Table 2 Results of the multivariate analyses performed with the point approach

Species	Genetic distance	Spatial predictors (MEM)	Environmental factors	r (with MEMs)	β (with MEMs)	U (without MEMs)	C (without MEMs)
*C. elaphus*	BCD R² = 0.264*	84 positive	Elevation	− 0.488	− 0.461*	0.1739	0.0640
		0 negative	Agricultural areas	0.214	0.175*	0.0199	0.0257
			Broad leaved forests	0.068	0.024	0.0004	0.0043
	a _R R² = 0.211*	83 positive	Elevation	− 0.437	− 0.362*	0.0828	0.1080
		0 negative	Artificial areas	0.263	0.126*	0.0131	0.0560
			Coniferous forests	− 0.331	− 0.060	0.0021	0.1075
	LKC R² = 0.160*	83 positive	Elevation	0.138	0.163*	0.0218	− 0.0029
		0 negative	Agricultural areas	− 0.362	− 0.250*	0.0402	0.0907
			Broad leaved forests	0.264	0.182*	0.0200	0.0497
S. *scrofa*	BCD R² = 0.193*	98 positive	Elevation	0.288	0.233*	0.0371	0.0458
		0 negative	Artificial areas	− 0.350	− 0.111*	0.0068	0.1157
			Broad leaved forests	0.268	0.255*	0.0487	0.0228
			Mixed forests	0.177	0.107*	0.0100	0.0212
	a _R R² = 0.034*	117 positive	Artificial areas	− 0.086	− 0.022	0.0004	0.0069
	a _R R² = 0.034*	0 negative	Broad leaved forests	0.183	0.175*	0.0266	0.0069
	LKC	124 positive	Mixed forests	− 0.179	− 0.241*	–	–
	R² = 0.043*	0 negative

For each species and each genetic distance, the table first provides the number of retained positive and negative spatial predictors (MEM) and the list of retained environmental factors. For each environmental factor, the table also reports: the Pearson’s correlation coefficient (r), the β weights (β) and the corresponding p-value with Benjamini–Hochberg correction after suppression of spatial autocorrelation in model residuals, as well as the unique and common contributions (U and C) of environmental factors to the variance in the dependent variable. Unique and common contribution of each factor were derived from CA (commonality analysis) performed in absence of spatial predictors (MEM). Genetic distance values were estimated with MAPI and environmental values were obtained by averaging environmental raster values falling in each corresponding hexagonal cell. Spatial predictors were computed as Eigenvector maps from a spatial connectivity model based on a Delaunay triangulation. (*) refers to significant determination coefficients R² or β-values (p-values < 0.05 after Benjamini–Hochberg correction)

Search