Effects of Lactic Acid Bacteria on Fermentation Quality and Microbiome of Leymus chinensis Silage

TL;DR

This study shows how adding specific bacteria improves the quality of silage made from Leymus chinensis, a type of grass used as animal feed.

Contribution

The novel contribution is identifying how Lactiplantibacillus plantarum and Lentilactobacillus buchneri affect fermentation and microbial communities in Leymus chinensis silage.

Findings

Lactiplantibacillus plantarum and Lentilactobacillus buchneri improved silage fermentation by lowering pH and ammonia-nitrogen levels.

Lactiplantibacillus plantarum showed higher lactic acid content, while Lentilactobacillus buchneri increased acetic acid content.

Microbial diversity decreased during ensiling, with Lactiplantibacillus plantarum causing the most significant reduction in diversity.

Abstract

This study investigated the distinct effects of Lentilactobacillus buchneri (LB) and Lactiplantibacillus plantarum (LP) inoculants on the fermentation characteristics and bacterial community succession of Leymus chinensis silage. Treatments included distilled water (CON), LB, and LP, applied at a concentration of 1 × 106 cfu/g of fresh matter (FM). Compared with the CON group, the fermentation quality was improved by the inoculations, the markedly (p < 0.05) lower pH and NH3–N were found in the LB and LP treatments. The significantly (p < 0.05) highest LA and AA contents were detected in the LP and LB treatments, respectively. The bacterial diversity, reflected by Shannon and Chao1 indices, decreased throughout the ensiling process, with the LP group exhibiting the most pronounced reduction. Furthermore, beta‐diversity analysis revealed distinct microbial community structures among the treatments. While fresh L. chinensis was dominated by Proteobacteria (48.03%), Firmicutes (26.23%), and Actinobacteriota (23.72%), the microbiome shifted dramatically after 60 days of ensiling to be predominantly Firmicutes (94.96%–99.79%), the genus Rhodococcus, Microbacterium, Enterococcus and Leuconostoc, Weissella, and Enterococcus were markedly (p < 0.05) enriched in the CK treatment during fermentation from 3 to 14 days, while from 30 days, the genus Lactobacillus as the dominant genus, especially in the LB and LP treatments. In conclusion, both additives facilitate L. chinensis ensiling by regulating the microbiome, yet L. plantarum demonstrates superior efficacy in optimizing fermentation quality.

Ensiled Leymus chinensis are an important feedstuff for ruminants in large parts of the world. These results suggested that the addition of Lactobacillus plantarum or Lactobacillus buchneri regulated the ensiling performance by shaping the microbial community and metabolic pathways.

Figures

Characterization of the bacterial microbiota in Leymus chinensis silage. Panels (A) and (B) display α‐diversity based on Shannon and Chao1 indices, respectively. (C) Assessment of β‐diversity. Bacterial composition is shown at the phylum level in (D) and genus level in (E). (F) A cladogram illustrating microbial species with significant inter‐treatment differences (default score of 3). Different groups are distinguished by color, with taxonomic classification radiating from the phylum (innermost layer) to the genus (outermost layer).

Differences of microbial taxa in Leymus chinensis silage with different treatments. FM, fresh materials; CK: samples without inoculants; LB: samples inoculated with Lentilactobacillus buchneri; LP: samples inoculated with Lactiplantibacillus plantarum. The linear discriminant analysis (LDA) effect size (LEfSe) analysis of L. chinensis silage bacterialbiomarkers associated with inoculants for different fermentation time (A–F). (A) Fermentation process at Day 3. (B) Fermentation process at Day 7. (C) Fermentation process at Day 14. (D) Fermentation process at Day 30. (E) Fermentation process at Day 60.

Mantel test analysis correlating the top 20 microbial taxa in Leymus chinensis silage with fermentation quality indices. Red and blue lines/cells denote positive and negative correlations, respectively. The size of the squares reflects the magnitude of the correlation coefficients, while color intensity corresponds to statistical significance.

Tables

Table 1. Constituent characteristics of Leymus chinensis prior to ensiling.

Item	Leymus chinensis
Dry matter (%)	54.33 ± 0.02
Crude protein (%DM)	8.21 ± 0.11
Acid detergent fiber (%DM)	38.70 ± 0.28
Neutral detergent fiber (%DM)	64.29 ± 0.60
Water soluble carbohydrate (%DM)	4.95 ± 0.26
Lactic acid bacteria (lg cfu/g FM)	3.79 ± 0.09
Yeast (lg cfu/g FM)	6.29 ± 0.09
Coliform bacteria (lg cfu/g FM)	6.73 ± 0.11
Aerobic bacteria (lg cfu/g FM)	6.88 ± 0.13

Table 2. Fermentation characteristics of ensiled Leymus chinensis.

Item	Day	Treatment			p‐value
		CON	LB	LP	Inoculations	Fermentation	Interactions
pH	3	4.85 ± 0.03aA	4.23 ± 0.01aB	4.29 ± 0.01aB	< 0.0001	0.5130	< 0.0001
	7	4.75 ± 0.01bA	4.17 ± 0.03bC	4.25 ± 0.01bB
	14	4.64 ± 0.04cA	4.17 ± 0.01bB	4.16 ± 0.02cB
	30	4.60 ± 0.01cA	4.13 ± 0.01cB	4.10 ± 0.01dB
	60	4.58 ± 0.01cA	4.11 ± 0.01cB	4.08 ± 0.01dC
LA (g/kg FM)	3	4.26 ± 0.29dC	5.68 ± 0.09eB	6.89 ± 0.17eA	< 0.0001	< 0.0001	< 0.0001
	7	5.71 ± 0.05cC	6.15 ± 0.13 dB	7.17 ± 0.09dA
	14	6.22 ± 0.09bC	6.86 ± 0.11cB	8.28 ± 0.19cA
	30	6.98 ± 0.08aC	8.09 ± 0.11bB	9.62 ± 0.16bA
	60	7.09 ± 0.05aC	8.60 ± 0.07aB	10.54 ± 0.16aA
AA (g/kg FM)	3	2.03 ± 0.06eB	2.46 ± 0.01eA	1.13 ± 0.02eC	0.0480	< 0.0001	< 0.0001
	7	2.44 ± 0.06dBC	3.40 ± 0.16dA	2.57 ± 0.09dAB
	14	2.91 ± 0.02cC	3.64 ± 0.07cA	3.49 ± 0.14cAB
	30	3.56 ± 0.03bC	3.95 ± 0.03bB	4.12 ± 0.09bA
	60	3.99 ± 0.16aC	6.56 ± 0.11aA	4.38 ± 0.09aB
PA (g/kg FM)	3	0.65 ± 0.06dA	0.73 ± 0.06eA	0.60 ± 0.03eA	0.0435	< 0.0001	< 0.0001
	7	0.99 ± 0.08cC	1.39 ± 0.06dA	1.07 ± 0.03dB
	14	1.47 ± 0.07bC	2.25 ± 0.09cA	1.87 ± 0.03cB
	30	1.62 ± 0.10bC	2.73 ± 0.09bA	2.09 ± 0.05bB
	60	2.20 ± 0.20aC	4.00 ± 0.17aA	2.86 ± 0.05aB
NH3‐N (g/kg FM)	3	1.62 ± 0.16dA	1.43 ± 0.02dB	1.44 ± 0.01eB	< 0.0001	< 0.0001	< 0.0001
	7	2.47 ± 0.04cA	1.45 ± 0.01dB	1.50 ± 0.01dB
	14	2.90 ± 0.07abA	1.73 ± 0.03cB	1.70 ± 0.07cB
	30	2.87 ± 0.07abA	2.12 ± 0.08bB	1.97 ± 0.01bC
	60	2.92 ± 0.01aA	2.44 ± 0.03aB	2.09 ± 0.06aC

Table 3. Bacterial alpha diversity in fresh and ensiled Leymus chinensis.

Ensiling days	Treatment	Sequences number	ASVs	Good's coverage
3 day	FM	69738	1501	> 0.9999
	CON	68617	671	> 0.9999
	LB	46524	146	> 0.9999
	LP	43770	366	> 0.9999
7day	CON	62973	242	> 0.9999
	LB	51903	356	> 0.9999
	LP	44121	98	> 0.9999
14 day	CON	61711	192	> 0.9999
	LB	49113	233	> 0.9999
	LP	45338	192	> 0.9999
30 day	CON	53903	351	> 0.9999
	LB	59511	122	> 0.9999
	LP	46640	225	> 0.9999
60 day	CON	66168	425	> 0.9999
	LB	57349	100	> 0.9999
	LP	47474	405	> 0.9999

Full text

Introduction

1

Socioeconomic development has exacerbated the imbalance between forage supply and livestock demand (Bilotto et al. 2021; Liu et al. 2025). China has become the largest consumer of the feedstuff (Wang et al. 2024). Strategic approaches focused on maximizing forage and grass efficiency are required to satisfy the surging requirements of the animal husbandry sector (Boval and Dixon 2012; Ren et al. 2024).

Leymus chinensis (L. chinensis) (Trin.) Tzvel., is one of the Gramineae perennial grasses, as a major forage resource for ruminants due to the excellent palatability (Du et al. 2020). Nevertheless, grazing and haymaking systems alone are often inadequate to secure a continuous, year‐round availability of high‐quality L. chinensis for ruminant livestock (Li et al. 2022). Widely acknowledged as an effective approach for conserving forage nutrients, ensiling functions through lactic acid bacteria (LAB)‐mediated anaerobic fermentation that lowers the pH to arrest mold and spoilage (Dong et al. 2020; Sukphun et al. 2024). However, the fermentation quality of silage is often constrained by the intrinsic chemical profile and the epiphytic microbial community of the fresh forage (Nascimento Agarussi et al. 2019; You et al. 2025). Poor fermentation quality in L. chinensis silage, defined by elevated pH levels and inadequate lactic acid (LA) concentrations, is frequently attributed to the low abundance of epiphytic LAB on the fresh forage (Zhang et al. 2016). Numerous LAB inoculations have been developed to improve the fermentation quality and preservation efficiency.

Prior studies have classified LAB inoculants into homofermentative and heterofermentative types based on their fermentation end products (Muck et al. 2018; Xiao et al. 2023). The homofermentative LAB can degrade glucose, pentoses, and xylose through Emden–Meyerhoff pathway, phosphoketolase pathway, and pentosephosphate pathway, and heterofermentative LAB degrade glucose, pentoses, and xylose through pentosephosphate pathway to produce the various end‐products (Du et al. 2026). The homofermentative and heterofermentative LAB strains, including Lactiplantibacillus plantarum, Lactiplantibacillus pentosus, and Lentilactobacillus buchneri, are widely used by rapidly decreasing the pH, suppressing the undesirable microorganisms, improving the aerobic stability and preserving nutritional compositions (You et al. 2025). Global research has extensively examined how LAB inoculants affect the bacterial communities of crops such as alfalfa, corn, barley, and wheat (Gharechahi et al. 2017; Keshri et al. 2019; Liu et al. 2019; Yang et al. 2019); however, information regarding the bacterial community dynamics of L. chinensis treated with LAB inoculants throughout the entire fermentation period is still scarce.

Based on the current understanding of the homofermentative‐ and heterofermentative‐ LAB characteristics and L. chinensis silage, the author hypothesized that the dynamic changes in L. chinensis silage occur in correlation with longitudinal alterations in microbial community structure during the fermentation process. Therefore, the present study investigated the ensiling properties and microbial community succession in the L. chinensis silage prepared without or with LAB using the 16S rRNA amplicon sequencing.

Methods

2

Materials and Ensilage Preparation

2.1

The L. chinensis was sampled and collected during the heading stage on September 3, 2023 in Huhhot, China. The fresh L. chinensis had a dry matter (DM) content of 54.33%, and its chemical composition is detailed in Table 1. The raw material was partitioned into three distinct treatments after being shredded into segments of roughly 2 cm by a manual forage chopper (Xianglong Co. Ltd., Linyi, China). The L. chinensis were treated separately using different methods, including distilled water (control, CON); L. buchneri (LB, heterofermentative LAB inoculation) and L. plantarum (LP, homofermentative LAB inoculation). Both inoculants (obtained from Jiangsu Lvke Biotechnology Company, Gaoyou, China) were applied at 1 × 10^6^ cfu/g of fresh weight. A total of 300 g of the chopped L. chinensis per replicate, treated with or without inoculants, was kept at room temperature (22°C–25°C) after being vacuum‐sealed in 40 cm × 28 cm polyethylene bags.

Nutritional Composition and Fermentation Product Analyses

2.2

Sampling was conducted at Days 3, 7, 14, 30, and 60 to evaluate fermentation quality and microbial dynamics. To determine dry matter (DM), fresh L. chinensis was oven‐dried at 65°C for 72 h. For the subsequent nutritional evaluation, the dried material was pulverized through a 1‐mm screen using a Taisite FW100 mill (Taisite Instrument Co. Ltd., Tianjin, China); crude protein (CP) content was then determined following the AOAC standard procedures (Rodrigues et al. 2010). Neutral detergent fiber (NDF) and acid detergent fiber (ADF) concentrations were determined to evaluate the fiber profiles, following the analytical procedures established by the previous method (Van Soest et al. 1991). The concentration of water‐soluble carbohydrates (WSC) was quantified via the anthrone‐sulfuric acid colorimetric assay (Maharjan et al. 2018).

The chopped fresh L. chinensis (10 g) were sampled and homogenized with deionized water (90 mL) in a glassware graduated cylinder. The resulting extract was filtered through layers of sterile cheesecloth, previously rinsed cheesecloth to ensure clarity. Following calibration of the probe, the pH of the L. chinensis silage was ascertained from the filtrate via a portable pH meter. The filtrate was further subjected to high‐performance liquid chromatography (HPLC) to quantify the levels of organic acids (Shui and Leong 2002). The ammonia‐N (NH_3_‐N) content was determined according to the previously published the phenol‐hypochlorite method (Kleinschmit et al. 2005).

Microbiological Analysis

2.3

The genomic DNA of both fresh and ensiled L. chinensis was extracted following the cetyltrimethylammonium bromide (CTAB) protocol. Subsequently, the concentration and purity of the isolated DNA were verified with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA). High‐throughput sequencing was conducted on the V3–V4 hypervariable segments of the 16S rRNA gene, utilizing an Illumina NovaSeq. 6000 system (Illumina Inc., San Diego, CA, USA). The processing of raw sequencing data was facilitated by the Qiime2 pipeline, where the DADA2 plugin was implemented to filter out low‐quality reads and generate amplicon sequence variants (ASVs) (Callahan et al. 2016). Next, the ASVs were annotated for taxonomy according to the SILVA database (Schloss 2020). To visualize bacterial community shifts, Bray–Curtis dissimilarity matrices were employed to perform Principal Coordinate Analysis (PCoA) within the R environment (v3.3.1), enabling the visualization of microbial community clustering with confidence intervals represented by ellipses. Furthermore, treatment‐level differences were statistically validated via PERMANOVA (R v2.5.4), while T‐tests were used to discern significant disparities (p < 0.05) in the microbiome of L. chinensis silage (Omontese et al. 2022). The R package (version 1.8.4) was used to calculate Mantel tests, and the network of correlation coefficient was generated using an online platform (http://www.omicsmart.com).

Statistical Analysis

2.4

A General Linear Model (GLM) was implemented via SAS 9.0 software (version 9.0, SAS Institute Inc., Cary, NC) to assess how bacterial inoculants and the length of fermentation influenced the results. The statistical model used was *Y_ij_ * = μ + *α_i_ * + *β_j_ * + (α × β)_ ij _ + *ε_ij_ *, where *Y_ij_

• is the dependent variable, μ is the overall mean, *α_i_
• is the inoculant effect, *β_j_
• is the time effect, (α × β)_ ij _ is the interaction between inoculant and time, and *ε_ij_
• represents the residual error. Statistical significance was pre‐defined at a threshold of p < 0.05.

Results

3

Characteristics of the L. chinensis Prior to Ensiling

3.1

The nutritional and microbial compositions of L. chinensis are shown in Table 1. The contents of WSC, CP, ADF, and NDF of L. chinensis were 4.95%, 8.21%, 38.70%, and 64.29% of DM, respectively. The L. chinensis contained low desirable LAB (3.79 log cfu/g of fresh weight) and high coliform bacteria (6.73 log cfu/g of fresh weight) counts.

Fermentation Quality of L. chinensis Inoculated Without or With L. buchneri or L. plantarum

3.2

The fermentation attributes of L. chinensis silage (Table 2) were significantly altered by both the treatment groups and fermentation timeframe (p < 0.05). Notably, a significant interplay between the additives and ensiling duration was detected for all chemical indicators, including pH, NH_3_‐N, and individual organic acids. As fermentation progressed, pH levels declined across all groups; however, the CON treatment maintained a higher pH compared to the inoculated groups, with the LP treatment achieving the lowest final pH. This acidification was driven by the continuous accumulation of organic acids, particularly LA and acetic acid (AA). By day 60, the LP and LB treatments exhibited the highest concentrations of LA and AA, respectively. Propionic acid (PA) content similarly increased over the 60‐day period, peaking in the LB group, whereas butyric acid‌ (BA) remained below detectable limits. Consistent with organic acid trends, NH_3_‐N levels rose as ensiling advanced, with the highest concentration recorded in the CON treatment at the end of the experiment.

Parameters of Bacterial Community of L. chinensis Inoculated Without or With L. buchneri or L. plantarum

3.3

A total of 1,831,365 refined reads were generated through 16S rRNA gene profiling of 33 specimens, encompassing both raw and ensiled L. chinensis. As detailed in Table 3, the sequence depth for individual samples fluctuated between 40,249 and 76,256. The Shannon and Chao1 indexes for these samples were displayed in Figure 1A,B. Microbial alpha‐diversity diminished markedly throughout the duration of fermentation, with the LP group demonstrating the most significant loss in bacterial complexity. The β‐diversity analysis, visualized via PCoA, revealed a sharp divergence among the various treatments, explaining 71.31% of the microbial variations (Figure 1C). The PERMANOVA results (R = 0.5312, p = 0.001) further confirmed the temporal succession of the bacterial populations during fermentation. Proteobacteria, Firmicutes, and Actinobacteriota were identified as the core phyla in raw L. chinensis, representing an aggregate abundance of more than 97% (Figure 1D), and the major bacterial were Lactobacillu (24.16%), Sphingomonas (16.95%), Methylobacterium‐Methylorubrum (8.16%), and unclassified_f_Microbacteriaceae (7.09%) at genus level (Figure 1E). Nevertheless, the bacterial compositions were shifted from Proteobacteria (48.03%), Firmicutes (26.23%) and Actinobacteriota (23.72%) in the fresh L. chinensis materials to Firmicutes (94.96%–99.79%) after ensiling 60 days (Figure 1E). The bacterial community of L. chinensis silage exhibited a consistent pattern of succession throughout fermentation. By Day 60, a sharp increase in the relative abundance of Lactobacillus was observed, coinciding with a decline in Sphingomonas and Methylobacterium‐Methylorubrum (Figure 1E). Notably, Lactobacillus dominated the microbial community from Day 3 to Day 60, particularly in the LB and LP treatments. To identify specific taxa associated with different treatments and times, LEfSe analysis was employed. As illustrated in Figure 1F, the fresh forage contained the highest number of discriminative biomarkers. However, by Day 3 of ensiling, undesirable taxa such as Enterobacterales and Gammaproteobacteria were significantly depressed in the LB and LP groups. To future examine the impact of the fermentation process on bacterial taxa in L. chinensis silages treated with or without inoculants, a LEfSe analysis at various fermentation period was performed, as shown in Figure 2A–E. In the fermented L. chinensis silage for 3 days, Rhodococcus, Microbacterium, Enterococcus, and Leuconostoc were enriched in the CK treatment. After 7 days and 14 days of fermentation process, the genus Weissella and Enterococcus were enriched in the CK treatments, and the genus Lactobacillus were concentrated in the LB and LP treatments. Ensiling for 30 days and 60 days, the genus Lactobacillus were also enriched in the LB and LP treatments.

Characterization of the bacterial microbiota in Leymus chinensis silage. Panels (A) and (B) display α‐diversity based on Shannon and Chao1 indices, respectively. (C) Assessment of β‐diversity. Bacterial composition is shown at the phylum level in (D) and genus level in (E). (F) A cladogram illustrating microbial species with significant inter‐treatment differences (default score of 3). Different groups are distinguished by color, with taxonomic classification radiating from the phylum (innermost layer) to the genus (outermost layer).

Differences of microbial taxa in Leymus chinensis silage with different treatments. FM, fresh materials; CK: samples without inoculants; LB: samples inoculated with Lentilactobacillus buchneri; LP: samples inoculated with Lactiplantibacillus plantarum. The linear discriminant analysis (LDA) effect size (LEfSe) analysis of L. chinensis silage bacterialbiomarkers associated with inoculants for different fermentation time (A–F). (A) Fermentation process at Day 3. (B) Fermentation process at Day 7. (C) Fermentation process at Day 14. (D) Fermentation process at Day 30. (E) Fermentation process at Day 60.

Correlations Between the Dominant Bacterial Community and Fermentation Characteristics of L. chinensis Silage Treated With L. buchneri or L. plantarum Based on Mantel Tests

3.4

Mantel tests were utilized to assess the association between the fermentation characteristics of L. chinensis silage and the microbial community structure of the plastisphere (Figure 3). The g_Lactobacillus was significantly correlated with all of the parameters of the fermentation quality, including pH (r = 0.470, p < 0.01), LA (r = 0.465, p < 0.01), PA (r = 0.204, p < 0.01) and AA (r = 0.188, p < 0.05). Several bacterial genera, including Enterococcus, Methylobacterium‐Methylorubrum, Microbacterium, and Nocardioides, showed significant correlations with pH (r = 0.411, 0.233, 0.273, and 0.273, respectively, p < 0.01) and LA (r = 0.370, 0.412, 0.316, and 0.457, respectively, p < 0.01).

Mantel test analysis correlating the top 20 microbial taxa in Leymus chinensis silage with fermentation quality indices. Red and blue lines/cells denote positive and negative correlations, respectively. The size of the squares reflects the magnitude of the correlation coefficients, while color intensity corresponds to statistical significance.

Discussion

4

Ensiling plays an undeniable role in preserving the nutritional compositions and extending the availability of forage and grass for ruminants. Nevertheless, the alternations of L. chinensis ensiled without or with L. buchneri or L. plantarum still remains limited. To investigate the dynamic changes in L. chinensis silage, we tracked its fermentation parameters and microbial community (via 16S rRNA sequencing) over a 60‐day period, comparing treatments with L. buchneri, L. plantarum, and an untreated control.

The successful preservation of L. chinensis silage was evidenced by a final pH below 4.70 across all treatment groups after the 60‐day fermentation period (Cui et al. 2020). Compared with the CON group, the application of microbial inoculants significantly altered the accumulation of LA, AA, and PA throughout the fermentation period. This finding aligns with previous studies indicating that additives exert a positive effect on organic acid production (Yan et al. 2022). Higher levels of LA and AA were found as a result of the LP and LB treatments, respectively, indicating that the fermentation was improved by the addition of microbial inoculants (Xiao et al. 2023). The significant disparity in LA and AA concentrations between the LP and LB treatments can be attributed to the distinct metabolic routes—namely the Embden–Meyerhof, phosphoketolase, and pentose phosphate pathways—utilized by L. plantarum and L. buchneri (Burgé et al. 2015; Du et al. 2026), resulting in the lower pH in the LP treatment and higher AA content in the LB treatment, which is in agreement with the previous reports (Ni et al. 2017; Xiao et al. 2023). The distinct metabolic pathways between homofermentative‐ and heterofermentative‐ LAB strains account for the variations observed in organic acid profiles (Stokes and Chen 1994; You et al. 2025). The BA concentrations were associated with the undesirable microorganisms and was lower than the detectable level in the current study, indicating that the undesirable microorganisms were depressed by the acidic environmental (EFSA Panel on Biological Hazards 2011; Yan et al. 2022). The higher NH_3_‐N content in this study indicated that the protein degradation occurred during ensiling as the plant enzymes and microbial metabolism. The NH_3_‐N content enriched in the CON group followed by the LB and LP groups, which could be explained by the CON treatment obtained the higher abundance of the genus Enterobacteriaceae and some amino acids could break down by the decarboxylation and deamination activities by the Enterobacteriaceae (Yuan et al. 2020). Conversely the CON treatment was hard to produce an acid environment because the NH_3_‐N content has a buffering capacity to prevents the pH drop (Yan et al. 2022).

Changes in the silage bacterial community were tracked via 16S rRNA sequencing across the ensiling process of L. chinensis. The metabolic pathways and ensiling performance of L. chinensis silage can be influenced by the microbial community dynamics. The Goods's coverage of all samples was higher than 0.99, indicating that the sequencing depth could reflect the bacterial community (Ren et al. 2024). Compared to the fresh L. chinensis, all treatments showed a consistent decline in microbial diversity and richness, pointing to a suppressive effect of the ensiling conditions on microbial communities (Yan et al. 2025). This decline in diversity and richness in the LB and LP groups likely resulted from the rapid domination of inoculated LAB and the consequent suppression of competing microbes (Yan et al. 2022; Xu et al. 2024). As depicted in Figure 1C, PCoA revealed distinct clustering between the fresh material and the three treatment groups, although no clear separation was observed with prolonged fermentation. This indicates that the ensiling process significantly altered the bacterial community structure, which is similar with the previous report (Xiao et al. 2023; You et al. 2025). Notably, the dominant phylum shifted from Proteobacteria in fresh forage to Firmicutes across all treatments, consistent with previous findings that Firmicutes are well‐adapted to anaerobic and acidic conditions (Cui et al. 2023). Previous studies indicated that the Proteobacteria is comprised of the genera Pantoea, Pseudomonas, Sphingomonas and Erwinia in silage (Ogunade et al. 2018; Romero et al. 2017). In the current study, the abundances of the Sphingomonas, Methylobacterium‐Methylorubrum and unclassified_o_Enterobacterales were the main microorganisms in the fresh materials and were markedly decreased after the ensiling progress as these microbes tend to be depressed in the acidic (pH < 5.40) conditions (Yang et al. 2025). The differences of the main microorganism could be contributed to the fresh materials. Early in the ensiling process (3 days), Lactobacillus emerged as the dominant genus across all groups, accounting for the overwhelming majority (80%) of the microbial population. The concentration of Lactobacillus increased as fermentation progressed, largely due to its high acid tolerance, which allowed it to thrive as pH levels declined (Wang et al. 2018).

Fermentation quality is governed by biochemical processes driven by the silage microbiota. Network analysis has emerged as a popular method for characterizing the interactions between specific nodes and variables (Barberán et al. 2012), and analyzing the relationships between the microbial community and ensiling performance could help us to better understand the responsibility of the key bacteria for improving the ensiling performance (Fang et al. 2022). The g_Lactobacilluswas significantly correlated with all of the parameters of the fermentation quality, which is in agreement with the previous report (Fang et al. 2022; Du et al. 2023; You et al. 2025), resulting in increasing the organic acid levels and decreasing the pH value in L. chinensis silage. Enterococcus is a cocci LAB genus. The g_Enterococcus was positively correlated with pH value, which could be contributed to the genus only growth in a pH > 4.5 environment (McGarvey et al. 2013; Xu et al. 2021). Owing to their complex cellular fatty acid profiles, Nocardioides are extensively applied in antibiotic synthesis and the bioremediation of pollutants, might result in the increase in the LA concentrations (Wang et al. 2024).

Conclusions

5

Conclusively, this research elucidated the impact of L. buchneri or L. plantarum on L. chinensis. This work demonstrates the significant impact of LAB inoculations regulation and supports the application of LAB additives for high‐quality silage production. Results indicated that both LAB strains ameliorated fermentation performance; however, L. plantarum exhibited superior efficacy in lowering the pH and NH_3_‐N levels and increasing LA concentration. The fermentation quality of the L. chinensis silage was dominated by the genus Lactobacillus, which could provide evidence for the importance of LAB in the L. chinensis silage preservation.

Author Contributions

Xiaowei Jiang: investigation, writing – original draft, visualization, data curation. Lichao He: investigation. Zhaorui Han: investigation. Sen Zong: investigation. Shuai Du: conceptualization. Hongxin Wu: supervision, funding acquisition. Yanzi Xiao: supervision, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.