Controls of riverine CO2 over an annual cycle determined using direct, high temporal resolution pCO2 measurements

Autonomous CO2 sensors were deployed in the Clark Fork River, Montana, USA, to characterize the partial pressure of CO2 (pCO(2)) during an annual cycle. A total of 23,941 measurements were made spanning the period 2002-2006. These data were compiled into a composite data set covering similar to 309 days, giving an unprecedented yearlong view of the carbon cycle dynamics of a riverine system. Seasonal pCO(2) varied from a winter minimum of similar to 100 mu atm to a fall maximum of similar to 900 mu atm. The pCO(2) changed by as much as 460 mu atm during a diel period, much larger than the range of the seasonal mean, in contrast to most other aquatic ecosystems where seasonal variability dominates. The diel pCO(2) amplitude was primarily controlled by the net ecosystem production (NEP) throughout the year, although heating/cooling and air-water exchange significantly altered the diel pCO(2) (and pH) magnitude. Although infrequent, rain events contributed similar to 21% to the cumulative short-term changes in inorganic carbon through CO2-enriched runoff. The seasonal cycle was controlled by temperature, NEP, and discharge. The Clark Fork River maintained pCO(2) levels that were supersaturated with respect to the atmosphere for the majority of the year. River-to-atmosphere CO2 gas exchange was estimated to be between 4.7 and 7.1 mol C m(-2) yr(-1). The loss of CO2 to the atmosphere arises from net heterotrophy that averaged 13.8 mmol m(-2) d(-1). The time series also captured important episodic events including macrophyte sloughing that led to a pulse of respiration that represented 7% of the annual CO2 gas efflux and cloudy periods that occurred every 7-18 days that dramatically decreased the pCO(2) through cooling.