Explaining the global pattern of protected area coverage: relative importance of vertebrate biodiversity, human activities and agricultural suitability

Aim  Twelve per cent of the Earth’s terrestrial surface is covered by protected areas, but neither these areas nor the biodiversity they contain are evenly distributed spatially. To guide future establishment of protected areas, it is important to understand the factors that have shaped the spatial arrangement of the current protected area system. We used an information‐theoretic approach to assess the ability of vertebrate biodiversity measures, resource consumption and agricultural potential to explain the global coverage pattern of protected areas. Location  Global. Methods  For each of 762 World Wildlife Fund terrestrial ecoregions of the world, we measured protected area coverage, resource consumption, terrestrial vertebrate species richness, number of endemic species, number of threatened species, net primary production, elevation and topographic heterogeneity. We combined these variables into 39 a priori models to describe protected area coverage at the global scale, and for six biogeographical realms. Using the Akaike information criterion and Akaike weights, we identified the relative importance and influence of each variable in describing protected area coverage. Results  Globally, the number of endemic species was the best variable describing protected area coverage, followed by the number of threatened species. Species richness and resource consumption were of moderate importance and agricultural potential had weak support for describing protected area coverage at a global scale. Yet, the relative importance of these factors varied among biogeographical realms. Measures of vertebrate biodiversity (species richness, endemism and threatened species) were among the most important variables in all realms, except the Indo‐Malayan, but had a wide range of relative importance and influence. Resource consumption was inversely related to protected area coverage across all but one realm (the Palearctic), most strongly in the Nearctic realm. Agricultural potential, despite having little support in describing protected area coverage globally, was strongly and positively related to protection in the Palearctic and Neotropical realms, as well as in the Indo‐Malayan realm. The Afrotropical, Indo‐Malayan and Australasian realms showed no clear, strong relationships between protected area coverage and the independent variables. Main conclusions  Globally, the existing protected area network is more strongly related to biodiversity measures than to patterns of resource consumption or agricultural potential. However, the relative importance of these factors varies widely among the world’s biogeographical realms. Understanding the biases of the current protected area system may help to correct for them as future protected areas are added to the global network.