Correcting for Phylogenetic Autocorrelation in Species Sensitivity Distributions

Advancing exposure assessment approaches to improve wildlife risk assessment

The exposure assessment component of a Wildlife Ecological Risk Assessment aims to estimate the magnitude, frequency, and duration of exposure to a chemical or environmental contaminant, along with characteristics of the exposed population. This can be challenging in wildlife as there is often high uncertainty and error caused by broad-based, interspecific extrapolation and assumptions often because of a lack of data. Both the US Environmental Protection Agency (USEPA) and European Food Safety Authority (EFSA) have broadly directed exposure assessments to include estimates of the quantity (dose or concentration), frequency, and duration of exposure to a contaminant of interest while considering "all relevant factors." This ambiguity in the inclusion or exclusion of specific factors (e.g., individual and species-specific biology, diet, or proportion time in treated or contaminated area) can significantly influence the overall risk characterization. In this review, we identify four discrete categories of complexity that should be considered in an exposure assessment-chemical, environmental, organismal, and ecological. These may require more data, but a degree of inclusion at all stages of the risk assessment is critical to moving beyond screening-level methods that have a high degree of uncertainty and suffer from conservatism and a lack of realism. We demonstrate that there are many existing and emerging scientific tools and cross-cutting solutions for tackling exposure complexity. To foster greater application of these methods in wildlife exposure assessments, we present a new framework for risk assessors to construct an "exposure matrix." Using three case studies, we illustrate how the matrix can better inform, integrate, and more transparently communicate the important elements of complexity and realism in exposure assessments for wildlife. Modernizing wildlife exposure assessments is long overdue and will require improved collaboration, data sharing, application of standardized exposure scenarios, better communication of assumptions and uncertainty, and postregulatory tracking. Integr Environ Assess Manag 2024;20:674-698. © 2023 SETAC.

Bio-QSARs 2.0: Unlocking a new level of predictive power for machine learning-based ecotoxicity predictions by exploiting chemical and biological information

Practical, legal, and ethical reasons necessitate the development of methods to replace animal experiments. Computational techniques to acquire information that traditionally relied on animal testing are considered a crucial pillar among these so-called new approach methodologies. In this light, we recently introduced the Bio-QSAR concept for multispecies aquatic toxicity regression tasks. These machine learning models, trained on both chemical and biological information, are capable of both cross-chemical and cross-species predictions. Here, we significantly extend these models' applicability. This was realized by increasing the quantity of training data by a factor of approximately 20, accomplished by considering both additional chemicals and aquatic organisms. Additionally, variable test durations and associated random effects were accommodated by employing a machine learning algorithm that combines tree-boosting with mixed-effects modeling (i.e., Gaussian Process Boosting). We also explored various biological descriptors including Dynamic Energy Budget model parameters, taxonomic distances, as well as genus-specific traits and investigated the inclusion of mode-of-action information. Through these efforts, we developed Bio-QSARs for fish and aquatic invertebrates with exceptional predictive power (R squared of up to 0.92 on independent test sets). Moreover, we made considerable strides to make models applicable for a range of use cases in environmental risk assessment as well as research and development of chemicals. Models were made fully explainable by implementing an algorithmic multicollinearity correction combined with SHapley Additive exPlanations. Furthermore, we devised novel approaches for applicability domain construction that take feature importance into account. We are hence confident these models, which are available via open access, will make a significant contribution towards the implementation of new approach methodologies and ultimately have the potential to support "Green Chemistry" and "Green Toxicology".

A benchmark dataset for machine learning in ecotoxicology

The use of machine learning for predicting ecotoxicological outcomes is promising, but underutilized. The curation of data with informative features requires both expertise in machine learning as well as a strong biological and ecotoxicological background, which we consider a barrier of entry for this kind of research. Additionally, model performances can only be compared across studies when the same dataset, cleaning, and splittings were used. Therefore, we provide ADORE, an extensive and well-described dataset on acute aquatic toxicity in three relevant taxonomic groups (fish, crustaceans, and algae). The core dataset describes ecotoxicological experiments and is expanded with phylogenetic and species-specific data on the species as well as chemical properties and molecular representations. Apart from challenging other researchers to try and achieve the best model performances across the whole dataset, we propose specific relevant challenges on subsets of the data and include datasets and splittings corresponding to each of these challenge as well as in-depth characterization and discussion of train-test splitting approaches.

Physiological variables in machine learning QSARs allow for both cross-chemical and cross-species predictions

A major challenge in ecological risk assessment is estimating chemical-induced effects across taxa without species-specific testing. Where ecotoxicological data may be more challenging to gather, information on species physiology is more available for a broad range of taxa. Physiology is known to drive species sensitivity but understanding about the relative contribution of specific underlying processes is still elusive. Consequently, there remains a need to understand which physiological processes lead to differences in species sensitivity. The objective of our study was to utilize existing knowledge about organismal physiology to both understand and predict differences in species sensitivity. Machine learning models were trained to predict chemical- and species-specific endpoints as a function of both chemical fingerprints/descriptors and physiological properties represented by dynamic energy budget (DEB) parameters. We found that random forest models were able to predict chemical- and species-specific endpoints, and that DEB parameters were relatively important in the models, particularly for invertebrates. Our approach illuminates how physiological properties may drive species sensitivity, which will allow more realistic predictions of effects across species without the need for additional animal testing.

ESASeedPARAM: A seed treatment model for threatened and endangered bird species in the United States

The USEPA, National Marine Fisheries Service, and Fish and Wildlife Service are required to assess the risks of pesticides undergoing registration or reregistration to threatened and endangered (i.e., listed) species. Currently, the USEPA lacks a refined model to assess the risks of seed treatments to listed bird species. We developed the Endangered Species Assessment Seed Treatment Probabilistic Avian Risk Assessment Model (ESASeedPARAM) to incorporate species-specific diets, body weights, and food ingestion rates for potentially exposed listed bird species. The model also incorporates information on dissipation of seed residues after planting, and metabolism and elimination by birds during exposure. The ESASeedPARAM estimates hourly intake from ingestion of treated seeds for up to 50 days after planting. For each simulated bird, maximum retained dose (= body burden) and maximum rolling average total daily intake are estimated for acute and chronic exposure, respectively. The model is probabilistic and estimates exposure and risk for 20 birds on each of 1000 fields. The model accounts for interfield variation in the amount of waste grain on the soil surface in tilled, reduced till, and untilled fields. To estimate the fate of each bird from acute exposure, a random value is selected from the appropriate dose-response relationship and compared with the maximum retained dose. If acute exposure exceeds the randomly chosen effects value, mortality is assumed. For chronic risk, the most sensitive No Observed Adverse Effects Level (NOAEL) and Lowest Observed Adverse Effects Level (LOAEL) for an apical endpoint (survival, growth, reproduction) are compared with maximum rolling average total daily intake. In this article, we describe a case study conducted with the ESASeedPARAM for imidacloprid used as a seed treatment on wheat and soybean. Integr Environ Assess Manag 2023;19:527-546. © 2022 SETAC.

The applicability of species sensitivity distributions to the development of generic doses for phytosanitary irradiation

Ionizing radiation is used as a phytosanitary treatment to prevent the introduction of pests through trade. Generic doses are a valuable means to increase the number of pest-commodity combinations that can be treated using phytosanitary irradiation. Generic doses allow for the treatment of the entire taxa for which the dose has been approved, allowing for the treatment of untested species. As such, the approval of a generic dose requires substantial supporting data and careful consideration of the risks involved. We adopt the Species Sensitivity Distribution (SSD) framework, already in widespread use in the field of ecotoxicology and environmental risk assessment, to evaluate generic doses for phytosanitary irradiation treatments. Parametric SSDs for Curculionidae and Tephritidae were developed using existing data on efficacious phytosanitary irradiation treatments. The resulting SSDs provided estimates of the taxa coverage expected by the generic dose, along with the margin of uncertainty. The SSD analysis lends support to the existing 150 Gy generic dose for Tephritidae and a proposed 175 Gy generic dose for Curculionidae. The quantitative estimates of risk produced by the SSD approach can be a valuable tool for phytosanitary rule making, improving the process for generic dose development and approval.

Extrapolating Acute Contact Bee Sensitivity to Insecticides Based on Body Weight Using a Phylogenetically Informed Interspecies Scaling Framework

Plant protection products, including insecticides, are important for global food production but can have adverse effects on nontarget organisms including bees. Historically, research investigating such effects has focused mainly on the honeybee (Apis mellifera), whereas less information is available for non-Apis bees. Consequently, a comprehensive hazard (sensitivity) assessment for the majority of bees is lacking, which in turn hinders accurate risk characterization and consequently bee protection. Interspecies sensitivity extrapolation based on body weight might be a way to improve the situation, but in the past such approaches often ignored the phylogenetic background of the species used, which in turn potentially reduces the robustness of such results. Published acute contact sensitivity data (median lethal dose per bee) of bees to insecticides, their body weight, and their phylogenetic background were used to build interspecies scaling models to predict bee sensitivity based on their weight. The results indicate that 1) bee body weight is a predictor of acute contact bee sensitivity to a range of insecticides, and 2) phylogeny (nonindependence of data points) needs to be considered in cross-species analysis, although it does not always confound the observed effects.  Environ Toxicol Chem 2021;40:2044-2052. © 2021 SETAC.

Integrating Exposure and Effect Distributions with the Ecotoxicity Risk Calculator: Case Studies with Crop Protection Products

Risk curves describe the relationship between cumulative probability and magnitude of effect and thus express far more information than risk quotients. However, their adoption has remained limited in ecological risk assessment. Therefore, we developed the Ecotoxicity Risk Calculator (ERC) to simplify the derivation of risk curves, which can be used to inform risk management decisions. Case studies are presented with crop protection products, highlighting the utility of the ERC at incorporating various data sources, including surface water modeling estimates, monitoring observations, and species sensitivity distributions. Integr Environ Assess Manag 2021;17:321-330. © 2020 Syngenta Crop Protection, LLC. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).

Recent Developments in Species Sensitivity Distribution Modeling

The species sensitivity distribution (SSD) is a statistical approach that is used to estimate either the concentration of a chemical that is hazardous to no more than x% of all species (the HCx) or the proportion of species potentially affected by a given concentration of a chemical. Despite a significant body of published research and critical reviews over the past 20 yr aimed at improving the methodology, the fundamentals remain unchanged. Although there have been some recent suggestions for improvements to SSD methods in the literature, in general, few of these suggestions have been formally adopted. Furthermore, critics of the approach can rightly point to the fact that differences in technical implementation can lead to marked differences in results, thereby undermining confidence in SSD approaches. Despite the limitations, SSDs remain a practical tool and, until a demonstrably better inferential framework is available, developments and enhancements to conventional SSD practice will and should continue. We therefore believe the time has come for the scientific community to decide how it wants SSD methods to evolve. The present study summarizes the current status of, and elaborates on several recent developments for, SSD methods, specifically, model averaging, multimodality, and software development. We also consider future directions with respect to the use of SSDs, with the ultimate aim of helping to facilitate greater international collaboration and, potentially, greater harmonization of SSD methods. Environ Toxicol Chem 2021;40:293-308. © 2020 SETAC.