Modeling the effects of endocrine disrupting chemicals on Atlantic croaker: understanding biomarkers and predicting population responses

A number of environmental stressors have been shown to interfere with reproductive and behavioral processes of fish by interfering with endocrine function. Most biomarkers of endocrine disturbance tend to be static measurements from dynamic systems making them difficult to evaluate within the context of an individual, or subtle effects that do not relate well to endpoints of ecological significance. I present an approach that uses a series of models, based on Atlantic croaker, to extrapolate laboratory results to indicators of individual and population health. First, I created a physiologically based model that simulates vitellogenesis in a female fish. The model simulates the major biochemical reactions from the secretion of gonadotropin to the production of vitellogenin. I simulated the effects of three environmental stressors that affect vitellogenin production differently. Model simulations demonstrated that it is possible to relate contaminant-induced changes in biomarkers to vitellogenin production and fecundity. A field application of the vitellogenesis model showed potential utility in interpreting field-measured biomarkers and to infer potential population hazards. Uncertainty analyses identified parameters that contributed most to variability of predictions. Second, I used a statistical model linked to an individual-based model to convert changes in behavior of ocean larvae exposed to two different contaminants to population relevant endpoints. Each contaminant imposed different effects and the effects were largely driven by impaired foraging abilities. Finally, I developed a matrix population model that realistically simulated two distinct populations of Atlantic croaker: Gulf of Mexico and Mid-Atlantic Bight. Simulations incorporated contaminant induced changes that were predicted by the other models, and compared population dynamics for 100 years under baseline conditions and under two separate contaminant scenarios. Predictions generated from the matrix model suggested that contaminant exposures at higher levels than observed in field measurements have the potential to impact populations, and that contaminant residency time within fish and the number of individuals exposed, interact with site-specific factors and life history traits, to determine population effects. The bottom-up approach employed here suggests that it is possible to scale laboratory effects to the population and provides a framework from which to base future model development and testing