A novel approach to assessing natural resource injury with Bayesian networks

Freya Elizabeth Rowland; Christopher James Kotalik; Bruce G. Marcot; Jo Ellen Hinck; David M. Walters

doi:10.1002/ieam.4836

Home/Archive/USGS

USGS · 70249413

A novel approach to assessing natural resource injury with Bayesian networks

Freya Elizabeth Rowland·Christopher James Kotalik·Bruce G. Marcot·Jo Ellen Hinck·David M. Walters

View original source ↗Export metadata ↓

Abstract

Quantifying the effects of environmental stressors on natural resources is problematic because of complex interactions among environmental factors that influence endpoints of interest. This complexity, coupled with data limitations, propagates uncertainty that can make it difficult to causally associate specific environmental stressors with injury endpoints. The Natural Resource Damage Assessment and Restoration (NRDAR) regulations under the Comprehensive Environmental Response, Compensation, and Liability Act and Oil Pollution Act aim to restore natural resources injured by oil spills and hazardous substances released into the environment; exploration of alternative statistical methods to evaluate effects could help address NRDAR legal claims. Bayesian networks (BNs) are statistical tools that can be used to estimate the influence and interrelatedness of abiotic and biotic environmental variables on environmental endpoints of interest. We investigated the application of a BN for injury assessment using a hypothetical case study by simulating data of acid mine drainage (AMD) affecting a fictional stream-dwelling bird species. We compared the BN-generated probability estimates for injury with a more traditional approach using toxicity thresholds for water and sediment chemistry. Bayesian networks offered several distinct advantages over traditional approaches, including formalizing the use of expert knowledge, probabilistic estimates of injury using intermediate direct and indirect effects, and the incorporation of a more nuanced and ecologically relevant representation of effects. Given the potential that BNs have for natural resource injury assessment, more research and field-based application are needed to determine their efficacy in NRDAR. We expect the resulting methods will be of interest to many US federal, state, and tribal programs devoted to the evaluation, mitigation, remediation, and/or restoration of natural resources injured by releases or spills of contaminants

Explore related subjects

Integrated Environmental Assessment and Management Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Freya Elizabeth Rowland, Christopher James Kotalik, Bruce G. Marcot, Jo Ellen Hinck, David M. Walters. 2023-09-04. A novel approach to assessing natural resource injury with Bayesian networks. https://doi.org/10.1002/ieam.4836

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2026

Composite samples for perfluorooctanesulfonate in fish may change contaminant threshold decisions

The widespread occurrence and persistence of the per- and polyfluoroalkyl substances (PFAS) in the environment may pose significant health risks to humans. Fish consumption is a direct pathway for exposure, prompting extensive monitoring. In the United States and Europe, fish consumption advisories for PFAS generally focus on perfluorooctanesulfonate (PFOS), the most frequent and abundant PFAS compound reported in fish. However, PFAS analysis is costly, leading monitoring programs to frequently use composite samples (i.e., combining different fish of the same species). This financial saving comes at an analytical cost of lost information on individual variation. It is unclear how this may influence threshold-based decisions. To test this, we used a modeling approach (i.e., Monte Carlo) based on aggregated data distributions from freshwater fish species. We simulated the measurement of PFOS for individual fish compared to composite means (three, five, and 10 fish) at different levels of error structure. These values were then compared to a typical do-not-eat threshold of 200 ng•g wet mass −1 and four metrics were estimated: (i) the probability of detecting a PFOS concentration above the threshold, (ii) the maximum simulated PFOS concentration, (iii) the number of false negative samples, and (iv) the rate of false negative samples. These results show that the probability of assigning PFOS levels above or below the threshold is shifted by masking individual value contributions in a composite sample. Because true individual contaminant loads are variable, composite sampling is a more conservative metric for contaminant thresholds. The model suggested that the greater the variability, the lower the detection probability for truly contaminated samples. Our study may help to inform sampling for monitoring programs aimed at managing human exposure to PFAS.

Integrated Environmental Assessment and Management↗

USGS2025

Comparative toxicity of aircraft deicers, runway deicers, and road salt in winter airport runoff

Aircraft deicer and anti-icer formulations and airfield pavement deicer formulations are commonly used at airports that experience freezing precipitation. In addition, road salt is often applied in urban areas surrounding airports. This mixture of deicers in receiving streams can be toxic to aquatic organisms. A 17-year study of deicers and toxicity in runoff samples from sites in and near Milwaukee Mitchell International Airport included bioassay testing of Ceriodaphnia dubia , Pimephales promelas , Raphidocelis subcapitata , and Aliivibrio fischeri concurrently with quantification of freezing-point depressants that represent presence of the various deicers. Propylene glycol was monitored to represent aircraft deicers and anti-icers, acetate and formate were monitored to represent airfield pavement deicers, and chloride was monitored to represent road salt. Toxicity endpoints (e.g., lethal concentrations and inhibitive concentrations) were compiled for 19 airport deicer products and sodium chloride. Comparison of freezing-point depressant concentrations with toxicity endpoints indicated that propylene glycol, acetate, and chloride exceeded endpoint concentrations in multiple samples: more than 70% of 492 runoff samples exceeded at least one endpoint concentration with exceedances of propylene glycol and acetate-based deicer endpoints as well as road salt endpoints. Validation of these results were observed with measured toxic effects on C. dubia , P. promelas , and R. subcapitata in bioassays conducted on undiluted samples: samples with multiple deicers exceeding endpoint thresholds had greater toxicity than samples with individual deicer endpoint exceedances, and samples with no endpoint exceedances typically exhibited less toxicity than samples with only one endpoint exceedance. A comparison of additivity of toxicity quotients with bioassay results fit a log-logistic regression curve, indicating that the condition of tested organisms degraded with increasing deicer presence. These findings suggest that concurrent reduction of runoff from aircraft, airfield pavement, and road salt deicers will be needed to eliminate aquatic toxicity in adjacent watersheds.

Wisconsin↗

USGS2025

Metal-rich lacustrine sediments from legacy mining perpetuate copper exposure to aquatic-riparian food webs

Historic copper mining left a legacy of metal-rich tailings resulting in ecological impacts along and within Torch Lake, an area of concern in the Keweenaw Peninsula, Michigan, USA. Given the toxicity of copper to invertebrates, this study assessed the influence of this legacy on present day nearshore aquatic and terrestrial ecosystems. We measured the metal (Co, Cu, Ni, Zn, Cd) and metalloid (As) concentrations in sediment, pore water, surface water, larval and adult insects, and two riparian spider taxa collected from Torch Lake and a nearby reference lake. Overall, elevated metal and metalloid concentrations, particularly Cu, were measured in all sediment samples and some surface and pore water samples collected from Torch Lake. For instance, Cu concentrations in the Torch Lake sediment were ∼200% higher than the reference lake and all measured concentrations exceeded predicted effects concentrations by at least ninefold. Within larval insect tissues, we observed 160% higher Cu concentrations than measured in the reference lake, and Cu was the only measured element above predicted effects concentrations in Torch Lake. Adult insects collected at both lakes had similar metal concentrations irrespective of exposure levels. Yet we found 100% higher copper concentrations in Torch Lake riparian spiders, demonstrating elevated exposure risk to insectivores across the aquatic-terrestrial boundary. Our results highlight that other metals in the mixture may not be as concerning to adjacent riparian ecosystems, but copper remains a contaminant of concern in Torch Lake 60 years after mining ceased.

Michigan↗