Effects of the Gully Land Consolidation Project on soil properties and carbon stocks of farmland in the Chinese Loess Plateau

Description of the study area

The Gutun watershed (36°46′39″–37°3′34″N, 109°41′02″–109°56′58″E) is located 46 km east of Yan’an City, Shanxi Province (Fig. 1). It covers approximately 24 km² and has an average elevation ranging from 956 to 1220m above sea level. The region is characterized by a continental monsoon climate with an annual mean temperature and precipitation of 9.7 °C and 541 mm, respectively. 70% of the precipitation falls between June and September (Wang et al., 2008; Yu et al., 2019a; Zhao et al., 2019b). According to the World Reference Base for Soil Resources (WRB) classification system, the soils in the watershed primarily belong to “Calcic Cambisols”, which are typical soils of the CLP.

The Gutun watershed is not only a typical representation of the gully terrain in the CLP but also a successful representation of the GLC project. According to field surveys, as early as the 1940s, local farmers undertook small-scale land reclamation through constructing sluice ditches and incising slopes to fill gullies, in order to expand arable land and enhance grain production. In 2011, the government officially named these land development measures “Gully Land Consolidation” (GLC) and initiated large-scale implementation. As the GLC project progressed over different periods, a series of newly reclaimed farmland with varying cultivation ages has been established in the watershed, including farmland cultivated for 5 years (R5), 15 years (R15), 35 years (R35), 60 years (R60), and at least 70 years (R70). The cultivation ages of these farmland were verified through interviews with local elders. Except for the R35, which was planted as a forest nursery, the others were planted with maize (Yu et al., 2019a).

Sample collection and laboratory analysis

To ensure that the collected soil samples represented all farmland treated by the GLC project in the Gutun watershed, we collected soil samples from farmland with varying cultivation ages, all of which have been artificially created since the 1940s. Soil samples were collected from three 1 m × 1 m quadrats in each type of reclaimed farmland located by GPS from March to April 2017, except R70, because only two 70-year-old farmland were available. Before sampling, all litter was removed from each sampling site. Every site was excavated to a depth of 500 cm until the calcification or overflow layer was reached (Yu et al., 2019a). Undisturbed soil samples were collected using 275 soil core rings with a diameter of five cm. BD was measured every 10 cm from 0 to 20 cm depth and every 20 cm from 20 to 500 cm depth. Another 275 samples were also collected to measure SWC, particle size, TC, and SIC content.

BD was calculated based on the volume-mass relationship of each soil sample dried to a constant weight at 105 °C (Wang, Shao & Liu, 2013). The calculation formula is as follows: (1)${\displaystyle BD=\frac{M_d}{V_d}}$

where, M_d and V_d represent the dry mass and volume of the soil after being dried to a constant weight, respectively.

A small portion of each soil sample was dried at 105 °C to a constant weight, and SWC was determined based on the mass loss before and after drying (Zhao et al., 2019b). The remaining soil samples were air-dried for the particle size, TC, and SIC content analysis.

Weighed air-dried soil of approximately 0.2 g and added 10 mL of 10% H₂O₂, then heated on a 95 °C hot plate to remove organic matter until bubbling ceased. Next, 10% hydrochloric acid (HCl) was added to remove carbonates, continuing until no further bubbling occurred. The sample was then washed with distilled water until neutral, left to stand for 12 h, and the supernatant was decanted. Finally, 10 mL of 1 mol L⁻¹ sodium hexametaphosphate was added, and the sample was sonicated for 10 min. Soil particle size was measured using a Mastersizer 3000 (Malvern Instruments, England). Soil particles were classified as clay (<2 µm), silt (2–50 µm), and sand (50–2,000 µm) based on particle size.

The remaining air-dried soil samples were sieved through a 2 mm mesh. 10 mg of each sieved sample was weighed, and wrapped in aluminum foil. TC content was measured using an elemental analyzer at a combustion temperature of 950 °C (Vario EL cube, Elemental Analyzer, Elementar, Germany). SIC refers to the carbonate minerals in the soil, such as CaCO₃ and MgCO₃ (Wu et al., 2009; Liu et al., 2017), it can be determined by measuring the CO₂ released during acid hydrolysis of the soil sample. Approximately 10 g of sieved soil was placed in a Y-tube, with 2 mL of concentrated phosphoric acid added to the other side. The tube was sealed and vacuumed to 10⁻³ atm, allowing the sample and phosphoric acid to mix. The mixture was heated at 110 °C for 1 h on a hot plate, and the CO₂ released during the process was measured using a vacuum line. The SIC content was calculated from the CO₂ released. The SOC content was obtained by subtracting SIC from TC (Yu et al., 2019a). All experiments were conducted at the Xi’an Accelerator Mass Spectrometry Center.

Calculation of carbon stocks

Carbon stock, i.e., carbon mass per unit area for a given depth, was calculated according to the following equations (Grimm et al., 2008): (2)${\displaystyle SO{C}_{stock}=SOC\times BD\times \left(1-F_{content}\right)\times \Delta d\times 0.01}$ (3)${\displaystyle SI{C}_{stock}=SIC\times BD\times \left(1-F_{content}\right)\times \Delta d\times 0.01}$ (4)${\displaystyle T{C}_{stock}=SO{C}_{stock}+SI{C}_{stock}}$

where, SOC_stock, SIC_stock, and TC_stock represent SOC, SIC, and TC stock (kg m⁻²), respectively; SOC and SIC represent SOC and SIC content (g kg⁻¹), respectively; BD is soil bulk density (g cm⁻³); F_content is the proportion of non-soil particles > 2 mm diameter (dimensionless). The occurrence of coarse fragments in the loessal soil of the CLP is rare, so F_content is usually negligible; Δd is the thickness of the layer (cm). Carbon stock per unit area for a given depth was calculated by summing carbon stock over all layers.

Data classification and sources

The soil data of farmlands analyzed in this study were categorized into three groups based on whether the study area had been treated by the GLC project: (1) data from areas treated by the GLC project in the Gutun watershed (GT-T), (2) data from untreated areas within the CLP (CLP-U), and (3) data from untreated areas outside the CLP (Other-U). Detailed information is provided in Table 1.

The CLP-U areas include the untreated areas in the Gutun watershed (GT-U), Linfen-Yanchi climatic zones, Huining, Lanzhou, Yangling, Jiegou watershed, Longtan watershed, Laoyemanqu watershed, Fufeng, and Yongshou. The farmland soil data from these areas were obtained from relevant references in Table 1.

The Other-U areas include the Yanqi Basin, Urumqi, Hebei Plain, the upper Yellow River Delta, Quzhou, Zhengzhou, Northeast China Plain, Lishu, and Sanjiang Plain. The farmlands soil data from these areas were also obtained from relevant references in Table 1.

Statistical analysis

Basic data processing was conducted using Microsoft Excel 2021. The sampling map was drawn using ArcMap 10.2. Statistical significance of differences among certain parameters was tested using one-way analysis of variance (ANOVA) in SPSS 26.0, with a significance level set at p < 0.05. Data visualization was performed using Origin 2025.

The GLC project changed the farmland soil texture

The mean proportions of sand, silt and clay in the farmland soil at depths of 0–500 cm in the GT-T watershed were 21.0%, 76.0% and 3.0% respectively (Fig. 2A), and the soil texture was predominantly silty loam (Fig. 2B). From R5 to R60, the sand proportion increased from 21.1% to 30.5%, while the silt and clay proportions decreased from 76.0% and 3.0% to 67.3% and 2.2%, respectively. By R70, the proportions of clay, silt, and sand had returned to the levels observed in R5 (p > 0.05) (Table 2). The proportions of sand, silt, and clay in the 0–100 cm layer were 24.5%, 73.0%, and 2.5%, respectively. In the 0–300 cm layer, these proportions were 22.7%, 74.5%, and 2.8%, and in the 0–500 cm layer, they were 21.5%, 75.6%, and 2.9%, respectively. With increasing depth, the proportion of sand decreased, while the proportions of silt and clay increased (Table 3).

As shown in Fig. 2A, the proportion of silt in the 0–100 cm (74.0%), 0–300 cm (74.0%), and 0–500 cm (76.0%) soil layers was the highest in the GT-T watershed compared to untreated areas (the CLP-U and Other-U), which was 1.0 to 1.4, 1.3 to 2.2, and 1.1 to 2.1 times higher than in these areas, respectively. In the 0–100 cm layer, the soil texture of farmland in the GT-T watershed was similar to that in the Hebei Plain, Yanqi Basin, and Laoyemanqu watershed, all of which were silty loam (Fig. 2B). In the 0–300 cm profile, the soil texture of farmland in Linfen-Yanchi climatic zones of the CLP-U areas was sandy loam (Fig. 2B), which differed significantly from that in the GT-T watershed. In the 0–500 cm profile, the soil texture of farmland in the GT-T, GT-U, and Laoyemanqu watersheds were silt loam, with the GT-T watershed exhibiting the highest silt proportion and the lowest proportions of clay and sand (Fig. 2B). Compared to farmland in the Guanzhong areas of the CLP, such as Yongshou and Fufeng (loam), significant differences in soil texture were observed (Fig. 2B).

The results above demonstrate that the GLC project has significantly altered soil composition in newly reclaimed farmland, manifesting as decreased proportions of clay and sand alongside increased silt content. This textural evolution has led to a trend in soil texture shifting from silt loam to silt, thereby enhancing soil and water conservation capacity. These textural modifications were primarily attributed to anthropogenic disturbances during loess backfilling operations and subsequent mechanical compaction processes. Specifically, these anthropogenic engineering practices disintegrated soil coarse aggregates into finer particulates, thereby inducing a decrease in both soil porosity and particle diameter (Ma et al., 2020).

The GLC project enhanced the water conservation capacity of deeper soil layer

There were no significant differences in SWC at depths of 0–500 cm among R5 (17.3 ± 3.4%), R15 (19.8 ± 3.9%), R60 (19.2 ± 7.9%), and R70 (18.1 ± 2.8%) (p > 0.05), but they were significantly higher than R35 (13.9 ± 25 g kg⁻¹) (p < 0.05) (Table 2). The SWC in the 0–100 cm layer of farmland in the GT-T watershed was 14.3 ± 3.7%, slightly lower than that in the Northeast China Plain (15.5 ± 5.7%), but higher than that in the CLP-U areas, including the GT-U, Jiegou, Longtan, and Laoyemanqu watersheds (Table 1, Fig. 3A). The SWC in the 0–300 cm and 0–500 cm depths of farmland in the GT-T watershed were 16.5 ± 2.2% and 17.7 ± 5.2%, respectively, which was 1.1–1.9 times and 1.1–2.0 times higher than that in the corresponding depths of farmland in the CLP-U areas (Fig. 3A). Thus, across the entire 500 cm depth, the SWC in the GT-T watershed remained at a higher level.

To investigate the vertical profile variation of SWC in the 0–500 cm profile of farmland in the GT-T watershed, the SWC at each 100 cm depth interval was analyzed and compared with that in the CLP-U areas, as shown in Fig. 3B. SWC in the GT-T watershed increased with depth and was 1.2 to 1.6 times higher in the 100–500 cm layers compared to the surface (0–100 cm), showing significantly greater moisture retention than CLP-U areas (Fig. 3B, Table 3). In contrast, in the CLP-U areas, such as the Longtan and Laoyemanqu watersheds, SWC decreased with increasing depth, while the GT-U and Jiegou watersheds showed little variation. The mean SWC across the 0–500 cm profile in the GT-T watershed (18.6%) was 1.6–2.8 times higher than that in the CLP-U areas, such as the GT-U (10.5%), Jiegou (11.3%), Longtan (6.6%), and Laoyemanqu (8.3%) watersheds.

In the GT-T watershed, soil water storage in the 100–500 cm layer (2,606 mm) accounted for 85% of the total soil water storage (3,054 mm) in the 0–500 cm depth. This indicated that the GLC project promoted greater water stored in the deeper soil (100–500 cm). This may be attributed to the land in the gully, filled to a depth of approximately 2 m by the GLC project, which not only reduced the residence time of runoff from the adjacent slopes, but also decreased the runoff volume and velocity within the gully, thereby promoting more water to infiltrate and be stored in the deeper soil. Additionally, a significant negative correlation was observed between SWC and the sand proportion, while a significant positive correlation was found with the silt proportion (Fig. 4). These results suggest that coarser soil particles in the surface layers hinder water conservation. In contrast, a higher silt proportion in the deeper layers provided favorable conditions for soil water storage after the GLC project implementation.

The GLC project increased soil BD in farmland

Soil BD in the 0–500 cm depth of farmland in the GT-T watershed varied from 1.0 to 1.8 g cm⁻³ (mean = 1.5 ± 0.2 g cm⁻³). Significant differences in BD were observed among farmland with different cultivation ages (p < 0.05). The lowest BD value was found in R35 (1.2 ± 0.1 g cm⁻³), while R70 exhibited the highest BD value (1.6 ± 0.2 g cm⁻³) (Table 2). However, no significant variations in BD were observed with increasing depth (p > 0.05) (Table 3).

The BD of 0–100 cm (1.4 g cm⁻³) and 0–300 cm (1.5 g cm⁻³) soil layers in the GT-T watershed was 1.0 to 1.2 and 1.1 to 1.3 times higher than in untreated areas, respectively. Specifically, the BD of the 0–100 cm soil layer in the GT-T watershed was similar to that in the GT-U watershed (1.4 g cm⁻³) and Northeast China Plain (1.4 g cm⁻³), but higher than that in the Other-U areas, such as Yanqi Basin (1.3 g cm⁻³), Sanjiang Plain (1.3 g cm⁻³), and Huining county (1.2 g cm⁻³) (Fig. 5A, Table 1). In the 0–300 cm, the mean BD was slightly higher in the GT-T watershed than that in the GT-U watershed (1.4 g cm⁻³), and significantly higher than that in the Linfen-Yanchi climatic zones (1.2 g cm⁻³) (Fig. 5A, Table 1). The variation in BD with depth in the GT-T and GT-U watersheds was analyzed (Fig. 5B). It was observed that the mean BD in the 0–100 cm layer in the GT-T watershed was slightly higher than that in the GT-U watershed. However, this difference became significantly more pronounced in the 100–300 cm layer.

The higher soil BD values in the GT-T watershed can be attributed to soil compaction during the GLC project, caused by machinery (Håkansson & Lipiec, 2000; Hamza & Anderson, 2005; Lu et al., 2019). During compaction, larger soil particles were broken into smaller particles, reducing the pore space between particles and resulting in denser soil. As shown in Fig. 4A, a significant negative correlation between BD value and sand proportion, while a significant positive correlation with silt and clay proportions, further supports the idea that soils with smaller particles are more compact.

Is a higher BD value an advantage or a disadvantage? According to the study by Dong et al. (2021), a BD value of 1.5 g cm⁻³ was considered optimal for cultivated land, as it promoted crop growth and optimal yield. In this study, the mean BD values of the 0–100 cm layer (1.4 g cm⁻³) and 0–300 cm layer (1.5 g cm⁻³) in the GT-T watershed were closer to this optimal value compared to farmland in the CLP-U and Other-U areas, which may be more favorable for crop growth and achieving optimal yield. However, the relationship between BD and soil health remains debated, because some studies indicated that lower BD values typically reflected better soil structure, facilitating gas exchange and improving permeability (Zhang et al., 2018; Lu et al., 2019). Therefore, further investigation is necessary to determine the optimal BD value for farmland in the GT-T watershed.

The GLC project increased the SICS: SOCS ratio in the 0–100 cm soil layer

In the 0–500 cm soil profile, the TC contents of farmland in the GT-T watershed, ranked from high to low were as follows: R35 (15.9 g kg⁻¹) > R5 (14.9 g kg⁻¹) > R60 (15.1 g kg⁻¹) > R70 (14.3 g kg⁻¹) > R15 (12.0 g kg⁻¹) (Table 2). The mean TC content was 14.4 ± 3.4 g kg⁻¹, with SOC and SIC contents of 2.0 ± 1.1 g kg⁻¹ and 12.4 ± 3.2 g kg⁻¹, respectively. The TCS values of 0–500 cm depth ranked from high to low were as follows: R5 (117.7 kg m⁻²) >R60 (106.8 kg m⁻²) > R35 (97.3 kg m⁻²) > R70 (94.2 kg m⁻²) > R15 (76.2 kg m⁻²) (Table 2). The mean TCS value was 98 ± 15.4 kg m⁻², with SOCS and SICS values of 14.6 ± 3.7 kg m⁻² and 83.8 ± 13.1 kg m⁻², respectively. The SICS:SOCS ratio was 5.7, indicating that the carbon storage was predominantly in the form of SIC (Table 3). With increasing depth, both SICS and TCS decreased at each 100 cm layer, while the SOCS of each soil layer below 100 cm was higher than that in the 0–100 cm layer (Table 3).

In the 0–100 cm layer, SOCS and SICS in the GT-T watershed were 2.5 kg m⁻² and 19.0 kg m⁻², respectively. Compared to the CLP-U and Other-U areas, SOCS of farmland in the GT-T watershed was the lowest, while SICS was relatively higher (Fig. 6A). In the 0–300 cm layer, SOCS and SICS in the GT-T watershed were 8.7 kg m⁻² and 55.5 kg m⁻², respectively, similar to the values found in Linfen-Yanchi climatic zones (Fig. 6A). In the 0–500 cm layer, SOCS and SICS in the GT-T watershed were 14.6 kg m⁻² and 83.8 kg m⁻², respectively, showing significant differences with that in Guanzhong areas, such as Yongshou and Fufeng (Fig. 6A). The TCS in the 0–300 cm and 0–500 cm layers in the GT-T watershed were 3 times and 4.6 times higher than that in the 0–100 cm layer, respectively. The SOCS in the GT-T watershed was generally lower than that in the CLP-U and Other-U areas, except for the Linfen-Yanchi climatic zones in the 0–300 cm soil layer.

In the 0–100 cm layer, the SICS:SOCS ratio reached 7.6 in the GT-T watershed, which was significantly higher than in deeper layers and all comparison areas (Fig. 6B). This can be attributed to the lower SOCS values. Compared to the CLP-U areas (such as Yangling, Lanzhou, and Linfen-Yanchi climatic zones) and Other-U areas (such as Lishu, the Northeast China Plain, Zhengzhou, Quzhou, the upper Yellow River Delta, the Hebei Plain, Urumqi, and the Yanqi Basin), the SOCS in the 0–100 cm depth in the GT-T watershed decreased by 0.7–10.2 kg m⁻² (Fig. 6A), but the SICS:SOCS ratio was 1.3-19 times higher than that in these areas (Fig. 6B). These results indicated that SOC loss occurred in the 0–100 cm soil layer of farmland in the GT-T watershed, leading to a significant increase in the SICS:SOCS ratio. Furthermore, compared to the SOC content in the 0–100 cm layer in the GT-U watershed (3.2 ± 0.4 g kg⁻¹) (Zhao, 2022), the SOC content in the GT-T watershed (2.0 ± 1.1 g kg⁻¹) decreased by 1.3 g kg⁻¹, further confirming that the implementation of the GLC project resulted in SOC loss in the 0–100 cm layer.

The SOC loss in the 0–100 cm layer in the GT-U watershed was likely due to the heavy machinery compaction of backfilled soil during the implementation of the GLC project, which disrupted soil macroaggregates in the surface layer. The disruption of macroaggregates increased the contact between soil organic matter and microorganisms, thereby accelerating the mineralization of SOC. Additionally, the insufficient input of external organic matter during the geomorphic reshaping process further contributed to the significant loss of SOC in the surface soil (Ahirwal & Maiti, 2016; Liu et al., 2024). The SOC loss may indicate a reduction in soil organic matter and available nutrients (Liu et al., 2024), which is detrimental to crop growth. Therefore, in the management of newly reclaimed farmland, special attention should be given to the protection and replenishment of SOC in the 0–100 cm soil layer.