First, do no harm: A systematic review of deforestation spillovers from protected areas

Land-use restrictions in protected areas (PAs) might have unintended spillover effects on non-target, neighboring areas. In the case of leakage, land-use change that would have occurred in the PA is displaced to an unprotected area where it would not have otherwise occurred. The resultant 'leakages' can offset benefits achieved inside PAs, confound impact assessments, and exacerbate the problems of opportunistic protection of lands. Conversely, in the case of 'blockage', the unprotected surroundings experience less land-use change than would have otherwise occurred due to a positive spillover effect from nearby protected areas. Little is known about the magnitude, ubiquity, and predictability of spillovers that have already occurred in the global PA network. Here we systematically review the literature and collate the existing evidence to quantify deforestation spillovers. We calculated deforestation rates within 3,398 PAs, most of which were found in tropical and subtropical moist forests, and in their unprotected adjacent surroundings and compared these rates to a baseline derived from the wider landscape. Of the 2,575 PAs that effectively restricted deforestation rates within their bounds relative to the baseline, 11.8% showed leakage and 54.8% exhibited blockage. Deforestation rates in the remaining 33.4% were indistinguishable from their respective baselines. Linear modelling of the correlates of leakage and blockage showed that PA-specific characteristics like size and IUCN category were uninformative, whereas national-scale socioeconomic factors like population density and GDP were useful predictors. Although spillovers from land-use restrictions are ultimately driven by socioeconomic factors, their ecological consequences are such that PA assessments should routinely and explicitly account for displaced impacts to the unprotected surroundings. (C) 2019 The Authors. Published by Elsevier B.V.