Temperature sensitivity (Q10) of ecosystem respiration (Re) is a critical parameter for predicting global terrestrial carbon dynamics and its response to climate warming. However, the determination of Q10 has been controversial. In this study, we scrutinized the underpinnings of three mainstream methods to reveal their relationships in estimating Q10 for Re in the Heihe River Basin, northwest China. Specifically, these methods are Q10 estimated from the long-term method (Q10_long), short-term method (Q10_short), and the low-frequency (Q10_lf) and high-frequency (Q10_hf) signals decomposed by the singular spectrum analysis (SSA) method. We found that: 1) Q10_lf and Q10_long are affected by the confounding effects caused by non-temperature factors, and are 1.8 ± 0.3 and 1.7 ± 0.3, respectively. 2) The high-frequency signals of the SSA method and short-term method have consistent roles in removing the confounding effects. Both Q10_short and Q10_hf reflect the actual response of respiration to temperature. 3) Overall, Q10_long has a larger variability (1.7 ± 0.3) across different biomes, whereas Q10_short and Q10_hf show convergence (1.4 ± 0.2 and 1.3 ± 0.1, respectively). These results highlight the fact that Q10 can be overestimated by the long-term method, whereas the short-term method and high-frequency signals decomposed by the SSA method can obtain closer and convergent values after removing the confounding effects driven by non-temperature factors. Therefore, it is recommended to use the Q10 value estimated by the short-term method or high-frequency signals decomposed by the SSA method to predict carbon dynamics and its response to global warming in Earth system models.
Read full abstract