You are here:
Identifying uncertainty trade-offs of ecological risk assessment objectives to guide model complexity
Citation:
Raimondo, Sandy, M. Etterson, N. Pollesch, K. Garber, A. Kanarek, W. Lehmann, AND J. Awkerman. Identifying uncertainty trade-offs of ecological risk assessment objectives to guide model complexity. Society of Environmental Toxicology and Chemistry (SETAC) Europe, On Line, On Line, IRELAND, May 03 - 07, 2020.
Impact/Purpose:
We present a framework for the development of population models for ERAs that aligns the trade-offs of generality, realism, and precision of the ERA objective with the uncertainty of model output. While the framework presents an approach that will guide the development of population models of varying levels of complexity based on the objective of the ERA, the principles that define the framework can be applied to other types of models used in ERA.
Description:
Risk managers are charged with interpretation of Ecological Risk Assessments (ERA) and need to ensure decisions reflect a strong scientific basis of the models used and understand uncertainties of both the models and assessments. Model uncertainty is directly connected to model complexity, which is a critical element of good modelling practice; however, there is little guidance on selecting the level of model complexity appropriate for specific ERA objectives. We present a framework that provides guidance for developing and applying models of varying complexity in regulatory decision-making using population models. The framework is centered on the trade-offs of generality, realism, precision, and complexity for both ERAs and models with consideration of resource investment. The framework considers what uncertainties are acceptable for an assessment based on its objectives and uses those uncertainties to guide model development. In doing so, our framework links the trade-offs of an ERA objective with those of model building. Also central to our framework is the concept that complexity is not independent of these trade-offs, increasing as models move both from general to realistic, and from lower to higher precision. Complexity can increase such that knowledge-based, or qualitative, uncertainties are reduced by adding mechanisms or model functions that represent real-world scenarios. Complexity may also increase such that the level of precision, or confidence in model output, is increased through the addition or improvement of mathematical functions that reduce quantitative uncertainty. Application of the framework first identifies ERAs by their level of generality, realism, precision, and associated complexity which is then used to guide commensurate trade-offs in model development. For both ERAs and models, realism increases with species, spatial, and temporal specificity and inclusion of real-world complexity, whereas generality is based on surrogate species, contains less defined spatial and temporal scales, and has broad applicability across locations. To demonstrate these concepts, we use case studies that represent ERAs and associated models that represent various trade-offs within the framework. While the framework presents an approach that will guide the development of population models of varying levels of complexity based on the objective of the ERA, the principles that define the framework can be applied to other types of models used in ERA.
