Trophic Cascades and Compensation: Differential Responses of Microzooplankton in Whole-Lake Experiments

Food webs in three lake basins were manipulated by altering fish communities to either reduce or increase the abundance o f Daphnia. These basins were subsequently fertilized with nitrogen and phosphorus for two years. W e tested three predictions about the response o f heterotrophic flagellates, ciliates, and rotifers (collectively, microzooplankton) derived from prior studies. W e predicted that ( 1 ) microzooplankton would increase with lake fertilization, ( 2 ) lakes with abundant Daphnia would have lower increases in microzooplankton, and ( 3 ) both increases in resource availability and suppression by Daphnia would determine microzooplankton dynamics. Contrary to the first prediction, microzooplankton did not increase with fertilization relative to the reference lake, except in the low-Daphnia system. The second prediction was supported, as Daphnia prevented microzooplankton from increasing in the fertilized lakes with the strength o f the Daphnia ef fect being greater than anticipated. Because o f this strong ef fect , microzooplankton dynamics were in all but one case most strongly related to suppression by Daphnia rather than to a combined ef fect o f resources and suppression. The microzooplankton communities were differentially affected by the trophic cascade. Heterotrophic flagellates appeared to be limited by a variety o f predators. Even in the low-Daphnia fertilized lake, mortality was apparently high. Ciliates and rotifers increased in the low-Daphnia fertilized lake and were strongly suppressed otherwise. These experiments indicate that small-scale, short-term experiments and larger-scale comparative analyses may be inadequate for assessing the strength o f trophic interactions. The potential for community-level responses, not well assessed except at the ecosystem scale, may alternatively dampen or enhance the impacts o f trophic cascades in food webs. Key +turds: ciliates; Daphnia; ecosystem experiments; jagellates; lakes; rotqers; trophic cas cades; zooplankton. I N T R O D U C T I O N cades originating from piscivorous and planktivorous Predicting the outcome o f food web interactions is fishes determine the size structure o f zooplankton comcomplicated in at least two ways. First, behavioral, munities, which may in turn influence phytoplankton physiological, and morphometric responses o f individbiomass (Carpenter et al. 1985). uals often act to ameliorate predation, thus limiting the Compensatory responses may also dampen the action impact o f one trophic group upon another (Sih 1987, o f trophic cascades. For example, in studies o f lake Polis et al. 1996). In addition, changes in food webs food webs, simple predictions about lower trophic levmay ramify through trophic networks, creating comels are not always upheld when fish populations are plex dynamics among the trophically linked populaexperimentally manipulated, because o f changes in the tions (Abrams et al. 1996). These individual and popintensity o f vertebrate and invertebrate predation on ulation responses can engender indeterminate and often zooplankton as well as changes within phytoplankton surprising results (Brown et al. 1986, Yodzis 1988). communities (e.g., Carpenter and Kitchell 1993). While it is possible to quantify the direct and indirect Strong effects o f zooplankton on phytoplankton hinge effects o f food web perturbations on populations exmost often on the presence or absence o f large species perimentally (Morin et al. 1988, Wootton 1994) or with o f the cladoceran Daphnia (Leibold 1989, Carpenter modeling approaches (Pimm 1991), the generality o f and Kitchell 1993). When large Daphnia are abundant, these responses for specific types o f communities is phytoplankton are often suppressed relative to their usually unknown. Nevertheless, strong food web inabundance in the absence o f a large grazer (Pace 1984, teractions can also result in dramatic and predictable Mazumder 1994, Carpenter et al. 1996). These intertrophic-level responses (Paine 1980). Such interactions actions, however, occur within a food web o f multiple have been well documented in lakes where trophic caspathways dictated by the trophic diversity o f microorganisms as well as interactions involving other zooManuscript received 16 September 1996; revised 7 march plankton and fish (Porter 1996). Included within these 1997; final version received 7 April 1997. pelagic food webs are flagellated and ciliated protozoa 139 Ja~iuary1998 TROPHIC CASCADES AND COMPENSATION and rotifers, which are both abundant and important constituents of the plankton community in terms of secondary production and nutrient cycling (Stockner and Porter 1988). These organisms, referred to collectively here as microzooplankton, can also be simultaneously potential competitors and prey of Daphnia. Previous studies have documented the importance of system productivity and predators as regulators of the abundance, biomass, and composition of microzooplankton communities. Heterotrophic flagellate, ciliate, and rotifer abundances increase across gradients of primary production, and simple regression models predict these changes (Pace 1986, Sanders et al. 1992, Gasol and Kalff 1995). Zooplankton predators also limit microzooplankton abundance and may regulate their dynamics (Pace and Funke 1991, Riemann and Christoffersen 1993, Burns and Schallenberg 1996). Daphnia kill, consume, and compete for food with microzooplankton, with the largest species having the greatest effect (Gilbert 1988a, b, Wickham and Gilbert 1991, 1993, Jurgens 1994). Microzooplankton populations are typically reduced to very low densities in enclosure experiments with high Daphnia populations (Pace and Funke 1991, Christoffersen et al. 1993, Jurgens et al. 1994, Marchessault and Mazumder, in press). Microzooplankton mortality is much higher In lakes with species of Daphnia of mean adult body lengths of 1 mm, -10 mg dry mass (Pace and VaquC 1994). Negative correlations have also been observed between the abundance of Daphnia and heterotrophic flagellates within and among lakes (Gude 1988,Weisse et al. 1990, Gasol and VaquC 1993).As for phytoplankton, Daphrzin appears to be a key genus that when present may strongly limit the abundance and biomass of microzooplankton (Pace and Funke 1991, Jurgens 1994).Lake trophic cascades arising from piscivorous fish should suppress microzooplankton by enhancing Daphnia. There is also the possibility, however, for compensation, as microzooplankton species resistant to predation may come to dominate communities when Daphnia is abundant. For example, some rotifers are less susceptible to Daphnia because of size, elongated spines, sturdy loricas, or rapid escape responses (Gilbert 1988a, b, Jack and Gilbert 1993). Understanding of the interactions of microzooplankton, Daphnia, and lake trophic conditions is based on experimental studies in enclosures over relatively short periods of time (days to weeks), or alternatively, on among-lake comparisons. There is uncertainty about whether the results from experimental and comparative studies extrapolate to predict the results of food web changes in lakes. For example, enclosure studies may exaggerate Daphnia effects because populations increase to unusually high densities when fish predation is eliminated (Pace and Cole 1994).Further, short-term experiments may not allow sufficient time for compensatory mechanisms that ameliorate predatory effects. Comparative studies, while describing general patterns of variation, may provide only weak predictions about changes within individual systems. Food web disturbances of whole lakes bridge the uncertainties associated with comparative and enclosure studies, allowing tests at key scales of interest-lakes and years (Carpenter and Kitchell 1993). Two common disturbances to lake food webs arise from increases in nutrient loading and shifts in fish communities that alter trophic cascades. Here, we explore the joint effects of nutrient loading and food web structure on microzooplankton in whole-lake experiments carried out over four years. Fish manipulations were designed to either suppress or enhance the abundance of Daphnia (Carpenter et al. 1996). Lakes with and without Daphnia were then fertilized during the final two years of the experiment. We ask here if we can predict at least qualitatively the net responses of microzooplankton to the combined manipulations by examining three general predictions. First, microzooplankton should increase with fertilization based on the comparative studies documenting increased abundances across gradients from oligotrophic to eutrophic systems. Second, increases should be reduced in fertilized lakes with sustained high Daphnia populations based on experimental results and the comparative study of heterotrophic flagellates by Gasol and VaquC (1993). The literature is not specific, however, about what abundance of Daphnia constitutes a "high" population. The third prediction concerns the relative importance of resources vs. suppression by Daphnia in determining abundance. The literature cited above suggests that both factors are important, and we test this prediction by fitting time series of microzooplankton dynamics with models that include measures of resources and Daphnia. Study sites and lake experimental manipulations Experiments were carried out in Paul, Peter, and Long lakes, located at the University of Notre Dame Environmental Research Center in Gogebic County, Michigan. Whole-lake manipulations and results for phytoplankton, zooplankton, and bacteria have been reported elsewhere (Carpenter et al. 1996, Pace and Cole 1996) and are only briefly summarized here. All three lakes are relatively small (<5 ha), steepsided basins that stratify strongly during the May-September period considered in this study. Long Lake was divided in the spring of 1991 into three basins using two neoprene curtains drawn top to bottom, across narrow portions of the lake. We studied two of these basins and hereafter refer to them as East and West Long lakes, for convenience. While these basins were chosen for their similarity in physical, chemical, and biological conditions, they diverged during the course of the experiment and responded quite differently to the manipulations as described below. We therefore consider 140 MICHAEL L. PACE ET AL. Ecology, Vol. 79, No. I TABLE1. Sampling frequency and depths for measurements of the abundance of heterotrophic flagellates, ciliates, and rotifers in the experimental lakes. Years Lake Group Depth PL, PR H. flagellates Epi, Meta EL, WL All All H. flagellates H. flagellates Ciliates EDI pi, Meta Epi PL, PR Rotifers E$ PL, PR Rotifers Epi, Meta EL, WL Rotifers Epi All Rotifers Epi, Meta Notes: PL = Paul Lake, PR = Peter Lake, EL = East Long Lake, and WL = West Long Lake; Epi meta = metalimnion, as defined in method.^. "Biweekly" means every other week. Frequency Weekly Weekly Biweekly Weekly Weekly Weekly Weekly Weekly