Plant production-community composition relationship associated with fungal guilds depends on grassland disturbance in surface and subsurface soils

Purpose Soil fungal guilds have been proven to inuence the plant community composition-production relationship, but not much is known about their effects on surface and subsurface soils under different disturbances. Methods Here, we assessed the functional characteristics of three fungal groups using the Ribosomal Database Project (RDP) classier and data available in FUNGuild, and we characterized the community of saprotrophic, mycorrhizal, and potential plant pathogenic fungi in surface (0–10 cm) and subsurface soils (10–20 cm) of temperate grasslands under different management practices. Results We found that grassland disturbances decreased plant aboveground production and changed plant community composition. In surface soil, antagonistic interaction between potential plant pathogens and saprotrophic fungi drove the plant community composition-production relationship. In subsurface soil, this relationship was driven by antagonistic interaction between mycorrhizal fungi and potential plant pathogens. These ndings revealed that under grassland disturbances, the surface soil fungal communities were more strongly associated with plant community composition-production relationship than those from the subsurface soil were. Potential plant pathogens played an important role in plant community composition-production relationship. This knowledge is important for predicting the shifts in ecosystem functions as a consequence of changes in soil fungal groups during grassland management. use drives small-scale homogenization of plant- and leafhopper communities and promotes generalists.


Introduction
Grazing and mowing are the main ways of human disturbance to grassland ecosystems (Gong et al. 2014), and they are the main grassland management strategies in China (Liu et al. 2018). The grazing intensity increases, the habitat, and productivity of the grassland ecosystem show a signi cant downward trend, reducing biodiversity (Eldridge & Delgado-Baquerizo 2017). Mowing signi cantly changes the species richness of aboveground plants (DÍAz et al. 2007), increasing plant community stability (Yang et al. 2012). Selective grazing, trampling and return of nitrogen from livestock in grazing areas cause habitat heterogeneity, while uniform harvesting in mowing areas highly affects species communities causing habitat homogeneity (Fóti et al. 2017;Chisté et al. 2018). Appropriate disturbance levels tend to promote grass growth, maintain stable plant community composition and productivity, and increase heterotrophic microbial activity (Hamilton & Frank 2001;Schönbach et al. 2011); however, excessive disturbance reduces the complexity of plant community composition and decreases soil microbial community diversity and plant productivity (Bardgett et al. 1998;Schönbach et al. 2011). Thus, grassland disturbance affects both plant community composition-production relationships and subsurface microbial communities (Vannucchi et al. 2015;Frąc et al. 2018). Thus, when evaluating the versatility of sustainable grassland ecosystem management, both above-and below-ground biodiversity must be considered (Putten et al. 2016; Wang et al. 2020b).
Grassland disturbances (grazing and mowing) in uence the soil microenvironment, causing changes in the soil nutrient availability and signi cantly affecting the fungal community structure and soil fungal activity (Ingram et al. 2008;Józefowska et al. 2018;Xun et al. 2018). These disturbances also change the interactions between fungal guilds (Grau et al. 2017), reduce grassland carbon sequestration, and cause changes in grassland production, diversity, and their potential relationships. However, Chen et al. (2018) reported that soil fungal diversity regulated the plant community composition-production relationship, and that it was not consistent in surface and subsurface soil. In semi-arid grasslands, exogenous nutrient addition induces soil environment and plant community changes, and plays a major role in shaping fungal diversity in surface soil. Conversely, in subsurface soil, fungal diversity is primarily controlled by exogenous nutrient addition and changes in plant community attributes (i.e., aboveground plant production and plant community composition). However, it is unknown how various disturbances that alter aboveground grassland structure impact the community assembly, abundance, and diversity of soil fungi, which have wide-ranging effects on grassland health, biogeochemistry, and regeneration (Liu et al. 2015;Tardy et al. 2015).
Fungal functional guilds play an important role in driving soil processes and can be used to predict soil abiotic factors and plant community characteristics (Chen et al. 2020). Many studies have shown that plant community composition-production relationships are controlled by complex networks of plants, soil, and microorganisms (Yang et al. 2015;Lozano et al. 2017); this relationship can also be in uenced by a series of biotic and abiotic factors (Guo et al. 2019). In greenhouse experiments, Yang et al. (2015) found that the strong effects of soil biota in soils do not simply accrue in experimental monocultures, but can occur in low diversity assemblages which are more realistic representations of what occurs in nature, as they contribute to spatial and temporal patterns of abundance in natural plant communities through negative plant-soil feedback. Fungi play important roles in the ecosystem through interactions of saprophytic fungi, mycorrhizal fungi, and potential plant pathogen fungi; these relationships are the determinants of the close relationship between plant community composition and production in terrestrial ecosystems (Heijden et al. 2008;Schnitzer et al. 2011). For example, potential plant pathogens can contribute to the maintenance of plant diversity by specifically suppressing dominant plants (Schnitzer et al. 2011). The results of the study by Yang et al. (2018) highlighted the importance of arbuscular mycorrhizal fungi (AMF) in structuring natural above-ground production under various biodiversity loss scenarios, and indicated that AMF will be able to modify vegetation dynamics in Page 4/27 response to the future plant diversity loss. Saprophytic fungi play a role in plant decomposition. Previous studies have examined how individual disturbances impact the community composition of soil fungi in grasslands. For example, grazing led to a signi cant decrease in fungal species diversity and abundance of pathotrophs, saprotrophs, and symbiotrophs (Yin et al. 2019). Long-term destocking can be considered as an important predictor of fungal functional guilds, and changes in these microorganisms were associated with grazing cessation-induced shifts in plants and soil (Wang et al. 2020a). The amount of litter produced by grazing signi cantly affects the relative abundance of fungal saprophytes, while longterm overgrazing can cause changes in the development of the soil AMF system and adversely affect grassland ecosystem production (Guo et al. 2016;Hamonts et al. 2017). Thus, soil abiotic conditions can modify plant community composition-production relationships via changes in the biotic interplay among fungal groups as well as via changes in their interactions with plants. However, to our knowledge, empirical evidence on how fungal communities in surface and subsurface soils integrate the linkages of plant community composition and production under grassland disturbances is lacking.
The Inner Mongolian grassland is one of the best-known rangelands in the eastern part of the Eurasian steppe (Wang 2004). Here, Baoyin et al. (2014) concluded that the best practice for haymaking from grassland should be based on the rule of 'mowing once a year' and a light grazing early in the season may increase hay quality mowed in the autumn if the grassland grows well early in the season in highproduction years. Here, we examined the impacts of different grassland disturbances (grazing, mowing, and grazing+mowing) on the plant community and soil fungal guilds in different soil horizons of steppe grasslands in China. We further characterized the relationships between the characteristics (richness, abundance, and composition) of three fungal guilds (saprophytic fungi, mycorrhizal fungi, and potential plant pathogens) and plant community composition-production relationship. We tested the following two hypotheses: rst, the fungal guilds primarily involved in plant community composition-production relationships differ between surface and subsurface soils; second, plant community compositionproduction relationships are mainly in uenced by fungal functional guilds in the surface soil. and Stipa krylovii Roshev in concert taken up more than 90% of the whole above-ground biomass.

Study design
The experiment was established in 2012 to study the effects of grassland for mowting and grazing on steppe ecosystems. In a randomized block design (Fig. S1), each of the four grassland management treatments had three replications, resulting in a total of 12 paddocks (33.3 × 33.3 m 2 ). The experiment included four treatments: no treatment (Control: enclosure and grazing prohibited since 2012), grazing (G: grazing twice in June & August each year), grazing + mowing (GM, grazing once in June and mowing once in August each year), and mowing (M, mowing once in August each year). Six Ugandan sheep were grazed in the grassland paddock on the 20 th of the designated month. The ock was removed once the average remaining height of L. chinensis was 6 cm, which corresponded to moderate. Mowing treatments were implemented every year at the time of peak plant biomass (20th of the designated month) at the stubble height of approximately 6 cm (Fig. S1).

Plant community investigation, soil sampling and analysis
Soil samples and plant were collected in belated August 2019. Plant communities were surveyed within each plot using three 1 × 1 m sample boxes to measure plant species abundance and above-ground biomass. In each sample box, plant species were identi ed and the abundance of each species was calculated by bundle or stem. Plant above-ground biomass (AGB) was found out through baking them at 65• C for 48 h to keep their weight unvarying and mowing all living plants to the land.
After the vegetation community investigation, an auger of 7 cm was used by us in diameter to gather surface (0-10 cm) and subsurface (10-20 cm) soils from each quadrat. The samples were placed in liquid nitrogen immediately and took back to the lab. A portion of the fresh soil samples were stored frozen at -80°C and used to measure soil bacterial and fungal communities. A portion of the fresh soil samples were stored frozen at -20°C and used to measure soil moisture (SM), bulk density (SBD), pH, ammonium nitrogen (NH4 +-N), nitrate nitrogen (NO3--N) and available phosphorus (SAP). A portion of the fresh soil samples were air dried and used to measure soil total carbon (STC), total nitrogen (STN) and separate (sandy, 50-2000 μ m; silt, 2-50 μ m; and clay, < 2 μ m). NO 3 --N and NH 4 + -N were determined using a Flow-Solution analyzer (AA3; SEAL, Germany). Soil total carbon (STC) and total nitrogen (STN) were determined with the element analyzer (Vario MACRO CUBE, Germany). The soil pH was measured using a pH meter (UV-1800; Shimadzu) for a 1: 5 soil: water suspension. SAP was determined using an ultraviolet-visible (UV-vis) light spectrophotometer (UV-1800; Shimadzu) via the molybdenum blue method. The soil separate was measured by Microtrac (USA). According to USDA, the soil separate was   (Table S1). To investigate plant community composition-production relationships, we focused on the roles of three functional guilds, namely saprotrophic fungi, potential plant pathogens, and mycorrhizal fungi.

Data analyses
One-way ANOVA was used to calculate the differences in plant aboveground production, richness of fungal functional guilds, and properties of surface and subsurface soils under different grassland disturbances using the 'ANOVA' function in the multcomp package in R (v. 3.3.2; R Core Team, 2017).
Paired t-test (using the 't. test' function in R) was used to test the overall differences among the plant, microbial, and soil properties. Using the 'cap scale' function of the vegan package in R, principal coordinate analysis (PCoA) and analysis of similarity (ANOSIM) were carried out based on Bray-Curtis distance. The rst coordinate of the PCoA (PCoA1) was used to represent the major variation in plant community composition (PCOM) and fungal functional composition in subsequent analyses.
Relationships between plant community composition (PCoA1 of plants) and aboveground production were examined using linear regression analysis. Random forest (RF) classi cation (Breiman 2001) was conducted to predict which of the three fungal functional groups in surface and subsurface soils were primarily associated with plant community composition and aboveground production in terms of abundance, richness, and community composition. The importance of each fungal characteristic in which the percentage increased in the mean square error (MSE) was determined using the 'randomForest' function of the randomForest package v. 4.6-14 in R. Signi cance level of the entire model and crossvalidated R 2 values were examined using 5,000 permutations of the response variables (plant community composition and aboveground biomass) via the 'a3' function of the A3 package. Subsequently, structural equation modeling (SEM) was performed using the AMOS 22.0 software (SPSS, IBM, NY, USA) to obtain a systematic understanding of direct and indirect relationships between fungal characteristics, soil abiotic variables, and plant components. Additionally, SEM was used to evaluate whether the relationships between soil fungal characteristics, plant community composition, and aboveground production were maintained after simultaneously accounting for multiple factors. A priori model was first established based on the known effects and relationships among these variables (Table  S2; Fig. S2). The data set was fitted to the model using the maximum likelihood estimation method. The initial model was adjusted and re-evaluated by stepwise removal of non-signi cant relationships until all pathways represented signi cant contributors to the nal model. It should be noted that there is no single universally accepted goodness of t test for SEM. Thus, evaluation of goodness of t of the nal model was based on the chi-squared (v2) test, and the t was deemed acceptable when 0 ≤ v2/df ≤ 2 and 0.05 < P ≤ 1.00 and when the root-mean-square error of approximation (RMSEA) was 0 ≤ RMSEA ≤ 0.05 and 0.10 < P ≤ 1.00.
To mitigate deviations from normality, before the analysis, the abundances of rhizosphere fungal guilds were square transformed, and aboveground biomass and soil properties (soil moisture, soil bulk density, pH, NO 3 --N, NH 4 + -N, SAP, STN, STC, soil C/N, and soil separate) were Ln transformed. All analyses were conducted independently for the surface and subsurface soils.

Plant and soil characteristics in grasslands under different disturbances
Plant aboveground production was signi cantly lower in the disturbed grasslands than in the control grasslands (Fig. 1a). A negative linear relationship was observed between plant community composition and plant aboveground production (Fig. 1b). Disturbances altered the composition of plant functional groups, with a substantial decrease in the relative abundance of annual/biennial forbs and a substantial increase in perennial forbs (Fig. 1c). Plant community composition differed between the disturbance treatments and the control according to principal coordinate analyses (PCoA) based on the Bray-Curtis distance ( Fig. 1d; PERMANOVA: F = 6.1706, R 2 = 0.481, P = 0.002).
In the surface soil (0-10 cm), grazing, grazing+mowing, and mowing resulted in signi cantly higher soil bulk density and soil C:N than those in the control, but with no differences among these variables under grazing, grazing+mowing, and mowing (Table S3). Among the three disturbances, grazing resulted in the lowest STN and STC, whereas mowing resulted in the highest STN and STC (P < 0.05; Table S3). In the subsurface soil (10-20 cm), no signi cant differences in any soil physical and chemical factor were found among the disturbances and the control (Table S3). Signi cantly lower soil moisture and silt content were observed in the surface soil than in the subsurface soil, whereas the content of sand was higher in the surface soil than in the subsurface soil (P < 0.05) (Table S3).
In the surface soil, grazing signi cantly reduced the richness of saprophytic fungi and potential plant pathogens compared to those in the control, and grazing+mowing signi cantly reduced the richness of saprophytic fungi compared to that in the control. However, there were no differences in the richness of mycorrhizal fungi between control and disturbed grasslands (Fig. 2c). However, in the subsurface soil, the richness of all fungal functional guilds was not signi cantly changed by grazing, grazing+mowing, and mowing compared to that in the control (Fig. 2d).

Relationships between plant production-community composition and associated fungal guilds
Correlations were observed between pairs of fungal guilds in both surface and subsurface soils, and they varied substantially between the horizons (Fig. S4). In the surface soil, according to the random forest analysis and prediction results, plant aboveground production was co-predicted by abundance of potential plant pathogens and abundance of saprotrophic fungi, while community composition was copredicted by abundance of potential plant pathogens and PCoA axis 1 for saprotrophic fungi (SaPCoA1) (Fig. S5a, b). In the subsurface soil, the random forest analysis indicated that the richness of saprotrophic fungi and PCoA axis 1 for potential plant pathogens (PpPCoA1) were the main fungal variables for predicting plant aboveground production. Richness of mycorrhizal fungi and PpPCoA1 were the main fungal variables for predicting plant community composition (Fig. S5c, d). Concerning the guild characteristics identi ed as the key predictors of both plant aboveground production and community composition, richness of mycorrhizal fungi showed no correlation with PpPCoA1 in the subsurface soil ( Fig. S4).
In the surface soil, the abundance of potential plant pathogens had signi cantly positive linear relationship with plant aboveground production and a signi cantly negative linear relationship with plant community composition (Fig. 3a, b). Partial regression analysis was carried out without considering the in uence of soil abiotic variables, and the results indicated that the abundance of potential plant pathogens had a signi cantly positive linear relationship with plant aboveground production. Furthermore, abundance of saprotrophic fungi had a signi cantly negative linear relationship with plant community composition (Fig. 3a, b; Table S4). The VPA showed that the combined effects of abundance of potential plant pathogens, abundance of saprotrophic fungi, and edaphic variables explained more of the variation in plant aboveground production (34.92%) than either of these factors did alone; the combined effects of abundance of potential plant pathogens and abundance of saprotrophic fungi explained 22.32% of the variation in plant aboveground production (Fig. 3c). Edaphic variables alone explained more of the variation in plant community composition (73.31%), followed by the combined effects of the abundance of potential plant pathogens, abundance of saprotrophic fungi, and edaphic variables, which explained 36.34% of the variation in plant aboveground production (Fig. 3d). Together, plant aboveground production and community composition accounted for most of the variation in the abundance of potential plant pathogens and abundance of saprotrophic fungi in the surface soil (Fig. 3e, f).
In the subsurface soil, although the richness of mycorrhizal fungi showed no correlation with either plant aboveground production or community composition, PpPCoA1 had a signi cantly negative linear relationship with plant aboveground production and no correlation with plant community composition (Fig. S6d, f). After controlling for the effects of shared edaphic variables, the relationships of PpPCoA1 and richness of mycorrhizal fungi with both plant aboveground production and plant community composition were still signi cant (Fig. 4a, b; Table S5). The VPA showed that PpPCoA1 explained most of the variation in plant aboveground production (21.69%); however, compared to PpPCoA1 and richness of mycorrhizal fungi, edaphic variables explained more of the variation in plant community composition (Fig. 4c, d). Moreover, plant aboveground production and community composition together accounted for most of the variation in PpPCoA1 in the subsurface soil (Fig. 3e).
SEM explained 93.8% and 60.6% of the variation in plant aboveground production and community composition in the surface soil, respectively ( Fig. 5a; Table S6). Furthermore, it explained 97.8% and 78.8% of the variation in the abundance of potential plant pathogens and abundance of saprotrophic fungi in the surface soil, respectively (Fig. 5a). In the surface soil, plant aboveground production was directly and signi cantly negatively affected by NO 3 --N and positively affected by NH 4 + -N, whereas plant community composition was directly and signi cantly negatively affected by soil available phosphorous and positively affected by NO 3 --N (Fig. 5a). Direct and positive impacts of soil moisture and NO 3 --N, and negative impacts of soil bulk density, available phosphorous, and NH 4 + -N on the abundance of potential plant pathogens were observed. Soil NH 4 + -N and NO 3 --N directly and positively affected the abundance of saprotrophic fungi. In contrast, soil moisture, bulk density, and available phosphorous directly negatively affected the abundance of saprotrophic fungi (Fig. 5a). Abundance of saprotrophic fungi and plant community composition had indirect effects on plant aboveground production by in uencing the abundance of potential plant pathogens. Abundance of potential plant pathogens had signi cantly positive effects on plant aboveground production (Fig. 5a). Furthermore, plant community composition and abundance of saprotrophic fungi had signi cantly negative effects on the abundance of potential plant pathogens (Fig. 5a). SEM explained 94.1% and 40.7% of the variation in plant aboveground production and community composition in the subsurface soil, respectively ( Fig. 5b; Table S7).
Furthermore, it explained 6.6% and 66.2% of the variation in PpPCoA1 and richness of mycorrhizal fungi in the subsurface soil, respectively (Fig. 5b). In the subsurface soil, plant aboveground production was directly and signi cantly positively affected by NO 3 --N and NH 4 + -N (Fig. 5b). Richness of mycorrhizal fungi was directly and signi cantly positively affected by soil moisture, available phosphorous, and NO 3 --N (Fig. 5b). PpPCoA1 directly and signi cantly positively affected plant community composition as well as plant aboveground production. Plant aboveground production was also signi cantly negatively affected by PpPCoA1 (Fig. 5b). Furthermore, richness of mycorrhizal fungi had negative effects on PpPCoA1 (Fig. 5b).

Discussion
Associations of fungal guilds from the two soil horizons with plant production and community composition under disturbances In contrast, no signi cant changes were observed in the richness of mycorrhizal fungi under the grazing, mowing, and grazing+mowing treatments. However, these changes were less pronounced in the subsurface soil than in the surface soil. Grassland disturbances signi cantly altered the community composition of potential plant pathogens in both the surface and subsurface soil, as well as the community composition of saprotrophic fungi in the subsurface soil. Grassland disturbance in uences plant indexes (including plant species diversity, abundance, aboveground biomass, and functional composition), which can affect the spread and infection of potential plant pathogens (Rottstock et al. 2014). Therefore, the addition of exogenous nutrients to the surface soil, especially the return of feces and urine, was the main factor causing the changes in abundance and community composition of saprotrophic fungi and potential plant pathogens. Mycorrhizal fungi obtain carbohydrates from host plants (Parniske 2008; Ji & Bever 2016); therefore, improving plant aboveground production can increase the richness of mycorrhizal fungi (Wagg et al. 2011). In the present study, we observed no signi cant changes in the richness and community composition of mycorrhizal fungi and dominant plant species.
The species-energy theory explains that species number increases with the increase in total available energy (Whittaker 2006). In natural ecosystems, structure and function of the surface soil microbiome may be strongly altered as a consequence of shifts in plant communities related to disturbances (Bahram et al. 2020). In the present study, we found a negative relationship between plant production and community composition, which was consistent with the ndings of Fraser et al. (2015). Furthermore, we observed positive associations between the abundance of potential plant pathogens (APp) and plant production (APP), as well as negative associations between the abundance of saprophytic fungi in the surface soil and plant production. There is an antagonistic relationships between the abundance of saprophytic fungi and potential plant pathogens in surface soil, which may be related Interaction between fungal guilds driving the plant production-community composition relationship vary between soil horizons In the present study, signi cant differences in the interactions between functional fungal communities were observed between different soil horizons. This was supported by previous observations of natural habitats with contrasting abiotic conditions (Grau et al. 2017), indicating that the interactions among fungal functional groups are environmentally dependent. Chen et al. (2019) found that the abundances of saprotrophic and mycorrhizal fungi were positively correlated at sites with high nutrient contents, and that the abundances of mycorrhizal fungi and potential plant pathogens were negatively correlated at sites with low nutrient contents. In the present study, in the surface soil, plant production and community composition could jointly explain the higher rates of variation in pathogenic and saprophytic fungi, suggesting that plants also simultaneously in uence fungal communities. The abundance of potential plant pathogens and saprophytic fungi together explained greater rates of variation in plant production and community composition, suggesting that these factors have a strong correlation with plant production and community composition. Yang et al. (2019) reported that high organic C and N contents in the soil were bene cial to the propagation of soil fungi, and they observed a high positive correlation between soil fungal diversity and organic C and N levels. Therefore, in the present study, the increased

Conclusion
Empirical evidence has improved our understanding of the importance of soil fungal communities in the plant community composition-production relationship. In the present study, the grassland disturbance experiment conducted over the past decade has allowed us to observe stable relationships between plants, soils, and microorganisms, and our ndings can be used as a basis for future studies on the relationships between plants, soil, and microbes. Our study showed that under grassland disturbance conditions, plant community composition and production were correlated with a soil horizon-related interplay between fungal guilds that have distinct biotic interactions with plants (specifically saprotrophs, mycorrhizal fungi, and potential plant pathogens). These co-existing fungal groups interact to drive plant community composition-production relationships, with fungal pathogens playing an important driver role.
Furthermore, different fungal mechanisms related to the linkages of plant composition and production between soil horizons illustrate how the overall negative linkage of plant composition and production can be maintained under grassland disturbance conditions.

Declarations
To be used for all articles, including articles with biological applications Funding This work was nancially supported by the National Natural Science Foundation of China (31860681) and the China Agriculture Research System (CARS-34).

Con icts of interest/Competing interests
We declare that we have no nancial and personal relationships with other people or organizations that can inappropriately in uence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as in uencing the position presented in, or the review of, the manuscript entitled, "Plant production-community composition relationship associated with fungal guilds depends on grassland disturbance in surface and subsurface soils".

Availability of data and material
All data generated or analyzed during this study are included in this published article [and its supplementary information les].

Code availability
All models and code generated or used during the study appear in the submitted article.     Details for model tness are presented below each gure. Signi cance levels are as follows: *, P < 0.05;