Chemicals

Glucose, mannose, galactose, xylose, and arabinose were obtained from Fluka (Milwaukee, WI) and Alfa Aesar (Ward Hill, MA). The reducing reagents (cysteine and glutathione) were purchased from Sigma-Aldrich (St. Louis, MO). Unless otherwise specified all the other reagents used in this study were of analytical grade and obtained from Sigma-Aldrich (St. Louis, MO).

Inoculum and anaerobic medium

The original microbial inoculum used in this experiment was collected from a full-scale continuously stirred tank reactor (CSTR) (Nordre Follo Wastewater Treatment Plant, Vinterbro, Norway) running with sewage sludge and food waste at thermophilic temperature (~ 62 °C). This inoculum was used to run sequential batch bottles fed with untreated and steam-exploded birch. The digestate from the sequential batch bottles was used as inoculum, which has a dry matter (DM) content of 3.8%, the volatile solid (VS) content of 2.1%, and the pH of 7.8. Anaerobic medium was prepared from a mixture of mineral buffer solution, potassium hydrogen phosphate, trace elements, selenite, and vitamins according to Angelidaki et al. with a slight modification [26]. Briefly, a stock solution of mineral buffer solution was prepared (concentration/L): 100-g NH₄Cl, 10-g NaCl, 10-g MgCl₂·6H₂O, and 5-g CaCl₂·2H₂O. A stock solution (1 L) of potassium hydrogen phosphate was prepared from 200-g K₂HPO₄·3H₂O. A stock solution of the vitamin solution was prepared (concentration/L): 2-mg biotin, 2-mg folic acid, 10-mg pyridoxine–HCl, 5-mg thiamine–HCl, 5-mg riboflavin, 0.1-mg vitamin B12, 5-mg nicotinic acid, 5-mg p-aminobenzoic acid, 5-mg lipoic acid, and 5-mg calcium pantothenate. The trace element and selenite solution contained (L⁻¹): 2-g FeCl₂·4H₂O, 0.05-g ZnCl₂, 0.05 MnCl₂·4H₂O, 0.05-g H₃BO₃, 0.05-g CoCl₂·6H₂O, 0.038-g CuCl₂·2H₂O, 0.05-g AlCl₃, 0.092-g NiCl₂·6H₂O, 0.05-g Na₂MoO₄·2H₂O, 0.5-g ethylenediaminetetraacetate, 1-mL concentrated HCl, and 0.1-g Na₂SeO₃·5H₂O. A mixture of anaerobic medium was prepared in 975 mL of distilled water from stock solutions of mineral buffer solution (10 mL), potassium hydrogen phosphate solution (2 mL), vitamin solution (1 mL), and trace element and selenite solution (1 mL). The mixture of all ingredients was boiled (except cysteine, bicarbonate, and sulfide), and then, it was cooled to room temperature under 80% N₂:20% CO₂ gas mixture to maintain neutral pH. The solution was dispensed under the same gas atmosphere into serum vials and autoclaved. The cysteine and sulfide were sterilized separately by filtration. The medium was reduced using (L⁻¹) 0.5-g cysteine and 0.5-g Na₂S, and then, Na₂CO₃ (2.6 g/L) was added. The final pH of the anaerobic medium was 7.0.

Raw material

Birch (Betula pubescens) wood chips originated from a tree harvested in 2009 in Norway (60.7°North, 10.4°East). The birch tree trunk was debarked and chipped to produce 20–30 mm chip fractions. These fractions were dried at room temperature and subsequently milled to pass a sieve of 6 mm (SM 2000, Retsch, Haan, Germany) and stored at room temperature and dry conditions. The DM and VS contents of the dried birch were 94.9 and 94.8% (fresh biomass weight), respectively. DM is the sum of VS and ash.

Cellulolytic bacteria culture used for bioaugmentation

The strain Caldicellulosiruptor bescii DSM 6725 was revived from the freeze-dried culture that was obtained from the DSMZ (Braunschweig, Germany). It was grown in DSMZ 516 medium with the following modification. The mineral solution contained (L⁻¹): 0.5-g NH₄Cl, 0.5-g KH₂PO₄, 0.33-g KCl, 0.33-g MgCl₂·6H₂O, 0.14-g CaCl₂·2H₂O, 0.5-g yeast extract, 5-g cellobiose, 0.5-mL resazurin (0.05% w/v), 5-mL vitamin solution, and 1-mL trace-element solution. The vitamin solution contained (L⁻¹): 4-mg biotin, 4-mg folic acid, 20-mg pyridoxine–HCl, 10-mg thiamine–HCl, 10-mg riboflavin, 10-mg nicotinic acid, 10-mg calcium pantothenate, 0.2-mg vitamin B12, 10-mg p-aminobenzoic acid, and 10-mg lipoic acid. The trace-element solution contained (L⁻¹): 1.5-g FeCl₂·4H₂O, 0.07-g ZnCl₂, 0.1 MnCl₂·4H₂O, 0.006-g H₃BO₃, 0.19-g CoCl₂·6H₂O, 0.002-g CuCl₂·2H₂O, 0.024-g NiCl₂·6H₂O, and 0.036-g Na₂MoO₄·2H₂O. The mixture of all ingredients was prepared (except cysteine, carbonate, cellobiose, and sulfide) and boiled, and then, it was cooled to room temperature under CO₂ gas. The solution was dispensed under the same gas atmosphere into serum vials and autoclaved. The cellobiose, cysteine, and sulfide were sterilized separately by filtration using a 0.22-μm-pore-size sterile filter (Millipore Filter Corp., Bedford, MA). The medium was reduced using (L⁻¹) 0.5-g cysteine and 0.5-g N₂S, and then, Na₂CO₃ (1 g/L) was added. The final pH was 7.0. The cultures were incubated at 65 °C for 3 days under static conditions. Cell growth was monitored by optical density (680 nm) using a spectrophotometer (Hitachi U-1900, Hitachi High-Technologies Corporation, Tokyo, Japan). After 3 days of incubation (OD680 of approximately 0.54), the C. bescii culture was harvested in the exponential phase and used as a supplementary inoculum (bioaugmentation) to set up the biogas batch experiments described below.

Steam-explosion (SE) pretreatment

SE pretreatment was conducted using a steam-explosion unit designed by Cambi AS (Asker, Norway) situated at Norwegian University of Life Science. In a previous study [6], the optimal steam-explosion conditions of birch for biogas production were found to be pretreatment at 210 °C and 10-min residence time. Therefore, we used the same pretreatment conditions in this study. The pretreated material was stored in plastic bags at 4 °C until the start of the biogas experiment. The DM and VS contents of the steam-exploded birch were 35.0 and 34.9%, respectively.

Batch experiments to test the biogas production potential of steam-exploded birch with and without applied bioaugmentation

Acid-insoluble lignin and carbohydrate analysis

Samples for carbohydrate and acid-insoluble lignin content analysis were prepared using a standard NREL two-stage acid hydrolysis protocol [28]. Acid hydrolysis generates soluble sugars and acid-insoluble lignin residues, where the later was dried overnight at 105 °C in an oven (Heratherm oven, Thermo Scientific, MA, USA) and weighed to obtain the acid-insoluble lignin (Klason lignin) content. The soluble sugars were analyzed for carbohydrate constituents by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) (Dionex ICS-3000, Dionex, Sunnyvale, CA, USA). Separation of soluble sugars was achieved utilizing a CarboPac-PA1 2 × 250 mm analytical column equipped with a CarboPac PA1 2 × 50 mm guard column (both from Dionex, Sunnyvale, CA), operated at 30 °C, with Milli-Q water as a mobile phase with a flow rate of 0.250 mL/min. The total run time was 35 min. External calibration curves were established using the standard solutions of arabinose, galactose, glucose, mannose, and xylose. The standard solutions were prepared from their corresponding monosaccharide (> 99%) obtained from Sigma-Aldrich.

Biogas composition and calculation

The biogas production was periodically monitored by measuring the gas pressure in the headspace of the batch bottles using a digital pressure transducer (GMH 3161, Greisinger Electronic, Regenstauf, Germany). After recording the pressure in the batch bottles, the overpressure was released by penetrating the septum with a needle. To avoid excessive dissolution of CO₂ with possible effects on pH, the overpressure was always kept below 200 kPa (Holliger et al. [27]). The biogas composition (CH₄ and CO₂) was analyzed by gas chromatography (GC) using a gas chromatograph (3000A Micro GC, Agilent Technologies, Wilmington, USA) equipped with a thermal conductivity detector (TCD). For separation of gases, two parallel capillary columns containing different coatings (MolSieve 5 Å PLOT, 10 m × 0.32 mm × 12 μm and PLOT Q, 10 m × 0.32 mm × 10 μm) were used. The operational temperature of sample inlet was kept the same for both columns at 60 °C. The operational injector and column temperatures for the MolSieve 5 Å PLOT were 90 and 70 °C, respectively, while the other column was operated at 50 and 45 °C, respectively. Both columns were connected to TCD with helium applied as a carrier gas. Certified standard mixture of CO₂ and CH₄ in nitrogen (AGA, Norway) was used for calibration. Using the measured overpressure, headspace volume of the bottles and methane concentration as input, the ideal gas law was applied for calculating the volume of methane produced. The volume of the methane produced for substrate-amended bottles with and without supplied bioaugmentation was reported after correcting the background methane production from the negative controls (inoculum with applied bioaugmentation and inoculum alone, respectively). The volume of methane produced was reported at standard temperature and pressure (0 °C and 1 atm). The average results of the biological triplicates are presented with standard deviations.

Other analytical methods

The DM and VS of the substrates were analyzed according to the standard methods (APHA, 2005).

Microbial community analysis

At the completion of the study, the samples collected from all batch bottles were analyzed for their microbial community composition. The genomic DNA was extracted from the samples stored at − 20 °C. 1 mL of the sample was centrifuged at 10,000×g for 5 min to remove the supernatant. Afterwards, the pellet was resuspended in 300 µL of S.T.A.R. buffer (Roche Diagnostics, Penzberg, Germany) to stabilize the nucleic acids in the sample. Cells were mechanically disrupted in a MagNa Lyser instrument (Roche Diagnostics GmbH, Mannheim, Germany) by adding 0.25 g of acid-washed glass beads and bead-beating twice at 6500 rpm and room temperature for 20 s each time. Thereafter, the sample was centrifuged at 13,000×g for 5 min to recover the DNA from the supernatant. The DNA was extracted using the MagMidi kit (LGC Genomics, UK) for the KingFisher Flex robot (ThermoFisher Scientific, Wilmington, USA) according to the manufacturer’s protocol, and its concentration was measured by Qubit fluorometer with Quant-iT dsDNA Br assay kit (Invitrogen, USA). The DNA quality was evaluated with the Nanodrop ND 1000 spectrophotometer (ThermoFisher Scientific, Wilmington, USA).

The extracted DNA was amplified using polymerase chain reaction (PCR) primers Pro341F/Pro805R: 5′-CCTACGGGNBGCASCAG-3′/5′-GACTACNVGGGTATCTAATCC-3′ [29], which target the V3–V4 hypervariable regions of bacterial and archaeal 16S rRNA gene. The PCR mixture (25 µL) contained 2.5 µL of DNA template (5 ng/µL), 12.5 µL of iProof HF Master Mix (BIO-RAD, USA), 0.625 µL of each primer (10 µM), and 8.75 µL nuclease free water. The PCR cycles were: an initial denaturation step at 98 °C for 3 min, followed by 30 cycles consisting of 98 °C for 10 s, 55 °C for 30 s, and 72 °C for 30 s, with a final elongation step at 72 °C for 5 min. Agencour tAMPure XP (Beckman Coulter, USA) was used for purification of the PCR products. The size and purity of amplicons were checked by electrophoresis on 1% w/v agarose gel. Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA) was used to index the PCR-amplified samples, according to manufacturer’s protocol. The barcoded amplicons were quantified using a Qubit fluorometer with Quant-iT dsDNA BR assay kit (Invitrogen, USA), and each amplicon was adjusted to equimolar concentration according to the Illumina protocol for 16S Metagenomic Sequencing Library Preparation. Finally, Illumina MiSeq sequencer (Illumina, San Diego, CA, USA) was used to sequence the denatured DNA using MiSeq reagent kit v3 (600-cycle).

Sequence data from the samples were analyzed with Quantitative Insight Into Microbial Ecology (QIIME) 1.9.1 software package [30]. Previously, the downstream analysis, paired-end reads from every sample were merged using PEAR program [31], followed by quality filtering using PRINSEQ [32] at mean quality score of 30 and a minimum length of 350 bp. The primers sequences were trimmed by Mothur [33]. Chimeric sequences were removed, followed by clustering into operational taxonomic sequences (OTUs) at 97% sequence identity by USEARCH [34, 35], which is implemented in QIIME, using the Greengenes database gg_13_8 [36]. Raw sequences are made available at the Sequence Read Archive (SRA) with the accession numbers SRR5921530–SRR5921558 as part of BioProject PRJNA394663.

Data analysis

Microsoft PowerPoint 2010 (Microsoft, WA 01060, USA) and Origin 8.0 (Origin Lab, WA 01060, USA) were used for graphing.

Feedstock characteristics

The DM content decreased from 94.8% in untreated birch to 35.0% in steam-exploded birch, as steam is added to the biomass during the pretreatment. Since the pH of steam-exploded birch was very low (3.0), it was adjusted to 7.5 by adding NaOH prior to the biogas experiments. The low pH can be explained by the release of organic acids from the degradation of hemicelluloses during SE pretreatment [38]. The VS content of the untreated and pretreated birch was similar (99.8% of DM).

Enhanced biogas production by combined SE and bioaugmentation

The effect of SE pretreatment on methane production was investigated using batch experiments run with untreated and steam-exploded birch for 50 days (Fig. 1). The data presented in Fig. 1 are the net methane production from the birch substrate after correcting for the background methane production from the inoculum or inoculum with C. bescii. SE pretreatment clearly influenced the methane production rate and yield, and the lag phase (Table 3). Following SE pretreatment, the final methane yield was increased from 81- to 179-mL CH₄/g VS (increased by 118%). While the initial rate of methane production seems similar for both pretreated and untreated materials, the rate substantially increased in bottles fed with pretreated material after day 13. The maximum rate of methane production also increased from 4.2-mL CH₄/g VS/day for bottles fed with untreated birch to 9.9-mL CH₄/g VS/day for steam-exploded birch (Table 3). Although the lag phase was relatively longer for bottles fed with pretreated birch, the rate and yield of methane production had significantly increased following SE. Thus, SE had improved the accessibility of the birch material to anaerobic bacteria. SE has been shown to result in hemicelluloses removal, lignin relocalization with some structural modification together with broken fiber and porous surface materials [8]. Such transformations of biomass could contribute to reduction of biomass recalcitrance and explain the higher methane yield obtained from the stream-exploded birch in this study.

Bottles	Measured CH4 yield, mL CH4/g VS	Calculated values from the model
		CH4 yield, mL CH4/g VS	max CH4 rate, mL CH4/g VS/day	Lag phase, day	R 2a	RMSEa
Untreated + 0%	81 ± 7	79 ± 6	4.2 ± 1	2.20 ± 0.66	0.9892	3.52
Pretreated + 0%	179 ± 7	177 ± 6	9.9 ± 2	8.00 ± 0.32	0.9899	8.11
Pretreated + 2%	197 ± 9	197 ± 8	15.5 ± 3	8.19 ± 1.14	0.9896	9.52
Pretreated + 5%	196 ± 5	196 ± 5	13.3 ± 4	7.25 ± 2.05	0.9902	8.89
Pretreated + 10%	189 ± 2	189 ± 2	13.2 ± 2	6.47 ± 1.22	0.9896	8.87
Pretreated + 15%	188 ± 8	188 ± 7	13.9 ± 2	7.26 ± 0.72	0.9878	9.66

^aR² and RMSE were calculated from the average values of the measured methane production during 50 days and the one calculated using the model

The effect of combined SE and bioaugmentation with C. bescii on methane production was also investigated (Fig. 1). The methane yield had increased significantly (by 130–140%) by combined SE pretreatment and bioaugmentation in comparison with bottles fed with untreated substrate (Table 3). The maximum methane production rate was increased from 4.2-mL CH₄/g VS/day for bottles fed with untreated birch to 13.2–15.5-mL CH₄/g VS/day for all bottles running with steam-exploded material and bioaugmented with C. bescii.

The bottles with and without bioaugmentation (0–15% v/v loading of C. bescii) were also compared to evaluate the effect of bioaugmentation alone on methane production (Fig. 1). The methane yield was similar during the first 3 days with and without bioaugmentation but higher in the former afterwards. The enhancement in methane yield by bioaugmentation alone reached between 38 and 48% at the timepoint, where more than 60% of methane was produced (day 18). This enhancement was reduced later and were 5–10% on day 50. The highest methane improvement on day 50 was 10% in bottles bioaugmented with lower dosages of C. bescii (2 and 5%). The maximum methane production rates were slightly improved and the lag phase periods were slightly shorter in bioaugmented bottles compared to the non-bioaugmented bottles. The observed positive effects of bioaugmentation may indicate an enhanced hydrolysis of steam-exploded materials by C. bescii and subsequent improvement in acidogenesis, acetogenesis, and finally methanogenesis. C. bescii was isolated from thermal springs of Kamchatka in Russia [40] and has been proven to be capable of hydrolyzing a variety of polysaccharides, including crystalline cellulose and untreated plant biomass [23, 24].

The extent of solubilization of the untreated birch and pretreated birch with and without the applied bioaugmentation is presented in Fig. 2. It clearly shows that the solubilization was increased by more than twofold when birch was pretreated. The solubilization was also increased when combined SE pretreatment and bioaugmentation was employed in comparison with pretreatment alone. The increase in solubilization is consistent with the observed improvement in methane production (Fig. 1 and Table 2).

Biogas intensification by bioaugmentation with cellulolytic bacteria for improving hydrolysis has been reported in several biogas studies from untreated lignocellulosic substrates [15, 16] and only few studies from pretreated materials [41, 42]. In a batch experiment fed with untreated straw, bioaugmentation increased the methane yield by 27% in a 4% amended enrichment culture obtained from sheep rumen fluid containing cellulolytic bacteria [15]. In another bioaugmentation experiment with Clostridium cellulolyticum, long adaptation period (more than 10 days) and high amount of culture (25%) were required to enhance the methane yield by 7.6–13% [16]. In another recent study [43], individual and mixed cultures of Ruminococcus flavefaciens 007C, Pseudobutyrivibrio xylanivorans Mz5T, Fibrobacter succinogenes S85, and Clostridium cellulovorans were used to enhance methane production of brewer spent grain. The highest methane improvement reached up to 18% when P. xylanivorans Mz5T was used for bioaugmentation, whereas the other hydrolytic bacteria increased the methane yield by only 5–7%. Despite the positive effects of bioaugmentation on methane production from untreated lignocellulosic substrates, the methane yield in bioaugmented bottles is rather low in comparison with the theoretical methane expected from the carbohydrate composition of the substrates. Thus, combination of pretreatments and bioaugmentation strategies seems more effective for the conversion of the carbohydrates in the biomass into methane, as demonstrated in this study (up to 140% improvement in methane yield). It should be noted that SE pretreatment contributed to the major share of methane enhancement by 118%, while bioaugmentation notably increased the initial methane production rate by up to 44%.

Other few biogas studies also demonstrated the beneficial effects of combined pretreatment and bioaugmentation for biogas production [41, 42]. Combined thermal pretreatment (150 °C for 2 h) and bioaugmentation with an enriched culture containing lignocellulolytic microorganisms (Clostridium stercorarium and Bacteroides cellulosolvens) can substantially increase the methane yield of sludge by up to 246% compared to the control (untreated sludge), whereby the thermal pretreatment contributed the major share of methane enhancement by 223% [42]. In another study by Sträuber et al. [41], combined calcium hydroxide pretreatment (10% w/w of Ca(OH)₂ per fresh weight of straw; stored at for 24 h at 22 °C) and two alkali-tolerant, lignocellulolytic environmental enrichment cultures, was employed to improve biogas production from wheat straw. The methane potential of the pretreated straw with and without bioaugmentation was 36% higher than that of untreated straw. Bioaugmentation only accelerated the methane production rate during the first week without enhancing the final methane yield. Overall, this and previously published studies supported the beneficial effects of combined pretreatment and bioaugmentation with cellulolytic bacteria for reducing biomass recalcitrance and improving hydrolysis, which contributed to ultimately improved methane production.

Microbial community analysis

The bacterial and archaeal communities in the samples from thermophilic bottles were analyzed by amplicon sequencing of 16S rRNA genes. As shown in Fig. 3, a total of 66 genera were identified which harbored ≥ 0.1% of the reads in one or more of the sequences. Of these, 15 genera were found to be highly abundant comprising at least 1% in at least one reactor (Fig. 4). The relative abundance of the OTUs comprising at least 1% in at least one reactor at the phylum level is shown in Fig. 5. The dominant bacterial phyla in all thermophilic bottles fed with pretreated substrate was Firmicutes (50–70% of total reads), whereas Firmicutes (23–40%) and Dictyoglomi (20–34%) dominated the bottles fed with untreated birch and without substrate (inoculum) (Fig. 5). Members of Firmicutes have been repeatedly identified as the main phyla in various anaerobic digesters along with Bacteroidetes [44–52]. Members of the phylum Bacteroidetes were detected in only the bottles fed with untreated birch at lower abundance (1.7%). The dominance of Firmicutes and absence of Bacteroidetes in all the thermophilic bottles except those fed with untreated birch suggest the importance of the former for the degradation of lignocellulosic biomass under thermophilic conditions. In a previous study of biogas digesters treating manure and straw, the relative abundance of Bacteroidetes decreased significantly from 13.2 to 16.6 to 0.4% with increase in operating temperature from 44 to 52 °C [53], whereas the relative abundance of Firmicutes increased during the increase in operating temperatures [53]. This suggests a competitive advantage of Firmicutes over Bacteroidetes under thermophilic conditions. The dominance of Dictyoglomi in all other samples except those from bottles running with pretreated material is less clear. Members of Dictyoglomi are known to have a cellulolytic activity [54] and have been detected previously in thermophilic biogas digesters fed with food waste (Hagen et al. [55]).

There are also other members of bacteria present in some of the biogas bottles with minor abundance (Fig. 5). For example, phylum “Atribacteria” OP9 lineage was present in inoculum and those bottles fed with untreated birch at low abundance (about 2%) compared to Dictyoglomi and Firmicutes. Members of OP9 have been observed in few thermophilic biogas digesters [55, 56] and suggested to hydrolyze complex carbohydrates such as cellulose and hemicellulose [57]. Proteobacteria was detected at about 0.1% in inoculum and bottles fed with untreated substrate, with all the members belonging to the known glucose utilizing Alphaproteobacteria.

Archaeal sequences detected in all the thermophilic bottles belonged to only the hydrogenotrophic Methanobacteriaceae, accounting for 30–48% of the total sequence reads (Fig. 5). Interestingly, sequences affiliated to acetoclastic methanogens were below the detectable level in any of the bottles, suggesting that acetoclastic methanogens are less important for methane production in anaerobic digesters above 60 °C [12]. A previous study showed that the hydrogenotrophic Methanothermobacter thrived when the operating temperature of biogas digesters was increased from 55 to 65 °C, whereas the mixotrophic Methanosarcina were no longer detectable at 65 °C [12]. In the absence of acetoclastic methanogens, methane production from acetate follows a two step pathway, i.e., acetate oxidation to CO₂ by the syntrophic acetate oxidation (SAO) pathway followed by the reduction of CO₂ into methane by hydrogenotrophic methanogenesis (HM). In the previous biogas digesters operated at temperatures above 60 °C [12], SAO-HM played key role for the conversion of acetate into methane. This study also supported the dominance of SAO-HM for methane production at high temperature (see the discussion below).

The abundance of the bacterial and archaeal community hardly changed in all the negative controls without substrate (inoculum with or without C. bescii) (Figs. 3, 4). This is not surprising as the remaining substrate in the inoculum is very recalcitrance to degradation by indigenous microbes and C. bescii, also confirmed by the very low methane production in these bottles (data not shown). The microbial changes were also minor when the bottles were fed with untreated substrate. As shown in Fig. 4, the relative abundance of Dictyoglomus was increased from 22% in bottles without substrate (inoculum + 0%) to 34% in bottles fed with untreated birch (untreated + 0%). This genus may be responsible for the degradation of part of the carbohydrate fraction of untreated birch. The relative abundance of Methanothermobacter and unclassified sequences of the phylum Methanobacteriaceae hardly changed when untreated birch was used as substrate compared to inoculum alone.

The bacterial communities were affected substantially when steam-exploded birch was used for biogas production (Figs. 3, 4). The relative abundance of Firmicutes increased from 40% in the inoculum + 0% bottles to 62% in the bottles fed with steam-exploded birch without bioaugmentation (pretreated + 0%) (Fig. 5). Whereas the phylum Dictyoglomi decreased significantly from 22% in the inoculum + 0% to 3% in the pretreated + 0% bottles. These results indicate that members of Dictyoglomi were outcompeted by the Firmicutes and the latter likely was responsible for the hydrolysis of steam-exploded birch. Members of the phylum Firmicutes were entirely represented by the class Clostridia, which previously has been associated with hydrolysis, acidogenesis, and acetogenesis steps [58]. The diversity within Clostridia was high, comprising several genera within the dominant orders Thermoanaerobacterales, Clostridiales, Natranaerobiales, and the less abundant orders such as SHA-98 and OPB54 (Fig. 4). Sequences belonging to the uncultured order OPB54 accounted for 3.7% of the total reads in the pretreated + 0% bottles, but below 1% in the inoculum + 0% bottles. OPB54 was previously identified in low abundance in thermophilic laboratory-scale digesters treating stillage [59] and in an enrichment culture amended with lignocellulosic biomass [60]. Culture-independent approaches using DNA-based stable isotopes probing (SIP) revealed that OPB54 was the most abundant putative SAO bacterium in biogas digesters fed with high levels of acetate [61]. The relative abundance of uncultured-order SHA-98 was lower in the pretreated + 0% bottles compared to the inoculum + 0% bottles, but its function in the methanogenic environment is still unknown. Within the class Clostridia, the genus Ethanoligenens, unclassified members of the family ML1228J-1, and Ruminococcaceae were detectable in all bottles except those fed with steam-exploded birch.

The relatively higher abundance of members of Clostridiales in the bottles fed with steam-exploded birch suggests their role in the hydrolysis of the carbohydrate fraction of steam-exploded birch. Members of Clostridiales are ubiquitous in biogas digesters operating with mono- and co-digestion of lignocellulosic materials [51, 52, 62] and have been reported as the main cellulose degrader in biogas digesters [63]. Within the order Clostridiales, the abundance of the genus Clostridium increased from 3% in the inoculum + 0% to 10% in the pretreated + 0% bottles, whereas the genus Caldicoprobacter remained stable (around 3.6%) (Fig. 4). Caldicoprobacter contains several xylanolytic bacteria capable of fermenting various sugars into acetate, lactate, ethanol, H_2, and CO₂ [64]. Some members of the Clostridium are known syntrophic acetate oxidizing bacteria (SAOB), like the mesophile C. ultunense [65]. Other members of the Clostridium contain the formyltetrahydrofolate synthetase (FTHFS)-encoding gene for formyltetrahydrofolate synthetase, which is a key enzyme involved in reductive acetogenesis and also the reverse reaction (i.e., SAO), suggesting their SAO capability [66]. In the absence of detectable amount of acetoclastic methanogens in the bottles fed with steam-exploded birch, members of Clostridium and other SAOB coupled with hydrogenotrophic methanogens likely contributed to acetate conversion into methane.

The relative abundance of the three genera of the order Thermoanaerobacterales, namely, Coprothermobacter, Thermacetogenium, and Thermovenabulum, was increased in the bottles treating steam-exploded birch (Fig. 4). Coprothermobacter was the most dominant (41%) genus among all the sequence reads in the pretreated + 0% bottles. Its relative abundance increased by almost 1.6 folds in pretreated + 0% bottles compared to inoculum + 0% bottles. In a previous study of high-rate (HRT of 2–4 days) and high-temperature (60 and 65 °C) anaerobic digesters, Coprothermobacter spp. was identified as a main acetate degrader in syntrophic association with hydrogenotrophic Methanothermobacter using stable carbon isotopic analysis combined with pyrosequencing methods [12]. Thermacetogenium was below 1% in the inoculum + 0% and 1.38% in the pretreated + 0% bottles. Thermacetogenium phaeum is a known SAOB [67], which was originally isolated from a thermophilic anaerobic reactor treating kraft-pulp wastewater [68] and found later in many thermophilic biogas digesters [12, 44, 69, 70]. Overall, the higher abundance of bacteria capable of SAO (Clostridium, Coprothermobacter, Thermacetogenium, and uncultured order OPB54) as well as the hydrogenotrophic Methanothermobacter (see the discussion below) in these thermophilic bottles fed with steam-exploded birch suggests the importance of SAO-HM for efficient degradation of steam-exploded birch into ultimately methane.

As discussed above, the archaeal sequences detected in all thermophilic bottles belonged to only the hydrogenotrophic Methanobacteriaceae (Fig. 5), but there were differences in the relative abundance of the sequences at genus level (Fig. 4). Methanothermobacter was the dominant genus in the bottles fed with steam-exploded birch with and without bioaugmentation, whereas most of the reads (30–37%) in the rest of bottles were not affiliated to a known genus of the Methanobacteriaceae. These changes in archaeal community at genus level are less clear, but the availability of the hydrogenotrophic substrates (H₂ and CO₂) due to the improved degradation of pretreated materials (Fig. 1) could give a competitive advantage for the Methanothermobacter to dominate in bottles fed with steam-exploded birch. Members of Methanothermobacter have been reported as a dominant methanogens in thermophilic biogas digesters (55–65 °C) [12, 44, 55].

Although C. bescii was not detectable in the bioaugmented bottles at the final day of the experiment, the microbial community structures were clearly affected by the bioaugmentation (Fig. 4). The relative abundance of the microbial communities of the bottles treating pretreated materials with (pretreated + 10 and + 15%) and without bioaugmentation (pretreated + 0%) was relatively similar, but differs from those bottles bioaugmented with minimum dosages of C. bescii (pretreated + 2 and + 5%). For instance, the relative abundance of Methanothermobacter was higher in the pretreated + 2% and pretreated + 5% bottles, which reached up to 33–38%, compared to the other bottles at lower abundance (13–27%). The abundance of the known hydrolytic bacterium Caldicoprobacter was also higher by twofold in these bioaugmented bottles (pretreated + 2 and + 5%) compared to the others (pretreated + 10 and + 15%). The increase in the abundance of the hydrogenotrophic Methanothermobacter and the hydrolytic bacterium Caldicoprobacter was well correlated with higher methane production (Table 3 and Fig. 1), suggesting that minimum amount of C. bescii was needed to influence the microbiota for efficient degradation of steam-exploded materials into ultimately methane. The reason why the relative abundance of the microbial communities of the bottles fed with pretreated birch with applied higher dosage of bioaugmentation (10 and 15%) and without bioaugmentation was similar and differs from lower bioaugmentation dosage (2 and 5%) is unclear.

There were also changes in bacterial community capable of SAO in bottles fed with pretreated substrate with and without bioaugmentation (Fig. 4). For instance, the abundance of Coprothermobacter was significantly reduced from 36 to 41% in the pretreated + 0%, pretreated + 10% and pretreated + 15% bottles to 10–22% in the pretreated + 2% and pretreated + 5% bottles. Clostridium was higher in the pretreated + 5% and pretreated + 10% bottles (around 20%) and lower (10–13%) in the rest of the bottles. The decrease and increase of some members of SAOB community without reduction in their combined abundance in these bottles may suggest the functional resilience of the microbial community to temporal changes in the production rate and levels of VFAs. Although we did not measure the concentration of VFAs in each reactor, their concentration and production rate could differ among some of the bottles fed with pretreated material, irrespective of bioaugmentation, as their methane production rate and yield varied (Fig. 1 and Table 3).

Overall, the combined SE pretreatment and bioaugmentation with minimum amount of C. bescii contributed to significant increase in methane production (by 140%) and changes in the microbial communities. Interestingly, the bioaugmenting culture (C. bescii) was not detectable at the end of the experiment. According to the observed higher methane production rate in our bioaugmented bottles after 9 days (Fig. 1), we speculated that C. bescii remained active over 9 days before it was finally outcompeted by the indigenous microbes for nutrients or due to environmental stress (e.g., different pH, VFAs) [14]. Similar results have been reported in a previous bioaugmentation study, where the bioaugmenting culture (P. xylanivorans Mz5T) was not detectable at the end of the experiment, while the methane production was improved by 18% and microbial community was affected during bioaugmentation [43].

The enhanced methane production rate and yield by adding only small amounts of the C. bescii culture is particularly suitable for upscaling the bioaugmentation process as the cost of cultivating C. bescii culture then is limited. Of course, the implication of the present findings for lab-scale semi-continuously fed digesters such as CSTRs needs to be investigated further before scaling up to industrial biogas plants. The challenges of microbial washout particularly in CSTRs and out-competition by indigenous microorganisms during bioaugmentation process should be addressed in future studies to favour survival and prolonged activity of the exogenous (bioaugmenting) microbes. A recent study by Kovacs et al. used a hydrolytic bacterium from the same genus as our study (i.e., Caldicellulosiruptor saccharolyticus) to enhance biogas production in CSTRs’ running with a mixture of pig slurry (25% w/v) and chopped sweet sorghum (75% w/v) [21]. Under similar organic loading rate (OLR, 4 g total organic solids/L/day), the addition of C. saccharolyticus led to enhanced biogas production, similar to the results obtained in our batch experiments. However, the concentration of the bioaugmenting bacterium was gradually reduced and finally disappeared after 2–3 weeks. Bioaugmentation at the higher OLR of 8-g total organic solids/L/day gave the same beneficial effect of enhanced biogas production as at the lower OLR without dilution of C. saccharolyticus over the monitored 24 days. The study demonstrates that changing operating conditions like OLR can be employed as a simple and economical strategy to favour the survival and prolonged activity of the bioaugmenting microbes without the need to carry out the bioaugmentation again and again.