Modeling the impact of climate change on corvus species distribution in Somaliland: Bayesian spatial point process approach for conservation

Introduction This study aimed to predict the spatial distribution of Corvus edithae (Somali crow) in Somaliland and explore its relationship with climatic covariates. Methods We applied a log-Gaussian Cox process model, utilizing the R-INLA package. The model integrated spatial climatic covariates (mean annual temperature, precipitation, temperature extremes, solar radiation, wind speed) alongside structured and unstructured random effects to address spatial autocorrelation and unexplained variability. Results Posterior mean estimates for climatic covariates showed wide 95% credible intervals encompassing zero, indicating substantial uncertainty regarding their specific effects. Conversely, the model revealed highly influential and statistically significant spatially structured (autocorrelation) and unstructured random effects. Model predictions indicated generally low occurrence intensities, with coastal areas exhibiting the highest expected densities, suggesting their importance as potential core habitats. Convergence diagnostics indicated model reliability. Discussion The findings underscore that unmeasured spatial factors and environmental heterogeneity are dominant drivers of Corvus edithae distribution, outweighing the influence of the broad-scale climate variables tested. This study provides a robust Bayesian spatial point process framework for conservation ecology, particularly where spatial patterns are prominent and data may be uncertain.