Wkdivextinct Report 2018

ICES WKDIVEXTINCT REPORT 2018

ICES ADVISORY COMMITTEE

ICES CM 2018/ACOM:48

REF. ACOM

Report of the Workshop on extinction risk of MSFD biodiversity approach (WKDIVExtinct)

12–15 June 2018

ICES HQ, Copenhagen, Denmark

International Council for the Exploration of the Sea Conseil International pour l’Exploration de la Mer

H. C. Andersens Boulevard 44–46 DK-1553 Copenhagen V Denmark Telephone (+45) 33 38 67 00 Telefax (+45) 33 93 42 15 www.ices.dk [email protected]

Recommended format for purposes of citation:

ICES. 2018. Report of the Workshop on extinction risk of MSFD biodiversity approach (WKDIVExtinct), 12–15 June 2018, ICES HQ, Copenhagen, Denmark. ICES CM 2018/ACOM:48. 43 pp.

The material in this report may be reused using the recommended citation. ICES may only grant usage rights of information, data, images, graphs, etc. of which it has own- ership. For other third-party material cited in this report, you must contact the original copyright holder for permission. For citation of datasets or use of data to be included in other databases, please refer to the latest ICES data policy on the ICES website. All extracts must be acknowledged. For other reproduction requests please contact the General Secretary.

The document is a report of an Expert Group under the auspices of the International Council for the Exploration of the Sea and does not necessarily represent the views of the Council.

ICES WKDIVExtinct REPORT 2018 | i

Contents

Executive summary 1

1 Introduction ...... 2 1.1 Background ...... 2 1.2 Findings from the workshop WKDIVAgg ...... 3 1.3 Critical evaluation of proposals by WKDIVAGG ...... 8

2 Definitions of extinction risk ...... 10 2.1 IUCN red list ...... 10 2.2 Local v. global ...... 13 2.2.1 Global – regional – national scale ...... 13 2.2.2 How to deal with shifting baselines? ...... 14 2.2.3 Species vs. populations/stocks ...... 15 2.2.4 Natural and anthropogenic processes ...... 15 2.2.5 Consideration of variable life-history strategies ...... 17 2.2.6 Caveats ...... 18 2.3 Threatened species management, an example from Norway ...... 19

3 Methods to identify risks ...... 19 3.1 Le Mans modelling ...... 19 3.2 Geometric mean abundance and living planet index ...... 22

4 Inadequacy of monitoring data, especially in the Black Sea ...... 25

5 Integrating red list species in aggregated GES indicators ...... 26

6 Conclusions and Recommendations ...... 27 6.1 WKDIVAGG outcomes on aggregation, implications on extinction risk ...... 27 6.2 Detailing extinction risk ...... 27 6.3 Including extinction risk in GES and aggregation ...... 28 6.4 Conclusion ...... 28

7 References ...... 29

Annex 3: WKDIVExtinct ToRs ...... 31

Annex 2: List of participants ...... 33

Annex 3: Agenda ...... 34

Annex 4 IUCN summary of criteria sheet ...... 35

Annex5 WKDIVExtinct Report Reviews ...... 36

ICES WKDIVExtinct REPORT 2018 | 1

Executive summary

The Workshop on extinction risk of MSFD biodiversity approach (WKDIVEXTINCT) was set up to: a ) consider if the methods recommended by WKDIVAGG on integration of the status of each species, followed by integration of species to provide status of species group, could lead to the failure to alert of a potential extinction risk to a species. b ) if required, suggest adaptions or additional rules to alert the risk of extinction of a species within a species group. The workshop brought together 11 experts from 8 countries, chaired by Dave Reid, Ireland, and held at ICES HQ, Copenhagen, Denmark, 12–15 June 2018. The workshop examined both the preceding workshop WKDIVAGG and a number of issues surrounding the definition and determination of extinction, as well as the differences between “extinction” and “extirpation”, and global v. local extinction. The broad conclusion was that none of the proposed approaches to aggregation of yes/no GES assessments were fully satisfactory, as they would all tend to produce a large number of false alarms in terms of GES. In addition, it was concluded that the risk of extinctions could be considered as largely independent of the method of aggregation. The recommended solution to this was to develop a “red list” approach, broadly based on that of the IUCN red list. The IUCN red list provides criteria for es- tablishing an increasing level of threat to any species, and can also be applied in a local context as well as global. Advice on both is provided by the IUCN. Local “red lists” have also been compiled at European, RFMO, and EU MS levels. It was recognized that many of these species would not normally be included in routine monitoring and reporting due to lack of data. The workshop proposed that species under the highest level of threat “critically endangered”, should always be included in any aggregation to the species group level or higher. A species shown to have moved from a lesser category of threat to a higher one should also be included in the periodical GES assessment. Two approaches for including these species were proposed. Any species identified as above should be considered as being below GES. It could then be included in any chosen aggregation process based on WKDIVAGG etc., as for any other species. However, it would be “flagged” that this was the case, and that remedial action, or other management decisions could lead to a changed status. Alternatively, the aggregated GES for a species group could be determined, and “red flags” attached to highlight the status of one or more threatened species. In either case management action would be required. Essentially this approach represents a “safety net” to avoid extinction risks under any aggregation approach. The workshop also noted that the data needs for much of this approach were quite demanding, and that in many cases there would be insufficient data to make a full determination. In particular cases e.g. the Black Sea, data availability is quite minimal and this would be an issue for determining both extinction risk, and for GES by species in general.

2 | ICES WKDIVExtinct REPORT 2018

1 Introduction

1.1 Background The main background to the meeting was the findings from WKDIVAGG 1–4 May 2018. The outcome of this workshop is summarised in chapter 1.2. and represents the conclusions of that group, and not any evaluation of that report by WKDIVEXTINCT. Based on WKDIVAGG, WKDEVEXTINCT was asked to identify if there would be a “failure to alert of a potential extinction risk to a species” from applying the aggregation approaches proposed. WKDIVEXTINCT was also asked to “suggest adaptions or additional rules” if required. The Terms of Reference of the workshop, list of participants and agenda can be found in annexes 1,2 and 3.

ICES WKDIVExtinct REPORT 2018 | 3

1.2 Findings from the workshop WKDIVAgg In May 2018, the workshop WKDIVAGG met to consider the aggregation and integration approaches for the assessment of GES for species biodiversity under the MSFD. This can be summarised as in Figure 1.2.1.

Figure 1.2.1. Integration approach of species biodiversity indicators under the revised MSFD commission decision 2017.

The WKDIVAGG workshop provided guidance on an appropriate method(s): a) to integrate across criteria for each species of bird, fish and cephalopod (excepting commercial species and species on Habitats Directive annexes); b) to aggregate species within species groups for an overall assessment of status per species group for MSFD Descriptor 1; c) to aggregate from species group to the level of birds, mammals, reptiles, fish and cephalopods for an overall presentation of the extent to which GES has been achieved for these higher groups. WKDIVAGG considered the previous ICES advice on integration together with the new Commission Decision and suggested integration methods to integrate status from criteria to species, species to species group and from species groups to ecosystem component. Important considerations included the robustness of the methods to measurement error to avoid false alarms (recording a poor status where the actual status is good) and missed alarms (recording a good status where the actual status is poor), the applicability of the methods to data limited situations and the information inherent in specific patterns in species not in GES. The participants agreed that three aspects required careful consideration: i) the treatment and possible propagation of false alarms raised due to random measurement error; ii) the importance of patterns in status; and iii) the issue of data-poor species. A species for which an indicator is measured with low precision (high measurement error) will be particularly prone to false alarms (recording a poor status where the actual status is good) and missed alarms (recording a good status where the actual status is poor). In cases where precision is known, this can be incorporated in weighting procedures when determining the status at criteria level, but in many cases, the precision of the indicator is unknown and no correction can be made.

4 | ICES WKDIVExtinct REPORT 2018

Patterns in alarms convey important information. For example, having a poor status for a species in one year may not be a concern, but a series of poor status years will be. An example was suggested of a one-off flooding event reducing breeding success greatly in one year, as opposed to a consistently low breeding success during a longer period. Similarly, a low status of a species in one subregion may not be a concern but low status in a number of areas will. The amount of data supporting the status assessment differs greatly between species, both in terms of quality and the length of available time-series. It was considered high priority that the methods suggested should be applicable in data limited situations, except that where data are very limited, in which case it may be necessary to use expert judgement to derive estimates, e.g. of the coefficient of variation (CV). Further, the participants pointed to the inherent trade-off in the methods between sim- plicity/explainability/transparency and robustness to variation in data. It was encour- aged that methods should be applicable, practical and pragmatic. To limit the degree to which non-GES elements are accepted, WGDIVAGG considered a limit case where a number of indicators each with a median just above the level defining good status. These indicators each have almost a 50% probability of falling below the good status level, equivalent to flipping a coin. The probability of getting a specific number of ‘good status’ indicators can be readily estimated using the binomial distribution

Figure 1.2.2. For each number of successful species, the estimated probability of a species being successful is given on the y-axis together with the estimated 95% confidence interval.

Figure 1.2.2 shows an example of how the estimated probability of success changes with the number of ‘good status’ species. In this case, we know the true probability to be 0.5, a result which is unlikely (outside the 95% confidence intervals) when the number of success is 9 or 10. Hence, if we have 10 indicators with each almost 50% risk of falling below good status, the most likely outcome is that 5 will be good and 5 will not be good. It is unlikely that 9 or 10 of these will be good, but it is still not unlikely that 8 of them will be at good status. Conversely, to conclude that the probability of good of the 10 indicators is more than 50%, we require 9 or 10 to be good. Figure 2 shows the

ICES WKDIVExtinct REPORT 2018 | 5

lowest number that leads to reject a hypothesis of 50% probability of good status of individual indicators for sample sizes from 1 to 30. Below 5, there are no outcomes leading to rejecting the 50% chance of being of not-good, and hence if there are fewer than 5 elements (e.g. species), no elements are allowed to be at non-GES. The resulting number of required good status elements is given in table 1. This table shows the highest number of non-good that can be accepted (highest probability of being below good) and should therefore be used as a limit, not as a target for the number of ‘not good’ elements (e.g. of species in a species group).

Figure 1.2.3. Lowest number that is unlikely using this method for sample sizes from 1 to 30. Below 5, there are no unlikely results, and hence if less than 5 species occur, no species are allowed to be at non-GES. P0=<0.05, dashed line represents number of success according to the OOAO approach.

In conclusion, the method addresses the issue of measurement error to lower the occurrence of false alarms. However, this makes the method sensitive to missed alarms where patterns in species status exist, and to avoid this, the method needs to be supplemented by additional measures. WKDIVAGG concluded that probabilistic methods have a high potential for application within the MSFD, especially when integrating more than 5 elements. This implies that probabilistic methods may be most appropriate for determining proportional thresholds when integrating species within a species group (e.g. wading birds, demer- sal shelf fish, etc.). However, further testing is required to analyse the consequences of applying probabilistic integration methods with the available data beyond the limit case presented here. WKDIVAGG considered it key to consider uncertainty in the estimated indicator and status assessment in integration in order to avoid excessive numbers of false alarms resulting from high measurement error of one or more indicators. The group considered that to limit indicators to only those showing low measurement error would se- verely limit the coverage of rare species and ecosystem components and hence was not the preferred option to address the issue.

6 | ICES WKDIVExtinct REPORT 2018

Recommended integration methods The recommended integration methods for the different levels of integration are shown in the table below for HD species, D3 species, and other species. The group did not come to a conclusion on a single method for integrating criteria to species, recom- mending that further investigation and testing is required. Two options were considered appropriate for integration at one or more levels. Both methods contain an aspect of proportion of species in good status. From criteria to species, five different options were proposed, each of which should be evaluated to determine the sensitivity to false alarms and risk of missing alarms. It was considered essential to retain the information on the proportion of species which were considered relevant to the species group, but were not assessed due to lack of data. This often includes species which have previously been abundant but are now too scarce to assess. The issue of vagrant species and species on the edge of their distribution area was also discussed — in general, attempting to influence abundance of a species by managing human impacts at the edge of the distribution area is unlikely to produce measurable results. It is integral to the assessments that methods using proportional or averaging methods are supplemented by safeguards, monitoring whether the non-GES status species are consistent from one group, one area or one species through time. All integration options should be tested on real-world case studies (population dynamics datasets and/or Member States assessment reports) to test and conclude on advantages and disadvantages of each and to agree on a recommended option. However, the raw data for all indicators was not available in the group.

ICES WKDIVExtinct REPORT 2018 | 7

Table 1.2.1. Integration methods recommended

Level Integration method(s) From species HD species: OOAO. group to Option 1: ecosystem OOAO component Option 2: Proportion of all species in good status above agreed level (across species groups) From species to HD species: OOAO. species group Proportional (determined by probability) if number of species in the group >5 (see Table 7), OOAO if the number of species is 1 to 5.

From criteria to HD species: OOAO. species D3 species: As assessed under D3. (populations) Other species: One of options 1-5 below. If D1C1 and D1C2 are both in good status, then determine the average D1C2- D1C5, weighted so the weight of D1C2= weight of D1C3-C5 together. If this average is in good status, the species is in good status. If D1C1 and D1C2 are both in good status, then determine the average D1C3- D1C5. If this average is in good status, the species is in good status. If the weighted average of D1C1 to D1C5 is in good status, the species is in good status. The average is weighted to ensure that the weight of D1C2= weight of D1C1=weight of D1C3-C5 Conditional/OOAO Population model determines weights. This method can be used to indicate appropriate weights of each of the criteria. For example, for long-lived species, abundance may be little affected by one poor recruitment year and it may therefore be desirable to down-weight this. If criterion D1C1 is not used, for options 1 and 2, D1C2 should be in good status and then the weighted average is considered. For option 3, the weight of D1C2=weight of D1C3–C5 combined.

Table 1.2.2. Number of species required to be at good status for the species group to be at GES, based on the number of species in a species group

Number of Number of species required Number of Number of species required species in the to be at good status for the species in the to be at good status for the species group species group to be at GES species group species group to be at GES 1 to 5 1 to 5 18 13 6 5 19 14 7 6 20 14 8 7 21 15 9 7 22 16 10 8 23 16 11 9 24 17 12 9 25 17 13 10 26 18 14 11 27 19 15 11 28 19 16 12 29 20 17 12 30+ 21

8 | ICES WKDIVExtinct REPORT 2018

1.3 Critical evaluation of proposals by WKDIVAGG WKDIVAGG found that for the aggregations from species level to species-group level and from species-group level to ecosystem components, WKDIVAGG was successful in constraining the range of methods to consider to just two options: (i) based on a threshold for the proportion of elements at the lower level in GES, computed from the binomial distribution, (ii) one-out-all-out (OOAO). The proposed methods do not fully solve the problems associated with integration for the following reasons: i ) The proportion-based method requires knowledge of the probability of false alarms for the assessment of the lower levels. While these probabili- ties can be estimated when species-level assessments are based on just a single criterion (Greenstreet et al. 2012), this is not possible when species level assessments are derived in the more complex ways required by COMDEC (EU 2017). ii ) The rationale underlying the method of WKDIVAGG to compute the table of proportions is intransparent, and could lead to counter-intuitive results. For instance; 1 ) Confidence intervals should plausibly be one-sided : 2 ) The case n=30 is far from the asymptotic result (which is 50% of species at GES), so there should be no "+". 3 ) If one iterates the procedure for aggregation over two levels, the condition on the output of the first (95% quantiles) does not match the condition on the input of the second (50% quantiles). iii ) The methods are prone to false alarms. For example, for a system that is in GES, the probability of achieving an assessment of being in GES for a species group consisting of 6 assessed species can be as low as 11%[Computed in R using the command dbinom(x = 5,size=6,prob=0.5) + dbinom(x = 6,size=6,prob=0.5)]. This is particularly true for OOAO. In order to avoid false alarms at aggregate level, status assessment at species level need to be designed such as to generate false alarms with very low probability. This appears incompatible with Habitats Directive and CFP criteria as well as with the precautionary approach. WKDIVEXTINCT considers that the issues of estimations of probability around a GES status evaluation and that of dealing with “false alarms” still require further work. WKDIVEXTINCT tested the proportion-based method proposed by WKDIVAGG for the particular case of bird species (Table BIRDS). A test case where the ecosystem component is in GES and counterfactually assumed that the probability p for false non-GES assessments of bird species is identical across species and known to be p=0.85 was considered. From this, we computed the probability of obtaining the correct assessment of GES through a two-step integration (bird species to species groups, then groups to the bird ecosystem component). The simulation code is shown in Box CODE. The results (listed in Table BIRDS) depend on the number of assessed species considered for each ecosystem component, and therefore differs between regions. In the Arctic, the assessment “GES” would be obtained with only 8.8% probability, in the North Sea with 61.4% probability. There is no simple relation between the total number of species assessed and the probability of a correct assessment. Such dependencies, next to the uncertainty regarding the actual values of p, might make interpretation of assessment results difficult.

ICES WKDIVExtinct REPORT 2018 | 9

Region Total number of assessed Probability of correct assessment species in bird ecosystem “GES” provide system in GES component

Arctic 28 0.088

Baltic 51 0.410

Celtic Seas 59 0.281

OSPAR Greater North 83 0.614 Sea

Box CODE. R code used to generate Table BIRDS

### Consider a situation in which everything is at GES, but assessments of ### species do not always give GES due to measurement error and natural ### fluctuations.

### We look are the case of birds. ### Here are the numbers of species assessed in each species group: ### Numbers from Bryone (pers. comm.) Arctic <- c(2,5,14,5,2) NS <- c(28,20,15,10,10) CS <- c(19,13,11,8,8) Baltic <- c(7,14,12,9,9)

### This generates the table provided by WKDIVAGG n.required<-data.frame(n=qbinom(p=0.975,size=seq(1,40),prob=0.5))

p.species.assessed.ges <- 0.85 ## input species.group.size <- NS; ## input

n.required[species.group.size,1] ## min # of GES species required

## probability for this to happen at GES p.species.group.assessed.ges <- 1-pbinom(n.required[species.group.size,1]-1, size=species.group.size, prob=p.species.assessed.ges); p.species.group.assessed.ges

### For integration to ecosystem component, use ### OOAO because there are 5 or less species groups: p.ecosystem.component.assessed.ges <- prod(p.species.group.assessed.ges)

round(p.ecosystem.component.assessed.ges,3) ## print the result

10 | ICES WKDIVExtinct REPORT 2018

2 Definitions of extinction risk

2.1 IUCN red list The IUCN Red List Categories and Criteria (IUCN 2012a) are intended to be an explicit, objective and easily understood framework for classifying species at high risk of global extinction. Global extinction has occurred when exhaustive surveys in known and/or expected habitat of a taxon, at appropriate times (diurnal, seasonal, annual), through- out its historic range have failed to record an individual. Surveys should be over a time frame appropriate to the taxon’s life cycle and life form. The IUCN criteria can also be applied to taxonomic and population units below the species level and for this reason IUCN use the term “taxon” rather than “species” to describe the taxonomic scale of categorization and has developed specific guidelines for application of IUCN Red List Criteria at regional levels (IUCN 2012b). The IUCN quantification of extinction risk involves five different criteria that are used to allocate species into 9 different Red List categories among which two contain species that are either Extinct (EX) or Extinct in the Wild (EW), three contain species that are threatened by extinction by being either critically endangered (CR), endangered (EN) or vulnerable (VU), two contains species that are not considered threatened by being either near threatened (NT) or considered to be of least concern (LC); and two are data deficient species (DD) or species that have not been evaluated (NE) (IUCN 2012a), Fig- ure 2.1.1. To ensure a consistent use of these criteria and help evaluators and data sup- pliers, a booklet of IUCN Red List Categories and Standards (IUCN 2012a) as well as more recently updated Guidelines for Using the IUCN Red List Categories and Criteria have been published (IUCN 2017).

Figure 2.1.1 The IUCN Red List Categories. From IUCN (2017)

The five criteria for assessing whether a taxon belongs to a threatened category can briefly be summarized as:

ICES WKDIVExtinct REPORT 2018 | 11

1 ) Population size reduction measured over the longer of 10 years or three generations, 2 ) Geographic range in the form of either the extent of occurrence and/ or area of occupancy below a specified limit, and at least two out of three additional conditions regarding: a) coherence of distribution range, b) continuing decline in distribution range, c) temporal fluctuations in either of these. 3 ) Small population size of mature individuals and either observed or projected continuing decline that is either above a specified rate, or has reduced the population of mature individuals below a certain level. 4 ) Very small or restricted population of mature individuals, 5 ) Quantitative analyses of the probability of extinction risk providing high probability of extinction. To list a particular taxon in any of the categories of threat requires only one of these criteria to be met. However, a taxon will be assessed against as many of the five criteria as the available data allow, only to be placed in the highest category of threat that the criteria permit. Although being based on quantitative criteria the system is deliberately flexible, to allow taxa to be assessed even though very little information is available. A more elaborated summary sheet of the thresholds used in connection with the criteria is presented in Annex 1. Note, however, that the IUCN Guidelines (IUCN 2017) contain much additional information necessary for the correct application of the criteria in actual assessments, and that IUCN therefore stresses that the summary sheet cannot be used for assessing threat status. Assessments should and must be based on the text provided in the latest version of the IUCN Red List Categories and Criteria (IUCN 2012a), the IUCN Guidelines for Using the IUCN Red List categories (IUCN 2017) and the Guidelines prepared by the IUCN SSC Regional Applications Working Group (e.g. IUCN 2012b). Regarding taxonomic scale it is important to note that the taxonomic unit used should be specified, and that taxonomic units below the rank of variety should not be assessed, with the exception of the assessment of subpopulations. Before taxa below the level of species can be included in the Red List the assessment of the full species is however required. If a subpopulation exchanges individuals with other subpopulations its assessment must follow regional guidelines, but these should again not be applied at a very small geographic scale. Note also that the categorization process should only be applied to wild populations inside their natural range, and to populations resulting from benign introductions. Note also that there are specific guidelines pertaining to fisheries (IUCN 2017, p.66). The data used in the assessment are often uncertain due to natural variability and measurement error. When using such data the recommendation is that the attitude of the assessor should be precautionary but realistic (IUCN 2017). There are, however, considerations in IUCN (2017) as to whether the categorization process will be sufficiently responsive under climate change, in particular whether the time frames in the criteria are too long to ensure that affected species will be adequately categorized to trigger appropriate and timely management action. Some preliminary analysis suggest that the assessment criteria allow for sufficiently rapid assessment, but further guidelines on how to assess species subject to climate change are never the less provided. Unfortunately, the guidelines do not include provisions for dealing with species that move outside their natural area of occurrence due e.g. to environmental or climate

12 | ICES WKDIVExtinct REPORT 2018

change, and start to appear in a new region. The IUCN does not categorize non-indigenous species with respect to extinction risk. The European Red List of Marine Fishes (Nieto et al. 2015) and the European Red List for Mammals (Temple and Terry 2007), for example, does only considers fish species that are native or have become natural- ized since before AD 1500. This may be important for assessing local extinction risk for species that have or will become established in the marine waters of the EU member states, and in particular for areas bordering the marine waters of non-EU states, e.g. in the Mediterranean and along the Portuguese and Spanish Atlantic coasts (ICES Sub- area IX). It is finally important to remember that the designation of species to different categories of threat cannot necessarily be used to determine the priority of conservation actions. The category reflects the current situation, whereas numerous other factors not included in the assessment will be necessary to determine the priorities of conservation actions. These include considerations pertaining to the costs, logistics, and likelihood of success of different conservation strategies for rebuilding threatened populations.

The European Red Lists The EU has supported the development of European Red Lists for a number of taxonomic groups. These Red Lists complement the reporting under the Habitat and Birds Directives because they consider all species within a taxonomic group, not just the species listed in the Annexes of the Directives and protected by EU legislation. All vertebrates in Europe have presently been assessed according to the IUCN criteria. These assessments have resulted in the categorization shown in Table 2.1.1. The critically endangered marine fish species are all chondrichthyans, and so are many of the endangered and vulnerable marine fish species. One marine mammal, the grey whale, Eschrichtius robustus, is Regionally Extinct. It formerly occurred in the North Atlantic and adjacent waters, but was extirpated by hunting. Among the six threatened marine mammals the northern right whale, Eubalaena glacialis, and the Mediterranean Monkseal, Monachus, are Critically Endangered, the sei whale, Balaenoptera borealis, and the blue whale, Baleanoptera musculus, are Endangered, and the sperm whale, Physeter catodon, and the harbour porpoise, Phocaena, are Vulnerable. The two threatened cephalopods are Opisthoteuthis calypso and Opisthoteuthis massyae.

ICES WKDIVExtinct REPORT 2018 | 13

Table 2.1.1. The number of Marine Fish, Birds Mammals and Cephalopods on the European Red List. Based on data in 1Nieto et al. (2015) for the European region, 2Birdlife International (2015) for the EU area, 3Temple & Terry (2007) for the European region and 4http://discover.iucnredlist.org/ also for the European area. Where available the numbers in brackets show the number of the species in the category that are endemic to Europe. The selection of birds on the European Red List that can be characterized as marine was done by Professor Mark Tasker (pers. comm.).

Marine Red List Category Marine Fish1 Marine Birds2 Cephalopods4 Mammals3 Extinct 0 0 0 0 Extinct in the Wild 0 0 0 0 Regionally Extinct 0 1 1 0 Critically endangered 15(2) 5 2 0 Endangered 22(4) 11 2 0 Vulnerable 22(2) 24 2 2 Near threatened 26(3) 10 1 0 Least concern 699(120) 115 7 39 Data Deficient 204(53) 0 12 34 Total 988(184) 176 27

2.2 Local vs. global In the frame of the MSFD, assessments have to be made on a national and on a regional scale. Sufficient data are needed to make indicators operational. Rare species are usually those for which the least data are available. This would include both, naturally rare species and species which have become increasingly rare over time (and thus may be categorised as threatened). Sensitive species or species that react to anthropogenic impact are required to be explicitly taken into account in the assessment by the revised Commission Decision (COMDEC 2017, p. 73).

2.2.1 Global – regional – national scale The spectrum from endemic species with a small distribution range to wide-ranging marine species must be accounted for. It complicates the management of ocean wildlife as species often traverse multiple management jurisdictions. The great mobility of many marine animal species may help them to better follow their climatic envelope and to colonize and recolonize habitats, as long as source population refuges are kept available (MacCauley et al. 2015). Extinction can be viewed in different scales from national/regional to global of which global extinction is the worst. IUCN consider in their red list, both the global scale and where appropriate the population scale. National red lists mainly focus on the status of a species (or a population) in national waters. Also the Regional Seas Conventions have their red lists. Whereas the term “extinction” often is used to refer to the death of the last individual of a taxon, e.g. a species, several other levels of extinction have been defined. For the marine environment McCauley highlighted three categories including local, ecological and commercial extinction. Local extinction, interchangeably used with “extirpation” refers to a local or regional extinction. This could for example be the result of a range contraction. Such contractions can result from the direct elimination of vulnerable local subpopulations or from region-wide declines in abundance (McCauley et al. 2015). Because a species' potential range may be very large, especially in migrating marine animals, a range contraction

14 | ICES WKDIVExtinct REPORT 2018

can be seen as an early sign of deteriorating conditions. This could be due to natural or anthropogenic reasons. Range contraction is a well-known phenomenon from terrestrial animals and might become of increasing importance in marine species (cf. McCauley et al. 2015). These are arguments for assessing the status on the basis of populations rather than species only (see section 3 of this chapter) and for assessing in the whole distribution area instead of concentrating on species’/populations’ core areas only. When considering varying spatial scales, the IUCN 2012 guidance states “Populations of long-lived individuals that have ceased to reproduce within the region (e.g. as a result of a deteriorating environment) should be regarded as potentially capable of reproduction and consequently should not be classified as RE (regionally extinct).“ (IUCN 2012). The capacity to reproduce should in general not be a factor determining “extinction”. However, the capacity to breed or recover may have been lost before the actual extinction. For this reason, the terms ecological and functional extinction (used as synonyms) are useful: Ecological or functional extinction describes a situation in which the abundance has decreased to values at which a depleted species cannot perform its functional role any more. Reductions in the abundance of some marine animals have been well documented. Marine vertebrates (fish, seabirds, sea turtles, and marine mammals) have declined in abundance by on average 22% in the last four decades. Of these, certain baleen whales have declined by 80 to 90%. Such ecological extinctions are well known in terrestrial environments and have been demonstrated to be just as disruptive as species extinctions (McCauley et al., 2015). Commercial extinction of a stock is reached when the further exploitation of the stock is not considered economically viable.

2.2.2 How to deal with shifting baselines? In general, the distribution and abundance of a population are positively correlated (e.g. Rindorf & Lewy 2012). Threats or impacts on a population become first visible at the periphery of its range. It would be a false concept of protection to concentrate only on core distribution area due to shifting baselines during a decline. Changes occurring at the margins result in a decreasing distribution area. Dependent on the selection of the baseline period the relation to the size of the actual distribution can vary a lot (this can be shown for various sharks and rays in the North Sea). This factor is accounted for in the revised Commission Decision 2017/848 in the relevant criteria:  D1C4 (primary criterion): The species distributional range and, where relevant, pattern is in line with prevailing physiographic, geographic and climatic conditions.  D1C5 (primary for HD species): The habitat for the species has the necessary extent and condition to support the different stages in the life history of the species. This implies that also the margins belong to the distribution area and species occurring on the fringe of their distribution must also be taken into account in national and regional assessments. Further, a species is considered established if at least one develop- mental stage of the species regularly has a partial habitat in the area or is a regular migrant. So not only areas with regular reproduction determine whether a species should be assessed (Thiel et al. 2013).

ICES WKDIVExtinct REPORT 2018 | 15

2.2.3 Species vs. populations/stocks It is important to maintain biodiversity for many different reasons. This includes the full range of genetic diversity of which genetically differentiated populations or subpopulations are possible scales for assessment purposes. One of the main reasons for maintaining biodiversity is to provide the resilience needed to withstand and react to environmental changes which can occur naturally or can be driven by anthropogenic factors (such as climate change). Also it is possible that within populations certain inherited traditions of complex learned behaviours can be conserved such as in killer whales (Orcinus orca). Further, there is incomplete understanding of potentially valuable ecosystem services provided by all taxa. In range contractions, changes often become visible in vulnerable populations or subpopulations on the margin of the distribution area. Thus it is important to take populations into account when assessing the status of a species. One population cannot simply replace another which has undergone extirpation. De- pendent on distance and mobility, individuals of another population can reach an abandoned area. If population pressure is high and the dispersal is large enough it is possible to recolonise the area formerly colonised by the extinct population. It is also possible that other species occupy the niche in the meantime and also that the “replacing” population would occupy a different niche. The latter is currently being observed in a terrestrial ecosystem in eastern corn buntings (Emberiza calandra) recolonising an area abandoned by the extirpated western population of the same species. However, individuals of the replacing population now live in pastureland whereas the extirpated population was native to fields. Fish are managed in stocks for practical reasons which may not reflect underlying differences in populations. North Sea cod stock consists of 3 populations (genetic differences but managed as only 1 stock) (Holmes et al 2008), also whiting which is managed in the North Sea as one stock has a number of genetically distinct populations (de Cas- tro et al 2013). In sandeel it is the contrasting situation. This species is managed in 6 stocks in the North Sea which do not have different genetics (). As a consequence, defining conservation targets for stocks on the basis of underlying populations is not a straightforward task.

2.2.4 Natural and anthropogenic processes Ecological communities naturally turn over their species composition. Numerous studies have demonstrated continuous turnover of community composition, implying local extinctions, while total species richness remains unchanged (Magurran et al. 2015; Brown et al., 2000; Parody et al., 2001; Magurran et al., 2018; Dornelas et al., 2014; Gotelli et al., 2017). Representative examples are show in Figure 2.2.1 for marine fish and Fig- ure 2.2.2 for a global meta-analysis. It is important to realise that change in community composition, including local extinctions, can be a natural and desirable response of ecosystems to environmental change. Without turnover, communities would be less resilient. Yoccoz et al. (2018) conclude from these observations: Studies of single species or single groups may therefore lead to biased understanding or prediction of diversity changes, since other species or groups may show opposite responses.

16 | ICES WKDIVExtinct REPORT 2018

They warn that “without an increasing effort, we may often wrongly extrapolate diversity changes based on single measures, regions, or emblematic taxonomic groups.” Model simulations show that natural turnover (e.g. decline of Jaccard Similarity) of species tends to be faster in local communities than in the metacommunity they are part of (Figure 2.2.3). At the same time, local communities can maintain their species richness even when richness at metacommunity level declines (Figure 2.2.3). From this point of view, the question if “biodiversity is maintained” (Descriptor 1) is best assessed at the largest spatial scale at which management measures can be enacted. At smaller scales, natural species turnover and anthropogenic decline of biodiversity can be hard to discern.

Fig 2.2.1, Temporal trends in αdiversity (rarefied species richness) and β diversity (Jaccard similarity, relative to the initial year), in each latitudinal band, over the duration of study. Trend lines (OLS regression) are colour coded red if significantly negative (P <0.05, n = 28) and grey if not significant. There were no significant positive slopes. (from: Magurran et al. 2015).

ICES WKDIVExtinct REPORT 2018 | 17

Fig 2.2.2. A meta-analysis by Dornelas et al. (2014) establishes a global pattern of progressive change in community composition (implying local extinctions) without changes in species richness.

Figure 2.2.3. Output of a metacommunity model simulation of species turnover (Dinner O'Sullivan, 2018). At each time-step, a new species is added to the metacommuity and population dynamics are simulated to equilibrium. Both regional richness (black line) and local richness (red, purple, blue) remain stable in this process, after an equilibrium has been reached (left panel, T < 500). The change in species composition (decline of Bray-Curtis similarity) is faster at local than at regional scale (right panel). Progressive removal of one species per time-step from the metacommunity (left panel, T > 500), starting with the rarest, notably affects local richness only after substantial biodiversity loss.

2.2.5 Consideration of variable life-history strategies When assessing extinction risk, differences are to be expected between e.g. species vagrant in a large area, indigenous species, and endemic species. It may be necessary to consider variable management responses as a function of differ- ing species characteristics. Selective forces, both natural and anthropogenic, act at multiple scales that are perceived at different scales of not just space, but also through time. Some examples for consideration include  r vs. K selected species  short vs. long lived species  small vs. large bodies species  cultural preferences

18 | ICES WKDIVExtinct REPORT 2018

Despite technically different characteristics, there are common examples that trans- cend the above categories such that they often fall on parallel spectrums with extreme examples being the great whales which are K-selected, long lived, large bodied species with strong cultural significance whereby social pressure can exist for preservation at the level of the individual. In contrast, forage fish perhaps fall at the other end of the spectrum for each character. It should be acknowledged that some species perhaps do not fall equally across these continuum, for example long lived deep-sea corals or r- selected large bodied tuna species.

2.2.6 Caveats MSFD works principally at MS and regional sea levels, and so extinctions recorded in monitoring programs might be only local, even when globally or regionally a species is not extinct. Local extinction, also called extirpation, may occur through natural processes, see Ma- gurran et al. (2015). The reasons and consequences of a local extinction are various but can be illustrated by two idealized examples: a local extinction caused by an unfavourable change in environmental conditions (whether caused by natural or anthropogenic processes) or an extinction caused by competition with a newly invading species (in practice, the two mechanisms can reinforce each other). The first example represents a situation where the extirpation may cause a change in ecosystem functioning. A shift into unfavourable environmental conditions could be crucial for species occurring on the edge of their distribution range, for example Atlantic cod in the Baltic Sea. In that case a slight hydrological change could improve or deteriorate conditions for its reproduction. For species which play an important role for an ecosystem, e.g. the extinction of a habitat builder or a top predator, this might affect the entire foodweb. In the second case, competition with an invading species, there is a high probability that the species’ functional role will be taken over by the newcomer. Local extinction of a species might also have wider consequences for the marine ecosystem. For example, when it affects the connectivity between different local populations, the species’ gene pool might become depauperate. Local extinction may also represent loss of subspecies. While globally the species may not be at risk of extinction, the local population might represent a unique set of characteristics, e.g. behaviour, biology, ecological interactions or even social aspects (existence value). In addition, any local extinction might be a step towards global extinction, as the species’ range is become more restricted. The relevance of a local extinction also depends on the condition of the ecosystem it inhabits. A complex and healthy state of the ecosystem gives resilience to changes occurring in the environment. An ecosystem that is more degraded and under a higher pressure has a lower potential to adapt to species loss. One way to assess the severity of a local extinction is to estimate the time it would take for the ecosystem to recover from the loss after anthropogenic pressure have been re- moved (Rossberg et al. 2017). Prior to local extinction, the recovery potential of a species depends on duration and complexity of its life cycle, its mobility, biology or even behaviour. Life-history traits that favour recovery are small-bodied individuals, fast growth, high fecundity etc. After an extirpation, community recovery might require recolonization by the extirpated species from areas where the species is still extant. Recovery time then depends on factors affecting the potential for recolonization, including:

ICES WKDIVExtinct REPORT 2018 | 19

1 ) Whether remaining extant populations are in the vicinity and not separated by barriers to migration or propagule transport (eggs, larvae etc.) 2 ) The size of remaining extant populations 3 ) Whether the local community is in a favourable state, e.g. absence of competitors occupying the species’ niche, and presence of suit- able habitat conditions and habitat providing species, e.g. seagrass beds for seahorses

2.3 Threatened species management, an example from Norway Norway is one of few countries which use the IUCN red list as a base for fishery management actions. In Norway, all species which are landed (including data-poor species) are evaluated every five years against the IUCN Red List Categories and Criteria by the Norwegian Biodiversity Centre. The Directorate Fisheries then evaluate present harvesting levels and considers the need for regulatory interventions (Gullestad et al. 2017).

3 Methods to identify risks

3.1 Le Mans modelling

Introduction Robert Thorpe presented some results from a multispecies management strategy evaluation in the North Sea, in which management outcomes are jointly evaluated in terms of risk and reward. In this study, a stock was defined to be at risk if its biomass fell below 10% of its unexploited biomass (virgin biomass or B0), but other thresholds could be used including complete extinction (biomass=0 for all future time, or on an exponential decay trajectory). Whilst the risk scores were used to help constrain multispecies MSY and inform management trade-offs in this study, they could be used as a proxy for “extinction risk” if suitably defined, and might show whether risks were increasing with time or whether some stocks were disproportionately at risk.

Model framework The North Sea LeMans model is a size-structured multispecies model that represents the North Sea fish community in terms of 21 species structured by size (Thorpe et al., 2015, 2016, 2017). There are 32 possible size classes which are common to all stocks. Thus the longest fish, cod, may be present in all length classes, while smaller species such as sprat may only be present in, say, five. Individuals progress through length classes as they grow and mature at a threshold length at maturity. Reproduction is described with a hockey-stick spawner recruit relationship (Barrowman and Myers, 2000), which determines the number of recruits entering the smallest size class as a function of the biomass of mature individuals. Species’ dynamics are linked via predation mortality (M2) which is an emergent property depending upon predator and prey abundance as determined by a diet matrix and the preferred predator/prey weight ra- tio. Individuals are also susceptible to residual natural mortality (M1) and fishing mortality (F). Parameterisation and validation of the model are described in Thorpe et al. (2015).

20 | ICES WKDIVExtinct REPORT 2018

The consequences of parameter uncertainty were assessed by developing 78125 models (5 variants for each of 7 key parameters covering i) diet matrix, ii) predator size-pref- erence, iii) non-predation natural mortality, iv) spawner-recruit initial steepness, v) ef- ficiency of growth, vi) asymptotic length, and vii) maximum recruitment), with parameters drawn from ranges that spanned literature estimates. Details of the parameter choices and their underlying rationale can be found in Thorpe et al. (2015), Supple- mentary Table S1. Models in the unfiltered ensemble were screened against stock assessment estimates of SSB to identify plausible models. The screening criteria were i) all species should persist in the absence of fishing, and b) the mean predicted SSB of assessed species after 30 years simulated fishing at average 1990-2010 rates (ICES, 2012), should be within a factor of two of the SSB estimated in ICES (2012). Stochastic recruitment around the deterministic hockey-stick spawner-recruit relationship was simulated using a lognormal distribution scaled so as to both preserve mean recruitment levels and reproduce variability of similar magnitude to that found in the ICES stock–recruit database (see Thorpe et al., 2017, Supplementary Information). Each parameter combination was tested three times, and accepted if all three simulations were plausible, giving rise to a filtered ensemble (FE) of 63 members. We used a factor of two because biomass estimates from assessments are uncertain, and because a range of processes including environmental factors influence abundance in the real world. However, a previous analysis suggests that changing this factor, which affects the FE size, has only modest impacts on predictions made by the filtered ensemble for FE sizes between 50 and 1000 (Thorpe et al., 2015, Supplementary Information).

Evaluating multispecies MSY Using the LeMans model as the operating model, we ran the MSE simulations for 50 years and defined management outcomes in terms of an average of the last 30 years of the simulation. We focused on the expected outcomes for risk and reward, and present results showing the gross economic yield (expected catch x value), and risk of stock depletion adjusted to reflect the need to avoid concentrating risk on a few stocks in the community to an unacceptable degree. This is the multispecies analogue of the aggregation issues addressed by WKDIVAgg. By defining a community risk in the manner described we can be sure it meets the ICES definition of precautionary (i.e. if any single stock has a high chance of depletion, the community risk will also be high, irrespective of the risks to the other twenty stocks). This is the equivalent of saying we will take note of extinction risk to a single stock, however many stocks are in the community. We defined stocks to be depleted when their biomass was below a certain fraction of the estimated unfished biomass (B0). In previous work we have defined a stock to be at risk when its biomass falls below 10% of B0 (Thorpe et al., 2016, 2017), but other definitions have been suggested (Smith et al., 2009) so we also consider 15% and 20% of B0. Results were expressed as the ensemble mean (across 63 ensemble members). Each calculation was repeated 100 times to take account of stochastic recruitment variation (Appendix; Thorpe et al., 2017). We considered 5 potential MSY candidates, one based upon single species assessments from 2012, one based on a 21-stock Nash equilibrium (Thorpe et al., 2017) and three based upon the bottom, middle, and top of the ICES PGY ranges. Expected long-term outcomes are shown in Figure 3.1.1 for “at risk” when B< 0.1 B0 (left) and B < 0.2 B0 (right).

ICES WKDIVExtinct REPORT 2018 | 21

Figure 3.1.1: Long-term expected outcomes for strict (left) and less strict (right) definitions of “at risk” when managing the 21-stock community to one of 5 different candidates for a multispecies MSY. Outcomes are expressed relative to the ICES 2012 single species assessments for lower (green), middle (cyan) and upper (magenta) PGY ranges, and for the Nash equilibrium (gold). Out- comes in the bottom right quadrant are better, in the top left are worse, and in the other two quad- rants involve trading off risk and reward.

We found that the managing to the middle of the PGY ranges was (unsurprisingly) similar to the ICES 2012 assessments, the bottom of the ranges was safer, and the top was worse. The multispecies Nash equilibrium was however better than the single species assessments, particularly for strict definitions of “at risk”. The results show the potential of the risk/reward approach to evaluate multispecies outcomes.

Further model development Since the MSE was carried out, we have further developed the model framework to incorporate i) food dependent growth (i.e. growth rates that depend upon prey availability and/or environmental factors), ii) stock recruitment relationships that allow for unlimited recruitment, iii) additional stocks and fleets, iv) consideration of spatial overlap, v) fitting to survey rather than assessment data and vi) predation from seals and cetaceans. These developments make the model framework more useful for answering questions of extinction, because they increase the number of mechanisms that can be considered (Table 3.1.1).

22 | ICES WKDIVExtinct REPORT 2018

Table 3.1.1: Some possible mechanisms behind species extinction, and how they can be addressed in the modelling framework

.Issues for future consideration The multispecies model framework presented here has the potential to be useful in addressing estimation of extinction risk, provided we are clear about what is meant by “extinction”, whether it means actual extinction (zero biomass) or depletion to low levels (a small percentage of virgin biomass, say 1%). We also need information on how species distributions are likely to change over time (from species habitat distribution models), changes in the primary productivity of the ecosystem in question, candidate invasive species, the temperature sensitivity of metabolism, potential links between the environment and recruitment, and whether there is likely to be an increase in anoxic zones in the future.

3.2 Geometric mean abundance and living planet index As an alternative method for aggregations of species status information to species groups and to ecosystem components, WKDIVEXTINCT considered the indicator pop- ularly known as Living Planet Index (LPI). The indicator is defined as the geometric mean of the abundances of a defined set of species (“Geometric Mean Abundance”). The indicator is normalized to 100% at either the start or the end of the indicator time- series. An example is show in Figure 3.2.1. The proportional rates of growth (or decline) of the indicator equals the average rates of increase or decline of the component species, over both short and long time periods. If, due to natural variability or in response to environmental change, some populations increase in abundance while others decline, such that overall “biodiversity is maintained” in accordance with Descriptor 1 of the MSFD, the indicator does not change its value. A statistically significant decline of the indicator, on the other hand, can result from pressures on the ecosystem, such as overexploitation or loss of habitats, resources, or subpopulations, that are likely to go along with regional biodiversity loss and contribute to biodiversity loss globally. The LPI was initially developed by the Zoological Society of London for the Worldwide Fund for Nature’s (WWF) Living Planet Report (Buckland et al 2011) which is produced biannually and submitted to the Convention of Biological Diversity to provide a global

ICES WKDIVExtinct REPORT 2018 | 23

index on biodiversity trends. It now forms part of the suite of “Indicators for the Stra- tegic Plan for Biodiversity 2011-2020 and the Aichi Biodiversity Targets” (CBD, 2016). Other uses of LPI-like indicators include the annual Arctic Species Trends Index (Barry and Helgason 2017) from the Circumpolar Biodiversity Monitoring Program of the in- tergovernmental Arctic Council, Environmental Data Compendium of the Nether- lands Statistical Office (Figure 3.2.1), and the assessment of wild-bird populations in the UK (https://www.gov.uk/government/statistics/wild-bird-populations-in-the-uk). The European Topic Centre – Inland, Coastal & Marine is also developing the use of the LPI for in the forthcoming European Environment Agency’s thematic marine biodiversity report. Among the strengths of the indicator are its easy interpretation and the possibility to aggregate or disaggregate it depending on the species groupings of interest. However, the indicator suffers from the same challenges as other indicators based on averages or geometric averages when applied to species groups consisting of a low number of species: high impact of highly variable species indicators makes the indicator sensitive to single species with large change in numbers and to species for which abundances are estimated with large uncertainty, and the ability for several not-GES species to be masked by a single species with a very high abundance. To address the latter issue, if used for communication, this indicator should as a minimum be accompanied by a figure showing the proportion of species in good and not-good status together with the proportion of species not included due to lack of data.

Figure 3.2.1. LPI Trend in Netherlands North Sea Marine fauna, 1990-2015, Biodiversity indicators for the Netherlands Environmental Data Compendium (http://www.clo.nl/indicatoren/nl1575- trend-mariene-fauna----living-planet-index)

In order to determine environmental status from an LPI time-series, one can ask if, in the assessment year, the indicator has fallen by a statistically significant amount below the value from any previous assessment year. When the indicator value for the final year is fixed at 100%, the relevant confidence interval of the indicator time-series can easily be computed using the standard bootstrapping technique (Fig 3.2.2).

24 | ICES WKDIVExtinct REPORT 2018

Figure 3.2.2. Illustrative examples or LPI time-series (thick line) with confidence intervals (shaded area) in comparison with the final indicator value (dotted line). The case on the right would be assessed as GES, the case on the left as not GES, because the rate of change in the final year is significantly smaller than zero and the final point lies significantly below the values during the period 2000-2005.

ICES WKDIVExtinct REPORT 2018 | 25

4 Inadequacy of monitoring data, especially in the Black Sea

The assessment of extinction risk of the Black Sea species in relation to MSFD criteria requires availability of long-term monitoring data. However, no regular datasets on fish species, birds and marine mammals are available to assess extinction risk at regional level for the Black Sea in spite of the framework for conservation actions provided by the Bucharest Convention, the Biodiversity and Landscape Conservation Protocol and the Black Sea Strategic Action Plan. The existing formal IUCN assessments from 2005 – 2009 are not updated due to the serious limitations or deficiencies of existing time-series datasets. Because some of the Black Sea species represent subpopulations of Mediterranean or Atlantic population, their regional conservation status might be different from that in other areas. However, population identity is not always clear For the assessment of abundance and extinction risk of the Black Sea species, regular basin-wide monitoring surveys are needed. All three ecosystem components (fish, birds and marine mammals) in the Black Sea are currently inadequately covered by monitoring programmes at regional level. Monitoring programmes, where these are available, focus mainly on commercial fish species. Further, monitoring of species under Descriptors D1 and D3 of MSFD and CFP is only required for Bulgaria and Roma- nia, which limits the availability of data to a small fraction of the Black Sea. Most of the data on non-commercial fish species are gathered from bycatches. Birds are monitored by NGOs affiliated to BirdLife International, therefore some datasets might be available in these organisations, but they are not reported to the Black Sea Commission. Ma- rine mammal populations are not regularly monitored in all Black Sea countries; basically stranded individuals are counted under ongoing projects of short duration. The levels of accidental bycatch of non-target species (birds and marine mammals) are unknown. Priority for assessments should be given to collect further data on all ecosystem components, with particular focus on potentially sensitive fish (e.g. sturgeons, elasmobranchs, shads), seabirds, and marine mammals (the three cetacean subspecies, endemic for the Black Sea: Delphinus delphis ponticus, Phocoena relicta and Tursiops trun- catus ponticus).

26 | ICES WKDIVExtinct REPORT 2018

5 Integrating red list species in aggregated GES indicators

Species sensitive to anthropogenic impact should be explicitly taken into account in the assessment. The revised Commission Decision sets out a number of scientific criteria such as ecological relevance and additional ‘practical’ criteria including that species to be assessed are present in sufficient numbers or extent to be able to estimate a suit- able indicator for assessment. However, rare and threatened species are usually those for which least monitoring data are available and which are prone to extinction. This could create an unfortunate incentive to not include rare species in poor status to achieve GES at the species group level. Rare or endangered species, however, should not be excluded on this basis, and e. g., in D1 assessments all mammals and reptiles listed in Annex II to Directive 92/43/EEC must be included. WKDIVExtinct recommends to include a ‘safety net’ for threatened species in the aggregation from species to species group to account for species or population listed under agreed IUCN red list criteria which are not included in the national reporting of MS. If the safety net for such species which are not included in indicator assessments signals a problem there are two options how to include this in the aggregation on the species group level (depending on aggregation criteria used): 1. The aggregated assessment turns to red when threatened species are identified which fails to meet agreed IUCN conditions and thus are in need of management actions to be taken. 2. To not include red list species in the aggregation, but to include “red flags” with a concrete message that one or more species have failed to meet agreed IUCN conditions and management actions are required. The (indicator based) overall aggregation outcome for the species group can still be green. The following table summarises arguments for and against either of these options:

ARGUMENTS FOR OPTION ARGUMENTS AGAINST OPTION Option 1: “red Mandatory for management to Risk of setting impossible goals because” take specific actions where progress towards these goals may be difficult to judge due to data gaps No matter which successful actions have been taken in order to improve assessment results a few threatened species inhibits the aggregated assessment to turn green. This could discourage management action. Option 2: “green Successful management actions Additional incentive is needed for but” for species assessed within the management actions of flagged frame of indicators are visible at species as no actions are required an earlier stage when GES is reached.

ICES WKDIVExtinct REPORT 2018 | 27

6 Conclusions and Recommendations

6.1 WKDIVAGG outcomes on aggregation, implications on extinction risk

It should be noted that adequate evaluation of both GES and the aggregation to species groups, as well as adequate implementation of the red list approach require access to some minimum amount of quantitative data for the species concerned. There will be data deficiencies in all areas. For the particular case of the Black Sea, the situation is described in Chapter 4.

WKDIVextinct examined the proposals suggested by WKDIVAGG on integration of the status of each species and integration of species to provide status of species group and ecosystem component (see Chapter 1.2), also in the context of possible failure to alert of a potential extinction risk to a species. It was clear that none of the aggregation approaches were able to robustly avoid failure to alert for risks or create false alarms (Chapter 1.3). Simulations of the effectiveness of different aggregation approaches have not yet been completed at the time of the workshop. In addition, an integration and assessment approach using Geometric Mean Abun- dances (widely known as the Living Planet Index) was presented (Chapter 3.2). While it is similar to other averaging methods in that it tends to miss alarms more than generate false alarms, the method is robust to natural species turnover (Chapter 2.2.4). The group considered that extinction risks could be missed regardless of the integration approach. Many threatened species would tend not to be included in the species level aggregations due to lack of data, because they have no key ecological functional role, because the species was sparsely distributed and therefore only a small proportion of the species was present in any MS EEZ or because current monitoring does not adequately record them, and as a result many may not be included in national lists of species for inclusion. So the requirement is for a separate process to explicitly address extinction risk.

6.2 Detailing extinction risk WKDIVextinct examined how potential extinction risk could be evaluated. IUCN provides a comprehensive list of species in the context of extinction risk. IUCN defines species as “threatened”, which is divided into three categories; vulnerable, endangered and critically endangered. The last category would represent the species at most risk of extinction. In the context of GES determination for a species group or higher aggregation level, assessment should include a component representing these “threatened” species. Where a species is critically endangered it should be directly included in the evaluation process. When a species is shown to have moved from a lesser category of threat to a higher one it should be flagged in the periodical GES assessment. This would obvi- ously include any species re-categorised as critically endangered, but equally also for those moved from lower threat levels to any category of “threatened”. Criteria for allocating a species to any of these categories have been established by IUCN. The IUCN evaluation process and the red list are presented in more detail in Chapter 2.1.

28 | ICES WKDIVExtinct REPORT 2018

6.3 Including extinction risk in GES and aggregation WKDIVEXTINCT was asked for adaptation or additional rules to include extinction risk in GES and aggregation. The group proposed the use of the IUCN red list classifi- cation as an additional criterion for the determination of GES at the species group level at the appropriate spatial scale. We would propose that where a “threatened” species is considered as part of a particular species group, GES determination for that species group should flag this species as “outside GES” in all cases where the species is “critically endangered” or has moved to a higher threat level. There are two alternative ways of including extinction risk in GES and aggregation. The species group GES can be considered as within or out of GES using the chosen aggregation rules as appropriate. So the inclusion of one or more critically endangered species in the aggregation could change the status of the species group from green to red. Alternatively, GES could be determined without these species, but these could be specifically flagged for action. For example, potentially a species group could be evaluated as green, but with one or more “red flags”. The pros and cons of the two approaches are detailed in Chapter 5. The group proposed that these red list species be included in species group aggregations for GES at both regional and national levels. It is assumed that a species in Euro- pean waters which is threatened at a global (IUCN) level, would also be threatened at a European, or national level. We would suggest that this be a mandatory part of the reporting. This could also include species and categories provided in red lists from the Regional Sea Conventions. So, in this context, the red list could also be adjusted for specific local conditions, see below.

Red List Minus As the MSFD operates at the level of MS actions, the global red list can be adapted to local specificities. So for instance, while a species may be critically endangered globally, it may never have been present in, say, the Baltic Sea, on a regular basis. There would be no reason to then include it in GES assessment in those waters.

Red List Plus It should be open to MS to include additional species in their “national red lists”. E.g. where the species is locally highly endangered, or where it is considered as unique or important. In addition, environmental change may bring in non-indigenous species that are not currently assessed by IUCN but may need protection locally (see Chapter 2.2). This type of red list could also incorporate national legislation requirements for threatened species (an example from Norway is given in Chapter 2.3) It may also be possible to evaluate extinction risk using ecosystem modelling, and one example of this is presented in Chapter 3.1.

6.4 Conclusion The inclusion of red list species in GES assessment can be seen as a “safety net” in the GES aggregation and reporting process. The main advantages of this approach are that it is largely independent of the choices for aggregation, makes the consideration of extinction risk explicit in the evaluation and ensures that species with very low abundance are not excluded from the evaluation due to lack of observations.

ICES WKDIVExtinct REPORT 2018 | 29

7 References

Barrowman, N.J. & Myers, R.A. 2000. Still more spawner-recruit curves: the hockey stick and its generalisations. Canadian Journal of Fisheries and Aquatic Science, 57, 665–676.

Barry T, Helgason H (2017). Arctic Species Trend Index (ASTI) : Marine. Version 1.3. Conserva- tion of Arctic Flora and Fauna. Occurrence dataset https://doi.org/10.15468/il2vr5 accessed via GBIF.org on 2018-06-13.

BirdLife International (2015). European Red List of birds. Publications Office of the European ion.

Brown, J. H., Ernest, S. M., Parody, J. M., & Haskell, J. P. (2001). Regulation of diversity: mainte- nance of species richness in changing environments. Oecologia, 126(3), 321-332.

Buckland, S. T., A. C. Studeny, A. E. Magurran, J. B. Illian, and S. E. Newson. 2011. The geometric mean of relative abundance indices: a biodiversity measure with a difference. Ecosphere 2(9):100. doi:10.1890/ES11-00186.1

CBD 2016. Indicators for the Strategic Plan for Biodiversity 2011-2020 and the Aichi Biodiversity Targets, CBD/COP/DEC/XIII/28, adopted 12 December 2016.

de Castro, C., P. J. Wright, et al. (2013). "Evidence for substock dynamics within whiting (Mer- langius merlangus) management regions." ICES Journal of Marine Science: Journal du Con- seil 70(6): 1118-1127.

Dinner O'Sullivan, J., Knell, R., Rossberg A. G. (2018). A mechanism for metacommunity-scale biodiversity regulation. Unpublished manuscript.

Dornelas, M., Gotelli, N. J., McGill, B., Shimadzu, H., Moyes, F., Sievers, C., & Magurran, A. E. (2014). Assemblage time series reveal biodiversity change but not systematic loss. Science, 344(6181), 296-299.

Gotelli, N. J., Shimadzu, H., Dornelas, M., McGill, B., Moyes, F., & Magurran, A. E. (2017). Com- munity-level regulation of temporal trends in biodiversity. Science advances, 3(7), e1700315.

Gullestad, P., Abotnes, A. M., Bakke, G., Skern-Mauritzen, M., Nedreaas, K., Søvik, G. 2017. To- wards ecosystem-based fisheries management in Norway – Practical tools for keeping track of relevant issues and prioritising management efforts. Marine Policy, 77: 104-110.

Holmes, S. J., P. J. Wright, et al. (2008). "Evidence from survey data for regional variability in cod dynamics in the North Sea and West of Scotland." ICES Journal of Marine Science 65(2): 206- 215.

IUCN (2012a). IUCN Red List Categories and Criteria: Version 3.1. Second edition. Gland, Swit- zerland and Cambridge, UK: IUCN. iv + 32pp. Available at http://www.iucnredlist.org/technical-documents/red-list-documents.

IUCN (2012b). Guidelines for Application of IUCN Red List Criteria at Regional and National Levels: Version 4.0. Gland, Switzerland and Cambridge, UK: IUCN. iii + 41pp.

IUCN (2017). IUCN Standards and Petitions Subcommittee. Duidelines for Using IUCN Red List Categorties and Criteria. Available at http://www.iucnredlist.org/technical-documents/red- list-documents.

Nieto, A., Ralph, G.M., Comeros-Raynal, M.T., Heessen, H.J.L. and Rijnsdorp, A.D. and others, 2015. European Red List of marine fishes. Publications Office of the European Union.

Magurran, A. E., Dornelas, M., Moyes, F., Gotelli, N. J., & McGill, B. (2015). Rapid biotic homog- enization of marine fish assemblages. Nature communications, 6, 8405.

Magurran, A. E., Deacon, A. E., Moyes, F., Shimadzu, H., Dornelas, M., Phillip, D. A., & Ram- narine, I. W. (2018). Divergent biodiversity change within ecosystems. Proceedings of the National Academy of Sciences, 115(8), 1843-1847.

30 | ICES WKDIVExtinct REPORT 2018

Parody, J. M., Cuthbert, F. J., & Decker, E. H. (2001). The effect of 50 years of landscape change on species richness and community composition. Global Ecology and Biogeography, 10(3), 305-313.

Rindorf, A. & Lewy, P. 2012 Estimating the relationship between abundance and distribution. Can. J. Fish. Aquat. Sci. 69: 382–397. doi:10.1139/F2011-153

Rossberg, A. G., Uusitalo, L., Berg, T., Zaiko, A., Chenuil, A., Uyarra, M. C., Borja, A., and Lynam, C. P. (2017). Quantitative criteria for choosing targets and indicators for sustainable use of ecosystems. Ecological Indicators, 72, 215—224.

Temple, H.J. and Terry, A. (Compilers). 2007. The Status and Distribution of European Mam- mals. Luxembourg: Office for Official Publications of the European Communities. viii + 48pp, 210 x 297.

Thiel, R.; Winkler, H.; Böttcher, U.; Dänhardt, A.; Fricke, R.; George, M.; Kloppmann, M.; Schaarschmidt, T.; Ubl, C. & Vorberg, R. 2013 Rote Liste und Gesamtartenliste der eta- blierten Neunaugen und Fische (Petromyzontida, Elasmobranchii & Actinopterygii) der marinen Gewässer Deutschlands, 5. Fassung, Stand August 2013. Naturschutz und Biolo- gische Vielfalt 70(2): 11 – 76

Thorpe, R.B., Jennings, S., and Dolder, 2017. P.J. Risks and benefits of catching pretty good yield in multispecies mixed fisheries. ICES Journal of Marine Science, 74(8), 2097-2106, https://doi.org/10.1093/icesjms/fsx062

Thorpe, R.B., Dolder, P.J., Reeves, S., Robinson, P. and Jennings, S. 2016. Assessing fishery and ecological consequences of alternate management options for multispecies fisheries ICES Journal of Marine Science, 73(6), 1503–1512, https://doi.org/10.1093/icesjms/fsw028

Thorpe, R.B., Le Quesne, W.J.F., Luxford, F., Collie, J.S., and Jennings, S. 2015. Evaluation and management implications of uncertainty in a multispecies size-structured model of population and community responses to fishing, Methods in Ecology and Evolution, 6, 49-58.

Yoccoz, N. G., Ellingsen, K. E., & Tveraa, T. (2018). Biodiversity may wax or wane depending on metrics or taxa. Proceedings of the National Academy of Sciences, 115(8), 1681-1683

ICES WKDIVExtinct REPORT 2018 | 31

Annex 1: WKDIVExtinct ToRs

WKDIVEXTINCT - Workshop on extinction risk of MSFD biodiversity approach 2017/2/ACOM43 The Workshop on extinction risk of MSFD biodiversity approach (WKDIVEXTINCT), chaired by Dave Reid, Ireland, will be established and will meet in ICES HQ, Copenhagen, Denmark, on 12–15 June 2018 to: a) consider if the methods recommended by WKDIVAGG on integration of the status of each species, followed by integration of species to provide status of species group, could lead to the failure to alert of a potential extinction risk to a species. b) if required, suggest adaptions or additional rules to alert the risk of extinction of a species within a species group.

WKDIVEXTINCT will report by 20 June 2018 for the attention of ACOM.

Supporting information

Priority High. This workshop and its sister WKDIVAGG are required to help address a request from the European Commission DGENV to provide guidance to the MSFD Central Implementation Strategy (CIS) in support of WGGES. Scientific justification DGENV made the following request to ICES: Guidance on an appropriate method(s): a) to integrate across criteria for each species of bird, fish and cephalopod (excepting commercial species and species on Habitats Directive annexes). b) to aggregate species within species groups for an overall assessment of status per species group for MSFD Descriptor 1. c) to aggregate from species group to the level of birds, mammals, reptiles, fish and cephalopods for an overall presentation of the extent to which GES has been achieved for these higher groups. The risk of these suggested methods of failing to assess population trends of sensitive, threatened or vulnerables specis needs to be evaluated. What is the possibility that overall MSFD D1 species assessments will fail to flag deletirious declines in species at risk? For example, if the threshold value for GES is set at 75% of species within the group, but there is significant risk of extinction of other species within the group, how should this affect the application of the aggregation rules?

32 | ICES WKDIVExtinct REPORT 2018

ICES is requested to re-evaluate the previous ICES guidance1 in light of the requirements of the new Decision.

Reference: Commission Decision (EU) 2017/848 of 17 May 2017 laying down criteria and methodological standards on good environmental status of marine waters and specifications and standardised methods for monitoring and assessment, and repealing Decision 2010/477/EU.

Resource requirements Supported by a request from European Commission DGENV. Support from EU MSFD CIS member countries and stakeholders and ICES Member Countries. Input also expected of staff from Regional Sea Conventions. Participants The participant researchers will come from both the ICES and the EU MSFD CIS networks. Secretariat facilities Requiring standard ICES secretariat and logistical support, including meeting room and IT. Financial No financial implications as supported by request from EU DGENV. ICES work order 2002-37. Linkages to advisory This request was made directly to the advisory committee. committees Linkages to other Links to WGECO, WGBIODIV, WGMIXFISH. committees or groups Linkages to other Links to all Regional Sea Conventions, EU DGENV, and MSFD CIS organizations countries and stakeholders.

1 http://www.ices.dk/sites/pub/Publication%20Reports/Advice/2016/Spe- cial_Requests/EU_Guidance_on_method_to_aggregate_species_within_spe- cies_groups_D1.pdf

ICES WKDIVExtinct REPORT 2018 | 33

Annex 2: List of participants

Workshop on extinction risk of MSFD biodiversity approach (WKDIVExtinct) 12–15 June 2018

Name Institute Country e-mail

Anna Rindorf DTU Aqua Denmark [email protected] Axel Rossberg Queen Mary UK [email protected] University of London, School of Biological and Chemical Sciences Dave Reid Marine Institute Ireland [email protected] (Chair) David Vaughan Joint Nature UK [email protected] (via Conservation correspondence) Committee Filip Svensson Swedish Sweden [email protected] University of Agricultural Sciences (SLU), Department of Aquatic Resources Henrik Gislason DTU Aqua Denmark [email protected] Katarzyna Spich National Marine Poland [email protected] Fisheries Research Institute Marina Institute of Bulgaria [email protected] Panayotova Oceanology - BAS Robert Thorpe Cefas UK [email protected] Sean Hayes NOAA, USA [email protected] Northeast Fisheries Science Center

Valeria Abaza National Romania [email protected] Institute for Marine Research and Development, Marine Ecology and Biology Department

34 | ICES WKDIVExtinct REPORT 2018

Annex 3: Agenda

Tuesday June 12th

10.00 Opening of the meeting

Adoption of Agenda

Background information  Quick Review of WKDIVAgg  First discussion of implications for sensitive species and extinction risks

13.00 Lunch

Round of presentations on relevant work from participants 14.00  Participants can discuss, if they wish so, relevant parts of their work - 10 minutes each  Discussion around these, and decisions on relevance to ToR

15.30 Coffee

15.45 Identification of likely species groups and individual species at risk

17.00 Adjourn

Wednesday June 13th

9.00 Are the proposed aggregations from WKDIVAgg likely to expose any of these species to risk?

Discussion

11.00 Coffee

Is the issue 1. failure to alert of a potential extinction risk to a species (easy?) or 2. to identify and quantify the species and the risks (more difficult)

13.00 Lunch

14.00 What adaptions or addional rules would be appropriate to alert the risk of extinction

17.00 Adjourn

Thursday June 14th

9.00 TBC

11.00 Coffee

11.30 TBC

13:00 Lunch

14.00 Report Drafting

15:15 Coffee

15.30 Report Drafting

17.00 Adjourn

Friday June 15th

9.00 Report drafting

11:00 Coffee

11.15 Resolving remaining issues

ICES WKDIVExtinct REPORT 2018 | 35

Annex 4: IUCN summary of criteria sheet

36 | ICES WKDIVExtinct REPORT 2018

Annex5: WKDIVExtinct Report Reviews

Review of report WKDIVEXTINCT

W. Nikolaus Probst WKDIVEXTINCT reviewed the results and conclusions by WKDIVAGG. However, by focusing on a single variant of multiple possible methods, WKDIVEXTINCT did not capture the major issues associated with probabilistic aggregation methods. The true problems of probabilistic methods are the unknown confidence in the majority of single element assessments (i.e. p in the binomial distribution), the ignorance to meaningful patterns in multiple assessment results and the potential masking of negative single element assessments. By focusing on a single probabilistic integration method, the text in chapters 1.2 and 1.3 becomes difficult to follow and is most likely will not understandable to persons, which have not participated in any of the workshops. Instead of dwelling on the issues of the specific probabilistic approach developed during WKDIVAGG, WKDIVEX- TINCT should have briefly recapped the conclusions on probabilistic (and other aggregation) methods drawn by WKDIVAGG. Further, WKDIVEXTINCT does not elaborate on the pros and cons of the One-out-all- out (OOAO) and does not voice any opinion on this aggregation method. WKEXTINCT suggest to implement a safety net for threatened species, as their status may not become apparent in integrated assessments when the majority of assessed species is at GES. WKDIVEXTINCT suggests that threatened species/populations should be included into the integrated assessment, either as turning the integrated assessment into non-GES if one or more threatened species are non-GES, or by including warning flags. WKDIVEXTINCT states that threatened species should be included into integrated assessments, even if they are not listed by single MS. I see two problems with these suggestions: 1 ) Into which assessment should threatened species be included, if they are not already part of single MS list? This could only be a regional assessment. Or should MS report on threatened species which they consider to live outside their national waters. E.g. should Poland report on sharks and rays in the Kattegat, as this is still considered as part of the Baltic by HELCOM? 2 ) Creating a new category of assessment results i.e. “GES”, non- GES” and “flagged” does not comply to the MSFD and does not make things easier to agree on integrated results. The conclusion on necessary management actions could be very different from MS to MS. Unfortunately, WKDIVEXTINCT remains vague, on how either of these solutions could be implemented. Here, some visualised examples wold have been helpful. In- stead of merging threatened and non-threatened species into the same integration while trying to treat them differently, it would be better to develop a deliberate biodiversity indicator anchored within the Regional Seas Conventions, which MS can use in their national reports (similar ot LFI, MML or MTL). This indicator should summarise the conservation status of species based on IUCN criteria and categories as has been suggested by Dulvy et al. (2006) and has been reworked by WGECO in 2012 and WKIND 2013. Also there was a suggested indicator by Magath et al. 2015 to OPSAR

ICES WKDIVExtinct REPORT 2018 | 37

(for which unfortunately I do not have any reference ,as it was rejected by OSAPR BDC).

References

Dulvy, N.K., Jennings, S., Rogers, S.I., Maxwell, D.L. (2006) Threat and decline in fish: an indicator of marine biodiversity. Canadian Journal of Fisheries and Aquatic Sciences 63, 1267-1275.

Review of the WGDIVAGG and WKDIVEXTINCT reports

Eider Andonegi The work developed by ICES in the context of two workshops (WKDIVAGG and WKDIVEXTINCT) to address the EU request to provide guidance on the most appropriate method to integrate criteria, species, species group to higher groups of birds, mammals, reptiles, fish and cephalopods for a Good Environmental Status assessment. The work done by these two groups is quite complete and based on a detailed review of the existing methods and their applicability of all the existing possibilities within the Decision that is being reviewed. WKDIVAGG carried out a detailed review of the existing aggregation methods and identified the most appropriate to be used under three different cases: species under the HD, species under D3 and other species. The group did not particularly address the case of the species under D3, but it was recognized that more information should be provided in order to adequately integrate them in D1. The group assumed the term ‘species’ to be referred to stocks or populations. The group did not come to a conclusion on a single method for integrating criteria to species but recommends a set of integration methods for each specific case. They rec- ommend using indicators with low measurement errors and to include uncertainty in the assessments, recognizing that could be difficult to be estimated in some cases. An- yway, it highlights that all integration options should be tested on real-world case studies (population dynamics datasets and/or Member States assessment reports) to test and conclude on advantages and disadvantages of each and to agree on a recommended option. I would strongly encourage the use of this recommendation in advice drafting. WKDIVAGG report says that “participants agreed that three aspects required careful consideration: i) the treatment and possible propagation of false alarms raised due to random measurement error; ii) the importance of patterns in status; and iii) the issue of data-poor species”. WKDIVEXTINCT critically evaluated the methods recommended by WKDIVAGG and proposed new approaches for assessing the risk of extinction that should be included in the assessment of GES and aggregation (some information still missing). Anyway, they only provide a detailed evaluation in one of the methods, while others should have also been examined. The group highlights that none of the aggregation approaches were able to robustly avoid failure to alert for risks or create false alarms. They also highlight the need of further research for getting better probability estimates around GES and deal with false alarms. Both group recognize that the data needs for much of this approach are quite demanding, and that in many cases there will be insufficient data to make a full determination. If expert judgment is used, WKDIVAGG recommended that a further review would

38 | ICES WKDIVExtinct REPORT 2018

be required once data are available. However, it should be stressed that expert judgement was not recommended for assessments in general (only accepted by the group for assessing uncertainties related to indicators when this information is not available), and this should be stressed in drafting advice. WKDIVEXTINCT also proposed two different alternatives to integrate extinction risk to GES and aggregation, which can be a first step but is quite vague and should be carefully analyzed in each individual assessment to be able to provide good information to inform recommendations for advice.

Review ICES WKDIVAGG and WKDIVExtinct

Andrea Belgrano WKDIVExtinct provided a valuable review (critical evaluation) on the methods proposed by WKDIVAGG. In particular in relation to the two aggregation options WKDIVExtinct concluded that the estimations of probability around the GES status assessment and when dealing with “false alarms” need further analysis. This aspect may also be used in drafting the advice. The proportion-based method proposed by WKDIVAGG was tested by WKDIVExtinct using bird species as a case study. This example showed clearly the underlying com- plexities in understanding the relationship between the total number of species assessed and the probability of producing a correct assessment due to the large uncertainty around the estimated, p values. WKDIVExtinct provide valuable inputs regarding the shifting baselines giving example for species vs. populations/stocks, and on natural and anthropogenic processes. They have also considered the life-history strategies variability as an important aspect when assessing extinction risks. The caveats section provides also valuable input in particular with reference to local extinction. The proposed methods to identify risks are sound, and the point regarding the inadequacy of monitoring data, e.g. Black Sea, is appropriate. WKDIVExtinct, concluded that is important to include extinction risk in GES and aggregation, and in particular that red list species should be included in species group aggregations for GES assessments at the regional and national level. In conclusion both workshops addressed fully the request from the European Com- mission DGENV and provided further guidance on how to proceed, the results form a good base for drafting the advice. However, aspects related on how to assess the functioning of the ecosystem components with respect to GES status using aggregation methods in a broader dynamical context, e.g. foodwebs, needs further considerations.

ICES WKDIVExtinct REPORT 2018 | 39

Synthesis of reviewers comments for WKDIVAGG & WKDIVEXTINCT

Drafted by ICES secretariat. Three independent researchers with expertise in MSFD and biodiversity issues were asked to review the reports of WKDIVAGG and WKDIVEXTINCT. The workshop were held to address a request for advice on the aggregation methods for MSFD de- scription 1 biodiversity of species and species groups. They were asked to:  identify errors in the analysis  comment on the completeness of that advice and if an aspect is found to be missing, draft text to address the point in question  agree that the conclusions are justified by the evidence presented  comment on the use of the reports from the workshops for the drafting of advice to answer the request Two reviewers reviewed both reports and one only WKDIV EXTINCT.

Summary of reviews Reviewer 1 (both reports) appeared to accept that the reports were comprehensive and were a good basis to provide advice on the request. The review stated “both workshops addressed fully the request from the European Commission DGENV and provided further guidance on how to proceed, the results form a good base for drafting the advice.” However, the reviewer criticised the premise of the MSFD approach to species and ecosystem diversity in relation to ecosystem function, in that the species aggregation approach could not account well for issues such as foodwebs. Reviewer 2 (both reports) commented that the work is quite complete and based on a detailed review of the existing methods and their applicability of all the existing possibilities within the Decision that is being reviewed. They also commented the WKDI- VAGG did not come to a conclusion on a single method for integrating criteria to species but recommends a set of integration methods for each specific case using indicators with low measurement errors and to include uncertainty in the assessments. The reviewer highlighted that there was still a need to test the aggregation on real-world case studies. With regards to WKDIVEXTINCT, the reviewer felt that only one of the suggested WKDIVAGG approaches had a detailed evaluation and the others should have also been examined. The reviewer warned against too much use of expert judgement. Reviewer 3 (WKDIVEXTINCT only) was more critical. They struggled to follow some of the methods used and also highlighted that only one of the WKDIVAGG approaches was explored. “WKDIVEXTINCT did not capture the major issues associated with probabilistic aggregation methods. The true problems of probabilistic methods are the unknown confidence in the majority of single element assessments (i.e. p in the binomial distribution), the ignorance to meaningful patterns in multiple assessment results and the potential masking of negative single element assessments.” With regards to the proposed safety net for threatened species, the reviewer highlighted two implementation problems i) how to com- pile the national or regional lists, ii) that either of the two proposed solutions (within the GES assessment or flags) did not fit well into the current MSFD and the revised decision. The reviewer proposed that the use of a threatened species indicator would solve this implementation challenge.