| Understanding of species distributions is essential to inform a variety of conservation decisions, including reserve design, assessment of species' extinction risk, management of populations of invasive species, identification of potential reintroduction sites, among others. Although some regions of the world have long term geographically comprehensive surveys of some taxa allowing to accurately determining their species distribution, for most regions of the world and taxa occurrence data comes from ad-hoc surveys such as natural history collections and recently, citizen-science initiatives. This situation may be particularly acute in biodiversity hotspots, which are in dire need of conservation actions. Can we inform conservation decision making based on these incomplete, biased and noisy datasets? In this thesis, I use species distribution modeling as a tool to assess the current conservation status, extinction risk and the threat posed by global climate change on birds of conservation concern in Colombia, a data-poor but highly diverse country. Specifically, I developed a novel method to incorporate covariate uncertainty, stemming from locational errors, in species distribution models. Besides fully accounting for covariate uncertainty, in some cases this new method can improve distribution predictions over existing approaches. Then, I developed a conceptual framework for the use of species distribution models in IUCN extinction risk assessments, which was used to assess the extinction risk of 127 species of birds of Colombia using geographic range size criteria. Lastly, I evaluated the effects of climate change on species distribution, community structure, and conservation in protected areas for 157 birds of conservation concern. This thesis provides both methodological tools and a framework for the use of species distribution models in conservation assessments in data poor regions. |
