Threshold and weighted-distance methods: a combined multiscale approach improves explanatory power of forest carabid beetle abundance in agricultural landscape.

Context: To analyze the scales at which landscape structure influences ecological processes, two approaches with different underlying ecological assumptions exist; the usual threshold method and the weighted-distance method. Objectives: We used abundance of species to test if the combination of weighted-distance and threshold approaches improves the explained variance of landscape metrics. Methods: We developed a workflow using the two approaches to calculate metrics computed at multiple scales. The latter was developed using weighted metrics based on different weighted-distance functions, and one metric could be selected for more than one spatial scale. Then, we tested the explained variance of species distribution (the activity-density of Abax parallelepipedus) by these two approaches applied independently and then together in modeling a specific ecological response. Results: The combination of metrics computed at multiple scales calculated by both weighted-distance and threshold method improved the predictive performance of the models. More precisely, adding metrics derived from the weighted-distance method to the threshold method significantly increased the explained variance when using the same environmental variables. The mean R² values of the selected model for the threshold method was 0.34 ± 0.10, 0.49 ± 0.11 with the weighted-distance method, and reached 0.71 ± 0.07 with the two methods combined. These results demonstrate the importance of combining metrics using the weighted-distance method and the threshold method. In addition, activity-density was better explained by metrics selected at multiple scales. Conclusions: This study highlights the importance of combining threshold and weighted-distance method at several scales to improve the explanation of ecological responses based on species abundance. [ABSTRACT FROM AUTHOR]