Most random-encounter-model density estimates in camera-based predator-prey studies are unreliable

Abstract

Population estimates are often required for identifying relationships between predators and their prey and to inform conservation and management actions. The random encounter model (REM) estimates population density of wildlife lacking individually unique markings, based on photographs or videos from remote camera-traps. However, the REM has strict sampling and input requirements that can be problematic, particularly for predators and other species which use landscapes non-randomly. Using data from a predator and its co-occurring prey, we found that placing cameras to target the predator, which may be implemented to achieve minimum sample sizes, inflated both predator and prey density estimates. Further, borrowing movement velocity (day range) values from other studies, species, or time periods caused substantial changes in density estimates. A comprehensive literature review revealed that 91% of REM density estimates in published predator–prey studies used data from non-random cameras or borrowed movement velocities and therefore did not satisfy REM requirements. Consequently, most REM density estimates from predator–prey ecology studies are likely not of the quality or reliability necessary for informing effective wildlife conservation or management.