Increasing importance of ecosystem respiration for ecosystem carbon exchange dynamics from weekly to interannual timescales

Abstract

 Vegetation carbon uptake is a major sink for anthropogenic greenhouse gas emissions, yet inferring longer-term behaviour of ecosystems as carbon sinks or sources is still difficult. Here, using a time series decomposition technique and eddy covariance data, we show that while at subseasonal scales net ecosystem CO2 uptake (NEP) is closely related to photosynthetic uptake, there is an increased importance of ecosystem respiration for determining NEP on longer time scales.  The interannual evolution of net ecosystem CO2 uptake (NEP) is insufficiently understood and often not well captured in state-of-the-art vegetation models and data products. This lack of understanding may in part be due to different drivers between interannual and seasonal scales affecting the two terms balancing NEP - photosynthetic uptake of CO2 (GPP) and ecosystem respiration (Reco).Here, we extract timescale specific carbon flux dynamics at 20 long-running FLUXNET eddy covariance sites (>13 years) using time series decomposition to relate variability in GPP and Reco to NEP at subseasonal to interannual timescales. The results indicate that relations between NEP and GPP or Reco, respectively, are not constant across different timescales, but that GPP and Reco exert differential control on NEP between sub-seasonal, seasonal, and longer timescales. Overall, the fraction of variance in NEP explained by GPP variance is decreased at longer timescales, while the fraction of NEP variance explained by Reco variance and by the covariance of GPP and Reco generally increases at longer timescales. Regarding GPP, we find that the slopes between NEP and GPP, which could be interpreted as scale specific apparent carbon use efficiencies (NEP/GPP) are highest and most consistent at subseasonal scales, while generally smaller in magnitude and less constrained at interannual scales. This indicates that GPP and NEP are generally more strongly and directly linked at the subseasonalscale. Regarding Reco, we find a positive relationship between NEP and Reco at the seasonal scale. This is counterintuitive given NEP = GPP – Reco, but similar to spatial relations in other studies and likely related to GPP seasonality as a common driver. In contrast, the subseasonal and interannual NEP-Reco relations are mostly negative, as would be expected since higher respiratory loss would generally indicate lower ecosystem carbon retention, i.e. lower NEP.  The timescale specific relations extracted here based on direct ecosystem CO2 exchange measurements suggest an increased importance of ecosystem respiration for long-term carbon source or sink behavior for some ecosystems. These results give insight into ecosystem functioning, as well as demonstrate the utility of time series decomposition as a diagnostic of ecosystem dynamics at different timescales. Such information may eventually serve as a basis to infer turnover times of ecosystem carbon pools and better characterize interannual ecosystem carbon dynamics.