The Effects of Vegetation Restoration and Season On Soil Enzyme Activity and Microbial Communities in Karst Rocky Desertication Area

Aims The process of karst rocky desertication has been closely related to improper land use in southwest China. Now this habitat is the subject of an important ecological restoration project. However, the changes in soil properties and microbial characteristics in response to this vegetation restoration remain poorly understood. Methods We investigated four vegetation types, including dragon fruit, Chinese pepper, walnut teak, with corn as a control, in southwest China, in 2019. We measured the impacts of these vegetation types on soil properties and microbial biomass, enzyme activity, and microbial community composition (using high-throughput sequencing technology). Results The different vegetation types had signicantly different impacts on soil exchangeable Ca 2+ , soil organic carbon and available nutrients. The vegetation types also signicantly affected microbial biomass. Soil enzyme activity, including b-1,4-glucosidase, b-1,4-N-acetylglucosaminidase, alkaline phosphatase, and catalase, were signicantly different among vegetation types. All vegetation types were dominated by the bacterial phyla Acidobacteria, Proteobacteria, and Actinobacteria and the fungal phylum Ascomycota, except for corn which was dominated by the fungal phylum Mucoromycota. Non-metric multidimensional scaling (NMDS) showed that the vegetation type exhibited different microbial b-diversity, especially in winter. The vegetation type, season, and soil properties collectively explained 46% and 59% of soil bacterial and fungal community composition, respectively. The bacterial-fungal interactions under the six vegetation types were distinctly different between summer and winter. Conclusions Compared with traditional corn, the restoration of natural vegetation partially reversed KRD by improving soil properties, increasing microbial biomass, and differentiating the microbial community structures in the different vegetation types.


Introduction
Surface and near-surface karst outcrops occupy 20% of the world's ice-free dry land (Ford and Williams 2013). Karst areas are extremely valuable resources and host a rich variety of plants and animals (Gutiérrez et al. 2014), supply water to 25% of the planet's population (Ford and Williams 2013) and are closely associated with rural poverty ). However, karst rocky deserti cation (KRD) with degradation of soil ecosystems and plant communities has occurred in many countries and regions (Jiang et al. 2014). The exposed karst area in southwest China is one of the world's three largest continuous areas of carbonate rocks (Yuan 2008). It provides a variety of unique ecological niche and is one of the world's 34 biodiversity hotspots (Myers et al. 2000). However, it possesses 10 million ha of KRD due to improper land use (National Forestry and Grassland Administration of China 2018; Jiang et al. 2014). The process of KRD is positively related to vegetation degradation from secondary forest to sparse shrub and grassland, but this degradation could be partly reversed by ecological restoration (Liu et al. 2009). The Chinese "14th Five-Year Plan" (2021-2025) proposes promotion of a comprehensive and scienti c management plan to tackle KRD. The process of halting KRD had previously involved ecological restoration techniques to restore degraded ecosystems (Yuan 2008).
Soil microorganisms are the key drivers of biogeochemical processes in the atmosphere, hydrosphere, lithosphere, and biosphere (Bardgett et al. 2008; al. 2018; Eugene 2011). Using 16S rRNA amplicon metagenomics, Avitia et al. (2021) showed that the composition of the soil microbial community changed when grasslands transitioned to woody plant cover. Zhao et al. (2018) found that areas afforested with Robinia pseudoacacia over a 42, 27 and 17 year chronosequence increased their microbial diversity and altered community structure compared to farmland. In non-karst ecosystems, microbial community structures and activities are also affected by climate (temperature and precipitation) (

Study site
The study site is located within the Guanling-Zhenfeng demonstration area for the reduction and control of karst rocky deserti cation, Guizhou Province, China (25°39'13" ~ 25°41'00" N, 105°36'30" ~ 105°46'30" E). The site is characterized by a humid subtropical monsoon climate with a mean annual temperature of 18.4°C and a mean annual precipitation of 1100 mm, 83% of which falls from May to October. The lithology is dolomitic limestone of the Middle Triassic system, and the soil type is calcareous. This site suffers from serious soil erosion due to anthropogenic disturbance, especially related to cultivation, which has led to KRD, with a rock exposure rate of 70% per year ) and a thin, discontinuous soil layer. The native forest cover has diminished (Cheng et al. 2020) and some economically important revegetation has now been carried out to reverse the effects of KRD, including the planting of dragon fruit (Hylocereus undatus), Chinese pepper (Zanthoxylum bungeanum), walnut (Carya cathayensis), teak (Tectona grandis), honeysuckle (Lonicera japonica), and paper mulberry (Broussonetia papyrifera).

Study design and eld sampling
This study sampled soils underneath dragon fruit (DF), walnut (WN), teak (TW), and Chinese pepper growing in depressions (CPD) and on sloping sites (CPS) (Fig. 1). Traditionally farmed corn (Zea mays) (CN) growing on a nearby terraced slope served as a comparison. Detailed information on the six sample sites is shown in Table 1. Within each site, six plots were selected with two plots each on the upper, middle, and lower slopes. The upper, middle, and lower slope plots were 10 m apart within a site for the DF, TW and CN, and 5 m apart for the WN, and CPD and CPS. Five soil samples were collected per plot at a depth of 20 cm, along an S-shaped transect, in June and November 2019, and mixed to form one composite sample. ml 0.5 M K 2 SO 4 in a rotatory shaker (200 rpm). The fumigated subsamples were extracted using a similar method to the unfumigated subsamples. All extracts were ltered and analyzed using the potassium dichromate-dilution heat colorimetric method. Extracellular enzyme activity, including C-acquiring β-1,4-glucosidase (ΒG), N-acquiring β-1,4-N-acetylglucosaminidase (NAG), and organic P-acquiring alkaline phosphatase (AKP), were determined by microplate enzyme assay (Cui et al. 2018). Enzyme activity was expressed as nanomoles of substrate released per hour per gram of soil organic matter (nmol g SOM −1 h −1 ). Catalase (CAT) activity was determined using the KMnO 4 titration method (Guan 1986  index, and community composition. We constructed a priori SEM based on current knowledge, model modi cation indices, and stepwise removal of nonsigni cant relationships (de Vries and Bardgett 2016). We used a minimum set of parameters to assess model t using the multigroup modeling approach with the R package lavaan (Rosseel 2012), including soil properties, enzyme activity, and microbial biomass. We used the rst two NMDS axes as proxies for bacterial and fungal community composition. Pearson's correlation heatmaps were constructed using BMKCloud (www.biocloud.net).

Soil physico-chemical properties
The results of the two-way ANOVA showed that the soil properties were signi cantly different between all vegetation types (  Values (Means ± standard error) followed by the different letter are signi cantly different within columns in the same season (P < 0.05). *, **, and *** indicate signi cant differences at P < 0.05, P < 0.01, and P < 0.001, respectively. ns mean no signi cance.
Except for soil pH and AP content, soil properties changed signi cantly with the seasons (  Table S1). The MBN/TN ratio was also signi cantly affected by the interaction of vegetation type and season (P < 0.001, Table S1). Vegetation type also signi cantly in uenced enzyme activity, including ΒG, NAG, AKP, and CAT (

Microbial community composition and network analysis
In summer, the microbial diversity indexes of bacteria and fungi, including ACE, Chao, and Shannon, showed no signi cant difference among all vegetation types (  At the bacterial phylum level, all vegetation types were dominated by Acidobacteria (35% relative abundance), Proteobacteria (25%), and Actinobacteria (15%) (Fig. 2a). The relative abundance of Acidobacteria was highest under CPD (40%) and lowest under TW (29%). Under CN, the relative abundance of Proteobacteria decreased from 31% in summer to 22% in winter, while that of Actinobacteria increased from 10-18% over the same period. The relative abundance of Proteobacteria and Actinobacteria under WN showed the same trend as under CN. However, under CPD, the relative abundance of Proteobacteria increased from 19-26%, accompanied by a decrease in Actinobacteria from 35-30%. At the fungal phylum level, the most abundant phylum was Ascomycota under CPD and CPS, WN, and TW (average: 56%), while Mortierellomycota was most abundant under CN (38%) (Fig. 2b). From summer to winter, the relative abundance of Ascomycota decreased from 67-36%, while that of Mortierellomycota increased from 6-53% under DF. The relative abundance of Mortierellomycota was more than 33% under CN and CPD, but lower than 4% under TW and CPS.
For both bacteria and fungi, the vegetation type led to signi cant differences in community composition (Fig. 3, P < 0.001), and there was a little overlap among the six vegetation types in summer (Fig. 3a and c). The bacterial and fungal community structures changed from summer ( Fig. 3a and c) to winter ( Fig. 3b and d), and the tighter clustering can be seen in the bar plots in Fig. 3. Due to their greater arti cial disturbance, the Bray-Curtis distance between DF and CN was closer than among the other vegetation types, especially in winter ( Fig. 3b and d). Meanwhile, the distance between CPD and CPS was greater than that between WN and CPS due to locational differences in Chinese pepper planting. Therefore, the effect of vegetation type on the composition of the bacterial and fungal communities was greater in winter than in summer, based on the larger PERMANOVA R2 values (Fig. 3).
We then performed network analyses to assess the impact of vegetation type and season on microbial interactions. The soil microbial network patterns differed among the six vegetation types and showed clear changes from summer to winter (Fig. 4a-l, Table S4). The microbial taxa showed higher network connectivity (i.e. network degree) under CN and CPD (Fig. 4c-f) than in other vegetation types. Bacterial taxa had higher network degrees than fungal taxa, especially under CN and CPD in winter (Fig. 4d, f, (Fig. 4, Table S4). The average path distances under CN (5.917) and CPD (5.809) were lower than under the other vegetation types, with values > 7 (Table S4). Moreover, the proportion of negative network edges (mainly representing bacteria-fungi interkingdom correlations) sharply declined from 39.2-8.6% under CPD (Fig. 4c, d) and from 40.3-8.8% under CN (Fig. 4e, f) from summer to winter, respectively.

Structural equation model (SEM) and Pearson's correlation heatmap
The SEM model was a reasonable t to our data (Fig. 5). The model showed that 63% and 84% of the variance in the rst and second soil fertility NMDS axes was explained by vegetation type, season, and soil moisture (Fig. 5). Soil moisture had signi cant and positive correlations with the SOC and TN levels (0.555 and 0.598, P < 0.01) ( Table S2). The model explained 83% and 71% of the variance in the rst and second soil enzyme activity NMDS axes. Season, soil fertility, soil moisture, soil pH and Ca 2+ , and microbial biomass directly affected soil enzyme activity. The effects of season and soil Ca 2+ were positive, whereas those of soil fertility, moisture, and soil pH had a negative effect on enzyme activity, which was also supported by the Pearson's correlation (Table  S2). The model explained 56% and 75% of the variance of MBC and MBN, respectively, which were directed in uenced by season, soil pH, and fertility.
Thirty-ve percent of the variance in the bacterial Shannon diversity index was explained by soil pH alone (Fig. 5a, path coe cient = 0.661***). The SEM explained 46% and 58% of the variance in the rst and second bacterial community NMDS axes. Soil pH and fertility showed direct positive effects, whereas vegetation type and soil Ca 2+ showed direct negative effects on bacterial community NMDS1. Vegetation type directly and positively in uenced bacterial community NMDS2. Due to direct and indirect effects mediated by soil fertility, vegetation type showed a stronger effect on bacterial community composition than soil pH and Ca 2+ . The fungal Shannon diversity index was directly and negatively in uenced by MBN (-0.659) and soil fertility NMDS2 (0.603) (Fig. 5b).
The SEM explained 59% and 42% of the variance of the rst and second fungal community NMDS axes. Vegetation type and soil Ca 2+ showed positive effects while season and soil fertility negatively affected fungal community composition.
For bacterial phyla, the SOC and TN contents showed signi cant and positive correlations with the abundances of Entotheonellaeota, Armatimonadetes, and Actinobacteria (P < 0.05), but were negatively correlated with the abundances of Nitrospirae and Proteobacteria (Fig. 6a, P < 0.05). The soil Ca 2+ , AP, AK, and

Effects of vegetation type and season on soil physico-chemical properties
Our results showed that some vegetation types, including CPS, CPD, and WN, could alter the physico-chemical properties of soils (Table 2). In both summer and winter, the higher water storage capacities under these vegetation types could prevent soil erosion and deserti cation (Jiang et al. 2014). The SOC is a perfect proxy for judging improvements in soil quality due to vegetation restoration (Lal 2004

Effects of vegetation type and season on soil microbial biomass and enzyme activity
In this study, the contents of soil MBC and MBN were signi cantly impacted by the vegetation type and season ( Table 3) (Table S1). The MBC/MBN ratio represents the structure and state of a soil microbial community (Joergensen and Brookes 2005). It is widely accepted that the C/N ratio in microbial biomass is about 6.5 for bacteria and 5-17 for fungi (Cleveland and Liptzin 2007). Our results showed that the dominant bacterial effect under CN is greater than under the other vegetation types in summer. This study also indicated that the soil microbial community is dominated by bacteria in summer (MBC/MBN < 6.5) but by fungi in winter (MBC/MBN > 6.5) (Table S1). Our results also showed that organic inputs, especially high quality organic matter, could favor vegetation types which restore soil microbial biomass.
Soil enzyme activity plays a key role in mineralization and transformation of organic matter, involving C, N, and P cycling in soil ecosystems (Chen et al. 2020; Kumar and Maiti 2011; Liu et al. 2021). In this study, the activities of BG, NAG, AKP, and CAT showed signi cant differences among the vegetation types (P < 0.001, Table 3). The NAG and AKP activities also changed signi cantly from summer to winter and were signi cantly affected by the interactions between vegetation type and season (P < 0.001). Compared with WN, CPD, and CPS, the activities of BG, NAG, and AKP were higher under CN, TW, and DF and had lower SOC contents. This can be explained by the fact that enzyme activity is strongly affected by the root system when vegetation is intensively planted. Cui et al. (2018) showed that inconsistencies between variations in microbial nutrient ratios and ecoenzyme ratios could be explained by the impacts of root systems. Bell et al. (2014) indicated that the roots of gramineous plants produce more extracellular enzymes to meet their nutrient requirements. Moreover, ecoenzymes produced by roots can enter the soil after root death (Rillig et al. 2007). In this study, soil enzyme activity was directly affected by MBN (P < 0.01, Fig. 5). However, NAG activity showed a negative correlation with MBN (P < 0.01, Table S2), which was attributed to the fact that NAG decomposes microbial residues to provide available N for soil microorganism and plant growth . We found that the increases in NAG and AKP activity were higher in winter than in summer, perhaps because of the higher SOC content (Kumar and Maiti 2011). Overall, the enzyme activity responses implied that vegetation restoration could result in different soil nutrient cycles among the six vegetation types.

Effects of vegetation type and season on the microbial community composition and network
The numbers of bacterial and fungal OTUs differed, but not signi cantly so, among all vegetation types (Table S3). However, the distribution of bacterial and fungal phyla was clearly different between vegetation types (Fig. 2). Our results showed that Acidobacteria, Proteobacteria and Actinobacteria were dominated under all vegetation types in both seasons (Fig. 2a), as reported by Liao et al. (2018). Liao et al. (2018) also found that some members of the Actinobacteria (e.g., Solirubrobacteraceae) signi cantly increased their relative abundances following land-use conversion in degraded karst ecosystems. The distribution of bacterial phyla was also affected by the karst terrain. For example, the relative abundance of Acidobacteria on ridges and in depressions were higher than on slopes, while the relative abundance of Proteobacteria showed the opposite trend ). Regarding the fungi, the most abundant phylum under corn was Mortierellomycota, while Ascomycota dominated under the other vegetation types (Fig. 2b). Vegetation restoration therefore clearly altered the relative composition of the soil microbiota, compared with traditional CN planting.
Compared with traditional CN planting, other vegetation types, including WN, CPD, and CPS, signi cantly increased microbial biomass (Table 2)  biomass were higher during the shrubland phase than in the forest phase due to changes in soil conditions (i.e., reduced pH) and resource availability (i.e., reduced SOC). Hu et al. (2016a) showed that planting of Pepino (Solanum muricatum) increased the diversity and abundance of bacterial communities in karst areas. In this study, the microbial diversity was not only affected by vegetation type, but also by substrate and environmental factors.
The vegetation type affected the microbial community composition in both summer and winter (Fig. 3). Avitia et al. (2021) also reported that differences in microbial β-diversity were mainly driven by vegetation type in Arizona. Zheng et al. (2021) found that forest type drove latitudinal differences in AM fungal βdiversity. Regardless of vegetation type, we observed greater variation and more overlap of microbial community composition in summer than in winter ( Fig. 3). This difference was perhaps due to the higher temperatures in summer ( Fig. 3a and c). This result was consistent with the ndings of Shen et al. (2021), which indicated that differences in microbial community composition were greater under warm than cool conditions. Similarly, our results implied that higher summer temperatures increase microbial richness and β-diversity (Table 3 and Fig. 5). In spite of the vegetation species being the same, soils under CPS showed greater bacterial and fungal compositional dissimilarities compared with CPD, than did CPS compared with WN (Fig. 3). This difference appears to be due to location: WN and CPS were found on slopes at similar elevation, while CPD growing in depressions occupied lower elevations. Therefore, in this study in karst areas, microbial β-diversity is affected by spatial heterogeneity, including topography and elevation, as well as by vegetation type and season and winter (Fig. 4). The bacterial taxa present showed higher α-diversity (Table 4) and network connectivity (Fig. 4) than the fungal taxa, according to the dominant vegetation type (Xiong et al. 2021). Our results indicated that the proportion of negative network edges decreased from summer to winter under CPD (Fig. 4c, d) and CN (Fig. 4e,f), primarily due to changes in soil fertility ( Table 2) and reduced competition between bacteria and fungi (Xiong et al. 2021).

Structural equation model and Pearson's correlation under six vegetation types
In karst areas, the microbial community composition is affected by changes in soil properties following vegetation restoration (Li et al. 2018b). Our results showed that some soil properties were signi cantly correlated with microbial community characteristics (Fig. 6). We found that SOC was positively and signi cantly correlated with TN level (Table S2), and that the higher SOC and TN contents under WN, CPD, and CPS led to increased microbial biomass ( Table   2 and 3). This effect was principally attributed to the coupling of the biogeochemical cycles of C, N, and P in terrestrial ecosystems (Delgado-Baquerizo et al. 2013). However, the higher SOC contents did not lead to higher F/B ratio as implied by other factors, such as root system changes, which may affect microbial community structure (Li et al. 2018b). This study showed that the bacterial phyla Nitrospirae and Gemmatimonadetes were signi cantly correlated with NO 3 − -content. Both N and P are limiting elements in karst areas  and are the main factors affecting variations in N-cycling microorganisms (Li et al. 2018c). In karst soils, the C and N contents signi cantly affected the phoD-harboring bacterial community structure under long-term fertilization ). Therefore, vegetation restoration, including vegetation type and management practices, can alter the composition of microbial communities through changes in soil properties.
In this study, our SEM showed that vegetation restoration and season directly and indirectly affect soil microbial biomass, diversity, and composition mainly through alterations of plant type and soil properties (Fig. 5). Microbial characteristics can be directly affected by soil fertility, which was mainly explained by season and by vegetation type and its cultivation and fertilization ( Table 2, Fig. 6 and 5). Soil nutrient availability (e.g., AK, AP, and AN) indirectly affects soil microbial growth, leading to temporal variations in microbial diversity and activity (Yang et al. 2017). We found that both vegetation type and season had a negative effect on soil Ca 2+ content ( Table 2  Soil properties signi cantly affect bacterial community composition during karst vegetation degradation and restoration (Tang et al. 2019). In this study, the vegetation type and soil properties (soil fertility, pH, and Ca 2+ ) directly affected bacterial community composition (Fig. 5). The bacterial community was more responsive to soil pH than the fungal community, as also shown by some other studies ( In this study, vegetation type and soil Ca 2+ had direct positive effects on fungal community composition, while season and soil fertility had negative effects (Fig. 3). The seasonal effects on fungal communities were perhaps directly driven by temperature and precipitation. These climatic factors determine fungal survival and colonization (Teng et al. 2021). The SEM explained 34.7% and 19.6% of the Shannon diversity index found in our study for bacteria and fungi, respectively (Fig. 5). Soil pH was the only contributor to the bacterial Shannon diversity index (0.661***) (Fig. 5). However, bacterial and fungal β-diversity showed signi cant differences among the six vegetation types, especially in winter (Fig. 3). Our results implied that microbial α-diversity may be improved by increasing aboveground diversity using mixed planting of different vegetation types.

Conclusions
In this study, we investigated the impacts of four restored vegetation types in Guizhou province, southwest China, in 2019. properties collectively explained 46% and 59% of the bacterial and fungal community compositions, respectively. Soil pH and fertility exerted the strongest direct effects on bacterial communities. The SOC and TN contents were signi cantly positively correlated with the abundances of Entotheonellaeota, Armatimonadetes, and Actinobacteria (P < 0.05). The soil exchangeable Ca 2+ content exerted direct effects on both the bacterial and fungal communities.
Therefore, vegetation restoration partly reversed the effects of KRD by improving soil properties, increasing microbial biomass, and differentiating the microbial community structures compared with traditional CN vegetation. Figure 1 The study sampled soils underneath dragon fruit (DF), walnut (WN), teak (TW), and Chinese pepper growing in depressions (CPD) and on sloping sites (CPS) Figure 3 For both bacteria and fungi, the vegetation type led to signi cant differences in community composition (Fig. 3, P < 0.001), and there was a little overlap among the six vegetation types in summer ( Fig. 3a and c). The bacterial and fungal community structures changed from summer ( Fig. 3a and c) to winter ( Fig. 3b and d), and the tighter clustering can be seen in the bar plots in Fig. 3. Due to their greater arti cial disturbance, the Bray-Curtis distance between DF and CN was closer than among the other vegetation types, especially in winter ( Fig. 3b and d). Meanwhile, the distance between CPD and CPS was greater than that between WN and CPS due to locational differences in Chinese pepper planting. Therefore, the effect of vegetation type on the composition of the bacterial and fungal communities was greater in winter than in summer, based on the larger PERMANOVA R2 values (Fig. 3).

Figure 4
The soil microbial network patterns differed among the six vegetation types and showed clear changes from summer to winter ( Fig. 4a-l, Table S4). The microbial taxa showed higher network connectivity (i.e. network degree) under CN and CPD ( Fig. 4c-f) than in other vegetation types. Bacterial taxa had higher network degrees than fungal taxa, especially under CN and CPD in winter (Fig. 4d, f, Table S4). In summer and winter, the average number of nodes under CN (641) was lower than under the other vegetation types (DF 764, CPD 754, WN 746, CPS 734, TW 660), while the average number of links under CN (2333) was lower than under CPD (2764), but higher than under the other vegetation types (Fig. 4, Table S4). The average path distances under CN (5.917) and CPD (5.809) were lower than under the other vegetation types, with values > 7 (Table S4). Moreover, the proportion of negative network edges (mainly representing bacteria-fungi interkingdom correlations) sharply declined from 39.2% to 8.6% under CPD (Fig. 4c, d) and from 40.3% to 8.8% under CN (Fig. 4e, f) from summer to winter, respectively.

Figure 5
Thirty-ve percent of the variance in the bacterial Shannon diversity index was explained by soil pH alone (Fig. 5a, path coe cient = 0.661***). The SEM explained 46% and 58% of the variance in the rst and second bacterial community NMDS axes. Soil pH and fertility showed direct positive effects, whereas vegetation type and soil Ca2+ showed direct negative effects on bacterial community NMDS1. Vegetation type directly and positively in uenced bacterial community NMDS2. Due to direct and indirect effects mediated by soil fertility, vegetation type showed a stronger effect on bacterial community composition than soil pH and Ca2+. The fungal Shannon diversity index was directly and negatively in uenced by MBN (-0.659) and soil fertility NMDS2 (0.603) (Fig. 5b).

Figure 6
For bacterial phyla, the SOC and TN contents showed signi cant and positive correlations with the abundances of Entotheonellaeota, Armatimonadetes, and Actinobacteria (P < 0.05), but were negatively correlated with the abundances of Nitrospirae and Proteobacteria (Fig. 6a, P < 0.05). The soil Ca2+, AP, AK, and NO3--N contents were positively correlated with Nitrospirae and Gemmatimonadetes: in particular, the NO3--N content was strongly signi cantly correlated with them (r = 0.69, P < 0.001). Meanwhile, the soil Ca2+ content was negatively correlated with the abundances of Actinobacteria, Entotheonellaeota, and Firmicutes (r < -0.4, P < 0.001). The soil NH4+-N content, and the AKP and NAG enzyme activities, were positively correlated with the abundance of Actinobacteria and Chloro exi, but negatively correlated with the abundance of Proteobacteria. CAT enzyme activity was negatively correlated with the abundance of Rokubacteria and Verrucomicrobia, while the activity of glucosidase was negatively correlated with the abundance of Entotheonellaeota. For the fungal phyla, the soil Ca2+, AP, AK, and NO3--N contents were positively correlated with the abundance of Mortierellomycota (Fig. 6b, P < 0.001). However, the soil Ca2+, AK, and NO3--N contents showed signi cant and negative correlations with the abundance of Ascomycota (P < 0.01). The soil AP content was positively correlated with the abundances of Chytridiomycota and Zoopagomycota. Glycosidase and NAG activities were positively correlated with the abundance of Glomeromycota.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.