Differentiating refilling and transpiration from night-time sap flux based on time series modelling

Key message Although night-time transpiration occurs most of the time, stem refilling contributed more to the amount of night-time sap flux, which indicates that there was no water loss in Schima superba at night-time. It has been historically shown to be difficult to identify the night-time transpiration (E-n) and stem refilling (R-n) components from night-time sap flux (NFt). We applied an autoregressive moving-average (ARMA) model with exogenous variables (ARMAX) fitting E-n to distinguish NFt of Schima superba for avoiding autocorrelation. In total, 23 optimum models were chosen at the node moments of the night in the dry and wet seasons. Models performed seasonal variations in involving environmental factors and build time. Vapor pressure deficit (VPDt) alone or with wind speed (WSt) drove positively NFt in most of the time in both seasons. SMt dominated NFt only at the beginning of the night-time in the dry season. E-n occurred 1 h later and was lower in the wet than in the dry season (ca. 1.09 kg h(-1) vs 1.82 kg h(-1)), and R-n is the opposite (ca. 2.36 kg h(-1) vs 1.9 kg h(-1)). This may cause by greater water storage deficit in the trunk due to stronger day-time transpiration. We found that NFt was minor compared to day-time sap flux (DFt), and the mean ratio of night-time sap flow (Q(n)) to daily sap flow (Q(w)) was only 0.02. Our results showed that there were no seasonal differences on the contribution of NFt to the 24-h F-t, and no water loss at the daily scale in Schima superba in both seasons (R-n > E-n). This study first quantifies E-n and R-n, and reveals seasonal variations in tree night-time water use and provides a basis for better understanding of E-n function.