Prioritizing wildlife conservation along habitat gradients in Sumatra

Managing protected areas (PAs) requires measurable indicators to assess effectiveness. The status of populations and guilds of multiple species are potential indicators that should be useful in biodiversity-rich tropical countries. We quantified such indicators using data from an intensive camera trap survey of seven sites at the forest-farmland interface of Kerinci Seblat National Park, Sumatra, Indonesia. Surveys between 2014 and 2016 covered 671 camera locations set along habitat gradients comprising primary to degraded forest and lowland to sub-montane forest. We ran Bayesian multi-species occupancy models that incorporated landscape covariates and patrol intensity to generate four population parameters: relative abundance, probability of habitat use, species richness and detection probability. Model-derived beta coefficients summarized at the guild-level were extrapolated using detailed spatially-explicit data on landscape covariates to produce multi-guild occurrence maps to explore the role of habitats in supporting multiple overlapping functional groups. From 55,856 trap nights, we recorded 33 species from six guilds: carnivores; frugivores; granivores; herbivores; insectivores; and omnivores. All guilds were negatively correlated with elevation and positively correlated with primary forest. Five areas with high multi-guild overlap were identified and recommended for increased protection and other conservation measures, such as increasing the frequency of SMART patrols. Our data-driven guild-level approach for improving conservation practice has high relevance to other biodiversity-rich countries. Further utility of this guild approach, with potential future refinement and improvement, should greatly assist PA managers with improving area-based conservation effectiveness, such as higher patrol frequencies and or prioritizing wildlife, and habitat and ecosystem inventory, under-pinned by enhanced research, and cost-efficient budget allocation.