Row whole-plant corn (Qingfeng 4) was obtained from a commercial plant base located in Liupanshui City, Guizhou Province (104.83 °E, 26.60 °N) on September 30, 2020 during the dough stage. The plant material was chopped to a theoretical cut length of 1–2 cm by a precision chopper. After mixing thoroughly, the chopped whole-plant corn was divided into three parts for: (i) ensiling in BS (7 m × 24 m with three 3.2 m high walls, and a capacity of 280 tones fresh crop weight), (ii) ensiling in RB (bales were immediately wrapped with 8 layers of 0.75 m wide and 0.025 mm thick black plastic film, dimensions: 0.8 m diameter × 0.6 m length, with a maximal capacity of 190 kg fresh crop weight), and (iii) ensiling in SB (dimensions: 0.5 m wide × 0.6 m height × 1.0 m length, and with a maximal capacity of 200 kg fresh crop weight). After 90 days of fermentation, silages were opened for 9 days of aerobic exposure. The density of the BS ranged from 496 to 504 kg/m³, while that of the RB and SB groups ranged from 592 to 604 kg/m³ and 644 to 655 kg/m³, respectively. The ambient temperature ranged from 0 to 7.5°C. Samples (about 1 kg each) from the middle of silages were taken at the same six time points (0, 1, 3, 5, 7, and 9 days of aerobic exposure) during the unloading of the BS, RB and SB, respectively. For BS, samples were collected from 3 silos as three replicates at 6 different sampling times. For RB and SB, totally 36 silos were taken for ensiling, all silos were opened and samples were collected from 3 silos of each treatment at one sampling time. In total, the study comprised 3 BS, 18 RB, and 18 SB. The samples (3 treatments × 6 time points × 3 replicates) were used to determine the chemical composition, fermentation quality, aerobic stability, microbial community and mycotoxin contents. One part of silage was sampled in 50 mL cryogenic vials and stored in a −80°C freezer for microbial community analysis, and another part was stored in a −20°C freezer for chemical composition, fermentation quality, and mycotoxin content analysis.

20 g of silage samples were combined with 180 mL of distilled water and stored in a 4°C refrigerator for 1 day, then four layers of cheese cloth were used to filter the silage, for the determination of the fermentation profile. The pH was measured using a pH meter (PHSJ-3F, CANY, Shanghai, China). Concentrations of lactic acid, acetic acid, propionic acid and butyric acid were determined by high-performance liquid chromatography, according to the methods described by Wu et al. (2022). The ammonia-N content was measured using phenol-hypochlorite colorimetry (Broderick and Kang, 1980).

WPCS was dried at 65°C for 48 h in a dry oven to measure the dry matter (DM) content. Then grounded into powder for total nitrogen, crude protein (CP), water-soluble carbohydrate (WSC), neutral detergent fiber (NDF) and acid detergent fiber (ADF) analyses. The CP content was calculated via the total nitrogen content multiplied by 6.25, and the total nitrogen was analyzed by a Kjeldahl nitrogen analyzer (Kjeltec 8400 Analyzer; Foss, Sweden). According to the methods described by Turula et al. (2010), WSC was determined by colorimetric after-reaction with anthrone reagent. The NDF and ADF were measured based on Van Soest procedures, a heat stable alpha amylase was used in the NDF procedure, the results were expressed on a DM basis including residual ash. Aerobic stability was measured after 90 days of ensiling. About 3 kg of silages from each silo were taken and mixed thoroughly, then put into separate new plastic silos (capacity 10 L) without compaction and uncovered. The geometric-center of the silage masses and the ambient temperature were recorded with sensors every 2 h. The aerobic stability was determined based on the time when the temperature of silage exposed to air exceeded the ambient temperature by 2°C (Ranjit and Kung, 2000).

WPCS (20 g) was homogenized with distilled water (180 mL), and subsequently filtered through two layers of medical gauze and centrifuged for 5 min at 8,000 g/min to collect microorganism cells. The total genomic DNA was extracted via the HiPure Soil Kit (QIAGEN, Inc., Venlo, Netherlands), then the purity, concentration, and integrity of the isolated DNA samples were determined by agarose gel electrophoresis. Thereafter, the 16S rRNA V5–V7 regions of genomic DNA was amplified via Pyrobest DNA Polymerase (TaKaRa, DR500A) with the universal primers of 799F (AACMGGATTAGATACCCKG) and 1193R (ACGTCATCCCCACCTTCC). ITS genes of regions (ITS1_other) were amplified using the specific primers ITS1-F (CTTGGTCATTTAGAGGAAGTAA) and ITS2 (GCTGCGTTCTTCATCGATGC) with Barcode. AMPure XP Beads (Beckman Coulter, Indianapolis, IN, United States) was adopted for the purification of the polymerase chain reaction (PCR) products, and the ABI StepOnePlus Real-Time PCR System (Life Technologies, United States) was used for quantification (Toju et al., 2012). Following the generate sequencing libraries according to the manufacturer’s instructions, the library quality was performed on the Illumina HiSeq 2500 platform by Gene Denovo Biotechnology Co., Ltd. (Guangzhou, China). After high-throughput sequencing and the filtration of chimera and low-quality sequences, the bioinformatics analyses of the microbial community were mainly performed using the QIIME (Version 2.15.3) and R software (Version 4.0.0).

Mycotoxins were extracted simultaneously from 2 g of a dried silage sample using a 40 mL centrifuge tube with 20 mL of an acetonitrile: water solution (80:20 v/v). After the samples were horizontally shaken for 40 min, 1.0 g of NaCl and 2.0 g of anhydrous magnesium sulfate was added to obtain phase separation. After shaking for another 1 min and centrifugation at 6,000 g/min for 5 min, the upper acetonitrile phase was recovered. The mixture was evaporated to dryness under a nitrogen steam at 40°C and re-dissolved in methanol: formate in water solution (10:90, v/v), then the final extract was filtered through a filter (Millipore Corporation, Bedford, United States; HV 0.45 μm). A 20 μL sample was injected into the HPLC with MS/MS system according to the methods described by Gallo et al. (2010).

Results are reported as the mean and the standard error of the mean (SEM). The data related to fermentation quality, chemical composition, mycotoxin levels and alpha diversity were subjected to two-way analysis of variance. Aerobic stability of WPCS after ensiling were subjected to one-way analysis of variance. When there were significant differences (p < 0.05), the group means were further compared with Duncan’s multiple range tests. The statistical analyses were performed using SPSS 26.0 (SPSS, Chicago, IL, United States). Alpha diversity metrics (Chao1, Shannon and Good’s coverage) were calculated with QIIME software (Version 2.15.3). Sample ordination based on the beta diversity was examined using the principal coordinate’s analysis (PCoA). The relative abundances of microbial communities at the phylum and genus levels were also analyzed. The resultant correlation matrix was analyzed by “corrplot” in R language (method = Spearman).

The chemical composition of the WPCS was shown in Table 1. The content of DM first increased and then decreased significantly over time (p < 0.05), and the highest DM content was observed at 3–5 days. The content of DM in each group was in the order of SB > RB > BS at any stage of aerobic exposure. The WSC content decreased after 9 days of aerobic exposure (p < 0.05), and the WSC content in the RB and BS groups was lower than that in the SB group. The CP content of the three groups showed a similar decreasing trend with increasing aerobic exposure duration. Compared to the RB and BS groups, the SB group retained the highest CP content after 9 days of aerobic exposure. There were significant increases in NDF and ADF contents during the exposure period in all groups (p < 0.05). The BS group had higher NDF and ADF contents than the other groups after 5 days of aerobic exposure. It was also found that the aerobic exposure days, ensiling methods, and their interaction significantly affected the contents of DM, CP, NDF, and ADF in the WPCS (p < 0.001).

As shown in Table 2, the pH of the WPCS was significantly affected by the prolonged aerobic exposure duration (p < 0.001). The pH value of SB group ranged from 3.78 to 4.01 from d 0 to d 9, while BS and RB group had increased over 4.20 at d 3 and d 5, respectively. Although the contents of LA decreased with prolonged aerobic exposure times, the SB group had a higher concentration than the other groups. Lower PA and AA contents were also found in the SB group (p < 0.05), and PA was not detected until day 7. Furthermore, BA was not detected in the RB and SB groups. The content of ammonia-N in all groups were increased after 9 days of aerobic exposure (p < 0.05). The lowest ammonia-N content was found in the SB group on all days of aerobic exposure.

The different ensiling methods significantly (p < 0.001) influenced the aerobic stability of the WPCS (Figure 1). The BS group spoiled within 174 h, RB group was stable for 182 h, which was (p > 0.05) longer than BS group (8 h). The SB group showed the strongest aerobic stability (202 h) among all silage.

The results of the mycotoxin levels are summarized in Table 3. The AFB1, ZEN, T-2, DON and FB1 concentrations showed a significant interaction between ensiling methods and aerobic exposure days (p < 0.001). Overall, the concentrations of mycotoxins continually increased with increasing aerobic exposure (p < 0.05), and the BS group had higher concentrations than the RB and SB groups after 9 days of exposure (p < 0.05). AFB1, T-2 and FB1 concentrations were significantly lower in the SB group than in the other groups (p < 0.05), while the contamination levels of ZEN and DON did not show significant differences between the RB and SB groups after 9 days of oxygen exposure (p > 0.05).

The Good’s coverage values were greater than 99%, indicating that the sampling depth adequately captured most of the bacterial and fungal communities. The alpha diversities of the WPCS were evaluated through Shannon and Chao1 indexes (Table 4). The ensiling methods significantly affected the alpha diversities of the microbial community (p < 0.001). The RB group had lower Shannon and Chao1 indexes than the BS and SB groups in the bacterial community (p < 0.05). However, the SB groups maintained lower Shannon and Chao1 indexes in the fungal community (p < 0.05).

To analyze the distribution and structure of bacterial and fungal communities in WPCS samples at different aerobic times, principal coordinates analysis (PCoA) based on Bray–Curtis distance at the OTU level was conducted (Figure 2). Good separation and differences in microbial communities were observed between different ensiling methods, and the microbial community of SB group showed less variation during aerobic exposure stages.

Alterations in the bacterial community at the phylum level after aerobic exposure are presented in Figure 3A. Overall, Proteobacteria and Firmicutes were the dominant phyla in all of the samples, covering more than 80% of the total sequences observed. The abundance of Proteobacteria increased dramatically in the BS and RB groups within 1 day of aerobic exposure, and Proteobacteria then became the most abundant phylum with prolonged exposure days. However, Firmicutes remained the most abundant phylum in the SB group despite exposure to the air.

As shown in Figure 3B, the relative abundances of the bacterial community were further analyzed at the genus level. Lactobacillus and Acetobacter were the most dominant genera in all treatments. The abundance of the genus Lactobacillus in the BS group decreased significantly with increasing exposure time, while the relative abundance of Acetobacter increased. The relative abundance of Acetobacter in the RB group increased to 60.41%, which was significantly higher than that in the other groups after 9 days of aerobic exposure. Furthermore, the abundance of Lactobacillus decreased dramatically on the first day of aerobic exposure, and then kept a stable relative abundance in the RB group. However, Lactobacillus was still the most abundant bacteria in the SB group during aerobic exposure, and its relative abundance remained above 53.93%. Conspicuous changes were observed that the relative abundance of Weissella in the BS group was lower than that in the other groups, while the relative abundance of Bacillus in the SB group increased after 9 days of exposure, and was significantly higher than that in the other groups.

Changes in the fungal community at the phylum level during aerobic exposure are shown in Figure 4A. The phyla Ascomycota, Basidiomycota, Anthophyta, Mortierellomycota, and Mucoromycota were identified in the WPCS. Overall, the predominant ITS detected sequence belonged to the Ascomycota phylum and was followed by lower Basidiomycota, together covering more than 88% of the total sequences observed. Ascomycota was the most abundant phylum in all groups and was also the dominant phylum in the WPCS during the aerobic exposure process.

As shown in Figure 4B, at the genus level, Kazachstania and Pichia were the dominant genera in the RB group. The abundance of Kazachstania in the RB group increased dramatically after 1 day of exposure and then continuously decreased until 9 days. Conversely, the abundance of Pichia first decreased within 1 day of aerobic exposure, and then continuously increased until day 9. Pichia became the dominant fungus instead of Kazachstania after 7 days of exposure. However, neither Kazachstania nor Pichia showed a remarkable dominance in fungal communities in the BS group from days 1–7. Fusarium increased significantly, ranging from 0.64 to 16.72%, from days 5–9 in the BS group. In addition to BS and RB groups, Kazachstania was the most abundant fungus in the SB group, followed by Candida, Pichia and Fusarium during the whole aerobic exposure. Moreover, the relative abundance of Kazachstania was significantly higher in the SB group than in the other groups from days 7–9. The relative abundance of Pichia was observed to be lower in the SB group than in the other groups after aerobic exposure.

Spearman’s correlation between the bacterial community and fermentation is shown in Figure 5A. Specifically, the pH value was negatively correlated with the relative abundances of Lactobacillus and Weissella (p < 0.001) and positively correlated with the Acetobacter abundance (p < 0.01). The LA concentration was significantly positively associated with the abundances of Lactobacillus and Weissella (p < 0.001). The AA content was positively correlated with Acetobacter and Exiguobacterium abundances (p < 0.05) but negatively correlated with Bacillus and Ureibacillus abundances (p < 0.001). The contents of PA and BA were both negatively associated with Lactobacillus, Weissella and Franconibater abundances (p < 0.001) and positively correlated with Bacteroides, Thermoactinomyces and Dysgonomonas abundances (p < 0.001). Finally, the ammonia-N concentration was negatively correlated with Lactobacillus abundances (p < 0.001) but positively correlated with Acetobacter abundances (p < 0.01).

Correlation analysis was performed to illustrate the relationship between bacteria and fungi at the genus level (Figure 5B). Lactobacillus and Weissella were positively correlated with Kazachstania (p < 0.01) but negatively correlated with Fusarium (p < 0.05). Candida was positively associated with Lactobacillus (p < 0.01) but negatively associated with genus Acetobacter (p < 0.001). Pichia was positively correlated with the genera Acetobacter (p < 0.001) but negatively correlated with Lactobacillus (p < 0.01).

Aerobic deterioration is a process that causes nutrient degradation (De Melo et al., 2023). The DM content increased first after aerobic exposure. This was caused by the water and volatile organic compounds in silage volatilized after exposed to the external environment (Merkevičiūte-Venslovė et al., 2023). As expect, the contents of DM, WSC and CP decreased after 9 days of aerobic exposure in this study. DM loss was achieved through the activity of yeasts and molds consuming residual WSC and LA during aerobic exposure (Muck et al., 1991). The higher contents of DM, WSC and LA in the SB group indicated the nutrients were well stored. Limited air exposure due to strong compaction, may illustrate the phenomenon of reduced nutrient degradation in the SB group. Conversely, the contents of ADF and NDF increased during aerobic exposure, which was indirectly caused by the loss of WSC in the WPCS (Randby et al., 2020).

Fermentation characteristics illustrate the fermentation quality of silages (Hao et al., 2022). The pH value of silage is a basic indicator to evaluate microbial activity, and well-fermented WPCS should not exceed a pH of 4.2 (Kung et al., 2018). During 1 days of aerobic exposure, silage was in an unstable state due to the fierce competition between LAB and yeasts. The previous study demonstrated that various organic acids were produced to compete with yeasts and improve the aerobic stability during aerobic exposure (Hu et al., 2020). Hence, the pH value was decreased at 1 day of aerobic exposure. After exposed to the air, the environment changed to aerobic conditions, and oxygen activated the lactate-assimilating yeasts in the WPCS. These yeasts consumed the LA concentration and increased pH value during aerobic exposure (Kung et al., 2003). However, the SB group did not exceed the pH value of 4.2 after 9 days of exposure. This result could be due to the higher LA content and relative abundance of lactic acid bacteria (LAB) observed in SB group than in BS and RB groups (Figure 3B). The higher density of SB may also reduce the entry of oxygen (Tian et al., 2020). Moreover, BA was not observed in the RB and SB groups. The presence of BA may be caused by the metabolic activity of harmful microorganisms, accompanied by substantial DM losses in silages. The contents of ammonia-N in all groups increased with prolonged aerobic exposure. This result is probably due to protein hydrolysis, which is typically caused by extensive protease (Wang et al., 2022). Higher contents of ammonia-N and lower CP were observed in the BS group than in the other groups, illustrating that amino acid deamination was relatively active in BS silages. Overall, with higher contents of LA and lower pH value and PA and ammonia-N contents, the SB group was considered to have lower fermentation degradation after aerobic exposure than the RB and BS groups. In the study, the better aerobic stability of SB group (202 h) was consistent with its lower pH value and higher silage density among all groups. The greater silage density can enhance the aerobic stability of corn silage (Gallo et al., 2018), and the low pH can also improve the aerobic stability to inhibit spoilage microorganisms (Ranjit and Kung, 2000).

Aflatoxins (AFs) are produced by several species of Aspergillus section Flavi (Sweeney and Dobson, 1998). AFB1 is considered the most toxic and carcinogenic AF, since it is highly toxic and causes carcinogenic and mutagenic effects in mammals. In the present study, the contents of AFB1 increased with prolonged exposure times, which was similar to the phenomenon reported by Ferrero et al. (2019). Moreover, lower AFB1 contamination was observed in SB than in RB and BS groups. Proper silage management and well-preserved WPCS are essential to inhibit massive concentrations of AFB1. In addition, a higher abundance of LAB (Figure 3B) may also be considered to play a crucial role to limit the accumulation of mycotoxins in SB (Lahtinen et al., 2004).

Fusarium toxins are known to be frequently found in corn and animal feed globally (Vaičiulienė et al., 2022). T-2, DON, FB1 and ZEN are secondary metabolites primarily produced by several members of Fusarium (Ogunade et al., 2018). The accumulation of oxygen contents allows for spoilage and toxigenic microorganisms to grow continually with prolonged aerobic exposure (Driehuis et al., 2018). Therefore, the concentrations of mycotoxins produced by undesirable microorganisms increased significantly in all groups. Higher Fusarium toxin contamination was observed in the BS group than the other groups after 9 days of exposure. This may be caused by the higher relative abundance of Fusarium observed in the BS group (Figure 3B). Furthermore, the bunker was prepared on bare ground and thus was highly exposed to the environment (Alonso et al., 2013). A larger exposed area from the ground may cause aerobic microorganism survival and increase the content of mycotoxins in BS silage. Niderkorn et al. (2006) reported that LAB could detoxify Fusarium mycotoxins. Thus, the higher abundance of Lactobacillus in the SB group than in the other groups may result in reduced accumulation of Fusarium mycotoxins. Overall, the contents of AFB1 T-2, ZEN, FB1 and DON in all groups did not exceed the guidance level of adult ruminants after 9 days of exposure (Commission of the European Community, 2006). The silages were opened for exposure from Dec. to Jan., and ambient temperature was detected from 0 to 7.5°C. High temperatures are known to promote aggressive fungal growth (Ogunade et al., 2018). Hence, the lower temperature in the current study may have limited the synthesis of mycotoxins.

The quality of natural silage depends on the complex composition of the microflora, and the composition of the bacterial community in silage is different at various stages (Yang et al., 2019; Ran et al., 2022). The dominant bacteria at the phylum level were Firmicutes and Proteobacteria in the WPCS during aerobic exposure (Hu et al., 2018), which was supported by the present study. Proteobacteria became the dominant phylum instead of Firmicutes in the BS and RB groups after 1 day of exposure. This result indicated that the WPCS involved a notable shift in the bacterial community from Firmicutes to Proteobacteria after the environment changed from anaerobic to aerobic (Liu et al., 2019). At the genus level, Acetobacter showed a predominance of the BS and RB groups during oxygen exposure. Acetobacter are often regarded as an initiation of aerobic deterioration, which causes silage deterioration and further limits Lactobacillus proliferation (Guan et al., 2018). A high abundance of Acetobacter is usually found in bunker silages (Wang et al., 2014), which was observed in the current study. However, Lactobacillus still dominated the structure of the bacterial community in the SB group, although the abundance of Lactobacillus decreased after 9 days of oxygen exposure. Lactobacillus is well known as a LAB in maize silages and has been reported by many studies (Guan et al., 2018; Hu et al., 2018; Huang et al., 2021). Lactobacillus could inhibit fungal growth, and prevent them from becoming dominant microorganisms in the early stage of aerobic exposure (Guan et al., 2020). In addition, Lactobacillus has an excellent ability to remove mycotoxins (Hathout and Aly, 2014). A higher relative abundance of Bacillus was also detected in the SB group during aerobic exposure. Some Bacillus species can produce bacteriocin to inhibit the growth of some undesirable microbes (Lara et al., 2016). Moreover, these bacteria have also been found to be effective in the removal of mycotoxins from liquid medium (Hathout and Aly, 2014). Therefore, the lower mycotoxin level and undesirable microbes in the SB group than in the other groups may be caused by the higher relative abundances of Bacillus and Lactobacillus.

Due to the response to oxygen exposure, various fungi began to compete with the dominance of the microbial community in silages. The majority of fungi in corn silage belonged to Ascomycota, followed by Basidiomycota (May et al., 2001). At the genus level, Kazachstania, Pichia and Candida were the top 3 most abundant fungi in the WPCS (Xu et al., 2019). Kazachstania, Pichia, and Candida belong to the Saccharomycetes, regarded as yeasts frequently through culture-based methods (Wang et al., 2020). It is well known that Pichia species are often considered the major initiators of silage aerobic deterioration. The predominant fungal genus shifted to Pichia in the RB group after aerobic exposure, illustrating that the composition of the fungal community was significantly changed during oxygen exposure. Kazachstania is usually observed as the dominate genus in WPCS after aerobic exposure and is associated with the deterioration of silage (Dolci et al., 2011; Xu et al., 2019). Candida can assimilate LA and promote WPCS deterioration (Pahlow et al., 2003). To respond to oxygen exposure, yeasts became active and utilized various organic acids, resulting in a continual increase in pH value. Thus, the proliferation of massive spoilage microorganisms in silages was promoted, leading to silage deterioration and mycotoxin accumulation (McAllister et al., 1995). It is also worth noting that Fusarium appeared in silages during aerobic exposure. This result may also cause mycotoxin accumulation, as shown in Table 3.

Silage quality was influenced by the bacterial community through a series of metabolites. The Lactobacillus abundance had a positive correlation with the concentration of LA (Huang et al., 2021), and Weissella and Lactobacillus abundance showed a negative correlation with the pH value (Yang et al., 2019). The results of these studies were supported by the current research. Acetobacter can consume ethanol to produce AA after exposure to air (Nanda et al., 2001). This can be proven by the content of AA being positively correlated with Acetobacter, and higher contents of AA were found in the RB group. The positive correlation between Acetobacter and ammonia-N contents illustrated that the existence of Acetobacter probably caused the degradation of protein in this study.

It is also important to explore the relationship between bacteria and fungi to understand the decomposition of WPCS. Yeasts play a vital role in facilitating the symbiosis of various microorganisms (Alvarez-Martin et al., 2008). After aerobic exposure, there was a synergistic effect between yeasts and LAB (Roostita and Fleet, 1996). Kazachstania and Candida can metabolize organic acids, and metabolites associated with nutrients help LAB grow. In addition, LAB could also promote a suitable environment for Kazachstania and Candida to multiply after aerobic exposure (Wang et al., 2020). The accumulation of Fusarium-derived mycotoxins is caused by the metabolic activity of toxigenic strains of Fusarium genus, and AFB 1 is mainly produced by toxigenic Aspergillus species (Kalúzová et al., 2022). However, the direct relationship between the Aspergillus and AFB 1 in WPCS was not found. Similarly, the previous study also reported that no correlations were observed between fungal DNA and mycotoxin contents (Vandicke et al., 2021). A negative correlation of Fusarium with Lactobacillus and Weissella was found in this study. It is known that Fusarium cannot tolerate a low pH environment (Cheli et al., 2013). Therefore, the higher relative abundance of Lactobacillus and lower pH in the SB group than in the other groups may have inhibited the proliferation of Fusarium and further limited the accumulation of Fusarium mycotoxins.