Density of wild felids in Sonora, Mexico: a comparison of spatially explicit capture-recapture methods

Abstract

Estimates of population density are crucial for wild felid conservation and are commonly conducted using camera trapping. Spatially explicit capture-recapture (SCR) survey designs often use two cameras per station to photograph both flanks of a spotted felid to confirm identities. However, if the sampling grid is inadequate, and detector devices are unable to capture an animal's complete movements, density may be overestimated. Density analyses incorporating identification inputs from both flanks derived from unpaired camera stations may offer a cost-effective solution by doubling the number of stations available to researchers. During 2008, we surveyed 164 sites for jaguars, ocelots, and bobcats using unpaired camera stations (mean trap nights per camera = 32.66, inter-trap distance = 800 m) on private ranching lands in Sonora, Mexico. We obtained 15, 52, and 229 detections of jaguars, ocelots, and bobcats from 9, 35, and 87 stations, respectively. SCR density estimates (individuals/100 km(2)) derived from a maximum likelihood multi-session model and a Bayesian spatial partial identity model (SPIM) were similar: 4.61 and 1.54 (jaguar), 4.66 and 4.33 (ocelot), and 15.22 and 15.88 (bobcat), respectively. Due to insufficient recaptures of jaguars, only SPIM provided a credible estimate of jaguar density. Jaguar density was one of the lowest reported across their distribution, which was expected given Sonora's location at the periphery of the jaguar's range. However, ocelot and bobcat populations appear to be healthy in Sonora, even within unprotected ranchlands. We recommend the use of SPIM to estimate the density of spotted felids with any dataset containing few recaptures in unpaired camera surveys.