Effect of terrace forms on water and tillage erosion on a hilly landscape in the Yangtze River Basin, China

Geomorphology 216 (2014) 114–124

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

Effect of terrace forms on water and tillage erosion on a hilly landscape in the Yangtze River Basin, China J.H. Zhang ⁎, Y. Wang, Z.H. Zhang Key Laboratory of Mountain Surface Processes and Ecological Regulation, CAS; Institute of Mountain Hazards and Environment, Chinese Academy of Sciences and Ministry of Water Conservancy, Chengdu 610041, China

a r t i c l e

i n f o

Article history: Received 6 November 2013 Received in revised form 12 March 2014 Accepted 23 March 2014 Available online 31 March 2014 Keywords: Terrace form Soil particle redistribution Tillage erosion Water erosion Yangtze River Basin China

a b s t r a c t In long-term agricultural practices, cultivated land on long slopes in hilly areas is sometimes terraced to diminish soil loss caused by water and facilitate farming operations. In the Upper Yangtze River Basin, the forms of the terraces vary with the landform, soil resources and traditional techniques used in different areas. This study assessed the pattern of soil redistribution and soil particle size redistribution for two major types of terraces: with and without embankments. Samples of 137Cs and soil texture were collected at the upper and lower parts of terraces, and at an interval of 5 m along the transect of the toposequence of individual terraces. The non-embankment terrace (NET) landscape showed increasing 137Cs downslope, whereas the embankment terrace (ET) landscape exhibited an abrupt change in 137Cs at the embankment. Tillage erosion dominated the soil redistribution, with smaller contributions made by water erosion in the ET landscape. In addition to tillage erosion, water erosion played an important role in the soil redistribution in the NET landscape, resulting in a net soil loss at both the upper and lower parts of the terrace. Soil fine particle fractions exhibited a trend of gradual increase along the transect of the toposequence in the NET landscape, and a similar fraction of fine particles was found at the upper and lower parts of the terrace. The establishment of embankments at the lower end of the terrace obstructed the formation and development of overland water flow, thereby creating a line of null downslope soil transport leading to tillage-induced soil accumulation on the upslope side of the embankment with little granulometric sorting. In the NET landscape, soil loss seemed to increase as a result of tillage erosion (an important delivery mechanism for water erosion), accelerating the on-site soil loss. The embankments at the lower end of the sloping terrace played a crucial role in the soil redistribution patterns of the toposequence, resulting in a shift from a water-dominated erosion process to a tillage-dominated process and a distinct pattern of soil particle size distribution. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Soil particle redistribution generally arises from soil erosion and deposition on hillslope landscapes. Water and wind cause soil particle movement on slopes, resulting in variation in soil features and especially surface soil properties. The assessment of soil particle redistribution on agricultural landscapes is meaningful for developing and correcting erosion models, establishing sediment budgets and implementing precise agricultural practices (Ampontuah et al., 2006). The maintenance of soil productivity is also associated with soil particle size distribution, as fine particle fractions play an important role in adsorbing and transporting SOC and nutrients (Walling et al., 2002; Zhang et al., 2006; Ge et al., 2007). Soil particles move along slopes with granulometric sorting during the water-induced erosion and deposition processes (Rhoton et al., 1979;

⁎ Corresponding author. Tel./fax: +86 28 8523 8973. E-mail address: [email protected] (J.H. Zhang).

http://dx.doi.org/10.1016/j.geomorph.2014.03.030 0169-555X © 2014 Elsevier B.V. All rights reserved.

Stone and Walling, 1996; Ni and Zhang, 2007). Numerous studies of soil redistribution have been conducted on long hillslopes, such as those in North America and Europe, and have documented that soil redistribution is controlled not only by water and wind erosion, but also by tillage erosion (Lindstrom et al., 1992; Govers et al., 1994; Lobb et al., 1995; Van Muysen et al., 1999; De Alba, 2001). The magnitude of soil redistribution due to tillage on hummocky land in temperate climates may often exceed that of water erosion (e.g., Govers et al., 1996). In terms of shoulder slope landscape positions, it has been estimated that tillage erosion accounts for at least 70% of the total soil erosion in southwestern Ontario, Canada (Lobb et al., 1995). Based on radionuclide (137Cs) tracer studies that have analysed soil redistribution over approximately 40 years, tillage erosion and deposition rate estimates frequently exceed 10 Mg ha−1 yr−1 at the eroding and aggrading sites of intensively cultivated land (Govers et al., 1996; Van Oost et al., 2003). In terms of long-term agricultural practice, farmers are aware that terracing can efficiently prevent soil and water losses in fields, and have developed a variety of terraces by dissecting long slopes into

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

several short segments, such as level, sloping and reverse sloping terraces. In the hilly areas of the Upper Yangtze River Basin, sloping terraces are dominant, and there are two forms of terrace construction, one with an embankment and one without. Despite the embankment, embanked terraces save on labour and are cost-efficient, as the original geomorphological shape is only slightly transformed with a small amount of construction engineering required. Terracing is effective at decreasing soil erosion by water, but may cause more intensive soil redistribution by tillage unless a levelled surface is created. A few studies have documented the severity of tillage translocation and erosion at the upper slope positions of linearly terraced hillslopes through the use of 137 Cs or physical tracers (Quine et al., 1999; Zhang et al., 2004a,b). Radiocaesium data have shown that soil translocation on short terraced slopes is more intense than that on long slopes (Zhang et al., 2006). Hoeing is used extensively to till fields, and unilateral downslope tillage is normally performed by overturning and pulling soil in the downslope direction to save time and energy. This has a positive effect on downslope soil translocation. Water erosion was once commonly assumed to be a unique soil redistribution process in the hilly areas of the Upper Yangtze River Basin. Recent studies there have demonstrated that tillage erosion is also an important soil redistribution process (Zhang et al., 2006, 2009), and that tillage erosion is a non-negligible process of soil redistribution in agricultural fields in other parts of the world (Lindstrom et al., 1990, 1992; Lobb et al., 1995; Govers et al., 1996; Poesen et al., 1997; Quine et al., 1999; Quine and Zhang, 2002; Heckrath et al., 2005). However, few studies have examined soil particle size distribution by water and tillage erosion in different types of terraces. Studying the features of soil redistribution along the transect of a toposequence comprising a

115

series of terraces rather than individual terraces would increase our understanding of soil redistribution mechanism. The objectives of this study are (1) to examine the soil redistribution rates and features of the two contrasting types of terraces, i.e., with and without embankments; (2) to determine patterns of water and tillage erosion along the toposequence with different types of terraces; and (3) to establish mechanisms for the horizontal and vertical distributions of soil particles for the two types of terraces. 2. Materials and methods 2.1. Study sites The study area was situated in the Upper Yangtze River Basin in China, an area that presents a variety of geomorphological settings consisting mainly of hills and mountains. Two hilly areas were selected for comparison. One was located in Jianyang County in the Sichuan Basin (30° 26′ N and 104° 28′ E) and the other in Zhongxian County (30° 25′ N and 108° 11′ E) in Chongqing Municipality (Fig. 1). The annual temperatures averaged 17.4 °C at the Jianyang site and 19.2 °C at the Zhongxian site, and the mean annual precipitation amounts were 872 and 1150 mm, respectively. Given the differences in landforms, soil resources, traditional techniques and other attributes, different forms of terraces were established in the two areas. Due to a serious shortage of land resources at the Zhongxian site, embankments were not established at the lower end of the sloping fields. Instead, the slope length changed through the dissection of long slopes into short slope segments, forming non-embankment terraces (NETs). At the Jianyang site, terraces were constructed on the contour and separated by mound embankments

Fig. 1. Map showing the study site locations.

116

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

that normally had a breadth of 40–60 cm, forming embankment terraces (ETs). Tiered terraces along the toposequence were measured and found to have slope gradients of 0.05–0.24 and 0.07–0.22 m m−1 with a mean of 0.16 and 0.17 m m−1 at the Jianyang and Zhongxian sites, respectively, and slope lengths ranging from 10.6 to 20.6 m and 4.2 to 12.0 m with a mean of 15.7 and 6.7 m. Given the steep geomorphologic shape at Zhongxian, longer and steeper slopes appeared between the two terraces amounting to 14.5 m in length and a slope steepness of approximately 10 m m−1, and formed an important belt for the transmission of overland water flow. The parent material at the two sites was derived from Jurassic-age mudstone and sandstone, and the subsequent soils were classified as Regosols in the FAO soil taxonomy. In these two areas, hoeing tillage is predominant in long-term agricultural practice, which is characterised by overturning and pulling soil downslope during tillage operations. The crop rotations were similar between the two areas, with both including wheat (Triticum aestivum L), corn (Zea mays L), sweet potato (Ipomoea batatas (L.) Lam), peanut (Arachis hypogaea L.) and rape (Brassica napus L.).

Three cores were collected in the soil profile for each sampling point, which included all of the soil layers from surface to bedrock. For the two types of profile sampling, the cores were segmented into subsample sections in 5 cm increments from soil surface to bedrock. The soil samples were air dried, crushed and passed through a 2 mm mesh sieve to remove any gravel. The same layer subsamples of the b 2 mm fraction for the three soil profiles were composited and packed into plastic beakers with 320 cm 3 volumes, and the 137 Cs activity was measured using a hyperpure lithium-drifted germanium detector, which was described in detail in a previous study (Zhang et al., 2008). The original measurements of the 137Cs activity expressed on a per-unit mass basis (Bq kg− 1) were converted into the inventory (Bq m− 2) using the total weight of the bulked core sample and the cross-sectional area of the sampling device. The soil particle size fractions for each subsample were analysed using the pipette method (Liu, 1996). 2.3. Calculation of erosion rates

2.2. Soil sampling and determination of soil samples Soil samples were taken along the transects of toposequences containing series of terraces with four slope segments at the two study sites. The toposequence at the Jianyang site (Fig. 2a) had a total horizontal length of 63 m with a mean gradient of 23%, and that at the Zhongxian site (Fig. 2b) had a horizontal length of 28 m with a mean gradient of 69%. The coordinates of each sampling point and their elevations were measured using a survey-grade differential global positioning system (DGPS). Soil samples were collected using a hand-operated core sampler measuring 6.8 cm in diameter to determine the 137Cs activity and soil particle size fraction. From the upper and lower slopes of each terrace, at a distance of 1 m from the immediate boundary of the terraced field, three core samples were collected in repetition at an interval of 80 cm on the contour, extending to the depth of the bedrock. Forty-eight samples were collected from the two series of terraces. To examine the soil redistribution in the individual NET fields, an additional field was selected in the middle portion of the toposequence, and more dense sampling points were established along the transect of the field. In terms of the 137Cs soil sampling, the core sampler was used to make physical determinations, and samples were collected down to bedrock at 5 m intervals along the transect of the slope.

a

Embankment

Sampling point 0

5

10 m

b Sampling point 0

5

10 m

Fig. 2. Contrasting terrace series: (a) with embankments and (b) without embankments.

A flow line down a toposequence comprises individual terraces along the transect of the toposequence, and each terrace can be approximated as a linear slope. Therefore, tillage erosion rates and the relevant parameters can be expressed as follows:   R ¼ 10· Q s;out –Q s;in =Li

ð1Þ

Q s;out ¼ Dt;i ·ρb;i ·ðk1 þ k2 Si Þ

ð2Þ

Q s;in ¼ Dt;i−1 ·ρb;i−1 ·ðk1 þ k2 Si−1 Þ

ð3Þ

where R is the tillage erosion rate (t ha−1 yr−1); Qs,out and Qs,in are the downslope soil flux leaving and entering the i-th terrace due to tillage, respectively (kg m− 1 yr− 1); Li is the slope length of the i-th terrace (m); ρb,i and ρb,i − 1 refer to the pre-tillage dry soil bulk density at the i-th and (i − 1)-th terraces, respectively (kg m− 3); Dt,i and Dt,i − 1 refer to the depth of the tillage layer (m); k1 and k2 are the displacement distance coefficients (m), representing a year level given a major tillage operation per year at the two sites; and Si and Si − 1 are the slope gradient of the i-th and (i − 1)-th terraces (m m−1), respectively. The coefficients k1 and k2 were obtained from previous studies: 0.1539 and 0.4954 at Zhongxian, respectively (Zhang et al., 2009), and 0.1066 and 0.4902 at Jianyang, respectively (Zhang et al., 2004a). Published data on 137Cs reference inventories near the study sites were used to compare the difference in 137Cs loss/gain for the cultivated soils. The 137 Cs reference inventories from the uneroded sites were measured as 1984 ± 152 (SD) Bq m−2 (n = 9) (Zheng et al., 2007) and 1791 ± 205 (SD) Bq m−2 (n = 12) (Wen et al., 2005) in the sample measuring year, and were 1769 and 1596 Bq m−2 with the radioactivity decay corrected to 2008 levels for the Jianyang and Zhongxian sites, respectively. A number of conversion models have been developed, and offer simple to sophisticated approaches with different degrees of accuracy. Sophisticated conversion models consider the effects of the temporal pattern of 137Cs fallout, tillage erosion and dispersion effects due to tillage translocation (Li et al., 2011). Although the simple conversion models generally produce greater structural uncertainties than the sophisticated models (Tiessen et al., 2009; Li et al., 2010, 2011), the sophisticated models often require additional parameters that were not available in the study sites. As such, the total erosion rates were estimated using the simplified mass balance model (MBM1) (Walling and He, 1999). We used previously published 137Cs data for the two cultivated sites (Zhang et al., 2012). The water erosion rates were calculated by subtracting the tillage erosion rates from the total erosion rates.

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

117

2.4. Statistical analysis A simple linear regression was used to test the correlations between two variables, e.g., particle size fraction vs. landscape position; particle size fraction vs. 137Cs concentration; particle size fraction vs. slope length. An analysis of variance was performed to detect the significance of differences in particle size fraction and 137Cs inventory between the upper and lower parts of the terrace. 3. Results 3.1. Features of soil erosion derived from 137Cs data on terraces In terms of the ETs, the 137Cs inventories exhibited an abrupt opposing change on both sides of the terrace boundary. Similar 137Cs inventories (360, 391 and 371 Bq m−2) were found in the upper parts of each terrace except for the toe portion of the toposequence, and the 137Cs inventories in the lower parts of each terrace were 4.2, 2.2 and 2.7 times those in the upper parts for the top, middle, and lower terraces, respectively. The toe terrace had the highest mean value of 137Cs inventory (1126 Bq m− 2) as a result of the depressional location of the whole toposequence (Table 1; Fig. 3a). However, in terms of the NETs except for the toe terrace, the 137Cs inventories displayed a trend of gradual increase along the transect of the toposequence (Table 1; Fig. 3b). In contrast to the toposequence with the ETs, the toposequence of the NETs had lower 137Cs inventories on the toe terrace than on the immediate upper terrace (i.e., the lower terrace). In terms of the NETs, the mean 137Cs concentrations of the soil profile showed an increasing trend from the top to toe terraces of 0.75, 1.46, 1.97 and 1.89 Bq kg−1, respectively (Table 1). However, such a distribution trend of 137Cs concentrations was not found along the transect from the top to toe terraces for the toposequence with embankments (2.39, 1.96, 1.51 and 2.26 Bq kg− 1, respectively; Table 1). Furthermore, in the lower parts of the terraces, a remarkable difference was found in the 137Cs concentrations between the upper and lower parts of the ETs, especially given the higher 137Cs concentrations in the surface soil (0–15 cm). However, in terms of the NETs, there were similar 137Cs concentrations between the upper and lower parts of all of the terraces except for the toe terrace (Fig. 4e–g). In terms of the 137Cs depth distribution, the depth of 137Cs attainment should have been consistent with the depth of the cultivated soil layer, that is, about 20 cm of tillage depth, with no erosion or deposition occurring, as tillage vertically mixes the soil within the cultivated soil layer. In the NET landscape, the depth of 137Cs attainment was approximate to the quite shallow soil profile depths on the top terrace, where a severe soil loss occurred due to tillage. Although the soil layer reached 40 cm deep, 137Cs attainment was only found at a depth of 30 cm on the toe terrace, showing a slight deposition in the toe portion

Fig. 3. Distribution of 137Cs inventories down the toposequence: (a) with embankments and (b) without embankments. (Upper and lower in each terrace represent the upper and lower parts of the terrace, respectively).

of the toposequence. However, in the ET landscape, a deep 137Cs profile distribution at the lower parts of the terrace (except for the middle terrace) was observed with a 137Cs attainment depth of 40 or 45 cm (Fig. 4a–d), suggesting that marked deposition occurred at the lower parts of the ET. 3.2. Distribution of soil particle size In the NET landscape, the b0.002 mm clay fraction of surface soils tended to increase from the top to bottom portions of the toposequence,

Table 1 Mean 137Cs concentrations of the soil profile at different landscape positions (±standard deviation). Landscape type

137

Slope position

Cs concentration/inventory

Top Embankment terrace

Upper ⁎ Lower ⁎ Mean

Non-embankment terrace

Upper Lower Mean

Bq Bq Bq Bq Bq Bq Bq Bq Bq Bq Bq Bq

kg−1 m−2 kg−1 m−2 kg−1 m−2 kg−1 m−2 kg−1 m−2 kg−1 m−2

1.76 ± 360 3.01 ± 1539 2.39 950 0.44 ± 70 1.05 ± 335 0.75 203

0.65 0.46

0.66 0.40

⁎ Upper and lower in each terrace represent the upper part and the lower part of the terrace, respectively.

Middle

Lower

1.06 353 2.86 863 1.96 608 1.59 724 1.33 546 1.46 635

1.34 ± 371 1.68 ± 963 1.51 667 2.05 ± 772 1.89 ± 1129 1.97 951

± 0.46 ± 0.74

± 0.35 ± 0.78

Toe 0.35 0.38

0.96 0.67

2.3 ± 0.97 1239 2.21 ± 1.3 1013 2.26 1126 1.4 ± 0.19 656 2.2 ± 0.96 1020 1.89 838

118

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

Fig. 4. Depth distribution of 137Cs concentrations for the two contrasting terrace series: a, b, c and d represent the upper, middle, lower and toe terraces with embankments, respectively; e, f, g and h represent the upper, middle, lower and toe terraces without embankments, respectively.

and was found to be positively and highly significantly correlated with the distance from the hilltop (r = 0.92, p b 0.01; Fig. 5). Although the relationship between the 0.05–0.002 mm silt fraction and slope length did not reveal any clear pattern for either landscape (r = − 0.42, p N 0.1; data not shown), the b0.05 mm particle size fraction was significantly correlated with the slope length in the NET landscape (r = 0.77, p b 0.05; Fig. 5). However, the b0.002 mm clay fraction of the surface soils did not show such a changing pattern in the ET landscape, and the correlations between these two variables did not yield any significant

results (r = −0.55; p N 0.1). Furthermore, the b 0.05 mm particle size fraction in the NET landscape did not follow a similar relationship to that in the ET landscape (Fig. 5). The soil particle size fractions exhibited different distribution patterns along the transect of the toposequence between the two landscapes. In the NET landscape, a significant difference in the b0.002 mm clay fraction of the surface soils was found between the upper and lower parts of the terrace (p b 0.02; Table 2). In addition, the b 0.05 mm particle size fraction of the surface soils showed an

<0.002 mm clay fraction (%)

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

119

40

30

bc

b

b

30.59

31.46

31.76

5

10

15

a

ac

20 26.11

28.72

10

0

0

20

Distance from the hilltop (m) Fig. 5. Percent of particle size fractions versus slope length.

Fig. 6. Distribution of soil particle size fractions within the selected individual terrace. The different letters between the bars indicate a statistically significant difference at a 95% confidence level.

overall significant difference between the two positions (p = 0.07–0.0005; Table 2). However, in the ET landscape, there were no significant differences between the upper and lower parts of the terrace in the b0.002 mm clay fraction (p = 0.07–0.66) and the b0.05 mm particle size fraction (p = 0.42–1.00; except for the middle terrace). A similar soil particle composition between the upper and lower parts of the terraces in the ET landscape was found when the two landscapes were compared. However, a slight change in the soil clay distribution in the individual terraces along the transect of the sloping terrace was revealed when a more elaborate investigation was conducted. On the individual terraces of the ET landscape, the b0.002 mm clay fraction of surface soils was smaller in the top and toe slope positions than in the middle slope portions. An analysis of variance showed a significant difference in the b0.002 mm clay fraction between the top and middle slope positions (p b 0.0001) and between the toe and middle slope positions (p = 0.003). However, the b0.002 mm clay fractions were similar among the different landscape positions for the middle slope portions (p N 0.10; Fig. 6). In the NET landscape, the depth distribution of the b 0.002 mm clay fraction in the soil profiles was found to decline gradually with the soil depth in the transect portions from the upper part of the middle terrace to the upper part of the toe terrace (Fig. 7), except in the upper part of the lower terrace. The b0.002 mm clay fraction was negatively and significantly correlated with soil depth (p b 0.05, b0.01, b0.001 and b0.05 for the upper and lower parts of the middle terrace, the lower part of the lower terrace and the upper part of the toe terrace, respectively). However, in most cases, the b0.002 mm clay fraction of the NETs did not follow such a pattern of depth distribution, and the depth distributions of this fraction had the opposite trend (Fig. 7), with a positively weak or significant correlation between the clay

fraction and soil depth (e.g., p b 0.10, b0.10, b 0.001 and b 0.01 for the upper part of the middle terrace, the lower part of the lower terrace and the upper and lower parts of the toe terrace, respectively). For both landscapes, the greatest fraction of b 0.002 mm clay was present in the lower part of the toe terrace. However, the b0.002 mm clay fraction in the lower part increased by 127% in the NET landscape, but only by 26% in the ET landscape compared with that of the respective upper parts (Table 2). Although the b0.05 mm particle fraction showed a lower increment in the lower part than the b0.002 mm clay fraction, the increment of the NETs (6.6%) was greater than that of the ETs (1.7%), as shown in Table 2. 3.3. Effects of soil redistribution on soil particle fractions In the ET landscape, the 137Cs data showed that notable soil redistribution occurred over a short distance within the toposequence, with erosion occurring at the upper parts of the terrace and deposition present at the lower parts, as previously stated. However, the soil particle composition did not exhibit significant differences between the upper and lower parts of the terrace, where a large difference in 137Cs inventories was observed (e.g., from the top to lower terraces in the ET landscape; Tables 1 and 2). No significant correlations were found between the 137Cs concentrations and the b0.002 mm clay fraction (r = 0.19, p N 0.1; Fig. 8) or between the 137Cs concentrations and the b0.05 mm particle fraction (r = 0.007, p N 0.1; data not shown). In contrast, there was a highly significant correlation in the NET landscape between the 137Cs concentrations and the b0.002 mm clay fraction (r = 0.51, p b 0.001; Fig. 8), except for the lowest location of the toposequence (the lower part of the toe terrace). The b0.002 mm clay fraction at the lowest location of the toposequence increased by a

Table 2 ANOVA of differences in soil particle size fractions between the up and down slope parts of each terrace. Landscape type

Particle size

Statistic item

Toposequence Top terrace

Embankment terrace

b0.002 mm b0.05 mm

Non-embankment terrace

b0.002 mm b0.05 mm

Mean Sig. P Mean Sig. P Mean Sig. P Mean Sig. P

± SD ± SD ± SD ± SD

Middle terrace

Lower terrace

Toe terrace

Upper ⁎

Lower ⁎

Upper

Lower

Upper

Lower

Upper

20.46 ± 0.74 0.444 75.87 ± 3.54 0.422 5.48 ± 0.57 0.019 44.43 ± 3.53 0.066

18.53 ± 3.83

14.24 ± 2.01 0.067 57.23 ± 2.36 0.0003 8.60 ± 1.11 0.006 49.11 ± 0.83 0.0005

18.16 ± 1.85

16.85 ± 5.36 0.245 75.62 ± 6.40 0.999 9.26 ± 0.26 0.007 44.64 ± 1.91 0.020

12.58 ± 0.87

13.72 ± 0.659 77.48 ± 0.798 11.51 ± b0.0001 56.52 ± 0.040

73.92 ± 1.31 7.23 ± 0.57 50.38 ± 2.10

75.32 ± 0.96 12.28 ± 0.35 54.37 ± 0.26

⁎ Upper and lower in each terrace represent the upper part and the lower part of the terrace, respectively.

75.62 ± 5.05 12.24 ± 0.99 50.61 ± 2.00

Lower 2.71

17.33 ± 0.30

1.09

78.82 ± 8.37

0.75

26.08 ± 0.36

1.82

60.22 ± 1.11

120

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

mean of 196.2% (ranging from 112.4% to 376.0%) compared with the other positions, indicating a noticeably depositional process at the lowest location with granulometric sorting in the landscape. In comparison, the other landscape positions exhibited a water-eroded process as a result of their small fractions of the b 0.002 mm clay. As a result, the 137Cs concentrations may have been associated with fine particle fractions only at landscape positions where water erosion was the dominant erosional process (Table 3).

4. Discussion 4.1. Patterns of sediment deposition Previous studies have reported that a total 137Cs inventory less or more than the reference inventory indicates soil erosion or sediment deposition (Walling and He, 1999). In this study, the toe slope position of the NETs exhibited sediment deposition due to a significantly large fine

Fig. 7. Depth distribution of the soil particle size fractions at different landscape positions. The letters a, b, c and d represent the upper, middle, lower and toe terraces with embankments, respectively. The letters e, f, g and h represent the upper, middle, lower and toe terraces without embankments, respectively. The numbers 1 and 2 represent the upper and lower parts within each terrace, respectively.

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

121

Fig. 7 (continued).

particle fraction. However, this ‘deposition’ is only a relative term, as the 137 Cs inventory at this position (1020 Bq m−2; Table 1) was apparently lower than the local 137Cs reference inventory (1596 Bq m−2). For both landscapes, the ‘real’ deposition was seldom observed in accordance with the 137Cs data, suggesting an erosion-dominant situation in these areas. In terms of the ‘relative’ deposition, there was a great difference in 137Cs inventories between the upper and lower parts of the terrace in the ET landscape, and there was no significant difference in soil

particle size distribution between these two positions (p = 0.07–0.66; see the preceding text). This may be attributable to the entire soil transport without granulometric sorting, induced by tillage erosion as the dominant process of soil redistribution in the ET landscape (Table 3). The embankment played a ‘zero transport’ role in the soil redistribution in the ET landscape, leading to relatively homogenous soil redistribution on the individual terraces due to the mixture effects of the tillage. However, the soil at the lower end of the sloping terraces in the NET

122

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

are more than the local reference inventory, as in the case of sediment deposition (Walling and He, 1999). In the context of this study, however, the depth distribution of the 137Cs was greater than 30 cm in the ‘depositional’ positions of the terrace, and the 137Cs inventories were lower than the local reference inventory. Consequently, deposition was present mostly at the lower parts of the terrace in terms of 137Cs depth distribution. This may be associated with the erosion and deposition processes that may have alternately occurred at a depositional position, and with the water-induced net sediment deposition that typically occurs during the later phase of surface runoff events. In such cases, for the site where water erosion was dominant, the soil in the depositional area was mainly derived from the plough layer of the upslope contributing area, leading to a similar 137Cs concentration to that of the soil in the plough layer. As a result, the soil properties of the new plough layer were related to those of the soil located upslope, where the 137Cs concentrations were previously lower due to soil loss (De Alba et al., 2004; Zhang et al., 2008). In most cases except for the lower part of the terrace, the 137Cs inventories in the NET landscape were only dependent on the extent of the 137Cs depth distribution, as the 137Cs concentrations were similar among the subsample layers (5 cm slices). Across the NET landscape except for the toe terrace, despite a significant difference in the fine particle fraction of the surface soils between the upper and lower parts of the terrace, the 137Cs concentrations of the surface soils were quite similar to those in the ET landscape (p N 0.43, N0.25 and N 0.90 for the top, middle and lower terraces, respectively; Table 1). This indicates that factors other than the fine particle fraction also affected the 137Cs concentrations of the soil, such as the strong surface adsorption ability of the particles derived from sedimentary mudstone. 4.2. Variations in the soil particle composition of the surface soil 137

Fig. 8. Cs concentrations versus the b0.002 mm clay fraction on the terrace landscapes: (a) with embankments and (b) without embankments.

landscape was easily transported down the toposequence step by step, from the top to toe portions due to runoff, resulting in a certain amount of soil clay accumulation on the upslope sides of the boundary. The NET landscape exhibited a much greater difference in fine particle fractions between the upper and lower parts of the toe terrace. This is mainly ascribed to the selective removal of soil fine particles (i.e., granulometric sorting) from the soil texture at the upper ends of the NET, where there was an overland water flow from the immediate upper terraces. As such, different patterns of fine particle redistribution resulted mainly from the dominant soil transport process of tillage erosion in the ET landscape, and the dominant granulometric sorting process of water erosion in the NET landscape. In general, the depth of 137Cs attainment is greater than the depth of the plough layer (normally up to 20 cm deep) when 137Cs inventories

In the ET landscape, variations in the soil clay content occurred close to both sides of the field boundary, with a smaller b 0.002 mm clay fraction at the two ends of the terrace than in the middle portions. However, the reductions in fine particles at the two ends of the sloping terraces were subject to different mechanisms. The reduction in clay fractions at the upper end resulted mainly from the incorporation of parent material and bedrock fragments with coarse particles into the till layer (surface soil layer). Drastic soil losses affected the upper slope position due to intense tillage, and the soil layer became thin where an embankment interrupted soil transport and prevented replacement. To maintain the soil layer thickness required for fundamental soil productivity, farmers have to till into parent material or bedrock. The mudstone and shale distributed widely in the study areas were prone to crushing by hoeing, and therefore it would be practical to incorporate bedrock fragments into the till layer. This mechanism was documented in tillage simulation studies performed in a Regosol in the hilly areas (Zhang

Table 3 Landform parameters and soil erosion rates. Landscape type

Embankment terrace

Non-embankment terrace

Item

Slope length; m Slope gradient; m m−1 Water erosion rate; t ha−1 yr−1 Percentage of total erosion; % Tillage erosion rate⁎; t ha−1 yr−1 Percentage of total erosion; % Slope length; m Slope gradient; m m−1 Water erosion rate; t ha−1 yr−1 Percentage of total erosion; % Tillage erosion rate⁎; t ha−1 yr−1 Percentage of total erosion; %

Toposequence Top terrace

Middle terrace

Lower terrace

Toe terrace

20.6 0.18 7.0 33 14 67 4.2 0.07 3.6 3 102 97

10.6 0.18 17 32 36 68 5.0 0.22 7.8 18 36 82

14.7 0.24 13 29 32 71 5.5 0.20 14 – −0.1 –

16.8 0.05 8.0 35 15 65 12.0 0.18 40 – −11 –

Symbol ‘–’ represents deposition. ⁎ Tillage erosion rates were estimated with a mean tillage depth of 15 cm, in which soil is transported.

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

et al., 2008; Su et al., 2012). Others have found that the SOC content of the newly formed surface soil was lower than in the original surface soil, as subsoils are poor in SOC (Papiernik et al., 2005). Furthermore, higher clay contents are present in shoulder slope positions than in footslope positions in the presence of clay-enriched parent material (Heckrath et al., 2005). Studies have demonstrated that the transformation of soil properties in the surface soil layer is closely associated with the incorporation of subsoil into the surface layer by tillage. Unlike the tops of the slopes, the toes exhibited apparent water erosion (e.g., rills) along with granulometric sorting, as severe water erosion normally occurs in the middle to lower slope positions (e.g., Lobb et al., 1995; Govers et al., 1996). Other studies have shown that a selective removal of fine particles occurs at the slope scale due to water erosion, suggesting that the footslope position is much richer in fine particles (Kosmas et al., 2001; Salvador-Blanes et al., 2006). In areas with Regosols, rills as a visible sign start forming at a distance of a few metres from the hilltop, and stop close to the lower end of the terrace with a gradually increasing cross-sectional area of rills. One recent study captured the heads of rills emerging at a distance of about 6 m from a hilltop under a gradient of 10° and simulated rainfall of 130 mm hr−1 (Yan et al., 2010). Our results confirm the results of Rhoton et al. (1979) who suggest that the b 2 μm clay fraction is preferentially removed. They also complement the results of Stone and Walling (1996), who concluded that b2 μm clay and b63 μm silt-sized particles are eroded preferentially and that the majority of sand-sized particles remain on site. However, our finding is inconsistent with that reported by Ampontuah et al. (2006) in the UK, who indicated that soil becomes coarser from the summit through the eroding side-slope to the bottom of the slope, with the upper slope soils containing higher clay to fine silt (b16 μm) content than the bottom slope soils. Other researchers have reported similar results (e.g., Le Bissonnais et al., 1993; Cai et al., 2002; Toy et al., 2002), explaining that the aggregating and bonding effect of clay (i.e., the cohesive force between clay-sized particles) resists detachment by runoff and that the coarser fractions (e.g., 16–63 μm) are relatively easily detached. The differences in the patterns of soil particle size distribution observed in other studies could be attributed to bedrock features in our case, which was mudstone sedimentary rock with a fine-grained texture. 4.3. Effects of terrace boundaries on soil particle size distribution As stated, the clay content of the surface soils in the ET landscape was similar at every position of the middle portions of the individual terraces, except for a variation present at the two ends of each terrace. Such a pattern of soil particle size distribution typically results from soil redistribution by tillage. Tillage excavates subsoil material and incorporates it into surface horizons in upper slope positions where surface soil is lost, and then transports it downslope. Tillage evenly mixes soil within the till layer, and the middle portion of the sloping terrace retains a stable soil flux by receiving soil that has been translocated from upslope. In this way, it acts as a conveyor belt (Govers et al., 1996; De Alba et al., 2004) in the middle portion, where the soil particle composition always remains constant. However, its conveyor belt function ends when the soil is translocated to the toe slope position, causing the soil particle composition to change at this position. While tillage erosion is the dominant soil redistribution process, water erosion acted as a minor process in the ET landscape, resulting in the selective removal of fine particles at the lower ends of the sloping terrace. A slight decrease in the clay content at the toe slope position may be associated with an increase in the slope gradients at the lower end section of the sloping terrace, where a drainage ditch was constructed on the contour to prevent waterlogging and collect depositional sediments. The upslope side of the ditch could not be directly established given the non-consolidation and instability of the soil. Therefore, a side slope was formed locally at the lower end section on

123

the upslope side of the ditch, i.e., on the opposite side of the field embankment, an area that had a greater slope gradient than the other lower parts of the terrace. However, despite the increased slope gradients, the clay fractions did not decline much as a result of the slope length limitations (normally only 1–2 m long). 5. Conclusions The 137Cs data showed that soil redistribution patterns in the two contrasting series of terraces, i.e., with embankments and without embankments, were different. Tillage erosion was a major contributor to the total soil erosion in the ET landscape, leading to erosion at the upper parts of each terrace and deposition at the lower parts. However, water erosion had an effect comparable to that of the soil redistribution by tillage erosion in the NET landscape, resulting in net soil losses at both the upper and lower parts of the terrace. The soil fine particle fractions exhibited a trend of gradual increase in the line of the toposequence in the NET landscape, with a significant linear correlation between the b0.002 mm clay fraction and distance from the hilltop. Similar fine particle fractions were found at the upper and lower parts of the terrace in the ET landscape. This suggests that the establishment of embankments at the lower end of the terrace obstructed the formation and development of overland water flow, creating a line of zero downslope transport of soil and resulting in a tillage-induced soil accumulation on the upslope side of the embankment with little granulometric sorting. Contrary to what was expected, the soil loss in the NETs did not decrease, and even appeared to increase. Acknowledgements The authors wish to acknowledge the financial support for this study provided by the 135 Strategic Program of the Institute of Mountain Hazards and Environment, CAS (SDS-135-1206) and the National Natural Science Foundation of China (41271242). References Ampontuah, E.O., Robinson, J.S., Nortcliff, S., 2006. Assessment of soil particle redistribution on two contrasting cultivated hillslopes. Geoderma 132, 324–343. Cai, D., De Roock, M., Jin, K., Schiettecatte, W., Wu, H., Gabriels, D., Hartmann, R., Cornellis, W., 2002. Nutrient load in runoff from small plots: laboratory and field rainfall simulation test on Chinese loess soils. Proc. 12th ISCO Conf., Beijing 2002, pp. 160–163. De Alba, S., 2001. Modelling the effects of complex topography and patterns of tillage on soil translocation by tillage with mouldboard plough. J. Soil Water Conserv. 56, 335–345. De Alba, S., Lindstrom, M., Schumacher, T.E., Malo, D.D., 2004. Soil landscape evolution due to soil redistribution by tillage: a new conceptual model of soil catena evolution in agricultural landscapes. Catena 58, 77–100. Ge, F.L., Zhang, J.H., Su, Z.A., Nie, X.J., 2007. Response of changes in soil nutrients to soil erosion on a purple soil of cultivated sloping land. Acta Ecol. Sin. 27, 459–464. Govers, G., Vandaele, K., Desmet, P.J.J., Poesen, J., Bunte, K., 1994. The role of tillage in soil redistribution on hillslopes. Eur. J. Soil Sci. 45, 469–478. Govers, G., Quine, T.A., Desmet, P.J.J., Walling, D.E., 1996. The relative contribution of soil tillage and overland flow erosion to soil redistribution on agricultural land. Earth Surf. Process. Landf. 21, 929–946. Heckrath, G., Djurhuus, J., Quine, T.A., Van Oost, K., Govers, G., Zhang, Y., 2005. Tillage erosion and its effect on soil properties and crop yield in Denmark. J. Environ. Qual. 34, 312–323. Kosmas, C., Gerontidis, S., Marathianou, M., Detsis, B., Zafiriou, T., Van Muysen, W., Govers, G., Quine, T., Van Oost, K., 2001. The effects of tillage displaced soil on soil properties and wheat biomass. Soil Tillage Res. 58, 31–44. Le Bissonnais, Y., Singer, M.J., Bradford, J.M., 1993. Assessment of soil erodibility: the relationship between soil properties, erosion processes and susceptibility to erosion. In: Wicherek, S. (Ed.), Farm Land Erosion: In Temperate Plains Environment and Hills. Elsevier, Amsterdam, pp. 87–96. Li, S., Lobb, D.A., Tiessen, K.H.D., McConkey, B.G., 2010. Selecting and applying cesium-137 conversion models to estimate soil erosion rates in cultivated fields. J. Environ. Qual. 39, 204–219. Li, S., Lobb, D.A., Kachanoski, R.G., McConkey, B.G., 2011. Comparing the use of the traditional and repeated-sampling-approach of the Cs-137 technique in soil erosion estimation. Geoderma 160, 324–335. Lindstrom, M.J., Nelson, W.W., Schumacher, T.E., Lemme, G.D., 1990. Soil movement by tillage as affected by slope. Soil Tillage Res. 17, 255–264. Lindstrom, M.J., Nelson, W.W., Schumacher, T.E., 1992. Quantifying tillage erosion rates due to moldboard plowing. Soil Tillage Res. 24, 243–255.

124

J.H. Zhang et al. / Geomorphology 216 (2014) 114–124

Liu, G.S., 1996. Soil Physical and Chemical Analysis & Description of Soil Profiles. Chinese Standard Press, Beijing (in Chinese). Lobb, D.A., Kachanoski, R.G., Miller, M.H., 1995. Tillage translocation and tillage erosion on shoulder slope landscape positions measured using Cs-137 as a tracer. Can. J. Soil Sci. 75, 211–218. Ni, S.J., Zhang, J.H., 2007. Variation of chemical properties as affected by soil erosion on hillslopes and terraces. Eur. J. Soil Sci. 58, 1285–1292. Papiernik, S.K., Lindstrom, M.J., Schumacher, J.A., Farenhorst, A., Stephens, K.D., Schumacher, T.E., Lobb, D.A., 2005. Variation in soil properties and crop yield across an eroded prairie landscape. J. Soil Water Conserv. 60, 388–395. Poesen, J., van Wesemael, B., Govers, G., Martinez-Fernandez, J., Desmet, P., Vandaele, K., Quine, T., Degraer, G., 1997. Patterns of rock fragment cover generated by tillage erosion. Geomorphology 18, 183–197. Quine, T.A., Zhang, Y., 2002. An investigation of spatial variation in soil erosion, soil properties, and crop production within an agricultural field in Devon, United Kingdom. J. Soil Water Conserv. 57, 55–65. Quine, T.A., Walling, D.E., Chakela, Q.K., Mandiringana, O.T., Zhang, X., 1999. Rates and patterns of tillage and water erosion on terraces and contour strips: evidence from caesium-137 measurements. Catena 36, 115–142. Rhoton, F.E., Smeck, N.E., Wilding, L.P., 1979. Preferential clay mineral erosion from watersheds m the Maumee River Basin. J. Environ. Qual. 8, 547–550. Salvador-Blanes, S., Cornu, S., Couturier, A., King, D., Macaire, J.-J., 2006. Morphological and geochemical properties of soil accumulated in hedge-induced terraces in the Massif Central, France. Soil Tillage Res. 85, 62–77. Stone, P.M., Walling, D.E., 1996. The Particle-Size Selectivity of Sediment Mobilization from Devon Hillslopes. In: Anderson, M.G., Brooks, S.M. (Eds.), Advances in Hillslope Processes, Vol. 1. John Wiley and Sons Ltd., New York, pp. 507–527. Su, Z.A., Zhang, J.H., Qin, F.C., Nie, X.J., 2012. Landform change due to soil redistribution by intense tillage based on high-resolution DEMs. Geomorphology 175–176, 190–198. Tiessen, K.H.D., Li, S., Lobb, D.A., Mehuys, G.R., Rees, H.W., Chow, T.L., 2009. Using repeated measurements of Cs-137 and modelling to identify spatial patterns of tillage and water erosion within potato production in Atlantic Canada. Geoderma 153, 104–118. Toy, T.J., Foster, G.R., Renard, K.G., 2002. Soil Erosion: Processes, Prediction, Measurement, and Control. John Wiley and Sons Ltd., New York. Van Muysen, W., Govers, G., Bergkamp, G., Roxo, M., Poesen, J., 1999. Measurement and modelling of the effects of initial soil conditions and slope gradient on soil translocation by tillage. Soil Tillage Res. 51, 303–316.

Van Oost, K., Govers, G., Van Muysen, W., 2003. A process-based conversion model for caesium-137 derived erosion rates on agricultural land: an integrated spatial approach. Earth Surf. Process. Landf. 28, 187–207. Walling, D.E., He, Q., 1999. Improved models for estimating soil erosion rates from cesium-137 measurements. J. Environ. Qual. 28, 611–622. Walling, D.E., Russell, M.A., Hodgkinson, R.A., Zhang, Y., 2002. Establishing sediment budgets for two small lowland agricultural catchments in the UK. Catena 47, 323–353. Wen, A.B., Qi, Y.Q., Wang, Y.C., He, X.B., Xiong, J.X., Zhang, X.B., 2005. Study on erosion and sedimentation in Yangtze Three Gorge Region. J. Soil Water Conserv. 19, 33–36 (in Chinese). Yan, D.C., Wen, A.B., Shi, D.L., Ju, L., He, X.B., 2010. Critical slope length of rill occurred in purple soil slope cultivated land in Three Gorges Reservoir area. J. Yangtze River Sci. Res. Inst. 27, 58–61 (in Chinese). Zhang, J.H., Frielinghaus, M., Tian, G., Lobb, D.A., 2004a. Ridge and contour tillage effects on soil erosion from steep hillslopes in the Sichuan Basin, China. J. Soil Water Conserv. 59, 277–284. Zhang, J.H., Lobb, D.A., Li, Y., Liu, G.C., 2004b. Assessment for tillage translocation and tillage erosion by hoeing on the steep land in hilly areas of Sichuan, China. Soil Tillage Res. 75, 99–107. Zhang, J.H., Quine, T.A., Ni, S.J., Ge, F.L., 2006. Stocks and dynamics of SOC in relation to soil redistribution by water and tillage erosion. Glob. Chang. Biol. 12, 1834–1841. Zhang, J.H., Nie, X.J., Su, Z.A., 2008. Soil profile properties in relation to soil redistribution by intense tillage on a steep hillslope. Soil Sci. Soc. Am. J. 72, 1767–1773. Zhang, J.H., Su, Z.A., Nie, X.J., 2009. An investigation of soil translocation and erosion by conservation hoeing tillage on steep lands using a magnetic tracer. Soil Tillage Res. 105, 177–183. Zhang, J.H., Ni, S.J., Su, Z.A., 2012. Dual roles of tillage erosion in lateral SOC movement in the landscape. Eur. J. Soil Sci. 63, 165–176. Zheng, J.J., Zhang, X.B., He, X.B., 2007. Combining WEPP model with 137Cs to study spatial pattern of soil redistribution on purple slope-land in Sichuan Hilly Basin. J. Soil Water Conserv. 21, 19–23 (in Chinese).