The experiment was conducted in Huangbilong Village (29.02°N, 119.43°E), Jiangtang Township, Wucheng District, Jinhua City, Zhejiang Province, China, from June 2015 to May 2020. The area has a subtropical monsoon climate, with an annual average temperature of 17.3°C to 18.2°C and annual precipitation of 1,109.0 to 1,305.2 mm. A rice–rapeseed rotation system was used. The growing season of rice was from June to October, whereas that of rapeseed was from November to May of the following year. The soil type was loam soil. Before the experiment, topsoil (0-20cm) chemical properties were the following: pH 6.55, organic matter 32.1 g·kg−1, TN 1.24 g·kg−1, available N (AN) 193.39 mg·kg−1, available P 5.98 mg·kg−1, and available K 72.5 mg·kg−1.
Plots in the experiment were 2.7 m × 7.5 m. There were two treatments: CK (i.e., without straw return) and straw return (ST). There were three plots per treatment, for a total of six plots. Treatments were randomly distributed among the plots. In ST, average rice straw returned to the field was 8,100 kg·ha−1, whereas rapeseed straw input was approximately 6,150 kg·ha−1. The depth of returned straw was 20 cm. In the rice season, N fertilizer amount was 270 kg·ha−1, with 40% applied as basal fertilizer, 30% as tiller fertilizer, and 30% as ear fertilizer. Potassium fertilizer (K2O) amount was 108 kg·ha−1, of which 50% was tiller fertilizer and 50% was ear fertilizer. Phosphate fertilizer (P2O5) was applied at 67.5 kg·ha−1 as base fertilizer. In the rapeseed season, N fertilizer was applied one time at 51.75 kg·ha−1 as base fertilizer. Potassium fertilizer amount was 112.5 kg·ha−1, with 40% applied as basal fertilizer and 60% as topdressing. Other field management measures were consistent with local practices.
Soil sample collection
After harvest in May 2020, the five-point sampling method was used to collect soil cores (4-cm diameter) from each plot at five different depths (0–20, 20–40, 40–60, 60–80, and 80–100-cm). After removing debris and plant residues, soil samples from the same soil layer at the five sample points in each plot were mixed evenly. One portion of the samples was stored at 4°C in a refrigerator and used to measure DOC, microbial biomass C (MBC), MBN, ammonia N (NH4+-N), nitrate N (NO3−-N), and potential N fixation rate (PNFR). A second portion was stored at −20°C until DNA extraction. The remaining portion was air-dried and screened through a 2-mm mesh to determine SOC, TN, AN, and particulate organic carbon (POC). Mineral-associated organic C was determined by subtracting POC from total SOC.
Soil chemical properties
Dissolved organic C was measured by the deionized water extraction method (Haynes and Francis, 1993). Microbial biomass C and MBN were extracted by the chloroform fumigation–extraction method (Vance et al., 1987). Particulate organic C was measured by sodium hexametaphosphate dispersion (Cambardella and Elliott, 1992). Before measurement of SOC and TN, soil was air-dried and ground to pass through a 0.15-mm sieve. An automatic element analyzer (Vario Isotope Cube, Elementar, Jena, Germany) was used to measure SOC and TN (Nelson and Sommers, 2005). Soil NH4+-N and NO3−-N were extracted with 2 mol·L−1 KCl and measured on an AA3 continuous flow analyzer (Auto Analyzer 3, Bra+Luebbe, Hamburg, Germany) (Liu, 1996). Available N content was determined by the alkaline solution diffusion method after samples were air-dried, ground, and sieved (Page et al., 1982).
Measurement of potential nitrogen fixation rate
According to the steps in Hsu and Buckley (2009), 15N labeling was used to determine PNFR. First, 5.0 g of fresh soil sample was put in a 100-mL serum bottle, and soil water content was adjusted to 60% of field water holding capacity. The bottle was vacuumed and then filled with 20% O2 (v/v) and 80% 15N2 (v/v). This process was repeated three times for each sample. Unlabeled N2 was used in the control. Samples were incubated for 7 days at room temperature in a dark room. After incubation, samples were freeze-dried and screened through a 0.15-mm sieve. Total N content and 15N abundance were determined by an elemental analyzer–isotopic ratio mass spectrometer technique. Based on the difference in 15N content between labeled and unlabeled samples, the PNFR (µg(N)∙kg−1∙d−1) was calculated.
Abundance and population composition of soil nitrogen-fixing bacteria
DNA was extracted from soil samples, 1.0–1.5 g, with an E.Z.N.A.®soil DNA kit (Omega, Norcross, GA, USA). A Nanodrop 2000UV-VIS spectrophotometer (Thermo Scientific, Waltham, MA, USA) was used to determine DNA concentration and purity, which was then stored at −80°C. Sample DNA was measured using a MiSeq sequencing platform at Shanghai Majorbio Bio-pharm Technology (Shanghai, China), and the primer sequences were nifHF (5′-AAAGGYGGWATCGGYAARTCCACCAC-3′) and nifHR (5′-TTGTTSGCSGCRTACATSGCCATCAT-3′) (Rösch et al., 2002).
Raw sequences were processed using the Quantitative Insights Into Microbial Ecology (QIIME) website (http://qiime.org/scripts/assign_taxonomy.html) and the UPARSE platform (version 7.0.1090; http://drive5.com/uparse/). Standard primer sets and bar codes were excluded, and sequences with quality scores less than 20 were removed. Sequences shorter than 50 bp or those containing unresolved nucleotides were also removed. Extracting nonrepetitive sequences from the optimized sequences was convenient in reducing redundant computation in the intermediate process of analysis, and single sequences without duplication were removed. According to 97% similarity, nonrepetitive sequences (excluding single sequences) were clustered into operational taxonomic units (OTUs), with chimeras removed in the clustering process to obtain representative sequences of OTUs. A representative sequence was selected for each OTU under default parameters, and each representative sequence was assigned taxonomy using the Ribosomal Database Project (RDP) Classifier (Ou et al., 2019). Taxonomic information was obtained, and the community composition of each sample was determined at phylum and genus levels. The OTUs in each replicate were analyzed for richness and diversity. The sobs, Chao, Simpson, and Shannon indices were calculated to assess the alpha diversity of N-fixing bacteria, with sobs and Chao indices reflecting community richness and Shannon and Simpson indices estimating community diversity (Zheng et al., 2018).
Fluorescence quantitative PCR was used to determine the copy number of nifH genes in sample DNA, and the primers were the same as those used for sequencing. The PCR procedure was as follows: pre-denaturation at 95°C for 3 min; then, 35 cycles of denaturation at 95°C for 30 s, annealing at 57°C for 30 s, and extension at 72°C for 30 s; and a repair extension at 72°C for 8 min. Temperature range of the melt curve was 65°C to 95°C; the temperature was increased in increments of 0.5°C and held for 5 s to collect data.
Statistical analyses were conducted using SPSS 20.0 software for Windows (SPSS Inc., Chicago, IL, USA). Levene’s test (Lin et al., 2018) confirmed that all data met the assumptions of normality and homogeneity of variance required for ANOVA. One-way ANOVA and least significant difference (LSD) were used to assess significance of effects of ST and soil depth on soil C and N contents, PNFR, nifH gene copy number, and diversity of N-fixing bacterial populations. Pearson correlations were used to test relations between different SOC and N components and PNFR, nifH gene copy number, and N-fixing bacteria in R with the “corrplot” package (https://corrplot.r-forge.r-project.org).
Differences in N-fixing bacterial populations under different treatments were analyzed using principal coordinates analysis (PCoA) based on a Bray–Curtis distance matrix in R with the “vegan” package (Oksanen et al., 2019). Variance inflation factors were computed to confirm collinearities among environmental properties. Usually, environmental properties that have variance inflation factors (VIFs) greater than 10 are of little value. Environmental properties were selected by the functions envfit (permu = 999) and vif.cca in R with the “vegan” package and then were removed until all VIF values were less than 10 (Ji et al., 2020) (Table S1). Redundancy analysis was performed to analyze the relations between N-fixing bacterial populations and selected soil properties (DOC, POC, TN, NH4+-N, NO3−-N) and assess the effects of long-term straw return on composition and structure of N-fixing bacterial populations throughout the soil profile (Oksanen et al., 2019). The P and R2 values indicated the significance and importance of the effect of a soil variable on N-fixing bacterial populations. In addition, variance partitioning analysis (VPA) was also performed in order to quantify the contributions of soil organic C (SOC, POC, MOC, DOC, MBC) and N (TN, NH4+-N, NO3−-N) components to variation in composition of N-fixing bacterial populations (Ji et al., 2020). Results were plotted using Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA), GraphPad Prism 8.0 (GraphPad Software, California, USA), and R (v 4.1.0).