2.8. Characterizing the soil water use pattern
Soil water use by jujube trees in a given layer during the growing season was quantified on the basis of daily changes in soil water storage (SWS) in that layer (Broedel et al., 2017; Christina et al., 2017) because the groundwater is far beyond the plant’s rooting depth at our site. The SWS for each layer was computed by multiplying the mean soil water content in this layer () by the layer depth (Di ). In the 0 layer, the soil water content at 10 cm was used as the mean soil water content. For other depth intervals, the mean soil water content was defined as the arithmetic mean of soil water content at the upper and lower depths. Soil water loss in the 0 layer is generally the result of soil evaporation (He et al., 2013) although this may be small in mulched plots. Nevertheless, in order to ensure consistency between the different plots, the SWS in the 0-10 cm layer was excluded when quantifying soil water use here. We then computed daily changes () in SWS at eight depth intervals (10-20, 20-40, 40-60, 60-100, 100-160, 160-220, and 220) during the growing season in each year. A negative indicates the use of soil water by jujube trees. The sum of negativein one layer over the growing season was defined as the total water uptake during the growing season from that layer.
Because of the non-uniform depth intervals between the different layers considered, the concept of depth density of water use (DWU) was introduced in order to compare soil water use at different depths. The DWU is defined as the amount of water use per centimeter depth. Furthermore, we calculated the proportion of water use source (PWU) for the shallow (10) and deep (100-280 cm) layers relative to those of the profile as a whole (10) in different years and growing periods to reveal temporal patterns at annual and seasonal scales.
One-way analysis of variance (ANOVA) was used to examine the impacts of combinations of mulching and terracing on SWS, transpiration and mid-day leaf water potential, using the nlme package in R (v.3.5.3, R Core Team 2019). The graphs presented in this paper were drawn using Origin 2016 (OriginLab, Massachusetts, USA).