Turbulent mixing and phytoplankton spring bloom development in a deep lake

A one-dimensional (1-D) mechanistic phytoplankton model combined with a 1-D hydrodynamic model was applied to simulate phytoplankton growth during winter and spring in deep monomictic Upper Lake Constance. Modeled chlorophyll a concentrations agree well with data from the 8-y time period considered. In particular, the interannual variation in the timing of phytoplankton growth is adequately simulated by the model. The onset of phytoplankton blooms in Upper Lake Constance is not sensitive to variations in the photosynthetically active radiation, the sinking velocity of the algae, or the effect of water temperature on biological process rates, but is primarily determined by turbulent diffusion (i.e., by the transition from strong mixing in winter and early spring to weak mixing). The transition in mixing conditions and thus also the beginning of phytoplankton population growth correlates with the build up of the first slight temperature stratification. Simulations performed without consideration of phytoplankton loss due to grazing overestimate algal biomass soon after the onset of algal growth. Including grazing by zooplankton substantially improves the agreement between model and data and suggests that ciliate grazing in particular leads to a significant reduction of phytoplankton abundance in spring after the start of the algal bloom. The mixing period in oceans and deep lakes can be considered to be the aquatic analogue of barren land (Reynolds 1997). During the winter mixing period primary productivity is limited by low temperature, the seasonal minimum of solar radiation, the maximum of its reflection at the water surface, and by turbulent mixing that transports phytoplankton below the euphotic zone. Only with increasing solar radiation, temperature, and thermal stratification does the phytoplankton bloom start to develop (Bleiker and Schanz 1997; Gaedke et al. 1998; Tian et al. 2003b) and forms the basis of further seasonal succession (e.g., Sommer et al. 1986), and the development of plankton trophic structure and food webs (Straile 2005). Understanding and predicting the onset and development of the phytoplankton bloom is hence of fundamental importance in predicting the response of aquatic ecosystems to environmental change. The start of the phytoplankton bloom in temperate regions is associated with the onset of stratification, increased solar radiation, and a warming of surface waters. Especially in deep water bodies, it is the absence of deepwater mixing that strongly determines light availability for phytoplankton (Tian et al. 2003b). Consequently, plankton blooms can develop immediately after the onset of stratification (Bleiker and Schanz 1997; Gaedke et al. 1998; Tian et al. 2003b) or in the absence of stratification when wind-driven vertical mixing is absent or very weak (Townsend et al. 1992; Bleiker and Schanz 1997). In more shallow systems, the increase in light availability due to increasing hours of daylight (Tian et al. 2003a) or the melting of the ice cover (Adrian et al. 1999) results into the plankton bloom. The aim of this study is to investigate to what extent the onset and early development of the phytoplankton spring bloom depends on seasonal changes in turbulent mixing, solar radiation, water temperature, and grazing. The analysis is based on a mechanistic model that is applied to the conditions in Upper Lake Constance. Two different modeling approaches are commonly employed to increase understanding of aquatic systems. On the one hand, the response of phytoplankton development is investigated on the basis of minimal models (Gragnani et al. 1999; Huisman et al. 2002, Huppert et al. 2002). The aim of such studies is, for example, to identify the range of turbulent diffusivities and sinking rates that allow phytoplankton populations to survive (e.g., Huisman et al. 2002), to characterize the nutrient regimes resulting in phytoplankton blooms (Huppert et al. 2002), or to investigate the competition between phytoplankton species in a dynamic environment where the phytoplankton concentration affects underwater light intensity (e.g., Huisman et al. 2004). These minimal models are not designed for a direct comparison with field data but are intended to provide principal insights into ecological mechanisms. On the other hand, complex ecosystem models are used to simulate plankton, nutrients, and other water constituents under field conditions. The aim of these studies is to provide models that enable a prognosis of the plankton development under changing environmental conditions due to climate change or anthropogenic impacts. The model design is usually validated by a direct comparison of Acknowledgments We thank H. Rossknecht (Institut für Seenforschung, Langenargen) for the long-term temperature data set, all scientists and technicians involved in the collection and compilation of the biological data set, and Kristine Schalau for the preparation of the thermistor data. The data set at Sta. BM was established within the Special Collaborative Programme (SFB) 248 ‘‘Cycling of Matter in Lake Constance,’’ supported by Deutsche Forschungsgemeinschaft (DFG). We also thank two anonymous referees for their constructive comments which helped to improve the manuscript. This research was conducted within the program AQUASHIFT (SPP 1162) funded by the DFG. Additional support was by the European Union within the framework of the European Commission project ‘CLIME’ (EVK1-CT-2002-00121). Limnol. Oceanogr., 52(1), 2007, 286–298 E 2007, by the American Society of Limnology and Oceanography, Inc.