Sweet Corn Stalk Treated with Saccharomyces Cerevisiae Alone or in Combination with Lactobacillus Plantarum: Nutritional Composition, Fermentation Traits and Aerobic Stability

Simple Summary The inclusion of Saccharomyces in the ration is beneficial to ruminants. We investigated the effects of inoculating a high-dose S. cerevisiae (108 cfu/g) on the nutritional composition and fermentation traits of sweet corn stalk. A high-dose S. cerevisiae inoculum increased the crude protein concentration of sweet corn stalk silage but decreased the silage quality. Thus, a high-dose S. cerevisiae inoculum is not conducive to obtaining high-quality corn stalk silage. Abstract This study examined the effects of a high-dose Saccharomyces cerevisiae inoculant alone or jointly with Lactobacillus plantarum on nutrient preservation, fermentation quality, and aerobic stability of sweet corn stalk silage. Fresh stalks (231 g dry matter (DM)/kg) were chopped and subjected to the following treatments: (1) deionized water (Uninoculated; U); (2) S. cerevisiae at 1 × 108 cfu/g of fresh forage (S); and (3) S. cerevisiae at 1 × 108 cfu/g plus L. plantarum at 1 × 105 cfu/g (SL). Treated stalks were ensiled in 5-litre laboratory silos for 30, 60, and 90 day. The S and SL silages had a greater (p < 0.001) pH and greater crude protein, ammonia nitrogen/total nitrogen, neutral detergent fibre, acid detergent fibre, and ethanol contents at all three ensiling periods than the U silage. Acetate, propionate and volatile fatty acids in the S and SL silages after 30 and 90 day of ensiling were greater (p < 0.05) than those in the U silage, but they were lower (p < 0.05) in the S and SL silages than in the U silage after 60 day. The lactate and V-score of the S and SL silages were lower (p < 0.001) than those of the U silage at all three ensiling periods. Compared with the U group, the aerobic stability of the S silage after 90 day of ensiling decreased (p < 0.05), and the aerobic stability of the SL silage was unaffected (p > 0.05). Overall, the quality of sweet corn stalk silage was not improved by inoculation with 108 cfu/g of S. cerevisiae alone or in combination with 1 × 105 cfu/g of L. plantarum.

height of 15-20 cm. The stalks were immediately chopped to an approximate length of 2 cm using a forage cutter (9Z-20, Wanying Machinery Equipment Co., Ltd., Zhengzhou, Henan Province, China). The chopped forage was treated with (1) deionized water (0.5 mL/kg of fresh forage) as the uninoculated control (U); (2) S. cerevisiae at 1 × 10 8 cfu/g of fresh forage (Procreatin-7, Lesaffre, Marcq-en-Baroeul, France; S); and (3) S. cerevisiae at 1 × 10 8 cfu/g and L. plantarum at 3 × 10 5 cfu/g (Taiwan Yaxin Biotechnology Co. Ltd., Taiwan, China; SL). Procreatin-7 is a commercially available active dry yeast (15 × 10 9 cfu/g of S. cerevisiae) powder. The application rate of S. cerevisiae was calculated according to the viable cells labelled in the instructions and sprinkled onto the forage with constant mixing by hand. Similarly, the theoretical inoculation amount of L. plantarum was diluted in sterile deionized water and applied at the rate of 0.5 mL/kg of fresh forage with a sprayer under constant manual mixing. Approximately 3.5 kg of treated forage was packed into 5-litre laboratory silos made of high-density polyethylene (19 cm in diameter × 27.5 cm in height) to achieve a packing density of 158 kg of dry matter (DM)/m 3 and sealed airtight with a layer of polyethylene membrane and a thick screw top. Eighteen replicates of each treatment were then stored in an enclosed warehouse at ambient temperature (25 ± 1 • C) for 30, 60 and 90 day.

Sampling and pH Determination
After 30, 60, and 90 day of ensiling, six silos from each treatment were randomly opened. The content in each silo was mixed thoroughly and sampled for analysis. Silage extract was prepared by adding 25 g of corn stalk silage to 225 mL deionized water, storing the sample for 24 h at 4 • C and then manually homogenizing the sample for 2 min according to the method described by Wang et al. [24]. The pH of the water extract was measured using a pH meter (FiveEasy Benchtop pH Meter FE20, Mettler-Toledo, Columbus, OH, USA). Then, the solution was filtered through Whatman 54 filter paper, and the filtered water extract was subdivided into 10 mL aliquots. Two aliquots were used directly for ammonia and water-soluble carbohydrates (WSCs) analysis, and three aliquots were frozen at −20 • C after adding 2 mL of 0.3 mol/L metaphosphoric acid for ethanol and organic acid analyses.

Determination of Chemical Composition
Approximately 300 g of silage samples in each silo were used in the determination of chemical composition, including neutral detergent fibre (NDF), acid detergent fibre (ADF), total nitrogen (TN), and starch. The DM of the silage samples was dried using a forced-air oven at 60 • C for 48 h and then ground with a pulveriser (FW-100, Beijing Yongguangmin Ltd., Beijing, China) prior to being passed through a 1-mm screen for further analysis. The DM concentration was corrected for the loss of volatile compounds according to Porter and Murray [25], and the variables after ensiling were presented on the basis of corrected DM. Crude protein (CP) was calculated by determining TN using a flow injection technology (AutoAnalyzer3, Flow Injection Analysis System, Seal Analytical, Norderstedt, Germany) according to the digestion procedure of the AOAC-2001.11 method (AOAC 2002) [26] and using a fixed conversion factor (6.25). Ether extract (EE) was also analysed using an automatic extraction apparatus (SOX416, C. Gerhardt GmbH & Co. KG, Königswinter, Germany) according to the AOAC 2003.05 method (AOAC 2006) [27]. The amounts of NDF and ADF were measured in a Fibretherm FT 12 Fiber Analyzer (C. Gerhardt GmbH & Co. KG) using the methods described by Mertens [28] with the inclusion of a heat-stable amylase and sodium sulphite. Both NDF and ADF were expressed inclusive of residual ash, and hemicellulose was calculated as the difference between NDF and ADF. Starch was analysed using the amyloglucosidase hydrolysis method as described by Wang et al. [29]. WSCs analysis based on the water extract of silage plant tissue homogenate was performed according to DuBois et al. [30].

Determination of Ammonia Nitrogen, Ethanol and Organic Acid Contents
After the frozen water extract was thawed, ammonia nitrogen (NH 3 -N) concentration was determined by the phenol-hypochlorite procedure as described by Chen et al. [31]. For the determination of ethanol and organic acid, the thawed frozen water extract was centrifuged at 15,000× g for 10 min at 4 • C, then the supernatant was filtered using a 0.22-µm syringe filter and transferred into a 2.0-mL glass chromatograph vial. Then, lactate was analysed by high-performance liquid chromatography (Agilent HPLC 1290, Agilent, Santa Clara, CA, USA) equipped with a Zorbax SB-C18 column (150 mm × 4.6 mm ID, 5 µm) and a diode array detector (G4212A, Agilent) and qualified by an external standard method with five standard solutions (0.01, 0.05, 0.10, 0.15 and 0.20 mg/mL). Mobile phase A was a 19.1 mmol/L H 3 PO 4 solution, and mobile phase B was HPLC grade acetonitrile. Both mobile phases were applied at a flow rate of 1 mL/min. The column temperature was maintained at 25 • C, and the detection wavelength was 210 nm. The system calibration and integrity were evaluated through the periodic injection of standard lactate solutions. Ethanol and volatile fatty acids (VFAs, including acetate, propionate and butyrate) were analysed using a gas chromatograph (Agilent 7890A, Agilent Inc.) equipped with a flame ionization detector and a DB-FFAP column (30 m × 0.25 mm × 0.2 µm), and the detailed procedures were described by Playne [32].

Background Microbial Population Analysis
Approximately 5 g of frozen fresh forage was ground in liquid nitrogen, and 100 mg subsamples were used for DNA extraction using a commercial kit (QIAamp fast DNA stool mini kit; QIAGEN GmbH, Hilden, North Rhine-Westphalia, Germany) according to the manufacturer's instructions. The concentration and purity of total DNA were measured using an ND-2000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). The extracted DNA was diluted with ultrapure water to 1 ng/µL, and PCRs were performed in triplicate. Each reaction was performed in a 20 µL mixture containing 0.8 µL of each primer (10 µM), 10 ng of template DNA, 2 µL of 2.5 mM dNTPs, 0.4 µL of FastPfu polymerase (TransGen Biotech Co., Ltd., Beijing, China) and 4 µL of 5 × FastPfu Buffer (TransGen Biotech Co., Ltd.). Specific primers containing an Illumina adaptor and barcode sequences were designed for the 16S rDNA V3-V4 hypervariable regions (341F, 5'-CCTAYGGGRBGCASCAG-3'; 806R, 5'-GGACTACNNGGGTATCTAAT-3') of the bacterial genomic DNA. The PCR products were excised from 2% agarose gels and purified using a QIAquick Gel extraction kit (QIAGEN GmbH). Amplicons from each reaction mixture were quantified fluorometrically, normalized and pooled at equimolar ratios based on the concentration of each amplicon. The amplicons were sequenced (paired-end) on an Illumina HiMiSeq PE300 platform by Novogene Company (Beijing, China). The 16S sequence reads were demultiplexed, the adaptors and barcodes were trimmed, and contigs were obtained based on the overlaps of pair-ended sequences using QIIME (v 1.7.0). The contigs were filtered according to quality using QIIME (v 1.7.0) as described by Caporaso et al. [33]. Chimeras were filtered using UCHIME v 11 against the Gold database [34]. The high-quality contigs were uniquified and clustered into operational taxonomic units (OTUs) at 97% similarity using USEARCH v.10 [35]. Representative sequences were defined based on the abundance of each OTU. A taxonomy analysis was performed using the SILVA database v.132 [36] with a minimum support threshold of 80%.

Aerobic Stability and Quality Assessment
After 90 day of ensiling, the silos were opened, and samples were collected. Two kg of silage was returned to the silo for the measurement of aerobic stability. Silages were covered with two layers of cheesecloth to prevent drying and contamination but to allow the penetration of air. Thermocouple probes (i500-E3TW, Yuhuanzhituo Instrumental Co., Ltd, Hangzhou, Zhejiang Province, China) were placed at the geometric centre of each sample to collect the temperature every 60 min; the ambient temperature was synchronously detected by another probe. Aerobic stability was denoted as the length of time that elapsed before the silage and ambient temperatures differed by more than 2 • C [37]. The V-score evaluation system [38] was applied to assess the quality score of the corn stalk silage based on the parameters of NH 3 -N and organic acids.

Data Analysis
Statistical analyses were performed using SPSS 19.0 statistical software (IBM Corp., Armonk, NY, USA, 2010). The linear mixed procedure was performed to examine the differences as follows: Y ijk = µ + T i + P j + T i × P j + e ijk , where Y ijk is the dependent variable, µ is the overall mean, T i is the fixed effect of the treatment, P j is the fixed effect of the ensiling period, T i × P j is the interaction between T i and P j , and e ijk is the random residual error. Aerobic stability data were analysed by the mixed model with the fixed effect of treatments. Differences among means were tested using the Bonferonni comparison [39], and significant differences were declared at p < 0.05. Data are presented as the means and standard error of the means (SEM).
A principal component analysis (PCA) was performed using SIMCA 14.0 software (Umetrics AB, Umeå, Sweden) based on the variables of the pH value, the chemical composition, organic acids, and ethanol. PCA-X and partial least squares-discriminant analysis (PLS-DA) were auto-fitted using seven-fold cross validations. The eigenvalue similarity level and significance level for DModX and Hotelling's T2 were both at 0.05. The variable importance in the projection (VIP) score in the PLS-DA model was analysed, and the "VIP scores greater than one" rule was generally used as the criterion for important variable selection [40], that is, a higher VIP score indicates a greater importance of this variable for explaining the difference among groups.

Chemical Composition and Microbial Population of Fresh Forage
The fresh stalk contained 247.1 g DM/kg as fed, 134.1 g/kg WSCs, 667.8 g/kg NDF, 426.3 g/kg ADF, 80.8 g/kg starch, 18.1 g/kg EE, and 72.3 g/kg CP on a DM basis. In the fresh stalk, the top five genera of the epiphytic microorganism population were Pseudomonas, Pantoea, Klebsiella, Raoultella, and Enterobacter (Table 1), which accounted for 73.18% of the total sequences. The majority of the population belonged to the phyla Proteobacteria, Bacteroidetes Firmicutes and Actinobacteria, which accounted for 86.96, 1.59, 0.81 and 0.29% of the total sequences, respectively. The epiphytic LAB included 0.35% Lactobacillus, 0.24% Lactococcus, 0.22% Weissella and a trace amount (0.018%) of Enterococcus.

Chemical Composition of Silage
The DM content in the SL group was lower (p < 0.05) than that in the U and S groups after 30 day of ensiling (Table 2), while it was similar between treatments after 30 and 60 day of ensiling. The concentrations of CP, NH 3 -N/TN, NDF and ADF in the S and SL groups during 90 day of ensiling increased (p < 0.001) compared with the U group, while the starch and WSCs concentrations in the S and SL groups decreased (p < 0.001) compared with those in the U group. Compared with the hemicellulose concentrations in the U group, the hemicellulose concentrations in the S and SL groups were unaffected (p > 0.05) at all three periods; only the concentration of hemicelluloses in the S group was lower than that in the SL group after 90 day of ensiling.

Fermentation Traits, Aerobic Stability and V-Score
The pH values in the S and SL silages were greater (p < 0.001) than those in the U group at all three ensiling periods (Table 3), and the pH values in the SL silages were highest (p < 0.001) at all three ensiling periods. The concentrations of acetate, propionate and VFAs in the S and SL groups were greater (p < 0.05) than those in the U group after 30 and 90 day of ensiling but lower (p < 0.05) in the S and SL groups than in the U group after 60 day of ensiling. The butyrate concentrations in the S and SL groups after 60 day of ensiling were lower (p < 0.05) than those in the U group, but they were higher in the SL group after 90 day of ensiling than in the U and S groups. However, the lactate concentration and lactate:acetate ratio in the S and SL groups after 30 and 90 day of ensiling were lower (p < 0.001) than those in the U group, while those values in the S and SL groups after 60 day of ensiling were higher (p < 0.001) than those in the U group. The ethanol concentrations in the S and SL silages were nine-to 29-fold greater (p < 0.001) than those of the U group at all three ensiling periods. The V-scores of the S and SL groups at all three ensiling periods were lower (p < 0.001) than those in the U group. Aerobic stability in the S group after 90 day of ensiling was lower (p < 0.05) than that in the U and SL groups. A-C Means within the same row followed by different lowercase superscript letters are significantly different (p < 0.05); a-c Means within the same column followed by different uppercase superscript letters are significantly different (p < 0.05); 1 U, no inoculant; S, Saccharomyces cerevisiae at 1 × 10 8 cfu/g; SL, S. cerevisiae at 1 × 10 8 cfu/g and Lactobacillus plantarum at 1 × 10 5 cfu/g. 2 SEM, standard error of means. 3 VFAs include acetate, propionate and butyrate.

Principal Component Analysis and VIP Score
PCA-X analysis indicated the difference between the uninoculated and inoculated groups ( Figure 1A, cumulative R 2 X = 0.798, cumulative Q 2 = 0.615), but no clear separation among the three ensiling periods was found. This separation was further verified by the PLS-DA model ( Figure 1B, cumulative R 2 Y = 0.705, cumulative Q 2 = 0.654) after a valid permutation test (Q 2 intercepts = −0.335), which showed the difference between the uninoculated and inoculated groups. According to the VIP score (>1) shown in Figure 2, the order of importance in fermentation variables was listed as follows: pH > NH 3 -N/TN > WSCs > CP > ethanol > lactate. The pH and NH 3 -N/TN (VIP score > 1.5) were the most important variables for explaining the differences among treatments.

Discussions
The present study indicated that the S. cerevisiae inoculant at a rate of 10 8 cfu/g alone or together with L. plantarum affected the pH, chemical composition, organic acid profile, and ethanol content of corn stalk silage and led to a lower V-score. In contrast to our hypothesis, the combined inoculation of high-dose S. cerevisiae and L. plantarum further raised the pH and NH 3 -N levels, which did not facilitate the storage of fresh stalk.
The chemical composition of fresh sweet corn stalk in this study was typical of low DM (<25 g/kg) and high WSCs (>130 g/kg), but the level of WSCs was slightly lower than that of another breed of sweet corn stalk (171 g/kg) [41]. The NDF, ADF and CP concentrations were similar to those of corn stover in previous studies [42]. The background microbial population detected in this study mostly comprised the phyla Proteobacteria, Firmicutes and Bacteroidetes, which is consistent with the background microflora in 10% bloom alfalfa [43] and in corn stover [44]. Lactobacillus, Lactococcus, Weissella, and Enterococcus were the only four LAB detected in this study and accounted for 0.83% of the total sequences, which was much lower than the relative abundance of background Lactobacillaceae (2.43%) from 16S rDNA sequence data reported by Romero et al. [45].
PCA analysis is an effective tool to identify discriminative metabolites or markers after experimental intervention and has recently been used to assess silage quality [46]. Although the VIP score of the detected variables did not provide any additional biological significance, it deepened our understanding of the experimental results and demonstrated that the pH and NH 3 -N/TN among the detected variables were the most influential parameters to differentiate whether the silage had been inoculated with S. cerevisiae. The NH 3 -N/TN concentration reflects the extent of organic nitrogen degradation or proteolysis during silage [47], which is usually lower than 10% for gramineous crop silage [48]. The NH 3 -N/TN concentration of the uninoculated corn stalk silage in this study was comparable to the data reported by Li et al. [42] for corn stover silage. However, the NH 3 -N/TN concentration in the inoculated silages in this study was increased by two-fold compared with the uninoculated silage. Nitrogen sources, including inorganic nitrogen and free amino acids, are degraded by S. cerevisiae to yield ammonia as a precursor nitrogen source for protein anabolism [22], which may explain the higher concentration of NH 3 -N/TN in the silages containing S. cerevisiae in the present study. Generally, the conversion between inorganic nitrogen and organic nitrogen in silage does not influence TN and thus, does not influence the CP concentration because CP is estimated from the TN concentration. However, the CP sources in the silages inoculated with S. cerevisiae differed from the uninoculated silage in that the protein was derived not only from plant proteins but also from the microbial proteins of the inoculants, especially from S. cerevisiae cells with a protein content between 40% and 60% [49]. Hence, the high application size of S. cerevisiae (32.5 g/kg DM) was estimated to provide silage with an additional 13 to 19 g/kg DM of microbial protein, which almost explained the increase in CP in the inoculated silages. Similarly, Kamphayae et al. [23] reported that the CP content of the fermented end-product increased in proportion to the inclusion of increased brewer's yeast during the short ensiling of a total mixed ration containing cassava pulp. However, it is not clear whether the increase in the CP concentration was accompanied by the proliferation of the S. cerevisiae population and the increase in true microbial protein.
Inoculation with S. cerevisiae alone or in combination with L. plantarum had little effect on the starch content, although significant differences occurred between the uninoculated silage and the silage inoculated with S. cerevisiae alone. This was due to the slower rate of degradation of the starch caused by the limited amount of amylase naturally secreted by S. cerevisiae [50]. However, S. cerevisiae and LAB, including L. plantarum, can effectively consume WSCs [51]; thus, the WSCs in the silages inoculated with high-dose S. cerevisiae alone or combined with L. plantarum in the present study were reduced. Structural carbohydrates, including hemicellulose, cellulose, and lignin, cannot be directly utilized by S. cerevisiae unless they are effectively saccharified [52]. The increase in NDF and ADF in the silages inoculated with S. cerevisiae could be largely due to the relative decrease in fermentable carbohydrates, including WSCs and starch, and DM loss. Unfortunately, we did not determine DM loss.
When forage with high moisture is ensiled, it is difficult to lower the pH value below the critical threshold of 3.8-4.2 [20,47] and inhibit the growth of undesired microorganisms such as butyric acid-producing Clostridium. Although the moisture content of fresh forage was up to 75% in this study, the pH value was below 3.8 in the control silages during 90 day of ensiling, which is indicative of a favourable ensiling process. However, the pH values of silages inoculated with S. cerevisiae alone or jointly with L. plantarum ranged from 4.3 to 4.6 during 90 day of ensiling. Many studies have confirmed that the pH value in silage is primarily dependent on lactic acid concentration and is partially affected by the VFA concentration. However, the concentrations of lactate and VFAs in the inoculated silages were inadequate in explaining the high pH. We postulated that the pH in the inoculated silages was mainly affected by the high levels of NH 3 -N/TN. Because ammonia is easily dissolved in water, the ammonia solution is weakly alkaline (the pKb for ammonia is 4.74), which counteracts the role of lactate in reducing the pH in the inoculated silages. However, Duniere et al. [18] reported no effect on pH value and NH 3 -N when S. cerevisiae was applied at a dose of 10 3 cfu/g of fresh forage. This inconsistency is possibly due to the difference in application dose of S. cerevisiae. Additional supportive evidence is provided by Kamphayae et al. [23], who observed an increase in pH when the proportion of liquid brewer's yeast included in the mixed silage of cassava pulp and rice straw increased. Thus, for a high inoculation dose of S. cerevisiae, a method by which the ammonia production can be reduced in silage must first be identified.
Previously, inoculation with S. cerevisiae had no impact on the fermentation products (acetate, propionate, lactate, succinic acid, ethanol, etc.) of corn silage at the application rate of 10 3 to 10 5 cfu/g [18,19], but inoculation with S. cerevisiae at the dose of 10 8 cfu/g in corn stalk silages affected the ethanol and organic acid contents in the present study. High concentrations of ethanol in silages (>3-4%) are often associated with high numbers of yeasts [48], and the increase in the ethanol levels of the inoculated silages in this study reflected the active anaerobic metabolism of S. cerevisiae, suggesting that the survival or even proliferation of S. cerevisiae occurred during ensiling in this study, which was confirmed by 16S rDNA sequencing analysis in previous studies [18,19]. A high concentration of ethanol could lower the palatability of silage and affect milk flavour [48]; however, this weakness is not an issue if the ensiling of sweet corn stalk is used as a pre-treatment process for biomass ethanol, and this notion was tested with L. plantarum A1 with ferulic acid esterase activity in dry corn stover [42]. In addition, the acetate and propionate concentrations in the inoculant silages after 30 and 90 day of ensilage increased, which may be attributed to the high pH that cannot inhibit the metabolism of facultative aerobic bacteria, such as epiphytic Pantoea, Klebsiella, and Enterobacter. However, the reason for the abrupt decline in acetate and propionate in inoculated silages on 60 day of ensiling remains unclear and needs to be further verified. In contrast to the uninoculated silage, lactate was reduced in the silages inoculated with S. cerevisiae. Despite the fact that S. cerevisiae cannot utilize lactate [20], S. cerevisiae stimulates the growth of lactate-utilizing bacteria [51], which may partly explain the decrease in lactate. On the other hand, S. cerevisiae competes for the utilization of WSC substrates with LAB, which may limit lactate production by LAB. Notably, compared with the uninoculated silage, S. cerevisiae inoculant had no obvious effect on butyrate production during the first 60 day of ensiling; the butyrate increased only in the silage inoculated with the combined inoculants of S. cerevisiae and L. plantarum after 90 day of storage, which partly impaired the fermentation quality.
Aerobic stability is one of the key traits of silage preparation. After opening the silos, the elapsed time for aerobic deterioration depends on the initial microbial compositions [41], lactate concentration and pH [53][54][55], and higher amounts of VFAs (acetate, propionate, and butyrate) and NH 3 -N inhibit the growth of aerobic bacteria, yeasts, and moulds [48]. The reduced aerobic stability of the silage inoculated with S. cerevisiae alone was mainly due to the high pH value (4.3-4.4) that could not inhibit the growth of aerobic spoilage bacteria after opening. This result is consistent with the outcome of a high pH and poor stability observed by Kamphayae et al. [23] when a high proportion of liquid brewer's yeast was included. In contrast, a lower inoculation dose (10 3 -10 5 cfu/g) of S. cerevisiae did not affect the pH and aerobic stability of corn silage [18,19]. Therefore, poor aerobic stability is directly related to the high pH associated with NH 3 -N levels. However, aerobic stability was not influenced by the joint inoculation of S. cerevisiae and L. plantarum compared with the control in our study. The greater levels of propionate and butyrate may prolong the aerobic stability because propionate and butyrate have been accepted to have strong antifungal characteristics [48].

Conclusions
This research provides a new exploration of high-dose S. cerevisiae as an inoculant to ensile fresh forage. At 10 8 cfu/g, although the S. cerevisiae inoculant increased the CP concentration of the corn stalk silage, it did not improve the quality of the corn stalk silage, as the silage had a high pH and a high concentration of NH 3 -N/TN, irrespective of L. plantarum. Further studies should optimize the inoculum size of S. cerevisiae in silage and preventative measures to inhibit ammonia production.