Occupancy of Terrestrial Mammal Populations in U.S. National Parks of the Southwest

North American mammal populations face many novel and powerful threats that are changing quickly over space and time. To complicate management and conservation decisions further, few established monitoring programs that can reliably detect changes in multiple mammal populations across species exist. Knowledge about the persistent threats and long-term trends of mammal populations are of immense importance to the public, the scientific community, as well as federal and state agencies charged with managing and protecting natural resources. We used an existing camera-based protocol to estimate occupancy of several terrestrial mammal species in seven U.S. National Parks in the Sonoran Desert. We surveyed 241 sites for 365 days to evaluate the detection efficiency (detected vs expected species) in different park units, evaluate a set of environmental variables with the potential to influence occupancy (ψ) and detection (p) probabilities of terrestrial mammals, and to estimate statistical power of the protocol to detect changes in occupancy. On average, we detected 76% (95% CI: 64% – 88%) of medium-to-large sized, native, mammals known to be present in surveyed park units. Mean occupancy across all park units ranged from 0.79 (0.10 – 0.86) for coyote (Canis latrans) to 0.12 (0.04 – 0.28) for black bear (Ursus americanus). Mean detection probability was highest for black-tailed jackrabbit (Lepus californicus) (0.35, 0.27 – 0.45) and lowest for mountain lion (Puma concolor) (0.06, 0.02 – 0.19). For many species, occupancy decreased as elevation and slope increased, and detection probability increased as precipitation increased during the study period. In addition, statistical power to detect changes in occupancy between two surveys was influenced mostly by occupancy and the number of sites surveyed. Power to detect changes in occupancy only increased to around 35 survey occasions (deployment days) above which it was relatively insensitive to increases in the number of survey occasions. Similarly, species with a higher initial occupancy achieved power to detect a change in occupancy faster (fewer survey occasions) and with less survey effort (fewer sites) than species with lower initial occupancy, whereas power was insensitive to changes in detection probability above 0.40. The camera trapping protocol we evaluated is sufficient to detect changes in occupancy with high power for species that are common (ψ = 0.80 – 0.99) and relatively easy to detect (p = 0.20 – 0.99), but not for species that are rare (ψ = 0.10 – 0.50) or difficult to detect (p = 0.10 – 0.19). For rare species, we suggest increasing detection probability by using lures and baits, increasing the duration of the survey period and surveyed sites, or strategically placing cameras in known areas of high activity. We conclude that the camera trapping design is well suited for simultaneously and cost-effectively monitoring terrestrial mammal communities within and between park units over the long term. This monitoring protocol has potential to inform conservation and management of mammals within the NPS and many other areas of North America