Soil carbon dioxide (CO2) flux is a main determinant of net ecosystem carbon exchange. Providing better insight into the variation of soil respiration (Rs) at the diurnal scale and its response to environment factors is of crucial importance for predicting soil CO2 flux and carbon-climate feedbacks. Here we investigated the diel variation in the Rs using 2004 – 2013 continuous Rs measurements (at half-hourly intervals) in a temperate deciduous broadleaf forest. We employed the Projection to Latent Structures (PLS) modeling to examine which environmental factors dominate the daytime and nighttime Rs flux. The PLS has the capability to avoid model overfitting in the presence of large number of independent variables and multicollinearity, which ensuring the modelling-involved variables independence, meanwhile, reflecting the distinct among them.
Results/Conclusions
Dynamic and similar diel variations of Rs were observed in this forest ecosystem with fluctuating from the minimum at about 6:00—9:00 h to the maximum at 16:00—18:00 h across different months. The nighttime Rs fluxes (2.12±1.54 μmol m−2 s−1) were not statistically different from the daytime Rs (2.10±1.50 μmol m−2 s−1). We also derived that the Rs flux measured at 12:00−13:00 h proffer the lowest bias to daily mean Rs flux. Further analysis using Projection to Latent Structures (PLS) modeling reveals that environmental factors indirectly influence the Rs by altering vegetation productivity (reflected by LAI) in daytime, whereas the Rs variation in the nighttime was the result of direct influence of temperature. Our study reveals differential causal relationship between Rs and its influencing factors during daytime and nighttime, which is of great implications for understanding the variation and controls on Rs at the diel timescale in temperature forests.
