Predictions of phytoplankton depletion by benthic bivalves in shallow, tidally driven estuaries must account for the formation of concentration boundary layers near the bed. These regions of low phytoplankton concentration result from the dynamic interactions of bivalve siphonal currents (excurrent jets and incurrent sinks) with the overlying turbulent boundary layer. To study the near- bed hydrodynamics of the benthic boundary layer, we conducted controlled experiments with the species Tapes japonica and Potamocorbula amurensis. Two different sets of experiments were conducted in two laboratory flumes, one with model clams and one with live animals.

Using multiple jets and sinks to represent bivalve siphonal currents, model experiments were performed to study the mixing characteristics of phytoplankton-depleted fluid. PLIF (Planar Laser Induced Fluorescence) and LDV (Laser Doppler Velocimetry) were used as the main diagnostics to characterize respectively the concentration and velocity fields. Refiltration fractions were determined by monitoring the concentration of dye ingested by incurrent siphons. Results show that refiltration fractions can be as high as 48%, and are a function of several dimensionless parameters: animal spacing, S/d_o, velocity ratio, , siphon height, hs/do, and crossflow Reynolds number, Rex, based on the distance from the boundary layer trip, x. (S is the mean distance between animals, do is the excurrent siphon diameter, h_s is the animal siphon height, is the excurrent jet velocity, and is the freestream velocity.) We found that a good estimate of refiltration, n, based on animal spacing, is (n S/d_o) 2-3. Differences in concentration profiles calculated from PLIF images for different flow conditions are likely due to the relative influence of four sources of turbulence in the flow: boundary layer shear, boundary roughness, jet in a crossflow, and multiple jet interactions.

Results of experiments with live animals indicate that the presence of actively feeding bivalves causes a velocity defect region close to the bed in the longitudinal velocity profile. The results of live animals observations are used to validate the assumptions used in the design of the model clams and to describe feeding behavioral responses to changes in crossflow velocity. Finally, implications for field observations of concentration boundary layer formation are discussed and our refiltration results are incorporated into a conceptual mass-balance model for food depletion at the bed.