Byeong-Ju Lee · Soo Hyung Eo

Abstract Deforestation or clear-cut logging affects forest ecosystems, including soil microbial communities. The purpose of this study was to investigate the effects of clearcut logging on the soil bacterial community in a temperate deciduous broad-leaved forest on Mt. Sambong, South Korea.We investigated the physicochemical characteristics and bacterial diversity of the soil in clear-cut logging and control sites. The available phosphorus (AP) level in soil was significantly lower in the clear-cut sites than in the control; however, the other physicochemical properties of soil were similar at the two sites. By examining the bacterial 16S rRNA gene using next-generation sequencing,we found that the number of bacterial taxa at the species and phylum level were similar at the control and clear-cut sites. Consistent with the high resilience of bacterial communities and absence of change in the soil physicochemical properties—with the exception of AP—we found similar levels of bacterial diversity at the two sites.Although most taxa showed similar composition ratios at the control and the clear-cut sites, some taxa such as Deltaproteobacteria, Ktedonobacteriales, Myxocccales,Polyangiaceae, Pedosphaera_f, and Solibacter showed differences after clear-cut logging. We conclude that AP was significantly associated with those bacterial taxa that showed differences in their composition ratios following clear-cut logging.

Keywords Forest soil · Metagenomics · Pyrosequencing ·Soil physicochemical characteristics

Introduction

Prokaryotes have an estimated biodiversity of about a million species (Chapman 2009). Some microorganisms are thought to occur in populations numbering more than ten times those of vertebrate species, and are present not only in the atmosphere,soil,and rivers but also in extreme environments such as the deep ocean, glaciers, and ice shelves (Torsvik et al. 2002; Saleem et al. 2016).Microorganisms are global in their distributions and are essential to ecosystem functions such as decomposition,mineralization, and retention of nutrients(Yim et al.2009;Attard et al.2010;Sˇtursovet al.2012;Haque et al.2012;Bowles et al. 2014).

Soil-dwelling forest microbes influence each other through interactions that include cooperation, competition,symbiosis, and parasitism. For example, decomposition of organic matter in forest soils occurs through the cooperation of various microbial species rather than by a single microbial species. Microbial species can affect the processes of organic degradation and nutrient synthesis (Damon et al. 2012). Changes in the underground ecosystem owing to microbial communities can influence the composition of vegetation communities (Bartelt-Ryser et al.2005), affecting plant growth and productivity (Li et al.2000; Wiggins and Kinkel 2005; Cameron et al. 2013;Souza et al. 2015). In the case of parasitic pathogens that infect plants in the soil, the bacterial composition of the entire population is very important in outbreaks of plant disease (Berendsen et al. 2012; Hernndez-Salmerón et al.2014). Thus, to better understand overall forest ecosystem functions it is critical to understand the composition of soil microbial communities because they interact with each other and affect forest ecosystems.

Human activity regarding forests alters forest vegetation and this, in turn, affects soil microbial communities(Nielsen et al. 2010; Coats et al. 2014). In particular,deforestation or clear-cut logging of forests causes, by definition, removal of trees from forests. Following deforestation, underground microbial communities are affected not only by decreased input of plant litter and root exudate but also by changes in associated environmental parameters such as temperature and moisture conditions(Ahn et al. 2012; Holden and Treseder 2014). Post-harvest changes in forest nutrient levels and microclimates affect the succession patterns of soil microbial communities(Torsvik et al. 2002). Finally, forest harvesting causes changes in microbial community diversity by altering the composition of microbial species(Kourtev et al.2002;Van der Putten et al. 2007; Bakker et al. 2014).

Using next-generation sequencing (NGS), specific characteristics of soil microbial communities have been used to index ecosystem stability and to evaluate the effects of forest disturbance (Banning et al. 2011). By applying metagenomic analysis to forest soils, Hartmann et al.(2012) showed that deforestation has a significant impact on the composition of bacterial communities, their species richness, and their community homogeneity. Rodrigues et al. (2013) showed that the diversity of soil bacterial species increased after a field was converted to grassland following deforestation. Other research was based on measurements of PCR-based denaturing gradient gel electrophoresis (DGGE) or culture-based methods, and these approaches did not yield sufficient species numbers to accurately describe microbial communities (Janssen 2006;Dowd et al. 2008).

Our aim was to quantify the changes in forest soil bacterial diversity and community structure after clear-cut logging. We used NGS to identify soil bacterial diversity.Soil physicochemical characteristics, including particle size, pH, and levels of organic matter, were analyzed and correlated with the characteristics of the soil bacterial community. We hypothesized that clear-cut logging in forests will affect the physicochemical characteristics and bacterial diversity of the forest soil. We hoped to provide insights into how to manage forests in terms of soil biodiversity by describing the effects of clear-cut logging on soil bacterial diversity.

Materials and methods

Study site

We sampled soils from the Sambongsan Leading Forest Management Complex (SLFMC; 1456 ha area; 36°28′N,127°31′E) at Mt. Sambong, Korea to investigate the effect of clear-cut logging on soil bacterial communities. Our sampling site was a typical deciduous broad-leaved forest in northeast Asia,with Quercus mongolica as the dominant species. Our site was at 482 masl with average annual temperature and precipitation of 12.5 °C and 1239 mm,respectively(Kang et al.2010).Clear-cut logging of a 7-ha plot at SLFMC was carried out in April 2013 and we used this as our sampling site.

Soil sampling

In September 2013, soils were sampled at three randomly selected points within the clear-cut site and at a control site,for a total of six sampling points. Soil sampling sites were located at least 50 m from the boundary dividing the logged and control sites. At each site, the three sampling points were ＞1 m away from the main root or stump of trees to avoid the effects of the rhizosphere of any given tree. At each sampling point, we collected 500 g of soil within 30 cm of the surface using a soil auger (5.6 cm diameter and 35 cm height) after removing the litter layer by trowel. Soil samples were transported on ice and later stored at - 20 °C pending physicochemical analyses and DNA extraction.

Soil physicochemical analysis

For physicochemical analyses, the sampled soils were airdried and sieved to select particles of ＜2 mm diameter.We determined the percentage of sand, silt, and clay from the samples using an ASTM 151H Soil Hydrometer (Gilson,USA).Soil pH was determined using a pH meter HM-30R (DKK-TOA, Japan). Organic matter (OM) and total nitrogen content (TN) were determined by the Dumas combustion method using a vario Max CN Element Analyzer (Elementar, Germany). Available phosphorus (AP)was determined by the Lancaster method using a Cary 4000 analyzer (Varian, Australia). Cation exchange capacity(CEC)was quantified by using the 1N ammonium acetate method using Kjeltec auto 8400 analyzer (FOSS,Denmark). Exchangeable cations (K+, Na+, Ca2+and Mg2+) were determined using an atomic absorption spectrometer AA280FS (Varian, Australia). Soil physicochemical analyses were performed in accordance with the manual ‘Soil tests and plant analysis (Korea Forest Research Institute 2014)’ by the Korea Forest Promotion Institute (Seoul, Korea). Soil texture composition at the logged and control sites was compared using the Chi square test.Soil pH,organic matter,total nitrogen content,available phosphorus, cation exchange capacity, and exchangeable cations (K+, Na+, Ca2+and Mg2+) at the logged and control sites were compared using the twotailed t test at α = 0.05.

DNA extraction and sequencing

DNA was extracted from 0.5 g of soil samples using a Power Soil DNA extraction kit (Mobio, Carlsbad, CA,USA). The extracted DNA was amplified using primers targeting regions V1 to V3 of the bacterial 16S rRNA gene.The primers used were 9F (5′-CCTATCCCCTGTGTGCC TTGGCAGTC-TCAG-AC-AGAGTTTGATCMTGGCTC AG-3′;the underlined sequence indicates the target region primer and ‘AC’ represents a common linker) and 541R(5′-CCATCTCATCCCTGCGTGTCTCCGAC-TCAG-XAC-ATTACCGCGGCTGCTGG-3′; ‘X’ indicates a unique barcode used for identification of each subject;http://oklbb.ezbiocloud.net/content/1001). Polymerase chain reaction (PCR) was performed under the following conditions: initial denaturation at 95 °C for 5 min, followed by 30 cycles consisting of denaturation at 95 °C for 30 s, primer annealing at 55 °C for 30 s,and extension at 72 °C for 30 s, with a final elongation step at 72 °C for 5 min. The amplified products were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA,USA). Equal concentrations of purified products were pooled and short fragments (non-target products) were removed via the Ampure beads purification kit (Agencourt Bioscience, MA, USA). DNA quality and product size were assessed on a Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA) using a DNA 7500 chip. Mixed amplicons were conducted by emulsion PCR and then deposited on Picotiter plates. Sequencing was carried out at Chunlab, Inc. (Seoul, Korea), with the GS Junior Sequencing system (Roche, Branford, CT, USA) according to the manufacturer’s instructions.

Fig. 1 Comparison of soil physicochemical characteristics between control and clear-cut sites at SLFMC, Korea

Data analysis

Our data analysis was carried out as previously described(Chun et al. 2010; Hur et al. 2011; Kim et al. 2012a).Briefly, sequencing reads from the different samples were divided by the unique barcodes of each PCR product. We then removed barcode regions, linkers, and primers from the original sequencing reads. We excluded any read containing two or more ambiguous nucleotides and limited further analyses to reads that were longer than 300 bp with a high-quality score (average score ≧25). We eliminated chimeric sequences that were detected by the Bellerophon method (Huber et al. 2004). Complete hierarchical taxonomic classification was performed by BLASTN search against the EzTaxon-e database (http://eztaxon-e.ezbio cloud.net) containing 16S rRNA gene sequences of type strains that have valid published names and representativespecies-level phylotypes (Kim et al. 2012b). Subsampling was performed to compensate for bias owing to variation in the number of sequencing reads per sample.

Table 1 Number of operational taxonomic units (OTUs) at control and the clear-cut sites

To identify bacterial operational taxonomic units(OTUs) from clear-cut and control forest soils and to understand their evolutionary relationships, we conducted phylogenetic analysis using MEGA v7.0.14 (Kumar et al.2016). A phylogenetic tree was created by the neighborjoining method.We measured OTU richness at the species to phylum levels and calculated the Shannon diversity index (H) (Shannon 1948; H =-∑Piln Pi, where Pirepresents the proportion of OTUs belong to the ithtaxon)at the phylum level. The OTUs richness and Shannon diversity indices were compared between the two sites by using the two-tailed t test. We identified shared or exclusively unique bacterial OTUs from each of the clear-cut and control sites,employing Venn diagrams.We calculated the composition ratio(%)of bacterial OTUs in each sample and compared them between the two sites by using the twotailed t test. An arcsine square root transformation of each taxonomic composition ratio (%) was used to adjust our data to a normal distribution. We conducted canonical correspondence analysis (CCA) to investigate the relationships between bacterial OTU numbers and soil physicochemical characteristics.

Fig. 2 Phylogenetic tree constructed with all soil bacterial species at control and clear-cut sites, using the neighbor joining method.Colors of each tip were associated with taxa in the phylum level. The bar graphs around phylogenetic tree represent presence/absence of each OTU at the control(red)or the clear-cut (blue) site

Results and discussion

Soil physicochemical analysis

The soil percentages of sand (30.0% in the control vs.32.9% in the clear-cut site), silt (51.9% vs. 50.4%), and clay (18.1% vs. 16.7%) indicated that the textural class of the forest soil was silt loam and proportions were similar at the logged and control sites(χ2= 0.64,P = 0.73;Fig. 1a).The available phosphorus (AP) in soils was significantly lower(t = 30.2,P ＜0.05)at the clear-cut site compared to the control with an average of 7.3 mg/kg in the control vs.3.0 mg/kg in the clear-cut samples (Fig. 1e). Soil pH,organic matter (OM), total nitrogen content (TN), cation exchange capacity (CEC), and exchangeable cations (K+,Na+,Ca2+and Mg2+)were similar at the control and clearcut sites (Fig. 1b–d, f–j). AP in forest soils tends to decrease after deforestation (Vitousek et al. 2010), and total phosphorus(P)migration in the soil,including that of AP, increases after deforestation (Deng et al. 2017).

Soil bacterial diversity and composition

NGS of six soil samples provided a total of 37,226 sequence reads (with a mean of 6204 reads, ranging from 4307 to 11,108) after quality control and primer trimming.The sequencing reads identified a total number of 1353 and 1305 bacterial OTUs from the control and clear-cut sites,respectively, at the species level. For sample-size normalization of sequence reads, we extracted 4307 reads from each sample.This procedure yielded an average number of 745 bacterial OTUs from each sample-size normalized subsample.

There were 28 phyla, 70 classes, 135 orders, 260 families,524 genera,and 1353 species across the three control samples, and there were 30 phyla, 76 classes, 145 orders,265 families,513 genera,and 1305 species across the three clear-cut sites(Table 1 and Fig. 2).Proteobacteria(33.7%)was the most dominant bacterial phylum in the soil samples from SLFMC, followed by Acidobacteria (29.5%), Actinobacteria (6.8%), Planctomycetes (6.3%), Chloroflexi(4.6%), Verrucomicrobia (4.0%), Cyanobacteria (2.8%),Bacteroidetes (2.3%), and Elusimicrobia (1.5%).

Mt. Sambong had a typical forest soil bacterial community structure dominated by Proteobacteria and Acidobacteria.Proteobacteria represent the bacterial taxon that dominates most environmental samples. Acidobacteria are known to occupy a high proportion of the soil microbial biomass (Madigan et al. 2008). Proteobacteria were more abundant than Acidobacteria at our study sties. Deciduous forests, including those at our study site, represent highly nutritive environments, and the proportion ofProteobacteria that prefer nutrient-rich conditions tends to be higher than that for Acidobacteria(Fierer et al.2007).In addition, soils at our sites were slightly acidic (mean pH 5.17), a soil condition favored by the Proteobacteria(Hartman et al. 2008; Lauber et al. 2009). Actinobacteria are mainly distributed in soil or aquatic environments and are associated with plant growth and fine root development(Nimaichand et al. 2016). Planctomycetes are abundant in aquatic environments but are frequently also found in soils(Xia et al. 2016).

Table 2 Top 5 most abundant soil bacterial taxa at the phylum to genus level at control and clear-cut sites

Mean species richness was 755 at the control and 734 at the clear-cut site(t = 0.24,P = 0.58).The average phylum richness was 23.3 at the control and 24.0 at the clear-cut site (t = - 0.29, P = 0.39). Diversity at the phylum level was 2.0 at the control and 2.1 at the clear-cut site(t = - 0.89, P = 0.29). Although soil bacterial diversity was similar at our control and clear-cut sites, we cannot rule out the possibility that our results were biased by small sample sizes. However, given the high resilience of bacterial communities and the similarity of soil physicochemical properties at our logged and control sites it is not extraordinary that we recorded similar bacterial diversity at the two sites.

Fig. 3 Venn diagram of soil bacterial taxa at the phylum to species level present at the control and clear-cut sites

In general,vegetation changes owing to deforestation or clear-cut logging affect soil microbial diversity (Hartmann et al. 2012; Lee-Cruz et al. 2013; Holden and Treseder 2014; Navarrete et al. 2015). However, bacteria have higher resistance and resilience to changes in vegetation than do other microorganisms such as fungi (Busse et al.2006; Lee-Cruz et al. 2013; Hynes and Germida 2013;Overby et al. 2015). We did not study fungal diversity in this study so it remains unknown whether fungal diversity was affected by logging. Bacterial community were,however, unaffected by clear-cut logging.

Despite the occurrence of clear-cut logging, there was no difference in the physicochemical characteristics of the soil;consistent with this,we did not observe any significant difference in soil bacterial diversity in this study.It is well known that soil bacterial communities are significantly influenced by the physicochemical properties of soils(Lauber et al. 2008; Xia et al. 2016). In particular, soil pH is related to soil characteristics that affect soil bacterial communities,such as the availability of soil nutrients,rates of absorption of exchangeable cations, levels of organic matter, and salinity (Brady and Weil 2002). We recorded no significant difference in pH between our control and logged sites. Soil bacterial diversity did not differ, in part,because the soil physicochemical properties,including pH,showed no change after clear-cut logging.

We determined the top five most abundant bacterial taxa at each OTU level (Table 2). At the phylum level, Proteobacteria accounted for 34.9% and 32.4% of bacterial OTUs at the control and clear-cut sites, respectively.Similarly, Acidobacteria were 29.2% and 29.8% of bacterial OTUs at the control and clear-cut sites,respectively.At the class level, Alphaproteobacteria were the most abundant class present at both sites. At the order level, Acidobacteriales, Rhodospirillales, and EU445199_o were the most abundant, with a slightly different ratio observed at the different sites. Acidobacteriaceae and Koribacter were the most abundant family and genus, respectively, at the control and clear-cut sites. The proportions of all of these taxa were similar at the control and clear-cut sites.

Among the taxa recorded in all three samples from both the control and clear-cut sites, we examined the common bacterial OTUs occurring at both sites or at either the control or the clear-cut site (Fig. 3 and Table 3). At the phylum level, 16 taxa were recorded at both sites, 3 taxa(OD1, OP3, and Bacteria_uc) were unique to the control site, and 2 taxa (Firmicutes and WS5) were unique to the clear-cut site. At the class level, 34 taxa were recorded atboth sites, 8 were unique to the control site, and 5 were unique to the clear-cut site. The proportion of taxa that were exclusively observed at the control or clear-cut sites was ＜1% (Table 3).

Table 3 Most abundant bacterial taxa detected exclusively unique in the control and the clear-cut sites

Numbers of major bacterial taxa were similar at the control and clear-cut sites but proportions of some taxa differed (Fig. 4). The proportion of the class Deltaproteobacteria was 6.9% at the clear-cut site, significantly lower than at the control (6.2%; t = 5.22, P ＜0.01;Fig. 4). In contrast, the proportion of the class Ktedonobacteria was 2.2% at the clear-cut site, higher than at the control site (1.6%; t = - 2.84, P ＜0.05). At order level, the proportion of Myxococcales was 2.7% at the clear-cut site,lower than 3.7%at the control site(t = 3.29,P ＜0.05). The proportion of Ktedonobacterales was 1.5%at the control, lower than 1.8% at the clear-cut site(t = - 3.68, P ＜0.05). The proportion of the genus Solibacter was 2.7% at the control, lower than 3.3% at the clear-cut site (t = - 3.56, P ＜0.05). For other taxa, proportions of the total population were similar the control and clear-cut sites.

Fig. 4 Relative species richness (%) of soil bacteria taxa at control and clear-cut sites

Fig. 5 Canonical correspondence analysis(CCA)of bacterial species richness and soil physicochemical characteristics. Soil textures (clay,sand,silt)and exchangeable cations(K+,Na+,Ca2+and Mg2+)were excluded because of multicollinearity with other soil characteristics.Bottom and left axes show CCA value of bacterial taxon. Top and right axes show CCA value of soil chemical characteristics

Relationships between soil bacteria and physicochemical characteristics

To investigate the relationship between the species numbers of each bacterial taxon and soil physicochemical properties,we conducted CCA.The first and second factors(CCA1 and CCA2), accounted for 83.3% and 9.7% of the inertia (Fig. 5), respectively, meaning that the two-dimensional CCA map was adequate to describe the relationship between soil bacterial taxa composition and physicochemical characteristics. In particular, AP showed the highest correlation with CCA1 (r = - 0.89). AP showed a negative correlation with Ktedonobacteria, Ktedonobacterales,and Solibacter,all of which were recorded in greater numbers at the clear-cut than at the control site.In contrast, AP was positively correlated with Deltaproteobacteria,Myxococcales,Pedosphaera_o,Polyangiaceae,and Pedoshpaera_f, which were recorded in low proportions at the clear-cut site.

These results show the association of AP with bacterial taxa showing a difference in the taxa proportions after clear-cut logging. Clear-cut logging reduced AP in soils and the low amount of AP is considered to have this limited the distribution and abundance of the bacterial taxa Deltaproteobacteria, Myxococcales, Pedosphaera_o,Polyangiaceae and Pedoshpaera_f. Because members of Solibacter can contribute to the soil microbial phosphorus turnover and phosphorus availability in soils by themselves(Bergkemper et al. 2016), it makes sense that the composition ratio of Solibacter was high in soils with low AP content.

Conclusion

Soil physicochemical features and bacterial communities were similar at clear-cut logging and control sites on Mt.Sambong. Among the soil characteristics, only differences in the proportions of some bacterial taxa and available phosphorus were verified. These results indicate that the impacts of clear-cut logging on Mt.Sambong:(1)were not large enough to change soil characteristics; (2) caused short-term changes from which soils had recovered by the time of our sampling;or(3)might yet be effective but had shown no change at the time of our study. This study suggests that forest management activities such as thinning,harvesting or clear-cut logging may not have a significant impact on soil bacterial communities. Long-term monitoring of soil bacteria is recommended because the impact of clear-cut logging on soil characteristics may not yet be apparent. In addition, studies of various organisms in soil such as soil fungi and microfauna as well as soil bacteria should be considered in terms of forest underground biodiversity and forest management.

AcknowledgementsThis study was carried out with the support of‘R&D Program for Forestry Technology (Project No.S211316L020130)’ provided by Korea Forest Service.