The partitioning of net ecosystem exchange of CO2 (NEE) measured via the eddy covariance technique into gross photosynthesis (GPP) and ecosystem respiration (Reco) can be challenging and is often associated with unknown amounts of uncertainty. In order to estimate GPP from NEE, typically nighttime measurements of Reco are extrapolated to the day using a temperature function. Then Reco is subtracted from NEE to obtain GPP (i.e., GPP = NEE - Reco). However, recent evidence employing stable carbon isotopes of CO2 has shown that this practice can strongly overestimate GPP and Reco, from 10-100%. This is due to reduced leaf respiration during the day, aka the Kok effect. One key parameter for partitioning NEE using stable carbon isotopes is the isotopic composition of carbon respiration by the ecosystem, δ13Cr. δ13Cr is typically estimated using a Keeling plot with nighttime data either sampled via a profile method or single-inlet sampling where measurements are filtered by vertical wind velocity. In both techniques, nighttime measurements are assumed to contain information about the fractionated carbon sources from Reco, including heterotrophic respiration and autotrophic respiration (roots, stems and leaves). We propose to compare the two Keeling plot techniques (profile vs single inlet sampling) with a modeling approach where δ13Cr is estimated using nighttime and daytime measurements of the δ13C of CO2 emissions from leaves and profile measurements within the soil and at the soil surface. By sampling beneath the ground surface, isotopic measurements will be unaffected by wind patterns above ground thereby allowing more accurate measurements of the isotopic composition of soil respiration.
Preliminary data comparing these approaches in an alfalfa field suggests that the δ13Cr can vary substantially at the daily to sub-daily time step. For example, using the Keeling plot approach with a single inlet, δ13Cr varied over the course of 2 weeks by 5 per mil. We are currently evaluating the modeling and profile techniques in a restored wetland. These data will help assess uncertainty associated with the δ13Cr parameter thereby improving our ability to accurately partition NEE into Reco and GPP.