AbstractIn their natural environment, planktonic grazers encounter considerable size variation in prey species. As a quantitative representation of feeding on diverse prey, theoretical studies use feeding kernels to describe grazing intensity as a function of body size, a factor that is a strongly discriminative property of prey. However, kernel functions are in general based on heuristic and poorly tested choices, lack an unequivocal definition, and are often based on experiments using mono-specific prey rather than a broad prey spectrum found in nature. The work reported here seeks to fill these theoretical gaps by exploring the distinction between the ingestion kernel and the selection kernel. While the ingestion kernel describes which size classes can be potentially used by a consumer, the selection kernel depicts the actual size-dependent grazing on prey assemblages. Simple biomechanical laws show that the ingestion kernel takes a log-normal shape with a universal width (1/EMBED Equation.DSMT4 in log-diameter-space). Collected experimental data support the predicted constant value of the ingestible logarithmic size range across plankton taxa. The selection kernel resolves behavioral modifications during the capture process. In particular, the inverse of the variable kernel width, defines consumer selectivity as a quantitative behavioral trait. Small kernel width thus large selectivity values, represent an optimization by consumers towards high food availability. Optimality in selectivity is tested using observations on copepod grazing. Integral grazing rates that incorporate prey diversity and adaptive consumer selectivity are demonstrated to provide a sound mechanistic basis for size-based plankton models.