Modeling tree spatial distributions after partial harvesting in uneven-aged boreal forests using inhomogeneous point processes

• We examined the spatial position of trees in partially cut forest stands.
• We report that systematic patterns can be identified.
• We show that such patterns affect the further growth of trees.
• We developed a stand structure simulator using a Gibbs point process model.

Old-growth forests are characterized by their compositional and structural complexity, which contributes to local biodiversity. To maintain these old-growth attributes in boreal forests, traditional even-aged management systems based on clear-cutting are gradually shifting to ecosystem-based management, which includes partial harvesting in uneven-aged stands. We investigated the spatial structure of trees within 31 plots that had been established in Canadian boreal forest after selection cutting. We used an approach based on heterogeneous point processes to investigate the relationship between tree density and distance to the nearest logging trail. Tree density was positively and log-linearly related with distance to logging trails, with large trees being particularly scarce near the trails. Using a statistical approach based on mixed models with random plot effects that accounted for distance to the nearest trail, we assessed how the spatial pattern of trees affected the diameter increment of trees. Seven-year diameter increments of trees with diameters <14 cm were negatively related to the distance to logging trails. Accordingly competitive interactions among trees left after partial harvesting, which aim at preserving old-growth forest attributes, were different from those of natural boreal forests. Based on these results, a stochastic model was applied using heterogeneous Gibbs point processes to realistically describe and simulate the location of residual trees after selection cutting in boreal stands that were represented by forest inventories without spatial information on tree location.