Wege: A New Metric for Ranking Locations for Biodiversity Conservation

Aim In order to implement effective conservation policies, it is crucial to know how biodiversity is distributed and one of the most widely used systems is the Key Biodiversity Areas (hereafter KBA) criteria, developed by the International Union for Conservation of Nature (IUCN). Here we develop a tool to rank Key Biodiversity Areas in a continuous scale to allow the ranking between KBAs and test this tool on a simulated dataset of 10 000 scenarios of species compositions of reptiles and mammals in eight locations in Mozambique. Location Mozambique, Africa Methods We compare the KBA criteria with four priorisation metrics (weighted endemism, extinction risk, evolutionary distinctiveness and EDGE score) to rank the biodiversity importance of eight sites with a randomly generated species composition of reptiles and mammals in Mozambique. Results We find that none of these metrics is able to provide a suitable ranking of the sites surveyed that would ultimately allow prioritization. We therefore develop and validate the “WEGE index” (Weighted Endemism including Global Endangerment index), which is an adaptation of the EDGE score (Evolutionarily Distinct and Globally Endangered) and allows the ranking of sites according to the KBA criteria but on a continuous scale. Main conclusions For our study system, the WEGE index scores areas that trigger KBA status higher and is able to rank their importance in terms of biodiversity by using the range and threat status of species present at the site. Prioritization may be crucial for policy making and real-life conservation, allowing the choice between otherwise equally qualified sites according to the KBA categories. WEGE is intended to support a transparent decision-making process in conservation.

the EDGE score (Evolutionarily Distinct and Globally Endangered) and allows the ranking of 27 sites according to the KBA criteria but on a continuous scale. 28

Main conclusions 29
For our study system, the WEGE index scores areas that trigger KBA status higher and is able 30 to rank their importance in terms of biodiversity by using the range and threat status of species 31 present at the site. Prioritization may be crucial for policy making and real-life conservation, 32 allowing the choice between otherwise equally qualified sites according to the KBA categories. 33 WEGE is intended to support a transparent decision-making process in conservation.

38
INTRODUCTION 39 In order to protect biodiversity and promote conservation, the decision-making process should 40 be based more on scientific research and data, and less on expert judgement not supported by 41 scientific studies (Sutherland et al., 2004). Threats to biodiversity such as conversion and 42 degradation of natural habitats, and invasion by non-native species and overexploitation, have 43 the potential of completely decimating biodiversity at local scales (Biofund, 2018;Mucova et 44 al., 2018). Therefore, in recent years there has been an increased awareness of the value of 45 protecting particular sites of high biological value, instead of focusing on large extensions of 46 land (Butchart et al., 2012). Such decisions may ultimately determine whether biodiversity is 47 preserved or lost. Thus, conservation planning should not only encompass the concepts of 48 global conservation prioritization (Myers et al., 2000), but also include a more local-scale 49 approach. 50 51 The Global Standards for the Identification of Key Biodiversity Areas (KBA) is an attempt to 52 gather a consensus on the distribution of key biodiversity by highlighting sites that contribute 53 significantly to the global persistence of biodiversity (IUCN 2004). The criteria and 54 methodology for identifying KBAs was created by the IUCN World Commission on 55 Protected Areas (IUCN, 2016). KBAs can vary considerably in size, and the criteria aim to 56 address aspects of biodiversity operating from regional to relatively local scales. The 57 categorization of areas is based on criteria such as presence and proportional inclusion of 58 threatened species and ecosystems, species' distribution ranges, ecological integrity and 59 irreplaceability. However, indices that directly measure biodiversity such as species richness 60 conservation. However, the accuracy of such indices is highly dependent on the quality and 71 availability of data, making poorly sampled areas particularly hard to evaluate (Faith, 1992 We assumed the presence of ≥10 reproductive units whenever a species was present in a 143 location. 144 To which we addressed by using the following conditions: 145 Presence of a CR or EN species with a distribution of 100 000 km 2 or less (corresponding to a 146 presence in one thousand 0.1-degree cells), presence of a VU species with a distribution of 147 10 000 km 2 or less (corresponding to a presence in one hundred 0.1-degree cells) and 148 presence of any species with a distribution of 1000 km 2 or less (corresponding to a presence 149 in 10 0.1-degree cells). 150

Biodiversity indices 152
To test whether we could use widely used biodiversity metrics to rank our locations, we 153 calculated the scores of four indices: WE, EDGE score, ER and ED and compared the ranking 154 of such metrics to our new index, WEGE. 155 Metrics such as EDGE, ER and ED, were calculated by summing the values of the species in 156 each community randomly generated. 157 To compare the different ranking of the different metrics for each of the 10 000 scenarios we 158 tested how often the different indices prioritize areas that trigger KBA status. 159 By using eight fictional locations, the number of areas triggering KBA status vary between 0 160 and 8 and the perfect ranking scores would vary between 1 for scenarios with 1 KBA and 36 161 for scenarios with 8 KBAs (1+2+3+4+5+6+7+8) (Appendix S1). 162 163 By comparing the distance between the obtained rankings from the different metrics and the 164 perfect ranking score we are able to compare the performance of the different indices at 165 ranking KBAs. restricted species in the group, the simulation was able to create the eight possible scenarios, 213 while mammals due to most species being widespread, there were less KBA trigger species to 214 trigger KBA status, thus, no scenario with eight areas qualified as KBA was generated. 215 In order to compare the ranking of KBAs between WEGE, WE, ER, ED and EDGE, we 216 summed all the ranking scores, where the perfect ranking score takes the lowest possible 217 value. Thus, the lower the sum of the ranking of the metrics, the shortest the distance to the 218 perfect ranking. In both vertebrate groups, WEGE outperformed the other metrics, followed 219 by WE, ER, EDGE and ED. In reptiles the difference between WEGE and WE was much 220 smaller, 494.14 to 566.61 (Fig. 2 B) than in mammals, 89.09 to 435.36 (Fig. 2 B). This 221 difference in performance is related to the fact that in mammals, unlike reptiles there are more 222 widespread endangered species. These species have the potential of triggering KBA status and 223 are weighted by WEGE, unlike WE, which doesn't take into account the conservation 224

parameter. 225
In order to test the sparsity of the ranking scores we normalized the data where a score of 0 is 226 the perfect score while the score of 1 is the worst. Both reptiles and mammals had most of 227 their WEGE scores closer to 0 when compared to the other metrics (Fig 1. C and Fig 2. C). 228 For both vertebrate groups tested in this study, reptiles and mammals, and for all KBAs sizes 229 tested (0.1 by 0.1, 0.5 by 0.5 and 1 by 1 degrees)

IUCN's KBA and priorisation indices 237
The IUCN's KBA uses a set of guidelines to check whether a particular site triggers a KBA 238 status, unlike biodiversity metrics which attempt to quantify different spectra of biodiversity. 239 Hence, different biodiversity metrics are expected to weight sites differently. The biodiversity 240 of specific sites should arguably not be assessed by just summing the number of species 241 existing in each location, but also taking into account other factors such as genetic diversity, 242 distribution ranges or conservation status (Magurran, 1988;Barthlott et al., 1999). Otherwise, 243 the presence of many widespread species producing a high SR would mask the importance of 244 vulnerable or endangered micro-endemic taxa (restricted to very few sites). 245

246
The fact that SR and PD indices are known to be highly correlated with sampling effort 247 In this study we tested the ranking of KBAs using WE, ER, ED, EDGE and our proposed 272 index WEGE. Locations that have higher WE are areas in which the species composition 273 contain species and more restricted ranges. Locations that score higher in ER are areas that 274 contain more species with higher threat status. Locations with higher ED are areas that house 275 species which have a higher evolutionary distinctiveness. Locations which score higher in 276 terms of EDGE are areas that have a composition of species with both high evolutionary 277 distinctiveness and threat levels. Finally, locations with higher WEGE scores, will be 278 locations with a combinations of range restricted and threatened species. 279 Regarding our analysis, using 10 000 simulated scenarios of species compositions of reptiles 280 and mammals, the WEGE index outperformed WE, ED, ER and EDGE both at overall sum of 281 ranking scores and density of scores closer to the perfect score. The results were consistent for 282 both vertebrate groups and KBA sites sizes tested. The second-best metric was WE, followed 283 by ER, EDGE and in last place ED. Interestingly, our results show that using ER alone would 284 be a more efficient way of ranking KBAs when compared to EDGE. Even though, both 285 EDGE scores and the KBA initiative are focused on the preservation of biodiversity, 286 according to our study they prioritize different sets of species. 287 The use of EDGE scores to rank sites is only expected to be efficient when the threats are metric is to combine conservation scoring of each species with a measure of the relative 302 importance of the site in question for each species. This could also be achieved by combining 303 a conservation score which incorporating evolutionary history such as e.g. PE rather than WE, 304 but since KBA by design weigh all species equally irrespective of their evolutionary 305 uniqueness we chose to select a measure with the same lack of taxonomic weighing. By 306 incorporating WE in the EDGE score formula and creating the WEGE index, we obtained an 307 index in line with the IUCN KBAs standards criteria compared to the WE, ED, ER and 308

EDGE. 309
The WEGE index can be used either to find suitable candidates' areas to be considered as 310 KBA's or as a mechanism of weighting the importance of biodiversity of particular KBA's as 311 well as areas outside KBA's. Additionally, it uses a simpler methodology by employing only 312 two metrics instead of a set of seven conditions (A1a -e and B1 and B2). Finally, WEGE can 313 act as a complement in the process, by which, sites selected using IUCN's KBA can now be 314 ranked objectively according to their biodiversity importance. 315

316
Complementing the categorical ranking of locations can bring great advantages when 317 prioritizing efforts with limited resources. IUCN's criteria lack this aspect by attributing a 318 binary system where one particular site either triggers KBA status or not. By using WEGE, 319 we rank sites within the same category and enabling the decision-making process to be 320 objective and transparent as possible. Getting conservation actions applied to any given area 321 usually demands a great deal of effort, so sensitivity to removing sites from KBA status is less 322 likely to be of high societal priority, but still, this methodology can highlight areas which even 323 though they trigger KBA status, their score is low and might be on the cusp of losing their 324 restricted species, will change if species become non-threatened or get their range 326 considerably expanded. Consequently, lower performing WEGE sites have higher odds of 327 losing their KBA status. One example that illustrates this scenario is the species 328 Cryptoblepharus ahli Mertens, 1928, described by Mertens (1928), synonymized to the 329 widespread species Cryptoblepharus africanus by Brygoo (1986)  status. Importantly, however, the two criteria in WEGE clearly still measure distinct processes 341 which for instance can be seen by the existence of widespread but endangered species like the 342 already mentioned Bluefin tuna or highly restricted and least concern as the Mount Mabu 343 Pygmy Chameleon. By combining the two we show that we get a better measure than solely 344 relying on IUCN criteria or solely on WE. Mozambique is a developing country that struggles to conciliate its rich biodiversity with the 381 for the mining industry, and the high potential economic gain that could follow. The country 382 also has one of the highest corruption levels in the world, and unbiased methods to quantify 383 biodiversity are a crucial parameter for a transparent decision-making process in conservation. 384 The selection of sites as KBAs is expected to have multiple uses, including conservation 385 planning support and priority-setting at national and regional levels (IUCN, 2016). Therefore, 386 the use of the WEGE index, allowing the ranking of key biodiversity areas is expected to by 387 association support a transparent ranking of sites in regards to conservation. 388 389

Supporting Information 390
Methods used for calculating indices, r packages used, KBA guidelines and raw data 391 (Appendix S1), is available online.