In uence of Various Additives On The Fermentation Quality And Bacterial Community of High-Moisture Whole-Plant Quinoa Silage


 To explore the potential of whole-plant quinoa (WPQ) as a high-protein source for livestock feed, this study evaluated the effects of additives on the fermentation quality and bacterial community of high-moisture WPQ silage. High-moisture WPQ was ensiled either untreated (control) or treated with cellulase (E), molasses (M), LAB inoculant (L), a combination of cellulase and LAB inoculant (EL), and a combination of molasses and LAB inoculant (ML). The fermentation quality and bacterial community after 60 days of ensiling were analyzed. Naturally fermented WPQ exhibited acetic acid-type fermentation dominated by enterobacteria, with low lactic acid content (21.1 g/kg DM), and high pH value (6.43) and NH3-N production (182 g/kg TN). Adding molasses alone or combined with LAB inoculant shifted the fermentation patterns toward increased intensity of lactic acid fermentation, lowering the pH value (<4.60), NH3-N content (<130 g/kg TN) and total abundance of enterobacteria (<19%), and increasing the lactic acid content (>61.5 g/kg DM), lactic/acetic acid ratio (>1.42) and the relative abundance of Lactobacillus (>70.5%). The results suggested that the lack of fermentable sugar could be the main factor of restricting extensive lactic acid fermentation in WPQ silage. Supplementing fermentable sugar or co-ensiling with materials with high WSC content and low moisture content could be beneficial strategies for producing high-quality WPQ silage.


Introduction
Quinoa (Chenopodium quinoa Willd.) is native to the Andean region of South America and is a traditional food for the indigenous peoples. In most recent years, quinoa seed has received much attention because of its exceptional nutritional value and potential health bene ts (Vega-Gálvez et al. 2010). Quinoa is rich in proteins, lipids, bres, vitamins and minerals, and contains signi cant amounts of bioactive compounds such as phytosterols, phytoecdysteroids, bioactive peptides and phenolic compounds ). Besides high nutritional value, quinoa can survive under harsh conditions, withstanding a wide range of temperatures (-4∼38℃) and pH (6.0~8.5), low rainfall (50 mm/year) and high salinity (40 ms/cm) ). These characteristics allow this plant to grow well in many marginal conditions. The vegetative part of quinoa plant is rich in protein and has far more biomass accumulation compared with the grain (Basra et al. 2014). Therefore, whole-plant quinoa (WPQ) is believed to have great potential for feeding animals. In some developed regions of livestock husbandry, fresh WPQ has been used to feed animals ). However, the WPQ biomass is largely accumulated in very short time during harvest. The main stem axis and high moisture content make quick drying of a large biomass of WPQ di cult, limiting its use as a dried forage. Based on economical and practical feasibility, ensiling could be the best option for WPQ preservation. This technique depends on epiphytic lactic acid bacteria (LAB) converting water soluble carbohydrates (WSC) to organic acids mainly lactic acid under anaerobic conditions, which creates an acid environment to inhibit spoilage organisms and achieves the goal of maintaining the original quality of the moist forage as much as possible. However, the WPQ is lower in WSC and extremely higher in moisture content compared with the traditional forage crops, such as Italian ryegrass and whole-crop maize (Li et al. 2013). These traits may be the restriction factors for the success of ensiling under natural fermentation conditions. Hence, silage additives or appropriate measures are required at the time of WPQ ensiling.
Homofermentative LAB is the most common biological additive in silage preservation. Previous studies have shown that LAB inoculation successfully directed the fermentation, reduced the dry matter (DM) loss and improved the fermentation quality of alfalfa silage (Guo et al. 2018; Zheng et al. 2017).
Chemical additives, such as cellulase and molasses, are also often used as fermentation stimulants to increase the amount of readily fermentable sugars for LAB. Adding cellulase produced more WSC and improved the organic matter digestibility of silage by the degradation of plant cell walls (Mu et al. 2020; Tian et al. 2014). Adding molasses to king grass silage enhanced the fermentation by promoting the lactic acid production during the early stages of ensiling . However, the possible bene cial effects of additives on silage fermentation depend on the properties of the forage crops being ensiled ). To our knowledge, few or no studies have identi ed the effects of the additives on WPQ silage. The information is valuable in terms of producing high-quality WPQ silage.
Silage fermentation is a dynamic process of microbial community succession and metabolite changes.
Recent advances in culture-independent analyses, such as high-throughput sequencing technology, have enabled microbial communities to be de ned with a degree of detail that is impossible using classical microbiology (Dong et al. 2019b). Understanding of the microbial communities involved in the ensiling process would provide an insight into approaches to improve the forage conservation (Tian et al. 2021;Zhao et al. 2021). Therefore, the objective of this study was to investigate the effects of different additives on the fermentation quality and bacterial community of high-moisture WPQ silage.

Plant Materials and silage preparation
Quinoa was grown at Xinyang Agricultural Experiment Station of Yancheng City (N33°52′, E120°44′, Yancheng, China). The quinoa was planted in three experimental blocks with same tillage, irrigation and fertilization practices. These experimental blocks were kept as replicates throughout the whole experiment. After nine weeks of growth, the entire plant was harvested at the early blooming stage 10 cm above ground level and then was chopped into a theoretical length of 1~2 cm using a forage cutter. The chopped WPQ from each experimental block was used for ensiling after 20 h of wilting on a plastic lm placed at a ventilated lobby.
The wilted WPQ was divided into six groups. The six groups were randomly assigned to one of the following treatments: (1) control; (2) cellulase (E); (3) molasses (M); (4) LAB inoculant (L); (5) a combination of cellulase and LAB inoculant (EL); (6) a combination of molasses and LAB inoculant (ML). The application rate of molasses was 1% on fresh matter (FM). The cellulase was a mixture of cellulase and a hemicellulose (supplied by Oddfoni Biological Technology Co., Ltd., Nanjing, China) and applied at a rate of 0.05 mg/g FM as suggested by Chen et al. (2019). The LAB inoculant was supplied by Institute of Grass Ensiling and Processing of Nanjing Agricultural University, mainly consisting of Lactobacillus plantarum. The application rate of LAB was 1.0×10 6 colony forming units (cfu)/g of FM, according to the manufacturer's speci cation. The additives were all diluted in deionized water and applied in liquid forms. The control silage was received with equal volume of deionized water. The treated WPQ (about 200 g) were packed into vacuum-sealing polyethylene plastic bags (20×30 cm) and heat-sealed after vacuum completed in the bag. A total of 18 bags (6 treatments× 3 replicates) were prepared and stored at room temperature (20~25℃). These bags were opened after 60 days of ensiling and sampled for further analysis.

Chemical and microbial analysis
The pre-ensiling and ensiled WPQ were sampled for chemical composition and microbial population analysis. Approximately 100g sample was oven-dried for 48 h at 60°C for DM measurement and ground to pass 1-mm screen with a laboratory pulverizer (FW100, Taisite Instrument Co., Ltd. For microbial population analysis, ten grams of sample was thoroughly mixed with 90 mL of sterilized saline solution on a shaker at 120 rpm for 2 h. After that, one hundred microliters of solution were used and serially diluted with sterilized saline solution to 10 −2~1 0 −5 for culture-medium plating. The LAB and Enterobacteriaceae were, respectively, counted on de Man, Rogosa, Sharpe and Violet Red Bile Glucose Agar mediums after anaerobically incubated at 37°C for 48 h. Aerobic bacteria were counted on nutrient agar under aerobic conditions at 37°C for 24 h. Yeasts were determined on potato dextrose agar under aerobic conditions at 30°C for 3 days. The remaining solution was ltered into a 50-mL centrifuge tube with 4 layers of medical gauze and stored at -80°C for DNA extraction.

Bacterial community analysis
The frozen solution for DNA extraction were thawed at 4°C and then centrifuged at 12,000×g for 30 min to obtain a pellet for subsequent DNA extraction. The DNA extraction was conducted using the Raw sequences were processed using FLASH (version 1.2.11). The QIIME quality control process (version 1.7.0) was used to discard low-quality sequences (quality scores <20). Chimeric sequences were identi ed and removed using UCHIME (version 1.7.0). Only sequences at least 200 bp long after quality ltering were grouped into operational taxonomic units (OTUs) at 97% similarity level. The alpha-diversity (Shannon, ACE, Chao1 and Coverage indices) and beta-diversity were analyzed using QIIME. Community structures of bacteria was analyzed from phylum to genus levels using the Silva database with a con dence threshold of 70%. Principal coordinate analysis (PCoA) was constructed to visualize the variation in microbial communities between samples. To illuminate the interactions among microbes, correlations of microbes were analyzed by Spearman's rank correlations between the abundant genera (top 10) and a network was created with Ctytoscape (version 3.9.0) to visualize the correlations.
Spearman correlation heatmap was constructed by the R software (version 3.3.1) to show the relationships of bacterial community and fermentation parameters. The sequences data reported in this study was archived in the Sequence Read Archive (SRA) with the accession number PRJNA795324.

Statistical Analyses
Data on chemical and microbial compositions of fresh and ensiled WPQ were tested to one-way analysis of variance (ANOVA) using the SPSS 22. Tukey's multiple comparison was used for the means separation. Signi cant differences were declared when P < 0.05.

Effects of additives on fermentation characteristics and chemical composition of WPQ silage
The effects of additives on the fermentation characterstics of WPQ silage are given in Table 2. Additive affected (P<0.05) fermentation parameters except for propionic acid and ethanol content. Naturally fermented WPQ exhibited acetic acid-type fermentation, with high pH value (6.34) and acetic acid content (85.4 g/kg DM) as well as low lactic/acetic acid ratio (0.25). Among the additives, addition of M and ML to WPQ silage increased the intensity of lactic acid production, indicated by increases (P<0.05) in lactic acid content and lactic/acetic acid ratio, and decreases (P<0.05) in pH value and acetic acid contents. Furthermore, M and ML addition decreased (P<0.05) the propionic acid and NH 3 -N contents, whereas E and EL addition increased (P<0.05) the NH 3 -N contents. The effects of additives on the chemical composition of WPQ silage are shown in Table 3. The DM, WSC and CP contents were affected (P<0.05) by the additive. Compared with other silages, the DM and CP content was higher (P<0.05) in the M-treated silages (M and ML silages). Addition of ML increased (P<0.05) the residual WSC content compared with control silage.

Correlation analysis of bacterial community and silage characteristics
Spearman correlations between bacterial community and silage characteristics are presented in Fig.   5. Lactobacillus statistically positively correlated with lactic acid, lactic/acetic acid ratio and WSC, while negatively correlated with pH, acetic acid and propionic acid. Citrobacter, Providencia, Proteus and Enterococcus were grouped together. They were positively correlated with acetic acid, propionic acid and pH, whereas negatively correlated with WSC, CP, lactic acid and lactic/acetic acid ratio. In addition, Proteasu also showed statistically positive correlation with NH 3 -N.

Discussion
The WPQ characteristics before ensiling The crude protein content of WPQ was as high as 191 g/kg DM, a level comparable with that of alfalfa harvested at squaring stage (Wang and Yu, 2020). High crude protein content, coupled with the high biomass production, suggests that WPQ can be used as a high-protein source for livestock feed. It is generally considered that the ideal moisture range for ensiling is 600-700 g/kg FM since such a moisture can maintain the vigorous growth of LAB as well as prevent the undesirable clostridial fermentation (Du et al. 2021). However, Moser (1995) reported that the stems dried 10 to 15 times slower than the grass leaves. Large amount of stem-stored water increases the di culty of wilting WPQ to an ideal moisture in a short time. After 20 h of wilting, the moisture content of WPQ was still as high as 858 g/kg DM, which could be the great challenge for making high-quality WPQ silage. The WSC content and epiphytic LAB count of the material are another two determinants of fermentation quality ). Zhang et al. (2015) suggested that the minimum WSC level for successful fermentation is 60 g/kg DM. Based on criterion, fresh WPQ failed to meet the requirement. However, the LAB number was adequate, exceeding the recommended level (5 log 10 cfu/g FM) as suggested by Mu et al. (2020).

Effects of additives on fermentation characterstics and chemical composition of WPQ silage
The lactic acid is the desired fermentation product as it is the main driver of lowering the silage pH. However, naturally fermented WPQ (control silage) exhibited acetic acid-type fermentation indicated by high pH value (6.43) and acetic acid content (85.4 g/kg DM) as well as low lactic/acetic acid ratio (0.25). Moderate content of acetic acid (30~40 g/kg DM) in silage can be bene cial, because they inhibit yeasts, resulting in improved stability when silage is exposed to air (Kung et al. 2018). However, as acetic acid is a weaker acid than lactic acid, excessively high production of acetic acid (>60 g/kg DM) represents ine cient fermention and is accompanied by poor DM recovery in silage ). In particular, poorly fermented silages with high acetic acid contents also undergo substantial protein degradation and result in accumulations of detrimental compounds (e.g., biogenic amine) that decrease the intake and negatively affect the animal production (Muck, 2010). High acetic acid contents is frequently observed in extremely wet silages dominated by enterobacteria, clostridia or heterolactic acid bacteria (Kung et al. 2018). This is because high level of moisture dilutes the fermentation acids and more lactic acid production by LAB is required to inhibit the activity of acetic acid-producing microbes. In the experiment, high moisture, together with the low WSC content and relatively high buffering capacity, may hamper the pH decline and allow the acetic acid-producing microbes to be active for long time during WPQ ensiling. Among the additives, adding molasses (M and ML) successfully shifted the fermentation patterns toward increased intensity of lactic acid production, indicated by lower pH values (<4.56), higher lactic acid content (>60.5 g/kg DM) and lactic/acetic acid ratio (>1.40) than the control silage, suggesting the improvement of fermentation quality. Cellulase addition failed to affect the fermentaion in WPQ silage. Commercial cellulases are known to have the optimum activity at pH 4.5-5.4 (Henderson and McDonald, 1977). The marginal effect of cellulase addition on silage fermentation was probably because the high silage pH (>5.4) was not within the optimum range for cellulase activity. This may weaken the function of cellulase in the silage. Similarly, Kung et al. (1991) have observed that if pH optima have not been achieved in the silage, cellulase effect will be partially negated. LAB inoculation also did not elicit the signi cant effects on silage fermentation. The LAB used in the experiment belongs to lactobacilli and is capable of quickly producing large amounts of LA by fermenting a wide variety of substrates. Little value of LAB inoculation to the WPQ silage re ects that LAB number or activity may be not the restricting factor for lactic acid fermentation. Overall, the responses of fermentation to the additives suggest that the lack of fermentable sugar could be the main factor of restricting extensive lactic acid fermentaion in WPQ silage.
The presence of propionic acid, butyric acid and ethanol in silages are unacceptable given that their generation is an energy-waste metabolism. The results showed the propionic acid contents (<9.00 g/kg DM) in all WPQ silages were within the acceptable range, as suggested by Kung et al. (2018). Moist forage that is poorly preserved often contains high concentrations of butyric acid due to the clostridial activity (Zheng et al. 2017). In the study, trace quantities of butyric acid (< 5g/kg DM) suggested that there was little or no clostridial activity in the WPQ silages. Ethanol has little preservation effect during ensiling, and it causes extremely higher DM and energy losses. According to Kung et al. (2018), over 30-40 g/kg DM of ethanol production in silage is associated with the action of yeast. The results showed that ethanol contents in all WPQ silages were < 20 g/kg DM, suggesting that ethanol was mainly produced by microbes such as heterolactic acid bacteria and enterobacteria.
The proteolysis by plant and microbial enzymes lowers the nutritive value of ensiled forage by degrading forage protein into non-protein fractions, such as peptides, free amino acid and NH 3 (Guo et al. 2008). As an end-product, NH 3 -N can be taken as an indicator of protein degradation in silage. Generally, over 100-2018). High NH 3 -N content (162 g/kg TN) in the control silage indicated that WPQ protein was badly preserved under natural fermentaion conditions. Adding molasses (M and ML) decreased the propionic acid and NH 3 -N contents. It indicates that molasses addition is bene cial for suppressing ine cient fermentations and improving the protein preservation of WPQ silage. By contrast, cellulase increased the NH 3 -N contents compared with control silage. Similar phenomenon has been reported by Kung et al. (1991). They ascribed it to the N contributions from the enzyme complex.
The molasses-treated silage had higher DM content than other silages, due to the addition of molasses. Higher residual WSC is nutritionally desirable because it is rapidly digestible in the rumen (McDonald et al. 1991). Addition of ML to WPQ silage increased the residual WSC contents, similar to the ndings of Guo et al. (2014) and Ebrahimi et al. (2014). It was presumably because more fermentable sugars plus high-activity LAB resulted in faster rates of lactic acid production and pH decline, quickly suppressing the growth of undesirable microbes and thus retaining more WSC in the silage. The CP content increased with the molasses addition, which may be linked with the decreased extent of protein degradation. Compared fresh WPQ, ensiled WPQ had lower NDF and ADF contents. Similarly, Kung et al. (1991) reported that ber degraded naturally during ensiling as a result of some silage micro ora producing extracellular cellulases and hemicellulases. NDF and ADF represent the less digestible ber portions for animals. The decreased NDF and ADF contents in ensiled WPQ silage suggested the improvement of nutritional value.

Effects of additives on bacterial community of WPQ silage
Ensiling is a bacterial-driven process in which the types and abundances of bacteria involved play a critical role in fermentation quality (Guan et al. 2018;He et al. 2020). As far as we know, this is the rst report concerning the bacterial community of WPQ silage. Good's coverage (>99%) indicated that sequencing depth had adequately captured most of the bacterial communities in the samples. Alpha diversity re ects the bacterial diversity and species richness in a single sample. The Shannon indice are used to measure bacterial diversity, whereas Chao1, ACE indices and OTUs number are measures of species richness. Generally, considerable reductions in bacterial diversity and species richness are expected to occur during successful ensiling because complex microbial communities in fresh materials will be replaced by the development of LAB (Du et al. 2021). However, ensiled WPQ had comparable or increased bacterial diversity compared with the fresh WPQ sample. It indicated high abundances of non-LAB microbes present in the WPQ silages. All silage samples had lower species richness than did fresh WPQ sample. This might be attributed to the decline in relative abundance of epiphytic aerobic bacteria, which are unable to survive under the anaerobic conditions (Yuan et al. 2020).
PCoA analysis is a method to explore and to visualize similarities or dissimilarities of the bacterial communities. The clear separation between fresh WPQ samples and silage samples suggests that microbial communities greatly changed during ensiling process. Among the silage samples, M and ML silage samples were clearly separated from the control silage samples, suggesting that these treatments signi cantly affected the microbial community in WPQ silage.
Proteobacteria is a major phylum of Gram-negative bacteria that includes a wide variety of pathogenic genera, such as Escherichia, Salmonella, Vibrio, Helicobacter, Yersinia and Legionellales. On the phylum level, the dominance of Proteobacteria suggested the large amount of undesirable microbes present in fresh WPQ. After ensiling, Firmicutes increased dramatically and became abundant in the WPQ silage. As all LAB belong to Firmicutes, the increased abundances of Firmicutes are bene cial to the silage fermentation. However, for control and L-treated silages, Proteobacteria remained to be most abundant phylum, suggesting the failure of LAB to dominate the microbita during ensiling.
The LAB are desirable bacteria contributing to the fermentation quality of silage. Members of the Leuconostocs, Lactococcus, Pediococcus and Weissella genera are the LAB most frequently detected on standing plants and contributes to the initial decline in silage pH (Dong et al. 2019b;). Similar to those reported in forage (Ali et al. 2020;Cai, 1999), LAB genera represent only a very small fraction of the epiphytic microbiota in the fresh WPQ. Silage fermentation is an anaerobic, closed, solidfermentation system. High abundance of Lactobacillus is frequently observed in this system because of high acid-resistant nature. Lactobacillus can quickly ferment a wide variety of substrates to produce large amount of lactic acid. Therefore, they are crucial bacteria responsible for the largest amount of lactic acid production and pH decline in silage (Costa et al. 2021). In the study, adding M and ML increased the relative abundance of Lactobacillus, explaining why lactic acid contents increased in these WPQ silages. However, it was observed that cellulase addition (E and EL) also promoted the development of Lactobacillus to some extents but did not result in a signi cant increment of lactic acid production. Such a discrepancy might be explained by the fact that sugars released from cellulose hydrolysis did not reach the signi cant level to affect the fermentation patterns in WPQ silage. Dong et al. (2020) stated that under WSC-de cient conditions, facultatively heterofermentative strains of Lactobacillus (e.g. Lactobacillus plantarum) process heterofermentative activity rather than homofermentative activity.
Silage fermentation is a complex process involving interactions among many factors. Ideal fermentation is not always obtained, and sometimes undesirable bacteria may dominate the silage. Enterobacteria represent a major group of undesirable bacteria in silage. They compete with LAB for the available sugars, and in addition they can degrade protein and result in production of toxic compounds such as biogenic amines and branched fatty acids (Muck, 2010). In this study, over 5 genera of Enterobacteria were detected in large quantities in the WPQ silage, and the total abundance was > 50% in the control silages. The principal fermentation product of Enterobacteria is acetic acid, not lactic. It explained the high acetic acid production and substantial protein degradation in the WPQ silage. Enterobacteria will not proliferate at low pH, and their population will be controlled by acidi cation (McDonald et al. 1991). Therefore, Enterobacteria was suppressed with the increased lactic acid production by Lactobacillus (Fig. 4). Among the Enterobacteria, Citrobacter, Morganella and Providencia were particularly suppressed. Citrobacter share all the general properties and biochemical characteristics of the family Enterobacteriaceae. They are found in a variety of environmental sources, including soil and water, and in the human intestines. Citrobacter has been found to associate with the losses of polyunsaturated fatty acid, α-tocopherol and β-carotene in alfalfa silage (Zong et al. 2021). Morganella previously belongs to family Enterobacteriaceae. They conduct anaerobic respiration and found to associate with the increase in acetic acid content in high-moisture Italian ryegrass silage (Li and Nishino, 2013). Providencia is closely related to the Morganella. These bacteria have been abundantly dectected in oat and barley silages (Liu et al 2019; Jia et al. 2021). Some members of Providencia were described to associate the diseases in animals (Jia et al. 2021).

Correlation analysis of bacterial community and silage characteristics
Studying the correlations between bacterial community and silage characteristics would give us a deep understanding of the key bacteria to silage quality. Lactobacillus positively correlated with lactic acid, lactic/acetic acid ratio and WSC, while negatively correlated with pH, acetic acid and propionic acid. It con rmed that Lactobacillus have played a crucial role in improving the fermentation quality and nutrients preservation of WPQ silage. Citrobacter, Providencia and Morganella are known to be able to produce acetic acid as the main product. Positive correlations of acetic acid with them suggests that they are main contributors to the acetic acid production in WPQ silage. Proteus has been identi ed as NH 3producing bacteria in fermented skate (Raja kenojei). Positive correlation between Proteus and NH 3 -N, suggests that their important role in producing NH 3 -N during WPQ ensiling. Enterococcus are cocci LAB and can only grow in pH > 4.5 (Cai, 1999). Weak acid-resistance of this LAB may explain their positive correlation with pH, acetic acid, propionic acid and butyric acid, and negative correlations with lactic acid and WSC.

Conclusion
In conclusion, naturally ensiled WPQ with high moisture content was prone to acetic acid-type fermentation dominated by enterobacteria, resulting in poor fermentation quality and substantial protein degradation. Adding molasses alone or combined with LAB successfully shifted the fermentation patterns toward increased intensity of lactic acid production by promoting Lactobacillus and decreasing the abundance of enterobacteria. The results reveal that the lack of fermentable sugar could be the main factor of restricting extensive lactic acid fermentaion in WPQ silage. Supplementing fermentable sugar or co-ensiling with materials with high WSC content and low moisture content could be bene cial for producing high-quality WPQ silage.        Relationships among top 10 genera during WPQ ensiling. A connection stands for a signi cant (P < 0.05) and strong (Spearman's |p| >0.5) correlation. Size of each node is proportional to the relative abundance, and the nodes are colored by phylum. The thickness of each connection (edge) between two nodes is proportional to the value of Spearman's correlation coe cient (p). The color of the edges corresponds to a positive (red) or negative (blue) relationship.