Options for reducing uncertainty in impact classification for alien species
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Options for reducing uncertainty in impact classification
for alien species
DAVID A. CLARKE ,1 DAVID J. PALMER,1 CHRIS MCGRANNACHAN,1 TREENA I. BURGESS,2
STEVEN L. CHOWN ,1 ROHAN H. CLARKE,1 SABRINA KUMSCHICK,3,4 LORI LACH ,5
ANDREW M. LIEBHOLD ,6,7 HELEN E. ROY,8 MANU E. SAUNDERS ,9,10 DAVID K. YEATES,11
MYRON P. ZALUCKI,12 AND MELODIE A. MCGEOCH 1,13,
1
School of Biological Sciences, Monash University, Clayton, Victoria 3800 Australia
2
Centre for Climate Impacted Terrestrial Ecosystems, Harry Butler Institute, Murdoch University, 90 South Street,
Murdoch 6150 Australia
3
Centre for Invasion Biology, Department of Botany & Zoology, Stellenbosch University, Matieland, South Africa
4
Cape Town Office, South African National Biodiversity Institute, Claremont, South Africa
5
College of Science and Engineering, James Cook University, PO Box 6811, Cairns, Queensland 4870 Australia
6
USDA Forest Service Northern Research Station, Morgantown, West Virginia 26505 USA
7
Faculty of Forestry and Wood Sciences, Czech University of Life Sciences, Praha 6 - Suchdol, CZ 165 21 Czech Republic
8
UK Centre for Ecology & Hydrology, Wallingford OX10 8BB UK
9
School of Environmental and Rural Science, University of New England, Armidale, New South Wales 2351 Australia
10
UNE Business School, University of New England, Armidale, New South Wales 2351 Australia
11
CSIRO Australian National Insect Collection, PO Box 1700, Canberra, Australian Capital Territory 2601 Australia
12
School of Biological Sciences, University of Queensland, Brisbane, Queensland 4072 Australia
13
Department of Ecology, Environment and Evolution, La Trobe University, Bundoora, Melbourne, Victoria 30186 Australia
Citation: Clarke, D. A., D. J. Palmer, C. McGrannachan, T. I. Burgess, S. L. Chown, R. H. Clarke, S. Kumschick, L. Lach,
A. M. Liebhold, H. E. Roy, M. E. Saunders, D. K. Yeates, M. P. Zalucki, and M. A. McGeoch. 2021. Options for reducing
uncertainty in impact classification for alien species. Ecosphere 12(4):e03461. 10.1002/ecs2.3461
Abstract. Impact assessment is an important and cost-effective tool for assisting in the identification and
prioritization of invasive alien species. With the number of alien and invasive alien species expected to
increase, reliance on impact assessment tools for the identification of species that pose the greatest threats
will continue to grow. Given the importance of such assessments for management and resource allocation,
it is critical to understand the uncertainty involved and what effect this may have on the outcome. Using
an uncertainty typology and insects as a model taxon, we identified and classified the causes and types of
uncertainty when performing impact assessments on alien species. We assessed 100 alien insect species
across two rounds of assessments with each species independently assessed by two assessors. Agreement
between assessors was relatively low for all three impact classification components (mechanism, severity,
and confidence) after the first round of assessments. For the second round, we revised guidelines and gave
assessors access to each other’s assessments which improved agreement by between 20% and 30% for
impact mechanism, severity, and confidence. Of the 12 potential reasons for assessment discrepancies iden-
tified a priori, 11 were found to occur. The most frequent causes (and types) of uncertainty (i.e., differences
between assessment outcomes for the same species) were as follows: incomplete information searches (sys-
tematic error), unclear mechanism and/or extent of impact (subjective judgment due to a lack of knowl-
edge), and limitations of the assessment framework (context dependence). In response to these findings,
we identify actions that may reduce uncertainty in the impact assessment process, particularly for assess-
ing speciose taxa with diverse life histories such as Insects. Evidence of environmental impact was avail-
able for most insect species, and (of the non-random original subset of species assessed) 14 of those with
evidence were identified as high impact species (with either major or massive impact). Although uncer-
tainty in risk assessment, including impact assessments, can never be eliminated, identifying, and commu-
nicating its cause and variety is a first step toward its reduction and a more reliable assessment outcome,
regardless of the taxa being assessed.
v www.esajournals.org 1 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Key words: EICAT; impact assessment; invasive alien species; mechanisms of impact; reducing uncertainty; risk
communication.
Received 23 September 2020; revised 30 November 2020; accepted 10 December 2020; final version received 27 January
2021. Corresponding Editor: Debra P. C. Peters.
Copyright: © 2021 The Authors. This is an open access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
E-mail: M.McGeoch@latrobe.edu.au
INTRODUCTION assigning relative impact severities to alien spe-
cies (Turbe et al. 2017, Gonzalez-Moreno et al.
With a changing climate and growing interna- 2019, Vila et al. 2019), producing evidence-based
tional trade, the range expansion of alien and data that can then be used in risk assessment and
invasive alien species (i.e., invasive alien species prioritization. Semi-quantitative protocols are a
being those that have a negative impact within valuable and necessary tool in contexts, such as
their recipient environment) is predicted to con- invasion biology, where decisions must consider
tinue (Seebens et al. 2017). Ongoing arrival and multiple lines of evidence, be made transparently
establishment of alien species necessitates a and where available evidence is incomplete (Gre-
triage approach to their management, where the gory et al. 2012). One such tool, Environmental
most damaging, or those most likely to cause Impact Classification for Alien Taxa (EICAT), is a
damage, is allocated high priority for surveil- formal protocol for using peer-reviewed litera-
lance and control (McGeoch et al. 2016). To this ture to assess any environmental impacts (i.e.,
end, research on the impacts of alien species has negative effects on the native environment) that
been growing steadily (Crystal-Ornelas and an alien species has inflicted and with which an
Lockwood 2020), accompanied by the develop- impact severity and mechanism is assigned
ment of various assessment frameworks for clas- (Hawkins et al. 2015). However, few studies have
sifying their impacts to assist in risk assessments yet explored the potential types and causes of
(Roy et al. 2018, Gonz alez-Moreno et al. 2019, uncertainty associated with semi-quantitative
Vila et al. 2019). tool outcomes for alien species impacts, and how
Risk assessments are generally performed using these could affect the final classification outcome
multiple data types that vary in reliability, and (but see Kumschick et al. 2017a, who compared
therefore, acknowledging, accounting, and com- results from two different applications of EICAT
municating associated uncertainty is an essential for alien amphibians). Better understanding of
component of the process (Harwood and Stokes the causes of uncertainty in impact classification
2003). For example, assessments by the Intergov- is an essential step in testing these protocols and
ernmental Panel on Climate Change (IPCC), that identifying general solutions to strengthen their
use multiple lines of evidence, are conducted reliability and value to risk analysis (Milner-Gul-
using a formal and agreed upon treatment of land and Shea 2017, Rueda-Cediel et al. 2018,
uncertainty. This method assesses the type, Latombe et al. 2019).
amount, quality, and consistency of evidence, While uncertainty cannot be removed entirely
along with the level of agreement, to assign confi- from semi-quantitative methods such as EICAT,
dence in the form of a likelihood scale (Mastran- identifying which causes of uncertainty are redu-
drea et al. 2011). Impact assessments of alien cible (and acknowledging those that are practi-
species, which contribute to full risk assessments, cally irreducible; Regan et al. 2002) is an essential
likewise should be accompanied by estimates of step in improving the reliability and value of the
uncertainty (Roy et al. 2018), although this does information they generate. Given the inherent
not always occur (Caton et al. 2018). nature of uncertainty in ecology more broadly,
Semi-quantitative decision protocols are now a and the need to account for it, frameworks for
widely used and accepted approach for identifying and classifying various types of
v www.esajournals.org 2 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
uncertainty have been found to be useful (Mil- recommendations for how best to communicate
ner-Gulland and Shea 2017). For example, the and mitigate general types of uncertainty associ-
uncertainty typology developed by Regan et al. ated with semi-quantitative methods for classify-
(2002) has been used to assess uncertainty in con- ing the severity of alien species impacts.
servation decision making in the face of climate
change (Kujala et al. 2013) and to assess uncer- METHODS
tainty in the alien species listing process and
what effect it may have on estimating the iden- Several tools and methods exist in support of
tity and number of such species in countries risk assessment to ensure that the process and its
(McGeoch et al. 2012). In both cases, application outcomes are as transparent and as evidence
of the typology helped to identify tactics for based as possible. One of these tools is Environ-
improving the transparency and repeatability of mental Impact Classification of Alien Taxa
environmental decisions. Regan et al. (2002) (EICAT). EICAT and other impact assessment
identify two broad types of uncertainty, with methods for IAS are not risk assessments them-
multiple categories under each, that is, (1) epis- selves, but rather tools used in support of evi-
temic uncertainty, or that brought about by dence-based RA’s that encompass a range of
uncertainty in determinate facts, and (2) linguis- other socio-economic and context-specific infor-
tic uncertainty, or that caused by the inherent mation (Hawkins et al. 2015, Gonzalez-Moreno
variation in our use of language (Regan et al. et al. 2019). Our intention here is to evaluate the
2002, Burgman 2005, Latombe et al. 2019). Lin- types of uncertainty associated with classifying
guistic uncertainty has been a significant hin- species based on evidence of their impact. This
drance to progress in invasion biology to date exercise is therefore intended as a contribution to
(Roy et al. 2018, Vil
a et al. 2019). The uncertainty improve the robustness of impact assessment
framework of Regan et al. (2002) provides a tools. We do so by (1) selecting 100 insect species
widely relevant framework for evaluating uncer- to assess, (2) independently applying the EICAT
tainty (McGeoch et al. 2012, Kujala et al. 2013). protocol to each species twice, (3) describing the
Here, we use this framework to identify and differences found between assessments for each
classify the types of uncertainty and their causes species and the reasons for these, (4) using
that may occur when assessing the environmen- Regan’s et al. (2002) uncertainty framework to
tal impacts of alien species. That is, we provide a assign each difference to an uncertainty type and
practical, rather than theoretical, approach to analyze their relative frequency, and (5) discuss
addressing uncertainty in impact assessments. ways in which each uncertainty type could be
Using insects as a case study, we apply the pub- reduced by refining structured decision protocols
lished protocol, EICAT, for classifying the for alien species impacts.
impacts of alien species (Blackburn et al. 2014,
Hawkins et al. 2015). Insects were selected as a Species pool and those assessed
model taxon for multiple reasons: (1) Alien inva- The subset of alien insect species assessed
sive insect species pose significant risks to the was deliberately selected to be likely to encom-
environment and economy and require impact pass as many species as possible with adequate
risk assessments to prioritize management (Brad- peer-reviewed evidence of impact, given the
shaw et al. 2016, Lovett et al. 2016, Suckling et al. demonstrated paucity of impact evidence for
2019); (2) they are one of the most species-rich, alien species (Crystal-Ornelas and Lockwood
abundant, functionally diverse taxonomic groups 2020). The initial set (~2800 species) consisted
(Stork et al. 2015, Noriega et al. 2018), and (3) of all insect species present in the Global Regis-
they are a group to which EICAT has not ter for Introduced and Invasive Species
previously been applied. We use differences (GRIIS), which provides verified species by
found between independent assessments of 100 country occurrence records outside their native
insect species to identify, interpret, and classify range at country scale and an attribute for
causes and types of uncertainty. Incorporating local evidence of impact (Pagad et al. 2018; as
the lessons learnt from performing impact classi- at June 2017). In addition, alien insects known
fication on alien insects, we conclude with to impact the environment were extracted from
v www.esajournals.org 3 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
relevant reviews (Kenis et al. 2009, Vaes-Petig- Database (Blackburn et al. 2014, Hawkins et al.
nat and Nentwig 2014, McGeoch et al. 2015, 2015).
Cameron et al. 2016, Bertelsmeier et al. 2017, EICAT assessments have, to date, been pub-
Evans et al. 2017). Species were then ranked by lished for alien birds (Evans et al. 2016), amphib-
the number of times they were referred to ians (Kumschick et al. 2017a), selected
across all sources, that is, multiple sources des- gastropods (Kesner and Kumschick 2018), mam-
ignated them as invasive or having a negative mals (Hagen and Kumschick 2018) and bamboo
environmental impact. The top 100 species species (Canavan et al. 2019). Here, we used the
were selected for assessment in this way prior EICAT protocol as published (Hawkins et al.
to the literature search for each species. Given 2015), although these guidelines have subse-
the assumed positive relationship between pest quently been revised by the IUCN via an online
severity and available literature on the species, consultation process and the involvement of one
we consider this suite of 100 species to incor- of the co-authors (SK; IUCN 2020a). Our interest
porate a best-case scenario for evidence avail- here was to identify forms of uncertainty in the
able on which to implement EICAT and impact assessment process, rather than to evalu-
interpret the results accordingly. ate EICAT per se, which is similar to a range of
other semi-quantitative impact assessment
Structured decision protocol protocols (Kumschick et al. 2017b). We adopted
EICAT is a protocol with published guidelines previously suggested modifications to the mech-
for its implementation to ensure that it is applied anisms used to assess the environmental impacts
in a consistent, transparent, and comparable way of alien insects. Specifically, (1) the mechanism
across taxa and between assessors (Blackburn grazing/herbivory/browsing was altered to her-
et al. 2014, Hawkins et al. 2015). EICAT has now bivory as grazing and browsing are not relevant
been adopted by the IUCN as a standard for the for insects, and (2) the addition of a relevant
purpose of assigning severity of impact cate- mechanism, facilitation of native species
gories and impact mechanisms to IAS, with (whereby an invasive species negatively affects
strong parallels to the method used for the IUCN one native species by positively affecting another
Red List of Threatened Species. The EICAT sever- native species) was considered important for
ity of negative impacts is classified as minimal capturing this frequently encountered form of
concern (MC), minor (MN), moderate (MO), indirect impact of invasive insects on biodiver-
major (MR), or massive (MV; Hawkins et al. sity (McGeoch et al. 2015).
2015). The final severity attributed to an alien
species at the end of the assessment process is its Implementation of EICAT
maximum realized impact anywhere within its The general steps involved following the pre-
introduced range; that is, EICAT is only con- assessment and assessment steps of the EICAT
cerned with the most severe realized (as opposed protocol as a research exercise, using the detailed
to potential) impact for a species. Two other cate- guidelines, decision charts, and definitions pub-
gories are possible, data deficient (DD) for cases lished in Hawkins et al. (2015). We discussed and
where there is insufficient evidence to perform conducted the research (including the assess-
an assessment, or not alien (NA) for when a ments unless otherwise specified below) and the
given species does not in fact have an alien distri- first and senior authors analyzed and reported
bution. Mechanisms of impact are ascribed for the results back to the authorship group for dis-
each assessment where impact evidence was cussion and interpretation.
found. For example, a given species may have a To measure the level of congruence between
negative impact by preying upon native species assessment outcomes and to identify reasons for
and would, therefore, be ascribed the mechanism the differences that occurred, each of the 100
of predation for that specific piece of evidence. alien insect species was assessed independently,
These mechanisms, of which there were 13 using peer-reviewed literature, for environmen-
(Blackburn et al. 2014, Hawkins et al. 2015), tal impacts by two people. The ECAT guidelines
expand on those identified by Kumschick et al. specify that assessments may be conducted by
(2012) to align with the Global Invasive Species individuals or groups, in person or via email
v www.esajournals.org 4 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
(Hawkins et al. 2015). Our rationale for individu- departing from these only where it was neces-
ally conducting the assessments was that by sary to clarify, elaborate, or modify details for the
doing so we were able to use differences found purpose of reducing uncertainty. The purpose of
between independent assessments to identify revisions to the guidelines was to clarify
where and why uncertainty arises in the impact particular protocol points that were identified as
assessment process. Comparing individual potentially resulting in differences between
assessments rather than group assessments also assessments, particularly to reduce the more
allowed us to complete assessments for many common and readily addressed misunderstand-
more species than would otherwise have been ings that arose in the first-round assessments
possible. Here, the rationale was that including (such as the need to first confirm evidence of the
many species, representing a broad range of life existence of alien populations of each species).
histories and available information, would likely Fourth, the clarified guidelines, along with the
yield a broader array of potential reasons for dif- initial results from both assessors for each
ferences found between assessments of the same species, were then used by the assessors in the
species. By controlling the assessments and their second assessment round. Differences can occur
revisions using the process outlined below, we for one or more outcome of each assessment, that
were able to exclude as far as possible potential is, differences in allocated impact mechanism,
influence of one assessor over the judgment of severity, confidence level, or any combination of
another, as well as maximize the comparability/ the three. Fifth, assessors could then either refine
uniformity of the process across each of the spe- their initial assessment considering this new
cies assessed. This suited the purpose of this information or leave their assessment as origi-
study, which was to identify types and sources of nally provided (Burgman et al. 2011). Either deci-
uncertainty. sion required justification. The sixth and final
The entire process can be generalized by the step required assessors to provide possible rea-
following steps. First, assessors discussed the sons for why, in their view, initial assessments
protocol and performed a series of pilot assess- resulted in differences.
ments as a group to familiarize everyone with
the EICAT process and protocol guidelines Application of the uncertainty framework
(Hawkins et al. 2015). Second, impact assess- We used the information generated from the
ments were performed independently by two above process to understand and classify the
assessors for each species. Literature searches assessment uncertainty, by treating the differ-
were performed first by using ISI Web of Science ences in assessment outcomes across assessors as
(webofknowledge.com) using the species name types of uncertainty and classifying them using
(scientific including synonyms) as the search the uncertainty typology of Regan et al. (2002).
terms, after which assessors used supplementary Using the detailed definitions of types of uncer-
searches and sources as needed. Each species tainty provided by Regan et al. (2002), causes of
assessment resulted in three main outputs with uncertainty were first broadly classified as either
supporting literature evidence: (1) the mecha- epistemic or linguistic in origin. Epistemic uncer-
nisms of impact (i.e., how the alien species is tainty is that associated with the knowledge of a
having a negative effect, e.g., via competition), system’s state and consists of six main types
(2) the size of the impact (the EICAT magnitude (measurement error, systematic error, natural
of impact severity category), and (3) the confi- variation, inherent randomness, model uncer-
dence in the evidence supporting the chosen tainty, subjective judgment; Regan et al. 2002).
mechanism and impact severity (Hawkins et al. There are five main types of linguistic uncer-
2015). Third, based on the outcome of the first tainty associated with the omnipresent variation
round where differences in outcomes occurred in our use of language (vagueness, context
between assessors, the guidelines were clarified dependence, ambiguity, theoretical indetermi-
where points of misunderstanding or differences nacy, under-specificity; Regan et al. 2002). Other
in understanding arose. Our intention here was studies of uncertainty using this typology have
to apply EICAT critically, adhering to the guideli- modified it by, for example, including a third
nes as far as possible and elaborating on or broad type of uncertainty (human decision
v www.esajournals.org 5 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
uncertainty) which contain causes of uncertainty 100
such as subjective judgment (Kujala et al. 2013). Assessment
round
We chose to follow McGeoch et al. (2012) who One
Assessor agreement (%)
Two
75
used the original typology with minor adjust-
ments for application to listing of invasive alien
species, distinguishing between subjective judg- 50
ment per se and subjective judgment due to a
lack of knowledge, and between systematic error
per se and systematic error due to a lack of 25
knowledge. The rationale for this is akin to the
dual pathway of forming a subjective probability
(Burgman 2005), where people’s subjectivity 0
Mechanism Severity Confidence
stems from either a lack of, or despite, available Component
knowledge. Instances of both epistemic and lin-
guistic uncertainty were in some cases simultane- Fig. 1. Agreement in environmental impact assess-
ously possible as the reason for differences ment outcomes between two independent assessors
between assessment outcomes (McGeoch et al. across two rounds of assessment. Agreement increased
2012). Therefore, for each species where there across all assessment components in the second round.
was a difference between assessors regarding The assessment of severity of impact improved most
mechanism and/or severity of impact attribution, and mechanism of impact least across the two assess-
the reasons and their associated uncertainty type ment rounds.
were identified using the definitions in Regan
et al. (2002) and McGeoch et al. (2012). We did
not assess the uncertainty associated with differ- most frequently identified (Table 1), and agree-
ences in the chosen confidence level due to its ment on herbivory as the primary mechanism of
strong dependence on the chosen mechanism impact occurred in 38% (n = 14) of instances.
and severity. We did, however, assess the correla- The remaining eight mechanisms were less fre-
tion between the amount of disagreement in quently attributed (Table 1). For example, para-
impact severity and the mean level of confidence sitism was attributed to six species, five of which
for a given species (Appendix S1: Fig. S1). There were cases of agreement. Facilitation of native
was no correlation between the two variables in species was also attributed to six species; how-
the first or second round (Appendix S1: Fig. S1). ever, all six were cases that differed between
An assignment of low confidence does not neces- assessors. Multiple (n = 14) instances occurred
sarily lead to disagreement between assessors where the assessors agreed on a mechanism for a
and therefore was not used as a proxy for asses- species, but one assessor also assigned additional
sor disagreement. mechanisms. The assignment of multiple mecha-
nisms in EICAT occurs when the same level of
RESULTS impact severity is assigned to more than a single
mechanism by an assessor for a given species.
First-round assessments led to agreement Taking a conservative approach here, these cases
levels of between 32% and 44% for each of the were considered differences in assessment out-
three EICAT assessment components, that is, come. After the second round of assessments,
mechanism, severity of impact, and confidence agreement on the mechanism of impact
(Fig. 1). Mechanism of impact had the highest increased from 44% to 65% (Fig. 1). Little change
level of agreement, with 44% of species attribu- between rounds occurred for the allocation of
ted with the same primary mechanism of impact mechanisms of impact for most mechanisms
by both independent assessors. The level of except for herbivory (37–53%), transmission of
agreement, however, varied across mechanisms, disease (33–100%), and other (6–40%; Table 1).
as did the frequency with which each mechanism Categories of impact severity varied in their
was attributed (Table 1; Appendix S1: Fig. S2). levels of agreement (Fig. 2A). Data deficient
Herbivory, competition, and predation were (DD) was the impact severity classification with
v www.esajournals.org 6 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Table 1. Instances of agreement and disagreement between assessors for mechanism of impact for both assess-
ment rounds.
Round 1 Round 2
Mechanisms of impact Agree Disagree Total Agree Disagree Total
Herbivory 14 23 37 17 15 32
Competition 22 21 43 17 19 36
Predation 11 13 24 9 12 21
Other 1 14 15 6 9 15
Transmission of disease 3 6 9 7 0 7
Parasitism 5 1 6 5 0 5
Interaction with other aliens 2 8 10 1 6 7
Facilitation of natives 0 6 6 0 6 6
Hybridization 2 0 2 2 0 2
Chemical/physical/structural ecosystem impacts 0 1 1 0 1 1
None 5 14 19 5 7 9
Note: As more than one mechanism can be attributed to a species, the instance totals within and between assessment rounds
do not sum to 100.
A B
R
R
M
M
M
O
MO
MV MV
NA NA
DD
DD
MN
MN
MC
MC
Fig. 2. Chord diagrams showing the variation in agreement or disagreement in the assessment of severity of
impact between independent assessors in each assessment round (A, B). The segment span width of a given
impact severity category (MV, massive; MR, major; MO, moderate; MN, minor; MC, minimal concern; DD, data
deficient; and NA, not alien) is proportional to the number of species for which that severity was assigned by at
least one assessor. The width of the links connecting segments reflects the number of species for which there was
difference between assessors involving those specific impact severities. The low density of links in B compared
with A reflects the increase in agreement across assessments, revealing also moderate followed by major and
data deficient as the most frequently assigned categories.
highest agreement (see Appendix S1: Table S1 Assessor agreement on impact severity increased
for full details of impact severity assessments). from 34% to 70% across rounds. As in the first
Agreement on impact severity in round 1 round of assessments, variation in levels of
occurred for 34% of species; that is, a species was agreement among severity of impact categories
assigned the same severity of impact by both remained (Appendix S1: Table S1). Similarly, the
assessors (assessment outcomes were as likely to distribution of agreement across the severity of
agree as to differ, v2 = 1.37, df = 1, P = 0.24). impact categories differed little between
v www.esajournals.org 7 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Anoplolepsis gracilipes
Pheidole megacephala
Apis mellifera
Adelges tsugae
Xyleborus glabratus
Aphis gossypi
Bombus terrestris
Paratrechina longicornis
Wasmannia auropunctata
Linepithema humile
Lymantria dispar
Lysiphlebus testaceipes
Merizodus soledadinus
Myrmica rubra
Pachycondyla chinensis
Solenopsis invicta
Vespula vulgaris
Andricus quercuscalicis
Anoplophora chinensis
Anoplophora glabripennis
Aulacaspis yasumatsui
Cactoblastis cactorum
Diaphorina citri
Homalodisca vitripennis
Orthotomicus erosus
Paratrechina fulva
Solenopsis richteri
Technomyrmex albipes
Tetropium fuscum
Apis cerana
Quadrastichus erythrinae
Bessa remota
Apis mellifera scutellata
Calliphora vicina
Cicindelidia trifasciata
Cinara cupressi
Danaus plexippus
Severity
Harmonia axyridis
Icerya purchasi MV
Larinus planus
Lasius neglectus MR
Ochlerotatus japonicus
Pieris rapae
Pteromalus puparum MO
Pterostichus melanarius
Solenopsis geminata MN
Tapinoma melanocephalum
Torymus sinensis MC
Species
Trechisibus antarcticus
Trechus obtusus
Trichopoda pilipes
DD
Urophora affinis NA
Urophora quadrifasciata
Vespa velutina
Vespula pensylvanica
Zizina labradus
Bactrocera tryoni
Chilo partellus
Confidence
Polistes dominula
Scyphophorus acupunctatus NA
Thaumetopoea processionea
Trissolcus basalis Low
Vespula germanica
Aedes albopictus
Anopheles quadrimaculatus Medium
Forficula auricularia
Microctonus aethiopoides High
Monomorium destructor
Monomorium pharaonis
Myzus ascalonicus
Solenopsis papuana
Tuta absoluta
Xylosandrus compactus
Chaetosiphon fragaefolii
Polistes chinensis
Rhyzopertha dominica
Acanthoscelides obtectus
Bruchus rufimanus
Drosophila subobscura
Rhopalosiphum maidis
Sitophilus oryzae
Sitotroga cerealella
Bruchus pisorum
Leptinotarsa decemlineata
Liriomyza huidobrensis
Megachile rotundata
Trogoderma granarium
Andricus corruptrix
Andricus kollari
Andricus lignicola
Anthonomus grandis
Aspidiotus nerii
Ceratitis capitata
Penicillaria jocosatrix
Stictocephala bisonia
Heteronychus arator
Chromaphis juglandicola
Callosobruchus chinensis
Propylea quatuordecimpunctata
Saissetia oleae
Megachile apicalis
NA
MV MR MO MN MC DD NA
Severity
Fig. 3. Final assessment results of impact severity and the associated confidence rating for all 100 insect
v www.esajournals.org 8 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
(Fig. 3. Continued)
species. Species for which there remained difference after both rounds of assessment are shown with two differ-
ent colored tiles in the row associated with them. Most instances of difference occurred at the interface between
major and moderate, and between minor or minimal concern and data deficient. Category abbreviations are mas-
sive, MV; major, MR; moderate, MO; minor, MN; minimal concern, MC; data deficient, DD; not alien, NA. For
additional information on species identities, see Appendix S2.
assessment rounds (Appendix S1: Table S1). uncertainty type that occurred most frequently,
Assessors were more likely to agree than dis- followed by subjective judgment as a result of
agree (v2 = 83.64, df = 1, P < 0.001), and most lack of knowledge and context dependence
differences (22/30) between impact severity cate- (Fig. 4).
gories in the second round involved categories Final assessments (after the completion of both
adjacent on the severity scale, as opposed to the rounds) revealed evidence of environmental
first round which had much less structure impact was inadequate or unavailable for 14 spe-
(Fig. 2B, Appendix S1: Fig. S2). For example, in cies (Fig. 3, Appendix S2: Table S1), that is,
the second round, if there was difference in a where both assessors agreed on the species being
Moderate impact, it involved one assessor data deficient. Including the additional 10 spe-
assigning either a Minor (one category lower) or cies where assessors disagreed on data availabil-
Major (one category higher) impact (Fig. 2B, ity equates to approximately 25% of alien insects
Fig. 3). Exceptions to this occurred for two spe- assessed as data deficient by at least one assessor.
cies (Diaphora citri, Tetropium fuscum) in which These species were distributed across five orders:
one assessor assessed severity as Minor and the Coleoptera (n = 9), Diptera (n = 2), Hemiptera
other assessed it as Major, as well as for six spe- (n = 5), Hymenoptera (n = 7), and Lepidoptera
cies (Orthotomicus erosus, Solenopsis richteri, Chae- (n = 1). Of those species for which environmen-
tosiphon fragaefolii, Polistes chinensis, Rhyzopertha tal impact information was available and asses-
dominica, and Thaumetopoea processionea) in which sors agreed on impact severity (n = 56), 25 were
one assessor assigned an impact severity and the Hymenoptera. This order was most frequently
other thought there was insufficient data to do so represented in each impact severity category,
or that there was no evidence of alien popula- except for Minimal Concern where there were no
tions in the case of T. processionea (Fig. 3). Hymenoptera. It was also the only order for
Eleven possible causes of uncertainty (Table 2) which there was agreement on a Massive impact,
identified a priori occurred across the 100 species that is, for both the yellow crazy ant (Anoplolepsis
assessed during the implementation of the proto- gracilipes) and the big-headed ant (Pheidole mega-
col (Table 2). Most frequently, different use of ref- cephala). Few species were attributed this highest
erence material was part of, if not the entire level of impact, and only five species were
cause of difference in a species assessment, clo- assigned this category by at least one assessor
sely followed by limitations of the assessment (Fig. 3). Considering only those species for which
framework and the mechanism or extent of there was agreement on severity and excluding
impact being unclear in the literature (Fig. 4). those that were data deficient, the number of spe-
However, the frequency of these causes of uncer- cies across impact severity categories approxi-
tainty differed depending on the assessment mates a symmetric distribution with Moderate
component (impact mechanism versus severity). (n = 26) the most attributed category
For example, although extrapolation of evidence (Appendix S1: Table S1). Including instances of
and deviation from assessment protocol occurred difference alters the distribution of impact severi-
similar numbers of times (14 and 12 times, ties, which reflects the uneven frequency with
respectively), all but one of the former cases were which each category was associated with assess-
related to differences in severity, while all ment differences. Agreement on impact mecha-
instances of the latter were related to differences nisms was most common for herbivory (n = 17)
in mechanisms (Fig. 4). Systematic error was the and competition (n = 17; Table 1). Hemiptera
v www.esajournals.org 9 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Table 2. Causes and types of uncertainty that may lead to differences between independent assessors when per-
forming environmental impact assessments.
Cause of uncertainty
(description used in Figures) Explanation Uncertainty type Species example
1. Human error Performing impact assessments using Measurement error (i) (E) Xylosandrus
incorrect information, for example, compactus
using the wrong species (e.g., (Coleoptera:
Xyleborus glabratus) in an Curculionidae)
assessment can result in
misinformation
2. Incomplete information Non-comprehensive literature Systematic error (ii) (E) Drosophila
searches (incomplete search) searches can lead to different subobscura
assessments for the same species. (Diptera:
Drosophilidae)
3. Documented data and Evidence is not uniform in its Systematic error as a result Cinara cupressi
knowledge not readily or widely availability to all assessors. of lack of knowledge (Hemiptera:
accessible (knowledge Language barriers can also prevent (ii_a) (E) Aphididae)
inaccessible) the use of some evidence. This can
result in either relevant pieces of
evidence being overlooked and/or
different evidence used between
assessors
4. Species identification The dynamic nature of taxonomy can Systematic error (ii)† (E) Apis mellifera
lead to errors of commission and scutellata
omission. Accepted species names (Hymenoptera:
may change over time leading to Apidae)
multiple synonyms for a species
causing evidence to be missed.
Additionally, insufficient taxonomic
resolution can lead to the erroneous
inclusion of evidence; specifically,
organisms with many sub-species
5. Information regarding a) Inclusion of impact evidence based Systematic error as a result N/A
indigenous and/or alien range is on course resolution knowledge of of lack of knowledge Chaetosiphon
insufficient (range uncertain) alien range may lead to (ii_a) (E) fragaefolii
overestimation of impacts. Subjective judgment as a (Hemiptera:
b) Information on alien range is result of lack of Aphididae)
scarce or variable leading to a knowledge (iii_a) (E)
subjective decision whether to
conclude impact evidence is within
an alien range
6. Limitations of assessment Lack of clarification in the assessment Context dependence (iv) Orthotomicus erosus
framework (assessment limits) framework as to what constitutes (L) (Coleoptera:
the inclusion of certain evidence Curculionidae)
suitable for assessing environmental
impacts
7. Species designation as invasive Definitions for what constitutes an Vagueness (v) (L) Danaus plexippus
(invasiveness criteria) alien or invasive species can be (Lepidoptera:
vague and vary widely in the Nymphalidae)
literature, and not consistent with
the definition used in the
assessment. This can result in
differences in opinion as to whether
a species requires an impact
assessment or not
8. Unclear mechanism and/or Information obtained from available Subjective judgment as a Monomorium
extent of impact (mechanism/ evidence can be unclear regarding result of lack of destructor
impact unclear) how a species is having an knowledge (iii_a) (E) (Hymenoptera:
environmental impact, leading to an Formicidae)
assessor’s subjective interpretation
of the evidence available
9. Extrapolation of evidence Based on their expert knowledge, Subjective judgment (iii) Adelges tsugae
assessors may reach conclusions (E) (Hemiptera:
beyond what the available evidence Adelgidae)
states
v www.esajournals.org 10 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
(Table 2. Continued.)
Cause of uncertainty
(description used in Figures) Explanation Uncertainty type Species example
10. Deviation from assessment Ascribing impact mechanisms to a Measurement error (i) (E) Aspidiotus nerii
protocol (deviation from species despite there being (Hemiptera:
protocol) insufficient evidence to carry out an Diaspididae)
assessment
11. No apparent cause Difference between assessments has Natural variation (vi) (E) Anoplolepsis
no apparent cause other than a gracilipes
natural variation in evidence (Hymenoptera:
interpretation Formicidae)
Notes: Causes of uncertainty were identified from McGeoch et al. (2012), checked for relevance to impact classification, and
assigned to the relevant uncertainty type following Regan et al. (2002). Species examples provided are those encountered dur-
ing the implementation of impact classification for insects and are discussed further in the text. Uncertainty types are catego-
rized as epistemic (E) or linguistic (L).
† Although classified as systematic error, this cause of uncertainty would be more accurately considered a form of theoreti-
cal indeterminacy as per Regan et al. (2002).
were somewhat of an exception, and here, trans- independently conducted assessments, with the
mission of disease occurred as often as herbivory. reasons for these differences associated with 11
Although these two mechanisms were most fre- different causes and six different types of uncer-
quent, all orders except one (Dermaptera) were tainty. However, levels of agreement increased
associated with at least four mechanisms of substantially following discussion between
impact (Appendix S1: Fig. S3). This declined assessors of reasons for the uncertainty involved.
slightly when considering only instances of This process enabled us to identify several practi-
agreement, with Lepidoptera decreasing to two cable recommendations for reducing the uncer-
attributed mechanisms. tainty associated with assessing and classifying
the impact of alien species.
DISCUSSION Alien invasive insects have negative impacts
within and across multiple environmental set-
With alien species introduction events tings (Kenis et al. 2009, McGeoch et al. 2015).
expected to continue (Seebens et al. 2017), the This can add uncertainty to decisions on which
prioritization of species of most concern for man- evidence to include (context dependence,
agement authorities is likely to become increas- Table 2, Cause 6, Type iv). For example, the
ingly important. Information obtained from alien Mediterranean pine beetle has been unintention-
species impact classifications can assist in both ally introduced to multiple countries, most likely
prioritizing those species to manage, and those via the timber trade (Brockerhoff et al. 2006,
in which to invest research. However, uncer- Haack 2006). It is known for causing damage to
tainty that arises in the process of conducting conifers, particularly pines, often leading to tree
impact assessments for alien species means that mortality (Sangu € esa-Barreda et al. 2015). How-
the results may not be reproducible and, as we ever, all such evidence of a negative impact
show here, independent assessments can arrive inside its alien range takes place within managed
at different conclusions for the same species, and forestry plantations (Stephens and Wagner 2007).
single assessments may produce biased out- As such, both assessors for O. erosus, who dif-
comes. Neither of these alternatives is desirable fered in opinion as to which evidence to include,
when decision makers rely on such information came to entirely different conclusions, even after
to guide action and investment. Here, we the second round of assessments (data deficient
showed how examining alternative outcomes for vs. massive). Similar discrepancies occurred in
individual species provides insight on the type the assessments of other species (citrus long-
and extent of uncertainty involved, and potential horned beetle (Anoplophora chinensis), Asian
solutions for reducing it. We found that follow- long-horned beetle (A. glabripennis)). Similar
ing the protocol guidelines resulted in relatively arguments can be made for agricultural crops,
high levels of disagreement between depending on the extent to which native
v www.esajournals.org 11 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Mechanism Severity
Incomplete search (2)
Mechanism/impact unclear (8)
Assessment limits (6)
Extrapolation of evidence (9)
Cause of uncertainty
Deviation from protocol (10)
Human error (1)
No apparent cause (11)
Invasiveness criteria (7)
Knowledge inaccessible (3)
Range uncertain (5b)
Species identification (4)
Range uncertain (5a)
0 5 10 15 20 0 5 10 15 20
Frequency
Mechanism Severity
Systematic error (ii)
Subjective judgment (iii_a)
Context dependence (iv)
Type of uncertainty
Measurement error (i)
Subjective judgment (iii)
Natural variation (vi)
Vagueness (v)
Systematic error (ii_a)
0 5 10 15 20 250 5 10 15 20 25
Frequency
Fig. 4. Frequency with which (A) causes of uncertainty and (B) types of uncertainty varied within- and
between-impact assessment components (mechanism and severity of impact). Bar lengths highlight the dominant
causes (A) or types (B) of uncertainty. The number and Roman numerals alongside the causes and types of uncer-
tainty are those used in Table 2, and ii_a and iii_a are the sub-types related to lack of knowledge.
v www.esajournals.org 12 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
biodiversity or native relatives within them are Uncertainty associated with severity of impact
impacted by the alien insect. Such impacts can Agreement on severity of impact increased
thus pose a potential risk to native biodiversity from 34% to 70% following the second round of
more widely. Without formal direction and clari- assessments, suggesting that this assessment
fication at the outset of the assessment process, component is quite amenable to a reduction in
differences based on subjective interpretations of uncertainty. Incomplete information searches
systems suitable for inclusion are inevitable, (systematic error, Table 2, Cause 2, Type ii) were
albeit avoidable. Inclusion of an alien species in a common cause of difference, and the most fre-
impact classification should, we argue, be based quent reason for why independent assessors
on what is being negatively affected more so assigned different impact severities. There are
than where; that is, the context within which the multiple possible explanations for why different
impact is occurring is less relevant in some cases. pieces of evidence were selected for use, not least
A different situation arises for species that of all are different interpretations of what consti-
may arguably never negatively affect the native tutes evidence appropriate for inclusion in the
environment, and this assumption led to some impact assessment (i.e., limitations of assessment
species being assigned minimal concern despite framework, Table 2, Cause 6, Type iv).
the absence of evidence (subjective judgment, Even with definitions for each severity cate-
Table 2, Cause 9, Type iii). Two groups of alien gory (Hawkins et al. 2015), assigning impact
insects, given their life histories, may lead an severity appears to be a decision open to greater
assessor to assign minimal concern without evi- interpretation than assigning impact mecha-
dence (i.e., rather than data deficient). The first nisms. One explanation may be the naturally
group are those species that are primarily stored unbounded nature of the relationship between
product pests. For example, the bean weevil successively more severe impact categories,
(Acanthoscelides obtectus) is an economically where even with definitions for each impact
important pest of legumes (Soares et al. 2015). severity category there is inevitably scope for
While damaging, its impact almost entirely subjective interpretations of degree. For example,
occurs on stored products, and it is a species pri- an alien species with a minor impact is one that
marily of economic concern with a very low like- negatively affects the fitness (or performance, see
lihood of having an environmental impact. The Fig. 5) of a native species (Hawkins et al. 2015).
second group are monophagous insects that only However, a decline in fitness of individuals is
attack and exist in agricultural crop environ- likely to lead to a decline in population size,
ments and/or have a long history of no evidence which constitutes the next most severe impact
of environmental impact. The broad bean beetle category (moderate). The difference between a
(Bruchus rufimanus) is a widespread crop pest, negative impact on the fitness of individuals in a
specifically targeting the faba bean (Vicia faba L.; population and a population-level effect is a mat-
Clement et al. 2002). Given the long history of ter of degree (Ricciardi et al. 2013, Blackburn
evidence for this species only being of concern et al. 2014). Thus, if we assume a given impact
for specific crop types, one may conclude that it severity begets the next most severe category,
is of minimal environmental concern, even then there are two alternative pathways to arrive
though the lack of evidence would otherwise at a given severity category, that is, either via evi-
lead to a data-deficient classification. Such con- dence of no impact or via no evidence of an
clusions are an instance of using absence of evi- impact at the next most severe category (the deci-
dence as evidence of absence, generally sion tree in Fig. 5 enables these alternatives to be
undesirable but logically valid under certain cir- distinguished). For example, an assessor may
cumstances (Sober 2009). The decision to classify conclude that a given species is having a minor
a species as minimal concern for environmental impact either because (1) there is evidence, it is
impact without evidence, however, should occur not causing population declines, or (2) because
outside of the impact assessment process in the there is no evidence, it is causing population
form of additional information that may comple- declines. For example, Hadley and Betts (2012)
ment the final assessment outcome, as suggested found that a proposed lack of negative conse-
for accommodating expert input. quences of habitat fragmentation on pollination
v www.esajournals.org 13 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
Decision tree for assigning an impact severity category to species of environmental concern
: No
reliable evidence that it has or had
Focal species
individuals existing in a wild state in
a region beyond its native
geographic range.
Does the species have alien
populations? : Best available
evidence indicates that a given taxon
has individuals existing in a wild state
outside of its native range but there
is inadequate information to classify
an impact or insufficient time has
NA NA elapsed since introductions have
Does the species have an become apparent.
environmental impact?
: Unlikely to
have caused deleterious impacts
on the native biota or abiotic
environment
DD Is the species negatively affecting
MC : Causing reductions in
the fitness* of native species? the fitness of individuals in the native
biota, but not declines in native
population sizes
: Causing declines in
Is the species negatively affecting the population size of native species,
the population of native species? but not changes to the structure of
communities or abiotic/biotic
ecosystem composition
MN MN : Causing the local
population extinction of at least one
Is the species causing local extinction native species, and leads to
of native species leading to community reversible changes in the structure of
level effects? communities and the abiotic/biotic
ecosystem composition.
: Causing the
MO MO replacement and local exitinction of
native species, and produces
irreversible changes in the structure
Are the community level effects
of communities and abiotic/biotic
irreversible? ecosystem composition.
Information is unavailable
MR MR
Evidence indicates yes
MV
Evidence indicates no
Fig. 5. Decision tree for improving the transparency and repeatability of assigning EICAT impact severity
v www.esajournals.org 14 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
(Fig. 5. Continued)
categories. This framework is based on the assumption that the different categories (MC–MV) are hierarchically
related; that is, a negative effect at one level necessitates a negative effect at the level below it (dashed arrows)
and may mean existence of an impact at a higher impact severity level, although no evidence may yet be avail-
able to support designation of this higher level of impact. Under this assumption, there can often be two path-
ways by which an impact category can be assigned, with the followed path dependent on information
availability. For example, in the case of the two pathways for assigning a moderate (MO) impact; if there is infor-
mation to support the conclusion that there is a negative effect on the population size of one or more native spe-
cies, but no effect on community structure, then the appropriate category to assign is MO. Alternatively, there
may be information on native population-level negative effects but no information (lack of evidence) available on
the effects at the community level, in which case the category assigned would also be MO. In general terms, these
two alternative pathways represent evidence of no impact (right) and no evidence of impact (left). In the most
recent version of the EICAT implementation guidelines (IUCN 2020a, b), the term fitness has been replaced by
performance.
dynamics was a result of the absence of evidence, other predation, or one assessor provided both
and not evidence of an absence of such conse- competition and predation because based on the
quences. Hawkins et al. (2015) state that an evidence, they were not able to distinguish
impact severity category is assigned if there is no between them. For example, the evidence of
evidence for the next most severe category, environmental impact for the harlequin lady-
implying the absence of evidence case (2 above). beetle (Harmonia axyridis; Roy et al. 2012, Masetti
However, explicitly distinguishing between these et al. 2018) and the Argentine ant (Linepithema
two scenarios is important (Altman and Bland humile; Menke et al. 2018) often inferred the
1995, Alderson 2004), even if the former occurs impact mechanism could be competition or pre-
rarely by comparison, because it informs the dation. In situations such as this where the evi-
strength of evidence available and confidence dence itself was sometimes unclear on the
placed in it. Specifying such reasoning during mechanism of impact, it is difficult to avoid dif-
assessments by making the decision process fering subjective interpretations by independent
explicit (as shown in Fig. 5) could reduce uncer- assessors (i.e., subjective judgment as a result of
tainty, as well as make the process more readily lack of knowledge, Table 2, Cause 8, Type iii_a).
repeatable by specifying the basis for the final Another common reason for differences
decision, that is, that there was, or was not, evi- between assessors on impact mechanisms was
dence for the next most severe category (Fig. 5). the use of different lines of evidence (systematic
error, Table 2, Cause 2, Type ii). Multiple reasons
Uncertainty associated with mechanism of impact for this are possible. First, for many insects the
One of the more common causes of difference quantity of literature is large, and a literature
in assigning a mechanism of impact was a lack of search can return thousands of studies for a sin-
clarification, on the part of the authors of the gle species (e.g., some species assessed had
studies used as evidence, as to how an alien spe- search returns in the order of 14,000). This
cies was negatively affecting the native environ- increases significantly the time needed to per-
ment. In such cases, assessors were required to form an assessment, and the goal is to be thor-
make a subjective interpretation as to what ough. Nonetheless, simple oversight can result in
mechanism was most likely, based on the context mistakenly missing, misinterpreting, or includ-
of the study. Similarly, it was often difficult to dif- ing a certain piece of evidence (increasing the
ferentiate between two mechanisms, for exam- chance of human error and incomplete informa-
ple, competition and predation (Sandvik et al. tion searches as the cause of uncertainty; Table 2,
2004), where the evidence suggested that either Cause 1, Type ii). For example, differences in the
choice was valid. This led to situations where first round of assessments for the shot-hole borer
either one assessor chose competition and the (Xylosandrus compactus) occurred because of the
v www.esajournals.org 15 April 2021 v Volume 12(4) v Article e03461CLARKE ET AL.
accidental use of evidence pertaining to a differ- new mechanism of impact. This could subse-
ent species (Xyleborus glabratus; Table 2). By con- quently reduce uncertainty associated with
trast, the Mediterranean pine beetle impact mechanism decisions.
(Orthotomicus erosus) assessments differed as a
result of different interpretations of the EICAT Consequences of uncertainty in impact
guidelines (Table 2). Both examples become classification for IAS
causes of uncertainty for reasons other than sim- Not all causes of uncertainty have equally sig-
ply incomplete information searches. This shows nificant consequences for the outcome of an
how uncertainty can compound during the assessment. For example, deviation from assess-
assessment process, starting with one uncer- ment protocol (measurement error, Table 2,
tainty type (in this case measurement error) that Cause 10, Type i) in some instances was simply
leads to another (context dependence). However, the situation where one of the assessors,
this also implies that by addressing some of the although assessing a species as data deficient
more manageable causes of assessor difference, did actually provide a mechanism of impact,
uncertainty propagation could feasibly be rather than assigning none (e.g., oleander scale,
reduced (see recommendations below on embed- Aspidiotus nerii; Table 2). This tended to occur
ding formal systematic review protocols into when the assessor was familiar with the species
environmental impact assessment protocols to and providing an educated judgment as to the
facilitate the transparency, efficiency, and most likely mechanism of impact. For example,
repeatability of literature screening and inclusion Lepidoptera are likely to affect via herbivory
(Siddaway et al. 2019, Haddaway et al. 2020)). and carabids by predation (or competition). Dis-
The complexity of assessing environmental crepancies of this kind are unlikely to have any
impacts extends beyond the broad range of eco- serious consequences for species prioritization,
logical systems that species invade to the range and by extension resource allocation, and can
and complexity of mechanisms of impact inva- be readily avoided. Extrapolating evidence to
sive alien species can have. One of the most fre- assign an impact severity beyond that provided
quent mechanism categories was other; that is, by the available evidence is, however, more
the evidence suggested a mechanism not cur- problematic (i.e., subjective judgment, Table 2,
rently listed under the EICAT protocol. For Cause 9, Type iii). This was one of the more fre-
example, the gall wasp (Andricus quercuscalicis) quent causes of uncertainty associated with
was assessed as having an impact by altering the impact severity. Subjective judgment of this
sex ratio of native species (Scho €nrogge et al. form stemmed primarily from the expert knowl-
2000), a mechanism that does not easily fit within edge of those performing assessments on spe-
any of the other mechanism categories. Another cies with which they are particularly familiar.
example is the Asian honeybee (Apis cerana) that For example, impact severity for the hemlock
robs food stores of native species (Hyatt 2012) woolly adelgid (Adelges tsugae) differed as a
and was noted as possibly fitting under multiple result of this cause of uncertainty (Table 2).
mechanism categories. Differences in mechanism Extended familiarity with a given species may
choice between assessors involving the use of reduce objectivity leading to a more liberal, but
other often also involved the mechanism “inter- not necessarily wrong, interpretation of impact
action with other alien species” or “facilitation of severity. This subjectivity can result in an assess-
native species.” This may imply that other is ment at odds with an assessor who is less famil-
often used to denote some form of indirect inter- iar with the same species, but who therefore
action that is associated with a negative environ- bases their assessment on a more objective inter-
mental impact. If the outcome of an assessment pretation of the evidence. We suggest that this
is the assignment of other for the mechanism of cause of uncertainty could be reduced by explic-
impact, then the assessor should provide a con- itly incorporating, within the assessment frame-
cise description of what they interpret the mech- work, opportunity for expert opinion to be
anism to be. Such information would be useful provided in addition to and alongside the
not only for the assessment itself, but also for assessment based on published evidence alone
potentially updating the framework to include a (Morgan 2014; see recommendations below).
v www.esajournals.org 16 April 2021 v Volume 12(4) v Article e03461You can also read
