1. Variable retention harvests or 'green-tree retention systems' are important to the management of coniferous forests in the temperate zone. Green-tree retention leaves large live trees after harvest (i.e. residual trees) to increase structural diversity of the regenerating stand and provide mature forest habitat that develops sooner than in typical even-aged management by clear-cutting. The seed-tree system is one method of harvesting that leaves a few wind-firm seed trees standing singly, or in groups, to provide seed to regenerate an area naturally. Green-tree retention may increase biological diversity and help to enhance functional links among forest structures or ecological processes. 2. This study was designed to test the hypothesis that the abundance (e.g. basal area and density of residual trees and amount of vegetation) and diversity (e.g. species diversity and structural diversity of the herb, shrub and tree layers) of various aspects of stand structure will decline with lower levels of tree retention. In particular, abundance and diversity should be greater in sites with green-tree retention than in clear-cut sites. 3. Stand structure attributes were measured from 1996 (immediately after harvesting) to 1999 in replicated clear-cut, single seed-tree, group seed-tree, patch-cut and uncut forest sites in mixed forests of Douglas fir Pseudotsuga menziesii-lodgepole pine Pinus contorta in southern British Columbia, Canada. 4. In terms of abundance, clear-cut, single seed-tree and group seed-tree sites had similar mean basal areas of residual trees, ranging from 0.1 to 2.5 m(2) ha(-1), with significantly higher levels on patch-cut (23.4 m(2) ha(-1)) and uncut forest (39.0 m(2) ha(-1)) sites. Mean densities of residual trees also followed this pattern, ranging from 0.7 to 16.3 stems ha(-1) on the clear-cut to group seed-tree sites, to 769.4 and 2050.0 stems ha(-1) on the patch-cut and uncut forest sites, respectively. However, mean volume (m(3) ha(-1)) of fallen wood (i.e. dead wood or woody debris) was similar among sites, ranging from 116.7 in the single seed-tree to 210.2 in the patch-cut sites. Contrary to our hypothesis, the mean index of total crown volume of herbs, shrubs, mosses and lichens was similar among sites. 5. Mean species richness of herbs, shrubs and total plants was similar among sites. Mean richness of trees in seed-tree sites was similar to that in uncut forest in three of four post-harvest years. The mean species diversity of herbs was similar among sites, but that of shrubs and trees was generally lowest in the patch-cut sites and similar among the other sites. 6. The mean structural richness and diversity of herbs was lowest in the uncut forest and similar at the other sites 3 and 4 years post-harvest. The mean structural richness of trees and total plants in the patch-cut and uncut forest sites had more vegetation layers than the other sites. 7. Our hypothesis was partly supported in terms of basal area and density of residual trees and structural richness of total vegetation layers across a gradient of tree retention from uncut forest to clear-cut sites. However, this pattern was not supported for abundance or diversity of understorey vegetation. Thus, the structural attributes of residual trees and layers of vegetation, but not the abundance, species richness or species diversity of understorey vegetation, lend support to the aims of variable retention harvesting, at least up to 4 years post-harvest.
