Predicting wastewater treatment plant influent in mixed, separate, and combined sewers using nearby surface water discharge for better wastewater-based epidemiology sampling design

For wastewater sample collection approaches supporting public health applications, few high hydrologic activity normalizing guidelines currently consider readily available environmental flow data that may earlier capture information regarding periods of influent mixing and dilution of wastewater with groundwater and runoff. This study aimed to identify wastewater sampling rules for high hydrological activity events, allowing for an earlier decision point in the control of dilution before sample collection. We defined the sampling rules via data-driven models (Random Forest and linear regression) using environmental data (i.e., wastewater treatment facility influent rates, nearby stream discharge flow, and precipitation). These models were applied to five treatment plants in Jefferson County, Kentucky (USA) in mixed, separate, and combined sewers with different population sizes. We proposed cutoffs of 10 %, 25 %, and 50 % flow conditions for orientation towards public health samples. The results showed a strong nonlinear relationship between nearby stream discharge and treatment facility flow rates, which was used to infer the hydrological conditions that produce high volumes of diluted wastewater in the sewer system. Accumulated Local Effects and SHapley Additive exPlanations aided in deciphering the relationship between the predictors and response variables of the Random Forest models. The influent rate to the treatment plant from the previous day and two USGS stream gages were needed to adequately predict the degree of infiltration and inflow mixing on a given day. Surface water discharge data can be used to provide an earlier workflow decision point during wet weather periods to improve understanding of flow conditions for wastewater-based epidemiological studies to inform laboratory analysis and data interpretation. Not only total flow, but also the specific proportions of infiltration and inflow to wastewater volume in influent should be considered when analyzing data for normalization purposes, and our method provides a starting point for doing so rapidly and at low cost.

NABat: A top-down, bottom-up solution to collaborative continental-scale monitoring

Collaborative monitoring over broad scales and levels of ecological organization can inform conservation efforts necessary to address the contemporary biodiversity crisis. An important challenge to collaborative monitoring is motivating local engagement with enough buy-in from stakeholders while providing adequate top-down direction for scientific rigor, quality control, and coordination. Collaborative monitoring must reconcile this inherent tension between top-down control and bottom-up engagement. Highly mobile and cryptic taxa, such as bats, present a particularly acute challenge. Given their scale of movement, complex life histories, and rapidly expanding threats, understanding population trends of bats requires coordinated broad-scale collaborative monitoring. The North American Bat Monitoring Program (NABat) reconciles top-down, bottom-up tension with a hierarchical master sample survey design, integrated data analysis, dynamic data curation, regional monitoring hubs, and knowledge delivery through web-based infrastructure. NABat supports collaborative monitoring across spatial and organizational scales and the full annual lifecycle of bats.

Sampling microfibres at the sea surface: The effects of mesh size, sample volume and water depth

Microfibres are one of the most ubiquitous particulate pollutants, occurring in all environmental compartments. They are often assumed to be microplastics, but include natural as well as synthetic textile fibres and are perhaps best treated as a separate class of pollutants given the challenges they pose in terms of identification and contamination. Microfibres have been largely ignored by traditional methods used to sample floating microplastics at sea, which use 300-500 μm mesh nets that are too coarse to sample most textile fibres. There is thus a need for a consistent set of methods for sampling microfibres in seawater. We processed bulk water samples through 0.7-63 μm filters to collect microfibres in three ocean basins. Fibre density increased as mesh size decreased: 20 μm mesh sampled 41% more fibres than 63 μm, and 0.7 μm filters sampled 44% more fibres than 25 μm mesh, but mesh size (20-63 μm) had little effect on the size of fibres retained. Fibre density decreased with sample volume when processed through larger mesh filters, presumably because more fibres were flushed through the filters. Microfibres averaged 2.5 times more abundant at the sea surface than in water sampled 5 m sub-surface. However, the data were noisy; counts of replicate 10-L samples had low repeatability (0.15-0.36; CV = 56%), suggesting that single samples provide only a rough estimate of microfibre abundance. We propose that sampling for microfibres should use a combination of <1 μm and 20-25 μm filters and process multiple samples to offset high within-site variability in microfibre densities.

An inconvenient dataset: bias and inappropriate inference with the multilevel model

The multilevel model has become a staple of social research. I textually and formally explicate sample design features that, I contend, are required for unbiased estimation of macro-level multilevel model parameters and the use of tools for statistical inference, such as standard errors. After detailing the limited and conflicting guidance on sample design in the multilevel model didactic literature, illustrative nationally-representative datasets and published examples that violate the posited requirements are identified. Because the didactic literature is either silent on sample design requirements or in disagreement with the constraints posited here, two Monte Carlo simulations are conducted to clarify the issues. The results indicate that bias follows use of samples that fail to satisfy the requirements outlined; notably, the bias is poorly-behaved, such that estimates provide neither upper nor lower bounds for the population parameter. Further, hypothesis tests are unjustified. Thus, published multilevel model analyses using many workhorse datasets, including NELS, AdHealth, NLSY, GSS, PSID, and SIPP, often unwittingly convey substantive results and theoretical conclusions that lack foundation. Future research using the multilevel model should be limited to cases that satisfy the sample requirements described.