Lactic Acid Bacteria as Microbial Silage Additives: Current Status and Future Outlook

Silage making is not a novel technique. However, the agricultural industry has made great strides in improving our understanding of—and efficiency in—producing high-quality silage for livestock. Silage microbiology research has been using the newest molecular techniques to study microbial diversity and metabolic changes. This chapter reviews important research that has laid the foundation for field-based utilization of silage inoculants. We also outline areas of current, and future, research that will improve global livestock production through the use of silage.


Introduction
Fermentation of forage is harder to control than other fermentation processes such as industrial fermentation of food. Whole plants cannot be manipulated to remove contaminating microorganisms, and this can lead to important variations in the quality of the forage. Harvesting machinery can also contribute to the inclusion of soil or manure particles as contaminants. Other factors have an impact on silage quality, which include harvesting management, packing rate, weather events during harvest, selection of the ensiling structure, and selection of a microbial or chemical additive to preserve the crops. Figure 1 provides an overview of the interactions between the main parameters involved in the production of high-quality silage.
This chapter will evaluate the recent published literature and will expand on the current knowledge in the study of the microbiota, search for silage inoculants, issues with aerobic instability, and understanding nonusers of forage inoculants. We will also review important research areas of microbial inoculants: fiber digestibility, analyzing "big data" functional studies, co-ensiling with by-products or food-processing wastes, and how lactic acid bacteria (LAB) used as forage additives influence animal performance.

Microbiota diversity during ensiling
Characterization of the different microbial species observed throughout the different phases of the ensiling process was traditionally performed using culture-dependent methods, following the isolation of strains and the determination of their taxonomic classification. The use of selective media has several shortcomings, including limited knowledge on how composition of the different defined culture media influences the growth of organisms within the targeted species range. Dormant or inactive cells (viable but nonculturable) may not have been accurately measured [1].
New techniques based on DNA profiling have helped understanding the microbial diversity of silage within specific families or genera [2]. These techniques were diverse and included denaturating gel electrophoresis [3] or metabolic fingerprinting by Fourier transform infrared spectroscopy [4].
Next generation sequencing (NGS) technologies provide more complete details on microbiota diversity. The first application of NGS in silage was performed on ensiled grass to help understand how inoculation would influence the microbial communities [5]. Three years passed before a second paper would be published using NGS studying spatial and temporal microbial variations in commercial bunkers [6]. Several more papers or communications were performed afterward (see Table 1).
One of the complexities facing ensiling of forage is that several factors will influence the size and diversity of the microbial community at harvest. Microbial diversity will change according to the plant species, weather conditions during growth and prior to harvesting, fertilization management, physiological state of the forage, and so on. As an example of the potential variation, important differences in the composition of the epiphytic bacterial population were observed from different organs of whole plant corn in the weeks prior to harvesting (Figure 2). Leaves, silk, Ensiling involves several biochemical and microbiological descriptors that are influencing silage quality and could be controlled by different management criteria (boxed elements), which are directly influencing the main fermentation parameters of forage as well as animal productivity. ( and tassels harbored different proportions of the main epiphytic bacterial families even though the variation in microbiota composition was small between the sampling periods. Cytophagaceae and Methylobacteriaceae were mainly observed on the leaves, while Enterobacteriaceae and Pseudomonadaceae were observed on silk, cob, and tassel [7]. Published results of microbiome analysis were performed from varied forages from temperature and tropical regions, including pure strands of legumes or grasses and mixed forages. Several studies performed time-based samplings to describe changes in the microbial communities in relation to the fermentation periods [7][8][9] ( Table 1). Generally, the relationship between the time of fermentation and the microbial composition was similar to the general succession pattern previously reported by culture-dependent microbiological techniques. For example, with corn silage inoculated with either Lactobacillus plantarum or Lactobacillus buchneri and/ or Lactobacillus hilgardii, it was possible to observe that the succession to Firmicutes was rapid, in a matter of hours after sealing the experimental mini-silos. A second observation was that Leuconostocaceae (mainly Weissella sp.) was the dominant operational taxonomic unit (OTU) during early fermentation. In both studies, there were important changes in bacteria richness during the fermentation, with either values below 50 OTUs after incubation of 30 days [9] or decreasing throughout fermentation to a similar level of OTUs [8]. In both studies, fungal richness dropped throughout fermentation. These changes in microbial population were also observed in samples collected on farms. Under commercial conditions, comparing silage made from the same forage between sites is difficult since differences in dry matter (DM), packing density, and other physical parameters will influence efficiency of the fermentation and the microflora. Associating those parameters to NGS studies could improve the understanding of this process. It will then be possible to comprehend how other physical variables may contribute, e.g., the impact of high or low temperature on microbial succession, the impact of length of storage, length of time at a high temperature, and the impact of DM variations within the same forage.
To date, most of the data collected from experimental silos was performed with incubation periods shorter than 100 days and at a temperature around 20-25°C. These conditions offer an initial set of parameters but must be expanded to simulate real-life conditions in silos, which could include variances of more than 20°C above ambient temperatures during fermentation and long fermentation periods [10].
Most of the published studies included a comparison between control and a microbial silage additive or between different strains of LAB. The general trend on microbial diversity is that inoculation with LAB reduces the microbial diversity, but the impact differs in relation to the forage and the species of LAB. As observed by Wang et al. [11], microbial diversity was influenced by the inoculation of Moringa oleifera differentially for each of the four LAB species inoculated as well as from the temperature of incubation.
Comparisons between studies tend toward similar changes in microbial composition. To facilitate comparisons, it will be necessary to standardize DNA isolation and preparation of the amplicons prior to sequencing. By summarizing the main methodology information from different trials ( Table 1), it was observed that some studies did not include fungal diversity, and the amplified DNA region differed. Most bacterial studies were performed following the amplification of the V3-V4 region, but there was a trend toward using the V4 or V4-V5 region, which offers potential for longer DNA strand and improves comparison scores against the database. Using a good quality database is also a critical step that is often overlooked during analysis [12]. The drawback of the current methodology for amplicon-based metagenomic is that the amplified region is short and does not provide enough coverage of the complete 16S rRNA gene. Two published studies were able to gather near complete fragments by sequencing the 16S rRNA gene on a PacBio sequencer instead of the Illumina model [12,13]. This expanded the analysis of diversity to the species, or even subspecies, level.
Currently, no study has tried to mix the potential offered by polymerase chain reaction (PCR)-based profiling technology-like PCR-DGGE-with NGS capacities. Instead of amplifying with universal primers, primers targeting regions of lower variations within ribosomal DNA, or in other genes, provide more precise results allowing higher similarity scores at the species level.
Microbial communities continuously evolve during the storage period, even during the anaerobic stable phase. By improving our knowledge on the succession between communities, genus, species, and even strains, it will be possible to refine how strains are selected as microbial silage additives. This could easily allow selection of strains for particular forage species or climatic conditions.

Searching for new forage inoculants in temperate and tropical forages
The fermentation capability-or the acidification potential-depends directly on the DM content, at the level of water-soluble carbohydrates (WSC), and, inversely, on the buffering capacity of a given forage [14]. Due to their compositions, the ensiling potential is completely different among the different families of forages: tropical (C4), temperate (C3) grasses, and legumes.
Studies conducted by Wilkinson [15] with C3 grasses have concluded that the minimal concentration of the WSC should be at least 2.5-3.0% of the fresh forage. Below 2% of WSC of fresh crop weight, forages are prone to undesirable fermentations. The average level of WSC found by Zopollatto et al. [16] in a review of microbial additives in Brazil for tropical grasses was only 1.6%, far from the minimum for a good fermentation.
Tropical grasses provide large quantities of DM, which can reach up to 30 tons of DM per hectare. This great yield, however, comes at the optimal stage of maturity in terms of nutrients with other types of challenges: wilting is an issue, and the excess moisture can lead to important losses of nutrients through effluent production [17]; additionally, its nutritive value sharply declines as maturity advances.
The microflora existing on the vegetative parts of plants consist mainly of microorganisms considered undesirable from the point of view of the fermentation process. These include anaerobic bacilli of the genus Clostridium; aerobic bacteria of the genus Bacillus; coliform bacilli, including Escherichia coli, Enterobacter spp., Citrobacter spp., and Klebsiella spp.; as well as bacteria of the genus Listeria, Salmonella, and Enterococcus (E. faecium, E. faecalis, E. mundtii, E. casseliflavus, E. avium, and E. hirae); and the occurrence of actinomycetes. Species of Clostridium are responsible for large losses because they produce CO 2 and butyric acid instead of lactic acid. Yeast and molds also form a large group [18].
Concerning the presence of LAB, Pahlow et al. [19] found in grasses that L. plantarum, L. casei, E. faecium and Pediococcus acidilactici were the most frequently observed species. However, with the development and the use of DNA sequencing profile techniques, it is possible to identify hundreds of species as mentioned earlier. Most of the studies done by scientific groups were based on the efforts to find any microorganisms, especially bacteria, able to drive a good fermentation and inhibit undesirable and detrimental microorganisms.
Zielińska et al. [20] demonstrated that microbial inoculants altered many parameters of silages, but the strength of the effects on fermentation depends on specific characteristic of an individual strain. Several research teams have been searching for new strains able to perform better than the ones currently on the market. For example, Agarussi et al. [21] searched for new promising strains for alfalfa silage inoculants and isolated Lactobacillus pentosus 14.7SE, L. plantarum 3.7E, Pediococcus pentosaceus 14.15SE, and a mixture of L. plantarum 3.7E and P. pentosaceus 14.15SE. The authors concluded that all of the tested strains had a positive effect on at least one chemical feature of the silage during the fermentation process, although the most promising strain found in that trial was the P. pentosaceus 14.15SE.
Moreover, Saarisalo et al. [22] searched for LAB capable of lowering the pH of grass silages with low proteolytic activity. The researchers found a potential strain of L. plantarum, which was effective in reducing the deamination in silages.
Besides aiming to enhance silage fermentation, aerobic stability has been an important topic in the last 20 years. During silage feedout, accelerated growth of spoilage organisms (yeasts) results in high temperatures and nutrients and DM losses, leading to increased silage deterioration [23]. According to McDonald et al. [24], even though yeasts can grow from 5 to 50°C, the optimum growth of most species occurs at 30°C. Other spoiling microorganisms, such as molds and Clostridium bacteria, grow between 25 and 37°C, respectively. Considering the specific temperature and humidity ranges of different microbes for growth, it is possible to see that tropical climates are more prone to spoilage than temperate ones.

Improving aerobic stability using forage inoculants
Silage feedout is the final phase of the ensiling process. At that moment, oxygen can slowly diffuse inside the silage mass. Diffusion speed will be influenced by different factors, including the level of humidity, porosity, and temperature of the silage [25].
The process of aerobic deterioration of silage involves a shift to aerobic metabolism in some microorganisms and the reactivation of strict aerobes that were dormant. Reduce nutritional value due to oxidation of the fermentation products, of carbohydrates, amino acids, and lipids to H 2 O, CO 2 , and heat. Simultaneously, the higher metabolic activity will increase the silage temperature, accelerating microbial growth. Several microorganisms are involved, but yeast and acetic acid bacteria are adapted to tolerate the initially low pH conditions and thus able to exploit this niche before pH increases following the catabolism of the organic acids. Crops with higher levels of easily accessible carbohydrates are more prone to aerobic deterioration, i.e., corn, sorghum, and sugarcane, since these sugars can be readily fermented by spoilage microorganisms in the presence of oxygen.
Following the isolation of a L. buchneri strain [26], researchers described its unique metabolic pathway, which consisted of converting moderate amounts of lactate under low pH to equal parts of acetate and 1,2-propanediol [27]. The latter chemical is an intermediate in the potential synthesis of propionic acid. L. buchneri does not have the gene to complete the reaction, so another species of LAB has to be involved to convert 1,2-propanediol to an equimolar amount of 1-propanol and propionic acid [28]. This conversion was initially observed in silage by Lactobacillus diolivorans [29], but other members of the buchneri group also possess the genetic system [30], like Lactobacillus reuteri [28].
Compared to lactic acid, the key feature of acetic and propionic acids in improving aerobic stability of silage is based on the difference in pK a between these weak acids and lactic acid, which is a stronger acid, with a pK a of 3.86. At higher pK a , 4.76 for acetic acid and 4.86 for propionic acid, these weak organic acids will have a low dissociation level under most ensiling conditions, thus allowing for passive diffusion inside the yeast or other microorganism cytoplasm. Once inside the cytoplasm, propionic acid will dissociate to the corresponding salt since internal pH is above pK a value. The same process is also possible for acetic acid. Constant pumping of the protons released inside the cytoplasm causes physiological stresses impacting several metabolic pathways in yeast cells [31].
Length of fermentation and establishment of heterofermentative LAB population are now considered critical toward the establishment of a good aerobic stability level. The facultative, or obligate heterofermentative, strains of LAB have lower growth rates than homofermentative strains, including rods like L. plantarum or coccids of the genera Leuconostoc, Enterococcus, or Lactococcus. The growth conditions after several days of ensiling are also more restrictive for physiological activities considering the low pH usually encountered. The strains succeeding the earlier colonizer need to be more tolerant to both acidity and osmotic stresses, simultaneously. Observation of the succession of different species of LAB during the anaerobic stability phase often leads to high abundance of LAB belonging to the L. buchneri taxonomical group [32], leading to specific adaptation to this ecological niche by these strains. Although few physiological studies on L. buchneri strains had been published, Heinl and Grabherr recently published a complete analysis of the genetic potential of the strain CD034 compared to other genomes from public databases [33]. One of the comparisons performed aimed to describe how the genetic system of this species can cope with high concentration of organic acids, including lactic acid. The anaerobic conversion system of lactic acid to 1,2-propanediol (to acetic acid and CO 2 under aerobic condition) represents one of those properties. It is possible to extend these observations to the results gathered from transcriptomic analysis on the strain L. buchneri CD034 [34] following the aeration of culture grown under anaerobic conditions. The team described the functions of 283 genes induced by the presence of oxygen. They also observed physiological adaptation related to changing oxygen concentration. Genes required by lactic acid fermentation systems were hardly affected.
Co-inoculation with different heterofermentative strains has recently been tested in the field or in commercialized conditions. This was the case for L. buchneri and L. diolivorans, tested on the fermentation of sourdough [35]. The authors showed an increase in the accumulation of propionic acid following the inoculation with both strains together. Co-inoculation of L. buchneri and L. hilgardii was tested in different ensiling trials [36,37] inducing better fermentation and higher aerobic stability level. L. hilgardii, an obligate heterofermentative strain, was not only previously observed as a contaminant of wine but also represents one of the dominant LAB strains in water kefir [38]. Strains of this species are often observed in sugar cane silage [39,40] and provide increased aerobic stability levels for this challenging crop. Improvement in fermentation and aerobic stability of sugarcane silage allowed increasing DM intake and milk yield [41].
Two recent meta-analyses [42,43] provided a complete overview of the impact of inoculation of LAB and described the importance of fermentation and aerobic stability in relation to the specificities of the forages and the activity of homofermentative, facultative heterofermentative, and obligate heterofermentative strains. In particular, the meta-analysis of Blajman et al. [42] analyzed the role of inoculation on reducing the amount of yeast in silage.
Improving aerobic stability to reduce overall losses during the storage and feedout is one of the main reasons to apply microbial inoculants on the forage at the time of ensiling. The value of silage inoculants is important, but optimal management of silos at all steps of the ensiling process is critical.

Improving adoption of forage inoculant use by increasing awareness of the economic value of forage inoculants
According to the 2017 National Agricultural Statistics Survey [44] census report, approximately 120,000,000 tons of whole-plant corn alone was harvested for silage in the United States. Even with this huge quantity of silage, there is little reliable survey data about the use of forage inoculants.
Based on an independent market survey of U.S. beef and dairy producers, two thirds of respondents indicated that forage additives used on their operations are microbial based. The main reason for their use is to minimize mold and spoilage in silage. Other reasons cited include preventing heat damage and increasing herd productivity [45]. Most inoculant users plan on continuous using and investing in this technology each year (personal communication).
Product performance, ease of use, and cost are the main influencers on the purchasing decision of inoculants. In addition, nutritionists and consultants are important sources for providing information on forage inoculants and the most involved outside sources in the purchase decision (personal communication).
Most producers do not have a detailed understanding of the different types of inoculant products, but they instead recognize the value and return on investment (ROI) that these technologies can bring to their operation. Value-added services and education offered by inoculant companies are also reasons to purchase, especially for larger producers.
Producers may often choose not to purchase forage inoculants due to the cost of the products. Other top reasons that influence purchase decisions are (1) not believing inoculants work, (2) lack of knowledge, or (3) lack of specific equipment for inoculating the forage. With all these factors in mind, there is a strong need for proper education on the application and showing the cost-to-benefit calculation of these forage additives (personal communication).
Even though some producers are nonusers, they believe that inoculants have the potential to improve consistency of silage quality, enhance ration quality, and increase feedout stability. In the same question, just 40% answered that improving ROI is one of the most important benefits of purchasing inoculants. Even though some producers do not associate inoculants with contributing to overall herd ROI and profitability, they positively associate the word "fresh" to silage having a good smell and high palatability (personal communication).
During typical field and harvest management conditions, silage losses are easily reported between 15 and 20%. If inoculant use can reduce DM losses by 5 percentile points, there would be savings of $2000 (US$) per thousand tons of silage, assuming the silage is valued at $40.00 (US$) per ton FM. Moreover, silage with high degree of deterioration not only has less overall tonnage to be fed, but the feed is also of lower nutritional quality.

Optimizing fiber and carbohydrate digestibility
The main metabolic activity of LAB during the ensiling process consists of reducing soluble carbohydrates to organic acids to acidify and preserve the forage for long-term storage. It has been observed that animal performance has been increased following the use of microbial inoculants, even if no or small changes in silage fermentation parameters were observed [2]. Future research is needed to explain why these improvements are observed. Yet, past research has made several important advancements.
As discussed previously, inoculation with LAB contributes to important modifications of the silage microbiota, for both the bacterial and the fungal communities. Some of these modifications could partly explain the contribution of the inoculant to one or more nutritional characteristics of silage. This could also support the theory of an indirect positive impact of these nutritional characteristics to the rumen microbial population and functions.
The rumen environment may also be affected by LAB forage inoculants. Some strains of LAB used as inoculants were shown to survive in the rumen fluid [46] and shift gas production toward other products or microbial cells [47]. Weinberg et al. [48] observed that LAB inoculants applied at ensiling, or into the rumen, had the potential to increase DM and fiber digestibility.
Studies using different inoculants showed increases in animal performance and milk production [49]. Mohammed et al. were also able to quantify elevated levels of L. plantarum in the rumen of cows eating the treated silage [50].
To help explain this improved animal performance, results from the studies of LAB used as a human probiotic may offer some clues. In a review of the metabolism of oligosaccharides and starch by lactobacilli, Gänzle and Follador [51] described limitations of the conversion of oligosaccharides since most related enzymes in LAB are active intracellularly and their substrates must be transported inside the cells to hydrolyze (Figure 3). By studying the genome of several LAB species, they report that most lactobacilli could generally metabolize α-glucans. They would require contribution of a trans-membrane transporter in order to hydrolyze small oligosaccharides. Like some other lactobacilli, L. plantarum genome includes a gene encoding for an extracellular amylase with endoamylases activity. The presence of this amylase in the genome is strain specific as reported by Hattingh [52] for strains of L. plantarum isolated from barley.
Selecting strains with a functional trait, for example, fiber-or starch-degrading functions, represents the initial step in the development of a new inoculant. The strain has to cope with the different stresses of silage and also compete against epiphytic LAB and other microorganisms. The function has also to be expressed under the targeted microbial niche. The extracellular enzymes then have to be optimized for the acidic conditions and cope with the specific nature of polysaccharide substrates.
Access by fibrolytic enzymes to cellulose is difficult due to steric hindrance of the lignin-hemicellulose-homocellulose matrix. Improving cellulose degradation was targeted by selecting a LAB strain producing ferulate esterase [53]. This enzyme releases ferulic acid from arabinoxylans, improving access to other fibrolytic enzymes of the lignin-cellulose layer within cell walls.
More research is needed in this area. The complexity and dynamic of the microbial communities following the inoculation provide an important challenge in understanding the impact and role of the key players involved in this beneficial effect of microbial silage additive [54].

Improving animal performance with LAB forage inoculants
The expected effects of using a LAB forage additive are improved fermentation and enhanced feedout stability, which in turn lead to better recovery of nutrients and DM. However, expectations from producers are often beyond better silage characteristics, such as improvements in feed efficiency and, subsequently, animal performance.
Scientific evidence shows positive impact from the use of microbial inoculants on increases animal performance and production, in addition to enhancing the fermentation. However, these improvements are difficult to quantify. Some of the existing theories are that these bacteria may have a beneficial influence in the rumen environment, including altering the fermentation profile and interacting with the animal's existing digestive microbiota [48] and inhibiting undesirable microorganisms, which subsequently help reduce the potential for toxin production [55].
Oliveira et al. [43] analyzed 31 studies-including animal performance results. This meta-analysis showed that microbial inoculation at a rate of at least 10 5 colony-forming units (CFU) of LAB per gram of forage significantly increased milk production by 0.37 kg/d, increased DM intake, and had no effect on feed efficiency and total tract DM digestibility. Furthermore, the contents of milk fat and milk protein tended to be higher for cows fed inoculated silage. The effects on increased milk production due to LAB inoculation happened regardless of the type of forage and diet, inoculant bacterial species and application rate (10 5 vs. 10 6 CFU/g of forage), and level of milk production.
Among the animal performance trials, there are cases when the inoculant had no effect on the silage fermentation compared to untreated silage, although animal productivity was increased [56]. Therefore, this indicates that some LAB strains are positively affecting the rumen microbial community and the digestive tract environment, resulting in improved effects on animal performance.
Recent research has described these effects by evaluating the impact of inoculated silages in the populations of the rumen microbial community, but no significant changes were observed [51]. However, nitrogen efficiency seemed to be improved due to lower levels of milk urea nitrogen in cows fed inoculated silage and greater ruminal DM digestibility on the inoculated silage ration [57]. Since LAB were shown to attach to the fiber inside the rumen [58], isolation methodology needs to be adapted to target the correct ecological niche.
Changes in nitrogen compounds during ensiling are expected. For example, over half of the true protein in alfalfa is degraded to soluble nonprotein compounds initially by the plant's own proteases, and then later by microbial activity within the cow, resulting in inefficient nitrogen use to the cow [59].
Specifically, in the corn kernel or other cereal grain, a protein matrix (prolamins) around the starch granules partially prevents ruminal starch digestion. It has been reported that a slow and continuous breakdown of the prolamins during the storage phase makes the starch more digestible with longer storage time [60]. The authors explained that this effect is due to natural proteolytic mechanisms. This event, however, requires months of storage for the optimum level of starch digestibility in the rumen, in which it is not always feasible in commercial operations. One alternative solution would be to shorten the time necessary for storage to help enhance starch digestibility by inoculation with bacteria that possess high proteolytic activity, but, to date, limited research has been reported and results are inconsistent.
Improvement of fiber digestibility has to be considered in relation to the activity of silage inoculants. Some strains of LAB have been reported to produce the enzyme ferulic acid esterease, which breaks the esterease bond between the lignin and the hemicellulose fraction, leading to more digestible fiber portions for the rumen microorganisms [61]. However, data from animal performance or production studies did not show consistencies in the improvements [61,62]. While in vitro and in situ effects may be conceivable, the expression of this phenomenon within in vivo environments needs additional research to be better understood.
There is still a need to better understand how the microbial additives for ensiling positively affect animal performance, so this should be used as criteria for a new generation of this type of additive.

Understanding the impact of ensiling on a global scale
Silage represents an important part of animal diets. Challenges in production, reducing losses, and the impact on agricultural practices are often overlooked compared to other nutritive benefits provided. Microbial activity during fermentation produces several compounds besides the desirable organic acids. Some of those compounds were identified as negatively influencing air quality around farms. They are classified as alcohols, esters, and aldehydes [63,64]. Production and volatilization of these compounds contribute to a reduction in quality of the stored feed, inducing ground-level ozone, and influence emission of greenhouse gases by the agricultural sector [65].
Forage characteristics and yield potential are influenced by several factors, including geographic and meteorological conditions. New analytical technologies and statistical methodologies now allow more comprehensive understanding of ensiling techniques and analyze productivity and nutritional quality on a broader scale.
Comparison between farms is always challenging, even between neighboring farms, since they could differ on animal husbandry, genetics of the herd, field management, harvesting periods, type and size of silos, management of the silos, and so on. On a broader geographic area, these differences will be minimized by the inclusion of higher numbers of farms, up to a point that patterns of variations could be analyzed. This type of analysis was performed by Gallo et al. in two recent studies [65,66]. The team used a multivariate analysis technique, Principal Component Analysis (PCA), to evaluate ensiling of corn silage on 68 dairy farms [66] and generated a fermentation quality index to rank the silage [67]. Using 36 variables measured on every individual samples, they were able to group the silage according to quality parameters in relation to silo management techniques to discriminate between well-preserved and poorly preserved forages.
At the farm level, quality parameters from silage and feed analysis reports could be analyzed to identify trends in animal health and performance. Different types of data could be collected and analyzed to understand the main variations in milk quality and yield on a yearly or multi-year basis. Linking milk quality parameters to farm management practices was performed following the analysis of milk constituent using Fourier transformed mid-infrared spectroscopy results gathered from 33 farms [68]. The difference between observed high and low de novo fatty acid composition of milk allowed characterizing differences in feeding management (one or two feeding periods-fresher silage) and higher animal management scores (freestall stocking-lower housing density).
Up to now, few data analysis included data specific to silage fermentation beside the main fermentation acids. This is truer for other parameters related to silage production and management, including yield from the field, management of the silos, losses during fermentation, or type of silage additive used. This needs to be addressed considering important changes to the microbiota following the inoculation discussed previously and to differentiate in other fermentation chemicals or their relationship with the nature of the additives applied, as observed by Daniel et al. [69].

Increasing the understanding of the fermentation process
Compared to other research domains in agricultural and environmental sciences, using new sequencing technologies to understand the dynamics of the microbial communities in silage is recent. McAllister et al. [12] published a review providing a technological and methodological overview. Currently, the number of trials performed using this technique is small enough that repetitions between geographical regions and over time are nonexistent.
Amplicon-based metasequencing represents the entry level of the -omic techniques. For silage research, the industry could also consider metagenomic, proteomic, transcriptomic, or epigenomic as a potential area of study. A review of the possibilities offered by metabolomics in agriculture was recently published [70].
Since ensiling is based on the fermentation of forage crops, knowledge of the metabolic activity of the forage prior to ensiling would be useful. A review by Rasmussen et al. [71] provides an insight into how plants are coping with physiological changes due to breeding strategies, associations with endophytes or rhizobia, responses to nutrients, and, more interestingly, on the metabolic responses to the osmotic stress. Harvesting and wilting will directly influence plant cell activities and nutrient cycling. The authors reported that amino acids, fatty acids, and phytosterols generally decrease following the water stress, while sugars and organic acids increased. Since the fermentation process requires fermentable sugars for optimal acidification of the forage, wilted plants may respond positively toward ensiling. We need to consider the speed of those changes in concentration of metabolites during wilting compared in order to propose a model of the response to an osmotic stress. Ould-Ahmed et al. [72] provided some knowledge on this response to wilting while studying changes in fructan, sucrose, and some associated hydrolytic enzymes, concluding there is a positive effect toward ensiling requirements from the different metabolites.
Metabolomic profiling of silage was performed in a study aiming to understand the role of inoculation with L. plantarum or L. buchneri in alfalfa silage against a noninoculated control [13]. The authors were able to distinguish all three inoculation treatments by a PCA of the 102 metabolites surveyed. The major metabolites observed were related to amino acids, organic acids, polyhydric alcohols, and some derivatives. One of the main observations was an increase in free amino acids and 4-aminobutyric acid following the inoculation with L. buchneri and a decrease in cadaverine and succinic acid following the inoculation with L. plantarum.
Testing the same two LAB strains on whole plant corn silage instead of alfalfa, Xu et al. [32] observed a total of 979 chemical substances, from which 316 were identified and quantified. The PCA allowed separating the three inoculation treatments along the first axis, representing nearly 80% of the variations between samples. The second axis was able to further distinguish how inoculation with L. buchneri influenced the fermentation. Inoculation with either L. plantarum or L. buchneri contributes to increase the concentration of amino acids and phenolic acids, 4-hydroxycinnamic acid, 3,4-dihydroxycinnamic acid, glycolic acids, and other organic acids. Inoculation with L. buchneri also induces higher concentration of 2-hydroxybutanoic acid, saccharic acid, mannose, and alpha-d-glucosamine-1-phosphate, among others. Other substances were increased by ensiling without specific impact of the inoculants, such as catechol and ferulic acid that could have antioxidant functions.
Metabolomic studies can also be used in defining a metabolomic signature specific of different forage and silage on feed efficiency of ruminants. With the aim of identifying feed efficiency traits in beef cattle, Novais et al. [73] investigated how serum metabolomic profiles could be used to predict feed intake and catabolism. They identified different molecules having feed efficiency role. Two molecules from the retinol pathway, vitamin A synthesis, were significantly associated with feed efficiency (higher concentration of retinal and lower concentration of retinoate). DOI: http://dx.doi.org/10. 5772/intechopen.89326 Besides the studies of Guo et al. [13] and Xu et al. [32], one other study combined different -omic techniques in understanding the ensiling process. The first glimpse of that study was presented at the International Silage Conference in 2018 [8] with data on microbiota dynamic between 1 and 64 days of fermentation of corn silage. Analysis of the amplicon-based metasequences, metagenomic, and metabolomic data set is currently underway.
The potential of transcriptomic was also shortly covered by the in vitro trial of Eikmeyer et al. [34], which aimed to understand induction of genes in L. buchneri CD034 under different incubation settings. It is expected that additional studies performed directly under ensiling conditions may be published in the next few years.
Metabolomic data have shown how inoculation of LAB strains induces changes to the ensiled forage that goes beyond the simple production of lactic and acetic acids from the fermentation of sugars under anaerobic conditions. Increases in a whole array of molecules were observed, but the change also extends to the fibers and is either a direct or an indirect effect of the inoculant. Inoculation of alfalfa by L. plantarum or Pediococcus pentosaceus strains increased the release of different hemicellulose polysaccharides, including homogalacturonan, rhamnogalacturonan, and arabinogalactan from the cell walls [74].
These new technologies will allow greater understanding of the impact of bacterial inoculants on improvements of the silage and their contribution in the induction of specific genes and proteins by other members of the microbial community at different stages of the ensiling process.

Co-ensiling forage with food processing waste and TMR conservation
Food processing residues represent high-energy organic material already used in some way that could include either food-processing residues from food industries or distiller's grains from the ethanol production. These residues could easily be used by farms closely located to the production site, but their relatively high humidity content renders them prone to a rapid deterioration. New ensiling techniques allow mixing them with low moisture forage or grain in order to perform a fermentation that is enclosed in a kind of total mixed ration (TMR) acidic conservation.
Aiming to use a bakery co-product waste, Rezende et al. [75] tested possibilities of re-hydration, treating it with acid whey or water and levels of urea. The authors found that the resulting silages had reduced populations of molds and yeast by acidification process. However, the initial population of these microorganisms was high, mainly accounting of Penicillium and Aspergillus spp. Inoculating with a bacteria that could produce antifungal chemicals, including acetic and propionic acids, might be considered for this kind of co-product.
TMR silage is an important source of ruminant feed. This practice has been more common in some places, where companies or producers mix wet co-products with dry feeds to prepare TMR that is then preserved as silage. Based on conventional criteria, aerobic deterioration could occur easily in TMR silage, because lactic acid prevails during fermentation and any sugars remaining unfermented can serve as substrates for the growth of yeasts. However, some trials [76,77] have been shown that when added concentrate, the brewer's grains or soybean curd residue, the main co-products used in TMR preserved do not show heating in the TMR. For the trial with brewers' grain-based TMR, the main bacteria found in the stable silages were L. buchneri, but for the soybean curd-based TMR, the main LAB found were P. acidilactici and L. brevis [78], showing potential association of those bacteria to preserve TMR silages. A similar trial was performed by Ferraretto et al. [79] to test how the process influenced luminal in vitro starch digestibility. They used dry ground corn to adjust the humidity level of wet brewers' grain and observed an increase in digestibility of the starch from the combined feed.
Nishino and Hattori [80] evaluated two bacterium-based additives in wet brewer's grains stored as a TMR in laboratory silos with lucerne hay, cracked maize, sugar beet pulp, soya bean meal, and molasses. The additives tested were the homofermentative LAB, L. casei, and the heterofermentative LAB L. buchneri. This last one was responsible for controlling yeast growth and the homolactic one helped in the fermentative profile of the ensiled TMR.

Final comments
General microbiology techniques have helped to understand the basic dynamic of microbial communities, the diversity of species, the biochemical pathways involved at each phase of the fermentation process, and the metabolic functions of the main spoiling agents involved in degrading the nutritional quality of the silage. NGS helped observe microbial communities, and metabolic profiling does not cease to evolve. This fact directly influences the nutritional characteristics of the silage.
In this chapter, the authors reviewed the main research activities that helped the agricultural industry understand silage, as it is known today and also pointed to experimental techniques that will continue to improve the understanding of metabolic pathways and functional aspects of the ensiling process. It is clear that these techniques will allow the scientific community to discover new inoculants that will combine our knowledge of silage fermentation, understand nutritional quality, improve rumen function, and contribute to better animal health. We are looking forward to the third generation of forage inoculants and seeing their positive impact.