A selection harvesting algorithm for use in spatially explicit individual-based forest simulation models

Abstract

There is growing interest in using spatially explicit, individual-based forest simulation models to explore the ecological and silvicultural consequences of various harvesting regimes. However, simulating the dynamics of managed forests requires harvesting algorithms capable of accurately mimicking the harvest regimes of interest. Under selection silviculture, trees are harvested individually or in small groups, with the aim of retaining trees across a full range of size classes. An algorithm that reproduces selection harvesting must therefore be able to recreate both the spatial and the structural patterns of harvest. Here we introduce a selection harvest algorithm that simulates harvests as a contagious spatial process in which the cutting of one tree affects the probability that neighboring trees are also cut. Three simple and intuitive parameters are required to implement this process: (1) the probability of cutting a “target” tree ( P t) (often a function of tree size), (2) the probability of cutting its nearest neighbor ( P n), and (3) the total number of target trees to cut ( N t). Specification of these parameters allows representation of both the spatial and the structural patterns of harvest expected under selection silviculture. Based on this simple process, we built two different versions of the harvesting algorithm. An “empirical” algorithm was designed and calibrated to reproduce the observed spatial and size distribution of stumps (harvested trees) at a study site in central Ontario, and was successful in reproducing harvesting patterns found in the field, notably variability in the cluster size of harvested trees. The “user-defined” algorithm implements alternative harvesting regimes (user-defined harvest targets), which may differ in the intensity of harvesting, the size-specificity of harvesting, and the spatial pattern of harvesting. We show that the user-defined harvesting algorithm succeeds in meeting harvest targets specified by the user (e.g., size class distribution and basal area of trees harvested), while simultaneously adjusting the gap size specified (i.e., the distribution of harvested trees per cluster). Incorporation of this harvesting algorithm into spatially explicit, individual-based models will permit analyses of long-term responses of forest stands to harvesting scenarios that more realistically capture the complex patterns of within-stand variability generated by selection silviculture as practiced in actual managed forests.