Conservation is an inherently human endeavor - initiated, designed, and deployed by humans to alter future behavior or undo previous impacts to affect positive changes in biodiversity. Predicting where conservation will occur in the future, however, remains a challenge. We propose a conceptual model where conservation action is determined by gradients of ecological value, individual willingness, and institutional ability. Conservation action may occur at any location along this 3-dimensional continuum, but becomes increasingly likely as ecological values, individual willingness, and institutional ability simultaneously approach their maxima. We demonstrate this approach using available high-resolution, spatially explicit data on demographics, economic drivers, institutional characteristics, and environmental conditions to evaluate the degree to which the spatial coincidence of these factors affects the likelihood of conservation. We use Bayesian hierarchical models that treat past conservation action as probabilistic outcomes of the interaction of ecological, institutional, and social covariates to identify key explanatory variables influencing the likelihood of conservation action using multi-model inference and hierarchical variance partitioning to evaluate the relative importance of each factor in explaining past conservation. We then implement these models in a GIS to generate probabilistic surfaces of the likelihood conservation action to identify where conservation is likely in the future.
Results/Conclusions: We evaluated conservation actions taken by individuals, municipalities, and federal agencies in the United States at different data resolutions and spatial scales. Our results suggest that it is possible to identify areas where conservation action is likely; however, the relative importance of ecological, social, and institutional factors driving those actions varies geographically and by action type. For example, membership in some conservation organizations is explained by both demographic factors and the distance from intact natural habitats. In contrast, designation of county open space tends be better explained by population demographics and municipal economic indicators rather than ecological condition. Finally, the relationship between our predictors and federal agency conservation action is highly dependent on the scale at which covariates are summarized. Our results can help characterize trade-offs among conservation actions, identify socio-political opportunities and impediments to conservation, and generate hypotheses regarding the effects of changing human communities on conservation likelihood.