Advancing Thermophilic Anaerobic Digestion of Corn Whole Stillage: Lignocellulose Decomposition and Microbial Community Characterization

: Converting corn grains into bioethanol is an expanding practice for sustainable fuel production, but this is accompanied by the production of large quantities of by-products such as whole stillage. In the present study, the influence of advanced wet oxidation and steam explosion (AWOEx) pretreatment on biogas production and lignocellulose decomposition of corn whole stillage (CWS) was evaluated using semi-continuous thermophilic reactors. The digestion of the CWS was shown to be feasible with an organic loading rate (OLR) of 1.12 ± 0.03 kg VS/m 3 day and a hydraulic retention time (HRT) of 30 days, achieving a methane yield of 0.75 ± 0.05 L CH 4 /g VS fed for untreated stillage and 0.86 ± 0.04 L CH 4 /g VS fed for pretreated stillage, corresponding with an increase in methane yield of about 15%. However, the reactors showed unstable performance with the highest investigated OLRs and shortest HRTs. Under optimal conditions, the conversion efficiencies of COD, cellulose, hemicellulose, and lignin were 88, 95, 97, and 59% for pretreated CWS, and 86, 94, 95, and 51% for untreated CWS, respectively. Microbial community analysis showed that Proteiniphilum , MBA03, and Acetomicrobium were the dominant genera in the digestate and were likely responsible for the conversion of proteins and volatile fatty acids in CWS.


Introduction
Global energy consumption has increased from 434 exajoules (EJ) in 2000 to 628 EJ in 2019 and is projected to reach about 760 EJ in 2050.Fossil fuels represent over 80 percent of the global energy supply.However, oil reserves are being depleted at an alarming rate of 4 billion tons annually, and fossil fuels contribute significantly to global warming, being responsible for almost 90% of total CO 2 emissions and 75% of global greenhouse gas (GHG) emissions.Thus, the focus is increasing on the transition from fossil-based energy to renewable energy sources.Renewable energy sources could potentially provide two-thirds of the global energy demand, resulting in smaller increases in the Earth's surface temperature to below +2 • C before 2050, which is required to prevent irreversible ecological disruptions due to climate change [1].Among renewable energy sources, biomass currently represents the largest share, accounting for 10% of the world's energy supply and 13% of energy consumption [2].Due to the various conversion processes, different forms of energy are produced from biomass including heat, liquid fuels, and gaseous fuels.Bioethanol derived from corn is already an important addition to conventional gasoline in the United States, the largest producer of bioethanol, producing 15 billion gallons of bioethanol in 2021 [3].
The largest amount of US bioethanol is produced from sources such as corn kernels and other agricultural biomasses [4].Corn bioethanol production is a multi-step process that includes the following: (1) dry milling; (2) cooking/liquefaction, where corn mash is formed; (3) saccharification, where starch hydrolyzes into glucose monomers; (4) fermentation, where yeast is added, and bioethanol is produced; and (5) distillation where the bioethanol is purified and recovered [5].During the bioethanol distillation step, a fermentation residue called whole stillage is generated.In a conventional corn bioethanol plant, whole stillage is further separated by centrifugation into wet distillers' grains (WDG) and thin stillage (supernatant).A portion of the thin stillage is generally recycled as backset water and used in the liquefaction process.The remaining portion of the thin stillage is turned into syrup after evaporation.The WDG is combined with the syrup to become wet distillers' grains with soluble (WDGS).As a last step, the WDGS enters a dryer and turns into distiller's dried grains with solubles (DDGS) [6].It has been reported that the production of DDGS from dry milling reached about 22.2 million tons in 2021 in the United States alone, an increase of 10% over 2020 [7].As reported, each bushel of corn yields about 10.22 L of bioethanol and 8.16 kg of DDGS in a dry-milling bioethanol plant [4].
The evaporation and drying steps that produce DDGS consume large amounts of energy [6], corresponding to a third of the energy needed for a dry-milled bioethanol plant [8].This increases the operational cost of the corn bioethanol plants, demonstrating the need for alternatives to valorize the whole stillage while bypassing the energy-intensive drying processes.DDGS is commonly used as animal feed (e.g., cattle, swine, and poultry) due to its high nutritional value (e.g., protein, Ca, S, and P) [4].However, the digestion efficiency of DDGS by monogastric animals is low due to an imbalance of nutrients and high levels of crude protein and dietary fiber content (non-starch polysaccharides, NSP) [9,10].Reportedly, parts of NSP are not digestible by animals and are detrimental to nutrient digestibility due to their structural composition [10].Accordingly, thermal (140-160 • C) pretreatment has been recommended to improve the digestive and fermentative properties of the dietary fiber of whole stillage [10].On the other hand, due to its high fertilizer value, stillage has been used as a fertilizer on farmland.However, the possibility of polluting farmland and the likelihood of soil acidification (low pH of around 4) hinder its widespread application [11].Also, stillage requires thickening to facilitate its transportation, which again entails a costly evaporation process.
Among the alternative valorization approaches is anaerobic digestion (AD), which can be performed without upfront evaporation and stillage drying.The produced biogas can considerably minimize natural gas consumption in dry-milling bioethanol plants with a significant decrease in the quantity of effluent solids.It has been reported that the produced biogas from thin stillage alone can replace more than 50 percent of the energy required for operating the ethanol plant [12].The energy production from whole stillage can be expected to be significantly higher than for thin stillage.AD can be integrated into the bioethanol plant, to avoid the need for transportation of the materials.Overall, the whole stillage substrate has only been researched to a limited extent and, therefore, a detailed and broader investigation is required.To date, most of the previous studies on whole stillage have been conducted using batch tests compared to far broader fermentation studies while converting thin stillage.This has led to limited knowledge and use of the AD process in treating the whole stillage.AD could have great potential to reduce energy use and increase decarbonization of bioethanol production.A few studies focusing specifically on corn whole stillage (CWS) digestion were conducted by Eskicioglu et al. [13], Eskicioglu and Ghorbani, [14], and Gyenge et al. [15].Volatile solids (VS) reduction in the range of 82-83% was reported for thin stillage [16] compared to 82.5-86% for whole stillage [13].Hydraulic retention times (HRTs) from 15 to 40 days were found to be suitable for thin stillage digestion [16,17].Both mesophilic (35-37 • C) and thermophilic (55-60 • C) digestion environments have been examined.The thermophilic conditions had higher methanogenic activity as well as a higher methane content than the mesophilic conditions [13].Ther-mophilic digestion appears to be a suitable process for stillage processing as the stillage leaves the distillation step at a temperature of around 90 • C [18].
Various strategies have been used for improving stillage digestion, such as the addition of additives (e.g., cobalt addition) [12], graphene addition [19], co-digestions [20], and bioaugmentation [21].Among the best-known strategies for enhancing biogas production of organic material is the use of pretreatments, which have been shown to boost the biogas yield of biomass substrates significantly.However, to the authors' knowledge, the effect of pretreatment on biogas improvement in CWS has not been previously investigated.Most studies on pretreatment have focused on improving the properties of DDGS (e.g., nutritive value) as an animal feed [9,10].Since corn fiber contains approximately 35-40% hemicellulose, 15-20% cellulose, and 5-10% lignin [22,23], a tailored pretreatment could potentially make the cellulose and hemicellulose more amenable to subsequent bioprocessing by AD.This could potentially make more valuable renewable energy to supplement the operations of the bioethanol plant and lower costs while decarbonizing the whole operation.Among the prevailing pretreatment methods, steam explosion pretreatment has been found effective in hydrolyzing hemicellulose into monomer sugars as well as increasing the surface area and porosity of lignocellulosic fibers [24].The wet oxidation and steam explosion (AWOEx) pretreatment is typically performed under conditions involving a temperature of 160 to 200 • C, a pressure of 15 to 35 bar, and oxygen of 3 to 20 for a short time (5 min-30 min) [25].High biogas yields along with efficient conversion of cellulose and hemicellulose can be obtained from lignocellulosic biomass pretreated by wet oxidation followed by steam explosion [26][27][28].In this study, the influence of advanced wet oxidation and steam expdlosion (AWOEx) pretreatment on biogas production, digestate properties, and solids reduction in CWS was examined using semi-continuous thermophilic reactors.In the present study, particular emphasis was placed on the effect of wet explosion on the decomposition efficiency of lignocellulosic compounds (cellulose, hemicellulose, and lignin) during AD, which has not been previously studied.Furthermore, the effect of different HRTs on AD performance was investigated.Finally, the bacterial community responsible for stable AD operation of whole stillage, which had not been studied before, was analyzed.

Feed Substrate and Inoculum
CWS was received from a local corn bioethanol plant.CWS was collected in 5-gallon plastic buckets and kept cold at 4 • C until use.A sample of the stillage substrate was pretreated with AWOEx as described below.An active thermophilic inoculum was obtained from an active AD reactor treating dairy cow manure operated in the laboratory at 55 • C at an HRT of 15 days.Characteristics of untreated and treated CWS as well as inoculum are presented in Table 1.

Advanced Wet Oxidation and Steam Explosion (AWOEx) Pretreatment
VS and total solids (TS) values were determined following standard methods [29,30].The pretreatment by AWOEx was performed at a temperature of 175 • C, an oxygen dosage of 5%, and a retention time of 20 min [27].AWOEx is a thermo-chemical method that incorporates wet oxidation (e.g., use of oxidizing agents such as air or pure O 2 ) and steam explosion (e.g., hot steam above 160 • C and 10 bars followed by explosive rupturing of the biomass) with the aim of decomposition of lignocellulosic biomass recalcitrant structure (mainly hemicellulose) and solubilization of lignin [31].

Reactors Setup and Operation
Both untreated and pretreated whole stillages were digested.Two identical reactors were operated, each reactor had a total volume of 2 L and an active working volume of 1.5 L. The VS concentration in the inoculum was 2.2%, and the inoculum was evenly distributed among the reactors.The digestion was performed under thermophilic conditions (55 • C) using a shaking water bath.At start-up, the reactors were purged with a gas mixture of 70% N 2 and 30% CO 2 for about 5 min to ensure anaerobic conditions.Each reactor was equipped with a feed line and biogas sampling port.The biogas sampling port was connected to the reactor headspace, while the feed line extended to the bottom of the digester.Raw whole stillage contained TS of 11.01%, VS of 10.3%, and VS/TS of 94%.Pretreated whole stillage contained TS of 9.05%, VS of 8.40%, and VS/TS of 93% (Table 1).Before feeding the reactor, both pretreated and untreated whole stillage adjusted to about 5% of TS, while the VS contents were in the range of 4.21-4.61%.The reactors were operated in a semi-continuous feeding mode.The systems were fed daily over an approximate 24 h period.Before feeding, excess sludge is drawn out and sampled to monitor pH values and volatile fatty acids (VFAs).The applied OLRs ranged from 1.12 to 3.35 kg VS/m 3 .d.The pH of the bioreactors, the biogas production rate, and the biogas composition were monitored daily.VFAs were measured twice a week.The effect of various HRTs (20, 10, and 30 days) on the biogas yield and VS reduction was evaluated starting with 20 days of HRT, and when the steady-state digestion was reached (2 retention times with stable performances) and all data had been logged, the process was moved to the new operational conditions.To control the desired HRT, a specified amount of excess sludge is withdrawn from the reactor every day and fed with the same amount.

Characterization of Whole Stillage, Reactor Feed, and Inoculum
Both non-pretreated and pretreated stillages were analyzed using high-performance liquid chromatography (HPLC) (Agilent Dionex 3000 technology, Santa Clara, CA, USA) to individual VFAs and C5 and C6 sugars.HPLC analysis was performed at 60 • C using an Aminex HPX-87H column (BIO-RAD, Hercules, CA, USA).The mobile phase was 5 mM H 2 SO 4 .The flow rate was 0.6 mL/min, and the injection volume was 10 µL.VFAs and sugars were analyzed with a refractive index detector (Shodex RI-101, New York, NY, USA).Before VFA analysis, the samples were centrifuged for 10 min at 14,000 rpm in an Eppendorf Centrifuge 5417R.The supernatant was then diluted as required followed by filtration (0.45 µm) using an Adamas-Beta Syringe filter.Samples were analyzed in triplicate and the results were averaged.
The total and soluble chemical oxygen demand (COD) of the untreated and pretreated whole stillage was determined according to the standard closed reflux protocol [30] using high-range reagent vials (Hanna dichromate COD method).The COD vials were heated using a CR 25 COD reactor (Rocker, Linkou Dist., New Taipei, Taiwan), while colorimetric measurements were performed with a genesys150 UV-visible spectrophotometer at a UV absorbance of 620 nm.A calibration curve was prepared using CHEMetrics COD standard solution.To obtain homogeneous samples, pretreated and untreated stillages were thoroughly mixed using a high-rpm Syvio kitchen blender (West Hollywood, CA, USA).
Medium-chained fatty acids (isocaproic and caproic acids) were quantified in a Gas Chromatograph-Flame Ionization Detector (GC-FID) system (TRACE GC Ultra, Thermo Scientific, Waltham, MA, USA) equipped with a dual split/splitless injector and a capillary chromatographic column.The carrier gas was hydrogen with a 2.50 mL/min flow rate.Injections were carried out in splitless mode (splitless time 1 min) with a split flow of 20 mL/min and 200 • C injector temperature.
The structural composition of carbohydrates of pretreated, untreated feed, and reactor effluents was performed according to the National Renewable Energy Laboratory (NREL) standard procedure [29].The dry biomass sample (0.3 g) was placed in a pressure tube, and then 3 mL of H 2 SO 4 72% was added.The sample was mixed well and incubated at 30 • C for 1 h using a water bath.Subsequently, the samples were diluted to 4% and autoclaved at 121 • C for one hour.Acid-soluble lignin was analyzed using a UV-vis spectrophotometer at a wavelength of 320 nm.The amount of structural carbohydrates was measured using the HPLC system (Sunnyvale, CA, USA).Samples were analyzed in triplicate and the results were averaged.

Biogas Production Measurements and Composition Analysis
The biogas production rate (volumetric amount) was measured and recorded daily using a water displacement approach, while the biogas composition was measured daily using a mass spectrometer universal gas analyzer (UGA 100, SRS, Stanford Research System, Sunnyvale, CA, USA), equipped with an ionizer, a quadrupole, and a Faraday cup detector.For biogas analysis, 10 mL of biogas were withdrawn from the reactor through a biogas sampling port that was installed on the biogas line using three-way stopcocks with Luer-lock fittings.

Digestate Characterization
Effluents of pretreated and untreated reactors were characterized in terms of organic matter concentration, VFA, COD, and structural composition of lignin and carbohydrates using the standard methods described above in Section 2.4.1.

DNA Isolation
Digestate samples were collected at the end of each run, corresponding to the regimes of 20-, 10-, and 30-day HRT, respectively, from the 1.5 L AD reactor treating untreated whole stillage.Due to the close performance pattern of pretreated and untreated stillage, microbial structure analysis was performed on untreated stillage samples only.The digestate samples were centrifuged at 8000 G for 8 min, and the supernatants were discarded.The pellets were resuspended in 10 mL of phosphate-buffer saline (PBS) and reprecipitated at 8000 G for 8 min, this process was repeated two times to remove water-soluble contaminants.The pellets were then transferred to microcentrifuge tubes and preserved at −20 • C for later lysis and DNA isolation.Cell lysis was performed by freezing at −80 • C and thawing at 25 • C three times.The use of lysozyme was avoided to prevent the overrepresentation of bacteria with thin peptidoglycan cell walls during taxa quantification.DNA isolation was conducted using standard extraction with phenol-chloroform-isoamyl alcohol (25:24:1) and precipitation with bioethanol [32].The DNA extracts were purified using the MicroElute DNA-cleanup kit (D6296-00, EZNA, Norcross, GA, USA) according to the instructions of the manufacturer.The DNA quality and concentration were measured in a fluorometer QuBit 2.0 (Invitrogen, Waltham, MA, USA) using the "high sensitivity dsDNA" and "high sensitivity RNA" protocols.The DNA in all samples was normalized to 20 ng/µL before proceeding to the amplification of the 16S rRNA gene by polymerase chain reaction (PCR).

Taxonomic Identification of Amplicon Sequence Variants
The analysis of the sequencing reads was performed using QIIME 2 2023.5 [33] operating on the Conda environment in Ubuntu 22.04.The forward and reverse sequences were denoised, clustered, and chimera-filtered using the "denoise-single" method from DADA2 [34] with a truncation length of 250 bp.The pre-trained classifier "Silva 138-99" was used for phylogenetic annotation of the 16S rDNA sequences [35,36].A total of 685 taxa were identified across all samples.Later, the taxa table, frequency table, and metadata files generated in QIIME 2 were imported into R using the "qiime2r" package [37].Data normalization and visualization were conducted in R using the "phyloseq" and "phylosmith" packages [38].The alpha diversity metrics of richness, abundance-based coverage estimator (ACE), Shannon, Simpson, and Fisher were calculated for each sample of different HRTs.

Statistical Analysis and Calculations
Results were evaluated by one-way analysis of variance (ANOVA), followed by Tukey's HSD test for pairwise analysis.Differences were considered statistically significant at a confidence level of p < 0.05.The results described in the tables and figures represent means and standard deviations of the results, where applicable.

Feed Substrate Properties
The main physicochemical characteristics of pretreated and untreated stillage are listed in Table 1.After pretreatment, the VS/TS is reduced due to condensation of some of the injected steam into the material.Also, a material loss of 1.4% can be calculated due to the pretreatment intensity.After pretreatment with AWOEx, the pretreatment liquor was analyzed, and acetic acid was found to have increased by 21%, while propionic acid increased by 132% compared to the non-pretreated samples.Acetate is generated mainly from the breakdown of acetyl groups in hemicellulose due to high temperature and pressure.Propionate can be produced from the α-hydroxyl group of lactic acid through a hydrogenolysis reaction [39].The high concentration of VFAs makes the pretreated material a readily available substrate for methanogenic bacteria.This is because these acids are a precursor to methane production.However, the concentration of butyric acid decreased after pretreatment of the substrate compared with acetic and propionic acids.Although pretreatment increases common AD intermediates, a balance of VFA rations is required to maintain optimal conditions for microbial activity and biogas production.
In terms of solubilized sugars in the pretreatment liquor, xylose, a degradation product of hemicellulose, increased by 1037.5%, while arabinose increased by 152.6%.Cellobiose, a disaccharide derived from cellulose, decreased by 37% after pretreatment, while the glucose concentration increased from below the detection limit to 90 mg/L.After pretreatment, hemicellulose concentration decreased by 47%, while cellulose decreased by approximately 4.8% and the lignin concentration increased by about 8%.The rise in lignin concentration after pretreatment can be elucidated by the phenomenon of pseudo-lignin formation which originates from aromatics and influences K-lignin contents but is not derived from the native lignin.This phenomenon has also been reported to cause loss of sugars during steam explosion pretreatment.The greater reduction in hemicellulose fraction is due to its easier dissolvability during the pretreatment compared to cellulose, which is left behind and, therefore, as found for lignin, will make a larger contribution in the pretreated material.Hydrolysis of hemicellulose and depolymerization of lignin make cellulose more accessible to enzymes due to increased surface area, decreased crystallinity, and enhanced porosity, resulting in improved biogas production and carbon conversion during digestion.The reduction in lignin and hemicellulose was further confirmed by the decrease in neutral detergent (NDF) fiber from 26.50 to 16.78% and the increase in acid detergent fiber (ADF) from 8.98 to 9.88%.The obtained TCOD ranged from 112 to 116 g/L, which is in line with results reported in the literature for whole stillage (117 g/L) [15].Compared to the untreated stillage, the soluble COD concentration of the pretreated stillage increased by more than 10%.

Biogas Production and Composition
Figure 1 shows the daily biogas production rate for pretreated and untreated whole stillages under different HRTs.For the pretreated stillage, 2 L•L −1 •d −1 was produced with 20 days of HRT (OLR of 1.67 kg VS/m 3 d) under the steady-state conditions, while the production of biogas of the same substrate was 1.5 L•L −1 •d −1 with 30 days of HRT (OLR of 1.12 kg VS/m 3 d).When operating at 10 days of HRT (OLR of 3.35 kg VS/m 3 d), up to 3 L of biogas per L reactor can be produced from the pretreated stillage.However, the production of biogas dropped sharply after 6 days and the general performance deteriorated, resulting in VFA accumulation.When operating at 10 days of HRT, the untreated stillage showed the same pattern as the pretreated stillage, with biogas production reaching approximately 2.5 L•L −1 •d −1 in the first 6 days of operation before sharply declining.In contrast to the 10-day HRT, biogas production from the untreated stillage under 20-day HRT and 30-day HRT was more stable and had an average production rate of 1.70 and 1.30 L•L −1 •d −1 , respectively.When treating a similar substrate at thermophilic temperatures, Eskicioglu et al. [13] reported 1.6 L•L −1 •d −1 using semi-continuous bioreactors at 60-day SRT and OLR of 1.93 kg VS/m 3 d.In contrast, the biogas production when operating at higher OLR (4.33 kg VS/m 3 d) at mesophilic conditions was only about 0.70 L•L −1 •d −1 [13].These results may indicate that the digestion of CWS at an elevated level of OLR can be problematic.
When operating at 10 days of HRT, pH dropped, biogas production declined, methane content was reduced by 8%, and CO 2 increased by 7% for the reactors of both pretreated and non-pretreated substrates after 6 days of operations (Figures 2-4).To restore the reactor performance, feeding was stopped for two days and then resumed.When the feeding was resumed, the biogas production increased until it reached its highest levels after which it decreased again (Figure 1).The feeding was stopped once again for two days to restore the performance of the reactor.It is worth noting that there was foam formed in the headspace of the bioreactors at 10 days HRT.The foam was more noticeable in the reactor processing untreated stillage compared to the reactor digesting pretreated stillage.
results may indicate that the digestion of CWS at an elevated level of OLR can be problematic.When operating at 10 days of HRT, pH dropped, biogas production declined, methane content was reduced by 8%, and CO2 increased by 7% for the reactors of both pretreated and non-pretreated substrates after 6 days of operations (Figures 2-4).To restore the reactor performance, feeding was stopped for two days and then resumed.When the feeding was resumed, the biogas production increased until it reached its highest levels after which it decreased again (Figure 1).The feeding was stopped once again for two days to restore the performance of the reactor.It is worth noting that there was foam formed in the headspace of the bioreactors at 10 days HRT.The foam was more noticeable in the reactor processing untreated stillage compared to the reactor digesting pretreated stillage.
When comparing biogas production from pretreated and non-treated stillages, it was found that the biogas production rate of pretreated stillage was increased by more than 15% on average (the average was calculated from the total operational time of the HRT) for 30-day HRT compared to untreated stillage (control), while the average increase in the biogas for 20 days of HRT was 13.7% compared to control.When the reactor was operated at a shorter HRT (10 days), the increase in biogas production reached about 25% compared to control, before it quickly declined due to deteriorating reactor performance.No matter whether the substrate is pretreated or not, no significant fluctuation in the biogas production was observed when the reactors were operated at a lower OLR (30-day HRT) compared to the higher OLRs (10-and 20-day HRT) (As shown in Figures 1 and 2).
Figure 2 exhibits the biogas yield of pretreated and untreated whole stillage under 20-day HRT, 10-day HRT, and 30-day HRT, respectively.From the results reported in Figure 2, the average biogas yield for the pretreated stillage was 1.16 ± 0.12 L biogas/g VSfed for 20-day HRT, 0.56 ± 0.26 L biogas/g VSfed for 10-day HRT, and 1.36 ± 0.07 L biogas/g VSfed for 30-day HRT, at average at stable operations.The mean biogas yields of the untreated substrate at stable operations were 1.03 ± 0.11, 0.53 ± 0.17-, and 1.20 ± 0.06 L biogas/g VSfed for Figures 1 and 2, the 30-day HRT appears to be the HRT to choose for whole stillage digestion in terms of biogas yield (1.2-1.4L biogas/g VSfed), methane content (70%), and VS conversion (87%), compared with 20-and 10-day HRT.Schaefer and Sung [17] reported similar findings when treating corn thin stillage at a thermophilic range (55 When comparing biogas production from pretreated and non-treated stillages, it was found that the biogas production rate of pretreated stillage was increased by more than 15% on average (the average was calculated from the total operational time of the HRT) for 30-day HRT compared to untreated stillage (control), while the average increase in the biogas for 20 days of HRT was 13.7% compared to control.When the reactor was operated at a shorter HRT (10 days), the increase in biogas production reached about 25% compared to control, before it quickly declined due to deteriorating reactor performance.No matter whether the substrate is pretreated or not, no significant fluctuation in the biogas production was observed when the reactors were operated at a lower OLR (30-day HRT) compared to the higher OLRs (10-and 20-day HRT) (As shown in Figures 1 and 2).
Figure 2 exhibits the biogas yield of pretreated and untreated whole stillage under 20day HRT, 10-day HRT, and 30-day HRT, respectively.From the results reported in Figure 2, the average biogas yield for the pretreated stillage was 1.16 ± 0.12 L biogas/g VS fed for 20-day HRT, 0.56 ± 0.26 L biogas/g VS fed for 10-day HRT, and 1.36 ± 0.07 L biogas/g VS fed for 30-day HRT, at average at stable operations.The mean biogas yields of the untreated substrate at stable operations were 1.03 ± 0.11, 0.53 ± 0.17-, and 1.20 ± 0.06 L biogas/g VS fed for Figures 1 and 2, the 30-day HRT appears to be the HRT to choose for whole stillage digestion in terms of biogas yield (1.2-1.4L biogas/g VS fed ), methane content (70%), and VS conversion (87%), compared with 20-and 10-day HRT.Schaefer and Sung [17] reported similar findings when treating corn thin stillage at a thermophilic range (55 • C) using a CSTR reactor, where the system performance was more stable at 30 days of HRT compared to shorter 12-day HRT, where their AD system failed.
In addition to biogas yield, the average CH 4 yields in this study at 0 • C, 1 atm (STP) of pretreated stillage were 0.72 ± 0.07, 0.32 ± 0.15, and 0.86 ± 0.04 L CH 4 /g VS fed for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.Whereas the average methane yields at 0 • C, 1 atm of untreated stillage were 0.65 ± 0.07, 0.31 ± 0.10, and 0.75 ± 0.05 L CH 4 /g VS fed for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.Methane production between treated and untreated whole stillage was significantly different (p < 0.01) under HRTs of 20 days and 30 days.However, methane production was not significantly different (p < 0.05) between treated and untreated whole stillage under 10-day HRT.This is mainly due to the unstable performance of shorter HRT.Under steady-state conditions, the methane yields observed in our study are slightly higher than the values found by Eskicioglu and Ghorbani [14] where the specific methane yield ranged from 0.40 to 0.50 L CH 4 /g VS added (at STP) determined by biochemical methane potential (BMP) assays under mesophilic range (35 ± 2 • C).However, the methane yield of the same substrate under thermophilic conditions (55 • C) was 0.60 ± 0.08 L CH 4 /g VS added (at STP) [13], which is to some extent in line with the methane yield observed in this study.The higher biogas yields obtained after pretreating the feed material in our study may be explained by the depolymerization of nonstarch polysaccharides (NSP) during the AWOEx process into monomers and oligomers, which become accessible to acidogenic and other fermentative microorganisms.After pretreatment, arabinoxylans decomposed leading to a significant increase in free sugars.Reportedly, NSP is not digestible by animals due to its cross-linkage [10], thus its valorization via AD coupled with a prehydrolysis step with AWOEx might be preferred.The relatively higher methane production of untreated stillage achieved in this study can be attributed to the lower OLR used as well as to the high fat (8.2-7.1%), protein (32.79-29.96%),and lactic acid (3.6-4.3%)contents of CWS substrate (Table 1).As reported in the literature, methane production from protein and lipids can reach 0.50 L CH 4 /g VS and 1.01 L CH 4 /g VS [40], respectively.Also, CWS contains a relatively high concentration of total volatile fatty acids (3.8-3.9%)(Table 1).VFAs have a high methane potential and could yield up to 1.29 g CH 4 /g COD fed [41].Thus, due to its higher lipid and fatty acid content, CWS represents an attractive source of methane, and its relatively high methane production is well justified.The findings of our study also suggest that using a low loading rate is the optimal option for treating whole stillage to methane.Since whole stillage contains significant amounts of lactic acid and VFAs (3.8-3.9%) as fermentation products, overloading could lead to VFA build-up and process problems.From the above results, it can be concluded that methane production from corn bioethanol whole stillage is promising and could be an economical processing option for the corn bioethanol industry while increasing their decarbonization score of importance for the final credit of the fuels produced.
Fermentation 2024, 10, x FOR PEER REVIEW 9 of 21 °C) using a CSTR reactor, where the system performance was more stable at 30 days of HRT compared to shorter 12-day HRT, where their AD system failed.In addition to biogas yield, the average CH4 yields in this study at 0 °C, 1 atm (STP) of pretreated stillage were 0.72 ± 0.07, 0.32 ± 0.15, and 0.86 ± 0.04 L CH4/g VSfed for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.Whereas the average methane yields at 0 °C, 1 atm of untreated stillage were 0.65 ± 0.07, 0.31 ± 0.10, and 0.75 ± 0.05 L CH4/g VS fed for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.Methane production between treated and untreated whole stillage was significantly different (p < 0.01) under HRTs of 20 days and 30 days.However, methane production was not significantly different (p < 0.05) between treated and untreated whole stillage under 10-day HRT.This is mainly due to the unstable performance of shorter HRT.Under steady-state conditions, the methane yields observed in our study are slightly higher than the values found by Eskicioglu and Ghorbani [14] where the specific methane yield ranged from 0.40 to 0.50 L CH4/g VSadded (at STP) determined by biochemical methane potential (BMP) assays under mesophilic range (35 ± 2 °C).However, the methane yield of the same substrate under thermophilic conditions (55 °C) was 0.60 ± 0.08 L CH4/g VSadded (at STP) [13], which is to some extent in line with the methane yield observed in this study.The higher biogas yields obtained after Figure 3a,b display the biogas composition of pretreated and untreated whole stillage under various digestion conditions.The average methane content of the pretreated stillage was 68.87%, 62.62%, and 70.08% for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.The average methane content of the untreated stillage was 69.67%, 63.52%, and 69.91% for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.Similar results have been reported for biogas production from CWS, where the methane concentration reached 61-67% [15].Under 20-day and 30-day HRTs, the methane content and biogas composition did not differ significantly between the treated and untreated whole stillages.However, the methane content in 10-day HRT was significantly different (p < 0.01) from the methane content in 20-day and 30-day HRT.The average carbon dioxide content of the untreated stillage was 27.70%, 33.09%, and 26.19% for 20-day HRT, 10-day HRT, and 30-day HRT, respectively, while the mean for the untreated stillage was 27.19%, 32.58%, and 26.88% for 20-day HRT, 10-day HRT, and 30-day HRT, respectively.When the volatile fatty acids in the system increased and pH decreased at 10 days of HRT, a decline in the biogas production was found as well as a decrease in the methane content (Figure 3a,b and Figure 4), indicating a negative impact on the methanogenic community in the reactor.It has been reported in the literature that the integration of a bioethanol plant with anaerobic digestion using thin stillage as a sole substrate is expected to boost the net energy balance ratio from 1.3 to 1.8 [16], along with a reduction in the overall GHG emissions of the bioethanol plant.Ziero et al. [42] reported a 65% methane content of the thin stillage, and the resulting biogas was estimated to generate 53% of the total electricity of a stand-alone corn bioethanol plant when using a combined heat and power system.This percentage is expected to be much higher when using whole stillage as substrate, due to the higher energy potential of the raw material resulting in higher amounts of methane-rich biogas compared to thin stillage.

Volatile Fatty Acids
Figure 4a,b show the volatile fatty acids (VFAs) concentrations and pH values under the tested HRTs.Largely pretreated and untreated stillages showed a similar pattern in terms of VFA concentrations under the examined conditions.In the first 13 days of operation of 20-day HRT (OLR of 1.67 kg VS/m 3 d), the propionic acid concentration increased from 1.7 to 2.8 g/L and remained at this level throughout the operation period.Similar findings were reported by Eskicioglu et al. [13] when treating whole stillage at thermophilic conditions and an OLR of 1.93 kg VS/m 3 d, and 60-day SRT, resulting in an increase in propionic acid from 0.3 to ca. 3 g/L, whereafter it stabilized at this concentration.For both untreated and pretreated stillages, the concentration of acetic acid was high at first, which led to a slight decrease in the pH, after which it decreased to a stabilized level of 0.5 g/L.A similar increase in propionic acid occurred when switching to an OLR of 3.35 kg VS/m 3 d (10-day HRT), but the increase this time was as high as 3.9 g/L, suggesting an imbalance between the bacterial populations degrading this acid and the concentration of  Largely pretreated and untreated stillages showed a similar pattern in terms of VFA concentrations under the examined conditions.In the first 13 days of operation of 20-day HRT (OLR of 1.67 kg VS/m 3 d), the propionic acid concentration increased from 1.7 to 2.8 g/L and remained at this level throughout the operation period.Similar findings were reported by Eskicioglu et al. [13] when treating whole stillage at thermophilic conditions and an OLR of 1.93 kg VS/m 3 d, and 60-day SRT, resulting in an increase in propionic acid from 0.3 to ca. 3 g/L, whereafter it stabilized at this concentration.For both untreated and pretreated stillages, the concentration of acetic acid was high at first, which led to a slight decrease in the pH, after which it decreased to a stabilized level of 0.5 g/L.A similar increase in propionic acid occurred when switching to an OLR of 3.35 kg VS/m 3 d (10-day HRT), but the increase this time was as high as 3.9 g/L, suggesting an imbalance between the bacterial populations degrading this acid and the concentration of hydrogen in the system.In this case, however, the increase in propionic acid was more pronounced for the pretreated stillage than for the untreated stillage.Within two days without feeding, the system recovered, and the propionic acid concentration decreased to around 2.4 g/L.However, when the feeding was started again, the concentration of propionic acid increased again indicating that this organic loading rate was not sufficient for stable degradation of intermediates produced during the degradation of whole stillage.Tukey's HSD post hoc test showed that the concentration of total VFAs was significantly higher in the 10-day HRT condition (p < 0.01) compared to the other two HRT conditions (20 days and 30 days).
As shown in Figures 1-4, the increase in the concentration of VFAs while the reactor was operated at 10 days HRT was accompanied by a decrease in pH, a sharp decline in biogas production, and a deterioration of biogas content.The performance problems might mainly be attributed to the increase in propionic acid concentrations.Eskicioglu et al. [13] observed a significant decrease in biogas production due to propionic acid accumulation of about 3 g/L with propionic acid to acetic acid ratio of 2.6.Due to the higher concentration of lactic acid in the stillage substrate (3.6 g/L in the pretreated stillage and 4.3 g/L in the untreated), propionic acid is probably produced in the bioreactor and accumulates as the conditions are unfavorable for degradation of this acid.This was further confirmed by our microbial community analysis.Bo et al. [43] reported a similar accumulation of propionic acid due to the higher concentration of lactic acid in the feed when evaluating the effect of lactic acid on methanogenesis.
Under the longest HRT (a 30-day HRT, OLR of 1.12 kg VS/m 3 d), VFAs of both pretreated and non-pretreated substrates showed a different trend compared to the shorter HRTs.The propionic acid concentration dropped immediately to a lower concentration, while the concentration of butyric acid was almost doubled.It is also worth noting that VFAs did change immediately after the retention time was changed from 10 days to 30-day HRT.

Organic Matter Conversions (VS, COD, and Carbohydrates)
Table 2 summarizes the characteristics of the raw materials and organic matter conversion efficiencies of the effluent digestate under the tested conditions.VS degradation of pretreated stillage was 81%, 75%, and 86% at HRT of 20, 10, and 30 days, respectively.Whereas the conversion of VS for untreated stillage was 77%, 73%, and 84% at HRT of 20, 10, and 30 days, respectively.Analysis of the biomass composition of the effluent showed that most of the carbohydrates were converted to biogas, with conversion efficiencies ranging from 83 to 95.2% for cellulose, 90.6-97% for hemicellulose, and 31.7-58.9%for lignin depending on the conditions applied.As shown in Table 2, the cellulose, hemicellulose, and lignin conversion efficiencies were higher for the pretreated material than the untreated material, where the increase in decomposition ranged from 1.1 to 7.8% for cellulose, 1.5-2.4% for hemicellulose, and 1.2-27.3%for lignin.One-way ANOVA analysis indicated that the p-value was less than 0.05 implying that the effect of pretreatment was significant.Post hoc Scheffé analyses showed that the lignocellulosic conversion efficiency was significantly higher in the pretreated whole stillage compared to the untreated whole stillage (p < 0.01).Compared to carbohydrates, lignin conversion was significantly higher in treated material compared to untreated material.Lignin degradation was approximately 32% higher in the treated materials compared to the untreated materials at 20-day HRT.However, the difference in lignin degradation between treated and untreated whole stillage was only about 16% at 30-day HRT, suggesting that not only the pretreatment but also the longer HRT influenced the effectiveness of degradation.One possible explanation for the higher lignocellulose decomposition after pretreatment is the complete hydrolysis of hemicellulose resulting in lignin breakage leaving the cellulose accessible to microorganisms.It has been noted that the longer the HRT, the greater the degree of lignocellulose conversion.The higher degradation efficiency at longer HRT is justified because longer HRT provides microorganisms more time to metabolize complex organic materials such as lignin and cellulose.In addition to increasing digestion time, longer HRT can also help dilute inhibitory substances, thus resulting in a stable process, which suggests better carbon conversion.Interestingly, it further showed that the differences between the degree of conversion decreased between pretreated and untreated materials when the HRT increased.This indicates that the longer the retention time, the more of the lignocellulosic material will be converted into the untreated stillage.
The TCOD conversion efficiencies showed a similar trend as that of lignocellulose and were in the range of 73-87.6%,where the longest HRT and lowest OLR achieved the highest conversion efficiency.In longer HRT, the pretreated substrate had a higher COD conversion compared to the untreated substrate.In longer HRT, the microbial community may be more balanced, with a wider range of microorganisms able to degrade different types of organic complexes, leading to more efficient COD conversion.The COD conversions observed in this investigation are slightly higher than those reported in previous results.Eskicioglu et al.'s [13] results showed VS and TCOD conversion efficiencies of 86.4% and 85.7%, respectively, when digested with CWS under thermophilic conditions and an OLR of 1.93 kg VS/m 3 d.The results of the present study are also slightly higher compared to those reported for the corn thin stillage.Lee et al. [16] reported TCOD conversion of 86% and VS reduction of 82% at 30-day HRT (OLR of 1.9 kg VS/m 3 d) when digesting corn thin stillage under mesophilic conditions.Due to the high degradation efficiency of cellulose and hemicellulose, lignin appears to be the largest contributor to the increase in total COD in the systems as its conversion only ranged between 31.7 and 58.9%.These results were also confirmed by the high lignin concentration in the reactor effluents, where lignin concentrations ranged between 24% and 35% in the effluent (Table 2).-indicates no detected concentration or below the detection limit.

Microbial Community Structure
The taxa identified in all 20-, 10-, and 30-day HRT samples of untreated stillage digestate corresponded to more than 99% of the total amplicon sequence variants (ASVs).This indicated a high coverage in the identification of bacterial taxa during the microbial community analysis.In general, more than 50% of the bacterial communities in the whole stillage digestate were members of the classes Clostridia, Limnochordia, Bacteroidia, and/or Synergistia (See Figure 5).These classes are composed of anaerobic chemoorganotrophic bacteria.Clostridia, Limnochordia, Bacteroidia, and Synergistia have been found in AD reactors treating other substrates like food waste [44], playing a major role in the degradation of carbohydrates, lipids, and proteins to VFA, H 2, and CO 2 .Clostridia and Bacteroidia are also ubiquitous in natural anaerobic environments, where they are found" for instance, in rumen and lake sediments [45]. in the degradation of carbohydrates, lipids, and proteins to VFA, H2, and CO2.Clostridia and Bacteroidia are also ubiquitous in natural anaerobic environments, where they are found,, for instance, in rumen and lake sediments [45].To further widen the understanding of the role of different bacteria in the AD process of whole stillage, the relative abundance of the twenty most dominant taxa was analyzed at the genus level.As shown in Figure 6, Proteiniphilum, MBA03, and Acetomicrobium were the dominant genera for 10-, 20-, and 30-day HRT, respectively.Shifts in the abundance of these genera were detected across all samples.For instance, the relative abundances of Proteiniphilum were ca.17.2, 2.4, and 14.6% for 10-, 20-, and 30-day HRT, respectively.The relative abundances of the genus MBA03 across samples were ca.2.8, 16.0, and 2.1% for 10-, 20-, and 30-day HRT, respectively.The relative abundances of Acetomicrobium were ca.9.3 and 17.1% for 10-and 30-day HRT, respectively.Acetomicrobium was not detected in the sample corresponding to 20 days of HRT.The high relative abundance of Proteiniphilum and Acetomicrobium can be attributed to the high protein (32.79-29.96%)and total volatile fatty acid concentration (3.8-3.9%) in the whole stillage.To further widen the understanding of the role of different bacteria in the AD process of whole stillage, the relative abundance of the twenty most dominant taxa was analyzed at the genus level.As shown in Figure 6, Proteiniphilum, MBA03, and Acetomicrobium were the dominant genera for 10-, 20-, and 30-day HRT, respectively.Shifts in the abundance of these genera were detected across all samples.For instance, the relative abundances of Proteiniphilum were ca.17.2, 2.4, and 14.6% for 10-, 20-, and 30-day HRT, respectively.The relative abundances of the genus MBA03 across samples were ca.2.8, 16.0, and 2.1% for 10-, 20-, and 30-day HRT, respectively.The relative abundances of Acetomicrobium were ca.9.3 and 17.1% for 10-and 30-day HRT, respectively.Acetomicrobium was not detected in the sample corresponding to 20 days of HRT.The high relative abundance of Proteiniphilum and Acetomicrobium can be attributed to the high protein (32.79-29.96%)and total volatile fatty acid concentration (3.8-3.9%) in the whole stillage.Proteiniphilum is a genus of anaerobic, proteolytic, chemoorganotrophic bacteria.Most members of Proteiniphilum, such as P. acetatigens, P. propionicum, and P. saccharofermentans, grow on organic substrates that are present in whole stillage, including starch, maltose, lactose, glucose, and lactic acid [46].Proteiniphilum bacteria produce acetic, propionic, and isovaleric acids as main metabolic products [45][46][47].Whole stillage is particularly rich in protein and lactic acid (as shown in Table 1), thus, the high abundance of Proteiniphilum in the samples for 10-and 30-day HRT may be explained by both the proteolytic and lactate oxidation capacities of Proteiniphilum, which allowed this genus to develop over time in the AD reactor.The high abundance of Proteiniphilum coupled with a high OLR was likely responsible for the accumulation of propionic acid when the reactor was operated at 10 days of HRT.There were other acidogens with a high relative abundance that also may have contributed to the accumulation of propionic acid in 10 of days HRT including DTU014 (10.7%),Defluviitoga (8.7%), and Tepidanaerobacter (6.9%).
Accumulation of propionic acid is a common symptom of reduced methanogenesis during AD.Therefore, some of the fluctuations that were observed in the methane content in the period corresponding to 10 days of HRT, were likely associated with a high abundance of acidogens, such as Proteiniphilum.The instabilities observed in 10 days of HRT were later alleviated when the OLR was decreased, this occurred together with an increase in relative abundance of Acetomicrobium in the bioreactor in 30 days of HRT.Acetomicrobium is a genus of acetogenic bacteria that ferments sugars and VFA into acetic acid, hydrogen, and carbon dioxide [48].Captivatingly, it has been found that the AD reactor shifted from an unstable (10 days HRT) to a stable operational regime (30 days of HRT) when the microcosmos was partially dominated by Acetomicrobium (17.1%), concurred with a decrease in propionate concentration and an increase in methane yield (see operational data in Figures 1 and 2).These observations suggest that Acetomicrobium was highly Proteiniphilum is a genus of anaerobic, proteolytic, chemoorganotrophic bacteria.Most members of Proteiniphilum, such as P. acetatigens, P. propionicum, and P. saccharofermentans, grow on organic substrates that are present in whole stillage, including starch, maltose, lactose, glucose, and lactic acid [46].Proteiniphilum bacteria produce acetic, propionic, and isovaleric acids as main metabolic products [45][46][47].Whole stillage is particularly rich in protein and lactic acid (as shown in Table 1), thus, the high abundance of Proteiniphilum in the samples for 10-and 30-day HRT may be explained by both the proteolytic and lactate oxidation capacities of Proteiniphilum, which allowed this genus to develop over time in the AD reactor.The high abundance of Proteiniphilum coupled with a high OLR was likely responsible for the accumulation of propionic acid when the reactor was operated at 10 days of HRT.There were other acidogens with a high relative abundance that also may have contributed to the accumulation of propionic acid in 10 of days HRT including DTU014 (10.7%),Defluviitoga (8.7%), and Tepidanaerobacter (6.9%).
Accumulation of propionic acid is a common symptom of reduced methanogenesis during AD.Therefore, some of the fluctuations that were observed in the methane content in the period corresponding to 10 days of HRT, were likely associated with a high abundance of acidogens, such as Proteiniphilum.The instabilities observed in 10 days of HRT were later alleviated when the OLR was decreased, this occurred together with an increase in relative abundance of Acetomicrobium in the bioreactor in 30 days of HRT.Acetomicrobium is a genus of acetogenic bacteria that ferments sugars and VFA into acetic acid, hydrogen, and carbon dioxide [48].Captivatingly, it has been found that the AD reactor shifted from an unstable (10 days HRT) to a stable operational regime (30 days of HRT) when the microcosmos was partially dominated by Acetomicrobium (17.1%), concurred with a decrease in propionate concentration and an increase in methane yield (see operational data in Figures 1 and 2).These observations suggest that Acetomicrobium was highly beneficial for the AD reactor operating at 30 days of HRT, regulating the levels of propionic acid, and producing more acetic acid, hydrogen, and carbon dioxide to be used by methanogenic archaea.A previous study also showed the enrichment of Acetomicrobium populations over time in anaerobic digesters fed with high loads of propionic acid [49].Regulation between acidogenic and acetogenic communities is key to achieving stable operations in anaerobic digesters.Future research on the gene expression of Proteiniphilum, Acetomicrobium, and methanogenic archaea in bioreactor systems could potentially produce a broader understanding of the role of these genera in the degradation of protein-rich substrates during AD.
A higher relative abundance of Syntrophomonas (5.2%) at 30 days of HRT was observed.This was probably due to an increased concentration of butyric acid in the AD reactor, along with the fact that the feed substrate contains approximately 2 g/L of butyric acid (see Tables 1 and 2).Syntrophomonas are butyric acid-degrading bacteria that are typically found in syntrophic association with hydrogenotrophic methanogenic archaea for interspecific hydrogen and electron transfer [50].Thus, a higher presence of Syntrophomonas in 30-day HRT could be due to the higher viability of methanogenic archaea at these HRT conditions.
The microbiome at 20 days of HRT was considerably different from the rest of the samples.Uncultured MBA03 (class Clostridia) was the dominant genus in 20 days of HRT, with a relative abundance of 16.0%.Other genera with high relative abundance in 20 days of HRT were Pseudomonas (8.2%), Ruminofilibacter (7.0%), Hydrogenispora (6.1%), Ruminiclostridium (5.6%), and HN-HF0106 (5.5%).All these genera have been described to possess important cellulose-and hemicellulose-degrading roles during thermophilic anaerobic digestion [51].The abundance of carbohydrate-degrading microorganisms could be explained by the fact that the inoculum came from a bioreactor processing manure substrate, and these microbes were probably over-represented in the inoculum.Indeed, whole stillage contains a relatively low amount of cellulose (11.7%) and is rich in protein, starch, and lactic acid.Consequently, many of the cellulosic genera were displaced over time by other microbes, such as Proteiniphilum, capable of using the specific substrates in whole stillage.Pseudomonas uses bioethanol as a carbon source when grown in anaerobic conditions [52].Interestingly, the presence of bioethanol (0.20 ± 0.01 g/L) in the whole stillage may have provided conditions for the development of Pseudomonas in 20 days HRT.Pseudomonas uses alcohol dehydrogenases to convert bioethanol into acetaldehyde, which is later transformed into acetate via NAD + -linked enzymes [52].The presence of Pseudomonas was perhaps beneficial to the AD process of whole stillage, as this microbe may be able to convert bioethanol into acetic acid, thereby alleviating potential inhibition from bioethanol and producing a substrate for aceticlastic methanogens.The role of Pseudomonas might have been replaced with other bioethanol-resistant acidogenic and acetogenic bacteria over time and be the reason these microbes were no longer detectable.
Table 3 summarizes the diversity metrics within the bacterial community across HRT conditions.There was a decrease in bacterial richness and diversity from 20-to 10-day HRT.In contrast, the bacterial community diversity was less affected in the transition from 10-to 30-day HRT, and using the Simpson index showed that the diversity was slightly higher at 30-day HRT than at 10-day HRT.The decrease in the bacterial diversity at 10 days of HRT may be related to natural ecological succession in the AD reactor driven by an increase in OLR and dilution rate.This ecological succession favored the development of beneficial acetogenic bacteria such as Acetomicrobium and Syntrophomonas.Regardless of an arguable decrease in diversity over time in the AD bioreactor, the Shannon index in 30 days of HRT (4.07) was high compared with previous lab-scale work at thermophilic conditions (ranging from 0.7 to 4.9) [53].This shows that high diversity in the bacterial community was maintained in the AD reactor despite the instabilities observed at 10 days of HRT.A high diversity of bacteria and archaea might be important to achieve a balanced transfer of metabolites across the microbial groups participating in the four stages of AD.Overall, a diverse microbiome seems to be important to achieve stable methane production.More research focusing on the possible correlation between alpha diversity metrics and the stability of the AD reactor could potentially be of great interest.

Conclusions
Anaerobic digestion of both pretreated and untreated CWS from corn bioethanol plant was carried out in this study.Pretreatment improved biogas production by more than 15%, mostly due to the decomposition of hemicellulose into its monomeric constituents such as xylose and arabinose as well as decomposition of parts of the lignin fraction.It was found that a high OLR is not appropriate when digesting whole stillage anaerobically.It has been found that an OLR of 1.12 kg VS/m 3 d was suitable for whole stillage digestion in terms of high biogas yield and high organic matter conversion (86% VS conversion).Compositional analysis of the effluent showed that the degradation of cellulose, hemicellulose, and lignin during the digestion process ranged between 83 and 95.2%, 90.6-97%, and 31.7-58.9%,respectively.The longer the HRT, the greater the degree of lignocellulose conversion but the smaller the conversion differences between treated and untreated materials.This emphasizes that pretreatment is of greater importance when operating systems with lower HRTs.The microbial community analysis showed that Proteiniphilum, MBA03, and Acetomicrobium were the dominant genera in the whole stillage digestate.It indicates that Proteiniphilum was primarily responsible for the conversion of protein-rich whole stillage substrate, while Acetomicrobium was likely responsible for regulating propionic acid levels in the AD reactor.Overall, this work shows that thermophilic AD of the whole stillage was clearly feasible using a low OLR and relatively long HRT and that the applied pretreatment indeed had a noticeable effect on the hemicellulose and lignin solubilization evident by the higher concentration of soluble COD, aliphatic acids, and monosaccharides in the pretreatment liquor as well as the increased biogas production.A techno-economic analysis would be needed to determine if pretreatment would be an economically viable solution for AD of whole stillage.

Figure 1 .
Figure 1.Daily biogas production rate of pretreated and untreated whole stillage under different HRTs.

Figure 1 .
Figure 1.Daily biogas production rate of pretreated and untreated whole stillage under different HRTs.

Figure 2 .
Figure 2. Biogas yield of pretreated and untreated whole stillage under different HRTs.

Figure 2 .
Figure 2. Biogas yield of pretreated and untreated whole stillage under different HRTs.

Figure 4 .
Figure 4. Volatile fatty acids concentration in the systems of (a) pretreated and (b) untreated stillages over the digestion time and different HRTs.

Figure 4 .
Figure 4. Volatile fatty acids concentration in the systems of (a) pretreated and (b) untreated stillages over the digestion time and different HRTs.

3. 3 .
Figure4a,b show the volatile fatty acids (VFAs) concentrations and pH values under the tested HRTs.Largely pretreated and untreated stillages showed a similar pattern in terms of VFA concentrations under the examined conditions.In the first 13 days of operation of 20-day HRT (OLR of 1.67 kg VS/m 3 d), the propionic acid concentration increased from 1.7 to 2.8 g/L and remained at this level throughout the operation period.Similar findings were reported by Eskicioglu et al.[13] when treating whole stillage at thermophilic conditions and an OLR of 1.93 kg VS/m 3 d, and 60-day SRT, resulting in an increase in propionic acid from 0.3 to ca. 3 g/L, whereafter it stabilized at this concentration.For both untreated and pretreated stillages, the concentration of acetic acid was high at first, which led to a slight decrease in the pH, after which it decreased to a stabilized level of 0.5 g/L.A similar increase in propionic acid occurred when switching to an OLR of 3.35 kg VS/m 3 d (10-day HRT), but the increase this time was as high as 3.9 g/L, suggesting an imbalance between the bacterial populations degrading this acid and the concentration of hydrogen in the system.In this case, however, the increase in propionic acid was more pronounced for the pretreated stillage than for the untreated stillage.Within two days

Figure 5 .
Figure 5. Relative abundance of bacterial classes in the samples collected from the AD reactor at three different operational regimes defined by HRTs of 10, 20, and 30 days.Classes with less than 1% abundance were grouped and tagged as "<1% abundant".

Figure 5 .
Figure 5. Relative abundance of bacterial classes in the samples collected from the AD reactor at three different operational regimes defined by HRTs of 10, 20, and 30 days.Classes with less than 1% abundance were grouped and tagged as "<1% abundant".

Figure 6 .
Figure 6.Relative abundance of the bacterial 16S rDNA ASVs at genus level across samples taken at three regimes in the operation of the thermophilic anaerobic digester fed with untreated whole stillage (10, 20, and 30 days of HRT).Heatmap displays the twenty most abundant taxa across all samples.

Figure 6 .
Figure 6.Relative abundance of the bacterial 16S rDNA ASVs at genus level across samples taken at three regimes in the operation of the thermophilic anaerobic digester fed with untreated whole stillage (10, 20, and 30 days of HRT).Heatmap displays the twenty most abundant taxa across all samples.

Table 1 .
Characteristics of untreated and pretreated whole stillage and inoculum.
-indicates no measured concentration or below the detection limit, * indicates dissolved sugar in the liquid.

Table 2 .
Effluent characteristics of pretreated and untreated substrates under tested digestion HRT conditions.

Table 3 .
Summary of bacterial diversity within the AD reactor processing whole stillage at the various HRT regimes.