Introduction

Forage preservation via ensiling has become a global practice because it provides a consistent, reliable, and predictable feed supply for ruminant production systems. Unavoidable losses of highly digestible nutrients caused by plant respiration, plant microbial proteolytic activity, clostridial fermentation, microbial deamination, and decarboxylation of amino acids may negatively affect conservation efficiency, increase energy and nutrient losses, and cause an accumulation of anti-nutritional compounds in silage1.

Alfalfa is a forage crop of great importance due to its worldwide use, high nutritional value and digestibility2. However, high concentrations of organic acids, salts, proteins, and minerals result in a high buffering capacity3. The high buffering capacity and CP, in combination with low water-soluble carbohydrates (WSC) concentrations, indicate that the ensiling properties of alfalfa are not ideal, as suggested by Muck4.

Therefore, the use of microbial inoculants as starters for alfalfa silage is recommended5. Zielińska et al.6 demonstrated that microbial inoculants altered many parameters of silages, but the strength of the effects on fermentation depended on the strain characteristics. One of the main challenges in the industry is the extent of variability in the effects of inoculant bacteria on the fermentation and preservation of silage, silage quality and animal performance, which were noted in several studies5,7. The lack of inoculant effects on the process may be related to the ability of the inoculated bacterium to grow rapidly in the forage mass and effectively compete with the epiphytic flora as well as the presence of adequate substrate, and it may also be related to specificities between the forage, the microorganisms present in the inoculant and the weather conditions8.

Muck9 reported that the major international companies producing inoculants are based in Europe and North America. Therefore, these products have been developed for cool-season grasses, whole-crop corn and alfalfa. The inoculants may or may not be effective when used on warm season grasses or tropical legumes, which suggests that environmental conditions affect the physiology and metabolism of the inoculated strains and may influence their effects on the fermentation process.

Oliveira et al.1 reported that Lactobacillus plantarum is the most commonly used silage inoculant. However, some lactic acid bacteria (LAB) species are also selected as silage inoculants because of their faster growth at high pH values (>5) compared to L. plantarum. The authors suggested that more research was needed on the effects of infrequently used LAB as individual silage inoculants on silage fermentation because little is known about their related effects on silage quality.

Inoculants containing synergistic mixtures of LAB are used via the addition of microorganisms that act during different phases of fermentation. Some Pediococcus strains are more tolerant to high dry matter (DM) conditions than Lactobacillus spp. and exhibit a wider range of optimal temperatures and pH values for growth10. Silages treated with one or more bacteria often have a lower pH value and acetic acid, butyric acid, and ammonia nitrogen (NH3-N) contents and also a higher lactic acid content and better DM recovery compared to untreated silages5.

Based on that, in the search for new promising strains for silage inoculants, the purpose of the present study was to investigate the effects of L. pentosus 14.7SE, L. plantarum 3.7E, P. pentosaceus 14.15SE and a mixture of L. plantarum 3.7E and P. pentosaceus 14.15SE on the chemical composition, fermentative profiles and in vitro DM digestibility of alfalfa silage under tropical conditions after 1, 3, 7, 14, 28 and 56 days of fermentation.

Methods

Location and climatic conditions

The experiment was performed between June and August 2016 at the Department of Animal Science of the Federal University of Vicosa (Viçosa, MG, Brazil), located at 20°45′S latitude, 42°52′W longitude 648 m above sea level. The annual precipitation and average temperature the year of the experiment were 1235.4 mm and 20.7 °C, respectively.

Experimental design

The experiment was performed using a completely randomized design (with three replicates) based on a 6 × 6 factorial assay (6 inoculants × 6 fermentation periods). The periods (P) were 1, 3, 7, 14, 28, and 56 days after fermentation. The following inoculants (I) were evaluated: 1- Control (CTRL); 2- Commercial inoculant (CI); 3- Lactobacillus pentosus 14.7SE (LPE); 4- Lactobacillus plantarum 3.7E (LP); 5- Pediococcus pentosaceus 14.15SE (PP); and 6- L. plantarum 3.7E + P. pentosaceus 14.15SE (LP + PP). The commercial inoculant Silobac (CHR Hansen’s®, Hørsholm, Denmark), which contains L. plantarum, P. pentosaceus, maltodextrin, sodium aluminosilicate and whey, was used to compare its effectiveness with the new strains.

Characterization of the inoculants

The three wild strains of LAB used in this study belong to the microorganism bank of the Forage Laboratory of the UFV and were isolated from wilted and non-wilted alfalfa silages. The sequences of the strains are deposited in the GenBank database with the following access numbers: L. pentosus 14.7SE - MH924298; L. plantarum 3.7E - MH924275; and P. pentosaceus 14.15SE - MH924301.

Growth tests at different temperatures (15 and 45 °C), pH (3.5, 4.0, 4.5 and 8.5 at 37 °C), salt concentrations (40 and 60 g/L of NaCl at 37 °C), gas production and antimicrobial activities were performed in a previously study11.

The efficiency in reducing pH was measured using a potentiometer after 24 h at 37 °C in alfalfa broth. Alfalfa broth was obtained from 100 g of herbage crushed in 400 ml of distilled water in an industrial blender for 1 min and was filtered and sterilized (121 °C, 15 min). The strains were activated twice in MRS broth for 24 h, and one more time in tubes containing 3 ml of the alfalfa broth for 24 h. A sample (10%) of the inoculum was added to a tube containing 5 ml of the alfalfa broth (pH = 5.87). Samples after 24 h incubation were analyzed for metabolite production (lactic, acetic and propionic acids) using HPLC (Shimadzu Scientific Instruments, Columbia, MD) according to Siegfried et al.12. The characteristics of the strains are presented in Table S1.

The strains were selected based of their metabolite production, ability to induce a fast drop in pH, growing capacity in different conditions and broad-spectrum antimicrobial activity against pathogenic and harmful spoilage organisms, such as Listeria monocytogenes 19117, Listeria monocytogenes 7644, and Escherichia Coli K12.

Silage production

The harvesting of alfalfa cv. Crioula (Medicago sativa cv. Crioula) was performed using a costal brush when the plants were at the early bud stage. Fresh alfalfa was wilted to a DM content of approximately 368 g/kg as fed and chopped into approximately 1.5-cm length particles.

Novel strains were cultured in MRS broth for 14 h, which was the average time that showed the maximum number of cells. Each inoculum was standardized using a spectrophotometer (630 nm) at an optical density of 0.05, in 20 ml of MRS broth, and the amount needed to reach the theoretical application rate of 105 colony forming units (cfu)/g of fresh weight were centrifuged at 1,000 g for 10 min. The supernatant was discarded.

Three replicated piles (each pile treated individually) containing approximately 10 kg of fresh alfalfa were prepared for each treatment (total of 18 piles). Inoculants were diluted in 15 ml of sterilized alfalfa broth plus 35 ml of water, which was sprayed uniformly on chopped forage. A total of 500 g of fresh alfalfa was packed into nylon-polyethylene bags (25 × 35 cm; Doug Care Equipment Inc., Springville, CA), and the air was evacuated from the bags using a vacuum sealer (Eco vacuum 1040, Orved, Italy). The same amount of alfalfa broth and water were applied to the CTRL silages. A total of 108 bags were prepared and stored in the laboratory at room temperature (range, 23–27 °C). Three bags from each treatment were opened 1, 3, 7, 14, 28, and 56 d after fermentation.

Fermentative profile

Twenty-five grams of the forage and silage samples from each mini-silo were homogenized in 225 ml of sterile Ringers solution (Oxoid, Basingstoke, UK) in an industrial blender for 1 min. The aqueous extract was divided in two portions: one portion was used to measure the pH using a potentiometer and determine the concentrations of NH3-N13, WSC14 and organic acids, as described previously.

Quantification of microbial populations

The second portion of the aqueous extracts was used to quantify the LAB, enterobacteria, yeast and mold populations. Serial dilutions were made in Ringers solution and plated using the plate technique in different culture media. Cultivation of the LAB population was performed on MRS agar (DifcoTM Lactobacilli MRS Agar®) at 37 °C for 48 h. Culture of enterobacteria was performed on VRB agar (Violet Red Bile) at 37 °C for 24 h, and the cultivation of mold and yeast was performed in Dextrose Potato Agar media containing a 1.5% tartaric acid solution (10% w/v) at 25 °C for 96 h. The cfu was determined on plates containing 25 to 250 colonies.

Chemical composition

Alfalfa samples before ensiling and their silages were dried in a forced-air oven at 55 °C for 72 h and milled in a Willey mill with a 1-mm sieve for determination of the DM (method 934.01) and CP (method 984.13), as described by the AOAC15.

Acid detergent fiber (ADF)15, (method 973.18) and neutral detergent fiber (NDF) using heat-stable α-amylase without sodium sulfite were only analyzed in the forage (day 0) and silage samples after 56 days of fermentation and corrected for residual ash16. Corrections of the NDF and ADF for nitrogen compounds were performed according to Licitra et al.17.

The microbial populations and chemical composition characteristics of alfalfa forage prior the ensiling processes are shown in Table 1.

Table 1 Chemical composition (g/kg of DM, unless otherwise stated) and microbial populations (log cfu/g of fresh weight) of alfalfa forage before ensiling.
Full size table

In vitro dry matter digestibility

In vitro DM digestibility (IVDMD) was performed on alfalfa silage samples from 56 d of fermentation. Dried 1-mm screen samples (0.5 g) were weighed in duplicates on F57 bags (Ankom Technology Corp.). Fermentation was performed in vitro using the DaisyII rotating jar in an incubator (Ankom Technology Corp.), according to methods described by Tilley and Terry18 and adapted by Holden19. The analyses were replicated on two different occasions.

A composited inoculum was prepared with rumen fluid and rumen solids (pH = 6.09) collected from 3 cannulated lactating Holstein cows in mid-lactation that were fed a diet containing 70 g/kg of corn silage and 30 g/kg of concentrate mix (DM basis). Bags were removed from the jars after 48 h of fermentation, rinsed, and dried in a forced-air oven at 55 °C for 48 h. The coefficients of IVDMD was determined.

Statistical analysis

The data were analyzed using the MIXED procedure of SAS® (v. 9.4 SAS Institute Inc., Cary, NC). The general model was given by Yijk = μ + Ii + Pj + (IP)ij + eijk, where Yijk = response variable; μ = overall mean; Ii = effect of inoculant i; Pj = effect of period j; (IP)ij = effect of the interaction between the level i of factor I and level j of factor P; and eijk = random residual term. The estimated means were compared using Tukey’s test considering a significance level of 0.05.

Results

The P-values and standard error of the mean of fermentation characteristics and microbial populations of alfalfa silages are shown in Table 2. There was an effect (P < 0.05) of I and P on DM content and yeast population. CP content was affected only by P. Lower DM were observed in CTRL and LP silages than the other silages (360.5 vs. 365.5 g/kg as fed). However, this difference was biologically insignificant (Fig. 1). LP, PP, and LP + PP silages had lower yeast counts than CTRL (3.76 vs. 4.46 log cfu/g of fresh weight) (Fig. 1). The CP content and yeast population were reduced during the fermentation period (Fig. 2).

Table 2 Significance (P-values) of the tested effects and the standard error of the mean (SEM) for the fermentation profile variables of alfalfa silages treated with inoculants at different fermentation periods.
Full size table
Figure 1
figure 1

Effect of the microbial inoculants on the dry matter content (a) and yeast population (b) of alfalfa silages. a,bMeans followed by different letters are significantly different according to Tukey’s test (P < 0.05). CTRL = Control (without inoculant); CI = Commercial inoculant - Silobac; LPE = Lactobacillus pentosus; LP = Lactobacillus plantarum; PP = Pediococcus pentosaceus; LP + PP = Lactobacillus plantarum + Pediococcus pentosaceus.

Full size image
Figure 2
figure 2

Effect of the fermentation period on the dry matter (a), crude protein (b) and yeast population (c) of alfalfa silages. a–cMeans followed by different letters are significantly different according to Tukey’s test (P < 0.05). CTRL = Control (without inoculant); CI = Commercial inoculant - Silobac; LPE = Lactobacillus pentosus; LP = Lactobacillus plantarum; PP = Pediococcus pentosaceus; LP + PP = Lactobacillus plantarum + Pediococcus pentosaceus.

Full size image

The I × P interaction affected (P < 0.05) the pH, WSC, NH3-N, LAB, enterobacteria and mold populations (Table 2). The characteristics of alfalfa silages as a function of microbial inoculant within each fermentation period are shown in Table 3.

Table 3 Average fermentation profile, chemical composition and microbial populations of alfalfa silages as a function of the microbial inoculant within each fermentation period.
Full size table

The LAB population was not affected by I only at 14 d of fermentation. The lowest counts were observed in CTRL silages in the first week of fermentation. Viable LAB counts increased from 6.54 log cfu/g in fresh forage to greater than 9 log cfu/g of fresh weight in silages. The population subsequently declined slowly after this peak. However, the LAB peak value for the inoculated silages occurred as early as 3 d after fermentation, and the peak was observed only after 28 d in non-inoculated silages. Lower counts of LAB was observed in PP silages at 28 d than other silages, except LP. However, PP silages had lower counts than LP at 56 d.

The I × P interaction did not affect the enterobacteria population at days 1 and 28. CTRL silages had the highest population 7 and 14 d after fermentation. The new strains controlled more efficiently the population at the end of the fermentation resulting in lower counts of these microorganisms. The mold population was not affected by interaction at 1 and 56 d. CTRL silages had the highest population of this microorganism at day 3 and higher counts than PP and LP + PP at 7 and 14 d of fermentation.

Non-inoculated silages had higher concentrations of residual WSC from 3 to 14 d, but no differences between silages was observed after 14 d. The pH was not affected by I only on the first day of fermentation. CTRL silages had the highest pH 3 and 7 d after ensiling. Inoculated silages had the highest pH decline rates in the first week of fermentation. PP resulted in the lowest pH values from 14 d of fermentation.

Inoculants affected the concentration of NH3-N at all fermentation periods. Lower values were observed for LP and PP silages at 28 d compared to CI (65.5 vs. 85.7 g/kg TN). PP silages had lower concentrations than LP and LPE (69.4 vs. 84.8 g/kg TN) after 56 d of fermentation.

The I × P interaction also affected the production of lactic, acetic, propionic and butyric acids (P < 0.05) (Table 2). The acid concentrations as a function of microbial inoculant within each fermentation period are shown in Table 4. An I × P effect was observed on the concentrations of lactic acid at 3, 7 and 28 d. CTRL silages had the lowest concentrations at 3 and 7 d. PP silages had the highest lactic acid values on day 28 of fermentation (40.6 vs. 34.9 g/kg DM).

Table 4 Average organic acid production of alfalfa silages as a function of the microbial inoculant within each fermentation period.
Full size table

The interaction did not affect the acetic acid concentration only at 1 d after fermentation. LP + PP and CTRL silages had higher concentrations than PP silages at days 3 and 7, respectively. The highest concentrations were observed in LP silages (9.3 g/kg DM) at day 14. Lower values were observed in LP and PP silages compared to CTRL (6.9 vs. 11.3 g/kg DM) at 28 d. PP silages had the lowest acetic acid production comparing to the new strains at 56 d.

Propionic acid was affected by the interaction at 3 and 7 d after fermentation. CTRL and CI silages had the highest concentration at day 3, and CTRL and LPE had the highest concentrations at day 7. There was a slight increase in butyric acid concentration in CTRL silages only 7 days after fermentation.

The chemical composition and IVDMD of alfalfa silages at 56 d of fermentation are presented in Table 5. The DM, CP, NDF and ADF of silages at 56 d of fermentation were unaffected (P > 0.05) by I and averaged 364 g/kg as fed, 169.2, 388.0 and 259.0 g/kg DM, respectively. Silages inoculated with PP had higher coefficients of digestibility than LP silages (0.644 vs. 0.611).

Table 5 Average (with the respective standard error of the mean and ANOVA-based P-value) chemical composition and coefficient of the in vitro dry matter digestibility of alfalfa silages treated with microbial inoculants at 56 d of fermentation (g/kg DM, unless otherwise stated).
Full size table

Discussion

Silage is a very complex fermentation matrix that exhibits variability in natural microbiota, chemical composition and nutrients, such as WSC and the nitrogenous components available for microbes20. The occurrence of desirable silage fermentations is guided by the amount and type of microorganisms present in the plant and the DM content, buffering capacity and WSC of the forage3.

The average WSC of raw alfalfa was lower than the 40–60 g/kg DM recommended by Mahanna21 as adequate for the occurrence of good fermentation of silage. However, studies on alfalfa silage also reported WSC between 10 and 40 g/kg DM22,23. The LAB counts were higher than the minimum established by Muck24 (5.0 log cfu/g fresh weight) as adequate for the occurrence of good fermentation of silage.

The faster increase in LAB counts observed in inoculated silages in early fermentation indicated that the LAB strains were competitive among the epiphytic communities. Microbial changes during this phase in successfully fermented silages are primarily the result of the disappearance of enterobacteria and the development of a dominant LAB population. The speed of this shift closely correlates with the rate of pH decline and lactic acid production25. The reduction in the LAB population after the peak in all assessed silages was expected because low pH and the lack of fermentable substrates result in bacterial death26.

The reduction of pH is related to the conservation of the ensiled material. The fast initial acidification promotes a decrease in the enzyme-mediated proteolytic activity of the plant itself and controls the growth of enterobacteria and clostridia, which grow until an inhibitory concentration of non-dissociated acids and/or sufficiently low pH are reached27.

In our study, the acidification induced by epiphytic bacteria fermentation in the CTRL silages reached the similar values of inoculated silages pH after 14 d. The highest pH values on days 3 and 7 in the CTRL silages reflected the low epiphytic LAB counts and its low efficiency in initiating fermentation and controlling undesirable microorganisms compared to the LAB strains, as suggested by Davies et al.28. The final pH values of all the silages were within the range of 4.48–4.88, which is considered adequate for legume silages, which usually stabilize when the pH drops to between 4.5 and 4.929.

The changes in the WSC contents are related to the use of these carbohydrates by bacteria as substrates for growth, which results in the synthesis of primarily lactic acid30. As expected, the WSC concentrations of all silages were reduced during the fermentation. The highest residual WSC content in the CTRL silages in the first week of fermentation reflects the lower fermentation intensity in these silages, as evidenced by the lower LAB counts and lactic acid concentrations and higher pH values.

Zielińska et al.6 found that some LAB strains developed more intensively in ensiled plants because of their role in the partial hydrolysis of starch, cellulose and xylans. This capacity may explain the lowest pH values of the PP silages with similar residual WSC concentrations at day 28 of fermentation, which was reflected by the highest conversion of substrate into lactic acid in these silages during the same period.

The reduction of CP content during the fermentation process was due to the plant and microbial proteolytic processes in the ensiled material, which change the nitrogenous compounds in silages and results in an increase in soluble N and NH3-N31, as observed in our study. According to Langston et al.32 proteolysis results in the formation of peptides and amino acids. The NH3-N formation is a reflection of amino acid deamination, which characterizes the end of a putrefactive process. The NH3-N concentrations were different between silages in our study, but no differences in CP concentrations were observed between inoculants, which suggests that the production of NH3-N resulted from different intensities of the deamination of free amino acids in the material.

The higher NH3-N concentrations of silages inoculated with LPE and LP strains compared to PP on the last day of fermentation were reflected by the higher pH and acetic acid concentrations in these silages, which indicated the growth of undesirable microorganism. Kung10 and Oliveira et al.1 reported that P. pentosaceus strain inoculation did not affect NH3-N concentrations in silage, which may be related to its slower growth rate than other bacteria, but this effect was not observed in our study.

The main acids identified in the silages are acetic, butyric and lactic because these acids represent the highest concentrations of acids33. Kung et al.31 demonstrated that lactic acid was generally found at the highest concentration in silages during the ensiling process and contributed the most to the decrease in pH during fermentation because it is approximately 10 to 12 times stronger than the other major acids. The concentrations of lactic acid in the silages were 20 to 40 g/kg DM, which are the concentrations commonly found in legume silages that were also reported by these authors.

The lower production of lactic acid in CTRL silages in the first week of fermentation reflects the lower LAB counts and their ability to dominate the fermentation, as discussed previously. Muck and Kung5 found that silages treated with homofermentative bacteria resulted in lower silage pH compared to untreated silages because of the greater production of lactic acid, which may be more evident in legume than corn silage.

Although lower LAB counts were observed in the PP silages at day 28, higher lactic acid values were produced, which shows the efficiency of substrate utilization and the persistence of acidification of the strain. PP silages had lower pH than other inoculated silages at day 56 of fermentation with the same lactic acid concentration, which may be attributed to the lower production of NH3-N and acetate in these silages and the reduced buffering effects of these compounds on the ensiled material34.

The contents of acetic and butyric acids are primary negative indicators of the quality of the fermentation process and also correspond to silages that showed marked losses of dry matter and energy during fermentation. Lower concentrations of acetic acid in PP silages at the end of the fermentation period may result in higher DM recovery, and it indicates the predominance of homolactic fermentation compared to the LP, LPE, and LP + PP silages. The highest concentrations of propionic acid at 3 and 7 d in CTRL silages may have resulted from secondary fermentations, especially because the concentrations of lactic acid were lower in this silage during this period.

Enterobacteria are generally the second most numerous bacterial group of the epiphytic microbiota active in the silo. Their population and rate of decline are used as indicators of silage quality because these microorganisms are main competitors with LAB for available sugars and result in gas losses and a reduction in the nutritional value of the silages30.

The dominance of LAB, the faster drop in the pH induced by higher lactic acid production and the synergistic effects of the acids produced during the fermentation in our study resulted in the reduction of enterobacteria counts in all inoculated silages until 14 days of fermentation. The same reasons are attributed to the reducing of yeast population throughout the fermentation periods.

The studied variables did not affect the chemical composition of silages at 56 d of fermentation, which suggests that the attendant improvements in silage characteristics are often lacking even when the concentrations of supposedly explanatory metabolites increase in response to bacterial inoculation. This effect may occur because the explanatory metabolites only explain a fraction of the variability in the response to an inoculant35. The measurement of IVDMD is used to analyze the nutrient digestibility of feed ingredients15,36.The values of IVDMD in our study were similar to Nadeau et al.37 and Rodrigues et al.23, also in alfalfa silage.

In conclusion, all of the novel strains tested had a positive effect on at least one chemical property of the silage during the fermentation process. However, the addition of P. pentosaceus alone had a positive influence on all of the evaluated parameters and changed the characteristics of the silages; particularly, the strain enhanced the lactic acid content and decreased the pH, deteriorating microorganisms, and NH3-N and acetic acid concentrations, which resulted in a better silage quality that surpassed the commercial inoculant. We suggest that this strain has potential for use as a silage inoculant, but it must be tested in different forages and in combination with other additives, such as heterofermentative bacteria or chemical additives. The results obtained at the laboratory scale must also be confirmed under more practical conditions.