Hierarchical Bayesian Approach To Reduce Uncertainty in the Aquatic Effect Assessment of Realistic Chemical Mixtures

Species in the aquatic environment differ in their toxicological sensitivity to the various chemicals they encounter. In aquatic risk assessment, this interspecies variation is often quantified via species sensitivity distributions. Because the information available for the characterization of these distributions is typically limited, optimal use of information is essential to reduce uncertainty involved in the assessment. In the present study, we show that the credibility intervals on the estimated potentially affected fraction of species after exposure to a mixture of chemicals at environmentally relevant surface water concentrations can be extremely wide if a classical approach is followed, in which each chemical in the mixture is considered in isolation. As an alternative, we propose a hierarchical Bayesian approach, in which knowledge on the toxicity of chemicals other than those assessed is incorporated. A case study with a mixture of 13 pharmaceuticals demonstrates that this hierarchical approach results in more realistic estimations of the potentially affected fraction, as a result of reduced uncertainty in species sensitivity distributions for data-poor chemicals.

Assessing predictive uncertainty in comparative toxicity potentials of triazoles

Comparative toxicity potentials (CTPs) quantify the potential ecotoxicological impacts of chemicals per unit of emission. They are the product of a substance's environmental fate, exposure, and hazardous concentration. When empirical data are lacking, substance properties can be predicted. The goal of the present study was to assess the influence of predictive uncertainty in substance property predictions on the CTPs of triazoles. Physicochemical and toxic properties were predicted with quantitative structure-activity relationships (QSARs), and uncertainty in the predictions was quantified with use of the data underlying the QSARs. Degradation half-lives were based on a probability distribution representing experimental half-lives of triazoles. Uncertainty related to the species' sample size that was present in the prediction of the hazardous aquatic concentration was also included. All parameter uncertainties were treated as probability distributions, and propagated by Monte Carlo simulations. The 90% confidence interval of the CTPs typically spanned nearly 4 orders of magnitude. The CTP uncertainty was mainly determined by uncertainty in soil sorption and soil degradation rates, together with the small number of species sampled. In contrast, uncertainty in species-specific toxicity predictions contributed relatively little. The findings imply that the reliability of CTP predictions for the chemicals studied can be improved particularly by including experimental data for soil sorption and soil degradation, and by developing toxicity QSARs for more species.

Do interspecies correlation estimations increase the reliability of toxicity estimates for wildlife?

For warm-blooded species, the hazardous dose of a chemical (HD50) is an upcoming and important characteristic in the assessment of toxic chemicals. Generally, experimental information is available for a limited number of warm-blooded species only, which causes statistical uncertainty. Furthermore, when small datasets contain an unrepresentative sample of species, they can cause systematic uncertainty in chemicals' hazardous doses. The number of species can be enlarged with interspecies correlation estimation (ICE) models, but these are uncertain themselves. The goal of this study is to quantify the possible gain in reliability of the HD50 values for warm-blooded wildlife species after enlargement of the sample size with ICE predictions. For 1137 chemicals, we compared systematic uncertainty and statistical uncertainty between HD50 values based on experimental data (HD50(Ex)) and on datasets combining experimental data and ICE predictions (HD50(Co)). HD50(Ex) values ranged between 1.0×10(-1) and 9.5×10(3)mgkg(wwt)(-1), and HD50(Co) values between 1.1×10(0) and 6.1×10(3)mgkg(wwt)(-1). For over 97 percent of the chemicals, HD50(Ex) values exceeded HD50(Co) values, with a systematic uncertainty (i.e. the ratio of HD50(Ex)/HD50(Co)) of typically 3.5. The limited availability of experimental toxicity data, predominantly for mammals, resulted in a systematic underestimation of the wildlife toxicity of a chemical. Statistical uncertainty factors (i.e. the ratio of the 95th/5th percentile) quantified the statistical uncertainty in the HD50 values. The statistical uncertainty factors ranged between 1.0×10(0) and 2.5×10(22) for the experimental dataset, and between 4.8×10(0) and 1.1×10(2) for the combined dataset. For all sample sizes, median statistical uncertainty factors were the largest for combined datasets. However, combining experimental toxicity data with ICE predictions makes it possible to reduce the upper limit of the range for statistical uncertainty factors. We conclude that, by combining experimental data with ICE model predictions, the validity of the HD50 value can be improved and high statistical uncertainty can be reduced, particularly in cases of limited toxicity data, i.e. data for mammals only or a sample size of n≤4.

Parameter uncertainty in modeling bioaccumulation factors of fish

We quantified the uncertainty due to biota-related parameters in estimated bioaccumulation factors (BAFs) of persistent organic pollutants for fish through Monte Carlo simulations. For this purpose, the bioaccumulation model OMEGA (Optimal Modeling for EcotoxicoloGical Applications) was parameterized based on data from the existing literature, analysis of allometric data, and maximum likelihood estimation. Lipid contents, fractions of food assimilated, the allometric rate exponent, normalized food intakes, respiration and growth dilution rates, and partial mass transfer resistances in water and lipid layers were included as uncertain parameters. The uncertainty in partial resistances was particularly important in the estimation of the rate constants for chemical intake from water by fish. Uncertainties in the fractions of food assimilated and partial water layer resistances from and to food were particularly important in the estimation of the rate constants of chemical intake from food. The uncertainty in the model outcomes for the bioaccumulation factors for fish was a factor of 10 (ratio of 95th and fifth percentile estimates), which was mainly caused by the uncertainty in the lipid fraction. For chemicals with a K(OW) of 10(3) to 10(6), the uncertainty in the lipid contents of fish accounted for more than 50% of the uncertainty in the estimated bioaccumulation factor. For chemicals with a high K(OW) (10(7) and higher), the fractions of food assimilated and partial resistances also contributed to uncertainty in the estimated bioaccumulation factor (up to 60%). A case study showed that uncertainty in estimated BAF for nonpersistent substances can be dominated by uncertainty in the rate constants for metabolic transformation.

Cumulative energy demand as predictor for the environmental burden of commodity production

Cumulative energy demand has been used as a methodology to assess life cycle environmental impacts of commodity production since the early seventies, but has also been criticized because it focuses on energy only. During the past 30 years there has been much research into the development of more complex single-score life cycle impact assessment methodologies. However, a comprehensive analysis of potential similarities and differences between these methodologies and cumulative energy demand has not been carried out so far. Here we compare the cumulative energy demand of 498 commodities with the results of six frequently applied environmental life cycle impact assessment methodologies. Commodity groups included are metals, glass, paper and cardboard, organic and inorganic chemicals, agricultural products, construction materials, and plastics. We show that all impact assessment methods investigated often provide converging results, in spite of the different philosophies behind these methodologies. Fossil energy use is identified by all methodologies as the most important driver of environmental burden of the majority of the commodities included,with the main exception of agricultural products. We conclude that a wide range of life cycle environmental assessment methodologies point into the same environmental direction for the production of many commodities.

The impact of an additional ecotoxicity test on ecological quality standards

The present study aims to support decisions on whether or not to perform an extra toxicity test in order to improve environmental quality standards (EQSs). The impact of an additional ecotoxicity test was analyzed by predicting new ecotoxicity values with three different estimation methods and adding them to existing species sensitivity distributions (SSDs) on which the EQSs are based. The results show that EQSs are likely to increase due to increasing sample size, but the change also depends on the number of toxicity values available, the estimation method used and the representativeness of the species tested. The management consequences are illustrated in a case study on contaminated freshwater sediment in the Netherlands. It is shown that a slight increase of the EQS can result in a large reduction of sediment remediation costs without impairing regulatory protection levels. The paper identifies indicators that can be used to evaluate the potential impact of an extra ecotoxicity test.

Separation of uncertainty and interindividual variability in human exposure modeling

The NORMTOX model predicts the lifetime-averaged exposure to contaminants through multiple environmental media, that is, food, air, soil, drinking and surface water. The model was developed to test the coherence of Dutch environmental quality objectives (EQOs). A set of EQOs is called coherent if simultaneous exposure to different environmental media that are all polluted up to their respective EQOs does not result in exceeding the acceptable or tolerable daily intake (ADI or TDI). Aim of the present study is to separate the impact of uncertainty and interindividual variability in coherence predictions with the NORMTOX model. The method is illustrated in a case study for chlorfenvinphos, mercury and nitrate. First, ANOVA was used to calculate interindividual variability in input parameters. Second, nested Monte Carlo simulation was used to propagate uncertainty and interindividual variability separately. Lifetime-averaged exposure to chlorfenvinphos, mercury and nitrate was modeled for the Dutch population. Output distributions specified the population fraction at risk, due to a particular exposure, and the reliability of this risk. From the case study, it was obtained that at lifelong exposure to all media polluted up to their standard, 100% of the Dutch population exceeds the ADI for chlorfenvinphos, 15% for mercury and 0% for nitrate. Variance in exposure to chlorfenvinphos, mercury and nitrate is mostly caused by interindividual variability instead of true uncertainty. It is concluded that the likelihood that ADIs of chlorfenvinphos and mercury will be exceeded should be further explored. If exceeding is likely, decision makers should focus on identification of high-risk subpopulations, rather than on additional research to obtain more accurate estimates for particular parameters.