Effects of Soil Microorganisms on Carbon Sequestration under Different Mixed Modification Models in Pinus massoniana L. Plantation

: In forests, microbial populations in the soil can directly influence the decomposition of carbon from surface plants, promoting carbon storage and stability. However, in sustainable forest management, it is still unclear how soil microorganisms under different plantation types affect organic carbon sequestration and whether the mechanisms of influence are the same. In this research, we focused on four mixed forests and pure Pinus massoniana -planted forest in the state-owned forest farm of Dushan County. Three replicated plots were set up for each model, and soil samples were collected from different layers (0–20 cm, 20–40 cm, and 40–60 cm), totaling 45 samples. We elucidated the effects of soil microorganisms on carbon sequestration under five mixed modification models of P. massoniana and further explored the mechanisms by which microbial functional communities regulate soil carbon sequestration under different mixed models through molecular sequencing and collinear network analysis. Variance analysis indicated that the soil organic carbon (SOC) of the same soil layer varied significantly, and there were also significant differences in the composition of soil bacterial and fungal microbial communities. Moreover, the bacterial community was more sensitive to changes in the vegetation environment, while the fungal community structure was more resistant to changes in the soil environment. Correlation analysis indicated that the diversity and composition of the bacterial community had more positive effects on soil organic carbon than those of the fungal community. Linear fitting and redundancy analysis (RDA) showed that particulate organic carbon (POC) in soil had the strongest correlation with SOC content. Soil microorganisms affected the storage and stability of soil carbon mainly by regulating the conversion of litter (carbon sources) into POC. The soil environment of different mixed models had different effects on soil carbon accumulation. Both correlation and collinearity network analyses indicated that soil microbial functional groups could enhance carbon storage by regulating readily oxidizable carbon (EOC) and POC content in mixed forest plantations. The results of our study provide a sound basis for replanting a reasonable forest model structure to improve forest carbon storage.


Introduction
Soil organic carbon (SOC) is a significantly important carbon store in forest ecosystems, with its content serving as a key indicator of soil fertility and quality.It is instrumental in regulating soil physicochemical properties, supplying nutrients, enhancing soil structure, and ensuring soil stability [1,2].Soil carbon is mainly composed of three components: active Forests 2024, 15, 1053 2 of 13 carbon, slow carbon, and inert carbon [3].Among them, labile organic carbon is more susceptible to decomposition and transformation than the slow-release and inert forms.This allows it to swiftly react to external environmental shifts.Slow-release and inert carbon play a pivotal role in storage and stabilization [4,5], and the mutual conversion mechanism between soil carbon components regulates the stability of the forest soil carbon pool [6].Expanding forests through artificial reforestation is consistently regarded as an effective strategy for enhancing SOC sequestration [7].Pure artificial coniferous forests, with their high productivity and straightforward management, dominate as the primary model of artificial reforestation in China [8].However, low-quality, monoculture coniferous forests can lead to ecological issues, including loss of species diversity, prevalent pests and diseases, and reduced productivity (decreased carbon sequestration capacity) [9].Studies indicate that intermingling pure artificial coniferous forests with native broad-leaved species can boost the forest ecosystem's carbon sequestration capacity [10,11].This is attributed to the varied impacts of different tree species' litter and root exudates on soil physicochemical properties and microbial community structures, which in turn influence changes in SOC content [12,13].Consequently, establishing mixed coniferous and broad-leaved forests is emerging as one of the most promising forest management strategies globally [14].
Soil microbes play an important role in the forest soil carbon cycle [15].Their metabolites play a key role in maintaining the versatility of ecosystems, for example, through soil physical structure and SOC sequestration [7].Soil carbon is the outcomes of the balance between microbial assimilation and decomposition.Forest ecosystems typically manage the fluctuations in soil carbon storage by influencing the formation of soil organic matter, microbial growth and metabolism, as well as the mineralization and cycling of nutrients [16].However, research on how soil microorganisms influence soil carbon storage after the replanting of broadleaf tree species is limited.Thus, clarifying the coupling relationship among "mixed model, soil microorganisms, and soil organic carbon" is essential for a profound understanding of the mechanisms driving carbon storage and for scientifically assessing forest soil carbon sinks/sources.P. massoniana is the most widely distributed and largest resource of artificial afforestation species in Guizhou Province, China.It is also the main object of adjusting the structure of tree species [17,18].Despite the strong carbon sequestration potential of P. massoniana forests, their homogenous species and simplistic model structure lead to substantial pest and disease infestations, resulting in low ecological stability and weak ecological functions and preventing them from fully realizing their comprehensive benefits within the ecosystem.During the transformation of P. massoniana plantations, it was discovered that mixing with appropriate proportions of native broad-leaved tree species can enhance the degradation of lignin in fallen leaves, thus improving soil quality and enhancing soil carbon sequestration [8].
However, there is still a lack of research on following questions: What will happen to soil microorganisms and organic carbon after the pure P. massoniana plantation is adjusted to mixed plantations?Do microbial changes cause changes in soil carbon?Do soil microorganisms and organic carbon change the same under different mixing models?To answer these questions, four mixed models of P. massoniana were selected (P.massoniana + Liriodendron chinense, MS; P. massoniana + Machilus nanmu, MN; P. massoniana + Cinnamomum camphora, MX; and P. massoniana + Idesia polycarpa, MS), with P. massoniana as the control group.The aim of this research is to offer a scientific foundation for the sustainable management of P. massoniana plantations, thereby optimizing their carbon sequestration potential and ecosystem service functions.

Overview of the Study Area
The research site is located in the state-owned forest farm of Dushan County, Guizhou Province, China (Figure 1).Dushan County is located in the southeast of Qiannan Buyi and Miao Autonomous Prefecture of Guizhou Province, and is located in the southernmost part  C, annual precipitation is 1430 mm, and the frost-free period lasts for 297 days.The state-owned forest farm in Dushan County spans approximately 18,588 hectares.Of this, 15,839.14hectares are forest land, making up 85.21% of the total area, while the remaining 2748.86 hectares are non-forest land, representing 14.79% of the area.The forest area is 13,564.32hectares.The forest coverage rate is 72.96%, and the total forest stock volume is 865,465.48stere, which makes it the largest state-owned forest farm in Guizhou.In the forest farm, the predominantly exposed rocks are sand shale and carbonate rocks, with the soil primarily being composed of siliceous and ferric aluminum yellow soils.The soil is typically found in the middle layer and above, possessing model rate fertility and predominantly exhibiting acidic or slightly acidic properties.The plantation trees are mainly P. massoniana, Pinus elliottii, Cinnamomum camphora, and Liriodendron chinensis.The research site is located in the state-owned forest farm of Dushan County, Guizhou Province, China (Figure 1).Dushan County is located in the southeast of Qiannan Buyi and Miao Autonomous Prefecture of Guizhou Province, and is located in the southernmost part of Guizhou Province.The county has an average altitude of 850-1100 m and falls within the subtropical warm monsoon climate.The average annual temperature is 15 °C, annual precipitation is 1430 millimeters, and the frost-free period lasts for 297 days.The state-owned forest farm in Dushan County spans approximately 18,588 hectares.Of this, 15,839.14hectares are forest land, making up 85.21% of the total area, while the remaining 2748.86 hectares are non-forest land, representing 14.79% of the area.The forest area is 13,564.32hectares.The forest coverage rate is 72.96%, and the total forest stock volume is 865,465.48stere, which makes it the largest state-owned forest farm in Guizhou.In the forest farm, the predominantly exposed rocks are sand shale and carbonate rocks, with the soil primarily being composed of siliceous and ferric aluminum yellow soils.The soil is typically found in the middle layer and above, possessing model rate fertility and predominantly exhibiting acidic or slightly acidic properties.The plantation trees are mainly P. massoniana, Pinus elliottii, Cinnamomum camphora, and Liriodendron chinensis.

Plot Setup and Sampling
The sampling time for field investigation was from 15-30 May 2023.Soil samples were collected from four different models (MS, MN, MX, and MS) and unmodified P. massoniana plantations (M0), and three plots with an area of 0.0667 hectares were set for each mixed model (three repeated plots per mixed model, totaling fifteen plots).All plots were ensured to have the same soil parent rock, site conditions, model age, and growth, and the mixing year was the same.The mixing ratio was 1:1, and the spacing of each sample was more than 500 m.The sample plot information is shown in Table S1.
In the sample plot, soil profiles were excavated using a five-point sampling method, each measuring 1.2-1.5 m in length and 0.8-1.0m in width.Soil samples were collected from the excavated profiles at depths of 0-20, 20-40, and 40-60 cm, from bottom to top (three replicates per layer of soil sample, totaling forty-five samples).Soil samples from the same layer were thoroughly mixed, and 500 g aliquots were taken using the quartering method.

Plot Setup and Sampling
The sampling time for field investigation was from 15-30 May 2023.Soil samples were collected from four different models (MS, MN, MX, and MS) and unmodified P. massoniana plantations (M0), and three plots with an area of 0.0667 hectares were set for each mixed model (three repeated plots per mixed model, totaling fifteen plots).All plots were ensured to have the same soil parent rock, site conditions, model age, and growth, and the mixing year was the same.The mixing ratio was 1:1, and the spacing of each sample was more than 500 m.The sample plot information is shown in Table S1.
In the sample plot, soil profiles were excavated using a five-point sampling method, each measuring 1.2-1.5 m in length and 0.8-1.0m in width.Soil samples were collected from the excavated profiles at depths of 0-20, 20-40, and 40-60 cm, from bottom to top (three replicates per layer of soil sample, totaling forty-five samples).Soil samples from the same layer were thoroughly mixed, and 500 g aliquots were taken using the quartering method.

Determination and Analysis of Soil Physical and Chemical Properties
After the collected soil was naturally air-dried, 200 g of soil sample was taken.After grinding, the soil was screened with a 0.1 mm mesh and then used to determine the soil water content, pH, carbon, nitrogen, phosphorus, and other contents.Soil pH was measured by potential generation.Soil bulk density (BD) was measured by ring knife (200 cm 3 ) weighing method.The carbon content was determined by concentrated H 2 SO 4 and K 2 Cr 2 O 7 using the external heating method [19].The TN content was determined by semi-micro nitrogen determination method [20].The content of TP was determined by molybdenum-antimony-resistance colorimetric method [21].

Determination and Analysis of Soil Organic Carbon Components
Readily oxidized organic carbon (ROC) was determined by potassium permanganate oxidation method, granular organic carbon (POC) was determined by wet screen method and potassium dichromate external heating method, soluble organic carbon (DOC) was determined by distilled water extraction method, and microbial biomass carbon (MBC) was determined by chloroform fumigation method [22].
The formula for calculating soil carbon storage is as follows: where C s represents soil carbon reserves (t•hm −2 ); SOC indicates SOC content (g•kg −1 ); BD is the soil density (g•cm −3 ); D is the thickness of soil layer (cm).

DNA and Illumina Sequencing
We extracted the total DNA from five models with 30 samples using the Magnetic Soil And Stool DNA Kit from Tiangen Biotech Co. (Beijing, China).(It is worth noting that, to ensure the purity of the total DNA extracted, a 1% agarose gel electrophoresis test was conducted to detect the concentration and purity of the extracted genes, ensuring support for subsequent research.)The bacterial community PCR was specifically amplified in the "V4 variable region" using universal primers 515F (5 ′ -GTGCCAGCMGCCGCGGTAA-3 ′ ) and 806R (5 ′ -GGACTACHVGGGTWTCTAAT-3 ′ ), while the fungal community was amplified using ITS5-1737F (GGAAGTAAAAGTCGTAACAAGG) and ITS2-2043R (GCT-GCGTTCTTCATCGATGC).The PCR reaction system and detailed amplification process were followed as carried out in the previous study [23].The purity of the PCR testing and recovery were performed using the Universal DNA Kit from Tiangen Biotech Co. Purity was tested by electrophoresis of 2% agarose gel electrophoresis before recovery.The 16S rRNA and ITS genes that met the purity requirements were indexed using the library kit, and then, the library quality was evaluated using the Tiangen Biotech Co Bioanalyzer system.Qualified libraries were sent for sequencing on the Illumina platform (Illumina, San Diego, CA, USA) [24].
The raw data after sequencing were processed on the Bioincloud platform (Microeco Tech Co., Ltd., Shenzhen, China) with quality control and treatment, and the specific process was as follows: (1) Based on the Qiime2 dada2 plug-in, the quality control (filtered), denoising (correcting the sequencing errors), merging, and de-chimeric of the 30 samples' full raw sequences were performed to form the ASV table [25].(2) Species classification was performed on the ASV table data using sklearn algorithms, and the specific plug-in was used to compare the ASV sequences with the NCBI database, with a similarity of 99%, ultimately obtaining the species classification information table [26].(3) The species needed for the study were selected for analysis and processing.After the above quality control and de-chimeric treatment, a total of 1,491,546 fungal ASVs and bacterial ASVs were obtained.

Data Analysis and Processing
One-way ANOVA was used to assess the impact of various mixed cropping patterns on soil carbon fractions and microbial alpha diversity in SPSS 26.0.We also examined the differences in carbon composition and physicochemical properties across different soil layers, applying the one-way ANOVA (LSD) method for significance testing (p = 0.05).Linear fitting was utilized to establish the close relationship between soil carbon components and SOC.Charts were created using Origin 2022.Utilizing Bray-Curtis distance, principal coordinate analysis (PCoA) was applied to predict how mixed cropping patterns affect fungal and bacterial community structures.Microbial community differences were initially compared on the cloud platform of Bioincloud platform and were subsequently visualized using R software (V4.3.2;https://www.r-project.org/) for enhanced presentation.The "rdaca.hp"package in R was utilized to categorize functional modules based on microbial physiological traits and nutritional strategies.The redundancy analysis (RDA) within the "vegan" R package was employed to explore potential associations between functional modules and relevant environmental factors, generating a heatmap.Based on the relative abundance of key microbial species in the sample, co-occurrence analysis was conducted to calculate the Spearman's rank correlation coefficient, determining species associations and further assessing the relationship between the top 10 dominant microbial functional modules and SOC.RDA testing was performed using Canoco5, with the significance level set at p < 0.05.

Analysis of Soil Carbon Changes and Correlation between Carbon Components and Organic Carbon Storage in Different Models
The results indicated that soil carbon sequestration was influenced by different mixing models (Figure 2).The SOC content in all models exhibited a decreasing trend with increasing soil depth.In the 0-60 cm soil layer, the SOC content in the MN model was relatively higher, and the surface SOC content was significantly greater than that in M0 (p < 0.05).With the exception of ME, the SOC content in the other three mixed models was slightly lower than that in the M0 model, but the difference was not statistically significant (p < 0.05).Across all models, the highest carbon lability activity occurred in the surface layer (0-20 cm), with M0 exhibiting the highest carbon lability activity; however, there were no significant differences observed with respect to soil depth and mixed models (p < 0.05).In the five models, EOC, POC, DOC, and MBC were all enriched in the surface layer (0-20 cm) of soil.The highest EOC and POC levels were found in the M0 model; however, except for MS and MX, these differences were not significant (p < 0.05).In terms of performance, the DOC and MBC decreased with soil depth, and the content of the four mixed models was higher than M0, though the difference was not significant.There is a positive correlation between EOC, POC, DOC, MBC, and SOC (Figure 3), in which EOC and POC were significantly positively correlated (p < 0.01), and POC had a greater impact on SOC storage content than did EOC.In addition, the correlation between DOC and SOC (R 2 = 0.015) was more pronounced than the correlation between MBC and SOC (R 2 = 0.002).There is a positive correlation between EOC, POC, DOC, MBC, and SOC (Figure 3), in which EOC and POC were significantly positively correlated (p < 0.01), and POC had a greater impact on SOC storage content than did EOC.In addition, the correlation between DOC and SOC (R 2 = 0.015) was more pronounced than the correlation between MBC and SOC (R 2 = 0.002).
differences in different mixed planting patterns within the same soil layer, while a indicate significant differences in different soil layers within the same forest model.SOC: soil organic carbon; POC, particulate organic carbon; MBC, microbial biomass C; EOC, readily oxidizable organic carbon; DOC, dissolved organic C; CL, carbon lability.
There is a positive correlation between EOC, POC, DOC, MBC, and SOC (Figure 3), in which EOC and POC were significantly positively correlated (p < 0.01), and POC had a greater impact on SOC storage content than did EOC.In addition, the correlation between DOC and SOC (R 2 = 0.015) was more pronounced than the correlation between MBC and SOC (R 2 = 0.002).

Microbial Community Composition
In all mixed models, Basidiomycota (45.88%-68.18%),Ascomycota (22.12%-39.89%),Mortierellomycota (2.58%-8.72%),Mucoromycota (0.27%-5.44%), and Chytridiomycota (0.01%-6.19%) were the dominant phyla in the fungal community, with notably higher abundances in MN, ME, and M0 (Figure 4).Basidiomycota had the highest abundance in MX and M0 models, while the lowest abundance was observed in MS, and Ascomycota had the highest abundance in ME and MN models.In terms of total fungal species abundance and alpha diversity index (Figure 5; fungal), except for MS, the alpha diversity index of the other three model species was higher than M0, and chao1 was significantly lower than ME, with no significant difference in Shannon index.Acidobacteria (6.94%-26.52%),Actinobacteria (4.48%-7.88%),Firmicutes (3.21%-8.72%),and Proteobacteria (29.76%-47.62%)dominated the bacterial community.Proteobacteria showed the highest abundance in the M0 model, while Acidobacteria had the opposite abundance.In terms of total bacterial community abundance and alpha diversity index (Figure 5; bacteria), M0 had the highest species diversity, followed by MX; in addition, the M0 model of the chao1 index was significantly different from the mixed models (p < 0.05), and the Shannon index M0 was significantly different from the MS and MN models (p < 0.05).Based on the Bray-Curtis distance, the soil microbial community abundance was ranked, and the analysis results showed that there were significant differences in the microbial community composition in various mixed models (Figure 6).abundance and alpha diversity index (Figure 5; fungal), except for MS, the alpha diversity index of the other three model species was higher than M0, and chao1 was significantly lower than ME, with no significant difference in Shannon index.Acidobacteria (6.94%-26.52%),Actinobacteria (4.48%-7.88%),Firmicutes (3.21%-8.72%),and Proteobacteria (29.76%-47.62%)dominated the bacterial community.Proteobacteria showed the highest abundance in the M0 model, while Acidobacteria had the opposite abundance.In terms of total bacterial community abundance and alpha diversity index (Figure 5; bacteria), M0 had the highest species diversity, followed by MX; in addition, the M0 model of the chao1 index was significantly different from the mixed models (p < 0.05), and the Shannon index M0 was significantly different from the MS and MN models (p < 0.05).Based on the Bray-Curtis distance, the soil microbial community abundance was ranked, and the analysis results showed that there were significant differences in the microbial community composition in various mixed models (Figure 6).

Correlation Analysis of Soil Carbon Components and Microbial Communities
By clustering and statistical comparative analysis of the soil microbial community structure in all experimental groups, the fungal sequences were classified into 581 genera in 26 phyla and the bacteria into 21 phyla 422 genera.In order to better demonstrate the relationship between the main dominant groups of microorganisms and soil carbon components, this study used Gephi to construct the species interaction network.The network analysis results showed that fungi and bacteria both had eight dominant modules (Figure 7A,B).Fun_mod 6 positively correlated with SOC storage and POC, Fun_mod 1 positively correlated with ROC and carbon lability, and Fun_mod 7 positively correlated with DOC.Fun_mod 2 positively correlated with POC (p < 0.05).The relative abundance of Bac_mod 7 significantly positively correlated with CL (p < 0.01) (Figure 7C).Therefore, Fun_mod 1,2,6,7 and Bac_mod 7 could be considered as the main interpretation modules of the soil carbon pool.The dominant bacterial population of Fun_mod 1 was Epicoccum (74.07%), the dominant bacterial population of Fun_mod 2 was Rhizoctonia (64.74%), and the dominant bacterial population of Fun_mod 6 was Cystolepiota (Figure 7D).Fun_mod 7's dom-

Correlation Analysis of Soil Carbon Components and Microbial Communities
By clustering and statistical comparative analysis of the soil microbial community structure in all experimental groups, the fungal sequences were classified into 581 genera in 26 phyla and the bacteria into 21 phyla 422 genera.In order to better demonstrate the relationship between the main dominant groups of microorganisms and soil carbon components, this study used Gephi to construct the species interaction network.The network analysis results showed that fungi and bacteria both had eight dominant modules (Figure 7A,B).Fun_mod 6 positively correlated with SOC storage and POC, Fun_mod 1 positively correlated with ROC and carbon lability, and Fun_mod 7 positively correlated with DOC.Fun_mod 2 positively correlated with POC (p < 0.05).The relative abundance of Bac_mod 7 significantly positively correlated with CL (p < 0.01) (Figure 7C).Therefore, Fun_mod 1, 2, 6, 7 and Bac_mod 7 could be considered as the main interpretation modules of the soil carbon pool.The dominant bacterial population of Fun_mod 1 was Epicoccum (74.07%), the dominant bacterial population of Fun_mod 2 was Rhizoctonia (64.74%), and the dominant bacterial population of Fun_mod 6 was Cystolepiota (Figure 7D).Fun_mod 7's dominant bacteria group was Xylodon, accounting for 57.30% abundance.Bac_mod 7's dominant bacteria group was Ensifer, accounting for 18.55% abundance.Taking the above dominant genus ASVs, as the response variable, we performed RDA analysis on SOC components as explanatory variables (Figure 7E).The results showed that the first and second main axes accounted for 25.75% of the variation of dominant modules.Epicoccum, Rhizoctonia and Cystolepiota positively correlated with DOC, MBC, and SOC; Ensifer positively correlated with POC, EOC, and CL; and Xylodon positively correlated with carbon storage, but the correlation was not significant (p < 0.05).

Soil Carbon Composition and Reserve Change under Mixed Modification Model
Carbon in the soil primarily originates from the decomposition of litter, substances secreted by plant roots, and detritus produced by root activities, with the highest concentration in the topsoil layer [27].This study demonstrates that the SOC content gradually diminishes with increasing soil depth, exhibiting a distinct vertical distribution pattern [28].This suggests that tree species composition mainly influences SOC in the surface layer [29,30].This is attributed to the progressive decrease in root distribution with depth, which slows down the migration of organic matter derived from the decomposition of surface litter and plant roots [31].Furthermore, significant differences exist in SOC content among different models, with the pattern MN > M0 > ME > MS > MX.This difference is attributed to varying growth characteristics and nutrient storage abilities among models, leading to differences in balanced control of nutrient input and output by vegetation, resulting in variations in soil carbon accumulation and release rates [32].Additionally, carbon lability, a marker of soil carbon pool stability, was highest in M0 and lowest in MX, suggesting that pure P. massoniana plantations exhibit the least stable soil carbon pools among the different models, with mixed and transformed forests proving more effective at retaining organic matter in the soil [33].The change trend of the soil carbon component also varies under different models of P. massoniana plantation and is similar to SOC.These differences are related to the quantity, quality, and decomposition rate of litter from various vegetation types [31,34].Particulate organic carbon primarily stems from plant and animal residue decomposition as well as forest community litter layers.Linear fitting demonstrated a significant positive correlation between POC and SOC during mixed transformation of artificial pure plantations, indicating that plant decomposition remains a major contributor to SOC content [35].The finding further corroborates earlier conclusions.

Soil Microbial Community Composition Interacts with Soil Carbon
Section 3.1 showed that carbon components of all models were mainly enriched in the surface layer (0-20 cm), and the carbon lability activity was highest in the surface layer.Yu et al. also showed that soil carbon in forest ecosystems was mainly in the surface layer [36].Therefore, the microbial part is represented by the surface layer (0~20 cm).In this study, there was no significant difference in the composition of soil fungal communities among different mixed models, but the dominant fungal groups were Basidiomycota, Ascomycota, and Mortierellomycota.This is consistent with the studies on the dominant microbial communities in most forest ecosystems [37,38].Ascomycota and Basidiomycota were used as indexes of land-use efficiency for accumulation and storage of SOC [37].In this study, the abundance of Basidiomycota and Ascomycota showed a mixed pattern higher than that of P. massoniana pure forest.These indicate that soil carbon is more easily accumulated and stored in MN, ME, and M0 models under the control of microorganisms, which is consistent with the above conclusions on the change of carbon content.Different from fungi, the composition of soil bacterial community in different mixing models in this study was significantly different, which is consistent with the study of Liu et al. [18].The results showed that the abundance and uniformity of bacteria had a more positive impact on soil carbon storage [18].Acidobacteria, Actinobacteria, and Proteobacteria are the dominant bacterial communities.Consistent with previous studies, Firmicutes comprise the most common bacterial communities in soil [39].Firmicutes is a kind of eutrophic microorganism.They prefer a nutrient-rich soil environment and regulate the degradation process of refractory lignin in soil [33].During the study, it was found that the mixed model except for MX increased the relative abundance of Firmicutes, while the M0 model decreased the relative abundance.Moreover, correlation studies showed that the abundance of Firmicutes was significantly positively correlated with soil TN (Figure S1), indicating that changes in the relative abundance of Firmicutes were related to changes in soil nutrients, especially soil nitrogen content.
In the visualization of microbial networks, species with similar traits can be grouped into functional modules.This approach facilitates the exploration of the relationship between functional gene modules and soil ecological functions, thereby aiding in the understanding of the connection between complex microbial communities and changes in the soil environment [40].The network visualization showed that Fun_mod 6 was positively correlated with SOC storage, Fun_mod 1 was positively correlated with carbon lability, and the relative abundance of Bac_mod 7 was significantly positively correlated with CL (p < 0.01).Therefore, Fun_mod 1, 6 and Bac_mod 7 can be considered as the microbial functional modules that play a decisive role in the process of soil carbon pool change.
Ensifer is a root nodal endophyte that has attracted much attention from microecological researchers in recent years.Unlike Rhizobium, this non-symbiotic bacteria does not form root nodules in the root nodules but produces plant hormones that affect the development of roots, thus promoting the absorption of carbon and nitrogen and other nutrients in the soil by plants [41,42].Ensifer was the dominant bacterial group in Bac_mod 7. The correlation showed that the conversion of artificial pure forest into mixed forest may lead to the increase of EOC and POC contents, which leads to the change of microbial physiological characteristics (Figure 7C,E).In our study, POC represented greatest correlation with SOC.Soil carbon was mainly stored in the POC bank derived from the decomposition of soil surface plants [43], which originates from the decomposition of surface plants in the soil.This suggests that when transitioning from a pure coniferous forest to a mixed coniferous and broad-leaved forest, soil microorganisms primarily influence the storage and stabilization of soil carbon by regulating the transformation of surface plants (carbon sources) into POC.This study clarified the complex relationship between mixed patterns, soil microorganisms, and soil carbon components, which is of great value in guiding the improvement of forest carbon sink efficiency through tree species structure adjustment in Guizhou Province in the future.

Forests
. The county has an average altitude of 850-1100 m and falls within the subtropical warm monsoon climate.The average annual temperature is 15

Figure 1 .
Figure 1.Location of the test site of the state-owned forest farm in Dushan County.

Figure 1 .
Figure 1.Location of the test site of the state-owned forest farm in Dushan County.

Figure 2 .
Figure 2. Soil carbon component content in different mixing models.A, B, and C indicate significant differences in different mixed planting patterns within the same soil layer, while a indicate significant differences in different soil layers within the same forest model.SOC: soil organic carbon; POC, particulate organic carbon; MBC, microbial biomass C; EOC, readily oxidizable organic carbon; DOC, dissolved organic C; CL, carbon lability.

Figure 2 .
Figure 2. Soil carbon component content in different mixing models.A, B, and C indicate significant differences in different mixed planting patterns within the same soil layer, while a indicate significant differences in different soil layers within the same forest model.SOC: soil organic carbon; POC, particulate organic carbon; MBC, microbial biomass C; EOC, readily oxidizable organic carbon; DOC, dissolved organic C; CL, carbon lability.

Figure 3 .
Figure 3. Correlation between soil carbon components and SOC in different models.

Figure 3 .
Figure 3. Correlation between soil carbon components and SOC in different models.

Figure 4 .
Figure 4. Circos plot of soil fungal and bacterial communities in different mixed patterns; (A) relative abundance of phylum levels of fungal communities and (B) that of bacterial communities.

Figure 4 .
Figure 4. Circos plot of soil fungal and bacterial communities in different mixed patterns; (A) relative abundance of phylum levels of fungal communities and (B) that of bacterial communities.

Figure 4 .
Figure 4. Circos plot of soil fungal and bacterial communities in different mixed patterns; (A) relative abundance of phylum levels of fungal communities and (B) that of bacterial communities.

Figure 5 . 13 Figure 5 .
Figure 5. Box plots of alpha diversity index in soil fungal and bacterial communities.lowercase a, b, c, d indicate significant differences in soil microbial alpha index under different mixed planting modes.

Figure 6 .
Figure 6.A 3D plot of soil fungal and bacterial communities based on Bray-Curtis distances.

Figure 6 .
Figure 6.A 3D plot of soil fungal and bacterial communities based on Bray-Curtis distances.