Background/Question/Methods Twenty-three closed-canopy, upland forests on the national forest in Wisconsin were inventoried for compositional and structural components. One objective of this monitoring effort {plots originally established in 1991} was to discover and rank features and interactions that influence understory cover , richness, and diversity.  Understory composition was determined by use of quadrats and a “walk through” to add species not found in the plots.   All forests are located on dry-mesic to mesic sites and were widely scattered.  The structural features that were tested included overstory basal area, midstory basal area, volume of coarse woody debris, snag density, shrub frequency, maple sapling density, percent hemlock in the overstory and percent conifer overstory.  Graphing and correlation were the initial phase of analysis.   When these suggested a possible connection, stepwise regression was employed to formally test and rank the potential predictor variables

Results/Conclusions Unexpectedly, not a single structural variable had any relation to understory richness or diversity (p > .25).   In contrast, both overstory and midstory basal area  were significantly and negatively related to understory cover.  The model had an adjusted R-square of .474 and was highly significant (p = .001); standardized coefficients indicate roughly equal influence of the two variables.  Previously work suggested  that deer browsing had a positive influence on both cover and richness, whereas exotic earthworm biomass had a negative impact on richness (p = .05) but not cover (p > .5). Ordination suggested moderately strong separation among forest types, but these differences have  waned over  the past 17 years.  Collectively, these results indicate  that biotic interactions have a much stronger effect than structure. Furthermore, these trends suggest that the regional species pool is  converging and most species are not dispersal limited.