3464 valuating environmental impacts of genetically modified crops oa

7/30/2019 3464 Valuating Environmental Impacts of Genetically Modified Crops OA

http://slidepdf.com/reader/full/3464-valuating-environmental-impacts-of-genetically-modified-crops-oa 1/194

Olivier Sanvido, Andreas Bachmann,

Jörg Romeis, Klaus Peter Rippe, Franz Bigler

Valuating environmental impacts of

genetically modified crops – ecological

and ethical criteria for regulatory

decision-making






decision-making






decision-making

Olivier Sanvido *

Andreas Bachmann +

Jörg Romeis*

Klaus Peter Rippe +

Franz Bigler *

* Forschungsanstalt Agroscope Reckenholz-Tänikon ART

+ Ethik im Diskurs gmbh



Publiziert mit Unterstützung des Schweizerischen Nationalonds zur Förderung

der wissenschatlichen Forschung.

Bibigrfh Inrmtinn dr Duthn Ntinbibithk

Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliograe;

detaillierte bibliograsche Daten sind im Internet abrubar über http://dnb.d-nb.de.

ISBN 978-3-7281-3443-1 (Printausgabe)Download open access:

ISBN 978-3-7281-3464-6/DOI 10.3218/3464-6

www.vd.ethz.ch

© 2012, vd Hochschulverlag AG an der ETH Zürich

Das Werk einschliesslich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung ausserhalb

der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlags unzulässig und strabar.

Das gilt insbesondere ür Vervielältigungen, Übersetzungen, Mikroverlmungen und die Einspeicherung

und Verarbeitung in elektronischen Systemen.

This work is licensed under a creative commons license



TaBle o coNTeNTs

..........................................................................................................................................................

1 eXecUTIVe sUMMaRY 9

..........................................................................................................................................................

2 INTRoDUcTIoN 15

2.1 Existing denitions o environmental damage 16

2.2 Diculties to apply existing criteria or environmental damage 18

2.3 Research questions and goals o the project 18

2.4 Relation between ethics and ecology in the context o the given research question 19

2.5 Methods and approaches o the project 212.6 Outline o the report 23

..........................................................................................................................................................

3 a GeNeRal DeINITIoN o DaMaGe 27

..........................................................................................................................................................

4 cHaRacTeRIZING THe PRoTecTIoN Goal “BIoDIVeRsITY” 33

4.1 Background 34

4.2 What is biodiversity? 344.3 Environmental protection goals as specied by Swiss legislation 38

4.4 The ecological relevance o biodiversity 42

..........................................................................................................................................................

5 oPeRaTIoNal DeINITIoN o PRoTecTIoN Goals 49

5.1 Diculty to nd an unambiguous denition o the term biodiversity 50

5.2 A matrix or an operational denition o biodiversity in agricultural landscapes 51

5.3 Application o the matrix or an operational denition o protection goals 57

..........................................................................................................................................................

6 THe Use o “THResHolDs” To eValUaTe RIsKs aND cHaNces 69

6.1 Thresholds in ethics 70

6.2 Thresholds in ecology 72

6.3 Using thresholds in regulatory decision-making 75

..........................................................................................................................................................

5

Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



..........................................................................................................................................................

7 a IRsT BaselINe aPPRoacH – DIeReNTIaTING eecTs o DIeReNT PesT 79

MaNaGeMeNT PRacTIces IN MaIZe

7.1 The challenge o selecting an appropriate baseline 80

7.2 Case study 1: Eects o Bt-maize on nontarget arthropods 81

7.3 Approach to dierentiate eects o various pest management practices 87

on nontarget arthropods

..........................................................................................................................................................

8 a secoND BaselINe aPPRoacH – DIeReNTIaTING eecTs o DIeReNT 107WeeD MaNaGeMeNT PRacTIces IN MaIZe

8.1 Dierences to the case study with Bt-maize 108

8.2 Case study 2: Eects o GMHT maize on armland biodiversity 108

..........................................................................................................................................................

9 eTHIcal ReeReNce sYsTeM To assess BIoDIVeRsITY 121

9.1 The deciencies o the Ethical Matrix 122

9.2 Introduction to the ethical reerence system 126

9.3 Explanations on the ethical reerence system 1279.4 Application o the proposed approach 138

..........................................................................................................................................................

10 RecoMMeNDaTIoNs oR DecIsIoN-MaKeRs 143

10.1 Ethical considerations when evaluating eects o GM plants 144

10.2 Protection goals 145

10.3 Thresholds 146

10.4 Comparative assessments and baselines 147

10.5 Weaknesses o the regulatory system or GM crops 14910.6 Regulatory burden or GM crops 150

10.7 Acknowledgements 151

..........................................................................................................................................................

11 ReeReNces 153

..........................................................................................................................................................

6




..........................................................................................................................................................

aNNeXes 165

ANNEX 1: LIST OF WORKSHOP PARTICIPANTS 166

ANNEX 2: WHAT IS RISK? 169

..........................................................................................................................................................

7



http://slidepdf.com/reader/full/3464-valuating-environmental-impacts-of-genetically-modified-crops-oa 10/194Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



eXecUTIVe sUMMaRY CHAPTER 1




1 eXecUTIVe sUMMaRY

The legal ramework regulating the approval and use o genetically modi-ed (GM) plants demands that regulatory decisions determine what kind o

environmental changes are relevant and represent environmental damage1.

The current debate on the impacts o GM crops on biodiversity illustrates

that consensus on criteria2 that would allow a commonly accepted evalua-

tion o environmental damage is presently lacking. Especially in Europe, GM

crops have been a constantly debated issue, and the interpretation o scien-

tic data is controversially discussed by the dierent stakeholders involved.

Considering the vast amount o scientic data available, one can argue that

the current debate is not primarily due to a lack o scientic data, but more toa lack o clear criteria that allow one to put a value on the impacts o GM crops

on biodiversity.

Ultimately, any decision by regulatory authorities on what they judge being

unacceptable is based on the relevant legal rameworks. Usually, such decisions

are taken in a political context that weighs scientic, ethical and economical

criteria with cultural, religious, aesthetic and other relevant social belies and

practices. Terms such as risk and saety are linked to a conception o damage.

Damage, however, cannot be dened on a purely scientic basis. The normative

character o the term “damage” implies that both choice and denition o whatconstitutes a risk are impossible without a value judgment. Damage has thus

to be dened together with an ethical evaluation as ecological analyses alone

cannot discover “correct” or “objective” criteria or damage.

The project VERDI (Valuating environmental impacts o GM crops – ecolog-

ical and ethical criteria or regulatory decision-making) is an interdisciplinary

collaboration between environmental biosaety and ethics that intends to oer

European policy-makers and regulatory authorities guidance on how decision-

making related to GM crops could be improved. Concentrating on environmental

impacts o GM crops on biodiversity, the project addresses both the ecologicaland the ethical questions involved in nding an operational approach to the

evaluation o environmental damage.

Two case studies with the currently most prevalent GM traits (insect-resist-

ance based on Bt and herbicide tolerance) are used to discuss the open questions

involved. Considerable scientic data on the environmental impacts o these two

traits have been gathered in the past 15 years, allowing one to determine how

such an evaluation could be perormed to be generically applicable to dierent

types o GM crop traits, including new applications o biotechnology.

1 In the ollowing, the two terms “damage” and “harm” are used interchangeably

2 The term criteria is used hereater in the sense o a standard on which a judgment

or decision ma e based.

10




Stakeholders rom dierent European countries including regulatory authorities, agricultural biotech companies and academia were invited to two

expert workshops. In the rst workshop, current approaches and challenges

in the regulatory decision-making process related to GM crops were analyzed.

The second expert workshop aimed at determining what particular eects o

GM crops on biodiversity are to be judged as unacceptable harm based on an

ecological and an ethical valuation comparing eects o GM crops on biodi-

versity to environmental eects o current agricultural management practices.

Results and eedback obtained during the work-shops were used to elaborate a

synthesis o the relevant ethical and ecological aspects when evaluating impactso GM crops on biodiversity.

The results obtained reveal that both protection goals and baselines are

two consistently emerging issues when discussing a denition o damage.

Protection goals as specied by existing legislation are the exclusive starting

point or a denition o damage or regulatory authorities. Yet, the legislative

terms to describe the protection goal “biodiversity” are too vague to be scien-

tically assessed. Two matrices are proposed to address this problem. The rst

matrix introduces the Ethical Reerence System (ERS), being the rst step in

a systematic process aiming at the specication and justication o plausibleethical criteria to evaluate the risks o GM crops on biodiversity. The second

matrix allows or an operational denition o biodiversity by speciying the

areas o protection as well as assessment and measurement endpoints based

on a number o dened criteria. While the initial proposal regarding what to

protect has to be ramed by the regulatory authorities, the operational deni-

tion o protection goals should be dened in a transparent process involving a

dialogue between all relevant stakeholders. The presented matrix can thereby

be used as a tool to structure the dialogue, especially when dening both assess-

ment and measurement endpoints. The process could include stakeholder meet-ings where stakeholders would compile and rank dierent conservation goals

and ecosystem services.

Baselines are recognized to be the second crucial point in decision-making

processes as they determine what makes a change to be damage. Due to their

vague denition, the use o baselines as a decision support tool nevertheless

remains ambiguous, necessitating a more precise characterization. Common

to all baseline denitions is the term “comparison.” Theoretically, decisions

are always taken relative to a comparator that determines the current practice

EXECUTIVE SUMMARY 11




that is judged being acceptable. According to this baseline conception, theimpact o a specic technology can only be compared i the impacts o current

practices are known. However, as GM and non-GM-based management prac-

tices are regulated according to dierent regulatory rameworks in Europe, a

direct comparison o the eects o a GM cropping system to its conventional

non-GM counterpart is impossible under the current Swiss and EU regulatory

rame-work. The baseline approach is thus principally not always applicable

as, or example, the EU regulations demands that one assess potential indirect

or cumulative long-term eects o GM crops while this is not a requirement or

conventional pest management practices. Nevertheless, an important point toconsider in such a comparison reers to the act that all regulatory rameworks

dierentiate between “intended” eects o a specic management practice and

“unintended” eects that are to be minimized. This dierentiation allows or

the outlining o a generic scheme that permits one to evaluate whether the

eects o dierent agricultural management practices are to be regarded as

intended eects (which are judged acceptable) or as unintended eects that

could represent environmental damage. This dierentiation may help to over-

come the principal diculties o the initial baseline conception. In a rst case

study with Bt-maize, a fow chart is presented that can help risk assessors todierentiate the eects o various pest management practices used or Euro-

pean Corn Borer management in maize on the arthropod auna in agricultural

landscapes.

In a second case study with GM herbicide-tolerant maize, criteria or regu-

latory decisions or noninsecticidal GM crops were determined. In contrast to

Bt-maize, the assessment o the direct eects o the genetic modication is not

the primary concern or noninsecticidal GM crops. Rather, changes in agricultural

management may be the primary cause o possible indirect eects on armland

biodiversity. The evaluation o indirect impacts prior to approval o the GM cropis challenging. It might be dicult to perorm such an evaluation within the time

rame normally available or pre-market risk assessment as long time periods

are usually needed or indirect environmental changes to become apparent.

These types o eects might, moreover, only become apparent during the large

scale cultivation o GM crop events under real agricultural management. The

establishment o risk mitigation measures thus appears to be a valid option to

increase the level o saety. The goal o these risk mitigation measures should be

to avoid the risk o reduced crop yields and the long-term build-up o problematic

12




weed communities while supporting a sustainable degree o armland biodiversity. Four risk management options are proposed that can help to achieve

a balance between agricultural production and the support o desired noncrop

plants in arable elds.

All technologies that could potentially harm the environment should be

evaluated according to the same legal criteria, or example, according to their

novelty and not to the process o their development. Hence, what constitutes

environmental damage should not be dened by the technology causing it,

but by the type o damage that should to be avoided. The elaborated ethical

and ecological criteria herein may allow a generally acceptable evaluation o damage that can be applied to a wide range o dierent GM crops. The criteria

could help regulatory authorities to improve decision-making and to take

more accurate and coherent decisions. This may ultimately avoid decisions

on environmental risks o GM crops being arbitrary in comparison to other

technologies.

EXECUTIVE SUMMARY 13




INTRoDUcTIoNCHAPTER 2




2 INTRoDUcTIoN

2.1 exiting dinitin nvirnmnt dmg

The legal rameworks regulating the approval and use o GM crops require regu-

latory authorities to decide what kind o environmental changes are relevant

and represent environmental damage.3 The current debate on the impacts o

GM crops on biodiversity illustrates that consensus on criteria that would allow

a commonly accepted evaluation o environmental damage is presently lacking.

Especially in Europe, GM crops have been a constantly debated issue and the

interpretation o scientic data is debated controversially by the dierent stake-

holders involved. Considering the vast amount o scientic data available, onecan argue that the current debate is not primarily due to a lack o scientic data,

but more due to a lack o clear denitions regarding how to put a value on the

impacts o GM crops on biodiversity.

Given the complexity o the questions involved, a concise and commonly

accepted denition o “environmental damage” is currently missing. A number o

denitions have been proposed (Box 1) that all entail some challenges regarding

their practical application (see section 2.2).

Common to all proposed denitions are the ollowing three eatures:

• Damage is occurring to a natural resource or resource service such as theconservation and sustainable use o biodiversity,

• Damage is measurable by some means,

• Damage is characterized by an adverse change that is either signicant,

severe or exceeding the natural range o variability

The common eatures lead to three main questions that need to be

answered when approaching a denition o environmental damage. These

questions are not addressed in sucient detail in the existing denitions

mentioned above: • What needs to be protected?

• What is to be measured?

• What is adverse?

To answer the rst question, one needs to dene the protection goals that

should not be harmed and more particularly one must to nd an applicable de-

nition o the concept o biodiversity. It can be noted that the concept o biodi-

versity is well dened in theory, though there are a number o diculties that

3 In the ollowing, the two terms “damage” and “harm” are used interchangeably.

16




2.2 Diiuti t ppy xiting ritri r nvirnmnt dmg

The diculty to nd an unambiguous denition o environmental damage is

primarily due to the challenge o applying the criteria that have thus ar been

proposed to assist regulatory authorities when evaluating environmental damage

in actual situations o decision-making. A number o authors have proposed

to evaluate damage according to criteria such as “spatial and temporal extent,”

“severity” and “reversibility” o eects (Ammannet al., 2000; ACRE, 2002; Nöh, 2002)

Decisions determining whether observed eects ulll the proposed criteria

or environmental damage are dicult to take due to methodological limits indata collection and analysis. Scientic methods are only partially capable o

adequately evaluating the magnitude o change o a particular environmental

resource. Environmental sciences can help to assess the abundance o a partic-

ular species group, but it is usually dicult to determine whether an observed

change in a species group is exceeding the natural variation o the species group.

This is because appropriate baseline data is oten lacking and because agro-

ecosystems are dynamic and subject to constant change. Long time periods are

usually needed or environmental changes to become apparent and it may be

impossible to determine whether observed changes will be reversible at somelater point in the uture.

A practicable denition o environmental damage necessitates criteria that are

less prone to the methodological challenges posed by scientic methods. Decision

criteria to evaluate eects o GM crops on biodiversity could be dened using an

approach where GM crop eects are compared to known eects o current agri-

cultural management practices. By putting a relative value on the eects o GM

crops in comparison to known environmental eects o current crop manage-

ment practices, decision criteria would be placed in a context where practical

experiences exist. Provided that this approach is scientically valid and ethically justiable, the approach would allow one to decide whether experienced GM crop

eects are ecologically signicant and why they are judged to be unacceptable.

2.3 Rrh qutin nd g th prjt

Ultimately, any decision by regulators on what they judge being unacceptable

is based on the existing legal rameworks. In addition to considering the actual

legal basis, such decisions are usually taken in a political context that weighs

18




scientic, ethical and economical criteria with cultural, religious, aesthetic andother relevant social belies and values. Concentrating on environmental impacts

o GM crops on biodiversity, the project addresses both the ecological and the

ethical questions involved in nding an operational approach or the evalua-

tion o environmental damage. The project aims at oering guidance about how

decision-making related to GM crops could be improved to all stakeholders

involved in the process o risk assessment o GM crops (policy-makers, regula-

tory authorities, agricultural biotech companies and scientic expert panels).

The main goals o the project are:

• To identiy the main challenges or regulators when assessing and judgingenvironmental impacts o GM crops including the most important ethical

and ecological knowledge gaps or the interpretation o scientic data

• To analyze types and magnitudes o biodiversity impacts o current crop

management practices and (known) impacts o GM crops on biodiversity

• To perorm a comparative ethical and ecological valuation o impacts

o current crop management practices and impacts o GM crops on

biodiversity

• To develop decision-criteria and guidance or an ethical and ecological

evaluation o impacts o GM crops on biodiversity considering the experi-enced impacts o current crop management practices

2.4 Rtin btwn thi nd gy in th ntxt th givn rrh qutin

Terms such as risk and saety are linked to a conception o damage. The notion

o damage or benet depends on our negative or positive valuation o impacts.

The normative character o the term “damage” implies that both choice and de-

nition o what constitutes a risk are impossible without a value judgment. What

we choose and dene to be a risk is based on a certain normative background.Damage must thereore be dened together with an ethical evaluation as ecolog-

ical analyses alone cannot discover “correct” or “objective” criteria or damage.

Natural sciences can thereby only determine the probability that a certain

impact will occur or the likely consequence i a certain impact has occurred.

Ethics is necessary to clariy the concept o damage in general and the concept

o environmental damage in particular. Both are evaluative concepts reerring to

changes or states o aairs that must be assessed negatively. On the basis o this

conceptual analysis, it becomes possible:

INTRODUCTION 19




• to critically analyze the appropriateness o legal risk concepts (given that arisk is a unction o probability and extent o damage); and

• to develop scientically sound and intersubjectively acceptable criteria4 or

the valuation o environmental damage that can be shared and accurately

communicated between dierent individuals.

Ecology as a science needs to rely on ethics i it strives to evaluate possible

and real impacts o GM crops on biodiversity because concepts such as damage

or risk, which play a pivotal role in this evaluation, are inherently value-

laden. The evaluation o impacts o GM crops on biodiversity in the contexto current agricultural systems includes both ethical and ecological questions

that need to be claried. From an ecological perspective, or example, not

every environmental impact is o ecological signicance and leading to rele-

vant impacts on biodiversity. From an ethical point o view, not every ecologi-

cally relevant impact on biodiversity is necessarily considered to be morally

wrong. Whether it is wrong or not primarily depends on the importance that

one ascribes to biodiversity. While there is agreement among ethicists that

biodiversity is valuable, there is no agreement on the importance o this value

(compared to other relevant values) and whether it is an inherent or just aninstrumental value.

The outlined research questions necessitate a true interdisciplinary

approach, which, however, includes some challenges. One diculty is that

ecology and ethics use dierent approaches. A challenge in nding an

adequate approach to environmental damage lies in the inherent diculty

to combine the descriptive scientic approach with the normative approach

used in ethics. Natural sciences try to establish the empirical acts o a matter

(what is the case), whereas ethics determine what ought to be the case. Prob-

ably even more challenging is the act that central concepts such as damageor risk are understood in dierent ways by ethicists and ecologists. In the

present project, it took a considerable amount o time to realize that there are

diverging interpretations o these concepts and to agree on common deni-

tions o these concepts. The common understanding now makes it possible

to proceed to the nal phase o this project, concretizing the ethical and

ecological criteria or the evaluation o impacts o GM crops on biodiversity

in a way that will make them applicable or regulators in actual situations o

decision-making.

20

4 Intersubjectively acceptable criteria are normative criteria that a rational person,

that is, a well-inormed person being committed to the best rationale available, would

accept ater a critical assessment. Such criteria are thus inherently universal and

independent rom a particular, subjective context.Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



2.5 Mthd nd pprh th prjt

A particular strength o the project was its ocus on two expert workshops

inviting regulators, members o biosaety committees and scientists with a

background in ethics, ecology and agriculture rom dierent European coun-

tries to combine their eorts in order to create an elaboration o solutions

to the open questions addressed. The project ollowed a stepwise approach

consisting o ve project phases that were centered on the two expert

workshops:

Phase 1: Review and analysis o current knowledge

The rst project phase consisted o an analysis o the relevant scientic litera-

ture to prepare the rst expert workshop that took place in phase 2 o the

project and to select an appropriate method to be used or the group discus-

sions during the workshop. In order to benet the most rom the discus-

sions planned or the rst workshop, the project team decided to make use o

proessional workshop moderation. The moderator team rom Genius GmbH

(Darmstadt, Germany) recommended the use o the Crea-Space method or

the workshop, a method supporting the development o creative potentials inteams and larger groups (El Hachimi and von Schlippe, 2003). The tool is meth-

odologically derived rom Organizational Development and serves to provide a

ramework or the achievement o sel-organization processes. In the context

o the present project, it was deemed to be the ideal tool to eectively introduce

issues derived rom rst results to a constructive discussion level. In order to

avoid a bias towards a predetermined position supported by the project team

workshop, participants were thus not provided with any background documen-

tation prior to the workshop.

Phase 2: First expert workshop: problem identifcation or decision-making

In order to analyze current regulatory decision-making processes, regula-

tors, members o biosaety committees and representatives o the agricultural

biotechnology industry rom dierent European countries were invited to a

rst expert workshop in June 2008. Workshop participants were supplemented

by scientists with a background in ethics, ecology and agronomy (see Annex 1:

List o workshop participants). The aim o the rst workshop was to determine

current approaches and challenges when evaluating scientic data on impacts

INTRODUCTION 21




o GM crops on biodiversity. The discussion during the group works and in theplenum illustrated that both protection goals and baselines were two consist-

ently emerging issues. Protection goals as specied by existing legislation were

regarded as the exclusive starting point or regulatory authorities or a deni-

tion o damage. Any negative impact on these protection goals would conse-

quently constitute damage. Baselines were recognized as being the crucial

point o any decision-making process o determining what makes a change to

be judged as damage.

Phase 3: Comparative ethical and ecological evaluationPhase 3 aimed at developing solutions or the challenges that had been identi-

ed by invited experts during the rst expert workshop when evaluating impacts

o GM crops on biodiversity. Four approaches were developed to help regulatory

authorities addressing the challenges o vague denitions or both protection

goals and baselines. Prior to the second expert workshop, experts were provided

with our background documents proposing solutions as to how the question o

unclear denitions o both protection goals and baselines could be addressed.

The rst two approaches aimed at enabling a more generic denition o protec-

tion goals in the context o agricultural management. The two latter approachesproposed a methodology or a comparative environmental risk assessment that

can be used to approach the question o an appropriate baseline. For the compar-

ative risk assessment, two currently commercialized GM crops (Bt-maize and

GM herbicide tolerant maize) were used as case studies to specically discuss

the environmental impacts o these GM crops in comparison to current pest and

weed management practices.

Phase 4: Second expert workshop: criteria or evaluation o impacts

o GM crops on biodiversity

The proposed approaches o both the environmental biosaety and the ethics

project parts were presented and discussed during the second expert work-

shop taking place in June 2009 in Engelberg, Switzerland, involving many o the

experts that had attended the rst expert workshop (see Annex 1: List o work-

shop participants). The background documents elaborated during phase 3 o

the project were critically discussed with all participating experts during

the group works taking place in the second workshop. All documents were

not meant to be conclusive as experts were invited to provide eedback and

22




criticism to the approaches proposed. The background documents werediscussed in our group works to determine both strengths and weaknesses o

the our proposed approaches:

• Group work 1: Ethical reerence system to assess biodiversity

• Group work 2: Operational denition o protection goals

• Group work 3: Comparative environmental risk assessment Bt-maize

• Group work 4: Comparative environmental risk assessment GMHT maize

For each group work, participants were randomly grouped into three

working groups. In each group, a rapporteur took notes on the main results o the discussion and presented the results in a short presentation to the plenum

in a plenary session ollowing the group work.

Phase 5: Synthesis of workshop results – preparation of guidance document

During the last step o the project, results and eedback obtained during the

second workshop were used to elaborate a synthesis o the relevant ethical

and ecological aspects when evaluating impacts o GM crops on biodiversity.

The results o the our group works and the eedback collected by experts were

used to improve the approaches proposed (see sections 8 – 9). The project results were revised and compiled into a guidance document that can be used or an

inormed decision-making process or the evaluation o impacts o GM crops

on biodiversity. The guidance document summarizes criteria to assist decision-

making on the relevance o impacts o GM crops on biodiversity.

2.6 outin th rprt

The rst part o the report contains an executive summary (chapter 1), a general

introduction (chapter 2) and a general denition o the term damage (chapter 3)that is mainly infuenced by ethical considerations.

The second part o the report (chapters 4 – 8), which was written by Agro-

scope ART, is primarily dealing with the ecological conception o environ-

mental damage:

Chapter 4 presents the general denition o the term “biodiversity,” intro-

duces the dierent components o biodiversity and discusses the diverse moti-

vations to preserve biodiversity. Moreover, environmental protection goals

as specied by Swiss legislation are described with regard to agriculture in

INTRODUCTION 23




general and more specically in view o the use o genetic engineering. Finally,the ecological relevance o biodiversity is discussed or species and habitat

diversity and or ecosystem unctions.

Chapter 5 presents an approach as to how operational protection goals or the

evaluation o impacts o GM crops on biodiversity could be dened. A matrix is

introduced that lists all entities o biodiversity that require protection. The matrix

lists actors that need to be considered when dening corresponding assessment

and measurement endpoints. It is stressed that clearly dened endpoints are

necessary or regulatory decision-making as they speciy what deserves protec-

tion. It is recommended that one dene indicators and parameters that can bemeasured to determine whether harm to the protection goals specied occurred.

Chapter 6 discusses the issue o using thresholds in environmental deci-

sion-making. It is recognized that the legal rameworks regulating the use o

GMOs operate according to the thresholds-concept (i.e., risks are acceptable as

long as they do not exceed a certain specied threshold). It is nevertheless also

emphasized that regulatory authorities do not provide clear threshold values,

which challenges decision-making processes.

Chapter 7 discusses how a baseline or the comparison o dierent agri-

cultural management practices could be dened to determine whether thecultivation o GM crops is better, equal or worse than current practices. Pest

management in maize is used as a case study to compare dierent GM and

non-GM based cropping systems. It is concluded that it is impossible to per-

orm a generic comparison o dierent pest management practices, mainly

because these are regulated based on dierent legal rameworks. Instead, an

approach is proposed that allows dierentiating between intended and unin-

tended eects o the pest management practice applied. The proposed fow

chart can urther be used to determine which type o unintended eects may

represent environmental damage.Chapter 8 uses the eects o dierent weed management practices in maize

to discuss how one can cope with rather vaguely denable damages on arm-

land biodiversity that might occur rom changes in agricultural management

practices. Since long time periods may elapse beore these damages become

apparent, appropriate risk management options might be a valuable option to

Reduce the risks o these damages occurring.

The third part o the report (chapter 9), which was presented by Ethik im

Diskurs, discusses the topic rom the ethical point o view.

24




Chapter 9 introduces the ethical reerence system (ERS) as a useul toolor regulatory decision-making that allows speciying and justiying plausible

ethical criteria or the evaluation o environmental impacts o GM crops on

biodiversity. It is emphasized that the ERS provides regulators with a general

orientation grid by describing the dierent ethical theories that may underlie

the protection o biodiversity. The ERS helps to determine which ethical

position one intuitively avors and acilitates a more refected understanding o

the ethical views enshrined in the law.

The report ends with a chapter summarizing general recommendations to

decision-makers (chapter 10)Finally, Annex 2, which was presented by Ethik im Diskurs, character-

izes the term “risk” and discusses it rom a general point o view, and more

specically in the context o the given research questions. Although the report

mainly ocuses on the question as to how one can nd criteria to determine

what constitutes environmental damage, the term “risk” is a recurring topic

when discussing denitions o damage. Under certain circumstances (e.g., i

it is dicult to clearly characterize which impacts constitutes damage),

relying on risk mitigation measures may be an adequate measure to approach

this question.

INTRODUCTION 25




a GeNeRal DeINITIoN

o DaMaGe

CHAPTER 3




28

3 a GeNeRal DeINITIoN o DaMaGe

In everyday language, something is called damage or harm i state S2 repre-sents a change that is valued negatively compared to an initial state S1. I, or

instance, a house burns to the ground and its inhabitants lose their belongings,

or i a healthy person alls ill and suers rom the illness, this is called damage

or harm. It is important to realize that damage is an evaluative notion, even

though the evaluation is based on the establishment o acts. For this reason,

there can be no value-ree scientic risk assessment. Science can ascertain

the actual state o something and it can describe and explain changes and its

related ecological implications. However, science can only answer the ques-

tion rom a scientic perspective as to how this state or these changes shouldbe evaluated.

To be more exact, damage as an evaluative notion does not denote a

change that is evaluated negatively, but a change that ought to be evaluated

negatively. In order to determine whether a state or an event is damage, we

need a reerence system. The purpose o this system is to name the values

or goods that allow justied negative evaluations o changes. These may be

“zero values” such as reedom rom pain and suering (whereby the negative

change essential or damage would be the occurrence o pain and suering)

or positive values such as pleasure, beauty, bodily integrity or lie (in the senseo being alive).

The evaluative dimension o the notion o damage consists o two aspects.

On the one hand, suering damage is always bad or the individuals (entities)

aected. That is the prudential dimension which reers to what is good or bad

or an individual living organism. On the other hand, damage is usually some-

thing that should not to be inficted on others or something the others should

be protected rom. This is the ethical dimension. It is also the normative basis o

the legal understanding o damage.

From an ethical point o view, damage is not necessarily bad or repre-hensible. Rather, damage is either neutral or relevant. “Neutral” means that

nobody can be held responsible or it. A congenital disease, or instance,

may damage the aected child, even though it is the unpredictable result o

a natural genetic process. “Relevant” means that the damage is to be evalu-

ated negatively or positively in a moral sense. I the evaluation is negative, the

infiction o damage is morally inadmissible; i it is positive, the infiction is

morally admissible or even required. The latter is the case, or example, when

a teacher gives a student a bad grade. This may damage the student: she may




A GENERAL DEFINITION OF DAMAGE 29

eel bad or have to repeat a course. However, i the grade is adequate to herperormance, the infiction o damage is morally justied. On the other hand,

torturing someone just or the un o it implies inficting a kind o damage that

is morally prohibited.

What kind o entities can be damaged? In everyday language, there are

dierent kinds o damage: economic damage, political damage, environmental

damage, damage to persons, animals, plants or engines, construction damage,

re damage – to name just a ew. These damages can be roughly classied into

three main groups: systemic damage, unctional damage and damage o indi-

vidual beings. Environmental damages, or instance, are systemic damages,engine damage is a unctional damage and damage to persons is a damage o an

individual being (a person).

I we were to undertake a thorough and comprehensive conceptual analysis

o the notion o damage, the ollowing questions would have to be answered:

• Can there be systemic damages? Is it plausible to assume, or instance, that an

ecosystem as such is damaged i its homeostatic equilibrium is disrupted?

• Can a loss o biodiversity as such be damage?

• Can plants or any other nonsentient living beings be damaged as such?

• Can nonliving entities (a car) or parts thereo (a car engine) be damagedas such?

In order to answer these questions, one would have to clariy what condi-

tions must be met or the correct application o the notion o damage. This

would be tantamount to an ontological analysis o the conditions o damage.

The main questions to be tackled would be:

• Can nonexisting entities (or instance, dead persons) be damaged as such?

• I only existing entities can be damaged as such, what is an existing entity?

• I something exists only nominally, that is, as an abstract entity (such as, orexample, an ecosystem), can it be damaged as such?

• Does damage presuppose the ability o experiencing it?

This analysis o the general notion o damage would be necessary beore

proceeding to the analysis o the specic notion o environmental damage.

Unortunately, most analyses perormed by ecologists and ethicists do not

ollow this two step procedure. This is why their discussion o this topic is

rather unsatisactory.5

5 One would be able to show that none o the conceptions or criteria o ecological

damage that play a role in this discussion, especially with regard to GMOs – such as,

evolutionary integrity, selective advantage, natural variation, similarity, impairment

o biodiversity – are convincing.Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



cHaRacTeRIZING THe PRoTecTIoN

Goal “BIoDIVeRsITY”

CHAPTER 4




34

4 cHaRacTeRIZING THe PRoTecTIoN Goal “BIoDIVeRsITY”

4.1 Bkgrund

Protection goals as specied by existing legislation are the exclusive starting point

or regulators or a denition o damage related to the evaluation o GM crops.

Typically, legal rameworks make relatively vague specications on the question

regarding what it is to be protected rom harm resulting rom human activities. Both

rom an ethical and ecological point o view, the legislative terms used to describe

the protection goals “environment” and “biodiversity” are however too vague to be

scientically assessed. The use o criteria to evaluate damage necessitates nding

an operational way o how to characterize the protection goal “biodiversity.” Ina rst step, the term “biodiversity” is thus examined more closely, ollowed by a

discussion o the environmental protection goals as specied by Swiss legisla-

tion both rom a general point o view and more specically in relation to the use

o genetic engineering. Finally, a proposal is made as to how the protection goal

“biodiversity” in agricultural landscapes could be operationally dened.

4.2 Wht i bidivrity?

4.2.1 Gnr dinitinAlthough the term biological diversity – or biodiversity – has been extensively

used in the past decades, the underlying concepts and denitions are anything but

uniorm. Probably one o the most prevalent denitions is given by the Conven-

tion on Biological Diversity (CBD) stating “Biological diversity means the vari-

ability among living organisms from all sources including (…) terrestrial, marine

and other aquatic ecosystems and the ecological complexes of which they are part;

this includes diversity within species, between species and of ecosystems” (CBD,

1992). As such, biodiversity is essentially an abstract concept, albeit one o which

most would say that they have some intuitive understanding. Biodiversity is otentermed as the variety o lie on earth and the natural pattern it orms (CBD, 2000).

The range o possible interpretations o such a conception o biodiversity is not

simply wide, but it is so wide that it becomes exceedingly dicult to comprehend

(Gaston, 1996). As a consequence, the concept o biodiversity is imprecise and it

risks being dened so broadly that it equates to the whole o biology. Unortu-

nately, a denition treating biodiversity as more or less all living things on earth

is o little use to policy-making where alternatives have to be selected in light o

given conditions to guide and determine present and uture decisions.




CHARACTERIZING THE PROTECTION GOAL “BIODIVERSITY” 35

4.2.2 Ditinguihing btwn dirnt mpnnt bidivrity A number o schemes have been proposed to distinguish the major eatures o

biodiversity and to better characterize what constitutes the “variety o lie.” A more

specic denition is given by Redord and Richter (Redord and Richter, 1999),

dening biodiversity as “the variety of living organisms, the ecological complexes in

which they occur, and the ways in which they interact with each other and the phys-

ical environment.” The authors propose to dene biodiversity in terms o dierent

components – genetic, population / species and ecosystems – each o which has

compositional (the identity and variety o elements), structural (the physical

organization or pattern o elements) and unctional attributes (ecological andevolutionary processes). The three latter attributes reer to an approach initially

suggested by Noss (Noss, 1990) to characterize biodiversity in terms o ecological

processes recognizing that biodiversity is not simply the number o genes, species,

ecosystems or any other groups o things in a dened area (as dened by the (CBD,

1992). Although such a scheme may be helpul in acilitating a practicable approach

to rening the concept o biodiversity, one has to recognize that even the rened

concept (Table 1) has its operational limits. Many o the terms used to describe

the dierent attributes o biodiversity are still too broad to be eectively used in a

decision-making process that aims at quantiying biodiversity changes.

4.2.3 Mtivtin t prrv bidivrity

The primary motivation to preserve biodiversity is oten purely in the sel-interest

o mankind (CBD, 2000). The global loss o biodiversity is said to threaten ood

supply, the source o wood, medicine, energy and essential ecological unctions

as well as opportunities or recreation and tourism.

The Secretariat o the CBD lists a vast number o “goods and services”

provided by ecosystems:

• Provision o ood, uel and ber • Providing o shelter and building material

• Purication o air and water

• Detoxication and decomposition o wastes

• Stabilization and moderation o Earth’s climate

• Moderation o foods, droughts, temperature extremes and the orces o wind

• Generation and renewal o soil ertility including nutrient cycling

• Pollination o plants including many crops

• Control o pests and diseases




36

• Maintenance o genetic resources as key inputs to crop varieties and live-

stock breeds, medicines and other products • Cultural and aesthetic benets

• Ability to adapt to change

Today, ecosystem services are usually described according to the deni-

tion given by the Millennium Ecosystem Assessment (Millennium Ecosystem

Assessment, 2005; EASAC, 2009). Ecosystem services are thereby dened as

the benets people obtain rom ecosystems and they are categorized into our

broad categories:

Table 1: Concept o the term “biodiversity” according to its three main components and the

corresponding attributes (based on Noss, 1990; Kaennel, 1998; Redord and Richter, 1999).

Gnti divrity

Phenotypic diversity

Variety, cultivar

Subspecies

Genetic structure

Gene fow

Genetic processes

spi divrity

Species richness

Species abundance

Species evenness

Species density

Endangered species

Threatened species

Population structure

Surrogate species

Indicator species

Keystone species

Flagship species

Umbrella species

eytm divrity

α, β, γ-diversity

Landscape types

Communities

Ecosystems

Landscape patterns

Habitat structure

Landscape processes

Land-use trends

Ecosystem processes /

services

Parasitism, predation

Pollination

Soil processes

Nutrient cycles (C, N, P, S)

Biomass production

cmpitin

(Identity and variety

o elements)

strutur

(Physical organization,

pattern o elements)

funtin

(Ecological and

evolutionary processes)





• Supporting services, which provide the basic inrastructure o lie, includingthe capture o energy rom the sun, the ormation and maintenance o soils

or plant growth and the cycling o water and nutrients (these services

underlie all other categories).

• Regulating services, which maintain an environment assisting human

society, managing the climate, pollution and natural hazards such as disease,

food and re.

• Provisioning services providing the products on which lie depends (ood,

water, energy) and the materials that human society uses or ashioning its

own products. • Cultural services providing landscapes and organisms that have signicance

or humankind because o religious or spiritual meanings they contain or

simply because people nd them attractive.

The need to preserve biodiversity is moreover oten linked to dierent

obligations (Kunin and Lawton, 1996; SCNAT, 2006):

• Moral and ethical obligations: biodiversity represents a heritage that has to

be preserved or uture generations.

• Human well-being: many organisms (fowers, birds, butterfies) bringpleasure to many people and enrich their lives. Biodiversity is urthermore

important or personal recreation and regeneration as people enjoy and

relax in a natural, diverse environment.

• Insurance: biodiversity can be useul as a potential resource or survival

in a changing environment by providing new drugs, ood stus or genetic

resources or crops and arm animals.

• Ecosystem services: organisms provide essential services maintaining the

lie-support systems o the planet (see above).

• Economy: biological resources are the pillars upon which human civiliza-tions are built as they support such diverse industrial sectors as agriculture,

pharmaceuticals, construction, waste treatment and tourism.

• Preservation o the homeland (“Heimat”): Species and habitats in a particular

country constitute an important part o the homeland and create identity.

Most people would certainly agree that a number o the mentioned obli-

gations are justied, but there are inevitably also debates on the question

whether all o these obligations are valid and which are to be preerred over




others. However, such a discussion opens up a vast number o topics that arenot relevant or the ultimate purpose o the present report. Discussing the

dierent value positions relevant to the dierent motivations and determining

which are ethically relevant and deendable is thereore beyond the scope o

this report. The presented Lists are thereore to be seen as an unranked list

o values that shape the discourse about policy goals when trying to nd an

operational denition o biodiversity (see section 5).

4.3 envirnmnt prttin g piid by swi gitin

4.3.1 Gnr nvirnmnt prttin g rtd t bidivrity

In Switzerland, the protection o biodiversity is laid down in a number o

national legislations based on international environmental treaties6 and on the

Swiss Federal Constitution. The overall environmental policy goal is thereby

to preserve and to promote native species and their habitats. The legislation is

complemented by a number o enactments o the Swiss Federal Council that

have a binding status or the ederal authorities such as the “Landschatskonzept

Schweiz” (landscape concept Switzerland) (Buwal, 1999). Therein, a number o

more detailed environmental policy goals are specied. In general, anthropogenicinfuences on the environment should not lead to additional Red List species and

to a reduction o widespread species. Moreover, threatened species and their

habitats should be preserved, their conservation status should not decrease and

the number o Red List species should diminish yearly by 1%. Detailed lists o

protected species and habitats are listed in respective legal texts (NHG, SR 451;

NHV, SR 451.1).7

4.3.2 envirnmnt prttin g rtd t griutur

According to the Swiss Federal Constitution, agriculture has our main goals: (1)to assure the provisioning o the population, (2) to preserve the natural resources

on which lie depends, (3) to maintain cultural landscapes and (4) to support a

decentralized settlement o the country. In order to achieve these policy goals,

the Federal Oce or the Environment (FOEN) and the Federal Oce or Agri-

culture (FOAG) specied a number o environmental policy goals or the agricul-

tural sector. The policy goals are dened based on current legal requirements as

6 Such as the Convention on Biological Diversity, the Convention on the Conservation

o European Wildlie and Natural Habitats and the International Treaty on Plant

Genetic Resources or Food and Agriculture

7 Similar legal texts exist in the European Union, e.g., the Council Directive 92/42/EEC

o 21May 1992 on the conservation o natural habitats and o wild auna and fora

(European Commission 1992)

38




laid down in various acts, ordinances, international treaties and decisions o theSwiss Federal Council (Bau/BLW, 2008). Based on these premises, agriculture

has to contribute substantially to the conservation and promotion o biodiversity.

Biodiversity is thereby classied into three main aspects: (1) species and habitat

diversity, (2) genetic diversity within species and (3) unctional biodiversity.

4.3.2.1 Species and habitat diversity

An important policy goal o Swiss agriculture is to preserve and promote native

species that are typical or agricultural landscapes (i.e., species that mainly

occur on agricultural land or depend on agricultural use). The environmentalpolicy goals or agriculture give very precise indications on which species

and habitat deserve protection within agricultural landscapes as two types

o species groups are specied deserving special considerations: (1) target

species are locally or regionally occurring species that should be preserved

and promoted as they are threatened on a national level. In addition, Swit-

zerland carries particular responsibility within Europe or the conservation

o these species, while (2) character species are characteristic or a particular

region and they are representative or a specic habitat (i.e., they serve as an

indicator or quality o the habitat they occur in). The environmental policy goals or agriculture moreover provide detailed species lists or both target and

character species that were elaborated by a working group involving multiple

stakeholders rom agriculture and nature conservation (Bau/BLW, 2008). The

lists cover a large number o taxa that deserve special consideration. The aunal

taxa include mammals, birds, reptiles, amphibian, beetles, bees, butterfies, lace-

wings, dragonfies, grasshoppers and mollusca, while the foral taxa consist o

fowering plants, erns, mosses, lichens and ungi. The report urther species a

number o habitats that should be preserved and promoted as they are typical

or agricultural land uses (e.g., ecological compensation areas, meadows andpastures, hedgerows, ruderal areas etc). Given that the character species listed

are representative or a wide range o these habitats, the approach combines

the conservation o species and habitat diversity.

4.3.2.2 Genetic diversity within species

Genetic diversity is a prerequisite or the long-term survival o wild species.

The Swiss environmental policy goals or agriculture demand the preserva-

tion o genetic diversity o the species and habitats that are to be protected as





set out above. In addition, the national action plan on the conservation andsustainable use o plant genetic resources or ood and agriculture aims at

preserving a broad genetic diversity o the crop plants present in Switzerland

and their wild relatives in view o ood production and security. Plant genetic

resources are the starting point or any plant breeding program that aims at

developing tailor-made varieties or uture needs. Apart rom economic bene-

ts, plant genetic resources also have ecological value (e.g., by being adapted

to local conditions such as being resistant to plant diseases) and cultural value

(e.g., by representing traditional regional production).

4.3.2.3 Functional biodiversity

According to the Swiss environmental policy goals or agriculture, agricultural

production has to preserve the ecosystem services provided by biodiversity. The

part o the biosphere providing the desired ecosystem services is termed “unc-

tional biodiversity.” Human society obtains a wide array o important benets

rom biodiversity and associated ecosystems. These ecosystem services are

essential to human well-being and to sustaining lie on earth since these serv-

ices operate on such a large scale, and in such complex ways, that most services

could not be replaced by technology (Daily, 1999; CBD, 2000; Daily et al., 2000).Functional biodiversity covers ecosystem services such as soil ertility, natural

pest regulation and pollination by insects (see section 4.2.3). The economic value

o 17 ecosystem services or the entire biosphere has been estimated to $33

trillion per year (Costanza et al., 1997). The production o 84% o crop species

cultivated in Europe, or example, directly depends on insect pollinators, espe-

cially bees (Gallai et al., 2009), while predators and parasitoids ulll relevant

ecological unctions by contributing to the natural regulation o arthropod pest

populations within crop elds in agricultural landscapes. The total economic

value o pollination worldwide is estimated to amount to €153 billion, whichrepresented 9.5% o the value o the world agricultural production used or

human ood in 2005 (Gallai et al., 2009). The value o natural pest control attrib-

utable to insects in the United States is estimated to be US $4.5 billion annually

(Losey and Vaughan, 2006).

4.3.2.4 Programs to assess Swiss environmental policy goals or agriculture

To determine whether the current Swiss environmental policy goals or agricul-

ture are met, there is a need or programs assessing the status o the dierent

40




components o biodiversity. The status o implementation o these programs varies considerably among the dierent components o biodiversity. Several

programs are running that assess the state o species and habitat diversity

in Switzerland. Instruments assessing species and habitat diversity include

among others the Red List species, the Biodiversity Monitoring Switzerland

(BDM, 2009), and the Swiss Bird Index (Keller et al., 2008). These instruments

assess a number o dierent indicators and allow one to obtain a relatively

good estimation o the state o species diversity and to a lesser extent o habitat

diversity. Only a ew activities have been started or genetic diversity. Genetic

diversity is primarily collected and inventoried or arable crops, vegetables,ruit trees and or orage grasses as part o the national action plan on the

conservation and sustainable use o plant genetic resources or ood and agri-

culture. Although not yet collected, there are plans to inventory the genetic

diversity o wild plants that are relatives o crop plants and or wild plants used

or medicinal purposes or as ornamental plants (Häner and Schierscher, 2009).

No inormation is currently available on the state o unctional biodiversity as

no programs or its assessment are implemented.

4.3.3 spii prttin g rtd t gnti nginring In Switzerland, the use o genetic engineering is regulated on the constitutional

level (BV, SR 101). According to Article 120 BV, humans and their environment

shall be protected against the misuse o genetic engineering. The rationale

or this specic regulation is ounded on the novelty o the technology and

the uncertainties related to the consequences o the transormation process

o genetically modied organisms (GMOs). The Swiss Federal Law relating to

Nonhuman Gene Technology specically prescribes the protection o humans,

animals and the environment rom abuses o genetic engineering (GTG, SR

814.91). In particular, the law mandates that GMOs intended or use in theenvironment may only be marketed i experiments in contained systems or

eld trials have shown that they: (a) do not impair the population o protected

organisms or organisms that are important or the ecosystem in question; (b) do

not lead to the unintended extinction o a species o organism; (c) do not cause

severe or permanent impairment o nutrient fows; (d) do not cause severe or

permanent impairment o any important unctions o the ecosystem in ques-

tion, in particular the ertility o the soil; and (e) do not disperse, or rather, their

traits do not spread in an undesired way.





4.4 Th gi rvn bidivrity

4.4.1 spi nd hbitt divrity

Species richness is oten used as a surrogate or overall biodiversity even though

the nature and identication o species continues to be controversial, though

no one thinks that species richness is all there is to biodiversity (Maclaurin and

Sterelny, 2008). Habitats with a high species richness are requently regarded

as being more valuable than habitats with a lower species richness. However,

species richness is not always an appropriate indicator or the ecological value

o habitats as the habitat type plays an important role in the evaluation. Gener-ally spoken, species-poor habitats oten contain common, widespread species,

whereas species-rich habitats are more likely to contain rare and threatened

species. Yet, some habitats such as moorland are characterized by being inher-

ently species poor and by containing particularly adapted, rare species. Certain

disturbances, such as increased nutrient inputs, might lead to a local increase in

biodiversity in these habitats due to the invasion o less specialized and more

widespread species that would not be able to survive in the originally nutrient-

poor environment. As the invading species are generally more competitive,

they usually replace the initial, rare species that can typically be ound in thisparticular habitat type. From a nature conservation point o view, the increased

biodiversity ound in these typical habitats is not desired as the specialized, rare

species represent a particularly valuable component o biodiversity (Duelli, 1994;

Kägi et al., 2002). Similarly, recent results o the Swiss Biodiversity Monitoring

showed a slight increase in species richness in meadows when compared to the

rst assessment in 2001 (BDM, 2009). However, the analyses also showed that the

increase was mainly due to the occurrence o widespread and generalist species

already present in intensively managed, nutrient rich meadows. The net increase

in species richness was thus not particularly valued as rare and specialist speciescharacteristic or agricultural landscapes were missing.

Further examples show that increases in species richness are not always

regarded as a positive development rom a nature conservation point o view.

Despite a global perception o declining biodiversity, managers are requently

presented with situations where biodiversity is increasing due to the arrival o

invasive species (Thompson and Starzomski, 2007). Invasions by alien species

are however regarded as being one o the major direct causes o biodiversity

loss (among other actors such as land use change, pollution, unsustainable

42




natural resources use and climate change) (Slingenberg et al., 2009). Losseso native species may not result in net changes in biodiversity i exotic species

move into an area and replace them. To the public and conservation managers,

however, the loss o particular, valued native species is o concern, more so than

the net change in biodiversity (Thompson and Starzomski, 2007).

4.4.2 eytm untin nd pi divrity

Rationales or the protection o biodiversity are oten based on the argument

that species provide ecological goods and services that are essential or human

welare. Some consensus appears to be crystallizing on the overall signicanceo maintaining ecosystem integrity and unction (Gaston, 1996). This puts more

emphasis to ecological processes than is immediately evident rom many deni-

tions o biodiversity, or example, the one put orward by the CBD (CBD, 1992)

that does not encompass unctional diversity enabling most ecosystem services.

The preservation o unctional diversity also necessitates other conservation

strategies than would be the case i primarily individual species would have to

be preserved. One o the central questions in ecology is thereore how biodiver-

sity relates to ecological unctions. Several hypotheses have been ormulated to

describe the relationship between ecosystem unctions and species diversity: • Diversity-stability hypothesis (Macarthur, 1955; Elton, 1958): Early theo-

ries established the axiom that diverse ecological communities are the most

stable. The rate o ecosystem processes is highest with the greatest number

o species. As the number o species in the system increases, the ability to

recover rom disturbances is higher.

• Rivet hypothesis (Ehrlich and Ehrlich, 1981): The rivet hypothesis assumes

that the loss o species has an increasingly critical eect on the unction o

an ecosystem. All species are equally important in maintaining the system’s

stability and every species has an equal and additive eect on unction. Whilethe eect o the loss o one species may be relatively small, the loss o several

species can lead to the complete collapse o the unctioning ecosystem.

• Redundancy hypothesis (Walker, 1992): This hypothesis predicts that not

all species are o concern as some are redundant. A loss o species rich-

ness may be o little consequence to ecosystem unctioning because unc-

tional groups comprise many dierent species. Losing most species may

have little eect, but the loss o some species may result in a dramatic

destabilization o the system.





• Idiosyncratic hypothesis (Lawton, 1994): This hypothesis suggests that thereis no clear relationship between species number and ecosystem unctions

since the contribution o species to unction is unpredictable as both the

identity and order o species loss will aect unction dierentially. Ecosystem

unction changes when diversity changes, but the magnitude and direction

o change are unpredictable as the roles o individual species are complex

and varied.

• Insurance hypothesis (McNaughton, 1977; Naeem and Li, 1997): According

to this hypothesis, biodiversity insures ecosystems against declines in their

unctioning. More diverse ecosystems are more likely to contain some speciesthat can withstand perturbations and thus compensate or unctional loss o

other species being reduced or eliminated by the perturbation.

There is rather little scientic evidence rom experimental studies

supporting both the diversity-stability hypothesis and the rivet hypothesis.

Evidence in support o a linear dependence o ecosystem unction on diver-

sity (i.e., that even rare species contribute to unction) is practically nonex-

istent (Schwartz et al., 2000). There is nevertheless convincing evidence that

species-rich systems deliver ecosystem services more reliably than species-poor ones. Empirical data suggests that the ecological stability is generated

more by a diversity o unctional groups than by species richness, which

supports the insurance hypothesis (Peterson et al., 1998). Ecosystem serv-

ices typically do not depend on the presence o specic species, especially

not rare, narrowly distributed species (Maclaurin and Sterelny, 2008). At

least among species within the same trophic level, rarer species are likely

to have small eects. Species are thus not o equal importance to ecosystem

services and species richness per se is not necessarily a key element o

ecosystem unctioning. Moreover, landscape complexity may compensateor biodiversity loss because o local management intensity. The diversity o

arable weeds, or example, is higher in organic than in conventional elds,

but only in simple landscapes, as landscape diversity enhances species diver-

sity in conventional elds to a similar diversity level (Roschewitz et al., 2005).

In contrast to what may be expected, introducing diverse habitats (and less

intensive practices such as organic arming) has only a great eect in simple

landscapes and will positively infuence resilience (i.e., the capacity to main-

tain ecosystem services ater disturbance), whereas complex landscapes are

44




already characterized by a high biodiversity sustaining ecosystem services(Tscharntke et al., 2005). Negative impacts o intensive arming such as herbi-

cide applications only occur in simple landscapes where colonization o arable

weeds rom the surrounding is limited, whereas complex landscapes appear to

mitigate local anthropogenic weed elimination.

Hence, ecosystem properties may be insensitive to species loss as (i)

ecosystems may have multiple species carrying out similar unctional

roles, (ii) some species may contribute relatively little to ecosystem prop-

erties, or (iii) properties may be primarily controlled by abiotic environ-

mental conditions (e.g., weather events) (Hooper et al., 2005). The mostdramatic changes in ecosystem services are likely to come rom altered

unctional compositions o communities and rom the loss (within the

same trophic level) o locally abundant species rather than rom the loss

o already rare species (Diaz et al., 2006). Interestingly, most conservation

eorts concentrate on rare species. Rare species, however, may not interact

strongly with other species in the ecosystem, nor be able to replace the eect

o a dominant species. In conservation terms, it is more crucial to identiy

which species are likely to have disproportionate eect i they are lost. Espe-

cially in agroecosystems, species richness is oten less important or ecosystemunctions than the presence o a small subset o species (Shennan, 2008).

High diversity o unctional groups may allow reorganizations ater distur-

bances due to a higher number o insurance species. Agricultural landscapes

must be a mosaic o well connected early and late successional habitats to

support a high biodiversity and thereby the capacity to recover rom minor

and major small- and large-scale disturbances (Tscharntke et al., 2005). From

an ecosystem service conservation perspective, it is thus more important to

protect species which have no unctional equivalent and to maintain diver-

sity within unctional groups than to ocus on conserving particular individualspecies (Thompson and Starzomski, 2007).

4.4.3 summry th gi rvn bidivrity

In the ollowing, we will argue that there are two main motivations or the

conservation o biodiversity: (1) the protection o rare and threatened species

and (2) the unctioning o ecosystem services based on genetic and species diver-

sity (ecological resilience). Genetic diversity (i.e., biodiversity within species) is

thereby crucial to a species’ ability to adapt to its environment. An important





oPeRaTIoNal DeINITIoN

o PRoTecTIoN Goals

CHAPTER 5




50

5 oPeRaTIoNal DeINITIoN o PRoTecTIoN Goals

5.1 Diiuty t ind n unmbiguu dinitin th trm bidivrity

One o the goals o the present study is to obtain a practicable conception o biodi-

versity that can be used to describe the agricultural context prevalent in Swit-

zerland. Scientically, the term “biodiversity” can be dierentiated into dierent

components (genetic, species and habitat diversity) and into dierent attributes

such as composition, structure and unction (see section 4.2). Although the scien-

tic classication provides a more precise denition o biodiversity, the rene-

ment also substantially broadens the conception o biodiversity without acili-

tating its operational use in decision-making. Despite its multiple dimensions,biodiversity is usually equated with species richness only (Feld et al., 2009). Not

surprisingly, the unctional, structural and genetic components o biodiversity

are still poorly addressed. This may, to a certain extent, be due to scientic limita-

tions o nding appropriate indicators or these components, but also refects the

act that biodiversity assessment and monitoring until recently has been mainly

driven by conservation biologist. This becomes apparent when looking at the

programs implemented in Switzerland to assess biodiversity (see section 4.3.2).

The biodiversity components most oten measured today are species richness

and habitat diversity, while genetic diversity is primarily covered by assessingthe number o livestock breeds and crop varieties (BDM, 2009). Moreover, there

are literally no routine monitoring programs that assess ecosystem services or

ecosystem unctions.

I even a rened concept o biodiversity (see Table 1) remains dicult to

apply in practice, how can the concept then be made operational to support policy

decisions? The variety o dierent denitions shows that biodiversity cannot be

dened as a clearly measurable quantity, that is, it is impossible to provide an

index allowing to rate ecosystems or collections o entities according to their

degree o diversity (Norton, 2008). One has to recognize that it is probably simi-larly impossible to nd an unambiguous scientic denition o the term biodi-

versity. The goal should thus be to nd an approach that enables stakeholders

agreeing upon the way orward. The denition must allow one to nd a common

language that allows communication about what deserves protection because

it is specically valued. As with other metaconcepts, such as, sustainability or

ecosystem health, biodiversity achieves operative sense only when it is clearly

dened and used as a means to inorm specic management decisions in specic

ecosystems using specic indicators (Failing and Gregory, 2003).




OPERATIONAL DEFINITION OF PROTECTION GOALS 51

Focusing on the management context necessitates going beyond generaldenitions to set scale and context-specic management objectives. At this stage,

managers (i.e., regulatory authorities) need to deal with the dicult question o

what is desired. Ultimately, policy decisions on what is to be protected are based on

the legal rameworks regulating the conservation o biodiversity (see section 4.3).

Nevertheless, policy decisions are inevitably linked to practical and nancial

constraints, making it impossible to conserve all components o biodiversity in

the same manner or in the way it would be desired. The protection o biodiversity,

as laid down in the legislation, is a man-made concept refecting social values.

When thinking about an apparently scientic concept such as biodiversity, oneis orced to conclude that the concept remains ambiguous until it is clear what

society is caring about (Norton, 2008). One could thus argue that a policy-rele-

vant denition has to be guided by both social and scientic criteria as what

deserves protection is also dened by what people care about. Determining

what truly deserves protection will thus inevitably be based on the motivations

or biodiversity conservation, on the relevance that is given to biodiversity to

serve these motivations (see section 4.2.3) and on the implicit values under-

lying the protection o biodiversity (see Box 2). Science and empirical evidence

can thereby guide the discourse about policy goals and indicate what ecologicaltheories tell us about the relevance o biodiversity (see section 4.4).

5.2 a mtrix r n prtin dinitin bidivrity in griutur ndp

Protection goals, as laid down in the existing legal rameworks, are usually

described too vaguely to be scientically assessed and to support regulatory

decisionmaking. In the ollowing, a matrix or an operational denition o protec-

tion goals is proposed that consists o a three-step process speciying protection

goals, assessment endpoints and measurement endpoints (Table 2). The rst twosteps allow a more explicit characterization o the protection goal “biodiversity

in agricultural landscapes” by speciying a number o assessment endpoints

that describe the entities deserving protection in more detail. In the third step,

measurement endpoints are dened that represent a measurable ecological

characteristic, which can be related to the assessment endpoint chosen. The

matrix presented herein will not be able to deliver a denite answer to the chal-

lenge o dening ully operational protection goals. The matrix should rather

be seen as a tool that supports an operational denition o biodiversity. A true




52

operational denition o protection goals will need to be elaborated in a trans-parent process involving all relevant stakeholders (regulators, applicants and

scientists) (see section 5.3). The matrix is urthermore not restricted to GM plants

and may thus orm the basis or any evaluation o environmental impacts on

biodiversity in agricultural landscapes.

5.2.1 Dinitin prttin g

The rst step or the denition o protection goals aims at identiying the dierent

areas o protection to be considered as laid down in existing legal rameworks.

The area o protection has been divided into the two principal categories, that is,“biodiversity conservation” and “ecosystem services” (Table 2). Biodiversity conser-

vation covers red list species and species o high conservation or cultural value

rom a range o dierent taxa including mammals, birds, amphibians, insects (such

as butterfies) and plants. In addition to species diversity, biodiversity conservation

also involves protected habitats and landscapes.

Protection in this category primarily ocuses on habitats as the term “land-

scape” is oten less clearly dened and habitats are the units usually listed in the

legislation (European Commission, 1992; NHV, SR 451.1). The operational deni-

tion o these two categories will ultimately necessitate compiling detailed lists oreach group o species and habitats to be explicitly protected.

The second area o protection relates to ecosystem services that are essen-

tial to human existence. Ecosystem services relevant in an agricultural context

include pollination, pest regulation, decomposition o organic matter, soil

nutrient cycling, soil structure and water regulation and purication.

5.2.2 Dinitin mnt ndpint

The second step aims at perorming an operational denition or the protec-

tion goals specied in step 1 (Table 2). Each protection goal is dened by oneor more assessment endpoints (Raybould, 2006, 2007). An assessment endpoint

is thereby dened as an “explicit expression o the environmental value that is

to be protected as set out by existing legal rameworks” (Suter, 2000). Assess-

ment endpoints are thus the valued attribute o an environmental entity

deemed worth o protection. Broad assessment endpoints tend to be less

valuable than more specic ones, but endpoints should not be too restrictive,

either. It is important to note that an assessment endpoint is not an indicator

(i.e., it is not an environmental measure generated by a monitoring program





that intends to be indicative o some environmental conditions), but the thingitsel to be protected. An operational denition o assessment endpoints neces-

sitates dening the ollowing criteria:

• An ecological entity being representative o the area o protection selected

(e.g., predators and parasitoids representing pest regulation)

• An attribute to be protected, or example, “abundance” or “ecological

unction”

• A unit o protection (individuals, populations, communities, guilds9)

• A quantiable spatial scale o protection10 (GM crop eld, other arable land,

nonagricultural habitats) • A quantiable temporal scale o protection (present cropping season,

ollowing cropping season, time o consent o the GM event11)

• A denition o the type o eects that are regarded to be harmul (i.e., a rele-

vant decrease in abundance or a relevant disturbance in ecological unction)

Ideally, assessment endpoints should also satisy the ollowing criteria

(adapted rom (Suter, 2007):

• As assessment endpoints are the basis or decision-making, they should

refect policy goals and societal values • Ecological entities selected should be ecologically relevant

• Ecological entities selected should be potentially highly exposed and respon-

sive to exposure

• Assessment endpoints should be operationally defnable (i.e., it should be

clearly specied what must be measured or modeled)

• Assessment endpoints should have the appropriate scale to the site or action

assessed

• Good techniques must be available to risk assessors (e.g., standard toxicity

tests, monitoring methods) to assess assessment endpoints

9 A guild denes a group o species that exploit the same class o environmental

resources in a similar way.

10 The proposed spatial scales allow a more unambiguous classication than a

classication using eld, eld borders, landscapes as the two latter terms are oten

not clearly dened. Using biodiversity conservation o plants as an example, the

proposed scales allow one to determine whether weeds are regarded a protection

goal. Weeds are only regarded to be worth o protection in case the spatial scale

o protection is extended to GM crop elds and other arable land. In case the

spatial scale or valued plants is restricted to nonagricultural habitats, weeds areexplicitly not regarded as a protection goal.

11 According to Swiss and EU legislation, the approval o a GM variety is granted

or 10 years.




Table 2: Matrix or an operational denition o the protection goal biodiversity

or an environmental risk assessment o GM plants

54

Biodiversity

conservation

Ecosystem services /

unctions

Protected habitats

Pollination

Pest regulation

Decomposition o

organic matter

Soil nutrient

cycling (N, P)

Soil structure

Water regulation

and purication

Mammals

Birds

Amphibians

Valued insects

(e.g., butterfies)

Valued plants

Habitats listed

in legislation

Pollinating insects

Predators and

parasitoids

Soil invertebrates,

soil microorganisms

Soil microorganisms

Soil invertebrates

Fish

Aquatic invertebrates

Algae

Red List species

Species o high

conservation /

cultural value

Abundance

Ecological unction

Area o protection Ecological entity Attribute Unit o protection

Individuals

Populations

Communities

Guilds

PRoTecTIoN Goals

cRITeRIa foR THe oPeRaTIoNal DefINITIoN of THe PRoTecTIoN Goal

assessMeNT eNDPoINTs





be conducted as such that the dened risk hypothesis is conrmed with themaximum possible accuracy and probability. This requires that the hypoth-

esis is ocusing on detecting ecologically relevant eects. Apart rom proper

experimental design and statistical planning (Fairweather, 1991; Perry et al.,

2003; EFSA, 2009a), the rigorous testing o a hypothesis needs to answer the

question regarding under what conditions the existence o a supposed risk

is most likely revealed. For example, only i a particular stressor (e.g., the

Bt-toxin) causes a relatively strong eect under eld conditions, it will not

be blurred by a number o infuencing actors which will cause dierent,

overlapping smaller eects. Given that the infuence o the various actorsis hardly distinguishable, it could become very dicult to unambiguously

determine the causality between a particular eect and the actor causing

it. The likelihood to detect a relevant eect in an environmental multiacto-

rial setting (as typical or environmental monitoring programs) might thus be

much lower than detecting one in a more controlled setting with only a ew

actors involved. Hence, testing a risk hypothesis in a more controlled setting

such as a laboratory or greenhouse might generally be more rigorous than

testing the hypothesis under more ecological conditions (Raybould, 2007;

Romeis et al., 2008a).

5.3 appitin th mtrix r n prtin dinitin prttin g

5.3.1 apprhing n prtin dinitin th prttin g bidivrity

Operational protection goals are a prerequisite or regulatory decision-making

as they speciy what deserves protection. Assessment endpoints thereby speciy

the ecological entities and their related attributes to be specically protected

while measurement endpoints enable to select those indicators that are to be

used to determine with the maximum possible rigor the likelihood that harmmight occur (or has occurred) to the protection goals chosen. While the de-

nition o protection goals is essentially a generic process applicable to agri-

culture management at large, the more precise denition o both assessment

and measurement endpoints has to be carried out on a case-by-case basis or

each specic GM crop. A thorough problem ormulation is thereby an essential

prerequisite or the operational denition o these endpoints where the nature

o the crop introduced, the genetically modied trait and the receiving environ-

ment are considered (Romeis et al., 2008a; Wolt et al., 2010).




Given that protection goals are set by existing legal rameworks, theinitial proposal what to protect has to be ramed by the regulatory authorities

involved in the risk analysis o GM crops. There is general consensus that biodi-

versity protection goals, as specied by existing legislation (such as Red List

species, protected habitats, etc.) (see section 4.3), are to be protected. There is,

however, controversy surrounding the necessity o the protection o “common”

species that are not explicitly listed in the legislation, but that ulll important

ecosystem services and may be reduced by common agricultural practices. The

operational denition o protection goals should thereore ideally be dened in

a transparent process involving a dialogue between all relevant stakeholders(regulators, applicants and scientists). The matrix presented herein (Table 2)

can thereby be used as a tool to structure the dialogue, especially when dening

both assessment and measurement endpoints. The process could include stake-

holder meetings where stakeholders would compile and rank dierent conser-

vation goals and ecosystem services.

Agreeing on endpoints (and thus policy targets) or biodiversity conserva-

tion and ecosystem services is likely to be a dicult task or at least two reasons.

First, the complexity o the underlying system means that endpoints are di-

cult to dene in a way that is both operational or public policy and possible tocommunicate to policy-makers and the public. Reducing these diculties by using

vague terms such as “ecological integrity” will not solve the problem. Second, the

problems are exacerbated by the act that endpoints or biodiversity conservation

and ecosystem unctioning will sometimes contradict each other. Setting ecologi-

cally-based endpoints may require balancing competing goals (e.g., the protection

o rare species vs. the preservation o common species) that are guaranteed to

be a source o controversy both or scientists and the public. The content o the

matrix will to a great extent depend on the motivations (see section 4.2.3) and

on the value systems underlying the preservation o biodiversity (see Box 2) o

the various stakeholders involved in the operational denition o protection goals.

Stakeholders will inevitably have competing motivations and values. Failures to

disclose diering motivations and value systems will have important policy impli-

cations as the operational denition o protection goal necessitates to agree on

common policy goals. The operational denition o protection goals will thereore

inevitably also include a number o challenges. Some o these challenges or a

number o ecological entities are discussed in the next section. To address these

challenges, the motivations and the values underlying the protection o biodiversity

58




need to be explicitly stated and assessment and measurement endpoints need to beadapted to these motivations and values. The Ethical Reerence System presented

in section 9.2 can thereby be a tool supporting such an exercise.

Nevertheless, it is important to consider that stakeholder approaches have

limitations when it comes to decision making. Setting priorities and making

inormed decisions requires both scientic and value judgments (Failing and

Gregory, 2003). These are distinct judgments to be made by dierent individuals.

Nonscientic stakeholders should be asked to make only value judgments in

consultative processes, but not scientic judgments or which they are not quali-

ed and inormed. Evaluating the relevance o the data is a task that necessitatesexpert knowledge (see section 9.1). This is thus the task o scientists and not

o the public as the latter lacks the necessary knowledge. It is crucial to recog-

nize that decision-making requires that one considers sound science and not

democratic consideration o dierent viewpoints and interests o stakeholders.

This is because decision-making is an executive orce and not a legislative orce.

Democratic considerations have been considered during the drating o the

legislation, but once in place, the legislation has to be carried out by the regula-

tory authorities considering sound scientic principles.

Box 2: Dierent value systems underlying the preservation o biodiversity

The necessity to preserve biodiversity may, to a certain extent, be based on a scientic

rationale (Chapin et al., 2000; McCann, 2000; Tilman, 2000; Diaz et al., 2006), but the

question “why?” remains ultimately a normative one that has to be answered based on

the prevalent societal and ethical value systems. The process o nding an operational

denition o biodiversity will inevitably necessitate decision-makers to make a trade-o

as the dierent value systems oten contradict each other. These contradictions become

apparent i one compares three possible value systems that may have an infuence on

the motivation to preserve biodiversity: (1) intrinsic value, (2) demand value and (3)option value.

Intrinsic value

The idea that biodiversity is intrinsically valuable enjoys wide support as it refects that

we care or nature not as a resource, but rather as a good in itsel (Maclaurin and Sterelny,

2008). The preamble to the CBD, or example, gives an intrinsic value to biological diversity

and the ecological, genetic, social, economic, scientic, educational, cultural, recreational

and aesthetic values o biological diversity and its components (CBD, 1992).





Dmnd vuOne o the simplest motivations to preserve biodiversity derives rom broadly utilitarian

theories o environmental ethics (see section 9.3.1), that is, rom the idea that the moral

worth o an action is determined solely by its contribution to overall utility (e.g., to the

maximization o happiness or pleasure as summed among all persons). The value o an

ecosystem and its components thereby equates with the resources and services they provide

to humans – they have a demand value that warrants the investment required or their

conservation (Maclaurin and Sterelny, 2008). Such an approach would lead the society to

place a high value on protecting the basic ecological mechanisms on which humans depend.

A demand value approach does urthermore not tie value to diversity per se as importance is

more tied to specic uses. This may include the importance as a resource, a crucial ecolog-

ical unction or a rather obscure attribute as being loved by the general public. Following

a rather strict utilitarian theory, the demand value approach aces some challenges as the

theory can hardly aggregate individual cost benet trade-os into a collective assessment –

benets to some always impose costs on others.

optin vu

Option value, being the third utilitarian theory, links utility much more closely to diversity.

Option value is an insurance concept borrowed rom economics that is based on two basic

ideas. First, species (or ecosystems) that are not o value to us at present may becomevaluable at some point in the uture. Second, as our knowledge improves (and as circum-

stances change) we might discover new ways in which they can be valuable (Maclaurin

and Sterelny, 2008). The crucial point about option value is making diversity valuable. The

approach suggests preserving a biodiversity as rich and representative as possible as it is

not known in advance which species will prove to be important. This is probably also one o

its most obvious limitations as the approach, understood in its most broad interpretation,

would demand that everything has to be preserved just in virtue o being useul in some

conceivable uture event. This would result in the ineective goal o preserving biodiversity

in all possible respects. One solution would be to ocus not on mere possibilities, but on

probabilities that a certain eature could prove to be valuable under some circumstances

(Maclaurin and Sterelny, 2008). Species loss may not be important in terms o the role o the

species now, but rather in terms o that species being available to ulll a unctional role in

a uture environment (Thompson and Starzomski, 2007). Clearly, the option value approach

also suggests that many species do not have high enough option value to justiy major

expenditures on their conservation. The approach recognizes that many species are o great

value, but it does not imply that all species or all biological systems, are o important value

and should to be saved.

60




5.3.2 exmp r m td prttin gIn the ollowing, the proposed matrix or an operational denition o biodiversity

(see section 5.2) is completed or some selected protection goals to illustrate how

such a denition could be perormed. The operational denition thereby consists

o speciying the area o protection and o dening assessment and measurement

endpoints (Table 3). Here, the protection goals are dierentiated into the two main

areas o protection, that is, biodiversity conservation and ecosystem services.

5.3.2.1 Defnition o assessment endpoints or biodiversity conservation

The rst step in the denition o assessment endpoints aims at identiying thedierent areas o protection as laid down in existing legal rameworks. For biodi-

versity conservation, the area o protection includes Red List species and species

o high conservation and cultural value. In the present example, the ecological

entities to be protected are restricted to valued plants and valued insects (in

particular, butterfies being a species group with a high conservation value where

more than 50% o the species occurring in Switzerland are Red List species).

Red List species are dened or both plants and insects (Duelli, 1994;

Gonseth, 1994; Moser et al., 2002). Similarly, species o high conservation value

and cultural value are specied in the environmental policy goals or agri-culture speciying target species12 and character species13 (Bau/BLW, 2008)

(see section 4.3.2). The unit o protection or biodiversity conservation o all o

these species is usually specied on the basis o populations or communities

and the attribute to be protected is the abundance o the respective popula-

tions and communities.

The rst diculty when dening assessment endpoint or biodiversity

conservation is the denition o the spatial scale o protection. Logically,

the spatial scale o protection or valued plants and valued insects (such as

butterfies) should be set to nonagricultural habitats as agricultural eldscannot be regarded as typical habitats or Red List species and species o

high conservation and cultural value. Typically, the primary aim o agricul-

tural elds is the production o ood and eed and not the production o

biodiversity. Yet, the promotion or conservation o biodiversity in agricul-

tural elds is a complex topic that is hotly debated in Europe. It is argued

that in many European countries, agriculture and natural habitats are inti-

mately mixed with around 70% o the land area being classied as agricul-

tural land (Hails, 2002). Consequently, the conservation o “common” species

12 Target species are locally or regionally occurring species that should be preserved

and promoted as they are threatened on a national level.

13 Character species are characteristic or a particular region and they are

representative or a specic habitat.





and communities within the armed landscape is seen as an emerging para-digm (Marshall et al., 2003). Hence, there is controversy about the question

whether certain common species (although not being explicitly listed in the

legislation) should be regarded as a protection goal since eects on these

species might translate to higher trophic levels. These higher trophic levels

may include Red List species or species o high conservation or cultural value.

A prominent example or the confict surrounding the protection o “common”

species are agricultural weeds that on the one hand have the potential to

reduce agricultural yield, but that do also orm an essential part o ood webs

in agricultural landscapes contributing to armland biodiversity (Watkinsonet al., 2000; Heard et al., 2005). The question is thus what characterizes a

“valued” plant species. Though there is a need to promote biodiversity not

only in nonagricultural habitats, we believe that classiying certain common

species (such as weeds) as a generic protection goal might lead to a number

o conficts. Rather, a balance between adequate weed control and the oppor-

tunity to retain some plants to support biological diversity should be searched

(see section 8.2.2).

A second diculty is linked to the denition o the temporal scale o protec-

tion. On the one hand, many legal rameworks such as the Swiss Gene Tech-nology Act require that the protection o biodiversity and its sustainable use is

permanent (GTG, SR 814.91). It is moreover known that long time periods may

be needed beore ecological eects become apparent. In the UK, it took several

decades beore marked reductions in armland biodiversity were noticed (Fuller

et al., 1995; Robinson and Sutherland, 2002). On the other hand, regulatory deci-

sions have oten to be taken within shorter timerames than decades, which

makes it necessary to nd a compromise or a measurable temporal scale. We

propose to dene the temporal scale according to the time o consent that is

granted or GM events, which is currently 10 years (European Community, 2001;GTG, SR 814.91). Ater the 10-year period, regulatory authorities can decide

whether impairment in a particular protection goal has occurred and they can

use this inormation when deciding on the renewal o consent.

5.3.2.2 Defnition o measurement endpoints or biodiversity conservation

The rst step when dening measurement endpoints or biodiversity conser-

vation consist in selecting appropriate indicator species. The term indicator is

used here according to Duelli and Obrist (2003), who dened that “an indicator

62





should be a measurable portion o an entity that correlates with this larger entity. ”On the one hand, biodiversity indicators can be dened on a more general basis

to be indicative or a larger entity such as “biodiversity in agricultural land-

scapes.” For such biodiversity assessments, several indicators such as fowering

plants, birds, and butterfies are usually combined to assure the quality o the

data obtained (Jeanneret et al., 2003; BDM, 2009). On the other hand, indicator

species can also be specically selected to test a particular risk hypothesis. I, or

example, a GM crop expresses a toxin with a specic toxicity against a particular

group o organisms (such as Cry1Ab against Lepidoptera), the risk hypothesis

should ocus on testing nontarget lepidopteran species as an eect is most prob-ably occurring in this species group (Romeis et al., 2008a).

A critical aspect when dening operational protection goals or biodiver-

sity conservation is linked to the diculty to dene measurable endpoints

or Red List species. Red List species can oten not be chosen as the primary

indicator as they can either not be tested under laboratory conditions due to

their conservation status or they occur too rarely in agricultural landscapes

to be easily monitored and to allow or proper statistical analyses (Aviron

et al., 2009). Laboratory toxicity studies and monitoring programs thereore

need to rely on surrogate species that are intended to be representative or aspecic group o Red List species. Although the utility o the concept o surro-

gate species is a constantly debated issue14 both in conservation biology (Caro

and O’Doherty, 1999) and ecotoxicological testing (Cairns, 1983), the surro-

gate approach is still the standard practice mandated or environmental risk

assessment o pesticides and GMOs in most countries. Hence, the choice o

surrogate species is most likely inevitable as it is impractical to assess biodi-

versity as a whole.

The last step in the denition o measurement endpoints or biodiversity

conservation consists o the choice o appropriate parameters or early andhigher tier studies. The term parameter is hereby used as being subsidiary

to the term indicator since eects on a given indicator are oten assessed by

the measurement o several parameters that are able to show changes in the

indicator chosen (Sanvido et al., 2005) (see section 5.2.3). “Mortality,” “devel-

opment time” or “reproduction” could, or example, be possible parameters

or early tier studies, while parameters or higher tier studies could be “abun-

dance” or “diversity.”

14 The critiques o this approach center on the question as to which species is best

suited to represent a specic group o species.




64

Biodiversity

conservation

Ecosystem services /unctions

Pollination

Pest regulation

Decomposition o

organic matter

Valued insects

(e.g., butterfies)

Valued plants

Pollinating insects

Predators and

parasitoids

Soil invertebrates,

soil microorganisms

Red List species

Species o high

conservation /

cultural value

Abundance

Ecological unction

Area o protection Ecological entity Attribute Unit o protection

PRoTecTIoN Goals

cRITeRIa foR THe oPeRaTIoNal DefINITIoN

of THe PRoTecTIoN Goal

assessMeNT eNDPoINTs

Population

Population

Guild

Guild

Guild

Individuals

Populations

Communities

Guilds

Table 3: Examples how to use the matrix or an operational denition o protection

goals (biodiversity) or some selected protection goals




in biological control unctions, or example, could be surveyed indirectly by recording unusual pest outbreaks (Sanvido et al., 2009). Soil invertebrates and

soil microorganisms could be surveyed by assessing decomposition o organic

matter (Knacker et al., 2003; Zurbrügg et al., 2010) or parameters such as soil

respiration and microbial biomass (Römbke, 2006; Ferreira et al., 2010).

Parameter selection or early tier studies should consider the known mode

o action o an insecticidal protein against the sensitive targets. In the case o

Cry1Ab proteins, or example, sensitive Lepidoptera larvae are killed relatively

quickly ater ingestion o the protein. Consequently, one would select “mortality”

as the appropriate parameter. In the case o an insecticidal protein that is knownto reduce the ecundity, but that is also known to have no lethal eect, the param-

eter “reproduction” would be a more appropriate endpoint than “mortality.”





THe Use o “THResHolDs”

To eValUaTe RIsKs aND cHaNces

CHAPTER 6




70

6 THe Use o “THResHolDs” To eValUaTe RIsKs aND cHaNces

The idea that risks are acceptable as long as they do not exceed a certain prede-ned threshold is a concept that is oten emerging in discussions related to deci-

sion-making in risk assessments. The use o thresholds is thereby oten linked

to the issue o balancing risks and benets or more precisely risks and chances.15

When discussing the application o thresholds and the issue o balancing risks

with chances, it is important to consider that both concepts have their oundation

in specic ethical theories (i.e., either deontology or utilitarianism). Depending

on the ethical theory that is underlying the relevant legal rameworks, there

may be dierences how these concepts may be applied in a decision-making

process. The term “threshold” is urthermore oten used in dierent connota-tions in ethics and ecology. In the ollowing, the implications o these dierences

or regulatory decision-making are shortly described.

6.1 Thrhd in thi

What are the ethical criteria or acceptable risks? One option that underlies

many legal rameworks is that the risks individuals or populations are exposed

to (without their consent) are acceptable as long as they do not exceed a certain

predened threshold. Another option is that risks are acceptable (or even oblig-atory) i they are necessary or realizing the greatest expected net benet. The

rst criterion is advocated by deontological theories, the second by utilitarian

theories. It is important to note that, contrary to a widespread belie, the deon-

tological and the utilitarian criterion o risk acceptability cannot be combined

since they are mutually exclusive.

6.1.1 Dntgi pprh

According to deontological approaches (see section 9.3.2), decisions are based on

an absolute normative threshold dening a limit that no risk may exceed (Figure 1).

The main idea underlying the deontological criterion is that there is a duty o

care according to which risks should be reduced to a point where the occurrence

o harm is not to be expected. Requiring zero risk, however, cannot be justied

since this would make social lie impossible. That is why a normative threshold

is introduced. Risks exceeding this threshold are prohibited, irrespective o

the chances associated with them, while risks below the threshold are accept-

able. All options where the risk remains below the specied threshold (B & C)

15 Despite the requent use o the term “benet“ as the opposite o the term “risk,” the

term “chance“ would be the correct antonym to risk. The correct antonym to “benet“

is “damage“ as both terms describe a actual statement whereas both the terms “risk“

and “chance“ incorporate a probability component, namely, the probability that a

damage (rom a risk) or that a benet (rom a chance) might occur, respectively.




72

or populations to certain risks, this must be done, even i these risks are very high.For each option, the total chance is compared to the total risk. There is an obligation

to choose the option with the highest expected benet (total chance minus total

risk). When ollowing a utilitarian approach, there is no normative threshold.

Utilitarianism needs a theory o intrinsic values to be able to determine risks

and chances and to weigh them up against each other. Chances reer to the prob-

ability o intrinsic value (benet) occurring, risks to the probability o intrinsic

disvalue (damage) occurring. Weighing risks and chances and summing them

up, however, presupposes a common scale, a common currency, as it were, that

allows adding and subtracting them. However, what does this currency consisto? There have been several attempts o solving this problem (or instance, by

using a monetary currency as in willingness to pay or willingness to compen-

sate), but all o them have shown to be problematic.

A urther problem aecting the determination o risks or damage is how

conficting protection goals should be weighed. O course, the law can settle this

problem by at. However, i policy makers want to preserve legal coherence,

they should try to determine the relative value o dierent protection goals by

deriving this value rom higher-order protection goals. Take, or instance, the

protection goals biodiversity and water. I there is a confict between these twoprotection goals, we need to know which o them has greater relative value with

regard to a higher-order protection goal such as the maintenance o ecosystem

stability. In many cases, the law does not suciently speciy the relation between

higher-order protection goals and subordinate protection goals. This problem

does not seem unsolvable, however. What would be needed in our example is a

determination o the instrumental, that is, unctional value o water and biodi-

versity with regard to ecosystem stability in a way that makes it clear which o

the two protection goals should be assigned priority. This determination should

be specic enough to be applicable in concrete situations where regulators mustchoose between these conficting protection goals.

6.2 Thrhd in gy

An ecological threshold can be described as a point or a zone where there is

a relatively rapid change rom one ecological condition to another (Huggett,

2005; Luck, 2005). According to this theory, at a threshold, even small changes in

environmental conditions will lead to large responses in ecosystem state variables




THE USE OF “THRESHOLDS” TO EVALUATE RISKS AND CHANCES 73

(Groman et al., 2006; Suding and Hobbs, 2009). One idea underlying the

threshold concept is its practical use or management decisions as exceeding

specic ecological thresholds would indicate that an ecosystem is moving away

rom a desired state.16

The term threshold is closely linked to the concept o resilience, which is

oten dened as the capacity o ecological systems to recover rom disturbances(Walker, 1995). Resilience is thereby equated to the time needed or a system to

return to a global equilibrium state ollowing a perturbation. In the ecological

literature, it is thus associated with the concept o ecological stability. 17 Accord-

ingly, some authors postulate a threshold o biodiversity below which ecosystems

lose the sel-organization that enables them to provide ecological services

(Perrings and Opschoor, 1994). Ecosystems would thereby become increasingly

16 The question as to what denes a desired state includes a subjective value

judgement, a question which can not be answered rom a purely scientic and

objective perspective.

17 The term “ecological stability” is one o the most nebulous terms in ecology as up to

163 dierent denitions o the term have been counted (Grimm and Wissel 1997,

Muradian 2001).

Figure 2: Graphical scheme illustrating the utilitarian approach to balancing

risks and chances

Chance

Risk

Acceptable

Not

acceptable

Not

acceptable

a cB

Highest

expected

benet




74

unstable as diversity decreases. As threshold patterns ollow a discontinuous,nonlinear behavior, ecosystems would not be robust enough to resist species

deletion and lose their capacity to absorb perturbations once species diversity

alls below the stability threshold. This would require that ecosystems have a

“saturation point,” that is, ecosystem services would not increase above a certain

diversity level even when biodiversity is increased. Such a pattern is quite

similar to the rivet hypothesis (Ehrlich and Ehrlich, 1981), which suggests that

ecosystems can initially aord continual removal o its component without expe-

riencing a loss o unction, while at a certain point only one additional species

extinction leads to a loss o the unction (see section 4.4).In a second denition o resilience, the existence o multiple “stable” states

in ecosystems is emphasized.18 Resilience is thereby dened as the ability o a

system to withstand changes and to maintain its structure and patterns beore

fipping to another stable state (Holling, 1973; Peterson et al., 1998; Gunderson,

2000). One key distinction between the two denitions o resilience lies in the

assumptions regarding the existence o multiple stable states. The rst deni-

tion, where resilience is essentially dened as the return time to equilibrium,

assumes that there exists only one stable state. The second one, in contrast,

presumes the existence o multiple stable states and the tolerance o the systemto perturbations that acilitate transitions among them (Gunderson, 2000).

The existence o multiple stable states and transitions between these states

have been described in a range o ecological systems, such as or example

the transition rom grassland or abandoned arable land to shrubland and

subsequently to woodland (Harmer et al., 2001; Kuiters and Slim, 2003). Both

ecological theories and empirical evidence clearly show that alternative stable

states can coexist side-by-side (Scheer and Carpenter, 2003). Landscapes

oten comprise a mosaic o patches with dierent alternative stable vegetation

types that remain unaltered or decades until an extreme event triggers a shitin the pattern

In theory, ecological thresholds must not be restricted to species extinction

leading to the collapse o ecosystem unctions, but thresholds can also be related

to habitat loss. This other theoretical pattern predicting ecological thresholds

is linked to habitat ragmentation and metapopulation models (Hanski, 1998).

Species loss may also occur due to habitat loss and ragmentation, in particular

due to the loss o habitat connectivity. Connectivity loss thereby disables the

dispersal and exchange o individuals between metapopulations occurring in

18 Ecosystems are obviously never stable in the sense that they do not change. There are

always constant fuctuations o natural populations due to seasonality and environmental

conditions (Scheer and Carpenter 2003).




THE USE OF “THRESHOLDS” TO EVALUATE RISKS AND CHANCES 75

dierent habitat patches. As the loss o habitat connectivity is a highly nonlinearprocess, theory predicts that species loss occurs ater the degree o connectivity

has allen below a certain critical threshold. The same may be true or ecosystem

services such as biological control unctions o natural enemies where habitat

ragments support a less diverse community o natural enemies resulting in lower

predation or parasitation rates on pest populations (Kruess and Tscharntke,

1994; With et al., 2002).

6.3 Uing thrhd in rgutry diin-mking

What are the conclusions or regulatory decision-making when considering the

above mentioned ethical and ecological considerations or the use o thresholds

and in particular regarding the question under what circumstances one is

allowed to balance risks and chances? The ethical considerations are insoar

relevant as many legal rameworks such as the Swiss Gene Technology Law

are mostly based on deontological considerations. Dierent options are thereby

acceptable as long as their associated risks do not exceed a certain predened

threshold. The ethical considerations are also important when answering the

question when to balance risks and chances. Again, the deontological approachunderlying many legal rameworks determines that balancing risks and chances

between dierent options is permitted or those options where the risks o each

option remains below the given threshold. Balancing risks and chances is thus

not per se prohibited.

Using a threshold to dene damage would be an elegant approach as it

would allow more or less unambiguous decisions what represents unaccept-

able harm. In principle, every indicator value that would exceed the threshold

would be damage. Unortunately, thresholds are dicult to apply in practice as

the legal rameworks regulating the use o genetic engineering in Switzerlanddo not dene such thresholds. Without such thresholds, a coherent application

o the law is, in principle, impossible as regulatory authorities have no clearly

quantied indications that allow them to ensure the protection o human

health and the environment. Similarly, applicants (i.e., usually the companies

that intend to market a GMO) ace regulatory uncertainty as they are not able

to predict the overall costs and the time necessary or product approval. This

is in contrast to, or example, the regulation o plant protection products which

provides standard criteria or the evaluation and approval o these products




76

in Annex 6 o the respective ordinance (PSMV, SR 916.161). I such thresh-olds are missing, it is clearly the responsibility o the regulatory authorities to

provide applicants with such thresholds as the authorities are responsible or

the execution o the law.

Based on these theoretical considerations, the dicult question to answer

rom an ecological point o view is how to dene such thresholds. We argue

that ecological thresholds or regulatory decision-making o GM crops are

dicult to apply in practice. One o the major actors inhibiting the use o

ecological thresholds in environmental management is the lack o general

principles or applying these concepts to dierent kinds o response variablesand ecosystems (Groman et al., 2006). In nature, populations usually fuctuate

around some trend or stable average and ecosystems are assumed to respond

smoothly to gradual change in external conditions. Occasionally, however, such

a scenario is interrupted by an abrupt shit to a dramatically dierent regime.

Dramatic regime shits are known or a range o ecosystems including lakes,

coral rees, oceans, orests and arid land (Scheer et al., 2001; Scheer and

Carpenter, 2003). While both theoretical hypotheses and practical examples

support the existence o discontinuities in ecological processes, it is very di-

cult to support such patterns with empirical data as most experimental studiesare not able to show a clearly discontinuous relationship between stability and

diversity (Muradian, 2001). Most studies ace the diculty to clearly dene the

characteristics o dierent alternative stable states, mainly as these are highly

dependent on the chosen temporal and spatial scales. Hence, each stressor

and ecosystem response must be evaluated independently, a process that is

usually not appropriate or regulatory decision-making as it requires years o

site-specic research. Ecological sciences are currently more able to predict

the magnitude o change (i.e., the possible alternative state) than the precise

threshold value (Muradian, 2001). With growing ecological experiences, thresh-olds or certain ecological groups may become available. Nevertheless, one has

to recognize that or most ecological indicators, ully operational thresholds

will probably rarely be available. This inevitably challenges policy-makers as

they cannot precisely rely on dened ecological thresholds that would acili-

tate decision-making processes. I thresholds are missing, damage should be

dened by using a baseline approach (see sections 7 and 8). In principle, the

baseline approach should allow one to determine when a change has to be

regarded a damage as the denition o damage is not depending on a precise




a IRsT BaselINe aPPRoacH –

DIeReNTIaTING eecTs o DIeReNT PesT MaNaGeMeNT PRacTIces IN MaIZe

CHAPTER 7




80

7 a IRsT BaselINe aPPRoacH – DIeReNTIaTING eecTso DIeReNT PesT MaNaGeMeNT PRacTIces IN MaIZe

7.1 Th hng ting n pprprit bin

The discussions during the rst expert workshop showed that baselines were

recognized as being a crucial point o any decision-making process as they serve

as a reerence to determine what makes a change a damage. Moreover, the base-

line approach seems to be the only practicable way o evaluating damage as the

threshold approach appears to have serious limitations that impede its practical

use in regulatory decision-making (see section 6.3).

Most people consider having an intuitive notion o the term “baseline.”However, looking at existing denitions in dierent texts, it becomes apparent

that these denitions are anything but uniorm. As a basic denition, a baseline

can be dened as “an initial set o critical observations or data used or compar-

ison or a control” (Merriam-Webster Online, 2011). The EU Directive 2004/35/EC

on environmental liability denes the term as “a reerence to assess the signif-

cance o any damage that has adverse eects on reaching or maintaining the avo-

rable conservation status o protected habitats or species. The signifcance has

to be assessed by reerence to the conservation status at the time o the damage,

the services provided by the amenities they produce and their capacity or naturalregeneration. Signifcant adverse changes to the baseline condition should be deter-

mined by means o measurable data” (European Commission, 2004). More speci-

cally, reerring to the assessment o GMOs, the European Food Saety Authority

denes the term baseline as “the current status quo, e.g., current conventional

cropping or historical agricultural or environmental data. The baseline serves as a

point o reerence against which any eects arising rom the placing on the market

o a GMO can be compared” (EFSA, 2006).

Due to the vague denitions o the term “baseline,” the use o the baseline

concept as a decision support tool remains ambiguous necessitating a moreprecise characterization. A common point pertinent to all denitions is the term

“comparison” – semantically decisions should thereore theoretically always be

taken relative to a comparator. Ultimately, the question as to what represents

environmental damage is nevertheless a legal problem. Legally, the baseline or

any evaluation o damage rom GMOs should be set by what is already regarded

representing damage today. This implies that potential damage resulting rom

GMOs should be evaluated according to the same criteria as any damage

that might occur rom other technologies, or example, chemical pesticides.19

19 Provided that there is a legal ramework regulating the technology (e.g., pesticides).

The approach is not valid or techniques that are not regulated such as the use o

mowing or tillage machinery.




A FIRST BASELINE APPROACH 81

The implicit baseline or the evaluation o impacts o GM crops on biodiver-sity corresponds thus to current agricultural management practices where the

impact o the GM cropping system are evaluated against the impacts caused by

the cropping practices that are replaced. This leads to the question how impacts

o dierent agricultural management practice could be compared to determine

whether the impacts o the GM cropping system are lower, equal or higher than

those o current practices. A direct comparison is dicult as these impacts vary

depending on a number o dierent actors. A simple, easily applicable generic

comparison o the GM cropping practice with the most common current pest

management practice is almost impossible and beyond the scope o this report.Comparative impact assessments can thus only be perormed on a case-by-case

basis where the nature o the crop, specic aspects o the pest management prac-

tices applied and the receiving environment are considered. Dierent regulatory

rameworks are moreover used as a basis to evaluate the eects o dierent pest

management practices, which makes it almost impossible to directly compare

the eects o a GM cropping system to its conventional non-GM counterpart.

Nevertheless, an important point to consider in each comparison reers to the

act that all regulatory rameworks dierentiate between “intended” eects o

the cropping practice applied and “unintended” eects that are to be minimized.Such a dierentiation should allow outlining a generic scheme that permits to

evaluate whether the eects o dierent pest management practices are to be

regarded as intended eects (which are regarded acceptable) and unintended

eects that could represent environmental damage. This dierentiation will in

the ollowing be used to establish an approach that can help to dierentiate

between these two types o eects (see section 7.3).

7.2 c tudy 1: et Bt-miz n nntrgt rthrpd

The impacts o genetically modied Bt-maize on arthropod biodiversity in

comparison to current pest management practices used or European Corn Borer

management in maize in central and western Europe are taken as an example

to show how it is necessary to dierentiate between dierent types o eects

when evaluating environmental damage. Current crop protection practices

considered include two current (non-GM) practices (insecticides and biological

control using Trichogramma). Given that both Bt-maize and current pest control

practices display insec ticidal properties, this case study will concentrate on the





which is still relatively small (approx. 4 – 5%) compared to the total area (3.3million hectares) where ECB damage is regularly above the economic injury

level in the European Union (EU-27). In addition, arming practices are used

to help manage insect pests by physically destroying pests essentially through

stalk destruction in the Fall and ploughing o maize stubbles prior to planting.

7.2.1.3 European Corn Borer control using Bt-maize

Bt-maize is genetically modied maize that carries a gene rom the soil bacte-

rium Bacillus thuringiensis (Bt). The dierent strains o B. thuringiensis contain

varying combinations o Cryproteins (so called Bt-toxins) and each o these insec-ticidal proteins is known to have a very selective toxicity against dierent groups

o arthropods (Höte and Whiteley, 1989; Schnep et al., 1998; de Maagd et al., 2001).

Bt Cry proteins are currently the only insecticidal proteins that are commercially

used in GM crops. The gene inserted into Bt-maize expresses the insecticidal

protein Cry1Ab that coners resistance to Lepidoptera (moths and butterfies).

It was initially developed to control the ECB, but has shown to be also eective

against various other lepidopteran pests, such as the Mediterranean Corn Borer

(Sesamia nonagrioides), which is restricted to the southern parts o Europe.

7.2.1.4 Environmental hazards associated with insecticides

Broad-spectrum insecticides usually have moderate to highly toxic eects on a

wide range o organisms (European Union, 2009a). Environmental impacts o

insecticides strongly vary depending on the product applied, its active substance,

the application rate, the ormulation and the application technique. For the present

case study, two active substances were selected that are used or ECB management

in maize in Europe. The rst active substance (Lambda-Cyhalothrin; a pyrethroid)

is an older broad-spectrum insecticide that is still widely used, while the second is

a more selective insecticide (Indoxacarb; an oxadioazine). Acute toxicity data or

the two active substances Lambda-Cyhalothrin and Indoxacarb are taken rom

review reports rom the EU Pesticides database (European Union, 2009a) and rom

Pesticide actsheets o the U.S. Environmental Protection Agency (EPA, 2009),

Pollinator, benecial arthropods, and butterfies: Broad-spectrum insecticides

such as Lambda-Cyhalothrin are highly toxic to a wide range o invertebrates

including honey bees. A repeatedly observed eect o many broad-spectrum

insecticides is the promotion o secondary pest outbreaks. These generally




84

occur when insecticide applications kill natural enemies that were controllinga herbivore species that was not a pest beore. These species can then increase

to densities that cause damage. The new classes o more selective insecticides

such as Indoxacarb have several advantages over the older broad-spectrum

insecticides. Most are less likely to harm natural enemies and other nontarget

species. Indoxacarb, or example, is very eective against lepidopteran larvae,

but allows most predators and immature parasitic wasp attacking these larvae

to survive. Indoxacarb is also practically nontoxic to bees. Insecticides generally

tend to have immediate, but predominantly short-term (2–3 months) eects on

nontarget arthropods, as aected population may be able to recover due to therecruitment o new individuals rom unaected populations.

7.2.1.5 Environmental hazards associated with biological control

using Trichogramma

The egg parasitoid Trichogramma brassicae is the only invertebrate used in

augmentative biological control in maize (Bigler et al., 2010). Two ecological

concerns are discussed related to the use o this exotic species in biological

control programs in Europe. The rst concern relates to the establishment o

T. brassicae under elds conditions ollowing its dispersal into nontarget habitats.Because o their broad host range, the second concern relates to the question is

whether mass releases o T. brassicae may have ecologically signicant direct and /

or indirect eects on nontarget arthropods. Given that T. brassicae can success-

ully overwinter in diapause under Swiss climatic conditions, one can assume that

the species would establish in most regions o Europe i nontarget hosts are avail-

able. T. brassicae density increases during mass releases, but drops to prerelease

densities 2–3 weeks ater the last release, indicating that only a small raction (less

than 10%) o the released population disperses out o maize elds to nontarget

habitats. Permanent establishment occurs only in limited areas o seminatural and

natural habitats making eects on native Trichogramma populations unlikely. As

T. brassicae has a clear host plant and habitat preerence, the parasitoid does not

successully attack arthropod host eggs on other plants than maize. The eect on

lepidoptera and aphid predators under semield and eld conditions is thus very

low. Short-term mortality caused by T. brassicae in o-crop habitats is likely, but

has a negligible to low population and community eect, as the magnitude is likely

to remain below a 40% reduction in the size o nontarget lepdoptera populations.

Quantitative eects will moreover be temporary and o transient nature.




7.2.1.6 Environmental hazards associated with Bt-maizeGiven the insecticidal properties o the Cry1Ab toxin, most concerns relate to

the question as to whether Bt-maize harms organisms other than the pest(s)

targeted by the toxin. As o May 2009, 89 original research articles have been

published addressing eects o Bt-maize (or the puried toxin) on dierent

groups o organisms both in the laboratory and in the eld (Naranjo, 2009). Most

published studies have been perormed by public research ater the approval

o the respective Bt-maize events in 1996. In addition, studies perormed by the

applicants (i.e., the company marketing the crop) have been conducted prior

to the market approval o Bt-maize (EPA, 2001; Mendelsohn et al., 2003; OECD,2007). Data or the Bt-toxin Cry1Ab are taken rom the U.S. EPA Biopesticides

registration action document on Bacillus thuringiensis plant-incorporated

protectants (EPA, 2001).

Pollinators: Honey bees are the most important pollinators o many agricultural

crops worldwide. Because o their importance to agriculture, they have been a

key test species used in environmental saety assessments o Bt-maize. These

assessments have involved comparisons o honey bee larval and adult survival

on puried Cry proteins or pollen collected rom Bt-maize versus survival onnon-Bt-maize material. A recent meta-analysis combining the results o 39

studies assessing dierent Cry proteins ound no statistically signicant eect

o Bt Cry protein treatments on survival o honey bees.

Benecial arthropods: Another important group o nontarget organisms

providing ecological and economic services within agricultural systems are

parasitoids and predators (so-called natural enemies). One hypothesis is that

Bt-maize could alter the biological control unctions o benecial arthro-

pods, which are important or controlling herbivorous insect populations inmaize. Nontarget organisms have to ingest the insecticidal protein expressed

in Bt-maize in order to be directly aected. Ingestion can occur via several

ways o exposures. Exposure can either occur by eeding on plant mate-

rial (e.g., leaves, pollen), by eeding on insects that have previously ed on

GM crops (and thereore contain the toxin) or via exposure through the envi-

ronment, e.g., when toxins rom plant residues persist in the soil. In order to be

aected by the toxin, natural enemies would moreover need to be sensitive to

Cry1Ab. Overall, several meta-analyses have shown that the majority o studies





86

conducted in the laboratory and in the eld show no adverse eects on nontargetorganisms resulting rom direct toxicity o the expressed Bt-toxins (Wolenbarger

et al., 2008; Naranjo, 2009). Some laboratory studies revealed indirect prey-quality

mediated eects due to Bt-maize. Under eld conditions, these secondary trophic

eects may be caused by changes in the availability and / or the quality o target

herbivores with specialist natural enemies (oten parasitoids) depending entirely

on the target pest (i.e., the ECB). Such eects are common or any pest control

method; they are o minor concern within an environmental risk assessment

context and they should be dierentiated rom direct eects o a toxin. The weight

o considerable data available rom eld studies gives evidence that Bt-maize hasno ecologically relevant eects on a number o taxa, especially in comparison

with alternative pest control measures such as broad-spectrum insecticides. The

results conrm the high specicity o Cry1Ab having a very narrow range o

activity restricted to the target pest and closely related species. Biodiversity is

thus higher in Bt-maize elds in comparison to maize elds treated with tradi-

tional insecticides having a broader range o activity.

Butterfies: A particular ocus was laid on butterfies as Cry1Ab coners resist-

ance to Lepidoptera and as butterfies are considered to be a species group witha high aesthetic value serving as symbols or conservation awareness. Butterfy

larvae can be exposed to Cry1Ab in the vicinity o Bt-maize elds (up to approx-

imately 2 meters rom the border) when eeding on host plant leaves naturally

dusted with maize pollen during anthesis (fowering). Especially the case o the

Monarch butterfy ( Danaus plexippus) caused much public interest and led to a

debate over the risks o Bt-maize as one study had ound that when pollen rom

a commercial variety o Bt-maize (event Bt11) was spread on milkweed leaves in

the laboratory and ed to monarch butterfy larvae, ater our days almost hal o

the tested larvae died (Losey et al., 1999). Extensive ollow-up studies showedthat initial reports o toxicity o high doses o Cry1Ab toxin to butterfies in the

laboratory did not necessarily mean that there would be exposure to toxic levels

in the eld (Sears et al., 2001; Dively et al., 2004). The proportion o Monarch

butterfy population exposed to Bt-pollen in the eld was estimated to be less

than 0.8%. Although similarly detailed data is not available or Europe, the data

available provide condence that commercial cultivation o currently avail-

able Bt-maize varieties constitutes a low risk or European butterfy species.

This conclusion is based on the assumption that larval exposure to Cry1Ab is





relatively low or European butterfy species considering the low expressionlevel o the toxin in pollen o the only approved Bt-maize event MON810, the

act that most pollen is deposited within a ew meters rom the eld border, and

that maize is not considered a host plant or nontarget butterfies.

7.3 apprh t dirntit t vriu pt mngmnt prti

n nntrgt rthrpd

The fow chart presented in Figure 3 proposes an approach as to how the eects

o dierent pest management practices on the arthropod auna can be dieren-tiated and compared. Eects are thereby considered within the whole system

“agricultural landscape”21 and cover both biodiversity conservation (species and

habitat diversity) and unctional biodiversity (i.e., ecosystem services) issues. The

rst part o the fow chart relates to crop protection and has to be answered based

on the spectrum o activity o the respective crop protection method applied.

Therein, the rst two questions help to dierentiate between intended eects

on target pests and other (potential) pests and unintended eects on nontarget

arthropods that may represent environmental damage. The second part o the

fow chart (Questions 3 to 11) has to be answered based on the species groupspotentially aected. Biodiversity entities potentially aected are determined

based on the Swiss environmental protection goals related to agriculture where

biodiversity is classied into three main aspects: species and habitat diversity,

genetic diversity within species and unctional biodiversity (see section 4.2.3).

Genetic diversity is thereby regarded as an important part o unctional diversity

as it is crucial to a species’ ability to adapt to its environment.

7.3.1 crp prttin

Existing regulatory rameworks acknowledge the act that agricultural manage-ment involves crop protection and that these practices inevitably comprise certain

environmental impacts. The regulation aims at ensuring that occurring impacts

are restricted to target pests, while possibly excluding unintended eects on

nontarget organisms. Three types o eects can be dierentiated when comparing

the eects o dierent crop protection practices: (1) intended eects on pests, (2)

inevitable trophic eects due to the absence o these pests and (3) unintended

eects on the arthropod auna in agricultural crops and the wider landscape.

21 Realistically, adverse eects on arthropods occur with a higher probability in crop elds

rather than outside the crop or on a landscape scale. This is basically because the exposure

o nontarget arthropods to the stressor (e.g., the insecticidal protein) is highest in crop elds.

Whether eects occurring in the crop translate to the landscape scale is depending on

a number o actors such as the magnitude o eects and the type o species aected.





Potential environmentaldamage on ecosystemservice

No concern as eectcan be mitigated byrisk management

No

Possibility o mitigating eect

Resiliencepossible?

Relevant eect onecosystem service?

No

No concern as eect islikely to be temporary

No concern asecosystem service is not

signicantly aected

Considermagnitudeo eect

No

No

Yes

Key speciesgroup or ecosystemservice?

No concern asecosystem service is

likely to be stable

Yes

Yes

Yes

Intndd eect (or posi-tive side eect) o thetoxic compound applied

Trophic eect due tothe absence o pest(s)

No concern as eect isdue to the crop protec-tion method applied

No concern as eectis an invitb conse-quence o crop protection

Ecosystem services





24 International Union or Conservation o Nature and Natural Resources

7.3.2 Bidivrity nrvtinQuestion 3: Does the species have a protection status?

The rst question to be answered when determining whether eects on

arthropods are aecting biodiversity conservation is whether these eects

are related to species having a particular protection status (Question 4 in

Figure 3). Three categories o species have a particular legal protection status

in Switzerland: (a) the IUCN24 Red List o threatened species being the most

comprehensive resource detailing the global conservation status o plants and

animals (Rodrigues et al., 2006), (b) species threatened on a national level and

(c) species characteristic or a specic habitat and / or region. Red Lists are a

legally binding instrument o biodiversity conservation assigning these species

an absolute protection status (NHV, SR 451.1). The two latter categories o

species are related to the overall environmental policy goal in Switzerland to

preserve and promote domestic species and their respective habitats (Duelli,

1994; Bau/BLW, 2008).

Groups o Red List species that are assigned with a specic protection

status are characterized by the availability o sucient expert knowledge to

perorm a judgment on their conservation status. For many groups o inverte-

brates, however, detailed data and expert knowledge is oten missing and theirconservation status is not known. Although being o conservation interest,

some arthropod groups are, or example, not listed in the Red Lists. Regula-

tory decision-making based on Red Lists thereore has limitations or inverte-

brates as decisions can only be made or those taxa that are listed and explicitly

protected. Ultimately, one has thus to accept that we can only evaluate what is

perceived to be important by having been listed.

Question 4: Is the species a charismatic species?

In case an eect is not related to a species having a particular protection status,the next question to be answered is whether the eect is associated with a charis-

matic species (Question 5 in Figure 3). Although not having a particular legal

protection status, a number o specie groups are regarded as being o conservation

value primarily because o their aesthetic, cultural or other relevant value. For

the arthropod auna, these include, or example, a number o butterfy and

ladybird beetle species that are not explicitly listed as either Red List species

or species threatened on a national level.




92

25 Zone A – North: Denmark, Estonia, Latvia, Lithuania, Finland, and Sweden

Zone B – Center: Belgium, Czech Republic, Germany, Ireland, Luxembourg, Hungary,

Netherlands, Austria, Poland, Romania, Slovenia, Slovakia, and United Kingdom

Zone C – South: Bulgaria, Greece, Spain, France, Italy, Cyprus, Malta, and Portugal

In case a particular eect is neither related to a species with a protectionstatus nor to charismatic species, one can reasonably assume that the species

aected are judged to be o no particular value. Any eect is thus o minor

concern or biodiversity conservation and is thereore not regarded as repre-

senting environmental damage.

7.3.2.1 Consider magnitude o eect

In case a particular eect relates to either species with a protection status or to

charismatic species, one has to determine the magnitude o the environmental

impact o the specic pest management practice. The impact o a specic pestmanagement practice depends on a number o dierent infuencing actors

that vary depending on the nature o the environmental stressor (Table 4).

These actors can only be assessed on a case-by-case basis, taking into account

the conditions prevalent in a particular receiving environment. Ideally, such

an assessment would need to estimate all infuencing actors. In many cases,

however, it may not be necessary to assess all actors as the occurrence o an

eect may become highly unlikely in the absence o certain decisive actors

(e.g., because the spectrum o activity and the mode o action o the stressor

exclude a toxic eect on a particular group o valued species). Similarly, it may sometimes not be necessary to assess all actors experimentally as meaningul

estimations can be deduced rom existing data or by theoretical considera-

tions. This might be particularly relevant or those cases where it is impossible

to obtain sucient individuals or a meaningul experimental assessment, or

example, in the case o a specic Red List species. Assessments in the case

o Red List species might nevertheless also be carried out experimentally by

using appropriate surrogates.

Even or a specic crop such as maize and a pest such as the European

Corn Borer, an application-based assessment can oten only be perormed ona regional basis as pest management decisions greatly vary among regions in

Europe (Meissle et al., 2010). Dierent approaches are possible to dene what

characterizes a region. According to the new regulation on plant protection

products (European Union, 2009b), or example, the EU is divided into three

dierent geographical zones.25 Other approaches dene more detailed biogeo-

graphical regions, such as the Natura 2000 network speciying nine regions

or the SEAMzones approach proposing 12 environmental zones across the

EU (Hazeu et al., in press). Regulatory authorities demanding or regional





assessments nevertheless need to consider that these need to remain practi-cable as assessments become increasingly impracticable the more regions have

to be considered.

Question 5: Is there a relevant eect on population level?

A relevant impact on biodiversity conservation is characterized by an explicit

reduction in population size o a particular species or by a clear reduction o

its area o distribution (Duelli, 1994). Such an evaluation can o course only

be perormed i the prior abundance and distribution o the species is known.

Unortunately, such data is oten not available, in particular, or most invertebrategroups that count or approximately 98% o all animal species (Duelli, 1994). The

conservation status o most species having a particular protection status is there-

ore usually deduced rom current knowledge on their specic habitat require-

ments. According to the Swiss legal rameworks, the value o a specic habitat

is judged based on the rareness o the species (potentially) being present in

this habitat (NHV, SR 451.1). In general, the rarer the species present, the more

value is given to the habitat rom a biodiversity conservation point o view. This

is based on the rationale that in case o a widespread impairment o a specic

habitat, one can reasonably assume that the species specically depending onthis habitat will be impaired too.

According to the Red List species concept, every loss is principally to be

regarded as damage as the legal rameworks assign these species a priority

status. With regard to biodiversity conservation, this would imply a zero tolerance

towards clear and widespread reductions in population size o Red List species

(i.e., a reduction o the total number o individuals o that taxon). In practice,

however, reductions in population size are most oten not abrupt, but continuous,

meaning that a particular species will not go extinct immediately. The IUCN

works with dierent Red List categories that describe the conservation statuso a particular species ranging rom extinct (EX) to least concern (LC). Species

in the categories critically endangered (CR), endangered (EN) and vulner-

able (VU) are denoted as threatened (Table 5) (Gärdenors, 2001; IUCN, 2001;

Rodrigues et al., 2006). Instead o immediate extinctions, biodiversity losses are

thereore evaluated based on transers between dierent threatened categories.

Moving a taxon rom a higher to a lower risk category is usually regarded as

being positive while transers rom categories o lower risk to higher risk indi-

cate biodiversity losses.




94

Bt-miz

sTRessoR-RelaTeD

Spectrum o activity and mode o action o the stressor

produced by the transgenic trait

Susceptibility o organism to the insecticidal protein

Amount o stressor present in the environment based on

the ollowing actors a–d

Spatial distribution o Bt-maize in the landscape

(depending on the percentage o maize cultivation in

a given region and the adoption rate o Bt-maize)

Amount o insecticidal protein expressed in a specic

plant tissue based on temporal expression levels and

patterns

Dispersal o the stressor in the environment based on

pollen distribution patterns, root exudates, crop residues

(i stressor is expressed in these tissues)

Degradation o the stressor in the environment based on

environmental ate

Temporal duration o exposure based on timing, dura-

tion and intensity o the stressor being present in the

environment

PRoTecTIoN Goal-RelaTeD

Number o individuals / proportion o the populationexposed to the stressor based on the ollowing actors a–c

Occurrence and spatial distribution (regional,

landscape, habitat) o a valued species or its sensitive

lie stages

Occurrence and spatial distribution o the species’ host

plants (i applicable)

Amount o plant tissue consumed by organisms in the

sensitive lie stage

Table 4: Factors infuencing the

magnitude o environmental impacts

o Bt-maize, insecticide applications

and biological control organisms

that have to be considered when

determining whether ecologically

relevant eects on population level

o valued species occur (ecological

relevance is determined based

on questions 5 – 7 and 9 – 11 in

Figure 3, respectively) (adapted

and expanded based on Sears

et al., 2001).

8

1

2

3

a

b

c

d

4

5

a

b

c





Intiid

Spectrum o activity and mode o action o active ingre-

dient o the insecticide

Susceptibility o organism to the active ingredient

Amount o stressor present in the environment based on

Percentage and spatial distribution o maize elds in

the landscape where the insecticide is applied

Amount o insecticide active ingredient present on crops

and / or host plants (depending on deposition rate)

Dispersal o the insecticide in the environment based

on ormulation, application technique, drit actor and

vegetation distribution actor

Degradation o the active ingredient in the environment

based on environmental ate o the active substance

and ormulation

Temporal duration o exposure based on timing,

duration and application rate o the insecticide

Number o individuals / percentage o the populationexposed to the stressor based on


landscape, habitat) o a valued species or its

sensitive lie stages

Occurrence and spatial distribution o the species’

host plants (i applicable)

Amount o stressor ingested or contacted by organisms

in the sensitive lie stages

Bigi ntr gnt (Trichogramma)

Host range o T. brassicae

Parasitation rate o T. brassicae

Number o adult parasitoids present in the

environment based on

Percentage and spatial distribution o maize elds

in the landscape where T. brassicae is released

Amount o emales o T. brassicae deployed

in maize elds

Dispersal rate o parasitoids outside o maize elds

Survival o adult parasitoids in the environment

Temporal duration o exposure based on timing,

duration and number o applications o T. brassicae

Number o individuals / percentage o the populationexposed to the stressor based on


landscape, habitat) o a valued species or its

sensitive lie stages

Occurrence and spatial distribution o the species’

host plants (i applicable)

Number o hosts encountered by adult parasitoids




96

10

PoPUlaTIoN ReDUcTIoN

(past, present and / or projected)

Observed reduction that is clearly reversible AND

understood AND ceased

Observed reduction that may not have ceased OR may not

be clearly understood OR may not be clearly reversible

Projected or suspected population reduction

up to a maximum o 100 years

Observed reduction where the time period includes both

the past and the uture

GeoGRaPHIc RaNGe (either B1 or B2)

Extent o Occurrence (EOO) (see ootnote 23)

Area o Occupancy (AOO) (see ootnote 24)

and 2 o the ollowing 3

Severely ragmented locations

Continuing decline

Extreme fuctuationssMall PoPUlaTIoN sIZe aND DeclINe

Number o mature individuals (and either C1 or C2)

Estimated continuing decline up to a maximum o

100 years

Continuing decline

(a i) o mature individuals in largest subpopulation

(a ii) or % mature individuals

(b) extreme fuctuations in the number o mature

individuals

VeRY sMall oR ResTRIcTeD PoPUlaTIoNs

Either Number o individuals

or Restricted AOO

QUaNTITaTIVe aNalYsIs

Probability o extinction in the wild is at least

Table 5: Summary o the ve criteria

(A – E) used to evaluate i a taxon

belongs to a threatened Red List

species category (modied based

on IUCN, 2008).

9

A

A1

A2

A3

A4

B

B1

B2

C

C1

C2

D

E





critiy ndngrd

> 90%

> 80%

> 80%

> 80%

< 100 km2

< 10 km2

= 1

n.a.

n.a.

< 250

25% in 3 years or 1 generation

< 50

90–100%

n.a.

< 50

n.a.

50% in 10 years or 3 generations

(max. 100 years)

endngrd

> 70%

> 50%

> 50%

> 50%

< 5000 km2

< 500 km2

≤ 5

n.a.

n.a.

< 2500


< 250

95–100%

n.a.

< 250

n.a.


(max. 100 years)

Vunrb

> 50%

> 30%

> 30%

> 30%

< 20,000 km2

< 2000 km2

≤ 10

n.a.

n.a.

< 10,000


< 1000

100%

n.a.

< 1’000

AOO > 20 km2 or # locations ≤ 5

10% in 100 years





Question 6: Is it possible to tolerate biodiversity losses o species witha protection status or o charismatic species?

Rare species in agricultural landscapes are oten remains o ancient agricul-

tural practices. Historically, low intensity land management has resulted in a

rich assemblage o species in agricultural landscapes (Stoate et al., 2009). Over

centuries, anthropogenic infuences shaped the initial woodland prevalent in

Central and Western Europe into a cultural landscape dominated by rural activi-

ties. Dierent agricultural management practices divided the landscape into a

multitude o diverse habitats that allowed a wide variety o species to colonize

the newly created habitats. This resulted in an increase o biodiversity prior tothe industrialization o agriculture in the middle o the last century. It has been

estimated that 50% o all species in Europe depend on agricultural habitats,

including a number o threatened and endemic species (Stoate et al., 2009). Yet,

agricultural intensication over the past 60 years has led to signicant impacts

on the environment. The mechanization o agriculture has acilitated degrada-

tion in habitat quality and increasing habitat homogeneity. This led to the elimi-

nation o many landscape elements and habitats that were typical or a number

o specialized taxa. The species richness and habitat diversity o nearly all arm-

land has urthermore declined due to more intensive eld management, that is,among others higher pesticide and ertilizer use and the simplication o crop

rotations (Stoate et al., 2001; Robinson and Sutherland, 2002; Tilman et al., 2002).

Species present in agricultural landscapes today tend to be habitat generalists.

One may thus argue that the ew rare species occurring in agricultural land-

scapes are particularly valuable as their habitats are predominantly threatened

rom disturbances and habitat loss.

When evaluating widespread biodiversity losses in agricultural landscapes,

one may ask whether certain reductions in the population size o protected or o

charismatic species are tolerable. In the ollowing, we will argue that local biodiversity losses o protected species may, to a certain extent, be tolerable i popula-

tions o that taxon still occur or may develop in other, undisturbed habitats. The

argument is related to the dilemma that Red Lists are oten used (or even laid

down in existing legal rameworks) on a national or on a regional level, but were

initially developed to assess extinction risk o entire populations o species at

the global level. The use o Red Lists at the national level may thus be misleading

as a population can be distributed over more than one country, which makes it

dicult to estimate the extinction risk o the part o the population resident





In conclusion, eects on arthropod biodiversity are only to be regardedas potential ecological damage in case biodiversity losses are related to either

species with a protection status or to charismatic species. It is moreover impor-

tant to determine whether the population losses are widespread and o a high

magnitude and whether certain actors apply that might allow tolerating the risk

or biodiversity losses (e.g., in case they do not relate to endemic species that are

particularly characteristic or Switzerland). Finally, damage only occurs i it is

impossible to mitigate the biodiversity losses.

7.3.3 eytm rviI an eect on arthropod biodiversity is not related to a biodiversity conser-

vation issue, it may be related to an ecosystem service. Ecosystem services

are those properties o ecosystems (e.g., climate regulation, nutrient cycling,

pollination, and provision o ood and ber) that either directly or indirectly

benet human activities (Hooper et al., 2005; EASAC, 2009). Rationales or the

protection o biodiversity are oten based on the argument that species provide

these ecological goods and services that are essential or human welare (see

section 4.2.3).

Question 8: May certain species be lost without impairing

the ecosystem function?

In contrast to biodiversity conservation issues, where the ocus o conserva-

tion is clearly laid on specic species to be protected, it is less clearly dened

how many and which species provide a specic ecosystem service (see section

4.4.2). One o the most important ecosystem services provided by arthropods in

agricultural landscapes is the biological control o arthropod pests within crops

elds (Kruess and Tscharntke, 1994; Wilby and Thomas, 2002; Kremen, 2005;

Tscharntke et al., 2005). A crucial question when determining what representsenvironmental damage in this context is thus whether certain species may be lost

without impairing the ecosystem unction as the species are redundant (Ques-

tion 9 in Figure 3). This leads to the question as to whether there is a unctional

relationship between species diversity and ecosystem unction, that is, whether

more species generally lead to more stable ecosystem services. Available data

show that species richness oten seems to be not so important or agroecosystem

unction, as even only one or a ew species might ensure a particular unction

(Hooper et al., 2005; Moonen and Barberi, 2008; Shennan, 2008)(see section 4.4).




102

There is nevertheless convincing evidence that species-rich systems deliverecosystem services more reliably than species-poor ones. This is because in

diverse communities, redundancy among species can buer processes in response

to changing environmental conditions and species losses. Ecosystem services

typically do, however, oten not depend on the presence o specic species, espe-

cially not rare, narrowly distributed species. They more likely depend on common

and widespread species than on rare species (Diaz et al., 2006; Maclaurin and

Sterelny, 2008). From a unctional point o view, reductions in population size o

a redundant species may thus be tolerable and be o no concern i the unction

is ullled by other species. In case a species is not redundant, one has to deter-mine the magnitude and the spatial extent o the impact to evaluate whether its

loss constitutes environmental damage (Question 10 in Figure 3).

7.3.3.1 Consider magnitude o eects

The actors to be considered when taking into account the magnitude o eects

on ecosystem services are principally comparable to those relevant or biodiver-

sity conservation issues (see Table 4) as unctions are ultimately also carried our

by specic species groups.

Question 9: Is there a relevant eect on an ecosystem service?

Potential ecological damage on ecosystem unctions occurs i there is a relevant

impact on population level o a nonredundant species responsible or a particular

ecosystem service. This is essentially the case i local reductions in population

size are not buered in due time by migration rom other populations occurring

in nearby habitat patches. Changes in ecosystem unctions are usually dicult

to assess as the precise nature o a particular unction is oten unknown. It is, or

example, oten not known how much a particular species contributes to a speci-

ic unction such as biological control o pest species. In contrast to biodiversity conservation issues, where population reductions o valued species can normally

be assessed directly, ecosystem unctions usually have to be assessed indirectly.

By analyzing the general state o ecosystem unctions, one obtains a more accu-

rate indication whether unctional biodiversity is aected. An indirect survey

can be perormed by recording an indicator able to demonstrate a ailure in the

unction. Failures in biological control unctions, or example, can be assessed

more accurately by observing unusual pest outbreaks rather than by monitoring

specic groups o natural enemies (Sanvido et al., 2009).




104

regulatory rameworks. This leads to the somehow irrational situation that thesame impacts may be judged dierently depending on what pest management

practice caused them. This contradicts the initial assumption that the baseline

or any evaluation o damage rom GM crops is legally set by what is already

regarded representing damage today. In our opinion, this dilemma is one o the

most important open question that has to be resolved when discussing what

baseline has to be considered when evaluating environmental impacts o GM

crops. As long as pest management practices having similar impacts on the

environment are regulated dierently, it will not be possible to nd a reason-

able solution to determine a baseline suitable or regulatory decision-makingo GM crops. This dilemma is particularly incoherent when considering that the

use o broad-spectrum insecticides is generally known to have wider environ-

mental impacts than the use o Bt-maize (Romeis et al., 2008b; Wolenbarger

et al., 2008; Naranjo, 2009).

The rst question that arises rom the present case study is whether the

risks o the dierent pest management practices or biodiversity should be

evaluated based on their specic impacts on single groups o species or based

on an overall sum o impacts. A comparison based on the overall sum o impacts

on biodiversity might actually be useul in practice to allow decision-makerscomparing the ecological ootprint o dierent agricultural management prac-

tices, but current legislation does oten not allow decisions based on averaged

risk. Regulatory decisions are usually based on absolute risks, meaning that i

a technology is harmul it will not get approval because another technology is

more harmul. Similarly, risks are not averaged – a technology does not obtain

approval just because there is no risk or certain groups o organism i there is

a high risk or other groups o organisms. The case study is thus probably more

relevant or a comparison o the ecological impacts o dierent crop protec-

tion methods on a particular group o organisms than or a comparison o theoverall risks.

Second, the question arises as to what agricultural management practice

is to be considered as a comparator when perorming such an evaluation or

Bt-maize. One could argue that a “no treatment” scenario would be an appro-

priate baseline considering that only a small percentage (approx. 5 – 20%) o

the maize area in Europe is sprayed with insecticides against the ECB and 5%

is treated with Trichogramma (see section 7.2.1). However, this argument is in

our opinion not a logical rationale supporting a “no treatment” scenario as a





realistic baseline since the act that approximately 75% o the maize area is nottreated against the ECB is primarily an economic decision taken by armers.

This decision is oten due to ECB damage remaining below the economic injury

level and technical diculties to apply insecticides in tall maize. This results

in a lack o protability o the control methods available. Since armers would

have the option to use approved insecticides or Trichogramma, these should

also be used as a comparator or a current management practice.




a secoND BaselINe aPPRoacH –

DIeReNTIaTING eecTs o DIeReNT WeeD MaNaGeMeNT PRacTIces IN MaIZe

CHAPTER 8




A SECOND BASELINE APPROACH 109

economically acceptable threshold level. Weed control in maize is alwaysnecessary in temperate climates during the critical growth stage. Since maize

develops slowly ater seeding in comparison to native weeds, the crop has

to be protected rom weed competition between the 2- and 8-lea stages to

prevent yield losses (Ammon, 1993; Häni et al., 2006; Dewar, 2009). Weeds may

be controlled by mechanical weeding and / or by the use o synthetic herbicides.

Weed control in conventional maize cultivation typically includes tillage prior

to sowing and usually up to two pre- and post-emergence herbicide treatments.

In Switzerland, or example, approximately 85% o the conventional maize area

is ploughed prior to sowing, while only 15% are cultivated using conservationtillage practices (Häni et al., 2006).

8.2.1.2 Hazards o current weed management practices in maize

Conventional weed management practices in maize are usually based on using

an herbicide mix containing a combination o three to our active ingredients

(Devos et al., 2008; Dewar, 2009). This allows armers to establish locally adapted

herbicide regimes that consider the weed species present, as well as soil type

and conditions.

Direct toxic eects o herbicides

Although there have been herbicides that were toxic to humans and dangerous

to handle or armers, most o these products are no longer used in Euro-

pean agriculture and they have been replaced by newer products. Many newer

herbicides used in conventional maize possess little or no direct toxicity to

mammals, birds, earthworms, bees and benecial arthropods as herbicides

target highly specic biological or biochemical processes within plants. Some

o the herbicides used in conventional cropping systems may show direct toxic

eects on aquatic organism.

Indirect eects o herbicides

The intensive use o herbicides may promote a number o indirect environmental

eects such as the development o resistances in weeds, the enhancement o soil

erosion and the contamination o surace water. These concerns are however

independent rom the use o either conventionally bred or GMHT varieties, but

a result o the herbicide regime applied.




110

Impacts o tillage on soilMost armers use tillage (ploughing) to prepare the soil or planting. Excessive

tillage, however, is known to cause soil structure changes, increase the suscep-

tibility to soil erosion and reduce soil moisture. Loss o top soil due to tillage

may cause environmental degradation that can last or centuries. Ploughing is

also known to be the most destructive cultivation method aecting invertebrate

populations through physical destruction, desiccation, depletion o ood and

increased exposure to predators (Stoate et al., 2001).

8.2.1.3 Weed management using GM herbicide-tolerant maizeGenetically modied herbicide-tolerant (GMHT) crops permit the use o broad

spectrum herbicides such as glyphosate (Roundup Ready®) or gluosinate-

ammonium (Liberty®) at the postemergence phase. Growing GMHT crops

allows growers to use one broad-spectrum herbicide controlling a wide range

o both broadlea and grass weeds instead o several herbicides. GMHT maize

is presently primarily cultivated in the United States. Up to a ew years ago,

GMHT maize varieties only accounted or about 10 – 20% o the U.S. maize

acreage mainly because GMHT varieties were not available in many o the

most popular hybrid varieties and as these varieties were not approved orimport into Europe (Gianessi, 2005). Slower adoption o GMHT maize may also

have been due to the act that many armers are able to control weeds in maize

by conventional weed management at moderate costs. Farmers using GMHT

maize varieties generally have dicult-to-control weed problems that require

more costly programs, especially as conventional herbicide regimes may not

always provide consistent season-long control. With the increasing adoption

o stacked GM maize varieties in the U.S., the use o GMHT in combination

with insect-resistance is growing. In the European Union, GMHT maize is

currently not approved or commercial cultivation, but several applicationsare pending. Several GMHT maize events are approved or importation into

the EU as ood and eed.

A similar herbicide regime as with GMHT maize can be applied when using

Cleareld® maize varieties that are tolerant to the broad-spectrum herbicide

imidazolinone. Given that these varieties have been developed by traditional

breeding, the use o these varieties is however not specically regulated as or

GM crops.




112

8.2.1.5 Uncertainties related to the environmental impacts o GMHT maizeAs no GMHT maize has been approved or commercial cultivation in the EU,

there are currently theoretical experiences on how weed management in

Europe would change under real agricultural conditions (Devos et al., 2008;

Dewar, 2009). Environmental impacts due to crop management changes are

usually dicult to assess because they are oten caused by many interacting

actors and do only show up ater an extended period o time. Considering

the widespread eects modern agricultural systems had in the last decades,

changes in management practices are probably among the most infuential

actors that could lead to biodiversity changes. It is, however, crucial to bearin mind that management changes are not limited to the adoption o GMHT

crops. They do occur in any (non-GM) crop management strategy, and could

also be caused by the adoption o new pesticides, cultivation techniques or

crop varieties.

8.2.1.6 Potential environmental chances o growing herbicide-tolerant crops

Mitigation o negative environmental impacts o soil tillage

The negative environmental impacts caused by tillage operations can be miti-

gated by the application o conservation tillage. GMHT varieties acilitatearmers to adopt conservation tillage practices (Dewar, 2009; NRC, 2010). Since

weed control can be done during the postemergence phase, there is no need

or pre-seeding tillage and direct-seeding techniques can be applied. These

conservation tillage practices leave a layer o plant residues on the soil sur-

ace, preventing soil erosion, reducing evaporation and increasing the ability

o the soil to absorb moisture. A richer soil biota develops that can improve

nutrient recycling. Earthworm populations are generally higher in no-till elds

than in ploughed elds. There is also evidence that conservation tillage can

provide a wide range o chances to armland biodiversity by improving agri-cultural land as habitat or wildlie (Holland, 2004). The greater availability o

crop residues and weed seeds can improve ood supplies or insects, birds, and

small mammals. In addition to a reduction in soil erosion and degradation, less

requent soil cultivation also results in a decrease in the emission o green-

house gases, partly arising rom a reduction in uel use (Brookes and Baroot,

2005; Burney and Lobell, 2010).




Lower toxicity o the herbicides used with herbicide-tolerant cropsAlthough new herbicides show a relatively low toxicity, glyphosate and gluosi-

nate ammonium are less toxic to human health and the environment compared

to the replaced herbicides. In addition, they do not move readily to ground water,

resulting in ewer losses o chemicals by leaching and run-o rom the eld (Duke,

2005; Fernandez-Cornejo and Caswell, 2006; Devos et al., 2008; Dewar, 2009).

8.2.2 a rik mnt pprh r GMHT miz

Discussions among participants o the second expert workshop showed

consensus that there were no undamental dierences in how to approach a riskassessment or GMHT maize and Bt-maize apart rom the problem ormulation

that has to consider indirect eects or GMHT maize as well. An environmental

risk assessment o GMHT crops thus raises two major concerns: (1) direct toxic

eects and (2) indirect ood web eects.

8.2.2.1 Direct toxic eects o GMHT maize

Two types o direct toxicity have to be considered related to GMHT crops: (a)

eects resulting rom the genetic modication due to the expression o a novel

protein enabling the herbicide tolerance and (b) toxic eects caused by theherbicide applied.

Effects caused by the novel protein: Tolerance to glyphosate is enabled by the

introduction o a gene coding or 5-enolpyruvylshikimate-3-phosphate synthase

(EPSPS) rom Agrobacterium sp. As EPSPS enzymes occur in a wide range o

plants and ungi, and in some microorganisms, humans and the environment

have a long history o dietary exposure to these proteins. No adverse eects

have been reported with their intake (EFSA, 2009b). The same applies or the

gene coding or PPT-acetyltranserase (PAT) that is responsible or toleranceto gluosinate-ammonium. The PAT gene was originally isolated rom Strep-

tomyces viridochromogenes, an aerobic soil actinomycete. The PAT enzyme is

thereore naturally occurring in the soil and there are extensive studies showing

the saety and specicity o this enzyme.

Toxic eects caused by the herbicide applied: There is currently a debate among

dierent regulatory bodies in the EU under what regulation direct eects o

the herbicides used with GMHT crops have to be assessed. In principle, both





in environmental chances compared with the conventional practice.Depending on the herbicide management chosen, it can either enhance

weed seed banks and Autumn bird ood availability, or provide early

season chances to invertebrates and nesting birds (Ammon, 1993;

Dewar et al., 2003; May et al., 2005).

c. A technical approach o leaving a proportion o rows o arable elds

unsprayed to promote the growth o fowering and seedling arable

weeds (Pidgeon et al., 2007). Model calculations or GMHT sugar beets,

or example, have shown that leaving one row in every 50 unsprayed

would achieve a weed seed production equivalent to current conven-tional practice.

2. A more specic option aiming at managing low populations o “benecial

weeds” in the eld. Benecial weed species are characterized by a low level

o competition with the crop and by having a potential value as a resource or

higher trophic groups (Storkey and Westbury, 2007). Under standard weed

management practices, such less competitive annual broadlea plants tend

to have a selective disadvantage compared to competitive perennial species

including many grass weed species.31 Grass weed species tend to be relatively

poor resources or higher trophic groups. Lists o benecial weed speciesshowing a positive number o associated insect species, being part o the

diets o armland birds and showing a low to moderate competitive ability

with crops have been published (Marshall et al., 2003). Weed management

practices can be adapted to promote benecial weeds by relying on selective

herbicides that control grass weed inestations while leaving most broadlea

species (Storkey and Westbury, 2007).

8.2.3 appitin th prpd pprh

The above mentioned considerations show that one o the main challenges oran ERA o GMHT crops is developing a meaningul problem ormulation. In

contrast to an ERA or an insecticidal crop such as Bt-maize, where the risk

consists in more or less probable adverse eects on other organism than the

target pest, the range o indirect impacts on armland biodiversity that could

occur due to the adoption o GMHT crops is very large. This makes it di-

cult to ormulate clear risk hypotheses that could be tested, especially as the

conceivable impacts are oten more or less vague concerns than risks in the

pure sense o the term. At least rom an ecological point o view, perorming a


31 The shit rom broad-lea weed species to grass weeds has urthermore been avored

by the shit rom Spring to Autumn sowing dates due to the planting o Winter cereals

(Storkey and Westbury 2007).




ETHICAL REFERENCE SYSTEM TO ASSESS BIODIVERSITY 125

35 One could raise the objection that ethical considerations taking place within the law, as

it were, must take into account the “act o pluralism,” that is, the need to recognise a

multiplicity o perspectives in the ethical debate, in the sense that none o the pertinent

theories can simply be dismissed as irrational. This objection could be justied by

arguing that a modern constitutional state cannot be guided by the undamental

ethical principles o one specic ethical theory. Instead it is required to listen to the

dierent standpoints in order to base its own activity on a minimum ethical consensus.

This is the only way that the state can avoid being in thrall to particular groups andpreserve its ideological neutrality. This argument, however, is not plausible. I ethical

refections are a part o the executive process, there is no point in demanding that the

regulatory authorities should strive or a minimal consensus. Rather, the aim should

be to ensure that the best possible arguments prevail. For this reason, the authorities

involved should base their decisions on expert opinion, that is, on the expertise o

proessional ethicists.

decision-making such as the EM. As ar as risk assessment is concerned,ethics could either clariy on which ethical theory or theories the legal criteria

or acceptable risks are based or it could contribute to developing clearer and

more concrete criteria or acceptable risks. The ormer would help regulators

to better understand the normative basis o their decisions. The latter would

help them to better justiy and improve the quality o their decisions.

This critique o the EM, based on a scientic – and thereore allibilistic –

conception o ethics, was the starting-point or preparing the second workshop.

The conclusions to be drawn rom the ailure o the EM to deliver what wasoriginally hoped or are clear. In order to develop ethical criteria or regulatory

decision-making regarding the question o acceptable risks o GM crops or

biodiversity, one must proceed as ollows:

• Systematically develop possible ethical criteria or assessing changes in

biodiversity and evaluating the risks o biodiversity loss in agriculture.

• As these criteria are theory-based, the next step is to choose the most plau-

sible systematic normative ethical theory and spell out what this entails or

the acceptability o risks o biodiversity changes in agriculture.

• Apply these criteria to the issue o evaluating environmental impacts o GMcrops on biodiversity (in the context o a comparative risk assessment).

The time-consuming rst step was taken beore and during the second work-

shop. Special care was taken to ormulate all systematically relevant ethical options

o assessing changes in biodiversity and the risks associated with it. The result

is a new “ethical matrix” which, as the discussions during the second workshop

clearly showed, does not only have the advantage o systematic completeness. It

also allows regulators to nd out or themselves which normative ethical theory

they avor. Moreover, it helps them understand which ethical theory or ethical

considerations underlie the legal regulations on which they base their decisions.




130

Direct

Indirect

Direct

Indirect

Deontological

Utilitarian

Deontological

Utilitarian

Bintri

entri





such as octopuses and squids). Whether invertebrates and plants are sentient iscontentious. In the ollowing, we will assume that invertebrates and plants are

not sentient, that is, they are not able to eel pleasure and pain. From a patho-

centric perspective, the value o biodiversity is purely instrumental deriving

rom its unction or the survival or well-being o sentient beings.

3. Biocentrism: All and only living beings count morally or their own sake.

According to biocentrism, it ought to be respected that living beings have a

“good o their own.” This kind o good is dened as the one being essential or

the fourishing o living beings. It is usually based on the telos (purpose) o arespective species (that can only realize itsel in individual beings). It is what

constitutes their inherent worth (Eigenwert). That is, or example, why we do

not respect an indoor plant i we let it wither.

Harm is thus not bound to pain and suering as in pathocentrism, but

to fourishing. A living being is harmed to the extent that it is hindered rom

leading a species-appropriate lie – a lie that is typical or ideal o beings o its

kind – or rom unolding its individual capacities and talents. From a biocen-

tric viewpoint, biodiversity does not have a value o its own. It is valuable only

insoar as it is indispensable or useul or the survival or well-being o individual living beings o any kind.

4. Ecocentrism: Not only individual living beings, but also nonliving individual

entities such as stones and collective entities such as populations, habitats,

ecosystems, rivers or species count morally or their own sake, irrespective o

their importance or individual living beings.

Ecocentrism implies that not only individual living beings, but also entities

such as ecosystems or populations or nature as a whole can be in a good or a

bad condition or state, and that they can be harmed as such. This is becausethey also have a good o their own and can thereore fourish or be prevented

rom fourishing. In ecocentrism, biodiversity is regarded as a real – as opposed

to a nominal – entity.39 This entity ought to be protected as such because o its

inherent worth. This protection may be morally required even i it is against the

interests o living beings, human beings included.

39 Real means that the concept “biodiversity” is equivalent to an actual entity “out there”

in the world. “Nominal” means that the concept “biodiversity” is just an abstract

concept o classication (i.e., certain eatures or groups o eatures in the world are

subsumed under one term without claiming that there is something out there that is

common to all o these eatures thereby constituting biodiversity).




9.3.2 Dntgy Deontological theories claim that the ethical assessment o actions must not be

based on their consequences alone. From a deontological point o view, certain

acts are morally wrong in themselves, that is, irrespective o their consequences

(such as the intentional killing o an innocent person). Such acts are prohibited

even i they would increase or maximize net benet. Moral prohibitions o this

deontological kind serve as constraints with regard to utilitarian considerations

o total utility.

One way to conceptualize this idea o constraints is to reer to moral rights.

To say, or example, that an individual being has a moral right to lie means thatthis right must be respected even i disrespecting it would lead to an increase o

the total amount o value. In other words, this being should not to be killed even

i killing it would increase or maximize total utility.

Concerning the ethical assessment o risks, deontological theories imply

dening threshold values. Risks that are above a threshold value are morally

prohibited, whereas risks below such a value are permissible or acceptable.

The general risk ethical criterion is the criterion o due diligence or duty o

care. According to this criterion, other beings may only be exposed to a risk i

the risk-exposing person (or institution) has taken all necessary precautionary measures to avoid – with the highest probability easible – the occurrence o a

potential harm.

The general idea underlying the criterion o duty o care is as ollows: the

concept o harm is dened by reerence to moral rights – inringing on these

rights means causing harm to the right-holder. Such harm is morally unjustied

even i by inficting it, one increases total utility.41 Which rights exist depends on

the normative background theory. Most theories would agree on moral rights,

such as, among others, the right to dispose o one’s own body and one’s own lie,

a right to bodily integrity, a right to reely choose one’s own way o lie, a right toproperty and a right to lie.

To extend the harm principle – the prohibition to inringe on other being’s

rights without their consent – to all situations where others are exposed to a

risk, that is, to demand zero risk and thus to require that others ought not to be

exposed to any risk, cannot work in real lie or a undamental reason: it would


41 This is a provisional conception. It may be more plausible to decouple the concept o

harm completely rom the concept o the inringement or violation o moral rights.Consequently, there would be no moral harm. Harm would always be prudential

harm. As ar as moral rights are concerned, their violation would, o course, remain

morally wrong, but it would not be a harm inficted on the being(s) whose right(s)

is (are) violated. This conception, however, would not change the risk criterion as

sketched above. The core idea would still be using threshold values regarding the

violation o moral rights in order to separate acceptable rom inacceptable risks.




The risk o a reduction in biodiversity outside agricultural land due to cultiva-tion practices is irrelevant in this context, insoar as no individual property right

is aected. There is no probability that this right may be violated. From an anthro-

pocentric-deontological perspective, there is no other moral right that could play

a role in this context as there is no individual moral right to biodiversity.

With regard to the indirect assessment, the question is whether the armer

exposes other human beings on his land or on other armer’s land or outside

arable land to a risk. In order to be a risk, there must be a probability that biodi-

versity losses would violate certain moral rights o these humans. O the rights

mentioned above, the only relevant right is the right to lie and limb (bodily integrity). It is questionable, however, whether there is any probability (higher

than zero) that this right could be negatively aected by a reduction in biodiver-

sity. In the case that this would be possible, the deontological approach would

again determine a threshold value. Again, the decisive point would be that the

probability o a harm occurring must be very low.

Utilitarian anthropocentrism

According to utilitarianism, one ought to choose the action that maximizes the

expectation value or all those being aected – in this case, or all human beings who are aected.43 As ar as the armer’s own land is concerned, there are two

scenarios. In the rst scenario, he is the only human being aected by a loss o

biodiversity. Supposing that utilitarianism advocates duties to onesel, that is,

a moral duty to maximize one’s own happiness, one could argue that he must

calculate the possible (economic) gains and the possible costs and then choose

the option with the greatest expectable utility (total amount o chances minus

total amount o risk). It is then perectly conceivable that a reduction in biodi-

versity could be preerable to the status quo; or instance, i reduction means the

elimination o pests which in turn increases the probability o higher yields. Inthe second scenario, we try to calculate the expectation value or human beings

who are on this armer’s land. This case seems to be morally irrelevant because it

is hard to see what the possible benet or possible harm engendered by the loss

o biodiversity could consist o.

I other armer’s arable land is aected, the situation becomes more compli-

cated. What must be taken into account are the possible positive and negative

impacts o a reduction in biodiversity caused by all armers. The same reduction

in biodiversity that could be preerable to the status quo in the case o armer A


43 This kind o utilitarianism would have to argue that only the pleasure and pain o

human beings (or only the ullment o human desires) counts morally.




138

(because it has a higher expectation value) might have to be assessed as unde-sirable in the case o armer B because it has a lower expectation value than

the status quo. What utilitarianism requires in such a situation is guring out

which o the ollowing two options is the best (better than the other option i

there are only two). Option 1: the total sum o the expectation value o armer

A and the expectation value o armer B resulting rom the reduction in biodi-

versity. Option 2: the total sum o the expectation value o armer A and the

expectation value o armer B based on the status quo ante. I the expectation

value o option 1 is lower than the expectation value o option 2, the loss o

biodiversity must be judged negatively. In that case, there is a duty to reversethe loss o biodiversity, even i this means that armer A is worse o as a result

(in utilitarianism distributive considerations are irrelevant).

The same procedure has to be applied i we take the environment outside

agriculture into consideration.

9.4 appitin th prpd pprh

As mentioned beore, the ERS is the rst step in a systematic process aiming at the

specication and justication o plausible ethical criteria to valuate environmentalimpacts o GM crops on biodiversity. This step is important in two respects.

1. It lays the systematic oundation or the determination and justication o

the targeted ethical criteria. As pointed out above, the ull development o

these criteria would require choosing the most plausible systematic norma-

tive ethical theory and then applying the respective criteria to the ques-

tion o risk and harm regarding biodiversity. This, however, would be beyond

the scope o this project since this is the most basic systematic question o

normative ethics in general. What can be said, nonetheless, is the ollowing:

Irrespective o which theory turns out to be the most plausible one, all theo-ries share some common ground with regard to the assessment o the risks

associated with GM crops. They all agree – albeit or dierent reasons – that

the commercial use o GM crops can only be allowed i there is sucient

knowledge regarding the risks involved. Deontologists would argue that

this is required because without this knowledge it is not possible to deter-

mine whether the risks exceed the threshold value separating permissible

rom impermissible risks. Utilitarians would argue that without this knowl-

edge, it is not possible to determine whether the expected net benet o an




agriculture with GM crops (or the net benet o specic GM crops) wouldexceed the expected net benet o an agriculture without GM crops.

Furthermore, they all agree that the current scientic risk knowledge does

not yet suce to justiy the commercial release o GM crops. Given this

knowledge it is still not possible to determine in a suciently reliable

way whether the risks o these crops (especially regarding worst cases)

are above or below the threshold. Nor is it possible to determine whether

the commercial use o (specic) GM crops would lead to an (long term)

increase o the net benet in agriculture (as compared with conventional

agriculture). Thus, they would all plea or more risk research based on thestep-by-step procedure.

2. The ERS may be useul or certain practical purposes. Three are worth

mentioning:

a. By providing regulators with a general orientation grid, the ERS enables

them to see which ethical position they intuitively avor.

b. The ERS helps regulators to question their point o view and to modiy it

accordingly. It also acilitates a better and more refected understanding

o the ethical view(s) enshrined in the law. As a consequence, regula-

tors are better able to communicate their position to the “outside world”(industry, scientic community, media, and public at large).

c. The ERS improves comprehension o certain tensions, or instance

the tension between the two aspects o protection that are essential

or biodiversity: the protection o rare and threatened species on the

one hand, and the unctioning o ecosystem services on the other hand.

These two aspects may lead to dilemmas or decision makers because

the relevant legal regulations are based on two dierent ethical theo-

ries that exclude each other. Biodiversity conservation is usually based

on an ecocentric approach according to which biodiversity is intrinsi-cally valuable (and rarity is valuable in itsel). This implies that espe-

cially rare species are entities that deserve protection irrespective o

their value or ecosystem stability and or human beings. The protec-

tion o ecosystem services is mainly based on an anthropocentric

approach according to which biodiversity is only instrumentally valu-

able, namely, insoar as it is necessary or human well being. I only the

anthropocentric approach were relevant or the legal interpretation o

biodiversity, there would be no tension between the two main aspects





RECOMMENDATIONS FOR DECISION-MAKERS 145

10.2 Prttin g

Protection goals as specied by existing legislation are the exclusive starting

point or regulators or a denition o damage. Policy decisions on what ulti-

mately has to be protected are based on the existing legal rameworks. Nearly all

legal rameworks demand the conservation and sustainable use o biodiversity.

However, due to practical and nancial constraints, it is impossible to conserve

all components o biodiversity in the same manner. Hence, one needs to be

able to decide which components o biodiversity deserve particular protection.

Both rom an ethical and rom an ecological point o view, the legislative termsused to describe the protection goal “biodiversity” are too vague to be scien-

tically assessed. To dene scientically measurable characteristics or each

ecological entity deserving protection, we propose to rst dene assessment

endpoints that are an explicit expression o the environmental value that is to

be protected. In a second step, measurement endpoints that represent a meas-

urable ecological characteristic that can be related to the particular assess-

ment endpoint are to be dened. We have specied a matrix listing entities o

biodiversity that need protection as well as criteria that need to be considered

when dening corresponding assessment and measurement endpoints. Thepresented matrix can be used as a tool to structure the dialogue between the

dierent stakeholders, especially between regulators and applicants.

Regulatory authorities need to be aware that setting endpoints will require

balancing competing goals that will be a source o controversy. Biodiversity

conservation, or example, usually ocuses on rare and threatened species,

while restoring ecosystem services necessitates concentrating on ubiquitous

species as a species on the verge o extinction is likely to have less signi-

cant ecological relevance. Similarly, although there is general consensus that

the ecological entities specied by legislation to deserve protection have to berespected (such as Red List species and protected habitats), there is contro-

versy over the necessity to protect “common” species that are not explicitly

listed in the legislation, but that are reduced by common agricultural practices

(such as agricultural weeds that may be an essential part o ood webs in agri-

cultural landscapes contributing to armland biodiversity, but that also reduce

agricultural yield).




2ND

RecoMMeNDaTIoN

Regulators should actively promote approaches that enable stakeholders to agree on

what deserves protection because it is specically valued. Regulatory authorities can

use our matrix speciying protection goals as well as assessment and measurement

endpoints as a starting point or discussing generic and operational biodiversity

protection goals among all stakeholders o GM plants. The matrix can thereby be

adapted to regional or country specic needs. We recommend that science and

empirical evidence support the discourse about policy goals and indicate what

ecological theories tell us about the relevance o biodiversity.

146

10.3 Thrhd

Using a threshold to dene damage would be an elegant approach as it would

allow more or less unambiguous decisions which represent unacceptable

harm. In principle, every indicator value that would exceed the threshold would

indicate damage. Unortunately, the threshold principle is currently not appli-

cable in practice as the legal rameworks regulating the use o genetic engi-

neering in Switzerland and in the European Union (EU) do not dene specicthreshold values. This is mainly due to the diculty to dene thresholds or

ecological indicators. Ecological thresholds must be evaluated independently

or every single species or species group – a procedure that is usually not

appropriate or regulatory decision-making. With growing ecological experi-

ences on the cultivation o GMOs, thresholds or certain ecological groups may

become available. Nevertheless, one has to recognize that or most ecological

indicators ully operational thresholds will probably rarely be available in the

near uture. These complexities inevitably challenge decision-makers as they

generally have no clearly quantied thresholds that allow them to decide whatrepresents damage.




3RD

RecoMMeNDaTIoN

In theory, thresholds would be an elegant approach to dene damage as every indi-

cator value that would exceed the threshold would indicate damage. Hence, where

available, ecological thresholds should be used or decision-making. However,

regulators should keep in mind that ecological thresholds are seldom available in

practice as ecological sciences are most oten not able to provide precise threshold

values that would indicate when damage occurs. I thresholds are missing, damage

should be dened by using a baseline approach. The baseline allows one to deter-

mine when a change has to be regarded as a damage without relying on a precise,

xed threshold as the denition o damage is perormed by comparing two dierentstates. The rst state (e.g., the impacts caused by current agricultural management

practices) is thereby indicating what is accepted. Damage occurs i the dierence

between the accepted state and the state to be evaluated is judged to be suciently

large to be adverse.

Note, however, that rom an ethical point o view, this understanding o base-

line is problematic insoar as normatively speaking the baseline should include

criteria o acceptability. What is acceptable and what is actually accepted may, but

must not coincide.


10.4 cmprtiv mnt nd bin

The project showed that generic comparative assessments o dierent agricul-

tural management practices are very complex and dicult to perorm because

the impacts o cropping systems depend on a variety o dierent actors that

can oten only be assessed on a case-by-case basis. It is thereore not possible

to reduce these impacts to one common “currency” that would allow a generic

comparison. This challenges the initial assumption o the project that compar-ative assessments are the only way to allow a coherent evaluation o envi-

ronmental risks o GM crops. Every generic comparison inevitably results in

omitting important details that infuence the character and the magnitude o a

specic environmental impact o a particular technology .

It is important to recognize that regulatory authorities do oten not have

a legal basis or ormal obligation to perorm a comparative assessment as

existing agricultural technologies are evaluated according to dierent regula-

tory rameworks. More precisely, the regulatory ramework or pesticides uses




4TH RecoMMeNDaTIoN

Future political and legislative processes in Switzerland and in the EU should

attempt to harmonize the legal rameworks regulating GMOs and other agricultural

management practices that serve similar purposes (such as the use o pesticides).

Agricultural technologies should be assessed based on their individual properties

and on their risk or the environment and not just because they have been created

by a specic methodology or technology. Regulators rom dierent authorities

should initiate discussions to create a legal ramework or a ormal process thatallows or a comparative approach in situations where the comparison o technolo-

gies is appropriate and where there is a risk that the same protection goals may be

adversely aected.

148

other evaluation criteria than the one or GMOs. Hence, regulatory authoritiesoten reuse to compare the eects o GM crops to the eects caused by, or

example, pesticides. As regulators are orced to restrict their judgments to

the GM legislation, this leads to the irrational situation that the same envi-

ronmental impacts may be judged dierently depending on what agricultural

management practice caused them. In our opinion, this incoherence is one o

the main reasons or the current paralysis o the regulatory ramework or GM

crops in Switzerland and in the EU.

As long as technologies having similar environmental impacts ace dierent

regulatory regimes, a comparative assessment cannot be perormed and it isimpossible to draw a baseline or regulatory decisions. Baselines (i.e., the compa-

rator which serves as a basis or comparison) are nevertheless essential or deci-

sion-making as they dene what makes a change to be regarded damage. Legally,

the baseline or the evaluation o damage rom GMOs should be set by what is

already regarded representing damage today. I GMOs would be regulated on the

same legal basis as other agricultural management practices such as pesticides,

a comparison would be possible as the baseline would be characterized by those

environmental impacts that are already approved and thus implicitly accepted.




5TH RecoMMeNDaTIoN

We rmly believe that it is the responsibility o regulatory authorities to ensure that

the approval process o new technologies is coherent and perormed according to

transparent criteria. Regulatory authorities need to provide applicants with precise

operational protection goals and they need to clearly indicate what other relevant

inormation they may need or the evaluation o environmental impacts o GM crops.

Protection goals have to be dened by regulatory authorities in a consensus process

involving all stakeholders as applicants alone cannot address the public and seek

or consensus.


10.5 Wkn th rgutry ytm r GM rp

The legal requirements in Switzerland and in the EU require that risks or

the environment are assessed prior to the approval o GM crops. In order to

assess the saety o GM crops, applicants o new GM crops need clear indica-

tions rom regulatory authorities regarding protection goals (i.e., regarding the

environmental entities to be protected rom damage). Unortunately, regulatory

authorities in Switzerland and in the EU do currently not provide applicants with

sucient inormation on operational protection goals when evaluating environ-

mental risks o GM crops.The GMO regulation in the EU is marked by a poor communication

between applicants and the responsible regulatory authorities. Regulatory

authorities are usually not allowed to communicate with applicants to discuss

scientic questions and other issues relevant or preparing dossiers or the

application o GM crops. The process o communication between regulatory

authorities and applicants is practiced very dierently in other countries such

as Australia, New Zealand, Canada and the U.S., where applicants seek advice

rom regulators at an early stage and during the process o dossier preparation.

These practices clearly allow a more coherent application o the legal ramework and it would be helpul i such practices would also become accepted in

Switzerland and in the EU.




ReeReNcesCHAPTER 11




154

11 ReeReNces

acRe (2002). The criteria used by ACRE to gaugeharm when giving advice on the risks o releasing

genetically modied organisms to the environment,

Department or Environment, Food and Rural Aairs –

Advisory Committee on Releases to the Environment,

London. pp. 19

acRe (2007). Managing the ootprint o agriculture:

towards a comparative assessment o risks and

benets or novel agricultural systems, Department

or Environment, Food and Rural Aairs – Advisory

Committee on Releases to the Environment,

London. pp. 97. www.dera.gov.uk/ENVIRONMENT/acre/sewiderissues/acre-se-060317drat.pd

ammnn K, Runr a Mzyd P, nd Jt Y (2000).

Konzept und praktische Lösungsansätze zur ökolo-

gischen Begleitorschung. in “Nachhaltige Land-

wirtschat und grüne Gentechnik” (Schulte E and

Käppeli O, eds.), BATS – Fachstelle ür Biosicher-

heitsorschung und Abschätzung von Technikolgen,

Basel. pp. 96–112.

ammn HU (1993). Von der Unkrautbekämpung

zur Regulation der Grünbedeckung im Mais.Landwirtschaft Schweiz , 6: (11–12) 649–660.

avirn s, snvid o, Rmi J, Hrzg , nd

Bigr (2009). Case-specic monitoring o butter-

fies to determine potential eects o transgenic

Bt-maize in Switzerland. Agriculture, Ecosystems

and Environment , 131:137–144.

Bu/BlW (2008). Umweltziele Landwirtschat –

hergeleitet aus bestehenden rechtlichen Grund-

lagen, Umwelt-Wissen Nr. 0820. Bundesamt ür

Umwelt, Bern. pp. 221

Brtz R, Hink U, nd Kwrik I (2010). Proposed

denition o environmental damage illustrated by

the cases o genetically modied crops and invasive

species. Conservation Biology , 24: (3) 675–681.

BDM (2009). Zustand der Biodiversität in der

Schweiz: Ergebnisse des Biodiversitäts-Monitorings

Schweiz (BDM), Umwelt-Zustand 0911. Bundesamt

ür Umwelt, Bern. pp. 112

Buhmp Tl, nd chidr J (2001). Principleso biomedical ethics, Oxord University Press,

Oxord UK. pp. 472.

Bkmn V, d Bkkr e, Brnzk H, Bun o,

Dbnd M, rbrg eM, d Gr R, Ingnip

HW, ln J, Mphm B, Prbrg Nin a,

Tmkin s, Thrtnn e, Mir K, skrupinki B,

Brm Wa, Kir M, nd sndø P (2006).

Ethical biotechnology assessment tools or agricul-

ture and ood production, Agricultural Economics

Research Institute (LEI), The Hague NL. pp. 59.

www.ethicaltools.ino/content/ET1%20Final%20Report%20(Binnenwerk%2059p).pd

Bnbrk c (2001). Do GM crops mean less pesticide

use. Pesticide Outlook , 12: (October) 204–207.

Bnbrk c (2003). Impacts o genetically engi-

neered crops on pesticide use in the United States:

the rst eight years, BioTech InoNet. pp. 42

Bigr , Bbndrir D, nd Vn lntrn Jc (2010).

Risk assessment and non-target eects o egg para-

sitoids in biological control. in “Egg parasitoids in

agroecosystems with emphasis on Trichogramma”

(Consoli FL, Parra JRP and Zucchi RA, eds.), Vol. 9,

Springer, Dordrecht NL.

Birkhr a (1980). The German risk study or

Nuclear Power Plants. pp. 11. http://www.

iaea.org/Publications/Magazines/Bulletin/

Bull225_6/225_604792333.pd

Bktr c, Guihrd l, Kihingr o, Girrdin

P, Gn MB, nd Gird G (2009). Comparison o

methods to assess the sustainability o agricultural

systems. A review. Agronomy for Sustainable

Development , 29: (1) 223–235.

Bhn M, Mn a, stt s, angvin ,

Grifth B, Krgh PH, Znidri M, nd Dzrki

s (2008). A qualitative multi-attribute model or

economic and ecological assessment o genetically

modied crops. Ecological Modelling, 215: (1–3)

247–261.




REFERENCES 155

Brimnr Ta, Givn GJ, nd stphnn GR (2005).Infuence o herbicide-resistant canola on the envi-

ronmental impact o weed management.

Pest Management Science, 61: (1) 47–52.

Brk G, nd Brt P (2005). GM crops: the global

economic and environmental impact – the rst nine

years 1996–2004. AgBioForum, 8: (2&3) 187–196.

Burny JaD, s J, nd lb DB (2010). Greenhouse

gas mitigation by agricultural intensication.

Proceedings of the National Academy of Sciences of

the United States of America, 107:12052–12057.

Buw (1999). Landschatskonzept Schweiz,

Bundesamt ür Umwelt, Wald und Landschat,

Bern. pp. 56

BV (sR 101). Bundesverassung der Schweiz-

erischen Eidgenossenschat, Systematische

Sammlung des Bundesrechts, Bern. pp. 90

cirn J (1983). Are single species toxicity tests

alone adequate or estimating environmental

hazard. Hydrobiologia, 100:47–57.

cndf M, Bigr , cmpb P, Himbh U,

shmuk R, angi G, Bkkr , Brwn K, cri G,

Dintr a, rti D, rtr R, Gthmnn a, Hn s,

Md-Brigg M, Mndri M, Numnn P, Pqu-

ini e, Pw W, Rbut JN, Rmijn K, shr

B, Thim T, Ur a, Vrgnt c, nd Vgt H (2000).

Principles or regulatory testing and interpretation

o semi-eld and eld studies with non-target

arthropods. Anzeiger für Schädlingskunde –

Journal of Pest Science, 73: (6) 141–147.

cr TM, nd o’Dhrty G (1999). On the useo surrogate species in conservation biology.

Conservation Biology , 13: (4) 805–814.

crpntr J, ft a, Gd T, Hmmig M, ontd D,

nd snku s (2002). Comparative environmental

impacts o biotechnology-derived and traditional

soybean, corn, and cotton crops, Council or

Agricultural Science and technology, Ames, Iowa.

pp. 200

cBD (1992). Convention on Biological Diversity,United Nations Environment Programme, Rio de

Janeiro. pp. 31

cBD (2000). Sustaining lie on earth – how the

Convention on Biological Diversity promotes nature

and human well-being, United Nations Environment

Programme, Montreal. pp. 20

cBD (2009). MOP 4 Decision BS-IV/12 Liability and

Redress under the Cartagena Protocol on Biosaety,

Convention on Biological Diversity, Montreal. pp. 24

chmbrin De, fur RJ, Bun RGH, Dukwrth Jc,

nd shrubb M (2000). Changes in the abundance o

armland birds in relation to the timing o agricultural

intensication in England and Wales.

Journal of Applied Ecology , 37: (5) 771–788.

chpin s, Zvt es, evinr VT, Nyr Rl,

Vituk PM, Rynd Hl, Hpr DU, lvr s,

s oe, Hbbi se, Mk Mc, nd Diz s (2000).

Consequences o changing biodiversity.

Nature, 405: (6783) 234–242.

ctnz R, darg R, dGrt R, rbr s, Gr M,

Hnnn B, limburg K, Nm s, oni RV, Pru

J, Rkin RG, suttn P, nd vndnBt M (1997).

The value o the world’s ecosystem services and

natural capital. Nature, 387: (6630) 253–260.

Diy G (1999). Developing a scientic basis or

managing earth’s lie support systems Conservation

Ecology , 3: (2) 14.

Diy Gc, sdrqvit T, aniyr s, arrw K, Dgupt

P, ehrih PR, k c, Jnn a, Jnn Bo,

Kutky N, lvin s, lubhn J, Mr KG, simpnD, strrtt D, Timn D, nd Wkr B (2000).

Ecology – The value o nature and the nature o

value. Science, 289: (5478) 395–396.

d Mgd Ra, Brv a, nd crikmr N (2001).

How Bacillus thuringiensis has evolved specic

toxins to colonize the insect world.

Trends in Genetics, 17: (4) 193–199.




156

Dv Y, cugnn M, Vrguht s, Buk R,Hrt G, sturbut W, nd Rhu D (2008).

Environmental impact o herbicide regimes used

with genetically modied herbicide-resistant maize

Transgenic Research, 17:1059–1077.

Dwr aM (2009). Weed control in glyphosate-

tolerant maize in Europe. Pest Management

Science, 65: (10) 1047–1058.

Dwr aM, My MJ, Wiwd IP, Hyk la,

chmpin GT, Grnr BH, snd RJN, Qi aM, nd

Pidgn JD (2003). A novel approach to the use o genetically modied herbicide tolerant crops or

environmental benet. Proceedings of the Royal

Society of London Series B-Biological Sciences, 270:

(1513) 335–340.

Diz s, rgin J, chpin s, nd Timn D (2006).

Biodiversity loss threatens human well-being.

Plos Biology , 4: (8) 1300–1305.

Divy GP, R R, sr MK, Hmih Rl, stny-

Hrn De, cvin DD, Ru JM, nd andrn Pl

(2004). Eects on monarch butterfy larvae

(Lepidoptera : Danaidae) ater continuous exposure

to Cry1Ab-expressing corn during anthesis.

Environmental Entomology , 33: (4) 1116–1125.

Dui P (1994). Rote Liste der geährdeten Tierarten

in der Schweiz, Bundesamt ür Umwelt, Wald und

Landschat, Bern. pp. 97

Dui P, nd obrit MK (2003). Biodiversity indica-

tors: the choice o values and measures. Agricul-

ture, Ecosystems and Environment , 98:87–98.

Duk so (2005). Taking stock o herbicide-resistantcrops ten years ater introduction.

Pest Management Science, 61: (3) 211–218.

easac (2009). Ecosystem services and biodiversity

in Europe, European Academies Science Advisory

Council, London UK. pp. 70. www.easac.eu/docu-

ment.asp?id=90&pageno=&detail=5&parent=

esa (2006). Guidance document o the scientic

panel on genetically modied organisms or the

risk assessment o genetically modied plants and

derived ood and eed. The EFSA Journal, 99:1–100.

esa (2009). Scientic opinion on statisticalconsiderations or the saety evaluation o GMOs,

European Food Saety Authority, Parma. pp. 62

esa (2009b). Scientic opinion: Applications

(reerences EFSA-GMO-NL-2005-22, EFSA-GMO-

RX-NK603) or the placing on the market o the

genetically modied glyphosate tolerant maize

NK603 or cultivation, ood and eed uses, import

and processing and or renewal o the authorisation

o maize NK603 as existing products, both under

Regulation (EC) No 1829/2003 rom Monsanto.

The EFSA Journal, 1137: (1–50).

ehrih PR, nd ehrih aH (1981). Extinction. The

causes and consequences o the disappearence o

species, Random House, New York. pp. 305.

e Hhimi M, nd vn shipp a (2003). Crea-

Space – eine Methode zur Entwicklung des krea-

tiven Potenzials in Teams und grösseren Gruppen.

Organisationsberatung – Supervision – Coaching,

2:137–144.

etn cs (1958). The ecology o invasions by

animals and plants, Methuen and Co. Ltd., London.

pp. 181.

ePa (2001). Biopesticides registration action docu-

ment – Bacillus thuringiensis plant-incorporated

protectants, U.S. Environmental Protection Agency,

Washington DC. pp. 481. www.epa.gov/pesticides/

biopesticides/pips/bt_brad.htm

ePa (2009). Pesticides: Topical & Chemical Fact

Sheets. U.S. Environmental Protection Agency

http://www.epa.gov/pesticides/actsheets/

eurpn cmmiin (1991). Council Directive o 15

July 1991 concerning the placing o plant protection

products on the market (91/414/EC), Commission

o the European Communitites, Brussels. pp.

eurpn cmmiin (1992). Council Directive

92/42/EEC o 21 May 1992 on the conservation

o natural habitats and o wild auna and fora,

Commission o the European Communities,

Brussels. pp. 66




eurpn cmmiin (2002). Guidance documenton terrestrial ecotoxicology under Council Directive

917414/EEC, Directorate E – Food Saety: plant

health, animal health and welare, international

questions, Brussels. pp. 39

eurpn cmmiin (2004). Directive 2004/35/

CE o the European parliament and o the council o

21 April 2004 on environmental liability with regard

to the prevention and remedying o environmental

damage, European Parliament and the Council o

the European Union, Strassburg. pp. 19

eurpn cmmunity (2001). Directive 2001/18/

EC o the European Parliament and o the Council o

12 March 2001 on the deliberate release into the

environment o genetically modied organisms and

repealing Council Directive 90/220/EEC, European

Parliament and the Council o the European Union,

Brussels. pp. 38

eurpn Unin (2009). EU Pesticides database.

http://ec.europa.eu/sanco_pesticides/public/

index.cm

eurpn Unin (2009b). Regulation (EC) No

1107/2009 o the European Parliament and o the

Council o 21 October 2009 concerning the placing

o plant protection products on the marketand

repealing Council Directives 79/117/EEC and

91/414/EEC, European Parliament and the Council

o the European Union, Brussels. pp. 50

iing l, nd Grgry R (2003). Ten common

mistakes in designing biodiversity indicators

or orest policy. Journal of Environmental

Management , 68: (2) 121–132.

irwthr PG (1991). Statistical power and

design requirements or environmental monitoring.

Australian Journal of Marine Freshwater Research,

42:555–567.

d cK, d siv PM, su JP, d B , Bugtr R,

Grndin U, Hring D, lvr s, Muntrd o,

Prd I, Prt M, Rmbk J, sndin l, Jn KB,

nd Hrrin P (2009). Indicators o biodiversity and

ecosystem services: a synthesis across ecosystems

and spatial scales. Oikos, 118: (12) 1862–1871.

rnndz-crnj J, nd cw M (2006). The rst decade o genetically engineered crops

in the United States, United States Department

o Agriculture – Economic Research Service,

Washington DC. pp. 30

rnndz-crnj J, nd M Brid WD (2002).

Adoption o bioengineered crops, Economic

Research Service, United States Department

o Agriculture, Washington, DC. pp. 61

rrir ePD, d snt HP, ct JR, D-Pi H,

nd Rumjnk NG (2010). Microbial soil qualityindicators under dierent crop rotations and

tillage management. Revista Ciencia Agronomica,

41: (2) 177–183.

ur RJ, Grgry RD, Gibbn DW, Mrhnt JH,

Win JD, Bii sR, nd crtr N (1995).

Population declines and range contractions among

lowland armland birds in Britain. Conservation

Biology , 9: (6) 1425–1441.

Gi N, s JM, stt J, nd Viir Be

(2009). Economic valuation o the vulnerability o

world agriculture conronted with pollinator decline.

Ecological Economics, 68: (3) 810–821.

Gärdnr U (2001). Classiying threatened species

at national versus global levels. Trends in Ecology &

Evolution, 16: (9) 511–516.

Gtn KJ (1996). What is biodiversity. in “Biodiver-

sity – a biology o numbers and dierence” (Gaston

KJ, ed.), Blackwell Science, Oxord UK. pp. 1–9.

Gini lP (2005). Economic and herbicide

use impacts o glyphosate-resistant crops.Pest Management Science, 61: (3) 241–245.

Gii D (2000). Philosophical theories o prob-

ability, Routledge, London, New York. pp. 240.

Gnth Y (1994). Rote Liste der geährdeten

Tagalter der Schweiz. in “Rote Liste der geähr-

deten Tierarten der Schweiz” (Duelli P, ed.),

Bundesamt ür Umwelt, Wlad und Landschat,

Bern. pp. 48–51.

REFERENCES 157




158

Grimm V, nd Wi c (1997). Babel, or the ecolog-ical stability discussions: An inventory and analysis

o terminology and a guide or avoiding conusion.

Oecologia, 109: (3) 323–334.

Grmn P, Brn J, Btt T, Gd a, Gdmn

I, Gundrn l, lvinn B, Pmr M, Pr H,

Ptrn G, P N, Rjki D, Rynd J, Turnr M,

Wthr K, nd Win J (2006). Ecological thresh-

olds: The key to successul environmental manage-

ment or an important concept with no practical

application? Ecosystems, 9: (1) 1–13.

GTG (sR 814.91). Bundesgesetz über die

Gentechnik im Ausserhumanbereich (Gentechnikge-

setz), Systematische Sammlung des Bundesrechts,

Bern. pp. 14

Gundrn lH (2000). Ecological resilience – in

theory and application. Annual Review of Ecology

and Systematics, 31:425–439.

Hking I (2001). An introduction to probability

and inductive logic, Cambridge University Press

Cambridge. pp. 320.

Hi Rs (2002). Assessing the risks associated

with new agricultural practices. Nature, 418:

(6898) 685–688.

Hjk a (2009). Interpretations o probability,

Stanord Encyclopedia o Philosophy. pp. http://

plato.stanord.edu/entries/probability-interpret

Hänr R, nd shirhr B (2009). First step

towards CWR conservation in Switerland. Crop

wild relative (7) 14–17.

Häni , Ppw G, Rinhrd H, shwrz a, nd

Tnnr K (2006). Pfanzenschutz im integrierten

Ackerbau, 6th/Ed., Landwirtschatliche Lehr-

mittelzentrale, Zollikoen.

Hnki I (1998). Metapopulation dynamics.

Nature, 396: (6706) 41–49.

Hrmr R, Ptrkn G, Krr G, nd Putn P (2001).

Vegetation changes during 100 years o development

o two secondary woodlands on abandoned arable

land. Biological Conservation, 101: (3) 291–304.

Hw c, Hughtn aJ, obrn Jl, Ry DB,crk sJ, Prry JN, Rthry P, Bhn Da, Brk DR,

chmpin GT, Dwr aM, Hrd Ms, Wiwd IP,

Dni Re, Yung MW, Prih aM, stt RJ,

irbnk lG, nd squir GR (2003). Responses

o plants and invertebrate trophic groups to

contrasting herbicide regimes in the Farm Scale

Evaluations o genetically modied herbicide-

tolerant crops. Philosophical Transactions of

the Royal Society of London Series B-Biological

Sciences, 358: (1439) 1899–1913.

Hzu GW, Mtzgr MJ, Mühr ca, Prz-sb M,Rntzdr c, nd andrn e (in pr). Euro-

pean environmental stratications and typolo-

gies: An overview. Agriculture, Ecosystems and

Environment doi:10.1016/j.agee.2010.01.009.

Hrd Ms, Rthry P, Prry JN, nd irbnk lG

(2005). Pedicting longer-term changes in weed

populations under GMHT crop management.

Weed Research, 45: (5) 331–338.

Höt H, nd Whity HR (1989). Insecticidal crystal

proteins o Bacillus thuringiensis. MicrobiologicalReviews, 53: (2) 242–255.

Hnd JM (2004). The environmental conse-

quences o adopting conservation tillage in Europe:

reviewing the evidence. Agriculture Ecosystems &

Environment , 103: (1) 1–25.

Hing cs (1973). Resilience and stability o

ecological systems. Annual Review of Ecology

and Systematics, 4:1–23.

Hpr DU, chpin fs, ew JJ, Htr a, Inhuti P,

lvr s, lwtn JH, ldg DM, lru M, Nm s,

shmid B, st H, symtd aJ, Vndrmr J,

nd Wrd Da (2005). Eects o biodiversity on

ecosystem unctioning: A consensus o current

knowledge. Ecological Monographs, 75: (1) 3–35.

Huggtt aJ (2005). The concept and utility o

‘ecological thresholds’ in biodiversity conservation.

Biological Conservation, 124:301–310.




160

Mrh eJP, Brwn VK, Btmn ND, lutmn PJW,squir GR, nd Wrd lK (2003). The role o weeds

in supporting biological diversity within crop elds.

Weed Research, 43: (2) 77–89.

My MJ, chmpin GT, Dwr aM, Qi a, nd

Pidgn JD (2005). Management o genetically

modied herbicide-tolerant sugar beet or spring

and autumn environmental benet Proceedings

of the Royal Society B-Biological Sciences,

272: (1559) 111–119.

Mcnn Ks (2000). The diversity-stability debate.Nature, 405: (6783) 228–233.

MNughtn sJ (1977). Diversity and stability

o ecological communities – comment on role

o empiricism in ecology. American Naturalist ,

111: (979) 515–525.

Mi M, Murn P, Mu T, Bigr , Pn X,

Viidi P, ott s, antihi D, Ki J, Pink Z,

Drnr Z, vn dr Wid R, Grtn J, czmbr e,

admzyk J, Thibrd JB, Mndr B, crdn

Nin G, Pun RT, Zimmrmnn o,

Vrhw a, nd odnburg e (2010). Pests,

pesticide use and alternative options in Euro-

pean maize production: current status and

uture prospects. Journal of Applied Entomology ,

134:357–375.

Mndhn M, Kugh J, Vituzi Z, nd Mtthw

K (2003). Are Bt crops sae? Nature Biotechnology ,

21: (9) 1003–1009.

Mphm B, Kir M, Thrtnn e, Tmkin s,

nd Mir K (2006). Ethical Matrix: manual,

Agricultural Economics Research Institute (LEI),

The Hague NL. pp. 45. http://www.ethicaltools.

ino/content/ET2%20Manual%20EM%20(Binnen-

werk%2045p).pd

Mrrim-Wbtr onin (2011). Merriam-Webster

Online Dictionary copyright © 2005 by Merriam-

Webster, Incorporated Merriam-Webster Inc.,

P.O. Box 281, Springeld, MA www.m-w.com

Minium eytm amnt (2005). Eco-systems and human well-being: Synthesis, Island

Press, Washington DC. pp. 137

Mnn ac, nd Brbri P (2008). Functional biodi-

versity: an agroecosystem approach. Agriculture

Ecosystems & Environment , 127: (1–2) 7–21.

Mr D, Gygx a, Bäumr B, Wyr N, nd P

R (2002). Rote Liste der geährdeten Farn- und

Blütenpfanzen der Schweiz, Bundesamt ür

Umwelt, Wald und Landschat, Bern; Zentrum des

Datenverbundnetzes der Schweizer Flora, Chambésy; Conservatoire et Jardin botanique de la Ville de

Genève, Chambésy. pp. 118

Murdin R (2001). Ecological thresholds: a survey.

Ecological Economics, 38: (1) 7–24.

Nm s, nd li sB (1997). Biodiversity enhances

ecosystem reliability. Nature, 390: (6659) 507–509.

Nrnj se (2009). Impacts o Bt crops on non-

target invertebrates and insecticide use patterns.

CAB Reviews: Perspectives in Agriculture, Veteri-

nary Science, Nutrition and Natural Resources, 4:

(11) 1–11.

NHG (sR 451). Bundesgesetz über den Natur-

und Heimatschutz, Systematische Sammlung des

Bundesrechts, Bern. pp. 24

NHV (sR 451.1). Verordnung über den Natur-

und Heimatschutz, Systematische Sammlung des

Bundesrechts, Bern. pp. 30

Nihu R (1984). The uture role o risk assessment

in nuclear saety. pp. 7. www.iaea.org/Publications/Magazines/Bulletin/Bull254/25402043036.pd

Nöh I (2002). Bewertung von Umweltwirkungen

gentechnisch veränderter Organismen (GVO).

UWSF – Zeitschrift für Umweltchemie und

Ökotoxikologie, 14: (3) 155–163.

Nrtn BG (2008). Towards a policy-relevant

denition o biodiversity. in “Saving biological

diversity” (Askins RA, Dreyer GD, Visqilio GR and

Whitelaw DM, eds.), Springer. pp. 11–20.




162

Rmi J, Mi M, Rybud a, nd HmihRl (2008b). Impact o insect-resistant genetically

modied crops on non-target arthropods. in “Envi-

ronmental impact o genetically modied/novel

crops” (Ferry N and Gatehouse AMR, eds.), CABI,

Wallingord, UK. pp. 163–196.

Rhwitz I, Gbri D, Thrntk T, nd Thi

c (2005). The eects o landscape complexity

on arable weed species diversity in organic and

conventional arming. Journal of Applied Ecology ,

42: (5) 873–882.

snvid o, Rmi J, nd Bigr (2009). An

approach or post-market monitoring o potential

environmental eects o Bt-maize expressing

Cry1ab on natural enemies. Journal of Applied

Entomology , 133:236–248.

snvid o, Widmr , Winzr M, nd Bigr

(2005). A conceptual ramework or the design o

environmental post-market monitoring o geneti-

cally modied plants. Environmental Biosafety

Research, 4:13–27.

shr M, crpntr s, y Ja, k c, nd

Wkr B (2001). Catastrophic shits in ecosystems.

Nature, 413: (6856) 591–596.

shr M, nd crpntr sR (2003). Catastrophic

regime shits in ecosystems: linking theory to

observation. Trends in Ecology & Evolution,

18: (12) 648–656.

shnp e, crikmr N, Vn Ri J, lru D, Bum J,

itn J, Zigr DR, nd Dn DH (1998).

Bacillus thuringiensis and its pesticidal crystal

proteins. Microbiology and Molecular Biology

Reviews, 62: (3) 775–806.

shwrtz MW, Brighm ca, Hkm JD, lyn KG,

Mi MH, nd vn Mntgm PJ (2000). Linking

biodiversity to ecosystem unction: implications

or conservation ecology. Oecologia, 122: (3)

297–305.

scNaT (2006). Die Schweiz braucht eine Biodi-versitätsstrategie!, Swiss Biodiversity Forum,

Berne. pp. 4. www.biodiversity.ch/downloads/

Argumente_D.pd

sr MK, Hmih Rl, stny-Hrn De, obr-

hur Ks, Pnt JM, Mtti HR, sigrid BD,

nd Divy GP (2001). Impact o Bt corn pollen on

monarch butterfy populations: A risk assessment.

Proceedings of the National Academy of Sciences

of the United States of America, 98: (21)

11937–11942.

shnnn c (2008). Biotic interactions, ecological

knowledge and agriculture. Philosophical Transac-

tions of the Royal Society B-Biological Sciences,

363: (1492) 717–739.

singnbrg a, Brt H, vn dr Windt H,

Rdmkr K, eihr l, nd Turnr K (2009).

Study on understanding the causes o biodiversity

loss and the policy assessment ramework, ECORYS

Nederlands BV, Rotterdam NL. pp. 206. http://

ec.europa.eu/environment/enveco/biodiversity/

pd/causes_biodiv_loss.pd

sRU (2004). Environmental report 2004: ensuring

environmental protection capacity, German Advisory

Council on the Environment. pp. 115

stt c, Bdi a, Bj P, Btmn ND, Hrzn I,

vn Drn a, d sn GR, Rky l, nd Rmw c

(2009). Ecological impacts o early 21st century

agricultural change in Europe – A review. Journal of

Environmental Management , 91: (1) 22–46.

stt c, Btmn ND, Brrh RJ, crvh cR,

d sn GR, nd edn P (2001). Ecological impacts

o arable intensication in Europe. Journal of

Environmental Management , 63: (4) 337–365.

strky J, Bhn Da, Hughtn aJ, chmpin

GT, Prry JN, Pppy GM, nd Wiwd IP (2008).

Providing the evidence base or environmental risk

assessments o novel arm management practices.

Environmental Science & Policy , 11: (7) 579–587.




strky J, nd Wtbury DB (2007). Managingarable weeds or biodiversity. Pest Management

Science, 63: (6) 517–523.

suding KN, nd Hbb RJ (2009). Threshold models

in restoration and conservation: a developing rame-

work. Trends in Ecology & Evolution, 24: (5) 271–279.

sutr GW (2000). Generic assessment endpoints

are needed or ecological risk assessment.

Risk Analysis, 20: (2) 173–178.

sutr GW (2007). Ecological risk assessment,

Second/Ed., CRC Press, Boca Raton FL, USA. pp. 643.

swit MJ, Iz aMN, nd vn Nrdwijk M (2004).

Biodiversity and ecosystem services in agricultural

landscapes – are we asking the right questions? Agri-

culture, Ecosystems and Environment , 104:113–134.

Thmpn R, nd strzmki BM (2007).

What does biodiversity actually do? A review or

managers and policy makers. Biodiversity and

Conservation, 16: (5) 1359–1378.

Timn D (2000). Causes, consequences and ethicso biodiversity. Nature, 405: (6783) 208–211.

Timn D, cmn KG, Mtn Pa, Nyr R,

nd Pky s (2002). Agricultural sustainability

and intensive production practices. Nature, 418:

(6898) 671–677.

Thrntk T, Kin aM, Kru a, stn-

Dwntr I, nd Thi c (2005). Landscape

perspectives on agricultural intensication and

biodiversity– ecosystem service management.

Ecology Letters, 8: (8) 857–874.

VGVl (sR 817.022.51). Verordnung des EDI über

gentechnisch veränderte Lebensmittel, Systema-

tische Sammlung des Bundesrechts, Bern. pp. 12

Wkr B (1995). Conserving biological diversity

through ecosystem resilience. Conservation

Biology , 9: (4) 747–752.

REFERENCES 163

Wkr BH (1992). Biodiversity and ecologicalredundancy. Conservation Biology , 6: (1) 18–23.

Wtkinn aR, rktn RP, Rbinn Ra, nd

suthrnd WJ (2000). Predictions o biodiversity

response to genetically modied herbicide-tolerant

crops. Science, 289: (5484) 1554–1557.

Wi l (2002). Charakterisierung der Risiken

der Kernenergienutzung. pp. 147. www.uni-

saarland.de/ak7/ze/AKE_Archiv/AKE2004H/

AKE2004H_Vortraege/AKE2004H_05Weil_

Charakterisierung_derKernenergieRisiken/AKE2004H_05aWeil_habil2002_Risiken-derKern-

energienutzung.pd

Wiby a, nd Thm MB (2002). Natural enemy

diversity and pest control: patterns o pest

emergence with agricultural intensication.

Ecology Letters, 5: (3) 353–360.

With Ka, Pvuk DM, Wrhuk Jl, ot RK, nd

ihr Jl (2002). Threshold eects o landscape

structure on biological control in agroecosystems.

Ecological Applications, 12: (1) 52–65.

Wnbrgr ll, Nrnj se, lundgrn JG, Bitzr

RJ, nd Wtrud ls (2008). Bt crop eects on unc-

tional guilds o non-target arthropods: a meta-anal-

ysis. PLoS One, 3: (5) e2118. doi:10.1371/journal.

pone.0002118.

Wt JD, K P, Rybud a, itzptrik JW, Bur-

hik M, Gry a, oin ss, shimnn J, sr M, nd

Wu (2010). Problem ormulation in the environ-

mental risk assessment or genetically modied

plants. Transgenic Research, 19:425–436.

Zurbrügg c, Hönmnn l, Mi M, Rmi J,

nd Nntwig W (2010). Decomposition dynamics

and structural plant components o genetically

modied Bt maize leaves do not dier rom leaves

o conventional hybrids. Transgenic Research, 19:

(2) 257–267.




aNNeX 1: lIsT o

WoRKsHoP PaRTIcIPaNTs

Annexes




166

aNNeX 1: lIsT o WoRKsHoP PaRTIcIPaNTs

andr Bhmnn (c-nvnr) Ethik im Diskurs

Katzenbachstrasse 192

CH - 8052 Zürich

[email protected]

rnz Bigr (cnvnr)

Agroscope Reckenholz Tänikon Research Station ART

Reckenholstrasse 191

CH - 8046 Zürich

[email protected]

Mrk BhnInstitute Joze Stean

Department o Knowledge Technologies

Jamova 39

SI - 1000 Ljubljana

[email protected]

Jérôm crtt

Laboratoire Sols et Environnement

UMR INPL-INRA 1120

2 av. de Forêt de Haye BP 172

FR - 54505 Vandoeuvre-lès-Nancy

[email protected]

Ynn Dv

European Food Saety Authority

Largo N. Palli 5/A

IT - 43100 Parma

[email protected]

Ptr Dui

Swiss Federal Institute or Forest,

Snow and Landscape Research

Zürcherstrasse 111

CH - 8903 Birmensdor [email protected]

Brbr ekbm

Swedish University o Agricultural Sciences

Department o Entomology

PO Box 7044

SE - 75007 Uppsala

[email protected]

ahim GthmnnFederal Oce o Consumer Protection

and Food Saety (BVL)

Reerat 404, Mauerstrasse 39–42

DE - 10117 Berlin

[email protected]

Mr M.c. Gikn

National Institute or Public Health

and the Environment (RIVM)

P.O. Box 1

NL - 3720 BA Bilthoven

[email protected]

Mtthi Kir

National Committee or Research Ethics

in Science and Technology (NENT)

Prinsensgate 18 - PO Box 522 Sentrum

NO - 0105 Oslo

[email protected]

liij Kdin

Vilnius University

Department o Microbiology and Plant Physiolog

Chiurliono str. 21/27LT - 03127 Vilnius

[email protected]

Juin Kindrrr*

Delt University o Technology

Department o Biotechnology

Julianalaan 67

NL - 2628 BC Delt

[email protected]

Jz Ki

Szent Istvan UniversityPlant Protection Institute

Pater Karoly utca 1

HU - 2100 Gödöllö

[email protected]

Pu Hnning Krgh

University o Aarhus, NERI

Department o Terrestrial Ecology

Vejlsøvej 25

DK - 8600 Silkeborg

[email protected]




ANNEX 1: LIST OF WORKSHOP PARTICIPANTS 167

* participated in the rst expert workshop;

+ participated in the second workshop

Zjk Kun (es rpprtur)*University o Zagreb

Department o Chemistry

HR - Zagreb

[email protected]

Pr Nin Kudk*

University o Aarhus

Department o Integrated Pest Management

Forsøgsvej 1

DK - 4200 Slagelse

[email protected]

Jui link

Swiss Expert Committee or Biosaety

c/o Federal Oce or the Environment

CH - 3000 Berne

[email protected]

cpr linntd

Norwegian Biotechnology Advisory Board

Rosenkrantz gt. 11 - P.O. Box 522 Sentrum

NO - 0105 Oslo

[email protected]

J Prry*

Oaklands Barn

Lug’s Lane, Broome

UK - Norolk NR35 2HT

[email protected]

annik Pyir*

Monsanto Europe S.A.

Avenue de Tervuren 270–272

BE - 1150 Brussels

[email protected]

an Rybud

Syngenta

Jealott’s Hill International Research Centre

UK - Bracknell RG42 6EY

[email protected]

Ku Ptr Ripp (c-nvnr) +

Ethik im Diskurs

Restelbergstrasse 60

CH - 8044 Zürich

[email protected]

Jörg Rmi* Agroscope Reckenholz Tänikon Research Station ART


CH - 8046 Zürich

[email protected]

Dmtri G. Rupki*

Aristotle University o Thessaloniki

School o Agriculture

Dept. o Genetics and Plant Breeding

GR - 54124 Thessaloniki

[email protected]

oivir snvid (c-nvnr)



CH - 8046 Zürich

[email protected]

Jrmy swt

6 The Green, Willingham

UK - Cambridge CB4 5JA

[email protected]

Hn vn Tri*

Archeon

Strandstigen 20 A 4

FIN - 00330 Helsinki

helena.troil@archeon.

stn Win*

Linköping University

Department o Medical and Health Sciences

IHS Campus US

SE - 58183 Linköping

[email protected]

Mih Winzr*



CH - 8046 Zürich

[email protected]

ann-Gbri Wut suy

Federal Oce or the Environment (FOEN)

CH - 3000 Bern

[email protected]




aNNeX 2: WHaT Is RIsK?Annexes




170

aNNeX 2: WHaT Is RIsK?

Andreas Bachmann and Klaus Peter Rippe, Ethik im Diskurs

Intensive discussions within the project team regarding the conception o risk

have shown that this concept needs to be claried more thoroughly. The main

reason is that the standard approach used in environmental risk assessment to

assess risk as a unction o exposure and hazard must be questioned rom a scien-

tic theoretical point o view as it does not indicate where the aspect o probability,

which is essential or an adequate denition o risk, comes into play. The debate

within the project team on this issue resulted in the ormulation o some prelimi-

nary theses by the two ethicists, Andreas Bachmann and Klaus Peter Rippe, thatmust be provocative to any risk researcher. In the ollowing, we want to present

these theses being ully aware o their provocative and controversial character.

Wht i rik?

Risk as a technical term is characterized as a unction o the extent (or the magni-

tude) and the likelihood (or probability) o harm.44 Both variables – probability and

magnitude o harm – are equally essential. Harm denotes something that should

be evaluated negatively. Thereore, the concept o risk is inherently value laden.

For this reason, determining risks is not a purely scientic issue. At best, empiricalscience can determine the probability o a uture event occurring. It cannot deter-

mine, however, how this act – the act that something will occur with a certain

probability – should be evaluated, i.e., whether it is a risk, that is acceptable or not.

This technical denition o risk must be distinguished rom two nontech-

nical usages. First, in everyday language, the term risk oten only reers to the

probability that harm will occur. The term risk is used in this way, or example,

when we say that there is only a small risk o being killed in a plane crash.

Second, the term risk at times only reers to the possible magnitude o harm.

This is the case, or example, when we say that lung cancer is one o the majorrisks that aect smokers.

It is important to keep in mind that the technical concept o risk always reers

to both, the magnitude and the probability o harm. To give an example: since the

risk o being killed implies a great harm, it can only be considered to be small i

the probability o occurrence o this harm is extremely low.

44 I unction is understood as a product then risk is a product o the probability and the

extent o harm. In other words: risk is an expectation value. The ollowing examplemay illustrate what this means: I in situation a) there is a 1% probability o 10’000

people dying and in situation b) a 50% probability o 200 people dying the risk, i.e.

the expectation value in both cases is 100 dead people. Identical expectation values

should be evaluated identically. This means that i the risk o situation b) is deemed

inacceptable this also applies to the risk o situation a).





What is the relation between risk and hazard? Hazard is usually denedas the intrinsic ability or the potential o something (an individual being, a

substance, an action, an event etc.) to cause harm. Without hazard, there can be

no harm and thus no risk. In this sense, hazard is a necessary condition o risk;

but it is not an aspect or a component o risk. A toxin, or instance, may have

the intrinsic ability to kill people. However, it becomes a risk only i someone

is exposed to it. I this happens, its potential to cause harm may turn into a real

harm; i.e., there is a certain probability that harm will actually occur.

Every hazardous entity is a source o risks. The hazard, i.e., the harm poten-

tial related to this source exists irrespective o whether anybody is exposed to therisk source.45 A risk, however, emerges only when someone is actually exposed

to the risk source. Thereore, exposure is another necessary condition o risk:

without exposure, there is no risk. Exposure in this sense is a digital concept:

either one is exposed to a risk source or one is not exposed. As long as nobody

is exposed to a potentially lethal toxin, or example, there is no risk since there

is no probability o a harm (death in our example) occurring. In this respect,

the likelihood o being exposed is thereore irrelevant. Those who use the term

“likelihood o exposure” usually reer to a certain percentage o a population

that will presumably be exposed to a hazard. This kind o likelihood, however,

45 Hazard designates the harm potential, i.e., the intrinsic ability to cause harm. It is

thereore completely independent o the probability o harm occurring. For example,

i the harm potential o a big avalanche is greater than the harm potential o a small

avalanche, this means that the big avalanche can – has the intrinsic ability to – cause

more harm than the small one. The ormer could, let us say, lead to the complete

destruction o a protection orest whereas the latter could only lead to a destruction

o a small part o this orest. This hazard exists irrespective o any probability o

the respective harm occurring. Another example may help to urther clariy this

point: Toxic substance A and toxic substance B both have the ability to kill. This is their hazard. The probability that death occurs i a person ingests either o these

substances may vary because substance A is more toxic than substance B. Hence,

it may be true that substance A is more likely to kill this person than substance B.

However, this does not aect the hazard, i.e., the intrinsic ability to cause harm.

Rather, it presupposes the determination o the hazard: Even though both substances

can be deadly (hazard), substance A is more likely to kill a person than – the same

quantity o – substance B. I this is plausible, however, are there not two dierent

hazards relating to these substances? The answer is no because, as ar as the intrinsic

ability to kill individual persons is concerned, substance A and substance B are

identical. In colloquial language, the answer may be dierent. It is completely normal

to say that substance A is more hazardous than substance B, that is, when a person

is exposed to these substances, substance A is more likely to kill this person than the

same quantity o substance B. Thus understood, however, hazard is just another wordor risk (not in a technical sense though, but rather in the everyday sense mentioned

above reerring to the probability that a harm will occur). For this reason, it seems

advisable to orego the concept o hazard altogether and to use the notions o harm

potential and risk instead, the latter being dened as a unction (the product) o

probability and magnitude o harm.




172

is not part o the risk itsel. In order to determine a risk, the only relevant actoris the number o individuals (or the percentage o a population) actually being

exposed.46 O course, certain eatures or parameters o exposure may have an

eect on probability. For instance, temporal duration or requency o exposure

(or magnitude o dose) may play a role: the more requent and the longer an

individual or a population is exposed to a risk source, the higher the probability

is o harm occurring. However, this kind o exposure is not equal to the actual

probability. Rather, the connection is inerential insoar as inormation regarding

temporal duration, requency or magnitude o dose can be used to estimate the

likelihood o the occurrence o harm.As shown, neither hazard nor exposure – understood as the bare act o

being exposed to a risk source – is a component o a risk. Yet, in order to deter-

mine a risk, we have to know the hazard (i.e., the intrinsic ability o a risk

source to cause harm) as well as the (approximate) number o individuals or

the percentage o a population actually exposed to this source.47 This knowl-

edge is necessary because without it, the probability and the magnitude o

harm associated with a risk source cannot be determined. However, urther

knowledge is required to determine probability and magnitude o harm. Some

examples – nuclear power plants, avalanches, ecotoxicology – may be useul toclariy this point.

Nur pwr pnt

The hazard o nuclear power plants, that is, their intrinsic ability to cause harm,

is extremely great. Worst case scenarios assume that a core melt accident could

lead to the death o thousands o people. This hazard exists independently o

whether anybody would be harmed by such an accident; that is, it would also

exist i a remotely controlled nuclear power plant was situated in the middle o

a deserted area so that nobody would be killed by a core melt. I this were thecase, nobody would be exposed to this risk source. Consequently, the possible

harm could not turn into a probable harm and there would thus be no risk. In

reality, however, the situation is quite dierent since many atomic power plants

are situated in densely populated areas, which means that countless people are

actually exposed to this risk source. In order to determine the risk or these

46 In many cases, an exact number cannot be ascertained. Thus, the number o

individuals (or the percentage o a population) actually being exposed to a risk sourcemust be estimated. The term “likelihood o exposure” reers to this estimate. I the

exact number (or the exact percentage) cannot be provided, this entails an uncertainty

with regard to the determination o a risk.

47 As mentioned, exposure also comprises aspects such as how much o a substance

someone encounters. This may have an infuence on the probability o a specied

harm occurring.





people, we have to know the probability and the magnitude o harm. Suppose theprobability o a reactor meltdown accident is 1:33,000 years. Suppose, urther-

more, that the number o dead people (including late atalities) in case o such

an accident amounts to 50,000. Given that quantiable risks can be expressed

in terms o expectation values (i.e., the product o the two variables probability

and magnitude o harm), the risk (the expectation value) amounts to 1.5 dead

persons per year.

A major problem is whether, and i so how, the probability and the magnitude

o harm can be calculated. A purely mathematical calculation is almost never

possible. Usually, the two variables o a risk must be estimated whereby the esti-mates are based on experience and/or model calculations. How these variables

are determined depends on the case at hand.

Regarding the risk o a meltdown accident in a nuclear power plant, the

analysis o the probability component is a complex procedure (Birkhoer, 1980;

Weil, 2002).48 It basically consists o recording the possible accident sequences

induced by an initiating event by means o event trees and then calculating

the probabilities o ailure o the dierent saety systems ultimately leading

to a core melt by means o ault tree analyses using Boolean logic. One prob-

lematic aspect o this approach is that its results are aficted with uncertain-ties resulting rom unveriable assumptions and models. Another problem-

atic aspect is that experts do not agree on the concept o probability. Some o

them avor a subjectivistic probability theory according to which probability is

a measure o degree o belie. Others avor an objectivistic probability theory

according to which probability reers to the relative requency o an event

(Birkhoer, 1980; Weil, 2002).

Depending on the direction in which the radioactive radiation released by a

core melt moves, the magnitude o harm (expressed in lives lost) is great(er) or

small(er). What is thus required or the risk assessment o a Maximum CredibleAccident (MCA) – beyond calculating the probability o a core melt accident – is to

determine the probability o the direction in which the radiation would move in

case o an MCA. MCA reers to the worst possible magnitude o harm (“hazard”):

the greatest number o people possibly harmed (i.e., killed) in the short and

long run by a core melt accident and its associated discharge o radioactivity.

Given the state o scientic knowledge, it is clear that this harm, as well as the

probability o its occurrence, can only be roughly estimated, but not precisely

determined (Niehaus, 1984; Weil, 2002).

48 In the ollowing, only the risks o purely technical ailures are taken into consideration.

We are well aware, however, that other aspects such as human ailure or external

actors (or instance, earthquakes or tsunamis, or a combination o both) are also

relevant or the determination o the risk.Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



174

This example highlights that when we are about to perorm a risk assess-ment, we do not only want to know what may or could happen, we want to know

the probability with which it will happen. In our example, we want to know the

probability o a core melt accident occurring and we want to know the probability

o the radioactive radiation released by this accident to harm (kill) the greatest

possible number o people.

This example also shows that risk assessments can be replete with uncer-

tainties regarding the probabilities and the magnitude o harm. These uncer-

tainties limit the accuracy and reliability o risk statements. Hence, they should

be reduced as ar as possible. In many cases, however, it will not be possible tocompletely eliminate them.

avnh

While risks o nuclear power plants are mainly induced by a certain technology,

the risks associated with avalanches may be regarded as primarily induced by

natural processes.49 The ocial scientic terminology (see www.sl.ch) in this

area is muddled as it equates hazard with the probability o an avalanche occur-

ring and risk with potential harm. Both terms are used in a wrong way. Hazard

means the potential o an avalanche to cause harm. The aspect o probability isnot included. And risk is not potential harm, but probable harm. Furthermore,

since risk always contains a harm component, it must not be conounded with

the probability o some event occurring. The use o the risk concept is only

adequate i the event in question must be considered harmul. Avalanches are

only risks thereore, i they threaten individuals (or material assets) that would

be harmed with a certain probability in case they would be triggered.

As ar as the probability o triggering avalanches is concerned, European

avalanche bulletins content themselves with qualitative statements, distin-

guishing between ve levels: low, limited, medium, high and very high. Theselevels are rather vaguely dened. “Low,” or instance, means that due to generally

very stable snow, avalanches triggered by persons are unlikely and conned to

very ew extremely steep slopes. “Very high” means that due to generally unstable

snow, even on gentle slopes, many large spontaneous avalanches are likely to

occur. To reach this degree o precision, a great amount o data and experience is

required. Since avalanches are complex phenomena, even this does not seem to

suce, however, to determine probabilities quantitatively – much less to predict

avalanches with certainty.

49 Even though they can be considerably increased or decreased by human behavior.




As ar as the hazard o avalanches is concerned, the decisive criterion isavalanche size. There are our sizes: slu, small, medium, large. “Small,” or

example, means that the avalanche can bury, injure or kill a person. “Large”

means that also large trucks and trains as well as large buildings and orest

areas can be buried and destroyed.

Knowing the level o probability and hazard, however, is not sucient to

determine the risks associated with avalanches. The aspect o probability

mentioned above solely reers to the likelihood that avalanches are triggered by

spontaneous events or human activity. The probability o harm remains unknown.

The aspect o hazard solely reers to possible harm, not to probable harm. Whatis lacking is the determination o the probabilities related to the dierent levels

o hazard. For that purpose, the component o exposure needs to be analyzed.

Exposure has two aspects. On the one hand, the aspect o being exposed to a risk

source: i someone is not exposed to an avalanche in the sense that he cannot

possibly be harmed or reasons o spatial distance, there is no corresponding

risk. On the other hand, exposure has characteristics such as temporal duration

or geographical location that aect the probability o being harmed: depending

on the route skiers or climbers choose, or instance, the probability o being

harmed (i.e., injured or killed) by an avalanche is greater or lower. Again, giventhe state o knowledge, these risks can only be determined qualitatively. They

cannot be quantied because neither the probability o being injured or killed

by an avalanche nor the number o persons being exposed can be precisely

(numerically) ascertained.

etxigy

In ecotoxicology, risk is commonly interpreted as a unction o exposure and

hazard. This interpretation is not equivalent to the standard denition o risk

being a unction o probability and magnitude o harm since probability is notthe same as exposure and magnitude o harm is not the same as hazard.

As explicated beore, hazard is the intrinsic ability or the potential o

something – a toxic substance in our case – to cause harm. In other words,

hazard reers to the possible harm such a substance may bring about. Risk

assessments, however, do not aim at determining possible harm. Rather, they

are targeted at determining (more or less) probable harm. To reach this goal,

exposure analysis is the next necessary step. Again, the two aspects o expo-

sure must be careully distinguished. First, there is only a risk i someone is





176

exposed to the hazardous risk source. Second, every exposure is characterizedby certain properties (exposure actors), such as, temporal duration, requency,

dose, number o individuals exposed etc. These properties must be known in

order to determine the probability o harm. They are, however, neither equal to

the actual probability nor do they suce as such to calculate this probability.

What is also required is a stochastic deliberation refecting the act that there

is not enough causal knowledge or making deterministic predictions.

An example may serve to illustrate how this might work. A population

o 100 sh is exposed to a certain dose o a toxic substance. Within a certain

period o time, 50% o the sh die. This is the harmul eect caused by the toxicsubstance. From this, a stochastic conclusion regarding risks can be drawn

(provided the sample is large enough), namely, the conclusion that the prob-

ability o dying amounts to 50% (under laboratory conditions) or every sh o

a population exposed to this dose o a toxic substance.50

There is an important dierence between the statement ”dose x o a toxin

will lead to the death o 50% o the members o the exposed population” and‚

“there is, ceteris paribus, a 50% probability or each member o the exposed

population to die o dose x o a toxin.” The rst is a prediction based on a cause-

eect-relationship. It states a certainty as opposed to the second statement which expresses a probability implying an uncertainty whether the uture event

actually will take place. The ormer is a deterministic prediction, and the latter

is a probabilistic prediction. The aim o risk assessments are accurate proba-

bilistic predictions. Deterministic predictions by contrast are not part o such

an assessment. I deterministic predictions are possible, we should not speak o

“risk analysis” or “risk assessment.”

This caveat is important or the ollowing reason. Given that we live in a caus-

ally determined world,51 it would be possible to predict any event with certainty

i we knew all relevant causal actors.52 Under such conditions, there would be norisks. The world, however, is so complex that, given the current state o science,

it is impossible in many cases to make reliable predictions based on our knowl-

edge o cause-eect-relationships. In this situation, we have to content ourselves

with probabilistic predictions, especially with regards to complex systems. In

justiying such predictions, three limiting actors play a crucial role:

50 “Every sh” does not reer to the single sh as an individual, but as a member

o a certain sh species.51 I quantum mechanics is true, physical processes at subatomar scales cannot be

explained deterministically. At this level, thereore, only probabilistic predictions

would be possible.

52 Thus, uncertainty is always due to lack o inormation or knowledge. There is no

uncertainty that a possible increase in knowledge could not eventually eliminate,

provided determinism is true.Olivier Sanvido et al.: Valuating environmental impacts of genetically modified crops © vdf Hochschulverlag 2012



• Time axis: Observation periods are usually rather limited. It is thus dicultto predict (in a probabilistic sense) what may happen in the long-term.

• Amount o data: Sucient data or making reliable probability statements

is oten not available. Yet, without sucient data, the risk level cannot be

determined. Given a sample o 100 persons o whom 50 are smokers and 50

are nonsmokers it is not possible to derive any scientically sound statement

regarding the (absolute) risk o lung cancer or smokers in general.

• Causal actors: Laboratory experiments oten do not – or, given the current

state o knowledge, cannot – detect all relevant causal actors. And even i

all relevant causal actors are known, their interaction may be too complexto allow deterministic predictions.

It may be argued that the toxic eects o a substance can be determined

more reliably in the laboratory than in the reality o a complex system because

the causal actors can be controlled more accurately within the laboratory.

However, what we want to know in the end are the eects o the toxin under real

environmental conditions, that is, under conditions in which a whole array o

other actors may infuence the eects o the toxin. These eects are too complex

to be deterministically predicted, or instance, by extrapolating them out o theavailable limited knowledge. At this point, we have to resort to risk analysis, ulti-

mately aiming at scientically sound probabilistic predictions. However, here we

ace yet another problem. In the laboratory, the hazardousness o a toxin can be

ascertained under specic and controlled conditions. In eld trials, the results

achieved in the laboratory can be tested, although only under restricted condi-

tions and or a limited amount o time. At most, eld trials thus enable a more

complex causal analysis. However, this is not enough to make suciently reli-

able risk statements as eld trials cannot generate the amount o data necessary

to make sound probabilistic predictions about eects in complex environmentalsystems. Presumably, the only way to achieve this is to monitor real systems,

whereby “monitoring” means collecting the data required to make well-ounded

probabilistic predictions.

What does this mean with regards to the risk assessment o GMO? It is impor-

tant to note that the approach sketched thus ar is only tenable in this respect i

certain assumptions concerning GMO in general, GM crops in particular prove

to be plausible.53 The basic question is: I a crop such as maize is changed by

inserting a oreign gene into its genome, is the GM maize just the old maize plus


53 The ollowing remarks are owed to critical comments o Alan Raybould on the rst

drat o this chapter.




178

the new trait or should it be regarded as a new kind o maize? How one answersthis question has considerable repercussions on environmental risk assessment.

The approach outlined in this chapter builds on the assumption that the trans-

genic maize is a new kind o maize. It cannot be maintained i this maize is just

the old kind with a new trait. What is meant by this dierence and what are the

implications or risk assessment?

Suppose a oreign gene expressing a toxic protein is inserted into a plant.

I this GM plant B is the conventional plant A plus this new trait, that would

entail the ollowing:

• As ar as risks are concerned there is no dierence between plant A andplant B regarding the already known risks o plant A. Hence, i the risks o

plant A are suciently known and deemed to be acceptable, there is no need

or risk tests on plant B.

• Risk tests are only required with regards to the new gene expressing a toxic

protein. This can be done in the laboratory, or instance, in order to examine

the eects o the toxin on non-target organisms. Given that it is clear what is

meant by harm in this context and what needs protecting rom harm (targets

or protection), this allows testing whether the GM crop in question is envi-

ronmentally sae. It can then be argued that it is sae i the likelihood o unac-ceptable harm occurring is minimal. This can be ascertained by exposing the

non-target organisms to a dose o this toxin that is (much) higher than these

organisms will encounter under real conditions. I the toxin even in this case

does not cause any harm that is regarded as inacceptable the GM plant as

a whole can be considered sae (enough). Field trials are thereore not nec-

essary as regulatory authorities may permit commercial-scale cultivation o

the crop without requiring more data.

There are two main objections to this view o the relation between the nature o a GM crop and risk assessment:

• This approach is not a risk model aimed at the determination o risks as the

product o probability and harm. It is a causal model aiming at the detection

o cause-eect-relations. Depending on the dose a toxin has certain impacts

on target and non-target organisms deemed to be positive or negative (or

neutral). These relations between cause and eect can be ascertained in the

laboratory. To ensure that the GM plant is sae (enough), the targets or protec-

tion are exposed to a dose considerably higher than they will be exposed to




in the eld. I no negative eect is observed, the conclusion is drawn that therisk – the probability o unacceptable harm – is minimal. Only at this point, risk

seems to play a role. However, what risk means is no more than the general

epistemic proviso that we are not inallible and that it cannot be absolutely

ruled out, thereore, we may have overlooked something.54 This, however, is

not what is meant by risk assessment, at least not i risk is dened in terms o

probability and (magnitude o) harm. To be sure, this does not show that the

concept o GM crops underlying this approach is inadequate. It only shows

that this concept is not satisactorily connected to risk assessment, understood

as a genuinely probabilistic approach, but to a causal model which seeks topredict the eects o cultivating (certain) GM crops (or instance, on non-

target organisms). Those in avor o this approach should thereore orego the

concept o risk and use a deterministic, not a probabilistic language.

• A second objection is that conceiving o a GM plant as the conventionally

bred template plus trait X resulting rom genetic modication is inadequate.

A GM plant is not the sum o the original plant’s traits and trait X. It cannot

be ruled out that this trait substantially changes the character o the GM

crop in question due to eects on the genetic level (multiple insertions, dele-

tion, ller DNA) and/or to epigenetic mechanisms (gene silencing, positionaleects, and pleiotropic eects). The question is what this means or risk

assessment. The main point is that the GM crop may not just show unin-

tended but unexpected traits. For this reason, lab tests trying to determine

the risks associated with this crop are o limited signicance. They must be

supplemented by eld trials and by post-market monitoring. At this point,

it is important to emphasize that environmental risk assessment should not

be conceived o as a pragmatic tool or decision-making under conditions

o limited time and money. Rather, its purpose is to generate the scientic

risk knowledge necessary to decide whether the risks associated with GMplants are acceptable or not. Whatever is necessary to ulll this task must be

done – regardless o how long it takes and how much it costs. O course, eld

trials as well as post-market monitoring should be based as ar as possible on

clear risk hypotheses. It does not make sense to collect large amounts o data

i these data do not help improve risk knowledge. Yet, as long as we do not

understand all the environmental actors that could infuence the GM plant

and the interplay between these actors and the structure o the plant eld

trials are necessary. This is so even i it were true that in many or even most


54 In German, this is expressed by the phrase: “mit an Sicherheit grenzender

Wahrscheinlichkeit.”




180

cases, the environment mitigates rather than exacerbates potentially harmuleects. One central aim o risk assessment consists precisely o determining

the probability o those – maybe not very common – cases in which harmul

eects are exacerbated and not mitigated by the environment.

Above, an argument was mentioned according to which a GM crop is sae

(enough) i the probability o unacceptable harm occurring is minimal. However,

whether this argument is valid cannot be decided by empirical science since it

concerns a normative question that lies beyond the reach o science: the question

o acceptability. Thus, even i it were true that the risks o GM crops – includingthe risks concerning worst cases – can be ascertained in the laboratory, this

would not entail that the commercial release o a GM crop is admissible i the

likelihood o unacceptable harm is minimal.

As ar as the normative criterion or risk acceptability is concerned, there are

two prominent approaches that reject the idea that activities or technologies that

have the ability to cause inacceptable harm are acceptable, all the same provided

the probability o this harm occurring is minimal.

• The rst approach is based on the maximin or minimax criterion. According

to this criterion, unacceptable harm means harm that must be prevented atall costs. This implies that the likelihood must be zero, whenever possible.

And this, in turn, entails that the respective activity – in our case, the

commercial release o a GM crop – must be prohibited given that this is the

only way to lower the likelihood o the harm occurring to zero. However,

almost zero is not enough.

• The second approach is based on a threshold criterion according to which

risk as the product o probability and magnitude o harm must not exceed

a predened threshold value. This implies that even i the probability o a

harm occurring is extremely low, this per se does not justiy the risk – orrather the activity associated with this risk – because the harm may be so

severe that the risk is still above the threshold.

Regarding some quite well-known GM plants, or example, Bt-maize, one

could argue that by now we do have enough data and knowledge to conclude

that the risks associated with them are so low as to be negligible and there-

ore acceptable. This conclusion can only be valid, however, i the probability

regarding the worst case, i.e., the greatest possible harm, can be determined with




sucient precision. “Sucient precision” does not require quantitative speci-cation. Qualitative determination may also be acceptable. For this to be the case,

however, one condition must be met: it must be possible to categorize the prob-

ability in a way that allows comparing it to the probabilities o other risks, or

instance, the probability o a strong earthquake – deemed to be low or very low.

Otherwise, it remains too vague, making it impossible to decide whether the

corresponding risk is acceptable or not.55

The same problem arises with regard to unintended eects o GM plants.

In this case, the analysis o worst case scenarios should aim at estimating the

probability with which, or example, pleiotropic and epigenetic eects triggera dynamic that would result in an irrevocable destabilization or even complete

collapse o ecosystem equilibrium. The main problem here is that on the one

hand, it is inadmissible to extrapolate rom the laboratory to the reality o complex

ecological or agricultural systems whose causal actors are scarcely known.56 On

the other hand, it is not possible in this case to derive probabilities o specic

events that have never occurred rom known eects. It is an open question, then,

as to how the probability o a maximum possible harm occurring can be deter-

mined, i it can be determined at all.57 Since this is an important point, it is neces-

sary to take a closer look at the conventional deterministic approach o assessingtoxic risks o GM plants. In our context, the main point to discuss is whether this

approach is able to give due consideration to the aspect o probability.


55 It is oten argued that the risks o a GM crop are acceptable i they are not (signi-

cantly) greater than the risks o its conventional counterpart. This normative argu-

ment is untenable or the ollowing reason. Logically, the act that B (a GM crop) is

not worse than A (the respective non-GM crop) does not make A good or acceptable.

It does not imply anything about the value o A. Even i A were very bad (harmul)

and utterly inacceptable, it would still be true that B is not worse than A. Obviously,

this does not make B acceptable. So this kind o comparison is empty with regard to

the question o the acceptability o B. In other words, it only works i A is acceptable.

However, how do we know that? The point is: Acceptability is a normative concept

that cannot be determined just by looking at legal regulations. There are many legal

regulations o risks that are accepted; and yet the risks regulated are inacceptable.

Acceptability must be logically distinguished rom acceptance. In this sense, it is a

normative or ethical concept. Ultimately, the dierent ethical risk theories provide the

standards or acceptable risks. Thereore, the baseline or acceptability cannot be an

existing legal standard regulating conventional agricultural products. Acceptability

must reer to normative standards that are independent o legal regulations even

though they may happen to be enshrined in the law.

This reers to the worst case scenario outlined above: the collapse o eco systems

resulting rom a dynamic due to pleiotropic or epigenetic eects. The point is simply this: whatever we may be able to test in a lab, it will not suce to allow deriving the

probability o this worst-case occurring. In other words, it will not suce to justiy the

conclusion that the likelihood o the worst-case happening is so low as to be negli-

gible and thus acceptable.

We will come back to this point at the end o this section.

56

57




182

A common test o acute toxicity is the LD50 test. In this test, an animalpopulation is exposed to a certain dose o a toxin leading to the death o 50% o

the individuals o this population. The toxic substance is the source o hazard.

Hazard reers to the ability or potential o this substance to bring about a certain

harm – death in the case o LD50. Exposure denotes the act that a population

has been exposed to a hazard over a certain period o time.

LD50 tests are o a deterministic kind. Once the test has established the dose

leading to the death o 50% o the exposed population, it is possible to make

predictions based on this result. These predictions are deterministic – as opposed

to the probabilistic predictions typical o risk assessments: they allow predictinga uture event, namely, that ceteris paribus 50% o a population exposed to a

certain dose o a toxin will die thereo, with certainty because they are based on

the knowledge o cause-eect-relations.

This deterministic approach underlies ecotoxicological tests. Regarding bio-

diversity, or example, the hazard o a stressor on the environment is charac-

terized in the laboratory or under eld conditions using a number o surrogate

species that are thought to be indicative or biodiversity. The intention o the tests

is to determine the (detrimental) eects o the stressor on the species chosen at

given levels o exposure. To this point, this is a purely deterministic proceduredetermining causalities; even though it is commonly called “risk assessment.”

The main question then is how the results o these tests can be transerred

to agricultural systems. Basically, there are two options: either it is justied to

directly transer them or a direct transer is not justied and some intermediate

steps are required.

To justiy a direct transer, the ollowing ontological statement would have to

be true: the same causal actors determine the eects o the stressor – irrespective

o whether we are in the laboratory or in the open system. This implies that there

are no additional – maybe as yet unknown – environmental actors that could infu-

ence the (harmul) eects o the stressor under real conditions. I that were true,

we could, given that we are able to exactly determine the exposure actors, predict

the harm a GM plant would cause once it would be commercially cultivated. There

would be no need to draw upon probabilities below one and above zero as the

probability actor would be one or zero. Probability one means it is 100% certain

that x will occur. Probability zero means it is 100% certain that x will not occur.

The aspect o uncertainty regarding the occurrence o harm would not exist, and

thereore there would be no risks since uncertainty is essential or any risk.




In this context, the proponents o the deterministic approach argue thatenvironmental risk assessment is basically an extrapolation o the results o the

hazard characterization and the exposure assessment perormed. Extrapola-

tion, however, is an ambiguous notion. For instance, i a vehicle covers a distance

o 200 meters in 12 seconds and 1000 meters in 60 seconds, one can extrapolate

that in 90 seconds it would cover 1500 meters. Understood as a prediction, this

is true i the speed o this vehicle is not infuenced (i.e., changed) by any addi-

tional internal or external causal actors. In that case, extrapolation enables a

kind o deterministic prediction. This, however, presupposes that we know all the

relevant causal actors. Otherwise, the extrapolation is aficted with uncertainty and we have to resort to a probabilistic analysis. In that case, extrapolation only

allows probabilistic predictions.

Applied to the assessment o GM crops, this means that an extrapolation

in the deterministic sense is only possible i 1) all causal actors relevant in

view o the possible harm caused by these crops and active under laboratory as

well as real conditions are known; and i 2) these causal actors are identical. I

conditions 1 and 2 are met, we should avoid speaking o risks and risk assess-

ment since risks essentially imply uncertainty. I conditions 1 and 2 are not met,

we should avoid speaking o a “deterministic approach” in the context o riskassessment and use the term “probabilistic approach” instead.

Given the current state o scientic knowledge, it is more plausible to

understand extrapolation in a probabilistic sense. The main reason is that, as

mentioned beore, we do not know all causal actors possibly relevant under

the environmental conditions o open agricultural systems and thereore

cannot transer laboratory results or results o eld trials directly to these

systems. In other words: it cannot be ruled out that the eects o GM crops

within these systems are shaped by infuencing actors we do not know. That is

why or the time being, sae deterministic predictions regarding these eectsare out o reach.58

The deterministic approach does not seem to be able to take the aspect

o probability adequately into account. This is no surprise as the approach is

geared to causal analysis, allowing reliable deterministic predictions based on

cause-eect relationships. In this approach, there is no room or uncertainties


58 Accurate probabilistic predictions concerning the risks o GM crops also require causal

knowledge, without, however, being reducible to this knowledge. It is important to notethat causal knowledge is not always necessary. The lung cancer risk o smokers, or

instance, can be probabilistically predicted on the basis o statistical knowledge alone.

Statistical knowledge does not presuppose causal knowledge: even i one has no clue

regarding the causal connection between smoking and lung cancer one can make an

accurate probabilistic prediction.




184

and thus or probabilities except probability = 1 (100%) or 0 (0%). The proba-bilistic approach seems better suited to handle the uncertainties we are aced

with due to a lack o causal knowledge.59

The main diculties concerning the calculation or estimation o prob-

abilities have already been mentioned. In conclusion, it may be helpul to list

them once again:

• In order to reach sound stochastic conclusions about possible harmul

eects in complex environmental systems, a large amount o data is required.

A good example to illustrate this point is meteorology. Weather orecasts

are probabilistic predictions. They are probabilistic because the weathersystem is too complex and too dynamic to allow deterministic predictions,

at least given the current state o meteorological knowledge. In order to

make probabilistically accurate orecasts, two things are needed: rst large

amounts o (relevant) weather data; and second, probability calculations.

Good weather orecasts are thus based on enough data-based inormation

on one hand, and on the other hand, on meteorological and mathematical

knowledge backed by sucient computing capacity.60

What would adhering to a probabilistic risk approach mean with regard

to environmental risk assessment in the area o GM crops? Given that agri-cultural systems are as complex and dynamic as weather systems, and given

that a GM plant is a new kind o plant and not just the old plant plus the

new trait, the ollowing conclusion can be drawn: a ull-fedged scientic

risk analysis necessitates a) a large amount o (relevant) data that cannot

be accumulated by a laboratory experiment or even a eld trial restricted

to a ew sampling sites, but only by monitoring real large-scale agricultural

systems where GM crops are cultivated; and b) ecological and mathemat-

ical knowledge backed up by sucient computing capacity.

59 It is important to distinguish between two dierent risk-related uncertainties.

1) The uncertainty regarding a uture event happening; 2) Uncertainties regarding

the determination o a risk. To give an example, when a person plays Russian

Roulette, there is a probability o 1 in 6 that this person will be killed by a pistol

bullet. Thus, regarding the determination o the risk, there is no uncertainty since

the probability can be mathematically calculated and the possible extent o harm

is unequivocally clear (provided that death is harm). This does not detract rom the

act, however, that it remains uncertain whether this person will die. In this respect,

only a probabilistic prediction is possible since we do not have the causal knowl-

edge required or a (justied true) deterministic prediction.

60 This does not mean that scientic weather orecasts are necessarily more precise

than orecasts made by “weather prophets” who base their predictions on subjective

experience and historical knowledge. Yet, it is beyond reasonable doubt that scientic

orecasts are incomparably more reliable and trustworthy than orecasts made by

lay people. The reason is that they are grounded in a much better understanding o

the weather system, understood as “deterministic chaos.”




• In view o unintended (and unexpected61

) eects o GM plants, the proba-bilistic approach is aced with a dierent problem when it comes to esti-

mating the probability o the worst case scenario occurring. The problem

is, we cannot directly iner the probability o this worst case on the basis o

known eects.

Thus ar, we have outlined the problems to be overcome when trying to calcu-

late probabilities with regard to the risks associated with GM crops. It has not

been made suciently clear, however, which methods may help to solve these

problems. The remainder o this chapter will be devoted to this topic.The question to be answered is: how can statements such as, “The prob-

ability o harming non-target organisms by commercially cultivating Bt-maize

is (very) high/ (very) low (or alternatively, is x%)” or “The probability o a

complete collapse o ecosystem equilibrium (harm) caused by pleiotropic and

epigenetic eects is (very) high/ (very) low (or, alternatively, is x%)” be justi-

ed? What methods can be used in order to calculate or at least reasonably

estimate these probabilities?

MethodsThe standard ormula o the classical conception o probability is

.

That is, probability p regarding event E is equal to the ratio o the number

o avorable cases N(E) to that o all the cases possible N.

This ormula allows calculating probabilities quantitatively in two kinds

o cases:

• In cases where physical conditions are irrelevant and p can be calculatedpurely mathematically (lottery 62)


61 Insoar as unexpected (and maybe even unexpectable) eects result rom the

dynamic and complex interplay between the GM plant and the open agricultural

system, they can only be detected by monitoring this system. O course, they must

rst maniest themselves beore one can start thinking about the risks associated

with them.

62

The probability o getting the correct six numbers in lottery 6/49 is 1/14 million. Thisprobability is calculated by means o mathematics alone. This probability means

that getting six right is extremely unlikely – irrespective o the number o players

taking part in the lottery. However, the more players there are, the greater is the

probability that one o them will win. The same event whose occurrence is extremely

unlikely or each player becomes very likely (weak law o large numbers).

p(E) = N(E)

N




186

• In cases where the calculation o p can be based on physical symmetries (aircoin, air die, roulette etc.63).

I there is no physical symmetry, as in the case o GM crops, the method

o the classical conception cannot be applied. At this point, the method o the

requency account comes into play. This method is also based on the ormula o

the classical conception since the requency conception is, structurally speaking,

the a posteriori version o the classical conception: ”it gives equal weight to each

member o a set o events, simply counting how many o them are ‘avorable’ as

a proportion o the total. The crucial dierence, however, is that where the clas-sical interpretation counted all possible outcomes o a given experiment, nite

requentism counts actual outcomes” (Hajek, 2009).

I, or instance, an asymmetrical die is biased in avor o number our, the

appropriate way to nd out the probability o this number occurring is to conduct

a suciently long series o throws in order to ascertain p via relative requency.

In other words: in empirical cases like these, what is needed to ascertain p are

the data required to determine relative requency in a reliable manner since this

requency is (approximately) equal to p.64 Suppose, or instance, a parachutist

worries about the saety o his two parachutes. He wants to know the probability that neither o them will open. In order to calculate this probability, we have to

know the probability o each parachute not to open. This is something we can

know only a posteriori, that is, based on experiences made with this product.

Suppose it is known that the parachute opens in 999 o 1000 cases. Thus, p(E

not open) would be 1/1000. I this applies to both parachutes, the probability we

are looking or would be 1/1000 x 1/1000 = 1/1,000,000 (using the rule o multi-

plication). Statistically speaking, this means that in one in a million cases, both

parachutes remain closed.

How do we nd out that the parachutes in question open in 999 o 1000cases? The answer given is by experience. In this context “experience” has

dierent aspects. 1) Parachutes are a ully developed technology. Thus, i

someone constructs a new parachute, it is possible to derive some conclu-

sions regarding the probability o it not unctioning based on the accumulated

63 Physical symmetry is never completely perect and hence always a kind o idealization.

Yet it makes sense to suppose its existence in these cases since it allows calculating p

accurately enough or practical purposes. This also applies in the case o the air coin

even though the result p=1/2 does not take into account the possibility o the coin

landing “edge.”

64 In case p is known, it is very likely according to the law o large numbers that the

proportion o avourable events, (i.e., the relative requency), will be approximately

equal to p.




knowledge o parachute technology. 2) Tests perormed beore market launch 3).Monitoring o the product ater market launch. Since the technology is already

ully developed, 1) and 2) are presumably sucient to come to a plausible esti-

mate o p (provided the new parachute is not revolutionary). However, the exact

probability can only be ascertained by collecting enough data ater market

launch to produce reliable statistics based on relative requency.

Basically, this approach is applicable in a wide range o cases, especially

cases belonging to the realm o the natural sciences. Take meteorology once

again as an example. What is meant by a orecast statement such as, “There is

a 30% probability o rain tomorrow?” It means, there is rain on 30% o all thedays characterized by the same (or similar) weather conditions as tomorrow.

“Tomorrow” here does not reer to a single event, but to an instance o a type.

In order to be correct, the orecast presupposes a quantity o data large enough

to acilitate reliable statistical statements concerning the probability o rain

under conditions dened by variables such as air pressure, cloud types, and

temperature.

Regarding natural events, the problem o determining the probability o

single cases is oten not as urgent as it may seem. O course, we want to know

how the weather will be like tomorrow. Thus, we reer to a single case: a uniqueuture event (there is only one tomorrow). It is true that the method o rela-

tive requency is unable to assign probabilities to single cases. However, this

does not imply that there is always a relevant dierence between statistical

probabilities and single case probabilities. Regarding the weather orecast, the

dierence between tomorrow as a single case, a token, and tomorrow as an

instance o a type may be negligible since the relevant variables may enable

quite specic probabilistic predictions so that single case and instance o a type

practically coincide.

This may be dierent when we consider, or instance, probabilities pertainingto the lie o a particular individual. Consider, or example, customer proles

used by insurances to calculate the lie risk o their clients. According to the

requency conception, insurances should take into account all variables that may

have an impact on lie expectancy such as age, sex, and health. The more vari-

ables being used, the more individualized the corresponding probability state-

ment. Nonetheless, the probability is only an approximation to an individual case

since its calculation is based on the principle that the narrowest (most speci-

ed) reerence class or which reliable statistics is available should be chosen.





188

Again, this does not necessarily mean that there are practically relevant dier-ences between this probability and the probability o the single case. Intuitively,

however, there may be such dierences. Thereore, the question arises o how

probabilities pertaining to individual persons can be determined.

Imagine the ollowing case (Gillies, 2000): Mr. Miller, 60 years old, has been

smoking three packets o cigarettes per day or thirty years. What is the prob-

ability that he will live until the age o 70? Suppose there is a reliable statistics

or these kinds o cases. The probability o Mr. Miller living until the age o 70 is

the requency o those in this class who have lived until the age o 70. It seems

reasonable, however, to adjust this probability i we learn that Mr. Miller comesrom a very large amily who all smoke three packets o cigarettes per day, but

none o whom died beore the age o 70. The question is, how much would the

probability change i there were no statistical data available concerning indi-

viduals who belong to such unusual amilies? In this case, we seem to be orced

to resort to a rather subjective and vague estimate.65

However, the probability regarding adverse eects o (specic) GM crops on

biodiversity is primarily a statistical, not a single case probability. From a meth-

odological viewpoint, this means that this probability should be determined by

means o the requentist approach. This approach is based on experience. Inthe case o GM crops, experience comprises the same aspects as in the para-

chute example: 1) Conclusions regarding p that are based on the knowledge

we already have. 2) Tests beore commercialization o GM crops (in the labora-

tory; eld trials); 3) Monitoring ater market launch. The main questions to be

answered are: what kind o and how much inormation do we need to derive

statistically valid probabilistic conclusions? And, how do we get this inorma-

tion? Some o the diculties in answering these questions have already been

mentioned. In light o the methodological deliberations just made, one di-

culty regarding the quality o the data required can be ormulated more clearly.The requentist method is a quantitative method. It is based on the idea that

probability can and must be determined quantitatively. This does not neces-

sarily imply that every probability must be expressed by an exact number.

It may also be the case that we can only speciy a particular range. Maybe

the probability o an event is above 1% and below 10%. And within that band,

65 In some cases, the best method to use in order to estimate singular probabilities may

be Bayes’ rule. This rule allows adjusting subjective probabilities in a controlled way inthe light o new experience. However, in most cases, the data resulting rom the new

experience can also be interpreted objectively (i.e., in a requentist manner) by orming

appropriate reerence classes. Bayes’ rule is to be avored only in cases where we have

good reasons to assume that there is a practically relevant dierence between type-

and token-probability.




it is impossible to assign probabilities. However, what about qualitative prob-ability statements? One might think that according to requentism, such state-

ments cannot be scientically sound and are thereore inadmissible. However,

we could also argue that qualitative characterizations o probabilities such as

“low” or “high” can be acceptable, provided that they are not completely arbi-

trary. However, this only makes sense i they are implicitly connected in some

way to numbers. On the one hand, we have here a conventional element: to call

a probability o 40% low is not plausible given the way we use the word “low”

in this context. On the other hand, qualitative characterizations may reer to

an approximate range within which they cannot be exactly localized. I noteven such a range can be determined, possible qualitative probability state-

ments remain vague. They are then expressions o subjective belies that are

not intersubjectively justiable.

Given a liberal understanding o the requentist method, it may, in the

sense just indicated, be admissible to estimate probabilities even i this

only allows qualitative characterizations. One should be aware, however,

that this becomes problematic i one wants to know the probability o the

greatest possible harm occurring (worst case). Take nuclear power plants as

an example. In this case, the probability o the worst case cannot be derivedrom experience in the sense discussed above. There has hardly been any

Maximum Credible Accidents (MCA). So it is not possible to produce a reli-

able statistics in the sense o relative requency. There is, however, another

option, namely, probabilities can be determined sequentially. As pointed out

beore, this is done by a complex component analysis. The probabilities o a

malunction o the dierent saety systems are calculated and then multiplied

(Weil, 2002). This presupposes that these probabilities can be determined at

least approximately. And this, in turn, presupposes enough experience with

the specic technologies being used. I such an analytical procedure is notapplicable due to lack o experience regarding these specic technologies,

then a worst case probability cannot be derived rom probabilities associated

with known eects. In this case, the only remaining option is to base probabil-

ities on model assumptions or scenarios. This may still allow an approximate

preliminary calculation o the overall probability. The more assumptions that

have to be made, however, the greater is the uncertainty concerning this prob-

ability, until a point is reached where probability statements become so vague as

to be virtually empty and useless.





The question regarding GM crops is whether there is any way in which plau-sible probabilities o worst-case scenarios can be determined. Since in scenarios

o this kind, GM crops as complex biological organisms are situated in complex

ecosystems whose behavior in many respects is scarcely known, a component

analysis is not possible. This means that these probabilities cannot be derived

(extrapolated) rom known eects and their probabilities. The only option seems

to be to operate with scenarios that are based on knowledge and on assumptions

that have not been scientically veried. Whether this allows determining the

probabilities o worst cases with the precision minimally required is a question

that needs urther consideration.

190




The debate on the possible impact o genetically modifed (GM) crops on biodiversity shows that so ar

there is no consensus on generally accepted assessment criteria or environmental harm. This debate

stems primarily not rom a shortage o data, but rather rom the absence o criteria or assessing the

eects o GM plants on biodiversity. Since there are no exact assessment criteria, regulatory decision-

making processes are oten not transparent and can be difcult to understand. This increases the

danger that decisions on environmental risks rom GM plants may appear arbitrary.

The VERDI Project (Valuating environmental impacts o genetically modifed crops – ecological and

ethical criteria or regulatory decision-making) is a interdisciplinary collaboration between biosaety

experts and risk ethicists. Its aim is to develop recommendations or decision makers and regulatory

authorities, thus helping to improve the regulation o GM plants. The results show that both the

unambiguous description o protection goals and the establishment o a basis o comparison are

two essential criteria when defning harm.

The book presents a proposal how criteria or the evaluation o GM crops could be developed. The

b k i di d ll h i l d i h d b b f d i k i i i

3464 valuating environmental impacts of genetically modified crops oa

Documents