Ensilage and Secondary Fermentation of Maize Stalk and Their Effect on Methane Production and Microbial Community Dynamics in Anaerobic Digestion

Abstract

Ensilage is an efficient storage method for preserving maize stalks for use as biogas feedstocks. However, maize stalk silages are susceptible to secondary fermentation, which degrades feedstock quality. This study explored the effects of ensilage and secondary fermentation on methane production from maize stalk and microbial community dynamics in anaerobic digestion (AD). Both ensilage and secondary fermentation decreased the specific methane yield (SMY) of maize stalks. Ensilage inhibited the acidogenesis process in AD. Secondary fermentation reduced bacterial richness and hydrolytic activity, and thus decreased the SMY of silage. After 6 months of ensilage, 97.06% organic dry matter (ODM) and 94.28% methane yield were preserved. SF greatly reduced the storage efficiency by causing 34.11% ODM loss and 52.60% methane yield loss in 40 days. Losses in ODM or methane yield during air exposure followed the Zwietering-modified Gompertz model. Metagenomic analysis showed a shift from Ruminoccoccaceae and Lachnospiraceae to Rikenellaceae in AD of maize stalk silage following secondary fermentation. Carnobacteriaceae, Moraxellaceae, Lachnospiraceae, Porphyromonadaceae, and Corynebacteriaceae were positively correlated with the content of water-soluble carbohydrates, whereas Anaerolineaceae and Ruminococcaceae were positively correlated with total organic acid content in stalks.

1. Introduction

Maize is widely planted in China and the average annual yield of maize was 2.65 million tons during 2014–2023 [1]. The annual production of maize stalk (including the stem and leaves above ground) can be estimated at 5.30 million tons based on a grain-straw ratio of 0.5 [2]. Corn stalks are returned to the cropland as a recycling solution or used as animal feed, household fuel or industrial feedstock. However, a large proportion of corn stalk is discarded or burned, causing environmental pollution. Anaerobic digestion (AD) is a treatment technology that converts biomass into biogas, which is composed primarily of methane and carbon dioxide. Biological conversion of corn stalk via AD technology into biogas recovers energy and produces digestates as fertilizers. Previous studies have shown that maize stalk has great potential to be utilized as feedstock for AD as it is abundant, cheap, easily collected, and contains a high content of hemicellulose [3,4]. However, maize stalk is harvested once a year and its proper storage is essential to ensuring a constant year-round supply of feedstock to biogas plants. Ensilage is an appropriate way of preserving energy crops for biogas production via AD. During the ensiling process, lactic acid bacteria ferment water-soluble carbohydrates (WSCs) into organic acids, mainly lactic acid, which lowers the pH and inhibits the activity of other bacteria, thus preserving energy and nutrients [4,5]. Ensiling allows seasonally harvested crops to be available throughout the entire year. The methane yield (MY) of energy crops can be almost completely preserved for 1 year by proper ensiling [6]. The advantages of ensilage over dry storage include lower fire risk, lower dry matter loss, and enhanced digestibility for biogas production [5,7].

The main principle of ensilage is the maintenance of acidic and anaerobic conditions [8]. However, air exposure of silage is inevitable in farm management [9], which usually happens during the feed-out phase, when the silo is opened and before silage feeding into the digester. In addition, it happens before silo-opening if the silage is stored under suboptimal conditions, e.g., low density of material or air diffusion through inferior or damaged silo cover. In the latter case, air exposure is usually unnoticeable and lasts for a long time. Oxygen triggers the growth of aerobic microorganisms that had been dormant under anaerobic conditions, which leads to the secondary fermentation (SF) of silage. Usually, the SF is initiated by lactic-acid tolerant yeasts, which consume lactic acid and cause the pH to rise. Then, other aerobic microbes, such as molds and bacteria, develop later among a succession of microbial changes [10]. SF results in the consumption of organic acids and WSC, as well as lignocellulose decomposition and protein degradation, leading to significant dry matter (DM) loss [10,11]. It has been reported that the DM loss can reach approximately 30% in poorly managed silage after 2 months of storage [12]. Substantial losses in methane yield can be caused by SF due to decreased degradability and mass loss of silage [8,12]. Understanding the dynamics of losses in dry matter and methane yield following air exposure is important for silage management on farms, especially for determining the feeding-out rate. However, the changes in these losses had not been well described.

While some research effort has been focused on the effect of ensilage and SF on methane production from energy crops [4,13,14], the reported results are not consistent. Ensilage has been reported to improve methane yield potential due to the conversion of WSC into more easily degradable organic acids and alcohols and the hydrolysis of lignocellulose [14,15,16], while in other cases, ensilage decreased the methane yield potential of feedstocks [17,18]. The inhibition effect of ensilage on methane production has not been explored. Understanding the SF process and the mass and methane loss during SF is crucial for silage management prior to AD. It has been reported that SF results in a decrease in the methane yield potential of silages, while an unexplained increase has been also observed in other cases [19,20]. The mechanism underlying these effects still remains poorly understood, because few detailed information about chemical or microbial dynamics during the storage process was provided. In addition, the effect of ensilage and SF on the methane fermentation profile and the microbial community dynamics has not been well explored yet. More studies should be performed to provide deeper insight into how ensilage and SF affect AD in order to improve the storage efficiency of silages.

This study aimed to describe the losses in dry matter and methane yield following SF and better reveal how ensilage and SF affect the methane production from maize stalk. Physico-chemical and microbial changes in maize stalk were assessed following 6 months of ensilage and 40 days of air exposure, and their effects on fermentation characteristics, microbial dynamics, and methane yield of stalks in AD were determined. The Zwietering-modified Gompertz model was used to describe the dynamics of dry matter or methane yield loss following air exposure.

2. Materials and Methods

2.1. Experimental Design

The experimental operation included the following three processes: (1) Ensilage of freshly harvested maize stalks for 6 months, and detection of the chemical and microbial changes during ensiling. (2) SF of the well-preserved maize stalk silage following 40 days of air exposure, and detection of chemical and microbial changes during air exposure. (3) Batch AD test of maize stalks before ensilage, on silage silo-opening, and after air exposure for different times. In addition, the fermentation profile and methane yield were determined to analyze the effects of ensilage and secondary fermentation on AD of maize stalks. The experimental details for these three processes are described below.

2.2. Harvest and Ensilage of Maize Stalks

Maize stalks were harvested with a sickle, leaving < 10 cm of stubble above the ground at the full-ripening stage from an experimental site located in Zhuozhou, Hebei Province, China. Immediately after harvesting, the cobs and husks were removed from the stalks manually, and the stalks were chopped to a particle size of approximately 1 cm and sealed in 30 L buckets (3 replicates) with a compaction density of 850 g WW/L. The buckets were stored at room temperature for 6 months. The material was weighed before and after ensiling.

2.3. Secondary Fermentation of Maize Stalks

After ensilage for 6 months, maize stalk was removed from the buckets and mixed thoroughly, and then it was subjected to SF testing. Silage containing 100 g of dry matter was placed loosely in each polystyrene box (Guoheng, Chengdu, China) and the boxes were placed within a thermostatic incubator (Yiheng, Shanghai, China), in which the ambient temperature was held at 25 ± 0.5 °C. The boxes were covered, but not sealed, to allow air to enter and prevent the evaporation of water. Thermocouples (Yokogawa, Tokyo, Japan) were placed in the middle of the silage. The temperature in each box was recorded every hour by a data logger (DX1006, Yokogawa, Tokyo, Japan). The air exposure was terminated after 40 days when the temperature remained around the environment temperature and the dry matter loss almost stopped increasing. The silage in 3 random boxes was weighed and totally sampled after 0, 2, 3, 7, 12, 22, or 40 days of air exposure and marked as S0, S2, S3, S7, S12, S22, and S40, respectively. A part of each sample was subjected to microbial enumeration and DM and organic dry matter (ODM) content analysis immediately after collection. The rest of the samples were frozen at −20 °C for chemical analysis and batch AD testing.

2.4. Anaerobic Digestion of Maize Stalks

Maize stalks before ensilage, upon silage silo-opening, and after air exposure were subjected to batch AD tests. Inoculum (average chemical characteristics: pH 8.0, DM 6.24%, ODM 4.29%) was obtained from a 200 m³ continuous stirred tank reactor at a biogas plant treating dairy manure and maize stalks (Beijing, China). The reactor was operated at 37 °C with a hydraulic retention time of 35 days. The sludge was collected and allowed to sit for 15 days until biogas production declined. The sludge was screened with a sieve (5 mm i.d.) before it was used as inoculum. Batch tests were conducted in triplicate in 1 L glass bottles. Firstly, 500 mL of inoculum was added to each bottle, after which substrate was added to the bottle at an ODM inoculum: substrate ratio of 2.5:1 and the total ODM reached 40.04 g/L in digesters. Assays using inoculum only were conducted as controls. Then, the bottles were flushed with N₂ for 5 min to remove O₂ and closed with silicone rubber caps. Finally, the reactors were incubated in a constant temperature room at 37 ± 2 °C for 40 days, during which they were shaken manually for 2 min every day to resolve the sediment and scum layers.

The pressure in the headspace of each digester was measured to allow for the calculation of the biogas volume produced in the digester using a 3150 WAL-BMP-Test system pressure gauge with 0.1% accuracy (WAL Mess-und Regelsysteme GmbH, Oldenburg, Germany), as described by Zhang et al. [21]. Methane content in biogas was determined by gas chromatography (GC-2010 plus, SHIMADZU, Kyoto, Japan) with a stainless steel column of TDX-01 (packed with a carbon molecular sieve, 2 m × 3 mm) and a thermal conductivity detector. Nitrogen was used as the carrier gas (flow rate, 30 mL/min). The temperatures of the column oven and detector were set at 40 °C and 100 °C, respectively.

Methane volume was normalized to standard temperature and pressure conditions, i.e., dry gas, 0 °C, 101.3 hPa. Specific methane yields (SMYs) were calculated as the sum of the methane volume produced over a period of 40 days with reference to the ODM added to the digesters in the batch test (ODM_added):

SMY (l_N/kg ODM_added) = (Methane volume − Methane volume_ck)/ODM_added

(1)

Methane yields were calculated with reference to the original ODM (ODM_origin), i.e., the amount of ODM in the original fresh material before ensilage, with organic mass loss considered:

Methane yield (l_N/kg ODM_origin) = (Methane volume − Methane volume_ck) /ODM_added × ODMR

(2)

Methane yield recovery (MYR)was calculated as follows:

Methane yield recovery (%) = SMY_Silage/SMY_FM × ODMR

(3)

where SMY_FM is the specific methane yield of fresh maize stalk before ensilage (l_N/kg ODM_added), SMY_Silage is the specific methane yield of maize stalk silages during storage (l_N/kg ODM_added), and ODMR is the organic dry matter recovery (%).

2.5. Kinetic Model Fitting of Organic Dry Matter (ODM) Loss and Methane Yield Loss During Secondary Fermentation

ODM and methane yield were lost due to the silage degradation during the SF phase. Data of ODM loss or methane yield loss fit the Zwietering-modified Gompertz model [22], which was used to predict the ODM loss or methane yield loss after silage was exposed to air. The equations were as follows:

y₁ = a₁ × exp{−exp[b₁ × e × (c₁ − t₁)/a₁ + 1]}

(4)

where y₁ is the total ODM loss (g/kg), a₁ is the potential ODM loss (g/kg), b₁ is the maximum ODM loss rate (g/kg·d), c₁ is the lag phase (d), t₁ is the duration of air exposure (d), and e is 2.7183;

y₂ = a₂ × exp{−exp[b₂ × e × (c₂ − t₂)/a₂ + 1]}

(5)

where y₂ is the total methane yield loss (l_N/kg OMD_origin), a₂ is the potential methane yield loss (l_N/kg OMD_origin), b₂ is the maximum methane yield loss rate (l_N/kg OMD_origin·d), c₂ is the lag phase (d), t₂ is the duration of air exposure (d), and e is 2.7183.

2.6. Chemical Analysis

Maize stalk samples were oven-dried at 100 °C for 16 h to a constant weight immediately after sampling and burned at 550 °C in a muffle furnace (Lichen, Shanghai, China) to determine DM, ODM, and ash content. DM and ODM were corrected for the loss of volatile components according to the method described by Porter and Murray [23].

The dried samples were milled and sieved with a 1 mm screen, after which they were used for determination of water-soluble carbohydrate (WSC), total nitrogen, neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) content according to AOAC [24] and Van Soest et al. [25]. Crude protein (CP) was calculated as follows: total nitrogen × 6.25.

Five grams of frozen maize stalk samples were suspended in 45 mL of distilled water and incubated at 4 °C for 10 h. Then, the extractions were filtered through two layers of cheesecloth and used for determination of pH, organic acid content, alcohol content, soluble chemical oxygen demand (sCOD), nitrate-nitrogen (Nitrate-N), and ammonia nitrogen (NH₃-N) content. The pH was determined by a portable pH meter (Model SX-610; San Xin Instrumentation, Shanghai, China). Organic acids and alcohols were measured by high performance liquid chromatography using a Shimadzu (Kyoto, Japan) HPLC system equipped with an Aminex HPX-87H column (300 mm × 7.8 mm; Bio-Rad Laboratories, Hercules, CA, USA) and an SPD-M20A detector (Shimadzu, Kyoto, Japan). The mobile phase was 5 mM H₂SO₄ (pH = 2.2, flow rate = 0.6 mL/min). The column oven was set at 40 °C. The extractives were filtered through an aperture of 0.22 μm. NH₃-N and Nitrate-N contents were analyzed with a flow analyzer according to APHA [26]. The soluble chemical oxygen demand (sCOD) analysis was conducted as previously reported [8]. Potassium dichromate was used as an oxidant. The extractives were centrifuged at 8000 rpm for 10 min, after which the supernatant was diluted and subjected to measurement using a COD analyzer (Model ET99731, Lovibond, Dortmund, Germany).

The anaerobic sludge collected from the AD digesters was centrifuged for 5 min at 8000 rpm and the supernatant was used for pH and organic acid content determination, and the analysis methods were the same as those of maize stalks.

2.7. Microbial Analysis of Maize Stalks

2.7.1. Hydrolytic Enzyme Activity Determination

Carboxymethyl cellulase (CMCase) activity and xylanase activity of the stalk were determined using the extractions (as stated in Section 2.5) according to previously reported methods [8]. Carboxymethyl cellulose (CMC) was used as a substrate for CMCase. Birchwood xylan was used as a substrate for xylanase. Glucose and xylose were used as the standards for the CMCase and xylanase measurements.

2.7.2. Microbial Enumeration

Microbial enumeration was performed with silages exposed to air for 0, 2, 3, 7, 12, 22, or 40 days as previously described [8,27]. Five grams of silage samples were extracted with 45 mL of distilled water and the serial 10-fold dilutions of the extractives were spread onto different culture plates. The plates were incubated for 3–5 days at 30 °C and then colonies were counted from plates of appropriate dilutions containing a minimum of 30 colonies. Aerobic bacteria were enumerated on a beef extract peptone agar and yeasts on PDA culture. Enumeration of cellulolytic microbes (CMs) was performed on CMC-Congo red agar (0.05 g MgSO₄, 0.2 g (NH₄)₂SO₄, 0.1 g KH₂PO₄, 0.05 g NaCl, 0.02 g Congo Red, 2.0 g CMC sodium salt, 2.0 g agar, 100 mL distilled water); each colony with a surrounding halo zone was counted. Microbial data were transformed (log10) before statistical analysis.

2.8. Microbial Analysis of Anaerobic Digestion (AD) Samples

2.8.1. Quantitative Real-Time Polymerase Chain Reaction of Bacteria in Anaerobic Digestion (AD)

DNA extraction was carried out on inoculum (day 0) and samples from the batch tests (day 2, 7, 18, and 30) with silage were exposed for 0, 7, 12, or 40 days. Total DNA was extracted from 1 mL of fermentation broth sample using E.Z.N.A. soil DNA Kit (Omega Bio-tech, Norcross, GA, USA). For quantitative real-time polymerase chain reaction (PCR) of total bacteria, the 63F (5’-GCAGGCCTAACACATGCAAGTC-3’) and 335R (5’-CTGCTGCCTCCCGTAGGAGT-3’) primers were used according to Castillo et al. [28]. The PCR reaction was conducted using the SYBR Green PCR SuperMix-UDG Kit (Invitrogen, Carlsbad, CA, USA). The reaction mixture (20 µL) included 10 µL Ultra SYBR Mixture, 0.4 µL Rox Dye (10×), 0.4 µL Primer-F (final concentration, 10 µM), 0.4 µL Primer-R (final concentration, 10 µM), 1.0 µL template, and 7.8 Ul PCR-grade water. The amplification protocol was as follows: 2 min at 50 °C and 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and combined annealing and extension for 1 min at 60 °C. The standard curve construction of bacteria was performed according to the methods reported by Hua et al. [29]. Quantitative PCR reactions were performed on an ABI 7500 system (Model 7500, Applied Biosystems, Waltham, MA, USA). All DNA samples were analyzed with each primer set in duplicate. The reported units were in copies/mL for fermentation broth.

2.8.2. High-Throughput Sequencing Analysis

The bacterial communities (after 2 days of fermentation) from digesters fed with the silage exposed for 0, 7, 12, and 40 days were analyzed by sequencing the V3–V4 hypervariable region of the 16S rRNA gene. The V3–V4 region was amplified using the following universal primer sets: 336F and 806R (5′-GTACTCCTACGGGAGGCAGCA-3′ and 5′-GTGGACTACHVGGGTWTCTAAT-3′) according to Zheng et al. [30]. The amplicon libraries were sequenced using the Illumina Miseq and a 2 × 300 bp paired-end protocol. Raw data were processed and analyzed using the Trimmomatic (version 0.40) and Flash (version 1.2.7) software platform. After demultiplexing, the reads were assigned species-equivalent operational taxonomic units (OTUs) at 97% sequence similarity. The remaining OTUs were denominated at the phylum and genus levels according to the Silva database, http://www.arb-silva.de/documentation/release-115 (accessed on 20 June 2021).

2.8.3. Hydrolytic Enzyme Activity Determination

Fermentation broth liquid during AD was centrifuged for 5 min at 8000 rpm and the supernatant was used for CMCase activity and xylanase activity, and the analysis methods were the same as those of maize stalks as stated in Section 2.7.1.

2.9. Data Analysis

Data processing and statistical analyses were performed using Microsoft Excel 2010 (Redmond, WA, USA) and Origin 8.6 (Origin Lab Corp., Northampton, MA, USA). Analysis of variance (univariate ANOVA) was used to detect differences between treatments. A p value < 0.05 was considered significant.

3. Results and Discussion

3.1. Ensilage and Secondary Fermentation of Maize Stalk

3.1.1. Characteristic Changes in Maize Stalk During Ensiling

The chemical characteristics of fresh maize stalk (FM) are shown in

Table 1. Changes in chemical characteristics of maize stalk during ensiling and air exposure.

Items	FM	S0	S2	S3	S7	S12	S22	S40
DM (g/kg)	301.0 ± 1.1	305.1 ± 10.3	299.4 ± 12.3	286.4 ± 0.8	266.3 ± 13.2	249.7 ± 12.1	251.2 ± 2.8	457.0 ± 4.8
ODM (g/kg)	276.6 ± 1.5	279.3 ± 11.4	274.0 ± 11.8	260.2 ± 1.0	240.2 ± 12.1	222.8 ± 11.0	220.9 ± 3.2	400.2 ± 5.7
Ash (g/kg)	24.7 ± 0.8	25.7 ± 0.1	25.4 ± 0.1	26.2 ± 0.1	26.1 ± 0.4	26.9 ± 0.2	30.3 ± 0.2	56.8 ± 0.3
NDF (%DM)	65.45 ± 0.65	64.59 ± 0.91	65.52 ± 2.10	70.22 ± 0.92	76.04 ± 1.04	73.27 ± 1.29	74.61 ± 1.64	62.33 ± 2.45
ADL (%DM)	3.69 ± 0.78	6.19 ± 0.56	5.78 ± 0.80	5.00 ± 0.80	7.73 ± 0.97	7.51 ± 0.58	5.65 ± 1.65	5.12 ± 1.37
CP (%DM)	5.35 ± 0.02	5.45 ± 0.02	5.62 ± 0.06	5.82 ± 0.21	5.87 ± 0.20	5.28 ± 1.31	5.27 ± 1.14	4.39 ± 0.05
WSC (%DM)	16.00 ± 0.21	1.30 ± 0.10	1.19 ± 0.11	1.50 ± 0.15	1.11 ± 0.10	0.78 ± 0.13	1.12 ± 0.05	1.08 ± 0.03
sCOD (%DM)	36.04 ± 1.43	27.53 ± 0.39	26.13 ± 2.66	18.26 ± 8.19	11.08 ± 0.31	8.10 ± 0.42	12.39 ± 0.34	15.16 ± 0.26
NH3-N (‰DM)	0.30 ± 0.03	0.94 ± 0.01	0.38 ± 0.00	0.24 ± 0.00	0.35 ± 0.00	0.39 ± 0.01	0.11 ± 0.00	0.05 ± 0.01
Nitrate-N (‰DM)	0.41 ± 0.02	0.11 ± 0.00	0.11 ± 0.00	0.20 ± 0.00	0.21 ± 0.00	0.24 ± 0.00	0.22 ± 0.00	0.20 ± 0.00
ODMR (%)	NA	97.06	94.18	87.84	81.15	72.98	65.24	62.95

NA: not applicable. Values are means ± standard deviations (n = 3).

. The DM content (301.0 g/kg WW) and WSC content (16.00% DM) of the maize stalk were within the proper ranges for ensiling (Table 1).

The silage was well preserved via ensiling, as indicated by its low pH of 3.9 (Figure 1) and high lactic acid content of 11.91% DM (

Table 2. Changes in organic acids of maize stalk during ensiling and air exposure.

Items	FM	S0	S2	S3	S7	S12	S22	S40
Formic acid (%DM)	0.05 ± 0.01	---	---	---	---	0.11 ± 0.00	0.14 ± 0.01	0.13 ± 0.00
Lactic acid (%DM)	---	11.91 ± 0.11	10.84 ± 0.02	3.94 ± 0.01	0.80 ± 0.12	0.59 ± 0.02	0.93 ± 0.01	0.00 ± 0.00
Acetic acid (%DM)	1.68 ± 0.01	3.24 ± 0.02	1.94 ± 0.00	0.50 ± 0.00	0.57 ± 0.00	1.02 ± 0.00	0.85 ± 0.01	1.04 ± 0.01
Propionic acid (%DM)	0.53 ± 0.01	0.15 ± 0.00	0.07 ± 0.00	0.32 ± 0.01	0.33 ± 0.00	0.40 ± 0.00	---	0.07 ± 0.01
Ethanol (%DM)	0.02	---	---	---	---	---	---	---

---: under detection. Values are means ± standard deviations (n = 3).

). No butyric acid was found, indicating the absence of undesired SF during ensiling. However, the high content of lactic acid and low content of antifungal organic acids (acetic and propionic acids) in the maize stalk silage indicated that the silage is susceptible to SF once exposed to air. The NH₃-N content increased from 0.30% to 0.94% DM (Table 1) due to protein degradation by plant enzymes and microbes during ensiling [31]. The sCOD content decreased from 36.04% to 27.53% DM (Table 1), as a result of water-soluble component consumption by plant respiration and aerobic microbe activity during the initial stage of ensiling. A low ODM loss of 2.94% was observed after 6 months of ensiling (Table 1). In previous studies, the reported the ODM loss ranged from 4.6 to 14.2% for maize [11].

3.1.2. Characteristic Changes, Microbial Dynamics, and Organic Dry Matter (ODM) Loss of Silage During Air Exposure

The pH, temperature, and microbial dynamics of silage following 40 days of air exposure are shown in Figure 1. SF occurred rapidly after silage exposure to air and was characterized by pH rise, temperature rise, and proliferation of yeasts, in agreement with previous reports that upon exposure to air, SF is often initiated by lactate-oxidizing yeasts [10]. Yeasts consume lactic acid and WSC, causing temperature and pH rises [9]. After 36 h of air exposure, the temperature of silage exceeded the ambient temperature by 2 °C (Figure 1a), indicating the start of SF. The temperature reached a peak of 31.3 °C at 50 h, and then gradually decreased to similar degrees with the ambient temperature (Figure 1a). Increasing the pH of silage facilitates the proliferation of molds, clostridia, and other aerobic microorganisms [9,10]. After the pH of silage rose to 5.9 (day 3), the number of bacteria and CMs increased (Figure 1b). The growth of CMs was accompanied by increased CMCase and xylanase activity, which indicated that the degradation of cellulose and hemicellulose happened (Figure 1b). Degradation of complex carbohydrates increases the bioavailability of WSC to other opportunistic aerobic microbes, which thus proliferate and further reduce silage mass [32]. The second and third peaks of temperature might be related to the increased carbon sources for aerobic microbes from polysaccharides hydrolysis.

Lactic acid, acetic acid, and ethanol were degraded rapidly during air exposure (Table 2). Consumption of organic acids and ethanol was related with the decrease in soluble chemical oxygen demand (sCOD). The relative content of CP, NDF, and ADL increased from day 0 to day 7 as a result of these changes (Table 1). However, NDF and ADL content decreased from day 7 to day 40 (Table 1), which might be related to the increased hydrolytic enzyme activity during air exposure. CP also decreased from 5.87% to 4.39% DM from day 7 to day 40 due to microbial degradation (Table 1). NH₃-N and Nitrate-N are both indicators of proteolytic activity. At the initial stage of exposure, the rise in pH and temperature promoted the conversion of NH₃-N into free ammonia and its volatilization, decreasing the NH₃-N content to 0.24% DM (day 3), and then the NH₃-N content increased to 0.39% DM (day 12), indicating progressive protein decay. The increase in Nitrate-N was observed during SF. NH₃-N can be converted into nitrate (NO₃⁻), nitric oxide (NO), N₂O, and dinitrogen (N₂), thus progressively reducing biomass nitrogen content [33].

Following SF, the ODM content of maize stalk silage decreased from 279.3 to 220.9 g/kg WW from day 0 to day 22 as a result of continuous consumption of organic matter, and increased to 400.2 g/kg WW by day 40 as a result of water evaporation during prolonged exposure to air (Table 1). SF resulted in the dramatic ODM loss of 15.91% within 7 days; after 40 days of air exposure, the ODM loss reached 34.11% (Table 1). The accumulated temperature rise of silage was linearly and positively related to the ODM loss (R²=0.93), in agreement with previous studies [8,11]. The Zwietering-modified Gompertz model [22] well fitted the data of ODM loss (R²= 0.98) during the SF phase (Figure 2). The predicated ODM loss (Model ODML) after 40 days of exposure was 335.96 g/kg by the model, similar to the detected value of 341.1 g/kg (Figure 2). The maximum ODM loss rate (a1) was estimated at 23.37 g/kg·d. The lag phase (c1) is 7.7 h, which indicates the time interval between air exposure and occurrence of ODM, and can be used as a parameter to reflect the aerobic stability of silage in further study. Results showed that silage ODM loss following air exposure can be predicted by the Zwietering-modified Gompertz model. The models can be taken into consideration to determine the feeding-out rate of silage.

3.2. Effect of Ensiling and Air Exposure on Methane Production

3.2.1. Fermentation Dynamics in Anaerobic Digestion (AD) of Maize Stalks

The daily methane yields of maize stalks before ensiling, on silo-opening, and during SF are shown in Figure 3. Two peaks were observed. With FM, S0, S2, S3, S7, S12, S22, and S40, the first daily methane yield peaks were 42.91 (day 3), 40.14 (day 2), 44.84 (day 2), 45.80 (day 2), 24.60 (day 3), 26.80 (day 3), 24.40 (day 5), and 18.70 (day 4) l_N/kg ODM, respectively (Figure 3). Conversion of the carbohydrate to more easily degradable components during ensiling brought forward the methane yield peak. SF tended to delay and decrease the first methane peak due to the consumption of water-soluble components. The second peaks appeared around day 14–22 (Figure 3).

CMCase and xylanase are the main enzymes that convert lignocellulose to reducing sugars. Their activity levels reflected the growth and activity of hydrolytic microorganisms. The hydrolytic enzyme activity during AD was shown in Figure 3. S0 showed a higher CMCase but lower xylanase than FM, which remains to be explained (Figure 3). Following the SF process, the enzyme activity during AD increased at a slower rate and peaked at a lower level (Figure 3), indicating that the lignocellulose hydrolysis was negatively affected by silage deterioration. The silages S12, S22, and S40 showed higher lignocellulosic enzyme activity on day 0 of AD than other treatments, due to the epiphytic enzymes produced by aerobic microbes during SF (Figure 3). However, their enzyme activity declined quickly within 2 days (Figure 3), perhaps as a result of environmental shock. AD is a multistep process that involves the interaction of many groups of microorganisms responsible for, respectively, hydrolysis, acidogenesis, acetogenesis, and methanogenesis [34]. High hydrolytic activity ensures abundant monomers as substrates for downstream acidogenesis, while low hydrolytic activity negatively affects the acidogenesis efficiency, and thus limits the acetogenesis and methanogenesis. It has been reported that the degradation rate (%) of feedstock and biogas yield was positively related to the CMCase and xylanase [35]. With poorly degradable feedstock, the hydrolysis stage is likely to be the limiting step [36]. In this study, the hydrolytic activity levels were positively related with organic acid production and methane production during AD.

The changes in pH and organic acid content of fermentation broth in AD were shown in Figure 4. The pH of the fermentation liquid varied between 7.2 and 8.0 (Figure 4), which was within a suitable range for methanogens in AD [37]. In digesters treating FM, most of the acetic acid was consumed after 4 days, and propionic acid and butyric acid increased during 0–5 and 0–10 days, respectively, which reflects the continuous conversion of rich WSC into organic acids. For silages before serious corruption (S0, S2, S3, S7, and S12), formic acid, lactic acid or acetic acid were consumed quickly, which contributed to the high methane yield observed during the first few days of AD. In digesters treating S0, S2, S3, and S7 silages, propionic acid or butyric acid content increased during the first 3 days and then decreased quickly (Figure 4). Propionic and butyric acids can be converted into acetic acid, providing a rich substrate for methanogens, and thus stimulated methane production. The transitory accumulation in propionic acid and butyric acid content was related to the active polymer hydrolysis and monomer fermentation activity during the first few days of AD. The most active accumulation of propionic acid and butyric acid was observed in the S2 digesters. However, although S0 showed higher hydrolytic enzyme activity than S2, it showed lower increase in propionic acid than S2 and no increase in butyric acid concentration (Figure 4). This indicated that the acidogenesis which converts monomers into volatile fatty acids (VFAs) might be inhibited in S0 digesters. For S22 and S40, total organic acid concentrations remained very low throughout the digestion process (Figure 4), consistent with their low hydrolytic activity and low methane yield.

3.2.2. Methane Yield Potential of Maize Stalks During Ensiling and Air Exposure

The changes in methane yield potential of maize stalks following ensilage and SF are shown in

Table 3. Methane yield potential and methane yield recovery of maize stalks during ensiling and air exposure of maize stalk.

Maize Stalk	Specific Methane Yield (lN/kg ODMadded)	Methane Yield (lN/kg DMorigin)	Methane Yield Recovery (%)
FM	370.18 a ± 2.09	NA	NA
S0	359.59 b ± 3.11	349.01 ± 3.02	94.28
S2	362.99 ab ± 3.78	341.86 ± 3.56	92.35
S3	355.23 b ± 2.82	312.05 ± 2.48	84.30
S7	307.87 d ± 3.34	249.85 ± 2.71	67.49
S12	325.67 c ± 6.24	237.68 ± 4.55	64.21
S22	274.36 e ± 5.01	179.00 ± 3.27	48.35
S40	245.07 f ± 3.74	154.27 ± 2.35	41.68

^a–f Different letters following the numbers indicate a significant difference (p < 0.05). NA: not applicable. Values are means ± standard deviations (n = 3).

. Ensiling storage was a complex process and many factors will affect the methane production of energy crop silage. Ensiling has been reported to increase methane yield potential due to the formation of easily degradable components such as organic acids and ethanol and partial hydrolysis of lignocellulose [5]. However, these positive effects were not observed in this study. Silage (S0) showed a lower methane yield potential than fresh maize stalk before ensiling (Table 3). This might be due to the loss of water-soluble components during ensiling as indicated by the lower sCOD of S0 than FM. In addition, the high lactic acid in S0 silage might have an inhibitory effect on methane production. In the ensiling process, the fermentation pattern of lactic acid bacteria can be divided into homo- and hetero-fermentative types. The homo-fermentative lactic acid bacteria utilize one molecule of hexose to generate two molecules of lactic acid. The hetero-fermentative lactic acid bacteria utilize hexoses to generate lactic acid, acetic acid, ethanol, and CO₂ or convert pentoses into equivalent lactic acid and acetic acid. It has been reported that when the homo-fermentative activity of lactic acid bacteria was enhanced in the silage with additives, the methane yield per ODM decreased by 13% compared with control [17]. The initial lactic acid concentration reached 3.13 g/L and total organic acids concentration reached 4.55 g/L in the S0 digester. Zhao et al. [7] reported that when the initial lactic acid concentration reached a similar value of 3.57 g/L, an inhibitory effect on methane production was observed in the batch digestion of maize stalk silage. It has been reported that when the total VFA concentration reached 4 g/L, glucose fermentation was inhibited [36]. The inhibitory effect of ensilage on methane production happened more likely in the acidogenesis stage, as stated in Section 3.2.1.

S2 showed slightly higher methane yield than the S0 silage perhaps because the lactic acid inhibitory was alleviated, but the difference was not significant (p > 0.05). However, the methane yield potential tended to decrease following the process of SF; after exposure to air for 3, 7, 14, 22, or 40 days, the SMY of silage samples was decreased by 1.21, 14.38, 10.69, and 23.70%, respectively (Table 3), in comparison with that of the silage before air exposure (S0). The trend of decreased methane yield potential after exposure to air has been reported in previous studies. The consumption of easily degradable components, such as WSC, alcohols and organic acids, was considered as an important reason for the decreased methane yield potential [11,21]. The decrease in acetic acid, which is a direct substrate for methanogens and can be quickly converted into methane, negatively affects the methanogenesis process; lactic acid, propionic, acid and butyric acid are intermediate products of anaerobic fermentation and readily converted to acetic acid. The loss in these acids negatively affects the acetogenesis; more importantly, the loss in WSC also negatively affected the growth and activity of polymer degraders as indicated by the bacterial richness (as stated in Section 3.2.3) and hydrolytic activity, and therefore decreased the biodegradation rate of maize stalk, and reduced the substrate for downstream microbes. There was an increase in methane yield potential after air exposure for 12 days. The SMY of S12 was 4.3% higher than that of S7 (Table 3), partially compensating the ODM loss. This change might have been caused by the increased abundance of cellulose- and hemicellulose-degrading enzymes in silage after 12 days of exposure, which broke the lignocellulosic structure and enhanced the biodegradability of organic matter. However, the increased digestibility of feedstock cannot compensate for the negative effects of nutrient loss; therefore, the SMY of S12 was lower than that of S0 (Table 3). With advanced deterioration, the positive effect of air exposure on SMY was abolished, and SMY of S22 and S40 continued to decrease (Table 3). This might have been caused by the aggravated consumption of relatively easily degradable components, increase in recalcitrant components or production of some microbe-inhibiting components.

3.2.3. Quantitative Polymerase Chain Reaction of Bacteria in Anaerobic Digestion (AD) of Maize Stalks

Bacteria in maize stalk fermentation broth were quantified using real-time PCR; samples from digesters treating maize stalks before ensiling (FM), silage before air exposure (S0), and silage at the early (S7), middle (S12), and late (S40) stages of SF were analyzed.

Rich water-soluble components in FM favored the rigorous bacterial proliferation, while ensilage negatively affected bacterial growth in AD of stalks. FM showed the highest bacteria amount on day 2 (Figure 5). On day 2 and day 7 in AD, the abundance of bacteria in FM was higher than that in S0 (Figure 5). The amount of bacteria in the S0 digesters increased more slowly than that in the FM digesters (Figure 5). This might be related to the loss of soluble carbon sources due to plant respiration and aerobic microbial activity during ensiling, as indicated by the lower sCOD in S0 than in FM. The changes in bacterial abundance corresponded to the methane potential of stalks. Similar results have been reported by Zheng et al. [30], indicating that the changes in the bacterial population corresponded to changes in the methane yield of switchgrass and that anaerobic digesters did not contain large bacterial populations when the methane production process suffered from varying degrees of inhibition.

SF of silage negatively affected bacterial proliferation in AD. Changes in the bacterial population corresponded to changes in hydrolytic enzyme activity, organic acid production, and methane production of the detected stalk samples. The abundance of bacteria in the S0 digesters was greater than that in the digesters treating silages subjected to air exposure (S7, S12, S40) throughout the AD process (except on day 2, when it was lower than that of the S40 digesters) (Figure 5), in accordance with higher hydrolytic activities and higher methane yield of S0 than other silages. The abundance of bacteria in the S7 digesters was much lower than that of the S0 digesters as a result of insufficient water-soluble components, in accordance with the lower level of hydrolytic enzyme activity. The bacterial richness in the S12 digesters was higher than S7 on day 2 and day 7 of AD, whereas it was lower than that in the S7 digesters on day 18 and day 30 (Figure 5). The increased biodegradability of the S12 silage promoted bacterial proliferation in the early stage of AD. Similarly, it has been reported that the high abundance of bacteria was related with higher degradability of lignocellulose in anaerobic digesters fed with corn stalks and cow dung [38]. The S40 digesters showed a higher bacterial richness than the S7 digesters on day 2 of AD (Figure 5). This might be related to the increased abundance of water-soluble components in stalks. However, after a slight increase on day 7, the bacterial abundance in the S40 digesters decreased quickly (Figure 5). This is possibly due to the fact that the degradable component content in the S40 silage was lower and consumed relatively quickly, thus hampering the growth of bacteria in the later stage of AD.

Proliferation of polymer degraders was related to the nutrients, especially the water-soluble components provided by the silage stalk. Maize stalks with rich water-soluble components could facilitate the growth of polymer degraders, enhance the hydrolytic activity, and provide rich substrate for downstream microbes. Following ensilage and SF, the loss of water-soluble components was an important reason for the decrease in methane yield potential since it negatively affected the microbial hydrolysis of feedstocks. Particularly, the negative effect of water-soluble component loss exceeded the positive effect of sugar-to-acid conversion during ensilage.

3.2.4. Metagenomic Analysis of Bacterial Community in Anaerobic Digestion (AD) of Maize Stalks Following Ensilage and Air Exposure

High-throughput analysis was used to measure the diversity of the bacteria in the four biogas reactors, which were fed with samples of fresh maize stalk (FM), silages before air exposure (S0), silage at the early (S7), middle (S12), and late (S40) stage of aerobic deterioration, respectively. A total of 186, 786 raw tags and 166, 055 clean tags after quality trimming were obtained for bacterial sequences. A total of 801 OTUs were assessed from all detected digesters. The Chao community richness index was from 407 to 658, and the Shannon community diversity index was from 4.45 to 6.50. FM showed the lowest Chao and Shannon indexes. In silages, the Chao indexes from S0 and S7 were relatively high, while those from S12 and S40 were relatively low. The Shannon index of silage decreased following deterioration.

The bacterial kingdom mainly comprises the phyla Firmicutes, Chloroflexi, Bacteroidetes, and Proteobacteria, and is followed by Synergistetes, Actinobacteria, and Atribacteria (Figure 6). Clostridia, Anaerolinea, and Bacteroidia were the most prevalent classes. These bacteria are common dominant groups in biogas plant microbiota, which play important roles in hydrolysis and acidogenesis [39,40,41].

The bacterial analysis revealed a predominance of microorganisms belonging to the Lactobacillales order (45.27% of bacteria), which were mostly classified into Trichococcus genus (45.13% of bacteria) in maize stalk prior to ensiling (FM). After ensiling, the proportion of Trichococcus decreased significantly, accounting for only 0.62% of the proportion of bacteria in S0. Trichococcus is a lactic acid bacterium which widely exists in activated sludge, marsh, soil, and methane fermentation systems and produces acid using a variety of sugars and polysaccharides [42,43]. The stalk before ensiling contains high soluble sugar, which is favorable for the fermentation of the bacteria to produce acid. In the process of ensiling, soluble sugar is converted into lactic acid, acetic acid, etc. The S0 sample has lower soluble sugar content and higher content of organic acid such as lactic acid and acetic acid, which might be the reason for the decrease in Trichococcus abundance.

Comparative analysis showed that Clostridia had the sharpest drop in the digesters treating deteriorated silage, followed by Gammaproteobacteria, Bacteroidetes_vadinHA17, and Deltaproteobacteria (Figure 7). The decrease in abundance of these bacterial groups was mainly responsible for the decrease in hydrolytic enzyme activity (as stated in Section 3.2.5) and methane production from silages S7, S12, and S40. Clostridia play an important role in cell wall degradation. These groups are usually abundant in AD systems [40,44]. Bacteroidetes_vadinHA17 constitute a candidate division and their functions will not be well known until members of these groups are isolated and characterized or until full genomic information is available [45]. Deltaproteobacteria have been found in the intestinal tracts of termites and cockroaches [46] and biogas reactors [47]. Gene analysis indicates that species belonging to Deltaproteobacteria are capable of reductive acetogenesis (homoacetogenesis) from CO₂ and H₂, and biosynthesis of amino acids through nitrogen fixation and ammonia assimilations [46]. Their higher abundance in the S0 digesters might be related to the high ammonia nitrogen in the corn stalk.

Bacteroidia and Synergistia (except for S40) showed higher abundance in digesters treating deteriorated silages (Figure 7). Bacteroidia hydrolyze polysaccharides and proteins, ferment sugars, and produce VFAs [39]. Synergistaceae (Synergistia class) was the sole family found of the Synergistetes phylum. However, many OTUs were assigned to unclassified genera. Some species from Synergistaceae were reported to ferment peptides and a range of amino acids to acetate, propionate, and hydrogen [44]. The Synergistaceae were present in higher proportions in deteriorated silage, which was conducive to the growth of hydrogenotrophic methanogens.

Redundancy analysis can clarify the relationship between substrate composition and bacteria, enabling a clear description of the microbial community in response to the utilization of the organic matter in maize stalks. As shown in Figure 8, the five maize stalk samples were clustered at different positions in the two-dimensional map, indicating that there were differences in bacterial community among them. WSC was positively correlated with FM, organic acids were positively correlated with S0, and ADL were positively correlated with S7, S12, and S40. The effects of environmental factors at the family level were as follows: WSC > total organic acids > ADL > cellulose+hemi-cellulose > protein. The abundance of Carnobacteriaceae, Moraxellaceae, Lachnospiraceae, Porphyromonadaceae, and Corynebacteriaceae were positively correlated with the WSC content. Anaerolineaceae and Ruminococcaceae were positively correlated with total organic acid content. Synergistaceae, Peptostreptococcaceae, and Rikenellaceae were closely positively correlated with ADL.

3.2.5. Methane Yield Loss of Maize Stalks During Ensiling and Air Exposure

Taking ODM loss into consideration, methane yields (l_N/kg ODM_origin) were well preserved by 94.28% after anaerobic storage, whereas it was greatly lost during SF (Table 3). After 7 days of air exposure, the methane yield of the maize stalk was decreased by 26.79%. Herrmann et al. [11] have reported that methane yields of maize silages decreased by 5–19% after 7 days of air exposure. After 40 days of air exposure, only 41.68% of methane yield of maize stalk was preserved. A negative linear correlation was observed between methane yield (MY, l_N/kg ODM_origin) and accumulated temperature rise above ambient temperature (AT, °C) of silages following SF: MY = −2.2529AT + 359, R² = 0.94. The loss in the methane yield of maize silage was 2.25 l_N/kg ODM_origin per 1 °C increase in silage temperature above ambient per day. The negative correlation has also been reported in a previous study [11], indicating that the cumulative temperature rise per 1 °C corresponded to a methane yield loss of 1.1 l_N/kg ODM_origin.

The Zwietering-modified Gompertz model [22] well fitted the data of methane yield loss (R² = 0.948) during the SF phase. The predicated methane yield loss after 40 days of exposure was 188.16 l_N/kg ODM_origin by the model, similar to the detected value 194.74 l_N/kg ODM_origin (Figure 2). The maximum methane yield loss rate (b2) was 11.59 l_N/kg ODM_origin per day by the model. The lag phase (c2) was 0.446 day (10.7 h), which indicated the time interval between the air exposure of silage and the occurrence of methane yield loss. Results showed that the methane yield loss following air exposure can be predicted by the modified Gompertz model. The Gompertz model was first introduced by Gompertz in 1825 [48] and Zwietering et al. [22] proposed the modified Gompertz model by reparametrizing the model, which made the parameters of the Gompertz model more biologically meaningful. The Zwietering-modified Gompertz model is widely used for describing microbial growth kinetics and bioproduction kinetics (e.g., hydrogen, methane, caproate, butanol, and hexanol production) [49,50,51]. As known to the authors, this is the first time that the modified Gompertz model has been used to simulate the ODM loss or methane yield loss during air exposure of maize stalks. These models can be taken into consideration to determine the feeding-out rate of silage.

4. Conclusions

The ODM and methane yield of the maize stalk were well preserved via ensilage by 97.06% and 94.28%, respectively. SF greatly reduced the storage efficiency of the maize stalk by causing 34.11% ODM loss and 52.60% methane yield loss in 40 days and should be avoided. Both ensilage and SF decreased the methane yield potential of maize stalk, with water-soluble component consumption as an important reason. SF negatively affected the bacterial growth and hydrolytic activity in AD. Positive effects of SF on methane production were also observed including lactic acid inhibition alleviation and lignocellulose decomposition, which are worthy of further study. The Zwietering-modified Gompertz model well fitted and predicted the losses in ODM and methane yield during air exposure. The bacterial community shifted from Lactobacillales to Anaerolineales in AD after ensilage and from Ruminoccoccaceae and Lachnospiraceae to Rikenellaceae following the secondary fermentation of silage. Carnobacteriaceae, Moraxellaceae, Lachnospiraceae, Porphyromonadaceae, and Corynebacteriaceae were positively correlated with the content of water-soluble carbohydrates, whereas Anaerolineaceae and Ruminococcaceae were positively correlated with total organic acid content in stalks.