Response of soil moisture to vegetation and trade-off analysis in the hilly area of the Loess Plateau, China

It is said that in a synergistic relationship between vegetation and soil moisture (SM), the latter may be consumed excessively, while at the same time improving the ecological environment, such as in the large-scale artificial vegetation planting areas on the Loess Plateau. At present, the possibility of vegetation restoration and its impact on the water ecology of the Loess Plateau have received extensive attention. In this study, piecewise linear regression (PLR) and nonparametric conditional quantile regression (NCQR) models have been innovatively applied to vegetation survey data and soil moisture data from 50 field samples to explore the trade-off relationship between soil moisture and various vegetation parameters (VPs) and the response of soil moisture to the vegetation. The results showed that precipitation and topography are significantly linearly related to soil moisture in the sampling plots. Affected by irrational planting, the soil moisture benefit at 20–40 cm is significant weaker than that at 0–20 cm (plot 4, 5, 26 and 44). Some sampling plots excessively consumed soil water to promote vegetation planting. Although there was stronger soil water infiltration by shrubs, it was noted that the vegetation planting plan in some grass plots and shrub-grass plots were unreasonable and needed to be adjusted due to the clear risk of a soil water deficit determined by the trade-off values (represented by the root mean square deviation), plant species, a multiparameter evaluation. With the increase in soil moisture in the sampling plots, the negative correlation between the trade-off (soil moisture and vegetation) and soil moisture was transformed into a positive correlation (the inflection point was 8.5–10%). In addition, the benefit of soil moisture gradually changed from “shortage” to “appropriateness” and even “excess” (the inflection point was 15–20%). The fitting results of the two models are optimal and show clear applicability for evaluating trade-off relationships. Among the VPs involved in the calculation, the Shannon–Wiener index (H), Margarlef index (D1) and Fractional Vegetation Cover index (FVC) demonstrated better regional representation and stability.