The effect of the hard clam Mercenaria mercenaria (L.) on the exchange of dissolved nutrients (silicate, phosphate, ammonium, nitrate+nitrite) and gases (oxygen, carbon dioxide) across the sediment-water interface was examined in 1983 and 1984 using experimental mesocosms (13 m3), designed to simulate a shallow coastal ecosystem, that allow for reciprocal biogeochemical interactions between water column (5 m) and bottom sediments (∼30 cm deep). Benthic, fluxes, measured during a spring-summer warming period, were compared for mesocosms maintained either with added M. mercenaria (16 per m2, treatment) or without M. mercenaria (control) as a component of the benthic community. Fluxes were within the range observed for the (control) community in situ in mid-Narragansett Bay and exhibited the pattern of increasing flux with increasing temperature observed in nature. For dissolved inorganic (DIN=ammonia+nitrate+nitrite), analytical problems allowed comparison of fluxes only at the higher temperatures (∼17° to 22°C); average DIN fluxes were 57% higher in mesocosms with clams. Fluxes of other nutrients and gases were modelled as exponential functions of temperature (∼9° to 22°C) using a linear regression of the natural logarithm of flux upon temperature in both treatment and control situations. Differences between regression slopes and intercepts for conditions with and without clams were assessed by analysis of covariance. Slopes of the linearized temperature-flux relationships were not significantly different between treatments for any flux measured, suggesting that presence of the clams did not alter the basic relationship of flux and temperature normally observed with this community. For oxygen consumption and silicate production, the regression intercepts were significantly higher in the clam treatment, suggesting that the level of flux was generally elevated in the presence of M. mercenaria by about 20% (oxygen) and 86% (silicate) at any given temperature. Data for carbon dioxide and phosphate showed similar trends to oxygen and silicate, but the variance in fluxes was larger and regression intercepts for the two treatments were not significantly different.