Maximum temperatures determine the habitat affiliations of North American mammals

Edited by Nils Stenseth, Universitetet i Oslo, Oslo, Norway; received March 16, 2023; accepted October 14, 2023
December 4, 2023
120 (50) e2304411120

Significance

While many studies have explored the effects of climate and land-use change in isolation, few have investigated how they interactively affect mammal distributions. If climatic conditions dictate species’ capacities to persist in anthropogenic landscapes, increasing temperature and precipitation variability may impede efforts to conserve biodiversity in anthropogenic landscapes. We leveraged a continental-scale dataset to study variation in how North American mammals respond to forest cover and anthropogenic modification across their ranges. Species were consistently more sensitive to human modification, and more dependent on forests, in hotter regions. This suggests that decisions about enhancing the conservation value of anthropogenic landscapes versus protecting natural areas should consider how species’ habitat affiliations are likely to shift under a changing climate.

Abstract

Addressing the ongoing biodiversity crisis requires identifying the winners and losers of global change. Species are often categorized based on how they respond to habitat loss; for example, species restricted to natural environments, those that most often occur in anthropogenic habitats, and generalists that do well in both. However, species might switch habitat affiliations across time and space: an organism may venture into human-modified areas in benign regions but retreat into thermally buffered forested habitats in areas with high temperatures. Here, we apply community occupancy models to a large-scale camera trapping dataset with 29 mammal species distributed over 2,485 sites across the continental United States, to ask three questions. First, are species’ responses to forest and anthropogenic habitats consistent across continental scales? Second, do macroclimatic conditions explain spatial variation in species responses to land use? Third, can species traits elucidate which taxa are most likely to show climate-dependent habitat associations? We found that all species exhibited significant spatial variation in how they respond to land-use, tending to avoid anthropogenic areas and increasingly use forests in hotter regions. In the hottest regions, species occupancy was 50% higher in forested compared to open habitats, whereas in the coldest regions, the trend reversed. Larger species with larger ranges, herbivores, and primary predators were more likely to change their habitat affiliations than top predators, which consistently affiliated with high forest cover. Our findings suggest that climatic conditions influence species’ space-use and that maintaining forest cover can help protect mammals from warming climates.
Habitat conversion is reducing terrestrial biodiversity and transforming ecological communities worldwide (1, 2). Though many species are able to persist in the farms, towns, and production forests that typify modern working landscapes (3), the communities that occupy natural versus anthropogenic landscapes are often distinct (2). Much effort has been expended to quantify interspecific variation in species responses to habitat change and, in doing so, identify the “winners” and “losers” associated with ongoing global change (47). For example, classic studies placed North American birds and mammals into three categories based on the degree to which they are preadapted to persist in urban areas: urban adapters (i.e., species that now heavily utilize anthropogenic habitats), urban avoiders (i.e., those that are restricted to natural environments), and urban exploiters (i.e., habitat generalists that can exploit both anthropogenic and natural areas) (8, 9).
However, recent evidence suggests that species are not monolithic with respect to their tolerance of anthropogenic habitats. That is, populations of the same species may exhibit very distinct responses to anthropogenic land-use in different regions (10). For example, birds in the Atlantic Forests of Brazil appear to be more sensitive to deforestation near the edge of their ranges compared to their range centers (11). Thus, placing species into simple categories, such as avoiders, adapters, and exploiters, may mask substantial variation in how species respond to habitat conversion.
Though the intraspecific variation in species responses to habitat conversion is rarely quantified, many nonmutually exclusive mechanisms may cause different populations of the same species to use anthropogenic habitats in some regions but not others. First, populations may have adapted to anthropogenic landscapes in regions with a long history of anthropogenic modification but not in more recently impacted areas (1214). Second, some populations may persist in anthropogenic landscapes by regularly receiving colonists from nearby natural areas, i.e., via source-sink dynamics (15, 16). Third, anthropogenic landscapes vary across space, in terms of both intensity of development and the patterns of fragmentation, with some landscapes composed of heterogeneous mosaics of diversified farms and small settlements, where many organisms can thrive, versus others dominated by large cities and intensive agriculture, where biodiversity often plummets (3, 17).
Yet another possibility is that species’ climatic niches constrain their ability to use anthropogenic landscapes in some regions but not others (18, 19). When temperatures spike (or plummet), many species can behaviorally thermoregulate by seeking thermally buffered microclimates (20). The presence of suitable microclimates can therefore drive species’ distributions (21), allowing species to persist in areas that would otherwise regularly exceed their thermal tolerances. Critically, anthropogenic land-uses are often much more structurally homogeneous than natural areas, resulting in reduced thermal buffering capacity and fewer potential climate refugia (20, 2224). For example, converting shaded tropical forests to open agricultural fields can cause local temperatures to increase by over 10 °C relative to nearby shaded forests (25, 26). Habitat conversion can also alter moisture regimes, either causing declines in moisture availability, for example, by removing tree canopies and drying the understory, or increases in moisture availability through irrigation (27). Correspondingly, several studies suggest that species may switch their habitat affiliations along macroclimatic gradients. For example, Neotropical herpetofauna are more likely to occupy agriculture at higher elevations because low-elevation sites regularly experience much more severe local temperature spikes (28, 29). Similarly, tropical birds seem to shift their habitat affiliations along climate gradients, using anthropogenic habitats more often in wetter (30, 31) or more seasonal regions (32).
If climatic conditions dictate species’ capacities to persist in anthropogenic landscapes, then increasingly frequent and severe temperature spikes and precipitation variability (33) may impede efforts to conserve biodiversity in anthropogenic landscapes (3). At the same time, identifying contexts in which species may be able to use working landscapes—both now and with ongoing climate change—may help improve the efficacy of farmland biodiversity conservation programs (34). However, while many studies have explored the effects of climate and land-use in isolation, the interactive effects of climate and land-use change are less understood yet likely a primary driver of species distributions in the Anthropocene (26, 35, 36).
We leveraged a continental-scale mammal monitoring program to study intraspecific variation in terrestrial mammal responses to forest cover and human modification. Our work was guided by three questions. First, are species’ use of forest and anthropogenic habitats consistent across continental scales? We predicted that, although some species always avoid or affiliate with anthropogenic habitats, some will affiliate with anthropogenic habitats in some regions and natural habitats in others. Second, do macroclimatic conditions explain spatial variation in species responses to land-use? We predicted that mammals use forests more often—and anthropogenic habitats less often—in drier regions and in regions with hotter temperature because forests can thermally buffer the understory and retain moisture in dry areas (27). Finally, can species traits elucidate which species are most likely to show climate-dependent habitat associations? We predicted that climate-dependent habitat associations would be more common in a) herbivores and primary predators because top predators may be especially persecuted by humans and thus sensitive to human modification everywhere (37), b) diurnal species because high temperatures in anthropogenic habitats are tempered at night, c) large-bodied species, because physiological cooling costs often increase with body size (38), and d) narrow-ranging species because, on the one hand, they are often assumed to have narrow abiotic tolerances (39) and thus may need to switch their habitat affiliations to cope with thermal extremes. On the other hand, however, narrow-ranged species are also often ecological specialists and thus may be unable to shift their habitat preferences across different regions.

Results

We analyzed camera trap data collected in 2019 and 2020, consisting of over 150,000 presence records of 29 species across 2,485 sites (camera trap locations) and covering a gradient of land-cover types and climatic conditions (SI Appendix, Tables S1–S4). First, we generated a hexagonal grid across the continental United States (453 cells of >10,000 km2), and selected grid cells with at least one camera trap deployment (122 grid cells with 1 to 153 camera locations per cell; SI Appendix, Fig. S1). We then built community occupancy models that account for imperfect detection to 1) quantify intraspecific variation in mammalian response to land-cover among grid cells, 2) determine whether macroclimatic conditions explain spatial variation in responses to land-cover, and 3) assess whether species traits elucidate which species exhibit interactions between climate and land-cover.

Variability in Response to Land-Cover.

In model set 1, we assessed intraspecific variation in species’ responses to land-cover by modeling responses as a random effect over grid cells and using the random effect SD as a measure of that variation. We found that species’ responses to both land-cover variables (i.e., forest proportion and human modification) varied strongly and significantly across their US ranges. At the community level, the probability of occupancy did not systematically increase or decline in response to forest proportions (μforest = 0.12; 90% Bayesian credible intervals, BCI: 0.4 to 0.1) or human modification (μmodification = 0.13; 90% BCI: 0.35 to 0.6). Average species-level effects (i.e., average effects across all hexagons within a species’ US range) indicated that four species responded significantly positively and ten species responded significantly negatively to forest proportions. Likewise, nine species responded positively and three species responded negatively to human modifications. Based on the proportion of hexagons with negative/positive coefficients for land-cover, we classified five species as forest specialists (positive response in at least 75% of grid cells), five species as open specialists (negative response in at least 75% of grid cells), and 19 species as generalists. With respect to human modifications, six species were human affiliated (positive response in at least 75% of grid cells), five species were nature affiliated (negative response in at least 75% of grid cells), and 18 species were generalists.
These average responses, however, masked significant variations in how species responded to land-use across their ranges. At the community level, the variation across hexagonal grid cells in the community response to forest (σforest) was 1.8 (90% BCI: 0.6 to 5.7) and to human modification (σmodification) was 2 (90% BCI: 0.1 to 8.8). Similarly, SDs across hexagonal grid cells for species-specific responses to forest cover and human modification (ωLk) were above 1 for all species, and the 90% highest posterior density (HPD) for all species did not include zero (Fig. 1). Species that exhibited more spatial variation in their responses to forest cover also tended to exhibit more variation in their response to human modification (Fig. 1; Spearman’s rank correlation coefficient = 0.82). Together, these results indicate that North American mammals exhibit significant intraspecific variation in land-cover responses across space.
Fig. 1.
Spatial variation in land-cover effects on mammal space-use. (A) SD of land-cover effects on species probability of site use, estimated by a community occupancy model fit to camera-trap data of 29 species (black circles). Circles show the posterior mode of species-specific parameters—i.e., how a species’ response to land-cover varies across the species distribution within the United States; the red triangle indicates the mode community average coefficient. Gray lines show 90% Bayesian credible intervals. (B) Maps depict spatial variation in land-cover responses for the eastern cottontail Sylvilagus floridanus, as an example. Each grid cell is colored according to the predicted effect of land-cover on occupancy probability, as calculated by adding the estimates of the species-specific average response to the species-specific spatial random effect of hexagons in our occupancy model. The Top map shows cells in which forest cover is estimated to increase (green), have no effect on (white), or decrease (brown) cottontail occupancy. The Bottom map shows areas of positive (purple), neutral (white), and negative (brown) effects of human modification on occupancy. The eastern cottontail is also highlighted with a green outline in panel A.

Climate-Dependent Response to Land-Cover.

In model set 2, we used interactions between climate and land-use variables to assess whether climate mediates variation in species’ response to land-cover. We found that macroclimatic conditions mediated species’ responses to forest cover and human modification across their US ranges, with the maximum temperature of the warmest month being the strongest climatic predictor of species’ land-cover associations. In locations with higher maximum temperatures, on average, species space-use was more associated with forested habitats and more sensitive to human modification (Figs. 2 and 3, and SI Appendix, Figs. S2–S4). Specifically, in hot regions (i.e., maximum temperature of 40 °C), models predicted that increasing forest proportion from 0 to 1 or decreasing human-modified area from 0.8 to 0 would result in a 50% and 66.7% increase in occupancy for the average mammal. However, in cooler regions (i.e., maximum temperature of 20 °C), increasing forest cover and decreasing human-modified areas by the same amount would decrease average mammal occupancy by 55% and 42.1%.
Fig. 2.
Interactions between land-use and climate for terrestrial mammal communities (29 species) across the contiguous United States, predicted by a community occupancy model. In hotter regions, the community became more forest-affiliated (Top plot). Eleven species (eastern gray squirrel Sciurus carolinensis and puma Puma concolor are shown as examples) were more forest affiliated in hotter regions (A), whereas 18 species (e.g., American badger Taxidea taxus, bobcat Lynx rufus, and white-tailed deer Odocoileus virginianus) were consistent in their forest affiliations across the maximum temperature gradient (B). There were no species that became less forest affiliated in hotter regions (C). Colors represent maximum temperatures of the warmest months.
Fig. 3.
Maximum temperatures mediate mammal species responses to human modification. In hotter regions, the community became less affiliated with anthropogenic habitats (Top). One species (mule deer Odocoileus hemionus) became more affiliated with anthropogenic habitat in hotter regions (A), 20 species (e.g., gray wolf Canis lupus, puma Puma concolor, and raccoon Procyon lotor) were consistent in their habitat affiliations across the maximum temperature gradient (B), and eight species (e.g., black bear Ursus americanus, coyote Canis latrans, and Virginia opossum Didelphis virginiana) became less affiliated with anthropogenic habitats in hotter regions (C). Colors represent maximum temperatures of the warmest months.
Occupancy models from model set 2 also provided insight into which species were most likely to change their associations with forest or human-modified lands across macroclimatic gradients (SI Appendix, Figs. S3 and S4 and Table S5). Eleven of the 29 species in the community were significantly more forest affiliated in hotter regions (i.e., the 90% BCI of the interaction term did not include 0). This included both forest specialist species (as defined by results from model set 1) that became even more affiliated with forests in hot regions (e.g., puma Puma concolor) and generalist species, which were forest affiliated in hot regions and open-habitat affiliated in cooler regions (e.g., eastern gray squirrel Sciurus carolinensis). No species were significantly more forest affiliated in cooler regions, and 18 species did not significantly change their association with forest cover across the temperature gradient. Correspondingly, in hotter regions, eight species were significantly less associated with human-modified areas and 20 species did not change their association with human modification. Only one species (mule deer Odocoileus hemionus) showed the reverse trend, exhibiting a positive association with human modification in hotter regions and negative associations in cooler regions.
Trends in relationships with average annual temperatures mimicked those in maximum temperatures at the species level but were not significant at the community level (SI Appendix, Table S6). In contrast, land-use did not interact with minimum temperatures to affect species or community-level space-use. For precipitation, there were no community-level responses; that is, on average across the community, species were no more likely to use forests or human-modified areas in wetter versus drier regions. That said, some species did exhibit significant interactions between land-cover and precipitation (SI Appendix, Table S5). Specifically, eight of the 29 species in the community were significantly more forest affiliated in dry regions (i.e., had a significant negative interaction term). This included both forest specialist species (as defined by results from model set 1) that became even more affiliated with forests in dry regions (two species, puma and mule deer), and generalist species, which were forest affiliated in dry regions and open-habitat affiliated in wetter regions (six species, e.g., Virginia opossum Didelphis virginiana). On the other hand, four species were significantly more forest affiliated in wetter regions (i.e., had a significant positive interaction term; e.g., fox squirrel Sciurus niger), and 17 species did not significantly change their association with forest cover across the precipitation gradient. Correspondingly, in dry regions, five species were significantly less associated with human-modified areas (e.g., elk Cervus canadensis), 20 species did not significantly change their habitat associations, and four species (e.g., mule deer) showed the reverse trend, exhibiting positive association with human modification in drier regions and negative associations in wetter regions.

Effects of Traits on Interactions between Land Cover and Climate.

In model set 3, we assessed whether species’ traits explain the interaction strength between climate and land-cover using three-way interactions. We found mixed evidence that species traits could elucidate which species were most likely to exhibit interactive effects of land-use and maximum temperature on occupancy (SI Appendix, Table S7). Specifically, diurnal species were more likely to shift their affiliations with anthropogenic habitats across maximum temperature gradients, compared to species with nocturnal or crepuscular activity patterns (βTCmodification = 0.49; 90% BCI: 0.91 to 0.07). However, neither trophic level, body mass, nor range size affected interactions between maximum temperature and human modification (SI Appendix, Table S7). In contrast, species with larger body masses were more likely to shift their affiliations with forest cover across maximum temperature gradients (βTCforest = 0.20; 90% BCI: 0.0 to 0.43). There was also a tendency for species with larger range sizes and herbivores and primary predators to be more likely to shift their affiliations with forest cover across maximum temperature gradients (βTCforest = 0.18; 90% BCI: 0.04 to 0.42 and βTCforest = 0.37; 90% BCI: 0.76 to 0.01 for range size and trophic levels, respectively) compared to top predators, which were more consistently affiliated with high forest cover across their ranges (Fig. 4). Finally, unlike for human modification, temporal activity patterns did not influence interactions between maximum temperature and forest cover.
Fig. 4.
Interactions between land-cover and maximum temperatures may be partially mediated by species traits. The plots show how the probability of occupancy changes with forest proportions and human modifications in areas with different maximum temperatures. Responses are predicted by four community occupancy models with a three-way interaction term between forest cover or human modification, maximum temperatures, and each trait. Thus, the plots show marginal, and not conditional, effects. Species range size and body mass are continuous (and predictions are graphed at 3 discrete levels). Trophic level (herbivores and primary predators—i.e., levels 1 and 2, versus top predators—i.e., level 3) and activity time are binary (diurnal and cathemeral versus nocturnal and crepuscular). Models suggest herbivores, primary predators, species with larger range sizes, and larger species may be more likely to exhibit forest-climate interactions. Likewise, species with diurnal activity patterns are more likely to exhibit interactions between human modification and temperatures.

Discussion

Our results suggest that intraspecific variation in habitat preferences is pervasive among North American mammals, with all studied species exhibiting significant spatial variability in their use of forests and human-modified habitats across the continent. Specifically, we found that mammals switch their habitat affiliations to become more associated with forests and more sensitive to human modification in warmer regions. Species traits partly elucidated which species were most likely to exhibit significant interactions between land-use and climate, with larger, wider-ranging, and diurnal herbivores and primary predators more likely to use forests or avoid human-modified habitats in warmer regions.

Intraspecific Variation in Mammalian Responses to Land-Cover.

Though much research has sought to guide conservation priorities by characterizing which species are likely to thrive versus decline in human-modified habitats (57), intraspecific variation in species responses is only beginning to be acknowledged in macroecological studies (10). Our results suggest grouping species into broad categories—such as urban avoiders, adapters, and exploiters—may mask substantial variation in how species respond to habitat conversion. In some senses, this finding is unsurprising: Population-level space-use is the outcome of individual habitat-use decisions, and the resource/habitat selection literature is replete with examples of individuals exhibiting different habitat selection strategies (40) due to intrinsic, e.g., sex; (41) and extrinsic, e.g., habitat availability; (42) factors.
Nonetheless, failing to recognize that populations exhibit predictable variability in habitat preferences across space has important implications for conservation (43). On the one hand, environmental managers may divert resources toward improving human-modified landscapes in areas where species are largely restricted to natural habitats. On the other hand, assuming a species will always be restricted to natural areas may lead to missed opportunities for conserving biodiversity alongside people: for example, by planting hedgerows, retaining small patches of natural habitats, or reducing agrochemical applications in human-modified landscapes (3).

Maximum Temperatures Mediate Mammalian Responses to Land-Cover.

Maximum temperatures, much more so than average temperatures or minimum temperatures, played a key role in determining how species respond to forest cover and human modification. As noted, closed forest canopies can thermally buffer the understory, thereby guarding against regional climate extremes (44). In contrast, many anthropogenic habitats lack closed canopies and thus also lack the local thermal buffering capacity needed to provide organisms with climate refugia (20, 2224). The microclimate buffering hypothesis posits that species’ habitat preferences may be at least partially driven by variations in climate buffering capacities among habitat types (22). Correspondingly, multiple tracking studies have shown that large mammals change their habitat selection preferences during heat waves. For example, moose Alces alces increasingly occupied forests, and ibex Capra ibex moved upslope when temperatures spiked (45, 46). Birds too are known to occupy more shaded areas within their home ranges during heat waves (47, 48). Behavioral thermoregulation is also a well-known phenomenon among herpetofauna, though temperature-induced changes in habitat use often occur at smaller scales, e.g., retreating into shade, burrows, or tree crevasses (49, 50). In particularly warm regions, organisms may thus pre-emptively increase their use of thermally buffered habitats—as we report here—because in open areas without microclimate refugia, heat waves would exact direct physiological stress on organisms (e.g., hyperthermia), and limit their activity times, thus preventing organisms from foraging or reproducing (38, 51, 52). Another possibility is that temperature spikes may aggravate already lower resource availability in human-modified areas, causing organisms to retreat into natural habitats (5355).
Unlike maximum temperatures, we found precipitation gradients to have mixed effects on species’ habitat preferences. We had hypothesized that higher moisture availability in forests might cause species to specialize in forests in drier regions (27), as seems to be the case for some Neotropical birds (30). However, we observed no significant interactions between habitat affiliation and precipitation at the community level. At the species level, we found eight species were more forest affiliated and five species were less human-affiliated in dry regions, aligning with our initial hypothesis. However, four and four species exhibited the reverse trend, becoming less forest affiliated and more human-affiliated in dry regions. Some of this variation could be due to anthropogenic water subsidies: In dry regions, irrigated agriculture or residential yards may actually have more moisture availability than nearby natural areas. Indeed, several studies have shown that anthropogenic water subsidies during drought or in dry regions can strongly influence wildlife distributions (56, 57).

Traits Explain Some Variation in Interactions between Climate and Land-Use.

Our results suggest that some species exhibit more intraspecific variation in habitat preferences than others and species that tend to be more spatially variable in their forest affiliation are also more spatially variable in their sensitivity to anthropogenic modification. The traits we examined exhibited some explanatory power in differentiating species that were more versus less likely to exhibit significant interactions between climate and land-use and generally confirmed our expectations. First, larger species showed stronger interactions between maximum temperatures and forest cover. This may be because larger species tend to have increased cooling costs and thus face greater difficulties thermoregulating (38). Second, top predators showed marginally weaker interactions between maximum temperatures and forest cover than herbivores or primary predators, likely because top predators are often more subject to human persecution and thus relatively sensitive to human modification and more restricted to forest habitat everywhere (37). Third, species with larger ranges showed marginally stronger interactions between maximum temperatures and forest cover, lending support for the possibility that small-ranged species may be so specialized that they are unable to shift their habitat preferences in different regions (39). We had originally hypothesized that narrow-ranged species might be more likely to shift their habitat affiliations across thermal gradients as small ranges can be associated with narrow thermal tolerances. Future work might look at thermal tolerances more directly as this has been shown to outperform surrogate range size measurements in prior studies (58). Finally, diurnal species showed stronger interactions between climate and land-use than nocturnal species, maybe because the highest temperatures organisms experience occur during the day. For our focal species, traits were not correlated (SI Appendix, Table S8); thus, the tendency for stronger interactions between land-use and climate for species with a particular trait is unlikely to be a confounding effect of other trait effects. Ultimately, however, the strong community-level patterns indicate that heightened sensitivity to habitat modification in hotter regions may be a general phenomenon among North American mammals.

Limitations.

Our study is not without some limitations. First, though studying natural variation across large spatial or temporal scales can indicate important ecological insights when experiments are not feasible (59), our study is nonetheless correlational. It is thus possible that unmeasured factors could drive some of the climate-associated habitat shifts that we observed. For example, our definition of “forest” is broad, and temperature-dependent affiliations with forest could thus reflect changes in species preferences for different forest types along temperature gradients (60). Similarly, if particularly intensive agricultural systems or highly urbanized cities were more likely to occur in hot locations, and diversified farms occurred in cool locations, then species might shift their habitat affiliations to avoid human-dominated habitats in warm locations not because of temperature but rather due to local land-use intensity. Moreover, while we did not detect strong correlations between land-use and climatic variables used in our analyses, forest proportion and maximum temperature showed a weak negative correlation (SI Appendix, Table S4). For pumas—the only forest specialist in our analysis whose forest affiliation became significantly stronger with increasing temperature—the tendency of forest habitat to decline with increasing temperature may force individuals into smaller forest patches, thus increasing forest use. However, this mechanism is unlikely to explain the shift—from avoiding to preferring forest as temperature increases—which we observed in most of the species with significant changes in forest affiliation across the maximum temperature gradient. Yet another possibility is that trends could partially result from shifts in species interactions across climate or land-use gradients (61, 62). For example, anthropogenic resource subsidies are thought to increase the spatiotemporal co-occurrence of antagonistically interacting mammals (55). If anthropogenic resource subsidies are more abundant in cooler than warmer regions, then antagonistic species interactions could force species into natural habitats in warm regions but not cool regions. Second, we limited our analysis to wide-ranging species to ensure sufficient variability in spatial covariates across the species’ range; however, narrow-range species may be more specialized compared to species included here (63). Thus, intraspecific variation in habitat preferences may not be as common across a wider swath of North American mammal species as our study suggests. Third, to quantify the effects of traits on climate-dependent habitat preferences, we constructed a rather complex model with multiple random effects alongside two- and three-way interactions. It is possible that the weak trait effects that we observed arose because of model complexity.

Conclusions.

While many studies have explored the effects of climate and land-use in isolation, the interactive effects of climate and land-use are less understood but still critical to understanding species distributions (26, 35, 36). Our findings reaffirm that high temperatures may prevent species from exploiting habitats that they might otherwise occupy (64). Looking forward, understanding how and when species track their niches across multiple axes—such as the habitat and climate niches investigated here—will have major implications for elucidating the biogeography of the Anthropocene (65). With ongoing climate change, hot temperature extremes—and as a consequence, maximum temperatures experienced by wildlife—will become even more frequent and extreme (33). Our work suggests these climate extremes may further constrain species distributions and compromise efforts to conserve biodiversity in human-dominated landscapes. More optimistically, however, our results may also provide avenues for developing forward-looking conservation strategies by identifying how human-modified habitats might be managed to provide the microclimatic refugia species need to persist by protecting forests or other cool shady habitats.

Materials and Methods

Camera Trap Data.

To study spatial variation in how species space-use is affected by habitat conversion, we used the Snapshot USA dataset, a coordinated survey of camera traps from 196 sampling arrays (different projects contributing data) across the contiguous United States (66, 67). Details on sampling arrays and camera models are reported elsewhere (66, 67). Briefly, sampling arrays were active for a minimum of 400 trap nights (a minimum of ten cameras for at least 40 d and nights). Within an array, cameras were placed 200 to 5,000 m apart from one another, and the entire array was located in one major habitat type (e.g., forest, grassland, etc.). Cameras were positioned 30 to 50 cm off the ground and programmed to take 1 to 10 photographs per trigger. Traps were active from August through November in all lower 48 states in 2019 and 2020. The resulting camera trap data were uploaded to the eMammal Data Management System (emammal.si.edu), and all species identifications were reviewed by at least one expert.
We generated a hexagonal grid across the United States with a diameter of 200 km and centroids separated by approximately 150 km and asked whether species responses to land-cover vary among grid cells or whether they are consistent. To address this question, we filtered the Snapshot USA dataset to only include wild terrestrial mammals with a minimum 350,000 km2 range size (equivalent to 10 hexagons) within the contiguous United States (68, 69). We opted for large spatial scales to ensure an adequate amount of data in sampled hexagon grid cells. Each species also had to be detected within a minimum of five hexagons. These filtering steps helped ensure that there would be variation in both land-use and climatic conditions across the species’ US range, as well as a minimum level of information needed to estimate species associations with these environmental variables. Finally, we only included species that weigh at least 300 g (70), as smaller species are not detected as reliably with the camera trap setup described above and also not as reliably identified (71, 72).
To construct the observation histories, we defined 7 d as one camera trap occasion for which each species could be detected or not. Then, for each site (camera trap location) i, we aggregated binary occasion-specific species detections over all occasions to frequencies; that is, how many times each species k was detected out of the total number of 7-d occasions site i had been surveyed. We did not include any camera traps with active times shorter than seven days. Species observation histories included all sites at which a species was ever detected as well as sites where species were not detected but that were inside the species’ geographic range (68, 69).

Covariates.

To quantify the association between species’ space-use and land-use, we extracted land-cover covariates at a 1-km buffer around each camera site. We determined the forest proportion within each 1-km buffer using the National Land-Cover Database 2016 (73) by summing all pixels categorized as deciduous, coniferous, or mixed forest divided by the total number of pixels comprising the buffer. We determined the anthropogenic impact for each 1-km buffer from the global human modification layer (74). To do so, we used the 30-m resolution human modification map to quantify the average level of human modification across all cells that fell within each 1-km buffer. We also extracted covariates at 5- and 10-km buffers to determine whether or not spatial patterns held at multiple spatial resolutions. Because covariate values were highly correlated across the spatial scales (SI Appendix, Table S3), we used the 1-km covariates in model building. To quantify the interacting effects of land-use and climate, we extracted the maximum temperature of the warmest month, minimum temperature of the coldest month, mean annual temperature, and total annual precipitation at each camera trap location from WORLDCLIM, downloaded at a 1-km resolution (75). Total annual precipitation was highly correlated with precipitation in the wettest month and quarter (r = 0.81 and 0.84, respectively). We standardized all covariates before model fitting. The range of spatial covariates present within species’ US ranges is summarized in SI Appendix, Table S2. To quantify how climate-dependent habitat associations change with species traits, we extracted information on species’ trophic levels (76), diurnal activities (70), body masses (70), and because of the spatial extent of our analysis, we used range sizes within the continental United States instead of the global range sizes (68, 69) (SI Appendix, Table S1).

Model Definition.

We used community occupancy models to quantify spatial variation in species’ habitat affiliations. Occupancy models use species detections and nondetections from repeated visits to sampled sites to estimate the probability of species occurrence and its relationship with environmental covariates while accounting for imperfect detection (77). Thus, they consist of ecological and observation submodels. Community occupancy models jointly analyze data from multiple species, allowing species-specific and community-level parameter estimates (78). All species-specific parameters are drawn from a common distribution governed by community hyperparameters, allowing estimations for rarer species (79). We built three sets of community occupancy models in a Bayesian framework to quantify whether associations between species’ space-use and land-cover are consistent across their US ranges (Set 1), whether climate variables mediate variation in species’ responses to land-cover (Set 2), and whether species traits explain the interaction between climate and land-cover (Set 3).

(i) Ecological submodel.

We modeled the true occupancy state of species k at site i, zik (1 if present, and 0 otherwise) as a Bernoulli trial with a probability of occupancy ψik:
zikBernoulli(ψik).
[1]
Spatial variation in occupancy probability can be modeled as a logit-linear function of covariates ci:
logit(ψik)=β0k+βckci.
[2]
Here, β0k are species-specific random intercepts that allow species to vary in their baseline occupancy probabilities. Random intercepts were drawn from a normal distribution, where μ0 and σ0 are the community mean and SD.
β0kNormal(μ0,σ0).
[3]
Similarly, species-specific effects of covariate c were normally distributed with community hyperparameters:
βckNormal(μc,σc).
[4]
Our three model sets included year as a species-specific binary covariate on occupancy probability to account for potential temporal variation. In model set (1), we quantified spatial variability of species’ response to land-use by modeling occupancy probability as a function of species-specific land-cover effects. To account for spatial autocorrelation, we included species-specific random effects of sampling array a, ϵak, including only arrays within each species’ US range and drawing from a normal distribution with SD τ across species.
logit(ψik)=β0k+βYkYeari+βLhikLandcoveri+ϵak.
[5]
ϵakNormal(0,τ).
[6]
For each species k, the land-cover effect at site i, βLhik, depends on which hexagon h site i is located in. Specifically, we modeled each species’ set of βLhik as a normal random effect δLhk with SD ωLk:
βLhik=μLk+δLhk.
[7]
δLhkNormal(0,ωLk).
[8]
We only included hexagons within each species’ US range. The species-level intercept μLk describes the average response of species k to land-cover L across hexagons and comes from a normal distribution with hyperparameters that are estimated in the model (μβ, σβ). The SD ωLk quantifies the intraspecific variability in species’ responses to land-cover across the species’ US range. We modeled ωLk as coming from a shared gamma distribution with hyperparameters a and b:
ωLkGamma(a,b).
[9]
μLkNormal(μβ,σβ).
[10]
We classified species as habitat specialists when the βLhik for a given land-cover was either positive or negative for at least 75% of the hexagons within that species’ US range and as a generalist for that land-cover otherwise. We used this classification in interpreting the variation in species’ responses to land-cover in model set (2).
In model set (2), we quantified how climate affects species’ responses to land-use by modeling occupancy probability as a function of climate, land-cover, and their interactions, with species-specific coefficients, and a species-specific random effect of hexagon h, δhk, again including only hexagons within a species’ US range. Significant interaction effects βIk indicate climate-dependent species associations with land-cover class.
logit(ψik)=β0k+βYkYeari+βLkLandcoveri+βCkClimatei+βIkLandcoveriClimatei+δhk.
[11]
In model set (3), we quantified whether species traits affect the interactions between climate and land-use in species space-use. To do so, in addition to the covariates in model set (2), we modeled occupancy probability as a function of a species trait βTTraitk, the interaction of that trait with climate βTC, and land-use βTL, and a three-way interaction between trait, climate, and land-use βTCL. Significant three-way interaction effects indicate a change in the interaction effect between climate and land-use based on a species’ trait.

(ii) Observation submodel.

Because nondetection of a species at a sampling site can either be caused by a true absence or by a failure to detect, repeated visits over the number of occasions to sampling sites can be used to estimate detection probability pik conditional on occupancy (pik = 0 where zik = 0, i.e., the species is not present). Observations yik are assumed to be from a binomial trial, where ni is the number of sampling occasions at site i.
yikBinomial(pikzik,ni).
[12]
Heterogeneity in detection was modeled as a logit-linear function of 1) sampling effort (i.e., the number of days with active camera trapping at site i, Xi) to account for variation in sampling intensity across sites, 2) species identity α0k, to account for some species being more detectable than others, and 3) a species-specific random effect of sampling array a, ϵak, including only arrays within each species’ US range. Sampling array was included to account for potential differences in setup techniques and equipment used by the different projects that contributed data, as well as differences due to habitat in which an array was located, and were drawn from a normal distribution with SD τ shared across species.
logit(pik)=α0k+αXi+ϵak.
[13]
ϵakNormal(0,τ).
[14]
Species-specific intercepts α0k were drawn from a normal distribution with community parameters:
α0kNormal(θ0,τ0).
[15]

Model Run and Diagnostics.

We fit the three model sets separately for the two land-use covariates (forest proportion and human modification) for a total of 2, 8, and 8 model runs for model sets 1 to 3, respectively. In model set (2), we combined each land-use with individual climate variables (annual precipitation and minimum, maximum, and mean temperatures). Because of the model complexity, we only included one climate variable and one land-use variable at a time. Similarly, in model set (3), we included one of the four traits at a time, combined with either forest proportion or human modification and maximum temperature. We only implemented models for maximum temperature based on analyses from the model set (2). To ensure our main results are robust to prior choice, we also implemented our models with a potentially less informative Beta(1,1) prior on the community average intercepts for the probability of detection and occupancy and found that parameter estimates were not sensitive to the choice of this prior (SI Appendix, Fig. S5). All model definitions are provided in SI Appendix.
We used Markov chain Monte Carlo (MCMC) simulation with NIMBLE [version 0.11.1; (80)] in R version 4.0.5 (81) to fit our models. We ran four chains of 200,000 iterations and discarded the first 50,000 samples as burn-in and further thinned the remaining samples by 20. We assessed convergence using the potential scale reduction value for all parameters and by inspecting the mixing of the chains using trace plots (82). We tested whether the model adequately fit the data by calculating a Bayesian P-value (83) (SI Appendix, Table S9). We report the posterior mean of all estimates with 90% Bayesian credible intervals (BCI); we interpret coefficients as significant when their 90% BCI does not overlap zero. For SDs (model set 1), we report the posterior mode, because of the expected skewed posterior distributions. Because values of SDs are bounded by zero, we estimated the density using a truncated kernel and evaluated a 90% credible interval using the 90% highest posterior density interval (84). This allowed a value of zero to be included in the 90% credible interval. We used 90% intervals throughout the results to ensure stable posterior densities (85).

Data, Materials, and Software Availability

Data used in this study are publicly available (66, 67).

Acknowledgments

Author contributions

M.T., R.S., and D.S.K. designed research; M.T., R.S., R.K., J.A., E.F., and D.S.K. performed research; M.T. and R.S. analyzed data; R.K. contributed to the preparation of the data and interpreting the results; J.A. and E.F. helped with funding and discussed the results; and M.T., R.S., and D.S.K. wrote the paper.

Competing interests

The authors declare no competing interest.

Supporting Information

Appendix 01 (PDF)

References

1
T. Newbold et al., Global effects of land use on local terrestrial biodiversity. Nature 520, 45–50 (2015).
2
T. Newbold et al., Global patterns of terrestrial assemblage turnover within and among land uses. Ecography 39, 1151–1163 (2016).
3
C. Kremen, A. M. Merenlender, Landscapes that work for biodiversity and people. Science 362, eaau6020 (2018).
4
M. L. McKinney, J. L. Lockwood, Biotic homogenization: A few winners replacing many losers in the next mass extinction. Trends Ecol. Evol. 14, 450–453 (1999).
5
M. Dornelas et al., A balance of winners and losers in the anthropocene. Ecol. Lett. 22, 847–854 (2019).
6
G. N. Daskalova, I. H. Myers-Smith, J. L. Godlee, Rare and common vertebrates span a wide spectrum of population trends. Nat. Commun. 11, 1–13 (2020).
7
J. P. Suraci et al., Disturbance type and species life history predict mammal responses to humans. Global Change Biol. 27, 3718–3731 (2021).
8
R. B. Blair, Land use and avian species diversity along an urban gradient. Ecol. Appl. 6, 506–519 (1996).
9
C. H. Nilon, L. W. VanDruff, Analysis of small mammal community data and applications to management of urban greenspaces. Proc. Natl. Sym. Urban Wildlife 2, 53–59 (1986).
10
C. Banks-Leite, M. G. Betts, R. M. Ewers, C. D. L. Orme, A. L. Pigot, The macroecology of landscape ecology. Trends Ecol. Evol. 37, 480–487 (2022).
11
C. D. L. Orme et al., Distance to range edge determines sensitivity to deforestation. Nat. Ecol. Evol. 3, 886–891 (2019).
12
M. G. Betts et al., Extinction filters mediate the global effects of habitat fragmentation on animals. Science 366, 1236–1239 (2019).
13
M. R. Lambert, K. I. Brans, S. Des Roches, C. M. Donihue, S. E. Diamond, Adaptive evolution in cities: Progress and misconceptions. Trends Ecol. Evol. 36, 239–257 (2021).
14
B. C. Verrelli et al., A global horizon scan for urban evolutionary ecology. Trends Ecol. Evol. 37, 1006–1019 (2022).
15
H. R. Pulliam, Sources, sinks and population regulation. Am. Nat. 132, 652–661 (1988).
16
J. J. Gilroy, D. P. Edwards, Source-sink dynamics: A neglected problem for landscape-scale biodiversity conservation in the tropics. Curr. Lands. Ecol. Rep. 2, 51–60 (2017).
17
C. Sirami et al., Increasing crop heterogeneity enhances multitrophic diversity across agricultural regions. Proc. Natl. Acad. Sci. U.S.A. 116, 16442–16447 (2019).
18
J. J. Williams, T. Newbold, Vertebrate responses to human land use are influenced by their proximity to climatic tolerance limits. Divers. Distrib. 27, 1308–1323 (2021).
19
J. J. Williams, R. Freeman, F. Spooner, T. Newbold, Vertebrate population trends are influenced by interactions between land use, climatic position, habitat loss and climate change. Global Change Biol. 28, 797–815 (2022).
20
A. J. Suggitt et al., Extinction risk from climate change is reduced by microclimatic buffering. Nat. Clim. Change 8, 713–717 (2018).
21
S. J. Frey, A. S. Hadley, M. G. Betts, Microclimate predicts within-season distribution dynamics of montane forest birds. Divers. Distrib. 22, 944–959 (2016).
22
H. Kim, B. C. McComb, S. J. Frey, D. M. Bell, M. G. Betts, Forest microclimate and composition mediate long-term trends of breeding bird populations. Global Change Biol. 28, 6180–6193 (2022).
23
B. R. Scheffers, T. A. Evans, S. E. Williams, D. P. Edwards, Microhabitats in the tropics buffer temperature in a globally coherent manner. Biol. Lett. 10, 20140819 (2014).
24
J. Lenoir, T. Hattab, G. Pierre, Climatic microrefugia under anthropogenic climate change: Implications for species redistribution. Ecography 40, 253–266 (2017).
25
R. A. Senior, J. K. Hill, P. González, L. K. del Pliego, DP Edwards. Goode, A pantropical analysis of the impacts of forest degradation and conversion on local temperature. Ecol. Evol. 7, 7897–7908 (2017).
26
A. J. Nowakowski, L. O. Frishkoff, M. Agha, B. D. Todd, B. R. Scheffers, Changing thermal landscapes: Merging climate science and landscape ecology through thermal biology. Curr. Lands. Ecol. Rep. 3, 57–72 (2018).
27
J. J. Williams, T. Newbold, Local climatic changes affect biodiversity responses to land use: A review. Divers. Distrib. 26, 76–92 (2020).
28
L. O. Frishkoff, E. A. Hadly, G. C. Daily, Thermal niche predicts tolerance to habitat conversion in tropical amphibians and reptiles. Global Change Biol. 21, 3901–3916 (2015).
29
L. O. Frishkoff, E. Gabot, G. Sandler, C. Marte, D. L. Mahler, Elevation shapes the reassembly of anthropocene lizard communities. Nat. Ecol. Evol. 3, 638–646 (2019).
30
L. O. Frishkoff et al., Climate change and habitat conversion favour the same species. Ecol. Lett. 19, 1081–1090 (2016).
31
D. S. Karp et al., Agriculture erases climate-driven -diversity in neotropical bird communities. Global Change Biol. 24, 338–349 (2018).
32
U. Srinivasan, P. R. Elsen, D. S. Wilcove, Annual temperature variation influences the vulnerability of montane bird communities to land-use change. Ecography 42, 2084–2094 (2019).
33
IPCC, Summary for Policymakers: Climate Change 2022\_ Impacts, Adaptation and Vulnerability\_Working Group II contribution to the Sixth Assessment Report of the Intergovernamental Panel on Climate Change (2022), p. 37.
34
P. Batáry, L. V. Dicks, D. Kleijn, W. J. Sutherland, The role of agri-environment schemes in conservation and environmental management. Conser. Biol. 29, 1006–1016 (2015).
35
J. M. Travis, Climate change and habitat destruction: A deadly anthropogenic cocktail. Proc. R. Soc. B: Biol. Sci. 270, 467–473 (2003).
36
C. Sirami et al., Impacts of global change on species distributions: Obstacles and solutions to integrate climate and land use. Global Ecol. Biogeogr. 26, 385–394 (2017).
37
W. J. Ripple et al., Status and ecological effects of the world’s largest carnivores. Science 343, 1241484 (2014).
38
E. A. Riddell, K. J. Iknayan, B. O. Wolf, B. Sinervo, S. R. Beissinger, Cooling requirements fueled the collapse of a desert bird community from climate change. Proc. Natl. Acad. Sci. U.S.A. 116, 21609–21615 (2019).
39
B. S. Pease, K. Pacifici, R. Kays, B. Reich, What drives spatially varying ecological relationships in a wide-ranging species? Divers. Distrib. 28, 1752–1768 (2022).
40
M. Leclerc et al., Quantifying consistent individual differences in habitat selection. Oecologia 180, 697–705 (2016).
41
J. G. Kie, R. T. Bowyer, Sexual segregation in white-tailed deer: Density-dependent changes in use of space, habitat selection, and dietary niche. J. Mammal. 80, 1004–1020 (1999).
42
A. Mysterud, R. A. Ims, Functional responses in habitat use: Availability influences relative use in trade-off situations. Ecology 79, 1435–1441 (1998).
43
B. J. Sigel, W. Douglas Robinson, T. W. Sherry, Comparing bird community responses to forest fragmentation in two lowland central American reserves. Biol. Conser. 143, 340–350 (2010).
44
P. De Frenne et al., Forest microclimates and climate change: Importance, drivers and future research agenda. Global Change Biol. 27, 2279–2297 (2021).
45
M. Melin et al., Moose (Alces alces) reacts to high summer temperatures by utilizing thermal shelters in boreal forests-an analysis based on airborne laser scanning of the canopy structure at moose locations. Global Change Biol. 20, 1115–1125 (2014).
46
T. H. Mason, F. Brivio, P. A. Stephens, M. Apollonio, S. Grignolio, The behavioral trade-off between thermoregulation and foraging in a heat-sensitive species. Behav. Ecol. 28, 908–918 (2017).
47
S. J. Cunningham, J. L. Gardner, R. O. Martin, Opportunity costs and the response of birds and mammals to climate warming. Front. Ecol. Environ. 19, 300–307 (2021).
48
L. K. Neumann, S. D. Fuhlendorf, C. A. Davis, S. M. Wilder, Climate alters the movement ecology of a non-migratory bird. Ecol. Evol. 12, e8869 (2022).
49
A. L. Burrow, R. T. Kazmaier, E. C. Hellgren, D. C. Ruthven III, Microhabitat selection by Texas horned lizards in southern Texas. J. Wild. Manag. 65, 645–652 (2001).
50
O. Attum, A. Kramer, S. M. B. El Din, Thermal utility of desert vegetation for the Egyptian tortoise and its conservation implications. J. Arid Environ. 96, 73–79 (2013).
51
B. Sinervo et al., Erosion of lizard diversity by climate change and altered thermal niches. Science 328, 894–899 (2010).
52
J. C. Wingfield et al., How birds cope physiologically and behaviourally with extreme climatic events. Philos. Trans. R. Soc. B: Biol. Sci. 372, 20160140 (2017).
53
E. K. Edwards, N. J. Mitchell, A. R. Ridley, The impact of high temperatures on foraging behaviour and body condition in the western Australian magpie Cracticus tibicen dorsalis. Ostrich 86, 137–144 (2015).
54
A. Fuller et al., How dryland mammals will respond to climate change: The effects of body size, heat load and a lack of food and water. J. Exp. Biol. 224, 1–11 (2021).
55
N. A. Gilbert, J. L. Stenglein, J. N. Pauli, B. Zuckerberg, Human disturbance compresses the spatiotemporal niche. Proc. Natl. Acad. Sci. U.S.A. 119, e2206339119 (2022).
56
B. M. Myers, E. J. Questad, M. D. Hubbell, D. J. Moriarty, Trends in bird species richness in the midst of drought. Western Birds 50, 74–91 (2019).
57
L. N. Rich, S. R. Beissinger, J. S. Brashares, B. J. Furnas, Artificial water catchments influence wildlife distribution in the Mojave desert. J. Wild. Manage. 83, 855–865 (2019).
58
C. A. Waldock, A. De Palma, P. A. Borges, A. Purvis, Insect occurrence in agricultural land-uses depends on realized niche and geographic range properties. Ecography 43, 1717–1728 (2020).
59
N. G. Yoccoz, J. D. Nichols, T. Boulinier, Monitoring of biological diversity in space and time. Trends Ecol. Evol. 16, 446–453 (2001).
60
P. B. Reich et al., Temperature drives global patterns in forest biomass distribution in leaves, stems, and roots. Proc. Natl. Acad. Sci. U.S.A. 111, 13721–13726 (2014).
61
S. A. Chamberlain, J. L. Bronstein, J. A. Rudgers, How context dependent are species interactions? Ecol. Lett. 17, 881–890 (2014).
62
B. Howes et al., Deforestation alters species interactions. Nat. Sci. 3, e20220027 (2023).
63
R. A. Slatyer, M. Hirst, J. P. Sexton, Niche breadth predicts geographical range size: A general ecological pattern. Ecol. Lett. 16, 1104–1114 (2013).
64
C. M. Mccain, R. K. Colwell, Assessing the threat to montane biodiversity from discordant shifts in temperature and precipitation in a changing climate. Ecol. Lett. 14, 1236–1245 (2011).
65
M. W. Tingley, W. B. Monahan, S. R. Beissinger, C. Moritz, Birds track their Grinnellian niche through a century of climate change. Proc. Natl. Acad. Sci. U.S.A. 106, 19637–19643 (2009).
66
M. V. Cove et al., SNAPSHOT USA 2019: A coordinated national camera trap survey of the United States. Ecology 102, 2019–2020 (2021).
67
R. Kays et al., SNAPSHOT USA 2020: A second coordinated national camera trap survey of the united states during the COVID-19 pandemic. Ecology 103, e3775 (2022).
68
IUCN Red List of Threatened Species. version 2022-2 (2022). https://www.iucnredlist.org.
69
U.S. Geological Survey - Gap Analysis Project Species Range Maps CONUS 2001: U.S. Geological Survey data release (2018). https://doi.org/10.5066/f7q81b3r.
70
D. T. Cox, A. S. Gardner, K. J. Gaston, Diel niche variation in mammals associated with expanded trait space. Nat. Commun. 12, 1–10 (2021).
71
R. Kays et al., Which mammals can be identified from camera traps and crowdsourced photographs? J. Mammal. 103, 767–775 (2022).
72
M. Tourani, E. N. Brøste, S. Bakken, J. Odden, R. Bischof, Sooner, closer, or longer: Detectability of mesocarnivores at camera traps. J. Zool. 312, 259–270 (2020).
73
C. Homer, J. Fry, The national land cover database. US Geol. Surv. Fact Sheet 3020, 1–4 (2012).
74
D. M. Theobald et al., Earth transformed: Detailed mapping of global human modification from 1990 to 2017. Earth Syst. Sci. Data 12, 1953–1972 (2020).
75
S. E. Fick, R. J. Hijmans, WorldClim 2: New 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315 (2017).
76
K. E. Jones et al., PanTHERIA: A species-level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology 90, 2648 (2009).
77
D. I. MacKenzie et al., Estimating site occupancy rates when detection probabilities are less than one. Ecology 83, 2248–2255 (2002).
78
R. M. Dorazio, J. A. Royle, Estimating size and composition of biological communities by modeling the occurrence of species. J. Am. Stat. Assoc. 100, 389–398 (2005).
79
E. F. Zipkin, A. Dewan, J. Andrew Royle, Impacts of forest fragmentation on species richness: A hierarchical approach to community modelling. J. Appl. Ecol. 46, 815–822 (2009).
80
P. de Valpine et al., Programming with models: Writing statistical algorithms for general model structures with nimble. J. Comput. Graph. Stat. 26, 403–413 (2017).
81
R development core team: A language and environment for statistical computing. R Foundation Statistical Computational, Vienna Austria (2020).
82
S. P. Brooks, A. Gelman, General methods for monitoring convergence of iterative simulations. J. Comput. Graphical Stat. 7, 434–455 (1998).
83
A. Gelman, J. B. Carlin, H. S. Stern, D. B. Rubin, Bayesian Data Analysis (Chapman and Hall/CRC, 1995).
84
L. O. Frishkoff, P. De Valpine, L. K. M’Gonigle, Phylogenetic occupancy models integrate imperfect detection and phylogenetic signal to analyze community structure. Ecology 98, 198–210 (2017).
85
M. D. Edge, Statistical Thinking from Scratch: A Primer for Scientists (Oxford University Press, 2019).

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 120 | No. 50
December 12, 2023
PubMed: 38048469

Classifications

Data, Materials, and Software Availability

Data used in this study are publicly available (66, 67).

Submission history

Received: March 16, 2023
Accepted: October 14, 2023
Published online: December 4, 2023
Published in issue: December 12, 2023

Keywords

  1. mammal community
  2. climate change
  3. occupancy
  4. biodiversity conservation
  5. habitat loss

Acknowledgments

Author Contributions
M.T., R.S., and D.S.K. designed research; M.T., R.S., R.K., J.A., E.F., and D.S.K. performed research; M.T. and R.S. analyzed data; R.K. contributed to the preparation of the data and interpreting the results; J.A. and E.F. helped with funding and discussed the results; and M.T., R.S., and D.S.K. wrote the paper.
Competing Interests
The authors declare no competing interest.

Notes

This article is a PNAS Direct Submission.
Although PNAS asks authors to adhere to United Nations naming conventions for maps (https://www.un.org/geospatial/mapsgeo), our policy is to publish maps as provided by the authors.

Authors

Affiliations

Department of Ecosystem and Conservation Sciences, University of Montana, Missoula, MT 59812
Department of Wildlife, Fish, and Conservation Biology, University of California, Davis, CA 95616
Department of Wildlife, Fish, and Conservation Biology, University of California, Davis, CA 95616
Department of Ecological Dynamics, Leibniz Institute for Zoo and Wildlife Research, Berlin 10315, Germany
Department of Forestry and Environmental Resources, North Carolina State University, Raleigh, NC 27607
North Carolina Museum of Natural Sciences, Raleigh, NC 27601
Jorge Ahumada
Moore Center for Science, Conservation International, Arlington, VA 22202
Center for Biodiversity Outcomes, Julia Ann Wrigley Global Institute of Sustainability, Arizona State University, Tempe, AZ 85281
Eric Fegraus
Moore Center for Science, Conservation International, Arlington, VA 22202
Department of Wildlife, Fish, and Conservation Biology, University of California, Davis, CA 95616

Notes

1
To whom correspondence may be addressed. Email: [email protected].

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

View Options

View options

PDF format

Download this article as a PDF file

DOWNLOAD PDF

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Personal login Institutional Login

Recommend to a librarian

Recommend PNAS to a Librarian

Purchase options

Purchase this article to access the full text.

Single Article Purchase

Maximum temperatures determine the habitat affiliations of North American mammals
Proceedings of the National Academy of Sciences
  • Vol. 120
  • No. 50

Media

Figures

Tables

Other

Share

Share

Share article link

Share on social media