Cyanobacterial blooms are expected to increase in occurrence and intensity as a consequence of future climate change, as well as human-induced eutrophication, leading to adverse effects on public health and local economics. In order to efficiently mitigate cyanobacterial blooms towards protecting public health, it is very important to understand the factors that promote cyanobacterial blooms, as well as their potential health impact. This study explores the effect of two potential promoting factors, land use and zebra mussel invasion, on cyanobacterial bloom intensity using data from the 2007 National Lake Assessment (NLA). First of all, the effect of invasive zebra mussels on cyanobacterial community structure and microcystins levels was investigated, using large-scale data from the USEPA National Lake Assessment and the U.S. Geological Survey Aquatic Nuisance Species database. ANOVA Based on Dissimilarities tests (Adonis) showed that there was a significant difference in cyanobacterial communities between lakes located in areas with and without established zebra mussel populations. Lakes located in areas with zebra mussels also had significantly higher microcystins levels and cyanobacteria abundance, but lower concentrations of phosphorous. Structural equation modeling suggested that the total effect of zebra mussel establishment resulted in an overall 1.40-fold increase in microcystins levels, which presumably resulted from three contributing factors: 1) a 0.86-fold decrease in microcystins levels through total phosphorus decrease, 2) a 1.06-fold increase through an increased cyanobacteria abundance, and 3) a 1.53-fold increase through other ways, such as an increase in cyanobacteria toxicity. Secondly, another promoting factor, land use of the lake basin, was investigated. A spatial modeling approach was used to explore relationships among watershed land-use, nutrient concentrations, cyanobacterial abundance, and microcystins production with the NLA dataset. As expected, nitrogen and phosphorus concentrations were found to be lower (p<0.05) in lakes surrounded by undeveloped watersheds (e.g., forested) as opposed to those dominated by agriculture. Logistic regression analysis revealed that the presence of both cyanobacteria and Microcystis was unrelated to in-lake total nitrogen and phosphorus concentrations. However, for lakes where cyanobacteria or Microcystis were present, their abundances were positively related to both in-lake total nitrogen and phosphorus concentrations. This study suggests that efforts to minimize land-use activities, which seemingly lead to nutrient runoff, could help reduce in-lake nutrient concentrations, as well as the prevalence of toxic cyanobacteria.
Then, we investigated the potential effect of cyanobacterial blooms on nonalcoholic liver disease mortality using satellite remote sensing data. A Bayesian spatial regression, implementing a negative binomial model, was used to analyze the relationship between cyanobacterial bloom coverage and death from nonalcoholic liver disease. Risk of death from nonalcoholic liver disease increased by 0.3% (95% Bayesian confidence interval 0.1% to 0.5%) with each 1 % increase in bloom coverage in the affected county after adjusting for age, gender, educational level, and race. The significant association between cyanobacterial blooms and nonalcoholic liver disease in this study suggested a population level health impact from cyanobacterial blooms.
In the last part of the study, two MODIS-based indicators were investigated for their performance in monitoring cyanobacteria abundance and toxin levels at two drinking water plant intakes located in Lake Erie during 2013. Good correlations were observed between toxic cyanobacteria, microcystins, and MODIS-retrieved bloom indicators for the intake in the western lake (where blooms were much more serious), but not for the central lake intake where blooms were milder. The microcystins levels showed a Spearman’s correlation of 0.815 (p<0.05) with MODIS- retrieved chlorophyll-a at the Toledo intake point in western Lake Erie. Both Microcystis abundance and toxic Microcystis abundance also showed a significant positive correlation with MODIS-retrieve chlorophyll-a for the Toledo plant, as well as the two locations combined. These results demonstrate the potential for satellite remote sensing, at a regional scale, to contribute to monitoring of cyanobacterial blooms, as a preliminary warning, to protect human health in bloom-impacted water bodies.