We conducted two runs of gaseous emission measurement experiments at the Ruminants Feeding Standard Research and Development Center. Four Thai native cattle with the initial average body weights (BW) 313.8 ± 20.4 kg (Run 1) and 354.7 ± 21.6 kg (Run 2) were used for the experiment. The cattle were fed a restricted amount (2% of BW, dry matter [DM] basis) of a diet comprised of 70% Pangola grass and 30% commercial concentrate to meet their digestible energy requirements. The cattle were kept in individual tie stall pens equipped with a ventilated head-hood system, and they were fed at 09:30 and 17:00 each day. Water and a mineral block were freely accessed.

The emission of CH₄ was measured by the head-hood system installed at the Ruminants Feeding Standard Research and Development Center for 6 days with three periods (Supplementary Figure S1, Suzuki et al., 2008). Briefly, a constant airflow from the hood into the measurement system was made by blowers, and the concentrations of CO₂ and CH₄ were measured continuously by a non-dispersive infrared sensor (VA-5000, Horiba, Japan; IR-200, Yokogawa Electric, Tokyo). The cattle were always placed in the ventilated head-hood system for the entire days during the experimental period. The feed intake including the amount of leftover feed was recorded, and the total feces and urine were collected, weighed and samples (manure, 1 kg head⁻¹ d⁻¹; urine, 350 g head⁻¹ d⁻¹) were stored in a freezer (−20°C) for further analysis.

During the experimental period, manure was collected every day and accumulated for 3 weeks in the chamber for the subsequent GHG emission measurement as described below.

CH₄ and N₂O emissions from manure were measured using a dynamic chamber system as described (Osada and Fukumoto, 2001; Maeda et al., 2013) (Supplementary Figure S2). The chamber system was designed to estimate the total GHG emission from manure. It consisted of a polyvinyl chloride (PVC) chamber equipped with an air-blowing ventilator and a gas-sampling port on the ventilation exhaust. The chamber was 3 m width, 3 m depth, and 2 m height. Four 10-cm-dia. vent holes were installed in the upper part of the chamber and connected by PVC pipe to the ventilation blower, which was installed outside the chamber. The airflow was measured by a micromanometer and kept constant throughout the experimental period. Fresh air was introduced under the skirt of the chamber.

Air samples for the determination of the CH₄ and N₂O concentrations were collected in 15-ml vials by an automatic gas-sampling unit (Akiyama et al., 2009). The CH₄ concentrations were determined with a gas chromatograph equipped with a flame ionization detector (GC-8A, molecular sieve 5A column; Shimadzu, Japan), and the N₂O concentrations were determined with a gas chromatograph equipped with an electron capture detector (GC-14B, Porapak Q column, Shimadzu, Japan). The respective rates of emission were estimated as described (Fukumoto et al., 2003). The gas concentration of the ambient air was subtracted from that of the outlet air and then multiplied by the ventilation rate.

To test the reliability of chamber system, we conducted a recovery test by introducing a known amount of CH₄ into the chamber three times: 17 g of standard pure CH₄ gas (99.99%) was introduced into the chamber, and the CH₄ concentration was determined every 5 min for 2 h and every 10 min thereafter. The concentration of CH₄ was determined as described above. The recovery rates were 97–98% (Supplementary Table S1).

Manure was collected during the feeding experiment and accumulated in the chamber on a waterproof concrete floor, until the mass of manure reached ∼500 kg in total. Rice straw (25 kg) was mixed into one manure heap, while the other manure heap did not receive any bulking agent (control heap). Both manure heaps were mixed completely by hand at the beginning of the storage. At the start of the experiment, each 3-m², 0.2-m-high heap had a volume of 0.18 m³. For the monitoring of the manure microbiome, the manure heaps were mixed every 2 weeks to obtain homogenized samples. The temperatures of the manure and ambient air were measured hourly with an automated thermocouple (TR-73U, T&D, Matsumoto, Japan). Fresh samples (1 kg) were taken at every mixing, at the start and the end of the experiments.

The value of the total solids (TS) was measured after drying the samples at 105°C for 24 h. Volatile solids (VS) were measured after the samples were processed at 600°C for 2 h. The levels of crude protein (CP), ether extract (EE), and crude ash were determined by the standard method (Chemists and Horwitz, 1990). The gross energy was determined by an automatic adiabatic bomb calorimeter (C6000; IKA-Werke, Staufen, Germany). The neutral detergent fiber exclusive of residual ash without the inclusion of sodium sulfite (NDF) was determined as described (Van Soest et al., 1991). The apparent digestibility for nutrients was calculated according to the formula: (nutrient consumed − nutrient in feces)/nutrient consumed.

For the measurement of the inorganic-N in the manure, the pH, and the electrical conductivity, 5 g of fresh manure was placed in a 50-ml polypropylene tube with 40 ml of deionized water and then shaken (200 rpm, 15 min) and centrifuged (3000 g, 10 min). The supernatant was collected, and the inorganic-N (NH₄ ⁺, NO₂ ⁻ and NO₃ ⁻) concentrations in the supernatant were measured by the colorimetrical method (Bio-Rad, Hercules, CA, United States). The pH and electrical conductivity (EC) were determined with calibrated electrodes (Horiba, Fukuoka, Japan).

DNA was extracted from 0.2 g of manure samples using Isofecal for Beads Beating (Nippon Gene, Tokyo), quantified by a NanoDrop Lite spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States), and stored at −20°C until further analysis. Partial fragments of the 16S rRNA gene (the V4 hypervariable region) were amplified by a two-step polymerase chain reaction (PCR). Primers 515F and 806R (Caporaso et al., 2011) with Illumina adapter overhang sequences were used for the first-round PCR with 20 cycles, and indexes were attached to the amplicon with eight additional cycles. Each 20-µl PCR mixture contained 0.2 µl TaKaRa ExTaq HS DNA polymerase (TaKaRa Bio, Shiga, Japan) with 2 µl of buffer (10×buffer), 1.6 µl of 2.5 mM dNTP mix, 1 µl of each forward and reverse primer (10 mM), and 1 µl of template DNA.

The first-round PCR conditions were as follows: 94°C for 2 min; 20 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 30 s; and a final round of 72°C for 5 min. The PCR products were purified using an Agencourt AMPure XP purification system (Beckman Coulter, Indianapolis, IN, United States), and used for the second-round PCR with the following conditions: 94°C for 2 min; eight cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s; and a final round of 72°C for 5 min. Tag-indexed PCR products were purified again, and their quality and quantity were checked by an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States) and a Qubit 2.0 Fluorometer and dsDNA HS Assay Kit (Life Technologies, Carlsbad, CA, United States), respectively. Qualified amplicons were pooled in equal amounts and sequenced with a 250-bp paired-end sequencing protocol (Illumina, San Diego, CA, United States).

Raw sequence reads were processed by Qiime2-2019.7 (Bolyen et al., 2019). Paired-end sequences were merged and quality-filtered by DADA2 (Callahan et al., 2016), and the denoised feature table and amplicon sequence variants (ASVs) were used for the taxonomic diversity analysis. Taxonomic classifications were assigned using a naïve Bayes classifier trained on the Greengenes 13_8_99% database, and mitochondria or chloroplast sequences were removed (Bokulich et al., 2018). The statistical analyses for the diversity metrics and the principal component analysis were performed through QIIME 2 (diversity “core-metrics-phylogenetic”).

PICRUSt was used for predicting the function of the manure microbiome (Langille et al., 2013). The closed-reference OTUs were normalized by copy number, and a new matrix of predicted functional categories was created with the KEGG database. We used STAMP to analyze the PICRUSt output file (Parks et al., 2014). The DNA sequences from this study were deposited in the DDBJ Sequence Read Archive (accession numbers: DRR347213 to DRR347239).

The qPCR assays were performed with iTaq Universal SYBR Green Supermix and CFX96 (Bio-Rad) with 20 µl of a reaction mix that contained 20 ng of template DNA. The primer pairs for amplifying bacterial 16S rRNA gene, bacterial and archaeal amoA gene, and bacterial denitrification genes (nirS, nirK and nosZ) were used. The PCR conditions for each reaction are summarized in Supplementary Table S3. An external standard curve was prepared using serial dilutions of a known copy number of the plasmid pGEM-T Easy vector (Promega, Madison, WI, United States) containing each gene. The insert gene for the 16S rRNA gene and nirS was Paracoccus denitrificans (NCIMB 16712), and that for the amoA gene was Nitrosomonas europaea (NBRC 14298). Plasmids containing the cloned nirK gene (AB441832) were used for the standard curve for these genes.

The chemical analysis and gaseous concentration data were analyzed by an analysis of variance (ANOVA) using the general linear model procedure described by SAS (SAS Institute, 2001). Tukey’s multiple range comparison tests were used to separate the means. Probability (p)-values <0.05 were considered significant.

The results of the CH₄ emission measurement and feed digestibility are summarized in Table 1. In Runs 1 and 2, the cattle’s dry matter intake (DMI) values during the experiment were 5.42 kg/day ± 0.35 kg/day (n = 4) and 5.64 kg/day ± 0.35 kg/day (n = 4) and the emissions of CH₄ were 155.0 L/day ± 15.7 L/day and 174.1 L/day ± 22.1 L/day, respectively. The digestibility of DM, organic matter (OM), CP, ethyl extract (EE), NDF and gross energy (GE) ranged from 46.45 ± 1.95% to 81.3 ± 6.62%; all values were in the normal range, indicating that the values and manure obtained in this study can be used as representative of the typical beef manure in the local production of beef cattle.

The temperature profiles of the manure with and without the addition of rice straw are summarized in Supplementary Figure S3. The maximum temperatures in the manure heap without rice straw were 43.8 and 47.4°C for Runs 1 and 2, respectively, whereas the max. temperatures reached 66.2 and 65.1°C in the heap with rice straw mixed into it. These temperature increases are due to the active degradation of organic matter in the manure. It can thus be said that the addition of rice straw, which is abundantly available from local farmers, can enhance the active organic matter degradation of the manure.

The gaseous emission (CH₄ and N₂O) profile during the manure storage is provided in Figure 1. We detected significant emissions (especially for CH₄) during the manure accumulation period. The manure heap with rice straw had lower CH₄ and N₂O emission peaks in both runs. In Run 2, the manure mixed with rice straw did not have obvious N₂O emission peak, and the manure-only heap had an emission peak between weeks 4 and 6.

The total gaseous emissions in these runs are summarized in Table 2. The CH₄ emission in the control heap was 5.91 g/kgVS ± 2.39 g/kgVS, whereas the rice straw-mixed manure emitted 6.32 g/kgVS ± 3.69 g/kgVS, showing no significant difference between the treatments due to the large variation between two runs. These values can be converted to 4.58 kg-CH₄ head⁻¹ yr⁻¹ ± 0.8 kg-CH₄ head⁻¹ yr⁻¹ with no addition control, and 4.47 kg-CH₄ head⁻¹ yr⁻¹ ± 2.23 kg-CH₄ head⁻¹ yr⁻¹ for the rice straw-amended manure.

Regarding the emission of N₂O, the control heap emitted 1.19 gN₂O-N/kgN_initial ± 0.39 gN₂O-N/kgN_initial and the rice straw-added manure emitted 0.75 gN₂O-N/kgN_initial ± 0.72 gN₂O-N/kgN_initial; the difference between the treatments was not significant. These values could be converted into 14.19 gN₂O-N head⁻¹ yr⁻¹ ± 4.16 gN₂O-N head⁻¹ yr⁻¹ and 8.89 gN₂O-N head⁻¹ yr⁻¹ ± 8.39 gN₂O-N head⁻¹ yr⁻¹ in the control and the rice straw-added heaps, respectively.

Based on the above-described findings, the fates of the energy contained in the feed are summarized in Table 3. The results show that 6.87 ± 0.11% of the total energy in the feed was emitted and lost as enteric CH₄, and 43.1 ± 1.7% was emitted as manure. The manure was further degraded by the microbes and emitted as CH₄, which accounted for 0.69 ± 0.26% of the total energy contained in the feed.

We monitored the manure microbial community every 2 weeks (at every turning event) during the 12-week storage (Supplementary Figure S4). In both the control and rice straw treatments, the bacterial/archaeal community significant changed over time. In the comparison of the values obtained at the beginning (week 0) and those at the end of the storage (week 12), the relative abundance (n = 2) increased significantly in the phylums Actinobacteria (4.6 ± 2.1% to 8.3 ± 0.5% in the control heap and 4.8 ± 1.3% to 14.0 ± 3.8% in the rice straw-mixed heap), Chloroflexi (1.6 ± 1.7% to 12.7 ± 1.8% in the control heap and 1.8 ± 0.7% to 10.0 ± 4.1% in the rice straw-amended heap), and Planctomycetes (1.7 ± 0.5% to 3.5 ± 2.0% in the control heap and 1.4 ± 0.1% to 2.9 ± 1.5% in the rice straw heap). Over the same period, the relative abundance decreased significantly in the phylums Firmicutes (41.6 ± 3.2% to 25.3 ± 9.6% in the control heap and 35.4 ± 4.4% into 25.1 ± 5.2% in the rice straw heap) and Bacteroidetes (12.5 ± 8.1% to 9.7 ± 2.4% in the control heap and 13.7 ± 5.1% to 10.8 ± 1.5% in the rice straw heap) and the archaeal phylum Eurarchaeota (4.3 ± 0.3% to 0.6 ± 0.2% in the control heap and 3.0 ± 0.6% to 0.2 ± 0.1% in the rice straw heap).

Significant between-treatment differences were observed at the order level especially in Run 1 (Figure 2). The abundance of the order Methanobacteriales decreased significantly during the process in both treatments, but the decrease of their relative abundance was much faster in the rice straw-mixed heap than the control heap-, indicating that the rice straw amendment significantly enhanced the decay of the methanogens. Another significant difference between treatments was the abundance of Roseiflexales, which proliferated in the rice straw heap; it increased in particular from 1.2 to 12.1% between weeks 0 and 2. Mixing in the rice straw also affected other orders: the relative abundances of Actinomycetales, Cyatophagales, Bacillales were significantly higher, and those of the orders Clostoridiales, Bacteroidales, and TG3-1 were significantly lower in the rice straw heap compared to the control heap.

These significant differences between treatments were not so consistent for some of the orders in Run 2. The relative abandance of Roseiflexales increased in both treatments, and TG3-1 were more abundant in the rice straw heap at the end of the storage; the abundance of Methanobacteriales decreased significantly in both treatments. The results of the principal component analysis (PCA) on the microbial communities in Runs 1 (Figure 2C) and 2 (Figure 2D) well summarize this inconsistency. The plot for Run 1 shows that mixing rice straw into the manure induced significant differences in the microbial community, whereas for Run 2 the results of both treatments are plotted in similar positions.

The changes in several diversity indices are illustrated in Supplementary Figure S5. All five indices show that a significant reduction in microbial diversity occurred in the rice straw heap at day 14, at the beginning of the storage period; the values were restored on day 28 and were relatively stable thereafter. This significant reduction of diversity was estimated to be induced by the significant temperature increase in the rice straw heap.

The results of the estimated function of the manure microbiome by PICRUst are shown in Figure 2E and Supplementary Figure S6. The addition of rice straw significantly affected the function of the manure microbiome, in 65 (Run 1) and 60 (Run 2) of the 328 features in total. In Run 1 with clear difference between the treatments, the methane metabolism (p = 0.042) and some other microbiome functions such as glycolysis/gluconeogenesis (p < 0.001) were significantly affected by addition of rice straw, but the addition of rice straw did not have a significant effect on nitrogen metabolism (p = 0.368). In Run 2, the mixing in of rice straw significantly affected the metabolism of fatty acids (p < 0.01) including propionate (p = 0.01) and butanoate acid (p = 0.015), which are the precursors of CH₄, but the addition of rice straw did not significantly affect the CH₄ metabolism (p = 0.227) or N metabolism (p = 0.638). Since the metabolisms of CH₄ (p = 0.038) and N (p = 0.046) changed significantly over time (Supplementary Figure S7)—especially the methane metabolism, which jumped up in week 2 in both treatments—these high variations within the treatment might have masked differences produced by the two treatments.

Figure 3 shows the results of the qPCR for total bacteria, methanogens (mcrA), nitrifiers including ammonia-oxidizing bacteria (AOB) and archaea (AOA) by amoA gene, and denitrifiers (nitrite reducers: nirK and nirS, N₂O reducer: nosZ). The amount of total bacteria was relatively stable in both treatments and both runs, ranging from 7.5 × 10¹¹ to 5.7 × 10¹² copies g⁻¹ TS. In Run 1, the abundance of methanogens (mcrA) ranged from 5.3 × 10⁸ to 5.3 × 10⁹ copies g⁻¹ TS in the control heap, which tend to be higher than the rice straw-added heap (1.9 × 10⁸ to 1.3 × 10⁹ copies g⁻¹ TS). This result agrees well with the 16S rRNA gene amplicon sequencing data, and both sets of results well support the concept that methanogens were inactivated by the mixing of rice straw into the manure.

Regarding nitrifiers, AOB-amoA were more abundant than AOA-amoA, indicating that AOB were the main nitrifiers in the beef cattle manure. Both nitrifiers significantly increased during the storage period: in the control heap, AOB-amoA increased from 1.4 × 10⁸ to 5.5 × 10⁹ and AOA-amoA increased from 1.1 × 10⁶ to 1.4×10⁸ copies g⁻¹ TS. These results were also obtained in the rice straw-added heap in Run 1, which showed increases of AOB-amoA from 8.2 × 10⁷ to 1.0 × 10¹⁰ copies g⁻¹ TS and AOA-amoA from 3.5 × 10⁵ to 8.5 × 10⁸ copies g⁻¹ TS. This was not consistent in Run 2, in which AOB-amoA and AOA-amoA increased from 4.6 × 10⁶ to 2.2 × 10⁹ copies g⁻¹ TS and from 3.1 × 10⁶ to 4.0 × 10⁸ copies g⁻¹ TS in the control heap; in the rice straw heap, AOB-amoA and AOA-amoA were stable or even decreased, ranging from 1.5 × 10⁷ to 4.8 × 10⁷ copies g⁻¹ TS and from 2.7 × 10⁹ to 2.6 × 10⁵ copies g⁻¹ TS, respectively.

Concerning the denitrifiers, the abundances of nirK and nirS were relatively stable or slightly increased during the storage in both runs with both treatments. The nirK abundance ranged from 4.1 × 10⁸ to 3.2 × 10⁹ copies g⁻¹ TS in Run 1 but higher in Run 2, from 3.1 × 10¹⁰ to 1.8 × 10¹¹ copies g⁻¹ TS. The abundance values of nirS were more stable, indicating that it was not affected by the treatment, ranging from 5.4 × 10⁹ to 1.1 × 10¹¹ copies g⁻¹ TS in both runs. Regarding the N₂O reducer, the nosZ abundance generally increased slightly during the storage in both treatments, from 3.1 × 10⁸ to 2.0 × 10¹⁰ copies g⁻¹ TS in Run 1 and from 4.7 × 10⁸ to 9.3 × 10⁹ copies g⁻¹ TS in Run 2.