Paula2021 Article IntegratingAdverseEffectAnalys

BioControl (2021) 66:113–139
https://doi.org/10.1007/s10526-020-10053-8 (0123456789().,-volV)
( 01234567
89().,-volV)
REVIEW
Integrating adverse effect analysis into environmental risk

assessment for exotic generalist arthropod biological control
agents: a three-tiered framework
Débora P. Paula . David A. Andow . Barbara I. P. Barratt . Robert S. Pfannenstiel .
Philippa J. Gerard . Jacqui H. Todd . Tania Zaviezo . Maria G. Luna .
Claudia V. Cédola . Antoon J. M. Loomans . Andy G. Howe . Michael D. Day .
Clark Ehlers . Chris Green . Salvatore Arpaia . Eizi Yano . Gabor L. Lövei .
Norihide Hinomoto . Eliana M. G. Fontes . Carmen S. S. Pires . Pedro H. B. Togni .
James R. Nechols . Micky D. Eubanks . Joop C. van Lenteren
Received: 1 May 2020 / Accepted: 23 September 2020 / Published online: 11 November 2020
International Organization for Biological Control (IOBC) 2020, corrected publication 2021
Abstract Environmental risk assessments (ERAs) biological control. To safely consider exotic GAB-
are required before utilizing exotic arthropods for CAs, an ERA must include methods for the analysis of
biological control (BC). Present ERAs focus on potential effects. A panel of 47 experts from 14
exposure analysis (host/prey range) and have resulted countries discussed, in six online forums over
in approval of many specialist exotic biological 12 months, scientific criteria for an ERA for exotic
control agents (BCA). In comparison to specialists, GABCAs. Using four case studies, a three-tiered ERA
generalist arthropod BCAs (GABCAs) have been comprising Scoping, Screening and Definitive Assess-
considered inherently risky and less used in classical ments was developed. The ERA is primarily based on
expert consultation, with decision processes in each
tier that lead to the approval of the petition or the
Débora. P. Paula and David A. Andow contributed equally to subsequent tier. In the Scoping Assessment, likelihood
the study.
of establishment (for augmentative BC), and potential
Handling Editor: Eric Wajnberg effect(s) are qualitatively assessed. If risks are iden-
tified, the Screening Assessment is conducted, in
Electronic supplementary material The online version of which 19 categories of effects (adverse and beneficial)
this article (https://doi.org/10.1007/s10526-020-10053-8) con- are quantified. If a risk exceeds the proposed risk
tains supplementary material, which is available to authorized
users. threshold in any of these categories, the analysis
D. P. Paula (&) E. M. G. Fontes C. S. S. Pires B. I. P. Barratt J. H. Todd

Department of Biological Control, Embrapa Genetic Better Border Biosecurity, Wellington, New Zealand
Resources and Biotechnology, Brası́lia,
DF 70770-901, Brazil R. S. Pfannenstiel
e-mail: debora.pires@embrapa.br Pests, Pathogens and Biocontrol Permitting, Plant Health
Programs, USDA APHIS PPQ, 4700 River Road, Unit
D. A. Andow 133, Riverdale, MD 20737, USA
Department of Entomology, University of Minnesota, 219
Hodson Hall, St. Paul, MN 55108, USA P. J. Gerard
AgResearch, Ruakura Research Centre,
B. I. P. Barratt PB 3123 Hamilton, New Zealand
AgResearch, Invermay Research Centre,
PB 50034, Mosgiel, New Zealand
123
114 D. P. Paula et al.
moves to the Definitive Assessment to identify area to attain short-term control of a pest. Conserva-
potential non-target species in the respective cate- tion BC has a large unrealized potential throughout the
gory(ies). When at least one potential non-target world (Wyckhuys et al. 2013) and does not require an
species is at significant risk, long-term and indirect environmental risk assessment (ERA). Therefore,
ecosystem risks must be quantified with actual data or because classical and augmentative BC (using exotic
the petition for release can be dismissed or withdrawn. BCAs) require an ERA due to their potential for
The proposed ERA should contribute to the develop- causing environmental harm, they are only justified
ment of safe pathways for the use of low risk when the local natural enemies do not provide
GABCAs. sufficient control.
Due to the recognized non-target effects caused by
Keywords Augmentative Biocontrol Biosafety the historical releases of exotic generalist arthropod
Classical Invertebrates Non-target species biological control agents (GABCAs), arthropod BC
has been largely restricted since the 1950s to the use of
specialist natural enemies because they have a narrow
host range (e.g., Nechols et al. 1992; van Lenteren
Introduction et al. 2020). More recently, heightened concerns over
non-target effects have resulted in an even greater
There is a worldwide demand to reduce pesticide use concentration on specialist natural enemies and partly
in crop production (e.g., van Lenteren et al. 2018) in explains the lower number of introductions worldwide
which biological control (BC) has been a key compo- since the 1990s (Cock et al. 2016). Even though
nent through conservation, classical and augmentation generalist and specialist BCAs have resulted in many
BC. Conservation BC implements practices that outstanding successes (Cock et al. 2016), specialists
enhance pre-existing natural enemies. Classical BC continue to be favored today. Thus, expanding the
introduces an exotic biological control agent (BCA) scope of biological control to enable the consideration
into a new environment with the expectation that it of GABCAs could increase the value of biological
will establish and provide long term control of an control worldwide. For example, several pests have no
exotic pest, while augmentative BC programs release a suitable specialist natural enemies, or their natural
BCA (native, naturalized or exotic) into a localized
J. H. Todd A. G. Howe
The New Zealand Institute for Plant and Food Research Forest Industries Research Centre, University of the
Limited, Private Bag 92169, Auckland 1142, New Sunshine Coast, GPO Box 267, Brisbane,
Zealand QLD 4001, Australia
T. Zaviezo M. D. Day
Facultad Agronomı́a e Ing. Forestal, Pontificia Department of Agriculture and Fisheries,
Universidad Católica de Chile, Santiago, Chile GPO Box 267, Brisbane, QLD 4001, Australia
M. G. Luna C. V. Cédola C. Ehlers
CEPAVE (CONICET - UNLP), Boulevard 120 entre 60 y Environmental Protection Authority, Level 10, 215
64, 1900 La Plata, Argentina Lambton Quay, Wellington 6011, New Zealand
A. J. M. Loomans C. Green
National Plant Protection Organization (NPPO), Department of Conservation,
Netherlands Food and Consumer Product Safety Private Bag 68908, Newton, Auckland 1145, New
Authority (NVWA), Geertjesweg 15, Zealand
6706 EA Wageningen, The Netherlands
S. Arpaia
A. G. Howe Division Bioenergy, Biorefinery and Green Chemistry,
Department of Geosciences and Natural Resource ENEA Research Centre Trisaia, S.S. 106 Jonica km 419.5,
Management, University of Copenhagen, Rolighedsvej 75026 Rotondella, MT, Italy
23, 1958 Frederiksberg C, Denmark
123
Integrating adverse effect analysis into environmental risk assessment 115
enemies are difficult to rear or exert insufficient breadth, habitat use, and their interactions with other
biological control (Nechols et al. 1992). In contrast to species, yet diet breadth is poorly known for many of
classical BC introductions, there has been a marked them, especially predatory arthropod species in their
shift towards considering GABCAs for augmentative natural habitats. For example, unexpected oligophagy
BC in greenhouse production (e.g., van Lenteren of some generalist predators has been demonstrated by
2012). Nevertheless, the use of exotic GABCAs has molecular gut content analysis (e.g., Paula et al. 2016).
been handicapped by a few problematic releases Initial steps toward a full spectrum ERA have focused
(Cock et al. 2016), substantial data gaps on existing on methods to better predict host/prey range of
and potential GABCAs, and the lack of suitable and generalists. These include semi-quantitative ranking
acceptable ERA methodology for regulatory bodies methods to determine the species most ‘‘at risk’’ in the
and BC practitioners. receiving environment (Todd et al. 2015; Barratt et al.
ERA methodologies have been developed to ensure 2016) and methods to identify species that might be
the safe use of exotic BCAs (e.g., van Lenteren et al. harmed by competition from BCAs (McGrath et al.
2003, 2006; EPPO 2018). These favor specialist BCAs 2020).
by using methods that regard the use of generalist To assess risks of exotic GABCAs, a comprehen-
natural enemies as too risky (Bigler et al. 2006; van sive ERA should include adverse effects analysis,
Lenteren et al. 2006). Based on species biology and which at present is rudimentary (e.g., NAPPO 2015;
ecology, van Lenteren et al. (2006) proposed a tiered EPPO 2018; etc.). The present focus on host/prey
scoping assessment, which relied on a determination range testing identifies non-targets that could be
that the BCA had a low likelihood of establishment adversely affected, but it does not always address the
and a narrow host/prey range, to rapidly identify BCAs critical questions of whether the BCA is likely to have
with low environmental risk. In EPPO (2018), the first an adverse effect on the non-target and to what degree.
tier is an Express Environmental Impact Assessment Consequently, two major scientific issues must be
(EIA), which poses three questions. In the subsequent addressed: criteria for determining the potential for
Full EIA of the EPPO (2018) schema, the applicant is harmful/adverse effects and for identifying non-target
requested to address several issues related to non- species/communities/ecosystem processes most at
target effects using qualitative scores, all of which can risk, and methods and models for conducting effects
be answered with much greater certainty for specialist analysis. In addition, a harmonized ERA methodology
BCAs. These approaches eliminate from considera- that can be used for any BCA, including exotic
tion most exotic generalist natural enemies as BCAs, GABCAs, is also needed as the existing guidelines and
even ones that are unlikely to cause significant harm to regional agreements (e.g., FAO, NAPPO, EPPO) have
non-target species (Lynch et al. 2001; Andreassen not prevented unintended spread of exotic BCAs
et al. 2009). GABCAs vary significantly in diet across borders (Petit et al. 2009; Stahl et al. 2019).
E. Yano J. R. Nechols
Center for Ecological Research (CER), Kyoto University, Department of Entomology, Kansas State University, 123
Hirano 2-509-3, Otsu, Shiga 520-2113, Japan Waters Hall, 1603 Old Claflin Place, Manhattan,
KS 66506, USA
G. L. Lövei
Department of Agroecology, Aarhus University,
M. D. Eubanks
Flakkebjerg Research Centre, Forsøgsvej 1,
Department of Entomology, Texas A&M University,
4200 Slagelse, Denmark
TAMU 2475, College Station, TX 77843-2475, USA
N. Hinomoto
J. C. van Lenteren
Laboratory of Ecological Information, Graduate School of
Laboratory of Entomology, Department of Plant Sciences,
Agriculture, Kyoto University, Kyoto 606-8502, Japan
Wageningen University and Research (WUR),
PO Box 16, 6700 AA Wageningen, The Netherlands
P. H. B. Togni
Departamento de Ecologia, Universidade de Brası́lia
(UnB), Campus Universitário Darcy Ribeiro, Brası́lia,
DF 70910-900, Brazil
123
To enable consideration of the untapped potential with anonymous participation, except for the co-chairs
for the safe use of generalists, we propose a three- of the organizing committee. Forum materials pre-
tiered ERA (Scoping, Screening and Definitive sented background information about discussion
Assessments) building on previous methods to topics, framing and providing case studies to instan-
improve and develop criteria to evaluate the biosafety tiate the issues (Table 1). At the end of the six online
of any exotic GABCA and other BCAs. In this work, discussions, the workshop Using Generalist arthropod
exposure analysis (host/prey range assessment) is biological control agents: Ensuring effectiveness and
complemented with effects analysis (potential effects safety was organized at the 2019 annual meeting of the
assessment). With the structured use of the ‘‘best Entomological Society of America in Saint Louis,
available information’’ in this ERA, we aim to support Missouri, USA, attracting 50 participants from
regulation of exotic GABCAs and sensitize the industry, academia and regulatory agencies to discuss
broader scientific community and stakeholders regard- remaining topics that needed more thorough
ing the knowledge gaps that are impeding the contin- consideration.
ued improvement of ERA methodologies for the safe
use of exotic GABCAs.
The proposed comprehensive environmental risk
assessment
Expert panel working group
The proposed three-tiered ERA serves as a structure
The three-tiered ERA for exotic GABCAs was for a petition for an exotic GABCA (Fig. 1). The first
developed by a multidisciplinary expert panel of tier is a Scoping Assessment which aims to determine
specialists and stakeholders, during a 12 month if the exotic GABCA can be considered ‘highly
process (2018–2019) of structured online discussions, unlikely to have an adverse effect’ or if further risk
culminating in a public symposium and workshop. An assessment steps are needed. The second tier is a
organizing committee, composed of experts from four Screening Assessment, which is only required when
countries (Brazil, New Zealand, The Netherlands and the Scoping Assessment determines that an exotic
USA, see Supplementary Material), formed the panel GABCA needs additional assessment. The aim of this
and guided the online discussions. Professionals with tier is to identify the most prominent potential adverse
expertise in biological control, ecology, invasive effects associated with the exotic GABCA and the
species, regulation, risk assessment and species con- likelihood and magnitude of these effects to determine
servation, from academia, government and industry which, if any, merit a definitive assessment. The last
were invited without financial incentive. Sixty-six tier is the Definitive Assessment and is only needed
experts from 14 countries accepted the invitation. when the Screening Assessment determines that
Forty-seven experts participated at least once (see additional assessment is needed. The first two tiers
Supplementary Material), and 28 experts from 14 are heavily based on expert consultation, while the last
countries participated in at least three of the six tier is based on expert consultation and on data to be
forums. The forum summary reports 1 to 3 and 6 can provided. Depending on the number and complexity of
be provided upon request to the corresponding author. the identified potentially significant adverse effects, at
Summary reports 4 and 5 are provided in Supplemen- any point in the three-tiered ERA process, an applicant
tary Material. may choose to withdraw the petition and avoid
additional costs. Throughout the methodology, addi-
tional details (e.g., number of specimens to be
Online discussion forums evaluated by a taxonomist, whether the exotic
GABCA is evaluated as a population or species, etc.)
The six online discussions were held in English during could be determined by whomever may adapt the
two weeks every other month from November 2018 to methodology. In addition, because regulatory frame-
November 2019 using the JotForm platform (https:// works and authorities vary considerably around the
www.jotform.com/) accessed by individual private world, we have not specified who would be respon-
links. The Delphi method (Sackman 1974) was used sible for the costs and procedures in the ERA, such as
123
Table 1 Case studies used in the online discussions on criteria for environmental risk assessment of exotic generalist arthropod
biological control agents
Species Function Native region Introduced areas Rationale for case References
study
Amblyseius Predatory mite sold Eastern USA in 1983 Permit declined in Arthurs et al. (2009),
swirskii (Athias commercially in Mediterranean and released Brazil in 2018 for Cédola and Polack
Henriot, 1962) Europe and North region or tested since commercial use as (2011), Kade et al.
(Acarina: America for 2005 in parts augmentative BCA (2011), Sato and
Phytoseiidae) augmentative of the Asia, in greenhouse Mochizuki (2011),
biological control in Africa and production due to a EPPO (2020)
greenhouse and Argentina lack of an ERA in
nursery crops the country
Fopius arisanus Solitary egg-larval Southeast Asia Several Interest in using it to Vargas et al.
(Sonan, 1932), parasitoid of fruit- countries in control fruit flies in (1993, 2007),
(Hymenoptera: feeding Tephritidae Oceania, organic farms in Waterhouse (1993),
Braconidae) Borneo, India, Northeast of Brazil Holler et al. (1996),
Malaysia, Ovruski et al.
Taiwan, (2000), Wang et al.
Thailand, (2004), Carmichael
Guinea, et al. (2005), Rousse
Guatemala et al. (2005)
Harmonia Predatory lady beetle Northeast Asia 38 countries in Vast record of e.g., Roy et al. (2016)
axyridis introduced in many North and invasions and
(Pallas, 1773) countries to control South establishment, as
(Coleoptera: aphids Americas, well as associated
Coccinellidae) Africa and adverse effects
Europe
Macrolophus Predatory mirid sold Mediterranean Augmentative Permit declined in Castañé et al. (2011),
pygmaeus commercially in basin biological New Zealand in 2014 EPA (2013, 2014)
(Rambur, 1839) Europe for control agent for augmentative
(Hemiptera: augmentative in Europe BCA for greenhouse
Miridae) biological control in tomatoes due to
greenhouses likely to establish
outside of the
greenhouses
in the case data acquisition (last tier) is required to Intended use and likely benefits
quantify risks on non-target species.
The intended use of the exotic GABCA, including the
intended target pest(s) and the crop or commodity
Tier 1: Scoping Assessment attacked must be provided, as well as time of the year
(augmentative BC) and geographic area of the
The Scoping Assessment (Fig. 2) relies on expert intended release. The potential benefits from the
consultation to use existing scientific information on a exotic GABCA release must be described, including
case-by-case basis to evaluate the following six the likely level of control of the target pest(s),
sections. It focuses on determining the potential replacement or reduction of existing plant protection
benefits and adverse environmental effects of a actions (e.g., pesticides) and protection of biodiver-
GABCA. sity. A literature review of the host/prey species and
habitat use of the exotic GABCA should also be
conducted to enable understanding and evaluation of
the general risk–benefit tradeoff, which significantly
affects the acceptance of the exotic GABCA risks.
123
Fig. 1 Overall proposed tiered environmental risk assessment for exotic GABCAs. The flowchart symbols indicate: oval a beginning or
end, rectangle a step of the process, diamond a decision, arrow direction of flow
Species identification 4–10, 11–30 and [ 30 species) or taxonomic related-

ness of the host/prey (e.g., same genus) (e.g., van
Confirmation of the identity of the exotic BCA must be Lenteren et al. 2003). These are easy to implement
certified by a taxonomic specialist to reduce or unambiguously but are not directly related to risk. An
eliminate the possibility of sibling or cryptic species approach based on the functional relationship between
(Le Hesran et al. 2019). The specimens must be the exotic GABCA and its hosts/prey is more closely
deposited in a curated permanent collection and the related to risks and may be critical for risk assessment
taxonomist must provide a certificate of identification but requires a level of detail that is typically unavail-
with the method used to identify the species. In the able. One example is the McMurtry et al. (2013)
case of a taxon without a single species name (species classification of phytoseiid predatory mites based on
complex, subspecies, biotypes), distinction from other the functional similarity of the prey species. Func-
entities of the same rank must be provided. tionally specialized phytoseiids feed only on func-
tionally similar species, such as only Tetranychus
Level of polyphagy mites, tydeoid mites, or tetranychiid web-nest pro-
ducing mites. None of these functionally specialized
As there is a diet gradient from specialist to generalist predators would be considered specialists by the
BCAs, it is crucial to assess the level of polyphagy. general classifications mentioned above. However,
General classifications of polyphagy are mostly based because the prey species all have functionally similar
on the number of host/prey (e.g., groups of 0, 1–3, ecologies, the assessment of risks will be similar for all
123
Fig. 2 Tier 1: Scoping Assessment. The flowchart symbols indicate: oval a beginning or end, rectangle a step of the process, diamond a
decision, arrow direction of flow, black background indicates an expert consultation process. AE adverse effect; NT non-target species
the species, which will greatly simplify the assess- GABCA to complete the Scoping Assessment, the
ment. In a case where there is insufficient biological ERA continues to the Screening Assessment where
knowledge about the level of polyphagy of the exotic such knowledge must be provided.
123
Status of establishment of exotic GABCA establishment of an augmentative exotic

GABCA’’ section of the Scoping Assessment.
The importance of the verification of the establishment C. Reintroductions that do not meet condition A2 or
(or lack thereof) of an exotic GABCA resides in the A3 should continue with a simplified ERA (see
fact that, once established, a permit for introduction Supplementary Material), which is focused on
may not be necessary and, in many jurisdictions, a evaluating if these conditions can be met or if a
permit for commercial release will not require an full ERA is necessary.
ERA. Also, verification of the geographic distribution
of establishment is recommended, as the distribution
Potential establishment of an augmentative exotic
could significantly affect permit approval for move-
GABCA
ment and release. To prove that the exotic GABCA is
already established, evidence of successful reproduc-
For an augmentative exotic GABCA, either of the
tion and persistence must be demonstrated through
following can be used to demonstrate that it is highly
records of presence (e.g., labeled specimens in
unlikely to establish, including any environment
collections, records in publications, biopesticide reg-
besides open field:
istration certificates, ecological samples, internet-
based recording schemes) of adults and, where pos- A. The release comprises only sterile individuals or
sible, immatures. The minimum time frame for only one sex of a non-parthenogenetic species.
records of presence to demonstrate establishment of B. The exotic GABCA cannot complete its life cycle
the exotic GABCA will be defined by a specialist, outside a protected environment in the geograph-
designated by the regulatory agency, based on the ical region where it is intended for release due to
species life cycle and biology, and it should be greater any one of the following: (1) absence of suit-
than the time needed to complete at least three able host/prey; (2) asynchrony of the predicted life
generations and one annual cycle. If establishment is cycle with suitable hosts or their host plants or
in doubt or cannot be demonstrated, the ERA contin- prey; (3) abiotic or climate conditions (including
ues to ‘‘Potential establishment of an augmentative conditions anticipated for the future, e.g. possibly
exotic GABCA’’ section of the Scoping Assessment. next ten years) that prevent survival at some time
If the exotic GABCA is established, the ERA during an annual cycle.
continues as follows:
If it cannot be shown that the exotic GABCA is highly
A. Reintroductions of an established exotic species
unlikely to establish, the ERA continues to ‘‘Qualita-
are considered to have low environmental risk and
tive non-target species assessment’’ section of the
need no further risk assessment if all of the
Scoping Assessment. Otherwise the exotic GABCA
following conditions are met: (1) previous release
can be considered highly unlikely to establish and
was authorized by a competent authority based on
therefore highly unlikely to have an adverse effect on
a valid ERA (i.e., no significant changes in
the environment. In this case, there is no need for
biological or regulatory circumstances); (2)
further ERA.
adverse effects were not observed during testing
or after release where the exotic GABCA is
Qualitative non-target species assessment
established on known at risk non-target species
(usually identified using criteria such as phyloge-
For augmentative releases of exotic GABCA that may
netic, ecological, biological and/or socio-eco-
establish, including outside a protected environment,
nomic criteria; Sands and van Driesche 2004;
or classical releases of exotic GABCA, the following
Kuhlmann et al. 2006; Barratt et al. 2016); (3) the
must be provided: (1) a list of the known host/prey
source of the exotic population to be released is
range and species with which the exotic GABCA
the same as the established population.
directly interacts wherever it occurs (native range and
B. Reintroductions that do not have a valid ERA
any place it has established); (2) a summary of the
(condition A1) should proceed to ‘‘Potential
status of the phylogenetic relationships for the known
host/prey species and species with which it directly
123
interacts. Then, the following two requirements must establish the exotic GABCA. For augmentative
be met to determine that an adverse non-target effect is releases, the exact probability of establishment is
highly unlikely: greater than 0, so as a worst-case assumption, LE is
also set to 1. The potential AE is quantified for each of
A. No record of adverse effect on any non-target
the 19 categories of effects (Table 2), using readily
species elsewhere in the world. This assumes that
available data and conservative assumptions by
if there are no direct effects on any non-target
experts.
species, then there are no indirect effects of the
exotic GABCA, because all indirect effects must
Adverse effect (AE)
involve at least one direct effect (Messing et al.
2006). In addition, the record(s) of adverse effects
In the 19 categories of effects (Table 2) each category
must be on the population of the non-target
is designated with the subscript i, with i = 1, …, 19.
species, and not merely records of parasitism/
These categories of effects were designed to allow
predation.
independent discrimination of interdependent or cor-
B. Lack of an expected adverse effect in the
related factors so that their individual importance can
geographic region where it is intended to be
be clearly judged. The AE associated with each
released due to any of the following:
category of effect i is estimated by an expert consul-
(1) Absence of non-target species in the tation panel through the ‘adverse effect’ form (AE
intended release region based on the known form, Table 2), which follows elicitation methods
host/prey in the natural range or other areas described by Kynn (2008) and Vanderhoeven et al.
where it was previously introduced, for all (2017). In the AE form, each of the categories of effect
the 19 categories of effects in Table 2. These is scored according to Likelihood of effect (LEfi) and
categories expand on a shorter list in Snyder Magnitude of effect (MEfi) on a non-target species in
and Evans (2006). category i. AEi is then estimated by:
(2) Lack of encounter (spatial or temporal
AEi ¼ LEfi MEfi ð1Þ
overlap) with the potential non-target spe-
cies in the intended release region. We proposed ordinal scales (Table 2) to score LEfi and
(3) Existence of a biological constraint limiting MEfi. The odd number of choices anchored to specific
the interaction of potential non-target spe- values is to enhance score accuracy (O’Hagan 2005;
cies with the exotic GABCA (e.g., unrec- Morgan 2014). Each category of effect is scored with
ognized chemical/biological cues, the highest, lowest and median scores (in that order).
ovipositor not long enough to reach non- Scoring the extremes first eliminates arbitrary anchor-
target species; etc.). ing bias for the median score. Anchoring bias is a
cognitive bias that occurs when an individual favors
If it cannot be determined that the exotic GABCA is their initial considerations over more comprehensive
highly unlikely to have an adverse effect, the ERA sources of information. If the median is scored first,
continues to the Screening Assessment. Otherwise, the arbitrary information, possibly unique to each expert
exotic GABCA is considered highly unlikely to have will bias the estimate. Instead, the median is anchored
an adverse effect on the environment and, hence, to both extremes, which will force the expert to
requires no further ERA. consider the median to be a score between these
values, thereby eliminating arbitrary anchoring bias.
The scoring of extremes also allows consideration of
Tier 2: Screening Assessment different conditions/scenarios that might result in
different LEfi or MEfi, and provides the regulator an
In this tier (Fig. 3), risk is estimated and characterized indication of the associated uncertainty (Morgan
using worst case assumptions for the Likelihood of 2014). A brief justification for the scores must be
establishment (LE) and potential Adverse effect (AE). provided to clarify to the regulator the technical
For classical releases, LE is set to 1 (certain to knowledge and experience used, and the level of
establish), because the intent of the release is to subjectivity (O’Hagan 2005). This is particularly
123
122
123
Table 2 Adverse effect (AE) form for the Tier 2 Screening Assessment in which each expert should fill the fields of each of the 19 categories of effects of the exotic GABCA on
potential non-target species for a particular receiving environment. The scales are provided below the table. The fields are filled with an answer of an expert (in italic) as an
example of an environmental assessment of the exotic Harmonia axyridis in the USA
Categories of effects on non-target species Likelihood of effect (LEfi)
High Low Median Justification Median Non-target Stakeholder Possible harm
score score score score in possibly affected harmed (who
climate cares)
change
1
1. Reduction of a native top predator 0 0 0 Not many 0 Unknown Unknown
predators eat
coccinellids
2. Reduction of natural enemies via:
2
2a. Exploitative competition 5 4 4 This occurs 4 Co-occurring Biodiversity Reduction in
among coccinellids conservationists coccinellid diversity
coccinellids and abundance
3
2b. Asymmetrical competition 0 0 0 Unknown 0 Unknown Unknown
4
2c. Intraguild predation 6 6 6 Commonly occurs 6 Same as 2a Same as 2a Extirpation and
reduction in
coccinellid diversity
and abundance
2d. Immunity from shared natural enemies 2 1 2 Harmonia not 2 Same as 2a Same as 2a Same as 2a
with native species 5 parasitized
much compared
to natives
2e. Co-introduction of new pathogens that 3 1 2 Possibly 2 Same as 2a Same as 2a Same as 2a
infect native species 6 introduced a
virulent Nosema
2f. Reproductive interference with native 4 2 2 Interspecific 2 Hippodamia Same as 2a Reduction in
species 7 mating convergens abundance
8
2g. Hybridization with another strain 1 0 0 Strain difference 0 Harmonia Farmers Reduced biological
known, but effect strains control
unknown
2h. Reduced biological control 0 0 0 Harmonia as 0 Co-occurring Same as 2h Same as 2g
effective or coccinellids
complementary and aphid
D. P. Paula et al.
to natives parasitoids
3. Reduction in herbivory with:
3a. Improved biological control 6 5 6 In soybean 6 Aphis glycines Soybean growers Reduction in
abundance (benefit)
Table 2 continued
Categories of effects on non-target species Likelihood of effect (LEfi)
High Low Median Justification Median Non-target Stakeholder Possible harm
score score score score in possibly affected harmed (who
climate cares)
change
3b. Release of undesired plant (weed) 0 0 0 Harmonia as 0 Unknown Unknown

population from herbivory effective or
complementary
to natives
3c. Competitive suppression of a plant by a 0 0 0 Harmonia as 0 Unknown Unknown
released plant9 effective or
complementary
to natives
3d. Reduced insecticide use 5 2 4 Probably reduces 4 Aphis glycines Soybean growers, Reduced insecticide
insecticide use environmentalists use
in soybean
4. Reduction in valued species:
4a. Species of conservation interest 1 0 0 Virtually no 1 Karner blue, Legal protection Take
(endangered, threatened or rare endemics) habitat overlap, Poweshiek
spillover skipperling
Integrating adverse effect analysis into environmental risk assessment
possible
4b. Beneficial arthropods (e.g., pollinators, 0 0 0 None known 0 Unknown Unknown
detritivores)
4c. Reduction in some other valued species10 0 0 0 None other known 0 Unknown Unknown
5. Increase in herbivory via:
5a. Direct herbivory 6 5 5 On some fruit 5 Grapes Grape farmers Crop loss (economic)
crops during fall
5b. Enhanced mutualism between exotic 0 0 0 No mutualisms 0 Unknown Unknown
GABCA and another organism11 with herbivores
6. Increase in a damaging organism vectored 0 0 0 Not a vector 0 Unknown Unknown
by the exotic GABCA12
(a) Overall assessment of the quality and 6 4 5
quantity of the information available for you
to use to indicate the scores of the categories :
(b) Overall assessment of the reliability of your 5 4 4
judgmentà:
Optional: List in order of priority additional information and/or research that would reduce the uncertainty of your assessment
123
123
Table 2 continued
124
Categories of effects on non-target species Magnitude of effect (MEfi)
123
High Low Median Non-target considering Median score Justification for
score score Score climate change considering climate change
climate change median score
1
1. Reduction of a native top predator 1 0 0 None 0
2. Reduction of natural enemies via:
2a. Exploitative competition 2 3 2 2 Unknown 2 Interaction strength
may increase or
decrease
3
2b. Asymmetrical competition 0 0 0 None 0
4
2c. Intraguild predation 4 3 4 Unknown 4 Same as 2a
5
2d. Immunity from shared natural enemies with native species 1 0 0 Unknown 0 Same as 2a
6
2e. Co-introduction of new pathogens that infect native species 2 0 2 Unknown 2 Same as 2a
7
2f. Reproductive interference with native species 2 0 1 Same species as now 1 Same as 2a
8
2g. Hybridization with another strain 1 0 0 Same species as now 0 Same as 2a
2h. Reduced biological control 2 0 1 Unknown 1 Same as 2a
3. Reduction in herbivory with:
3a. Improved biological control -3 -2 -3 Same species as now -3 Same as 2a
3b. Release of undesired plant (weed) population from herbivory 0 0 0 None 0
3c. Competitive suppression of a plant by a released plant9 0 0 0 None 0
3d. Reduced insecticide use -2 -2 -2 Same species as now -2 Same as 2a
4. Reduction in valued species:
4a. Species of conservation interest (endangered, threatened or rare 2 1 2 Same species as now 2 May have greater
endemics) encounter
4b. Beneficial arthropods (e.g., pollinators, detritivores) 0 0 0 None 0
4c. Reduction in some other valued species10 0 0 0 None 0
5. Increase in herbivory via:
5a. Direct herbivory 2 0 1 Same species as now 1 Same as 2a
5b. Enhanced mutualism between exotic GABCA and another organism11 0 0 0 None 0
6. Increase in a damaging organism vectored by the exotic GABCA12 0 0 0 None 0
(a) Overall assessment of the quality and quantity of the information 5 3 4
available for you to use to indicate the scores of the categories :
(b) Overall assessment of the reliability of your judgmentà: 5 4 4
Optional: List in order of priority additional information and/or research that would reduce the uncertainty of your assessment
D. P. Paula et al.
Table 2 continued
Likelihood of effect scale Magnitude of effect scale Quality and quantity of information Reliabilityà
0 = Very highly unlikely -4 = Massively beneficial: widespread, large, 0 = Very poor (worst 5% of cases) 0 = Totally unreliable (0–0.05 chance of being
(0–0.05) consistent correct)
1 = Highly unlikely (0.05—0.2) -3 = Highly beneficial: widespread, large, 1 = Poor (next worst 15% of cases) 1 = Unreliable (0.05—0.2 chance of being
variable correct)
2 = Unlikely (0.2–0.4) -2 = Beneficial: local or small and variable 2 = Below average (next worst 20% 2 = Slightly unreliable (0.2–0.4 chance of
of cases) being correct)
3 = Neither unlikely or likely -1 = Slightly beneficial: local and small 3 = Average (middle 20% of cases) 3 = Neutral (0.4–0.6 chance of being correct)
(0.4–0.6)
4 = Likely (0.6–0.8) 0 = Neutral, neither beneficial or adverse: 4 = Above average (next best 20% of 4 = Slightly reliable (0.6–0.8 chance of being
cases) correct)
5 = Highly likely (0.8–0.95) 1 = Slightly adverse: local and small 5 = Good (next best 15% of cases) 5 = Reliable (0.8–0.95 chance of being
correct)
6 = Very highly likely 2 = Adverse: local or small and variable 6 = Very good (best 5% of cases) 6 = Totally reliable (0.95–1.0 chance of being
(0.95–1.0) correct)
3 = Highly adverse: widespread and large, but
variable
4 = Massively adverse: widespread and large and
consistent
Integrating adverse effect analysis into environmental risk assessment
1
Also known as alpha predator or apex predator, it is a predator at the top of a food chain, i.e. with no natural predators
2
Organisms directly compete for resources by exploiting and using the resources so they are unavailable to other organisms
3
Unequal division of resources amongst competing organisms. This often occurs when one species uses a resource earlier in the season than the other species, or modifies the
resource so that later it does not become available to the other species
4
Predation/parasitism on other natural enemies
5
By being immune to shared natural enemies, the exotic GABCA is not as affected by them as are the native species. This can exacerbate of competition and intraguild predation
6
The exotic GABCA may bring in new pathogens that are avirulent to it, but are virulent in native species
7
Interactions among species that reduce the reproductive fitness of one or both species
8
Mating between strains of the BCA that may produce descendants that are fertile or sterile
9
A plant that is not a weed may be released from herbivory, and the released plant may competitively suppress a desired plant
10
These include species of cultural concern
11
A formicid GABCA could enhance the mutualism with sap-sucking pest species, resulting in greater crop damage
12
These could be plant or animal pathogens or symbionts transmitted to other species that cause mortality or sterility

Quality and quantity of information; and àReliability
125
123
the magnitude of the effect should be based on the size

of the harm, not merely on the effect on the non-target
species.
Finally, an overall assessment of the quality and
quantity of the information available for scoring and
reliability of the judgment (second order uncertainty)
is also provided, which allows the regulator ways to
evaluate if the answers were based more on general
experience/knowledge or on actual data for the exotic
BCA (Morgan 2014). This empowers the regulator to
weigh the entire response of an expert when experts
give highly different scores, which may happen. As an
option, the expert may list in order of priority,
additional information and/or research that would
reduce the uncertainty of the assessment. This pro-
vides an opportunity for the regulator to become aware
of (or sensitized to) research areas that need to be
encouraged (Morgan 2014).
The AE form completed by the experts can be
combined by the regulator into a single score using
fuzzy systems. A fuzzy system is proposed because it
accounts for and quantitatively preserves uncertainty
related to lack of knowledge. This is more realistic
Fig. 3 Tier 2: screening assessment. The flowchart symbols than standard logic for evaluating the safety of an
indicate: oval a beginning or end, rectangle a step of the process,
diamond a decision, arrow direction of flow. Black background environmental stressor when there are knowledge
indicates an expert consultation process. LE likelihood of gaps. Fuzzy systems are based on the concept of
establishment, AE adverse effect, i a category of effect, LEf membership. For example, suppose for some likeli-
likelihood of effect, MEf = magnitude of effect, R = risk hood of an effect, LEfi, that minimum score is 0.1, so it
cannot be less than 0.1 (membership of 0) and could
important when the same category of effect receives possibly be greater than 0.5 (membership of 1), but we
highly variable scores among experts, because it have no information to know if it could possibly be
provides a way for the regulator to moderate or weight between 0.1 and 0.5. Under standard logic, we would
scores of different experts. The possible effects of have to assume that it was possible or not (0 or 1),
plausible climate change scenarios are also scored between 0.1 and 0.5, but under fuzzy systems, we can
(only as a change in the median score) to communicate specify a membership between 0 and 1 to reflect the
the relative stability (or dependence) of the score on knowledge gap. Depending on the assumption, stan-
present conditions or circumstances. dard logic would over or underestimate the true value
The scientific name(s) of some non-target species because it ignores the knowledge gap.
possibly affected, the stakeholders concerned, and a To apply a fuzzy system to the scores in the AE
description of the possible harm must also be form, the high, low and median scores for each LEfi
provided. The non-target species list does not need and MEfi are converted into fuzzy sets (Pedrycz and
to be comprehensive, but is needed to ensure that the Gomide 2007). The fuzzy sets for LEfi and MEfi for
estimated values have specificity and are not based on each responding expert are multiplied together (Rah-
generalities. The identification of stakeholders and man 2016) to produce a fuzzy set for adverse effect
possible harm is essential to ensure that the effect is i (AEi) for each expert. The AEi fuzzy sets of all
actually adverse and to ensure that the estimate of the experts are averaged into an aggregated fuzzy set and
magnitude of the effect is based on more than an defuzzified by the centroid method to produce an
impact on a non-target. For an effect to be considered estimated value of AEi (Table 3). The centroid is the
adverse, it must harm some group of stakeholders, and weighted mean AEi, weighted by the membership in
123
Table 3 Adverse effect (AE) calculated for each category of characterize the acceptability of the risk, six risk threshold
effect obtained from 13 experts answers to the AE form for the options were tested. Categories of effects marked with an X
Harmonia axyridis example (Table 2) and, consequently, the have risk scored above the proposed threshold and, therefore,
associated risk score using the upper 0.5 k-cut parameter. To would need to be evaluated in the Definitive Assessment
Potential effects and Adverse effect Risk Threshold options
categories of effects on non- (R)
target species Centroid Upper Centroid R[6 RC5 RC4 R C 3* R C 4** R C 4***
0.5 k- for
cut climate
change
1. Reduction of a native top 0 0 0 0

predator
2. Reduction of native natural
enemies via:
2a. Exploitative competition 9.7 11.7 9.7 11.7 X X X X X X
2b. Asymmetrical 0 0 0 0
competition
2c. Intraguild predation 22.2 24 22.2 24 X X X X X X
2d. Immunity from shared 0.5 1.1 0.5 1.1
natural enemies with
native species
2e. Co-introduction of new 2.5 4.1 2.5 4.1 X
pathogens that infect
native species
2f. Reproductive 2.9 5.2 2.9 5.2 X X X X X
interference with native
species
2g. Hybridization with 0.2 0.5 0.2 0.5
another strain
2h. Reduced biological 0 0 0 0
control
3. Reduction in herbivory
with:
3a. Improved biological -15.5 -18.0 -15.5 -18.0
control
3b. Release of undesired 0 0 0 0
plant (weed) population
from herbivory
3c. Competitive suppression 0 0 0 0
of a plant by a released
plant
3d. Reduced insecticide use -7.4 -9.1 -7.4 -9.1
4. Reduction in valued
species:
4a. Species of conservation 0.5 1.1 1.4 1.1
interest (endangered,
threatened or rare
endemics)
4b. Beneficial arthropods 0 0 0 0
(e.g., pollinators,
detritivores)
4c. Endemic species or 0 0 0 0
species of cultural value
123
Table 3 continued
Potential effects and Adverse effect Risk Threshold options
categories of effects on non- (R)
target species Centroid Upper Centroid R[6 RC5 RC4 R C 3* R C 4** R C 4***
0.5 k- for
cut climate
change
5. Increase in herbivory via

5a. Direct herbivory 5.5 8.7 5.5 8.7 X X X X X X
5b. Enhanced mutualism 0 0 0 0
between exotic GABCA
and another organism
6. Increase in a damaging 0 0 0 0
organism vectored by the
exotic GABCA
Categories with risk (R) above 3 4 5 4 4 4
the threshold
*Only for category ‘‘Reduction in a valued species’’, otherwise R C 5; **only when LEfi C 4, otherwise R C 5; ***only when
MEfi C 3, otherwise R C 5
the fuzzy set. The variation among expert scores is Material). This resulted in six proposed thresholds
measured by a k-cut of the aggregated AEi fuzzy set (Table 3), which were subsequently evaluated by 13
with membership equal to 0.5 (this is like a SD) experts using the well-known case of the predatory
(Table 3). As the Screening Assessment uses conser- Asian ladybeetle Harmonia axyridis (Pallas) (Coleop-
vative assumptions, the upper value of the k-cut is tera: Coccinellidae) established in the USA. The
used as an estimate of AEi. The fuzzy set calculations thresholds R C 5, R C 3*, R C 4** and R C 4***
should be automatically provided by embedded indicated the same categories for additional assess-
formulas in the compiled AE form so that regulators ment and were intermediate between the R [ 6 and
or experts would not be required to calculate them. R C 4 (Table 3). Experts considered threshold R [ 6
to be too high and threshold R [ 4 to be too low. The
Risk characterization thresholds R C 3*, R C 4** and R C 4*** were
considered unnecessarily complex. A category of
Having set the LE = 1, and estimated the AEi (Table 3) effect with R \ 5 means that in the worst case, the
using conservative assumptions, the risk (Ri) associ- effect would be either likely and slightly adverse (local
ated with each category of effect i can be estimated by: and small), unlikely and adverse (local or small and
variable), or highly unlikely and potentially massively
Ri ¼ LE AEi ¼ AEi ; ð2Þ
adverse (widespread, large and consistent). Hence, the
which is given above as the upper 0.5 k-cut of the AEi threshold R C 5 was selected as the risk threshold for
average fuzzy set. In addition to calculating a risk, risk potential adverse effects to need a definitive assess-
characterization involves determining which risks are ment. However, experts believed that the threshold for
important enough to merit definitive assessment (Tier valued species should be lower because these species
3). need additional protection. A category of effect with
To determine thresholds to characterize the risk, we R \ 4 indicates that in the worst case, the effect would
first tested the acceptability of various risk thresholds be either unlikely and slightly adverse (local and
by allowing experts to consider the results of the small) or highly unlikely and adverse (local or small
Screening Assessment for exotic GABCAs with which and variable). Experts agreed that for valued species
they were familiar (forum 5 in Supplementary
123
analysis in the Definitive Assessment, the H. axyridis

example in the USA (Table 3) indicates that there were
four categories of effects with risks above the thresh-
old: exploitative competition (R = 11.7), intraguild
predation (R = 24), reproductive interference with
native species (R = 5.2) and direct herbivory
(R = 8.7, on grapes). Subsequently, if a petition was
continued, the risks associated with those four cate-
gories of effects would need to be evaluated in a
definitive assessment. In the case that all of the
potential categories of effect for the exotic GABCA
have risks \ 4 for valued species and \ 5 otherwise,
the exotic GABCA can be considered highly unlikely
to have a significant adverse effect on the environ-
ment, and there is no need for further ERA. For
example, for the oligophagous parasitic wasp Fopius
arisanus (Sonan, 1932), (Hymenoptera: Braconidae)
in Brazil, there was no category of effect with risk
above the threshold. Therefore, F. arisanus would be
considered highly unlikely to have a significant
adverse effect in Brazil and the ERA would be
concluded.
Benefit scores (negative risks) for both F. arisanus
and H. axyridis were high for improved biological
control (R = - 10.7 and - 18.0) and reduced insec-
ticide use (R = - 8.2 and - 9.1), respectively.
Therefore, for F. arisanus, the potential benefits may
outweigh the potential risks of a release. For H.
axyridis, the four potential risks identified were not
outweighed by the two potential benefits. These case
studies exemplify that risk characterization in this
Screening Assessment takes into consideration three
kinds of effects: adverse effects (risk scores above the
thresholds), benefits (risk scores below zero) and
neutral (risk scores between zero and the thresholds,
Fig. 4 Tier 3: Definitive Assessment. The flowchart symbols i.e. insignificant AE), and provides the regulator the
indicate: oval a beginning or end, rectangle a step of the process,
diamond a decision, arrow direction of flow. Black background opportunity to compare and weigh them.
indicates an expert consultation process. i = category of
adverse effect, j = a non-target species; Effij = overall short-
term effect of an exotic GABCA on a non-target species j; Tier 3: Definitive Assessment
LEnij = Likelihood of encounter between the exotic GABCA
with the non-target species j; Lij = likelihood of an effect on
non-target species j after encounter; Mij = magnitude of an The Definitive Assessment (Fig. 4) provides a quan-
effect on non-target species j after encounter tification of the risks of an exotic GABCA for the
category(ies) of effect(s) that scored R C 4 for
the threshold should be R C 4 needing definitive endangered, threatened, or rare endemic species, and
assessment. R C 5 otherwise, in the Screening Assessment. It
Having established the risk thresholds of R C 4 for proceeds by:
endangered, threatened, or rare endemic species, and
R C 5 otherwise for effects that require further
123
A. Identifying non-target species with potentially that relevant non-target species can be outside of the
significant risk(s). If a species list is not already host/prey range and would not require host/prey range
available, research in the receiving environment test(s).
must be conducted to create the list. The applicant For some categories of effects, host/prey range tests
is responsible for providing a reliable list of may be needed to determine the potential non-target
species to be confirmed by the regulatory body. species, such as for intraguild predation, direct
B. Specifying the mechanistic pathway(s) of adverse herbivory and species of conservation concern. The
effect(s). This is achieved using interaction net- applicant, the regulatory agency or a third party could
works involving the exotic GABCA and the be responsible for collecting the necessary data, but
identified non-target species. policies could be considered so that the process is
C. Identifying assessment and measurement end- transparent, that small companies and BCAs for local
point(s). An assessment endpoint is an attribute and small market crops are not excluded and that
(e.g., abundance, distribution, etc.) of the non- potential conflicts of interest are appropriately man-
target species that is assessed via a measurement aged. The non-target species identification process
endpoint, which is a quantifiable indicator of the presented here provides a rough quantitative risk
assessment endpoint. estimate, building on Kuhlmann et al. (2006) and
D. Generating estimates of risk component(s). This PRONTI (Barratt et al. 2016).
includes establishment of the exotic GABCA, The objective of the NT form is to link character-
encounter between the exotic GABCA and the istics of the non-target species with elements of risk to
non-target species, and effects on that species or identify species that may have greater short-term
ecosystem services. effects and might need additional testing to determine
E. Characterizing risk. This involves combining the long-term effects. This is achieved by making realistic
components and then interpreting if it is accept- estimates of components that are used to estimate the
able or not. short-term effect(s) of the exotic GABCA on each
potential non-target species using expert consultation
and scientific data. The short-term effect of the exotic
Non-target species identification
GABCA on a non-target species is the proportion of
non-target individuals in a generation that is expected
In previous non-target species selection methods (e.g.,
to be killed/reduced/affected by an established popu-
Kuhlmann et al. 2006; Todd et al. 2015), the potential
lation of the exotic GABCA (e.g., Hopper 2001).
non-target species are listed at the beginning of the
Long-term effects on a population require careful
ERA, resulting in a large number of non-target
analysis of how such short-term effects interact with
species, which are then filtered in subsequent steps.
the mechanisms regulating the non-target population,
Unlike those methods, the initial list of potential non-
to determine if the short-term effects actually affect
target species to be examined in the Definitive
the long-term equilibrium population size. Thus, a
Assessment is based on the results of the risk
short-term effect is highly unlikely to underestimate
characterization from the Screening Assessment for
the long-term effect on the non-target population.
each category of effect in which the risk scored above
These short-term effects are then extended to deter-
the threshold, i.e. only potentially significant adverse
mine if any of the non-target species require an
effects, will be considered. For example, for H.
evaluation of long-term population and indirect
axyridis in the USA (Table 3), experts would look
ecosystem effects. As indirect ecosystem effects will
for non-target species in four categories of effects with
generally require a long-term change in the population
risk above the threshold, and select representative test
of the non-target, using short-term effects for these
species for each category using the ‘non-target species
indirect ecosystem effects is also unlikely to underes-
form’ (NT form, Table 4). For the case of exploitative
timate them as well. This has the advantage of
competition, the potential non-target species would be
delaying or avoiding the need to evaluate long-term
other natural enemies with which H. axyridis is likely
population and indirect ecosystem effects, which can
to compete for similar prey. This category and the
be expensive and time-consuming, to the last steps of
reproductive interference category emphasize the fact
the Definitive Assessment. Finally, the estimated
123
Table 4 Non-target (NT) species form to evaluate short term are automatically calculated. If estimates cannot be provided,
effects of an exotic GABCA on a non-target species j in expert should note with an explanation. Alternatively, expert
category of effect i (Effij) for the exotic Harmonia axyridis in may provide the highest score possible and note this in
the USA, considering the category of effect ‘‘2c. Intraguild comments. More columns should be added for additional
predation’’ with R = 24 from the Screening Assessment. Each species. Cma = Coleomegilla maculata; Cmu = Cycloneda
expert should provide estimates (between 0 and 1) for each munda
parameter (answers are exemplified in italics). Numbers in bold
NT species
Cma Cmu
Likelihood of Encounter (LEnj)

E1. Proportion of the geographic range of NT species that is included in the 1.0 1.0
predicted geographic range of the exotic GABCA1
E2. Proportion of the season when a vulnerable stage(s) of the NT species is 1.0 1.0
available that occurs when the exotic GABCA is active2
E3. Proportion of the habitats (in the E1 geographical range) used by a NT 0.90 0.90
species that are also used by the exotic GABCA or that can be reached by
dispersal of the exotic GABCA.3
E4. Proportion of plants in habitats used by the NT species that are expected to 0.80 0.60
be searched by the exotic GABCA (niche component).4
E5. Likelihood that the non-target species will be found by the exotic GABCA 0.50 0.30
on the plants that it searches (i.e., those plants satisfying E4).5
LEnj = E1j 9 E2j 9 E3j 9 E4j 9 E5j Lower 0.36 0.16
bound
Uncertainty in LEnj: provide a lower and upper bound of LEnj that you believe Upper 0.10 0.10
will give a 95% credibility interval for these LEnj estimates bound
Likelihood (Lj) of a short-term effect on non-target species j 0.70 0.40
A1. Likelihood that the exotic GABCA recognizes and attacks the NT species6 1.0 1.0
A2. Likelihood that the exotic GABCA successfully kills (or damages) the non- 0.50 0.50
target individual (or plant part) after attack (A1).7
Lj = A1j 9 A2j 0.50 0.50
Uncertainty in Lj: provide a lower and upper bound of Lj that you believe will
give a 95% credibility interval for these Lj estimates
Lower bound 0.30 0.30
Upper bound 0.70 0.70
Magnitude (Mj) of a short-term effect on non-target species j
These are set to 1 for the case that the exotic GABCA successfully kills (or
damages) the NT individual it attacks. This can be modified for other less
severe effects, such as trait-mediated effects
Mj = 1 1.00 1.00
8
Overall short-term effect
Effj = LEnj 9 Lj 9 Mj 0.18 0.08
Uncertainty in Effj
9
Lower bound 0.03 0.03
10
Upper bound 0.49 0.28
Valuation (V) of short-term effects
Legally protected species (endangered, threatened or special concern)
Enter an estimated population size of attacked stage of the species
Enter the degree of threat: endangered = 1; threatened = 1.5; special
concern = 2
Significance of species (small is more significant) none none
or
123
Table 4 continued
NT species
Cma Cmu
Rare endemic species

Enter the degree of endemism (proportion of the total land area that is
suitable habitat)
Enter the relative population size in the endemic areas (relative to another
endemic species, expressed as a proportion of that species)
Significance of species (small is more significant) none none
or
All other species
List why the species is valued (e.g., income, biological control service, Biological Biological
pollination service, culture, etc.) control control
service service
Significance: estimate how much the Effj for that species would reduce this 8.00 7.00
value (1–10 scale, 1 is 0–10%, 10 is 90–100%). (large is more significant)
List the stakeholder(s) who may be harmed by the loss of value Soybean Soybean
producers producers
Based on Effj and V, would you select which (if any) Selected Not selected
NT species for additional assessment?
1
This can be estimated from maps of the known or predicted geographical distribution of the NT species and exotic GABCA
2
This is only for the NTs satisfying E1, i.e., in the part of the geographic range where they overlap. This can be estimated by knowing
the vulnerable stage of the NT species and when it occurs seasonally relative to the predicted seasonal activity of the exotic GABCA
in the region of geographic overlap
3
This is only for the NTs satisfying E2, i.e., with seasonal overlap with the exotic GABCA. This can be estimated from the habitats
used by the NT species (habitats that are both suitable and occupied by the NT) that are also used by the exotic GABCA, and NT
habitats that are close enough to GABCA habitats that the exotic GABCA will disperse into the NT habitat. The NT and GABCA
may overlap geographically and temporally, but if they do not use the same habitat, they will not encounter each other
4
This is only for habitats satisfying E3. For NT plants, the likelihood that the exotic GABCA will find the plant in the habitats
satisfying E3. For NT plants, this can be estimated approximately from the proportion of NT plants in the habitat. For NT species on
plants, this can be estimated from the proportion of plants used by the NT species in the habitat, assuming random search by the
exotic GABCA. If the exotic GABCA prefers these plants the proportion will be higher than random, and if it disprefers these plants,
the proportion will be lower. For weed-free monocultures, this parameter might be 1, but for vegetationally diverse habitats, it might
be much less than 1
5
This parameter is valid only for non-plant NTs. This niche component should be considered for the entire vulnerable stage of the NT
species assuming that the exotic GABCA does search a plant that has the NT species. This score will be affected by the expected
density of exotic GABCA, how rapidly the NT species can be found, and how rapidly the exotic GABCA leaves the plant before
finding it. Several factors may affect this likelihood, such as non-target characteristics that make it highly accessible (e.g., release
attractive semiochemicals) or inaccessible (e.g., hiding, feeding on a plant part that the exotic GABCA does not search) to the exotic
GABCA. Inaccessible species will have a likelihood near 0
6
Encounter does not always result in attack. Attack is affected by: (1) Presence or absence of NT defense behaviors or characteristics
that reduce attack rates (e.g., crypsis, aposematic coloration, hardness of chorion or cuticle, kicking, dropping, removing host/prey
location cues used by the exotic GABCA, irritating regurgitant, toxic reflex bleeding, etc.). (2) For NT plants, characteristics that
make the plant easy (e.g., semiochemicals that call in the GABCA) or hard (e.g., repellent semiochemicals, occurence in
microhabitats [shade, sun, aspect] not normally searched by the GABCA) to find for the exotic GABCA
7
Many attacks of predators fail. For an exotic GABCA parasitoid, this can be reduced by host physiological defenses, such as
encystment and encapsulation
8
This can be modified for other less severe effects, such as trait-mediated effects
9,10
Multiplication of all the lower and upper bounds and Mj, respectively
123
short-term effects are then evaluated to determine if on that species. Valuation of rare species (e.g.,
any are large enough to require additional testing. endangered, threatened, and rare endemics) is based
The short-term effect of an exotic GABCA on a on the desire to preserve such species. Hence, all other
non-target species j in category of effect i (Effij) is things being equal, the rarer the species and the greater
assessed by independently estimating the: the effect of the GABCA (Effj), the greater is the need
to protect the species against effects by the exotic
• Likelihood of encounter (LEnij) between the
GABCA, and the higher is its value. If the number of
exotic GABCA with the non-target species j,
individuals harmed is predicted to be very small (such
which is determined from geographic, temporal,
as less than the fecundity of a single female) and the
habitat and niche use overlap.
number is predicted to be a small part of the population
• Likelihood (Lij) of an effect on non-target species
(such as less than 0.1% of the population), then such a
j, which is the probability of successful attack or
species would probably not need to be selected for
interaction.
additional testing. Such a small effect would be very
• Magnitude (Mij) of an effect on non-target species
difficult to measure precisely enough to determine that
j after encounter, which is the consequence of the
an exotic GABCA would actually adversely affect the
attack or interaction.
rare species. For legally protected species (endan-
gered, threatened or special concern) and some rare
These estimates are combined (Table 4) to estimate an
endemics where there is a population size estimate
overall short-term effect on a non-target species j as
(available in the documentation supporting the orig-
follows.
inal listing or the recovery plan), this estimate can be
Effij ¼ LEnij Lij Mij ð3Þ
used to evaluate the relative rarity and potential
Quantitative data, if they exist, should be used to severity of Effj. For most rare endemics, population
estimate any of those components of the risk. Other- estimates are not available, so relative rarity should be
wise, estimates should be based on expert judgments. estimated, perhaps by comparison with another, better
This is similar to stepwise approaches in ERA in many known endemic.
fields, including biological control (Andow et al. Valuation of all other species in the remaining
1995; Olckers 2003). categories of potential adverse effect is based on
Large Effij are not automatically more significant identifying the value, assessing the relative impact of
than small ones, because significance depends on their Effj on that value and identifying the stake-
value to society. Valuations are used to determine holder(s) who may be harmed by the reduction in
which species, if any, should be selected as a test value. There are many possible economic and envi-
species in the final part of the Definitive Assessment ronmental values that a non-target species may
(Table 4). For each non-target species j, the short-term contribute, including pollination, biological control,
effects (Effj) are used to determine relative valuations biodiversity value, cultural values (beliefs, traditions,
to make them comparable across the species and rituals), symbolic value, aesthetic value, and income
facilitate the selection process. Values include reduc- value. The value(s) that the non-target species con-
tion in ecosystem services, erosion of biodiversity, tributes is important to identify so that it is possible to
cultural, symbolic, aesthetic, or income values, and make a concrete assessment of the significance of
harm to protected species. All of these values require a Effj. To assess significance, estimate how much the
human agent/stakeholder who is harmed by the loss of Effj would reduce the value, using a 1–10 scale (1 is
value, as specified in the Screening Assessment 0–10% reduction in value, 10 is 90–100% reduction in
(Table 2). This implies that valuations may differ value). Lastly, identify the stakeholder(s) who may be
amongst jurisdictions. In the interest of environmental harmed by this reduction in value. The regulatory
justice, effects on stakeholders who lack power, authority may wish to consider stakeholder salience to
legitimacy and urgency may be valued more highly value some non-target species more than others.
than those with power, legitimacy and urgency Stakeholder salience is the prominence of stakehold-
(Mitchell et al. 1997). ers to an issue and is based on power, legitimacy and
Valuation relies on both the value of the species that urgency. Highly salient stakeholders have the power to
could be affected and the size of the short-term effect affect the issue, are recognized by others as having a
123
risk(s) associated with the non-target species should

be provided to support assessment and measurement
endpoint(s), an analysis model, and the need for
gathering any further data, as follows.
Specifying the mechanistic pathway(s) of adverse

effect(s)
To enable quantitative evaluation of the risk to the

non-target effects that remain, the pathway(s) by
which the effect is expected to occur should be
specified. This guides the quantification as the strength
of the pathway(s) is an important component of the
Fig. 5 Some possible pathways by which an exotic GABCA
risk, because the risk is the product of the strength of
(eGABCA) could affect a non-target (NT) pollinator. (1)
eGABCA directly adversely affects NT pollinator, by consum- the pathway(s) and the effect of the pathway(s) on the
ing it or disrupting its behavior. (2) eGABCA indirectly non-target population(s). The pathways can be spec-
adversely affects NT pollinator by suppressing Pest 1, which ified with the ecological interactions between the
had facilitated the NT pollinator, possibly by the release of
semiochemicals that attracted the NT pollinator. (3) eGABCA
exotic GABCA and the non-target species (Puccia and
indirectly adversely affects NT pollinator via another natural Levins 2013). An example of three possible pathways
enemy (NT NE), releasing Pest 2, which suppresses the Plant. by which an exotic GABCA could affect a non-target
Black solid lines are direct effects involved in the pathways. pollinator are shown in Fig. 5.
Gray solid lines are other direct effects. Dotted lines are indirect
effects. Arrow represent positive interactions and circles
represent negative interactions, using the notation of Puccia Assessment and measurement endpoints
and Levins (2013)
An assessment endpoint must be proposed for risk
legitimate interest or concern in the issue, and have an quantification. One logical assessment endpoint for
urgent need to address the issue. The regulatory long-term effects on a non-target population is the
authority may choose to value high and/or low predicted reduction in the long-term population size
salience (and uncertainty therein) as a part of the by the exotic GABCA. However, other endpoints for
valuation process. the long-term effect on a non-target population and
Finally, at the end of the NT form, each expert indirect ecosystem effects could be proposed with
should select the non-target species (if any) for further sufficient justification. A measurement endpoint could
assessment through risk quantification keeping in be the same as the assessment endpoint, or it could be,
mind that parameters equal to zero or close to zero for example, other population dynamics parameters
means the risk is zero or close to zero. During the that are related to the long-term population size.
discussion forums, 16 experts completed the NT form
using familiar exotic GABCAs and NT species, Risk quantification analysis plan
illustrating a range of cases from no to all NT species
selected for further assessment (forum 4 in Supple- Several possibilities could be considered to enable risk
mentary Material). It is possible that the short-term quantification for the non-target species identified:
effects on all of the species in all of the categories are
A. Based on expert recommendations, the regulatory
considered too insignificant to merit risk quantifica-
agency could, for each category of effect, suggest
tion. In this case, the exotic GABCA can be considered
experiments, tests and/or models that could be
by the regulator highly unlikely to have a significant
used to assess long-term population effects and
adverse effect on the environment for all categories of
indirect ecosystem effects, as partially exempli-
effects. Otherwise, the risk(s) in which a category(ies)
fied in Table 5.
of effect(s) had a species identified as potentially
B. Long-term population effects could be evaluated
significantly affected need to be quantified to enable a
comparatively, i.e. compare with a similar non-
final regulatory decision. Data to quantify the
123
Table 5 Suggestions of experiments, tests and/or models to assess some indirect ecosystem effects for risk quantification in the
Definitive Assessment for exotic generalist arthropod biological control agents (GABCAs)
Indirect effect Test
Behavioral/evolutionary changes Laboratory tests (literature review), biology of GABCAs (or related surrogates)
Change in species richness and/or abundance/ Laboratory tests (e.g., competition)

evenness Computer modelling (e.g., food webs to identify species at risk)
Potential distribution maps (accounting for climate change)
Nutrient cycling Literature review on decomposers
Reduced insecticide/herbicide use Likelihood of establishment and impact on target sampling (pheromone/sticky
trap)
Compare to other releases
Area of origin comparison (food webs)
target species for which the population effect is close (same genus) to the non-target species, has
known from the same or similar exotic GABCA. functionally similar ecologies, uses the same habitat,
This may require the availability of a database of is a common species, is relevant to the receiving
known effects. environment, is well-studied taxon, is easy to rear. The
C. Long-term population effects could be evaluated use of surrogates must be determined on a case-by-
using models, involving intrinsic growth rates and case basis in consultation with the regulatory
density dependence (Barlow et al. 2004). This authority.
approach has the advantage of being able to
account for compensatory mortality. Closure of the ERA
D. Long-term population effects and indirect ecosys-
tem effects could be evaluated by an expert panel The requested data for risk quantification are submit-
using expert solicitation methods as has been ted to the regulatory agency. The regulatory agency
conducted in other ERAs (e.g., Harris et al. 1994). conducts the final risk characterization with technical
input from experts on a case-by-case basis, i.e. this
In cases where a valued species is at risk, additional time with no pre-established risk threshold. The
information may be provided, including: (1) a mitiga- regulatory agency will complete its assessment pro-
tion proposal to offset the adverse effects, (2) data to cess (e.g., after public comment or public hearing) and
show that the exotic GABCA does not suppress the determine an outcome which could be to: approve the
valued species, (3) comparative results to demonstrate petition, approve the petition with conditions, not
that the exotic GABCA has a lower effect on the approve/decline the petition or request further
valued species than the current methods or products on information.
the market, and/or (4) analyses to show that not using
the exotic GABCA has greater risk on the valued
species than using the exotic GABCA. An improved ERA?
The use of surrogate species might be appropriate
instead of the actual non-target species when: (1) non- The three-tiered ERA outlined in this paper could
target at risk species are threatened, endangered or rare support the regulatory system in many countries as it
endemic; (2) the experiments themselves would would facilitate a transparent, scientific approach to
jeopardize the species; (3) if the ERA is performed assessing the risks of exotic GABCAs on a case-by-
in the native range of the exotic GABCA; or (4) it is case basis. It expands on the scoping assessment
very difficult or impossible to test the actual non-target originally proposed by van Lenteren et al. (2003) and
species. The criteria for a species to be selected as a the two-tiered method recently published by EPPO
surrogate are all of the following: is taxonomically (2018), implementing major improvements. First, the
123
three tiers provide several opportunities to use some of these host/prey may not be considered
published information and expert judgement to significant enough to merit additional assessment.
provide upper bounds on the potential risks of This may occur because the species itself has low
introducing or commercially releasing an exotic social value or because the effect is considered
GABCA before additional data are needed, and focus insignificant, such as the effect of augmentative
these data collecting activities on the most critical releases of Trichogramma nubilale Ertle & Davis
issues. Our Scoping Assessment differs from the van 1975 (Hymenoptera: Trichogrammatidae) on the
Lenteren et al. (2003) and EPPO (2018) approach by endangered Karner blue butterfly (Lycaeides melissa
evaluating the possible existence of risks instead of samuelis (Nabokov 1944) (Lepidoptera: Lycaenidae),
likely risk. If risks might exist, then additional Andow et al. 1995) or T. brassicae Bezdenko 1968 on
assessment is required, while under the previous native Lepidoptera and the parasitoid Lydella thomp-
scoping assessments, risks must be judged to be likely soni Herting, 1959 (Diptera: Tachinidae) (Lynch et al.
to require additional assessment. The present Scoping 2001).
Assessment entails biosafety questions with greater Third, the methodology offers explicit methods to
specificity and rigor, and it results in a more risk- analyze the potential effects of an exotic GABCA on
averse determination than previous methods. It non-target species. In the Screening Assessment, these
includes all of the elements of the EPPO (2018) are based on the semi-quantitative analysis of expert
express EIA, except the EPPO (2018) explicitly judgements of the likelihood and magnitude of an
considers the balance between benefits and risks, and adverse effect via an interaction characterized by the
considers an exotic BCA highly unlikely to have a category of effect. For example, for an effect on a non-
significant adverse effect on the environment if the target species via exploitative competition, the likeli-
benefits are likely to ‘‘significantly’’ exceed the risks. hood is the probability that the exotic GABCA and
Our method allows consideration of a risk–benefit non-target species will occur together, and the mag-
balance, but does not specify how this balance would nitude is the size of the effect the exotic GABCA
affect the acceptability of risk. might have on the non-target assuming that they co-
Second, the present methodology uses categories of occur. In the species selection process of the Definitive
effects to guide the ERA instead of considering risks to Assessment, likelihood and magnitude are estimated
any non-target species. The 19 categories of effects quantitatively. The likelihood of co-occurrence is
(Table 2) are the documented ways an exotic BCA can separated into spatial overlap, temporal overlap,
affect the environment. They enable a comprehensive habitat overlap, niche overlap, and likelihood of
evaluation of potential environmental risks in the encounter in the niche. The magnitude of the effect
Scoping and Screening Assessments without requiring is the reduction in survival and/or reproduction of an
a detailed evaluation of specific risks to specific non- individual non-target species, if it co-occurs with the
target species. As most exotic BCAs will affect only a exotic BCA. By breaking down the adverse effect into
relatively small subset of these 19 categories, this its components, it is possible to construct an estimate
focuses subsequent steps in the ERA on issues of while at the same time reveal knowledge gaps.
greatest concern. It also delays identification of Fourth, it introduces in the Screening and Definitive
potential at risk species until much later in the ERA, Assessments an explicit reliance on the social valua-
avoiding some unneeded work to compile compre- tion of adverse effects. Although there is considerable
hensive species lists that might be affected (e.g., variation in how to value individual species, valuation
Kuhlmann et al. 2006) by focusing only on compiling is essentially a social process. That is, the social
lists of those species associated with the important significance of an adverse effect should be considered
categories of risk(s). This approach divides the process when determining the significance of an adverse
of non-target species identification developed by Todd effect. For example, a 20% long-term reduction in
et al. (2015) into two parts, one associated with the the population of some common non-target arthropod
categories of effects and another associated with the herbivore might not be considered socially significant
prioritization and identification of non-target species if the non-target will still be common and is not
within a category. Finally, this approach implies that endangered.
not all host/prey are considered at risk, as an effect on
123
As several countries and jurisdictions allow a References

comparison of risks and benefits of proposed intro-
ductions of exotic BCAs, methods to incorporate a Andow DA, Lane CP, Olson DM (1995) Use of Trichogramma
in maize-estimating environmental risks. In: Lynch JM,
formal assessment of benefits still needs to be Hokkanen HH (eds) Benefits and risks of introducing
developed. In our Scoping Assessment, risks may be biocontrol agents. Blackwell, London, pp 101–118
poorly characterized and highly uncertain compared to Andreassen LD, Kuhlmann U, Mason PG, Holliday NJ (2009)
benefits, so a simple risk–benefit comparison might Host range testing of a prospective classical biological
control agent against cabbage maggot, Delia radicum, in
not be appropriate, as it may favor the more certain Canada. Biol Control 48:210–220
benefits over the highly uncertain risks. In addition, Arthurs S, McKenzie CL, Chen J, Doğramaci M, Brennan M,
there may be trade-offs that would need to be taken Houben K, Osborne L (2009) Evaluation of Neoseiulus
into consideration where, for example, the economic cucumeris and Amblyseius swirskii (Acari: Phytoseiidae)
as biological control agents of chilli thrips, Scirtothrips
benefits are maybe substantial if the exotic BCA dorsalis (Thysanoptera: Thripidae) on pepper. Biol Control
provides significant pest control. In the Screening 49:91–96
Assessment, the semi-quantitative assessments of Barlow ND, Barratt BIP, Ferguson CM, Barron MC (2004)
benefits and risks may be difficult to compare. They Using models to estimate parasitoid impacts on non-target
host abundance. Environ Entomol 33:941–948
are based on ordinal scales that may not be directly Barratt BIP, Todd J, Malone LA (2016) Selecting non-target
comparable, so methods to weigh the benefits and risks species for arthropod biological control agent host range
are needed. testing: evaluation of a novel method. Biol Control
We recognize that this is a conceptual proposal for 93:84–92
Bigler F, Babendreier D, Kuhlmann U (2006) Environmental
an improved ERA for all exotic BCAs, including impact of invertebrates for biological control of arthro-
GABCAs. Although the expert panel agreed that it is pods: methods and risk assessment. CAB International,
based on sound scientific principles, its practicality for Wallingford
both regulators and applicants requires testing against Carmichael AC, Wharton RA, Clarke AR (2005) Opiine (Hy-
menoptera: Braconidae) parasitoids of tropical fruit flies
other species in novel environments. It would be (Diptera: Tephritidae) of the Australian and South Pacific
useful to conduct several case studies to reveal region. Bull Entomol Res 95:545–569
ambiguities and generate estimates of the cost of the Castañé C, Arnó J, Gabarra R, Alomar O (2011) Plant damage to
tiers and the time it takes to complete them. Further- vegetable crops by zoophytophagous mirid predators. Biol
Control 59:22–29
more, it is essential to involve the regulatory commu- Cédola C, Polack A (2011) First record of Amblyseius swirskii
nity and potential applicants, perhaps initially to (Acari: Phytoseiidae) from Argentina. Rev Soc Entomol
suggest relevant case studies, and thereafter to eval- Arg 70:375–378
uate the results. This proposed ERA for exotic Cock MJW, Murphy ST, Kairo MTK, Thompson E, Murphy RJ,
Francis AW (2016) Trends in the classical biological
GABCAs, and other BCAs, aims to enable evaluation control of insect pests by insects: an update of the BIOCAT
of their environmental safety, while focusing greater database. BioControl 61:349–363
efforts on their specific risks needing greater over- EPA (New Zealand) (2013) Application to import for release or
sight, rigorously and without prejudice. We hope these to release from containment new organisms, Macrolophus
pygmaeus, application APP201254, 2013 November 20.
efforts provide a pathway for the assessment and https://www.epa.govt.nz/assets/FileAPI/hsno-ar/
possible approval for the safe use of exotic GABCAs APP201254/APP201254-EPA-BCA-application-
and other BCAs. Macrolophus-2013-FINAL.pdf. Accessed 6 Feb 2019
EPA (New Zealand) (2014) EPA decision for application
Acknowledgements We are grateful for all the experts and APP201254, 2014 April 17. https://www.epa.govt.nz/
stakeholders worldwide who contributed to this work. We thank assets/FileAPI/hsno-ar/APP201254/APP201254-
Embrapa (grant number SEG 12.13.12.005.00.00), USDA APP201254-Decision-FINAL-2014-04-17.pdf. Accessed
(NIFA grant 2019-67013-29406) and the New Zealand Better 6 Feb 2019
Border Biosecurity Research Collaboration for their financial EPPO (European and Mediterranean Plant Protection Organi-
support. zation) (2018) PM 6/04 (1) decision-support scheme for
import and release of biological control agents of plant
Compliance with ethical standards pests. EPPO Bull 48:352–367
EPPO (2020) PM 6/3(4) List of biological control agents widely
Conflict of interest The authors declare that they have no used in the EPPO region—2020 version. https://www.
conflict of interest. eppo.int/media/uploaded_images/RESOURCES/eppo_
123
standards/pm6/pm6-03-2020-en.pdf. Accessed 12 Sept agents. The Secretariat of the North American Plant Pro-
2020 tection Organization, Ottawa, p 14
Harris HJ, Wegner RB, Harris VA, Devault DS (1994) A Nechols JR, Kauffman WC, Schaefer PW (1992) Significance of
method for assessing environmental risk: a case study of host specificity in classical biological control. In: Kauff-
Green Bay, Lake Michigan, USA. Environ Manag man WC, Nechols JR (eds) Selection criteria and ecolog-
18:295–306 ical consequences of importing natural enemies. Thomas
Holler TC, Gerónimo F, Sivinski J, Gonzalez JC, Stewart J Say Publications in Entomology, Entomological Society of
(1996) Mediterranean fruit fly parasitoid aerial release test America, Lanham, pp 41–52
in Guatemala. In: 28 WGFFWH meeting, Viña del Mar, O’Hagan A (2005) Research in elicitation. Department of
Chile, pp 74–75 Probability and Statistics, School of Mathematics,
Hopper KR (2001) Research needs concerning non-target University of Sheffield, Sheffield
impacts of biological control introductions. In: Wajnberg Olckers T (2003) Assessing the risks associated with the release
E, Scott JK, Quimby PC (eds) Evaluating indirect eco- of a flowerbud weevil, Anthonomus santacruzi, against the
logical effects of biological control. CAB International, invasive tree Solanum mauritianum in South Africa. Biol
Wallingford, pp 39–56 Control 28:302–312
Kade N, Gueye-Ndiaye A, Duverney C, Moraes GJ (2011) Ovruski S, Aluja M, Sivinski J, Wharton R (2000) Hyme-
Phytoseiid mites (Acari: Phytoseiidae) from Senegal. nopteran parasitoids on fruit-infesting Tephritidae (Dip-
Acarologia 51:133–138 tera) in Latin America and the southern United States:
Kuhlmann U, Schaffner U, Mason PG (2006) Selection of non- diversity, distribution, taxonomic status and their use in
target species for host specificity testing of entomophagous fruit fly biological control. J Integr Pest Manag 5:81–107
biological control agents. In: Second international sym- Paula DP, Linard B, Platt AC, Srivathsan A, Timmermans
posium on biological control of arthropods, Davos, MJTN, Sujii ER, Pires CSS, Souza LM, Andow DA,
Switzerland, pp 12–16 Vogler AP (2016) Uncovering trophic interactions in
Kynn M (2008) The ‘heuristics and biases’ bias in expert elic- arthropod predators through DNA shotgun-sequencing of
itation. J R Stat Soc A 171:239–264 gut contents. PLoS ONE 11(9):e0161841
Le Hesran S, Ras E, Wajnberg E, Beukeboom L (2019) Next Pedrycz W, Gomide F (2007) Fuzzy systems engineering:
generation biological control—an introduction. Entomol toward human-centric computing. Wiley, New York
Exp Appl 167:579–583 Petit JN, Hoddle MS, Grandgirard J, Roderick GK, Davies N
Lynch LD, Hokkanen HMT, Babendreier D, Bigler F, Burgio G, (2009) Successful spread of a biocontrol agent reveals a
Gao ZH, Kuske S, Loomans A, Menzler-Hokkanen I, biosecurity failure: elucidating long distance invasion
Thomas MB, Tommasini G, Waage JK, van Lenteren JC, pathways for Gonatocerus ashmeadi in French Polynesia.
Zeng QQ (2001) Insect biological control and non-target BioControl 54:485–495
effects: a European perspective. In: Wajnberg E, Scott JK, Puccia CJ, Levins R (2013) Qualitative modeling of complex
Quimby PC (eds) Evaluating indirect ecological effects of systems: an introduction to loop analysis and time aver-
biological control. CAB International, Wallingford, aging. Harvard University Press, Cambridge
pp 99–125 Rahman MM (2016) Multiplication and division of triangular
McGrath Z, MacDonald F, Walker G, Ward D (2020) A fuzzy numbers. DIU J Sci Technol 11(2):49–53
framework for predicting competition between native and Rousse P, Harris EJ, Quilici S (2005) Fopius arisanus, an egg-
exotic hymenopteran parasitoids of lepidopteran larvae pupal parasitoid of Tephritidae—an overview. Biocontrol
using taxonomic collections and species level traits. News Info 26:59–69
BioControl. https://doi.org/10.1007/s10526-020-10025-y Roy HE, Brown PM, Adriaens T, Berkvens N, Borges I, Clu-
McMurtry JA, De Moraes GJ, Sourassou NF (2013) Revision of sella-Trullas S, Comont RF, De Clercq P, Eschen R, Estoup
the lifestyles of phytoseiid mites (Acari: Phytoseiidae) and A, Evans EW, Facon B, Gardiner MM, Gil A, Grez AA,
implications for biological control strategies. Syst Appl Guillemaud T, Haelewaters D, Herz A, Honek A, Howe
Acarol 18:297–320 AG, Hui C, Hutchison WD, Kenis M, Koch RL, Kulfan J,
Messing R, Roitberg B, Brodeur J (2006) Measuring and pre- Handley LL, Lombaert E, Loomans A, Losey J, Lukashuk
dicting indirect impacts of biological control: competition, AO, Maes D, Magro A, Murray KM, Martin GS, Martin-
displacement and secondary interactions. In: Bigler F, kova Z, Minnaar IA, Nedved O, Orlova-Bienkowskaja MJ,
Babendreier D, Kuhlmann U (eds) Environmental impact Osawa N, Rabitsch W, Ravn HP, Rondoni G, Rorke SL,
of invertebrates for biological control of arthropods: Ryndevich SK, Saethre M-G, Sloggett JJ, Soares AO, Stals
methods and risk assessment. CAB International, R, Tinsley MC, Vandereycken A, van Wielink P, Viglá-
Wallingford, pp 64–77 šová S, Zach P, Zakharov IA, Zaviezo T, Zhao Z (2016)
Mitchell R, Agle B, Wood D (1997) Toward a theory of The harlequin ladybird, Harmonia axyridis: global per-
stakeholder identification and salience: defining the prin- spectives on invasion history and ecology. Biol Invasions
ciple of who and what really counts. Acad Manag Rev 18:997–1044
22:853–886 Sackman H (1974) Delphi assessment: expert opinion, fore-
Morgan MG (2014) Use (and abuse) of expert elicitation in casting, and group process (no. RAND-R-1283-PR). Rand
support of decision making for public policy. PNAS Corp, Santa Monica
111:7176–7184 Sands DPA, van Driesche RG (2004) Using the scientific liter-
NAPPO (2015) RSPM 12 Guidelines for petition for first release ature to estimate the host range of a biological control
of non-indigenous entomophagous biological control agent. In: van Driesche RG, Sands DPA (eds) Assessing
123
host ranges of parasitoids and predators used for classical van Lenteren JC, Bueno VH, Colmenarez YC, Luna MG (2020)
biological control: a guide to best practice. Forest Health Biological control in Latin America and the Caribbean: its
Technology Enterprise Team, Forest Service, USDA, rich history and bright future. CAB International,
Morgantown, pp 15–23 Wallingford, p 522
Sato Y, Mochizuki A (2011) Risk assessment of non-target Vanderhoeven S, Branquart E, Casaer J, Dhondt B, Hulme PE,
effects caused by releasing two exotic phytoseiid mites in Shwartz A, Strubbe D, Turbé A, Verreycken H, Adriaens T
Japan: can an indigenous phytoseiid mite become IG prey? (2017) Beyond protocols: improving the reliability of
Exp Appl Acarol 54:319–329 expert-based risk analysis underpinning invasive species
Snyder WE, Evans EW (2006) Ecological effects of invasive policies. Biol Invasions 19:2507–2517
arthropod generalist predators. Annu Rev Ecol Evol Syst Vargas RI, Stark JD, Uchida GK, Purcell M (1993) Opiine
37:95–122 parasitoids (Hymenoptera: Braconidae) of Oriental fruit fly
Stahl J, Tortorici F, Pontini M, Bon M-C, Hoelmer K, Marazzi (Diptera: Tephritidae) on Kauai Island, Hawaii: Islandwide
C, Tavella L, Haye T (2019) First discovery of adventive relative abundance and parasitism rates in wild and orchard
populations of Trissolcus japonicus in Europe. J Pest Sci guava habitats. Environ Entomol 22:246–253
92:371–379 Vargas RI, Leblanc L, Putoa R, Eitam A (2007) Impact of
Todd JH, Barratt BIP, Tooman L, Beggs JR, Malone LA (2015) introduction of Bactrocera dorsalis (Diptera: Tephritidae)
Selecting non-target species for risk assessment of ento- and classical biological control releases of Fopius arisanus
mophagous biological control agents: evaluation of the (Hymenoptera: Braconidae) on economically important
PRONTI decision-support tool. Biol Control 80:77–88 fruit flies in French Polynesia. J Econ Entomol
van Lenteren JC, Babendreier D, Bigler F, Burgio G, Hokkanen 100:670–679
HMT, Kuske S, Loomans AJM, Menzler-Hokkanen I, van Wang XG, Bokonon-Ganta AH, Ramadan MM, Messing RH
Rijn PCJ, Thomas MB, Tommasini MG, Zeng QQ (2003) (2004) Egg-larval opiine parasitoids (Hym., Braconidae) of
Environmental risk assessment of exotic natural enemies tephritid fruit fly pests do not attack the flowerhead-feeder
used in inundative biological control. BioControl 48:3–38 Trupanea dubautiae (Dipt., Tephritidae). J Appl Entomol
van Lenteren JC, Bale J, Bigler F, Hokkanen HMT, Loomans 12:716–722
AJM (2006) Assessing risks of releasing exotic biological Waterhouse DF (1993) Pest fruit flies in the Oceanic Pacific. In:
control agents of arthropod pests. Annu Rev Entomol Waterhouse DF (ed) Biological control: Pacific pro-
51:609–634 spects—supplement 2. Australian Centre for International
van Lenteren JC (2012) The state of commercial augmentative Agricultural Research, Canberra, pp 4–47
biological control: plenty of natural enemies, but a frus- Wyckhuys KA, Lu Y, Morales H, Vazquez LL, Legaspi JC,
trating lack of uptake. BioControl 57:1–20 Eliopoulos PA, Hernandez LM (2013) Current status and
van Lenteren JC, Bolckmans K, Köhl J, Ravensberg WJ, potential of conservation biological control for agriculture
Urbaneja A (2018) Biological control using invertebrates in the developing world. Biol Control 65:152–167
and microorganisms: plenty of new opportunities.
BioControl 63:39–59
123

Paula2021 Article IntegratingAdverseEffectAnalys

Uploaded by

Copyright:

Available Formats

Paula2021 Article IntegratingAdverseEffectAnalys

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Paula2021 Article IntegratingAdverseEffectAnalys

Uploaded by

Copyright:

Available Formats

BioControl (2021) 66:113–139

Integrating adverse effect analysis into environmental risk

D. P. Paula (&) E. M. G. Fontes C. S. S. Pires B. I. P. Barratt J. H. Todd

Species identification 4–10, 11–30 and [ 30 species) or taxonomic related-

Status of establishment of exotic GABCA establishment of an augmentative exotic

3b. Release of undesired plant (weed) 0 0 0 Harmonia as 0 Unknown Unknown

Categories of effects on non-target species Magnitude of effect (MEfi)

the magnitude of the effect should be based on the size

1. Reduction of a native top 0 0 0 0

5. Increase in herbivory via

analysis in the Definitive Assessment, the H. axyridis

Likelihood of Encounter (LEnj)

Rare endemic species

risk(s) associated with the non-target species should

Specifying the mechanistic pathway(s) of adverse

To enable quantitative evaluation of the risk to the

Change in species richness and/or abundance/ Laboratory tests (e.g., competition)

Nutrient cycling Literature review on decomposers

As several countries and jurisdictions allow a References

You might also like