Seedbed proportions in and outside skid trails: Temporal variation following selection cutting in northern hardwood forests

Partial harvesting during the snow-free season disturbs the forest floor and modifies seedbed characteristics. Quantitative information is lacking about the distribution of changes in seedbed proportions between areas located in and outside skid trails, and about how long these changes persist after harvest. These effects could interact with species' seedbed requirements and seed input and influence spatio-temporal patterns of seedling establishment in forests, which would have important implications for regeneration dynamics. The objectives of this study were to determine how selection cutting affects seedbed proportions in and outside skid trails in northern hardwood stands, how these seedbed proportions vary over time following harvest, and how seedbed proportions in selection cuts compare with unharvested stand conditions. We sampled 12 sugar maple-dominated stands in southeastern Quebec, Canada. Two had not been harvested in the recent decades, while 10 had been harvested through selection cutting, 1-3 years earlier. A total of 3600 quadrats were sampled to determine the proportions of 8 seedbed types and whether or not a quadrat was located in a skid trail. This was the case for 24.1% of the quadrats in selection cuts. Outside skid trails, seedbed proportions in selection cuts did not vary from those in unharvested stands (P ⩾ 0.097). In these stands, leaf litter was the most abundant substrate, covering 87.3% of the forest floor, followed by rotten wood (4.9%) and fresh wood (3.0%). Humus, rocks and live tree bases occupied 1-2% the forest floor, while mineral soil and moss covered less than 1%. In selection cuts, proportions of rotten wood and live tree bases were lower in skid trails than outside, and this difference persisted 13 years after harvest. In 1- and 2-year-old cuts, the proportion of litter was lower in skid trails than outside, but not later on. The proportions of mineral soil and disturbed humus increased sharply in the skid trails after harvest (17.1-fold and 2.7-fold increases, respectively), but these effects lasted only 3 years for mineral soil and 1 year for humus. Power functions were used to model variation in litter, mineral soil and disturbed humus proportions as a function of time since harvest. We discuss the implications for regeneration dynamics of the marked, but short term increase in mineral soil and disturbed humus availability. Spatially explicit models that can simulate the mid- to long-term availability of substrates in both time and space would be useful for assessing the long-term implications of harvesting on seedbed proportions and regeneration patterns.