A model-based method to analyze ecotoxicology experiments in mesocosm
Résumé
The mesocosms (experimental ecosystems) improve the ecological relevance of the studies on the effects of xenobiotics on aquatic species. However, in mesocosm studies, the number of replicates is limited by practical and financial constraints. In addition, high levels of biological organization are characterized by a high variability of descriptive variables. This variability and the poor number of replicates have been recognized as a major drawback for detecting significant effects of chemicals in mesocosm studies. These characteristics induce a large uncertainty on the probabilistic distributions of the endpoints (location parameter, variability and shape) in mesocosm experiments. To reduce this uncertainty, a solution would be to optimize the knowledge used to define the expected probabilistic distribution of the endpoints for a given experiment in control conditions. A way to combine all the information available is to develop, calibrate and validate a model of for the population studied in mesocosm. This model will be fed by the characteristics of a given experiment (initial points, environmental scenario), the characteristics of the species that was chosen (ethology, ecology, population dynamics) and the characteristics of the experimental system (for example species composition of the mesocosm). We evaluated here this methodology on experiments conducted in lotic mesocosm and focused on the three-spined stickleback (Gasterosteus aculeatus)population dynamics. An individual based-model (IBM) was developed from a bioenergetic model following the Dynamic Energy Budget theory in order to represent the life-cycle of sticklebacks (DEB model). Sensitivity analyses were performed to explore the DEB-IBM, and the parameters of the model were adjusted using two dataset with two different environmental scenarii. The predictability of the model was then tested using two other datasets. Finally, we observed on a case-study that mesocosm data analysis was improved by using the model-based methodology proposed. To conclude, designing an individual-based model is very promising for improving mesocosm data analysis.