The thinning trial Cryptomeria japonica var. sinensis plantation is located at Yangziling Forest Farm, Yaan City, Sichuan Province, China (29°47′37″N, 102°56′18″E). It belongs to the subtropical monsoon mountain climate with abundant rainfall. The annual mean air temperature, precipitation and humidity are 13.1°C, 1,800 mm and 79%, respectively. The plantation was established with a spacing of 2.0 m × 1.5 m in April 2006 with an area of approximately seven hectares. The soil is yellow loam with a pH value of 4.3–4.7. The mean altitude of the site is 1,539 m, and the slope is approximately 15°.

The thinning trial was carried out in October 2014, and arranged in a randomized complete block design with three replicates and four thinning intensity treatments, including no thinning (control), light-intensity thinning (LIT: 20% of the trees removed), moderate-intensity thinning (MIT: 30% of the trees removed) and strong-intensity thinning (HIT: 40% of the trees removed). Each plot was 600 m² in size and was surrounded by buffer zones (5 m) to reduce potential edge effects. The stand density and height (m) and diameter at breast height (DBH) (cm) of each tree were measured for each plot just after thinning (October 2014) and October 2019. The tree growth performance of each treatment is presented in Table 1.

Four 2 m × 2 m subplots in each plot and one 1 m × 1 m quadrat in each subplot were established to assess the diversity of the shrub and herb layers, respectively. In each subplot or quadrat, the number was counted, and coverage and frequency were measured for each understory plant species in October 2019. The Shannon index and Pielou index were calculated (Wang G. et al., 2021), and the relative species abundance was calculated to indicate the species richness for the shrub and herb layers.

In each plot, soil profiles were excavated routinely at five points, and soils were sampled with a cutting ring (100 cm³) at topsoil layers (0–15 cm) to measure soil bulk density (SBD). Two topsoil samples were then collected at each point. One was placed on ice in the field and was then transported promptly to the laboratory and stored at −80°C for DNA extraction, and the other was dried at room temperature for physicochemical analysis. For each purpose, soils from five points were mixed equally as one sample for each plot before further treatment or analysis.

SBD was measured by the cutting ring method (Grossman and Reinsch, 2002), and soil moisture (SM) was determined by oven drying soils at 105°C for 24 h. Soil pH was measured using a pH meter (soil to water ratio was 1:2.5) (Kenworthy et al., 1976). The total nitrogen content (TN) was measured by the Kjeldahl method using a 2,300 Kjeltec Analyzer Unit (FOSS, Sweden) (Bremner and Mulvaney, 1982). The contents of total phosphorus (TP) and total potassium (TK) were determined via the ascorbic acid colorimetric method and atomic absorption method, respectively. The soil organic matter content (SOM) was assessed using the dichromate wet combustion method and a visible spectrophotometer (Lefroy et al., 1993). The alkaline hydrolysis method, molybdenum blue colorimetric method (Tan et al., 2014), and a flame photometer (Lu, 2000) were used to measure the contents of available nitrogen (AN), available phosphorus (AP) and available potassium (AK), respectively.

Soil DNA was extracted in triplicate from 0.25 g of each sample using an E.Z.N.A.® soil DNA Kit (Omega Bio-Tek, Norcross, GA, U.S.) following the manufacturer’s instructions. The extraction quality of DNA was detected by 1% agarose gel electrophoresis, and the concentration and purity of DNA were determined by a NanoDrop2000.

The V3-V4 region of the bacterial 16S RNA gene and partial fungal ITS regions were amplified by PCR using the primers 338F (5′- ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′- GGA CTACHVGGGTWTCTAAT-3′), and ITS1F (5’-CTTGGTCATTTA GAGGAAGTAA-3′) and ITS2R (5’-GCTGCGTTCTTCATCG ATGC-3′), respectively (Manter and Vivanco, 2007). All amplifications were performed in 20 μl mixtures containing 4 μl of 5 × FastPfu Buffer, 2 μl of 2.5 mM dNTPs, 0.8 μl of each primer (5 μM), 0.4 μl of FastPfu Polymerase and 10 ng of template DNA. The amplification program included initial denaturation at 95°C for 3 min, followed by 27 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s, with a final extension at 72°C for 10 min. However, the PCR cycle number was 35.

The PCR products were purified using an AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), and were quantified using QuantiFluor™-ST (Promega, USA) (Zhou et al., 2019). Subsequently, all the PCR products were pooled with equal molarity. Finally, the amplicons were pair-ended sequenced on an Illumina MiSeq platform (Illumina, San Diego, USA) by the Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China).

The raw sequences were quality-filtered and merged using Fastp¹ (version 0.19.6) and FLASH² (version 1.2.11), respectively. Operational taxonomic unit (OTU) clustering and chimera removal were conducted using Upraise software³ (version 11) based on the 97% similarity threshold (Edgar, 2013). The taxonomic assignments of 16S rRNA and ITS sequence reads were determined using the bacterial SILVA reference database (Release138⁴) and the Unite reference database (Release 8.0⁵), respectively. For both databases, the Ribosomal Database Project (RDP) Classifier⁶ (version 2.11) was used to perform taxonomic annotation for OTU representative sequences, and the confidence threshold was set to 0.7 to obtain taxonomic annotation results. The sample sequence was flattened according to the minimum number to obtain standardized data for calculation of the Shannon index and Chao index according to Cao et al. (2022), and compositions of the soil microbial community were analyzed at the phylum level.

The functional annotation and prediction of metabolic or other putative ecological functions were assessed based on the Tax4Fun-KEGG (Wang et al., 2020) and FAPROTAX databases (Lu et al., 2022) for bacteria. Tax4Fun converts the SILVA-based OTU counts into functional or metabolic profiles and computed metabolic reference profiles based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The FUNGuild database (Nguyen et al., 2016) was adopted to predict the ecological and biological functions of ITS genes for fungi, which is a flat database hosted by GitHub.⁷

One-way analysis of variance (ANOVA) and a least significant difference multiple range test (p < 0.05) were performed to assess the significance that thinning affected the assayed soil physicochemical properties (SM, SBD, pH, SOM, TK, TP, TN, AN, AK and AP), understory vegetation diversity (Shannon index, Pielou index, species richness), soil microbial community diversity indices (Shannon index and Chao index), soil microbial community compositions and the relative abundance of the genes associated with different functional categories. Partial least squares discriminant analysis (PLS-DA), a supervised method, was performed on OTU data to discriminate the microbial community profiles of the four thinning treatments (Pérez-Enciso and Tenenhaus, 2003; Ma et al., 2017). Biomarker analysis was performed by linear discriminant analysis (LDA) for effect size (LEfSe) using the Kruskal–Wallis test and All-Against-All (more strict) to determine the significance of differences in soil microbial species among the four thinning treatments. LDA was performed to evaluate the difference in each microbial taxon with a threshold value of 3.5. Redundancy analysis (RDA) was used to test the relationships of microbial groups with understory vegetation and soil properties using Monte Carlo permutations (999 repetitions). Rarefaction curves, PLS-DA, correlation heatmap, and RDA were conducted with R 4.2.0 software.

The Shannon index, Pielou index and species richness of the shrub layer under the light-(LIT), moderate-(MIT) and high-(HIT) intensity thinning treatments were remarkably higher than those under the control (p < 0.05). While significant difference was absent in the herb layer among the four treatments (p ≥ 0.05) (Table 2).

Thinning significantly influenced the soil moisture (SM), soil bulk density (SBD), pH value, and contents of total nitrogen (TN), total potassium (TK), total phosphorus (TP), available potassium (AK) and available phosphorus (AP) (Table 2). The pH value, SBD and TP were significantly higher in the control than in the LIT, MIT and HIT treatments. The SM and TK in the LIT treatment were significantly higher than those in the control, MIT and HIT treatments. The TN was significantly higher in the MIT treatment than that in other treatments. The AP of LIT and HIT was higher than that of the control and MIT treatments. Significant differences were absent in the contents of soil organic matter (SOM) and available nitrogen (AN) among the four thinning treatments.

A total of 318,732 bacterial and 819,288 fungal sequence reads were obtained from the complete dataset of 12 samples and were clustered into 3,994 and 2,807 OTUs, respectively. The rarefaction curve for thinning treatments of Shannon index on OUT level had been well captured at the amount of randomly drawn sequencing data (Figure 1). The Shannon and Chao indices of the bacterial community under the four thinning treatments ranged from 5.677 to 6.101 and from 2,485 to 3,110, respectively, and both indices of the fungal community ranged from 3.287 to 4.571 and from 1,028 to 1,402, respectively (Table 3). All of them were strongly influenced by thinning.

The top three bacterial phyla in the bacterial community were Proteobacteria (relative abundance: 34.93%), Acidobacteria (21.82%), and Actinobacteria (19.57%) (Figure 2A). Rhizobiales (17.21%) and Gammaproteobacteria_Incertae_Sedis (2.1%) were the most abundant orders observed within the Alphaproteobacteria and Gammaproteobacteria classes, respectively (Supplementary Table S1). Significant differences were mostly absent in the relative abundances of Proteobacteria, Acidobacteria and Verrucomicrobiota among the four treatments, while the relative abundance of Actinobacteriota in the LIT, MIT and HIT treatments was significantly higher than that in the control (14.3 ± 3.25%) (Figure 2A).

Ascomycota (55.69%), Basidiomycota (14.72%), and Rozellomycota (11.85%) represented most of the observed fungal phyla (Figure 2B). The relative abundances of Ascomycota in the LIT (75.05 ± 3.89%) and MIT (86.46 ± 7.40%) treatments were significantly higher than those in the HIT (59.28 ± 2.75%) and control (20.06 ± 2.62%) treatments (Figure 2B). The relative abundances of Basidiomycota, Rozellomycota and Mortierellomycota in the control were 23.33 ± 2.05%, 24.10 ± 1.51% and 13.76 ± 2.89%, respectively, and were all significantly higher than those in the LIT, MIT and HIT treatments (Figure 2B).

The PLS-DA analysis illustrated that the bacterial communities of the four thinning treatments were clearly separated from each other (Figure 3A), and the same were the fungal communities under the control and HIT treatments, while those under the LIT and MIT treatments were clustered together (Figure 3B).

The LEfSe algorithm was used to determine the taxa in the soil microbial communities (Supplementary Figure S1). In total, 22 bacterial clades presented statistically significant differences (LDA > 3.5, p < 0.05) in all soil samples. There were six differentially abundant taxa in the control (Supplementary Figure S1a), and the most important contribution to the control was the phylum Firmicutes, which accounted for 83%. The LIT treatment had the fewest biomarkers, with only two abundant bacterial clades. For the MIT treatment, the greatest contribution was made by four biomarkers, all of which came from the phylum Firmicutes, including the class Actinobacteria, with LDA scores of 4.50. The HIT treatment contained 17 biomarkers, and the categories with the highest contribution were mainly Acidimicrobia, Planococcaceae, Chitinophagales, and Rhodospirillales.

For fungi, 73 clades were determined by the LEfSe algorithm, which exhibited significant differences under four thinning treatments with an LDA threshold of 3.5 (Supplementary Figure S2). The control accounted for 72.6% of the fungal clades, while the LIT, MIT and HIT treatments had one, 15 and four clades, respectively. The proportion of the phylum Rozellomycota was highest in the control, followed by Basidiomycota (phylum) and Tremellomycetes (class). Specifically, the LIT treatment was only rich in Meripilaceae (family). The clades of the MIT and HIT treatments were mainly concentrated in Ascomycota.

Overall, the soil microorganisms composition was significantly different among the four thinning treatments, which to some extent reflects the impact of thinning on the soil microbial community.

The Pielou index in the herb layer was positively correlated with microbes, especially fungi (Figure 4). The Shannon index and species richness in the shrub layer were significantly negatively correlated with the relative abundances of Zoopagomycota, Rozellomycota, Chytridiomycota, Basidiomycota and Mortierellomycota (Figure 4B). According to the redundancy analysis (RDA), the total interpretation degree of the understory vegetation index to the bacterial genus level was 54.11% (Figure 5A), and three indicators in the shrub layer showed a significant correlation with fungal community composition (Figure 5B).

SBD and pH value were significantly and positively correlated with the relative abundance of Chloroflexi, while SM and AP were significantly and negatively correlated with it. The relative abundance of Firmicutes was negatively correlated with TN and SOM (Figure 4A). SOM and TN were significantly and positively correlated with the relative abundance of Ascomycota. They were significantly and negatively correlated with the relative abundances of Zoopagomycota, Rozellomycota, Chytridiomycota, Basidiomycota and Mortierellomycota, respectively (Figure 4B).

Figure 6A shows that the first and second ordination axes of RDA explained 61.32 and 16.19% of the variance in the soil bacterial community, respectively, and TP and TN explained the largest proportion of the variance. The pH value, SBD, TN, TK, TP and AP appeared to be the most important environmental parameters for the fungal community, and the variance explained by the first and second ordination axes was 93.73 and 2.68%, respectively (Figure 6B).

A total of 40 groups at level 2 of KEGG orthologs (KO) were detected (Supplementary Table S2). There were 6,053 KOs across all samples, and the most abundant functional pathways were presented in Supplementary Figure S3. The abundances of metabolism, environmental information processing and genetic information processing accounted for a relatively high proportion of the four thinning treatments, with mean values of 60.07%, 22.04%, and 10.03%, respectively. Significant differences were found in bacterial functional pathways among the thinning treatments. Analysis of level 2 metabolic pathways showed that the abundances of bacterial functions such as membrane transport, signal transduction, energy metabolism, and cell motility increased significantly with increasing thinning intensity.

The abundances of nucleotide metabolism, translation, replication, and repair were opposite to the abovementioned changes (Supplementary Table S2). The proportion of cellulolysis was significantly increased in the LIT, MIT and HIT treatments. The relative abundances of nitrogen fixation and phototrophy in the LIT and MIT treatments were lower than those in the control (Figure 7A).

Most fungi could not be classified (classification = ‘Unknown’), and 56.27%, 81.46%, 82.37%, and 70.91% of the relative abundance of fungal functional groups were unclassified in the control, LIT, MIT and HIT treatments, respectively (Supplementary Figure S4). The most abundant fungal parasite, undefined saprotrophs, had a mean abundance of 14.26% in the control, almost three times as much as it did in the LIT or MIT treatment (Supplementary Figure S4). Thinning significantly enhanced the abundance of soil saprotrophs and lichenized-undefined saprotrophs, and the abundance of soil saprotrophs in HIT was 5 times higher than that in the control (Figure 7B). The abundance of endophytes, plant pathogens and ectomycorrhizal fungi decreased with increasing thinning intensity.

In the present study, the Shannon index, the Pielou index and the species richness of the shrub layer in the HIT, MIT and LIT treatments were notably higher than those in the control. The main reason is that thinning results in more or larger canopy gaps and improves understory light conditions, which increases plant diversity (Tamura and Yamane, 2017). However, the effect of thinning on the richness of the herb layer was not as obvious as those in previous reports (Son et al., 2004; Muscolo et al., 2021), and the species richness was greater in the shrub layer than in the herb layer. These results were presumably related to the fact that shrubs are taller than herbaceous plants and can embrace light first, as the amount of light reaching the understory is one of the most important limiting factors affecting the species of understory plants (Yu et al., 2022). Additionally, the lack of seed resources of understory vegetation species in and around the plot due to the relatively high tree density before thinning (Ali et al., 2019b) and the altitude of the site might be another reason for the lower number of vegetation species.

Understory vegetation cover and richness increased with increasing thinning intensities in the present, which led to an increase in the amount of litter (He and Barclay, 2000; Teste et al., 2012), followed by changes in soil physical attributes and nutrient cycling (Elliott et al., 2015; Xiao et al., 2018; Yang et al., 2019). The roots and dead parts of the newly occurred herbs and shrubs are rich in lignin and other substances that enhance the looseness of the topsoil layer (Osman, 2013), which was the main reason why the soil bulk density decreased as the thinning intensity increased in the present study. Then, the humification of litter increased the contents of soil organic acids and organic matter (Sayer, 2006; Zou et al., 2017). The soil pH value difference in the present study was related to the changes in soil organic acid generation activity after thinning, and soil nutrients, especially the contents of soil organic matter (SOM) and total nitrogen (TN) were increased after thinning. Compared with the control, the SOM and TN under the high-intensity thinning treatment (HIT) increased by 19.87% and 35.29%, respectively (Table 2). Thus, it can be seen that LIT and MIT treatments were conducive to improving the understory vegetation diversity and soil physicochemical properties of young C. japonica var. sinensis plantation.

With increasing thinning intensities, the Shannon and Chao indices of soil bacteria and fungi first decreased and then increased in the C. japonica var. sinensis forest (Table 3). This was different from the study of Dang et al. (2018), in which the Shannon and Chao indices in soil bacteria and fungi did not vary significantly with thinning intensities. Cai et al.’s (2020) study on the thinning of Larix gmelinii var. principis-rupprechtii plantation showed that the Shannon index of bacterial communities were higher in the medium-intensity than in other treatment. Lin et al. (2016) demonstrated that soil fungal communities did not change significantly after 21 months of thinning in a Japanese cedar (Cryptomeria japonica) plantation in Taiwan. The reason for inconsistent results among these studies may be related to the difference in species composition and distribution pattern of the understory vegetation since microbial community composition varies with vegetation species (Overby et al., 2015). Thinning promotes the growth of planted trees and understory vegetation, which is the main reason for changing the composition and decomposition rate of litter and therefore affects the activity of soil microorganisms. This was also demonstrated in a study of mature Pinus contorta forests, which found that the heterogeneity of understory plant species and their rhizosphere resources (e.g., root exudates, nutrients) influenced the patterns of variation in belowground microbial communities (McIntosh et al., 2013). In addition, the changes in the soil bacterial and fungal communities may be related to their adaptability to the changed microenvironment in the stands after thinning.

Thinning significantly changed the soil bacterial community structure in the C. japonica var. sinensis plantation at the phylum level, and the relative abundances of Proteobacteria and Actinobacteriota increased with increasing thinning intensity (Figure 2A). This could be explained by the enhanced soil nitrogen content under thinning treatments (Table 2), as in Wang et al.’s (2018) study on Chinese fir (Cunninghamia lanceolata) plantations, in which nitrogen addition could increase the relative abundance of Proteobacteria and Actinobacteria. The study of Cai et al. (2020) showed that these two phyla had higher abundance in the no thinning treatment of Larix plantations. The most reasonable explanations are that the bacterial composition is closely related to the soil conditions, climate and sampling time (Green et al., 2008), and the adaptability of soil microorganisms in the community to soil environmental changes is different (Navarrete et al., 2015; Ren et al., 2016). Accordingly, the phylum level of bacteria may respond differently to changes in forest density. This study found that soil bacterial communities at the phylum level were mainly composed of Proteobacteria (relative abundance 34.93%) in four treatments (Figure 2A), and the second dominant phylum was Acidobacteriota, followed by Actinobacteriota and Chloroflexi. Coincidentally, the LEfSe algorithm showed that Bacilli (Firmicutes) may be a potential biomarker of bacteria in the control (Supplementary Figure S1). With increasing thinning intensities, Gammaproteobacteria (Proteobacteria) and Acidimicrobia (Actinobacteriota) may be potential biomarkers for light-intensity (LIT) and high-intensity (HIT) thinning treatments, respectively. For moderate-(MIT) intensity thinning treatments, Actinobacteriota and Chloroflexi may be potential biomarkers. Proteobacteria show a significant positive correlation with the soil organic carbon concentration, indicating that increases in root secretions or plant litter decomposition products stimulate their growth (Huhe et al., 2017). Actinobacteria species are recognized as degraders of cellulose, chitin and other complex carbon compounds, which gives them a central role in the carbon cycle and in the turnover of organic matter (Yergeau et al., 2010; Zhou et al., 2017). Acidobacteria and Chloroflexi also play an important role in the decomposition of organic matter and nutrient cycling (Bryant and Frigaard, 2006; Eichorst et al., 2018). Firmicutes can reduce the abundance of plant pathogens (Ali et al., 2019a), and they also embrace sulfate- and iron- reduction abilities (Gupta et al., 2018). These results demonstrate that changes in soil bacterial community composition are directly related to soil characteristics, especially changes in carbon and nitrogen.

The abundance of Ascomycota increased significantly, while Basidiomycota, Rozellomycota, and Mortierellomycota decreased at the soil fungal phylum level with increasing thinning intensities (Figure 2B). Ascomycota is a potential biomarker of the LIT, MIT and HIT treatments, and as the most abundant fungus, Ascomycota accounted for up to 75% (MIT treatment) of fungi in the C. japonica var. sinensis plantation, in which Archaeorhizomycetes had the highest abundance (Supplementary Table S1), and increased with increasing thinning intensities. Archaeorhizomycetes are a type of saprophytic fungi (Rosling et al., 2011) that play an important role in the cycling of carbon derived from living or dead roots in soil (Rosling et al., 2013). The LEfSe algorithm showed that fungi had the most taxonomic groups in the control, which were 53, 3.53 and 13.25 times that of the LIT, MIT and HIT treatments, respectively. Rozellomycota, Basidiomycota, and Tremellomycetes (Basidiomycota) may be potential biomarkers for the control (Supplementary Figure S2). Basidiomycota are found in almost all terrestrial environments, and many of them obtain nutrients by decaying wood, leaf litter and other dead organic matter. Therefore, they play an indispensable role in the carbon cycle. C. japonica var. sinensis had dense branches with a dark and moist environment in the control plots, and it was easy to cause the bottom branches to wither away and fall down. Branches of conifer species are more difficult and need a longer time to be decomposed than shrubs and herbs, and the present study was conducted 5 years after thinning, which may be the reason for the higher abundance of Basidiomycota in the control.

The relationship of microbial community structure with understory vegetation and soil is one of the vital goals of microbial ecology (van der Heijden et al., 2008; Fu et al., 2018). In this study, the Pielou index of the herb layer was positively correlated with the abundance of Chloroflexi, Planctomycetota and Verrucomicrobiota, and was negatively correlated with the abundance of Actinobacteria (Figure 4A). The Shannon index and species richness in the herb layer and the Shannon index, Pielou index and species richness in the shrub layers were almost negatively correlated with the abundance of the top 10 bacterial phyla, and the bacterial community was more likely to be significantly correlated with the soil properties than vegetation. This demonstrated that soil physicochemical properties had more important effects on bacterial community structure than did understory vegetation diversity. We also found that the pH value and soil bulk density (SBD) were significantly positively correlated with the relative abundance of Chloroflexi. One study on the depth profiles of microbial communities in high-elevation soils demonstrated that soil pH was an important driver of forming soil bacterial communities in a given region (Chu et al., 2016). However, Dang et al. (2018) revealed that pH was not correlated with dominant members of bacterial communities in Pinus tabuliformis Carriere plantations. The SM and available phosphorus content (AP) were the main driving factors of the soil bacterial community structure in the C. japonica var. sinensis plantation; they were positively correlated with the abundance of Proteobacteria and negatively correlated with the abundance of Chloroflexi. Proteobacteria can improve soil fertility and sustainability (Rousk et al., 2009; Niu et al., 2020). Chloroflexi is slow-growing and was once classified as an oligotrophic group, and its growth and development may be limited by soil nutrient accumulation (Xu et al., 2021; Zhang et al., 2022), which is why TN, AP and the content of total potassium (TK) were negatively correlated with the abundance of Chloroflexi.

The correlation heatmap of understory vegetation diversity and soil properties with the main phylum of fungi showed that the relative abundances of the main phylum fungi, except Ascomycota and Glomeromycota, were negatively correlated with the indices of the shrub layer and positively correlated with SBD, pH value and the content of total phosphorus (TP). RDA further confirmed that SBD, pH value, TK, TP, TN, AP and shrub diversity affected the abundance of fungal communities. Overall, soil parameters were the most important factors affecting the soil fungal community structure. This was in accordance with Adamo et al.’s (2021) study in 42 pure and mixed pine forests, which showed that soil chemistry significantly affected the variability of soil fungal communities.

Thinning had significant effects on the relative abundances of the dominant fungal communities rather than the dominant bacterial communities, which might indicate that the bacterial community has stronger resistance to changes in stand density than fungal communities (Wang C. Q. et al., 2021). Some studies have also reported that fungal communities have a more obvious response to plant–soil feedback than bacterial communities (Dang et al., 2018; Hou et al., 2021). This is mainly because the bacterial community has a smaller ecological niche in soil and a weaker symbiotic relationship with plants than the fungal community (Shan et al., 2017). Therefore, changes in soil properties, directly or indirectly caused by the intensity of thinning, were less responsive to bacterial than to fungal community diversity.

Understory vegetation diversity and contents of soil nutrients in the C. japonica var. sinensis plantation increased with increasing thinning intensities. Different bacterial and fungal species had different strategies for adaptation to microenvironmental changes in the thinned plantations. Analysis of microbial function prediction showed that 60% of bacteria were involved in metabolic pathways, some of which are known to cause human disease, and the remaining bacteria were involved in genetic information processing, environmental information processing, organismal systems and cellular processes. The abundances of membrane transport, nucleotide metabolism and signal transduction in the LIT, MIT and HIT plots were higher than those in the control. Bacteria often have a strong relationship with human diseases (Geng et al., 2020; Wang et al., 2020). In this study, bacterial infectious diseases accounted for the highest proportion of human diseases, up to 10%. Fortunately, the proportion of human diseases decreased with increasing thinning intensities. This inferred that the change in the bacterial community caused by thinning in the C. japonica var. sinensis plantations might reduce the possibility of harm to humans.

Thinning significantly increased the functional groups ‘soil saprotrophs’ while decreasing the functional groups ‘endophytes’ and ‘plant saprotrophs’ in fungi. This proved that soil nutrients under the C. japonica var. sinensis plantation could be improved by thinning since the abundance of soil saprotrophs was positively correlated with soil fertility (Kyaschenko et al., 2017). This was further confirmed by RDA, from which a significant positive correlation was found between soil properties and the fungal community. Based on the composition and community function of soil bacteria and fungi, it has been predicted that higher thinning intensities are beneficial for soil properties and soil microbial relationships in Cryptomeria japonica var. sinensis plantations.