An organizing challenge in the study of food webs is to understand how perturbations to one trophic level extend to other levels.  On Yellowstone’s Northern Range, riparian willows (Salix sp.) have been heavily browsed by elk (Cervus elaphus), leading to an alternative state in the plant community where willows are rare and small in stature.  Wolf reintroduction in 1995 has likely reduced elk numbers and may have changed their foraging behavior, leading to reduced herbivory on willows. However, whether a reduction in herbivory is sufficent to allow for willow recovery remains unclear because many Northern Range streams are incised and lack the populations of beaver that occupied them in the early 1900s. These changes in the landscape may stabilize the alternative state.  In 2001, we began a manipulative experiment to investigate how herbivory and water table decline influence willow growth at four sites in the Northern Range.  Initial results from the first five years of this experiment suggested that water table depths may constrain the potential for willow recovery even when herbivory is eliminated.  Here we build on those analyses with additional data.  We explored willow height dynamics using Bayesian hierarchical models to describe changes in plant height.  We used these models to describe individual willow height dynamics, and to evaluate the importance of covariates.

Results/Conclusions

We show that continuous predictors such as water table mean and variance through the growing season as well as overwinter browsing intensity explain individual plant height gains through time.  Annual growth rates of willows are shaped by interactions between plant height and browsing intensity as well as interactions between water table depth and browsing intensity.  In addition to linear effects, the best models included random effects explaining site-level and year-to-year variation.   These results suggest that willow recovery is inhibited not only by water table depth, but also by high variability in the water table throughout the growing season.   The effect of temporal variance provides additional evidence of the positive effect of beaver dams on willow growth dynamics.  Our models also indicate that willow height recovery in areas adjacent to incised streams will be slow even if browsing pressure is severely reduced.  This work provides additional support to our earlier findings that relaxed herbivory resulting from reduced predation pressure by wolves on elk does not lead to a rapid state change in willow communities in Yellowstone National Park.