It has been hypothesized that partial retention of the overstory can mediate loss and accelerate recovery of forest-dependent species following logging.  We studied how level and spatial pattern of retention affected responses of vascular and non-vascular plants to harvest of mature coniferous forests in the Pacific Northwest.  The DEMO experiment employs a randomized block design replicated at five locations in western Oregon and Washington.  Each block includes a control (100% retention) and four treatments that comprise a factorial design with two levels of retention (40 vs. 15% of original basal area) and two spatial patterns (trees aggregated in 1-ha patches vs. dispersed).  We addressed three questions:  How do level and pattern of retention affect composition and diversity of forest understory plants?  Do forest aggregates serve as refugia for species sensitive to disturbance or environmental stress?  Are bryophyte responses to retention influenced by substrate type/quality?  Permanent plots systematically distributed within experimental units and across forest-aggregate boundaries were used to examine vascular plant responses 6-7 yr after harvest.  Bryophyte abundance and richness were sampled 8 yr after harvest on three substrates — forest floor, coarse woody debris (CWD), and tree bases — across a gradient of dispersed retention (15 to 100%).

Results/Conclusions

Both early- and late-seral herbs showed significant responses to level of retention, with larger changes at 15 than at 40% retention.  In contrast, pattern of retention had minimal effect on treatment-scale responses.  In aggregated units, small changes within forest patches were balanced by large changes in adjacent harvest areas.  Thus, on average, changes were comparable to those in dispersed treatments.  Effects on community composition were consistent with these trends:  controls and forest aggregates showed little variation over time; dispersed treatments, moderate variation; and harvested areas of aggregated treatments, the largest changes in species composition.  Forest aggregates serve as refugia for most shade-tolerant herbs.  Only one of 29 common species showed significant declines within the aggregates.  This contrasts with early responses to harvest:  in year two 35% of species exhibited edge-related declines.  At lower levels of dispersed retention, bryophytes showed significantly greater declines on CWD than on the forest floor, suggesting an indirect effect of overstory removal on substrate quality.  Our results suggest that a combination of aggregates and dispersed trees — at levels greater than current minimum standards — may be a viable strategy for conserving the diversity and abundance of species associated with late-seral forests.