Estuaries Vol. 22, No. 3A, p. 584-591 September 1999 Tower-Based Conditional Sampling for Measuring Ecosystem- Scale Carbon Dioxide Exchange in Coastal Wetlands J. L. HEILMAN 1 D. R. COBOS E A. HEINSCH C. S. CAMPBELl. K.J. MCINNES Department of Soil and Crop Sciences Texas A & M University College Station, Texas 77843 ABSTRACT: Long-term measurements of CO~ exchange between coastal wetlands and the atmosphere are necessary to improve our understanding of the role these ecosystems play in the global carbon cycle, and the response of these systems to environmental change. We conducted research to adapt and evaluate tower-based conditional sampling as a method for measuring net CO 2 exchange (NCE) at the ecosystem scale on a continuous basis. With conditional sampling, NCE is determined from the product of the standard deviation of vertical wind velocity, the difference in CO 2 concen- tration between updrafts and downdrafts in the constant flux portion of the boundary layer above the surface, and an empirical coefficient. We constructed a system that used a sonic anemometer to measure vertical wind velocity (w) and con~ol a high-speed three-way valve that diverted air from updrafts and downdrafts into separate sample lines, depending on the direction of m An infrared gas analyzer was used to measure the concentration difference. The conditional sampling system was installed and tested in a marsh in the Nueces River Delta near Corpus Christi, Texas, as part of a long-term study of effects of freshwater inflow on CO~ flux. System accuracy was evaluated by comparing conditional sampling measurements of water vapor flux with independent estimates obtained with the Bowen ratio method. Average daily flux estimates for the two methods agreed to within 13%. Measurements showed that freshwater inflow due to flooding of the Nueces River increased NCE by increasing CO~ assimilation and decreasing CO 2 efflux. Over a 65-d period, daily NCE varied from a maximum gain of 0.16 tool COz m -2 d -I during flooding to a maximum loss of -0.14 moi CO2 m 2 d x when the marsh dried. Our study showed that conditional sampling was well suited for quantifying CO 2 exchange in coastal wetlands on a diel, daily, and seasonal basis. Introduction The response of coastal wetland ecosystems to environmental change has received increasing at- tention in light of the continued rise in atmospher- ic CO2 concentration. Elevated CO2 concentration is expected to stimulate carbon assimilation (Drake and Leadley 1991; Gifford et al. 1996), though in- teractions between increased assimilation and cli- mate change could produce unexpected changes in the carbon balance of these systems. One of the most direct methods for evaluating ecological con- sequences of environmental change is measure- ment of net carbon dioxide exchange (NCE) be- tween the surface and the atmosphere (Koch and Mooney 1996). Long-term measurements of car- bon exchange are necessary to determine whether a given wetland ecosystem is a net source or sink of atmospheric carbon, and whether carbon source-sink relationships and primary productivity are altered by environmental change. Few such 1 Corresponding author; tele: 409/845-7169; fax: 409/845- 0456; e-mail: j-heilman@tamu.edu. measurements have been made in coastal wetlands (Happell and Chanton 1993; Pulliam 1993). The most comprehensive studies of net CO2 flux in a coastal wetland system have been conducted in the Chesapeake Bay, Maryland, using open-top chambers (Curtis et al. 1989; Drake and Leadley 1991; Drake 1992; Drake et al. 1996). Open-top chambers are a valuable tool, particularly for COx enrichment studies, but they fail to provide an en- vironment free of chamber effects. They are gen- erally warmer and more humid, with attenuated light and low wind speed (Drake et al. 1989). Chambers are small so that multiple chambers and scaling (Kim and Verma 1991) may be necessary to obtain an ecosystem-level flux. Meteorological measurements of CO2 flux in the constant flux portion of the boundary layer above the surface are an alternative to chambers for mea- suring NCE. Unlike chambers, meteorological measurements do not alter the physical environ- ment, and since the spatial scale of meteorological measurements is large (Schmid 1994), the mea- surements simultaneously integrate all sources, 9 1999 Estuarine Research Federation 584