Assessing the biotic and abiotic factors influencing the distribution and intensity of cyanobacterial harmful algal blooms in Dog Lake, South Frontenac using UAV imaging and eDNA

Abstract

Cyanobacterial harmful algal blooms (CHABs) deteriorate water quality and cause widespread economic and ecological damage to aquatic ecosystems. Efforts in mitigating and monitoring CHABs face challenges from their complex predictors, dynamic spatial and temporal distributions, and scope. Traditional monitoring techniques do not have the resolution, coverage, or frequency to characterize small scale CHABs. Novel approaches with unmanned aerial vehicles (UAV) surveillance and environmental DNA (eDNA) sampling allow us to examine CHAB distribution, intensity, and potential toxicity. We conducted in situ field experiments in Dog Lake, South Frontenac, Ontario, Canada. In February and March of 2019, we collected lacustrine eDNA samples from the top 0.5 m and bottom 0.5 m of sites with lakebed depths ranging from 1-19 m, as well as from ice cover. We used qPCR to quantify the concentration of the Microcystis aeruginosa mcyE and 16S rRNA genes. Our results indicated overwintering M. aeruginosa showed depth preference, with high concentrations at lake surface in deeper areas of the lake. Shallow areas of the lake had higher eDNA concentrations overall. Ice cover was a potential source of spring inoculum. In May to October, we conducted UAV surveys and collected localized weather data, water variables, eDNA, and toxin samples from Milburn Bay, Dog Lake. We used a cyanobacterial signal enhancement algorithm to quantify the area and distribution of CHABs. We created a multiple linear regression model, with CHAB coverage as the independent variable, and various topographic, weather, and biotic variables as predictors after model selection for the strongest predictor time periods with best subsets selection. Wind reduced CHAB particle size and coverage. 16S rRNA eDNA concentration correlated with microcystin concentrations. Overall, we concluded that the timing and duration of weather variables had a notable impact on the predictive value of environmental variables on CHABs at a local scale. UAV surveys had a cost of approximately $50 per mission, while eDNA sampling had a cost of approximately $5 per sample. We demonstrated that RGB-UAV surveying in tandem with eDNA sampling are a powerful, novel approach to CHAB monitoring.