Interactive effects of flooding and white-tailed deer herbivory on tree seedling recruitment in floodplain forests of the Upper Mississippi River

Background/Question/Methods

Floodplain forests are important components of aquatic-terrestrial transition zones, providing habitat, moderating nutrient fluxes, and stabilizing river corridors. Along the Upper Mississippi River (UMR), floodplain forests have been lost and degraded as a result of a modified hydrologic regime and restoration efforts have met numerous challenges. Some studies indicate that in addition to flooding, white-tailed deer (Odocoileus virginianus) browsing can severely reduce tree growth rates in forest restorations, to the point that reestablishment of a functional forest is questionable. Yet, little is known about the effects of herbivores on tree growth in highly productive and periodically inundated floodplain forest restoration sites. To investigate the effects of flooding and herbivory on tree seedling recruitment, we coupled a tightly controlled study of the interactive effects of flooding and herbivory on tree growth and survival with a regional survey of deer browsing in newly restored floodplain forests along the UMR. In the controlled experiment, we established exclosures along a flood inundation gradient and measured browse intensity, tree height, and mortality. In the regional survey, we measured browsing intensity and plant preference as they related to tree density.

Results/Conclusions

In the controlled experiment, tree height declined in response to percent of stems browsed annually (P<0.05), but deer browsing did not influence tree survival (P>0.05). Percent of the growing season in which water levels were higher than mean tree height decreased tree growth increment (P<0.05), and tree mortality increased with flood duration (P<0.05). Overall, browsing appears to limit tree height, which in turn affects an individual’s ability to survive flooded conditions. Regionally, browsing by deer increased with tree density (P<0.05). Browsing levels ranged from <1% to 67% and averaged 30% of all available stems. Isolated restoration sites typically had lower browse intensities than sites surrounded by forest. Certain tree species such as American elm, shagbark hickory, and silver maple were preferentially browsed, suggesting the potential for long-term herbivore effects on forest composition.  Our results suggest that floodplain forest restoration in the UMR may be limited by both flooding and deer herbivory on a regional basis.