Background/Question/Methods   Cyanobacteria form a natural component of aquatic communities, but can occasionally exhibit large blooms leading to unsightly surface scum, foul smells and harmful cyanotoxins. Ecologists and managers are interested in understanding the conditions promoting blooms, as blooms can significantly change ecological dynamics within lakes and affect recreational use. Many cyanobacteria species are capable of fixing atmospheric nitrogen, making them good competitors in low nitrogen environments. Thus, resource competition between cyanobacteria and other non-fixing phytoplankton for nitrogen (N) and phosphorus (P) is considered a major factor underlying cyanobacteria blooms. Previous studies have debated the relative importance of absolute levels of resources (N and P), and their ratio (N:P) in driving cyanobacteria abundance. Other hypotheses focus on cyanobacteria grazing resistance, buoyancy, light competition, and temperature tolerance. We test hypotheses regarding the influence of N, P and N:P, as well as abiotic conditions in driving cyanobacteria blooms. Adopting a mixture model approach for zero inflated data, we examine factors related to cyanobacteria presence/absence as well as abundance. Our work analyzes these hypotheses on a large scale, using data on phytoplankton identity and abundance, water chemistry and physical properties from 543 lakes, surveyed in the US EPA National Eutrophication Survey (1973).

Results/Conclusions   While resource-ratio theory predicts that cyanobacteria should be competitively excluded in high N and low P conditions, we find that nutrient levels are generally poor predictors of cyanobacteria presence/absence. Instead, cyanobacteria presence seems to be strongly influenced by positive effects of temperature, time of year and latitude. These patterns are consistent with widely observed seasonal succession dynamics, and suggest that factors such as temperature and grazing resistance play a larger role than nutrient levels in determining cyanobacteria presence. Where present, however, cyanobacteria abundance is significantly explained by total N levels and to a lesser extent N:P ratios. Making use of re-sampling techniques, we show that patterns in earlier studies that clearly suggested an important role of N:P in cyanobacteria presence/absence may have arisen from using data sets containing disproportionately few lakes with high N:P levels. Overall, we conclude that predicting cyanobacteria blooms requires studying the abiotic factors governing where cyanobacteria can occur at all, as well as understanding the resource conditions that enable them to flourish in the face of competition.