Strain‐Specific Fermentation Performance of Lactic Acid Bacteria Isolated From Maize and Napier Fodder During Maize Ensiling

TL;DR

This study identifies and tests lactic acid bacteria from maize and Napier fodder, finding that specific strains improve silage quality during fermentation.

Contribution

The study demonstrates the strain-specific efficacy of native Limosilactobacillus fermentum and Bacillus subtilis as bio-inoculants for maize silage.

Findings

Inoculated silage had significantly lower pH and ammonia-nitrogen levels compared to controls.

L. fermentum improved crude protein retention, while combined inoculation increased DM loss.

Inoculated silage showed higher LAB counts and suppressed yeast and mold growth.

Abstract

This study integrated phenotypic and molecular characterisation of LAB isolated from whole‐crop maize and Napier fodder and silage with laboratory‐scale ensiling to evaluate their fermentation efficacy in maize silage. Twenty‐one LAB isolates were screened, of which seven were confirmed by 16 s rRNA gene sequenced as Limosilactobacillus fermentum and
Bacillus subtilis
. The isolates exhibited broad physiological tolerance, growth at 15°C–43°C, pH 4.0–9.5, 7% NaCl, and the ability to ferment multiple carbohydrates. Selected strains (
L. fermentum
PQ482012 and
B. subtilis
PQ482016) were evaluated as silage inoculants in maize fodder over 45 days under controlled conditions. Inoculated silages exhibited significantly improved fermentation, with reduced pH (4.04–4.09; p = 0.004) and lower ammonia‐nitrogen (4.20–6.10 mg/100 mL; p = 0.001) compared to uninoculated control (pH, 4.19; 7.25 mg/100 mL).
L. fermentum
inoculation resulted in higher crude protein retention (92.20 g/kg of dry matter; p = 0.001), whilst combined inoculation enhanced acidification but caused greater DM loss (48.65 g/kg; p = 0.001). Inoculated silage also exhibited higher LAB counts, suppression of yeast, and absence of moulds. Overall, the native
L. fermentum
and
B. subtilis
strains exhibit strong strain‐specific fermentation traits, showing promise as bio‐inoculants for enhancing maize silage quality.

Isolation and identification of lactic acid bacteria from maize and Napier fodder and silage using physiological and biochemical tests, and Sanger sequencing. The isolated strains of Limosilactobacillus fermentum BLRI 1 and
Bacillus subtilis
BLRI 24 were utilised as an inoculant in maize silage.

Figures

16S rRNA profiles of 27F and 1492 R primers generated from Bacteria, M: Denotes 1 kb DNA ladder.

Phylogenetic analysis of (A) Bacillus subtilis and (B) Limosilactobacillus fermentum bacteria based on 16S rRNA genes by the neighbour‐joining method with 1000 times robustness by MEGA11 software of isolated lactic acid bacteria (LAB) from maize and Napier fodder and silage.

Fermentation characteristics and nutrient conservation of maize silage after 35 days of ensiling with selected Lactobacillus inoculants. (A) Dry matter (DM) in fodder and silage, (B) The pH values of fodder and silage, (C) Crude protein (CP), (D) ammonia‐nitrogen (NH3‐N), and (E) DM loss in silage. Values represent the mean ± standard deviation of each treatment (n = 3). a–c superscript letters in the same column indicate significant differences (p < 0.05; Duncan's test). Control, no microbial inoculants; LF, Maize inoculated with Limosilactobacillus fermentum PQ482012 strain; BS, Maize inoculated with Bacillus subtilis PQ482016 strain; Combo, Maize inoculated with the 1:1 mixer of LF and BS; SEM, Standard error of the mean; DM, Dry matter; CP, Crude protein; NH‐N, ammonia nitrogen.

Microbial composition (Log10CFU/g) of maize fodder inoculated with Limosilactobacillus fermentum PQ482012 and Bacillus subtilis PQ482016 strains during ensiling. Values are expressed as mean ± SD (n = 3). a–c superscript letters in the same column indicate significant differences (p < 0.05; Duncan's test). Control, no microbial inoculants; LF, Maize inoculated with L. fermentum PQ482012 strain; BS, Maize inoculated with B. subtilis PQ482016 strain; Combo, Maize inoculated with a 1:1 mixer of LF and BS; SEM, Standard error of the mean; ND, not detected.

Tables

TABLE 1. Morphological, physiological, and biochemical characteristics of Limosilactobacillus and Bacillus strains isolated from maize and Napier fodder and silage.

Isolates ID	S1				S2							S4				WMP		MS	NS
	1	2	11	19	24	27	28	29	30	33	34	36	38	39	44	45	47	49	50	51	52
Characteristics
Cell shape	R	R	R	R	R	R	R	R	R	R	R	R	R	R	R	R	R	R	CB	CB	C
Gram stain	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Catalase	+	+	−	+	−	−	−	−	−	−	+	−	−	+	+	−	−	−	−	−	−
Acid from glucose	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Gas from glucose	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Fermentation type	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero	Hetero
Growth at temperature (°C)
5	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
10	W	W	+	W	W	−	W	W	W	W	W	W	W	W	W	W	−	−	W	W	−
15	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
30	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
37	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
40	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
43	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	W
45	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	W	−	W	W	W	−
50	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
Growth in NaCl (%)
2.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
3.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
4.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
6.5	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
7.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
8.0	W	W	+	W	+	+	W	W	W	+	W	+	W	W	W	−	−	−	−	−	−
10.0	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
Growth at pH
3.0	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
3.5	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−
4.0	W	W	+	W	+	+	W	+	+	+	W	+	+	W	W	−	−	−	−	−	−
4.5	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	−	−	−	W	W	+
5.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
6.5	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
7.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
8.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
8.5	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
9.0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
9.5	+	W	W	W	+	W	W	+	+	+	W	W	+	W	W	+	+	+	−	−	−

TABLE 2. The sugar fermentation characteristics of lactic acid bacteria (LAB) isolates identified from maize and Napier fodder and silage.

Sugars	Isolates
	1, 2, 19, 34, 39, and 44	11, 27, 28, 47, and 49	24 and 29	30, 33, 36, and 38	45	50 and 51	52
Glucose, Dextrose, Sucrose, Arabinose, Ribose, Melibiose, Cellobiose, and Raffinose	+	+	+	+	+	+	+
Sorbitol	−	−	−	−	−	−	−
Bacterial species	Limosilactobacillus spp./L. fermentum	Bacillus spp.	Bacilus spp./B. subtilis	B. subtilis	B. subtilis/B. coagulanse	Bacillus spp./B. megaterium	Staphylococcus spp.

TABLE 3. Molecular identification of maize fodder isolates based on 16S rRNA gene sequencing and sequence similarity analysis (NCBI BLAST analysis).

Isolates	Name of bacteria	Strain name	Accession number	Number of compared bases	Resemblance (%)
1	Limosilactobacillus fermentum	CAU:3341	MF354239	1402	97.43
2	Limosilactobacillus fermentum	CAU:3341	MF354239	1402	97.99
19	Limosilactobacillus fermentum	CAU:3341	MF354239	1402	98.90
24	Bacillus subtilis	MJ14	FJ532063	1475	99.26
29	Bacillus subtilis	MJ14	FJ532063	1475	98.99
30	Bacillus subtilis	MJ14	FJ532063	1475	99.73
44	Limosilactobacillus fermentum	CAU:3341	MF354239	1402	98.97

TABLE 4. The registered strains with accession numbers of isolates identified from maize fodder.

Isolates	Name of bacteria	Strain name	Accession number	Base pairs
S1‐1	Limosilactobacillus fermentum	BLRI 1	PQ482012	1517
S1‐19	Limosilactobacillus fermentum	BLRI 19	PQ482013	1518
S1‐2	Limosilactobacillus fermentum	BLRI 2	PQ482014	1518
S2‐24	Bacillus subtilis	BLRI 24	PQ482016	1169
S2‐29	Bacillus subtilis	BLRI 29	PQ482017	1169
S2‐30	Bacillus subtilis	BLRI 39	PQ482018	1170
S4‐44	Limosilactobacillus fermentum	BLRI 44	PQ482015	1517

Full text

Introduction

1

Ensiling is an effective forage preservation method in regions with seasonal feed shortage. Maize (* Zea mays L*.) is widely used for silage production due to its high biomass yield, favourable carbohydrate content, and good ensiling characteristics, which together support efficient fermentation (Liu et al. 2025). Properly fermented maize silage improves feed palatability, digestibility, and preserves more nutrients, thereby boosting ruminant productivity (Kung et al. 2018). The success of ensiling largely depends on quickly establishing anaerobic conditions and the dominance of lactic acid bacteria (LAB), which convert water‐soluble carbohydrates into lactic acid, resulting in a rapid decline in pH and inhibition of spoilage microorganisms (Ma et al. 2025; Okoye et al. 2023).

The LABs are a diverse group of Gram‐positive, non‐spore‐forming, anaerobic or microaerophilic bacteria that naturally occur on the forage surface (Xia et al. 2023). However, their population density and species composition are influenced by crop type, maturity, harvesting conditions, and environmental factors (Okoye et al. 2023; Soundharrajan et al. 2023). Inconsistent epiphytic LAB populations can result in poor fermentation, nutrient loss, and silage spoilage. Consequently, isolation and selection of efficient LAB with strong fermentative and stress‐tolerant properties are crucial for improving silage quality. Previous studies have identified LAB strains such as Lactoplantibacillus plantarum, Levilactobacillus brevis, Pediococcus pentosaceous, Weissella cibaria, and Lactocaseibacillus rhamnosus as promising silage inoculants (Disanayaka et al. 2025; Dong and Yuan 2024; Peng et al. 2023; Pereira et al. 2019; Wang et al. 2017). These strains can enhance fermentation efficacy and aerobic stability by accelerating fermentation kinetics, increasing lactic acid production, reducing pH, and suppressing undesirable microorganisms, including Clostridia, yeast, and moulds. These improvements lead to reduce the production of butyric acid and ammonia‐nitrogen (NH_3_‐N), which are indicators of protein degradation and poor fermentation (Disanayaka et al. 2025; Monteiro et al. 2021).

Certain LAB strains possess antimicrobial properties against yeasts and moulds, thereby enhancing aerobic stability after silo opening; however, their effectiveness depends on the fermentation process under specific environmental and substrate conditions (Tahir et al. 2025). Additionally, Bacillus subtilis is a spore‐forming bacterium that can also produce lactic acid and tolerate harsh environmental conditions (Bai et al. 2022), making it potentially useful as a silage inoculant and complementing the role of LAB. Because maize has high fermentable sugars and low buffering capacity, monitoring key fermentation indicators, such as dry matter (DM), crude protein (CP), NH_3_‐N, and pH, and microbial populations is essential for evaluating silage quality and stability (Kung et al. 2018). Although commercial inoculants such as Lentilactobacillus buchneri, Lacticaseibacillus casei, L. plantarum, Lactococcus lactis, and L. rhamnosus are widely used (Disanayaka et al. 2025; Gallo et al. 2022; Yang et al. 2024); their performance may vary under different agro‐climatic conditions, particularly in stress factors such as temperature fluctuations and osmotic pressure (Dong and Yuan 2024). In contrast, indigenous or epiphytic LAB strains adapted to the local environment often exhibit superior competitiveness and fermentation performance (Wang, He, et al. 2018).

In this work, the objectives of the present study were to phenotypically and molecularly identify LAB isolates, followed by laboratory‐scale mini‐silo trials to evaluate fermentation performance, chemical composition, and microbiological properties. The present study focused on indigenous strains of Limosilactobacillus fermentum (LF) and Bacillus subtilis (BS) isolated from local fodder and silage sources to evaluate their efficacy as maize silage inoculants. These strains may possess adaptive traits and antimicrobial potential that enhance fermentation quality and silage stability by producing organic acids and other bioactive compounds. This approach aims to contribute to the development of effective, host‐specific silage inoculants tailored to maize ensiling conditions.

Materials and Methods

2

Isolation of Lactic Acid Bacteria From Fodder and Silage

2.1

Epiphytic lactic acid bacteria (LAB) were isolated from maize and Napier fodder and silage, where silage samples were collected after 60 days of ensiling. In total, 33 samples were collected (Napier: 12 fodder, 3 silage; maize: 15 fodders, 3 silage), chopped into 3–5 cm pieces, and macerated (20 g) in 180 mL sterile (0.85% NaCl; Merck, Darmstadt, Germany). Ten‐fold serial dilutions were plated on deMan, Rogosa, and Sharpe (MRS) agar (Condalab, Spain) and incubated anaerobically at 37°C for 48 h using a carbon dioxide (CO_2_) incubator (Memmert, ICO105, United States of America, USA). Colony counts were performed following the methods described by Cai et al. (2014). Presumptive LABs were identified based on colony morphology, Gram staining, and catalase activity (3% hydrogen peroxide, H_2_O_2_; Sigma‐Aldrich, Germany) following Negm El‐Dein et al. (2024). Colonies that were Gram‐positive, catalase‐negative, white/creamy white, and grey in colour were selected (Fossi et al. 2022). Isolates were purified by repeated streaking on MRS agar and preserved at −80°C in 40% glycerol (Sigma‐Aldrich, Germany) for further examination.

Morphological, Physiological, and Biochemical Tests

2.1.1

Gram staining was performed using overnight cultures grown in MRS broth (Condalab, Spain). Culture cells (1 mL) were centrifuged (3000 rpm, 10 min, 4°C) using a benchtop centrifuge machine (Hettich GmbH, Germany), washed in sterile water, smeared, heat‐fixed, and subjected to Gram staining. Catalase activity was determined by mixing a loopful of culture with a drop of 3% (w/v) H_2_O_2_ on a glass slide. Immediate effervescence indicated catalase positivity, whereas no bubble formation indicated catalase negativity. The fermentation type (homofermentative or heterofermentative) was assessed by carbon dioxide (CO_2_) gas production in Durham tubes (HiMedia, India, Cat#PW012) containing MRS broth supplemented with glucose (Sigma‐Aldrich, Germany). Tubes were inoculated with 50 μL of overnight culture and incubated at 37°C for 72 h. Gas accumulation in Durham tubes indicated heterofermentative metabolism. Growth under different conditions was examined in MRS broth incubated at various temperatures (5°C, 10°C, and 15°C for 5 days; 30°C, 43°C, 45°C, and 50°C for 48 h), and at different pH levels (3.0–9.5, adjusted with 1 N hydrochloric acid (HCl) or NaOH; Merck, Germany, Cat# 100317 & 106498). Salt tolerance was tested in MRS broth containing NaCl (Merck, Germany) at 2.0%, 3.0%, 4.0%, 6.5%, 7.0%, 8.0%, and 10.0% (w/v) at 37°C for 48 h (Shah et al. 2018). Growth was monitored by turbidity changes in the broth. Phenotypic identification was performed according to Bergey's Manual of Determinative Bacteriology. Carbohydrate fermentation profiles were determined using eight sugars (Sigma‐Aldrich, Germany): glucose (cat#G8270), sucrose (cat#S7903), arabinose (cat#A3256), ribose (cat#R7500), melibiose (cat#5500), cellobiose (cat#C7252), sorbitol (cat#1876), and raffinose (R0514). Overnight activated cultures were centrifuged at 10,000 rpm for 10 min, and cell pellets were inoculated into sugar fermentation broth (10 mL) MRS broth with phenol red indicator (Merck, Germany) and inverted Durham tubes supplemented with 200 μL of sterile sugar solution (5%, w/v, filter sterile, 0.22 μm; polyvinylidene fluoride, Sigma‐Aldrich, Germany). Then the tubes were incubated to allow for fermentation at 37°C for 24 h. The change of colour of the sugar solution from red to yellow by producing acid or acid and gas in the Durham tubes was then observed. A colour change from red to yellow indicated acid production, whilst gas accumulation in Durham tubes indicated gas production. Broth with cells but no sugar and broth with sugar but no cells were included as controls.

Molecular Identification of Lactic Acid Bacteria

2.1.2

Genomic deoxyribonucleic acid (DNA) was extracted from 24 h MRS broth cultures using an automated DNA extractor (Maxwell, 16; Promega, USA) and quantified with a NanoDrop spectrophotometer (ND2000; Thermo Scientific, USA). The 16S rRNA gene was amplified using universal primers 27F (27F, 5′‐AGA GTT TGA TCM TGG CTC AG‐3′) and 1492R (1492R, 5′‐CGG TTA CCT TGT TAC GAC TT‐3′) in a 20 μL PCR master mixture with Promega master mix (Promega, USA). The cycling conditions were: 95°C for 3 min; 35 cycles of at 95°C for 30 s, 48°C for 30 s, 72°C for 90 s; and a final extension at 72°C for 5 min using a PCR machine (Gene Atlas, G2; Astec, Japan). Amplicons were visualised by agarose gel electrophoresis (Figure 1), then purified, and sent for commercial sequencing (Macrogen Inc., Korea). Assembled sequences were analysed with BioEdit version 7.2 software and compared with National Center for Biotechnology Information (NCBI) GenBank sequences using Basic Local Alignment Search Tool (BLAST). Multiple sequence alignment was performed using the ClustalW programme (Thompson 1997), and a phylogenetic tree was constructed with MEGA 11 software (Tamura et al. 2021) by the neighbour‐joining method with 1000 bootstrap replications (Saitou and Nei 1987). Nucleotide substitution rates were calculated according to (Ohta and Kimura 1972).

16S rRNA profiles of 27F and 1492 R primers generated from Bacteria, M: Denotes 1 kb DNA ladder.

Maize Silage in Experimental Silos

2.2

Two strains, Limosilactobacillus fermentum PQ482012 (BLRI 1) and Bacillus subtilis PQ482016 (BLRI 24), previously isolated and identified, were selected to evaluate their efficacy on maize silage. The maize (approximately 3 months old) was harvested from the Fodder Field of Bangladesh Livestock Research Institute (23°55′11.3″ N 90°25′14.2″ E) during the late winter season. The maize was chopped into 1–2 cm pieces using a local straw‐cutting machine (CIMMYET, DANIDA type, BLRI‐modified) and quickly transported to the laboratory. Inoculants were prepared according to Jung et al. (2024) with slight modifications. Briefly, isolates were cultured in MRS for 24 h, centrifuged, and washed with phosphate‐buffered saline (PBS, Sigma‐Aldrich, Germany) to obtain cell pellets. A liquid inoculum was prepared in saline solution containing a minimum bacterial cell concentration of 10^6^–10^7^ colony‐forming unit (CFU)/ml by adjusting the optical density (OD_600nm_) of 0.78–82 using a Double Beam UV–VIS Spectrophotometer (Electronics, 2377, India). Twenty laboratory‐scale mini silos (1 kg capacity) were prepared with the following treatments: (i) Control, without inoculants (equal volume of distilled water); (ii) LF, maize inoculated with L. fermentum PQ482012; (iii) BS, maize inoculated with B. subtilis PQ482015; and (iv) Combo, maize inoculated with the 1:1 mixer of LF and BS. Each silo consisted of 1 kg of chopped maize mixed with 80 mL of distilled water containing inoculum. The silos were made of cylindrical plastic containers (15 cm in height and 10 cm in diameter) and were vacuum‐sealed using a Panasonic MC‐CL601 vacuum cleaner to maintain anaerobic conditions. All silos were weighed and stored at room temperature for 45 days, with four replicates per treatment.

Chemical Analysis

2.3

Fresh maize and silage samples were analysed for dry matter (DM) and pH. Approximately 200 g of silage was collected from the top, middle, and bottom of each silo, pooled, and subsampled. Portions were stored at −20°C for further chemical and proximate analyses. Silage samples were collected after 60 days of fermentation. One subsample was weighed and oven‐dried (Memmert ULE 400, Germany) at 55°C for 72 h, whilst another was used for water extracts to determine the pH, NH_3_‐N, and microbial counts. Dried samples were ground (Willey grinder, 30‐mesh) and stored in labelled plastic containers for analysis of DM (AOAC 1990), crude protein (CP; AOAC 1990), neutral detergent fibre (NDF; Van Soest et al. 1991), and water‐soluble carbohydrates (Weiß and Alt 2017). For pH determination, 25 g fresh or ensiled maize was homogenised in 225 mL of 0.1% (w/v) sterile peptone water (Difco, Becton Dickinson, USA) and shaken for 20 min. The 5 mL of filtrate (through four layers of cheesecloth) was centrifuged (10,000 rpm, 15 min, 4°C), and the supernatant was collected for NH_3_‐N calculation. pH was measured with a bench‐top pH metre (Mettler Toledo, SevenCompact, Switzerland).

Microbial Analysis

2.4

Water extracts were also used for the enumeration of microorganisms. Sequential 10‐fold dilutions were plated to quantify LAB, yeasts, and filamentous fungi. LAB were enumerated on MRS agar supplemented with nystatin (4 mL/L) and incubated at 30°C for 72 h. Yeasts and moulds were plated in Potato Dextrose Agar (Difco; Becton Dickinson, USA) and incubated at 28°C for 72 h. Colonies were counted on plates containing 30–300 CFU, and yeasts were distinguished from moulds based on colony morphology and cell structure.

Statistical Analysis

2.5

The data (proximate composition, pH, NH_3_‐N contents, and microbial counts) were analysed in a complete randomised design (CRD) with four treatments (control, LP, BS, and Combo) and three independent replications (n = 3). The results of proximate composition, fodder and silage pH, and microbial counts were expressed as mean ± standard error. Before analysis, all data were tested for normality (Shapiro–Wilk test) and homogeneity (Levene's test). Data meeting the assumptions of parametric analysis were subjected to analysis of variance (ANOVA) using a general linear model (GLM) at a 95% confidence level, whilst the Duncan's Multiple Range Test (DMRT) value compared means. Microbial data were converted to log_10_‐transformed (CFU/g) before analysis. All data analysis was performed using SPSS software (Version 25; IBM‐SPSS, USA, 2023) and R‐Studio (Version 4.3.1; R Foundation for Statistical Computing, Australia) for line graphs, bar plots, and box plots.

Results and Discussion

3

LAB Strains Characteristics

3.1

Lactobacillus and Bacillus are commonly found and identified in native forage and silage produced through natural and inoculated LAB (Disanayaka et al. 2025; Peng et al. 2023; Yang et al. 2024). Whilst different LABs dominate forage and silage samples, studies have shown that L. fermentum and B. subtilis have been identified in Napier and maize fodders and their silage (Ayodele et al. 2024; Puntillo et al. 2020; Zhang et al. 2024). However, identifying differences amongst taxonomically related species can be challenging based on morphological, physiological, and biochemical tests (Wang et al. 2019). As a result, this study aimed to assess and screen the LAB from epiphytes through morphological, physiological, and biochemical profiling. The identified LAB isolates from maize and Napier fodder and silage are depicted in Table 1, whilst sugar fermentation profiles are summarised in Table 2. A total of 21 isolates were identified, including 17 from maize fodder, one from maize silage, and three from Napier silage, highlighting the diversity of LAB in these sources (Silva et al. 2025; Yang et al. 2024). The table indicated that most isolates were Gram‐positive, rod‐shaped, catalase‐negative, and heterofermentative, indicating their Limosilactobacillus spp. characteristic. In contrast, catalase‐positive rod‐shaped and coccobacillus heterofermentative isolates were identified as Bacillus spp., which was supported by the sugar fermentation tests (Table 2). Growth response to temperature showed the mesophilic characteristics, with robust growth between 15°C and 45°C, whilst all grew well at 15°C–43°C and weakly survived at 10°C and 45°C, but did not grow at 5°C or 50°C. Salt tolerance up to 7.0% NaCl and growth across a wide pH range (4.0–9.5, optimum at 5.0–8.5) demonstrated strong adaptability to the silage environment. Rod‐shaped isolates generally grew at pH 4.0–9.5, whereas cocci and cocco‐bacilli grew between pH 4.5–9.0. These characteristics highlighted the dominance of Limosilactobacillus spp. in maize and Napier fodder and silage, consistent with the previous reports that hetero‐fermentative LAB are key players in silage (Ayodele et al. 2024; Bao et al. 2024) due to their lactic and acetic acid production, osmo‐tolerance, and pH resilience (Ogunade et al. 2018; Okoye et al. 2023). Their ability to grow under moderately acidic conditions (pH 4.0–5.0) is particularly important for survival during the early fermentation stage, whilst salt tolerance ensures persistence under variable storage conditions (Muck et al. 2018). All isolates were able to utilise a broad range of carbohydrates, including glucose, dextrose, sucrose, arabinose, ribose, melibiose, cellobiose, and raffinose, but none utilised sorbitol (Table 2). Based on their biochemical profiles, the isolates were grouped into different species; isolates 1, 2, 19, 34, 39, and 44 were identified as Limosilactobacillus spp. ( L. fermentum ), whilst isolates 11, 27, 28, 47, and 49 were classified as Bacillus spp. Isolates 24, 29, 30, 33, 36, and 38 were further characterised as B. subtilis , whilst isolates 45, 50, and 51 resembled Bacillus spp. (* B. coagulans/B. megaterium *), and isolate 52 was identified as Staphylococcus spp. The results demonstrated the heterogeneity of microbial population in maize and Napier silage, dominated by L. fermentum alongside several B. subtilis . The ability of all isolates to ferment common sugars indicates metabolic flexibility, a trait advantageous in the carbohydrate‐rich silage environment (Guo et al. 2022).

16S rRNA Gene Sequence Analysis

3.2

Molecular identification of these isolates was performed by sequencing the 16S rRNA gene, and phylogenetic trees were constructed to assess genetic divergences. A total of 22 identified isolates were subjected to commercial Sanger sequencing (16 s rRNA gene sequencing) for further confirmation, of which seven were successfully confirmed. Isolates 1, 2, 19, and 44 clustered closely with L. fermentum , showing 98%–99% sequence similarity and 98% bootstrap support (Table 3). These isolates were assigned under accession number MF354239 and strain designation CAU:3341. Similarly, isolates no 24, 29, and 30 exhibited 98%–100% sequence similarity to B. subtilis, with 97% BLAST support, and were assigned accession number FJ532063, strain MJ14. The phylogenetic tree showed B. subtilis and L. fermentum isolates clustering with their respective reference sequences (Figure 2). The representative sequences of L. fermentum (four isolates) and B. subtilis (three isolates) were deposited in GenBank and registered under accession numbers PQ482012, PQ482013, PQ482014, PQ482015, PQ482016, PQ482017, and PQ482018, respectively (Table 4). Consistent with the present findings, Zhang et al. (2024) identified the B. subtilis from whole‐plant corn silage using morphological, physiological, and biochemical profiling, along with 16S rDNA gene sequencing. Furthermore, more recent studies have confirmed the isolation of L. fermentum from natural silage; for example, Bao et al. (2023) successfully isolated and identified L. fermentum from native grass silage with genetic and phenotypic characterisation consistent with our findings.

Phylogenetic analysis of (A) Bacillus subtilis and (B) Limosilactobacillus fermentum bacteria based on 16S rRNA genes by the neighbour‐joining method with 1000 times robustness by MEGA11 software of isolated lactic acid bacteria (LAB) from maize and Napier fodder and silage.

Chemical Composition and Fermentation Characteristics of Maize Fodder

3.3

The chemical composition and fermentation profiles of maize fodder are critical for optimising its nutritional values for livestock. After 35 days of ensiling maize with L. fermentum (LF) and B. subtilis (BS) inoculants, the chemical and fermentation characteristics are presented in Figure 3. The DM content of fresh fodder did not differ significantly (p = 0.204) amongst the treatments (control, LF, BS, and combo groups), with values ranging from 248.27 to 250.20 g/kg of DM (Figure 3A). This indicates that the pre‐ensiling had consistent initial DM content (Serva 2024), ensuring similar moisture levels across all groups. However, silage DM content was significantly affected by the treatments (p = 0.004). The control group has the highest DM content (204.62 g/kg), which was statistically similar to the LF (204.08 g/kg) and BS (202.63 g/kg) groups. The combo group had a significantly lower DM content (200.11 g/kg) than the control, supporting the findings of Su et al. (2024), which indicated enhanced microbial activity of Lactobacillus and Bacillus synergistically during fermentation, improved water retention. Furthermore, Yuan et al. (2022) proposed that the over‐conversion of nutrients through Lactobacillus fermentation into lactic acid, volatile fatty acids, and NH_3_‐N leads to a decrease in silage DM content. The high‐quality and well‐preserved silage should have a pH range of 3.8 to 4.2 (Ahmadi et al. 2019; Guo et al. 2021). In this study, the initial pH of the fresh fodder before ensiling ranged from 5.68 to 5.75, with the Combo group exhibiting significantly higher pH (5.75, p < 0.001) than other treatments (Figure 3B). All treatments showed a marked decrease in pH upon ensiling, with significant differences amongst them (p = 0.001). The LF‐ and BS‐treated silage had lower pH values (4.04 and 4.09, respectively) compared to the control and Combo, highlighting better fermentation and higher organic acid production (Jung et al. 2024). Similarly to the present study, Su et al. (2024) found that adding L. plantarum and B. subtilis to corn stalk silage led to a quicker pH drop to about 3.8 over 30 days. CP retention in silage is a critical indicator of silage nutritional value. In this study, Boxplot analysis (Figure 3C) revealed significant variation in silage CP levels amongst treatments (p = 0.001), with values ranging from 85.10 g/kg in the control to a significantly higher 92.20 g/kg in the LF treatment. Notably, all inoculated groups (LF, BS, and combo) exhibited higher levels of CP than the control, demonstrating that inoculation with LAB effectively reduced proteolytic activities during ensiling (Uher et al. 2019). The uninoculated silage showed lower CP values of 85.10 g/kg, likely due to increased proteolytic activity, leading to reduced nitrogen retention (He et al. 2020). The LF treatment, inoculated with L. fermentum , achieved the highest CP retention due to its rapid acidification, as indicated by a lower silage pH (4.04), which inhibits plant protease and spoilage microbes, minimising protein degradation (Bao et al. 2024). DM loss is an important parameter that reflects the extent of fermentation, thus directly impacting silage yield and energy conversion efficacy. As shown in the boxplot (Figure 3D), DM loss was significantly influenced by the treatments (p = 0.001). The Combo group experienced the highest DM loss (48.65 g/kg), whereas the LF, BS, and control treatments showed comparable and lower losses ranging from 44.37 to 45.64 g/kg. The NH_3_‐N is a key indicator of protein degradation during silage fermentation, with elevated levels reflecting extensive proteolysis and poor fermentation. In this study, the concentrations of NH_3_‐N (mg/100 mL) varied significantly (p = 0.001) amongst the different treatments (Figure 3E). The control group showed the highest concentration at 7.25, whilst the combo treatment showed the lowest NH_3_‐N concentration (4.20), followed by BS (5.13) and LF (6.10). Similarly, Wang et al. (2019) and Wang et al. (2017) observed lower NH3‐N concentrations and pH levels in silage made from Italian ryegrass ( Lolium multiflorum Lam.), tall fescue ( Festuca arundinacea Schred.), and oat ( Avena sativa L.) after 30 days, using four Lactobacillus strains derived from straw silage. The highest NH_3_‐N level was in the control, indicating an extensive protein breakdown resulting from the absence of microbial intervention (Bao et al. 2024; Su et al. 2024). The lower NH_3_‐N in the inoculated treatments is likely due to a quicker pH decline, especially in the combo group, which may have experienced synergistic effects that improved acidification and reduced unwanted microbial activity. These results align with the findings of Su et al. (2024), who reported that lactic acid production by L. plantarum and B. subtilis minimises protein hydrolysis and NH_3_‐N accumulation in silage. Although LF‐treated silage had the highest CP content, its NH_3_‐N level was higher than BS and combo, suggesting that these contents are not always directly correlated, and that the proteolysis depends not only on pH but also on microbial dynamics and plant enzyme activities (Jin et al. 2024). Despite having low NH_3_‐N levels and the highest silage pH, the elevated loss in Combo may be attributed to more excessive or inefficient fermentation or enhanced microbial respiration, which promotes greater carbohydrate degradation and nutrient loss (Jatkauskas et al. 2018; Kim et al. 2021). A previous study by Carvalho et al. (2021) noted that the synergistic fermentation of LAB, such as heterofermentative L. fermentum and B. subtilis , results in lactic acid, acetic acid, ethanol, and CO_2_, leading to greater DM loss. In contrast, the enhanced preservation in the LF and BS groups may be attributed to their capacity to promote rapid acidification, reduce unwanted microbial activity, and minimise gas and effluent loss (Gonda et al. 2023). However, insufficient lactic acid production by hetero‐lactic LAB or microbial antagonism in combination with L. fermentum and B. subtilis may increase the pH of silage in high moisture fodders (Figure 3A,B), which favours the growth of undesirable microorganisms (Jung et al. 2024; Sun et al. 2023).

Fermentation characteristics and nutrient conservation of maize silage after 35 days of ensiling with selected Lactobacillus inoculants. (A) Dry matter (DM) in fodder and silage, (B) The pH values of fodder and silage, (C) Crude protein (CP), (D) ammonia‐nitrogen (NH3‐N), and (E) DM loss in silage. Values represent the mean ± standard deviation of each treatment (n = 3). a–c superscript letters in the same column indicate significant differences (p < 0.05; Duncan's test). Control, no microbial inoculants; LF, Maize inoculated with Limosilactobacillus fermentum PQ482012 strain; BS, Maize inoculated with Bacillus subtilis PQ482016 strain; Combo, Maize inoculated with the 1:1 mixer of LF and BS; SEM, Standard error of the mean; DM, Dry matter; CP, Crude protein; NH‐N, ammonia nitrogen.

Microbial Composition of Silage

3.4

Silage fermentation by LAB produces organic acids like lactic acid, which preserve silage by suppressing harmful microbes over time (Ridwan et al. 2023). In this study, Lactobacillus and Bacillus strains were used to evaluate their effectiveness in preserving maize fodder by suppressing fungal growth and improving the microbial quality of silage. The microbial count in maize silage after 35 days of ensiling is depicted in Figure 4. The population of LAB was significantly higher in LF‐treated silage (8.01 ± 0.12 log_10_ CFU/g) than in the control, BS, or combo group (p = 0.036). This highlights the competitive advantage and colonisation efficacy of L. fermentum, which likely contributed to the silage pH and better protein retention observed in the LF group (Kim et al. 2021; Su et al. 2019). Similarly, the results of (Jung et al. 2024) indicated that fermentation with L. brevis improved fermentation profiles, primarily due to the increased LAB count in Alfalfa silage. In contrast, silage inoculated with B. subtilis showed significantly higher Bacillus counts (8.29 ± 0.21 log_10_ CFU/g; p < 0.001) compared to other treatments. This reflects the successful establishment of B. subtilis, which is known for its enzymatic activity, which can degrade fibre and potentially enhance silage quality (Guo et al. 2022). Yeast counts were significantly reduced in the BS and combo group (6.58–6.62 log_10_ CFU/g) compared to the control and LF inoculated silage (p < 0.001), suggesting that B. subtilis may have exerted antifungal effects that helped suppress spoilage yeasts. Mould growth was not detected (ND) in any of the inoculated treatments, whilst it remained detectable (2.88 ± 0.12 log_10_ CFU/g) in the control, indicating that microbial inoculants, especially combinations of LF and BS, can effectively inhibit fungal contaminations (Bao et al. 2024). Consistent with the present findings, Soundharrajan et al. (2023) reported that inoculation with LAB strains, particularly L. plantarum ‐KCC‐34, KCC‐48, P. pentosaceus ‐KCC‐44, KCC‐53, and L. rhamnosus –KCC‐54, either individually or in combination, significantly increased the LAB population in silage whilst simultaneously reducing yeast and mould counts. In this context, previous studies (Bai et al. 2024; Bao et al. 2024) supported the current findings, reporting that Lactobacillus strains were effective in shifting harmful bacterial and fungal populations to LAB during ensiling.

Microbial composition (Log10CFU/g) of maize fodder inoculated with Limosilactobacillus fermentum PQ482012 and Bacillus subtilis PQ482016 strains during ensiling. Values are expressed as mean ± SD (n = 3). a–c superscript letters in the same column indicate significant differences (p < 0.05; Duncan's test). Control, no microbial inoculants; LF, Maize inoculated with L. fermentum PQ482012 strain; BS, Maize inoculated with B. subtilis PQ482016 strain; Combo, Maize inoculated with a 1:1 mixer of LF and BS; SEM, Standard error of the mean; ND, not detected.

Conclusions

4

Silage preservation relies on microbial fermentation, where LAB and Bacillus spp. boost nutrient retention and inhibit spoilage. In this study, 21 LAB isolates from maize and Napier fodder were characterised and identified as L. fermentum and B. subtilis , with sequences in GenBank (PQ482012–PQ482018), showing strong adaptability to silage conditions and a positive influence on maize silage fermentation. Inoculation with L. fermentum enhanced silage acidification and protein preservation, whilst B. subtilis effectively suppressed undesirable bacteria, although their combined application indicated potential microbial interaction affecting dry matter loss. The study was limited to lab‐scale, short‐term fermentation without in vivo animal effects, long‐term stability, or detailed product analysis. Future work should include genome sequencing, metabolomics, and farm trials to better understand the mechanisms of these strains. Their potential as bio‐preservatives makes L. fermentum and B. subtilis promising for sustainable silage and feed safety.

Author Contributions

Sardar Muhammad Amanullah, Md. Moklesur Rahman: conceptualization. Md. Moklesur Rahman, Sardar Muhammad Amanullah, Md. Ahsanul Kabir, Md. Zulfekar Ali, Md. Shamim Ahmed: data curation. Md. Moklesur Rahman, Sardar Muhammad Amanullah: formal analysis. S. M. Jahangir Hossain: project administration. Sardar Muhammad Amanullah: supervision. Sardar Muhammad Amanullah: funding acquisition. Sardar Muhammad Amanullah and S. M. Jahangir Hossain: visualisation. Md. Moklesur Rahman: roles/writing – original draft. Sardar Muhammad Amanullah: writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.