Bacterial Diversity, Chemical Composition, and Fermentation Quality of Alfalfa-Based Total Mixed Ration Silage Inoculated with Lactobacillus reuteri and Lentilactobacillus buchneri

Abstract

This study examined the effects of lactic acid bacteria (LAB) from extremely cold environments on the bacterial diversity, chemical composition, and fermentation quality of alfalfa-based TMR silage. The TMR was treated with Lactobacillus reuteri A4-2 (Lr A4-2) and Lentilactobacillus buchneri 9-2 (Lb 9-2) at an application rate of 1.0 × 10⁵ cfu/g fresh material, respectively, and the control received the same volume of distilled water. The TMR was ensiled for 7, 15, 30, 60, and 210 days. The Lr A4-2 treatment produced higher lactic acid (62.53 g/kg DM at 210 days) and maintained a diverse bacterial community throughout the ensiling compared with the control and Lb 9-2 treatment. The Lb 9-2 treatment increased acetic acid (51.42 g/kg DM at 210 days) and formed a distinct bacterial community profile. The 16S rRNA sequencing revealed a shift from initial Weissella dominance to Lactobacillus-dominated communities across the treatments by day 210. Alpha diversity indices decreased over time, with Lr A4-2 treatment maintaining higher diversity. Principal coordinate analysis demonstrated significant temporal shifts in bacterial community composition among treatments (p < 0.01). The results suggest that different heterofermentative LAB strains modulated the fermentation and microbial balance in alfalfa-based TMR silage in different ways.

1. Introduction

The total mixed ration (TMR) system has become increasingly popular in livestock management, as it offers a well-balanced nutrient supply that fulfills the dietary requirements of ruminants [1]. However, TMR generates heat during the preparation and feeding due to the high moisture content, affecting its nutritional value [2]. Ensiling is widely adopted and effective for preserving green feed worldwide [3]. Like ordinary silage, when TMR is ensiled, it facilitates the extended storage of the thoroughly mixed feed [4]. The ensiling process promotes anaerobic fermentation, primarily through the action of lactic acid bacteria (LAB), which convert water-soluble carbohydrates (WSC) into lactic acid (LA). In addition, this conversion lowers the pH and inhibits the proliferation of harmful microorganisms during the ensiling process [5].

Alfalfa (Medicago sativa L.) is a favored perennial legume forage among livestock producers due to its high nutritional value, particularly its elevated levels of crude protein (CP) [6,7,8]. Alfalfa is widely used in ruminant production, especially in regions with suitable climatic conditions, such as temperate zones with adequate moisture and sunlight [9]. For this study, alfalfa was harvested in Dingxi, Gansu Province, in northern China. This region features a temperate continental climate, marked by cold winters and warm summers. The second and third cuts of alfalfa are typically harvested in July and August, marked by unpredictable weather conditions that complicate alfalfa processing due to the rainy season. As a result, ensiling is the preferred method for preserving alfalfa during this time. However, achieving satisfactory fermentation quality when ensiling alfalfa can be challenging due to its high buffering capacity and low dry matter (DM) and WSC levels [10]. Wilting and adding additives often enhance fermentation quality and optimize microbial populations in ensiled alfalfa [11,12]. Alternatively, incorporating fresh alfalfa into TMR is a practical way of making high-quality silage.

In addition to alfalfa, other key components of TMR include corn silage and concentrates, which provide essential carbohydrates and energy sources, respectively. Corn silage, with its higher DM and WSC content, complements alfalfa by improving the overall fermentation quality of the TMR [13]. Concentrates, such as grains and protein supplements, ensure that the TMR meets the nutritional requirements of ruminants by providing a balanced mix of nutrients [14]. Considering the global scarcity of grain feed supplies and the corresponding rise in feed prices, strategically incorporating food by-products, such as apple pomace, as livestock feed presents a viable solution to these challenges [15]. Many studies have demonstrated that apple pomace can be used to produce aromatic compounds, dietary fiber, citric acid, pectin, and seasoning and can be used as animal feed [16,17]. However, apple pomace has a high moisture content and significant sugar concentration. If the residual by-product is not processed promptly, it may pose considerable public health and environmental risks, including spoilage, pest infestations, effluent discharge, and generation of unpleasant odors [16,18]. While dried apple pomace facilitates easier transportation and storage, drying is energy-intensive and often dependent on fossil fuels, increasing feed costs and reducing practicality. As a result, fermented TMR emerges as a beneficial preparation method because it can effectively use high-moisture food by-products [15].

The microbial composition, particularly the populations of LAB, plays a vital role in the fermentation quality of silage or TMR silage [19]. Advanced DNA sequencing methods have revolutionized our understanding of the diverse microbial communities in different silages and their impact on fermentation quality [20]. Previous studies have indicated that the inoculation of LAB additives during the ensiling [21,22,23] optimizes the bacterial community and enhances the fermentation quality of the final silage [12,23,24,25]. However, how LAB additives affect the fermentation of TMR composed of various raw materials is currently unclear. Therefore, this study aimed to investigate the fermentation characteristics, chemical characteristics, and alterations in the bacterial community with progression in ensiling alfalfa-based TMR inoculated with two different LAB strains.

2. Materials and Methods

2.1. TMR Silage Preparation

To prepare the TMR, alfalfa (Medicago sativa L.) was harvested approximately 10 cm above the ground using a mechanical harvester during its early flowering stage. The alfalfa was randomly selected from a well-established field in Dingxi, Gansu Province, China. According to NASEM [14], the nutritional requirements of basal TMR prepared for 600–800 kg beef cattle fattening are presented in

Table 1. The formula and chemical composition of TMR before ensiling.

Items	Dietary Treatment 2
	Lr A4-2 TMR	Lb 9-2 TMR	Control TMR
Ingredients (% DM)
Whole-crop corn silage	17	17	17
Apple pomace	19	19	19
Fresh alfalfa	17	17	17
Concentrate Corn grain (50%) Soybean meal (10%) Cotton seed meal (14.5%) Wheat bran (14.5) Flax oil residue (4%) Calcium carbonate (2%) 1 Premix (5%)	47	47	47
Total	100	100	100
3 Chemical composition (% DM)
DM	45.8	45.8	45.8
aNDF	26.7	26.7	26.7
ADF	18.5	18.5	18.5
CP	16.7	16.7	16.7
Ca	1.17	1.17	1.17
P	0.71	0.71	0.71

¹ Premix provided (per kg of premix): Vitamin A, 625,000 IU; Vitamin D3, 200,000 IU; Vitamin E, 1250 IU; Cu, 625 mg; Fe, 9500 mg; Zn, 3700 mg; Mn, 3700 mg; I, 250 mg; Se, 12.5 mg; Co, 50 mg. ² Lr A4-2 TMR, total mixed ration inoculated with L. reuteri A4-2 TMR; Lb 9-2 TMR, total mixed ration inoculated with L. buchneri TMR, control, total mixed ration without inoculation. ³ Chemical composition of the TMR before inoculation and ensiling. DM, dry matter; aNDF, neutral detergent fiber, assayed with a heat-stable amylase and expressed. Including residual ash; ADF, acid detergent fiber; CP, crude protein.

. Gansu Zhongtian Sheep Industry Co., Ltd. (Dingxi, China), provided the dried apple pomace used in the experiment. The corn silage had been ensiled in a semi-underground silo for one year before the experiment started and the pH fell below 4.0.

All roughage components were chopped into small pieces using a mechanical forage cutter (Toyohira Agricultural Machinery, Sapporo, Japan) and mixed thoroughly using a horizontal mixer to ensure uniform distribution of each component of TMR. Then, the mixed feed was divided into 15 piles and treated with (50 mL) distilled water as a control, Lactobacillus reuteri A4-2 (Lr A4-2), and Lentilactobacillus buchneri 9-2 (Lb 9-2), respectively, with each treatment replicated 5 times. The inoculants were applied in a water solution before bailing, using handheld sprayers, to achieve a standard of 1 × 10⁵ cfu/g of fresh weight, as applied in current practice. Each treated pile was then baled using a bale wrapper machine with a front-conveyor (RX-DK5252C, Zhengzhou Muchang Agricultural Machinery Manufacturing Co., Ltd., Zhengzhou, China) and wrapped with 4 to 6 layers of stretch film (MY-100, green forage stretches film, Zhengzhou, China; 250 mm wide and 25 mm thick) to form cylindrical bales that were 0.6 m tall × 0.6 m diameter, which served as experimental silos. The silos were subsequently stored and allowed to ensile for the designated periods of 7, 15, 30, 60, and 210 days; then, sampling was conducted from one bale per treatment at each ensiling duration. However, samples were collected in triplicate per treatment per ensiling duration from each bale using an electric sampler (HN35, TianJin HaoNiu Bio-Tech Co., Ltd., Tianjin, China). Additionally, approximately 500 g of the TMR sample was preserved as a fresh sample at −20 °C for further analysis.

2.2. Fermentation and Chemical Analysis of TMR

The TMR silage samples from each treatment were divided into three separate lots (500 g each) for each day of ensiling. Twenty grams of samples from the first lot were homogenized with 180 mL of distilled water using a juice extractor for 30 s, followed by filtration through four layers of medical gauze. The pH of the resulting filtrates was immediately measured using a glass electrode pH meter (Orion Star™ A111, Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, the filtrates were acidified to a pH of 1–2 using 7.14 mol/L H₂SO₄ and filtered through a 0.22 μm dialyzer. These filtrates were used to quantify the concentrations of lactic acid (LA), acetic acid (AA), and propionic acid (PA) following the methodology outlined in our previously published work [26]. The remaining filtrate was centrifugated (12,000× g for 15 min at 4 °C) and subsequently frozen at −20 °C to determine ammonia nitrogen (NH₃-N) and WSC. The concentration of NH₃-N in the filters was determined using the phenol-sodium hypochlorite method following the procedure of Broderick and Kang [27]. The concentration of WSC in the samples was determined using a colorimetric method following a complete reaction with anthrone reagents [28].

For the second lot of samples, the DM content was assessed using the AOAC method 943.01 [29]. The dried samples were ground through a 1 mm sieve and stored in plastic sample bags for subsequent analysis of the fiber fractions, including amylase-treated neutral detergent fiber (aNDF), acid detergent fiber (ADF), and CP. The concentrations of aNDF and ADF were determined through digestion in neutral and acid detergent solutions, respectively, utilizing the batch procedures specified for an ANKOM 2000 Fiber Analyzer (ANKOM Technology Corporation, Fairport, NY, USA). During the aNDF analysis, heat-stable alpha-amylase and sodium sulfite were added to mitigate the effects of starch and protein, respectively, while the aNDF content included the residual ash. Nitrogen was quantified using a Kjeldahl automated apparatus (K9805, Shanghai Analytical Instrument Co., Ltd., Shanghai, China), and CP was derived by multiplying the Kjeldahl nitrogen by 6.25 [30].

2.3. DNA Extraction, PCR Amplification, and Sequencing

Samples (third lot) of TMR ensiled for 7, 15, 30, 60, and 210 days were collected for microbial community analysis, with triplicates selected for each treatment group. Briefly, a 15 g forage sample was homogenized with 70 mL of sterile 0.85% sodium chloride solution and agitated at 120 rpm for 20 min. The resulting liquid was then filtered through four layers of medical-grade cheesecloth and centrifuged at 10,000 rpm for 6 min to isolate microbial cells. The extraction of microbial DNA from the forage samples was performed according to the manufacturer’s instructions for the DNA isolation kit (MP Biomedicals, Solon, OH, USA). The concentration and purity of the extracted DNA were assessed using established methods [31]. PCR amplification, sequencing of amplicons, and the subsequent data analysis were conducted by Shanghai Majorbio Bio-Pharm Technology Co., Ltd. The specific primers 27F (5′-AGRGTTYGATYMTGGCTCAG-3′) and 1492R (5′-RGYTACCTTGTTACGACTT-3′) were employed to amplify the bacterial 16S rRNA gene.

The PCR amplification and purification of the PCR products were conducted following previously reported methodologies [32]. Circular consensus reads were generated using the PacBio Sequel II System (Pacific Biosciences, Menlo Park, CA, USA) by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). The extraction, filtration, and optimization of sequences were performed as previously described [33]. The optimized reads were classified into operational taxonomic units (OTUs) using UPARSE 7.1, applying a sequence identity threshold of 97%. The representative sequences were then annotated through a QIIME-based wrapper of the RDP-classifier v.2.2, utilizing the SILVA bacterial 16S rRNA database with a confidence cutoff of 0.8 [32]. Alpha diversity, as measured by the Shannon index, was estimated by using QIIME software (http://qiime.org/scripts/assign_taxonomy.html). Principal coordinates analysis (PCoA) was employed to visually represent the diversity of microbial community structures across different groups. Additionally, Spearman correlation analysis (SCA) was conducted to investigate the relationships between environmental factors and the predominant microbiota of TMR, utilizing the Majorbio online platform. Detrended correspondence was performed using R package 4 (version 3.2.5) before redundancy analysis (RDA).

2.4. Statistical Analysis

The chemical composition of TMR before ensiling was reported as a mean (N = 3). The analysis of TMR silage fermentation parameters, chemical composition, and alpha diversity indices during the ensiling process was conducted using the general linear model from the Statistical Package for Social Science (SPSS 21.0, SPSS, Inc., Chicago, IL, USA), following 3 × 5 factorial treatments design that included three inoculation treatments and five ensiling days. The model is expressed as follows:

Y_ijk = µ + T_i + D_j + (T × D) _ij + e_ijk

(1)

In Equation (1), Y_ijk denotes the observation; μ represents the overall mean; T_i indicates the effect of the three treatments (i = 1, 2, 3); D_j signifies the impact of the five ensiling days (j = 1, 2, 3, 4, 5); (T × D)_ij reflects the interaction between treatments and ensiling days; and e_ijk accounts for the random residual error. A significance level of p ≤ 0.05 was established. The mean of the fermentation characteristics, chemical composition, and alpha diversity indices (ACE, Chao1, Shannon, Coverage, and Simpson) for the treatment within the same ensiling duration were separated using Tukey’s multiple comparisons when the interaction was significant (p < 0.05). The relative abundances of microbial communities at both the genus and species levels were analyzed using two-way ANOVA in SPSS version 21.0. Additionally, R studio (Version 1.0.136) was employed to compute Spearman correlation coefficients and to create heatmaps to examine the correlations between the relative abundances of bacterial species and each fermentation and chemical parameter, with a significance threshold set at p ≤ 0.05.

3. Results

3.1. Fermentation Profile of TMR Silage Inoculated with Different Lactic Acid Bacteria Strains

The fermentation characteristics of TMR silage over different ensiling durations and inoculation effects are presented in

Table 2. Fermentation characteristics of TMR silage on different ensiling days.

Item	pH	LA	AA	PA	WSC	NH3-N
		g/kg DM
Ensiling duration
7 d	4.360 ABC	32.44 C	16.99 C	4.336 A	34.66 A	14.48 B
15 d	4.313 C	37.09 BC	20.62 C	6.193 AB	27.96 A	18.14 B
30 d	4.320 ABC	43.88 ABC	26.43 BC	8.590 ABC	25.43 A	21.92 B
60 d	4.380 B	51.1 AB	35.73 AB	14.29 BC	21.06 A	25.86 A
210 d	4.513 A	53.73 A	44.70 A	15.17 C	16.98 A	26.95 A
Inoculation
Control	4.370 a	42.03 a	25.21 a	4.860 a	30.46 a	18.11 a
Lr A4-2	4.382 a	49.08 a	28.15 a	3.080 a	25.38 b	14.60 b
Lb 9-2	4.380 a	39.64 a	33.32 a	5.070 a	38.76 a	12.57 c
SEM 1	0.034	1.425	1.196	0.260	2.420	0.425
p-value
T	<0.001	<0.001	<0.001	<0.001	0.001	0.001
D	0.856	<0.001	<0.001	<0.001	0.001	0.001
T × D	0.870	<0.001	<0.001	<0.001	0.001	0.001

LA, Lactic acid; AA, acetic acid; PA, propionic acid; WSC, water-soluble carbohydrate; NH₃N, ammonia nitrogen; T, treatment; D, ensiling days; T × D, interaction between treatment and ensiling days. Control, TMR without inoculants; Lr A4-2, L. reuteri A4-2, applied at 1 × 10⁵ cfu/g of fresh TMR; Lb 9-2, L. buchneri 9-2, applied at 1 × 10⁵ cfu/g of fresh TMR. ¹ SEM, standard error of the means. ^A–C Means in the same row with different uppercase letters differed for the ensiling duration (p < 0.05). ^a–c Means in the same column with different lowercase letters differed for treatments (p < 0.05).

. The pH values showed minimal variation across treatments, with no significant practical differences observed. However, the LA concentrations varied significantly, reaching the highest level of 53.73 g/kg DM in the 210 d ensiling period, compared to 32.44 g/kg DM at 7 d (p < 0.001). The AA levels also increased progressively with longer ensiling durations, peaking at 44.70 g/kg DM by 210 d, up from 16.99 g/kg DM at 7 d (p < 0.001). Similarly, the PA concentrations rose from 4.336 g/kg DM at 7 d to 15.17 g/kg DM by 210 d (p < 0.001). The water-soluble carbohydrate (WSC) levels declined over time, starting at 34.66 g/kg DM at 7 d and decreasing to 16.98 g/kg DM by 210 d (p = 0.001). The NH₃-N concentrations increased steadily with ensiling duration, reaching 26.95 g/kg DM at 210 d, from 14.48 g/kg DM at 7 d (p = 0.001). In terms of inoculation effects, the LA concentration peaked at 49.08 g/kg DM in the Lr A4-2 treatment, compared to 42.03 g/kg DM in the control group and 39.64 g/kg DM in the Lb 9-2 treatment at 210 d of ensiling (p < 0.001). The AA concentrations were highest in the Lb 9-2 treatment at 33.32 g/kg DM, compared to 25.21 g/kg DM in the control and 28.15 g/kg DM in Lr A4-2 (p < 0.001). The PA levels were similar across treatments, with no significant differences observed. The WSC concentrations were highest in the Lb 9-2 treatment at 38.76 g/kg DM, while Lr A4-2 had the lowest at 25.38 g/kg DM (p < 0.001). The NH₃-N levels were lowest in the Lb 9-2 treatment at 12.57 g/kg DM, followed by Lr A4-2 at 14.60 g/kg DM, and highest in the control at 18.11 g/kg DM (p < 0.001). Significant interactions between treatment and ensiling duration were observed for all the measured parameters (p < 0.001) except for pH (p = 0.870), indicating that the effects of inoculation varied depending on the length of the ensiling period.

3.2. Chemical Characteristics of TMR Silage Inoculated with Different Lactic Acid Bacteria Strains

The chemical composition of TMR silage was assessed over a 210 d ensiling period, as detailed in

Table 3. Chemical composition of TMR silage on different ensiling days.

Item	DM, g/kg FW	aNDF	ADF	CP	Ash
	g/kg DM
Ensiling duration
7 d	444.3 AB	275.6 A	192.1 A	173.8 A	102.7 A
15 d	450.8 A	290.6 A	197.7 A	177.4 A	101.1 A
30 d	439.2 B	290.0 A	198.9 A	181.3 A	110.5 A
60 d	429.8 AB	279.6 A	196.5 A	187.5 A	104.5 A
210 d	433.6 AB	191.8 A	190.7 A	714.1 A	106.1 A
Inoculation
Control	496.8 a	291.1 b	194.1 a	202.2 a	84.86 a
Lr A4-2	497.3 a	300.1 a	200.1 a	203.9 a	81.06 a
Lb 9-2	491.5 a	300.5 a	203.8 a	200.4 a	77.03 a
SEM 1	0.622	0.802	0.607	0.467	0.667
p-value
T	<0.001	0.057	0.432	0.017	0.373
D	0.099	0.096	0.202	0.086	0.339
T × D	0.394	0.351	0.387	0.345	0.092

DM, dry matter; aNDF, neutral detergent fiber (NDF assayed with a heat-stable amylase and expressed inclusive of residual ash); ADF, acid detergent fiber; CP, crude protein; T, treatment; D, ensiling days; T × D, interaction between treatment and ensiling days. Control, TMR without inoculants; Lr A4-2, L.reuteri A4-2, applied at 1 × 10⁵ cfu/g of fresh TMR; Lb 9-2, L. buchneri 9-2, applied at 1 × 10⁵ cfu/g of fresh TMR. ¹ SEM, standard error of the means. ^A,B Means in the same row with different uppercase letters differed for the ensiling duration (p < 0.05). ^a,b Means in the same column with different lowercase letters differed for treatments (p < 0.05).

. The DM content varied significantly among the treatments, with the control group showing the highest DM content at 496.8 g/kg FW, followed closely by Lr A4-2 at 497.3 g/kg FW and Lb 9-2 at 491.5 g/kg FW (p < 0.001). The CP content slightly increased over time, with no significant differences observed among treatments by day 210 (p = 0.017). The ash content, aNDF, and ADF levels were not significantly influenced by treatments or the interaction between the treatment and the ensiling duration (p > 0.05).

3.3. Diversity Parameters of the Bacterial Community of TMR Silage

Table 4. The variations in bacterial alpha diversity of TMR silage.

Item	Shannon	ACE	Chao1	Simpson	Coverage
Ensiling duration
7 d	2.66 A	357 A	455 A	0.14 B	0.994 C
15 d	2.45 A	350 A	308 B	0.15 B	0.996 B
30 d	1.23 B	216 A	184 C	0.46 AB	0.997 A
60 d	0.76 B	244 A	162 C	0.66 A	0.998 A
210 d	0.75 B	211 A	145 C	0.67 A	0.998 A
Inoculation
Control	1.41 a	315 a	251 ab	0.513 c	0.996 a
Lr A4-2	1.67 b	300 a	239 a	0.322 a	0.997 a
Lb 9-2	1.63 a	302 a	262 b	0.430 b	0.997 a
SEM 1	0.135	47.24	30.71	0.039	0.001
p-value
T	0.009	0.753	0.866	<0.001	0.916
D	0.001	<0.001	<0.001	<0.001	<0.001
T × D	0.001	0.236	0.208	<0.001	0.296

Control, TMR without inoculants; Lr A4-2, L.reuteri A4-2 treatment; Lb 9-2, L.buchneri 9-2, treatment. FM represents fresh material. T, treatment; D, ensiling days; T × D, interaction between treatment and ensiling days. ¹ SEM, standard error of the means. ^A–C Means in the same row with different uppercase letters differed for the ensiling duration (p < 0.05). ^a–c Means in the same column with different lowercase letters differed for treatments (p < 0.05).

explains the results for the alpha diversity indexes. The Shannon index revealed significant effects for the treatments (p = 0.009), ensiling days (p = 0.001), and their interaction (p = 0.001). Lr A4-2 maintained higher bacterial diversity after 60 days compared to the control and Lb 9-2 treatments. The ACE and Chao1 indices showed significant effects of ensiling days (p < 0.001), but no significant differences were observed among treatments or their interactions (p > 0.05). The Simpson index increased significantly over time, with Lr A4-2 exhibiting lower dominance after 60 days of ensiling (p < 0.001). Coverage values remained high across all treatments and ensiling durations, indicating robust sequencing depth. Regarding treatment effects, the control group had a Shannon index of 1.41, while Lr A4-2 and Lb 9-2 showed higher values of 1.67 and 1.63, respectively (p = 0.009). The Simpson index was lowest in the Lr A4-2 treatment at 0.322, followed by Lb 9-2 at 0.430, and was highest in the control at 0.513 (p < 0.001). The ACE and Chao1 indices did not differ significantly among the treatments (p > 0.05). The interaction between the treatment and the ensiling duration significantly influenced the Shannon and Simpson indices (p = 0.001), indicating that the effects of inoculation on bacterial diversity varied with the length of the ensiling period.

Figure 1 shows the parameters of the bacterial community composition in TMR silage. The taxonomic distribution of bacterial sequences at the genus level in TMR exhibited dynamic changes across different treatments over various ensiling days. In the fresh material (FM), the bacterial community was dominated by Weissella spp., Lactiplantibacillus spp., and Companilactobacillus sp. The control treatment showed a more complex colony composition at 7 d and 15 d, but starting from 30 d, it was mainly dominated by Lactobacillus spp. The Lr A4-2 treatment demonstrated a rapid dominance of Lactobacillus spp.; in addition to this, Companilactobacillus sp. also accounted for a large proportion at 15 d and 30 d, but by 60 d and 210 d, pronounced dominance of Lactobacillus spp. was observed. Notably, the Lb 9-2 treatment maintained a higher proportion of Lactobacillus spp. and Lentilactobacillus spp. after 15 d. All treatments converged to a Lactobacillus-dominated community structure by 210 d, with a minimal presence of other genera. However, the rate and pattern of this convergence differed significantly among treatments (p < 0.001).

A comprehensive analysis of bacterial species distribution in the TMR silage revealed significant microbial population shifts over time (Figure 2). In the FM, the bacterial community was dominated by Companilactobacillus farciminis, Levilactobacillus namurensis, and Lactiplantibacillus plantarum. The control group exhibited a transient increase in Companilactobacillus farciminis at 15 d; however, inconsistent stability in microbial diversity was observed over time, with Lactobacillus acetotolerans maintaining dominance. The Lr A4-2 treatment caused a temporary increase in Companilactobacillus farciminis and Levilactobacillus namurensis at 15 and 30 d, indicating a transient shift in microbial structure. However, the microbial composition stabilized at 60 and 210 d with Lactobacillus acetotolerans dominance. The Lb 9-2 treatment showed a noticeable rise in Companilactobacillus farciminis and Levilactobacillus namurensis on ensiling days 15 and 30. However, at 60 and 210 d the communities resembled the control group and were mainly dominated by Lactobacillus acetotolerans.

Temporal dynamics exhibited distinct patterns across different ensiling periods, with the variation explained by principal coordinates PC1 and PC2 fluctuating over time, as shown in Figure 3. PC1 and PC2 accounted for 77.41% of the total variation, with FM exhibiting a distinct clustering on d 7. As the ensiling process progressed, community differentiation became more pronounced, peaking on day 210, when PC1 accounted for 80.08% of the variation. Treatment-specific patterns were observed, with the Lr A4-2 and control treatments displaying increasingly similar community structures over time, while Lb 9-2 consistently maintained a distinct profile.

The Spearman correlation heatmaps are shown in Figure 4, and the RDA analysis revealed dynamic relationships between microbial genera, fermentation characteristics, chemical composition, and additive treatments, as shown in Figure 5. Over the ensiling duration of 7 and 15 d, Lentilactobacillus spp. and Secundilactobacillus sp. showed a strong positive correlation with CP and LA. However, Secundilactobacillus sp. and Acetobacter sp. displayed a strong negative correlation with AA and WSC. TMR silages treated with Lb 9-2 enhanced positive correlations between AA, CP, DM, and Lentilactobacillus spp.; in contrast, LA and NH₃-N in the silages treated with Lr A4-2 and the control treatments negatively correlated with Loigolactobacillus sp. and Companilactobacillus sp. By 3 and 60 d of the ensiling period Lentilactobacillus spp., Weissella spp., and Weizmannia spp. correlated positively with AA, while Companilactobacillus sp. and Lactiplantibacillus spp. exhibited a positive correlation with LA and CP, respectively, whereas a negative correlation was observed between lactobacillus spp. and WSC. Additive Lb 9-2 maintained strong positive correlations between AA, DM, and Lentilactobacillus spp., while WSC treated with Lr A4-2 positively correlated with Levilactobacillus spp., but LA treated with Lr A4-2 and the control group negatively correlated with Lactobacillus spp. Long-term ensiling (210 days) revealed more pronounced patterns, with Lb 9-2-treated silages showing positive correlations between AA and Lentilactobacillus spp., as well as CP and Caldibacillus sp. Conversely, negative correlations were observed between AA and Lactobacillus spp., as well as between WSC and both Companilactobacillus sp and Loigolactobacillus sp.

4. Discussion

Bacterial inoculants significantly modulated the TMR fermentation profile. The treatment with Lr A4-2 demonstrated enhanced LA production, achieving a 49.08 g/kg DM concentration at 210 d (53.73 g/kg DM). This finding corroborates the previously published results [34], highlighting targeted bacterial inoculation’s efficacy in improving silage fermentation. In contrast, the Lb 9-2 treatment produced the highest levels of AA and PA, aligning with already conducted research [35] and emphasizing heterofermentative bacteria’s role in silage preservation. The increased production of different organic acids in different treatments indicates that there may exist a different function for inoculums in extending the shelf life of TMR silage.

The fermentation characteristics of TMR silage exhibited distinct patterns across various treatments and ensiling durations. All treatments demonstrated a similar trend in pH, characterized by an initial decline followed by a slight increase toward the conclusion of the ensiling period. This trend aligns with typical silage fermentation processes documented in previous studies [36]. The significant effect of treatment on pH indicates that the lactic acid bacteria (LAB) inoculants played a crucial role in the acidification process, which is vital for maintaining silage quality. The LA concentration increased significantly in the Lr A4-2 treatment, reaching the highest peak at 49.08 g/kg on d 210 (53.73 g/kg DM). This finding is consistent with recent research highlighting the effectiveness of Lactobacillus strains in promoting lactic acid production in silage [37]. The elevated LA levels in the Lr A4-2 treatment suggest a more efficient fermentation process, which is essential for suppressing undesirable microorganisms and preserving nutrients [38]. The Lb 9-2 treatment produced the highest AA concentration by d 210, surpassing the control and Lr A4-2 treatments. This observation is consistent with the known heterofermentative characteristics of L. buchneri, which generates substantial amounts of acetic acid during prolonged fermentation periods [36]. The increased AA concentration in Lb 9-2-treated silage may enhance aerobic stability, as acetic acid possesses antifungal properties that inhibit the proliferation of spoilage microorganisms upon exposure to air [36,39]. In addition, the Lb 9-2 treatment generally maintained higher and more stable crude protein levels; however, protein degradation is a common challenge encountered during the ensiling process, primarily resulting from the combined actions of plant proteases and microbial proteolytic activity [40].

The taxonomic distribution of bacterial sequences exhibited dynamic changes across treatments and ensiling periods. Initially, Weissella spp. was prevalent in the fresh material; however, it transitioned to a Lactobacillus-dominated community by the conclusion of the ensiling process across all treatments. This succession pattern is characteristic of silage fermentation and has been documented in various forage types [21,40]. In line with previously reported data, this swift dominance of Lactobacillus spp. indicates successful inoculation and an efficient fermentation process [37]. In contrast, the Lb 9-2 treatment exhibited a more gradual increase in Lactobacillus spp. abundance while maintaining a higher proportion of Lentilactobacillus spp. throughout the ensiling period. The observed shifts in bacterial community composition are consistent with previous findings [41], which reported that LAB inoculation significantly altered the microbial community structure in whole-crop barley silage. The dominance of Lactobacillus species in LAB-inoculated treatments by day 210 aligns with prior research [21], which confirmed that Lactobacillus became the predominant genus in TMR silage inoculated with LAB strains. The addition of apple pomace in the TMR may have influenced the dynamics of the bacterial community, as apple pomace contains pectin and other fermentable carbohydrates that can serve as substrates for various bacterial species [42]. This additional nutrient source could have supported the growth of diverse bacterial populations, potentially contributing to the observed differences in community composition among treatments.

The alpha diversity analysis revealed a general decline in bacterial diversity over time across all treatments, with the control group exhibiting the most significant reduction. This decrease in diversity is a well-documented phenomenon in silage fermentation, attributed to the acidic conditions that favor acid-tolerant bacterial species [21]. Notably, the Lr A4-2 treatment maintained a higher Shannon index throughout the ensiling period than the control and Lb 9-2 treatments, suggesting that inoculation with L. reuteri A4-2 contributed to the preservation of bacterial diversity to some extent. The results from the Simpson index further demonstrated the dominance of specific bacterial species over time across all treatments. The lower Simpson index observed in the Lr A4-2 treatment indicates that inoculation with L. reuteri A4-2 promoted a more balanced bacterial community. This equilibrium may enhance the fermentation process’s stability and improve the silage’s quality, as indicated by recent investigations into microbial community dynamics in silage [40,43].

Incorporating apple pomace into the TMR likely influenced the nutritional quality of the silage. Apple pomace is rich in soluble sugars and phenolic compounds, which can affect both the fermentation process and the nutritional profile of silage [42]. Recent research has shown that including apple pomace in ruminant diets can enhance animal growth and lactation performance [43]. Furthermore, the polyphenols present in apple pomace may improve the antioxidant capacity of the silage, potentially enhancing its nutritional value and stability [44]. However, it is essential to consider the optimal inclusion rate of apple pomace in TMR silages.

The results of this study have significant implications for silage management practices, particularly in regions characterized by cold climates. Applying LAB inoculants isolated from such environments as high and cold regions shows potential for enhancing the fermentation quality and bacterial community composition of alfalfa-based TMR silage containing apple pomace. The rapid dominance of beneficial bacteria in the inoculated treatments indicates that these strains are well-suited to the silage environment and can effectively outcompete undesirable microorganisms. The differences observed between the Lr A4-2 and Lb 9-2 treatments underscore the necessity of selecting appropriate LAB strains based on the desired characteristics of the silage. While L. reuteri A4-2 facilitated rapid acidification and elevated lactic acid levels, L. buchneri 9-2 increased acetic acid production and enhanced aerobic stability. This finding is consistent with recent research advocating the combination of heterofermentative LAB strains to optimize silage quality [38,44]. The inclusion of apple pomace in the TMR is particularly noteworthy, as it exemplifies an effective strategy for utilizing by-products from the food industry. This approach aligns with contemporary efforts to enhance resource efficiency in animal feed production, as evidenced by the work of Fang et al. [42], who explored the effects of substituting commercial materials with apple pomace in TMR silage.

5. Conclusions

In conclusion, this research demonstrates compelling evidence for the significant impact of LAB inoculants on the bacterial diversity, chemical composition, and fermentation quality of alfalfa-based TMR silage incorporating apple pomace. The results underscore the critical importance of considering both ensiling duration and LAB strain selection in developing effective silage management strategies. The observed enhancements in fermentation quality and microbial community structure suggest that the LAB strains utilized in this study may serve as efficacious silage inoculants, particularly for the ensiling of TMR. Importantly, the present results suggest that different heterofermentative LAB strains modulated the fermentation and microbial balance in alfalfa-based TMR silage in different ways. The strain Lr A4-2 performed well in promoting TMR silage fermentation quality via high lactic acid production, whereas the strain Lb 9-2 was good at improving aerobic stability because of the high acetic acid concentration in its inoculated TMR silage.