Impacts of 10 Years of Elevated CO2 and Warming on Soil Fungal Diversity and Network Complexity in a Chinese Paddy Field

Climatic change conditions (elevated CO2 and warming) have been known to threaten agricultural sustainability and grain yield. Soil fungi play an important role in maintaining agroecosystem functions. However, little is known about the responses of fungal community in paddy field to elevated CO2 and warming. Herein, using internal transcribed spacer (ITS) gene amplicon sequencing and co-occurrence network methods, the responses of soil fungal community to factorial combinations of elevated CO2 (550 ppm), and canopy warming (+2 °C) were explored in an open-air field experiment for 10 years. Elevated CO2 significantly increased the operational taxonomic unit (OTU) richness and Shannon diversity of fungal communities in both rice rhizosphere and bulk soils, whereas the relative abundances of Ascomycota and Basidiomycota were significantly decreased and increased under elevated CO2, respectively. Co-occurrence network analysis showed that elevated CO2, warming, and their combination increased the network complexity and negative correlation of the fungal community in rhizosphere and bulk soils, suggesting that these factors enhanced the competition of microbial species. Warming resulted in a more complex network structure by altering topological roles and increasing the numbers of key fungal nodes. Principal coordinate analysis indicated that rice growth stages rather than elevated CO2 and warming altered soil fungal communities. Specifically, the changes in diversity and network complexity were greater at the heading and ripening stages than at the tillering stage. Furthermore, elevated CO2 and warming significantly increased the relative abundances of pathotrophic fungi and reduced those of symbiotrophic fungi in both rhizosphere and bulk soils. Overall, the results indicate that long-term CO2 exposure and warming enhance the complexity and stability of soil fungal community, potentially threatening crop health and soil functions through adverse effects on fungal community functions.


Introduction
The global atmospheric carbon dioxide (CO 2 ) concentration has increased from 280 in 1850 to 400 ppm today and is predicted to reach 550 ppm by the middle of this century, accompanied by an increase in global mean temperature by 2 °C [53].As the most important staple crop, rice (Oryza sativa) provides food for more than half of the world's population [46].Paddy fields were the important nutrient cycling interface between the land and atmospheric CO 2 [13,55].Elevated atmospheric CO 2 could increase rice grain production and the biomass [34,61], which could be potentially offset by concurrent warming [16,61].Higher CO 2 levels promote photosynthesis and increase the amount of root exudates, thereby stimulating the growth and activity of microorganisms and changing the structure and function of microbial communities [15,24,58].
Soil fungi play an important role in regulating soil ecosystem functions (Yu et al., 2016), and fungal diversity contributes to ecosystem stability and maintains plant diversity [2,57].Hayden et al. [32] conducted climate change simulation studies on soil microorganisms in Australian grasslands and found that elevated CO 2 significantly increased the gene abundance and changed the composition of the fungal communities.CO 2 enrichment was found to increase the gene abundance and diversity of soil fungi in an agricultural ecosystem [38,39].However, Tu et al. [56] reported that long-term elevated CO 2 did not significantly alter the overall structure and species richness of the fungal community but significantly increased community evenness and diversity.Liu et al. [39] reported that warming significantly decreased the abundance of soil fungi in wheat rhizosphere.Recently, it was shown that the diversity of the soil fungal community was lower under warming conditions [1].Deslippe et al. [23] showed that long-term warming resulted in a significant increase in the relative abundance of fungi in Arctic tundra soil.However, studies have also shown that elevated CO 2 and warming do not considerably affect the abundance, community composition, or activity of soil fungi [3,8].Thus, there are still some contradictions concerning the dynamic changes in soil fungal communities with elevated CO 2 levels or warming.
Microbial interactions can form complex networks that collectively act on ecosystem functions [4,43].According to previous studies, the soil microbial correlation network model can vary with environmental factors (e.g., drought, warming and elevated CO 2 concentration).Tu et al. [56] analyzed the responses of soil fungal communities to long-term elevated CO 2 in a Minnesota field and found that elevated CO 2 increased the complexity of the fungal community network.Warming was found to significantly alter the diversity, structure, and network complexity of soil fungal communities [69].Recently, it was shown that long-term warming increased the complexity and abundances of keystone species of a microbial network [68].Another study reported that elevated CO 2 simplified the soybean rhizosphere soil fungi network structure by changing the keystone species members.However, the symbiotic pattern of fungal networks and their responses to elevated CO 2 and warming in paddy soil are still not fully understood [56].The aim of this study is to determine the effects of elevated CO 2 , warming, and their combination on the diversity and network complexity of fungal communities in a paddy field.It was hypothesized that the diversity and network complexity would increase when exposed to elevated CO 2 , but the positive impacts could be offset by warming.To test this hypothesis, a 10-year open free-air experiment with factorial elevated CO 2 (550 ppm) and warming (by + 2 °C) was conducted in a wheat and rice rotation ecosystem.

Site Description and Experimental Setup
The field experiment with free-air CO 2 enrichment and warming facilities was established in 2010 in Guli Township (31°30′N, 120°33′E), Changshu, Jiangsu Province, China.In this region, the traditional cropping system is the rotation of summer rice and winter wheat.The area has a typical subtropical monsoon climate, with an average annual temperature of 16 °C and average annual precipitation of 1100-1200 mm.The soil was derived from clay lacustrine deposit and classified as Gleyic Stagnic Anthrosol, with pH (H 2 O) of 7.0, soil organic carbon (SOC) content of 16.2 g•kg −1 , and total nitrogen (TN) content of 1.9 g•kg −1 .
The climate change treatments included ambient environmental conditions (CK); CO 2 increased to 550 ppm (CE); temperature increased by 2 °C (WA); and combined elevated CO 2 and temperature (CW).Each treatment consisted of three replicates, and each replicate was conducted in an octagonal ring of 8-m diameter (an area of approximately 50 m 2 ).In total, 12 octagonal rings were set up in this experimental field, and each ring was buffered by 28 m of open field to minimize any treatment cross-over effects (Fig. S1).For the CO 2 enrichment treatments (CE and CW), pure CO 2 gas was pumped from a storage tank and injected through perforated tubes surrounding the rings.For the warming treatments (WA and CW), 12 infrared heating lamps were mounted 1.2 m above the rice canopy in each ring.The treatments with elevated CO 2 and warming were maintained consistently over the entire rice growing period.More details of the experimental facility layout, performance, and operation are described by Liu et al. [37].

Rhizobox Application and Sample Collection
Rice (Oryza sativa L. cv.Changyou 5) was transplanted with a density of 26 hills per m 2 on 10th June, 2020, and harvested on 27th October, 2020.Rhizosphere soil was collected using a rhizobox inserted in each ring plot 2 days before rice transplantation.The dimensions of the rhizobox were 8 × 8 × 15 cm (length × width × height) (Fig. S2).The rhizobox was divided into three sections by a nylon net with a pore diameter of 30 μm to keep in the roots, allowing water and nutrient passage [66].To ensure full contact between soil and roots, two rice plants were planted in each rhizobox.The soil in the middle compartment of the rhizobox was treated as rhizosphere soil, and the left and right sides were bulk soil.Soil samples were collected at the tillering, heading, and ripening stages during the rice growth season.The rhizoboxes containing rice plant were lifted from the field soil.The rice was gently separated from the soil and carefully collected through washing using sieves and sterile water.Rhizosphere soil was washed off from roots by vortexing in sterile phosphate-buffered saline (PBS) solution.After 10 min of centrifugation at 5000 g at 25 °C, we removed the supernatant and collected the pellet for DNA extraction.A total of 72 samples (four treatments × three replicates × three stages × two kinds) were collected from the experimental facilities.

Soil Total DNA Extraction and Bioinformatic Analysis
The total DNA was extracted from 0.5 g soil using a SPINeasy DNA Kit for Soil (MP Biomedicals, LLC), according to the manufacturer's instructions.The concentration and quality of total DNA were assessed by 1% agarose gel electrophoresis and a NanoDrop spectrophotometer.The fungal ITS1 region was amplified with the primers ITS1F (5′-CTT GGT CAT TTA GAG GAA GTAA-3′) and ITS2 (5′-GCT GCG TTC TTC ATC GAT GC-3′) [18].The PCR amplification product was purified with the SanPrep Column PCR Product Purification Kit (Sangon Biotech Co., China) and quantified using QuantiFluor™-ST (Promega, USA).The purified amplicon libraries were sequenced on the Illumina MiSeq platform.The raw sequences were quality-filtered using the Quantitative Insight into Microbial Ecology (QIIME) software [17] to remove barcodes, primers, fuzzy bases, and low-quality sequences (less than 200 bp).For the remaining sequences, the ITS1 region was detected and extracted with fungal ITSx software before processing [7].The chimeric sequences were detected and removed using the Mothur software [50], and the non-chimeric sequences were clustered into operational taxonomic units (OTUs) with a 97% similarity cutoff using the QIIME software.According to the NCBI GenBank database, the representative sequences in OTU were classified and identified using the BLAST algorithm.Alpha diversity, including OTU richness and Shannon index, was calculated using the Mothur software.The functional prediction of the fungal community was conducted using the FUNGuild database [42,47].

Network Construction and Analysis
Molecular ecological networks were used to analyze the intradomain correlation of microbial species at multiple taxon levels.To assess the impact of climate change on the complexity and specificity of the soil fungal community, network analysis was conducted for each treatment.All analyses were conducted in a pipeline http:// mem.rcees.ac.cn: 8081 available online [28].The OTUs which occurred in more than half of the samples were retained without log-transformation prior to obtaining the Spearman correlation coefficient matrix.Based on the random matrix theory (RMT), a uniform threshold was determined for each microbial network.After network construction, the topology characteristics and network randomization were implemented in MENAP.Among-module and intermodule connectivity was computed based on the detected modules, and nodes were assigned to network hubs, module hubs, connectors, or peripherals.Finally, networks were visualized in Gephi (version 0.9.2; https:// gephi.org/) with a Fruchterman-Reingold layout algorithm.
To determine the role of nodes in the network, the topological role of each node of the microbial network was defined according to the within-module connectivity (Zi) and among-module connectivity (Pi) of the nodes of the molecular ecological network of the soil microbial community.Network nodes were divided into four categories: (1) peripherals (Zi ≤ 2.5, Pi ≤ 0.62), with few connections, which are basically connected to the internal nodes of the module; (2) module hub (Zi > 2.5, Pi ≤ 0.62), which is highly connected with nodes inside the module; (3) connector (Zi ≤ 2.5, Pi > 0.62), which is highly connected with nodes of other modules; and (4) network hub (Zi > 2.5, Pi > 0.62), which is highly connected to the nodes of other modules as well as those inside the module.It is generally believed that nodes with Zi > 2.5 or Pi > 0.62 are key nodes and play an important role in the connection with nodes within or between modules [49,54].

Statistical Analysis
All statistical analyses were performed using SPSS 20.0 (SPSS Inc., USA), and data visualization was carried out in R (version 4.1.2).The significant difference between climate change treatments (P < 0.05) was tested by one-way ANOVA, followed by Duncan's test.Principal co-ordinate analysis (PCoA) was conducted to determine the fungal community distribution using the R package "vegan" with the "Adonis" function.Repeated measures ANOVA was employed to determine the primary effects of elevated CO 2 , warming, and growth stage.

Diversity and Richness of the Soil Fungal Community
After quality filtering, a total of 4,757,273 qualified reads were obtained from 72 samples, with 52,697-70,616 sequences per sample (Table S1).The coverage of each sample was 99.96 to 100.00%, indicating that the sequencing intensity was sufficient to detect the fungal diversity in all samples (Table S3).
As shown in Table 1, the OTU richness and the Shannon index were significantly increased under elevated CO 2 but only slightly reduced under warming treatment, and these effects were influenced by growth stages.In rice rhizosphere and bulk soils, elevated CO 2 increased the OTU richness and Shannon index by 9.0-9.7% and 5.3-10.4% (P < 0.001), respectively.The increases in OTU richness and Shannon under elevated CO 2 were much larger at the heading and ripening stages than at the tillering stage (P > 0.05).Warming had no effect on OTU richness or Shannon index in bulk soil, but it decreased OTU richness in the rhizosphere (P = 0.023).At the ripening stage, warming had no effect on OTU richness, whereas at the tillering and heading stages, OTU richness was reduced by 15.68% and 14.59%, respectively.There was no interactive effect of elevated CO 2 and warming on OTU richness and Shannon index for both rhizosphere and bulk soils: the positive effects of elevated CO 2 on OTU richness and Shannon index were moderated by warming.
Additionally, there was a significant effect of interaction between CO 2 and rice growth stage on Shannon index in bulk soil (P < 0.001).

Structure Composition of the Soil Fungal Community
Ascomycota (29.5-68.8%),Rozellomycota (2.5-21.8%),Mortierellomycota (1.4-15.9%),and Basidiomycota (1.5-8.3%) were the dominant fungal phyla in rice rhizosphere and bulk soils across all treatments (Fig. 1A).Elevated CO 2 significantly decreased the relative abundance of Ascomycota by 11.6-16.7%and increased that of Basidiomycota by 39.6-57.4% in rhizosphere and bulk soil (Table S3).Warming resulted in a significant increase in the relative abundances of Mortierellomycota and Basidiomycota, and the interaction between CO 2 and warming had a significant effect only on the relative abundance of Basidiomycota (P < 0.001).Additionally, a significant effect of interaction between CO 2 and growth stage on the relative abundance of Basidiomycota was observed: elevated CO 2 increased the relative abundance of Basidiomycota at the heading and ripening stages but had no impact at the tillering stage.PERMANOVA and PCoA results revealed that the soil fungal communities were significantly affected by elevated CO 2 , warming, and growth stage (Table 2 and Fig. 1B).The fungal communities in the tillering stage were clearly separated from those in the heading and ripening stages along the PCoA1 axis (Fig. 1B).The PCoA showed a clear separation between CK and the other treatments (CE, WA, and CW) in both rhizosphere and bulk soils (Fig. S3).
As shown in Fig. S4, the relative abundances of the top 10 fungal genera (> 1%) were determined under elevated CO 2 and warming.The fungal genera were significantly affected by climate change only in the rhizosphere.Elevated CO 2 significantly increased the relative abundances of Pyrenochaetopsis, Cosmospora, and Metarhizium but reduced those of Pseudeurotium, Cladorrhinum, and Podospora.Warming significantly increased the relative abundances of Cladorrhinum and Podospora but decreased those of Fusicolla and Sutellinia in the rhizosphere.

Functional Composition of the Soil Fungal Community
Seven trophic modes were detected in this study.Among them, saprotrophs (61.7-82.9%),saprotroph-symbiotrophs (5.1-33.3%),and pathogens (1.5-15.1%)were the most abundant functional guilds across all treatments (Fig. S5).Individual functional guilds in rhizosphere and bulk soils showed diverse responses to elevated CO 2 and warming (Table 3).The relative abundances of saprotrophs and symbiotrophs were significantly decreased, whereas those of pathotrophs, pathotroph-saprotrophs, and saprotrophsymbiotrophs in both rhizosphere and bulk soils were increased.Warming significantly increased the relative abundances of pathotrophs and saprotroph-symbiotrophs and decreased those of symbiotrophs and pathotroph-saprotrophsymbiotrophs.Notably, significant effects of interactions between growth stage and CO 2 /warming were observed on the relative abundances of most fungal functional guilds.

Characteristics of the Fungal Co-occurrence Network
Based on the random matrix theory (RMT) method, cooccurrence networks of the soil fungal community were constructed for climate change treatments and growth stages, respectively (Fig. 2 and Fig. S6).Compared to CK, elevated CO 2 and warming greatly increased the complexity of the fungal network in both rhizosphere and bulk soils.Elevated CO 2 , warming, and their interaction strongly increased the edges, linkage density, average degree, especially the negative correlation, and decreased the average path distance and modularity (Table 4).Further analysis showed that the nodes, edges, linkage density, average degree, average clustering coefficient, as well as negative correlation of fungal networks were higher in the heading and ripening stages than in the tillering stage (Table S4).The network topologies of both rhizosphere and bulk soils differed significantly from those of randomly generated networks across all treatments (t test: P < 0.01), suggesting that the fungal co-occurrence network constructed in this study has features of small-world and modularity (Table 4 and Table S4).

Topological Role of the Fungal Network Nodes
To identify the role of the nodes in each network, Zi and Pi were calculated for each node.In the rhizosphere, one module hub and connector were observed in CK and CE treatments, two module hubs, and three connectors in WA, and one module hub and two connectors in CW (Fig. 3).The module hubs and connectors in CK, CE, and CW treatments were Ascomycota and Basidiomycota, whereas Ascomycota and Mortierellomycota were the module hubs and connectors in the WA treatment (Table S5).In bulk soil, there was one connector, namely Ascomycota, in CK, and there were five connectors, including Ascomycota and Basidiomycota, in WA and CW treatments.The module hub in CW belonged to Chytridiomycota (Table S5).Further analysis showed that there was one connector at the tillering stage, two module hubs, and two connectors at the heading stage, and three module hubs at the ripening stage in the rhizosphere (Fig. 4).The connectors at the tillering and heading stages belonged to Ascomycota, while the module hubs at the tillering, heading, and ripening stages belonged to Ascomycota, Basidiomycota, and Rozellomycota, respectively (Table S6).However, there was no module hub or connector at the tillering and heading stages, whereas one module and two connectors were observed at the ripening stage in the bulk soil.

Discussion
Soil fungal communities are formed by aboveground vegetation through root exudates during plant growth and development [9,14].This study found significant differences in the α-diversity indices of the soil fungal community across different rice growth stages.The OTU richness and Shannon diversity were higher at the heading and ripening stages than at the tillering stage, indicating greater physiological metabolic activity of rice at the latter stages of rice.The PCoA also showed that the soil fungal community of samples from the heading and ripening stages grouped together and were clearly separated from those at the tillering stage, suggesting that the fungal community is significantly influenced by rice growth stages.This is in agreement with Hannula et al. [31], who reported that different plant growth stages showed different soil nutrient contents and soil temperatures.Thus, there is a greater influence on soil fungal composition and diversity at the rapid vegetation growing stages.Breidenbach et al. [13] and Edwards et al. [26] found that rice growth stage can influence microbial community structure in the rhizosphere.In this study, elevated CO 2 significantly increased fungal OTU richness and Shannon diversity in both rhizosphere and bulk soils, consistent with the findings of previous studies [38,56].Elevated CO 2 stimulates C3 plant photosynthesis and root exudate production, leading to a greater soil organic C and C:N ratio, which facilitates fungal growth [10,11].In the same experiment, it was found that fungal α-diversity in rice and wheat soils increased by increasing the input of organic C and reducing the soil pH [30].In this study, warming had no effect on OTU richness and Shannon index, except for a significant decrease in OTU richness of the rhizosphere soil (P = 0.023).This is in agreement with the finding of Lorberau et al. [41] that warming does not alter fungal richness and diversity.However, a  significant increase in fungal diversity under warming was found in an alpine meadow [60].The distinct responses may be due to the differences in plant host and warming conditions (air canopy warming vs soil warming).In this experiment, the +2 °C warming of the air canopy resulted in a small increase in soil temperature (< 1 °C, [38]), which could be in the range of fungal growth fluctuations.
This study also found that elevated CO 2 and warming significantly changed the composition of the soil fungal communities in both rhizosphere and bulk soils.Elevated CO 2 and warming may indirectly shift the soil microbial diversity and composition by affecting plant and root growth and altering soil environmental factors (e.g., temperature, moisture, pH, and available C, among others) [25,29,63].In this study, Ascomycota, Rozellomycota, Mortierellomycota, and Basidiomycota were the dominant phyla across the treatments and growth stages.Elevated CO 2 significantly decreased the relative abundance of Ascomycota, which is consistent with the study by Tu et al. [56] in a grassland region.Lauber et al. [35] found that the abundance of Ascomycota was greater in soils with a higher soil pH, and suggesting that the decrease in soil pH under elevated CO 2 treatment in this rice field can inhibit the abundance of Ascomycota [30].In this study, the relative abundance of Basidiomycota under elevated CO 2 and warming were significantly increased in rhizosphere and bulk soils.CO 2 enrichment increase the input of soil organic carbon by higher photosynthesis and root production resulting with a Fig. 3 Role distribution of the fungal network nodes in rice rhizosphere (A) and bulk soils (B) under climate change treatments.CK, the ambient; CE, CO 2 enrichment; WA, canopy warming; CW, CO 2 enrichment plus canopy warming higher C:N ratio [10,11], which is favoring the growth of Basidiomycota.Under warming condition, the litter decomposition rate, plant biomass, and ecosystem C fluxes were generally increased by field experiment and meta-analysis [20,64], which provides higher nutrients for Basidiomycota.In addition to being influenced by climate change, rhizosphere soil microbes are also largely impacted by plant species [31].Shi et al. [52] showed that the composition of enriched and excluded microbes in the rhizosphere differed during different growth stages of soybean, further illustrating the regulation of eukaryotic microbes in the rhizosphere by plant root exudates.Compared with the tillering stage, the relative abundance of Basidiomycetes increased and that of Rozellomycota decreased at the heading and ripening stages, and the growth stage significantly changed the community composition of the soil fungi in rhizosphere and bulk soils.This is consistent with the report that climate change affects soil microbial communities that perform different functions at different growth stages [33].Further analysis indicated that the dominant fungal genera in different growth stages responded differently to elevated CO 2 and warming.
The observed OTUs were categorized into fungal functional guilds by the FUNGuild annotation tool [47].The distribution patterns of the fungal functional guilds were clearly influenced by elevated CO 2 , warming, and growth stage.Based on the above results, both elevated CO 2 and warming significantly increased the relative abundances of pathotrophic fungi.Although there is no study about plant to the ecosystem functions [68].In the present study, the co-occurrence networks of the soil fungal community in the heading and ripening stages were more complex than those in the tillering stage, based on the higher numbers of nodes and edges as well as the higher linkage density, average degree, and clustering coefficient in both rhizosphere and bulk soils.The more complex network indicates that in the later rice stages, soil nutrient availability for fungal communities is increased due to the high quality and quantity of root exudates and plant residues [36].The present results indicate that elevated CO 2 and warming altered the topological parameters of the fungal ecological network.In previous studies, average degree and linkage density have been commonly used to assess the complexity of microbial networks [22,45,59].Compared with CK, elevated CO 2 and warming treatments increased the linkage density, average degree, and edge number and decreased the modularity and average path distance of co-occurrence networks, indicating that the complexity of soil fungal networks was improved by elevated CO 2 and warming.These findings are consistent with previous studies reporting that elevated Fig. 4 Role distribution of the fungal network nodes in rice rhizosphere (A) and bulk soils (B) in the three growth stages.CK, the ambient; CE, CO 2 enrichment; WA, canopy warming; CW, CO 2 enrichment plus canopy warming pathogens under elevated CO 2 in agricultural ecosystems, a recent study using a global meta-analysis and a 9-year field experiment found that warming increased the abundances of fungal plant pathogens [21].Soil pathogenic fungi might proliferate under warming, affecting the functions and structure of the forest [40].In this study, the relative abundances of symbiotrophic fungi were significantly decreased under elevated CO 2 and warming.Symbiotrophic fungi provide nutrients and water for plant host under environmental stress, which plays an important role in soil health and crop production [51].In paddy soil, reduction in symbiotrophic fungi under elevated O 3 was ascribed to the decrease in plant photosynthesis and nutrient availability [62].Although FUNGuild is highly accurate, the ecological functions of many fungi remain unknown.In particular, the ecological functions of soil fungi under global climate change conditions need to be further studied and verified.
The microbial co-occurrence network constructed in this study is characterized by scale-free, small world, and modularity.The topological properties are used to define the complexity of the network, which is closely related CO 2 and warming increase the complexity of fungal networks [56,68].The positively correlated connections in the co-occurrence network represent the existence of mutual synergistic relationships among microorganisms, whereas the negatively correlated connections represent potential antagonistic effects [12,19].In this study, elevated CO 2 and warming increased the negative correlation in both rice rhizosphere and bulk soils, indicating that climate change conditions stimulated competitive relationships among fungal compositions.Ma et al. [44] found that microbial network complexity facilitated the growth of microbial flora, leading to more efficient use of soil nutrients.Previous studies found that elevated CO 2 and warming increased the contents of soil organic carbon, total nitrogen, and root exudates in rice paddy soil [30,38,65], which may increase the competition for soil nutrients among microbes.Additionally, previous studies found that the complexity of microbial networks is positively correlated with α-diversity [19,27], indicating that the increase in fungal OTU richness and Shannon index value under elevated CO 2 in this study may have led to the enhanced network complexity.Higher complexity of the microbial network means stronger stability of the whole microbial community, and the competitive relationships will also further enhance the stability [48,59,68].A recent study reported that long-term warming increased the complexity and stability of a microbial network in grassland soil, which are important for maintaining ecosystem functions [68].
This study also screened keystone species of the fungal community by analyzing the topology of the co-occurrence network.A total of six module hubs and 18 connectors were detected in all molecular ecological networks, which can be regarded as key nodes that play essential roles in forming the network structure [5].The number of module hubs and connectors was higher in the warming treatments than in the control, indicating that the fungal network is more complex under warming.This was further supported by the higher node and edge numbers as well as the increased linkage density, average degree, and clustering coefficient under warming conditions, compared to the control.These findings are in agreement with Zhou et al. [69], who reported that long-term warming can increase the abundance of keystone species and the complexity of the microbial network in the grassland ecosystem, which may be closely related to ecosystem functions.
The higher number of module hubs and connectors under warming condition suggests that the interactions, as well as energy and nutrient flows among the soil fungal community, were more efficient, compared to the control [67].In particular, most of the key nodes were affiliated to the phyla Ascomycetes and Basidiomycetes.Previous studies found that Ascomycota can decompose mainly degradable organic matter in soil, while Basidiomycota can decompose substances such as lignin and cellulose [6].However, the network was connected by Ascomycota and Basidiomycota, which are both parasitic and saprophytic organisms, facilitating nutrient and energy flow in this network.Moreover, the nodes categorized as module hubs and connectors in the warming network were different from those in the control, indicating that each OTU played different roles in warming and ambient networks.Overall, the warming-induced change in the network structure and the topological roles of the keystone OTUs may be related to soil nutrient availability.However, further studies are needed to determine the ecological functions of the keystone species.

Conclusions
This study shows that elevated CO 2 , warming, and growth stages clearly altered the diversity, composition, and network structure of fungal communities in both rhizosphere and bulk soils.Elevated CO 2 resulted significantly positive effects on fungal communities by increasing the diversity and network complexity, while warming had a little negative effect on fungal communities.The interaction of elevated CO 2 and warming had relatively positive effects on fungal communities, although warming offset part of the positive effects of elevated CO 2 .Elevated CO 2 and warming increased the network complexity and negative correlation of the fungal community, suggesting that both elevated CO 2 and warming enhanced the stability of the soil fungal community.Furthermore, the individual functional composition of the soil fungal community showed diverse responses to elevated CO 2 and warming.The responses of soil fungal communities were greater at the later growth stages than at the tillering stage, whereas α-diversity and network complexity were greatly increased at the heading and ripening stages.Overall, soil fungal communities were considerably altered by elevated CO 2 , warming, and growth stages, potentially influencing ecosystem function and threatening food production under future climate change conditions.

Fig. 1
Fig. 1 Relative abundance (A) and principal coordinate analysis (PCoA) (B) of the fungal community in rice rhizosphere and bulk soils under climate change treatments.CK, the ambient; CE, CO 2 enrichment; WA, canopy warming; CW, CO 2 enrichment plus canopy warming

Fig. 2
Fig. 2 The complexity and interactions of the soil fungal community in rice rhizosphere (A) and bulk soils (B) under climate change treatments.The size of a node is proportional to the relative abundance of OTUs, and the colors of nodes indicate different fungal phyla of

Table 1
The OTU richness and Shannon diversity of the soil fungal community in rice rhizosphere and bulk soils under climate change treatmentsThe various lowercase letters in a column each block indicate significance differences between climate change treatments at the P < 0.05 level CK the ambient, CE CO 2 enrichment, WA canopy warming, CW CO 2 enrichment plus canopy warming

Table 2
The statistical test of multifactorial permutational analysis of variance (PERMANOVA) to analyze the differences of fungal composition in rhizosphere and bulk soils with elevated CO 2 (CO 2 ), warming, and growth stage (stage)

Table 3
Effects of elevated CO 2, warming and growth stage on the relative abundance of fungal functional guilds in rhizosphere and bulk soils ns no significant (P > 0.05)***, **, * indicate significant levels with P < 0.001, P < 0.01, and P < 0.05, respectively A Main effects of elevated CO 2 calculated as ((CE + CW)/(CK + WA)−1) × 100 averaged across the three stages B Main effects of warming calculated as ((WA + CW)/(CK + CE)−1) × 100 averaged across the three stages

Table 4
Topological properties of co-occurrence network in rice rhizosphere and bulk soils under climate change treatments Parameters of random networks were generated from randomly rewired (100 times) empirical networks.The presented parameters are mean values and standard derivations of random networks.Significant differences between empirical networks and random networks were determined by t-test