Strong relationships between size and other traits have long motivated studies of the size structure and dynamics of planktonic food webs. Size structured ecosystem models (SSEMs) are often used to represent the behavior of these ecosystems, with organism size as a first order approximation of the axis of biological diversity. Previous studies using SSEMs have reported the emergence of localized “peaks” in the size spectrum, a phenomenon that will be referred to in this study as “quantization”. However, SSEMs that are used routinely in Earth System Models (ESMs), they tend to be too coarsely discretized to resolve quantization. Observational studies of plankton biomass have also shown qualitatively similar patterns, with localized peaks along the size spectrum. The conditions under which quantization occurs and the ecosystem parameters that control the locations of the biomass “peaks” along the size spectrum have not been systematically explored. This study serves to simultaneously advance our understanding of the constraints on quantization in size-structured ecosystems, and to suggest an approach to discretizing SSEMs that leverages quantization to select a greatly reduced number of size classes. A size-structured model of the pelagic food web, similar to those implemented in global models, is used to investigate the sensitivity of biomass peaks to predator–prey interactions, and nutrient forcing. This study shows that the location of biomass peaks along the size spectrum is strongly controlled by the size selectivity of predation, and the location of biomass peaks along the size spectrum is less sensitive to variations in nutrient supply, external ecosystem forcing, and vertical heterogeneity. Taking advantage of a robust localization of biomass peaks, the dynamics of a continuous planktonic size spectrum to be represented using a few selected size classes, corresponding to locations of the peaks along the size spectrum. These findings offer an insight on how to approach discretization of size structured ecosystem model in Earth system models.