SPECIAL FEATURE | ECOLOGY SUSTAINABILITY SCIENCE OPEN ACCESS Socio-ecological gap analysis to forecast species range contractions for conservation Nyeema C. Harrisa,1, Asia Murphyb, Aalayna R. Greenc meza , Siria Ga , Daniel M. Mwamidid, and Gabriela C. Nunez-Mire Edited by A. Michelle Lawing, Texas A&M University, College Station, TX; received March 18, 2022; accepted June 23, 2022 by Editorial Board Member Nils C. Stenseth Conservation requires both a needs assessment and prioritization scheme for planning and implementation. Range maps are critical for understanding and conserving biodi- Significance versity, but current range maps often omit content, negating important metrics of varia- tion in populations and places. Here, we integrate a myriad of conditions that are Geospatial approaches are spatially explicit across distributions of carnivores to identify gaps in capacity necessary increasingly vital for conservation for their conservation. Expanding on traditional gap analyses that focus almost exclu- with applications of gap analysis sively on quantifying discordance in protected area coverage across a species’ range, our informing decision-making and work aggregates threat layers (e.g., drought, human pressures) with resources layers resource allocation. We extend (e.g., protected areas, cultural diversity) to identify gaps in available conservation capac- traditional assessments by ity (ACC) across ranges for 91 African carnivores. Our model indicated that all species incorporating both the spatial have some portion of their range at risk of contraction, with an average of 15 percentage distribution of threats and range loss. We found that the ACC differed based on body size and taxonomy. Results resources to build an index of deviated from current perceptions of extinction risks for species with an International available conservation capacity Union for Conservation of Nature (IUCN) threat status of Least Concern and yielded insights for species categorized as Data Deficient. Our socio-ecological gap analysis across a species’ range. Using 91 presents a geospatial approach to inform decision-making and resource allocation in con- African carnivores, we identified servation. Ultimately, our work advances forecasting dynamics of species’ ranges that are locations within each species’ increasingly vital in an era of great socio-ecological change to mitigate human–wildlife range at risk of contraction due to conflict and promote inclusive carnivore conservation across geographies. a deficit of resources available to potentially thwart present threats. geospatial j Africa j carnivore j human pressure j protected area Our results raise new concerns for African carnivores, particularly Present species ranges reflect historic conditions and extant pressures coupled with biol- ogy and resource availability (1, 2). Despite an exhaustive focus on climate change small-bodied species, contrary to inducing range shifts, the heterogeneity of conditions within a species’ range is complex current perceptions of their and extends beyond environmental properties to delimit distributions and alter ecology extinction risks. (3, 4). Now, in an era of immense anthropogenic influence, social factors stemming from human values, interactions with nature, and land use are increasingly governing range dynamics and threatening the persistence of species (5, 6). Implicit in conserva- tion efforts to preserve biodiversity is the maintenance of individual and overlapping species ranges. The fluidity of spatial patterns of biodiversity results from those envi- ronmental, ecological, macroevolutionary, and social processes causing species’ ranges Author affiliations: aApplied Wildlife Ecology Lab, Yale School of the Environment, Yale University, New Haven, CT to shift with specific concern for contraction at varying temporal scales (7). Thus, 06511; bDepartment of Environmental Studies, University building approaches and tools that aid in forecasting range dynamics remains a funda- of California, Santa Cruz, CA 95064; cDepartment of Natural Resources and the Environment, Cornell mental tenet of conservation biogeography research, necessitating the amalgamation of University, Ithaca, NY 14853; dLaboratory for Analysis of vast amounts of data for mapping (Fig. 1). Socio-ecological Systems in a Global World, Institute of Environmental Science and Technology, Autonomous Geospatial analytical approaches are now integral to conservation planning and University of Barcelona, 08193 Barcelona, Spain; and e decision-making (8). Applied applications of mapping have extended beyond species Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607 presence to identify areas critical for natural processes, research, and conservation investment (9–11). Areas classified as highly vulnerable or “gaps” in protection help prioritize resource allocation. Traditional gap analysis at the species level overlay the Author contributions: N.C.H. designed research; N.C.H., A.M., A.R.G., S.G., D.M.M., and G.C.N.-M. performed distribution of protected areas with the distribution of a species range or their available research; G.C.N.-M. contributed geo-processing analytic habitat to quantify spatial discordance in coverage (12, 13). Such gap analysis tools; N.C.H., A.M., and G.C.N.-M. analyzed data; N.C.H. wrote the paper and all authors provided edits; N.C.H. approaches have been applied to a range of species and varying ecosystems to guide solicited and managed the team; and A.M., A.R.G., S.G., conservation efforts, e.g. seahorses (14), cacti (15), and rivers (16). Species or ecosystem and D.M.M. compiled data. representation and human pressures within existing protected areas provide another The authors declare no competing interest. lens for conservation assessments (17–19). Coad et al. (20) reported that only 4 to 9% This article is a PNAS Direct Submission. A.M.L. is a guest editor invited by the Editorial Board. of terrestrial vertebrates are adequately represented within existing global protected Copyright © 2023 the Author(s). Published by PNAS. areas. Ahmadi et al. (21) coupled gap analysis with network analysis using graph theory This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives to consider the spatial configuration of conservation areas for large felids in Iran. These License 4.0 (CC BY-NC-ND). approaches still largely omit considering how the distribution of threats affect the vul- 1 To whom correspondence may be addressed. Email: nerability of species to extirpation and range contraction, although a gap analysis for nyeema.harris@yale.edu. giant panda (Ailuropoda melanoleuca) extended beyond only protected area coverage to Published February 6, 2023. PNAS 2023 Vol. 120 No. 7 e2201942119 https://doi.org/10.1073/pnas.2201942119 1 of 9 Fig. 1. Species ranges vary in space, time, and in various characteristics. Temporal dynamics highlight potential changes in the amount and location occu- pied. Ranges may remain stable (no change), contract, expand, or shift. Regardless of the range outcome, the heterogeneous distribution of resources and threats contribute and compromise respectively to influence the available conservation capacity (ACC) of a species. disparities in management considering timber extraction and stressor is not accurate. This is especially true given that the effec- other human disturbances in China (22). tiveness of biodiversity conservation projects is directly contingent Land-use changes, including the expansion of urbanization upon local and regional sociopolitical dimensions including gov- and agricultural production, have and will continue to alter the ernance (35). Therefore, conservation strategies in response to ecology, evolution, behavior, and distribution of species (23, socio-ecological changes must then adequately incorporate these 24). Consequently, understanding the occurrence of these socio-ecological dimensions. threats across a species’ range could facilitate forecasting which Many carnivores face global uncertainty and operate in populations may be vulnerable to habitat loss and where con- dynamic socio-ecological processes that underscore their impor- flicts will emerge between conservation and development agen- tance (36, 37). Carnivores connect systems through a myriad of das. For example, Strona et al. (25) considered projected oil pathways including their regulatory capacity in shaping eco- palm expansion to assess range loss of primates across Africa. As systems, charisma that contributes to gross domestic product captured in the human footprint index (HFI), threats are not through tourism, and complex direct and indirect human inter- perturbing species in isolation but instead can operate synergis- actions (38). These species also often have wide distributions tically across scales (26, 27). Mapping HFI or other threats is that intersect varying land management regimes that increase particularly promising for species of conservation concern to encounters with people and our domesticates. Africa contains a understand current and future risks (28, 29). As pressures third of the world’s carnivore species that persist in an environ- mount, anticipating where contractions occur in species’ ranges ment fraught with anthropogenic pressures and rich biocultural via population extirpations can aid in devising effective mitiga- diversity (39). Africans have coexisted with carnivores for tion strategies. However, the singular focus on the distribution millennia and exhibit intricate relationships with carnivores, as of threats, or alternatively protected areas for gap analysis, is evident from the ubiquitous inclusion of carnivores in oral tra- insufficient and misrepresents actual conservation needs. Such ditions and folktales (Box 1). For example, Africa’s iconic lions approaches exclude how various resources available throughout (Panthera leo) serve as tokens of power, strength, and fierceness. a species’ range can influence their probability of persistence. As such, body parts of lions are used by tribal traditional healers Areas rich in biodiversity often have improved ecosystem func- for various spiritual and bodily ailments. The Xhosha of South tion and possess certain safeguards to prevent collapse (30). Africa, Zimbabweans, the Samburu of Kenya, and the Gogo of Recent work also highlights the contribution of indigenous Tanzania all use lion fat as a potion to ward off wicked spirits, land management to habitat and species conservation through exorcize evil spirits, repel creditors from debtors, and give cour- their stewardship practices (e.g., ref. 31). Wildlife provide rec- age, respectively (40). As such, advancing any conservation reation services, provision food and materials, and generate rev- agenda for African carnivores requires knowing and engaging enue that contributes to national economies that motivate Africa’s indigenous communities as key partners across the spe- their conservation (32, 33). Despite the pervasive narrative of cies’ range. Here, we propose an innovative socio-ecological gap degradative anthropic factors, humans are also (though not analysis that includes both threat and resource components to equally) contributing to conservation efforts, and considering determine if existing capacities are sufficient to promote stable geographies with higher cultural diversity may enhance the capac- geographic ranges for African carnivores (Fig. 1). ity for conservation. For example, customary laws and traditional Our model incorporates the distribution of habitat, biodiver- ceremonies of the Nharira community in central Zimbabwe sity, and protected area along with cultural diversity as resources include biodiversity protection (34). Thus, the blanket inclusion that inform conservation capacity across the geographic ranges of of human density as inherently and exclusively an environmental 91 African carnivores. African species in the order Carnivora 2 of 9 https://doi.org/10.1073/pnas.2201942119 pnas.org resources that reflect social, environmental, and ecological fac- Box 1. tors at two scales, individual grid cells within the range (ACCi) and the entire range (ACCj) for a species (see Eq. 1). All maps are presented in the supplementary materials (41). Our final list of African carnivores with available range maps from the Inter- national Union for Conservation of Nature (IUCN) included 91 species with only extant ranges used for analyses (table S1 in ref. 41). Range sizes (mean = 748,920 km2, SE = 101,645) and body sizes (mean = 7.95 kg; SE = 1.99) varied as well as current IUCN threat status among these carnivores with 21 species classified as Endangered (EN) or Vulnerable (VU). While the majority (n = 58, 64%) of African carnivores are classified as Least Concern (LC), range shifts are still plausible for these species in future. Africa’s rich biocultural diversity induces complex interactions The distribution of threats and available resources that con- between nature and local people. In Ethiopia, communities vary in tribute to species persistence were highly variable among Afri- their relationship with spotted hyenas (C. crocuta), with some viewing can carnivores and throughout their ranges (table S2 and figure them as abnormal and perceived agents of evil while others have a S2 in ref. 41). We aggregated five geospatial layers of human spiritual connection (65). Carnivore pelts provide resources to support modifications as well as drought and hunting pressures to inves- ornamentation for rituals and medicinal practices for African commu- tigate these threats across carnivore distributions throughout nities. Traditional cultures in Nigeria use leopard skin to treat snake Africa (table S3 in ref. 41). We found risks from drought and bites and convulsion and believe the skin, teeth, and hair of the spot- hunting pressure were the most prevalent threat, as their nor- ted hyena can combat adultery and promiscuity (66). In Tanzania, malized mean values (meanNV) were the highest across carni- otters’ parts are believed to have medicinal and witchcraft properties, vore ranges on average. The different modifications induced while some west and central African communities believe carrying otter fur renders them invisible to their enemies (67). The Mbuti from human land use were more limited across the distribution people of Democratic Republic of Congo use skins of civets and gen- of African carnivores. For 8% of carnivores including the ets for ornamental and medicinal purposes (68). Similar traditional endangered Ethiopian wolf Canis simensis, agriculture was a cultural uses have also been reported in local communities in Camer- widespread threat across their distributions with meanNV rang- oon, where skins are sold in the markets as love charms and have dec- ing from 0.483 to 0.782. However, we also found that metrics orative, spiritual, and medicinal uses (69). Overall, as market value of biocultural diversity, such as mammal, bird, and amphibian increases for pelts and use shifts from small-scale village level to more diversity (B_DIV) and ethnic diversity (C_DIV), were wide- commercial purposes, concerns of overexploitation mount for carni- spread resources across carnivore ranges. Specifically, C_DIV vore species in these regions. comprised the highest meanNV compared to other resources for 12% of African carnivores including two EN species (Eupleres major and Mungotictis decemlineata). For 11% of the species in range in body size from ∼250 g (dwarf mongoose) to over our analysis, habitat was also a widely distributed resource that 200 kg (African lion). African carnivores also vary across ecologi- contributed to portions of carnivore ranges being stable. cal attributes including sociality, arboreality, nocturnality, and Our capacity gap analysis revealed several potential outcomes dietary breadth that influence their extinction risks. Geographic with varying implications for conservation of African carnivores ranges for these species are wide, small, contracting, and frag- (figure S1 in ref. 41). On average, 81% of carnivore ranges mented. Such diversity of species characteristics makes African were stable due to the surplus of resources compared to threats carnivores a useful model to build and test the conceptual frame- across ranges compared to 15% of carnivore ranges having a work of a capacity gap analysis in conservation planning. Concep- deficiency that could lead to contractions across IUCN threat tually, species with ranges composed of smaller capacity gaps (i.e., categories based on their ACCj value. We also found propor- more available conservation capacity, ACC) should be less threat- tions of African carnivore ranges where threats and resources ened with extinction. For carnivore conservation to be effective were comparably low (mean = 3% of ranges; range: 0 to 16%), in an era of extreme environmental changes resources must be subject to rapid transitions that could induce range contrac- pooled across disciplines that enhance traditional gap analysis. tions. Interestingly, a small proportion of African carnivore We use the availability of these resources compared to the distri- ranges were classified as uncertain due to comparably high bution of threats including hunting pressure, human modifica- amounts of threats and resources in specific areas of their distri- tion, and drought to determine capacity scenarios that enable bution (mean = 1.3% of ranges; range: 0 to 4%). As such, forecasting locations within a species’ range at risk of contraction cooperation between natural and social agendas becomes essen- (figure S1 in ref. 41). Simply put, areas across a species’ distribu- tial in these locations across a species’ range. Ultimately, fore- tion where threats outweigh resources are at immediate risk of casting range dynamics informs expectations of spatial patterns contraction. We argue that only when both the distribution of in biodiversity as well as conservation prioritization and strate- threats and the distribution of resources are simultaneously con- gies at the individual species level. sidered can one identify where there are gaps in capacity for miti- Perceptions of extinction risks guide conservation efforts. Our gation. Ultimately, assessing the ACC across a species’ range is approach considers the heterogeneity of threats and resources necessary to both combat threats and anticipate range dynamics. throughout a species’ range to forecast range dynamics. Species var- ied widely in the ACCi across their ranges (Fig. 2). Additionally, Results our results represent an independent validation of current knowl- edge on the conservation status of African carnivores (Fig. 3). We assessed the ACC for 91 African carnivore species across Three carnivores in our dataset were Data Deficient under the eight families based on the coupled distribution of threats and IUCN classification: Dologale dybowskii, Genetta abyssinica, and PNAS 2023 Vol. 120 No. 7 e2201942119 https://doi.org/10.1073/pnas.2201942119 3 of 9 megalotis) currently exists within two disjoint ranges, one in eastern Africa (Kenya and Tanzania) and one in southern Africa (Namibia, Botswana, South Africa), and has 25% of its current distribution at-risk of contraction. Overall, we found the majority of Least Concern species (98%) had portions of their range vulnerable to contraction with six species having over one-third of their distributions in Africa at risk. We further investigated whether species characteristics affected the ACCj index value for species using structural equa- tion modeling. Specifically, we included body size, range size, IUCN threat status, and population trend as well as carnivore family. We found a significant positive relationship for body size (standardized β = 0.22; P = 0.0274), indicating that larger carnivores have more conservation capacity. We also found a taxonomic signature with varied relationships among carnivore families with a positive effect of Eupleridae, the Malagasy carni- vore guild (standardized β = 0.57; P < 0.001), on ACCj. Inter- estingly, Malagasy carnivores also had significant positive and negative indirect paths to IUCN threat status and population trend, respectively. Contrary to expectations though, ACCj was not significantly influenced by the current threat status or pop- ulation trends (figure S3 in ref. 41). Discussion Despite advancements in geospatial analysis for conservation, we developed a socio-ecological gap analysis that enhances tra- ditional approaches to anticipate how ranges of African carni- vores may change in the future due to the distribution of ACC. We explicitly incorporate resources (e.g., protected areas) and threats (e.g., drought pressure) to better understand how extir- pation risk might vary spatially and temporally across the geo- graphic ranges of 91 African carnivores. We found that the most common threats across species were drought and hunting pressures, while the biocultural diversity metric was a valuable resource for many species. Although an average of 81% of a species’ geographic range was classified as stable due to a sur- plus of resources compared to threats, we found that, on aver- age, 15% of a species’ range was at risk of contraction due to burgeoning threats. In addition, we found that many species Fig. 2. Percentage of a species’ range classified by ACC outcomes: stable, uncertain, dynamic, or contraction (left to right) for African carnivores that are classified as Least Concern under the IUCN threat across IUCN threat categories. schema had the highest percentages of their range at risk of contraction. Our conceptual framework expanded approaches to assess conservation status and anticipate range dynamics Genetta poensis. All three of these species are small-bodied with based on an ACC index. Additionally, species with higher relatively small geographic ranges (78,960 km2, 83,225 km2, and ACC may serve as an umbrella, broadening effects of single- 21,070 km2, respectively). Our results indicated which threats species conservation efforts. were persistent across their ranges: drought and agricultural modi- The IUCN Red List (42) remains the premier source of fication for D. dybowskii, drought and hunting pressure for information of conservation status along with the Living Planet G. abyssinica, and only hunting pressure for G. poensis. While Index with time-series abundance data to guide prioritization immediate concerns may not exist for G. poensis based on the efforts (43). However, we did not find strong evidence of align- ACCj index, our results raise some concern for the other two ment between ACC and IUCN conservation status, specifically Data Deficient African carnivores, as 7% of both ranges are at in that carnivore species categorized as Least Concern did not risk of contraction. A range loss of 7% for a species with a small, have low risk of range contractions and those categorized as isolated geographic range could have major consequences for their Endangered did not have the highest ACC. While most African future persistence as well as major socio-ecological ramifications. carnivores are classified as Least Concern, they occur in regions Most concerning are species with high-capacity gaps (i.e., undergoing rapid environmental and social change that may negative ACCi) throughout considerable portions of their range threaten the persistence for specific populations. Notably, these (Fig. 2). In some cases, these species are currently classified results highlight that our current prioritization of species status accordingly with IUCN threat classification of Endangered. For may be masked by conspicuous local abundance of species that example, Grandidier’s vontsira (Galidictis grandidieri) and the are still vulnerable to extirpation. Additionally, prior investiga- Ethiopian wolf have small, restricted ranges occupying a single tions of range dynamics highlighted concerns of range contrac- country with 19% and 33% of their ranges, respectively, at risk tions among large carnivores (44). Despite several iconic large of contraction. However, the LC bat-eared fox (Otocyon carnivores of Africa including the African lion and spotted 4 of 9 https://doi.org/10.1073/pnas.2201942119 pnas.org Fig. 3. Exemplar results for African species with the distribution of ACC outcomes to forecast range dynamics. EN and LC denotes Endangered and Least Concern species according to the IUCN Redlist, respectively. (Photo source: iNaturalist; photo credit: Felis nigripes – South Africa, Niall Perrins; Ictonyx striatus – Malawi, markusgmeiner; Crossarchus platycephalus – Central African Republic, Rod Cassidy; Vulpes zerda – Western Sahara, Martina Milanese; Helogale purvula – Kenya, Zarek Cockar; Canis mesomelas – South Africa, jandutoit). hyena (Crocuta crocuta), the carnivore community of Africa is leopard (Panthera pardus) and cheetah (Acinonyx jubatus) popula- composed mainly of a diverse assortment of small carnivores. tions (48); according to results of our model these populations We found that smaller carnivores had less ACC. These small are not at risk of extirpation. The distribution of African cultures carnivores perform vital ecosystem services, support local com- explicitly represents opportunities for enhanced capacity, a munities, and disproportionately receive less funding and con- type of social capital, that could contribute to conservation. His- servation attention compared to large carnivores (45). torically, designated government protected areas and national Traditional range maps limit decision-making for conservation parks have been an extension of Eurocentric values and colonial because of omitted information (3). However, knowing a place conservation practices, representing a barrier to traditional ecolog- requires knowing its inhabitants and their interactions; cultural ical knowledge and practice (49). Additionally, positioning diversity and linguistic diversity is often coupled with biodiversity the nations of Africa as assets may help reduce potentially exacer- (46, 47). Culturally relevant maps that incorporate people’s activ- bating environmental injustices and social inequities resulting ities and values in places can also enhance collaboration and effec- from traditionally exclusionary conservation interventions. tiveness of conservation efforts. For example, some carnivore Geography remains critically important to conservation species serve as totems of reverence and power within many planning with continued advancement of quantitative and African communities, which governs their behavior and level of modeling approaches (50). However, we first acknowledge exploitation (Box 1). Though extractive practices are viewed by that results of any analysis reliant on range maps requires vali- Western, European-centric values as challenges for conservation, dation before local-level interventions can occur for species a more inclusive ideology would recognize such use indicates val- conservation. Our work relied exclusively on IUCN range ues and opportunities for cooperative management. This is espe- maps to assess conservation capacity in our socio-ecological cially true when sustainability is embedded within practices, as gap analysis, but we recognize these maps are fraught with evident by Daasanach elders in Kenya reporting reuse of carni- inconsistencies and errors such as occurrence records far vore skins in successive cultural ceremonies so as to not harm beyond the species’ actual range or omission of known PNAS 2023 Vol. 120 No. 7 e2201942119 https://doi.org/10.1073/pnas.2201942119 5 of 9 occurrence points (51). Another persistent challenge in our simply through a single disruptive lens of anthropogenic pressures analysis is reconciling the scale of divergent socio-ecological but rather acknowledging that cultural value and heritage of Afri- data layers. In our case, relevant information both in relation can peoples can fundamentally contribute to species conservation to spatially explicit threats and resources were omitted because amid extractive customs. As such, by seeking, incorporating, and of scale. On one end, the Environmental Justice Atlas (https:// respecting the traditional ecological knowledge of people in ejatlas.org/) provides a global representation of environmental places, conservation can progress to a more inclusive practice and activities that are often extractive and interact with human protect species’ ranges under varying global change scenarios. well-being that cause social conflict. These point data are not presented at a scale that reflects the geographic footprint of Materials and Methods activities and reporting varies, causing discrepancies in inter- Datasets. We obtained a species list from the IUCN Red List of extant terrestrial pretation, despite shared consequences for wildlife that would African carnivores excluding Otariidae and Phocidae species. Our query originally inform vulnerability risks for African carnivores. However, yielded 105 species that were then verified through a literature search and some of the oil and gas mining coverage is captured within the occurrence records in the Global Biodiversity Information Facility (GBIF) database. human modification dataset we used in our analysis. We also We removed Canis lupus and Canis aureus because of limited potential range in attest to not presenting a comprehensive model of all pressures Africa and unresolved taxonomy of African wolves in North Africa (56). We also on the continent of Africa threatening species, which could be excluded Viverricula indica as no range map was available through IUCN. Our expanded in subsequent analysis to enhance forecasting range final list of carnivores with available IUCN range maps included 91 species across dynamics. On the other end, much relevant resource information eight families with only extant ranges used for analyses. The dataset comprised such as conservation spending or environmental governance that 21 species with higher risk of extirpation or extinction based on current assess- could inform capacity requires finer spatial scale datasets instead ments that warranted classifications of Endangered (EN) or Vulnerable (VU). See of being constrained to the country level (52, 53). Additionally, table S1 in ref. 41 for the full species list with associated traits. distinct from government protected areas, the recognition of All spatial data were downloaded and compiled from open-access platforms, indigenous lands in conservation efforts underscores an immedi- publications, and government geospatial data repositories (table S3 in ref. 41). ate need to develop this geospatial layer across scales (31). Threat layers included human modification, drought, and hunting pressure. The Another limitation in our approach involves excluding the drought data were reported as a Standardized Evapotranspiration Index with val- temporal patterns in resources or threats that would alter the ues ranging from 2 to 2. Human modification to terrestrial ecosystems spatial distribution of conservation capacity. Conservation included spatially explicit datasets on five major anthropogenic stressors: urban/ maps are often static representations of dynamic, interacting built-up infrastructure; agricultural production and timber harvest; energy pro- processes (54). We did not project changes in ACC and instead duction and mining; transportation and service corridors; and human intrusion, focused on building the conceptual framework to categorize natural system modification, and pollution. These data were extracted for each current contributions across ranges. Our approach also explic- stressor group for Africa from 2017 global datasets at 1,000-m resolution (57). Given the vast uncertainty in climate projection models for Africa, we did not itly identifies locations of spatial refugia across a species’ range. include climate change per se as a threat variable. Instead, we used a standard- Populations in these locations are more likely to be stable and ized precipitation evapotranspiration index as a metric of drought for Africa in potentially serve as source populations in the maintenance of 2016 at 1-mo temporal resolution and 5-km spatial resolution (58). From Zhu genetic and phenotypic diversity. However, the ACC index we et al. (59) global model of hunting-gathering populations of present-day climate, developed could be used for back-casting and forecasting his- we extracted hunting time as an index of hunting pressure across Africa at spatial toric and future ranges, respectively. For example, examining resolution of 2° latitude. The estimate of hunting time was calculated in part historic ranges could inform whether contractions could have from information on population density, animal biomass availability, daily dry been predicted by applying the ACC index at distinct time matter intake of meat, and energy reserves based on consumption the previous periods in the past. Results from our model yielded insights day. However, the model omits considerations of small animals such as birds, into anticipating where ranges may contract to target interven- reptiles, insects, and small mammals that would underestimate food availability tions and identify partners necessary to promote extirpation and estimates of meat in the diet. avoidance. Resource layers included habitat, protected area, biodiversity, and cultural diversity (table S3 in ref. 41). Habitat was derived from the European Space Conclusion Agency CCI Land cover 2020 classification at 300-m resolution. To capture the breadth of habitat requirements and conditions suitable for generalist carnivore Overall, incorporating the “texture” in the form of ACC across a species, we identified all land-cover types that could represent suitable habitat species’ range highlighted the spatial consequences of heterogene- for these species. We then combined these land-cover types into six broader cat- ity in the distribution and intensity of threats as well as the distri- egories: forest, grassland, shrub, mixed, sparse, and flooded vegetation all as bution and availability of resources. Because species’ ranges and habitat with individual values of 1 (table S4 in ref. 41). Protected Area (PA) cover- the conditions encompassed within them are dynamic, an effec- age was obtained from the World Database of Protected Areas as polygons for tive strategy must devise spatially explicit and temporally relevant protected areas and community managed areas (version February 2022). Biolog- actions that incorporate various knowledge systems and informa- ical diversity (B_DIV) was calculated as species richness for mammals, birds tion types. We demonstrated the importance of threats being excluding seabirds, and amphibians at 10-km resolution from 2018 using IUCN contrasted with resources to assess capacity and conclude the and Birdlife International range maps (https://biodiversitymapping.org/). Cultural extent of risks to prioritize interventions. Our approach presents and ethnic diversity (C_DIV) was calculated as ethnic group richness throughout an elegantly simple strategy to forecast future range dynamics. Africa (60). Individual polygons from this dataset could represent up to three eth- Using African carnivores as the model system, we further demon- nic groups and data occur at 7.5-km resolution. While these data were compiled strated the complexity of ecological, environmental, and social originally for application in understanding and anticipating civil conflict, we repurpose for a positive application as cultural beliefs can aid in conservation factors that challenge the persistence of biodiversity in Africa. efforts (55). Therefore, capturing the representation of different ethnicities could Conservation planning must acknowledge the interconnectedness reflect opportunities that enhance capacity. of human and natural systems as well as stress the necessity of social–ecological approaches in conservation practice (55). We Geoprocessing Spatial Data. Because the spatial data obtained for threat and employed an interdisciplinary approach by including people not resource variables varied widely in format, resolution, and spatial projection, we 6 of 9 https://doi.org/10.1073/pnas.2201942119 pnas.org Fig. 4. Workflow diagram of geoprocessing for spatial threat and resource layers to create the ACC across a species’ range. completed several preprocessing steps prior to analysis that depended on the variable, although future analysis could scale particular variables based on their format of the data (Fig. 4). Data stored as polygons (e.g., PA) were processed to ecological importance for a given species or group of species, if this information be represented in a numerical raster format, specifying the cell size of the output is known. to be 5 km2. Data on the location and settlement patterns of C_DIV were also stored in polygon vector format and therefore converted into raster files with the Sj ACCj = number of distinct ethnic groups whose settlements overlap the output cell rang- nj  nj     [1] ing 0 to 5. Habitat data were processed similarly to C_DIV in that the output ras- ∑ Rij ∑ Tij Sj = ∑   100 ter displayed the number of suitable habitats (i.e., forest, grassland, shrub, i=1 xij yij mixed, sparse and flooded vegetation; table S4 in ref. 41). We did not create separate land-cover variables for each classification for a binary classification of ACCj represents the global level as the total capacity gap for species j where R is each habitat type to avoid zero inflation and instead aggregated for each 5-km2 the sum of normalized resources values and T is the sum of normalized threat cell ranging from 0 to 6. values across n locations of a species’ geographic range. Because all resource The remaining data resources and threats were contained in numerical raster and threat variables may not be present at each location and to make all varia- files at varying spatial resolutions. All raster files containing continuous data bles that were present weighted equally, we divided R and T by xij and yij repre- were resampled to a spatial resolution of 5 km2 through bilinear interpolation of senting the number of resources and threats included, respectively. ACCi were cells. Raster files containing discrete numerical data were resampled to the same mapped for each 5-km2 grid cell across the species range. Positive values indi- spatial resolution (5 km2) using the nearest-neighbor method. Bilinear interpola- cate a surplus of available resources that presumably can combat threats across tion is the preferred method for resampling raster files of continuous data, as it landscapes, while negative values signal a deficit of resources and raise concerns mitigates the “staircase effect” and blockiness that results from using the for the local persistence of species. ACCi values that resulted in differences nearest-neighbor method when resampling continuous data. On the other hand, between resources and threats of 0.05). We expected that if con- servation status reflects extirpation risks, then conservation status will be a signifi- Analysis. We assessed the relative contribution of a species’ IUCN threat status cant predictor of the ACC index. on ACCj compared to other explanatory variables including range size, taxonomic Data, Materials, and Software Availability. All data sources and products family, body size, and IUCN population trends using structure equation models (SEM) in the piecewiseSEM package (62). SEMs unite multiple predictor and from analyses are available in the supporting information including individual response variables into a causal network, often visualized through path dia- species ACC maps that have been deposited in Dryad (https://datadryad.org/ grams. Geographic range size was estimated as the number of cells in the spe- stash/share/ZamtKCx8Q7XsrLsfBo21yvGAddTUKOQ aco46QwjmxBM) (41). cies geographic range layer taken from IUCN and average body weight derived ACKNOWLEDGMENTS. We express our deepest appreciation to the Special from Myhrvold et al. (63). Range, average body size, and ACC were all continu- Issue organizers, Drs. J. McGuire, M. Lawing, S. Diaz, and N.C. Stenseth for all ous variables. IUCN status of Least Concern (1), Data Deficient (2), Near Threat- their effort and coordination. We thank G. Gadsden for assistance with Box 1 ened (3), Vulnerable (4), and Endangered (5) as well as population trends of graphic creation. We also salute the many African communities and their cultural Decreasing (1), Unknown (2), Stable (3), and Increasing (4) were ordered cate- diversities that have contributed to preservation of carnivores across the conti- gorical variables. We found no indication of multicollinearity using the variance nent for millennia. 1. J. Bridle, A. Hoffmann, Understanding the biology of species’ ranges: When and how does 32. G. M. Mace, K. Norris, A. H. Fitter, Biodiversity and ecosystem services: A multilayered relationship. evolution change the rules of ecological engagement? Philos. Trans. R. Soc. Lond. B Biol. Sci. 377, Trends Ecol. Evol. 27, 19–26 (2012). 20210027 (2022). 33. F. Isbell et al., Linking the influence and dependence of people on biodiversity across scales. 2. E. Polaina, M. Gonzalez-Suarez, E. Revilla, The legacy of past human land use in current patterns of Nature 546, 65–72 (2017). mammal distribution. Ecography 42, 1623–1635 (2019). 34. E. Mavhura, S. Mushure, Forest and wildlife resource-conservation efforts based on indigenous 3. N. C. Harris, S. Gamez, G. I. Gadsden, R. Malhotra, Textured species range maps enhance knowledge: The case of Nharira community in Chikomba district, Zimbabwe. For. Policy Econ. 105, interdisciplinary science capacity across scales. Front. Ecol. Environ. 20, 319–326 (2022). 83–90 (2019). 4. A. Johnston et al., Comparing abundance distributions and range maps in spatial conservation 35. D. R. Magness et al., Management foundations for navigating ecological transformation by planning for migratory species. Ecol. Appl. 30, e02058 (2020). resisting, accepting, or directing social–ecological change. Bioscience 72, 30–44 (2022). 5. M. Di Marco, L. Santini, Human pressures predict species’ geographic range size better than 36. W. J. Ripple et al., Status and ecological effects of the world’s largest carnivores. Science 343, biological traits. Glob. Change Biol. 21, 2169–2178 (2015). 1241484 (2014). 6. A. M. C. Santos et al., Current climate, but also long-term climate changes and human impacts, 37. J. T. Bruskotter et al., Modernization, risk, and conservation of the world’s largest carnivores. determine the geographic distribution of European mammal diversity. Glob. Ecol. Biogeogr. 29, Bioscience 67, 646–655 (2017). 1758–1769 (2020). 38. J. Lozano et al., Human-carnivore relations: A systematic review. Biol. Conserv. 237, 480–492 (2019). 7. M. Pacifici et al., Global correlates of range contractions and expansions in terrestrial mammals. 39. IUCN, The IUCN Red List of Threatened Species, Version 2021-3. https://www.iucnredlist.org Nat. Commun. 11, 2840 (2020). Accessed 16 January 2022. 8. D. Zurell et al., Spatially explicit models for decision-making in animal conservation and 40. N. Sifuna, The future of traditional customary uses of wildlife in modern Africa: A case study of restoration. Ecography 41, 1–16 (2021). Kenya and Botswana. Adv. Anthropol. 2, 31–38 (2012). 9. N. Loiseau et al., Global distribution and conservation status of ecologically rare mammal and bird 41. N.C. Harris et al., Textured range maps based on available conservation capacity (ACC) for 91 species. Nat. Commun. 11, 5071 (2020). African carnivores. Dryad. https://datadryad.org/stash/share/ 10. A. Zizka, C. D. Barratt, C. D. Ritter, T. Joerger-Hickfang, V. M. A. Zizka, Existing approaches and future ZamtKCx8Q7XsrLsfBo21yvGAddTUKOQaco 46QwjmxBM. Deposited May 31, 2022. directions to link macroecology, macroevolution and conservation prioritization. Ecography. 42. A. S. L. Rodrigues, J. D. Pilgrim, J. F. Lamoreux, M. Hoffmann, T. M. Brooks, The value of the IUCN 11. S. Andrefou€et, O. Bionaz, Lessons from a global remote sensing mapping project. A review of the Red List for conservation. Trends Ecol. Evol. 21, 71–76 (2006). impact of the Millennium Coral Reef Mapping Project for science and management. Sci. Total 43. J. Loh et al., The Living Planet Index: Using species population time series to track trends in Environ. 776, 145987 (2021). biodiversity. Philos. Trans. R. Soc. Lond. B Biol. Sci. 360, 289–295 (2005). 12. A. S. L. Rodrigues et al., Global gap analysis: Priority regions for expanding the global protected- 44. C. Wolf, W. J. Ripple, Range contractions of the world’s large carnivores. R. Soc. Open Sci. 4, area network. Bioscience 54, 1092–1100 (2004). 170052 (2017). 13. O. Venter et al., Targeting global protected area expansion for imperiled biodiversity. PLoS Biol. 45. C. Marneweck et al., Shining the spotlight on small mammalian carnivores: Global status and 12, e1001891 (2014). threats. Biol. Conserv. 255, 109005 (2021). 14. X. Zhang, A. C. J. Vincent, Conservation prioritization for seahorses (Hippocampus spp.) at broad 46. L. J. Gorenflo, S. Romaine, R. A. Mittermeier, K. Walker-Painemilla, Co-occurrence of linguistic and spatial scales considering socioeconomic costs. Biol. Conserv. 235, 79–88 (2019). biological diversity in biodiversity hotspots and high biodiversity wilderness areas. Proc. Natl. Acad. 15. B. Goettsch, A. P. Duran, K. J. Gaston, Global gap analysis of cactus species and priority sites for Sci. U.S.A. 109, 8032–8037 (2012). their conservation. Conserv. Biol. 33, 369–376 (2019). 47. B. R. Tershy, K.-W. Shen, K. M. Newton, N. D. Holmes, D. A. Croll, The importance of islands for the 16. J. Major, D. Perry, C. Aslan, R. McManamay, Identifying gaps in protected areas to expand protection of biological and linguistic diversity. Bioscience 65, 592–597 (2015). integrated riverine ecosystem conservation. Conserv. Sci. Pract. 3, e470 (2021). 48. D. M. Mwamidi et al., Contemporary pastoral commons in East Africa as OECMs: A case study from 17. S. H. M. Butchart et al., Shortfalls and solutions for meeting national and global conservation area the Daasanach community. Parks 24, 79–88 (2018). targets. Conserv. Lett. 8, 329–337 (2015). 49. P. Kashwan, R. V. Duffy, F. Masse, A. P. Asiyanbi, E. Marijnen, From racialized neocolonial global 18. K. R. Jones et al., One-third of global protected land is under intense human pressure. Science conservation to an inclusive and regenerative conservation. Environment 63, 4–19 (2021). 360, 788–791 (2018). 50. E. Di Minin, R. A. Correia, T. Toivonen, Quantitative conservation geography. Trends Ecol. Evol. 37, 19. N. Politi, L. Rivera, S. Martinuzzi, V. C. Radeloff, A. M. Pidgeon, Conservation prioritization when 42–52 (2022). species distribution data are scarce. Landsc. Urban Plan. 210, 104067 (2021). 51. C. J. Marsh et al., Expert range maps of global mammal distributions armonized to three 20. L. Coad et al., Widespread shortfalls in protected area resourcing undermine efforts to conserve taxonomic authorities. J. Biogeogr. 49, 979–992 (2022). biodiversity. Front. Ecol. Environ. 17, 259–264 (2019). 52. A. Waldron et al., Reductions in global biodiversity loss predicted from conservation spending. 21. M. Ahmadi et al., Species and space: A combined gap analysis to guide management planning of Nature 551, 364–367 (2017). conservation areas. Landsc. Ecol. 35, 1505–1517 (2020). 53. M. A. Titley, S. H. M. Butchart, V. R. Jones, M. J. Whittingham, S. G. Willis, Global inequities and 22. B. Yang et al., Gap analysis of giant panda conservation as an example for planning china’s political borders challenge nature conservation under climate change. Proc. Natl. Acad. Sci. U.S.A. national park system. Curr. Biol. 30, 1287–1291.e2 (2020). 118, e2011204118 (2021). 23. R. P. Powers, W. Jetz, Global habitat loss and extinction risk of terrestrial vertebrates under future 54. W. L. Geary et al., Threat webs: Reframing the co-occurrence and interactions of threats to land-use-change scenarios. Nat. Clim. Chang. 9, 323–329 (2019). biodiversity. J. Appl. Ecol., 10.1111/1365-2664.13427 (2019). 24. E. C. Ellis et al., People have shaped most of terrestrial nature for at least 12,000 years. Proc. Natl. 55. R. M. Niemiec et al., Integrating social science into conservation planning. Biol. Conserv. 262, Acad. Sci. U.S.A. 118, e2023483118 (2021). 109298 (2021). 25. G. Strona et al., Small room for compromise between oil palm cultivation and primate conservation 56. M. Krofel et al., Towards resolving taxonomic uncertainties in wolf, dog and jackal lineages of in Africa. Proc. Natl. Acad. Sci. U.S.A. 115, 8811–8816 (2018). Africa, Eurasia and Australasia. J. Zool., 10.1111/jzo.12946 (2021). 26. O. Venter et al., Sixteen years of change in the global terrestrial human footprint and implications 57. C. M. Kennedy, J. R. Oakleaf, D. M. Theobald, S. Baruch-Mordo, J. Kiesecker, Managing the for biodiversity conservation. Nat. Commun. 7, 12558 (2016). middle: A shift in conservation priorities based on the global human modification gradient. Glob. 27. K. Kimmel, M. Clark, D. Tilman, Impact of multiple small and persistent threats on extinction risk. Change Biol. 25, 811–826 (2019). Conserv Biol., 10.1111/cobi.13901 (2022). 58. J. Peng et al., A pan-African high-resolution drought index dataset. Earth Syst. Sci. Data 12, 28. V. J. Tulloch et al., Why do we map threats? Linking threat mapping with actions to make better 753–769 (2020). conservation decisions. Front. Ecol. Environ. 13, 91–99 (2015). 59. D. Zhu, E. D. Galbraith, V. Reyes-Garcıa, P. Ciais, Global hunter-gatherer population densities 29. M. Di Marco, O. Venter, H. P. Possingham, J. E. M. Watson, Changes in human footprint drive constrained by influence of seasonality on diet composition. Nat. Ecol. Evol. 5, 1536–1545 (2021). changes in species extinction risk. Nat. Commun. 9, 4621 (2018). 60. N. B. Weidmann, J. K. Rød, L.-E. Cederman, Representing ethnic groups in space: A new dataset. 30. T. H. Oliver et al., Biodiversity and resilience of ecosystem functions. Trends Ecol. Evol. 30, 673–684 J. Peace Res. 47, 491–499 (2010). (2015). 61. A. C. Bovik, The Essential Guide to Image Processing (Academic Press, 2009). 31. C. J. O’Bryan et al., The importance of Indigenous Peoples’ lands for the conservation of terrestrial 62. J. S. Lefcheck, piecewiseSEM: Piecewise structural equation modelling in r for ecology, evolution, mammals. Conserv. Biol. 35, 1002–1008 (2021). and systematics. Methods Ecol. Evol. 7, 573–579 (2016). 8 of 9 https://doi.org/10.1073/pnas.2201942119 pnas.org 63. N. P. Myhrvold et al., An amniote life-history database to perform comparative analyses 67. D. W. De Luca, N. E. Mpunga, Small Carnivores of the Mt Rungwe-Kitulo Landscape, with birds, mammals, and reptiles. Ecology 96, 3109–3109 (2015). Southwest Tanzania: Presence, Distributions and Threats (Small Carnivore Conservation, 2013), 64. J. Fox, S. Weisberg, An R Companion to Applied Regression (SAGE Publications, 2018). pp. 67–82. 65. J. K. Young, D. L. Coppock, J. A. Baggio, K. A. Rood, G. Yirga, Linking human 68. F. Maisels, E. Keming, M. Kemei, C. Toh, The extirpation of large mammals and implications for perceptions and spotted hyena behavior in urban areas of Ethiopia. Animals (Basel) 10, 2400 montane forest conservation: The case of the Kilum-Ijim Forest, North-west Province, Cameroon. (2020). Oryx 35, 322–331 (2001). 66. S. K. Timothy, D. U. Habib, A. E. Ayodeji, Survey of zoological materials used in traditional 69. K. S. Bobo, F. F. M. Aghomo, B. C. Ntumwel, Wildlife use and the role of taboos in the conservation medicine in Sabon Gari and Zaria local government areas, Kaduna state, Nigeria. J. Complement. of wildlife around the Nkwende Hills Forest Reserve; South-west Cameroon. J. Ethnobiol. Med. Res. 8, 32–39 (2018). Ethnomed. 11, 2 (2015). PNAS 2023 Vol. 120 No. 7 e2201942119 https://doi.org/10.1073/pnas.2201942119 9 of 9