

Heimbach F. 1997. Field tests on the side effects ofpesticides on earthworms: Influence of plotsize and cultivation practices. Soil Biol Biochem29:671–676.

Holland JM. 2002. Carabid beetles: their ecolo-gy, survival and use in agroecosystems. In: Hol-land JM, editor. The agroecology of carabidbeetles. Andover (UK): Intercept. p 1–40.

Holland JM, Southway S, Ewald JA, BirkettT, Begbie M, Hart J, Parrot D, Allcock J.2002. Invertebrate chick food for farmlandbirds: spatial and temporal variation in differ-ent crops. Aspects Appl Biol 67:27–34.

Hommen U, Schäfers C, Ross-Nickoll M,Ratte T. 2004. Auswertung der wichtigsten inDeutschland durchgeführten Monitoringstudi-en zu Auswirkungen von Pflanzenschutzmit-teln auf Nichtzielorganismen. Fraunhofer, IME.

Inglesfield C. 1989. Pyrethroids and TerrestrialNon-target Organisms. Pestic Sci 27:387–428.

(ISO) International Organization for Stan-dardization. 1999. Soil quality – Effects ofpollutants on earthworms (Eisenia fetida). Part3: Guidance on the determination of effects infield situations. Geneva: ISO. Report nr 11268-3. 8 p.

Jergentz S, Mugni H, Bonetto C, Schulz R.2004. Runoff-related endosulfan contamina-tion and aquatic macroinvertebrate response inrural basins near Buenos Aires, Argentina. ArchEnviron Contam Toxicol 46:345–352.

Jergentz S, Pessacq P, Mugni H, Bonetto C,Schulz R. 2004. Linking in situ bioassays andpopulation dynamics of macroinvertebrates toassess agricultural contamination in streams ofthe Argentine pampa. Ecotoxicol Environ Saf59:133–141.

Jones A, Hart ADM. 1998. Comparison of labora-tory toxicity test for pesticides with field effectson earthworm populations: a review. In: Shep-pard SBJ, Holmstrup M, Postuma L, editors.Advances in earthworm ecotoxicology. Pen-sacola (FL): SETAC. p 247–267.

Jumel A, Coutellec MA, Cravedi JP, LagadicL. 2002. Nonylphenol polyethoxylate adjuvantmitigates the reproductive toxicity of fome-safen on the freshwater snail Lymnaea stagnalisin outdoor experimental ponds. Environ ToxicolChem 21: 1876–1888

w o r ks h o p sy n t h e s i s 7 · re f e re n c e s f – l

57

CONTENTS

Kedwards TJ, Maund SJ, Chapman PF. 1999.Community level analysis of ecotoxicologicalfield studies: I. Biological monitoring. EnvironToxicol Chem 18:149–157.

Kendall RJ, Lacher TEJ. 1994. Wildlife toxicolo-gy and population modeling. Chelsea (MI):Lewis. 576 p.

Kennedy PJ, Conrad KF, Perry JN, Powell D,Aegerter J, Todd AD, Walters KFA, Pow-ell W. 2001. Comparison of two field-scale ap-proaches for the study of effects of insecticideson polyphagous predators in cereals. Appl SoilEcol 17:253–266.

Kreuger J. 1998. Pesticides in stream water with-in an agricultural catchment in southern Swe-den, 1990–1996. Sci Total Environ 216:227–251.

Lagadic L, Caquet T, Fourcy D, Heydorff M.2002. Évaluation à long terme des effets de ladémoustication dans le Morbihan; suivi de l’im-pact écotoxicologique des traitements sur lesinvertébrés aquatiques entre 1998 et 2001. Con-vention de Recherche, Conseil Général du Mor-bihan.

Lagadic L, Hanson ML, Graham DW, CaquetT, Knapp CW, Coutellec MA, Azam D. 2004.Influence of connectivity on the recovery ofaquatic ecosystems exposed to chemical stres-sors. Fourth SETAC World Congress, Portland,14–18 November 2004 (Platform Presentation –Abstract nr 756).

Leonard AW, Hyne RV, Lim RP, Chapman JC.1999. Effect of endosulfan runoff from cottonfields on macroinvertebrates in the NamoiRiver. Ecotoxicol Environ Saf 42:125–134.

Liess M. 1994. Pesticide impact on macroinverte-brate communities of running waters in agri-cultural ecosystems. Verh Internat Verein Lim-nol; Proceedings of the International Associationof Theoretical and Applied Limnology25:2060–2062.

Liess M. 1998. Significance of agricultural pesti-cides on stream macroinvertebrate communi-ties. Verh Internat Verein Limnol; Proc Int AssocTheor Appl Limnol 26:1245–1249.

Liess M, Schulz R. 1999. Linking insecticide con-tamination and population response in an agri-cultural stream. Environ Toxicol Chem18:1948–1955.

Liess M, Schulz R, Liess MHD, Rother B,Kreuzig R. 1999. Determination of insecticidecontamination in agricultural headwaterstreams. Water Res 33:239–247.

Liess M, Von der Ohe P. 2005. Analyzing effectsof pesticides on invertebrate communities instreams. Environ Toxicol Chem 24:954–965.

Mather JG, Christensen OM. 1998. Behaviour-al aspects of the ‘New Zealand flatworm’, Artio-posthia triangulata, in relation to speciesspread. Pedobiologia 42:520–531.

Matthews RA, Landis WG, Matthews GB.1996. The community conditioning hypothesisand its application to environmental toxicology.Environ Toxicol Chem 15:597–603.

McKelvey RA, Wright JP, Honegger JL. 2002.A comparison of crop and non-crop plants assensitive indicator species for regulatory test-ing. Pest Manag Sci 58:1161–1174.

Migula P. 2000. Validity and links between enzy-matic effects and animal population demogra-phy in stressed environments. In: KammengaJLR, editor. Demography in ecotoxicology. Lon-don, New York: Wiley & Sons. p 219–240.

Mineau P. 2002. Estimating the probability of birdmortality from pesticide sprays on the basis ofthe field study record. Environ Toxicol Chem21:1497–1506.

Muller K, Bach M, Hartmann H, Spiteller M,Frede HG. 2002. Point- and non point-sourcepesticide contamination in the Zwester Ohmcatchment, Germany. J Environ Qual31:309–318.

Newton I. 1986. The sparrowhawk. Calton, UK: T.& A.D. Poyser. 396 p.

Newton I, Wyllie I. 1992. Recovery of a spar-rowhawk population in relation to decliningpesticide contamination. J Appl Ecol29:476–484.

Nijboer R. 2000. Natural communities of Dutchinland aquatic ecosystems. Wageningen, TheNetherlands. Report nr EC-LNV nr AS-06.

Nijboer RC, Johnson RK, Verdonschot PFM,Sommerhauser M, Buffagni A. 2004. Estab-lishing reference conditions for Europeanstreams. Hydrobiologia 516:91–105.

Petersen BS. 1996. The distribution of birds inDanish farmland. An analysis of distributionand population densities of 14 farmland speciesin relation to habitat, crop and pesticide use.Ministry of Environment and Energy, DanishEnvironmental Protection Agency. Report nr 17.

7 · re f e re n c e s l – r

58

pa rt b

CONTENTS

Petersen BS, Jacobsen EM. 1997. Populationtrends in Danish farmland birds. A modelling ofpopulation changes 1976–1996 with special ref-erence to the effects of pesticide use. Ministryof Environment and Energy, Danish Environ-mental Protection Agency. Report nr 34.

Ponge JF, Gillet S, Dubs F, Fedoroff E, HaeseL, Sousa JP, Lavelle P. 2003. Collembolancommunities as bioindicators of land use inten-sification. Soil Biol Biochem 35:813–826.

Potts GR. 1986. The partridge: pesticides, preda-tion and conservation. London: Collins. 274 p.

Rajapakse N, Silva E, Kortenkamp A. 2002.Combining xenoestrogens at levels below indi-vidual No-observed-effect concentrations dra-matically enhances steroid hormone action.Environ Health Perspect 110:917 – 921.

Rodgers JRJH, Dunn A. 1992. Developing designguidelines for constructed wetlands to removepesticides from agricultural runoff. EcologicEng 1:83–95.

Römbke J, Breure AM. 2005. The ecological clas-sification and assessment of soils. Ecotoxicol En-viron Saf. Forthcoming.

Ruf A, Beck L, Dreher P, Hund-Rinke K, Rom-bke J, Spelda J. 2003. A biological classifica-tion concept for the assessment of soil quality:“biological soil classification scheme” (BBSK).Agric Ecosys Environ 98:263–271.

Scheffer M. 1998. Ecology of shallow lakes. Lon-don (UK): Chapman & Hall. 357 p.

Schouten AJ, Breure AM, Mulder C, RutgersM. 2004. Nematode diversity in Dutch soils,from Rio to a biological indicator for soil quali-ty. In: Cook RC, Hunt DJ, editors. Nematologymonographs and perspectives. Volume 2. Lei-den, The Netherlands: Brill Academic Publish-ers. p 469–482.

Schulz R. 2004. Field studies on exposure, ef-fects, and risk mitigation of aquatic non-pointsource insecticide pollution: A review. JEnviron Qual 33:419–448.

Schulz R, Liess M. 1999a. Validity and ecologicalrelevance of an active in situ bioassay usingGammarus pulex and Limnephilus lunatus. Envi-ron Toxicol Chem 18:2243–2250.

Schulz R, Liess M. 1999b. A field study of the ef-fects of agriculturally derived insecticide inputon stream macroinvertebrate dynamics. AquatToxicol 46:155–176.

Schulz R, Peall SKC, Dabrowski JM,Reinecke AJ. 2001. Current-use insecticides,phosphates and suspended solids in theLourens River, Western Cape, during the firstrainfall event of the wet season. Water S.A.27:65–70.

Schulz R, Thiere G, Dabrowski JM. 2002. Acombined microcosm and field approach toevaluate the aquatic toxicity of azinphosmethylto stream communities. Environ Toxicol Chem21:2172–2178.

Scott-Fordsmand JJ, Weeks JM. 2000. Bio-markers in earthworms. Rev Environ ContamToxicol 165:117–159.

Sherratt TN, Roberts G, Williams P, Whit-field M, Biggs J, Shillabeer N, Maund SJ.1999. A life-history approach to predicting therecovery of aquatic invertebrate populationsafter exposure to xenobiotic chemicals. EnvironToxicol Chem 18: 2512–2518

Sibly RM, Newton I, Walker CH. 2000. Effectsof dieldrin on population growth rates of spar-rowhawks 1963–1986. J Appl Ecol 37:540–546.

Sönnichsen H. 2002. Untersuchungen über dieBedeutung von Flora und Fauna der Parzellen-gräben in ackerbaulich (landwirtschaftlich)genutzten Flächen auf der Insel Nordstrand.Braunschweig.

Sotherton NW, Holland JM. 2002. Indirect ef-fects of pesticides on farmland wildlife. In:Hoffman DJ RB, Burton GA, Cairns J, editors.Handbook of ecotoxicology. Boca Raton (FL):Lewis. p 1173–1195.

Spurgeon DJ, Sandifer RD, Hopkin SP. 1996.The use of macro-invertebrates for populationand community monitoring of metal contami-nation – Indicator taxa, effect parameters andthe need for a soil invertebrate prediction andclassification scheme (SIVPACS). BioindicatorSystems for Soil Pollution 10:96–110.

Tarrant KA, Field SA, Langton SD, HartADM. 1997. Effects on earthworm populationsof reducing pesticide use in arable crop rota-tions. Soil Biol Biochem 29:657–661.

Thiere G, Schulz R. 2004. Runoff-related agri-cultural impact in relation to macroinverte-brate communities of the Lourens River, SouthAfrica. Water Res 38:3092–3102.

Thompson HM. 2003. Behavioural effects of pes-ticides in bees – Their potential for use in riskassessment. Ecotoxicology 12:317–330.

w o r ks h o p sy n t h e s i s 7 · re f e re n c e s r – w

59

CONTENTS

van Straalen NM, van Rijn JP. 1998. Ecotoxico-logical risk assessment of soil fauna recoveryfrom pesticide application. Rev Environ ContamToxicol 154:83–141.

Vogt H. 1994: Side-effects of pesticides on benefi-cial organisms: Comparison of laboratory,semi-field and field results. IOBC/WPRS Bull17. 178 p.

Vogt H. 2000. Sensitivity of non-target arthropodspecies to plant protection products accordingto laboratory results of the IOBC WG Pesticidesand Beneficial Organisms IOBC/WPRS Bull23:3–15

Von der Ohe PC, Liess M. 2004. Relative sensi-tivity distribution of aquatic invertebrates to or-ganic and metal compounds. Environ ToxicolChem 23:150–156

Walter H, Consolaro F, Gramatica P,Scholze M, Altenburger R. 2002. Mixturetoxicity of priority pollutants at no observed ef-fect concentrations (NOECs). Ecotoxicology11:299–310.

Wick M, Freier B. 2000. Long-term effects of aninsecticide application on non-target arthro-pods in winter wheat – a field study over 2 sea-sons. J Pest Sci 73:61–69.

Wiles JA, Frampton GK. 1996. A field bioassayapproach to assess the toxicity of insecticideresidues on soil to Collembola. Pest Sci47:273–285.

Wiles JA, Jepson PC. 1994. Sublethal effects ofdeltamethrin residues on the within-crop be-havior and distribution of Coccinella-Septem-punctata. Entomologia Experimentalis et Appli-cata 72:33–45.

Wogram J, Liess M. 2001. Rank ordering ofmacroinvertebrate species sensitivity to toxiccompounds by comparison with that of Daph-nia magna. Bull Environ Contam Toxicol67:360–367.

7 · re f e re n c e s w

60

pa rt b

CONTENTS


appendix aABSTRACTS OFPLATFORMPRESENTATIONS

CONTENTS

CONTENTS

stressand disturbance

in nature

Sibly, R

School of Animal and Microbial Sciences,University of Reading, UK

In this paper, I use the terms “stressors” to meanenvironmental factors that decrease populationgrowth rate when first applied (Sibly and Calow1989; Sibly and Hone 2002) and “disturbances” tomean short-acting stresses. Examples of naturaldisturbances include fires, floods, avalanches, hur-ricanes, grazing, and trampling. Natural stressorsinclude pH, food availability, moisture levels, andother climatic variables.

All natural populations are subject to stress anddisturbance. However, documentation of the ef-fects of stress and disturbance in nature is far fromeasy. The difficulties arise in part because oftenthe occurrence of stress and disturbance cannot bepredicted in advance, and thus research methodsto study them cannot be planned and put in placeas needed. In addition, ascription of a particularchange in population size to a particular stressor isextremely difficult. In general, identification ofstressors requires lengthy, replicated experiments.Even then, identification of the cause of popula-tion change in particular cases often remains con-troversial.

The study of stress is generally easiest at the in-dividual level. Populations are in general a littleharder because the effects of interactions betweenindividuals also have to be taken into account.These may produce effects such as density depen-dence, i.e., decline in population growth rate asthe population increases. Communities introducemany further complexities as well (Van Straalen2003). In this paper, I shall deal mainly with popu-lation responses, although towards the end I shallconsider how levels of disturbance affect commu-nities. The response of communities to disturbanceis dealt with in this volume in the paper by Ratte.

There are a number of examples of the effectsof stress on population growth rate at low popula-tion density. Classic studies include those ofTilman et al. (1981), showing that decreasingavailability of silicates reduces population growthrate in the diatom Asterionella formosa, and ofDaniels and Allan (1981), showing similarly thestressful effects of dieldrin on the water flea Daph-nia pulex. However, in many ways the most im-pressive of such studies is one of the earliest, thatof Birch (1953). Birch not only demonstrated thestressful effects of temperature and moisture levelson 2 species of grain beetle but also produced aclassic diagram indicating the regions of “nichespace” in which each species could prosper (i.e.,increase). And he related this conception of the“ecological niche” to the geographic regions ofAustralia in which the 2 species occur; as predict-ed, the warm-adapted species occurred in thenorth and the cold-adapted in the south. This im-portant approach has, I think, received too littleattention in recent years, though there are exam-ples that reproduce some of its features. Ranta(1979), for instance, delineated the ecologicalniches of 3 species of water flea, and Ellenberg(1988) the ecological niches of the northern Euro-pean trees.

How fast a population can recover from a peri-od of temporary stress depends on the species’ lifehistory and body size; generally, small animals re-cover faster than large ones. Thus, for instance,the maximum population growth rate for largemammals and man is about 6% per year, whereassmall rodents can increase 20-fold per year and in-sects more than 100-fold. The relationships be-tween maximum population growth rate and bodysize and generation time are documented, for in-stance, in Campbell (1996) and in Caughley andSinclair (1994). Periods of population increase arenecessarily of limited duration, and in the absenceof further disturbance, population abundancemust eventually reach equilibrium, representingthe carrying capacity of the environment.

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s p l e n a r y k e y n ote

63

Plenary keynote

CONTENTS

6363

The effect of chronic stress is generally to re-duce carrying capacity. Such effects are seen par-ticularly clearly in plant species, which cannotmove to avoid stressors; thus, plant distributionsmay be limited by moisture levels, pH, wind, andso on (Ellenberg 1988). There are few examples ofthe effects of stressors on the carrying capacitiesof mammals and birds. Some examples in whichpredators (foxes) reduce the carrying capacity ofthe environments of several marsupial species arepresented by Sinclair and Krebs (2002). An exam-ple in which dieldrin affected the carrying capaci-ty of sparrowhawks is considered by Sibly et al.(2000).

The effect of chronic stress is, however, not al-ways a reduction in carrying capacity. There areexamples in nature and in microcosm experimentsin which populations compensate for the effects ofstress (Forbes et al. 2001). This may happen if thestressors reduce the competition for food so thatsurvivors do better than they would otherwisehave done. In both the sparrowhawk and the pere-grine, for instance, the use of DDT in the yearsafter 1945 had an immediate effect on eggshellthickness and a demonstrable effect on fledgingsuccess in the sparrowhawk (Newton 1986; Rat-cliffe 1993). Population sizes, however, did not de-cline at this time. Indeed, declines were not seenuntil after the introduction of the cyclodiene pesti-cides 10 years later. Because sparrowhawk andperegrine breed at ages 1 to 2 and 2 years, respec-tively, and have relatively short life expectancies(about 1.3 and 3.5 years, respectively), one has toconclude that DDT did not reduce adult popula-tion sizes in the UK, even though it did reduce ju-venile survival. Perhaps the smaller numbers offledglings were individually more successful in ob-taining food because they had fewer competitors.In this way, the population could compensate forthe effects of DDT reducing juvenile survival.

This study of the effects of DDT shows an inter-action between stress and density. At low densitythe effects of stress would be a reduction in popu-lation growth rate, but in larger populations,around carrying capacity the population is able tocompensate for the effects of stress. A number ofmicrocosm studies have recently shown similar ef-fects (Forbes et al. 2001). Thus, Liess (2002) hasshown that at low but not high population density,fenvalerate reduces emergence success in the cad-dis fly Limnephilus lunatus, and Forbes et al.(2003) obtained similar effects on population

growth rate exposing the larvae of the marinepolychaete worm Capitella sp. I to fluoranthene.

In considering the effects of disturbance in na-ture, it is proper also to bear in mind the importantidea that biodiversity may be highest at intermedi-ate levels of disturbance (Connell 1978). Connellsuggested that at low levels of disturbance, diver-sity is reduced because many species may be out-competed by a few dominant species; thus, with-out disturbance, the beech Fagus sylvatica woulddominate the forests of much of central Europe. Atthe other extreme, where there is too much distur-bance, forests cannot persist. These disturbancescan result, for example, from fire, herbivory, tram-pling, or floods. In general, therefore, we expect tofind highest biodiversity at intermediate levels ofdisturbance.

In conclusion, I would like to endorse the pro-posal of Nico Van Straalen (2003) that ecotoxicolo-gy is best viewed in relation to stress ecology. I seea number of advantages to this idea. In the firstplace, it will broaden the scope of our studies andincrease scientific interest in them. There will be aconsequent increase in the scientific status of thediscipline. For too long, there has been a rift be-tween “pure” and “applied” ecology, and this rifthas wasted opportunities for scientific progress. Inparticular, ecotoxicology has an immense amountto offer in monitoring populations in the field andin linking predictions from studying microcosmsand mesocosms to what happens in the field.

There is, however, much more that we shoulddo. Using existing knowledge of ecological niches,we should develop landscape maps that will chartthe distribution of key species throughout Europe.These maps will allow us to identify the risks ofpesticide exposure. They will also make a hugeand fundamental contribution to ecology. Second,we should initiate general ecological studies ofhow populations and communities recover fromdisturbances. The immediate effects of short-termexposure to pesticides are reductions in abun-dance. Subsequent recovery may often be inde-pendent of the identity of the pesticide, dependinginstead on (1) the species composition and life-his-tory characteristics of the survivors, (2) the mini-mum population sizes to which the populationsare reduced, and (3) the nearby habitats and com-munities from which immigration may occur.

Stress and disturbance are core processes with-in ecology that merit far more attention than theyhave so far received. Many of the studies have

p l e n a r y k e y n ote

64

a p p e n d i x a

CONTENTS

come – and will continue to come – from ecotoxi-cology.

R e f e r e n c e sBirch LC. 1953. Experimental background to the

study of the distribution and abundance of in-sects. I. The influence of temperature, moistureand food on the innate capacity for increase ofthree grain beetles. Ecology 34:698–711.

Campbell NA. 1996. Biology, 4th edn. MenloPark, California: Benjamin Cummings.1206 p.

Caughley G, Sinclair ARE. 1994. Wildlife Ecolo-gy and Management. Cambridge (Massachu-setts): Blackwell Science. 334 p.

Connell JH. 1978. Diversity in tropical rainforestsand coral reefs. Science 199:1302–1310.

Daniels RE, Allan JD. 1981. Life table evaluationof chronic exposure to pesticide. Can. J. Fish.Aquat. Sci. 38:485–494.

Ellenberg H. 1988. Vegetation Ecology of CentralEurope, 4th edn. Cambridge (UK): CambridgeUniversity Press.

Forbes VE, Sibly RM, Calow P. 2001. Toxicantimpacts on density-limited populations: a criti-cal review of theory, practice and results. Eco-logical Applications 11:1249–1257.

Forbes VE, Sibly RM, Linke-Gamenick I. 2003.Joint effects of a toxicant and population densi-ty on population dynamics: an experimentalstudy using capitella sp. I (polychaeta). Ecologi-cal Applications 13:1094–1103.

Liess M. 2002. Population response to toxicants isaltered by intraspecific interaction. EnvironToxicol Chem 21:138–142.

Newton I. 1986. The sparrowhawk. Calton, UK: T& AD Poyser. 396 p.

Ranta E. 1979. Niche of Daphnia species in rockpools. Archives fur Hydobiologie 87:205-223.

Ratcliffe D. 1993. The Peregrine Falcon, 2ndedn. Calton (UK): T & AD Poyser.

Sibly RM, Calow P. 1989. A life-cycle theory ofresponses to stress. Biol. J. Linn. Soc.37:101–116.

Sibly RM, Hone J. 2002. Population growth rateand its determinants: an overview. Phil. Trans.Biol. Sci. 357:1153–1170.

Sibly RM, Newton I, Walker CH. 2000. Effectsof dieldrin on population growth rates of spar-rowhawks 1963–1986. Journal of Applied Ecolo-gy 37:540–546.

Sinclair ARE, Krebs CJ. 2002. Complex numeri-cal responses to top-down and bottom-upprocesses in vertebrate populations. Phil. Trans.Biol. Sci. 357:1221–1232.

Tilman D, Mattson M, Langer S. 1981. Compe-tition and nutrient kinetics along a temperaturegradient: an experimental test of a mechanisticapproach to niche theory. Limnol. Oceanogr.26:1020–1033.

Van Straalen N. 2003. Ecotoxicology becomesstress ecology. Environ Sc Technol 37:324A–330

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s p l e n a r y k e y n ote

65

CONTENTS

linking terrestrialvertebrate risk assessment

to effects in the field

Clook, M.1 and Fletcher, M.2

1) Pesticide Safety Directorate,Department for environment, food and rural affairs, York, UK

2) Central Science Laboratory,Department for environment, food and rural affairs, York, UK

IntroductionCurrently in the European Union (EU), pesticidesare assessed via the EU Directive 91/414/EEC (seeAnonymous 1991). This directive and the associat-ed annexes cover the risk to the operator, con-sumer, and environment. Annex II outlines whatdata are required on the active substance, whileAnnex III indicates the data required for the asso-ciated product. Annex VI, or the Uniform Princi-ples, outlines, among other issues, the decision-making criteria that need to be considered prior toan active substance being placed on Annex I andthe associated product being authorised.

The risk to the environment covers both thefate and behaviour of an active substance (i.e., ex-posure) as well as its possible effects to non-targetorganisms. Non-target organisms consideredunder Directive 91/414/EEC include the following:birds, mammals, aquatic life (including fish,aquatic invertebrates, algae, and aquatic plants),non-target arthropods, honeybees, earthworms,soil macroinvertebrates, soil microbial processes,and terrestrial non-target plants.

The risk assessment carried out for non-targetorganisms currently takes a single point estimateof toxicity as well as exposure. This results in ei-ther a toxicity-exposure ratio (TER) or a hazardquotient (HQ), which is then compared to a regu-latory trigger value in the Uniform Principles ofDirective 91/414/EEC (Council Directive94/43/EC). If the relevant trigger value isbreached, then no authorisation can be granted“unless it is clearly established through an appro-priate risk assessment that under field conditionsno unacceptable impact occurs after use of the

plant protection product according to the pro-posed conditions of use”. This “appropriate risk as-sessment” usually takes the form of further infor-mation on either the toxicity of the compound orthe exposure of non-target organisms to the com-pound.

Outline of bird and mammalrisk assessment

The above is a very general overview regardingecotoxicological regulatory risk assessment. Ofmore relevance to this workshop is how the risk tobirds and mammals is assessed. Outlined below isa brief summary of the current risk assessmentprocedure for birds and mammals.

When the risk to birds or mammals is assessed,the process as outlined in SANCO/4145/2000 is fol-lowed (see Anonymous 2002). For the first-tier as-sessment, the “estimated theoretical exposures”(ETEs) for the acute, short-term, and long-termexposure scenarios are determined. The followingcalculation is used to determine the ETE:

ETE = (FIR/bw) × C × Av × PT × PD(mg/kg bw/d)where:

ETE = estimated theoretical exposure,FIR = food intake rate,bw = bodyweight,C = concentration on treated food,Av = avoidance,PT = proportion of diet obtained from

treated area, andPD = proportion of food type in diet.

Once the ETE is calculated, it is then comparedwith the appropriate toxicity endpoint and a TERis determined. If the resulting TER is less than theappropriate trigger value presented in the UniformPrinciples, then the assessment may be refinedusing the various steps outlined in the document(see Anonymous 2002). For example, specificresidue data can be used in place of the generic in-formation used in the first-tier calculation. Like-wise, information on the proportion of diet ob-

o b s e r va t i o n o f e f f e c t s i n t h e f i e l d

66

a p p e n d i x a

Observationof effects in the field

CONTENTS

tained from the treated areas together with theproportion of food type in the diet can be used toproduce more realistic estimates of exposure. Inaddition, it may be possible to carry out a fieldstudy to address the whole issue regarding the riskto birds and/or mammals. Outlined below is amore detailed consideration of the types of fieldtrials that may be conducted to address concernsregarding the risk to birds and mammals from theuse of pesticides.

Types of field studies that can be usedin regulatory risk assessment

Field trials conducted to address a regulatory re-quirement may take several different designs, andhence, for the purposes of this paper, it is pro-posed that regulatory field trials can fit in to 3 cat-egories: ecological, residue, and effects. These aredefined below.

Ecological field trialsThese are field trials that are designed to pro-vide information on the ecology of the speciesunder consideration. For example, these typesof field studies may entail radio-tracking birdsto determine how long a particular speciesspends in a specific type of field. Therefore,this type of information can be used to refinePT. Likewise, “ecological field trials” can alsoprovide information on the types of food that aparticular species may eat when in the treatedarea. This information can therefore be used torefine PD.

Residue field trialsThese are trials that are carried out to provide amore realistic estimate of the concentration of

pesticide on treated food items. Therefore, thisinformation can be used to refine C.

Effects field trialsThese are trials that are designed to detect anyeffects of the pesticide in the field on birdsand/or mammals and hence are used to refinethe whole assessment.

It should be noted that 1 field trial can be designedto cover more than one of the above issues.

It is assumed that of primary interest to this work-shop is the latter type of field trial, that is, the ef-fects field trial. There is no standardised protocolfor this type of study, and hence each is designedon a case-by-case basis to answer specific ques-tions that have arisen during the risk assessmentprocess. However, when a field study is designed,several issues need to be considered. Of primaryimportance is what it is that the trial is trying toaddress. The trial should therefore be designed toaddress the specific question that has arisen fromthe regulatory risk assessment. For example, whatis the level of mortality of small mammals follow-ing the use of a certain seed treatment?

Outlined in Table 1 are some of the other issuesthat need to be considered prior to commencingan effects field trial. Included in Table 1 is a briefexplanation of each issue. In addition to those is-sues highlighted in Table 1, it is assumed that thefield trial will be based on the proposed use andhence be conducted on the appropriate crop and atthe appropriate rate.

How are field trials usedin regulatory risk assessments?

Of the 3 types of field trials outlined above, the useof residue field trials is relatively straightforward

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s o b s e r va t i o n o f e f f e c t s i n t h e f i e l d

67

CONTENTS

Suitability of site The site should be climatically, biologically and geographically representative of where theproduct will be used.

Capture of animals This is potentially useful in demonstrating any population level effects.

Carcase searching This is used to indicate the level of mortality, however there is a need to demonstrate theefficiency of carcase searching. Ideally need to have an indication as to what is considered tobe a lethal level in carcases so that cause of death can be determined.

Chemical analysis of potential fooditems

Potentially useful in demonstrating that exposure has taken place in the field. Informationcan be fed back in to the above equation and hence C revised appropriately.

Faecal analysis Potentially useful in demonstrating exposure to treated food items.

Radiotracking This can provide an indication of the level of mortality as well as provide additionalinformation regarding PT.

Table 1 Issues that need to be considered when designing a field study.(Information on protocol design as well as issues to consider are available in a number of publications including –Greaves et al. (1988), Somerville and Walker (1990) and Anon (1990); there also is a protocol by the U.S. EPA (OPTTS850.2500 – Field testing of terrestrial wildlife).)

in that residue data gathered from appropriate lo-cations can be fed into the above equation andhence C can be revised (see Sections 5.2 and 5.3 ofAnonymous 2002 for further information). Like-wise, information from ecological field trials canbe fed into the above equation to refine PT and PDappropriately (see Section 5.6 of Anonymous[2002] for further information). The output fromboth of these types of field trials is quantitativeand hence can be fed into the risk assessment on aquantitative basis. An illustration of how these fac-tors can be revised in light of more relevant data isprovided in Appendix IV of Anonymous (2002).

Due to its qualitative nature, it is not possible tofactor information from an effects field trial intothe above equation. The following is provided asan example of how the findings from an effectsfield trial may be considered in a regulatory set-ting.

The initial risk assessment indicates that for afictitious insecticide applied to cereals at an earlygrowth stage, a high risk, that is, TERacute is lessthan 1, is predicted. A field trial has been conduct-ed to investigate the level of mortality of geese fol-lowing application of the insecticide. In trying todetermine this level, a site that is frequented bygeese has been chosen, and the protocol has beendesigned to determine likely exposure levels intreated vegetation. It has also been designed to de-termine residues in faeces as well as the likelylevel of mortality. This latter factor is assessed viacasualty searches. During the field trial, no mor-talities were found and no abnormal symptomswere observed. Exposure was confirmed by analy-sis of the cereal shoots as well as faeces analysis.

From a regulatory perspective, what does thisstudy tell us? One thing it does not tell us is the in-secticide’s level of acceptability: this is not a ques-tion that any field trial can answer. Does it providean answer to the original question, that is, what isthe level of mortality of geese following applica-tion of the insecticide? Taken on face value, it doesindicate that, following exposure, no mortality ofgeese occurred. However, it should be noted thatthere are concerns regarding several issues of thetrial. The first issue is the representativeness of thesite: Is this site representative of all sites wheregeese will feed? Was the trial representative interms of the time and hence rate that the geesespent feeding on the field? The residues indicateexposure; however, is this 1 site representative ofall sites that could be treated? Due to the above

questions, effects field trials provide only a quali-tative indication of the potential level of impact.

In conclusion, ecological field trials and effectsfield trials can provide useful information that canbe used on a quantitative basis in refining the riskassessment. However, in considering the overallusefulness of effects field trials in the regulatoryrisk assessment, the key issue is whether they areof sufficient robustness to provide either a deci-sion maker or a policy maker (Anonymous 2000)with sufficient information to make a decision.Scientists alone cannot determine whether this isthe case, and hence the usefulness of effects fieldtrials will depend upon an interaction between sci-entific advisers, decision makers, and policy mak-ers.

Monitoring schemesThere are 2 types of monitoring schemes: proac-tive and reactive. In a proactive monitoringscheme, the regulator requests that some form ofmonitoring be carried out once the product hasbeen approved. This can take the form of “growerquestionnaires”. These simply seek information onwhat happens after a particular product is used.The information is extremely crude and hence ofpotentially limited usefulness in making regulato-ry decisions. However, this information can aiduser-education schemes in ensuring the safe use ofa product. Reactive monitoring is used once an in-cident occurs. One example of this type of moni-toring is the UK Wildlife Incident InvestigationScheme (WIIS), which investigates deaths ofwildlife, including beneficial insects, pets, andsome livestock when there is strong evidence sug-gesting that pesticide poisoning may be involved.

The WIIS provides a unique means of post-reg-istration surveillance of pesticide use, and thusregistrations can be revised if necessary. In addi-tion, it provides a measure of the success of thepesticide registration process and helps in the veri-fication and improvement of the risk assessmentsmade in the registration of compounds. Evidencefrom the WIIS may also be used to enforce legisla-tion on the use of pesticides and the protection offood, the environment, and animals.

Animals that are suspected of being poisonedare submitted for post-mortem examination toeliminate cases of trauma, disease, and starvationand to examine tissues for signs of poisoning. Afield inquiry is carried out to assess the possiblecause of the poisoning, its extent, and what


68

a p p e n d i x a

CONTENTS

species are involved. Using the post-mortem andfield inquiry reports, suitable analyses are carriedout to detect residues of possible compounds. Ifresidues are detected, the significance of these aredetermined and reported to the regulators or en-forcement authorities.

The WIIS relies on the poisoned animal beingfound and recognised as a possible pesticide-poi-soning incident. It must then be reported to theWIIS. Annual reports of findings are published(e.g., Barnett et al. 2002).

By examining the facts of an incident, it is oftenpossible to establish how the incident occurred. In-cidents may arise from one of 3 ways: � through approved use, where the product has

been used according to the specified conditionsfor use;

� through misuse, which occurs from careless oraccidental or wilful failure to adhere to the cor-rect practise of using a product, or

� through abuse, which results from deliberate,illegal attempts to poison animals.

Confirmed approved-use incidents are of thegreatest value to regulators because the product,being used in a correct manner, has resulted in thedeath of wildlife. An assessment can be made ofthe conditions of the incident, and remedial actioncan be taken to prevent incidents under the sameconditions in the future. However, caution shouldbe used in interpreting and hence using these datain the risk assessment. As part of a large researchproject carried out in the UK, the reporting rate ofcasualties was investigated via the use of teleme-try, and it was concluded that when interpretingWIIS data, it should always be remembered thatonly a small proportion of incidents are reportedand that the proportion reported is smaller forsmaller species. Therefore, for large species suchas the woodpigeon, a lack of reported incidentsmay be taken as evidence of a lack of substantialmortality, but the possibility that occasional mor-talities may be occurring cannot be ruled out.While for small species of birds, a lack of reportedpoisoning incidents does not rule out the possibili-ty that substantial mortality may be occurring.

But what about incidents resulting from misuseof pesticides? Are these relevant in the regulatoryrisk assessment? They may be relevant if the inci-dent arose from misleading or unclear instruc-tions.

C o n c l u s i o n sField studies and monitoring programmes can beused to aid the regulatory process. However, cau-tion is required to ensure that the informationthey provide is used appropriately. In determiningthe value of field trials, views of decision makersand policy makers should be sought to ensure thatthe trials are sufficient to address their needs.

R e f e r e n c e sAnonymous. 1990. Recommendations of an inter-

national workshop on terrestrial field testing ofpesticides. In: Somerville L, Walker CH, edi-tors: Pesticide effects on terrestrial wildlife.London, New York, Philadelphia: Taylor &Francis. p 353–393.

Anonymous. 1991. Council Directive concerningthe placing of plant protection products on themarket (91/414/EEC). (Seehttp://europa.eu.int/comm/food/fs/ph_ps/pro/legal/dir91-414-eec_en.pdf)

Anonymous. 2000. Policy, risk and science: se-curing and using scientific advice. Prep by Ox-ford Economic Research Associates (OXERA)Ltd. Contract Research Report for the UKHealth and Safety Executive nr 295/2000 (En-vironment Agency R and D Technical ReportP414.). 119 p.


Barnett EA, Fletcher MR, Hunter K, SharpEA. 2002. Pesticide Poisoning of Animals 2001:Investigations of suspected Incidents in theUnited Kingdom. Department for Environment,Food and Rural Affairs, London (UK), 43 p.

Greaves MP, Greig-Smith PW, Smith BD, edi-tors. 1988. Field method for the study of envi-ronmental effects of pesticides, British CropProtection Council (BCPC) Publications, Mono-graph nr 40, Thornton Heath. 370 p.

Somerville L, Walker CH, editors. 1990. Pesti-cide Effects on Terrestrial Wildlife. London:Taylor & Francis, 404 p.

(USEPA) U.S. Environmental ProtectionAgency. 1996. Ecological Effects Test Guide-lines. OPPTS 850.2500 Field Testing for Terres-trial Wildlife. EPA 712-C-96–144. 41 p.


69

CONTENTS

http://europa.eu.int/comm/food/fs/ph_ps/pro/legal/dir91-414-eec_en.pdf

http://europa.eu.int/comm/food/fs/ph_ps/pro/legal/dir91-414-eec_en.pdf

field studiesof pesticide effects

on terrestrialinvertebrates

Holland, J.M., Ewald J.A.and Aebischer N.J.

The Game Conservancy Trust, Fordingbridge, UK

Insecticides can cause toxic and sublethal effectsto non-target invertebrates, leading to a decline intheir abundance, diversity, and range inhabited.Most data exist on the short-term, within-seasonimpact of insecticide applications generated fromfield trials conducted for registration purposes orindependently to quantify the impact on speciesnot previously tested (e.g., Moreby et al. 2001).Data generated for registration are rarely pub-lished, although they could be widely used byagronomist and conservation advisors. Arablefield trials are usually conducted within singlefields using a replicated block approach basedupon 1-ha plots. The data from such trials have tobe treated with a degree of caution, and the fol-lowing reasons were discussed by Brown (1998).Whether an insecticide causes mortality, for exam-ple, is determined by the susceptibility of each or-ganism and the level of exposure, and this is deter-mined by the organism’s activity and the spraycoverage (Wiles and Jepson 1994). There may alsobe sublethal effects that will eventually cause apopulation reduction, but these may not appearduring the monitoring period. The outcome is fur-ther complicated as a consequence of reinvasionby untreated invertebrates from adjacent un-sprayed areas (Duffield and Aebischer 1994), nat-ural population fluctuations in response to otherunrelated factors (Holland 2002), the heteroge-neous distribution of invertebrates (Holland et al.1999), and biased sampling methodologies. In ad-dition, invertebrate abundance is always lower infield centres of conventionally managed arablecrops (Holland et al. 1999) where insecticide trialsare typically conducted, and therefore only a fewspecies will be collected in sufficient numbers forstatistical analysis. Trials conducted along theedge of a cereal field are more likely to encountera greater number and diversity of arthropods, asfound by Moreby et al. (2001).

There is plenty of evidence that the impact offield-scale applications can be highly variable,with considerable variation occurring between

species and higher taxa, between trials conductedusing the same methodology, and between years(Mead-Briggs 1998; Moreby et al. 2001). A typicalresult may find that some non-target taxa are sus-ceptible while others suffer considerable mortality(Figure 1); moreover, there may also be indirect ef-fects that can be equally damaging (Sotherton andHolland 2002). However, the impacts on ecosys-tem functioning and the food chain are most im-portant ecologically, but these impacts are rarelyconsidered in field trials because they are usuallyof insufficient spatial and temporal scale to detectsuch effects. The best evidence of examining pesti-cide effects through the food chain can be found inthe studies of the grey partridge conducted by TheGame Conservancy Trust (Potts 1986). Grey par-tridge populations have declined severely since the1970s, and a reduction in their invertebrate foodsupplies was identified as one of the major con-straints on chick survival. Many of their key inver-tebrate food items are phytophagous, but thesewere substantially reduced by herbicides and in-secticides, resulting in poor chick survival. Whenmore broad-leaf weeds along with their associatedinvertebrates were allowed to survive by selective-ly spraying the outer 6 m of cereal fields and insec-ticides were excluded from this area, the improve-ment in food supplies was sufficient to increasethe survival rate of grey partridge chicks (Sother-ton 1991). Even so, the long-term effects on thewildlife food chain remain poorly quantified, al-

though these effects may be considerably affectingmany species (Campbell et al. 1997).

In addition to a shortage of long-term data,there is also an absence of landscape-scale studies,although pesticides can be applied over extensive


70

a p p e n d i x a

CONTENTS

Figure 1. Variation between taxa in susceptibility to differ-ent insecticides. Reprinted with permission from Morebyet al. 2001. Copyright Society of Environmental Toxicologyand Chemistry (SETAC).

areas. The overall impact of pesticides was exam-ined in an extensive, long-term study conductedby The Game Conservancy Trust. The cerealecosystem was monitored annually over an area of62 km2 in West Sussex from 1970 until present day.Each year the abundance of beneficial insects andweeds was measured in 100 cereal fields duringJune, along with counts of the grey partridge inAugust. Pesticide usage and cropping are alsoknown. All data were incorporated into a geo-graphical information system (GIS). An extensiveanalysis of this database was conducted by Ewaldand Aebischer (1999, 2000), but only some of thekey relevant findings are presented here. Theanalyses included, among others,

1) comparison of pesticide inputs over thestudy area with the national average, and

2) comparison of the abundance of five groupsof invertebrates eaten by farmland birdswitha. average number of herbicide applica-

tions; b. annual proportion of area treated with

fungicides and insecticides; c. different groups of herbicides, fungi-

cides, and insecticides; and d. pesticide use in the previous year.

Cropping and year were taken into account in theanalyses.

The trends in pesticide use were similar to thenational average, with an increase in the areastreated with herbicides, fungicides, and insecti-cides, although these have reached a plateau inthe last decade. The efficacy of herbicides has alsoincreased, and more weed species are being classi-fied as susceptible to the herbicide regime. The in-crease in autumn-sown cereals also led to a corre-sponding increase in autumn-applied herbicidesand insecticides. As expected, the occurrence andabundance of broad-leaved weeds in fields treatedwith herbicide was 13% less than in untreatedfields, although there was considerable variationbetween species. Only Stellaria media (chickweed)and Myostis arvensis (field forget-me-not) showeda significant decline in abundance over the studyperiod.

In the area where pesticides applications wereknown, the annual abundance of 3 arthropodgroups (Araneae and Opiliones, Symphyta andLepidoptera, and Chrysomelidae and Curculion-idae) decreased significantly over the study periodand was negatively related with the broad mea-

sure of pesticide intensity (annual number of her-bicide applications, proportion of the study areatreated with fungicides or insecticides). When her-bicide, fungicide, and insecticide use were consid-ered simultaneously and year of study and croptype were controlled for, the pattern was for all 5invertebrate groups (Araneae and Opiliones, Cara-bidae and Elateridae, Chrysomelidae and Cur-culionidae, Symphyta and Lepidoptera, and non-aphid Hemiptera) to have a negative relationshipwith insecticide use, significantly so for the lasttwo groups. All groups except Carabidae and Ela-teridae declined with fungicides, significantly sofor Araneae and Opiliones. There was a consistentpattern of a negative relationship with the use in-secticides in autumn and winter as well as inspring and summer. The abundance of 4 inverte-brate groups declined as the application ofpyrethroids increased, while all 5 groups showed anegative relationship with organophosphorus in-secticides (Figure 2). None showed this relation-

ship with the selective insecticide pirimicarb.Numbers of invertebrates in all 5 groups showed adecrease after insecticide use in the previous year.In general, there was no relationship between anyof the measures of herbicide usage with inverte-brate numbers, but the power of this analysis waslower because herbicide inputs were widespreadbefore the start of the study and remained consis-tent over the study period.

This extensive dataset includes data on the abun-dance of many invertebrate families or specieswithin each field and consequently will allow theirrange across the study area to be investigated. Al-


71

CONTENTS

Figure 2. Comparison of the densities of invertebrategroups in insecticide-treated fields with those in untreat-ed fields, in relation to the type of insecticide used: Sus-sex study 1970-1995. (* P<0.05, ** P<0.01, *** P<0.001; Fromdata of Ewald and Aebischer, 1999).

though the spatial component of this dataset hasyet to be fully explored, it is evident that some ofthe invertebrates have declined in abundancewithin fields or have become absent from some(Holland 2002). Relatively few other studies con-sider long-term effects or scale. To reduce the im-pact of reinvasion, 10-ha plots were located withina 100-ha field and the impact of an insecticide ap-plication was evaluated for 1 year after the appli-cation (Wick and Freier 2000). The insecticide re-duced numbers of invertebrates within the year,but there were no effects 1 year afterwards. Toeliminate the effect of reinvasion, Kennedy et al.(2001) used barriered plots. Insecticides reducednumbers of epigeal invertebrates, but the numbersand diversity were lower while variability washigher within the barriered plots, restricting possi-ble analyses.

Further evidence of the direct and indirect ef-fects of pesticides will be gained from the IndirectEffects of Pesticides Project (1999–2004) fundedby the Pesticide Safety Directorate of the Depart-ment for Environment Food and Rural Affairs(DEFRA) in the UK (Holland et al. 2002; Hart et al.2002). One of the main components was to con-duct a large-scale field experiment to assess therelative importance of food availability on the de-mography of farmland bird populations. Two keydeterminants affecting survival and reproductionwere considered: (1) supplies of seed food foradults in winter, which is affected by the level ofherbicide inputs, and (2) abundance of inverte-brate food for nestlings in summer, which is deter-mined by insecticide and herbicide inputs. To ma-nipulate these food supplies, 4 experimental treat-ments were implemented: (1) extra seed over win-ter, (2) high insecticide inputs during the summer,(3) extra seed over winter and high insecticide in-puts during the summer, and (4) control sites,each on a 1-km2 block of arable farmland. Thesetreatments were repeated at 3 study sites. Dataanalysis is currently underway

In conclusion, the ecological consequences ofpesticide usage can be far-reaching, with the ex-tent of effect being influenced by, for example, thespatial and temporal scale of the applications, thepesticides’ spectrum of activity, and the speciespresent. In addition, there may be interactionswith other components of the farming system, in-cluding cropping and type of tillage. Not onlyshort-term consequences but also those that mayoccur in the long term must be considered.

R e f e r e n c e sBrown K. 1998. The value of field studies with

pesticides and non-target arthropods. The 1998Brighton Conference – Pests and Diseases, 2,575–582.

Campbell LH, Avery MI, Donald P, Evans AD,Green RE, Wilson JD. (1997) A review of theindirect effects of pesticides on birds. JNCC Re-port nr 227.

Duffield SJ, Aebischer NJ. 1994. The effect ofspatial scale of treatment with dimethoate oninvertebrate population recovery in winterwheat. Journal of Applied Ecology 31:263–281.

Ewald JA, Aebischer NJ. 1999. Pesticide use,avian food resources and bird densities in Sus-sex. Peterborough, UK: JNCC. Report nr 296.148 p

Ewald JA, Aebischer NJ. 2000. Trends in pesti-cide use and efficacy during 26 years of chang-ing agriculture in southern England. EnvironMonit Assess 64:493–529.

Hart J, Murray AWA, Milsom TP, Parrot D,Allcock J, Watola GV, Bishop JD, Robert-son PA, Holland JM, Bowyer A, Birkett T,Begbie M. 2002. The abundance of farmlandbirds within arable fields in relation to seeddensity. Aspects of Applied Biology 67: 221–228.

Holland JM. 2002. Carabid beetles: their ecolo-gy, survival and use in agroecosystems. In: Hol-land JM, editor. The agroecology of carabidbeetles. Andover (UK): Intercept. p 1–40.

Holland JM, Southway S, Ewald JA, BirkettT, Begbie M, Hart J, Parrot D, Allcock J.2002. Invertebrate chick food for farmlandbirds: spatial and temporal variation in differ-ent crops. Aspects Appl Biol 67:27–34.

Holland JM, Perry JN, Winder L. 1999. Thewithin-field spatial and temporal distributionof arthropods in winter wheat. Bulletin of Ento-mological Research 89:499–513.

Kennedy PJ, Conrad KF, Perry JN, Powell D,Aegerter J, Todd AD, Walters KFA, Pow-ell W. 2001. Comparison of two field-scale ap-proaches for the study of effects of insecticideson polyphagous predators in cereals. Appl SoilEcol 17:253–266.


72

a p p e n d i x a

CONTENTS

Mead-Briggs M. 1998. The value of large-scalefield trials for determining the effects of pesti-cides on the non-target arthropod fauna of ce-real crops. In: Haskell PT, McEwen P, editors.Ecotoxicology; Pesticides and Beneficial Organ-isms. Dordrecht (The Netherlands): KluwerAcademic Publishers. p182–190.

Moreby SJ, Southway S, Barker A, HollandJM. 2001. A comparison of the effect of newand established insecticides on non-target in-vertebrates of winter wheat fields. Environ Tox-icol Chem 20:2243–2254.

Potts GR. 1986. The partridge: pesticides, preda-tion and conservation. London: Collins. 274 p.

Sotherton NW. 1991. Conservation Headlands: apractical combination of intensive cereal farm-ing and conservation. In: Firbank LG, Carter N,Darbyshire JF, Potts GR, editors. The Ecology ofTemperate Cereal Fields. Oxford (UK): Black-well Scientific Publications, p 373–397.

Sotherton NW, Holland JM. 2002. Indirect ef-fects of pesticides on farmland wildlife. In:Hoffman DJ RB, Burton GA, Cairns J, editors.Handbook of ecotoxicology. Boca Raton (FL):Lewis. p 1173–1195.

Wick M, Freier B. 2000. Long-term effects of aninsecticide application on non-target arthro-pods in winter wheat – a field study over 2 sea-sons. J Pest Sci 73:61–69.

Wiles JA, Jepson PC. 1994. Sublethal effects ofdeltamethrin residues on the within-crop be-havior and distribution of Coccinella-Septem-punctata. Entomologia Experimentalis et Appli-cata 72:33–45.

field studiesof herbicide effects

on terrestrialplants

De Jong, F.M.W.1 and De Snoo, G.R.2

1) National Institute for Public Health and the Environment,Bilthoven, The Netherlands

2) University of Leiden, The Netherlands

In the recent past increasing attention has beengiven to the side effects of herbicides on non-tar-get plants outside the treated area. In Europe, onlyrecently has guidance been published concerningthe inclusion of non-target terrestrial plants in thefirst tier of the registration procedure for pesti-cides. Given this background, higher-tier studies inthe framework of the registration procedure arescarce, and only a few field studies aimed at ef-fects of herbicides on terrestrial non-target plantsare available.

In the present paper the results of 3 types offield studies, conducted in the Netherlands, arepresented: (1) bioassay experiments, (2) field mar-gin studies, and (3) a large-scale field experimentwith natural vegetation and a dose-response de-sign.

The bioassay studies were conducted on an ex-perimental field, using young plants of a dicotyle-donous species (Brassica napus) and a mono-cotyledonous species (Poa annua). The effects ofwind-drift of a number of herbicides with differentmodes of action were studied on growth and sur-vival of the test plants (De Jong and Udo de Haes2001, De Jong 2001). Drift deposition was mea-sured as well. From the results, a deposition effectcorrelation was derived. An example of a sensitivecombination of herbicide and species is shown inFigure 1.


73

CONTENTS

Figure 1. Modelled effect of deposition of herbicides onplant growth (De Jong 2001).

From the model in Figure 1 an EC5 of 0.15% ofthe field dose and an EC50 of 5% of the field dosewere calculated for effects of herbicides on plantgrowth.

In the field margin experiment, the effects ofthe actual herbicide use in arable fields were stud-ied on ditchbank vegetation adjacent to thesprayed fields (De Snoo and Van der Poll 1999). Inthe experiment the effects of drift adjacent tofields with and without unsprayed field margins of3 m and 6 m wide (resulting in, respectively, low-

and high-drift figures) were studied with numberand cover of species as effect parameter.

Figure 2 shows that the number of dicotyledonspecies adjacent to an unsprayed buffer zone ishigher (Wilcoxon, P < 0.001) than the number ofspecies directly adjacent to the parcel (10.2 com-pared to 5.9 species). Drift deposition in ditch-banks adjacent to a weed parcel is 5% to 25% ofthe field dose. When a buffer zone is used, the de-position is diminished to less than 0.1% of the fielddose.

In the large-scale field experiment, short-term(weeks), mid-term (within the season), and long-term (3 years) effects of a broad-spectrum herbi-cide were studied on non-target vegetation onditchbanks and road verges. In this case the vege-tation was directly sprayed with a range ofdosages. Four locations were studied, with a treat-ment range of 0%, 2%, 4%, 16%, 32%, and 64% of

the field dosage and 5 replicates per treatmentlevel.

For the short-term effects, phytotoxic symp-toms were studied. For the mid-term effects within

the season and the long-term effects between dif-ferent years, the parameters biomass, number ofspecies, and community structure were used as ef-fect parameters.

For a discussion of the results, 3 time scales aredistinguished: short-term effects on plant growthand phytotoxic symptoms, effects on the vegeta-tion parameters within the season, and recovery inthe next year.

The results (see Table 1) show that short-term ef-fects are present at low dosages (<5%, bioassaysand field experiment) of the maximum field dose.In the case of a contact herbicide, individual plantsare able to recover from phytotoxic symptoms,even when exposed to the highest dosage (positivecontrol).

Regarding effects on species number within theseason, in one year a significant effect on speciesnumber was found at low dosages at two of the lo-cations, while at the other two locations no effectswere found at all, not even at the highest dose. Inthe second year no effects on species number werefound at any of the locations at any dosage. Fur-ther analyses of these data are being carried out todetermine whether the effects are a false positiveor whether an ecological explanation can be foundfor the large differences between the years.

Concerning recovery, assessed in the spring ofthe next year before application, all parametersare in the range of the control at all dosages.


74

a p p e n d i x a

CONTENTS

Figure 2. Number of species in a ditchbank adjacent toa winter wheat parcel with and without a 3-6 m buffer-zone.

Dosage After first spraying After second spraying2% dosage 7% ( 4–14%) 3% ( 0– 5%)4% dosage 14% (10–31%) 11% ( 5–14%)

16% 51% (43–57%) 48% (27–52%)32% 65% (50–67%) 63% (54–70%)64% 78% (69–85%) 80% (74–87%)

Table 1. Phytotoxic effects of a broad spectrum herbicideon non-target vegetation (years and sites combined)

All Sites 2000 before 2000 after 2001 before 2001 after 2002 beforeP-value > 10 0.002 > 0.10 0.008 > 0.10NOED community > 64% 32% > 64% 32% > 64%

Table 2. Effects of a broad spectrum herbicide on non-target vegetation community.

Table 2 shows the effect on the communitylevel, using principle response curves as themethod of statistical analysis. The effects showthat the NOEDcommunity within the year was32% of the field dose. On the individual sites, how-ever, 16% of the maximum field dose was found asa no-observed-effect dose (NOED) on several occa-sions. In the year after application, no effects werefound.

Table 2. Effects of a broad-spectrum herbicideon non-target vegetation community.

From the results it was concluded that clearshort-term effects on growth and phytotoxic symp-toms can be found at low dosages of herbicides.The distance from a treated plot is several metres.Vegetation effects within the year are generallyfound at higher dosages, meaning that the inclu-sion of unsprayed buffer zones and other drift-re-ducing measures could prevent these kinds of ef-fects. In the field experiment, full recovery wasfound in the next year for all dosages and effectparameters.

R e f e r e n c e sDe Jong FMW. 2001. Terrestrial field trials for

side-effects of pesticides (Thesis). Leiden (TheNetherlands), Leiden University.

De Jong FMW, Udo de Haes HA. 2001. Develop-ment of a field bioassay for the side-effects ofpesticides on vascular plants. Environ Toxicol16:397–407.

De Snoo GR, Van der Poll RJ. 1999. Effect ofherbicide drift on adjacent boundary vegeta-tion. Agric Ecosyst Environ 73:1–6.

landscape-levelpesticide risk assessment

using freshwatermacroinvertebrate

community structurein the orchard agriculture

of altes land, germany

Christoph Schäfers1, Udo Hommen1

Michael Dembinski2, Juan Gonzalez-Valero3

1) Fraunhofer IME, Schmallenberg, Germany2 ) PLANULA, Hamburg, Germany · 3) Syngenta

Objectives:� To develop a method to investigate landscape

level effects of pesticides.� To detect relevant differences in the macroin-

vertebrate community structure.� To determine whether there is a correlation of

differences to the exposure to pesticides.

Approach:Areal approach using multivariate statistics. Inves-tigation of 40 sites per region in spring, summer,and autumn (5 sampling in 1998 until 2000);macroinvertebrate communities and a set of po-tentially relevant influence factors. Pesticide expo-sure was not measured, but calculated. Intention:less exact for single sites, detection of changes rel-evant for the landscape via statistics.

Investigated regions:Altes Land near Hamburg, apple orchards� Ditches with nearly identical morphology� One culture on one side of a ditch, homoge-

neous distance� Nearly identical pesticide applications follow-

ing recommendations of the local orchard re-search institute (timing, product, dosing)� Very homogeneous situation

� Intensive use of pesticides (20 to 40 applica-tions per year)

� Special legal status: exempted from distanceconditions� Very high exposure

Unique situation for an effect study: ideal for theproof of principle

Region of Braunschweig� Heterogeneous ditches, intensive field cultures

� “Normal” situation: application of themethod


75

CONTENTS


76

a p p e n d i x a

CONTENTS

Figure 1: Decision loop to assessing causes for macroinvertebrate community structure differences. Aim: Excluding vari-ability caused by other relevant influences than pesticides to enhance statistical power to detect pesticide effects.

Figure 2: Factors determining differences in the macroinvertebrate community structure in waters of the agriculturallandscape. Factors excluded by the selection of homogeneous sites in the “Altes Land” are striked out.


77

CONTENTS

Figure 3: grouping of sites by community structure viaTWINSPAN

Figure 4: Calculated potential of exposure (by distanceand water volume) resulted in 3 groups of sites. Meadowsites excluded as controls, organic sites excluded as ex-posed.

Figure 5: Principal response curve analysis of the sites(organic sites excluded): Recovery trends during winter

Figure 5: Species number significantly reduced at sites ofhigh exposure potential

Figure 6: number of Red List species comparable at siteswith no or low, but clearly reduced at sites with high ex-posure potential. Occurrence of these species regarded asIndication of the “value” of the communities in the “AltesLand”.

the lourens river,south africa: a case study

for a mediterraneanagricultural catchment

Schulz, R.

Syngenta AG, Ecological Sciences, Jealott’s Hill InternationalResearch Centre, Bracknell, UK

IntroductionThe assessment of pesticides in the field requiresunderstanding of both exposure and the resultingbiological effects in natural surface waters due tonormal farming practises. This contribution focus-es on insecticide exposure and effects in aquaticecosystems. Besides a general description of theavailable literature, results from the Lourens Riverin the Western Cape of South Africa will be pre-sented as a case study for a Mediterranean agricul-tural catchment.

ExposureMore than 60 reports of insecticide-compound de-tection in surface waters due to agricultural non-point-source pollution have been published in theopen literature during the past 20 years, aboutone-third of which have been undertaken in thepast 3.5 years. Recent reports tend to concentrateon specific routes of pesticide entry, such asrunoff, but there are very few studies on spraydrift–borne contamination. Reported insecticideconcentrations are negatively correlated with thecatchment size, and all concentrations >10 µg/L(16 out of 127) were found in smaller-scale catch-ments, <100 km2.

The Lourens River in the Western Cape ofSouth Africa receives non-point-source pesticidecontamination from a 400-ha orchard area. Stormrunoff and spray drift are the two most importantroutes of entry for current-use organophosphateinsecticides. A variety of methods have been usedto quantify transient pesticide contamination, andmeasured concentrations are well in accordancewith predictions using simple spray-drift andrunoff models. The GIS-implemented models wereused to predict exposure concentrations and loadson a catchment level. Storm runoff events duringthe pesticide-application period leads to highshort-term exposure concentrations, and runoffhas been shown to be more important than spraydrift as a route of entry for pesticides into surfacewaters.

Effects Field studies on effects of insecticide contamina-tion often lack appropriate exposure characterisa-tion. About 15 of the 42 effect studies reviewed re-vealed a clear relationship between quantified,non-experimental exposure and observed effectsin situ on abundance, drift, community structure,or dynamics. Azinphos methyl, chlorpyrifos, andendosulfan were frequently detected at levelsabove those reported to reveal effects in the field.However, it is important to note that for almost allof the studies that seem to establish a clear link be-tween exposure and effect, the pesticide concen-trations measured in the field were not highenough to support an explanation of the observedeffects based simply on acute toxicity data.

Microcosm experiments using communitiesfrom the Lourens River were combined with sur-veys on the invertebrate fauna in the field, sug-gesting the potential to link these different ap-proaches in order to identify causal exposure-ef-fect relationships. Invertebrate surveys in the fieldsuggest that transient pesticide levels affect thecommunity composition at exposed sites. Howev-er, at this level of complexity it becomes evidentthat it is very difficult to clearly separate pesticidesfrom other contributing factors such as increasedturbidity during runoff events.

Conclusions and recommendationsConsiderations for future studies should includebut not be restricted to:� regular and effective exposure characterisa-

tion,� measurement and understanding of confound-

ing factors, � combination of endpoints (e.g., in situ, abun-

dance, drift), and� risk mitigation and ecosystem and landscape

features.

R e f e r e n c e sJergentz S, Mugni H, Bonetto C, Schulz R.

2004. Runoff-related endosulfan contamina-tion and aquatic macroinvertebrate response inrural basins near Buenos Aires, Argentina. ArchEnviron Contam Toxicol 46:345–352.

Schulz R. 2004. Using a freshwater amphipod insitu bioassay as a sensitive tool to detect pesti-cide effects in the field. Environ Toxicol Chem22:1172–1176.


78

a p p e n d i x a

CONTENTS

Schulz R. 2004. Field studies on exposure, ef-fects, and risk mitigation of aquatic non-pointsource insecticide pollution: A review. JEnviron Qual 33:419–448.

Schulz R, Thiere G, Dabrowski JM. 2002. Acombined microcosm and field approach toevaluate the aquatic toxicity of azinphosmethylto stream communities. Environ Toxicol Chem21:2172–2178.

Schulz R, Liess M. 1999. A field study of the ef-fects of agriculturally derived insecticide inputon stream macroinvertebrate dynamics. AquatToxicol 46:155–176.

Schulz R, Liess M. 1999. Validity and ecologicalrelevance of an active in situ bioassay usingGammarus pulex and Limnephilus lunatus. Envi-ron Toxicol Chem 18:2243–2250.

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s

7979


CONTENTS

exposure monitoringfor aquatic

risk assessment

Jenny Kreuger

Swedish University of Agricultural Sciences, Div. Water QualityManagement, P.O. Box 7072, SE-750 07 Uppsala, Sweden

Monitoring of pesticide exposure in the aquaticenvironment has been performed for various rea-sons, resulting in a multitude of different monitor-ing strategies. Commonly, monitoring is done tocomply with regulatory requirements such as com-pliance with drinking water standards or with dif-ferent conventions (e.g., HELCOM, PARCOM, theWater Framework Directive) or due to post-regis-tration requirements laid down by the regulatoryauthorities as a condition for approval. Moreover,there have been research-based monitoring pro-grammes to develop scientific understanding ofpesticide fate in the environment, preferably alsousing the results to calibrate and validate exposuremodels. Monitoring can also be used as a tool tofollow-up on policy decisions, (e.g., checking riskmitigation programmes). Furthermore, there aresome examples of exposure monitoring for risk as-sessment also considering ecological endpoints.However, most studies are short term (1–3 years),thus resulting in a general lack of long-term moni-toring studies.

In a report on the presence of residues and theimpact of plant protection products in the EU(SSLRC 1997), data from field monitoring studieswere compiled and analysed. The overall conclu-sion was that monitoring and data collection with-in Europe were uncoordinated, preventing system-atic interpretation of information with respect todetermining the presence and impact of pesticides.Monitoring was usually not targeted for location,timing, or a specific active ingredient with respectto effects on non-target organisms. Some generalconclusions were as follows.� Certain monitoring data were classified as con-

fidential or presented only in a summary.

� Older, no longer registered persistent pesticidesoften used key resources.

� There were varying study designs, i.e., littleconsistency in scale of study, sampling site se-lection, timing of sample collection, target pes-ticides (other than the triazines), and detectionlimits.

� There was a lack of usage data. � There was no link between chemical and bio-

logical monitoring. � Environmental quality standards (EQS) were

derived for only a limited number of pesticidesand varied often by several orders of magni-tude between countries.

Nevertheless, monitoring results have demonstrat-ed a widespread presence of pesticides in streamsand rivers throughout Europe. Public concern isfocused on possible negative impacts of pesticideson aquatic life or on human health. Proper estima-tion of any hazard that pesticides may pose is de-pendent upon knowledge of both exposure andtoxicity. For adequate exposure assessment as partof a risk evaluation, good-quality data are neededon pesticide exposure patterns and characteristics.The ecological effects of pesticides on flora andfauna in surface waters are dependent on bothpeak concentrations and the duration of the expo-sure.

Transport of pesticides from cultivated fields tosurrounding surface waters generally occursthrough runoff or drainage and is induced by rainor irrigation. However, wind drift from nearbyspraying applications as well as incautious actionsduring the handling of pesticides are also sourcesfor pesticides entering stream waters. Pesticideconcentrations in streams and rivers leaving largercatchments are generally much lower than thosein runoff water from edge-of-field sites. Neverthe-less, transported amounts in percentage of that ap-plied are often independent of the size of the studyarea. In most monitoring studies there has been alack of site-specific data relating to pesticide oc-currence in the water with ongoing activities inthe catchment area.

a b i l i t y to p re d i c t e f f e c t s : l i n k i n g t i e r s a n d e x t ra p o l a t i o n

80

a p p e n d i x a

Ability to predict effects:linking tiers and extrapolation

CONTENTS

The objective of many monitoring investiga-tions has been to determine whether concentra-tions of pesticides could be detected in surface wa-ters on single occasions. This target is insufficientto assess the ecological risks posed by pesticides insurface waters. To constitute a basis for exposureassessment, monitoring in the future should im-prove the sampling and analytical strategies to in-crease evaluation possibilities. Also, when regula-tory pollution-control measurements are deter-mined, transport calculations are useful in evalu-ating possible changes. Minimum backgrounddata for adequate evaluation should include thefollowing: catchment size, land use pattern, soiltype, precipitation, water flow rate, amount andtype of pesticides used, and spraying season. Forbetter recommendations to the users on how tominimise losses of pesticides to the water bodies,there is also a strong need to increase our knowl-edge of the different transport pathways within acatchment, including all possible processes (spills,runoff, leaching, wind drift, etc.).

Swedish monitoring experiencesA pesticide monitoring study was initiated in thespring of 1990 to examine the loss of pesticidesfrom an agricultural catchment in southern Swe-den under normal management practises (Kreuger1998). The catchment has an area of 9 km2 consist-ing of 95% arable land. Information on pesticideusage and handling within this area was collectedannually from the farmers. About 35 different sub-stances were used annually, and 85% to 95% (byweight) of these were included in the analyses. Atthe outlet of the watershed, an automatic watersampler collected, on a weekly basis, time-inte-grated water samples from May to November.Also, at different sites within the catchment, sam-ples were collected on occasion to assess pointsources. The overall objectives were to monitor oc-currence and long-term trends of pesticideresidues in stream water, to explore the reasonsfor pesticide contamination, and (from 1995) tominimise losses of pesticides to water by imple-menting risk mitigation practises.

Results have demonstrated a 90% reduction inpesticide concentrations since the onset of mitiga-tion measurements, although applied amounts inthe area have remained quite constant (Figure 1).Decreasing levels can primarily be attributed to anincreased awareness among farmers on betterhandling and application routines. During recent

years, correct handling and application procedureshave been developed further as integrated parts ofprogrammes by authorities and in trade agree-ments (i.e., the industry imposing conditions onfarmers in exchange for buying the sugar beet har-vest), giving growers economic incentive to min-imise environmental risks when pesticides areused.

Since 2002, the Swedish national monitoringprogramme for pesticides in the aquatic environ-ment comprises:� 4 small agricultural catchments (8–17 km2), in-

cluding stream water with continuous samplingfrom May to November; shallow ground water(3–6 m deep); and information on pesticide useat a field level, and

� 2 rivers (90 km2 and 500 km2).The analysis includes ca. 70 different pesticides,including some metabolites.

Results demonstrate that even in rather smallcatchments (8–17 km2), a large number of pesti-cides were applied (26 to 54 pesticides per catch-ment) and that many of these also were present instream water leaving these catchments. In addi-tion, a number of pesticides with no registered usewere detected, presumably originating from previ-ous applications or possibly from unreportedusage. The presence of a multitude of differentpesticides (sometimes more than 20 pesticides in asingle sample) has to be considered when poten-tial effects of pesticides on non-target organismsare evaluated.

Future challenges� The use of chemical vs. biological monitoring

(a combination would be preferable). Consid-erations include how to cope with a multitudeof pesticides occurring and how to select rele-vant control sites.

� Development of relevant, well-defined qualitycriteria for the aquatic environment.

� Development of guidelines on how to applythese criteria. For example, should the criteriabe different for different ecosystems or re-gions? Are they applicable only to data fromwell-defined monitoring programmes? Shouldwe use single values, a number of consecutivevalues, or an average over a certain time peri-od? What should the minimum size of thecatchment or edge-of-field site be?

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s a b i l i t y to p re d i c t e f f e c t s : l i n k i n g t i e r s a n d e x t ra p o l a t i o n

81

CONTENTS

R e f e r e n c e sKreuger J. 1998. Pesticides in stream water with-

in an agricultural catchment in southern Swe-den, 1990–1996. Sci Total Environ 216:227–251.

SSLRC. 1997. Further Analysis of Presence ofResidues and Impact of Plant Protection Prod-ucts in the EU. Report for the Commission ofEuropean Communities & the Dutch Ministryfor the Environment within the project “Sus-tainable Use of Plant Protection Products(PPPs)”. http://europa.eu.int/comm/environment/ppps/history.htm

exposure modellingfor aquatic

risk assessment

Brown, C.

Cranfield University, Cranfield Centre fro EcoChemistry, Silsoe, UK

Models are used extensively to predict pesticideexposure for regulatory risk assessment. The mod-els are used to estimate field behaviour on thebasis of mainly laboratory measurements and offerseveral advantages over direct determination ofexposure in the field. Modelling is faster andcheaper, it is possible to assess fate and exposureunder the full range of possible use conditions,and the analysis can be repeated to account forchanges in use or impact of mitigation measures.Nevertheless, it is critical to establish the extent towhich model predictions for environmental con-centrations of pesticides match measurements inthe field. Several studies have addressed this ques-tion in the EU and the US. In assessing the implica-tions for risk assessment, it is also necessary toconsider a number of generic issues related to theuse of models and evaluation against field data:� Variability: Spatial variability associated with

fate processes and exposure profiles is general-ly ignored in modelling for risk assessment,which has almost always been deterministic.

� Scale: Exposure modelling typically considersthe field as the unit of assessment with timespans of days to weeks. Predictability of thesystem generally improves for measures that in-tegrate across larger spatial and temporalscales (e.g., annual loadings within smallcatchments).

� Data availability: Models provide continuousoutput that is difficult to evaluate against mon-itoring information that may comprise just afew measurements in space and/or time.

� User subjectivity: Studies have shown that deci-sions made by the user result in uncertaintiesin the model output (Brown et al. 1996;Boesten 2000). Recent developments byFOCUS reduce but do not eliminate subjectivityin modelling.

Simulation of in-field concentrations in soil shouldbe relatively accurate because there is a primaryinput to the system, although studies have shownsignificant variability in deposition across thetreated area. Beulke et al. (2000) evaluated amodel of soil degradation based on the equations


82

a p p e n d i x a

CONTENTS

Figure 1. Time-weighted mean concentration for the sumof pesticides in stream water leaving the catchment dur-ing May-September 1992-2002 (bars) and applied amountof pesticides included in the analytical procedures duringcorresponding years (line).

http://europa.eu.int/comm/environment/ppps/history.htm

http://europa.eu.int/comm/environment/ppps/history.htm

of Walker (1974) against published data from 178studies. Laboratory half-lives were corrected on adaily basis for soil temperature and moisture con-tent, and the simulated residue for each study wasoutput at the time so that 50% degradation wasobserved in the field. Most simulated residues fellwithin the measured value ±50% and there was amarked tendency for the model to overestimatefield concentrations (Figure 1). Similar work in theUS using the PRZM degradation routines for asmaller dataset concluded that simulated concen-trations across the whole degradation curve can beexpected to be within a factor of 3 of measuredvalues 50% of the time and within a factor of 1080% of the time (Jones and Russell 2001).

FOCUS has recommended that the models tobe used for different fate processes relevant forbasic aquatic risk assessment are spray-drift ta-bles, MACRO (drain flow), PRZM (surface runoff),and TOXSWA (fate in surface water). The timingof pesticide loading to surface water via drain flowand surface runoff is generally well predicted bythe models, and most attention has focused on theability to simulate chemical concentrations andloads. It has been shown that MACRO is better ableto simulate drainage concentrations from loamsand clay loams (accuracy within an order of mag-nitude) than from heavy clay soils, where loadings

are larger and fate is dominated by the highly het-erogeneous process of preferential flow.

The most comprehensive evaluation of simula-tions for pesticide transport via surface runoff anderosion was undertaken by the FIFRA Environ-mental Model Validation Task Force (Jones andRussell 2001). The PRZM model was used to simu-

late data from 8 field trials. It was found that sim-ulations for concentrations and loads of pesticidesin in-field runoff and erosion were generally with-in an order of magnitude of measured values.Model accuracy improved for multievent simula-tions.

By comparison with other exposure models,evaluation of TOXSWA is in its infancy. To date,only fitting of the model to laboratory water-sedi-ment data or fate data from small ponds have beenreported. Ongoing research in the Netherlands,France, and the UK will generate evaluationsagainst field measurements. Evidence from regula-tory simulations and catchment-level water-quali-ty models suggests that fate in water is of lesserimportance in predicting aquatic exposure than isaccurate characterisation of pesticide inputs towater and of advective losses within the waterbody.

Regulatory modelling adopts a simplified ap-proach to predicting exposure to pesticides that isdesigned to provide a protective estimation for usein risk assessment. Scenario selection considers


83

CONTENTS

Figure 1. Simulated pesticide concentration in soil at thetime of measurement of 50% loss in 178 published fieldstudies (Beulke et al., 2000).

Figure 3. Relative accuracy of MACRO leaching simula-tions for different soil types (Brown et al., 1999)

Figure 2. Rank plot of the accuracy of simulations for pes-ticide concentration in soil measured in 8 field studies(Jones & Russell, 2001).

vulnerable situations and exposure estimated onthe basis of maximum and time-weighted averageconcentrations and assumes that full bioavailabili-ty of the environmental residue will be conserva-tive. Given the limitations outlined above and theaccuracy of model simulations for site-specificconditions, the question appears to be not to whatextent the models are able to match real exposurelevels, but whether the exposure estimates basedon deterministic modelling provide a protectiveinput on which to base the risk assessment.

R e f e r e n c e sBeulke S, Dubus IG, Brown CD, Gottesburen

B. 2000. Simulation of pesticide persistence inthe field on the basis of laboratory data – A re-view. J Environ Qual 29:1371–1379.

Boesten JJTI. 2000. Modeller subjectivity in esti-mating pesticide parameters for leaching mod-els using the same laboratory data set. AgricWat Manag 44:389–409.

Brown CD, Baer U, Gunther P, Trevisan M,Walker A. 1996. Ring test with the modelsLEACHP, PRZM-2 and VARLEACH: variabilitybetween model users in prediction of pesticideleaching using a standard data set. Pest Sc47:249–258.

Brown CD, Beulke S, Dubus IG. 1999. Simulat-ing pesticide transport via preferential flow: acurrent perspective. In: del Re AAM, Brown C,Capri E, Errera, G, Evans SP, Trevisan M., edi-tors. Proc. XI Symposium on Pesticide Chem-istry, Human and Environmental exposure toxenobiotics, September 1999, Cremona, Italy, p73–82.

Jones RL, Russell MH. 2001. FIFRA Environ-mental Model Validation Task Force. Final Re-port. American Crop Protection Association.Washington. p 768.

Walker A. 1974. A simulation model for predic-tion of herbicide persistence. J Environ Qual3:396–401.

predictability andacceptability of

effects of pesticidesin freshwater ecosystems

Brock, T.C.M.

Alterra Green World Research, Wageningen University andResearch Centre, Wageningen, The Netherlands

IntroductionEstablishing causal relationships between pesti-cide exposure concentrations and effects in natur-al ecosystems is difficult due to the many intrinsicenvironmental factors that can hinder this process.In conjunction with observational field studies,controlled laboratory studies and experiments inmicrocosms or mesocosms can be used to investi-gate possible causal relationships between (combi-nations of) pesticides and ecological effects. Theecological realism of these controlled experimentsis improved when the species and/or communitiesused in these experiments are representative ofthe species and communities at risk in the field.This also applies when the experimental exposureregime is realistic in terms of normal agriculturalpractise, which may imply the repeated applica-tion of a single compound or the application of arealistic combination. In addition, by incorporat-ing the modulating factors characteristic of thenatural ecosystem at risk in the experimentalstudy design, field monitoring studies and con-trolled experiments complement each other.

Irrespective of whether the effect assessment isbased on an observational study in the field or ona controlled semi-field experiment, it is importantto realise that the properties of aquatic ecosystemsvary in space and time. Consequently, an impor-tant issue in the assessment and management ofecological risks of pesticides in freshwater ecosys-tems is the predictive value of the current tieredrisk assessment procedure. In other words, can thederived critical threshold levels and ecological re-sponses be extrapolated in space and time? Anoth-er burning issue is whether the acceptable concen-trations proposed for individual compounds suffi-ciently protect the aquatic ecosystem when morethan one pesticide or stressor is present. In addi-tion, the “what if” and “so what” questions may beraised when the “ecological significance” of pre-dicted or measured pesticide concentrationsand/or effects in different types of surface waterare evaluated.


84

a p p e n d i x a

CONTENTS

Rules of thumb derived from higher-tiertests with individual compounds

The first part of this paper will focus on the fieldrelevance of effect assessments based on speciessensitivity distributions (SSDs) and micro- andmesocosm experiments. The presented data for in-dividual compounds are based on a literature re-view and studies performed at Alterra (Wagenin-gen UR, The Netherlands) and the University ofSheffield (UK), with special reference to insecti-cides and herbicides.

Overall, the compiled single-species toxicitydatabase reveals that SSDs can be used to predictecological threshold levels for direct toxic effects,at least when the toxic mode of action of the com-pound and the exposure concentrations underfield conditions are known. The studies of Maltbyet al. (2002, 2005), Van den Brink et al. (2002),and Schroer et al. (2004) indicate that:� SSDs and HC5 values for the same compound

may be comparable between different freshwa-ter habitats and geographical regions, at leastwhen based on the sensitive taxonomic groupand taxa characteristics for these habitats andregions.

� Laboratory single-species toxicity data can beused to generate SSDs that are representativeof (semi)-natural assemblages.

� In general, the HC5 derived from short-termarthropod toxicity data, or short-term primaryproducer toxicity data, may be used to estimatethe ecological threshold concentration foracute insecticide and herbicide exposures, re-spectively.

� In the absence of sufficient long-term NOECdata, SSDs based on short-term L(E)C50 datamay be used to derive HC5 values for chronicexposure if an appropriate safety factor is ap-plied (e.g., based on the acute-to-chronic ratioobserved for the sensitive standard test speciesin the basic dataset).

Overall, the compiled micro- and mesocosm data-base (Brock, Lahr et al. 2000; Brock, Van Wijn-gaarden et al. 2000) reveals that: � Ecological threshold levels of the same com-

pound usually differ by less than a factor of 10between different micro- or mesocosm experi-ments, a least when similar exposure regimesare tested.

� The ecological threshold levels for herbicideswith a photosynthesis- or growth-inhibiting

mode of action are generally ≥0.1 TUalgae.1TUalgae = geometric mean EC50 of the mostsensitive standard test alga (TU = toxic unit).

� In insecticide (semi-)field studies the ecologi-cal threshold level is usually higher than(0.01–0.1) × EC50 of the most sensitive stan-dard test species (in most cases 0.01–0.1TUDaphnia).

� The recovery of affected populations dependson dynamics in exposure concentrations in thetest system, on generation time and dispersalproperties of the population affected, and onthe ecological infrastructure of the surround-ings of the locality exposed (e.g., extent of iso-lation).

� At exposure concentrations above the criticalthreshold level for direct toxic effects (usually>0.1–1 TU), the magnitude and type of indirecteffects may differ considerably between differ-ent micro- and mesocosm experiments

� Insight into the types of indirect effects thatcan be expected in freshwater ecosystems canbe obtained by combining results of ecotoxico-logical experiments, knowledge on the struc-ture and functioning of the ecosystem at risk,and food web modelling.

Multistress and pesticidesTo address the issue of multistress in freshwaterecosystems, the results of a complex study in ex-perimental ditches are presented. These experi-mental ditches sufficiently resemble outdoor fieldsituations in the polder landscape of the Nether-lands. This study focussed on the impact of a real-istic exposure to a range of pesticides commonlyused in the cultivation of potatoes in the Nether-lands. The main experimental aims were to pro-vide information on the ecological impact of driftof a realistic package of pesticides into surfacewater and to evaluate the effectiveness of drift-re-duction measures in mitigating risks. The pesti-cides selected and the dosage, frequency, and tim-ing of application were based on normal agricul-tural practise for potato crops. During the growingseason two herbicides (prosulfocarb andmetribuzin), one insecticide (lambda-cy-halothrin), and two fungicides (chlorothalonil andfluazinam) were applied 1, 1, 2, 4, and 8 times, re-spectively. Applications were made at 0.2%, 1%,and 5% of the label-recommended rates. At the0.2% level, no consistent treatment-related effectscould be observed. At the 1 % level, only short-


85

CONTENTS

lived effects occurred. In the 5% treatment, cleareffects on macroinvertebrates were observed,which could mainly be attributed to the insecticidelambda-cyhalothrin. The results of this complexstudy suggest that the current aquatic risk assess-ment procedure in the Netherlands (based on indi-vidual compounds, the Uniform Principles, and adrift emission of 1%) may sufficiently protectaquatic ecosystems stressed by a realistic combina-tion of pesticides (Arts et al. personal communica-tion).

The results of the above-mentioned study are inline with another crop-based microcosm experi-ment performed in Wageningen (Van Wijngaardenet al., accepted). In this microcosm experiment,the treatment regime was based on a realistic ap-plication scenario in tulip cultivation. The fungi-cide fluazinam, the insecticide lambda-cy-halothrin, and the herbicides asulam andmetamitron were applied at concentrations equalto 0%, 0.2%, 0.5%, 2%, and 5% spray-drift emis-sion of label-recommended rates. Again, at the0.2% level no consistent treatment-related effectscould be observed. The 0.5% treatment regime re-sulted in short-term effects. Pronounced effectswere observed at the 2% and 5% treatment levels.In this multistress experiment, and for the end-points measured (community metabolism andspecies composition and abundance), the first-tierrisk assessment procedure for individual com-pounds was adequate for protecting sensitive pop-ulations exposed to a realistic combination of pes-ticides. Spray drift–reducing measures seem to beefficient in protecting aquatic ecosystems in agri-cultural areas.

The “ecological significance”of pesticide stress

Guidance on how to deal with risks of chemicals inthe environment is provided not only in a regulato-ry context but also by concepts based on science,ethics, and aesthetics. To sharpen the discussion,four completely different perceptions to considerecological risks of toxicants in non-target habitatscan be recognised(1) the pollution preventionprinciple, (2) the ecological threshold principle,(3) the recovery principle, and (4) the functionalredundancy principle.

It is concluded that consensus on well-definedprotection criteria, which may differ among typesof ecosystems, is needed for tuning of EU policy(e.g., Uniform Principles, Water Framework).

A c k n o w l e d g m e n t sThis contribution depends highly on the input ofthose mentioned in the references.

R e f e r e n c e sBrock TCM, Lahr J, Van den Brink PJ. 2000.

Ecological risks of pesticides in freshwaterecosystems. Part 1: herbicides. Alterra-Reportnr 088, Wageningen, 124 p

Brock TCM, Van Wijngaarden RPA, VanGeest GJ. 2000. Ecological risks of pesticidesin freshwater ecosystems. Part 2: Insecticides.Alterra-Report nr 089, 131 p

Maltby L, Blake N, Brock TCM, Van denBrink P. 2005. Insecticide species sensitivitydistributions: the importance of test species se-lection and relevance to aquatic ecosystems.Environ. Toxicol. Chem 24:379–388

Maltby L, Brock TCM, Blake N, Van denBrink P. 2002. Addressing interspecific varia-tion in sensitivity and the potential to reducethis source of uncertainty in ecotoxicologicalassessments. Department for Environment,Food, and Rural Affairs (DEFRA), UK. Projectnr PN0932, 22 p.

Schroer AFM, Belgers D, Brock TCM, MaundSJ, Van den Brink PJ. 2004. Acute toxicity ofthe pyrethroid insecticide lambda-cyhalothrinto invertebrates of lentic freshwater ecosys-tems. Arch. Environ. Contam. Toxicol.46:324–335

Van den Brink PJ, Brock TCM, Posthuma L.2002. The value of the species sensitivity distri-bution concept for predicting field effects:(non-)confirmation of the concept using semi-field experiments. In: The use of Species Sensi-tivity Distributions in Ecotoxicology. Suter GW,Traas TP, Posthuma L, editors. Boca Raton(FL): Lewis Publishers. p 155–193.

Van Wijngaarden RPA, Cuppen JGM, ArtsGHP, Crum SJH, van den Hoorn MW, vanden Brink PJ, Brock TCM. 2004. Aquatic riskassessment of a realistic exposure to pesticidesused in bulb crops, a microcosm study. Environ.Toxicol. Chem. 23:1479–1498


86

a p p e n d i x a

CONTENTS

predictingeffects of pesticides

in streams

Matthias Liess

UFZ Centre for Environmental Research, Departmentof Chemical Ecotoxicology, Leipzig, Germany

AbstractOur aim was to find patterns related to the effectsof pesticides in aquatic invertebrate communitycomposition in central European streams . To re-duce the site-specific variation of community de-scriptors due to environmental factors other thanpesticides, species were classified and grouped ac-cording to their vulnerability to pesticides. Theywere classified as species at risk (SPEAR) andspecies not at risk (SPEnotAR). Ecological traitsused to define these groups were (1) sensitivity totoxicants, (2) generation time, (3) migration abili-ty, and (4) presence of aquatic stages during timeof maximum pesticide application. Results showedthat measured pesticide concentrations of 1:10 ofthe acute 48-h median lethal concentration (LC50)of Daphnia magna led to a short- and long-term re-duction of abundance and number of SPEAR and acorresponding increase in SPEnotAR. Concentra-tions of 1:100 of the acute 48-h LC50 of D. magnacorrelated with a long-term change of communitycomposition. However, number and abundance ofSPEAR in disturbed stream sections are greatly in-creased when undisturbed stream sections are pre-sent in upstream reaches. This positive influencecompensated for the negative effect of high con-centrations of pesticides through recolonisation.The results emphasize the importance of consider-ing ecological traits and recolonisation processeson the landscape level for ecotoxicological risk as-sessment.

IntroductionThe EU Uniform Principles for the assessment ofpesticides require that if the preliminary risk char-acterisation indicates potential concerns, registra-tion cannot be granted unless it can be demon-strated that “… under field conditions no unac-ceptable impact on the viability of exposed organ-isms …” occurs. To date, such assessments havebeen made by conducting higher-tier studies,which have included a range of laboratory andsemi-field experiments. Therefore, it is still notclear to what extent pesticides change population

dynamics and community structures in the field.Recently, some studies have quantified pesticideexposure, adverse effects on aquatic life, and re-covery of these invertebrate communities in thefield. Mortality of 6 mayfly species in an Aus-tralian river was linked to endosulfan contamina-tion due to runoff (Leonard et al. 2000). Other in-vestigations also found a link between mortality ofseveral invertebrate species and insecticide con-centrations in streams (Liess and Schulz 1999;Thiere and Schulz 2004). Several invertebratespecies that declined in abundance due to pesti-cides were found to recover within a year (Liessand Schulz 1999). Nevertheless, most existingstudies lack (1) sufficient numbers of investiga-tions in various streams to evaluate the frequencyof potentially harmful events in a specific region,(2) evaluation of long-term effects on invertebratecommunities, and (3) quantification of the recov-ery of impacted communities due to recolonisationfrom undisturbed stream sections. The inclusion ofhabitat quality may put the risks resulting fromcontamination in context with other stressors.

According to these open questions, the aim ofthe present investigation was to find patterns incommunity composition that were related to theeffect of pesticides. As other environmental factorsmay mask possible effects, a new approach thataims at reducing variability in community charac-terisation is presented. A detailed description ofthis work can be found in Liess and Von der Ohe(2005).

Results and DiscussionEnvironmental parameters at investigated sites During the investigation period (1.8 years per site,20 sites total), pesticides were detected in 125runoff events at 18 of the 20 sites. Most of the cont-aminated-runoff events occurred in May, followedby June and July. The four pesticides contributingthe most to the toxic units (TU(D.magna)) wereparathion-ethyl, azoxystrobin, kresoxim-methyl,and ethofumesat in decreasing contribution.Water-quality standard parameters would not beexpected to have deleterious effects on the inverte-brate fauna.

Correlating Environmental parametersand community descriptors The measure for toxic stress of pesticides TU(D.

magna) best described the variance of communitydescriptors related to SPEAR. In general, the num-


87

CONTENTS

ber and abundance of SPEAR correlated negativelywith TU(D. magna). In contrast, the average num-ber and abundance of SPEnotAR did not correlatewith toxic stress. Other parameters contributing tothe variability of SPEAR are length of forestedstream sections, type of substrate, and coveragewith submersed plants.

Temporal changes in community structureThe abundance of SPEAR decreased from April toMay at sites with values of TU(D. magna) exceeding –2to –1 compared with sites where TU(D. magna) valueswere below –4. Furthermore, an increase in abun-dance of SPEnotAR occurred from April to June atsites where TU(D. magna) values were greater than –3to –2 compared with the sites where TU(D. magna),values were below –4.

Contribution of uncontaminatedstream sections to recoveryThe presence of forested stream sections >200 min length and <4000 m upstream of the investigat-ed sites had a strong influence on the interceptand slope of the correlation between TU(D. magna)

and SPEAR in April. When forested stream sec-tions were present, the numbers of SPEAR tendedto be greater. At the same time, the reduction ofSPEAR with increasing TU(D. magna) was greaterthan at sites without forested stream sections (AN-COVA, P < 0.05). However, the positive influence

of forested stream sections upstream of the investi-gated sites compensated for the negative effect ofhigh TU(D. magna) on SPEAR. Indeed, sites with TU> –2 and forested stream sections contained anumber and abundance of SPEAR similar to thoseat sites with TU < –3 without forested stream sec-tions (Figure 1). The described differences of siteswith and without forested stream sections wereapparent only in April—the time period before thehighest toxic units have been measured. The dif-ferences of sites with and without forested streamsections were not detectable in June—the time pe-riod directly after the highest toxic units have beenmeasured.

Finding patterns in community compositionIn the present investigation, we reduced the prob-lem of natural variability by grouping species ac-cording to their sensitivity to pesticides (Von derOhe and Liess 2004) and their life-cycle traitsknown to influence recovery from toxicant stress.The approach of grouping SPEAR has the advan-tage of reducing the variability of the site-specificcommunity characterisation and increasing theability to detect the effect of pesticides on commu-nity composition However, the SPEAR approachalso has disadvantages: because species-level dataare aggregated according to sensitivity and life-cycle traits related to recovery, the effect of a pesti-cide cannot be assigned to any particular speciesor taxon.

Temporal changes in communitystructure: reduction in SPEARSites characterised by high TUs (between –1 and 0based on the 48-h LC50(D. magna)) showed a 75% re-duction of SPEAR from April to May, when thehighest concentrations of pesticides were mea-sured. Other investigations of streams in agricul-tural areas also reported that pesticides from sur-face runoff can cause acute mortality of benthicinvertebrates when they reach the range of the 48-h LC50(D. magna). No indication was found in thepresent investigation that parameters other thanpesticides (e.g., hydrodynamic stress, water-quali-ty parameters) might be responsible for the ob-served short-term reduction of sensitive species. Inagricultural areas, hydrodynamic stress in streamsdue to increased current velocity and suspendedparticles during runoff events can occur frequentlythroughout the year. Hence, this stressor is proba-bly not responsible for the short-term reduction of


88

a p p e n d i x a

CONTENTS

Figure 1: Relation between toxic units(Daphnia magna) and thenumber of species at risk in April (SPEAR(number)-April).Sites are differentiated on the basis of the presence offorested stream sections closer than 4000 m upstream ofthe study site (filled circles; linear regression, r2 = 0.70,p<0.01) or absence of such sites (open circles; linear re-gression, r2 = 0.70, p=0.01). Confidence bands show the95% confidence limit for the respective means. The slopesof the two regression lines differ (analysis of covariance,p<0.05). Reprinted with permission from Liess and Vonder Ohe 2005. Copyright Society of Environmental Toxi-cology and Chemistry (SETAC).

individuals that occurred only during May. A de-tailed discussion of the topic can be found in Liessand Von der Ohe (2005).

Temporal changes in community structure:alteration of SPEnotAR and long-term changesSites characterised by low TUs (below –3, based onthe 48-h LC50(D. magna)) showed a 60% reduction ofSPEnotAR between April and June. This reductionwas not observed at sites where TU values werehigh (above the range of –3 and –2). This patterncould result from an indirect positive effect of pes-ticides on SPEnotAR due to negative effects of pesti-cides on sensitive species. Such negative effectsmight occur even within the range of sublethal ef-fects (TU(D. magna)). However, previous investiga-tions support the idea that short-term exposure toconcentrations that are more than 100 times lowerthan concentrations causing acute mortality cancause long-term effects (Liess 2002).

Factors other than pesticides clearly can influ-ence benthic invertebrate community structure,too, but we found no evidence that this occurredin relation to the endpoint of SPEAR at the sites westudied. However, because the levels of contami-nation may have been insufficiently quantified, itremains uncertain at which concentration thesechanges occur. A detailed discussion of these top-ics can be found in Liess and Von der Ohe (2005).

Contribution of uncontaminatedstream sections to recoveryThe length of forested stream sections upstream ofthe investigated site did relate significantly to thenumber and proportion of SPEAR in April. Thepositive effect of upstream forested stream sec-tions on SPEAR at the downstream sites was notdue to lower concentrations of contaminants atdownstream sites; hence we suggest that the posi-tive effect of forested stream sections on SPEAR atdownstream sites can be attributed to in-streamrecolonisation by invertebrates from the undis-turbed stream sections where diversity is greater.

Cumulative risk For the streams we studied, habitat quality seemedas important as toxicity for community composi-tion. Thus, landscape information increases pre-dictability in the assessment of risk due to pesti-cides. We suggest that the geographical unit of as-sessment should be extended to include landscaperecovery potential.

R e f e r e n c e sLeonard AW, Hyne RV, Lim RP, Pablo F, Van

den Brink PJ. 2000. Riverine endosulfan con-centrations in the Namoi river, Australia: linkto cotton field runoff and macroinvertebratepopulation densities. Environ Toxicol Chem19:1540–1551.

Liess M. 2002. Population response to toxicants isaltered by intraspecific interaction. EnvironToxicol Chem 21:138–142.

Liess M, Schulz R. 1999. Linking insecticide con-tamination and population response in an agri-cultural stream. Environ Toxicol Chem18:1948–1955.

Liess M, Von der Ohe P. 2005. Analyzing effectsof pesticides on invertebrate communities instreams. Environ Toxicol Chem 24:954–965.

Thiere G, Schulz R. 2004. Runoff-related agri-cultural impact in relation to macroinverte-brate communities of the Lourens River, SouthAfrica. Water Res 38:3092–3102.

Von der Ohe PC, Liess M. 2004. Relative sensi-tivity distribution of aquatic invertebrates to or-ganic and metal compounds. Environ ToxicolChem 23:150–156


89

CONTENTS

from laboratory to field:extrapolation of pesticide

risks to non-targetinvertebrate communities

Römbke, J.1 and Frampton, G.2

1) ECT Ökotoxikologie, Floersheim, Germany2) University of Southampton, School of Biology, Southampton, UK

IntroductionIn this contribution an overview is given on thecurrent knowledge of the relationship between thelaboratory and field level when the potential ef-fects of plant protection products (PPPs) on non-target terrestrial invertebrates are tested. In par-ticular, the following extrapolation steps whengoing from laboratory tests to field studies are ad-dressed: � test species versus species in the field, � artificial versus natural substrates, and� acute versus chronic endpoints.According to Directive 91/414/EEC, the risk of pes-ticides to non-target terrestrial invertebrates isroutinely assessed for 3 invertebrate groups,namely, bees, non-target (above-ground) arthro-pods (NTAs), and the below-ground soil fauna. Re-cently, a functional endpoint was added, coveringimpacts on organic matter breakdown at a higher-tier level.

At lower tiers of risk assessment, the bees arerepresented by the honeybee, NTAs by two testspecies (Aphidius rhopalosiphi and Typhlodromuspyri), and the soil fauna mainly by earthworms.Depending upon the lower-tier risk and the usagepattern of the pesticide, additional species may betested at intermediate tiers (e.g., Carabidae forNTAs or Collembola for soil fauna). Standardguidelines for the highest (field) tier of risk assess-ment are currently available only for earthwormtests (ISO 1999), but similar guidance is given forbees and some NTAs by other organisations suchas the International Organisation for BiologicalControl of Noxious Animals and Plants (IOBC)(Candolfi et al. 2000). The risk to soil fauna is as-sessed in-crop, whereas for NTAs off-crop risk as-sessment is also required (EPPO 2003).

Problems of extrapolationThe prediction of pesticide risks to terrestrial non-target invertebrates poses a challenge in regulato-ry risk assessment. Terrestrial invertebrate com-munities are complex, involving hundreds of inter-

acting species whose ecological roles and impor-tance are temporally and spatially dynamic but notalways well understood. Worst-case assessment ofpesticide risks in the laboratory, on the otherhand, necessarily focuses on a small subset ofspecies under controlled conditions. Key difficul-ties are the realistic prediction of indirect effects,long-term effects, impacts of multiple stressors,and recovery.

Modelling approaches are increasingly beingused to address some of these issues but arestrongly limited by the availability and quality ofempirical data. Risk assessment methods are avail-able for use with small datasets (e.g., probabilisticapproaches, species sensitivity distributions), butthe data shortage invoking the use of such ap-proaches could preclude their validation. Some-times the underlying assumptions for the use ofprobabilistic approaches are not satisfied (e.g.,species sensitivity distributions based on non-ran-dom selection of species).

Despite the fact that some tests still need to bestandardised, it is no longer true that in compari-son to aquatic ecotoxicology the number of terres-trial guidelines is too low. However, the availabili-ty of data (access to existing data or the produc-tion of new data) still limits any general discussionof the existing ecological risk assessment (ERA)scheme for PPPs.

Effects of PPPs on bees� Standard guidelines for all test levels are

available.� Results of field tests are based mainly on “ex-

pert judgement”.� Monitoring studies are not required, but are re-

gionally performed (e.g., in the UK).The UK monitoring shows that no pesticides classi-fied as having “low risk” were implicated in poi-soning incidents (Aldridge and Hart 1993), while,for example, pyrethroids (classified as “high risk”in the laboratory) are of low hazard in the field(Inglesfield 1989). Therefore, the current ERAscheme for bees seems to be successful. However,in the public domain potential side effects of PPPson bees are intensively discussed (e.g., in France).

Lessons learned from bee testing According to the Bee Assessment Scheme of

EPPO (2003), field test results should be regardedas decisive when conclusions from lower-tier testsconflict with those from field tests. In addition,monitoring schemes should be used for the valida-


90

a p p e n d i x a

CONTENTS

tion and refinement of ERA methods and results inother non-target groups. However, such an imple-mentation will be more difficult due to higherspecies numbers, lack of experience, etc. (Lewis etal. 1998).

Effects of PPPs on Non-target Arthropods� Standard guidelines for all test levels are avail-

able (but relatively few for field tests).� Results of field tests are mainly based on “ex-

pert judgement”.� Monitoring studies are not required but are re-

gionally performed (e.g., in the UK).� The results of field as well as monitoring stud-

ies are difficult to assess because in-field andoff-field situations have to be distinguished,high numbers of species from various taxa withclearly diverse biological properties in very dif-ferent exposure situations are potentially in-volved.

� Assessment of higher-tier results is still underdiscussion (in particular, a definition of theterm “ecologically significant effects” is lack-ing).

However, some recent compilations have shownthat it is possible to identify “standard test species”(T. pyri and A. rhopalosiphi), which are on averagethe most sensitive ones, and to recommend triggervalues for higher tier testing if data are both suffi-cient and available. The following issues need more discussion: � Is the database, in particular from the field,

sufficient (e.g., for Aphidius sp.)?� How can multiple applications be considered?� Is the sensitivity (but also biology) of standard

test species comparable to the sensitivity ofarthropod off-field communities? Is the ecolog-ical knowledge of off-crop communities suffi-cient? What about their exposure conditions?

� How can an “ecologically relevant” period andrecovery be defined?

� How can local and regional variability (both interms of exposure and effects) be included inERA schemes?

� Is there a need for more guidance on fieldstudy design (depending on the aims and localconditions)? How can the effects of persistentPPPs plus the different recovery strategies beincluded?

� Assuming that the species level is the most ap-propriate one to address, is the taxonomic

knowledge about these species sufficient, andcan it be handled in a practical way?

Effects of PPPs on soil(below-ground) invertebrates

� Standard guidelines for all test levels are avail-able (but just one for field tests).

� Results of field tests are partly based on “expertjudgement”.

� Monitoring studies are not required but areperformed locally, usually as part of researchprojects studying the role of earthworms andother ecosystem engineers.

Due to the limited ecological knowledge (exceptfor earthworms and, probably, Collembola) and tothe problems related to field studies in general, ex-pert judgement is clearly needed for these higher-tier studies.

Concerning species sensitivity, the followingconclusions can be drawn from some reviews (e.g.,on earthworms; Jones and Hart 1998) and an on-going project sponsored by DEFRA (WEBFRAM 5):� No general pattern is visible (e.g., the compost

worm E. fetida is not always less sensitive thanother lumbricid species).

� Sensitivity differences between taxonomicgroups occur.

� Within one taxonomic group (i.e., the familylevel), sensitivity differs at about a factor of 10(earthworms: 22 out of 25 pesticides), but ex-ceptions are possible (up to a factor of 72 withPPPs; Jones and Hart 1998).

� The sensitivity ratio (the factor describing sen-sitivity differences for one chemical) for soil in-vertebrates is: SR95:5 = 437 (Elmegaard andJagers op Akerhuis 2000).

While the influence of different species sensitivi-ties can be substantial, different soils usually havea smaller impact on the lab-to-field extrapolation(e.g., the influence of the high organic amount inOECD artificial soil is covered by a factor of 2 forchemicals with a log Kow >2; EPPO 2003). How-ever, it must be taken into account that certain soilproperties can already stress individual species;therefore, the extrapolation to, for example, acidsoils might be problematic. In addition, the differ-ent behaviour of individual species can alter theexposure to PPPs considerably.

There have been several attempts to derive ra-tios between acute and chronic test results as wellas between chronic laboratory tests and field tests


91

CONTENTS

(e.g., Barber et al. 1998; Heimbach 1998), but stillthe amount of data is small and for the time being,“artificial” values are used.

Looking at the effects of PPPs in the field, it isgood news that so far there is only one exampleshowing that soil invertebrates were eradicateddue to the impact of chemicals (species of theearthworm genus Scherotheca by heavy metals inparts of southern France; Abdul-Rida and Bouche1995). However, because detailed monitoring stud-ies are rare it is difficult to know whether suchlong-term and large-scale effects are genuinely in-frequent or simply are not being detected (e.g.,long-term effects of synthetic pyrethroids onarthropods were revealed by a unique time seriesof data in the UK (Ewald and Aebischer 1999).

Concerning the use of PPPs in English farm-lands, some general results are listed (for detailssee, e.g., Frampton 1997):� Community responses differ clearly among

groups and species of arthropods. � Short-term effects of insecticides nearly always

occurred, while long-term effects were very sitespecific and group specific.

� Non-conventional agriculture (i.e., a reduceduse of PPPs) favours soil invertebrates, but thevariability in time and scale is high and effectsof, for example, tillage are often higher thanthose of PPPs (e.g., earthworms: Tarrant et al.1997).

In a case study with Oligochaeta and the fungicidecarbendazim, in which laboratory, semi-field (i.e.,terrestrial model ecosystems), and field studieswere performed at 4 European sites, it could beshown that the increasing realism when going tohigher-tier studies can lower the resultingPEC–PNEC ratios considerably (Römbke et al.2004; Weyers et al. 2004; see also Weyers et al.this volume). However, the high variability ofthese complex tests is difficult to assess (wouldone study be sufficient to cover, e.g., different geo-graphical localities?).

Recovery of soil invertebrates in the field afterthe use of PPPs has been studied with Collembolaand earthworms (for a review, see Van Straalenand Van Rijn 1998). It seems that for the formergroup, the most astonishing result of these studiesis that the potential for recovery from adjacent,untreated field margins or unsprayed areas of crophas been overestimated (e.g., Frampton 2002). Forearthworms, some more details can be given:

� Recovery strongly depends on the persistenceof the PPP (Jones and Hart 1998): DT50 <50days = recovery within 6 to 12 months; DT50>50 days = no reliable estimation possible.

� The role of immigration for recovery is notclear. Migration of up to 10 m/year is possible,but repellent effects are known (e.g., for beno-myl: Heimbach 1997; Mather and Christensen1998).

� The species composition is rarely considered,but for the ecosystem, anecics are usually moreimportant than endogeics.

� Other stress factors related to agricultural prac-tise (or even the diffuse impact of other chemi-cals) influence the populations as well.

As a rule of thumb, any effect of more than 50%impact on earthworm populations or importantspecies under farm conditions should cause con-cern.

Effects of PPPs on soil functions (processes)� Standard guidelines for microbial (laboratory

level) and decomposition tests (litter bag) areavailable (e.g., EPFES 2003).

� Trigger values are currently arbitrary.� Monitoring (as opposed to testing) of these

functional processes is not required and ishardly done at all.

The litter-bag test is triggered by the persistence ofa compound and, partly, by effects in single-species laboratory tests. Since these two endpoints(mortality or reproduction of organisms and or-ganic matter breakdown) are not directly linked,research on the structure-function relationship isneeded. Two possibilities exist:� A field test with soil invertebrates (i.e., Collem-

bola and mites) in analogy to or in combinationwith the standard earthworm field guideline(ISO 1999).

� A monitoring approach (see also the accompa-nying contribution by T. Ratte): comparison ofpotentially impacted sites with reference valuesderived from “unaffected” sites in analogy toaquatic concepts like RIVPACS or BEAST(Reynoldson et al. 2000). In the following fig-ure, the general approach is described (Ruf etal. 2003).


92

a p p e n d i x a

CONTENTS

Recommendations� Improvement of data availability (by either col-

lating existing data or generating new data).� Collation of (ecotoxicological and ecological)

data of the most important (test) species in acentral database.

� Use of these data in modelling (note theorder!).

� Clarification of the aim of higher-tier studies(e.g., by defining what is an “ecologically rele-vant” recovery time).

� Provision of guidance on (regionally differenti-ated) field studies as well as monitoring studies(see, e.g., ISO draft guidelines on field sam-pling of soil invertebrates).

� Investigation of the relationship between struc-tural and functional endpoints.

� Identification of the role and guidance on theperformance of monitoring studies.

R e f e r e n c e sAbdul Rida AMM, Bouche MB. 1995. The eradi-

cation of an earthworm genus by heavy metalsin southern France. Appl. Soil Ecol. 2:45–52.

Aldridge CA, Hart ADM. 1993. Validation of theEPPO/CoE risk assessment scheme for honey-bees. Proc. 5th International Symposium on theHazards of Pesticides to Bees October 26–28,1993; Wageningen, The Netherlands, p 37–41.

Barber I, Bembridge J, Dohmen P, Edwards P,Heimbach F, Heusel R, Romijn K, Rufli H.1998. Development and evaluation of triggersfor earthworm toxicity testing with plant pro-tection products. In: Sheppard SC, BembridgeJD, Holmstrup M, Posthuma L, editors. Ad-vances in earthworm ecotoxicology. Pensacola(FL): SETAC. p 269–278.

Candolfi M, Blümel S, Forster R, BakkerFM, Grimm C, Hassan SA, Heimbach U,Mead-Briggs M, Reber B, Schmuck R, VogtH. editors. 2000. Guidelines to evaluate side-effects of plant protection products to non-tar-get arthropods. IOBC/OILB Publication,158 p.

Elmegaard N, Jagers op Akerhuis GAJM.2000. Safety Factors in Pesticide Risk Assess-ment. Differences in Species and Acute –Chronic Relations. National Environmental Re-search Institute NERI Technical Report nr 325,Silkeborg (Denmark). 60 p

Frampton GK. 1997. The potential of Collembolaas indicators of pesticide usage: Evidence andmethods from the UK arable ecosystem. Pedobi-ologia 41:179–184.

Frampton GK. 2002. Long-term impacts of anorganophosphate-based regime of pesticides onfield and field-edge Collembola communities.Pest Manag. Sci. 58:991–1001.

EPFES. 2003. Guidance Document: Effects of PlantProtection Products on Functional Endpoints inSoil (EPFES). Römbke JHF, Hoy S, Kula C,Scott-Fordsmand J, Sousa P, Stephenson G,Weeks J, editors. Pensacola (FL): SETAC. 92 p.

EPPO. 2003. Environmental risk assessmentschem for plant protection products.OEPP/EPPO Bull. 33: 99–101.

Ewald JA, Aebischer NJ. 1999. Pesticide use,avian food resources and bird densities in Sus-sex. Peterborough, UK: Joint Nature Conserva-tion Committee JNCC. Report nr 296. 148 p

Heimbach F. 1997. Field tests on the side effects ofpesticides on earthworms: Influence of plotsize and cultivation practices. Soil Biol Biochem29:671–676.

Heimbach F. 1998. Comparison of the sensitivitiesof an earthworm (Eisenia fetida) reproductiontest and a standardized field test on grassland.In: Sheppard SBJ, Holmstrup M, Postuma L,editors. Advances in earthworm ecotoxicology.Pensacola (FL): SETAC. p223–245.

Inglesfield C. 1989. Pyrethroids and TerrestrialNon-target Organisms. Pestic Sci 27:387–428.

Jones A, Hart ADM. 1998. Comparison of labora-tory toxicity test for pesticides with field effectson earthworm populations: a review. In: Shep-pard SBJ, Holmstrup M, Postuma L, editors.Advances in earthworm ecotoxicology. Pen-sacola (FL): SETAC. p 247–267.


93

CONTENTS

(ISO) International Organization for Stan-dardization. 1999. Soil quality – Effects ofpollutants on earthworms (Eisenia fetida). Part3: Guidance on the determination of effects infield situations. Geneva: ISO. Report nr 11268-3. 8 p.

Lewis GB, Stevenson JH, Oomen PA. 1998.Honey bees in Europe: lessons learned forother terrestrial non-target arthropods. In: Eco-toxicology: Pesticides and beneficial organisms.Haskell PT, McEwen P. editors. London (UK):Chapman & Hall. p 248–256.

Mather JG, Christensen OM. 1998. Earthwormsurface migration in the field: Influence of pes-ticides using benomyl as test chemical. In:Sheppard SC, Bembridge JD, Holmstrup M,Posthuma L, editors. Advances in earthwormecotoxicology. Pensacola (FL): SETAC. p327–340.

Reynoldson TB, Day KE, Pascoe T. 2000. Thedevelopment of the BEAST: a predictive ap-proach for assessing sediment quality in theNorth American Great Lakes. In: Wright JF,Sutcliffe DW, Furse MT. editors. Assessing thebiological quality of freshwaters. RIVPACS andother techniques. Ambleside (UK), FreshwaterBiological Association. p 165–180.

Römbke J, Jones SE, van Gestel CAM, Kool-haas J, Rodrigues J, Moser T. 2004. Ring-Testing and Field-Validation of a TerrestrialModel Ecosystem (TME) – An Instrument forTesting Potentially Harmful Substances: Effectsof carbendazim on earthworms Ecotoxicology13:105–118.

Ruf A, Beck L, Dreher P, Hund-Rinke K, Rom-bke J, Spelda J. 2003. A biological classifica-tion concept for the assessment of soil quality:“biological soil classification scheme” (BBSK).Agric Ecosys Environ 98:263–271.

Tarrant KA, Field SA, Langton SD, HartADM. 1997. Effects on earthworm populationsof reducing pesticide use in arable crop rota-tions. Soil Biol Biochem 29:657–661.

van Straalen NM, van Rijn JP. 1998. Ecotoxico-logical risk assessment of soil fauna recoveryfrom pesticide application. Rev Environ ContamToxicol 154:83–141.

Weyers A, Sokull-Klüttgen B, Knacker T,Martin S, van Gestel CAM. 2004. Use of Ter-restrial Model Ecosystem Data in Environmen-tal Risk Assessment for Industrial Chemicals,Biocides and Plant Protection Products in theEU. Ecotoxicology 13:163–176.


94

a p p e n d i x a

CONTENTS

ability topredict effects in birds:

linking tiers andextrapolation

Mineau, P. and Whiteside, M.

National Wildlife Research Centre,Canadian Wildlife Service, Ottawa, Canada

Currently, most of the regulatory effort in pesti-cide avian risk assessment centres on the develop-ment of models incorporating laboratory-deter-mined toxicity as well as estimates of food-itemresidues, consumption rates, proportion of forag-ing time in treated areas, and any avoidance re-sponse to treated foodstuffs. These models can beof a deterministic nature but, increasingly, naturalvariation and model uncertainty are incorporatedinto the risk assessment process through proba-bilistic tools such as Monte Carlo simulations.These assessments are typically carried out chemi-cal by chemical or, in some cases, on a cluster ofpesticides used for the same purpose.

Toxicity has been relatively easy to model inavian risk assessment because several approacheshave been developed to process data obtained ondifferent species in order to put any prediction oftoxicity on a reasonably firm quantitative basis.Unfortunately, these techniques are often ignoredby regulatory bodies in favour of arbitrary andhighly stochastic single-species toxicity endpoints.Exposure, on the other hand, has proven particu-larly difficult to model, and such models tend to bevery data intensive. Worse, estimates of exposure(hence, the very structure of existing models) havehad almost no verification or validation against ac-tual field outcomes. Fundamental assumptions in-herent in the models, such as the belief that mostof the exposure comes from the ingestion of conta-minated food, are critical to model validity; yet theavailable evidence (Mineau et al. 1990; Driver etal. 1991) suggests that, for pesticide sprays at least,these assumptions are often incorrect.

A single directed field study by itself may notbe sufficient to dispel a presumption of high riskthat is placed on a pesticide. This deficiency re-sults from the stochastic variability encountered inmost field situations as well as from our inabilityto detect impacts every time they occur (in partbecause of the difficulty of finding evidence of animpact such as carcasses). Recognising this, theUnited States Environmental Protection Agency

(USEPA) proposed that a large number of fieldsneeded to be monitored in order to increase theconfidence of a finding of no effect (Fite et al.1988). This requirement meant that substantialcosts needed to be expended for each pesticideunder presumption of causing avian harm (essen-tially mortality). This was one of the reasons fieldstudies eventually fell out of favour with the USpesticide regulatory system. Yet, the logic wasprobably sound: several field studies are needed touncover avian mortality that is either sporadic oreven regular, but with a low frequency of occur-rence.

A recent analysis of the avian pesticide studyrecord (Mineau 2002) showed that it was possibleto generate empirically based models to predictthe likelihood of visible avian mortality. Mortalityis certainly not the only endpoint of importance inavian risk assessment, but it is nevertheless theone that has attracted the most attention. Themost accurate model developed by Mineau (2002)requires the following inputs: the “toxic potential”of the pesticide, a measure that incorporates HD5(sensu Aldenberg and Slob 1993; values fromMineau et al. 2001 as well as application rate); thedermal toxicity index of the pesticide (calculatedfrom rat data as well as physicochemical con-stants); and Henry’s law constant, a reflection ofthe potential inhalation exposure. This model wasable to classify safe and lethal applications in thesample with better than 80% accuracy. We com-pared the output of this reality-based model to thegeneric (it stays the same regardless of the crop)small insectivore Tier 1 toxicity–exposure ratio(TER) calculation proposed in various drafts of theEU guidance document on risk assessment forbirds and mammals under Council Directive91/414/EEC (e.g., EU 2002).

For all the pesticides represented in Mineau’s2002 sample of field studies, we calculated therate of application that is expected to lead to avianmortality 10% of the time or following 1 in 10 ap-plications. This choice is entirely arbitrary: thisfrequency of wildlife incidents may not be accept-able in the case of very broad uses or wherespecies of conservation interest are targeted. How-ever, we think there would be general agreementthat a pesticide-use pattern giving rise to mortalityin 1 out of every 10 fields should warrant morescrutiny than Tier 1 can offer. For illustration pur-poses, we used the model developed by Mineau(2002) for field crop and pasture applications. We


95

CONTENTS

then assumed that this calculated rate was the raterequested for registration and determined whetheror not the registration request would be sent toTier 2 (TER <10) or passed through without fur-ther investigation (TER >10). TERs were calculat-ed with northern bobwhite, Japanese quail, ormallard LD50 values. Based on precedence, weconsider these to be the most likely species thatwould be tested for a new pesticide registrationbeing submitted to the EU. We used either themaximum values available or a geometric mean ofavailable values reflecting the chance element in-volved in selecting a toxicity value or a systematicbias on the part of registrants. Also, in order to re-move any remaining doubts surrounding themodel or the arbitrary choice of a 10% risk of mor-tality, we computed TERs for all available fieldstudies with a determination of bird mortality ordebilitation.

Based on the 2001 suggestion of a FIR (food in-take ratio) of 1.03 and a RUD (residue level calcu-lated for a 1 kg/ha a.i. application) of 11 mg/kg,there was a high “failure rate” in that between29% and 44% of approximately 30 organophos-phorus and carbamate insecticides would havesuccessfully passed Tier 1, even though the fieldevidence suggests that, at the proposed rate of ap-plication, they kill birds 10% of the time.

In response to comments, the 2002 draft guid-ance document proposed abandonment of the 11-ppm RUD for “small” insects typically consumedby birds and adoption of a 52-ppm RUD, the maxi-mum value proposed by Hoerger and Kenaga(1972) as an interim value until the issue of insectresidues could be sorted out. Predictably, the “Tier1 failure rate” for the same sample of insecticidesdropped to between 0% and 16%, depending onthe starting point toxicity value.

Not surprisingly, the TERs calculated for the in-dividual field studies were also misleading for ahigh proportion of field studies. (Here, we com-bined all studies, whether on field crops, orchards,or woodlots, because the draft assessment guide-lines do not specifically distinguish between thesescenarios.) Using the maximum of the LD50 valuesavailable for the bobwhite, the Japanese quail, ormallard, TERs calculated with an insect RUD of 11ppm would have failed in the case of 11 separateinsecticides and would not have predicted ob-served field mortality or severe debilitation fol-lowing some applications of acephate, azinphos-methyl, chlorpyrifos, diazinon, dimethoate, feni-

trothion, fenthion, methamidophos, methomyl,mevinphos, phoxim, and propoxur. Repeat calcula-tions with a RUD of 52 ppm still would not havepredicted field studies reporting either mortalityor debilitation with acephate, chlorpyrifos,dimethoate, fenitrothion, phoxim, and propoxur.

Fiddling with model constants such as the RUDcan clearly be used to bring Tier 1 predictions morein line with available field evidence. However, aslong as the underlying problem is not solved, thisalso will result in a much larger number of “falsepositives”. Based on a number of field studieswhere no mortality of debilitation was seen, we es-timate that the proportion of false positives couldeasily be as high as 40–70% of pesticides, depend-ing on whether a RUD of 11 or 52 was used. We be-lieve the only scientifically defensible way ofachieving the proper Tier 1 balance between falsepositives and false negatives is to have a goodmodel in the first place. Based on our earlier work,this would mean considering exposure routesother than the ingestion of contaminated food. Itis not a coincidence that several of the insecticideswhose risk was most easily underestimated basedon documented field studies (e.g., acephate, chlor-pyrifos, dimethoate, and fenitrothion) are amongthose products with the highest dermal toxicity in-dices. Another (phoxim) has the highest Henry’slaw constant of the products examined, suggest-ing, again, that the impact documented in the fieldresulted from exposure other than food intake. Itis clearly in the interests of both regulators and in-dustry to put more emphasis in this area and fixthe problem.

R e f e r e n c e sAldenberg T, Slob W. 1993. Confidence limits

for hazardous concentrations based on logisti-cally distributed NOEC toxicity data. EcotoxicolEnviron Saf 25:48–63.


Driver CJ, Ligotke MW, Van Voris P, McVeetyBD, Greenspan BJ, Drown DB. 1991. Routesof uptake and their relative contribution to thetoxicologic response of northern bobwhite(Colinus virginianus) to an organophosphatepesticide. Environ Toxicol Chem 10:21–33.


96

a p p e n d i x a

CONTENTS

Fite EC, Turner LW, Cook NJ, Stunkard C.1988. Hazard evaluation division, standardevaluation procedure: guidance document forconducting terrestrial field studies. UnitedStates Environmental Protection Agency,EPA/540/09-88/109, Washington DC (USA), p1–66.

Hoerger FD, Kenaga EE. 1972. Pesticidesresidues on plants: correlation of representa-tive data as a basis for estimation of their mag-nitude in the environment. EnvironmentalQuality. New York: Academic press. p 9–28.

Mineau P, Sundaram KMS, Sundaram A,Feng C, Busby DG, Pearce PA. 1990. An im-proved method to study the impact of pesticidesprays on small song birds. J Environ Sci HealthB25:105–135.

Mineau P, Baril A, Collins BT, Duffe JA,Joermann G, Luttik R. 2001. Pesticide acutetoxicity reference values for birds. Rev EnvironContam Toxicol 170:13–74.

Mineau P. 2002. Estimating the probability of birdmortality from pesticide sprays on the basis ofthe field study record. Environ Toxicol Chem21:1497–1506.


97

CONTENTS

ecosystem dynamicsand stability:

are the effects ofpesticides ecologically

acceptable?

Ratte, H.T1; F. Lennartz2 & M. Roß-Nickoll1

1) RWTH, Aachen University, Aachen, Germany2) Research Institut gaiac, Aachen

AbstractThis paper focuses on the non-target communitiesin agro-ecosystems that could undergo frequentperturbations by agricultural practises. Some fre-quently discussed concepts of theoretical commu-nity ecology and some general conclusions arebriefly described. Among these conclusions arethat (1) a tendency for inherent stability increasesas complexity decreases; (2) relatively stable envi-ronments support complex but fragile communi-ties, while relatively variable environments allowthe persistence of only simpler, more robust com-munities; (3) complex, fragile communities of rel-atively constant environments (e.g., the tropics)are more susceptible to outside, unnatural distur-bance than the simpler, more robust communitiesthat are more accustomed to disturbance (e.g., inmore temperate regions).

The main conclusion of the paper will be thatin the context of an acceptability discussion, thegeneral considerations should be followed by anapplied approach, as communities exhibit differ-ent structures across the landscapes. Thus, the ac-ceptability of effects can be discussed only on thebasis of the local communities and agriculturalpractises. An approach is proposed on how to es-tablish an inventory of reference habitats or com-munities, by means of which unacceptable effectsin non-target habitats could be identified.

IntroductionThe acceptability of anthropogenic effects onecosystems largely depends on values developedby human societies rather than on conclusions de-

rived from ecological science. Valued properties ofnatural ecosystems undergo changes in time andare quite different among those who use nature(e.g., for agriculture, sports, game fishing, or com-mercial fishing) and those who want to protect na-ture for ethical reasons.

Plant protection products (PPPs) are applied todo their job on target sites in agricultural ecosys-tems1, but non-target sites are often affected unin-tentionally, resulting in perturbations of terrestrialand aquatic communities or habitats surroundingthe cropped areas. With respect to the target sitesof PPPs, it has long been recognised that there is aneed to monitor the sustainable maintenance ofthe soil functions necessary for growing crops andthus to protect the composition and functioning ofthe associated in-soil fauna.

The attractiveness of the cultivated landscapecomes through a mix of various structural compo-nents in addition to the cropped areas, such asvegetation strips, hedges, ponds, small rivers, andwoods, which contribute to the characteristicimage of the landscape. These components formhabitats for a number of biotic communities con-tributing to the agro-ecosystem. Although in ac-cordance with current agricultural practises, theapplication of PPPs is also expected to impact non-target communities (e.g., their structure [diversi-ty] and functioning). Consequently, the followingissues arise: how these communities are affected,whether they will remain structurally stable in fu-ture, and how community composition (diversity)and functioning of the agro-ecosystem can bemaintained in a sustainable manner.

Quick journey throughtheoretical community ecology

I shall briefly examine whether these questions (orat least some of them) can be answered theoreti-cally by the ecological science, following the linesgiven in modern textbooks (e.g., Begon et al.1990). Community ecology deals with the speciescomposition or structure of communities and withthe pathways followed by energy, nutrients, and

i m p l i c a t i o n s f o r r i s k a s s e s s m e n t

98

a p p e n d i x a

Implicationsfor risk assessment

CONTENTS

other chemicals as they pass through them (thefunctioning of communities). Unfortunately, dif-ferent terminology has developed between plantand animal ecologists. The term “community” ismost frequently used by zoologists in both a gener-al and specific sense, whereas botanists (plant so-ciologists) use “association” as the fundamentalunit representing a plant community of definitefloristic composition. An association is composedof a number of stands that are concrete units ofvegetation observed on a site. Zoologists havebeen more concerned with functional relation-ships such as food webs and energy flow throughthe community, whereas botanists have been moreconcerned with taxonomic and structural relation-ships in the community and the way these changein space and time. The often more-comprehensivestudies of zoologists have dealt with plants as ani-mal food, and botanists have tended to ignore theanimals.

Important community characteristics arespecies diversity, growth form and structure, dom-inance of species, relative abundance of species,and trophic structure. Species diversity comprisesall species of plants and animals living in a partic-ular community. A species list is a simple measureof species richness or species diversity. The differ-ent growth forms determine the stratification, orvertical layering of the community (e.g., trees,shrubs, herbs, and mosses and broad-leaved andneedle-leaved trees). Dominant species exert amajor influence on the community by virtue oftheir size, numbers, or activities. Dominant speciesare highly successful ecologically and determine toa considerable extent the conditions under whichthe associated species must grow. The relativeabundance is the relative proportion of species in acommunity. Trophic structure describes the path-ways of material and energy in a community: whoeats whom?

Among the methods to analyse the compositionof communities in space are gradient analysis, or-dination, and classification. Experience tells usthat homogeneous environments hardly exist innature. Most environments contain within themgradients of conditions or available resources.There is much evidence that the existence of onetype of organism in an area immediately diversi-fies it for others. Changes in species and environ-mental conditions in time are either seasonal cy-cles or so-called successions. These are long-termchanges and run through various stages, from the

first colonisation of new land to the stabilisation ofcommunities. The final or mature stage is calledclimax, and there is a debate on whether there isone definite climax stage at a certain geographicalarea (mono-climax) or multiple climaxes (poly-cli-max). If a community develops anew on bare land,a so-called primary succession takes place, where-as a secondary succession occurs after a distur-bance. With regard to agricultural activities andPPP applications, it is important that both activi-ties prevent the natural succession or put succes-sion back to an earlier stage at least in the targetsites.

Because impacts of agricultural measures, inparticular of PPPs, cannot be completely prevent-ed, one has to study whether communities of non-target sites remain stable and sustain the associat-ed perturbations in future times. Some major as-pects of stability of ecological systems are de-scribed as follows:� Resilience describes the speed with which a

community returns to a former state.� Resistance is the ability of the community to

avoid displacement arising from a perturba-tion.

� Local stability is the tendency of a communityto return to its original state (or somethingclose to it) when subjected to a small perturba-tion.

� Global stability is achieved if this tendency isshown when the community is subjected to alarge perturbation.

� Robustness describes the stability of a commu-nity in a wide range of environmental factors.

� Fragility describes the stability of a communityin a narrow range of environmental factors.

The “conventional wisdom” states that “an in-crease in complexity leads to an increase in stabili-ty”. This statement, however, has been under-mined by more recent work and mathematicalmodelling. No clear-cut relationship was found be-tween the complexity of a community and its in-herent stability. The stability appears to depend onthe nature of the community itself, the way inwhich it is disturbed, and the way in which stabili-ty is assessed. Nonetheless, a tendency for inher-ent stability to increase as complexity decreaseswas obvious. Among other tendencies observedare that relatively stable environments supportcomplex but fragile communities, whereas rela-tively variable environments allow the persistence

a b st r ac t s o f p l at f o r mp r e s e n tat i o n s i m p l i c a t i o n s f o r r i s k a s s e s s m e n t

9999

CONTENTS

of only simpler, more robust communities. Thelikelihood exists that complex, fragile communi-ties of relatively constant environments (e.g., thetropics) are more susceptible to outside, unnaturaldisturbance (and are more in need of protection)than the simpler, more robust communities thatare more accustomed to disturbance (e.g., in moretemperate regions).

Probably, there is a parallel in the properties ofthe community and the properties of the compo-nent populations. In stable environments popula-tions will be subject to a high degree of K-selec-tion; in variable environments they will be subjectto a relatively high degree of r-selection. The K-se-lected populations (high competitive stability, highinherent survivorship, low reproductive output)will be resistant to perturbations, but once per-turbed will have little capacity to recover (low re-silience). The r-selected populations will have lit-tle resistance but a higher resilience.

An approach to acceptabilityThe rationale behind ongoing discussions on eco-logical stability, long carried out by risk assessorsand those applying the PPPs, appears to be thereasoning that resistant or resilient communitiesmight have developed in the agricultural areasthat can sustain impacts brought about by thePPPs. However, this hypothesis cannot be derivedor proven by theoretical community ecology, espe-cially not the acceptability of impacts. In practise,we are faced with concrete habitats embedded in alandscape that is comprised of gradients of condi-tions as well as of fauna and flora. The speciescomposition varies across local and global scales.Questions on the impact of PPPs and the accept-ability of their effects can be answered only at alocal scale. Recovery from effects also dependslargely on the specific local conditions (e.g., onwhether allogeneic recovery sources such as meta-populations exist). Thus, the conclusion here willbe that applied ecology has to be put in place togenerate the basic information on a local scale,which then can be used to discuss acceptability.

I shall now propose an approach, the central el-ement of which is the development of target im-ages for the major habitats and communities innon-target areas of the agro-ecosystem. The ap-proach is similar to one performed for small rivers(e.g., in Germany) and is based on target imagesfor rivers of the various geological areas and con-crete reference rivers coming close to the target

image, though influenced by humans. A riverunder consideration can be compared with the ref-erence, and measures can be taken when its quali-ty does not come close enough to that of the refer-ence. This approach shall be performed in fulfil-ment of the requirements of the EU water frame-work directive.

According to this directive, the terrestrial ap-proach would be to� generate an inventory of habitats and commu-

nities in non-target areas of the agro-ecosystemspecific for the geographical regions;

� generate agreement on target images and refer-ence sites;

� define boundaries below which deviations fromthe reference are seen as unacceptable;

� compare the habitats and communities in theconsidered region with their reference sites;

� come to conclusions on the acceptability ofstructural deviations from the reference; and

� come to conclusions about measures to betaken to remediate the habitats, depending onthe conclusions reached above.

This approach is illustrated in Figure 1. The varietyof non-target habitats appears to be small, andthus this task could be performed in relatively lit-tle time. A recurrent monitoring comparison of

local habitats and communities with their refer-ences would lead to sustainable preservation ofthe cultivated landscape and maintenance of theirtypical diversity of habitats and species.

R e f e r e n c eBegon M., Harper J.L., Townsend C.R. 1990.

Ecology: Individuals, Populations and Commu-nities. Oxford (UK): Blackwell Scientific Publi-cations, p 945.


100

a p p e n d i x a

CONTENTS

100

Figure 1: Proposed approach to acceptability


101

the role of data frommonitoring studies in the

regulatory framework

Streloke, M.1, Tarazona, J.V.2 and Reinert, W.3

1) Federal Office for Consumer Protection and Food Safety (BVL),Braunschweig, Germany

2) Department of the Environment, INIA, Madrid, Spain3) European Commission, DG Health and Consumer Protection,

Brussels, Belgium

Annexes II and III of EU Directive 91/414/EEC donot explicitly require data from monitoring stud-ies. Consequently, in Annex VI no specific assess-ment schemes for monitoring data are given. Thisabsence is not surprising, because an applicantmust show that a product is safe before it is placedon the market, and current regulations are dealingwith this situation. Monitoring data under realisticuse conditions can only be collected later on. How-ever, monitoring data have been required undernational regulations for special purposes (e.g., toconfirm that risk mitigation measures are effec-tive). Below, examples for monitoring are present-ed.

Based on the state of the art in ecochemistryand ecotoxicology, the methods used to assess therisk for non-target organisms lead to conservativerisk predictions, and it is very likely that unaccept-able risks are precluded. However, a certain de-gree of uncertainty is inevitable, and therefore thepredictions will never be perfect. Furthermore, thetools currently used are focused on single com-pounds and disregard combinations of effects. Or-ganisms living under real-field conditions inte-grate all effects and all interactions. Consideringthat the sustainability of these populations is theultimate protection aim, it is reasonable to con-duct such monitoring studies from a scientificpoint of view. Additionally, these data should beused to demonstrate to the public that the use ofplant protection products is safe. The interpreta-tion of results should be conducted by an indepen-dent board of experts to ensure transparency andreliability. It is recommended that the authoritiesresponsible for the authorisation of plant protec-tion products identify typical vulnerable areaswhere continuous monitoring studies should beconducted.

Standard environmental risk assessments arebased mainly on toxicity data from laboratory testsand on more-or-less sophisticated exposure esti-mates. During the past few years, especially in the

area of plant protection products, the risk esti-mates have become more realistic because, for ex-ample, data from higher-tier tests have been used.However, even when these modern tools are used,it is unclear whether the risk estimates match theactual risk prevailing in reality, that is, to whichdegree the risk estimate is overprotective or under-protective. Monitoring data can answer this ques-tion, although their interpretation is difficult andno precise numbers are to be expected. In a broad-er sense, data from monitoring studies can be usedfor a “reality check” of risk assessment schemes.

CONTENTS

geographical differencesin the evaluationand prediction of

the effects of pesticides

José V. Tarazona

Department of the Environment, INIA

The regulatory environmental risk assessment ofpesticides is conducted mostly through the extrap-olation of laboratory data to natural systems. Inaddition, the evaluation covers generic agronomicpractises. However, the biological characteristicsand the pressure of natural ecosystems and com-munities associated with agricultural land are verydifferent among regions.

Fortunately, direct effects of pesticides onaquatic systems and terrestrial wildlife in Europehave been significantly reduced due to regulatorymeasures, farmers’ training, and an increase in en-vironmental concerns.

This contribution will review the potential dif-ferences among geographical regions, particularlythe specific differences of the Mediterranean re-gion, in the assessment of effects on pesticides inthe field.

The geographic dimensionGeographic differences are associated with signifi-cant agricultural differences. Four main aspectsshould be considered:

Regional crops: The connection of some cropsto climatic and related conditions is so high thatthe crops are exclusively associated with a certainregion and even with specific areas of a region.Rice, olive trees, and citrus trees are typical exam-ples of Mediterranean crops.

Environmental, climatic, and agricultural dif-ferences: Environmental differences among re-gions include climate, geography, landscape char-acteristics, etc. All these factors modify the rele-vance of different exposure routes, as well as thetype of scenario required for the assessment.

Differences in agricultural practises: Communi-ties exposed off-crop are challenged by the wholelist of pesticides used in the field and must followthe use pattern required for pest control in thearea. The treatments differ among areas due todifferences in the pests to be controlled and to dif-ferences in cropping characteristics, traditions,and regulatory and management practises. For ex-ample, the use of insecticides is particularly impor-

tant for Mediterranean conditions, where the cli-mate allows rapid development of insect pests.

Ecological differences: The biological commu-nity is obviously adapted to the geographical area.The differences are related not only to differencesin species but also to the role of each taxonomicgroup within the ecosystem. All main taxonomicgroups considered in the current pesticide assess-ment are present in all geographical areas. The se-lected species for tier I assessment are generic, andthere is no information to justify specific trends inthe sensitivity of species among the areas. Thus,no fixed sensitivity tendencies for direct effectscan be substantiated from current knowledge.However, differences in the ecological roles are ev-ident, and therefore differences regarding indirectand long-term effects can be hypothesised. Theconservation of habitats and species also presentssignificant differences among regions.

Several examples revealing the role of these differ-ences when assessing the effects of pesticides inthe field are discussed below.

Regional crops: RiceRice is a very special crop because the in-crop as-sessment combines aquatic and terrestrial taxa.The biodiversity of rice paddies is also very high,and most of the European area is located in thevicinity or even within areas of high ecological rel-evance. Wetland birds can get most of their foodfrom rice paddies. Large bird mortalities byorganophosphates have been described in ricepaddies, and management plants for the use ofagrochemicals in rice are assumed to be essentialelements for an overall sustainable management.

Environmental differences: the roleof exposure routes for aquatic systems

Spray drift is assumed to be largely responsible foreffects of pesticides on aquatic systems. Spray driftis a very local and punctual event, intimately asso-ciated with the application of the pesticide by thefarmer.

The effects, if any, can be observed immediate-ly or at least shortly after application. Monitoringincidents is relatively easy. However, the relevanceof spray drift in the Mediterranean region is rela-tively low, while runoff and soil erosion can be thelargest contributors to pesticide surface-water con-tamination. The events are associated with rain-fall, not pesticide application, and monitoring pro-


102

a p p e n d i x a

CONTENTS

102


103

grammes require significant efforts to identify ef-fects on aquatic communities and to quantify therole of pesticide application versus all other fac-tors affecting the ecosystems.

Differences in agricultural practises:Herbicide applications

Climatic conditions offer the possibility of specificuses of pesticides. The use of herbicides for facili-tating harvesting is typical for several crops, in-cluding citrus and olive trees. The climatic condi-tions may produce different results from apparent-ly similar practises. For example, in some areas,plant growth is associated with a very specific part

of the year, and the application of a herbicide canmaintain a bare soil for several months, within thetreated area and that affected by drift. Therefore,recovery of the non-target populations can be dra-matically affected.

Ecological differencesTwo aspects will be mentioned: the ecological rele-vance of indirect effects, such as the eutrophica-tion risk associated with direct effects on zoo-plankton in Mediterranean systems, and the con-sequences on the European biodiversity, related tothe effects on endangered species.

Environmental, climatic and agricultural differences.

CONTENTS

Eutrophication processes are very common inseveral parts of Europe. Every year, hundreds ofepisodes of algal blooms and associated fish mor-talities are reported in Spain. Light, temperature,reduced flow, large nutrient loads, etc., are obvi-ously factors affecting algal growth and are not re-lated to pesticide use. However, recent studieshave confirmed that under Mediterranean condi-tions, the role of zooplankton in controlling algalpopulations is a key element for ecosystem effects.Toxic concentrations of several pesticides havebeen reported in areas associated with eutrophica-tion problems, and a role of pesticides as an addi-tional element cannot be excluded.

Effects on biodiversity are particularly relevantwhen endangered species are affected. In the lastfew years, a large number of poisoning incidentson prey birds have been reported in Spain, includ-ing the most endangered species. A total of 454cases in black vulture (Aegypius monachus), 16 inbearded vulture (Gypaetus barbatus), and 72 in im-perial eagle (Aquila adalberti) have been reported.Half of these poisoning have been diagnosed aspesticide poisoning, with carbamates, organophos-phates, and some organochlorinated pesticidesbeing the main responsible agents. Most of thesepoisonings are associated with the illegal use ofpesticides for poisoning wild carnivores. However,effects associated with normal use of the pesticideby farmers have been suggested by other authors.

DiscussionDue to the particular conditions of the Mediter-ranean area, indirect and long-term effects associ-ated with the use of pesticides have become a criti-cal element for assessing the real impact of thesechemicals on communities and ecosystems.

A quantitative assessment of the relevance ofthese effects cannot be conducted with the currentlevel of information, but there are several indica-tors suggesting a significant contribution of pesti-cides to the overall anthropogenic hazards.


104

a p p e n d i x a

CONTENTS


appendix bLIST OF POSTERPRESENTATIONS

CONTENTS

CONTENTS

From laboratory to field: extrapolationof pesticide risks to non-target invertebratescommunities

Amorim M1, Soares A,1 and Römbke J2

1) Department of Biology, University of Aveiro, Portugal2) ECT Ökotoxikologie GmbH, Flörsheim, Germany

Aquatic risk assessment of realistic pesticideregimes in semi-field experiments

Arts GHP, van Wijngaarden RPAand Brock ThCMAlterra Green World Research, Wageningen Universityand Research Centre, Wageningen, The Netherlands

Statistical Evaluation of the availableEcotoxicology data on plant protection productsand their Metabolites (“SEEM-Project”)

Auteri D, Visentin S, Alberio P, Redolfi E,Azimonti G, Mangiarotti M and Maroni MInternational Center for Pesticides and Health Risk Prevention,Busto Garolfo (Milano), Italy

Evaluation of methylparathion impact infreshwater open mesocosm on biocoenotic andbivalve biomarker responses

Bassères A and Guerniou ATotal, Groupement de Recherches de Lacq, Lacq, France

Chemical-biological monitoring in drainageditches in the orchading region “Altes Land”.1. Residues of a.i. in surface water

Bischoff G1, Stähler M1, Fricke K2

and Pestemer W1

1) Federal Biological Research Center for Agriculture and Forestry,Institute for Ecotoxicology and Ecochemistry in PlantProtection, Berlin-Dahlem and Kleinmachnow, Germany

2) Center for Research and Fruit Consultation for Fruit Growingof the Chamber of Agriculture Hannover, Jork, Germany

Entry of pesticides into surface waters: Resultsof the Lamspringe run-off monitoring project

Bischoff G1, Rodemann B,2 and Pestemer W1

1) Federal Biological Research Center for Agriculture and ForestryInstitute for Ecotoxicology and Ecochemistry in PlantProtection, Berlin-Dahlem, Germany

2) Federal Biological Research Center for Agriculture and Forestry,Institute for Plant Protection in Field Crops and Grassland,Braunschweig, Germany

Monitoring of terbuthylazine in surface watersadjacent to maize fields with potential run-offto prove the efficacy of vegetated buffer zones

Bischoff G1, Pestemer W1, Rodemann B2

and Küchler T3

1) Federal Biological Research Center for Agriculture andForestry Institute for Ecotoxicology and Ecochemistry inPlant Protection, Berlin-Dahlem, Germany

2) Federal Biological Research Center for Agriculture andForestry, Institute for Plant Protection in Field Cropsand Grassland, Braunschweig, Germany

3) Syngenta Agro GmbH, Maintal, Germany

Higher-tier risk assessment of fomesafen andfomesafen-nonylphenol polyethoxylate adjuvantmixture. 1. Fate and effects on planktoniccommunities

Caquet Th1, Lacroix G2, Deydier-Stephan L3, LeRouzic B3, Cravedi J-P4, Lescher-Moutoué F2,Azam D5, Heydorff M1 and Lagadic L1

1) UMR INRA-ENSAR Écobiologie et Qualité des HydrosystèmesContinentaux, Équipe Écotoxicologie et Qualité des Milieux,Rennes, France

2) UMR CNRS Laboratoire d’Écologie de l’École NormaleSupérieure, Paris, France

3) UMR CNRS Université de Rennes 1, ‘Ecobio’, Rennes, France4) UMR INRA-ENSAT/INPT-ENVT ‘Xénobiotiques’, Toulouse, France5) Unité Expérimentale Écologie et Écotoxicologie Aquatique,

INRA, Rennes, France

Higher-tier risk assessment of fomesafenand fomesafen-nonylphenol polyethoxylateadjuvant mixture. 2. Effects onmacroinvertebrate communities

Caquet Th1, Roucaute M1, Heydorff M1,Azam D2 and Lagadic L1


2) Unité Expérimentale Écologie et Écotoxicologie Aquatique,INRA, Rennes, France

Predicted versus observed effectsof pesticides in a paddy rice field

Dohmen GP and Kubitza JBASF-AG, Agricultural Center Limburgerhof, Germany

Predicting population recovery in aquaticsystems using toxicity data combined with bothchemical monitoring and functionaldeterminations

Farrelly EBraddan Scientific Ltd., Harrogate, UK

l i st o f p o st e rp r e s e n tat i o n s

107

CONTENTS

108

a p p e n d i x b

CONTENTS

Assessing pesticide risk on birds: a case study(Parco Agricolo Sud, Milano)

Finizio A1, Verro R1, Sala S1, Auteri D2

and Vighi M1

1) Department of Environmental and Landscape Science,University of Milano Bicocca, Milano, Italy

2) International Center for Pesticides and Health Risk Prevention,Busto Garolfo (Milano), Italy

Development of community toleranceto the antifouling agent Irgarol 1051 in marineperiphyton after years of coastal watercontamination

Grönvall F1, Barceló D2, Blanck H1, Martinez K2

and Nihlén P1

1) Botanical Institute, Göteborg University, Göteborg, Sweden2) Environmental Chemistry, CID-CSIC, Barcelona, Spain

Higher-tier risk assessment of fomesafen andfomesafen-nonylphenol polyethoxylate adjuvantmixture. 3. Changes in individual performancesof the snail Lymnaea stagnalis

Jumel A1*, Coutellec M-A1, 2, Cravedi J-P3

and Lagadic L1


2) UMR ‘Ecobio’ CNRS-Université de Rennes 1, Rennes, France3) UMR 1089 INRA-ENSAT/INPT-ENVT ‘Xénobiotiques’,

Toulouse, France*) Present address: INRA-SCRIBE, Rennes, France

Effect of drift potential ondrift exposure in terrestrial habitats

Koch H, Weißer P, Landfried M and Strub OState Agency for Agronomy and Plant Protection,Mainz, Germany

Probabilistic risk assessment of Mancozebfungicide in beneficial arthropods

Kramer VJ and Miles M Dow AgroSciences, European Development Centre, Abingdon,UK

Multi-tiered assessment of the effects of chemicalmosquito control on invertebrates in coastalwetlands of Morbihan (Brittany, France)

Lagadic L, Caquet Th, Fourcy D*, Jumel A* andHeydorff MUMR INRA-ENSAR Écobiologie et Qualité des HydrosystèmesContinentaux, Équipe Écotoxicologie et Qualité des Milieux,Rennes, France

*) Present address: INRA-SCRIBE, Rennes, France

Linking insecticide contamination andpopulation response in an agricultural stream

Liess M1 and Schulz R2

1) UFZ Centre for Environmental Research, Departmentof Chemical Ecotoxicology, Leipzig, Germany

2) Syngenta AG, Ecological Sciences, Jealott’s Hill InternationalResearch Centre, Bracknell, UK

Population response to toxicantsis altered by intraspecific interaction

Liess MUFZ Centre for Environmental Research, Departmentof Chemical Ecotoxicology, Leipzig, Germany

Effects of pesticides in persistently contaminatedareas – Detoxification enzymes as the endpointsof the pesticide effects in invertebratesinhabiting a gradient of metal pollution

Migula PDepartment of Animal Physiology and Ecotoxicology,University of Silesia, Katowice, Poland

Landscape exposure assessmentof chlorpyrifos in a citrus area

Padovani L, Capri E and Trevisan MIstituto di Chimica Agraria ed Ambientale, Università Cattolicadel Sacro Cuore,Piacenza, Italy

Ecological impact of mixtures of antifoulingagents on marine microalgal communitiesand the prediction of combined effects

Porsbring T1, Arrhenius Å1, Blanck H1,Kuylenstierna M2 and Scholze M3

1) University of Göteborg, Botanical Institute,Department of Plant Physiology, Göteborg, Sweden

2) Kristineberg Marine Research Station,450 34 Fiskebäckskil, Sweden

3) Centre for Toxicology, The School of Pharmacy,University of London, 29/39 Brunswick Square Bloomsbury,London, United Kingdom

A critical review of pesticide relatedmonitoring studies in Germany

Schäfers C1, Hommen U1 and Streloke M2

1) Fraunhofer Institute for Molecular Biology and Applied Ecology,Schmallenberg, Germany

2) Federal Office for Consumer Protection and Food Safety (BVL),Braunschweig, Germany

l i st o f p o st e rp r e s e n tat i o n s

109

CONTENTS

Three-year field study on effects oflambda-cyhalothrin on carabid beetles

Schenke D1, Baier B1 and Heimbach U2

1) Federal Biological Research Center for Agriculture and Forestry,Institute for Ecotoxicology and Ecochemistry in PlantProtection, Berlin-Dahlem, Germany

2) Federal Biological Research Center for Agriculture and Forestry,Institute for Plant Protection in Field Crops and Grassland,Braunschweig, Germany

Modelling toxic pressure in streams –An indicator of the potential aquaticexposure caused by pesticide runoff

Schriever CA, von der Ohe PC and Liess MUFZ Centre for Environmental Research,Department of Chemical Ecotoxicology, Leipzig, Germany

A review of the analysis of morphologicalalterations of individuals as bioindication toolsin aquatic ecosystems: examples withChironomidae (Diptera) larvae

Servia MJ1, Cobo F2 and Gonzalez M2

1) UMR INRA-ENSAR Ecobiologie et Qualité des HydrosystèmesContinentaux, Equipe Ecotoxicologie et Qualité des Milieux,Rennes, France

2) Departamento de Bioloxia Animal, Facultad de Bioloxia,Universidade de Santiago de Compostela, Spain

Effects of short-term climatic variationson fluctuating asymmetry levels in Chironomusriparius larvae at a polluted site

Servia MJ1, Cobo F2 and Gonzalez M2

1) UMR INRA-ENSAR Ecobiologie et Qualité des HydrosystèmesContinentaux, Equipe Ecotoxicologie et Qualité des Milieux,Rennes, France

2) Departamento de Bioloxia Animal, Facultad de Bioloxia,Universidade de Santiago de Compostela, Spain

Field studies in the risk assessmentof atrazine in amphibians

Solomon KR1 and du Preez LH2

1) Centre for Toxicology, University of Guelph, Canada2) School of Environmental Sciences and Development,

Potchefstroom University for CHE, Potchefstroom, South Africa

Approaches for active biologicalmonitoring of pesticides

Stähler M and Pestemer WFederal Biological Research Centre for Agriculture and Forestry,Institute for Ecotoxicology and Ecochemistry in PlantProtection, Kleinmachnow, Germany

Chemical-biological monitoring in drainageditches in the orcharding region “Altes Land”.2. Ecotoxicological evaluation and surveyof aquatic zoocenoses

Süss A, Mueller ACW and Pestemer WFederal Biological Research Centre for Agriculture and Forestry,Institute for Ecotoxicology and Ecochemistry in PlantProtection, Kleinmachnow, Germany

Functional landscape classificationfor the development of a GIS-based local riskassessment with realistic exposure scenarios

Trap M1, Schad T2 and Kubiak R1

1) Ecology Department of State Education & Research CenterNeustadt, Neustadt, Germany

2) Bayer CropScience, Institute for Metabolism andEnvironmental Fate, Monheim, Germany

Relative Sensitivity Distribution (RSD)of aquatic invertebrates to organic andmetal compounds

von der Ohe PC and Liess MUFZ Centre for Environmental Research,Department of Chemical Ecotoxicology, Leipzig, Germany

The effects of the herbicide metsulfuronmethyl on the ecology of macrophyte-dominated experimental ecosystems

Wendt-Rasch L1, Gustavsson K2 and Woin P1

1) Department of Ecology, Chemical Ecology and Ecotoxicology,Lund University, Sweden

2) Department of System Ecology, Stockholm University, Sweden

Environmental risk assessment for plantprotection products in the EU using laboratory,terrestrial model ecosystem and field data

Weyers A1, Sokull-Klüttgen B1, Knacker T2,Martin S3, van Gestel CAM4 and Römbke J2

1) European Chemicals Bureau, Joint Research Centre,Ispra (VA), Italy

2) ECT Oekotoxikologie GmbH, Flörsheim, Germany3) Federal Environmental Agency, Berlin, Germany4) Vrije Universiteit, Amsterdam, The Netherlands

Magnitude and time course of residuesin arthropods as potential food itemsfor terrestrial vertebrates –obtaining data from field experiments

Wolf C, Fuelling O, Giessing B, Kuppels U,Neumann C, Nuesslein F and Wilkens SBayer CropScience, Ecotoxicology, Monheim, Germany

Habitat use and time budgets (PT) of terrestrialvertebrates – hop yards in Bavaria: an exampleof obtaining data from field experiments

Wolf C, Fuelling O, Giessing B and Wilkens SBayer CropScience, Ecotoxicology, Monheim, Germany

110

a p p e n d i x b

CONTENTS


CONTENTS

appendix cREPORTSOF THEBREAKOUTGROUPS

CONTENTS

1 GENERAL COMMENTSABOUT EFFECTS OFPESTICIDES IN THE FIELD

The fact that agriculture per se (including landmanagement, cattle rearing, pesticide use, fertiliz-ers, irrigation, etc.) has an impact on water bodiesin agronomic landscapes is generally acknowl-edged. Well-performed monitoring studies fromthe real world (“the field”) presented during thisworkshop showed that there can be effects on non-target aquatic organisms in the field from pesticideuse. At the workshop the examples presented inplenary were discussed. However, several funda-mental questions remain:� There are relatively few examples of field stud-

ies and there is little information available onhow widespread the described effects are.

� There is often much uncertainty over the mag-nitude of exposure that produces the observedeffects.

� There are difficulties in attributing a particularobservation of effect solely to pesticides, as theeffects could often also be the result of otherstressors (including natural environmental fac-tors) having their effects upon the ecosystem.Documented effects concern mainly insecti-cides and arthropods. Data on other pesticidesand other groups of organisms are scarce orwere not presented/discussed at the workshop.

2 FIELD AND MONITORING STUDIES

2.1 DEFINITION OF FIELD/MONITORING STUDIES

These studies generally fall into two categories:� Field study: an investigation into the impact of

specific products or active substances on a spe-cific water body and introduced artificially forthe purpose of the experiment under controlledconditions;

� Monitoring study: an investigation into theoverall impact of pesticide use (or misuse) onwater bodies

Within the context of this workshop, the focus ismainly on monitoring studies. However, field stud-ies will often provide valuable information for pre-dicting potential effects in the field and for inter-pretation of monitoring data.

Monitoring studies should include both chemi-cal (exposure) and biological (effects) monitoringin order to attempt to establish causality accordingto widely accepted epidemiological criteria. It wasnoted that not all the criteria need to be met at alltimes (e.g., it may be difficult to link pesticideresidues to indirect effects). In order to avoid spu-rious correlations, a combined approach usingexperimentation (such as mesocosm studies)and field observation is often useful to establishcausality, although this may not always be fea-sible.

2.2 WHY DO WE NEED FIELD/MONITORING STUDIES?

There are several reasons why field monitoringstudies may be useful:� They are a reality check for the overall risk as-

sessment process and as such may help to iden-tify areas of concern that are not sufficientlyprotected that require further investigation andalso identify areas where the present processmay be overprotective;

� Measurements in the field are explicitly fo-cused on ecosystem components or processesthat we wish to protect and are made under re-alistic conditions, rather than the surrogatesfor exposure and effect that are used at lowertiers in the pesticide risk assessment frame-work;

� They may help to demonstrate the extent towhich risk mitigation measures really work inthe field, in order to provide confidence tothose who implement them. Conversely, they

r e p o r t s o f t h eb r e a ko u t g r o u p s a qu a t i c e c o s y ste m s

113

Aquatic Ecosystems

CONTENTS

may sometimes demonstrate that these mea-sures are not effective enough and thereforejustify the development of modified mitigationmeasures;

� Repeated field monitoring studies (i.e. at regu-lar time intervals) may give information on thetemporal trends of the environmental impact ofagricultural use of pesticides at a local/regionalscale;

� They can play a role in refinement of the riskassessment and in risk management;

� Data from monitoring studies have the advan-tage of being relatively easy to communicate tothe public and other stakeholders.

� Currently biological monitoring is the onlymeans to identify potential impacts from multi-ple stressors in the field.

Controlled field and outdoor experiments (e.g.,micro and meso-cosms) can be useful predictivetools during the pre-authorization risk assessmentand for post-authorization testing of hypotheses.This contrasts with monitoring studies that areuseful during the post-authorization phase as acheck on pre-authorization predictions, and alsopotentially useful for re-registration purposes.Monitoring results may help to develop more ap-propriate experimental designs (e.g., for commonmixtures of chemicals, and to identify potentialconfounding factors.)

2.3 EXPOSURE MONITORING

Routine monitoring for pesticides and other sub-stances in surface waters is usually performed bywater authorities and regulatory bodies for legisla-tive reasons unconnected with pesticide autho-rization and control. As a result, many of theseprogrammes have poor temporal resolution (e.g.,monthly) in relation to likely pesticide peak con-centrations, with sampling locations more fre-quently found in larger water bodies (e.g., at theinlet of drinking water plants) and biased towardswater-soluble herbicides (which are easily de-tectable with multi-residue methods and mostlikely to exceed drinking water standards). Rou-tine chemical monitoring is often therefore of lim-ited value for the prediction of ecological effects,although it is acknowledged that under the WaterFramework Directive this should be augmented bybiological monitoring for environmental impacts.

The published data from case studies on agricul-tural insecticide exposure in surface waters haverecently been reviewed (Schulz, 2004).

On occasions the pesticide concentrations mea-sured in chemical monitoring programmes differsfrom those predicted by regulatory models (e.g.,(Crane et al., 2003); or see www.bestrijdingsmiddelenatlas.nl). This may be due to various rea-sons such as, � Sampling during a low vulnerability situation

(e.g., in relation to timing of rainfall);� Sampling from different types of water bodies

(flowing rather than static water bodies);� Peak concentrations missed when monitoring

by grab sampling;� Differences in the dissipation of the pesticide in

water in the field compared to that predictedon the basis of laboratory water-sediment stud-ies;

� Point source contamination.� Poor agricultural practice (e.g., misuse, mainly

buffer zone violation) � Conservatism in the initial tiers of the risk as-

sessment process (e.g., “worst-case” 90th per-centile modelled environmental exposure) re-sulting in differences between predicted andmeasured pesticide concentrations.

Well-designed monitoring studies can be used todemonstrate that the risk assessment process issufficiently precautionary in estimating exposurefrom single, or many, pesticide applications in thesame catchment. Targeted chemical monitoring isan appropriate means of refining exposure andcould be used to validate exposure models. For ex-ample, targeted catchment studies can be used tocalibrate models that include product use data andlandscape information, which can then be used forgeneric exposure assessment (e.g., the PESTSURFdevelopment in Denmark). However, chemicalmonitoring data on their own are of limited usewithout establishing a correlation between possi-ble chemical causes and biological effects. Ifcausality can be established, then it may be possi-ble to examine the correlation between critical ef-fect concentrations detected in the field and thosepredicted from exposure estimates under Directive91/414.

a qu a t i c e c o s y ste m s

114

a p p e n d i x c

CONTENTS



2.4 EFFECTS MONITORING

Biological monitoring studies are useful for inves-tigating the effects of multiple stressors, as organ-isms can integrate these effects. However, it is im-portant to identify:� What it is possible to observe and measure in

the field; and � Whether it is possible to discriminate the im-

pacts of individual stressors.

A major problem in environmental monitoring isthe variability caused by the multitude of stressorsinfluencing the specific local and temporal ecosys-tem. Thus, due to the “noise” in the data the powerto detect differences can be low. Despite this, stud-ies by Leonard et al. in Australia (Leonard et al.,1999; Leonard et al., 2000), Liess (platform pre-sentation), Liess and Schulz in Germany (Liess,1994, 1998; Liess & Schulz, 1999; Liess et al., 1999;Schulz & Liess, 1999a, 1999b), Schulz et al. inSouth Africa (Schulz et al., 2001; Schulz et al.,2002), and in Argentina (Jergentz et al., 2004) andSchaefers et al. in Germany (Altes Land; Schaefers,platform presentation) demonstrate that respons-es can be detected from within the variability ofthe data. (Schulz, 2004) has critically reviewedthe available field studies on insecticide effects insurface waters (see Table 1).

It is likely that monitoring studies may not de-tect small and subtle effects, so it is importantwhen designing a study to understand what onewishes to measure (and protect) and the level ofdifference that it is necessary to detect. Such deci-sions may be informed by taking species life-histo-ry strategies (e.g., r versus K-selected, short versuslong generation times; Liess, platform presenta-tion) and potential species or ecosystem sensitivityinto account.

2.5 LIMITATIONS OF AVAILABLE DATA

The studies presented at the workshop concentrat-ed on insecticides and macroinvertebrates, so gen-eralizations to other organism types or compoundgroups are difficult. The number of focused, suit-able monitoring data is rather low and some thesestudies had limitations (e.g., absence of chemicalmonitoring, or no clear evidence of cause and ef-fect). There is also evidence of macrophyte loss

from the agricultural landscape (Williams et al.,1998).

3 WHAT FEATURES ARE ESSENTIALIN A SUCCESSFUL FIELD/MONITORING STUDY AND WHY?

Monitoring studies should include both chemical(exposure) and biological (effects) monitoring inorder to attempt to establish causality.

3.1 QUANTIFYINGENVIRONMENTAL VARIABLES

3.1.1 PROBLEMS INQUANTIFYING EXPOSURE

One potential issue in interpreting monitoringdata is that exposure may be poorly characterized.Field measurements should be made relative to ap-plication time to detect the peak concentrationsand to aid interpretation of observed biological ef-fects. For rainfall-driven inputs (runoff and drain-flow) it is important to sample at the start of thehydrogeological event to capture the peak pesti-cide concentrations. For drift inputs, it is neces-sary to sample at the time of application.

Effects arising from mixtures require a more com-plex exposure monitoring design that can detectmultiple chemicals. As complete a knowledge aspossible of exposure concentrations is desirableand also of other factors associated with landscapestructure (e.g., river basin structure) and agricul-tural practice (e.g., fertilizer application, land usepractices, etc.) which may confound effective in-terpretation. It is important to measure these po-tential causal factors as well as pesticide concen-trations so that their role can be assessed. Estab-lishing cause and effect may be difficult, and theuse of additional experimentation (e.g., laboratoryor in situ bioassays, or mesocosms) may be usefulto validate monitoring data. Controlled experi-mental studies in combination with monitoringstudies can often help to test hypotheses and to in-terpret field data.

In some circumstances validated/verified modelestimation may be more appropriate than inade-quate measurement, e.g., for rapidly dissipating


115

CONTENTS

substances. However, a problem with reliance onmodelling alone or on targeted chemical analysesis that legacy pesticides and undetected, yet toxi-cologically important, substances may be missed.Nevertheless there is the potential to combinemodelling with monitoring data to provide furtherinsights into quantifying exposure.

3.1.2 POINT VERSUS DIFFUSE SOURCES

There may be a poor correlation between pointsource and diffuse pesticide contamination, assome of the exposure to pesticides detected byfield monitoring may arise from point sources(e.g., waste water treatment works, or impropercleaning of equipment in the farmyard) in addi-tion to field applications. As a result, effects ob-served in a monitoring study may not necessarilyinvalidate a single-substance risk assessment.

Comparison of effects observed in field moni-toring studies with the risk assessment for an ac-tive substance or product is difficult because, ingeneral, the field exposure is insufficiently charac-terized (e.g., it may miss the peak contamination,potential sample deterioration during storage ormatrix interference and presence of undetectedcontaminants or those exerting biological effectsbelow analytical detection levels).

Risk assessment of pesticides currently does notconsider point sources because under Directive91/414 only product use according to Good Agri-cultural Practice (GAP) is considered. ModelledPECs therefore currently do not consider exposurefrom poor agricultural practice. However, thisissue needs to be kept in perspective as, for exam-ple, chemical monitoring data from the Nether-lands indicate that only a few chemicals exceedlower and/or higher-tier maximum permissibleconcentrations (see: www.bestrijdingsmiddelenatlas.nl) and these may be associated withunknown sources. In addition, experience fromGermany with chemical monitoring in groundwa-ter indicates that many cases of exceedences werefrom point source contamination (Bach et al.,2001; Bach et al., 2000; Muller et al., 2002). How-ever, the limitations of routine chemical monitor-ing apply (see 2.3).

3.1.3 TEMPORAL AND SPATIAL SCALES

The scale of monitoring and observation needs tobe considered. At a smaller scale (e.g., small wa-terbodies) there may be high variability, so suffi-cient sites and times of sampling should be used todistinguish cause and effect relationships with ad-equate statistical power. Similarly, at a larger scalelocalized impacts may not be detected.

3.1.4 SITE-SPECIFIC BASELINEINFORMATION

Site-specific, context-dependent and ecologicallyimportant biological, physical, and chemical fac-tors (e.g., water chemistry, habitat, and site histo-ry) that could be confounding variables should bemeasured in field monitoring programmes. Thismay include pesticide metabolites and degrada-tion products with known toxicological signifi-cance.

The use of long-term databases on communitycomposition at a regional scale (e.g., specieschecklists) may also give an opportunity to detectchanges in regional biodiversity in a particulararea. However, without additional regional infor-mation (e.g., changes in water management, landuse, etc.) elucidation of causality and robust inter-pretation may be compromised.

3.2 QUANTIFYINGBIOLOGICAL RESPONSE

3.2.1 DETECTING RESPONSES

Loss of species richness (or loss at another taxo-nomic level) was considered to provide strong evi-dence of an adverse biological effect in a surfacewaterbody. However, species should be grouped, ifpossible, according to their:� life cycle traits to account for differences in re-

covery/recolonization potential;� level of tolerance to particular pesticide modes

of action (the percentage of species within thecommunity that are sensitive to the particulartoxicant should then be estimated);

� the species at risk (SPEAR) concept developedby Liess (see case study in this report) was con-sidered a potentially valuable tool.


116

a p p e n d i x c

CONTENTS



Some variability in measured ecological effectsmay be accounted for by categorizing species ac-cording to ecological traits (e.g., sensitivity to pes-ticides, reproduction potential, resilient life stages,high O2 requirements, voltinism, microhabitat,etc.), and then comparing effects in these groupsacross different exposure concentrations.

Similarly, categorizing habitats with referenceto the organisms within them (e.g., organisms as-sociated with oxygen-rich microhabitats) may helpto elucidate the effects of other stressors and thusisolate possible effects of pesticides. Functionalendpoints should also be considered but are usual-ly less sensitive than structural ones.

Endpoints for which tolerance or resistance hasbeen shown to have developed should be avoided.The presence and abundance of sensitive speciesin natural, functioning ecosystems was considereda reliable endpoint. For example, a study in north-ern Germany (Sönnichsen, 2002) of ephemeralditches found no differences in communities be-tween ditches within arable areas compared tothose alongside meadows. This was attributed tothe high recolonization potential of the systemsand, possibly, the ability of the specific communityto withstand periodic drought. Species with a lowrecovery/recolonization potential were not found.Similar hypotheses were proposed by Lagadic toexplain the absence of effects of insecticides on in-vertebrate communities in coastal wetlands sub-jected to chemical mosquito control (Lagadic,1999).

3.2.2 REPRESENTATIVENESSOF THE SYSTEM

The system that is monitored should be represen-tative of the agricultural landscape. However, thepresence of extreme disturbance (e.g., very fre-quent maintenance, intense chemical stress) has tobe avoided. The system should have the capacityto respond to stressors, and representatives of themain taxa and of the different life cycle strategies(multi- to semivoltine species, r- and K-species)should be present in sufficient numbers for ade-quate statistical analysis. Other important biologi-cal issues like, for example, the number of genera-tions should also be considered. The normal rangeof variability in measured endpoints should be un-

derstood for both the measured system and forthose that it is intended to represent.

The challenge is in defining such representative-ness, but there are examples in the UK where clas-sification of water bodies has been systematicallyachieved (e.g., UK aquatic landscapes project:(Anonymous, 2003).

3.2.3 CONTROL AND REFERENCE SITES

A key issue for all field monitoring studies is estab-lishing the control or reference site. Often there isno “control” site, and water bodies can containadapted populations due to stress from pesticidesor other contaminants. It may be easier to find ref-erence sites in flowing water systems (e.g., compa-rable pristine upstream parts of the contaminatedwaterbody), although it was acknowledged thatcommunities may be inherently different at differ-ent sites for reasons other than pesticide use.

Establishing a “true” uncontaminated referencesite is probably unachievable and therefore findinga reference site with a clearly different exposureregime from the test site is required. One possibili-ty might be to use “relative” references, i.e. siteswith a gradient of contamination, from the lowestto the highest measured levels. However pollutioninduced community tolerance (PICT) caused byother stressors should be avoided in the testedsites and in the control sites, respectively. Refer-ence sites should be included in either time orspace (it is acknowledged that this may be verydifficult), although using an ‘internal’ reference inwhich comparisons at the same site are madethrough time may be appropriate.

To establish a control, the intensity of anthro-pogenic activities (i.e., the degree of disturbance)could be used as an indicator to aid site selection.Investigators should not be too concerned if waterbodies have been modified by anthropogenic activ-ities rather than being truly natural, as the formerhave many of the properties of truly natural sys-tems (e.g., certain types of ditches may simulatethe properties of natural water bodies). Referencesites may be better at a more integrated level(community tolerance as opposed to species abun-dance, ecosystem functioning, etc.).


117

CONTENTS

In places where there are other recognized im-pacts occurring (e.g., elevated nutrient concentra-tions) it will usually be impossible to demonstrateconclusive correlations between observed impactsand individual stressors.

3.2.4 CONFOUNDING VARIABLES

It is important to establish what the potential mul-tiple stressors in a system are before beginning astudy so that they can be accounted for in thestudy design and interpretation. Confoundingvariables (e.g., unusual weather or fluctuating or-ganism populations) cannot be completely avoidedin field studies, as they are part of the natural sys-tems; however, they should be monitored in orderto aid the interpretation of results. Differences be-tween reference and treatment sites (e.g., neigh-bouring cultures, shading, differences in upstreamareas, stream or pond morphology, nutrient situa-tion) should be avoided. Routes of exposure exam-ined in laboratory and monitoring studies shouldalso be understood before a study begins so thatrelationships between laboratory and field obser-vations can be interpreted. This may be easier toachieve for short-term effects (e.g., lethality) thanfor long-term effects, because exposure to stres-sors (including pesticides) and the resulting ef-fects are more difficult to characterize over longertime periods. Hence, the statistical power of fieldtests is generally less than for mesocosm studiesbecause of greater variability and more confound-ing factors.

Biological effects observed in the field could resultfrom multiple stresses, which may make determin-ing pesticide exposure as the cause very difficult.On the other hand, the absence of observed effectscould be due to pollution induced community tol-erance (PICT) or to poor statistical power. There isa need for consensus on what kinds of effect moni-toring studies are capable of detecting.

4 WHAT FEATURES ARE DESIRABLEIN A FIELD/MONITORING STUDY?

Monitoring studies should include both chemical(exposure) and biological (effects) monitoring inorder to attempt to establish causality. Additional-ly, in order to avoid spurious correlations, a com-

bined approach using experimentation (in situstudies, biomarkers or mesocosms) and field ob-servation is often useful to establish causality.However, a problem with field studies based on insitu bioassays is that they tend to be more sensitivebecause of increased exposure in cages and lack ofescape opportunities. (e.g., (Schulz, 2003) forGammarus pulex; (Lagadic et al., 2002) for chi-ronomids). Despite this, in situ bioassays can beuseful systems for assessing potential bioavailabili-ty and exposure, particularly when natural popu-lations show variable demographics (perhaps dueto seasonality).

Sub-organism biomarkers may provide a good in-dicator of ecosystem stress in general (De Coen &Janssen, 2003; Hanson & Lagadic, 2003; Sibley etal., 2000). However, difficulties in attributing a re-sponse to a single toxicant may make results frombiomarkers difficult to interpret. Therefore, selec-tion of biomarkers on the basis of the mode-of-ac-tion of pesticides is essential in order to attempt todiscriminate between various stressors (Lagadic etal., 2002). In addition, difficulties in extrapolatingfrom biomarker responses to population level end-points have led some to question their relevance.The selection of biomarkers should be based ontheir physiological role, and more specifically ontheir metabolic implication in individual perfor-mances (growth, reproduction, etc.) (Liess &Schulz, 1999).

5 HOW SHOULD FIELD MONITORINGDATA BE INCORPORATEDINTO A RISK ASSESSMENT?

The general consensus was that it is difficult to en-visage monitoring studies being incorporated intodecision making with respect to registration for asingle active substance. However, it could be usedfor fine-tuning potential mitigation measures inpost-registration activities.

The principal benefit of monitoring data withinthe current risk assessment framework is as acheck on the pre-authorization assessment. Therisk assessment considers single compounds in iso-lation, but there is uncertainty over whether or notit is also conservative for multiple stresses (includ-ing multiple pesticide exposure) that may occur inthe field.


118

a p p e n d i x c

CONTENTS

In one field study (Liess, platform presentation), itwas shown that the most toxic substance (highestToxic Units) gave the best correlation betweencontamination and community impact, while forall contaminants, the additive exposure concentra-tions gave a poorer correlation. This would sup-port the view that the single chemical risk assess-ment is sufficient for the pesticides, habitat typeand ecological receptors examined in this study.

This view is supported by controlled microcosm/mesocosm experiments in which the impact of arealistic exposure to pesticides used in, e.g., tulipand potato fields was simulated (Arts and Brock,presentations; Van Wijngaarden et al., 2005).

The need for a clear definition of the protectionaims underlying risk assessment for pesticides wasidentified. One approach may be to establish a ref-erence condition (i.e., the “natural” status) thatfulfils the protection aims for water bodies. How-ever reference locations may be difficult to locatein nations with historically modified surface wa-ters (i.e., most developed countries). Definition ofthe reference condition may therefore be simply apragmatic decision based upon a combination ofscientific judgement, ‘public acceptability’ and evi-dence-based determination of ‘good quality’ inagricultural landscapes. There are examples avail-able of such definitions for aquatic communities inDutch drainage ditches in the Netherlands (Ni-jboer, 2000; Nijboer et al., 2004).

6 HOW SHOULD POPULATIONRECOVERY BE INTERPRETED,ESPECIALLY FROMUNCONTAMINATED AREAS?

6.1 GENERAL

Recovery is part of the description of how naturalsystems respond to a stressor and should be con-sidered within the context of both specific land-scape factors and extrapolation to wider land-scapes. The term recovery includes both internalsystem recovery and recolonization from externalsources. Assessing recovery requires considerableknowledge about different landscape factors andhow they interact with organism life histories,

although many questions remain in this area ofecology:

� How does recovery differ across systems andlandscapes?

� There are some typical ecosystems in the agri-cultural landscape – is it acceptable to takethese as baselines?

� Is the possibility or probability of recovery re-duced if a large part of the landscape is ad-versely affected by pesticide exposure?

� What percentage of an interconnected ditchsystem should be protected to facilitate recolo-nization?

Measurement of “recovery” in monitoring studiesmay be difficult because removal of keystonespecies could lead to alternative stable states (e.g.,macrophyte-dominated versus algal-dominatedaquatic systems (Scheffer, 1998)). Also, (Matthewset al., 1996) have suggested that ecosystems pos-sess a ‘memory’ of stressful events and that there isa lack of return to original conditions after chemi-cal exposure (i.e., irreversible effects). If an effecthas been of sufficient magnitude to alter theecosystem irreversibly, it would probably be con-sidered an unwanted effect even if there wasecosystem stability.

Whether populations are r or K-selected will clear-ly influence the speed and capacity for system re-covery (as shown in Liess and Brock, platform pre-sentations). Adaptation to stress may not be achange to less sensitive species, but to thosespecies that have better recovery/recolonizationpotential (e.g., shorter generation time and aeriallife stage). For example, in the field study present-ed by Liess the main recolonization observed inthe field was in-stream transport by organism driftfrom small, forested upstream areas. This suggeststhat the area required as a source of recolonizationmay not need to be large to have a significant posi-tive effect.

Recovery in flowing systems might be faster thanin isolated water bodies, and it is possible to makereasonable predictions for interconnected waterbodies that recovery will usually occur within 1-2years, even from substantial natural perturbations,provided the stressor is removed. It is much harderto make predictions for isolated water bodies, asthe recovery processes in these are poorly under-stood. However, as there will often be pesticideconcentration gradients within larger water bod-


119

CONTENTS

ies, there will frequently be refugia and thus re-covery potential from within the system.

In summary, it is important to define recoveryendpoints a priori and, when considering recov-ery, monitoring studies should attempt to distin-guish and quantify both internal recovery and re-colonization by dispersal. This may be very diffi-cult as it requires incorporation of site-specificpopulation dynamics, life history information foraffected populations, and metapopulation dynam-ics in the landscape.

6.2 MODELLING RECOVERY

Few data are available about recolonization (e.g.,reproduction, drift, and dispersal) that could beused in population models. However, data on theprinciples of aquatic organism population recoveryare available. Using very simple models it shouldbe possible to differentiate between landscapeswith isolated aquatic systems and those with inter-connected and flowing water bodies For example,data on recovery potential of organisms are avail-able at the family level on the PondFX website(www.ent3.orst.edu/PondFX). Data at the specieslevel are being collected in a cooperative projectbetween The Ponds Conservation Trust, UFZ Cen-tre for Environmental Research and Alterra. Theuse of relatively simple recolonization/popula-tions models also have the potential to assist in hy-pothesis testing.

It is therefore possible to use generalised informa-tion on recolonization statistics for differentspecies to generate hypotheses and design experi-ments. Modelling and landscape analysis can beused to identify water bodies that are most likelyto be impacted and then targeted monitoring canbe undertaken to investigate whether or not theimpacts are observed.

6.3 RECOVERY AND RISK ASSESSMENT

Knowledge about population recovery can help usto understand how observed recovery in meso-cosms should be interpreted within a wider envi-ronmental context. Despite its importance, recov-ery is currently not generally considered in the ini-tial tiers of the risk assessment process, although

at higher tiers it is considered (e.g., NOEAEC;(91/414/EEC, 2002; Anonymous, 2002). There arealso conceptual difficulties in incorporating recol-onization into a risk assessment, as only singlestressors are considered.

The FOCUS Landscapes & Mitigation workinggroup is attempting to develop ecological informa-tion associated with the established surface waterscenarios to inform risk assessment.

6.4 MITIGATION

The evidence from the Altes Land study presentedin this report demonstrated differing impacts withchanges in spray direction, thus indicating the po-tential benefits of effective risk mitigation. Fur-thermore, Liess’s study in this report also demon-strated that relatively small areas in streams thatdo not suffer impacts (e.g., a forested area) cansignificantly mitigate impacts in connected waterbodies. In addition, the workshop also heard of aGerman terrestrial mitigation system (using GISinformation). Although this was designed for a dif-ferent aim (the proportion of the landscape cov-ered by woodland scrub was correlated with thepotential for recolonization), the principle and useof GIS could be useful in assessing or refining ap-proaches to mitigation in the aquatic environment.

7 WHAT GENERAL CONCLUSIONS CANBE DRAWN ABOUT THE LEVEL OFAGREEMENT BETWEEN LABORATORYBASED RISK ASSESSMENT AND FIELDEFFECTS.

Specific targeted experiments do show effects ofpesticides in the field. This has been shown in theAltes Land (see case study in this report) in caseswhere spraying was at a short distance from and inthe direction of the ditch, in the Schulz studiesfrom South Africa (where the exposure was con-sistent with predictions from spray drift tables asused in the EU), and in the Liess and Schulz stud-ies following run-off events. For various reasons(demonstration of causality, or poor characterisa-tion of exposure), it is not always possible to corre-late observed biological effects with the risk as-sessment process (e.g., uncertainty over actualfield exposure). Biological impacts observed in


120

a p p e n d i x c

CONTENTS

www.ent3.orst.edu/PondFX

monitoring could be used to indicate the need forfurther investigation. When the “actual” exposureis well understood, monitoring studies do not un-derestimate the risk.

Mesocosm studies may provide fairly good effectspredictions for known exposures (e.g., those com-pounds or other stressors that are experimentallyapplied), but these predictions are less robustwhen exposure is more complicated (e.g., un-known compounds or stressors present). They areprobably also predictive of direct effects, but as in-direct effects are usually context-dependent it isdifficult to be sure that generic tests such as meso-cosms are protective in all cases.

Controlled experiments are predictive of thresholdeffects concentrations, but recovery in the field isoften context-dependent (e.g., voltinism in theparticular assemblage, hydrological connectivity,latitude, etc.). Insertion of species into mesocosms(e.g., univoltine species) may help us to under-stand potential recovery. However such datashould be carefully considered since they are notfully representative of natural conditions wherethe time lag between contamination and recolo-nization may be much greater compared to the ex-perimental test duration.

Monitoring data may help us to design bettergeneric mesocosm studies (including inter-labora-tory mesocosm ring tests) or, at least, to under-stand the limitations of specific mesocosm de-signs. For example, from monitoring we may dis-cover important differences in the timing of algalblooms, generation time, etc., across differenthabitats or geographical regions, which must betaken into account in pesticide approval or re-reg-istration.

8 WHAT FURTHER RESEARCH ISREQUIRED TO IMPROVE THECONDUCT AND INTERPRETATIONOF A FIELD/MONITORING STUDYFOR RISK ASSESSMENT?

� More well-designed field studies are required.They should include monitoring studies with awider geographic distribution to account fordifferences in routes of exposure, life cycleecology, etc. (Examples to date have been from

Northern Europe; recovery may be different indifferent climates, but pest/disease pressuremay also be different.)

� If field studies reveal correlations between pes-ticide exposure and biological effects thenthese should be regarded as hypotheses. Subse-quently mesocosm or other focused studies canthen be used for hypothesis-testing.

� Additional taxonomic, life history, genetic andother ecological information (e.g., regional dif-ferences in structure and function of communi-ties) to generate and interpret results.

� Better understanding of the mechanisms thatunderpin species sensitivities. For example,there was an interesting contrast in the sensi-tivities of molluscs between the Altes Land re-sults (potentially quite sensitive) and foreststream results (insensitive) presented at theworkshop and summarised in this report.

� Additional information on landscape and othergeographical factors and development of ap-propriate population models. This may benefitfrom integration of GIS data.

� Studies intermediate between mesocosm andlarge scale monitoring studies: a need to re-solve causality at an intermediate level and toaddress issues such as multiple stressors (in-cluding natural environmental factors).Methodological guidance on this (e.g., qualitycriteria) may be required.

� Studies of recovery: better understanding ofthe mechanisms associated with recovery, in-cluding generating data that can be used to pa-rameterise landscape/metapopulation models.

� Modelling of landscape scale impacts to helpaddress the question of ‘What magnitude of im-pact is significant?’

� Experiments in which stressors are deliberatelyadded to previously minimally impaired loca-tions, and in which stressors are selectively re-moved (if possible) from locations in whichpesticides are used.

R e f e r e n c e sAnonymous. 2002. SANCO Guidance Document

on Aquatic Ecotoxicology in the frame of theDirective 91/414/EEC.

Anonymous. 2003. Aquatic ecosystems in the UKagricultural landscape. The Ponds Conserva-tion Trust, Oxford Brookes University, GipsyLane, Headington, Oxford, OX3 0BP.


121

CONTENTS

Bach M, Huber A, Frede HG. 2001. Input path-ways and river load of pesticides in Germany –a national scale modeling assessment. WaterScience and Technology, 43, 261–268.

Bach M, Huber A, Frede HG, Mohaupt V,Zullei-Seibert N. 2000. Schätzung der Ein-träge von Pflanzenschutzmitteln aus der Land-wirtschaft in die Oberflächengewässer Deutsch-lands Schmidt-Verlag, Berlin.

Crane M, Whitehouse P, Comber S, Watts C,Giddings J, Moore DRJ, Grist E. 2003.Evaluation of probabilistic risk assessment ofpesticides in the UK: chlorpyrifos use on topfruit. Pest Management Science, 59, 512–526.

De Coen WM, Janssen CR. 2003. The missingbiomarker link: Relationships between effectson the cellular energy allocation biomarker oftoxicant-stressed Daphnia magna and corre-sponding population characteristics. Environ-mental Toxicology and Chemistry, 22, 1632–1641.

Hanson ML, Lagadic L. 2003. Chitobiase activityas an indicator of aquatic ecosystem health.SETAC Globe, 4, 37–38.

Jergentz S, Mugni H, Bonetto C, Schulz R.2004. Runoff-related endosulfan contamina-tion and aquatic macroinvertebrate response inrural basins near Buenos Aires, Argentina.Archives of Environmental Contamination andToxicology, 46, 345–352.

Lagadic L. 1999. Biomarkers in Invertebrates.Evaluating the effects of chemicals on popula-tions and communities from biochemical andphysiological changes in individuals. In Bio-markers: A Pragmatic Basis for Remediation ofSevere Pollution in Eastern Europe. (ed D.B.Peakall, Walker, C.H. & Migula, P.), Vol. Series.2, Vol. 54, pp. 153-175. Kluwer academic pub-lishers, Dordrecht, The Netherlands.

Lagadic L, Caquet T, Fourcy D, Heydorff M.2002. Évaluation à long terme des effets de ladémoustication dans le Morbihan. Suivi del’impact écotoxicologique des traitements surles invertébrés aquatiques entre 1998 et 2001.Convention de Recherche Conseil Général duMorbihan.

Leonard AW, Hyne RV, Lim RP, Chapman JC.1999. Effect of endosulfan runoff from cottonfields on macroinvertebrates in the NamoiRiver. Ecotoxicology and Environmental Safety,42, 125–134.

Leonard AW, Hyne RV, Lim RP, Pablo F,Van den Brink PJ. 2000. Riverine endosulfanconcentrations in the Namoi River, Australia:Link to cotton field runoff and macroinverte-brate population densities. Environmental Toxi-cology and Chemistry, 19, 1540–1551.

Liess M. 1994. Pesticide impact on macroinverte-brate communities of running waters in agri-cultural ecosystems. Verh Internat Verein Lim-nol; Proceedings of the International Associationof Theoretical and Applied Limnology,2060–2062.

Liess M. 1998. Significance of agricultural pesti-cides on stream macroinvertebrate communi-ties. Verh Internat Verein Limnol; Proceedings ofthe International Association of Theoretical andApplied Limnology, 26, 1245–1249.

Liess M, Schulz R. 1999. Linking insecticide con-tamination and population response in an agri-cultural stream. Environmental Toxicology andChemistry, 18, 1948–1955.

Liess M, Schulz R, Liess MHD, Rother B,Kreuzig R. 1999. Determination of insecticidecontamination in agricultural headwaterstreams. Water Research, 33, 239–247.

Matthews RA, Landis WG, Matthews GB.1996. The community conditioning hypothesisand its application to environmental toxicology.Environmental Toxicology and Chemistry, 15,597–603.

Muller K, Bach M, Hartmann H, Spiteller M,Frede HG. 2002. Point- and non point-sourcepesticide contamination in the Zwester Ohmcatchment, Germany. Journal of EnvironmentalQuality, 31, 309–318.

Nijboer R. 2000. Natural communities of Dutchinland aquatic ecosystems., Rep. No. ReportEC-LNV no AS-06, Wageningen.

Nijboer RC, Johnson RK, Verdonschot PFM,Sommerhauser M, Buffagni A. 2004. Estab-lishing reference conditions for Europeanstreams. Hydrobiologia, 516, 91–105.

Scheffer M. 1998. Ecology of Shallow Lakes Chap-man & Hall, London.

Schulz R. 2003. Using a freshwater amphipod insitu bioassay as a sensitive tool to detect pesti-cide effects in the field. Environmental Toxicolo-gy and Chemistry, 22, 1172–1176.

Schulz R. 2004. Field studies on exposure, ef-fects, and risk mitigation of aquatic nonpoint-source insecticide pollution: A review. Journalof Environmental Quality, 33, 419–448.


122

a p p e n d i x c

CONTENTS

Schulz R, Liess M. 1999a. Validity and ecologicalrelevance of an active in situ bioassay usingGammarus pulex and Limnephilus lunatus. En-vironmental Toxicology and Chemistry, 18,2243–2250.

Schulz R, Liess M. 1999b. A field study of the ef-fects of agriculturally derived insecticide inputon stream macroinvertebrate dynamics. Aquat-ic Toxicology, 46, 155–176.

Schulz R, Peall SKC, Dabrowski JM,Reinecke AJ. 2001. Spray deposition of two in-secticides into surface waters in a SouthAfrican orchard area. Journal of EnvironmentalQuality, 30, 814–822.

Schulz R, Thiere G, Dabrowski JM. 2002.A combined microcosm and field approach toevaluate the aquatic toxicity of azinphosmethylto stream communities. Environmental Toxicol-ogy and Chemistry, 21, 2172–2178.

Sibley PK, Chappel MJ, George TK, SolomonKR, Liber K. 2000. Integrating effects of stres-sors across levels of biological organization: Ex-ample using organophosphorus insecticidemixtures in field-level exposures. Journal ofEcosystem Stress and Recovery, 7, 117–130.

Sönnichsen H. 2002. Untersuchungen über dieBedeutung von Flora und Fauna der Parzellen-gräben in ackerbaulich (landwirtschaftlich)genutzten Flächen auf der Insel Nordstrand,Braunschweig.

Williams PJ, Biggs J, Barr CJ, Cummins CP,Gillespie MK, Rich TCG, Baker A, Baker J,Beesley J, Corfield A, Dobson D, Culling AS,Fox G, Howard DC, Luursema K, Rich M, Sam-son D, Scott WA, White R, Whitfield M. 1998.Lowland Pond Survey 1996. Department of the En-vironment, Transport and the Region, London.


123

CONTENTS

1 TO WHAT EXTENT AREEFFECTS OF PPP SEEN UNDERFIELD CONDITIONS?

The group discussed both dedicated field trials(controlled experiments conducted with adequatecontrols and often reference treatments in repli-cated plots) and monitoring (planned or un-planned observations of effects occurring in thefield or treated area as a consequence of the use ormisuse of pesticides). A short discussion was heldto clarify the difference between these two activi-ties, and the essential and desirable features ofstudies of either kind were listed separately.The group surveyed some of the general featuresthat characterise good field trials and monitoringstudies before going into specific aspects (Table 1).These general features were summarized by AndyHart, as follows:� Ecological relevance: endpoints measured

should reflect the major ecological propertiesof the system

� Representativeness: the study system should betypical with respect to climatic conditions, forthe pesticides, biological properties, geograph-ic position and agricultural systems.

� Bias/accuracy: the study should be realistic andthe measurements should accurately reflect thetrue structure and function of the system

� Causality: the observations should be designedso as to reveal wherever possible the mecha-nisms underlying their occurrence.

� Detectability: the measurements should bedone with sufficient replication and power soas to maximize the likelihood of observing ef-fects if any.

The group discussed the introductory lecturesgiven in the morning with the aim of drawinglessons from them, which would be relevant forthe issue of terrestrial invertebrates and plants.

Crucial points of the presentations:� Clook paper. The UK wildlife incidence scheme

seems to work adequately for birds and mam-mals and to a certain extent also for bees. Ingeneral they provide the most useful informa-tion for species of larger body mass and of per-ceived functional/economic importance. Themost useful aspect of these schemes is to raisepublic awareness of adverse effects on non-tar-get species.

� A honeybee monitoring system is also opera-tional in Germany.

� No such system is effective for below-groundorganisms or non-target arthropods (NTA).

� The group discussed the possible application ofthe honeybee scheme to other NTAs? Effects-exposure assessment for bees is possible but forother species this is not considered possible (orvery difficult), because developed and acceptedschemes are lacking; moreover multiple com-pounds are used, more than one species mustbe observed and the time scale when they areapplied is very diverse. It is difficult to finddead individuals of many NTA’s (methodologi-cal problems).

� Monitoring of bees is mainly done to providecentralized data for incident reporting and forregulatory bodies appraisal of wider scale ef-fects.

� Holland paper. Difference in sensitivity to pesti-cides between species within the same taxo-nomic group is often bigger than between dif-ferent taxonomic groups: so how are we goingto deal with this? We need to have data onmany species not only on single ones. This iscollaborated by the view that the diversity ofarthropod species may be greater in the fieldedges than in the field and concentrating onhigher taxa may not give the full impact. Thepower of monitoring schemes is greatly in-creased when it takes whole communities orguilds into account, because this maximises thelikelihood of detecting effects. However, ac-cording to ESCORT 2 field trials for arthropods

te r re st r i a l i n ve r te b ra te s a n d p l a n t s

124

a p p e n d i x c

TerrestrialInvertebrates and Plants

CONTENTS

should focus on key/relevant NTA species or onspecies where issues are predicted. Need fortaxonomic tools for identification. Lack of lifehistory and generic data on species. Strength ofmonitoring data is the availability of long-termdata (trend analyses possible). Cause-relation-ship is sometimes difficult to establish. Onlyone sample time per year could eventually beproblematic as indicated by the effect of inver-tebrate life stage on grey partridge.

� De Jong paper. There was a large amount ofvariability among the plots before treatment.Random assignment of treatments to plotsseems to be problematic. The group discussedthat, in general, information on life-historystrategies, phenology and habitat requirements

of wild plants are readily available, which is anadvantage of the use of plant in monitoringschemes. How important are buffer zones? Thedata presented showed how important bufferzones were in reducing toxic effects of herbi-cides on NTP. From the results it is clear thatshort term effects were seen on growth andphytoxicity symptoms at low exposure ratesand that full recovery occurred. Are we under-estimating effects of herbicides on plant feed-ers? In- or off-crop? Unsprayed field marginsare important for diversity and refuge. More-over, margins are still a very important reser-voir of invertebrates.

In the context of regulatory testing protocols in re-spect of non- crop plant species, recent findings by

r e p o r t s o f t h eb r e a ko u t g r o u p s te r re st r i a l i n ve r te b ra te s a n d p l a n t s

125

CONTENTS

Features Field Monitoring

Clear Objectives E E

Definition of endpoints E E

Pre-treatment observations E D

Replications E 1 D

Control E D 2

Reference item D 3 Na

Exposure assessment D E

Statistics E D

Duration Shorter typically 1 crop season or one year Longer generally several years

Area/plot size/location E 4 D

No. of observation high D D

Representatives / Geography E E

Meteorological data E E

Recovery included D D

High taxonomic resolution E E

Sampling methods E E

Surrounding E E

Functional endpoints in addition tostructural endpoints

D D

GAP E D

Table 1: Essential or desirable features in a successful field/monitoring studyE = essential feature; D = desirable feature; Na = not applicable1) Replication within experiment2) Following options are possible: a) nearby site(s) differing only in potential impact but not in site properties (e.g. cli-

mate, vegetation, soil type) – ideal but in reality difficult to find. b) “virtual” reference site(s), derived from a number of“undisturbed” sites or from time series, linking site properties with species distribution – this approach worked in lim-nology and plant sociology, see Ratte presentation. When an internal control is used (“before – after”-approaches), itneeds to be assured that the tested/monitored community is not adapted to chemical stress and reacts sensitive tothe substance of concern.

3) Reference item or exposure assessment needed4) Guidance document available see Candolfi et al. 2000, J Pest Science 73 (6): 141-147.

McKelvey et.al. 2002 (Pest Manag Sci 58: 1161-1174)suggest that crop species sensitivity to herbicidesis adequate to represent non-crop species responseregardless of chemical class or exposure.

S u m m a r y o f d i s c u s s i o n :1) Very limited monitoring data on invertebrates

other than honeybees are available.2) Limited field trial data are available in the open

literature, since most of the performed studiesbelong to the company’s assets.

3) More effort to collect life history data is neededfor invertebrates.

4) Monitoring data are useful but we need toknow and define the limitations.

5) Identification of species is an issue. For someinvertebrate groups (e.g. carabids) good identi-fication keys are available, however, for someothers, (e.g. oribatids) good keys are missing.Computer aided identification keys may behelpful. At least a strong need for good educat-ed taxonomists were identified.

6) Comprehensive studies needed: multiple para-meter studies are needed.

7) Need to make more effort to determine expo-sure level in the environmental compartmentslikely to be at risk and in the target and non-target species inhabiting that compartment. Ex-posure data particularly of residues in fooditems are extremely important also to refine therisk assessment. Exposure assessment is essen-tial in monitoring studies but less so in fieldstudies where application is highly controlledand we have reference compounds. The validityof monitoring studies without exposure assess-ment is very questionable.

8) GIS could be used for space and time resolu-tion; recovery and local extinction are verymuch dependent on spatial arrangement oftreatment and dispersal of invertebrates be-tween plots. GIS could also be very helpful inchoosing appropriate sites for the field andmonitoring trials.

2 LIST OF ECOLOGICALLY IMPORTANTEFFECTS OR ENDPOINTS WHICHWERE THOROUGHLY INVESTIGATEDAND IDENTIFIED WHETHER EFFECTSWERE OBSERVED OR NOT

� Population dynamics� Species composition /community structure� Biomass (plants, animals, potentially microbes)� Organic matter breakdown� Behavioural studies (e.g. repellency)� Extent, time, duration and space of effects

Lesson learned from the presentations:� Brock paper. The author is of the opinion that

NOEC and LOEC can be predicted more or lessreliably from lab studies (if exposure is similar)but what will happen above the threshold de-pends on several factors (e.g. structure of sys-tem, species, and life cycle). This paper also in-dicates that current risk assessment practicesfor individual compounds using appropriatespray drift estimates is adequate to protect sen-sitive populations exposed to realistic combina-tions of pesticides. Limited data are available inthe literature to verify this for terrestrial inver-tebrates (e.g. ESCORT 2). Some data are avail-able for soil organisms (earthworms, e.g. He-imbach 1992). Chemical monitoring data arenot necessarily linked to effects seen in theaquatic system: poor correlation seen withsampling results. The main determinant of re-covery was the generation time especially withaquatic insect species. For the soil compart-ment, this is not known since no chemical mon-itoring was done.

� Kreuger & Brown. Effect assessment for plantswould be improved if we would have chemicalmonitoring data. Basic terrestrial in- and off-crop exposure data and information concern-ing the main exposure routes of the species(e.g. where, when, how much, see Kochposters) are lacking and need absolutely to begenerated. Moreover, better usage of existingdata (e.g. residue data), arising at times of eco-logical significance rather than for determiningcrop residues at harvest, could improve the as-sessments. The bulk of contamination in theSwedish study seems to be due to point sourcerelease. Exposure assessment is an importantfactor when predicting effects of terrestrial in-vertebrates for off-crop deposits, droplet size of


126

a p p e n d i x c

CONTENTS

spray drift is the critical factor in determiningthe extent of drift deposits.

� Liess. Monitoring should focus on long-term ef-fects. Parameters other than toxicology are veryimportant in the effects seen and measured.The species’ physiological sensitivity as well astheir recolonisation potential (both implement-ed in the species at risk concept, SAR) need tobe taken into account when assessing the ef-fects on biocenosis level. Need to define thespecies at risk concept!

� Römbke. Earlier expert meetings agreed thateffects on earthworms and non-target arthro-pods in the field smaller than 50% are usuallyfalling in the range of natural fluctuation andcan often hardly be demonstrated with statisti-cal certainty. Effects above 50% reduction areconsidered critical. No fundamental long-termeffects on NTA were seen in the UK studies(Boxworth-Scarab) but the studies were criti-cised. Number of tests available for soil organ-isms (soil micro-organisms, earthworms, mites,Collembola) seems to be sufficient to deter-mine potential effects on soil organisms butvalidation of the data generated (particularlyhigher-tier tests) and used for the risk assess-ment needs to be done. This does not apply forNTA where only data on part of the fauna po-tentially exposed is assessed (beneficial arthro-pods). For plants there are only 1st tier tests.

� Mineau: Underestimation of exposure may leadto incorrect risk assessment. Consideration ofall exposure routes is essential for a correct riskassessment. We acknowledge the importance ofdifferent exposure routes for different speciesand for chemical substances with particular at-tributes.

3 HOW DO DATA FROM MONITORINGSTUDIES COMPARE WITH OUTPUTFROM REGULATORY STUDIESAND/OR RISK ASSESSMENT?

3.1 WHAT APPROACHES AREAPPROPRIATE FOR INCORPORATINGFIELD/MONITORING DATA IN RISKASSESSMENT?

There are appropriate test systems and risk assess-ment tools in place for the use of field trials forhoneybees, earthworms and non-target arthro-

pods (e.g. ISO 1999, IOBC 2000, EPPO 2002). Ap-propriate monitoring tools are available for honey-bees. Number of test available for soil organisms(earthworms, mites, Collembola, see EU Terrestri-al Guidance Document 2002) seems to be suffi-cient to determine potential effects on soil organ-isms but validation of the data (especially fielddata) generated and their usage in the risk assess-ment needs to be done. For non-target plants andsoil invertebrates other than earthworms, highertier study and risk assessment tools are lacking (ci-tation). Bee monitoring data were successfullyused to validate the testing and risk assessmentscheme (EPPO 2003).

The following approaches were considered to beappropriate for incorporation of monitoring datain risk assessment:a) Post-registration soil organisms, NTA or NTP

monitoring could be used to validate the cur-rent risk assessment procedures.

b) Such test could also be used to check the effec-tiveness of risk reduction (mitigation) mea-sures imposed.

c) Monitoring data can be used to reduce the un-certainty by enlarging the scale and “quality” ofdata available (e.g. spatial and taxonomic scalein different regions).

d) Monitoring data can provide public confirma-tion of absence of significant ecological effectson a field scale.

3.2 HOW TO INCLUDE RECOVERY ANDRE-COLONIZATION – ESPECIALLYFROM UNCONTAMINATED AREA

Recovery and re-colonisation data cannot easily beimplemented as part of the risk assessmentprocess. The spatial and temporal importance ofthe components of recovery was recognised. Forexample, recovery by immigration will be a func-tion of the scale of the treated area. Conversely, re-covery by emergence of fresh individuals fromwithin treated areas will be unaffected by the scaleof treatment. In monitoring studies these process-es are likely to be included by definition. The in-tensity of pesticide use and spatial distributionpatterns particularly for extensive cropping pat-terns with frequent perturbations, needs be incor-porated in the risk assessment. Regional risk as-sessment is probably the future way to go. Spatial

r e p o r t s o f t h eb r e a ko u t g r o u p s te r re st r i a l i n ve r te b ra te s a n d p l a n t s

127

CONTENTS

explicit modelling, using geographic informationsystems combined with biogeography and disper-sal data of the target/non-target species, could bean effective way of achieving this. Reference dataon the non-disturbed habitat/community is neces-sary to provide a benchmark for assessing changesand measuring recovery.

3.3 WHAT GENERAL CONCLUSIONCAN BE DRAWN, ON THE BASIS OFCURRENT EVIDENCE, ABOUT THELEVEL OF AGREEMENT BETWEENLAB-BASED RISK ASSESSMENTAND FIELD EFFECTS?

Laboratory data are considered to be worst-caseand therefore appropriate to screen for harmless-ness. If you get the exposure of the lab study cor-rect (i.e. route and level of exposure) the predic-tion of the field effect should probably be good forthe initial or threshold value. Consideration of ex-posure duration (appropriate test design for thespecies) and measuring the correct endpoints (e.g.acute, chronic, food web, function) are essential.In addition to that, indirect and long-term effectsmay also depend on factors, which cannot be in-corporated in lab tests like landscape, climate,species composition and so on.

3.4 FURTHER RESEARCH REQUIRED TOIMPROVE THE CONDUCT ANDINTERPRETATION OF AFIELD/MONITORING STUDY FOR RISKASSESSMENT

The following research areas need further atten-tion:

� Life-history, life-cycle, dispersal and other eco-logical data including food sources for soil in-vertebrate and NTA are needed

� More knowledge should be generated on whatare the key species in the different agro-ecosys-tems

� Species distribution and behaviour data. Inter-actions between species in trophic networks.

� Taxonomy� Exposure data especially drift deposit and dis-

tribution patterns in terrestrial off-crop habi-

tats (e.g. hedges) and probabilities of being ex-posed

� Relate mode of action and species-specificmetabolic pathways to effects observed

� Determine maximum duration of effects thatwill not cause long term population decline(development of population models would bevery useful for this).


128

a p p e n d i x c

CONTENTS

1 GENERAL COMMENTS ABOUTEFFECTS OF PESTICIDES IN THE FIELD

Agricultural fields are inhabited by a broad rangeof avian and mammalian species. Mammals andbirds may have their nests or burrows there, orvisit fields in search for food, or just pass through.Therefore they inevitably come into contact withpesticides. When trying to predict effects then theexposure assessment turns out to be a hard nut tocrack. The reason is that the exposure for an invid-ual, expressed as dose or body burden, is not onlya function of the environmental concentration butalso depends on the animals‘ behaviour which ishighly variable. Therefore there is a need for mon-itoring studies not only to find out what happensin the field but also to improve exposure modelsand risk assessment schemes by supplying inputdata and data for calibration.

Besides direct (toxic) effects pesticides may exertindirect effects by altering the habitat and the foodavailability. Indirect effects may be influenced byagricultural measures other than pesticide use.Monitoring studies may be designed to determineeither direct or indirect effects. Sometimes, how-ever, it is challenging to find the causal networkbehind an observed effect.

Monitoring studies have clearly shown that thereare direct as well as indirect effects of pesticideson birds and mammals. The extent of such effectsremains unclear, however it appears that indirecteffects are of more concern than direct effects.

2 FIELD MONITORING STUDIES

2.1 TYPES OF STUDIES

There are no standardized procedures for fieldstudies and monitoring projects, instead there is abroad spectrum of approaches. For the purpose ofthis workshop the group considered the useful

classification which was presented in the keynotelecture by Mark Clook.� Effects field studies� Ecological field studies (also called ‚generic‘

field studies‘)� Residue field studies� Proactive monitoring studies� Reactive monitoring studies

2.2 INVENTORY OF ENDPOINTS

The following collections comprise endpoints thathave been part of field/monitoring studies or thatare often discussed. The aim of the workshop hasnot been to discuss methods and endpoints in de-tail, therefore a few references only are given(Greaves et al., 1988; Somerville and Walker, 1990;Maltby et al., 2001; Pastorok et al., 2001; Munns etal., in press).

a) Individual level� Behavioural responses

� Avoidance� Food collection, foraging plot choice, feed-

ing rates � Time budgetting� Courtship behaviour

� Survival (mortality, debilitation)� Reproductive performance� Physiological parameters

� Analysis of faeces for the substance and/ormetabolites

� Enzyme-assays� Fat content� Egg shell thickness� Sub cellular endpoints (e.g. genetoxicity)

� Tissue residues

b) Population level� Abundance (including spatial distribution)� Vital rates� Population growth rate� Return, recovery, extinction rates

r e p o r t s o f t h eb r e a ko u t g r o u p s b i rd s a n d m a m m a l s

129

Birds and Mammals

CONTENTS

� Population structure� Inbreeding coefficient

c) Community level� Shifts in species dominance

In the case of birds and mammals, population andcommunity endpoints are challenging due to thelarge home ranges of the animals and long genera-tion times (Kendall and Lacher, 1994). Sometimescommunity parameters are used to characterizethe suitability of the study area rather than to in-terpret them in terms of effects.

3 LIMITATIONS AND DESIRABLEFEATURES OF FIELD/MONITORINGSTUDIES

3.1 LONG-TERM SPECIES-SPECIFICMONITORING STUDIES

Such studies are usually not designed to examinejust pesticide effects but instead look at the ecolo-gy of individual species. However, pesticide effectsare sometimes detected (which depends onwhether there is a variation with regard to pesti-cide use, or any obvious hints on causal effects).

An example of a thorough investigation is the par-tridge survival project (Sussex study). The studywas started in 1968 as part of a suite of work to in-vestigate the reasons for the decline of the GreyPartridge. The extensive monitoring has continuedever since. This data was used in the 1980s to iden-tify causes behind the decline which turned out tobe the reduction in availability of chickfood (Potts,1986). This lead to the Cereal and Game birds pro-ject and from this the development of Conserva-tion Headlands. This later work included a multi-tude of methods, best summarised in Sotherton,1991. In the late 1990s the Sussex data was re-analysed to look at impact of pesticides (Ewaldand Aebischer, 1999; 2000; Holland, presenta-tion). The study can briefly be described as fol-lows:� Triggered by observed population decline of

the Grey Partridge� Aimed at general effects of farming practice

and indirect effects of pesticides, but excludeddirect toxic effects of pesticides

� Multitude of methods and techniques (radio-tagging, diet composition via faecal analysis,nesting parameters)

� Length of monitoring (20 years) was importantfor its success

� There could be demonstrated the chain ofcausal links from pesticide use to insect abun-dance to chick survival to population size

� Results have not been used for the assessmentof individual pesticides but led to managementmeasures (recommendations with regard toconservation headlands).

Similar long-term monitoring studies:� Corn bunting study: There was the same hy-

pothesis of cause and effects as in the partridgeproject, and it was also proven (Brickle et al.,2000)

� Turtle Dove study: The study compared ecolog-ical parameters in the 1990s to the 1960s; therecould be demonstrated a change in diet, an in-crease in foraging distance and a decrease inthe number of nesting attempts (Browne andAebischer, in press)

� Sparrowhawk study (Newton, 1986; Newtonand Wyllie, 1992; Sibly et al., 2000)

� Studies on yellowhammer (Morris et al., inpress)

� Studies commissioned by the Danish Environ-mental Protection Agency (Petersen, 1996; Pe-tersen and Jacobsen, 1997)

3.2 FARM SCALE EXPERIMENT

The Breakout Group identified farm-scale experi-ments as a distinctive type of field study, in whichlarge blocks of farmland comprising multiplefields are subjected to contrasting treatments. Dueto their cost, there are few examples of such stud-ies. A well-known example for terrestrial verte-brates is the Boxworth Project (Greig-Smith et al.,1992). Avian endpoints studied at Boxworth in-cluded abundance, reproductive performance, fre-quency of nest visits by birds feeding young, dietcomposition (from faecal samples), andcholinesterase inhibition in nestlings. The projectalso measured abundance of wood mice usingmark-release-recapture methods. The project wasaffected by a number of limitations. The most im-portant, recognised in the project report, was thelack of replication: due to cost limitations there

b i rd s a n d m a m m a l s

130

a p p e n d i x c

CONTENTS

was only one block of farmland for each treat-ment. This made it difficult to assess whether dif-ferences between the blocks were due to the treat-ments or to confounding factors. In addition, de-spite the relatively large size of the blocks, thenumber of active bird nests available for study atthe time of any particular pesticide applicationwas small. The only vertebrate endpoint thatshowed a clear pesticide effect was the abundanceof wood-mice, which decreased markedly and thenrecovered following application of methiocarb slugpellets.

Other examples of farm-scale studies involvingvertebrate endpoints include:� UK organic farming studies (Anonymous, 1995)� Danish yellowhammer and skylark studies (Pe-

tersen et al., 1995; Odderskær et al., 1997)� UK Game Conservancy study on conservation

headlands (Sotherton, 1991)

3.3 REGULATORY FIELD STUDIES

Regulatory field studies are usually undertaken todetermine the direct effect following the use of aspecific compound. To that end they are plannedas controlled field studies in contrast to monitor-ing studies. To be useful for pesticide regulation itmust be possible to demonstrate the causal rela-tionship of observed effects to the pesticide.Typical study designs include one or more of thefollowing endpoints:� Mortality (carcass searching)� Censuses (e.g. CMR)� Reproductive performance (observation of nest

boxes) � Biochemical endpoints (provide a good link

with the pesticide, but are not available for allkinds of pesticide chemistries)

Carcass searching is useful if it is carried out in anappropriately and the pitfalls are avoided: Effi-ciency and power must be demonstrated to be suf-ficient, observer disturbance must be avoided, andthe cause of death has to be determined. Biochem-ical endpoints (biomarkers) may provide a goodlink with pesticide exposure or effects, however,are not always available. Finally, the representa-tiveness of the study in terms of the proposed useis crucial, for example is the location and environ-ment, climate, diversity of birds/mammals etc allappropriate to the proposed use of the compound?

3.4 INCIDENT SCHEMES

The Breakout Group discussed reactive monitoringschemes, they defined this type of scheme as onethat considers whether the death of a bird ormammal has been due to the use of a pesticide andif so whether it is due to the correct use, misuse orabuse of a compound.

The scheme considered by the Breakout Group indetail was the UK Wildlife Incident InvestigationScheme (WIIS) (Barnett et al., 2002). It was notedthat this scheme covers other organisms apartfrom birds and mammals. It was agreed that WIISwas a ‘gold standard’ as regards reactive monitor-ing schemes.

The Group identified the following strengths:� a safety net for the regulatory systems in that it

highlights some incidents which had not beenpredicted by the regulatory system,

� it is countrywide, � highlights the need for risk management issues,� highlights issues regarding misuse and abuse

The Group identified the following limitations:� Low probability: It was agreed that there was a

low probability of an incident being reported;this is due to the fact that in the first placethere is a low probability of an organism beingfound, that this carcass will then be reported toWIIS and then that the incident will be accept-ed by WIIS and assessed.

� Biased: The Scheme only measures lethality; itis more likely to consider large conspicuousspecies, species of conservation interest/value,and those that form large flocks rather thansmall species.

Another successful incidents scheme is establishedin France. It is operated by a national network(SAGIR) and generally aims at wildlife health, notonly at pesticide poisonings. Its focus had beenmainly on game species but the scheme is now ex-tended also to other species. Apart from that theorganisation of incident investigation schemes inEurope seems to be underdeveloped (deSnoo etal., 1999).


131

CONTENTS

4 HOW SHOULD FIELD/MONITORINGDATA BE USED IN RISK ASSESSMENT?

4.1 INCIDENTS DATA

From the nature of these data they can only beused retrospectively. If incidents come up they flaga risk; however, in the absence of incidents no con-clusions can be drawn, i.e. a lack of reported inci-dents does not mean a lack of incidents.

4.2 EFFECTS FIELD STUDIES

Concern was raised about the usefulness of fieldeffect studies, particularly relating to the difficultyof proving a lack of effect. However, such studiesshould not be excluded and when conducted andused in risk assessments they must meet at leastthe following quality requirements:� The endpoints must be relevant.� A solid database is necessary for a quantitative

evaluation. The study design and statisticalpower must be sufficient to detect effects.

� The trial should not be biased in any way (e.g.exposure reduced due to disturbance by studypersonnel)

� The link between cause and effect should beclear

� The study needs to be representative of the fullrange of conditions expected in normal use.

If effects are seen then this confirms risk. Howev-er, the absence of effects does not prove that theproduct is safe and would not immediately over-turn the lab-based risk assessment. However, theresults could be valuable if there are additionalendpoints that help to explain why there havebeen no effects (e.g. biomarkers, residue analysis)and if it can be demonstrated that the study condi-tions are genuinely worst case. The interpretationis facilitated if the study is accompanied by otherfield type studies, e.g. avoidance/feeding behav-iour and residue studies.

4.3 ECOLOGICAL/RESIDUE FIELD STUDIES

The main purpose of these data is to improve riskassessment procedures by feeding input parame-ters into exposure models (deterministic and prob-abilistic).

5 HOW SHOULD POPULATION EFFECTS,RECOVERY, AND RECOLONIZATIONBE CONSIDERED

With birds and mammals losses of a few individu-als may already be considered unacceptable byrisk managers, and thus regulatory action may betaken prior regardless of any adverse effects at thepopulation level. Nevertheless, even in those casesassessors should include a statement in the riskcharacterization outlining potential population ef-fects. Experience from risk assessments has shownthat the issue of recovery and recolonization main-ly becomes relevant with small mammalianspecies which are sedentary in the treated area,for example small rodents in pastures or orchards.

6 GENERAL CONCLUSION ABOUTTHE LEVEL OF AGREEMENT BETWEENLAB-BASED RISK ASSESSMENT ANDFIELD/MONITORING STUDIES

6.1 ARIOUS FIELD WORK ANDMONITORING PROGRAMMES

Apart from the studies covered by the meta-analy-sis (see below) there are detailed data from laband field on a range of compounds. The followinglist gives examples, but is not exhaustive:� Fonofos: effects from seed treatment (Hart et

al., 1999)� Dieldrin/Aldrin: effects from seed treatment

use (Murton and Vizoso, 1963). � Fenitrothion (Pauli et al., 1993)� Rodenticides: barn owl and polecat monitoring

study (Newton et al., 1999)� Methiocarb: effects on wood mice from mollus-

cicide use (Greig-Smith et al., 1992)� Azinphos-methyl: effects on small mammals

(Edge et al., 1996)

Without going deep into the details of individualstudies the group made the following general ob-servations:� Field observations include evidence of effects

not covered by standard risk assessment (otherroutes of exposure, indirect effects).

� The quality of field data is not always sufficientto draw any conclusions

� Even where data are good it is not necessarilypossible to link findings with risk assessment


132

a p p e n d i x c

CONTENTS

due to confounding factors. Often there arecase-specific or compound-specific factors mak-ing it difficult to make statements with regardto the general suitability of risk assessment pro-cedures (e.g. avoidance, persistence, secondarypoisoning).

� There could be valuable material in the litera-ture which should be searched for and evaluat-ed.

6.2 META-ANALYSIS OF PIERRE MINEAU

The study is published (Mineau, 2002) and waspresented to the workshop as a keynote lecture.The breakout group considered the study the bestavailable and suited for this purpose. However, afew points have to be observed: Only organophos-phate and carbamate pesticides are covered; withregard to location there is bias towards NorthAmerican studies Despite these issues the groupfelt able to draw the following conclusions:� According to this study the current EU Tier-I-

risk assessment procedure (Anonymous 2002)produces a small proportion of false negatives(i.e. TERacute >10 for pesticides which havecaused mortality in field studies); when basedon quail toxicity it is about 5% of the com-pounds that were included in the study, howev-er, the scheme produces a high rate of falsepositives (70%). Note: The figures of 5% and 70% are associatedwith some assumptions and uncertainties: first-ly, with regard to the data which are availablefor running the risk assessment scheme (e.g.number and kind of species tested), secondlywith regard to a potential bias in the field dataset as mentioned above, and thirdly with re-gard to the definition of a “positive” from thefield data set (critical level of mortality rate,observed vs. modelled probabilities)

� The study suggests that the predictive power ofthe Tier-I-risk assessment scheme is not satis-factory, because it misclassifies a substantialproportion of pesticides. Improving the para-meters for the current risk assessment model(e.g. RUD = Residue per unit dose) is desir-able, however, tuning the parameters to a lowerrate of false negatives probably will result in anincrease of false positives. It rather appearsthat the model itself has certain deficits (e.g.the exposure part of the model). If that were

true then working on the model structure couldenhance the discriminatory power.

� There are strong indications that dermal expo-sure is not negligible as the current risk assess-ment scheme supposes, at least fororganophosphate pesticides.

7 WHAT FURTHER RESEARCH ISREQUIRED TO IMPROVE THECONDUCT AND INTERPRETATIONOF A FIELD/MONITORING STUDYFOR RISK ASSESSMENT

� Proactive monitoring is currently rarely usedwith regard to birds and mammals, but its po-tential should be explored.

� Ecological data (‘generic’ field studies or litera-ture reviews) should be generated to help inter-pret classical style effects studies and to pro-duce input data for exposure modelling.

� Consideration could be given to establish“benchmark cases” that could serve as refer-ences for other compounds and that could beused for validation of laboratorybased risk as-sessment schemes.

R e f e r e n c e sAnonymous. 2004. Assessing the indirect effects

of pesticides on birds. Draft Final report forPN0925. Defra, UK

Anonymous. 1995. The effect of organic farmingregimes on breeding and winter bird popula-tions. Parts I-IV. BTO Research Report 154.

Anonymous. 2002. Guidance Document on RiskAssessment for Birds and Mammals

Under Council Directive 91/414/EEC, Docu-ment SANCO/4145/2000 – final.http://europa.eu.int/comm/food/fs/ph_ps/pro/wrkdoc/wrkdoc19_en.pdf

Barnett EA, Fletcher MR, Hunter K,Sharp EA. 2002. Pesticide Poisoning of Ani-mals 2001: Investigations of suspected Inci-dents in the United Kingdom – A report of theEnvironmental Panel of the Advisory Commit-tee on Pesticides. Defra publication.


133

CONTENTS

http://europa.eu.int/comm/food/fs/ph_ps/pro/wrkdoc/wrkdoc19_en.pdf

http://europa.eu.int/comm/food/fs/ph_ps/pro/wrkdoc/wrkdoc19_en.pdf

Brickle NW, Harper DGC, Aebischer NJ,Cockayne SH. 2000. Effects of agricultural in-tensification on the breeding success of cornbuntings Miliaria calandra. Journal of AppliedEcology 37, 742–755.

Browne SJ, Aebischer NJ. In press. Habitat use,foraging ecology and diet of Turtle Doves Strep-topelia turtur in Britain. Ibis 145.

De Snoo GR, Scheidegger NMI, De Jong FMW.1999. Vertebrate wildlife incidents with pesti-cides: A European survey. Pesticide Science 55,47–54.

Edge WD, Carey RL, Wolff JO, Ganio LM,Manning T. 1996. Effects of guthion 2S on Mi-crotus canicaudus: a risk assessment validation.Journal of Applied Ecology 33, 269–278.

Ewald JA, Aebischer NJ. 1999. Avian food, birdsand pesticides. JNCC Report No. 296, JNCCCommunications, Peterborough, 103 pp.

Ewald JA, Aebischer NJ. 2000. Trends in pesti-cide use and efficacy during 26 years of chang-ing agriculture in southern England. Environ-mental Monitoring and Assessment 64,493–529.

Greaves MP, Greig-Smith PW, Smith BD (eds).1988. Field method for the study of environ-mental effects of pesticides. BCPC MonographNo. 40, BCPC Publications, Thornton Heath,370 pp.

Greig-Smith PW, Frampton GK, Hardy AR(eds). 1992. Pesticides, cereal farming and theenvironment: the Boxworth Project. HMSO,London, UK.

Hart A, Fryday S, McKay H, Pascual J,Prosser P. 1999. Understanding risks to birdsfrom pesticide-treated seeds. In: Adams, N. &Slotow, R. (Eds), Proc. 22 Int. Ornithol. Congr.Durban: pp. 1070–1087. Birdlife South Africa,Johannesburg.

Kendall RJ, Lacher TE (eds). 1994. Wildlifetoxicology and population modeling. SETACSpecial Publication, Lewis Publishers, BocaRaton, 576pp.

Maltby L, Kedwards TJ, Forbes VE, GrasmanK, Kammenga JE. 2001. Linking individual-level responses and population-level conse-quences. In: Baird DJ, Burton GA Jr (eds) Eco-logical variability: separating natural from an-thropogenic causes of ecosystem impairment.SETAC, Pensacola FL, USA, pp. 27–82.

Mineau P. 2002. Estimating the probability of birdmortality from pesticide sprays on the basis ofthe field study record. Environmental Toxicolo-gy and Chemistry 21, 1497–1506.

Morris AJ, Wilson JD, Whittingham MJ,Bradbury RB. In press. Evidence for indirecteffects of pesticides on breeding yellowham-mers Emberiza citrinella. Agriculture, Ecosys-tems and Environment.

Munns WRJ, Gervais J, Hoffman AA, HommenU, Nacci DE. In press. Modeling approaches topopulation-level ecological risk assessment.SETAC.

Murton RK, Vizoso M. 1963. Dressed cereal seedas a hazard to wood-pigeons. Ann. appl. Biol.,52: 503–517.

Newton I. 1986. The Sparrowhawk. Poyser, Cal-ton.

Newton I, Wyllie I. 1992. Recovery of a spar-rowhawk population in relation to decliningpesticide contaimination. Journal of AppliedEcology 29, 476–484.

Newton I, Shore RF, Wyllie I, Birks JDS,Dale L. 1999. Empirical evidence of side-ef-fects of rodenticides on some predatory birdsand mammals. In: Cowand DP, Feare CJ (eds.)Advances in vertebrate pest management, Fi-lander, Fürth, 347–368.

Odderskær P, Prang A, Elmegaard N, Ander-sen PN. 1997. Skylark reproduction in pesticidetreated and untreated fields. Comparative stud-ies of skylark Alauda arvensis breeding perfor-mance in sprayed and unsprayed spring barleyfields. Bekæmpelsesmiddelforskning fraMiljøstyrelsen 32, Ministry of Environment andEnergy, Denmark.

Pascual JA, Hart ADM. 1997. Exposure of cap-tive feral pigeons to fonofos-treated seed in asemi-field experiment. Environmental Toxicol-ogy and Chemistry, 16: 2543–2549.

Pastorok RA, Bartell SM, Ferson S,Ginzburg LR (eds). 2001. Ecological model-ing in risk assessment. CRC Press Lewis Pub-lishing, London.

Pauli BD, Holmes SB, Sebastien RJ, Rawn GR.1993. Fenitrothion risk assessment. CanadianWildlife Service Tech. Report Series No. 165, 75pp.


134

a p p e n d i x c

CONTENTS

Petersen BS, Falk K, Bjerre KD. 1995. Yel-lowhammer studies on organic and convention-al farms. Comparative analyses of clutch size,nestling growth and foraging behaviour in rela-tion to pesticide sprayings. Bekæmpelsesmid-delforskning fra Miljøstyrelsen 15, Ministry ofEnvironment and Energy, Denmark.

Petersen BS. 1996. The distribution of birds inDanish farmland. An analysis of distributionand population densities of 14 farmland speciesin relation to habitat, crop and pesticide use.Bekæmpelsesmiddelforskning fraMiljøstyrelsen 17, Ministry of Environment andEnergy, Denmark.

Petersen BS, Jacobsen EM. 1997. Populationtrends in Danish farmland birds. A modelling ofpopulation changes 1976–1996 with special ref-erence to the effects of pesticide use. Bekæm-pelsesmiddelforskning fra Miljøstyrelsen 34,Ministry of Environment and Energy, Denmark.

Potts GR. 1986. The partridge: pesticides, preda-tion and conservation. London: Collins.

Sibly RM, Newton I, Walker CH. 2000. Effectsof dieldrin on population growth rates of spar-rowhawks 1963–1986. Journal of Applied Ecol-ogy 37, 540–546.

Somerville L, Walker CH (eds). 1990. Pesticideeffects on terrestrial wildlife. Taylor & Francis.London, New York, Philadelphia, 404 pp.

Sotherton NW. 1991. Conservation Headlands: apractical combination of intensive cereal farm-ing and conservation. In: Firbank LG, Carter N,Darbyshire JF, Potts GR (eds) Ecology of tem-perate cereal fields, pp. 373–397. Oxford: Black-well Scientific Publications.


135

CONTENTS


136

i m p r e s s u m

CONTENTS

IMPRESSUM

Proceedings from the workshop:

“Effects of Pesticides in the Field”· EPiFA SETAC publication

EditorsMatthias LiessColin BrownPeter DohmenSabine DuquesneAndy HartFred HeimbachJenny KreugerLaurent LagadicSteve MaundWolfgang ReinertMartin StrelokeJosé V. Tarazona

PublisherSETAC, puplished: July 2005

Design & LayoutDarius Samek2medien, BerlinJohanna MichelAlexa Sabarth

PrintingDruckerei Hermstein, Berlin

isbn1-880611-81-3

Effects of Pesticides in the Field is one of manySETAC publications that offer timely reviews and newperspectives on current topics relating to broad environ-mental toxicology and chemistry issues. SETAC publi-cations often are based on the collaborative efforts of topscientists and policymakers, and they undergo extensiveprepublication reviews. SETAC assumes an active leader-ship in the development of educational programs andpublishes the peer-reviewed, international journals, En-vironmental Toxicology and Chemistry and IntegratedEnvironmental Assessment and Management. For moreinformation, contact the SETAC Office nearest you;details about the Society and its activities can be foundat www.setac.org.


SETAC OfficeAvenue de la Toison d’Or 67Brussels, BelgiumT 322 772 72 81 F 322 770 53 83E [email protected]

www.setac.org

Environmental Quality Through Science®

www.setac.org

www.setac.org

EU & SETAC EUROPE WorkshopOctober 2003Le Croisic, France

Editors:Matthias Liess

Colin BrownPeter Dohmen

Sabine DuquesneAndy Hart

Fred Heimbach

Jenny KreugerLaurent LagadicSteve MaundWolfgang ReinertMartin StrelokeJosé V. Tarazona

EFFECTS OF

PESTICIDESFIELDIN THE

SETA

CEf

fect

s o

f Pe

stic

ide

s in

th

e F

ield

EPIF

About the editors

Matthias Liess leads the Effect Propagation Group,Aquatic Ecotoxicology, UFZ-Centre for EnvironmentalResearch, Leipzig, Germany.

Colin Brown, Professor of Environmental Sciencewithin the Environment Departmentof the Universityof York, UK.

G. Peter Dohmen senior scientist at the AgriculturalResearch Center, BASF-AG, Germany.

Sabine Duquesne, research scientist at the Effectpropagation group, Aquatic Ecotoxicology, UFZ-Centrefor Environmental Research, Leipzig, Germany.

Andy Hart leads the Risk Analysis Team at theCentral Science Laboratory (CSL), York, UK.

Fred Heimbach, senior scientist at the Institute ofEnviromental Biology in the Crop Protection Divisionof Bayer CropScience, Monchheim, Germany.

Jenny Kreuger, Assistant Professor at the SwedishUniversity of Agricultural Sciences, Division of WaterQuality Management.

Laurent Lagadic, research director at the INRA(National Institute for Agronomic Research) inRennes, France.

Steve Maund, Global Lead for aquatic ecology,Syngenta, Basel, Switzerland

Wolfgang Reinert, European Commission, Healthand Consumer Protection – Directorate General, UnitChemicals, Contaminants and Pesticides, Brussels,Belgium.

Martin Streloke, senior scientist, Federal Office ofConsumer Protection and Food Safety (BVL),Braunschweig, Germany.

José V. Tarazona, Director of the Department of theEnvironment. Spanish National Institute forAgricultural and Food Research and Technology (INIA),Madrid, Spain.

The EU Uniform Principles for the assessment of plantprotection products (PPP’s) require that if the prelimi-nary risk characterization indicates potential con-cerns, it has to be granted that “... under field condi-tions no unacceptable impact on the viability ofexposed organisms ...” occurs. To date, such assess-ments have been made by conducting higher-tierstudies. The aims of the workshop, were to: (i) reviewavailable field monitoring studies addressing theenvironmental effects of PPP’s due to agriculture, (ii)compare observed effects of PPP’s under field condi-tions with the impact predicted on the basis of thecurrent risk assessment guidance, and (iii) identifyrequirements for future monitoring studies and thedesign for higher tier tests.

EEffffeeccttss ooff PPeessttiicciiddeess iinn tthhee FFiieelldd –– EEPPIIFF is one of many SETAC publications that offer timelyreviews and new perspectives on current topics relat-ing to broad environmental toxicology and chemistryissues. SETAC publications often are based on the col-laborative efforts of top scientists and policymakers,and they undergo extensive prepublication reviews.SETAC assumes an active leadership in the develop-ment of educational programs and publishes thepeer-reviewed, international journals, EnvironmentalToxicology and Chemistry and Integrated Environ-mental Assessment and Management.For more information, contact the SETAC Office near-est you; details about the Society and its activitiescan be found at www.setac.org.

Administrative Offices


SETAC OfficeAvenue de la Toison d’Or 67Brussels, BelgiumT 322 772 72 81 F 32 2 770 53 83E [email protected]

E n v i r o n m e n ta l S c i e n c e

“Effects of Pesticides in the Field” – EPIF

Editors:

Matthias Liess Colin Brown Peter DohmenSabine Duquesne Andy HartFred Heimbach

Jenny Kreuger Laurent Lagadic Steve MaundWolfgang ReinertMartin Streloke José V. Tarazona

Documents

(PDF) Effects of Pesticides in the Field - EPiF - ResearchGate