The effects of hydrodynamics on phytoplankton dynamics in shallow estuaries has been studied with three numerical models. The 1D phytoplankton model provided a means of studying in detail the effects of density stratification, vertical turbulent mixing, and sinking on phytoplankton bloom initiation in the presence of light limitation and benthic grazing. It was found that strain-induced periodic stratification (SIPS) does not increase the likelihood of a bloom beyond that of a constantly unstratified water column, whereas persistent stratification does. In addition to the persistence of the stratification, the "details" of the stratification (i.e. surface layer depth, pycnocline thickness, vertical density difference, tidal current speed, total depth) are important in determining conditions which promote the onset of phytoplankton blooms. The surface and bottom layers of a stratified water column are not truly decoupled; turbulent and advective leakage can be responsible for the loss of a significant percentage of the biomass produced in the surface layer.

The Pseudo-2D model allowed us to hypothesize as to the relative importance of various processes in a coupled channel/shoal system. Processes identified as "First Order'' (capable of controlling the occurrence of a systemwide bloom) are lateral exchange, benthic grazing in the shoal, and light attenuation in the shoal. Processes identified as "Second Order'' (capable of controlling only systemwide bloom magnitude or the occurrence of a local bloom) are channel stratification, benthic grazing in the channel, and light attenuation in the channel.

TRIM-BIO, a depth-averaged hydrodynamic model (TRIM2D) with incorporated depth-averaged phytoplankton dynamics, allowed us to explore the effects of realistic horizontal transport and bathymetry. Under grazing-limited conditions, the deeper regions may be the most productive, while, under light-limited conditions, the shallower regions may be associated with the highest effective growth rates. Very shallow regions experience the greatest semidiurnal and spring/neap variability in effective growth rates, due to the interaction between temporal shallowing/deepening of the water column and benthic grazing. The following "global'' mechanisms may control the occurrence and location of a bloom: 1) import to a region; 2) export from a region (residence time); 3) "buffer zones'' which receive phytoplankton from the shallowest areas on ebb tide, allowing the phytoplankton to avoid extremely low or negative effective growth rates nearshore in the presence of grazing; and 4) "lateral sloshing'' of phytoplankton during ebb tides from the shoals into the channel, where effective growth rates are commonly negative.