Microbial community and network differently reshaped by crushed straw or biochar incorporation and associated with nitrogen fertilizer level

Straw returning has been demonstrated as a beneficial approach for the utilization of renewable biomass source, which contributes to reducing environmental pollution and strengthening the sustainability of agriculture. However, information on how microorganisms respond to different straw return modes (SRMs) at varying nitrogen fertilizer levels (NFLs) in the black soil is still limited. The community composition, network pattern, and modular function of bacteria and fungi are investigated under three SRMs, including straw removal (CK), crushed straw incorporation (SD), and biochar incorporation (BC) at three NFLs (0, 144, and 240 kg N ha−1, respectively) mainly using Illumina MiSeq technique based on a long‐term maize field experiment. Results showed that bacterial richness, diversity, and fungal richness decreased with NFL reduction. However, these decreases can be compensated by SD and BC, demonstrating superiority for BC at reduced NFLs. SD and BC differed in their effects on the bacterial and fungal abundances (showing increments only in SD) and fungal Shannon diversity (remaining stable only in BC irrespective of NFLs). Microbial communities were substantially affected by SRMs and interacted with NFLs, which were driven by soil NH4+‐N, available potassium, total nitrogen, and pH. In addition, SD induced a network characterized by its highly complex (average degree 10.259 vs. 3.364) and stable structure (average clustering coefficient 0.503 vs. 0.239), Ascomycota as predominating keystone taxa, and abundant N‐cycling related bacteria, while BC formed a network comprising a superior modular structure (modularity 2.599 vs. 0.912), dominant symbiotic fungi, and soil bulk density as specific shaping factor, indicating that network pattern, keystone taxa, modular function, and determining factors shifted between SD and BC co‐occurrence networks. These results deepen insights into the response divergence of bacteria and fungi to SRMs and NFLs, providing a scientific basis for selecting the suitable strategy for sustainable straw utilization in the black soil area.


| INTRODUCTION
According to statistics, the annual production of maize straw, one of the main types of crop straws, peaked at 67.83 million tons in 2020 in China (Mao et al., 2023). As an abundant and renewable biomass source, the proper utilization of maize straw can contribute to the reduction in global warming and environmental pollution (Koul et al., 2022;Shi et al., 2023;Zhao et al., 2018). Returning crop straws into the field is currently recognized as a promising and effective method to recycle nutrients, increase soil fertility, and improve crop yields Joseph et al., 2021;Wang et al., 2018). However, comparing and determining the best straw returning mode (SRM) for soil health and crop production is crucial to guide the utilization of crop straw and address the demand for economic efficiency. In recent years, crushed straw and biochar incorporation are the two common SRMs widely applied in the world Siedt et al., 2021). Crushed straw direct incorporation showed the advantages of saving labor and time and was regarded as a feasible SRM Turmel et al., 2015). Meanwhile, biochar incorporation has been considered as another promising SRM due to its large surface area and specific porous structure, high hydrophilicity and cation exchange capacity, and dominant recalcitrant components, which showed higher carbon sequestration potential and disease resistance compared to other SRMs (Ajayi et al., 2016;Lehmann et al., 2011;Yin et al., 2022).
The black soils, also known as Mollisols, Chernozems, Kastanozems, or Phaeozems, are recognized as the most fertile and productive soil and characterized by their rich organic matter and strong macroaggregate stability (Liu et al., 2012;Zhou et al., 2016). On a world-wide basis, most of the world's 916 million ha of black soils occur in four regions, in which the Northeast Plain of China is best represented by the eastern belt of black soil across central Asia Liu et al., 2012). However, the organic matter content has sharply decreased due to long-term intensive and destructive cultivation, leading to loss of stable aggregates, degradation of soil biological functions, and suppression of crop yield. These phenomena have seriously threatened the sustainability of agriculture in the black soil regions. Meanwhile, the overuse of chemical N fertilizer (NF) in agricultural farmlands has also resulted in the reduction of soil fertility, degradation of soil functions, and environmental pollutions, including soil acidification, soil microbial activities and diversity loss, N leaching, N 2 O emissions, and ammonia volatilization (Gong et al., 2022;Wang et al., 2019;Yang, Muhammad, et al., 2022). In recent years, straw direct incorporation and biochar incorporation are the two common SRMs applied in the black soil region of Northeast China to protect black soils resources (Fan & Wu, 2021;Guo et al., 2022).
One of the positive effects of straw returning was the capture and reuse of nutrients that would have otherwise been lost during the straw removal from the agroecosystem, reducing the requirement of chemical fertilizers (Gao et al., 2023). Furthermore, straw returning can retain soil nutrients for a longer term compared with the use of chemical fertilizers alone (Siedt et al., 2021). Considering chemical NF requirements, previous studies have demonstrated that straw returning has a considerable potential to reduce its usage Wang et al., 2020;Yin et al., 2018). Controversially, returning straw with a relatively high C/N ratio may need additional NF inputs to overcome the temporary reduction of the available N contents in soil (Alijani et al., 2019;Dianatmanesh et al., 2022). Numerous insights into the impact of straw returning on N nutrient in soil are already available, and straw returning can generally affect the following: (1) the concentrations of different N forms; (2) N retention, availability, dynamics, leaching, emissions, and N-cycle; (3) crop N uptake and NF use efficiency, through acting on soil properties or directly on the fertilizer itself (Borchard et al., 2019;Nguyen et al., 2017;Xia et al., 2018).
The responses of soil microorganisms, which are vital players of soil fertility and agroecosystem stability, must be explained to gain insight into the complete picture of these effects (Frąc et al., 2018;Zhang et al., 2018Zhang et al., , 2023. Soil microorganisms are an integral part of the soil and perform a series of important ecological functions, including litter decomposition, soil organic matter formation, and soil nutrient cycling (Hernández-Lara et al., 2023;Nkongolo & Narendrula-Kotha, 2020;Wu et al., 2022). Straw returning can impact the soil microbial diversity and community structure by altering space for their colonization, providing abundant carbon sources and other nutrients, and influencing other soil environmental factors (Cong et al., 2020;Siedt et al., 2021;Zhao et al., 2018). In addition, fertilization K E Y W O R D S bacteria and fungi, biochar, chemical nitrogen fertilizer, diversity and function, network and modular pattern, straw return modes practices impacted a diverse set of soil microbial properties Wang et al., 2019). Previous reports have suggested that a combination of straw and chemical fertilizers is considered the most beneficial treatment for soil bacteria communities and functions (e.g., increases in the richness of bacteria and fungi or the rate of N 2 fixation in soil); however, the effect is not always positive, which depends on straw characteristics, soil properties, and experiment conditions Gao et al., 2023;Yang, Muhammad, et al., 2022). Notably, the turnover rate, stress tolerance, and nutritional requirements of bacteria markedly differ from fungi. Considering straw decomposition, soil bacteria (e.g., phyla Proteobacteria, Acidobacteria, and Actinobacteria) play a big role in the degradation of easily decomposed organic compounds in the early stages, while soil fungi (e.g., phyla Ascomycota and Basidiomycota) mainly degrade highly recalcitrant organic compounds in later phases (Marschner et al., 2011;Li et al., 2015;Yang, Jing, et al., 2019;Yang, Ma, et al., 2019). In addition, soil bacteria and fungi respond differently to the NF input because the adaption of many fungal species to N limitation is better than bacteria, while the mineral N retained by bacteria is more than fungi (Gougoulias et al., 2014;Rousk & Frey, 2015;Wu et al., 2022;Zhou et al., 2016). The underlying mechanism of soil microbial responses to straw returning should cover microbial species interactions and potential ecological functions (Cui et al., 2022;Jiang et al., 2017;Morriën et al., 2017). The enhanced complexity in bacterial network caused by straw returning has been previously reported (Du, Zhang, et al., 2022;Yang, Bao, et al., 2022). However, whether and how cooccurrence network pattern of bacteria and fungi respond to different SRMs at varying NF levels (NFLs) in the black soils remain unclear.
The main objectives of the current study are as follows: (1) evaluate and compare the effects of different SRMs and NFLs on bacterial and fungal abundance, diversity, richness and community structure; (2) determine and compare the effect of soil microbial network patterns, keystone taxa, and modular functions by different SRMs at varying NFLs; (3) explore the links between soil environmental variables and soil microbial response at the community and key network module levels based on a long-term field experiment in the black soils. We hypothesized that microbial community and network shift under different conditions of straw incorporation combined with nitrogen fertilizer application. Therefore, this study is expected to provide theoretical reference for the optimal form of straw management and NF input and suggestions for the sustainable agricultural production.

| Experimental site, design, and soil sampling
A long-term maize (Zea mays L.) field experiment (since 2016) was conducted at the research and development base of conservation tillage (43°19′ N, 124°14′ E), Chinese Academy of Sciences, Gaojia Village, Dafangsheng Township, Lishu County, Jilin Province, China. The region has a temperate semi-humid continental monsoon climate with a mean air temperature, relative humidity, and total precipitation values of 5.8°C, 69%, and 577 mm, respectively. The soil of the experimental site was classified as Mollisol with a clay soil quality comprising initial properties of 11.30 g kg −1 soil organic carbon, 1.20 g kg −1 total N, 13.6 mg kg −1 available P, 218 mg kg −1 available K, and pH 7.1 under sampling at 0-20 cm depth. The experiment was conducted with three replications in randomized complete block design in split-plot arrangement. The SRMs and NFLs were the main plots and the subplots, respectively. The three SRMs are as follows: straw direct incorporation (SD; maize residues were harvested and mechanically crushed into 2-3 cm and directly mixed with soil in top 0-20 cm with rotary tillage at 10,000 kg hm −2 , which was equal to 30% of the seasonal production of maize straw amounts), biochar (from Jinhefu Biochar Company) incorporation (BC, maize straws were pyrolyzed under oxygen-limited conditions based on the 30% production amount of straw), and traditional planting (CK; maize straw was removed after crop harvesting). Meanwhile, the three NFLs included control (0%, N0), 144 kg N ha −1 (60%, N60), and 240 kg N ha −1 (100%, N100, recommended dose). Therefore, the treatments at the N0 level were CKN0, SDN0, and BCN0, whereas those at the N60 level were CKN60, SDN60, and BCN60. Meanwhile, the treatments at the N100 level were CKN100, SDN100, and BCN100. Phosphorus (110 kg P 2 O 5 ha −1 ) and potassium (110 kg K 2 O ha −1 ) fertilizers were applied at equal amounts for each treatment. Each subplot was 6.5 × 10 m with a buffer zone (two rows) between two adjacent treatments. In 2019, the maize cultivar "Xianyu-335" was sown on May 11th and harvested on October 9th. Five random soil samples (0-20 cm in depth) were collected from each sub-plot at the jointing stage of maize. After collection, soil samples were then mixed, sieved, and divided into the following two parts: microbial (stored at −80°C) and soil properties (air-dried at room temperature) analyses. Soil properties, including total nitrogen (TN), total carbon (TC), NH 4 + -N (AN), available phosphorus (AP), available potassium (AK), soil organic carbon (SOC), pH, water content (WC), and bulk density (BD), were determined in accordance with our previous work .

| Soil DNA extraction, quantitative PCR (Q-PCR), and Illumina MiSeq sequencing
The soil DNA extraction, PCR amplification, amplicons purification, library preparation, and sequencing were performed at GENEWIZ, Inc. The bacterial 16S rRNA gene and fungal internal transcribed spacer (ITS) fragment amplicons were sequenced on an Illumina MiSeq platform (Illumina), in which the universal primers 338F (ACTCC TAC GGG AGG CAGCA) and 806R (TCGGA CTA CHV GGG TWT CTAAT) and barcode primers ITS1F (CTTGG TCA TTT AGA GGA AGTAA) and ITS2R (GCTGC GTT CTT CAT CGATGC) were used, respectively. The raw sequence data have been accessioned in the sequence read archive of NCBI, United States (SRA accession: PRJNA985386 and PRJNA986449). Sequencing data were processed following the previously used method of Sun et al. (2023). Moreover, the bacterial and fungal functional groups were predicted by the FAPROTAX (Louca et al., 2016) and FUNGuild (Nguyen et al., 2016) functional annotations dataset, respectively. Moreover, the soilDNA was extracted and then the abundance of total bacteria and total fungi was assessed in triplicate by an ABI Real-Time 7500 system (Applied Biosystems) using primer sets F338/R806 and FITS1/RITS2, respectively. The Q-PCR reaction was performed in 25 μL reaction mixtures following the protocols reported previously (Yao et al., 2017;Zhou et al., 2016). Briefly, the copy numbers of 16S rRNA gene and ITS fragment were calculated using a regression equation to convert the cycle threshold value to the known number of copies in the standard curves, which were generated using serial dilutions of a plasmid inserting the 16S rRNA gene or ITS fragment. Meanwhile, negative controls were performed by replacing soil DNA with sterilized Milli-Q water.

| Statistical analyses
The results of gene or fragment copy numbers of 16S rRNA and ITS, alpha diversity indexes, and relative abundant (RA) of dominant taxa were analyzed using two-way ANOVA under three SRMs and three NFLs. The least significant difference was used to evaluate the differences among all treatments at p < 0.05. In order to visualize the microbial structural differences among the treatments, non-metric multidimensional scaling (NMDS) and analysis of similarities (ANOSIM) were carried out using the VEGAN package in R (version 3.6.1). To visualize the interaction among microorganisms, two integrated networks for two SRMs (SD vs. BC) based on combined samples (samples from different NFLs) were constructed in accordance with the works of Deng et al. (2012) and Jiao et al. (2020). Meanwhile, seven sub-networks, which included soil environmental factors, were also constructed for each key module in two integrated networks. Only OTUs with top 50 RA and strong connections (Spearman's correlation |r| > 0.60) with statistical significance (p < 0.01) were selected for network analysis. The networks were constructed and visualized in Gephi (v0.9.3), and network topological parameters were then calculated (Shi et al., 2016). Subsequently, the topological roles of nodes (proposed as keystone taxa, KT) in the network were inferred on the basis of two parameters: among module connectivity (Pi) with a threshold value of 0.62 and within module connectivity (Zi) with a threshold value of 2.5 (Olesen et al., 2007). Redundancy analysis (RDA) and Spearman correlation analysis were performed to examine the relationship between the microbial community or sub-community with the soil environmental factors.

| Abundances of soil bacteria and fungi
The copy numbers of bacterial 16S rRNA gene and fungal ITS fragment were significantly affected by NFLs (p < 0.01) and SRMs (p < 0.001), respectively (Figure 1a,b). As NFLs reduced from N100 to N0, the abundances of 16S rRNA gene (ranged from 7.51 × 10 6 to 3.40 × 10 6 copies g −1 dry soil) and ITS fragment (ranged from 6.30 × 10 5 to 3.15 × 10 5 copies g −1 dry soil) decreased in CK without straw incorporation. Compared with CK, SD resulted in their increments while BC demonstrated the opposite, both showing a declining trend with NFL reduction. The B/F ratio was insignificantly affected individually by SRMs or NFLs (p > 0.05). However, SD decreased this ratio by 53.1% compared with BC at the N100 level, while BC reduced this ratio by 24.6% compared with SD at the N0 level.

| Diversities of soil bacteria and fungi
Chao1 richness and Shannon diversity indexes were characterized in this study to assess the alpha diversity of microbial community. The bacterial Chao1 richness and Shannon diversity were significantly affected by SRMs (p < 0.001) and NFLs (p < 0.01), respectively; the interactive effects between SRMs and NFLs were also significant (Table 1). In this concern, compared with CKN100, the input of NF at reduced levels (CKN60 and CKN0) significantly decreased bacterial Chao1 richness (by 33.0% and 31.8%) and Shannon diversity (by 1.9% and 2.1%), respectively (p < 0.05). However, compared with CK series (including CKN100, CKN60, and CKN0), bacterial Chao1 richness and Shannon diversity were increased by SD and BC, respectively, specifically by BC at N60 and N0 levels (p < 0.05). Fungal Chao1 richness index was affected by SRMs (p < 0.01) and NFLs (p < 0.05) but without significant interaction (p > 0.05). Similarly, NF input at low levels reduced fungal Chao1 richness in CK without straw incorporation; however, this index was increased by SD or BC, specifically by BC at N60 and N0 levels (p < 0.05).
Differently, fungal Shannon diversity was significantly affected individually by SRMs (p < 0.001) but did not significantly respond to NFLs (p > 0.05). Compared with CK series, fungal Shannon diversity was decreased by SD but was maintained stable by BC regardless of NFLs.
The NMDS results showed that the bacterial samples of SD series (including SDN100, SDN60, and SDN0) and BC series (including BCN100, BCN60, and BCN0) tended to be located to the left and right of the graph, respectively; however, the samples were not clearly separated in the dimension of NFLs (Figure 2a), which indicated that soil bacterial communities were more susceptible to SRMs than NFLs. The separation pattern of fungal samples was consistent with that in bacterial samples (Figure 2b). F I G U R E 1 Effects of straw return modes (SRMs) and N fertilizer levels (NFLs) on (a) 16S rRNA gene copy number, (b) ITS fragment copy number, and (c) bacteria-to-fungi ratio. Their two-way ANOVA results are shown in the figures, and "ns" means "not significant". The blue, red, and gray curves and corresponding light color area fill represent the mean and deviations (n = 3) of control without straw amendment (CK), crushed straw incorporation (SD), and biochar incorporation (BC), respectively.
T A B L E 1 Richness and diversity of soil microorganisms and their two-way ANOVA results under different straw return modes (SRMs) and N fertilizer levels (NFLs). Note: Different letters in column indicate significant differences between samples at p < 0.05 (LSD). Values are means ± SD (n = 3). N100, N60, and N0, indicate that amount of N fertilizer applied at 100%, 60%, and 0%, respectively.

| Community compositions of soil bacteria and fungi
According to the RA of the assigned bacteria phyla and genera (Tables S1 and S2), the top five abundant phyla included Proteobacteria (32.5%-36.0%), Acidobacteria (22.1%-27.7%), Actinobacteria (9.3%-14.2%), Bacteroidetes (7.7%-10.2%), and Gemmatimonadetes (5.5%-7.4%), accounting for approximately 85.9%-87.2% of the microbial population identified in all treatments. Sphingomonas (5.2%-8.4%), RB41 (4.1%-5.3%), Rubrobacter (0.9%-2.2%), MND1 (1.3%-1.5%), and Nitrospira (0.9%-1.7%) were the most dominant genus with known classification in all treatments. The results of heat map and cluster analysis showed that SD increased RAs of Bacteroidetes, Nitrospirae, Verrucomicrobia, and Patescibacteria while decreased RAs of Actinobacteria, Gemmatimonadetes, and F I G U R E 2 Non-metric multidimensional scaling (NMDS) and analysis of similarities of soil bacterial (a, c) and fungal (b, d) community composition under different straw return modes and N fertilizer levels. CK, control without straw amendment; SD, crushed straw incorporation; BC, biochar incorporation; N100, N60, and N0, indicate that amount of N fertilizer applied at 100%, 60%, and 0%, respectively. Chloroflexi ( Figure 3a). Differently, the RAs of most phyla were slightly affected by BC. Confirming via twoway ANOVA, the RAs of Actinobacteria, Nitrospirae, and Patescibacteria were significantly affected by SRMs and NFLs (p < 0.05). Similarly, the interactive effects on Acidobacteria, Actinobacteria, and Gemmatimonadetes between SRMs and NFLs were also significant (p < 0.05). Therefore, without straw amendment, the RA of Actinobacteria significantly increased from 11.3% to 13.4% as the amount of NFLs decreased from N100 to N0 levels; it was significantly reduced by up to 9.3% by SD regardless of NFLs but was significantly increased to 14.2% by BC at the N60 level (p < 0.05). The RA of Nitrospirae showed a decreased trend as NFLs reducing; it was significantly increased by SD regardless of NFLs but was unaffected by BC at each NFL. At genus level, the effect of SD was different from BC on the RA of bacteria (Figure 3b). The RAs of most genera were decreased by SD, especially Rubrobacter and Gaiella, while others (i.e., Nitrospira and Pseudarthrobacter) were increased. Confirming via two-way ANOVA (Tables S1 and S2), the RAs of Rubrobacter, Nitrospira, Gaiella, and Bryobacter were significantly affected by SRMs and NFLs (p < 0.05). In this concern, the RA of Nitrospira revealed a decreased trend with NFL reduction (1.4%, 1.1%, and 1.1% in CKN100, CKN60 and CKN0, respectively); it was increased by SD, especially at the N100 and N0 levels, while remained unaffected by BC at each NFL. The RA of Gaiella was significantly increased from 0.70% to 0.93% with NFLs reduction from N100 to N0 without straw amendment; it was significantly reduced by SD regardless of NFLs while was insignificantly affected by BC at each NFL.
Ascomycota (32.9%-53.4%) and Basidiomycota (21.1%-46.9%) were the top two dominant fungal phyla across all soil samples. At the genus level, genera Geomyces and Thielaviopsis were dominant, and their RAs varied from 2.3% to 7.7% and from 1.2% to 5.5%, respectively, across all samples. The RAs of Ascomycota (p < 0.05) and Basidiomycota (p < 0.05) were significantly affected individually by SRMs. In this concern, the RA of Ascomycota significantly decreased by SD at the N0 level but was unaffected by BC. Inversely, the RA of Basidiomycota significantly increased by SD at the N0 level but was unaffected by BC. However, two SRMs did not affect the RAs of the two fungal phyla at the N100 and N60 levels. At the genus level, the RAs of Geomyces, Trichoderma, and Myrmecridium were significantly affected by SRMs and FNLs (p < 0.05). Therefore, the RA of Geomyces showed an increased trend with NFLs reduction without straw amendment (5.0%, 3.7%, and 3.9% in CKN100, CKN60, and CKN0, respectively); it was significantly decreased by SD at the N100 level while was significantly increased by BC at the N0 level. The RA of Trichoderma was significantly increased by BC at the N100 level but was significantly decreased by SD and BC at the N60 level. At the N0 level, the RA of Myrmecridium was significantly increased without straw amendment but was significantly decreased by SD and BC. Furthermore, members of Ascomycota, soil-borne plant-pathogenic genera, such as Fusarium and Gibberella, were not significantly affected by each SRM across all NFLs.

F I G U R E 3
Hierarchically clustered heat map analysis of bacterial (a, b) and fungal (c, d) community compositions at phylum (a, c) and genus (b, d) levels under different treatments. CK, control without straw amendment; SD, crushed straw incorporation; BC, biochar incorporation; N100, N60, and N0, indicate that amount of N-fertilizer applied at 100%, 60%, and 0%, respectively.

| Network patterns and function
Network analyses were performed to explore the interactions among members of bacteria and fungi in soils under two SRMs across all NFLs (Figure 4a,b, respectively). The edge number of SD (436) was 2.95 times that of BC (148), and the positive correlations among microbial taxa in the SD (64.0%) were higher than in the BC (59.5%) network, indicating that the networks were characterized by cooccurrence. The average degree (10.259), graphic density (0.122), and average clustering coefficient (0.503) of the SD network were substantially higher than that of the BC network (3.364, 0.039, and 0.239, respectively), indicating that the SD network was highly complex and stable compared to the BC network. The modularity of the SD and BC was 0.912 and 2.599, respectively, indicating that the SD and BC networks exhibited a modular structure. However, the BC network had a superior modular structure.
Further topological analysis on the connections (within and among module connectivities) revealed an interesting shift of KT (module hubs and connectors) from F I G U R E 4 Networks of microbial communities (a, b) and keystone taxa (c, d) in crushed straw (a, c) or biochar (b, d) incorporated soils. The size of each node is proportional to the number of connections (i.e., degree); the thickness of each connection between two nodes (i.e., edge) is proportional to the absolute value of Spearman's correlation coefficients. Key modules are labeled with different colors in the respective network. SD to BC networks (Figure 4c,d). Overall, one module hub, namely Mrakia, together with four other connectors, were identified in the BC network; meanwhile, one module hub, namely Pseudozyma, together with 30 other connectors, were also identified in the SD network. The SD network showed that 63.3% of KS belonged to fungi, and Chaetomium (phylum Ascomycota), Glomus (phylum Glomeromycota), and Mucor (phylum Zygomycota) showed the top three average degrees; whereas, 36.7% of KS belonged to bacteria, and Gaiella (phylum Actinobacteria) and Nitrospira (phylum Nitrospirae) showed the top two average degrees, indicating their possible key roles in maintaining the SD network. Additional details of KT were summarized in Table S3.
Microbial genera constructing key modules (SD-M0, SD-M1, SD-M2, BC-M0, BC-M1, BC-M2, and BC-M3) of SD and BC networks were classified and sorted into several dominant predicted functional groups according to FAPROTAX and FUNGuild databases (Table S4). Among these groups, N-cycling was the main functional group in all three key SD modules. In SD-M2, the RA of the functional group related to N cycling (nitrite_oxidization, ni-trogen_fixation, and ureolysis) was significantly reduced (p < 0.05). Saprotroph was the primary trophic mode in BC-M1 module and its RA was significantly reduced by BC, while the RA of symbiotroph mode in BC-M2 was significantly increased by BC (p < 0.05). In addition, as the major functional components, the RA of N cycling functional group in BC-M2 (nitrogen_fixation, ureolysis, nitrite-oxidation, aerobic_ammonia_oxidation, nitrite_ respiration, nitrogen_respiration, nitrous_oxide_denitrification) and BC-M3 (nitrate_reduction) tended to increase or decrease, respectively.

| Relationships between microbial community and soil environmental factors
The RDA results showed that bacterial community structure was separated by SRMs, and their differences were mainly driven by AP, AK, AN, and WC ( Figure 5a and Table S5). Meanwhile, AN, AK, and AP were the main drivers that significantly affected and controlled fungal community structure (Figure 5b). The correlation analysis of the top 20 bacterial/fungal genera with soil environmental factors revealed the presence of strong microbial species-environment relationships, especially in SD ( Figure S1). Four factors, namely AK, AN, TN, and pH, had considerable influences on the SD bacterial/fungal community. Different from SD, few factors had relatively strong correlations with bacterial/fungal taxa in BC.
Furthermore, the drivers for the key modules of the two networks showed specific patterns (Figure 5c). As the key module of SD network, SD-M0 mainly comprised fungi classified as Ascomycota and showed negative correlations with pH, AN, AK, TC, TN, and SOC, and demonstrated positive correlation with WC. Most microorganisms in SD-M1 and SD-M2 modules, which were dominated by bacteria and fungi classified as Proteobacteria and Ascomycota, respectively, were correlated positively with pH, AN, AK, and TN, while other genera in the two key modules were negatively correlated with these drivers. As key modules of the BC network, BC-M2 was mainly dominated by bacteria belonging to Proteobacteria and demonstrated positive relationship with pH and TN, while BC-M3 was mainly dominated by bacteria belonging to Actinobacteria and demonstrated negative relationship with AN, AP, TN, and SOC. Interestingly, members of BC-M0 all belonged to fungi classified as Ascomycota and were correlated negatively with WC, AK, TC, TN, and SOC. BC-M1 was also mainly dominated by fungi classified as Ascomycota and showed a positive relationship with BD and AP but was negatively correlated with TC and AK. These results indicated that the sensitivity of modular microbial groups varied in response to SRMs across all NFLs.

| Effects on microbial abundance, richness, diversity, and community
In agricultural ecosystems, soil microorganisms are sensitive to fertilization regimes, which can directly increase the soil available nutrients used for the growth of microorganism Yan, Wang, et al., 2021;Yan, Xue, et al., 2021). Particularly, soil microorganisms are generally N limited; therefore, their community structure can be influenced by N availability (Cederlund et al., 2014;Fan et al., 2022). In this study, the abundances of bacterial 16S rRNA gene and fungal ITS fragment decreased as NFLs reducing regardless of SRMs (Figure 1). In addition, bacterial Chao1 richness, Shannon diversity, and fungal Chao1 richness decreased as NFLs reducing (Table 1). These results are consistent with previous observations which found that bacterial abundance showed the same trend as the αdiversity, both demonstrating increments by fertilization treatments Yan, Wang, et al., 2021;Yan, Xue, et al., 2021). The decrease in microorganism diversity as NFLs reducing revealed that the vital microbial processes and soil ecosystem functions are simple and sensitive . However, SD and BC can lead to the increase in bacterial richness and diversity, especially at reduced NFLs (Table 1). These results indicated that SD and BC can compensate for the loss of bacterial diversity caused by NFLs reduction, while the compensation effect of BC on the loss of bacterial diversity was better than that of SD. These results are consistent with observation of Yang, Muhammad, et al. (2022), who found that straw return improves soil bacterial diversity and richness. The development of bacterial diversity under SD could be explained by the release F I G U R E 5 Associations between soil environmental variables and microorganisms at bacterial genus (a), fungal genus (b), and key modular (c) levels. CK, control without straw amendment; SD, crushed straw incorporation; BC, biochar incorporation; N100, N60, and N0, indicate that amount of N fertilizer applied at 100%, 60%, and 0%, respectively. Soil environmental factors include total nitrogen (TN), total carbon ( of diverse nutrients and soluble organic matter during the decomposition of crop straw, providing space, and influencing other soil environmental factors, which are beneficial to the proliferation and growth of microorganisms and produce stable and healthy soil ecosystem (Du, Zhang, et al., 2022;Tang et al., 2021). Different from fungal richness, the fungal Shannon index was significantly decreased by SD at reduced NFLs while remained stable in BC at each NFL (Table 1). Reports indicated that the fungal diversity was negatively impacted by maize straw return  while was insignificantly influenced by long-term biochar addition (Yao et al., 2017), both consistent with the obtained results. The loss of fungal diversity caused by SD at reduced NFLs indicated that SD may lead to a slightly stable agroecosystem with low N input (Bani et al., 2018;Yang, Jing, et al., 2019;Yang, Ma, et al., 2019;Zhou et al., 2016). Saprotroph fungi can easily utilize the labile C, which is contained in straw with a high proportion; therefore, their competitive capacity with pathotrophs and symbiotrophs was then enhanced, and the fungal diversity was finally decreased (Dai et al., 2021;Tian et al., 2019). In addition, SD had the lowest B/F ratio at the N100 level, while BC had the lowest B/F ratio at the N0 level (Figure 1). A low B/F ratio indicates a highly sustainable agricultural system Zhang et al., 2018). In this concern, the obtained results showed that the adaptability of bacteria and fungi to the reduction in NFLs differed between two SRMs. This difference indicates that BC is better than SD in reducing NF usage, while SD is highly suitable to improve the sustainability of maize cropping systems without NF reduction. In addition to the analysis of richness and diversity, the assessment of microbial biodiversity as soil quality indicators should also consider the population structure (Frąc et al., 2018). In this study, Proteobacteria was the most dominant bacterial phyla across all soils (Table S1), which were considered to be the most common bacterial phyla in soil amended with straw or straw biochar in previous studies (Cong et al., 2020;Liu et al., 2021). Proteobacteria is the most important bacterial group in soil; it actively participates in the straw decomposition and then finally optimizes the rhizosphere microenvironment Naseem et al., 2018;. Previous study found that the RA of Proteobacteria was the highest in SD treatment . The RA of Proteobacteria in the current study showed an increase trend in BC and SD; SRMs also significantly altered the RAs of genus belonging Proteobacteria, such as Sphingomonas, Steroidobacter, and Acidibacter (Figure 3 and Table S2). Sphingomonas can degrade soil toxic substances, resist multiple pathogens, promote nutrient absorption, and maintain soil nitrogen balance (Chaparro et al., 2012;Hernández-Lara et al., 2023). The RA of Sphingomonas in this study was affected by SRMs and associated with NFLs, which was significantly increased by BC at the N60 level but was significantly decreased by SD at the N0 level (Table S2). A previous study also suggested that the effect of straw return on the RA of Proteobacteria was related to the nitrogen level (Du, Zhang, et al., 2022). The sufficient air and organic nutrient conditions provided by BC might be responsible for the increase in Sphingomonas (Li, Song, et al., 2022;Li, Zheng, et al., 2022).
Considering fungi, Ascomycota and Basidiomycota dominated all soil samples, which was similar to the result of Yao et al. (2017), who reported that these phyla are the most common fungal phyla in the black soil of northeast China (Table S1). In this study, SD reduced the RA of Ascomycota and inversely increased the RA of Basidiomycota, but BC did not change the RAs of both ( Figure 3 and Table S1). The plausible explanation for the different responses between two fungal phyla lies in their differences in degradation efficiency of organic matter, despite being proven to be the key decomposers in soils (Cui et al., 2022;Fan & Wu, 2021;Frąc et al., 2018). Different from Basidiomycota, which are prone to degrade cellulose, lignin, and other refractory substances, Ascomycota mainly utilize degradable organic compounds (Cong et al., 2020;Liu, He, et al., 2022;Lundell et al., 2010). This finding was also supported by the result from Song et al. (2020), who found that the RAs of Basidiomycota (increased) and Ascomycota (decreased) were significantly affected by biochar addition, respectively. In addition, SRMs have a considerable influence on the RAs of several fungi at genus level, which was more evident than the effect of NFLs. The RAs of Acremonium and Trichoderma were increased by SD (at the N0 level) and BC (at the N100 level), respectively ( Figure 3 and Table S2). As a well-known plant growth-promoting fungus, Trichoderma can activate the plant immune system against pathogens (Fernández-González et al., 2020;Li, Song, et al., 2022;Li, Zheng, et al., 2022). Acremonium was also considered to be a beneficial fungus and was found to be dominant in soils treated with organic residues (Hernández-Lara et al., 2023). Concerning the responses of microbial community to SRMs and NFLs, the current study may hint that BC could have positive effects on bacterial structure especially at low N levels, while the reshape of fungal structures by SD could be beneficial for straw degradation.

| Effects on microbial network and modular function
The interactions within microbial communities are associated with ecological functions, and the network analysis is a powerful tool to discover the existing complex interactions among microorganisms (Banerjee et al., 2016;Du, Trivedi, et al., 2022). As identified by the network analysis, SD network is more complex and robustly stable than BC, as reflected by the high number of edge, average degree, graphic density, and average clustering coefficient (Figure 4). This result was partly consistent with the observation of Du, Zhang, et al. (2022), who found that straw had positive effect on the bacterial network pattern. The interactions of soil microbial communities became strong in response to straw return, which demonstrates that the degradation of complex organic matter in SD soil may require a variety of microorganisms to work together (Yang, Bao, et al., 2022). In addition, a highly developed and connected microbial network may imply the enhancements of the efficient carbon utilization, nutrient cycling function, and functional redundancy (Mougi & Kondoh, 2012;Morriën et al., 2017;Wagg et al., 2019). Therefore, the exchange of nutrients among soil microbes and the resistance to disturbance could have possibly been promoted or enhanced by SD. By contrast, the reduction in the abundance of microorganisms or the strong competition for the same resource in the BC network may lead to a simple and weak microbial network (Fierer & Lennon, 2011;Wang et al., 2023;Yao et al., 2017).
Modularity is important for microbial community function and stability considering the resource allocation, niche sharing, or habitat heterogeneity (Yan, Xue, et al., 2021;Wagg et al., 2019). As identified by the network analysis (Figure 4), BC network demonstrated a superior modular structure, implying that niche sharing among microbes in BC soil was maximal. KT, including module hubs and connectors, can markedly affect microbiome function and composition, such as cycling of different substances and exchanges of nutrients and metabolites (Banerjee et al., 2016;Du, Zhang, et al., 2022;Yang, Jing, et al., 2019;Yang, Ma, et al., 2019). In the current study (Figure 4 and Table S3), Glomus, Chaetomium, Mucor, and Gaiella in SD-M0, Pseudozyma and Nitrospira in SD-M1, and Mrakia in SD-M2 were identified as the main KT in the SD network, while Fusarium in BC-M0, Mrakia in BC-M1, Massilia in BC-M2, and Solirubrobacter and Ramicandelaber in BC-M3 were identified as KT in the BC network, indicating their potential as indicators and the shifting of their ecological roles by different SRMs. Gaiella likely serves a role in nitrogen cycling in cultivated farm ecosystems (Li, Song, et al., 2022;Li, Zheng, et al., 2022). Nitrosospira are main ammonia-and nitrite-oxidizing bacteria and function as soil N cycle Yang, Jing, et al., 2019;Yang, Ma, et al., 2019). In current study (Tables S1 and S2), the RA of Nitrospirae showed a decreasing trend with the reduction in NF amount; however, it was significantly increased by SD regardless of NFLs. The RA of Nitrospira showed a similar respond trend to Nitrospirae. The enrichments of Nitrospirae and Nitrospira may be due to their adaptation to soil environmental changes during straw incorporation or the stimulated proliferation due to the increased soil nutrient availability associated with straw decomposition . In the current study (Table S4), the RA of N-cycling functional group in SD-M2 was significantly reduced, which could possibly be attributed to the absence of Nitrospira in this module. Previous studies have shown that biochar addition plays a key role in soil N-cycling processes (such as N fixation, nitrification and denitrification) mediated by microbes Yin et al., 2020). However, in the current study (Table S4), the N-cycling functional group at key module levels was not significantly changed by BC. These opposing results may be partly attributed to the differences in the biochar properties, soil and crop types.
The identified KT and key modules in this study (Tables S3 and S4) were predominated by Ascomycota (fungal phylum), suggesting that fungi played a key role in the SD network. Soil fungi vary in their mechanisms for nutrient acquisition and substrate preference; therefore, they could form complex interactions with each other as well as other soil organisms and finally influence the overall soil microbial community and ecosystem stability (Frąc et al., 2018;Yang, Jing, et al., 2019;Yang, Ma, et al., 2019). The capability of fungi was generally larger than bacteria to degrade recalcitrant organic C and sequestrate C (Fan & Wu, 2021). Mrakia could decompose surrounding cell remnants and function as a carbon source supplier . Mucor could transform decaying material into simple sugars by producing lipolytic and proteolytic enzymes (Duan et al., 2019). Considering the functionality of the fungi, saprophytic fungi play key roles in degradation of organic matter and C-cycling in soils (Hernández-Lara et al., 2023). For example, as a saprophyte fungus, Chaetomium participates in decomposing organic matter . A previous study showed that Chaetomium was enriched by long-term straw mulching (Qiu et al., 2020). Dai et al. (2018) found that the proportion of the saprotrophic fungi was increased by biochar addition. In this study (Table S4), saprotroph mode was dominant trophic mode in all three SD key modules but was significantly reduced in BC-M18, hinting that the capacity of saprothophic fungi to utilize the labile or recalcitrant C differed between two SRMs (Dai et al., 2021). Different from saprophytic fungi, symbiotrophic fungi play key roles in the quality, nutrition, and health of crops . For example, as arbuscular mycorrhizal fungi, which account for the majority of symbiotrophs, Glomus could suppress fungal pathogens and improve plant growth Frąc et al., 2018). Pseudozyma also could induce a plant defense response (Kitamoto, 2019). In this study (Table S4), the RA of symbiotroph category in BC-M219 was significantly increased, which suggested that BC could be beneficial for symbiotrophic fungi growth.

| Drivers of microbial communities and sub-communities
The current study (Figures 1-4) found the effects of SD and BC on soil organisms markedly differed mainly due to the differences in soil environmental factors. RDA and correlation analysis of main bacterial/fungal taxa with soil environmental factors ( Figure 5 and Table S5) revealed that the AN and TN were the key drivers for the fungal community composition under straw or biochar straw returning. As a main phylum of fungi, Ascomycota' growth rate depended on the N availability, and its RA was significantly negatively affected by TN in soil (Fontaine et al., 2011;Qiu et al., 2020). The obtained findings confirmed that the RAs of dominant fungi at genus levels were negatively correlated with AN and TN in soil incorporated with maize straw, which was consistent with the previous findings Su et al., 2020). The drivers for the key modules of two networks showed specific patterns in this study (Figure 4). For example, WC was the driver for the SD-M0, SD-M1, and BC-M0, which suggested that microbial module community and function was related to soil moisture. Straw return can enhance soil water content and water retention, and then influence the substrates and energy substances available to soil microorganisms, finally improving their growth and activities (Hagemann et al., 2017;Liu, Li, et al., 2022;Siedt et al., 2021). In addition, BD was the only shared driver for the BC-M1 and BC-M3 but not for SD modules. Reports have been indicated that the decreased BD caused by biochar addition can increase micropore volume and enhance oxygen and available water content as well as soil hydraulic conductivity, subsequently improving microbial activities (Adhikari et al., 2022;Zhang et al., 2018). Soil pH was the common factor driving the microbial composition of all three key modules of SD (SD-M0, SD-M1, and SD-M2), suggesting that microbial structure in the SD network is more sensitive to soil pH variation than is BC. It was previously reported that the fungal diversity and abundance were dependent on soil pH; the fungal communities were sensitive to variation in soil pH (Feng et al., 2016;Zhou et al., 2016). We also found that the soil pH is a powerful factor influencing the fungal community structure ( Figure S1).

| CONCLUSION
This study explores the effects of SRMs and NFLs on the diversity and structure of bacterial and fungal communities, co-occurrence networks, and key modules patterns, as well as their driving soil environmental factors. The dominant bacterial phyla were Proteobacteria, Acidobacteria, and Actinobacteria while the dominant fungal phyla were Ascomycota and Basidiomycota in the black soil of northeast China. Microbial communities were significantly affected by SRMs and depended on NFLs, which were driven by soil AN, AK, TN, and pH. SD and BC can compensate for the loss of bacterial richness and diversity and fungal richness caused by NFL reduction while the compensation effect of BC was better than that of SD, especially at reduced NFLs. However, fungal Shannon diversity responded differently to SD and BC, which remained stable only in BC. In addition, network pattern, keystone taxa, modular function, and driving factors shifted between SD and BC microbial networks. Concerning the responses of microbial abundance, community and network to SRMs and NFLs, the current study indicates that BC is better than SD in reducing NF usage, while SD is highly suitable to improve the sustainability of maize cropping systems without NF reduction. These results deepen insights into the divergence of bacterial and fungal taxa response to SRMs and NFLs, providing a scientific basis for the selection of suitable strategy for sustainable straw utilization. In the future, crop root exudates and microbial function based on metagenomic sequencing should be considered to better understand the influence of straw returning on soil ecosystem and assist in the efficient use of straw resources.