Seeing the forest for the heterogeneous trees: stand-scale resource distributions emerge from tree-scale structure.

Forest ecosystem processes depend on local interactions that are modified by the spatial pattern of trees and resources. Effects of resource supplies on processes such as regeneration are increasingly well understood, yet we have few tools to compare resource heterogeneity among forests that differ in structural complexity. We used a neighborhood approach to examine understory light and nutrient availability in a well-replicated and large-scale variable-retention harvesting experiment in a red pine forest in Minnesota, USA. The experiment included an unharvested control and three harvesting treatments with similar tree abundance but different patterns of retention (evenly dispersed as well as aggregated retention achieved by cutting 0.1- or 0.3-ha gaps). We measured light and soil nutrients across all treatments and mapped trees around each sample point to develop an index of neighborhood effects (NI). Field data and simulation modeling were used to test hypotheses that the mean and heterogeneity of resource availability would increase with patchiness because of greater variation in competitive environments. Our treatments dramatically altered the types and abundances of competitive neighborhoods (NI) in each stand and resulted in significantly nonlinear relationships of light, nitrogen and phosphorus availability to NI. Hence, the distribution of neighborhoods in each treatment had a significant impact on resource availability and heterogeneity. In dense control stands, neighborhood variation had little impact on resource availability, whereas in more open stands (retention treatments), it had large effects on light and modest effects on soil nutrients. Our results demonstrate that tree spatial pattern can affect resource availability and heterogeneity in explainable and predictable ways, and that neighborhood models provide a useful tool for scaling heterogeneity from the individual tree to the stand. These insights are needed to anticipate the outcomes of silvicultural manipulations and should become more holistically integrated into both basic ecological and management science.