Translator Disclaimer
15 October 2021 Habitat diversity influences puma Puma concolor diet in the Chihuahuan Desert
Charles H. Prude, James W. Cain III
Author Affiliations +

Habitat heterogeneity and corresponding diversity in potential prey species should increase the diet breadth of generalist predators. Many previous studies describing puma Puma concolor diets in the arid regions of the southwestern United States were focused within largely xeric locations, overlooking the influence of heterogeneity created by riparian forests. Such habitat heterogeneity and corresponding prey diversity could influence prey availability and puma diet composition. We examined seasonal prey composition of pumas occupying areas with different habitat conditions representing riparian areas adjacent to the Rio Grande and xeric Chihuahuan Desert uplands in southern New Mexico. We collected prey composition data from 686 kill sites made by 17 (9 males and 8 females) GPS-collared pumas from 2014 to 2018. Diet composition included 32 different avian, aquatic, small mammal, and ungulate prey species. Prey composition varied, with more ungulate prey consumed by pumas inhabiting the upland desert areas and more aquatic prey consumed in the riparian bosque. Prey composition differed between seasons, with ungulate prey decreasing and aquatic prey increasing during the hot–dry season. Prey composition also varied between puma sex and habitat with females in the desert uplands consuming more small mammals than either males or females in riparian areas. The diverse diets of the pumas inhabiting the heterogeneous landscapes in southern New Mexico provide additional evidence that pumas have broad diets that are strongly influenced by the habitat and prey community that their home range encompasses.

Animal behavior, primary productivity, and other environmental conditions influence the abundance, distribution, and vulnerability of prey species (Luttbeg et al. 2003). Areas with heterogeneous habitat conditions often have higher prey abundance when compared to more homogeneous habitats (Kerr and Packer 1997). Heterogeneity in habitat conditions affects habitat use and diet for both predators and prey (Hebblewhite et al. 2005, Gorini et al. 2012). Prey often benefit from habitat heterogeneity because the increased diversity in forage can enhance their ability to meet seasonal nutritional and energetic demands compared to homogenous habitats. Additionally, heterogeneous habitats may reduce predation risk, as prey can select areas with conditions that impede foraging by predators (Warfe and Barmuta 2004, Lecomte et al. 2008). Prey can also exploit habitat heterogeneity to mitigate predation risk from multiple predators. For example, elk Cervus elaphus can select areas with more rugged terrain and dense vegetation to evade cursorial predators (e.g. wolves; Canis lupus) or use areas with less vegetation cover and with higher visibility to evade stalking and ambush predators (e.g. puma; Puma concolor) (Kohl et al. 2019). On the other hand, predators can benefit from habitat heterogeneity because of the increased diversity, abundance, and in certain conditions also vulnerability of prey (Schooley et al. 1996, Bhattarai and Kindlmann 2012).

Puma is a widely distributed predator, occupying areas from the Andean Mountains in southern Argentina to the Yukon and Northwestern Territories in northern Canada (Currier 1983, Mulders et al. 2001, Jung and Merchant 2005, Elbroch and Wittmer 2014). Across their distribution range, pumas inhabit areas with diverse environmental conditions ranging from the marshy Florida Everglades (Maehr et al. 2002), densely vegetated neotropical forests (Novack et al. 2005), and the deserts in North and South America (Franklin et al 1999, Logan and Sweanor 2001, Choate et al. 2018). Puma morphology, physiology and behavior allow them to thrive in widely varying environmental conditions (Logan and Sweanor 2001). The generalist diets and adaptability to various environmental conditions allow pumas to exploit the diversity of prey within heterogeneous landscapes (Tattersall et al. 2002). This is especially true in areas where habitat heterogeneity increases the amount of stalking cover that enhances the ability of pumas to ambush prey (Lehman et al. 2017, Smith et al. 2019).

Pumas prey opportunistically on the most abundant and assailable species across their distribution range (Anderson 1983, Logan and Sweanor 2001), and consume a variety of prey species ranging in size from beetles (likely in association with larger prey items; Chrysomelidae spp., Cashman et al. 1992) and rodents (Cunningham et al. 1999) in Arizona, to feral horses Equus caballus and moose Alces alces (Knopff et al. 2009, Bacon et al. 2011) in Alberta, Canada. In South America, puma commonly prey on guanaco Lama guanicoe, vicuna Vicugna vicugna, European hare Lepus europaeus, lesser rhea Pterocnemia pennata, tapir Tapirus terrestris and pudu Pudu pudu (Iriarte et al. 1991, Franklin et al. 1999, Hernandez-Guzman et al. 2011, Azevedo et al. 2016, Gelin et al. 2017). In Central America, the most common prey are the white-tailed deer Odocoileus virginianus, collared peccary Pecari tajacu, coatimundi Nasua narica, nine-banded armadillo Dasypus novemcinctus, and various lagomorph species (Lepus spp., Sylvilagus audubonii; Nunez et al. 2000, de la Torre and de la Riva 2009). Pumas in North America frequently kill large ungulates such as deer Odocoileus spp., elk Cervus elaphus, pronghorn Antilocapra americana and bighorn sheep Ovis canadensis, and a variety of smaller mammals such as beaver Castor canadensis, coyote Canis latrans, raccoon Procyon lotor and skunk Mephitidae spp. Although the extent of livestock depredation by puma varies widely across their distribution range, cattle Bos tarus, sheep Ovis aries and goats Capra aegagrus, are also depredated by puma throughout the Americas in areas with ranching and agriculture (Polisar et al. 2003, Rominger et al. 2004). Despite having an extremely diverse diet, many studies have reported deer to be the preferred prey of pumas across different ecoregions (Iriarte et al. 1990, de la Torre and de la Riva 2009, Villepique et al. 2011), comprising, in many cases, more than 50% of consumed prey (Logan and Sweanor 2001, Wilckens et al. 2015).

Previous studies on puma diets in the arid regions of the southwestern United States primarily occurred in areas where the landscape is dominated by upland desert (Cunningham et al. 1999, Logan and Sweanor 2001, Choate et al. 2018). We sought to determine seasonal variation in prey composition and quantify differences in prey composition of pumas occupying the two habitat types including the mesic riparian bosque along the Rio Grande and xeric uplands. To assess the relationship between habitat heterogeneity and puma diet composition, we conducted a four-year study (2014–2018) examining puma diet through field investigation of kill sites at two study areas in the Chihuahuan Desert adjacent to the Rio Grande in south-central New Mexico. We predicted that this heterogeneity in vegetation would result in increased prey diversity and puma diet composition.

Material and methods

Study area

We conducted this study on the Armendaris Ranch (AR) and Sevilleta National Wildlife Refuge (SNWR) in south-central New Mexico (Fig. 1). The AR located 24 km east of Truth or Consequences, New Mexico, is a 146 854 ha private bison Bison bison ranch. The AR is bordered by the San Andres Mountains on White Sands Missile Range (WSMR) to the east, the Bosque del Apache National Wildlife Refuge to the north, the Rio Grande, and Elephant Butte Reservoir to the west. Elevation ranges from 1340 m along the Rio Grande to 2083 m in the Fra Cristobal Mountains. Vegetation types on the AR are comprised mostly of Chihuahuan desert scrub and desert grasslands with sparse pinyon–juniper Pinus edulis, Juniperus spp. woodlands at higher elevations in the Fra Cristobal Mountains. The landscape is primarily desert, except for the lush strip of riparian bosque bordering the Rio Grande and edges of Elephant Butte Reservoir. Common plant species in the desert upland areas include creosote bush Larrea tridentata, fourwing saltbush Atriplex canescens, ocotillo Fouquieria splendens, longleaf ephedra Ephedra trifurca, gramma grasses Bouteloua spp., juniper Juniperus deppeana, J. monosperma, prickly pear Opuntia spp. and cholla cacti Cylindropuntia spp. Whereas common plant species in the Rio Grande riparian bosque include cottonwood Populus wislizeni, desert willow Chilopsis linearis, willow Salix exigua and non-native salt cedar Tamarix ramosissima and Russian olive Elaeagnus angustifolia. Mean annual precipitation is 23.7 cm (SD ± 7.6) and mean annual snowfall is 8.6 cm (SD ± 15.5). Temperatures range from an average daily minimum of 5.3°C (SD ± 3.1) in January to an average daily maximum 30.6°C (SD ± 2.3) in July (climate data from Elephant Butte Dam, Truth or Consequences, NM, 1908–2019; WRCC 2018a).

Ungulates common in the xeric uplands on the AR include mule deer O. hemionus, pronghorn, non-native gemsbok Oryx gazella, and collared peccary. In addition, desert bighorn sheep O. c. mexicana occupy the Fra Cristobal Mountains. Potential prey species inhabiting the riparian areas adjacent to the Rio Grande include beaver, raccoon, Rio Grande wild turkey Meleagris gallopavo intermedia, and various aquatic species such as spiny softshell turtle Apalone spinifera and non-native common carp Cyprinus carpio. Other predators or potential scavengers of puma prey kills on the AR include coyote, bobcat Lynx rufus, gray fox Urocyon cinereoargenteus, golden eagle Aquila chrysaetos, and transient black bears Ursus americanus. The bosque bordering Elephant Butte Reservoir and the Rio Grande also provide habitat for migratory waterfowl in winter, which increases potential prey at that time (Kelly and Finch 1999).

The SNWR, located 30 km north of Socorro, New Mexico, is a 93 077 ha wildlife refuge managed by US Fish and Wildlife Service. The SNWR is approximately 75 km north of the AR (Fig. 1). The landscape at the SNWR is comparable to the AR and comprises xeric upland desert areas and riparian bosque bordering the Rio Grande. Elevation ranges from 1432 m along the Rio Grande to 2529 m in the Los Pinos and Sierra Ladrones mountain ranges. The xeric upland areas consist of Chihuahuan Desert Scrub, Great Plains Short Grass Prairie, Colorado Plateau Shrub Steppe at lower elevations, and pinyon–juniper woodland in the Los Pinos and Sierra Ladrones. The vegetation within the Rio Grande bosque is not only nearly identical to the AR but also has some restored wetland and waterfowl management areas. In the uplands, the SNWR has more pinyon pine, oak Quercus grisea, Q. gambelii, and juniper than the AR. The temperatures range from an average daily minimum of 2.1°C (SD ± 4.3) in January to an average daily high of 25.2°C (SD ± 2.6) in July. Mean annual rainfall is 20.6 cm (SD ± 6.6), with a mean annual snowfall of 11.8 cm (SD ± 12.2; climate data from Bernardo, NM, 1936–2019; WRCC 2018b).

Figure 1.

(A) Location of Sevilleta National Wildlife Refuge (north), Armendaris Ranch (south), and Rio Grande riparian bosque habitat (blue) in south-central New Mexico where we collected data on puma kills and prey diversity. Predation data were collected from GPS-collared pumas from 2016 to 2018 at the Sevilleta National Wildlife Refuge and from 2014 to 2018 at the Armendaris Ranch. (B) Enlarged section of Rio Grande riparian bosque habitat bordering the Armendaris Ranch near Fort Craig, NM in which kill site and diet data were collected from GPS-collared pumas from 2014 to 2018.


Common mammals in the upland desert areas at the SNWR include elk, non-native aoudad Ammotragus lervia, Rocky Mountain bighorn sheep, and desert bighorn sheep at higher elevations and feral horses Equus caballus, pronghorn, mule deer, and gemsbok at lower elevations. Common predators include coyote, bobcat, gray fox, and resident populations of black bear. Public access to the SNWR is restricted, however, some waterfowl and upland game bird hunting is permitted. Both study areas border private and public lands (i.e. state trust lands, Bureau of Reclamation, and Bureau of Land Management), most of which are used for livestock ranching, hunting and agriculture.

The abundance, availability and vulnerability of various prey species can change seasonally in our study area. To account for seasonal differences in prey composition, we used long-term climate data (1936–2019) to classify seasons at both study areas as the cool–dry (CD, November–March), hot–dry (HD, April–June), and hot–wet (HW, July–October) seasons.

Capture and monitoring

We primarily used Aldrich and Fremont foot snares to capture pumas from January 2014 to June 2018 on the AR, and from November 2015 to April 2017 on the SNWR. We monitored snare sets using cellular cameras (Verizon Black-hawk, Covert Scouting Cameras, Lewisburg, KY) and we used VHF trap-site transmitters (TBT-503-3, Telonics Inc., Mesa AZ) to monitor snares in areas lacking cellular service. We programmed the cellular cameras to send an SMS picture message alert immediately upon activity at the snare and tested the cameras for functionality by sending a remote command to the cameras to send a real-time image of snare sites daily at 07:00 and 18:00 h (MST). When using VHF trap-site transmitters, we checked the VHF signal every 6–12 h, depending on the weather conditions. We checked the snare transmitters more frequently during periods with hot (above 32°C) and cold (below 0°C) ambient temperatures to reduce the risk of stress or mortality from hyperthermia or hypothermia, respectively. We also used hounds to capture pumas in areas that provided suitable hunting conditions for hounds and safe escape structures (trees or boulders) for pumas. We mostly used hounds to recapture pumas to exchange collars with low batteries or those that were malfunctioning. Upon capture, we immobilized pumas with a pneumatic dart gun using 5 mg kg–1 ketamine combined with 0.08 mg kg–1 medetomidine. We used 0.3 mg kg–1 atipamezole as the antagonist for medetomidine (Kreeger et al. 2002). During processing, we recorded the age, sex and weight of each captured animal. We estimated the age using tooth wear and pelage patterns (Shaw 1986). We collared pumas older than 10–12 months with a GPS-Iridium collar (G2110E, Advanced Telemetry Systems, Isanti, MN). We marked captured pumas with a visual identification pattern (i.e. reflective color, letter or number) attached to the collar and ear-tagged each puma with a numbered tag. We closely monitored vital rates of all captured pumas for complications during capture and post-release. All capture and handling procedures follow acceptable methods (Sikes et al. 2016) and were approved by the New Mexico State University Institutional Animal Care and Use Committee (Protocol 2015-015).

We programmed the collars deployed on the AR to collect 16 GPS fixes per day; hourly intervals during crepuscular and nocturnal periods when pumas are characteristically more active (i.e. 19:00 – 07:00 h; Sweanor et al. 2008, Lewis et al. 2015, Soria-Dίaz et al. 2016) and then at 3-h intervals during the daytime (i.e. 10:00, 13:00 and 16:00 h) when pumas are less active. We programmed the collars on the SNWR to collect eight GPS fixes per day at 3-h intervals. The GPS data were transmitted via the Iridium satellite system every 12 h (i.e. 06:00, 18:00 h MST).

Prey composition data collection

We used GPS clusters to identify potential prey kill sites and to determine diet composition. At the AR, we defined a cluster, or potential kill and feeding location, as ≥6 consecutive crepuscular or nocturnal locations within a 50-m radius, whereas on the SNWR, to account for the 3-h fix interval, we defined a cluster as ≥2 consecutive crepuscular or nocturnal locations within a 50-m radius. Thus, any location where a puma spent six consecutive crepuscular or nocturnal hours within a 50-m radius was considered a cluster and subject to field inspection. To make efficient use of limited field resources, we used broader temporal and more restricted distance characteristics than some other studies that used GPS cluster analysis to identify predator kill sites. We used AnimalClusters.R (ver. 1.1) developed by Daniel and Kindschuh (2016) and program R (ver. 3.1.2; < www.r-project. org>) to identify GPS clusters. We then investigated clusters in the field as soon as possible to prevent loss of kill evidence caused by scavengers and weathering, which was generally within 7–14 days of the cluster start date. We also prioritized smaller clusters for visitation to minimize the loss of evidence from clusters that might contain remains of smaller prey species. We located clusters in the field by navigating to the centermost GPS fix within the defined cluster and then outwardly searched the surrounding area within 50 m of each GPS location in the cluster in a spiral-like fashion. Thus, we examined each location in the cluster for evidence of a kill (i.e. carcass remains, hair, bone fragments, blood, drag marks, disturbed vegetation, and soil; Shaw 1986). We classified clusters as kill sites if they contained evidence of a kill.

At each kill site, we used tooth wear, pelage patterns and the morphological characteristics of the carcass to estimate the age class of prey. For ungulate prey, we classified ages as neonate (<1 year), yearling (1–2 years), sub-adult (2–4 years), adult (4–6 years), (older than 6 years), and unidentified for prey that lacked evidence of age. For non-ungulate prey, we classified age as neonate (younger than 1 year), adult (older than 1 year), and undefined. We used genitalia or secondary sexual characteristics to identify prey sex when possible. We determined if the prey had been killed by a puma or scavenged by examining the carcass and site for evidence of puma predation (i.e. bite marks to the neck or throat, carcass cache, subcutaneous hematomas on neck or throat, tracks near carcass; Shaw 1986). We also used the rate of decomposition of the carcass relative to the GPS location fix times and dates from the cluster (Wilckens et al. 2015). We inspected the carcass remains for signs of malady, injury, deformity or anything that could have increased its susceptibility to puma predation.


At both study areas, some pumas remained in the riparian bosque habitat, others exclusively used xeric upland areas and some utilized both areas, regularly moving between riparian bosque and xeric uplands. To account for variation in the predominant use of one vegetation cover type over others by GPS-collared pumas, we used satellite imagery in ArcGIS 10.6 (Esri 2018: 10.6. Redlands, CA) to digitize the boundary between the riparian bosque along the Rio Grande and the xeric uplands (Fig. 1). We then classified each puma as being riparian, upland, or mixed based upon the proportion of their total GPS fixes within the upland and riparian areas: pumas with more than 75% of their cumulative GPS fixes within the riparian area were classified as riparian, pumas with more than 75% of their cumulative fixes in the upland areas were classified as upland, and pumas with less than 75% of their cumulative fixes in either riparian or upland areas were classified as mixed.

We categorized prey species into four prey classes: avian (all non-waterfowl avian species), aquatic (all species with habitat requirements associated with water in the Rio Grande, including fish, turtles, waterfowl, beaver and muskrat), small mammal (all non-ungulate mammals), and ungulate prey. We combined beaver and muskrat with aquatic species because their populations in our study area would not exist without the aquatic habitat created by the Rio Grande. We did not document any beaver or muskrat kills outside of the riparian habitat and all kills were located at very close proximities to water, identical to the other aquatic species documented (carp, turtles). We then calculated the proportion of kills in each prey class for individual pumas within each season and year. Because our response variables were proportional, we then used the logit transformation on the data before analysis. We used multivariate analysis of variance (MANOVA) to examine differences in the proportion of each prey type by puma sex, predominant habitat type (i.e. riparian, upland, mixed) and season (i.e. cool-dry, hot-dry and hot-wet). We then used Turkey's HSD post hoc analysis to further assess differences in prey class composition between seasons and puma habitat types. Due to low sample sizes, we conducted all analyses with α = 0.1 to reduce the chance of committing a type II error. All statistical analyses were conducted using SPSS (IBM SPSS Statistics for Windows, ver. 25.0).


We captured 11 pumas (7 male and 4 females) on the AR between February 2014 and June 2018 and 5 pumas (1 male and 4 females) on the SNWR from November 2015 through December 2017 (Table 1). Only one female puma was captured using hounds, the others were captured with snares. Data were also collected from one male puma (LM7) that was originally captured by another researcher on the Ladder Ranch near Hillsboro, New Mexico but dispersed to the AR shortly after capture. Most of the pumas in this study were classified as adults (>3 years) however we did collect data from 3 subadult (18 months to 3 years) pumas at the AR (2 females, 1 male). We classified 3 males and 4 females as being upland pumas, 2 males and 4 females as riparian pumas and 4 males as mixed pumas (Table 1). We monitored the pumas for 5582 telemetry days (n = 17 pumas, mean = 328 days/puma ± 226 days [SD]; Table 1). Female pumas were generally monitored for a longer period (3442 total days; mean = 430 days/female ± 200 days [SD]) than males (2140 days; mean = 237 days/male ± 217 [SD]). We monitored pumas for 2457 telemetry days during the cool-dry seasons, 1195 telemetry days during the hot-dry seasons, and 1930 telemetry days during the hot-wet seasons.

Table 1.

Puma sex, age, monitoring period, habitat classification and prey class proportions for satellite collared pumas captured on the Armendaris Ranch and Sevilleta National Wildlife Refuge in south-central New Mexico, 2014–2018. Puma habitat classification based upon the proportion of fixes within habitat type; riparian bosque or upland desert.


We investigated 1073 GPS clusters, of which 686 (64%) were kills or feeding sites. The remaining 387 cluster locations we investigated were classified as bed sites (n = 247; 23%), scat sites (n = 13; 0.01%), hunting sites (n = 45; 4%), scavenge sites (n = 2; 0.002%), water locations (n = 3; 0.003%) or unknown (n = 77; 7%). We found 531 kills on the AR (77%) and 155 kills at SNWR (23%). Female pumas killed 403 prey animals (59% of total kills) and males killed 283 (41% of total kills).

We documented 32 different prey species at kill sites ranging from small aquatic prey (e.g. common carp, waterfowl), to large ungulates (e.g. gemsbok, mule deer; Table 2). Mule deer were the most common prey species (n = 195; 28%), followed by coyote (n = 84; 12%), beaver (n = 70; 10%), raccoon (n = 51; 0.07%), carp (n = 49; 0.07%) and gemsbok (n = 35; 0.05%). Bighorn rams (n = 12; 44%) and lambs (n = 10; 37%) were killed more than ewes (n = 5; 19%); upland, riparian, and mixed puma all killed bighorn sheep. Prey composition included 18 kills of avian species (0.03%), 158 kills of aquatic species (23%), 192 kills of small mammal species (28%), and 318 ungulate kills (46%; Table 2). We were unable to identify the age and/or sex of many of the small mammals, ungulate neonates, and some of the aquatic prey because pumas would consume nearly the entire carcass, leaving only hair, hooves, scales, or some larger bone fragments. For the carcasses that we were able to collect age information, there were 55 neonates (8%), 46 yearlings (7%), 68 sub-adults (10%), 275 adults (40%), and 28 mature animals (4%). There were 214 kills with insufficient remains to adequately estimate the age of the animal (31%). We were able to identify the sex for 76 male (11%) and 55 female (8%) prey, most of which were adult ungulates (n = 118; 90%). There were 555 kills (81%) that lacked genitalia or secondary sexual characteristic to determine the sex. We documented 305 kills during the cool-dry season (44%), 101 during the hot-dry season (15%) and 280 during the hot-wet season (41%).

Table 2.

Puma kills by species documented at GPS clusters from collared pumas at the Armendaris Ranch and Sevilleta National Wildlife Refuge in south-central New Mexico, 2014–2018.


Mule deer were the most common prey species during the hot-wet (n = 110; 39%) and cool-dry (n = 74; 24%) seasons, but were the fourth most common species at kill sites (n = 11; 11%) during the hot-dry season behind carp, beaver, and coyote. Coyote were the second most common prey species during the cool-dry and hot-wet seasons (n = 48; 16% and n = 21; 11%) and the third species during the hot-dry season (n = 13; 13%). Beaver were common during all three seasons: (cool-dry (n = 34; 11%), hot-dry (n = 14; 14%) and hot-wet (n = 22; 8%). Carp were the most frequent prey species located at GPS clusters during the hot-dry season (n = 22) and comprised 22% of all kills during the hot-dry season. The proportion of raccoons at kill sites was higher during the cool-dry season (n = 35; 11%), compared to hot-dry (n = 3; 3%) and hot-wet (n = 13; 5%) seasons. There were also more waterfowl kills during the cool-dry season (n = 11; 4%), compared to the hot-dry (n = 3; 3%) and hot-wet (n = 7; 3%) seasons.

Prey composition differed between puma habitat classifications for all prey types (aquatic, F2,51=22.3, p < 0.001; avian, F2,51 = 5.24, p=0.01; small mammal, F2,51 = 2.75, p=0.077; ungulate, F2,51 = 4.05, p=0.026). Kill sites for pumas predominantly occupying the riparian corridor consisted of four times as many aquatic prey than mixed pumas and more than 10 times higher than upland pumas. Riparian pumas also consumed 2–4 times as many avian preys than both mixed and upland pumas (Fig. 2). Kill sites from upland pumas were comprised of 2–3 times as many ungulates as riparian and mixed pumas using both areas (Fig. 2). Small mammal prey were more prevalent at the kill sites of upland (21 total, mean proportion=0.217 ± 0.051 [SE]) and riparian pumas (18 total, mean proportion=0.249 ± 0.063 [SE]) compared to mixed pumas that used both areas (12 total, mean proportion=0.064 ± 0.049 [SE]; Fig. 2).

Figure 2.

Mean proportion of puma kill sites by prey class and puma habitat class based on GPS-collared pumas in the Armendaris Ranch and Sevilleta National Wildlife Refuge, south-central New Mexico, 2014–2018. Pumas are categorized into habitat classes based on the proportion of their GPS fixes within the upland desert and Rio Grande riparian bosque habitats. Error bars represent 90% confidence intervals.


For all puma types, the proportion of kills sites that were ungulates also differed by season (F2,51 = 2.61, p = 0.087). Ungulate prey were 3–4 times more common at kill sites during the cool-dry and hot-wet seasons than during the hot-dry season (Fig. 3). The proportion of kills composed of small mammal prey differed by puma habitat classes and puma sex (puma habitat class × puma sex interaction; F1,52 = 3.32, p = 0.077, Fig. 4). Upland female pumas consumed the highest proportion of small mammal prey, 2–3 times as many as did upland, riparian and mixed males; and approximately 6% more than riparian females. The proportion of kill sites composed of avian prey were dependent on puma habitat class, season, and sex (puma habitat class × season × sex interaction, F6,23 = 2.62, p = 0.087) with upland female pumas having a higher proportion of avian prey during the hot-dry season (Table 3).

Ungulate prey comprised the highest mean proportion of kills across all three seasons with the highest during the hot-wet season (0.524 ± 0.091 [SE]). There was more aquatic prey killed during the hot-dry season (n = 40 aquatic prey, n = 31 ungulate prey), however, the mean proportion of ungulates (0.235 ± 0.079 [SE]) in the combined diet was still higher than that for aquatic prey (0.189 ± 0.087 [SE]). Small mammal prey had the second highest mean proportion during the cool-dry (0.239 ± 0.057 [SE]) and hot-wet (0.155 ± 0.041 [SE]) seasons but had a slightly lower mean proportion than aquatic prey during the hot-dry season (0.183 ± 0.075 [SE]). Avian prey represented the lowest mean proportion of the diet across all three seasons with the highest proportion during the hot-dry season (0.041 ± 0.031 [SE]) and lowest during the hot-wet season (0.012 ± 0.009 [SE]; Fig. 3).


We identified high variability in puma prey composition as a result of different habitat conditions and prey availability between the mesic riparian bosque along the Rio Grande and surrounding Chihuahuan Desert. The diet breadth documented in many previous puma studies is often less than 20 different prey species. Approximately 15 different species were consumed by jaguars and pumas in Sonora, Mexico (Rosas-Rosas et al. 2003), 17 different species in northeast Oregon (Clark et al. 2014), 13 species in the badlands of North Dakota (Wilckens et al. 2015), 15 species in the Maya Biosphere Reserve, Guatemala (Novack et al. 2005) and 10 species in Banff National Park, Canada (Knopff et al. 2010). Harveson et al. (2000) reported pumas utilizing 10 different prey species in a heterogeneous south Texas landscape that was comprised of 42% riparian and 58% upland habitat. However, Elbroch and Quigley (2019) reported pumas consuming more than 40 different species in the Greater Yellowstone Ecosystem, Wyoming. Most previous puma diet studies in desert areas occurred mostly in areas that lacked wetland habitat and had little or no aquatic prey available (Logan and Sweanor 2001, Choate et al. 2018). Seven of the 32 prey species (22%) that we documented did not occur outside of the riparian bosque.

Figure 3.

Mean seasonal proportion of puma kill sites by prey classes of GPS-collared puma at the Armendaris Ranch and Sevilleta National Wildlife Refuge in south-central New Mexico, 2014–2018. Error bars represent 90% confidence intervals.


Figure 4.

Mean proportion of puma kills by prey class, puma sex, and puma habitat class based on GPS-collared pumas at the Armendaris Ranch and Sevilleta National Wildlife Refuge, south-central New Mexico, 2014–2018. Pumas are categorized into habitat classes based on the proportion of their GPS fixes within the upland desert and Rio Grande riparian bosque habitats. Error bars represent 90% confidence intervals.


Although we documented higher diversity in prey composition than in many previous puma studies in desert biomes, our results are still similar in that large ungulate, primarily deer, are the most common prey consumed (Logan and Sweanor 2001, de la Torre and de la Riva 2009, Villepique et al. 2011, Wilckens et al. 2015). The specific prey composition of pumas restricted to the upland areas strongly suggests that had the landscape been homogenous desert without the riparian bosque, our results would have closely resembled the ungulate dominated diets documented by Logan and Sweanor (2001) in the nearby San Andres Mountains (i.e. ungulates, primarily mule deer, composed 92% of the diet). However, Logan and Sweanor (2001) did not use GPS collars during their study, so kills were likely biased towards larger prey items, whereas in our study we identified several kills associated with smaller prey using GPS cluster analyses. The diets of the upland pumas in our study consisted of 70% ungulate prey, 28% small mammal prey, and 2% aquatic and avian prey. Prey composition at kill sites of riparian pumas was similar to those in South American neotropical areas where puma diet is mostly comprised of smaller prey items due to the increased abundance of small prey species (Iriarte et al. 1990, Monroy-Vilchis et al. 2009, Gómez-Ortiz et al. 2011). The diets of riparian pumas in this study consisted of only 26% ungulate prey and 74% aquatic, small mammal, and avian prey. Beaver was a common prey for riparian pumas and comprised 42% of the 158 aquatic species kills. Only four male pumas were classified as mixed habitat users and their diet was more similar to that of the upland pumas with 62% ungulate prey and 38% small mammal and aquatic prey. Female pumas were spent 90–100% of their time within their chosen habitat and utilized all prey classes. Whereas males utilized both habitats more generally, spending 53–96% of their time within a single habitat type but kill composition was less diverse compared to females.

Elk kills were uncommon and only comprised 2% of the total kills we documented. Elk occurred at lower densities in our study areas and were generally located in agricultural or wetland areas near the Rio Grande (i.e. Bosque del Apache NWR, agricultural areas near Socorro, NM; Fig. 1) and at higher elevations on the SNWR, which limited their availability as potential prey. Gemsbok, an elk-size non-native ungulate, occurred at higher densities (Bender et al. 2019) and were frequently preyed upon by male pumas (3 males were responsible for 89% of gemsbok kills we documented) and infrequently by female pumas (2 females killed 3 gemsbok). Predation of adult gemsbok was unexpected, as only 3 neonate gemsbok kills were documented by Logan and Sweanor (2001) between 1985 and 1995 in the nearby San Andres Mountains. Gemsbok evolved with African lion Panthera leo predation in the arid and semi-arid regions of southern Africa. As a result of which, gemsbok have thicker skin and muscular tissue in their neck protecting their spine and spear-like horns averaging 60–150 cm in length as weaponry to defend against predators (Logan and Sweanor 2001, Edgington 2009). Many of the gemsbok kills that we documented were neonates; however, one mature male puma killed 29 adult gemsbok on WSMR which comprised 58% (n = 29) of his total kills. Bighorn sheep only represented 8% (n = 27) of the ungulate kills. However, the low contribution of bighorn sheep to the prey composition was almost certainly influenced by an active management program that included the lethal removal of pumas that killed multiple (5) bighorn sheep in the Fra Cristobal and Ladron mountains. Bighorn sheep were preyed upon throughout the year, with a slight increase during lambing season from February through May. All but one of the bighorn sheep kills were made by male pumas in the Fra Cristobal and Caballo mountains, the exception being one ram killed by a female puma in the Pino Mountains on the SNWR. Although pumas regularly utilized areas with livestock, mostly cattle, we only documented a few instances of livestock predation and most were beef calves and a feral goat killed in the bosque along the Rio Grande.

Table 3.

Puma kills by age and sex documented at GPS clusters from collared pumas at the Armendaris Ranch and Sevilleta National Wildlife Refuge in south-central New Mexico, 2014–2018.


The increased proportion of ungulate prey during the hot-wet season is coincident with the increased availability of mule deer fawns during fawning season (July–September). Fawns and yearlings comprised 55% (n = 60) of the mule deer kills and 21% of the total kills during the hot-wet season. These findings are consistent with Logan and Sweanor (2001) on WSMR and with the findings of Kay (2018) in the nearby Gallinas Mountains near Corona, New Mexico. During the hot-dry season, we documented an increase in aquatic prey consumption; this time coincides with the spawning season of carp. During spawning, carp are more susceptible to puma predation as they use shallower waters (1–4 feet in depth) to spawn. We speculate that carp were typically caught in shallower water of the Rio Grande in areas where the riverbank was flat and provided ambush cover (i.e. vegetation, driftwood snags) for pumas. There were a few instances in which carp became trapped as flooded areas adjacent to the Rio Grande dried, allowing pumas to easily catch them. One young female puma (ARF02) seemed to specialize (Elbroch and Wittmer 2013) in killing turtles as she was responsible for 15 (94%) of the spiny-softshell turtle kills. The majority of the spiny-softshell turtle kills occurred during August–September which is typically when the flooded areas adjacent to the Rio Grande become dry, forcing the turtles to travel back to the Rio Grande. August is also when female turtles lay their eggs in nests burrowed in dry sandy areas (Stebbins 2003), which may have also increased their vulnerability to puma predation (Stebbins 2003). Although the availability of waterfowl increases considerably during the cool-dry season, there was only a slight increase in waterfowl kills compared to other seasons.

Like most studies of predator diet composition using GPS cluster analysis, we were faced with tradeoffs, both between GPS fix interval and collar battery life and when selecting a cluster definition (i.e. number of points, timing and distance) that would improve detection of kill sites from small and large prey species. The longer temporal component of our cluster definition and prioritizing clusters initiated during nocturnal or crepuscular periods likely enhanced our ability to detect larger prey items (Wilckens et al. 2015, Vogt et al. 2018). However, our cluster definition may have also biased against detection of some smaller prey species. The shortened nighttime GPS fix interval and promptness in field investigation of smaller clusters likely improved our ability to locate some of the smaller prey species (Knopff et al. 2009). Nonetheless, we found that very small prey items such as lagomorphs and rodents were difficult to detect using GPS cluster investigation and are therefore likely to be underrepresented in our data (Bacon et al. 2011). For the rabbit kills that we were able to locate, typically only feet, ears (jackrabbit), or a few tufts of fur remained as evidence similar to other studies (Elbroch and Wittmer 2013). The small aquatic prey kills were easier to locate due to more carcass remnants as pumas did not eat feathers (waterfowl), shells (turtles), or scales and gill plates (carp). We were unable to investigate some of the clusters that occurred on private lands outside of our study areas and WSMR as promptly due to access restrictions. The delay may have reduced our ability to detect smaller non-ungulate prey at those sites.

Previous research on pumas across ecosystems ranging from northern Canada to the southern tip of South America indicates that puma are generalist predators and frequently kill the most abundant and or vulnerable prey species in the areas they occupy. The heterogeneity across a relatively small spatial scale in our unique study system, increased prey diversity for pumas. Pumas show similar adaptive responses across much larger spatial scales. The diverse diets of the pumas in our study provide additional evidence that pumas are predators that utilize a multitude of prey species and are capable of inhabiting extremely diverse habitats. Pumas have broad diets that are strongly influenced by the habitat and prey community that their home ranges encompass. Additionally, puma diet is likely to be more diverse in areas with heterogeneous habitat conditions that support a wider variety of prey species. This is especially true in western North America where xeric habitat conditions typically do not support higher densities of ungulate prey and pumas are forced to exploit a variety of smaller species.

Acknowledgements –

Comments by B. D. Jansen, G. Harris and K. A. Logan improved earlier drafts of this manuscript.

Funding – Funding for this project was provided by New Mexico Department of Game and Fish, US Fish and Wildlife and Turner Enterprises Inc. We thank M. Keeling, R. Passernig, D. Martin, R. Thompson and J. Zemke for assistance with captures and fieldwork. T. Waddell, E.D. Edwards at the Armendaris Ranch, K. Granillo and J. Erz at the Sevilleta National Wildlife Refuge and C. Kruse with Turner Enterprises Inc. provided logistical support. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Permits – All capture and handling procedures follow acceptable methods (Sikes et al. 2016) and were approved by the New Mexico State University Institutional Animal Care and Use Committee (Protocol 2015-015).

Data availability statement

Data are available from the Dryad Digital Repository: <> (Prude and Cain III 2021).



Anderson, A. E. 1983. A critical review of literature on puma Puma concolor. – Colorado Division of Wildlife Special Report No. 54. Google Scholar


Azevedo, F. et al. 2016. Puma Puma concolor predation on tapir Tapirus terrestris. – Biota Neotrop. 16: 1–4. Google Scholar


Bacon, M. M. et al. 2011. Do GPS clusters really work? Carnivore diet from scat analysis and GPS telemetry methods. – Wildl. Soc. Bull. 35: 409–415. Google Scholar


Bender, L. et al. 2019. Population dynamics and control of exotic South African oryx in the Chihuahuan Desert, south-central New Mexico. – Hum.–Wildl. Interact. 13: article 19. Google Scholar


Bhattarai, B. P. and Kindlmann, P. 2012. Habitat heterogeneity as the key determinant of the abundance and habitat preference of prey species of tiger in the Chitwan National Park, Nepal. – Acta Theriol. 57: 89–97. Google Scholar


Cashman, J. L. et al. 1992. Diets of mountain lions in southwestern Arizona. – Southwest. Nat. 37: 324–326. Google Scholar


Choate, D. et al. 2018. Cougar dispersal and natal homing in a desert environment. – W. N. Am. Nat. 78: 221–235. Google Scholar


Clark, D. A. et al. 2014. Cougar kill rates and prey selection in a multiple-prey system in northeast Oregon. – J. Wildl. Manage. 78: 1161–1176. Google Scholar


Cunningham, S. et al. 1999. Diet selection of mountain lions in southeastern Arizona. – J. Range Manage. 52: 202–207. Google Scholar


Currier, M. J. 1983. Felis concolor.– Mammalian Species 8: 1–7. Google Scholar


Daniel, D. L. and Kindschuh, S. R. 2016. AnimalClusters.R source code (Ver. 1.1) [Source code]. – . Google Scholar


de la Torre, J. A. and de la Riva, G. 2009. Food habits of pumas Puma concolor in a semiarid region of central Mexico. – Mastozool. Neotrop. 16: 211–216. Google Scholar


Edgington, R. 2009. The safari of the southwest: hunting, science and the African oryx on White Sands Missile Range, New Mexico, 1969–2006. – West. Hist. Quart. 40: 469–491. Google Scholar


Elbroch, L. M. and Quigley, H. 2019. Age-specific foraging strategies among pumas, and its implications for aiding ungulate populations through carnivore control. – Conserv. Sci. Pract. 1: e23. Google Scholar


Elbroch, L. M. and Wittmer, H. U. 2013. The effects of puma prey selection and specialization on less abundant prey in Patagonia. – J. Mammal. 94: 259–268. Google Scholar


Franklin, W. et al. 1999. Ecology of the Patagonia puma Puma concolor patagonica in southern Chile. – Biol. Conserv. 90: 33–40. Google Scholar


Gelin, M. L. et al. 2017. Response of pumas Puma concolor to migration of their primary prey in Patagonia. – PLoS One 12: e0188877. Google Scholar


Gómez-Ortiz, Y. et al. 2011. Is food quality important for carnivores? The case of Puma concolor. – Anim. Biol. 61: 277–288. Google Scholar


Gorini, L. et al. 2012. Habitat heterogeneity and mammalian predator–prey interactions. – Mammal Rev. 42: 55–77. Google Scholar


Harveson, L. A. et al. 2000. Prey use by mountain lions in southern Texas. – Southwest. Nat. 45: 472–476. Google Scholar


Hebblewhite, M. et al. 2005. Spatial decomposition of predation risk using resource selection functions: an example in a wolf–elk predator–prey system. – Oikos 111: 101–111. Google Scholar


Hernandez-Guzman, A. et al. 2011. Food habits of Puma concolor (Carnivora: Felidae) in the Parque Nacional Natural Purace, Colombia. – Rev. Biol. Trop. 59: 1285–1294. Google Scholar


Iriarte, J. et al. 1990. Biogeographic variation of food habits and body size of the America puma. – Oecologia 85: 185–190. Google Scholar


Iriarte, J. A. et al. 1991. Feeding ecology of the Patagonia puma in southernmost Chile. – Rev. Chil. Hist. Nat. 64: 145–156. Google Scholar


Jung, T. S. and Merchant, P. J. 2005. First confirmation of cougar, Puma concolor, in the Yukon. – Can. Field-Nat. 119: 580–581. Google Scholar


Kay, J. H. 2018. Top–down and bottom–up influences on central New Mexico mule deer (Odocoileus hemionus). – MS thesis, New Mexico State Univ., Las Cruces, New Mexico. Google Scholar


Kelly, J. F. and Finch, D. M. 1999. Use of saltcedar vegetation by landbirds migrating through the Bosque Del Apache National Wildlife Refuge. Rio Grande ecosystems: linking land, water and people. – In: Proc. US Dept of Agriculture, Forest Service. RMRS-P-7. USDA-FS, Ogden, UT, pp. 222–230. Google Scholar


Kerr, J. T. and Packer, L. 1997. Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. – Nature 385: 252–254. Google Scholar


Kindschuh, S. R. et al. 2016. Efficacy of GPS cluster analysis for predicting carnivory sites of a wide-ranging omnivore: the American black bear. – Ecosphere 7: e01513. Google Scholar


Knopff, K. et al. 2009. Evaluating global positioning system telemetry techniques for estimating cougar predation parameters. – J. Wildl. Manage. 73: 586–597. Google Scholar


Knopff, K. H. et al. 2010. Cougar kill rate and prey composition in a multiprey system. – J. Wildl. Manage. 74: 1435–1447. Google Scholar


Kohl, M. T. et al. 2019. Do prey select for vacant hunting domains to minimize a multi-predator threat? – Ecol. Lett. 22: 1724–1733. Google Scholar


Kreeger, T. J. et al. 2002. Handbook of wildlife chemical immobilization, International edition. – Wildlife Pharmaceuticals Inc., Fort Collins, Colorado, USA. Google Scholar


Lecomte, N. et al. 2008. Predator behaviour and predation risk in the heterogeneous Arctic environment. – J. Anim. Ecol. 77: 439–447. Google Scholar


Lehman, C. P. et al. 2017. Characteristics of successful puma kill sites of elk in the Black Hills, South Dakota. – Wildl. Biol. 2017: wlb.00248. Google Scholar


Lewis, J. S. et al. 2015. Interspecific interactions between wild felids vary across spatial scales and levels of urbanization. – Ecol. Evol. 5: 5946–5961. Google Scholar


Logan, K. and Sweanor, L. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. – Island Press, Washington, DC. Google Scholar


Luttbeg, B. et al. 2003. Prey state and experimental design affect relative size of trait- and density-mediated indirect effects. – Ecology 84: 1140–1150. Google Scholar


Maehr, D. S. et al. 2002. Florida panther dispersal and conservation. – Biol. Conserv. 106: 187–197. Google Scholar


Monroy-Vilchis, O. et al. 2009. Food niche of Puma concolor in central Mexico. – Wildl. Biol. 15: 97–106. Google Scholar


Mulders, R. et al. 2001. Cougars (Puma concolor) in the Northwest Territories and Wood Buffalo National Park. – Arctic 54: 185–187. Google Scholar


Novack, A. J. et al. 2005. Foraging ecology of jaguar (Panthera onca) and puma (Puma concolor) in hunted and non-hunted sites within the Maya Biosphere Reserve, Guatemala. – J. Zool. 267: 167–178. Google Scholar


Nunez, R. et al. 2000. Food habits of jaguars and pumas in Jalisco, Mexico. – J. Zool. 252: 373–379. Google Scholar


Polisar, J. et al. 2003. Jaguars, pumas, their prey base and cattle ranching: ecological interpretations of a management problem. – Biol. Conserv. 109: 297–310. Google Scholar


Prude, C. H. and Cain III, J. W. 2021. Data from: Habitat diversity influences puma Puma concolor diet in the Chihuahuan Desert. – Dryad Digital Repository, <>. Google Scholar


Rominger, E. M. et al. 2004. The influence of mountain lion predation on bighorn sheep translocations. – J. Wildl. Manage. 68: 993–999. Google Scholar


Rosas-Rosas, O. et al. 2003. Food habits of pumas in northwestern Sonora, Mexico. – Wildl. Soc. Bull. 31: 528–535. Google Scholar


Schooley, R. L. et al. 1996. Can shrub cover increase predation risk for a desert rodent? – Can. J. Zool. 74: 157–163. Google Scholar


Shaw, H. G. 1986. Mountain lion field guide. – Arizona Game and Fish Dept Special Report 9, Phoenix, USA. Google Scholar


Sikes, R. S. and the Animal Care and Use Committee of the American Society of Mammalogists 2016. 2016 Guidelines of the American Society of Mammalogists for the use of wild mammals in research and education. – J. Mammal. 97: 663–688. Google Scholar


Smith, J. A. et al. 2019. Habitat complexity mediates the predator–prey space race. – Ecology 100: e02724. Google Scholar


Soria-Dίaz, L. et al. 2016. Activity pattern of puma Puma concolor and its main prey in central Mexico. – Anim. Biol. 66: 13–20. Google Scholar


Stebbins, R. C. 2003. A field guide to western reptiles and amphibians. – Houghton Mifflin Harcourt. Google Scholar


Sweanor, L. L. et al. 2008. Puma and human spatial and temporal use of a popular California state park. – J. Wildl. Manage. 72: 1076–1084. Google Scholar


Tattersall, F. H. et al. 2002. Is habitat linearity important for small mammal communities on farmland? – J. Appl. Ecol. 39: 643–652. Google Scholar


Villepique, J. et al. 2011. Diet of cougars Puma concolor following a decline in a population of mule deer Odocoileus hemionus: lack of evidence for switching prey. – Southwest. Nat. 56: 187–192. Google Scholar


Vogt, K. et al. 2018. Suitability of GPS telemetry for studying the predation of Eurasian lynx on small- and medium-sized prey animals in the northwestern Swiss Alps. – Eur. J. Wildl. Res. 64: article73. Google Scholar


Warfe, D. M. and Barmuta, L. A. 2004. Habitat structural complexity mediates the foraging success of multiple predators. – Oecologia 141: 171–178. Google Scholar


Wilckens, D. T. et al. 2015. Mountain lion (Puma concolor) feeding behavior in the Little Missouri Badlands of North Dakota. – J. Mammal. 97: 373–385. Google Scholar


WRCC 2018a. Climate data from Elephant Butte Dam, Truth or Consequences, NM. – <>, accessed 15 December 2020. Google Scholar


WRCC 2018b. Climate data from Bernardo, NM. – <>, accessed 15 December 2020. Google Scholar
© 2021 The Authors.
Charles H. Prude and James W. Cain III "Habitat diversity influences puma Puma concolor diet in the Chihuahuan Desert," Wildlife Biology 2021(4), wlb.00875-, (15 October 2021).
Accepted: 17 August 2021; Published: 15 October 2021

Back to Top