Comparative models of mussel bioenergetics and their validation at field culture sites

Abstract

Models of bioenergetics of the blue mussel (Mytilus edulis) were constructed to simulate growth in suspended culture at field sites in Upper South Cove (Nova Scotia) and Marennes-Oléron Bay (France). Two models with contrasting complexity of feeding behaviour were used to test whether simple empirical formulations of the energy budget (statistical model) were adequate to simulate growth compared to a more fully mechanistic model. In the statistical model, ingestion was related to a single food source (particulate organic matter, POM) and absorption efficiency via laboratory feeding studies from the literature. In the mechanistic model, filtration, particle rejection and selection, and absorption efficiency were related to phytoplankton, detrital food and total particulate load. Respiration terms were identical between models to facilitate comparisons between feeding behaviour. Measured tissue trajectories of cultured mussels were used to groundtruth model predictions. The Nova Scotia grow-out site was characterized by low turbidity and particulate organic matter (POM), and seasonally high chlorophyll, while the Marennes-Oléron site had high turbidity (up to 180 mg l −1) and POM, but similar chlorophyll to the Nova Scotia site. Results of the simulations indicated that for Nova Scotia, the statistical model provided a realistic growth trajectory, whereas the mechanistic model underpredicted growth during the first half of the year due to low phytoplankton biomass. Use of fluorometer records rather than water samples as a measure of chlorophyll improved the accuracy of the growth prediction. For Marennes-Oléron, the statistical model was unsuccessful since its POM-ingestion relationship did not allow for limitation by turbidity. The mechanistic model applied to this site was reasonable in predicting a growth trajectory, but it was sensitive to the functional curve of particle rejection as well as the maximum daily ingestion. Simulations at both sites were sensitive to the POC content of POM and the absorption efficiency of detritus. These simulations demonstrate that simple formulations of food and feeding will suffice in predicting growth where seston values are not extreme (e.g., high turbidity). Although the separation of suspended food source into POM and chlorophyll may be convenient, it is not clear how organic particle selection occurs, considering the aggregated nature of particles in the field, and the close association of POM with silt grains. The sensitivity of models to both food quality and feeding formulation requires better knowledge of variance in suspended food sources and the ability of bivalves to deal with natural particle mixtures.