Landscape-scale estimation of forest ungulate density and biomass using camera traps: Applying the REST model

Abstract

Landscape-scale information on animal density is being increasingly recognized as a fundamental parameter for enhancing wildlife conservation practices and management. It is important to obtain such information on forest ungulates because they have profound impacts on vegetation and also often constitute important prey for human hunters and large predators. In this study, we incorporated habitat covariates into a recently-developed likelihood-based approach (random encounter and staying time [REST] model) and test the potential of estimating the density and biomass of forest ungulates at landscape scales exclusively from camera traps. We targeted four duiker species (subfamily: Cephalophinae) in Central Africa and determined the effects of habitat covariates using Bayesian model averaging. The density of sympatric duikers largely varied across species, with each species exhibiting different spatial patterns; thus, sympatric duikers might exhibit space partitioning at the landscape scale. Topography might be a key factor determining spatial variation in duiker density, within and among species. Yet, total duiker biomass (kg) did not vary largely, and even remained high in naturally- and anthropogenically-disturbed forests; thus, disturbed forests may still be of value to human hunters and large predators. Through determining the habitat-density relationships, this study provides a novel approach for predicting animal density at landscape scales. Given the difficulty in implementing classic line-transect surveys in sloped areas, our approach might provide a viable way of estimating the density of ungulates occupying a wide variety of habitats.