Drone-based water sampling and characterization of three freshwater harmful algal blooms in the United States

Abstract

Freshwater harmful algal blooms (HABs), caused mostly by toxic cyanobacteria, produce a range of cyanotoxins that threaten the health of humans and domestic animals. Climate conditions and anthropogenic influences such as agricultural run-off can alter the onset and intensity of HABs. Little is known about the distribution and spread of freshwater HABs. Current sampling protocols in some lakes involve teams of researchers that collect samples by hand from a boat and/or from the shoreline. Water samples can be collected from the surface, from discrete-depth collections, and/or from depth-integrated intervals. These collections are often restricted to certain months of the year, and generally are only performed at a limited number of collection sites. In lakes with active HABs, surface samples are generally sufficient for HAB water quality assessments. We used a unique DrOne Water Sampling SystEm (DOWSE) to collect water samples from the surface of three different HABs in Ohio (Grand Lake St Marys, GLSM and Lake Erie) and Virginia (Lake Anna), United States in 2019. The DOWSE consisted of a 3D-printed sampling device tethered to a drone (uncrewed aerial system, or UAS), and was used to collect surface water samples at different distances (10–100 m) from the shore or from an anchored boat. One hundred and eighty water samples (40 at GLSM, 20 at Lake Erie, and 120 at Lake Anna) were collected and analyzed from 18 drone flights. Our methods included testing for cyanotoxins, phycocyanin, and nutrients from surface water samples. Mean concentrations of microcystins (MCs) in drone water samples were 15.00, 1.92, and 0.02 ppb for GLSM, Lake Erie, and Lake Anna, respectively. Lake Anna had low levels of anatoxin in nearly all (111/120) of the drone water samples. Mean concentrations of phycocyanin in drone water samples were 687, 38, and 62 ppb for GLSM, Lake Erie, and Lake Anna, respectively. High levels of total phosphorus were observed in the drone water samples from GLSM (mean of 0.34 mg/L) and Lake Erie (mean of 0.12 mg/L). Lake Anna had the highest variability of total phosphorus with concentrations that ranged from 0.01 mg/L to 0.21 mg/L, with a mean of 0.06 mg/L. Nitrate levels varied greatly across sites, inverse with bloom biomass, ranging from below detection to 3.64 mg/L, with highest mean values in Lake Erie followed by GLSM and Lake Anna, respectively. Drones offer a rapid, targeted collection of water samples from virtually anywhere on a lake with an active HAB without the need for a boat which can disturb the surrounding water. Drones are, however, limited in their ability to operate during inclement weather such as rain and heavy winds. Collectively, our results highlight numerous opportunities for drone-based water sampling technologies to track, predict, and respond to HABs in the future.