GESTE results.

Results of seven regression analyses performed in GESTE to examine how much of the variation in genetic structure among Pristionchus pacificus populations on La Réunion Island is explained by variation in environmental factors. The first column shows the environmental factor(s) tested in a given analysis. The second and third columns correspond to the model which best explains the data (where the constant term corresponds to bias that is not accounted for by the terms in the model), and its probability (P), respectively. The remaining four columns give the probability of each individual factor in the highest probability model, with the factor in the far left column corresponding (from left to right) with labels G1, G2 and/or G3 in subsequent columns, and G1*G2 referring to the interaction between factors G1 and G2. Thus, in the first three tests, the probability of the individual factors (environmental variables) in the model is less than that of the a model including only random (constant) effects, whereas in the subsequent two-factor tests, the highest probability models were those that included both factors and their interaction (see Results for further information).