Consistency of COSEWIC species at risk designations: freshwater fishes as a case study more

959 Consistency of COSEWIC species at risk designations: freshwater fishes as a case study James R. Lukey and Stephen S. Crawford Abstract: The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) is responsible for the assessment of Canadian wildlife at risk. The COSEWIC assessment process is primarily based on five quantitative criteria developed by the International Union for the Conservation of Nature, but allows for further modification of designations under certain conditions. This study investigated the consistency of designations predicted using the quantitative COSEWIC criteria compared with observed designations reported by COSEWIC. A total of 49 COSEWIC designations for freshwater fishes from 2000 to 2007 were compared for consistency in decision-making. Overall, there was a 57.1% agreement between predicted and observed designations. A substantial number (35.1%) of COSEWIC designations were downlisted from ‘‘Endangered’’ or ‘‘Threatened’’ without sufficient explanation to justify the modifications. For the cases of uplisting, these differences were associated with qualitative criteria not effectively represented in our algorithm. Recommendations are offered to improve the transparency and accountability of COSEWIC decision-making, including enhancements to reporting and the explicit incorporation of uncertainty in the COSEWIC risk assessment protocol. ´ ` ´ ´ ´ ´ Resume : Le Comite sur la situation des especes en peril au Canada (COSEPAC) est responsable de l’evaluation de la ´ ´ ´ ` faune canadienne menacee. Le processus d’evaluation du COSEPAC est principalement base sur cinq criteres quantitatifs ´ mis au point par l’Union internationale pour la conservation de la nature, mais il permet des modifications subsequentes ´ ´ ´ ´ ´ ` de la designation dans certaines conditions. Notre etude examine la coherence entre les designations predites a partir des ` ´ ´ ´ ´ ´ criteres quantitatifs du COSEPAC et les designations realisees de fait par le COSEPAC. Nous avons examine la coherence ´ ´ dans les prises de decision dans un ensemble de 49 designations faites par le COSEPAC pour les poissons d’eau douce de ` ´ ´ ´ 2000 a 2007. Globalement, il y a une concordance de 57,1 % entre les designations predites et observees. Un nombre sig´ ´´ ´ ` ´ nificatif (31,5 %) des designations du COSEPAC ont ete reduites de « en voie de disparition » a « menace » sans les ex´ ´ plications necessaires pour justifier ces modifications. Dans les cas d’augmentation de la cote, les differences sont ´ ` ` associees a des criteres qualitatifs non effectivement compris dans l’algorithme. Nous formulons des recommandations ´ ´ pour ameliorer la transparence et la responsabilisation de la prise des decisions au COSEPAC, en particulier une augmen´ tation de la diffusion et l’incorporation explicite de l’incertitude dans le protocole d’evaluation des risques du COSEPAC. ´ [Traduit par la Redaction] Introduction Setting priorities is important for the conservation of species, given that approximately 17 000 species are considered threatened worldwide, and there are limited conservation resources typically available for consultation, listing, conservation, and recovery (James et al. 1999; IUCN 2008). One key priority is the assessment of risk of extinction and the designation of conservation status (Regan et al. 2005, 2004). Three general methods are often used, each with specific characteristics. Qualitative criteria rely on best available information and expert opinion (Master 1991; Regan et al. 2004). Point scoring approaches utilize attributes that are Received 13 August 2008. Accepted 20 April 2009. Published on the NRC Research Press Web site at cjfas.nrc.ca on 30 May 2009. J20724 J.R. Lukey.1 Department of Integrative Biology, University of Guelph, Guelph, ON N1G 2W1, Canada. S.S. Crawford. Department of Integrative Biology, University of Guelph, Guelph, ON N1G 2W1, Canada; Chippewas of Nawash Unceded First Nation, R.R. #5, Wiarton, ON NOH 2T0, Canada. 1Corresponding author (e-mail: jlukey@uoguelph.ca). scored and summed to indicate conservation priority (Millsap et al. 1990; Regan et al. 2005). Quantitative rules with explicit thresholds have been developed to help designate risk categories (IUCN 2001). The outcome of these different approaches can vary widely, with potentially serious management consequences for prioritizing protection efforts. The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) was created in 1977, based on the need for a national classification of wildlife species at risk (COSEWIC 2004a). In 2002, the Canadian Species at Risk Act (SARA) identified COSEWIC to be responsible for the scientific assessment of wildlife species at risk (SARA 2002). COSEWIC assesses species (or designatable units (DUs)) from 12 taxonomic groups: terrestrial mammals, marine mammals, birds, reptiles, amphibians, marine fishes, freshwater fishes, arthropods, molluscs, vascular plants, mosses, and lichens (Mooers et al. 2007). As of April 2007, 759 species had been assessed, of which 425 were formally listed under SARA (COSEWIC 2007a; Government of Canada 2008a). COSEWIC designations are made at Species Assessment Meetings, where committee members first consider five factors leading to status designation for a candidate species: (i) eligibility of species to be assessed; (ii) sufficiency of information to determine status; (iii) adequacy of the species Published by NRC Research Press Can. J. Fish. Aquat. Sci. 66: 959–971 (2009) doi:10.1139/F09-054 960 Fig. 1. Committee on the Status of Endangered Wildlife in Canada (COSEWIC) decision-making process, from candidate species to final designation, showing the five quantitative criteria in Phase I (A–E) and the four modifying criteria in Phase II (F–I). Risk designation codes are as follows: DD, Data Deficient; NR, Not at Risk; SC, Special Concern; TH, Threatened; EN, Endangered. Broken lines represent decision-making based on nonquantitative criteria based on expert opinion (i.e., criteria that could not be effectively represented in the computer algorithm of the COSEWIC protocol used in this study). Can. J. Fish. Aquat. Sci. Vol. 66, 2009 Fig. 2. Possible inconsistencies between species at risk designations using the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) protocol and the assigned COSEWIC risk designations. (a) Full Consistency: species designated at the same level as the protocol; (b) Uplisting Hypothesis: species designated at a higher level than the protocol; (c) Downlisting Hypothesis: species designated at a lower level than the protocol; (d) Combined Effects: species designated at higher or lower levels than the protocol. Risk designation codes are as follows: NR, Not at Risk; SC, Special Concern; TH, Threatened; EN, Endangered. assessment report; (iv) preliminary designation based on the quantitative assessment; and (v) evaluation of preliminary designation leading to final designation. The quantitative assessments are based on the revised Red List and Criteria and Categories developed by the International Union for the Conservation of Nature (COSEWIC 2004a). COSEWIC determines the status of candidate species in two steps (Fig. 1): Phase I is based on five quantitative criteria (A–E), and Phase II is based on four modifying criteria (F–I). After the two assessment phases are complete, COSEWIC designates the species as Data Deficient (DD), Not at Risk (NR), Special Concern (SC), Threatened (TH), or Endangered (EN) (COSEWIC 2006a). According to the COSEWIC (2005f) Operations and Procedures Manual, the final step of the assessment process represents a culmination of decisionmaking in which ‘‘the Committee considers all the information, analysis, and discussion presented at the meeting, and evaluates if the status category suggested by the application of the criteria and guidelines is consistent with the definition of the status category used by COSEWIC. If there is inconsistency, the status representing the most appropriate definition will take precedence, and any variance between the status definition and the quantitative criteria will be explained.’’ The Committee may make detailed, confidential notes of these discussions; however, these are not made available to the public (COSEWIC 2005f). Ultimately, the COSEWIC species designation is forwarded to the Government of Canada as a recommendation for official listing under SARA, which if approved for TH or EN status would automatically trigger legal protection, development and implementation of recovery plans, and protection of critical habitat (SARA 2002). For the purpose of this study, the term ‘‘protocol’’ will refer specifically to COSEWIC’s use of the five quantitative assessment criteria (A–E) of Phase I and rescue effect criterion (F) of Phase II. To date, few studies have quantitatively examined consistency in the species at risk designation process (e.g., Akca¸ kaya et al. 2000; Regan et al. 2005). In this study, we discriminate between two potential sources of bias in the risk designation process. The Uplisting Hypothesis states that COSEWIC guards against the possibility of under-protection (type II error) by consistently choosing designations higher than determined by the protocol (Fig. 2b). This type of bias can be important because it can waste limited resources protecting a species that does not actually require protection. In contrast, the Downlisting Hypothesis states that COSEWIC guards against the risk of over-protection (type I error) by consistently choosing designations lower than determined by the protocol (Fig. 2c). This type of bias can be important because it could leave some bona fide species at risk without the protection afforded by SARA designation. These two hypotheses can combine to predict a bimodal distribution for under- and over-designated species (Fig. 2d). The goal of this study was to evaluate the degree of consistency between risk designations generated by COSEWIC’s stated protocol (predicted) and the risk designations actually reported by COSEWIC (observed), using freshwater fishes as a case study. To achieve this goal, it was necessary to satisfy the following specific objectives: (i) develop a computer algorithm to simulate the COSEWIC protocol; (ii) collate COSEWIC reports for the freshwater fishes and Published by NRC Research Press Lukey and Crawford 961 extract input data for the protocol; (iii) execute the algorithm using extracted input data to generate predicted designations; (iv) collect COSEWIC assessment reports to identify observed designations; and (v) evaluate the level and pattern of consistency between predicted and observed designations. Materials and methods An algorithm was programmed in MATLAB (The MathWorks Inc., Natick, Massachusetts, www.mathworks.com) to emulate the COSEWIC decision-making process (Fig. 1), using 35 input variables defined by the COSEWIC protocol. For Phase I, each of the five quantitative criteria (A–E) was assessed independently; however, the overall risk designation was established as the highest contributing rank. For Phase II, when the rescue effect (criterion F) was engaged, the overall designation from Phase I could be modified down one level (EN to TH; TH to SC). Criteria G, H, and I of the modifying criteria were not included in the algorithm because these were all derived from qualitative expert opinion that could not be effectively coded. The final risk designations were assigned as ordinals: DD (–1), NR (0), SC (1), TH (2), EN (3). COSEWIC species assessment reports for freshwater fishes (species or DUs) evaluated from 2000 to 2007 were compiled from the Species at Risk Public Registry (Government of Canada 2008b). Input data were extracted from the COSEWIC reports for each of the 35 input variables. In cases where data were recorded in the COSEWIC report as a range (minimum–maximum), the midpoint was used as the input value. In the few cases where data were recorded in the COSEWIC report as either greater than or less than some threshold with no additional information, a value 25% greater than or 25% less than the stated value was arbitrarily used, respectively. The algorithm was executed for all of the selected taxa, resulting in the generation of predicted COSEWIC designations. COSEWIC risk status recommendations were obtained from the Species at Risk Public Registry, and these were used as the observed species designations. In cases where variables were expressed with uncertainty (minimum–maximum, less than, greater than), a sensitivity analysis was conducted to determine the extent to which differences between the algorithm predictions and the observed COSWIC designations were due to the uncertain estimates. The variables occasionally expressed with uncertainty included percent decline (PD), extent of occurrence (EO), area of occupancy (AO), number of locations (LC), and number of mature individuals (MI). In cases where a range of values was identified in the COSEWIC report, the minimum value; 25th, 50th, and 75th percentiles; and the maximum value were used to determine designation probabilities. For those cases with a ‘‘less than’’ value (e.g., <9000 mature individuals), the minimum value was set at 0, and the maximum was set at the stated value (e.g., 9000). For cases with a ‘‘greater than’’ value (e.g., >10 000 km2), the minimum value was set as the minimum value, and the maximum was arbitrarily set at five times the minimum value. For percent decline ‘‘greater than’’ values, the maximum was always set at 100%. If more than one variable was expressed with un- certainty, all possible combinations of the bounds and percentiles described above were evaluated. For cases with a single uncertain variable, there were five possible combinations; for two uncertain variables, there were 25 possible combinations; for three uncertain variables, there were 125 combinations; and for the maximum of four uncertain variables, there were 625 combinations. Values were constrained by COSEWIC rules such that extent of occurrence could never be lower than the area of occupancy. The probabilities of designation for each level (NR, SC, TH, and EN) were calculated for each species or DU, and the proportions were compared with the original algorithm designation and the observed COSEWIC designation. A Kolmogorov–Smirnov goodness-of-fit test was used to determine if there was a significant difference of total risk designation frequencies between the predicted and observed designations. The Kolmogorov–Smirnov test was used because it is a nonparametric test that has greater power than c2 tests, and it works well even with small sample sizes (Sokal and Rohlf 1981). A paired t test was used to determine if there was a difference between the ordinal rankings of the predicted and observed designations. A generalized linear model (GLM) was used to determine if there was a relationship between agreement (predicted vs. observed) and other factors that could have affected the COSEWIC designations: year of assessment; family of species; presence or absence of report summary table; and taxonomic status (DU = 0, species = 1). Statistical analyses were conducted using SPSS 15.0 (SPSS Inc., Chicago, Illinois) and SAS 9.1 (SAS Institute Inc., Cary, North Carolina). Results A total of 54 COSEWIC reports and assessments for freshwater fishes were compiled, consisting of 15 families and 41 species, with 7 species separated into multiple DUs (Table 1). Five of the observed COSEWIC designations were found to be DD and were not included further in this analysis. The remaining 49 assessments were examined for differences between predicted and observed COSEWIC designations. The algorithm predicted designations of 10 NR, 2 SC, 6 TH, and 31 EN, while COSEWIC reports revealed observed designations of 2 NR, 15 SC, 12 TH, and 20 EN. The distribution of predicted versus observed designations is shown for each of the 49 taxa assessed (Table 2); the shaded diagonal represents complete agreement between predicted and observed designations. Overall, COSEWIC designations matched the predicted designations for 57.1% (28/49) of the assessments. Most of the 21 differences were found in the uplisting of the lowest designation level (NR to SC, n = 8) and in the downlisting of the highest designation level (EN to TH, n = 8). Eight of the ten predicted NR designations were uplisted by COSEWIC to SC (Table 3), based on a variety of reasons associated with the application of expert opinion modifying criteria (G, H, I) in Phase II (Fig. 1). Two of the six predicted TH designations were observed at the lower level of SC, while 11 of the 31 predicted EN designations were observed at a lower level (8 TH and 3 SC) (Table 4). Only 3 of the 21 differences were more than a single designation level, and these were all downlisted from EN to SC. Published by NRC Research Press 962 Table 1. Freshwater fish species (or designatable units) assessed by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) during the period 2000–2007, with predicted and observed COSEWIC risk designations. Designation Family Petromyzontidae Species Ichthyomyzon fossor Lampetra richardsoni Acipenser brevirostrum Acipenser fulvescens Distribution Saskatchewan River, Nelson River Great Lakes, upper St. Lawrence Canada Canada Nelson River Red–Assiniboine rivers, Lake Winnipeg Saskatchewan River Western Hudson Bay Winnipeg River, English River Lake of the Woods Southern Hudson Bay, James Bay Great Lakes, upper St. Lawrence Canada Canada Canada Canada Canada Lake Simcoe Upper Great Lakes Canada Canada Canada Canada Alberta British Columbia Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada Newfoundland Year 2007 2007 2000 2005 2006 2006 2006 2006 2006 2006 2006 2006 2004 2003 2005 2006 2005 2005 2005 2004 2007 2005 2003 2006 2006 2000 2007 2001 2002 2006 2007 2006 2002 2005 2006 2005 2004 2002 2002 2003 Predicted DD NR EN EN EN EN EN EN EN TH NR EN NR EN TH NR SC DD EN SC DD EN TH EN NR EN EN EN EN TH EN EN EN NR EN TH EN DD EN NR Observed DD SC EN SC EN EN EN EN EN SC SC TH SC EN TH SC SC DD SC SC DD EN TH TH SC EN EN TH EN TH EN EN EN SC SC TH EN DD EN SC Difference Uplist Downlist Acipenseridae Lepisosteidae Anguillidae Esocidae Salmonidae Acipenser medirostris Acipenser transmontanus Lepisosteus oculatus Anguilla rostrata Esox americanus vermiculatus Coregonus clupeaformis Coregonus kiyi kiyi Coregonus laurettae Coregonus nigripinnis Coregonus reighardi Coregonus zenithicus Oncorhynchus clarkii lewisii Salvelinus fontinalis timagamiensis Clinostomus elongatus Hybognathus argyritis Notropis anogenus Notropis percobromus Rhinichthys cataractae Rhinichthys osculus Catostomus sp. Minytrema melanops Moxostoma carinatum Moxostoma duquesnei Moxostoma hubbsi Noturus insignis Noturus stigmosus Fundulus diaphanous Downlist Uplist Downlist Uplist Uplist Downlist Downlist Uplist Cyprinidae Downlist Can. J. Fish. Aquat. Sci. Vol. 66, 2009 Published by NRC Research Press Catostomidae Uplist Downlist Ictaluridae Fundulidae Uplist Lukey and Crawford Table 2. A comparison of the predicted and observed number of species in each risk category for COSEWIC designations of 49 freshwater fishes assessed during the period 2000–2007. Predicted Observed EN TH SC NR Total Designation Predicted EN EN EN EN EN EN DD NR NR EN EN TH NR EN NR 0 0 8 2 10 SC 0 0 2 0 2 TH 0 4 2 0 6 EN 20 8 3 0 31 Total 20 12 15 2 49 963 Difference Uplist Downlist Downlist Downlist Downlist Downlist Observed EN EN EN EN TH TH DD NR SC TH TH SC NR TH Downlist Note: Risk designation codes: NR, Not at Risk; SC, Special Concern; TH, Threatened; EN, Endangered. The shaded diagonal represents complete consistency between predicted and observed designations. Note: Risk designation codes: NR, Not at Risk; SC, Special Concern; TH, Threatened; EN, Endangered. Myoxocephalus quadricornis Myoxocephalus thompsonii Species Gasterosteus sp. Lepomis gulosus Etheostoma blennioides Percina copelandi Morone saxatilis A total of 31/49 (63.3%) assessments had at least one variable that was expressed with uncertainty (Appendix A, Table A1). EO was uncertain in 25 assessments, AO and MI were each uncertain in 14 assessments, and PD and LC were each uncertain in 5 assessments. In 26/31 (83.9%) of the cases expressed with uncertainty, there was complete (P = 1.00) agreement between predicted designations without and with uncertainty. In 4/5 (80%) of the remaining cases where uncertainty yielded multiple designations, the maximum probability matched the designation without uncertainty. The lone exception was shortjaw cisco (Coregonus zenithicus), where the designation without uncertainty was TH, while the designation with uncertainty was highest for EN at P = 0.60 (Appendix A, Table A1). The overall distributions of predicted and observed designations were found to be significantly different (Kolmogorov–Smirnov, P = 0.013). At the individual species level, the predicted and observed ordinal designations were not significantly different (paired t test, P > 0.05). There was no apparent relationship between designation consistency and year of assessment, family of species, or presence of summary table in the COSEWIC report (GLM, P > 0.05). There was a significant relationship between designation consistency and taxonomic status (GLM, P < 0.01). Assessments for taxa at the level of DUs exhibited consistent designations in only 6 of 18 (33.3%) of the cases, while assessments at the level of species were consistent in 22 of 31 (71.0%) of the cases. Of the 12 DUs where there was disagreement, seven were predicted at a higher status than observed. Cottus sp. Distribution Canada Canada Canada Canada Canada St. Mary River, Milk River Canada Western Canada Great Lakes, western St. Lawrence Bay of Fundy Southern Gulf of St. Lawrence Canada Canada Canada Year 2002 2002 2006 2006 2000 2005 2003 2006 2006 2004 2004 2005 2006 2002 Discussion The goal of this study was to evaluate the degree of consistency between risk designations generated by COSEWIC’s stated protocol (predicted) and the risk designations actually reported by COSEWIC (observed), using freshwater fishes as a case study. In general, the observed COSEWIC risk designations for 43% of the freshwater fishes were found to be inconsistent with the protocol of quantitative criteria A–E that could be effectively represented in this analysis, a result that supports consideration of previous assertions that COSEWIC designations are not highly repeatable (Andelman et al. 2004). In virtually all cases where Published by NRC Research Press Table 1 (concluded). Family Gasterosteidae Moronidae Centrarchidae Percidae Cottidae 964 Reason ‘‘. . .about 50% of the streams it is known to inhabit are subjected to ongoing chemical treatment for sea lamprey control which causes mortality to its larval stage.’’ (COSEWIC 2007b) ‘‘The increased access to relatively unimpacted populations and the likelihood of increased hydroelectric development in some areas are causes for concern. . .’’ (COSEWIC 2006c) ‘‘This species is globally at risk and is of concern in Canada because of exploitation and habitat loss due to damming of rivers.’’ (COSEWIC 2004c) ‘‘. . .abundance in the Upper St. Lawrence River and Lake Ontario has declined by approximately 99% since the 1970s. . . Possible causes of the observed decline, including habitat alteration, dams, fishery harvest, oscillations in ocean conditions, acid rain, and contaminants, may continue to impede recovery.’’ (COSEWIC 2006b) ‘‘Populations are stressed by hybridization and competition with introduced species. . . expanding urban development, agricultural activities and resource-based industries are expected to lead to additional stresses associated with habitat loss and degradation. . .’’ (COSEWIC 2006d) ‘‘. . .restricted to southwestern Ontario. The greatest threat to this species is habitat degradation through increased erosion and turbidity.’’ (COSEWIC 2005d) ‘‘The population is separated from others by a major barrier to movement. . . very limited area of occupancy. The possibility of range expansion is limited by steep gradients and impassable rapids and/or falls. Habitat degradation resulting from proposed logging would negatively impact populations in some areas.’’ (COSEWIC 2003) ‘‘Populations have been exterminated in 2 lakes in Quebec due to eutrophication of these lakes, and may be in decline in Lake Huron, possibly in relation to the introduction of zebra mussel.’’ (COSEWIC 2006e) Can. J. Fish. Aquat. Sci. Vol. 66, 2009 input variables were expressed with uncertainty, the sensitivity analysis demonstrated that our representation of the uncertainty did not have an effect on the predicted designation. Overall, the differences found between predicted and observed designations reflected changes of a single designation level, but strongly suggest that there were additional COSEWIC decision-making processes not described in the stated protocol. It is important to recall that in all cases where a final COSEWIC designation varies from preliminary designations, reasons for the different designation must be explained (COSEWIC 2006a). We searched further through the COSEWIC reports to identify explanations for the 8 observed cases of uplisting and the 13 observed cases of downlisting. In the majority of these cases, while COSEWIC did make some explicit reference to reasons for uplisting or downlisting, most of the statements were not informative enough to justify a change in COSEWIC risk designation. As mentioned above, some of the differences between predicted and observed designations may result from the final round of COSEWIC discussions, during which the Committee evaluates the appropriateness of the status resulting from the application of quantitative and modifying criteria and selects the ‘‘most appropriate definition’’ (COSEWIC 2005f). However, the confidential notes of these expert discussions are not made available to the public and thus cannot be represented in a post hoc analysis. A lack of transparency and accountability in this phase of decision-making could lead to serious concerns about the rationale for final COSEWIC risk category designations (Andelman et al. 2004; Keith et al. 2004). Year of assessment could have influenced designation consistency because the COSEWIC process is periodically reviewed and revised to provide more accurate assessments of taxa (SARA 2002). Detailed technical summary tables were not presented in all COSEWIC reports, especially those from the early period of 2000–2002. Presence of summary tables could have increased designation consistency by providing all of the important decision-making information in one location so that assessors would not need to search and extract input variables from the report text. However, neither the year of assessment nor the presence of a technical summary had a major effect on consistency between predicted and observed designations. Taxonomic status of the assessment (species versus DU) could also have influenced designation consistency, because recognition of a DU below the species level requires information on the discreteness of the unit, such as genetic information or natural barriers causing a disjunction of the population (COSEWIC 2008). A significant relationship was indeed found between taxonomic status and designation consistency; however, we actually found that species separated into DUs were less likely to exhibit consistent risk designations, compared with species that were assessed at the level of species. It has been hypothesized that separating species into DUs for assessment of extinction risk implicitly adds a precautionary factor in endangered species decisionmaking (Green 2005). It is interesting to note that of the six DUs that were predicted to be NR, five received an SC designation; this could be considered a form of extra precaution employed by COSEWIC. Logically, COSEWIC downlisting can only occur for desPublished by NRC Research Press Table 3. Freshwater fish species assessed by COSEWIC during the period 2000–2007 that were uplisted from Not At Risk (NR) to Special Concern (SC), with stated reasons for the uplisting. Observed SC SC SC SC SC SC Designation Predicted NR SC NR NR NR NR NR NR Distribution Great Lakes, upper St. Lawrence Southern Hudson Bay, James Bay Canada British Columbia Westslope cutthroat trout Common name Northern brook lamprey Lake sturgeon Green sturgeon American eel Spotted sucker Banded killifish Newfoundland Canada Canada Oncorhynchus clarkii lewisii Species Ichthyomyzon fosser Acipenser fulvescens Acipenser medirostris Anguilla rostrata Minytrema melanops Fundulus diaphanus Myoxocephalus thompsonii Deepwater sculpin Great Lakes, western St. Lawrence NR SC Lukey and Crawford Table 4. Freshwater fish species assessed by COSEWIC during the period 2000–2007 that were downlisted from Endangered (EN) or Threatened (TH), with stated reasons for the downlisting where available. Designation Species Acipenser brevirostrum Acipenser fulvescens Common name Shortnose sturgeon Lake sturgeon Distribution Canada Predicted EN Observed SC Reasons ‘‘. . .although the area of occupancy is <500 km2, and the species is known from only one location in Canada, there is no evidence of continuing decline or extreme fluctuations in the area of occupancy, extent or quality of habitat, or number of mature individuals.’’ (COSEWIC 2005c) Not stated (COSEWIC 2006c) ‘‘Meets criteria for Endangered A2abcd, but designated Threatened A2abcd because a quarter of the populations have been lost, more than half of the remaining populations are either stable or recovering’’ (COSEWIC 2006c) ‘‘The extirpation in lakes Huron and Michigan occurred more than three generations in the past. The remaining population in Lake Superior appears to be stable, and supports a small, regulated fishery. Other threats, such as the introduction of exotic species, which impacted populations in the lower lakes do not appear to be important in Lake Superior.’’ (COSEWIC 2005b) ‘‘Native populations have been reduced by almost 80% ... decline rate in the last 3 generations is not known’’ (COSEWIC 2006d) Not stated (COSEWIC 2001) ‘‘Met criterion B2 (AO of 178.5 < 500 sq. km) and a) for severely fragmented but no continuing decline or extreme fluctuations.’’ (COSEWIC 2006f) Not stated (COSEWIC 2000) ‘‘the area of occupancy (6 km2) is less than 500 km2, there are fewer than 5 locations, and there is evidence of decline in AO, quantity and quality of habitat and number of populations, but the rate is of decline is not known’’ (COSEWIC 2005a) ‘‘Met criteria for Endangered, A2bc, but designated Threatened, A2bc; D2, because the one remaining spawning population does not appear to be at imminent risk.’’ (COSEWIC 2004b) ‘‘Met criteria for Endangered, B2ac(iv), but designated as Threatened, B2ac(iv); D2, because of the high degree of resilience evident in recent spawner abundance estimates.’’ (COSEWIC 2004b) ‘‘Met criterion for Threatened D2, but there is a possibility of a rescue effect from populations in the United States. Therefore, designated Special Concern.’’ (COSEWIC 2005e) Not stated (COSEWIC 2002) Lake of the Woods Great Lakes, upper St. Lawrence TH EN NR TH Coregonus kiyi kiyi Upper Great Lakes kiyi Canada EN SC Oncorhynchus clarkii lewisii Hybognathus argyritis Moxostoma carinatum Cottus sp. Cottus sp. Westslope cutthroat trout Western silvery minnow River redhorse Cultus pygmy sculpin Eastslope sculpin Striped bass Alberta Canada Canada Canada St. Mary River, Milk River Bay of Fundy Southern Gulf of St. Lawrence Canada Canada EN EN EN EN EN TH TH SC TH TH Morone saxatilis EN EN TH EN TH TH SC TH Lepomis gulosus Percina copelandi Warmouth Channel darter Published by NRC Research Press Note: NR, not at risk; SC, special concern. 965 966 Can. J. Fish. Aquat. Sci. Vol. 66, 2009 ignations that have levels below them (i.e., EN and TH rather than NR). The 13 downlistings observed in this study support the Downlisting Hypothesis that analysts guard against the risk of over-protection (type I error) by consistently choosing risk designations that would be lower than concluded by the protocol. COSEWIC decision-making allows for the downlisting of risk designations if there is a clear indication of the rescue effect from extralimital populations (COSEWIC 2006a). However, for 6 of 13 (46%) downlisted designations, COSEWIC justification for the downlisting did not directly correspond to any of the stated possible reasons for designating at a lower risk level. For example, the eastslope sculpin (Cottus sp.) was downlisted from EN to TH because rate of decline was unknown (COSEWIC 2005a, 2006a). The eastslope sculpin had an extent of occurrence that was estimated to be below 5000 km2, an area of occupancy estimated to be below 500 km2, there were less than five locations, and there was evidence of decline, which all should qualify for EN designation under criterion B (COSEWIC 2005a). Furthermore, two of the downlistings were contradictory to the data presented in the COSEWIC report. River redhorse (Moxostoma carinatum) was apparently not listed as EN because there were ‘‘no continuing decline or extreme fluctuations’’ (COSEWIC 2006f); however, the river redhorse was reported to exceed the EN threshold for criterion B with an area of occupancy of less than 500 km2, a severely fragmented population and decline in the Quebec range, and decline in number of populations (COSEWIC 2006f). Shortnose sturgeon (Acipenser brevirostrum) was apparently not listed as EN because of lack of evidence of decline. However, shortnose sturgeon passed the EN threshold for criterion B with an extent of occurrence less than 5000 km2, an area of occupancy of less than 500 km2, only one location, and declines in both extent of occurrence and area of occupancy (COSEWIC 2005c). Four of the downlisted designations did not give any reasons for downlisting, while one case was downlisted because of the possibility of the rescue effect, which is considered to lower the chance of extinction of a subpopulation owing to immigration of individuals from other subpopulations (Brown and Kodric-Brown 1977). Similarly, COSEWIC uplisting can only occur for designations that have levels above them (i.e., NR and TH rather than EN). All of the eight uplistings that we observed were associated with modification from NR to SC, and these could have been accounted for by the nonquantitative, modifying criteria (G, H, I) that were not represented in our protocol algorithm. For these reasons, we do not see strong support for the Uplisting Hypothesis that analysts guard against the possibility of under-protection (type II error) by consistently choosing designations higher than determined by the protocol. However, it is important to note that contrary to COSEWIC (2006a) requirements, most of the statements provided to explain the uplistings from NR to SC were not sufficiently informative to justify the uplistings. There are implications for species at risk management resulting from inconsistent risk designations. Under the Canadian Species at Risk Act, COSEWIC designations are forwarded as recommendations to the responsible minister (Environment Canada or Fisheries and Oceans Canada) for consideration in association with other social, economic, po- litical, and legal factors (SARA 2002). Downlisting can lead to species at risk receiving less protection than they require for recovery, and this can substantially increase the probability of extinction (Rohlf 1991; Prato 2005). When COSEWIC downlists a species to NR, it is recommending that the species receive no protective measures under SARA at all (COSEWIC 2006a). COSEWIC downlisting to SC leads to a recommendation that the species receive some degree of management (SARA 2002), but not the enhanced measures of TH or EN (e.g., automatic prohibitions, protection of critical habitat). Conversely, when COSEWIC uplists a species to SC, it is recommending the expenditure of limited government funds for species protection and recovery, when the species may not in fact require those efforts. Based on the results of this study, a number of methodological recommendations can be considered for improving consistency in COSEWIC decision-making. First, the modification of designations should be conducted in a more transparent and accountable manner, especially with regard to the discussions of Committee experts that are currently not released to the public. The designation process should closely follow the COSEWIC protocol, and changes should be clearly justified, especially when COSEWIC expert discussions lead to postquantitative modifications. Second, although many of the more recent COSEWIC reports provide some explanation of criteria-specific designations, a complete breakdown of Phases I and II should be required, including the rationale for the final Committee decision. Finally, COSEWIC needs to develop practical tools to explicitly incorporate uncertainty in its decision-making process. In our next pair of studies, we will present analyses of how (i) uncertainty in information availability (presence or absence) and (ii) uncertainty in quantitative states of nature relative to thresholds combine to pose special challenges and opportunities for COSEWIC to make better decisions regarding species at risk designations. It is our hope that these kinds of sensitivity analyses will assist in the continued development of effective decision-making support tools for COSEWIC risk designations. Acknowledgements Financial support for this research project was provided by the Chippewas of Nawash Unceded First Nation and the Department of Integrative Biology, University of Guelph. Andrew MacDougall and Tom Nudds reviewed early drafts of the manuscript, while Nick Mandrak provided essential insight into the practical application of the COSEWIC protocol. Dan Gillis and Dongzhi Qu assisted with statistical analyses, and Gabriel Smith provided technical support with computer programming. We thank John Reynolds and an anonymous referee for valuable comments on an earlier draft of the manuscript. References Akcakaya, H.R., Ferson, S., Burgman, M.A., Keith, D.A., Mace, ¸ G.M., and Todd, C.R. 2000. Making consistent IUCN classifications under uncertainty. Conserv. Biol. 14(4): 1001–1013. doi:10.1046/j.1523-1739.2000.99125.x. Andelman, S.J., Gross, C., and Regan, H.M. 2004. A review of protocols for selecting species at risk in the context of US Forest Published by NRC Research Press Lukey and Crawford Service viability assessments. Acta Oecol. 26(2): 75–83. doi:10. 1016/j.actao.2004.04.005. Brown, J.H., and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effect of immigration on extinction. Ecology, 58(2): 445–449. doi:10.2307/1935620. COSEWIC. 2000. COSEWIC assessment and status report on the Cultus pygmy sculpin, Cottus sp., in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_cultus_pygmy_sculpin_1100_e.pdf [accessed 2 October 2008]. COSEWIC. 2001. COSEWIC assessment and status report on the western silvery minnow, Hybognathus argyritis, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_western_silvery_minnow_1101_e. pdf [accessed 2 October 2008]. COSEWIC. 2002. COSEWIC assessment and update status report on the channel darter, Percina copelandi, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/ files/cosewic/sr_channel_darter_e.pdf [accessed 2 October 2008]. COSEWIC. 2003. COSEWIC assessment and update status report on the banded killifish, Fundulus diaphunus, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/ files/cosewic/sr_banded_killifish_e.pdf [accessed 2 October 2008]. COSEWIC. 2004a. About COSEWIC: brief history [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.cosewic.gc.ca/eng/sct6/sct6_3_e. cfm#hist [accessed 21 July 2008; 6 February 2009]. COSEWIC. 2004b. COSEWIC assessment and status report on the striped bass, Morone saxatilis, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_striped_bass_e.pdf [accessed 2 October 2008]. COSEWIC. 2004c. COSEWIC assessment and update status report on the green sturgeon, Acipenser medirostris, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_green_sturgeon_e.pdf [accessed 2 October 2008]. COSEWIC. 2005a. COSEWIC assessment and status report on the ‘‘Eastslope’’ sculpin, Cottus sp., in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_eastslope_sculpin_e.pdf [accessed 2 October 2008]. COSEWIC. 2005b. COSEWIC assessment and update status report on the Lake Ontario kiyi, Coregonus kiyi orientalis, and Upper Great Lakes kiyi, Coregonus kiyi kiyi, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_kiyi_e.pdf [accessed 2 October 2008]. COSEWIC. 2005c. COSEWIC assessment and update status report on the shortnose sturgeon, Acipenser brevirostrum, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_shortnose_sturgeon_e.pdf [accessed 2 October 2008]. COSEWIC. 2005d. COSEWIC assessment and update status report on the spotted sucker, Minytrema melanops, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ot- 967 tawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/ files/cosewic/sr_spotted_sucker_e.pdf [accessed 2 October 2008]. COSEWIC. 2005e. COSEWIC assessment and update status report on the warmouth, Lepomis gulosus, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_warmouth_e.pdf [accessed 2 October 2008]. COSEWIC. 2005f. COSEWIC operations and procedures manual. April 2005. Committee on the Status of Endangered Wildlife in Canada. Ottawa, Ont., Canada. COSEWIC. 2006a. COSEWIC’s assessment process and criteria [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.cosewic.gc.ca/eng/sct0/ assessment_process_e.cfm [accessed 20 September 2007; updated 17 April 2009]. COSEWIC. 2006b. COSEWIC assessment and status report on the American eel, Anguilla rostrata, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/sr_american_eel_e.pdf [accessed 2 October 2008]. COSEWIC. 2006c. COSEWIC assessment and status report on the lake sturgeon, Acipenser fulvescens, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/cosewic/ sr_Lake_Sturgeon_0807_e.pdf [accessed 2 October 2008]. COSEWIC. 2006d. COSEWIC Assessment and status report on the Westslope cutthroat trout, Oncorhynchus clarkii lewisi, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_oncorhynchus_clarkii_lewisi_e.pdf [accessed 2 October 2008]. COSEWIC. 2006e. COSEWIC assessment and update status report on the deepwater sculpin, Myoxocephalus thompsonii, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_deepwater_sculpin_e.pdf [accessed 2 October 2008]. COSEWIC. 2006f. COSEWIC assessment and update status report on the river redhorse, Moxostoma carinatum, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/ files/cosewic/sr_river_redhorse_e.pdf [accessed 2 October 2008]. COSEWIC. 2007a. COSEWIC annual report [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/virtual_sara/files/ cosewic/car_COSEWIC_annual_report_2007_e.pdf [accessed 3 November 2009]. COSEWIC. 2007b. COSEWIC assessment and update status report on the northern brook lamprey, Ichthyomyzon fossor, in Canada [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ virtual_sara/files/cosewic/sr_ichthyomyzon_fossor_e.pdf [accessed 2 October 2008]. COSEWIC. 2008. Guidelines for recognizing designatable units below the species level [online]. Committee on the Status of Endangered Wildlife in Canada, Ottawa, Ont. Available from www.cosewic.gc.ca/eng/sct2/sct2_5_e.cfm [accessed 3 November 2009; 27 January 2009]. Government of Canada. 2008a. Schedule 1: List of wildlife species at risk [online]. Species at Risk Public Registry, Government of Canada, Ottawa, Ont. Available from www.sararegistry.gc.ca/ species/schedules_e.cfm?id=1 [accessed 22 July 2008; updated 9 January 2009]. Published by NRC Research Press 968 Government of Canada. 2008b. Species at Risk Public Registry [online]. Available from www.sararegistry.gc.ca [accessed 5 June 2008; updated 9 January 2009]. Green, D.M. 2005. Designatable units for status assessment of endangered species. Conserv. Biol. 19(6): 1813–1820. doi:10.1111/ j.1523-1739.2005.00284.x. IUCN. 2001. IUCN Red List Categories and Criteria. Version 3.1 [online]. International Union for Conservation of Nature Species Survival Commision, Gland, Switzerland, and Cambridge, UK. Available from www.iucnredlist.org/static/ categories_criteria_3_1 [accessed 15 November 2008]. IUCN. 2008. Summary statistics for globally threatened species [online]. International Union for Conservation of Nature, Cambridge, UK. Available from www.iucnredlist.org/static/stats [accessed 15 November 2008]. James, A.N., Gaston, K.J., and Balmford, A. 1999. Balancing the Earth’s accounts. Nature (London), 401(6751): 323–324. doi:10. 1038/43774. PMID:16862091. Keith, D.A., McCarthy, M.A., Regan, H.M., Regan, T., Bowles, C., Drill, C., Craig, C., Pellow, B., Burgman, M.A., Master, L.L., Ruckelshaus, M., Mackenzie, B., Andelman, S.J., and Wade, P.R. 2004. Protocols for listing threatened species can forecast extinction. Ecol. Lett. 7(11): 1101–1108. doi:10.1111/j.14610248.2004.00663.x. Master, L.L. 1991. Assessing threats and setting priorities for conservation. Conserv. Biol. 5(4): 559–563. doi:10.1111/j.15231739.1991.tb00370.x. Millsap, B.A., Gore, J.A., Runde, D.E., and Cerulean, S.I. 1990. Setting priorities for the conservation of fish and wildlife species in Florida. Wildl. Monogr. 111: 1–57. Can. J. Fish. Aquat. Sci. Vol. 66, 2009 Mooers, A.Ø., Prugh, L.R., Festa-Bianchet, M., and Hutchings, J.A. 2007. Biases in legal listing under Canadian endangered species legislation. Conserv. Biol. 21(3): 572–575. doi:10.1111/j.15231739.2007.00689.x. PMID:17531035. Prato, T. 2005. Accounting for uncertainty in making species protection decisions. Conserv. Biol. 19: 806–814. Regan, T.J., Master, L.L., and Hammerson, G.A. 2004. Capturing expert knowledge for threatened species assessments: a case study using NatureServe conservation status ranks. Acta Oecol. 26(2): 95–107. doi:10.1016/j.actao.2004.03.013. Regan, T.J., Burgman, M.A., McCarthy, M.A., Master, L.L., Keith, D.A., Mace, G.M., and Andelman, S.J. 2005. The consistency of extinction risk classification protocols. Conserv. Biol. 19(6): 1969–1977. doi:10.1111/j.1523-1739.2005.00235.x. Rohlf, D.J. 1991. Six biological reasons why the Endangered Species Act doesn’t work and what to do about it. Conserv. Biol. 5(3): 273–282. doi:10.1111/j.1523-1739.1991.tb00139.x. SARA. 2002. Bill C-5, an act respecting the protection of wildlife species at risk in Canada. [online] Available from laws.justice. gc.ca/en/s-15.3/text.html [accessed 14 May 2009; updated 19 May 2009]. Sokal, R.R., and Rohlf, F.J. 1981. Biometry. W.H. Freeman and Company, San Francisco, Calif. Appendix A Table A1 appears on the following pages. Published by NRC Research Press Lukey and Crawford Table A1. Freshwater fish species and designatable units (DUs) assessed by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) (2000–2007), with variables including expressions of uncertainty. Probability of designation under uncertainty Species–DU Petromyzontidae Northern brook lamprey – Great Lakes and upper St. Lawrence Morrison Creek lamprey Acipenseridae Shortnose sturgeon Lake sturgeon – Nelson River Variable AO EO AO EO AO AO LC MI PD AO LC MI PD EO AO MI PD AO MI AO AO LC AO LC AO LC EO AO PD MI MI Uncertainty >26, <31 km2 <100 km2 <50 km2 <500 km2 <200 km2 <40 000 km2 <5? <3 000 Close to 100% <250 000 km2 <5 Very few >50% SK and AB, >80% MBa >400 000 km <400 000 km2 <3 300 >98% <300 000 km2 Probably low 1 000s <1 000 km2 <100 000 km2 2+ <1 000 000 km2 >10 <600 000 km2 <70 <500 000 km2 12 000–30 000 km >50% <9 000 129 412–2 000 000 Min–Max 26–31 0–100 0–50 0–500 0–200 0–40 000 0–5 0–3 000 90–100 0–250 000 0–5 0–1 000 50–100 400 000–2 000 000 0–400 000 0–3 300 98–100 0–300 000 1 000–5 000 0–1 000 0–100 000 2–10 0–1 000 000 10–50 0–600 000 0–70 0–500 000 12 000–30 000 50–100 0–9 000 129 412–2 000 000 Algorithm designation NR EN NR 1.00 1.00 SC TH EN COSEWIC designation SC EN EN EN 1.00 1.00 SC EN Lake sturgeon – Red–Assiniboine rivers, Lake Winnipeg EN 1.00 EN Lake sturgeon – Saskatchewan River EN 1.00 EN Lake sturgeon – western Hudson Bay EN 1.00 EN Lake sturgeon – Winnipeg River, English River Lake sturgeon – Lake of the Woods Lake sturgeon – southern Hudson Bay, James Bay Published by NRC Research Press EN TH NR 1.00 0.48 0.80 0.52 0.20 EN SC SC Lake sturgeon – Great Lakes, upper St. Lawrence Green sturgeon White sturgeon Salmonidae Upper Great Lakes kiyi EN NR EN 1.00 1.00 TH SC 1.00 EN EN 1.00 SC 969 970 Table A1 (continued). Probability of designation under uncertainty Species–DU Bering cisco Shortjaw cisco Variable EO AO PD EO AO AO MI AO EO AO MI EO AO AO EO AO MI EO AO MI Uncertainty <1 000 km2 <1 000 km2 >30% >1Â106 km2 >175 000 km2 <30 000 km2 29 400–122 900 <2 000 km2 <500 km2 <100 km2 500–1 000 <30 000 km2 <20 km2 <0.03 km2 <100 km2 <20 km2 <1 000 <1 600 km2 <700 km2 Low thousands or hundreds <50 000 km2 <200 km2 Probably more than 10 000 <2 600 km2 <6 km2 <100 km2 <100 km2 12 000–94 000 Min–Max 0–1 000 0–1 000 30–100 1 000 000–5 000 000 175 000–875 000 0–30 000 29 400–122 900 0–2 000 0–500 0–100 500–1 000 0–30 000 0–20 0–0.03 0–100 0–20 0–1 000 0–1 600 0–700 100–5 000 Algorithm designation SC TH NR SC 1.00 TH EN COSEWIC designation SC TH 0.40 0.60 Westslope cutthroat trout – British Columbia Westslope cutthroat trout – Alberta Aurora trout NR EN EN 0.80 0.20 1.00 1.00 SC TH EN Cyprinidae Pugnose shiner Nooksack dace Catostomidae Spotted sucker EN EN EN 1.00 1.00 1.00 EN EN EN Ictaluridae Northern madtom EN 1.00 EN Cyprinodontidae Banded killifish – Newfoundland Percidae Greenside darter Published by NRC Research Press EO AO MI 0–50 000 0–200 10 000–50 000 NR 0.76 0.24 SC Can. J. Fish. Aquat. Sci. Vol. 66, 2009 NR 1.00 NR Cottidae Eastslope sculpin Cultus pygmy sculpin Gasterosteidae Limnetic Enos Lake stickleback EO AO EO AO MI 0–2 600 0–6 0–100 0–100 12 000–94 000 EN EN 1.00 1.00 TH TH EN 1.00 EN Lukey and Crawford Table A1 (concluded). Probability of designation under uncertainty Species–DU Benthic Enos Lake stickleback Misty Lake lentic stickleback Misty Lake lotic stickleback Variable MI MI EO AO Uncertainty 30 000–47 000 >10 000 lake, >4000 outlet stream <9 km2 0.008 to 0.04km2 Min–Max 30 000–47 000 14 000–70 000 0–9 0.008–0.04 Algorithm designation EN EN NR SC TH EN 1.00 1.00 COSEWIC designation EN EN EN 1.00 EN Note: Percent decline (PD), extent of occurrence (EO), area of occupancy (AO), number of locations (LC), and number of mature individuals (MI) were assessed for all combinations of the values at the minimum; 25th, 50th, and 75th percentiles; and the maximum. Probability of designation under uncertainty were calculated for Not at Risk (NR), Special Concern (SC), Threatened (TH), and Endangered (EN) and compared with (i) the predicted designation determined by the deterministic algorithm used in this study and (ii) the observed COSEWIC designation. a SK, Saskatchewan; AB, Alberta; MB, Manitoba. Published by NRC Research Press 971