Ecosystem respiration (RE), the release of carbon from an ecosystem into the atmosphere, is a key component of the terrestrial ecosystem carbon budget and plays an important role in the global carbon balance. Analyzing the interaction between RE components is essential to understand RE and to accurately evaluate the ecosystem carbon budget. There are many methods for separating RE into autotrophic respiration (Ra) and heterotrophic respiration (Rh), but each approach has disadvantages. Large RE data have obtained through long-term eddy covariance measurements, while the interaction between Ra and Rh is poorly documented, which inhibits the accurate assessment of global carbon budget. In this study, we used an empirical statistical method to separate RE into its two components and to examine component relationships and seasonal dynamics at three ChinaFLUX sites: 1) Changbaishan temperate mixed forest (CBS); 2) Dinghushan subtropical evergreen broad-leaf forest (DHS); and 3) Haibei shrub meadow (HBGC). The applicability and sensitivity of this method in typical ecosystems of China were also evaluated. The method used in this study was based on the ratio of Ra to RE (Ra/RE). The range of Ra/RE was obtained by calculating two ratios: the ratio of RE to net ecosystem productivity (NEP) (RE/NEP) and that of net primary productivity (NPP) to gross primary productivity (GPP) (NPP/GPP). Within the range of Ra/RE, 1000 Ra/REs were randomly selected and the value of Ra/RE used in this study was set as the mean of the 1000 random Ra/REs. Ra and Rh were then calculated using Ra/RE and RE.
Our study showed that the RE separating method produced consistent results with those obtained through static chamber/gas chromatographic techniques at the same sites, as well as with biomass surveys and theoretical speculation. The interaction of RE components was sensitive to the variation of NPP/GPP: a ten-percent increase of NPP/GPP led to a five-percent decrease of Ra/RE. In all three ecosystems, RE and its components showed similar seasonal dynamics, with a single-peak pattern achieving its maximum midway through the growing season. The ratio of Rh to RE (Rh/RE) also showed different seasonal dynamic among the three ecosystems. In CBS, Rh/RE increased during the first half of the year, reached its peak during the growing season then decreased. However, Rh/RE in HBGC decreased during the first half of the year and increased again later in growing-season. In DHS, Rh/RE was relatively stable at approximately 0.5. Seasonal dynamics of Ra/RE were opposite to those in Rh/RE. The annual total Rh accounted for 60% of the RE in HBGC, suggesting that a large proportion of emitted carbon was released by Rh in this ecosystem. In CBS and DHS, Rh was only 49% of RE, indicating that the release of carbon through Ra and Rh was nearly the same in these two forest ecosystems. Results indicate that this statistical method, which requires detailed observations of NPP/GPP, can successfully separate RE into Rh and Ra and can provide necessary data for the detailed analysis of the ecosystem carbon cycle.