Changes of soil quality induced by different vegetation restoration in the collapsing gully erosion areas of southern China

Hui Wen,Shimin Ni,Junguang Wang,Chongfa Cai

Key Laboratory of Arable Land Conservation(Middle and Lower Reaches of Yangtze River),Ministry of Agriculture and Rural Affairs,College of Resources and Environment,Huazhong Agricultural University,Wuhan,430070,China

Keywords:Vegetation restoration Collapsing gully Soil saturated hydraulic conductivity Soil quality index

A B S T R A C T Understanding the influence of collapsing gully management restoration on soil quality and function is essential to the protection of the regional ecological environment in the collapsing gully erosion area.The primary objective of this study was to construct soil quality index(SQI)to assess the influence of different vegetation restoration types on soil quality in collapsing gully restoration.The influence of five vegetation restoration types on soil properties was investigated by using a path analysis,a comprehensive soil quality index(SQI),and a general linear model(GLM).Vegetation restoration was shown to significantly increase the saturated hydraulic conductivity(Ks),mainly due to the effect of the physical parameters of bulk density,soil cohesion,and soil water content.Meanwhile,pH,Ks,soil organic matter(OM),and sand content were revealed as reasonable indicators to evaluate the influence of vegetation restoration on soil quality.Moreover,vegetation restoration was found to significantly improve the soil quality,with the highest SQI value for natural restoration mixed forest(NF),followed by replanted arboreal forest(RA)and replanted scrubland(RS),which were all significantly higher than the SQI value of the erosion area(EA)in the collapsing gully.Additionally,vegetation type explained the most substantial proportion of total variability(46.41%),and restoration time showed a positive correlation with SQI.The results of this study can provide a reference for the restoration and protection of the regional ecological environment in the collapsing gully area.

1.Introduction

Land is mainly a non-renewable resource on a time scale related to agricultural production and economic development(Lal,2015).Soil erosion leads to the removal of soil nutrient and degradation of soil structure is considered to be a major environmental and social issue affecting the ability of a range of soil functions worldwide(Singh & Panda,2017).As a widespread phenomenon of soil erosion,collapsing gully is one of the most severe land erosion problems in Southern China(Liao et al.,2019),which is usually characterized by the emergence of fine,shallow and cut trenches on the slope with increasing runoff and erosion and is currently considered to be an important target for soil erosion control in China(Liao et al.,2019;Wang,Zhang,et al.,2019;Zhong et al.,2013).The collapsing gully erosion occurs mainly in the hilly granite areas of tropical and subtropical South China(Liao et al.,2019;Xia et al.,2016),where the locally known as Benggang and the tropical monsoon climate conditions can promote the weathering and soil erosion of bedrock and accelerate the occurrence of collapsing gully(Tao et al.,2017).The area of collapsing gully soil erosion in the seven provinces of southern China has been reported to be more than 1200 km2,with a total annual erosion of more than 60 million tons,affecting the surrounding environment of nearly 20 thousand km2(Feng et al.,2009).

In this context,in order to effectively control water and soil erosion,protect water and soil resources,and restore regional ecological functions,some scholars have carried out a lot of research on the prevention and management of collapsing gully erosion and analyzed the formation mechanism(Xiao et al.,2014;Xu,1996).The fractal features of soil particle size distribution in alluvial fans of collapsing gully were studied in terms of soil fractal dimension,physico-chemical properties and their relationship with different land-use patterns(Deng et al.,2018).The structural,slope directional and morphological characteristics were also investigated in collapsing gullies(Deng et al.,2020;Lin et al.,2017).Due to erosion and the disturbance of certain human activities,soil structure varies from undisturbed soil in collapsing gully area.The highest degree of soil degradation was found during the active stage of collapsing gully,and the large amount of sediment produced from erosion can damage roads,farmlands,and rivers adjacent to the permanent gully(Feng et al.,2019).The relationships between the soil properties and the effects of rainfall and slope on soil erosion have also been investigated by Zhu et al.(2020)and Jiang et al.(2014).

Soil quality,as one of the many components of the agroecosystem,refers to the ability of the soil to function within the boundaries of ecosystem,maintain the quality of the bioenvironment,and promote the health of animals and plants(Andrew et al.,2002,2004;Karlen et al.,2003).The collapsing gully area of southern china is particularly susceptible to severe soil erosion due to its geotechnical property and geographical environment,and soil quality has been significantly reduced as a consequence of soil degradation causing by anthropogenic and natural factors.

Exposed slopes are often found in the area of collapsing gully erosion.Vegetation control measures are used to management collapsing gully to increase the strength of the surface soil,increase the soil’s anti-erosion ability,and can also intercept and maintain rainfall(Feng et al.,2019).It can also effectively improve the underground ecosystem,increase the soil nutrient content,improve the soil physical structure,and increase the soil biomass and activity through the input of litter and roots(Neris et al.,2012;Zhang et al.,2010,2019).Therefore,vegetation has been widely used to reduce soil erosion,soil degradation,while providing opportunities for studying the mechanisms for its effect on soil quality at different stages of recovery(Guo et al.,2018;Wang et al.,2012),which may facilitate the understanding of the relationship between plant species and soil characteristics and the selection of the most suitable species for restoration(Lin et al.,2015).Previous studies have shown that the effects of vegetation restoration on soil properties are different with the different vegetation types.Hao et al.(2020)assessed the effect of vegetation restoration on soil infiltrability in heavy-textured soil and found that soil organic matter,aggregate stability,soil total porosity,and non-capillary porosity have been significantly ameliorated in the order of grass-planted＞shrubplanted＞bare soil.Moreover,the effects of vegetation on soil physical properties varied with restoration years.Different fundamental parameters such as soil bulk,total phosphorus,protease activity and soil shear strength are often used to evaluate characteristics of soil quality changes under different vegetation treatments(Raiesi & Kabiri,2016;Tang et al.,2019).Vegetation restoration can also be evaluated by other parameters,including saturated hydraulic conductivity,aggregate stability,and macroporosity(Gu et al.,2019).

Soil quality index(SQI)was an appropriate method for evaluating the reclaimed mine spoil(Mukhopadhyay et al.,2016),and also a practical tool for evaluating and monitoring soil quality in a vegetation restoration area for its quantitative flexibility and accuracy(Tang et al.,2019;Yu et al.,2018;Zhang et al.,2019).Nevertheless,soil quality cannot be evaluated by individual soil properties,thus soil quality index(SQI)was established to comprehensively assess soil quality(Andrews et al.,2002;Nosrati& Collins,2019).Some researchers used the smallest data set to determine soil quality values to analyze the changes in soil quality in different regions and assess their impact on the ecological environment(Mukhopadhyay et al.,2016;Zhang et al.,2019).Feng et al.(2019)identified the total nitrogen,total phosphorus,pH,soil cohesion and root density as major factors affecting the soil degradation and spatial variability in soil quality in a collapsing gully erosion area.

The effects of vegetation restoration on soil quality have focused on the influence of different soil tillage practices(Raiesi & Kabiri,2016),land use(Gu et al.,2019;Neris et al.,2012),and agricultural soil management(Masto et al.,2008),as well as the impact of vegetation restoration on the soil characteristics in Loess Plateau(Deng et al.,2018).Little attention has been paid to the effect of vegetation restoration on soil quality in the collapsing gully area after management,which is of great significance for studying the process of soil and ecological environment restoration in the collapsing gully area.Additionally,the mechanisms for the effect of these factors on this area also remain poorly understood.

The objectives of this study were to(i)evaluate the changes trend of soil saturated hydraulic conductivity under different vegetation restoration,(ii)identify the predictors for changes in soil saturated hydraulic conductivity based on the soil characteristics,and(iii)evaluate the changes of soil quality under different vegetation restoration and main impact factors.Specifically,the effect of vegetation restoration on the soil basic properties of collapsing gully soil was first systematically evaluated.Then,the relationship between soil physical properties and saturated hydraulic conductivity was investigated.Finally,the minimum data set(MDS)established based on the results of existing and previous studies was used to determine the SQI and assess the change trend of soil quality in different vegetation restoration.

2.Materials and methods

2.1.Study sites

This study was carried out in the Jingouxing watershed(latitudes 26°10′N to 26°13′N and longitudes 115°9′E to 115°13′E)in the north of Gan County,Jiangxi Province,China(Fig.1).This area is characterized by a humid subtropical climate with an average annual temperature and annual rainfall of 19.3°C and 1394 mm,respectively.The rainy season occurs during April-July.The parent material is mainly weathered granitic rocks,with the rock mass weathering rapidly and the weathered crust having abundant fracture surfaces and joint surfaces.The soil types are primarily red soil,paddy soil,and alluvial soil,with the vegetation species being mostly Pinus massoniana and Dicranopteris linearis.It is a typical collapsing gully region with an area of 95.8 hm2.In this region,eight typical collapsing gullies were selected,and in each collapsing gully,three spatial locations were divided from top to bottom in the units of slope surface as upper slope(US),middle slope(MS),and lower slope(LS).

2.2.Soil sampling and laboratory analysis

Seven collapsing gullies that represent the vegetation restoration after management in this region were selected from the watershed:3 with replanted arboreal forest(RA)for 8-13 years(RA8,RA10,RA13),2 with replanted scrubland for 3-9 years(RS3,RS9),1 with artificial grassland for 2 years(AG2),1 with natural restoration mixed forest for 30 years(NF30),and one collapsing gully erosion area(EA)as reference(Table 1).

Table 1 Information on the selected sites.

Fig.1.Location of study sites in the north of Gan County,Jiangxi,China.

The soil samples were collected from the surface layer(0-20 cm)of each slope surface among adjacent vegetation and organized in triplicate through an“S”sampling strategy in the summer of 2018.Soil samples(5-6)were collected at the spatial location of each study site and a total of 121 soil samples were obtained for the determination of soil physical properties.At each sampling site,soil samples(～1-2 kg)were collected and transferred to the laboratory to remove the crop roots and weeds,followed by air-drying,crushing,and sieving to determine the physical and chemical properties.The bulk density(BD),capillary porosity(CP),and water content(WC)of the soil were measured separately by the steel ring(100 cm3in volume,5.02 cm in diameter and 5.05 cm in height)and gravimetric method,and these indices are widely used to indicate soil quality(Raiesi,2017;Zhang et al.,2011).A steel ring(60 cm3in volume,6.18 cm in diameter and 2.00 cm in height)was used to measure the soil cohesion(SC),and the shear rate was calculated using the Mohr-Coulomb shear strength criterion equation(Zhang et al.,2018):

where τ(kPa)is the shear strength;SC(kPa),the soil cohesion;P(kPa),the normal stress acting on the failure surface(σ=50,100,150 and 200 kPa);φ(°),the angle of internal friction.

All soil samples were measured and analyzed in the summer and autumn of 2018,and all tests were repeated three times.Soil mechanical composition was measured using the pipette method and the soil particle size was classified according to USDA(United States Department of Agriculture)standards.pH was measured using a pH meter model(soil to water ratio of 1:2.5).Soil organic matter(OM)was measured using the K2Cr2O7wet oxidation method.Soil cation exchange(CEC)was measured by the CH3COONH4exchange method.The contents of total nitrogen(TN)and available nitrogen(AN)in soil extracts were measured based on a Flow Injection Analysis System(FIAS 400 PerkinElmer,Inc,CT,USA)equipped with an AS90 Autosampler(PerkinElmer).Flame atomic absorption spectrophotometry was used to determine the content of available potassium(AK).Total phosphorus(TP)and available phosphorus(AP)were measured by the Olsen method.Saturated hydraulic conductivity(Ks)was measured by the constant water head method using distilled water under a 5 cm head(Reynolds et al.,2002).

2.3.Assessment of soil quality index

SQI is a decision tool that can be effectively combined with various information for multi-objective decision-making.The selection of SQI should be based on the functional characteristics of the soil and the management objectives defined by the system,i.e.,the appropriate indicators selected for assessment of the soil functions can help to achieve management goals(Andrew et al.,2002).The calculation of SQI mainly consists of the following three steps:the selection of the MDS,the score of the selected indicators,and the calculation of the SQI comprehensive value(Andrew et al.,2004;Brejda et al.,2000).Principal component analysis(PCA)and Person’s correlation analysis were used to select the MDS indicators,coupled with the purpose of the research.MDS was composed of the components with eigenvalues≥1 and each explaining＞5%of data variation(Andrew et al.,2003;Bastida et al.,2006;Wu et al.,2019).Then,these selected soil parameters were transformed into unitless scores by using a nonlinear scoring function(Masto et al.,2008).The sigmoidal function(Eq.(1))was used as follows(Andrews et al.,2002):

where Siis the score of the selected soil indicator;α,the maximum score of 1;xi,the value of the indicator;x0,the mean value of each soil indictor;b values of--2.5 and 2.5 were used for a“more is better”or a“less is better”curve,respectively(Bastida et al.,2006).The weights of the soil quality factor(Wi)were calculated with the score coefficient by the following equation(Xu et al.,2009):

where Ciis the score coefficient of the soil quality factor i;n,the number selected from the MDS.Furthermore,the SQI is calculated by the following equation(Masto et al.,2008):

where Siis the indicator score calculated by Eq.(2);Wi,the weighting values calculated by Eq.(3).

2.4.Statistical analysis

Analysis of variance(ANOVA;T-test)and least significant difference(LSD)were used to assess the differences of the soil properties among the eight sites at a significance level of P<0.05.Pearson correlation analysis was also used to determine the optimal factors and correlation matrices among soil indicators.Statistical analysis and data processing were performed using SPSS 22 and Origin 2021,respectively.ArcGIS 10.6 was used to generate the study area map.

3.Results

3.1.Soil physicochemical indicators under different vegetation restoration types

Table 2 shows the mechanical composition(%)for different vegetation types.The content of clay and silt showed an increase with increasing restoration time.In the 8 study sites,natural restoration mixed forest(NF30)and replanted arboreal forest 13(RA13)had a similar soil texture,sand content(9.12%-21.86%)and silt content(33.06%-51.83%),which were significantly lower than those at the other sites.Artificial grassland(AG2)was significantly higher(P<0.05)than the other sites in sand content(12.27%-20.27%),and collapsing gully erosion area(EA)was significantly higher than the other sites in silt content(85.73%)in the middle slope(MS).

Table 2 Mechanical composition(%)for different vegetation types at different study sites.

The effects of different vegetation types on the soil physical properties are shown in Fig.2.Bulk density(BD)was shown to decrease with increasing restoration time,with a significantly lower(P<0.05)value in the NF30 site than in the AG2 site.Moreover,with the increase of elevation at different slopes,the BD value showed an upward trend first,followed by a decrease in the sites of NF30 and Replanted arboreal forest 9(RA8).The RA13 had the lowest BD(1.15 g cm-3)in the upper slope in all the 8 study sites.In Fig.2b,NF30 was seen to be significantly higher(P<0.05)than the other sites in soil water content(WC),with the maximum value being 2.12-fold the minimum value.Moreover,except for RA8 and AG2,the other restoration sites showed the highest WC in the upper slope and the lowest WC in the middle slope.The soil cohesion(SC)exhibited a consistent trend with WC,showing an increase with increasing recovery time(Fig.2c).Specifically,the SC value was highest in NF30 and lowest in AG2.Except for RA8,AG2,Replanted scrubland 3(RS3)and EA,the other four recovered sites had the highest SC value in the lower slope and the lowest value in the middle slope.AG2 had the lowest SC value in the middle slope in all the eight vegetation-recovered sites.

The accumulation of nutrients on the soil surface is the result of biological processes regulated by plants and soil organisms,as well as the complex interaction between the atmosphere and biochemical processes(Chen et al.,2015).Fig.3 shows the effects of different vegetation restoration on the soil chemical properties.In Fig.3a,NF30 was seen to have the highest value of soil total phosphorus(TP),which was significantly(P<0.05)higher than that of the other sites.Except for Replanted scrubland 9(RS9),RA8 and AG2,the other sites showed a significantly lower TP value in the middle slope(MS)than in the other locations.Similar to TP,available phosphorus(AP)(Fig.3b),total nitrogen(TN)(Fig.3c)and available nitrogen(AN)(Fig.3d)exhibited significant difference in the eight study sites.All the four indicators had the highest value in NF30 and the lowest value in AG2.The TP was seen to be the highest for NF30(0.78 g kg-1)and the lowest for AG2(0.04 g kg-1).Additionally,AG2 had a significantly lower TN value(0.13 g kg-1-0.15 g kg-1)than the other restoration sites at the three locations.

After three years of vegetation restoration,the accumulation of AN also increased significantly when compared with soil organic matter(OM)(Fig.3f)and AP.NF30 exhibited the highest AN value(72.23 mg kg-1-78.20 mg kg-1),with no significant difference between EA and AG2 in the AN values(6.82 mg kg-1-20.02 mg kg-1)versus(6.82 mg kg-1-8.73 mg kg-1).Additionally,EA and AG2 also showed no significant difference in the value of soil available potassium(AK)(Fig.3g).Generally,soil cation exchange capacity(CEC)is an indicator for its fertility and nutrient retention capacity(Brady&Weil,2007).In Fig.3e,CEC was seen to have no significant difference in RA10,RA8 and RS9,with the lowest value observed in EA(6.30 cmol kg-1-6.75 cmol kg-1).At the lower slope,soil AK generally increased with increasing vegetation restoration time,while CEC did not increase significantly until eight years of vegetation restoration.

3.2.Evaluation of saturated hydraulic conductivity

Fig.4 shows the soil KSunder different vegetation types in the eight study sites.The KSvalues varied significantly(P<0.05)andranged between 0.014-0.123 cm min-1,probably due to the difference of vegetation restoration types and restoration time.The KSvalues increased gradually with increasing vegetation recovery time.Of all the vegetation types,NF30 was significantly higher(P<0.05)than EA or AG2 in the KSvalue.Compared with other slopes,the middle slope showed a lower KSvalue in most study sites,with the lowest value observed in the EA middle slope,which was 0.12 times the KSvalue of NF30.

Fig.2.Soil physical indicators under different vegetation types in the eight study sites at upper(US),middle(MS)and lower slopes(LS).(a)Bulk density(BD);(b)Soil water content(WC);(c)Soil cohesion(SC).Different uppercase letters indicate significant difference under different parts of vegetation types(one-way ANOVA,P<0.05).Different lowercase letters indicate significant difference among different study sites at the same parts(one-way ANOVA,P<0.05).

Fig.3.Soil chemical indicators under different vegetation types in the eight study sites at the upper(US),middle(MS)and lower slopes(LS).(a)Total phosphorus(TP);(b)Available phosphorus(AP);(c)Total nitrogen(TN);(d)Available nitrogen(AN);(e)Cation exchange capacity(CEC);(f)Organic matter(OM);(g)Available potassium(AK).Different uppercase letters indicate significant difference under different parts of vegetation types(one-way ANOVA,P<0.05).Different lowercase letters indicate significant difference among different study sites at the same parts(one-way ANOVA,P<0.05).

Fig.4.Saturated hydraulic conductivity for the soil in the upper(US),middle(MS)and lower slope(LS)of the eight study sites.Different uppercase letters indicate significant difference under different parts of vegetation types(one-way ANOVA,P<0.05).Different lowercase letters indicate significant difference among different study sites at the same parts(one-way ANOVA,P<0.05).

Fig.5.Correlation analysis of main soil physical properties under different vegetation types in the eight study sites.Clay:clay content(%);Silt:silt content(%);Sand:sand contents(%);BD:bulk density(kg.m-3);CP:capillary porosity(%);WC:water content(%);SC:soil cohesion(kPa);Ks:saturated hydraulic conductivity(cm.min-1).Indicated values represent the correlation coefficients.The green color indicates a positive correlation,and the orange color indicates a negative correlation at significant levels of*P≤0.05;**P≤0.01.

3.3.Effects of soil physical properties on saturated hydraulic conductivity

Fig.5 shows the correlation coefficients of 8 main soil physical properties.BD was found to have a significant negative correlation(P<0.01)with most of measured soil physical properties.Meanwhile,clay content was positively related to most soil properties(except for silt content,sand content and BD).CP was negatively correlated with other soil physical properties except for clay content,WC,SC and KS.SC was significantly related to most soil properties except for sand content.The KSwas shown to have a significant positive(P<0.01)correlation with WC and SC,in contrast to a negative relation(P<0.01)with BD.

The optimal factors affecting KSwere selected by stepwise regression.The results showed that when WC(X1),BD(X2),and SC(X3)were selected,the regression equation was expressed as follows:

The conceptual path model explained 87%of the variance in Ks(Fig.6a).Because of changes in water content(WC),soil cohesion(SC)could also cause a change in bulk density,thus affecting the value of soil saturated water conductivity.SC was seen to have the highest direct effect on KS(standardized path coefficients of 0.76,P<0.001),followed by BD(standardized path coefficient of--0.30,P<0.001),with the lowest effect from WC(standardized path coefficient of 0.02)(Fig.6a).Moreover,WC and SC had the direct positive effect on KS,in contrast to the negative effect for BD.In terms of indirect effects,SC had the largest influence(standardized path coefficients of--0.44,P<0.01),followed by WC(standardized path coefficients of--0.40,P<0.05),and both of them partially affected KSby influencing BD.Additionally,SC was higher in direct effect than in indirect effect,while WC was lower in direct effect than in indirect effect,and both of them affect KSthrough BD,indicating the essential impact of BD on KS.This agreed with the previous finding that BD and WC are important for KS,due to their influence on KSthrough＞0.25 mm aggregate stability and microporosity(Gu et al.,2019;Wu et al.,2016).SC was still observed to have the most obvious total effect on Ks,with a standardized path coefficient of 0.89(Fig.6b).

3.4.Evaluation of SQI

The PCA was used to group the experiment soil indicators(Table 3,Fig.9).Eigenvalues of the first two PCs explained 81.42%(PC1=64.88%,PC2=16.54%)of the total variance and were all≥1,indicating that they could explain most information of the data structure.In the first principal component(PC-1),the highly weighted indicators were TP,AP,TN,AN,OM,AK,Clay,Silt,KSand SC.In the second principal component(PC-2),pH and Sand content were the highly weighted indicators.The redundant variables were eliminated by correlation analysis within each PC(Fig.7).In PC-1,TP,AP,TN,AN,AK,Clay,Silt,KSand SC were highly correlated(P<0.01)with OM,so only OM was retained in the MDS.In PC-2,there was no significant correlation between pH and Sand(P＞0.05),and the correlation coefficient was less than 0.6,so pH and Sand were retained in the MDS.Considering SC is a vital soil mechanical property for predicting wind shear resistance,this parameter was still retained in the MDS,and KSwas selected for the MDS due to its lowest sum of correlation.Finally,OM,KS,SC,pH and Sand were selected in the MDS for calculation of SQI.

Table 3 Principal component analysis of soil quality indicators.

The SQI results for different vegetation types with a different restoration time in the eight study sites are shown in Fig.8.In Fig.8a,SQI was seen to be the largest for NF30(0.66),followed by RA13(0.60),RA10(0.57),RA8(0.51),RS9(0.46),RS3(0.45),EA(0.26),and AG2(0.23)in the eight different sites.In the three different spatial locations,SQI showed the lowest value in the middle slope due to its large spatial variability.Collectively,the SQI was 2.26-to 3.58-fold higher in NF30 than in AG2,and the SQI for EA was 0.14,0.31 and 0.33 in the middle,upper and lower slopes,respectively.In a similar vegetation type,the SQI was lower in the middle slope than in the other two slopes(Fig.8b).

Fig.6.Path diagram for the relationship between soil properties and saturated hydraulic conductivity.(a)Conceptual path model.The numbers on arrows are standardized path coefficients,with solid arrows indicating positive and dashed arrows negative;numbers in bold indicate a significant effect(*P<0.05;**P<0.01;***P<0.001);percentages close to the boxes indicate the variance explained by the model(R2).(b)The standardized path coefficients for direct and indirect effects of SC,WC and BD on saturated hydraulic conductivity.

4.Discussion

4.1.Changes of soil properties under different vegetation types

A number of studies have shown that vegetation restoration can reduce BD,increase soil shear strength,and improve soil nutrients(Gu et al.,2019;Hao et al.,2020;Zhu et al.,2020).In the present study,no significant change was observed in soil texture(Table 2)for most vegetation restoration types except for AG2.Under different types of vegetation restoration,the sand content gradually decreased while the clay content and silt content gradually increased,indicating that vegetation can prevent the loss of fine soil particles(Gu et al.,2019).Moreover,AG2 showed the highest BD value and the lowest SC value,mainly due to the short recovery time(2 years)of artificial grassland and the poor soil structure.In contrast,NF30 was lower in soil BD value and higher in SC and WC values.The lower soil bulk density(BD)of NF30 may be related to higher biological activity,root penetration,nutrient incorporation,etc.As reported previously,naturally restored trees were rich in soil biodiversity,high in vegetation,and could enhance soil porosity by vegetation roots(Feng et al.,2019).The increase in soil SC and WC after vegetation restoration illustrated the improvement of soil structure.

Fig.7.Correlation coefficients of 12 soil properties determined for different vegetation types in the eight study sites.TP:total phosphorus(g.kg-1);AP:available phosphorus(mg.kg-1);TN:total nitrogen(g.kg-1);AN:available nitrogen(mg.kg-1);OM:soil organic matter(g.kg-1);AK:available potassium(mg.kg-1);Clay:clay content(%);Ks:saturated hydraulic conductivity(cm.min-1);Silt:silt content(%);SC:soil cohesion(kPa);Sand:sand contents(%).Indicated values represent the correlation coefficients.The green color indicates a positive correlation,and the orange color indicates a negative correlation at significant levels of*P≤0.05;**P≤0.01.

KSis related to most soil physical parameters,including soil moisture content,soil texture and soil porosity(Franzluebbers,2002;Wang et al.,2017).Stepwise regression results showed BD,WC,and SC as the main soil physical parameters that affect KS,indicating that Ks could also be used as an indicator of soil properties changes(Fig.6).Meanwhile,the indirect effect of WC on KSthrough soil BD was higher than its direct effect,indicating that BD has a great influence on KS(Neris et al.,2012;Tang et al.,2019).Compared with the replanted scrubland,replanted arboreal forest increased KS(Fig.4),indicating that the soil permeability is better in the replanted arboreal forest than in the replanted scrubland,which is the reason for the decrease of runoff in the replanted arboreal forest area(Hu et al.,2008).

Fig.8.(a)Soil quality index(SQI)under different vegetation types in the eight study sites.(b)Soil quality index(SQI)at different spatial locations in the eight study sites.

Fig.9.Results of principal component analysis.

In this study,the physicochemical indicators of the soil varied in the seven sites with four different vegetation types,indicating that the vegetation type plays an important role in the nature of the soil(Zhang et al.,2011).The soil properties are most significantly improved in the study sites with 3 years of artificial vegetation restoration versus collapsing gully erosion area,indicating that vegetation restoration can change the soil physical properties and increase the nutrients significantly(Jia et al.,2017).

The overall stability of replanted arboreal vegetation was higher than that of shrubs,indicating that arboreal forest is more resistant to soil erosion than shrubs in vegetation-restored areas.Compared with EA soil,AG2 soil was poor in soil structure and nutrients.These differences in effects can be attributed to the shorter vegetation restoration time,the lower grassland root consolidation effect on the soil and certain human disturbance on grassland.

4.2.Evaluation of SQI and indicators

MDS,an effective evaluation method because of its small data volume,high accuracy and fast processing speed,has been widely used in many related studies(Raiesi,2017;Yu et al.,2018).In this study,the five soil parameters(OM,SC,KS,pH and Sand content)were finally selected for SQI calculation because we considered them as the most important indicators of soil quality.Among the five indicators,OM was slightly higher in the weighted value and greater in the contribution to SQI than the other indictors.The shear strength of soil,which can directly reflect the difficulty of soil deformation under external force,is also an important index to characterize the soil mechanical properties,and texture has also been reported to have a significant correlation with soil quality(Zhang et al.,2018;Zhu et al.,2020).In this study,soil cohesion(SC)was selected as an important indicator only second to OM.Additionally,soil infiltration is the basis for discussing surface runoff and has an impact on other physical properties,so soil KSis an important indicator for evaluating soil erosion resistance,and has also been selected into the minimum data set(Hao et al.,2020;Wang,Zhang,et al.,2019).A previous study showed that lowering soil pH can provide a more suitable environment for soil optimization and should be used as an important parameter for evaluating ecological restoration(Mahar et al.,2016).The results in Fig.7 showed that pH was significantly related to clay content and nutrient content(P<0.01).

I fell in love with the minister s son the winter I turned fourteen. He was not Chinese, but as white as Mary in the manger. For Christmas I prayed for this blond-haired boy, Robert, and a slim new American nose. When I found out that my parents had invited the minister s family over for Christmas Eve dinner, I cried. What would Robert think of our shabby Chinese Christmas? What would he think of our noisy Chinese relatives who lacked proper American manners? What terrible disappointment would he feel upon seeing not a roasted turkey and sweet potatoes but Chinese food?

4.3.Effects of vegetation types on soil quality

Several studies have shown that revegetation can significantly improve the physical,chemical,and biological properties of the soil(Hao et al.,2020;Yu et al.,2018;Zhu et al.,2020),which is consistent with the results of our study.Furthermore,in the process of vegetation restoration,many problems arise from the different governance patterns of collapsing gully(Feng et al.,2019).Previous studies have demonstrated that some replanted shrubland species grow well initially,but once water supply is developed,they tend to degenerate(Li,2000).Improper vegetation management and planting may exacerbate the severity of groundwater shortage and soil degradation.Soil quality can be affected by many factors,such as its parent material,environmental factors like terrain and rainfall,land-use vegetation types and human activities(Hao et al.,2020;Karlen et al.,2006).

Vegetation types vary in their ability to improve soil quality(Yu et al.,2018).Marzaioli et al.(2010)pointed out that mixed forests have the highest SQI,followed by shrubs and farmland,with the highest SQI for NF30,which is consistent with the observation of this study(Fig.8a).Specifically,the average SQI value was highest in NF30 while lowest in AG2.The improvement of soil quality in shrubland and arboreal can be attributed to the increased vegetation coverage and biodiversity after revegetation,which can improve the soil structure and the absorption of nutrients in plant roots and increase the input of soil nutrients(Leung et al.,2015;Moscatelli et al.,2007).In this paper,the SQI was significantly higher in RA13 than RS9 or RA8 and in the order of RA13＞RA8＞RS9,indicating a better recovery in replanted arboreal forest than in replanted shrubland.The lowest soil quality for AG2 can be attributed to the following reasons:(i)short restoration time;(ii)loose soil surface layer due to weak root consolidation of vegetation;(iii)more human interference;(iv)lower soil organic matter input by the corresponding vegetation.

In this study,the soil quality was the lowest in the middle slope(Fig.8b),probably due to less vegetation coverage on the middle slopes.Moreover,based on the average SQI at all the three soil locations,the SQI was higher in replanted arboreal forests(RA13(0.60)and RA10(0.57))than in replanted scrubland(RS9(0.45)and RS3(0.46)),indicating that both arboreal vegetation and shrubland vegetation are suitable for soil remediation,but planting arbor was more effective.The results of the general linear model(GLM)showed that vegetation type accounts for 46.41% of the SQI variation,which was the largest compared with other indicators(Fig.10).However,the interaction of soil location with vegetation type and soil location also caused the change of SQI by 24.36%and 17.64%,respectively,and other uncertain factors explained 3.07%of SQI changes.In this study,it is shown that vegetation restoration can improve soil quality and optimize the ecological environment by improving the basic characteristics of the soil,and replanting arboreal forests can better restore soil quality than replanting shrublands.This indicates that plant cover,which has a direct impact on soil quality,is of great significance for improving soil parameters and reducing soil erosion.

In PCA analysis,NF30,RA13,RA10,and RA8 were grouped(Fig.9),probably due to their good soil properties,including low BD,but high soil nutrients,SC and KS.Previous studies have shown that replanting arboreal vegetation can improve soil quality better than replanting shrubland vegetation(Marzaioli et al.,2010),probably due to the following three reasons:(i)a higher arboreal coverage can effectively intercept part of the rainfall and reduce the loss of nutrients under the forest;(ii)the surface litter can also provide nutrients to the soil;(iii)deep arboreal roots facilitate nutrient absorption and utilization(Yan et al.,2020).The results of this study show RS9 and RS3 have lower soil quality than RA13,RA10 and RA8,with the lowest values for AG2 in all the vegetation types.Similar to our results,Zhang et al.(2011)mentioned that,compared with artificial vegetation selected by humans,natural vegetation is produced through spontaneous natural succession without any artificial influence,enabling it to better adapt to natural environmental conditions and restore soil structure.With the increase of recovery time(years),soil quality is on the rise,and soil nutrients are further enhanced.However,there is no significant change in the soil quality during the initial restoration of artificial grassland,probably because the plant grows slowly in that period,and the grass litter is relatively small.Meanwhile,the grassroots system is shallow,the soil consolidation is weak,and the plant also needs to absorb some nutrients for growth.We found that the soil quality was the lowest in the middle slope layer of the replanted vegetation(Fig.8b).Due to the small vegetation cover and the influence of certain human activities,the soil nutrient content was lower in the middle slope than in the other locations.Runoff can take away a lot of soil nutrients and fine particles,and soil nutrients show an upward trend from the middle slope to the lower slope,which agreed with the findings of Ni and Zhang(2007)and Zhang et al.(2016).

Fig.10.Results obtained from the general linear model(GLM),showing the integrative effects of vegetation type,location,interaction of vegetation type and location,and the elevation of the study area on soil quality index(SQI).

5.Conclusion

Effects of vegetation types with different restoration time(years)on the soil quality of vegetation-restored collapsing gully was evaluated by SQI.There are significant differences between the soil physicochemical properties in eight different vegetation restoration sites,indicating that changes in vegetation restoration has a considerable impact on soil properties.The soil bulk density was significantly reduced,and the saturated hydraulic conductivity was significantly increased.The main soil physical parameters affecting soil saturated hydraulic conductivity include soil bulk density,water content and soil cohesion.

The results of the study showed that KS,pH,SC,OM,and sand content could be used as effective indicators for evaluating the effect of vegetation restoration on soil quality.SQI values were slightly higher in replanted arboreal and replanted scrubland areas than in the erosion area in the collapsing gully,implying that planting cash crops in the vegetation-restored collapsing gully is not only suitable for restoring soil quality,but also brings certain economic benefits.The SQI value of the replanted artificial grassland did not change significantly during the initial period of restoration.However,as the foundation of landscape development,artificial grassland cultivation is superior to local tourism development and can lay a foundation for soil restoration by bringing certain economic and social benefits.

The GLM results indicate that selecting the suitable vegetation type for soil restoration is vital for improving soil quality,and when selecting vegetation for collapsing gully restoration in the future,the stereoscopic collocation of arboreal-shrub vegetation is revealed as the first choice for vegetation restoration.Furthermore,SQI is shown as an effective and practical tool for evaluating and monitoring soil quality.However,for a more comprehensive and accurate assessment of soil quality,future studies should consider the plant growth parameters,land surface roughness,and the variation trend of soil SQI at different levels,as well as the differences between vegetation types under the same recovery time(years).

Declaration of competing interest

We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.

Acknowledgments

We thank the financial support for this research provided by the National Natural Science Foundation of China(41630858),the National Key Research and Development Program of China(2017YFC0505404)and the National Natural Science Foundation of China(41771304).We also thank the editor and anonymous reviewers for their valuable comments and suggestions on improving this manuscript.