Finding the functional grain: comparing methods for scaling resistance surfaces

The influence of landscape features on the movement of an organism between two point locations is often measured as an effective distance. Typically, raster models of landscape resistance are used to calculate effective distance. Because organisms may experience landscape heterogeneity at different scales (i.e. functional grains), using a raster with too fine or too coarse a spatial grain (i.e. analysis grain) may lead to inaccurate estimates of effective distance. We adopted a simulation approach where the true functional grain and effective distance for a theoretical organism were defined and the analysis grains of landscape connectivity models were systematically changed. We used moving windows and grains of connectivity, a recently introduced landscape graph method that uses an irregular tessellation of the resistance surface to coarsen the landscape data. We then used least-cost path metrics to measure effective distance and found that matching the functional and analysis grain sizes was most accurate at recovering the expected effective distance, affirming the importance of multi-scale analysis. Moving window scaling with a maximum function (win.max) performed well when the majority of landscape structure influencing connectivity consisted of high resistance features. Moving window scaling with a minimum function (win.min) performed well when the relevant landscape structure consisted of low resistance regions. The grains of connectivity method performed well under all scenarios, avoiding an a priori choice of window function, which may be challenging in complex landscapes. Appendices are provided that demonstrate the use of grains of connectivity models.