4. Discussion
Building bioclimate models for a range of freshwater fishes in Great Britain under current and future climate scenarios predicted that the climate space of each species would shift, with centroid displacement always in a north-westerly direction. Predictions revealed that even under low emission climate change scenarios and the two dispersal scenarios, alterations in climate space were generally consistent at the family level, with constriction for the freshwater life-stages of native salmonid fishes and a large increase in the climate space of most cyprinid species, including the non-native C. carpio, other than under the no dispersal scenario when their distribution would be unchanged. Across all the modelled fishes, annual mean temperature was the most important variable in the ensemble model predictions. There was, however, some inter-family variability across the importance of the climatic variables, with a combination of annual mean precipitation and temperature variables being most important for the two salmonid species. For all other modelled species, temperature variables were most important, with their combined importance to the ensemble models being a minimum of 92%. The limitation of the modelling architecture to the use of only two dispersal scenarios limited the distribution predictions to scenarios of no dispersal and full dispersal. It is highly probable that some dispersal within and between some catchments of species will occur naturally and/or by anthropogenic means (Conti et al., 2015). This is thus a limitation of the method used and so it is recommended that future work incorporates the ability of freshwater fish to disperse in relation to hydrological connectivity and the changes in their environments. The modelling approach was based on bioclimatic variables, using the assumption that the influence of climate on the biology and ecology of the fishes was the key determinant of their distribution pattern (Woodward and Beerling, 1997; Hampe, 2004).