Catena 137 (2016) 52–60

Contents lists available at ScienceDirect

Catena

j ourna l homepage: www.e lsev ie r .com/ locate /catena

Interaction of soil water storage dynamics and long-term naturalvegetation succession on the Loess Plateau, China

Yong-wang Zhang a, Lei Deng b, Wei-ming Yan b, Zhou-ping Shangguan a,⁎a State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil andWater Conservation, Chinese Academy of Sciences andMinistry ofWater Resources, Yangling,Shaanxi 712100, Chinab Institute of Soil and Water Conservation, Northwest A&F University, Yangling, Shaanxi 712100, China

⁎ Corresponding author.E-mail address: shangguan@ms.iswc.ac.cn (Z. Shanggu

http://dx.doi.org/10.1016/j.catena.2015.08.0160341-8162/© 201 Elsevier B.V. All rights reserved.5

a b s t r a c t

a r t i c l e i n f o

Article history:Received 23 April 2015Received in revised form 7 July 2015Accepted 26 August 2015Available online 16 September 2015

Keywords:DynamicsSoil waterSoil clay and siltThe Loess PlateauVegetation restoration

Soil water is a key terrestrial water resource, particularly in arid and semi-arid regions of the world such as theLoess Plateau of China. Information on the dynamics of soil moisture following vegetation restoration is essentialfor managing water resources and can be helpful for adjusting relevant government policies. To evaluate the re-sponse of soil water storage (SWS) to long-term natural vegetation succession (~160 a), we examined the soilmoisture for different restoration ages in the Ziwuling forest region, which is located in the central part of theLoess Plateau. Our results showed that the SWS decreased with long-term natural vegetation restoration. Thebulk density (BD), soil water content (SWC), and clay and silt content presented similar trends to those of theSWS throughout the entire vegetation succession. The SWS was significantly and positively correlated with theSWC and aeration porosity (P b 0.05). The SWC was lower in the upper soils (0–50 cm) than in the deepersoils (N50 cm) at every restoration stage, as was the SWS; however, the SWS in the 200–300 cm soil layer wasthe highest (164.61–212.80 mm) compared to other layers in all restoration stages. These results are expectedto help improve the understanding of the response of deep soil water to long-term natural vegetation restorationand to provide insights into the dynamics of deep soil water influenced by vegetation.

© 201 Elsevier B.V. All rights reserved.5

1. Introduction

Vegetation succession can lead to the recovery of deteriorated soilproperties (Jia et al., 2005; Zhao et al., 2010). To control soil erosionand ecosystem degradation, China's government has initiated severalvegetation restoration projects on the Loess Plateau due to its infamouserosion (Deng et al., 2012); thus, a large area of agricultural land in theLoess Plateau has been converted into other uses during the past fewdecades. For example, farmland has been converted into grasslands,shrublands or forests with natural vegetation (Deng et al., 2014; Fenget al., 2013; Fu et al., 2006; Zhou et al., 2012). The study of recovery pro-cesses is of great significance, as it could deepen the understanding ofthe relationship between the succession of vegetation and the evolutionof soil ecological function and provide recommendations for eco-environmental reconstruction or rehabilitation.

Soil water is a significant terrestrial water resource, particularly inarid and semi-arid regions of the world such as the Loess Plateau ofChina, where groundwater is buried below the thick unsaturated loess-ial soil (Tsunekawa et al., 2014) and concentrated precipitation is lost inthe form of overland flow (Yang, 2001). The quantity of the soil–waterresources depends to a large extent on soil depth. Water resources in

an).

deep soil profiles are relatively stable for vegetation growth due to theinsulating effect of the upper soil (Wang et al., 2012b). Water resourcesmay be used differently by the various vegetation types (forest, grass-land, and crops) due to differences in the distribution of root systems,the characteristics of transpiration, and the amount of water taken upby roots (Wang et al., 2012b). Large spatial variation in deep soilwater may thus be induced by consumption by vegetation (Yang et al.,2012a, 2012b). Additionally, Yang et al. (2014) reported that soilwater decreases drastically after vegetation restoration, with no signifi-cant difference in near-surface soil moisture among the vegetationtypes but significant differences in the sub-surface and deep soil layers.Previous studies have mainly focused on soil water in the shallowlayers, whereas soil water in deeper layers has largely been ignoreddue to the high costs of labor and time required for such investigation(Gao and Shao, 2012; Tombul, 2007; Wang et al., 2013; Zhu et al.,2009; Zhu, 2014. Therefore, water resources in deep soil profiles playan important role in ensuring a well-established vegetation cover insemi-arid regions, and understanding the response of deep soil waterto vegetation restoration is essential for estimating the productivityand sustainability of semi-arid ecosystems.

The recovery of vegetation is very important in accelerating theamelioration of soil quality. Soil physical properties are usually recog-nized as important soil quality indicators (Boix-Fayos et al., 2001).Much research has been done recently on the influence of vegetation

53Y. Zhang et al. / Catena 137 (2016) 52–60

recovery or different land-use patterns on soil physical properties (Fuand Chen, 2000; Fu et al., 2003; Stolte et al., 2003). However, changesin soil physical properties are still under study during the long-termrecovery of vegetation. Research into changes in soil physical propertiesis considered necessary for understanding the ecological consequencesof vegetation recovery. In the semi-arid area of the Plateau, vegetationrecovery or reconstruction is always limited by the shortage ofwater re-sources. Transformation of precipitation into soil water is also influ-enced by soil physical properties, such as bulk density, porosity, andhydraulic conductivity. There is little literature concernedwith this par-ticular issue, especially the long-term change of soil physical propertiesunder natural restoration in the Plateau. Recently, Chinese scientistshave increased the attention paid to the succession of secondary forestson the Loess Plateau (Cheng et al., 2012; Jia et al., 2005; Wang et al.,2010a; Zhao et al., 2010). While many studies have focused on changesin the aboveground vegetation of secondary forests in the central part ofthe Loess Plateau (Wang et al., 2010a; Zou et al., 2002) and soil carbonand nitrogen dynamics (Jia et al., 2005; Deng et al., 2013, 2014), fewstudies have focused on changes to soil water following long-term nat-ural vegetation restoration, although studies have been done on soilwater in other regions of the Loess Plateau (Chen et al., 2008; Jia andShao, 2013). In addition, Cao et al. (2009) reported that the conse-quences of large-scale vegetation restoration have been associatedwith an increased severity in water shortages; therefore, informationon the dynamics of soil water following long-term vegetation restora-tion is essential for managing water resources and would be helpfulfor adjusting relevant government policies.

In this study, we hypothesized that the soil water storage variedwith long-term natural vegetation restoration ages throughout succes-sion. In the Ziwuling forest region of the Loess Plateau, there is an intactseries of naturally recovering vegetation succession. We chose thisstudy area to provide a scientific foundation for constructing the eco-environment and rehabilitating thewater storage and regulation capac-ity of the soil reservoir. Therefore, the specific objectives of the studywere to investigate the following: (1) the dynamics of deep soil waterstorage with the succession of long-term natural vegetation restorationfromgrassland to forests, (2) the effects of natural vegetation successionon the soil water content, and (3) the soil factors affecting soil waterstorage.

Fig. 1. Location of the study s

2. Materials and methods

2.1. Study area

The study was conducted at the Lianjiabian Forest Farm of theHeshui General Forest Farm of Gansu (35°03′–36°37′ N, 108°10′–109°18′E, 1211–1453 m a.s.l.), covering a total area of 23,000 km2 inthe hinterland of the Loess Plateau in the Ziwuling forest region(Fig. 1). The area's annual temperature is 10 °C, annual rainfall is587mm, accumulative temperature is 2671 °C, and annual frost-free pe-riod is 112–140 days. The region's soils are largely loessial, havingdevel-oped from primitive or secondary loess parent materials, which areevenly distributed at thicknesses of 50–130 m above red earthconsisting of calcareous cinnamon soil (Jia et al., 2005). Soil pH rangesfrom 7.92 to 8.31. The area is covered in species-rich uniform forestswith a forest canopy density ranging between 80% and 95% (Chenget al., 2012). The natural biomes of the region are deciduous broadleafforests, of which the climax vegetation is the Quercus liaotungensisKoidz. Throughout the region, Populus davidiana Dode and Betulaplatyphylla Suk communities dominate the pioneer forests; Sophoradavidii (Franch.) Skeels, Hippophae rhamnoides (Linn.), Rosa xanthinaLindl and Spiraea pubescens Turcz are the main shrub species; andBothriochloa ischaemum (Linn.) Keng, Carex lanceolata Boott, Potentillachinensis (Ser) and Stipa bungeana Trin are the main herb species.Based on previous research in the study area, secondary forests natural-ly regenerated on abandoned land from grassland to shrubland and cli-max forest (Q. liaotungensis) over approximately 150 years (Wang et al.,2010a; Table 1).

2.2. Experiment design and soil sampling

A field surveywas undertaken between July 15 andAugust 15, 2014.The sampling areas were determined according to the size of the com-munities. There were five 20m× 20m plots chosen in each forest com-munity, five 5 m × 5 m plots in the shrub communities, and five2 m × 2m plots in the herbaceous communities. The largest relative el-evation difference between two plots was less than 120 m. Most of theplots faced north and had a slope gradient of less than 20°. All the soilssurveyed developed from the same parent materials and have had

ite in the Loess Plateau.

Table 1Geographical and vegetation characteristics at different restoration stages in the Ziwuling forest region of the Loess Plateau. G1 and G2 represent the grass restoration stages, S stands forthe shrub restoration stage, F1 represents the early forest stage, and F2 represents the climax forest stage. Numbers in parentheses following the succession stage are the ages since crop-land abandonment. G, S and F stand for grassland, shrub and forest, respectively.

Successional stageLatitude Longitude Altitude Aspect Slope Coverage

Main plant species(N) (E) (m) (°) (%)

G1 (10 a) 36°05′04.0″ 108°31′37.4″ 1348 NE 14 85 Lespedeza bicolorG2 (30 a) 36°05′08.8″ 108°31′38.9″ 1365 NE 8 85 B. ischaemumS (50 a) 36°04′14.4″ 108°32′01.4″ 1354 NE 18 90 H. rhamnoidesF1 (110 a) 36°03′05.3″ 108°32′31.8″ 1437 NE 10 90 P. davidiana, Q. liaotungensisF2 (160 a) 36°02′57.5″ 108°32′13.7″ 1449 NE 18 95 Q. liaotungensis

Fig. 2. Soil bulk densities (BDs) in the different soil layers in each restoration stage (seeTable 1). Values are in the form of the Mean ± SE, and the sample size n = 5. Differentlower-case letters above the bars mean significant differences in the same soil layers atthe different restoration stages (P b 0.05).

54 Y. Zhang et al. / Catena 137 (2016) 52–60

vegetation for varying lengths of time. Five soil sites were selected fromgrowing vegetation approximately 10, 30, 50, 110 and 160 years old,and all the stages of succession discussed are the result of natural reveg-etation. The distance between the soil sites was approximately 1 km.Basic site information is shown in Table 1.

Soil samples were taken at five points. These were the four cornersand center of the soil sampling sites described above. Soil sampleswere taken using a drill to a depth of 1 m at 10-cm intervals and of 1–5 m at 20-cm intervals. Samples were stored in sealed aluminumcases to prevent potential moisture loss before they were transportedto the laboratory formeasurement of soil water content using the gravi-metric approach,which uses oven drying for 24h at 105 °C. Undisturbedsoil cores were collected using a soil bulk sampler with a 5.0 cm diame-ter and a 5.0 cm high stainless steel cutting ring (3 replicates) for mea-suring soil bulk density and porosity in the top soil layer (0–50 cmat 10-cm intervals). To measure particle size, disturbed soil samples from thetop soil layer were sieved through a 2 mm screen, and roots and otherdebris were removed; each sample was air-dried and stored at roomtemperature.

2.3. Laboratory assay

Soil water content (SWC) was measured gravimetrically andexpressed as a percentage of soil water to dry soil weight (Jia et al.,2012). Soil bulk density (BD)was calculated depending on the inner di-ameter of the core sampler, sampling depth and oven dried weight ofthe composite soil samples (Jia et al., 2005). Total porosity (TP), inactiveporosity (IP), aeration porosity (AP) and capillary porosity (CP) werecalculated using the saturated moisture content, fieldmoisture capacityand wilting water content (Ghanbarian-Alavijeh andMillán, 2009). Soilparticle sizes were determined using theMasterSizer 2000method. Theproportions of the clay (b0.002 mm), silt (0.002–0.02 mm), and sand(N0.02 mm) contents were then calculated.

2.4. Soil water storage

This study used the following equation to calculate soil water stor-age (SWS) within a depth of 5 m (Jia and Shao, 2013):

s ¼ θv � h � 10

where s is the SWS at a specific depth (mm), θv is the volumetric soilwater content at a specific depth (cm3 cm−3) and h is the soil depth in-crement (cm).

2.5. Statistical analysis

One-way ANOVA was used to analyze the means of the same soillayers across the different restoration stages. Differences were evaluat-ed at the 0.05 significance level. When significance was observed atthe P b 0.05 level, the Least Significant Difference (LSD) test was usedto carry out themultiple comparisons. Pearson's test was adopted to de-terminewhether there were significant correlations between soil waterstorage and the soil properties measured in the study.

3. Results

3.1. Soil bulk density, porosity and particle composition

Soil bulk densities (BDs) in the different soil layerswere significantlydifferent under the different succession stages following vegetation res-toration (P b 0.05) (Fig. 2). With vegetation succession, the BD variedmainly in the topsoil layers (0–10, 10–20 and 20–30 cm), where it de-creased significantly with vegetation succession (P b 0.05). In the 30–40 and 40–50 cm soil layers, the BD showed no significant differenceamong the four restoration stages (P N 0.05) in which the restorationage was greater than 30 years, and it was highest in the G1 restorationstage (P b 0.05). Generally, the soil BD is higher in deeper soil layersthan in upper layers.

The soil total porosity (TP), inactive porosity (IP), aeration porosity(AP) and capillary porosity (CP) of the different soil layers under the dif-ferent succession stages of vegetation after farmland abandonment aredisplayed in Fig. 3. Following vegetation restoration, the TP increasedsignificantly (P b 0.05) in the 0–10 cm soil layer. It first increased andthen decreased in the 10–20 cm soil layer and was stable at the F1stage. The TP in the 20–30 cm soil layer did not increase until the F2stage and gradually increased in the 30–40 and 40–50 cm soil layers,and it decreased from the S stage and became stable at the F1 stage.The IP decreased initially and then increased significantly, and the low-est value occurred in the S stage in the 0–10 and 10–20 cm layers. Thevariation trend in the 20–30 cm layer was the same as that in theupper two layers; however, the lowest value occurred in the F1 stage.The IP showed no significant variation in the 30–40 cm layer, and itwas highest in the F2 stage in the 40–50 cm layer. The AP also initially

Fig. 3. Soil total porosity (TP), inactive porosity (IP), aeration porosity (AP) and capillary porosity (CP) in the different soil layers in each restoration stage (see Table 1). Values are in theform of the Mean ± SE, and the sample size n = 5. Different lower-case letters above the bars mean significant differences in the same soil layers at the different restoration stages (P b

0.05).

55Y. Zhang et al. / Catena 137 (2016) 52–60

increased and then decreased significantly in the 0–10 and 10–20 cmsoil layers, and the highest value occurred in the S stage in both layers.In other soil layers (20–50 cm), the AP no longer had obvious variations.The CP increased significantly in the 0–10 cm layer, and the highestvalue occurred in the F1 stage; in other soil layers, it displayed no obvi-ous variation.

The contents of clay (b0.002 mm), silt (0.002–0.02 mm) and sand(0.02–2 mm) of the different soil layers varied significantly under thedifferent vegetation succession stages after abandonment of agriculturalproduction (Table 2). Generally, in the five layers, the sand content washighest, followed by the silt content, and the clay content was the low-est. With the vegetation restoration, the clay and silt content decreased,whereas the sand content increased significantly (P b 0.05) in the 0–10 cm layer. In the 10–20 and 20–30 cm soil layers, the clay and silt con-tents were highest in the G2 stage and lowest in the S stage; conversely,the highest sand content occurred in the S stage, and the lowest oc-curred in the G2 stage. In the 30–40 cm soil layer, the clay content atgrass stages was higher than that at shrub and forest stages, as wasthe silt content; however, the sand content was highest in the S stage.In both the 40–50 and 0–50 cm soil layers, the clay, silt and sand con-tents showed no significant variation (P N 0.05).

3.2. Soil water storage

Vertical variations in soil water content (SWC) over the 0–500 cmsoil depth decreased significantly with vegetation succession(P b 0.05) (Fig. 4(a)). Within the 0–500 cm soil layer, the SWC wasthe lowest at the forest stage and highest at the grass stage (Fig. 4(a)).The SWC did not decrease until approximately 300 cm soil depth at

grass and shrub stages and until 200 cm soil depth at forest stages andthen decreased slightly and gradually stabilized (Fig. 4(a)).

Vertical variations in the soil water storage (SWS) over the 0–500 cm soil depth decreased significantly with vegetation restoration(P b 0.05) (Fig. 4(b)). In the early grass stage, the vertical SWS varieddrastically; it did not increase until the 220–240 cm soil layer andthen decreased obviously, and when it developed to the later grassstage, the drastic variation trend was diminished (Fig. 4(b)). The SWSbelow 200 cm soil depth in the shrub stage was the lowest among allthe stages. The SWS did not decrease until approximately 300 cm soildepth at grass and shrub stages and until 200 cm soil depth at foreststages and then decreased slightly and gradually became stable(Fig. 4(b)).

The soil water storage in the different soil layers was obviously dif-ferent in each restoration stage (Fig. 4(c)). Generally, in the 0–500 cmsoil layer, from the grass stage to the forest stage, the SWS showed a sig-nificantly (P b 0.05) decreasing trend; in the 0–100 and 100–200 cm soillayers, there were similar variations in the five vegetation restorationstages, and the SWS was highest (164.61–212.80 mm) in the 200–300 cm soil layer compared to the other four soil layers (0–100, 100–200, 300–400 and 400–500 cm) in all five vegetation succession stages.

3.3. Relationship between soil water storage and soil properties

The SWS and BD showed a negative relationship in the 0–10 cmsoil layer but a positive relationship in the 10–20, 20–30, 30–40and 40–50 cm soil layers (P N 0.05) (Fig. 5). The SWS and SWCshowed an obvious positive relationship in the 0–10 and 40–50 cmsoil layers (P b 0.05) and an extreme positive relationship in the30–40 cm soil layer (P b 0.01).

Table 2The content of clay (b0.002mm), silt (0.002–0.02 mm) and sand (0.02–2 mm) in the dif-ferent soil layers in each restoration stage (P b 0.05) (see Table 1). Values are in the formofthe Mean ± SE, and the sample size n = 5. Different lower-case letters mean significantdifferences in the same soil layers at the different restoration stages (P b 0.05).

Soil depth(cm)

Restorationstage

Soil particle size (%)

Clay(b0.002 mm)

Silt(0.002–0.02 mm)

Sand(0.02–2 mm)

0–10

G1 12.31 ± 0.07b 30.08 ± 0.16b 57.61 ± 0.23bG2 14.13 ± 0.02a 33.33 ± 0.05a 52.54 ± 0.07cS 9.92 ± 0.1c 27.48 ± 0.29c 62.61 ± 0.39aF1 9.78 ± 0.07 cd 27.05 ± 0.18c 63.17 ± 0.25aF2 9.02 ± 0.09d 26.38 ± 0.23c 64.60 ± 0.32a

10–20

G1 14.06 ± 0.09a 32.98 ± 0.18b 52.96 ± 0.28bG2 14.57 ± 0.09a 34.98 ± 0.22a 50.45 ± 0.31cS 11.19 ± 0.03c 30.43 ± 0.13c 58.38 ± 0.15aF1 12.61 ± 0.05b 30.73 ± 0.04c 56.66 ± 0.08aF2 12.84 ± 0.02b 33.74 ± 0.02ab 53.43 ± 0.03b

20–30

G1 13.04 ± 0.02b 33.41 ± 0.02a 53.56 ± 0.03cG2 15.35 ± 0.05a 33.39 ± 0.11a 51.26 ± 0.13dS 12.05 ± 0.02c 30.64 ± 0.1b 57.30 ± 0.1aF1 13.43 ± 0.03b 31.32 ± 0.05b 55.25 ± 0.08bF2 13.78 ± 0.02b 34.74 ± 0.06a 51.48 ± 0.05d

30–40

G1 14.39 ± 0.01a 37.04 ± 0.08a 48.56 ± 0.09bG2 14.20 ± 0.03a 32.58 ± 0.1b 53.22 ± 0.12aS 12.20 ± 0.03b 32.62 ± 0.06b 55.18 ± 0.08aF1 14.00 ± 0.02a 31.47 ± 0.04b 54.53 ± 0.05aF2 12.78 ± 0.03b 37.91 ± 0.07a 49.31 ± 0.07b

40–50

G1 12.76 ± 0.02a 32.73 ± 0.06a 54.51 ± 0.07aG2 13.00 ± 0.07a 30.33 ± 0.11a 56.67 ± 0.18aS 12.20 ± 0.03a 33.54 ± 0.08a 54.26 ± 0.07aF1 14.37 ± 0.01a 31.98 ± 0.07a 53.66 ± 0.06aF2 13.59 ± 0.01a 37.79 ± 0.05a 48.62 ± 0.05a

0–50

G1 13.31 ± 0.04ab 33.25 ± 0.1a 53.44 ± 0.14aG2 14.25 ± 0.05a 32.92 ± 0.12a 52.83 ± 0.16aS 11.51 ± 0.04b 30.94 ± 0.13a 57.55 ± 0.16aF1 12.84 ± 0.04ab 30.51 ± 0.08a 56.65 ± 0.1aF2 12.40 ± 0.04ab 34.11 ± 0.09a 53.49 ± 0.1a

56 Y. Zhang et al. / Catena 137 (2016) 52–60

The SWS and TP showed a positive relationship in the 0–10, 10–20and 40–50 cm soil layers and a negative relationship in the 20–30 and30–40 cm soil layers, but they were not significant relationships

Fig. 4. Vertical variation of the (a) soil water content (SWC) and (b) soil water storage (SWS) thstage (see Table 1). Values are in the formof theMean±SE, and the sample size n=5.Differentdifferent restoration stages (P b 0.05).

(Fig. 6). The SWS and IP showed a positive relationship in the 0–10,10–20, 20–30 and 30–40 cm soil layers and a negative relationship inthe 40–50 cm soil layer (Fig. 6). The SWS and AP showed a significantnegative relationship in the 30–40 cm soil layer (P b 0.05) and a positiverelationship in the 0–10, 10–20 and 40–50 cm soil layers. The SWS andCP showed a positive relationship in the 0–10, 10–20 and 40–50 cm soillayers, and a negative relationship in the 20–30 and30–40 cmsoil layers(Fig. 6).

The SWS and clay content showed a negative relationship in the 0–10 and 40–50 cm soil layers and a positive relationship in the 10–20,20–30 and 30–40 cm soil layers (Fig. 7). The SWS and silt contentshowed a similar relationship as the SWS and clay content (Fig. 7).However, the SWS and sand content showed a positive relationship inthe 0–10 and 40–50 cm soil layers and a negative relationship in the10–20, 20–30 and 30–40 cm soil layers (Fig. 7).

4. Discussion

Vegetation types and structures can have significant influences onthe soil water content (Wang et al., 2012a). This study indicated thatthe SWS in the 0–500 cm soil layers significantly decreased along thevegetation succession stages from the early grass stage to the climaxforest stage (Fig. 4). This trend is probably because the SWC showedthe same pattern with the vegetation restoration. The SWC had a signif-icant positive relationship with the SWS (P b 0.05) (Fig. 5). As expectedand previously studied, the vertical variation in soil moisture tended tobe higher near the soil surface as a result of the frequent exchange ofwater and energy (Jia and Shao, 2013). Our study also showed thatthe SWC in the 0–100 cm soil layer was highly variable. In contrast,Chen et al. (2008) showed that the SWC was less variable in the0–400 cm soil layer at the plot scale, where there was less variabilityin, e.g., soil types, plants and topography. The SWC at all the successionstages had similar profile distribution patterns that initially increasedsignificantly and then decreased slightly with increasing soil depth.This trend was contrary to the results of Wang et al. (2012b). Thismay be due to the different soil textures, investigation periods and veg-etation types. Generally, at any particular point in the landscape, theSWC would be influenced by four factors: (1) the precipitation history;

rough the 0–500 cm soil depth and (c) SWS in the different soil layers in each restorationlower-case letters above the barsmean significant differences in the same soil layers at the

Fig. 5. Relationship between the soil water storage (SWS) and the soil bulk density (BD) and soil water content (SWC) in five soil layers at soil depths of 0–50 cm through the vegetationsuccession gradient. Capped horizontal and vertical lines are the SE.

57Y. Zhang et al. / Catena 137 (2016) 52–60

(2) the texture of the soil, which is related to the soil structure, soil hy-drology and water holding capacity; (3) the topographic features, suchas slope gradient, which can affect runoff and infiltration, and aspect,which can affect local climate parameters and plant water demands;and (4) the vegetation type and degree of land cover, which influence

Fig. 6. Relationship between the soil water storage (SWS) and the soil total porosity (TP), inactidepths of 0–50 cm through the vegetation succession gradient. Capped horizontal and vertical

evapotranspiration and deep percolation (Mohanty and Skaggs, 2001).However, soils at grass stages had significantly higher SWC than thoseat shrub or forest stages (P b 0.05), which is partly in agreement withthe results of Wang et al. (2006). Soils at forest stages tended to havelowerwater contents due to the higher root densities of trees compared

ve porosity (IP), aeration porosity (AP) and capillary porosity (CP) in five soil layers at soillines are the SE.

Fig. 7. Relationship between the soil water storage (SWS) and the contents of clay (b0.002 mm), silt (0.002–0.02 mm) and sand (0.02–2 mm) in five soil layers at soil depths of 0–50 cmthrough the vegetation succession gradient. Capped horizontal and vertical lines are the SE.

58 Y. Zhang et al. / Catena 137 (2016) 52–60

with grasses at these depths and the resulting greater transpiration abil-ity (Wang et al., 2010b). Thus, land use can significantly affect the soilwater cycle in the soil–plant–atmosphere system, and water resourcesare used differently in the various soil layers between the surface anddepths of 500 cm or deeper (Yang, 2001; Chen et al., 2008).

In our study, we showed that the BD decreased with vegetation suc-cession, indicating that natural recovery improves the physical proper-ties of soil to some extent at a soil depth of 0–50 cm (Wang et al.,2012c). However, there were no significant differences in the BD ofthe 0–20 cm soil layers at the early stages of vegetation restoration(b30 years) (P N 0.05), indicating that any improvement in the soilphysical conditions associated with natural recovery occurred beyond30 years. In addition, in our study, lower sand contents in 0–20 cmsoil layers at grass stages were in accordance with the higher BD, andclay contents showed the opposite pattern. However, this differencedid not induce a change in soil texture because therewere no significantbetween-vegetation-type differences in silt contents, including in thelower layers below 0–20 cm. Clay contents in the 0–40 cm soil layersdid not decrease until after the later grass stage that had been recover-ing for 30 years (P b 0.05), which implied that long-term natural vege-tation recovery (N30 years) can ameliorate soil texture to someextent. This result indicates that the soil physical degradation resultingfrom soil desertification is difficult to reverse in a short period (Wanget al., 2012c; Li and Shao, 2006).

In the Loess Plateau, porosity is a key attribute of soil structure affect-ing the soil reservoir under natural vegetation recovery (Zhao et al.,2010). In this study, the TP and CP of the surface soil showed the sametrend, which was that both of them increased with the developmentof vegetation succession, consistent with the results that soil porosityincreases gradually with the natural vegetation succession, which si-multaneously increases soil water storage capacity and soil waterretaining capacity (Zhao et al., 2010; Li and Shao, 2006; Zhou et al.,2003). Udawatta et al. (2006) and Udawatta et al. (2008b) examinedthe effect of differentmethods of vegetationmanagement on soil poros-ity parameters and found that the soil porosity parameters in forestswere much more favorable than those in grasslands and farmlands.The soil porosity parameters in natural grassland were more favorablethan those in recovered grassland and farmland (Udawatta et al.,

2008a). These results were, in part, consistent with the conclusions ofthis study. This result means that natural vegetation recovery could re-markably improve soil pore characteristics, indicating that soil proper-ties improved after vegetation restoration because increases in macro-pore volume imply increases in hydraulic conductivity and water-holding capacity, which promote effective infiltration of precipitation,aeration of deeper soil layers and, thus, plant-root growth and vegeta-tion development (Wang et al., 2008). The vegetation types changedalong with the vegetation restoration. Through the natural vegetationsuccession, the vegetation changes in diversity during the evolution ofaboveground populations from herb plants to scrub and trees(Table 1). The transformation of vegetation litter into organic mattercauses soil physical and chemical properties to change, thereby affectingsoil pores (Zhao et al., 2010). The results of this study showed that in the0–50 cm depth, the soil pore parameters tended to decrease with in-creasing soil depth, which was consistent with previous conclusions(Udawatta et al., 2008a; Udawatta and Anderson, 2008b; Udawattaet al., 2006; Rachman et al., 2005). The results of the study could providea helpful reference for assessing the capacity of soil reservoirs and carry-ing out research related to soil structures.

In addition, the BD, SWC, and clay and silt contents presented similartrends to those of the SWS along the vegetation restoration stages in thestudy. The BD, SWC, and clay and silt contents all decreased with vege-tation restoration, and the SWS was significantly and positively corre-lated with other soil properties, i.e., the SWC and aeration porosity(P b 0.05) (Figs. 3, 5 and 6). The BD, porosity, SWC and soil particle com-position are all commonly measured soil physical properties. Correla-tion analysis of parameters of the 0–50 cm layers showed that the soilphysical properties in the study area were strongly interrelated(Table 3). The BD was negatively related to most other measured soilphysical properties, such as the SWC, sand content and IP, but positivelyrelated to the clay and silt contents (P b 0.05). This indicated that thevegetation restoration process that reduced bulk densitywould also im-prove the other soil physical properties (Wang et al., 2012c). Increasesin total porosity were strongly positively related to increases in aerationporosity, capillary porosity and inactive porosity. The clay content wasalso significantly correlated with the soil water content negatively andthe bulk density positively, indicating that the clay content was

Table 3Pearson's correlation between the SWS and soil properties.

Soil depth(cm)

Factor Factor R2 Soil depth(cm)

Factor Factor R2

0–10

SWS SWC 0.943⁎

20–30

TP IP 0.974⁎⁎

BD SWC −0.918⁎ TP AP 0.921⁎

Clay BD 0.928⁎ TP CP 0.999⁎⁎

Silt BD 0.915⁎ IP CP 0.984⁎⁎

Sand BD −0.922⁎ AP CP 0.901⁎

Clay SWC −0.921⁎ Silt Sand −0.922⁎

Silt SWC −0.911⁎

30–40

SWS SWC 0.976⁎⁎

Sand SWC 0.917⁎ SWS AP −0.900⁎

TP AP 0.963⁎⁎ BD Sand −0.886⁎

TP CP 0.994⁎⁎ SWC AP −0.963⁎⁎

AP CP 0.929⁎ TP AP 0.946⁎

Clay Silt 0.994⁎⁎ TP CP 0.999⁎⁎

Clay Sand −0.998⁎⁎ AP CP 0.936⁎

Silt Sand −0.999⁎⁎ Silt Sand −0.948⁎

10–20

BD IP 0.948⁎

40–50

SWS SWC 0.950⁎

TP CP 0.962⁎⁎ TP CP 0.999⁎⁎

Clay Sand −0.938⁎ Silt Sand −0.961⁎⁎

Silt Sand −0.972⁎⁎

Note: SWS soil water storage, BD bulk density, TP total porosity, IP inactive porosity, APaeration porosity, CP capillary porosity.⁎ Correlation significant at the 0.05 level (2-tailed).⁎⁎ Correlation significant at the 0.01 level (2-tailed).

59Y. Zhang et al. / Catena 137 (2016) 52–60

sensitive to the soil water content in the study area. Although it was rel-atively low, the clay content strongly influenced the soil water-holdingcapacity. The silt content showed similar correlations to those of otherphysical properties with the clay content, while the sand contentshowed opposite correlations.

5. Conclusions

Long-term vegetation restoration had significant effects on the SWSfrom grassland to forest. The SWS significantly decreased with vegeta-tion restoration. The SWC, associated with the SWS through the 0–500 cm soil depth, significantly decreased in the early stages of restora-tion (0–50 a) and tended to decrease slightly in the later forest stage(50–160 a). The SWC and SWS were lower in the upper soils than inthe lower soils through the different restoration stages; however, theSWS in the 200–300 cm soil layer was higher than that in the otherlayers at all restoration stages. In addition, the BD, soil porosity andsoil particle composition were better along the vegetation restoration.As the study area, the Ziwuling forest region of the Loess Plateau, hasan intact series of naturally recovering vegetation succession, the con-clusions are universal in semi-arid areas. The results of the study couldprovide a helpful reference for assessing the capacity of soil reservoirsand carrying out research related to soil structures.

Acknowledgments

This study was sponsored by the Major Program of the NationalNatural Science Foundation of China (41390463), the National NaturalScience Foundation of China (41501094), the National Key TechnologyR&DProgram (2015BAC01B03) and the Science and Technology ServiceNetwork Initiative of the Chinese Academy of Sciences (KFJ-EW-STS-005).

References

Boix-Fayos, C., Calvo-Cases, A., Imeson, A.C., Soriano-Soto, M.D., 2001. Influence of soilproperties on the aggregation of some Mediterranean soils and the use of aggregatesize and stability as land degradation indicators. Catena 44, 47–67.

Cao, S., Chen, L., Yu, X., 2009. Impact of China's grain for green project on the landscape ofvulnerable arid and semi-arid agricultural regions: a case study in northern ShaanxiProvince. J. Appl. Ecol. 46 (3), 536–543.

Chen, H.S., Shao, M.A., Li, Y.Y., 2008. The characteristics of soil water cycle and water bal-ance on steep grassland under natural and simulated rainfall conditions in the LoessPlateau of China. J. Hydrol. 360 (1-4), 242–251.

Cheng, J.M., Cheng, J., Shao, H.B., Zhao, L.P., Yang, X.M., 2012. Soil seed banks and forestsuccession direction reflect soil quality in Ziwuling Mountain, Loess Plateau, China.Clean Soil Air Water 40, 140–147.

Deng, L., Shangguan, Z.P., Li, R., 2012. Effects of the grain-for-green program on soil ero-sion in China. Int. J. Sediment Res. 27, 131–138.

Deng, L., Wang, K.B., Chen, M.L., Shangguan, Z.P., Sweeney, S., 2013. Soil organic carbonstorage capacity positively related to forest succession on the Loess Plateau. Catena2013 (110), 1–7.

Deng, L., Wang, K.B., Shangguan, Z.P., 2014. Long-term natural succession improves nitro-gen storage capacity of soil on the Loess Plateau, China. Soil Res. 52, 262–270.

Feng, X., Fu, B., Lu, N., Zeng, Y., Wu, B., 2013. How ecological restoration alters ecosystemservices: an analysis of carbon sequestration in China's Loess Plateau. Sci. Rep. 3,2846. http://dx.doi.org/10.1038/srep02846.

Fu, B., Chen, L., 2000. Agricultural landscape spatial pattern analysis in the semi-arid hillarea of the Loess Plateau, China. J. Arid Environ. 44 (3), 291–303.

Fu, B.J., Hu, C.X., Chen, L.D., Honnay, O., Gulinck, H., 2006. Evaluating change in agriculturallandscape pattern between 1980 and 2000 in the Loess hilly region of Ansai County.Agric. Ecosyst. Environ. 114 (2–4), 387–396.

Fu, B., Wang, J., Chen, L., Qiu., Y., 2003. The effects of land use on soil moisture vari-ation in the Danangou catchment of the Loess Plateau, China. Catena 54 (1–2),197–213.

Gao, L., Shao, M., 2012. Temporal stability of shallow soil water content for three adjacenttransects on a hillslope. Agric. Water Manag. 110, 41–54.

Ghanbarian-Alavijeh, B., Millán, H., 2009. The relationship between surface fractal dimen-sion and soil water content at permanent wilting point. Geoderma 151 (3–4),224–232.

Jia, Y.H., Shao, M.A., 2013. Temporal stability of soil water storage under four types ofrevegetation on the northern Loess Plateau of China. Agric. Water Manag. 117,33–42.

Jia, G.M., Cao, J., Wang, C.Y., Wang, G., 2005. Microbial biomass and nutrients in soil at thedifferent stages of secondary forest succession in Ziwulin, northwest China. For. Ecol.Manag. 217, 117–125.

Jia, X.X., Wei, X.R., Shao, M.A., Li, X.Z., 2012. Distribution of soil carbon and nitrogenalong a revegetational succession on the Loess Plateau of China. Catena 95,160–168.

Li, Y.Y., Shao, M.A., 2006. Change of soil physical properties under longterm naturalvegetation restoration in the Loess Plateau of China. J. Arid Environ. 64, 77–96.

Mohanty, B.P., Skaggs, T.H., 2001. Spatio-temporal evolution and time-stable characteris-tics of soil moisture within remote sensing footprints with varying soil, slope andvegetation. Adv. Water Resour. 24 (9-10), 1051–1067.

Rachman, A., Anderson, S.H., Gantzer, C.J., 2005. Computed-tomographic measurement ofsoil macroporosity parameters as affected by stiff-stemmed grass hedges. Soil Sci.Soc. Am. J. 69, 1609–1616.

Stolte, J., van Venrooij, B., Zhang, G., Trouwborst, K.O., Liu, G., Ritsema, C.J., Hessel, R., 2003.Landuse induced spatial heterogeneity of soil hydraulic properties on the Loess Pla-teau in China. Catena 54 (1–2), 59–75.

Tombul, M., 2007. Mapping field surface soil moisture for hydrological modeling. WaterResour. Manag. 21 (11), 1865–1880.

Tsunekawa, A., Liu, G.B., Yamanaka, N., Du, S., 2014. Restoration and Development of theDegraded Loess Plateau, China. Springer Verlag, Japan.

Udawatta, R.P., Anderson, S.H., 2008b. CT-measured pore characteristics of surface andsubsurface soils influenced by agroforestry and grass buffers. Geoderma 145,381–389.

Udawatta, R.P., Anderson, S.H., Gantzer, C.J., 2006. Agroforestry and grass buffer influenceon macropore characteristics: a computed tomography analysis. Soil Sci. Soc. Am. J.70, 1763–1773.

Udawatta, R.R., Anderson, S.H., Gantzer, C.J., 2008a. Influence of prairie restoration on CT-measured soil pore characteristics. J. Environ. Qual. 37, 219–228.

Wang, Z.H., Hou, Y., Fang, H., Yu, D., Zhang, M., Xu, C., Chen, M., Sun, L., 2012a. Effects ofplant species diversity on soil conservation and stability in the secondary successionphases of a semi-humid evergreen broadleaf forest in China. J. Soil Water Conserv. 67,311–320.

Wang, L., Mu, Y., Zhang, Q.F., Jia, Z.K., 2012c. Effects of vegetation restoration on soilphysical properties in the wind–water erosion region of the northern Loess Plateauof China. Clean Soil Air Water 40 (1), 7–15.

Wang, K.B., Shao, R.X., Shangguan, Z.P., 2010a. Changes in species richness andcommunity productivity during succession on the Loess Plateau (China). Pol.J. Ecol. 58, 549–558.

Wang, Y., Shao, M., Shao, H., 2010b. A preliminary investigation of the dynamic char-acteristics of dried soil layers on the Loess Plateau of China. J. Hydrol. 381, 9–17.

Wang, Y., Shao, M.a., Liu, Z., Horton, R., 2013. Regional-scale variation and distributionpatterns of soil saturated hydraulic conductivities in surface and subsurface layersin the loessial soils of China. J. Hydrol. 487, 13–23.

Wang, Y., Shao, M.A., Liu, Z., Warrington, D.N., 2012b. Regional spatial pattern of deep soilwater content and its influencing factors. Hydrol. Sci. J. 57 (2), 265–281.

Wang, L., Wang, Q.J., Wei, S.P., Shao, M.A., Li, Y., 2008. soil desiccation for loess soils onnatural and regrown areas. For. Ecol. Manag. 255, 2467–2477.

Wang, Y.Q., Zhang, X.C., Cong, W., Wei, Q.C., 2006. Spatial variability of soil moisture onslope-land under different land uses on the Loess Plateau. Trans. CSAE 22 (12),65–71 (in Chinese with English abstract).

Yang, W., 2001. Soil water resources and afforestation in Loess Plateau. J. Nat. Resour. 16(5), 433–438 (in Chinese).

Yang, L., Wei, W., Chen, L., Mo, B., 2012b. Response of deep soil moisture to land use andafforestation in the semi-arid Loess Plateau. J. Hydrol. 475, 111–122.

Yang, L., Wei, W., Chen, L., Jia, F., Mo, B., 2012a. Spatial variations of shallow and deep soilmoisture in the semi-arid Loess Plateau. Hydrol. Earth Syst. Sci. 16 (9), 3199–3217.

60 Y. Zhang et al. / Catena 137 (2016) 52–60

Yang, L., Wei, W., Chen, L.D., Wang, J.L., 2014. Response of temporal variation of soilmoisture to vegetation restoration in semi-arid Loess Plateau, China. Catena 115,123–133.

Zhao, S.W., Zhao, Y.G.,Wu, J.S., 2010. Quantitative analysis of soil pores under natural veg-etation successions on the Loess Plateau. Sci. Earth Sci. 53, 617–625.

Zhou, Y.D., Wu, J.S., Zhao, S.W., Guo, S.L., Lu, P., 2003. Change of soil organic matter andwater holding ability during vegetation succession in Ziwuling region. Acta Bot. Bore-al.-Occident. Sin. 23, 895–900.

Zhou, D., Zhao, S., Zhu, C., 2012. The grain for green project induced land cover change in theLoess Plateau: a case study with Ansai County, Shanxi Province. Ecol. Indic. 23, 88–94.

Zhu, H.D., 2014. Soil moisture response to environmental factors following precipitationevents in a small catchment. Catena 120, 73–80.

Zhu, Y., Shao, M.A., Shao, H., 2009. Evaporation process in soil surface containing calcicnodules on the northern Loess Plateau of China by simulated experiments. CleanSoil Air Water 37 (11), 866–871.

Zou, H.Y., Liu, G.B., Wang, H.S., 2002. The vegetation development in north ziwuling forestregion in last fifty years. Acta Bot. Boreal.-Occident. Sin. 22, 1–8 (in Chinese).