Evaluation of CORDEX-CORE RCMs in representing BioClim variables for mosquito distribution models in Mediterranean regions

An adequate representation of the distribution of mosquitoes transmitting vector-based diseases under future climate change conditions is essential to estimate the occurrence of those diseases in areas such as the Mediterranean. For this purpose, species distribution models are used to establish a statistical relationship between recent and future environmental conditions and the spread of species. The 19 BioClim variables have been widely used to drive these models because they represent key features (e.g. hot, cold, dry, humid) of different temporal dimensions (quarter or month). The scientific community has paid less attention to the difficulties that arise when using BioClim variables, including how well the climate model can reproduce them. This study intended to assess the skill of 10 different high-resolution regional climate models (RCMs) of the Coordinated Regional Climate Downscaling Experiment’s CORE initiative that represent BioClim variables in Mediterranean regions. The skill analysis was implemented using the metrics of spatial correlation, spatial standard deviation and the mean absolute percentage error. The result reveals that REGCM4 and REMO15 are the best-performing RCMs in terms of standard deviation and the mean absolute percentage error, and RACMO and CCLM4 are the best-performing RCMs in terms of spatial correlation. The result shows that temperature-based BioClim variables are better represented by the majority of RCMs compared to precipitation models. A huge uncertainty remains when it comes to the representation of quarter-based variables. Hence, future studies should be focused on improving the representation of BioClim variables, especially those related to precipitation, by applying appropriate bias correction techniques. The best-performing RCMs of this study will be used to derive different species distribution models of Anopheles mosquitoes over the Mediterranean regions to estimate the future distribution of vector-based diseases.