Humans are altering the most basic patterns of biodiversity at a startling rate. Over the past centuries, we appropriated much of the land surface for our own use, in turn degrading the habitat of many species. We are now entering a new era, in which climate change is expected to match, or perhaps dwarf the effects of land-use change on biodiversity. Indeed, a host of global studies have suggested that moving forward climate change will outpace local land-use change in terms of the net impact on biodiversity. However, we argue that this conflicts with decades of localized research and points to a mismatch in the spatial scale of the available data, the types of models used, and the mechanisms considered by these global analyses. Using over 3 million observations of bird species across the entirety of Costa Rica, and 500,000 observations of birds over 20 years and 48 sites situated along gradients of both climate and land-use we are able generate models of bird occurrence and abundance for 251 species.
Results/Conclusions
Unlike traditional species distribution models, which show only 5-10% of the variance explained by land-use our hybrid models with local data show that local land-use is a more important predictor than climate for 163 of our 251 species, while climate is only more important for 9 species. Moreover, categorical land-cover failed to perform as well as our locally descriptive vegetation characteristics for 93% of our species, indicating pitfalls with the current status quo of assessing species responses to land-use. These modified models show a much higher sensitivity to changes in land-use than traditional models, placing the impacts of relatively modest changes in vegetation characteristics on par with climate change scenarios. These results underscore the need for more precise data on our environment and biodiversity to ensure that we are safeguarding species and the benefits they provide under a changing climate and landscape.