Author: Cal D Buelo,  Ph.D. Candidate (University of Virginia)

Designation: 

Description:
Algal blooms in aquatic ecosystems represent regime shifts between clear-water and algae dominated states. Theory and modeling studies suggest ecosystem statistical properties can provide early warning of approaching blooms in temporal and spatial measurements of key state variables. Previous field tests of these predictions have been successful but focused on either time series or spatial data, not both, and focused on systems known to be approaching a bloom. We added nutrients to a lake to induce an algal bloom, while simultaneously monitoring an adjacent, unmanipulated reference lake. We collected high frequency temporal data on bloom variables from stationary rafts, as well as high resolution spatial data with a mobile sensor platform, to directly compare temporal and spatial indicators. Fertilization caused a large bloom, followed by a crash and a smaller bloom. Temporal early warning statistics (EWS) performed better than analogous spatial EWS. Prior to the bloom, rolling window standard deviation increased and generated alarms in multiple variables. In contrast, spatial EWS were highly variable between sampling dates with a high degree of overlap between experimental and reference lakes. Analyzing data from this and past experiments both with and without nutrient additions, we find that temporal EWS have higher rates of true positives when approaching a bloom than false positives when a bloom is not imminent. This study suggests temporal EWS may provide reliable warning of algal blooms consistent with theory and potentially useful to management.

Category: Scientific Program Abstract > Special Sessions > SS60 Contextualizing abrupt change using big environmental data at freshwater and marine ecosystems