Complex interactions between the zebra mussel, Dreissena polymorpha, and the harmful phytoplankter, Microcystis aeruginosa

We report a reversal in the sign of the herbivore–phytoplankton interaction between the zebra mussel (Dreissena polymorpha) and Microcystis aeruginosa, a harmful planktonic cyanobacterium. A pair of large-scale manipulations of mussel density in the same lake in consecutive years showed that when phosphorus concentrations were very low (mean total phosphorus [TP] ; 3 mg L21), the effect of Dreissena on the biomass of M. aeruginosa was monotonically negative across the full range of sustainable mussel densities. When the enclosures were fertilized with phosphorus (mean TP ; 9 mg L21), there was a monotonically positive effect of Dreissena on M. aeruginosa across the same mussel gradient. These contrasting results indicate that D. polymorpha feeds on M. aeruginosa, as shown in previous laboratory feeding experiments, but that the positive effects of D. polymorpha on M. aeruginosa can be larger than the negative effects of consumption. A sign reversal in the interaction between these two species is congruent with highly variable patterns in the response of M. aeruginosa to D. polymorpha invasion across lake and river systems in North America. A well-established paradigm in limnology holds that the taxonomic composition of summer phytoplankton assemblages shifts with phosphorus enrichment toward greater dominance by cyanobacteria (Smith 1986; Trimbee and Pre1 Corresponding author ([email protected]). Acknowledgments We thank C. Scheele, J. Chiotti, A. Schuerer, E. Milroy, B. Hanna, E. McConnell, L. Bartner, T. Toda, and M. Machavaram for assistance in the field and laboratory, N. Consolatti for logistical support, T. Coon and D. Hayes for lending equipment, S. Hu and A. Tessier for providing pre-Dreissena phytoplankton data for Gull Lake and cultured Ankistrodesmus, and S. Peacor and C. Klausmeier for critical comments. Funding was provided by the Michigan Sea Grant College Program, project R/NIS-4, under grant NA76RG0133 from the Office of Sea Grant, National Oceanic and Atmospheric Administration (NOAA), U.S. Department of Commerce, the State of Michigan, the Kalamazoo Community Foundation, and the National Science Foundation (DEB-9701714, DBI-9602252). This paper is contribution 1168 of the W. K. Kellogg Biological Station. pas 1987; Watson et al. 1997). It is also widely accepted that harmful species of cyanobacteria (temperate species within the genera Anabaena, Aphanizomenon, Microcystis, and Oscillatoria) are more likely to attain bloom densities in lakes and rivers that are nutrient rich (Paerl 1988). Given this background, recent reports of increases in harmful cyanobacteria (specifically, the toxin-producing colonial species Microcystis aeruginosa) in coastal areas of the Great Lakes (Vanderploeg et al. 2001; Nicholls et al. 2002) are surprising, given the success of phosphorus abatement programs in these habitats (Bierman et al. 1984; Bertram 1993; Vanderploeg et al. 2001; Nicholls et al. 2002). Increases in M. aeruginosa are coincident with invasion of the Great Lakes by the zebra mussel (Dreissena polymorpha), a filter-feeding bivalve (Vanderploeg et al. 2002). A recent survey of inland lakes (Raikow et al. 2004) shows a strong positive association between Dreissena invasion and the relative abundance of M. aeruginosa, but only in lakes with relatively low nutrients (total phosphorus [TP] 5 10–25 mg L21). That study also reported no relationship between TP and cyanobacterial dominance in invaded lakes, in contrast to the positive re897 Zebra mussels and Microcystis aeruginosa Fig. 1. Relative abundance (as percentage of total phytoplankton biomass) of Microcystis aeruginosa in Gull Lake before and after zebra mussel invasion. Zebra mussels were first observed in Gull Lake in 1994. For each year, one to three samples collected during the first half of August were analyzed. lationship observed in lakes that lack the invader (Trimbee and Prepas 1987; Watson et al. 1997; Downing et al. 2001). Given the radical implications of these recent observations with respect to the general understanding of the drivers of harmful cyanobacteria, there is a critical need for experimental field tests of the hypothesis that zebra mussels have positive effects on the biomass of M. aeruginosa. Despite significant interest in the consequences of Dreissena invasion in North America (Nalepa and Fahnenstiel 1995; MacIsaac 1996; Caraco et al. 1997), there is a relative paucity of reports regarding field manipulations of Dreissena density (Heath et al. 1995; Thayer et al. 1997; Stewart et al. 1999; Jack and Thorp 2000; Wilson 2003). None of these field experiments provide direct information about the response of M. aeruginosa to manipulations of Dreissena. Observations of phytoplankton species composition before and after zebra mussel invasion suggest that the Dreissena–M. aeruginosa interaction may be complex, in the sense that Dreissena may have opposite effects in different habitats. For example, the biomass of M. aeruginosa in the Bay of Quinte in Lake Ontario increased dramatically after Dreissena invasion (Nicholls et al. 2002), whereas the biomass of M. aeruginosa in the Hudson River declined dramatically after invasion (Smith et al. 1998). The results of laboratory feeding experiments are similarly conflicting, with mussels having relatively high and low selectivities for different clones of M. aeruginosa within and across studies (Vanderploeg et al. 2001). M. aeruginosa forms colonies that vary greatly in size, and some strains produce intracellular toxins that may act as herbivore deterrents, so vulnerability to herbivory may be highly variable within this species (Vanderploeg et al. 2001). In this paper, we demonstrate both negative and positive effects of Dreissena on the biomass of M. aeruginosa in a pair of large-scale experiments conducted in the same lake in consecutive years. Logistical, legal, and ethical constraints limited us to conducting mussel-removal manipulations in an already-invaded lake, rather than mussel-addition manipulations in an uninvaded lake. Consequently, we assumed at the outset that the effects of mussel invasion on M. aeruginosa could be reversed (by reducing mussel density) within the restricted time frame of our enclosure experiments.