Abstract
Retention forestry involves leaving single or groups of unharvested trees within harvest areas. Patch retention, which resembles structures such as unburned patches remaining after wildfire, is one practice implemented within the framework of Ecosystem-based Forest Management (EBM), which seeks to use natural forests as a model and minimize differences in natural and managed forests. Despite the widespread adoption of patch retention practices, few comparisons of the attributes of postfire and postharvest islands, or their drivers, have been made. Given the importance of deadwood in forests to a variety of ecosystem functions, we sought to compare the local bioenvironmental drivers of deadwood (snags, CWD) C stocks in islands remnants in postfire and postharvest forests a decade after disturbance. We also determined whether their relative effects are consistent across deadwood types (snags, CWD) and disturbance regimes using generalized additive mixed models with study site as random factor in all cases. A candidate model with initial stand volume (ISV), basal area of live trees, and size heterogeneity of live trees best predicted snag and CWD C stocks in both disturbance types, but their relative importance was inconsistent. The ISV had significantly (p < 0.05) positive effects on C stocks in snags and CWD across disturbance types, but its relative effects was higher in retention islands than fire islands. In all cases, stand density of remnant live trees was negatively related to deadwood C stocks. Conversely, the size heterogeneity of remnant live trees significantly boosted deadwood C stocks in fire islands but not in harvest islands. The results imply consideration for the stocking level of candidate forest areas for retention patches as this drives the evolution of deadwood accumulation in the postharvest islands.