Throughout eastern North America, white-tailed deer herbivory is a key disturbance that can be expressed in terms of periodicity, severity, extent, intensity. Exclosure studies manipulate this disturbance, allowing researchers to determine the extent to which deer alter plant community composition and habitat structure. Here, I take advantage of a 20-year study of deer exclusion in a hemlock-northern hardwood stand in N Wisconsin. I test two hypotheses: the difference in species composition and percent cover between browsed and unbrowsed plots differs radically, and year-to-year variability in species composition and percent cover is greater in browsed plots. I sampled species composition annually beginning in 2006 in adjacent control-exclosure plots along permanent line-intercept transects. In each year I calculated the Bray-Curtis distance among all pairwise combination of sites, with species weighted by percent cover. I used ANOSIM to analyze multivariate differences between exclosure and control plots, and PERMDISP to test for differences in interannual variability between exclosure and control plots.

Results/Conclusions

ANOSIM reveals distinct community types: a low biomass community dominated by sedges and grasses, and a high biomass community dominated by broadleaved herbs and woody species (ANOSIM R = 0.95; P < 0.001). Within-plot composition varied slightly from year to year, with neither treatment nor control exhibiting greater interannual variability (PERMDISP; P = 0.56). Distinct community types persisted in each year of sampling (ANOSIM R = 1.0). Graminoid-dominated control plots probably reflects the emergence of a grazing lawn in this forest understory environment, with browsing creating a positive feedback loop conferring a competitive advantage to graminoids. The elimination of deer browsing from exclosure plots resulted in community composition divergence that might be an example of alternate stable states in a grazing system.