Garlic (Allium sativum L.) is a flavoring ingredient universally used with an annual production of approximately 20 million tons in the world (Lee et al., 2017). The by-products of garlic processing (garlic skin and straw) account for 25–30% of the weight of the raw material (Arzanlou and Bohlooli, 2010) but are mostly discarded or incinerated, which causes environmental pollution (Kallel et al., 2014). It was reported that garlic skin contains similar antimicrobial and antioxidative compounds as the garlic bulb, such as alliin (Louis et al., 2012), N-trans-coumaroyloctopamine and N-trans-feruloyloctopamine (Wu et al., 2015), and polyphenols (Kallel et al., 2014). There is increased interest in utilizing garlic skin in animal husbandry in recent years. Although garlic skin has a slightly pungent odor, Zhu et al. (2021) find that supplementation of 80 g/kg dry matter (DM) garlic skin did not influence the DM intake and improved the growth performance of lambs. However, garlic skin is not suitable to be fed alone owing to its relatively low nutritional values.

Pennisetum hydridum (P. hydridum), also known as king grass or Hybrid Pennisetum, is a perennial gramineous forage species, which has been widely planted in the tropics and subtropics of China. P. hydridum has high fresh biomass yields (85–300 t ha^–1) and strong resistance to environmental threats (Tan et al., 2021) and is commonly used as a forage source for ruminants because of its high palatability and nutritional values (Li et al., 2014). Because of the seasonal imbalance availability of P. hydridum, ensiling is an ideal preservation approach that allows year-round quality forage supplements for ruminants. However, the moisture content of fresh P. hydridum can be higher than 80%, which enhances the risk of clostridial fermentation, extensive proteolysis, and nutrient loss during ensiling (Yan et al., 2019). Wilting prior to ensiling is commonly used to reduce the moisture of the fresh materials, but it may cause extra nutritional losses, and the effectiveness highly depends on weather conditions (Franco et al., 2017).

Mixing with a low-moisture material is demonstrated to be an efficient way to optimize the moisture content and improve the fermentation quality of high-moisture silages (Yan et al., 2019; He et al., 2020b). Besides the active compounds, garlic skin is characterized by low-moisture content. It is reasonable to hypothesize that addition of garlic skin could improve the fermentation quality of high-moisture P. hydridum silage by its active compounds and low moisture content and, thus, the inhibition of undesirable microbial growth during ensiling. To the best of our knowledge, no study has been conducted to estimate the effects of the addition of garlic skin on the silage fermentation quality. Therefore, the objectives of the current study were to investigate silage fermentation characteristics and bacterial community of high-moisture fresh P. hydridum when ensiled with the garlic skin at different ratios.

DM content is a key factor affecting the silage fermentation quality. The inappropriate DM content of the ensiling silage results in seepage and an undesirable fermentation (DM < 28%) or difficulty to pack and poorly fermented (DM > 40%) (Guyader et al., 2018). The DM content of P. hydridum in the current study (16.35%) is similar to that reported in a previous study (Li et al., 2014) and is far lower than the ideal DM content for silage processing, implying the difficulty to obtain high-quality silage without further treatment. Thus, mixing high-moisture P. hydridum with dry materials, e.g., garlic skin, could be an efficient strategy to reduce the moisture content and improve the fermentation quality. The CP content of P. hydridum is higher than that of ryegrass (10.74% DM) and dried corn stover (6.08% DM, Yan et al., 2019), suggesting that P. hydridum is good forage for ruminants. The relatively low CP content of garlic skin (4.06 % DM) suggests that the high mixing ratio of garlic skin might reduce the nutritional value of the silage. The P. hydridum has stemmy structures and accordingly has relatively high fiber content, which is in line with the previous study (Shah et al., 2017). Additionally, the epiphytic microbial community and WSC content are two crucial factors that affect the fermentation quality. For well-preserved silage, epiphytic lactic acid bacteria should reach counts of at least 10⁵ cfu/g of FM (Cai et al., 1999), and WSC content should be greater than 5% DM for propagation and growth of lactic acid bacteria during ensiling (Smith, 1962). In the current study, although both P. hydridum and garlic skin had sufficient WSC concentration and epiphytic lactic acid bacteria, medium counts of undesirable microorganisms, e.g., yeasts, molds, and coliform bacteria, were distributed in both forages, which implies the risk of poor fermentation.

As expected, the differences of DM and CP content among silage treatments were highly related to the DM and CP content of P. hydridum and garlic skin in the current study. In addition, DM contents in all silages consistently decreased as the ensiling progressed, which is in accordance with previous studies (Li et al., 2018; Dong et al., 2020). The DM reduction primarily occurs at the early stage of ensiling due to the microbial breakdown of substrates into liquid and gases, which would be gradually inhibited as the ensiling progressed (Desta et al., 2016). Nevertheless, CP contents remained relatively stable during the 60 days of ensiling, which is consistent with the results reported in a previous study (He et al., 2020b). The CP content is calculated based on the total nitrogen of the silage and the proteolysis during ensiling mainly changes the protein to non-protein nitrogen compounds instead of total nitrogen (Fijałkowska et al., 2015), which may partly explain the relatively stable CP content in the current study. Similar to the results reported by Li et al. (2018), the NDF and ADF contents decreased over a 60-days fermentation. A previous study shows that the addition of cellulase could reduce the content of NDF and ADF in P. hydridum silages (Li et al., 2014). In the future, it would be interesting to evaluate if the addition of garlic skin combined with cellulose could further decrease the NDF and ADF content of the silage.

The pH value is a simple but important criterion to evaluate the silage fermentation quality. pH ≤ 4.2 is generally regarded as an indicator for well-persevered silages (McDonald et al., 1991). High-moisture silage (>70%) and moderate pH (>4.5) favor clostridial fermentation (Muck, 1988). In line with that, the poor fermentation quality of high-moisture P. hydridum silage (GS0) was observed in the current study. The final pH value of the GS0 silage (5.08) after 60 days fermentation was still greater than the threshold pH value despite the consistent decline of pH during the fermentation process. The accumulation of acetic, propionic, and butyric acid content and decreased lactic acid in GS0 silage indicate the failure to inhibit the growth of spoilage organisms, such as clostridia that could ferment lactic acid to butyric acid (He et al., 2020b). Generally, the butyric acid content of silage < 10 g/kg DM is considered desirable, and excessive butyric acid in silage can be associated with reduced silage intake and increased risk of clinical ketosis (Driehuis et al., 2018). The butyric acid of GS0 silage, however, was greater than 20 g/kg DM after 60-days fermentation. During ensiling, the protein is degraded to free amino acids, amines, and ammonia by the proteolytic plant enzymes in the early phase and by bacteria when the pH drops below 5.0, which decreases the efficiency of nitrogen utilization for ruminants (Fijałkowska et al., 2015). Thus, the highest NH₃-N content of GS0 silage among groups reflects the greatest proteolysis during ensiling. Collectively, these results indicate the GS0 silage is poorly fermented and not suitable to be used as ruminant feed.

In the current study, the addition of garlic skin significantly improved the fermentation quality of silage as indicated by lower pH value, acetic acids, and NH₃-N contents, which is similar to the effects of silage additives (e.g., sucrose, glucose, molasses, or glucose) on P. hydridum silage fermentation quality (Li et al., 2014). Moreover, propionic and butyric acid were not detectable in the P. hydridum silages mixed with garlic skin throughout the ensiling process. Clearly, these results indicate that garlic skin inhibited the growth of clostridia and changed the fermentation type. Of note, the DM content of GS10 silage was still considerably lower than the appropriate DM content for the good fermentation quality, but the fermentation quality has been significantly improved. Besides the low moisture content, the garlic skin may exert beneficial effects via its active constituents, e.g., phenylpropanoid (Ichikawa et al., 2003) or alliin (Louis et al., 2012), which have been reported to possess antibacterial activity (Choi et al., 2007; Hemaiswarya and Doble, 2010). Interestingly, the lactic acid contents of GS10 were greater but the pH value was higher than that of those of GS20 and GS30 after 60 days of ensiling. The pH of silage is most correlated with the lactic acid content and the buffering capacity of the materials (Kung et al., 2018), and P. hydridum has a relatively high buffering capacity (Yahaya et al., 2004), which could be the reason for the higher pH value of GS10 silage. In the current study, the counts of yeasts and molds were 2.53–4.09 log₁₀ cfu/g FM in both fresh materials, but the population was decreased to below the detection level (<2 log₁₀ cfu/g FM) after 14 days of ensiling in the current study. Coliform bacteria are the major competitor of lactic acid bacteria and are responsible for silage nutrient loss (Rauramaa et al., 1987). The counts of coliform bacteria were still high after 60 days of ensiling in all silages (3.88–4.11 log₁₀ cfu/g FM) though the counts gradually declined with the decrease of pH during ensiling. It is suggested that coliform bacteria cannot be completely inhibited if the pH > 4.0 (Ni et al., 2015), and thus, the pH of the silages in the current study may not be low enough to inhibit its growth. To the best of our knowledge, the current study is the first to investigate the effects of garlic skin on silage fermentation quality. The findings not only provide a simple way to improve the fermentation quality of high-moisture silage without wilting or additive addition, but also open new avenues for the usage of garlic skin, which is mainly treated as industrial waste. Moreover, the strong aromatic odor of the well-fermented silage covered the pungent odor of garlic skin, which could potentially increase its palatability.

Basically, ensiling is a complex and dynamic microbial fermentation process. Ren et al. (2019) observed that the Chao index increased in the sugarcane top silage from days 2 to 60 and then decreased after days 60, which is similar to the findings in the current study. Moreover, the bacterial diversity declined from days 30 to 60 of ensiling in the P. hydridum silage mixed with garlic skin, but not in the P. hydridum silage ensiled alone. The high bacterial diversity of fresh material is gradually decreased with the lactic acid bacteria becoming predominant during successful silage fermentation (McDonald et al., 1991), which could be the reason for the reduction of bacterial diversity in P. hydridum silage mixed with garlic skin. These results show an increase in anaerobic stability in GS10, GS20, and GS30 silages but not in GS0 silage, which is in line with the fermentation quality data as mentioned above. Additionally, the principal coordinate analysis revealed that the GS0 silage was significantly separated from GS10, GS20, and GS30 at both days 30 and 60 of ensiling. Therefore, the improved fermentation quality in silages mixed with garlic skin seems to be related to the shift of the microbial community from undesirable microorganisms to favorable microorganisms.

The addition of garlic skin considerably changed the relative abundance of bacterial communities compared with P. hydridum silage ensiled alone. Enterobacter is a common genus during silage fermentation and is undesirable due to its competition for WSC and NH₃-N production (Muck, 2010). As discussed above, the high abundance of Enterobacter at days 30 of ensiling could be related to the high moisture content of P. hydridum and medium pH value in the current study. He et al. (2020b) and Wu et al. (2020) also observed a high relative abundance of Enterobacter in Pennisetum purpureum silage (4.68–51.33%) and in high-moisture corn stalk silage (28.9–39.3%), respectively. However, the relative abundance of Enterobacter at days 60 of ensiling decreased, which could be associated with the decreased pH value caused by the significantly increased relative abundance of Lactobacillus (45.8–68.2%). Ensiling is an anaerobic process, and the higher relative abundance of anaerobes, e.g., Lactobacillus, produces lactic acid to inhibit the acid-intolerant Enterobacter (Dunière et al., 2013). Therefore, the addition of garlic skin in P. hydridum silage shifted the mixture of homo- and hetero-lactic fermentation to a more homo-lactic fermentation at days 60 of ensiling, which is more efficient in lactic acid production and results in fewer nutrients lost.

In the current study, the relative abundance of Clostridium was greatest in GS0 silages, which led to considerable nutrient losses and proteolysis and coincided with the highest butyric acid and NH₃-N content in GS0 silages as aforementioned. He et al. (2020b) also observed that Clostridium was the first dominant genus in the high-moisture stylo silage. In addition, garlic skin increased the relative abundance of Kosakonia and Pantoea at days 30 of ensiling, and the relative abundance of Kosakonia was not substantially inhibited by the decreased pH value at days 60 of ensiling. Kosakonia is a novel reclassified branch from Enterobacter and is related to the fixation of nitrogen and production of xylanase (Chin et al., 2017). Pantoea has been reported to have negative correlations with silage pH, NH₃-N concentration, and yeast and mold counts (Ogunade et al., 2018). He et al. (2020a) propose that Kosakonia and Pantoea might contribute to the reduction of NH₃-N of silage. Furthermore, at days 60 of ensiling, the relative abundances of Prevotella and Caproiciproducens in GS0 silage were significantly greater than other silages. Prevotella is a dominant proteolytic bacterium in the rumen and could produce propionate during fermentation (Chen et al., 2019). Caproiciproducens is a recently defined genus and is also observed in Italian ryegrass and corn stover mixed silage (Yan et al., 2019) and stylo silage (Wang C. et al., 2019). The greater content of NH₃-N, acetate, and butyrate could be related to the Prevotella and Caproiciproducens in GS0 silage, but the exact biological function of these bacteria in silage fermentation should be further investigated in future studies.

The bacterial functions prediction analysis showed that the several relative abundance pathways (i.e., signal transduction, energy metabolism, nucleotide metabolism, etc.) were significantly different in GS10 silage compared with other silages at days 30 of ensiling. These differences might be related to the greater growth of Lactobacillus in GS10 silage at days 30 of ensiling. Bai et al. (2021) also observed the opposite trend in nucleotide and energy metabolism in alfalfa silage, but the reason for this phenomenon is not clear and needs further investigation. At days 60 of ensiling, GS0 silage had lower relative abundance pathways of lipid metabolism and greater relative abundance pathways of metabolism of cofactors and vitamins compared with other silage treatments. Liu et al. (2020) suggest that the decreased Firmicutes/Bacteroidetes ratio may inhibit lipid metabolism–related gene expression. The significantly lower relative abundance of Lactobacillus, which is a classical type of Firmicutes, in GS0 silage could explain the lower lipid metabolism in the current study. Notably, the data about the predicted bacterial function should be interpreted cautiously, and direct mRNA measurements or transcriptomics should be employed to verify these metabolic pathways involving in silage fermentation (Keshri et al., 2018).

The current study illustrates that co-ensiling of high-moisture P. hydridum with garlic skin could significantly decrease pH value, butyric acid, NH₃-N, and increase lactic acid production. Moreover, the relative abundances of favorable bacteria, such as Lactobacillus, were increased while that of undesirable bacteria, such as Clostridium and Enterobacter, were decreased. These results suggest that the addition of garlic skin could be a simple way to improve the fermentation quality and preserve protein of high-moisture P. hydridum silage via optimizing the moisture content and, thus, the inhibition of undesirable microbial growth during ensiling. Future research focusing on the utilization in animal husbandry and the related economic analysis are needed for better application in practice, and metagenomic and metabolomic analysis should also be considered to better characterize bacterial species and their functionality during ensiling.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/, PRJNA755145.

JC, GH, and ZZ designed the study and wrote the manuscript. HX, HQ, HZ, and YS performed the experiments. YL analyzed the data. YZ was involved in the manuscript revision. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.