Impact of molasses and microbial inoculants on fermentation quality, aerobic stability, and bacterial and fungal microbiomes of barley silage

This study aimed to investigate the effects of microbial inoculants (L) and molasses (M) on the bacterial and fungal microbiomes of barley silage after the aerobic stage. The addition of molasses and microbial inoculants improved the aerobic stability of barley silage. The ML silage, which had a low pH value and high lactic and acetic acid contents, remained aerobically stable for more than 216 h. The ML silage exhibited low bacterial and high fungal diversities. Microbial inoculants and molasses enriched the abundance of Lactobacillus in silage after aerobic exposure. The enrichment of L. buchneri was significant in ML silage at days 5 and 7 during the aerobic stage. The abundance of harmful microorganisms, such as aerobic bacterial including Acinetobacter, Providencia, Bacillus, and yeasts including Issatchenkia, Candida, and Kazachstania, were suppressed in ML silage. M and L had an impact on bacterial and fungal microbes, resulting in the improvement of fermentation quality and reduction of aerobic spoilage in barley silage.


Results
Effect of ML on quality of barley silage during aerobic stage. The pH values of the control (CK), M silage and ML silage were within the range of 3.87-3.94 after 60 days of ensiling (Table 1). When exposed to air, the pH values of the CK and M silages increased rapidly with prolonged exposure but not of the ML silage. Changes in pH values were relatively constant in barley silage prepared with ML from day 0 to 7 of aerobic exposure. Similarly, the concentration of fermentation products, such as LA and AA, decreased gradually with prolonged exposure in the CK and M silage but not in the ML silage. Changes in the concentrations of fermentation products were significant during aerobic exposure of barley silage (p < 0.05). LA and AA contents were higher in the ML silage than the CK and M silages at days 2, 5, and 7 of aerobic exposure (p < 0.05). In addition, as shown in Fig. 1, the ML silage was aerobically stable for more than 216 h, whereas the CK and M groups were stable for only 85 h and 96 h, respectively, indicating that the aerobic stability of barley silage was improved by the addition of ML. The single addition of M minimally improved aerobic stability of silage. Although yeast and AB counts increased gradually with prolonged aerobic exposure, the LAB counts did not significantly decrease in ML silage ( Table 2). Compared with the CK silage, the counts of LAB, yeast, and AB in barley silage prepared with M only    www.nature.com/scientificreports www.nature.com/scientificreports/ changed significantly after 5 and 7 d of exposure (p < 0.05). Nevertheless, the yeast and AB counts of the ML group were significantly lower than those of the CK and M groups during aerobic exposure (p < 0.05).
Effect of ML on bacterial and fungal diversity in barley silage during aerobic stage. In this study, the operational taxonomic units and Chao 1 indices, which represented the richness of bacterial communities, had notably decreased in barley silage treated with ML at 2 and 5 d during aerobic stage (Table 3). Furthermore, the Shannon index of bacteria was not affected by M treatment only at 5 and 7 d during aerobic stage in barley silage, as compared with the CK silage. However, the Shannon index was lower in ML silage than that in the CK and M silages at 5 and 7 d during aerobic stage, suggesting that there was less bacterial diversity of barley silage in the ML silage than that in the CK and M silages.
Among the three groups, the operational taxonomic units, Chao 1, and Shannon indices of fungi were the highest in ML silage after 7d of exposure. Nevertheless, these indices decreased in the ML silage during aerobic stage. These findings show that the richness and diversity of fungal microbes had increased in barley silage treated with ML after aerobic exposure.
Effect of ML on bacterial microbiomes in barley silage during aerobic stage. Firmicutes was the most abundant phyla with more than 97% relative abundance in ML silage during aerobic stage ( Fig. 2A). However, the abundance of Firmicutes was decreased, whereas that of Proteobacteria was increased, with prolonged exposure in CK and M silages. Among the three groups, CK silage had the highest and the lowest abundances of Proteobacteria and Firmicutes, while the ML silage had the lowest and the highest abundances of Proteobacteria and Firmicutes at 5 and 7 d during the aerobic stage, respectively. In the CK, M, and ML groups, the abundances of Firmicutes at 7d of aerobic exposure were 1.544%, 26.291%, and 97.325%, respectively.
Changes at the genus level in bacterial community structures among the CK, M and ML silages during aerobic stage was summarized in Fig. 2B. Similar to Firmicutes at the phylum level, Lactobacillus was the dominant bacterial genus with a relative abundance of more than 95% in ML silage after aerobic exposure. Meanwhile, the abundance of Lactobacillus was high in the CK and M silages at 0 and 2d during aerobic stage. In CK silage at 5 d during aerobic stage, the dominant bacterial genera were Enterococcus, Acinetobacter, Providencia, and Empedobacter with relative abundances of 32.864%, 26.401%, 16.509%, and 13.024%, respectively, while that of Lactobacillus was only 0.708%. Furthermore, in M silage at 5 d during aerobic stage, Lactobacillus, Providencia, Enterococcus, Acinetobacter, Empedobacter, and Bacillus were the dominant bacterial genera with the abundances of 48.463%, 19.687%, 12.636%, 7.143%, 6.091%, and 4.755%, respectively. After 7 d of aerobic exposure, Acinetobacter was the prominent genus at 50.226% relative abundance in CK silage, which was accompanied by the emergence of Sphingobacterium with an abundance of up to 17.041% and a decrease in Providencia abundance to 5.717%. Meanwhile, in M silage, Lactobacillus abundance decreased from 48.463% at day 5 of exposure to 0.187% at day 7 of exposure. The abundances of Providencia and Stenotrophomonas in M silage increased to 38.866% and 17.031% at 7 d during aerobic stage, respectively, suggesting that they were the dominant genera.   www.nature.com/scientificreports www.nature.com/scientificreports/ Differences in bacterial community structures among the groups were identified by linear discriminant analysis (LDA) of effect size (LEfSe), which was used to calculate the relative abundances of bacterial genera in barley silage at 5 and 7 d during aerobic stage (Fig. 2C,D). LEfSe was used to identify the taxa that most likely explain the differences in bacterial community structures among the CK, M, and ML groups. The results revealed significant differences in LDA scores among the three groups. Enrichment of L. buchneri was significant in ML silage at 5 and 7 d during aerobic stage, suggesting that L. buchneri is a potential indicator of aerobic stability. Moreover, enrichment of Bacillus was significant in M silage at 5 and 7 d during aerobic stage. For Empedobacter, enrichment was significant in CK silage at 7 d during aerobic exposure, whereas for Enterococcaceae and Acinetobacter, enrichment was significant in CK silage at 5 d during aerobic exposure.
The results of correlation analysis of bacterial communities, silage fermentation, and microbial counts of barley silage during aerobic stage are shown in Fig. 3A. When exposed to air for 7 days, Lactobacillus, Weissella, and Pediococcus were positively correlated with LA, AA, and PA concentrations, and LAB counts but negatively correlated with pH values and AB. Acinetobacter was positively correlated with pH values, as well as yeast and AB counts, but negatively correlated with LA concentrations and LAB counts. Stenotrophomonas and Empedobacter were positively correlated with pH values, as well as yeast and AB counts, but negatively correlated with LA and PA concentrations, and LAB counts. Figure 3B shows a heatmap of microbial function pathways. A hierarchical cluster heatmap was generated to visualize the distribution of microbial communities in the CK, M, and ML groups. The heatmap results suggest that alanine aspartate and glutamate metabolism, pyruvate metabolism, amino sugar and nucleotide sugar metabolism, starch and sucrose metabolism, glycolysis, purine metabolism, pyrimidine metabolism, and aminoacyl tRNA biosynthesis, which are involved in amino acid metabolism, carbohydrate metabolism, nucleotide metabolism, and translation, respectively, were more abundant in the ML group than in the CK group. As compared with the ML group, the microbiota in the CK group had higher functional performance related to bacterial secretion systems, bacterial motility proteins, two-component systems, and adenosine triphosphate binding cassette transporters, which are involved in environmental information processing.

Effect of ML on fungal microbiomes in barley silage after aerobic stage. Fungal community com-
position at the phylum level in ML silage after aerobic stage was significantly different with CK silage (Fig. 4A). The members of Ascomycota phylum were more abundant in M and CK silages with an abundance of more than 97% during the aerobic stage. Basidiomycota in the ML silage was the preponderant fungi at the early stage of exposure, with abundances of 62.562% and 78.714% at 0 and 2 d, respectively, during the aerobic stage. Ascomycota was the most abundant group in the ML silage at 5 d of aerobic exposure. Figure 4B shows the differences in fungal community structures at the genus level among barley silages after the aerobic stage. The predominant fungal genera in the CK silage at 2 d during the aerobic stage were Issatchenkia, Kazachstania, and Candida with abundances of 80.580%, 12.055%, and 4.449% respectively. Meanwhile, Issatchenkia, Monascus, and Kazachstania were the major fungi in M treated silage at 2 d during aerobic stage with the abundances of 55.310%, 35.024%, and 5.487%, respectively. Compared with CK silage, the addition of M decreased the abundance of Issatchenkia and increased the abundance of Monascus and Candida at 0 and 2 d during aerobic stage. Furthermore, the abundances of Issatchenkia, Candida, and Kazachstania decreased in the ML silage after aerobic exposure, as compared with those of the CK and M silages. However, other genera (less than 1% sequenced) were the predominant fungi in ML silage at 2 d during aerobic stage at 82.5% abundance, followed by Issatchenkia (8.95%). This result was in accordance with fungal diversity of barley silage, which showed that ML silage had the greatest fungal diversity at 2 d during aerobic stage among the three groups, as shown in Table 3. Differences in www.nature.com/scientificreports www.nature.com/scientificreports/ fungal community structures among the groups was investigated using LEfSe analysis, which was used to calculate the relative abundance of fungi in barley silage at 0 and 2 d during aerobic stage. As shown in Fig. 4C,D, there were significant differences in fungal community structures between the ML and CK silages, which confirmed that significant enrichment of Sporidiobolus was a indicator of aerobic stability in the ML silage at 0 d during aerobic stage, although the abundance of Kazachstania in CK silage was significantly enriched at 2 d during aerobic stage. The relationship between the fungal community structure and fermentation quality (Fig. 5) was determined by canonical correlation analysis. Issatchenkia was positively correlated with pH values and yeast counts, but negatively correlated with LA concentrations and LAB counts. Meanwhile, Sporidiobolus was negatively correlated with pH values. The addition of M combined with microbial inoculants might have decreased the abundance of  www.nature.com/scientificreports www.nature.com/scientificreports/ Issatchenkia, increased the abundance of others fungi (less than 1% sequence), decreased the pH value and content of yeast, and increased the contents LA and LAB of barley silage after aerobic exposure.

Discussion
The pH value was one of the main factors that influenced the extent of fermentation and quality of silage. A low pH value of 3.94 or less at the terminal point of silage indicates good fermentation during ensiling and aerobic exposure. The decline in microbial growth, which may favor the growth of undesirable anaerobic organisms and subsequently decrease aerobic stability, was effectively inhibited in silage under low pH conditions 12,13 . Consistently, in this study, the pH values of barley silage prepared with ML were low, suggesting ML silage has high aerobic stability. During aerobic exposure, the pH values were always less than 4.0 in ML silage, while the pH value reached 4.85 in barley silage added with inoculants at day 7 of aerobic exposure 3 . The yeast and AB counts of the ML group were significantly lower than those of the CK and M groups during aerobic exposure. Meanwhile, the counts of yeast and AB in ML silage were also less than those of barley silage added with inoculants which reported by Liu et al. 3 at day 5 and 7 of aerobic exposure. Yeasts, as primary initiators of aerobic spoilage in silage, assimilate lactate and degrade it into CO 2 and water, which is accompanied with nutrient loss in silage exposed to air. As a result of lactate depletion by yeasts, pH increases, rendering silage more susceptible to spoilage 14 . Compare to LA, it is easier for yeasts to be inhibited by AA 15 . According to the report by Danner et al. 16 , silages with high concentrations of AA have low populations of yeasts and molds and high aerobic stability. Barley silage treated with ML had the highest AA content during the aerobic stage in this study. The lowest yeast population was observed at 7 d of aerobic exposure, indicating that the aerobic stability of ML silage was the highest among the groups. However, these results do not conform to those of Kim et al. 10 , who reported that silages treated with M and L. plantarum had the lowest AA content, the highest populations of yeasts, and the highest aerobic stabilities during the ensiling of silage.
After aerobic exposure, the diversity and richness of bacteria in ML silage were lower than that in CK and M silages, which might be attributed to the predominant roles of L. plantarum. This result is supported by the findings of a previous study conducted by Méndez-García et al. 17 , who indicated that limited microbial diversity was due to low pH values in acidic environments. A similar result was reported by Yang et al. 18 . Lower bacterial diversity and greater aerobic stability were also observed in silages with L. buchneri and PA than those of CK silages 19 . However, bacterial diversity did not decrease in M silage, as compared with CK silage, but was evidently reduced in ML silage during aerobic exposure in our study. Together, these results show that the richness and diversity of microbial communities in silage were affected by the addition of ML during aerobic exposure.
Firmicutes are gram-positive bacteria with genomes having low G + C contents and most degrade macromolecular compounds, such as cellulose, protein, and starch 20

. McGarvey et al. reported that Firmicutes and
Proteobacteria were the dominant bacteria in alfalfa silage 21 . Proteobacteria is the largest phylum of bacteria and includes pathogenic bacteria, such as Escherichia coli, as well as Vibrio and Helicobacter spp., in addition to Erwinia, Sphingomonas, Methylobacter, Pseudomonas, and Agrobacterium spp. 22 . Romero et al. indicated a relatively high abundance of Firmicutes (99.8%) and a low abundance for Proteobacteria (0.07%) in wilted whole-crop oats ensiled for 217 days 20 . In this experiment, the bacterial community structure in silages changed substantially with prolonged exposure. Similar to the findings of Peng et al. 22 , in this study, the abundance of Firmicutes was significantly decreased, whereas that of Proteobacteria increased at 7 d during the aerobic stage in the CK and M silages. During aerobic exposure, the dominant phylum in ML silage was Firmicutes, with the abundance of >97%, which was significantly higher than that of barley silage added with inoculants. According to the findings of Liu et al. 3 , the dominant phylum in barley silage added with inoculants was Firmicutes, with the abundance of >99% at day 0 and 2 of aerobic exposure. However, Firmicutes and Proteobacteria were dominant in barley silage added with inoculants, with the abundance of 78.9% and 23.0% at day 5 of aerobic exposure, 20.8% and 69.2% at day 7 of aerobic exposure, respectively. According to a report by Xu et al. 23 , the abundance of Firmicutes predominated in spoiled silage after aerobic exposure, but was subsequently replaced by Proteobacteria and Actinobacteria.
Lactobacillus prevented aerobic spoilage due to the production of LA and the reduction of pH value 24 . In this study, the abundance of Lactobacillus predominated in ML silage throughout aerobic exposure, reaching more than 95%. At day 0 and 2 of aerobic exposure, Lactobacillus was dominant in CK silage and M silage, with the abundance of >94%. However, the abundance of Lactobacillus in M silage was only 48.2% and 1.2% at day 5 and 7 of aerobic exposure, respectively. Liu et al. 3 reported that the abundance of lactobacillus for dominant genera in barley silage with inoculants kept more than 99% at day 0 and day 2 of aerobic exposure, but the abundance decreased to 71.4% and 11.6% at day 5 and 7 of aerobic exposure, respectively. Therefore, the addition of molasses combined with microbial inoculants could improve the structure of bacteria community which was predominated by Lactobacillus in barley silage during aerobic exposure. Pang et al. 25 found that Lactobacillus was the prominent genus in sorghum, forage paddy rice, and alfalfa silages. M provides additional fermentable substrates for LAB, which promotes domination in the microbial communities of silage, thereby directing the metabolism to homofermentative LAB 26 . Li et al. 27 reported that AA and PA might exert direct effects on bacterial physiology and indirect effects on the microbial community structure. In this study, the AA and PA concentrations in the ML group might be sufficiently high to effectively inhibit the growth of aerobic microorganisms that cause spoilage. With the addition of ML during the ensiling of barley silage, Lactobacillus, as beneficial bacteria, became the absolute dominant genus and competitively inhibited the growth of harmful bacteria, such as Empedobacter and Acinetobacter. The low bacterial diversity observed in ML silage, as indicated by the low Shannon index of 1.518, was a result of the high abundance of Lactobacillus species at 95.458% and low pH value (3.98) of barely silage after 7 d of exposure. Lactobacillus buchneri is a heterofermentative species that produces a variety of products and ferments LA into AA 2,28,29 . The conversion of LA to AA results in high concentrations of AA, which is regarded as an inhibitor of yeasts and molds. In this study, at 5 and 7 d during aerobic stage, LA and AA concentrations were higher in the ML silage than the CK and M silages. Hence, barley silage ensiling with ML had the best aerobic stability among the three groups. This result is possibly due to L. buchneri from ML silage, which plays an important role in pH, fermentation products, and microbial community dynamics during aerobic exposure.
Graf et al. 30 deduced that the proliferation of bacilli normally occurs in the later period of aerobic spoilage, as indicated by the detection of Bacillus spores at the outermost layers of grass and maize silages. According to Liu et al. 31 , Stenotrophomonas possibly caused aerobic deterioration of corn straw silage. Acinetobacter are aerobic non-fermenting bacteria that exist in various environments except silage 32 . According to Liu et al. 31 , AB in corn straw silage upon aerobic exposure mainly comprises Acinetobacter, Stenotrophomonas, and Bacillus, which possibly cause aerobic deterioration of corn straw silage. In the present study, these aerobic bacteria were positively correlated with pH values, as well as yeast and AB counts, and promoted the growth of spoilage microorganisms.
As compared with the ML group, the microbiota in the CK group had higher functional performance related to bacterial secretion systems, bacterial motility proteins, two-component systems, and adenosine triphosphate binding cassette transporters, which are involved in environmental information processing. This result is in agreement with those of Keshri et al. 33 , who reported that silage treated with L. buchneri, which was stable after aerobic exposure, differed from untreated silage at day 90 in terms of three pathways; namely, base-excision repair, pyruvate metabolism, and transcription machinery. Notably, a high degree of pyruvate metabolism was also observed in LAB-supplement barley silage, suggesting a possible reason for the enhancement of silage stability 3 . Amino acids were metabolized by the dominant microflora in L. plantarum supplement and untreated silages during the ensiling 33 . The addition of L. buchneri increased the accumulation of polyols and free amino acids, which was possibly due to the unique metabolic pathways of sugar fermentation and amino acid biosynthesis executed by L. buchneri 34 . In the present study, the changes in the carbohydrate and amino acid metabolic pathways may have contributed to the inhibition of aerobic spoilage in ML silage. As compared with the CK silage, the higher abundances of carbohydrates and amino acids were responsible for the improved aerobic stability of LAB-inoculated barley silages 3 .
The members of Ascomycota phylum were more abundant in M and CK silages with an abundance of more than 97% during the aerobic stage. These results were similar to the findings of Romero et al. 20 , who reported that Ascomycota was the most abundant group, and Basidiomycota was the second most abundant in oat silage. In our study, Basidiomycota in the ML silage was the preponderant fungi at the early stage of exposure during the aerobic stage. Ascomycota was the most abundant group in the ML silage at 5 d of aerobic exposure, which was accompanied with a decrease in the abundance of Basidiomycota, suggesting that the fungal community structure was significantly modified by ML treatment at the early stage of aerobic exposure. Yeasts, such as Issatchenkia, Candida, and Saccharomyces spp, are considered basic initiators of aerobic spoilage in barley silage 35 . The abundances of Issatchenkia, Candida, and Kazachstania decreased in the ML silage after aerobic exposure, as compared with those of the CK and M silages, suggesting the addition of ML affected fungal community composition. Issatchenkia was the main microorganism causing aerobic deterioration of corn and barley silages 35,36 . Liu et al. 3 , indicated the addition of inoculants could decrease the abundances of Issatchenkia at day 0 of aerobic exposure, but not affect the abundances of Issatchenkia at day 2, 5 and 7 of aerobic exposure. Our results suggested the addition of molasses combined with microbial inoculants could affect fungal community composition and decrease the abundances of Issatchenkia in barley silage, however, single addition of inoculants had little effect on fungal community composition during aerobic exposure. The abundance of harmful microorganisms, such as aerobic bacteria including Acinetobacter, Providencia, Bacillus, and yeasts including Issatchenkia, Candida, and Kazachstania, were inhibited in the ML silage. Sporidiobolus can be exploited for biotechnological production of β-carotene using conventional fed-batch fermentation 37 . Sporidiobolus and Sporobolomyces produce carotenoids, which are effective antioxidants having potential antimicrobial properties 38 . In addition, similar results reported by Dolci et al. showed that the dominant yeast species in corn silage was Kazachstania exigua after exposure to air, as determined by denaturing gradient gel electrophoresis profiles of fungal DNA and RNA 39 . According to Lu et al., Kazachstania unispora was cardinal among the dominant species of yeasts in aerobically deteriorating corn silage 40 . conclusion ML silages remained aerobically stable for more than 216 h with stable pH values and LA contents. Bacterial diversity was reduced, while fungal diversity was increased in the ML silage during aerobic exposure. The abundance of Lactobacillus was enriched, while that of harmful microorganisms, such as aerobic bacterial including Acinetobacter, Providencia, Bacillus, and yeasts including Issatchenkia, Candida, and Kazachstania, were lowest in the ML silage. The addition of ML could increase AA content, reduce yeast and AB counts, change bacterial and fungal microbes, improve fermentation quality, and reduce aerobic spoilage of barley silage.

Methods
preparation of barley silage. Barley was harvested from a farm located in Yancheng, Jiangsu, China (33°19′ N, 120°45′ E) in May 2017. Barley was chopped using fully automatic harvester. We used inoculants and molasses as additives for barley silage. The inoculants were prepared by our laboratory and comprised L. plantarum, L. casei, and L. buchneri (2:1:1) 3 , which were isolated form ryegrass silage, rice straw silage, corn silage, respectively. The NCBI accession number of inoculants were MK106013, MK106014, MK106015, respectively. The combined inoculants were added at 5 × 10 5 colony forming units (CFU) g −1 of fresh material (FM). The chopped barley was mixed and divided into equal portions for three treatments comprising no additive (CK group), 2 g of M per 100 g of FM (M group), or 2 g of M per 100 g of FM + inoculants (ML group). After thorough mixing, barley was ensiled in 1 L silos.
After 60 days of ensiling, silos were opened. Fermentation quality, microbial counts, chemical compositions, bacterial and fungal community structures were analyzed at 0, 2, 5, and 7 d during aerobic exposure. The aerobic stability was determined using a previously described method 41 .

Scientific RepoRtS |
(2020) 10:5342 | https://doi.org/10.1038/s41598-020-62290-7 www.nature.com/scientificreports www.nature.com/scientificreports/ fermentation quality and chemical composition. The pH value and organic acid concentrations were determined by homogenizing 20 g of silage in 180 mL of sterilized water at 4 °C for 24 h and then filtrating the homogenate through four layers of cheesecloth. The pH of the filtrate was measured using a glass electrode pH meter (Mettler-Toledo AG, Zurich, Switzerland). LA, acetic acid (AA), and propionic acid (PA) and ethanol were determined by high performance liquid chromatography (Agilent 1260; Agilent Technologies, Santa Clara, CA, USA). The analytical conditions were as follows: column, Carbomix ® H-NP5 (Sepax Technologies, Inc., Newark, DE, USA); oven temperature, 50 °C; mobile phase, 2.5-mM H 2 SO 4 ; flow rate, 0.5 mL/min). Approximately 100 g of each sample was dried at 55 °C for 48 h and weighed to determine the dry matter content of barley silage. Both neutral and acid detergent fibers were analyzed using the method described by Van Soest et al. 42 .
Microbial counts. Twenty grams of each sample in 180 mL of physiological saline was blended and serially diluted. Microbial counts of LAB, yeasts, and aerobic bacteria (AB) were measured according to the study of Liu et al. 3 . The culture medium used were lactobacilli de Man, Rogosa, Sharpe agar (CM361; Oxoid Ltd.), potato dextrose agar (CM0139; Oxoid Ltd.), and nutrient agar (CM1160; Oxoid Ltd.). Lactic acid bacteria were incubated at 30 °C for 48 h under anaerobic conditions, whereas yeasts and aerobic bacteria were incubated at 30 °C for 24 h. Colonies were counted in CFU/g FM −1 as viable numbers of microorganisms.
Bacterial and fungal community. Total DNA was extracted using the PowerSoil ® DNA Isolation Kit (MO BIO Laboratories, Carlsbad, CA, USA) in accordance with the manufacturer's protocol. The V3-V4 region of bacterial 16S rDNA and the fungal internal transcribed spacer 1 region were amplified, sequenced, and analyzed according to the methods described by Liu et al. and Zhang et al. 3,43 . The raw reads were deposited in the European Nucleotide Archive (ENA) database under the accession number ERP118410 and ERP118523.
Statistical analysis. Fermentation quality, microbial counts, nutritional value were analyzed by two-way analysis of variance for a 3 × 4 (treatment × storage period)-factorial arrangement of treatments and compared with Tukey's test using IBM SPSS Statistics for Windows, version 19.0. (IBM Corporation, Armonk, NY, USA). A probability (p) value of <0.05 was considered statistically significant.