Protected areas are now the last strongholds for many imperiled mammal species

The global network of terrestrial protected areas (PAs) has experienced a fourfold expansion since the 1970s. Yet, there is increasing debate around the role of the global PA estate in covering and sustaining threatened species, with serious ramifications for current PA financing and the setting of post‐2020 global conservation targets. By comparing “past” (1970s) and current distribution range of 237 mammals, and measuring the proportion of range covered by PAs in the past and in the present, we show that a small number of PAs have now become the last bastions of hope for ensuring the persistence of many mammal species. For 187 species (∼79% of those analyzed) the proportion of range covered by PAs has doubled over the time period, with 10% of all species now having most of their current range protected. This increase in proportional protection over time is largely due to a retreat of species distribution (outside existing PAs) and, in smaller part, to PA expansion. It is clear that adequately resourcing those PAs critical in sustaining mammal species is now essential, to avert a worldwide rapid mammal loss.


INTRODUCTION
A fourfold expansion in the global network of terrestrial protected areas (PAs) has occurred since the 1970s, and today PAs cover ∼15% of the terrestrial planet (UNEP-WCMC & IUCN, 2016). Recent works have shown the number of threatened vertebrates included in PAs is increasing as a consequence of an expanding PA estate (Venter et al., 2014) and there is evidence that some PAs have been effective at maintaining, or even enhancing wildlife populations (Geldmann et al., 2013;Walston, Stokes, & Hedges, 2016). For example, species like the tiger (Panthera tigris) and lion (Panthera leo) have clearly benefited from the use This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Conservation Letters published by Wiley Periodicals LLC of fences and antipoaching measures employed within PAs (Geldmann et al., 2013;Karanth, Nichols, Kumar, & Hines, 2006). Yet, there is increasing conjecture around the role of the global PA estate in covering and sustaining imperiled species, as PA coverage is biased toward locations that are less expensive to protect and sometimes of limited importance for biodiversity (Visconti et al., 2019;Venter et al., 2014).
Signatory nations of the Convention of Biological Diversity (CBD) are now focusing on generating a new, post-2020 PA target that includes a further increase of global PA coverage (Convention on Biological Diversity, 2020). But the uncertainty of how effective PAs have been in saving imperiled species has serious ramifications for current PA financing (Watson, Dudley, Segan, & Hockings, 2014), and is likely to impact international debate on the best post-2020 strategy for PAs (Visconti et al., 2019). As such, assessments around the effectiveness of the PA estate to conserve biodiversity are increasingly important for policy discussions and financing (Coad et al., 2019).
We collected data on past (1970s) distribution ranges for a total of 237 mammal species (Table S1) by combining a public database on past distribution maps for terrestrial non-volant mammals (Pacifici et al., 2019) with literature-based information on bat distributions. We measured species' range changes, and the past and present overlay of the ranges with PAs at different spatial resolutions for sensitivity testing (see Supporting Methods). Due to the 50-year timeframe considered, the changes in distribution we observed are related to relatively recent human activities (Pacifici et al., 2020) and may underestimate historical contractions in species' ranges.
Our results show that many imperiled species have now become almost completely dependent on PAs for their persistence. We chose mammals because they are a wellstudied taxon which has highly suffered from human impact in recent years despite having a crucial role in providing and maintaining services and functions associated with sustaining balanced ecosystems (Di Marco, Venter, Possingham, & Watson, 2018;Pacifici et al., 2020). Our analysis represents all major taxonomic mammal orders and biogeographic regions, with the exception of the Palearctic region for which we have a relatively lower number of species ( Figure S1). Although large-bodied orders are usually better represented in mammal distribution analyses as they are more studied (Monsarrat & Kerley, 2018), we improved the Pacifici et al. (2019) databasewhich did not include volant mammals-so as to balance the number of large charismatic species and small lessknown mammals and obtain more generalizable results ( Figure S2, Table S4). Small mammals (< 3 kg) represent 53% of the species in our database, although our data might underrepresent those species that live in very small areas and for which local extinctions are not realistically contemplated

P R O T E C T E D A R E A S A R E E S S E N T I A L F O R M A N Y I M P E R I L E D M A M M A L S
There have been clear increases in PA coverage for 224 (i.e., 94%) of the mammal species in our analysis ( Figure 1a, Table S2), with an average 12% additional coverage since 1970s. For 187 species (∼79%) the proportion of range covered by PAs has at least doubled in the last 50 years, with 10% of all species now having more than half of their distribution range protected (Figures 1a, 2). This pattern of increased protection relative to range size was especially true in the tropics ( Figure S3), with some areas, like the Sunda Islands and Central-Eastern Africa, standing out for the high number of charismatic species surviving now almost exclusively in PAs ( Figure 2).
Importantly, the majority of species (n = 147, 62%) experienced contractions of their distribution range alongside an increase in percentage PA cover (Table S2). Largebodied mammals (here defined as those species with body mass >20 kg, following MacDonald & Willis, 2013) are the most significant examples of the trend of net loss of distribution range and proportional increase in PA coverage (Figure 1b, c). The increased in percentage coverage of PAs was only due to a net increase in the extent of PAs within the range of the species (i.e., no range reduction) in 33% of cases, while for 53% of species, the increase in proportional protection over time is almost entirely due to a net decline of their distribution outside existing PAs (Table S2). For example, ∼80% of the range of the Indian rhinoceros (Rhinoceros unicornis) is currently protected compared to 3% in the 1970s, after the species has lost a dramatic 99.7% of its distribution in just 50 years (Table S1); about 87% of the remaining individuals currently occur in just two PAs (Kaziranga National Park in India and Chitwan National Park in Nepal; Walston et al., 2016). Likewise, the percentage range protection of the Addax in Africa (Addax nasomaculatus) has increased from 3% in the 1970s up to 59% today, after a 97% range contraction ( Figure 2).
For 26 other mammal species, we found no reduction in species' distribution and an increase in range protection from PA expansion (Table S2). An example is the puma (Puma concolor), a widespread species that occurs from Canada to Chile. This species showed no significant changes in its distribution but a 17% increase in PA coverage (Table S2). Puma is a very adaptable species, found in every major habitat type of the Americas (Nielsen, Thompson, Kelly, & Lopez-Gonzalez, 2015), and it widely occurs in Brazil, the country where annual rate of land protection is the highest in the world (UNEP-WCMC et al., 2018). There were also a few (n = 14) species that had no change in percentage range protected, mostly located in eastern and Midwestern United States and Mexico ( Figure S1); these are some of the areas with the lowest PA expansion over the past two decades (UNEP-WCMC et al., 2018).

AN URGENT NEED TO IDENTIFY THOSE JEWELS IN THE PROTECTED AREA ESTATE
We did not set out to prove PAs are effective at preventing future extinctions, but to simply assess if (and why) mammal species are increasingly found in protected land and F I G U R E 1 Changes in protected area coverage within mammal ranges since the 1970s. (a) Comparison between current and past protected area coverage within species ranges divided by body mass and locomotion: large-bodied non-volant mammals, ≥20 kg, in red; medium-bodied nonvolant species, 3-15 kg, in orange; small-bodied nonvolant species, ≤3 kg, in black; bats in blue. (b) Past and present relationship between species' range size and protected area coverage; arrows direction goes from past to current data, with error bars representing 95% confidence intervals. (c) Trends of large-bodied species (arrows direction goes from past to current data) to explore the policy consequences of this. Our results are unequivocal, and supported also by a sensitivity analysis ( Figure S4)-there are clear increases in PA coverage for 94% of the mammal species, and for 79% of species the proportion of range covered by PAs has at least doubled in the last 50 years (Table S2). The most likely explanation is that PAs, especially when well resourced, managed, and wellplaced within the wider landscape, have every chance in halting the threatening processes causing mammal decline (Coad et al., 2019;Edgar, Stuart-Smith, Willis, & Nature, 2014;Watson et al., 2014). Another key aspect that emerged from our work is that when dealing with spatial analysis on species distribution and PAs, the resolution and resampling technique must be carefully selected. We found for those species with very small, fragmented distributions, simply resampling polygons at a coarser resolution may lead to a loss of ecological signal (Table S3, Figure S4).
There is growing body of species-specific evidence around why PAs are effective at maintaining mammal populations. For example, despite a disastrous 82% reduc-tion in global distribution (Table S1), the tiger population in Rajiv Gandhi (Nagarahole) National Park in India has increased by 400%, due to improved reserve management aimed at maintaining high prey densities (Karanth et al., 2006). Likewise, South-African lions subpopulations showed an overall 12% increase since 1993, with a peak of +1,400% in the well-budgeted Tembe Elephant Park (South Africa, Bauer et al., 2015), despite a global species decline of 43%.
In Latin America, the amount of range under protection of the white-lipped peccary (Tayassu pecari), a declining species classified as Vulnerable by the International Union for Conservation of Nature (IUCN), has increased sevenfold since the 1970s (Suárez et al., 2009; Figure 1c). In the Yasuni Biosphere Park and Reserve in Ecuador, the species has been the subject of an intensive commercial bushmeat trade in the last 3 decades, with peaks in the years 2005-2007(Suárez et al., 2009. A successful collaboration between TRAFFIC, the Ministry of Environment and local communities planned targeted actions for the F I G U R E 2 Top 20 species of mammals threatened by range reduction. The figure highlights species with the highest percentage range contraction since 1970s. Red circles represent species for which protected areas represent the last bastions (> 50% range is protected). Blue circles show species with very high rates of range decline, and relatively small portions of the current distribution protected (< 50%). In each big circle, percentage values at the top indicate range loss, while those at the bottom represent the amount of range currently protected. Protected areas are represented in green. Small dots represent the centroid of the range of the other species in the original sample. A full species list of the top 20 mammals threatened by range reduction (big circles) is provided in Table S1 different consumer groups, targeted to reducing the consumption of wild meat while providing alternative sources of protein to local populations (Phelps, Biggs, & Webb, 2016).
The saiga antelope (Saiga tatarica) lost > 70% of its range (Table S2) and is currently listed as Critically Endangered by the IUCN. Population estimates for Mongolia, where animals are mostly concentrated in two PAs (Sharga Nature Reserve and Mankhan Nature Reserve), suggest that the population is growing (Milner-Gulland, 2015), probably due to rigorous conservation efforts since 1993 when reserves have been established to protect most of the remaining distribution range of the species (IUCN SSC Antelope Specialist Group, 2018). What these examples show is identifying those PAs that play a vital role for highly threatened species is a priority: they should get appropriate levels of funding to ensure effective management and be upsized and upgraded if needed. Our analysis, while not claiming to be comprehensive across all threatened species, indicates that many PAs in East and South-East Asia and Sub-Saharan Africa are a priority for conservation investment, as they contain enormous percentages of threatened terrestrial mammals ranges. Efforts should be made to systematically identify PAs across every continent which play a disproportionally important role in maintaining species current distributions; this should be taken into account when threatened species belonging to any taxonomic group are strictly related to PAs management and effectiveness, as these PAs are essential to these species' long-term security.
For those international policy forums that deal with PAs, greater clarity must be made in specific protected area text that showcases the roles different PAs play in abating the conservation crisis. This is especially timely, given the world's governments will soon adopt a post-2020 biodiversity framework. This plan will likely contain biodiversity targets that define the global conservation agenda for at least the next decade and, while it is still unknown what the actual targets will be (and still the subject of much debate; Visconti et al., 2019), it is clear that PAs will play an important role in this wider conservation agenda. Our research provides more evidence that conservation policy development and implementation must be built on a platform that is clear on the overall goals of conservation (Visconti et al., 2019). This is because some PAs play more of a critically important role for mammal conservation than others, and these must be identified and targeted for resources. Therefore, policy makers must be clear on the different roles PAs will play in achieving these goals. By clearly specifying the different roles PAs play within the wider conservation agenda, there will be a shift in focus toward objective assessments on PAs through the world, focusing on which ones need more resourcing and what are the best ways to identify key funding gaps and priorities (Coad et al., 2019). At the same time, it will lead to more effective monitoring and reporting of PAs by nations, including states and trends of threatened species, which should underpin the adaptive management of PAs.

PROTECTED AREAS ARE NECESSARY BUT NOT SUFFICIENT TO HALT MAMMAL DECLINE
PAs are not the one-and-only panacea when it comes to effective conservation for imperiled mammals. The Earth surface has been drastically altered and fragmented by human activities, with natural lands usually converted into other land cover types (Haddad et al., 2015). This has led wildlife to be increasingly confined in evermore rare and sparse habitat patches, often only found in PAs, facing problems of isolation, increased vulnerability to stochastic events-for example, genetic drift, environmental and demographic stochasticity-and higher probability of local extinction due to the reduction of their populations (Newmark, 1995;Pacifici et al., 2020). For example, the Cape buffalo (Syncerus caffer caffer) populations inhabiting smaller isolated parks in Kenya and Uganda showed signs of genetic erosion, presumably because of the lower ecological carrying capacities of these reserves, hence restricting the effective population size of animals, and the small ratio between the size of the PA and the surrounding human-dominated landscape which reduces gene flow (Heller, Okello, & Siegismund, 2010).
Large carnivores isolated in small PAs have proven to be more vulnerable to human-induced mortality rather than to stochastic processes. For these mammals, conflicts with humans on reserve borders are the main cause of mortality, and species with relatively larger home ranges have a higher probability of going locally extinct (Woodroffe & Ginsberg, 1998). Conservation measures that aim only to combat stochastic events are therefore unlikely to avert extinction for these species with large area requirements. Instead, actions that seek to maximize PA size and connectivity between PAs are essential, as are efforts to mitigate carnivore persecution on PA borders and in buffer zones.

WITHOUT WELL-FUNDED PROTECTED AREAS, THE FUTURE LOOKS GRIM FOR MAMMALS
Our research provides clear evidence of dramatic range declines for many mammal species, with distribution shrinkage toward PAs. The rate at which species have disappeared from their past (1970s) range has largely surpassed the rate by which PAs have been created, pointing to the obvious fact that we are running out of time to save these species. As the global community comes together and debates new post-2020 conservation goals (where protected area targets and future financing will play a central role), it is vital that the role that those well-placed PAs play in ensuring many mammal species persist in the wild is formally recognized. As species like the African wild dog (Lycaon pictus), dhole (Cuon alpinus), lion, and tiger face continuing conflict with people even inside well cited PAs (Woodroffe & Ginsberg, 1998; respectively 14%, 3%, 12%, 4% of their range losses have been inside PAs; Table S2), it is critical that PAs must be well financed and managed. This is the only chance for highly threatened, exploited species to survive Many of the species we found retreating toward PAs are heavily affected by illegal trade, a market worth US $7-$23 billion a year (Nellemann et al., 2016). Mountain Nyalas (Tragelaphus buxtoni), for example, are widely persecuted for their meat and horns (Sillero-Zubiri, 2013) and today survive almost exclusively within PAs. Mountain Nyala is one of the few species in our list that lost vast portions of its range also within PAs (63%; Table S2). The situation is similar for white rhinoceros (Ceratotherium simum), black rhinoceros (Diceros bicornis), Indian rhinoceros, and Sumatran rhinoceros, whose horns are mainly used for traditional medicine and dagger handles in Asia (Biggs et al., 2013), with the vast majority of their current distributions (65-89%) being in PAs (Table S1). To ensure the long-term survival of mammals, global conservation policies must focus on securing those critically important PAs and, at the same time, reward those efforts that ensure reexpansion and recolonization of wildlife populations on territories beyond protected area boundaries. This means focusing on retaining Earth's remaining intact ecosystems that contain key PAs and prioritizing efforts to restore habitat corridors between isolated reserves that best provide opportunities movement and genetic exchange.

A C K N O W L E D G M E N T S
This study was funded by the ARC Centre of Excellence for Environmental Decisions. MDM acknowledges support from the EU H2020 programme (H2020-MSCA-IF-2017-793212).

AUTHORS' CONTRIBUTIONS
MP, MDM, and JEMW designed the study. MP analyzed the data. All authors discussed the results and wrote the paper.

D ATA A C C E S S I B I L I T Y S TAT E M E N T
Data sources for past ranges are available at https: //onlinelibrary.wiley.com/doi/10.1002/ecy.2747/suppinfo. IUCN current ranges are available upon request at https:// www.iucnredlist.org/. Current ranges not taken from the IUCN Red List for sensitive species are available from the corresponding author upon reasonable request.