Reconciling top-down and bottom-up estimates of ecosystem respiration in a mature eucalypt forest

Ecosystem respiration (Reco) arises from interacting autotrophic and heterotrophic processes constrained by distinct drivers. Here, we evaluated up-scaling of observed components of Reco in a mature eucalypt forest in southeast Australia and assessed whether a land surface model adequately represented all the fluxes and their seasonal temperature responses. We measured respiration from soil (Rsoil), heterotrophic soil microbes (Rh), roots (Rroot), and stems (Rstem) in 2018-2019. Reco and its components were simulated using the CABLE-POP (Community Atmosphere-Biosphere Land Exchange-Population Orders Physiology) land surface model, constrained by eddy covariance and chamber measurements and enabled with a newly implemented Dual Arrhenius and Michaelis-Menten (DAMM) module for soil organic matter decomposition. Eddy-covariance based Reco (Reco.eddy, 1,439 g C m−2 y−1) was slightly higher than the sum of the respiration components (Reco.sum, 1,295 g C m−2 y−1) and simulated Reco (1,297 g C m−2 y−1). The largest mean contribution to Reco was from Rsoil (64%) across seasons. The measured contributions of Rh (49%), Rroot (15%), and Rstem (22%) to Reco.sum were very close to model outputs of 46%, 11%, and 22%, respectively. The modeled Rh was highly correlated with measured Rh (R2 = 0.66, RMSE = 0.61), empirically validating the DAMM module. The apparent temperature sensitivities (Q10) of Reco were 2.22 for Reco.sum, 2.15 for Reco.eddy, and 1.57 for CABLE-POP. This research demonstrated that bottom-up respiration component measurements can be successfully scaled to eddy covariance-based Reco and help to better constrain the magnitude of individual respiration components as well as their temperature sensitivities in land surface models.