4. Discussion
The current study found that the SDC values of sand soil under all four treatments were higher than those of loessal soil. Previous studies have similarly found that SDC may be influenced by soil type (Su et al., 2014). Line and Meyer (1989) showed that resistance of soil to erosion increased with increasing soil clay particle content. In the current study, the clay contents of sand soil and loessal soil were 6.82% and 17.59%, respectively. Thus, it may be concluded that the lower SDC for loess soil was due to its higher clay content. However, the response of SDC of sand soil to hydraulic parameters was weaker compared to that of loessal soil. In particular, there was a weak relationship between the hydraulic parameters and the SDC of sand soil with roots. Thus, the SDC of soil may be strongly affected by the root system distribution and soil properties. Soil particles bounded by the root system are likely to be strongly consolidated and therefore would not easily be detached by water erosion, whereas soil not consolidated by the root system would be highly susceptible to erosion. In addition, the root system increases the roughness of the soil surface, thereby reducing the energy of runoff, decreasing flow velocity and increasing soil water infiltration (Zheng et al., 2009). Soil that is not consolidated by the presence of a root system experiences stronger scouring erosion, resulting in the rapid detachment of erodible soil particles. Since the root system ofB. ischaemum is mainly distributed in the surface soil, the roots of this plant protect the surface soil as well as the soil in deeper layers. Under this scenario, there would be little variation in SDC with increasing flow intensity until such a time that the flow is sufficiently strong to dislodge the surface soil layer bounded by the concentrated root system. Therefore, predicting the SDC of sand soil containing a root system requires both hydraulic parameters as well as a parameter reflecting root weight. However, due to the higher clay content and lower erodibility of loessal soil, the soil detachment characteristics of loessal soil both with and without a root system are similar. Hence, models that considered only hydraulic parameters were better able to predict the SDC of loessal soil with roots.
Although the SDC of soils both with and without roots increased under the freeze-thaw effect, the effect was not significant. This result is consistent with that of previous research (Sun et al., 2018). The stability of soil aggregate is an effective index to measure the SDC of soil subjected to freeze-thaw (Bryan, 2000; DiAz-Zorita et al., 2002) as the soil aggregate is broken down by successive freeze-thaw cycles (Lehrsch et al., 1991; Oztas and Fayetorbay, 2003), and it has been shown that soil SDC is increased with the destruction of soil macro-aggregate (Edwards, 1991). However, once the number of freeze-thaw cycles exceed 5, factors such as flow and slope become increasingly important for explaining the contribution of freeze-thaw to soil SDC. This is why the contribution of freeze-thaw to the SDC was not significant (Sun et al., 2018). The clear consolidation effect of the root system on soil results in a significant negative correlation between root density and SDC (Wang et al., 2014), indicating that the root system reduces the SDC of soil significantly. The present study found that SDC of both unfrozen soil and freeze-thaw soil was reduced by 97.82–99.66% due to the presence of a root system. This result is supported by the research results of Gyssels et al. (2006), who found that the roots of cereal plants reduced SDC by > 90%. The main reason for root systems reducing SDC is due to consolidation of soil particles and the increase in soil adhesion and stability by the presence of roots (Baets et al., 2006, 2007). A second reason is the influence of the root system on increasing soil infiltration capacity by increasing the soil surface roughness and thereby reducing the velocity of surface runoff (Gyssels et al., 2006; Viles, 1990). Thus, the SDC would be decreased by root system.
Although soil detachment was increased and decreased by freeze-thaw and the root system, respectively, the inhibition effect of the root system dominated when the effects of freeze-thaw were combined with that of the root system. The SDC under the effect of freeze-thaw in combination with the root system was 5 times higher than that of soil under only the influence of the root system, but 295 times lower than that of the soil affected only by freeze-thaw. On the one hand, the root system significantly increased resistance to overland flow velocity during the process of erosion, thereby weakening the SDC driven by stream scour. On the other hand, the effect of soil disturbance through the freeze-thaw cycle was weakened by the root system to some extent (Gao et al., 2015). The freeze-thaw process has been shown to weaken bare soil by 20.6%, whereas it only weakened soil containing a root system by 7.3% (Li et al., 2012). Meanwhile, the present study found that combined factors of freeze-thaw and the root system explained 36.9% of SDC variation, whereas freeze-thaw as a single factor only explained 10.64% of SDC variation. This result suggests hints at the complexity of the composite effect of freeze-thaw and the root system on soil detachment and the need for further research.
Shear stress and stream power had positive linear relationships with SDC, whereas unit energy of the water carrying section and unit stream power showed power function relationships with SDC. This result is consistent with the previous research (Wang et al., 2016; Zhang et al., 2002, 2003, 2008). The present study found that stream power was the hydraulic parameter best able to explain SDC under the effect of freeze-thaw and the root system. This result is consistent with the findings of Cao et al. (2009) and Su et al. (2014). Since the root system is an important parameter that cannot be ignored when representing the process of soil detachment within model development (Wang et al., 2016), the present study developed an SDC prediction model for different soil types and treatment measures based on stream power and root weight. While the model developed in the present study is similar to those developed in previous studies (Li et al., 2015; Wang et al., 2014), some important differences exist. First, since soil parameters were not considered in the present study, only root weight and stream power were included in the prediction model. Secondly, the present study improved the regression relationship between the root system and SDC as an improved correlation between the SDC and the square of root weight was obtained. This result may be related to change to soil permeability resulting from the root system (Gyssels et al., 2006). Although the presence of a small root system in the soil limits the consolidation effect of the root system on soil particles, mechanical interludes of the root system can destroy soil structure, thereby increasing water infiltration. This can result in the collapse of the soil structure, particularly in low-viscosity soils such as sand, resulting in an increase in stream scour-driven SDC after water infiltration volume exceeds a certain critical value. However, an increase in the root system above a certain threshold density will result in great improvement in the soil consolidation capacity of the root system, thereby enhancing soil cohesion and reducing the detachment and transport capacity of stream scour on soil particles and reducing SDC (Xu et al., 2019). Therefore, the present study identified a prediction model based on the square of root weight and stream power that had a higher prediction accuracy compared to models developed in previous research (NSE = 0.8488) (Li et al.,2015). The model developed in the present study could provide accurate prediction for SDC under different conditions in the seasonal freeze-thaw region. Of course, further field tests are required to verify the practicability of this model in areas subject to seasonal freeze-thaw.