Predictions of Impacts of Climate Change on Animal Species Suffer From Model Discrepancies

One of the claimed impacts of climate change is that future climates will disrupt the environments for species, causing them to become at risk from local extinction and the need to migrate to new regions. One way to assess this risk is to use species distribution models. Such models predict habitat quality (based on frequency of presence) for a species as a function of habitat and climate variables. In this study logistic regression habitat suitability models were developed for four animals in Spain: an amphibian, a reptile, a bird, and a mammal, based on current climate. These models were reasonably accurate at predicting current distributions of the species. Two climate models, CGCM2 and ECHAM4 were used to simulate the climate of Spain through 2100, each using two IPCC emissions scenarios, A2 and B2. The downscaling of the model data across Spain had been previously conducted by the Spanish National Meteorological Agency. To compare the favorability maps generated by the climate simulations applied to the species distribution models, the authors used fuzzy sets to assess overlap in spatial distributions.

The study says: "our results show that discrepancies between predictions were more attributable to the lack of knowledge concerning the effect of oceans and atmosphere on climate (general circulation models) than to the putative effect of different emissions scenarios on future climate." The study found that the two climate models they tested produced species distribution results that differed more than the results from the two emissions scenarios, showing that there is not sufficient model agreement to allow models to be used to infer species habitat impacts over the next 100 years. The authors state: "our conclusion is that species distribution models in climate warming scenarios are still not useful for informing emission policy planning." This suggests that the claims of model agreement do not extend to the regional scale, the very scale at which impacts must be assessed.