Comparing connectivity outcomes from random-walk penetration into the matrix, gap crossing, and a combination of both.

We depict random-walk penetration into the matrix (open points, dashed lines), gap crossing (×, full lines), and the combination of both (full squares, bold lines), for the average total number of interpatch movements during home range (first line of subplots), the proportion of patches that received visits during home range (second line), the average total number of interpatch movements during dispersal (third line), and the proportion of patches that received visits during dispersal (lowest line of subplots). Results on the first and third column are for the least fragmented landscape (50% forest cover, clustered – cf. Figure 2f), results on the second and fourth column are for the most fragmented landscape (10% forest cover, fragmented – cf. Figure 2a). Results in the two left columns of subplots are for cell-quality dependent mortality, and in the two right-hand columns mortality is fixed per habitat type. X values represent the location of the 50% decay point of the sigmoid function (b in equation 1) along the interior-exterior forest ecotone. The point at X = 0 is depicted only for gap crossing but serves as reference both for no gap crossing and no usage of the edge. Note the different y-axis scales when the number of movements are counted (see red markings). Results for the probability of interpatch movements (per bird, per step) are not shown, as they were qualitatively similar to those obtained from the total number of interpatch movements.