High-resolution, physical-biological models of coastal and shelf regions typically use a single functional phytoplankton group, which limits their ability to represent ecological gradients (e.g., highly productive shelf systems adjacent to oligotrophic regions), as these are dominated by different functional phytoplankton groups. We implemented a size-structured ecosystem model in a high-resolution, regional circulation model of the northeast North American shelf and adjacent deep ocean in order to assess whether the added functional complexity of two functional phytoplankton groups improves the model's ability to represent surface chlorophyll concentrations along an ecological gradient encompassing five distinct regions. We used satellite-derived SST and sea-surface chlorophyll for our model assessment, as these allow investigation of spatial variability and temporal variations from monthly to interannual, and analyzed three complementary statistical measures of model-data agreement: model bias, root mean square error and model efficiency (or skill). All three measures were integrated for the whole domain, for distinct subregions and were calculated in a spatially explicit manner. Comparison with a previously published simulation that used a model with a single phytoplankton functional group indicates that the inclusion of an additional phytoplankton group representing picoplankton markedly improves the model's skill.