Modeling food choice in the two suspension-feeding bivalves, Crassostrea virginica and Mytilus edulis

Abstract

Suspension-feeding bivalves are able to sort and select food particles from a complex mixture. Recent reports have indicated that this selection is mediated by interactions between lectins present in mucus covering the feeding organs and carbohydrates associated with the surface of microalgae. In this study, several statistical methods were evaluated to predict the likelihood for a given microalga to be ingested or rejected based upon its cell surface carbohydrate signature. First, the carbohydrate signatures of 16 microalgae were characterized using 10 different lectins. In June 2014, a subset of microalgae (12 species) was then used in feeding experiments where different pairs of microalgae were presented to oysters (Crassostrea virginica) and mussels (Mytilus edulis) to evaluate selection. Results show that cell surface carbohydrates are good predictors for particle fate. Specifically, microalgae rich in glucose/mannose residues were preferentially selected by both oysters and mussels. Several statistical methods for predicting the likelihood of a given alga being ingested or rejected were evaluated, and a decision tree that accurately models selection in the two bivalves is proposed even though the model warrants further validation with different species or in various seasons. Overall, these findings provide