Top-down control from the bottom: Regulation of eutrophication in a large river by benthic grazing

We use a 15-yr record from a 100-km stretch of the tidal, freshwater Hudson River to examine the controls of phytoplankton biomass. Across years, seasonal mean chlorophyll a (Chl a) and maximum Chl a varied by approximately 15-fold, with mean growing-season Chl a as high as 24 mg L21 and maximum Chl a as high as 120 mg L21. For 3 of the 15 yr the river would be classified as eutrophic on the basis of lake standards for Chl a. Year-toyear variation in Chl a was not closely related to either nutrients (phosphorus or nitrogen concentrations) or hydrologic flow. Annual variation in grazing by the invasive zebra mussel (Dreissena polymorpha) explained 90% of the variation in mean Chl a; however, the maximum Chl a reached during the growing season was not significantly related to grazing and, even with high grazing, blooms of phytoplankton that contained high proportions of potentially toxic cyanobacteria occurred. Further, zebra mussels decreased dissolved oxygen concentrations even while increasing production of submersed aquatic vegetation. The results from the Hudson add to a growing literature that suggests that ecosystem changes linked with high phytoplankton biomass depend on a diverse range of system characteristics as well as whether phytoplankton are controlled by top-down or bottom-up mechanisms. For lakes, reservoirs, and estuaries it is well established that high N and P inputs frequently result in high phytoplankton biomass and to problems associated with this high biomass (Schindler 1978; Smith 2003). However, it is now recognized that resistance to eutrophication can vary greatly between systems (Cloern 2001). As a group, large rivers, which tend to have high N and P loads from both point and nonpoint sources, may show strong resistance to both high phytoplankton biomass and many of the associated consequences of high phytoplankton biomass. In large rivers biomass of phytoplankton can potentially be kept low by high advective losses associated with short residence times (Soballe and Kimmel 1987; Wetzel 2001) and strong light limitation resulting from the combination of a well-mixed water column and turbid conditions (Cole et al. 1992). Additionally, the well-mixed conditions may make river phytoplankton more susceptible to control by benthic grazers than are phytoplankton in stratified systems (Sullivan et al. 1991). Finally, because of turbulent conditions in most rivers, even bottom waters exchange relatively rapidly with the atmosphere (Caraco et al. 2000) and, thus, low oxygen conditions resulting from phytoplankton decomposition are less likely. The potential resistance that rivers show to eutrophication could explain why there has been comparatively little research on the controls of phytoplankton biomass or consequences of high phytoplankton biomass in riverine systems (Smith 2003). At times, however, phytoplankton biomass can reach values well above 15 mg L21 chlorophyll a (Chl a) (Basu and Pick 1996; Welker and Walz 1998; Smith 2003),