Microbial community redundancy in anaerobic digestion drives process recovery after salinity exposure Academic research paper on "Biological sciences"

Accepted Manuscript

WATER RESEARCH

Microbial community redundancy in anaerobic digestion drives process recovery after salinity exposure

Jo De Vrieze, Marlies E.R. Christiaens, Diego Walraedt, Arno Devooght, Umer Zeeshan Ijaz, Nico Boon

PII: S0043-1354(16)30986-1

DOI: 10.1016/j.watres.2016.12.042

Reference: WR 12592

To appear in: Water Research

Received Date: 31 October 2016

Revised Date: 23 December 2016 Accepted Date: 24 December 2016

Please cite this article as: De Vrieze, J., Christiaens, M.E.R., Walraedt, D., Devooght, A., Ijaz, U.Z., Boon, N., Microbial community redundancy in anaerobic digestion drives process recovery after salinity exposure, Water Research (2017), doi: 10.1016/j.watres.2016.12.042.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bacteria Archaea

NMDS1 NMDS1

MANUSCRIPT Water Research Version 2

Title: Microbial community redundancy in anaerobic digestion drives process recovery after salinity exposure

Running title: Microbial redundancy in anaerobic digestion

Jo De Vrieze1, Marlies E.R. Christiaens1, Diego Walraedt1, Arno Devooght1, Umer Zeeshan Ijaz2, Nico Boon1®

1Center for Microbial Ecology and Technology (CMET), Ghent University, Coupure Links 653, B-9000 Gent, Belgium

Infrastructure and Environment Research Division, School of Engineering, University of Glasgow, UK

H Correspondence to: Nico Boon, Ghent University; Faculty of Bioscience Engineering; Center for Microbial Ecology and Technology (CMET); Coupure Links 653; B-9000 Gent, Belgium; phone: +32 (0)9 264 59 76; fax: +32 (0)9 264 62 48; E-mail: Nico.Boon@UGent.be; Webpage: www.cmet.ugent.be.

26 Abstract

27 Anaerobic digestion of high-salinity wastewaters often results in process inhibition due to the

28 susceptibility of the methanogenic archaea. The ability of the microbial community to deal

29 with increased salinity levels is of high importance to ensure process perseverance or recovery

30 after failure. The exact strategy of the microbial community to ensure process endurance is,

31 however, often unknown. In this study, we investigated how the microbial community is able

32 to recover process performance following a disturbance through the application of high-

33 salinity molasses wastewater. After a stable start-up, methane production quickly decreased

34 from 625 ± 17 to 232 ± 35 mL CH4 L-1 d-1 with a simultaneous accumulation in volatile fatty

35 acids up to 20.5 ± 1.4 g COD L-1, indicating severe process disturbance. A shift in feedstock

36 from molasses wastewater to waste activated sludge resulted in complete process recovery.

37 However, the bacterial and archaeal communities did not return to their original composition

38 as before the disturbance, despite similar process conditions. Microbial community diversity

39 was recovered to similar levels as before disturbance, which indicates that the metabolic

40 potential of the community was maintained. A mild increase in ammonia concentration after

41 process recovery did not influence methane production, indicating a well-balanced microbial

42 community. Hence, given the change in community composition following recovery after

43 salinity disturbance, it can be assumed that microbial community redundancy was the major

44 strategy to ensure the continuation of methane production, without loss of functionality or

45 metabolic flexibility.

47 Keywords: 16S rRNA gene, biogas, Illumina sequencing, methanogenesis, microbiome, salt

1. Introduction

Anaerobic digestion (AD) of nitrogen- and or salt-rich feedstocks, such as animal manure (Usack and Angenent 2015), slaughterhouse waste (Franke-Whittle and Insam 2013, Pitk et al. 2013) and aquaculture sludge (Zhang et al. 2016) leads to high total ammonia nitrogen (TAN) and salt, mainly monovalent cations (Na+ and K+), concentrations in the mixed liquor of the digester. This may encounter process stability issues, e.g. volatile fatty acid (VFA) accumulation and/or a variable biogas production and composition that have to be avoided to guarantee continuous high biogas production rates. The concentrations of total ammonium and salt that cause process disturbance in AD have been determined in numerous studies, but this led to a very broad concentration range that provokes 50% inhibition of methane production, i.e. 1.7-14 g N L"1 for TAN (Chen et al. 2008) and 4.4-17.7 g L"1 for Na+ (Feijoo et al. 1995). Several factors also influence the degree of toxicity of both salt and TAN, including pH, temperature, presence of other cations, and organic loading rate (Chen et al. 2008, Fang et al. 2011, Garcia and Angenent 2009, Moestedt et al. 2016, Rajagopal et al. 2013). Especially for TAN, which can be present as ammonium ion (NH|) and the free ammonia (NH3) form, both an increase in pH and temperature engage a shift to free ammonia. Free ammonia is, in general, more toxic than the ammonium ion, due to its ability to freely migrate through the microbial cell membrane, thus, influencing intracellular pH and proton transport across the membrane (Gallert et al. 1998).

The accumulation of VFA and instable methane production as a result of an increase in TAN and/or salt concentration is the consequence of the overall high susceptibility of the methanogenic (archaeal) community to high salt and total/free ammonia concentrations, compared with the bacterial community (De Vrieze et al. 2012). This is related with often observed shifts from acetoclastic to hydrogenotrophic methanogenesis with increased total/free ammonia and salt concentrations (Niu et al. 2013, Schnurer and Nordberg 2008,

Werner et al. 2014). Hence, an increased level of microbial community dynamics could be observed in response to process disturbances that are accompanied with (sudden) changes in operational parameters (Goux et al. 2015, Li et al. 2015b, Niu et al. 2015a, Regueiro et al. 2014a).

To determine to which extent microbial community dynamics influence functional stability, several aspects need to be taken into account. As the microbial community shows a constant degree of dynamics, even at a constant feeding regime and operational conditions (De Vrieze et al. 2013, Fernandez et al. 1999), it is not always straightforward to differentiate between natural fluctuations and actual shifts with respect to process disturbances. If the shift in the microbial community is confirmed as beyond natural fluctuations, its influence on process performance needs to be estimated. For example, a shift from acetoclastic methanogenesis to syntrophic acetate oxidation coupled with hydrogenotrophic methanogenesis often maintains process stability (Werner et al. 2014, Westerholm et al. 2011). This shift, however, often coincides with a decrease in acetate oxidation and even methane production (Schnurer et al. 1999). It is, therefore, important to estimate to which extent community dynamics can prevent process failure. A shift in the microbial community in response to a specific disturbance could coincide with a change in (species) diversity (Briones and Raskin 2003, Curtis and Sloan 2004). Microbial community diversity and functional stability often do not show a positive correlation (Baho et al. 2012, Dearman et al. 2006, McCann 2000), which emphasizes the equal importance of functional resilience and/or redundancy to ensure process stability in AD (Niu et al. 2015a, Werner et al. 2011). The microbial community can either maintain a stable composition (resistance), temporarily change in composition (resilience) or shift to a new composition (redundancy) in response to a disturbance (Allison and Martiny 2008). After the disturbance has disappeared, the return to a microbial community with similar functional stability as the original state has to be confirmed to ensure resistance to future perturbations.

99 In this research, molasses wastewater with a high salinity and nitrogen content was used as

100 feedstock to provoke a disturbance in AD after which the feedstock was switched again to

101 waste activated sludge with low salinity and nitrogen content. This was carried out to test the

102 ability of the microbial community to return to its original state and performance. After this,

103 an alternative disturbance was applied through the addition of urea to evaluate the functional

104 resilience or redundancy of the microbial community following the previous disturbance. The

105 archaeal and bacterial community were evaluated in terms of functional performance

106 (methane production) and ecology with respect to the different disturbances.

2. Material and methods

2.1. Inoculum and feedstocks

The anaerobic sludge inoculum was obtained from a full-scale digester treating brewery wastewater (Van Steenberge, Ertvelde, Belgium) (Table 1). The molasses wastewater was acquired from the company AVEBE (Veendam, The Netherlands), and the waste activated sludge (WAS) was collected from the municipal wastewater treatment plant Dendermonde (Dendermonde, Belgium) (Table 2). Both the molasses wastewater and WAS were stored at 4 °C until use.

2.2. Experimental set-up

Glass Schott bottles with a liquid volume of 800 mL and a total volume of 1 L were used as digesters. A rubber stop was used to ensure anaerobic conditions, and a plastic tube was installed as gas outlet through the rubber stop. The biogas was collected by means of water displacement in inverted plastic tubes (Figure S1). To avoid CO2 in the biogas from dissolving in the water, the pH was maintained below a value of 4.3. The inoculum was diluted with tap water until a volatile suspended solids concentration of 10 g L-1 was obtained. The digesters were operated in a continuous stirred tank reactor mode, and incubated in a temperature controlled room at 34 ± 1 °C. Manual feeding took place three times a week.

2.3. Digester operation: start-up and experimental phases.

In the start-up phase (day 0-34), the nine digesters were operated in the exact same way (Table 3 & S1). Initially (day 0-20), only WAS sludge was used as feedstock, and the hydraulic retention time (HRT) was decreased from 40 to 20 days, while from day 21-34 molasses wastewater was slowly added at an HRT of 20 days to avoid overloading of the digester, and to guarantee a smooth transition between the two feedstocks.

The experimental phase consisted of 3 main phases (Table 3). In Phase 1 (day 35-83), molasses was used as single feed in each of the nine digesters. A differentiation between the three digesters, each of which were run in triplicate, was initiated in Phase 2 (day 84-167). To Digester A, molasses was continued to be added as single feedstock, whilst for the other two digesters (Digester B and C), the feed was switched back to the same WAS that was used during the start-up phase. Glycerol was added as a co-substrate to Digester B and C to obtain a similar organic loading rate in all digesters, because the COD content of the WAS was too low to ensure an organic loading rate of 2.0 g COD L-1 d-1, as was the case in Digester A. Glycerol was selected as co-substrate, because crude glycerol is often co-digested with other feedstocks to increase methane production (Fountoulakis et al. 2010). In Phase 3 (day 168245), molasses was still the only feed for Digester A. In both Digester B and C, WAS and glycerol also still served as substrate, but to Digester C, first NH4Cl and then urea was added (Figure S2) to slowly increase the free ammonia concentration to obtain a final additional TAN concentration of 2.0 g N L-1. The organic loading rate and HRT were maintained at 2.0 g COD L-1 d-1 and 20 days, respectively, during the entire experimental phase in each of the digesters.

The volumetric biogas production and composition were measured three times per week, and reported at standard temperature (273 K) and pressure (101,325 Pa) conditions (STP). The pH in the mixed liquor of each digester was also measured three times per week. Samples for total VFA, TAN and conductivity measuring were taken once per week. Conductivity can be considered an overall estimation of salinity in the mixed liquor. Samples were collected from the inoculum and the mixed liquor at the end of each phase for microbial community analysis, and stored at -20°C until analysis.

2.4. Microbial community analysis

The DNA extraction was carried out as described by Vilchez-Vargas et al. (2013). The quality of the DNA extracts was validated by means of agarose gel electrophoresis and via PCR, using the protocol of Boon et al. (2002) with the bacterial primers P338 and P518r (Muyzer et al. 1993). The quality of the PCR product was determined with agarose gel electrophoresis to ensure that no inhibition of the PCR took place. The DNA extracts were sent to LGC Genomics GmbH (Berlin, Germany) for Illumina sequencing on the Miseq platform. The amplicon sequencing and data processing were carried out as described in SI (S4).

2.5. Statistical analysis

Separate tables containing the abundances of the OTUs (operational taxonomic units) and the taxonomic assignments were generated for the bacteria and archaea (Supplementary file 2). The R Studio version 3.2.3. software (http://www.r-project.org) (R Development Core Team 2013) was used for statistical analysis. The packages phyloseq (McMurdie and Holmes 2013) and vegan (Oksanen et al. 2016) were used for microbial community analysis. The community composition of the biological replicates was statistically compared by means of analysis of variance (ANOVA, aov function) to validate that both the bacterial and archaeal community did not significantly differ between biological replicates. Heatmaps were generated using the weighted average values of the biological replicates by means of the pheatmap package. Differences in order-based Hill's numbers (Hill 1973) between the digesters were defined via ANOVA and the post-hoc Dunn's Test of Multiple Comparisons (dunn.test package) with Benjamini-Hochberg correction. Separate non-metric distance scaling (NMDS) plots of the bacterial and archaeal community data were created based on the (un)weighted Unifrac and Bray-Curtis distance measures. Significant differences in community composition between digesters and/or between different phases were identified by means of pair-wise Permutational ANOVA (PERMANOVA) with Bonferroni correction,

using the adonis function (vegan). The identification of a subset of OTUs that best describe the overall community pattern was carried out via a stepwise exploration according to description of the forward selection/backward elimination algorithm (Clarke and Warwick 1998). This was done in two steps. First, the algorithm was run for 50 times, randomly selecting 60% of the total OTUs for inclusion in the stepwise exploration. In the second step, samples similarities were identified using a much reduced number of subset of OTUs by maximising correlation (pearson correlation) of the Bray-Curtis distances measure between samples considering the subset of OTUs and between the samples considering all OTUs.

2.6. Chemical analyses

Total solids (TS), VS, total suspended solids (TSS), volatile suspended solids (VSS), Kjeldahl nitrogen (TKN) and TAN were determined by means of standard methods (Greenberg et al. 1992). The free ammonia concentration was calculated based on the TAN concentration, pH and temperature (Anthonisen et al. 1976). A C532 pH and C833 conductivity meter (Consort, Turnhout, Belgium) were used to measure pH and conductivity, respectively. The biogas composition was measured with a compact gas chromatograph (Global Analyser Solutions, Breda, The Netherlands), and concentrations of different VFA were also analysed by means of gas chromatography (GC-2014, Shimadzu®, The Netherlands), as described in SI (S5).

2.7. Data deposition

Data deposition: the sequences reported in this paper have been deposited in the European Nucleotide Archive (ENA) database (Accession numbers LT624945-LT625937 for bacteria and LT625938-LT626064 for archaea).

3. Results

3.1. Impact of feed composition on digester performance

The shift in feedstock from WAS to molasses wastewater during the start-up resulted in an increase in methane production to 416 ± 15 mL CH L-1 d-1 on day 35 (Figure 1). The pH reached a value of 7.37 ± 0.04 (Figure S3), and VFA remained below the detection limit (Figure S4), indicating a stable start-up. During Phase 1, feeding of molasses wastewater initially resulted in an increase in methane production with a maximum value of 625 ± 17 mL CH4 L-1 d-1 on day 37 (Figure 1), and this corresponded with a COD conversion efficiency of 89.3 ± 2.4 %. After this, methane production decreased to a value of 232 ± 35 mL CH4 L-1 d-1 (33.2 ± 5.0 % conversion of the COD in the molasses wastewater to CH4) on day 84. A concomitant increase in total VFA concentration to 20.5 ±1.4 g COD L-1 was observed, which corresponded with 51.2 ± 4.0 % of the COD in the molasses wastewater converted to VFA (Figure S4). This points to a strong inhibition of the AD process. This decrease in methane production and increase in VFA concentration coincided with an increase in conductivity to 26.8 ± 0.7 mS cm-1 on day 84 (Figure 2). Similarly, an increase in the TAN concentration (Figure S5) could be observed, while free ammonia concentration (Figure S6) initially increased to a maximum value of 81 ± 12 mg N L-1 after which it decreased to 34 ± 3 mg N L-1 at the end of Phase 1.

The switch to WAS and glycerol as main feedstock in Phase 2 in Digester B resulted in an increase in methane production to 248 ± 17 mL CH4 L-1 d-1 on day 168, while VFA decreased to values below detection limit towards the end of Phase 2, indicating process recovery. In contrast, methane production in Digester A in which molasses wastewater was still used as feedstock decreased further to a value of 62 ± 15 mL L-1 d-1 (8.9 ± 2.1 % conversion of the COD in the molasses wastewater to CH4) and VFA further accumulated to 30.3 ± 1.8 g COD L-1, which corresponded with 75.7 ± 5.2 % of the COD in the molasses wastewater converted

to VFA. The conductivity reached a value of 29.6 ± 0.6 mS cm-1 in Digester A on day 168, while it was a factor 6 lower in Digester B, with only 4.68 ± 0.12 mS cm-1. A similar observation was made for TAN, with values of 1.5 ± 0 and 0.7 ± 0 g N L-1 in Digester A and B, respectively.

In the final phase (Phase 3), Digester A showed similar values to Phase 2 in terms of methane production and VFA concentration. Similarly, Digester B maintained stable methane production and VFA remained below the detection limit, which indicates that process recovery was maintained. The addition of NH4Cl and urea to Digester C resulted in an increase in TAN and free ammonia concentration, yet, this did not result in a decrease in methane production, and VFA remained below 0.5 g COD L-1 up to day 196. Towards the end of Phase 3, however, a minor increase in VFA concentration up to 1.2 ± 0.8 g COD L-1 or 3.0 ± 2.0% of the COD in the molasses wastewater was observed, while this was not the case for Digester B.

3.2. Microbial community changes in response to feed composition and operational conditions

Microbial community analysis resulted in an average of 11,762 ± 4,650 reads and 994 OTUs for the bacterial community, while 37,018 ± 19,424 reads and 127 OTUs were obtained for the archaeal community. No significant differences (P < 0.01) were observed in bacterial and archaeal community composition between biological replicates of the same digester.

3.2.1. Microbial community composition

The bacterial community was mainly dominated by the Firmicutes, Actinobacteria, Proteobacteria and Bacteroidetes phyla (Figure 3a). The Inoculum did not show a distinct dominance of any of the bacterial phyla, with the exception of a slightly higher relative

abundance (23.1 %) of the Proteobacteria phylum. In contrast, Actinobacteria mainly dominated Digester B and C, representing on average 38.4 ± 6.0 % of total bacterial reads, and this was mainly related with the increased relative abundance of Actinomycetales order (Figure S7). Firmicutes dominated Digester A, with an average of 65.5 ± 6.6 % of total bacterial reads, and the Clostridiales and Lactobacillales were the two main orders. A slight increase in Firmicutes relative abundance was, however, observed in Digester C, compared with Digester B, which was related with a relative increase in Lactobacillales. The difference in dominance between the Actinobacteria and Firmicutes phylum was clearly reflected in the composition of the feedstocks, i.e. WAS (47.4% of total reads for Actinobacteria) and molasses wastewater (93.8 % of total reads for Firmicutes), respectively. The archaeal community contained both acetoclastic and hydrogenotrophic methanogens, yet, their relative abundance strongly differed between the different digesters (Figure 3b). Digester B and C were mainly dominated by OTUs belonging to the acetoclastic Methanosaeta genus (35.9 ± 4.2 % of total archaeal reads), which was also the case for the inoculum sample (49.3 %), while Digester A was dominated by OTUs belonging to the hydrogenotrophic Methanocorpusculum (Phase 1, 35.4 %) and Methanobrevibacter (Phase 2 & 3, 54.7 ± 17.5 %) genera. Digester B also showed an increased relative abundance of the Methanosarcina genus (20.0 %). No methanogens were detected in the WAS and molasses wastewater feedstocks.

3.2.2. Microbial community dynamics

The evolution in community composition over time, and with respect to the different digesters was analysed via NMDS of the weighted Unifrac distance measure (Figure 4). Both for the bacterial and archaeal community, four distinct clusters could be identified, containing the Inoculum sample (Cluster 1), Digester A in Phase 1 (Cluster 2), Digester A in Phase 2 & 3

(Cluster 3) and Digester B and C in Phase 2 & 3 (Cluster 4). A significant difference in community composition was observed between Digester B and C on one hand and Digester A on the other hand for bacteria (P = 0.0006) and archaea (P = 0.009). A significant difference was also observed between the Inoculum and Digester B and C for bacteria (P = 0.0012) and archaea (P = 0.0006). This indicates a strong deviation from the original community, both for archaea and bacteria, despite similar operational conditions for the Inoculum and Digester B in Phase 2 & 3. Similar observations were made based on the unweighted Unifrac and Bray-Curtis dissimilarity measures (Figure S8).

For the bacterial community, only OTU851 (Unclassified Bacteria) showed a significant correlation (P = 0.001) with the Bray-Curtis dissimilarity matrix. In contrast, in the archaeal community four OTUs, i.e. OTU5 (Methanobrevibacter, P = 0.003), OTU28 (Methanobacterium, P = 0.001), OTU267 (Methanobacterium, P = 0.022), and OTU372 (Methanoregula, P = 0.001) were identified with a significant correlation with the Bray-Curtis dissimilarity matrix.

3.2.3. Microbial community redundancy and functionality

Microbial community redundancy, i.e. the change in microbial community organization in response to a disturbance (Werner et al. 2011) was evaluated via alpha diversity analysis by means of the Hill diversity order numbers (Hill 1973).

A clear difference could be observed between the three Hill diversity orders for the bacterial community. A significantly lower richness (H0) was observed between the Inoculum sample and Digester B in Phase 2 (P = 0.0152) and Phase 3 (P = 0.0203) (Figure 5a), while this was not the case for H1 and H2 (Figure 5b & c). The diversity in Digester B Phase 2 (P = 0.0050 for H0 and 0.0106 for H1), Digester B Phase 3 (P = 0.0052 for H0 and 0.0104 for H1), and Digester C Phase 3 (P = 0.0157 for H0 and 0.0408 for H1) was significantly higher, compared

308 with Digester A. The H2 diversity in the Inoculum was significantly higher than for Digester

309 A in Phase 1 (P = 0.0184) and Phase 2 (P = 0.0192).

310 The archaeal species richness (H0) was significantly lower in the Inoculum compared with

311 Digester B in Phase 2 (P = 0.0287), and Digester B (P = 0.0319) and Digester C (P = 0.0313)

312 in Phase 3 (Figure 5a), while this was not the case for H1 and H2 (Figure 5b & c). Each of the

313 three order numbers of the archaeal community was significantly lower in Digester A than

314 Digester B and C (Phase 2 & 3). No significant differences were detected between Digester B

315 and C, both for bacteria and archaea (P > 0.2).

4. Discussion

A clear shift in community composition was detected with respect to the shifts between WAS and molasses wastewater as main feedstock, both for the bacterial and archaeal community. Despite a retrieval of similar operational parameters in Phase 3, the microbial community did not return to its initial state following the disturbance phase, despite similar conditions. A decrease in alpha diversity was observed during the disturbance period (Phase 1), yet, it increased again following stabilization of the AD process in the final phase (Phase 2 & 3). The application of a mild disturbance in Phase 3 did not entail a strong effect on process or community stability.

4.1. Salt stress drives bacterial and archaeal community composition

Stable methane production was initially observed during the start-up phase, as VFA remained below the detection limit. Stable methane production was, however, not maintained, because during Phase 1 methane production and pH decreased, and VFA accumulated, indicating process failure. Both the increase in ammonia concentration and/or salinity (as measured by conductivity) could be considered responsible for this. As the TAN concentration did not reach values higher than 1.9 g N L-1 in Phase 1 & 2, and also free ammonia remained below 90 mg N L-1, ammonia toxicity was most likely not the main cause of process failure (Angelidaki and Ahring 1993, Rajagopal et al. 2013, Sung and Liu 2003). In contrast, salinity reached values up to 35 mS cm-1 in Phase 1, while it has been postulated that overall conductivity in AD should be maintained below 30 mS cm-1 to avoid salt stress (Chen et al. 2008, De Vrieze et al. 2012). The problem of salt inhibition during AD of molasses or vinasses wastewaters has been reported in numerous studies (De Vrieze et al. 2014, De Vrieze et al. 2015b, Fang et al. 2011, Syutsubo et al. 2013), thus, supporting our findings. The switch in feedstock from WAS to molasses also resulted in a shift in the microbial

community. The increase in Firmicutes, more specifically the Clostridiales and Lactobacillales orders, in Digester A and the increase in Actinobacteria in Digester B and C appears to be a direct consequence of the feedstock microbial composition itself, i.e. the molasses wastewater and WAS, respectively. This is in line with other studies in which the importance of the feedstock for shaping the community in AD has been demonstrated (Li et al. 2015a, Lu et al. 2013, Zhang et al. 2014, Ziganshin et al. 2013). In line with the results in our study, the increased relative abundance of the Actinobacteria often has been observed in sludge digesters (Chouari et al. 2005, De Vrieze et al. 2015c, Sundberg et al. 2013). The microbial composition of the feedstock is, however, not the only factor that determined the microbial community in the digesters. A dominance of the Clostridiales order often has been observed at suboptimal conditions for methanogenesis (e.g. increased ammonia and salt concentrations), irrespective of feedstock composition (Alsouleman et al. 2016, De Vrieze et al. 2015c, Muller et al. 2016). The shift in the methanogenic community from the acetoclastic Methanosaeta to the hydrogenotrophic Methanobrevibacter and Methanocorpusculum is also related with the increased salinity and process deterioration in general (Goberna et al. 2015, Walter et al. 2016). This potentially reflects the shift from acetoclastic methanogenesis to syntrophic acetate oxidation coupled with hydrogenotrophic methanogenesis, as reported earlier for Clostridiales and Methanobrevibacter or other hydrogenotrophic methanogens (Muller et al. 2016, Werner et al. 2014).

4.2. The microbial community does not return to its original composition following the disturbance phase

The significant shift in microbial community composition between the Inoculum and Digester A in Phase 1 was as anticipated, given the strong change in operational conditions. In Digester B in Phase 2 & 3, by adjusting the feed, operational conditions were readjusted to the

initial situation, yet, neither the bacterial, nor the archaeal community returned to its original composition. This is related with the high diversity of the bacterial community. Multiple bacterial species are able to occupy the same niche in AD, for example the fermentation of carbohydrates into VFA (Xia et al. 2015), thus, a different community can be used to obtain a similar process performance, as observed earlier (Goux et al. 2015). This was confirmed by the operational parameters in Digester B in Phase 2 &3. It also relates with the often observed (high) bacterial community dynamics over time, despite constant or similar operation (De Vrieze et al. 2016, Klang et al. 2015, Pycke et al. 2011). An alternative explanation might be the change in feedstock, e.g. the addition of glycerol to the feed, which has been shown to potentially influence microbial community structure (Zhang et al. 2014). However, it was shown by De Vrieze et al. (2015c) that digesters with similar operational conditions contained closely related microbial communities.

The significant change in the overall archaeal community between the Inoculum and Digester B in Phase 2 & 3 was not anticipated, given the much lower archaeal diversity compared with bacteria, as observed in this and other studies (Nelson et al. 2011, Sundberg et al. 2013). The archaeal community dynamics as a function of time at stable conditions are in most cases also lower than for bacteria (De Vrieze et al. 2013, Regueiro et al. 2014b, Town et al. 2014), as long as no (severe) process disturbance, as was the case in Phase 1, leads to (partial) inhibition of the methanogens (Poirier et al. 2016, Williams et al. 2013). Methanosaeta can be considered a key methanogen in the AD processes, because it is the, thus far, only known acetoclastic methanogen with a sufficiently low Ks value to efficiently convert acetate directly to methane at low acetate concentrations (Conklin et al. 2006, De Vrieze et al. 2012). This allows Methanosaeta to become dominant at low acetate concentrations, which is in contrast with Methanosarcina that has a higher Ks value and, thus, lower affinity for acetate (Conklin et al. 2006, De Vrieze et al. 2012). Despite the fact that a clear revival of the Methanosaeta

genus can be observed in Digester B, following the disturbance in Phase 1, its relative abundance is clearly lower compared with the Inoculum. This 'hiatus' in the methanogenic community was taken in by Methanoregula. The Methanoregula OTU372 significantly correlated with the Bray-Curtis dissimilarity matrix, which confirms its clear contribution to the methanogenic community shift. As Methanoregula is a hydrogenotrophic methanogen that can also use formate as substrate (Yashiro et al. 2011), it cannot take over the acetoclastic methanogenesis. Hence, this points to a partial shift in methanogenesis pathway caused by the disturbance, even though operational conditions, such as pH, TAN and VFA concentration, were similar in digester B in Phase 3 compared with the Inoculum. The inability of Methanosaeta to regain its previously uncontested position in the methanogenesis process was most likely hampered by its low growth rate (Conklin et al. 2006, De Vrieze et al. 2012), and the absence of Methanosaeta sp. in the WAS sludge, making feedstock inoculation not possible.

4.3. The microbial community retains its redundancy and metabolic potential following recovery after process disturbance

Rather than maintaining a constant composition, the resilience, resistance or redundancy of the microbial community is important to sustain process performance (Allison and Martiny 2008, Niu et al. 2015b, Werner et al. 2011). In this research, during process disturbance, the microbial community was unable to sustain the methane production process, most likely due to the high salinity, caused by the molasses wastewater. In Phase 2 & 3, Digester B regained process stability, reflected in the increase in methane production and removal of residual VFA. Both on the bacterial and archaeal level, the community did not rebound to its original composition. This demonstrates microbial community redundancy, rather than resilience or resistance as the major strategy behind AD process recovery (Allison and Martiny 2008,

417 Langer et al. 2015).

418 The metabolic potential of the microbial community was estimated by basic alpha diversity

419 analysis through the Hill diversity order numbers. The different diversity order numbers, both

420 for bacteria and archaea, indicated a similar, and sometimes even significantly higher

421 diversity in Digester B and C (Phase 2 & 3) in comparison with the Inoculum. Hence, this

422 proves that the metabolic potential of the microbial community is retained after process

423 disturbance.

424 To verify to which extent the microbial community would be able to deal with other mild

425 disturbances after process recovery, a mild (free) ammonia disturbance was provoked in

426 Digester C. Both methane production and pH values were maintained at similar levels as

427 Digester B, and only a minor accumulation of VFA was observed towards the end of Phase 3.

428 The overall bacterial and archaeal community did not show a significant change in

429 composition compared with Digester B. However, an increase in relative abundance of the

430 Firmicutes phylum and Methanosarcina genus, as observed earlier for (free) ammonia

431 disturbance in AD (De Vrieze et al. 2015a, Niu et al. 2015a), again demonstrates the

432 redundant character of the microbial community.

434 5. Conclusions

435 The application of a disturbance, in this case molasses wastewater, resulted in a significant

436 shift in bacterial and archaeal community composition, following recovery after the

437 disturbance. Despite the fact that the microbial community did not return to its original

438 position, microbial diversity was recovered. Hence, microbial redundancy was shown to be

439 the major strategy behind process recovery, continued operational stability and microbial

440 flexibility in AD in response to a common disturbance, i.e. most likely a high salinity.

442 Acknowledgements

443 Jo De Vrieze is supported as postdoctoral fellow from the Research Foundation Flanders

444 (FWO-Vlaanderen). Marlies E.R. Christiaens received support from Ghent University &

445 ESA/BELSPO for MELiSSA. Umer Zeeshan Ijaz is funded by NERC independent research

446 fellowship NE/L011956/1. The authors would like to thank Tim Lacoere for his assistance

447 with the molecular work, and Ruben Props and Frederiek-Maarten Kerckhof for their useful

448 suggestions. The authors also kindly acknowledge Jose Maria Carvajal Arroyo, Ioanna

449 Chatzigiannidou and Jeet Varia for critically reading the manuscript.

References

Allison, S.D. and Martiny, J.B.H. (2008) Resistance, resilience, and redundancy in microbial communities. Proceedings of the National Academy of Sciences of the United States of America 105, 11512-11519.

Alsouleman, K., Linke, B., Klang, J., Klocke, M., Krakat, N. and Theuerl, S. (2016) Reorganisation of a mesophilic biogas microbiome as response to a stepwise increase of ammonium nitrogen induced by poultry manure supply. Bioresource Technology 208, 200204.

Angelidaki, I. and Ahring, B.K. (1993) Thermophilic anaerobic digestion of livestock waste: the effect of ammonia. Applied Microbiology and Biotechnology 38(4), 560-564. Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S. and Srinath, E.G. (1976) Inhibition of Nitrification by Ammonia and Nitrous Acid. Journal Water Pollution Control Federation 48(5), 835-852.

Baho, D.L., Peter, H. and Tranvik, L.J. (2012) Resistance and resilience of microbial communities - temporal and spatial insurance against perturbations. Environmental Microbiology 14(9), 2283-2292.

Boon, N., De Windt, W., Verstraete, W. and Top, E.M. (2002) Evaluation of nested PCR-DGGE (denaturing gradient gel electrophoresis) with group-specific 16S rRNA primers for the analysis of bacterial communities from different wastewater treatment plants. Fems Microbiology Ecology 39(2), 101-112.

Briones, A. and Raskin, L. (2003) Diversity and dynamics of microbial communities in engineered environments and their implications for process stability. Current Opinion in Biotechnology 14(3), 270-276.

Chen, Y., Cheng, J.J. and Creamer, K.S. (2008) Inhibition of anaerobic digestion process: A review. Bioresource Technology 99(10), 4044-4064.

Chouari, R., Le Paslier, D., Daegelen, P., Ginestet, P., Weissenbach, J. and Sghir, A. (2005) Novel predominant archaeal and bacterial groups revealed by molecular analysis of an anaerobic sludge digester. Environmental Microbiology 7(8), 1104-1115. Clarke, K.R. and Warwick, R.M. (1998) Quantifying structural redundancy in ecological communities. Oecologia 113(2), 278-289.

Conklin, A., Stensel, H.D. and Ferguson, J. (2006) Growth kinetics and competition between Methanosarcina and Methanosaeta in mesophilic anaerobic digestion. Water Environment Research 78(5), 486-496.

Curtis, T.P. and Sloan, W.T. (2004) Prokaryotic diversity and its limits: microbial community structure in nature and implications for microbial ecology. Current Opinion in Microbiology 7(3), 221-226.

De Vrieze, J., Gildemyn, S., Vilchez-Vargas, R., Jâuregui, R., Pieper, D., Verstraete, W. and Boon, N. (2015a) Inoculum selection is crucial to ensure operational stability in anaerobic digestion. Applied Microbiology and Biotechnology 99(1), 189-199.

De Vrieze, J., Hennebel, T., Boon, N. and Verstraete, W. (2012) Methanosarcina: The rediscovered methanogen for heavy duty biomethanation. Bioresource Technology 112, 1-9. De Vrieze, J., Hennebel, T., Van den Brande, J., Bilad, R.i.M., Bruton, T.A., Vankelecom, I.F.J., Verstraete, W. and Boon, N. (2014) Anaerobic digestion of molasses by means of a vibrating and non-vibrating submerged anaerobic membrane bioreactor. Biomass and Bioenergy 68(0), 95-105.

De Vrieze, J., Plovie, K., Verstraete, W. and Boon, N. (2015b) Co-digestion of molasses or kitchen waste with high-rate activated sludge results in a diverse microbial community with stable methane production. Journal of Environmental Management 152(0), 75-82. De Vrieze, J., Raport, L., Roume, H., Vilchez-Vargas, R., Jâuregui, R., Pieper, D.H. and

Boon, N. (2016) The full-scale anaerobic digestion microbiome is represented by specific marker populations. Water Research 104, 101-110.

De Vrieze, J., Saunders, A.M., He, Y., Fang, J., Nielsen, P.H., Verstraete, W. and Boon, N. (2015c) Ammonia and temperature determine potential clustering in the anaerobic digestion microbiome. Water Research 75(0), 312-323.

De Vrieze, J., Verstraete, W. and Boon, N. (2013) Repeated pulse feeding induces functional stability in anaerobic digestion. Microbial Biotechnology 6(4), 414-424. Dearman, B., Marschner, P. and Bentham, R.H. (2006) Methane production and microbial community structure in single-stage batch and sequential batch systems anaerobically co-digesting food waste and biosolids. Applied Microbiology and Biotechnology 69(5), 589-596. Fang, C., Boe, K. and Angelidaki, I. (2011) Anaerobic co-digestion of desugared molasses with cow manure; focusing on sodium and potassium inhibition. Bioresource Technology 102(2), 1005-1011.

Feijoo, G., Soto, M., Mendez, R. and Lema, J.M. (1995) Sodium inhibition in the anaerobic digestion process: Antagonism and adaptation phenomena. Enzyme and Microbial Technology 17(2), 180-188.

Fernandez, A., Huang, S.Y., Seston, S., Xing, J., Hickey, R., Criddle, C. and Tiedje, J. (1999) How stable is stable? Function versus community composition. Applied and Environmental Microbiology 65(8), 3697-3704.

Fountoulakis, M.S., Petousi, I. and Manios, T. (2010) Co-digestion of sewage sludge with glycerol to boost biogas production. Waste Management 30(10), 1849-1853. Franke-Whittle, I.H. and Insam, H. (2013) Treatment alternatives of slaughterhouse wastes, and their effect on the inactivation of different pathogens: A review. Critical Reviews in Microbiology 39(2), 139-151.

Gallert, C., Bauer, S. and Winter, J. (1998) Effect of ammonia on the anaerobic degradation of protein by a mesophilic and thermophilic biowaste population. Applied Microbiology and Biotechnology 50(4), 495-501.

Garcia, M.L. and Angenent, L.T. (2009) Interaction between temperature and ammonia in mesophilic digesters for animal waste treatment. Water Research 43(9), 2373-2382. Goberna, M., Gadermaier, M., Franke-Whittle, I.H., Garcia, C., Wett, B. and Insam, H. (2015) Start-up strategies in manure-fed biogas reactors: Process parameters and methanogenic communities. Biomass & Bioenergy 75, 46-56.

Goux, X., Calusinska, M., Lemaigre, S., Marynowska, M., Klocke, M., Udelhoven, T., Benizri, E. and Delfosse, P. (2015) Microbial community dynamics in replicate anaerobic digesters exposed sequentially to increasing organic loading rate, acidosis, and process recovery. Biotechnology for Biofuels 8, 18.

Greenberg, A.E., Clesceri, L.S. and Eaton, A.D. (1992) Standard Methods for the Examination of Water and Wastewater American Public Health Association Publications, Washington.

Hill, M.O. (1973) Diversity and Evenness: A Unifying Notation and Its Consequences. Ecology 54(2), 427-432.

Klang, J., Theuerl, S., Szewzyk, U., Huth, M., Tolle, R. and Klocke, M. (2015) Dynamic variation of the microbial community structure during the long-time mono-fermentation of maize and sugar beet silage. Microbial Biotechnology 8(5), 764-775.

Langer, S.G., Ahmed, S., Einfalt, D., Bengelsdorf, F.R. and Kazda, M. (2015) Functionally redundant but dissimilar microbial communities within biogas reactors treating maize silage in co-fermentation with sugar beet silage. Microbial Biotechnology 8(5), 828-836. Li, J.B., Rui, J.P., Yao, M.J., Zhang, S.H., Yan, X.F., Wang, Y.P., Yan, Z.Y. and Li, X.Z. (2015a) Substrate Type and Free Ammonia Determine Bacterial Community Structure in Full-Scale Mesophilic Anaerobic Digesters Treating Cattle or Swine Manure. Frontiers in

Microbiology 6, 10.

Li, L., He, Q., Ma, Y., Wang, X.M. and Peng, X.Y. (2015b) Dynamics of microbial community in a mesophilic anaerobic digester treating food waste: Relationship between community structure and process stability. Bioresource Technology 189, 113-120. Lu, X.Y., Rao, S., Shen, Z.Y. and Lee, P.K.H. (2013) Substrate induced emergence of different active bacterial and archaeal assemblages during biomethane production. Bioresource Technology 148, 517-524.

McCann, K.S. (2000) The diversity-stability debate. Nature 405(6783), 228-233.

McMurdie, P.J. and Holmes, S. (2013) phyloseq: An R Package for Reproducible Interactive

Analysis and Graphics of Microbiome Census Data. Plos One 8(4), e61217.

Moestedt, J., Muller, B., Westerholm, M. and Schnurer, A. (2016) Ammonia threshold for

inhibition of anaerobic digestion of thin stillage and the importance of organic loading rate.

Microbial Biotechnology 9(2), 180-194.

Muller, B., Sun, L., Westerholm, M. and Schnurer, A. (2016) Bacterial community composition and fhs profiles of low- and high-ammonia biogas digesters reveal novel syntrophic acetate-oxidising bacteria. Biotechnology for Biofuels 9, 18. Muyzer, G., Dewaal, E.C. and Uitterlinden, A.G. (1993) Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Applied and Environmental Microbiology 59(3), 695700.

Nelson, M.C., Morrison, M. and Yu, Z.T. (2011) A meta-analysis of the microbial diversity observed in anaerobic digesters. Bioresource Technology 102(4), 3730-3739. Niu, Q.G., Kubota, K., Qiao, W., Jing, Z.Q., Zhang, Y.L. and Yu-You, L. (2015a) Effect of ammonia inhibition on microbial community dynamic and process functional resilience in mesophilic methane fermentation of chicken manure. Journal of Chemical Technology and Biotechnology 90(12), 2161-2169.

Niu, Q.G., Qiao, W., Qiang, H. and Li, Y.Y. (2013) Microbial community shifts and biogas conversion computation during steady, inhibited and recovered stages of thermophilic methane fermentation on chicken manure with a wide variation of ammonia. Bioresource Technology 146, 223-233.

Niu, Q.G., Takemura, Y., Kubota, K. and Li, Y.Y. (2015b) Comparing mesophilic and thermophilic anaerobic digestion of chicken manure: Microbial community dynamics and process resilience. Waste Management 43, 114-122.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara R.B., Simpson, G.L., Solymos, P., Stevens, M.H.H. and Wagner, H. (2016) Vegan: Community ecology package. R package version 2.3-4.

Pitk, P., Kaparaju, P., Palatsi, J., Affes, R. and Vilu, R. (2013) Co-digestion of sewage sludge and sterilized solid slaughterhouse waste: Methane production efficiency and process limitations. Bioresource Technology 134, 227-232.

Poirier, S., Bize, A., Bureau, C., Bouchez, T. and Chapleur, O. (2016) Community shifts within anaerobic digestion microbiota facing phenol inhibition: Towards early warning microbial indicators? Water Research 100, 296-305.

Pycke, B.F.G., Etchebehere, C., Van de Caveye, P., Negroni, A., Verstraete, W. and Boon, N. (2011) A time-course analysis of four full-scale anaerobic digesters in relation to the dynamics of change of their microbial communities. Water Science and Technology 63(4), 769-775.

R Development Core Team (2013) R: A Language and Environment for Statistical Computing. 3.0 ed. Vienna, Austria: R Foundation for Statistical Computing. Rajagopal, R., Masse, D.I. and Singh, G. (2013) A critical review on inhibition of anaerobic digestion process by excess ammonia. Bioresource Technology 143, 632-641.

Regueiro, L., Carballa, M. and Lema, J.M. (2014a) Outlining microbial community dynamics during temperature drop and subsequent recovery period in anaerobic co-digestion systems. Journal of Biotechnology 192, Part A, 179-186.

Regueiro, L., Veiga, P., Figueroa, M., Lema, J.M. and Carballa, M. (2014b) Influence of transitional states on the microbial ecology of anaerobic digesters treating solid wastes. Applied Microbiology and Biotechnology 98(5), 2015-2027.

Schnurer, A. and Nordberg, A. (2008) Ammonia, a selective agent for methane production by syntrophic acetate oxidation at mesophilic temperature. Water Science and Technology 57(5), 735-740.

Schnurer, A., Zellner, G. and Svensson, B.H. (1999) Mesophilic syntrophic acetate oxidation during methane formation in biogas reactors. Fems Microbiology Ecology 29(3), 249-261. Sundberg, C., Al-Soud, W.A., Larsson, M., Alm, E., Yekta, S.S., Svensson, B.H., Sorensen, S.J. and Karlsson, A. (2013) 454 pyrosequencing analyses of bacterial and archaeal richness in 21 full-scale biogas digesters. Fems Microbiology Ecology 85(3), 612-626. Sung, S.W. and Liu, T. (2003) Ammonia inhibition on thermophilic anaerobic digestion. Chemosphere 53(1), 43-52.

Syutsubo, K., Onodera, T., Choeisai, P., Khodphuvieng, J., Prammanee, P., Yoochatchaval, W., Kaewpradit, W. and Kubota, K. (2013) Development of appropriate technology for treatment of molasses-based wastewater. Journal of Environmental Science and Health Part aToxic/Hazardous Substances & Environmental Engineering 48(9), 1114-1121. Town, J., Annand, H., Pratt, D., Dumonceaux, T. and Fonstad, T. (2014) Microbial community composition is consistent across anaerobic digesters processing wheat-based fuel ethanol waste streams. Bioresource Technology 157, 127-133.

Usack, J.G. and Angenent, L.T. (2015) Comparing the inhibitory thresholds of dairy manure co-digesters after prolonged acclimation periods: Part 1-Performance and operating limits. Water Research 87, 446-457.

Vilchez-Vargas, R., Geffers, R., Suarez-Diez, M., Conte, I., Waliczek, A., Kaser, V.S., Kralova, M., Junca, H. and Pieper, D.H. (2013) Analysis of the microbial gene landscape and transcriptome for aromatic pollutants and alkane degradation using a novel internally calibrated microarray system. Environmental Microbiology 15(4), 1016-1039. Walter, A., Probst, M., Hinterberger, S., Muller, H. and Insam, H. (2016) Biotic and abiotic dynamics of a high solid-state anaerobic digestion box-type container system. Waste Management 49, 26-35.

Werner, J.J., Garcia, M.L., Perkins, S.D., Yarasheski, K.E., Smith, S.R., Muegge, B.D., Stadermann, F.J., DeRito, C.M., Floss, C., Madsen, E.L., Gordon, J.I. and Angenent, L.T. (2014) Microbial Community Dynamics and Stability during an Ammonia-Induced Shift to Syntrophic Acetate Oxidation. Applied and Environmental Microbiology 80(11), 3375-3383. Werner, J.J., Knights, D., Garcia, M.L., Scalfone, N.B., Smith, S., Yarasheski, K., Cummings, T.A., Beers, A.R., Knight, R. and Angenent, L.T. (2011) Bacterial community structures are unique and resilient in full-scale bioenergy systems. Proceedings of the National Academy of Sciences of the United States of America 108(10), 4158-4163.

Westerholm, M., Muller, B., Arthurson, V. and Schnurer, A. (2011) Changes in the Acetogenic Population in a Mesophilic Anaerobic Digester in Response to Increasing Ammonia Concentration. Microbes and Environments 26(4), 347-353.

Williams, J., Williams, H., Dinsdale, R., Guwy, A. and Esteves, S. (2013) Monitoring methanogenic population dynamics in a full-scale anaerobic digester to facilitate operational management. Bioresource Technology 140(0), 234-242.

Xia, Y., Chin, F.Y.L., Chao, Y.Q. and Zhang, T. (2015) Phylogeny-structured carbohydrate metabolism across microbiomes collected from different units in wastewater treatment process. Biotechnology for Biofuels 8, 12.

650 Yashiro, Y., Sakai, S., Ehara, M., Miyazaki, M., Yamaguchi, T. and Imachi, H. (2011)

651 Methanoregula formicica sp. nov., a methane-producing archaeon isolated from

652 methanogenic sludge. International Journal of Systematic and Evolutionary Microbiology 61,

653 53-59.

654 Zhang, W., Werner, J.J., Agler, M.T. and Angenent, L.T. (2014) Substrate type drives

655 variation in reactor microbiomes of anaerobic digesters. Bioresource Technology 151, 397656 401.

657 Zhang, X.D., Tao, Y., Hu, J.M., Liu, G., Spanjers, H. and van Lier, J.B. (2016)

658 Biomethanation and microbial community changes in a digester treating sludge from a

659 brackish aquaculture recirculation system. Bioresource Technology 214, 338-347.

660 Ziganshin, A.M., Liebetrau, J., Proter, J. and Kleinsteuber, S. (2013) Microbial community

661 structure and dynamics during anaerobic digestion of various agricultural waste materials.

662 Applied Microbiology and Biotechnology 97(11), 5161-5174.

664 Tables:

665 Table 1 Characteristics of the anaerobic sludge that was used to inoculate the digesters (n=3).

Parameter

pH 7.33 ± 0.04

Total solids, TS (g L-1) 59.5 ± 3.4

Volatile solids, VS (g L-1) 49.3 ± 3.2

Conductivity (mS cm-1) 5.71 ± 0.47

Total VFA 0 ± 0

Acetate 0 ± 0

Propionate 0 ± 0 Total ammonia nitrogen, TAN (mg L-1) 220 ± 5

Free ammonia (mg L-1) 0.45 ± 0.01

667 Table 2 Characteristics of the waste activated sludge and molasses wastewater (n=3).

Parameter Waste activated sludge Molasses wastewater

pH (-) 6.83 ± 0.04 5.63 ± 0.04

Total solids, TS (g kg-1) 58.7 ± 0.2 52.6 ± 0.6

Volatile solids, VS (g kg-1) 20.2 ± 0.2 36.7 ± 1.7

Total COD (g kg-1) 29.6 ± 2.0 40.0 ± 1.4

Conductivity (mS cm-1) 2.70 ± 0.36 13.40 ± 0.45

Total VFA (mg COD kg-1) 0 ± 0 560 ± 19

Acetate (mg COD kg-1) 0 ± 0 560 ± 19

Propionate (mg COD kg-1) 0 ± 0 0 ± 0

Total ammonia nitrogen, TAN (mg N kg-1) 249 ± 7 132 ± 10

Kjeldahl nitrogen, TKN (mg N kg-1) 1650±70 3290 ± 90

COD:N ratio 18.0 ± 1.4 12.2 ± 0.5

TS:VS ratio 2.90 ± 0.03 1.43 ± 0.07

COD:VS ratio 1.47 ± 0.10 1.09 ± 0.06

669 Table 3 Overview of the main difference between the treatments in the different phases of the

670 experiment. The organic loading rate and hydraulic retention time were kept constant at 2.0 g

671 COD L-1 d-1 and 20 days, respectively, in each treatment. WAS = waste activated sludge.

Phase Period Feed composition

Digester A Digester B Digester C

Start-up Day 0-34 WAS + molasses WAS + molasses WAS + molasses

Phase 1 Day 35-83 molasses molasses molasses

Phase 2 Day 84-167 molasses WAS + glycerol WAS + glycerol

Phase 3 Day 168-245 molasses WAS + glycerol WAS + glycerol +

urea/NH4Cl

673 Figures:

Time (days)

675 Figure 1 Methane production in the three different digesters. Error bars represent standard

676 deviations of the biological replicates. In the Start-up phase and Phase 1, the average value

677 (n=9) was calculated over Digester A, B and C, as these were still biological replicates. In

678 Phase 2, the average value (n=6) was calculated over Digester B and C, as these were still

679 biological replicates, while Digester A (n=3) was considered separately, due to a different

680 treatment. In Phase 3, Digester A, B and C (each n=3) were considered separately, due to a

681 different treatment.

o 3 "O

e 10 o

Start-up Phase 1 Phase 2 Phase 3 -rj

V -Digester A

\ -Digester B

\ -Digester C

100 150

Time (days)

Figure 2 Conductivity in the mixed liquor of the three different digesters. Error bars represent standard deviations of the biological replicates. In the Start-up phase and Phase 1, the average value (n=9) was calculated over Digester A, B and C, as these were still biological replicates. In Phase 2, the average value (n=6) was calculated over Digester B and C, as these were still biological replicates, while Digester A (n=3) was considered separately, due to a different treatment. In Phase 3, Digester A, B and C (each n=3) were considered separately, due to a different treatment.

C P <—f-ru eu

CTQ re

CTQ CD

t/J CZ!

>—». »—k.

CTQ OQ

CZ) CZ)

(—K (—K

>-! i-i

Nitrospira

Acidobacteria

Planctomycetes Verrucomicrobia WS3

Actinobacteria

Bacteroidetes

Proteobacteria

Firmicutes

Chloroflexi

Synergistetes

Thermotogae

»—k. »—k. »—k.

OQ OQ OQ

re re re

czi czj czj

re re re

■—i ■—t i

> od o

Methanosphaera

Thermogymnomonas

Methanobrevibacter

Methanocorpusculum

Methanosphaerula

Methanoculleus

Methanosarcina

Methanoregula

Methanospirillum

Methanolinea

Methanomethylovorans

Methanobacterium

Methanosaeta

cr p W3 re cr p czi re cr p czi re hd ha tr tr p p czi czi re re tr p CZ) re

to i to i w u> 1 1 UJ i

O i a i a 1 1 o o i U

oq' rt> CZ) oq' re CZ) oq' re CZ) oq' oq' re re CZ) CZ) oq' re CZ)

re i-t re i-t re i-t re re i-t i-t re i-i

> > dd > dd o

Figure 3 Heatmap showing the square root transformed relative abundance of (a) the bacterial community at the phylum level, and (b) the archaeal community at the genus level. Weighted average values of the biological replicates are presented. The colour scale ranges from 0 (white) to 80% (red) relative abundance. In Phase 1, the average value (n=9) was calculated over Digester A, B and C, as these were still biological replicates. In Phase 2, the average value (n=6) was calculated over Digester B and C, as these were still biological replicates, while Digester A (n=3) was considered separately, due to a different treatment. In Phase 3, Digester A, B and C (each n=3) were considered separately, due to a different treatment.

▲ ▲

CM CO Q

Treatment Phase

• • Inoculum

▲ Digester A • Phase l

■ Digester B • Phase 2

+ Digester C • Phase 3

▲ ▲

▲ ▲ ▲ ▲ A 1

^ I ▲

(b) ▲

0.00 NMDS1

0.0 0.1 NMDS1

Figure 4 Non-metric distance scaling (NMDS) analysis of the weighted Unifrac distance indices of the (a) bacterial, and (b) archaeal community at OTU level. Different symbols and colours are used for different digesters and phases, respectively.

708 Figure 5 Alpha diversity of the samples for each digester in the different phases. The three

709 Hill order diversity numbers (a) H0 (richness, number of OTUs), (b) H1 (exponential value of

710 the Shannon index) and (c) H2 (inverse Simpson index) were calculated both for bacteria and

711 archaea. Error bars represent standard deviations of the biological replicates. In Phase 1, the

712 average value (n=9) was calculated over Digester A, B and C, as these were still biological

713 replicates. In Phase 2, the average value (n=6) was calculated over Digester B and C, as these

714 were still biological replicates, while Digester A (n=3) was considered separately, due to a

715 different treatment. In Phase 3, Digester A, B and C (each n=3) were considered separately,

716 due to a different treatment.

Time (days)

Time (days)

h-1 •

Cu: CTQ

. I—».

! CTQ CD

>—k. »—k.

CTQ CTQ

Ct> CD

r-K (—K

>-! i-i

Nitrospira

Acidobacteria

Planctomycetes Verrucomicrobia WS3

Actinobacteria

Bacteroidetes

Proteobacteria

Firmicutes

Chloroflexi

Synergistetes

Thermotogae

»—k. »—k. »—k.

CTQ CTQ CTQ

re re re

CO CO CO

re re re

■—i ■—t i

> ro o

id tr p CO re id cr p CO re ^d cr p CO re

^ to I to i

O I p i P

aq' re CO aq' re CO aq' re CO

re i-t re i-t re i-t

> > ro

Methanosphaera

Thermogymnomonas

Methanobrevibacter

Methanocorpusculum

Methanosphaerula

Methanoculleus

Methanosarcina

Methanoregula

Methanospirillum

Methanolinea

Methanomethylovorans

Methanobacterium

Methanosaeta

Hd id id

tr cr cr

CO CO CO

re re re www

h-k• h-k• h-»•

OQ OQ OQ

re re re

CO CO CO

I-»- I-»-

re re re

I-» H» !-(

> ro o

▲ ▲

Treatment Dhase

• • Inoculum

▲ Digester A • Phase l

■ Digester B • Phase 2

+ Digester C • Phase 3

0.00 NMDS1

CM CO Q

▲ ▲

▲ ▲ ▲ ▲ A 1

R- | ▲

(b) ▲

Highlights manuscript "Microbial community redundancy in anaerobic digestion drives process recovery after salinity exposure."

• Anaerobic digestion of high-salinity molasses wastewater causes process inhibition.

• A shift in substrate resulted in complete process recovery.

• The bacterial and archaeal community maintained their metabolic potential.

• Microbial community redundancy was the major strategy to ensure process stability.