Skip to main content
NCBI home page

This is a preprint.

It has not yet been peer reviewed by a journal.

The National Library of Medicine is running a pilot to include preprints that result from research funded by NIH in PMC and PubMed.

bioRxiv logoLink to bioRxiv
[Preprint]. 2025 Sep 8:2025.09.04.674278. [Version 1] doi: 10.1101/2025.09.04.674278

Uncovering historical small mammal biodiversity among the Madrean Sky Islands

Damien C Rivera 1, Nathan S Upham 1
PMCID: PMC12439921  PMID: 40964259

Abstract

The Madrean Sky Island Archipelago is a system of 54 mountains with isolated woodland habitat above 1,600 meters, primarily in the Sonoran Desert. These mountains harbor a wide variety of native small mammals spanning 11 families of bats, rodents and shrews. Improved understanding of Madrean Sky Island biodiversity will potentially advance studies of biogeography, phylogenetic relationships, host-symbiont interactions, and processes of community assembly in this ecoregion. However, which species are found in each sky island and how their populations are genetically related remain open questions. To establish the current knowledge baseline, we used voucher specimen geocoordinates and elevations to summarize the extent and timing of past collecting efforts for small mammals in woodland habitats across the Madrean Sky Islands. In total, 97 species of small mammals (39 bats, 54 rodents, 4 shrews) from 9,541 specimens were collected from 1884 to 2023. Of these historical specimens, 79% come from five sky islands (Chiricahuas, Pinaleños, Huachucas, Animas, and Santa Catalinas) and only 25 sky islands in the Madrean system have any recorded specimens. Mexico’s 25 sky islands are mostly unsampled (only the San Luis, Sierra dos Ajos, and Sierra La Mariquita have any specimens) and several of Arizona’s larger sky islands have fewer than 40 specimens (Galiuros, Canelo Hills, Santa Teresas, Mules, and Dragoons). A large majority of small mammal specimens (87%) were collected prior to 1980, meaning they were collected without DNA/RNA preservation as a priority. This distributional summary is the current basis for all derived biodiversity knowledge of Madrean Sky Island small mammals, illustrating clear gaps regarding most species of woodland-dwelling bats, rodents, and shrews. This work lays the foundation for future fieldwork and voucher specimen preservation in the Madrean Sky Islands, especially from undersampled mountains where biodiversity assumptions are unconfirmed.

Keywords: biogeography, Chiroptera, Eulipotyphla, museum specimens, natural history collections, Rodentia, sky islands, species richness

Introduction

“Mountains—those islands in the sky surrounded by a sea of desert.”

– Edward Abbey (Desert Solitaire, 1968: 129)

Envisioning forested desert mountains as ‘sky islands’ dates back to Heald’s (1967) account of Arizona’s Chiricahua Mountains, and Abbey’s (1968) ode to Utah’s La Sal Mountains. The metaphor has inspired generations of naturalists, contributing to the foundations of metapopulation ecology and habitat island biogeography (reviewed in Matthews 2021). The Madrean Sky Island Archipelago, containing the Chiricahuas among other mountains, is globally one of the most extensive of the so-called ‘stepping stone’ archipelagos that connect two mountain chains (Warshall 1995). Spanning between the Colorado Plateau and the Sierra Madre Occidental, the Madrean Archipelago contains somewhere between 40 (Warshall 1995) and 65 (Moore et al. 2013) forested mountains, each one isolated from others by the surrounding Sonoran and Chihuahuan Deserts. Here we present a 54-mountain definition of the Madrean Sky Islands designed for studying woodland-dwelling species relative to the surrounding ‘sea’ of greater aridity (desert grassland and desert scrub; Fig. 1, Table 1).

Figure 1.—

Figure 1.—

Open in a new tab

Map of the Madrean Sky Island region showing the 54 mountains defined in this study, with numbered labels corresponding to Table 1. The Colorado Plateau lies directly North of the figure and the Sierra Madre Occidental lies South/Southeast of the figure.

Table 1.—

List of all 54 mountains in the Madrean Sky Islands region as defined in this study, with the label numbers corresponding to Figure 1.

Label Mountain Area above 1,600 meters (km2) Maximum elevation (m)
1 Superstition 43.28 1902
2 Pinal 144.36 2392
3 Santa Teresa 327.76 2497
4 Santa Catalina 414.78 2806
5 Galiuro 501.57 2322
6 Pinaleño 747.2 3267
7 Peloncillo North 40.3 1982
8 Peloncillo Central 38.11 1982
9 Rincon 264.1 2646
10 Winchester 234.28 2315
11 Dos Cabezas 218.99 2485
12 Baboquivari 69.59 2262
13 Sierrita 24.12 1887
14 Santa Rita 341.77 2869
15 Whetstone 183.1 2331
16 Little Dragoon 39.81 2037
17 Dragoon 252.4 2275
18 Swisshelm 83.45 2181
19 Chiricahua 1769.87 2990
20 Peloncillo South 903.55 2095
21 Animas 739.77 2598
22 Little Hatchet 35.47 2017
23 Big Hatchet 109.32 2542
24 Alamo Hueco 89.17 2090
25 Atascosa 11.09 1948
26 Patagonia 201.43 2200
27 Canelo Hills 416.09 1968
28 Huachuca 357.84 2875
29 Mule 265.56 2241
30 Pajarito 8.89 1800
31 Sierra Chivato 192.43 2195
32 Sierra San Jose 123.24 2539
33 Sierra La Ceniza 13.61 1822
34 Sierra de Embudos 24.31 1921
35 San Luis 1722.44 2515
36 Sierra Avispas 5.39 1746
37 Sierra Cibuta 43.72 2079
38 Sierra El Pinito West 52.24 2264
39 Sierra La Mariquita 239.5 2500
40 Sierra de los Ajos 719.64 2616
41 Sierra La Madera 38.22 2044
42 Sierra Azul 296.54 2452
43 Sierra El Manzanal 28.7 1919
44 Cerro Bacoachi 3.79 1707
45 Sierra San Antonio 112.39 2213
46 Sierra Buenos Aires 94.7 2474
47 Sierra El Pinito East 37.42 2230
48 Cerro Bellotal 35.63 1942
49 Sierra Purica 11.5 1925
50 Sierra La Sandia 137.04 2179
51 Sierra Jacaral 224.8 1908
52 Sierra Aconchi 37.43 2190
53 Sierra de Oposura 133.79 2345
54 Sierra de Las Guijas 4.11 1823
Open in a new tab

Intersecting biomes make the Madrean Sky Islands a biodiversity hotspot for plants, insects, birds, mammals, and other taxa (Deyo et al. 2013). However, distributional knowledge of the Madrean biota remains highly incomplete. Historical efforts to characterize species distributions have focused on flora or fauna of the more accessible mountains (e.g. Santa Catalinas, Lange 1960; Moore et al. 2013; Chiricahuas, Duncan 1990;(Cahalane 1939); Huachucas, Hoffmeister and Goodpasters 1954; Bowers and McLaughlin 1996). A few individual species have also been studied across multiple mountains (e.g., Mexican Woodrat, Neotoma mexicana Sullivan 1994; Canyon Tree Frog, Hyla arenicolor Barber 1999; Transvolcanic Jay, Aphelocoma ultramarina McCormack et al. 2008; Yarrow’s Spiny Lizard, Sceloperous jarrovii Wiens et al. 2019; American Black Bear, Ursus americanus Atwood et al. 2011; Gould et al. 2022). Given recent losses of forested habitats in the Madrean Archipelago from fire, logging, and climate change (Williams et al. 2010; Yanahan and Moore 2019), there is growing urgency for a more integrative multi-species and multi-mountain understanding of the region’s biodiversity. Identifying gaps in the historical knowledge of species’ geographical and elevational distributions is therefore a high priority.

An accurate understanding of small mammal biodiversity is particularly imperative in the Madrean region given their potential to serve as zoonotic disease vectors (e.g., hantaviruses: Astorga et al. 2025; lung fungi: Salazar-Hamm et al. 2022) and indicators of ecosystem health (Pearce and Venier 2005; Leis et al. 2008; Russo et al. 2021). Understanding how many species live in a certain area and how their distributions are influenced by environmental changes can help predict the risk of pathogen spillover into human populations (Olival et al. 2017; Johnson et al. 2020; García-Peña and Rubio 2024). Unfortunately, substantial gaps exist in mammal distributional knowledge across this region. This “dearth of published data” on mammal biodiversity was highlighted for the Madrean Archipelago ca. 20 years ago (Koprowski et al. 2005: 412), with particular sampling gaps noted for native species of bats, rodents, and shrews as compared to larger ungulates and carnivores. While the endangered Mount Graham Red Squirrel (Tamiasciurus fremonti grahamensis) has been intensively studied (e.g., Merrick et al. 2021; Mahoney and Pasch 2024), most small mammal species in this region are comparatively unknown. The efforts of Findley et al. (1976) and Hoffmeister (1986) in assembling the Mammals of New Mexico and Mammals of Arizona, respectively, still represent the best distributional knowledge for many Madrean sky-island species despite their outdated taxonomies. Advances in the digitization of museum specimen records over the past two decades now present an outstanding opportunity to revisit prior work on the distribution of Madrean Sky Island small mammals, aiming to set a baseline for future studies.

We seek to address three related questions. First, what is the most functional definition of the Madrean Sky Islands for woodland-dwelling species; i.e., how many separate mountaintop islands does this system contain? Varying answers have been given to this seemingly simple question, calling for a specific geospatial definition. Second, where in the Madrean Sky Islands have small mammal species been historically studied versus not? Museum voucher specimen records are the most reliable means of addressing this question, given that small mammals are cryptic and mostly nocturnal and therefore difficult to identify from photographs. Third, which mountains are historically understudied relative to their size? Here we present evidence of the need for targeted field surveys to fill geographic, elevational, and taxonomic gaps of small mammals in the Madrean Sky Islands.

Methods

Study system

To define the Madrean Sky Islands study system, we used NASA’s ASTER Global Digital Elevation Model (DEM) within ARCGIS Pro v3.1.2 to delimit all areas above 1,600 meters (m) in the study region. This elevational cutoff was based on Moore et al. (2013) and roughly corresponds to the lowest elevation of Madrean oak-woodland habitat. The 1,600-m cutoff served as a starting point to isolate the forested mountain areas south of the Superstitions, east of the Baboquivaris, west of the Alamo Hueco and Big Hatchets, and north of the Sierra de Las Guijas. We then performed manual cleanup to refine this delimitation, as follows. First, we eliminated mountain ranges that have less than 1 kilometer squared (km2) area above 1,600 m (i.e., Sierra El Humo and Sierra San Juan). Then, we determined whether contiguous areas of land above 1,600 m constituted separate ranges in the context of the dispersal propensity of small mammals. To resolve any boundary conflicts, we used satellite imagery (Google Earth; Airbus imagery 5/21/2022–11/19/2023) to identify areas of discontinuous oak woodland coverage. For example, while the Galiuro and Winchester Mountains remain continuous in terms of area above 1,600 m, a significant gap in oak woodland habitat between them constitutes separating these sky islands. Small, discontinuous fragments of area above 1,600 m were assigned to the nearest sky island if they were within 5 km of each other; otherwise these fragments were eliminated from the study. Using these criteria, we defined 54 isolated forest habitats as constituting the sky islands of the Madrean system and formalized this definition as individual DEM raster and vector polygon layers (see Supplementary Data SD1-SD2).

Within this 54-mountain definition, we made one notable exception to separately retain the Canelo Hills and Huachuca Mountains, even though they are continuously connected in terms of both land area above 1,600 m and woodland habitat. The Huachucas reach elevations high enough to maintain pine and mixed conifer forests while the Canelo Hills comprise almost exclusively of oak woodland without elevations above 2,000 m. We reasoned that combining historical records from the steep and heterogenous Huachucas with those from the lower and more uniform Canelo Hills would be misleading. The stark difference in ruggedness as well as the dense historical sampling of the Huachucas relative to the few Canelo Hills specimens also supported this decision (see results).

Specimen records

We summarized natural history specimen records from public databases to assess the historical sampling efforts of small mammals in the Madrean Sky Islands. Searches of the Global Biodiversity Information Facility (GBIF) database (23 January, 2024) allowed us to maximize the number of collections and specimen records included in the study. We focused our searches upon the taxonomic orders of Rodentia, Chiroptera, and Eulipotyphla using a geographic boundary polygon that encompassed the entire study region (bounded at four points: 33.76114, −111.7944; 27.05179, −113.06772; 27.00817, −105.62368; 34.39024, −106.02152). This approach returned 122,324 total specimens (bats: n=21,118; rodents: n=99,844; shrews: n=1,362), including those collected in both lowland and highland habitats. We intersected all specimen records with the polygons of 54 mountains in the Madrean Sky Islands to filter records to areas above 1,600 m, which reduced the number of specimens to 10,340 total (bats: n=2,260; rodents: n=7,901; shrews: n=179). We also excluded specimens with a coordinate uncertainty of 10 km or greater to maximize both the amount and accuracy of data included in this study, which further reduced the number of specimens to 9,802 (bats: n=2,194; rodents: n=7,429; shrews: n=179). This approach struck a balance between retaining known highland species while excluding known lowland species. Without this uncertainty-based filtering, the coordinates of a strictly lowland species might mistakenly be interpreted to the highest reaches of a mountain (e.g., Chaetodipus penicillatus at 2,500 m in the Huachuca Mountains, MSB:Mamm:184809). Among the remaining specimens that had a coordinate uncertainty <10 km, most (n=8,224) lacked any elevation data. To resolve this, we interpolated elevation using the ASTER DEM layer, favoring the interpolated value even among specimens that did have reported elevation data to ensure consistency in subsequent analyses.

A final filtering step removed specimens that were only identified at the genus level, given that this lack of taxonomic certainty may indicate reliability issues with the record. Some specimens lacked a specific identification in GBIF’s ‘Species’ data field, but did contain a complete specific identification in the ‘verbatimScientificName’ field. For these specimens (n=840, most from University of Michigan Museum of Zoology), we added their verbatim identification to the ‘Species’ data field for inclusion in our study. For graphing specimen collections over time, we excluded specimens that lacked a date altogether or had been assigned a range of dates spanning multiple years (n=383; typically listed as either January 1st, 1800 or January 1st, 1900). Finally, we updated the taxonomy of the specimens to match that of the Mammal Diversity Database v2.2 (MDD 2025), which included minor spelling changes (e.g., Erethizon dorsatus to Erethizon dorsatum), re-assigned genera (e.g., Aeorestes cinereus to Lasiurus cinereus) and synonymizations (e.g., Corynorhinus rafinesquii pallescens to Corynorhinus townsendii).

Sampling bias

To identify potential biases of where historical collectors have focused their efforts in the Madrean Sky Islands, we evaluated the ruggedness, area, and elevational profile of each mountain. We calculated the Terrain Ruggedness Index for the individual DEMs, which returned the number of raster pixels in each of 6 ruggedness categories (sensu Riley et al. 1999). We used these pixel counts to generate a weighted average and standard deviation of ruggedness per sky island. To quantify the specimen-area ratio (s/km2) as a measure of “sampling density”, we first intersected the individual polygons with the filtered rodent, bat, and shrew specimen records to identify historical specimens from each sky island individually. Then, we used the attribute table for each mountain’s DEM to calculate the area, multiplying the total number of pixels >1,600 m for each sky island by the size of one pixel in m2 (adjusted for longitude and latitude). Lastly, we compared the elevational profiles of mountains using the DEMs to quantify the amount of land area (km2) per unit elevation (m). We applied five vegetation categories to these elevational profiles, following the guidance of Bennett et al. 2004, as follows: oak woodland from 1,600–1,800 m; pine-oak woodland from 1,801–2,200 m; pine forest from 2,201–2,550 m; mixed conifer forest from 2,551–3,000 m; and limber and alpine fir forests >3,000 m (only found in the Pinaleño Mountains). We used these vegetation categories to estimate the total land area available to each habitat type across the sky islands, excluding sky islands that contained less than 1 km2 of a given habitat category. Also excluded from these analyses was the highly variable chaparral habitat that occurs between 1,250 m and 2,590 m depending on mountain slope and aspect.

Results

Study system

Using our definition of the Madrean Sky Islands as containing 54 unique mountain ranges with woodland habitats above 1,600 m, we find that this region encompasses 13,210 km2 of isolated habitat within a sea of desert in the Southwestern United States and Northern Mexico (Fig. 1; Table 1). The international border roughly bisects these sky islands, with 25 mountains lying primarily in the state of Sonora (SO) compared to 24 in Arizona (AZ) and 5 in New Mexico (NM). The highest elevations in the Madrean Sky Islands are found at the peaks of the Pinaleño Mountains (AZ) while the largest sky island by area is the Chiricahua Mountains (AZ). The smallest sky island by both maximum elevation and area is Cerro Bacoachi (SO).

Comparison of elevational profiles (Fig. 2) reveals that each mountain range varies greatly in slope, aspect, maximum elevation, and total area. For example, the Peloncillo South Mountains (max. elevation=2,095 m, area=921 km2) contain extensive oak woodland habitat and limited pine-oak woodland, but fail to reach the higher elevations needed for mixed conifer forests. On the other hand, the Rincon Mountains (max. elevation=2,646 m, area=265 km2) are less than one third of the area of the Peloncillo South Mountains, yet reach high enough elevations to support mixed conifer forests. In terms of habitat variation, even though the Chiricahua Mountains (max. elevation=2,990 m, area=1752 km2) are 90% larger than the Peloncillo South Mountains overall, the latter contain 48 km2 more oak woodland.

Figure 2.—

Figure 2.—

Open in a new tab

Elevation per unit land area above 1,600 m for six mountains in the Madrean Sky Island region of Arizona. The five ecoregions are overlaid with their typical upper range limit relative to elevation.

Specimen records

After filtering specimen records for quality and locational accuracy, we find that 9,541 specimens of small mammals have been collected in the Madrean Sky Islands (Fig. 3). Only 25 of the 54 mountains have one or more historical specimen records. Of those, 7,635 specimens were collected in Arizona, 1,761 in New Mexico, and 145 in Sonora (no specimens were collected in the small Chihuahuan regions of the Madrean Sky Islands). In the order Rodentia, we identify 7,228 specimens comprising six families (Cricetidae, Erethizontidae, Geomyidae, Heteromyidae, Muridae, and Sciuridae), 22 genera, and 54 species (Fig. 4). These specimens were collected from 25 sky islands. In the order Chiroptera, we identify 2,135 specimens comprising five families (Molossidae, Mormoopidae, Natalidae, Phyllostomidae, and Vespertilionidae), 20 genera, and 39 species (Fig. 5). These specimens were collected from 20 sky islands. Under the order Eulipotyphla, we identify just 178 specimens from one family (Soricidae), two genera, and four species (Notiosorex crawfordi, Sorex arizonae, Sorex monticolus, and Sorex vagrans) collected from 9 sky islands (see Supplementary Data SD3-SD5).

Figure 3.—

Figure 3.—

Open in a new tab

Historical voucher specimens sampled from areas greater than 1,600 m in the Madrean Sky Islands, as separated by (A) Chiroptera, and (B) Eulipotyphla and Rodentia.

Figure 4.—

Figure 4.—

Open in a new tab

Frequency of rodent specimens sampled at >1,600 m in the Madrean Sky Islands (A) by genus with colors corresponding to taxonomic families, and (B) by the 10 most frequently sampled species. Both charts are on a square root scale.

Figure 5.—

Figure 5.—

Open in a new tab

Frequency of bat specimens sampled at >1,600 m in the Madrean Sky Islands (A) by genus with colors corresponding to taxonomic families, and (B) by the 10 most frequently sampled species (Leptonycteris yerbabuenae and Myotis auriculus are equally the 10th most common). Both charts are on a square root scale.

Considering the timing of collecting activity in the Madrean Sky Islands, the earliest specimen was collected in the Huachuca mountains by E. A. Mearns in 1884 (FMNH:Mammals:4943) and only 265 specimens were collected across the sky islands before 1900. The majority of specimens (5,922; 65%) were collected between 1950 and 1980, encompassing 84% of bats, 59% of rodents, and 48% of shrews ever collected in the Madrean Sky Islands (Fig. 6; Supplementary Data SD6). In the ensuing 43-year period (1981–2023), only 1,142 specimens were collected in this region (12% of the total).

Figure 6.—

Figure 6.—

Open in a new tab

The accumulation of specimens over time acrossChiroptera, Rodentia, and Eulipotyphla. The colored lines begin where the first specimen for each order was collected (1884, 1894, and 1931 respectively) and the vertical dashed line indicates the year 1980.

Preserved specimens of small mammals from the Madrean Sky Islands are presently housed in 43 different natural history collections, mainly in the southwestern United States. The most extensive of these collections is the Museum of Southwestern Biology (University of New Mexico) with 3,378 specimens, followed by the University of Arizona Museum of Natural History (1,954 specimens), University of Michigan Museum of Zoology (1,090 specimens), and Arizona State University Natural History Collection (394 specimens).

Sampling bias

We find that sky-island area and maximum elevation significantly influence where historical collectors focused their sampling. Collection efforts leading to the preservation of small mammal specimens have been concentrated in a few, large sky islands in the Madrean system, resulting in a positive log-linear relationship between sky-island size and specimens (Fig. 7A; Y=−3.524+1.354X, R2=0.3578, P=0.0016). Similarly, we find a positive linear relationship between the maximum elevation of a sky island and the specimen-area ratio (Fig. 7B; Y=−3.664+0.0017X, R2=0.3762, P=0.0011). Interestingly, we found no significant relationship between the specimen-area ratio and sky-island ruggedness (R2=0.0791, P=0.1731), even though ruggedness is positively related to maximum elevation (R2=0.2901, P=0.0055). The average specimen-area ratio across all 54 sky islands is 0.32 s/km2, or 0.68 s/km2 after excluding the 29 mountains without specimens (Supplementary Data SD7-SD8). The most densely sampled sky islands are the Huachucas (3.74 s/km2), Santa Catalinas (2.15 s/km2), Chiricahuas (1.76 s/km2), Pinaleños (1.71 s/km2), and Santa Ritas (1.33 s/km2); only three other sky islands have >1.0 s/km2 (the Animas, Alamo Hueco, and Pinal Mountains). We recover notably low sampling density on several large sky islands: the San Luis Mountains (2,515 km2, 0.09 s/km2), the Sierra de los Ajos (735 km2, 0.12 s/km2), Galiuros (503 km2, 0.07 s/km2), and Canelo Hills (423 km2, 0.04 s/km2).

Figure 7.—

Figure 7.—

Open in a new tab

(A) Biases in historical sampling as measured by the number of small mammal specimens relative to mountain area above 1,600 meters. Both axes are shown on a log10 scale, indicating a multiplicative relationship between sky-island area and number of specimens. (B) Biases in historical sampling as measured by the density of sampling relative to sky-island maximum elevation on a linear scale. In both panels, the 29 sky islands without any historical specimens were excluded and sky islands with a specimen-area ratio greater than 1 are shown in brown.

Discussion

The dearth of published research highlighted by Koprowski et al. (2005) has remained a reality for most species of small mammals in the Madrean Sky Islands. However, thanks to two decades of advances in the digitization of museum specimen records, we are now able to synthesize data from 43 natural history collections to identify gaps in the distributional knowledge of this diverse small mammal fauna. The ‘virtual flora and fauna’ of Deyo et al. (2013) was the last attempt to summarize distributional knowledge of taxa across the Madrean Sky Islands, but the surprisingly few records of mammals (n=217) they recovered was likely an artifact of sparse georeferencing efforts at that time. Their work preceded the maturation of integrated databases like VertNet and GBIF (Guralnick et al. 2016; Heberling et al. 2021), which now enable greater synthesis of historical specimen records. In contrast, our search and quality filtering steps revealed 9,541 georeferenced voucher specimens of small mammals (Fig. 3). These specimens are allocated to 97 species from 12 families of bats, rodents, and shrews (Figs. 4, 5). The over 40x growth in digitized records from this region illuminates several biases in historical collecting, including 29 of 54 mountains in the Madrean Sky Islands for which no small mammals have ever been vouchered.

To put this sampling history in regional context, a study by Jones et al. (2021) in the neighboring Gila Mountains region of New Mexico found 12,505 specimen records from 77 species of bats (n≈2,500), rodents (n≈9,250; totals estimated from figures), and shrews across nearly twice the study area (24,383 km2 versus 13,210 km2 in the Madrean Sky Islands). Both regions experienced similar peaks of collecting activity prior to 1980, but the large pulse of Gila collecting from 2012–2020 (led by the Museum of Southwestern Biology) differs from the comparatively dormant collecting in the Madrean Sky Islands over the same period (Fig. 6 compared to fig. 6c in Jones et al. 2021). Critically, the paucity of post-1980 specimens from the Madrean Sky Islands means that frozen tissues and holistic specimen preparations are mostly unavailable (both preservation modes were popularized subsequently; Galbreath et al. 2019). Overall, Madrean small mammals have been less extensively studied than those of the Gila Mountains despite the former’s ~2x higher density of species per km2 (0.0072 versus 0.0032). Recent efforts to re-survey the Santa Catalina Mountains using holistic preservation (2021–2023 led by the Arizona State University Natural History Collections; see Rowsey et al. this volume) have begun bridging this shortfall, but parallel efforts on nearby sky islands are sorely needed.

Digitally defining the mountains of the Madrean Sky Islands was a key step in this study that now enables comparable future studies in this biodiverse region (Fig. 1). Querying how woodland-dwelling taxa of different kinds have responded to past climatic changes in this region, particularly the cycles of forest expansion/contraction since the end-Pliocene (Betancourt et al. 1990), is now more replicable given the included geospatial resource (Supplementary Data SD1-SD2). However, the 54-mountain definition of the Madrean Sky Islands presented here differs from three previous totals: “about 40” in Warshall (1995: 6); “55 ranges/complexes” in Deyo et al. (2013: 296); and “~65 mountain ranges” in Moore et al. (2013: 148). Those differences, and some nuances of our own definition, underscore the inherent subjectivity of dividing a complex, rugged landscape into discrete units. We followed the lead of Moore et al. (2013) in choosing an initial definition of 1,600 m, but then deviated from that work by joining several of the mountains that they also acknowledged were part of ‘Madrean Sky Island complexes.’ The result better reflects the ability of woodland-dwelling small mammals to disperse through continuous oak woodland (Sutherland et al. 2000; Selonen and Wistbacka 2017). For example, we joined the Patagonia and San Antonio Mountains, which in Moore et al. (2013) appear to only be separated by the U.S.-Mexico international border (Supplementary Data SD9). In the case of the mountain complex consisting of Chiricahua, Pedregosa, Swisshelm, and Dos Cabezas, these ranges are connected by varying degrees of oak woodland and so under that criterion could be considered one unit (as Warshall 1995 chose to do). We chose to join the Chiricahua and Pedregosa Mountains owing to their elevational and forest connectivity above 1,600 m while separating the Swisshelm and Dos Cabezas Mountains since they lack such elevational connectivity. Satellite imagery does show sparse oak woodland growth below 1,600 m connecting these latter two mountains from the Chiricahuas (Google Earth; Airbus imagery 5/21/2022–11/19/2023), but we reasoned that this thin forest likely presents a strong enough dispersal barrier to warrant separate mountain recognition.

We found clear patterns of uneven among-mountain sampling of small mammals across the Madrean Sky Islands that makes accurate biodiversity assessments difficult. Past collecting efforts have concentrated mainly on 8 sky islands that collectively contain 87% of known rodent, bat, and shrew specimens. These more densely sampled mountains all contain >1.0 specimens/km2 (Fig. 7A), consistent with the expectation of positive specimen-area relationships if larger mountains attract more collecting efforts. However, the comparatively sparse sampling of other large sky islands (e.g., San Luis and Sierra de los Ajos in Mexico, Galiuros and Canelo Hills in the U.S.) indicates that mountain area is not the only factor explaining historical collecting. Indeed, the maximum elevation of a mountain is also strongly associated with increased sampling density (Fig. 7B), perhaps in part owing to the development of paved roads to areas above 1,600 m on some of the taller mountains (e.g. the Pinaleños, Santa Catalinas, and Santa Ritas). Road accessibility, a known bias in biodiversity surveys (Kadmon et al. 2004; Monsarrat et al. 2019), was likely critical in promoting surveys of mountains with established roads. Yet roads may also threaten biodiversity by promoting land conversion (Bennett 2017; Barrientos et al. 2021). Thus, sky-island access roads may paradoxically be associated with both increased biodiversity knowledge (as measured in specimens and resulting publications) and potential threats, a tradeoff that land managers must closely consider.

Even for sky islands with comparatively dense historical sampling, small mammal diversity is still likely underestimated. For example, the Huachuca Mountains have the highest sampling density of any Madrean Sky Island at 3.67 s/km2, yet over a century of fieldwork on the mountain failed to find the Cliff Chipmunk (Neotamias dorsalis) until a breeding population was identified in 2007 (Cudworth and Koprowski 2010). Furthermore, a 2023 field survey of the Santa Catalina Mountains (2.15 s/km2) identified the first specimen of the Fulvous Harvest Mouse (Reithrodontomys fulvescens) ever recorded on that sky island (Rowsey et al. this volume). We suggest that the status of the Madrean Sky Islands as a ‘biogeographic crossroads’ (Spector 2002) raises the likelihood that additional species at the northern or southern edge of their distributions have gone undetected on multiple mountains. Surveying elevational gradients across adjacent sky islands may present a useful test of the hypothesis that species have ‘rare edges and abundant cores’ (Brown 1984; Martin et al. 2024). Future work that mitigates the identified sampling biases for small mammals may be able to accomplish this goal.

Comparing elevational profiles among sky islands shows striking heterogeneity in the extent of different forest types across mountains (Fig. 2), suggesting that biodiversity knowledge gained from one sky island might not apply to all others. For example, the Peloncillo South and Patagonia Mountains have similar maximum elevations (2,095 and 2,200 m, respectively) but the former contains ~4.5x more forested area than the latter across a sloping oaken plateau. These disparate elevational profiles are associated with more total specimens (587 vs. 95, respectively) and more apparent species diversity in the former than the latter (27 rodent, 19 bat, and 1 shrew species compared to 12, 5, and 0). While this positive correlation between sky-island area and species richness is consistent with Frey et al. (2007), no systematic method exists for identifying which species will be lost and which will be retained as the sky-island area shrinks. Hence, we expect each additional sky island for which the resident small mammal fauna is surveyed adds value to the total biodiversity knowledge of the Madrean system.

Potential biases in our study pertain to three issues: the accuracy of digital specimen records, unevenness in digitization efforts, and the historical methods used to obtain those specimens. First, while online biodiversity data aggregators like GBIF have advanced the ability to synthesize specimen records globally (Heberling et al. 2021), GBIF data requires careful validation to correct errors and outdated taxonomies (Zizka et al. 2020). Because 94% of the specimens in our study’s initial download predate the widespread use of handheld GPS devices (Kumar and Moore 2002), their coordinate data presumably came from secondary georeferencing using field notes and locality descriptors to assign latitude and longitude. Our filtering based on coordinate uncertainty thresholds aimed to alleviate specious records in the dataset, but this approach still relies on the accuracy of initial georeferencing (Murphey et al. 2004). Second, unevenness among natural history museums in the extent and accuracy of specimen digitization is another source of bias. Although the GBIF search returned relevant results from most major collections, we found that expected specimens from the American Museum of Natural History (AMNH) were completely absent. We identified 921 small mammal specimens from a direct search of the AMNH Mammal Collection that were likely collected above 1,600 m in the Madrean Sky Islands (based on the locality data field), but georeferencing these specimens was beyond the scope of our study and unlikely to alter the main findings (a majority of specimens are from the Chiricahuas, Pinalenos, Santa Catalinas, and Huachucas, see Supplementary Data SD10). Lastly, the lack of transparency regarding which trapping techniques were used to capture specimens on each sky island presents another bias. For example, mountains surveyed using Museum Special snap traps may yield greater diversity and abundance of small mammals per trapnight than those using Sherman live traps (Eulinger and Burt 2011). The unevenness of bat-netting efforts is another potential bias; however, bat specimens were recovered from most of the same mountains as non-volant taxa (Fig. 3) and bats and rodents have had roughly parallel specimen accumulation curves through time (Fig. 6). The apparent rarity of shrews likely reflects the infrequent use of pitfall traps, drift fences, and other methods (Maddock 1992), rather than actual low abundances in the Madrean Sky Islands.

Taxonomic changes and historical difficulties in identifying small mammals present other challenges to this study. The external measurements that are often used to differentiate related species can be prone to error, potentially leading to misidentification (Blackwell et al. 2006). Indeed, in our review of specimen records, we identified 11 species whose identification requires further investigation (Supplementary Data SD11). For example, the 18 specimens identified as Leptonycteris curasoae in the Madrean Sky Islands are presumably either L. nivalis or L. yerbabuenae, given that L. curasoae is endemic to Colombia and Venezuela (Cole and Wilson 2006), but making this determination would require physically examining those specimens (all housed in the MSB Mammal Collection). Similarly, our dataset includes 1 specimen of Peromyscus beatae, a species of deer mouse native to central Mexico with no known distribution in the Mexican portion of the Madrean Sky Islands (Álvarez-Castañeda et al. 2017), indicating that this specimen may be a misidentified individual of P. boylii (Tiemann-Boege et al. 2000). So too with the 11 specimens of P. truei that are recorded from sky islands in Arizona (Chiricahuas and Santa Catalinas) when their known distribution is thought to be restricted to the Animas Mountains of New Mexico south into the Sierra Madre Occidental (Hoffmeister 1981; Cook 1986). Investigation is needed to determine if these specimens constitute a modern range expansion, historical undersampling, or previous misidentification. These issues highlight the steadfast importance of museum specimens as physical records, each one providing a critical opportunity to revisit such issues, collect new data (e.g., measurements, DNA sequences), and thereby gain new insights even hundreds of years later.

Overall, we find that woodland-dwelling small mammals are surprisingly understudied throughout the Madrean Sky Islands. These forested mountains exist at the crossroads of the Nearctic and Neotropical biogeographic realms, which presents a diverse natural laboratory of repeated forest isolation and reconnection for studying how ecological and evolutionary processes unfold (Marshall 1957, Warshall 1995, Spector 2002, Koprowski et al. 2005, Deyo et al. 2013, Moore et al. 2013). We find that only 8 of the 54 mountains have had substantial historical sampling of voucher specimens, and 29 mountains are apparently still unsampled. The multiplicity of threats now impacting this region from fires, drought, and human disturbance signals an imperative to document basic biodiversity patterns across the sky islands (Misztal and Hansen 2013; Peters et al. 2018; Yanahan and Moore 2019; Love et al. 2023). Geopolitical conflicts such as the construction of a steel wall along the US-Mexico international border, bisecting as many as six sky islands, also impact biodiversity in this region. This wall directly blocks dispersal routes for larger mammals and indirectly impacts smaller mammals via increased light pollution, human presence, and road infrastructure (Harrity et al. 2024; Marín and Koprowski 2025). Increased fire severity, notably in more mesic and high elevation biomes, has also played a role in the degradation of sky island habitat, which has the potential to decrease diversity among small mammal communities (O’Connor et al. 2014; Villarreal et al. 2019; Culhane et al. 2022). Therefore, future studies should focus on elevational sampling in un- or under-sampled sky islands, particularly in Mexico where only two mountains (San Luis, Sierra de Los Ajos) have digital records of >10 small mammal specimens. So too should future studies focus on cryopreserving tissues and preparing holistic specimens (sensu Galbreath et al. 2019) since even the comparatively well-sampled sky islands mostly lack post-1980 collection efforts. Having multiple thoroughly sampled elevational gradients in the Madrean Sky Islands would enable replicated tests of which factors are the most important determinants of species richness (Chen et al. 2017) and metacommunity structure (Presley et al. 2012), two long-standing eco-evolutionary questions. Finally, work is needed to query the population-genetic histories of montane small mammals in this system, particularly how differing dispersal abilities affect gene flow between conspecific populations on different mountains. Apt candidates for this research would be some of the most common sky-island species in museums, such as Peromyscus boylii or Corynorhinus townsendii.

Acknowledgments

We owe thanks to the generations of mammalogists that have contributed invaluable specimens from which we derive small mammal biodiversity knowledge in the Madrean Sky Islands. Dakota M. Rowsey helped to inspire and guide this work. Special thanks to Luisa Zamora Chavez, Alexander Fenlon, Gilma De Leon, Jessica Neumeier, Ryan Nguyen, and Morgan Pierce for helpful comments that improved this manuscript. Funding support was provided by the National Institutes of Health (1R35GM156919 to NSU) and Arizona State University, as well as the American Society of Mammalogists (Grant-in-Aid of Research to DCR).

Footnotes

Supplementary Data

Supplementary Data SD 1.—54 individual sky island vector polygon shape layers and supporting files for use in GIS software.

Supplementary Data SD 2.—54 individual sky island DEM raster layers and supporting files for use in GIS software.

Supplementary Data SD 3.—Table of all Chiroptera specimens included in the study with original GBIF data fields as well as columns added during the study (e.g., Species_MDD).

Supplementary Data SD 4.—Table of all Eulipotyphla specimens included in the study with original GBIF data fields as well as columns added during the study (e.g., Species_MDD).

Supplementary Data SD 5.—Table of all Rodentia specimens included in the study with original GBIF data fields as well as columns added during the study (e.g., Species_MDD).

Supplementary Data SD 6.—Combined table of all specimens included in the study with the exception of specimens that lacked a date altogether or had been assigned a range of dates spanning multiple years (n=383).

Supplementary Data SD 7.—Data matrix summarizing area, mean ruggedness, maximum elevation, and specimen counts for each sky island.

Supplementary Data SD 8.—Raw area and ruggedness data for each sky island.

Supplementary Data SD 9.—Changes made to the mountains included in the Madrean Sky Islands compared to Moore et al. 2013.

Supplementary Data SD 10.—Table of specimens from the American Museum of Natural History that are likely relevant to the study, but lack geocoordinates.

Supplementary Data SD 11.—List of 11 species found in the Madrean Sky Islands that need further examination. These observations fall outside of the currently understood ranges of these species. Both P. truei and P. difficilis are known to occur on specific sky islands, however, some specimens have been reported outside of these known mountains, warranting additional verification. Specimens of L. ega and M. keenii likely belong to L. xanthinus and M. evotis respectively, but verification of specimens is needed in those cases as well.

References

  1. Abbey E. 1968. Desert Solitaire. University of Arizona Press. [Google Scholar]
  2. Álvarez-Castañeda ST, Álvarez T, González-Ruiz N. 2017. Keys for Identifying Mexican Mammals. JHU Press. [Google Scholar]
  3. Atwood TC, Young JK, Beckmann JP, Breck SW, Fike J, Rhodes OE, Bristow KD. 2011. Modeling connectivity of black bears in a desert sky island archipelago. Biol Conserv. 144(12):2851–2862. doi: 10.1016/j.biocon.2011.08.002. [DOI] [Google Scholar]
  4. Barber PH. 1999. Phylogeography of the canyon treefrog, Hyla arenicolor (Cope) based on mitochondrial DNA sequence data. Mol Ecol. 8(4):547–562. doi: 10.1046/j.1365-294x.1999.00593.x. [DOI] [PubMed] [Google Scholar]
  5. Barrientos R, Ascensão F, D’Amico M, Grilo C, Pereira HM. 2021. The lost road: Do transportation networks imperil wildlife population persistence? Perspect Ecol Conserv. 19(4):411–416. doi: 10.1016/j.pecon.2021.07.004. [DOI] [Google Scholar]
  6. Bennett PS, Kunzmann MR, Graham LA. 2004. Mar 5. Descriptions of Arizona Vegetation Represented on the GAP Vegetation Map. [Google Scholar]
  7. Bennett VJ. 2017. Effects of Road Density and Pattern on the Conservation of Species and Biodiversity. Curr Landsc Ecol Rep. 2(1):1–11. doi: 10.1007/s40823-017-0020-6. [DOI] [Google Scholar]
  8. Betancourt JL, Van Devender TR, Martin PS. 1990. Packrat Middens: The Last 40,000 Years of Biotic Change. University of Arizona Press. [accessed 2025 Jul 18]. https://www.jstor.org/stable/j.ctv21wj578. [Google Scholar]
  9. Blackwell GL, Bassett SM, Dickman CR. 2006. Measurement Error Associated with External Measurements Commonly Used in Small-Mammal Studies. J Mammal. 87(2):216–223. doi: 10.1644/05-MAMM-A-215R1.1. [DOI] [Google Scholar]
  10. Bowers JE, McLaughlin SP. 1996. Flora of the Huachuca Mountains, a Botanically Rich and Historically Significant Sky Island in Cochise County, Arizona. J Ariz-Nev Acad Sci. 29(2):66–107. [Google Scholar]
  11. Brown JH. 1984. On the Relationship between Abundance and Distribution of Species. Am Nat. 124(2):255–279. doi: 10.1086/284267. [DOI] [Google Scholar]
  12. Cahalane VH. 1939. Mammals of the Chiricahua Mountains, Cochise County, Arizona. J Mammal. 20(4):418–440. doi: 10.2307/1374590. [DOI] [Google Scholar]
  13. Chen Z, He K, Cheng F, Khanal L, Jiang X. 2017. Patterns and underlying mechanisms of non-volant small mammal richness along two contrasting mountain slopes in southwestern China. Sci Rep. 7(1):13277. doi: 10.1038/s41598-017-13637-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Cole FR, Wilson DE. 2006. Leptonycteris curasoae. Mamm Species.(796):1–3. doi: 10.1644/796.1. [DOI] [Google Scholar]
  15. Cook J. 1986. Jun 30. The Mammals of the Animas Mountains and Adjacent Areas, Hidalgo County, New Mexico. Occas Pap. https://digitalrepository.unm.edu/occasionalpapers/11. [Google Scholar]
  16. Cudworth NL, Koprowski JL. 2010. First Record of a Range Extension of the Cliff Chipmunk (Tamias dorsalis) into the Huachuca Mountains. West North Am Nat. 70(3):418–420. doi: 10.3398/064.070.0316. [DOI] [Google Scholar]
  17. Culhane K, Sollmann R, White AM, Tarbill GL, Cooper SD, Young HS. 2022. Small mammal responses to fire severity mediated by vegetation characteristics and species traits. Ecol Evol. 12(5):e8918. doi: 10.1002/ece3.8918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Deyo NS, Van Devender TR, Smith A, Gilbert E. 2013. Documenting the Biodiversity of the Madrean Archipelago: An Analysis of a Virtual Flora and Fauna. USDA For Serv Proc. [Google Scholar]
  19. Duncan DK. 1990. Small Mammal Inventory of Chiricahua National Monument, Cochise County, Arizona. No. 30. Cooperative National Park Resources Studies Unit, School of Renewable Resources, University of Arizona. [Google Scholar]
  20. Eulinger KG, Burt MS. 2011. Comparison of Captures Between Sherman Live Traps and Museum Special Kill Traps. Southwest Nat. 56(2):241–246. doi: 10.1894/F14-PS-32.1. [DOI] [Google Scholar]
  21. Findley JS, Harris AH, Wilson DE, Jones C. 1976. Mammals of New Mexico. Q Rev Biol. 51(1):155–155. doi: 10.1086/409191. [DOI] [Google Scholar]
  22. Frey JK, Bogan MA, Yates TL. 2007. Mountaintop Island Age Determines Species Richness of Boreal Mammals in the American Southwest. Ecography. 30(2):231–240. [Google Scholar]
  23. Galbreath KE, Hoberg EP, Cook JA, Armién B, Bell KC, Campbell ML, Dunnum JL, Dursahinhan AT, Eckerlin RP, Gardner SL, et al. 2019. Building an integrated infrastructure for exploring biodiversity: field collections and archives of mammals and parasites. J Mammal. 100(2):382–393. doi: 10.1093/jmammal/gyz048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. García-Peña GE, Rubio AV. 2024. Unveiling the impacts of land use on the phylogeography of zoonotic New World Hantaviruses. Ecography. 2024(10):e06996. doi: 10.1111/ecog.06996. [DOI] [Google Scholar]
  25. GBIF.org (23 January 2024) GBIF Occurrence Download 10.15468/dl.x64dfs [DOI]
  26. GBIF.org (23 January 2024) GBIF Occurrence Download 10.15468/dl.td65sz [DOI]
  27. GBIF.org (23 January 2024) GBIF Occurrence Download 10.15468/dl.wttwuq [DOI]
  28. Gould MJ, Cain III JW, Atwood TC, Harding LE, Johnson HE, Onorato DP, Winslow FS, Roemer GW. 2022. Pleistocene–Holocene vicariance, not Anthropocene landscape change, explains the genetic structure of American black bear (Ursus americanus) populations in the American Southwest and northern Mexico. Ecol Evol. 12(10):e9406. doi: 10.1002/ece3.9406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Guralnick RP, Zermoglio PF, Wieczorek J, LaFrance R, Bloom D, Russell L. 2016. The importance of digitized biocollections as a source of trait data and a new VertNet resource. Database. 2016:baw158. doi: 10.1093/database/baw158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Harrity EJ, Traphagen M, Bethel M, Facka AN, Dax M, Burns E. 2024. USA-Mexico border wall impedes wildlife movement. Front Ecol Evol. 12. doi: 10.3389/fevo.2024.1487911. [accessed 2025 Jul 18]. https://www.frontiersin.org/journals/ecology-and-evolution/articles/10.3389/fevo.2024.1487911/full. [DOI] [Google Scholar]
  31. Heald WF. 1967. The Chiricahuas, Sky Island. Marguerite Bantlin Pub. [Google Scholar]
  32. Heberling JM, Miller JT, Noesgaard D, Weingart SB, Schigel D. 2021. Data integration enables global biodiversity synthesis. Proc Natl Acad Sci. 118(6):e2018093118. doi: 10.1073/pnas.2018093118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hoffmeister DF. 1981. Peromyscus truei. Mamm Species.(161):1–5. doi: 10.2307/3503851. [DOI] [Google Scholar]
  34. Hoffmeister DF. 1986. Mammals of Arizona. Tucson: University of Arizona Press. [Google Scholar]
  35. Hoffmeister DF, Goodpasters WW. 1954. The mammals of the Huachua Mountains, southeastern Arizona 24. Ill Biol Monogr V 24 No 1. [Google Scholar]
  36. Johnson CK, Hitchens PL, Pandit PS, Rushmore J, Evans TS, Young CCW, Doyle MM. 2020. Global shifts in mammalian population trends reveal key predictors of virus spillover risk. Proc R Soc B Biol Sci. 287(1924):20192736. doi: 10.1098/rspb.2019.2736. [DOI] [Google Scholar]
  37. Jones AK, Liphardt SW, Dunnum JL. 2021. An overview of the mammals of the Gila region, New Mexico. Therya. 12(2):213–236. doi: 10.12933/therya-21-1123. [DOI] [Google Scholar]
  38. Kadmon R, Farber O, Danin A. 2004. Effect of Roadside Bias on the Accuracy of Predictive Maps Produced by Bioclimatic Models. Ecol Appl. 14(2):401–413. doi: 10.1890/02-5364. [DOI] [Google Scholar]
  39. Koprowski JL, Edelman AJ, Pasch BS, Buecher DC. 2005. A Dearth of Data on the Mammals of the Madrean Archipelago: What We Think We Know and What We Actually Do Know. [Google Scholar]
  40. Kumar S, Moore KB. 2002. The Evolution of Global Positioning System (GPS) Technology. J Sci Educ Technol. 11(1):59–80. doi: 10.1023/A:1013999415003. [DOI] [Google Scholar]
  41. Lange KI. 1960. Mammals of the Santa Catalina Mountains, Arizona. Am Midl Nat. 64(2):436–458. doi: 10.2307/2422673. [DOI] [Google Scholar]
  42. Leis SA, Leslie DM, Engle DM, Fehmi JS. 2008. Small mammals as indicators of short-term and long-term disturbance in mixed prairie. Environ Monit Assess. 137(1–3):75–84. doi: 10.1007/s10661-007-9730-2. [DOI] [PubMed] [Google Scholar]
  43. Love SJ, Schweitzer JA, Woolbright SA, Bailey JK. 2023. Sky Islands Are a Global Tool for Predicting the Ecological and Evolutionary Consequences of Climate Change. Annu Rev Ecol Evol Syst. 54(Volume 54, 2023):219–236. doi: 10.1146/annurev-ecolsys-102221-050029. [DOI] [Google Scholar]
  44. Maddock AH. 1992. Comparison of Two Methods for Trapping Rodents and Shrews. [accessed 2025 Apr 14]. https://brill.com/view/journals/ijee/38/3-4/article-p333_14.xml. [Google Scholar]
  45. Mahoney SM, Pasch B. 2024. Evolutionary lability of food caching behaviour in mammals. J Anim Ecol. 93(7):862–875. doi: 10.1111/1365-2656.14093. [DOI] [PubMed] [Google Scholar]
  46. Mammal Diversity Database. 2025. Mammal Diversity Database (version 2.2) [data set]. Zenodo. Available from: 10.5281/zenodo.15659662 [DOI] [Google Scholar]
  47. Marín G, Koprowski JL. 2025. Mammals on the edge: how the border wall, roads, and riparian areas impact mammal populations along the US-Mexico boundary. Sci Total Environ. 988:179833. doi: 10.1016/j.scitotenv.2025.179833. [DOI] [PubMed] [Google Scholar]
  48. Marshall JT. 1957. Birds of Pine-Oak Woodland in Southern Arizona and Adjacent Mexico. Pac Coast Avifauna No32. [Google Scholar]
  49. Martin PR, Robinson OJ, Bonier F. 2024. Rare edges and abundant cores: range-wide variation in abundance in North American birds. Proc R Soc B Biol Sci. 291(2015):20231760. doi: 10.1098/rspb.2023.1760. [DOI] [Google Scholar]
  50. Matthews TJ. 2021. On The Biogeography of Habitat Islands: The Importance of Matrix Effects, Noncore Species, and Source-Sink Dynamics. Q Rev Biol. 96(2):73–104. doi: 10.1086/714482. [DOI] [Google Scholar]
  51. McCormack JE, Bowen BS, Smith TB. 2008. Integrating paleoecology and genetics of bird populations in two sky island archipelagos. BMC Biol. 6(1):28. doi: 10.1186/1741-7007-6-28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Merrick MJ, Morandini M, Greer VL, Koprowski JL. 2021. Endemic Population Response to Increasingly Severe Fire: A Cascade of Endangerment for the Mt. Graham Red Squirrel. BioScience. 71(2):161–173. doi: 10.1093/biosci/biaa153. [DOI] [Google Scholar]
  53. Misztal LW, Hansen L. 2013. Responding to Climate Change Impacts in the Sky Island Region: From Planning to Action. Proc RMRS. 2013:60–67. [Google Scholar]
  54. Monsarrat S, Boshoff AF, Kerley GIH. 2019. Accessibility maps as a tool to predict sampling bias in historical biodiversity occurrence records. Ecography. 42(1):125–136. doi: 10.1111/ecog.03944. [DOI] [Google Scholar]
  55. Moore W, Meyer WM, Eble JA, Franklin K, Wiens JF, Brusca RC. 2013a. Introduction to the Arizona Sky Island Arthropod Project (ASAP): Systematics, Biogeography, Ecology, and Population Genetics of Arthropods of the Madrean Sky Islands. Proc RMRS. 2013:144–168. [PMC free article] [PubMed] [Google Scholar]
  56. Murphey PC, Guralnick RP, Neufeld D, Ryan JA. 2004. Georeferencing of museum collections: A review of problems and automated tools, and the methodology developed by the Mountain and Plains Spatio-Temporal Database- Informatics Initiative (Mapstedi). [Google Scholar]
  57. NASA/METI/AIST/Japan Spacesystems, U.S./Japan ASTER Science Team. 2019. ASTER Global Digital Elevation Model NetCDF V003 [data set]. NASA EOSDIS Land Processes Distributed Active Archive Center. Available from: 10.5067/ASTER/ASTGTM_NC.003 [DOI] [Google Scholar]
  58. O’Connor CD, Falk DA, Lynch AM, Swetnam TW. 2014. Fire severity, size, and climate associations diverge from historical precedent along an ecological gradient in the Pinaleño Mountains, Arizona, USA. For Ecol Manag. 329:264–278. doi: 10.1016/j.foreco.2014.06.032. [DOI] [Google Scholar]
  59. Olival KJ, Hosseini PR, Zambrana-Torrelio C, Ross N, Bogich TL, Daszak P. 2017. Host and viral traits predict zoonotic spillover from mammals. Nature. 546(7660):646–650. doi: 10.1038/nature22975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Pearce J, Venier L. 2005. Small mammals as bioindicators of sustainable boreal forest management. For Ecol Manag. 208(1):153–175. doi: 10.1016/j.foreco.2004.11.024. [DOI] [Google Scholar]
  61. Peters R, Ripple WJ, Wolf C, Moskwik M, Carreón-Arroyo G, Ceballos G, Córdova A, Dirzo R, Ehrlich PR, Flesch AD, et al. 2018. Nature Divided, Scientists United: US–Mexico Border Wall Threatens Biodiversity and Binational Conservation. BioScience. 68(10):740–743. doi: 10.1093/biosci/biy063. [DOI] [Google Scholar]
  62. Presley SJ, Cisneros LM, Patterson BD, Willig MR. 2012. Vertebrate metacommunity structure along an extensive elevational gradient in the tropics: a comparison of bats, rodents and birds. Glob Ecol Biogeogr. 21(10):968–976. doi: 10.1111/j.1466-8238.2011.00738.x. [DOI] [Google Scholar]
  63. Riley SJ, DeGloria SD, Elliot R. 1999. Index that quantifies topographic heterogeneity. Intermt J Sci. 5(1–4):23–27. [Google Scholar]
  64. Rowsey DM, Smith SM, Zamora Chavez LJ, Rivera DC, Hess SC, Jones MF, Bucci ME, Mohammadian S, Alston JM, Baez JR, et al. In Prep. Revisiting the small mammal community of an iconic Madrean Sky Island (Santa Catalina Mountains, Arizona, USA). J Mammal. [Google Scholar]
  65. Russo D, Salinas-Ramos VB, Cistrone L, Smeraldo S, Bosso L, Ancillotto L. 2021. Do We Need to Use Bats as Bioindicators? Biology. 10(8):693. doi: 10.3390/biology10080693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Selonen V, Wistbacka R. 2017. Role of breeding and natal movements in lifetime dispersal of a forest-dwelling rodent. Ecol Evol. 7(7):2204–2213. doi: 10.1002/ece3.2814. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Spector S. 2002. Biogeographic Crossroads as Priority Areas for Biodiversity Conservation. Conserv Biol. 16(6):1480–1487. doi: 10.1046/j.1523-1739.2002.00573.x. [DOI] [Google Scholar]
  68. Sullivan RM. 1994. Micro-Evolutionary Differentiation and Biogeographic Structure Among Coniferous Forest Populations of the Mexican Woodrat (Neotoma mexicana) in the American Southwest: A Test of the Vicariance Hypothesis. J Biogeogr. 21(4):369. doi: 10.2307/2845756. [DOI] [Google Scholar]
  69. Sutherland G, Harestad A, Price K, Lertzman K. 2000. Scaling of Natal Dispersal Distances in Terrestrial Birds and Mammals. Conserv Ecol. 4(1). doi: 10.5751/ES-00184-040116. [accessed 2025 Jul 18]. https://www.ecologyandsociety.org/vol4/iss1/art16/. [DOI] [Google Scholar]
  70. Tiemann-Boege I, Kilpatrick CW, Schmidly DJ, Bradley RD. 2000. Molecular Phylogenetics of the Peromyscus boylii Species Group (Rodentia: Muridae) Based on Mitochondrial Cytochrome b Sequences. Mol Phylogenet Evol. 16(3):366–378. doi: 10.1006/mpev.2000.0806. [DOI] [PubMed] [Google Scholar]
  71. Villarreal M, Haire S, Iniguez J, Cortes C, Poitras T. 2019. Distant neighbors: recent wildfire patterns of the Madrean Sky Islands of southwestern United States and northwestern Mexico. Fire Ecol. 15. doi: 10.1186/s42408-018-0012-x. [DOI] [Google Scholar]
  72. Warshall P. 1995. The Madrean sky island archipelago: a planetary overview. US Department of Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment Station. [Google Scholar]
  73. Wiens JJ, Camacho A, Goldberg A, Jezkova T, Kaplan ME, Lambert SM, Miller EC, Streicher JW, Walls RL. 2019. Climate change, extinction, and Sky Island biogeography in a montane lizard. Mol Ecol. 28(10):2610–2624. doi: 10.1111/mec.15073. [DOI] [PubMed] [Google Scholar]
  74. Williams AP, Allen CD, Millar CI, Swetnam TW, Michaelsen J, Still CJ, Leavitt SW. 2010. Jun 4. Forest responses to increasing aridity and warmth in the southwestern United States. doi: 10.1073/pnas.0914211107. [accessed 2025 Mar 26]. https://www.pnas.org/doi/epdf/10.1073/pnas.0914211107. [DOI] [Google Scholar]
  75. Yanahan AD, Moore W. 2019. Impacts of 21st-century climate change on montane habitat in the Madrean Sky Island Archipelago. Divers Distrib. 25(10):1625–1638. doi: 10.1111/ddi.12965. [DOI] [Google Scholar]
  76. Zizka A, Antunes Carvalho F, Calvente A, Rocio Baez-Lizarazo M, Cabral A, Coelho JFR, Colli-Silva M, Fantinati MR, Fernandes MF, Ferreira-Araújo T, et al. 2020. No one-size-fits-all solution to clean GBIF. PeerJ. 8:e9916. doi: 10.7717/peerj.9916. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from bioRxiv are provided here courtesy of Cold Spring Harbor Laboratory Preprints

RESOURCES