Interannual variation in diatom bloom dynamics: roles of hydrology, nutrient limitation, sinking, and whole lake manipulation.

Spring development of diatoms in Ford Lake, Michigan, USA was markedly different in 2004 from 2005 and 2006. In 2004, diatom biovolume surpassed 15 mm(3)l(-1) but in 2005 and 2006 maximum biovolume was less than 5 mm(3)l(-1). Soluble reactive silica (SRSi) in 2004 fell below 5 microM whereas in 2005 and 2006, SRSi remained above 30 microM. Taxonomic composition was similar among years and consisted mainly of Asterionella, Cyclotella, Fragilaria, Aulacoseira, and Synedra. Bioassay experiments in 2005 demonstrated that P rather than Si was the element most limiting biomass development. However, P supply rate did not account for the differences among years. Model simulations of Si uptake, washout rates, and sinking implicated hydrologic differences among years as the cause of differential success by diatom populations in April of each year. Bioassay experiments performed after overturn demonstrated that diatoms could grow well in unamended lake water, but they did not flourish in the lake; model simulations implicated sinking losses as the reason. In summer 2006, we performed a selective withdrawal of hypolimnetic water from the outlet dam and weakened density stratification. An Aulacoseira bloom resulted in early to mid-August, depleting SRSi to less than 30 microM. The lake, which had been acting as a P source, changed to a P sink during the bloom, and cyanobacteria did not develop as they had in all previous years. Stoichiometric calculations indicate that the net SRSi uptake and the net DP uptake during the induced bloom were consistent with diatom production.