Air-dried corn stover was collected post-harvest from Yuzhong county (Lanzhou, PR China). The moisture content of air-
-dried corn stover chopped to lengths of 15-20 mm on fresh mass basis was 10.23%. The cabbage waste was collected from the Qilihe vegetable market (Lanzhou, PR China).

Representative samples of silage after varied intervals were oven-dried and ground through a 1-mm screen prior to chemical analysis. Dry matter content was determined by oven drying the material at 105 °C (Keweiyongxing Instrument Co. Ltd, Beijing, PR China) to constant mass, and subsequently ground in a cutting mill (Kanghe Machinery Manufacturing Co. Ltd, Jiangyin, PR China) for further analysis. Water-soluble carbohydrates were determined by the anthrone-sulfuric acid method (20). The neutral and acid detergent fibre and acid detergent lignin content were determined according to the Van Soest method (21). Hemicellulose content was calculated as the difference between neutral and acid detergent fibre, while cellulose content was determined as the difference between acid detergent fibre and acid detergent lignin. Total nitrogen content was determined by the Kjeldahl method according to Tangka (22). The crude protein content was calculated as nitrogen mass fractions multiplied by conversion constant (6.25) (11).

The silage extract was prepared by mixing 25 g of various silage samples with 225 mL of distilled water in a homogenizer (JJ-2; Guohua Instruments Co. Ltd, Suzhou, PR China). The slurry mixture was then filtered through four layers of cheesecloth, and the filtrate was used for determination of pH, ammoniacal nitrogen, lactic acid and volatile fatty acid contents. The pH was measured immediately using a pH meter (UB-10; Denver Instruments Co. Ltd, Beijing, PR China). Part of the volume was centrifuged (H2500R; Xiangyi Instrument Co. Ltd, Changsha, PR China) at 3000×g for 20 min at 4 °C and the supernatant was filtered through a 0.22-μm syringe filter (MET Laboratories Inc., Baltimore, MD, USA) and used to quantify lactic acid and volatile fatty acids. Lactic acid was determined using a biosensor (SBA-
-40X; Bosheng Biotechnology, Jinan, PR China), and 0.4 µL of samples were analyzed to determine ethanol and volatile fatty acid contents using gas chromatograph (GC-2014C; Shimadzu, Kyoto, Japan), which was equipped with a pillar KB-Wax and flame ionization detector. The temperatures of sample injector, detector and column were set at 250, 250 and 130 °C, respectively, and nitrogen gas was used as the carrier gas. All of the chemical analyses were conducted in triplicate, and the results were expressed on dry mass basis, except for dry mass content (expressed as % on fresh mass basis) and ammoniacal nitrogen (expressed as % of total nitrogen) (11).

The silages were randomly subsampled from various layers and then mixed to generate a homogeneous sample for microbial community analysis. High-throughput 16S rDNA sequencing technology was adopted to analyze the changes of population structure during ensiling. Total DNA was extracted using Water DNA isolation kit (FuJi Biotechnology Co. Ltd, Chengdu, PR China) according to the manufacturer’s instructions. For sequencing, the 16S rRNA genes were amplified with primers 338F (5-ACTCCTACGGGAGGCAGCA-3) and 806R (5-GGACTACHVGGGTW TCTAAT-3) targeting the bacterial and archaeal V3-V4 region in the 16S rRNA gene with a 10-mer barcode at the 50 end of primer 338F. To minimize PCR bias, two PCR reactions were set up for each sample, and the PCR products in the replicate reactions were pooled (13). The amplicons from each sample were pooled with an equimolar concentration and sequenced with the Illumina MiSeq platform. The details of PCR and related procedures were described previously by Li et al. (13). Sequencing reads with primer sequences were removed.

Data analysis from each ensiling period was performed using SPSS v. 20 software (23), and one-way analysis of variance (ANOVA) with Tukey’s multiple comparison test at a 0.05% significant level was used to determine the effects of additives.

The chemical compositions of air-dried corn stover, cabbage waste and their mixture before storage are shown (Table 1). The content of dry matter in air-dried corn stover was found to be 89.8%, while neutral detergent fibre, acid detergent fibre and acid detergent lignin contents in air-dried corn stover were 76.5, 46.5 and 7.6%, respectively. The air-dried corn stover used in this study contained approx. 39.0% cellulose and 30.0% hemicellulose, which represent biodegradable polysaccharides that can be converted into bioenergy through biological conversion. It is noteworthy that the moisture content in air-dried corn stover was merely 10.2%, which is relatively low. Furthermore, the content of water-soluble carbohydrates was 8.1%, which was slightly less than the typical values (i.e. 8-10%), due to free sugar consumption and water evaporation during wilting and field drying (24). The moisture and water-soluble carbohydrate content in cabbage waste were 91.4 and 14.6%, respectively, which were both significantly higher than those of air-dried corn stover. Therefore, the trans-seasonal mixed storage of air-dried corn stover and cabbage waste was feasible on the basis of strong complementarity in terms of nutrient constituents and moisture content.

Fig. 1 shows that the relative abundance of undesirable microorganisms or non-lactic acid bacteria in cabbage waste was about 99.1%, mainly including Pseudomonas (48.4%), Pantoea (17.1%), Flavobacterium (16.3%), Yersinia (6.5%), Enterobacter (3.6%) and Exiguobacterium (0.04%), which indicated that LAB were scarcely present in pre-ensiled cabbage waste. Once discarded and exposed to outdoor weathering, cabbage waste is perishable and causes heavy odour and leachate, and thus utilization of this material is essential for environmental protection. With respect to the relative abundance of LAB in air-dried corn stover, nearly 29.8% of the community was comprised mainly of Carnobacterium, Enterococcus and Lactococcus. Thus, while cabbage waste brings nutrients and moisture, air-dried corn stover contributes LAB, underscoring the complementarity of both pre-ensiled materials. These results are supported by Li and Nishino (25), who reported that Pseudomonas, Lactobacillus and Rahnella are often found in pre-ensiled wilted Italian ryegrass crops. Heron et al. (26) also reported that Erwinia and Rahnella often dominate in fresh grass, but these genera were not discovered in air-dried corn stover and cabbage waste, indicating that bacterial populations may vary according to pre-ensiled material variety, geographic area, climate or growing stage.

As can be seen in Table 2, the dry matter content in negative control group, and treatments with formic (both high and low dosage) or acetic (high dosage) acid showed a decreasing trend. However, with the addition of low dosage of acetic acid the dry matter was observed to increase first and then decrease. Smaller dry matter reduction than in the control indicated that the addition of formic or acetic acid was beneficial for preserving dry matter, especially during 90-170 days of storage. Furthermore, it is well-known that high water-soluble carbohydrate content is desirable for the silage process, as it provides readily accessible substrates for bacteria to produce lactic acid and other organic acids, which can reduce the pH of silage and facilitate the preservation of feedstock (27). When the period of ensiling was extended, the water-soluble carbohydrate content in all five groups was found to significantly (p<0.05) decrease, compared to their content before storage. However, the water-soluble carbohydrate content increased significantly (p<0.05) after addition of formic or acetic acid compared with the negative control group during storage (except the silage treated with low dose of formic acid on day 170), which is in agreement with the results found by Li et al. (19), who indicated that application of formic or acetic acid results in increased water-soluble carbohydrate and crude protein contents. On the other hand, substantial changes in the amounts of structural carbohydrates occurred during ensiling. The contents of dry matter and structural carbohydrates were found to be increased to some extent, which was congruent with the results reported by Van Ranst et al. (28). In previous work it was concluded that the content of dry matter and structural carbohydrates tended to decrease with storage duration (6); however, this trend was inconsistent among various silage types.

The air-dried corn stover before and after ensiling showed distinct differences in their chemical compositions. With time, the holocellulose content increased in negative control group and treatments with formic acid (both high and low dosage), but it first increased and then decreased in silages treated with acetic acid (both high and low dosage). Holocellulose content during ensiling did not increase in silages treated with acetic or formic acid. Instead, the holocellulose content in the majority of treatments decreased to different extents. A similar trend was found for the cellulose content in silages treated with acetic or formic acid. However, the holocellulose content was well preserved in silage treated with acetic acid compared with negative control group during storage for 170 days. The acid detergent lignin content in silages treated with acetic or formic acid was shown to have an increasing tendency compared with the negative control. The addition of formic acid likely contributed to the decrease of neutral detergent fibre and the addition of acetic acid was conducive to the decrease of acid detergent fibre. Pakarinen et al. (9) studied the chemical composition of whole crop corn before and after 4- and 8-month ensiling, and the results indicated that the water-soluble carbohydrate and cellulose contents increased with the addition of formic acid after 4 months, while the lignin content also increased regardless of the used additives. Weinberg and Chen (14) reported that the expected increase in neutral detergent fibre content occurred only with wheat silage from the milk stage, probably due to the acid hydrolysis of hemicelluloses, which kept neutral detergent fibre contents constant over time.

The fermentation quality during 170 days of mixed ensiling, including the changes in pH, the contents of lactic, acetic, propionic and butyric acids, as well as the ratios of lactic to acetic acid and ammoniacal nitrogen to total nitrogen are shown in Table 3. The pH range of 3.7-4.2 is generally considered beneficial for the preservation of cereal forages (29). Over the extended storage period, the pH was observed to increase first and then subsequently decrease in negative control and treatments with a low dose of formic acid and high doses of formic and acetic acids, wherein the pH after treatment with acetic acid promptly dropped to less than 4.2 on day 30, and remained low until 170 days of ensiling, suggesting that the mixed silages with the addition of acetic acid were well preserved with significantly (p<0.05) lower pH values than that of control, resulting in growth inhibition or reduced survival of yeasts and moulds. It should be noted that no significant differences (p>0.05) with respect to the pH value after treatment with formic acid compared with the negative control were observed. On average, pH values were higher in mixed silages treated with formic acid than in those treated with acetic acid (4.21 vs. 4.01; p<0.05), and with the increase of formic acid mass fraction (i.e. from 0.3 to 0.6%), the pH in ensiled forages increased accordingly. Earlier criteria for the effective preservation of ensiled crops included a high content of lactic acid and a pH below 4.5 after the fermentation phase. The five types of mixed silages all had a lower pH in the range of 3.83-4.58, and were all below 4.5 except for the pH of the silage treated with high dose of formic acid (4.58). In well-preserved silage, LAB dominated the fermentation, rapidly producing the low pH conditions that help to preserve the silage. Lactic acid is the major organic acid responsible for decreasing silage pH because it has a lower dissociation constant (pK_a=3.86) (30). The changes in pH values were the result of changes in organic acid content. Considerable lactic acid content was obtained in the five types of mixed silages, wherein the lactic acid content in silage treated with acetic acid tended to be greater than that of negative control and silage treated with formic acid after 30, 90 and 130 days of storage, respectively, suggesting that the addition of acetic acid could ensure rapid and vigorous fermentation by promoting the production of lactic acid. Statistical analysis showed that the pH and lactic acid content were significantly affected by the addition of acetic acid, but this was not the case with formic acid. The lactic acid content of the silages was influenced significantly by the dose of additives throughout the storage. Moreover, the lactic acid content of all mixed silages was higher than that (2.0%) of good-quality silages (except for silages treated with a low dose of formic acid on day 30, high dose of formic acid on days 30-90, and high dose of acetic acid on day 170), which were all well preserved at low pH (24). With prolonged storage period, a trend of increased acetic acid content in the five types of mixed silages was observed.

Acetic acid is a short-chain fatty acid with substantial antifungal activity capable of reducing yeast and mould growth, and this is enhanced with decreasing pH value. The acetic acid content in silage treated with acetic acid was significantly greater (p<0.05) than that of negative control and silage treated with formic acid during storage for 30-130 days. Rapid decrease in the pH of the silage treated with acetic acid over time was partly due to the accumulation of acetic acid, especially in the silage treated with high dose of acetic acid. The propionic acid content in all silages after 170 days of storage was less than 10%, which is in agreement with the findings of Levital et al. (31). Butyric acid is usually indicative of low-quality silage, and is produced by undesirable clostridial fermentations (32). Content of butyric acid in silage reported by Nkosi et al. (33) was lower than 1%, representing well-preserved silage. In the work presented here, the butyric acid contents of five types of silages were all lower than 0.5% during 170 days of storage, indicating good preservation. It has been reported that propionic acid is produced from lactic acid, and the presence of butyric acid in silage is a sign of fermentation by undesirable microorganisms, which should be avoided at all costs (34). Furthermore, the ratio of lactic to acetic acid and the ratio of lactic to total organic acids in all five silages decreased with the prolonged storage, and the three silages (negative control and those treated with low dose of formic or acetic acid) had higher ratio of lactic to acetic acid than those treated with high dose of formic and acetic acids. The ratio of lactic to acetic acid in negative control and treatments with low dose of formic or acetic acid was more than 3:1 when stored for 30 to 60 days (Table 3). Kung Jr and Ranjit (35), as well as Stokes and Chen (36) reported that the ratio of lactic to acetic acid of more than 3:1 was an indication of a homofermentative lactic bacteria dominant fermentation. Thus, considering this, the results found here suggest that homofermentative lactic acid bacteria dominated in negative control and silages treated with low dose of formic or acetic acid between 30 and 60 days. The ratio of lactic to acetic acid decreased after 90 days due to fermentation of pentose sugars released from hemicellulose by acid hydrolysis (36). In the present study, the lower pH, the ratio of lactic to total organic acids, and lactic acid contents in silages treated with formic or acetic acid indicate that the addition of formic or acetic acid during ensiling could alter or suppress fermentation.

Ammoniacal nitrogen, expressed as percentage of total nitrogen, gradually showed an increasing trend over time, indicating that protein degradation and deamination occurred during the later period of storage. The ratio of ammoniacal nitrogen to total nitrogen in silage treated with acetic acid was significantly lower (p<0.05) than that of negative control over 30-90 days, suggesting that proteolysis was reduced by lowering plant enzyme activity or inhibiting undesirable microorganisms with the addition of acetic acid, potentially reducing the number of yeasts that cause spoilage in the presence of air. Some authors have described the benefits of the inhibition of spoilage organisms (37), demonstrating that lowered pH value in ensiled forage can effectively inhibit proteolysis because plant enzymes are quickly inactivated under these conditions (15). The ratio of ammoniacal nitrogen to total nitrogen in silage treated with formic acid, on average, was significantly higher (p<0.05) than that of negative control when stored between 90 and 170 days due to the higher pH values and weakened inhibitory effects, but ammonium content in all five silages was lower than the recommended value of 10% of total nitrogen, indicating good silage, which demonstrated that the total nitrogen or crude protein in silages were well preserved (38). However, a greater reduction of the ratio of ammoniacal nitrogen to total nitrogen in silages treated with acetic acid was observed than in those treated with formic acid, which helps to explain the lower pH values achieved with the acetic acid treatment. In brief, organic acids, such as formic or acetic acid, represent feasible methods for promoting a rapid decline of silage pH and improving the fermentation quality of ensiled forage, with the application of acetic acid being especially efficacious.

The microbiota profile is another indicator of silage quality. Importantly, good quality silage should not contain any pathogenic and spoilage bacteria. The recovered reads of all five silages summed up to 593 875, and these reads were clustered into a total of 5669 operational taxonomic units (OTU) of genus at a 3% dissimilarity level (Table 4). Silages treated with formic or acetic acid had higher OTU numbers than negative control on day 30, the silage treated with high dose of acetic acid had a higher OTU number on day 60, silages treated with formic or acetic acid had lower OTU numbers on days 90 and 170, and the silage treated with high dose of formic acid had a lower OTU number on day 130. Another bacterial community richness estimator, Chao, was utilized to estimate the number of OTUs (39). The Chao index showed a similar trend to OTUs. These two indices showed that with respect to the bacterial community richness, the higher OTUs indicated a richer bacterial community. The diversity index of microbial population, as represented by the Shannon index (39), varied within the range of 3.27-4.58 for negative control, 1.96-5.15 for treatment with low dose of formic acid, 3.26-3.93 for treatment with high dose of formic acid, 2.01-4.23 for treatment with low dose of acetic acid and 2.43-3.69 for treatment with high dose of acetic acid. The coverage values were around 0.99, suggesting that most of bacterial species were detected. Results from the Shannon index, Chao index and reads of observed species (Table 4) indicated that most samples had a high bacterial biodiversity.

There were seven bacterial phyla, namely, Actinobacteria, Bacteroidetes, Firmicutes, Proteobacteria, Verrucomicro, Cyanobacteria and others (Fig. 2). Proteobacteria (65.3%) and Firmicutes (33.8%) were dominant in air-dried corn stover, while the relative abundance of Bacteroidetes (0.8%) and Actinobacteria (0.2%) was low. Proteobacteria (80.2%) and Bacteroidetes (18.6%) were dominant in cabbage waste, and the relative abundance of Firmicutes (0.9%), Cyanobacteria (0.2%) and Actinobacteria (0.1%) was low.

Overall, Firmicutes and Proteobacteria were still dominant in all five silages after 170 days. An increasing trend of Actinobacteria and Firmicutes in silages treated with formic or acetic acid was observed in comparison with negative control. With prolonged storage period, relative abundance of Actinobacteria first increased, followed by a decrease in negative control and silages treated with formic and high dose of acetic acid, while in the silage treated with low dose of formic acid it reached a maximum (0.9%) on day 130. Similarly, Actinobacteria in silage treated with low dose of acetic acid was observed to decrease first and then increase, with the relative abundance reaching a maximum (1.0%) on day 170. However, with prolonged storage period, Firmicutes was observed to decrease and then increase in negative control and silages treated with low dose of formic or acetic acid, while it increased first and then decreased in silages treated with high dose of formic or acetic acid. Firmicutes in silages treated with high dose of formic acid (78.1%) and silages treated with high dose of acetic acid (97.0%) reached maximum values on day 90. The most abundant bacteria on the phylum level in the mixed silages, with or without additives, were Firmicutes.

The dominant genus in air-dried corn stover included Enterobacter (47.1%), Carnobacterium (27.7%), Pantoea (10.1%), Yersinia (4.8%), Exiguobacterium (3.5%), Pseudomonas (2.3%) and so on (Fig. 3), in cabbage waste it included Pseudomonas (48.4%), Pantoea (17.1%), Flavobacterium (16.3%), Yersinia (6.5%), Enterobacter (3.6%) and Carnobacterium (0.9%). The relative abundance of non-LAB in cabbage waste was 99.1% (minimal LAB presence). Enterobacteria are non-spore forming, facultative anaerobes that can ferment lactic acid to acetic acid and other products, thus causing a loss of nutritional value in silage (40). After being discarded, cabbage waste is easily decomposed and mixed with air-dried corn stover, and is complementary to lactic acid bacteria.

Relative abundance of LAB was higher than of non-LAB in silages with added formic or acetic acid than in the negative control, except for the silage with high dose of formic acid on day 30, with low dose of acetic acid on day 60 and with low dose of formic acid on day 130. With prolonged storage, the amount of LAB species increased, and the relative abundance of LAB in silage with high dose of acetic acid increased first, from 69.2 to 96.8%, and then remained in the range of 96.1-96.8%. The relative abundance of some harmful bacteria such as Flavobacterium and Pseudomonas in all five silages decreased significantly in comparison with air-dried corn stover and cabbage waste, and the presence of harmful bacteria such as Enterobacter, Pantoea, Yersinia and Exiguobacterium was not observed. Parvin et al. (41) observed a shift in bacterial community structure from Enterobacter to Lactobacillus, Pediococcus and Lactococcus after 30 days of silage by comparing bacterial communities over time. Li and Nishino (42) also reported that Lactobacillus, Pediococcus, Weissella and Klebsiella were found in both the pre-ensiled crop and the silage by the analysis of evolution in bacterial communities in whole corn silage. In fact, great variations were found among the relative abundances of detected bacteria before and after fermentation of silages treated with formic or acetic acid in different ensiling phases. Before fermentation, Carnobacterium was the most abundant species in the silage samples. After fermentation, Lactobacillus (83.8%) and Enterobacter (49.9%) appeared to be the dominant bacteria in the silages (Fig. 3).

This shows that the addition of formic or acetic acid can help to establish LAB as dominant bacteria during storage, and effectively inhibit or reduce the growth of harmful bacteria. It was found that Firmicutes was the dominant phylum during this process, while Lactobacillus was the dominant observed genus, which is in agreement with the findings of Ennahar et al. (43).

The main LAB in air-dried corn stover were Carnobacterium and Enterococcus, with respective fractions of 96.7 and 3.3% (Fig. 4). In cabbage waste, the majority of LAB were Carnobacterium and Enterococcus, with relative fractions of 96.9 and 1.8%, respectively, while also containing a small amount of Lactobacillus (0.9%) and Leuconostoc (0.5%). Paralactobacillus and Pediococcus also appeared in all five silages after day 170, while Lachnobacterium appeared earlier in silage treated with high dose of formic acid on day 30. Furthermore, Lactobacillales appeared in all treatments after day 130. Lactobacillus was the dominant LAB genus throughout the whole storage period, and the proportion of LAB increased from 0.9% in cabbage waste on day 0 to 91.0% in silage treated with low dose of formic acid on day 130. Overall, LAB populations as a whole reached a maximum in silage treated with high dose of formic acid on day 90. Tohno et al. (44) reported that lactic acid bacteria typically associated with silage belong to the genera Lactobacillus, Lactococcus, Pediococcus, Leuconostoc, Enterococcus and Weissella. LAB are characterized by their acid tolerance and final pH values reached 3.8 at the end of the corn silage fermentation stage. To conclude, the addition of formic or acetic acid, especially acetic acid, has mostly positive effects on fermentation during mixed silage of air-dried corn stover and cabbage waste. Apart from regulating the pH and the content of microorganism metabolites (volatile fatty acids and ammoniacal nitrogen), the formic or acetic acid additives also modulate the bacterial composition and community structure in these fermented silages.