Abstract
Predicting range shifts of marine species under different CO2 emission scenarios is of paramount importance to understand spatial potential changes in a context of climate change and to ensure appropriate management, in particular in areas where resources are critical to fisheries. Important tools which use environmental variables to infer range limits and species habitat suitability are the species distribution models or SDMs. In this work, we develop an ensemble species distribution model (e-SDM) to assess past, present and future distributions under Representative Concentration Pathway (RCP) 8.5 of nine demersal species and hotspot areas for their two life stages (adult and juvenile) in the Adriatic and Western Ionian Seas in four time windows (1999-2003, 2014-2018, 2031-2035 and 2046-2050). The e-SDM has been developed using three different models (and sub-models), i.e. (i) generalized additive models (GAM), (ii) generalized linear mixed model (GLMM), (iii) gradient boosting machine (GBM), through the combination of density data in terms of numbers of individuals km 2 and environmental variables. Then, we have determined the changes in the aggregation hotspots and distributions. Finally, we assess gains and losses areas (i.e. occupation area) in the future climate change scenario as new potential range shifts for the nine species and their life stages. The results show that densities of some key commercial species, such as Merluccius merluccius (European hake), Mullus barbatus (red mullet), and Lophius budegassa (anglerfish) will be shifting northwards.