Herbivory has played a major role in dictating vegetation abundance and species composition at Kingman Marsh in Anacostia Park, Washington, D.C., since restoration of this tidal freshwater wetland was initiated in 2000. The diverse and robust vegetative cover that developed in the first year post-reconstruction experienced significant decimation in the second year, after the protective fencing was removed, and remained suppressed throughout the five-year study period. In June 2009 a herbivory study was initiated to document the impacts of herbivory by resident and nonmigratory Canada geese (Branta canadensis) to vegetation at Kingman Marsh. Sixteen modules consisting of paired fenced plots and unfenced control plots were constructed. Eight of the modules were installed in vegetated portions of the restoration site that had been protected over time by pre-existing fencing, while the remaining eight modules were placed in portions of the site that had not been protected over time and were basically unvegetated at the start of the experiment. Exclosure fencing was sufficiently elevated from the substrate level to allow access to other herbivores such as fish and turtles, while hopefully excluding mature Canada geese. The study was designed with an initial exclosure elevation of 20 cm. This elevation was chosen based on the literature, as adequate to exclude mature Canada geese, while maximizing access to other herbivores such as fish and turtles.

Repeated measures analysis of variance (ANOVA) was used to analyze the differences between paired fenced and unfenced control plots for a number of variables including total vegetative cover. Differences in total vegetative cover were not statistically significant for the baseline data collected in June 2009. By contrast, two months after the old protective fencing was removed from the initially-vegetated areas to allow Canada geese access to the unfenced control plots, total vegetative cover had declined dramatically in the initially-vegetated unfenced control plots, and differences between paired fenced and unfenced control plots were statistically significant. These differences have remained steady and significant throughout the remainder of these first three years of the study.

Total vegetative cover has followed a somewhat different path in the initially-unvegetated modules, where cover in the fenced plots did not significantly exceed cover in the unfenced control plots until the August 2010 sampling event. In spite of the slow start in the initially-unvegetated modules, differences between paired fenced plots and unfenced control plots have remained significant and even increased significantly over time. This indicates that total vegetative cover in the initially-unvegetated fenced plots and unfenced control plots is continuing to diverge over time as vegetation increases in the protected plots compared to the basically unvegetated unfenced control plots.

Total vegetative cover has been composed almost entirely of native species during the first three years of the study, with cover by exotics averaging less than 1% during each sampling event.

Species richness did not differ significantly between fenced plots and unfenced control plots during 2009, the first year of the study. Since August 2010, species richness has remained significantly greater in the fenced plots than in the unfenced control plots. These differences have remained relatively steady over time for both the initially-vegetated and initially unvegetated modules.

During the study it became apparent that our elevated fence plots were more accessible to mature geese than we had expected. Even after lowering the exclosure fencing to 15 cm in 2010 and 10 cm in 2011, we documented geese inside exclosures in both years. Nonetheless the data indicate that even at 10 cm, we have limited the numbers of mature geese entering the fenced plots, rather than totally preventing their access through low spots in the uneven substrate surface. At an exclosure elevation of 10 cm and with a soft, mucky substrate, we are assuming that non-goose herbivores such as fish and turtles still have free access to the fenced plots. Annual wildrice (Zizania aquatica), known from previous studies to be especially palatable to Canada geese, has seen the greatest impact from partial access to the fenced plots by mature geese, moving from an overwhelming dominant in the initially-vegetated plots to a minor presence there by August 2011. Interestingly, pickerelweed (Pontederia cordata), also known to be highly palatable to Canada geese, has so far shown only minor herbivory in the fenced plots. By August 2011, pickerelweed had actually increased to significantly greater cover levels in the fenced plots compared to the unfenced control plots.

In conclusion, the first three years of data document that vegetation exposed to full herbivory by resident and nonmigratory Canada geese for three years in the unfenced control plots showed significantly lower total vegetative cover and species richness compared to the vegetation in the fenced plots, which experienced reduced herbivory by resident and nonmigratory Canada geese. These effects were documented for modules located in both initially-vegetated and initially-unvegetated habitats.