RESUMEN
Effective conservation requires understanding the processes that determine population outcomes. Too often, we assume that protected areas conserve wild populations despite evidence that they frequently fail to do so. Without large-scale studies, however, we cannot determine what relationships are the product of localized conditions versus general patterns that inform conservation more broadly. Leopards' (Panthera pardus) basic ecology is well studied but little research has investigated anthropogenic effects on leopard density at broad scales. We investigated the drivers of leopard density among 27 diverse protected areas in northeastern South Africa to understand what conditions facilitate abundant populations. We formulated 10 working hypotheses that considered the relative influence of bottom-up biological factors and top-down anthropogenic factors on leopard density. Using camera-trap survey data, we fit a multi-session spatial capture-recapture model with inhomogenous density for each hypothesis and evaluated support using an information theoretic approach. The four supported hypotheses indicated that leopard density is primarily limited by human impacts, but that habitat suitability and management conditions also matter. The proportion of camera stations that recorded domestic animals, a proxy for the extent of human impacts and protected area effectiveness, was the only predictor variable present in all four supported models. Protected areas are the cornerstone of large felid conservation, but only when the human-wildlife interface is well managed and protected areas shelter wildlife populations from anthropogenic impacts. To ensure the long-term abundance of large carnivore populations, reserve managers should recognize the ineffectiveness of "paper parks" and promote contiguous networks of protected areas that offer leopards and other large mammal populations greater space and reduced human impacts.