Abstract
Fish spawning is often used as an integrated measure of reproductive toxicity, and an indicator of aquatic ecosystem health in the context of forecasting potential population-level effects considered important for ecological risk assessment. Consequently, there is a need for flexible, widely-applicable, biologically-based models that can predict changes in fecundity in response to chemical exposures, based on readily measured biochemical endpoints, such as plasma vitellogenin (VTG) concentrations, as input parameters. Herein we describe a MATLAB® version of an oocyte growth dynamics model for fathead minnows (Pimephales promelas) with a graphical user interface based upon a previously published model developed with MCSim software and evaluated with data from fathead minnows exposed to an androgenic chemical, 17β-trenbolone. We extended the evaluation of our new model to include six chemicals that inhibit enzymes involved in steroid biosynthesis: fadrozole, ketoconazole, propiconazole, prochloraz, fenarimol, and trilostane. In addition, for unexposed fathead minnows from group spawning design studies, and those exposed to the six chemicals, we evaluated whether the model is capable of predicting the average number of eggs per spawn and the average number of spawns per female, which was not evaluated previously. The new model is significantly improved in terms of ease of use, platform independence, and utility for providing output in a format that can be used as input into a population dynamics model. Model-predicted minimum and maximum cumulative fecundity over time encompassed the observed data for fadrozole and most propiconazole, prochloraz, fenarimol and trilostane treatments, but did not consistently replicate results from ketoconazole treatments. For average fecundity (eggs•female(-1)•day(-1)), eggs per spawn, and the number of spawns per female, the range of model-predicted values generally encompassed the experimentally observed values. Overall, we found that the model predicts reproduction metrics robustly and its predictions capture the variability in the experimentally observed data.