Abstract:

Objective: The objective of this study is to quantify the time-lag responses of net ecosystem exchange of CO₂ (NEE) to environmental factors in a Populus×euramericana plantation, in order to provid scientific evidence for improving the accuracy of estimating carbon source/sink in the ecosystem. Method: We used the eddy covariance technique and micrometeorological sensors to continuously measure NEE and environmental elements, such as air temperature (T_a), soil temperature at 5 cm depth (T₅), photosynthetically active radiation (PAR), vapor pressure deficiency (VPD) and soil volumetric water content at 25 cm depth (VWC₂₅), during the growing season (April-October) in 2014 in a P.×euramericana plantation adjacent to the Chaobai River in the northern China. We quantified the time-lag responses of NEE to these environmental factors by using regression analysis and wavelet cross-correlation analysis. Besides, partial correlation analysis and principal component analysis were used to evaluate the importance of time lag on the relationship between NEE and the environmental factors based on the determination coefficient (R²), Lin's concordance correlation coefficient (LCCC) and root mean square error (RMSE). Result: Our result showed that PAR was the dominant environmental factor regulating NEE. Additionally, VWC₂₅ had no significant effect on NEE (P=0.151 5) due to the high-level groundwater and sufficient water supply in this riparian ecosystem. Although NEE showed the same diurnal variation with PAR, it reached the peak 2.5, 2 and 2.5 h earlier than T_a, T₅ and VPD, respectively. Interestingly, there were no significant time-lag relationships between NEE and environmental factors at night. After eliminating the time lag, the partial correlation between NEE and T₅ was significant (P=0.0051) compared to the insignificant relationship before (P=0.224 8). Moreover, the principal component analysis indicated that the relationship between NEE and the climatic elements could be more accurately described by logarithmic regression model. After eliminating the time lag, the model R² and LCCC increased by 8.1% and 2.0%, respectively, and RMSE reduced by 0.417 μmol·m^-2s^-1. Conclusion: Therefore, the diurnal dynamics of NEE in this P.×euramericana plantation were controlled by T_a, T₅, PAR, and VPD with distinct time-lag phenomenon in the daytime. It is vital for eliminating time lag to evaluate the environmental effects on NEE and improve the model accuracy.

Key words: net ecosystem exchange of CO₂, environmental factors, time lag, wavelet analysis, Populus×euramericana plantation