Advances in climate models from CMIP3 to CMIP5 do not change predictions of future habitat suitability for California reptiles and amphibians

Understanding how predicted species responses to climate change are affected by advances in climate modeling is important for determining the frequency with which vulnerability assessments need to be updated. We used ecological niche models to compare predicted climatic habitat suitability for 132 species of reptiles and amphibians in California, USA under the previous and current generations of climate simulations from the Coupled Model Intercomparison Project (CMIP3 and CMIP5). We used data from seven global climate models for future (2014–2060) predictions under the following greenhouse gas emissions scenarios: SRES A2 for CMIP3 and RCP 8.5 for CMIP5. Ensembles of these climate models predicted a warmer and slightly wetter future California on average: CMIP3 + 2 °C mean annual temperature, +15 mm annual precipitation, CMIP5 + 2.5 °C mean annual temperature, +24 mm annual precipitation. CMIP3 and CMIP5 ensembles differed in where precipitation changes were predicted to be largest, with CMIP3 predicting greatest increased precipitation in the northern deserts and CMIP5 predicting greatest increased precipitation in the northern mountains. Under both sets of climate models (CMIP3 and CMIP5), mean habitat suitability within species ranges was predicted to decrease in the future. The degree of predicted decline was similar on average for CMIP3 and CMIP5, −15 % and −13 % respectively, suggesting that conclusions drawn from previous studies using ensembles of CMIP3 models are robust, at least for California. However, the effect of CMIP3 vs. CMIP5 on future mean habitat suitability depended strongly on which GCM was used: three GCMs predicted little change in future habitat suitability between CMIP3 and CMIP5 (MIROC, CNRM, GFDL), three predicted greater reductions in habitat suitability under CMIP3 (MPI, GISS, IPSL), and one predicted greater reductions in habitat suitability under CMIP5 (MRI). We conclude that habitat suitability assessments under CMIP3 made using more than 3 GCMs are likely to remain broadly applicable, while those made using 3 or fewer may be conservation priorities for re-evaluation under CMIP5.