Prototypic automated continuous recreational water quality monitoring of nine Chicago beaches.

Predictive empirical modeling is used in many locations worldwide as a rapid, alternative recreational water quality management tool to eliminate delayed notifications associated with traditional fecal indicator bacteria (FIB) culturing (referred to as the persistence model, PM) and to prevent errors in releasing swimming advisories. The goal of this study was to develop a fully automated water quality management system for multiple beaches using predictive empirical models (EM) and state-of-the-art technology. Many recent EMs rely on samples or data collected manually, which adds to analysis time and increases the burden to the beach manager. In this study, data from water quality buoys and weather stations were transmitted through cellular telemetry to a web hosting service. An executable program simultaneously retrieved and aggregated data for regression equations and calculated EM results each morning at 9:30 AM; results were transferred through RSS feed to a website, mapped to each beach, and received by the lifeguards to be posted at the beach. Models were initially developed for five beaches, but by the third year, 21 beaches were managed using refined and validated modeling systems. The adjusted R(2) of the regressions relating Escherichia coli to hydrometeorological variables for the EMs were greater than those for the PMs, and ranged from 0.220 to 0.390 (2011) and 0.103 to 0.381 (2012). Validation results in 2013 revealed reduced predictive capabilities; however, three of the originally modeled beaches showed improvement in 2013 compared to 2012. The EMs generally showed higher accuracy and specificity than those of the PMs, and sensitivity was low for both approaches. In 2012 EM accuracy was 70-97%; specificity, 71-100%; and sensitivity, 0-64% and in 2013 accuracy was 68-97%; specificity, 73-100%; and sensitivity 0-36%. Factors that may have affected model capabilities include instrument malfunction, non-point source inputs, and sparse calibration data. The modeling system developed is the most extensive, fully-automated system for recreational water quality developed to date. Key insights for refining and improving large-scale empirical models for beach management have been developed through this multi-year effort.

Interaction of bacterial communities and indicators of water quality in shoreline sand, sediment, and water of Lake Michigan.

Shoreline sand harbors high concentrations of fecal indicator bacteria (FIB) that may be resuspended into the water column through washing and resuspension. Studies have explored coastal processes that influence this sand-water flux for FIB, but little is known about how microbial markers of contamination or the bacterial community interact in the sand-water interface. In this study, we take a three-tiered approach to explore the relationship between bacteria in sand, sediment, and overlying water at three shoreline sites and two associated rivers along an extended freshwater shoreline. Samples were collected over two years and analyzed for FIB, two microbial source tracking (MST) markers (Catellicoccus marimammalium, Gull2; Bacteroides HF183), and targeted metagenomic 16S rRNA gene analysis. FIB was much higher in sand than in water at all three sites. Gull2 marker was abundant in shoreline sand and water while HF183 marker was mostly present in rivers. Overall bacterial communities were dissimilar between sand/sediment and water, indicating little interaction. Sediment composition was generally unfavorable to bacterial resuspension. Results show that FIB and MST markers were effective estimates of short-term conditions at these locations, and bacterial communities in sand and sediment reflected longer-term conditions. Findings are useful for locating contamination sources and targeting restoration by evaluating scope of shoreline degradation.

Environmental DNA (eDNA): A tool for quantifying the abundant but elusive round goby (Neogobius melanostomus).

Environmental DNA (eDNA) is revolutionizing biodiversity monitoring, occupancy estimates, and real-time detections of invasive species. In the Great Lakes, the round goby (Neogobius melanostomus), an invasive benthic fish from the Black Sea, has spread to encompass all five lakes and many tributaries, outcompeting or consuming native species; however, estimates of round goby abundance are confounded by behavior and habitat preference, which impact reliable methods for estimating their population. By integrating eDNA into round goby monitoring, improved estimates of biomass may be obtainable. We conducted mesocosm experiments to estimate rates of goby DNA shedding and decay. Further, we compared eDNA with several methods of traditional field sampling to compare its use as an alternative/complementary monitoring method. Environmental DNA decay was comparable to other fish species, and first-order decay was lower at 12°C (k = 0.043) than at 19°C (k = 0.058). Round goby eDNA was routinely detected in known invaded sites of Lake Michigan and its tributaries (range log10 4.8-6.2 CN/L), but not upstream of an artificial fish barrier. Traditional techniques (mark-recapture, seining, trapping) in Lakes Michigan and Huron resulted in fewer, more variable detections than eDNA, but trapping and eDNA were correlated (Pearson R = 0.87). Additional field testing will help correlate round goby abundance with eDNA, providing insight on its role as a prey fish and its impact on food webs.