Study area

Longji Rice Terraces (109°32'–110°14' E, 25°35'–26°17' N) is located in Longji Town, northwest of Guilin City, Guangxi, China. It is a field developed on Longji Mountain, which belongs to the Yuecheng Mountain range and is the northwest branch of the Nanling Mountain system. The Longji Rice Terraces mainly include Longji Guzhuang Village terraced fields (LGVTF), Ping’an Zhuang terraced fields (PZTF), and Jinkeng Red Yao terraced fields (JRYTF). Longji Rice Terraces are in a subtropical monsoon climate zone affected by the convergence of cold air from the north and warm and humid air from the south, with more frontal rain and fog weather and an annual precipitation was 1800–1900 mm. The elevation of the Longji Rice Terraces is 300–1100 m. Longji Mountain is covered with lush vegetation, so it can store water well. The upper layer of Longji Mountain is permeable sandy loam, which can store water, and the lower layer is impermeable sedimentary rocks, shale bedrock, and slate with good water impermeability, located approximately 1.5–2 m below the surface soil. The excess rainwater that infiltrated the sandy loam flowed out of the cracks in the mountain and down to each terrace. The farming methods of the Longji Rice Terraces are primitive pair plowing and traditional flood irrigation.

Sample collection

Soil samples were collected from Longji Rice Terrace in November 2021. The sampling locations and information are shown in Fig. 1 and Table 1, respectively. The sample IDs, beginning as “P” and “D”, were taken from PZTF and JRYTF, respectively. The sampling sites on the PZTF (open triangles and pentagons) and JRYTF (solid triangles) are shown in Fig. 1. Triangles and pentagons represent the surface soil and core sites, respectively. Red and yellow lines represent the sampling sites along the slope direction and contour lines, respectively. Fourteen paddy field surface soil samples (0–5 cm, marked by open triangles) and two paddy field soil cores (P1#1 with a depth of 50 cm and P1#2 with a depth of 40 cm, marked by open pentagons) were collected from the PZTF, and 19 paddy field surface soil samples (0–5 cm, marked by solid triangles)) were collected from the JRYTF. Surface soil samples were collected along a specified slope direction. Samples from the PZTF and JRYTF were collected from high and low elevations along the southwest and southeast slopes, respectively. All soil cores were sliced at 5 cm intervals.

Fig. 1Full-screen View

(Color online) Map showing the soil sampling locations of Longji Rice Terraces in the northwest of Guilin, Guangxi, South China. Triangles and pentagons represent sites of surface soils and cores, respectively. The red lines represent the sampling sites along the slope direction. (a) PZTF. The yellow line represents the contour line at an elevation of 860 m. (b) JRYTF. The yellow line represents the contour line at an elevation of 970 m.

Sample preparation and measurement

The samples were heated at 100 ℃ for 12 h until dryness, then crushed and sieved. After that, the ¹³⁷Cs activities of the 50 g fine soil samples were measured by an γ spectrometer equipped with a high-purity germanium detector and a lead shielding chamber at a time of 86400 s. The detailed methodology and spectrometer are described in our previous work [37, 38, 39]. Briefly, A HPGe detector (GEM-C5970; ORTEC, USA) with a counting efficiency of 45% and FWHM of 1.7 keV on 1.33 MeV energy of ⁶⁰Co was used. ¹³⁷Cs content was determined by detecting the 661.62 keV signature γ-ray. The detection limit for ¹³⁷Cs was 0.04 mBq/g. 1 g of pretreated soil sample was calcined at 550 ℃ for 12 h to determine the organic matter content. The organic matter content was calculated as the mass loss before and after the ashing procedure, and was expressed as the percentage loss on ashing for the soil sample.

2.5 g soil sample was weighed for plutonium isotope analysis. Details of the chemical separation and purification processes have been described in previous studies [37, 38, 39]. In brief, 1.82 pg ²⁴²Pu was added as a chemical yield tracer to monitor and calculate the chemical recovery of the plutonium isotope analysis, the main analysis processes including digestion, filtration, adjustment of acidity and valence, purification of plutonium with AG1-X8 anion exchange resin and re-purification of plutonium with AG MP-1M anion exchange resin. Finally, the concentrated plutonium solution was placed in a 1.5mL sample bottle with 4% nitric acid and carried to the College of Radiology and Protection, Soochow University, for measuring ^239+240Pu activity and ²⁴⁰Pu/²³⁹Pu atom ratio using sector-field ICP-MS (Finnigan Element 2, Bremen, Germany). The chemical recovery of Pu ranged from 60% to 80%, and the limits of detection for ICP-MS calculated from 3 times the standard deviation of the blank solution (4% HNO₃) were 0.09 fg/mL for ²³⁹Pu and 0.14 fg/mL for ²⁴⁰Pu with a counting time of 150 s. The concentration of plutonium isotopes was calculated by subtracting the blank from the measured results of samples, and the measured signal intensities of ²³⁹Pu and ²⁴⁰Pu in procedure blank were 8.5 cps and 6.4 cps, respectively.

Activities of 239+240Pu and 137Cs in soil samples

3.1.1

240Pu/239Pu atom ratios, activities of 239+240Pu and 137Cs in surface soil samples

The mean atom ratio of ²⁴⁰Pu/²³⁹Pu in the surface soil samples was 0.195±0.014 (n=33), ranging from 0.162±0.013 to 0.232±0.033 (Table 1), which agreed well with the value of the global fallout (0.178±0.023 (0–30°N)) [40]. This result is consistent with those of previous reports [41, 42, 43, 44, 45, 46]. The result is higher than the ²⁴⁰Pu/²³⁹Pu atom ratio for nuclear weapons-grade plutonium (0.01–0.07, e.g., 0.028-0.037 for the Nagasaki atomic bomb explosion) [47, 48], and lower than those for the Chernobyl nuclear accident (0.403-0.412) [49] and Fukushima nuclear accident (0.303-0.330) [9], indicating there is no regional radioactive contamination in the studied area, so all the Pu and Cs are sourced from the global fallout, then they can be used as tracers for soil erosion research.

The activities of ^239+240Pu and ¹³⁷Cs in surface soil samples are listed in Table 1. It can be seen that the activities of ^239+240Purange from 0.089±0.007 to 0.734±0.046 mBq/g, with a mean value of 0.264 ± 0.132 mBq/g (n=33), similar to that of surface soils (0.365 ± 0.014 - 0.819 ± 0.066 mBq/g) from Shaodong of Hunan Province in China [43], and within the data range (0.005–1.988 mBq/g) of Chinese surface (≤ 10 cm) soil [50]. ^239+240Pu activities in the surface soil of PZTF (0.089 ± 0.006-0.335±0.011 mBq/g, with a mean of 0.253±0.064 mBq/g)and JRYTF(0.089±0.007-0.734±0.046 mBq/g, with a mean of 0.273±0.168 mBq/g) are similar. The activities of ¹³⁷Cs in the surface soil samples range from 1.80±0.13 to 7.88±0.28 mBq/g, with a mean value of 4.67±1.38 mBq/g (n=33) which was slightly higher than that of 3.39 mBq/g from Puding County of Guizhou Province [51]. Because all samples were collected from a limited spatial range with a diameter of 1000 m, the large variation in ¹³⁷Cs activity indicates local alteration after deposition.

As shown in Fig. 2a, the activity of ^239+240Pu in the surface soils of the PZTF first increased and then decreased from an elevation of 930 m to 790 m. The maximum value of PZTF appeared at elevations of 860 and 870 m. Fig. 2b indicates that ^239+240Pu activities in the surface soils of JRYTF, except for the samples at the highest elevation of 1080 m, also showed the same trend as PZTF. The maximum value of the JRYTF appeared at an elevation of 890 m. The same trend in ¹³⁷Cs activity is observed in Fig. 2. Similarly, the ¹³⁷Cs inventories of sloping farmland in the southeast of the Qinghai-Tibet Plateau showed a fluctuating increasing and decreasing trend along elevation [25]. All possible factors affecting plutonium and ¹³⁷Cs concentrations included soil texture, organic matter content, soil moisture, particle size, elevation, and slope gradient. All samples were collected from a small area of rice fields under the same basic environmental conditions, which had the same soil texture, soil moisture, and grain size; therefore, these factors could be excluded. Elevation was also not a major factor affecting the plutonium and ¹³⁷Cs concentrations in the studied area because the elevation span of the studied sites was within 300 m, and the concentrations of plutonium and ¹³⁷Cs did not increase or decrease monotonically with elevation. A strong correlation between ³⁹⁺²⁴⁰Pu activity concentration and organic matter content was reported in our previous study [39] and by other researchers [52]. However, the surface soil samples of Longji Rice Terraces were all collected from rice paddies with little variation in organic matter content (Table 1), suggesting that landform is more likely to be the main factor in ^239+240Pu activity variation relative to organic matter. Landform factors, such as slope gradients, are among the most important factors affecting the spatial variation in soil erosion rates [53, 54]. Fig. 3a shows the distributions of ^239+240Pu and ¹³⁷Cs in surface soils at an elevation of 860 m in the PZTF. The activities of ^239+240Pu and ¹³⁷Cs were 0.215±0.016-0.335±0.009 mBq/g and 3.40±0.18-6.31±0.25 mBq/g respectively, with the mean values of 0.270±0.046 mBq/g (n=7) and 5.27±1.09 mBq/g (n=7). Fig. 3b shows the distributions of ^239+240Pu and ¹³⁷Cs in the surface soils at an elevation of 970 m in the JRYTF. The activities of ^239+240Pu and ¹³⁷Cs were 0.214±0.017-0.734±0.046 mBq/g and 3.44±0.18-4.92±0.22 mBq/g respectively, with the mean values of 0.432±0.172 mBq/g and 4.18±0.49 mBq/g. At the same elevation and with little variation in organic matter content, the activities of ^239+240Pu and ¹³⁷Cs at different sites showed discrepancies, indicating that the slope gradient was a significant factor in the discrepancy.

Fig. 2Full-screen View

Distributions of ^239+240Pu and ¹³⁷Cs activity in the surface soils of paddy field from Longji Rice Terraces. (a) PZTF. (b) JRYTF.

Fig. 3Full-screen View

Distributions of ^239+240Pu and ¹³⁷Cs activities in the surface soils of equal elevation paddy field in Longji Rice Terraces. (a) With an elevation of 860 m in PZTF. (b) With an elevation of 970 m in JRYTF.

Similar to the staircase gradient, here, we define the angle (α) between the slope line of adjacent sampling sites and the horizontal plane as the slope gradient of the terraces between adjacent sampling sites (Fig. 4). The slope gradient was determined by the width (b) of the horizontal terraces at each level and the height (h) between them and the upper or lower terraces, which reflected the steepness of the terraced section. Fig. 5a shows the slope gradient distribution between two adjacent sampling sites in the PZTF. The slope gradient first decreased and then increased with elevation from 900 to 790 m, which is the opposite of the distribution trends shown for the activities of ^239+240Pu and ¹³⁷Cs, indicating that the larger the slope gradient, the lower the activities of ^239+240Pu and ¹³⁷Cs and the greater the soil erosion. This is consistent with the results for sloping cultivated fields in Nanchong of Sichuan and Kaixian of Chongqing [55]. Khodadadi et al. [24] and Su et al. [25] reported that soil erosion affected the distribution of plutonium and ¹³⁷Cs in hillslope soils. Li et al. [53] and Van Oost et al. [54] also reported that slope gradient is the most important factor affecting the spatial variability of soil erosion rates.

Fig. 4Full-screen View

Schematic diagram of the slope gradient between adjacent sampling sites of the terraces.

Fig. 5Full-screen View

Distribution of slope gradient in the adjacent surface soil sample site in Longji Rice Terraces. (a) PZTF. (b) JRYTF.

The activities of ^239+240Pu and ¹³⁷Cs at an elevation of 930 m (P1#1_0-5) were lowest, except for that with an elevation of 790 m (P16). However, the slope gradient (9.4°) between this site P1#1_0-5 and the sampling site P11 was low (Fig. 2a and Fig. 5a). This is inconsistent with the rule that the greater the slope gradient, the lower are the activities of ^239+240Pu and ¹³⁷Cs. This may be due to the fact that the slope gradient between this site (P1#1_0-5, located at the top of the terrace slope closest to and directly below the observation platform) and the observation platform should be the largest. Serious soil erosion usually occurs at the top of slopes in sloping and terraced farmlands without soil material supplement [25]. This slope gradient exacerbated the scouring of the soil at this site by the rapid runoff above it (especially the concentrated rainwater from the observation platform), and there was insufficient soil replenishment at this site, resulting in the lowest activities of ^239+240Pu and ¹³⁷Cs. The slope gradient (9.2°) between the sample sites with elevations of 860 m (PH11) and 840 m (P14) was smaller than that (18.4°) between the samples with elevations of 860 m and 870 m (P13), whereas the maximum activity of ^239+240Pu and ¹³⁷Cs was not at an elevation of 840 m but at the elevations of 860 m and 870 m place. This is due to the influence of existing trails, which are paved with stones between sites at elevations of 860–840 m (Fig. 1a). The trail impeded the movement of the soil to lower elevations from an elevation of 860 m, whereas the concentrated water flowing on the trail intensified the scouring of the soil at an elevation of 840 m. This is similar to a previous report [32] that the concentrated water flow of rural roads may scour the soil and affect the distribution of ¹³⁷Cs and the soil erosion rate.

Figure 5b shows the slope gradient distribution between two adjacent sites in the JRYTF. The slope first decreased and then increased with elevation from 905 m to 810 m, which is the opposite of the distribution trends shown for ^239+240Pu and ¹³⁷Cs dependence on elevation. The sampling site at an elevation of 1070 m (DA8) was close to the observation platform. Due to intense precipitation, irrigation, and other factors, surface runoff is accelerated by gravity and flows downward from the upper terraces, which greatly increases the runoff erosive power [56], resulting in lower activities of ^239+240Pu and ¹³⁷Cs at the sampling site (DA8). From elevations of 1070 m to 905 m (DA2), the activities of ^239+240Pu and ¹³⁷Cs increased with a decrease in elevation, which may have been caused by the gradual decrease in the erosivity of the surface runoff due to the gradual increase in the soil carried by the surface runoff.

3.1.2

Vertical distributions of ^239+240Pu and ¹³⁷Cs activities in soil cores

The vertical distributions of ^239+240Pu and ¹³⁷Cs activities in the PZTF soil core (P1 #1) at an elevation of 930 m and the soil core (P1 #2) at an elevation of 860 m are shown in Fig. 6a and 6b, respectively. The activities of ^239+240Pu and ¹³⁷Cs in soil core P1#1 tended to be uniformly distributed within the upper 20 cm and declined exponentially below this depth. The distribution of ¹³⁷Cs activity in soil core P1#2 also tended to be uniformly distributed within the upper 20 cm, but the ^239+240Pu value fluctuated greatly within the upper 20 cm, and both declined exponentially below this depth. The difference in the distribution of Pu isotopes and ¹³⁷Cs activities in the upper 20 cm layers of soil core P1#2 may be due to rainwater immersion. At site P1#2, water immersion enhanced the reducibility of the soil, leading to a change in Pu (IV) to Pu (III), and the effect of oxide metals acting on Pu migration was weakened [57, 58, 59]. However, this soil environment had a very weak effect on ¹³⁷Cs because the reduction potential of Cs⁺/Cs was lower than that of the soil environment. Therefore, the difference in the distribution of Pu isotopes and ¹³⁷Cs in the upper 20 cm layers of soil core P1#2 may be caused by the impact of rainwater immersion on the oxidation and reduction of Pu in the soil environment. The oxidation and reduction of Pu in soil greatly influences Pu mobility [60]. The difference in the distribution of ^239+240Pu activity in the upper 20 cm layers from soil cores P1#1 and P1#2 may be due to discrepancies in the soil environment. P1#1 had fewer rainwater soaking times because of its location at the top of the terraces, which may have been less affected by rainwater immersion. Previous studies [21, 25, 55] have also shown that the vertical distributions of ^239+240Pu and ¹³⁷Cs in the plow layer were basically uniform. Therefore, the vertical distributions of ^239+240Pu and ¹³⁷Cs in the two soil cores indicated that the terraced plow layer depth was approximately 20 cm, and ^239+240Pu and ¹³⁷Cs were mixed uniformly within the upper 20 cm by tillage. The distribution of ^239+240Pu and ¹³⁷Cs activities below the plow layer reflects their downward migration. The relatively high ^239+240Pu and ¹³⁷Cs activities within the upper 20 cm layer are also consistent with the ¹³⁷Cs results of the plow layer in the earth-banked terraces of the Yimeng Mountains [56].

Fig. 6Full-screen View

Vertical distributions of ^239+240Pu and ¹³⁷Cs activities in soil cores from Longji Rice Terraces. (a) P1#1. (b) P1#2.

The mean activities of ^239+240Pu and ¹³⁷Cs in the soil core P1#2 (0.242±0.252 and 3.58±2.17 mBq/g, respectively) are twice those of P1#1 (0.121±0.080 and 1.84±1.46 mBq/g, respectively), and the maximum activities of ^239+240Pu and ¹³⁷Cs (0.693±0.046 and 5.77±0.24 mBq/g, respectively) are also higher than that of P1#1 (0.228±0.008 and 4.07±0.20 mBq/g, respectively), indicating a different extent of soil erosion at two sites.

Inventories of ^239+240Pu and ¹³⁷Cs in soil cores

Calculated the inventories by integrating ^239+240Pu content and ¹³⁷Cs content in all layers from the cores P1#1 and P1#2 (Fig. 6), we calculated that the inventories of ^239+240Pu in soil cores P1#1 and P1#2 were 81.51 Bq/m² (surface layer accounted for 14.87%) and 119.28 Bq/m² (surface layer accounted for 11.10%), respectively. And the ¹³⁷Cs inventories in soil cores P1#1 and P1#2 were 1232.86 Bq/m² (surface layer accounted for 16.42%) and 1827.61 Bq/m² (surface layer accounted for 19.16%), respectively.

Table 2 lists the locations, elevations, precipitations, and inventories of ^239+240Pu and ¹³⁷Cs of Longji Rice Terraces and the previous work. The mean inventories of ^239+240Pu and ¹³⁷Cs in Longji Rice Terraces were 100.40 Bq/m² and 1530.24 Bq/m², respectively. The mean inventory of ^239+240Pu was higher than the result (63 Bq/m²) of forest soil in Guiyang of Guizhou [42] but within the range (36-410 Bq/m²) in the 20 °N–30 °N band summarized by Cao et al. [63]. The mean inventory of ¹³⁷Cs was higher than that previous studies in Kaiyuan [61] and Libo [62], but within the range (414.42–3140.88 Bq/m², decay-corrected to 2021) in the 20 °N – 30 °N band summarized by Tagami et al. [64]. The global inventory of plutonium deposited at a specific location may be related to precipitation, elevation, latitude, and other factors [65]. As shown in Table 2, the latitudes of Longji Rice Terraces, Guiyang, Kaiyuan, and Libo are similar, and the elevation of Longji Rice Terraces is lower than that of Guiyang but similar to that of Libo. The precipitation of Longji Rice Terraces(1900 mm/y) was higher than that of Guiyang(1130 mm/y) and Libo(1470 mm/y), and the high plutonium inventory and ¹³⁷Cs inventory of Longji Rice Terraces may be due to the relatively high precipitation.

Soil erosion rate

The soil erosion rate of cultivated land can be estimated by a simplified mass balance model [4, 66, 67]: $I_{\text{m}}=I_{\text{f}}{\left(1-\frac{h}{H}\right)}^{N-1963},$ (1) where I_m is the measured inventory of ^239+240Pu or ¹³⁷Cs in the eroded soil core (Bq/m²), I_f is the background value of ^239+240Pu or ¹³⁷Cs inventory (Bq/m²) in the study area, H is the depth of the plow layer (cm), h is the erosion thickness (cm/a), and N is the sampling year. Thus, the soil erosion rate E (t/(ha·a)) is given by the expression [4]: $E=100\cdot h\cdot B,$ (2) where B is the soil bulk density (g/cm³).

A background inventory (I_f) of ^239+240Pu or ¹³⁷Cs in the study area is crucial for estimating soil erosion. The traditional approach is to take an inventory of an undisturbed and flat reference site, which represents the background inventory of the study area. However, this approach is not applicable to the Longji Rice terraces. If site P1#1 is taken as a reference site, we will obtain the unreasonable result that there is only soil deposition in the terraces, except for the top site P1#1 and the bottom site, because the lowest activity concentrations of ^239+240Pu and ¹³⁷Cs on the surface (Fig. 2a) appeared at 930 m (the top site P1#1) and 790 m (the bottom site). It is also inappropriate to use site P1#2 as a reference because it is impossible to determine whether the soil flowing into and out of site P1 #2 is in equilibrium. In addition, the inventories of ^239+240Pu and ¹³⁷Cs were higher than those of nearby areas (Table 2) because of the relatively high precipitation in the study area; therefore, neither of the sites listed in table 2 is an ideal reference site. It is also difficult to find an ideal undisturbed and flat reference site in the study area because of its unique erosional environment (special topography and geomorphological characteristics). Moreover, when the surface soil was eroded, the depth of the plow layer was approximately constant owing to the same tillage method; therefore, the deeper soil under the original plow layer was turned over and mixed evenly with the upper soil, which diluted the ^239+240Pu and ¹³⁷Cs in the soil of the original plow layer, resulting in the redistribution of ^239+240Pu and ¹³⁷Cs in the surface soil from the terraces. Therefore, the distributions of ^239+240Pu and ¹³⁷Cs in the surface soils from the studied area of Longji Rice terraces can reflect soil erosion, and the mean surface activities of ^239+240Pu and ¹³⁷Cs can be used as the background values of surface activity in the studied area. Therefore, we estimated the background values of ^239+240Pu and ¹³⁷Cs inventories from the mean activities of ^239+240Pu and ¹³⁷Cs, respectively, in surface soil.

The ^239+240Pu in the surface layers (0–5 cm) of soil cores P1#1 and P1#2 accounted for 14.87% and 11.10%, respectively, of the total soil core inventory, with a mean value of 12.98%. The ¹³⁷Cs in the surface layers of soil cores P1#1 and P1#2 accounted for 19.15% and 16.42% of the total soil core inventory, respectively, with a mean value of 17.79%. The inventories of ^239+240Pu and ¹³⁷Cs in the surface layer were 14.67 and 303.92 Bq/m², respectively, which were derived from the mean activity of ^239+240Pu (0.253 mBq/g) and ¹³⁷Cs (5.24 mBq/g) as well as the bulk density (1.16 g/cm³) in the surface samples from PZTF. Then the background values of 113.05 Bq/m² for ^239+240Pu and 1708.38 Bq/m² for ¹³⁷Cs were estimated by the proportion (12.98% for ^239+240Pu and 17.79% for ¹³⁷Cs, respectively) of inventory in the surface layer. The background values obtained using this method were slightly different from the mean inventories of ^239+240Pu and ¹³⁷Cs from cores P1 #1 and P1 #2 in the study area. For ^239+240Pu, the calculated background value was 113.05 Bq/m² and the mean inventory from P1#1 and P1#2 cores was 100.4 Bq/m², and the difference between them was only 11.6%. For ¹³⁷Cs, the difference between the calculated background value of 1708.38 Bq/m² and the mean inventory of 1530.24 Bq/m² was only 12.5%. Therefore, the background values of the study area estimated using our method are reasonable and credible.

Soil erosion rates were calculated using ^239+240Pu and ¹³⁷Cs methods (Table 3). The result showed that the erosion rates calculated by the ^239+240Pu method were13.04 t/(ha·a) for P1#1 site and –2.15t/(ha·a) for the P1#2 site, which is consistent with the values of 13.01 t/(ha·a) and –2.70 t/(ha·a) by ¹³⁷Cs method. The erosion rates were negative, indicating net soil deposition at site P1 #2.

P1#1 was near the observation platform, and the slope gradient between it and the observation platform was steeper, resulting in severe erosion at P1#1 and no soil replenishment at that location. Sampling site P1#2 was located on the contour at an elevation of 860 m (near PH11). The slope gradient was approximately 18.4°, which was gentler than the slope gradient between the upper terraces at higher elevations. Therefore, the eroded soil from the upper terraces was easily deposited in P1#2. Soil erosion rates at sampling site P1#1 were much higher than those at P1#2, again confirming that local landform features (e.g., slope gradient variation) are the main factors influencing soil erosion in the Longji Rice Terraces.

The mean erosion rates (i.e., the arithmetic mean of the erosion rates at sampling points P1#1 and P1#2) of Longji Rice Terraces estimated by the ^239+240Pu method (5.44 t/(ha·a)) and the ¹³⁷Cs method (5.16 t/(ha·a)) are basically consistent with that of the Dapotou depression in Yunnan Province (6.32 t/(ha·a)) [61], but higher than that of the cultivated land in Huanjiang County, Guangxi (0.22 t/(ha·a)) [69]. The Dapotou depression area (23° 36′48.04″–23° 37′28.10″ N) and the Longji Rice terraces are both hillside cultivated lands with similar latitudes, this is a possible reason why the mean erosion rates of both are similar. Higher precipitation (1900 mm/y for Longji Rice terraces and 1390 mm/y for Huanjiang County) may lead to higher soil erosion in the Longji Rice terraces than in Huanjiang County. In addition, the reported cultivated land in Huanjiang County has been partially reclaimed, and some of the cultivated land was converted from cultivated land to grassland in 2004. The diverse land-use types and high vegetation cover may be other reasons for the lower soil erosion rate in Huanjiang County. This may be due to the higher vegetation cover of grasslands compared to cultivated land, and the interlaced plant roots in grasslands can inhibit soil erosion [36]. Moreover, the mean soil erosion rate of Longji Rice Terraces is much lower of the steep sloping terraces (56.47 t/(ha·a)) in Zhongxian County of Chongqing [26], the earth-banked terraces (27 t/(ha·a)) in the Yimeng Mountains [56] and the cultivated land (19.4 t/(ha·a)) in Pingguo City of Guangxi [68]. The soil of the steep-slope terraces in Zhongxian County is purple with loose parent rock that easily disintegrates and has faster soil loss. The Longji Rice Terraces are known as a hundred-stage terrace system, whereas the Yimeng Mountain terrace study area is only a seven-stage terrace system, which may have led to a lower erosion rate of the Longji Rice terraces. This may be because the terraces changed the slope gradient of the original hillslope, flattening the terraced plots at all levels. The more steps there were, the shorter the slope length of each terrace, and the terrace fields could slow or prevent rapid surface runoff due to the interception of runoff water and a shorter slope length, which significantly reduced runoff and soil erosion rates [70]. In addition, Longji Rice Terraces are cultivated once a year, while in Yimeng Mountain, they are cultivated twice a year (wheat-corn rotation), and lower tillage frequency may be another reason for the lower erosion rate in Longji Rice terraces. It has been reported that there is a sinkhole in the cultivated land in Pingguo City, and obvious underground erosion is a possible reason why the soil erosion rate in this area is higher than that of the Longji Rice Terraces.

In summary, the results show that the differences in soil erosion rates in the above-mentioned areas are mainly related to precipitation, landform features, soil structure and properties, vegetation coverage, crop type, and tillage frequency.