Influence of vegetation restoration on soil physical properties in the

SOILS, SEC 2 • GLOBAL CHANGE, ENVIRON RISK ASSESS, SUSTAINABLE LAND USE • RESEARCH

ARTICLE

Influence of vegetation restoration on soil physical properties

in the Loess Plateau, China

Chaojun Gu

& Xingmin Mu

1,2

& Peng Gao

1,2

& Guangju Zhao

1,2

& Wenyi Sun

1,2

& John Tatarko

& Xuejin Tan

Received: 7 December 2017 / Accepted: 4 July 2018 / Published online: 2 August 2018

Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract

Purpose Extensive vegetation recovery has been implemented to control severe soil erosion on the Loess Plateau, China.

However, no systematic study has been done on the soil improvement benefit and preferable pattern of vegetation rehabilitation.

In this study, the effects of vegetation restoration on soil physical properties at ten sites with different vegetation types and varying

restoration periods were investigated.

Materials and methods The experiment was carried out in the Yanhe river basin in the hilly-gully region of the Loess Plateau.

Ten sites, including two replanted arboreal forests for 25 and 35 years, three replanted scrubland for 15, 30, and 45 years, four

secondary natural grassland for 10, 20, 30, and 40 years, and one farmland, were selected for soil sampling. Sampling was

conducted at 0–20 cm and 20–40 cm layers.

Results and discussion Vegetation restoration significantly decreased bulk density, and increased aggregate stability and

saturated hydraulic conducti vity (Ks), while their effects on porosity was complicated. The soil texture class did not

change with vegetation succession, but minor differences in sand, silt, and clay components were observed. Bulk

density, macroporosity, and > 0.25 mm aggregate stability were the principal physical parameters that affected Ks.

Moreover, bulk density had the most effect on Ks (− 0.84), while > 0.25 m m aggregate stability had the lowest

impact (0.38). Bulk density and Ks were correlated with most of the other soil phys ical properties. Repl anted

scrubland and secondary natural grassland had higher soil physical quality index (S

Dexter

) than replanted arboreal

forests and farmland.

Conclusions It is reasonable to take bulk density and Ks as the indicators to evaluate the effect of vegetation

restoration on soil physical properties. Planting shrubs and grassland is be tter than forest for eco -environment

rehabilitation in the study area. Results of this study provide a reference to regional eco-environmental rehabilitation

and conservation.

Keywords Bulk density

Saturated hydraulic conductivity

Soil physical properties

Soil quality

1 Introduction

China’s Loess Plateau is one of the most severely eroded areas

in the world due to its loose loess soils, steep slopes, high

intensity summer storms, and poor vegetation conditions (Li

et al. 2009). The annual average sediment discharge of the

Yellow River was 16 × 10

t between 1919 and 1960, indicat-

ing the seriousness of soil loss in the Loess Plateau (Mu et al.

2012). Soil erosion leads to the removal of soil nutrients and

degradation of soil structure, negatively impacting vegetation

development and impeding soil water transport capabilities

(Berger and Hager 2000). In order to control soil erosion and

restore the regional ecological functions, numerous soil and

Responsible editor: Saskia D. Keesstra

* Xingmin Mu

xmmu@ms.iswc.ac.cn

Institute of Soil and Water Conservation, Northwest A&F University,

Yangling 712100, Shaanxi, China

Institute of Soil and Water Conservation, Chinese Academy of

Sciences and Ministry of Water Resources,

Yangling 712100, Shaanxi, China

Rangeland Resources and Systems Research Unit, USDA–

Agricultural Research Service, Fort Collins, CO, USA

Journal of Soils and Sediments (2019) 19:716–728

https://doi.org/10.1007/s11368-018-2083-3

water conservation practices have been implemented over the

Loess Plateau since the 1950s, especially after the implemen-

tation of Grain-for-Green (GFG) project (Feng et al. 2012).

With the help of large-scale vegetation rehabilitation, veg-

etation coverage of the Loess Plateau increased from 28.8% in

the 1980s to 43.8% in 2012. Meanwhile, soil erosion has been

effectively controlled, and sediment discharge into the Yellow

River decreased significantly after the 1980s (7 × 10

t per

year during the 1980s–2000s), particularly after the GFG pro-

ject (4 × 10

t per year during 2000–2008) (Mu et al. 2012).

However, the flow of Yellow River declined significantly dur-

ing this period, which exacerbated water shortages in the re-

gion. Numerous researchers have studied the reasons for the

decrease in runoff (Gao et al. 2011). They found that extensive

forestation in this arid region is the primary cause (McVicar et

al. 2007). Large-scale vegetation recovery in the eroded re-

gion increased rainfall interception, surface ground coverage,

and caused changes in soil properties (Fu et al. 2003;Jiaoetal.

2011). Understanding the effects of vegetation restoration on

soil physical properties is important for studying the mecha-

nisms of hydrologic processes. Moreover, a study of vegeta-

tion recovery processes and their impacts on soil properties

would provide critical guide for eco-environmental restoration

or rehabilitation (Li and Shao 2006).

Many efforts have been made to study the effects of vege-

tation recovery on soil properties because of its importance in

eco-restoration and effectiveness (Cerdà 2000; Celik 2005;

Erktan et al. 2016). Previous studies have indicated that re-

vegetation can restore the integrity of disturbed ecosystems by

improving soil physical quality (e.g., decreasing soil bulk den-

sity, incre asing porosity and aggregate stability) (Jia et al.

2011;Lietal.2012). Nevertheless, the impacts of vegetation

restoration on soil properties may vary with the vegetation

type. Li et al. (2012) assessed the effects of vegetation resto-

ration on soil physical properties in the wind–water erosion

region of the Loess Plateau. The results showed that while soil

physical properties (e.g., bulk density, mean weight diameter,

macro-aggregates) have been significantly ameliorated in the

topsoil (0–20 cm layer) under secondary natural grasslands,

they changed adversely for replanted scrubland (e.g.,

Caragana korshinskii, Medicago sativa). Neris et al. (2012)

compared the infiltration rate of green forest and pine forest in

Tenerife with farmland soils. They found that infiltration rate

of the green forest was 11.9 times greater than that of the

farmland soils while pine forest was 2.8 times greater.

Moreover, vegetation effects on soil physical properties varied

with restoration periods (Li et al. 2007; Zhang et al. 2010).

The general conclusion was that the soil bulk d ensity de-

creased significantly with the recovery time (linear relation),

while soil organic matter increased but their relation differ

according to the regions (Zhang et al. 2010).

The Loess Plateau experienced extensive vegetation resto-

ration since the 1950s, and particularly during the recent

20 years (Gao et al. 2011). Their impacts on regional environ-

ment, such as climate, soil properties, and water resource have

been widely studied (Jiao et al. 2011). Soil acts as an indis-

pensable component of the soil–plant–atmosphere continuum

(SPAC). An evaluation of the changes in soil properties as a

consequence of this vegetation restoration efforts is important,

especially for fragile ecological areas. Previous studies have

looked at the influence of vegetation recovery on soil proper-

ties (Fu et al. 2003;Stolteetal.2003). Wilson et al. (2005)

assessed the changes of macropore flow characteristics for

four conditions (i.e., 1 year following tillage, 6 years follow-

ing tillage, 6 years following contour ditching, and greater

than 15 years following tillage) of the Loess Plateau. They

und that total number of macropores (> 1 mm), number of

large (> 5 mm) macropores, and the macroporosity increased

with revegetated time. Moreover, the soil matrix infiltration

rate was highest in the newly established (1 year) and the

oldest (> 15 years) revegetated areas. Xu et al. (2006) identi-

fied soil quality factors and indicators of the Loess Plateau by

32 soil properties. They found that organic matter, hydraulic

conductivity, and anti-scourability were the most important

indicators to characterize soil quality in this loess plateau.

However, changes in soil properties during long-term veg-

etation restoration on the Loess Plateau still require further

thorough studies. For instance, most studies rarely relate

changes i n soil physical properties with different vegeta-

tion types a nd restoration periods, and the choice of pref-

erable pattern of vegetation rehabilitation (e.g., Stolte et al.

2003). Furthermore, majority of published literature fo-

cused on soil physical properties such as bulk density, total

porosity, and aggregate, while less emphasis were on soil

functions such as the hydraulic conductivity and hierarchi-

cal pore (e.g., Jiao et al. 2011). Even though some articles

report hydraulic properties changed with vegetation resto-

ration, their relationships with other physical parameters

were limited (Li et al. 2012). Soil hyd raulic properties

(e.g., saturated hydraulic conductivity, Ks) are important

for soil water movemen t and runoff yield calculations

(Horton 1945), and play important roles in the understand-

ing hydrological response to vegetation restoration.

Following a general conclusion by researchers that the de-

crease in runoff in the yellow river is a consequence of

increased vegetation recover (Gao et al. 2011), it is becom-

ing more important to further investigate the effects of

vegetation restoration on soil physical properties and some

hydraulic functions.

The hilly-gully region of the Loess Plateau is a substantial

soil loss region, which supplies nearly 60% of total sediment

but only approximately 15% of total runoff into the Yellow

River (Wang et al. 2007). It is critical to study the effects of

vegetation restoration on soil physical properties both for the

evaluation of ecological restoration and understanding

of shifts in hydrological regime. For the reasons, we

J Soils Sediments (2019) 19:716–728 717

systematically determined effects of vegetation restoration

on soil physical properties in the hilly-gully region of the

Loess P lateau. Specifically, we evaluated the impacts of

soil physical properties on saturated hydraulic conductivity

and explored the effectiveness of vegetation restoration on

soil physical properties.

2 Materials and methods

2.1 Study area

This study was carried out in the Yanhe River basin in the hilly-

gully region of the Loess Plateau (middle reaches of Yellow

River) of China (Fig. 1a). The study area belongs to a temper-

ate continental semi-arid monsoon climate, with annual aver-

age precipitation and temperature of 538 mm and 9.9 °C, re-

spectively (1952–2015). The precipitation of the area is tem-

porally uneven, with 71.1% of the precipitation falling between

June and September (Fig. 1b). Vegetation restoration of the

basin started in the 1950s, and large-scale revegetation was

further implemented after the GFG project (1999). The vege-

tation cover increased by 2.58% per year after the GFG started

in 1999 and reached 58% in 2010 (Fig. 1c).

2.2 Site selection and sampling design

Ten sites (within 8 km of each other) representing the typical

vegetation restoration in this region were selected within the

basin. T wo sites were replanted to arboreal forests (Robinia

pseudoacacia L.) for 25 and 35 years; three replanted to scrub-

land (Korshinsk pea shrub) for 15, 30, and 45 years; four to

secondary natural grassland for 10, 20, 30, and 40 years; and

one was continuous farmland, used as a control (vegetation

types or treatments are defined in T able 1). The farmland has

been continuously planted to maize for more than 30 years. The

conventional tillage is performed manually by hoeing and resi-

due is removed for use as biofuel. All the vegetation was

replanted on the abandoned croplands. We assumed that local

soil properties are largely a consequence of plant growth and soil

protection during secondary succession, and the initial condi-

tions or management for the sites were similar (Jiao et al.

201 1). The chronosequence of the vegetation was investigated

to examine how soil properties change over time during resto-

ration. This chronological approach has been applied in other

ecosystem research (Fang and Peng 1997) and is considered

retrospective research because existing conditions are compared

with known original conditions and treatments. To reduce the

effects of gradient and elevation on the soil properties, all the

selected sites have the similar gradient (less than 10°) and ele-

vation (less than 200 m) (Table 1). Three plots were established

at each site (20 × 20 m in the forest sites and 10 × 10 m for other

sites) for soil sampling. Three points along the diagonal of the

plot were used for soil sampling and their average represents the

soil parameter value of the plot. The soil sampling was taken in

May, when there is no crop and tillage measure in farmland.

To determine soil physical properties, undisturbed samples

were obtained from the 0–20 cm and 20–40 cm soil layer at

each site. Three intact soil cores (5 cm diameter and 5 cm

height) were obtained using a ring knife in each soil layer,

which was used to measure the bulk density, soil water reten-

tion curve (SWRC), and Ks.

The bulk density was calculated as:

BD ¼ M

=V ð1Þ

where BD is the bulk density (g cm

−3

), M

is the mass of dry

soil (105–110 °C for 12 h) (g), and V is the volume of soil core

(cm

Soil water retention curve (SWRC) was measured in the

laboratory by a high-speed centrifuge using undisturbed soil.

Before the measurement, samples were first saturated in water

for 24 h. Then, soil water content (g g

−1

)atpressureheadsof

10, 100, 200, 400, 600, 800, 1000, 2000, 4000, 6000, 8000,

and 10,000 cm H

O were measured. To this end, the data was

used for fitting van Genuchten (1980) model (VG). The VG

model was expressed as follows (Van Genuchten 1980):

θ−θ

−θ

1 þ αψðÞ



ð2Þ

where S

is effective saturation, α is a scaling factor (cm

−1

O), n is pore size distribution parameter, and m =1− 1/n.

Three momentous parameters of SWRC were used to calcu-

late the soil porosity, namely, saturation moisture content

(SMC), field capacity (FC), and permanent wilting point

(PWC) (g g

−1

). FC and WC was the soil water content at

pressure heads of 300 and 1500 cm H

O, respectively. Soil

porosity was classified as inactive porosity (IP), microporosity

(MIP), and macroporosity (MAP) in this paper according to

the study of Luxmoore (1981) and Zhang et al. (2016), and

they were expressed as follows:

IP ¼

PWC  BD

 100% ð3Þ

MIP ¼

FC−PWCðÞBD

 100% ð4Þ

MAP ¼

SMC−FCðÞBD

 100% ð5Þ

where WD is the water density (g cm

−3

A soil physical quality index S

Dexter

(Dexter 2004)canbe

calculated based on the SWRC, and a higher S

Dexter

mean a

better soil physical quality. S

Dexter

was expressed as:

Dexter

¼ −n θ

−θ

ðÞ

1 þ m



− 1þmðÞ

ð6Þ

718 J Soils Sediments (2019) 19:716–728

Approximately 1 kg of composite sample was collected

at each plot from each sampling depth for soil texture and

aggregate stability measurement. Air-dried soil sieved

through 2 mm si eve was used for soil particle s iz e mea-

surement by the laser diffraction technique using a

MasterSizer 2000 (Malvern Instruments, Malvern,

England). The soil particle size was classified into sand

(2–0.05 mm), silt (0.05–0.002 mm), and clay (< 0.002 mm)

according to USDA classification system. Aggregate sta-

bility was measured based on air-dried aggregates and

was evaluated according to wet-sieving effects (Li and

Shao 2006).

Ks was measured by the constant water head method using

distilled water at a 5 cm water head (Reynolds et al. 2002).

When a constant flow rate had been established, percolated

water volume per unit time was measured and used to calcu-

late Ks according to Darcy’s Law (Eq. (7)).

v ¼ −Ks



ð7Þ

where v is the percolation velocity (m s

−1

), Ks is the saturated

hydraulic conductivity (m s

−1

), and dh/ds is the hydraulic

gradient (m m

−1

). The percolation velocity can be expressed

Fig. 1 The location of selected

sites (a), mean monthly

precipitation and temperature

during 1952–2015 (b), and

vegetation coverage (1980–2010)

J Soils Sediments (2019) 19:716–728 719

as v = Q/Awith Q being the water flux (m

−1

)andA the cross-

section area that the water flowed through (m

). The Ks was

transformed to 10 °C in this study to eliminate the effects of

experiment temperature on Ks (Chi and Wang 2009).

0:7 þ 0:03t

ð8Þ

where Ks

was the Ks at 10 °C, Ks

is the Ks at t °C, and t is the

temperature of measurement water. Hereafter, Ks denotes the

saturated hydraulic conductivity at 10 °C.

2.3 Data analysis

Analysis of variance (ANOVA) was used to assess the differ-

ences of t he soil properties among the ten sites. When

ANOVA indicated the differences were significant according

to the F values, a Duncan’stestatP < 0.05 was performed to

compare means of soil variables. All the soil variables were

tested whether the data followed a normal distribution and

variance homogeneity by Shapiro test (Shapiro and Wilk

1965) and Bartlett test (Bartlett 1954), respectively. If the data

failed to meet the two conditions, the non-parametric method

(Kruskal–Wallis test) was used for above analysis. Stepwise

regression and path analysis were employed to evaluate the

effects of other soil physical properties on the Ks. Principal

components analysis (PCA) was used to explore the relation

between vegetation restoration and soil properties. All the

statistical tests were performed with R version 3.3.3.

3 Results

3.1 Bulk density and porosity

Soils in vegetated areas had lower BD at the two different

depths (P < 0.01) (Fig. 2a). In the 0–20 cm layer, BD of the

40-year replanted scrubland (RS40) and natural grass (NG40)

were significantly lower than other treatments. For the RS and

NG treatments, the BD significantly decreased with tim e

while no significant difference was found between replanted

forestland of 35 years (RF35) and 25 years (RF25). In the 20–

40 cm layer, the difference between NG40 and other years

(NG30, NG20, and NG10) was significant, but no significant

differences for the other vegetation types were found for the

different replanted years.

Figure 3a shows the soil water content at each pressure

heads could be well fitted by VG model. The average coeffi-

cient of determination (R

) was 0.984, and the mean root-

mean-square error (RMSE) was 0.616 (Fig. 3a). A relative

big difference was found in the soil water characteristic curves

(SWRCs) at the low-pressure heads (< 100 cm) among the

sites compared with high-pressure heads (> 100 cm). The soil

water holding capacity of RS35 was the greatest among the

ten sites at low-pressure heads for 0–20 cm soil layer, while it

dropped sharply at the high-pressure heads (> 100 cm)

(Fig. 3b).

Three categories of soil porosity were calculated based

on SWRC. Inactive porosity (IP) and microporos ity

(MIP) at 0–20 cm soil layer were generally lower than

that of 20–40 cm (Fig. 2b, c). IP of RF35 was signifi-

cantly higher than other sample plots at 0–20 cm layer,

while it was the highest in RS40 f or 20–40 cm layer

(Fig. 2b). In both 0–20 and 20–40 cm soil layers, natural

secondary grassland soil had lower MIP while replanted

forest soil had higher MIP (Fig. 2c). However, the differ-

ence of MIP betw e en mo st ve ge tate d ar eas and far mlan d

was not significant. Macroporosity (MAP) at 0–20 cm

layer for most sample plots was higher than that of 20–

40 cm soil layer (Fig. 2d). Replanted scrubland and nat-

ural secondary grassland had higher MAP than replanted

forestland and farmlan d, especially for RS40, RS3 0,

NG30, and NG10.

Table 1 Description of the sample sites in the Yanhe watershed

ID Vegetation type Age (years) Latitude Longitude Elevation (m) Slope (°) SOM (g kg

−1

)

RF35 Replanted arboreal forest 35 36°46′08″ 109°16′23″ 1140 21 8.32

RF25 Replanted arboreal forest 25 36°44′41″ 109°15′03″ 1152 26 6.98

RS40 Replanted scrubland 40 36°47′16″ 109°16′17″ 1200 28 8.26

RS30 Replanted scrubland 30 36°44′37″ 109°15′13″ 1260 20 8.55

RS15 Replanted scrubland 15 36°43′52″ 109°14′46″ 1340 20 8.41

NG40 Secondary natural grassland 40 36°47′16″ 109°15′30″ 1200 25 8.61

NG30 Secondary natural grassland 30 36°47′51″ 109°16′46″ 1250 26 7.39

NG20 Secondary natural grassland 20 36°47′15″ 109°15′27″ 1230 25 6.57

NG10 Secondary natural grassland 10 36°47′16″ 109°16′24″ 1180 20 5.58

FA Farmland 0 36°44′23″ 109°15′0″ 1173 26 8.49

ID is the sample site identification and SOM is the soil organic matter in 0–20 cm layer. Age refers to the number of years since the cessation of farming

activities

720 J Soils Sediments (2019) 19:716–728

3.2 Soil particle size composition

Mechanical analysis (Table 2) showed that there was signifi-

cantly lower sand contents in soil from revegetated than farm-

land. Additionally, for sand contents in 0–20 cm layer, sec-

ondary natural grassland of 40 an d 30 years (i.e., NG40,

NG30) were lower than NG20 and NG10. Silt contents of

soils in most restored sites were significantly higher than the

farmland soil (FA). For each soil from revegetated, only NG30

was significantly higher than NG20. There was no significant

difference among the ten sites for clay content. However, there

was a trend where clay content increased the longer the plots

were vegetated. Changes of soil particle size composition in

the 20–40 cm layer were similar to the topsoil. Soil texture

Fig. 2 Soil bulk density (a), inactive porosity (b), microporosity (c), and macroporosity (d) for different sample plots. The same letter above the point

denotes they are not significantly different at the 0.05 probability level (Duncan’stest)

Fig. 3 The fitting of VG model (a) and the soil water retention curves of the 0–20 cm layer (b) for the ten study sites

J Soils Sediments (2019) 19:716–728 721

triangle showing texture class for both topsoil and subsoil

were silt loam texture class (Fig. 4a USDA classification sys-

tem). Moreover, sand content of soil for 20–40 cm layer was

generally higher than soil in the 0–20 cm (Fig. 4b).

3.3 Aggregate stability

Table 3 shows RF35 had the highest proportions of > 5 mm

aggregates, while FA had the lowest proportions in the 0–

20 cm layer. A significant difference between the soils of

farmland and replanted vegetation sites was also found for

2–5mm,2–1 mm, 0.5–1mm,and0.25–0.5 mm aggregates.

For the total percent aggregates (> 0.25 mm), replanted forest-

land had the highest values, followed by NG40. FA had the

lowest percent aggregates and was not significantly different

from NG20 and NG10. Aggregate stability in the 20–40 cm

soil layer was significantly lower than that of 0–20 layer, and

the difference of the sites was slight compared with the topsoil

(data not shown).

3.4 Saturated hydraulic conductivity

Figure 5 shows vegetation restoration soils had signifi-

cantly higher Ks than farmland in the topsoil, while the

differences were less f or the subsoil. Ks ranged from 13.8

(FA) to 27 mm h

−1

for (RS40) in the 0–20 cm layer and 7.2

(FA) to 12 mm h

−1

(RS30) in the 20–40 cm layer f or the ten

sites. Of all vegetation types, RS40 was significantly

higher than RS30 and RS15. No distinguishable differ-

ences were found for other vegetation types in the topsoil.

In the subsoil, there was no significant difference among

the restoration periods for the ea ch vegeta tion type.

Moreover, RF35, RF25 , and NG10 were not signi fic ant ly

different fro m f ar mla nd.

Table 2 Mechanical composition (%) for different vegetation recovery types

0–20 cm 20–40 cm

ID Sand (%) Silt (%) Clay (%) Sand (%) Silt (%) Clay (%)

(0.05–2mm) (0.002–0.05 mm) (< 0.002 mm) (0.05–2mm) (0.002–0.05 mm) (< 0.002 mm)

RF35 10.76 (0.90)cd 63.07 (1.05)abcd 26.17 (0.24) 12.61 (0.15)b 61.57 (0.01)cd 25.82 (0.14)

RF25 10.05 (0.16)d 64.05 (0.13)a 25.90 (0.28) 12.40 (0.02)bc 61.83 (0.03)bc 25.78 (0.05)

RS40 10.08 (0.53)d 63.24 (0.30)abc 26.68 (0.46) 11.88 (0.17)d 62.29 (0.09)ab 25.84 (0.14)

RS30 10.43 (0.25)cd 62.9 (0.11)abcd 26.67 (0.25) 12.54 (0.24)b 61.73 (0.15)cd 25.73 (0.28)

RS15 10.17 (0.19)d 63.39 (0.57)abc 26.44 (0.42) 12.06 (0.42)cd 62.07 (0.52)abc 25.88 (0.13)

NG40 10.04 (0.06)d 63.30 (0.45)abc 26.67 (0.50) 11.81 (0.39)d 62.03 (0.39)abc 26.16 (0.32)

NG30 10.16 (0.20)d 63.63 (0.24)ab 26.21 (0.10) 12.02 (0.23)cd 62.47 (0.42)a 25.52 (0.23)

NG20 11.59 (0.15)ab 62.34 (0.23)cd 26.07 (0.22) 11.91 (0.29)d 62.37 (0.02)a 25.72 (0.29)

NG10 11.19 (0.73)bc 62.59 (1.04)bcd 26.22 (0.70) 12.81 (0.11)ab 61.63 (0.20)cd 25.57 (0.32)

FA 12.3 (0.55)a 62.01 (0.99)d 25.69 (0.48) 13.25 (0.19)a 61.30 (0.36)d 25.45 (0.46)

Significance of ANOVA < 0.001 0.024 0.073 < 0.001 < 0.001 0.126

Mean values in the same column followed by the same letter are not significantly different at the 0.05 probability level (Duncan’s test)

Fig. 4 Soil texture triangle

showing the range of textures for

different vegetation recovery

types (1, sand; 2, loamy sand; 3,

silt; 4, sandy loam; 5, loam; 6, silt

loam; 7, sandy clay loam; 8, clay

loam; 9, silty clay loam; 10, sandy

clay; 11, silty clay; 12, clay)

722 J Soils Sediments (2019) 19:716–728

3.5 Correlation analysis of soil physical properties

BD showed a significant negative relationship to most mea-

sured soil physical properties (P < 0.01) (Table 4). Sand con-

tents were negatively related to most soil properties (except

for BD and MIP); however, silt were positively correlated

with most soil properties. Total aggregates were significantly

related to most soil physical properties except for soil porosity.

IP was significantly positively correlated with silt content.

MIP was significantly positive correlated with BD and sand

contents, while it was negatively correlated with most other

soil properties. MAP was just significantly positive correlated

with clay content and Ks. The Ks was negatively correlated

with BD and sand contents, while it was significantly posi-

tively related to other soil physical properties except for IP.

3.6 Effects of soil physical properties on saturated

hydraulic conductivity

A stepwise regression (forward) was performed to select the

optimal factors that influenced Ks. Results showed that when

BD (X

), MAP (X

), and > 0.25 mm aggregate stability (X

)

were included, the model had the highest efficiency. The re-

gression equation was expressed as:

Y ¼ −52:893X

þ 0:164X

þ 0:081X

þ 69:845 R

¼ 0:763; N ¼ 60; P < 0:001



ð9Þ

where Y is the saturated hydraulic conductivity.

Because of changes in MAP, > 0.25 mm aggregate stability

could also cause the changes in BD. Path coefficients were

conducted to determine the direct and indirect effects of the

soil physical properties on Ks. Figure 6 shows that BD had the

highest direct effects (− 0.84) on Ks, followed by MAP (0.43),

while the lowest was found in aggregate stability (0.38).

Moreover, the direct effect of BD on Ks was negative, while

it was positive for other factors. In terms of the indirect effects,

the high est indirect e ffect was exerted by MAP (− 0. 4 7) ,

followed by aggregate stability (− 0.38). They exerted influ-

ences on Ks via BD. Moreover, the indirect effect of MAP was

higher than its direct effects, which revealed the importance of

BD in Ks.

Table 3 Percent (%) of aggregates by size in the surface layer (0–20 cm) for different vegetation recovery types

ID > 5 mm 5–2mm 2–1mm 0.5–1mm 0.25–0.5 mm > 0.25 mm

RF35 15.08 (1.08)a 13.58 (0.18)ab 12.06 (0.80)b 10.59 (0.56)b 9.38 (1.01)b 60.68 (0.56)a

RF25 3.98 (2.51)b 9.38 (2.48)cd 16.61 (0.21)a 18.64 (3.97)a 15.56 (2.52)a 64.16 (1.72)a

RS40 13.42 (7.05)a 9.38 (1.81)cd 6.07 (1.09)c 4.98 (0.55)cde 5.53 (1.98)b 39.39 (11.13)c

RS30 12.70 (2.25)a 12.09 (3.11)abcd 6.93 (1.33)c 5.64 (1.56)cde 5.52 (2.08)b 42.88 (6.33)bc

RS15 13.19 (7.32)a 9.90 (1.09)bcd 6.75 (0.99)c 5.95 (0.92)cde 6.65 (2.80)b 42.45 (5.26)bc

NG40 11.18 (3.11)a 12.86 (1.88)abc 10.31 (2.67)b 8.19 (2.82)bc 6.71 (2.31)b 49.24 (6.54)b

NG30 11.68 (0.96)a 14.90 (3.83)a 5.62 (0.48)c 3.09 (0.64)e 8.13 (3.78)b 39.42 (5.11)c

NG20 9.35 (2.96)ab 11.39 (0.59)abcd 5.37 (0.77)c 3.11 (0.72)e 6.18 (0.12)b 35.40 (4.22)cd

NG10 4.45 (0.63)b 8.97 (0.52)d 6.91 (1.43)c 6.97 (1.73)cd 8.18 (1.48)b 35.49 (1.40)cd

FA 2.95 (0.81)b 9.10 (0.80)cd 6.09 (0.77)c 3.78 (1.52)de 5.98 (0.76)b 27.91 (0.83)d

Significance of ANOVA 0.003 0.011 < 0.001 < 0.001 < 0.001 < 0.001

Mean values in the same column followed by the same letter are not significantly different at the 0.05 probability level (Duncan’s test)

Fig. 5 Saturated hydraulic

conductivity of 0–20 cm and 20–

40 cm soil layer for all sample

plots. The same letter above the

point denotes they are not

significantly different at the 0.05

probability level (Duncan’stest)

J Soils Sediments (2019) 19:716–728 723

3.7 Linkages between vegetation restoration and soil

properties

Principal component analysis (PCA) revealed that the first two

principal components explained 65.8% (PC1 = 42.7%, PC2 =

23.1%) of the variance, indicating they could express most

information of the data structure. Figure 7a shows the scatter

plot for the study sites and the relationship between the soil

properties based on the first two principal components. The

sample plots could be divided into three categories. The first

category includes RS40, RS30, and NG40, which have high

clay contents, MAP, and Ks. However, BD, sand contents, IP,

and MIP of the first category were fairly low. RF35, RF25,

RS15, and NG30 belong to the second category. These sample

plots had higher aggregates stability, IP, MIP, and silt contents,

while they had r elatively lower MAP. The third category

covers FA, NG20, and NG10. This category had higher BD

and sand contents compared with others. In terms of the soil

physical quality index (S

Dexter

), all the sites were higher than

0.035, indicating the studied soil had relatively good quality

(Fig. 7b). For the sites, S

Dexter

of RS40 is the highest (0.112),

followed by RS30 (0.099) and NG40 (0.094). S

Dexter

of RS35

and FA were 0.052 and 0.043, respectively, and were signifi-

cantly lower than others.

4 Discussion

4.1 Changes in soil physical properties along with

vegetation restoration

A common perception is that vegetation restoration can de-

crease BD and increase the porosity, especially the CP (Celik

2005;Lietal.2007; Neris et al. 2012). In our study, the

highest BD and lowest porosity were found in farmland,

which was mainly credited to compaction from tillage and

harvest operations (Zhang et al. 2010). In contrasts, soils from

the re-vegetated areas had a lower BD and higher porosity.

This is partly due to the accumulation of organic matter in the

soils from the re-vegetated (Table 1, r = − 0.685, P <0.001).A

relatively higher organic matter in the farmland was mainly

the result of fertilization which produced a high rate of vege-

tation turnover (Wang et al. 2011).

Our results showed that the soil texture of all study sites did

not significantly change (Fig. 4), which implies the soil texture

was slow to transform in the Loess Plateau ( Li and Shao

2006). Although significant differences in sand, silt, and clay

were lacking, sands were generally lower with silt and clay

higher in the plots with longer years in vegetation. This would

indicate that vegetation protects the soil from erosion loss of

highly erodible finer particles. Aggregate stability plays an

Table 4 Correlation coefficient matrix (N = 60) of main soil physical properties in the vegetation recovery plots

BD Sand Silt Clay Aggregates IP MIP MAP Ks

BD 1

Sand 0.853** 1

Silt − 0.539** − 0.845** 1

Clay − 0.787** − 0.609** 0.089 1

Aggregates − 0.345** − 0.433** 0.368** 0.260* 1

IP 0.186 0.014 0.254* − 0.402** 0.084 1

MIP 0.365** 0.348** − 0.180 − 0.380** 0.091 − 0.441** 1

MAP − 0.443** − 0.356** 0.118 0.488** − 0.093 0.156 − 0.432** 1

Ks − 0.834** − 0.760** 0.445** 0.755** 0.830** − 0.251 − 0.323* 0.471** 1

BD bulk density, sand sand contents, silt silt content, clay clay content, Aggregates > 0.25 mm aggregate stability, IP inactive porosity, MIP micropo-

rosity , MAP macroporosity, Ks saturated hydraulic conductivity

*Significant at P ≤ 0.05

**significant at P ≤ 0.01

MAP

>0.25 mm

Aggregates

-0.47

-0.38

0.38

-0.84

0.43

Fig. 6 Path diagram for the relationship between soil properties and

saturated hydraulic conductivity. The magnitudes of the numerical

values in the figure indicate the level of effects on saturated hydraulic

conductivity. MAP, macroporosity; BD, bulk density; Ks, saturated

hydraulic conductivity

724 J Soils Sediments (2019) 19:716–728

important role in anti-erodibility and stabilizing soil structure

(Bissonnais 1996;Erktanetal.2016; Van Hall et al. 2017).

Areas with vegetation restoration had higher aggregate stabil-

ity than farmland (Table 3), which indicates that farmland is

more vulnerable to erosion than soils from the revegetated.

The low aggregate stability of farmland was mainly because

of anthropogenic disturbance resulting from agriculture

(Cerdà 2000). MAP of replanted RS40, RS30, and NG40

were higher while IP and MIP were lower than others, indi-

cating they have better porosity (Fig. 2b–d). BD was signifi-

cantly related to most other soil parameters (Table 4); hence, it

could be used as an indicator of soil structure evolution. The

reductions in the BD of soils from the re-vegetated illustrated

the amelioration effects on soil structure.

Vegetation restoration increased Ks compared with farm-

land (Fig. 5). The higher Ks indicated that soils fro m the

revegetated had better soil permeability than farmland, which

is a cause for runoff reduction in the region (Gao et al. 2011).

Ks correlated with most of the studied soil physical parame-

ters, which demonstrated that Ks also could be used as an

indicator of soil structure change. The stepwise regression

showed that BD, MAP, and > 0.25 mm aggregates stability

were the principal soil physical parameters that affect the Ks.

Furthermore, the indirect effects of MAP via soil BD were

higher than its direct effects, implying the great effect of BD

on Ks (Fig. 6). Many studies in the literature show that there is

a significant exponential relationship between Ks and BD

(Jabro 1992), which is in accordance with our results

(y = 64.435e

−4.806x

, R

= 0.7007, P < 0.001) (Fig. 8a). With

the succession of vegetation, soil BD decreased while porosity

and aggregate stability increased. These changes in the soil

properties enhanced the soil water conductive capability.

MAP was more closely related to Ks than IP and MIP, and

acting as an input of Eq. 9. MAP could form macropore flow

in the soil, which was shown to be a main determinant for Ks

(Ahuja et al. 1984). Biological macropore (e.g., root channels)

hasgoodconnectivityandlargeporediameterthatisbetterfor

the flow processes in the soil (Lado et al. 2004; Ajayi and

Horn 2016). Additionally, the length of vegetation time affects

Ks. Wilson et al. (2005) found that soil matrix infiltration

rate was highest in the newly established (1 year) and the

oldest (> 15 years) revegetated areas, and biogenic surface

crust was the main reason. In our study, Ks increased along

with vegetation time for each vegetation type because there is

no biogenic surface crust for our soil samples.

On the other hand, organic matter accumulation in vegetat-

ed areas plays an important role in Ks increases. Bissonnais

and Arrouays (1997) used a rainfall simulator and found that

reduction of the organic carbon content of a loamy soil below

1.5 to 2.0% decreased the soil infiltration rate. A quadratic

function was identified by others between soil organic matter

and Ks (Celik 2005;LiandShao2006)aswellasthisstudy

(i.e., y = − 0.001 4x

+ 0.0481x +0.0183, R

= 0.6545, P <

0.001) (Fig. 8b). Increases in organic matter were positive-

ly correlated to aggregate formation, reduced BD, and de-

creased the susceptibility of the soil to seal formation

(Lado et al. 2004 ). Those functions were conducive to

the increases of Ks. However, the quadratic relation indi-

cated that the role of organic matter on Ks was not mono-

tonic. There was a threshold value for soil organic matter.

When soil organic matter to affect Ks lower than the

threshold, it played a positive role in Ks. Otherwise, the

role was adverse. The main reason is that soil organic mat-

ter h as retention effect on soil moisture (Van Hall et al.

Dexter

Site

abc

RF35 RF25 RS40 RS30 RS15 NG40 NG30 NG20 NG10 FA

0.00

0.04

0.08

0.12

0.16

0.20

(a) (b)

-1.5 -1.0 -0.5 0.0 0.5 1.0

PC1

PC2

FA FA FA

NG20 NG20

NG10 NG10

MIP

MAP

Sand

Silt

Clay

Aggregates

RS40 RS40 RS40

RS30 RS30 RS30

RS15 NG40 NG40

RF35 RF35 RF35

RF25 RF25 RF25

RS15 RS15 NG40

NG30 NG30

Fig. 7 Results of principle component analysis (PCA) (a)andthesoil

physical quality index (S

Dexter

)(b) for different sample plots based on the

soil properties in 0–20 cm soil layer. BD, bulk density; Sand, sand

content; Clay, clay content; Silt, silt content; Ks, saturated hydraulic

conductivity; MAP, macroporosity; MIP, microporosity; IP, inactive

porosity; Aggregates, > 0.25 mm aggregates stability. The same letter

above the point denotes they are not significantly different at the 0.05

probability level (Duncan’stest)

J Soils Sediments (2019) 19:716–728 725

2017). If the retention effect is higher than the threshold,

the increases of soil organic matter would i mpede the in-

creases of Ks.

4.2 Effectiveness of vegetation restoration

in the Loess Plateau

Due to water shortage in the Loess Plateau, there are many

issues during vegetation restoration (Feng et al. 2012). Some

studies have found that a number of woody (e.g., Pinus

tabulaeformis, Platycladus orientalis) and shrub species

(e.g., Hippophae rhamnoide) grow well at the beginning,

but they often degrade once the initial water supply has been

exploited (Li 2000). Large-scale afforestation in the water-

limited arid and semiarid regions may increase the severity

of groundwater shortages. PCA analysis showed that RS40,

RS30, and NG40 were grouped together (Fig. 7a). They have

the best soil properties among the sites including low BD,

high MAP, and high Ks. However, RF35 and RF25 belong

to the second category (Fig. 7a), of which the soil properties

were of less quality, relative to RS40, RS30, and NG40. The

Dexter

showed similar result as PCA, i.e., RS40, RS30, and

NG40 had highest value while RF35 and FA had the lowest

(Fig. 7b). The findings indicated that the effectiveness of

replanted forest to ameliorate soil health was less than

replanted scrubland. Forest uses more soil water and form

drier soil layers (Li 2000), which leads to soil degradation.

We found that soil moisture of replanted forest was the highest

at 0–30 cm layer, while it sharply decreased thereafter (Fig. 9).

At depths lower than 100 cm, replanted forest had the lowest

soil moisture among the four land-use types (Fig. 9). The

results further indicated that forest consumed more soil water

than others, which was adverse to sustainability of vegetation

restoration. NG20, NG10, and FA were combined into the

same category, which implied that soil natural rehabilitation

of abandoned cropland needs a long time period. If the reha-

bilitation period is long enough (more than 40 years), the

secondary natural grassland could effectively improve soil

properties like the shrub (Fig. 7).

5 Conclusions

Vegetation restoration in the hilly-gully region of the Loess

Plateau, China had significant effects on soil physical proper-

ties and saturated hydraulic conductivity. The bulk density

decreased significantly, while aggregate stability increased

significantly during vegetation restoration. Change of soil po-

rosity was complex, which showed that inactive porosity and

microporosity was higher in replanted forest while

macroporosity was higher in replanted scrubland. Vegetation

restoration resulted in significant increases in soil saturated

hydraulic cond uctivity, but did not alter soil texture. Bulk

density, macroporosity, and > 0.25 mm aggregate stability

were the main soil physical parameters that influenced satu-

rated hydraulic conductivity. Bulk density and saturated hy-

draulic conductivity were strongly correlated with most mea-

sured so il physical properties. Hence, bulk density and

Fig. 8 Relationships between saturated hydraulic conductivity and bulk density (a) and soil organic matter (b)

Fig. 9 Average vertical soil moisture (%) distributions among different

land-use pattern (RF, replanted forest; RS, replanted scrubland; NG,

secondary nature grassland; FA, farmland)

726 J Soils Sediments (2019) 19:716–728

saturated hydraulic conductivity could be used as indicators to

evaluate the impact of vegetation restoration on soil proper-

ties. It is better to plant scrub or grassland than forest in the

study area for eco-environment rehabilitation. When selecting

the tree species for vegetation restoration on the Loess

Plateau, water conditions should be considered as a major

limiting factor.

Acknowledgments This work was supported by the National Natural

Science Foundation of China (41671285) and t he National Key

Research and Development Program of China (2016YFC0501707,

2016YFC0402401).

References

Ahuja L, Naney J, Green R, Nielsen D (1984) Macroporosity to charac-

terize spatial variability of hydraulic conductivity and effects of land

management. Soil Sci Soc Am J 48(4):699–702

Ajayi AE, Horn R (2016) Transformation of ex-arable land to permanent

grassland promotes pore rigidity and mechanical soil resilience. Ecol

Eng 94:592–598

Bartlett MS (1954) A note on the multiplying factors for various χ2

approximations. J R Stat Soc Ser B Methodol 16:296–298

Berger TW, Hager H (2000) Physical top soil properties in pure stands of

Norway spruce (Picea abies) and mixed species stands in Austria.

For Ecol Manag 136(1):159–172

Bissonnais YL (1996) Aggregate stability and assessment of soil

crustability and erodibility: I. Theory and methodology. Eur J Soil

Sci 47(4):425–437

Bissonnais YL, Arrouays D (1997) Aggregate stability and assessment of

soil crustability and erodibility: II. Application to humic loamy soils

with various organic carbon contents. Eur J Soil Sci 48(1):39–48

Celik I (2005) Land-use effects on organic matter and physical properties

of soil in a southern Mediterranean highland of Turkey. Soil Tillage

Res 83(2):270–277

Cerdà A (2000) Aggregate stability against water forces under different

climates on agriculture land and scrubland in southern Bolivia. Soil

Tillage Res 57(3):159–166

Chi CM, Wang ZC (2009) Effect of sand on saturated hydraulic conduc-

tivity and salts leaching of sodic soil. J Soil Water Conserv 23(01):

99–102 +116 (in Chinese)

Dexter AR (2004) Soil physical quality: part I. Theory, effects of soil

texture, density, and organic matter, and effects on root growth.

Geoderma 120(3):201–214

Erktan A, Cécillon L, Graf F, Roumet C, Legout C, Rey F (2016) Increase

in soil aggregate stability along a Mediterranean successional gradi-

ent in severely eroded gully bed ecosystems: combined effects of

soil, root traits and plant community characteristics. Pl ant Soil

398(1–2):121–137

Fang W, Peng SL (1997) Development of species diversity in the resto-

ration process of establishing a tropical man-made forest ecosystem

in China. For Ecol Manag 99(1):185–196

Feng X, Sun G, Fu B, Su C, Liu Y, Lamparski H (2012) Regional effects

of vegetation restoration on water yield across the Loess Plateau,

China. Hydrol Earth Syst Sci 16(8):2617–2628

Fu B, Wang J, Chen L, Qiu Y (2003) The effects of land use on soil

moisture variation in the Danangou catchment of the Loess

Plateau, China. Catena 54(1):197–213

Gao P, Mu XM, Wang F, Li R (2011) Changes in streamflow and sedi-

ment discharge and the response to human activities in the middle

reaches of the Yellow River. Hydrol Earth Syst Sci 15(1):1–10

Horton RE (1945) Erosional development of streams and their drainage

basins; hydrophysicalal approach to quantitative morphology. Geol

Soc Am Bull 56(3):275–370

Jabro J (1992) Estimation of saturated hydraulic conductivity of soils

from particle size distribution and bulk density data. Trans ASAE

35(2):557–560

Jia X, Shao M, Wei X, Horton R, Li X (2011) Estimating total net primary

productivity of managed grasslands by a state-space modeling ap-

proach in a small catchment on the Loess Plateau, China. Geoderma

160(3):281–291

Jiao F, Wen ZM, An SS (2011) Changes in soil properties across a

chronosequence of vegetation restoration on the Loess Plateau of

China. Catena 86(2):110–116

Lado M, Paz A, Ben-Hur M (2004) Organic matter and aggregate size

interactions in infiltration, seal formation, and soil loss. Soil Sci Soc

Am J 68(3):935–942

Li Y (2000) Effects of forest on water circle on the loess plateau. J Nat

Resour 16(5):427

–43

2 (in Chinese)

Li Y, Shao M (2006) Change of soil physical properties under long-term

natural vegetation restoration in the Loess Plateau of China. J Arid

Environ 64(1):77–96

Li X, He M, Duan Z, Xiao H, Jia X (2007) Recovery of topsoil physico-

chemical properties in revegetated sites in the sand-burial ecosys-

tems of the Tengger Desert, northern China. Geomorphology 88(3):

254–265

Li Z, Liu WZ, Zhang XC, Zheng F-L (2009) Impacts of land use change

and climate variability on hydrology in an agricultural catchment on

the Loess Plateau of China. J Hydrol 377(1):35–42

Li W, Yan M, Qingfeng Z, Zhikaun J (2012) Effects of vegetation resto-

ration on soil physical properties in the wind–water erosion region of

the northern Loess Plateau of China. Clean Soil Air Water 40(1):7–

Luxmoore R (1981) Comments and letters to the. Soil Sci Soc Am J 45:

671–673

McVicar TR, Li L, Van Niel TG, Zhang L, Li R, Yang Q, Zhang X, Mu X,

Wen Z, Liu W (2007) Developing a decision support tool for China's

re-vegetation program: simulating regional impacts of afforestation

on average annual streamflow in the Loess Plateau. For Ecol Manag

251(1):65–81

Mu X, Zhang X, Shao H, Gao P, Wang F, Jiao J, Zhu J (2012) Dynamic

changes of sediment discharge and the influencing factors in the

Yellow River, China, for the recent 90 years. Clean Soil Air Water

40(3):303–309

Neris J, Jiménez C, Fuentes J, Morillas G, Tejedor M (2012) Vegetation

and land-use effects on soil properties and water infiltration of

Andisols in Tenerife (Canary Islands, Spain). Catena 98:55–62

Reynolds W, Elrick D, Youngs E, Amoozegar A, Booltink H, Bouma J

(2002) 3.4 Saturated and field-saturated water flow parameters. In:

Dane JH, Topp GC (eds) Methods of soil analysis Part 4. Physical

methods. SSSA, Madison, pp 797–878

Shapiro SS, Wilk MB (1965) An analysis of variance test for normality

(complete samples). Biometrika 52(3–4):591–611

Stolte J, van Venrooij B, Zhang G, Trouwborst KO, Liu G, Ritsema CJ,

Hessel R (2003) Land-use induced spatial heterogeneity of soil hy-

draulic properties on the Loess Plateau in China. Catena 54(1):59–

Van Genuchten MT (1980) A closed-form equation for predicting the

hydraulic conductivity of unsaturated soils. Soil Sci Soc Am J

44(5):892–898

Van Hall R, Cammeraat L, Keesstra S, Zorn M (2017) Impact of second-

ary vegetation succession on soil quality in a humid Mediterranean

landscape. Catena 149:836–843

Wang H, Yang Z, Saito Y, Liu JP, Sun X, Wang Y (2007) Stepwise

decreases of the Huanghe (Yellow River) sediment load (1950–

2005): impacts of climate change and human activities. Glob

Planet Chang 57(3):331–354

J Soils Sediments (2019) 19:716–728 727

Wang Y, Fu B, Lü Y, Chen L (2011) Effects of vegetation restoration on

soil organic carbon sequestration at multiple scales in semi-arid

Loess Plateau, China. Catena 85(1):58–66

Wilson GV, Xu M, Chen Y, Liu G, Romkens MJM (2005) Macropore

flow and mass wasting of gullies in the Loess Plateau, China. Int J

Sediment Res 20(3):249

Xu M, Zhao Y, Liu G, Wilson GV (2006) Identification of soil quality

factors and indicators for the Loess Plateau of China. Soil Sc i

171(5):400–413

Zhang K, Dang H, Tan S, Wang Z, Zhang Q (2010) Vegetation commu-

nity and soil characteristics of abandoned agricultural land and pine

plantation i n the Qinling Mountains, China. For Ecol Manag

259(10):2036–2047

Zhang YW, Deng L, Yan WM, Shangguan ZP (2016) Interaction of soil

water storage dynamics and long-term natural vegetation succession

on the Loess Plateau, China. Catena 137:52–60

728 J Soils Sediments (2019) 19:716–728