Multi-scale strategies for the monitoring of freshwater cyanobacteria: reducing the sources of uncertainty.

Cyanobacterial blooms are a frequent phenomenon in eutrophic freshwaters worldwide and are considered potential hazards to ecosystems and human health. Monitoring strategies based on conventional sampling often fail to cover the marked spatial and temporal variations in cyanobacterial distribution and fluctuating toxin concentrations inherent to cyanobacterial blooms. To deal with these problems, we employed a multi-scale approach for the study of a massive Microcystis bloom in Tajo River (Spain) utilizing 1) remote sensing techniques, 2) conventional water sampling and 3) analysis of chemotypical subpopulations. Tajo River at the study area is influenced by high temperatures waters diverted upstream from a nuclear power plant, the presence of a dam downstream and a high nutrient load, which provide optimal conditions for massive cyanobacterial proliferation. MERIS imagery revealed high Chl-a concentrations that rarely fell below 20 μg L(-1) and moderate spatiotemporal variations throughout the study period (March-November 2009). Although the phytoplanktonic community was generally dominated by Microcystis, sampling points highly differed in cyanobacterial abundance and community composition. Microcystin (MC) concentrations were highly heterogeneous, varying up to 3 orders of magnitude among sampling points, exceeding in some cases WHO guideline values for drinking and also for recreational waters. The analysis of single colonies by MALDI-TOF MS revealed differences in the proportion of MC-producing colonies among points. The proportion of toxic colonies showed a highly significant linear correlation with total MC: biovolume ratio (r(2) = 0.9; p < 0.001), evidencing that the variability in toxin concentrations can be efficiently addressed by simple analysis of subpopulations. We propose implementing a multi-scale monitoring strategy that allows covering the spatiotemporal heterogeneities in both cyanobacterial distribution (remote sensing) and MC concentrations (subpopulation analysis) and thereby reduce the main sources of uncertainty in the assessment of the risks associated to bloom events.