Aim: Although it is recognized that ecological patterns are scale dependent, the exact scales over which specific ecological processes operate are still a matter of controversy. In particular, understanding the scales over which biotic interactions operate is critical for predicting changes in species distributions in the face of the ongoing biodiversity crisis. It has been hypothesized that biotic interactions operate predominately at fine grains, yet this conjecture has received relatively little empirical scrutiny. We use US woodpeckers as a model system to assess the relative importance of biotic interactions, environmental suitability and geographic proximity to other intraspecific occurrence sites, across scales.

Methods: We combined species occurrence data from the North American Breeding Bird Survey (BBS) with a large pair-wise interaction matrix describing known interactions among woodpeckers and other bird species. Using a logistic mixed modelling framework we then established the relative importance of biotic interactions as predictors of woodpecker occurrences in relation to environment and geographic proximity to intraspecific occurrence sites.

Results: We found that geographic proximity becomes a stronger predictor of woodpecker occurrence as grain becomes coarser, while environment is grain-invariant. As opposed to environment and geographic proximity, we found that when the focal species experienced positive biotic interactions, the importance of interactions decreased with increased grain. However, positive interactions remained important up to a grain size of entire BBS routes (c. 40 km), which is much coarser than the grain size used by most species distribution models. In contrast, when the focal species experienced negative interactions we did not find clear grain dependence.

Main conclusions: Biotic interactions (both positive and negative) are important predictors of species occurrences. While these interactions are strongest at fine grains, they can remain important even at coarse grains, and are thus critical for predicting distributional changes in the face of the ongoing biodiversity crisis.