rDEBtktd, an R-package for analysis and forward-prediction of sublethal effects

Environmental Risk Assessment (ERA) of chemicals is based on standard laboratory toxicity tests with living organisms which ensure controlled experimental conditions and reproducibility. These toxicity tests are usually carried out under constant exposure concentrations, which can be far from reality of environmental exposure regimes as foreseen by the practical use of chemicals. In that respect mechanistic effect modelling, such as Toxicokinectic – Toxicodynamic (TKTD) modelling, has recently been playing an increasing role in the extrapolation of effects from constant controlled exposure conditions to time-variable exposure, closer to real environmental conditions. Among TKTD models, models based on the Dynamic Energy Budget theory adapted for ecotoxicology (DEB-TKTD models) offer a comprehensive framework to analyse and extrapolate sublethal effects (growth and reproduction) of chemicals on individual organisms across their whole life cycle. While the EFSA Scientific Opinion on the state of the art of TKTD effect models (EFSA PPR, 2018. EFSA Journal;16(8):5377) considers DEB-TKTD models as valuable tools for ERA, their full acceptance by stake-holders still requires the development of standardized and user-friendly tools. To bridge this gap, we developed ready-to-use functions within a new R package “rDEBtktd”. This package takes advantage of the general Bayesian framework thus enabling the estimation of probability distributions for physiological DEB parameters and TKTD parameters, from which uncertainties can be easily quantified to be then propagated to forward-predictions for untested time-variable exposure scenarios. The physiological part of the DEB-TKTD model we implemented follows the original definition of the DEB model, which allows using the parameter values available for more than 1000 species in the Add-my-Pet database as prior information for the Bayesian inference process. This poster illustrates: (1) how to simply simultaneously estimate all the parameters of the DEB-TKTD model from one or several growth and reproduction datasets, (2) how to produce informative summaries to assess the results of the Bayesian inference and check all goodness-of-fit criteria, (3) how to make growth and reproduction predictions for untested time-variable exposure scenarios, (4) and finally the influence of both data quantity and design on the precision of parameter estimates. Environmental Risk Assessment (ERA) of chemicals is based on standard laboratory toxicity tests with living organisms which ensure controlled experimental conditions and reproducibility. These toxicity tests are usually carried out under constant exposure concentrations, which can be far from reality of environmental exposure regimes as foreseen by the practical use of chemicals. In that respect mechanistic effect modelling, such as Toxicokinectic – Toxicodynamic (TKTD) modelling, has recently been playing an increasing role in the extrapolation of effects from constant controlled exposure conditions to time-variable exposure, closer to real environmental conditions. Among TKTD models, models based on the Dynamic Energy Budget theory adapted for ecotoxicology (DEB-TKTD models) offer a comprehensive framework to analyse and extrapolate sublethal effects (growth and reproduction) of chemicals on individual organisms across their whole life cycle. While the EFSA Scientific Opinion on the state of the art of TKTD effect models (EFSA PPR, 2018. EFSA Journal;16(8):5377) considers DEB-TKTD models as valuable tools for ERA, their full acceptance by stake-holders still requires the development of standardized and user-friendly tools. To bridge this gap, we developed ready-to-use functions within a new R package “rDEBtktd”. This package takes advantage of the general Bayesian framework thus enabling the estimation of probability distributions for physiological DEB parameters and TKTD parameters, from which uncertainties can be easily quantified to be then propagated to forward-predictions for untested time-variable exposure scenarios. The physiological part of the DEB-TKTD model we implemented follows the original definition of the DEB model, which allows using the parameter values available for more than 1000 species in the Add-my-Pet database as prior information for the Bayesian inference process. This poster illustrates: (1) how to simply simultaneously estimate all the parameters of the DEB-TKTD model from one or several growth and reproduction datasets, (2) how to produce informative summaries to assess the results of the Bayesian inference and check all goodness-of-fit criteria, (3) how to make growth and reproduction predictions for untested time-variable exposure scenarios, (4) and finally the influence of both data quantity and design on the precision of parameter estimates.