A spatio-temporally explicit distribution model to support conservation strategies for isolated small-ranged species: a case study from Gal´apagos

Species Distribution Models (SDMs) represent the gold standard to investigate how environmental factors affect the occurrence of animal species and estimate their actual and potential distribution. However, SDMs often lack a formal hypothesis-driven selection of environmental predictors and require independence among observations. In contrast, actual occurrence data are often spatially and temporally autocorrelated, particularly when most observations originate from a limited number of surveys, as frequently happens in studies focusing on isolated, small-ranged species of high-conservation priority. Here, we explored the application of a hierarchical Bayesian spatio-temporal SDM to model the probability of occurrence of small-ranged species living in isolated and difficult-to-access areas, combining Integrated Nested Laplace Approximation (INLA) with Stochastic Partial Differential Equation (SPDE). To do so we used the Gal´apagos Pink Land Iguana (Conolophus marthae) as a model organism. This species is well suited for our scope in being critically endangered and endemic to an extremely small and remote area on Wolf Volcano on Isabela Island, Gal´apagos. Our approach enabled us to obtain reliable estimates of the effects of environmental variables on species distribution using opportunistic occurrence data, which exhibit a pronounced spatio-temporal structure. A spatially explicit cross-validation of the model demonstrated that the INLA-SPDE approach allows elucidating the relationship between small-ranged species and their areas of distribution, resulting in accurate predictions (Cohen’s K = 0.86; True skill statistics = 0.84). We also used the INLA-SPDE model to map the actual and potential distribution of the species across the Gal´apagos archipelago. Our approach provided a first formal measure of the range of C. marthae (ca. 42 km2) and indicated that abundance of trophic resources, availability of open areas, and terrain roughness limit species’ distribution. We also identified ca. 400 km2 of suitable areas clustered into three main sites outside the species’ actual range. These results constitute an essential starting point to evaluate a species translocation, a key conservation strategy outlined in the Conservation and Management Plan for this species.