Advancing Effect Modeling for ERA: Bridging Scales, Data Gaps, and Regulatory Challenges
Valery E. Forbes, Florida Atlantic University; Ines Haberle, Florida Atlantic University; Louise Stevenson, Oak Ridge National Laboratory; and Maxime Vaugeois, Syngenta Crop Protection
The evolving landscape of ecological risk assessment (ERA) demands innovative approaches to address the intricate interplay of environmental stressors, species vulnerabilities and ecosystem dynamics across varying spatial and temporal scales. This session, held during the SETAC North America 46th Annual Meeting in November in Portland, Oregon, explored cutting-edge applications and challenges for effect modeling in ERA, with a focus on bridging the gap between scientific advancements and regulatory implementation.
The session explored innovative techniques for addressing data gaps, particularly for modeling species with limited datasets, such as endangered species and those affected by reduced animal testing initiatives. Through oral and poster presentations, the session aimed to foster dialogue among modelers, risk assessors and regulators to advance the integration of effect modeling in ERA and support evidence-based environmental management decisions. The session was sponsored by the SETAC North America Interest Group on Effect Modeling, and presentations highlighted both the potential and limitations of effect models in ERA.
Ines Haberle, Florida Atlantic University, and coauthors presented their research on the development of a generalized individual-based population model (IBM) for freshwater mussels. The model was designed to be parameterized for distinct mussel life-history categories and builds upon prior work that combined Dynamic Energy Budget (DEB) modeling with life-history theory to capture sub-lethal effects of environmental stressors at the individual level. Previous work highlighted that freshwater mussels can be categorized into three distinct groups representing different life-history strategies – opportunistic, equilibrium and periodic – based on their life-history traits, which emerge from individual-level bioenergetic processes. The DEB-IBM integrates these processes into a population-level framework, enabling assessment of how effects such as reduced growth, delayed maturation or lower fecundity may influence long-term population dynamics. The DEB-IBM model, developed in the NetLogo programming environment, allows simulation of direct and indirect threats, such as pesticide exposure and sedimentation, incorporating realistic environmental drivers like temperature and food availability. It also provides a framework for comparing how different life-history strategies may buffer or exacerbate population-level consequences of stress. Because life-history characteristics drive population dynamics, population-level outcomes may differ across life-history categories even when individual-level responses appear similar. For instance, model simulations showed that populations of the equilibrium species (Ortmanniana ligamentina) took longer to recover after the same individual-level stress on assimilation than the opportunistic species (Truncilla donaciformis). Future work will focus on simulating additional species and more realistic exposure scenarios and will explore various mitigation strategies.
In the corresponding poster session, Lorena Rabello Martins, Florida Atlantic University, and coauthors presented a conceptual and computational model of the most complex phase of the freshwater mussel life cycle, the parasitic early life stage (larvae, i.e., glochidia) and the attachment to a fish host required for successful metamorphosis. The model considered mussel infection strategy (broadcast vs. luring) and, based on the mussels’ life history, the difference in reproductive output. Existing data on host compatibility were used to predict attachment outcomes across different host communities, and effects of theoretical stress – reduction of fecundity or attachment probability – were explored and compared between strategies. By identifying the conditions that enhance or constrain larval success, the model can support conservation planning efforts and help prioritize species and habitats for protection. Looking ahead, this approach could be incorporated into a broader population-level dynamics framework, providing deeper insight into how early life-stage processes influence long-term persistence of threatened mussel populations.
Maxime Vaugeois, Syngenta Crop Protection, and colleagues presented their talk on efforts to develop an agent-based model (ABM) for fathead minnows (FHM), a widely used standard test species for aquatic ERA in the USA and compared their model results to those of an integral projection model (IPM). The ABM is based on Dynamic Energy Budget theory and incorporates toxicokinetic-toxicodynamic modules to represent effects of stressors on the fathead minnow metabolism. The IPM is a size-structured model that uses readily measurable variables (weight and length) to represent growth, survival and reproduction (and their stress-driven changes) of fish in a population. Both models incorporate seasonal reproduction, density-dependence and time-variable chemical exposure scenarios, but they represent a few life-cycle processes differently (density-dependence and overwintering). The ABM and IPM were independently developed and parameterized using existing literature and laboratory data on FHM life history. They were applied to simulate population-level effects under the same time-varying exposure scenario. The ABM provided detailed insights into the individual-level responses to the exposure, whereas the IPM offered a more concise, distribution-based output to inform population-level perspectives. The direct comparison of the ABM and IPM outputs revealed minor differences in their predictions of population-level effects when density-dependence and overwintering were not considered. The comparison of the density-dependence and the overwintering processes between the ABM and the IPM showed marked differences on population dynamics, and further work is ongoing to analyze the impact on predictions of population-level risk assessment. The analysis shows how modeling choices may affect the final outputs and highlights strengths and limitations of each approach. Despite some differences, the work also provides a valuable example of model cross-validation, enhancing confidence in modeling results for ERA.
Vaugeois and colleagues also presented two posters on specific effect model applications. First poster introduced a tiered ecological risk assessment framework for the neonicotinoid insecticide thiamethoxam (TMX) on Chironomus riparius, a dipteran species recognized as highly sensitive to neonicotinoids among aquatic invertebrates. The presented framework integrates IBM, DEB and toxicokinetic-toxicodynamic (TKTD) modules to extrapolate laboratory data to real-world scenarios, assessing survival, growth, time to emergence, and reproduction under varying TMX concentrations and temperatures, while accounting for density-dependent effects like cannibalism. Worst-case TMX exposure profiles, derived from a geographically extensive and temporally extended water monitoring program, were used to determine exposure multiplication factors (EMFs) analogous to ECx values, quantifying risk at a pre-defined effect strength. The model evaluates the transferability of the EMFs from the individual to the population level and provides a comprehensive and more realistic risk assessment of TMX exposure in aquatic environments. The second poster shared the development of a mechanistic effect model of Lepidoptera, combining DEB theory with TKTD approaches. Here, DEB captures animal life cycles in terms of energy allocations for growth, development and reproduction, whereas TKTD incorporates chemical uptake, metabolism, lethal effects and/or sublethal effects on the DEB processes. The authors highlighted that such an approach allows for more accurate extrapolation from lab data to real-world scenarios, ultimately providing a stronger foundation for population-level risk assessments and supporting better regulatory decisions for endangered species protection.
Research by Jeffrey Minucci of the U.S. Environmental Protection Agency and colleagues, presented by Vaugeois because Minucci was unable to attend, discussed a variety of models developed for honey bees, bumble bees and non-Apis bees that span organism-level effects (e.g., toxicokinetic-toxicodynamic models), colony models and population models. In order to facilitate the integration of these models into the risk assessment process, the Bee Modelling Interest Group was established within the International Commission for Plant-Pollinator Relationships (ICPPR), with the aim of improving the acceptability and usability of exposure and effects models in bee risk assessments. The interest group provides a platform for exchange and discussion between stakeholders from different groups including regulators, industry, contract research organizations and academia. The presentation described three work groups that have been created based on a survey of members' perceptions of the urgency and feasibility of various topics: (1) improving characterization of exposure pathways, (2) incorporation of sublethal effects, and (3) establishing a bee model inventory. The presentation introduced the aims, activities and goals of the ICPPR Bee Modelling Interest Group and its three active work groups. This initiative demonstrates how tripartite collaborations can accelerate the adoption of mechanistic models within regulatory frameworks, ultimately enabling more accurate and efficient risk assessments that better protect populations while supporting sustainable agriculture.
Julann Spromberg, Getchell, addressed a very important aspect of models when being used in regulatory decision-making processes: the model rigor needs to match the types of decisions to be made, the time frame for reassessment, and the level of risk the regulator/agency deems appropriate. Model risk, defined as “the possibility the model is wrong or the output is misapplied,” may stem from data limitations, parameter estimation uncertainty, model misspecification or inappropriate use of a model. Along these lines, Spromberg proposed a decision framework that can assist regulators as they consider using models as a line of evidence in various regulatory contexts, applicable both if using an existing model, or if a new model is being constructed. She highlighted that acknowledging and managing model risk can increase confidence in using models in regulatory contexts and support progress toward utilizing models in regulatory decision-making.
The final part of the session was structured as a lively panel discussion moderated by Valery E. Forbes, Florida Atlantic University, on the overarching theme of “Opportunities and Challenges for Enhancing the Role of Effect Modeling in ERA.” Panelists Ines Haberle, Florida Atlantic University; Wayne G. Landis, Western Washington University; Charlie Menzie, Exponent; Dwayne R.J. Moore, Stone Environmental Inc.; Julann Spromberg, Getchell; and Maxime Vaugeois, Syngenta Crop Protection, enriched the conversation by bringing a wide range of backgrounds, affiliations and expertise to the table. Panelists were asked whether they had ever used mechanistic effect modeling in their own work, and if so, they were asked to elaborate on the kind of models and how they were used. Most, but not all, panelists had used effect models, and given the variety of sectors represented, the way they used the models also varied. Further discussion focused on the different degrees of acceptance of effect models versus exposure models in regulatory risk assessments, the latter of which has a much longer history of use. Panelists shared their perspective on issues related to model validation (and what constitutes sufficient validation in the eyes of different stakeholders), the use of effect models at higher versus lower ERA tiers, and how the usefulness and acceptability of effect models for regulatory ERAs could be facilitated. The session concluded with an important take-home message: more multi-stakeholder collaborations on model development and evaluation outside of specific regulatory decisions – such as is being done in the Bee Modeling Interest Group presented by Minucci et al. – are urgently needed.
Contact: [email protected]