Concurrent measurements of soil and ecosystem respiration in a mature eucalypt woodland : advantages, lessons, and questions

Understanding seasonal and diurnal dynamics of ecosystem respiration (Reco) in forests is challenging, because Reco can only be measured directly during night-time by eddy-covariance flux towers. Reco is the sum of soil respiration (Rsoil) and above-ground respiration (in theory, RAG = Reco - Rsoil). Rsoil can be measured day and night and can provide a check of consistency on Reco, as the difference in magnitude and time dynamic between Reco and Rsoil should be explained by RAG. We assessed the temporal patterns and climatic drivers of Rsoil and Reco in a mature eucalypt woodland, using continuous measurements (only at night for Reco) at half-hourly resolution over 4 years (2014-2017). Our data showed large seasonal and diurnal (overnight) variation of Reco, while Rsoil had a low diurnal amplitude and their difference (Reco - Rsoil, or RAG) had a low seasonal amplitude. This result implies at first glance that seasonal variation of Reco was mainly influenced by Rsoil while its diurnal variation was mainly influenced by RAG. However, our analysis suggests that the night-time Reco decline cannot realistically be explained by a decline of RAG. Chamber measurements of autotrophic components at half-hourly time resolution are needed to quantify how much of the Reco decline overnight is due to declines in leaf or stem respiration, and how much is due to missing storage or advection, which may create a systematic bias in Reco measurements. Our findings emphasize the need for reconciling bottom-up (via components measured with chambers) and direct estimates of Reco (via eddy-covariance method).