Effect of Soybean Meal on Nutritional Content, Fermentation Profile, and Bacterial Community Structure of Napier Grass Silage

Abstract

This study investigated the effects of soybean meal on fermentation characteristics, nutritional composition, bacterial community, and functional metabolic prediction in Napier grass silage. Napier grass was treated with soybean meal at 0% (CK), 10% (SA), 15% (SB), and 20% (SC) and ensiled for 7, 15, 30, 60, and 90 days. After 90 days of ensiling, SA, SB, and SC significantly increased (p < 0.05) lactic acid (LA) concentration, acetic acid (AA), the ratio of LA/AA, dry matter (DM), and crude protein (CP), and pH level while decreasing butyric acid (BA), ethanol, NH₃-N, NDF, ADF, and ADL compared to CK silage. At 7 days of ensiling, the SA, SB, and SC treatments increased the abundance of Leuconostoc, Pediococcus, and Klebsiella compared to the control. after 30 days of fermentation, the dominant genus shifted to Lactococcus in the SA, SB, and SC treatments, which was accompanied by a higher abundance of Klebsiella. In contrast, Lactobacillus became the dominant genus in the CK silage. Spearman’s rank correlation analysis revealed a positive correlation between DM content and Lactococcus and a negative correlation with NH₃-N concentration. pH and DM correlate negatively with Lactobacillus. LA, propionic acid, and AA show a negative correlation with Weissella. Enterobacter positively correlates with PA concentration. These findings demonstrate that SA improves the fermentation quality, and SB and SC could improve the nutritional content and microbial diversity abundance. We recommended ensiling Napier grass silage with SB (15%) doses and the best ensiling duration is 60 days.

1. Introduction

Ensiling is a widely adopted agricultural practice for preserving the forage crops during the off-season. During this process, lactic acid bacteria ferment water-soluble carbohydrates (WSC) into lactic acid, substantially reducing pH within the silo [1]. The resulting acidic environment effectively inhibited the growth of undesirable microorganisms while promoting the hydrolysis of hemicellulose and lignocellulose bonds. Research has demonstrated that various forage plants can be effectively ensiled with appropriate additives, providing a consistent and high-quality roughage source for ruminant livestock in semi-arid regions [2,3]. Furthermore, feeding ruminants with well-preserved silage has been shown to enhance overall dietary quality and improve animal performance [4,5,6].

Napier grass (Pennisetum purpureum Schumach.) is a perennial C4 grass, originating from African grasslands. It is widely used to feed ruminants in tropical and sub-tropical areas due to its high biomass yield and adaptability across various ecosystems. However, it has low protein, dry matter(DM), and WSC content [6,7], and a high fiber content [8]. The ensiling process is an effective way to preserve forage and improve its quality under optimal management [9]. Nevertheless, the ensiling of Napier grass presents some challenges, including inadequate fermentable substrates and a high moisture content, which can lead to unacceptable fermentation and effluent production [6,10]. Soybean meal is an effective additive with a high protein content, 40–49% DM and a well-balanced amino acid profile. It is widely used as a supplementary protein in animal feed, the production of antibiotics, and soy concentrate [11,12]. Additionally, it can serve as a fermented substrate [13].

Researchers have used various additives to achieve better quality Napier grass silage in terms of both fermentation and nutritional aspects, particularly regarding protein content, which is a crucial component for animal feed. For instance, adding urea effectively increased protein level. However, it may exacerbate unfavorable fermentation conditions, such as excessively high NH₃-N levels (>100 g/kg DM) and a raised pH level [10]. Additionally, the addition of dried Distillers’ grains with soluble (DDGS) effectively improved the nutritive value of Napier grass silage, due to their rich protein content and good absorption properties [14]. Additionally, wheat bran can also improve the fermentation and nutritional value of Napier grass [15]; however, both DDGS and wheat bran have low protein levels and high fiber content compared to soybean meal [16]. A study by Yunus et al. [17] found that adding soybean meal increased the pH, total nitrogen (TN), and ammonia nitrogen (NH₃-N/TN) levels, while reducing LA production. However, processing, such as heating, can effectively reduce the NH₃-N/TN levels and butyric acid and increase LA in young Napier grass (85 days of growth).

Therefore, soybean meal is a preferred additive due to its high nutritional value, especially protein, enhancing Napier grass silage quality. Although previous research indicated negative effects on young grass, the material used in this study was at a different growth stage. Furthermore, only a limited number of studies have investigated soybean meal as an additive for Napier grass silage. Thus, a better understanding of its effects on fermentation characteristics, nutritional indices, and microbial community structure is necessary to determine the optimal dose that will provide the best response. This study aimed to investigate the effects of soybean meal on the dynamics of bacterial community structure, nutritional content, and fermentation indices in Napier grass silage.

2. Materials and Methods

2.1. Silage Preparation

This study was conducted at Nanjing Agricultural University, Institute of Ensiling and Processing of Grass, China. Napier grass (Pennisetum purpureum Schumach.) was collected from the Nanjing Agricultural University experimental field, harvested at the end of November 2023 (after 150 days of growth), and then immediately chopped to a length of 2–3 cm using grass cutter machines. A 4 × 5 factorial design, four soybean meal doses (T) × five days of ensiling (D) with three replicates per dose, totaling 60 experimental silos, was used. The soybean meal doses were ensiled with Napier grass at inclusion rates of 0% (control, CK), 10% (SA), 15% (SB), and 20% (SC) on a fresh material basis (120 g). All samples were packed into a polyethene plastic bag (size: 22 cm × 32 cm), vacuum-sealed with a vacuum sealing machine (DZD-400, Nanjing Aomitai Technology Co., Ltd., Nanjing, China), then stored at room temperature. Subsequently, three silos from each dose were opened after 7, 15, 30, 60, and 90 days of ensiling to assess the fermentation quality.

2.2. Fermentation Characteristics and Chemical Compositions

A total of 50 g of fresh Napier grass silage was immediately dried in a forced-air oven at 65 °C for 48 h to calculate the DM content, then ground to pass through a 1 mm sieve and stored for chemical tests. TN content was analyzed using a Kjeltec 8400 Analyzer (FOSS Analytical AB, Höganäs, Sweden), and the CP was calculated as TN × 6.25 [18]. WSC content was enumerated via the Method described by Arthur Thomas [19]. The neutral detergent fiber (NDF) and acid detergent fiber (ADF) were quantified according to the Van Soest procedures [20]. A total of 30 g of fresh silage was placed in a 100 mL conical flask and covered with 60 mL of deionized water, then stored at 4 °C for 24 h. The liquid was filtered through two layers of cheesecloth and Whatman No.1 filter paper (Hangzhou Xinhua Co., Ltd., Hangzhou, China). Then, the pH was measured directly using a pH electrode (Hanna Instruments Italia Srl, Padua, Italy). The silage extract was stored in a freezer at −20 °C, before being taken (2 mL) and centrifuged at 13,000× g for 15 min at 4 °C, the supernatant was filtered through a 0.45 μm membrane for organic acids (lactic acid (LA), acetic acid (AA), propionic acid (PA), and butyric acid (BA)) and ethanol (ETH), according to [21], andanalyzed by Agilent high-performance liquid chromatography (HPLC) 1260 Agilent Technologies, Inc., Walderman, German, equipped with a refractive index detector maintained at (30 °C) and a carbomix^®H-NP Column 5:8% (5 µm, 7.8 × 300 mm). The mobile phase was 2.5 mM/L H₂SO₄, with a flow rate of 0.5 mL/min, and the column temperature was set at 55 °C. The phenol-hypochlorite method was used to measure NH₃-N in silage extract [22].

A ten-gram sample was homogenized in 90 mL of sterilized sodium chloride solution (0.85%) for 1 h, then filtered through two cheesecloth layers. The filtrate was serially diluted 10-fold for microbial enumeration, and the remaining solution was centrifuged at 8000× g for 10 min to investigate the microbial community diversity. Lactic acid bacteria (LAB)were counted on DeMan, Rogosa, and Sharp agar after 48 h of anaerobic incubation at 37 °C. Potato dextrose agar medium was used to determine the yeast content after 48 h of aerobic incubation at 30 °C. Aerobic bacteria were counted on a nutrient agar medium after incubation at 37 °C under aerobic conditions for 24 h.

2.3. Microbial Community Analysis

2.3.1. Bacterial DNA Extraction and PCR Amplification

A succession of microbes from throughout the fermentation phases are crucial for quality assessments. Fresh material, as well as material at seven and thirty days of ensiling, was taken. All samples from each dose group were analyzed to assess the dynamics of bacterial composition during the early fermentation phase, mid-phase, and before ensiling. Bacterial DNA was extracted via a Fast DNA^® SPIN kit for soil and amplified using a Fast Prep^® instrument (MP Biomedicals, Santa Ana, CA, USA). The manufacturer’s protocol for high-throughput sequencing was applied to the samples. The DNA samples were subjected to polymerase chain reaction (PCR) amplification, with the bacterial 16S rRNA V3–V4 regions being amplified using the primers 338F (ACTCCTACGGGAGGCAGCAG) and 806R (GGACTACHVGGGTWTCTAAT). The PCR settings were as follows: initial denaturation at 95 °C for 3 min, followed by 25 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, and elongation at 72 °C for 30 s [23].

2.3.2. High-Throughput Sequencing of Metagenomic DNA

After purification, the purified PCR amplicons were paired-end sequenced using the Illumina MiSeq PE300 platform (Illumina Inc., San Diego, CA, USA) at Majorbio Pharmaceutical Technology Co., Ltd. (Shanghai, China). All raw reads were checked to discard low-quality sequences using FLASH (Version 1.2.11) and QIIME (Version 1.7.0).

Operational taxonomic units (OTUs) were clustered based on a 97% sequence similarity cutoff using UPARSE (Version 7.1, https://drive5.com/uparse (accessed on 9 November 2025)). Chimeric sequences were then identified and removed using UCHIME. The ribosomal database project (RDP) classifier (http://rdp.cme.msu.edu/ (accessed on 9 November 2025)) was used to analyze bacterial compositions at the genus level using the SILVA138/16S rRNA bacteria and Unite database with a confidence threshold of 70%. Based on the analysis of the main coordinates (PCoA), the alpha diversity was estimated (Chao1, goods coverage, and Shannon index were calculated using Mouther (version 1.30.1, http://www.mothur.org/wiki/classify.seqs (accessed on 9 November 2025))).

Taxonomic classification was performed at the gene and phylum levels using the RDP algorithm to classify the representative sequencing of individual OTUs. The online platform of the Majobio I-sanger cloud platform (www.i-sanger.com). It was used to analyze all high-throughput sequence data [23].

2.4. Statistical Analysis of Data

Microbial data were converted to logarithmic form for presentation and statistical analysis. Data on silage characteristics (fermentation indices, chemical composition, and microbial counts) across the soybean meal doses, ensiling times, and their interaction were analyzed using two-way analysis of variance (ANOVA). Data on CP and carbohydrate components were analyzed using one-way ANOVA. This was performed using a general linear model procedure in SPSS software, version 27. Polynomial regression analysis used orthogonal contrasts (linear and quadratic) to examine differences in silage parameters across soybean meal doses and ensiling duration. Tukey’s multiple comparisons were used to determine statistical differences between means. Significant differences were noted when p < 0.05. The statistical significance of bacterial community structure difference among the soybean meal doses and ensiling durations (alpha and beta diversity indices) was assessed by Student’s t-test, ANOSIM with 999 permutations, and the Wilcoxon rank-sum test.

3. Results

3.1. The Chemical Composition and Microbial Population of Fresh Material

The fresh Napier grass (

Table 1. Characteristics of fresh Napier grass.

Items 1	Mean + SD 2
pH	5.75 ± 0.01
BC (mEq/kg DM)	41.12
DM (g/kg DM)	287.30 ± 0.03
WSC (g/kg DM)	53.30 ± 0.41
CP (g/kg DM)	40.82 ± 0.49
NDF (g/kg DM)	634.67 ± 0.27
ADF (g/kg DM)	364.89 ± 0.11
ADL (g/kg DM)	47.29 ± 0.70
Lactic acid bacteria (log10 CFU/g FM)	8.46 ± 0.20
Aerobic bacteria (log10 CFU/g FM)	9.14 ± 0.51
Yeasts (log10 CFU/g FM)	9.32 ± 0.99

¹ Abbreviations: BC, buffering capacity; DM, dry matter; WSC, water-soluble carbohydrate; CP, crude protein; NDF, neutral detergent fiber; ADF, acid detergent fiber; ADL, acid detergent lignin; mEq, milliequivalents; g/kg DM, gram/kilogram dry matter; CFU, colony-forming units; FM, fresh material of Napier grass; ² SD, standard deviation.

) exhibits a low content of buffering capacity, with low WSC, CP, and DM content. However, the microbial populations of lactic acid bacteria, aerobic bacteria, and yeast were 8.46, 9.14, and 9.32 log₁₀ CFU/g FM, respectively.

3.2. Effect of Soybean Meal on Silage Fermentation, Microbial Population, and Chemical Composition of Napier Grass

The concentrations of LA, AA, PA, and BA, and the LA/AA ratio (

Table 2. Volatile fatty acids of Napier grass silage treated with soybean meal doses and five ensiling times (days).

Items 1	T 2	Ensiling Days 3					SEM 4	p-Value 5			M C p 6
		7	15	30	60	90		T	D	T × D	L	Q
LA (g/kg DM)	CK	12.59 Bc	20.66 Ab	19.04 Bb	25.49 a	21.28 Bb	1.112	0.012	<0.001	<0.001	<0.001	<0.001
	SA	11.56 Bc	18.43 Ab	21.78 Ab	28.63 a	26.35 Aa	1.598				<0.001	0.002
	SB	12.76 Bc	14.21 Bc	22.14 Ab	27.11 a	24.61 Aab	1.499				<0.001	0.005
	SC	15.97 Ac	13.89 Bd	23.07 Ab	26.24 a	24.49 Aab	1.272				<0.001	0.009
	SEM 4	0.538	0.861	0.488	0.590	0.576
	p-value	0.003	<0.001	<0.001	0.894	0.014
AA (g/kg DM)	CK	1.99 Bb	3.15 b	3.81 ab	5.94 a	4.36 Bab	0.388	0.006	<0.001	0.300	<0.001	0.049
	SA	4.03 AB	4.66	5.52	5.91	6.00 A	0.256				0.004	0.373
	SB	3.91 ABb	4.04 b	5.14 b	6.61 a	5.17 ABb	0.273				<0.001	0.033
	SC	4.44 Aab	4.20 b	5.48 ab	5.89 ab	6.28 Aa	0.261				0.003	0.809
	SEM 4	0.372	0.212	0.283	0.155	0.248
	p-value	0.031	0.145	0.069	0.707	0.006
LA/AA (g/kg DM)	CK	3.05 b	6.56 Aa	5.05 ab	4.30 ab	4.89 Aab	0.400	0.034	0.002	0.026	0.566	0.097
	SA	2.87 b	4.14 Bab	4.02 ab	4.86 a	4.44 ABab	0.228				0.014	0.151
	SB	3.27 b	3.53 Bab	4.45 ab	4.10 ab	4.76 ABa	0.182				0.005	0.683
	SC	3.71	3.33 B	4.32	4.48	3.91 B	0.159				0.156	0.337
	SEM 4	0.347	0.406	0.230	0.111	0.133
	p-value	0.541	<0.001	0.444	0.809	0.015
PA (g/kg DM)	CK	0.00	0.63 B	0.03	0.39 B	0.65	0.105	<0.001	<0.001	<0.001	0.140	0.765
	SA	NDc	2.99 Aa	0.08 bc	0.29 Bbc	0.51 b	0.292				<0.001	<0.001
	SB	NDc	3.50 Aa	0.14 bc	0.80 Bb	0.38 bc	0.340				0.004	<0.001
	SC	NDc	3.70 Aa	0.46 bc	1.63 Ab	0.41 c	0.351				0.013	<0.001
	SEM 4	0.001	0.366	0.065	0.168	0.075
	p-value	0.217	<0.001	0.067	0.001	0.286
BA (g/kg DM)	CK	1.83 A	2.07 A	1.45	2.28	2.05 A	0.166	<0.001	<0.001	0.002	0.614	0.746
	SA	0.02 Bd	0.71 Bc	1.50 ab	1.78 a	1.37 ABb	0.166				<0.001	<0.001
	SB	0.02 Bc	0.70 Bb	1.67 a	1.63 a	1.13 Bb	0.163				<0.001	<0.001
	SC	0.02 Bd	0.67 Bc	1.59 b	1.75 a	1.65 ABab	0.175				<0.001	<0.001
	SEM 4	0.261	0.175	0.051	0.128	0.128
	p-value	0.005	<0.001	0.289	0.171	0.016

¹ Abbreviations: LA, lactic acid content; AA, acetic acid content; LA/AA, ratio lactic acid to acetic acid; PA, propionic acid; BA, butyric acid; ² CK, ensiling without additives; SA, 10% soybean meal; SB, 15% soybean meal; SC, 20% soybean meal; ³ the capital letters show the effect of the Treatment in the same column, and the small letters show the Effect of ensiling day in the same row, which differ at p < 0.05; ND, not detected; ⁴ SEM, standard error of the mean; ⁵ T, the effect of soybean meal doses; D, the effect of ensiling days; T × D, the interaction of soybean meal doses and ensiling days; ⁶ M C p, model contrast p-value; L: linear regression; Q, quadratic regression.

) were significantly influenced by soybean meal doses and ensiling duration, and their interaction (T × D). Adding different doses of soybean meal linearly increases (p < 0.05) the LA, AA, and PA, but linearly reduces the BA content in silage. The ensiling time significantly affects (p < 0.05) the organic acid content, with LA being highest linearly (p = 0.003) in SC at 7 d, and similarly leads to slight decreases linearly (p < 0.001) in SB and SC silage at 15 d. After 30 days of ensiling, both LA and AA showed a linear increase (p < 0.001) in SC, SB, and SA silages compared to the control silage. The best value was observed in SA silage at 90 days of ensiling. The LA/AA ratio decreased linearly (p < 0.001) in SA, SB, and SC silage during the first 15 days and after 90 days of ensiling, with lower (p = 0.015) LA/AA concentration observed in SC. Moreover, the concentration of PA was undetected across all soybean meal doses at the initial fermentation stage, 7d. After 15 days, it showed a linearly higher (p < 0.001) content, while it decreased after 30 days, but remained higher than the control; however, at 90 d, it appeared to have a low concentration. The BA content increases linearly (p < 0.001) with ensiling time, particularly between 7 and 60 days, but remained at a low concentration compared to the control silage, with the best values recorded in SB and SA silages after 90 days of ensiling.

Soybean meal doses and ensiling duration significantly (p < 0.001) affect the pH, DM, WSC, ETH, and NH₃-N (

Table 3. Chemical composition of Napier grass silage ensiled with soybean meal doses and five ensiling durations.

Items 1	T 2	Ensiling Days 3					SEM 4	p-Value 5			M C p 6
		7	15	30	60	90		T	D	T × D	L	Q
pH	CK	4.04 Da	3.66 Bb	3.98 a	4.06 Ba	3.97 Ba	0.043	<0.001	<0.001	<0.001	0.203	0.164
	SA	4.44 Ca	4.07 Ab	3.82 c	4.14 Bb	4.40 Aa	0.061				0.896	<0.001
	SB	4.55 Ba	4.24 Aab	4.06 b	4.25 ABab	4.41 Aab	0.052				0.309	0.001
	SC	4.63 Aa	4.08 Ab	4.07 b	4.37 Aa	4.42 Aa	0.059				0.544	<0.001
	SEM 4	0.066	0.067	0.053	0.039	0.057
	p-value	<0.001	< 0.001	0.296	<0.001	< 0.001
DM (g/kg FW)	CK	331.44 B	312.12 C	306.07 C	282.44 B	296.64 C	7.110	<0.001	0.020	0.033	0.073	0.418
	SA	376.79 AB	366.44 B	351.69 B	358.00 A	381.63 B	4.658				0.970	0.042
	SB	390.60 A	408.59 A	402.83 A	379.38 A	409.02 AB	3.965				0.745	0.839
	SC	384.74 ABb	418.61 Aab	442.52 Aa	395.55 Ab	426.84 Aab	5.680				0.066	0.055
	SEM 4	5.447	12.309	14.370	13.966	13.893
	p-value	0.044	<0.001	<0.001	<0.001	<0.001
WSC (g/kg DM)	CK	24.57 Ba	16.86 Db	16.94 Cb	9.47 Cd	10.39 Cc	1.409	<0.001	<0.001	<0.001	<0.001	<0.001
	SA	25.75 Ba	25.55 Ca	17.42 Cb	12.02 Ad	14.53 Bc	1.460				<0.001	<0.001
	SB	27.65 Aa	26.42 Bb	21.74 Bc	10.74 Be	15.85 Ad	1.654				<0.001	<0.001
	SC	28.48 Ab	32.82 Aa	23.29 Ac	12.14 Ae	15.53 Ad	1.996				<0.001	<0.001
	SEM 4	0.461	1.640	0.788	0.315	0.630
	p-value	<0.001	<0.001	<0.001	<0.001	<0.001
ETH (g/kg DM)	CK	3.75 c	5.90 Aabc	5.11 Abc	8.01 Aa	7.24 Aab	0.433	<0.001	<0.001	0.074	<0.001	0.288
	SA	2.47 d	3.20 Bcd	3.61 Bbc	4.84 Ba	4.47 Bab	0.235				<0.001	0.118
	SB	2.42 b	2.71 Bb	3.41 Bab	4.64 Ba	4.02 Ba	0.233				<0.001	0.227
	SC	2.80 c	3.08 Bbc	3.32 Bbc	5.37 Ba	4.37 Bab	0.266				<0.001	0.480
	SEM 4	0.199	0.386	0.236	0.405	0.426
	p-value	0.067	<0.001	0.002	<0.001	0.004
NH3-N (g/kg TN)	CK	51.49 Ad	47.75 Ae	66.41 Aa	63.17 Ab	57.30 Ac	1.808	<0.001	<0.001	<0.001	<0.001	<0.001
	SA	14.01 Bd	17.37 Bc	13.44 Cd	25.14 Ba	20.72 Bb	1.138				<0.001	0.957
	SB	13.96 Bb	11.04 Cc	16.42 Ba	12.17 Cc	11.86 Dc	0.509				0.013	0.004
	SC	8.64 Cd	10.29 Cc	12.31 Cb	7.51 De	13.19 Ca	0.555				<0.001	0.063
	SEM 4	4.954	4.430	6.560	6.315	5.346
	p-value	<0.001	<0.001	<0.001	<0.001	<0.001

¹ Abbreviations: DM, dry matter content; WSC, water-soluble carbohydrate content; ETH, ethanol; NH₃-N/TN, ammonia nitrogen/total nitrogen; FW, fresh weight; ² CK, ensiling without additives; SA, 10% soybean meal; SB, 15% soybean meal; SC, 20% soybean meal; ³ the capital letters show the effect of soybean meal doses in the same column, and the small letters show the effect of ensiling day in the same row differing at p < 0.05; ⁴ SEM, standard error of the mean, ⁵ T, the effect of soybean meal doses; D, the effect of ensiling days; T × D, the interaction of soybean meal doses and ensiling days; ⁶ M C p, model contrast p-value; L: linear regression; Q, quadratic regression.

). Compared with the control silage, the soybean meal doses SA, SB, and SC increase linearly (p < 0.001) the pH levels and the DM and WSC, while decreasing linearly (p < 0.001) the content of ETH and NH₃-N. At later incubation times, the pH was quadratically lower (p < 0.001) than that at 7 days. The most desirable pH values were observed at 15, 30, and 60 days in SA, SC, and SC silages. The DM content quadratically increased (p < 0.05) in SA and SB silages. The WSC content decreased linearly (p < 0.001) over the first 30 days of ensiling across all soybean meal doses. The ETH increased linearly (p < 0.001) with ensiling time, and a high concentration was observed at 60 d in all soybean meal doses silages. The NH₃-N fluctuated slightly between decreased and increased linearly (p < 0.001) in SB and SC silage after 30 d; overall, NH₃-N was lower (p < 0.05) in both SC and SB silage compared to SA and control silages.

The population of yeasts was significantly increased linearly (p < 0.05) by soybean meal doses compared with the control silage, particularly at higher doses (SC and SB), as shown in

Table 4. The microorganism population in Napier grass silage ensiled with soybean meal doses and five ensiling times.

Items 1	Treatment 2	Ensiling Days 3					SEM 4	p-Value 5			M C p 6
		7	15	30	60	90		T	D	T × D	L	Q
LAB (Log10 CFU/g FW)	CK	9.65 ABa	9.38 a	9.57 a	8.62 b	7.01 c	0.261	0.095	<0.001	0.182	<0.001	<0.001
	SA	9.47 Bab	9.96 a	9.02 ab	8.31 b	6.67 c	0.315				<0.001	0.005
	SB	9.89 Aa	9.68 a	9.19 b	8.36 c	7.45 d	0.234				<0.001	<0.001
	SC	9.92 Aa	9.72 ab	9.37 b	8.49 c	7.18 d	0.263				<0.001	<0.001
	SEM 4	0.062	0.085	0.090	0.062	0.148
	p-value	0.011	0.304	0.569	0.527	0.370
AR (Log10 CFU/g FW)	CK	9.46 a	8.42 Ba	6.18 c	6.54 Ac	4.58 d	0.459	0.063	<0.001	0.658	<0.001	0.525
	SA	9.58 a	9.65 Aa	7.16 b	5.60 Bc	4.57 d	0.536				<0.001	0.151
	SB	9.81 a	9.35 ABa	7.04 b	5.26 Bc	5.32 c	0.503				<0.001	0.046
	SC	9.56 a	9.51 ABa	7.13 b	5.33 Bc	4.88 c	0.521				<0.001	0.807
	SEM 4	0.090	0.178	0.155	0.167	0.148
	p-value	0.580	0.038	0.037	0.002	0.240
Yeast (Log10 CFU/g FW)	CK	8.97 Ba	8.72 a	6.82 b	6.13 bc	5.09 Bc	0.400	0.011	<0.001	<0.001	<0.001	0.724
	SA	9.20 ABa	9.58 a	7.38 b	6.06 c	5.05 Bd	0.454				<0.001	0.001
	SB	9.55 Aa	9.55 a	7.36 b	5.87 c	5.22 ABd	0.468				<0.001	0.152
	SC	9.54 Aa	9.70 a	7.15 b	5.66 c	5.57 Ac	0.467				<0.001	0.226
	SEM 4	0.078	0.162	0.083	0.091	0.070
	p-value	<0.001	0.055	0.135	0.080	0.005

¹ Abbreviations: LAB, lactic acid bacteria; AR, aerobic bacteria numbers; CFU, colony-forming units; FW, fresh weight. ² CK, silage control; SA, 10% Soybean meal; SB, 15% soybean meal; SC, 20% soybean meal; ³ The capital letters show the effect of the treatment in the same column, and the small letters show the that effect of ensiling day in the same row differ at p < 0.05; ⁴ SEM, standard error of mean; ⁵ T, the effect of soybean meal doses; D, ensiling day; T × D, the interaction of soybean meal doses and ensiling days; ⁶ M C p, model contrast p-value; L: linear regression; Q, quadratic regression.

. However, ensiling day linearly reduced (p < 0.001) the microbial population number after 30 days in LAB, aerobic bacteria, and yeast (Table 4).

3.3. Nutritional Composition of Napier Grass Silage

Soybean meal dose levels significantly affect the CP, ADF, NDF, and ADL content, as shown in

Table 5. Effect of soybean meal dose on the nutritional content of Napier grass silage when ensiled for 90 days.

Items 1	Experimental Silage 2				SEM 3	p-Value 4	p-Value of Contrast 5
	CK	SA	SB	SC		ANOVA	L	Q
a CP (g/kg TN)	40.59 D	144.74 C	166.57 B	201.50 A	17.311	<0.001	<0.001	<0.001
b NDF (g/kg DM)	610.61 A	562.11 C	574.53 B	575.20 B	5.229	<0.001	0.049	<0.001
c ADF (g/kg DM)	391.34 A	369.69 B	357.71 C	360.23 C	3.837	<0.001	<0.001	<0.001
d ADL (g/kg DM)	43.40 A	38.27 B	35.72 B	35.94 B	0.949	<0.001	<0.001	0.009
e HC (g/kg DM)	219.27 A	192.41 B	216.82 A	214.97 A	3.135	<0.001	0.713	<0.001
f CE (g/kg DM)	347.94 A	331.42 B	321.99 C	324.28 C	2.952	<0.001	<0.001	<0.001

¹ Abbreviations: CP, crude protein; NDF, neutral detergent fiber; ADF, acid detergent fiber; ADL, acid detergent lignin; HC, hemicellulose content; CE, cellulose content; g/kg DM, gram/kilogram dry matter; g/kg TN, gram/kilogram total nitrogen; ² CK, ensiling without additives; SA, SB, and SC, represent 10% 15% and 20% of soybean meal respectively; The capital letters show the effect of soybean meal doses in the same row, differing at p < 0.05; ³ SEM, standard error of the mean; ⁴ p-value, showing significant difference different between mean variables (p < 0.05). ⁵ L: linear regression; Q, quadratic regression, the lower-case letter in the item’s column, shown linear equation of each parameter ^a Y = 12.21 + 50.46 × x, R² = 0.885; ^b Y = 6.04 × 10² − 9.38 × x, R² = 0.335; ^c Y = 3.96 × 10² − 10.53 × x, R² = 0.785; ^d Y = 44.56 − 2.49 × x, R² = 0.719; ^e Y = 2.08 × 10² − 30.1 × x + 6.25 × x², R² = 0.345; ^f Y = 3.52 × 10² − 8.04 × x, R² = 0.773.

. The CP content increased linearly (p < 0.001) by 50.46% for each 1% of soybean meal added to Napier grass silage, the highest values observed in soybean meal doses SC, SB, and SA compared to the control silage. Furthermore, the NDF (p = 0.049), ADF (p < 0.001), and ADL (p < 0.001) content decreased linearly by 9.38%,10.53%, and 2.49%, respectively, for each 1% of soybean meal added to Napier grass silages. While linearly decreased (p < 0.001) CEL by 8.04% for each 1% of soybean meal added in Napier grass silage.

3.4. Effect of Soybean Meal on Bacterial Community of Napier Grass Silage

Bacterial Community Diversity and Abundance

The bacterial alpha diversity of Napier grass silage at 7 d and 30 d, as well as fresh material, is outlined in

Table 6. Alpha diversity of bacterial communities in Napier grass fresh material and silage ensiled with soybean meal over 7 and 30 days.

Ensiling Days 1	Treatments 2	Reads	OTUs	Shannon	Chaio1	Simpson	Coverage
0	FM	65,353	144	0.979	177.35	0.66	0.999
7	CK	54,204	106	1.732	127.04	0.29	0.9994
7	SA	62,279	98	2.258	119.92	0.15	0.9993
7	SB	56,435	105	2.15	133.72	0.17	0.9993
7	SC	63,933	98	2.19	119.79	0.17	0.9993
30	CK	52,312	93	1.19	119.74	0.50	0.9993
30	SA	48,542	124	1.95	148.14	0.22	0.9992
30	SB	50,453	116	2.01	163.78	0.21	0.999
30	SC	46,958	140	1.98	178.5	0.25	0.9989

¹ Abbreviations: 0: day before ensiling; 7: 7 days of ensiling, 30: 30 days of ensiling, ² FM: fresh material of mature Napier grass, CK: control, SA: 10% soybean meal, SB: 15% soybean meal, SC: 20% soybean meal, OTUs: operational taxonomic units.

. The number of sequences per sample ranged from 65,353 to 46,958 base pairs (bp), depending on soybean meal doses. The coverage value in all samples exceeded 99%, indicating that the sequencing depth was sufficiently reliable for the microbial community analysis. Regarding the term of operational taxonomic units (OTUs), at 7 d, the SA, SB, and SC silage had 98,105, and 98 OTUs, respectively, and the control had 106 OTUs.

At 30 d, OTUs showed an increase in all soybean meal dose silages, but a slight decrease in silage control compared to 7 d. The Shannon index was higher in SA, SC, and SB than in the silage control at 7 d. However, after 30 days, a decrease in SA, SB, and SC silage was observed, but CK still had a lower Shannon index. The Chaio1 and Simpson indices increased in the soybean meal dose and decreased in control silage after 30 days of ensiling, indicating that SC, SB, and SA silages have greater bacterial richness.

Studies on bacterial community dynamics during the ensiling process revealed species variation between silage groups, and fresh material, as well as differences in bacterial taxonomy at the phylum and genus level (Figure 1, Figure 2 and Figure 3). The bacterial community structure was visualized using principal coordinate analysis (PCoA) and non-metric multidimensional scaling (NMDS) based on Bray–Curtis distances (Figure 1). Both analyses revealed distance clustering patterns, clearly separating the microbial communities of fresh material from all ensiled samples. This indicates that the ensiling process itself significantly altered the bacterial composition. Furthermore, within the silage samples, separation was observed based on both the ensiling duration and the application of soybean meal dose. Specifically, after 30 days, the 20% soybean meal dose (SC) formed a distinct cluster, separate from the 10% (SA) and 15% (SB) doses, which clustered together. The PCoA (Figure 1A) showed that the first two principal coordinates (PCo1 and PCo2) accounted for 50.13% and 29.51% of the total variation, respectively. The NMDS plot (Figure 1B) corroborated this pattern with a satisfactory stress value (0.00), confirming the reliability of the observed ordination. Statistical validation using permutational multivariate analysis of variance (PERMANOVA) confirmed that the differences in bacterial community structure between the treatment groups were significant (p < 0.01).

The dominant phyla (Figure 2A) in the fresh Napier grass samples microbiota were Proteobacteria (99.15%). During ensiling, soybean meal doses as well as their interactions affect bacterial community dynamics. The Firmicutes increased in all silage groups, CK, SC, SB, and SA silage at 7 d. Subsequently, at 30 days of ensiling, Firmicutes showed a slight increase in CK, while in soybean meal doses, SA, SC, and SB decreased. In contrast, the abundance of proteobacteria was slightly increased in the SA, SC, and SB doses compared to the control silage.

Figure 2B illustrates the changed bacterial community abundance of Napier grass at the genus level. They identified more than five genera with relative abundances up to 1% in silage samples, while three genera appeared in fresh samples. The three high-abundance genera in the fresh sample were Acinetobacter (84.53%), Klebsiella (4.73%), and Enterobacteria (2.74%). But after ensiling, the abundance of Acinetobacter decreases in the early fermentation stage, while Klebsiella increases. In silage material, the dominant genera at seven days were Lactococcus (45.31%), Weissella (19.41%), Lactobacillus (18.58%), Klebsiella (5.08%), and Enterobacter (4.75%) in control silage. The dominant genera in SA were Lactococcus (43.27%), Leuconostoc (14.80%), Lactobacillus (11.39%), Klebsiella (10.54%), and Pediococcus (10.05%). Furthermore, in the SB dose, the dominant genera were Lactococcus (49.1%), Leuconostoc (18.70%), Klebsiella (9%), Weissella (7.04%), and Pediococcus (6.65%). In (SC) dose, the four dominant genera were Lactococcus (44.29%), Leuconostoc (19.38%), Klebsiella (9.06%), Pediococcus (8.09%), and Lactobacillus (7.72%). The ensiling time influences the bacterial dynamics.

After 30 days, the abundances of Lactobacillus (76.96%) increased, and those of Lactococcus (14.83%), Klebsiella (2.67%), and Enterobacter (1.82%) genera decreased in CK. In the SA group, the abundances of Lactococcus (49%), Klebsiella (21.1%), Lactobacillus (13.93%), and Enterobacter (5.25%) increased. While in (SB) silage, increase relative abundance of Lactobacillus (19.58%), Klebsiella (10.44%), however, reduce abundance of Leuconostoc (9.33%), and Lactococcus (47.63%), In SC silage increased the genera Lactococcus (61.14%), Klebsiella (12.47%), Lactobacillus (9.1%), and Enterobacter (4.21%).

The species differential analysis was practical in validating the changes in bacterial community among silage groups and fresh material (Figure 3). The soybean meal doses significantly influence the relative abundances of the following bacterial genera: Leuconostoc, Lactobacillus, Lactococcus, Weissella, and Pediococcus, with significant differences between silage groups at two ensiling days but not significant in the F sample. While Klebsiella, Enterobacter, Pantoea, and Enterococcus were significant in silage groups and fresh samples (F), the Acinetobacter genus was substantial only in the silage group (SA, SB, and SC) at 30 d and in the fresh sample (F).

3.5. Correlation Analysis Between Bacterial Community Genus and Fermentation Indices

The Spearman correlation analysis of bacterial genus and fermentation parameters (Figure 4). Lactococcus has a positive correlation with DM (p < 0.05) and a negative correlation with NH₃-N (p < 0.05). Lactobacillus correlated negatively with WSC, pH, and DM (p < 0.01). Weissella exhibited a positive correlation with pH and WSC (p < 0.001), and a negative correlation with PA, LA, and AA (p < 0.001). Enterobacter negatively correlates with NH₃-N (p < 0.001) and positively with PA and BA content (p < 0.01). Pediococcus and Leuconostoc are negatively correlated with the BA and PA (p < 0.01), while positively correlated with pH (p < 0.001).

3.6. Bacterial Function Prediction

PICRUSt2 functional predictive analysis of 16S and function abundance statistics (COG) (Figure 5) showed that the treatment affects the bacterial functions, slightly increasing the carbohydrate metabolism, energy metabolism, lipid transport, and metabolism, and decreasing the amino acid metabolism.

4. Discussion

The characteristics and microbial population of raw silage material primarily affect the fermentation process and overall quality of silage. For acceptable conservation, WSC should exceed 60 g/kg⁻¹ DM, dry matter >300 g/kg⁻¹ FM, and epiphytic LAB number > 10⁵ cfu g⁻¹ FM [24]. In this work, Napier grass showed WSC 53.30 g/kg⁻¹ DM, and epiphytic LAB >10⁵ cfu⁻¹ FM; however, its DM content of 287.30 g/kg FM (Table 1) indicates a moisture content > 70%, which may lead to unwanted fermentation or undesirable microbial growth. Additionally, the high fibrous fractions NDF and ADF, and low CP (40.82 g/kg DM) suggest that, as maturity advances, the cell wall components increase, thereby reducing the CP content [25].

4.1. Effect of Soybean Meal on the Fermentation Quality and Nutritional Composition of Napier Grass Ensiled for 90 Days

Good silage is achieved through stable fermentation, with LA as the main acid being produced; meanwhile, AA (1–3% DM), PA < 0.1% DM, and BA < 0.5–1.0% DM are present in grass silage with a DM content in the range of 25–35% DM [26,27].In this current work, the LA and AA linearly increased, and the PA content decreased linearly after 30 days in all soybean meal-dose silages. This finding suggests that dry soybean meal could affect the fermentation in the early phase by absorbing moisture, or may have limited directly available soluble sugar and a higher buffering capacity and DM content, which can slow the drop in pH and reduce theof LA and AA content in the early fermentation phase, at 7 d and 15 d.This May inhibit LAB activity [24], However, as incubation time progresses, the microorganisms hydrolyze the molecular weight of soybean meal, making it easier for the substrate to support fermentation, which may explain the fluctuations in LA production. The reduction in BA, ETH, and NH₃-N indicates that homolactic acid bacteria are the dominant genus in silage (Figure 2B). A previous study reported similar findings in Napier grass silage supplemented with dry crude soybean residue [28]. The DM and WSC content increased linearly in SA, SB, and SC silage (Table 3) because soybean meal acts as a moisture absorber, contributing to an increase in DM content. Additionally, it contained simple sugars; the reduction in WSC content occurred during incubation, indicating its conversion into LA and other products, as previously reported [29].

The pH value of high-quality silage should be lower than 4.2 [24]. In the current experiment, all soybean meal-dose silages (Table 3) had high pH values (ranging from 4.4 to 4.6 at 7 days and reaching 4.4 at 90 days) than this benchmark. However, the silage was still well-conserved, as indicated by the low NH₃-H (<26 g/kg of TN) and BA (<1.7 g/kg DM) concentrations. Despite this increase, it remains within the range of grass silage (25–35% DM) suggested by Kung et al. [27]. According to McDonald et al. [24], when the ratio of ammonia-N to total N is below 100 g/kg⁻¹ of TN, this is an indicator of protein-conservation and rich fermentation. In the present experiment, the NH₃-N/TN content was below 26 g/kg⁻¹ TN, in the SA, SB, and SC silage groups, providing evidence of limited proteolysis (Table 3). The dominance of lactic acid bacteria (LAB) compared to aerobic bacteria and yeasts (Table 4) is a key indicator of successful fermentation and good silage quality. This result aligns with previous findings [30].

The marked linear increase in CP content in soybean meal doses SC, SB, and SA compared to the control (Table 5) is attributed to the soybean meal, which is rich in protein (45 to 55% DM) [12]. The linear reduction in structural carbohydrates, NDF, ADF, and ADL in soybean meal doses in the current work suggests that these additives effectively dilute the overall fiber content through microbial activity during ensiling. The previous literature has mentioned that the addition of soybean meal to grass silage not only enhances the protein content but also improves the fermentation profile [31]. Overall, adding soybean meal to Napier grass silage improves the nutritional content and fermentation profile. Nonetheless, future research will be needed to achieve a balance between fermentation efficiency and nutritional value.

4.2. Effect of Soybean Meal on Bacterial Community Relative Abundance in Napier Grass Silage

Bacteria in Napier grass silage samples and fresh samples were sequenced using amplicon sequencing; the sequence number differed from 46,958 to 65,353 bp, for individual samples, and all of them had a coverage value of 99% (Table 4). This finding suggests that the sequencing depth was sufficient to adequately capture the microbial community diversity [29]. A high Shannon index was observed in the soybean meal dose group compared to the silage control after 7 d, and increased OTUs numbers, and increases in the Shannon and Chaio 1 indiceswere observed, in the treatment group compared to the control silage at 30 d. These findings suggest that the addition of soybean meal leads to an increase in bacterial community diversity, as stated in recent work [32]. The previous literature has mentioned that soybean meal serves as a nitrogen source, an essential nutrient for microbial growth [13].

The principal coordinate analysis (PCoA) and non-metric multidimensional scaling (NMDS) results exhibited differences in the configuration of the silage bacterial community in various fresh material and silage samples, indicating significant differences in bacterial composition among stowed silage and soybean meal efficiency, which could change the bacterial community structure. The shift from Proteobacteriota to Firmicutes after ensiling at the phylum level, resulted in a change in the environment’s anaerobic condition and reductions in pH [33]. The bacteria responsible for lactic acid production predominantly belong to the Firmicutes phylum, whereas the Proteobacteria are more common in fresh forage, as stated by Pang et al. [34].

Dynamic changes were observed in bacterial community abundance at the genus level (Figure 3). Lactococcus are the dominant genus in all silages. At the same time, Lactobacillus and Weissella were the most relevant genera in CK silage during the early fermentation stage, 7 d. This suggests that Lactococcus homofermentative LAB rapidly ferments sugar, converting it to LA and reducing the pH. However, it showed tolerance to a moderate pH level between 4.5 and 5, which represents a bridge between its initial alkalinity and the later increase in acidification. This action promotes the growth of LAB, including Lactobacillus and Weissella. These findings tend to support previous work [35]. The higher abundances of Leuconostoc and Klebsiella in soybean meal-dose silage could be attributed to the slower pH reduction (Table 2). This slower acidification likely results from the higher buffering capacity imparted by the protein-rich soybean meal. A study by Santos et al. [36] found that Klebsiella and Lactic acid bacteria compete during fermentation, increasing the acetic acid content in silage. Although some Klebsiella species are considered opportunistic pathogens and can compromise silage aerobic stability [37], their increased abundance in soybean meal-dose silages coincided with improved fermentation parameters (e.g., lower NH₃-N and BA). This suggests that the functional role of Klebsiella in silage may be complex and context-dependent, warranting further investigation. After 30 days of ensiling, the marked increase in Lactobacillus abundance 76% led to a decrease in Lactococcus by 14% and a decrease of 1.3% in Weissella, during CK silage. Additionally, there was a reduction in Leuconostoc and Pediococcus in SA, SB, and SC silage, because of the high LA and low pH levels at 30 d (Table 2 and Table 3). Du et al. [38] shared a similar view, stating that both genera contribute to LA production and are adapted to the moderately acidic conditions during the initial fermentation stage.

4.3. Linkages Between Fermentation Characteristics, Bacterial Community and Predicted Bacterial Function

The link between fermentation characteristics and bacterial community at the genus level is shown in Figure 4. Lactococcus exhibits a positive correlation with DM and a negative correlation with NH₃-N, suggesting that it helps to improve nutritional content. By inhibiting growth, undesirable bacteria responsible for protein breakdown can decrease the NH₃-N content in silage [39]. Lactobacillus correlated negatively with WSC, pH, and DM. This relationship indicates that it utilizes WSC as a fermentable substrate to produce LA, which can reduce the pH. In contrast, an inadequate amount of WSC results in a lower relative abundance of the Lactobacillus genus. Additionally, it prefers optimal moisture, as it is more abundant in the CK than in the SA, SB, and SC silage groups, which had high DM content. Weissella exhibited a positive correlation with pH and WSC, and a negative correlation with PA, LA, and AA. This suggests that Weissella acts as an initial colonizer, growing rapidly during the early fermentation stage when WSC levels are higher and the pH has not yet decreased substantially. The negative relationship with PA, LA, and AA is likely because it has low acid resistance (Figure 2B and Table 3). A study by Graf et al. [40] also showed a similar trend. Enterobacter negatively correlates with NH₃-N and positively correlates with PA and BA concentration. This suggests that Enterobacteria may inhibit urease activity, thereby reducing ammonia levels. As facultative anaerobic bacteria, they can convert soluble sugars and lactic acid to acetic acid and propionic acid [41]. Pediococcus and Leuconostoc negatively correlate with BA and PA, while they are positively correlated with pH, suggesting that Theos LAB ferments sugar to LA and AA; however, since the main product is LA, this considerably lowers the pH, which suppresses harmful microorganisms that are responsible for producing BA and PA. Their positive correlation with pH may be because they can grow rapidly under high pH conditions, as evidenced by their high abundance in the early fermentation phase at 7 d. A previous study assessed the high relative abundance of those bacteria in silage with a high pH [42,43].

As shown in Figure 5, the PICRUSt2 analysis (COG) classification revealed that the addition of soybean meal 10% (SA) led to a slight increase in the relative abundance of genes predicted to be involved in carbohydrate metabolism at the early fermentation stage (from 9.87% in CK to 10.01% in SA at 7 d). as well as increasing lipid transport and metabolism (from 2.39% in CK to 2.69% in SC), and a decreasing amino acid metabolism (from 9.41% in CK to 8.84% in SC). These results suggest that soybean meal supplementation alters microbial metabolic pathways, leading to enhanced carbohydrate and energy metabolism, and potentially improves lipid metabolism. These shifts are likely driven by substantial changes in microbial community structure and metabolic activity during ensiling. The previous literature stated that fermented soybean meal increased carbohydrate metabolism while decreasing amino acid metabolism [44].

5. Conclusions

The current study demonstrated that adding soybean meal to Napier grass silage significantly improves both fermentation quality and nutritional profile, further enhancing the bacterial community. Specifically reduce the ammonia nitrogen concentration and butyric acid, NDF, ADF, and ADL content, while increasing the CP, LA, AA, and DM content. Lactococcus was a dominant genus in the bacterial community of soybean meal-dosed Napier grass silage after 30 days of ensiling. These findings suggest that SA and SB doses could be recommended as appropriate levels to improve the fermentation quality of Napier grass silage. Nevertheless, the effects of soybean meal doses on aerobic stability, nutritional digestibility, and growth performance need further investigation.