Prioritizing Species for Conservation Planning

CONVENORS: Caroline Lees, Onnie Byers, Phil Miller

AIM: To explore the need for and potential solutions to, prioritizing species for conservation planning within the SSC.

BACKGROUND: The IUCN’s Species Survival Commission recently launched a new initiative aimed at increasing its participation and effectiveness in species conservation through species conservation planning. Of the 85,604 species assessed through the IUCN Global Red List, 24,307 are considered threatened with extinction. Resources are finite and decisions will need to be taken about which of these species to focus on, or at least which to focus on first. A number of initiatives, including some led by IUCN member countries and organizations, have considered the issue of species prioritization for conservation attention and have developed their own approaches. A sample is described in Table 1. We can learn much from these initiatives. In particular:

1) That there is no universally “right” outcome of species prioritization for conservation. Different prioritization goals and contexts will necessarily give rise to different priorities – that is, different priorities can be “right” for different circumstances.

2) That despite the necessary subjectivity embedded within specific prioritization schemes, it is possible to design systematic approaches that are transparent about where this subjectivity lies.

3) That those prioritizing species for conservation attention, though working in different contexts and towards different goals, will often cover some of the same ground and draw the same conclusions about what is important.

4) That where prioritization criteria require the de novo assembly or analysis of large amounts of data, the use of those criteria is likely to be limited to well-understood taxa.

5) That developing from scratch a prioritization scheme acceptable to a large group of stakeholders can take much time, energy and resources.

In this workshop we will explore the species prioritization needs of the SSC community in the area of conservation planning. We will invite participants to share their experiences of developing or implementing prioritization schemes and use this as a basis for discussing how we might move forward with one or more tools to assist prioritization for use within the SSC, and potentially beyond.

PROCESS:

Presentations:

• Summary of different approaches to prioritizing species for conservation attention.

Discussion: • Prioritization needs: what kinds of species planning prioritization problems are there, or might there be within the SSC? • Pros and cons of different schemes: from participants’ experiences of developing or implementing species prioritization schemes, what has or is working well and why? What has not and why? What were the biggest challenges? What else can we learn? • For the prioritization needs identified, are existing tools adequate? If not, what kinds of new tools might be most useful? Is an expert system an option? Would written guidelines be a better approach?

Some of the results of this session will be used to inform the subsequent workshop on multi-species planning.

PREPARATION: Participants are asked to familiarize themselves with Mace et al., 2007 and to come equipped with details and insights into any species prioritization schemes or tools with which they are familiar.

Table 1. Example schemes for prioritizing species for conservation attention

Note that all of the schemes exemplified here use the IUCN Red List or equivalent as a criterion, which automatically excludes the many species not yet assessed.

Overarching goal List of alternatives Criteria for selection/scoring Resulting priorities Initiative: Alliance for Zero Extinction (AZE) (88 NGOs. National Alliances now also exist.) See http://www.zeroextinction.org To defend against the most All species for Endangered or Critically 920 species prioritised to predictable species losses. which Endangered (IUCN) AND date - mammals, birds, endangerment and restricted to a single amphibians, reptiles, distribution are remaining site conifers, and reef- known. building corals.

Asian Species Action Partnership (ASAP) http://www.speciesonthebrink.org/ Reversing declines in the wild of Southeast Asian Critically Endangered, 174 species Asian species on the brink of species freshwater and terrestrial extinction. vertebrates, occurring regularly in Southeast Asia Initiative: Evolutionarily Distinct and Globally Endangered (EDGE) (Zoological Society of London) (see Isaac et al., 2007)

To maximise conservation of All species for Score = Evolutionary Portfolio approach phylogenetic diversity. which phylogenetic Distinctiveness (ED) X Global includes the top 100 uniqueness has Endangerment (IUCN). scoring species in each of been assessed: the major taxonomic mammals, EDGE species have a greater groups considered amphibians, corals than average score (Isaac et and birds. al., 2007)

Initiative: Method for the Assessment of Priorities for International Species Conservation (MAPISCo)

To identify species for which Species in the IUCN Ability to contribute to: ? targeted conservation action would Red List database (1) habitat and area have the broadest co-benefits for for which sufficient conservation (2) other species, habitats, wider data exist to allow sustainable harvesting of ecosystems, and ecosystem assessment against Overarching goal List of alternatives Criteria for selection/scoring Resulting priorities services. the criteria (?). fish, invertebrates and aquatic plants, (3) conservation of genetic diversity of wild relatives of cultivated plants and domesticated animals, (4) protection of the provisioning of ecosystem services (5) the prevention of species extinctions. Initiative: National prioritisation scheme for conservation action planning (New Zealand Dept. of Conservation) (NZ DOC) (see Joseph et al., 2009) To optimise allocation of All species native to Assessed (using NZ RL- ≈700 species prioritised conservation planning resources New Zealand. equivalent) as conservation for management planning towards the goal of ensuring the dependent OR as threatened out of ≈10,000 assessed. persistence of all New Zealand and declining, with threats species somewhere. understood and conservation action considered feasible.

READING:

Isaac N.J., Turvey S.T., Collen B., Waterman C., Baillie J.E. 2007. Mammals on the EDGE: Conservation Priorities Based on Threat and Phylogeny. PLoS ONE 2(3): e296. https://doi.org/10.1371/journal.pone.0000296

Joseph, L. N., R. F. Maloney, J. E. M. Watson, and H. P. Possingham. 2011. Securing nonflagship species from extinction. Conservation Letters 4:324-325.

Mace, G.M., Possingham, H.P., and Leader-Williams, N. 2007. Prioritizing Choices in Conservation. In: MacDonald, D.W. and Service, K. (Eds) Key Topics in Conservation Biology. Blackwell Publishing Ltd.

MAPISCo Project Team (2013) Method for the assessment of priorities for international species conservation. Newcastle University, Newcastle upon Tyne, UK. http://www.zeroextinction.org

Mammals on the EDGE: Conservation Priorities Based on Threat and Phylogeny Nick J. B. Isaac*, Samuel T. Turvey, Ben Collen, Carly Waterman, Jonathan E. M. Baillie

Institute of Zoology, Zoological Society of London, London, United Kingdom

Conservation priority setting based on phylogenetic diversity has frequently been proposed but rarely implemented. Here, we define a simple index that measures the contribution made by different species to phylogenetic diversity and show how the index might contribute towards species-based conservation priorities. We describe procedures to control for missing species, incomplete phylogenetic resolution and uncertainty in node ages that make it possible to apply the method in poorly known clades. We also show that the index is independent of clade size in phylogenies of more than 100 species, indicating that scores from unrelated taxonomic groups are likely to be comparable. Similar scores are returned under two different species concepts, suggesting that the index is robust to taxonomic changes. The approach is applied to a near-complete species-level phylogeny of the Mammalia to generate a global priority list incorporating both phylogenetic diversity and extinction risk. The 100 highest-ranking species represent a high proportion of total mammalian diversity and include many species not usually recognised as conservation priorities. Many species that are both evolutionarily distinct and globally endangered (EDGE species) do not benefit from existing conservation projects or protected areas. The results suggest that global conservation priorities may have to be reassessed in order to prevent a disproportionately large amount of mammalian evolutionary history becoming extinct in the near future. Citation: Isaac NJB, Turvey ST, Collen B, Waterman C, Baillie JEM (2007) Mammals on the EDGE: Conservation Priorities Based on Threat and Phylogeny. PLoS ONE 2(3): e296. doi:10.1371/journal.pone.0000296

INTRODUCTION [19]. Among mammals alone, at least 14 genera and three families Our planet is currently experiencing a severe anthropogenically have gone extinct since AD 1500 [20], and all members of a further driven extinction event, comparable in magnitude to prehistoric 19 families and three orders are considered to be in imminent mass extinctions. Global extinction rates are now elevated up to danger of extinction [2]. Many academic papers have suggested a thousand times higher than the background extinction rates ways to maximise the conservation of PD [e.g. 12,13,21–23] and shown by the fossil record, and may climb another order of measure species’ contributions to PD [e.g. 4,23–25], but these magnitude in the near future [1–3]. The resources currently have rarely been incorporated into conservation strategies. available for conservation are, unfortunately, insufficient to Therefore, it is possible that evolutionary history is being rapidly prevent the loss of much of the world’s threatened biodiversity lost, yet the most distinct species are not being identified as high during this crisis, and conservation planners have been forced into priorities in existing conservation frameworks. the unenviable situation of having to prioritise which species There are several reasons why PD has not gained wider accept- should receive the most protection–this is ‘the agony of choice’ [4] ance in the conservation community. First, although evolutionary or the ‘Noah’s Ark problem’ [5]. history consists of two distinct components (the branching pattern A range of methods for setting species-based conservation of a phylogenetic tree and the length of its branches), complete priorities have been advocated by different researchers or dated species-level phylogenies for large taxonomic groups have organisations, focusing variously on threatened species, restrict- only recently become available [26]. Early implementations of PD- ed-range endemics, ‘flagship’, ‘umbrella’, ‘keystone’, ‘landscape’ or based approaches were therefore unable to incorporate branch ‘indicator’ species, or species with significant economic, ecological, length data, and focused solely on measurements of branching scientific or cultural value [6–8]. To date, global priority-setting pattern [4]. Second, PD removes the focus from species and so exercises have tended to focus on endemic (or restricted range) may lack wider tangible appeal to the public; conserving PD may species [6,9,10], presumably because endemism is easier to be seen as less important than the protection of endemic or measure than competing methods. However, recent data show that endemism is a poor predictor of total species richness or the number of threatened species [11]. Academic Editor: Walt Reid, The David and Lucile Packard Foundation, It has also been argued that maximising Phylogenetic Diversity Conservation and Science Program, United States of America (PD) should be a key component of conservation priority setting Received January 16, 2007; Accepted February 19, 2007; Published March 14, [4,12–14]. Species represent different amounts of evolutionary 2007 history, reflecting the tempo and mode of divergence across the Tree of Life. The extinction of a species in an old, monotypic or Copyright: ß 2007 Isaac et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits species-poor clade would therefore result in a greater loss of unrestricted use, distribution, and reproduction in any medium, provided the biodiversity than that of a young species with many close relatives original author and source are credited. [15,16]. However, conserving such lineages may be difficult, since Funding: The authors did not receive any funding to conduct the research there is some evidence that they are more likely to be threatened described in this paper. with extinction than expected by chance [17]. This clumping of extinction risk in species-poor clades greatly increases the loss of Competing Interests: The authors have declared that no competing interests exist. PD compared with a null model of random extinction [18] and suggests that entire vertebrate orders may be lost within centuries * To whom correspondence should be addressed. E-mail: [email protected]

PLoS ONE | www.plosone.org 1 March 2007 | Issue 3 | e296 Mammals on the EDGE threatened species [16]. However, the current instability in species taxonomy [27] means that decisions based on PD might be more objective than those based on different species concepts [13,16,27]. Combining species’ conservation status with a measure of their contribution to PD is therefore desirable, because species can be retained as units but weighted appropriately [5,22]. This would generate a useful and transparent means for setting global priorities for species-based conservation [25]. This paper describes a new method for measuring species’ relative contributions to phylogenetic diversity [the ‘originality’ of species: ref 24]. We explore the statistical properties of the resulting measure, which we call Evolutionary Distinctiveness (ED), and test its robustness to changing species concepts. ED scores are calculated for the Class Mammalia, and combined with values for species’ extinction risk to generate a list of species that are both evolutionarily distinct and globally endangered (‘EDGE species’). The resultant list provides a set of priorities for mammalian conservation based not only on the likelihood that a species will be lost, but also on its irreplaceability.

Evolutionary Distinctiveness and its use in priority-setting In order to calculate ED scores for each species, we divide the total phylogenetic diversity of a clade amongst its members. This is achieved by applying a value to each branch equal to its length divided by the number of species subtending the branch. The ED of a species is simply the sum of these values for all branches from which the species is descended, to the root of the phylogeny. For Figure 1. Hypothetical phylogeny of seven species (A–G) with the examples in this paper, we have measured ED in units of time, Evolutionary Distinctiveness (ED) scores. Numbers above each branch such that each million years of evolution receives equal weighting indicate the length; numbers below show the number of descendent and the branches terminate at the same point (i.e. the phylogeny is species. MYBP, millions of years before present. ultrametric). The method could be applied to non-ultrametric doi:10.1371/journal.pone.0000296.g001 phylogenies if the conservation of other units [e.g. character diversity 28,29] was prioritised [although see ref 30]. uncertainty in node ages (see Materials and Methods). The The basic procedure for calculating ED scores is illustrated in approach is implemented using a dated phylogeny of the Class figure 1, which describes a clade of seven species (A–G). The ED Mammalia that is nearly complete (.99%) at the species level score of species A is given by the sum of the ED scores for each of [31]. We then combined ED and extinction risk to identify species the four branches between A and the root. The terminal branch that are both evolutionarily distinct and globally endangered contains just one species (A) and is 1 million years (MY) long, so (‘EDGE species’). We measured extinction risk using the receives a score of 1 MY. The next two branches are both 1 MY quantitative and objective framework provided by the World long and contain two and three species, so each daughter species Conservation Union (IUCN) Red List Categories [2]. We follow (A, B and C) receives 1/2 and 1/3 MY respectively. The deepest previous researchers in treating the Red List categories as intervals branch that is ancestral to species A is 2 MY long and is shared of extinction risk and by assuming equivalence among criteria among five species (A to E), so the total ED score for species A is [32,33, but see 34]. The resulting list of conservation priorities given by (1/1+1/2+1/3+2/5) = 2.23 MY. Species B is the sister (‘EDGE scores’) was calculated as follows: taxon of A, so receives the same score. By the same arithmetic, C has a score of (2/1+1/3+2/5) = 2.73 MY, both D and E receive (1/1+2/2+2/5) = 2.4 MY, and both F and G receive (0.5/1+4.5/ EDGE~ln(1zED)zGE ln(2) ð1Þ 2) = 2.75 MY. The example illustrates that ED is not solely determined by a species’ unique PD (i.e. the length of the terminal where GE is the Red List category weight [Least Concern = 0, branch). Species F and G are the top-ranked species based on their ED scores, even though each represents just a small amount of Near Threatened and Conservation Dependent = 1, Vulnera- unique evolutionary history (0.5 MY). This suggests that the ble = 2, Endangered = 3, Critically Endangered = 4, ref 32], here conservation of both F and G should be prioritised, because the representing extinction risk on a log scale. EDGE scores are extinction of either would leave a single descendant of the oldest therefore equivalent to a loge-transformation of the species-specific and most unusual lineage in the phylogeny [c.f. 15,24]. The ED expected loss of evolutionary history [5,25] in which each calculation is similar to the Equal Splits measure [25], which increment of Red List category represents a doubling (eln(2))of apportions branch length equally among daughter clades, rather extinction risk. For the purposes of these analyses, we did not than among descendent species. In order to represent a useful tool in priority setting, ED scores calculate EDGE scores for species listed as Extinct in the Wild must be applicable in real phylogenies of large taxonomic groups. (n = 4), domesticated populations of threatened species and 34 To do this, we modified the basic procedure described above to species (mostly of dubious taxonomic status) for which an control for missing species, incomplete phylogenetic resolution and evaluation has not been made.

PLoS ONE | www.plosone.org 2 March 2007 | Issue 3 | e296 Mammals on the EDGE

RESULTS species concept [36] (r2 = 0.65 on a log-log scale), in spite of the fact that there are substantial differences between the two: the Statistical properties of ED number of primate species differs by 50%. Furthermore, the We measured ED in clades of different sizes to test whether ED highest-ranking species do not change their identity: 45 of 58 scores from different taxonomic groups are likely to be biological species in the upper quartile of ED scores are also in the comparable. We found that most ED is derived from a few upper quartile as phylogenetic species. However, species that have branches near the tips (i.e. those shared with few other species) and been split into three or more species do tend to lose a large portion that virtually no ED is gained in clades above ,180 species of their ED. For example, the fork-marked lemur (Phaner furcifer)is (figure 2). Median ED in clades of 60 species is 88% of the total the second most distinct biological species of primate, with an ED accumulated using the whole tree (n = 10, figure 2). Moreover, the score of 38.33. It was split into four phylogenetic species [36] with rank order of ED scores is unaffected by the size of the clade under an ED score of 10.45 (Table S2), which is just inside the upper consideration, except in very small clades and among species with quartile. low overall ED (i.e. few of the lines in figure 2 cross one another). These findings suggest that ED scores of different taxonomic groups measured on separate phylogenies (i.e. with no nodes in ED and EDGE scores in mammals common) will be comparable, so long as each phylogeny is larger Mammal ED scores range from 0.0582 MY (19 murid rodents) to than a threshold size. Based on the scaling observed in figure 2, we 97.6 MY (duck-billed platypus, Ornithorhynchus anatinus). Scores are suggest a minimum species richness of 100 as a useful rule of approximately log-normally distributed, with a median of 7.86 thumb to ensure comparability among taxa. MY and geometric mean of 6.28 MY. Although most species (90% in figure 2) derive at least two- Evolutionary Distinctiveness is not evenly distributed among the thirds of their total ED from the terminal branch (which is not Red List categories. Least Concern species have significantly lower shared with others), this branch length is a poor predictor of total ED than the other categories (F1,4180 = 26.3, p,0.0001, using loge ED (r2 = 0.03 on a log-log scale). For species on short branches, transformed scores); there are no significant differences among the there is an order of magnitude difference between the length of the remaining categories. This suggests that species with low ED terminal branch and ED. For example, the pale-throated and scores tend to suffer from low levels of extinction risk, although the brown-throated three-toed sloths (Bradypus tridactylus and B. explanatory power of this model is extremely low (r2 = 0.006). variegatus) share a common ancestor thought to be just over EDGE scores range from 0.0565 (10 murid rodents) to 6.48 a million years old, but the total ED of both species is 20.4 MY (Yangtze River dolphin or baiji, Lipotes vexillifer) and are (Table S1) since they have few close living relatives. approximately normally distributed around a mean of 2.63 ED scores are also robust to taxonomic changes. For example, (60.017; figure 3). The 100 highest priority (EDGE) species ED scores in primates under the biological species concept [35] includes several large-bodied and charismatic mammals, including are tightly correlated with ED scores under the phylogenetic the giant and lesser pandas, the orang-utan, African and Asian

Figure 2. Scaling of ED scores with clade size for ten Critically Endangered mammal species. ED scores were calculated at each node between the tips and root for ten species in different orders. Species chosen are: the baiji (Lipotes vexillifer), sumatran rhino (Dicerorhinus sumatrensis), northern hairy- nosed wombat (Lasiorhinus krefftii), persian mole (Talpa streeti), Omiltemi rabbit (Sylvilagus insonus), Przewalski’s gazelle (Procapra przewalskii), black- faced lion tamarin (Leontopithecus caissara), Livingstone’s flying fox (Pteropus livingstonii), red wolf (Canis rufus) and northern Luzon shrew rat (Crunomys fallax). See Materials and Methods for further details. doi:10.1371/journal.pone.0000296.g002

PLoS ONE | www.plosone.org 3 March 2007 | Issue 3 | e296 Mammals on the EDGE elephants, four rhinoceroses, two tapirs, two baleen whales, by a decrease in ED. A good example is that of the ruffed lemurs a dugong and a manatee. However, many smaller and less (Varecia spp.), which consist of one Endangered biological species appreciated species also receive high priority, including sixteen (ED = 19.8; EDGE = 5.11) or two phylogenetic species (Endan- rodents, thirteen eulipotyphlans, twelve bats, four lagomorphs and gered and Critically Endangered; ED = 10.3; EDGE = 4.50 and an elephant shrew (Table S1). The top 100 also includes at least 37 5.20). Using the same approach, we estimate that the long-beaked species that would not qualify for most area-based definitions of echidna (Zaglossus bruijni) would fall from the second-ranked endemism, since they are listed as threatened under Red List priority to the 20th after the addition of two new congeners criterion A (reduction in population size) without qualifying for [suggested by 39]. Thus, EDGE scores for existing species are criteria B–D, which are based on population size or geographical robust to the ongoing discovery of new species. range. Whilst the highest-ranked species, by definition, are all EDGE priorities are also robust to several other forms of highly threatened (44 of the top 100 species are Critically uncertainty. Like all phylogenetic methods, the precise EDGE Endangered, a further 47 are Endangered), threat status alone scores are dependent on the topology and branch lengths of the does not guarantee a high priority. For example, 10 Critically phylogeny. However, errors in the phylogeny are unlikely to alter Endangered species (in the genera Gerbillus, Peromyscus and the identity of high-ranking species, particularly for clades of Crocidura), as well as 32 Endangered species, fail to make the top several hundred species. Topological uncertainty is usually 1000, whilst 130 Near Threatened species do. expressed in supertrees as polytomies, which are accounted for using simple correction factors. Likewise, branch length un- DISCUSSION certainty has been incorporated into the scoring system to down- weight the priority of species descended from nodes with It is important that conservation priority-setting approaches are imprecisely estimated ages (see Materials and Methods). These able to satisfy two conditions: they capture biodiversity and are developments make it possible to estimate robustly the contribu- robust to uncertainty. The method described herein satisfies the tion to phylogenetic diversity of species in poorly known clades. first condition because EDGE scores incorporate species value (in The other major source of uncertainty is in estimating extinction terms of originality, or irreplaceability) weighted by urgency of risk: most recent changes in Red List category have come about action (i.e. risk of extinction). Our approach satisfies the second through improvements in knowledge, rather than genuine changes condition because the scores are also robust to clade size, missing in status [32]. EDGE scores will inevitably be affected by future species and poor phylogenetic resolution. EDGE scores are also changes in extinction risk, although no more so than other easy to calculate, as all that is required is a set of Red List approaches using the Red List categories. assessments and a near-complete phylogeny containing at least A minority of mammal species could not be assigned EDGE 100 species. scores. Around 300 species are classified as Data Deficient and In particular, EDGE priorities are much less sensitive to could not be meaningfully included, although in reality they may taxonomic uncertainty than alternate methods. The current trend have a high risk of extinction [17]. By far the most likely candidate towards the adoption of the phylogenetic species concept among for high EDGE status following future Red List re-assessment is biologists [27] is likely to produce a large number of ‘new’ the franciscana or La Plata River dolphin Pontoporia blainvillei threatened and endemic species [37], potentially altering the (ED = 36.3 MY). In addition, fifty extant species are missing from distribution of hotspots [38] and distorting other biodiversity the phylogeny. The highest ranked of these are probably a pair of patterns [27]. The EDGE approach is robust to such distortion Critically Endangered shrews (Sorex cansulus and S. kizlovi); median because any increase in extinction risk due to splitting is balanced and maximum ED scores for the genus are 4.55 and 14.6 MY, giving potential respective EDGE scores of 4.49 and 5.52 for these species (cf. figure 3). A further 260 species have been described since the chosen taxonomy was published [40]. Of these, the recently described Annamite striped rabbit Nesolagus timminsi [41] is the sister species to the tenth-ranked Sumatran rabbit N. netscheri, so would be a high priority if similarly threatened. It has been suggested that species with few close relatives (i.e. high ED) are ‘relicts’ or ‘living fossils’ that have limited ability to generate novel diversity. This view implies that conservation efforts should instead be focused on recent radiations containing species with low ED scores (e.g. murid rodents), which represent ‘cradles’ rather than ‘museums’ of diversity [e.g. 16,42]. However, the assumption that we are able to predict future evolutionary potential is dubious and no general relationships between phylogeny and diversity over geological time have yet been established [43,44]. Furthermore, phylogenetic diversity is clearly related to character diversity [30], and so ED may be a useful predictor of divergent properties and hence potential utilitarian value [14]. Moreover, because species with low ED scores tend to suffer from low levels of extinction risk, phylogenetic cradles of mammalian diversity are likely to survive the current extinction crisis even without specific interventions. Focusing on lower risk Figure 3. Histogram of EDGE scores for 4182 mammal species, by threat category. Colours indicate the Red List category: Least Concern (green), species, at the expense of EDGE priorities, would therefore result Near Threatened and Conservation Dependent (brown), Vulnerable in a severe pruning of major branches of the Tree of Life (yellow), Endangered (orange) and Critically Endangered (red). comparable to that seen in previous mass extinction events doi:10.1371/journal.pone.0000296.g003 [45,46].

PLoS ONE | www.plosone.org 4 March 2007 | Issue 3 | e296 Mammals on the EDGE

The top 100 EDGE species span all the major mammalian measures, and that more work is carried out to prevent the clades [being distributed among 18 orders and 52 families imminent loss of large quantities of our evolutionary heritage. It is recognised by ref 35] and display a comparable range of hoped that this approach will serve to highlight their importance morphological and ecological disparity, including the largest and to biodiversity and emphasize the need for urgent conservation smallest mammals, most of the world’s freshwater cetaceans, an action. oviparous mammal and the only species capable of injecting venom using their teeth. However, around three-quarters of MATERIALS AND METHODS species-based mammal conservation projects are specifically aimed at charismatic megafauna [47], so conventional priority-setting Implementing ED scores for mammals tools may not be sufficient to protect high priority EDGE species. We used a composite ‘supertree’ phylogeny [31] to calculate ED This concern is supported by two additional lines of evidence. scores for mammals. The supertree presents several challenges to First, we found that species not found in protected areas [‘gap the estimation of ED when compared with the (unknown) true species’ defined by ref 48] tended to have higher EDGE scores phylogeny: poor resolution, missing species and uncertainty in than those found inside protected areas (logistic regression: node ages. Accordingly, we modified the basic procedure to 2 x 1,3994 = 69.46, p,0.0001). Second, an assessment of published control for these problems. conservation strategies and recommendations (including IUCN Phylogenetic information is poor in many mammalian clades Specialist Group Conservation Action Plans, captive breeding (especially bats and rodents, which together make up .60% of protocols and the wider scientific literature listed in the 1978–2005 species) and the whole supertree contains only 47% of all possible Zoological Record database) reveals that no species-specific nodes, many of which are polytomies (nodes with more than two conservation actions have even been suggested for 42 of the top daughter branches). Across the whole phylogeny, ,40% of species 100 EDGE species. Most of these species are from poorly known are immediately descended from bifurcations, ,20% from small regions or taxonomic groups and until now have rarely been polytomies (3–5 daughters), ,15% from medium-sized polytomies highlighted as conservation priorities. Little conservation action is (6–10 daughters) and the remainder from large polytomies with actually being implemented for many other top EDGE species, .10 daughters. Polytomies in supertrees result from poor or despite frequent recommendations in the conservation literature. conflicting data rather than a true representation of the speciation Indeed, the top-scoring EDGE species, the Yangtze River dolphin process, so the distinctiveness of branches subtending them is (Lipotes vexillifer), is now possibly the world’s most threatened overestimated [54], thus leading to biased ED scores. For example, mammal despite two decades of debate over a potential ex situ the common ancestor of species X, Y and Z is believed to be 1 MY breeding programme, and may number fewer than 13 surviving old, but the branching pattern within the clade is unknown. The individuals [49]. The lack of conservation attention for priority polytomy appears to show that each species represents 1 MY of EDGE species is a serious problem for mammalian biodiversity unique evolutionary history. In reality, the phylogeny is bi- and suggests that large amounts of evolutionary history are likely furcating, with one species aged 1 MY and the others sharing to be lost in the near future. This phenomenon of diversity slipping a more recent common ancestor. The bias induced by polytomies quietly towards extinction is likely to be much more severe in less can be corrected by estimating the expected ED of descendant charismatic groups than mammals. species under an appropriate null model of diversification. We The approach described in this paper can be used for achieved this by applying a scaling factor based on the empirical conservation in a number of ways. First, conservation managers distribution of ED scores in a randomly generated phylogeny of with limited resources at their disposal typically need to conserve 5000 species grown under constant rates of speciation (0.1) and populations of several threatened species. If all other factors were extinction (0.08). The mean ED score of species in 819 clades of equal, the management of the most evolutionarily distinct species three species was 0.81 of the clade age; ED scores for nodes of 2– should be prioritized. Second, a list of high-priority species 20 species scale according to (branch length) * (1.081–0.267 * requiring urgent conservation action can be generated easily. In ln{d}), where d is the number of descendent branches (n = 2873 this paper, we have selected the 100 highest-ranking species, but clades, r2 = 0.69). Quantitatively similar values were obtained in one might equally choose all threatened (Vulnerable and above) bifurcating clades of primates [1.117–0.246 * ln{d}, n = 78, ref 55] species with above average ED. This would result in a list of 521 and carnivores [1.139–0.269 * ln{d}, n = 101, ref 56]. (using median) or 630 (using geometric mean) ‘EDGE species’ that The mammal supertree contains 4510 of the 4548 (.99%) are both evolutionarily distinct and globally endangered. Third, extant species listed in Wilson & Reeder [35]. Although few in EDGE scores could also be used to weight species’ importance in number, the missing species need to be taken into account because selecting reserve networks, building on previous studies that have their absence will tend to inflate the ED scores of close relatives. used phylogenetic diversity [50–52] or threatened species [11] to For example, omitting species A from the phylogeny in figure 1 identify priority areas for conservation. The statistical properties of would elevate B from the joint lowest ranking species (with A) to EDGE scores (they are both normally-distributed and bounded at the joint highest-ranking (with C), with an ED score of (2/1+1/ zero) make them especially suitable for these kinds of analysis. In 2+2/4) = 3.5 MY. The problem is acute in real datasets since this way, the EDGE approach is not an alternative to existing missing species tend not to be a random sample: 22 of the 38 conservation frameworks [e.g. 6] but complements them. missing mammals are from the genus Sorex. We account for this The EDGE approach identifies the species representing most problem using a simple correction factor that allocates the missing evolutionary history from among those in imminent danger of species among their presumed closest relatives. For example, we extinction. Our methods extend the application of PD-based correct for the omission of the bare-bellied hedgehog (Hemiechinus conservation to a wider range of taxa and situations than previous nudiventris) by treating the other five Hemiechinus spp. as 6/5 = 1.2 approaches [4,5,13,22,24,25]. Future work might incorporate species, and we correct for the omission of both Cryptochloris species socioeconomic considerations [5,14] and the fact that a species’ by spreading the two missing species evenly between other value depends also on the extinction risk of its close relatives [53]. Chrysochloridae. However, our results suggest that large numbers of evolutionarily Variation among morphological and molecular estimates of distinct species are inadequately served by existing conservation divergence times (node ages) can lead to considerable uncertainty

PLoS ONE | www.plosone.org 5 March 2007 | Issue 3 | e296 Mammals on the EDGE in ED scores. To reduce the effects of this uncertainty, we EDGE scores for all species included in the mammal supertree estimated ED using three sets of branch lengths. One set was based [31] ranked by their EDGE score. Species that could not be on the best (i.e. mean) estimates of node age; the others were assigned EDGE scores are appended to the bottom of the list, derived from the upper and lower 95% confidence intervals sorted by status and ED score. Species taxonomy follows Wilson & around these dates. Species values of ED were calculated as the Reeder [35]. Red List categories follow the 2006 IUCN Red List geometric mean of scores under the three sets of branch lengths. [2]: CR = Critically Endangered, EN = Endangered, VU = Vul- The geometric mean was preferred since it down-weights species nerable, NT = Near Threatened, LC = Least Concern, CD = Con- whose scores are based on nodes with symmetrical but wide servation Dependent, DD = Data Deficient, NE = Not Evaluated. confidence intervals in estimate age, and is therefore more The NE category includes species in Wilson & Reeder [35] that conservative than the arithmetic mean. could not be matched with any species or subspecies names in the Red List. Tests of robustness Found at: doi:10.1371/journal.pone.0000296.s001 (0.42 MB To test whether ED scores are comparable among taxonomic PDF) groups, we examined how species’ ED accumulates as pro- Table S2 Evolutionary Distinctiveness for primates under two gressively larger clades are considered. If ED scores are truly species concepts. This table lists ED scores for primates under the comparable, their rank order will be independent of the size of the biological species concept i[.e. the taxonomy of ref 35], the clade considered. We randomly selected one Critically Endan- number of phylogenetic species into which the biological species gered species from each of ten mammal orders and measured the was split [36] and the estimated ED score of each phylogenetic cumulative ED score at each node between the species and the species. See Materials and Methods for further information. ED root of the mammal supertree, thus redefining and enlarging the scores are lower for phylogenetic species than biological species, clade (and so increasing the number of species it contained) at each even for taxa whose taxonomic status is the same under both step. concepts (i.e. the number of phylogenetic species is one). This Taxonomic changes have the potential to dramatically alter the occurs because the total number of species in the phylogeny is ED scores of individual species. Splitting a species in two reduces greater, so each receives a smaller share of the distinctiveness of the distinctiveness of all branches ancestral to the split, particularly ancestral branches. ED scores were calculated using just one set of those near the tips. If ED scores are highly sensitive to taxonomic branch lengths (the ‘best’ set), so differ from those in table S1. changes then it may be meaningless to apply them in setting conservation priorities. The effects of taxonomic changes on ED Found at: doi:10.1371/journal.pone.0000296.s002 (0.05 MB scores were therefore investigated in the primates, which have PDF) recently experienced considerable taxonomic inflation [27]. We compared primate ED scores under a biological species concept ACKNOWLEDGMENTS [35: 233 species] and a phylogenetic species concept [36: 358 We thank the PanTHERIA consortium for use of the mammal supertree. species]. We employed a single phylogeny [31], but changed the In particular, Olaf Bininda-Emonds generously provided us with electronic number of species represented by each tip. We calculated the copies of the topology and alternate branch lengths. We also thank Ana expected ED for multi-species tips by treating them as if they were Rodrigues for providing a list of species found in protected areas. We are also grateful to Guy Cowlishaw, Sarah Durant, Dan Faith, John Gittleman, descended from a polytomy of {n+r+1} descendent branches, Rich Grenyer, Kate Jones, Georgina Mace, Arne Mooers, Andy Purvis where n is the actual number of descendent branches and r is the and three anonymous referees for useful comments and discussion. number of species represented by the tip. Author Contributions SUPPORTING INFORMATION Conceived and designed the experiments: JB BC NI ST. Performed the Table S1 Evolutionary Distinctiveness and EDGE scores for experiments: NI. Analyzed the data: NI. Contributed reagents/materials/ mammals. This table shows Evolutionary Distinctiveness (ED) and analysis tools: NI ST CW. Wrote the paper: JB BC NI ST.

REFERENCES 1. Pimm SL, Russell GJ, Gittleman JL, Brooks TM (1995) The future of 10. Stattersfield AJ, Crosby MJ, Long AJ, Wege DC (1998) Endemic bird areas of biodiversity. Science 269: 347–350. theworld:prioritiesforbiodiversityconservation.Cambridge:BirdLife 2. IUCN (2006) 2006 IUCN Red List of Threatened Species. http://www. International. 846 p. iucnredlist.org. 11. Orme CDL, Davies RG, Burgess M, Eigenbrod F, Pickup N, et al. (2005) Global 3. Mace GM, Masundire H, Baillie JEM (2005) Biodiversity. Chapter 4 in: hotspots of species richness are not congruent with endemism or threat. Nature Millennium ecosystem assessment, 2005 Current state and trends: findings of the 436: 1016–1019. condition and trends working group Ecosystems and human well-being, vol1. 12. Witting L, Loeschcke V (1995) The optimization of biodiversity conservation. Washington DC: Island Press. Biological Conservation 71: 205–207. 4. Vane-Wright RI, Humphries CJ, Williams PH (1991) What to protect- 13. Faith DP (1992) Conservation evaluation and phylogenetic diversity. Biological systematics and the agony of choice. Biological Conservation 55: 235–254. Conservation 61: 1–10. 5. Weitzman ML (1998) The Noah’s Ark Problem. Econometrica 66: 1279–1298. 14. Crozier RH (1997) Preserving the information content of species: Genetic 6. Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) diversity, phylogeny, and conservation worth. Annual Review of Ecology and Biodiversity hotspots for conservation priorities. Nature 403: 853–858. Systematics 28: 243–268. 7. Mace GM, Collar NJ (2002) Priority-setting in species conservation. In: Norris K, 15. May RM (1990) Taxonomy as destiny. Nature 347: 129–130. Pain DJ, eds (2002) Conserving bird biodiversity: general principles and their 16. Mace GM, Gittleman JL, Purvis A (2003) Preserving the tree of life. Science 300: application. Cambridge: Cambridge University Press. pp. 61–73. 1707–1709. 8. Entwistle A, Dunstone N, eds (2000) Has the panda had its day? Priorities for the 17. Purvis A, Agapow PM, Gittleman JL, Mace GM (2000) Nonrandom extinction conservation of mammalian diversity. Cambridge: Cambridge University Press. and the loss of evolutionary history. Science 288: 328–330. 455 p. 18. Heard SB, Mooers AØ (2000) Measuring the loss of evolutionary history from 9. Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GVN, et al. extinction: phylogenetically patterned speciation rates and extinction risks alter (2001) Terrestrial ecoregions of the worlds: A new map of life on Earth. the calculus of biodiversity. Proceedings of the Royal Society of London Series Bioscience 51: 933–938. B-Biological Sciences 267: 613–620.

PLoS ONE | www.plosone.org 6 March 2007 | Issue 3 | e296 Mammals on the EDGE

19. McKinney ML (1998) Branching models predict loss of many bird and mammal 39. Flannery TF, Groves CP (1998) A revision of the genus Zaglossus (Monotremata, orders within centuries. Animal Conservation 1: 159–164. Tachyglossidae), with description of new species and subspecies. Mammalia 62: 20. MacPhee RDE, Flemming C (1999) Requiem æternam: the last five hundred 367–396. years of mammalian species extinctions. In: MacPhee RDE, ed (1999) 40. Wilson DE, Reeder DM, eds (2005) Mammal Species of the World. 3rd ed. Extinctions in near time: causes, contexts, and consequences. New York: Baltimore: Johns Hopkins University Press. 2142 p. Kluwer Academic/Plenum. pp. 333–371. 41. Averianov AO, Abramov AV, Tikhonov AN (2000) A new species of Nesolagus 21. Nee S, May RM (1997) Extinction and the loss of evolutionary history. Science (Lagomorpha, Leoporidae) from Vietnam with osteological description. Con- 278: 692–694. tributions from the Zoological Institute, St Petersburg 3: 1–22. 22. Barker GM (2002) Phylogenetic diversity: a quantitative framework for 42. Erwin TL (1991) An evolutionary basis for conservation strategies. Science 253: measurement of priority and achievement in biodiversity conservation. Bi- 750–752. ological Journal of the Linnean Society 76: 165–194. 43. Krajewski C (1991) Phylogeny and diversity. Science 254: 918–919. 23. Steel M (2005) Phylogenetic diversity and the greedy algorithm. Systematic 44. Kemp TS (1999) Fossils and evolution. Oxford: Oxford University Press. Biology 54: 527–529. 45. Hallam A, Wignall PB (1997) Mass extinctions and their aftermath. Oxford: 24. Pavoine S, Ollier S, Dufour AB (2005) Is the originality of a species measurable? Oxford University Press. Ecology Letters 8: 579–586. 46. Briggs DEG, Crowther PR (2001) Palaeobiology II. Oxford: Blackwell Science. 25. Redding DW, Mooers AØ (2006) Incorporating evolutionary measures into 47. Leader-Williams N, Dublin HT (2000) Charismatic megafauna as ‘flagship conservation prioritization. Conservation Biology 20: 1670–1678. species’. In: Entwistle A, Dunstone N, eds (2000) Has the panda had its day? 26. Bininda-Emonds ORP (2004) The evolution of supertrees. Trends in Ecology Priorities for the conservation of mammalian diversity. Cambridge: Cambridge and Evolution 19: 315–322. University Press. pp. 53–81. 27. Isaac NJB, Mallet J, Mace GM (2004) Taxonomic inflation: its influence on 48. Rodrigues ASL, Andelman SJ, Bakarr MI, Boitani L, Brooks TM, et al. (2004) macroecology and conservation. Trends in Ecology and Evolution 19: 464–469. Effectiveness of the global protected area network in representing species 28. Owens IPF, Bennett PM (2000) Quantifying biodiversity: a phenotypic diversity. Nature 428: 640–643. perspective. Conservation Biology 14: 1014–1022. 29. Diniz-Filho JAF (2004) Phylogenetic diversity and conservation priorities under 49. Dudgeon D (2005) Last chance to see…: ex situ conservation and the fate of the distinct models of phenotypic evolution. Conservation Biology 18: 698–704. baiji. Aquatic Conservation: Marine and Freshwater Ecosystems 15: 105–108. 30. Faith DP (2002) Quantifying biodiversity: a phylogenetic perspective. Conser- 50. Sechrest W, Brooks TM, da Fonseca GAB, Konstant WR, Mittermeier RA, et vation Biology 16: 248–252. al. (2002) Hotspots and the conservation of evolutionary history. Proceedings of 31. Bininda-Emonds ORP, Cardillo M, Jones KE, MacPhee RDE, Beck RMD, the National Academy of Sciences of the United States of America 99: et al. (2007) The delayed rise of present-day mammals. Nature, in press. 2067–2071. doi:10.1038/nature05634 51. Rodrigues ASL, Gaston KJ (2002) Maximising phylogenetic diversity in the 32. Butchart SHM, Stattersfield AJ, Bennun LA, Shutes SM, Akc¸akaya HR, et al. selection of networks of conservation areas. Biological Conservation 105: (2004) Measuring global trends in the status of biodiversity: Red List Indices for 103–111. birds. PLoS Biology 2: 2294–2304. 52. Brooks T, Pilgrim JD, Rodrigues ASL, da Fonseca GAB (2005) Conservation 33. Purvis A, Cardillo M, Grenyer R, Collen B (2005) Correlates of extinction risk: status and geographic distribution of avian evolutionary history. In: Purvis A, phylogeny, biology, threat and scale. In: Purvis A, Gittleman JL, Brooks T, eds Gittleman JL, Brooks T, eds (2005) Phylogeny and conservation. Cambridge: (2005) Phylogeny and conservation. Cambridge: Cambridge University Press. Cambridge University Press. pp. 267–294. pp. 295–316. 53. Hartmann K, Steel M (2007) Phylogenetic diversity: From combinatorics to 34. IUCN Red List Standards and Petitions Subcommittee (2006) Guidelines for ecology. In: Gascuel O, Steel M, eds (2007) Reconstructing evolution: new using the IUCN Red List categories and criteria. http://www.iucn.org/ mathematical and computational advances. Oxford: Oxford University Press, in webfiles/doc/SSC/RedList/RedListGuidelines.pdf. press. 35. Wilson DE, Reeder DM, eds (1993) Mammal species of the world: a taxonomic 54. Soutullo A, Dodsworth S, Heard SB, Mooers AO (2005) Distribution and and geographic reference. 2nd ed. Washington DC: Smithsonian Institute Press. correlates of carnivore phylogenetic diversity across the Americas. Animal 36. Groves CP (2001) Primate Taxonomy; Erwin DH, Funk VA, eds. Washington Conservation 8: 249–258. and London: Smithsonian Institute Press. 55. Purvis A (1995) A composite estimate of primate phylogeny. Philosophical 37. Agapow PM, Bininda-Emonds ORP, Crandall KA, Gittleman JL, Mace GM, et Transactions of the Royal Society of London Series B 348: 405–421. al. (2004) The impact of species concept on biodiversity. Quarterly Review of 56. Bininda-Emonds ORP, Gittleman JL, Purvis A (1999) Building large trees by Biology 79: 161–179. combining phylogenetic information: a complete phylogeny of the extant 38. Peterson AT, Navarro-Siguenza AG (1999) Alternate species concepts as bases Carnivora (Mammalia). Biological Reviews of the Cambridge Philosophical for determining priority conservation areas. Conservation Biology 13: 427–431. Society 74: 143–175.

PLoS ONE | www.plosone.org 7 March 2007 | Issue 3 | e296 CORRESPONDENCE

Securing nonflagship species from extinction Liana N. Joseph1,RichardF.Maloney2, James E.M. Watson1, & Hugh P. Possingham3

1 Wildlife Conservation Society, 2300 Southern Boulevard, Bronx, New York 10460, USA 2 Threatened Species Development, Research and Development Group, Department of Conservation, Christchurch, New Zealand 3 ARC Centre of Excellence for Environmental Decisions, The Ecology Centre, University of Queensland, St Lucia 4072, Australia

Received 17 December 2010 Accepted 6 February 2011

Editor James Blignaut doi: 10.1111/j.1755-263X.2011.00174.x

ning exercise, priority actions, and costs and feasibility for Introduction those actions, were identified for securing each of ∼660 A recent article in Conservation Letters by Verissimo of New Zealand’s most threatened species (Joseph et al. and colleagues provides clarity with respect to the 2009; O’Conner et al. 2009). The New Zealand govern- concept of flagship species. As the authors state, the ment is now in the position to state how much it will cost use of flagship species can offer a powerful tool for to secure all or a selection of these species from extinc- environmental organizations to raise money and raise tion. With this kind of information, it is possible to cal- public awareness generally. Regrettably, in many cases, culate the exact amount required to secure species and the money that is raised for flagship species is tied to make statements like: “...as little as $x million is needed spending solely on that species. Consequently, other non- to secure a given number of the most threatened species flagship threatened species are unlikely to benefit. The and $y million would secure a greater number.” Simi- use of flagship species creates a conundrum for those or- larly, these data can be used to demonstrate the expected ganizations that aim to secure the greatest number of gains of additional funding for threatened species. These threatened species from extinction. This goal will not figures give the Department of Conservation a powerful be achievable if the limited conservation budget is con- tool for seeking wider support for managing threatened strained to specific actions that only assist the few flagship species in New Zealand. species. The authors make a brief reference to this weak- The concept of saving large numbers of endangered ness of the flagship-species approach and suggest that species is commonly used to “sell” priority landscapes or solutions may include using the funds to pay for over- regions for conservation NGOs (e.g., Conservation Inter- heads that benefit multiple species or declaring upfront national’s Biodiversity Hotspots, Myers et al. 2000; Al- that funding will be spent on other species. We suggest liance for Zero Extinction sites, Ricketts et al. 2005). Yet, that there is another option: a marketing tool that may the example that we present here illustrates a method be attractive to donors and result in funding that is not for proving clear and fully costed opportunities to raise tied to a single species. funds for priority actions that will result in the recovery We believe that it is possible to raise funds by focus- of threatened species specifically. We suggest that mar- ing on the task of securing large numbers of threatened keting the ability to secure from extinction of large num- species rather than a single flagship species. We illustrate bers of species is an effective complementary tool to the the potential power of this type of marketing tool with a flagship-species approach that can be particularly useful species prioritization exercise recently undertaken by the for securing threatened species that will never be poten- New Zealand Department of Conservation. In this plan- tial flagship species.

324 Conservation Letters 4 (2011) 324–325 Copyright and Photocopying: c 2011 Wiley Periodicals, Inc. L. N. Joseph et al. Protecting non-flagship species

References O’Conner, S., Maloney R., Newman D., Joseph L., Possingham H. (2009) Development of a new approach to Joseph, L.N., Maloney R.F., Possingham H.P. (2009) threatened species management in New Zealand. Pages Optimal allocation of resources among threatened 21–41 in J. Baxter, C. Galbraith, editors. Species species: a project prioritization protocol. Conserv Biol 23, management:challenges and solutions for the 21st century. TSO, 328–338. Scotland, Edinburgh. Myers, N., Mittermeier R.A., Mittermeier C.G., da Fonseca Ricketts, T.H., Dinerstein E., Boucher T. et al. (2005) G.A.B., Kent J. (2000) Biodiversity hotspots for Pinpointing and preventing imminent extinctions. Proc Natl conservation priorities. Nature 403, 853–858. Acad Sci U S A 102, 18497–18501.

Conservation Letters 4 (2011) 324–325 Copyright and Photocopying: c 2011 Wiley Periodicals, Inc. 325 Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 17 6.5.2006 11:07pm

2 Prioritizing choices in conservation

______Georgina M. Mace, Hugh P. Possingham and Nigel Leader-Williams The last word in ignorance is the man who says of an animal or plant: ‘What good is it?’ If the land mechanism as a whole is good, then every part is good, whether we understand it or not. If the biota, in the course of aeons, has built something we like but do not understand, then who but a fool would discard seemingly useless parts? To keep every cog and wheel is the first precaution of intelligent tinkering. (Aldo Leopold, Round River, Oxford University Press, New York, 1993, pp. 145–6.)

tice in other sectors. In business, many com- Introduction panies spend about 10% of the value of their capital assets each year on maintaining those assets, although the figure varies depending on We are in the midst of a mass extinction in the type of asset. For example, 30% might be which at least 10%, and may be as much as spent for computers compared with 5% for 50%, of the world’s biodiversity may disappear buildings: contrast that with 0.02% for bio- over the next few hundred years. Conservation diversity! Furthermore, the scale of underin- practitioners face the dilemma that the cost of vestment in biodiversity may be exaggerated maintaining global biodiversity far exceeds the by the effects of poor governance, sometimes available financial and human resources. Esti- even corruption, on achieving success in mates suggest that in the late twentieth century conservation (Smith et al. 2003). Given such only US$6 billion per year was spent globally problems, conservation scientists and non- on protecting biodiversity (James et al. 1999), government organizations (NGOs) supporting even though an estimated US$33 trillion per international conservation efforts are begin- year of direct and indirect benefits were derived ning to develop systems to more effectively from ecosystem services provided by biodiver- target investment in biodiversity conservation sity, implying an asset worth US$330 trillion (Johnson 1995; Kershaw et al. 1995; Olson & (Costanza et al. 1997). Together these crude Dinerstein 1998; Myers et al. 2000; Possingham estimates suggest that there could be a 500- et al. 2001; Wilson et al. in press). fold underinvestment in conserving the world’s One fundamental resource allocation question biodiversity. However, even if these estimates facing conservation scientists and practitioners is are wildly wrong, the imbalance of funding is whether conservation goals are best met by man- seriously inconsistent with best business prac- aging single species as opposed to whole ecosys- Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 18 6.5.2006 11:07pm

18 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS tems (Simberloff 1998). Efforts in conservation clude judgments that cannot be made by scien- priority setting have historically concentrated on tists or managers alone. Involving wider ecosystem-based priorities – determining where societal and political decision-making processes and when to acquire protected areas (Ferrier et is vital to gain local support for, and ensure the al. 2000; Margules & Pressey 2000; Pressey & ultimate success of, all conservation planning. Taffs 2001; Meir et al. 2004). There has been comparatively little work on the question of how to allocate conservation effort between Single species approaches species. Despite the tension between ecosystem- based and species-based conservation, we believe there is merit in considering the issue of resource Species-based conservation management ap- allocation between species because: proaches have, until fairly recently, concentrated on a single species, such as keystone species, 1. a ‘fuzzy’ idea such as ecosystem manage- umbrella species, indicator species or flagship ment holds little appeal for the general pub- species (see Leader-Williams & Dublin 2000). lic, who prefer to grasp simpler messages conveyed by charismatic species such as Keystone and umbrella species differ in the im- tigers (Leader-Williams & Dublin 2000); portance of their ecological role in an ecosystem: 2. data on species, whether through direct counts of indicator species (Heywood 1. keystone species have a disproportionate 1995), or through assessments of threat effect on their ecosystem, due to their size or (Butchart et al. 2005), provide some of the activity, and any change in their population most readily available, repeatable and expli- will have correspondingly large effects on cit monitoring and analytic systems with their ecosystem (e.g. the sole fruit disperser which to assess the success or otherwise of of many species of tree); conservation efforts (Balmford et al. 2005). 2. umbrella species have such demanding 3. in practice, almost regardless of their ultim- habitat and/or area requirements that, if we ate goal, conservation bodies often end up can conserve enough land to ensure their directing conservation actions to species viability, the viability of smaller and more and species communities (see e.g. figure 1 abundant species is almost guaranteed. of Redford et al. 2003), probably because these are tangible and manageable compon- In contrast, ‘flagship species’ encompass ents of ecosystems. purely strategic objectives: The topic of setting priorities for conservation is 3. flagship species are chosen strategically to immense, so here we restrict ourselves to dif- raise public awareness or financial support ferent methods for setting priorities between for conservation action. species. We explore the issues that a systematic approach should consider, and we show how Furthermore, definitions for indicator species simple scoring systems may lead to unintended can encompass both ecological and strategic consequences. We also recommend an explicit roles: discussion of attributes of the species that make them desirable targets for conservation effort. 4. indicator species are intended either to Using a case study, we show how different represent community composition or to re- perspectives will affect the outcome, and so as flect environmental change. With respect to an alternative we present a method based the latter, indicator species must respond to on economic optimization. Ultimately, any the particular environmental change of con- decisions about ‘what to save first’ should in- Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 19 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 19

cern and demonstrate that change when predators have a different view (Leader- monitored. Williams & Dublin 2000). Such dissonance is best avoided by promoting locally supported One species may be a priority species for flagship species (Entwistle 2000). For example, more than one reason, depending on the situ- the discovery of a new species of an uncharis- ation or context in which the term is used. matic, but virus-resistant, wild maize, with its However, the terms ‘keystone’ and ‘umbrella’ possible utilitarian value for human food pro- are likely to remain more of a fixed character- duction, highlighted the conservation value of istic or property of that species. In contrast, the the Mexican mountains in which it was found term ‘flagship’ and, possibly to a lesser extent, (Iltis 1988). This increase in local public aware- ‘indicator’ may be more context-specific. ness led to the establishment of a protected area Promoting the conservation of a specific focal that conserves parrots and jaguars (Panthera species may help to identify potential areas for onca), orchids and ocelots (Leopardus pardalis), conservation that satisfy the needs of other spe- species that many consider charismatic. Hence cies and species assemblages (Leader-Williams this species of wild maize served as a strategic- & Dublin 2000). For example, the umbrella ally astute local flagship species. Another way species concept (Simberloff 1998) can represent of promoting local flagships is to prioritize those an efficient first step to protect other species. In species that bring significant and obvious local addition, minimizing the number of species benefits (Goodwin & Leader-Williams 2000), that must be monitored once a protected area such as the Komodo dragon, Varanus komodoen- has been created will reduce the time and sis (Walpole & Leader-Williams 2001), which money that must be devoted to its maintenance generates tourism. Similarly, species that can (Berger 1997). be hunted for sport, such as the African ele- Alternatively, conservation managers and phants (Loxodonta africana), may contribute dir- international NGOs may choose to focus on ectly to community conservation programmes the most charismatic ‘flagship’ species, which (Bond 1994). stimulate public support for conservation ac- Several questions can arise from promoting tion, and that in turn may have spin-off bene- conservation through single species (Simberloff fits for other species. For example, use of the 1998). One of these is how should individual giant panda (Ailuropoda melanoleuca) as a logo species be prioritized? The common response is by the World Wildlife Fund (WWF) has been to begin with species that are most at risk of widely accepted (Dietz et al. 1994) as a success- extinction, the critically endangered species. ful mechanism for conserving many other spe- Many countries and agencies take this ap- cies across a wide variety of taxonomic groups. proach. However, there may be no known Furthermore, other mammalian and avian management for some of these species, and if ‘flagships’ have been used to promote the con- there is, it may be risky and/or expensive. This servation of large natural ecosystems (i.e. Mit- can lead to a large share of limited conservation termeier 1986; Goldspink et al. 1998; Downer resources being expended with negligible or 1996; Johnsingh & Joshua 1994; Western uncertain benefit (Possingham et al. 2002). 1987; Dietz et al. 1994). On the other hand, when taking an ecosystem Nevertheless, the context of what may con- approach, managers might choose to focus on stitute a charismatic species can differ widely the keystone species that play the most signifi- across stakeholders. For example, the tiger cant role in the ecosystem. Unfortunately in (Panthera tigris) is among the most popular flag- many ecosystems we do not know the identity ship species in developed countries, but those of keystone species. Often, after intensive in developing countries who suffer loss of life study, they turn out to be invertebrates or and livelihood because of tigers or other large fungi (Paine 1995), groups that are unlikely to Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 20 6.5.2006 11:07pm

20 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS attract public or government support unless cies, it does highlight the need for more field- ways can be found to make them locally based studies to determine the most appropriate relevant. approach for conserving the most biodiversity. Another problem with single species conser- As a result of problems associated with single vation arises when the management of one focal species management, focus has been turning species is detrimental to the management of towards multiple species approaches. another focal species. For example, in the Ever- glades of Florida, management plans for two charismatic, federally listed birds are in conflict. Multispecies approaches One species, the Everglades snail kite (Rostrha- mus sociabilis plumbeus), has been reduced to some 600 individuals by wetland degradation Methods based on several focal species, or pro- and agricultural and residential development. tecting a specific habitat type, might be a more It feeds almost exclusively on freshwater snails appropriate means of prioritizing protected of the genus Pomacea and requires high water areas (Lambeck 1997; Fleishman et al. 2000; levels, which increase snail production. The Sanderson et al. 2002b). A frequent criticism of snail kite is thus an extreme habitat specialist setting conservation priorities based on a single (Ehrlich et al. 1992). The other species, the focal species is that it is improbable that the wood stork Mycteria americana, has been reduced requirements of one species would encapsulate to about 10,000 pairs by swamp drainage, habi- those of all other species (Noss et al. 1996; Basset tat modification and altered water regimes. et al. 2000; Hess & King 2002; Lindenmayer et al. Ironically, the US Fish and Wildlife Service op- 2002). Hence, there is a need for multispecies posed a proposal by the Everglades National strategies to broaden the coverage of the pro- Park to modify water flow to improve stork tective umbrella (e.g. Miller et al. 1999; Fleish- habitat on the grounds that the change would man et al. 2000, 2001; Carroll et al. 2001). be detrimental to the kite (Ehrlich et al. 1992) Among the different multispecies ap- (an added thought-provoking detail is that both proaches, Lambeck’s (1997) ‘focal species’ ap- species are common in South America). proach seems the most promising because it Another issue is that few studies have been provides a systematic procedure for selecting carried out to assess the effectiveness of one several focal species (Lambeck 1997; Watson focal species in adequately protecting viable et al. 2001; Bani et al. 2002; Brooker 2002; populations of other species (Caro et al. 2004). Hess & King 2002). In Lambeck’s (1997) in- For example, the umbrella-species concept is novative approach, a suite of focal species are often applied in management yet rarely tested. identified and used to define the spatial, com- The grizzly bear (Ursus arctos) has been recog- positional and functional attributes that must nized as an umbrella species but, had a pro- be present in a landscape. The process involves posed conservation plan for the grizzly bear in identifying the main threats to biodiversity and Idaho been implemented, taxa such as reptiles selecting the species that is most sensitive to would have been underrepresented (Noss et al. each threat. The requirements of this small 1996). Similarly, in a smaller scale study, the and manageable suite of focal species guide areas where flagship species, such as jaguar, conservation actions. The approach was tapir (Tapirus terrestris) and white-lipped pec- extended by Sanderson et al. (2002a), who cary (Tayassu pecari), were most commonly proposed the ‘landscape species approach’. seen did not coincide with areas of vertebrate They defined landscape species by their ‘use of species richness or abundance (Caro et al. 2004). large, ecologically diverse areas and their im- Although these results may not hold true for all pacts on the structure and function of natural other protected areas based around flagship spe- ecosystems . . . their requirements in time and Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 21 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 21 space make them particularly susceptible to 4. manage over periods of time long enough human alteration and use of wild landscapes’. to maintain the evolutionary potential of Because landscape species require large, wild species; areas, they could potentially serve an umbrella 5. accommodate human use and occupancy function (sensu Caro & O’Doherty 1999) – within these constraints (Grumbine 1994). meeting their needs would provide substantial Although the financial efficiencies inherent protection for the species with which they co- in managing an ecosystem rather than several occur. Like other focal-species approaches, this single species are attractive, this approach is method of setting priorities carries inherent also not without its problems. First, compared biases (Lindenmayer et al. 2002), and may be with a species, ecosystem boundaries are constrained by incomplete or inconsistent data. harder to define, so determining the location, size, connectivity and spacing of protected areas Ecosystem and habitat-based to conserve the full range of ecosystems, and approaches variation within those ecosystems, is more dif- ficult (Possingham et al. 2005). Second, indi- vidual species are more interesting to people and will attract greater emotional and financial Some conservation scientists believe that set- ting conservation priorities at the scale of eco- investments than ecosystems. Third, although ecological services are provided by ecosystems, systems and habitats is more appropriate for individual species often play pivotal roles in the developing countries with limited resources for conservation, inadequate information about provision of these services, particularly for dir- ect uses such as tourism or harvesting. Finally, single species and pressing threats such as the main problem faced by managers wishing habitat destruction. Logically, how much effort we place in conserving a particular ecosystem to implement an ecosystem approach is the lack of data available on how ecosystems function. should take into account factors such as: how This manifests itself in confusion about how threatened it is, how well represented that ecosystem is in that country’s protected area much of each ecosystem needs to be conserved to protect biodiversity adequately in a region. network, the number of species restricted to In contrast, for the better known single species, that ecosystem (endemic species), the cost of conserving the ecosystem and the likelihood the issue of adequacy can be dealt with using population viability analysis and/or harvesting that conservation actions will work. One can models (Beissinger & Westphal 1998; this vol- debate the relative importance of each of these factors – for example, some consider the the ume, Chapter 15). number of endemic species is paramount, whereas others prefer the notion of ‘equal rep- resentation’ whereby a fixed percentage of Systematic conservation planning every habitat type is conserved. The main goals of an ecosystem approach are to: Systematic conservation planning (or gap-an- alysis in the USA: Scott et al. 1993) focuses on 1. maintain viable populations of all native locating and designing protected areas that species in situ; comprehensively represent the biodiversity of 2. represent, within protected areas, all native each region. Without a systematic approach, ecosystem types across their natural range protected area networks have the tendency to of variation; occur in economically unproductive areas 3. maintain evolutionary and ecological processes; (Leader-Williams et al. 1990), leaving many Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 22 6.5.2006 11:07pm

22 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS habitats or ecosystems with little or no protec- threats. The degree to which different countries tion (Pressey 1994). The systematic conserva- use species-based planning as opposed to sys- tion planning approach can be divided into six tematic conservation planning depends on his- stages (Margules & Pressey 2000). torical, cultural and legislative influences. Even with systematic conservation planning, how- 1. Compile biodiversity data in the region of ever, the better surveyed species or species concern. This includes collating existing groups often feature as the units for assessment. data, along with collecting new data if ne- In other words, the conservation value of dif- cessary, and if time and funds permit. ferent areas is often assessed on the presence or Where biodiversity data, such as habitat conservation status of the species within it, maps and species distributions, are limited simply because these are the best known elem- more readily available biophysical data may ents of biodiversity. Systematic conservation be used that reflect variation in biodiversity, such as mean annual rainfall or soil type. planning approaches have become popular 2. Identify conservation goals for the region, and widespread, partly because they are sup- including setting conservation targets for ported by several decision-support software species and habitats, and principles for pro- packages (Possingham et al 2000, Pressey et al tected area design, such as maximizing con- 1995, Williams et al 2000, Garson et al 2002). nectivity and minimizing the edge-to-area ratio. 3. Review existing conservation areas, includ- ing determining the extent to which they Methods for setting conservation already meet quantitative targets, and miti- priorities of species gate threats. 4. Select additional conservation areas in the region using systematic conservation plan- Prioritizing species, habitats and ecosystems by ning software. their perceived level of endangerment has be- 5. Implement conservation action, including come a standard practice in the field of conser- decisions on the most appropriate form of vation biology (Rabinowitz 1981; Master 1991; management to be applied. Mace & Collar 1995; Carter et al. 2000; Stein 6. Maintain the required values of the conser- vation areas. This includes setting conserva- et al. 2000). The need for a priority-setting tion goals for each area, and monitoring key process is driven by limited conservation re- indicators that will reflect the success of sources that necessitate choices among a subset management (see below). of all possible species in any given geographical area, and distinct differences among species in Ultimately, conservation planning is riddled their apparent vulnerability to extinction or with uncertainty, so managers must learn to need for conservation action. This need has deal explicitly with uncertainty in ways that led to the development of practical systems minimize the chances for major mistakes (Mar- for categorizing and assessing the degree of vul- gules & Pressey 2000; Arau´ jo & Williams 2000, nerability of various components of biodiver- Wilson et al 2005), and be prepared to modify sity, particularly vertebrates (e.g. Millsap et al. their management goals appropriately through 1990; Mace & Lande 1991; Master 1991; Reed adaptive management. 1992; Stotz et al. 1996), and more recently Systematic conservation planning can com- ecoregions (Hoekstra et al. 2005). plement species-based approaches because it Methods used for assessing the conservation focuses on removing the threat of development status of species are varied but follow three and it compliments a long tradition of species general styles (Regan et al. 2004), rule-based, recovery plans that concentrate on mitigating point scoring and qualitative judgement. Per- Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 23 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 23 haps the best known system is that developed more of the criteria – for example because there by the IUCN (International Union for the Con- are less than 250 mature individuals of the servation of Nature and Natural Resources) – Norfolk Island green parrot (Cyanoramphus coo- The World Conservation Union – which uses a kii) in the wild it is immediately listed as endan- set of five quantitative rules with explicit gered under criterion D of the IUCN Red List thresholds to assign a risk of extinction (Mace protocol. A similar species can meet a higher & Lande 1991; IUCN 2001). Other methods category of threat if it meets alternative cri- adopt point-scoring approaches where points teria. For example, the orange-bellied parrot are assigned for a number of attributes and (Neophema chrysogaster) also has less than 250 summed to indicate conservation priority (Mill- mature individuals but it is listed as critically sap et al. 1990; Lunney et al. 1996; Carter et al. endangered, under criterion C2b, because the 2000). Other methods assess conservation sta- population is also in decline and all the individ- tus using qualitative criteria; judgements about uals are in a single subpopulation. One concep- a species’ status are determined intuitively tual problem with rule-based methods is that a based on available information and expert species that just missed out on being listed as, opinion (Master 1991). One widely applied sys- say, endangered on several criteria would be tem is the biodiversity status-ranking system ranked as vulnerable, equal with a species that developed and used by the Natural Heritage may have only just met the criteria for being Network and The Nature Conservancy (Master vulnerable. 1991; Morse 1993). This ranking system has The rule-based methods have the advantage been designed to evaluate the biological and that they are completely explicit about what conservation status of plant and animal species feature of the species led to it being listed as and within-species taxa, as well as of ecological threatened. In the IUCN system, assessors have communities. to list the criteria whereby the species qualified for a particular category of threat, and also have to provide documentation to support this infor- Rule-based methods mation – usually in the form of scientific sur- veys or field reports that detail the information Quantitative rule-based methods can be used to used. As a result, listings may be continually estimate the extinction risk of a species and updated and improved as new data become thus contribute to determining priority areas available. Normally this will allow a new con- for conservation action. For example, the sensus among experts, but in the exceptional IUCN Red List places species in one of the fol- cases where this is not agreed, the IUCN have a lowing categories: extinct (EX), extinct in the petitions and appeals process to resolve matters. wild (EW), critically endangered (CR), endan- For example, in 2001 some of the listings of gered (EN), vulnerable (V), near threatened marine turtle species were disputed among ex- (NT) or least concern (LC), based on quantita- perts. On this occasion, IUCN implemented tive information for known life history, habitat their appeal procedure and provided a new requirements, abundance, distribution, threats assessment (http://www.iucn.org/themes/ssc/ and any specified management options of that redlists/petitions.html). The wide use of the species, and in a data deficient (DD) category if IUCN system also means that there is an ever there are insufficient data to make an assess- increasing resource of best-practice documen- ment (IUCN 2001). The IUCN system is based tation and guidelines, which aid consistent and around five criteria (A to E) which reflect dif- comparable approaches by different species ferent ways in which a species might qualify for assessors (see http://www.iucn.org/themes/ssc/red- any of the threat categories (CR, EN, VU). A lists.htm). species is placed in a category if it meets one or Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 24 6.5.2006 11:07pm

24 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS

A problem with some point-scoring methods Point scoring method is that there is no explicit link to extinction risk, the weightings of each criteria, from 1 to 5 in The point scoring method for assigning conser- the example above, are completely arbitrary, vation priority involves assigning a series of and there is an infinity of ways in which the scores to each species based on different param- scores could be combined: adding, multiplying, eters relating to their ecology or conservation taking the product of the largest three values, status, which together will determine their and so forth. A related problem is that point- relative priority. One method of dealing with scoring methods can generate an artificially the scores is then to simply sum them to give an high ranking for a species when criteria are overall conservation priority, although this can interrelated. For example, a system that priori- be misleading. Beissinger et al. (2000) suggest tized species because they needed large home that a categorical approach based on a combin- ranges, had slow reproductive rates and small ation of scores might be more accurate in litter sizes might end up allocating unreason- determining overall conservation priority. ably high scores to any large-bodied species. An example of a point scoring system is that All three of these traits are associated with developed by Partners in Flight (PIF) in 1995 in relatively large body size, but they are not an effort to conserve non-game birds and their necessarily so much more vulnerable. habitats throughout the USA (Carter et al. 2000). The PIF system involves assigning a series of scores to each species ranging from 1 (low priority) to 5 (high priority) for seven param- Conservation status ranks method eters that reflect different degrees of need for conservation attention. The scores are assigned Status ranks are based primarily on objective within physiographical areas and the seven factors relating to a species’ rarity, population parameters are based on global and local infor- trends and threats. Four aspects of rarity are mation. Three of the parameters are strictly typically considered: the number of individuals, global and are assigned for the entire range of number of populations or occurrences, rarity of the bird: breeding distribution (BD), non-breed- habitat, and size of geographic range. Ranking ing distribution (ND) and relative abundance is based on an approximately logarithmic scale, (AR). Other parameters are threats to breeding ranging from 1 (critically imperiled) to 5 (dem- (TB), threats to non-breeding (TN), population onstrably secure). Typically species with ranks trend (PT) and, locally, area importance (AI). from 1 to 3 would be considered of conserva- The scores for each of these seven parameters tion concern and broadly overlap with species are obtained independently (Carter et al. 2000). that might be considered for review under the The PIF then uses a combination of approaches, Endangered Species Act or similar state or including the summing of scores, to determine international statutes. an overall conservation priority (Carter et al. The NatureServe system (Master 1991) is one 2000), with species that score highly on several example of a system that uses status ranks. parameters achieving high priority. Although Developed initially by The Nature Conservancy this method of defining bird species of high con- (TNC) and applied throughout North America, servation priority is thought to be reliable, like the NatureServe system uses trained experts other methodologies, it is hindered by the lack who evaluate quantitative data and make in- of data on species distribution, abundance and tuitive judgements about species vulnerability. populations trends, particularly in areas outside The aim of the NatureServe system is to deter- the USA to which many of these species migrate mine the relative susceptibility of a species or (Carter et al. 2000). ecological community to extinction or extirpa- Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 25 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 25 tion. To achieve this, assessments consider both question of how quickly action needs to be deterministic and stochastic processes that can taken. Hence, all other things being equal, the lead to extinction. Deterministic factors include critically endangered species will be most likely habitat destruction or alteration, non-indigen- to become extinct first if nothing is done. How- ous predators, competitors, or parasites, over- ever, this is by no means the only consideration harvesting and environmental shifts such as that should be used by a conservation planner. climate change. Stochastic factors include, en- How then should extinction risk be used for vironmental and demographic stochasticity, priority-setting? It may be easier to make the natural catastrophes and genetic effects (Shaffer analogy with a different system altogether. For 1981). example, the priority-setting systems used by NatureServe assessments are performed on a Triage nurses in hospital emergency depart- basic unit called an element. An element can ments categorize people according to how ur- be any plant or animal species or infraspecific gently they need to be seen; those seen first are taxon (subspecies or variety), ecological com- the ones that appear to have the most urgent munity, or other non-taxonomic biological and threatening symptoms. The symptoms can entity, such as a distinctive population (e.g. be very diverse, however, and some may turn evolutionarily significant unit or distinct popu- out upon inspection and diagnosis to be less lation segment, as defined by some agencies) or serious than might have been expected. Medical a consistently occurring mixed species aggrega- planning across the board would not use the tion of migratory species (e.g. shorebird migra- triage system to allocate resources. The same is tory concentration area) (Regan et al. 2004). true for conservation planning. As with ill and Defining elements in this way ensures that a injured people, our first sorting of cases should broad spectrum of biodiversity and ecological be according to urgency, and should also be processes are identified and targeted for conser- precautionary (i.e. take more risks with listing vation (Stein et al. 2000). This approach is be- species that are in fact not threatened than with lieved to be an efficient and effective approach failing to list those that really are). However, to capturing biodiversity in a network of once the diagnosis is made, and the manager reserves (e.g. Jenkins 1976, 1996). Assessment is reasonably sure that most critical cases are results in a numeric code or rank that reflects now known and diagnosed, a more systematic an element’s relative degree of imperilment or planning process should follow. risk of extinction at either the global, national or subnational scale (Master et al. 2000). Variables other than risk

Back to basics – extinction risk versus setting priorities Now we consider a whole range of new variables other than risk. Table 2.1 shows a range of vari- The discussion above has reviewed methods for ables – grouped under headings of biological categorizing species according to their conser- value (i.e. what biologists would consider), eco- vation priority. Running throughout is a ten- nomic value, social and cultural value, urgency dency to equate conservation priority with and practical issues. Under each of these head- extinction risk; yet these are clearly not the ings are a range of attributes that might contrib- same thing (Mace & Lande 1991). Extinction ute to a species priority. The first three columns risk is only one of a range of considerations that concern values, but the last two are rather dif- determine priorities for action or for conserva- ferent. Urgency is a measure that can be compli- tion funding. The threat assessment is really an cated to implement – i.e. high urgency may assessment of urgency, and an answer to the indicate that if nothing is done now, then it Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 26 6.5.2006 11:07pm

26 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS

Table 2.1 Classes and kinds of issues that are considered in priority-setting exercises for single-species recovery

Social and Biological value Economic value cultural value Urgency Practical issues

Degree of endemism Cost of management Scientific and Threat status Feasibility and logistics or recovery educational benefits ¼ extinction risk Relictual status Direct economic Cultural status Time limitation, Recoverability, i.e. benefits (e.g. ceremonial) i.e. opportunities reversibility of threats, will be lost later rate of species response Evolutionary Indirect economic Political status Timeliness, i.e. Popularity – will there uniqueness benefit (e.g. symbolic or likelihood of be support from the emblematic) success varies community? with time Collateral benefits to Ecological services Popularity Responsibility, i.e. how other species much is this also someone else’s responsibility? Collateral costs to Local or regional Land tenure other species significance Ecological uniqueness Governmental/agency jurisdictions Keystone species status Umbrella species status

will be too late. This measure is not a value score Interestingly, the criteria that biologists com- that can easily be added to the others, and a monly consider, and which form the basis of moderate score has little meaning. Practical most formal decision-processes, fall under one issues are also rather different, and will vary heading (biological benefits). Yet in practice, greatly in their nature and importance depend- the other five criteria (Table 2.2) also influence ing on the context. Some species that are con- real decisions. Would it not, therefore, be pref- sidered urgent cases may be extremely erable to incorporate these other criteria expli- impractical and/or costly to attend to. This set citly in the process of setting priorities? of considerations is probably not complete, but it does illustrate the point that there are more things to think about than extinction risk. Turning criterion-based ranks This initial classification by the value type is into priorities hard to manage in a priority-setting system. Therefore, in Table 2.2, we classify these into A potential next step would be to add the scores six criteria reflecting the nature of the attribute from Table 2.2. By allocating a score of 1, 2 or 3 (importance, feasibility, biological benefits, to each criterion and then adding the ranks, an economic benefits, urgency and chance of suc- overall priority could be calculated. We advise cess). This classification has the advantage that against this for several reasons. First, the differ- the different questions are more or less inde- ent variables are not equal; we might for ex- pendent of one another, and each addresses a ample wish to weight the biological issues question that public, policy-makers and scien- more highly. Second, they are not additive: as tists can all address, and for which they can mentioned earlier both urgency and chance provide at least relative scores. of success are all or nothing decisions. For Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 27 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 27

Table 2.2 Criteria for setting prorities. The different kinds of considerations from Table 2.1 are classified into six criteria (rows), each of which can be qualitatively assessed for a particular species

Criterion Explanation Subcriteria Scores

Importance ‘Does anyone care?’ A Social and cultural importance Important (I) measure of how much (including charisma) Moderately important (M) support there is likely to be Responsibility – Unimportant (U) how much of the species status depends on this project? Feasibility ‘How easy is this to achieve?’ Logistical and political, source of Feasible (F) An assessment of the difficulty funds, community attitudes Moderately difficult (M) associated with this project Biological Difficult (D) Benefits ‘What good will it do?’ A Reduction in extinction risk, Highly beneficial (H) measure of how much good increase in population size, extent Moderately beneficial (M) will result from the project. of occurrence Unclear benefits (U) Collateral biological benefits, to other species or processes Costs ‘What will it cost?’ An assessment Direct and indirect costs of project Expensive of the relative economic costs Direct and indirect social and Moderately costly of the project (or gains). In this economic costs and benefits that will Inexpensive criterion there are both postive flow from the project and negative aspects which have to be weighed against each other Urgency ‘Can it be delayed?’ A measure Extinction risk, potential for loss of Urgent of whether the project is time- opportunity if delayed Moderately urgent limited, or whether it can be Less urgent delayed Chance of ‘Will it work?’ An assessment of Will it meet its specified objectives? Achievable success whether or not the project will Uncertain work Highly uncertain

example, if chance of success is nil we would may lead to ambiguous results. An expedient not wish to invest in that species at all, so it process at this stage is to invite a range of experts would seem more logical to multiply other representing different perspectives to rate the scores by the chance of success. Third, although priorities explicitly. For example, given the pos- we have sorted the issues into more-or-less sible set of scores in Table 2.2, what set would independent categories, there still are associ- they most wish to see in the top priorities versus ations between them. For example, the feasibil- those lower down? This sounds complicated but ity and chance of success are likely to be in practice we think it is feasible. positively correlated, as are biological benefits A good example of this approach was devel- and importance. Hence, simple scoring can lead oped for UK birds by the Royal Society for the to double-counting, which is not what was in- Protection of Birds (Avery et al. 1995). Three tended. criteria were used: global threat, national de- Multicriteria decision-analysis is one decision- cline rates and national responsibility, and each making tool for choosing between priorities that was rated high, medium or low. However, by rate differently for separate criteria. There are simply adding these scores, globally endan- innumerable ways of carrying out a multicriteria gered species that are stable, and for which analysis, and the process can be complex and the UK has medium responsibility, had the Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 28 6.5.2006 11:07pm

28 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS

Conservation 1 ance and economic benefits and minimize risk priority 1 of failure. Hence the two profiles would look Global threat set 1 1 11 1 quite different (Fig. 2.2). Figure 2.2 illustrates 2 the different approaches – see how you would 2 222 1 score the criteria in Table 2.2 to make your own 2 set! 2 3 3 2 Here we are effectively creating a complex Responsibility rule set that maps any species into one of three categories without adding or multiplying National decline the scores for different criteria. The method Fig. 2.1 The conservation cube. (From Avery et al. suffers from its somewhat arbitrary nature. 1995.) Below we suggest that optimal allocation of funds between species can be achieved more same priority as globally secure species exhibit- rigorously if we place the problem within an ing slow decline in the UK. This would not be explicit framework in which we can apply most people’s choice; whatever their status in decision theory. the UK, a globally endangered species probably should be in the category of highest priority. Hence, Avery et al. (1995) set priorities using A decision theory approach – optimal their conservation cube (Fig. 2.1). Here they allocation evaluated each of the 27 possible circumstances into three categories for priority. In their A major problem with using scores or ranks for system, any globally threatened species and threatened species to determine funding and any species declining at a high rate nationally action priorities is that these methods were not are the highest priority. designed for that task – they were designed to This approach can be taken more generally determine the relative level of threat to a suite of using the six criteria in Table 2.2. By asking species (Possingham et al. 2002). Hence they what would be the criteria associated with top cannot provide the solution to the problem of priority species, it is possible to assemble a pro- optimal resource allocation between species – file. For example, whereas a species conserva- this problem should be formulated then solved tion ‘idealist’ might choose to ignore properly (Possingham et al. 2001). importance, feasibility, economic benefits and Optimal allocation is one simple and attract- chance of success, and to focus just on the most ive approach to prioritization that could inform urgent and most threatened forms, a more ‘pol- decisions about how to allocate resources be- itical’ approach would be to maximize import- tween species. It requires information about

Manager 1 Manager 2 Idealist Politician HML HML HML HML Importance Feasibility Biological benefits Economic benefits Urgency Chance of success

Fig. 2.2 Priority sets for four different people. The blocked out cells indicate the conditions under which assessors would choose to include species in their priority set, according to how they scored on the variables in Table 2.2 as H, high; M, medium; L, low. Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 29 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 29 the relationship between the resources allo- For poorly known taxa the curves showing cated to the species and the reduction in prob- this relationship would very much be a reflec- ability of extinction. Here we use expert tion of expert opinion, garnered by asking opinion and/or population models to estimate questions about how much it might cost to the relationship between percentage recovery give a particular species a 90% chance of not (measured, for example, in terms of probability becoming extinct in the long term. Given a set of not becoming extinct) and the funds allo- amount of money for a set period in the con- cated to that species. servation budget, the optimal allocation of

100 90 80 70 60 50 Scheme to maximise the total amount of recovery, given $ amount 40 Recovery (%) 30 20 10 0 0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000 6,000,000 (a) $ spent

Species accumulation

16 14 12 10 8 6 4 2 Expected species recovered 0 0.E+00 1.E+07 2.E+07 3.E+07 4.E+07 5.E+07 6.E+07 7.E+07 8.E+07 9.E+07 1.E+08 (b) $ in budget

Fig. 2.3 Optimal allocation. (a) Three curves show the expected recovery for three different species given certain amounts of investment. If the manager has a specified budget (in this case $1 million), the optimal allocation among species that achieves the greatest total amount of recovery will result if funds are allocated as shown by the vertical dotted lines (see Possingham et al. 2002). (b) Increasing investment leads to gradually increasing numbers of species recovered. Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 30 6.5.2006 11:07pm

30 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS funds can be determined between species. This When species are rated highly by the criteria occurs when the rate of gain of recovery for the two approaches give similar results, but at each species is equal, such that there is no low criterion scores there can be much variabil- advantage in shifting resources from one spe- ity. Perhaps the only general conclusion here is cies to another (see Fig. 2.3 and Possingham that inevitably the optimal allocation approach et al. 2002). The implicit objective is to maxi- will favour some species that, on the basis of the mize the mean number of recovered species criteria, would not be given high priority. In given a fixed budget and assuming all species practice, sensible management could use both are of equal ‘value’. approaches – the criteria to select high-priority Using the set of species plotted in Fig. 2.3, we sets and the financial algorithm to then maxi- estimate the costs of recovery, and then find mize the benefits from the finite resources avail- the optimal allocation of funds per species. able to conservation. The species accumulation curve shows the total expected number of species that can be recovered given a conservation budget. The al- Conclusions gorithm will tend first to select species that show large recovery for relatively low costs. Slow responders will be conserved later. Given Priority setting needs to consider a range of an annual budget basis, the more intractable variables, and although this undoubtedly oc- conservation problems may never be funded curs, it is not always transparent. Although because the selection process will always favour much effort has gone into biologically based allocation of resources to the species that pro- systems, in practice other societal value judg- vide the greatest gains for the smallest costs ments are often included. We suggest that, if (the low-hanging fruit). conservation goals are to be achieved, it is vital So how would these two approaches: cri- to be explicit about what these are, and to teria-based prioritization and optimal allocation decide upon them in an open and consultative of resources differ in practice? Obviously there manner before choices are made. is no general answer to this, other than a priori Different people and organizations, and differ- we do not expect them to be the same. The ent sectors in society, will make different choices outcome of a small case study, based on real in their value judgments. Approaches to under- species and the expertise of two real conserva- standing these choices are important so we can tion managers is shown in Fig. 2.4. interpret the differences in setting priorities.

4.5 4 3.5 3 2.5 2 1.5

Optimisation rank 1 0.5 0 012345 Criterion rank

Fig. 2.4 Comparison of priority ranks for18 species using the criteria-based method versus optimal allocation of funds. Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 31 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 31

We recommend using more than one decisions about resource allocation be formu- method to set priorities, and the comparison lated more explicitly in terms of objectives, can be informative. We also recommend that constraints and costs.

For if one link in nature’s chain might be lost, another might be lost, until the whole of things will vanish by piecemeal. (Thomas Jefferson (1743–1826) in Charles Miller, Jefferson and Nature, 1993.)

Brooker L. (2002) The application of focal species References knowledge to landscape design in agricultural lands using the ecological neighbourhood as a template. Landscape and Urban Planning 60: 185– Arau´ jo, M.B. & Williams, P.H. (2000) Selecting areas 210. for species persistence using occurrence data. Bio- Butchart, S.H.M., Stattersfield, A.J., Bennun, L.A., logical Conservation 96: 331–45. et al. (2005) Measuring global trends in the status Avery, M., Gibbons, D.W., Porter, R., Tew, T., of biodiversity: Red List Indices for birds. Public Tucker, G. & Williams, G. (1995) Revising the Library of Science Biology 2: 2294–304. British Red Data List for birds: the biological Carroll, C., Noss, R.F. & Paquet, P.C. (2001) Carni- basis of UK conservation priorities. Ibis 137(S1): vores as focal species for conservation planning in 232–9. the Rocky Mountain region. Ecological Applications Balmford, A., Bennun, L., ten Brink, B., et al. (2005) 11: 961–80. Science and the Convention on Biological Diver- Caro, T. & Doherty, G. (1999) On the use of surro- sity’s 2010 target. Science 307: 212–13. gate species in conservation. Conservation Biology Bani, L., Baietto, M., Bottoni, L. & Massa, R. (2002) 13: 805–14. The use of focal species in designing a habitat Caro, T., Engilis, A., Fitzherbert, E. & Gardner, T. network for a lowland area of Lombardy, Italy. (2004) Preliminary assessment of the flagship species Conservation Biology 16: 826–31. concept at a small scale. Animal Conservation 7: 63–70. Basset, Y., Novotny, V., Miller, S.E. & Pyle, R. (2000) Carter, M.F., Hunter, W.C., Pashley, D.N. & Rosen- Quantifying biodiversity: experience with paratax- berg, K.V. (2000) Setting conservation priorities onomists and digital photography in Papua New for landbirds in the United States: the Partners in Guinea and Guyana. Bioscience 50(10): 899–908. Flight Approach. Auk 117: 541–8. Beissinger, S.R. & Westphal, M.I. (1998) On the use Costanza, R., d’Arge, R., de Groot, R., et al. (1997) of demographic models of population viability in The value of the world’s ecosystem services and endangered species management. Journal of Wild- natural capital. Nature 387: 253–60. life Management 62: 821–41. Dietz, J.M., Dietz, L.A. & Nagagata, E.Y. (1994) The Beissinger, S.R., Reed, J.M., Wunderle, Jr, J.M., effective use of flagship species for conservation of Robinson, S.K. & Finch, D.M. (2000) Report of the biodiversity: the example of lion tamarins in Bra- AOU Conservation Committee on the Partners in zil. In Creative Conservation: Interactive Management of Flight species prioritization plan. Auk 117:549–61. Wild and Captive Animals (Eds P.J.S. Olney, G.M. Berger, J. (1997) Population constraints associated Mace & A.T.C. Feistner), pp. 32–49. Chapman & with the use of black rhinos as an umbrella species Hall, London. for desert herbivores. Conservation Biology 11:69–78. Downer, C. (1996) The mountain tapir, endangered Bond, I. (1994) Importance of elephant sport hunt- ‘flagship’ species of the high Andes. Oryx 30: 45–58. ing to CAMPFIRE revenue in Zimbabwe. TRAFFIC Ehrlich, P.R., Dobkin, D.S. & Wheye, D. (1992) Birds Bulletin 14: 117–19. in Jeopardy: the Imperiled and Extinct Birds of the Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 32 6.5.2006 11:07pm

32 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS

United States and Canada, including Hawaii and Puerto Union for the Conservation of Nature and Natural Rico. Stanford University Press, Stanford, CA. Resources, Gland, Switzerland. Entwistle, A. (2000) Flagships for the future? Oryx James, A.N., Gaston, K.J. & Balmford, A. (1999) 34: 239. Balancing the earth’s accounts. Nature 401: Ferrier, S., Pressey, R.L. & Barrett, T.W. (2000) 323–4. A new predictor of the irreplaceability of areas Jenkins, R.E. (1976) Maintenance of natural diver- for achieving a conservation goal, its application sity: approach and recommendations. In Transac- to real-world planning, and a research agenda tions of the 41st North American Wildlife and Natural for further refinement. Biological Conservation Resources Conference 21–25 March, Washington, DC 93: 303–25. (Ed. K. Sabol), pp. 441–51. Fleishman, E., Murphy, D.D. & Brussard, P.F. (2000) Jenkins, R.E. (1996) Natural Heritage Data Centre A new method for selection of umbrella species for Network: managing information for managing bio- conservation planning. Ecological Applications 10: diversity. In Biodiversity in Managed Landscapes: The- 569–79. ory and Practice (Eds R.C. Szaro & D.W. Johnston), Fleishman, E., Blair, R.B. & Murphy, D.D. (2001) pp. 176–92. Oxford University Press, New York. Empirical validation of a method for umbrella spe- Johnson, N.C. (1995) Biodiversity in the Balance: Ap- cies selection. Ecological Applications 11: 1489–501. proaches to Setting Geographic Conservation Priorities. Garson, J., Aggarwal, A. & Sarkar, S. (2002) Birds as Biodiversity Support Program, World Wildlife surrogates for biodiveristy: an analysis of a data set Fund, Washington, DC. from southern Que´bec. Journal of Biosciences Johnsingh, A.J.T. & Joshua, J. (1994) Conserving 27:347–60. Rajaji and Corbett National Parks: the elephant as Goldspink, C.R., Holland, R.K., Sweet, G. & Stjerns- a flagship species. Oryx 28: 135–40. redt, R. (1998) A note on the distribution of the Kershaw, M., Mace, G.M. & Williams, P.H. (1995) puku, Kobus vardoni, in Kasanka National Park, Threatened status, rarity and diversity as alterna- Zambia. African Journal of Ecology 36: 23–33. tive selection measures for protected areas: a test Goodwin, H.J. & Leader-Williams, N. (2000) Tourism using Afrotropical antelopes. Conservation Biology and protected areas: distorting conservation prior- 9: 324–34. ities towards charismatic megafauna? In Priorities Lambeck, R.J. (1997) Focal species: a multi-species for the Conservation of Mammalian Diversity: Has umbrella for nature conservation. Conservation the Panda had its Day? (Eds A. Entwistle & Biology 11: 849–56. N. Dunstone), pp. 257–75. Cambridge University Leader-Williams, N. & Dublin, H.T. (2000) Charis- Press, Cambridge. matic megafauna as ‘flagship species’. In Priorities Grumbine, R.E. (1994) What is ecosystem manage- for the Conservation of Mammalian Diversity: has the ment? Conservation Biology 8: 27–38. Panda had its Day? (Eds A. Entwistle & N. Dun- Hess, G.R. & King, T.J. (2002) Plannng for wildlife in stone), pp. 53–81. Cambridge University Press, a suburbanizing landscape I: selecting focal species Cambridge. using a Delphi survey approach. Landscape and Leader-Williams, N., Harrision, J. & Green, M.J.B. Urban Planning 58: 25–40. (1990) Designing protected areas to conserve nat- Heywood, V.H. (1995) Global Biodiversity and Assess- ural resources. Science Progress 74: 189–204. ment. Cambridge University Press, New York. Lindenmayer, D.B., Manning, A.D., Smith, P.L., et al. Hoekstra, J.M., Boucher, T.M., Ricketts, T.H. & (2002) The focal-species approach and landscape Roberts, C. (2005) Confronting a biome crisis: restoration: a critique. Conservation Biology global disparities of habitat loss and protection. 16: 338–45. Ecology Letters 8: 23–9. Lunney, D., Curtin, A., Ayers, D., Cogger, H.G. & Iltis, H.H. (1988) Serendipity in the exploration of Dickman, C.R. (1996) An ecological approach to biodiversity: what good are weedy tomatoes? In identifying the endangered fauna of New South Biodiversity (Ed. E.O. Wilson), pp. 98–105. National Wales. Pacific Conservation Biology 2: 212. Academy Press, Washington, DC. Mace, G.M. & Collar, N.J. (1995) Extinction risk IUCN (2001) IUCN Red List Categories: Version 3.1. assessment for birds through quantitative criteria. Species Survival Commission, International Ibis 137: 240–46. Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 33 6.5.2006 11:07pm

PRIORITIZING CHOICES IN CONSERVATION 33

Mace, G.M. & Lande, R. (1991) Assessing extinction Conservation Biology (Eds S. Ferson & M. Burgman), threats: toward a reevaluation of IUCN threatened pp. 291–305. Springer-Verlag, New York. species categories. Conservation Biology 5: 148–57 Possingham, H.P., Andelman, S.J., Noon, B.R., Margules, C.R. & Pressey, R.L. (2000) Systematic Trombulak, S. & Pulliam, H.R. (2001) Making smart conservation planning. Nature 405: 243–53. conservation decisions. In Research Priorities for Master, L.L. (1991) Assessing threats and setting con- Conservation Biology (Eds G. Orians & M. Soule´), servation priorities. Conservation Biology 5: 559–63. pp. 225–44. Island Press, Covelo, California. Master, L.L., Stein, B.A., Kutner, L.S. & Hammerson, Possingham, H.P., Andelman, S.J., Burgman, M.A., G. (2000) Vanishing assets: conservation status of Mendellin, R.A., Master, L.L. & Keith, D.A. (2002) US species. In Precious Heritage. The Status of Bio- Limits to the use of threatened species lists. Trends diversity in the United States (Eds B.A. Stein, L.S. in Ecology and Evolution 17: 501–7. Kutner & J.S. Adams), 93–118. Oxford University Possingham, H.P., Franklin, J., Wilson, K. & Regan, Press, New York. T.J. (2005) The roles of spatial heterogeneity and Meir, E., Andelman, S. & Possingham, H.P. (2004) ecological processes in conservation planning. In Does conservation planning matter in a dynamic Ecosystem Function in Heterogeneous Landscapes (Eds and uncertain world? Ecology Letters 7: 615–22. G.M.Lovett,C.G.Jones,M.G.Turner&K.C.Wea- Miller, B., Reading, R., Strittholt, J., et al. (1999) thers), pp. 389–406. Springer-Verlag, New York. Using focal species in the design of nature reserve Pressey, R.L. (1994) Ad hoc reservations: forward or networks. Wild Earth 8: 81–92. backward steps in developing representative re- Millsap, B.A., Gore, J.A., Runde, D.E. & Cerulean, serve systems? Conservation Biology 8: 662–8. S.I. (1990) Setting priorities for conservation of Pressey, R.L. & Taffs, K.H. (2001) Sampling of land fish and wildlife species in Florida. Wildlife types by protected areas: three measures of effect- Monographs 111: 1–57. iveness applied to western New South Wales. Bio- Mittermeier, R.A. (1986) Primate conservation prior- logical Conservation 101: 105–17. ities in the Neotropical region. In Primates: The Road Pressey, R.L., Ferrier, S., Hutchinson, C.D., Sivertsen, to Self-sustaining Populations (Ed. K. Benirschke), D.P. & Manion, G. (1995) Planning for negotiation: pp. 221–40. Springer-Verlag, New York. using an interactive geographic information system Morse, L.E. (1993) Standard and alternative taxo- to explore alternative protected area networks. In nomic data in the multi-institutional Natural Heri- Nature Conservation 4: the Role of Networks (Eds D.A. tage Data Center Network. In Designs for a Global Saunders, J.L. Craig & E.M. Mattiske), pp. 23–33. Plant Species Information System (Eds. F.A Bisby, G.F. Surrey Beatty and Sons, Sydney. Russell & R. Pankhurst), pp. 127–56. Oxford Rabinowitz, D. (1981) Seven forms of rarity. In The University Press, Oxford. Biological Aspects of Rare Plant Conservation (Ed. H. Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Synge), pp. 205–17. Wiley, Chichester. Fonseca, G.A.B. & Kent, J. (2000) Biodiversity Redford, K.H., Coppolillo, P., Sanderson, E.W., et al. hotspots for conservation priorities. Nature 403: (2003) Mapping the conservation landscape. Con- 853–8. servation Biology 17: 116–31 Noss R.F., Quigley, H.B., Hornocker, M.G., Merrill, T. Reed, J.M. (1992) A system for ranking conservation & Paquet, P.C. (1996) Conservation biology and priorities for neotropical migrant birds based on carnivore conservation in the Rocky Mountains. relative susceptibility to extinction. In Ecology and Conservation Biology 10: 949–63 Conservation of Neotropical Migrant Landbirds (Eds Olson, D.M. & Dinerstein, E. (1998) The global 200: a J.M.H. Hagan III & D.W. Johnston), pp. 524–36. representation approach to conserving the Earth’s Smithsonian Institution Press, Washington, DC. most biologically valuable ecoregions. Conservation Regan, T.J., L.L. Master & G.A. Hammerson. (2004) Biology 12: 502–15. Capturing expert knowledge for threatened species Paine, R.T. (1995) A conversation on refining the assessments: a case study using Natureserve con- concept of keystone species. Conservation Biology servation status ranks. Acta Oecologica 26: 95–107. 9: 962–64. Sanderson, E.W., Redford, K.H., Chetkiewicz, Possingham, H., Ball, I. & Andelman, S. (2000) C.L.B., et al. (2002a) Planning to save a species: Mathematical methods for identifying representa- the jaguar as a model. Conservation Biology 16: tive reserve networks. In Quantitative Methods for 58–72. Macdonald/Key Topics in Conservation Biology 1405122498_4_002 Final Proof page 34 6.5.2006 11:07pm

34 G.M. MACE, H.P. POSSINGHAM AND N. LEADER-WILLIAMS

Sanderson, E.W., Redford, K.H., Vedder, A., Ward, Walpole, M.J. & Leader-Williams, N. (2002) Eco- S.E. & Coppolillo, P.B. (2002b) A conceptual tourism and flagship species in conservation. model for conservation planning based on land- Biodiversity and Conservation 11: 543–547. scape species requirements. Landscape and Urban Watson, J., Freudenberger, D. & Paull, D. (2001) An Planning 58: 41–56. assessment of the focal species approach for con- Scott, J.M., Davis, F., Csuti, B., et al. (1993) Gap serving birds in varigated landscapes in southeast- analysis: a geographic approach to protection of ern Australia. Conservation Biology 15: 1364–73. biological diversity. Wildlife Monographs 123: 1–41. Western, D. (1987) Africa’s elephants and rhinos: flag- Shaffer, M.L. (1981) Minimum population sizes for ships in crisis. Trends in Ecology and Evolution 2: 343–6. species conservation. BioScience, 31, 131-134 Williams, P.H., Burgess, N.D. & Rahbek, C. (2000) Simberloff, D. (1998) Flagships, umbrellas, and key- Flagship species, ecological complementarity and stones: is single-species management passe’ in the conserving the diversity of mammals and birds in landscape era. Biological Conservation 83: 247–57. sub-Saharan Africa. Animal Conservation 3: 249–60. Smith, R.J., Muir, R.D.J., Walpole, M.J., Balmford, Wilson, K.A., Westphal, M.I., Possingham, H.P. & A. & Leader-Williams, N. (2003) Governance and Elith, J. (2005) Sensitivity of conservation plan- the loss of biodiversity. Nature 426: 67–70. ning to different approaches to using predicted Stein, B.A., Kutner, L.S. & Adams, J.S. (2000) species distribution data. Biological Conservation Precious Heritage. The Status of Biodiversity in the 122: 99–112. United States. Oxford University Press, New York. Wilson, K.A., McBride, M., Bode, M. & Possingham, Stotz, D.F., Fitzpatrick, J.W., Parker, III, T.A. & H.P. (In press) Prioritising the allocation of conser- Moskovits, D.K. (1996) Neotropical Birds: Ecology and vation resources between biodiversity hotspots. Conservation. University of Chicago Press, Chicago, IL Nature.

View publication stats METHOD FOR THE ASSESSMENT OF PRIORITIES FOR INTERNATIONAL SPECIES CONSERVATION

(MAPISCo)

Final Report

March 2013

Project Team

Matthew Grainger (Newcastle University)

Phil McGowan (Newcastle University) Ailsa McKenzie (Newcastle University) Jeroen Minderman (Newcastle University) Jon-Paul Rodriguez (IUCN-SSC) Alison Rosser (UNEP-WCMC)

Andy South (Freelance) Selina Stead (Newcastle University) Mark Whittingham (Newcastle University)

Steering Group

Vin Fleming (JNCC) Noel McGough (RBG Kew) Trevor Salmon (Defra) Michael Sigsworth (Defra) Andy Stott (Defra)

Dominic Whitmee (Defra)

MAPISCo Final report: Table of contents

Table of contents

1. EXECUTIVE SUMMARY ...... 4

2. INTRODUCTION ...... 9

2.1. Background ...... 9

2.2. Why is this method necessary? The link between policy demands and scientific capability ...... 10

2.3. Outline of the proposed project ...... 11

3. DEVELOPMENT OF THE METHOD ...... 12

3.1. Selection of co-benefits ...... 12

3.2. Focal co-benefits, data sets used and sources & category scoring...... 13 3.2.1. Habitat and area conservation (Aichi Targets 5 and 7) ...... 13 3.2.2. Sustainable harvesting (Aichi Target 6) ...... 15 3.2.3. Conservation of genetic diversity (Aichi Target 13) ...... 17 3.2.4. Protection of ecosystem services (Aichi Target 14) ...... 18 3.2.5. Preventing species extinction (Aichi Target 12) ...... 20

3.3. Database & prioritisation ...... 22 3.3.1. Database building ...... 22 3.3.2. Co-benefit weighting, re-scaling and priority score calculation ...... 22

3.4. Inclusion of red list threat classification data ...... 24

4. RESULTS - EXAMPLE PRIORITY LISTS ...... 27

4.1. Example 1: All species ...... 30 4.1.1. Summary findings ...... 30 4.1.2. Taxonomic composition ...... 30 4.1.3. Geographic composition ...... 30 4.1.4. IUCN threat categories and classifications ...... 30 4.1.5. Co-benefits ...... 31

4.2. Example 2: Taxonomic case study – birds ...... 36 4.2.1. Summary findings ...... 36 4.2.2. Orders and Families...... 36 4.2.3. Country/region ...... 36 4.2.4. IUCN Red List threat categories and classifications ...... 36 4.2.5. Co-benefits ...... 37 4.2.6. Key findings and how they relate to policy ...... 37

4.3. Example 3: Geographic case study – SE Asia ...... 44 4.3.1. General findings ...... 44 4.3.2. Taxonomic composition ...... 44 4.3.3. Geographic composition ...... 44 4.3.4. IUCN Red List threat categories and classifications ...... 44 4.3.5. Co-benefits ...... 45 4.3.6. Key findings and how they relate to policy ...... 45

5. USING THE METHOD ...... 50

2 MAPISCo Final report: Table of contents

5.1. Expandable –How does the does the priority list respond to the inclusion of additional co-benefit data? A plant example...... 50

5.2. Adaptable - How does the priority list respond to changes in co-benefit weightings? ...... 51 5.2.1. Sensitivity analysis ...... 51 5.2.2. Worked examples – threat status and ecosystem services ...... 51

5.3. Usable - Development of Graphic User Interface (GUI) ...... 54 5.3.1. GUI Development ...... 54 5.3.2. Constraints and legacy ...... 55 5.3.3. Future development options ...... 55 5.3.4. Future hosting options ...... 56

6. DISCUSSION ...... 57

6.1. Fit to original project brief ...... 57

6.2. How does the method compare with ‘business as usual’? ...... 58

6.3 How do co-benefits relate to IUCN threat status? ...... 58

6.4. Operating constraints ...... 59

6.5. Integration of MAPISCo into decision-making – next steps ...... 60 6.5.1. SCIENCE. Ensuring that the methodology fully accounts for scientific advances ...... 61 6.5.2. PRACTICAL - Maintenance of database and incorporation of additional data. Where will the database be housed? ...... 61 6.5.3. POLICY - Integrating MAPISCo into policy and resource allocation decisions ...... 63

6.6. Concluding remarks ...... 64

7. REFERENCES ...... 65

3 MAPISCo Final report: 1. Executive summary

1. Executive summary

Context. Biodiversity is declining at unprecedented levels globally, and meeting international targets aimed at halting these declines requires conservation efforts targeted not only at species but also at other aspects of biodiversity such as habitats, cultural values and ecosystem services. In spite of this wide range of targets requiring investment, resources available are declining. Against this backdrop, the UK Department for Environment, Food and Rural Affairs (Defra) sought, via this project, to develop a methodology for identifying species for which targeted conservation action would have the broadest consequential benefits (hereafter co-benefits) on other species, habitats, wider ecosystems, and ecosystem services.

Agreed scope and aims. The objective of the MAPISCo project was to develop a scoring method that enables species to be ranked based on their combined contribution to a selection of co- benefits linked to conservation targets (Aichi Targets). Concentration of conservation on these high-ranking species would, in theory, result in the largest associated biodiversity benefit. This methodology would be expandable, able to include further datasets should they become available, adaptable, with the weighting of co-benefits able to be altered in line with varying policy aspirations, and usable, ultimately able to be used by non-expert practitioners.

Selection of co-benefits. Five co-benefits were selected for inclusion in the methodology- (1) habitat and area conservation (Aichi Targets 5 and 7), sustainable harvesting of fish, invertebrates and aquatic plants (Aichi Target 6), (3) conservation of genetic diversity of wild relatives of cultivated plants and domesticated animals (Aichi Target 13), (4) protection of the provisioning of ecosystem services (Aichi Target 14) and (5) the prevention of species extinctions (Aichi Target 12). The selection of these co-benefits was based on the Aichi Targets of policy interest to Defra. They could also be linked in a scientifically defensible way with conservation effort on a species level AND have adequate data associated with them (from preliminary searches) to be related to species conservation lists, incurring as few taxonomic and geographic constraints as possible. However, it should be stressed that this methodological framework can be extended to include more co-benefits in the future, given, for example, specific policy needs (expandable).

Scoring. The methodology proposed here produces species lists ranked by their expected value in contributing to each of the five focal co-benefits under consideration. First, reliable data sources were identified that could be used to quantify the value of a given species to each of the five co- benefits. Species from these data sources were then added to a database and given a score for each of the five co-benefits. Details of how this was done are explained briefly in Table S1. The scores from each database were then combined to obtain an overall value that corresponds to species rank (see Box S1). Twelve data sources were found to be suitable for inclusion in the database at this stage. Many more were identified but later disregarded because of issues with data coverage and compatibility. The weighting, or importance, of each co-benefit can be adjusted in response to policy aspirations – this makes the methodology highly adaptable.

Results – example priority lists. The results generated by the database in its current format are constrained by data availability (e.g. only around 3% of all plant species have been categorised on the IUCN Red List while almost all bird species are included). For this reason in this section we present three different sets of results 1. All species, 2.Birds only (taxonomic case study) and 3. SE Asia only (geographic case study). The use of case studies allows us to focus on discrete sets of data, which, while not eliminating the constraints completely, allows a more meaningful

4 MAPISCo Final report: 1. Executive summary demonstration of how the database can be used. For each set of lists we have outlined the main findings from the method followed by a discussion of how some of the key findings could be related to policy actions.

Generally, the results indicate that “politically interesting” or flagship species often championed by interest groups do not generally rank highly (e.g. Polar Bear Ursus maritimus 45625th and White Rhinoceros Ceratotherium simum 51510th), because they are associated with only a small number of the co-benefits considered here. There are 1064 species which occur in the top 500 regardless of changes in the co-benefit weightings, these include 502 birds, 161 mammals, 158 amphibians, 151 fish, 54 plants, 20 reptiles and 18 “other” species. This reflects the need for more data on plants, reptiles and invertebrates. Both the Habitat and Ecosystem Services co-benefits are significantly negatively correlated to Threat Status, meaning that more traditional approaches to conservation (based on extinction risk- the IUCN Red List) do not capture more recent concerns about protecting a range of co-benefits from each species. The MAPISCo methodology successfully prioritises both extinction risk and contribution to co-benefits.

Using the method. Expandable. We demonstrate how additional species or co-benefit data can be added to the database, and outline how such changes impact on the ranking of priority lists. Adaptable. We examine the effect changing individual co-benefit weightings (i.e. making certain co-benefits “more important” in the calculation of priority lists than others) has on priority list ranking. Usable. Here we outline the development of a web-based interface, which, using a variety of tabs and graphics, allows users to fully explore the priority lists created by the methodology under a number of different scenarios. Importantly, it also enables user to investigate how varying individual co-benefit weightings impact upon rankings. We view this as a critical feature of the Graphical User Interface (GUI), as it makes it extremely adaptable to policy aspirations. Further investment in this project could see this tool becoming available (open source) to interested parties

Discussion and project legacy. We conclude that we have delivered a methodology that can prioritise species for conservation based on their expected contributions to a selection of co- benefits. Thus, higher-ranking species should make greater contributions to meeting relevant Aichi Targets (5-8 and 12-14). This methodology is expandable – additional datasets can be added to it should they become available, adaptable – co-benefit weightings can be altered to fit with individual policy aims and usable – the development of a graphic user interface will allow non- technical users to use the method.

The original ambitious conceptual development of MAPISCo was rooted in the desire to embed science firmly in international species policy. The core issue was that biodiversity spending can tend towards projects focussed on charismatic animals with little evidence scientific justification for such action. The method we present here yields priority lists based on available scientific evidence but there are major caveats. The most important is the paucity of data available for some taxa (especially plants). Whilst our analysis based on a well-known taxon (birds) for which all species are assessed on the Red List does yield potentially usable results other prioritisation results based on combining taxa are inevitably strongly constrained by data availability.

As a consequence of this we therefore outline a road-map to overcoming the challenges of linking science and policy effectively in biodiversity governance in a way that will help ensure that MAPISCo strengthens the UK’s ability to maximise the wider value to biodiversity of its spend on international species conservation.

5 MAPISCo Final report: 1. Executive summary

Conclusion. We have developed a methodology which provides a broad-brush mechanism for identifying species conservation priorities based on a selection of co-benefits. These co-benefits are based on currently available and accessible data that are accepted to be good quality and have the potential to be expanded as new data emerges. The project has demonstrated that the choice of co-benefits, the importance given to them and the data sources used has a strong effect on which species are identified as being higher priorities. Therefore, explicit policy decisions are required (and need to be documented) throughout the prioritisation process. This finding alone is a significant contribution to increasing engagement at the science-policy interface, because it shows how closely intertwined the two spheres are. This feature of MAPISCo is likely to make it more policy relevant than other prioritisation processes which are less sensitive to the practicalities of policy-making. Implied in the original project brief is an assumption of a relatively straightforward and linear science-policy interface, where policy asks a question, science answers it and then policy decides what action should be taken. In practice, while this assumption has proved broadly accurate, this must go hand in hand with meaningful dialogue between policy-makers and scientists so that the best information available is used to inform policy as soundly as possible. There is clear scope for Defra to build on the progress made in this project to allow scientific knowledge and practice to better support UK government objectives. Overall, there is significant potential for the methodology we have developed to become part of an iterative process where conservation science and policy continually inform each other to produce evidence-based scientific policy that is more relevant to society.

6 MAPISCo Final report: 1. Executive summary

Table S1. The data sources and scoring format used for each of the five co-benefits currently included in the methodology. With further development we envisage being able to include a greater number of data sources.

Which data source do scores come How were species scored? * Co-Benefit from?

1. Habitat/area 1) Important Bird Area (IBA) 1) Mean (average) number of co - conservation occurring species of conservation

concern (e.g. at high risk of extinction) in all of the IBA’s in which a species occurs. 2) Alliance for Zero Extinction (AZE) 2) Total number of species of conservation concern co-occurring with the target species in an AZE

2. Sustainable 3) “FishBase” data on commercial 3) 1-6 (1=no interest, 6=highly harvesting value in fisheries commercial 4) IUCN Red List listed as affected 4) 1-3 (1=unknown, 3=industrial) by aquaculture 5) “FishBase” for species used in 5) 1 or 0 (1 if listed, 0 if not) aquaculture 3. Conservation 6) Database of crop wild relatives 6) 1 or 0 (1 if listed, 0 if not) of genetic 7) Lists of wild relatives of 7) 1 or 0 (1 if listed, 0 if not) diversity domesticated animals 8) Plants listed as of medicinal use 8) 1-3 (least to most use) 4. Protection of 9) Carbon loss through 9) Estimate of loss of carbon through ecosystem deforestation (country-level) deforestation (tonnes/year) services 10) Freshwater availability (country- 10) Availability of freshwater per level) capita per year (m3/capita/year). 5. The 11) IUCN Red List (for animals) 11) & 12) 1-9 (1= extinct, 2= least prevention of 12) SRLI (for plants) concern, 9=critically endangered) extinctions

*These scores were rescaled to between 0 and 1 for the final ranking process and then standardized to give equal weighting between scores.

7 MAPISCo Final report: 1. Executive summary

Box S1: A worked example of the final priority score calculation

The final priority score for a species is the sum of the scores given for the five co-benefits. The method for calculating co-benefits scores is outlined below.

1) Mean taken of the scores assigned 2) The mean score calculated in step 1 is then standardised by taking away from original individual datasets (in this from it the mean of all the values in the entire co-benefit column, then example, two different datasets). dividing it by the standard deviation of that co-benefit mean (calculation of a “z score”). The resultant score will be positive if the individual species score is greater than the mean score and negative if the individual species score is smaller than the mean score.

Species Habitat Harvesting Genetic diversity Ecosystem Threat status Final Score Service Provisioning Francolinus 2) The mean score calculated in 1) is then standardised by taking away from it camerunensis (0.136+0.789)/2 (0+0+0)/3the mean(0+0+3)/3= of all the values (0.323+0.975)/2 in the entire co-benefitmax(0.778,0) column, then dividing it by =0.462 = 0 the standard0.333 deviation of that= co0.649-benefit mean (calculation= 0.778 of a “z score”). The resultant score will be positive if the individual species score((5.05)+( is greater-2.58)+ 0.462-0.07 0-0.25 than the0.333 mean- 0.24 score 0.11 and negative0.462- 0.55if the individual 0.462 species-0.45 score is(0.82)+(0.77) smaller 0.08 0.10 than the mean=0.82 score. 0.11 0.25 +(1.35)= = 5.05 =-2.58 =0.77 =1.35 5.42 5.05*1 -2.58*1 0.82*1 0.77*1 1.35*1

3) The new co-benefit score is then multiplied by a 4) Final score calculated by adding together the 5 co- weighting factor (in this case all co-benefits are benefit scores. This score is then used to rank species weighted equally (i.e. weighting set to 1). in the priority list.

8 MAPISCo Final report: 2. Introduction

2. Introduction

Capsule.

 With biodiversity declining at unprecedented levels globally, the choice of where and how to

invest biodiversity conservation effort is becoming increasingly difficult.  This project seeks to develop a methodology by which species can be prioritised for conservation based not only on individual species benefits, but also on the contribution their conservation may make to other species, habitats, wider ecosystems, and ecosystem services.  This project aims to help bridge the gap between the contrasting spheres of science and policy.

2.1. Background

Biodiversity is declining at unprecedented levels globally: rates of species extinctions are increasing while natural habitats are declining. This is largely as a result of anthropogenic pressures (Butchart et al. 2010; Hoffmann et al. 2010). As a consequence, negative impacts on humans accrue, not only through intrinsic loss of wildlife but also as a result of declines in and loss of the ecosystem services healthy natural systems underpin and provide (Millennium Ecosystem Assessment 2005; Cardinale et al. 2012). The choice of where and how to invest biodiversity conservation effort is becoming increasingly difficult as available resources are shrinking and the number of targets to which to contribute is growing. Against this backdrop, the UK Department for Environment, Food and Rural Affairs (Defra) seeks through this project to develop a scientifically robust and repeatable method to identify species for which targeted conservation action by the UK Government would have the broadest consequential benefits (hereafter termed co- benefits) for other species (or taxa), habitats, wider ecosystems, and ecosystem services. Key conservation action aimed at such species will maximise contributions to international species conservation treaties such as the Convention on Biological Diversity’s (CBD) Strategic Plan for Biodiversity 2011-2020, which established twenty international targets to safeguard global biodiversity. These are known as the Aichi Targets (COP 10 Decision X/2, see http://www.cbd.int/decision/cop/?id=12268), and aim to safeguard biodiversity in its broadest sense and at different levels. The targets of interest include preventing extinctions, conserving habitats, controlling invasive species, sustainable harvesting, and protection of ecosystem services.

As described in the original project brief (see Appendix 1), this method would involve the development of a scoring system where individual species are linked, via existing data sources (e.g. IUCN Red List, FishBase), to their expected contribution to various co-benefits (such as ecosystem service provision or genetic relatedness to domesticated plants and animals). Species recorded in the FishBase database, for example, as being important food sources would receive a high score for a “sustainable harvesting” co-benefit, whereas a species recorded as having little or no importance in harvesting would receive a low, or zero, score. This would enable individual species to be ranked within an overall priority list based on their contribution to all the co-benefits added together (the score for each co-benefit summed). Conservation action aimed at species ranked at the top of this list would, therefore, be expected to have their greatest co-

9 MAPISCo Final report: 2. Introduction sequential benefits to the environment (based on the co-benefits selected for inclusion in the method).

This methodology should be 1) expandable allowing the incorporation of future data, 2) adaptable to changing policy aims and 3) usable by non-technical practitioners. This methodology would then be available to and usable by a range of practitioners, and be adaptable to a wide range of policy and conservation goals.

2.2. Why is this method necessary? The link between policy demands and scientific capability

One key goal of this project is to link policy goals (i.e. the addressing of Aichi Targets) to real conservation actions via sound scientific method. The fulfilment of this goal requires a smooth transition from policy to science and back to policy - that scientifically robust findings can be used to explore policy aspirations, and that the resulting policy is based on sound science and clear decisions (Figure 1).

However, as the science and policy “spheres” tend to have very different rationales, time-lines and objectives, the transition between them is often far from straightforward. Koetz et al. (2008) synthesise several authors in arriving at their view of the issues at the heart of the science-policy interface. They suggest that while science objectively deals with the generation of knowledge, policy tends to be concerned with making subjective choices between different arguments, often tackling interests and values that ultimately conflict (see also Appendix 5. Rapid Assessment Report to support development of a Methodology for the Assessment of Priorities for International Species Conservation, a report commissioned by this project, subcontracted to UNEP-WCMC).

This project aims to help bridge the gap between science and policy spheres by taking an integrated approach. While the inputs of the methodology will be policy-driven (i.e. the selection of co-benefits to which individual species conservation will be related and how these co-benefits are individually weighted), the methodology used to address these policy questions will be based on the best available scientific evidence. The development of a usable “front end” for this methodology should enable the scientific finding to be used by policy makers and applied directly to the policy sphere.

1) Scope defined by 2) Scientific 3) Resources policy aspirations knowledge and data available and availability used to deployed by policy (e.g. Aichi Targets) identify 'priorities' makers

Figure 1. Classic view of the science-policy interface

10 MAPISCo Final report: 2. Introduction

2.3. Outline of the proposed project

The project brief underwent considerable development in the early phase of this project (see Appendix 1 and 2 for full details of the original brief and how it was amended). The final agreed aims of the project are as follows:

 To develop a scoring method which enables species to be ranked based on their contribution to a selection of conservation targets (or co-benefits), that would be expandable allowing the incorporation of future data, adaptable to changing policy aims and usable by non-technical practitioners.

 To test the results and usability of this methodology using case studies (taxonomic and geographic), testing the expandability and adaptability of the database with the addition of extra data sources and changes to co-benefit weighting.

 To develop a web-based tool (a graphical user interface GUI) so that the methodology can be demonstrated to and used by non-technical practitioners - usability.

 To consider the broader science-policy context in which MAPISCo sits and propose how it may become fully integrated in the future (the project legacy).

11 MAPISCo Final report: 3. Development of the method

3. Development of the method

Capsule.

 Co-benefits selected (by the steering group) for inclusion in the method: (1) Habitat and area conservation, (2) Sustainable harvesting of fish, invertebrates and aquatic plants, (3) Conservation of genetic diversity of wild relatives of cultivated plants and domesticated

animals, (4) Protection of the provisioning of ecosystem services, and (5) Prevention of

species extinctions.  These co-benefits can be changed or added to - this makes the method expandable.  Scoring method developed which produces lists in which species are ranked by their

expected value in contributing to each of the five co-benefits above.

 The weighting, or importance, of each co-benefit can be adjusted in response to policy. aspirations. This makes the method adaptable.

3.1. Selection of co-benefits

The original aim of this project was to link the conservation of individual species to a large suite of ecosystem co-benefits that could be related directly to the relevant Aichi Targets. The targets of interest included preventing extinctions, conserving habitats, controlling invasive species, sustainable harvesting, and protection of ecosystem services. However, preliminary work indicated that for many species groups, sufficient data linking their conservation to many of the suggested co-benefits are either not available or not easily accessible. Moreover, the full range of co-benefits set out in the original brief was likely too large for the project timeframe.

For these reasons, a subset of five co-benefits was selected for inclusion in the development of the methodology (see Box 1). This selection was made based on the contribution these co-benefits made to Aichi Targets of policy interest to Defra, their links with conservation effort on a species level and adequate data being available (from preliminary searches) to link them to species conservation lists.

However, it should be stressed that further co-benefits could be included in the future to incorporate specific policy needs (expandable).

Box 1. The five co-benefits selected for inclusion in the development of the methodology, and the Aichi Targets to which they contribute

1. Habitat and area conservation (Targets 5 and 7; hereafter termed “Habitat co-benefit”) 2. Sustainable harvesting of fish, invertebrates and aquatic plants (Target 6; hereafter “Harvesting co-benefit”) 3. Conservation of genetic diversity, in particular of wild relatives of cultivated plants and domesticated animals (Target 13; hereafter “Genetic Diversity co-benefit”) 4. Conservation of the provisioning of ecosystem services (Target 14; hereafter “ES co-benefit”) 5. Prevention of species extinctions (Target 12; hereafter “species extinction co-benefit”).

12 MAPISCo Final report: 3. Development of the method

3.2. Focal co-benefits, data sets used and sources & category scoring

In the following sections, for each of the five co-benefits selected for inclusion we discuss (i) the rationale for links to conservation effort on a species level, (ii) data sets chosen to make this link and (iii) quantitative scoring used to integrate each data source in the prioritisation methodology.

A note on data sources The majority of data sources were identified through discussions with experts at the project workshop (see Appendix 4). Many suggested data sources were unsuitable for use in the final methodology due to taxonomic or geographic biases in datasets, or general data accessibility issues. For example, for the habitat co-benefit, Biodiversity Hotpots (Myers et al. 2000) and the Global “200” Ecoregions project data (Olsen & Dinerstein 2002) could not be used as species associations made in the lists were taxonomically biased and the data were not easily available. For the harvest co-benefit, the Seas Around Us project (www.seaaroundus.org) was also rejected due to the data not being publicly accessible. For the genetic diversity co-benefit, the SEPASAL database (www.kew.org/ceb/sepasal) was rejected as data were both taxonomically and regionally biased as well as being difficult to access. Further examples of investigated but unsuitable databases are listed in Appendix 8, Table A8-1.

3.2.1. Habitat and area conservation (Aichi Targets 5 and 7)

Rationale

Target 5: “By 2020, the rate of loss of all natural habitats, including forests, is at least halved and where feasible brought close to zero, and degradation and fragmentation is significantly reduced.”

Target 7: “By 2020 areas under agriculture, aquaculture and forestry are managed sustainably, ensuring conservation of biodiversity.”

Both Aichi Target 5 and 7 relate to the conservation or sustainable management of natural and semi-natural habitats. Although conservation effort directed at species usually involves a degree of protection of the habitat(s) (Mace & Collar 2002), such contribution to habitat conservation is likely to be greater for some species than for others.

One way to link species-level conservation to habitat conservation is to focus on those species that are thought to be of disproportionate importance to their habitats or co-occurring species (i.e. “surrogate species” such as umbrella, keystone or indicator species; (Caro & O’Doherty 1999; Caro & Girling 2010). However, the effectiveness of surrogate species for conservation is widely criticised (e.g. Lindenmayer et al. 2002; Saetersdal & Gjerde 2011). More importantly, concrete evidence for the effectiveness of species as surrogates for habitats is limited and often highly context-dependent (Andelman & Fagan 2000), which means globally applicable lists of appropriate habitat surrogate species are not available.

Instead, in the current project we have chosen to link species to habitats by focussing on species that have previously been associated with, or used as “triggers” for the designation of Key Biodiversity Areas (KBAs) (Eken et al. 2004). Defined as “sites of global significance for biodiversity conservation “(Eken et al. 2004), KBAs are designated based on the conservation of habitat within them being important or even vital for the persistence of one or more

13 MAPISCo Final report: 3. Development of the method target species (Eken et al. 2004). However, conservation effort directed at these target species is likely to benefit the wider habitat, making KBAs important areas for a wide and varied number of species. There is also a general consensus that the conservation of KBA sites has many wider benefits (e.g. in terms of cultural value or provisioning of ES (e.g. Butchart et al. 2012; Larsen, Turner & Brooks 2012).

Theoretically, conservation effort directed at a species recorded in a KBA will also benefit non- target species co-occurring in that KBA, as well as having other localised benefits (e.g. ecosystem service provision). Therefore, conservation effort directed at species recorded in species-rich KBAs is likely to produce higher levels of habitat co-benefits than conservation of a species that occurs in a species poor KBA.

A “habitat score” for each individual species is calculated as follows: for each species associated with a KBA, we use the mean number of other species known to co-occur in all KBAs in which that individual species occurs. This is, in effect, a proxy measure of the expected contribution a given species may make to habitat conservation overall.

Data sources

The linking of species and habitats necessary in this approach relies upon the availability of species inventories for individual KBAs. However, these inventories are not always available for all types of KBA, or for all species groups. For this reason we have decided to include data from the two types of KBA for which most data is readily available: Important Bird Areas and Alliance for Zero Extinction sites.

 Important Bird Areas (IBAs) are sites identified as being globally important for the persistence of one or more populations of endangered bird species. Identified using standardised criteria (http://www.birdlife.org/datazone/info/bacritglob [Accessed 22 August 2012]), a site qualifies for IBA status if it holds or is thought to hold significant populations (or parts of populations) of bird species (1) listed as endangered (Critically Endangered, Endangered or Vulnerable) on the IUCN Red List, (2) with a restricted range (e.g. endemics), (3) that are (largely) restricted to a single biome, or (4) that are migratory or congregatory and for which the site is important during particular parts of the year (BirdLife International 2008, http://www.birdlife.org/datazone/info/ibacriteria [Accessed 22 August 2012]). To date, over 10,000 IBAs have been identified globally (http://www.birdlife.org/datazone/site [Accessed 22 August 2012]), in which 4847 species are listed to occur. The number of bird species listed per IBA site ranges from 1 to 247, with an average of 9 per site.

 Alliance for Zero Extinction (AZE) sites are sites which hold the last remaining population(s) of highly threatened species of mammals, birds and/or selected reptiles, amphibians and conifers. To qualify as an AZE site, a site must (1) hold at least one species listed as Critically Endangered or Endangered on the IUCN Red List, (2) hold all or the majority (>95%) of the known population of the species, and (3) must be geographically and politically discrete (Ricketts et al. 2005). To date, 587 AZE sites containing 920 species have been identified. The number of species listed per AZE site ranges from 1 to 22, with an average 1.6 species per site.

14 MAPISCo Final report: 3. Development of the method

Species-level score

All IBA and AZE inventories were sourced and the species included on them listed in a database in preparation for the assignment of a score.

For species listed as occurring in an IBA (or listed as a “trigger” species used to delineate an IBA), individual species scores were equal to the mean number of species recorded as co- occurring with that individual species across all IBAs for which there was a record. For example, if species A was recorded on three IBA inventories, and co-occurred with five species on one inventory, ten on another and twenty on the third, the score species A received would be mean of 5, 10 and 20 = 11.67. This resulted in each species receiving a value between 1 and 246, with a mean of 34.3.

Species listed as occurring in an AZE were attributed a score equal to the total number of species co-occurring with that species at the site in which it was recorded (not an average because, by definition, a given species occurs in only a single AZE site). This resulted in a score ranging from 1 to 22 (22 being the maximum number of species listed for one site) with a mean of 3.5. Species listed as occurring in both an IBA and an AZE were given the mean score. Those not listed as occurring in either an IBA or an AZE site were not allocated any score.1

For both scores, higher values indicate that, on average, a species occurs in AZE or IBA sites that hold larger numbers of other species. Conservation effort directed at species with such higher scores is therefore likely to both benefit their wider habitat as well as a larger number of co- occurring species.

3.2.2. Sustainable harvesting (Aichi Target 6)

Rationale

“By 2020 all fish and invertebrate stocks and aquatic plants are managed and harvested sustainably, legally and applying ecosystem based approaches, so that overfishing is avoided, recovery plans and measures are in place for all depleted species, fisheries have no significant adverse impacts on threatened species and vulnerable ecosystems and the impacts of fisheries on stocks, species and ecosystems are within safe ecological limits.”

It was assumed (in discussion and agreement with the steering group) that conservation effort directed at harvested fish species or species involved in aquaculture production would contribute to this target. Moreover, we assumed that such contributions would be stronger for species that are seen as having greater economic value.

Data sources

Thus, we used three data sources to identify species relevant to this target: 1) Commercial value of a species to fisheries. FishBase (Froese & Pauly 2012) provides a qualitative assessment of the economic value of 3111 fish species across countries in which they are harvested. The overall economic value or importance of each species is categorised in one of six categories, ranging from “no interest” to “highly commercial”. The definitions of the categories are given in Table 1.

15 MAPISCo Final report: 3. Development of the method

2) Species listed as affected by aquaculture on the IUCN Red List (IUCN 2012). As well as conservation status and extinction risk, the IUCN Red List (2012.1) records types of threats faced by species. One of these threat classifications is aquaculture. We identified 269 species listed under Threat Classification (v.3.1) 2.4 (Marine & freshwater aquaculture), which distinguishes between species impacted by “Industrial” (Classification 2.4.2), “Subsistence/artisanal” (Classification 2.4.1.) and “Unknown” levels of aquaculture (Classification 2.4.3).

3) Species listed as used in aquaculture production on FishBase (204 species, C. McOwen pers. comm.).

Table 1. FishBase commercial harvesting categories and scores attributed.

Category Category No. spp. % spp. score Definition Highly commercial 207 6.7 6 The species is very important to the capture fisheries (or aquaculture) of a country Commercial 1416 45.5 5 The species is regularly taken in the capture fisheries or regularly found in aquaculture activities of a country Minor commercial 1233 39.6 4 The species is of comparatively less importance in capture fisheries or aquaculture in a given country Subsistence fisheries 210 6.8 3 The species is consumed locally only, mostly by the fishers themselves Of potential interest 2 0.1 2 Of no interest 43 1.4 1

Species-level scores

All species were attributed a score for each of the three data sources listed above. First, species listed as commercially harvested in FishBase were attributed a score between 1 (for “no interest”) to 6 (for “highly commercial”), reflecting increasing conservation priority for species of greater economic interest (Table 1). Second, species listed as threatened by aquaculture on the IUCN Red List were attributed a score between 1 (for “unknown scale”) and 3 (for “industrial”) reflecting the increasing intensity of aquaculture threat and therefore the increasing potential for conservation effort directed at such species to benefit the target in question (Table 2). Third, species listed as used in aquaculture production on FishBase were attributed a score of 1, whereas other species were not attributed any value.

16 MAPISCo Final report: 3. Development of the method

Table 2. Species and classifications listed under Threat Classification (v.3.1) 2.4 (Marine & freshwater aquaculture) on the IUCN Red List (2012.1), and category scores attributed.

Category Category No. spp. % of spp. score Industrial 27 10 3 Subsistence/artisanal 16 5.9 2 Scale unknown 226 84 1

3.2.3. Conservation of genetic diversity (Aichi Target 13)

Rationale

“By 2020, the genetic diversity of cultivated plants and farmed and domesticated animals and of wild relatives, including other socio-economically as well as culturally valuable species, is maintained, and strategies have been developed and implemented for minimising genetic erosion and safeguarding their genetic diversity.”

It was assumed (in discussion and agreement with the steering group) that conservation effort directed at species extant in the wild will contribute to the preservation of (unique) genetic diversity of the targeted species. Following the central tenet of this target, we chose to focus on wild relatives of crops and domestic animals, and further expanded our consideration into plant species with a known medicinal use.

Data sources

We used the following data sources to identify species relevant to this target:

1) A database of wild relatives of plant crop species. This database (Vincent et al. in prep.) lists 1385 high priority crop wild relative (CWR) species. CWR are wild species closely related to crop species which have the potential to contribute valuable traits (e.g. disease resistance) to crops in the future. Vincent et al. define CWR as those species which are sufficiently similar genetically to allow crossing (either naturally or in the laboratory; the “gene pool concept”) or in some cases those species belonging to the same genus (the “taxon concept”).

2) Lists of wild relatives of domesticated animal species, compiled from the FAO World Watch List for Domestic Animal Diversity (FAO 2000) and from (McGowan 2010). The former document identified avian and mammal species representing domestic animal genetic resources at risk of loss, based on a range of survey- and monitoring- efforts. From these, we identified those species extant in the wild and/or listed as at risk from hybridisation on the IUCN Red List (IUCN 2012, in total 210 species). Among birds, Galliformes are particularly important economically and we therefore added species from family and genera identified in McGowan 2010 and listed on the IUCN Red List (IUCN 2012) as relatives of domesticated animals (in total 323 species). It should be noted that in light of a recent review (Owens & McGowan in prep), neither Cracidae (chachalacas, guans and curassows

17 MAPISCo Final report: 3. Development of the method

from Central and South America) nor Megapodidae (mound-builders from the Australasian region) receive scores for this co-benefit. This is because no successful hybrids between species from these families and domesticated poultry have been recorded (McCarthy 2006).

3) Plants species listed as used for medicinal purposes in the BGCI PlantSearch database (http://www.bgci.org/plant_search.php/ [Accessed 22 August 2012]) and Hawkins (2008). The BGCI PlantSearch database is compiled from information supplied by botanical gardens worldwide, including whether a given species has a medicinal use. The database holds 1788 species records listed as having medicinal use. Hawkins (2008) compiled information on many medicinal plant species from a range of sources and expert opinion questionnaires, and lists 429 medicinal plant priorities.

Species-level scores

All species were attributed a numerical score for each of the data sources outlined above. First, reflecting their status as CWR, species occurring in the CWR database were attributed a score of 1. Second, species occurring on our compiled list of relatives of domesticated animals were also attributed a score of one. Species not on either list were not attributed any score. The resulting binary scores reflect our limited ability using these data sources to distinguish further between the relevant species in terms of priority (e.g. a species either is a CWR or not). Third, species listed in the top 35 of Annex 5 of Hawkins (2008) were attributed a score of 3, remaining species in the same list were scored as 2 and other species with a known medical use (listed in the PlantSearch database) were scored as 1.

3.2.4. Protection of ecosystem services (Aichi Target 14)

Rationale

“By 2020, ecosystems that provide essential services, including services related to water, and contribute to health, livelihoods and well-being, are restored and safeguarded, taking into account the needs of women, indigenous and local communities, and the poor and vulnerable.”

To maximise contributions to the protection of ecosystem services (ES) by conservation effort directed at a species level, evidence is required that shows the relative importance of species to the provisioning of ES. While it is widely recognised that ecosystems provide a range of services and benefits to humans, and indeed that groups of species can be associated with broad service provision (Millenium Ecosystem Assessment 2005; UK NEA 2011), the evidence base linking individual species to particular services is limited and evidence showing the relative value of different species for a given service even more so. Where such evidence is available, it is often limited to a single species in a particular context (e.g. Vira & Adams 2009; MAPISCo Project Team 2012). Such context-dependent examples do not constitute the solid evidence base necessary for the taxonomic and geographic scope required for the present methodology. Moreover, examples of broad species groups providing essential services are prevalent and often cited. However, in such cases, often large numbers of species are involved, and their relative importance to the provision of the service in different contexts is unclear. For example, although it is well known that many insects are vital as pollinators of economically important crop species, the value of individual pollinating species has, in the vast majority of cases, not been quantified. This inability to

18 MAPISCo Final report: 3. Development of the method distinguish between such species in terms of their relative value to the service in question limits the use of such data in a species prioritisation methodology.

By contrast, there is a growing consensus that priorities for biodiversity conservation and for ES can be reconciled using area-based (as opposed to species-based) approaches, for example freshwater provisioning or carbon sequestration along with levels of biodiversity (Lamoreux et al. 2005; Goldman et al. 2008; Naidoo et al. 2008; Egoh et al. 2011; Fisher et al. 2011; Butchart et al. 2012). Thus, here we chose to make the link between species and the provisioning of a selection of ES by focusing on a larger, habitat- and country scale. The advantage of focusing on a country level is that broad measures of the provisioning of some ES are available at country level, and species occurrence data within broad habitats in countries is more readily available (and potentially more reliable) than finer-scale measures of distribution.

Data sources

We focus on two example ES – 1) estimated carbon loss through deforestation and 2) freshwater availability. We used the following data sets to link country-level measures of these two services to species level:

1) Carbon loss through deforestation (tonnes/year). To estimate this, we multiplied country- level estimates of (1) the stock of carbon in living forest biomass in 2010 (tonnes/hectare) with (2) the trend of the extent of primary forest between 2005-2010 (change in hectares), as available in the FAO Global Forest Resources Assessment 2010 (FAO 2010) database (http://countrystat.org/index.asp?ctry=for&HomeFor=for [Accessed 22 August 2012]).

2) Freshwater availability. As a measure of the availability of freshwater to people, we used the country-level estimated total renewable per capita freshwater supply in 2010, as obtained from the FAO AQUASTAT database (http://www.fao.org/nr/water/aquastat/data/query/index.html [Accessed 22 August 2012]). Lower values (<1000 m3/capita) indicate water scarcity (UNEP Vital Water Graphics: www.unep.org/geo/geo4/report/Glossary.pdf [Accessed 22 August 2012]).

3) Habitat- and country- occurrence. We used the data in the IUCN Red List (IUCN 2012) augmented by the Sampled Red List Index (SRLI) for plants (http://threatenedplants.myspecies.info/ [Accessed 22 August 2012], and S. Bachman pers. comm.) (to increase representation of plant species) to identify species occurrence in countries, and in (1) forest habitats (Habitat Classification 1) and wetland (inland) habitats (Habitat Classification 5). See section 3.2.5 (page 20) for more information on the data used from the Red List and SRLI for plants.

Species-level scores

We assumed that conservation effort directed at forest species occurring in countries with higher estimated rates of carbon loss through deforestation is more likely to make contributions to targets to reduce carbon loss or increase carbon sequestration. Similarly, because wetland and forest habitats are particularly important in controlling both the supply and quality of freshwater (Millenium Ecosystem Assessment 2005; Larsen, Turner, & Brooks 2012), we assumed that conservation effort directed at forest or wetland species occurring in countries with lower levels of per capita

19 MAPISCo Final report: 3. Development of the method freshwater supply is likely to make greater contributions to targets aiming to alleviate water scarcity or stress.

Accordingly, species occurring in forest habitats (IUCN Red List Habitat Classification 1) were attributed a score for the estimated carbon loss through deforestation, calculated as the average estimated carbon loss through deforestation across all countries in which the species occurs. Lower values indicate an association with higher rates of loss and therefore higher conservation priority. Similarly, species occurring in forest- or wetland habitats (IUCN Red List Habitat Classifications 1 and 5) were attributed a score for freshwater supply calculated as the average estimated supply across countries in which it occurs. Lower values indicate a greater association with higher levels of water scarcity, and therefore higher conservation priority.

3.2.5. Preventing species extinction (Aichi Target 12)

Rationale

“By 2020 the extinction of known threatened species has been prevented and their conservation status, particularly of those most in decline, has been improved and sustained.”

As the present prioritisation methodology aims to maximise co-benefits of the conservation of species, we considered Aichi Target 12 to be our “focal” target and assumed that conservation effort directed at more highly threatened species would contribute most to it.

Since revisions from the IUCN Red Data Books (Mace & Lande 1991), the IUCN Red List has grown to become not only the most comprehensive data set on the extinction risk of a wide range of species from various taxonomic groups, but also represents an effective data source for species occurrence and habitat classifications (IUCN 2012). Red List threat status assessments are made by experts according to well-documented standards (IUCN 2001) (http://www.iucnredlist.org/technical-documents/categories-and-criteria/2001-categories-criteria [Accessed 22 August 2012]). Species are placed into one of nine threat categories representing increasing extinction risk, based on a range of criteria including (1) declines in population size, (2) restrictions in geographic range, (3) small absolute population size or (4) analytical evidence of high extinction risk.

Data sources and species-level scores

For each species listed on the IUCN Red List v. 2012.1 (IUCN 2012) and/or the SRLI for plants (http://threatenedplants.myspecies.info/ [Accessed 22 August 2012], and S. Bachman pers. comm.), the most recent threat status assessment was obtained. Each category was attributed a default numerical score on a linear scale, from 1 for the lowest category (Extinct) to 9 for Critically Endangered (this scale was set by the larger number of categories in the Red List data which has a “Lower Risk” category not present in the SRLI plant data), so that higher scores represent a greater risk of extinction. Species not occurring on either list were not attributed any score. The SRLI data had a “Not Evaluated” category, which was attributed a score of zero. Extinct in the Wild was treated as a higher priority category by attributing the second-to-highest score in both cases, with the view that the key goal of the prioritisation methodology is to achieve in situ conservation and species currently only persisting ex situ therefore require substantial conservation effort. For precautionary reasons, Data Deficient species were scored between Near Threatened and Least

20 MAPISCo Final report: 3. Development of the method

Concern, which in both the Red List and SRLI scoring is near the middle of the score distributions. See Tables 3 and 4 for the default numerical scoring for the Red List and SRLI for plants, respectively.

Table 3. Threat categories and number of species in the IUCN Red List (2012.1), and scores attributed. Species “Not Evaluated”(NE) were scored 0.

Category code Category No. spp. % of spp. Category score CR Critically Endangered 3947 6.412 9 EW Extinct in the Wild 63 0.102 8 EN Endangered 5766 9.368 7 VU Vulnerable 10105 16.417 6 NT Near Threatened 3452 5.608 5 DD Data Deficient 10497 17.054 4 LC Least Concern 26922 43.738 2 EX Extinct 801 1.301 1

Table 4. Threat categories and number of species in the Sampled Red List Index (SRLI) for Plants, and scores attributed.

Category Category code Category No. spp. % of spp. score CR Critically Endangered 1813 10.930 9 EW Extinct in the Wild 31 0.187 8 EN Endangered 2688 16.204 7 VU Vulnerable 4998 30.130 6 NT Near Threatened 752 4.533 5 DD Data Deficient 1244 7.499 4 LR/cd Lower Risk 223 1.344 3 LR/lc Lower Risk 909 5.480 3 LR/nt Lower Risk 677 4.081 3 LC Least Concern 3162 19.062 2

EX Extinct 91 0.549 1

Although both the relative scores among species and the scale used (e.g. linear, exponential) are inevitably largely subjective, the translation of threat categories into numerical scores used here is similar to that used in previous prioritisation studies (e.g. Rodriguez et al. 2004, Butchart et al. 2012). Because of the qualitative nature of these scores, we suggest that in the final version of the present prioritisation methodology these scores can be altered to suit changing expert opinion or policy aspirations.

In addition to conservation status assessments, from the RL and SRLI for plants we obtained lists of countries in which each species occurs, and lists of species occurring in forest habitats (IUCN Red List Habitat Classification 1) and wetlands (Classification 5).

21 MAPISCo Final report: 3. Development of the method

3.3. Database & prioritisation

3.3.1. Database building

For a full explanation of the database structure see Appendix 6. Briefly, data from the sources described in section 3.2, page 13, were cleaned (errors removed) and compiled using R (v. 2.15.0, R Development Core Team 2012) and resulting tables stored in a SQLite relational database (v. 3.7.11). The resulting main output table contained one row per species, with either a relevant value for each data source, or a “blank” indicating the species does not occur in that data set.

3.3.2. Co-benefit weighting, re-scaling and priority score calculation

A worked example of how the final priority score for each species was calculated is provided in Box 2). Broadly, the score was calculated by following these steps

1) Individual species scores from each data set were rescaled to allow them to be compared on the same scale. First, each individual dataset score was divided by the maximum value for scores from that particular dataset. This allowed all scores to be assigned a value between 0 and 1. For example, a species receiving a score of 7 for the species extinction co-benefit (i.e. a threatened species) was rescored to 0.778 (raw score of 7 divided by the maximum score for that database of 9 [the score given to Critically Endangered species]), while a species scoring 2 for extinction risk (Least Concern) was rescored to 0.222 (raw score of 2 divided by the maximum score of 9). For database scores listed only as 0 or 1 (binary scores, such as those for “Aquaculture use” obtained from FishBase and species listed as “Crop Wild Relatives”) this transformation had no effect. For both ES data sources (estimated carbon loss through deforestation and freshwater availability) lower scores were associated with higher priorities, so their scales were inverted as well as rescaled.

2) Scores attaining to each co-benefit were then combined to create a “score per co-benefit”. For the prevention of species extinction (section 3.2.5, page 20), this co-benefit score equalled either the Red List conservation status score or the plant SRLI score, whichever was greater. All other co-benefits scores (Habitat, Harvesting, Genetic Diversity and ES Provisioning) were calculated by taking the mean of the individual dataset scores contributing to the co-benefit.

3) The overall co-benefit score for each species was then standardised. This was necessary because while the individual database scores were “rescaled” to between zero and one as described in 1), their position along this 0-1 scale was arbitrary. For example, for a species in receipt of a co-benefit score of 1 for harvesting (of which there could only be a score of 1 or 0 due to limitations in the data) and a score of 0.56 for extinction list, the harvesting score is not “twice” as important as the habitat extinction score – it is only twice as large as a result of the scoring method of the individual datasets. For scores to be standardised a z- score calculation was made – this is an accepted standardising technique. The quantity z represents the distance between the raw score and the population mean in units of the standard deviation. z is negative when the raw score is below the mean, positive when above. This means that the co-benefit scores will be related in terms of the overall mean score.

22 MAPISCo Final report: 3. Development of the method

a. First, the mean and standard deviation of all the scores given to individual species was calculated for each co-benefit. It is important to note that this mean is calculated with empty cells being treated as missing data rather than as zero data (i.e. not included in the calculation of the mean). This is important, as the database contain a large percentage of “missing data” given few species receive scores in all databases. Treating them as “zeros” biases the database towards species that have scores for more co-benefits. By treating them as missing data this is a more accurate representation of what is known – i.e. it is not known if there is a relationship, rather than there is no relationship (p 45).

b. Each co-benefit score was then standardised by taking this mean score away from it, and dividing it by the standard deviation of this mean. The “z-score” was negative when the raw score was below the mean and positive when it was above.

4) The new score per co-benefit was then able to be modulated further by multiplying each by a weighting factor between 0 (no contribution) and 1 (maximum contribution). These weights can be modified based on policy decisions regarding the importance of each co- benefit. The default (as set in the database) is that all are equally important.

5) The final composite priority score per species equalled the sum of the weighted co-benefit scores.

The resultant list was sorted by decreasing final priority score. We used seven broad taxonomic groups to present the results below: amphibians (class Amphibia), birds (class Aves), fish (classes Actinopterygii, Cephalaspidomorphi, Chondrichthyes, Myxini and Sarcopterygii), mammals (class Mammalia), plants (kingdom Plantae), and reptiles (class Reptilia). These groups were chosen because they represent a wide range of taxonomic groups and are relatively well represented in the combined data sources used (e.g. as opposed to insects). The Red List was used as the primary source of taxonomic data, although for species not included on the Red List additional taxonomic data was used from the other data sets, where available, or inferred from the type of data. Species not belonging to any of the above groups were grouped as “Other” (mainly invertebrates).

23 MAPISCo Final report: 3. Development of the method

Box 2: A worked example of the final priority score calculation

The final priority score for a species is the sum of the scores given for the five co-benefits. The method for calculating co-benefits scores is outlined below.

1) Mean taken of the scores assigned 2) The mean score calculated in step 1 is then standardised by taking away from original individual datasets (in this from it the mean of all the values in the entire co-benefit column, then example, two different datasets) dividing it by the standard deviation of that co-benefit mean (calculation of a “z score”). The resultant score will be positive if the individual species score is greater than the mean score and negative if the individual species score is smaller than the mean score.

Species Habitat Harvesting Genetic diversity Ecosystem Threat status Final Score Service Provisioning Francolinus 2) The mean score calculated in 1) is then standardised by taking away from it camerunensis (0.136+0.789)/2 (0+0+0)/3the mean(0+0+3)/3= of all the values (0.323+0.975)/2 in the entire co-benefitmax(0.778,0) column, then dividing it by =0.462 = 0 the standard0.333 deviation of that= co0.649-benefit mean (calculation= 0.778 of a “z score”). The resultant score will be positive if the individual species score((5.05)+( is greater-2.58)+ 0.462-0.07 0-0.25 than the0.333 mean- 0.24 score 0.11 and negative0.462- 0.55if the individual 0.462 species-0.45 score is(0.82)+(0.77) smaller 0.08 0.10 than the mean=0.82 score. 0.11 0.25 +(1.35)= = 5.05 =-2.58 =0.77 =1.35 5.42 5.05*1 -2.58*1 0.82*1 0.77*1 1.35*1

3) The new co-benefit score is then multiplied by a 4) Final score calculated by adding together the five weighting factor (in this case all co-benefits are co-benefit scores. This score is then used to rank weighted equally (i.e. weighting set to 1) species in the priority list.

3.4. Inclusion of red list threat classification data

While Red List “threat category” (e.g. Critically Endangered, Near Threatened etc.) was included as a co-benefit in the methodology (from the Red List for animals and from the SRLI list for plants), the nature of that threat was not. The Red List “Threats Classification scheme” assigns a threat type to each of the species listed within it. The threats take a hierarchical format, with main threat categories subdivided into a number of subcategories, some of which are subdivided further (Table 5 and Appendix 8, Table A8-2 for full explanation of categories). The addition of this data to the overall database would allow the production of a) lists containing major threat classifications for particular species groups or geographical regions, and/or b) lists of species per threat classification.

24 MAPISCo Final report: 3. Development of the method

Species data were downloaded from the IUCN Red List website for each major threat category (categories 1-12 Table 5). Only “first tier” threat classifications were used, rather than sub- categories, as these data were more robust. These data were then integrated into the main MAPISCo database using the unique binomial species name and or species identification number as a link between data tables. This enabled us to produce lists of threat data as shown in Table 6. Threat data was not scored in the current incarnation of the database because of uncertainties in the IUCN threat classification process for each focal taxon. Where data are better standardised (such as for the birds) there is the potential for these threats to be ranked and scored in accordance with policy aspirations (expandable).

Table 5. The IUCN threat classification scheme categories (see Appendix Table A8-2 for full list and definitions)

Main threat category Sub-category (number of further sub-categories) 1 Residential & commercial 1.1 Housing & urban areas development 1.2 Commercial & industrial areas 1.3 Tourism & recreation area 2 Agriculture & 2.1 Annual & perennial non-timber crops (4) aquaculture 2.2 Wood & pulp plantations (3*) 2.3 Livestock farming & ranching (4) 2.4 Marine & freshwater aquaculture (3) 3 Energy production & 3.1 Oil & gas drilling mining 3.2 Mining & quarrying 3.3 Renewable energy 4 Transportation & service 4.1 Roads & railroads corridors 4.2 Utility & service lines 4.3 Shipping lanes 4.4 Flight paths 5 Biological resource use 5.1 Hunting & collecting terrestrial animals (4) 5.2 Gathering terrestrial plants (4) 5.3 Logging & wood harvesting (5) 5.4 Fishing & harvesting aquatic resources (6) 6 Human intrusions & 6.1 Recreational activities disturbance 6.2 War, civil unrest & military exercises 6.3 Work & other activities 7 Natural system 7.1 Fire & fire suppression (3) modifications 7.2 Dams & water management/use (11) 7.3 Other ecosystem modifications 8 Invasive & other 8.1 Invasive non-native/alien species/diseases (2) problematic species, genes 8.2 Problematic native species/diseases (2) & diseases 8.3 Introduced genetic material 8.4 Problematic species/diseases of unknown origin (2) 8.5 Viral/prion-induced diseases (2) 8.6 Diseases of unknown cause 9 Pollution 9.1 Domestic & urban waste water (3) 9.2 Industrial & military effluents (3) 9.3 Agricultural & forestry effluents (4) 9.4 Garbage & solid waste 9.5 Air-borne pollutants (4) 9.6 Excess energy (4) 10 Geological events 10.1 Volcanoes 10.2 Earthquakes/tsunamis 10.3 Avalanches/landslides 11 Climate change & 11.1 Habitat shifting & alteration severe weather 11.2 Droughts 11.3 Temperature extremes 11.4 Storms & flooding 11.5 Other impacts 12 Other options 12.1 Other threat

25 MAPISCo Final report: 3. Development of the method

Table 6. Example output of the MAPISCo database with threat classifications added

Taxonomic

Species

group

ystem modifications ystem

1. Residential 1. Agriculture 2. 3.Energy Transport 4. use Resource 5. Disturbance 6. S 7. species Invasive 8. Pollution 9. events Geological 10. change Climate 11. Other 12. status Threat Habitat Harvesting diversity Gen. provisioning ES Score Rank Francolinus birds 1 0 0 0 1 0 0 0 1 0 1 0 0.78 0.46 0.33 0.65 5.42 1 camerunensis

Caprimulgus birds 0 0 0 0 0 0 0 0 0 0 1 0 0.78 0.57 0.66 3.88 2 prigoginei

Afropavo birds 1 0 0 0 1 0 0 1 0 0 1 0 0.67 0.36 0.33 0.66 3.73 3 congensis

Craugastor amphibians 1 0 0 0 0 1 0 0 0 0 1 0 1 0.50 0.62 3.58 4 polymniae

Ecnomiohyla amphibians 1 0 0 0 0 1 0 1 0 0 1 0 1 0.50 0.62 3.58 4 echinata

26 MAPISCo Final report: 4. Results - Example priority lists

4. Results - Example priority lists

Capsule.

 The results generated by the database in its current format are constrained by data (e.g. only

around 3% of all plant species have been categorised on the IUCN Red List while almost all bird species are included).  For this reason in this section we present three different sets of results 1. All species 2. Birds only (taxonomic case study)

3. SE Asia only (geographic case study)  These case studies allow us to focus on discrete sets of data, which, while not eliminating constraints completely, allows a more meaningful demonstration of how the database can be used.

 These results are then linked to policy actions.

In this section we present species priority lists generated using the method outlined in section 3. However, we do so with an important caveat - priority lists generated by the current version of the method are limited by the data used to calculate species scores.

As described in section 3.2 (page 13), only 12 data sources were deemed suitable for inclusion in the method. This has resulted in constraints on the database, both in terms of which species have been able to be included on the lists, and in terms of the numbers of co-benefits on which species on the lists can be scored. There are also geographical biases, with many more species being recorded from some regions than others.

If this method is to become fully integrated into conservation policy in the UK, these constraints must be addressed. The focus of further development should be on sourcing data for taxa and from regions which are under-represented by the current version of the database, and also on identifying those species groups within the lists which have small numbers of co-benefit scores (see Box 3, page 28, for full description of constraints in the database). This topic is further discussed in both sections 5 (page 50) and 6.4 (page 59).

For these reasons, in this section we have presented three different priority lists.

1. A list for all species included on this list 2. A list containing only bird species (taxonomic case study) 3. A list containing only data from SE Asia (geographic case study)

The use of case studies allows us to focus on discrete sets of data, which, while not eliminating constraints completely, allows a more meaningful demonstration of how the database can be used.

27 MAPISCo Final report: 4. Results - Example priority lists

Box 3. Description of constraints incurred in the “all species” database

The coverage of the database (the number of species on the database as a percentage of all the species 4 currently described globally) varies greatly between taxa. It is very good for birds (100.64% ), mammals (100.22%4) and amphibians (94.10%; Table B3-1), but poor for reptiles (38.39%) and fish (37.82%) and poorer still for plants (6.30%) and “other” species (1.02%; Table B3-1).

Table B3-1 (from RL Stats Table 2 2012 IUCN Red List website)

Estimated number of Number of species on the Coverage of the Taxonomic group described species database database (%) Amphibians 6771 6371 94.10 4 Birds 10064 10128 100.64 Fish 32400 12252 37.82 Mammals 5501 5513 100.224 Other 1305250 13298 1.02 Plants 307674 19398 6.30 Reptiles 9547 3665 38.39

* note, the coverage for the database is greater than 100% for mammals and birds because some species listed on the database are not considered full species by all authorities, particularly those species that have been domesticated.

The database is also constrained by the number of co-benefits for which individual species have received scores (mean 2.29 for birds and 1.25 for plants; Table B3-2). This is significant because missing scores do not result from a “zero impact”, but from missing data. There are also large differences for the individual co- benefits, with birds receiving by far the most scores for habitat, and fish for harvesting (Table B3-2).

Mean number of Proportion of species in each taxon scored on each co-benefit (%)

Taxonomic group co-benefits scored Ecosystem (max = 5) Threat status Habitat Harvesting Genetic Diversity Services Birds 2.29 99.37 48.36 0.32 2.38 78.65 Amphibians 1.98 99.98 7.97 0.09 0 90.03 Mammals 1.77 99.78 3.01 0.33 3.66 69.73 Reptiles 1.63 99.97 0.46 0.08 0.49 61.47 Fish 1.57 84.57 0 26.03 0 46.89 Other 1.57 100.00 0.02 0.44 0 56.64 Plants 1.25 85.66 0.13 0.42 15.83 22.71

The case studies were selected, based on a range of different criteria, outlined below -

Taxonomic case study. Birds were selected for number of reasons: 1) birds are a very well-studied and understood group of species; 2) unlike for other taxa, there is only one Red List authority responsible for the prioritisation of all bird species (BirdLife International), which means the classification of species for the Red List is more uniform and more robust than for other taxa; 3) for birds, unlike for other taxa, all species are evaluated on the Red List. This is not the case for other taxonomic groups.

Geographic case study. SE Asia. Given the constraints in data availability between taxa it is important to note that any region will be biased in its taxonomic coverage. However, we

28 MAPISCo Final report: 4. Results - Example priority lists

present an example based on one region as an example of the applicability of the method. We selected SE Asia as the case study region because large numbers of species on the “all species” database were recorded from this region. There was also good taxonomic coverage for these species (Figure 2). A further case study using UKOTs was also carried out (as described in the original contract specification), which is detailed in Appendix 7.

For each set of lists we have outlined the main findings from the method. This is followed, for the case studies (sections 4.2, page 36 and 4.3, page 44), by a discussion of how some of the key findings could be related to policy actions. This section has not been included for the “all species” section (section 4.1, page 30), as we believe there to be too much uncertainty in these results for them to be related directly to policy actions.

Figure 2. The percentage contribution of each taxon (amphibians, birds, fish, mammals, plants, reptiles and other) to the MAPISCo database for each region (not including European regions). Regions are ordered by the number of species on the database they contain (shown in brackets). See Appendix 8, Table A8-3 for how countries were assigned to regions.

29 MAPISCo Final report: 4. Results - Example priority lists

4.1. Example 1: All species

4.1.1. Summary findings

Combining all data sources, and with all co-benefit weighting set to 1, the final output list consisted of 70625 species. The top 500 species from the list is shown in Appendix 9, Table A9-1.The top ten species in the list are Francolinus camerunensis, Caprimulgus prigoginei, Afropavo congensis (bird species), Craugastor polymniae, Ecnomiohyla echinata, Megastomatohyla mixe, Plectrohyla calvicollina, Plectrohyla celata, Plectrohyla cyanomma, Plectrohyla sabrina (amphibians) with priority scores ranging from 5.43 to 3.59 (Table 7). The score resolution1 for the full list is 18.55% (13255 unique ranks for 70625 species).

4.1.2. Taxonomic composition

The overall list is made up of 14.34% birds, 27.47% plants, 7.81% mammals, 5.19% reptiles, 9.02% amphibians, 17.35% fish and 18.83% “other” species (Table 7). The highest scoring species for each taxon are shown in Table 7. This top 500 list consists of 116 (23.2%) amphibians, 242 (48.4%) birds, 36 (7.2%) fish, 79 (15.8%) mammals, 10 (2%) plants, 13 (2.6%) reptiles and 4 (0.8%) other species (Table 8).

4.1.3. Geographic composition

In the overall list 241 countries are represented, and 80 in the top 500. The ten countries for which the largest percentage of bird species on the list have been recorded are shown in Table 9. In the overall list, Indonesia has the largest percentage of species on the global list (2.05%) and Brazil has the largest percentage of species (19.12%) in the top 500 list.

4.1.4. IUCN threat categories and classifications

The threat status of species in the overall and top 500 lists are shown in Table 10. The makeup of the overall list mirrors the Red List, apart for species that have not been assessed by it (6.72%). The majority of species on the list are classed as Least Concern (40.01%), followed by Data Deficient (15.22%), Vulnerable (14.50), Endangered (8.33%) Critically Endangered (5.68%) and Near Threatened (5.14%). The remaining categories make up less than 6%. In the top 500, species classed as Critically Endangered contribute the biggest proportion (30.00%), followed by species classed as Endangered (26.20%), Vulnerable (21.20%), Least Concern and Near Threatened (10.40% each). Species in the remaining categories make up less than 2%. When the numbers of species falling in each Red List major threat category were compared with the top 500 list, 3.74% of all the bird species listed as Critically Endangered, 3.17% of species listed as Extinct in the Wild, 2.23% of species listed Endangered, 1.03% of species listed as Vulnerable and 1.43% of species listed as Near Threatened were also in the top 500.

1 “Score resolution” refers to the ability of a given output list (priority list) to distinguish between species in terms of priorities: for example, some species receive the same final priority score, which results in the same rank (and therefore equal priority) for all these species. The score resolution is calculated as the number of unique ranks divided by the total number of species on a given list.

30 MAPISCo Final report: 4. Results - Example priority lists

The threat classification which occurred most frequently in both the overall and top 500 lists was Biological resource use (23.94% and 46.0% of species respectively). This category includes hunting, fishing and logging activities. Agriculture and aquaculture was the threat category next most frequently recorded (17.02% and 40.8%), followed by Natural system modifications (10.86% and 19.4%), Residential and commercial development (10.43% and 16.2%), Pollution (10.41% and 10.2 %) and Invasive species (8.38% and 16.2%).The remaining six classifications made up less than 7% each (Table 11 and Figure 3). Looking at each species groups individually, the patterns are remarkably similar (Table 11). The top two threats across all species groups are Biological resource use and Agriculture and aquaculture. Beyond the top two, there are individual taxa variations. Birds, for example, are more threatened by Climate change than any other taxa in the list (9.36% vs. 7.16% and below); fish are more threatened by Pollution than any other taxa (13.49% vs. 11.81% and below) and amphibians by Residential and commercial developments (13.22% vs. 11.79% and below).

4.1.5. Co-benefits

In the overall list, 56.84% are scored for threat status, 32.34% for ES Provisioning, 4.84% for Habitat and Area Conservation, 2.92% for Sustainable Harvesting and 3.04% for Genetic Diversity. In the top 500 list, 33.53% of species are scored for conservation status and ES Provisioning, 24.28% for Habitat and Area Conservation, 2.62% for Sustainable Harvesting and 6.04% for Genetic Diversity.

31 MAPISCo Final report: 4. Results - Example priority lists

Table 7. Top 20 species listed in the database

Species name Taxonomic Group

Threat status Threat Habitat Harvesting diversity Gen. provisioning ES Score Rank Francolinus camerunensis birds 0.78 0.46 0.33 0.65 5.43 1 Caprimulgus prigoginei birds 0.78 0.57 0.66 3.89 2 Afropavo congensis birds 0.67 0.36 0.33 0.66 3.74 3 Craugastor polymniae amphibians 1.00 0.50 0.62 3.59 4 Ecnomiohyla echinata amphibians 1.00 0.50 0.62 3.59 4 Megastomatohyla mixe amphibians 1.00 0.50 0.62 3.59 4 Plectrohyla calvicollina amphibians 1.00 0.50 0.62 3.59 4 Plectrohyla celata amphibians 1.00 0.50 0.62 3.59 4 Plectrohyla cyanomma amphibians 1.00 0.50 0.62 3.59 4 Plectrohyla sabrina amphibians 1.00 0.50 0.62 3.59 4 Pseudoeurycea saltator amphibians 1.00 0.50 0.62 3.59 4 Pseudoeurycea smithi amphibians 1.00 0.50 0.62 3.59 4 Pseudoeurycea unguidentis amphibians 1.00 0.50 0.62 3.59 4 Thorius aureus amphibians 1.00 0.50 0.62 3.59 4 Thorius smithi amphibians 1.00 0.50 0.62 3.59 4 Habromys chinanteco mammals 1.00 0.50 0.62 3.59 4 Habromys ixtlani mammals 1.00 0.50 0.62 3.59 4 Habromys lepturus mammals 1.00 0.50 0.62 3.59 4 Calyptura cristata birds 1.00 0.24 0.96 3.14 19 Duellmanohyla ignicolor amphibians 0.78 0.50 0.62 2.68 20

Table 8. Number, percentage, highest ranks and scores and highest-ranking species in the complete and top 500 species lists.

Taxonomic No. spp. in % spp. In Highest Highest group No. spp. % spp. top 500 top 500 rank score Highest scoring species

Craugastor polymniae, Ecnomiohyla echinata, Megastomatohyla mixe, Plectrohyla calvicollina, Plectrohyla celata, Plectrohyla cyanomma, Plectrohyla sabrina, Pseudoeurycea saltator, P. Smithi, P. Unguidentis, Thorius Amphibians 6371 9.02 116 23.2 4 3.59 aureus, T. smithi Birds 10128 14.34 242 48.4 1 5.43 Francolinus camerunensis Fish 12252 17.35 36 7.2 30 2.05 Acipenser sturio Mammals 5513 7.81 79 15.8 4 3.59 Habromys chinanteco Other 13298 18.83 4 0.8 149 0.08 Elga newtonsantosi Plants 19398 27.47 10 2 149 0.08 Devillea flagelliformis Reptiles 3665 5.19 13 2.6 89 0.69 Ctenosaura oaxacana

32 MAPISCo Final report: 4. Results - Example priority lists

Table 9. The top 10 countries in the overall and top 500 lists and the number and percentage of species which have been recorded as occurring within them.

Overall list Top 500 list Country Country No. % Country No. % rank spp. spp. spp. spp. 1 Indonesia 6298 2.05 Brazil 528 19.12 2 Ecuador 6116 1.99 Indonesia 99 3.59 3 India 5340 1.74 Mexico 85 3.08 United 4 States 5245 1.71 India 62 2.25 5 China 5059 1.65 Congo, The Democratic Republic of the 55 1.99 6 Brazil 4867 1.59 China 54 1.96 7 Malaysia 4842 1.58 Cameroon 53 1.92 8 Mexico 4782 1.56 Thailand 44 1.59 9 Colombia 4651 1.52 Myanmar 43 1.56 10 Thailand 4522 1.47 Malaysia and Argentina 39 1.41

Table 10. The proportion of species in the overall and top 500 lists and the IUCN threat category in which they are listed. The end column shows the proportion of bird species listed from each Red List category which occur in the top 500 list.

Overall list Top 500

% of all spp. in threat category Threat Status No. spp. % spp. No. spp. % spp. on red list in top 500

CR 4009 5.68 150 30.00 3.74 DD 10672 15.11 6 1.20 0.06 EN 5882 8.33 131 26.20 2.23 EW 63 0.09 2 0.40 3.17 EX 801 1.13 0 0 0 LC 28258 40.01 52 10.40 0.18 LR/cd 255 0.36 0 0 0 LR/lc 1018 1.44 1 0.20 0.10 LR/nt 1015 1.44 0 0 0 NE 29 0.04 0 0 0 NT 3631 5.14 52 10.40 1.43 VU 10243 14.50 106 21.20 1.03 Not Assessed 4749 6.72 0 0 0

33 MAPISCo Final report: 4. Results - Example priority lists

Table 11. Percentage of species classified as threatened by each of the 12 IUCN threat classification categories, for the overall list and for each species group individually.

Percentage of species classified as threatened by each of the 12 categories overall list top 500 mammals plants birds fish reptiles other amphibians Biological resource use 23.94 46.00 26.21 23.52 22.88 28.71 23.91 23.38 23.71 Agriculture /aquaculture 17.02 40.80 18.84 17.89 18.69 14.23 19.16 14.86 21.33 Natural system modification 10.86 19.40 10.51 12.28 12.60 11.29 9.90 10.73 8.56 Residential/commercial development 10.43 16.20 10.69 11.62 8.03 9.08 11.79 11.07 13.22 Pollution 10.18 10.20 8.90 9.91 7.44 13.49 10.01 11.81 10.03 Invasive species 8.39 16.20 7.87 8.31 8.98 8.12 8.58 9.17 8.55 Climate change 6.25 10.40 4.81 5.36 9.36 5.85 5.66 7.16 5.10 Human intrusions/disturbance 4.21 5.40 4.23 4.20 4.11 3.58 4.08 5.35 4.05 Energy production/mining 3.68 6.60 4.63 3.83 4.34 3.59 3.88 3.52 2.88 Transportation/service corridors 2.27 4.20 2.63 2.37 3.03 1.68 2.38 2.32 1.95 Geological events 0.48 1.20 0.59 0.57 0.51 0.27 0.49 0.51 0.55 Other threats 0.10 0.60 0.11 0.14 0.03 0.13 0.16 0.11 0.07

34 MAPISCo Final report: 4. Results - Example priority lists

Figure 3. Percentage of species classified as threatened by each of the 12 IUCN threat classification categories a) the overall list; b) the top 500. Full names for each threat category are given in Table 11.

35 MAPISCo Final report: 4. Results - Example priority lists

4.2. Example 2: Taxonomic case study – birds

4.2.1. Summary findings

The bird list contained 10128 species, 14.34% of the overall database (for the top 500 list see Appendix 9, Table A9-2). The score resolution for the overall list was 61.57% (6326 unique ranks for species) and for the top 500 list 89% (445 unique ranks). The top five species on the list are all Galliformes; Cameroon Francolin Francolinus camerunensis (scoring 6.53), Congo Peacock Afropavo congensis (4.62), Nahan's Francolin Francolinus nahani (3.39), White-breasted Guineafowl Agelastes meleagrides (2.30) and Swierstra’s Francolin Francolinus swierstra (1.44) (for top 20 see Table 12).

4.2.2. Orders and Families

The overall list consisted of birds from 25 orders. Passeriformes were the most represented in the list at 57.73%, followed by Apodiformes (4.96%), Piciformes (4.63%), Psittaciformes (4.13%), and Galliformes (3.49%) (Table 13).

The top 500 list consisted of species from 19 orders. Of these, two made up almost three-quarters of the species listed – Galliformes (44.8%) and Passeriformes (33.6%), (Table 13). Sixty-eight individual families made up the top 500, with Phasianidae, making up the largest percentage (36.2%). The highest ranking species in each order is also shown in Table 13.

4.2.3. Country/region

Two hundred and thirty four countries are represented in the overall list, and 80 in the top 500. The ten countries for which the largest percentage of bird species on the list have been recorded are shown in Table 14. In the overall list, Colombia has the largest percentage of species on the global list (1.71%) and Brazil has the largest percentage of species (29.40%) in the top 500 list. The number and proportion of species recorded in each country, for both the overall and top 500 lists, is detailed in Appendix 9, Table A9-3.

4.2.4. IUCN Red List threat categories and classifications

The threat status of species in the overall and top 500 lists is shown in Table 15. The makeup of the overall list mirrors the Red List, apart for species that have not been assessed by it. The majority of species are classed as Least Concern (75.8%), followed by Near Threatened (8.69%), Vulnerable (7.18%), Endangered (3.84%) and Critically Endangered (1.95%). The remaining categories make up less than 3%. In the top 500, again species classed as Least Concern contribute the biggest proportion (36.6%), followed by species classed as Vulnerable (20.2%), Near Threatened (19.2%), Endangered (11.8%) and Critically Endangered (10.8%) followed by species in the remaining categories making up less than 2%. When the numbers in each category on the Red List were compared with the top 500 list, 27.41% of all the bird species listed as Critically Endangered by the Red List were in that top 500, 25% of species listed as Extinct in the Wild, 15.17% of species listed Endangered, 13.89% of species listed as Vulnerable and 10.91% of species listed as Near Threatened (Table 15).

36 MAPISCo Final report: 4. Results - Example priority lists

In the overall list the threat classification against which the majority of species are listed is Biological resource use (22.33%). Agriculture and aquaculture (21.56%), Natural system modifications (14.28%), Climate change (12.42%), Invasive species (9.45%) and Residential/commercial development (4.79%) make up the next largest proportions. The remaining 6 categories make up the remaining 15% (see Table 16 for full list). For the top 500 list, the threat classification against which the majority of species are listed is also Biological resource use (23.90%) followed by Agriculture and aquaculture (22.54%), Natural system modifications (12.10%), Invasive species (9.53), Climate change (9.23%), Energy production and mining (5.59%), Residential and commercial development (4.99%) and Transportation and service corridors (4.54%). The remaining four categories made up less than 8% of the threats listed Table 16).

4.2.5. Co-benefits

In the overall list, 43.29% are scored for Threat Status, 33.67% for ES Provisioning, 20.62% for Habitat and Area Conservation, 0.68% for Sustainable Harvesting and 1.73% for Genetic Diversity. In the top 500 list, 31.83% of species are scored for Conservation Status and ES Provisioning, 29.79% for Habitat and Area Conservation, 0.57% for Sustainable Harvesting and 5.98% for Genetic Diversity.

4.2.6. Key findings and how they relate to policy

Species from the order Galliformes made up the majority of the top 500 bird species on the priority list. This is not surprising considering that the majority of Galliformes are scored on three co- benefits more (mean number per species = 3.10). As one of the most threatened group of birds, with over 25% of species in the group being classified as Vulnerable, Endangered or Critically Endangered, Galliformes score highly on Threat Status. Over two-thirds of the Galliformes in the top 500 have a score for Genetic Diversity; this reflects their close genetic relationship to the domesticated chicken, guineafowl, pheasant and quail. Over two-thirds have a score for Ecosystem Services and over half for Habitat, which reflects forest being the predominant habitat of Galliformes on the top 500 list. These characteristics are scored highly by the current co-benefit scoring system.

In order to maximise conservation benefit for Galliformes species (if we accept the inherent constraints in the MAPISCo prioritisation) policy-makers could target resources to countries with a high number of Galliformes species from the top 500 list. These countries (shown in Figure 4) include Indonesia, Brazil, The Democratic Republic of the Congo and Malaysia.

37 MAPISCo Final report: 4. Results - Example priority lists

Table 12. The top 20 highest scoring species, all in the order Galliformes, the country(ies) in which they have been recorded, scores each of the five co- benefits as well as resultant priority score and rank. Note that all values have been rounded to three decimal points. All co-benefit weighting factors set to 1

(default).

Rank Species name English name Country

Threat Threat status Habitat Harvesti ng Gen. diversity ES provisio ning Score Cameroon 1 Francolinus camerunensis Cameroon 0.778 0.462 0 0.333 0.649 6.539 Francolin

2 Afropavo congensis Congo Peacock Democratic Republic of Congo 0.667 0.361 0 0.333 0.659 4.625

3 Francolinus nahani Nahan's Francolin Uganda & Democratic Republic of Congo 0.778 0.245 0 0.333 0.640 3.389

White-breasted 4 Agelastes meleagrides Cote d' Ivoire, Ghana, Liberia & Sierra Leone 0.667 0.230 0 0.333 0.606 2.301 Guineafowl Swierstra's 5 Francolinus swierstrai Angola 0.778 0.116 0 0.333 0.626 1.436 Francolin Plumed Angola, Cameroon, Central African Republic, Congo, Democratic Republic of Congo, 6 Guttera plumifera 0.222 0.340 0 0.333 0.584 1.400 Guineafowl Equatorial Guinea & Gabon

Angola, Cameroon, Central African Republic, Congo, Democratic Republic Congo, Equatorial 7 Agelastes niger Black Guineafowl 0.222 0.330 0 0.333 0.597 1.380 Guinea, Gabon, Nigeria

Crimson Horned- 8 Tragopan satyra Bhutan, China, India, Nepal 0.556 0.231 0 0.333 0.561 1.347 pheasant

9 Francolinus ochropectus Djibouti Francolin Djibouti 1 0.026 0 0.333 0.619 1.233

Edwards's 10 Lophura edwardsi Vietnam 1 0.007 0 0.333 0.595 0.754 Pheasant Spot-winged 11 Odontophorus capueira Argentina, Brazil, Paraguay 0.222 0.151 0 0.333 0.795 0.642 Wood-quail

12 Lophura hoogerwerfi Aceh Pheasant Indonesia 0.667 0.016 0 0.333 0.747 0.564

38 MAPISCo Final report: 4. Results - Example priority lists

Bornean Peacock- 13 Polyplectron schleiermacheri Indonesia, Malaysia 0.778 0.013 0 0.333 0.684 0.514 pheasant

Salvadori's 14 Lophura inornata Indonesia 0.667 0.012 0 0.333 0.747 0.501 Pheasant Grey-breasted 15 Arborophila orientalis Indonesia 0.667 0.007 0 0.333 0.747 0.434 Partridge Dark-backed 16 Odontophorus melanonotus Colombia, Ecuador 0.667 0.098 0 0.333 0.603 0.411 Wood-quail Angola, Cameroon, Central African Republic, Congo, The Democratic Republic of Congo, Cote 17 Francolinus lathami Forest Francolin d' Ivoire, Equatorial Guinea, Gabon, Ghana, Guinea, Liberia, Nigeria, Sierra Leone, Sudan, 0.222 0.249 0 0.333 0.608 0.329 Tanzania, Togo, Uganda Handsome 18 Francolinus nobilis Burundi, Democratic Republic of Congo, Rwanda, Uganda 0.222 0.225 0 0.333 0.629 0.177 Francolin

19 Cyrtonyx ocellatus Ocellated Quail El Salvador, Guatemala, Honduras, Mexico, Nicaragua 0.667 0.069 0 0.333 0.617 0.124

Tacarcuna Wood- 20 Odontophorus dialeucos Colombia, Panama 0.667 0.079 0 0.333 0.590 0.027 quail

39 MAPISCo Final report: 4. Results - Example priority lists

Table 13. Proportion of species in the overall and top 500 species lists by Order, along with the highest rank and score per order and highest scoring species.

Order Overall list Top 500 Highest rank Highest score Highest scoring species Red List status No. spp. % spp. No. spp. % spp.

PASSERIFORMES 4763 57.73 168 33.60 207 -8.230 Calyptura cristata CR APODIFORMES 409 4.96 11 2.20 242 -8.960 Schoutedenapus schoutedeni VU PICIFORMES 382 4.63 8 1.60 306 -11.920 Indicator pumilio NT PSITTACIFORMES 341 4.13 30 6.00 249 -10.210 Touit melanonotus EN GALLIFORMES 288 3.49 224 44.80 1 6.530 Francolinus camerunensis EN FALCONIFORMES 275 3.33 5 1.00 261 -10.923 Leptodon forbesi CR COLUMBIFORMES 270 3.27 3 0.60 415 -12.990 Columba albinucha NT CHARADRIIFORMES 214 2.59 1 0.20 498 -13.560 Charadrius thoracicus VU CORACIIFORMES 207 2.51 3 0.60 447 -13.220 Bycanistes cylindricus VU STRIGIFORMES 181 2.19 10 2.00 245 -9.660 Phodilus prigoginei EN GRUIFORMES 177 2.15 3 0.60 295 -11.740 Psophia viridis EN ANSERIFORMES 159 1.93 18 3.60 21 0.001 Cairina scutulata EN CUCULIFORMES 152 1.84 2 0.40 422 -13.060 Neomorphus squamiger VU CICONIIFORMES 111 1.35 1 0.20 450 -13.212 Bostrychia bocagei CR CAPRIMULGIFORMES 100 1.21 2 0.40 201 -7.435 Caprimulgus prigoginei EN PROCELLARIIFORMES 55 0.67 3 0.60 340 -12.374 Pterodroma magentae CR TROGONIFORMES 44 0.53 0 0.00 587 -14.253 Apaloderma aequatoriale LC PELECANIFORMES 35 0.42 1 0.20 458 -13.300 Fregata andrewsi CR TINAMIFORMES 32 0.39 1 0.20 401 -12.895 Crypturellus noctivagus NT PODICIPEDIFORMES 22 0.27 0 0.00 821 -14.956 Podiceps taczanowskii CR STRUTHIONIFORMES 12 0.15 6 1.20 29 -0.238 Casuarius casuarius VU SPHENISCIFORMES 10 0.12 0 0.00 1548 -16.062 Eudyptes robustus VU PHOENICOPTERIFORMES 5 0.06 0 0.00 2663 -17.468 Phoeniconaias minor NT GAVIIFORMES 5 0.06 0 0.00 3509 -17.928 Gavia adamsii NT COLIIFORMES 2 0.02 0 0.00 3105 -17.689 Colius castanotus LC

40 MAPISCo Final report: 4. Results - Example priority lists

Table 14. The top 10 countries in the overall and top 500 lists and the number and percentage of species that have been recorded as occurring within them.

Overall list Top 500 list Country Country No. spp. % spp. Country No. spp. % spp. rank 1 Colombia 1835 1.71 Brazil 147 29.40 2 Peru 1814 1.69 DR Congo 27 5.40 3 Brazil 1766 1.65 Cameroon 18 3.60 4 Ecuador 1647 1.54 Indonesia 15 3.00 5 Indonesia 1600 1.49 India, Uganda 11 2.20 Argentina, Liberia, 6 China 1269 1.19 10 2.00 Malaysia, Nigeria Colombia, Cote d' 7 India 1225 1.14 Ivoire, Ghana, 9 1.80 Paraguay Gabon, Myanmar, 8 DR Congo 1129 1.05 8 1.60 Sierra Leone Central African Republic, China, 9 Mexico 1103 1.03 7 1.40 Congo, Equatorial Guinea Angola, Guinea, Nepal, 10 Kenya 1098 1.03 6 1.20 Thailand, Viet Nam

41 MAPISCo Final report: 4. Results - Example priority lists

Table 15. The proportion of species in the overall and top 500 lists and the IUCN threat category in which they are listed. The end column shows the proportion of bird species listed from each Red List category that occur in the top 500 list.

Overall list Top 500 % of all spp. in threat category on red list in top Threat Status No. spp. % spp. No. spp. % spp. 500 CR 197 1.95 54 10.8 27.41 DD 60 0.59 4 0.8 6.67 EN 389 3.84 59 11.8 15.17 EW 4 0.04 1 0.2 25.00 EX 130 1.28 1 0.2 0.77 LC 7677 75.8 183 36.6 2.38 NT 880 8.69 96 19.2 10.91 VU 727 7.18 101 20.2 13.89 not classified 64 0.63 1 0.2 1.56

Table 16. Red List threat classifications of the species in the overall and top 500 lists.

Overall list Top 500 Threats No. spp. % spp. No. spp. % spp. Biological Resource use 1847 22.33 158 23.90 Transportation and service corridors 296 3.58 30 4.54 Pollution 294 3.55 26 3.93 Other threats 0 0 0 0 Natural system modifications 1181 14.28 80 12.10 Invasive species 782 9.45 63 9.53 Human intrusions and disturbance 281 3.40 21 3.177 Residential and commercial development 396 4.79 33 4.99 Geological events 35 0.42 3 0.45

Energy production and mining 349 4.22 37 5.59 Agriculture and aquaculture 1783 21.56 149 22.54 Climate Change 1027 12.42 61 9.23

42 MAPISCo Final report: 4. Results - Example priority lists

Galliformes

category 1 2 3 4 5 6 7 8

Figure 4. The global distribution of Galliformes species in the top 500 birds list. The categories relate to the number of Galliformes species (in the top 500) that are found in each country. Countries coloured white have no Galliformes listed in the top 500. (The map was produced using the package rworldmap South, A. (2011) rworldmap: A New R package for Mapping Global Data. The R Journal Vol. 3/1: 35-43.)

43 MAPISCo Final report: 4. Results - Example priority lists

4.3. Example 3: Geographic case study – SE Asia

4.3.1. General findings

The final output list for SE Asia contains 12496 species, which is 17.7% of the overall database. The score resolution for the full list is 23.66% (2956 unique ranks for 12496 species) and for the top 500 is 49.20% (246 unique ranks). The five highest priority species are the Togian Islands Babirus Babyrousa togeanensis (0.782), Anoa Bubalus depressicornis (0.782), Mountain Anoa B. quarlesi (0.782), Javan pig Sus verrucosus (0.782) and Aceh pheasant Lophura hoogerwerfi (0.503) (see Table 17 for the top 20 species).

4.3.2. Taxonomic composition

The overall list consists of 20.42% “other” species (2552 spp.), 20.27% plants (2533 spp.), 20.09% birds (2510 spp.), 18.17% fish (2271 spp.), 8.86% mammals (1107 spp.), 6.24% reptiles (780 spp.) and 5.95% amphibians (743 spp.) (Table 18). The top 500 list has a rather different composition, with birds and mammals contributing the largest proportions (36.8 %, 184 spp. and 35.0%, 175 spp. respectively), followed by plants (10.4%, 52 spp.), “other” species (6.8%, 34 spp.), amphibians (4.8%, 24 spp.), fish (4.2%, 21 spp.) and reptiles (2.0%, 10 spp.). The highest scoring species in each taxon are shown in Table 18. The highest scoring mammals, birds and fish are all in the top 20, while the highest scoring amphibian Duttaphrynus sumatranus is ranked at 86, “other” species Protosticta gracilis at 107, plant Taxus wallichiana at 44 and reptile Emoia ruficauda 193.

4.3.3. Geographic composition

The species in the overall and top 500 lists have been recorded in 11 countries - Brunei, Cambodia, Indonesia, Lao, Malaysia, Myanmar, Philippines, Singapore, Thailand, East Timor and Viet Nam. In the overall list, the greatest proportion of species have been recorded in Indonesia (18.28%) followed by Malaysia (14.05%), Thailand (13.13%), Viet Nam (11.42%), Myanmar (10.15%), Philippines (9.72%), Cambodia (6.40%), PDR Lao (6.39%), Singapore (5.64%), Brunei (3.55%) and East Timor (1.27%) (Table 19). In the top 500 list the country in which the largest proportion of species have been recorded is Indonesia (32.41%), followed by Malaysia (11.18%), Myanmar (10.15%), Thailand and Viet Nam (both 9.44%), Philippines (7.28%), Cambodia (6.46%), Myanmar (6.83%), Lao PDR (6.36%), Brunei (2.77%), Singapore (2.67%) and East Timor (1.85%) (Table 19). The top ranked species in each country are listed in Table 20.

4.3.4. IUCN Red List threat categories and classifications

The threat status of the overall and top 500 lists is shown in Table 21. The majority of species in the overall list are classed as Least Concern (47.7%), followed by Data Deficient (18.55%), Vulnerable (13.43%), Near Threatened (6.55%), Endangered (4.64%) and Critically Endangered (4.03%). The remaining categories make up less than 6%. In the top 500 list, species classed as Vulnerable made up the largest proportion (29%), followed by Critically Endangered (27%), Endangered (22%), Least Concern (14.2%), Near Threatened (6.6%), and Data Deficient (1.2%). No species in the top 500 were classed in the any of the remaining categories. In the overall list, the threat classification against which the majority of species are listed is Biological resource use (29.90%) followed by Agriculture and aquaculture (13.25%), Pollution (11.95%), Residential/commercial development (11.82%), Natural system modifications (8.46%), Climate change (7.34%), Human intrusions/disturbance (6.55%), Invasive species (6.18%), Energy

44 MAPISCo Final report: 4. Results - Example priority lists production and mining (2.74%). The remaining four categories make up less than 4% (see Table 22 for full list). For the top 500 list, the threat classification against which the majority of species were listed is also Biological resource use (25.6%), followed by Agriculture and aquaculture (20.79%), Natural system modifications (11.60%), Residential and commercial development and invasive species (both 8.35%), Climate change (7.64%), Pollution (5.20%), Energy production and mining (4.95%), Human intrusions and disturbance (3.68%), Transportation and service corridors (3.11%) and Geological events (0.71%) (Table 22).

4.3.5. Co-benefits

In the overall list, 100% are scored for Threat Status, 66.12% for ES Provisioning, 7.18% for Habitat and Area Conservation, 5.27% for Sustainable Harvesting and 1.58% for Genetic Diversity. In the top 500 list, 100% of species are scored for Conservation Status, 99.8% for ES Provisioning, 28.6% for Habitat and Area Conservation, 8.8% for Sustainable Harvesting and 13.6% for Genetic Diversity.

4.3.6. Key findings and how they relate to policy

In the top 500 list, the country in which the largest proportions of species have been recorded is Indonesia (32.41%). These 316 species (138 mammals, 112 birds, 23 plants, 20 amphibians, 11 fish, 9 “other” species and 3 reptiles) share similar threats (37.34% of these species are threatened by Biological resource use and 29.43% by Agriculture and aquaculture). For the highest ranked 10 species in the top 500 list (all from Indonesia), hunting and/or habitat destruction are the major threats listed by the IUCN Red List. Conservation actions that reduce habitat destruction and target unsustainable hunting of species in Indonesia should therefore benefit these priority species. Several species on the list, e.g. Javan warty pig Sus verrucosus and Sulawesi babirusa Babyrousa celebensis are illegally hunted in protected areas and therefore better community engagement and conservation law enforcement would benefit these species.

45 MAPISCo Final report: 4. Results - Example priority lists

Table 17. The top 20 highest scoring species, the order to which they belong, the country(ies) in which they have been recorded, scores each of the five co-benefits as well as resultant priority score and rank. Note

that all values have been rounded to three decimal points. All co-benefit weighting factors set to 1 (default).

Rank Species English name Taxon Country

Threat status Threat Habitat Harvesting diversity Gen. ES provisioning Score Babyrousa Togian Islands 1 mammals Indonesia 0.778 0.333 0.747 0.782 togeanensis Babirusa Bubalus 1 Anoa mammals Indonesia 0.778 0.333 0.747 0.782 depressicornis 1 Bubalus quarlesi Mountain Anoa mammals Indonesia 0.778 0.333 0.747 0.782

1 Sus verrucosus Javan Pig mammals Indonesia 0.778 0.333 0.747 0.782 Lophura 5 Aceh Pheasant Birds Indonesia 0.667 0.016 0.333 0.747 0.503 hoogerwerfi

6 Lophura inornata Salvadori's Pheasant Birds Indonesia 0.667 0.012 0.333 0.747 0.441

Arborophila Grey-breasted 7 Birds Indonesia 0.667 0.007 0.333 0.747 0.374 orientalis Partridge Babyrousa 8 Babiroussa mammals Indonesia 0.667 0.333 0.747 0.275 babyrussa Babyrousa 8 Sulawesi Babirusa mammals Indonesia 0.667 0.333 0.747 0.275 celebensis Callosciurus 8 Mentawai Squirrel mammals Indonesia 0.667 0.333 0.747 0.275 melanogaster Polyplectron Bornean Peacock- 11 Birds Indonesia, Malaysia 0.778 0.013 0.333 0.684 0.251 schleiermacheri pheasant Bubalus Mindoro Dwarf 12 mammals Philippines 1.000 0.333 0.608 0.208 mindorensis Buffalo

12 Sus cebifrons Visayan Warty Pig mammals Philippines 1.000 0.333 0.608 0.208

14 Lophura edwardsi Edwards's Pheasant Birds Viet Nam 1.000 0.007 0.333 0.595 0.172 Cambodia, Lao PDR, 15 Bos sauveli Grey Ox mammals 1.000 0.333 0.597 0.082 Thailand, Viet Nam Brunei Darussalam, Cambodia, Indonesia, Epinephelus 16 Estuary Cod Fish Malaysia, Myanmar, 0.556 0.611 0.614 -0.119 coioides Philippines, Singapore, Thailand, Viet Nam 17 Sus celebensis Celebes Pig mammals Indonesia 0.556 0.333 0.747 -0.233 Cambodia, Indonesia, Lao PDR, Malaysia, 18 Cairina scutulata White-winged Duck Birds 0.778 0.017 0.333 0.622 -0.407 Myanmar, Thailand, Viet Nam Brunei Darussalam, 19 Melanoperdix niger Black Partridge Birds Indonesia, Malaysia, 0.667 0.020 0.333 0.660 -0.441 Singapore Brunei Darussalam, Lophura 20 Crestless Fireback Birds Indonesia, Malaysia, 0.667 0.019 0.333 0.660 -0.448 erythrophthalma Singapore

46 MAPISCo Final report: 4. Results - Example priority lists

Table 18. Proportion of species in the overall and top 500 species lists by taxonomic group, along with the highest rank and score per group, highest scoring species and the country in which it is found.

Overall list Top 500 Taxonomic group No. Highest rank Highest score Highest scoring species Country (ies) No. spp. % spp. spp. % spp. Duttaphrynus sumatranus Amphibians 743 5.95 24 4.8 86 -2.056 Indonesia Sumartrian toad Lophura hoogerwerfi Aceh Birds 2510 20.09 184 36.8 5 0.503 Indonesia pheasant Brunei Darussalam; Cambodia; Epinephelus coioides Indonesia; Malaysia; Myanmar; Fish 2271 18.17 21 4.2 16 -0.119 Estuary Cod Philippines; Singapore; Thailand; Viet Nam Babyrousa togeanensis Mammals 1107 8.86 175 35 1 0.782 Indonesia Togian Islands Babirusa Protosticta gracilis Other 2552 20.42 34 6.8 107 -2.375 Indonesia Arthropod Taxus wallichiana Indonesia; Myanmar; Plants 2533 20.27 52 10.4 44 -0.842 Himalayan Yew Philippines; Viet Nam Emoia ruficauda Red-tailed Reptiles 780 6.24 10 2 193 -3.334 Philippines Swamp Skink

47 MAPISCo Final report: 4. Results - Example priority lists

Table 19. The 11 countries which feature in the overall and top 500 lists and the number and percentage of species which have been recorded as occurring within them.

Overall list Top 500 list Country Country No. spp. % spp. Country No. spp. % spp. rank 1 Indonesia 6298 18.28 Indonesia 316 32.41

2 Malaysia 4842 14.05 Malaysia 109 11.18

3 Thailand 4522 13.13 Myanmar 99 10.15

4 Viet Nam 3934 11.42 Thailand 92 9.44

5 Myanmar 3498 10.15 Viet Nam 92 9.44

6 Philippines 3347 9.72 Philippines 71 7.28

7 Cambodia 2205 6.40 Cambodia 63 6.46

8 Lao PDR 2202 6.39 Lao PDR 62 6.36

9 Singapore 1944 5.64 Brunei 27 2.77

10 Brunei 1223 3.55 Singapore 26 2.67

11 East Timor 436 1.27 East Timor 18 1.85

Table 20. The top ranking species in each country.

Highest Priority Rank Species Country score

Brunei -0.119 16 Epinephelus coioides, Estuary Cod

Cambodia 0.081 15 Bos sauveli, Grey Ox Babyrousa togeanensis, Togian Indonesia 0.782 1 Islands Babirusa People's Democratic Republic of Lao 0.081 15 Bos sauveli, Grey Ox Polyplectron schleiermacheri, Malaysia 0.251 11 Bornean Peacock-pheasant Myanmar 0.356 28 Epinephelus coioides, Estuary Cod Bubalus mindorensis, Mindoro Philippines 0.207 12 Dwarf Buffalo

48 MAPISCo Final report: 4. Results - Example priority lists

Table 21. The proportion of species in the overall and top 500 lists and the IUCN threat category in which they are listed.

Overall list Top 500 Threat Category No. spp. % spp. No. spp. % spp. CR 503 4.03 135 27.00 DD 2318 18.55 6 1.20 EN 580 4.64 110 22.00 EW 2 0.02 0 0.00 EX 7 0.06 0 0.00 LC 5960 47.70 71 14.20 NT 819 6.55 33 6.60 VU 1678 13.43 145 29.00 not evaluated 122 0.98 0 0.00 LR/cd 122 0.98 0 0.00 LR/lc 378 3.02 0 0.00 LR/nt 129 1.03 0 0.00

Table 22. Red List threat classifications of the species in the overall and top 500 lists.

Overall list Top 500 Threats No. spp. % spp. No. spp. % spp. Biological Resource use 4371 29.90 181 25.60 Transportation and service corridors 224 1.53 22 3.11 Pollution 1747 11.95 37 5.23 Other threats 6 0.04 0 0.00 Natural system modifications 1236 8.46 82 11.60 Invasive species 903 6.18 59 8.35 Human intrusions and disturbance 957 6.55 26 3.68 Residential/commercial development 1728 11.82 59 8.35 Geological events 34 0.23 5 0.71 Energy production and mining 401 2.74 35 4.95 Agriculture and aquaculture 1937 13.25 147 20.79 Climate Change 1073 7.34 54 7.64

49 MAPISCo Final report: 5. Using the method

5. Using the method

Capsule .

 Expandable. We demonstrate how additional species or co-benefit data can be added to the database, and outline how such changes impact on the ranking of priority lists.

 Adaptable. We examine the effect changing individual co-benefit weightings (i.e. making certain co- benefits “more important” in the calculation of priority lists than others) has on priority list ranking.  Usable. Here we outline the development of a web-based interface, which, using a variety of tabs and

graphics, allows users to fully explore the priority lists created by the methodology under a number of different scenarios. We view this as a critical feature of the GUI, as it makes it adaptable to policy aspirations.

5.1. Expandable –How does the does the priority list respond to the inclusion of additional co-benefit data? A plant example.

The method developed in this project has the capacity for additional datasets (either taxonomic or co-benefit) to be included within it should they become available. We see this expansion as a critical element of the method because at present, the number of datasets that contribute to the overall priority scores is relatively small (12 data sources). As discussed in section 4 (page 27), this has resulted in taxonomic and geographic constraints on the lists, which must be considered before this method can become fully integrated in conservation policy. One way to address these constraints is to add new co-benefit data to the database. At present, we believe we have included all currently available, verified data, by focussing on either the transcribing of existing datasets into a format usable by the method, or on the collection of new data, additional data would make a huge improvement to the database. As described in section 4 (page 27) this should focus on the taxa that are underrepresented in the current database - plants for example, are vastly underrepresented in the database in its current form – just 6.3% of all plants species currently described worldwide are on the priority list. This is not the case for other taxa, birds, the coverage for mammals and amphibians all approaching 100% (see Box 3 Table B3-1; page 28).

For this reason, we have chosen to investigate the effect additional plant data will have on the composition of priority lists

Method: To assess the effect of inclusion of further co-benefit data, we asked the IUCN SSC Palm Specialist Group (Bill Baker, Kew Gardens & IUCN-SSC Palm Specialist Group, pers. comm.) to use its specialist knowledge to score a selection of palm species based on their contribution to one of the five co-benefits – Harvesting. The group selected 64 species for inclusion in this assessment, based on the flagship species for palm conservation. However, only 52 of these species were already on the overall list. As this exercise was to address the addition of data to the list, we concentrated on these 52 species. The group were asked to score species on a scale of 0 to 2 where 0 is a species of zero value to harvesting and 2 is a species of the maximum value to harvesting. These values were then rescaled to fit with the original harvesting data. New co-benefit scores for harvesting were then calculated using these rescaled data set scores, following the procedure outlined in section 3 (page 12). Final priority scores for the full list were then calculated.

50 MAPISCo Final report: 5. Using the method

Results: For the 52 palm species previously included on the priority list, inclusion of the new harvesting co-benefit scores results in harvesting co-benefit scores being increased (mean value from 0.111 to 0.414). The mean final priority score for these species in the overall list also increased, from -7.043 to -1.224. This had a significant change in the overall species ranking within the list. Firstly, two palm species now ranked equal first at the top of the overall list - Carpoxylon macrospermum and Ceroxylon sasaimae (both previously ranked 37816th). Secondly, inclusion of this new data set increased the representation of plants in the top 500 by 4% (from 10 to 30 species), largely at the expense of fish (-2.4%) and amphibians (-1.8%).

Discussion: These results show clearly that the expansion of the database through the addition of new information will have considerable effects on species priority lists.

5.2. Adaptable - How does the priority list respond to changes in co-benefit weightings?

The facility to change the relative weight given to of each co-benefit in is built into the database. This allows the priority lists to be adapted to explore policy scenarios. If, for example, a policy wished to prioritise species based on their contribution to ecosystem services, the weighting this co-benefit was given in the methodology could be increased in relation to the other co-benefits. This would then give an ecosystem service-centric list. We illustrate this adaptability using a sensitivity analysis and by carrying out a worked example.

5.2.1. Sensitivity analysis

The species represented in the top 500 list changes as the co-benefit weightings are varied. We can illustrate the strength of this effect across the five co-benefits by decreasing the weight of one co-benefit by 0.1 intervals from 1 to 0 whilst keeping all others constant (at 1). This gives a total of 51 combinations (10 decreases in weighting by 0.1, for each of the 5 co-benefits, in addition to all co-benefit weights set to 1).

Decreasing the weighting of each co-benefit (relative to all others held at a constant weight of 1) from 1 to 0.1 results in a similar pattern of absolute change in rank for each co-benefit (see Figure 5 below). The greatest absolute change in species in the top 500 list occurs when Ecosystem services is reduced to a weight of 0.1 (the mean change in rank is 1588.01). In the 51 combinations of weightings tested, a total of 1064 different species occur in the top 500, with some species occurring in many or all iterations (up to 51 times). The distribution of species occurrence in the top 500 is distinctly bimodal: 504 (41.6%) species occurred in the top 500 in more than 30 iterations and 602 (49.4%) occurred fewer than 10 times. This suggests a surprising degree of stability of species representation in the top of the list irrespective of modest levels of variations in weighting. This most likely reflects the non-independence of the co-benefits outlined above. Appendix 9, Table A9-4 shows the list of species occurring more than 30 times in the top 500.

5.2.2. Worked examples – threat status and ecosystem services

If policy-makers wish to prioritise threat status above all other co-benefits then they could adjust the weight applied to it allowing it to have a greater influence on the final priority score. By setting the weight of Threat Status to 1.0 and all other co-benefits to 0.5, this changes the priority list in the following ways;

51 MAPISCo Final report: 5. Using the method

1. The percentage of Critically Endangered species in the top 500 increases by 24.2% compared to when all co-benefits are equally weighted (1.0). The percentage of Vulnerable, Near Threatened or Least Concern species decreases, but the number of Endangered species remains constant (see Table 23).

2. The taxonomic focus of the top 500 does not change extensively, with birds continuing to contribute the largest proportion of species, followed by amphibians, mammals and fish (Table 23)

If policy-makers wish to prioritise Ecosystem service provision and adjust the weight applied to it as described above, this changes the priority list in the following ways;

1. The percentage of Critically Endangered species in the top 500 decreases by 14.2%, compared to when all co-benefits are equally weighted (1.0). The percentage of Endangered and Vulnerable species also decreases, while the number of Near Threatened and Data Deficient species increase. The number of Least Concern species remains relatively constant (see Table 24).

2. The taxonomic focus of the top 500 list shifts from birds (48.4 % of species when all co- benefits are weighted equally) to amphibians (42.4 %) (see Table 24).

52 MAPISCo Final report: 5. Using the method

1500

1000 Status Hab Har GenD ES

500 Mean absolutechangeinrank Mean

0

-0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7 -0.8 -0.9 Change in weight (from 1)

Figure 5. Mean absolute change in the rank of species in the top 500, following decreases in co-benefit weightings.

Table 23. The proportion of species in the top 500 lists and the IUCN threat category in which they are listed when Threat Status or Ecosystem Services are given priority over other co-benefits (CBs).

IUCN Red list Threat Threat status weighted 1 (all other Ecosystem Services weighted 1 (all All CBs Status categories CBs weighted 0.5) other CBs weighted 0.5) weighted 1 CR 54.2 % 15.8% 30% EN 26.6% 14.2% 26.2% VU 12.6% 20.6% 21.2% NT 5.6% 14.6% 10.4% EW 0.4% 0.2% 0.4% LC 0.4% 9% 10.4% DD 0.2% 25.6% 1.2%

53 MAPISCo Final report: 5. Using the method

Table 24. The proportion of species in the top 500 lists in each taxonomic group when Threat status or Ecosystem services are given priority over other co-benefits (CBs).

Taxonomic Threat status weighted 1 (all Ecosystem Services weighted 1 (all All CBs group other CBs weighted 0.5) other CBs weighted 0.5) weighted 1 amphibians 29.8 42.4 23.2 birds 40.6 35.2 48.4 fish 6.4 0.6 7.2 mammals 17.2 14 16 other 1.6 1.6 0.8 plants 2.2 4.6 1.8 reptiles 2.2 1.6 2.6

5.3. Usable - Development of Graphic User Interface (GUI)

A GUI is defined as “a type of user interface that allows users to interact with electronic devices using images rather than text commands”. In the context of the MAPISCo database, we see a GUI functioning as a way of enabling non-technical users to explore the data within it without having to individually alter each component manually. We envisaged the GUI using a combination of lists, graphs and maps to display the data in an easily interpreted form, allowing users to investigate questions such as: What are the highest priority species? Where in the world does these species occur? What effect does altering particular co-benefit scores have on the overall ranking of the lists? Where does a particular species fall in the ranking?

5.3.1. GUI Development

The GUI was developed using the same open-source statistical environment as the original database - R. This allowed user interface and analysis routines to be integrated easily. R also has rich graphical routines, a rapid development time, good transparency of method, and the potential for modification by other team members. R is freely available for all common operating systems, relatively easy to install, used in universities worldwide and increasingly by major commercial organisations such as Google and the New York Times. Using the new R Package ‘Shiny’, released in November 2012 for user interface development allows user interfaces to be run locally as well as on a web server. In the latter case users do not need to have R installed.

An initial prototype user interface was presented by the developer Andy South at the MAPISCo steering group meeting in November 2012.The initial prototype allowed users to select a species from the list and then choose one of the following display options (tabs):

 Graphic: the position of this species in the MAPISCo priority ranking (with all weighting factors set to 1), as a bar chart with each individual species represented by one bar. Bars are coloured according to taxonomic group.

 Map: a world map showing which countries this species occurs in and an option to label the countries with their names.

54 MAPISCo Final report: 5. Using the method

 Score co-benefits: shows how the priority score for this species is calculated from the five co-benefits. This allows the user to see, for example, whether a particular species receives a high priority score because of its values for Threat Status, Habitat or Harvesting.

A subsequent version was made available in time for the CITES COP at the start of March 2013. The new version added the following functionality requested at the steering group meeting.

 Two stage selection process: first allowing users to select a taxonomic group, continent and country. A species list is displayed based on these selections allowing the user to then select an individual species and view the outputs outlined above.

 Weightings sliders for each of the five co-benefits. The sliders can be moved between zero and one (with one being the default starting value). Changing the weightings leads to re- calculation of the priority scores and all other UI components update with the re-calculated rankings.

 Rank Table tab: displays a table of the selected species list ranked by priority scores. The table contains the scientific name, taxonomic group, English name and scores for each of the five co-benefits as well as the overall priority score.

 About tab: gives a brief outline of the project, contact details for project participants and acknowledgements.

The current, development version, of the GUI can be viewed and tried out at: www.mapisco.org.uk.

5.3.2. Constraints and legacy

The resources available for this part of the project have been limited relative to the usual resources required to develop a fully featured, robust, useable software product. 14 days of developer time for GUI development were available from project funds. Therefore, the GUI that has been produced should be seen as a prototype to be developed further. We are keen to develop the user interface when funds can be sourced.

5.3.3. Future development options

With relatively few extra days development time the following options can be added to the GUI in the short term. Costed proposals for these options were provided to the project steering committee in February 2013 (options that were taken up at that stage and are included in the implementation described above are not included here).

 Outputting priority lists as a PDF or CSV file including metadata detailing the user, time of creation and weighting options selected.

 Produce a version of the user interface targeted specifically at Overseas Territories (OTs), only including species occurring in OTs and allowing specific OTs to be selected.

 Allow output of a single reference page for a species chosen, giving text, map and graphics, including specifying the co-benefit values and which weightings options chosen.

55 MAPISCo Final report: 5. Using the method

 Creation of a reference document/atlas containing a page for all species subject to a maximum PDF size of ~2GB. The page would have text, map, and a scorings graph.

 Provide a button that will link to other databases (WCMC, IUCN Red List) and/or image search for the selected species.

 Creation of an R package containing the MAPISCo database and analysis routines, documentation and helpfiles. Submission to international repository. This will make the database and methods easily accessible to researchers worldwide and will help to ensure project legacy.

These short term development options could provide a bridge to longer term developments for which there is considerable potential.

5.3.4. Future hosting options

The beta-test version of the GUI is currently hosted on a test server and redirected from the www.mapisco.org.uk domain name. There is no guarantee how long this test server will remain available. To make the user interface freely available online in the long term there are two options. The first option is to make it available on a project specific server running R and shiny thus incurring no extra costs beyond hosting. The second option is to use the Shiny hosting service, which would incur an as yet unknown monthly fee. The related issue of where the database should be hosted is considered in section 6.52, page 61.

56 MAPISCo Final Report: 6. Discussion

6. Discussion

Capsule  The method developed allows the prioritisation of species based on their contribution to five co- benefits - conservation effort directed at high ranking species is expected, therefore, to contribute most to biodiversity, via the selected co-benefits.  The database that we have compiled contains information on 70000 species that has been consolidated from a suite of databases held by other organisations. Creating, curating and updating primary databases is time-consuming and expensive and so the coverage of species and co-benefit scores is variable across major species-groups. For example, birds are well-covered, plants much less so.  There is clear scope for Defra to build on the progress made to date so that scientific knowledge and practice can better support UK government objectives. In order to do this, the database requires modest technical development and a permanent home, the scientific rationale linking species and co-benefits should be strengthened, and the policy arenas where it can be used should be defined more closely.

6.1. Fit to original project brief

The original contract specification for this project required the creation of a methodology to prioritise species conservation effort for the greatest contribution to “consequential benefits for other species (or taxa), habitats, wider ecosystems, and ecosystem services”. The methodology was to be: expandable allowing the incorporation of future data, adaptable to changing policy aims and usable by non-technical practitioners.

The methodology presented in this report meets the specification outlined above by focusing on a selection of five priority co-benefits (habitat conservation, genetic diversity, harvesting, species extinction risk and ecosystem services). The steps involved in developing a priority list of species for conservation investment included: (i) identifying 2-3 data sources which could be used to quantify the value of a given species to each co-benefit, (ii) computing standardised scores for each species on each co-benefit (across data sources), (iii) summing these scores to create a final ranked priority score, weighted as required. In theory, conservation effort directed at species ranked highly on the priority list could be expected to contribute most to the selected co- benefits. This would permit greater contributions to:

1) the prevention of species extinctions by on average focusing effort on more highly threatened species (Aichi Target 12),

2) the conservation of habitats by focusing effort on those species used to identify a selection of Key Biodiversity Areas in which larger number of species co-occur (Aichi Targets 5 and 7),

3) the promotion of sustainable harvesting by focusing on harvested species of the greatest economic value (Aichi Target 6),

57 MAPISCo Final Report: 6. Discussion

4) the conservation of genetic diversity of species of economic or social value, by focusing on wild relatives of crops and domesticated animals, and medicinal plant species (Aichi Target 13) and

5) the protection of ecosystem service provisioning by focusing on species occurring in forest- and wetland habitats in countries with higher estimated rates of carbon loss through deforestation or lower freshwater availability (Aichi Target 14).

Therefore, conservation of the species highlighted by our approach presents the greatest potential to contribute to Aichi 5-7 and 12-14, as well as being an effective way to help direct conservation policy to contribute to international conservation agreements.

6.2. How does the method compare with ‘business as usual’?

One of the original drivers for this project was the perceived view that resources were often directed towards a few charismatic species. In the current methodology “politically interesting” or flagship species often championed by interest groups do not generally rank highly (e.g. Asian Elephant Elephas maximus is ranked 397th, African Elephant 553rd, Tiger 1759th, Giant Panda Ailuropoda melanoleuca 9473rd, African Lion Panthera leo 6724th, Eastern Gorilla G. beringei 3819th, Lowland Gorilla Gorilla gorilla 2741st, Black Rhinoceros Diceros bicornis 37816th, Polar Bear Ursus maritimus 45625th and White Rhinoceros Ceratotherium simum 51510th). This is because they are associated with only a small number, if any, of the co-benefits considered here. The method we have devised is based on objective criteria which can be transparently adapted as policy aspirations change. This can be done by putting more or less value on each co-benefit by varying the associated co-benefit weighting. For example, clearly using objective criteria based on a range of co-benefits places a few of the charismatic species illustrated here in context: they are less likely to fulfil a range of goals as set out in the Aichi Targets which we focus on than a large number of other species.

6.3 How do co-benefits relate to IUCN threat status?

An aim of the MAPISCo approach was to prioritise species based on the consequential benefits associated with their survival in addition to their IUCN Red List threat status. We would therefore expect that resulting priority lists are not simply a reflection of threat status.

To test this expectation we subdivided the overall database into four sub-databases – 1) one containing all species which score on Red List Threat Status AND Habitat Conservation, 2) one containing all species which score on Red List Threat Status AND Harvesting, 3) one containing all species which score on Red List Threat Status AND Genetic Diversity, and 4) one containing all species which score on Red List Threat Status AND Ecosystem Services. This meant that each sub database contained data only for species that score on Red List Threat Status and each of the other four co-benefits in turn. For each of these four databases we created three new priority lists based on -. 1) Red List threat category scores alone (for which the scores will be 1 for Critically Endangered, 0.88 for Endangered etc). This represents how prioritisation decisions could be made if the Red List alone was used to rank species. 2) Scores from the co-benefit in that database

58 MAPISCo Final Report: 6. Discussion alone (e.g. scores for Genetic Diversity). 3) Scores from the threat category and co-benefit combined (the average of the two). The results of this are shown in Table 25 below.

Both the Habitat and Ecosystem Services co-benefits are significantly negatively related to Threat Status, meaning that more traditional approaches to conservation (based on extinction risk- the IUCN Red List) do not capture more recent concerns about protecting a range of co-benefits from each species.

When threat status and each co-benefit are combined, three of them (Habitat, Harvesting and Ecosystem Services) are positively correlated to threat status suggesting that the MAPISCo database does encompass extinction risk and these co-benefits as well. The exception is Genetic Diversity, which is negatively correlated to both IUCN status (although not significantly) and to IUCN status and Genetic Diversity suggesting that the relationship with this score and others is not straightforward. This is probably due to the scoring for this co-benefit which tends to be binary (either not related at all or quite highly related – section 3.2.3., page 17).

Table 25. Spearman’s rank correlation coefficient between IUCN status and each co-benefit (* = p < 0.05).

Genetic Ecosystem IUCN Habitat (n Harvesting diversity (n = services (n = Priority score status = 5553) (n = 1504) 811) 38946) Habitat -0.154* Harvesting 0.013 Genetic diversity -0.066 Ecosystem services -0.60* IUCN status + Habitat 0.890* 0.246* IUCN status + Harvesting 0.862* 0.450* IUCN status + Genetic Diversity 0.902* -0.328* IUCN status + Ecosystem services 0.863* 0.389*

6.4. Operating constraints

It is important to bear in mind that, in its current format, the method developed by this project is biased towards those taxa that have the greatest representation in the databases used to calculate priority scores. This is because some species and some co-benefits have been subject to more study and data collation than others. This issue is particularly acute for plants, which have a very low proportional representation in the current version of the database (this is discussed in full in section 4, page 27). This will inevitably result in the relative (overall mean) downgrading of plants in any species prioritisation process until more plants have been assessed on the Red List and in other databases. Therefore, we urge caution when using the method with all taxa: it is better to currently use it to ask specific questions such as prioritisation within well studied groups, such as birds.

59 MAPISCo Final Report: 6. Discussion

6.5. Integration of MAPISCo into decision-making – next steps

As biodiversity issues become mainstream in political processes there is a recognition that the interface between science and policy must be strengthened (e.g. Koetz et al. 2008). Drawing on experience with the Intergovernmental Panel on Climate Change (IPCC), the necessary elements include research produced by bodies external to the policy body; a structure for collecting new data and observations; and an assessment body to make information and knowledge accessible for policy makers (Lariguaderie & Mooney 2010a;b). Perhaps most importantly, it is vital to recognise that developing science-based policy is an iterative and adaptive process that relies on improving knowledge so as to reduce uncertainty coupled with dialogue between the research and policy community (Koetz, Farrell & Bridgewater (2011).

Thus, use of the MAPISCo methodology should be seen as an iterative process (Figure 6). As discussed in sections 4 and 5, the use of data sources to inform the co-benefits should be continually assessed, improved and expanded, in response to (changing) expert opinion. Because the lists generated by the current method are based on a relatively small number of data sources (12), this expansion is crucial to give greater robustness to decision-making. It would be sensible, therefore, for Defra to give priority to supporting the efficient collation and curation (and even collection) of such data. It may also be possible to develop formal means to incorporate expert opinion in the way in which data sets are used and scored (e.g. Howes, Maron, & Mcalpine 2010; Aguilera et al. 2011) (as with the Palms example, section 5.1, page 50).

Figure 6. Non-linear science-policy interface showing how MAPISCo may benefit from stronger dialogue between these two fields. Black arrows form part of the traditional “linear” interface, red arrows are the feedback required.

60 MAPISCo Final Report: 6. Discussion

To allow MAPISCo to be used successfully in the longer term, three areas require further attention: 1) Science, 2) Practical and 3) Policy. This will put MAPISCo on a sustainable footing and enable it to contribute to policy development.

6.5.1. SCIENCE. Ensuring that the methodology fully accounts for scientific advances

Linking species data to co-benefits

Information on the links between conservation of individual species and co-benefits is varied. Thus the ability to prioritise conservation on the basis of some co-benefits will be more limited in some species groups than others (see section 4, page 27). However, the knowledge base relating species conservation to co-benefits may improve in the future as interest in work on ecosystem services gains ground (including a thorough review commissioned by this project, see Appendix 3). If Defra is to progress with the MAPISCo method, further research may be appropriate refine the concepts and framework.

Conceptual advances

The analytical field of prioritisation in biodiversity conservation is moving rapidly and there are frequent advances in standardising variables and dealing with unknowns and uncertainty. It is important that relevant advances are tracked so that any necessary adjustments to the MAPISCo methodology can be made. Defra could achieve this by establishing a MAPSICo Secretariat, that undertakes the necessary surveillance or by commissioning regular updates.

Links to other Aichi Targets

The project has taken place within the context of Defra’s desire to maximise overall conservation benefit from its spend on species. Currently five co-benefits, linked to four Aichi Targets are included in the methodology so the priority lists are only relevant within the context of these particular targets. Different species would be ranked more highly if contributions to Target 9 (control of invasive species) or others were to be included.

Non-independence of data sets

There is a degree of overlap in type of data used for the calculation of co-benefit scores (e.g. extinction risk as based on the IUCN Red List categories is used for Threat Status calculations, but is also used in the identification of e.g. IBA and AZE species). In statistical analyses, such interdependence would be considered a problem. However, in this case, this interdependence results from the co-benefits themselves (and the Aichi Targets from which they are derived) being non-independent. For example, by addressing Aichi Target 12 (preventing species extinctions), many species relevant to Target 13 (Genetic Diversity) would also be covered (as many of these are listed on the IUCN Red List).

6.5.2. PRACTICAL - Maintenance of database and incorporation of additional data. Where will the database be housed?

There is an immediate need to determine where the database will be housed and what form technical support will be required. There are several options for hosting the database either within

61 MAPISCo Final Report: 6. Discussion

Defra, or with external hosts, such as the IUCN Red List Office (Cambridge), Newcastle University or UNEP-WCMC. The suitability of these options will depend on an assessment of the following factors:

 Cost;  Capability;  Ability to provide scientific support (see below);  Understanding of Defra’s policy needs and way of working; and  Duration of hosting contract that Defra proposes.

Scientific support needs

Some level of ongoing scientific support to Defra may be required to allow the MAPISCO methodology to be fully operational. It may be that the standards and other documentation (see below) together with training in the application and use of the methodology to provide prioritisation lists would be sufficient, rather than an ongoing ‘help desk’ approach. Defra will need to consider what support it is likely to require, for how long and in what form.

Define standards and documentation

To ensure that the database and methodology are used appropriately and to best effect it is desirable to develop technical documents that define the standards to be used and specify in what format any outputs should appear. This is also important to show that all queries run in MAPISCo are transparent and that the decisions on weightings are documented fully. Good examples of how this can be done and how useful it may be can be drawn from the IUCN Red List which has standard definitions and classifications (see http://www.iucnredlist.org/technical- documents/classification-schemes and links therein) and also produces outputs of searches in a standardised way with a recommended citation. With some consideration, it would be possible to produce a similarly standardised output from the graphical user interface that gives all decisions made on weighting and identifies the person running the query as the author.

Incorporating new data and updating existing data

Data availability and accessibility is a concern. A very limited number of datasets exist which contain the information required for the MAPSICo methodology. Fewer still have been brought together, been adequately assembled and documented, and then made accessible. This is important because the availability and choice of data sources included in the database has consequences for the ranking of species. Thus, the ability to prioritise conservation on the basis of some co-benefits will be more limited in some species groups than others.

The paucity of data, for some taxa such as plants, means that small improvements in the availability of data can have a large impact on the resulting species priority list. The inclusion of specialist data on palm tree species into the database resulted in a considerable change in their place in the species ranking from no species in the top 2000; to two in the top five (see section 5.2, page 51). By exploring other scenarios where small improvements in the availability and/or quality of data linking species to co-benefits, it will be possible to focus resource investment in the

62 MAPISCo Final Report: 6. Discussion gathering and/or collation of data that can maximise impact on prioritisation ability. In addition, the method derived here for ‘within-group’ prioritisation could be a useful way forward.

All suitable datasets have been incorporated in the current database. Addressing remaining gaps may require various approaches. For example, taxonomic coverage is not uniform, the specific content of the database (the data fields) vary and there is substantial need to improve the curation and accessibility of many datasets before they could be considered for inclusion in the MAPSICo database. Finally, there will be varying motivations of data holders to share their data with Defra on a gratis basis. A practical first step would be to look at institutions close to Defra (and current/future partners) and produce a detailed analysis of what their data holdings are and how they can be made accessible to MAPSICo. A strong candidate here would be Kew Gardens and the data currently being assessed by the IUCN Red List in Cambridge. Access to these sources may add significantly to the MAPSICo coverage. Filling other gaps would require a more strategic approach and this would depend on immediate Defra priorities.

6.5.3. POLICY - Integrating MAPISCo into policy and resource allocation decisions

Demonstrating the potential of MAPISCo

MAPISCo would benefit from external review by scientists and policy-makers. Therefore, it is now very important to demonstrate the method to policy makers and senior officials in Defra and other potential users. This could involve a demonstration of how lists can be generated, the sorts of decisions that can be taken on co-benefit weightings and the impact varying these may have, and how the outputs can be used. One or more workshops with potential end users would likely be the best way to promote the method. If this could be combined with working through one or more current Defra species issues, it would be a very strong demonstration of the method.

Strengthening the ability of MAPISCo to underpin policy

The strengthening of links between both policy requirements and the framing of the scientific inputs, and between MAPSICo’s outputs and the impact they have on policy decision-making will be key integrating MAPISCo into working policy. Providing a broader array of policy situations in which to demonstrate how the method may be used would be helpful, as this would allow the range of assumptions and decisions to be assessed through the range of weightings applied by policy makers to MAPISCo at the input stage. At the other, output, end of the process, it would allow much greater understanding of the range of uses to which the priority lists (and associated data) generated would then inform the decisions that have to be taken. Exploring this with a range of users in a variety of contexts would allow the potential and the limits of the method to be defined more clearly.

A further exploration and demonstration of the value of MAPISCo would be a clearly defined project in which priority lists would be generated with various weightings (reflecting different policy demands) and comparing these against existing Defra priorities. This would permit assessment of both how well aligned they are and, perhaps more informatively, reasons for variation between current priorities and various MAPISCo outputs. Such an assessment should help bring into sharp focus unstated assumptions or other factors that may need to be incorporated into the methodology to account for the full range of contexts in which it may be used.

63 MAPISCo Final Report: 6. Discussion

Ease of use

The methodology is far more likely to be used if it is intuitive and clear. Therefore the interface (graphical user interface: GUI see section 5.3, page 54 for further discussion of GUI legacy) and the explanatory documentation needs to be easy to understand. The need here is for a short period of testing documentation and the GUI are a good fit to the users. There may be a need for refinement based on this testing.

6.6. Concluding remarks

We have delivered a first version of a methodology that can identify priorities for species conservation efforts based on expected contributions to a selection of five co-benefits. Our finding that around 1064 species are commonly ranked highly irrespective of variations in how the co- benefits are weighted, suggests that the proposed methodology does provide a blunt tool for identifying species where conservation effort could be expected to make significant contributions to the Aichi Targets.

This project is at the cutting edge of the science-policy interface. Although species prioritisation efforts are common, the vast majority of previous efforts are either geographically or taxonomically limited (e.g. Dunn, Hussell, & Welsh 1999; Knapp, Russell, & Swihart 2003; Rodríguez, Rojas- Suárez, & Sharpe 2004; Jimenez-Alfaro, Colubi, & Gonzalez-Rodriguez 2010), and are often limited to biological considerations only (Mace & Collar 2002; Mace, Possingham, & Leader- Williams 2006).

Although this project set out to establish clear objective criteria to determine how to prioritise species conservation investments, we have also shown that the choice of both co-benefit weighting and the data sources has a strong effect on which species are identified as higher priorities. As a result, the development of this methodology has brought the mismatch between the data requirements and data availability/accessibility for ambitious species prioritisation exercises into sharp focus.

Implied in the original project brief is an assumption of a relatively straightforward and linear science-policy interface, where science can directly meet policy needs and inform policy changes. Both the mismatch between the data required to fully service the original project brief and our results presented here highlight the need for a re-evaluation of this interface. As we have outlined above, at this stage of the methodology, further guidance and refinement of policy aims are required for science to make progress. In other words, a number of feedbacks from science into policy and back again need to be incorporated into a non-linear interface. As a result, the methodology described here becomes part of an iterative process where conservation science and policy meet and continuously refine each other’s needs, rather than a final answer to global species prioritisation problems.

64 MAPISCo Final Report: 7. References

7. References

Aguilera, P.A., Fernández, A., Fernández, R., Rumí, R. & Salmerón, A. (2011) Bayesian networks in environmental modelling. Environmental Modelling & Software, 26, 1376–1388.

Andelman, S. & Fagan, W. (2000) Umbrellas and flagships: Efficient conservation surrogates or expensive mistakes? Proceedings of the National Academy of Sciences of the United States of America, 97, 5954–5959.

Butchart, S.H.M., Scharlemann, J.P.W., Evans, M.I., Quader, S., Arico, S., Arinaitwe, J., Balman, M., Bennun, L.A., Bertzky, B., Besancon, C., Boucher, T.M., Brooks, T.M., Burfield, I.J., Burgess, N.D., Chan, S., Clay, R.P., Crosby, M.J., Davidson, N.C., De Silva, N., Devenish, C., Dutson, G.C.L., Fernandez, D.F.D.Z., Fishpool, L.D.C., Fitzgerald, C., Foster, M., Heath, M.F., Hockings, M., Hoffmann, M., Knox, D., Larsen, F.W., Lamoreux, J.F., Loucks, C., May, I., Millett, J., Molloy, D., Morling, P., Parr, M., Ricketts, T.H., Seddon, N., Skolnik, B., Stuart, S.N., Upgren, A. & Woodley, S. (2012) Protecting Important Sites for Biodiversity Contributes to Meeting Global Conservation Targets. Plos One, 7.

Butchart, S.H.M., Walpole, M., Collen, B., van Strien, A., Scharlemann, J.P.W., Almond, R.E.A., Baillie, J.E.M., Bomhard, B., Brown, C., Bruno, J., Carpenter, K.E., Carr, G.M., Chanson, J., Chenery, A.M., Csirke, J., Davidson, N.C., Dentener, F., Foster, M., Galli, A., Galloway, J.N., Genovesi, P., Gregory, R.D., Hockings, M., Kapos, V., Lamarque, J.-F., Leverington, F., Loh, J., McGeoch, M.A., McRae, L., Minasyan, A., Morcillo, M.H., Oldfield, T.E.E., Pauly, D., Quader, S., Revenga, C., Sauer, J.R., Skolnik, B., Spear, D., Stanwell-Smith, D., Stuart, S.N., Symes, A., Tierney, M., Tyrrell, T.D., Vie, J.-C. & Watson, R. (2010) Global Biodiversity: Indicators of Recent Declines. Science, 328, 1164–1168.

Cardinale, B.J., Duffy, J.E., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P., Narwani, A., Mace, G.M., Tilman, D., Wardle, D.A., Kinzig, A.P., Daily, G.C., Loreau, M., Grace, J.B., Larigauderie, A., Srivastava, D.S. & Naeem, S. (2012) Biodiversity loss and its impact on humanity. Nature, 486, 59–67.

Caro, T. & Girling, S. (2010) Conservation by Proxy: Indicator, Umbrella, Keystone, Flagship, and Other Surrogate Species. Island Press.

Caro, T. & O’Doherty, G. (1999) On the use of surrogate species in conservation biology. Conservation Biology, 13, 805–814.

Dunn, E.H., Hussell, D.J.T. & Welsh, D.A. (1999) Priority-Setting Tool Applied to Canada’s Landbirds Based on Concern and Responsibility for Species. Conservation Biology, 13, 1404– 1415.

Egoh, B.N., Reyers, B., Rouget, M. & Richardson, D.M. (2011) Identifying priority areas for ecosystem service management in South African grasslands. Journal of Environmental Management, 92, 1642–1650.

Eken, G., Bennun, L., Brooks, T.M., Darwall, W., Fishpool, L.D.C., Foster, M., Knox, D., Langhammer, P., Matiku, P., Radford, E., Salaman, P., Sechrest, W., Smith, M.L., Spector, S. & Tordoff, A. (2004) Key biodiversity areas as site conservation targets. Bioscience, 54, 1110– 1118.

65 MAPISCo Final Report: 7. References

FAO. (2000) World Watch List for Domestic Animal Diversity. Food and Agriculture Organisation of the United Nations, Rome, Italy. Available at: http://www.fao.org/docrep/009/x8750e/x8750e00.HTM [Accessed 23/08/2012].

FAO. (2010) Global Forest Resources Assessment 2010: Main Report. Food and Agriculture Organisation of the United Nations, Rome. Available at: http://www.fao.org/forestry/fra/fra2010/en/ [Accessed 23/08/2012].

Fisher, B., Bradbury, R.B., Andrews, J.E., Ausden, M., Bentham-Green, S., White, S.M. & Gill, J.A. (2011a) Impacts of species-led conservation on ecosystem services of wetlands: understanding co-benefits and tradeoffs. Biodiversity and Conservation, 20, 2461–2481.

Froese, R. & Pauly, D. (2012) FishBase V. 08/2012. Available at: http://www.fishbase.org. [Accessed 22/08/2012].

Goldman, R.L., Tallis, H., Kareiva, P. & Daily, G.C. (2008) Field evidence that ecosystem service projects support biodiversity and diversify options. Proceedings of the National Academy of Sciences of the United States of America, 105, 9445–9448.

Hawkins, B. (2008) Plants for Life: Medicinal Plant Conservation and Botanic Gardens. Botanic Gardens Conservation International, Richmond, UK. Available at: http://www.bgci.org/ourwork/medplants/ [Accessed 23/08/2012].

Hoffmann, M., Hilton-Taylor, C., Angulo, A., Böhm, M., Brooks, T.M., Butchart, S.H.M., et al. (2010) The Impact of Conservation on the Status of the World’s Vertebrates. Science, 330, 1503 –1509.

Howes, A.L., Maron, M. & Mcalpine, C.A. (2010) Bayesian Networks and Adaptive Management of Wildlife Habitat. Conservation Biology, 24, 974–983.

IUCN (2001) IUCN Red List Categories and Criteria: Version 3.1. IUCN Species Survival Commission, Gland, Switzerland and Cambridge, UK. Available at: http://www.iucnredlist.org/technical-documents/categories-and-criteria [Accessed 22/08/2012].

IUCN 2012. The IUCN Red List of Threatened Species. Version 2012.2. http://www.iucnredlist.org [Accessed 23/08/2012].

Jimenez-Alfaro, B., Colubi, A. & Gonzalez-Rodriguez, G. (2010) A comparison of point-scoring procedures for species prioritization and allocation of seed collection resources in a mountain region. Biodiversity and Conservation, 19, 3667–3684.

Knapp, S.M., Russell, R.E. & Swihart, R.K. (2003) Setting priorities for conservation: the influence of uncertainty on species rankings of Indiana mammals. Biological Conservation, 111, 223–234.

Koetz, T., Bridgewater, P., van den Hove, S. & Siebenhüner, B. (2008) The role of the Subsidiary Body on Scientific, Technical and Technological Advice to the Convention on Biological Diversity as science–policy interface. Environmental Science & Policy, 11, 505–516.

Koetz, T., Farrell, K.N. & Bridgewater, P. (2011) Building better science-policy interfaces for international environmental governance: assessing potential within the Intergovernmental Platform for Biodiversity and Ecosystem Services. International Environmental Agreements: Politics, Law and Economics, 12, 1–21.

66 MAPISCo Final Report: 7. References

Lamoreux, J.F., Morrison, J.C., Ricketts, T.H., Olson, D.M., Dinerstein, E., McKnight, M.W. & Shugart, H.H. (2005) Global tests of biodiversity concordance and the importance of endemism. Nature, 440, 212–214.

Lariguaderie & Mooney. (2010a) The International Year of Biodiversity: an opportunity to strengthen the science–policy interface for biodiversity and ecosystem services. Editorial overview. Current Opinion in Environmental Sustainability, 2, 1–2.

Lariguaderie & Mooney. (2010b) The Intergovernmental science-policy Platform on Biodiversity and Ecosystem Services: moving a step closer to an IPCC-like mechanism for biodiversity. Current Opinion in Environmental Sustainability, 2, 9–14.

Larsen, F.W., Turner, W.R. & Brooks, T.M. (2012) Conserving Critical Sites for Biodiversity Provides Disproportionate Benefits to People. Plos One, 7.

Lindenmayer, D.B., Manning, A., Smith, P., Possingham, H., Fischer, J., Oliver, I. & McCarthy, M. (2002) The focal-species approach and landscape restoration: a critique. Conservation Biology, 16, 338–345.

Mace, G.M. & Collar, N.J. (2002) Priority-setting in species conservation. In: Conserving Bird Biodiversity: General Principles and their Application Conservation Biology. Cambridge University Press, Cambridge, UK.

Mace, G. & Lande, R. (1991) Assessing Extinction Threats - Toward a Reevaluation of IUCN Threatened Species Categories. Conservation Biology, 5, 148–157.

Mace, G.M., Possingham, H.P. & Leader-Williams, N. (2006) Prioritizing choices in conservation. In: Key Topics in Conservation Biology Blackwell Scientific Publications, Oxford, UK.

MAPISCo Project Team. (2012) What Is the Evidence Base Linking Species Groups and Ecosystem Services? Newcastle University, Newcastle upon Tyne, UK.

McGowan, P.J.K. (2010) Conservation Status of Wild Relatives of Animals Used for Food. Animal Genetic Resources / Resources génétiques animales / Recursos genéticos animales, 47, 115– 118.

Millennium Ecosystem Assessment. (2005) Ecosystems and Human Well-Being. Island Press, Washington, DC, USA. Available at: http://www.maweb.org/en/index.aspx [Accessed 23/08/2012].

Myers, N., Mittermeier, R. A., Mittermeier, C. G., G. da Fonseca, A. B. & Kent, J. (2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853-858

Naidoo, R., Balmford, A., Costanza, R., Fisher, B., Green, R.E., Lehner, B., Malcolm, T.R. & Ricketts, T.H. (2008) Global mapping of ecosystem services and conservation priorities. Proceedings of the National Academy of Sciences, 105, 9495 –9500.

Olsen, D.M. & Dinerstein, E. (2002) The global 200: priority Ecoregions for global conservation. Annals of the Missouri Botanical Garden, 89, 199–224.

Owens, L.L. & McGowan, P.J.K. (In prep) Poultry wild relatives. Report to the Rufford Foundation.

67 MAPISCo Final Report: 7. References

R Development Core Team. (2012) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Available at: http://cran.r-project.org/ [Accessed 23/08/2012].

Ricketts, T.H., Dinerstein, E., Boucher, T., Brooks, T.M., Butchart, S.H.M., Hoffmann, M., Lamoreux, J.F., Morrison, J., Parr, M., Pilgrim, J.D., Rodrigues, A.S.L., Sechrest, W., Wallace, G.E., Berlin, K., Bielby, J., Burgess, N.D., Church, D.R., Cox, N., Knox, D., Loucks, C., Luck, G.W., Master, L.L., Moore, R., Naidoo, R., Ridgely, R., Schatz, G.E., Shire, G., Strand, H., Wettengel, W. & Wikramanayake, E. (2005) Pinpointing and Preventing Imminent Extinctions. Proceedings of the National Academy of Sciences of the United States of America, 102, 18497– 18501.

Rodríguez, J.P., Rojas-Suárez, F. & Sharpe, C.J. (2004) Setting Priorities for the Conservation of Venezuela’s Threatened Birds. Oryx, 38, 373–382.

Saetersdal, M. & Gjerde, I. (2011) Prioritising conservation areas using species surrogate measures: consistent with ecological theory? Journal of Applied Ecology, 48, 1236–1240.

UK NEA. (2011) The UK National Ecosystem Assessment: Synthesis of the Key Findings. UNEP- WCMC, Cambridge, UK.

Vincent, H., Wiersema, J., Leon, B., Dobbie, S., Kell, S., Fielder, H., Castenada, N., Guarino, L., Eastwood, R. & Maxted, N. (in prep.) A prioritised crop wild relative inventory to help underpin global food security. In preparation.

Vira, B. & Adams, W.M. (2009) Ecosystem services and conservation strategy: beware the silver bullet. Conservation Letters, 2, 158–162.

68