New technologies as decision aids for the advancement of ecological risk assessment

Moore's law states that the number of transistors that can be placed on an integrated circuit doubles every two years (Moore, 1975). This has led to a steady increase in the processing power of computers over time, and technology is now enhancing and advancing software and scientific applications, which has enabled computationally intensive methods such as machine learning, data science, modeling, and simulation (Figure 1). The advancement of computers and data-driven algorithms is profoundly impacting people's lives. It is changing the way we work, the way we learn, and the way we interact with the world around us. This editorial will discuss how scientists can benefit from the latest technology advancements and related tools by incorporating them into the ecological risk assessment (ERA) to study ecosystems as a way to create refined assessments and accelerate the turnaround times. The incorporation and integration of new technologies as part of the ERA framework will help risk assessors keep up with the natural and anthropogenic stressors released into the environment. The second outcomes of phase 1 are conceptual models, which describe the relationships between ecosystems and their stressors, and often include a generalized diagram or model depicting the movement of contaminants from point and nonpoint sources resulting from natural and anthropogenic processes to endpoints. Establishing relationships between ecosystems and stressors can be daunting; this often represents a source of uncertainty in ERA. This is exemplified during the assessment of multiple stressors and adverse effects, complexity of ecosystems, variability, and scale. An important step to improve accuracy in conceptual models is the use of machine learning, as shown in Figure 1. For instance, researchers at Syracuse University (Beibei et al., 2023) have built machine-learning models that consider 32 watershed properties to predict how human activities (represented by population density and impervious surface area) affect rivers in the United States. The construction of robust models helps to reduce uncertainties when modeling multiple factors simultaneously. Furthermore, a linked open data approach has been used as a data mining tool to develop specific models (e.g., Eurostat/RapidMiner) using linked open data sources to suggest causality between health factors and alcohol consumption in European countries (Lausch et al., 2015). Alternative and more robust conceptual models can be developed using similar technologies that can enable the use of available data sets and information. Phase 2 of an ERA includes the analysis of exposure and effects and their relationships with the assessment points from phase 1 (Suter et al., 2004; USEPA, 1998). The phase 2 outcomes are exposure (exposure characterization) and stressor-response profiles (ecological effect characterization) (USEPA, 1998). Exposure characterization describes the magnitude and spatial and temporal patterns of exposure to stressors (e.g., contaminants) shown in the conceptual model. To this end, autonomous sensor technology has been adapted to estimate exposure levels from pollutants such as metals in lakes associated with adverse effects (Peixoto Mendes et al., 2023). Some of the advantages of this technology include their wireless, remote, ubiquity, and real-time capabilities (Randhawa et al., 2017), which can lead to data collection from remote or hazardous sites. As the affordability and accuracy of commercial sensors increase, their deployment and application also increase in monitoring activities and site identification (Randhawa et al., 2017). For example, real-time electrical conductivity data from autonomous sensors have been shown to be a good surrogate for total aqueous selenium levels in a Canadian boreal lake and for identifying key monitoring locations (Peixoto Mendes et al., 2023). Autonomous sensors can not only be deployed to measure water-quality parameters, but also, they can generate exposure estimates of levels of contaminants on a case-by-case basis, as shown for metals in aquatic ecosystems. The ecological effect characterizations in phase 2 evaluate the evidence that exposure to stressors is associated with an observed response of the assessment endpoint (Suter et al., 2004; USEPA, 1998). Risk assessments often rely on comparing exposure and toxicity data from a single medium or environmental compartment, thereby excluding other sources of information on compound distribution (Gobas et al., 2018). For example, in-house facilities such as flow-through artificial stream systems have been used to set up contaminant exposure scenarios for the assessment of behavioral and histological effects of environmental exposure to per- and polyfluorinated alkyl substances (PFAS) using wild fish and crayfish, which resulted in significant relationships between tissue concentrations of long-chain PFAS and crayfish, and fish critical swimming speed responses (Coy et al., 2022). Although this approach mimics some natural exposure conditions, it might also neglect other important factors, such as the impact of the weathering process on the target compounds by the time an assessment is conducted (Suter, 1997). To further integrate a larger set of exposure parameters, geospatial or mapping tools have been incorporated in more recent risk assessments (Ojha et al., 2022). For example, in a study using available public data to conduct a geospatial analysis to visualize and prioritize sampling locations and potential PFAS hotspot locations, the authors were able to incorporate data from current and past users of the chemical, according to their industry, into their geospatial regression model with 76% accuracy in predicting the likelihood of PFAS prevalence in public drinking water systems while making recommendations for locations where PFAS would also present a risk to the public (Ojha et al., 2022). To model exposure scenarios is crucial for chemicals widely used in a large number of industrial and nonindustrial sectors. It is worth mentioning that the benefit of geospatial technologies is their visualization capability (e.g., heat maps) to show potential exposure risks, which can assist in establishing the relationships between exposure and stressor-response profiles, and the risk communication with the stakeholders (Figure 1). Phase 3 of the ERA includes risk characterization by integrating exposure and stressor-response profiles to evaluate the likelihood of adverse ecological effects associated with exposure to stressor(s) (Suter et al., 2004; USEPA, 1998). One of the key outcomes of phase 3 is ecological risk estimates (USEPA, 1998). Examples of ERA's risk estimates can be the use of conservative estimates of exposure concentrations compared to conservative benchmarks (e.g., EC50 and NOEL), in which exposure concentrations of contaminants below benchmarks would be considered to pose a minimal risk (Suter, 1997). In this case, a hazard quotient (HQ) ratio between the ambient exposure concentration and the toxicologically effective concentration can be used to produce rankings of risk for an ecological effect from exposure to that contaminant (i.e., in general: HQ ≥ 1, high risk; HQ < 1, low risk) (USEPA, 1998). Most recently, refined assessments of risks, based on simulations and probabilistic models, have been used to determine risk estimates and, in some cases, to refine previous assessments (Giddings et al., 2014). For example, in a risk assessment focused on risks to aquatic organisms from exposure to chlorpyrifos, the authors used probabilistic analyses (e.g., joint probability curves) based on simulation models to compare exposure concentrations measured in surface waters (Giddings et al., 2014). Based on the predictions from probabilistic analysis using species sensitivity distributions, the authors concluded that risks from the direct effects of chlorpyrifos on fish were minimal across three different water bodies (Giddings et al., 2014). The authors also acknowledged that sufficient data are needed to conduct a probabilistic assessment of risks, and when this is not possible, the assessment of risks can still include HQ values (Giddings et al., 2014). Additionally, technologies such as virtual reality (VR) environments have the potential to offer experiences in laboratory settings or unique environments and situations (Figure 1). For example, VR has been used to communicate urban flood risks (Winkler et al., 2018). This can be incorporated into risk communication of chemical scenarios for the public. Furthermore, the three-dimensional (3D) visual space adds realism to the simulations and displayed environments. Here, users can engage in short-term flooding interventions or stay away from highly polluted sites, and by interacting with the software, they can improve their understanding of the chemical fate of fluid flow and see the effect of polluted water on environmental endpoints, ultimately helping the users to develop understanding related to the assessment and management of chemicals or flooding events. Three-dimensional simulations can also be beneficial in situations previously studied only with numerical methods and reduced visualization experience. Likewise, Gabcan et al. (2020) created a virtual environment scenario for a radioactive waste infiltration model applied to the Abadia de Goiás repository in Brazil to aid the understanding of the water infiltration inside a repository, allowing comparison of the variation of the amount of the radioactive material in the scenario of water infiltration as a function of the change in the 3D simulation model, producing a more realistic perception of the probable scenario by the user. A generalized diagram for the use of new technologies in an ERA was constructed from the traditional phases (Figure 1). Although some of the technologies discussed here are not widely utilized or fully developed (e.g., sensors) to be implemented immediately, it is expected that with the advancement of technology and computing capabilities in the coming years, their use will become more predominant in the way that ERAs are conducted. In a recent work by Rivetti and Campos (2023), they argue the need to “foster a transition toward the use of New Approach Methodologies (NAMs)” for nonanimal approaches that apply fit-for-purpose tools—based on the latest advances in science and technology—for human and environmental protection. This suggests that there is still a need to continue enhancing the ERA process as more contaminants are being released into the environment (Dale et al., 2008). It is expected that the incorporation and integration of new technologies as part of the ERA framework will help risk assessors keep up with the natural and anthropogenic stressors released into the environment. Federico Sinche Chele: Conceptualization; visualization; writing—original draft; writing—review and editing. Priscilla Jimenez-Pazmino: Visualization; writing—review and editing. Konstantin Läufer: Writing—review and editing. The authors thank their respective institutions for allowing time to complete the present work. The authors declare no conflicts of interest.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.