3.1. Rill Initiation Time with Different Crops Cover
The rill initiation time on tested slopes with millet, maize, and soybean are shown in
Table 3. Rill initiation times in the slopes of all three crops were greater than that on the smooth slope, in the order of millet (4.91 min) > soybean (4.11 min) > maize (3.35 min) > CK (2.67 min), which indicated that crops could significantly increase the time required to produce rills on sloping farmland, i.e., crops enhanced soil anti-scourability compared to the bare land. Notably, millet crop had the most enhanced soil anti-scourability, followed by soybean crop, with the least enhancement under maize crop.
Root systems did not always play a dominant role in the erosion process, as we observed overland flow in the early stages of the experiments when rills had not yet been formed. Therefore, we recorded runoff initiation time under different crop covers at the bottom end of the flume. According to the data, runoff initiation time was in the order of millet (3.99 min) > maize (2.98 min) > soybean (2.36 min) > CK (1.67 min). The results indicate that crops could delay runoff initiation time, which is consistent with the findings of previous studies [
39].
In addition, runoff initiation time in all slopes under the three crops were longer than the corresponding rill initiation times, which demonstrated that runoff initiation time was mainly influenced by the resistance effect of the slope surface on runoff, with no direct relationship with crop roots within the soil.
Table 4 lists aboveground plant part characteristics of different crops, including plant number, planting density, and average ground diameter. Considering the fact that stem cross-sections are generally circular, we found that the proportions of the cross-sectional areas of the different crop stems in the total slope area were in the following order: millet > maize > soybean, which is consistent with the order of runoff initiation time. In the case of overland flow, the larger the cross-sectional area of crop stems accounting for the slope area, the stronger the flow resistance effect of crops on the flow.
Figure 4 shows that rill initiation time in cropland with millet, maize, and soybean on 2° slopes were 10.47, 6.33, and 7.24 min, respectively, under a unit discharge rate of 0.1 L m
−1s
−1. When the slope gradient increased to 12°, rill initiation times in millet, maize, and soybean crops decreased to 4.64, 3.20, and 4.06 min, respectively. Such trends were also observed under other unit discharge rates, indicating that rill initiation time decreased with an increase in slope gradient. Moreover, differences in rill initiation time between millet, maize, and soybean crops, and bare surface on 2° slopes were 5.15, 1.01, and 1.92 min, respectively, under a unit discharge rate of 0.1 L m
−1s
−1, which decreased to 2.33, 0.89, and 1.75 min, respectively, on the basis of the 12° slope, indicating that the capacity of crops to hinder rill generation decreased with an increase in slope gradient, and unit discharge rate had a similar effect on the basis of a similar analysis.
Overall, the flow scouring effect was relatively low under low slope gradient and low unit discharge rate conditions. Conversely, under high slope gradient and high unit discharge rates, the scouring effect intensified, resulting in a reduction of rill initiation time and a decrease in the influence of crop cover on rill initiation time.
3.2. Correlations between Soil Detachment Rate, Slope Gradient, and Unit Discharge Rate
Slope gradient and unit discharge rate are two key variables that influence soil detachment.
Table 5 lists the detachment rates under different crop covers with different slope gradients and unit discharge rates. Soil detachment rates among different crop covers were significantly different (
p < 0.05) under different slope gradients and unit discharge rates.
Figure 5 illustrates how the two parameters influence soil detachment rate under different crop covers.
Generally, soil detachment rate increased with the rising of both slope gradient and unit discharge rate on all surfaces. For instance, the mean soil detachment rates under millet, maize, and soybean cover, and bare surface increased by 2.511, 5.711, 2.821, and 7.962 g m−2s−1, respectively, as slope gradient increased from 2° to 12°, and increased by 1.862, 2.879, 1.852, and 6.650 g m−2s−1, respectively, as unit discharge rate increased from 0.1 L to 0.3 L m−1s−1. The increasing effects of slope gradient and unit discharge rate on soil detachment rate were in the order of CK > maize > soybean > millet, implying that crops could attenuate the increasing soil erosion trends significantly with increases in slope gradient and unit discharge when compared with the bare surface. The weakening effect was the greatest under maize cover, and the least under millet cover.
We carried out multiple regressions to assess how slope gradient (
S) and unit discharge (
q) influence soil detachment rate (
Dr) with different crop covers. The regression equations are as follows:
The regression equations indicated that detachment rate was positively correlated with both slope gradient and unit discharge rate on all surfaces, including under crop cover and on bare surfaces, with the Dr-S and q relations being described as power functions. The coefficients of determination (R2) of all equations were high, indicating that soil detachment rate can be well predicted by a power function of slope gradient and unit discharge rate. In addition, the regression coefficients of slope gradient and unit discharge rate were quite close, and were 0.966 and 0.995 for millet, 1.350 and 0.982 for maize, and 0.946 and 0.944 for soybean, respectively, on slopes with millet, maize, and soybean, which indicated that both slope gradient and unit discharge rate have important effects on soil detachment rate in cropland, and the extents of the impacts were essentially similar. In addition, the regression coefficients of slope gradient and unit discharge were close to 1 under different crop covers, indicating that the effect of the change in slope and unit discharge is nearly linear in cropland.
In contrast, the regression coefficients of slope gradient and unit discharge rate on the bare surfaces were 2.024 and 2.106, respectively, which were nonlinear and significantly larger than those on cropped surfaces, indicating that the effects of the change in slope and unit discharge were related to root system but not crop density, and crops can weaken the increasing detachment rate trend with increases in slope gradient and unit discharge rate when compared with a bare surface. The results are consistent with our previous conclusion.
3.3. Soil Detachment Rate with Different Crop Covers
Across the four treatments, average soil detachment rate was the highest under CK (3.248 g m
−2s
−1), followed by under maize (2.498 g m
−2s
−1) and soybean (2.059 g m
−2s
−1), and the least under millet cover (1.782 g m
−2s
−1) (
Table 5). Soil detachment rates in crops on slopes were lower than that on the bare surface, with significant differences (
p < 0.05). The results are consistent with the findings of many previous studies [
41,
42,
43]. Moreover, soil anti-erodibility under various crop covers varied in the order of millet > soybean > maize. The results of the present analysis are based on a crop system, while the anti-scourability based on a single plant could present different results. According to the present experimental design, the planting densities of millet, maize, and soybean were 31.7, 3.7, and 11.0 individuals per square meter, respectively, with soil erosion rates in the order of maize (0.675 g m
−2s
−1) > soybean (0.187 g m
−2s
−1) > millet (0.056 g m
−2s
−1), as the difference of planting density of three crops was excluded. Consequently, not only the soil reinforcement capability of a single crop but also crop planting density influenced the erosion characteristics in croplands.
Root systems regulate soil erosion when rills are formed and when rill head gradually “cut down” as gully erosion intensifies. Both millet and maize have fibrous root systems, whereas soybean has a tap root system. Therefore, the effects of differences in root characteristics among different crops should be taken into account.
The main crop root system characteristics are listed in
Table 6. According to the root scanning results, average root diameter in the three crops was in the order of soybean (0.97 mm) > maize (0.85 mm) > millet (0.63 mm). In addition, RD, RLD, and RSAD of millet were the highest, followed by those of soybean and maize. Conversely, soybean had the largest RVD, and that of millet was slightly lower than that of soybean, with maize having the lowest RVD. According to the results of one-way ANOVA, root diameter and RVD had non-significant effects on soil detachment rate, whereas RD, RLD, and RSAD had significant effects (
p < 0.05) on soil detachment rate. However, some researchers have demonstrated that anti-scourability varies under different root diameter ranges [
44,
45]. Consequently, to further study how root characteristics influence soil detachment rate, we divided roots into diameter ranges at five intervals (0–0.5, 0.5–1, 1–2, 2–5, and >5 mm), and the root scanning data were reclassified and analyzed (
Figure 6). According to the results, there were still significant differences in root characteristics across different crops (
p > 0.05) in each of the root diameter intervals.
The root length densities of millet, maize, and soybean were the maximum in the 0–0.5 mm root diameter range, accounting for 65.4%, 61.9%, and 62.3% of the total root lengths, respectively. In addition, root length density was inversely proportional to root diameter in all three crops. Most crop roots were fine roots, with a root diameter of 0–0.5 mm. RSAD first increased and then decreased with an increase in crop root diameter. Specifically, millet RSAD was the largest (0.44 m2 m−3) in the 0.5–1 mm root diameter range, accounting for 33.6% of the total root surface area, and the smallest in the >5 mm root diameter range, accounting for 1.1% of the total root surface area. Soybean RSAD trends were largely consistent with those of millet, with the maximum (0.29 m2 m−3; 32.2%) in the 0.5–1 mm root diameter range, and the minimum (0.09 m2 m−3; 9.8%) in the range >5 mm root diameter range. In the case of maize, RSAD was the largest (0.21 m2 m−3; 34.3%) in the 2–5 mm root diameter range; nevertheless, the RSAD was not significantly different from those in other diameter intervals. In addition, RVD in both millet and maize increased first and then decreased with an increase in root diameter. RVD of millet in 2–5 mm root diameter range was the largest (96.04 cm3 m−3), accounting for 41.5% of the total volume, and the smallest in the >5 mm root diameter range, accounting for 4.4%. The largest RVD in maize (63.78 cm3 m−3; 45.1%) was observed in the 2–5 mm root diameter range, and the smallest (8.61 cm3 m−3; 1.7%) was observed in the 0–0.5 mm root diameter range. Unlike millet and maize RVDs, soybean RVD increased exponentially with an increase in root diameter, and the maximum in the > 5 mm range was 80.94 cm3 m−3 (32.6%), whereas the minimum in the 0–0.5 mm range was 17.22 cm3 m−3 (3.5%).
To investigate the potential influence of root characteristics under different diameter ranges on soil detachment rate, we conducted a correlation analysis between detachment rate and root parameter of tested crops using one-way ANOVA (
Table 7). According to the results, the correlations between all root parameters and soil detachment rate were negative, indicating that RLD, RSAD, and RVD were negatively correlated with soil detachment rate. Moreover, in the 0–0.5 mm and 0.5–1 mm root diameter ranges, the correlations between root parameters and soil detachment rate were highly significant (
p < 0.01). In addition, the correlations in the 0–0.5 mm root diameter range were more significant than those in the 0.5–1 mm root diameter range. In the 1–2 mm root diameter range, the correlations between soil detachment rate and RLD and RSAD were significant (
p < 0.05), although the correlation with RVD was not significant. In the 2–5 mm and >5 mm root diameter ranges, the correlations between soil erosion rate and RLD, RSAD, and RVD were not significant.
Data analysis showed that the correlations between crop root parameters and soil erosion rate were highly significant in the 0–1 mm root diameter range (
p < 0.01), and decreased gradually with an increase in root diameter, which indicates that crop roots in the 0–1 mm diameter range could significantly influence soil erosion in cropped slopes, as well as improve cropland anti-erodibility capacity. Such a finding has also been reported by Wang [
46]. Among the root parameters examined, soil detachment rate was more sensitive to RLD and RSAD than to RVD.