Table 4.

Results of Bayesian inference and model selection in SUNDER testing the relative effects of geographical and environmental variables on genetic differentiation among eleven populations of the Pyrenean Morales grasshopper

		Likelihood and (β i) for each model
		G	E	G + E
Environmental variable	CLIMDIS	−469.43 ( β G  = 14.95)	−471.44 (β E = 25.44)	−469.78 (β G = 25.50) (β E = 21.31)
	ELEVDIS	−460.11 ( β G  = 23.01)	−461.24 (β E = 31.08)	−460.63 (β G = 17.45) (β E = 31.11)

We separately tested the environmental variables [elevation (ELEV_DIS) and climatic (CLIM_DIS) dissimilarity matrices] against an IBD resistance distance matrix (i.e. equal conductance to all pixel values, equivalent to geographic distance). ‘G’ corresponds to models only considering geography, ‘E’ corresponds to models only considering the environmental variable, and ‘G + E’ corresponds to models considering both of them. For all runs, we show the likelihood of each model based on the validation dataset and the values of β _i parameter. The β _i parameter quantifies the magnitude of the effect of each variable on genetic covariance (small values correspond to a strong decreasing of the genetic covariance with increasing geographical or environmental distance, i.e. small values indicate an important effect of such variable). The most likely model for each comparison is in bold