The scale of effect of landscape context varies with the species’ response variable measured

Abstract

To detect an effect of landscape context on a species’ response, the landscape variables need to be measured within the appropriate distance from the species’ response, i.e. at the scale of effect. However, it is not clear what factors determine the scale of effect. Our objective was to test the prediction that the scale of effect should be smallest when the species’ response is fecundity, larger for abundance, and larger still for occurrence. We compared the scale of effect of two landscape variables (road density, forest proportion) on the three responses (fecundity, abundance, occurrence) for the wood frog (Lithobates sylvaticus) in eastern Ontario, Canada. We used egg mass surveys of 34 ponds to estimate fecundity (mean eggs/mass), abundance (number of masses), and occurrence (presence/absence of egg masses). We then empirically estimated the scale of effect of each landscape variable on each response. The scale of effect differed among responses, from 0.2 to 3.0 km radii; however, it did not vary in the predicted order. Furthermore, the order was not consistent between the two landscape variables. Our results show that the scale of effect of a landscape variable on a given species can differ for different response variables. However, they also suggest that these differences in the scale of effect are not predictable. Thus, the most reliable way to ensure a landscape context study is conducted at the correct spatial extent is to estimate the scale of effect empirically, rather than ‘guesstimating’ the extent a priori.