Considerations for using reproduction data in toxicokinetic-toxicodynamic modelling

TL;DR

This paper critically examines the challenges of using reproduction data in TKTD models, especially for aquatic invertebrates, and offers recommendations to improve model-data linkage for better environmental risk assessments.

Contribution

It highlights the complexities in linking laboratory reproduction data with TKTD models based on DEB theory and provides preliminary solutions for more accurate modelling.

Findings

Reproduction data often do not directly match model outputs.

Species with discrete clutch reproduction pose specific challenges.

Ignoring data-model discrepancies can impair model calibration and validation.

Abstract

Toxicokinetic-toxicodynamic (TKTD) modelling is essential to make sense of the time dependence of toxic effects, and to interpret and predict consequences of time-varying exposure. These advantages have been recognised in the regulatory arena, especially for environmental risk assessment (ERA) of pesticides, where time-varying exposure is the norm. We critically evaluate the link between the modelled variables in TKTD models and the observations from laboratory ecotoxicity tests. For the endpoint reproduction, this link is far from trivial. The relevant TKTD models for sub-lethal effects are based on Dynamic-Energy Budget (DEB) theory, which specifies a continuous investment flux into reproduction. In contrast, experimental tests score egg or offspring release by the mother. The link between model and data is particularly troublesome when a species reproduces in discrete clutches, and even more so when eggs are incubated in the mother's brood pouch (and release of neonates is scored in the test). This situation is quite common among aquatic invertebrates (e.g., cladocerans, amphipods, mysids), including many popular test species. We discuss these and other issues with reproduction data, reflect on their potential impact on DEB-TKTD analysis, and provide preliminary recommendations to correct them. Both modellers and users of model results need to be aware of these complications, as ignoring them could easily lead to unnecessary failure of DEB-TKTD models during calibration, or when validating them against independent data for other exposure scenarios.