a b s t r a c tThe spring phytoplankton bloom on the US Northeast Continental Shelf is a feature of the ecosystem production cycle that varies annually in timing, spatial extent, and magnitude. To quantify this variability, we analyzed remotely-sensed ocean color data at two spatial scales, one based on ecologically defined sub-units of the ecosystem (production units) and the other on a regular grid (0.5°). Five units were defined: Gulf of Maine East and West, Georges Bank, and Middle Atlantic Bight North and South. The units averaged 47 Â 10 3 km 2 in size. The initiation and termination of the spring bloom were determined using change-point analysis with constraints on what was identified as a bloom based on climatological bloom patterns. A discrete spring bloom was detected in most years over much of the western Gulf of Maine production unit. However, bloom frequency declined in the eastern Gulf of Maine and transitioned to frequencies as low as 50% along the southern flank of the Georges Bank production unit. Detectable spring blooms were episodic in the Middle Atlantic Bight production units. In the western Gulf of Maine, bloom duration was inversely related to bloom start day; thus, early blooms tended to be longer lasting and larger magnitude blooms. We view this as a phenological mismatch between bloom timing and the "top-down" grazing pressure that terminates a bloom. Estimates of secondary production were available from plankton surveys that provided spring indices of zooplankton biovolume. Winter chlorophyll biomass had little effect on spring zooplankton biovolume, whereas spring chlorophyll biomass had mixed effects on biovolume. There was evidence of a "bottom up" response seen on Georges Bank where spring zooplankton biovolume was positively correlated with the concentration of chlorophyll. However, in the western Gulf of Maine, biovolume was uncorrelated with chlorophyll concentration, but was positively correlated with bloom start and negatively correlated with magnitude. This observation is consistent with both a "top-down" mechanism of control of the bloom and a "bottom-up" effect of bloom timing on zooplankton grazing. Our inability to form a consistent model of these relationships across adjacent systems underscores the need for further research.Published by Elsevier Ltd.