Stream oxygen flux and metabolism determined with the open water and aquatic eddy covariance techniques

We quantified oxygen flux in a coastal stream in Virginia using a novel combination of the conventional open water technique and the aquatic eddy covariance technique. The latter has a smaller footprint (sediment surface area that contributes to the flux; 10 m), allowing measurements to be made at multiple sites within the footprint of the open water technique ( 1000 m). Sites included an unvegetated stream pool with cohesive sediment, a macrophyte bed with sandy sediment, and an unvegetated sand bed with rippled bedforms. Nighttime eddy covariance oxygen uptake was always smaller than uptake produced by the open water technique. At the pool and unvegetated sand bed sites, nighttime eddy covariance uptake was 20-fold smaller than open water uptake. At the macrophyte bed site, gross primary production quantified with the two techniques was similar but eddy covariance uptake was 2.4-fold smaller. The difference in oxygen uptake between eddy covariance and open water techniques could not be accounted for by uncertainties in the gas transfer velocity but could be accounted for by anoxic groundwater inflow through stream banks outside of the eddy covariance footprint. Nighttime oxygen uptake was also measured with eddy covariance in a tidal freshwater part of the stream, where pore space in the sandy sediment near the sediment–water interface was flushed with stream water at peak water velocities. As a result of this advective hyporheic exchange, nighttime oxygen flux increased fourfold with a doubling of water velocity. The recent discovery that inland waters may store or transform over half of net terrestrial ecosystem production en route to coastal oceans (Cole et al. 2007) has highlighted the contribution of inland waters to global carbon transformation. Low-order streams may be disproportionately important in this regard due to the large surface area that they represent, their high carbon dioxide concentrations, and the high gas transfer velocities that characterize them (Butman and Raymond 2011; Raymond et al. 2013). However, the contribution of in situ organic carbon mineralization to carbon dioxide evasion from streams and rivers is not always well constrained and appears to vary markedly from watershed to watershed. For example, while most of the dissolved organic matter that enters the upper Hudson River is mineralized to CO2 during transport toward the ocean (Cole and Caraco 2001), in situ mineralization in the Ottawa River may be insignificant (Telmer and Veizer 1999). Measurements of oxygen flux are integral to our understanding of carbon mineralization in rivers and streams (e.g., Battin et al. 2008) and one of the primary oxygen flux techniques applied to rivers and streams is the open water, or diel change technique (Odum 1956). The technique has been applied to quantify fluvial carbon transformation across biomes and across stream and river orders, where the greatest areal rates of oxygen uptake are observed in small streams (Battin et al. 2008). Because oxygen transformation in the water column of streams is very small relative to benthic rates, benthic flux largely determines stream ecosystem metabolism (Minshall et al. 1983). The vertical flux of oxygen across the sediment–water interface reflects the contributions of oxic metabolic processes and the reaction of oxygen with the reduced products of anoxic metabolic processes. Groundwater inputs, however, can complicate the interpretation of oxygen fluxes. As a result of groundwater inflow, oxygen balances in streams and rivers may reflect not only in situ metabolic processes but also organic carbon mineralization in soils and groundwaters (McCutchan et al. 2002; Hall and Tank 2005). To improve predictions of organic carbon mineralization in streams and rivers the controls and drivers of aquatic mineralization and primary production need to be better constrained. A proximal driver of metabolism in streams is water velocity (Odum 1956). For cohesive sediments, oxygen consumption is expected to increase linearly with water velocity where the rate of organic matter mineralization is limited by oxygen transport through the diffusive boundary layer (Nakamura and Stefan 1994). For permeable sediments, an *Correspondence: [email protected] 1 LIMNOLOGY and OCEANOGRAPHY Limnol. Oceanogr. 00, 2015, 1–25 VC 2015 Association for the Sciences of Limnology and Oceanography doi: 10.1002/lno.10103 increase in overlying water velocity stimulates hyporheic exchange, the advective exchange and mixing of pore water with overlying water. This mechanism is highly efficient in the mineralization of organic carbon (Huettel et al. 2003; Berg et al. 2013) and is one of the primary mechanisms of organic matter mineralization in streams (Grimm and Fisher 1984). Hyporheic exchange scales with sediment permeability and the overlying water velocity squared (Packman and Salehin 2003), and thus varies predictably over stream pools and riffles (Pusch 1996). The open water technique quantifies oxygen flux under in situ hydrodynamic conditions, but the footprint (the sediment area that contributes to the flux) in streams and rivers is commonly several hundreds to thousands of meters in length (Reichert et al. 2009). The footprint of the open water technique is described according to Reichert et al. (2009) as