Résumé
1. Models relating intake rate to food abundance and competitor density (generalized functional response models) can predict forager distributions and movements between patches, but we lack understanding of how distributions and small-scale movements by the foragers themselves affect intake rates. Using a state-of-the-art approach based on continuous-time Markov chain dynamics, we add realism to classic functional response models by acknowledging that the chances to encounter food and competitors are influenced by movement decisions, and, vice versa, that movement decisions are influenced by these encounters. We used a multi-state modelling framework to construct a stochastic functional response model in which foragers alternate between three behavioural states: searching, handling and moving. Using behavioural observations on a molluscivore migrant shorebird (red knot, Calidris canutus canutus), at its main wintering area (Banc d'Arguin, Mauritania), we estimated transition rates between foraging states as a function of conspecific densities and densities of the two main bivalve prey. Intake rate decreased with conspecific density. This interference effect was not due to decreased searching efficiency, but resulted from time lost to avoidance movements. Red knots showed a strong functional response to one prey (Dosinia isocardia), but a weak response to the other prey (Loripes lucinalis). This corroborates predictions from a recently developed optimal diet model that accounts for the mildly toxic effects due to consuming Loripes. Using model averaging across the most plausible multi-state models, the fully parameterized functional response model was then used to predict intake rate for an independent data set on habitat choice by red knot. Comparison of the sites selected by red knots with random sampling sites showed that the birds fed at sites with higher than average Loripes and Dosinia densities, that is sites for which we predicted higher than average intake rates. We discuss the limitations of Holling's classic functional response model which ignores movement and the limitations of contemporary movement ecological theory that ignores consumer-resource interactions. With the rapid advancement of technologies to track movements of individual foragers at fine spatial scales, the time is ripe to integrate descriptive tracking studies with stochastic movement-based functional response models.