Soil Phosphorus Availability Controls Deterministic and Stochastic Processes of Soil Microbial Community along an Elevational Gradient in Subtropical Forests

Abstract

Revealing the assembly mechanisms of the soil microbial community, which is crucial to comprehend microbial biodiversity, is a central focus in ecology. The distribution patterns of microbial elevational diversity have been extensively studied, but their assembly processes and drivers remain unclear. Therefore, it is essential to unravel the relationship between the deterministic and stochastic processes of the microbial community assembly and elevational gradients. Here, our study built upon previous physicochemical analyses of soil samples collected along an elevational gradient (900–1500 m) in Daiyun mountain, a subtropical forest located in southeastern China. Using the phylogenetic-bin-based null model analysis (icamp) and multiple regression on matrices approach, we explored the major drivers that influence the assembly processes of soil bacterial and fungal community across elevations. The results showed that: (1) bacterial rare taxa exhibited a broad habitat niche breadth along the elevational gradient; (2) homogeneous selection and homogenizing dispersal proved to be the most important assembly processes for the bacterial and fungal community; (3) soil phosphorus availability mediated the relative importance of deterministic and stochastic processes in the soil microbial community. Notably, the relative abundance of dominant microbial taxa controlled by homogeneous selection and homogenizing dispersal increased with increasing soil phosphorus availability. Collectively, the assembly processes of microbial elevational communities of the subtropical mountains in China can be explained to some extent by variations in the soil phosphorus availability. This conclusion provides valuable insights into the prediction of soil microbial diversity and phosphorus nutrient cycling in subtropical montane forests.

1. Introduction

Soil microbial communities play a crucial role in nutrient cycling and energy flow within terrestrial ecosystems [1,2]. Notably, in this regard, variations in microbial community species’ composition across space, known as microbial β diversity, is a central focus of ecology [3], and is typically determined by deterministic and stochastic processes [3,4]. Thus, uncovering the assembly processes of soil microbial communities can offer vital insights into the generation and maintenance of biodiversity [5].

Understanding the deterministic and stochastic processes that control soil microbial assembly is crucial for predicting the response of microbial β diversity to environmental changes [6]. Deterministic processes highlight the relationship between microbial β diversity with the biotic and abiotic factor, and are classified into homogeneous and heterogeneous selection at different local scales [3]. The homogeneous selection usually leads to a reduction in β diversity, while heterogeneous selection implies that community assembly may differ greatly with variations in environmental conditions over time and space [7]. In contrast, stochastic processes emphasize that microbial β diversity is primarily mediated by dispersal events, ecological drift, genetic mutations, and historical contingencies [8]. Homogenizing dispersal and dispersal limitation are regarded as the primary factors influencing regional species pools and community assembly [7,9]. Early views suggested that deterministic or stochastic processes independently determine microbial community structure [10,11]. Recent studies on microbial β diversity patterns generally indicate that deterministic and stochastic processes act on community assembly together, although their relative importance may vary [12]. However, it is widely acknowledged that assembly processes tend to operate at the finer biological organizations rather than at the whole communities [12,13]. It is likely that the diversification rate, dispersal ability, and susceptibility to drift variation vary substantially among different microbial individual lineages/taxa [14]. Hence, revealing environmental filtering and other assembly processes in microbial individual lineages [3] will help ascertain the underlying mechanisms of soil microbial β diversity.

The influence of abiotic factors on soil microbial community assembly varies with different regional scales [15]. Generally, large spatial scales exhibit greater environmental heterogeneity, resulting in enhanced environmental filtering and reduced ecological drift. Meanwhile, dispersal is a key contributor to the variability of a regional species pool [3,16]. However, the local scales have not been thoroughly explored due to the limited number of individuals and relatively complex microenvironment. Additionally, the microenvironment of small spatial scales may make the microbial communities more homogeneous [17], which enhances ecological drift [16]. Elevation strongly mediates environmental attributes (e.g., temperature and light), resulting in soil characteristics and vegetation alterations [18]. Consequently, elevational gradients provide a natural experimental platform for investigating the assembly of soil microbial communities [19,20]. Previous studies have mainly focused on the distribution patterns of soil microorganisms (bacteria, archaea, and fungi) and the ecological drivers that control microbial diversities across elevations [19,21,22]. However, our understanding of the assembly mechanisms of soil microorganisms along elevational gradients remains incomplete, which impedes our ability to fully comprehend how drivers control soil microbial community assembly.

The forest ecosystem is the most intricate terrestrial ecosystem [23]. The climatic conditions of montane forests depend on various factors, including topography, elevation, and the interplay between the forest nutrient cycle and global climate [24]. Similarly, the diversity and rarity of montane forests, community assembly, migration, and niche evolution collectively contribute to montane forest biodiversity patterns [24]. In particular, the habitat niche is a pivotal trait that impacts species sorting and dispersal in shaping ecological communities [25]. Taken together, these forest factors create an ideal experimental region for studying the assembly of soil microbial communities [26]. The studies revealed that the interplay of forest environmental factors dominates the community structure in phylogenetic and functional traits [27]. In subtropical forests, soil phosphorus limitation is a commonly observed phenomenon [28]. In nutrient-impoverished soils, soil microbial communities (bacteria or fungi) may be primarily driven by phosphorus or nitrogen availability [29], which can further restrict microbial community assembly [28]. In a long-term fertilization experiment, Ma et al. [30] found that alterations in soil stoichiometry (N/P) and phosphorus availability shape fungal communities by influencing fungal enzyme activity and functional gene abundance. Similarly, Peng et al. [31] identified that soil phosphorus plays a critical role not only in shaping the diversity and community structure of both abundant and rare bacteria but also in mediating the balance between the deterministic and stochastic processes. Thus, we hypothesize that phosphorus availability plays an essential role in the soil microbial community assembly of subtropical forest ecosystems. However, the drivers that determine the assembly processes of the dominant microbial community in the forest ecosystem are poorly resolved. Therefore, elucidating the correlations between dominant microbial taxa controlled by major assembly processes and environmental factors can provide vital insights into the complex mechanisms of the forest ecosystem.

Daiyun mountain belongs to a typical montane forest ecosystem on the southeastern coast of China [32]. The pronounced distinctions in climatic attributes between different elevational gradients create typical vegetation cover types. This distinctive feature offers an ideal setting for studying the soil microbial community along an elevational gradient [33]. In previous studies, we examined the stoichiometric characteristics of soil carbon, nitrogen, and phosphorus along the elevational gradient and found that soil phosphorus, as a major limiting factor, fluctuates with changes in elevation [34]. Thus, it is crucial to determine the extent to which these fluctuations in soil phosphorus levels impact the microbial community in Daiyun mountain. Although recent investigations have explored the structure and functional diversity of soil microbial communities in Daiyun mountain [33,35], there is still a significant knowledge gap regarding the assembly of these communities across elevations. To fill this knowledge gap, we used high-throughput metagenomic technologies to evaluate the effects of deterministic and stochastic processes on soil bacterial and fungal community assembly along the elevational gradients. Here, our main hypothesis is as follows: (1) homogeneous assembly processes are enhanced in soil microbial community along an elevational gradient, and homogeneous selection likely serves as the major assembly process of the microbial community. Furthermore, we propose that (2) soil phosphorus availability mediates the abundance and assembly of dominant microbial taxa as the microbial community structure is strongly associated with the soil and vegetation phosphorus content [36], and whether the mechanism exists in Daiyun mountain. To address the two hypotheses, we applied phylogenetic-bin-based null model analysis (icamp) [14] to quantitatively speculate the assembly mechanisms of soil microbial community in Daiyun mountain, while also further analyzing microbial individual lineages.

2. Materials and Methods

2.1. Study Location

This study was conducted in Daiyun Mountain National Nature Reserve (118°05′~118°20′ E, 25°38′~25°43′ N) located in the southeastern region of China, which lies within the transitional zone between subtropical and south subtropical forests [34] (Figure 1a). The reserve spans a total area of 134.72 km², with the highest peak of Daiyun mountain reaching an elevation of 1856 m. The major soil type in the area is mountain Ferric acrisols soil (900–1600 m a.s.l.), according to the World Reference Base for Soil Resources (WRB) [34]. In addition, this region belongs to a mid-subtropical montane climate with a mean annual temperature approximately ranging from 12.1 to 16.9 °C, and the mean annual relative humidity exceeding 80% [34]. The natural forest in this site is mainly composed of coniferous forest (e.g., Pinus massoniana Lamb, P. taiwanensis Hayata, Cunninghamia lanceolata (Lamb.) Hook) and evergreen broad-leaved forest (e.g., Castanopsis fabri Hance, C. eyrei (Champ.) Tutch) [33].

2.2. Soil Samples and Data Collection

In July 2018, seven standard elevational plots (20 m × 30 m) were set up along an elevation of 900–1500 m (DY9, DY10, DY11, DY12, DY13, DY14, and DY15) on the southern slope of Daiyun mountain [37]. The slopes were measured using a GPS device (Magellan eXplorist 510, Santa Clara, CA, USA). Each elevational plot was divided into three sample plots (10 m × 20 m), and within each sample plot, soils collected by five-point method were mixed to create a soil sample. In total, 21 soil samples were collected. Then, we packed the soil samples to ice boxes and brought them to the laboratory. All soil samples was sieved through 2 mm to remove plants roots, and animal residues, then divided them into two separate parts for soil DNA extraction and physicochemical analyses, respectively.

2.3. Soil Physicochemical Analyses

Soil pH was measured using pH potentiometry (water-soil ratio 2.5:1). Soil water content (SWC) was determined by the drying method. Soil temperature (ST) was measured every four hours using temperature recorders (DS1922L-F50 iButton, Maxim Integrated, San Jose, CA, USA). Soil available nitrogen (AN) content was determined by the alkaline solution diffusion method. Soil available phosphorus (AP) content was determined by the molybdenum antimony anti-colourimetric method. An elemental analyzer (Vario MAX cube CN, Elementar Analysensysteme GmbH, Langenselbold, Germany) was used to determine the total carbon (TC) and total nitrogen (TN) content in natural air-dried soil samples. Total phosphorus (TP) content and total potassium (TK) content were measured by the acid nitrification digestion of soil samples, and the resulting supernatant was analyzed using an inductively coupled plasma optical emission spectrometer (PE OPTIMA 8000, PerkinElmer, Waltham, MA, USA). The results of soil physicochemical contents related to the He et al. [33].

2.4. DNA Extraction and High-Throughput Amplicon Sequencing

Total soil DNA was extracted using the BioFast Soil Genomic DNA Extraction Kit (BioFlux Company, Hangzhou, China) following the manufacturer’s instructions. The extracted DNA was detected by 1% agarose gel electrophoresis, and the concentration of DNA was assessed by the NanoDrop 2000C spectrophotometer (Thermo Scientific, Waltham, MA, USA). Next, the purified DNA was collected by gel recovery kit (TianGen Biotech Co., Ltd., Beijing, China) and stored at −20 °C in the refrigerator. The V3-V4 region (338F/806R) of the 16S rRNA gene for the bacteria and the ITS1 region (ITS5-1737F/ITS2-2043R) of the fungal 18S rRNA gene were amplified by high-throughput amplicon sequencing technology, respectively. Polymerase chain reaction (PCR) was carried out according to the following thermal program: pre-denaturation at 95 °C for 5 min, denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s, extension at 72 °C for 40 s for 25 cycles, and final extension at 72 °C for 7 min. PCR products were electrophoresed on a 2% agarose gel, and then collected using a gel recovery kit (AxyPrepDNA, Axygen, Hangzhou, China). The samples were sequenced using the IonS5TMXL sequencing platform by the single-end sequencing technology.

2.5. Statistical Analyses

All packages for statistical analyses were performed in R v4.2.1.

2.5.1. Soil Microbial Community Structure Analysis

Quality filtering on the raw reads were performed under specific filtering conditions to obtain the high-quality clean reads according to the Cutadapt (V1.9.1) quality controlled process [38]. Next, the reads were compared with the reference database (Silva database) using the UCHIME algorithm (UCHIME Algorithm) to detect chimera sequences [39,40], and then the chimera sequences were removed [41]. Then, the clean reads were finally obtained. Sequences with ≥97% similarity were assigned to the same OTUs in Uparse software (Uparse v7.0.1001) [42]. For each OTU representative sequence, the Silva Database was used based on the Mothur algorithm to annotate taxonomic information [40]. OTUs’ abundance information was normalized using a standard of sequence numbers corresponding to the sample with the least sequences. In the study, the OTUs were classified into six different taxa (AAT, ART, MT, CRT, CAT, and CRAT) based on their relative abundance [43]. AAT and CAT were collectively identified as abundant taxa, while ART and CRT were recognized as rare taxa [44]. The dplyr package was used to calculate all OTUs across elevations, those with higher relative abundance and greater contributions to the major assembly processes were deemed as dominant microbial taxa. According to the annotation information, the relative abundance of the microbial phylum was selected for heatmap analysis using the pheatmap package. Additionally, the phylogenetic trees were constructed using FastTree in Qiime2 [45], before being annotated and visualized in the ggtree and treeio packages.

2.5.2. Habitat Niche Breadth Analysis

The habitat niche breadth of the microbial taxa is a crucial factor to consider when studying deterministic and stochastic processes, particularly in relation to the dispersal of microbial communities [46]. Microbial taxa with a wider niche breadth can exhibit greater metabolic flexibility and dispersal ability at the community level [46]. The spaa package was used to calculate the niche breadth (Levins’ niche breadth) and niche overlap in different microbial taxa [47]. Normality tests were performed in the base package, and the niche breadth of the three taxa was evaluated by the ‘LSD-test’ function (p < 0.05) in the agricolae package.

2.5.3. Null Model Analysis

Compared to other null model analyses, the phylogenetic-bin-based null model analysis (icamp) demonstrates a higher quantitative expression in the assessment of community assembly [14]. Based on the icamp model analysis, the assembly processes of the soil microbial community comprised homogeneous selection (HoS), heterogeneous selection (HeS), homogenizing dispersal (HD), dispersal limitation (DL), and drift (DR). The observed microbial taxa were initially classified into ‘bins’ according to their phylogenetic relationship. The β-net relatedness index (βNRI) and Raup Crick metric (RC) of each bin were used to assess the relative importance of different assembly processes within individual lineages [14]. Pairwise comparisons with βNRI < −1.96 were deemed to be controlled by homogeneous selection within each bin, whereas those with βNRI > 1.96 were considered to be influenced by heterogeneous selection. The remaining pairwise comparisons, where |βNRI| ≤ 1.96, were subdivided using the taxonomic diversity metric RC. Comparisons with RC < −0.95 were controlled by homogenizing dispersal, while those with RC > 0.95 were controlled by dispersal limitation. Values with |βNRI| ≤ 1.96 and |RC| ≤ 0.95 represented the impacts of drift and other assembly processes. Meanwhile, we assigned weights to each process within bins based on the relative abundance of each bin [14], enabling us to estimate the relative importance of each process at different elevations. To ensure the accuracy of the icamp model analysis, we used the confidence metric and 1000 permutations [14]. In addition, we employed bootstrapping in the icamp package to estimate the variation in the relative importance of each process at different elevations. To select the dominant microbial taxa controlled by major assembly processes [48], we used the dplyr package and visualized by the iTOL v5 [49].

2.5.4. Correlation Analysis

To identify the effects of environmental factors on major assembly processes, we firstly used the ‘varclus’ function in the Hmisc package to assess the redundancy (Spearman ρ² = 0.7) among environmental variables [50]. Secondly, the Mantel test in the vegan package was used to determine the correlation between environmental factors and major assembly processes [51], with the Mantel r (spearman coefficient) being calculated. To further investigate the correlation between dominant microbial taxa and environmental factors, we conducted multiple regression on matrics (MRM) analysis using the vegan package and the ecodist package [52]. Furthermore, the linear regression model was established using the ‘lm’ function in the stats package.

3. Results

3.1. Soil Microbial Community Composition and Structure

The soil bacterial community consisted primarily of Proteobacteria, Acidobacteria, Actinobacteria, Chloroflexi, Planctomycetes, and Verrucomicrobia along the elevational gradient. Proteobacteria (33.4% of all OTUs), Acidobacteria (8.1%), and Actinomycetes (6.3%) were the dominant bacterial phylum (Figure 1b). On the other hand, the fungal community was mainly composed of Basidiomycota, Ascomycota, Zygomycota, Glomeromycota and Chytridiomycota. Ascomycota (61.5% of all OTUs) and Basidiomycota (26.2%) were the dominant fungal phylum (Figure 1c). Among the total bacterial OTUs, 0.2% (16) were classified as abundant taxa, 7.1% (434) as moderate taxa, and 92.6% (5675) as rare taxa (Figure S1a). For the total fungal OTUs, abundant taxa accounted for 0.8% (32), moderate taxa for 0.3% (13), and rare taxa for 96.8% (3714) (Figure S1b).

The heatmap analysis revealed the distribution pattern of soil microbes across different elevations. The bacterial community at the elevation of 900 m (DY9B), 1000 m (DY10B) and 1100 m (DY11B) were clustered, indicating that their community structure was more similar to each other, and differed from those at the elevation of 1200 m (DY12B), 1300 m (DY13B), 1400 m (DY14B), and 1500 m (DY15B) (Figure S2a). Meanwhile, the fungal community structure of the seven elevations was classified into three groups (Figure S2b).

3.2. Habitat Niche Breadth

Differences in habitat niche breadth were observed among the overall, abundant, and rare taxa in the soil microbial community. In the bacterial community, the overall taxa exhibited the widest niche breadth, followed by the rare taxa and abundant taxa (Figure 2a). This suggests that the abundant taxa use limited resources within the bacterial community. For the soil fungal community, the rare taxa had the narrowest niche breadth (Figure 2b), which implied that the rare taxa are at a disadvantage in resource competition. Additionally, the niche breadths in the bacterial community were significantly broader (p < 0.05) than those in the fungal community for the three taxa (Figure 2c). The niche overlap between both the abundant and rare microbes exceeded 90%.

3.3. Soil Microbial Community Assembly Processes and Their Influencing Factors

The relative importance of assembly processes varied along the elevational gradient, with homogeneous selection (HoS) and homogenizing dispersal (HD) serving as the two most pivotal assembly processes for the soil microbial community at all elevations (Figure 3a,c). In the bacterial community, the relative importance of HoS (45.9%–78.5%) and HD (18.7%–52%) varied significantly across elevations. The relative importance of HoS (78.5%) was highest at an elevation of 1000 m. The relative importance of HD (52%) peaked at 1500 m, where the relative importance of HoS (45.9%) was the lowest. For the fungal community, the relative importance of HoS (37.5%–90.5%) was much higher than that of HD (9.4%–37.3%). The relative importance of HoS was 90.5% and 72.6% at the elevation of 1100 m and 1400 m, respectively. Meanwhile, the relative importance of fungal HD remained less than 40% at all elevations (Figure S3). Furthermore, we found that the deterministic and stochastic processes of the bacterial community tended to balance out with higher elevations (Figure 3b). For the fungal community, deterministic processes were found to be more important than stochastic processes (Figure 3d).

Using the Mantel test, we determined the contribution of the environmental factors mediating the two major assembly processes (Table S1). Soil available phosphorus (AP) was significantly and positively correlated with HoS (bacteria: r = 0.382, p < 0.05; fungi: r = 0.471, p < 0.05) and HD (bacteria: r = 0.497, p < 0.05; fungi: r = 0.499, p < 0.05); thus, AP served as the best predictor for HoS and HD along the elevational gradient. We observed a significant correlation between the soil total phosphorus (TP) content and HD (bacteria: r = 0.642, p < 0.05; fungi: r = 0.535, p < 0.05). Subsequently, the strong effect of soil temperature (ST) in the bacterial community was observed, as the relative importance of HoS (r = 0.400, p < 0.05) and HD (r = 0.602, p < 0.01) significantly varied with ST.

3.4. Soil Microbial Community Assembly Processes Based on Individual Lineages

Based on the icamp model, the bacterial 6126 OTUs were divided into 52 phylogenetic bins (Figure S4a) and the fungal 3836 OTUs were divided into 42 phylogenetic bins (Figure S4b). Of the bacterial bins, HoS dominated 31 bins (59.6% of bin numbers and 59.3% of relative abundance) and HD dominated 14 bins (26.9% of bin numbers and 31.6% of relative abundance, Table S2a). For the fungal bins, HoS dominated 19 bins (45.2% of bin numbers and 47.5% of relative abundance) and HD dominated 15 bins (35.7% of bin numbers and 35.0% of relative abundance, Table S2b). Thus, we found that the dominance of HoS was stronger than HD for both bacterial and fungal bins.

Of the bacterial and fungal bins, ten bins controlled by two major assembly processes were selected, respectively. The 10 bacterial bins constituted 19.3% of the total bacterial abundance, with 49.4% and 58.6% contributions for HoS and HD (Table S2a), respectively. Of these, five bins were mainly controlled by HoS (Group 1B), and five mainly by HD (Group 2B, Figure 4a). Comparatively, the 10 fungal bins accounted for 30.0% of the total fungal abundance, with 60.7% and 55.6% contributions for HoS and HD (Table S2b), respectively. Of these, five bins were mainly controlled by HoS (Group 1F), and five mainly by HD (Group 2F, Figure 4b). The selected bins (Group 1B, Group 2B, Group 1F, and Group 2F) were deemed the dominant microbial taxa. The dominant bacterial taxa controlled by HoS and HD included Proteobacteria, Acidobacteria, Actinobacteria, Verrucomicrobia, and Firmicutes (Table S3a). Additionally, the dominant fungal taxa controlled by HoS and HD included Basidiomycota, Ascomycota, and Zygomycota (Table S3b). More importantly, the dominant microbial taxa controlled by HoS and HD primarily belonged to rare taxa (Table S3).

3.5. Correlation between Environmental Factors and Dominant Microbial Taxa

Considering the interaction of environmental factors along elevational gradients, we further used the MRM model to establish the correlation between environmental factors and the dominant microbial taxa (Figure 4c). Our results revealed that AP was the most important variable for predicting the abundance of dominant microbial taxa (p < 0.01). Interestingly, the most important predictor of the dominant microbial taxa was consistent with the best factor predicting the major community assembly processes (Table S1, Figure 4c) and the most significant and positive linear regressions was found between AP and the relative abundance of dominant microbial taxa (p < 0.001). Soil water content (SWC), total potassium content (TK), total nitrogen content (TN), and slope (SLO) were observed to have a strong effect on the dominant bacterial taxa (p < 0.05). The total carbon content (TC) had a significantly negative correlation with the relative abundance of dominant bacterial taxa (p < 0.05). In addition, the TK and ST were the dominant drivers for the dominant fungal taxa (p < 0.05).

4. Discussion

4.1. Structure of Soil Microbial Community along the Elevational Gradient

In our study, the dominant bacterial phylum consisted mainly of Proteobacteria, Acidobacteria, and Actinobacteria (Figure 1b), and the dominant fungal phylum was Ascomycota and Basidiomycota (Figure 1c). These findings are consistent with the previous conclusions on the spatial distribution of soil microorganisms in subtropical montane forests [53,54]. Proteobacteria, as a major bacterial phylum, play an essential role in responding to soil nutrients [55]. In addition, Basidiomycota explains the spatial properties of the plant root community in environmental heterogeneity [56].

Most species in all ecosystems are rare species, especially in the microbial community [57]. Our results supported this view, as the majority of soil microbes along the elevational gradient belonged to rare taxa (Figure 1b,c). More importantly, bacterial rare taxa displayed a broader habitat niche breadth and made a greater contribution to the major assembly processes (Figure 2a, Table S3). A possible explanation for this is that bacterial rare taxa are more susceptible to environmental filtering than abundant taxa [58]. And, the β diversity of bacterial rare taxa is more strongly controlled by stochastic processes (i.e., dispersal) and weak environmental filtering [59], we suspect that more rare species may be involved in dispersal events. The habitat niche breadth of abundant taxa was broader than that of rare taxa in the fungal community (Figure 2b), which suggests that fungal abundant taxa have greater versatility in metabolism than rare taxa. The result is supported by the fact that the fungal abundant taxa expressed substantially broader response thresholds as well as stronger phylogenetic signals for the ecological preferences along environmental gradients compared to rare taxa [60]. Yet, as bacteria exhibit superior functionality in resource utilization and metabolism [61], they tend to be more resilient to environmental stressors and less susceptible to species sorting processes than fungi [62]. Our result may provide evidence for this view, where the niche breadth of all bacterial taxa was wider than that of fungal taxa along the elevational gradient (Figure 2c).

4.2. Mechanisms for the Soil Microbial Community Assembly

Homogeneous selection and homogenizing dispersal were the two most important assembly processes along the elevational gradient in Daiyun mountain (Figure 3a,c), corresponding to our first hypothesis. The habitat at small spatial scales can only sustain a limited number of individuals [17], and since homogeneous selection and homogenizing dispersal can reduce β diversity [63,64], we speculate that microbial community along the elevational gradient of Daiyun mountain may exhibit lower β diversity. Biotic factors play an important role in microbial community assembly [3]. For instance, soil fungal richness was found to mediate the assembly processes of bacterial communities, with stochastic processes showing a negative correlation with increased fungal richness [65]. Meanwhile, Bahram et al. [66] observed a decrease in the species richness of forest fungal communities with higher elevation. Remarkably, the fungal richness decreased with increasing elevation at the elevation from 1000 to 1500 m (Table S4), while the relative importance of stochastic processes in bacterial community increased (Figure 3b), which was consistent with the above views. Deterministic processes, stochastic processes, or a trade-off between the two may control the microbial community assembly across different spatiotemporal scales [67,68,69]. In our study, the results show that the trade-off between deterministic and stochastic processes controlled the bacterial community assembly (Figure 3b), whereas deterministic processes were the major assembly processes driving the fungal community (Figure 3d). In general, the assembly mechanisms of soil microbial communities are a combination of scale-dependent and species-dependent characteristics.

Homogeneous selection and homogenizing dispersal were mediated mainly by soil available phosphorus, the effect of phosphorus availability on major assembly processes was stronger in fungal community than in bacterial community (Table S1). Available phosphorus content was significantly and positively correlated with the relative importance of homogeneous selection and homogenizing dispersal (Table S1), which is a possible explanation for the fact that decreased soil phosphorus availability can limit microbial growth and community structure [28]. In general, soil phosphorus is mainly the limiting factor of forest ecosystem primary productivity in tropical and subtropical regions [70]. Additionally, the biogeochemical cycle of phosphorus in vegetation is an essential component of the geochemical cycle of phosphorus in montane systems. Plant leaf phosphorus can remarkably affect the spatial distribution of soil available phosphorus [36], which indirectly influences the assembly of soil microbial communities. This may be responsible for the fact that soil phosphorus availability mediated the assembly of soil microbial community in subtropical forests. Moreover, soil phosphorus remains the major limiting element in Daiyun mountain [34]. The extremely low utilization of soil phosphorus in this area is primarily due to the vertical and horizontal migration of phosphorus across the soil profile and surface runoff [71]. In the future, supplementing soil phosphorus content could be used to investigate the impact of phosphorus nutrient cycling on microbial community assembly at finer levels on a local scale.

Various environmental factors played a pivotal role in driving changes in the microbial community. Notably, a significant effect of soil temperature on the bacterial community assembly was observed (Table S1). This can be attributed to the fact that temperature influences the metabolic rate of microorganisms and their competitive interactions [72], thereby resulting in changes in the community assembly of microorganisms. In previous studies, soil pH has been a crucial factor affecting the structure of microbial communities [73,74]. However, our study investigated an elevational gradient on a local scale, and soil pH levels were relatively stable in the Daiyun mountain [34]. Therefore, we found that the soil pH had a low impact on the microbial community assembly along the elevational gradient (Table S1).

4.3. Assembly Processes of the Dominant Microbial Taxa

While most previous studies have focused on assembly processes at the community level, our study delves deeper by examining the mechanisms of microbial assembly at the individual level (Figure 4a,b). Of the all microbial bins, we selected five bins dominated by homogeneous selection and five bins dominated by homogenizing dispersal. These specific bins were identified as the dominant microbial taxa controlled by the major assembly processes (Table S3). We found that rare taxa played a greater role in the relative abundance of Group 2B and Group 2F, whereas abundant taxa had a stronger impact on the relative abundance of Group 1B and Group 1F (Table S3). This indicates that dominant rare taxa may have more individuals involved in dispersal events, while dominant abundant taxa may be more susceptible to environmental filtering along the elevational gradient. However, our findings are inconsistent with the conclusions of Liu et al. [58] and Jousset et al. [75]. One reasonable explanation is that our survey had a relatively restricted scope, as forest climate and topography have a notable impact on the spatiotemporal distribution of both abundant and rare taxa [76,77]. Furthermore, it is important to recognize that the construction of abundant and rare taxa is governed by different stochastic processes [77]. In short, microbial communities can be controlled by different ecological processes depending on the spatiotemporal scales of the study [78].

Most importantly, our results demonstrate that soil available phosphorus also drove changes in the dominant microbial taxa, through homogeneous selection and homogenizing dispersal (Table S1, Figure 4c). Additionally, the relative abundance of dominant microbial taxa increased significantly with increasing soil phosphorus availability (Figure 4c), supporting our second hypothesis. Water availability and land-use intensification can affect the biomass and function of soil microorganisms [79]. This rationale may provide support for our result, as the soil water content is an important driver of dominant bacterial community composition across elevations (Figure 4c). Based on the findings of Baldock et al. [80] and Carletti et al. [81], it has been suggested that soil mineral assemblages contribute to the stability and recalcitrance of soil carbon, rendering it inaccessible to soil bacteria and thereby selecting for specific bacterial taxa. This may explain the negative correlation observed between TC and dominant bacterial taxa (Figure 4c). The unique montane climate creates a habitat for the majority of species and plays a key role in regional biodiversity [24]. Typically, the generation and maintenance of montane biodiversity are mainly influenced by the regional area, productivity, and topography [24]. Our findings show that the slope and soil temperature had strong effects on microbial abundance and community structure (Table S1, Figure 4c).

4.4. Limitations and Prospects of the Study

While our study is grounded in elevational gradients within a relatively restricted local scale, it is an indicative investigation of subtropical forests. Future work should focus on proving a causal relationship between soil phosphorus availability and microbial assembly at a regional scale. Furthermore, several potential limitations should be considered in the context of the present study. Firstly, forest bacterial communities are primarily affected by soil abiotic variables, whereas fungal communities are mainly mediated by plant biomass [82]. It is necessary to consider that tree species richness has a vital impact on soil microbial diversity in subtropical forests [26]. In addition, understory vegetation not only fosters a positive interaction with microbial communities but also plays an important role in maintaining ecosystem stability [83]. Climate-vulnerable cryptogamic plants in montane cloud forests may also alter the structure and function of fungal community [84]. For example, the sampling site at the elevation of 1100 m lies in a C. lanceolata forest, which likely leads to enhanced deterministic processes in the soil fungal community (Figure 3d). Secondly, the diversity of animals in the forest ecosystem is known to be much higher than that in other terrestrial ecosystems [85]. The soil fauna and wildlife’s habits can change the microbial community assembly, and the microenvironment in soil fauna could also affect the surrounding soil microorganisms [86]. In a forthcoming study, investigating microbial community assembly will require considering biotic factors, physical factors, and biological correlation networks to more fully elucidate the ecological functions of the community.

5. Conclusions

Our study sheds light on the assembly mechanisms of soil microbial community along an elevational gradient in subtropical forests. At a local scale, homogeneous ecological processes play a major role in microbial community assembly. Soil phosphorus availability mediated the relative importance of homogeneous selection and homogenizing dispersal in bacterial and fungal community assembly. Specifically, the trade-off of deterministic and stochastic processes alters microbial β diversity by adjusting the relative abundance of dominant microbial taxa depending on soil phosphorus availability. Our findings represent an important advancement in revealing the underlying principles of microbial community assembly in subtropical forests by the assessment of environmental factors affecting the major assembly processes, which provides essential insights into the microbial distribution patterns. In the future, we need to investigate the nutrient cycling of phosphorus and the responses of plant and animal factors to soil microbial community assembly, which is crucial for understanding the generation and maintenance of forest biodiversity.