Impacts of Human Recreational Disturbances on Soil Bacterial Community Composition and Diversity in Urban Forest in Changchun, Northeast China

Abstract

Urban parks, as vital components of urban green infrastructure, can improve urban ecological environments, showcase urban culture, and offer spaces for human recreation and exercise. However, human activities in these parks also produce severe disruption to soil ecosystems. Studying the effects of recreational disturbances on soil properties and microbial communities is crucial for conserving urban biodiversity and maintaining ecosystem services. This study investigated the effects of human recreational disturbances (HRDs) on soil physicochemical properties and bacterial communities in four forest stands (Phellodendron amurense Rupr (Phe amu), Salix matsudana Koidz. (Sal mat), Pinus tabuliformis var. mukdensis (Pin tab), and Picea asperata Mast. (Pie asp)) in Changchun’s South Lake Park. The results showed that HRD significantly reduced soil water content (SWC) and total phosphorus (TP) while increasing soil bulk density (SBD) and pH. Soil organic carbon (SOC) and total nitrogen (TN) increased in Phe amu and Pie asp soils but decreased in Sal mat and Pin tab soils (p < 0.05). Electrical conductivity (EC) changes were inversely related to SOC and TN trends. Dominant bacterial phyla included Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi. HRD reduced bacterial species richness and diversity by 5.3% and 7.6%, respectively. SWC and SBD were key factors influencing bacterial community dynamics, with SBD affecting Bacteroidota, Proteobacteria, and Myxococcota, and SWC impacting Proteobacteria, Bacteroidota, and Actinobacteriota. These findings provide insights for urban park management, supporting soil microbial diversity and sustainable urban ecosystem development.

1. Introduction

As an important component of urban green infrastructures, urban forest plays vital roles in improving urban environment and promoting the sustainable development of the urban ecosystem [1]. It plays obvious roles in purifying local air quality [2,3], improving urban microclimate [4,5], creating urban recreational spaces [6], and enhancing the quality of life of urban residents [7]. With the acceleration of global urbanization processes and the growth of urban populations, people’s demand for a healthy and ecological environment in cities is increasing. Comprehensive and resident-friendly urban parks have become crucial venues for citizens’ leisure and recreation.

At the same time, the trampling and compaction caused by human recreational activities have severely affected the soil health of forest stands in parks, including soil physicochemical properties, soil microbial community composition and diversity, etc. Previous studies have demonstrated that grassland that has been heavily trampled exhibits 15%–30% higher soil bulk density and 20%–50% lower water content compared with undisturbed sites [8]. In addition, the heavy soil compaction caused by human activities also can lead to a significant reduction in soil porosity and an increase in soil bulk density (SBD), which in turn restrict the growth of plant roots. In the event of heavy rainfall, such soil property changes are prone to inducing surface runoff and elevating the risk of soil erosion [9]. Moreover, some studies have demonstrated that both anthropogenic and faunal disturbances can accelerate the decomposition of soil organic carbon (SOC), resulting in SOC content reduction [10,11], and induce pH decrease and salinity accumulation with elevated electrical conductivity (EC), which may subsequently modify soil microbial habitats by altering nutrient dynamics [12]. These changes ultimately disrupt the soil’s biogeochemical cycles within urban forest ecosystems [13].

Soil bacteria are the most dynamic part of soil ecosystems [14], accounting for 70%–80% of the total soil microorganisms [15]. Soil microbial composition, community structure, and diversity characteristics are important indicators for ecosystem stability. As a pioneering indicator of ecosystem restoration processes, soil microorganisms are extremely sensitive to changes in the soil micro-environment [16]. Human recreational disturbances in urban parks can seriously affect the physical and chemical properties of soil, which in turn affects the composition and diversity of soil microbial communities [17]. It has been shown that microbial diversity is significantly affected by environmental factors, such as forest vegetation type, soil type, temperature, and soil pH at landscape and regional scales [18,19,20,21]. Studies on the disturbance of urban forest soil by human recreational activities have mainly focused on alterations in soil physicochemical properties [22,23,24,25], but few studies have been conducted on the recreational disturbances of soil bacterial community composition and diversity in urban parks. Zheng et al. demonstrated that urban forest soils harbor distinct bacterial communities compared to natural forests, yet the specific impacts of recreational disturbances remain poorly quantified [26]. Similarly, Christel et al. identified significant variations in microbial abundance across different urban land uses, but noted limited understanding of how recreational pressures specifically reshape bacterial community assembly [27]. This understanding remains particularly limited in northeastern China’s urban ecosystems, especially given the region’s rapid urbanization and unique climatic conditions.

This study investigates the effects of human recreational disturbances on soil properties and bacterial communities across four dominant forest stands in Changchun’s South Lake Park, representing a typical urban ecosystem in Northeast China. Three hypotheses were put forward: (1) recreational disturbances significantly alter key soil physicochemical properties; (2) these disturbances reduce bacterial richness and diversity while reshaping community composition, with effects exceeding those observed in less intensively used urban green spaces; and (3) forest stand identity mediates disturbance impacts, reflecting species-specific influences on soil–microbe interactions within urban environments. Our objectives specifically address the scarcity of mechanistic understanding about recreational impacts on soil microbiomes in Northeast China’s unique cold urban ecosystems.

2. Materials and Methods

2.1. Study Area

The city of Changchun (43°05′~45°15′ N, 124°18′~127°05′ E) is located in the hinterland of Northeast China, within the transition zone between humid and semi-arid climate regions. (Figure 1) It has a temperate continental humid climate. The multi-annual average temperature is around 5.48 °C, and the annual precipitation is 400~1000 mm [28]. The soils in Changchun exhibit a spatial variability from east to west, forming a gradient dominated by black soil, meadow soil, black calcium soil, and dark brown soil [29].

South Lake Park (43°51′ N, 125°18′ E), located in the center of the main urban area of Changchun, was established in 1933 and spans approximately 2.38 km². It is the largest urban park in Northeast China. The forest area is about 1.26 km², accounting for about 56.8% of the total area [30]. With up to 200,000 visitors per day during the peak tourist season, South Lake Park is known as the most popular urban forest park in Changchun [30].

2.2. Plot Layout and Soil Sample Collection

Field surveys and soil sampling were conducted in June 2023. Four representative paired forest stands which experienced severe human recreational activity-disturbed areas (HRDs) and human recreational activity-undisturbed areas (HRUDs) were selected in South Lake Park, Changchun, including forest stands dominated by Phellodendron amurense Rupr (Phe amu), Salix matsudana Koidz (Sal mat), Pinus tabuliformis var. mukdensis (Pin tab), and Picea asperata Mast (Pie asp). Each sampling plot area was 100 m². Following standard forest soil sampling protocols, soil samples were collected at a 1.5 m distance from tree bases to avoid rhizosphere effects and obtain representative soil [31].

Disturbance classification was determined by quantitative soil bulk density (SBD) measurements combined with diagnostic field characteristics. Sites designated as human recreation-disturbed (HRD) displayed clear evidence of soil compaction (SBD: 1.45–1.68 g/cm³), accompanied by visible trampling traces and the absence of understory herbaceous vegetation. In contrast, human recreation-undisturbed (HRUD) sites showed no signs of human trampling, maintained intact herbaceous layers, and exhibited significantly lower SBD values (1.12–1.28 g/cm³). A threshold of >1.35 g/cm³ was established to differentiate disturbed conditions, representing a 20% increase above the mean baseline value of undisturbed sites.

Two complementary sampling approaches were employed: (1) SBD determination using the cut-ring method with steel rings (volume: 100 ± 2 cm³; height: 5 cm) inserted vertically at 0–5 cm depths after removing surface debris; (2) comprehensive soil sampling using a five-point method with a 50 mm diameter auger at 0–20 cm depth for physicochemical and microbial analyses. Each sampling point included three replicate ring cuts (n = 24, which was equal to 4 stands × 2 treatments × 3 replicates). Soil samples obtained through the five-point sampling method were processed to remove gravel and roots, then divided into two equal portions—one was air-dried in a cool, ventilated area for physicochemical analysis, while the other was stored at −80 °C for microbial community profiling.

2.3. Determination of Soil Physicochemical Properties

For soil physicochemical properties, the soil samples were sieved through 10-mesh and 60-mesh sieves. Soil bulk density (SBD) was determined using the cut-ring method. Soil cores were collected, dried at 105 °C, and then the SBD was calculated based on the mass-to-volume ratio of the dry soil. Soil water content (SWC) was determined by the desiccation method in an oven at 105 °C [32]. Soil pH was determined using the potentiometric method (PHS-3E pH meter), and the water-to-soil ratio was 2.5:1. Soil electrical conductivity (EC) was determined with the potentiometric method (DDS-11A conductivity meter), and the water-to-soil ratio was 5:1 [33]. Soil organic carbon (SOC) content was determined using the potassium dichromate external heating method [34]. Total nitrogen (TN) and total phosphorus (TP) content were determined by the sulfuric acid–hydrogen peroxide digestion method and then measured using a continuous flow analyzer (AA3, Germany) [35].

2.4. Determination of Soil Microbial Composition and Diversity

Soil bacterial species identification was performed using 16S gene sequencing. The 16S rRNA gene databases for bacteria were obtained from SILVA (Release 138, http://www.arb-silva.de (accessed on 15 August 2023)), RDP (Release 11.5, http://rdp.cme.msu.edu/ (accessed on 18 August 2023)). Screening for potential bacterial pathogens at the genus level (using a 1% relative abundance threshold) revealed no significant pathogenic taxa in any samples, indicating minimal public health concerns in the studied areas.

Taxonomic classification was conducted using the QIIME platform (http://qiime.org/scripts/assign_taxonomy.html (accessed on 16 August 2023)) and the RDP Classifier (version 2.11, http://sourceforge.net/projects/rdp-classifier/ (accessed on 18 August 2023)), with a confidence threshold of 0.7. Taxonomic assignment employed the following identity thresholds based on RDP Classifier recommendations for 16S rRNA gene fragments: species-level (≥97% similarity), genus-level (≥94%), family-level (≥90%), and phylum-level (≥80%).

PCR amplification was performed using the TransGen AP221-02 TransStart FastPfu DNA Polymerase on an ABI GeneAmp^® 9700 PCR System (Applied Biosystems, Foster City, CA, USA). All samples were processed under the same experimental conditions, with three technical replicates per sample. PCR products from the same sample were pooled and analyzed by 2% agarose gel electrophoresis. Target bands were excised and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), followed by elution with Tris-HCl buffer. Quality control measures included (1) negative extraction controls using sterile water instead of a soil sample; (2) positive controls with ZymoBIOMICS Microbial Community Standard (Irvine, CA, USA); and (3) no-template PCR controls to monitor contamination. Only samples with control results within expected parameters were included in downstream analysis.

Illumina MiSeq sequencing generated an average of 48,532 raw paired-end reads per sample (range: 41,203–55,891). After quality filtering (Q-score ≥ 20), adapter removal, and sequence splicing based on overlap relationships, there was an average of 38,416 high-quality reads per sample (range: 32,700–45,626). The mean read length was 420 bp (median: 415 bp) after primer trimming. For diversity analyses, samples were rarefied to 10,000 reads per sample using the ‘vegan’ package in R (v2.6-4) to ensure even sequencing depth. Rarefaction curves (see Section 3.2.1 for details) confirmed adequate sequencing coverage, with all curves reaching saturation.

OTU clustering was performed on the UPARSE platform (v7.0.1090, http://drive5.com/uparse/ (accessed on 18 August 2023)). Chimeras were identified and removed using VSEARCH (v2.22.1) with the ‘uchime_denovo’ algorithm under default parameters. Sequences identified as chimeras against the SILVA reference database (v138) were excluded from subsequent analysis. After sample differentiation, OTU clustering and taxonomic classification were conducted, followed by multi-diversity index analysis based on OTUs. Alpha diversity indices, including the Shannon index and Chao index, were calculated to assess species richness and diversity using mothur (v.1.30.2, https://mothur.org/wiki/calculators/ (accessed on 22 August 2023)). The OTU similarity threshold for index evaluation was set at 97% (0.97). Hierarchical clustering analysis was performed based on the beta diversity distance matrix, using the UPGMA (Unweighted Pair Group Method with Arithmetic Mean) algorithm to construct a dendrogram, which visually represents the degree of dissimilarity in community distribution among samples.

2.5. Data Statistical Analysis

The average values and standard deviations of the triplicate experimental results for each sampling point were calculated using Microsoft Excel 2019. Post hoc multiple comparison tests were conducted: the LSD (Least Significant Difference) method was used when the variance was homogeneous (χ²), while Tamhane’s T2 method was applied when the variance was heterogeneous. The effect of disturbance from human recreational activities on soil physicochemical properties was analyzed using a paired samples t-test. The above analyses were performed in SPSS 23.0 (SPSS, Chicago, FL, USA). In the Majorbio platform, Venn diagrams were generated and statistically analyzed using the R language (version 3.3.1). The Circos diagrams illustrating the relationships between samples and species were created using Circos-0.67-7 (http://circos.ca/ (accessed on 25 August 2023)). Heatmap visualization of microbial communities was performed using the heatmap package (version 1.0.8) in the R language (version 3.3.1). Redundancy analysis (RDA) was conducted using the rda function from the vegan package (version 2.4.3) in R. Partial Least Squares Structural Equation Modeling (PLS-SEM) was performed using the R programming environment with the complete dataset (n = 24 biological replicates), providing adequate statistical power for model estimation.

3. Results

3.1. Effects of Human Recreational Activity Disturbances on Soil Properties

Under disturbance by human recreational activities, SWC was significantly decreased by 52.6%, 81.2%, and 65.07% (p < 0.05) in the soils of Phe amu, Pin tab, and Pie asp stands, respectively, except for a 1.3% increase in the Sal mat stand compared with HRUD forest soils (Figure 2a). But, the SBD was significantly increased in all four forest stands (p < 0.05), with the following orders of Pie asp (62.1%), Phe amu (42.9%), Pin tab (41%), and Sal mat (13.1%) (Figure 2b).

For the HRD forest soil, TP content was slightly decreased compared with that in the HRUD soil of the Pin tab stand (p > 0.05), while it was significantly increased in the other three forest stands (Figure 2c; p < 0.05). However, the soil TN content and SOC content both significantly increased in the Phe amu and Pie asp stands, and decreased significantly in the Sal mat and Pin tab forest stands. Soil EC values had the opposite trend, which significantly increased in the Sal mat and Pin tab and decreased in the Phe amu and Pie asp forest stands (p < 0.05), respectively (Figure 2f). Soil pH values increased slightly in all forest stands (Figure 2g).

Human recreational disturbances differentially impacted soil properties across forest stands (Phe amu, Pin tab, Pie asp, and Sal mat). Pie asp exhibited the most severe alterations: a 62.1% increase in SBD, a 172.3% increase in SOC, and a 7.1% increase in pH (all p < 0.05). Sal mat showed minimal changes (a 1.3% increase in SWC and a 0.04% increase in pH). Intermediate responses occurred in Phe amu (a 52.6% decrease in SWC and a 60.2% increase in TN) and Pin tab (an 81.2% decrease in SWC and a 135.2% increase in EC). These species-specific responses necessitate tailored management approaches.

3.2. Effects of Human Recreational Activities on the Composition and Diversity of Soil Bacterial Communities

3.2.1. Analysis of Dilution Curves and Shannon Wiener Curves at OTU Levels

The results of the Shannon-Wiener curve and the dilution curve showed that most of the microbiome can be obtained from all samples with the available sequencing volume (Figure 3). The Shannon-Wiener curve was more pronounced to represent the bacterial species observed in the soil sample (Figure 3b). When the number of randomly selected sequences reached 10,000, the curve approached a horizontal trend, indicating that the measured sequences could basically cover the bacterial species in the soil. The Shannon–Wiener and dilution curves indicated that the sequencing depth was sufficient to capture the majority of bacterial species in the soil samples, with adequate coverage confirmed.

3.2.2. Effects of Human Recreational Activity Disturbances on the OTU Composition of Soil Bacteria

The number of soil bacterial OTUs in the HRUD of the Phe amu, Sal mat, Pin tab, and Pie asp soils was 7.6%, 11.5%, 14.2%, and 11.5% higher than that in the HRD, respectively (Figure 4). HRUD detected higher numbers of OTUs than HRD in all four stand soil samples (Figure 4).

3.2.3. Impacts of Human Recreational Activities Disturbances on Soil Bacterial Community Composition

Overall, a total of 38 phylum of bacteria were detected in HRD samples and HRUD samples (Figure 5). Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi were the dominant groups in HRD samples, which accounted for 38%, 16.8%, 14.5%, and 12.9% of the total bacteria. However, in the HRUD samples, the proportions of Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi were 27.7%, 21.8%, 18.7%. and 11.8%, respectively. The combined relative abundance of the four most abundant phyla (Actinobacteria, Proteobacteria, Acidobacteria, and Chlorobactomia) accounted for 78.8% of all sequences. Under disturbance by human recreational activities, Actinobacteriota increased by 37.2%, Chloroflexi increased by 8.7%, while Proteobacteria and Acidobacteriota decreased by 22.9% and 22.8%, respectively. Under disturbances by human recreational activities, the Actinobacteriota communities increased by 42%, 19.5%, 33%, and 54% in the Phe amu, Sal mat, Pin tab, and Pie asp forest soils, respectively (Figure 5). The Proteobacteria communities decreased by 31%, 17%, and 32% in Phe amu, Pin tab, and Pie asp, and with no change observed in Sal mat. The Acidobacteriota communities decreased by 17%, 75.7%, and 35% in the soils of Phe amu, Pin tab, and Pie asp, while they increased by 7% in Sal mat soil. The Chloroflexi communities decreased by 35% and 2.4% in the soils of Phe amu and Pin tab, but increased by 12.9% and 45% in Sal mat and Pie asp soils compared with the HRUD soils (Figure 5).

Statistical Analyses of Metagenomic Profiles (STAMPs), using Welch’s test, showed that the OTU of HRD soil differed obviously from that of HRUD soil at the phylum level of the four different forest stands; the abundance changes in OTUs of the top 15 phylum varied between the two groups (p < 0.05; Figure 6). For HRD forest soil, the OTU abundances of Bacteroidota, GAL15, and Bdellovibrionota in the Phe amu stand were significantly reduced by 57.4%, 60.9%, and 50.9%, respectively (Figure 6a). Meanwhile, the abundances of the OTUs of Acidobacteriota, Methylomirabilota, Verrucomicrobiota, Planctomycetota, Latescibacterota, NB1-j, Elusimicrobiota, and MBNT15 in the Sal mat stand was significantly reduced by 43.1%, 48.9%, 78.9%, 73%, 67.6%, 31.2%, 77.3%, and 55.5% (Figure 6b), respectively. Furthermore, in the Pin tab forest stand, the abundances of OTUs of Myxococcota, Bacteroidota, Latescibacterota, Desulfobacterota, Cyanobacteria, Dependentiae, NB1-j, MBNT15, Fibrobacterota, and SAR324_cladeMarine_group_B were significantly reduced (Figure 6c). And the abundance of OTUs in phyla of Actinobacteriota and Firmicutes were significantly higher than in the undisturbed group. The abundances of the OTUs of Proteobacteria, Bacteroidota, Myxococcota, Patescibacteria, Patescibacteria, Bdellovibrionota, Dependentiae, Elusimicrobiota, MBNT15, and WPS-2 in the Pic asp forest soil were significantly reduced, but Proteobacteria and Dependentiae increased by 135.5% and 177%, respectively (Figure 6d; p < 0.001).

3.2.4. Effects of Human Recreational Disturbance on Soil Bacterial Diversity

Human recreational disturbance induced a decreasing trend of bacterial diversity expressed as the Shannon index and Chao index in the four HRD forest stands (Figure 7a,b). The Chao indexes of the Pin tab and Pic asp soils were significantly decreased by 9.46% and 8.41% under human recreational disturbances (p < 0.05; Figure 7a). The Shannon’s indexes of the Pin tab and Sal mat soils were also significantly declined by 9.13% and 5.35% (p < 0.05; Figure 7b), respectively.

3.3. Correlation Between Soil Properties and Soil Bacterial Communities

3.3.1. Redundancy Analysis (RDA)

The results of RDA showed that Axes 1 and Axes 2 explained 37.05% and 5.00% of the total variations (Figure 8). It can be seen that SWC (r = 0.87097, p ≤ 0.001) had the greatest influence on the abundance of soil bacterial communities, followed by the SBD (r = 0.70213, p ≤ 0.05) samples of the four forest stands. SOC (r = 0.47222, p = 0.063), pH (r = 0.3834, p = 0.182), TP (r = 0.35972, p = 0.23), EC (r = 0.32434, p = 0.349), and TN (r = 0.14866, p = 0.802) had a lesser degree of influence on the soil bacterial communities of the four forest stands.

Under human recreational disturbance (HRD), soil microbial composition exhibited greater diversity and more dispersed distribution patterns compared to undisturbed soils (HRUD). In HRUD forest stands (Phe amu, Pie asp, and Sal mat), bacterial communities were primarily governed by SWC (r² = 0.7586, p = 0.001) and EC, with the relative abundances of Proteobacteria and Acidobacteriota showing particularly strong correlations with SWC. In contrast, HRD stands (Phe amu, Pin tab, and Pie asp) demonstrated significant changes in community structure driven by SBD (r² = 0.493, p = 0.002), SOC, and pH.

3.3.2. PLS-SEM Analysis of Recreational Disturbance Impacts on Soil Microbial Systems

Structural equation modeling revealed that human recreational disturbance exerted significant direct negative effects on soil bacterial communities (β = −0.865, p < 0.001), which subsequently reduced bacterial diversity (β = −0.674, p < 0.001). While disturbance significantly altered both chemical (β = 0.490, p < 0.05) and physical properties (β = 0.477, p < 0.05), these abiotic factors did not directly influence bacterial diversity (chemical: β = 0.194, p = 0.144; physical: β = 0.149, p = 0.292). The model explained 74.9% of the variance in the bacterial communities and 70.9% of the diversity.

4. Discussion

4.1. Effects of Human Recreational Activity Disturbances on Soil Properties

The impact of human recreational disturbance on urban forest soils can vary between forest stands, as observed in this study. Previous studies have shown that human recreational activities can significantly influence soil physicochemical properties, particularly through the assessment of soil compaction using bulk density [36]. These disturbances often lead to reduced SWC, which can in turn induce greater soil hardness [37]. The SWC usually decreases with the intensification of human recreational activities [38]. The results of our study of Phe amu, Pin tab, and Pie asp forest soils were consistent with the results mentioned above (Figure 2a), which indicates that disturbance by human recreational activities leads to a declining trend in soil’s water-holding function [39]. Previous studies on Salix showed that its high biomass production and extensive root systems were important factors that influenced the physical, chemical, and biological properties of the soil. Root systems consist of dense fine roots that maintain and improve soil conditions through nutrient cycling process, reducing soil erosion and stimulating soil biodiversity [39,40]. Therefore, the SWC of Sal mat soils with dense roots may exhibit distinct responses compared to the other three forest stands under human recreational disturbances in our study, as it was located near water bodies providing natural hydrological buffering against the impacts of recreational activities. In our study, we found that disturbance by human recreational activities caused significant increases in soil compaction (Figure 2b). The increased SBD can indirectly result in reduced soil porosity, increased soil compaction, and impaired soil aeration. Cilimburg’s research showed that human trampling significantly creased soil pH [41]. Soil EC reflects the concentration of soluble ions in the soil, which is increased by human recreational activities [42]. In this study, the results further confirm that parks with higher disturbance intensities in terms of human recreational activities had higher soil EC and pH values, which is consistent with previous studies. However, the contents of TP, TN,, and SOC were significantly increased in this study. This variation may be attributed to the management and maintenance measures for the forests in the park and some exogenous substances input in the process of human recreational activities. The characteristics of soil properties in the Pie asp stand exhibited the most obvious changes, primarily due to its location near the park entrance, with frequent human activities including group dancing, singing, and fitness exercises. In contrast, the Pin tab and Phe amu stands experienced moderate changes mainly owning to some disturbance from camping and tai chi practice. The Sal mat stand showed relatively minor impacts, with disturbances limited mainly to trail formation from foot traffic. Recreational disturbances significantly altered soil physicochemical properties, manifesting as reduced SWC and increased SBD, TP, and pH, with divergent responses in TN and SOC across the forest stands. This is the same as assumptions 1 and 3.

4.2. Effects of Human Recreational Activities on the Composition and Diversity of Soil Bacterial Communities

The composition and diversity of soil bacterial communities was dramatically changed under the influence of human recreational activities. In this study human recreational disturbances lead to the soil bacterial OTU richness decreasing across all forest stands (7.6%–14.2%), suggesting that trampling may simplify microbial community structure through physical compaction and habitat fragmentation (Figure 4). The results were consistent with the findings of Qianqian Liu et al. In all forest stands, Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi were the dominant groups (Figure 5). The results were similar to those for the forest soils in Dinghu Mountain [43] and Sanjiang Plain [27]. The dominant phyla in the present study—Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi—exhibited relative abundances and ecological influences consistent with prior reports [44,45,46,47], confirming their stable functional roles across varied soil environments.

One previous study indicated that the abundance of Proteobacteria was positively correlated with soil carbon content [48], and its proportion was higher in more nutrient-rich soils, which is similar to the results of this study. Under the influence of human recreational activities, the abundance of most microorganisms declined, while the number of operational taxonomic units (OTUs) of certain specific species showed an increasing trend. This indicates that human recreational activities lead to an overall reduction in the total number of soil microbial bacterial taxa and their diversity, but the impacts on soil bacterial community composition were not uniform. Specifically, the changes observed reflect a shift in the community structure rather than a decrease in the number of distinct communities.

While 16S rRNA amplicon sequencing reveals taxonomic shifts, it cannot directly quantify functional gene abundances or metabolic activities. Proteobacteria are known to play a significant role in organic matter decomposition [49], and their increased abundance in disturbed soils may reflect their adaptability to the nutrient-rich conditions created by organic matter inputs and the soil structure altered by to human disturbances [48,50,51,52]. Similarly, Actinobacteriota dominated in disturbed soils likely owning to their tolerance to low-moisture conditions caused by increased SBD [53]. The decline of Acidobacteriota, known for their role in oligotrophic metabolism and complex carbon degradation [54,55,56], coupled with the enrichment of copiotrophic Actinobacteriota, suggests a fundamental shift in carbon cycling pathways. In contrast, increases in Actinobacteriota, which preferentially metabolize labile carbon, could shift decomposition toward more rapid carbon turnover, influencing long-term soil carbon sequestration. Within the nitrogen cycle, reduced Acidobacteriota abundance may diminish nitrification activity, whereas the rise in denitrifying Proteobacteria could promote N₂O production, particularly in the anaerobic microsites resulting from recreational compaction [57]. Chloroflexi’s persistence may be attributed to their anaerobic adaptability in compacted soils [48,58]. These abundance shifts directly correlate with trampling-induced physicochemical changes (increased SBD, decreased SWC, and pH alterations) [59,60].

The reduced microbial diversity observed in disturbed soils signals impaired ecosystem functionality. In this study, human recreational disturbances consistently reduced bacterial diversity across all forest stands, as evidenced by the significant decreases in the Chao1 index (Pin tab: −9.46%; Pie asp: −8.41%) and Shannon index (Pin tab: −9.13%; Sal mat: −5.35%). These diversity reductions were accompanied by taxon-specific abundance shifts among the dominant phyla, ultimately altering microbial community composition and biomass. Such disturbances may consequently impair plant productivity through modified microbe-mediated nutrient cycling processes [61]. Reduced microbial diversity impairs organic matter decomposition [62], potentially decreasing CO₂ emissions, and diminishes potential denitrification activity in nitrogen cycling [63]. These changes may ultimately compromise soil nutrient cycling services and ecosystem health. Lower diversity diminishes metabolic versatility, potentially weakening organic matter decomposition and nutrient cycling services. Dominance shifts toward stress-tolerant phyla (e.g., Proteobacteria and Actinobacteriota) at the expense of acidophilic specialists (e.g., Acidobacteriota) reflect habitat filtering under recreational pressure. Such homogenized communities may decrease ecosystem resilience to further disturbances, as evidenced by studies linking diversity loss with reduced functional redundancy in soils [64,65]. Future studies should explicitly quantify these diversity–function relationships to predict long-term impacts on urban forest sustainability. These disturbances reduced bacterial richness (5.3% average decline) and diversity (7.6% average decline), as evidenced by consistently lower OTU numbers in the HRD samples across all stands. This is consistent with hypothesis 2.

4.3. Correlations Between Soil Properties and Soil Microorganisms

SWC is an important factor affecting microorganisms and is essential for maintaining normal microbial metabolism [66]. Changes in soil moisture conditions can affect oxygen content and substrate availability, which in turn affects the microbiota [67]. In this study, soil bacterial community structure was significantly affected by SBD, pH, SOC, and TN. According to a recent study by Zhang et al., SBD, pH, and SOC were identified as the dominant factors shaping bacterial community structure, which aligns well with our current findings [68]. This suggested that the changes in bacterial community structure were partially attributed to the changes in soil properties, which was consistent with the results of this study (Figure 8). SBD showed a significant increasing trend with the increasing intensity of human recreational activities. However, with the increase in the intensity of human recreational activities, the content of soil powder particles, clay particles, and the number of macroaggregates decreased, and the porosity of the surface soil layer decreased, which would induce a decrease in SWC [69]. The redundancy analysis revealed that SWC and SBD had the greatest influence on the soil bacterial species of the four stand types, and SOC, pH, TP, EC, and TN had a lesser influence on the soil bacterial species of the four stand types (Figure 8). This suggests that soil SBD, SWC, and other physical and chemical properties are not only the main environmental factors affecting plant growth and development, but are also important factors affecting the composition and community structure of soil bacteria phyla. Among these, the significant correlation with Bacteroidota was extremely strong [70]. Strains in the phylum Bacteroidota act as major decomposer of organic matter, which can increase the soil organic carbon content and provide energy for microbial growth and soil enzyme activity [71]. SBD was more significant with Proteobacteria phylum and usually there was a negative correlation between SBD and microbial abundance [72,73]. SWC was significantly positively correlated a with Proteobacteria and Bacteroidiota, while being negatively correlated with Actinobacteriota. It has been shown that Proteobacteria and Actinobacteriota are mainly involved in organic matter decomposition [39]. The relative abundance of Proteobacteria was positively correlated with soil carbon content [74], and the proportion was higher in more nutrient-rich soils [43].

Human recreational activities in this study exerted significant direct negative effects on soil bacterial communities, consequently reducing bacterial diversity (Figure 9). Liu et al. demonstrated that in highly disturbed areas, functional selection (e.g., pollutant degradation) leads to synchronized shifts in microbial community structure and function, whereas in less disturbed areas, functional attributes remain relatively stable despite compositional changes [75]. Although human activities and climate change indirectly affect microbial communities by altering soil physicochemical properties, these abiotic factors did not directly reduce bacterial diversity, indicating a degree of microbial adaptive capacity [76].

In conclusion, recreational activities primarily impact soil bacterial communities through environmental filtering and functional selection rather than through changes in soil properties. This mechanistic understanding provides a scientific basis for soil ecological conservation, highlighting the need to mitigate direct human-induced disturbances to microbial habitats.

4.4. Limitations and Future Research Directions

While this study provides insights into recreational impacts on soil bacterial communities, several limitations should be noted. Our sampling captured only a single time point, limiting insight into how microbial communities respond to seasonal changes in moisture and temperature. In addition, we focused solely on bacteria, leaving open questions regarding the response of fungi, archaea, and viruses—each integral to nutrient cycling. Furthermore, taxonomic data derived from 16S rRNA amplicon sequencing do not directly reflect functional traits or metabolic activity.

Future work should adopt multi-season sampling to resolve temporal dynamics, apply shotgun metagenomics to characterize functional genes, and incorporate enzyme activity assays to quantify process rates. Expanding sequencing efforts to include fungal ITS and archaeal 16S markers would offer a more integrated perspective. Together, these approaches would help clarify how recreational pressure influences the functioning of urban soil ecosystems.

5. Conclusions and Recommendations

Compared with forest soils undisturbed by human recreational activities, the SWC exhibited a significant decreasing trend under disturbance by human recreational activities, while the SBD, TP content, and pH values increased significantly. Soil TN and SOC content increased significantly in Phe amu and Pie asp soils, but decreased in the Sal mat and Pin tab soils. Soil EC values showed the opposite trend.

The dominant soil bacterial communities in this study were Actinobacteriota, Proteobacteria, Acidobacteriota, and Chloroflexi. The HRUD samples exhibited significantly higher OTU richness in soil bacterial communities compared to the HRD samples across all forest stands, with relative increases of 7.6% (Phe amu), 11.5% (Sal mat), 14.2% (Pin tab), and 11.5% (Pie asp), respectively. SWC and SBD had significant effects on soil bacterial community composition in the four forest stands. And SBD significantly affected the abundance of Bacteroidota, Proteobacteria, and Myxococcota in the soils, while SWC significantly affected the abundance of Proteobacteria, Bacteroidota, and Actinobacteriota.

Urban parks are windows for displaying urban culture and civilization that transmit regional culture. Therefore, in order to maintain the harmony of the ecological environment and human recreational activities, to prevent the destruction of habitats, and to coordinate the relationship between recreation and the environment, the following suggestions were put forward under the premise of guaranteeing the development of the ecological environment:

• Implement strategic zoning to separate high-intensity recreational areas from ecologically sensitive soil zones. Use permeable paving materials (e.g., anticorrosive wood or porous concrete) in activity zones to reduce soil compaction, and install clear signage to prohibit off-trail trampling.
• Integrate adaptive management practices: Limit visitor density in fragile soil areas through timed access or capacity controls. Regularly aerate compacted soils in high-use zones and apply organic mulch to restore microbial activity and nutrient cycling.
• Establish long-term monitoring protocols for soil health indicators, including bulk density, organic matter content, and microbial diversity. Use findings to adjust zoning boundaries or restrict activities in areas showing signs of degradation.
• Promote public engagement through educational campaigns, highlighting the link between soil biodiversity and ecosystem services (e.g., water filtration, carbon sequestration). Encourage community participation in soil restoration initiatives (e.g., native plantings).
• Integrate microbial conservation into urban planning. Recreational disturbances reduce soil bacterial diversity and impair microbial functions. Managing visitor access is therefore essential to preserving nutrient cycling and carbon storage services. Protecting soil microbial communities must become a priority in sustainable park management to maintain urban forest health.