g Global priorities for conservation across multiple dimensions of mammalian diversity Fernanda T. Brum’ Carlo Rondinini, Rafael Loyola’, and Ana D. Davidson®* “conservation Bogeography Laboratory Departamento di Gradvagio em Eclogis, Universi University, NY 11794; iss Federal Researeh ns ogia, nversidade Federal de Goss, 74680-900 Golania, Federal do Rio Grande do Sul 15007 Port Alegre, Bras “De re WSL, CH-8803 Birmensdor,Swrtzerlane; ‘Department of Blelogy, Auburn University at ', Catherine H. Graham“, Gabriel C. Costa‘, S. Blair Hedges’, Caterina Penone®, Volker C. Radeloft", as "Programa de 9 ment af Ecology and Evolution, Stony rook Montgomery, Montgomery AL 26124, Center for Bodvesiy, Temple University, Pailadelphi, PA 18128 institute of Fant Sclencen Unversity of Bern, », Department of Biology and Biotechnologies, Sapienza University of Natural Hetage Program, Colorada State Unversty, For Calling, CO 80822 Progra Eales by? (received for review Apr 1, 2017) Conservation priorities that are based on species distribution, endemism, and vulnerability may underrepresent biologically unique species as well as their functional roles and evolutionary histories. To ensure that priorities are biologically comprehensive, multiple dimensions of diversity must be considered. Further, understanding how the different dimensions relate to one another spatially is important for conservation prioritization, but the relationship remains poorly understood. Here, we use spatial conservation planning to (identify and compare priority regions {for global mammal conservation across three key dimensions of biodiversity—taxonomic, phylogenetic, and traits—and (i) deter- mine the overlap of these regions with the locations of threatened species and existing protected areas. We show that priority areas for mammal conservation exhibit low overlap across the three dimensions, highlighting the need for an integrative approach for biodiversity conservation. Additionally, currently protected areas poorly represent the three dimensions of mammalian bio- diversity. We identify areas of high conservation priority among and across the dimensions that should receive special attention for expanding the global protected area network. These his priority areas, combined with areas of high priority for other tax- ‘nomic groups and with socal, economic, and political consider- ations, provide a biological foundation for future conservation planning efforts. complementarity | phylogenetic dimension | sata conservation Prioritization | taxonomic dimension | rat dimension juman activities ate rapidly transforming the planet and are the primary causes of biodiversity loss (1-4). In response to growing concern about the future of biodiversity, the 10th Convention on Biological Diversity developed a strategic plan for 2011-2020 that introduced protection targets called the ‘Aichi Targets” to facilitate conservation action (5). One of the primary targets is to protect 17% of the global land surface (6), focusing on areas of particular importance for biodiversity and ecosystem services (7, 8). Traditionally, global priority areas have been identified based on richness, species endemism, and vul- nerability (9, 10). Although these features are important in ‘identifying key biodiversity regions, they focus on only one di- ‘mension of biodiversity taxonomic diversity. Selecting regions based on species richness may not be the best conservation strategy, because richness does not reflect complementatity and thus could lead to priority areas with similar assemblage com- position, at the cost of protecting unique assemblages (11, 12). Species are a product of ecological and evolutionary processes, and the species that we observe today represent only the tips of the tee of life. If we measure only species numbers, we might ‘ise unique ecological and evolutionazy information, Therefore the taxonomic dimension may not sufficiently capture other won pat orgie. 107%pna 1706461114 1, Rome 00185, Italy; NatureServe, Arlington, VA 22203; and ‘Colorado ica C Weight, Centre Vali, Stony Brook University, Stony Brook, NY, and accepted by Eeitorial Board Member Ruth S. Defies May 31,2017 facets of diversity, such as evolutionary history and functional traits (13-15). Evolutionary history captures the uniqueness of Tinages through deep time (16) and can influence species’ sus ceplibility to extinction, because extinction risk is phylogeneti cally nonrandom (17-19). The trait dimension reflec ecological, morphological, and physiological strategies of species (20). Therefore losses in all three dimensions could have large con- sequences for biodiversity and ecosystem function (13). There is clearly a need for spatial conservation priorities that account for multiple dimensions of biological diversy, including taxonomy, phylogeny, and traits (15, 21, 22). However, the re- lationships among these dimensions across the globe are poorly understood and vary depending on the metric used and the scale of the analysis (15, 23). Some studies have shown a high corre- lation among taxonomic, phylogenetic, and trait diversity of global mammals (23, 24), especially when using diversity indices lffected by species richness, ic., phylogenetic diversity (14) and functional diversity (20). For example, species richness was an effective surrogate for the functional and phylogenetic dimen- sions of local rodent assemblages in Manu (25). In contrast, Mazel et al, (15) identified strong geographical mismatches among global hotspots of taxonomic, phylogenetic, and fune- tional diversity of mammals. Low geographical congruence was also found in prioritization analyses based on taxonomic, phy- logenetic, and trait diversity of birds and mammals in Brazil 21). Significance ‘Approximately a quarter of all land mammals are currently threatened, mostly by human activities including habitat loss and harvesting. Here, we provide the fist biological map of priority areas that captures all three dimensions of mammalian biodiversity: taxonomic, phylogenetic, and traits, We find lim ited overlap in priority regions across the three dimensions and with currently protected areas, indicating that conservation planning should consider multiple dimensions of biodiversity ‘to maximize biodiversity conservation. Our complementarity: based prioritization provides a conservation solution that can bbe incorporated in future conservation planning efforts aimed at helping protect not only species but also evolutionary po: ‘ential and ecosystem function. ries PNAS Dc Submision. RL 3 gus eter ned by the Edo Tear contin upping information linea srs: prarglckupue PNAS | Jay 48,2017 | vol 104 | o.29. | 7641-7646 sustamamury ScuNCE‘The problem with the lack of congruence among dimensions is ‘that conservation planning based on only one dimension does not fally represent biodiversity (26, 27). Policy responses to bio- diversity loss require knowledge of how these dimensions relate to ‘ach other and should use a mose integrated approach by con- sidering each dimension independently and all thee jointly. Global priority regions for mammal conservation have been identified (9, 10, 28, 29), but few have included different [ie taxonomic, phylogenetic, and functional (15, 21)] dimensions of mammal biodiversity simultancously, and’ few have used a complementariy-based prioritization framework globally (30 33), Further, al prioritization analyses for mammals incorporate taxonomic, phylogenetic, and functional information by using alpha diversity indexes (15, 21, 23). Selecting priority areas for conservation based on high alpha diversity could omit some specieslineagesitraits, because alpha diversity indices donot take into account the differences in composition among sites (2). Complementarity is derived from beta diversity and is considered a core principle of systematic conservation planning to ensure that conservation actions are directed to all species, not just to those occusting in species-rich hotspots (7, 34). Complementarity-based selection of priosty areas allows the representation ofall biological features without much duplica- tion of sites with similar species, resulting in a more cost- effective conservation solution (12). For example, Strecker ct al (22) performed the frst complementanty-based prionti- zation, considering separately taxonomic, phylogenetic, and ait diversity in freshwater fish, highlighting the use of ths approach for the optimal allocation of limited conservation resources that incorporates multiple dimensions of biodiversity. ere we present a spatial conservation prioritization for terres- ‘rial mammals using taxonomic, phylogenetic, and trait information and a complementarty-based approach to ste selection. Our aims were (i) 0 identify and compare priority regions for global ter- restrial mammal conservation, based on taxonomic, phylogenetic, and trait dimensions and (j) to evaluate how those priority areas relate to threatened species’ distribution and existing protected areas. Our work on the global prioritization of mammalian bio diversity across the dimensions can be used as a biological layer from which priority areas can be identified and subsequently combined with political, social, and economic considerations (eg. ref. 31), Results and Discussion Priority Areas for Global Mammal Conservation Across the Three Dimensions of Biodiversity. We found that important areas for mammal conservation had strikingly diferent spatial patterns for the three dimensions of biodiversity (Figs. 1 and 2), For the top 17% of sites with the highest conservation values, priority areas based on taxonomic dimension were spatially scattered, whereas the phylogeny and tait-based priority areas were moze contig- sous (Fig. 1) for maps with all values see Fig. Si). Similarly, Carwardine et al. (35) and Dobrovolski et al. (2) found a scattered patter of global priority areas for mammals, and Albuguergue et al. (36) also found the same diffused priority areas for mammals, birds, and amphibians. The areas we iden- tified by using the taxonomic dimension were likely spatially scattered because of the prioritization algorithm we used, The core-area algorthen reduces biological loss by retaining the core of each species’ geographic range, even when they occur in species-poor regions, and it favors species with restricted ranges inthe final solution mn our study, priority areas based on phylogeny and traits were generally similar, but there were important differences as well Taxonomic dimension Phylogenetic dimension Fig. 1. The top 17% of cel selected according to zonation prioritization based onthe taxonomic dimension, phylogenetic dimension, ait dimension and ‘on the overlap across the three imensions.n che overlap map, ces in which all tvee dimensions overlap are shown in dark green, these in which ‘wo dimensions overlap are shown in light green, and those selected only by one dimension are shown in orange. 7682 | ww pnasorlegldov.107Spnas. 1706461114Overlap among the top 17% priority areas,across the three dimensions of biodiveristy [itcimersion [lJ 2 dimensions lf 2 dimensions Fig.2._ Venn diagram showing the proportion ofthe land surface where we ‘an observe the overlap between the top 17% priory areas acrox the ‘aronomic phylogenetic, and wait dmenrons of biodversty, referent t0 ‘the overlap map presented in ig. 1 For example, only 1.8% of the land was selected aban area of poy for both the taxonomic and Walt dimensions ‘The color scheme that ured inthe overlap map in Fg For example, parts of Australia were more valuable in the phylogeny- ‘based solution than in the tait-based solution. Almost hal of the native land mammals from Australia are either monotremes or ‘marsupials, lineages that diverged carly in mammalian evolution and have unique evolutionary history, making them phyloge- netically distinct, However, the ecological traits of Australian ‘marsupials are functionally similar to those of placentals and are classic examples of ecological convergence (317). The preponderance HI Existing protected areas HE overiap prionties [HB overiap priorities already protected € - Fig and areas of overlap that are already protected (orange). frum ea of these distinct lineages may explain why Australia had more areas of importance when we considesed priorities for the phy- logenetic dimension rather than trait dimension, and the results suggest we are successfully capturing different facets of biodiversity in our prioritization analyses, ‘The overlap of important areas across the three dimensions of biodiversity was low, only 4.6% of the global land area (Figs. 1 and 2). Many of the overlapping regions have been identified previously as important areas for conservation because of their high species richness and number of threatened and restricted. range species, including mammals and other vertebrates, inver tebrates, and trees (10, 31-33, 36, 38). These areas also were included in the global priority map for the expansion of pro- tected areas to achieve the goals of 17% of global protected land and triple the average protection of vertebrate species’ ranges (G9). Further, some of the areas that we identify here as im- portant for all dimensions of mammal biodiversity are alzeady recognized as Biodiversity Hotspots (40) and High Biodiversity Wildemess Areas (41). For example, the Tropical Andes in the Neotropics, Madagascar, Sundaland and Indo-Burma regions in Indo-Malay realm, and the forest of eastern Australia are bio- diversity hotspots because of their high vulnerability and itt placeability (40). Likewise, the Amazon and New Guinea are identified by Conservation International as High-Biodiversity Wilderness Arcas because of their mostly intact original vegetation cover and high species richness and endemism (41). We found that these areas are important for all three dimensions of biodiversity and that they harbor not only high species richness and endemism. ‘but also unique evolutionary lineages and distinct ecological traits, underscoring the importance of protecting these regions. ‘The conservation values according to our prioritization anal ysis for each of the different dimensions were correlated (Table 1 land Fig. 52), but correlation values were small, and the corre Iation plots did not show clear trends, again indicating the low congruence of conservation priorities among the three dimen- sions of mammal biodiversity. The conservation values were also positively correlated with species richness and with the number Of threatened species, showing that places with higher conser vation values forall dimensions usually had higher species richness (Table 1 and Fig. $3), and consistently harbored a greater number of threatened species (Table 1 and Fig. $4). The spatial mismatch among the taxonomic, phylogenetic, and functional components 3. The overlap across areas important forthe taxonomic, phylogenetic, and tat dimensions (brown), the current network of protected areas (green), PNAS | July 98,2017 | vol 114 | n0.29 | 7685 reoloey sustamanury ScuNCETable 1. Kendall rank correlation coefficients (x) between the conserva siven by the spatial prioritization analysis for each pair of dimension of biodiversity (species, phylogeny, and traits) and with species richness and the number of threatened mammal Conservation priority Number of Taxonomic Phylogenetic —Trait_=—=—Species threatened dimension dimension dimension richness species Conservation priory for 1.00 030, 025 034 oan taxonomic dimension Conservation priority for 030 1.00 o43 046 o37 phylogenetic dimension Conservation prorty for 025 043 1.00 050 034 teat dimension ‘All correlation coefficients were igifcan for P< 0.00%, ‘of biodiversity has been observed globally for birds (26), fresh- water fishes (22), and mammals (15). Such lack of congruence Iriggers a conservation conflict, because conservation strategies based only on taxonomic diversity do not include important phy- logenetic and functional hotspots (15, 26). Priotity Areas and Currently Protected Areas. Currently protected areas performed poorly in terms of protecting the species, phy logenetic diversity, trait dimensions of mammal biodiversity and. also in protecting threatened species (Table 2). The average conservation values of cells currently under protection were significantly lower than would be expected by randomly selecting areas around the globe, and this finding was consistent for each fof the three dimensions and for all three dimensions jointly (Fig, 'S5). Similarly, Guilhaumon et al, (27) found that the Mediter~ ranean marine protected areas system did not harbor more taxonomic diversity, phylogenetic diversity, or functional di- versity of coastal fish than would be expected by chance, Fenkins ct al. (38) showed that location of protected areas in the United ‘States contrasts with regions where high numbers of endemic and threatened species aze located, leaving unique species un- protected, Some of the priority areas highlighted by Jenkins eta, (G8) concur with those that we identify as important across the thyce dimensions of mammal biodiversity, especially in the western United States. Several studies (eg, refs. 42H) have shown that the global network of protected ateas is ineffective and is biased toward residual locations, these places are con- sidered cheap to protect because they ate remote or of low ‘economic value, making it easier to set them aside for protection (7, 44, 45). Our findings combined with these other studies at local and regional scales demonstrate that currently protected areas are not maximizing the coverage of biodiversity. Of the 4.6% of the atea that we indicated as having priority for conservation across all dimensions of mammal biodiversity Table 2 Percentage of global total ef species richness, phylogenetic diversity (14) trait (consensus map, Fig. 3), only 1% is currently protected, leaving. ‘vulnerable 3.6% of the most important areas for all three di- mensions of mammal diversity conservation, Considering that ~14% of the Earth’s land area is already protected (46), the 3.6% of unprotected land that we identified hexe as key areas for ‘mammal conservation could inform efforts to expand the system cf currently protected areas to achieve the Aichi Target 1. However, setting aside areas for conservation incurs socioeco- nomic and politcal costs, such as land acquisition value, social conflicts, and political willingness (7, 35). Therefore, the priority areas provide a key biodiversity layer for future conservation planning that should be considered along with the priority areas of other taxonomic groups and economic, social, and politcal considerations (1, 12, 35). They also highlight biologically impor- tant, unprotected regions of the globe that can be used to guide future conservation planning at both regional and global scales. Conclusion We show that priority areas for taxonomic, phylogenetic, and functional dimensions of biodiversity are strongly spatially mis- matched, suggesting that conservation planning based in only ‘one dimension of biodiversity could undermine the conservation of the other two, Effective conservation planning should main- tain taxonomic, phylogenetic, and functional diversity to ensure biodiversity persistence in a changing world (47). Here, we identify a set of areas that are high conservation priorities across all three dimensions; this information can be used to inform conservation actions to expand the current network of protected areas to achieve the 17% target proposed by the Convention on Biological Diversity strategic plan for 2011-2020. The current system of protected areas poorly protects species richness, phy- logenetic diversity, functional diversity, and threatened species the important areas that we identified could perform better for versity (20), and ‘number of threatened species that are retained within the top 17% of the important regions for each dimension (taxonomic, phylogenetic, and functional) and across all three dimensions (averlap priorities) from the zonation prioritization analyses and in currently protected areas Zonation prioritization Taxonemic —_Phylogeni curently protectes % of the global total dimension, % almension, % Trait dimension, % Overlap priorities areas, % Species richness 99 8 7 70 @ Phylogenetic diversity 99 20 a a0 4 Trait diversity 99 ot 80 0 7” Threatened species 99 n 56 s7 27 7646 | www prasoriegldovt0.107Ipnas. 706961114all these aspects of biodiversity. The small fraction of areas (4.6%) where there was overlap across the dimensions should receive special conservation attention, Those areas retain many restricted-range species that have distinct evolutionary history and unique traits. Our work presents a global effort to identify important arces for terrestrial mammal conservation across the taxonomic, phylogenetic, and functional dimensions of biodiversity using a complementarity-based analysis. The key areas identified here can be used as a biological foundation for future conservation planning, which also would have to account for other factors such as opportunity costs, social/political con- siderations, and funding for land purchase, Conserving bio- diversity beyond species identity (13, 48, 49) is crucial to ensure the provision of ecosystem services and their contribution to human well-being, the evolutionary potential for species to evolve and adapt, and the extraordinary diversity that exists across mammalian lineages. Methods Occurrence Dats, We used maps of “extent of suitable habitat" that were generated based on species" ranges and habitat preferences (50) 9 ca {ulate mammal species occurrence. We aggregated the occurrence infor Imation into datbution maps (oresencelabience) of 1° > 1° (110 110 km 2 the equator) and identified priority regions for the mammal taxe- omic dimension using geographic sistribution maps for 4547 terres tral mammal “Trae. We compiled specs level traits databace for 454 cerestia mammals (17, 51-54. From a total of 23 ats avaiable, we used 14 invinsicblogical tras inthe analyes We chose these tras bared on ecological importance, ‘thecorrelation among them andthe percentage of ming values. Traits were related to resource use (activity ace, habiat mode, trophic level, det breadth) speed of life history fooay mas iter sie, ters pe ear, gestion lensth weaning age, neanateboay mass, maximum if spar), ana population characterises cal group size, population density. Because removing spe {ies for which some trait data ae fckng can cause statistical bias an te {erpretation eror (55), we imputed missing Walt values Using misForest [Package misFores in 8 (S62 nonparametric approach basea on ranom fore, Ths method performs well on large databases with correlated vari. ables (56, 57. Phylogeny was not used in the imputation analysis. Because Some ofthe tats had high proportions of data missing, we ran che priori zation analyse removing tat ith more than 60% of ming data 33 o- bbustnestchedk We ound that removing ait that had high propotion: af rising data betoce imputation dd not change our rain resus To reprsent the ait dimension nthe provtization analyses, we followed ‘the framework of Strecker etal. 22), wing a grid cel x eraits mati To eteate this matrix ll Walt variables were converted inte binary format. For fategorial tats we asigned the pesenceabsence of each category. We spit quanceative eras Into 5% quanties and then converted them Into binary variables. We choose to ue St to incorporate mare trait varailty incur analyte) Once our datatet was converted to inary formst, we cre ed a binary species x traits matrix. By mulpying the species > wats and Species x gr calls (obtained rom accrrence dats) matin, we obtained & ‘raltx grid cells matrix in which each 1" 1” grid cell contained the number of species exhibiting @ vat value (eg, the number of nocturnal species in ‘hat ce, Then we generated a diioution map for each trait value nd Used those map in the prirtizaion analysis to find the priority regions for amma ta almenson, Phylogeny We used phylogenetic eigenvectors to represent the phylogenetic dimension in our analyses, thereby avoiding nonindependence sues a5o- Ciated with phylogenetic ees (SB). We used an interpolated, smoothed Phylogenetic tre of mammals 68) to obtain a phylogenetic estance matrix mong all species. Then we sjthesized the phylogenetic information in tlgenvectors by conducting 2 prindpal coordinate analysis (PCOA) based on phylogenetic astances between specs (PVR package In R es. 5,60). E= Genvecors from a phylogenetic datence matin reflect the different phy: logenetic relationships among specs in independent vectors (60. The fst eigenvectors tend te represent larger stances among species, expressing divergences doter tothe root of the phylogeny (60), and sbiequent «= genvecors tend to capture phylogenetic relationships case tothe terminal odes. For each species, we generated multiple eigenvector scores that represent relatedness of exch species to all other species a diferent phylogenetic levels. We used only axes that presented eigenvalues larger than 1%, because axes with eigenvalues less than 1% contain only a smal fraction of the total variation ofthe whole phylogeny, and we wanted to ‘void Including ow-epresentative axes Inthe analy. From 4.348 phy ‘genetic axes generated by the PCOA, we onl used 16 eigenvectors in our Snalyies. These eigenvectors contained 634 of the total variation in the phylogenetic distance mati. We ako tested whether out results were Eee f0 the numberof eigenvector by uring 16, 10, 200, and 250 «= ‘genveciors. Because the revuls ganerated by the differen sets were highly Conelated, the inclusion of more elgenvectors appeared not to provide any Elgniicant new information in the analyses ‘We pt the fist 16 phylogenetic eigenvectors into 5% quantiles, as we dd 20 samesze phylogenetic groups in which specs were grouped based on ‘her phylogenetic aint ina given phylogenetic level, Then we multiplied the binary matrixofspeces x phylogenetic groups bythe gtd cel specie matic resulting in amavix of ite x ylegenetic groups in which eah 1" ‘fd call contained the number of Species belonging to particular phyo- ‘genetic group Next, we generated a ditroution map foreach phylogenetic ‘roup and used those maps in the piriizaion analyst fad the prionty Felons forthe mammal phylogenetic dimension. Priorzation Analyses. Ve identted the important areas for mammal con- servation aos the dimensions of biedversity using Zonation (3). Zonation produces hierarchical priorzation of the sud) region based on the bi ‘logical value of ste (cell), accounting fr complementary by considering the representation level ofall species (or other biodvesty features). Zo- nation Ieratively removes calls wihese removal causes the smallest loss in Biodiversity reprerentatin across the verall remaining region wnt no cll left n the region. The hertenical pritization of the region i based on the order of cll removal, which is recorded and can be used late to selec. any siven top faction (eg, best 10%) of the region. Ths order of call femoval'saled “conservation value” and ranges frm 0 1, with O being ‘the first cll removed (Le, the least important for retaining biodiversity representation) and 1 being the fst cell removed from the region (the ‘most important “The basic cellremoval ul the Core-Ares Zanation(CAZ) algorithm. The {cAZ algrthm clelates the conservation values of eath cell based on the ‘marginale (te relative contrition to total dveriy) ofthe species phylogenetic grouptat value witn he higher proportion ofits rangein that {IL CAZ prirtizes sits by gathering a higher proportion of each dimension disrioution, thus favering rae specesphylogenetigroupstratvaluesin the final solution, even wien they acer in otherwise species poor regions. We analyzed each dimension separately because we wanted to evaluate the inaivigual elution generated by each mension and te determine how much they converge Falewing the Convention on Biologicl Diversity (3), which propoted that 17% of the terertlal areas should be protected by 2029, we focuree our analyser on the 17% of the worl with the higher conservation value Ge, cel with conservation value greater than of equal t0 0.83) In [addtion, we provide the ful results in Fi. 5 Protected Aras. To compare the prioity areas across the three dimensions laith curently protected areas, we Used the Intermatinal Union for the Conservation of Nature (UCN) and United Nations Envitonmental Pro gramme World Conservation Monitoring Center {UNEPAVCMC) data on slobal protected areas (62). We inchided only restvcted protected areas {lasted as HIV bythe IUCN In our analyses, We resampled protected areas atthe 11 grid cal level ‘ensons and between each dimension and spetes ness and the number (oF threatened species in two Ways {By evaluating how the conservation ‘Values generate in our prortization anahsitrelated between derions land Ui) by calculating how much of the global ammount of taxonomicphy- Togenetctait diversity was aptured by each spatal priory. Fist we cal Cilated the Kendal rank corelation to evalvate how the conservation Priority values of the cals fr the priaritzaton bated on each dimension {texonomiclphylogenetckraid related to each other and to species chert ‘ato the numberof threatened species, Then to evaluate hav much the Spatial priorities of each elmenson captured the exstingtaxonome func Alona, and phylogenetic diversity of terrestrial mammals, we calculated the Global amount of species richness, the number of threatened mammals, phylogenetic eiversity (he PD index) (1), and funcional aiversity (he FD inde) (20) Lat, we assessed how much of the global otal ofeach index was captured proportionally by the conrervation solution for each dimension ANAS | Huy 18,2017 | vel 114 | 0.29 | 7685 reoloey sustamanury ScuNCEoO) and by the overlapping areas among all dimensions. We also performed the ‘Fame calculation to determine how much of the global total of species ‘icine, numberof threatened mammal, phylogenetic der, and fine tional diversity are protected by the currely protected area network. The analyses were performed in R wing the "stats" package (63) fr correlation nals and the “picante” package (68) to calculate phylogenetic ang flnctional avery indexes, ACKNOWLEDGMENTS, We thank Antonin Machae, Ben Weinstein, and Brune Oliver for helpful discussions ané the wo anonymous reewere for their valuable suggestions to mprove this paper. Our research vat funded by she National Science Foundation Dimensions of dediversty Pro- (gam Grant DES-TT36SE6. We thank Bruce Young and Tom Brooks, cls n ‘Sie Sorte hens aeons a 2 Feith an ables coentenpente nanene vom testo ert ding Ban he ita a 0) aw se cane oS snc crepe er Sonia be pl crannies Cael on fhe ty eso etn 9 ee 27. Gammon tet 00 Remsen nora pyr and co dost 2, Robiges ea B000 lob gro Sa: my vgn expan he as, Sipps tana the ste tesa and mane manna: Deity, ve Dittece ret Ia hoe prot rgb mel exc 21, Boral hojl 2, Galhaunanf Gua SF, Dhicho JAF 013) blag ‘asl xparon an carves crverton apron. Bol Corer 1512-78 22 Dobrovelit Loyal a Forcea Sas, Onizriha J, Araujo Ma 2018) lb 7645 | www pnasorgiegldovt0.107Ipnas. 1706961114 the NSF DEB-1136586 grant that helped to support this research. FB. was upported by 3 Cooraiation forthe Improvement of Higher Eeucaton Per- onnel (CAPES) doctorate scholarshp, received CAPES Doctorate Scnlarshp Ex 7910/13-4 fo vsk CH.G's laboratory at Stony Brook University, and rently hole 2 postdoctral scholarship from © Conselho Nacional de Detenvalvimenta Cientiiea ® Tecnalogio (CNPa), Grant 15217220165. RR was supported by cha Grants 20852220167, 4799582013-7, 47094 2013, and 5036212016 and by 0 Botiario Group Foundation for Nature Protection Grant PROG_0008_7013. Ths papers convibution of the stat Nacional de Ciencia e Tecnologia in Ecology, Evolution, and Biodivers Conservation founded by Ministerio da Ciencia, Tecnologia, Invagoes ‘= ComunicagaevthPg and Finda de Amparo Pecquia do Estado de Coit {Grane 465610720145) VCR reerved funding fom the Terra and Aqua Pro- stam af the National Aeronautics and Space Administration dence 18.01 Minn Et a 201) Cla res fr rain carnare conseration der 2, erga ea 247 Lining beta destypatrst protected ree: Leone 3. Comarine| et a QOD! Costetfecve rote fo gobl mammal onseratin 1, Mouqserue eer? (15 cleat ues and envnmental eee this a, Farger Vander Wh Bo! See hessng rede ana mara Conergen 2, Jena Ch Van wa KS. orn Sl Secon 1 (203) US protected land isaten 39, Montene Pura, eta 2014 Global protected ares xpesioncompromied by 40, hye Mier RA itera Coa Fone GAB, Kent 208) Radverity 1, fatermer Re BoD res andes ceterton Prac at end 12, e Sout, tak GOT Protected ares an effective Bndveity contention 18, Nett sel 2055) Amphibian conser suerte 6, ea, G08) raging lobl protected aes expan for imple 2016 WNERNCME Cage UI 1, breney RL Cabeza M, as ME, coing RIM, Win KA 207} Coneiton pam 48 Foret Fetal 2007) Peering te evluorary potential f ors in Bevery 19, Histo Loyel D,Cancariso MV GOI) lbs and cleans logs steven a teresa mammal Weriving pares stn sel 5, Ronan #20") Global habit staity model of teresa maria 11 Jones KE at 0000 PanTHERIA A ipecessevel database of He Histon, a grography of extant and rece extn mamas Enlogy 308448 ‘2, Theca het 299) Human ageing genanic reaurces negra datas Sr 53, Pacis Met a. 203) Generation lrg fo amma. Nat Conse 57-8 Se Verde Aegis, Blonaerg SP Fete G0 fu) Phylogenet core f 55 Nakagawa 5, reletonR 2008 Ming incon The danger a igneing ming 5. Stthoven 6) tahnann G012 Misforet-oonparametremisng value inputs 157, Poona eta Gat luca of rising ast etn ta 58 Diiezha JA. de Sune CER, in LM (388) An egenvecar mathe fret trating lagen rein Evolution SI:247-122 6. Dh J ta. 0952) One sleton of plage gene: or eo (1 Mainen A tab 201) Zoran spat Cnveation aning Framework ard Sarma, Ur Maral, Helin Veron» Frogrorne 2018) Worl! baatae an Poteced Areas Susisbe st pen 163, R Development Core Team OIE) R.A language and envionment for satistal omutng I ounantion fr Satsteal Computing, Vine) 4, Kemal 3, eal B01) Pant Saale for inegratng phylogenies And wl,