Nitrogen content and C/N ratio in straw are the key to affect biological nitrogen fixation in a paddy field

Straw amendment can increase nitrogen fixation in paddy field, however, the efficiency of carbon sources with different biochemical properties to enhance N2 fixation and nitrogen fixation activity is still unclear. A 15N2-labelling system was used in the field environment to determine biological nitrogen fixation (BNF) under the addition of three kinds of straw. The nifH gene (DNA) and nifH RNA gene (cDNA) of soil were amplified by real-time fluorescent quantitative PCR. The diversity and community composition of nitrogen fixing microorganisms were studied by high-throughput sequencing. The study is expected to reveal how different carbon sources impact biological nitrogen fixation and its mechanism in the paddy field system. Results showed that the absolute abundance diazotrophs in the treatment of mature stage wheat straw (MWS) was 4.88 times as high as that in the CK treatment, but jointing stage wheat straw (JWS) and poplar branch (PB) did not induce any significant changes. Straw amendment had no impact on cyanobacteria abundance. The proportion of N2 fixation increased by MWS was 2.07 times, but which was much lower than the increase proportion of the heterotrophic diazotrophs, leading to a decrease of diazotrophic nitrogen fixation activity. Mature wheat straw addition increased biologically fixed nitrogen in paddy field by increasing the number of heterotrophic nitrogen fixing bacteria. The results indicated that to increase biological N2 fixation in paddy system, straws with low nitrogen content and high C/N ratio were recommended.


Introduction
Nowadays, in order to improve rice yield, a large quantity of chemical nitrogen fertilizer has been applied to paddy fields, which has brought environmental problems such as soil acidification, water eutrophication and greenhouse gas emissions (Deng et al. 2014;Mahmud et al. 2020). The researchers have studied various ways to reduce chemical nitrogen fertilizer while ensuring rice yield, such as applying different types of fertilizer (Lu 2020) and different fertilization methods (Zhang et al. 2021a, b). Biological nitrogen fixation (BNF) is considered to have the potential to partially replace chemical nitrogen fertilizer for the following two reasons (Rosenblueth et al. 2018). Firstly, BNF is one of the main sources of nitrogen in rice field before the invention of industrial synthetic nitrogen fertilizer, which has permitted moderate but constant productivity in rice fields for thousands of years (Ladha and Redddy 2003;Roger and Watanabe 1986). Secondly, almost all types of BNF have been found in rice fields (Roger and Watanabe 1986;Yoshida and Ancajas 1971). However, there are still many problems to improve BNF in paddy fields, such as BNF is an energetically and carbon expensive process and heterotrophic diazotrophs can only fix nitrogen when adequate supplies of carbon are available (Darian et al. 2019). Therefore, providing sufficient carbon source for nitrogen fixing microorganisms in paddy field may help to increase the amount of BNF.
Plant straw has high carbon content, and its application can provide available carbon source for microorganisms (including nitrogen fixing bacteria) in paddy soil. Crop residues and other organic amendments are usually added to the soil to improve soil SOC (Arunrat et al. 2020;Lal 2011). Some studies find that straw application may affectthe amount of BNF in paddy field (Hao et al. 2022;Tang et al. 2017). Generally, rice straw returning can increase the abundance of nitrogen fixing bacteria (Tang et al. 2021) and its nitrogen fixing activity (Rao 1976;Motohiko and Michio 2003). The nitrogen fixation activity in paddy soil is significantly increased by the addition of rice straw (0.6% w/w) by acetylene reduction method (Tanaka et al. 2006). By a microcosm 15 N 2 labeling method, 3 t ha −1 rice straw application in paddy soil increases nitrogen fixation by 14.86%, and a 6t ha −1 application increased nitrogen fixation by 32.43% (Charyulu et al. 1981). These studies have promoted our understanding of the impact of straw returning on BNF. However, in the previous studies, paddy soil was cultivated indoors (without rice planting), which was far from the field environment. Moreover, the cellulose, hemicellulose and lignin content in different straw are different, resulting in different carbon availability (Wang et al. 2020a, b). The application of straws with different biochemical compositions may have different effects on BNF, which was not investigated in the previous research.
BNF is a diazotrophs-driven process. Heterotrophic and phototrophic nitrogen fixing microorganisms under no straw and nitrogen additions contribute 49.7% and 50.3% to BNF in paddy fields, respectively (Bei et al. 2013). Heterotrophic nitrogen fixing bacteria rely more on external carbon sources than autotrophic nitrogen fixing bacteria, since phototrophic nitrogen fixing bacteria can use CO 2 as carbon source through photosynthesis (Stewart 1969). Therefore, straw returning may mainly affect heterotrophic nitrogen fixing bacteria rather than phototrophic nitrogen fixing bacteria such as cyanobacteria. It is generally believed that bacteria colonizing rice root (endophytes) can have higher efficacy to do N 2 fixation and higher commercial values than free-living bacteria (Pittol et al. 2016;Banik et al. 2019). Rice endophytic nitrogen fixing bacteria and rhizosphere nitrogen fixing bacteria are mainly heterotrophic nitrogen fixing bacteria, which rely heavily on rice for their carbon source (Rosenblueth et al. 2018). Indeed, the heterotrophic nitrogen fixing bacteria have been often inoculated into rice or paddy soil to increase the nitrogen source of rice (Yoneyama et al. 2017;Banik et al. 2019). However, the effects of straw with different biochemical composition on heterotrophic nitrogen fixing microorganisms and its mechanism are not clear. For example, whether straw returning mainly acts on the numbers of heterotrophic nitrogen fixing bacteria or its nitrogen fixing activity (i.e., the amount of nitrogen fixation and nifH gene expression), and which components of straw play a major role.
In this study, an airtight transparent growth chamber 15 N-labelling system was used (Bei et al. 2013) in the field environment to determine BNF under the addition of three different carbon substrates with differences in biochemical composition. The nifH gene of soil DNA was amplified by real-time fluorescent quantitative PCR (qPCR) to study the number of nitrogen fixing bacteria. The nifH RNA gene reverse transcription-polymerase chain reaction (RT-PCR) was used to study the nifH gene expression. The community composition of nitrogen fixing microorganisms was studied by high-throughput sequencing of nifH DNA gene. The study is expected to reveal how different carbon sources impact biological nitrogen fixation and diazotrophs in the paddy field system.

Site description, soil collection and soil properties
The experiment site is located in Xiaoji town, Jiangdu City, Jiangsu province, China (119°42'E, 32°35'N). The soil was classified as Inceptisol in the US soil taxonomy (or Shajiang Aquic Anthrosol based on Chinese soil taxonomy). The soil samples were randomly collected at several points along a S shape from the plough layer (0-15 cm) at the end of April. Visible plant materials in the soil were removed, and then the soil was air-dried, thoroughly homogenized and sieved (< 5 mm). The soil had a pH value of 6.1 (soil:water = 1:5, w:v), and it contained 7.4% sand, 78.0% silt and 14.6% clay. More physio-chemical properties of the soil samples can be found in Wang et al. (2019).

Carbon sources
In order to know clearly how biochemical compositions influence BNF and its mechanisms, 3 kinds of straw with large difference in biochemical compositions were chosen. They were jointing stage wheat straw, mature stage wheat straw, and poplar branch. The jointing wheat (Triticum aestivum L., cv. Ning Mai 16) straw was collected on 7 th April in 2018 from the field when the plant height was around 62 cm. After wheat seedlings were collected, they were killed immediately at 105 ℃ for 1 h and then dried at 75 ℃ for preservation. The mature stage wheat (Triticum aestivum L., cv. Ning Mai 16) straw was collected on 5 th June in 2017 after harvest. Branches (Diameter 0.3-0.8 cm) of poplar (Populus tomentosa) were also collected on 7 th April in 2018. The contents of cellulose, hemicellulose and lignin in the materials were analyzed by Van Soest method (Van Soest et al. 1991), total nitrogen and total carbon were determined by N/CN Elemental Analyzer (Vario Max CN, Germany). The results are shown in Table 1, and the correlation of carbon content, nitrogen content, C/N ratio with biochemical composition are shown in Table 2. These plant materials were ground by a crusher after 105 ℃ sterilization and drying, passed through a 2 mm sieve and stored for subsequent usage.
Field 15 N 2 labeling experiment A field 15 N 2 labeling experiment was conducted in the paddy field located in Xiaoji town, Jiangdu city, Jiangsu province, China from 23 rd July to 23 rd October in 2018. The airtight chambers (len gth × width × height = 1160 × 680 × 890 mm) (ITI-GCN Crop., Ltd, Nanjing, China) in 3 replicates Table 1 Main compositions of carbon materials mean ± standard error (n = 3). Data followed by the same lowercase letters indicate non-significant (Turkey's Test, P > 0.05) differences in the same column. Abbreviations: jointing stage wheat straw (JWS), poplar branch (PB) and mature stage wheat straw (MWS). "★" means: the soluble matter is the part that dissolved by detergent during the determination of cellulose, hemicellulose and lignin by Van Soest method were treated with 15 N 2 and installed in parallel in the rice field. The temperature and CO 2 concentration in the chambers were automatically controlled to keep in line with the ambient air temperature (ambient temperature ± 1 ℃) and CO 2 concentration (400 ± 20 ppm). Excessive O 2 generated by rice photosynthesis in the chambers was removed using iron powder to keep the O 2 concentration at around 21%. In addition, three replicates of all treatments were placed outside the chambers as unlabeled controls. More detailed information about the design and control system of this airtight 15 N 2 -labeled chamber is described by Bei et al. (2013). In this experiment, there were four treatments as follows: (1) 1.4 kg soil with no material amendment (CK), (2) 1.4 kg soil with 14 g jointing stage wheat straw amendment (JWS), (3) 1.4 kg soil with 14 g poplar branch amendment (PB), (4) 1.4 kg soil with 14 g mature stage wheat straw amendment (MWS). The sifted straw (< 2 mm) and the sifted air-dried (< 5 mm) soil were mixed evenly according to the weight, and then filled into a pot (length × width × height = 9 × 9 × 2 0 mm) with a height of about 15 cm. All treatments have three repetitions with each replication in each chamber, respectively. Before seedling transplantation, each pot was filled with water for 7 days, and the water surface was 1-2 cm above the soil surface. Two rice seedlings (Oryza sativa L., cultivar Wuyunjing 23) were transplanted to each pot. Each pot was added with 116.37 mg KH 2 PO 4 and 32.55 mg KCl (equivalent to 75 kg P 2 O 5 ha −1 and 75 kg K 2 O ha −1 ) before rice transplantation, while no N fertilizer was applied. In each pot, the water surface was kept 1-2 cm above the soil surface until 10 days before rice harvest.
After pots with different treatments were moved inside the chamber and air-tightness of the system was carefully checked, 40 L of the air in each chamber was replaced by 40 L of 15 N 2 (99 atom% 15 N). 15 N 2 was produced and purified following the method of Ohyama and Kumazawa (1981), and detailed information on 15 N 2 preparation was described by Ma et al. (2019b). The 15 N enrichment of N 2 within the three chambers was monitored by taking gas samples once a week during the 90-day labeling period.

Sampling and 15 N determination
After 90 days of 15 N 2 continuous labelling, the growth chambers were opened for soil and plant sampling. In each pot, rice plants were separated into aboveground and belowground parts (roots). The soil samples were divided into 0-1 cm, 1-5 cm, and 5-15 cm. Subsamples of soils from different treatments were stored in liquid N 2 just after sampling, transported to Nanjing lab and stored at -80 ℃ for further molecular analysis. The remained soil samples were oven dried and ground using a Retsch MM 400 mixer mill (Retsch, Haan, Germany). A Thermo Finnigan Delta plus Advantages Mass Spectrometer coupled with an elemental analyzer (Thermo Fisher Scientific Inc., Waltham, MA, USA) was used for analysis of total N content and 15 N-enrichment in soil and plants. A MAT 253 stable isotope ratio mass spectrometer (Thermo Fisher Scientific Inc., Bremen, Germany) was used for analysis of the 15 N-enrichment of N 2 in the 15 N 2 -labeled chambers. A Vario Max CN Analyzer (Elementar Analysensysteme GmbH, Hanau, Germany) was used to measure soil total C and N concentrations. Table 2 Correlation of carbon content, nitrogen content, carbon nitrogen ratio with biochemical composition "-" in front of the number indicates negative correlation, and no symbol in front of the number indicates positive correlation. "*" means P < 0.05, "**" means P < 0.01  (Poly et al. 2001) were used for quantifying nifH gene copies, the size of the PCR products was 362 bp. The PCR thermal cycling conditions were set as follows: an initial denaturation step of 10 min at 95 ℃, followed by 35 cycles of amplification (30 s at 95 ℃, and 35 s at 58 ℃, 45 s at 72 ℃), and a final extension step of 10 min at 72 ℃. The copy number of the target gene was quantified by qPCR analysis with a CFX96 Optimal Real-time system (Bio-Rad Laboratories, Inc. Hercules, CA). For sequencing library construction, the nifH gene was amplified using the same primer set polF and polR. A sample tagging approach was used to recognize each sample in one run of MiSeq Sequencing. A tag (12 bases), synthesized by Generay (Generay Biotech, China), was added to the 5'-end of the forward primer. The PCR procedure were as follows: an initial denaturation step of 3 min at 95 ℃, followed by 35 cycles of amplification (30 s at 95 ℃, and 30 s at 57 ℃, 30 s at 72 ℃), and a final extension step of 5 min at 72 ℃. The PCR products were verified on 1.5% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Bio-sciences, USA). Equimolar amounts of different PCR amplicons were added to a tube and evenly mixed. The mixed products were sequenced on the Illumina MiSeq platform (Illumina, USA). More detailed information about the Real-time qPCR and sequencing library construction can be found in Ma et al. (2019b). RNeasy® PowerSoil® Total RNA Kit (Qiagen) kit was used to extract total RNA from soil. Total RNA was further purified by using RNeasy ® MinElute ® cleanup kit (Qiagen). Prime Script™ RT reagent kit with gDNA eraser (Takara) kit was used to further remove DNA and then reversely transcribe the extracted total RNA to cDNA. The nifH gene in cDNA was quantified by qPCR analysis with a CFX96 Optimal Real-time system (Bio-Rad Laboratories, Inc. Hercules, CA), and the quantification method is the same as the quantitative amplification method of nifH gene in DNA.

Sequencing data analysis
The detailed sequence data processing method can be found in Wang et al. (2019). Briefly, the software PEAR was used to merge the raw paired-end nifH gene reads (Zhang et al. 2013). The sequences with low quality (average score < 20) were removed and then checked for chimera using UCHIME algorithm (Caporaso et al. 2010;Edgar et al. 2011). UPARSE were used to cluste Operational taxonomic units (OTUs) with 97% similarity cutoff (Edgar 2013). The representative sequences were compared with the Ribosomal Database Project (RDP) nifH gene database using FrameBot in Fungene (min length = 80 aa, identity cutoff = 80%) Wang et al. 2013Wang et al. , 2019. Then, an R microeco package (v0.6.5) was used for microbial community and diversity analysis and mapping (Liu et al. 2020).

Calculation and statistical analysis
The amount of 15 N fixed by BNF in the rice soil system during the experiment was calculated as (Zhang et al. 2021a Where %N fixed is the percentage of N derived from BNF in rice plant and soil. N i sample is the amount of N in rice plant and soil in the 15 N 2 -labeled chamber. Statistical differences of the data were analysed using the one-way ANOVA and Tukey's honestly significant difference (HSD) test in SPSS. Values with P < 0.05 were recognized as significantly different.

N-labelling growth chamber performance
The labeling period was from the seedling stage to the mature stage of rice growth for 90 days from July 23 to October 23, 2018. The abundance of 15 N 2 in the 15 N-labelling growth chambers decreased gradually from 9.37% to 0.80% over the whole rice season, with an average value of 3.73% (Fig. S1-1). The temperatures in the chambers were similar with the field temperature ( Fig. S1-2). The CO 2 concentration in the chambers was at 400 ± 20 μL L −1 during daytime, but it accumulated to a maximum of approximate 2000 μL L −1 at night due to plant respiration and a lack of photosynthesis (Fig. S1-3). The relative humidity in the chambers was between 50 and 80% ( Fig. S-4). The oxygen produced by photosynthesis in the chambers was eliminated by deoxidizer (iron powder), and the concentration was maintained at about 21% ( Fig. S1-5).

Fixed nitrogen in soil and rice
The N fixed in 0-1 cm soil layer and total soil in the MWS treatment was significantly higher than those in the other treatments (Fig. 1A). In 1-5 cm soil layer, there was no significant difference for the N fixed among all treatments. In 5-15 cm soil layer, the N fixed in the CK treatment was significantly higher than in the JWS and PB treatments, but there was no significant difference between CK and MWS treatment. In the CK treatment, the N fixed in the 0-1 cm soil layer and 5-15 cm soil layer were significantly higher than 1-5 cm soil layer, but in the other three treatments, the N fixed in the 0-1 cm soil layer was significantly higher than in the other soil layers (1-5 cm, Absolute abundance = Ralative abundance × nifH RNA gene copy number 5-15 cm). The N fixed in the root and whole rice plant in the MWS treatment was significantly higher than those of the other treatments (Fig. 1B). But there was no significant difference for aboveground plant among all treatments. In each treatment, except for the treatment of mature stage wheat straw addition, there was no significant difference in N fixed of root and aboveground in the other treatments (i.e., CK, JWS and MWS treatments). The total nitrogen fixation of MWS treatment was 15.8 kg ha −1 , which was significantly higher than that of the other three treatments (nitrogen fixation was 7.62, 6.22 and 7.44 kg ha −1 in CK, JWS and PB treatments, respectively) (Fig. 1C), while there was no significant difference in nitrogen fixation among CK, JWS and PB treatments (Fig. 1C).

Abundance of nifH DNA gene, nifH RNA reverse transcription gene and nitrogen fixation activity
The copy numbers of the nifH DNA gene were 0.35 × 10 8 , 0.24 × 10 8 , 0.54 × 10 8 and 1.70 × 10 8 g −1 dry soil in CK, JWS, PB and MWS treatments, respectively. The copy numbers of the nifH DNA gene of MWS treatment were significantly higher than those of the other treatments, and there was no significant difference among the other three treatments (Fig. 1E). The copy numbers of the nifH RNA reverse transcription gene were 0.21 × 10 5 , 0.21 × 10 5 , 0.50 × 10 5 and 6.00 × 10 5 g −1 dry soil in CK, JWS, PB and MWS treatments, respectively. The copy numbers of the nifH RNA reverse transcription gene of MWS treatment were significantly higher than those of the other treatments, and there was no significant difference among the other three treatments (Fig. 1D). The nitrogen fixation activity of MWS was the lowest and significantly lower than that of the CK treatment. The nitrogen fixation activity of JWS and PB treatments was also lower than that of CK treatment, but it was not statistically significant (Fig. 1F).
Correlation of nitrogen fixation amount with abundance of nifH DNA gene and nifH RNA reverse transcription gene The nitrogen fixation amount is positively correlated with the abundance of nifH DNA gene, but the statistics are not significant, while it is significantly positively correlated with the abundance of nifH RNA reverse transcription gene (Table 3). The change of nitrogen fixation amount (i.e., the nitrogen fixation in JWS, PB and MWS minus that in CK) is significantly positively correlated with the change (i.e., the nifH gene numbers and nifH reverse transcription gene number in JWS, PB and MWS minus that in CK) of the abundance of nifH DNA gene and nifH RNA reverse transcription gene (Table 3). The sequence number of each sample ranged from 2996 to 34306. To make the sequence number equal for each sample, each sample retained 2996 sequences for downstream analysis. By calculating the alpha diversity of samples among different treatments, the observed species and fish index of the CK treatment were significantly lower than those of JWS, PB and MWS treatment, but there was no significant difference among the treatments which were added organic carbon source materials. The shannon index of MWS treatment was significantly higher than CK treatment, while there was no significant difference in the other three treatments (Table S3). NMDS plots showed clear separations between CK treatment and the three straw addition treatments (Fig. S2).
Proteobacteria and cyanobacteria were the main phylum of nitrogen fixing bacteria, which accounted for 82.0-90.6% of the total nitrogen fixing bacteria in each treatment (Table 4). At the class level, Table 4 shows the 10 classes with the highest relative abundance. In each treatment, Alphaproteobacteria was the most abundant class. The relative abundance of Alphaproteobacteria in the CK treatment was higher than that of PB and MWS treatments, and significantly higher than that of JWS treatment. The absolute abundance of Gammaproteobacteria in the CK treatment was the lowest and significantly lower than that of JWS treatment. The absolute abundance of each class in the MWS treatment was significantly higher than that of other treatments, except for the absolute abundance of oscillatoriophycidae, nostocales, actinobacteria and stigonematales (Table 5). Table 3 Correlation of nitrogen fixation amount with abundance of nifH DNA gene and nifH RNA reverse transcription gene The number represents the positive correlation coefficient. Spearman correlations between biochemical composition of different carbon materials and the changes of nifH gene numbers, nitrogen fixation amount and nifH transcript gene number ("the change" means: the nifH gene numbers, nitrogen fixation amount and nifH reverse transcription gene number in JWS, PB and MWS minus that in CK). "*" means P < 0.05, "**" means P < 0.01 Abundance of nifH DNA gene Abundance of nifH RNA reverse transcription gene "The change" of abundance of nifH DNA gene "The change" of abundance of nifH RNA reverse transcription gene Nitrogen fixation amount 0.57 0.71* \ \ "The change" of nitrogen fixation amount \ \ 0.92** 0.85** Table 4 Relative abundance of the nifH gene for the taxonomic profiles at the class level of the nitrogen fixing microorganism in the soil (Top 10 of average relative abundance) mean ± standard error (n = 3). The same small letters indicate non-significant differences among different treatments (P > 0.05) in the same row. Abbreviations: Control (CK), jointing stage wheat straw (JWS), poplar branch (PB), and mature stage wheat straw (MWS) Relationships between the biochemical composition of different carbon materials and the changes of nifH DNA gene numbers, nitrogen fixation amount and nifH RNA reverse transcription gene numbers Spearman's rank correlation was used to evaluate the relationships between the biochemical composition of different carbon materials and the changes of nitrogen fixation activity, nifH gene numbers, nitrogen fixation amount and nifH transcript gene number (Fig. 2). The nitrogen and soluble matter content showed a significant negative correlation (P < 0.01) with the changes of nifH gene numbers, nitrogen fixation amount and nifH transcript gene number (subtracting the CK from the JWS, PB and MWS). However, the relationship between the C/N ratio and the changes of nifH gene numbers, nitrogen fixation amount and nifH RNA reverse transcription gene number were completely opposite to that between nitrogen and these changes. In addition, the cellulose content exhibited a significant negative correlation (P < 0.05) with nitrogen fixation activity and a significant positive correlation (P < 0.05) with nifH gene number and nitrogen fixation amount.

Soil properties and its correlation with BNF, DNA_nifH and cDNA_nifH
The addition of these three kinds of straw significantly increased the content of soil total carbon and total nitrogen, but did not significantly change the soil carbon nitrogen ratio (Table S4). Moreover, there was no significant correlation between soil total carbon, total nitrogen concentration, carbon nitrogen ratio and biological nitrogen fixation (BNF), nifH DNA gene copy number (DNA_nifH), nifH RNA reverse transcription gene copy number (cDNA_nifH).  2 Spearman correlations between biochemical composition of different carbon materials and the changes of nifH gene numbers, nitrogen fixation amount and nifH transcript gene number ("the change" means: the nifH gene numbers, nitrogen fixation amount and nifH reverse transcription gene number in JWS, PB and MWS minus that in CK). "*" means P < 0.05, "**" means P < 0.01

Discussion
Biological nitrogen fixation is a very energy consuming process (Darian et al. 2019), and the addition of carbon source can increase nitrogen fixation (Vitousek et al. 2002;Gupta et al. 2014). However, residues with different biochemical compositions have different impacts on diazotrophs and nitrogen fixation. The addition of mature stage wheat straw (MWS) significantly increased fixed nitrogen in the soil (Fig. 1A) and rice plant (Fig. 1B), and the total nitrogen fixation in rice-soil system (Fig. 1C). But, the addition of jointing stage wheat straw (JWS) and poplar branch (PB) did not significantly increase the nitrogen fixation (Fig. 1C). This might be due to the increased number of diazotrophs and expression of nifH gene in MWS treatment ( Fig. 1D and E), while the diazotrophs' number did not increase significantly in JWS and PB treatments. The change of fixed nitrogen is significantly positively correlated with the increase of nitrogen fixing microorganisms numbers and nifH gene expression (Table 3). The number of nitrogen fixing microorganisms and the nifH gene expression are the keys to determine the amount of nitrogen fixation. Microorganisms require carbon as energy source and nitrogen to maintain and develop cells, proteins, enzymes, and hormones (Damascene et al. 2020). Nitrogen fixing microorganisms can utilize the nitrogen source (N 2 ) in the air, so compared with other microorganisms, they have a stronger competitive advantage in the high C/N environment. In addition, higher available nitrogen content will inhibit nitrogen fixation activity (Kox et al. 2016;Zhang et al. 2021a, b). This is also illustrated by significant positive correlations between C/N and nitrogen fixation, and the copy number of nifH DNA gene and nifH RNA reverse transcription gene (Fig. 2). On the other hand, although the concentrations of soil total carbon and total nitrogen concentration increased significantly with the addition of straw, they had no significant correlation with biological nitrogen fixation, nifH DNA gene copy number, nifH RNA reverse transcription gene copy number. This may be because the addition of straw did not significantly change the ratio of soil carbon and nitrogen (Table S4).
Although the increase of the fixed nitrogen was promoted by increasing the number of nitrogen fixing bacteria and expression of nifH gene in MWS treatment, their increase is not proportional.
The nitrogen fixation, the number and the expression of the nitrogen fixing bacteria in MWS treatment was 2.07, 4.88 and 28.57 times as high as those in CK treatment, respectively (Fig. 1C, E, and D). This indicated that the fixed nitrogen did not increase in the same proportion with the number of nitrogen fixing bacteria as we expected. This disproportionate change in the number of nitrogen fixing bacteria and nitrogen fixation has also been reported in other studies. (Tang et al. 2017;Zhang et al. 2021a, b). MWS significantly increased the number of alphaproteobacteria, gammaproteobacteria, betaproteobacteria, delta/epsilon subdivisions and Clostridia by 3.78, 7.33, 3.75, 16.65 and 10.33 times, respectively, but it did not significantly change the number of Oscillatoriophycideae, Nostocales and Stigonematales (Table 5), which are found to be the major contributors to the fixed nitrogen in paddy ecosystem by NanoSIMS and 15 N 2 -DNA-stable isotope probing (Wang et al. 2020a, b;Ma et al. 2019a, b). This may be due to the fact that mature wheat straw can provide more available carbon sources for heterotrophic nitrogen fixing bacteria (Charyulu and Rao 1979;Roger and Watanabe 1986), but Oscillatoriophycideae, Nostocales and Stigonematales were belong to cyanobacteria, which can use CO 2 as a carbon source through photosynthesis. These results indicated that most of the increased alphaproteobacteria, gammaproteobacteria, betaproteobacteria, delta/epsilon subdivisions and Clostridia, which most of them belong to heterotrophic nitrogen fixing bacteria, are not involved in the biological nitrogen fixation process, or they only have very low nitrogen fixation activity (Fig. 1F). Jointing stage wheat straw and poplar branch did not significantly change the number of diazotrophs. This may be due to the relatively high nitrogen content of jointing stage wheat straw and poplar branch. On the one hand, available nitrogen can inhibit the biological nitrogen fixation process, which has been well reported (Paul et al. 2000;Kox et al. 2016;Reed et al. 2011;Zhang et al. 2021a, b). In this study, the nitrogen content of JWS and PB is 3.00 and 2.07 times that of MWS respectively. 14 g carbon materials of JWS, PB and MWS containing 0.273, 0.188 and 0.091 g nitrogen were applied to 1.4 kg soil, respectively (Table 1) (equivalent to 336, 232 and 112 kg N ha −1 application rate, respectively). Furthermore, this higher nitrogen content of JWS and PB leads to a lower C/N, which makes the nitrogen in the straw easier to release (Antil et al. 2014). Higher nitrogen content in the straw also increases the growth of other non-nitrogen-fixing microorganisms which might lead to the inhibition of nitrogen fixing microorganisms, since straw with high nitrogen content (or low C/N) is easier to be degraded by microorganisms (Antil et al. 2014). Therefore, the low nitrogen content and high C/N ratio of mature wheat straw may be the key to improve the number of nitrogen fixing bacteria and nitrogen fixation, but it is might not be a good strategy to amend straw to increase heterotrophic diazotrophs' N 2 fixation in paddy field.

Conclusion
Residue with low N content could increase BNF through increasing population of heterotrophic diazotrophs and their expression. The BNF increased to 2.07 times that of CK treatment, while the diazotrophs number increased to 4.88 times, indicating that the BNF increase was much lower than the population increase. Resides with high N content could not increase BNF. These results indicated that residues with low N content might be more promising in enhancing paddy BNF.