The impact of landslides on chemical and microbial properties of soil in a temperate secondary forest ecosystem

Yakun Zhang · Chen Miao · Jiaojun Zhu ·Tian Gao · Yirong Sun · Jinxin Zhang ·Shuang Xu · Kai Yang

Abstract In forest ecosystems, landslides are one of the most common natural disturbances, altering the physical,chemical and microbial characteristics of soil and thus further altering ecosystem properties and processes. Although secondary forests comprise more than 50% of global forests, the inf luence of landslides on the soil properties in these forests is underappreciated. Therefore, this study investigates the inf luence of landslides on the chemical and microbial nature of the soil. Study of these modif ications is critical, as it provides baseline evidence for subsequent forest revegetation. We selected four independent landslides and adjacent secondary forest stands as references in a temperate secondary forest in northeastern China. Soils were obtained from each stand at 0-10 cm and 10-20 cm depths to determine chemical and microbial properties. Soil total carbon (TC), total nitrogen (TN), nitrate (NO 3 - -N), available phosphorus (P), microbial biomass carbon (MBC), microbial biomass nitrogen (MBN), microbial biomass phosphorus(MBP) and phenol oxidase, exoglucanase, β-glucosidase,N-acetyl-β-glucosaminidase, L-asparaginase and acid phosphatase activities were 29.3-70.1% lower at the 0-10 cm soil depth in the landslide sites than at the secondary forest sites, whereas total phosphorus (TP) and ammonium(NH 4+ -N) were unaffected by the landslides. N-related enzymes, N-acetyl-β-glucosaminidase and L-asparaginase were reduced by more than 65% in the landslide sites, consistent with the decrease in nitrate concentration at the same 0-10 cm depth. At a depth of 10-20 cm, the variations in the soil properties were consistent with those at the 0-10 cm depth. The results demonstrated that soil chemical and microbial properties were signif icantly disrupted after the landslides, even though the landslides had occurred 6 years earlier. A long time is thus needed to restore the original C and nutrient levels. In temperate secondary forests, soil TC and TN contents were found to be more suitable for estimating the state of soil restoration than soil TP content.

Keywords Secondary forest · Landslide · Soil nutrients ·Microbial biomass · Enzyme activity

Introduction

Secondary forests are distributed worldwide, especially in temperate regions. For instance, secondary forests occupy 70% of the forest coverage in Northeast China (Zhu and Liu 2007). The functions of secondary forests involve in controlling soil erosion, conserving water sources, regulating the climate, and purifying the air (Yang et al. 2010b, 2012a, b).However, natural disturbances often occur in secondary forests. Such disturbances include landslides (Li et al. 2017a),wind and snow (Li et al. 2017b), insects (Taki et al. 2013),and f ires (Cohen et al. 2016). Landslides, whereby rocky masses separate from the ground and slide along a slope,often induced by precipitation and earthquakes (Walker et al.1996), are especially critical disturbances (Singh et al. 2001)because the signif icant soil erosion and sediment deposition changes the soil properties and cycling of ecosystem nutrients (Shiels et al. 2005; B?ońska et al. 2017).

Previous studies have focused on spatiotemporal variations in vegetation cover after landslides and on the role of tree species composition in landslide stabilization (Shiels et al. 2005; Gonzalez-Ollauri and Mickovski 2017; Balzano et al. 2019), but the inf luence of landslides on soil properties in forests is gaining more attention (Cheng et al. 2016;Schomakers et al. 2019; Yang et al. 2020). The most obvious alterations of soil properties resulting from landslides are decreases in carbon (C) and nitrogen (N), which cannot be recovered in a short time (Wilcke et al. 2003; Cheng et al.2016). Specif ically, Wilcke et al. ( 2003) observed that in the montane rainforest of Ecuador, the total carbon (TC) and total nitrogen (TN) of landslides soils were 72% and 83%lower, respectively, in comparison to those of undisturbed soil. Cheng et al. ( 2016) also found that in the Taiwanese valley, the contents of TC and TN after landslides were 91% and 97% lower, respectively, than those of unaff ected soils. Although many studies (Reddy and Singh 1993; Shiels et al. 2005; Begum et al. 2018) have shown that TC and TN decrease after landslides, the impacts of landslides on soil pH and other nutrients are still unclear. Cheng et al.( 2016) found that soil pH in landslides was higher than that in adjacent forests, whereas Van Eynde et al. ( 2017)observed that soil pH did not diff er signif icantly between the landslides and their adjacent stands. Cheng et al. ( 2016) also reported no marked gaps in exchangeable potassium (K),calcium (Ca), magnesium (Mg), and available phosphorus(P) between the landslides and the undisturbed areas. In contrast, Masyagina et al. ( 2019) reported that the availability of elements (N, P, K, Ca) in landslide soils was lower than in undisturbed soils. The diff erences among the above studies might be due to the spatial heterogeneity of the studied sites(Walker et al. 2013; Van Eynde et al. 2017).

Soil microbes involved in organic matter transformation and nutritional recycling are essential for maintaining the structure and composition of ecosystems (Zhong et al.2009). Soil microbial biomass and enzymatic activity are very sensitive to environmental changes. Therefore, they can be used to evaluate soil quality after perturbation (Xu et al. 2006; Yang et al. 2010b, 2012a, b). Singh et al. ( 2001)observed that in tropical forests, the soil microbial biomass carbon (MBC) of 1-, 15- and 58-year-old landslides was, respectively, 83.0%, 47.6% and 18.0% lower than in adjacent undisturbed forest soils. However, B?ońska et al.( 2016) found that 15 years after a landslide, the soil MBC of landslides was 7.5% higher than that of adjacent soils.Masyagina et al. ( 2019) also demonstrated that at a depth of 0 - 5 cm, soil MBC in 6- and 35-year-old landslides did not diff er from that in adjacent forest soils. The discrepancy in the above outcomes may be associated with the chronological age of the landslides and the regeneration of vegetation after the landslides. However, there are fewer studies on microbial biomass nitrogen (MBN) and microbial biomass phosphorus (MBP) than MBC. Singh et al. ( 2001) revealed that compared with those of adjacent tropical forest soils,the MBN and MBP of a 1-year-old landslide soil were 78%and 81% lower, respectively.

Soil enzymes are important in the conversion of soil nutrients, and their activity is related to soil chemistry and soil microbial biomass (Cai et al. 2018). Soil enzymes are divided into two categories, oxidases and hydrolases (Burns and Dick 2002). Oxidases mediate key ecosystem functions,lignin degradation and C decomposition (Sinsabaugh 2010).Hydrolases include proteases, ureases, cellulases, amylases,and phosphatases, which mainly participate in the hydrolysis of proteins, cellulose, and chitin (Burns and Dick 2002;Sinsabaugh et al. 2008). In landslide studies, researchers have considered the variations in enzymes in tropical climate zones (Shiels et al. 2005; Schomakers et al. 2019) but not in temperate climate zones. For example, B?ońska et al.( 2016) argued that MBC, MBN and the activity of dehydrogenase can be considered the main factors for evaluating the soil changes after landslides. Therefore, it is necessary to determine whether microbial properties are aff ected by landslides in temperate secondary forest ecosystems. Furthermore, it is essential to explore which microbial factors indicate soil property changes after landslides in secondary forest ecosystems.

Here, we assessed the chemical and microbial properties of landslide soil to those in adjacent secondary forest soil to determine the extent of landslide damage to secondary forest soils and provide basic data for future study of vegetation regeneration in secondary forest ecosystems. On the basis of the studies of Singh et al. ( 2001) and B?ońska et al. ( 2016),we assumed that TC, TN, total phosphorus (TP), available nutrients, MBC, MBN, MBP, and enzyme activities in landslides are lower than those in adjacent secondary forests and that the degrees of variation in the available nutrients and microbial properties are higher than those of the chemical properties.

Materials and methods

Study site

This study was undertaken in a typical secondary forest environment in the Qingyuan Forest CERN, a Chinese Ecosystem Research Network site in Liaoning Province overseen by the Chinese Academy of Sciences (41°51′ N, 124°54′ E,500-1100 m above sea level). The location has a monsoonal continental climate wherein the annual average temperature is 4.5 °C. The annual rainfall is between 700 and 850 mm,with 80% during the summer. The length of the vegetative season is 6 months (from April to September). An area with loamy clay soil was chosen for the sample plots (Yang et al.2012a, b).

The vegetation at the study site belongs to the Changbai f lora, and the primary forest is a broad-leaved Korean pine (Pinus koraiensis) forest. Due to long-term human and natural disturbance, almost all the primary forests have disappeared and large areas of secondary forests have formed in their place (Zhu et al. 2007). We selected four pairs of landslides and their adjacent secondary forests as the research stands. The secondary forests are dominated by tree species such asSophora japonica,Acer mono,Acer trif lorum,Quercus mongolica,Acer pseudosieboldianum,Juglans mandshurica,Fraxinus rhynchophylla, andFraxinus mandshurica.Actinidia arguta,Corylus mandshurica,Sorbaria sorbifolia, andLonicera japonicawere distributed in the shrub layer, andMeehania urticifolia,Urtica laetevirens,Artemisia argyi, andScopolia acutangulardominate the herb layer. The vegetation coverage of the landslides is 75%, and the landslide sites are mainly composed ofBetula,Salix pierotii,Populus, andPhellodendron amurense. The shrub layer is dominated byRubus idaeus,Schisandra chinensis,Aralia elata, andRibes nigrum, and the herb layer is dominated byOenothera biennis,Galium pseudoasprellum,Carex siderosticta, andViola collina.

Selection of landslides and soil sampling

During a f ield survey in August 2019, we selected four landslide stands that had formed during rainstorm in August 2013. The length of the landslides was approximately 300 m and the widest point of each was approximately 120 m.Adjacent secondary forests were selected as reference sites.To eliminate edge eff ects, a f ixed 20 m × 20 m plot was selected in the middle of each secondary forest and landslide stand, and a total of eight plots were established for litter and soil sampling (Fig. 1). Within each plot, we collected litter samples at nine individual 10 m × 10 cm sites, recorded the thickness of litter, mixed them to form a composite litter sample, and then brought the samples back to the laboratory and weighed them after oven drying at 60 °C. The average thickness of the litter layer on the landslides was 1.3 cm and 3.7 cm in the secondary forests. Soil samples were collected at the same nine individual sites in each plot and divided into 0-10 cm and 10-20 cm depths. Then, soil samples from the same soil depths were mixed to form a composite soil sample for each plot. Soil samples were brought back to the laboratory and sieved (2 mm) after removal of rocks and roots. Then, the samples were divided into three parts: (1)oven dried and sifted through a 0.15 mm sieve to analyze TC, TN, total P (TP), δ 13 C and δ 15 N values in the soil; (2)air dried for analysis of available P and pH; (3) stored at 4 °C to measure microbial biomass C, N and P and enzymatic activities and inorganic N (NH4+-N and NO3--N).

Litter and soil chemical properties

Fig. 1 Landslides and adjacent natural secondary forests selected in the Qingyuan Forest Chinese Ecosystem Research Network of the Chinese Academy of Sciences. Landslide sites are marked with red dots,secondary forest sites with green dots; study plots are labelled 1-4. Sampling sites in the landslide plots are shown on the right

The carbon and nitrogen contents of litter and the TC and TN contents of soil were measured using a Vario EL III elemental analyzer (Elementar Analysensysteme GmbH,Hanau, Germany). An isotope ratio mass spectrometer(IsoPrime 100 Isotope Ratio Mass Spectrometer, Germany)was used to measure the natural abundances of 13 C and 15 N.The litter P and soil TP concentrations were measured using the molybdenum-antimony colorimetric process after digestion with H2SO4-HClO4(Olsen and Sommers 1982). The stocks of litter C, N and P were calculated as follows. The mass of the litter at the 9 sites were summed and divided by the sampling area to obtain the amount of litter in each plot per square meter and then multiplied by the content of C, N and P of litter to get the stocks of litter C, N and P. Inorganic nitrogen (NH4+-N and NO3--N) was extracted with 2 M KCl (1:5 soil-liquid) and quantif ied using Auto-Analyzer III (Bran + Luebbe GmbH, Germany). Available P was measured using a molybdenum-antimony colorimetric system with 0.5 M NaHCO 3 (pH 8.5, 1:20 soil-liquid).Soil pH was measured using a glass electrode process (1:2.5 soil-water). The bulk density of the soil was ascertained by the cutting-ring method (Wang et al. 2011).

Soil microbial properties

The concentrations of MBC and MBN in the soil samples were determined by chloroform fumigation extraction(Vance et al. 1987). In an evacuated extractor, soil samples were fumigated for 24 h with ethanol-free chloroform.Fumigated and unfumigated soils were extracted with 0.5 M K2SO4(1: 4 soil to extractant) and analyzed using a TOC analyzer.

Soil MBC was calculated as MBC =EC/kC.whereECis the discrepancy between extracted organic C in fumigated soil and unfumigated soil andkC= 0.45 (Wu et al. 1990).

Soil MBN was computed as MBN =EN/kN.whereENis the discrepancy between extracted N in fumigated soil and unfumigated soil andkC= 0.54 (Brookes et al. 1985).

The molybdenum-antimony colorimetric procedure was used to calculate MBP (Brookes et al. 1982). The method of extraction of MBP in unfumigated and fumigated soil samples was the same as that of available P. Inorganic phosphate (25 μg mL -1 KH2PO4) was added to the soil sample to calculate the recovery rate (R%).

Soil MBP was computed as MBP =EP/(R% × 0.40).whereEPis the diff erence between extracted P in fumigated soil and unfumigated soil, R% is the P recovery coeffi cient, and 0.40 is the correction coeffi cient (Brookes et al. 1982).

The activity of phenol oxidase was determined by using L-3,4-dihydroxyphenylalanine as a substrate (Saiya-Cork et al. 2002). Three milliliters of 50 mM sodium acetate buff er and 4 mL of the substrate solution were added to 2 g of fresh soil and shaken at 37 °C for 40 min. After f iltration of the reaction products, absorbance of the f iltrates was measured at 460 nm (UV1800 Ultraviolet and Visible Spectrophotometer, JINGHUA Instruments, China).

The activities of exoglucanase, β-glucosidase and N-acetyl-β-glucosaminidase were measured using 4-nitrophenyl-β-D-cellobioside (Turner et al. 2002),4-nitrophenyl-β-D-glucopyranoside (Parham and Deng 2000) and 4-nitrophenyl-N-acetyl-β-D-glucosaminide (Von Mersi and Schinner 1996) as the respective substrates. Three replicates of 0.3 g of fresh soil were incubated with 1.2 mL of 50 mM sodium acetate buff er and 0.3 mL of substrate solution for, respectively, 2 h, 1 h, and 1 h at 37 °C. After incubation, 0.3 mL of 0.5 M CaCl2and 1.2 mL of 0.5 M NaOH were added to stop the reaction and develop color.The reaction products were centrifuged, then analyzed colorimetrically at 410 nm.

L-Asparaginase activity was determined as described by Frankenberger and Tabatabai ( 1991). Nine milliliters of Tris buff er, 0.2 mL of methylbenzene, and 1 mL of substrate solution were added to 5 g of fresh soil and incubated at 37 °C for 2 h. Then, 35 mL of KCl-Ag2SO4solution was added to terminate the reaction, and 20 mL of the sample solution was combined with 0.2 g of MgO for distillation.Five milliliters boric acid was combined with the indicator solution and used to absorb the distillate, which was then titrated with 5 mM H2SO4to the end stage.

Acid phosphatase activities were measured using p-nitrophenyl phosphoric acid hexahydrate as the substrate (Tabatabai 1994). Fresh soil (0.5 g) was inf iltrated with 2 mL of universal buff er (pH 6.5), 0.1 mL of methylbenzene and 0.5 mL of the substrate solution for 1 h at 37 °C. The reaction was then terminated with the addition of 0.5 mL 0.5 M CaCl2and 2 mL 0.5 M NaOH. The reaction products were diluted and colorimetrically tested at 420 nm.

Enzymatic activity is expressed as μmol g -1 h -1 in each case.

Statistical analyses

F tests and two-way analysis of variance (ANOVA) were applied to evaluate the signif icance of diff erences in litter chemistry and soil properties between forest type and soil depth, respectively. If main eff ects were signif icant, pairwise post-hoc comparisons of subgroup means were made using the Bonferroni test. If the interactions were signif icant, differences in the eff ects of forest type and soil depth were analyzed using one-way ANOVA. Statistical analyses were performed in SPSS 25.0 for Windows (IBM, Armonk, NY,USA). All diff erences in soil properties were considered signif icant atP＜ 0.05.

Results

Chemical properties of the litter layer in secondary forests and landslides

The litter layer of the secondary forest had signif icantly more N (P＜ 0.05) than that in the adjacent landslides (Table 1).However, no signif icant diff erences in C and P concentrations or C:N ratios were observed between secondary forestsand landslides. In comparison to those of adjacent secondary forests, the litter C, N, and P stocks declined by 47.4%,53.1%, and 45.0% in the landslides, respectively.

Table 1 Litter C, N, and P concentrations and stocks(mean ± SE, n = 4) in the landslides and adjacent secondary forest stands

Soil chemical properties in the secondary forests and landslides

The TC concentration between the secondary forest and the landslides diff ered signif icantly and tended to decline with soil depth (P＜ 0.05) (Table 2). At depths of 0-10 cm and 10-20 cm, the TC was 62.7% and 54.6% lower in soils collected from landslides compared to secondary forest soils from the same depths. In comparison to those in adjacent secondary forests, the concentrations of soil TN were also lower in the landslides (P＜ 0.05), but diff erent soil depths had no signif icant diff erences in TN concentration. The mean values of TN in the landslides were 57.1% and 50.8%lower than those in the secondary forest samples at depths of 0-10 cm and 10-20 cm, respectively. However, no discernible variation in soil TP was found between the secondary forest and landslide samples. The C:N ratio was stable with soil depth in the landslides and signif icantly lower than the C:N ratios in the adjacent secondary forest soils (P＜ 0.05).The soil pH was markedly greater in the landslides than in the secondary forest sites, but there was no discrepancy in pH between the two soil depths.

Similar to the soil C and N contents, the C and N stock reductions were signif icant between secondary forests and landslides (P＜ 0.05), but there was no signif icant diff erence between the 0-10 cm and 10-20 cm depths (Table 2). Specif ically, the amounts of C stocked in the landslides were 58.0% lower at the 0-10 cm depth and 45.8% lower at the 10-20 cm depth compared to those in the adjacent secondary forest soils, and the N stocks were also lower 49.5% and 42.6% at the two soil depths, respectively. We found that the P stocks showed the same pattern as the P concentrations.The NO3--N concentration was significantly different between the landslides and the secondary forests and decreased with soil depth (P＜ 0.05), while there was no variation in the concentrations of NH4+-N in either of the two stands studied (Fig. 2, Table 3). Compared with the secondary forests, the NO3--N concentrations at the 0-10 cm and 10-20 cm depths in the landslides were 76.3% and 68.5%lower, respectively. Unlike the soil TP, the available P concentrations at the 0-10 cm and 10-20 cm depths in the landslides were 68.8% and 34.6% lower than those at the same depths in the adjacent secondary forest soils.

Soil microbial properties in the secondary forests and the landslides

The soil isotopic composition of 13 C (δ 13 C) also diff ered signif icantly between the landslides and the secondary forests(P＜ 0.05) (Fig. 3). Higher values of δ 13 C were found in the landslide soils at 0-10 cm and 10-20 cm depths compared to those of the secondary forests. Nonetheless, there were no signif icant variations in the abundance of 15 N (δ 15 N)between landslides and secondary forests at either soil depth.

The results showed that landslides in the secondary forests reduced the microbial biomass C, N and P (Fig. 4,Table 3). The MBC, MBN, and MBP concentrations in the landslides were 62.4%, 46.4%, and 58.2%, respectively,lower than those in the secondary forests at the 0-10 cm soil depth. At the 10-20 cm soil depth, the MBC, MBNand MBP in the landslides were 56.7%, 49.5%, and 55.7%,respectively, lower than those in the secondary forests. No significant variations were observed between the landslide and secondary forest soils regarding the MBC:TC and MBN:TN ratios, while the MBP:TP ratio was signif icantly greater in the secondary forests than in the landslides(P＜ 0.05).

Table 2 Selected soil chemical properties and C, N and P stocks (mean ± SE, n = 4) at diff erent soil depths in the landslides and adjacent secondary forest stands

Fig. 2 Concentrations of NH 4 + -N ( a), NO 3 - -N ( b) and available P( c) at each soil depth in the secondary forests and landslides. Error bars indicate standard errors of the means ( n = 4). *Signif icant differences between landslides and secondary forests in the same depth( P ＜ 0.05)

Overall, all the soil enzymatic activities in the samples were signif icantly lower in the landslides than in the neighboring secondary forests at both 0-10 cm and 10-20 cm soil depths, and they decreased with soil depth (P＜ 0.05) (Fig. 5,Table 3). The activities of phenol oxidase, exoglucanase,β-glucosidase, N-acetyl-β-glucosaminidase, L-asparaginase,and acid phosphatase were 46.0%, 29.3%, 70.1%, 67.6%,70.0%, and 48.8%, respectively, lower in the landslides relative to those in the secondary forests at the 0-10 cm soil depth.

Discussion

Our analyses lead to two signif icant conclusions. First, the studied landslides reduced soil TC and TN levels, but not TP levels in their temperate secondary forest locations. Second, the landslides also led to decreases in soil MBC, MBN,MBP and enzymatic activity. The decreases in N-related enzymes such as N-acetyl-β-glucosaminidase (NAG) and L-asparaginase (LAP) coincided with decreases in NO3--N concentration.

Eff ects of landslides on soil chemical properties in the temperate secondary forest ecosystem

TC, TN and TP are important indexes for evaluating soil quality, and their reduction can indicate the degree of soil destruction (B?ońska et al. 2016). By comparing landslide and secondary forest soil samples, our results conf irmed that TC and TN concentrations after landslides were only 30%-50% of those of the adjacent secondary forests. Similar sharp decreases in TC and TN have been reported in temperate and tropical forests (Singh et al. 2001; B?ońska et al.2017; Schomakers et al. 2019). For example, Singh et al.( 2001) pointed out that in the forest ecosystem of the Nepal Himalayas, TC and TN concentrations in landslide soils were 24.1% and 25.8% of those in mature forests, respectively. B?ońska et al. ( 2017) also reported that in southern Poland, TC and TN concentrations in landslide soils were 67.9% and 44.4% of those in the adjacent landslide stands,respectively. There may be two main reasons that landslides remove most of the soil TC and TN in forest ecosystems.First, an extreme loss of surface soil occurs during the landslide disturbance, and second, the equilibrium among the input and output of C and nutrients is altered (Tian et al.2009). In detail, litter input and microbial decomposition are two major sources of organic matter in the soil of forest ecosystems (Crow et al. 2009; Cotrufo et al. 2015). Landslides change the composition of vegetation and reduce the amount of litter that enters the soil, resulting in a decrease in TC and TN (Chen et al. 2010). However, in contrast to the reductions in soil TC and TN, the soil TP in the studied landslides was similar to that in the adjacent secondary forests. Two possible reasons could explain the unchanged soil TP between the secondary forests and the landslides.One reason is the poor mobilization of P in soil. Soil TP is stable and not easily lost relative to C and N concentrations(Frizano et al. 2002; Yang et al. 2010a). The other reason is that there are no signif icant diff erences in P concentrations between secondary forest and landslide litter layers, which results in a lack of diff erences in soil TP (Chen et al. 2017).

Table 3 Eff ects of forest type, soil depth, and the interaction of forest type and soil depth on soil NH 4 + -N, NO 3 - -N and available P, microbial biomass C, N and P, and enzyme activities ( n = 4)

Fig. 3 The δ 13 C ( a) and δ 15 N ( b) at each soil depth in the secondary forests and the landslides. Error bars indicate standard errors of the means ( n = 4). *Signif icant diff erences between landslides and secondary forests at the same depth ( P ＜ 0.05)

Fig. 4 Soil MBC ( a), MBN ( b), and MBP ( c) and ratios MBC:TC( d), MBN:TN ( e) and MBP:TP ( f) in the adjacent secondary forests and the landslides. The standard error of the mean is represented by the error bars ( n = 4). *Signif icant diff erences between landslides and secondary forests at the same depth ( P ＜ 0.05)

Fig. 5 Activities of phenol oxidase ( a), exoglucanase ( b),β-glucosidase ( c), N-acetyl-β-glucosaminidase ( d), L-asparaginase( e) and acid phosphatase ( f) in soils from the adjacent secondary forests and the landslides. The standard error of the mean is represented by error bars ( n = 4). *Signif icant diff erences between landslides and secondary forests at the same depth ( P ＜ 0.05)

Landslides not only result in the loss of soil TC and TN but also lead to a drop in the concentrations of available N and P. NH4+-N and NO3--N are the main forms of soil available N. After landslides, available N is exposed to the soil surface and lost through enhanced leaching and runoff (Singh et al. 2001), which could explain why NO3--N concentrations in the landslides were markedly reduced in our study. Compared with those of NO3--N, the concentrations of NH4+-N were too low to change signif icantly since NH4+-N is f ixed by microbial biomass or absorbed by clay (Medorio-García et al. 2020). Furthermore, landslides may alter N mineralization and then inf luence soil available N. Singh et al. ( 2001) suggested that N mineralization decreases in the early stage of landslides because of low soil TC and microbial biomass, which leads to low N availability.Diff erent from TP, available P concentrations were signif icantly greater in the adjacent secondary forests than that in the landslides. The availability of phosphorus is determined by the ratio of mineralization to leaching, and an increase in leaching rates in landslide soils may lead to a decrease in available phosphorus (Lemanowicz 2018).

Eff ects of landslides on soil microbial properties in temperate secondary forest ecosystems

Microbes are important components of forest ecosystems since they drive the transformation and circulation of C and nutrients (Chen et al. 2010). Our results indicate that soil microbial biomass tends to undergo a rapid decline upon surface soil and vegetation removal by landslides. These results are opposite the diff erences in abundance of δ 13 C,which were higher in the landslide soils than in the adjacent secondary forest soils. Numerous studies have been conducted on isotopic variations in C and N in woodland environments and vegetation. The most fundamental observation is that δ 13 C and δ 15 N in soil increase with soil depth (Bostrom et al. 2007; Llorente et al. 2010; Haering et al. 2013).In forest ecosystems, stable isotope accumulation is inf luenced by microbial processes. Microbial assimilation and litter decomposition are generally considered to account for isotopic changes in organic matter (Feng 2002). This conclusion suggests that the δ 13 C values in landslide soils will steadily increase over time if microbial reactions preferentially use carbon sources produced by litter decomposition.In addition, when landslides occur, the surface soil is washed away, and the subsoil is exposed, resulting in a higher 13 C in the landslides (Haering et al. 2013). However, we did not observe the same pattern for δ 15 N between the landslide and secondary forest soils as for δ 13 C. Bostrom et al. ( 2007)suggested that although there is a tight link between the circulation of C and N, the gradient of δ 15 N builds up more quickly than the gradient of δ 13 C, which could be related to the deposition mechanism of 15 N-exhausted litter on topsoil(Llorente et al. 2010). Secondary forest litters contain more N than landslide litters, resulting in higher δ 15 N in the secondary forest surface soil.

According to several studies (Sparling 1992; Ullah et al.2013), the ratios of soil microbial biomass to soil nutrients serve as responsive measures of improvements in soil quality, as they signify the use of soil nutrients and dynamic changes in C, N, and P. Higher values of these ratios indicate that soil organic carbon and available N and P can be more readily used by soil microbes (Wardle 1992). In our study,variations in soil microbial biomass after landslides were strongly associated with soil C and N dynamics. Although signif icant diff erences in TC, TN and microbial biomass were identif ied between the landslide and the secondary forest soils, there were no marked diff erences in the MBC:TC and MBN:TN ratios between the two land types. This f inding may have resulted from the similar degrees of decline between MBC and MBN and between TC and TN after landslides. Through a global analysis of microbial biomass,Fierer et al. ( 2009) considered that variation in microbial biomass is aff ected by litter inputs and TC concentrations.Indeed, we observed lower contributions of TC and less litter in landslides than in secondary forests, implying that the type and amount of litter played a role in the decrease in microbial biomass after the landslides (Deng et al. 2016).However, the MBP: TP ratios of the landslides were 64%lower than those of the secondary forests because of the signif icant reduction in MBP in the landslides and lack of change in TP.

Soil enzymatic activity is a sensitive index indicating changes in soil nutrients. Soil enzymes are greatly aff ected by the growth, activity and function of microbes, since they are mainly derived from soil microbes (Cai et al. 2018). We evaluated three soil enzymes for their ability to degrade sugar. The enzymes were phenol oxidase, exoglucanase,and β-glucosidase, which all engage in the degradation of lignin and cellulose. Because lignin and cellulose contain no nitrogen or phosphorus, we also evaluated N-acetyl-βglucosaminidase, L-asparaginase, and acid phosphatase(Sinsabaugh et al. 2009). In comparison to those in neighboring secondary forests, all soil enzyme activities signif icantly decreased in the landslides which is similar to our f indings for soil TC, TN and microbial biomass. Specif ically,β-glucosidase, N-acetyl-β-glucosaminidase and L-asparaginase decreased by 70.1%, 67.6%, and 70.0%, respectively. The activity of β-glucosidase is rarely restricted by substrates but is positively correlated with concentrations of TC and MBC (Turner et al. 2002), so the signif icant decrease in the activity of β-glucosidase in the landslides compared to the secondary forests was possibly due to the massive loss of TC and MBC in the landslides. The enzyme N-acetyl-β-glucosaminidase is involved in the breakdown of chitin, a main component of fungal cell walls (Shi et al.2006), and L-asparaginase may hydrolyze L-asparagine to NH4+-N (Tabatabai 1994). Along with changes in plant species, major soil disruptions caused by landslides will greatly alter the soil microbial population. As a result, N-acetylβ-glucosaminidase and L-asparaginase activities are signif icantly lower in landslides than in secondary forests.Lemanowicz ( 2018) demonstrated that acid phosphatase activity was actively related to TN and accessible nitrogen concentrations in soil, which is compatible with our observations. The drop in soil TN decreases the demand for P by microorganisms, thus reducing the activity of acid phosphatase. Moreover, soil enzyme activity is aff ected by soil pH (Wardle 1992), which probably accounts for the lower activities we measured in the landslides.

Conclusion

Our study demonstrated that landslides resulted in a decrease in soil TC, TN, available nutrients, microbial biomass and enzyme activity. It is essential to comprehend the stages of succession on landslides and understand landslide soil restoration. Therefore, further studies should constantly monitor the dynamics of soil chemical and microbial properties after landslides, assess how plant species interact on landslides,and evaluate the patterns of the recovery of soil properties aff ected by landslides.

Acknowledgements We are grateful to Mrs. Fengqin Li for assistance with experimental design.