Land Use Affects Soil Water Balance and Soil Desiccation within the Soil Profile: Evidence from the Western Loess Plateau Case

Abstract

This study evaluated the properties of soil water dynamics and desiccation to a depth of 500 cm and tested the idea that land use affects soil drying in deep profiles. Four land use types were chosen: farmland, artificial forest and grassland, and abandoned land. Soil water content was most outstanding under long-term wheat fields, but average soil water content under artificial vegetation of Caragana korshinskii Kom. and alfalfa dropped to 6–8% within the 160–500 cm soil profile, very near to the 7.0% wilting point. Long-term continuous maize cultivation in a fully mulched ridge–furrow system significantly depleted soil water and resulted in a dried soil layer with a thickness of 240 cm. Short-term or long-term land abandonment enhanced soil reservoir and reduced soil water storage deficit degree compared to long-term maize and artificial vegetation. Soil water storage in the 160–500 cm soil profile was depleted by 240 and 464 mm under long-term maize and Caragana korshinskii Kom., respectively, by 267, 319, 381, and 463 mm under 5-, 10-, 20-, and 30-year alfalfa, and by −58, 278, 234, and 93 mm under 5-, 10-, 20-, and 30-year abandonment land, respectively, compared to long-term wheat. Based on the analysis of long-term experimental results, this study shows that the phenomenon of soil drying caused by long-term intensive maize production cannot be ignored in semi-arid areas and that natural re-vegetation under long-term abandonment, rather than artificial vegetation, may be the best type of vegetation reconstruction for this region based on soil water balances.

1. Introduction

Soil water is one of the most critical limiting factors in semi-arid re-vegetation. It is affected by various factors, including topography, soil, land, vegetation, precipitation, and other climatic conditions [1,2,3]. Plants strongly impact soil water because of their different abilities to access, transport, and transpire water [4]. Theoretically, soil water indirectly affects a range of ecological processes, including precipitation [5], precipitation interception, water retention, infiltration, runoff [6,7], soil erosion, solute transport [8], carbon sequestration, soil enzyme activities [9], and land-atmosphere interactions [1,2,10]. Plant water absorption must not exceed water recharge in a healthy plant–soil ecosystem; otherwise, soil dryness would result. Maintaining ecosystem health requires balancing soil water availability and plant water use [11]. Water managers in arid and semi-arid environments face significant difficulty in managing water resources in both time and space.

In the last 50 years, the natural grass, shrub, and forestland of the Loess Plateau have been rapidly transformed into farmland to serve a burgeoning population. The regional natural vegetation was nearly destroyed, resulting in significant soil and water loess and accelerating the degradation and fragmentation processes of farmland. In order to prevent further deterioration of the ecosystem, large-scale vegetation restoration campaigns on the Loess Plateau, such as the eco-project “The Grain to Green”, started by the Chinese government in 1998, attempt to repurpose cropland for forest and grass, resulting in a significant increase in vegetation cover on the Loess Plateau [11]. Caragana korshinskii Kom. and alfalfa (Medicago sativa L.) have become the preferred species for re-vegetation and have been widely planted. Long-term scientific evidence on soil water balance in the semi-arid western Loess Plateau is sparse, and whether forest or grasslands were the dominant natural flora on the Loess Plateau is still debated.

Due to excessive evapotranspiration, the newly introduced artificial re-vegetation drastically depleted soil water in the deep soil profile [12,13,14,15]. For example, in the southeast and north of the Loess Plateau, artificial forest and grassland significantly decreased soil water content. In contrast, deep soil water, on the other hand, cannot be replenished by rain [15,16,17,18,19]. The overconsumption of soil water by artificial re-vegetation has led to severe soil desiccation, resulting in a dried soil layer (DSL) in the soil profile [13,20,21]. Moreover, the DSL cannot recover in the short term in many areas of the Loess Plateau once it is formed [15,20,22]. Because of modifying the processes of water cycles in the soil-vegetation–atmosphere transfer system, the creation of DSLs have a negative impact on ecological and hydrological processes [3,11] and ultimately leads to soil quality degradation and vegetation degeneration [11,23]. According to soil water balances, native grasses may be more viable for restoration in the Chinese Loess Plateau in places with multi-year precipitation less than 440 mm, as suggested by several authors [22,24].

Due to the low economic benefits of agriculture production, agricultural land abandonment has become widespread across China’s Loess Plateau in the last 20 years because of the rise of industrialization. The conversion of cultivated land to abandoned land impacts the soil’s water content [25]. However, there was no uniform view on soil water content under abandonment land. When comparing abandoned fields to cropland, some authors found that soil water content dropped steadily with field age [24,26,27]. In contrast, others found that abandoned land soil water content increased significantly with the increase in abandoned years [28,29]. However, a study reported that as the number of abandoned years increased, soil water content declined initially and subsequently climbed [30]. The debate over these opposing viewpoints is likely to hinder progress in research toward vegetation reconstruction and sustainable water resource management. As a result, long-term research is required to investigate the issue further.

A deeper rooting depth in artificial vegetation means more water is stored and available in the deep soil. However, most studies on root depth remain confined to the uppermost (<3 m) soil layers. Research on deep profiles is scarce, and there have been few studies on processes in the deep soil layers on the semi-arid western Loess Plateau. Understanding the reaction of soil water content to land use is critical for advising the development and adoption of sustainable re-vegetation, considering all of the above elements. As a result, this research aimed to look at the characteristics of soil water dynamics and soil desiccation under various land use scenarios. It was proposed that: (1) the introduced artificial re-vegetation of Caragana korshinskii Kom. and alfalfa deplete soil water and form the DSL; (2) abandoned land increases SW in the short term (≤5 years); and (3) long-term continuous maize production would cause soil dryness in deep soil layers (≥300 cm).

2. Materials and Methods

2.1. Site Description

The research was conducted in a small watershed (35°28′ N, 104°44′ E, 1970 m above sea level) in Lijiabu Town, Aning District of Gansu Province, China (Figure 1). The station is located in the Yellow River’s middle reaches and in the gullies of China’s western Loess Plateau. The experimental site has a medium-temperate semi-arid environment with a mean annual air temperature of 6.4 °C, according to 50 years of meteorological data from the Anding District Meteorological Service (1970–2020). The mean annual cumulative sunshine hours, cumulative temperature (≥10 °C), frost-free duration, and pan evaporation were 2480 h, 2240 °C, 140 d, and 1531 mm, respectively. Long-term average annual rainfall averages 390 mm, with 75% of that falling between June and September. The depth of the water table exceeds 40 m. This soil has a sandy loam texture with approximately 50% sand at the experimental site. According to Chinese soil taxonomy [31], the aeolian soil at the experimental site is classed as Huangmian, a Calcaric Cambisol according to FAO soil classification (2006) [32]. This soil has a sandy loam texture at the experimental site with ≥50% sand. Soil characteristics for the 0–20 cm soil layer, based on analysis of soil samples collected from the experimental site, were 8.4 pH (measured in a suspension of soil in 0.01 M CaCl₂), 9.02 g kg⁻¹ soil organic matter (SOM), and 1.0 g kg⁻¹ total nitrogen (N). The average soil bulk density, permanent wilting point, and field capacity (soil water content at 0.03 MPa) in the site’s 0−300 soil profile are 1.21 g cm⁻³, 0.085 cm³ cm⁻³, and 0.256 cm³ cm⁻³, respectively.

2.2. Experiment Design

In the middle of the rainy season (early August) in 2020, four land use types were chosen: farmland, artificial forest and grassland, and abandoned land. The artificial forest chosen in this study is a 25-year-old Caragana korshinskii Kom. Farmland includes a 12-year-old continuous wheat (Triticum aestivum L.) monoculture field and a 12-year-old continuous high-yielding maize (Zea mays L.) monoculture field. Artificial grassland consists of 5-, 10-, 20-, and 30-year-old alfalfa grassland; alfalfa was the main species in the first 20 years, but succession by Agropyron Gaertn. took place after 20 years. Abandoned land includes 5-, 10-, 20-, and 30-year-old abandoned farmland, where natural plant regeneration was permitted after arable land abandonment, and Agropyron Gaertn. was the main species after abandonment. All land use types were rainfed, and all sampling sites were established on terraces that were built to produce food crops in the 1970s; surface runoff was considered negligible. Because groundwater was buried at a depth of 40–50 m in this region, capillary rise to the root zone and drainage from the root zone were likewise regarded as minimal. The characteristics of water testing sites of different vegetation types are shown in

Table 1. Characteristics of water testing sites of different vegetation types.

Land uses	Aspect	Ground Coverage (%)	Vegetation Types	Dominant Species
Farmland 1	south	–	12-year wheat monoculture	Triticum aestivum L.
Farmland 2	south	–	12-year maize monoculture	Zea mays L.
Artificial forestland	south	50	25-year-old shrub	Caragana korshinskii Kom.
Artificial grassland	south	90	5-year-old alfalfa	Medicago sativa L.
	south	90	10-year-old alfalfa	Medicago sativa L.
	south	90	20-year-old alfalfa	Medicago sativa L.
	south	90	30-year-old alfalfa	Medicago sativa L.
Abandoned land (Natural re-vegetation)	south	75	5-year-old abandoned land	Agropyron cristatum (L.) Gaertn.
	south	85	10-year-old abandoned farmland	Agropyron cristatum (L.) Gaertn.
	south	80	20-year-old abandoned farmland	Agropyron cristatum (L.) Gaertn.
	south	75	30-year-old abandoned farmland	Agropyron cristatum (L.) Gaertn.

.

2.3. Sampling and Measurements

By using an auger, soil samples were taken at 20 cm intervals within the first 200 cm depth and 30 cm intervals from 201 cm to 500 cm (4.5 cm inner diameter). In order to calculate soil water content (SW), the fresh weight of the samples was calculated immediately, and the dry weight was determined after drying in a forced-air oven at 105 °C until a consistent weight was achieved. All soil samples were taken on-site for 20 m × 20 m plots, and each vegetation type was replicated three times. Within soil profiles, soil field capacity (soil water content at 0.03 MPa), bulk density, and permanent wilting point were not directly measured in this study due to the practical difficulty of obtaining a large number of undisturbed sample cores from deep soil profiles. As the groundwater table remained at a depth of about 40 m below the surface, upward flow into the root zone was negligible, and drainage out of the root zone was not considered in this region [33]. According to previous studies, the average soil bulk density (1.20 g cm⁻³), permanent wilting point (0.071 g g⁻¹), and field capacity (0.256 cm³ cm⁻³) in the 0−300 cm soil profile in the site were employed in this study [15,33,34].

The stable soil water content was considered in this study as 60% of the field capacity. The amount of water stored in the soil (mm) was calculated as [15,35]:

$$SWS=\left(SDi\times BD\times SWi\right)\times 10$$

(1)

where SWS is the soil water storage (mm), SDi is the soil depth (m) in the ith soil layer, BD is the soil bulk density (g cm⁻³), and SWi is the gravimetric soil water content of the ith layer at each site (%), respectively.

The relative soil water deficit index (SWDI) was calculated to evaluate the soil water depletion among land use types [36]. The SWDI was calculated as:

$$SWDI=\frac{\left(SWc-SWt\right)}{\left(SWc-SWh\right)}\times 100\%$$

(2)

where SWc is soil water content in the pea–wheat rotation plot (%), SWt is soil water content in the treatment plot (%), SWh is the permanent wilting humidity (%), and the wilting point is defined as occurring when the soil water potential is −1.5 MPa. An SWDI greater than 0 indicates soil water depletion, while an SWDI less than 0 indicates soil water replenishment.

The relative quantity of soil water deficit was calculated as:

$$SWD=SWSi-SWSc$$

(3)

where SWD is the relative quantity of soil water deficit (mm), SWSi is the soil water storage in a given soil layer in the other treatment plot, and SWSc is the soil water storage in a given soil layer in the pea–wheat rotation plot.

The soil water deficit was assessed using the soil water storage deficit degree (SWD) and plant-available water storage (PAW) for each land use type. The degree of soil water storage deficiency was determined using the formulae below [15]:

$$SWD=\frac{\left(Fc-SWS\right)}{Fc}\times 100$$

(4)

where Fc is soil water storage at field capacity (mm).

The following formula was used to compute plant-available water storage in each soil layer [34,37]:

$$PAW=SWS-SWSh$$

(5)

where SWSh is the soil water storage at permanent wilting humidity (mm).

We used five indices to evaluate the extent of DSL in this study, including the soil desiccation index (SDI), DSL formation depth (DSLFD), DSL thickness (cm), mean SWC, and the amount of water deficit in DSL (g kg⁻¹) at each sampling site, to provide an efficient and quantifiable description of soil desiccation intensity. Because the research area’s yearly average infiltration depth is around 160 cm, which corresponds to k = 8, we only looked at the DSL for soil layers below 160 cm in this study.

The SDI was calculated as [38]:

$$SDI=\frac{SWs-SWi}{SWs-SWh}\times 100\%$$

(6)

where SWs is stable soil water content (%). A greater SDI value indicates a higher intensity of soil desiccation. Soil desiccation index was used to distinguish among the extremely desiccated soil (SDI > 100), strongly desiccated soil (75 < SDI ≤ 100), severe desiccation (50 < SDI ≤ 75), moderately desiccated soil (25 < SDI ≤ 50), mild desiccation (0 < SDI ≤ 25), and not desiccated (SDI < 0)

At each sampling site, the mean SW and the quantity of soil water deficit in DSLs were computed as follows [21]:

$$MSW=\frac{1}{x} {{\displaystyle \sum }}_{i=1}^N\mathit{SW}i,(\mathrm{if}\mathit{SW}i-\mathit{SWs}\text{≤}0,i=8,9,10,\cdots N)$$

(7)

$$SWDq={{\displaystyle \sum }}_{i=1}^N\left(SWs-SWi\right)\times BD\times Ti, (\mathrm{if}\mathit{SW}i-\mathit{SWs}\text{≤}0,i=8,9,10,\cdots N)$$

(8)

where MSW is the mean soil water content in DSLs (g kg⁻¹), SWDq is the quantity of water deficit in DSLs, Ti is the thickness of the ith soil sampling layer (cm), and x is the number of soil samples in the DSL (x < N), N is the number of sampled soil layers or soil depths (N = 20).

The following formulae were used to compute the thickness of DSLs (cm) at each sampling site [21]:

$$\mathrm{Thickness}\mathrm{of}\mathrm{DSLs}=Ti\times {{\displaystyle \sum }}_{i=1}^{N}f\left(SWi-SWs\right)$$

(9)

$$f\left(\mathit{SWs}-\mathit{SWi}\right)=\{\begin{array}{c}1, SWi-SWs<0\\ 0, SWs-SWi\ge 0\end{array} (i=8,9,10,\cdots N)$$

(10)

2.4. Data Analysis

For statistical analysis, we used SAS software (SAS Institute, Inc., Cary, NC, USA) and performed an analysis of variance (ANOVA) at a significance level of p ≤ 0.05. In a completely randomized design with three replications, all data were statistically examined using analysis of variance. The Fisher’s protected least significant difference test was used to determine the differences between means.

3. Results

3.1. Precipitation

The precipitation in the 2020 season (from August 2019 to July 2020) was 550 mm (Figure 2). The annual precipitation in the 2020 season was 157 mm less than the long-term average of 393 mm (2000–2019) compared to the long-term average of 393 mm (2000–2019). Based on the Standardization Administration of the People’s Republic of China (2008), the 2020 season (40% of precipitation anomaly) was an extremely wet year for precipitation.

3.2. Soil Water Content and Water Deficit

Soil water content varied among land use (Figure 3). Soil water content in artificial forestland (Caragana korshinskii Kom.) decreased from 9% to 5% in the topsoil (0–80 cm) and then stabilized at around 6% in the 80–500 cm soil layers, very close to the wilting point of 7.0 percent; soil water content in Caragana korshinskii Kom. was always significantly lower in the 0–500 cm soil layers than stable soil water content (Figure 3A). Soil water content under long-term wheat monoculture plot increased first from 12% to 21% in the 0–160 cm soil layers and then decreased to 16% in the 160–320 cm soil layers, and finally stabilized at 16–17% in 320–500 cm soil profiles. Soil water content under a long-term wheat monoculture plot was considerably greater in all soil layers than stable soil water content except for the 40 cm soil layers. Soil water content decreased first in the 0–140 cm soil layers (from 18% to 12%) before stabilizing at 10–12% in the 140–500 cm soil profiles in a long-term maize monoculture experiment (Figure 3A). In the deep soil profiles (140–500 cm), the soil water content in the long-term maize monoculture plot was consistently lower compared to stable soil water content except for the 230–290 cm soil layers.

In artificial grassland of alfalfa, soil water content was always lower than stable soil water content (Figure 3B). Soil water content gradually increased from 8% to 12% in 5-year-old alfalfa in the 0–500 cm soil layers. The soil water content of 10- and 20-year-old alfalfa remained at 8–9% in all soil layers, close to the wilting point of 7.0 percent. Soil water content in 30-year-old alfalfa ranged from 9% to 6%, with a downward trend in the 0–500 cm soil profiles. Overall, as alfalfa grows in age, soil water content declines.

Soil water content was highest in abandoned land that was 5 years old, followed by 30-year abandoned land, and lowest in abandoned land that was 20 and 10 years old (Figure 3C). In 5-year abandoned land, soil water content was increased from 12% to 20% with the increase in soil depth and was always greater than stable soil water content except for the surface 60 cm soil layers. In 10- and 20-year abandoned land, soil water content varied from 8% to 12% and was always less than stable soil water content. Soil water content was enhanced from 9% to 16 % in the 0–140 cm soil profiles and subsequently maintained at around 15% in the 140–500 cm soil layers in 30-year abandoned land; in the 0–100 cm soil levels, soil water content was lower than stable soil water content, but it was higher in the 100–500 cm soil layers.

Plant-available water storage and the degree of soil water storage deficiency were both impacted by land use (

Table 2. Effect of land use on plant-available water storage in different soil profiles.

Vegetation Type	Plant-Available Water Storage (mm)
	0–100 cm	101–200 cm	201–350 cm	351–500 cm	Sum (0–500 cm)
Long-term wheat	71.0 c a	135.3 a	194.2 b	174.6 b	575.1 a
Long-term maize	130.9 a	63.8 d	100.6 d	68.9 e	364.2 b
Caragana korshinskii Kom.	−4.5 h	−17.0 g	−20.0 i	−5.0 g	−46.5 f
5-year alfalfa	23.5 f	40.7 e	62.2 f	83.8 d	210.2 c
10-year alfalfa	23.7 f	27.5 fg	45.1 gh	55.3 e	151.6 d
20-year alfalfa	18.9 f	17.7 g	29.3 h	12.6 f	78.5 e
30-year alfalfa	9.5 g	5.0 h	−16.8 i	−16.0 g	−18.3 g
5-year abandoned land	90.7 b	99.1 c	219.7 a	227.4 a	636.9 a
10-year abandoned land	34.6 de	35.1 ef	56.1 g	82.3 d	208.1 c
20-year abandoned land	25.3 ef	48.1 e	86.2 e	86.8 d	246.4 c
30-year abandoned land	42.8 d	102.1 b	141.5 c	151.8 c	438.2 b

^a Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments.

and

Table 3. Effect of land use on soil water storage deficit degree in different soil layers.

Vegetation Type	Soil Water Storage Deficit Degree (%)
	0–100 cm	101–200 cm	201–350 cm	351–500 cm	0–500 cm
Long-term wheat	38.6 d a	13.8 e	16.1 e	21.2 d	22.4 f
Long-term maize	14.7 e	41.8 c	40.5 c	48.8 c	36.5 c
Caragana korshinskii Kom.	68.6 a	73.4 a	72.0 a	68.1 a	70.5 a
5-year alfalfa	57.8 bc	50.8 b	50.5 bc	44.9 c	51.0 bc
10-year alfalfa	57.9 bc	56.0 b	55.0 b	52.3 bc	55.3 bc
20-year alfalfa	59.7 bc	59.8 b	59.1 b	63.5 ab	60.5 ab
30-year alfalfa	63.2 ab	64.8 ab	71.1 a	70.9 a	67.5 a
5-year abandoned land	31.1 d	28.0 d	9.5 f	7.4 e	19.0 f
10-year abandoned land	53.2 c	53.0 b	52.1 b	45.3 c	50.9 bc
20-year abandoned land	57.0 bc	47.9 bc	44.3 c	44.1 c	48.3 c
30-year abandoned land	50.2 c	26.8 d	29.8 d	27.2 d	33.5 d

^a Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments.

). In topsoil (0–100 cm), the greatest plant-available water storage was found in long-term maize monoculture plot (131 mm), followed by long-term wheat monoculture plot, abandoned land (25–91 mm), and artificial grassland of alfalfa (10–24 mm), and was lowest in artificial forestland of Caragana korshinskii Kom. (−5 mm). With the increase in abandonment years, plant-available water storage first decreased and then increased. The plant-available water storage was most outstanding in the 5-year abandoned land and was lowest in 20-year abandoned land. With the increase in alfalfa growth years, plant-available water storage decreased gradually. In 101–200, 201–350, and 351–500 soil layers, the change in plant-available water storage was parallel to the 0–100 cm soil layer pattern except for wheat; in the 100–500 cm soil layers, long-term wheat monoculture plot had higher plant-available water storage than long-term maize monoculture plot (Table 2). Similarly, the pattern of plant-available water storage was parallel to the change in soil water storage deficit degree among land uses (Table 3).

With changes in land use, the relative soil water deficit index changed (Figure 4). The relative soil water deficit index for long-term maize plots was less than 0 in the 0–120 cm soil layers but greater than 0 in the 120–500 cm soil layers, indicating that, in comparison to long-term wheat plots, soil water was recharged in the 0–120 cm soil layers but depleted in the 120–500 cm layers (Figure 4A). When compared to long-term wheat, artificial vegetation of Caragana korshinskii Kom and alfalfa drained soil water in all soil profiles (Figure 4A,B). Soil water under 5-year abandoned land was recharged in the 60–120 and 260–500 cm soil layers but depleted soil water in other soil layers when compared to long-term wheat plots. Although soil water was depleted in 10–30 years of abandoned lands for all soil layers, the depletion degree of soil water decreased gradually with the increase in abandoned years (Figure 4C).

Overall, compared to long-term wheat, relative soil water storage in the 0–160 cm soil layer was increased by 28 and 4 mm under long-term maize and 5-year abandoned land, respectively, but was reduced by 102, 110, 121, and 136 mm under 5-, 10-, 20-, and 30-year alfalfa and by 93, 99, and 46 mm under 10-, 20-, and 30-year abandoned land, respectively (Figure 5). Caragana korshinskii Kom. depleted soil water storage by 164 mm in the topsoil (0–160 cm) compared to long-term wheat. In the 160–320 cm soil layer, the soil water storage layer was reduced by 114 mm under long-term maize, by152, 173, 186, and 236 mm under 5-, 10-, 20-, and 30-year alfalfa, and by 2, 164, 128, and 63 mm under 5-, 10-, 20-, and 30-year abandoned land, respectively, compared to long-term wheat. When compared to long-term wheat, Caragana korshinskii Kom. reduced soil water storage by 244 mm in the 160–320 cm soil layer. When comparing long-term maize and Caragana korshinskii Kom. to long-term wheat, soil water storage was reduced by 125 and 220 mm in the 320–500 cm soil layer, respectively (Figure 5A). When compared to long-term wheat, the 5-, 10-, 20-, and 30-year alfalfa decreased soil water storage by 115, 147, 196, and 232 mm in the 320–500 cm soil layer, respectively (Figure 5B). The relative change in soil water storage was 60, −114, 106, and −31 mm in the 320–500 cm soil layer under 5-, 10-, 20-, and 30-year abandoned land, respectively (Figure 5C).

3.3. Soil Desiccation Indices

The characteristics of soil desiccation were significantly affected by land use (Figure 6). We only evaluated DSL for soil layers below 160 cm in our study. The DSLs below 160 cm soil layers were not observed in long-term wheat and 5- and 30-year abandoned land. Long-term maize and 10- and 20-year abandoned land appeared in the DSLs; the thickness of the DSLs under long-term maize and 10- and 20-year abandoned land was 240 cm and 340 cm, respectively (

Table 4. Effect of land use on soil desiccation indices.

Vegetation Type	DSLFD a	Thickness of the DSL	Soil Water Deficit in the DSL	Mean SW in the DSL
	cm	cm	mm	g kg−1
Long-term wheat	−	0 c	−	−
Long-term maize	160 a b	240 b c	68.1 c	112.7 a
Caragana korshinskii Kom.	160 a	340 a	263.4 a	62.5 d
5-year alfalfa	160 a	340 a	66.7 c	110.5 a
10-year alfalfa	160 a	340 a	119.1 b	97.7 b
20-year alfalfa	160 a	340 a	180.9 b	83.2 c
30-year alfalfa	160 a	340 a	262.8 a	63.6 d
5-year abandoned land	−	0 c	−	−
10-year abandoned land	160 a	340 a	77.4 c	107.7 ab
20-year abandoned land	160 a	340 a	33.7 d	118.7 a
30-year abandoned land	−	0 c	−	−

^a DSLFD, the dried soil layer forming depth; DSL, dried soil layer; SW, soil water content. ^b We only analyzed the DSL for soil layers below 160 cm in this study because the annual average infiltration depth of the study area is about 160 cm. ^c Different lowercase letters in the same column represent significant differences at p ≤ 0.05 between treatments.

). Soil water deficit in the DSLs under long-term maize and 10- and 20-year abandoned land was 65, 77 and 34 mm, respectively, and mean soil water content in the DSLs was 113, 108, and 119 g kg⁻¹, respectively. All alfalfa formed the DSLs in all soil layers (160–500 cm). The degree of soil drying increased as the age of alfalfa stands increased; moreover, when the age of alfalfa stands reached 20 years, desiccation of the 1.6–5 m soil layers reached an extreme level. The thickness of the DSLs under 5–30 years of alfalfa was 340 cm, and average soil water content in the DSLs was 111, 98, 83, and 64 g kg⁻¹, respectively, under 5-, 10-, 20-, and 30-year alfalfa (Table 4). In the 160–500 cm soil levels, Caragana korshinskii Kom. produced severe DSLs. The thickness of the DSLs and average soil water content in the DSLs were 340 cm was 63 g kg⁻¹ under Caragana korshinskii Kom., respectively (Table 4).

4. Discussion

On the Loess Plateau, the average cover thickness of loess exceeds 200 m [39]. Thus, water storage in the deep soil layer (<200 cm) is an essential component of soil water resources [22]. Land use significantly affected soil water content, which was consistent with earlier research [1,10,25,38]. Artificial shrublands (Caragana korshinskii Kom.) presented an extremely DSL, mainly because the Caragana korshinskii Kom. has a deep root system. For example, Canadell et al. (1996) reported that the global rooting depths of shrubs is 5.1 m [40], but rooting depths of Caragana korshinskii Kom. are found to have roots reaching a depth of 22.4 m on the Loess Plateau [36]. As a result of deep root distribution, the depletion of soil water under Caragana korshinskii Kom. is 628 mm in the 0–500 cm soil layers compared to long-term wheat. Similar responses were reported for forestland s in Southern Spain [40]. It is likely because forestland has great soil water repellency on the soil surface and less soil hydraulic conductivity compared to Abandoned farmland and intensive farmland [41].

Some studies found that long-term maize production under double ridges and furrows mulched with a plastic film system can maintain soil water balance in topsoil (<300 cm) in semi-arid northwest China [42,43,44], but few studies pay attention to water balance in the deeper soil layers (>3 m). Interestingly, our results show that long-term production of maize severely depleted soil water content to a depth of 500 cm, and an apparent dried soil layer is formed; it is likely because maize is a high-water consumption crop compared to other crops, such as spring wheat, potato (Solanum tuberosum L.) and broad bean (Vicia faba L.). High water requirements for the long-term production of maize force plants to extract soil water from deeper profiles; meanwhile, limited rainfall is not enough to replenish soil water [45]. In the long run, the negative balance of soil water led to soil desiccation. Especially in this study, there is no turning point in the soil water content of maize field at a depth of 5 m, so it is speculated that the thickness of dried soil layer in maize field should extend > 5 m deep. The greatest soil water content was found in long-term wheat, mainly because low spring wheat yield saved soil to water because of low evapotranspiration; these results show that wheat is a suitable crop for water sustainability in this region.

The annual evapotranspiration of alfalfa is greater than the annual precipitation on the Loess Plateau [15,16,46]; high water demand forces alfalfa to absorb water from deeper soil layers through deep root distribution [15,37,47], thereby resulting in a severe soil dryness in deep soil, as it has been shown in the studies [15,16]. The high soil water absorption under alfalfa was attributed to its deep root distribution. On the semi-arid Loess Plateau, for example, a study discovered that alfalfa roots extend > 5 m deep after 2–5 years of growth; the greatest root depths under alfalfa are 15.5 m [48]. The loss of soil water under five-year-old alfalfa was 200 mm more than the original soil water storage, according to the previous study [12]. Wang et al. (2021) reported that soil water storage declined by 5 mm year⁻¹ in the topsoil (0–300 cm) [15]. When the age of alfalfa stands reached ≥10 years, the desiccation of the 0–5m soil layer reaches an extreme degree [15]. Compared to farmland, long-term alfalfa production can decrease soil water to a depth of 15 m on the Loess Plateau, and water deficit is challenging to restore [49]. Li (2002) found that after alfalfa grew for more than 8 years, average soil water content within the 300–1000 cm profile decreased from 22.1% to 13.6%, very close to the wilting point of 13.0 % [16]. As a result, the authors proposed that the optimal length of the alfalfa phase and annual crop rotation systems was suggested to be less than 8 years. Similar responses were reported for alfalfa on the Western Loess Plateau [15]. The negative balance of soil water resources caused by overconsumption by deep root vegetation and soil drying occurred in deep soil profiles in artificial shrublands (Caragana korshinskii Kom.) and grasslands (alfalfa), as has been shown in the previous studies [12,13,14,15]. In the long run, the negative balance of soil water would cause soil degradation, vegetation biomass, or stunted growth [3,15,48] and result in localized and regional vegetation degradation [3,11]. Therefore, the formation of dried soil layer under artificial vegetation and long-term maize cannot be ignored in the semi-arid western Loess Plateau.

In this study, the most significant soil water content was found under 5-year abandoned land, suggesting that short-term abandonment promotes soil water storage. On the one hand, it may be because the vegetation is a slow recovery process under natural restoration, so there is less water consumption in the first few years [24,27]; this is likely because the abandonment promotes soil water infiltration as a result of the restoration of vegetation and macro aggregate formation and increasing aggregates stability [50]. With the increase in abandoned years, soil water content decreased first at 5–20 years and then increased after 20 years, in agreement with previous studies in the west and southwest of the Loess Plateau [30,51] and in Southern Spain [41]. It is likely because vegetation growth gradually increased with the increase in abandoned years in the first 5 to 20 years (i.e., ground coverage), thus increasing soil water consumption and depleting soil water, resulting in the dried soil layer. Soil water content decreased steadily with field age after their abandonment in the south of the Loess Plateau (annual rainfall ≈ 1100mm), mainly because dwarf shrubs (e.g., Lespedezadahurica) and short rhizome grass (e.g., Bothriochloa ischaemum) appeared at the mid-succession stage and gradually increased in abundance during succession in their study [26]. In contrast, another study showed that soil water content of abandoned land increased significantly with the increase in abandoned years in the 0−400 cm soil layers [28]. This discrepancy may be due to the difference of dominant species in plant communities after abandonment since some studies have been conducted in the north of the Loess Plateau, where the dominant species in abandonment land is Agropyron cristatum (L.) Gaertn., Artemisia giraldii, Bothriochloa ischcemum, and Stipa bungeana [28]. Compared to 10- and 20-year-old abandoned land, soil water content increased, and soil desiccation degree decreased when the cultivated land was abandoned for 30 years. It is likely because the vegetation reached a stable period (e.g., ground coverage) after they went through a rapid recovery period, and more roots improved soil infiltration capacity and increased soil water storage after long-term abandonment [7,52]. These results suggest that except for long-term low water consumption crops, such as wheat, natural re-vegetation should be the best land use type for this region based on the soil water balances observed in this long-term field study.

5. Conclusions

From the analysis of long-term experimental results, this study shows that long-term continuous maize under a fully mulched ridge–furrow system significantly depleted soil water and caused soil drying; however, long-term wheat had the greatest soil water conditions, which suggests that long-term low water consumption crop production, such as wheat, is the best alternative for farmland use. In addition, to maintain high land productivity, a low water consumption crop (e.g., wheat)–maize rotation is needed for maize production in the <400 mm precipitation zone. Artificial vegetation of Caragana korshinskii Kom. and alfalfa resulted in extreme soil dryness and formed a dried soil layer, but short-term or long-term abandonment of land enhanced “soil reservoirs”, suggesting that natural re-vegetation under long-term abandonment is the best type of vegetation reconstruction rather than artificial vegetation for this region based on the soil water balances.