Effects of Bacteriocin-Producing Lactiplantibacillus plantarum on Fermentation, Dynamics of Bacterial Community, and Their Functional Shifts of Alfalfa Silage with Different Dry Matters

Abstract

This study investigated the effects of two bacteriocin-producing Lactiplantibacillus plantarum strains on fermentation, bacterial communities, and their functions of alfalfa silage with two dry matter (DM) contents of 355 (moderate DM) and 428 (high DM) g/kg fresh weight. Before ensiling, alfalfa was treated with (1) distilled water (control), (2) the commercial strain L. plantarum MTD/1, (3) bacteriocin-producing L. plantarum ATCC14917, and (4) bacteriocin-producing L. plantarum LP1-4, and ensiled for 3 d, 7 d, 14 d, 60 d, and 90 d, respectively. Application of ATCC14917 promoted lactic acid production in the moderate DM silage at the early fermentation stage (3 d). Silages treated with ATCC14917 and LP1-4 showed lower DM losses and non-protein nitrogen concentrations versus the control or MDT/1-treated silage (p < 0.05). During fermentation, a high proportion of Weissella cibaria was observed in the silages with high DM content from 3 to 60 d of ensiling, and the functions of carbohydrate and amino acid metabolisms of silage bacterial community were decreased by ATCC14917 before 60 d of ensiling. In addition, ATCC14917 also inhibited the growth of Aerococcus and Enterobacter in silage. Therefore, the bacteriocin-producing L. plantarum ATCC14917 has a great potential to improve alfalfa silage quality, nutritive value, and safety as well.

1. Introduction

Ensiling is a traditional way of preserving fresh forage for ruminants’ nutrition worldwide. As one of the main ingredients of cattle diets, high quality silage provides the animals with essential nutrients, such as crude protein (CP), digestible fiber, and micronutrients. As a leguminous forage, alfalfa (Medicago sativa L.) is considered a high-quality silage source for ruminants because of its high protein content and good palatability. However, the pollution of spoilage bacteria and nutrient loss during ensiling are still the obstacles to the production of high-quality silage, especially in legumes with low water-soluble carbohydrates (WSC) content and high buffering. In order to ensure and further improve the fermentation quality, different lactic acid bacteria (LAB) silage additives were developed to explore their effects on silage fermentation quality, aerobic stability, and animal productivity [1]. Typically, the ensiling process of forage crops is dominated by LAB, and alterations in microbial communities are closely related to silage fermentation [2]. Therefore, profiling of silage microbial communities can enhance our understanding of the biological process underlying silage formation and how LAB inoculants modulate this process. Bai et al. [3] and Xu et al. [4] reported that different LAB inoculants had different manners in modulating the dynamics of microbial community and their functional shifts during the ensiling process.

Currently, the innovative research on silage LAB is undertaken to fulfill human health requirements, such as animal feed safety and nutrition, and even animal welfare, precisely for human food safety and health [5]. Based on the existing literature, the application of bacteriocin-producing LAB can not only improve silage fermentation but also effectively control the growth of Listeria and enterobacteria in silages. Amado et al. [6] reported that inoculation with bacteriocin-producing Lactococcus lactis CECT 539 or Pediococcus acidilactici NRRL B-5627 before ensiling controlled the occurrence of Listeria monocytogenes in grass and maize silages. Amado et al. [7] further confirmed that the growth of Listeria in maize silage was completely inhibited after 8 d of ensiling by application of the purified pediocin SA-1 produced by Pediococcus acidilactici NRRL B-5627. Previous studies also reported the effective controlling of enterobacteria by application of LAB with bacteriocinogenic potential in ensiled forages [8,9]. Li et al. [10] reported that bacteriocin-producing inoculants improved alfalfa silage fermentation quality, reduced the growth of yeasts and molds, and improved the aerobic stability of the ensiled forage. Moreover, bacteriocin has the potential to enhance animal performance and feeding efficiency as a substitute for feeding antibiotics without drug resistance or harmful residues and is environmentally friendly [11,12]. Therefore, the application of bacteriocin-producing LAB inoculants has great potential in improving the safety and fermentation quality of silage. However, to the best of our knowledge, the effects of bacteriocin-producing LAB on bacterial community composition, and their functions in silage, as well as the efficacy of application of bacteriocin-producing inoculants at different silage dry matter (DM) contents, are still unclear.

Thus, the objective of this study was to investigate the effects of two bacteriocin-producing Lactiplantibacillus plantarum strains on fermentation characteristics, dynamics of bacterial community, and their functional shifts of alfalfa silage ensiled at different dry matter contents.

2. Materials and Methods

2.1. Strain Preparation

The inoculants used in the present study included (1) L. plantarum MTD/1 (NCIMB 40027), a non-bacteriocin-producing silage inoculant, purchased from Ecosyl Products Ltd., Stokesley, UK; (2) L. plantarum ATCC14917, a class IIa bacteriocin-producing strain purchased from American type culture collection [13]; and (3) L. plantarum LP1-4, a bacteriocin-like substance producing strain, isolated and screened from turbot intestine originally sampled in Jinzhou, China [14]. The three strains were activated twice at 1% (v/v) in MRS broth, and cultured at 37 °C for 18 h before making silage.

2.2. Alfalfa Silage Preparation

Alfalfa (“Zhongmu 2#”) at early flowering period was manually harvested from four experimental fields in Dingxi (N35°58′, E104°62′), Gansu province, China, on July 2022. The harvested alfalfa forages from each field were divided into two portions, and naturally wilted to DM contents of around 350 g/kg and 430 g/kg fresh weight (FW), respectively. The wilted forages from four harvested fields and with each DM content served as four experimental replicates. Thereafter, the wilted alfalfa forages from each field were chopped into small pieces using a hand hay cutter, roughly 1−2 cm in length, and the chopped alfalfa with each DM content was subsequently divided into five piles (five ensiling periods: 3 d, 7 d, 14 d, 60 d, and 90 d). The five piles from each of the four fields were then treated with one of the following treatments: (1) distilled water (control), (2) L. plantarum MTD/1 (MTD/1), (3) L. plantarum ATCC14917 (ATCC14917), and (4) L. plantarum LP1-4 (LP1-4) applied on 1 × 10⁵ colony-forming units (CFU)/g FW. Approximately 500 g of wilted alfalfa from each field was collected as fresh samples and frozen at –20 °C for subsequent analysis. The LAB were inoculated two consecutive times at a ratio of 2%. Each LAB culture was centrifuged at 8000× g for 5 min, and the precipitation was suspended in sterile water to achieve an application rate of 1 × 10⁸ CFU of viable cells/mL and evenly sprayed on chopped alfalfa at 5 mL/500 g FW. An equal volume of sterile water was used to treat the control. Each treated pile (approximately 500 g) was filled into vacuum plastic bags and sealed (density 0.91 to 0.93 g/cm3; vacuum degree 0.1 Mpa).

2.3. Chemical Composition and Fermentation Characteristics Analysis

At the opening time, a 20 g sample (including fresh samples and silage) from each mini-silo was taken randomly, soaked with 180 mL sterile water, and filtered with four layers of sterile gauze after homogenization. The pH of the silage filtrate was instantly determined with a pH meter (PHSJ-3F, CANY, Shanghai, China). A portion of the filtrate was acidified to pH 2.0 with 7.14 mM H₂SO₄, filtered with a 0.22-μm filter, and frozen at –20 °C for subsequent organic acid analysis. Organic acids (lactic, acetic, propionic acid, and butyric acid) were measured using high-performance liquid chromatography system (KC-811 column, Agilent Technologies Inc.; Santa Clara, CA, USA; oven temperature 50 °C; flow rate: 1 mL/min; SPD: 210 nm) according to the method of Yang et al. [15]. Another portion of non-acidified filtrate was added to trichloroacetic acid to eliminate true protein, and the supernatant was used to determine water-soluble carbohydrates (WSC), non-protein nitrogen (NPN), and ammonia nitrogen (NH₃-N) according to Thomas [16], Licitra et al. [17], and Broderick and Kang [18], respectively.

Approximately 80 g of fresh and silage samples were oven-dried at 60 °C for 72 h for subsequent chemical analysis. The dried samples were milled through a 1 mm sieve. To determine the DM contents, 20 g samples were accurately weighed and dried at 105 °C for three h. CP, neutral detergent fiber (aNDF; using heat-stable α-amylase) and acid detergent fiber (ADF) contents were determined using the methods described by Broderick and Kang [18] and Van Soest et al. [19], respectively.

2.4. DNA Extraction and SMRT Sequencing Analyses

Silage samples were used to extract total bacterial DNA after fermentation for 3 d, 7 d, 14 d, 60 d, and 90 d. Briefly, silage samples (15 g) were added to a sterile saline solution (50 mL) and shaken with a constant temperature shaker (zxy-48, Runhua Co., Ltd. Changzhou, China) at room temperature (25 °C) at 120 rpm for 10 min. The shaken silage was then filtered through 8 folds of medical gauze. The filtrate was centrifuged at 4 °C at 10,000× g for 10 min, and the precipitate was collected (bacterial cells). The total DNA was extracted using a Fast DNA SPIN for soil kit (Tiangen Biochemical Technology Co., Ltd., Beijing, China) following the manufacturer’s instructions. The concentration and purity of the extracted DNA were determined using a NanoDrop 2000 UV–vis Spectrophotometer (Thermo Scientific, Waltham, MA, USA), and 1% agarose gel electrophoresis was used to assess the quality of the extracted DNA. Polymerase Chain Reaction (PCR) amplification, purification, quantification, and bioinformatic analysis were performed according to the method described by Du et al. [20]. The diversity index was based on the Operational Taxonomic Unit (OUT). According to the analysis of the OTU clustering, silage samples were analyzed for unique information. The principal component analysis (PCA) was conducted to assess the structural variation of microbiota. Detrended correspondences were performed using R package 4 (version 3.2.5) prior to redundancy analysis (RDA). According to the phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt), functional genes of the bacterial communities were predicted.

2.5. Statistical Analysis

The data of fermentation parameters were analyzed by using the general linear model of SPSS 20.0 (IBM Co., Armonk, NY, USA) according to a 5 (period) × 4 (treatment) × 2 (DM) experimental design with the fixed factors of treatments, DM content, ensiling time and their interactions. Tukey’s multiple comparison test was employed to separate means when at least one of the interactions was significant at p < 0.05. Data on chemical composition of 90 d silage and the Shannon index of bacterial community for each ensiling time were subjected to the general linear model of SPSS 20.0 according to the 2 (DM content) × 4 (treatment) factor design with the fixed factors of treatment, DM content, and their interaction. The effect of DM within each treatment was analyzed using Tukey’s multiple comparison test when the interaction was significant at p < 0.05. Data on relative abundance of bacterial community functions were subjected to a one-way ANOVA, and the Tukey’s multiple comparison test was employed to compare the differences among treatments at each ensiling time. The significance was declared at p < 0.05.

3. Results

3.1. Chemical Composition of Alfalfa before Ensiling

The chemical composition of alfalfa prior to ensiling is displayed in

Table 1. Chemical composition of alfalfa prior to ensiling (n = 3).

Item 1	Alfalfa 2
	M	H
DM, g/kg FM	355	428
WSC, g/kg DM	31.5	36.7
CP, g/kg DM	177	172
aNDF, g/kg DM	366	391
ADF, g/kg DM	283	294
pH	6.17	6.49

¹ DM, dry matter; FM, fresh matter; WSC, water-soluble carbohydrates; CP, crude protein; aNDF, neutral detergent fiber with heat-stable α-amylase; and ADF, acid detergent fiber. ² M, moderate DM content; H, high DM content.

. The harvest fresh alfalfa was wilted to two different DM contents of 355 (moderate DM) and 428 (high DM) g/kg fresh weight, respectively. For the wilted alfalfa with the moderate and high DM contents, the pH values were 6.17 and 6.49, the CP concentrations were 177 and 172 g/kg DM, the WSC concentrations were 31.5 and 36.7 g/kg DM, the aNDF concentrations were 366 and 391 g/kg DM, and the ADF concentrations were 283 and 294 g/kg DM, respectively.

3.2. Dynamics of Fermentation Characteristics of Ensiled Alfalfa

There were additive × DM × ensiling time interactions (p = 0.001) for pH, lactic acid, acetic acid, and propionic acid (

Table 2. Dynamics of fermentation characteristics of alfalfa silage (n = 4).

Iterms 1	Additives 2	M						H					Mean	SEM 3	Effects 4
		3 d	7 d	14 d	60 d	90 d	Mean	3 d	7 d	14 d	60 d	90 d			A	D	T	A × D	A × T	D × T	A × D × T
pH, g/kg DM	Control	5.87 aA	5.46 aB	5.31 aC	4.78 aD	4.56 aE	5.19	6.04 aA	5.70 aB	5.48 aC	4.62 bD	4.59 aD	5.28	0.005	<0.001	0.24	<0.001	0.102	<0.001	0.062	<0.001
	MTD/1	4.33 dC	4.39 cAB	4.35 cC	4.52 bA	4.41 cAB	4.40	5.31 cA	4.92 cB	4.71 cC	4.45 cD	4.47 AbD	4.77
	ATCC14917	4.50 cA	4.45 cA	4.34 cAB	4.39 bA	4.22 dB	4.39	5.20 cA	4.80 cB	4.60 dC	4.42 cD	4.44 bD	4.69
	LP1-4	5.35 bA	5.01 bB	5.01 bB	4.55 bC	4.46 bC	4.88	5.83 bA	5.50 bB	5.31 bC	4.79 aD	4.63 aE	5.21
LA, g/kg DM	Control	0.00 cD	7.56 cC	11.7 bC	34.7 aA	21.3 bB	15.22	0.00 bC	13.9 cB	14.4 AbB	19.8 bA	21.5 abA	13.92	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	MTD/1	16.5 bC	27.3 abAB	29.4 aAB	33.9 aA	27.5 abB	26.72	16.9 aB	22.7 abA	19.0 abAB	22.3 aAB	19.8 bcAB	20.13
	ATCC14917	31.1 a	31.3 a	34.0 a	23.6 b	30.7 a	30.08	18.3 a	26.2 a	25.1 a	22.1 a	23.2 a	22.99
	LP1-4	16.5 bC	21.4 bBC	18.1 bC	28.4 abA	25.0 abAB	21.90	19.5 a	17.1 bc	16.6 b	20.3 ab	17.7 c	18.24
AA, g/kg DM	Control	0.00 cD	0.00 bD	4.34 aC	26.5 aA	12.0 aB	8.56	0.00 cD	0.00 bD	3.38 bC	10.9 aA	7.81 bB	4.42	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	MTD/1	2.19 aB	3.70 aB	4.06 aB	7.91 bcA	9.47 abA	5.46	1.17 bC	3.57 aC	6.87 abB	10.6 aA	8.53 abAB	6.15
	ATCC14917	2.37 a	4.16 a	4.36 a	4.96 Bc	6.56 Bb	4.65	2.49 aC	3.48 aC	8.91 aA	8.51 abA	9.62 aA	6.60
	LP1-4	1.73 bC	3.52 aC	1.66 bC	9.34 bB	11.5 aA	5.56	2.75 aC	2.50 aC	4.31 abBC	6.69 bAB	7.45 bA	4.74
PA, g/kg DM	Control	0.00 bD	2.35 bCD	4.05 bC	19.5 aA	13.0 aB	7.78	0.00 cD	1.73 bC	3.22 bB	9.76 aA	9.42 aA	4.83	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
	MTD/1	1.88 aB	2.04 bB	3.51 bB	11.1 bA	12.2 aA	6.14	1.47 bC	1.84 bC	4.08 abB	8.49 aA	8.71 abA	4.92
	ATCC14917	1.70 a	1.94 b	3.81 b	6.22 c	8.58 b	4.77	1.96 abC	1.83 bC	4.35 aB	6.56 bA	8.45 abA	4.63
	LP1-4	1.50 aD	2.95 aC	6.35 aB	11.5 bA	12.0 aA	6.88	3.13 aCD	2.55 aD	4.24 abC	8.57 aA	6.65 bB	5.03

¹ LA, lactic acid; AA, acetic acid; PA, propionic acid; DM, dry matter; M, moderate dry matter; and H, high dry matter. ² Control, distilled water; MTD/1, MTD/1 treatment; ATCC14917, ATCC14917 treatment; LP1-4, LP1-4 treatment. ^a–d Different lowercase letters indicate significant differences in treatments (p < 0.05). ^A–E Different uppercase letters indicate significant differences in ensiling periods (p < 0.05). ³ SEM, standard error of the means. ⁴ A, additive; D, dry matter; T, ensiling time; A × D, the interaction between additive and dry matter; A × T, the interaction between additive and ensiling time; D×T, the interaction between dry matter and ensiling time; A × D × T, the interaction among additives, dry matter, and ensiling time.

). Butyric acid was not detected in all treatments during ensiling in the present study. All inoculants decreased silage pH compared with the control group during the entire ensiling period at the moderate DM content, while increased and comparable pH values were observed in 60 d and 90 d silages respectively treated with LP1-4 at the high DM content. Although all inoculants increased lactic acid concentrations in 3 d and 7 d silages at both DM contents, different effects of inoculants on lactic acid concentrations were observed in the moderate and high DM silages ensiled for 14 d, 60 d, and 90 d, respectively. Inoculation with ATCC14917 increased lactic acid concentration in 14 d and 90 d silages at the moderate DM content relative to the control silage, while increased concentrations of lactic acid were observed in 14 d and 60 d silages treated with ATCC14917 at the high DM content. In the initial phase of ensiling (3−7 d), acetic acid concentrations in silages at both moderate and high DM contents were increased (p < 0.05) by inoculants, and decreased (p < 0.05) in the inoculants-treated silages on d 60 and in LP1-4-treated-silage on d 90 at the moderate DM; however, acetic acid concentrations in silages with the high DM content were reduced (p < 0.05) by LP1-4 on d 60 and enhanced (p < 0.05) by ATCC14917 on d 90.

3.3. Chemical Characteristics of Ensiled Alfalfa

There was an additive × DM interaction (p < 0.05) for DM content, DM loss, and concentrations of WSC, CP, and ADF (

Table 3. Chemical composition of alfalfa silage after 90 d of ensiling (n = 4).

Iterms 1	Additives 2	M	H	SEM 3	Effects 4
					A	D	A × D
DM loss, g/kg DM	Control	83.2 Aa	69.0 Ba	1.267	< 0.001	< 0.001	0.006
	MTD/1	68.7 b	63.7 a
	ATCC14917	51.9 Ac	19.0 Bc
	LP1-4	62.0 Abc	44.8 Bb
DM g/kg, FM	Control	337 Bc	409 Ac	0.11	<0.001	<0.001	<0.001
	MTD/1	341 Bbc	418 Abc
	ATCC14917	349 Aa	442 Ba
	LP1-4	346 Bab	426 Ab
WSC, g/kg DM	Control	5.43 Ac	4.76 Bd	0.126	0.007	<0.001	<0.001
	MTD/1	7.95 b	7.32 c
	ATCC14917	11.4 a	12.9 a
	LP1-4	7.19 Bb	9.95 Ab
CP, g/kg DM	Control	163 b	168 c	0.481	0.002	<0.001	0.036
	MTD/1	170 a	169 bc
	ATCC14917	171 a	176 ab
	LP1-4	171 Ba	177 Aa
NPN, g/kg DM	Control	581 Aa	592 Ba	6.734	0.028	<0.001	0.299
	MTD/1	409 b	379 b
	ATCC14917	316 Ac	260 Bc
	LP1-4	347 Abc	295 Bc
NH3-N, g/kg DM	Control	67.3 a	56.7 a	0.216	0.012	<0.001	0.299
	MTD/1	45.0 Bc	47.6 Ab
	ATCC14917	28.7 Bd	39.5 Ac
	LP1-4	54.7 Ab	47.2 Bb
aNDF, g/kg DM	Control	415 Aa	400 Ba	1.302	<0.001	0.031	0.127
	MTD/1	386 b	383 b
	ATCC14917	390 b	382 b
	LP1-4	383 b	385 b
ADF, g/kg DM	Control	355 a	353 a	1.226	< 0.001	0.942	0.001
	MTD/1	346 ab	330 b
	ATCC14917	335 b	334 b
	LP1-4	334 Bb	352 Aa

¹ DM loss, dry matter loss; FM, fresh matter; DM, dry matter; WSC, water-soluble carbohydrates; CP, crude protein; NPN, non-protein nitrogen; TN, total nitrogen; NH₃-N, ammonia nitrogen; aNDF, neutral detergent fiber with heat-stable α-amylase; and ADF, acid detergent fiber. M, moderate dry matter; H, high dry matter. ² Control, distilled water; MTD/1, MTD/1 treatment; ATCC14917, ATCC14917 treatment; and LP1-4, LP1-4 treatment. ³ SEM, standard error of the means. ⁴ A, additive; D, dry matter; A × D, the interaction between additive and dry matter. ^a–d Different lowercase letters indicate significant differences in treatments (p < 0.05). ^A–B Different uppercase letters indicate significant differences in dry matter contents (p < 0.05).

). All inoculated silages had lower DM losses compared with the control silage at the moderate DM content, while lower DM losses were only observed in silages treated with the two bacteriocin-producing strains at the high DM content. In addition, the three inoculants increased silage CP contents at the moderate DM content but the increased CP concentrations were only detected in the bacteriocin-producing strains’ inoculated silages at the high DM content. NPN concentrations in all three strain-treated-silages were lower (p < 0.05) than those in the control silage at both moderate and high DM; while NPN concentrations were decreased (p < 0.05) by ATCC17917 at the moderate DM, and the two bacteriocin-producing strain-treated-silages at the high DM compared with MTD/1. All three strain-treated-silages reduced (p < 0.05) NH₃-N concentrations, while the NH₃-N concentrations in ATCC14917 were the lowest (p < 0.05) at both moderate and high DM.

3.4. Dynamics of Bacterial Community of Ensiled Alfalfa

The additive × DM had an interaction (p < 0.001) on the Shannon index at each ensiling time (Figure 1). As expected, all inoculants decreased the Shannon index of silage bacterial community; however, for the silages with the moderate DM content, lower Shannon indexes were observed in MTD/1- and ATCC14917-treated-silages versus LP1-4-treated-silage at each ensiling time, while the Shannon index was the lowest in ATCC14917-treated-silage ensiled for 3 d, 14 d, and 60 d.

The bacterial community composition at species level is presented in Figure 2. For the silages with the moderate DM content, L. plantarum was the most abundant species in the MTD/1- and ATCC14917-inoculated silages fermented for 3 to 60 d of ensiling, while the most abundant bacterial species were L. plantarum and Lactobacillus buchneri in the three inoculant-treated-silages after 90 d of ensiling (Figure 2A). There was an increased abundance of Weissella cibaria in LP1-4 compared with the other groups at each fermentation period. Moreover, LP1-4 also increased the abundance of Weissella confusa in silage compared with the other groups during ensiling except for 90 d. Similar to the results from silages with the moderate DM content, L. plantarum was also the most abundant species in the inoculated silages with high DM content from 3 to 60 d of ensiling (Figure 2B). Unlike control silage with the moderate DM content, a relatively large proportion of W. cibaria was observed in the control group with the high DM content from 7 to 90 d, and a high relative abundance of L. buchneri was only found in LP1-4 silage ensiled for 90 d. The PCA dynamic analysis based on OTU is displayed in Figure 2C. The control group was clearly separated from the other groups during ensiling regardless of DM contents. LP1-4 with the moderate DM content was separated from the other groups during ensiling except for d 3, while at the high DM content, it was separated from the other groups during ensiling except for 14 d silage.

As undesirable bacteria accounted for a relatively low proportion in the composition; the 30 most abundant species were presented on the heatmap to analyze the dynamics of undesirable bacteria (Figure 3). The results showed that the abundance of Aerococcus urinaeequi in the silage with moderate DM content was decreased by ATCC14917 from d 3 to 14, and it was reduced by ATCC14917 from d 14 to 90 in the silage with the high DM content. The abundance of unclassified Enterobacter was lowered by application of MTD/1 in 3 d and 60 d silages with the moderate DM content, and lowered by ATCC14917 and LP1-4 in 14 d silage with the high DM content.

The RDA dynamic analysis on correlations among fermentation, bacterial species, and additive treatment at two DM contents is presented in Figure 4. Lactic and acetic acid contents in 3 d, 7 d, and 14 d silages treated with ATCC14917 and MTD/1 were positively correlated with L. plantarum at both DM contents, while lactic, acetic and propionic acids in 60 d silages were positively correlated with L. buchneri for all treatments except for the control group and Lp1-4 with the high DM content. For the 90 d silages, lactic, acetic, and propionic acids were also positively correlated with L. buchneri, especially in ATCC14917 and MTD/1 with the moderate DM content and in the control group with the high DM content. W. cibaria was positively correlated with pH during fermentation in the control group and Lp1-4 with the high DM content.

The predicted functions of bacterial community in silages within ensiling times are shown in Figure 5. At the initial fermentation stage of 3 d, a remarkable down-regulation in the function of carbohydrate metabolism was observed in ATCC14917 and MTD/1 with the moderate DM content compared with the control group and MTD/1 (p < 0.05). A lower abundance of carbohydrate metabolism was also observed in ATCC14917 silage fermented for 7 d and 14 d, respectively; whereas, the abundance of carbohydrate metabolism in ATCC14917 was only lower than MTD/1 in 60 d and 90 d silages (p < 0.05). In addition, application of ATCC14917 and Lp1-4 down-regulated the function of amino acid metabolism in 3 d, 7 d, and 14 d silages with the moderate DM content compared with the control group; however, an opposite result was observed in 90 d silage. For the silages with the high DM content, inoculation with ATCC14917 also down-regulated the function of carbohydrate metabolism in 3 d, 7 d and 60 d silages compared with the control group (p < 0.05). Similar to the results in silages with the moderate DM content, application of ATCC14917 and Lp1-4 down-regulated the function of amino acid metabolism in 3 d, 7 d, 14 d, and 60 d silages with the high DM content compared with the control and MTD/1-treated-silages.

4. Discussion

LAB inoculants are commonly used to improve silage fermentation quality and prevent the growth of undesirable microorganisms via a rapid accumulation of lactic acid and a quick acidification at the early stage of ensiling [21]. In the past decade, application of the bacteriocinogenic LAB in silage has been increasingly studied because these strains not only improve silage fermentation quality but also enhance silage safety [6,8,9]. In this study, all inoculants accelerated the lactic acid fermentation of ensiled alfalfa at both DM contents. Basically, silages treated with the bacteriocin-producing strain ATCC14917 and the commercial strain MTD/1 had comparable pH values during ensiling. However, the lowest pH was observed in 3 d and 90 d silages treated with ATCC14917 at the moderate DM content, which was consistent with the remarkable accumulation of lactic acid in the 3 d silage treated with ATCC14917 at the moderate DM content. Amado et al. [6] also reported that the forages treated with the bacteriocinogenic LAB had higher pH decline rates during the initial stage of ensiling. The DM content of fresh forage before ensiling has a pronounced effect on silage fermentation, and a higher DM content normally lead to a higher pH value and a lower lactic acid concentration in ensiled forages [22,23]. The present study also showed that alfalfa silages ensiled at the high DM content had higher pH values and lower lactic acid concentrations compared with the silages at the moderate DM content regardless of treatments. As expected, the two bacteriocin-producing strain-treated-groups at both moderate and high DM enhanced DM contents and lowered DM loss, with treatment of ATCC14917 being the most efficient. The result might be attributed to the depression of the growth of undesirable microorganisms by the antimicrobial peptides produced by the two strains [7]. A similar effect was also found in the silage N fractions, and the lowest NPN and NH₃-N content was observed in ATCC14917-treated-silage regardless of DM contents. This result was mainly due to the rapid decline in pH in ATCC14917 because proteolysis in ensiled forage can be effectively inhibited under low pH environment [24]. This allowed silage persist more undegraded protein, instead of the NPN with low utilization efficiency by ruminants, and prevented excessive N excretion [25]. The increased WSC concentrations in the two bacteriocin-producing strain-treated-silages at both moderate and high DM was consistent with the previous studies on the application of bacteriocin-producing LAB in silage [6,10].

In the present study, all the inoculated silages had lower Shannon indexes compared with the control group, indicating a reduction in bacterial species richness [3]. However, silage treated with LP1-4 at both moderate and high DM contents basically exhibited a higher Shannon index than silages treated with MTD/1 and ATCC14917 due to its lower acid-producing capacity during ensiling [26]. In addition, the pH values in ATCC14917 and MTD/1-treated-silages at both moderate and high DM were roughly similar, and both maintained a low level during the fermentation periods. However, the silage treated with ATCC14917 with the high DM content had a lower Shannon index than MTD/1 on d 3, d 14, and d 60, which might be attributed to the role of bacteriocin produced by ATCC14917.

As can be seen from the bacterial composition of silages with the moderate DM content, application of L. plantarum in MTD/1 and ATCC14917 rapidly promoted the silage fermentation from 3 to 14 d, and L. plantarum dominated the entire bacterial composition virtually, while a declining trend of L. plantarum was observed from d 60 to 90. Similar results were also reported by a previous study [24]. The high abundance of L. buchneri in MTD/1 and ATCC14917 at the end of fermentation (90 d) suggested that the sufficient acid-producing ability of MTD/1 and ATCC14917 provided favorable conditions for the growth of acid-tolerant L. buchneri at the late stage of fermentation [24]. Interestingly, inoculation with L. plantarum LP1-4 promoted a large proportion of W. cibaria in silages with the moderate DM content from 3 to 14 d. For the silages with the high DM content, a relatively large proportion of W. cibaria was also observed in the control silage during the whole ensiling period, but application of the three strains resulted in a remarkable decline in the abundance of W. cibaria in the silages ensiled from 7 to 60 d. In contrast with the results from the silages with the moderate DM content, the silage treated with LP1-4 at the high DM content promoted the growth of L. buchneri on d 90. A plausible explanation for the above results was that the degree of wilting on a high DM basis affected the composition of the bacterial community in the raw material of alfalfa [26].

Aerococcus and Enterobacter are two pathogens that cause human and animal infections, including urogenital infections, septicemia, pneumonia, and infective endocarditis [27,28]. The present results indicated that the antibacterial effect of ATCC14917 on pathogens was mainly at the early and middle stages of fermentation under the moderate DM condition, but mainly at the middle and late stages under the high DM condition. An explanation is that the fermentation process of ATCC14917 was delayed by the high DM, and its antibacterial periods were affected by the slow decrease in pH [29].

Based on the RDA analysis results, the organic acids were positively correlated with L. plantarum in 3 d, 7 d, and 14 d silages treated with MTD/1 and ATCC14917, which further confirmed that application of the two L. plantarum strains promoted silage fermentation at the early stage of ensiled alfalfa. However, the positive correlations between organic acids and L. buchneri in 60 and 90 d silages indicated that the fermentation was dominated by L. buchneri at the late stage of ensiling rather than L. plantarum. In addition, W. cibaria was positively correlated with pH during fermentation, indicating that it was not conducive to the fermentation of alfalfa silage [30].

PICRUSt2 can predict the metabolic pathways of microorganisms based on information from the KEGG database, thereby revealing the activities of microorganisms during silage fermentation. Basically, silage treated with ATCC14917 had the lowest abundance value of carbohydrate metabolism before 60 d of ensiling, indicating that the carbohydrate metabolism of the bacterial community was lowered by application of ATCC14917. This result might be one of the reasons that the lowest DM loss and highest WSC concentration were observed in ATTCC14917-treated-silage. In addition, the lower abundance of amino acid metabolism was found in silages treated with ATCC14917 and LP1-4 compared with the control and MTD/1-treated-silages before 60 d of ensiling, which was in accordance with the decreased NPN and NH₃-N in ATCC14917 and LP1-4 inoculated silages. It was reported that the proteolysis of ensiled forage is mainly caused by the plant enzymes, with a subsequent degradation of peptides and amino acids by microbial activity [31,32]. Therefore, inhibition of amino acid metabolism of the bacterial community in ensiled alfalfa by ATCC14917 and LP1-4 before 60 d of ensiling could also contribute to the decline in deamination of present silage besides inactivation of plant proteases by a rapid decline in silage pH [33].

5. Conclusions

Application of the bacteriocin-producing strain L. plantarum ATCC14917 effectively improved alfalfa silage fermentation quality. Both ATCC14917 and LP1-4 had better performance in inhibition of DM loss and degradation of protein in ensiled alfalfa compared with the non-bacteriocin-producing silage inoculant MTD/1 regardless of silage DM contents. The bacteriocin-producing strain L. plantarum ATCC14917 was the most effective in decreasing silage DM loss among the three inoculants used in the present study. The DM content had a pronounced effect on bacterial community successions in alfalfa silage, and a large proportion of W. cibaria was observed in silage with the high DM content from 3 to 14 d of ensiling. Application of ATCC14917 and MTD/1 increased the abundance of L. buchneri in the 90 d silage with the moderate DM content but not with the high DM silage. In addition, application of ATCC14917 decreased the functions of carbohydrate metabolism and amino acid metabolism of the bacterial community in alfalfa silage before 60 d of ensiling. An effective inhibition on the growth of Aerococcus and Enterobacter was also observed in ATCC14917-treated-silage. Therefore, the bacteriocin-producing L. plantarum ATCC14917 has great potential as an inoculant to improve the quality, nutritive value, and safety of alfalfa silage.