Yonghui Bai·Xuan Zha·Shifa Chen

Abstract We evaluated the effects of the number of years of restoration of vegetation on soil microbial community structure and biomass in degraded ecosystems.We investigated the microbial community structure by analyzing their phospholipid fatty acids then examined microbial biomass carbon and nitrogen by chloroform fumigation extraction of restoration soils over several years.The data were compared with those of highly degraded lands and native vegetation sites.The results show that the duration of vegetation on the sites substantially increased microbial biomass and shifted the microbial community structure even after only 4 years.However,microbial communities and biomass did not recover to the status of native vegetation even after 35 years of vegetation cover.A redundancy analysis and Pearson correlation analysis indicated that soil organic carbon,total nitrogen,available potassium,soil water content,silt content and soil hardness explained 98.4% of total variability in the microbial community composition.Soil organic carbon,total nitrogen,available potassium and soil water content were positively correlated with microbial community structure and biomass,whereas,soil hardness and silt content were negatively related to microbial community structure and biomass.This study provides new insights into microbial community structure and biomass that influence organic carbon,nitrogen,phosphorus and potassium accumulation,and clay content in soils at different stages of restoration.

Keywords Vegetation restoration·Soil microorganisms·Environmental factors·PLFA·Degraded red soil lands

Introduction

Soil is one of the most valuable resources(Reeves 1997).It is estimated that 2.5 cm of new topsoil is formed every 100 to 1000 years,equivalent to 0.4-4.0 Mg ha-1year-1(Pimentel et al.1976).Depending on site characteristics,it can take up to 200 years to form 1 cm of soil,while it may be eroded in a few minutes by even a moderate rain storm(Verheijen et al.2009).Degraded soils cover approximately 24% of the global land area(Pimentel et al.1995;Ball et al.2007).Apart from implications to global food security and environmental quality, land degradation reduces biological productivity and affects environmental,social,and economic development(Nunes et al.2012).

The causes for degradation are both natural and humaninduced.Natural causes include water and wind erosion,organic matter depletion,salinity,acidification,loss of soil biodiversity,crusting and sealing,waterlogging and compaction;anthropogenic causes are the result of land clearing and forest degradation, inappropriate agricultural practices,improper management of industrial effluents and waste,over-grazing,poor or inappropriate forest management,surface mining,urban sprawl,and commercial/industrial development(Douglas and Charles 2015).Erosion by water is the most widespread form of soil degradation worldwide(Oldeman et al.1991).Nearly all of Europe is affected by soil erosion(Blum 1990);losses due to water erosion are high in southern Europe and becoming an everincreasing problem in northern Europe.Soil degradation due to either water or wind is inextricably linked to the loss of soil organic matter,and in many North American soils,has decreased by 30% to 40% (Brady and Weil 2008).Water erosion of soils also occurs in the southern hemisphere.Degraded ecosystems such as those in northeastern Brazil and eastern Amazonia have been extensively studied(Denich et al.2004;Nunes et al.2012).

China is one of many countries suffering from severe soil erosion(Zhao et al.2013).Erosion occurs only in arid and semiarid regions such as on the Loess plateau,but is also common in humid subtropical China.The second most typically eroded soil area after the Loess plateau is the granite red soil region.Historically,due to low levels of economic development, humans have been destroying vegetation on this soil type to maintain their livelihoods.Excessive extraction of natural resources such as overlogging,grazing and farming are widely to blame for severe soil erosion in the red soil regions of southern China over the past six decades(Wang et al.2011a,b;Xie et al.2013).

Ecological restoration based on improving soil properties and increasing vegetation cover represents a promising approach to the recovery of soil productivity and sustainability (Cooke and Johnson 2002). A chronosequence approach has been frequently used to investigate changes in vegetation and soil properties over temporal or spatial scales(Kalinina et al.2015;Boecker et al.2015).It has been used to examine the effects of time on ecological succession, soil development, and vegetation recovery following disturbance. The chronosequence design can assist in our understanding of the recovery of soil microbiota over long time periods and can provide critical information to influence successional processes for restoration(Li et al.2010;Walker et al.2010;Gasch et al.2014).

Land degradation leads to a loss of productivity that results from the loss of soil organic matter,reduces soil fertility,and decreases soil microbial biomass and activities(Lal 1996;Carpenter et al.2001;Nunes et al.2012).Soil microorganisms are an important indicator of ecological health and play crucial roles in the decomposition and mineralization of plant and animal residues present in the soil.They respond more quickly to variations in biotic and abiotic factors than to changes in soil organic matter(Marinari et al.2006;Brookes et al.2008).Differences in soil microbial biomass may lead to shifts in the important functions of soil organic matter decomposition and nutrient cycling.Studies have shown that soil microorganisms play key roles in mediating ecosystem carbon and nitrogen cycling by utilizing aboveground plant litter,as well as belowground plant roots and soil organic matter as sources of nutrients and energy(Zeller et al.2001;Hargreaves and Hofmockel 2014;Huang et al.2014).Fresh plant exudate inputs stimulate microbial activities and soil organic matter decomposition(Cheng et al.2003;Kuzyakov et al.2009).For example,the key results indicates that the rate of soil organic matter decomposition is partially controlled by the microbial community, and that microbial biomass and activity are regulated by the availability of fresh carbon sources(Fontaine et al.2011;Huang et al.2014).When soil nutrient availability is high,microbes increase the rate of fresh organic matter decomposition, whereas when microbial decomposition of soil organic matter declines,soil organic matter decreases.The exhaustion of fresh organic matter and low soil nutrient availability enhances the microbial decomposition of SOM,eventually leading to its loss(Fontaine et al.2003,2011;Huang et al.2014).As a result,evaluating the effects of land restoration on soil microbial communities is critical to better understanding their effects on soil organic matter accumulation and decomposition,as well as on ecosystem nutrient cycling.Microbial composition and biomass are also directly affected by biotic factors(e.g.,soil fauna)and abiotic factors such as pH,moisture,and the environment(Dimitriu and Grayston 2010;Wu et al.2011;Chen et al.2012;Barnard et al.2013;Huang et al.2014;Banerjee et al.2016).

Vegetation restoration is widely accepted as an effective means of dramatically increasing soil fertility(Chen et al.2002;Zhang and Xu 2005),reducing soil and water loss in eroded red soil systems(Zheng et al.2008;Huang et al.2010)and has significantly improved soil qualities and environmental conditions(Xie et al.2008;Wang et al.2011a,b;Nunes et al.2012).During the period of restoring the site’s vegetation,changes in species composition and cover can alter litter input, root architecture, and the physical,chemical and biological properties of the soil(Ren et al.2017;Zhao et al.2017).In recent years,there has been a growing interest in investigating the effects of vegetation restoration on soil biological properties(Mummey et al.2002;Frouz and Novakova 2005;An et al.2009;Xu et al.2010).This is because microbial communities are crucial to the functioning of soils,and are responsible for establishing biogeochemical cycles,for energy transfer,and in forming soil structure(Diaz et al.1993;Preston et al.2001).Therefore,it is important to assess the impacts of vegetation restoration on soil microbial community structure and biomass during the restoration process.

However,how soil microbial properties develop during vegetation restoration remains poorly understood.Reforestation of degraded soils can improve physiochemical properties but little is known about the interactions between soil microbes and soil physicochemical properties in the southern red soil area.This research is based on a case study of Changting County,Fujian Province,a typical area of erosion of red soil in southern China.Methods such as closing hillsides for afforestation,i.e.,planting trees and grasses,have been successful and widely utilized to restore degraded ecosystems over the last decade(Yue and Chen 2003;Li et al.2008;Gao et al.2011;Zhong et al.2013).However, there is little information on soil microbial community structure and biomass evaluation along a chronosequence,especially in the degraded lands in of Changting County.

In the present work, we selected five vegetation restoration years:4,7,10,13,and 35 years,and two control plots,highly degraded land(HDL)and degraded land under natural vegetation(NV)on the same soil type to compare physicochemical characteristics,soil microbial communities, microbial biomass carbon (MBC) and microbial biomass nitrogen(MBN)over the restoration period.We hypothesized that:(1)after four years,vegetation restoration will substantially increase microbial biomass and alter the microbial community structure;(2)vegetation restoration after 35 years will return to native vegetation;and(3)key environmental factors affect the soil microbial community composition and biomass.

Materials and methods

Study area

Severe water erosion over more than 100 years has occurred in Changting County, an area of 3099 km2southwest of Fujian Province.The climate is subtropical monsoon with a mean precipitation of 1730.4 mm yr-1and a mean annual temperature of 18.3°C(Zou et al.2009;Bai et al.2014).The geomorphology is low mountains,hills and terraces.Dominant tree species are Pinus massoniana Lamb.and Dicranopteris dichotoma.The most common soil type is granite red soil,classified as Argi-Udic Ferrosols(Zhang et al.2011).

The HDL plot was dominated by Pinus massoniana and Dicranopteris dichotoma,whereas the 4-,7-,10-,13-and 35-year and the NV plots were dominated by Pinus massoniana,Dicranopteris dichotoma,Schima superba.,Liquidambar formosana,Lespedeza bicolor Turcz.,Paspalum wettsteinii, and Paspalum notatum Flugge (Table 1;Fig.1).HDL and NV sites were considered to be controls,as the NV sites contained a preserved forest patch with high plant diversity as a result of minimal human disturbance, and the HDL sites was a severely damaged ecosystem with low plant diversity. No ecological restoration measures have been implemented at this location.

Field sampling

Twenty-one sampling plots were examined in August 2016.In each plot,plant cover was removed from the topsoil in order take soil samples to measure bulk density using a 100 cm3stainless steel sampler at three randomly selected points and the upper 5 cm depth of the A horizon.Soil hardness(SH)was assessed at three random points in the upper 5 cm layer with a soil hardness tester(TYD-1).Soil sampling was carried out and microbial samples were sealed in plastic bags and kept in an ice box for transport to the soil-erosion testing laboratory.One set of subsamples was air-dried by passing through 1-mm sieves for the measurement of pH, soil organic carbon (SOC), total nitrogen(TN),total and available phosphorous(TP,AP),total and available potassium(TK,AK).Field-moist samples were sieved(2-mm mesh)and stored in sealed plastic bags at 4°C for microbial analysis.All microbiological determinations were performed within two weeks of sampling.One sample was passed through 2 mm sieves and stored at 4°C to determine microbial biomass nitrogen(MBN)and microbial biomass carbon(MBC).All other samples were passed through 2 mm sieves, stored at-80°C in polythene bags after freeze drying before phospholipid fatty acid(PLFA)analysis.

Soil physicochemical analysis

Soil sampling was carried out with a 100 cm3stainless steel soil sampler and oven dried at 105°C to determine soil moisture(Bao 2008).Bulk density(BD)was determined using a bulk sampler with a 100 cm3stainless steel cutting ring,and an oven drying method.Total porosity(tp)was calculated by Eq.(1)(Ma 1994).Soil texture was determined using a conventional Robinson’s pipette and sieving technique(Pansu and Gautheyrou 2003).Soil particles were classified into three size fractions of sand(2.0-0.05 mm),silt(0.05-0.002 mm),and clay(＜0.002 mm)(Jin et al.2013).

Soil chemical analyses were performed in a soil-erosion testing laboratory(Table 2).Soil pH was determined using a 1:2.5 soil/water extract. Soil organic carbon (SOC)determination used the oil bath-K2Cr2O7titration method.Total nitrogen(TN)was analyzed using the semi-micro Kjeldahl method;total phosphorus(TP)was determined colorimetrically after wet digestion with H2SO4+HClO4,and available phosphorus (AP) was extracted in a 0.5 mol L-1NaHCO3solution(pH 8.5).Total potassium(TK)was determined by the Cornfield method and available potassium (AK) by the CH3COONH4extraction method(Liu 1996).

Table 1 Main site characteristics of native vegetation(NV);degraded land under restoration of 35,13,10,7,and 4 years and highly degraded land(HDL)

Soil microbial analysis

Soil MBC and MBN were evaluated through the chloroform fumigation-extraction method(Vance et al.1987).In a typical procedure,a 5-g dry-weight-equivalent of moist soil stored prior to measurement at 4°C was fumigated with ethanol-free chloroform for 48 h at 25°C in the dark.Both fumigated and unfumigated samples were extracted with 100 mL of 0.5 M K2SO4by shaking for 30 min at 200 rpm and filtered.Organic C and N in the K2SO4-extracted solution was determined using a Liquid TOCII analyzer.MBC and MBN were calculated by Eqs.(2)and(3):

where EC and EN are the organic C and N extracted from the fumigated soil minus the organic C and N extracted from unfumigated soil(Brookes et al.1985;Wu et al.1990).

Soil microbial community structure was assessed by PLFA analysis based on Bossio and Scow(1998).The concentration of each fatty acid was based on the carbon internal standard 19:0.The samples were analyzed using a Hewlett-Packard 6890 Gas Chromatograph equipped with an Ultra 2-methylpolysiloxane column with N2as carrier gas and H2and air to support the flame.The types of individual fatty acids were expressed in nanomoles per gram (nm g) of dry soil; peaks were identified using bacterial fatty-acid standards and MIDI peak-identification software(MIDI,Inc.,Newark,DE,USA).Total phospholipid-derived fatty acids(PLFAs),total bacterial PLFAs(B), gram-positive (G+) bacteria, gram-negative (G-)bacteria,arbuscular mycorrhizal fungi(AMF),fungal(F),actinomycete PLFAs(ACT);bacterial stress index(BS)served as indicators and references in(Table 2).

Statistical analysis

Univariate statistical analyses were performed with SPSS 17 software.One-way ANOVA determined the significance of different restoration stages on soil physicochemical properties,microbial biomass,and various types of PLFAs.Pearson correlation analysis assessed the relationships of MBC and MBN with the environmental variables.A Redundancy Analysis (RDA) was performed using CANOCO software for Windows 5.0 to test the relationship between the soil microbial communities and environmental variables. The statistical significance of the RDA was tested using the Monte Carlo permutation tests(499 permutations).Statistical significance was determined at P ＜0.05.

Results

Soil physicochemical characteristics

The results show that soil physical indices(bulk density(BD),soil water content(SWC),total porosity(tp),soil hardness(SH),and soil particle composition)gradually improved with increasing restoration years.BD was significantly lowest in the NV plots compared with other restoration-implemented sites at this stage of soil development(Table 3).The highest levels of SWC were in the NV sites,and the lowest in the HDL sites.SWC ranged from 17 to 36% in all other sites and increased over time.Total porosity(TP)was 50% in the HDL plot,lowest in the 4-year plots and did not change after 4 years of restoration.The tp improved after 7,10,13,and 35 years.The highest improvement in tp was 78% in the NV plots.Soil hardness(SH) decreased significantly over the restoration years(Table 3).Sand content ranged from 61.7 to 72.9% ,silt content from 10.1 to 16.2% and clay content from 10.7 to 24.1% (Table 3).

Fig.1 Study areas in Changting County of Fujian Province and sample sites

The results show that nutrient accumulation increased with the number of restoration years.The highest concentrations of SOC,TN,TP,AP,TK,and AK were found in the NV plot,and decreased from the 35 year plot to the HDL plot.Soil pH was the highest in the HDL and decreased as the time of restoration increased(Table 3).

Table 2 Fatty acid markers of soil microbial population

Table 3 Soil physicochemical top soil(0-5 cm)properties in different restoration years

Soil microbial community composition and biomass

MBC, MBN, and total PLFA levels ranged from 11.0-112.5 mg kg-1, 6.5-15.5 mg kg-1, and 0.7-70.0 nmol g-1, respectively. Statistically significant differences in MBC,MBN,and total PLFAs were detected in all the plot soils except for MBN among the 10,7,4 year and HDL plots(Figs.2,3a).The quantities of PLFAs in the restoration stages decreased in the order of bacterial PLFAs ＞G+bacterial PLFAs ＞G-bacteria PLFAs ＞actinomycete PLFAs ＞fungal PLFAs ＞AMF PLFAs (Fig.3b-g). The abundance of AMF PLFAs differed unsignificantly among restoration stages for the 13-,10-,7-,4-year sites and HDL sites,but differed significantly among these years 35 year and NV sites.No significant differences were detected in G+bacterial PLFA in the 13-and 10-year soils,and between 7-and 4-year soils for G-bacterial PLFAs.The G+/G-ratio was significantly higher in HDL soils than in all the other plots.No significant differences were found between 10-and 7-year soils,as well as between 35-and 4-year soils(Fig.3h).The F/B(Fungal/Bacterial)ratio in 13-year soils was significantly higher than that in other restoration-stage soils,but there were no statistically significant differences between13-and 10-year soils or among 35-,7-,4-year,and HDL soils.The F/B ratios were 0.17,0.15,0.25,0.37,0.38,0.18,and 0.09 in the HDL,4-,7-,10-,13-,35-year,and NV soils,respectively(Fig.3i).The bacterial stress indices were lowest in the NV sites and highest in the HDL sites(Fig.3j).

Fig.2 Changes in MBC and MBN:native vegetation(NV);degraded sites under restoration for 35,13,10,7,4 years and highly degraded land(HDL).Different letters over the bars represent statistically significant differences(P ＜0.05)in the different restoration years

Relationships between microbial community composition and environmental factors

Fourteen environmental factors selected in our study,only six(TN,AK,SOC,SWC,silt content,SH)present in the ordination explained 98.4% of the total variability of the PLFAs(Fig.4).The variations of the PLFAs were strongly correlated with TN(F=15,P=0.002),AK(F=13.5,P=0.002), SOC (F=157, P=0.002), silt (F=12,P=0.002),SWC(F=6.3,P=0.002)and SH(F=15.8,P=0.002).The first ordination RDA axis(axis 1,horizontal),which was highly correlated with SOC,TN,AK,SWC, explained 95.6% of the total variations of the PLFAs.The second ordination RDA axis(axis 2,vertical)was mainly related to soil hardness(SH)and silt content,and explained 2.8% of the total variations of the PLFAs(Fig.4).Total nitrogen(TN),SOC,AK and soil water content(SWC)were strongly positively correlated with microbial community structure and biomass,whereas SH and silt content were negatively related.Pearson correlation analysis revealed that MBC and MBN were significantly positively correlated with SOC,TN,TK,TP,AK,AP and clay content,and negatively correlated with SH,BD,pH,silt and sand content(Table 4).

Discussion

Vegetation restoration effects on soil physicochemical properties

Longer years for recovery resulted in higher SWC,BD,TP,SH and soil particle composition were improved. A possible explanation is that,with the extension of recovery time,the ability of vegetation to intercept precipitation reduced the loss of clay and silt particles.Clay particles in Changting County are prone to erosion by water.Another possible explanation is that a variety of soil macrofauna(e.g.,varieties of earthworms and ants)are helpful to improve soil water content(SWC),bulk density(BD),total porosity(tp),soil hardness(SH)(Buse 1990;H?lldobler and Wilson 1990;Jordan et al.1999;Ponder and Tadros 2002;Edwards 2004;Cerdàand Jurgensen 2008;Cerdàet al.2009;Benckiser 2010;Blouin et al.2013).A further explanation is that these plants have deep root systems which can exploit cracks,voids and large pores,as well as enlarge small pores,all of which have a positive influence on root biomass(Rutigliano et al.2004;Jiao et al.2008).The relatively evenly distributed roots throughout the surface and litter layers are conducive to improving soil physical properties.There are more large pores and the soil quality is improved as the roots grow(Bai et al.2016;Zhao et al.2017;Zhang et al.2017).

Soil pH decreased with longer recovery years in agreement with previous vegetation recovery studies(Berthrong et al.2009;Kalinina et al.2015).The reasons were,first,changes occurred in root and ectomycorrhizal exudate quality,which often contain high amounts and varieties of organic acids(Grayston et al.1997).Secondly,the uptake of cations by tree roots increased(Jobbágy and Jackson 2003). Thirdly, possibly because of the organic acids released by the roots,soil microorganisms and exchangeable cations were lost to plant uptake(Alfredsson et al.1998).A final reason for the decrease in pH is the accumulation of soil organic matter.

Soil nutrients accumulated with longer restoration years(Deng et al.2015;Deng and Shangguan 2017),which agree with our findings. Vegetation changes in soil physicochemical properties occurred over time and indirectly affected the amount of runoff and soil loss(Janssens et al.1998;Pohl et al.2009).The duration of vegetation cover influenced soil chemical characteristics mainly through the quantity and chemical components of plantfoliage,litter,and roots(Sourkova et al.2005;Ostertag et al.2008;Marin-Spiotta et al.2009).Litter decomposition in different years is regulated by the interaction between decomposers and substrate quality,which were both directly affected by abiotic and biotic factors.Different vegetation restoration years have different vertical vegetation structures(aboveground cover,surface litter layer,and underground roots),leading to differences in rainfall interception and thereby reducing the occurrence of splash erosion and runoff accumulation,and increasing soil nutrients(Bochet et al.1998;Ross et al.1999;Calder 2001). Moreover, the years since restoration inducedchanges in litter quantity and quality.We believed that stable soil temperature and humidity were factors controlling litter decomposition rates.Salinas et al.(2011)found that soil temperatures explained 95% of the variation in the rate of decomposition.Suitable temperatures and humidity can accelerate litter and roots (coarse and fine roots)decomposition by enhancing litter microbial activity.At small scales(20 m×20 m),micro-meteorological conditions incorporate both soil temperature and soil humidity,which is particularly important for relatively infertile sites such as the HDL and 4 year plots(soil temperature and soil humidity in an unstable micro-meteorological environment).In general,legumes such as Lespedeza bicolor Turcz.have low nutrient-use efficiency and strong effects on N availability and supply in both natural and agricultural systems(Fornara and Tilman 2008).All plots were planted with L.bicolor.Soil organic carbon,and total nitrogen,total and available phosphorous,and available potassium levels were higher in sites where evergreen broad-leaved species (L. bicolor and Schima superba Gardn.et Champ.)were planted than in the HDL sites.

Fig.3 Total PLFAs(a),bacteria PLFAs(b),Fungal PLFAs(c),Actinobacteria PLFAs(d),Arbuscular mycorrhizal fungi PLFAs(e),Gram-positive bacterial PLFAs(f);Gram-negative bacterial PLFAs(g),Gram-positive/Gram-negative ratio(h),Fungal/Bacterial ratio(i),Bacterial stress index(j)in the different sites.Different lower letters over the bars explain statistically significant differences(P ＜0.05)among the different restoration years.PLFAs=phospholipid fatty acids

Fig.3 continued

Fig.4 RDA results of the relationship between PLFAs and soil physical-chemical factors.which are expressed by the different color arrows;solid blue arrows point to the PLFAs index,consisting of T,F,G+,G-,AMF,and A.Hollow red arrows represent soil physicalchemical factors,consisting of SH,silt content(silt),SOC,TK,AK,and SWC.Filled green circles represent HDL,NV and different restoration years.Notes:total PLFAs(T),fungal PLFAs(F),Grampositive bacterial PLFAs(G+),Gram-negative bacterial PLFAs(G-),arbuscular mycorrhizal fungal PLFAs (AMF), and actinomycete PLFAs(A)

Vegetation restoration effects on soil microbial community composition and biomass

Our results indicate that microbial community composition and biomass increased with time since restoration,even after 4 years.Our first hypothesis was supported and shows that vegetation restoration after 35 years on a degraded red soil will not recover to the status of native vegetation,which did not support our second hypothesis.

Soil microbial composition and biomass carbon and nitrogen were significantly higher in all plots compared with the HDL plot.High levels of microbial community composition and biomass were found in the soil under native vegetation,the NV plot.This may be attributed to the relatively high quantity and quality of litter which supported greater microbial activity due to the availability of above and belowground carbon(Nsabimana et al.2004).These findings may be due to higher soil organic carbon,total nitrogen,total and available potassium,and total and available phosphorous.Higher nutrient availability combined with higher organic matter inputs favored bacterial growth.This is consistent with our finding that time since reclamation enhanced soil organic C and total N accumulation(H?gberg et al.2003;Nakamura et al.2003;Yuan et al.2015).Carbon and nitrogen,as the major source of energy and cell materials for microorganisms,are closely related to their maintenance and growth as reported by Nsabimana et al.(2004),Wang et al.(2005)and Araujo et al.(2010).Soil carbon and nutrient availability,as determined by the quantity and quality of substrates by litter inputs and root exudations,were crucial factors that affected microbial biomass (Wardle 1998; Wang and Wang 2007;Yang et al.2010).Previous studies have shown that fungi are more efficient decomposers of recalcitrant plant compounds such as lignin,cellulose,and hemicellulose and produce a broader range of extracellular enzymes than do bacteria,thereby enhancing recalcitrant litter decomposition and soil organic matter formation(Cairney and Meharg 2002;Cusack et al.2011).Nitrogenfixing plants such as Lespedeza bicolor are widely planted to restore degraded ecosystems because they can rapidly improve soil nutrient levels and microbial activity(Nichols et al.2001;Sheoran et al.2010).Arbuscular mycorrhizal fungi form a key functional interface between roots and soils to promote growth by facilitating nutrient uptake and improving soil resistance to erosion(Brearley et al.2016;Umardhiah et al.2016),which benefits microbial accumulation.Another reason might be because micro-meteorological conditions under native vegetation are relatively stable,as the dense canopy buffers weather fluctuations,and microorganism growth may have adapted to the soil environment.Consistent soil temperatures and humidity,available nutrients for microbial colonization, and the composition and activity of decomposer communities,may contribute to faster decomposition of litter and fine roots and benefit changes in microbial community structure and biomass.

Relation between vegetation restoration and environmental and microbial parameters

Key environmental factors affected microbial community composition and biomass over the years since restoration,which supports our third hypothesis.

Redundancy analysis and Pearson correlation analysis of the relationships between soil properties and microbial communities in different restoration year soils indicate that total nitrogen,organic carbon,available potassium,soil water levels,silt content and soil hardness were the key factors affecting microbial community composition(Fig.4). Considering that soil properties were closely related to litter characteristics,soil properties influenced the microbial community structure(Bardgett and Shine 1999;Myers et al.2001;B??th and Anderson 2003).Vegetation restoration can improve soil quality and affect the soil ecosystem,especially microorganisms,by influencing moisture levels,temperature,pH,carbon content(Griffiths et al.1998),and total nitrogen(Paul and Clark 1989;Craine et al.2007).Soil nutrients can promote changes of microbial community structure and microbial biomass accumulation.On the one hand,total nitrogen(including ammonium nitrogen and nitrate nitrogen),dissolved organic carbon and available potassium were easily dissolved in water and carried away by surface runoff.On the other hand,these are also lost by dissolution and leaching. The loss of these nutrients limits microbial community and biomass.

In the highly degraded(HDL)sites,soil hardness was highest but decreased with root penetration with longer recovery years.Soil water content was the lowest in the HDL site due an absence of vegetation and was lost by intense evaporation.However,with the extension of vegetation recovery time,on degraded sites with native vegetation (NV) and on sites with long recovery times(35 years), soil water levels were relatively enhanced because of the interception of rainfall by ground vegetation and surface runoff infiltration.Higher soil moisture conditions and lower soil hardness are conducive to microbial growth and decomposition.

Soil texture is known to affect carbon accumulation and soil microbial properties(Buscot and Varma 2005).The composition of fine particles(clay and silt)is the most important index of soil structure.In this study,the silt content was negatively correlated with microbial structure and biomass.The reasons are,first,for highly degraded land in China’s Changting County or sites under 4 years of restoration,once soil erosion occurs,silt and clay particles are carried away by surface runoff.High litter cover occurs with time since restoration and reduce sediment transport,especially for clay-sized particles(Shi et al.2013).Secondly,the most common soil type of degraded lands is typical red granite,and microbial community structure and biomass may be more suitable in a clay habitat.Silt and sand are apparently not suitable but this should be verified in future research.Thirdly,micro-meteorological conditions under degraded native vegetation or on 35 year restoration sites are relatively stable,as the canopy buffers weather fluctuations.In particular,the loss of clay was reduced and microorganisms may have adapted to the relatively stable physical and chemical environment.

Conclusion

This study demonstrated that the longer the time since restoration was initiated,soil microbial biomass increased and altered microbial community structure.This is apparent even after only 4 years of recovery, although the recovery was relatively slow.Following 35 years of vegetation restoration,microbial communities and biomass still did not fully recover to their status under native vegetation.Variations in microbial community structure and biomass mostly depend on soil organic carbon, total nitrogen, available potassium, silt content, soil water levels,and soil hardness.This study provided new insights into microbial community structure and microbial biomass that influence soil organic carbon,nitrogen,potassium,and clay contents in soils at different stages of restoration.These findings can serve as a reference for future studies on the restoration of vegetation and reconstruction of eroded soils in the typical red granite soils of southern China.

AcknowledgementsWe thank our co-workers for their assistance with the fieldwork,with PLFA extractions and with manuscript writing,and all members of the Fujian Normal University.