Biogas production from lignocellulosic agricultural residues

Microbial approaches for enhanced efficiency

Tong Liu Faculty of Natural Resources and Agricultural Sciences

Department of Molecular Sciences Uppsala

Doctoral thesis Swedish University of Agricultural Sciences

Uppsala 2019

Acta Universitatis agriculturae Sueciae 2019:5

ISSN 1652-6880 ISBN (print version) 978-91-7760-328-3 ISBN (electronic version) 978-91-7760-329-0 © 2019 Tong Liu, Uppsala Print: SLU Service/Repro, Uppsala 2019

Cover: Scanning Electron Microscope (SEM) photo showing pure cultured Clostridium sp. Bciso-3 degrading cellulose, isolated from an industrial-scale anaerobic digester. (Photo: Tong Liu. Colorized by Johnny Isaksson)

Methane, produced through microbial anaerobic digestion of various organic materials, is seen as a promising sustainable bioenergy source with the potential to reduce the current dependence on fossil fuels. Among organic materials, lignocellulosic materials, especially agriculture residues, are highly interesting due to high abundance and potential for methane production. However, low nutrient content and highly recalcitrant structure often limit process efficiency. This thesis presents the results of in-depth studies conducted in order to obtain new information about lignocellulose-degrading bacteria in biogas processes and to identify ways to enable more efficient biogas production.

Different biogas processes were investigated in terms of their overall microbial community (bacteria and archaea) and potential lignocellulose degraders. The results showed that the biogas processes differed with regard to overall microbial community and chemical composition, but also composition of the cellulose-degrading bacterial community. These differences significantly influenced the degradation efficiency of both cellulose and wheat straw in batch digestion systems and also performance during start-up of semi-continuous stirred tank reactor (CSTR) processes. A positive correlation was found between lignocellulose degradation efficiency and relative abundance of Clostridium cellulolyticum. Ammonia level in the inoculum was identified as the most significant factor potentially affecting microbial community structure and methane production from lignocellulosic materials. Microbial and chemical composition of the original inoculum sources also influenced long-term degradation of lignocellulose in CSTR and appeared to influence residual methane potential. Different molecular methods for microbial community analysis were explored, with the aim of building an appropriate pipeline for in-depth studies of lignocellulose degraders in anaerobic reactors.

This thesis provides novel information about the microbial communities involved in degradation of lignocellulosic materials and possible connections to process parameters. This information could potentially enable biogas production to be steered towards a more efficient and controllable process for degradation and biogas production from agriculture residues and plant-based materials.

Keywords: anaerobic digestion, lignocellulose, glycoside hydrolase families 5 and 48, biomethane potential, continuous stirred-tank reactor, co-digestion, residual methane potential, next-generation amplicon sequencing, terminal restriction fragment length polymorphism (T-RFLP).

Author’s address: Tong Liu, SLU, Department of Molecular Sciences, P.O. Box 7015, 750 07 Uppsala, Sweden

Biogas production from lignocellulosic agricultural residues. Microbial approaches for enhanced efficiency

Abstract

Metan, som produceras genom mikrobiell nedbrytning av olika organiska material under anaeroba förhållanden, ses som en lovande hållbar bioenergikälla med potential att minska det nuvarande beroendet av fossila bränslen. I detta sammanhang representerar jordbruksrester, som finns tillgängligt i stor mängd, en stor metanpotential. Tyvärr har denna typ av material ofta ett lågt näringsinnehåll och ett högt innehåll av lignocellulosa, som är svårt att bryta ner och därför begränsar processens effektivitet. Denna avhandling presenterar resultat från studier som genomförts för att ta fram ny information om bakterier som bryter ner lignocellulosa i biogasprocesser. Målet var att identifiera sätt att möjliggöra en effektivare biogasproduktion.

Olika biogasprocesser undersöktes med avseende på sammansättningen av det mikrobiella samhället (bakterier och arkeer) och bakterier med potentiell förmåga att bryta ner lignocellulosa. För den mikrobiella analysen användes olika molekylära metoder. Resultaten visade att de olika biogasprocesserna var olika i avseende både till den kemiska sammansättningen och det mikrobiella samhället, inklusive de cellulosanedbrytande bakterierna. Dessa skillnader påverkade signifikant nedbrytningseffektiviteten av cellulosa och vetehalm i satsvisa metanproduktionsprocesser. Under dessa försök identifierades en negativ korrelation mellan nedbrytningseffektiviteten och halten ammoniak, samt en positiv korrelation med mängden av en specifik cellulosanedbrytande bakterie, Clostridium cellulolyticum. Uppstart av semi-kontinuerligt omrörda biogasreaktorer (CSTR) visade också tydliga skillnader i processprestanda beroende på ympens ammoniakhalt och på sammansättningen av det mikrobiella samhället. En koppling mellan låg nedbrytningseffektivitet och resterande metanpotential identifierades också.

Kunskap som genererats i denna avhandling kan potentiellt möjliggöra styrning mot en mer effektiv och kontrollerbar process för nedbrytning och biogasproduktion från jordbruksrester och växtbaserade material.

Nyckelord: anaerob nedbrytning (rötning), lignocellulosa, glykosidhydrolas familj 5 och 48, biometanpotential, CSTR, samrötning, restgas produktion, pyrosekvensiering, T-RFLP

Författarens adress: Tong Liu, SLU, Institutionen för molekylär vetenskap, P.O. Box 7015, 750 07 Uppsala, Sverige

Biogasproduktion från lantbrukets lignocellulosrika restprodukter. Mikrobiella tillvägagångssätt för ökad effektivitet

Sammanfattning

To me, myself, my mom: , my wife: , and my cat: Wasabi

Dedication

List of publications 9

List of figures 13

Abbreviations 14

1 Introduction 15 1.1 Hypothesis 17 1.2 Aim 17

2 Biogas production 19 2.1 Current status of biogas production in the EU and Sweden 22

3 Lignocellulosic materials as a substrate for biogas production 25 3.1 Structure of lignocellulose 25 3.2 Lignocellulosic substrate optimisation 27 3.3 Anaerobic digestion process configurations for lignocellulose-rich

material 28 3.4 Process regulating parameters 32 3.5 Process monitoring parameters 33

4 Microbial communities 37 4.1 Lignocellulose degraders in the anaerobic environment 39 4.2 Enzymatic depolymerisation of cellulose and hemicelluloses 41 4.3 Lignocellulolytic communities and influence of digester parameters 43

5 Microbial community analysis techniques 45 5.1 Methods applied to study lignocellulolytic communities 46

6 Conclusions and future perspectives 51

References 53

Popular science summary 65

Contents

Populärvetenskaplig sammanfattning 67

Acknowledgements 69

9

This thesis is based on the work contained in the following papers, referred to by Roman numerals in the text:

I Liu, T., Sun, L., Müller, B. and Schnürer, A.* (2016). The microbial community structure in industrial biogas plants influences the degradation rate of straw and cellulose in batch tests. Biotechnology for Biofuels 9, 1-20.

II Liu, T., Sun, L., Müller, B. and Schnürer, A.* (2017). Importance of inoculum source and initial community structure for biogas production from agricultural substrates. Bioresource Technology 245, 768-777.

III Liu, T., Sun, L., Nordberg, Å. and Schnürer, A.* (2018). Substrate-induced response in biogas process performance and microbial community relates back to inoculum source. Microorganisms 6, 80-99.

IV Ahlberg-Eliasson, K., Liu. T., Nadeau, E. and Schnürer, A.* (2018). Forage types and origin of manure in codigestion affect methane yield and microbial community structure. Grass and Forage Science 73, 740-757.

Papers I-IV are reproduced with the permission of the publishers.

* Corresponding author.

List of publications

10

In addition to Paper I-IV, I contributed to the following paper within the time frame of this thesis work:

Šafarič, L., Yekta, S.S.*, Liu, T., Svensson, B.H., Schnürer, A., Bastviken D. and Björn, A. (2018). Dynamics of a perturbed microbial community during thermophilic anaerobic digestion of chemically defined soluble organic compounds. Microorganisms 6, 105-118.

* Corresponding author.

I. Participated in planning the study and analysing the results. Performed some molecular work, monitored the reactors and was engaged in writing the article.

II. Participated in planning the study and analysing the results. Performed all molecular work and was involved in reactor operation. Main writer of the article.

III. Participated in planning the study and analysing the results. Performed all molecular work and was involved in reactor operation. Main writer of the article.

IV. Participated in analysing the results. Performed some molecularwork and was engaged in writing the article.

My contribution to the papers included in this thesis was as follows:

11

13

Figure 1. The anaerobic digestion process leading to biogas production. Figure 2. Biogas buses refuelling at a biogas station (Photo: Anna Schnürer,

Uppsala). Figure 3. Grass bedding mixed with cattle manure, an agricultural waste with

high potential for biogas production. Figure 4. Microstructure of a typical plant cell wall, indicating the relationship

between cellulose, hemicellulose and lignin. Figure 5. Sealed serum bottles on a rotary shaker in a biomethane potential

(BMP) test. Figure 6. Left: A series of continuously stirred tank reactors (CSTR). Right:

SLU full-scale biogas plant at Lövsta (Photo: Anna Schnürer). Figure 7. Operating parameters affecting microbial community and thus

potentially biogas production performance. Figure 8. Scanning electron microscope (SEM) image of material isolated from

an industrial-scale anaerobic digester, showing pure-cultured Clostridium sp. Bciso-3 degrading cellulose.

Figure 9. Structure of the cellulose chain. Figure 10. Left: Anaerobic serum bottles of lignocellulolytic microorganism-

enriched culture. Right: Anaerobic glass tubes with agar for picking single colonies.

List of figures

14

AD Anaerobic digestion

BMP Biomethane potential

CSTR Continuously stirred tank reactor

GH Glycoside hydrolase

HRT Hydraulic retention time

LC-MS Liquid chromatography-mass spectrometry

NGS Next-generation sequencing

OLR Organic loading rate

PCR Polymerase chain reaction

Qiime Quantitative insights into microbial ecology

qPCR Quantitative polymerase chain reaction

RMP Residual methane potential

SBS Sequencing by synthesis

SMP Specific methane production

T-RFLP Terminal restriction fragment length polymorphism

T-RFs Terminal restriction fragments

VFA Volatile fatty acid

WWTP Wastewater treatment plant

Abbreviations

15

The development of the petroleum industry has led to a rapid rise in the world economy in the past century. However, the underlying resource, fossil fuel, is recognised as a limited energy resource. Furthermore, emissions of greenhouse gases (e.g. fossil fuel-derived carbon dioxide (CO2) emissions) have become a global concern, since about 88% of global energy consumption derives from fossil fuels (Achinas et al., 2017; Agency, 2015). To meet the environmental challenges and overcome the dependence on fossil fuel, European Union (EU) member states have decided to increase the proportion of renewable energy to 20% of total consumption by 2020 (Karmellos et al., 2016). Progress towards this target is measured every two years and was proposed on 30 November 2016 to reach at least 27% renewables in final energy consumption in the EU by 2030 (Scarlat et al., 2018).

Biogas is seen as one of the most important renewable energy resources that can replace part of the fossil fuel-based energy used today, and it shows great potential and many advantages, including both climate and economic benefits (Meyer-Aurich et al., 2016). A biogas process can be implemented in small or large scale, which is important when designing flexible and sustainable energy solutions in both industrialised and developing countries (Holm-Nielsen et al., 2009). Materials that can be used for biogas production include various types of waste products, such as manure, straw, municipal wastewater, food waste etc., and dedicated energy crops (Vasco-Correa et al., 2018; Appels et al., 2011). Among these substrates, lignocellulosic materials, such as agricultural residues, are of great interest due to their high abundance and potential for biogas production (Azman et al., 2015). By controlled use of wastes in a biogas process rather than e.g. dumping household waste in landfill or storing farm manure in open tanks, it is possible not only reduce the number of waste deposits, but also to decrease emissions of carbon dioxide and other greenhouse gases (Borjesson & Mattiasson, 2008). The biogas produced, containing the energy carrier methane, can be used for production of heat,

1 Introduction

16

electricity and vehicle fuel after upgrading (removal of carbon dioxide and trace gases) (Holm-Nielsen et al., 2009). The residues left after biogas production are rich in mineral nutrients and can be used as a fertiliser during crop production to replace fossil energy-requiring mineral fertilisers, thus enabling recycling of nutrients between urban and rural areas (Vasco-Correa et al., 2018; Möller & Müller, 2012; Weiland, 2010; Holm-Nielsen et al., 2009).

Microorganisms are essential for degrading organic material to biogas, in a process that involves various anaerobic digestion pathways and requires the

combined activity of several groups of microorganisms with differing metabolic capacities (Angelidaki et al., 2011). To obtain a stable biogas process, all these conversion steps and microorganisms must work in a synchronised manner (Vanwonterghem et al., 2014a). When plant-based materials (e.g. agricultural residues) are used for biogas production, the first step of the microbiological process, hydrolysis, becomes rate-limiting. It has been suggested that the crystalline structure of the lignocelluloses obstructs degradation in the initial step, and thus the hydrolysis of these insoluble compounds becomes slow (Mulat & Horn, 2018; Lynd et al., 2002a; Noike et al., 1985).

Some of the obstacles with degradation of these types of materials can be overcome by various pre-treatment methods, making the material more accessible to microbial and enzymatic attack (Martínez-Gutiérrez, 2018). An alternative strategy is to increase the efficiency of the active microbial community. Numerous studies have been devoted to examining anaerobic cellulose-degrading bacteria and their enzymatic capabilities, in efforts to clarify the degradation mechanisms and identify ways to enhance degradation rates. Most of these studies have been performed on samples from gut and soil ecosystems (Tsavkelova & Netrusov, 2012; Lynd et al., 2002a) (Do et al., 2018; Ransom-Jones et al., 2012; Morrison et al., 2009b; Miron et al., 2001), while only a few have examined cellulose-degrading bacteria in biogas digesters (Jia et al., 2018; Bozan et al., 2017; Azman et al., 2015; Sun et al., 2013; Yan et al., 2012). Consequently, insufficient information is available on cellulose-degrading communities in biogas processes and on possibilities to enhance the degradation rate by ‘microbial steering’, i.e. by supporting the growth of highly efficient cellulose-degrading bacteria or communities.

17

1.1 Hypothesis The overall hypothesis tested in this thesis work was that: Increased knowledge of the microorganisms involved in hydrolysis of lignocellulosic materials can enable biogas production to be steered towards a more efficient and controllable process for degradation and biogas production from agriculture residues and plant-based materials.

1.2 Aim The main aim of this thesis was to provide novel information about lignocellulose-degrading bacteria in biogas processes and thereby enable a more efficient biogas production from lignocellulosic materials.

Specific objectives of the work described in Papers I-IV were to: 1. Search for correlations between the degradation rate of cellulose and

straw and the bacterial community structure, including potential cellulose-degrading bacteria (I).

2. Investigate the importance of the inoculum source for efficient biogas

production from lignocellulosic materials in a continuously operated process and the dynamics of the microbial community shaped by the substrates and operating parameters used (II, III).

3. Examine the impact of adding an energy-rich co-substrate to anaerobic

reactors operating with different lignocellulosic based substrates, regarding the reactor performance and microbial community (IV).

19

Biogas is the name given to a biologically produced specific gas mixture mainly composed of methane (52-85%), carbon dioxide (14-48%) and some small quantities of nitrogen, oxygen, hydrogen, hydrogen sulphide, ammonia and hydrocarbons (C2-C7) and some traces of organic compounds of sulphur, chlorine, fluorine, silicon etc. (Zamorska-Wojdyła et al., 2012). Methane (CH4) is an energy-rich and economically valuable energy resource. Methane can be produced through anaerobic digestion, a complex microbiological process requiring the combined activity of several groups of microorganisms with different metabolic capacities (Schnürer, 2016). At least four different groups of microorganisms (i.e. performing hydrolysis, acidogenesis, acetogenesis and methanogenesis) are involved (Schnürer, 2016) (Figure 1).

The substrate fed to a biogas process, such as manure, crop residues, food wastes or municipal sewage sludge, is mainly composed of polysaccharides (such as starch, cellulose, hemicellulose, pectin etc.), proteins and lipids. Most of these complex organic compounds are too large for a single organism to bring into the cell for its metabolism. Thus, in the first degradation step, the compounds are degraded (hydrolysed) to soluble sugars, peptides, amino acids and fatty acids, by the action of extracellular enzymes produced by microorganisms (Adekunle & Okolie, 2015). In the second step, the fermenting bacteria use these monomers as carbon and energy sources in their metabolism and, as a result, they produce alcohols, organic acids, carbon dioxide, hydrogen, hydrogen sulphide and ammonia (sometimes called intermediate products). These compounds can then be utilised by acetogens in the third step, producing mainly acetic acid, hydrogen and carbon dioxide. In the last step, methanogens (archaea) use mainly acetate, formate, methyl compounds, hydrogen and carbon dioxide as carbon and energy sources, forming carbon dioxide and methane (biogas) as the final products. According to the known methanogenic pathways, these methanogens can be categorised as hydrogenotrophic methanogens, acetoclastic methanogens and methylotrophic

2 Biogas production

20

methanogens (Kleinsteuber, 2018). The hydrogenotrophic methanogens perform a very important role, as they ‘pull’ many of the preceding oxidation reactions, e.g. oxidation of acids. These oxidation reactions are endergonic under standard conditions and can only proceed at a low partial pressure of hydrogen, i.e. in the presence of hydrogenotrophic methanogens. The hydrogen and carbon dioxide produced during the acidogenesis and acetogenesis steps can be converted to acetate through homoacetogenesis, which can also affect the partial pressure of hydrogen (Ye et al., 2014; Collet et al., 2005). The conversion of acetate to methane can proceed through two different pathways, depending on prevailing environmental conditions such as ammonia and volatile fatty acid (VFA) level and temperature: 1) the acetoclastic pathway, which involves acetoclastic methanogens cleaving acetate into methane and carbon dioxide; and 2) the syntrophic acetate oxidation (SAO) pathway, where acetate is first metabolised into hydrogen and carbon dioxide by syntrophic acetate-oxidising bacteria (SAOB) and is later used by hydrogenotrophic methanogens for methane production (Westerholm et al., 2016; Schnürer et al., 1999; Zinder & Koch, 1984).

21

Figure 1. The anaerobic digestion process leading to biogas production. Organic materials are

first hydrolysed to soluble organic compounds such as amino acids, fatty acids and sugars (1.

Hydrolysis). Then, depending on different kinds of microorganisms, these soluble organic

compounds are converted to intermediate products such as alcohols and fatty acids (2.

Acidogenesis). In the next step, the intermediate products are utilised by acetogens to form

hydrogen (H2), carbon dioxide (CO2) and acetate (3. Acetogenesis and syntrophy). Finally,

methanogens consume mainly CO2, H2 and acetate to produce methane (CH4) and CO2 as the

metabolic end-products (4. Methanogenesis). Diagram adapted from Pap et al. (2016) and

Schnürer et al. (2016) (Pap & Maróti, 2016; Schnürer, 2016).

Acetate

Soluble organic compounds

H2, CO2

Complex organic material

Intermediate products

CH4; CO2

1

2

3

3

4 4

2

3

2 Bacteria

Archaea

22

2.1 Current status of biogas production in the EU and Sweden

Biogas production has been continually increasing in the EU and its member states for some years. By 2015, there were more than 17,400 biogas plants installed in the EU, producing in total 18 billion m3 methane (equal to ~654 PJ), which corresponded to 50% of global biogas production in 2015 (Scarlat et al., 2018). The biogas production situation and applications vary between EU countries from several perspectives, including: 1) the production sources (i.e. landfill gas, wastewater treatment, anaerobic digestion and thermochemical processes); 2) the feedstock used (i.e. energy crops, agricultural residues, biowaste and municipal waste, industrial waste, sewage etc.); and 3) the downstream usage of the biogas (i.e. for electricity, heat and transportation) (Scarlat et al., 2018). For example, in Germany biogas is mainly produced through the anaerobic digestion process using around half energy crops and half agricultural waste (calculation based on wet weight of material), while in Sweden it is mainly produced from sewage sludge at wastewater treatment facilities (Stambasky et al., 2016). In contrast to other EU countries, the biogas produced in Sweden is mainly upgraded to vehicle fuel (65%), while less is used for generation of electricity and heat (Johan & Linus, 2018). Sweden is a world leader in the use of upgraded biogas in the transportation sector, where the amount used corresponds to 75% of all biogas used for vehicles in Europe (Scarlat et al., 2018) (Figure 2).

Figure 2. Biogas buses refuelling at a biogas station (Photo: Anna Schnürer, Uppsala).

23

Moreover, the EU has adopted the target of increasing the share of renewables to at least 27% of final energy consumption by 2030 (final, 2014). This target could be achieved by contributions from further development of the biogas sector. The increase in biogas production in the EU after 2003 was achieved in the first instance by the development of anaerobic digestion processes (treating various organic materials, including energy crops), followed by landfill gas and biogas from wastewater treatment (Scarlat et al., 2018). Organic wastes, especially agricultural wastes, have been highlighted as having great potential for future biogas production (Scarlat et al., 2018; Meyer et al., 2017). Within the EU target for 2030, Sweden has a more specific target of reaching 49% renewables in final energy consumption and reducing use of fossil fuels in the transport sector by 80% from 2010 to 2030 (https://2030-sekretariatet.se/english/). The theoretical biogas production potential in Sweden has been calculated to be about 54 PJ/year, which is nearly seven times the current annual production of biogas (around 7.6 PJ/year). The agricultural waste sector has again been suggested to represent a major part of this potential (Meyer et al., 2017) (Figure 3).

Figure 3. Grass bedding mixed with cattle manure, an agricultural waste with high potential for biogas production.

25

When lignocellulosic materials are used as a substrate for anaerobic digestion, the first step, hydrolysis, usually becomes rate-limiting for the whole process, due to the recalcitrant structure of the plant cell wall (Mulat & Horn, 2018; Lynd et al., 2002a). Moreover, lignocellulosic materials are characterised by low nutrient content, giving low methane yield compared with other substrates, such as food and municipal wastes (Li et al., 2013; Chynoweth et al., 1993). To overcome this disadvantage of using lignocellulosic materials, many approaches have been suggested, involving both substrate optimisation (e.g. pre-treatment and co-digestion) and optimisation of process configuration (e.g. improved process design). These are discussed in more detail below.

3.1 Structure of lignocellulose Lignocellulose is widely present in plants in the form of microfibrils in the cell wall, which makes plants strong (Li et al., 2009). It is abundant in most kinds of plants, comprising e.g. around 100% in cotton flower parts (Bayané & Guiot, 2010) and around 40-50% in different agriculture residues (e.g. rice straw, rice husk, maize stalks etc.) (Gani & Naruse, 2007). In the linear structure of microfibrils, acetal bonds provide a strong binding force between each cellulose unit. Each linear cellulose strain interacts with the neighbouring strains forming a sheet structure, which is similar to the β-sheet structure in the DNA molecule. These cellulose strains are covered by hemicellulose, which has several branched glucose structures that are further reinforced by the mesh of lignin (Figure 4). Lignin is a complex aromatic structure that cannot be significantly degraded by microorganisms in the anaerobic environment (Prochazka et al., 2012). The rigid structure of the plant cell wall,

3 Lignocellulosic materials as a substrate for biogas production

26

lignocellulose, is almost unreachable by enzymes produced by microorganisms (Akin, 1988), thus restricting the degradation efficiency (Bayané & Guiot, 2010). Consequently, the hydrolysis rate is the main limitation in biogas production using lignocellulosic materials (Mulat & Horn, 2018; Noike et al., 1985).

Figure 4. Microstructure of a typical plant cell wall, indicating the relationship between cellulose, hemicellulose and lignin (Modified from https://www.total.com).

27

3.2 Lignocellulosic substrate optimisation As mentioned above, the biodegradability of lignocellulosic materials can be increased by a pre-treatment with the purpose of removing lignin, hydrolysing hemicellulose, decreasing cellulose crystallinity, increasing the porosity of materials and making the material more accessible to microbial and enzymatic attack (Monlau et al., 2013). Different pre-treatment methods for lignocellulosic materials have been explored, for example mechanical, thermal, chemical and biological methods (Monlau et al., 2013). However, most pre-treatment methods require expensive specialist equipment with substantial energy requirements. In addition, toxic products such as furfurals, 5-hydroxymethylfurfural (HMF), organic acids and phenols may be formed and cause inhibition of the microbial process (Sawatdeenarunat et al., 2015).

Lignocellulose-rich materials typically also have a high carbon to nitrogen (C/N) ratio, low levels of micronutrients and, often, a low energy content (Li et al., 2013). However, through co-digestion, the substrate mixture can be designed to optimise the composition of nutrients, balance the C/N ratio etc. and achieve higher methane yields (Ebner et al., 2016; Macias-Corral et al., 2008; Lehtomäki et al., 2007; Sosnowski et al., 2003). Many substrates have been tested for co-digestion in biogas production from lignocellulose-rich material. For example, lignocellulose-rich cattle manure has been evaluated in co-digestion with food waste (Awasthi et al., 2018; Ebner et al., 2016) and stillage (Westerholm et al., 2012) and co-digestion has been shown to give enhanced methane yield compared with mono-digestion of the manure. Process stability and volumetric biogas yield from lignocellulose-rich materials with a low C/N ratio, such as corn stovers (Li et al., 2014), switchgrass (Zheng et al., 2015) and other agricultural residues, have been shown to improve when these materials are co-digested with nitrogen-rich animal manure (Neshat et al., 2017; Zhang et al., 2013).

When diluted agricultural residues (such as liquid manure) are used, co-digestion with lignocellulosic materials can also be applied to achieve a higher organic loading rate (OLR), with only minor effects on hydraulic retention time (HRT). This is particularly important, as relatively long hydraulic retention time is typically needed for degrading lignocellulose-rich materials (Neshat et al., 2017; Mata-Alvarez et al., 2014). Positive effects, such as increased methane yield, of combining lignocellulose-rich agricultural substrates with various high-energy co-substrates, including protein- and sugar-rich materials, have been demonstrated in several different studies (Ahlberg-Eliasson et al., 2017; Neshat et al., 2017; Mata-Alvarez et al., 2014) and in this thesis (III, IV).

28

An issue to consider when selecting a co-substrate for lignocellulosic materials is that some co-materials can result in decreased degradation efficiency. For example, a negative effect of proteins on anaerobic digestion of carbohydrate-rich materials has been observed and has been attributed to high ammonia levels (Breure et al., 1986). Similar results are presented in this thesis, with negative effects, specifically on cellulose degradation, observed following high levels of ammonia release during degradation of proteins (I, II). A decrease in the degree of degradation efficiency and specific methane production was also observed in this thesis work when digesting lignocellulose-rich material with milled feed wheat, resulting in elevated ammonia levels (III). The low degree of degradation efficiency is unfavourable, as it represents a loss of energy and could potentially lead to higher methane emissions during digestate storage (Liebetrau et al., 2013); (III).

3.3 Anaerobic digestion process configurations for lignocellulose-rich material

Various configurations can be used for biogas production depending on the practical needs (e.g. different production purposes, characteristics of the feedstock etc.). Depending on the feeding frequency, biogas plants are generally categorised as batch, fed-batch or continuous processes (Schnürer, 2016).

In a batch process, all the materials are added at once and the four steps of the biogas production process proceed in one reactor at the same time. The advantages of the batch process are that it is cheap, easy to operate and allows nearly 100% degradation of the organic material in a substrate. However, the batch process usually requires a long time to digest the organic material and toxic compounds such as ammonia can accumulate, since the internal reactor contents are not exchanged during the process (Schnürer & Jarvis, 2018; Raposo et al., 2012). Batch-type processes are typically used for small-scale production of biogas, mainly in Asia, but are also common in Germany, especially for dry materials (total solids content >15%), which are often rich in lignocellulosic material (Kothari et al., 2014; Rajendran et al., 2012).

Batch processes are also commonly used in laboratory biogas trials, for example during determination of the biomethane potential (BMP) of a certain substrate (Schnürer et al., 2017) (I, II, III, IV). BMP value is usually high for substrates like food waste, vegetable oil, and cheese whey while substrates like agriculture and forest residuals, containing high levels of lignocellulose, have

29

lower biomethane potential (Labatut et al., 2011). The final biomethane potential and the degradation efficiency (time to reach the final biomethane potential) of a substrate can be used to guide the set-up of the biogas reactor. The biomethane potential can also be used to evaluate the importance of different inocula for the degradation of various materials (Perrotta et al., 2017; Elbeshbishy et al., 2012) (I, II, III) (Figure 5). For lignocellulose-rich materials, a significant difference in degradation was seen in this thesis depending on the characteristics of inoculum, with different physicochemical and microbial components (I, III) (De Vrieze et al., 2015b; Gu et al., 2014).

Figure 5. Sealed serum bottles on a rotary shaker in a biomethane potential (BMP) test.

In a fed-batch process, materials are added successively over time, which allows for a more constant rate of gas production and a higher level of dilution of any toxic compounds accumulated compared with the batch process (Lim & Shin, 2013). However, the amount of gas produced rises quickly at the start of feeding and decreases gradually over time and the digester needs to be filled and emptied at intervals, which causes irregular gas production compared with a continuous process (Li et al., 2011). Examples of using lignocellulosic material in a methanogenic fed-batch process are rare in the literature. However, one study found that using a fed-batch process gave a higher methane yield than a batch or semi-continuous process when degrading lignocellulosic material (grass and maize silage) with anaerobic sludge from pig slurry fermentation after supplementation of rumen anaerobic fungi (Prochazka et al., 2012).

30

A continuous process is the most commonly used method in industrial biogas production (Schnürer et al., 2017). The major advantages of a continuous process are that the substrate is added continuously or semi-continuously, in parallel with removal of the reactor contents, thus giving constant production of biogas. Continuously stirred tank reactors (CSTR) are often used for a continuous process (or a semi-continuous process) (Moestedt et al., 2014; Usack et al., 2012). Continuously stirred tank reactors can be applied at different scales from a few litres (laboratory-scale) to hundreds of cubic metres (commercial or full-scale) (Schnürer, 2016). Thus, a CSTR can be used as a laboratory or pilot test system before scaling up (Kaparaju et al., 2009; Kaparaju et al., 2008) (Figure 6). Previous studies have also shown similar process performance during laboratory-scale and full-scale operation (Westerholm et al., 2018; Grim et al., 2015; Moestedt et al., 2014). There are many studies on the use of CSTR with lignocellulosic materials, focusing on various research questions, e.g. comparisons of methane yield using different lignocellulosic feedstocks (Martínez-Gutiérrez, 2018), evaluations of the importance of seeded inoculum (II, III), effects of co-digestion (Li et al., 2014; Comino et al., 2012; Nges et al., 2012) (IV), impacts of pre-treatment (Carrere et al., 2016), process operating parameters (e.g. hydraulic retention time, organic loading rate and temperature) (Shi et al., 2017; Zhou et al., 2017; Risberg et al., 2013) or feeding strategy (Mauky et al., 2015) (III) and determination of lignocellulolytic microbes (Yu et al., 2018; Zhou et al., 2017; Sun et al., 2015; Qiao et al., 2013; Lissens et al., 2004) (II, III). Continuously stirred tank reactors have also been shown to give higher efficiency of lignocellulosic material degradation than other types of continuous reactor configurations, such as the leach bed-upflow anaerobic sludge blanket (USAB) (Fu & Hu, 2016; Nizami & Murphy, 2010).

Figure 6. Left: A series of continuously stirred tank reactors (CSTR). Right: SLU full-scale biogas plant at Lövsta (Photo: Anna Schnürer).

31

The anaerobic digestion process configuration can also be categorised by different process stages, e.g. single-stage, two-stage and multiple-stage processes (Achinas et al., 2017). In a single-stage process, all materials are digested in a single reactor and the four degradation steps in the biogas production process take place at the same time and in the same chemical environment. To achieve better biogas performance, the biogas process can also be set up as a multiple-stage system (Ward et al., 2008). In the example of a two-stage anaerobic digester, all four degradation steps proceed in both digesters, but the second digester is fed with the reactor contents from the first reactor (Parawira et al., 2008). This type of design allows two digesters to work with different operating parameters (such as temperature, agitation speed etc.) and allows the first stage to focus on hydrolysis and acidogenesis. It has been used for complex substrates, such as lignocellulose-based materials (Akobi et al., 2016; Ward et al., 2008). Higher methane concentration and greater efficiency can be achieved with a two-stage process design compared with a single-stage design (Colussi et al., 2013; Parawira et al., 2008; Taherzadeh & Karimi, 2008).

However, an obvious drawback of multiple-stage anaerobic digestion is the high cost compared with the one-stage process. Thus, there are few commercial multiple-stage anaerobic digestion systems for processing lignocellulose-based materials in operation today (Achinas et al., 2017). Multiple-stage anaerobic digestion processes (as opposed to multiple-phase anaerobic digestion systems) sometimes also include recirculation of reactor contents (e.g. from methanogenic phase to hydrolytic phase) (Azbar & Speece, 2001). This has been applied as an additional approach to optimise the degradation of lignocellulosic materials. For example, by recirculating the reactor contents in an anaerobic digestion process, it is possible to: 1) achieve a longer retention time and consequently the time available for degradation can be prolonged (Estevez et al., 2014); 2) reach optimal conditions for hydrolytic bacteria (in terms of pH, water content and alkalinity), which are better maintained in the process (Dandikas et al., 2018); and 3) preserve micronutrients (Aslanzadeh et al., 2013). Thus, several studies using this concept for degradation of lignocellulosic material have found increased methane yield compared with a non-recirculating reactor.

32

3.4 Process regulating parameters In addition to digester configuration, different operating parameters, such as organic loading rate (OLR), hydraulic retention time (HRT), temperature and stirring, need to be considered during set-up of an anaerobic digestion process (Schnürer, 2016). Important parameters are e.g. OLR and HRT, which are often interlinked so that a higher OLR usually leads to shorter HRT. The organic loading rate can be defined in kilograms or grams of volatile solids (VS) per day and cubic metre or litre of reactor volume. Overloading with organic materials may cause accumulation of volatile fatty acids (VFAs), as the methanogenic step cannot keep up with the acidogenic and acetogenic steps (Franke-Whittle et al., 2014). However, due to the slow degradation rate, the risk of VFA accumulation due to overloading is relatively low when using lignocellulosic material compared with when using e.g. sugar-rich or lipid-rich materials (Schnürer & Jarvis, 2018; Cavaleiro et al., 2009). A high load can also reduce the HRT, which, as mentioned above, can have a negative impact on the degradation efficiency. Hydraulic retention time is defined as the time that the substrate remains in a digester. In a CSTR, the HRT can be approximated as volume of liquid phase divided by effluent flow rate. The HRT varies in different biogas digesters and normally ranges from 10 to 30 days, but is sometimes longer (Mao et al., 2015). The actual magnitude of the HRT applied is dependent on many different parameters, such as the characteristics of the input substrate and the operating temperature. Due to the intricate structure of lignocellulosic materials limiting the hydrolysis efficiency in an anaerobic digestion process, a comparatively long HRT (>30 days) is typically needed (Shi et al., 2017).

Another important parameter for the anaerobic digestion process is the operating temperature, where an appropriate temperature can potentially improve the methane production performance (Schnürer et al., 2017). For a digester operating with lignocellulosic materials, the operating temperature is usually set around 37 °C (Sawatdeenarunat et al., 2015) (I, II, III, IV). However, studies have shown that digestion of lignocellulosic materials at different temperatures is possible (Risberg et al., 2013; El-Mashad et al., 2004). A higher operating temperature has been suggested to give a higher hydrolysis rate of lignocellulosic material and even a higher methane yield (Moset et al., 2015; Labatut et al., 2014; Veeken & Hamelers, 1999). However, some studies have found no significant difference in performance and methane yield using the same substrate (manure/straw) at different temperatures (Risberg et al., 2013).

33

Feeding regime (e.g. different feeding intervals) is another parameter that could be used to achieve flexible and efficient biogas production in a continuous process (Mulat et al., 2016). However, the effect of feeding regime on biogas production performance can vary depending on the substrate used and the feeding interval (Piao et al., 2018; Ziels et al., 2018; Ziels et al., 2017; Mulat et al., 2016; Mauky et al., 2015; De Vrieze et al., 2013). Very few studies have investigated the effect of feeding regime on the degradation of lignocellulosic materials. In this thesis, a less frequent feeding regime that involved adding milled feed wheat as a co-substrate load all at once, compared with in two portions two hours apart, in CSTRs operating with a grass-manure mixture was found to give slightly higher cellulose conversion activity (III).

One additional parameter recently suggested to be of importance for the efficiency of degradation of lignocellulosic material and the final methane yield is the nature of the inoculum, including both physicochemical and microbial characteristics (Perrotta et al., 2017; De Vrieze et al., 2015a; De Vrieze et al., 2015b) (II, III). For example, in Paper I lower cellulose degradation efficiency was seen in batch processes seeded with inoculum taken from biogas plants fed with wheat-based stillage, slaughterhouse waste and grass, compared with inoculum from a process fed with mixed sludge. In Paper II, CSTRs operating with different inoculum sources showed a significant difference in degradation efficiency for a grass-manure mixture, especially in the initial phase of the process. In Paper III, it was concluded that the original inoculum can profoundly influence biogas production performance in the long term and affect microbial responses to process operation changes.

3.5 Process monitoring parameters

When an anaerobic digestion process is set up, parameters of the reactor contents, such as volatile fatty acid concentration, pH, alkalinity and ammonia level, and parameters of the gas phase, such as methane, carbon dioxide and hydrogen sulphide, are regularly monitored and can be used in combination to evaluate the biogas production performance (Schnürer et al., 2017). These parameters can be further subdivided into process efficiency measures, such as specific methane production (SMP), volatile solids reduction etc., and process stability measures, such as volatile fatty acid concentration, ammonia level etc.

Specific methane production is defined as the normalised volume of methane produced per gram of volatile solids in the substrate. A decreasing value of SMP for a substrate with a certain biomethane potential (BMP) may indicate less efficient substrate degradation (III). However, for a biogas plant,

34

the volumetric yield is often continuously recorded, while SMP might be less frequently considered. For example, decreased degradation efficiency caused by a recalcitrant substrate such as lignocellulosic materials can be masked by increased volumetric yield due to increased load, as shown in this thesis (III). Furthermore, several studies have shown that low efficient substrate degradation can increase the risk of methane emissions during storage of the digestate (i.e. residual methane potential, RMP), which is unfavourable from both an economic and an environmental perspective (Ahlberg-Eliasson et al., 2017; Ruile et al., 2015) (IV). This risk can be measured as the volatile solids (VS) reduction in reactor contents (i.e. VS of substrate compared with VS of the reactor contents). Moreover, a combination of BMP/RMP analysis has recently been proposed for the anaerobic digestion process, to better evaluate the degradation efficiency of the substrate (Li et al., 2017; Rico et al., 2015).

Volatile fatty acids are intermediate products produced during anaerobic digestion of organic compounds and the VFA concentration is considered one of the most important indicators for judging the stability of an anaerobic digestion process (see Chapter 2 of this thesis) (Drosg, 2013). Accumulation of VFAs can be caused by e.g. temperature fluctuations and substrate overloading (Schnürer et al., 2017; Ferguson et al., 2016). It can lead to a pH drop, which inhibits the methanogens and results in a decrease in methane production (Schnürer et al., 2017). When the rate of acidogenesis is higher than the rate of methanogenesis, acetate and propionate often accumulate more than other VFAs such as butyrate and valerate, as demonstrated in this thesis (II, III). A high propionate to acetate ratio can be used as an early indicator of a risk of process failure (Marchaim & Krause, 1993). Methods to remedy VFA accumulation in an anaerobic digestion process include reducing the organic loading rate, extending the hydraulic retention time and adding trace elements, aiming to rebalance the relative rate of the acidogenesis and methanogenesis steps (Choong et al., 2016; Ferguson et al., 2016; Moestedt et al., 2013).

The level of alkalinity indicates the buffering capacity within the anaerobic digestion process. When acids such as VFAs accumulate, the alkalinity typically shows a decrease before a pH drop (Drosg, 2013). Thus, the VFA/alkalinity ratio can be measured to monitor the stability of a reactor, especially when there is a high risk of acidification (Schnürer et al., 2017).

Ammonium is formed during the degradation of protein-rich materials. Free ammonia, in equilibrium with ammonium, is toxic to microbes and strongly inhibits the anaerobic digestion process (Westerholm et al., 2016) (I, II). However, a gradual increase in ammonia level permits development of ammonia-tolerant communities (Müller et al., 2016). Decreasing the temperature and the pH can push the equilibrium between ammonium (NH4+)

35

and ammonia (NH3) towards the ammonium side, and is thus often used to mitigate ammonia inhibition (Schnürer, 2016).

Another important factor is pH, which is affected by process parameters such as temperature, alkalinity, VFA concentration and ammonium level. (Fitamo et al., 2017; Shi et al., 2017; Franke-Whittle et al., 2014). Different microbes have different optimal growth pH ranges. For the acidogenic bacteria, a pH range down to 4.5-5.0 can be tolerated (Chandra et al., 2012), while the optimal pH range for methanogens is around 6.7-8.0. Thus, most single-stage biogas plants operate at around neutral pH to maximise the methanogenesis step (Schnürer et al., 2017).

When using lignocellulosic materials as the main substrate, ammonia/ammonium and VFAs are unlikely to accumulate due to the high C/N ratio, slow hydrolysis rate and relatively long hydraulic retention time and low organic loading rate applied in the anaerobic digestion process (Cavaleiro et al., 2009; Ward et al., 2008). However, process imbalances can still arise, as lignocellulosic materials are often combined with co-substrates, such as proteins, to balance the C/N ratio and improve the gas yield (Neshat et al., 2017) (III).

In addition, recent studies have suggested microbial monitoring as a possible way to evaluate and manage the process (Carballa et al., 2015; Lebuhn et al., 2015). By following the community dynamics or analysing specific key groups, such as the methanogens, and correlating their abundance to specific process parameters, it may be possible to predict instability or steer the community in a desired direction. These correlations are discussed in more detail in Chapter 4.

37

As mentioned in Chapter 2, biogas is produced by a complex network of microbes with differing and complementary metabolisms. Thus, to optimise and achieve better regulation of a biogas process, an in-depth understanding of the important microbial agents is needed (Kleinsteuber, 2018; Carballa et al., 2015; Lebuhn et al., 2015; Vanwonterghem et al., 2014b). In a typical methanogenic CSTR, members of the phyla Firmicutes and Bacteroidetes are often found to dominate the bacterial community, while members of the phylum Euryarchaeota tend to dominate the archaeal community (Güllert et al., 2016; Luo et al., 2016; Pore et al., 2016; Satpathy et al., 2016; Watanabe et al., 2016; Sun et al., 2015; Lu et al., 2014) (I, II, III, IV). However, some other bacterial phyla such as Proteobacteria, Chloroflexi and Fibrobacteres can also be abundant (Schnürer, 2016; Vanwonterghem et al., 2014a) as confirmed here (I, II, III, IV). Moreover, within the fungal community, the phylum Neocallimastigomycota has been shown to dominate (Dollhofer et al., 2015).

4 Microbial communities

38

Many recent studies have found that microbial communities can be shaped by the operating parameters of the anaerobic digestion process and can thus affect the biogas production performance (Grohmann et al., 2017; Pap & Maróti, 2016; Satpathy et al., 2016; Sun et al., 2016; Westerholm et al., 2016; De Francisci et al., 2015; Rui et al., 2015; Sundberg et al., 2013; Cardinali-Rezende et al., 2012; Kampmann et al., 2012). This was also demonstrated in Papers I-IV in this thesis (Figure 7).

Figure 7. Operating parameters affecting microbial community and thus potentially biogas production performance.

39

4.1 Lignocellulose degraders in the anaerobic environment

In studies using different isolation and molecular microbiological methods, various anaerobic lignocellulose degraders have been found in diverse anaerobic environments, including soil, anaerobic digesters, aquatic environments such as sludge and sediment, animal gut environments such as the rumen, termites, dung beetles, etc. (Saini et al., 2017; Azman et al., 2015; Dollhofer et al., 2015; Estes et al., 2013; Ransom-Jones et al., 2012; Morrison et al., 2009a; Lynd et al., 2002b; Leschine, 1995) (I, II, III, IV). These anaerobic lignocellulose degraders are widely distributed in genera within the bacteria and fungi domain, but have also been found recently in the archaea domain (Cragg et al., 2015).

Many types of anaerobic bacteria have been demonstrated to have the ability to degrade or potentially utilise lignocellulose as a carbon source. These can be found in genera such as Clostridium, Ruminococcus, Fibrobacter, Acetivibrio, Butyrivibrio, Halocella, Bacteroides, Spirochaeta, Thermotoga, Echinicola, Mahella, Marinilabilia, Prevotella, Flavobacterium and Streptomyces (Azman et al., 2015; Sun et al., 2013; Tsavkelova & Netrusov, 2012) (I, II, III) (Figure 8).

Figure 8. Scanning electron microscope (SEM) image of material isolated from an industrial-scale anaerobic digester, showing pure-cultured Clostridium sp. Bciso-3 degrading cellulose.

40

The relative abundance of these genera typically varies depending on the anaerobic environment. For example, the best-studied genus, Clostridium, has been found to be more abundant in landfilled sludge than genera such as Fibrobacter and Ruminococcus, but less abundant in the rumen (Ransom-Jones et al., 2012; Burrell et al., 2004). In anaerobic digestion processes operating with lignocellulosic materials as the main substrate, the relative abundance of different genera can also vary depending on factors such as the composition of the substrate, process configuration and operating parameters (Azman et al., 2015). However, the phyla Bacteroidetes and Firmicutes often dominate, followed by phyla such as Proteobacteria and Actinobacteria (Güllert et al., 2016; Azman et al., 2015; Sun et al., 2013). This was also the case in the anaerobic digestion processes studied in this thesis (I, II, III, IV). Recent studies using metatranscriptomics and metaproteomics approaches have revealed information on the active, lignocellulose degraders in the anaerobic digestion processes, rather than simply all microbes present. The results confirm the important roles of lignocellulose degradation by the genus Clostridium (Jia et al., 2018; Güllert et al., 2016; Lü et al., 2014). New knowledge on members of the genus Clostridium has also been used to guide the design of bioaugmentation strategies for improving the lignocellulose degrading efficiency and methane yield in different anaerobic digestion processes (Tsapekos et al., 2017; Poszytek et al., 2016).

For fungi, the best-studied anaerobic cellulase producers are members of the family Neocallimastigaceae, including the genera Neocallimastix, Orpinomyces and Piromyces (Cheng et al., 2018; Dollhofer et al., 2015; Viikari et al., 2009). These genera have been widely found in the gastrointestinal tract of ruminants and most non-ruminant herbivores (Dashtban et al., 2009), but have lately been identified also as part of the community in anaerobic digesters (Dollhofer et al., 2015). The first anaerobic lignocellulolytic fungus to be identified was Neocallimastix frontalis, isolated from sheep rumen fluid (Orpin, 1975; Braune, 1913). Later studies have demonstrated that members of the genera Neocallimastix, Orpinomyces and Piromyces are able to utilise different carbohydrates and produce hydrogen, carbon dioxide, acetate, formate, lactate and ethanol as metabolic end-products (Gruninger et al., 2014; Dashtban et al., 2009; Hodrová et al., 1998; Joblin & Naylor, 1989). Notably, these fungi can also develop an invasive rhizoid system that penetrates plant cell walls, combined with secretion of various carbohydrate-hydrolysing enzymes, thus improving the accessibility of plant structures to bacterial action (Dollhofer et al., 2015). Moreover, an ability of anaerobic fungi to degrade lignin has been reported in several studies (Dollhofer et al., 2015; Gruninger et al., 2014; Haitjema et al., 2014; Kazda et

41

al., 2014). This suggests the potential to enhance degradation of lignocellulosic material for biogas production by enhancing the growth of these fungi.

A few species of hyperthermophilic archaea belonging to the genus Pyrococcus, such as Pyrococcus furiosus, Pyrococcus horikoshii and Pyrococcus glycovorans, have also been found to produce endoglucanases such as glycoside hydrolase (GH) families 5 and 12 (Kishishita et al., 2015; Ando et al., 2002; Barbier et al., 1999). These archaea can live under extremely high temperatures (around 100 °C) and in high-salt environments, and could thus potentially be applied in a pre-treatment step before biogas production.

4.2 Enzymatic depolymerisation of cellulose and hemicelluloses

Lignocellulose is degraded by the collective action of multiple carbohydrate-active enzymes, including glycoside hydrolases, produced by microorganisms (Jia et al., 2018; Young et al., 2018; Cragg et al., 2015; Malherbe & Cloete, 2002). The glycoside hydrolases are classified based on amino acid sequence similarities and grouped into different enzyme families, such as GH 5, 6, 7, 8, 9, 10, 11, 12, 26, 44, 45, 48, 51, 60, 61 and 74 (Henrissat, 1991). Notably, most cellulases secreted by the anaerobic cellulose-degrading bacteria belong to GH families 5, 9 and 48 (endo-β-1,4-glucanase) (Vanwonterghem et al., 2016; Pereyra et al., 2010) (Figure 9).

Figure 9. Structure of the cellulose chain.

Depending on the environment (aerobic/anaerobic), the strategy used by microbes for cellulose degradation is somewhat different (Tomme et al., 1995). In the aerobic environment, fungi (such as the phyla Ascomycetes and Basidiomycetes) and bacteria (such as the genera Cellulomonas, Cellvibrio and

β-(1,4)

42

Cytophaga) typically use non-complexed cellulase systems, which secrete cellulase-hydrolysing enzymes (Malherbe & Cloete, 2002; Mullings & Parish, 1984). However, in the anaerobic environment, fungi (such as the family Neocallimastigaceae) and bacteria (such as the genus Clostridium) typically contain a relatively more complex cellulase system, including a membrane-bound enzyme complex (cellulosome) (Gruninger et al., 2014; Pereyra et al., 2010).

The cellulosome assists in the degradation process by synchronising different type of enzymes performing different reactions (Bayer et al., 2004). A typical cellulosome contains a scaffolding protein chain (without enzymatic activity), which has many enzyme binding domains, named cohesions. There are also different types of cohesins, e.g. Clostridium thermocellum has two, type I and type II. A corresponding domain on glycoside hydrolases, called dockerin, can selectively bind to the type-I cohesins of the primary scaffolding protein CipA. The terminal X-dockerin dyad of CipA can then bind to three types of type-II cohesins of anchoring scaffoldings, named SdbA, Orf2p and OlpB. These three types of type-II cohesins bind to the cell surface with an S-layer homology module (Bayer et al., 2008; Boisset et al., 1999; Bayer et al., 1998). Cellulose is bound by the carbohydrate binding module (CBM) on the scaffolding protein chain, which results in linkage of the cellulosic material and the cellulosome complex (Shoseyov et al., 2006).

In addition, recent studies have regrouped glycoside hydrolase family GH 61 and carbohydrate binding module CBM33 into a new family due to their capacity for catalysing oxidative cleavage of polysaccharides. This new family, which is called lytic polysaccharide monooxygenases (LPMOs) (Horn et al., 2012), has been found in fungi, bacteria and viruses (Chiu et al., 2015; Kohler et al., 2015; Vaaje-Kolstad et al., 2010). These enzymes have been demonstrated to specifically break and loosen the polysaccharide chains, which creates new attack points for cellobiohydrolases (CBHs), thus increasing the accessibility of cellulose to microorganisms (Johansen, 2016; Hemsworth et al., 2015). It is known that LPMOs require molecular oxygen (O2) for their activity (Johansen, 2016). However, a recent study has shown that hydrogen peroxide (H2O2) can act as a co-substrate instead of molecular oxygen, which suggests that LPMOs can work under anaerobic conditions (Bissaro et al., 2016). However, so far these enzymes have not been shown to be present in an anaerobic environment.

43

4.3 Lignocellulolytic communities and influence of digester parameters

In an anaerobic digestion process, lignocellulose degradation is usually not performed by a single fungus or bacterium, but by a complex microbial community (Jia et al., 2018; Young et al., 2018; Pereyra et al., 2010). In this thesis and in other studies, the composition and structure of the lignocellulolytic community (as part of the overall microbial community) has been shown to be influenced by process parameters such as temperature, volatile fatty acid concentration and ammonia content (Jia et al., 2018; Sun et al., 2013) (I, II, III).

Temperature is one of the most important factors shaping microbial communities. At different temperatures, community structure, diversity and the activity of microorganisms all vary and the stability of the reactor is then highly dependent on the resilience of the microbial community (Westerholm et al., 2018; Westerholm et al., 2017; Luo et al., 2016; De Vrieze et al., 2015c; Westerholm et al., 2015). Typically, higher operating temperature results in higher relative abundance of the phylum Firmicutes than of the phylum Bacteroidetes and lower microbial diversity compared with operation in mesophilic conditions (Westerholm et al., 2017; Luo et al., 2016; Westerholm et al., 2015; Moestedt et al., 2014). Studies specifically focusing on the response of the lignocellulolytic community to temperature changes in the anaerobic digestion process are rare. However, changing the operating temperature from 39 to 50 °C was shown to increase the ratio of Firmicutes to Bacteroidetes in a pilot-scale biogas reactor operating with rice straw (Yu et al., 2018). In another study, higher temperature (55 °C compared with 37 °C) resulted in an increase in the relative abundance of an uncultured order MBA08 (class Clostridia) and a decrease in community diversity in a CSTR process operating with steam-exploded straw and manure (Sun et al., 2015). Furthermore, temperatures below 4 °C and above 50 °C have been demonstrated to strongly decrease the degree of adhesion between bacteria and cellulose, thus potentially lowering the cellulose degradation efficiency (Miron et al., 2001).

Volatile fatty acid content has been shown to inhibit microbial groups to different degrees (Ma et al., 2015; Chen et al., 2014; Franke-Whittle et al., 2014). For the lignocellulolytic community, a negative correlation was seen in Paper I between the VFA content of the inoculum and the relative abundance of potential cellulose degraders, such as Mahella australiensis 50-1 BON and Echinicola vietnamensis. In Paper III, a decrease in the degrading efficiency of cellulose was also found to be associated with an increase in acetate and

44

propionate content. However, the correlation seen between VFA content and cellulose degradation could possibly be an indirect effect of ammonia inhibition, which often gives rise to accumulation of VFAs (III).

Ammonia level, combined with temperature, also significantly affects microbial community structure, both the overall structure (Hu et al., 2017; De Vrieze et al., 2015c) (I) and that of specific groups of microorganisms, such as the community of acetogenic bacteria (Moestedt et al., 2016), syntrophic acetate-oxidising bacteria (SAOB) (Müller et al., 2016) and the methanogens (Westerholm et al., 2016). At present, there is little information available in the literature regarding the impact of ammonia on the degradation of lignocellulose and on lignocellulolytic bacteria. However, in this thesis work, ammonia level was shown to be negatively correlated with the relative abundance of specifically C. cellulolyticum (I) in a batch process and with the cellulose degradation efficiency in both a batch (I) and semi-continuous process (II).

45

Various methods can be employed to study microbial community structure. These are normally categorised into: 1) culture-dependent techniques, including e.g. clone library, isolation and characterisation; or 2) culture-independent techniques. The culture-independent techniques can be further categorised based on different study purposes into: i) identification (cloning library, denaturing gradient gel electrophoresis (DGGE), terminal restriction fragment length polymorphism (T-RFLP), Sanger sequencing, microarray, next-generation sequencing etc.); ii) community dynamics changes (DGGE, T-RFLP, single-stranded conformation polymorphism (SSCP), Sanger sequencing, next-generation sequencing etc.); iii) quantification (quantitative polymerase chain reaction (q-PCR), fluorescence in situ hybridisation (FISH) etc.) (I, II, III, IV) functionality (advanced FISH, stable isotope probing (SIP), metatranscriptomics, metaproteomics etc.) (Cabezas et al., 2015). Alternatively, these techniques can be classified according to the level of gene products recovered from e.g. transcripts and proteins into: metagenomics, metatranscriptomics and metaproteomics) (Hassa et al., 2018; Kameshwar & Qin, 2018; Kleinsteuber, 2018; Aguiar-Pulido et al., 2016; Gawad et al., 2016; Goswami et al., 2016; Prince et al., 2014; Fry, 2004).

5 Microbial community analysis techniques

46

5.1 Methods applied to study lignocellulolytic communities

Isolation and cultivation of pure culture is a very important way to study a lignocellulolytic microorganism. Many potential lignocellulolytic microorganisms, including aerobic and anaerobic fungi and bacteria, have been isolated from various environments such as termites, rumen, paper mill, manure, wood fermenter, anaerobic digestion process etc. (Pereyra et al., 2010; König et al., 2006; Schwarz, 2001). Anaerobic bacteria can be cultivated in an anaerobic medium, using the following preparation steps: boiling the medium (to reduce the amount of oxygen), adding reducing agents, adding a substrate such as cellulose, cellobiose, filter paper etc. (to enrich lignocellulolytic microorganisms), and exchange of gas phase in the bottle to nitrogen gas (N2) or N2/CO2 (Westerholm et al., 2010). Isolation often starts with enrichment of lignocellulolytic microorganisms, followed by e.g. use of the agar shake method to pick single colonies from a dilution series from the previously enriched culture (Sun, 2015) (Figure 10).

Figure 10. Left: Anaerobic serum bottles of lignocellulolytic microorganism-enriched culture. Right: Anaerobic glass tubes with agar for picking single colonies.

Besides the culture-based method, some molecular tools have been employed to study lignocellulolytic communities. As mentioned in section 3.2, lignocellulose is degraded by different glycoside hydrolases, which are grouped in different families based on amino acid similarities (Henrissat, 1991). Based on this information, Pereyra et al. (2010) designed a degenerated primer set to specifically target the major glycoside hydrolase genes (cel5 and cel48) in anaerobic digestion processes (Pereyra et al., 2010). These primers were further adapted to quantitative PCR (qPCR) and revealed dynamic changes in these genes in two different biogas reactors (Pereyra et al., 2010). However, using qPCR can only show the overall changes in cel5 and cel48 genes in the reactor samples, and not the microbial community structures

47

containing these genes. Thus, the same primer sets were used in studies by Sun et al. (2013) and in work performed in this thesis (I, II, III), where the analysis was combined with T-RFLP and Sanger sequencing of clone libraries. These studies successfully revealed the population and structure of the potential lignocellulolytic degraders in anaerobic digestion processes set up with different inoculum sources and operated with agricultural substrates. The method of combining T-RFLP and sequencing of clone libraries has been widely used to study the microbial community structure in different ecosystems (Theuerl et al., 2018; Ramos et al., 2010; Dickie & FitzJohn, 2007; Wang et al., 2004). However, there are some limitations to this method. For example: 1) the principle behind T-RFLP is that the length of terminal restriction fragments (T-RFs) should vary with various microorganisms and restriction enzyme(s) used (Liu et al., 1997). However, one T-RF can represent several operational taxonomic units (OTUs) if they have the same number of bases at the first cutting site from the restriction enzyme; and 2) the community diversity is limited by the sequenced number of clones. These disadvantages can be somewhat mitigated by increasing the number of sequenced clones and using different enzymes in combination for the cutting. However, the method fails to provide as high resolution of the microbial community as next-generation sequencing.

Next-generation sequencing has been wildly applied for microbial community studies due to the advantages of including a high number of sequences per reaction, high max parallelisation and high throughput compared with Sanger sequencing (Ansorge, 2009; Morozova & Marra, 2008). Several recent studies, included Papers I-III in this thesis, have used different next-generation sequencing technologies (e.g. Roche454, illumina (Solexa), Ion Torrent and SOLiD) to scan the potential lignocellulolytic communities in e.g. dung beetles, termites, manure, anaerobic digestion processes fed with lignocellulosic materials etc. The aim of these studies has been either to identify previously undiscovered lignocellulolytic degraders or to investigate the correlation between the lignocellulolytic communities and the performance of an anaerobic digestion process (Ahlberg‐Eliasson et al., 2018; Chew et al., 2018; Vanwonterghem et al., 2016; Azman et al., 2015; Estes et al., 2013; Xia et al., 2013) (I, II, III) .

In addition, next-generation sequencing has been applied in functional genomics studies relating to lignocellulolytic degraders. For example, Wei et al. (2015) and Wang et al. (2015) first sequenced DNA samples extracted from a mesophilic and thermophilic biogas digester, respectively, using the GSFLX sequencing system (Roche 454). They recovered several novel glycoside hydrolase genes from these metagenome datasets and heterologously expressed

48

these genes in Escherichia coli to study their biochemical characteristics (Wang et al., 2015; Wei et al., 2015). Vanwonterghem et al. (2016) used the Illumina HiSeq platform and a gene-centric metagenomic approach to compare the glycoside hydrolase profiles over time in different anaerobic digestion environments (Vanwonterghem et al., 2016). These studies greatly expanded existing knowledge of possible application of the glycoside hydrolases and novel lignocellulolytic degraders, especially rare and uncultured species.

Moreover, when pure isolates are obtained, metagenome assembly and binning studies can be complemented by single-cell genomics with the help of next-generation sequencing (Yilmaz & Singh, 2012). Single-cell genomics can be used to assemble the genome of a bacterial species that is present at relatively low abundance in a metagenomics sample, or the genomes of completely unknown microorganisms (Gawad et al., 2016). For example, complete genome sequencing of the cellulolytic anaerobic bacteria Herbivorax saccincola Type Strain GGR1 and Herbinix luporum SD1D is reported by Alexander et al. (2018) and Daniela et al. (2016), respectively. Their results revealed the presence of abundant carbohydrate-active enzymes (CAZymes) in these two bacteria (Pechtl et al., 2018; Koeck et al., 2016).

In recent studies, there has been an increasing trend for employing combined meta-omics methods, including metagenomics, metatranscriptomics and metaproteomics, to analyse lignocellulolytic communities (Kleinsteuber, 2018). For example, Güllert et al. (2016) compared microbial community structure by: i) 16S rRNA gene tag sequencing (using the Roche 454 platform) and ii) taxonomic origin of the cellulolytic glycoside hydrolase genes retrieved by the metagenomic data (using the Illumina HiSeq 2500 platform). The results indicate differences in cellulose degradation efficiency between biogas fermenter contents, elephant faeces and cow rumen fluid, possibly caused by differences in amount of transcribed cellulase (Güllert et al., 2016).

Jia et al. (2018) reconstructed 107 population genomes from enrichment cultures and found only one sub-group to be highly transcribed in the metatranscriptomes. For the cellulose degraders, different genes were seen to be activated under different culture conditions. These findings deepen understanding of the relationship between a microbial population and the functional roles of active players in cellulosic biomass degradation (Jia et al., 2018).

Furthermore, metaproteomics have been applied to study the metabolic activity of the lignocellulolytic communities by extracting total proteins, which are then digested with e.g. trypsin, followed by liquid chromatography-mass spectrometry (LC-MS) analysis (Heyer et al., 2013). In a study combining metagenomics and metaproteomics, Hanreich et al. (2013) found that the

49

phylum Firmicutes seemed to play a major role for cellulose degradation, even though a fewer glycoside hydrolase genes were detected than in the phylum Bacteroidetes (Hanreich et al., 2013). Moreover, a comparison of the taxonomic community structure recovered from a metaproteomic dataset and 16S rRNA gene tag pyrosequencing, together with fluorescent in situ hybridisation analyses, has revealed detailed lignocellulolytic functions in Caldicellulosiruptor spp. and the key role of Clostridium thermocellum for cellulose degradation (Lü et al., 2014).

However, the use of metaproteomics to study lignocellulosic degradation groups in anaerobic digestion samples is still challenging in many ways (Heyer et al., 2013). For example, the identification of proteins largely relies on the metagenomic database (Speda et al., 2017b). The most used protein database, Swiss-Prot from the Universal Protein Resource (UniProt), contains around only 558 898 reviewed and annotated entries (last visited December 12, 2018), and most of these entries are not for bacteria and archaea. To overcome this problem, metaproteomic analysis can be performed based on a metagenome dataset recovered from the same samples (Hanreich et al., 2013; Rademacher et al., 2012). Another limitation is the sample complexity. To get high resolution in protein identification (i.e. identify as many proteins as possible), the extraction process needs to remove impurities such as humic organic matter, lipids, granules etc. (Keller & Hettich, 2009; Maron et al., 2007; Hofman-Bang et al., 2003). In addition, lignocellulosic bacteria are usually tightly attached to the fibres of biomass. Thus, the protein extraction method needs to be optimised in this regard to mitigate the loss of lignocellulosic bacteria. Several extraction methods have been tested in order to improve the protein yield from anaerobic digestate (Speda et al., 2017a). The biases that can be introduced by using different databases and purification methods have been evaluated in an ongoing work (not included in this thesis). Preliminary results showed a significant improvement on the quality of identified proteins by using customised metagenomic database and purification method. However, obtaining specific microbial proteins from highly redundant and abundant environmental protein pools remains a great challenge (Heyer et al., 2015).

51

Lignocellulosic materials, especially lignocellulosic residues, represent an important class of biomass that has not yet been fully utilised. Anaerobic digestion is believed to be one of the most feasible and economical tools for extracting the ‘hidden’ energy in lignocellulosic materials. Globally, several billion cubic metres of methane are produced yearly and demand is growing. The potential of using lignocellulosic materials to expand future production to meet this demand is impossible to ignore. However, the degradation efficiency of lignocellulosic materials in the anaerobic digestion process is still far from optimal. To increase use of lignocellulosic materials for methane production, a deeper understanding of the key agents in the degradation process, lignocellulosic microbes, is essential.

This thesis revealed the importance of the original inoculum for methane production using cellulose and wheat straw in a batch digestion system and also for the performance during start-up of semi-continuous stirred tank reactor (CSTR) processes. The microbial and chemical composition of the original inoculum sources was also revealed to influence the degradation of lignocellulose during long-term operation of CSTRs. Moreover, a positive correlation between the cellulose degradation rate of wheat straw and the level of Clostridium cellulolyticum was observed, indicating the possibility for steering the biogas production process to become more efficient by using a microbial approach. However, ammonia level appears to be one of the most important factors regulating the methane production performance of processes using lignocellulosic materials, possibly because it is a strong parameter shaping the microbial community structure and also the potential cellulose-degrading bacterial community. Lignocellulose-rich materials are often co-digested with energy-rich materials such as proteins in order to improve the C/N ratio. The data presented in this thesis suggest that degradation of proteins, giving high ammonia levels and high volatile fatty acid levels, results in

6 Conclusions and future perspectives

52

decreased lignocellulose efficiency. However, this decreased efficiency can be masked by increased volumetric yield due to increased load and higher energy content of the co-substrate. A low substrate degradation rate can potentially increase the risk of residual methane emissions during storage of the reactor digestate before use as a fertiliser.

The picture of the anaerobic lignocellulolytic microbial community is still far from complete. One important component of that community not covered in this thesis is the anaerobic fungi. Studies have shown that anaerobic fungi play an active role in lignocellulose degradation, even though their relative abundance in the overall microbial community is often low.

Furthermore, in this thesis only genomics-based analyses were performed and these are not sufficient to describe the anaerobic lignocellulolytic microbial community. Additional analyses relating to functions (e.g. proteomics and transcriptomics) are needed to fully identify the lignocellulose-degrading community and how to optimise it. Fortunately, with the rapid development in analytical methods and techniques and the corresponding growing databases, the cost of using transcriptomics- and proteomics-based approaches is becoming cheaper. When the complete guild is identified and a comprehensive and elaborate map of the lignocellulolytic microbial community becomes available, a customised inoculum adapted for each specific digestion task can be designed. This will help maximise methane production from the highly abundant lignocellulosic materials available world-wide.

53

Achinas, S., Achinas, V. & Euverink, G.J.W. (2017). A technological overview of biogas production from biowaste. (Engineering, 3).

Adekunle, K.F. & Okolie, J.A. (2015). A review of biochemical process of anaerobic digestion. Advances in Bioscience and Biotechnology, 6(03), p. 205.

Agency, I.E. (2015). World energy outlook special report 2015: Energy and climate change. (Final report. OECD/IEA, Paris: International Energy Agency.

Aguiar-Pulido, V., Huang, W., Suarez-Ulloa, V., Cickovski, T., Mathee, K. & Narasimhan, G. (2016). Metagenomics, Metatranscriptomics, and Metabolomics Approaches for Microbiome Analysis: Supplementary Issue: Bioinformatics Methods and Applications for Big Metagenomics Data. Evolutionary Bioinformatics, 12, p. EBO. S36436.

Ahlberg-Eliasson, K., Nadeau, E., Levén, L. & Schnürer, A. (2017). Production efficiency of Swedish farm-scale biogas plants. Biomass and Bioenergy, 97, pp. 27-37.

Ahlberg‐Eliasson, K., Liu, T., Nadeau, E. & Schnürer, A. (2018). Forage types and origin of manure in codigestion affect methane yield and microbial community structure. Grass and Forage Science, 0(0).

Akin, D. (1988). Biological structure of lignocellulose and its degradation in the rumen. Animal Feed Science and Technology, 21(2), pp. 295-310.

Akobi, C., Yeo, H., Hafez, H. & Nakhla, G. (2016). Single-stage and two-stage anaerobic digestion of extruded lignocellulosic biomass. Applied Energy, 184, pp. 548-559.

Ando, S., Ishida, H., Kosugi, Y. & Ishikawa, K. (2002). Hyperthermostable endoglucanase from pyrococcus horikoshii. Appl Environ Microbiol, 68(1), p. 430.

Angelidaki, I., Karakashev, D., Batstone, D.J., Plugge, C.M. & Stams, A.J. (2011). Biomethanation and its potential. Methods Enzymol, 494(16), pp. 327-351.

Ansorge, W.J. (2009). Next-generation DNA sequencing techniques. N Biotechnol, 25(4), pp. 195-203. Appels, L., Lauwers, J., Degrève, J., Helsen, L., Lievens, B., Willems, K., Van Impe, J. & Dewil, R.

(2011). Anaerobic digestion in global bio-energy production: potential and research challenges. Renewable and Sustainable Energy Reviews, 15(9), pp. 4295-4301.

Aslanzadeh, S., Rajendran, K., Jeihanipour, A. & Taherzadeh, M.J. (2013). The effect of effluent recirculation in a semi-continuous two-stage anaerobic digestion system. Energies, 6(6), pp. 2966-2981.

Awasthi, S.K., Joshi, R., Dhar, H., Verma, S., Awasthi, M.K., Varjani, S., Sarsaiya, S., Zhang, Z. & Kumar, S. (2018). Improving methane yield and quality via co-digestion of cow dung mixed with food waste. Bioresource Technology, 251, pp. 259-263.

Azbar, N. & Speece, R.E. (2001). Two-phase, two-stage, and single-stage anaerobic process comparison. Journal of Environmental Engineering, 127(3), pp. 240-248.

Azman, S., Khadem, A.F., Lier, J.B., Zeeman, G. & Plugge, C.M. (2015). Presence and role of anaerobic hydrolytic microbes in conversion of lignocellulosic biomass for biogas production. Crit Rev Environ Sci Technol, 45.

Barbier, G., Godfroy, A., Meunier, J.-R., Quérellou, J., Cambon, M.-A., Lesongeur, F., Grimont, P.A. & Raguénès, G. (1999). Pyrococcus glycovorans sp. nov., a hyperthermophilic archaeon isolated from the East Pacific Rise. International Journal of Systematic and Evolutionary Microbiology, 49(4), pp. 1829-1837.

References

54

Bayané, A. & Guiot, S.R. (2010). Animal digestive strategies versus anaerobic digestion bioprocesses for biogas production from lignocellulosic biomass. Reviews in Environmental Science and Bio/Technology, 10(1), pp. 43-62.

Bayer, E.A., Belaich, J.P., Shoham, Y. & Lamed, R. (2004). The cellulosomes: multienzyme machines for degradation of plant cell wall polysaccharides. Annual Review of Microbiology, 58, pp. 521-54.

Bayer, E.A., Chanzy, H., Lamed, R. & Shoham, Y. (1998). Cellulose, cellulases and cellulosomes. Current Opinion in Structural Biology, 8(5), pp. 548-557.

Bayer, E.A., Lamed, R., White, B.A. & Flint, H.J. (2008). From Cellulosomes to Cellulosomics. Chem Rec, 8(6), pp. 364-377.

Bissaro, B., Rohr, A.K., Skaugen, M., Forsberg, Z., Horn, S.J., Vaaje-Kolstad, G. & Eijsink, V. (2016). Fenton-type chemistry by a copper enzyme: molecular mechanism of polysaccharide oxidative cleavage. bioRxiv.

Boisset, C., Chanzy, H.D., Henrissat, B., Schulein, M. & Bayer, E.A. (1999). Hydroysis of model cellulose systems by cellulosomes and fungal cellulases. Abstracts of Papers of the American Chemical Society, 217, pp. U252-U252.

Borjesson, P. & Mattiasson, B. (2008). Biogas as a resource-efficient vehicle fuel. Trends Biotechnol, 26(1), pp. 7-13.

Bozan, M., Akyol, C., Ince, O., Aydin, S. & Ince, B. (2017). Application of next-generation sequencing methods for microbial monitoring of anaerobic digestion of lignocellulosic biomass. Appl Microbiol Biotechnol, 101(18), pp. 6849-6864.

Braune, R. (1913). Untersuchungen uber die im Wiederkauermagen vorkommenden Protozoen: Friedrich-Wilhelms-Universitat zu Berlin.

Breure, A., Mooijman, K. & Van Andel, J. (1986). Protein degradation in anaerobic digestion: influence of volatile fatty acids and carbohydrates on hydrolysis and acidogenic fermentation of gelatin. Appl Microbiol Biotechnol, 24(5), pp. 426-431.

Burrell, P.C., O'Sullivan, C., Song, H., Clarke, W.P. & Blackall, L.L. (2004). Identification, detection, and spatial resolution of Clostridium populations responsible for cellulose degradation in a methanogenic landfill leachate bioreactor. Appl Environ Microbiol, 70(4), pp. 2414-9.

Cabezas, A., de Araujo, J., Callejas, C., Galès, A., Hamelin, J., Marone, A., Sousa, D., Trably, E. & Etchebehere, C. (2015). How to use molecular biology tools for the study of the anaerobic digestion process? Reviews in Environmental Science and Bio/Technology, pp. 1-39.

Carballa, M., Regueiro, L. & Lema, J.M. (2015). Microbial management of anaerobic digestion: exploiting the microbiome-functionality nexus. Curr Opin Biotechnol, 33, pp. 103-111.

Cardinali-Rezende, J., Colturato, L.F., Colturato, T.D., Chartone-Souza, E., Nascimento, A.M. & Sanz, J.L. (2012). Prokaryotic diversity and dynamics in a full-scale municipal solid waste anaerobic reactor from start-up to steady-state conditions. Bioresource Technology, 119, pp. 373-383.

Carrere, H., Antonopoulou, G., Affes, R., Passos, F., Battimelli, A., Lyberatos, G. & Ferrer, I. (2016). Review of feedstock pretreatment strategies for improved anaerobic digestion: From lab-scale research to full-scale application. Bioresour Technol, 199, pp. 386-97.

Cavaleiro, A.J., Salvador, A.F., Alves, J.I. & Alves, M. (2009). Continuous high rate anaerobic treatment of oleic acid based wastewater is possible after a step feeding start-up. Environ Sci Technol, 43(8), pp. 2931-6.

Chandra, R., Takeuchi, H. & Hasegawa, T. (2012). Methane production from lignocellulosic agricultural crop wastes: A review in context to second generation of biofuel production. Renewable and Sustainable Energy Reviews, 16(3), pp. 1462-1476.

Chen, J.L., Ortiz, R., Steele, T.W.J. & Stuckey, D.C. (2014). Toxicants inhibiting anaerobic digestion: A review. Biotechnol Adv, 32(8), pp. 1523-1534.

Cheng, Y., Shi, Q., Sun, R., Liang, D., Li, Y., Li, Y., Jin, W. & Zhu, W. (2018). The biotechnological potential of anaerobic fungi on fiber degradation and methane production. World J Microbiol Biotechnol, 34(10), p. 155.

Chew, Y.M., Lye, S., Md. Salleh, M. & Yahya, A. (2018). 16S rRNA metagenomic analysis of the symbiotic community structures of bacteria in foregut, midgut, and hindgut of the wood-feeding termite Bulbitermes sp. Symbiosis, 76(2), pp. 187-197.

Chiu, E., Hijnen, M., Bunker, R.D., Boudes, M., Rajendran, C., Aizel, K., Oliéric, V., Schulze-Briese, C., Mitsuhashi, W., Young, V., Ward, V.K., Bergoin, M., Metcalf, P. & Coulibaly, F. (2015). Structural basis for the enhancement of virulence by viral spindles and their in vivo crystallization. Proceedings of the National Academy of Sciences, 112(13), p. 3973.

55

Choong, Y.Y., Norli, I., Abdullah, A.Z. & Yhaya, M.F. (2016). Impacts of trace element supplementation on the performance of anaerobic digestion process: A critical review. Bioresource Technology, 209, pp. 369-379.

Chynoweth, D.P., Turick, C.E., Owens, J.M., Jerger, D.E. & Peck, M.W. (1993). Biochemical methane potential of biomass and waste feedstocks. Biomass and Bioenergy, 5(1), pp. 95-111.

Collet, C., Gaudard, O., Péringer, P. & Schwitzguébel, J.-P. (2005). Acetate production from lactose by Clostridium thermolacticum and hydrogen-scavenging microorganisms in continuous culture—Effect of hydrogen partial pressure. Journal of Biotechnology, 118(3), pp. 328-338.

Colussi, I., Cortesi, A., Gallo, V., Rubesa Fernandez, A. & Vitanza, R. (2013). Improvement of methane yield from maize silage by a two-stage anaerobic process. Chemical engineering transactions, 32(1).

Comino, E., Riggio, V.A. & Rosso, M. (2012). Biogas production by anaerobic co-digestion of cattle slurry and cheese whey. Bioresource Technology, 114, pp. 46-53.

Cragg, S.M., Beckham, G.T., Bruce, N.C., Bugg, T.D.H., Distel, D.L., Dupree, P., Etxabe, A.G., Goodell, B.S., Jellison, J., McGeehan, J.E., McQueen-Mason, S.J., Schnorr, K., Walton, P.H., Watts, J.E.M. & Zimmer, M. (2015). Lignocellulose degradation mechanisms across the Tree of Life. Current Opinion in Chemical Biology, 29, pp. 108-119.

Dandikas, V., Marín Pérez, C., Koch, K., Lebuhn, M. & Gronauer, A. (2018). Influence of digestate recirculation on a two-phase anaerobic digestion of maize silage.

Dashtban, M., Schraft, H. & Qin, W. (2009). Fungal bioconversion of lignocellulosic residues; opportunities & perspectives. Int J Biol Sci, 5(6), pp. 578-595.

De Francisci, D., Kougias, P.G., Treu, L., Campanaro, S. & Angelidaki, I. (2015). Microbial diversity and dynamicity of biogas reactors due to radical changes of feedstock composition. Bioresource Technology, 176, pp. 56-64.

De Vrieze, J., Gildemyn, S., Vilchez-Vargas, R., Jáuregui, R., Pieper, D.H., Verstraete, W. & Boon, N. (2015a). Inoculum selection is crucial to ensure operational stability in anaerobic digestion. Appl Microbiol Biotechnol, 99(1), pp. 189-199.

De Vrieze, J., Raport, L., Willems, B., Verbrugge, S., Volcke, E., Meers, E., Angenent, L.T. & Boon, N. (2015b). Inoculum selection influences the biochemical methane potential of agro-industrial substrates. Microb Biotechnol, 8(5), pp. 776-86.

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

De Vrieze, J., Verstraete, W. & Boon, N. (2013). Repeated pulse feeding induces functional stability in anaerobic digestion. Microbial Biotechnology, 6(4), pp. 414-424.

Dickie, I. & FitzJohn, R. (2007). Using terminal restriction fragment length polymorphism (T-RFLP) to identify mycorrhizal fungi: a methods review. Mycorrhiza, 17(4), pp. 259-270.

Do, T.H., Le, N.G., Dao, T.K., Nguyen, T.M.P., Le, T.L., Luu, H.L., Nguyen, K.H.V., Nguyen, V.L., Le, L.A., Phung, T.N., van Straalen, N.M., Roelofs, D. & Truong, N.H. (2018). Metagenomic insights into lignocellulose-degrading genes through Illumina-based de novo sequencing of the microbiome in Vietnamese native goats' rumen. J Gen Appl Microbiol, 64(3), pp. 108-116.

Dollhofer, V., Podmirseg, S.M., Callaghan, T.M., Griffith, G.W. & Fliegerová, K. (2015). Anaerobic fungi and their potential for biogas production. In: Biogas Science and Technology Springer, pp. 41-61.

Drosg, B. Process monitoring in biogas plants. In: Proceedings of IEA Bioenergy Task2013. Ebner, J.H., Labatut, R.A., Lodge, J.S., Williamson, A.A. & Trabold, T.A. (2016). Anaerobic co-

digestion of commercial food waste and dairy manure: Characterizing biochemical parameters and synergistic effects. Waste Management (New York, N.y.), 52, pp. 286-294.

El-Mashad, H.M., Zeeman, G., van Loon, W.K.P., Bot, G.P.A. & Lettinga, G. (2004). Effect of temperature and temperature fluctuation on thermophilic anaerobic digestion of cattle manure. Bioresource Technology, 95(2), pp. 191-201.

Elbeshbishy, E., Nakhla, G. & Hafez, H. (2012). Biochemical methane potential (BMP) of food waste and primary sludge: influence of inoculum pre-incubation and inoculum source. Bioresource Technology, 110, pp. 18-25.

Estes, A.M., Hearn, D.J., Snell-Rood, E.C., Feindler, M., Feeser, K., Abebe, T., Dunning Hotopp, J.C. & Moczek, A.P. (2013). Brood Ball-Mediated Transmission of Microbiome Members in the Dung Beetle, Onthophagus taurus (Coleoptera: Scarabaeidae). PLoS One, 8(11), p. e79061.

56

Estevez, M.M., Sapci, Z., Linjordet, R., Schnürer, A. & Morken, J. (2014). Semi-continuous anaerobic co-digestion of cow manure and steam-exploded Salix with recirculation of liquid digestate. Journal of Environmental Management, 136, pp. 9-15.

Ferguson, R.M.W., Coulon, F. & Villa, R. (2016). Organic loading rate: A promising microbial management tool in anaerobic digestion. Water Res, 100, pp. 348-356.

final, C. (2014). Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: Brussels.

Fitamo, T., Treu, L., Boldrin, A., Sartori, C., Angelidaki, I. & Scheutz, C. (2017). Microbial population dynamics in urban organic waste anaerobic co-digestion with mixed sludge during a change in feedstock composition and different hydraulic retention times. Water Res, 118, pp. 261-271.

Franke-Whittle, I.H., Walter, A., Ebner, C. & Insam, H. (2014). Investigation into the effect of high concentrations of volatile fatty acids in anaerobic digestion on methanogenic communities. Waste Management, 34(11), pp. 2080-2089.

Fry, J.C. (2004). Culture-dependent microbiology. In: Microbial Diversity and Bioprospecting American Society of Microbiology, pp. 80-87.

Fu, X. & Hu, Y. (2016). Comparison of Reactor Configurations for Biogas Production from Rapeseed Straw. Bioresources, 11(4), pp. 9970-9985.

Gani, A. & Naruse, I. (2007). Effect of cellulose and lignin content on pyrolysis and combustion characteristics for several types of biomass. Renewable Energy, 32(4), pp. 649-661.

Gawad, C., Koh, W. & Quake, S.R. (2016). Single-cell genome sequencing: current state of the science. Nature reviews genetics, 17, p. 175.

Goswami, R., Mukherjee, S., Chakraborty, A.K., Balachandran, S., Sinha Babu, S.P. & Chaudhury, S. (2016). Optimization of growth determinants of a potent cellulolytic bacterium isolated from lignocellulosic biomass for enhancing biogas production. Clean Technologies and Environmental Policy, 18(5), pp. 1565-1583.

Grim, J., Malmros, P., Schnürer, A. & Nordberg, Å. (2015). Comparison of pasteurization and integrated thermophilic sanitation at a full-scale biogas plant–Heat demand and biogas production. Energy, 79, pp. 419-427.

Grohmann, A., Fehrmann, S., Vainshtein, Y., Haag, N., Wiese, F., Stevens, P., Naegele, H.-J., Oechsner, H., Hartsch, T., Sohn, K. & Grumaz, C. (2017). Microbiome dynamics and adaptation of expression signatures during methane production failure and process recovery247).

Gruninger, R.J., Puniya, A.K., Callaghan, T.M., Edwards, J.E., Youssef, N., Dagar, S.S., Fliegerova, K., Griffith, G.W., Forster, R., Tsang, A., McAllister, T. & Elshahed, M.S. (2014). Anaerobic fungi (phylum Neocallimastigomycota): advances in understanding their taxonomy, life cycle, ecology, role and biotechnological potential. FEMS Microbiol Ecol, 90(1), pp. 1-17.

Gu, Y., Chen, X., Liu, Z., Zhou, X. & Zhang, Y. (2014). Effect of inoculum sources on the anaerobic digestion of rice straw. Bioresour Technol, 158, pp. 149-55.

Güllert, S., Fischer, M.A., Turaev, D., Noebauer, B., Ilmberger, N., Wemheuer, B., Alawi, M., Rattei, T., Daniel, R., Schmitz, R.A., Grundhoff, A. & Streit, W.R. (2016). Deep metagenome and metatranscriptome analyses of microbial communities affiliated with an industrial biogas fermenter, a cow rumen, and elephant feces reveal major differences in carbohydrate hydrolysis strategies. Biotechnol Biofuels, 9(1), pp. 1-20.

Haitjema, C.H., Solomon, K.V., Henske, J.K., Theodorou, M.K. & O'Malley, M.A. (2014). Anaerobic gut fungi: Advances in isolation, culture, and cellulolytic enzyme discovery for biofuel production. Biotechnol Bioeng, 111(8), pp. 1471-82.

Hanreich, A., Schimpf, U., Zakrzewski, M., Schlüter, A., Benndorf, D., Heyer, R., Rapp, E., Pühler, A., Reichl, U. & Klocke, M. (2013). Metagenome and metaproteome analyses of microbial communities in mesophilic biogas-producing anaerobic batch fermentations indicate concerted plant carbohydrate degradation. Syst Appl Microbiol, 36(5), pp. 330-338.

Hassa, J., Maus, I., Off, S., Pühler, A., Scherer, P., Klocke, M. & Schlüter, A. (2018). Metagenome, metatranscriptome, and metaproteome approaches unraveled compositions and functional relationships of microbial communities residing in biogas plants. Appl Microbiol Biotechnol, pp. 1-19.

Hemsworth, G.R., Johnston, E.M., Davies, G.J. & Walton, P.H. (2015). Lytic Polysaccharide Monooxygenases in Biomass Conversion. Trends Biotechnol, 33(12), pp. 747-761.

Henrissat, B. (1991). A classification of glycosyl hydrolases based on amino acid sequence similarities. Biochemical Journal, 280(2), pp. 309-316.

57

Heyer, R., Kohrs, F., Benndorf, D., Rapp, E., Kausmann, R., Heiermann, M., Klocke, M. & Reichl, U. (2013). Metaproteome analysis of the microbial communities in agricultural biogas plants. N Biotechnol, 30(6), pp. 614-622.

Heyer, R., Kohrs, F., Reichl, U. & Benndorf, D. (2015). Metaproteomics of complex microbial communities in biogas plants. Microbial Biotechnology, 8(5), pp. 749-763.

Hodrová, B., Kopečný, J. & Káš, J. (1998). Cellulolytic enzymes of rumen anaerobic fungi Orpinomyces joyonii and Caecomyces communis. Res Microbiol, 149(6), pp. 417-427.

Hofman-Bang, J., Zheng, D., Westermann, P., Ahring, B.K. & Raskin, L. (2003). Molecular ecology of anaerobic reactor systems. In: Biomethanation I Springer, pp. 151-203.

Holm-Nielsen, J.B., Al Seadi, T. & Oleskowicz-Popiel, P. (2009). The future of anaerobic digestion and biogas utilization. Bioresour Technol, 100.

Horn, S.J., Vaaje-Kolstad, G., Westereng, B. & Eijsink, V. (2012). Novel enzymes for the degradation of cellulose. Biotechnol Biofuels, 5(1), p. 45.

Hu, J., Shao, J., Yang, H., Lin, G., Chen, Y., Wang, X., Zhang, W. & Chen, H. (2017). Co-gasification of coal and biomass: Synergy, characterization and reactivity of the residual char. Bioresource Technology, 244, pp. 1-7.

Jia, Y., Ng, S.-K., Lu, H., Cai, M. & Lee, P.K.H. (2018). Genome-centric metatranscriptomes and ecological roles of the active microbial populations during cellulosic biomass anaerobic digestion. Biotechnol Biofuels, 11(1), p. 117.

Joblin, K.N. & Naylor, G.E. (1989). Fermentation of woods by rumen anaerobic fungi. FEMS Microbiol Lett, 65(1‐2), pp. 119-122.

Johan, H. & Linus, K. (2018). Produktion och användning av biogas och rötrester år 2017: Eskilstuna: Statens Energimyndighet.

Johansen, K.S. (2016). Lytic Polysaccharide Monooxygenases: The Microbial Power Tool for Lignocellulose Degradation. Trends Plant Sci, 21(11), pp. 926-936.

Kameshwar, A.K.S. & Qin, W. (2018). Isolation and Screening of Cellulose-Degrading Microorganisms from Different Ecological Niches. In: Lübeck, M. (ed. Cellulases: Methods and Protocols. New York, NY: Springer New York, pp. 47-56.

Kampmann, K., Ratering, S., Kramer, I., Schmidt, M., Zerr, W. & Schnell, S. (2012). Unexpected stability of Bacteroidetes and Firmicutes communities in laboratory biogas reactors fed with different defined substrates. Appl Environ Microbiol, 78(7), pp. 2106-2119.

Kaparaju, P., Buendia, I., Ellegaard, L. & Angelidakia, I. (2008). Effects of mixing on methane production during thermophilic anaerobic digestion of manure: Lab-scale and pilot-scale studies. Bioresource Technology, 99(11), pp. 4919-4928.

Kaparaju, P., Ellegaard, L. & Angelidaki, I. (2009). Optimisation of biogas production from manure through serial digestion: Lab-scale and pilot-scale studies. Bioresource Technology, 100(2), pp. 701-709.

Karmellos, M., Kopidou, D. & Diakoulaki, D. (2016). A decomposition analysis of the driving factors of CO2 (Carbon dioxide) emissions from the power sector in the European Union countries. Energy, 94, pp. 680-692.

Kazda, M., Langer, S. & Bengelsdorf, F.R. (2014). Fungi open new possibilities for anaerobic fermentation of organic residues. Energy, Sustainability and Society, 4(1), pp. 1-9.

Keller, M. & Hettich, R. (2009). Environmental proteomics: a paradigm shift in characterizing microbial activities at the molecular level. Microbiology and molecular biology reviews, 73(1), pp. 62-70.

Kishishita, S., Fujii, T. & Ishikawa, K. (2015). Heterologous expression of hyperthermophilic cellulases of archaea Pyrococcus sp. by fungus Talaromyces cellulolyticus. J Ind Microbiol Biotechnol, 42(1), pp. 137-141.

Kleinsteuber, S. (2018). Metagenomics of Methanogenic Communities in Anaerobic Digesters. In: Stams, A.J.M. & Sousa, D. (eds) Biogenesis of Hydrocarbons. Cham: Springer International Publishing, pp. 1-23.

Koeck, D.E., Maus, I., Wibberg, D., Winkler, A., Zverlov, V.V., Liebl, W., Pühler, A., Schwarz, W.H. & Schlüter, A. (2016). Complete genome sequence of Herbinix luporum SD1D, a new cellulose-degrading bacterium isolated from a thermophilic biogas reactor. Genome Announcements, 4(4).

Kohler, A., Kuo, A., Nagy, L.G., Morin, E., Barry, K.W., Buscot, F., Canbäck, B., Choi, C., Cichocki, N., Clum, A., Colpaert, J., Copeland, A., Costa, M.D., Doré, J., Floudas, D., Gay, G., Girlanda, M., Henrissat, B., Herrmann, S., Hess, J., Högberg, N., Johansson, T., Khouja, H.-R., LaButti, K., Lahrmann, U., Levasseur, A., Lindquist, E.A., Lipzen, A., Marmeisse, R.,

58

Martino, E., Murat, C., Ngan, C.Y., Nehls, U., Plett, J.M., Pringle, A., Ohm, R.A., Perotto, S., Peter, M., Riley, R., Rineau, F., Ruytinx, J., Salamov, A., Shah, F., Sun, H., Tarkka, M., Tritt, A., Veneault-Fourrey, C., Zuccaro, A., Mycorrhizal Genomics Initiative, C., Tunlid, A., Grigoriev, I.V., Hibbett, D.S. & Martin, F. (2015). Convergent losses of decay mechanisms and rapid turnover of symbiosis genes in mycorrhizal mutualists. Nat Genet, 47, p. 410.

König, H., Fröhlich, J. & Hertel, H. (2006). Diversity and Lignocellulolytic Activities of Cultured Microorganisms. In: König, H. & Varma, A. (eds) Intestinal Microorganisms of Termites and Other Invertebrates. Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 271-301.

Kothari, R., Pandey, A., Kumar, S., Tyagi, V. & Tyagi, S. (2014). Different aspects of dry anaerobic digestion for bio-energy: An overview. Renewable and Sustainable Energy Reviews, 39, pp. 174-195.

Labatut, R.A., Angenent, L.T. & Scott, N.R. (2011). Biochemical methane potential and biodegradability of complex organic substrates. Bioresour Technol, 102(3), pp. 2255-64.

Labatut, R.A., Angenent, L.T. & Scott, N.R. (2014). Conventional mesophilic vs. thermophilic anaerobic digestion: A trade-off between performance and stability? Water Res, 53, pp. 249-258.

Lebuhn, M., Weiß, S., Munk, B. & Guebitz, G.M. (2015). Microbiology and molecular biology tools for biogas process analysis, diagnosis and control. In: Biogas Science and Technology Springer, pp. 1-40.

Lehtomäki, A., Huttunen, S. & Rintala, J. (2007). Laboratory investigations on co-digestion of energy crops and crop residues with cow manure for methane production: effect of crop to manure ratio. Resources, Conservation and Recycling, 51(3), pp. 591-609.

Leschine, S.B. (1995). Cellulose degradation in anaerobic environments. Annual Review of Microbiology, 49, pp. 399-426.

Li, C., Nges, I.A., Lu, W. & Wang, H. (2017). Assessment of the degradation efficiency of full-scale biogas plants: A comparative study of degradation indicators. Bioresource Technology, 244, pp. 304-312.

Li, S.Y., Srivastava, R., Suib, S.L., Li, Y. & Parnas, R.S. (2011). Performance of batch, fed-batch, and continuous A-B-E fermentation with pH-control. Bioresour Technol, 102(5), pp. 4241-50.

Li, T.L., Mazeas, L., Sghir, A., Leblon, G. & Bouchez, T. (2009). Insights into networks of functional microbes catalysing methanization of cellulose under mesophilic conditions. Environmental Microbiology, 11(4), pp. 889-904.

Li, Y., Zhang, R., He, Y., Zhang, C., Liu, X., Chen, C. & Liu, G. (2014). Anaerobic co-digestion of chicken manure and corn stover in batch and continuously stirred tank reactor (CSTR). Bioresource Technology, 156, pp. 342-347.

Li, Y., Zhang, R., Liu, G., Chen, C., He, Y. & Liu, X. (2013). Comparison of methane production potential, biodegradability, and kinetics of different organic substrates. Bioresource Technology, 149, pp. 565-569.

Liebetrau, J., Reinelt, T., Clemens, J., Hafermann, C., Friehe, J. & Weiland, P. (2013). Analysis of greenhouse gas emissions from 10 biogas plants within the agricultural sector. Water Science and Technology, 67(6), pp. 1370-9.

Lim, H.C. & Shin, H.S. (2013). Fed-batch Cultures: Principles and Applications of Semi-batch Bioreactors: Cambridge University Press.

Lissens, G., Verstraete, W., Albrecht, T., Brunner, G., Creuly, C., Seon, J., Dussap, G. & Lasseur, C. (2004). Advanced anaerobic bioconversion of lignocellulosic waste for bioregenerative life support following thermal water treatment and biodegradation by Fibrobacter succinogenes. Biodegradation, 15(3), pp. 173-183.

Liu, W.-T., Marsh, T.L., Cheng, H. & Forney, L.J. (1997). Characterization of microbial diversity by determining terminal restriction fragment length polymorphisms of genes encoding 16S rRNA. Appl Environ Microbiol, 63(11), pp. 4516-4522.

Lu, F., Bize, A., Guillot, A., Monnet, V., Madigou, C., Chapleur, O., Mazeas, L., He, P. & Bouchez, T. (2014). Metaproteomics of cellulose methanisation under thermophilic conditions reveals a surprisingly high proteolytic activity. ISME J, 8(1), pp. 88-102.

Lü, F., Bize, A., Guillot, A., Monnet, V., Madigou, C., Chapleur, O., Mazéas, L., He, P. & Bouchez, T. (2014). Metaproteomics of cellulose methanisation under thermophilic conditions reveals a surprisingly high proteolytic activity. ISME J, 8(1), pp. 88-102.

59

Luo, G., Fotidis, I.A. & Angelidaki, I. (2016). Comparative analysis of taxonomic, functional, and metabolic patterns of microbiomes from 14 full-scale biogas reactors by metagenomic sequencing and radioisotopic analysis. Biotechnol Biofuels, 9, p. 51.

Lynd, L.R., Weimer, P.J., Van Zyl, W.H. & Pretorius, I.S. (2002a). Microbial cellulose utilization: fundamentals and biotechnology. Microbiology and molecular biology reviews, 66(3), pp. 506-577.

Lynd, L.R., Weimer, P.J., van Zyl, W.H. & Pretorius, I.S. (2002b). Microbial cellulose utilization: fundamentals and biotechnology. Microbiol Mol Biol Rev, 66(3), pp. 506-77, table of contents.

Ma, J., Zhao, Q.-B., Laurens, L.L., Jarvis, E.E., Nagle, N.J., Chen, S. & Frear, C.S. (2015). Mechanism, kinetics and microbiology of inhibition caused by long-chain fatty acids in anaerobic digestion of algal biomass. Biotechnol Biofuels, 8(1), p. 141.

Macias-Corral, M., Samani, Z., Hanson, A., Smith, G., Funk, P., Yu, H. & Longworth, J. (2008). Anaerobic digestion of municipal solid waste and agricultural waste and the effect of co-digestion with dairy cow manure. Bioresource Technology, 99(17), pp. 8288-8293.

Malherbe, S. & Cloete, T.E. (2002). Lignocellulose biodegradation: Fundamentals and applications. Reviews in Environmental Science and Biotechnology, 1(2), pp. 105-114.

Mao, C., Feng, Y., Wang, X. & Ren, G. (2015). Review on research achievements of biogas from anaerobic digestion. Renewable and Sustainable Energy Reviews, 45, pp. 540-555.

Marchaim, U. & Krause, C. (1993). Propionic to acetic acid ratios in overloaded anaerobic digestion. Bioresource Technology, 43(3), pp. 195-203.

Maron, P.-A., Ranjard, L., Mougel, C. & Lemanceau, P. (2007). Metaproteomics: a new approach for studying functional microbial ecology. Microbial Ecology, 53(3), pp. 486-493.

Martínez-Gutiérrez, E. (2018). Biogas production from different lignocellulosic biomass sources: advances and perspectives. 3 Biotech, 8(5).

Mata-Alvarez, J., Dosta, J., Romero-Güiza, M., Fonoll, X., Peces, M. & Astals, S. (2014). A critical review on anaerobic co-digestion achievements between 2010 and 2013. Renewable and Sustainable Energy Reviews, 36, pp. 412-427.

Mauky, E., Jacobi, H.F., Liebetrau, J. & Nelles, M. (2015). Flexible biogas production for demand-driven energy supply - Feeding strategies and types of substrates. Bioresour Technol, 178, pp. 262-9.

Meyer-Aurich, A., Lochmann, Y., Klauss, H. & Prochnow, A. (2016). Comparative Advantage of Maize-and Grass-Silage Based Feedstock for Biogas Production with Respect to Greenhouse Gas Mitigation. Sustainability, 8(7), p. 617.

Meyer, A.K.P., Ehimen, E.A. & Holm-Nielsen, J.B. (2017). Future European biogas: Animal manure, straw and grass potentials for a sustainable European biogas production. Biomass and Bioenergy.

Miron, J., Ben-Ghedalia, D. & Morrison, M. (2001). Invited review: adhesion mechanisms of rumen cellulolytic bacteria. J Dairy Sci, 84(6), pp. 1294-1309.

Moestedt, J., Muller, B., Westerholm, M. & Schnurer, A. (2016). Ammonia threshold for inhibition of anaerobic digestion of thin stillage and the importance of organic loading rate. Microb Biotechnol, 9(2), pp. 180-94.

Moestedt, J., Nordell, E. & Schnürer, A. (2014). Comparison of operating strategies for increased biogas production from thin stillage. Journal of Biotechnology, 175, pp. 22-30.

Moestedt, J., Paledal, S.N., Schnurer, A. & Nordell, E. (2013). Biogas Production from Thin Stillage on an Industrial Scale-Experience and Optimisation. Energies, 6(11), pp. 5642-5655.

Möller, K. & Müller, T. (2012). Effects of anaerobic digestion on digestate nutrient availability and crop growth: a review. Engineering in Life Sciences, 12(3), pp. 242-257.

Monlau, F., Barakat, A., Trably, E., Dumas, C., Steyer, J.-P. & Carrère, H. (2013). Lignocellulosic Materials Into Biohydrogen and Biomethane: Impact of Structural Features and Pretreatment. Critical Reviews in Environmental Science and Technology, 43(3), pp. 260-322.

Morozova, O. & Marra, M.A. (2008). Applications of next-generation sequencing technologies in functional genomics. Genomics, 92(5), pp. 255-64.

Morrison, M., Pope, P.B., Denman, S.E. & McSweeney, C.S. (2009a). Plant biomass degradation by gut microbiomes: more of the same or something new? Curr Opin Biotechnol, 20(3), pp. 358-63.

Morrison, M., Pope, P.B., Denman, S.E. & McSweeney, C.S. (2009b). Plant biomass degradation by gut microbiomes: more of the same or something new? Curr Opin Biotechnol, 20(3), pp. 358-363.

60

Moset, V., Poulsen, M., Wahid, R., Højberg, O. & Møller, H.B. (2015). Mesophilic versus thermophilic anaerobic digestion of cattle manure: methane productivity and microbial ecology. Microbial Biotechnology, 8(5), pp. 787-800.

Mulat, D.G. & Horn, S.J. (2018). Biogas Production from Lignin via Anaerobic Digestion. In: Lignin Valorization, pp. 391-412.

Mulat, D.G., Jacobi, H.F., Feilberg, A., Adamsen, A.P.S., Richnow, H.-H. & Nikolausz, M. (2016). Changing feeding regimes to demonstrate flexible biogas production: effects on process performance, microbial community structure, and methanogenesis pathways. Appl Environ Microbiol, 82(2), pp. 438-449.

Müller, B., Sun, L., Westerholm, M. & Schnürer, A. (2016). Bacterial community composition and fhs profiles of low-and high-ammonia biogas digesters reveal novel syntrophic acetate-oxidising bacteria. Biotechnol Biofuels, 9(1), p. 1.

Mullings, R. & Parish, J. (1984). Mesophilic aerobic Gram negative cellulose degrading bacteria from aquatic habitats and soils. Journal of Applied Bacteriology, 57(3), pp. 455-468.

Neshat, S.A., Mohammadi, M., Najafpour, G.D. & Lahijani, P. (2017). Anaerobic co-digestion of animal manures and lignocellulosic residues as a potent approach for sustainable biogas production. Renewable and Sustainable Energy Reviews, 79, pp. 308-322.

Nges, I.A., Escobar, F., Fu, X. & Bjornsson, L. (2012). Benefits of supplementing an industrial waste anaerobic digester with energy crops for increased biogas production. Waste Manag, 32(1), pp. 53-9.

Nizami, A.-S. & Murphy, J.D. (2010). What type of digester configurations should be employed to produce biomethane from grass silage? Renewable and Sustainable Energy Reviews, 14(6), pp. 1558-1568.

Noike, T., Endo, G., Chang, J.E., Yaguchi, J.I. & Matsumoto, J.I. (1985). Characteristics of carbohydrate degradation and the rate‐limiting step in anaerobic digestion. Biotechnol Bioeng, 27(10), pp. 1482-1489.

Orpin, C.G. (1975). Studies on the Rumen Flagellate Neocallimastix frontalis. Microbiology, 91(2), pp. 249-262.

Pap, B. & Maróti, G. (2016). Diversity of Microbial Communities in Biogas Reactors. Current Biochemical Engineering, 3(3), pp. 177-187.

Parawira, W., Read, J.S., Mattiasson, B. & Björnsson, L. (2008). Energy production from agricultural residues: High methane yields in pilot-scale two-stage anaerobic digestion. Biomass and Bioenergy, 32(1), pp. 44-50.

Pechtl, A., Rückert, C., Maus, I., Koeck, D.E., Trushina, N., Kornberger, P., Schwarz, W.H., Schlüter, A., Liebl, W. & Zverlov, V.V. (2018). Complete Genome Sequence of the Novel Cellulolytic, Anaerobic, Thermophilic Bacterium <em>Herbivorax saccincola</em> Type Strain GGR1, Isolated from a Lab Scale Biogas Reactor as Established by Illumina and Nanopore MinION Sequencing. Genome Announcements, 6(6).

Pereyra, L.P., Hiibel, S.R., Prieto Riquelme, M.V., Reardon, K.F. & Pruden, A. (2010). Detection and quantification of functional genes of cellulose- degrading, fermentative, and sulfate-reducing bacteria and methanogenic archaea. Appl Environ Microbiol, 76(7), pp. 2192-202.

Perrotta, A.R., Kumaraswamy, R., Bastidas-Oyanedel, J.R., Alm, E.J. & Rodríguez, J. (2017). Inoculum composition determines microbial community and function in an anaerobic sequential batch reactor. PLoS One, 12(2), p. e0171369.

Piao, Z.H., Lee, J. & Kim, J.Y. (2018). Effect of substrate feeding frequencies on the methane production and microbial communities of laboratory-scale anaerobic digestion reactors. Journal of Material Cycles and Waste Management, 20(1), pp. 147-154.

Pore, S.D., Shetty, D., Arora, P., Maheshwari, S. & Dhakephalkar, P.K. (2016). Metagenome changes in the biogas producing community during anaerobic digestion of rice straw. Bioresource Technology, 213, pp. 50-53.

Poszytek, K., Ciezkowska, M., Sklodowska, A. & Drewniak, L. (2016). Microbial Consortium with High Cellulolytic Activity (MCHCA) for enhanced biogas production. Frontiers in Microbiology, 7.

Prince, A.L., Antony, K.M., Ma, J. & Aagaard, K.M. (2014). The microbiome and development: a mother's perspective. Semin Reprod Med, 32(1), pp. 14-22.

Prochazka, J., Mrazek, J., Strosova, L., Fliegerova, K., Zabranska, J. & Dohanyos, M. (2012). Enhanced biogas yield from energy crops with rumen anaerobic fungi. Engineering in Life Sciences, 12(3), pp. 343-351.

61

Qiao, J.-T., Qiu, Y.-L., Yuan, X.-Z., Shi, X.-S., Xu, X.-H. & Guo, R.-B. (2013). Molecular characterization of bacterial and archaeal communities in a full-scale anaerobic reactor treating corn straw. Bioresource Technology, 143, pp. 512-518.

Rademacher, A., Zakrzewski, M., Schlüter, A., Schönberg, M., Szczepanowski, R., Goesmann, A., Pühler, A. & Klocke, M. (2012). Characterization of microbial biofilms in a thermophilic biogas system by high-throughput metagenome sequencing. FEMS Microbiol Ecol, 79.

Rajendran, K., Aslanzadeh, S. & Taherzadeh, M.J. (2012). Household biogas digesters—A review. Energies, 5(8), pp. 2911-2942.

Ramos, C.G., Grilo, A.M., Sousa, S.A., Barbosa, M.L., Nadais, H. & Leitao, J.H. (2010). A new methodology combining PCR, cloning, and sequencing of clones discriminated by RFLP for the study of microbial populations: application to an UASB reactor sample. Appl Microbiol Biotechnol, 85(3), pp. 801-6.

Ransom-Jones, E., Jones, D.L., McCarthy, A.J. & McDonald, J.E. (2012). The Fibrobacteres: an Important Phylum of Cellulose-Degrading Bacteria. Microbial Ecology, 63(2), pp. 267-281.

Raposo, F., De la Rubia, M.A., Fernández-Cegrí, V. & Borja, R. (2012). Anaerobic digestion of solid organic substrates in batch mode: An overview relating to methane yields and experimental procedures. Renewable and Sustainable Energy Reviews, 16(1), pp. 861-877.

Rico, C., Muñoz, N. & Rico, J.L. (2015). Anaerobic co-digestion of cheese whey and the screened liquid fraction of dairy manure in a single continuously stirred tank reactor process: Limits in co-substrate ratios and organic loading rate. Bioresource Technology, 189, pp. 327-333.

Risberg, K., Sun, L., Leven, L., Horn, S.J. & Schnurer, A. (2013). Biogas production from wheat straw and manure - Impact of pretreatment and process operating parameters. Bioresource Technology, 149, pp. 232-237.

Rui, J., Li, J., Zhang, S., Yan, X., Wang, Y. & Li, X. (2015). The core populations and co-occurrence patterns of prokaryotic communities in household biogas digesters. Biotechnol Biofuels, 8(1), p. 158.

Ruile, S., Schmitz, S., Mönch-Tegeder, M. & Oechsner, H. (2015). Degradation efficiency of agricultural biogas plants – A full-scale study. Bioresource Technology, 178, pp. 341-349.

Saini, A., Aggarwal, N.K. & Yadav, A. (2017). Isolation and Screening of Cellulose Hydrolyzing Bacteria from Different Ecological Niches.

Satpathy, P., Steinigeweg, S., Cypionka, H. & Engelen, B. (2016). Different substrates and starter inocula govern microbial community structures in biogas reactors. Environ Technol, 37(11), pp. 1441-50.

Sawatdeenarunat, C., Surendra, K.C., Takara, D., Oechsner, H. & Khanal, S.K. (2015). Anaerobic digestion of lignocellulosic biomass: Challenges and opportunities. Bioresource Technology, 178, pp. 178-186.

Scarlat, N., Dallemand, J.-F. & Fahl, F. (2018). Biogas: Developments and perspectives in Europe. Renewable Energy, 129, pp. 457-472.

Schnürer, A. (2016). Biogas Production: Microbiology and Technology. In: Hatti-Kaul, R., Mamo, G. & Mattiasson, B. (eds) Anaerobes in Biotechnology. Cham: Springer International Publishing, pp. 195-234.

Schnürer, A., Bohn, I. & Moestedt, J. (2017). Protocol for Start-Up and Operation of CSTR Biogas Processes. In: McGenity, T.J., Timmis, K.N. & Nogales, B. (eds) Hydrocarbon and Lipid Microbiology Protocols: Bioproducts, Biofuels, Biocatalysts and Facilitating Tools. Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 171-200.

Schnürer, A. & Jarvis, Å. (2018). Microbiology of the biogas process. Schnürer, A., Zellner, G. & Svensson, B.H. (1999). Mesophilic syntrophic acetate oxidation during

methane formation in biogas reactors. FEMS Microbiol Ecol, 29. Schwarz, W. (2001). The cellulosome and cellulose degradation by anaerobic bacteria. Appl Microbiol

Biotechnol, 56(5), pp. 634-649. Shi, X.-S., Dong, J.-J., Yu, J.-H., Yin, H., Hu, S.-M., Huang, S.-X. & Yuan, X.-Z. (2017). Effect of

Hydraulic Retention Time on Anaerobic Digestion of Wheat Straw in the Semicontinuous Continuous Stirred-Tank Reactors. Biomed Res Int, 2017.

Shoseyov, O., Shani, Z. & Levy, I. (2006). Carbohydrate binding modules: biochemical properties and novel applications. Microbiol Mol Biol Rev, 70(2), pp. 283-295.

Sosnowski, P., Wieczorek, A. & Ledakowicz, S. (2003). Anaerobic co-digestion of sewage sludge and organic fraction of municipal solid wastes. Advances in Environmental Research, 7(3), pp. 609-616.

62

Speda, J., Johansson, M.A., Carlsson, U. & Karlsson, M. (2017a). Assessment of sample preparation methods for metaproteomics of extracellular proteins. Anal Biochem, 516, pp. 23-36.

Speda, J., Jonsson, B.-H., Carlsson, U. & Karlsson, M. (2017b). Metaproteomics-guided selection of targeted enzymes for bioprospecting of mixed microbial communities. Biotechnol Biofuels, 10(1), p. 128.

Stambasky, J., Pflüger, S., Deremince, B., Scheidl, S., de la Vega, N. & Conton, M. (2016). Annual Statistical Report of the European Biogas Association. Brussels: European Biogas Association.

Sun, L. (2015). Biogas production from lignocellulosic materials. Sun, L., Liu, T., Müller, B. & Schnürer, A. (2016). The microbial community structure in industrial

biogas plants influences the degradation rate of straw and cellulose in batch tests. Biotechnol Biofuels, 9(1), pp. 1-20.

Sun, L., Müller, B. & Schnürer, A. (2013). Biogas production from wheat straw: community structure of cellulose-degrading bacteria. Energy, Sustainability and Society, 3(1), pp. 1-11.

Sun, L., Pope, P.B., Eijsink, V.G.H. & Schnürer, A. (2015). Characterization of microbial community structure during continuous anaerobic digestion of straw and cow manure. Microbial Biotechnology, 8(5), pp. 815-827.

Sundberg, C., Al-Soud, W.A., Larsson, M., Alm, E., Yekta, S.S., Svensson, B.H., Sørensen, S.J. & Karlsson, A. (2013). 454 pyrosequencing analyses of bacterial and archaeal richness in 21 full-scale biogas digesters. FEMS Microbiol Ecol, 85(3), pp. 612-626.

Taherzadeh, M.J. & Karimi, K. (2008). Pretreatment of lignocellulosic wastes to improve ethanol and biogas production: a review. International journal of molecular sciences, 9(9), pp. 1621-1651.

Theuerl, S., Klang, J., Heiermann, M. & De Vrieze, J. (2018). Marker microbiome clusters are determined by operational parameters and specific key taxa combinations in anaerobic digestion. Bioresource Technology, 263, pp. 128-135.

Tomme, P., Warren, R.A.J. & Gilkes, N.R. (1995). Cellulose Hydrolysis by Bacteria and Fungi. In: Poole, R.K. (ed. Advances in Microbial Physiology37) Academic Press, pp. 1-81.

Tsapekos, P., Kougias, P.G., Vasileiou, S.A., Treu, L., Campanaro, S., Lyberatos, G. & Angelidaki, I. (2017). Bioaugmentation with hydrolytic microbes to improve the anaerobic biodegradability of lignocellulosic agricultural residues. Bioresource Technology, 234, pp. 350-359.

Tsavkelova, E.A. & Netrusov, A.I. (2012). Biogas production from cellulose-containing substrates: A review. Applied Biochemistry and Microbiology, 48(5), pp. 421-433.

Usack, J.G., Spirito, C.M. & Angenent, L.T. (2012). Continuously-stirred anaerobic digester to convert organic wastes into biogas: system setup and basic operation. J Vis Exp(65), p. e3978.

Vaaje-Kolstad, G., Westereng, B., Horn, S.J., Liu, Z., Zhai, H., Sørlie, M. & Eijsink, V.G.H. (2010). An Oxidative Enzyme Boosting the Enzymatic Conversion of Recalcitrant Polysaccharides. Science, 330(6001), p. 219.

Vanwonterghem, I., Jensen, P.D., Dennis, P.G., Hugenholtz, P., Rabaey, K. & Tyson, G.W. (2014a). Deterministic processes guide long-term synchronised population dynamics in replicate anaerobic digesters. ISME J, 8, p. 2015.

Vanwonterghem, I., Jensen, P.D., Ho, D.P., Batstone, D.J. & Tyson, G.W. (2014b). Linking microbial community structure, interactions and function in anaerobic digesters using new molecular techniques. Curr Opin Biotechnol, 27, pp. 55-64.

Vanwonterghem, I., Jensen, P.D., Rabaey, K. & Tyson, G.W. (2016). Genome‐centric resolution of microbial diversity, metabolism and interactions in anaerobic digestion. Environmental Microbiology, 18(9), pp. 3144-3158.

Vasco-Correa, J., Khanal, S., Manandhar, A. & Shah, A. (2018). Anaerobic digestion for bioenergy production: Global status, environmental and techno-economic implications, and government policies. Bioresource Technology, 247, pp. 1015-1026.

Veeken, A. & Hamelers, B. (1999). Effect of temperature on hydrolysis rates of selected biowaste components. Bioresource Technology, 69(3), pp. 249-254.

Viikari, L., Suurnäkki, A., Grönqvist, S., Raaska, L. & Ragauskas, A. (2009). Forest Products: Biotechnology in Pulp and Paper Processing. In: Schaechter, M. (ed. Encyclopedia of Microbiology (Third Edition). Oxford: Academic Press, pp. 80-94.

Wang, M., Ahrne, S., Antonsson, M. & Molin, G. (2004). T-RFLP combined with principal component analysis and 16S rRNA gene sequencing: an effective strategy for comparison of fecal microbiota in infants of different ages. J Microbiol Methods, 59(1), pp. 53-69.

63

Wang, M., Lai, G.-L., Nie, Y., Geng, S., Liu, L., Zhu, B., Shi, Z. & Wu, X.-L. (2015). Synergistic function of four novel thermostable glycoside hydrolases from a long-term enriched thermophilic methanogenic digester. Frontiers in Microbiology, 6, p. 509.

Ward, A.J., Hobbs, P.J., Holliman, P.J. & Jones, D.L. (2008). Optimisation of the anaerobic digestion of agricultural resources. Bioresource Technology, 99(17), pp. 7928-7940.

Watanabe, R., Nie, Y., Takahashi, S., Wakahara, S. & Li, Y.-Y. (2016). Efficient performance and the microbial community changes of submerged anaerobic membrane bioreactor in treatment of sewage containing cellulose suspended solid at 25° C. Bioresource Technology, 216, pp. 128-134.

Wei, Y., Zhou, H., Zhang, J., Zhang, L., Geng, A., Liu, F., Zhao, G., Wang, S., Zhou, Z. & Yan, X. (2015). Insight into dominant cellulolytic bacteria from two biogas digesters and their glycoside hydrolase genes. PLoS One, 10(6), p. e0129921.

Weiland, P. (2010). Biogas production: current state and perspectives. Appl Microbiol Biotechnol, 85(4), pp. 849-860.

Westerholm, M., Hansson, M. & Schnurer, A. (2012). Improved biogas production from whole stillage by co-digestion with cattle manure. Bioresour Technol, 114, pp. 314-9.

Westerholm, M., Isaksson, S., Karlsson Lindsjö, O. & Schnürer, A. (2018). Microbial community adaptability to altered temperature conditions determines the potential for process optimisation in biogas production. Applied Energy, 226, pp. 838-848.

Westerholm, M., Isaksson, S., Sun, L. & Schnürer, A. (2017). Microbial Community Ability to Adapt to Altered Temperature Conditions Influences Operating Stability in Anaerobic Digestion. Energy Procedia, 105, pp. 895-900.

Westerholm, M., Moestedt, J. & Schnürer, A. (2016). Biogas production through syntrophic acetate oxidation and deliberate operating strategies for improved digester performance. Applied Energy, 179, pp. 124-135.

Westerholm, M., Müller, B., Isaksson, S. & Schnürer, A. (2015). Trace element and temperature effects on microbial communities and links to biogas digester performance at high ammonia levels. Biotechnol Biofuels, 8(1), p. 154.

Westerholm, M., Roos, S. & Schnürer, A. (2010). Syntrophaceticus schinkii gen. nov., sp. nov., an anaerobic, syntrophic acetate-oxidizing bacterium isolated from a mesophilic anaerobic filter. FEMS Microbiol Lett, 309(1), pp. 100-104.

Xia, Y., Ju, F., Fang, H.H. & Zhang, T. (2013). Mining of novel thermo-stable cellulolytic genes from a thermophilic cellulose-degrading consortium by metagenomics. PLoS One, 8(1), p. e53779.

Yan, L., Gao, Y., Wang, Y., Liu, Q., Sun, Z., Fu, B., Wen, X., Cui, Z. & Wang, W. (2012). Diversity of a mesophilic lignocellulolytic microbial consortium which is useful for enhancement of biogas production. Bioresource Technology, 111, pp. 49-54.

Ye, R., Jin, Q., Bohannan, B., Keller, J.K. & Bridgham, S.D. (2014). Homoacetogenesis: A potentially underappreciated carbon pathway in peatlands. Soil Biology and Biochemistry, 68, pp. 385-391.

Yilmaz, S. & Singh, A.K. (2012). Single cell genome sequencing. Curr Opin Biotechnol, 23(3), pp. 437-443.

Young, D., Dollhofer, V., Callaghan, T.M., Reitberger, S., Lebuhn, M. & Benz, J.P. (2018). Isolation, identification and characterization of lignocellulolytic aerobic and anaerobic fungi in one- and two-phase biogas plants. Bioresource Technology, 268, pp. 470-479.

Yu, Q., Tian, Z., Liu, J., Zhou, J., Yan, Z., Yong, X., Jia, H., Wu, X. & Wei, P. (2018). Biogas Production and Microbial Community Dynamics during the Anaerobic Digestion of Rice Straw at 39–50 °C: A Pilot Study. Energy & Fuels, 32(4), pp. 5157-5163.

Zamorska-Wojdyła, D., Gaj, K., Hołtra, A. & Sitarska, M. (2012). Quality Evaluation of Biogas and Selected Methods of its Analysis. Ecological Chemistry and Engineering S, 19(1), pp. 77-87.

Zhang, T., Liu, L., Song, Z., Ren, G., Feng, Y., Han, X. & Yang, G. (2013). Biogas production by co-digestion of goat manure with three crop residues. PLoS One, 8(6), p. e66845.

Zheng, Z., Liu, J., Yuan, X., Wang, X., Zhu, W., Yang, F. & Cui, Z. (2015). Effect of dairy manure to switchgrass co-digestion ratio on methane production and the bacterial community in batch anaerobic digestion. Applied Energy, 151, pp. 249-257.

Zhou, J., Yang, J., Yu, Q., Yong, X., Xie, X., Zhang, L., Wei, P. & Jia, H. (2017). Different organic loading rates on the biogas production during the anaerobic digestion of rice straw: A pilot study. Bioresource Technology, 244, pp. 865-871.

64

Ziels, R.M., Beck, D.A.C. & Stensel, H.D. (2017). Long-chain fatty acid feeding frequency in anaerobic codigestion impacts syntrophic community structure and biokinetics. Water Res, 117, pp. 218-229.

Ziels, R.M., Sousa, D.Z., Stensel, H.D. & Beck, D.A.C. (2018). DNA-SIP based genome-centric metagenomics identifies key long-chain fatty acid-degrading populations in anaerobic digesters with different feeding frequencies. ISME J, 12(1), pp. 112-123.

Zinder, S.H. & Koch, M. (1984). Non-Aceticlastic Methanogenesis from Acetate - Acetate Oxidation by a Thermophilic Syntrophic Coculture. Archives of Microbiology, 138(3), pp. 263-272.

65

The development of the oil industry has led to a rapid rise in the global economy in the last century. However, fossil fuel is a limited and non-sustainable energy resource. It is believed that, if no alternatives are developed in the future, energy constraints on the international community will become the main bottleneck in economic development. In addition, emissions of greenhouse gases (e.g. fossil fuel-derived carbon dioxide) have become a global concern, with about 88% of global energy consumption originating from fossil fuels. To overcome the environmental challenges and the dependence on fossil fuel, governments world-wide have formulated various policies to encourage the use of renewable energy.

Against this background, anaerobic digestion technology is highly interesting. In this process, various types of organic materials can be degraded under anaerobic conditions (without oxygen) into the end-product biogas, a renewable energy source. Anaerobic digestion is a multi-functional technology and it can be used simultaneously for waste management, production of renewable energy and production of an organic fertiliser. In addition, the biogas process can be implemented at small or large scale, which is important when designing flexible and sustainable energy solutions in both industrialised and developing countries. Materials that can be used for biogas production include various types of waste products, such as manure, straw, municipal wastewater, food waste etc., and dedicated energy crops. By controlled use of wastes in a biogas process, rather than dumping household waste in landfill or storing manure in open tanks, it is possible to reduce the volume of unwanted wastes and also decrease emissions of carbon dioxide, methane and other greenhouse gases. The biogas produced, containing the energy carrier methane, can be used for production of heat, electricity or vehicle fuel after upgrading (removal of carbon dioxide and trace gases). The residues left after biogas production are rich in plant nutrients and can be used as a fertiliser in crop

Popular science summary

66

production, replacing fossil energy-requiring mineral fertilisers and enabling recycling of nutrients between urban and rural areas.

Microorganisms are essential for degrading organic materials to biogas, in a process that proceeds through various anaerobic digestion pathways and requires the combined activity of several groups of microorganisms. To obtain a stable biogas process, all these microorganisms must work in a synchronised manner.

Among the organic materials that can be used for biogas production, lignocellulosic materials, especially agriculture residues such as animal manure, straw, rice husks, corn stalks etc. are of great interest due to their high abundance world-wide. However, when lignocellulosic materials (including agriculture residues) are used for biogas production, the process efficiency is limited, due to the low nutrient content of these materials and the highly recalcitrant structure of their plant cell walls hindering microbial degradation. Thus, to achieve higher degradation efficiency of lignocellulosic materials for biogas production, a better understanding of the lignocellulose degraders (i.e. lignocellulolytic microorganisms) is needed.

This thesis examined possible lignocellulose degraders and studied the composition of their community. It also investigated possible links between changes in microbe community structure and environmental factors within the anaerobic reactor (e.g. process parameters, feedstock composition and feeding strategy). Different molecular methods (analysing DNA and proteins) for exploring the microbial communities were discussed, with the aim of building an appropriate pipeline for in-depth study of the lignocellulose degraders in biogas processes.

67

Utvecklingen av oljeindustrin har lett till en snabb ökning av världsekonomin under det senaste århundradet. Men fossilt bränsle är en begränsad och icke-hållbar energiresurs och användningen leder dessutom till utsläpp av växthusgaser. Fossila bränslen står för ca 88% av den globala energiförbrukningen. För att övervinna miljöutmaningarna och beroendet av fossila bränslen har regeringar över hela världen formulerat olika strategier för att uppmuntra användningen av förnybar energi.

Mot denna bakgrund är anaerob (syrefri) rötning en mycket intressant teknik. I denna process kan olika typer av organiska material brytas ned av olika mikroorganismer till slutprodukten biogas, en förnybar energikälla. Anaerob nedbrytning är en multifunktionell teknik som kan användas för både behandling av avfall och produktion av förnybar energi och av organiskt gödselmedel. Dessutom kan systemet sättas upp både i liten eller stor skala, vilket är viktigt vid utformningen av flexibla och hållbara energilösningar i både industri- och utvecklingsländer. Material som kan användas för biogasproduktion inkluderar olika typer av avfall, såsom gödsel, halm, kommunalt avloppsvatten, matavfall m.m. och dedikerade energigrödor. Genom kontrollerad användning av avfall i en biogasprocess, snarare än deponering på soptipp eller lagring av gödsel i öppna tankar, är det möjligt att minska volymen av oönskat avfall och samtidigt minska även utsläppen av koldioxid, metan och andra växthusgaser. Den biogas som produceras kan sedan användas för produktion av värme, el eller fordonsbränsle efter uppgradering (avlägsnande av koldioxid och spårgas). Resterna som blir kvar efter biogasproduktionen är rik på växtnäringsämnen och kan användas som gödningsmedel i jordbruket och då ersätta fossila energikrävande mineralgödselmedel. Genom att använda rötresten som gödningsmedel möjliggörs också kretslopp av näringsämnen mellan städer och landsbygdsområden.

Populärvetenskaplig sammanfattning

68

Olika mikroorganismer är avgörande för att anaerob nedbrytning av organiskt material till biogas ska fungera. Den mikrobiella processen fortskrider genom flera olika nedbrytningsvägar och kräver också aktivitet av flera olika grupper av mikroorganismer. För att erhålla en stabil biogasprocess måste alla dessa mikroorganismer också fungera på ett synkroniserat sätt. Bland de organiska material som kan användas för biogasproduktion är olika jordbruksrester, som djurgödsel, halm, blast etc. av stort intresse på grund av att dessa finns i stor mängd. Karaktäristiskt för detta material är att det ofta innehåller mycket lignocellulosa, vilket har ett lågt näringsinnehåll och en komplicerad struktur, något som hindrar mikrobiell nedbrytning. För att uppnå högre nedbrytningseffektivitet och biogasproduktion av denna typ av material behövs en bättre förståelse av de bakterier som bryter ner lignocellulosa i biogasprocesser.

Denna avhandling undersökte vilka bakterier som är närvarande i olika biogasprocesser och som potentiellt kan vara inblandade i nedbrytningen av lignocellulosa. Den utredde också möjliga kopplingar mellan sammansättningen på mikroorganismssamhället och driften av biogasprocessen. Frågor som belystes var till exempel; kommer sammansättningen av det cellulosanedbrytande bakteriesamhället påverkas av vilket material som bryts ner i biogasreaktorn? Är vissa bakterier viktigare än andra för att få en bra nedbrytning? För att studera mikroorganismerna i olika biogasprocesser användes olika s.k. molekylära metoder (analys av DNA). Det övergripande syftet med studierna var dels att hitta metoder att studera specifikt de bakterier som bryter ner lignocellulosa i biogasprocesser och dels att förstå vilka som är mest kritiska för att få en effektiv process samt vilka parametrar som påverkar dem.

69

This PhD programme was financially supported by the thematic research program MicroDrive, the Swedish Energy Agency (ERA-NET Bioenergy), the China Scholarship Council (CSC) and the Research Council of Norway. First, I would like to thank my main supervisor, Professor Anna Schnürer. I could not have done this work without your help. Your serious attitude to science and how you balance work and family made you an idol to me. You were always so patient and modest when we discussed my ideas and work, even though you have so much more knowledge than I. You are my superhero. You have the superpower that turns every negative thing into a positive. No matter what difficulties I met in my work or life, you always made them easy. I am very appreciative that I had the chance to work with you for these years. You really helped and taught me a lot. I also want to thank my co-supervisors: Bettina Müller, thanks a lot for all the academic discussions. No matter how busy you were, and how stupid my questions were, you always gave the most detailed answers. In case you don’t know, you are also my wife’s idol. She likes your life attitude. Mats Sandgren, thanks for supporting me a lot in my last protein project. I also got so much advice about the scientist’s life from you. When I came to you and asked a question, you always directly gave me the answer, sometimes even before I finished my question. I don’t know how you did it.

Acknowledgements

70

I also would like to thank Åke Nordberg, thanks for your ideas on the applications of the AD process. Thanks also for all the comments in our published paper. Hope we will have more chances to work in the same project in the future. Thanks also for inviting me to drink vodka at the Poland conference. Many thanks to all the present and past biogas group members: Maria Erikson, thanks for all your help in the laboratory, maybe you don’t know, but somehow you made our group feel like a home. Maria Westerholm and Oskar Karlsson Lindsjö, thanks for all the discussions and mini-workshops, and the coffee smell in my office. Jan Moestedt, thanks for the reactor sample collection. Thanks to Karin Ahlberg-Eliasson, we had a great time at the Beijing conference and it was so nice to publish a paper with you. Thanks to my PhD colleagues and friends He Sun and Abhijeet Singh for making our working atmosphere so happy, and of course for all the moments of brainstorming to solve problems. Special thanks to Li Sun, you helped me a lot not only at work but also in my personal life. You are the most reliable person I have ever met. Special thanks to Simon Isaksson for never saying no when I needed help. You are my first and very important Swedish friend. I can’t imagine my life in Sweden without you around. Please don’t move back to your hometown. Sorry I didn’t catch up with the baby making plan. Also, thanks for all the fun times we had together (see below!). To a guy who is actually not working in our group, but somehow feels like he belongs to our group, I would like to give a big thank you to my friend Anton Pallin. Thanks a lot for all the tennis/BBQ/kayak/hamburger testing/movie nights/whiskey nights/gambling nights (here also @Simon). It is terrible news to me that you have fallen in love with Alice and have decide to move to Germany. But I guess it is my fault, because this all happened when I went back to China for a vacation. Anyway, I wish you the best, and I hope you and Alice will move back to Sweden one day. Also, lots of thanks to members of the biogas group at Linköping University. Thanks to Annika Björn for giving me so many valuable comments at my half-time seminar, I read them word by word. Thanks to Luka Šafarič and especially Sepehr Shakeri Yekta for all the fun discussions, I learned a lot from

71

both of you. I am so happy that we published a paper together, and I believe we will have a lot of chances in the future to cooperate again. I also want to express my thanks to: Members of RISE: Xinmei, Maria, Johnny, and Leticia. It was so nice to work with you in the biogas lab. Colleagues from Biocentre: Jonas, for teaching me beer brewing techniques. Ludwig, for sharing whiskey and Taiwan coffee. Mikael, for teaching me how to teach students. Nils, for helping me solve IT problems. Miao, for helping me with protein data analysis. Bing, Yunkai, and Chen, for helping me with the protein Extraction. To my family, First, I want to thank my mom Dan, who is sometimes strict, but always unconditionally believes in and supports my decisions. I am very sorry that I had to leave you and come to such a faraway place to chase my dreams. I want to you know that you are still giving me strength and courage, even though we are far apart. Many thanks to my cat Wasabi, although you are unable to read this, I want to you know that your cuteness always cures my worries. I wish you a long life. Last but definitely not least, I want to give all my thanks to my wife, Xin. Your love (and the food you cooked) is the most powerful energy source supporting me to fight against all the difficulties, and your cuteness always cures my worries too. Having you with me is the greatest happiness in my whole life. People always say that couples get tired of each other in marriage. It is not true. We have been together for almost ten years, and I still believe: In another ten years, we still live just like today ( , ).

Sun et al. Biotechnol Biofuels (2016) 9:128 DOI 10.1186/s13068-016-0543-9

RESEARCH

The microbial community structure in industrial biogas plants influences the degradation rate of straw and cellulose in batch testsLi Sun1†, Tong Liu1†, Bettina Müller1 and Anna Schnürer1,2*

Abstract

Background: Materials rich in lignocellulose, such as straw, are abundant, cheap and highly interesting for biogas

production. However, the complex structure of lignocellulose is difficult for microbial cellulolytic enzymes to access,

limiting degradation. The rate of degradation depends on the activity of members of the microbial community, but

the knowledge of this community in the biogas process is rather limited. This study, therefore, investigated the deg-

radation rate of cellulose and straw in batch cultivation test initiated with inoculums from four co-digestion biogas

plants (CD) and six wastewater treatment plants (WWTP). The results were correlated to the bacterial community by

454-pyrosequencing targeting 16S rRNA gene and by T-RFLP analysis targeting genes of glycoside hydrolase families

5 (cel5) and 48 (cel48), combined with construction of clone libraries

Results: UniFrac principal coordinate analysis of 16S rRNA gene amplicons revealed a clustering of WWTPs, while the

CDs were more separated from each other. Bacteroidetes and Firmicutes dominated the community with a compara-

bly higher abundance of the latter in the processes operating at high ammonia levels. Sequences obtained from the

cel5 and cel 48 clone libraries were also mainly related to the phyla Firmicutes and Bacteroidetes and here ammonia

was a parameter with a strong impact on the cel5 community. The results from the batch cultivation showed similar

degradation pattern for eight of the biogas plants, while two characterised by high ammonia level and low bacterial

diversity, showed a clear lower degradation rate. Interestingly, two T-RFs from the cel5 community were positively

correlated to high degradation rates of both straw and cellulose. One of the respective partial cel5 sequences shared

100 % identity to Clostridium cellulolyticum.

Conclusion: The degradation rate of cellulose and straw varied in the batch tests dependent on the origin of the

inoculum and was negatively correlated with the ammonia level. The cellulose-degrading community, targeted by

analysis of the glycoside hydrolase families 5 (cel5) and 48 (cel48), showed a dominance of bacteria belonging the

Firmicutes and Bacteriodetes, and a positive correlation was found between the cellulose degradation rate of wheat

straw with the level of C. cellulolyticum.

Keywords: Biogas, Cellulose, Community composition, Glycoside hydrolases, cel48, cel5, Terminal restriction fragment

length polymorphism (T-RFLP), Next generation amplicon sequencing

© 2016 The Author(s). This article is distributed under the terms of the Creative Commons Attribution 4.0 International License

(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,

provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license,

and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/

publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Open Access

Biotechnology for Biofuels

*Correspondence: anna.schnurer@slu.se †Li Sun and Tong Liu contributed equally to this work

1 Department of Microbiology, Swedish University of Agricultural Science,

Uppsala BioCenter, P.O. Box 7025, 750 07 Uppsala, Sweden

Full list of author information is available at the end of the article

Page 2 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

Background

Anaerobic digestion (AD) transforms biodegradable

organic materials into a renewable energy resource,

biogas, which can be used for production of vehicle fuel

and/or for combined heat and electricity generation [1].

Furthermore, the residue after digestion is rich in nutri-

ents and can be used as a fertiliser in crop production [2].

In a number of studies, production of biogas has been

shown to offer significant advantages over other forms

of bioenergy production and it has been rated one of

the most energy-efficient and environmentally benefi-

cial technologies for bioenergy production [3, 4]. Biogas

can be produced from many different types of materials,

including various types of waste streams from the food

and feed industry, sludge from wastewater treatment

plants, plant residues and manure from agriculture and

energy crops [5, 6]. Plant-based biomass is very interest-

ing in this regard, with lignocellulosic residues being the

most promising material as these do not compete directly

with food and feed production [7]. They include residues

of agricultural plants, e.g. stalks, straw, husks, cobs, etc.

The total amount of lignocellulosic residues accumulated

annually in the world is estimated to be at least 10 billion

tons [8].

Unfortunately, biogas production from lignocellulose-

rich materials poses some challenges, as the complex

structure, consisting of cellulose, hemicellulose and

lignin cross-linked in a matrix structure, is very resist-

ant to microbial degradation [9–13]. These difficulties

can be overcome to some degree by a pre-treatment that

breaks up the complex structure and makes the mate-

rial more accessible for degradation, but the degradation

rate and biogas yields are still typically rather low [7].

Moreover, many pre-treatments are energy- and cost-

intensive, limiting large-scale application. Alternative less

energy-consuming approaches to improve the degrada-

tion of lignocellulosic materials include management of

the biogas process to ensure growth of microorganisms

with efficient degradation capabilities, for example by

adjustment of the solid retention time, using a two-stage

approach or by a co-digestion approach as reviewed by

Sawatdeenarunat and co-workers [7]. Bioaugmentation

with efficient cellulose-degrading bacteria and addition

of enzymes have also been suggested as promising meth-

ods to increase methane production from lignocellulosic

materials [14, 15].

The microbial degradation of organic materials for

biogas production requires at least four steps: hydroly-

sis, fermentation, acetogenesis and methanogenesis [16,

17]. In the first step, various hydrolytic microorganisms

degrade complex organic polymers to monomers such as

amino acids and sugars. In this step, microbes responsi-

ble for cellulose degradation use either free extracellular

or cell-anchored enzyme complexes including cellu-

losomes, the latter more commonly found in anaerobic

environments [10, 18]. A recent survey of around 1500

complete bacterial genomes revealed that  ~38  % of the

sequenced bacterial genomes encoded at least one cel-

lulase gene, with a small fraction containing more than

three cellulases, a prerequisite for effective degradation

of natural cellulose [19]. The genes necessary for degra-

dation of cellulose have been found in bacteria belonging

to several different phyla: Actinobacteria, Fimicutes, Bac-

teroidetes, Thermotogae, Choloroflexi and Proteobacte-

ria [19–21]. Cellulose-degrading bacterial communities

specifically in biogas processes have been investigated by

various methods. These include cultivation [22–27] and

different molecular techniques targeting bacterial groups

involved specifically in hydrolysis/acetogenesis [28] and

functional genes, i.e. the glycoside hydrolase [29], or

targeting the overall bacterial community [30–34]. The

majority of cultivated cellulose degraders isolated from

different anaerobic environments mainly belong to the

order Clostridiales [10, 20, 21]. Bacteria from this order

have also been shown to dominate in various AD pro-

cesses operating with various lignocellulosic materi-

als, such as wheat straw and cattle manure [31], maize,

green rye and chicken manure [35], and maize straw and

hay [36]. Clostridiales is also suggested to be of impor-

tance, specifically in the hydrolysis step, based on results

obtained using a metagenomic approach [30]. In addition

to Clostridiales, bacteria belonging to the order Bacteroi-

dales have been suggested to be involved in the degrada-

tion of lignocellulose materials, such as straw and hay, in

biogas processes [31, 36].

A range of studies have examined different technical/

chemical/thermal methods for enhanced degradation of

lignocellulosic materials and lately knowledge regarding

the microbiological mechanisms, and organisms involved

in the degradation of lignocellulose have also increased.

However, the correlation between the structure of the

cellulose-degrading community in a biogas process and

efficient degradation is still unclear and requires further

research. The aim of this study was, therefore, to search

for correlations between the degradation rate of cellu-

lose and straw and the bacterial community structure,

including potentially cellulose-degrading bacteria. An

additional aim was to obtain further information about

the cellulose-degrading bacterial population in different

AD processes. The degradation of straw and cellulose was

Page 3 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

analysed using a batch culture approach and with inocu-

lums from ten different Swedish biogas plants, operat-

ing with different substrates and with different operating

parameters. These inoculum samples were analysed both

before and after the batch cultivation to determine the

composition of potential cellulose-degrading bacteria by

targeting the glycoside hydrolase genes of family 5 and

family 48  glycoside hydrolases. The total bacterial com-

munity in the different biogas plants was also analysed,

using next generation amplicon sequencing targeting the

16S rRNA gene.

Results and discussion

Characterisation of inoculum

The inoculum samples investigated in this study origi-

nated from four co-digestion plants processing vari-

ous combinations of substrates (CD01–04) and six

wastewater treatment plants processing mixed sludge

(WWTP01–06). In general, the co-digesters had a longer

hydrolytic retention time (HRT) and higher levels of

VFA and total ammonium-nitrogen than the WWTP

(Table  1). However, one co-digestion plant (CD03) was

more similar to the WWTP in this respect. Moreover, the

free ammonia level in the three plants CD01, 02 and 04

was considerably higher (>0.218 g L−1) than in the other

plants (<0.063 g L−1).

BMP analysis

The final methane potential achieved with inoculum

from the different industrial-scale biogas plants varied

from 307 ± 54 to 376 ± 8 N mL CH4 g VS−1 for cellulose

and from 233 ±  38 to 316 ±  37 N mL CH4 g VS−1 for

straw (Table 2). The degradation rate and the time needed

to reach the final potential also varied, with a clearly

lower degradation rate in the tests initiated with inocu-

lum from CD01 to 02. The time needed to reach 50  %

of the final potential with these two inoculum samples

was 18–47 days and 28–45 days for cellulose and straw,

respectively, while for the other inoculum samples the

corresponding values were 5–8 and 8–19, respectively.

The methane potential values obtained for cellulose and

straw were in line with those reported in previous stud-

ies, illustrating that the inoculum samples evaluated in

this study came from fully functional biogas plants [37–

39]. Some of the tests showed large standard deviation,

most likely due to the non-homogeneous character of

Table 1 Operating data for the 10 industrial-scale biogas plants investigated in this study

CD 01–04 co-digestion plants, WWTP 01–06 wastewater treatment plantsa Total solidsb Volatile solidsc Hydraulic retention timed Temperaturee Total nitrogenf Volatile fatty acidsg Organic loading rateh Total ammonium nitrogeni Free ammonia, calculated according to Hansen et al. [109]j Source-separated municipal organic waste

Digester

code

TSa

(%)

VSb

(%)

HRTc

(day)

TMd

(°C)

Tot Ne

(g L−1 ww)

pH VFAf (g L−1) OLRg (vs

g L−1 day−1)

TANh

(g L−1 ww)

Ammoniai

(g L−1 ww)

Major sub-

strate

CD01 5.7 3.5 45 38 8.7 7.8 1.3 3.0 4.6 0.365 SSMOWj,

slaugh-

terhouse

waste

CD02 4.8 3.8 55 38 9.1 7.8 0.8 2.9 5.1 0.408 Thin stillage

CD03 2.6 2.0 30 37 2.6 7.3 <0.1 3.0 0.9 0.022 SSMOW

CD04 7.1 1.3 70 38 6.2 7.7 3.0 3.0 3.5 0.218 Grass, wheat-

based

stillage

WWTP01 2.7 1.9 17 38 2.5 7.3 <0.1 2.4 1.4 0.036 Mixed sludge

WWTP02 3.5 2.5 23 37 2.9 7.5 0.2 3.1 1.6 0.063 Mixed sludge

WWTP03 2.4 1.9 18 38 1.8 7.3 0.1 2.0 0.9 0.022 Mixed sludge

WWTP04 3.4 2.4 30 37 2.9 7.5 0.3 1.6 1.5 0.058 Mixed sludge

WWTP05 2.5 2.1 22 37 2.1 7.3 0.1 2.8 1.1 0.027 Mixed sludge

WWTP06 2.1 5.5 26 34 1.8 7.3 <0.1 1.1 1.1 0.023 Mixed sludge

Page 4 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

some inoculum samples making it difficult to divide them

evenly between the triplicate bottles. It is also worth not-

ing that the inoculum:substrate ratio inadvertently dif-

fered slightly between the different inoculum samples,

but was still within the range suggested as optimal for a

BMP test, i.e. 2–4. However, this difference could poten-

tially have had an impact on the degradation rate and

possibly also on the final potential in the test. The ratio

for the inoculum from WWTP04–06 was adjusted to 3.8,

3.2 and 2.4, respectively, while the rest of the tests started

with a ratio of 2. Even so, the degradation rate obtained

in the tests of CD01–02 was still lower than for CD03–

04 and WWTP01–03, despite all these sharing the same

ratio.

Bacterial communities

Diversity indicesAnalysis of the bacterial communities in the ten indus-

trial-scale biogas plants by 454 pyrosequencing resulted

in 36,523 sequences after quality trim and chimera check,

with a range from 2573 to 4915 sequences per sample.

Unique barcode was assigned to each replicate of one

sample. The triplicate sequencing analysis was evaluated

with unweighted UniFrac principal coordinate analysis

(PCoA), where no outlier was observed (data not shown).

The triplicates were then pooled in silico and randomly

subsampled according to the sample having the lowest

number of sequences (2500 sequences per sample). The

number of OTUs per sample ranged from 52 to 258, with

a comparatively lower value for the CD plants (Table 3).

The rarefaction curve revealed the same general trend,

i.e. a lower number of OTUs in the co-digestion plants

compared with the WWTP (Table  3; Fig.  1). Based on

the observed species and the Chao1 index, the sequenc-

ing covered 64.7–89.7 % of the total bacterial community.

The three co-digestion plants CD01, 02 and 04 had low

diversity and evenness of the bacterial community, as

indicated by low values of species richness, Shannon and

Simpson index. However, WWTP04 had similarly low

values in this regard (Table 3).

The low diversity in CD01, 02 and 03 correlated with

the comparably higher level of ammonia in these digest-

ers (two-sample t test and nonparametric Monte Carlo

permutations, n =  999, P  <  0.01). In biogas plants pro-

cessing protein-rich materials, the ammonia released

during the AD process is known to have a strong impact

on both the community structure and the diversity, most

probably because of the inhibitory effects of ammonia

[40–42]. High evenness and richness have been suggested

to be associated with good conversion efficiency of fatty

Table 2 Methane potential and time for degradation of straw and cellulose obtained in biochemical methane potential

tests using inoculum from different biogas plants co-digesting different substrates (CD) or sludge from wastewater treat-

ment plants (WWTP)

Inoculum Cellulose Straw

Days to reach  % of the final potential Final potential Days to reach  % of the final potential Final potential

100 % 80 % 50 % 100 % 80 % 50 %

CD01 41 24 18 319 ± 24 110 56 28 258 ± 47

CD02 110 71 47 307 ± 54 75 62 45 233 ± 38

CD03 12 6 5 347 ± 15 60 23 8 316 ± 37

CD04 20 10 8 348 ± 24 26 10 8 274 ± 17

WWTP01 57 15 7 350 ± 7 110 44 19 290 ± 9

WWTP02 36 13 7 314 ± 34 57 24 14 240 ± 38

WWTP03 27 11 6 322 ± 7 75 25 13 310 ± 76

WWTP04 29 15 8 325 ± 8 60 20 11 277 ± 11

WWTP05 45 9 6 376 ± 8 59 30 10 281 ± 32

WWTP06 49 13 8 324 ± 13 135 39 18 296 ± 8

Table 3 Summary of  observed OTUs, Chao1, Shannon

and Simpson index in 10 industrial-scale biogas plants

CD 01-04 co-digestion plants, WWTP 01-06 wastewater treatment plants

Sample Chao 1 OTUs Shannon Simpson

CD01 58 52 1.937 0.456

CD02 94 69 3.111 0.763

CD03 147 120 5.042 0.947

CD04 109 96 3.619 0.767

WWTP01 294 227 5.851 0.956

WWTP02 215 187 4.737 0.861

WWTP03 304 258 5.823 0.930

WWTP04 209 135 3.088 0.609

WWTP05 354 244 5.899 0.955

WWTP06 280 242 5.857 0.930

Page 5 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

acids to methane [43]. High initial evenness has also been

shown to be important for preserving functional stability

in microbial communities in general [44]. In the present

study, CD01, 02, 04 and WWTP04 had lower evenness

and diversity than the other digesters, implying that these

biogas plants might have a comparatively lower poten-

tial to reach high degradation efficiencies. Indeed in the

BMP tests started with inoculum from CD01 to CD02

a comparably lower methane production rates was seen

for both substrates (Table  2). However, both CD04 and

WWTP04 showed higher degradation rates, similar to

the rates obtained with inoculums having a higher bacte-

rial diversity (CD03, WWTP01–3 and WWTP05–6).

Phylogenetic analysis across samplesThe phylogenetic composition, as determined by PCoA

of unweighted UniFrac matrices (Fig. 2a), clearly revealed

a clustering of the WWTPs, which suggests close phylo-

genetic distance within these different plants, while the

CDs were more separated from each other and from

the WWTPs. Considering the relative abundance, as

revealed by the weighted UniFrac matrices, the plant

CD03 was more closely related to the WWTPs, while

the other three CDs plants were still distantly separated

from each other and from the WWTPs (Fig.  2b). This

confirms findings by Sundberg et al. [45] of two separate

clusters distinguishing co-digestion plants and plants

processing sewage sludge. In general, Firmicutes and

Bacteroidetes were identified as predominant phyla in

the industrial-scale biogas plants studied here, but other

phyla were also present (Fig.  3). For example, the phyla

Chloroflexi, Proteobacteria and OP8 were more associ-

ated with plants processing sewage sludge (Welch’s t-test,

P  <  0.01), while the co-digesting CD01 and 02 biogas

plants contained a large fraction of sequences belonging

to a unclassified cluster at phylum level. Bacteroidetes

and Firmicutes are the two dominant phyla commonly

found in various AD processes [21, 28, 31, 35, 36, 46, 47].

The phylum Firmicutes was detected in all biogas

plants, but at different relative abundance, from 5.2 to

67 % (Fig. 3). Within this phylum, the class Clostridia was

dominant, with relative abundance of 4.4–64  %. Other

classes were also present but at lower levels, such as

Bacilli (<2.7  %) and Erysipelotrichi (<0.4  %) (Additional

file 1: Figure S1). The order Clostridiales (3.8–56.3 %) was

dominant within the clostridia, while Thermoanaerobac-

terales (<0.6  %) and unclassified orders such as MBA08

and SHA-98 comprised the rest of this class (Additional

file  1: Figure S2). In anaerobic environments, Clostridi-

ales has been reported as the main cellulose degrader

[10]. This order has frequently been recovered from vari-

ous digesters operating with mono- and co-digestion of

lignocellulosic materials [31, 32, 46, 48–50]. However,

the class Clostridia also contains proteolytic members

and replacement of cellulolytic clostridia with proteo-

lytic members has been observed when using protein-

rich material as substrate with an inoculum originating

from a biogas plant processing pig slurry and maize silage

0

50

100

150

200

250

300

0 500 1000 1500 2000 2500

CD01

CD02

CD03

CD04

WWTP01

WWTP02

WWTP03

WWTP04

WWTP05

WWTP06

Fig. 1 Rarefaction analysis of bacterial communities in 10 industrial-scale biogas plants. CD 01–04 co-digestion plants, WWTP 01–06 wastewater

treatment plants

Page 6 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

[51, 52]. A genus Caldicoprobacter (OTU 481, Additional

file  1: Figures S4, S5) within Clostridiales was found at

a higher relative abundance (0.68  %) in CD01, 02 and

04 compared to the rest samples (<0.04  %). This genus

contains several xylanolytic bacteria [53–55]. The order

MBA08 was detected in four samples in the present study

CD01 (2.8 %), CD02 (1.5 %), CD04 (6.8 %) and WWTP06

(0.4 %) (Additional file 1: Figures S2, S5). This cluster was

first identified in a thermophilic laboratory-scale digester

[56] and later also in other thermophilic digesters [31,

57]. The presence of representatives from this cluster in

the mesophilic biogas plants included in this study sug-

gests that MBA08 contains organisms growing at wide

range of temperatures. The order SHA-98 was observed

a b

Fig. 2 Phylogenetic distance between samples as determined by a unweighted and b weighted UniFrac principal coordinate analysis (PCoA)

(red co-digestion plants, blue wastewater treatment plants)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100% Cloacimonetes

Unclassified

Bacteroidetes

Firmicutes

Ac�nobacteria

Chlorobi

Chloroflexi

Hyd24-12

OD1

OP8

OP9

Proteobacteria

SAR406

Spirochaetes

Synergistetes

Tenericutes

Verrucomicrobia

Minor groups

Fig. 3 Relative abundance of bacterial 16S rRNA gene at phylum level in 10 industrial-scale biogas plants. CD 01–04 co-digestion plants, WWTP

01–06 wastewater treatment plants

Page 7 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

at comparably higher levels in the high ammonia digest-

ers CD01 and CD02, with relative abundance of 5.5 and

8.8 %, respectively. The levels in the rest of the digesters

were 0.3–3.2 %. This order was represented by two domi-

nant OTUs, OTU11 representing the main OTU in the

high ammonia digesters CD01/02/04 (up to 7.9  %) and

OTU 35, comparably more abundant in all other digest-

ers (up to 2.6 %) (Additional file 1: Figure S5). In line with

previous findings by Sundberg and co-workers [45], the

family Clostridiaceae had higher relative abundance in

WWTPs (2.2–5.4 %) than in CDs (0.2–0.9 %) (Welch’s t test, P < 0.01) (Additional file 1: Figure S3). Notably, the

genus Gallicola, classified to the proposed family Tis-

sierellaceae [58], was present at high relative abundance

in plant CD04 (46.2 %) (Additional file 1: Figure S5). This

family was not detected in any other biogas plant except

WWTP06 (0.1  %) (Additional file  1: Figure S4). The

first representative of the genus Gallicola was isolated

from chicken manure and it has been shown to grow on

purines, such as uric acid, xanthine, 6,8-dihydroxypurine,

guanine and hypoxanthine [59].

The phylum Bacteroidetes had a relative abundance

of 10.1–69.2  % (Fig.  3). Within this phylum, the order

Bacteroidales (class Bacteroidia) was dominant in all

samples (Additional file  1: Figure S2, S3). Members of

the Bacteroidetes are able to degrade various polysac-

charides [60, 61]. In a study with batch fermentation of

straw and hay, the relative abundance of Bacteroidetes

was higher at the end than at the beginning of the batch

cultivation, indicating the importance of this phylum for

degradation of lignocellulose [36]. An increase in rela-

tive abundance of this phylum has also been observed in

response to straw addition in a laboratory-scale digester

initially operating with cattle manure [31]. However, the

Bacteroidetes have been shown to decrease in abun-

dance at high levels of ammonia [41, 42, 51, 62]. In line

with this, the lowest relative abundance of this phylum

was seen for CD01 and CD04, both with relatively high

ammonia levels (Fig. 3). However, CD02 also had a high

ammonia level but had a similar relative abundance of

Bacteroidetes as the low ammonia digesters. At fam-

ily level, the Bacteroidaceae (up to 2.3  %), Marinila-

biaceae (up to 0.2  %), Porphyromonadaceae (up to

9.2  %), Rikenellaceae (up to 0.7  %) and a few unclassi-

fied families were present (Additional file 1: Figure S3).

One unclassified family, SB-1, was also observed, with

higher abundance in CD03 (12.6  %) and WWTP01-06

(4.0–65.3 %) than in CD01, 02 and 04 (<0.2 %), possibly

suggesting sensitivity to high ammonia levels (Table  1;

Additional file  1: Figure S5). In addition, the Porphy-

romonadaceae were observed at relative abundance of

9.8  % in CD02 and 8.0  % in CD03, while other orders

represented less than 2.0 %.

The candidate phylum Cloacimonetes (formerly

WWE1) was observed at levels of less than 0.1 % in plants

CD01 and CD02, but ranged from 1.0 to 14.2  % in the

other eight biogas plants (Fig. 3; Additional file 1: Figure

S5). Representatives of this phylum were first discovered

in AD plants processing sewage sludge [63], but have

since been found in a full-scale plant fed energy crop

(mainly maize silage) [32] and a laboratory-scale digester

fed cattle manure as the sole substrate or co-digested

with wheat straw [31]. A study using stable isotopes sug-

gested that members of this phylum are engaged in either

cellulose hydrolysis or uptake of cellulose fermentation

products [64]. The uncultured cluster at phylum level

SAR406 was observed in CD03 with a relative abundance

of 8.2  %, while the level in CD01-02 and in all WWTP

was less than 1 % (Fig. 3; Additional file 1: Figure S5). The

candidate name Marinimicrobia has been proposed for

this cluster [65] and preliminary genome analysis sug-

gests that members within this phylum are proteolytic

and amino acid degraders [64, 66]. The phylum Proteo-

bacteria was present at a level of 3.4–9.6 % in the WWTP,

but less than 0.3 % in the CDs (Fig. 3; Additional file 1:

Figure S5). This phylum has previously been found in var-

ious digesters processing sewage sludge [45, 67–69], but

recently also in mono-digestion of fodder beet silage [46]

and in a process co-digesting food waste with Chinese sil-

vergrass [70]. Representatives of the uncultured phylum

Hyd24-12 corresponded to 12–25  % in WWTP 01, 05

and 06, but were much less abundant in the other biogas

plants (<2.3 %). This uncultured cluster has been found in

other methanogenic digesters [71, 72], but its function in

the methanogenic environment is still unknown [73]. The

phylum Chloroflexi was represented with relative abun-

dance of 6.5–11.5  % and 0.1–3.9  % in the WWTPs and

CDs, respectively, and was dominated by an uncultured

genus T78 (5.4–9.2 % in WWTPs and 0.1–3.0 % in CDs)

(Fig.  3; Additional file  1: Figure S5). This phylum has

been previously found in anaerobic digesters processing

sewage sludge [74] and in co-digestion of whey perme-

ate and cow manure [75], and members have been sug-

gested to be carbohydrate utilisers [76, 77]. In addition,

an uncultured cluster was found at high levels in CD01,

02 and 04, with a relative abundance of 73.7, 45.2 and

11.4 %, respectively (Fig. 3; Additional file 1: Figure S5).

This large cluster was represented by one single OTU

and had 84 % similarity based on 16S rRNA gene with an

uncultured Clostridium (OTU1). The dominance of the

community by such a large fraction represented by a sin-

gle OTU is somewhat surprising and, to our knowledge,

has not been reported previously for biogas digesters.

Considering the feedstock for these biogas plants, this

OTU might represent a protein-fermenting bacterium

enriched by the protein-rich feedstock. Alternatively, this

Page 8 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

bacterium has a selective advantage at the high levels of

the free ammonia in these digesters (0.218–0.408 g L−1).

Ammonia has in previous studies been shown to have a

strong selective pressure on the microbial community

[41, 42].

T-RFLP

cel5The cellulolytic community structures were investi-

gated using T-RFLP combined with clone library analysis

(Fig. 4a). Analysis of the inoculum samples revealed that

T-RF 275  bp was present in all digesters except CD04,

with a higher relative abundance in CD01–02 (51.1–

52.1  %) than in the other plants (<21.1  %). T-RF 362  bp

was also present at higher levels in CD01–02 (14.0–

24.8 %) than in the other samples, with detectable levels

only in samples WWTP02–04 (1.3–2.9  %). T-RF 85  bp

had the highest relative abundance in CD03 (76.1 %), fol-

lowed by WWTP01–04 (44.3–71.4  %), WWTP05–06

(14.5–15.4 %) and CD01–02 (2.8–9.2 %), while it was not

a

b

0%10%20%30%40%50%60%70%80%90%

100%

CD01

CD01

-cCD

01-s

CD02

CD02

-cCD

02-s

CD03

CD03

-cCD

03-s

CD04

CD04

-cCD

04-s

WW

TP01

WW

TP01

-cW

WTP

01-s

WW

TP02

WW

TP02

-cW

WTP

02-s

WW

TP03

WW

TP03

-cW

WTP

03-s

WW

TP04

WW

TP04

-cW

WTP

04-s

WW

TP05

WW

TP05

-cW

WTP

05-s

WW

TP06

WW

TP06

-cW

WTP

06-s

39638636233628627524021120118617917415613813012210496857874

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

CD01

CD01

-CCD

01-S

CD02

CD02

-CCD

02-S

CD03

CD03

-CCD

03-S

CD04

CD04

-CCD

04-S

WW

TP01

WW

TP01

-CW

WTP

01-S

WW

TP02

WW

TP02

-CW

WTP

02-S

WW

TP0 3

WW

TP03

-CW

WTP

03-S

WW

TP04

WW

TP04

-CW

WTP

04-S

WW

TP0 5

WW

TP05

-CW

WTP

05-S

WW

TP06

WW

TP06

-CW

WTP

06-S

440 432

357 344

328 321

316 301

296 288

277 268

254 247

238 231

224 209

202 198

178 171

164 157

149 141

130 123

116 109

102 94

87 80

73 67

62 54

Fig. 4 T-RFLP profile representing the community of a glycoside hydrolase gene family 5 (cel5) and b glycoside hydrolase gene family 48 (cel48) in

10 industrial-scale biogas plants: CD01–04/c/s and WWTP 01–06/c/s refer to inoculum analysed at the starting point of a batch cultivation and at

the end point using cellulose (-c) and straw (-s) as substrate, respectively

Page 9 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

detected at all in CD04. T-RF 396 bp dominated in CD04

(88.8  %) and WWTP05–06 (66.1–80.8  %), while T-RF

78  bp showed higher relative abundance in WWTP06

(13.1 %), WWTP04 (13.7 %) and WWTP03 (23.9 %), but

was present at lower levels (<8 %) in the other plants.

Incubation with cellulose and straw in the batch test

clearly changed the cel5 TRFLP profile compared to the

one observed at the beginning of the experiment. How-

ever, the response varied both with inoculum and with

substrate (Fig.  4a). In CD01, the relative abundances of

the dominant T-RFs 275 and 362  bp were lower after

the incubation with cellulose (28.3, 14.0  %) compared

to levels at the beginning of the batch test. Instead the

abundance of T-RFs 74, 78, 85, 156 and 396 bp were com-

parably higher (27.2, 6.0, 11.5, 10.6 and 7.4  %, respec-

tively) than at the starting point (0, 1.8, 2.8, 1.2 and 2.6 %,

respectively). For CD02, a similar decrease was seen

for T-RF 362 bp, while with this inoculum T-RF 275 bp

increased to 79.9 %. The other T-RFs (i.e. 74, 78, 85, 156

and 396  bp) that increased to a smaller extent in CD01

were not detected at the end point for this inoculum.

Using straw as substrate resulted in a similar T-RFLP

profile change as for cellulose for both CD01 and CD02.

For CD03 and WWTP01–04, the relative abundance of

the most dominant T-RF 85  bp decreased after incuba-

tion with both cellulose and straw. For WWTP03, the

level of T-RFs 78 and 85 bp was reduced to non-detect-

able, while instead T-RFs 275 and 362  bp increased

to 76.5 and 23.5  % for cellulose (83.6 and 16.4  % for

straw). However, at the end point using both substrates,

T-RFs 74, 78 and 156 bp increased their level for CD03,

WWTP01–02 and WWTP04, except T-RFs 74 and

156 bp for WWTP04c. For WWTP04c, the level of T-RFs

275 and 362  bp increased. T-RF 122  bp increased in

CD03c and WWTP01–02c, while T-RF 130 bp increased

in WWTP01–02c. For CD04 and WWTP05–06, the

most pronounced change was for the dominant T-RF

396  bp, which decreased to 7.9, 8.1 and 7.8  %, respec-

tively. Instead, for CD04c, T-RFs 211, 85, 156 and 74 bp

increased to 44.1, 26.5, 11.9 and 8.2  %, respectively. In

addition, for CD04 s T-RF 78 bp increased to 15.8 %. For

WWTP05, T-RFs 74, 78, 85 and 275 bp increased to 14.1,

10.5, 37.1 and 12.4  %, respectively, for cellulose, while

T-RFs 74, 85 and 275 bp increased to 23.4, 33.7 and 9.5 %,

respectively, for straw. The profile of WWTP06 was simi-

lar to that of WWTP05 but, in addition, T-RFs 156 and

78  bp increased to 16.1 and 30.5  % for WWTP06c and

WWTP06 s, respectively.

cel48The T-RFLP profile obtained by the cel48 primer set of

the inoculum samples showed a high relative abundance

of T-RF 62 in the majority of samples from the WWTPs,

with WWTP01, 02, 03 and 05 having relative abun-

dance of 28.6–81.5  % (Fig.  4b). The level in WWTP04

and WWTP06 was lower, 4.8 and 1.4  %, respectively.

This T-RF was only detected in two CDs (CD02–03) and

at low levels (1.5–2.3  %). For WWTP04, the dominant

T-RFs were instead 141 bp (21.2 %) and 238 bp (38.6 %),

while for WWTP06, the two most dominant T-RFs were

357 bp (48.2 %) and 328 bp (12.8 %). For CD01 and CD04,

the T-RFLP profile was dominated by T-RFs 357 bp (35.4

and 17.4  %) and 328  bp, (30.1 and 48.0  %). For CD02,

the level of T-RF 328 bp was similar (29.9 %), but T-RF

357  bp represented only 1.9  %. Instead, T-RFs 231, 224

and 73 bp were present at a level of 17.9, 11.5 and 10.2 %,

respectively. For CD03, the two most dominant T-RFs

were 247  bp (34.9  %) and 321  bp (33.0  %), which were

present at a relatively low level in all other biogas plants.

In contrast to the cel5 community, the end point cel48

community in most cases was not the same when using

cellulose and straw as substrate (Fig.  4b). For CD01, the

dominant T-RFs 328 and 357 bp decreased in both cases,

but T-RF 316 bp (47.9 %) increased in the cellulose incu-

bations and T-RF 296  bp (53.6  %) increased after incu-

bation with straw. For CD02, the dominant T-RF after

incubation with straw and cellulose was 268 bp (30.5 %)

and 357 bp (61.6 %), respectively. For CD03, the starting

point T-RF 247 bp disappeared for both end points, while

T-RF 321  bp increased in response to cellulose addition

(67.0 %), but declined in abundance when straw was used

as substrate (2.5  %). In addition, T-RF 62  bp increased

to 89.0 % when straw was used. For CD04, T-RF 357 bp

dominated after digestion with cellulose (90.9  %), while

T-RFs 321 bp (38.4 %), 328 bp (16.2 %) and 357 bp (14.1 %)

dominated in the straw cultures. For WWTPs, after

digestion with cellulose, the major peak was T-RF 62 bp

(13.0–94.6  %). In addition, T-RFs 296  bp (78.7  %) and

321  bp (45.1  %) were high in WWTP03 and 04, respec-

tively. For the end point with straw, T-RF 62 bp remained

as the major peak in WWTP02 (38.9) and 06 (74.1  %)

and T-RF 296 bp was also high in WWTP02 (41.4 %). For

WWTP01 s and 04–05 s, the dominant peak changed to

T-RF 164 bp, with a relative abundance of 94.0, 49.3 and

93.6 %, respectively. For WWTP03 s, T-RF 357 bp, with a

level of 36.3 %, was the major peak in this plant.

Clone libraries and phylogenetic analysisSequencing of 118 and 215 clones from cel5 and cel48

libraries resulted in 10 and 11 OTUs, respectively. All

OTUs have low similarity to characterised bacteria, with

one exception of OTU10, which partial sequence shared

100 % identity to Clostridium cellulolyticum. For twenty out

of 21 OTUs the closest cultivated bacterial phyla belonged

to Bacteroidetes or Firmicutes, one OTU was close to Act-

inobacteria (cel48 OTU06, Table  4). This is in agreement

Page 10 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

with previous studies investigating the bacterial community

in biogas digesters using the same primers that all analysed

clones were close to Bacteroidetes and Firmicutes [29].

Although the partial deduced amino acid sequences are as

short as 100 and 130 amino acid residues, respectively, both

trees are supported by high bootstrap values.

cel5T-RF 275  bp, represented by OTU07 and present in all

digesters except CD04, and T-RF 362 bp (OTU08), highly

abundant in CD01 and CD02, were both most closely

related to an uncultured bacterium (AEV59723), with

an identity of 77.0 and 76.0 %, respectively. Mahella aus-traliensis (phylum Firmicutes) was the closest cultivated

bacterium, with an identity of 57.4 and 58.4  %, respec-

tively (Table 4, Fig. 5). M. australiensis is able to ferment

different carbohydrates, including cellobiose [78]. It has

previously been detected in thermophilic digesters oper-

ating with chicken manure, where it is suggested to be

sensitive to ammonia [79, 80]. This species has also been

observed in the cel5 community in a CSTR digester fed

cow manure and steam-exploded straw operating at 44 °C

[29]. T-RF 85  bp, present at high level in most digest-

ers and represented by OTU02 and OTU03, was closely

related to an uncultured bacterium (AGW24153), with

an identity of 100 and 89.1 %, respectively. This OTU has

previously been observed during anaerobic digestion of

cow manure and steam-exploded straw at 37 °C [29]. The

Table 4 Clone sequences of cel5 and cel48 obtained from industrial-scale biogas processes

a Most closely related to uncultured bacterium AGW24153 (identity: 100 %) from laboratory-scale anaerobic reactorb Most closely related to uncultured bacterium AGW24153 (identity: 89.1 %) from laboratory-scale anaerobic reactorc Most closely related to uncultured bacterium ACV50344 (identity: 73.3 %) from lignocellulose-based sulphate-reducing bioreactord Most closely related to uncultured bacterium AGO64733 (identity: 75.0 %) from anaerobic digester sludgee Most closely related to uncultured bacterium AEV59723 (identity: 77.0 %) from laboratory biogas digester treating rice strawf Most closely related to uncultured bacterium AEV59723 (identity: 76.0 %) from laboratory biogas digester treating rice strawg Most closely related to uncultured bacterium AGO64695 (identity: 99.0 %) from anaerobic digester sludgeh Most closely related to uncultured bacterium AGO64692 (identity: 75.2 %) from anaerobic digester sludgei Most closely related to uncultured bacterium AGO64682 (identity: 85.6 %) from anaerobic digester sludgej Most closely related to uncultured bacterium AGO64673 (identity: 99.1 %) from anaerobic digester sludgek Most closely related to uncultured bacterium AGO64695 (identity: 99.0 %) from anaerobic digester sludge

Clone T-RFs (bp) Most closely related microorganism Identity (%) Accession number

Cel5

OTU01 56 Ruminococcus callidus 84.8 WP_021681794

OTU02a 85 Echinicola vietnamensis 58.0 WP_015264998

OTU03b 85 Echinicola vietnamensis 61.4 WP_015264998

OTU04c 106 Flavobacterium sp. 66.3 WP_007808671

OTU05 136 Niastella koreensis 73.7 AEV98714

OTU06d 211 Marinilabilia salmonicolor 66.3 WP_036163195

OTU07e 275 Mahella australiensis 57.4 AEE96311

OTU08f 362 Mahella australiensis 58.4 AEE96311

OTU09 375 Clostridium sp. 72.1 WP_033166154

OTU10 396 Clostridium cellulolyticum 100 WP_015924614

Cel48

OTU01 68 Clostridium stercorarium DSM 8532 57.9 AGI39871

OTU02 68 Clostridium sp. Iso6-17a 50.0 ADM52292

OTU03 177 Ruminococcus sp. CAG:254 97.5 CCZ84184

OTU04g 205 Clostridium sp. Iso6-24 79.0 ADM52293

OTU05 238 Ruminococcus sp. HUN007 83.2 WP_049962845

OTU06 238 Streptomyces griseorubens 61.0 WP_037642616

OTU07h 247 Acetivibrio cellulolyticus 71.2 WP_010681059

OTU08 290 Clostridium acetobutylicum 50.5 WP_010964229

OTU09i 321 Ruminiclostridium thermocellum 75.0 ACT46162

OTU10j 328 Clostridium termitidis CT1112 73.8 EMS73539

OTU11k 358 Clostridium straminisolvens JCM 21531 78.8 GAE90081

Page 11 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

closest cultivated bacterium was Echinicola vietnamensis, with an identity of 58 and 61.4 %, respectively. E. vietna-mensis was first isolated from seawater [81], and is able

to hydrolyse starch and grow up to 44 °C with 15 % NaCl.

T-RF 396 bp (OTU10), dominating in CD04, WWTP05

and WWTP06, was verified as C. cellulolyticum (identity

P-OTU05

P-OTU07

OTU06

Marinilabilia salmonicolor (WP 010663917)

P-OTU02

Flavobacterium johnsoniae UW101 (YP 001193127)

Echinicola vietnamensis DSM 17526 (YP 007223444)

OTU03

OTU02

P-OTU08

Flavobacterium sp. (WP 007808671)

OTU04

Niastella koreensis (WP 041348462)

OTU05

Acetivibrio cellulolyticus CD2 (ZP 09465414)

Clostridium cellulovorans 743B (YP 003842949)

Eubacterium siraeum 70/3 (CBK96866)

P-OTU01

OTU01

Ruminococcus callidus (WP 021681794)

Eubacterium cellulosolvens 6 (ZP 10167476)

P-OTU03

uncultured microorganism buffalo rumens (ACA61144)

uncultured microorganism buffalo rumens (ACA61160)

Clostridium cellulolyticum (WP 015924614)

OTU10

Clostridium sp. KNHs205 (WP 033166154)

OTU09

P-OTU04

uncultured organism human gut (ADD61911)

Marinilabilia salmonicolor JCM 21150 (ZP 11227339)

Bacteroides salanitronis DSM 18170 (YP 004257951)

OTU07

OTU08

Mahella australiensis 50-1 BON (YP 004463133)

P-OTU06

uncultured bacterium biogas digester (AEV59735)100

100

99

85

96

100

98

100

92

99

98

44

93

98

60

74

95

65

72

78

4466

77

42

55

36

28

37

30

45

30

26

15

23

0.1

Fig. 5 Phylogenetic tree of sequences from glycoside hydrolase gene family 5 retrieved from different industrial-scale biogas processes. OTUs

operational taxonomic units identified in this study, P-OTUs operational taxonomic units identified in Sun et al. [29])

Page 12 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

100  %). This is a well-studied non-ruminal mesophilic

cellulolytic bacterium, initially isolated from decayed

grass [82] that has the ability to degrade cellulose and

hemicellulose to acetate, ethanol, lactate and H2 [83].

Moreover, it has been shown that bioaugmentation with

C. cellulolyticum can increase the cellulose degrada-

tion efficiency of wheat straw during batch cultivation

[84]. T-RF 78 bp, which increased in most of the digest-

ers in response to cellulose and straw, was unfortunately

not found in the clone library in this study. However,

in a previous study investigating the cel5 community in

digesters operating with manure and steam-exploded

straw, a T-RF of the same size was shown to correspond

to a clone related to Eubacterium siraeum (49.5 % iden-

tity) [29]. The abundance of the organism in that study

was shown to increase from 14  to 52 % with increasing

temperature from 37 to 52  °C. This bacterium has been

isolated from human faeces and belongs to a genus com-

monly existing in the human gut and involved in cellulose

degradation [85] [86]. T-RF 211  bp, which increased in

the CDs in response to both straw and cellulose, was rep-

resented by OTU06. This OTU was related to an uncul-

tured bacterium (AGO64733; 75.0 % identity) previously

identified in a laboratory-scale reactor operating with

manure and straw [29] and grouped with P-OTU05 and

07, identified in the same study (Fig. 5). The closest cul-

tured relative, Marinilabilia salmonicolor (phylum Bac-

teroidetes, 66.3 % identity) was isolated in seawater and

has been shown to have the ability to degrade cellulose,

monomeric sugars and starch [87].

cel48T-RF 238  bp (OTU05), dominating in WWTP04, was

most closely related to Ruminococcus sp. (identity

82.3  %). T-RFs 358  bp (OTU11) and 328  bp (OTU10),

dominating in WWTP06, were most closely related to

uncultured bacterium AGO64695 (99.0  % identity) and

AGO64673 (99.1 % identity) as identified previously [29].

The closest cultivated relative was Clostridium stramini-solvens JCM 21,531(78.8 % identity) and Clostridium ter-mitidis CT1112 (73.8 % identity) for OTU11 and OTU10,

respectively (Fig.  6). Both C. straminisolvens and C. termitidis have previously been shown to have the abil-

ity to degrade a variety of monomeric sugars, as well as

cellobiose [88, 89]. Clostridium straminisolvens has been

detected in various types of biogas digesters, such as an

anaerobic thermophilic digester fed municipal waste [24]

and an anaerobic mesophilic digester fed pig manure,

rice straw [90, 91]. The T-RF with size 238  bp was also

found in digester WWTP03 at an abundance of 11.9 %. In

this case, the corresponding clone (OTU06) was closely

related to Streptomyces griseorubens (61 % identity). This

result highlights one drawback of T-RFLP analysis, with

several different sequences resulting in the same T-RF

size, and the importance of combining this method with

a clone library. S. griseorubens has been isolated from

soil in both India and China, and the ability to degrade

lignocellulose has been demonstrated [92–94]. Clones

OTU07 and OTU09 represented T-RFs 247  bp and

321 bp, respectively, which were present at comparatively

high levels in CD03. For these, two uncultured bacteria

(AGO64692; identity 75.2  % and AGO64682; identity

85.6  %), respectively, both retrieved from anaerobic

digester sludge, had the highest similarity. The closest

known bacteria were Acetivibrio cellulolyticus (71.2  %

identity) and Ruminiclostridium thermocellum (75  %

identity), respectively. A. cellulolyticus can utilise cellu-

lose, cellobiose and salicin and has previously been found

in a methanogenic enrichment culture from munici-

pal sewage sludge [95]. R. thermocellum, synonym of

Clostridium thermocellum, is a well-studied anaerobic

cellulose-degrading thermophilic organism which has

been suggested as a potential candidate for different bio-

technological applications [96]. It has been isolated from

plants and from cow and horse manure [97] and has been

demonstrated to play a key role in cellulolytic degrada-

tion in different types of biomethane production digest-

ers [98–100]. After incubation with straw, T-RFs 296 and

164 bp increased in abundance in CD01 and WWTP01,

respectively. These two T-RFs could not be found in the

clone libraries. However, in a previous study of labora-

tory-scale digesters operating with manure and straw at

44 °C [29], both T-RFs were found and suggested to rep-

resent species closely related to A. cellulolyticum (74.3 %

identity) and Ruminococcus champanellensis (61.8  %

identity), respectively. R. champanellensis is able to utilise

cellulose, cellobiose and xylan, but not starch and pectin

[101]. OTU02 (T-RF 68) and OTU08 (T-RF 290) grouped

together in the phylogenetic tree (Fig.  6) and showed

similarity to Ruminococcus, but with relatively low iden-

tity (50.0 and 50.5  %, respectively). OTU04, 09 and 11

grouped together with P-OTU03 and 08 from a previous

study investigating potential cellulose-degrading bacteria

in digesters operating with straw and cow manure [29].

Correlation of microbial community structure with process

parameters

To identify possible correlations between microbial com-

munity composition within the inoculum and biogas

process operating parameters (Table 1) and batch diges-

tion performance (Table  2) using cellulose and straw as

substrate, CCA was performed. On including the process

parameters (free ammonia, OLR, HRT and VFA) and

process performance (inverse of days needed to reach

50 and 80 % of the final methane potential for cellulose

and straw, C50/80 and S50/80), the first two dimensions

Page 13 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

P-OTU08

OTU11

OTU04

P-OTU03

OTU09

Clostridium thermocellum DSM 1313 (ACT46162)

Clostridium straminisolvens (ACV92097)

Clostridium sp. Iso6-17a (ADM52292)

OTU07

Acetivibrio cellulolyticus CD2 (ZP 09463651)

P-OTU07

P-OTU01

Streptomyces griseorubens (WP 037642616)

OTU06

Clostridium cellulovorans 743B (ZP 07630533)

P-OTU09

Clostridium cellulolyticum H10 (YP 002505088)

Clostridium josui (BAA32430)

OTU01

Clostridium stercorarium DSM 8532 (AGI39871)

uncultured bacterium sulfate-reducing reactors (ACV50351)

Clostridium termitidis CT1112 (EMS73539)

P-OTU02

OTU10

Clostridium acetobutylicum ATCC 824 (NP 347547)

OTU08

OTU02

Ruminococcus flavefaciens FD-1 (ZP 06145360)

P-OTU06

Ruminococcus albus 7 (YP 004105715)

P-OTU05

Ruminococcus champanellensis 18P13 (CBL17316)

P-OTU04

Ruminococcus sp. HUN007 (WP 044974973)

OTU05

Ruminococcus sp. CAG:254 (WP 022050983)

OTU0399

99

79

100

97

92

99

96

70

81

99

62

92

43

21

38

44

95

76

76

52

98

92

66

85

37

43

10

29

17

42

26

98 44

0.2

Fig. 6 Phylogenetic tree of sequences from glycoside hydrolase gene family 48 retrieved from different industrial-scale biogas processes. OTUs

operational taxonomic units identified in this study, P-OTUs operational taxonomic units identified in Sun et al. [29]

Page 14 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

of the CCA plot explained around 58.7  % (16S ampli-

con sequencing; Fig.  7a) and 79.6  % (cel5, Fig.  7b) of

the variation in relative abundance of OTUs and T-RFs,

respectively. For cel48, no clear clustering of community

structure or clear influence of parameter was found (data

not shown).

The CCA plot of amplicon sequencing data (Fig.  7a)

showed a similar pattern to the weighted UniFrac PCoA

plot (Fig.  2b). The free ammonia concentration was

identified as the one important factor influencing the

community structure, determined by analysis of both

the 16S rRNA and cel5 gene. Batch degradation perfor-

mance, i.e. the time to reach 50 and 80 % degradation of

both cellulose and straw, on the other hand, was nega-

tively associated with the level of free ammonia in the

original process in both analyses. At present, not much

information is available in the literature regarding the

impact of ammonia on the degradation of cellulose and

the obtained results so far are contradictory [21]. In the

16S rRNA amplicon dataset, a few OTUs were found to

be positively correlated with ammonia level, including

the highly dominant OTU in both digester CD01 and

CD02. This OTU was also negatively correlated with the

batch process performance. For the cel5 community, the

dominant T-RFs 362  bp showed a positive correlation

with ammonia concentration, while the T-RFs 396 and

85 bp were correlated with process performance. One of

these T-RFs (TRF 396) represented a clone with 100  %

similarity to C. cellulolyticum, recently shown to increase

the cellulose degradation efficiency of wheat straw during

bioaugmentation [84]. The batch test results cannot be

completely transferred to the industrial-scale process, i.e.

the degradation in the batch test might not be the same

as in the continuous full-scale system. Still the microbial

composition in the inoculums, shaped by the process

parameters in the full-scale plant, will most likely impact

on the outcome of the BMP test, as has been shown in

previous studies [102].

Conclusions

The overall bacterial communities within the investigated

co-digestion and WWTP plants were separated from

each other, probably owing to differences in substrate

and/or operating parameters used by these two groups

of biogas plants. Moreover, the diversity was lower in CD

compared with WWTP plants. Among the ten plants,

two showed clearly lower degradation efficiency of

straw and cellulose, measured during batch cultivation.

These two plants also had the lowest bacterial diversity

(species richness and evenness) and the highest level of

ammonia. Ten of 21 OTUs obtained from clone libraries

based on the glycosidase hydrolase gene sequence were

mainly distantly related to known organisms, while the

rest were related to partial sequences of unknown bac-

teria, although according to the phylogenetic analysis

still related to saccharolytic or cellulolytic bacteria. Sta-

tistical analysis identified ammonia as a parameter with

a strong impact on the cel5 community, while no clear

trend could be seen for the cel48 community. This indi-

cates that ammonia not only influences the methano-

genic community structure in biogas processes, but also

shapes the community of bacteria involved in the hydrol-

ysis step. Interestingly, two dominant T-RFs from the cel5

Fig. 7 Canonical correspondence analysis (CCA) of a the major OTUs

at genus level of the 16S rRNA gene and b the T-RFs of cel5 commu-

nity, within 10 industrial-scale biogas plants. CD 01–04 co-digestion

plants, WWTP 01–06 wastewater treatment plants. Ammonia free

ammonia, OLR organic loading rate, HRT hydraulic retention time, VFA

volatile fatty acids, C50/80 and S50/80: the inverse of days needed

to reach 50/80 % of final methane potential for cellulose and straw,

respectively

Page 15 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

community were positively correlated with batch process

performance, i.e. the degradation efficiency of straw and

cellulose. One of these T-RFs represented a clone with

100  % similarity to C. cellulolyticum, recently shown to

be of importance for the degradation of wheat straw. The

presence of this bacterium was negatively correlated to

the ammonia level, supporting the idea that ammonia

might have a negative impact on the degradation of lig-

nocellulosic material.

Methods

Biogas plants

In total, the main digesters of 10 different industrial-scale

biogas plants were investigated. These digesters were all

operating at mesophilic temperature (range 37–38  °C),

six processing sludge from wastewater treatment plants

(WWTP01–06) and four co-digesting various organic

waste fractions (CD01–04), such as agricultural waste,

source-separated organic municipal household waste and

slaughterhouse waste (Table  1). All investigated biogas

plants had been in operation for several years under

similar conditions regarding substrate and operating

parameters.

Bio-methane potential test

The methane potential of cellulose and straw was deter-

mined by biochemical methane potential (BMP) analysis

[103]. The substrate used in the BMP test was cellulose

(C6663, Sigma-Aldrich, MO, USA) and wheat straw

(mechanical chopped into 1–2  cm). Inoculum was col-

lected from the different biogas plants and incubated at

37 °C for 4–6 days prior to the BMP test to decrease gas

production from endogenous material. The batch anaero-

bic digestion was performed in 309 mL serum bottles, to

which substrate was added corresponding to 3 g volatile

solids (VS) per liter, and the inoculum:substrate ratio was

between 2:1 and 4:1. The final liquid volume was set to

193 mL in all bottles by adding tap water. For each inocu-

lum, triplicate bottles were set up for straw and for cel-

lulose. Triplicate bottles filled only with the inoculum,

corresponding to the same volume as in the test bottles,

were also set up to measure background production of

methane (CH4). All bottles were incubated at 37 °C on a

rotary shaker at 130 rpm. Total gas production and meth-

ane content were continuously monitored over 60  days

by pressure measurement combined with gas sampling

and gas chromatograph (GC) analysis of gas composi-

tion [103]. The accumulated methane production was

calculated and the gas production from the control was

deducted. The volumetric methane value was normalised

to standard temperature (273.15 K) and pressure (1 bar)

using the ideal gas law, and finally expressed as N  mL

CH4 g VS−1 [104]. The days needed to reach 50 and 80 %

of final methane potential was used as a measure of deg-

radation capacity.

Analytical methods

Content of total solids (TS) and VS in the inoculum sam-

ples was measured according to the international stand-

ard methods [105]. Total Kjeldahl nitrogen (TKN) and

ammonium-nitrogen (NH4-N) were analysed according

to the International Standardization Operation (ISO)

methods (ISO 10,694, 1995  and  ISO 13,878, 1998). The

concentration of free ammonia was calculated from the

NH4-N concentration, pH and temperature accord-

ing to Hansen et  al. [109]. The volatile fatty acid (VFA)

content in the digester samples was determined by high-

performance liquid chromatography (HPLC) analysis as

described previously [103].

DNA extraction

Samples for microbial community analysis were taken

from the inoculum on the start day of the BMP test and

at the end of the anaerobic batch test. A 15 mL sample

was withdrawn from each inoculum and each batch test.

The latter samples were designated CD 01–04c, WWTP

01–06c and CD 01–04  s, WWTP 01–06  s, for samples

from incubation with cellulose and straw, respectively.

All samples were stored at −20  °C until the extraction

of DNA. Total genomic DNA was extracted in triplicate

using the FastDNA Spin kit for soil (MP Biomedicals,

Santa Ana, CA, USA) according to the manufactur-

er’s instructions, with the modifications that aliquots

of 200  μL digester sample were used for extraction and

60 μL water was used in the final DNA elution. The con-

centrations of extracted DNA were measured using a

Nano Vue spectrophotometer (GE Healthcare, Bucking-

hamshire, UK).

454-pyrosequencing and 16S rRNA gene sequence analysis

The bacterial communities in inoculum from the 10

industrial biogas plants were investigated by amplifica-

tion of genomic DNA using polymerase chain reaction

(PCR) primers targeting the bacterial 16S rRNA gene and

integrated with 454 Life Sciences adaptors 8F (5′-CCT

ATC CCC TGT GTG CCT TGG CAG TCT CAG CAA

CAG CTA GAG TTT GAT CCT GG-3′) and 515R (5′-CCA TCT CAT CCC TGC GTG TCT CCG ACT CAG

NNN NNN NNT TAC CGC GGC TGC T-3′ [106]. Each

PCR contained 12.5 μL of Maxima Hot Start PCR Mas-

ter Mix (Fermentas, Thermo Fisher Scientific, Hudson,

NH, USA), 0.5  μM of each primer, 20  ng of DNA tem-

plate and 9.5 μL of sterile water (25 μL final volume). The

PCR protocol was as described in Sun et al. [31]. The size

and purity of amplicons were checked by electrophore-

sis on 2 % agarose gel. Short, non-specific amplification

Page 16 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

products were removed with AMP beads (AMPure XP,

Beckman Coulter Genomics, Danvers, MA, USA) using

the manufacturer’s protocol but with a modified bead:

DNA volume ratio of 0.7:1. The concentrations of puri-

fied products were measured using the Quant-iT dsDNA

BR Assay Kit (Invitrogen, Life Technologies Europe,

Stockholm, Sweden). All PCR products were pooled in

equal molar amounts and sequenced at the Swedish Insti-

tute for Infectious Disease Control in Solna (Stockholm,

Sweden), using the Roche/454 GS Titanium technol-

ogy platform. The 16S rRNA sequences were processed

as described previously [31] and deposited in the NCBI

Sequence Read Archive (SRA) under the accession num-

ber PRJNA290173.

T-RFLP

Primers targeting the glycoside hydrolase families 5

(cel5_392F 5′-GAG CAT GGG CTG GAA YHT NGG

NAA-3′ and cel5_754R 5′-CAT CAT AAT CTT TGA

AGT GGT TTG CAA TYT GDK TCC A-3′) and 48

(cel48_490F 5′ TNA TGG TTG AAG CTC CDG AYT

AYG G-3′ and cel48_920R 5′-CCA AAN CCR TAC

CAG TTR TCA ACR TC-3′) [86] were used to study

the cellulose-degrading bacterial community structures

in the different inoculum samples and at the end of the

batch test by terminal restriction fragment length poly-

morphism (T-RFLP) analysis. For the assay, the 5′end of

the cel5_754R and cel48_920R primer was labelled with

6-carboxyfluorescein (FAM). PCR amplification of tripli-

cate extractions was conducted using Maxima Hot Start

PCR Master Mix (Fermentas, Thermo Fisher Scientific,

Hudson, NH, USA). Each PCR contained 12.5 μL of cor-

responding reaction mix, 1  μL of each primer (0.5  μM

final concentration), 1  μL of DNA template (20 times

dilution) and 9.5  μL of sterile water. The PCR program

used for amplification of cel5 and cel48 included initiali-

sation at 95 °C for 5 min, denaturation at 95 °C for 1 min,

annealing for 30  s at 56  °C for cel48 (35 cycles) or at

52 °C for cel5 (45 cycles), elongation at 72 °C for 30 s, fol-

lowed by a final extension at 72 °C for 10 min. The pooled

FAM-labelled amplicons of cel5 and cel48 were puri-

fied with QIAquick gel extraction kit (Qiagen, Hilden,

Germany) and digested overnight at 37  °C with restric-

tion enzyme MboI (New England Biolabs, Wilbury Way

Hitchin, Herts, UK) and AluI (Fermentas, Thermo Fisher

Scientific, Hudson, NH, USA). Fluorescently labelled ter-

minal restriction fragments (T-RFs) were separated and

detected with ABI3730xl capillary sequencer (Applied

Biosystems, Cheshire, UK). GS ROX 500 internal size

standard (Applied Biosystems) was included in all assays.

The T-RFLP profiles were processed by Peak Scanner

software (1.0, Applied Biosystems) and the relative abun-

dance of the individual T-RFs was calculated by dividing

the peak area by the total area of all peaks. T-RFs shorter

than 70 bp (cel5) and 50 bp (cel48) constituting less than

1 % of the total peak area were excluded as background.

Clone library construction and sequencing analysis

Clone libraries were constructed for the cel5 commu-

nity with samples retrieved from CD01 and WWTP

02c/03/04/05c and for the cel48 community with samples

from CD 01/02/03 and WWTP 03/04/06, as described

previously [29]. In brief, triplicate PCRs were conducted

for each DNA extraction replicate using the primer (with-

out FAM label) and conditions described above. The

resulting nine PCR products per sample were pooled and

gel purified with QIAquick gel extraction kit (Qiagen,

Hilden, Germany) and ligated into pCR™4-TOPO® vector

(Invitrogen, Life Technologies, Grand Island, NY, USA),

followed by transformation of the ligation product into

TOP10 One Shot® chemically competent Escherichia coli (Invitrogen), according to the manufacturer’s instructions.

The sequences obtained were quality checked and edited

with the software package Geneious, version 5.6.5 (Bio-

matters Ltd., Auckland, New Zealand) and subsequently

assigned to operational taxonomic units (OTU) at the

threshold of 97 % nucleotide identity. The sequences were

compared with sequences available in the NCBI GenBank.

Alignment of cloned sequences and selected reference

sequences, as well as sequences from uncultured bacteria,

was conducted using the programme MUSCLE [107]. The

phylogenetic trees were constructed with the MEGA pro-

gramme version 5 using the maximum likelihood method

and WAG model [108]. The confidence of the trees was

tested by bootstrap resampling analysis for 1000 replicates.

All sequences were deposited in the NCBI GenBank data-

base under the accession number KT336110-29 for primer

pair cel5 and KT336130-99 for primer cel48.

Statistics

To investigate correlations between the microbial compo-

sition of the different biogas plants and the batch test per-

formance using inoculum from the corresponding plants,

canonical correspondence analysis (CCA) was performed

using the Vegan Community Ecology Package (ver-

sion 2.3-0, http://CRAN.R-project.org/package=vegan)

within R (A language and environment for statistical

computing, http://www.R-project.org/). Separate CCA

was performed using the amplicon sequencing and

T-RFLP data. Only major OTUs, i.e. OTUs represent-

ing ≥2 % of total sequences at genus level, were selected

for the assay and for T-RFLP the relative abundance of

each T-RF was used. The inverse of days needed to reach

50 and 80  % of final methane potential was used as a

measure of degradation performance. The process data

(Table 1) were included as environmental variables.

Page 17 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

Abbreviations

AD: anaerobic digestion; BMP: biochemical methane potential test; CCA:

canonical correspondence analysis; CD: co-digestion; cel5: glycoside hydrolase

genes of family 5; cel48: glycoside hydrolase genes of family 48; FAM: phos-

phoramidite fluorochrome 5-carboxy-fluorescein; GC: gas chromatograph;

HPLC: high pressure liquid chromatography; HRT: hydraulic retention time;

OLR: organic loading rate; OTU: operational taxonomic unit; PCoA: principal

coordinates analysis; PCR: polymeras chain reaction; TKN: total kjeldahl nitro-

gen; TRFLP: terminal restriction fragment length polymorphism; T-RF: terminal

restriction fragment; TS: total solids; VFA: volatile fatty acids; VS: volatile solids;

WWTP: waste water treatment plant.

Authors’ contributions

LS and TL participated equally in the planning of the study, the laboratory

work, the data interpretation and were mainly responsible for writing the

manuscript. BM contributed with analytical assistance during the microbio-

logical work, participated in data interpretation and reviewed the manuscript.

AS conceived, designed and coordinated the study, participated in supervi-

sion and data interpretation and in writing the manuscript. All authors were

involved in critical revision of the manuscript, and agree to be accountable for

all aspects of the work. All authors read and approved the final manuscript.

Author details1 Department of Microbiology, Swedish University of Agricultural Science,

Uppsala BioCenter, P.O. Box 7025, 750 07 Uppsala, Sweden. 2 Department

of Chemistry, Biotechnology and Food Science, Norwegian University of Life

Science, 1432 Ås, Norway.

Acknowledgements

The authors thank the staff at the large-scale biogas plants for assisting with

sampling and for information about operating parameters, Ulf Olsson for

assisting with the statistical analysis and Junfeng Liang for help with the BMP

test.

Availability of data and material

Sequence data have been deposited in publicly available databases and

information about accession numbers is given in the manuscript.

Competing interests

The authors declare that they have no competing interests.

Funding

This project was funded by the thematic research programme MicroDrive

(http://www.slu.se/microdrive), The Swedish Energy Agency (ERA-NET Bioen-

ergy) and the China Scholarship Council (CSC, File No.20 1,307,930,025) for the

financial support. This work was also funded in parts by the Research Council

of Norway, grant numbers 190877/S60 and 203402/E20.

Received: 28 February 2016 Accepted: 2 June 2016

Additional file

Additional file 1: Figure S1 Relative abundance of bacterial 16S

rRNA gene at class level in 10 industrial-scale biogas plants. CD 01-04:

co-digestion plants; WWTP 01-06: wastewater treatment plants: Figure

S2. Relative abundance of bacterial 16S rRNA gene at order level in 10

industrial-scale biogas plants. CD 01-04: co-digestion plants; WWTP 01-06:

wastewater treatment plants. Figure S3. Relative abundance of bacterial

16S rRNA gene at family level in 10 industrial-scale biogas plants. CD

01-04: co-digestion plants; WWTP 01-06: wastewater treatment plants.

Figure S4. Relative abundance of bacterial 16S rRNA gene at genus level

in 10 industrial-scale biogas plants. CD 01-04: co-digestion plants; WWTP

01-06: wastewater treatment plants. Figure S5. OTU heatmap based on

bacterial OTUs having relative abundance higher or equal to 0.2% in 10

industrial-scale biogas plants. CD 01-04: co-digestion plants; WWTP 01-06:

wastewater treatment plants.

References

1. Börjesson P, Mattiasson B. Biogas as a resource-efficient vehicle fuel.

Trends Biotechnol. 2008;26:7–13.

2. Möller K, Müller T. Effects of anaerobic digestion on digestate nutrient

availability and crop growth: a review. Eng Life Sci. 2012;12:242–57.

3. Lozanovski A, Lindner J, Bos U. Environmental evaluation and compari-

son of selected industrial scale biomethane production facilities across

Europe. Int J Life Cycle Assess. 2014;19:1823–32.

4. Börjesson P, Tufvesson LM. Agricultural crop-based biofuels—resource

efficiency and environmental performance including direct land use

changes. J Clean Prod. 2011;19:108–20.

5. Appels L, Lauwers J, Degrève J, Helsen L, Lievens B, Willems K, Van Impe

J, Dewil R. Anaerobic digestion in global bio-energy production: poten-

tial and research challenges. Renew Sust Energ Rev. 2011;15:4295–301.

6. Weiland P. Biogas production: current state and perspectives. Appl

Microbiol Biotechnol. 2010;85:849–60.

7. Sawatdeenarunat C, Surendra KC, Takara D, Oechsner H, Khanal SK.

Anaerobic digestion of lignocellulosic biomass: challenges and oppor-

tunities. Bioresour Technol. 2015;178:178–86.

8. Ioelovich M. Recent findings and the energetic potential of plant

biomass as a renewable source of biofuels—a review. Bioresources.

2015;10:1879–914.

9. Noike T, Endo G, Chang JE, Yaguchi JI, Matsumoto JI. Characteristics

of carbohydrate degradation and the rate-limiting step in anaerobic

digestion. Biotechnol Bioeng. 1985;27:1482–9.

10. Lynd LR, Weimer PJ, van Zyl WH, Pretorius IS. Microbial cellulose

utilization: fundamentals and biotechnology. Microbiol Mol Biol Rev.

2002;66:506–77.

11. Mata-Alvarez J, Macé S, Llabrés P. Anaerobic digestion of organic solid

wastes. An overview of research achievements and perspectives. Biore-

sour Technol. 2000;74:3–16.

12. Yang L, Xu F, Ge X, Li Y. Challenges and strategies for solid-state

anaerobic digestion of lignocellulosic biomass. Renew Sust Energ Rev.

2015;44:824–34.

13. Zheng Y, Zhao J, Xu F, Li Y. Pretreatment of lignocellulosic bio-

mass for enhanced biogas production. Prog Energy Combust Sci.

2014;42:35–53.

14. Čater M, Fanedl L, Malovrh Š, Marinšek Logar R. Biogas production from

brewery spent grain enhanced by bioaugmentation with hydrolytic

anaerobic bacteria. Bioresour Technol. 2015;186:261–9.

15. MerlinChristy P, Gopinath LR, Divya D. A review on anaerobic decompo-

sition and enhancement of biogas production through enzymes and

microorganisms. Renew Sust Energ Rev. 2014;34:167–73.

16. Zinder SH, Koch M. Non-aceticlastic methanogenesis from acetate:

acetate oxidation by a thermophilic syntrophic coculture. Arch Micro-

biol. 1984;138:263–72.

17. Angelidaki I, Karakashev D, Batstone DJ, Plugge CM, Stams AJM. Chap-

ter sixteen—biomethanation and its potential. In: Amy CR, Stephen WR,

editors. Methods enzymol. New York: Academic Press; 2011. p. 327–51.

18. Himmel ME, Xu Q, Luo Y, Ding S-Y, Lamed R, Bayer EA. Microbial enzyme sys-

tems for biomass conversion: emerging paradigms. Biofuels. 2010;1:323–41.

19. Medie FM, Davies GJ, Drancourt M, Henrissat B. Genome analyses

highlight the different biological roles of cellulases. Nat Rev Micro.

2012;10:227–34.

20. Koeck DE, Pechtl A, Zverlov VV, Schwarz WH. Genomics of cellulolytic

bacteria. Curr Opin Biotechnol. 2014;29:171–83.

21. Azman S, Khadem AF, Van Lier JB, Zeeman G, Plugge CM. Presence and role

of anaerobic hydrolytic microbes in conversion of lignocellulosic biomass

for biogas production. Crit Rev Environ Sci Technol. 2015;45:2523–64.

22. Yang JC, Chynoweth DP, Williams DS, Li A. Clostridium aldrichii sp. nov.,

a cellulolytic mesophile inhabiting a wood-fermenting anaerobic

digester. Int J Syst Bacteriol. 1990;40:268–72.

23. Palop ML, Valles S, Pinaga F, Flors A. Isolation and characterization of an

anaerobic, cellulolytic bacterium, Clostridium celerecrescens sp. nov. Int J

Syst Bacteriol. 1989;39:68–71.

24. Shiratori H, Ikeno H, Ayame S, Kataoka N, Miya A, Hosono K, Beppu T,

Ueda K. Isolation and characterization of a New Clostridium sp. That

performs effective cellulosic waste digestion in a thermophilic metha-

nogenic bioreactor. Appl Environ Microbiol. 2006;72:3702–9.

Page 18 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

25. Shiratori H, Sasaya K, Ohiwa H, Ikeno H, Ayame S, Kataoka N, Miya A,

Beppu T, Ueda K. Clostridium clariflavum sp. nov. and Clostridium caeni-

cola sp. nov., moderately thermophilic, cellulose-/cellobiose-digesting

bacteria isolated from methanogenic sludge. Int J Syst Evol Microbiol.

2009;59:1764–70.

26. Sleat R, Mah RA, Robinson R. Isolation and characterization of an

anaerobic, cellulolytic bacterium, Clostridium cellulovorans sp. nov. Appl

Environ Microbiol. 1984;48:88–93.

27. Sleat R, Mah RA. Clostridium populeti sp. nov., a cellulolytic species from

a woody-biomass digestor. Int J Syst Bacteriol. 1985;35:160–3.

28. Lebuhn M, Hanreich A, Klocke M, Schluter A, Bauer C, Perez CM.

Towards molecular biomarkers for biogas production from lignocellu-

lose-rich substrates. Anaerobe. 2014;29:10–21.

29. Sun L, Müller B, Schnürer A. Biogas production from wheat straw: com-

munity structure of cellulose-degrading bacteria. Energy Sustain Soc.

2013;3:15.

30. Krause L, Diaz NN, Edwards RA, Gartemann KH, Krömeke H, Neuweger

H, Pühler A, Runte KJ, Schlüter A, Stoye J, et al. Taxonomic composi-

tion and gene content of a methane-producing microbial community

isolated from a biogas reactor. J Biotechnol. 2008;136:91–101.

31. Sun L, Pope PB, Eijsink VGH, Schnürer A. Characterization of microbial

community structure during continuous anaerobic digestion of straw

and cow manure. Microb Biotechnol. 2015;8:815–27.

32. Lucas R, Kuchenbuch A, Fetzer I, Harms H, Kleinsteuber S. Long-term

monitoring reveals stable and remarkably similar microbial communi-

ties in parallel full-scale biogas reactors digesting energy crops. FEMS

Microbiol Ecol. 2015;91(3). doi:10.1093/femsec/fiv004.

33. Theuerl S, Kohrs F, Benndorf D, Maus I, Wibberg D, Schlüter A,

Kausmann R, Heiermann M, Rapp E, Reichl U, et al. Community shifts

in a well-operating agricultural biogas plant: how process varia-

tions are handled by the microbiome. Appl Microbiol Biotechnol.

2015;99:7791–803.

34. Guo Z, Zhou A, Yang C, Liang B, Sangeetha T, He Z, Wang L, Cai W,

Wang A, Liu W. Enhanced short chain fatty acids production from

waste activated sludge conditioning with typical agricultural residues:

carbon source composition regulates community functions. Biotechnol

Biofuels. 2015;8:192.

35. Schlüter A, Bekel T, Diaz NN, Dondrup M, Eichenlaub R, Gartemann

KH, Krahn I, Krause L, Krömeke H, Kruse O, et al. The metagenome of a

biogas-producing microbial community of a production-scale biogas

plant fermenter analysed by the 454-pyrosequencing technology. J

Biotechnol. 2008;136:77–90.

36. Hanreich A, Schimpf U, Zakrzewski M, Schlüter A, Benndorf D, Heyer R,

Rapp E, Pühler A, Reichl U, Klocke M. Metagenome and metaproteome

analyses of microbial communities in mesophilic biogas-producing

anaerobic batch fermentations indicate concerted plant carbohydrate

degradation. Syst Appl Microbiol. 2013;36:330–8.

37. Theuretzbacher F, Lizasoain J, Lefever C, Saylor MK, Enguidanos R,

Weran N, Gronauer A, Bauer A. Steam explosion pretreatment of wheat

straw to improve methane yields: investigation of the degradation

kinetics of structural compounds during anaerobic digestion. Bioresour

Technol. 2015;179:299–305.

38. Risberg K, Sun L, Levén L, Horn SJ, Schnürer A. Biogas production from

wheat straw and manure—impact of pretreatment and process oper-

ating parameters. Bioresour Technol. 2013;149:232–7.

39. Ferreira LC, Nilsen PJ, Fdz-Polanco F, Pérez-Elvira SI. Biomethane poten-

tial of wheat straw: influence of particle size, water impregnation and

thermal hydrolysis. Chem Eng J. 2014;242:254–9.

40. Werner JJ, Garcia ML, Perkins SD, Yarasheski KE, Smith SR, Muegge

BD, Stadermann FJ, DeRito CM, Floss C, Madsen EL, et al. Microbial

community dynamics and stability during an ammonia-induced

shift to syntrophic acetate oxidation. Appl Environ Microbiol.

2014;80:3375–83.

41. De Vrieze J, Saunders AM, He Y, Fang J, Nielsen PH, Verstraete W, Boon

N. Ammonia and temperature determine potential clustering in the

anaerobic digestion microbiome. Water Res. 2015;75:312–23.

42. Müller B, Sun L, Westerholm M, Schnürer A. Bacterial community com-

position and fhs profiles of low- and high-ammonia biogas digesters

reveal novel syntrophic acetate-oxidising bacteria. Biotechnol Biofuels.

2016;9:48.

43. Carballa M, Smits M, Etchebehere C, Boon N, Verstraete W. Correlations

between molecular and operational parameters in continuous lab-

scale anaerobic reactors. Appl Microbiol Biotechnol. 2011;89:303–14.

44. Wittebolle L, Marzorati M, Clement L, Balloi A, Daffonchio D, Heylen

K, De Vos P, Verstraete W, Boon N. Initial community evenness favours

functionality under selective stress. Nature. 2009;458:623–6.

45. Sundberg C, Al-Soud WA, Larsson M, Alm E, Yekta SS, Svensson BH,

Sørensen SJ, Karlsson A. 454 pyrosequencing analyses of bacterial and

archaeal richness in 21 full-scale biogas digesters. FEMS Microbiol Ecol.

2013;85:612–26.

46. Klocke M, Mähnert P, Mundt K, Souidi K, Linke B. Microbial community

analysis of a biogas-producing completely stirred tank reactor fed

continuously with fodder beet silage as mono-substrate. Syst Appl

Microbiol. 2007;30:139–51.

47. Lu X, Rao S, Shen Z, Lee PKH. Substrate induced emergence of different

active bacterial and archaeal assemblages during biomethane produc-

tion. Bioresour Technol. 2013;148:517–24.

48. Cirne DG, Lehtomäki A, Björnsson L, Blackall LL. Hydrolysis and micro-

bial community analyses in two-stage anaerobic digestion of energy

crops. J Appl Microbiol. 2007;103:516–27.

49. Wang H, Vuorela M, Keränen AL, Lehtinen TM, Lensu A, Lehtomäki

A, Rintala J. Development of microbial populations in the anaerobic

hydrolysis of grass silage for methane production. FEMS Microbiol Ecol.

2010;72:496–506.

50. Ziganshin AM, Liebetrau J, Proter J, Kleinsteuber S. Microbial commu-

nity structure and dynamics during anaerobic digestion of various agri-

cultural waste materials. Appl Microbiol Biotechnol. 2013;97:5161–74.

51. Kovács E, Wirth R, Maróti G, Bagi Z, Nagy K, Minárovits J, Rákhely

G, Kovács KL. Augmented biogas production from protein-rich

substrates and associated metagenomic changes. Bioresour Technol.

2015;178:254–61.

52. Kovács E, Wirth R, Maróti G, Bagi Z, Rákhely G, Kovács KL. Biogas produc-

tion from protein-rich biomass: fed-batch anaerobic fermentation of

casein and of pig blood and associated changes in microbial commu-

nity composition. PLoS ONE. 2013;8:e77265.

53. Bouanane-Darenfed A, Fardeau ML, Gregoire P, Joseph M, Kebbouche-

Gana S, Benayad T, Hacene H, Cayol JL, Ollivier B. Caldicoprobacter

algeriensis sp. nov. a new thermophilic anaerobic, xylanolytic bacterium

isolated from an Algerian hot spring. Curr Microbiol. 2011;62:826–32.

54. Yokoyama H, Wagner ID, Wiegel J. Caldicoprobacter oshimai gen. nov.,

sp. nov., an anaerobic, xylanolytic, extremely thermophilic bacterium

isolated from sheep faeces, and proposal of Caldicoprobacteraceae

fam. nov. Int J Syst Evol Microbiol. 2010;60:67–71.

55. Bouanane-Darenfed A, Ben Hania W, Hacene H, Cayol JL, Ollivier B,

Fardeau ML. Caldicoprobacter guelmensis sp. nov., a thermophilic,

anaerobic, xylanolytic bacterium isolated from a hot spring. Int J Syst

Evol Microbiol. 2013;63:2049–53.

56. Tang Y, Shigematsu T, Morimura S, Kida K. The effects of micro-aeration

on the phylogenetic diversity of microorganisms in a thermophilic

anaerobic municipal solid-waste digester. Water Res. 2004;38:2537–50.

57. Cheon J, Hong F, Hidaka T, Koshikawa H, Tsuno H. Microbial population

dynamics in a thermophilic methane digester fed with garbage. Water

Sci Technol. 2007;55:175–82.

58. Alauzet C, Marchandin H, Courtin P, Mory F, Lemée L, Pons JL, Chapot-

Chartier MP, Lozniewski A, Jumas-Bilak E. Multilocus analysis reveals

diversity in the genus Tissierella: description of Tissierella carlieri sp. nov. in

the new class Tissierellia classis nov. Syst Appl Microbiol. 2014;37:23–34.

59. Schiefer-Ullrich H, Andreesen J. Peptostreptococcus barnesae sp. nov.,

a Gram-positive, anaerobic, obligately purine utilizing coccus from

chicken feces. Arch Microbiol. 1985;143:26–31.

60. Hatamoto M, Kaneshige M, Nakamura A, Yamaguchi T. Bacteroides

luti sp. nov., an anaerobic, cellulolytic and xylanolytic bacterium

isolated from methanogenic sludge. Int J Syst Evol Microbiol.

2014;64:1770–4.

61. Robert C, Chassard C, Lawson PA, Bernalier-Donadille A. Bacteroides

cellulosilyticus sp. nov., a cellulolytic bacterium from the human gut

microbial community. Int J Syst Evol Microbiol. 2007;57:1516–20.

62. Rui J, Li J, Zhang S, Yan X, Wang Y, Li X. The core populations and co-

occurrence patterns of prokaryotic communities in household biogas

digesters. Biotechnol Biofuels. 2015;8:1–15.

Page 19 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

63. Chouari R, Le Paslier D, Dauga C, Daegelen P, Weissenbach J, Sghir A.

Novel major bacterial candidate division within a municipal anaerobic

sludge digester. Appl Environ Microbiol. 2005;71:2145–53.

64. Limam RD, Chouari R, Mazéas L, Wu TD, Li T, Grossin-Debattista

J, Guerquin-Kern JL, Saidi M, Landoulsi A, Sghir A, Bouchez T.

Members of the uncultured bacterial candidate division WWE1 are

implicated in anaerobic digestion of cellulose. MicrobiologyOpen.

2014;3:157–67.

65. Rinke C, Schwientek P, Sczyrba A, Ivanova NN, Anderson IJ, Cheng J-F,

Darling A, Malfatti S, Swan BK, Gies EA, et al. Insights into the phylogeny

and coding potential of microbial dark matter. Nature. 2013;499:431–7.

66. Nobu MK, Narihiro T, Rinke C, Kamagata Y, Tringe SG, Woyke T, Liu

W-T. Microbial dark matter ecogenomics reveals complex synergistic

networks in a methanogenic bioreactor. ISME J. 2015;9:1710–22.

67. Narihiro T, Sekiguchi Y. Microbial communities in anaerobic digestion

processes for waste and wastewater treatment: a microbiological

update. Curr Opin Biotechnol. 2007;18:273–8.

68. Chouari R, Le Paslier D, Daegelen P, Ginestet P, Weissenbach J, Sghir

A. Novel predominant archaeal and bacterial groups revealed by

molecular analysis of an anaerobic sludge digester. Environ Microbiol.

2005;7:1104–15.

69. Riviere D, Desvignes V, Pelletier E, Chaussonnerie S, Guermazi S, Weis-

senbach J, Li T, Camacho P, Sghir A. Towards the definition of a core

of microorganisms involved in anaerobic digestion of sludge. ISME J.

2009;3:700–14.

70. Wan S, Sun L, Sun J, Luo W. Biogas production and microbial commu-

nity change during the Co-digestion of food waste with chinese silver

grass in a single-stage anaerobic reactor. Biotechnol Bioprocess Eng.

2013;18:1022–30.

71. Zhang Y, Hu M, Li P, Wang X, Meng Q. Trichloroethylene removal and

bacterial variations in the up-flow anaerobic sludge blanket reac-

tor in response to temperature shifts. Appl Microbiol Biotechnol.

2015;99:6091–102.

72. Shu D, He Y, Yue H, Wang Q. Microbial structures and community

functions of anaerobic sludge in six full-scale wastewater treatment

plants as revealed by 454 high-throughput pyrosequencing. Bioresour

Technol. 2015;186:163–72.

73. Narihiro T, Nobu MK, Kim N-K, Kamagata Y, Liu W-T. The nexus of

syntrophy-associated microbiota in anaerobic digestion revealed by

long-term enrichment and community survey. Environ Microbiol.

2015;17:1707–20.

74. Ariesyady HD, Ito T, Okabe S. Functional bacterial and archaeal commu-

nity structures of major trophic groups in a full-scale anaerobic sludge

digester. Water Res. 2007;41:1554–68.

75. Hagen LH, Vivekanand V, Linjordet R, Pope PB, Eijsink VGH, Horn SJ.

Microbial community structure and dynamics during co-digestion of

whey permeate and cow manure in continuous stirred tank reactor

systems. Bioresour Technol. 2014;171:350–9.

76. Yamada T, Sekiguchi Y. Cultivation of uncultured Chloroflexi Subphyla:

significance and ecophysiology of formerly uncultured Chloroflexi

‘Subphylum I’ with natural and biotechnological relevance. Microbes

Environ. 2009;24:205–16.

77. Miura Y, Okabe S. Quantification of cell specific uptake activity of

microbial products by uncultured chloroflexi by microautoradiography

combined with fluorescence in situ hybridization. Environ Sci Technol.

2008;42:7380–6.

78. BonillaSalinas M, Fardeau ML, Thomas P, Cayol JL, Patel BKC, Ollivier

B. Mahella australiensis gen. nov., sp. nov., a moderately thermophilic

anaerobic bacterium isolated from an Australian oil well. Int J Syst Evol

Microbiol. 2004;54:2169–73.

79. Niu Q, Qiao W, Qiang H, Li YY. 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. Bioresour Technol. 2013;146:223–33.

80. Niu Q, Takemura Y, Kubota K, Li Y-Y. Comparing mesophilic and thermo-

philic anaerobic digestion of chicken manure: microbial community

dynamics and process resilience. Waste Manage. 2015;43:114–22.

81. Nedashkovskaya OI, Kim SB, Hoste B, Shin DS, Beleneva IA, Vancanneyt

M, Mikhailov VV. Echinicola vietnamensis sp. nov., a member of the

phylum Bacteroidetes isolated from seawater. Int J Syst Evol Microbiol.

2007;57:761–3.

82. Petitdemange E, Caillet F, Giallo J, Gaudin C. Clostridium cellulolyticum

sp. nov., a cellulolytic, mesophilic: species from decayed grass. Int J Syst

Bacteriol. 1984;34:155–9.

83. Giallo J, Gaudin C, Belaich JP, Petitdemange E, Caillet-Mangin F. Metabo-

lism of glucose and cellobiose by cellulolytic mesophilic Clostridium sp.

strain H10. Appl Environ Microbiol. 1983;45:843–9.

84. Peng X, Börner RA, Nges IA, Liu J. Impact of bioaugmentation on

biochemical methane potential for wheat straw with addition of

Clostridium cellulolyticum. Bioresour Technol. 2014;152:567–71.

85. Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, Sargent M,

Gill SR, Nelson KE, Relman DA. Diversity of the human intestinal micro-

bial flora. Science. 2005;308:1635–8.

86. Pereyra LP, Hiibel SR, Prieto Riquelme MV, Reardon KF, Pruden A. Detec-

tion and quantification of functional genes of cellulose-degrading, fer-

mentative, and sulfate-reducing bacteria and methanogenic Archaea.

Appl Environ Microbiol. 2010;76:2192–202.

87. Suzuki M, Nakagawa Y, Harayama S, Yamamoto S. Phylogenetic analysis

of genus Marinilabilia and related bacteria based on the amino acid

sequences of GyrB and emended description of Marinilabilia salmoni-

color with Marinilabilia agarovorans as its subjective synonym. Int J Syst

Evol Microbiol. 1999;49:1551–7.

88. Hethener P, Brauman A, Garcia JL. Clostridium termitidis sp. nov., a

cellulolytic Bacterium from the gut of the wood-feeding termite, Nasu-

titermes lujae. Syst Appl Microbiol. 1992;15:52–8.

89. Kato S, Haruta S, Cui ZJ, Ishii M, Yokota A, Igarashi Y. Clostridium stramini-

solvens sp. nov., a moderately thermophilic, aerotolerant and cellulolytic

bacterium isolated from a cellulose-degrading bacterial community. Int

J Syst Evol Microbiol. 2004;54:2043–7.

90. Wei Y, Zhou H, Zhang J, Zhang L, Geng A, Liu F, Zhao G, Wang S,

Zhou Z, Yan X. Insight into dominant cellulolytic bacteria from two

biogas digesters and their glycoside hydrolase genes. PLoS ONE.

2015;10:e0129921.

91. Park SK, Jang HM, Ha JH, Park JM. Sequential sludge digestion

after diverse pre-treatment conditions: sludge removal, methane

production and microbial community changes. Bioresour Technol.

2014;162:331–40.

92. Prasad P, Singh T, Bedi S. Characterization of the cellulolytic enzyme

produced by Streptomyces griseorubens (Accession No. AB184139)

isolated from Indian soil. J King Saud Univ Sci. 2013;25:245–50.

93. Feng HW, Zhi YE, Shi WW, Mao L, Zhou P. Isolation, identification and

characterization of a straw degrading Streptomyces griseorubens JSD-1.

Afr J Microbiol Res. 2013;7:2730–5.

94. Feng H, Zhi Y, Sun Y, Wei X, Luo Y, Zhou P. Draft genome sequence of

a novel Streptomyces griseorubens strain, JSD-1, active in carbon and

nitrogen recycling. Genome Announc. 2014;2:e00650.

95. Patel GB, Khan AW, Agnew BJ, Colvin JR. Isolation and characterization

of an anaerobic, cellulolytic microorganism, Acetivibrio cellulolyticus gen.

nov., sp. nov. Int J Syst Bacteriol. 1980;30:179–85.

96. Akinosho H, Yee K, Close D, Ragauskas A. The emergence of Clostridium

thermocellum as a high utility candidate for consolidated bioprocessing

applications. Front Chem. 2014;2:66. doi:10.3389/fchem.2014.00066.

97. MahalingeswaraBhat K, Wood TM. The cellulase of the anaerobic bacte-

rium Clostridium thermocellum isolation, dissociation, and reassociation

of the cellulosome. Carbohydr Res. 1992;227:293–300.

98. Lu F, Bize A, Guillot A, Monnet V, Madigou C, Chapleur O, Mazeas L, He P,

Bouchez T. Metaproteomics of cellulose methanisation under thermo-

philic conditions reveals a surprisingly high proteolytic activity. ISME J.

2014;8:88–102.

99. Lamed R, Setter E, Bayer EA. Characterization of a cellulose-binding,

cellulase-containing complex in Clostridium thermocellum. J Bacteriol.

1983;156:828–36.

100. Dumitrache A, Wolfaardt G, Allen G, Liss SN, Lynd LR. Form and func-

tion of Clostridium thermocellum biofilms. Appl Environ Microbiol.

2013;79:231–9.

101. Chassard C, Delmas E, Robert C, Lawson PA, Bernalier-Donadille A. Rumi-

nococcus champanellensis sp. nov., a cellulose-degrading bacterium

from human gut microbiota. Int J Syst Evol Microbiol. 2012;62:138–43.

102. De Vrieze J, Raport L, Willems B, Verbrugge S, Volcke E, Meers E,

Angenent LT, Boon N. Inoculum selection influences the biochemical

methane potential of agro-industrial substrates. Microb Biotechnol.

2015;8:776–86.

Page 20 of 20Sun et al. Biotechnol Biofuels (2016) 9:128

• We accept pre-submission inquiries

• Our selector tool helps you to find the most relevant journal

• We provide round the clock customer support

• Convenient online submission

• Thorough peer review

• Inclusion in PubMed and all major indexing services

• Maximum visibility for your research

Submit your manuscript atwww.biomedcentral.com/submit

Submit your next manuscript to BioMed Central and we will help you at every step:

103. Westerholm M, Hansson M, Schnürer A. Improved biogas production

from whole stillage by co-digestion with cattle manure. Bioresour

Technol. 2012;114:314–9.

104. Stromberg S, Nistor M, Liu J. Towards eliminating systematic errors

caused by the experimental conditions in Biochemical Methane Poten-

tial (BMP) tests. Waste Manag. 2014;34:1939–48.

105. APHA. Standard methods for the examination of water and wastewater,

20th edn. American Public Health Association, American Water Works

Association, Water Environment Federation; 1998.

106. Stanley D, Geier MS, Hughes RJ, Denman SE, Moore RJ. Highly variable

microbiota development in the chicken gastrointestinal tract. PLoS

ONE. 2013;8:e84290.

107. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy

and high throughput. Nucleic Acids Res. 2004;32:1792–7.

108. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5:

molecular evolutionary genetics analysis using maximum likelihood,

evolutionary distance, and maximum parsimony methods. Mol Biol

Evol. 2011;28:2731–9.

109. Hansen KH, Angelidaki I, Ahring BK. Anaerobic digestion of swine

manure: inhibition by ammonia. Water Res. 1998;32:5–12.

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier.com/locate/biortech

Importance of inoculum source and initial community structure for biogasproduction from agricultural substrates

Tong Liu1, Li Sun1, Bettina Müller, Anna Schnürer⁎

Department of Molecular Science, Swedish University of Agricultural Science, Uppsala BioCenter, P.O. Box 7025, SE-75007 Uppsala, Sweden

G R A P H I C A L A B S T R A C T

A R T I C L E I N F O

Keywords:BiogasLignocelluloseNext-generation amplicon sequencingTerminal restriction fragment lengthpolymorphism (T-RFLP)Glycoside hydrolase families 5 and 48

A B S T R A C T

This study evaluated the importance of inoculum source for start-up and operation of biogas processes. Threedifferent inocula with different community structure were used to initiate six laboratory continuous stirred tankreactor (CSTR) processes operated with a grass manure mixture as substrate. The processes were evaluated bychemical and microbiological analysis, by targeting the overall bacterial community and potential cellulose-degrading bacteria. As expected, the results showed a large difference in community structure in the inocula andin process performance during the first hydraulic retention time (HRT). However, the performance and overallmicrobial community structure became similar in the reactors over time. An inoculum from a high-ammoniaprocess, characterized by low diversity and low degradation efficiency, took the longest time to reach stabilityand final methane yield. The overall bacterial community was mainly shaped by the operating conditions but,interestingly, potential cellulose-degrading bacteria seemed mainly to originate from the substrate.

1. Introduction

Biogas is one of the most promising bioenergy alternatives for fossilfuel-based energy. Many biodegradable organic wastes can be used assubstrates to produce biogas, which eases the pressure on the en-vironment, waste treatment, and energy supply to cities (Weiland,2010). Among these substrates, lignocellulosic materials such as

agricultural residues are highly interesting due to high abundance andpotential for biogas production (Meyer et al., 2017).

Biogas is produced through anaerobic degradation, engaging var-ious microorganisms performing four major degradation steps: hydro-lysis, fermentation, acetogenesis, and methanogenesis (Schnürer,2016). The degree and rate of degradation and the biogas yield dependon the chemical and physical characteristics of the substrate, but also

http://dx.doi.org/10.1016/j.biortech.2017.08.213Received 3 July 2017; Received in revised form 29 August 2017; Accepted 31 August 2017

⁎ Corresponding author.

1 Tong Liu and Li Sun contributed equally to this work.E-mail address: anna.schnurer@slu.se (A. Schnürer).

on the chosen process parameters, such as temperature and retentiontime, that shape the composition of the different microbial groups andcommunities active in the process (Schnürer, 2016). When agriculturalresidues (stalks, straw, husks, cobs, grass, etc.) and manure are used assubstrates, the initial hydrolysis step is usually rate-limiting for thewhole degradation process, as the lignocellulose in these plant-basedmaterials is difficult for microorganisms to access and utilize (Azmanet al., 2015). Many attempts have been made to improve the degree ofdegradation, e.g., by using different pre-treatment techniques to im-prove the accessibility of the substrate for microorganisms (Carrereet al., 2016). Bioaugmentation with bacteria with high degradationefficiency for lignocellulosic material has also been evaluated, but withvarying results (Poszytek et al., 2016; Tsapekos et al., 2017).

Recent research related to the degradation of lignocellulose inbiogas processes has had a strong focus on the microorganisms in-volved, with the aim of further understanding and improving de-gradation. These studies have e.g., analyzed the whole bacterial andarchaea community by analyzing the 16 rRNA genes (Azman et al.,2015; Sun et al., 2016). However, it is difficult to address communitychanges in lignocellulose degraders exclusively based on a phylogeneticmarker, such as the 16S RNA gene. Thus, some recent studies have triedto target the cellulose-degrading community specifically by variousother molecular tools based on functional genes. For example, Pereyraet al. (2010) designed consensus degenerate hybrid oligonucleotideprimers to directly target the genes encoding glycoside hydrolase. Useof quantitative PCR (qPCR) has revealed higher abundance of thesegenes in lignocellulose-fed rather than ethanol-fed bioreactors (Pereyraet al., 2010). Sun et al. (2016) used the same primer sets to investigatethe potential cellulolytic bacterial community patterns in 10 full-scalebiogas plants through terminal restriction fragment length poly-morphism (T-RFLP) combined with clone libraries. The results showeda high correlation between lignocellulose degradation and presence ofglycoside hydrolase families 5 and 48, mainly representing bacteriabelonging to the phyla Firmicutes and Bacteroidetes. Moreover, stableisotope probing (SIP) combined with fluorescent in situ hybridization(FISH) has revealed that Firmicutes and Bacteroidetes play importantroles in cellulose degradation (Li et al., 2009).

In a previous study, we investigated degradation of straw and cel-lulose during batch cultivation using material from different full-scalebiogas plants as the inoculum source (Sun et al., 2016). The resultsshowed similar biogas yields but differences in the degradation rate, aswell as a correlation between degradation rate and the composition ofthe cellulose-degrading community. These results imply that the lig-nocellulolytic microbial community in the initial inoculum could be animportant factor for start-up of a biogas process and later on for reg-ulating the degradation rate of lignocellulose-rich material. The im-portance of inoculum source for biogas production from different sub-strates, such as food waste, sewage sludge, crops, microalgae, andmanure, has been demonstrated previously (De Vrieze et al., 2015b;Koch et al., 2017; Mahdy et al., 2017). Moreover, a recent survey in-vestigating the importance of the methanogenic population on start-upof a biogas process showed that different sources of inoculum withdifferent methanogenic composition and abundance resulted in dif-ferent biomethane potential, most likely as a result of differences inammonium tolerance in the initial community (De Vrieze et al., 2015b).However, most of the above-mentioned studies have investigated theimportance of inoculum source using a batch cultivation system andfew studies have investigated effects in a continuous digestion system.In addition, the importance of inoculum source for the degradation oflignocellulosic materials has not been specifically addressed.

Thus, the overall aim of the present study was to investigate theimportance of the inoculum source for efficient biogas production fromlignocellulose-rich material in a continuously operated process, theimpact of the initial microbial community structure on the degradationefficiency, and how these were affected by the substrates and operatingparameters used. The hypothesis tested was that choosing a suitable

inoculum gives an early advantage in biogas production and degrada-tion of lignocellulose. Inocula were taken from three different in-dustrial-scale biogas plants in Sweden showing differences in de-gradation of straw and cellulose. A manure-grass mixture was used assubstrate to operate the laboratory-scale CSTR reactors initiated withthe different inocula. The processes were monitored by analysis ofdifferent chemical parameters. The microbial community structureswere analyzed by Illumina sequencing targeting the 16sRNA genes andby T-RFLP analysis targeting glycoside hydrolase families 5 and 48. Thedegradation of the substrate and biogas production were also evaluatedusing batch cultivation, started with the different initial inocula andwith inocula retrieved from the laboratory-scale CSTR reactors at theend of the experiment.

2. Materials and methods

2.1. Biogas plants

Inocula for the laboratory-scale continuously stirred tank reactors(CSTR) was taken from three industrial-scale biogas plants in Sweden.One plant (GA) was associated with a wastewater treatment facility andoperated with sludge as the sole substrate. The second plant (GB) usedstillage from an ethanol production process as the main substrate, whilethe third (GC) was an agricultural biogas plant using mainly manureand grass silage, but occasionally other substrates. Information aboutthe plants is summarized in Table 1. Operating information on the full-scale plants (GA and GB, designated WWTP03 and CD02, respectively,in Sun et al. 2016), can also be found in our previous publications(Moestedt et al., 2016; Sun et al., 2016).

2.2. Anaerobic reactors

The laboratory-scale CSTR (8L, Dolly Belach) were initiated withinocula from the selected biogas plants. Each inoculum was used forduplicate reactors, giving in total six reactors designated GA1, GA2,GB1, GB2, GC1, and GC2. The reactors were started by filling them with5 L of the respective inoculum. From day 2, the reactors were fed with amanure-grass mixture 6 days a week, initially with a daily load of 0.6 gvolatile solids (VS) per liter. The load was then gradually increased to2.6 g VS/L/day, with an increase of around 0.5 g VS per week, reachingfull load after a total 37 of days of operation. The hydraulic retentiontime (HRT) was kept at 40 days during the whole period by adjustingthe volume by addition of tap water. The chemical composition of themanure-grass mixture used as substrate is summarized in Table 2. Thereactors were operated for 191 days, corresponding to 4.8 HRT at fullload. All reactors were operated at mesophilic (37 °C) temperature andwith a stirring speed of 90 rpm. Total gas production and CO2 contentin the gas were recorded daily and gas and liquid samples were takenevery week for analysis of methane, pH, and volatile fatty acids (VFA).Liquid samples (15 and 400 mL) were also taken and frozen at −20 °Cfor later microbial community structure analysis and for analysis oftotal nitrogen, carbon, phosphorus, potassium, magnesium, calcium,sodium, sulfur, organic nitrogen, and ammonium nitrogen (NH4

+-N).

Table 1Operating parameters of the three industrial-scale biogas plants located in Sweden.

Reactor code HRT(day)

TM (°C) Organic loading rate(OLR) (VS g/L/day)

Major substrate

GA 18 38 2.0 Mixed sludgeGB 55 38 2.9 Thin stillageGC 45 38 2.0 Agricultural

waste

T. Liu et al.

2.3. Anaerobic batch test

The methane potential of the substrate was analyzed by a bio-methane potential test (BMP test) (Westerholm et al., 2012), performedon two occasions (I, II). On the first occasion (I), samples from the full-scale biogas plants used for start-up of the CSTR (labeled GA0_0, GB0_0,and GC0_0) were used as inocula for the BMP test. The second (II) BMPtest was started with inocula taken from the CSTR after 154 days ofoperation (labeled GB1_154, GB2_154, GC1_154, and GC2_154), corre-sponding to 3 HRT at full organic load. The manure-grass mixture and acontrol group (cellulose) were tested on both occasions (I and II). Be-fore starting the tests, all inocula were kept at 37 °C for 7 days to de-crease biogas production from endogenous material. At the start of theBMP test, inoculum and substrate were mixed in a serum bottle(309 mL) under flushing with N2. The amount of inoculum and sub-strate corresponded to 12 g and 3 g VS/L, respectively, i.e., the in-oculum to substrate ratio was 4:1. On the first BMP test occasion, ratiosof 2:1 and 3:1 were also evaluated (data not shown), but 4:1 was foundto give the best rate in the test and was thus selected for use. Tap waterwas added to the bottles to reach a final liquid volume of 193 mL. Eachsubstrate was evaluated in triplicate bottles. In addition, to monitorbackground gas production from inoculum alone, three bottles were setup by adding the same amount of inoculum and water to reach the samefinal liquid volume, but without substrate. All bottles were incubated at37 °C on a rotary shaker at 130 rpm. Gas production was quantified bypressure measurements and the methane content was analyzed bysampling and analysis by gas chromatograph (GC). After each sampling,the pressure was released in the bottles. The gas/methane values werestandardized to normal atmospheric pressure (atm) and 0 °C (273.15 K,1 bar). The accumulated amount of methane was plotted over time andthe value obtained after leveling off was considered to be specific me-thane production (mL CH4/g VS).

2.4. Analytical methods

Liquid and gas samples from the reactors were preserved weekly todetermine pH value, methane (gas chromatography), and VFA content(high performance liquid chromatography) (Westerholm et al., 2012).The pH was determined directly after sampling. Selected liquid samples(400 mL) were used to measure total nitrogen and ammonium-nitrogenaccording to standard ISO methods 13878, 1998, and 11732, 2005,respectively. Total carbon was measured according to standard ISO10694, 1995 and total phosphorus, total sulfur, and total potassiumwere measured according to Swedish standard SS 28311, 1997. Totalsolid (TS) and volatile solid (VS) weight in inocula and substrate sam-ples were measured according to standard methods for examination of

water and wastewater (20th edn) published by American Public HealthAssociation in 1998.

2.5. DNA extraction

Samples were withdrawn from the inocula used to set up the CSTRon the day of starting the reactors and digestate from each anaerobicreactor was thereafter sampled every week (15 mL). All samples werestored at −20 °C. Samples from the starting time (inoculum, 0 day),digestate samples after 77, 119, and 154 days of operation (correlatingto 1, 2, and 3 HRT, respectively, of full load of substrate), and samplesof the manure-grass mixture itself were used to extract total genomicDNA. Each sample was extracted in triplicate, using 200 mg for eachextraction. The FastDNA Spin kit for soil (MP Biomedicals, Santa Ana,CA, USA) was used for the DNA extraction as described previously (Sunet al., 2016).

2.6. T-RFLP

To study the cellulose-degrading bacterial community structure, T-RFLP analysis was performed using the DNA extractions from themanure-grass mixture, the original inocula, and the digestate from day154 of operation of each continuous laboratory-scale CSTR. The genesof glycoside hydrolase families 5 and 48 were targeted by reverse sidefluorescein amidite (FAM) labeled primer pair cel5 and cel48, respec-tively. The T-RFLP analysis was performed according to the proceduredescribed previously (Sun et al., 2016). The fragment length patternsobtained were compared with in silico patterns of partial cel5 and cel48sequences obtained from clone libraries constructed and identifiedbased on amino acid level in our previous studies (Sun et al., 2016,2013). These include clone libraries constructed from two of the biogasplants (GA and GB) used as inoculum in the present paper.

2.7. Illumina sequencing and data analysis

The universal primer sets 515’F and 805R designed by Hugerthet al., (2014) were used to partially amplify the 16S rRNA genes forbuilding amplicon libraries for Illumina sequencing following methodsdescribed previously (Müller et al., 2016). The purified PCR productswere eluted with 20 μL EB buffer and quantified by using Qubit (In-vitrogen, Thermo Fisher Science, Waltham, MA, USA). The concentra-tions of the final PCR product were adjusted to 5 nM with EB buffer and2 μL of each final adjusted PCR product were pooled together. Illuminasequencing was performed at SciLifeLab in Stockholm, Sweden, usingMiSeq system. The raw DNA sequencing data obtained were submittedto National Center for Biotechnology Information database (NCBI)under accession number: from SRR5808316 to SRR5808429, thenanalysed though the open-source bioinformatics pipeline: QuantitativeInsights into Microbial Ecology (Qiime) with loaded module bioingo-tools, Qiime/1.8.0/1.9.1, SeqPrep and Cutadapt following by stepsdescribed previous (Müller et al., 2016).

3. Results and discussion

3.1. Inoculum characteristics

The inoculum used for the CSTR was taken from three full-scalebiogas plants using different operating parameters, which resulted indifferent characteristics of the inoculum (Tables 1 and 2). The inoculumused for GA originated from a wastewater treatment plant operating ata relatively short HRT (18 days) and with mixed sludge as a substrate,resulting in a relatively low level of NH4

+-N (0.7 g/L), TS (2.7%), andVFA (0.1 g/L). The biogas plants used for collection of inoculum for GBand GC were mainly fed with thin stillage and agricultural waste, re-spectively, and had a longer HRT, 55 and 45 days respectively(Table 1). The NH4

+-N and VFA level of GB were higher (4.1 and 0.8 g/

Table 2Information on the selected inocula and the manure-grass mixture substrate.

Manure-grass mixture GA0_0 GB0_0 GC0_0

TS (%) 13.6 2.7 4.1 6.3VS (%) 10.2 1.9 3.8 4.4pH 4.8 7.3 7.8 7.6VFA (g/L) 2.8 0.1 0.8 <0.1NH4

+-N (g/L) 0.7 0.7 4.1 2.5Ammonia (g/L) < 0.01 0.02 0.3 0.1Organic-N (g/L) 2.3 1.2 2.9 1.9Tot-N (g/L) 2.9 1.9 7.0 4.4Tot-C (g/L) 59.0 8.7 13.4 24.7C/N 20.1 4.5 1.9 5.6Tot-P (g/L) 0.4 0.7 0.9 0.8Tot-K (g/L) 2.7 0.1 3.1 3.3Tot-Mg (g/L) 0.4 0.1 0.1 0.6Tot-Ca (g/L) 1.1 0.8 0.9 1.1Tot-Na (g/L) 0.5 0.1 0.8 0.5Tot-S (g/L) 0.3 0.2 1.0 0.4

T. Liu et al.

L, respectively) and those of GC lower (2.5 and<0.1 g/L, respectively)than for GA. Moreover, the TS of GB and GC was 4.1 and 6.3%, re-spectively (Table 2). All reactors from which inoculum was taken op-erated at the same temperature, i.e., 37 °C. These operating conditionswere all typical for the different substrate categories. Biogas plantsusing sludge from wastewater treatment plants typically operate with acomparatively short HRT (around 15–20 days) and low levels of am-monia and VFA (Yadvika et al., 2004), while a longer HRT (around30–50 days) and comparatively higher levels of ammonia are typical inbiogas plants using nitrogen-rich substrate (such as thin stillage) andagricultural residues (such as fiber-rich material and manure) (Ahlberg-Eliasson et al., 2017; Westerholm et al., 2016).

3.2. Reactor performance and BMP- tests

In the first BMP test, based on the three different inocula used forstart-up of the CSTR processes, no significant difference (pairwise t-test,p > 0.05) in the substrate degradation rate of cellulose was seen. Thetime taken to reach 50%, 80%, and 100% of the final methane potentialwas 7–9, 13–14, and 30–35 days, respectively (Table 3). However, thefinal methane potential was different and ranged from 313 ± 8 to390 ± 5 mL CH4 g VS−1, with significantly higher values obtained inthe tests started with inoculum GA compared with GB and GC (pairwiset-test, p < 0.01) (Table 3). For the grass-manure mixture, both the rateand the final potential differed depending on the inoculum (Table 3).The initial rate was lower in the test started with inoculum GB than inthat started with GA or GC (pairwise t-test, p < 0.01). The final me-thane potential for the grass-manure mixture ranged from 441 ± 18 to548 ± 17 mL CH4 gVS−1, with the lowest and highest value for in-oculum GA and GC, respectively (student’s t-test, p < 0.01) (Table 3).This low and high degradation efficiency for cellulose in the batch testinitiated with inoculum GB and GA, respectively, was in line with re-sults from our previous study (Sun et al., 2016). The importance ofinoculum source for the degradation of cellulose was also shown in arecent study by Koch et al. (2017), illustrating differences in bothmethane production rate and yield in batch cultures initiated with in-oculum from biogas plants operating with bio-waste, sludge, ormanure/crops.

Consistent with the difference in the batch cultures, a differencebetween the inocula was also seen in the initial start-up phase of theCSTR reactors. Unsurprisingly, inoculum GC, taken from the plant usinga similar substrate as used in this study, showed the highest initial gasproduction (pairwise t-test, p < 0.01). During the first week of op-eration, the average specific methane production in GC1 and GC2reached values similar to those in the BMP test and was 458 ± 14 and405 ± 13 mL/g VS day (Fig. 1). The reactors initiated with the in-oculum GA, originating from the wastewater plant, showed sig-nificantly lower methane yields and reached values of 303 ± 14 and363 ± 17 mL CH4/g VS day, for GA1 and GA2, respectively. However,as indicated also in the BMP test, inoculum GB appeared less adapted tothe substrate used and initially reached only 61 ± 7 and 58 ± 7 mL/g VS day, respectively (Fig. 1).

On successively increasing the load from 0.6 to 2.6 g/L within40 days (i.e. 1 HRT), the performance of the reactors equalized and thespecific methane production leveled off at an average of214 ± 10 mL CH4/g VS day (Fig. 1).

During the start-up phase, a sharp increase in the concentration ofVFA, mainly represented by acetate, was seen in all reactors (Table 4).Biogas plants GA and GC reached a VFA peak at around day 50, while inGB the highest value appeared already at around day 20. This increasein VFA coincided with the increase in the percentage of CO2 and de-creasing concentration of CH4 in the gas (data not shown). Accumula-tion of VFA is a sign of process instability (Schnürer et al., 2017).However, the majority of the VFA was still acetate, suggesting that thisdisturbance was not strong. Typically, more severe disturbances arecharacterized by propionate accumulation and increasing propionate toacetate ratios (Schnürer et al., 2017). Moreover, by 62 days of opera-tion (i.e., 1.55 HRT), the VFA levels had decreased below 0.4 g/L andthe CO2 and CH4 concentrations had stabilized at levels of approxi-mately 40% and 52 ± 2%, respectively, in all reactors (Table 4). TheNH4

+-N level in GB and GC, initially around 4.1 and 2.4 g/L respec-tively, dropped during the process and after 173 days of operationreached a similar level to GA, which remained at around 1 g/L duringthe whole time of operation. The pH values were stable in the range7.4 ± 0.2 for all reactors during the whole operating process. Incombination, these results suggest that inoculum GB was less adapted tothe substrate used, likely explained by the comparably higher ammo-nium/ammonia level in this inoculum (Table 2). High NH4

+-N con-centration has a strong effect on the composition of the microbialcommunity, but has also been shown to promote a change in the me-thanogenic pathway from acetoclastic methanogenesis to syntrophicacetate oxidation (SAO), caused by inhibition of acetoclastic metha-nogens by NH3 (Westerholm et al., 2016). These changes in the me-thanogenic pathway temporarily impact on the overall methanogenesisrate. A high degree of SAO activity has previously been shown for thefull-scale plant from which inoculum GB was taken (Sun et al., 2014).Thus, changes in the methanogenic pathway might also be a factorcontributing to the comparatively slower rate of methane productionseen for inoculum GB. Although not explicitly confirmed, a negativecorrelation between NH4

+-N level and the rate of cellulose degradationhas also been suggested (Azman et al., 2015; Sun et al., 2016). It isworth noting that several studies have shown that ammonia-adaptedinoculum can also improve start-up and operation of some biogasprocesses, e.g., when exposed to ammonia stress (De Vrieze et al.,2015c; Mahdy et al., 2017).

After 3 HRT (i.e., after day 154) of full load with the grass-manuremixture, the BMP test was repeated using the digestate from the la-boratory-scale CSTR for inoculation. In this case, in line with the CSTRreactor performance, the results were more similar in the differentbatches compared with the first test run, both with regard to rate andfinal methane potential (Table 3). The rates of cellulose degradationwere similar to those in the first test run, but the final potential wasslightly lower and showed no significant difference (pairwise t-test,p > 0.05), reaching a value of 269 ± 28 mL CH4 g VS−1. For the

Table 3Final methane potential, and time taken to reach it, of the substrates cellulose and grass-manure mixture achieved by the different inocula in BMP test.

Cellulose Grass-manure mixture

Days to reach% of the final potential Final potential (mL CH4 gVS-1) Days to reach% of the final potential Final potential (mL CH4 gVS-1)

Test Inoculum 100% 80% 50% 100% 80% 50%I GA0_0 30 14 7 390 ± 5 96 22 6 441 ± 18

GB0_0 35 13 9 313 ± 8 105 32 15 511 ± 72GC0_0 35 14 7 328 ± 19 96 30 9 548 ± 18

II GA0_154 28 14 9 264 ± 36 52 15 5 361 ± 5GB0_154 28 14 9 261 ± 33 52 16 5 329 ± 27GC0_154 28 14 9 281 ± 17 52 17 7 382 ± 71

T. Liu et al.

grass-manure mixture, faster degradation was seen compared with thefirst test, irrespective of the inoculum source, with only approximately5, 16, and 53 days required to reach 50, 80, and 100% of the final

potential, respectively. However, compared with the first test run, thefinal yield was generally much lower, with mean value357 ± 45 mL CH4 g VS−1, although still with no significant differ-ences between the test started with different inocula (pairwise t-test,p > 0.05) (Table 3). The overall decrease in degradation efficiency inthe process over time, as illustrated in the batch test for both celluloseand the manure mixture, has also been shown in a previous study(Risberg et al., 2013). This change is possibly caused by a change in themicrobial community, shaped by the prevailing operating conditionsand the substrate. Still, all reactors, both batch semi-continuous,showed equal performance, suggesting that the initially different mi-crobial community had equalized, as discussed in the section below.

3.3. Bacterial communities

3.3.1. Diversity indicesIllumina sequencing of 22 triplicate total genomic DNA samples

saved from four time points, i.e., 0, 77, 119, and 154 days, resulted in atotal of 2 109 843 sequences, with 8288–88 793 sequences per sampleafter quality trim and chimera check. The triplicate samples weremerged in silico and then randomly subsampled based on the detectedlowest sequences of the sample (41 100 sequences per sample). Thenumber of observed operational taxonomic units (OTU) across samplesobtained from the rarefaction curve ranged from 365 to 1666, with thelowest value for the substrate (GS0_0) and the inoculum used for re-actors GB (GB0_0) (Table 5). In line with this result, these samples alsoshowed the lowest values of Chao1 (Table 5). According to Chao1 and

0

50

100

150

200

250

300

350

400

450

500

0 20 40 60 80 100 120 140 160 180 200

Aver

age

CH4 p

rodu

co

(mL/

g VS

day

)

Days

GA1

GA2

GB1

GB2

GC1

GC2

2. full load 3 HRT 1. full load started

Fig. 1. Average methane (CH4) production(mL/g VS day) of six continuous laboratory-scale biogas reactors started with three dif-ferent types of inoculum (A, B, C) and op-erated with a substrate of grass-manure at37 °C.

Table 4Total VFA (g/L) changes over time in the CSTR reactors (including acetate, propionate, I-butyrate, butyrate, I-valerate, and valerate). Concentration of acetate shown in brackets.

GA1 GA2 GB1 GB2 GC1 GC2

Days13 0 0 2.35 (1.90) 2.21 (1.94) 0 020 0 0 3.14 (2.52) 3.02 (2.51) 0 027 0 0 1.90 (1.44) 1.02 (0.83) 0.10 034 0 0 0.35 (0.23) 0.09 0.01 (0.01) 0.20 (0.20)42 0.50 (0.50) 0.10 (0.10) 0.34 (0.30) 0.35 (0.32) 1.27 (1.11) 1.09 (1.00)49 2.09 (1.91) 0.80 (0.80) 0.11 (0.08) 0.20 (0.20) 3.04 (2.63) 2.26 (2.04)55 0.89 (0.80) 1.03 (0.80) 0.03 (0.03) 0.30 (0.30) 3.77 (2.82) 4.50 (3.62)62 0.03 (0.03) 0.27 (0.20) 0.38 (0.21) 0.42 (0.31) 1.97 (1.63) 1.01 (0.64)104 0.06 (0.03) 0.08 (0.03) 0.04 0.04 0.07 0.26111 0 0 0.12 (0.03) 0.04 (0.04) 0.04 (0.04) 0.05 (0.05)181 0.05 (0.05) 0.04 (0.04) 0.11 (0.07) 0.05 (0.05) 0.04 (0.04) 0.03 (0.03)188 0.03 (0.03) 0.04 (0.03) 0.06 (0.04) 0.03 (0.03) 0.04 (0.04) 0.03 (0.03)

Table 5Summary of observed OTUs and values of Chao1, Shannon and Simpson indices.

Sample Chao1 Observed OTUs Shannon Simpson

GA0_0 727 562 5.790 0.934GA1_77 1396 1073 5.805 0.910GA1_119 1525 1268 6.270 0.942GA1_154 1575 1341 6.581 0.954GA2_77 1182 964 5.389 0.894GA2_119 1553 1326 6.583 0.948GA2_154 1541 1295 5.798 0.901GB0_0 639 409 4.792 0.903GB1_77 1564 1323 6.479 0.952GB1_119 1598 1367 6.636 0.961GB1_154 1569 1302 6.178 0.950GB2_77 1528 1294 6.724 0.960GB2_119 1652 1427 7.026 0.964GB2_154 1618 1439 6.473 0.953GC0_0 1157 936 6.292 0.949GC1_77 1822 1608 7.572 0.979GC1_119 1791 1609 7.709 0.978GC1_154 1812 1663 7.209 0.959GC2_77 1776 1557 7.516 0.978GC2_119 1726 1606 7.673 0.977GC2_154 1869 1666 7.119 0.963GS0_0 619 365 4.617 0.870

T. Liu et al.

observed OTU number, the sequencing coverage of the total microbialcommunity was 58.9–93.1%, with most of the samples reaching valuesabove 77% (Table 5). The lowest coverage was again obtained for thesubstrate (58.9%) and the inoculum (63.9%) for reactor GB. Based onthe value of the Shannon and Simpson indices, GB0_0 also showed thelowest diversity and evenness (Table 5). This comparatively lower di-versity observed for GB was in line with previous studies showing bothdecreased richness and evenness with increasing NH4

+-N levels (DeVrieze et al., 2015c; Müller et al., 2016; Sun et al., 2016). In addition,lower coverage in the high compared with the low NH4+-N levelsamples was also found in those studies.

During operation of the reactors, the diversity index values for theGA and GB reactors gradually increased and became similar andreached the same level as for GC, which did not change over time(Table 5). The higher diversity of GA0_0 and GC0_0 compared withGB0_0 may explain the higher methane yield in GA and GC within thefirst 3 HRT. The positive correlation between reactor function andmicrobial diversity has been shown previously in our own and otherstudies (De Vrieze et al., 2015a; Sun et al., 2016). The more diversemicrobial communities in GA and GC in the present study potentiallyallowed simultaneous activity of multiple metabolism pathways, re-sulting in more efficient degradation compared with GB (De Vriezeet al., 2015a). Still, over time the microbial diversity increased to asimilar level in all reactor samples, suggesting that, in a long-term,perspective operation management was more important than the initialmicrobial community structure for the performance of the process.

3.3.2. Phylogenetic analysisUnweighted UniFrac principal coordinate analysis (PCoA) revealed

clear separation of the microbial communities in the different inoculaand these were also separated from the community in the grass-manuremixture (Fig. 2). This is in line with previous findings from analysis ofdifferent full-scale biogas plants showing clear separation in microbialcommunity structure between plants operating with sewage sludge,manure, or co-digestion of other substrates (Sundberg et al., 2013). Thisseparation can be caused by differences in substrate composition, butalso in the chosen operating parameters, which in turn are set

depending on the substrate. Among different operating parameters,temperature and ammonium/ammonia have been shown in severalstudies to have a strong impact on development of the communitystructure (De Vrieze et al., 2015c; Müller et al., 2016; Sun et al., 2015).Consequently, the separation seen in this study was probably caused byboth the substrate characteristics and prevailing conditions in the re-actors.

The microbial community in the substrate was dominated by re-presentatives from the phyla Proteobacteria (77.8%) and Firmicutes(19.5%). In addition, the phyla Actinobacteria and Bacteroidetes werepresent at lower abundance, 1.5 and 0.6% respectively (Fig. 3). Thedifferent inocula were different from the substrate, but also from eachother. The phylum Proteobacteria that dominated in the grass-manuremixture was barely detectable in reactor samples (average relativeabundance 0.5%). This difference appeared to be the major reason forthe community separation of inocula and substrate (Fig. 2). The GAinoculum instead mainly consisted of members of the phylum Bacter-oidetes (34.8%), candidate phylum OP8 (20.2%), phylum Firmicutes(11.1%), and candidate phylum Hyd24-12 (7.7%). The inoculum fromGB and GC had a higher proportion of the phylum Firmicutes, withrelative abundance of 41.4 and 47.7%, respectively. In addition, GB hada higher proportion of the phyla Chloroflexi (32%) and Bacteroidetes(18.6%), whereas GC had a higher proportion of Bacteroidetes (41.3%)(Fig. 3). The observed community composition at phylum level, withdominance of Firmicutes and Bacteroidetes, is in line with previousfindings for biogas reactors (Schnürer, 2016). Members of these phylahave a wide metabolic capacity, most likely explaining their highabundance. The higher relative abundance of the phylum Firmicutes inGB and GC was most likely correlated with the higher ammonium/ammonia level in these two inocula, as has also been shown in otherstudies (De Vrieze et al., 2015c; Müller et al., 2016). The presence ofcandidate phyla Hyd24-12 and OP8 in the inoculum GA is also in linewith previous studies of sludge-based biogas plants (Kirkegaard et al.,2016; Sekiguchi, 2006). In contrast to previous studies reporting highabundance of the phylum Chloroflexi in biogas plants operating withsludge from WWTP (McIlroy et al., 2017; Sundberg et al., 2013), in thepresent study members of this phylum were mainly present in inoculum

Fig. 2. Phylogenetic distance between samples as determined by un-weighted UniFrac principal coordinate analysis (PCoA) (arrows illustratethe movement over time).

T. Liu et al.

GB, i.e., from the biogas plants operating with this stillage. To ourknowledge, high abundance of this phylum has not previously beendescribed at the high ammonia levels prevailing in this reactor. How-ever, despite the initial difference between the inocula, and in line withreactor performance, the microbial community structure phylogeneti-cally shifted over time and, at the end, all reactor samples were gath-ered in one cluster, still clearly separated from the community in thesubstrate (Fig. 2). The time line for this change differed between GAand GB/GC. After 77 days, i.e., 1.9 HRT, the communities in GB and GCwere already similar and did not change much more during the ex-perimental period. However, for GA, which was initially similar to GB,the change was slower and the community in these reactors becamesimilar to the others only after 154 days, i.e., after more than 3 HRT. Asdiscussed above, operating parameters of the reactor have been sug-gested as the main driving force shaping microbial community structureand the time required for stabilizing the adapted community structureis thus likely to vary. In a previous study also investigating differentinocula (De Vrieze et al., 2015a), similar equalization of the communitywere seen after 77 days (∼3.8 SRT). Contrasting results with regard tocommunity changes were obtained in a study investigating the im-portance of inoculum source for continuous operation of a biogas pro-cess using cellulose as a substrate (Han et al., 2016). Here, after8 months of operation (HRT 15 days), both the methane production andthe microbial communities were still different.

After operation of the reactors for 77 days, the communitieschanged and were mainly dominated by the Bacteroidetes(40.9–81.6%) and Firmicutes (12.3–48.3%). In addition, the phylaChloroflexi, Synergistetes, and Actinobacteria were detected at rela-tively low abundance (< 1.5%) across all reactor samples (Fig. 3).Moreover, unclassified bacterial phyla were detected in all reactorsamples at levels from 1.6 to 5.4% and at a lower level (0.03%) in thesubstrate. Proteobacteria, strongly dominating in the substrate, did notseem to have a strong impact on the final community in the continuousanaerobic biogas reactors and were only present at 0.04–1.69%. Withinthe phylum Bacteroidetes, the order Bacteroidales dominated across allreactor samples. The relative abundance of this order increased after77 days of operation, in line with the increase in the phylum

Bacteroidetes, compared with the initial level, i.e., the abundance in GAincreased from 33.9% (GA0_0) to 65.2 ± 8.8% (average value of GA1and GA2 from day 77 to 154), that in GB increased from 18.1% (GB0_0)to 60.9 ± 12.2% (average value of GB1 and GB2 from day 77 to 154),and that in GC increased slightly from 40.4% (GC0_0) to 52.3 ± 9.4%(average value of GC1 and GC2 from day 77 to 154). Within the orderBacteroidales, the community mainly consisted of the families Bacter-oidaceae, Porphyromonadaceae, and an unknown family (data notshown). This consistent change suggests that these two families arestrongly involved in grass-manure mixture degradation, a suggestionsupported by evidence that members of the family Bacteroidaceae playa key role in the hydrolysis and fermentation process in anaerobic en-vironments and are able to degrade lignocellulose (Azman et al., 2015).The family Porphyromonadaceae has also been shown to be involved inhydrolysis, saccharide fermentation, and acetogenesis in different typesof anaerobic reactors (Sakamoto, 2014). Within the phylum Firmicutes,the class Clostridia dominated in the inocula, with higher average re-lative abundance in GB0_0 and GC0_0 (40.6 and 46.1%, respectively)than in GA0_0 (9.7%). This class decreased over time in GB and GC(from 40.6 and 46.1% to 14.6% and 19.9%, respectively, by day 154)and stayed at a similar level in GA (11.7 ± 2.3%). However, the classClostridia, which contains many important cellulose-degrading bac-terial species (Azman et al., 2015; Pereyra et al., 2010), still dominatedin overall community of all reactors at the end of the experiment.Clostridium (genus level) was the most abundant identified bacteriumwithin the class Clostridia across all reactor samples Clostridium alsochanged in line with the class Clostridia, which decreased in GB andGC, but not in GA.

The phylum WWE1, also found in all original inoculum samples butat low average relative abundance (GA0_0: 7.3%, GB0_0: 0.6%, andGC0_0: 2.2%), stayed at the same level of relative abundance over thewhole time of operation in GB (0.3 ± 0.2%) and GC (0.9 ± 0.8%).However, in GA reactors the phylum WWE1 slightly decreased (day 77)and then increased to 11.7–14.7% (day 119), but decreased again atday 154 to reach levels of around 1.6–5.7%. There is no obvious ex-planation for this peak, based on the reactor performance data.However, the phylum WWE1 is commonly found in biogas plants

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100% [Thermi] [Caldithrix] WWE1 WS1 Verrucomicrobia Thermotogae Tenericutes TM7 Synergistetes Spirochaetes SR1 Proteobacteria Planctomycetes OP9 OP8 OD1 NKB19 Len sphaerae Hyd24-12 Fusobacteria Firmicutes Fibrobacteres Cyanobacteria Chloro exi Chlorobi Bacteroidetes Arma monadetes Ac nobacteria Acidobacteria k__Bacteria;Other Euryarchaeota Crenarchaeota Unclassi ed;Other

Fig. 3. Relative abundance of bacterial 16SrRNA gene at phylum level in the CSTRsamples (GA, GB and GC), arranged by time(day 0, 77, 119 and 154) and the substratesample (GS0_0).

T. Liu et al.

operating with wastewater sludge and its members have been shown todegrade amino acid and propionate (Pelletier et al., 2008).

The phyla Crenarchaeota and Euryarchaeota of the Archaea weredetected at very low relative abundance in all samples (range 0–0.45%in different samples). This low relative abundance of archaeal phylacompared with bacterial has been shown in other studies targeting the16rRNA gene in samples from different types of biogas reactor (Sunet al., 2015; Sundberg et al., 2013). Nevertheless, differences on genuslevel between substrate, inocula, and reactor samples were observed.The substrate was mainly represented by the genera Methanobrevibacter(order Methanobacteriales) and Methanosarcina (order Methanosarci-nales), and an unidentified genus belonging to the order Methanomi-crobiales. GA0_0 was instead dominated by the genus Methanosaeta(order Methanosarcinales) and two unidentified genera, belonging tothe family WSA2 and order Methanomicrobiales, respectively. InGB0_0, the phylum Euryarchaeota was only represented by the genusMethanosarcina, while in GC0_0 the genera Methanobacterium and Me-thanobrevibacter, both order Methanobacteriales, dominated. This dif-ference in methanogenic composition in the different reactors is in linewith previous findings, with the acetate-utilizing genus Methanosaetatypically dominating in WWTP plants and representatives from thegenus Methanosarcina and the orders Methanobacteria and Methano-microbiales being more commonly detected in co-digestion andmanure-based plants (Schnürer, 2016). After operation of the reactorsfor 77 days, the genera Methanobacterium and Methanosarcina becamedominant in all reactors. These two genera are both commonly detectedin biogas reactors and a positive correlation between their abundanceand good reactor performance, resulting in higher methane yield, hasbeen suggested (FitzGerald et al., 2015; Walter et al., 2016).

3.4. T-RFLP

As for the overall community, the community of potential cellulosedegraders, analyzed by T-RFLP, differed in the inocula, both for the cel5and cel48 communities, but still with more similar communities in GBand GC compared with GA (Fig. 4). According to results from ourprevious studies (Sun et al., 2016, 2013), Echinicola vietnamensis, Clos-tridium stercorarium, and Ruminococcus sp. HUN007 are the identifiedbacteria most closely related to the dominant T-RFs 81 bp (cel5,WP_015264998; 57.4% identity), 68 bp (cel48, AGI39871; 57.9%identity), and 238 bp (cel48, WP_049962845; 83.2% identity) in GAinoculum, with relative abundance of 32.9, 14.6, and 43.3%, respec-tively. Echinicola vietnamensis was originally isolated from seawater andis able to degrade starch (Nedashkovskaya et al., 2007). This T-RF haspreviously also been detected in biogas reactors operating with manureand wheat straw (Sun et al., 2013). Clostridium stercorarium is wellknown for its ability to degrade lignocellulose (Adelsberger et al.,2004). T-RFs representing C. stercorarium were also seen in low relativeabundance in our previous study, where slaughterhouse waste, grass-wheat based stillage, and mixed sludge were used as substrate in batchtests initiated with different inocula (Sun et al., 2016). Ruminococcus sp.HUN007 was first isolated from rumen fluid and later detected insamples from the wastewater treatment plant included in this study andin different co-digestion biogas plants (Liu et al., 2008; Sun et al.,2016). Marinilabilia salmonicolor, Herbinix sp. SD1D, and Clostridiumstraminisolvens are the identified bacteria most closely related to thedominant T-RFs 211 bp (cel5, WP_036163195; 66.3% identity), 328 bp(cel48, WP_058258585; 89.7% identity), and 358 bp (cel48, GAE90081;78.8% identity) in the GB and GC inocula, with relative abundance of35.1 and 27.5%, 17.1 and 62.9%, 9.7 and 13.5%, respectively (Fig. 4).Marinilabilia salmonicolor was first isolated in a marine environment,but has also been detected in biogas reactors operated with manure andhas been shown to have a positive correlation with the use of straw andcellulose as substrate (Sun et al., 2016, 2013). In other studies, M.salmonicolor has been shown to degrade monomeric sugars and starch(Nakagawa & Yamasato, 1996). Herbinix sp. SD1D is a cellulose-

degrading bacterium belonging to the family Lachnospiraceae, orderClostridiales. This bacterium was recently isolated from a thermophilicbiogas reactor and has been shown to be able to metabolize lig-nocellulose to acetate, ethanol, and propionic acid (Koeck et al., 2016).Clostridium straminisolvens has been seen in many anaerobic reactorsoperating with lignocellulose-rich material (Sun et al., 2016) and hasbeen shown to possess cellulosomes (Wei et al., 2015). In the batchassay performed with the initial inocula, GA resulted in a significantlyhigher methane yield from cellulose than GB. This higher yield waspossibly caused by the difference in the cellulose-degrading commu-nity, with GC and GB being more similar to each other than to GA. GAinitially also had a higher relative abundance of the class Clostridia,possibly also an explanation for the higher methane yield.

Operation with the grass-manure mixture shaped the potential cel-lulolytic communities in a similar trend as for the total community, i.e.,no significant difference was seen in the CSTRs after an operating timeof 145 days, despite the difference in the inocula, except that T-RF157 bp (cel48) was enriched only in the GA samples (Fig. 4b). This TRFrepresents a clone sequence with closest similarity to Ruminococcusflavefaciens (WP_051530684; 69.6% identity). T-RFs 74, 222, 228(cel5), and 321 bp (cel48) were all enriched in GA, GB, and GC reactorsfrom initial average levels of 3.9 ± 3.4%, 2.2 ± 0.5%, 0.4 ± 0.7%,and less than detected level, respectively, to 17.9 ± 11.9%,50.8 ± 9.8%, 22.3 ± 11.6%, and 5.2 ± 3.9%, respectively after154 days of operation (Fig. 4). In a previous study, these T-RFs havebeen shown to represent sequences most closely related to Clostridiumcellulovorans (WP_010075948; 60.7% identity), Prevotella buccae(WP_004346180; 55.1% identity), Bacteroides uniformis(WP_061411411; 67.5%, identity), and Clostridium thermocellum (AC-T46162; 75% identity), respectively (Sun et al., 2016). Prevotella buccaeis an anaerobic bacterium common in the rumen of various mammalianspecies (Shah et al., 2010). It can utilize various saccharides such asarabinose, cellobiose, fructose, glucose, lactose, maltose, mannose,rhamnose, salicin, and xylose (Mohammedi et al., 1998). This same T-RF (222 bp) has also been found in a reactor operating with straw andmanure and the corresponding clone in this case was identified (52.5%)as Eubacterium cellulosolvens (Sun et al., 2013). Bacteroides uniformis haspreviously been detected in various anaerobic reactors, and is able touse carbohydrates and peptones (Deublein & Steinhauser, 2011;Renouf & Hendrich, 2011). Clostridium thermocellum and C. cellulovoranshave frequently been detected in various lignocellulose-rich anaerobicenvironments and in biogas reactors investigated in our previous study,where cellulose and straw were used as substrate in batch cultures(Azman et al., 2015; Pereyra et al., 2010; Sun et al., 2016). In a recentstudy, bioaugmentation using C. thermocellum has been reported to giveenhanced biogas yield from wheat straw in batch cultures (Tsapekoset al., 2017).

Most interestingly, the potential cellulose-degrading bacteriaemerging at the end of the experiment appeared to originate from thegrass-manure mixture rather than the inoculum. Only the T-RF 328 bp(cel48) close to H. sp. SD1D, with high abundance in reactors GB and GCat the start, increased over time in these reactors (Fig. 4b). This con-tradicts the results of illumina sequencing, where the overall bacterialcommunity clearly separated from the substrate and seemed to be in-fluenced mainly by the chosen operating parameters. However, thestrong dominance of Proteobacteria in the substrate might have maskeda common community later established in the reactors. It is worthnoting, however, that the cellulose-degrading ability, as analyzed in thebatch test, appeared to decrease over time in the GA and GC reactors,suggesting that the community initially present in the inoculum wasstill more efficient with regard to cellulose degradation.

4. Conclusions

Start-up of CTRS reactors with three different inocula with differentcommunity structure and degradation capacity for cellulose and

T. Liu et al.

substrate (manure/grass) resulted in clear differences during the firstHRT of operation. A high ammonia inoculum with low diversity showedthe lowest performance. This confirms that inoculum source is im-portant for starting up biogas reactors. However, still over time, thereactors equalized in terms of both performance and structure of theoverall microbial community. Substrate and selected operating para-meters, and not the inoculum source, appeared to be the main driversfor both process performance and overall microbial community.

Acknowledgements

The authors thank the staff at the industrial-scale biogas plants forproviding inoculum samples and data on operating parameters. We alsoacknowledge Maria Erikson for help with operation of the reactors andchemical analysis.

Funding

This work was supported by the Swedish Energy Agency (ERA-NETBioenergy), the China Scholarship Council (CSC), [Grant No. 20

1,307,930,025, 2014] and the STandUp for Energy program.

Conflict of interest

None.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in theonline version, at http://dx.doi.org/10.1016/j.biortech.2017.08.213.

References

Adelsberger, H., Hertel, C., Glawischnig, E., Zverlov, V.V., Schwarz, W.H., 2004. Enzymesystem of Clostridium stercorarium for hydrolysis of arabinoxylan: reconstitution ofthe in vivo system from recombinant enzymes. Microbiology 150 (Pt 7), 2257–2266.

Ahlberg-Eliasson, K., Nadeau, E., Levén, L., Schnürer, A., 2017. Production efficiency ofSwedish farm-scale biogas plants. Biomass Bioenergy 97, 27–37.

Azman, S., Khadem, A.F., Lier, J.B., Zeeman, G., Plugge, C.M., 2015. Presence and role ofanaerobic hydrolytic microbes in conversion of lignocellulosic biomass for biogasproduction. Crit. Rev. Environ. Sci. Technol. 45.

Carrere, H., Antonopoulou, G., Affes, R., Passos, F., Battimelli, A., Lyberatos, G., Ferrer, I.,

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%cel5 T-RFLP pro e 476 412

404 400 388 384 364 360 356 292 280 276 272 244 240 232 228 224 222 216 211 208 164 156 144 132 128 124 112 100 96 92 88 85 80 74 72

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%cel48 T-RFLP pro e 358 328

324 321 316 296 292 252 248 238 236 232 224 220 216 212 208 172 168 160 157 152 140 120 116 112 100 96 92 88 84 80 76 72 68 60 56 52

a

b

Fig. 4. T-RFLP profile representing thecommunity of glycoside hydrolase gene fa-mily 5 (cel5) (a) and 48 (cel48) (b) in theCSTR samples (GA, GB, GC), arranged bytime (day 0 and 154) and the substratesample (GS0_0).

T. Liu et al.

2016. Review of feedstock pretreatment strategies for improved anaerobic digestion:from lab-scale research to full-scale application. Bioresour. Technol. 199, 386–397.

De Vrieze, J., Gildemyn, S., Vilchez-Vargas, R., Jáuregui, R., Pieper, D.H., Verstraete, W.,Boon, N., 2015a. Inoculum selection is crucial to ensure operational stability inanaerobic digestion. Appl. Microbiol. Biotechnol. 99 (1), 189–199.

De Vrieze, J., Raport, L., Willems, B., Verbrugge, S., Volcke, E., Meers, E., Angenent, L.T.,Boon, N., 2015b. Inoculum selection influences the biochemical methane potential ofagro-industrial substrates. Microb. Biotechnol. 8 (5), 776–786.

De Vrieze, J., Saunders, A.M., He, Y., Fang, J., Nielsen, P.H., Verstraete, W., Boon, N.,2015c. Ammonia and temperature determine potential clustering in the anaerobicdigestion microbiome. Water Res. 75, 312–323.

Deublein, D., Steinhauser, A., 2011. Biogas from Waste and Renewable Resources: AnIntroduction. John Wiley & Sons.

FitzGerald, J.A., Allen, E., Wall, D.M., Jackson, S.A., Murphy, J.D., Dobson, A.D., 2015.Methanosarcina play an important role in anaerobic co-digestion of the seaweed Ulvalactuca: taxonomy and predicted metabolism of functional microbial communities.PLoS One 10 (11), e0142603.

Han, S., Liu, Y., Zhang, S., Luo, G., 2016. Reactor performances and microbial commu-nities of biogas reactors: effects of inoculum sources. Appl. Microbiol. Biotechnol.100 (2), 987–995.

Hugerth, L.W., Wefer, H.A., Lundin, S., Jakobsson, H.E., Lindberg, M., Rodin, S.,Engstrand, L., Andersson, A.F., 2014. DegePrime, a program for degenerate primerdesign for broad-taxonomic-range PCR in microbial ecology studies. Appl. Environ.Microbiol. 80 (16), 5116–5123.

Kirkegaard, R.H., Dueholm, M.S., McIlroy, S.J., Nierychlo, M., Karst, S.M., Albertsen, M.,Nielsen, P.H., 2016. Genomic insights into members of the candidate phylum Hyd24-12 common in mesophilic anaerobic digesters. ISME J. 10 (10), 2352.

Koch, K., Lippert, T., Drewes, J.E., 2017. The role of inoculum’s origin on the methaneyield of different substrates in biochemical methane potential (BMP) tests. Bioresour.Technol. 243, 457–463.

Koeck, D.E., Hahnke, S., Zverlov, V.V., 2016. Herbinix luporum sp. nov., a thermophiliccellulose-degrading bacterium isolated from a thermophilic biogas reactor. Int. J.Syst. Evol. Microbiol. 66 (10), 4132–4137.

Li, T.L., Mazeas, L., Sghir, A., Leblon, G., Bouchez, T., 2009. Insights into networks offunctional microbes catalysing methanization of cellulose under mesophilic condi-tions. Environ. Microbiol. 11 (4), 889–904.

Liu, C., Finegold, S.M., Song, Y., Lawson, P.A., 2008. Reclassification of Clostridium coc-coides, Ruminococcus hansenii, Ruminococcus hydrogenotrophicus, Ruminococcus luti,Ruminococcus productus and Ruminococcus schinkii as Blautia coccoides gen. nov.,comb. nov., Blautia hansenii comb. nov., Blautia hydrogenotrophica comb. nov., Blautialuti comb. nov., Blautia producta comb. nov., Blautia schinkii comb. nov. and de-scription of Blautia wexlerae sp. nov., isolated from human faeces. Int. J. Syst. Evol.Microbiol. 58 (Pt 8), 1896–1902.

Mahdy, A., Fotidis, I.A., Mancini, E., Ballesteros, M., González-Fernández, C., Angelidaki,I., 2017. Ammonia tolerant inocula provide a good base for anaerobic digestion ofmicroalgae in third generation biogas process. Bioresour. Technol. 225, 272–278.

McIlroy, S.J., Kirkegaard, R.H., Dueholm, M.S., Fernando, E., Karst, S.M., Albertsen, M.,Nielsen, P.H., 2017. Culture-independent analyses reveal novel anaerolineaceae asabundant primary fermenters in anaerobic digesters treating waste activated sludge.Front. Microbiol. 8 (1134).

Meyer, A.K.P., Ehimen, E.A., Holm-Nielsen, J.B., 2017. Future European biogas: Animalmanure, straw and grass potentials for a sustainable European biogas production.Biomass Bioenergy.

Moestedt, J., Muller, B., Westerholm, M., Schnurer, A., 2016. Ammonia threshold forinhibition of anaerobic digestion of thin stillage and the importance of organicloading rate. Microb. Biotechnol. 9 (2), 180–194.

Mohammedi, I., Streichenberger, T., Beck, F., Chevalier, J.M., Motin, J., 1998. Prevotellabuccae bacteraemia associated with infection of a pseudoaneurysm. Eur. J. Vasc.Endovasc. Surg. 15 (2), 175–176.

Müller, B., Sun, L., Westerholm, M., Schnürer, A., 2016. Bacterial community compositionand fhs profiles of low-and high-ammonia biogas digesters reveal novel syntrophicacetate-oxidising bacteria. Biotechnol. Biofuels 9 (1), 1.

Nakagawa, Y., Yamasato, K., 1996. Emendation of the genus Cytophaga and transfer ofCytophaga agarovorans and Cytophaga salmonicolor to Marinilabilia gen. nov.: phylo-genetic analysis of the Flavobacterium-Cytophaga complex. Int. J. Syst. Evol.Microbiol. 46 (2), 599–603.

Nedashkovskaya, O.I., Kim, S.B., Hoste, B., Shin, D.S., Beleneva, I.A., Vancanneyt, M.,Mikhailov, V.V., 2007. Echinicola vietnamensis sp. nov., a member of the phylumBacteroidetes isolated from seawater. Int. J. Syst. Evol. Microbiol. 57 (Pt 4), 761–763.

Pelletier, E., Kreimeyer, A., Bocs, S., Rouy, Z., Gyapay, G., Chouari, R., Riviere, D.,Ganesan, A., Daegelen, P., Sghir, A., Cohen, G.N., Medigue, C., Weissenbach, J., LePaslier, D., 2008. “Candidatus Cloacamonas acidaminovorans”: genome sequence re-construction provides a first glimpse of a new bacterial division. J. Bacteriol. 190 (7),2572–2579.

Pereyra, L.P., Hiibel, S.R., Prieto Riquelme, M.V., Reardon, K.F., Pruden, A., 2010.Detection and quantification of functional genes of cellulose- degrading, fermenta-tive, and sulfate-reducing bacteria and methanogenic archaea. Appl. Environ.Microbiol. 76 (7), 2192–2202.

Poszytek, K., Ciezkowska, M., Sklodowska, A., Drewniak, L., 2016. Microbial consortiumwith high cellulolytic activity (MCHCA) for enhanced biogas production. Front.Microbiol. 7.

Renouf, M., Hendrich, S., 2011. Bacteroides uniformis is a putative bacterial species as-sociated with the degradation of the isoflavone genistein in human feces. J. Nutr. 141(6), 1120–1126.

Risberg, K., Sun, L., Leven, L., Horn, S.J., Schnurer, A., 2013. Biogas production fromwheat straw and manure – impact of pretreatment and process operating parameters.Bioresour. Technol. 149, 232–237.

Sakamoto, M., 2014. The Family Porphyromonadaceae. Springer, Berlin Heidelberg.Berlin, Heidelberg, pp. 811–824.

Schnürer, A., 2016. Biogas Production: Microbiology and Technology. In: Hatti-Kaul, R.,Mamo, G., Mattiasson, B. (Eds.), Anaerobes in Biotechnology. Springer InternationalPublishing Cham, pp. 195–234.

Schnürer, A., Bohn, I., Moestedt, J., 2017. Protocol for Start-Up and Operation of CSTRBiogas Processes. Springer, Berlin Heidelberg. Berlin, Heidelberg, pp. 171–200.

Sekiguchi, Y., 2006. Yet-to-be cultured microorganisms relevant to methane fermentationprocesses. Microb. Environ. 21 (1), 1–15.

Shah, H.N., Gharbia, S.E., Olsen, I. 2010. Bacteroides, Prevotella, and Porphyromonas. In:Topley &Wilson’s Microbiology and Microbial Infections. John Wiley & Sons, Ltd.

Sun, L., Liu, T., Müller, B., Schnürer, A., 2016. The microbial community structure inindustrial biogas plants influences the degradation rate of straw and cellulose inbatch tests. Biotechno. Biofuels 9 (1), 1–20.

Sun, L., Müller, B., Schnürer, A., 2013. Biogas production from wheat straw: communitystructure of cellulose-degrading bacteria. Energy Sustain. Soc. 3 (1), 1–11.

Sun, L., Müller, B., Westerholm, M., Schnürer, A., 2014. Syntrophic acetate oxidation inindustrial CSTR biogas digesters. J. Biotechnol. 171, 39–44.

Sun, L., Pope, P.B., Eijsink, V.G.H., Schnürer, A., 2015. Characterization of microbialcommunity structure during continuous anaerobic digestion of straw and cowmanure. Microb. Biotechnol. 8 (5), 815–827.

Sundberg, C., Al-Soud, W.A., Larsson, M., Alm, E., Yekta, S.S., Svensson, B.H., Sørensen,S.J., Karlsson, A., 2013. 454 pyrosequencing analyses of bacterial and archaealrichness in 21 full-scale biogas digesters. FEMS Microbiol. Ecol. 85 (3), 612–626.

Tsapekos, P., Kougias, P.G., Vasileiou, S.A., Treu, L., Campanaro, S., Lyberatos, G.,Angelidaki, I., 2017. Bioaugmentation with hydrolytic microbes to improve theanaerobic biodegradability of lignocellulosic agricultural residues. Bioresour.Technol. 234, 350–359.

Walter, A., Silberberger, S., Juárez, M.F.-D., Insam, H., Franke-Whittle, I.H., 2016.Biomethane potential of industrial paper wastes and investigation of the methano-genic communities involved. Biotechnol. Biofuels 9 (1), 1.

Wei, Y., Zhou, H., Zhang, J., Zhang, L., Geng, A., Liu, F., Zhao, G., Wang, S., Zhou, Z., Yan,X., 2015. Insight into dominant cellulolytic bacteria from two biogas digesters andtheir glycoside hydrolase genes. PLoS One 10 (6), e0129921.

Weiland, P., 2010. Biogas production: current state and perspectives. Appl. Microbiol.Biotechnol. 85 (4), 849–860.

Westerholm, M., Crauwels, S., Houtmeyers, S., Meerbergen, K., Van Geel, M., Lievens, B.,Appels, L., 2016. Microbial community dynamics linked to enhanced substrateavailability and biogas production of electrokinetically pre-treated waste activatedsludge. Bioresour. Technol. 218, 761–770.

Westerholm, M., Hansson, M., Schnurer, A., 2012. Improved biogas production fromwhole stillage by co-digestion with cattle manure. Bioresour. Technol. 114, 314–319.

Yadvika, Santosh, Sreekrishnan, T.R., Kohli, S., Rana, V., 2004. Enhancement of biogasproduction from solid substrates using different techniques––a review. Bioresour.Technol. 95 (1), 1–10.

T. Liu et al.

microorganisms

Article

Substrate-Induced Response in Biogas ProcessPerformance and Microbial Community Relates Backto Inoculum Source

Tong Liu 1, Li Sun 1, Åke Nordberg 2 ID and Anna Schnürer 1,* ID

1 Department of Molecular Science, Swedish University of Agricultural Sciences, Uppsala 75007, Sweden;tong.liu@slu.se (T.L.); li.sun@slu.se (L.S.)

2 Department of Energy and Technology, Swedish University of Agricultural Sciences,Uppsala 75007, Sweden; ake.nordberg@slu.se

* Correspondence: anna.schnurer@slu.se; Tel.: +46-018-673288

Received: 25 May 2018; Accepted: 2 August 2018; Published: 5 August 2018���������������

Abstract: This study investigated whether biogas reactor performance, including microbialcommunity development, in response to a change in substrate composition is influenced by initialinoculum source. For the study, reactors previously operated with the same grass–manure mixturefor more than 120 days and started with two different inocula were used. These reactors initiallyshowed great differences depending on inoculum source, but eventually showed similar performanceand overall microbial community structure. At the start of the present experiment, the substrate wascomplemented with milled feed wheat, added all at once or divided into two portions. The startinghypothesis was that process performance depends on initial inoculum source and microbial diversity,and thus that reactor performance is influenced by the feeding regime. In response to the substratechange, all reactors showed increases and decreases in volumetric and specific methane production,respectively. However, specific methane yield and development of the microbial community showeddifferences related to the initial inoculum source, confirming the hypothesis. However, the differentfeeding regimes had only minor effects on process performance and overall community structure,but still induced differences in the cellulose-degrading community and in cellulose degradation.

Keywords: anaerobic digestion; co-digestion; continuous stirred-tank reactor (CSTR); bio-methanepotential (BMP)-test; next-generation amplicon sequencing; terminal restriction fragment lengthpolymorphism (T-RFLP); qPCR; glycoside hydrolase families 5 and 48

1. Introduction

Biogas, produced via anaerobic digestion, represents a valuable renewable energy resource thatcan replace part of the fossil fuel-based energy used today, resulting in climate and economic benefits [1].Many types of organic materials can be used for biogas production, but agricultural residues (manureand crop residues, such as stalks, straw, husks, cobs, grass, etc.) are of particular interest due to highabundance and thus high gas potential [2]. However, the high content of lignocellulose and nutrientimbalances often limit the degradation efficiency of agricultural residues [3]. An additional limitationwith manure is high water content, making it difficult to achieve high organic loads and volumetricgas production [4]. Some of the obstacles with these types of materials can be overcome by variouspre-treatment methods, making the material more accessible to microbial and enzymatic attack [5],or by co-digestion with materials that provide complementary nutrients [6]. For manure-based biogasplants, co-digestion also offers possibilities to increase the organic load. By combining manure with ahigh-water content with a more energy-dense material, such as crop/crop residues, the organic loadcan be increased without significantly decreasing the hydraulic retention time (HRT) [4,6,7].

Microorganisms 2018, 6, 80; doi:10.3390/microorganisms6030080 www.mdpi.com/journal/microorganisms

Microorganisms 2018, 6, 80 2 of 19

A prerequisite to achieving efficient biogas production is an active microbial community inbalance [8]. Parameters shown to impact the community include operating parameters, such astemperature, organic loading rate (OLR), substrate composition, and feeding regime [9–11].Many studies have looked for correlations between microbial composition and reactor function, but mosthave not found consistent relationships [12,13]. Some studies suggest positive correlations betweendiversity and function [14,15], but a correlation between low diversity and high function has alsobeen reported [16]. Still, positive correlations are often seen in connection with a specific type ofsubstrate/environment. For example, for the function of processes operating with protein-rich materialsand consequently high ammonia concentrations, the importance of syntrophic acetate-oxidizingbacteria has been highlighted [17]. For the degradation of lipids, positive correlations with the levelof Syntrophomonas have been shown [18]. For cellulose degradation, positive correlations with thelevel of Clostridium cellulolyticum have been observed [12]. Feeding regime has also been shown toinfluence function, diversity, and community structure [9,19–23], with some studies showing positivecorrelations between more frequent feeding regime and higher microbial diversity [19,21,23], however,not necessarily resulting in better reactor function [9,22–24].

The aim of the present study was to increase the understanding of the relationshipbetween community structure and performance and efficiency of a biogas process operating withlignocellulose-rich substrates. More specifically, the aim was to investigate the importance of originalinoculum source, used during start-up, when adding a cosubstrate to biogas processes operated withthe same substrate and showing similarity in both performance and overall microbial composition.The hypothesis is that the reactors will respond differently to the change in substrate depending on themicrobial community present in the initial inoculum.

To test the hypothesis, reactors operated in a previous study were used in the present experimentalwork [15]. These reactors were initially started with different inocula characterized by differences incommunity structure and diversity, and fed a mixture of cow manure and silage grass. The reactorsinitially showed significant differences in degradation efficiency and methane yield but over time,after operation for more than 3 HRT, the processes became similar regarding both performance andoverall community structure and diversity, as analyzed by targeting 16S rRNA gene [15]. However,specific analysis of the potential cellulose-degrading community, targeting the genes encoding cel 5 and48 glycosidases, revealed that the reactors still differed in this regard at the end of the experiment [15].In the present study, these reactors were complemented with milled feed wheat (MFW) as an additionalcosubstrate. It was selected as a cosubstrate because its high total solids (TS) concentration allowedthe organic load to be increased without significantly altering the HRT. The MFW was either added allat once, together with the grass-manure mixture, or divided into two portions, in order to evaluatealso the effect of feeding regime. The reactors were operated for more than 3 HRT and their overallperformance regarding methane yield stability and changes in microbial community structure wereinvestigated. Both the total microbial community and potential cellulose degrading bacteria wereanalyzed. Degradation of the substrates and of pure cellulose was also investigated in batch culturesstarted with inoculum from the reactors at the beginning and end of the experimental period.

2. Materials and Methods

2.1. Laboratory-Scale Semi-Continuous Anaerobic Reactors

Two laboratory-scale continuous stirred-tank reactor (CSTR) processes in duplicate reactorswere initially started with inoculum from two different full-scale biogas processes (codes GB, GC)in Sweden [15]. Operating information on the full-scale plants can be found in our previouspublications [12,15]. Based on the inoculum origin, the reactors were named GB1 and GB2, andGC1 and GC2. In the 120 days before the start of the present experiment (day 0), the reactors were fedwith the same substrate, a grass-manure mixture (Table 1) [15], for six days a week (once a day), withan average daily load of 2.6 g volatile solids (VS)/L and 40-day HRT. After 42 days of operation in

Microorganisms 2018, 6, 80 3 of 19

the present study, MFW (Table 1) was added to all four reactors with average daily load graduallyincreasing from 0.6 to 1.7 g VS/L (6 days a week, from day 42 to 70) (Figure 1), resulting in a totalaverage daily load of 4.3 g VS/L and a HRT of 37 days. From day 70, the reactors were fed with thefull load of MFW (1.7 g VS/L day) in different feeding regimes: reactors GB1 and GC1 were fed allMFW at the same time as the grass-manure mixture, while GB2 and GC2 were fed the MFW in twoportions, with half the amount fed 2 h after adding the grass-manure mixture and the remaining halfafter another 2 h. In total, the reactors were operated for 231 days, corresponding to 4.5 HRT at a fullload of MFW. All reactors were operated at mesophilic (37 ◦C) temperature, a stirring speed of 90 rpm,and a HRT of 37 days. Samples of liquid (15 mL) were taken at day 0, 77, 106, 147, and 231 and frozenat −20 ◦C for later analysis of the microbial community structure.

Table 1. Composition of the grass-manure mix and milled feed wheat (MFW) substrates. Values (%)based on wet weight.

VS Crude Protein Starch Crude Fat Crude Fiber Ash

Manure-grassmixture 10.2 2.1 0.4 4.9 26.2 9.0

MFW 84.0 16.0 28.0 5.0 6.5 3.5

2.2. Anaerobic Batch Test

The methane potential of the substrate was analyzed by a bio-methane potential (BMP) test [15],performed on two time points. On the first time point (test I), digestate samples from the duplicatereactors, before MFW addition, were pooled and used as inoculum (GB0_0 and GC0_0). On thesecond time point (test II), inocula were taken from all the laboratory-scale reactors after 231 days ofoperation, corresponding to an operating period of 4.5 HRT at full MFW load, and used in separatetests (GB1_231, GB2_231, GC1_231, and GC2_231). The grass-manure mixture, MFW, and cellulose(control group) were evaluated on both time points, i.e., with all inocula. Before starting the tests,all inocula were kept at 37 ◦C for seven days to decrease biogas production from the endogenousmaterial. At the start of the BMP test, inoculum and substrate were mixed in a serum bottle (309 mL)under flushing with nitrogen gas (N2). The amount of inoculum and substrate was 12 g and 3 g VS/L,respectively, i.e., the inoculum to substrate ratio was 4:1 [25,26]. Tap water was added to the bottles toreach a final liquid volume of 193 mL. Each substrate was evaluated in triplicate bottles. Additionally,to monitor background gas production from inoculum alone, three bottles were initiated by addingthe same amount of inoculum and water to reach the same final liquid volume, but with no substrate.All bottles were incubated on a rotary shaker at 37 ◦C and 130 rpm. Gas production was quantifiedby pressure measurements, and the methane content was analyzed by sampling (2 mL) followedby analysis by gas chromatography (GC) [27]. After each sampling, the pressure in the bottles wasreleased. The biogas and methane values were standardized to normal atmospheric pressure (273.15 K,1 bar). The accumulated amount of methane was plotted over time, and the value obtained afterleveling off was considered the specific methane production (ml CH4/g VS). The efficiency of methaneproduction in each reactor was further evaluated by transferring the time (days) spent to reach 50%,80%, and 100% of the final methane potential.

2.3. Residual Methane Emissions Measurement

The residual methane potential of the digestate was measured twice, on the same time points asthe BMP tests, by incubating 50 mL digestate from the four laboratory-scale reactors, i.e., digestatefrom day 0 and day 231, at 37 ◦C for approximately 52 days, when the methane production leveled off.Sampling and analysis of methane production during the incubation were performed according to themethod described above.

Microorganisms 2018, 6, 80 4 of 19

2.4. Analytical Methods

2 mL gas samples were withdrawn weekly from each reactor to determine the methane contentby gas chromatography according to a previously described method [27]. At the same time, 15 mLliquid samples (digestate) were taken to determine pH [28] and volatile fatty acid (VFA) content byhigh-performance liquid chromatography (HPLC) [27]. The pH was determined directly after sampling.Liquid samples (400 mL) were also taken on two-time points and sent to a commercial analyticallaboratory (Agrilab, Uppsala, Sweden) for determination of the concentration of total nitrogen andammonium-nitrogen according to standard ISO methods 13878:1998 and 11732:2005, respectively.Total carbon was measured according to standard ISO 10694 and total phosphorus, total sulfur, andtotal potassium according to Swedish standard SS 28311. Composition of the MFW was determined byNord Mills Co. and that of the grass-manure mixture by the laboratory at the Department of AnimalNutrition and Management (Swedish University of Agricultural Sciences, Uppsala, Sweden). Starchcontent was determined by an enzymatic method according to Åman and Hesselman (1984). Crudeprotein content was analyzed according to the Nordic Committee on Food Analysis (1976) methodfor nitrogen determination in food and feed (Kjeldahl, No 6, 3rd Edn), using a 2520 Digestor, Kjeltec8400 Analyzer unit, and 8460 sample unit (FOSS Analytical A/S Hilleröd, Denmark). Crude fat wasdetermined according to the Official Journal of the European Communities method for determinationof crude oils and fat (Commission Directive 98/64/EC, 1998), using a Hydrotec 8000 and Soxtec 8000extraction unit (FOSS Analytical A/S Hilleröd, Denmark). Weight of total solids and volatile solidsin the inocula and substrate samples was measured according to international standard methodspublished by the American Public Health Association (1998).

2.5. DNA Extraction and Microbial Community Analysis

For DNA extraction, liquid samples (15 mL) from the semi-continuous processes were taken at thetime for starting the present study (inocula, 0 day) and after 77, 106, 147, and 231 days (around 1, 2, 3,and 5 HRT, respectively) of operation with the MFW and the grass–manure mixture. Aliquots of 200 mgin triplicate were used to extract total genomic DNA, as described previously [15]. The degenerateprimer sets 515’F and 805R were used to amplify the 16S rRNA genes of both archaea and bacteriato build amplicon libraries for next-generation amplicon sequencing [29]. The PCR products werepurified by using AMPure XP (Beckman Coulter, Inc. Brea, CA, USA) and eluted with 20 μL EBbuffer, and quantified by using Qubit (Invitrogen, Thermo Fisher Science, Waltham, MA, USA).The concentrations of the final PCR product were adjusted to 5 nM with EB buffer and 2 μL of each finaladjusted PCR product were pooled together. The next-generation amplicon sequencing was performedat SciLifeLab in Uppsala, Sweden, using MiSeq system. The raw DNA sequencing data obtained weresubmitted to the National Center for Biotechnology Information database (NCBI) under accessionnumber: from SRR5808389 to SRR5808384, and analyzed through the open-source bioinformaticspipeline: Quantitative Insights into Microbial Ecology (QIIME) with loaded module bioinfo-tools,QIIME/1.8.0/1.9.1, SeqPrep and Cutadapt [17]. Specifically, the adaptor and primer sequences weretrimmed using the following criteria: (1) Trim base from the 3′ end which had a quality below 10.(2) Remove read if it contained N base, was longer than 300 bp, or did not contain primer sequences.The trimmed paired end reads were further processed in QIIME/1.8.0/1.9.1 [30]. Join_paired_ends.pywas used for joining paired end reads with minimum overlap 150 bp bases, using the SeqPrep method(https://github.com/jstjohn/SeqPrep). The joined reads were used for splitting into libraries withno barcode errors allowed, and only one consecutive low-quality base call was allowed per read.Any read that with Phred quality scores below 20 were removed. Then, the operational taxonomicunits (OTUs) were assigned by using the open reference OTU pick strategy [31]. The criteria for OTUclustering was set to a threshold of 97% similarity and performed with Uclust against Greengenes coreset (gg_13_8) [32]. The most abundant sequence in each OTU was selected as a representative sequenceand further aligned against the Greengenes core set using PyNAST software [33]. The chimericsequences were discarded using ChimeraSlayer [34]. Taxonomy was assigned to each OTU using the

Microorganisms 2018, 6, 80 5 of 19

Ribosomal Database Project (RDP) classifier with a minimum confidence threshold of 80% [35]. Then,OTUs that at least be observed in three samples and contained at least 0.0025% of total reads wereretained and used to build the final OTU table. Alpha diversity (Chao1, Shannon and Simpson indexand richness) and Beta diversity (unweighted UniFrac distance matrix) analysis were performed usingQIIME/1.8.0/1.9.1 [30].

According to the result from sequencing, three methanogenic groups: Methanobacteriales,Methanomicrobiales, and Methanosarcinaceae were quantified by quantitative polymerase chainreaction (qPCR) using the primer sets MBT, MMB, and Msc, respectively [27]. The qPCR protocol andanalysis were performed as described previously [27]. The potential cellulose-degrading bacterialcommunity in the substrate and in the digestate after 154 days of operation was analyzed by terminalrestriction fragment length polymorphism (T-RFLP) targeting the genes of glycoside hydrolasefamilies 5 and 48, according to the procedure described previously [12]. The length patterns ofthe fragments obtained were compared with the sequences of clone libraries established in our earlierpublications [12,36].

3. Results and Discussion

3.1. BMP Test

In the first BMP test (test I), the final methane potential of all substrates tested reached a meanvalue of 357 ± 45 mL CH4/g VS and showed no significant difference between the different substratesor the different inocula (pairwise t-test, p > 0.05) (Table 2). These values were at the same level asobserved before for similar substrates, e.g., cow manure, grass, and cellulose [4,12]. The values in thesecond test were in the same range, but some differences could be seen. The average BMP value forcellulose and MFW reached 291 ± 46 and 300 ± 38 mL CH4/g VS, respectively, with a slightly highervalue for cellulose in GB1_231 compared with GB2_231 (Table 2 and Figure S1). For the grass–manuremixture the final methane potential in the second BMP test (test II) using inoculum from the GBreactors was significantly lower as compared with test I; 252 ± 29 (GB1_231) and 289 ± 11 (GB2_231)compared to 329 ± 27 mL CH4/g VS (GB0_0) (pairwise t-test < 0.01). For GC, however, the values forMFW remained more similar to those in test I; 327 ± 46 (GC1_231) and 375 ± 73 (GC2_231) comparedto 382 ± 71 mL CH4/g VS (GC0_0) (Table 2). Moreover, the average BMP value obtained for thegrass–manure mixture in GC1_231 and GC2_231 (353 ± 62 mL CH4/g VS) was significantly higherthan in GB1_231 and GB2_231 (270 ± 28 mL CH4/g VS) (student t-test, p < 0.02), suggesting theimportance of initial inoculum source.

The time to reach the final methane potential was in the first test (I) between 28–52 days, with thelongest time for the grass manure mixture. This suggests that the manure grass substrate will not befully degraded in the semi-continue process having 30 days retention time. Moreover, significantlylonger times were needed to reach the final potential, as well as 50% and 80% of this potential, for allthe substrates, in the second (II) compared with the first (I) test (student t-test, p < 0.01) (Table 2). Still,higher degradation efficiency for cellulose was seen in GB1_231 compared with GB2_231, which mightindicate a small effect of the different feeding approaches for this process (Table 2 and Figure S1).

Mic

roor

gani

sms

20

18,6

,80

6of

19

Tab

le2.

Fina

lmet

hane

pote

ntia

l(m

LC

H4/

gV

S)an

dtim

eta

ken

tore

ach

100%

,80%

,and

50%

ofth

ispo

tent

ialw

ithce

llulo

se,g

rass

–man

ure

mix

ture

,and

mill

edfe

edw

heat

(MFW

)sub

stra

tes

for

the

two

diff

eren

tino

cula

(GB,

GC

)in

the

biom

etha

nepo

tent

ial(

BMP)

test

.The

met

hane

valu

esw

ere

stan

dard

ized

tono

rmal

atm

osph

eric

pres

sure

(273

.15

K,1

bar)

.

Cell

ulo

seG

rass

-man

ure

mix

ture

MF

W

Day

sto

Reach

%o

fF

inal

Po

ten

tial

Fin

al

Po

ten

tial

Day

sto

Reach

%o

fF

inal

Po

ten

tial

Fin

al

Po

ten

tial

Day

sto

Reach

%o

fF

inal

Po

ten

tial

Fin

al

Po

ten

tial

Test

Inoc

ulum

100%

80%

50%

100%

80%

50%

100%

80%

50%

IG

B0_0

2814

926

1±

3352

165

329±

2728

144

282±

24G

C0_

028

149

281±

1752

177

382±

7128

135

307±

53

II

GB1

_231

119

5621

350±

4347

179

252±

2965

2715

278±

13G

B2_2

3111

961

3527

0±

7211

955

828

9±

1111

954

1932

2±

51G

C1_

231

119

6135

308±

411

945

1732

7±

4611

954

2533

3±

43G

C2_

231

119

6138

274±

3011

937

1537

9±

7311

939

2530

7±

14

Microorganisms 2018, 6, 80 7 of 19

3.2. CSTR Processes

This experiment was started with four reactors previously operated in another study [15],where they were initially started with two different inocula and were shown to have very differentperformance in the initial phase of operation, but similar performance by the end. In the presentstudy, the reactors showed similar initial performance after complementing the grass–manure mixturewith MFW. Irrespective of the feeding regime, co-digestion with MFW increased the total methaneproduction compared with the grass–manure mixture alone in all four semi-continuous processes.In the initial phase, the level increased gradually from 3818 ± 158 to 5317 ± 304 mL CH4/day (averagevalue for day 0–42 and day 56–112, respectively) and then increased rapidly and reached a peak of6669 ± 439 mL CH4/day on day 140. Thereafter, total methane production decreased gradually andstabilized at 5362 ± 205 mL CH4/day (day 182–231), i.e., after 3 HRT of operation with a full load ofMFW (Figure 1). This final value represented an increase of around 29% over the initial level beforeaddition of the MFW, as also observed in our previous study [15].

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0

1000

2000

3000

4000

5000

6000

7000

8000

0 50 100 150 200 250

Tota

l CH 4

pro

duct

ion

(mL/

day)

Time (d)

GB1

GB2

GC1

GC2

OLR g VS/L/Day

MFW addition started (42 d)

MFW full load Two feed regimes applied (70 d)

3HRT with full load of MFW (182 d)

Digestate taken for BMP-test I

Digestate taken for BMP-test II

Foaming (112 d)

Figure 1. Primary Y-axis: total average methane (CH4) production (mL/day) in four continuouslaboratory-scale biogas reactors originally started with two different types of inoculum (GB, GC) andcodigested with substrates of grass–manure and milled feed wheat (MFW) in two feeding regimes, fullload (GB1, GC1) and split load (GB2, GC2) at 37 ◦C. The methane values were standardized to normalatmospheric pressure (273.15 K, 1 bar). Secondary Y-axis: the total OLR (g VS of the substrates per Lreactor volume per day).

As expected, increasing the load by addition of MFW resulted in a significant increase involumetric gas production, thus giving more efficient use of available digester volume. Severalprevious studies have shown a similar positive effect of codigesting energy-dense materials withmanure [4,6,7,37]. However, an increase in OLR also resulted in foaming (day 112) and increased VFAlevels, suggesting some instability in the processes (Table S1). Still, this foaming was only temporaryand lasted for approximately two weeks, after which the problem stopped. Shortly after (on day 140)a peak was observed in the CH4 production (Figure 1, Figure S2; 312 ± 16 mL CH4/g VS). This peakmight be explained by nutrients accumulating and being converted to CH4 by microorganisms whenthe foam disappeared.

Even though an increase in total methane yield was obtained, both specific methane production(SMP, defined as the normalized volume of CH4 produced per g VS of the substrate) and degree ofdegradation (VS reduction) decreased in response to MFW addition (Figure S2). The SMP level in all

Microorganisms 2018, 6, 80 8 of 19

digesters was on average 296 ± 16 mL CH4/g VS before addition of MFW (day 0–42) and 249 ± 18 mLCH4/g VS after addition (day 182–231). This decrease was not expected as the BMP values for thegrass–manure mixture and the MFW were similar and thus the SMP should theoretically not decrease(Table 2). Still, the results were also supported by the BMP tests, illustrating decreased degradationefficiency in test II compared to test I (Table 2). Moreover, the corresponding VS reduction before andafter MFW addition was 74.1 ± 3.4% and 63.7 ± 3.5%, respectively. Thus, combined these resultssuggest that the increase in load by addition of MFW resulted in less efficient degradation than whenthe grass–manure mixture was used as the sole substrate.

Differences in feeding regime did not result in any statistically significant differences in volumetricor specific methane yield. However, there were some minor differences related to process performance,e.g., VFA accumulation. Accumulation of VFA started around day 112 in all reactors, when the OLRreached 4.3 g VS/L day (Table S1). The total level fluctuated somewhat but was highest at day 224(1.6–4.6 g/L). Accumulation of VFA is linked to less efficient biogas production and typically occurswhen there is an imbalance between different microbial degradation steps. A high propionate to acetateratio can be taken as an early indicator of a risk of process failure [38]. In this study, the propionate toacetate ratio showed some differences depending on the feeding regime, with GB1 and GC1 showingslightly higher values than GB2 and GC2 after day 203 (Table S1). The VFA accumulation was alsoassociated with the foaming event, but in that case, no differences related to the feeding regime wereobserved. Foaming can be triggered by many parameters such as the production of surface-activesubstances, abrupt degassing, viscosity, alkalinity, insufficient mixing, and accumulation of VFAs [39].In our previous study, the reactors had been operated with the grass–manure mixture for a long timewithout any VFA accumulation or foaming [15]. Thus, the instability in the present study was clearlycaused by the introduction of MFW as a substrate. In comparison with manure, the MFW had higherlevels of protein and starch (Table 1). When protein is degraded ammonium-nitrogen is released.In this study, the ammonium-nitrogen concentration increased from 1.05 ± 0.5 to 2.6 ± 0.12 g/L as aresult of MFW addition (average of all reactors, from day 0 to 224). Ammonium is in equilibrium withammonia, a well-known inhibitor of biogas processes (specifically by inhibiting methanogens) [40],and this could possibly have caused the VFA accumulation followed by foaming. However, takinginto account the pH (7.6–7.7) and temperature (37 ◦C), the level of free ammonia was calculated andfound to be at most only around 0.16 g/L. This level was still low and below levels previously shownto cause inhibition [41]. Thus, a more likely explanation for the foaming was the introduction of starchby the MFV addition, which is typically converted rapidly to VFAs [39]. Previous studies investigatingthe effect of feeding regime on reactor performance have reported somewhat contradictory resultsand no consistent influence on key process parameters such as gas yield, degree of degradation,and VFA levels [9,19,21–24]. For example, no clear effect of the feeding regimes (feed every 2 dayscompared with daily) was seen on VFA, ammonia level and methane yield [19], while higher levels ofVFA have been reported when feeding every 2 days compared with every 2 h [9] and every 6 h [21].This inconsistency in results regarding the effect of feeding regime can be explained by differences intype of substrate, OLR, and feeding frequencies, with 2–48 h between feedings. Still, Mulat et al. (2016)and Ziels et al. (2018) obtained slightly higher methane yield (14 and 20%, respectively) with a lessfrequent feeding regime [9,22].

In conclusion, only small differences were observed between reactors with differences in feedingregimes in the present study. However, there were differences between the GB and GC reactors, withsignificantly higher specific methane production for GC reactors in the period after day 182 (studentt-test p < 0.01) (Figure S2). This suggests that reactor performance was influenced by the originalinoculum used for the start-up of the reactors in our previous study, where GB reactors producedsignificantly less methane than GC reactors in the start-up phase (within 1 HRT) [15]. The poorperformance of GB reactors in our previous study was attributed to a higher ammonium-nitrogenlevel in the inoculum used for the start-up of these reactors [15]. In the present study theammonium-nitrogen level increased, but to the same level in all reactors (GB: from 1.1 to 2.5 g/L,

Microorganisms 2018, 6, 80 9 of 19

GC: from 1.0 to 2.5 g/L). Thus, a more likely explanation for the differing results obtained for GBand GC reactors is differences in the microbial community rather than the ammonia level per se,as discussed below.

3.3. Residual Methane Potential

For a production plant, the volumetric yield is highly important and thus continuously logged,while the specific yield (SMP) less often is considered. Thus, a decrease in degradation efficiency causedby a new cosubstrate, as shown in this study, can be somewhat hidden. Unfortunately, decreaseddegradation efficiency might increase the risk of methane emissions during storage of the digestate,as shown in several other studies [4,37,42]. To evaluate this risk, the residual methane production(RMP) was measured by incubation of digestate taken from all reactors before and after MFW addition(day 0 and day 231, respectively). The evaluation showed similar values for all reactors before MFWaddition, i.e., 71 ± 5 mL CH4/g VS, on average for reactors GB0_0 and GC0_0 (pairwise t-test, p > 0.5)(Figure S3). However, after operation with MFW, the RMP was significantly higher showing on average134 ± 12 mL CH4/g VS (pairwise t-test, p < 0.01), but with no significant differences between thereactors (Figure S3). Thus, MFW addition clearly increased the risk of methane emissions duringstorage, which was consistent with the decrease in the degree of degradation seen in the reactors andin the BMP tests.

3.4. Microbial Communities

3.4.1. Diversity Indices

After quality trim and chimera check, 3,311,869 sequences (from 15,874 to 116,439 per sample)were retained. The triplicate samples were merged in silico and then subsampled based on the detectedlowest sequences of the sample (41,100 sequences per sample). The number of observed OTUs acrosssamples obtained from the rarefaction curve varied from 958 to 1666, with the lowest values for the GCreactors at the end of the experiment. At the start of the experiment, there was no significant differencein Chao1, Shannon, and Simpson indices of the observed OTUs between all four reactors (Table 3).However, the indices varied over time (Table 3) and the values appeared to fluctuate consistentlywith methane production and VFA levels in all four semi-continuous reactors (Table 3). Additionof MFW appeared to cause an overall decrease in diversity compared with operation with only thegrass–manure mixture and this decrease was independent of the feeding regime.

Table 3. Summary of observed OTUs, Chao1, Shannon, and Simpson index values.

Sample Chao1 Observed OTUs Shannon Simpson

GB1_0 1571 1302 6.178 0.950GB2_0 1619 1439 6.473 0.953

GB1_77 1410 1159 5.291 0.898GB2_77 1485 1260 6.054 0.938GB1_106 1673 1364 5.749 0.921GB2_106 1773 1435 6.254 0.946GB1_147 1477 1232 6.181 0.943GB2_147 1544 1291 5.982 0.935GB1_231 1453 1104 5.410 0.887GB2_231 1449 1085 5.365 0.893GC1_0 1813 1664 7.210 0.959GC2_0 1872 1666 7.119 0.963GC1_77 1592 1299 5.583 0.891GC2_77 1650 1386 5.979 0.910

GC1_106 1718 1529 7.073 0.970GC2_106 1744 1509 6.748 0.955GC1_147 1775 1448 6.681 0.962GC2_147 1675 1439 7.202 0.972GC1_231 1414 989 5.431 0.872GC2_231 1444 958 5.054 0.831

Microorganisms 2018, 6, 80 10 of 19

Several previous studies have shown a correlation between high methane production and highdiversity of microbial community, especially when the process was operated with complex substratesor in fluctuating process conditions, suggesting that a more diverse microbial community allowsactivation of multiple metabolic pathways and consequently high methane production [14,15,43].However, under constant conditions, a specialized community can be expected to be more efficient [21].In the present study, the GC reactors showed significantly higher methane production than GB reactorsat the end of the experiment, displayed the greatest decrease in OTUs richness and in the Simpson andShannon indices.

Previous studies have also shown that microbial community diversity can be affected by differentfeeding regimes. Digesters fed with lower frequency (every 2 days compared with daily or every2 h) have been shown to form a more diverse microbial community [9,19]. In line with this, GB1 andGC1, receiving the MFW all at once, showed a slightly higher average number of observed OTUs andShannon index, respectively, than GB2 and GC2, but this difference was not statistically significant(Table 3).

3.4.2. Phylogenetic Analysis

The microbial community composition, analyzed by an unweighted UniFrac principal coordinateanalysis (PCoA), was similar at the beginning of the experiment. However, the community changedover time and at the end (day 231) a separation was seen between GC and GB reactors, suggesting theimportance of the original inoculum (Figure 2). For the different feeding regimes, however, no clearseparation between GB1, GB2 and GC1, GC2 was observed (Figure 2).

Figure 2. Phylogenetic distance between samples as determined by unweighted UniFrac principalcoordinate analysis (PCoA). Sample legend arranged by time (day 0, 77, 106, 147, and 231).

Irrespective of MFW addition, the phyla Bacteroidetes (67.4 ± 12.7%) and Firmicutes (24.4 ± 9.7%)dominated in all processes and at all-time points, followed by the phylum Actinobacteria (2.2 ± 1.8%)

Microorganisms 2018, 6, 80 11 of 19

(Figure S4). The phyla Tenericutes, Verrucomicrobia, Synergistetes, WWE1, and Proteobacteria werealso detected in all reactors, but in relatively low abundance (<1%) (Figure S4). This dominance of thephyla Bacteroidetes and Firmicutes has been seen in various anaerobic digesters in many previousstudies [15,44,45]. Members of these two phyla can utilize a broad range of organic compounds andare involved in the hydrolysis, fermentation, and acetogenesis steps of anaerobic digestion [2,46].

While no significant differences were seen on phylum level, MFW addition, independent offeeding regime, resulted in a similar shift in the overall microbial community pattern in all reactors atlower taxonomic level. The most pronounced change was an increase in the relative abundance of thegenus Paludibacter (family Paludibacteraceae, order Bacteroidales, phylum Bacteroidetes) from <0.1%(day 0) to an average of 49.9 ± 7.5% (day 231) (Figure 3). This increase was seen in all reactors but wasmore pronounced in GC1 and GC2 (on average 12.7% higher than in GB1 and GB2). This differencewas most likely the cause of the separation in the PCoA analysis at the last time point (i.e., day 231)(Figure 2). The genus Paludibacter was also found in the grass–manure mixture and in the originalinocula for GB and GC, as described in our previous study [15], but here only at very low relativeabundance (<0.1%). The genus Paludibacter is strictly anaerobic and can utilize various sugars such asarabinose, xylose, cellobiose, fructose, galactose, glucose, mannose, maltose, melibiose, glycogen, andsoluble starch while producing acetate and propionate as major fermentation end-products [47]. Thus,in this study, the high level of starch in MFW probably enhanced the growth of this genus. Membersof this genus have been found at various relative abundances in other anaerobic digesters and alsoin other anaerobic environments, such as cow manure, wetlands, sludge from alkali-hydrolyzed ricestraw, and plant residues in irrigated rice-field soil [47–52]. The genus Paludibacter was recently alsosuggested as a potential cellulose degrader [51,53].

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

GS0_

00

GB1_

0GB

2_0

GB1_

77GB

2_77

GB1_

106

GB2_

106

GB1_

147

GB2_

147

GB1_

231

GB2_

231

GC1_

0GC

2_0

GC1_

77GC

2_77

GC1_

106

GC2_

106

GC1_

147

GC2_

147

GC1_

231

GC2_

231

f__[Cloacamonaceae];Othero__[Cloacamonales];Othero__LD1-PB3;f__;g__f__Acholeplasmataceae;g__Acholeplasmaf__Dethiosulfovibrionaceae;g__HA73f__Pseudomonadaceae;g__Pseudomonasf__Pseudomonadaceae;Otherf__Moraxellaceae;g__Acinetobacterf__Moraxellaceae;Othero__Pseudomonadales;Otherf__Comamonadaceae;Otherf__Bradyrhizobiaceae;Othero__MBA08;f__;g__f__[Tissierellaceae];g__Sedimentibacterf__[Tissierellaceae];g__f__Syntrophomonadaceae;g__Syntrophomonasf__Lachnospiraceae;g__Coprococcusf__Clostridiaceae;g__SMB53f__Clostridiaceae;g__Clostridiumf__Caldicoprobacteraceae;g__Caldicoprobactero__Clostridiales;f__;g__o__Clostridiales;Otherc__Clostridia;Otherf__Turicibacteraceae;g__Turicibacterf__Streptococcaceae;g__Streptococcusf__Lactobacillaceae;g__Lactobacillusf__Lactobacillaceae;Otherf__Aerococcaceae;g__Facklamiaf__Aerococcaceae;Othero__Lactobacillales;Otherp__Firmicutes;Otherf__Porphyromonadaceae;g__Paludibacterf__Porphyromonadaceae;g__f__Porphyromonadaceae;Otherf__Marinilabiaceae;g__Ruminofilibacterf__Bacteroidaceae;clone "BF311"f__Bacteroidaceae;g__f__Bacteroidaceae;Othero__Bacteroidales;f__;g__o__Bacteroidales;Otherp__Bacteroidetes;Otherf__Coriobacteriaceae;g__f__Actinomycetaceae;g__Actinomycesk__Bacteria;Other

Figure 3. Relative abundance of bacterial 16S rRNA gene at genus level in the CSTR reactors (GB1,GB2, GC1, and GC2), arranged by time (day 0, 77, 106, 147, and 231 of operation) and the substratesample (GS0_0). Relative abundance <1% were filtered out. The genus names were represented as thefirst letter of the closest classified taxonomical level plus the taxonomic name.

Microorganisms 2018, 6, 80 12 of 19

With the increase in Paludibacter, the average relative abundance of an uncultured rumenbacterium clone BF311 (belonging to unclassified order Bacteroidales) gradually decreased in allreactors after the addition of MFW, from 20.0 ± 10.4% (day 77) to 0.9 ± 0.2% (day 231), but withno significant difference between reactors (Figure 3). This uncultured rumen bacterium clone BF311(GenBank: EU850525.1) is one partial sequence of 16S ribosomal RNA genes from a series of clonesmade by Satitmanwiwat et al. (2008, Thailand) (https://www.ncbi.nlm.nih.gov/nuccore/197940871).However, it was mistakenly assigned as genus BF311 in the Greengene database and thus wrongly citedby other studies [54–56]. Still, BF311 has been reported in cattle rumen and horse feces samples [54,55].However, to our knowledge no previous publication other than ours has found BF311 in biogasdigesters. In our previous study, the relative abundance increased differently in GB and GC reactors,from 0.5% to 15.1% and from 2.5% to 5.2%, respectively, when operated with the same grass–manuremixture as used in the present study for over 3 HRT (i.e., 154 days) [15]. BF311 has been suggestedto play an important role in lignocellulose degradation in rumen environments [55,56]. In this study,BF311 was possibly outcompeted by representatives from the genus Paludibacter.

Class Clostridia (phylum Firmicutes) also slightly decreased in response to MFW addition in allreactors, from average levels of 17.2 ± 4.2% to 10.6 ± 1.9%. However, the levels increased again aroundday 146 (to on average 26.0 ± 8.0%), i.e., in the period of VFA accumulation and slight reactor instability.During reactor recovery, the levels again decreased, but to different levels in the different reactors,ranging from 18.5% in GB reactors to 9.6% in GC. These changes in the class Clostridia were mainlycaused by two unclassified families and the genus Caldicoprobacter (family Caldicoprobacteraceae)(Figure 3). Members of this genus can utilize various sugars, but also xylan and pyruvate, and produceacetate, lactate, and hydrogen as end-products [57,58]. The genus Caldicoprobacter has also beenfound to be enriched in anaerobic digesters fed lignocellulosic biomass under both mesophilic andthermophilic conditions [59,60]. Moreover, it has been shown to dominate in an anaerobic digesterwith high total ammonium-nitrogen (5 to 25 g/L) and, as in this study, high VFA levels (>4 g/L) [61].

Moreover, a slight increase in the genus Clostridium (family Clostridiaceae, phylum Firmicutes)from 1.6 ± 0.2% (day 0) to 6.9 ± 2.6% (day 231) (Figure 3), irrespective of the total changes in the levelof Class Clostridia, was observed after MFW addition in all reactors. This genus contains organismsactive both during fermentation and anaerobic oxidation that can utilize proteins and carbohydrates,and their corresponding monomers, while producing different fatty acids as end-products of theirmetabolism [62–64]. This increase is probably directly related to MFW addition and the observedincreased in VFA level at the same time point [65]. A slight increase in relative abundance of thephylum Actinobacteria (mostly contributed by the family Coriobacteriaceae), from 1.2 ± 0.6% to3.1 ± 2.1%, was also seen at the time of VFA accumulation and foaming (day 146). This phylumcontains many acid-producing bacteria and has previously been found to increase in the deteriorativephase of an anaerobic process [44]. The family Coriobacteriaceae has been shown to dominate inan anaerobic digester operating with wastewater sludge and is suggested to convert lignocellulosehydrolysates into lactic acid and acetic acid [66,67].

Among the Archaea, the phyla Euryarchaeota and Crenarchaeota dominated (Figure S4).However, as seen in several other studies of biogas digesters [15,68,69], the total relative abundance ofArchaea was very low, in this study less than 0.3% across all samples (Figure S4). Thus based on thesequencing results it was difficult to reveal detailed information about the methanogenic community.Still, order Methanobacteriales, Methanomicrobiales, and Methanosarcinales, with the dominance oforder Methanobacteriales and Methanosarcinales, was detected in all four reactors before (day 0) andafter (day 231) the MFW addition (Figure S5). These three methanogenic groups are commonly foundin various anaerobic digesters [8,27], and members within these orders can all utilize hydrogen [8].Members of Methanosarcinales can in addition utilize methanol and acetate for methane formation [8].qPCR analysis furthermore revealed that all four reactors had a similar average gene abundance ofthese three methanogenic groups at the beginning (day 0) but at the end of the experiment (day 231) asignificantly lower abundance, compared to the starting point, was seen for all groups (student t-test,

Microorganisms 2018, 6, 80 13 of 19

p < 0.01) (Figure S6). However, a difference was observed in the reactors as a higher abundance of allthree methanogenic groups was seen in GC compared to GB reactors (Figure S6), which could alsocontribute the separation of GB and GC at the day 231 in the PCoA plot (Figure 2). Possibly the higherabundance of methanogens in reactor GC could also explain the slightly higher methane production inthese reactors. However, no clear difference could be observed for the different feeding regimes.

The different feeding regime showed no clear effect on the overall microbial community in thisstudy. Similar results were obtained in a previous study during operation of CSTRs fed with glucose(once and twice a day and every 2 days) [23]. In contrast, a slight increase in microbial communityrichness was observed in a study using a starch-rich synthetic substrate fed every two days comparedwith daily feedings [19]. This is consistent with findings in the present study of higher richness inGB1 and GC1 compared with GB2 and GB2. Similarly, previous studies evaluating different feedingregimes have found effects of certain microbial groups. For example, during co-digestion of manureand oleate, the community fraction of the genus Syntrophomonas was higher when the oleate was fedevery 2 days compared with every 6 h [22]. However, the interactions between feeding regime, digesterperformance (including methane production and process parameters), and microbial communityremain somewhat unclear, as various feeding regimes have been shown to cause changes in microbialdynamics without affecting digester performance and vice versa [9,19,21,70].

3.4.3. T-RFLP

In our previous study, the T-RFLP profiles for glycoside hydrolase families 5 (cel5) and 48 (cel48)genes differed between the original inocula used to start the GB and GC reactors [15]. After 3 HRT ofoperation with the grass–manure mixture, the various T-RFLP profiles became more similar andthe community in both GB and GC reactors and the cel5 and cel 48 profile were dominated byT-RFs 74, 222, 228bp, and T-RF 328bp, respectively, according to clone libraries represented byClostridium cellulovorans (WP_010075948, 60.7% identity), Prevotella buccae (WP_004346180, 55.1%identity), Bacteroides uniformis (WP_061411411, 67.5%, identity), and Herbinix sp. SD1D (WP_058258585,89.7% identity), respectively [15].

In the present study, the addition of MFW as a cosubstrate changed both the cel5 and cel48communities, not significantly in composition but somewhat more in relative abundance (Figure 4).For the cel5 community, T-RF 222bp became slightly more abundant across all reactor samples bythe end of the experiment (from 48.8 ± 11.6% to 56.8 ± 21.9%), while T-RF 228bp decreased from15.9 ± 7.2% to 1.2 ± 1.4%. T-RF 74bp showed no significant trend in response to MFW addition(Figure 4a). For the cel 48 community, T-RF 328bp increased in all reactor samples, from 67.5 ± 4.6%to 80.1 ± 13.7% (Figure 4b). The four bacteria representing these dominant T-RFs have been foundin various anaerobic environments and show potential lignocellulolytic capacity [71–74]. Prevotellabuccae, Bacteroides uniformis, and Herbinix sp. SD1D have also been shown able to utilize starch [75–77],most likely explaining the enrichment induced by MFW addition in the present study.

A different pattern in the T-RFLP profile was also seen in response to the feeding regime, withT-RF 216bp (4.6 ± 2.1%, cel5, not identified) and T-RF 321bp (16.7 ± 6.6%, cel48) mainly detected insamples where all MFW and the grass–manure mixture were fed simultaneously (Figure 4). T-RF 321bphas previously been shown to correspond to a clone most closely related to Clostridium thermocellum(ACT46162), with 75% identity [12]. This bacterium is reported to be a highly potent cellulose degraderand to be enriched in anaerobic digesters fed lignocellulose-rich materials [12,15,78]. Moreover,a species of this bacterium is reported to be capable of producing an extracellular amylase when grownon starch [79]. The higher abundance of this bacterium possibly explains the higher degradationefficiency of cellulose seen in GB1 and GC1 compared with GB2 and GC2 in BMP test II.

Microorganisms 2018, 6, 80 14 of 19

(a)

(b)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

cel5 T-RFLP profile 412 404

400 332

244 240

236 232

228 224

222 216

212 208

164 156

144 132

128 112

108 104

100 96

92 84

80 74

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%cel48 T-RFLP profile 358 352

332 328

324 321

316 252

247 244

238 236

168 164

160 157

152 140

120 116

112 100

96 92

88 84

80 76

72 68

64 52

Figure 4. T-RFLP profile representing the community of glycoside hydrolase gene family 5 (cel5) (a)and 48 (cel48) (b) in the reactor samples (GB1, GB2, GC1, and GC2), arranged by time (day 0, 147,and 231) and the substrate sample (GS0_0).

4. Conclusions

Addition of MFW to four semi-continuous processes that had been operated with a grass-manuremixture for ~200 days, and showed similar performance and microbial community structure, resulted

Microorganisms 2018, 6, 80 15 of 19

in a significant increase in volumetric methane production and a concomitant decrease in specificmethane production and substrate degradation efficiency. The magnitude of the decrease variedbetween the processes and appeared to relate to the initial inoculum used for startup. This mayhave been caused by differences in the microbial community prevailing in the initial inoculum,suggesting that the original inoculum can profoundly influence biogas production performance inthe long term and affect microbial responses to process operation changes, which confirmed ourhypothesis. Applying different feeding regimes for MFW addition had no clear influence on methaneproduction or overall microbial community structure, but had an impact on the development of thecellulose-degrading community. Adding the MFW load all at once rather than in two portions at 2-hintervals gave slightly higher cellulose conversion activity (as indicated by BMP tests), possibly causedby a higher abundance of Clostridium thermocellum.

Supplementary Materials: The following are available online at http://www.mdpi.com/2076-2607/6/3/80/s1,Figure S1: Accumulated methane (CH4) production (mL/g VS) using digestate taken from the end of the CSTRtest (GB1_231 and GB2_231), operating with cellulose as substrate, Figure S2: Specific average methane (CH4)production (mL/g VS day) of four continuous laboratory-scale biogas reactors originally started with two differenttypes of inoculum (GB and GC) and codigested with substrates of grass–manure and milled feed wheat (MFW) intwo feeding approaches at 37 ◦C, Figure S3: Residual methane (CH4) production (mL/g VS) of digestate takenfrom all reactors before (GB0_0 and GC0_0) and after (GB1_231, GB2_231, GC1_231, and GC2_231) addition ofmilled feed wheat (MFW), Figure S4: Relative abundance of bacterial 16S rRNA gene at phylum level in the CSTRsamples (GB1, GB2, GC1, and GC2), arranged by time (day 0, 77, 106, 147, and 231) and the substrate sample(GS0_0), Figure S5: Relative abundance of Archaea 16S rRNA gene based on next-generation amplicon sequencingat order level in the CSTR samples (GB1, GB2, GC1, and GC2), arranged by time (day 0 and 231), Figure S6:Average log gene abundance per mL sample obtained in qPCR analysis of the main methanogenic populations inthe CSTR samples (GB1, GB2, GC1, and GC2), arranged by time (day 0 and 231), Table S1: Total changes over timein volatile fatty acid concentration (VFA, g/L) (including acetate, propionate, I-butyrate, butyrate, I-valerate, andvalerate) in the different reactors. Pro/Ac = propionate to acetate ratio.

Author Contributions: Conceptualization, T.L., L.S., Å.N., and A.S.; Data curation, T.L., L.S., and Å.N.; Formalanalysis, T.L.; Investigation, T.L., L.S., and A.S.; Methodology, T.L., L.S., Å.N., and A.S.; Project administration,A.S.; Supervision, Å.N. and A.S.; Visualization, T.L.; Writing original draft, T.L.; Writing review & editing, T.L.,L.S., Å.N., and A.S.

Funding: This work was supported by the Swedish Energy Agency (ERA-NET Bioenergy), the China ScholarshipCouncil (CSC) [Grant No. 201307930025, 2014], and the STandUp for Energy program.

Acknowledgments: The authors thank Simon Isaksson for help with operation of the reactors andchemical analysis.

Conflicts of Interest: The authors declare no conflicts of interest. The funding sponsors had no role in the designof the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in thedecision to publish the results.

References

1. Meyer-Aurich, A.; Lochmann, Y.; Klauss, H.; Prochnow, A. Comparative advantage of maize-and grass-silagebased feedstock for biogas production with respect to greenhouse gas mitigation. Sustainability 2016, 8, 617.[CrossRef]

2. Azman, S.; Khadem, A.F.; Lier, J.B.; Zeeman, G.; Plugge, C.M. Presence and role of anaerobic hydrolyticmicrobes in conversion of lignocellulosic biomass for biogas production. Crit. Rev. Environ. Sci. Technol.2015, 45, 2523–2564. [CrossRef]

3. Lynd, L.R.; Weimer, P.J.; Van Zyl, W.H.; Pretorius, I.S. Microbial cellulose utilization: Fundamentals andbiotechnology. Microbiol. Mol. Biol. Rev. 2002, 66, 506–577. [CrossRef]

4. Ahlberg-Eliasson, K.; Liu, T.; Nadeau, E.; Schnürer, A. Forage types and origin of manure in co-digestionaffect methane yield and microbial community structure. Grass Forage Sci. 2018. [CrossRef]

5. Monlau, F.; Barakat, A.; Trably, E.; Dumas, C.; Steyer, J.-P.; Carrère, H. Lignocellulosic materials intobiohydrogen and biomethane: Impact of structural features and pretreatment. Crit. Rev. Environ. Sci. Technol.2013, 43, 260–322. [CrossRef]

Microorganisms 2018, 6, 80 16 of 19

6. Neshat, S.A.; Mohammadi, M.; Najafpour, G.D.; Lahijani, P. Anaerobic co-digestion of animal manures andlignocellulosic residues as a potent approach for sustainable biogas production. Renew. Sustain. Energy Rev.2017, 79, 308–322. [CrossRef]

7. Mata-Alvarez, J.; Dosta, J.; Romero-Güiza, M.; Fonoll, X.; Peces, M.; Astals, S. A critical review on anaerobicco-digestion achievements between 2010 and 2013. Renew. Sustain. Energy Rev. 2014, 36, 412–427. [CrossRef]

8. Schnürer, A. Biogas production: Microbiology and technology. Adv. Biochem. Eng. Biotechnol. 2016, 156,195–234. [CrossRef]

9. Mulat, D.G.; Jacobi, H.F.; Feilberg, A.; Adamsen, A.P.S.; Richnow, H.-H.; Nikolausz, M. Changing feedingregimes to demonstrate flexible biogas production: Effects on process performance, microbial communitystructure, and methanogenesis pathways. Appl. Environ. Microbiol. 2016, 82, 438–449. [CrossRef] [PubMed]

10. Pap, B.; Maróti, G. Diversity of microbial communities in biogas reactors. Curr. Biochem. Eng. 2016, 3,177–187. [CrossRef]

11. Zhao, C.; Mu, H.; Zhao, Y.; Wang, L.; Zuo, B. Microbial characteristics analysis and kinetic studies onsubstrate composition to methane after microbial and nutritional regulation of fruit and vegetable wastesanaerobic digestion. Bioresour. Technol. 2018, 249, 315–321. [CrossRef] [PubMed]

12. Sun, L.; Liu, T.; Müller, B.; Schnürer, A. The microbial community structure in industrial biogas plantsinfluences the degradation rate of straw and cellulose in batch tests. Biotechnol. Biofuels 2016, 9, 1–20.[CrossRef] [PubMed]

13. Vrieze, J.; Pinto, A.; T Sloan, W.; Zeeshan Ijaz, U. The active microbial community more accurately reflectsthe anaerobic digestion process: 16s rRNA (gene) sequencing as a predictive tool. Microbiome 2018, 6, 63.[CrossRef] [PubMed]

14. De Vrieze, J.; Gildemyn, S.; Vilchez-Vargas, R.; Jáuregui, R.; Pieper, D.H.; Verstraete, W.; Boon, N. Inoculumselection is crucial to ensure operational stability in anaerobic digestion. Appl. Microbiol. Biotechnol. 2015, 99,189–199. [CrossRef] [PubMed]

15. Liu, T.; Sun, L.; Schnürer, B.M.A. Importance of inoculum source and initial community structure for biogasproduction from agricultural substrates. Bioresour. Technol. 2017, 245, 768–777. [CrossRef] [PubMed]

16. Moestedt, J.; Muller, B.; Westerholm, M.; Schnurer, A. Ammonia threshold for inhibition of anaerobicdigestion of thin stillage and the importance of organic loading rate. Microb. Biotechnol. 2016, 9, 180–194.[CrossRef] [PubMed]

17. Müller, B.; Sun, L.; Westerholm, M.; Schnürer, A. Bacterial community composition and fhs profiles oflow-and high-ammonia biogas digesters reveal novel syntrophic acetate-oxidising bacteria. Biotechnol.Biofuels 2016, 9, 1. [CrossRef] [PubMed]

18. Ziels, R.M.; Beck, D.A.; Martí, M.; Gough, H.L.; Stensel, H.D.; Svensson, B.H. Monitoring the dynamicsof syntrophic β-oxidizing bacteria during anaerobic degradation of oleic acid by quantitative PCR.FEMS Microbiol. Ecol. 2015, 91. [CrossRef] [PubMed]

19. De Vrieze, J.; Verstraete, W.; Boon, N. Repeated pulse feeding induces functional stability in anaerobicdigestion. Microb. Biotechnol. 2013, 6, 414–424. [CrossRef] [PubMed]

20. Lv, Z.; Hu, M.; Harms, H.; Richnow, H.H.; Liebetrau, J.; Nikolausz, M. Stable isotope composition of biogasallows early warning of complete process failure as a result of ammonia inhibition in anaerobic digesters.Bioresour. Technol. 2014, 167, 251–259. [CrossRef] [PubMed]

21. Ziels, R.M.; Beck, D.A.C.; Stensel, H.D. Long-chain fatty acid feeding frequency in anaerobic co-digestionimpacts syntrophic community structure and biokinetics. Water Res. 2017, 117, 218–229. [CrossRef] [PubMed]

22. Ziels, R.M.; Sousa, D.Z.; Stensel, H.D.; Beck, D.A.C. DNA-sip based genome-centric metagenomics identifieskey long-chain fatty acid-degrading populations in anaerobic digesters with different feeding frequencies.ISME J. 2018, 12, 112–123. [CrossRef] [PubMed]

23. Piao, Z.H.; Lee, J.; Kim, J.Y. Effect of substrate feeding frequencies on the methane production and microbialcommunities of laboratory-scale anaerobic digestion reactors. J. Mater. Cycles Waste 2018, 20, 147–154.[CrossRef]

24. Mauky, E.; Jacobi, H.F.; Liebetrau, J.; Nelles, M. Flexible biogas production for demand-driven energysupply—Feeding strategies and types of substrates. Bioresour. Technol. 2015, 178, 262–269. [CrossRef][PubMed]

25. Strömberg, S.; Nistor, M.; Liu, J. Towards eliminating systematic errors caused by the experimental conditionsin biochemical methane potential (bmp) tests. Waste Manag. 2014, 34, 1939–1948. [CrossRef] [PubMed]

Microorganisms 2018, 6, 80 17 of 19

26. Dechrugsa, S.; Kantachote, D.; Chaiprapat, S. Effects of inoculum to substrate ratio, substrate mix ratioand inoculum source on batch co-digestion of grass and pig manure. Bioresour. Technol. 2013, 146, 101–108.[CrossRef] [PubMed]

27. Westerholm, M.; Hansson, M.; Schnurer, A. Improved biogas production from whole stillage by co-digestionwith cattle manure. Bioresour Technol. 2012, 114, 314–319. [CrossRef] [PubMed]

28. Hansen, K.H.; Angelidaki, I.; Ahring, B.K. Anaerobic digestion of swine manure: Inhibition by ammonia.Water Res. 1998, 32, 5–12. [CrossRef]

29. Hugerth, L.W.; Wefer, H.A.; Lundin, S.; Jakobsson, H.E.; Lindberg, M.; Rodin, S.; Engstrand, L.;Andersson, A.F. Degeprime, a program for degenerate primer design for broad-taxonomic-range PCRin microbial ecology studies. Appl. Environ. Microbiol. 2014, 80, 5116–5123. [CrossRef] [PubMed]

30. Caporaso, J.G.; Kuczynski, J.; Stombaugh, J.; Bittinger, K.; Bushman, F.D.; Costello, E.K.; Fierer, N.;Pena, A.G.; Goodrich, J.K.; Gordon, J.I. Qiime allows analysis of high-throughput community sequencingdata. Nat. Methods 2010, 7, 335. [CrossRef] [PubMed]

31. Rideout, J.R.; He, Y.; Navas-Molina, J.A.; Walters, W.A.; Ursell, L.K.; Gibbons, S.M.; Chase, J.; McDonald, D.;Gonzalez, A.; Robbins-Pianka, A. Subsampled open-reference clustering creates consistent, comprehensiveotu definitions and scales to billions of sequences. PeerJ 2014, 2, e545. [CrossRef] [PubMed]

32. Edgar, R.C. Search and clustering orders of magnitude faster than blast. Bioinformatics 2010, 26, 2460–2461.[CrossRef] [PubMed]

33. Caporaso, J.G.; Bittinger, K.; Bushman, F.D.; DeSantis, T.Z.; Andersen, G.L.; Knight, R. Pynast: A flexible toolfor aligning sequences to a template alignment. Bioinformatics 2009, 26, 266–267. [CrossRef] [PubMed]

34. Haas, B.J.; Gevers, D.; Earl, A.M.; Feldgarden, M.; Ward, D.V.; Giannoukos, G.; Ciulla, D.; Tabbaa, D.;Highlander, S.K.; Sodergren, E. Chimeric 16s rRNA sequence formation and detection in sanger and454-pyrosequenced PCR amplicons. Genome Res. 2011. [CrossRef] [PubMed]

35. Wang, Q.; Garrity, G.M.; Tiedje, J.M.; Cole, J.R. Naive bayesian classifier for rapid assignment of rRNAsequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 2007, 73, 5261–5267. [CrossRef][PubMed]

36. Sun, L.; Müller, B.; Schnürer, A. Biogas production from wheat straw: Community structure ofcellulose-degrading bacteria. Energy Sustain. Soc. 2013, 3, 1–11. [CrossRef]

37. Ahlberg-Eliasson, K.; Nadeau, E.; Levén, L.; Schnürer, A. Production efficiency of swedish farm-scale biogasplants. Biomass Bioenergy 2017, 97, 27–37. [CrossRef]

38. Schnürer, A.; Bohn, I.; Moestedt, J. Protocol for Start-Up and Operation of CSTR Biogas Processes.In Hydrocarbon and Lipid Microbiology Protocols: Bioproducts, Biofuels, Biocatalysts and Facilitating Tools;McGenity, T.J., Timmis, K.N., Nogales, B., Eds.; Springer: Berlin/Heidelberg, Germany, 2017; pp. 171–200.

39. Lindorfer, H.; Demmig, C. Foam formation in biogas plants—A survey on causes and control strategies.Chem. Eng. Technol. 2016, 39, 620–626. [CrossRef]

40. He, Q.; Li, L.; Zhao, X.; Qu, L.; Wu, D.; Peng, X. Investigation of foaming causes in three mesophilic foodwaste digesters: Reactor performance and microbial analysis. Sci. Rep. 2017, 7, 13701. [CrossRef] [PubMed]

41. Westerholm, M.; Moestedt, J.; Schnürer, A. Biogas production through syntrophic acetate oxidation anddeliberate operating strategies for improved digester performance. Appl. Energy 2016, 179, 124–135.[CrossRef]

42. Ruile, S.; Schmitz, S.; Mönch-Tegeder, M.; Oechsner, H. Degradation efficiency of agricultural biogasplants—A full-scale study. Bioresour. Technol. 2015, 178, 341–349. [CrossRef] [PubMed]

43. De Francisci, D.; Kougias, P.G.; Treu, L.; Campanaro, S.; Angelidaki, I. Microbial diversity and dynamicityof biogas reactors due to radical changes of feedstock composition. Bioresour. Technol. 2015, 176, 56–64.[CrossRef] [PubMed]

44. Wang, P.; Wang, H.; Qiu, Y.; Ren, L.; Jiang, B. Microbial characteristics in anaerobic digestion process of foodwaste for methane production-review. Bioresour. Technol. 2018, 248, 29–36. [CrossRef] [PubMed]

45. Ma, H.; Ye, L.; Hu, H.; Zhang, L.; Ding, L.; Ren, H. Determination and variation of core bacterialcommunity in a two-stage full-scale anaerobic reactor treating high-strength pharmaceutical wastewater.J. Microbiol. Biotechnol. 2017, 27, 1808–1819. [CrossRef] [PubMed]

Microorganisms 2018, 6, 80 18 of 19

46. Christy, P.M.; Gopinath, L.; Divya, D. A review on anaerobic decomposition and enhancement of biogasproduction through enzymes and microorganisms. Renew. Sustain. Energy Rev. 2014, 34, 167–173. [CrossRef]

47. Ueki, A.; Akasaka, H.; Suzuki, D.; Ueki, K. Paludibacter propionicigenes gen. nov., sp. nov., a novel strictlyanaerobic, Gram-negative, propionate-producing bacterium isolated from plant residue in irrigated rice-fieldsoil in Japan. Int. J. Syst. Evol. Microbiol. 2006, 56, 39–44. [CrossRef] [PubMed]

48. Chen, H.; Chang, S. Impact of temperatures on microbial community structures of sewage sludge biologicalhydrolysis. Bioresour. Technol. 2017, 245, 502–510. [CrossRef] [PubMed]

49. Zhang, Y.; Li, L.; Kong, X.; Zhen, F.; Wang, Z.; Sun, Y.; Dong, P.; Lv, P. Inhibition effect of sodiumconcentrations on the anaerobic digestion performance of Sargassum species. Energy Fuels 2017, 31, 7101–7109.[CrossRef]

50. Girija, D.; Deepa, K.; Xavier, F.; Antony, I.; Shidhi, P. Analysis of cow dung microbiota—A metagenomicapproach. Indian J. Biotechnol. 2013, 12, 372–378.

51. Dai, Y.; Yan, Z.; Jia, L.; Zhang, S.; Gao, L.; Wei, X.; Mei, Z.; Liu, X. The composition, localization and functionof low-temperature-adapted microbial communities involved in methanogenic degradations of cellulose andchitin from qinghai-tibetan plateau wetland soils. J. Appl. Microbiol. 2016, 121, 163–176. [CrossRef] [PubMed]

52. El-Bery, H.; Tawfik, A.; Kumari, S.; Bux, F. Effect of thermal pre-treatment on inoculum sludge to enhancebio-hydrogen production from alkali hydrolysed rice straw in a mesophilic anaerobic baffled reactor.Environ. Technol. 2013, 34, 1965–1972. [CrossRef] [PubMed]

53. Garg, R.; Srivastava, R.; Brahma, V.; Verma, L.; Karthikeyan, S.; Sahni, G. Biochemical and structuralcharacterization of a novel halotolerant cellulase from soil metagenome. Sci. Rep. 2016, 6, 39634. [CrossRef][PubMed]

54. Zhao, L.; Meng, Q.; Ren, L.; Liu, W.; Zhang, X.; Huo, Y.; Zhou, Z. Effects of nitrate addition on rumenfermentation, bacterial biodiversity and abundance. Asian-Austral. J. Anim. 2015, 28, 1433–1441. [CrossRef][PubMed]

55. Fernandes, K.A.; Kittelmann, S.; Rogers, C.W.; Gee, E.K.; Bolwell, C.F.; Bermingham, E.N.; Thomas, D.G.Faecal microbiota of forage-fed horses in New Zealand and the population dynamics of microbialcommunities following dietary change. PLoS ONE 2014, 9, e112846. [CrossRef] [PubMed]

56. Klevenhusen, F.; Petri, R.M.; Kleefisch, M.-T.; Khiaosa-ard, R.; Metzler-Zebeli, B.U.; Zebeli, Q. Changesin fibre-adherent and fluid-associated microbial communities and fermentation profiles in the rumen ofcattle fed diets differing in hay quality and concentrate amount. FEMS Microbiol. Ecol. 2017, 93. [CrossRef][PubMed]

57. Yokoyama, H.; Wagner, I.D.; Wiegel, J. Caldicoprobacter oshimai gen. Nov., sp. Nov., an anaerobic, xylanolytic,extremely thermophilic bacterium isolated from sheep faeces, and proposal of caldicoprobacteraceae fam.Nov. Int. J. Syst. Evol. Microbiol. 2010, 60, 67–71. [CrossRef] [PubMed]

58. Bouanane-Darenfed, A.; Hania, W.B.; Hacene, H.; Cayol, J.-L.; Ollivier, B.; Fardeau, M.-L. Caldicoprobacterguelmensis sp. Nov., a thermophilic, anaerobic, xylanolytic bacterium isolated from a hot spring. Int. J. Syst.Evol. Microbiol. 2013, 63, 2049–2053. [CrossRef] [PubMed]

59. Deng, Y.; Huang, Z.; Zhao, M.; Ruan, W.; Miao, H.; Ren, H. Effects of co-inoculating rice straw with ruminalmicrobiota and anaerobic sludge: Digestion performance and spatial distribution of microbial communities.Appl. Microbiol. Biotechnol. 2017, 101, 5937–5948. [CrossRef] [PubMed]

60. Mulat, D.G.; Huerta, S.G.; Kalyani, D.; Horn, S.J. Enhancing methane production from lignocellulosicbiomass by combined steam-explosion pretreatment and bioaugmentation with cellulolytic bacteriumCaldicellulosiruptor bescii. Biotechnol. Biofuels 2018, 11, 19. [CrossRef] [PubMed]

61. Poirier, S.; Desmond-Le Quéméner, E.; Madigou, C.; Bouchez, T.; Chapleur, O. Anaerobic digestion ofbiowaste under extreme ammonia concentration: Identification of key microbial phylotypes. Bioresour.Technol. 2016, 207, 92–101. [CrossRef] [PubMed]

62. Vos, P.; Garrity, G.; Jones, D.; Krieg, N.R.; Ludwig, W.; Rainey, F.A.; Schleifer, K.-H.; Whitman, W.B. Bergey’sManual of Systematic Bacteriology: Volume 3: The Firmicutes; Springer Science & Business Media: Berlin,Germany, 2011.

63. Poehlein, A.; Zverlov, V.V.; Daniel, R.; Schwarz, W.H.; Liebl, W. Complete genome sequence of Clostridiumstercorarium subsp. stercorarium Strain DSM 8532, a Thermophilic Degrader of Plant Cell Wall Fibers.Genome Announc. 2013, 1, e00073-13.

Microorganisms 2018, 6, 80 19 of 19

64. Wei, Y.; Zhou, H.; Zhang, J.; Zhang, L.; Geng, A.; Liu, F.; Zhao, G.; Wang, S.; Zhou, Z.; Yan, X. Insight intodominant cellulolytic bacteria from two biogas digesters and their glycoside hydrolase genes. PLoS ONE2015, 10, e0129921. [CrossRef] [PubMed]

65. Su, H.; Liu, L.; Wang, S.; Wang, Q.; Jiang, Y.; Hou, X.; Tan, T. Semi-continuous anaerobic digestion forbiogas production: Influence of ammonium acetate supplement and structure of the microbial community.Biotechnol. Biofuels 2015, 8, 13. [CrossRef] [PubMed]

66. Haroun, B.M.; Nakhla, G.; Hafez, H.; Velayutham, P.; Levin, D.B.; Derakhshani, H.; Khafipour, E.; Nasr, F.A.Significance of acclimatization for biohydrogen production from synthetic lignocellulose hydrolysate incontinuous-flow systems. Int. J. Hydrogen Energy 2016, 41, 14003–14014. [CrossRef]

67. Dewhirst, F.E.; Paster, B.J.; Tzellas, N.; Coleman, B.; Downes, J.; Spratt, D.A.; Wade, W.G. Characterizationof novel human oral isolates and cloned 16S rDNA sequences that fall in the family Coriobacteriaceae:Description of Olsenella gen. Nov., reclassification of Lactobacillus uli as Olsenella uli comb. Nov. Anddescription of Olsenella profusa sp. Nov. Int. J. Syst. Evol. Microbiol. 2001, 51, 1797–1804.

68. Sundberg, C.; Al-Soud, W.A.; Larsson, M.; Alm, E.; Yekta, S.S.; Svensson, B.H.; Sørensen, S.J.; Karlsson, A. 454pyrosequencing analyses of bacterial and archaeal richness in 21 full-scale biogas digesters. FEMS Microbiol.Ecol. 2013, 85, 612–626. [CrossRef] [PubMed]

69. Pore, S.D.; Shetty, D.; Arora, P.; Maheshwari, S.; Dhakephalkar, P.K. Metagenome changes in the biogasproducing community during anaerobic digestion of rice straw. Bioresour. Technol. 2016, 213, 50–53.[CrossRef] [PubMed]

70. Goux, X.; Calusinska, M.; Lemaigre, S.; Marynowska, M.; Klocke, M.; Udelhoven, T.; Benizri, E.; Delfosse, P.Microbial community dynamics in replicate anaerobic digesters exposed sequentially to increasing organicloading rate, acidosis, and process recovery. Biotechnol. Biofuels 2015, 8, 122. [CrossRef] [PubMed]

71. Sleat, R.; Mah, R.A.; Robinson, R. Isolation and characterization of an anaerobic, cellulolytic bacterium,Costridium cellulovorans sp. Nov. Appl. Environ. Microbiol. 1984, 48, 88–93.

72. Renouf, M.; Hendrich, S. Bacteroides uniformis is a putative bacterial species associated with the degradationof the isoflavone genistein in human feces. J. Nutr. 2011, 141, 1120–1126. [CrossRef] [PubMed]

73. Koeck, D.E.; Hahnke, S.; Zverlov, V.V. Herbinix luporum sp. Nov., a thermophilic cellulose-degradingbacterium isolated from a thermophilic biogas reactor. Int. J. Syst. Evol. Microbiol. 2016, 66, 4132–4137.[PubMed]

74. Maus, I.; Bremges, A.; Stolze, Y.; Hahnke, S.; Cibis, K.G.; Koeck, D.E.; Kim, Y.S.; Kreubel, J.; Hassa, J.;Wibberg, D. Genomics and prevalence of bacterial and archaeal isolates from biogas-producing microbiomes.Biotechnol. Biofuels 2017, 10, 264. [CrossRef] [PubMed]

75. Krieg, N.R.; Ludwig, W.; Euzéby, J.; Whitman, W.B. Phylum xiv. Bacteroidetes phyl. Nov. In Bergey’smanual®of systematic bacteriology: Volume four The Bacteroidetes, Spirochaetes, Tenericutes (Mollicutes), Acidobacteria,Fibrobacteres, Fusobacteria, Dictyoglomi, Gemmatimonadetes, Lentisphaerae, Verrucomicrobia, Chlamydiae, andPlanctomycetes; Krieg, N.R., Staley, J.T., Brown, D.R., Hedlund, B.P., Paster, B.J., Ward, N.L., Ludwig, W.,Whitman, W.B., Eds.; Springer: New York, NY, USA, 2010; pp. 25–469.

76. Benítez-Páez, A.; Gómez del Pulgar, E.M.; Sanz, Y. The glycolytic versatility of Bacteroides uniformis CECT7771 and its genome response to oligo and polysaccharides. Front. Cell. Infect. Microbiol. 2017, 7, 383.[CrossRef] [PubMed]

77. Koeck, D.E.; Maus, I.; Wibberg, D.; Winkler, A.; Zverlov, V.V.; Liebl, W.; Pühler, A.; Schwarz, W.H.; Schlüter, A.Complete genome sequence of Herbinix luporum SD1D, a new cellulose-degrading bacterium isolated from athermophilic biogas reactor. Genome Announc. 2016, 4, 00687-16. [CrossRef] [PubMed]

78. Gold, N.D.; Martin, V.J.J. Global view of the Clostridium thermocellum cellulosome revealed by quantitativeproteomic analysis. J. Bacterial. 2007, 189, 6787–6795. [CrossRef] [PubMed]

79. Swamy, M.V.; Ram, M.S.; Seenayya, G. B-amylase from Clostridium thermocellum SS8—A thermophilic,anaerobic, cellulolytic bacterium. Lett. Appl. Microbiol. 1994, 18, 301–304. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open accessarticle distributed under the terms and conditions of the Creative Commons Attribution(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

OR I G I N A L A R T I C L E

Forage types and origin of manure in codigestion affectmethane yield and microbial community structure

K. Ahlberg-Eliasson1,2 | T. Liu2 | E. Nadeau1,3 | A. Schn€urer2

1Swedish Rural Economy and Agricultural

Society, L€anghem, Sweden

2Department of Molecular Sciences,

Biocenter, Swedish University of

Agricultural Sciences, Uppsala, Sweden

3Department of Animal Environment and

Health, Swedish University of Agricultural

Sciences, Skara, Sweden

Correspondence

Karin Ahlberg-Eliasson, Swedish Rural

Economy and Agricultural Society, L€anghem,

Sweden.

Email: karin.eliasson@hushallningssallskapet.se

Funding information

Swedish Energy Agency, as part of the

project, Grant/Award Number: 36474-1;

V€astra G€otaland County Council and the

Chinese Scholarship Council (CSC)

Abstract

In farm-scale biogas systems, different kinds of manure are the most important sub-

strate for anaerobic digestion, but result in low biogas yields. Biogas production can

be increased by complementing the manure with forage crops, in codigestion. The

aim of this study was to evaluate grass-clover (GCS) and whole-crop barley silages

(WCB) in codigestion with manure from organic and conventional dairy production

systems on biogas production, microbial community, degree of degradation and gas

quality at different organic loading rates by addition of soya bean meal and wheat

grain, which are rich in protein and starch. Four continuous stirred anaerobic labora-

tory-scale reactors were used, and the codigestion resulted in additive effects on

biogas production, but no synergistic effects. The highest biogas yield was obtained

in reactors receiving WCB independently of manure types, for both experiments

(7,416 ml/day and 10,978 ml/day respectively). The degradation efficiency, mea-

sured as the reduction in volatile solids was, on average, six percentage units higher

in the reactors receiving manure from conventional compared with organic dairy

cows, probably because of a higher concentration of undigested fibre and proteins

in conventional cow manure. Microbiological analysis by illumina sequencing illus-

trated low impact of both manure types on the reactor community and only small

differences between the reactors receiving GCS and WCB. However, addition of

soya bean meal and wheat grain changed the community in all reactors. The ratio

between Firmicutes and Bacteroidetes was comparably higher in reactors having the

highest gas production and methane yield.

K E YWORD S

biogas, digestate, greenhouse gas emissions, microbiology, silage

1 | INTRODUCTION

Anaerobic digestion at farm level is an important tool for producing

renewable energy and reducing greenhouse gas emissions from the

agriculture sector (Holm-Nielsen, Al Seadi, & Oleskowicz-Popiel,

2009; Insam, G�omez-Brand�on, & Ascher, 2015; Pucker, Jungmeier,

Siegl, & Poetsch, 2013). Manure is a suitable substrate for biogas

production, as it is available in large amounts for many farms operat-

ing biogas plants. Manure is also feasible for the microbiological pro-

cess because of its content of trace minerals and stabilization of pH

in the biogas digester (Tufaner & Avs�ar, 2016). However, the

concentration of organic matter in manure is low, and the water

content is high, resulting in low organic loads in the digester and

consequently low volumetric gas production. One way to improve

the biogas yield from manure is to mix in more energy-dense sub-

strate(s) for codigestion. Codigestion with forage crops added to

manure, allows increased organic loading and, consequently,

increased gas yield, with only marginal effects on retention time

(Lehtomaki, Huttunen, & Rintala, 2007; Mata-Alvarez et al., 2014;

Sondergaard, Fotidis, Kovalovszki, & Angelidaki, 2015). Alternative

use of forage crops in codigestion with manure for biogas produc-

tion increases the agricultural land use and the environmental

Received: 1 September 2017 | Revised: 28 February 2018

DOI: 10.1111/gfs.12358

740 | © 2018 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/gfs Grass Forage Sci. 2018;73:740–757.

sustainability of farm operations. The chemical and physical parame-

ters in the crop, and the biogas yield, are influenced by cultivation,

harvest time, silage fermentation and pre-treatment (Heiermann,

Ploechl, Linke, Schelle, & Herrmann, 2009; Lehtomaki, 2006; McE-

niry & O’Kiely, 2013; Moller et al., 2008). Codigestion can improve

the carbon:nitrogen (C/N) ratio and nutrient balance, contributing

buffering capacity and, thus, stabilizing the pH, all potentially

improving the biogas process performance. Substrate composition

and operating parameters such as retention time and organic load

also influence development of microbiological communities in the

reactor (Schn€urer, 2016). For an efficient biogas production process,

the conditions must be set to secure the activity of groups with dif-

ferent metabolic function, which are often also dependent on each

other. For this reason, optimal codigestion could have a synergistic

effect in improving the overall degradation of organic matter in the

reactor. Codigestion can also result in so-called priming, caused by a

boost of enzymatic activity, triggered by easily available organic mat-

ter, resulting in improved degradation of recalcitrant material such as

lignocellulose (Aichinger et al., 2015; Insam & Markt, 2016).

In the European Union, biogas production is approx. 8% of total

renewable energy production, and about 50% of the biogas is pro-

duced using energy crops, mainly maize (European Comission,

2016b). However, while the potential for biogas is large in the EU,

there are still barriers to production. For crops specifically, the ILUC

criteria limiting cultivation of energy crops on land suitable for food

production are problematic for the development of biogas markets

(European Comission, 2016a). On the other hand, the Food and Agri-

culture Organization of the United Nations (FAO) clearly states that

modern agriculture structures that integrate food and energy sys-

tems are climate-friendly and highly important for sustainable food

production (FAO, 2014a,b). Recent research also highlighted the

importance of energy as part of sustainable intensification of agricul-

tural production (Rockstrom et al., 2017).

Farm-scale biogas production can be an important tool in a long-

term sustainable agricultural food chain. To balance the most likely

forthcoming regulation on biomass production for energy in the EU,

energy crops used in farm-scale biogas systems need to show low

environmental impact during cultivation, harvest and storage (Bauer

et al., 2010). Grass-legume forages and whole-crop cereals, as nurse

crops to undersown leys and as catch crops, show low environmen-

tal impact compared with other crops, such as maize, when used in

a biogas system (Borjesson, Prade, Lantz, & Bjornsson, 2015; Giss�en

et al., 2014; Luscher, Mueller-Harvey, Soussana, Rees, & Peyraud,

2014; Moller et al., 2008). However, most previous studies on codi-

gestion of manure and energy crops have focused on maize silage

and cereals, while less attention has been paid to other potential

crops, for biogas production (Wall, Allen, Straccialini, O’Kiely, & Mur-

phy, 2014). To our knowledge, there has been little research on

codigestion of whole-crop barley and grass-clover forage in a contin-

uous biogas system. Whole-crop barley and grass-clover forage are

common crops on a livestock farm as they have complementary

nutrient compositions for a diet formulation and would therefore be

relevant to test for biogas production in codigestion with manure.

Moreover, only few studies have evaluated the effect of manure

composition in codigestion with whole-crop barley and grass-clover

forages (Lehtomaki, Huttunen, & Rintala, 2007). The complementary

composition of the material used for codigestion influences the out-

come of the biogas process. Cattle manure composition varies and is

affected by the feedstuffs in the diet, and how well these are

degraded in the rumen. For this reason, manures produced in con-

ventional and organic dairy production systems differ (Mgbeahuruike,

Norgaard, Eriksson, Nordqvist, & Nadeau, 2016). In addition, codi-

gestion with this type of manure and crops has to our knowledge

not been studied at various loading rates. The hypothesis of this

study was that different forage types and origin of manure will

impact process efficiency, microbial community and, consequently,

the final gas production. To corroborate this hypothesis, codigestion

of cattle manure from conventional and organic dairy production

with grass-clover silage or whole-crop barley silage as forage crops

was investigated in two experiments, which differed in organic load-

ing rates (OLRs). The processes were evaluated in term of: (i) overall

biogas production, (ii) degree of degradation, (iii) gas quality, (iv)

digestate quality and (v) microbiological community structure.

2 | MATERIAL AND METHODS

2.1 | Experimental set-up

Two different methods, batch systems and continuous stirred tank

reactors (CSTR), were used to evaluate biogas production from the

different substrates. The substrates evaluated were two different

types of manure (organic cattle manure [OCM] and conventional cat-

tle manure [CCM]) and two different types of energy crops (grass-

clover silage [GCS] and whole-crop barley silage [WCB]). In addition,

the biomethane potential (BMP) from a substrate mix of soya bean

meal and wheat grain (SW) was measured in the batch system

(Table 1).

2.1.1 | Substrate sampling and handling

The whole-crop barley and grass-clover ley used in the experiment

were harvested at R�adde Experimental Farm, L€anghem Sweden

(57.60740N–13.25998W) in 2013. The whole-crop barley was har-

vested, at the late milk to early dough stage of maturity, on 18 July

and ensiled in round bales at 40% dry matter (DM), and the grass-

clover forage was harvested in the second cut on the 2nd of July at

a DM content of 39%. The grass-clover forage consisted of 40%

timothy (Phleum pratense L.) at head emergence stage, 15% perennial

ryegrass (Lolium perenne L.) at bloom stage, 3% Festulolium (Lolium

multiflorum L. 9 Festulolium arundinacea L.) at leaf stage, 31% red

clover (Trifolium pratense L.) and 11% white clover (Trifolium repens

L.) at bloom stage on a DM basis. A silage additive containing

sodium nitrite, hexamine and sodium benzoate was applied to the

forages at 2 L/tonne at harvest. The bales were opened in Novem-

ber 2014 and chopped using a precision chopper (John Deere,

7,350). The particle size of the silages was determined in a Penn

AHLBERG-ELIASSON ET AL. | 741

State Particle Size Separator (Heinrichs & Kononoff, 2002). The

mean particle size for the GCS was 6.5 � 2.5 mm and for WCB

6.3 � 2.2 mm. Silages were frozen at �18°C and thawed directly

before use in the trial.

Liquid manure from a conventional (CCM) and an organic (OCM)

dairy herd was collected directly from lactating dairy cows. The man-

ure was frozen in batches (3 kg) at �18°C and thawed directly

before use as substrate in the experiments. Before being used in the

CSTR or in the batch system, both types of manure were diluted

with water to a final content of DM of 8%. The mix of soya bean

meal and wheat grain was purchased from a local supplier of seeds

and mixed before use to a ratio of 50:50 based on volume. Chemical

composition of the different substrates is presented in Table 1.

2.1.2 | Batch system

The biomethane potential (BMP) for WCB, GCS, OCM and CCM, for

their mixes OCM:WCB, OCM:GCS, CCM:WCB and CCM:GCS and for

SW was determined using an automatic methane potential test sys-

tem (Bioprocess Control AB). In the substrate mixes, the contribution

of VS was set to 50:50 between the manure and the crop included.

The batches were incubated for 30 days at 38°C. The inoculum used

in the test was collected from a full-scale biogas plant operating at

mesophilic temperature using a substrate mix of grass silage and grain.

At the time for sampling, the pH value was approx. 7.5 and total VFA

concentration 300 mg/L. According to the standard set-up of BMP

tests, the substrate to inoculum ratio on a volatile solids (VS) basis

was set to 1:3, and the organic loading of substrate was 3 g VS/L

(Schn€urer, Bohn, & Moestedt, 2017). The total active volume of the

bottles was 400 ml, and tap water was used for dilution to reach the

final volume. Each substrate/substrate mix was run in triplicate bot-

tles, and a blank measuring background methane was also analysed in

triplicate using the same amount of inoculum, but without addition of

substrate. Cellulose (SIGMA, Cellulose fibres medium CAS 9004-34-6)

was used as a standard in triplicate bottles. Gas volume was calculated

at standard temperature and pressure.

2.1.3 | Continuous stirred tank reactors

For the CSTR experiment, four digesters (Dolly, Belach Biotechnol-

ogy AB) named R1-R4 were operated. The experiments were run in

two periods and designated Exp A and Exp B. During Exp A, the

digesters were operated with substrate mixtures of manure and

energy crop. R1 received a mix of OCM and WCB, R2 received a

mix of OCM and GCS, R3 received CCM and WCB, and R4 received

CCM and GCS. In Exp B, 6 g/day of SW (corresponding to approx.

25% of total added VS) was added as a substrate in R1, R2, R3 and

R4 to increase the OLRs (Table 2).

For the start-up of the reactors, inoculum from the same full-

scale biogas plant as for the batch cultures was used. At start, 6 L of

the inoculum was transferred to the digesters, and the OLRs were

successively increased from 1 g VS/L day to 2.8 g VS/L day during

57 days using the substrate mix of manure and energy crops. There-

after, when full load was reached, the first experiment (Exp A) was

started, running for 127 days. On day 128, Exp B was initiated, and

between day 128 and day 137, the OLRs were increased by 1 g VS/

L day, reaching a final OLRs of 3.6–3.8 g VS/L day, by addition of

SW to all four reactors, running for 106 days. The reactors operated

at 38°C, and the hydraulic retention time (HRT) for both of the

experimental periods was 30 days (Table 2).

The digesters were fed once a day for 6 days a week. Active vol-

ume used in the CSTR was set to 6 L of a total 10 L digester vol-

ume. Gas production (GP) was measured online with a volumetric

TABLE 1 Chemical composition of the substrates used in thebatch system and in the continuous stirred tank reactors in

experiment (Exp) A and B

Parameters Units WCB GCS CCM OCM SW

DM % of ww 40 43 10 11 95

VS % of ww 38 39 9 9 91

Crude protein g/kg DM 71 161 138 121 318

Starch g/kg DM 63 11 9 11 316

NDF g/kg DM 486 415 535 476 139

ADF g/kg DM 291 317 361 368 ND

ADL g/kg DM 53 78 99 121 ND

Crude fat g/kg DM 12 26 31 32 33

WSC g/kg DM 109 31 ND ND ND

VOS % of VS 72 77 38 36 ND

pH 4.2 4.6 7.0 7.5 ND

N-Tot g/kg DM 11.4 25.8 34.8 39.3 50.8

Am-N g/kg DM 0.7 2.0 11.3 18.1 ND

Ca g/kg DM 4.6 15.0 9.1 18.9 2.1

K g/kg DM 10.6 24.9 22.5 44.2 14.7

P g/kg DM 2.4 2.9 5.5 5.6 4.9

Mg g/kg DM 1.3 2.8 7.3 6.3 2.4

Na g/kg DM 1.3 1.1 3.0 5.3 0.2

S g/kg DM 1.2 2.0 4.3 3.2 2.6

Cu mg/kg DM 7.0 9.0 28.0 32.0 9.6

Fe mg/kg DM 618 147 1,223 1,218 99

Mn mg/kg DM 18 44 210 284 29

Zn mg/kg DM 17 24 162 122 43

Lactate g/kg DM 28.7 52.2 0.9 0.8 ND

Acetate g/kg DM 13.1 12.5 38.9 50.8 ND

Propionate g/kg DM 0.5 1.2 2.8 3.1 ND

Butyrate g/kg DM 0.2 0.1 0.9 0.8 ND

Ethanol g/kg DM 7.0 1.6 0.9 0.8 ND

Formic acid g/kg DM 1.1 0.1 0.9 0.8 ND

2,3-butanediol g/kg DM 1.7 3.9 0.9 0.8 ND

WCB, whole-crop barley silage; GCS, grass-clover silage; CCM, conven-

tional cattle manure; OCM, organic cattle manure; SW, soya bean meal

and wheat grain; DM, dry matter; VS, volatile solids; NDF, neutral deter-

gent fibre; ADF, acid detergent fibre; ADL, acid detergent lignin; WSC,

water-soluble carbohydrates (i.e. sugar); VOS, in vitro organic matter

digestibility; N-tot, total ammonia nitrogen; Am-N, Ammonia nitrogen;

ND, not determined.

742 | AHLBERG-ELIASSON ET AL.

gas counter calibrated by analysis of collected gas volume, using a

drum gas meter (Ritter TG0.5/5, Dr-Ing. Ritter Apparatebau GMBH

& Co. Kg). The content of carbon dioxide in the gas was measured

every day using a saccharometer filled with saturated sodium

hydroxide (7M). Concentrations of hydrogen sulphide, carbon diox-

ide, methane and oxygen were measured on nine occasions during

Exp A (day 88, 92, 94, 97, 105, 108, 115, 118, 121) and on five

occasions during Exp B (day 139, 143, 197, 216, 219) using a gas

analyser (Sewerin Multitec 540; PPM M€atteknik, Industriell Gasm€at-

ning AB, Hisings Backa). The methane content in the gas was also

determined by gas chromatography (Clarus 500, Perkin Elmer, USA,

Polyimide Uncoated capillary column 5 m 9 0.32 mm, FID detector)

on five occasions during Exp A (day 88, 94, 105, 110, 118) and on

four occasions during Exp B (day 140, 153, 174, 187). All the gas

samples were collected at the same time each day and at the same

interval (approx. 1 hr) after feeding. All volumetric gas values were

converted to standard conditions at pressure 1.01 bar and tempera-

ture 273.2 K. The HRT, OLRs, GP and specific methane production

(SMP) were calculated based on average weekly values.

2.1.4 | Analytical methods

Single substrates, substrate mixes and digestate from the CSTR

digesters were analysed in the laboratory of the Department of Ani-

mal Nutrition and Management, Swedish University of Agricultural

Sciences, Uppsala, Sweden. Concentrations of neutral detergent fibre

(NDF), acid detergent fibre (ADF), acid detergent lignin (ADL), starch,

water-soluble carbohydrate (WSC), crude protein, ash and in vitro

digestibility of organic matter (OM) were determined on dried sam-

ples that had been milled (Kamas Kvarnmaskiner AB, Malm€o, Swe-

den) to pass through a 1-mm screen. Concentrations of NDF, ADF

and ADL were determined according to Van Soest, Robertson, and

Lewis (1991). Starch was analysed by an enzymatic method accord-

ing to Aman and Hesselman (1984) and WSC according to Larsson

and Bengtsson (1983). Crude protein concentration was determined

as total N concentration using the Kjeldahl technique in a Tecator

Auto Sampler 1035 Analyzer (Tecator Inc, H€ogan€as, Sweden).

Ammonia-N concentration was determined by the FIA method using

the Tecator Kjeltec Auto Sampler System 1035 Analyzer (Tecator,

Application Note, ASN 50-01/92.) In vitro rumen digestibility of

organic matter (VOS) was determined by incubating untreated silage

samples in rumen fluid and buffer at 39°C for 96 hr (Lindgren, 1979,

1983). The DM and VS contents in the digestate and substrates

were determined and corrected for losses of volatile compounds

according to �Akerlind et al. (2011). The pH in the substrates was

measured with a Metrohm 654 pH meter (Metrohm AG, Herisau,

Switzerland). Minerals (phosphorus [P], calcium [Ca], magnesium

[Mg], sodium [Na], sulphur [S], iron [Fe], copper [Cu], manganese

[Mn] and zinc [Zn]) were analysed according to Balsberg-P�ahlsson

(1990) using a spectrophotometer (SPECTRO BLUE, Model: FMS26).

Digestate samples of 500 ml were taken prior feeding at day 111

for Exp A and day 213 for Exp B at the end of each experiment.

Organic acids and alcohols, in the crops, were quantified using HPLC

(Ericson & Andr�e, 2010). The content of volatile fatty acids (VFA) in

the reactor liquid was measured during the start-up phase of the

experimental phases, on seven occasions during Exp A (day 3, 13,

24, 41, 48, 63, 105) and on three occasions during Exp B (day 141,

153, 174), by HPLC analysis (Westerholm, Roos, & Schnurer, 2010).

Digestate samples were collected on day 68, 88, 101, 108, 154, 197,

217, analysed in triplicate, were used to calculate the degree of

degradation, that is reduction in VS after digestion compared with

the VS content in the substrate. Digestibility of NDF and protein

was calculated using Equation 1, M_in is the mass in gram of sub-

strate and M_out is the mass in gram, of digestate. Crude protein

content of the substrate (Table 1) was calculated as total N (N-

tot) 9 6.25.

Digestibility output ¼ 100� ððM in�M outÞ=M inÞ (1)

2.2 | Microbiological data

Samples for the microbiological analysis were taken from the two

types of manures, the inoculum used for start-up and the digestate

from the CSTR (day 94, 105, 174, 211).

2.2.1 | 16s RNA genes amplicon librariesconstruction

To build the amplicon libraries for illumina sequencing, the primer

sets 515`F and 805R were applied to amplify the 16S rRNA genes

(Hugerth et al., 2014), following the steps described by M€uller, Sun,

Westerholm, and Schn€urer (2016). For PCR amplification, 1.25 ll

TABLE 2 General operating parameters (OP) for experiments (Exp) A and B

Reactor name (substrate)aStart-uptime (days)

ExperimenttimeExp A/Exp B (days)

Temp(°C)

OLRsA/B

(g VS/L day)OLRsExp Ab (%)

OLRs ExpBc (%) HRT (days)

R1 (OCM:WCB) 57 127/106 38 2.8/3.8 52:48 39:36:25 30

R2 (OCM:GCS) 57 127/106 38 2.7/3.6 54:46 40:34:26 30

R3 (CCM:WCB) 57 127/106 38 2.7/3.8 52:48 39:36:25 30

R4 (CCM:GCS) 57 127/106 38 2.8/3.6 54:46 40:34:26 30

WCB, whole-crop barley silage; GCS, grass-clover silage; CCM, conventional cattle manure; OCM, organic cattle manure.aMain substrate used for each reactor (R1–R4). In Exp B, all reactors also received the mix of SW.bPercentages (VS basis) of manure and energy crop in Exp A (running for 127 days).cPercentage (VS basis) of manure, energy crop and soya bean meal/wheat grain (SW) in Exp B (running for 106 days).

AHLBERG-ELIASSON ET AL. | 743

(10 lM) of each primer, 12.5 ll of 29 Phusion MasterMix (Thermo

Scientific, Waltham, MA, USA) and 8 ll nuclease-free H2O were

added with 2 ll of extracted DNA from triplicate extractions to a

total volume of 25 ll. The PCR was programmed as 98°C 30 s, 20

cycles of 98°C 10 s, 60°C 30 s, 72°C 4 s and final elongation at

72°C 2 min. The PCR products obtained were checked by gel elec-

trophoresis (1.5% agarose, with 0.59 Tris-borate-EDTA buffer). They

were then purified (20 ll of the PCR products) using magnetic beads

according to Agencourt AMPure Purification Method (AMPure XP,

Beckman Coulter Genomics, Beverly, MA, USA) and eluted with

20 ll EB buffer. A second PCR, programmed as 98°C 30 s, 10 cycles

of 98°C 10 s, 62°C 30 s, 72°C 5 s and final elongation 72°C 2 min,

was applied to attach the barcodes and adapters for the later illu-

mina sequencing. For the second PCR, 1.25 ll of each primer, 10 ll

of cleaned PCR products from the last step and 12.5 ll of 29 Phu-

sion MasterMix were added. The same cleaning step was applied for

the second PCR products obtained, and Qubit (Invitrogen, Thermo

Fisher Science, Waltham, MA, USA) was used to quantify the final

PCR products. Finally, 2 ll portions of each final adjusted PCR pro-

duct (concentration of each sample was adjusted to 5 nM with EB

buffer) were pooled and sequenced using the MiSeq system (SciLife-

Lab in Stockholm, Sweden).

2.2.2 | Illumina sequencing and data analysis

The raw DNA sequencing data obtained were deposited to National

Centre for Biotechnology Information database (NCBI) Sequence

Read Archive (SRA) under accession number SRP119953 and anal-

ysed through the open-source bioinformatics pipeline, quantitative

insights into microbial ecology (QIIME) with loaded module bioingo-

tools: Qiime/1.8.0/1.9.1, SeqPrep and Cutadapt, following steps

described in M€uller et al. (2016).

2.3 | Statistical analysis

Data from the CSTR experiments Exp A and Exp B were analysed

separately in the mixed model procedure of SAS version 9.4 (2015).

Manure and forage were treated as fixed factors and week (Exp A

13 weeks, Exp B 13 weeks) as a random factor with repeated mea-

sures on the reactors. The reactors (R1–R4) were taken as experi-

mental units, and the covariance structure was AR (1). Main effects

of cattle manure and energy crop and their interactions on SMP and

GP were studied. When a significant F-value was detected at

p < .05, or a tendency for significance (.05 < p < .10) was detected,

differences between the substrates were tested by Tukey’s test. The

significance levels were set at p < .001, p < .01 and p < .05.

3 | RESULTS

3.1 | Biomethane potential

The BMP for WCB, GCS, OCM, CCM and SW was determined

after 30 days of incubation, when the methane production had

levelled off (Table 3). The methane production from the crops was

faster than the production from the manures. After 10 days, 81%

and 68% of the total BMP from WCB and GCS had been reached,

compared with 46% and 43% from the OCM and CCM respec-

tively. The SW reached a final BMP of 285 � 8 ml/g VS. The mix-

tures of manure and crop showed no differences in final BMP and

in methane production rates (Table 3). The calculated BMP values

for the mixtures (based on the BMP values for the single sub-

strates, and calculated as 50:50% contribution to the mixed sub-

strate) were in line with the experimental results of all

substrate mixtures except for OCM:WCB, which showed a calcu-

lated value for BMP of 242 ml/g VS (OCM + WCB), while reaching

272 � 21 ml/g VS in the test.

3.1.1 | Performance of CSTR reactors

The four CSTR reactors were stable and similar in gas production

in the initial start-up phase, before the start of Exp A. At the

start of Exp A, the reactors had received the substrates for two

retention times during an increase of OLRs to the final level of

2.8 g VS/L day. During the first retention time in the start-up

phase, R1 and R2 showed total VFA concentrations of around

3,000 mg/L. However, during the second retention time in the

start-up phase the total VFA concentration decreased to low

levels, that is <760 mg/L. The VFA concentrations during Exp A

were even lower (mean 110 mg/L). In the next experiment (Exp

B) after increasing the OLRs by addition of SW, the concentra-

tions were also low except for one measurement for R2 (day 153)

showing a total VFA concentration of 2,300 mg/L. The pH was

nearly the same in the reactors over the whole experimental per-

iod, resulting in a mean value for Exp A and Exp B of 7.4 � 0.1

and 7.7 � 0.1 respectively (Table 5).

There were significant (p < .01) differences in gas production

between the substrate mixes in Exp A, ranging from 6,857 to

TABLE 3 Biomethane potential (BMP) values of the differentsubstrate/substrate mixtures determined after incubation for

30 days and percentage of this potential reached after 10 and20 days. Standard deviations included

Substrate Day 10 (%) Day 20 (%) BMP day 30 (ml CH4/g VS)

WCB 81 89 262 � 62

GCS 68 88 235 � 35

OCM 46 84 222 � 11

CCM 43 89 253 � 27

OCM:WCB 56 89 272 � 21

OCM:GCS 57 94 272 � 3

CCM:WCB 55 91 262 � 31

CCM:GCS 58 92 290 � 17

SW 71 74 285 � 8

WCB, whole-crop barley silage; GCS, grass-clover silage; CCM, conven-

tional cattle manure; OCM, organic cattle manure; SW, soya bean meal/

wheat grain.

744 | AHLBERG-ELIASSON ET AL.

7,557 ml biogas per day (p < .01; Table 4). The WCB produced more

gas than the GCS for both manure types. The CCM produced more

gas than the OCM when combined with WCB, but no differences in

GP were found between the manure types when they were com-

bined with GCS. The SMP varied between 219 and 237 ml CH4/g

VS and showed a tendency for significance (p < .054) between the

substrates. The WCB tended to have higher SMP than GCS for both

manure types. Furthermore, the OCM had higher SMP than CCM,

when averaged across crops (p < .05). In Exp B, when the OLRs

were increased by approximately 1 g VS/L day by addition of SW,

all reactors showed an increase in gas production, with a range in

GP from 9,539 to 11,158 ml biogas per day and in SMP from 237

to 270 ml CH4/g VS. Both GP and SMP were higher for WCB than

for GCS when the crop was combined with OCM. However, when

the crop was combined with CCM, the WCB only had a higher GP

than the GCS (p < .001). Conventional cattle manure combined with

GCS (p < .001) produced more total gas than OCM:GCS, but no dif-

ferences in GP were found between the manure types when com-

bined with WCB. In terms of SMP, the OCM combined with WCB

had a significantly higher value than CCM:WCB, whereas no differ-

ences were found between the manure types when combined with

GCS (p < .001).

The methane content in the raw biogas in Exp A differed slightly

between the reactors in the order R1 > R2 > R3 > R4 (55% � 1.5%,

53% � 1.8%, 52% � 0.9% and 51% � 2.0% respectively). In Exp B,

the methane content was 54% � 0.9% in R1, R2 and R4 and

53% � 1.3% in R3. No significant differences were found between

the reactors. Measurements of hydrogen sulphide (H2S) in the raw

biogas showed values between 134 and 1,100 ppm, with a high

standard error of the mean (Table 4). Still, the levels varied some-

what over time and in general the reactors receiving CCM (R3, R4)

in the substrate mix, irrespective of the crop added as cosubstrate,

showed the highest values of H2S in both experiments (p < .05 and

p < .01 respectively). Moreover, the mean values of H2S in the raw

biogas increased on average by 13% in Exp B compared with Exp A

in reactors R1, R3 and R4. In contrast, the average content of H2S in

R2 in Exp B was 5% lower than in Exp A (Table 4).

3.2 | Digestibility and digestate composition

Contents of nutrients and minerals in the digestate from the four

CSTR differed between the reactors and between the experiments

(Table 5). In general, the level of all nutrients except Na, Mg and Fe

increased in the digestate between Exp A and Exp B. In Exp A, the con-

tent of NDF was highest in the reactors receiving WCB, with 413 g/

kg DM and 420 g/kg DM in R1 and R3 respectively. However, in Exp

B, the highest content of NDF was found for R2 (490 g/kg DM) and

R3 (483 g/kg DM). For protein, the highest concentrations during both

experiments were found in R2 and R4, receiving GCS in the mix. Total

ammonia content in Exp A was highest in R2 (29.5 g/kg DM) and low-

est in R1 (18.4 g/kg DM). In Exp B, the total ammonia content

increased compared with Exp A. However, the pattern shifted, and in

Exp B, the highest ammonia content was obtained in R3 (at 36.8 g/kg

DM) and the lowest content in R2 (at 33.2 g/kg DM; Table 5). The

content of Fe and Na in all reactors showed an average decrease by

26% and 4% in Exp B and in Exp A respectively. Concentrations of K,

P, Mg, S, Mn, Cu (except for R2) and Zn were higher in the digestate

from Exp B compared with Exp A. For Ca, the differences between the

experiments were small (Table 5).

The degree of degradation (DD), that is, the percentage of vola-

tile solids in the digestate compared with the substrate decreased in

the order R3 > R4 > R1 > R2. Values for Exp A were 55 � 2.1%,

52 � 1.9%, 47 � 1.4%, 44 � 1.2%, for R3, R4, R1 and R2 respec-

tively. After the SW addition, the DD increased in all reactors to val-

ues; 60 � 2.3%, 58 � 1.7%, 53 � 1.4%, 51 � 1.5%, for R3, R4, R1

and R2 respectively.

TABLE 4 Results† from the continuously operated reactors R1–R4‡ for experiments (Exp) A and B

Parameters Unit R1 R2 R3 R4 SEM p (M 3 C) p (M) p (C)

Exp A

GP ml biogas/day 7,276b 6,867c 7,557a 6,847c 136.5 <.01 <.05 <.001

SMP ml CH4/g VS 238(a) 230(b) 236(a,b) 219(c) 4.2 .054 <.05 <.001

H2S ppm 312(b) 385(b) 706(a) 812(a) 37.1 .501 <.05 .055

Exp B

GP ml biogas/day 11,158a 9,539c 10,788a 10,272b 227.6 <.001 .062 <.001

SMP ml CH4/g VS 270a 233c 246b 240b,c 5.3 <.001 <.05 <.001

H2S ppm 369(b) 377(b) 714(a) 955(a) 67.85 .086 <.01 .077

GP, gas production; SMP, specific methane potential; H2S, hydrogen sulphide; SEM, standard error of the mean; p (M 9 C), p - value for the interaction

between manure and crop; p (M), p - value for the main effect of manure averaged over crops; p (C), p -value for the main effect of crop averaged over

manure.†Least-square means with superscripts a–c differ significantly (p < .05), superscripts a–c within brackets indicate a tendency to significance

(.05 < p < .10).‡Substrate mixture in each reactor: R1, organic cattle manure and whole-crop barley silage; R2, organic cattle manure and grass-clover silage; R3, con-

ventional cattle manure and whole-crop barley silage; R4, conventional cattle manure and grass-clover silage. In Exp B, all reactors also received a mix

ture of soya bean meal and wheat grain.

AHLBERG-ELIASSON ET AL. | 745

Protein digestibility was calculated on mass balance basis, was

highest in R2 and R4 receiving the GCS in Exp A, whereas R1 and R3

(receiving WCB) had the highest protein digestibility in Exp B, where

SW was added. Digestibility of NDF in Exp A was highest in R3 and

R4. The NDF digestibility decreased between Exp A and Exp B in all

reactors, and this decrease was most pronounced in R2, with values

of 56% and 30% for Exp A and Exp B respectively (Figure A1).

3.3 | Bacterial communities

3.3.1 | Diversity indices

The microbial communities in the reactors, manure and inoculum were

analysed by illumina sequencing. After quality trim and chimera check,

3,564,730 sequences (from 10,946 to 209,787 per sample) remained

from a total of 55 triplicate total genomic DNA samples. The triplicate

samples were merged in silico and randomly subsampled based on the

detected lowest number of sequences in the sample. The number of

observed OTUs, according to the rarefaction curves, varied from 615 to

1,950, with the lowest values for the two different cattle manures

(CCM and OCM) and for the inoculum (Figure A2). The highest values

in all reactor samples (R1–R4) were found after 94 days of operation

(Table A2). In line with this, the lowest Chao1 value was found for the

two manure samples and for the inoculum, while the highest Chao 1

values were obtained for the reactor samples at day 105, that is at the

end of the first experiment (Exp A), except in sample R3, which reached

the highest value in sample R3221 in the second experiment (Table A2).

The sequencing coverage of the total bacterial community reached

73.1% to 92.6%, with the lowest and highest value for OCM and R294

respectively (Table A2). According to the Shannon and Simpson indices,

the inoculum and sample R2174 showed lower diversity and evenness

compared with the rest of the samples (Table A2) while the highest

value was seen in the all reactor samples for the first experimental per-

iod (Exp A). No significant difference in the Chao1, Shannon and Simp-

son indices was found after addition of SW (Exp B).

3.3.2 | Phylogenetic analysis across samples

Unweighted UniFrac principal coordinate analysis (PCoA) showed

that the community structures in the two different manures (CCM

and OCM) were similar to each other, but separated from the inocu-

lum and the reactor samples (Figure 1). A slight separation of the

communities was also observed between reactors receiving the same

crop but different manures in the first experiment (day 94 and 105).

The addition of SW affected the community in a similar way, that is,

all reactor samples slightly moved from the top left in the direction

of the bottom left in the diagram over time, and the four reactor

samples were still separated from each other (Figure 1).

TABLE 5 Contents of nutrients and micronutrients in the digestate obtained from the reactorsa R1–R4 in experiments (Exp) A and B

Parameters Units

Exp A Exp B

R1 R2 R3 R4 R1 R2 R3 R4

DM % of ww 5.8 5.8 4.7 4.9 6.4 6.6 5.5 5.7

VS % of ww 4.6 4.6 4.0 4.0 5.2 5.3 4.7 4.8

Protein g/kg DM 142 169 170 199 151 185 167 203

NDF g/kg DM 413 356 420 362 458 490 483 444

VOS % of VS 27 28 31 28 32 25 33 37

Tot-N g/kg DM 41.0 57.5 50.0 53.8 59.5 65.3 65.9 69.2

Am-N g/kg DM 18.4 29.5 22.4 23.8 34.5 33.2 36.8 34.9

Ca g/kg DM 19.9 26.0 12.8 19.7 17.5 22.8 11.3 17.3

K g/kg DM 45.1 54.7 34.2 43.3 48.9 53.8 34.7 42.1

P g/kg DM 6.8 7.1 7.6 7.4 7.7 7.4 7.4 8.0

Mg g/kg DM 6.1 7.0 7.8 8.4 6.2 6.5 6.9 7.9

Na g/kg DM 5.0 4.8 4.3 3.8 4.3 4.0 3.4 3.2

S g/kg DM 3.4 3.9 4.7 4.8 3.9 4.1 4.7 5.0

Cu mg/kg DM 29.8 33.9 30.2 31.1 28.5 29.2 26.9 28.7

Fe mg/kg DM 1,253 1,016 1,491 1,225 784 729 856 808

Mn mg/kg DM 277 291 236 238 278 272 211 223

Zn mg/kg DM 126 131 181 171 132 127 172 169

pH 7.3 7.3 7.7 7.6 7.4 7.4 7.7 7.7

Tot VFA g/L 0.11 0.09 0.09 0.05 0.19 0.06 0.85 0.09

DM, dry matter; VS, volatile solids; NDF, neutral detergent fibre; VOS, in vitro organic matter digestibility; Tot VFA, total volatile fatty acids.aSubstrate mixture in each reactor: R1, organic cattle manure and whole-crop barley silage; R2, organic cattle manure and grass-clover silage; R3, con-

ventional cattle manure and whole-crop barley silage; R4, conventional cattle manure and grass-clover silage. In Exp B all reactors also received a mix-

ture of soya bean meal and wheat grain.

746 | AHLBERG-ELIASSON ET AL.

Different bacterial community structures were observed in the

inoculum and the manures (Figure 2). The phyla Bacteroidetes, Firmi-

cutes and Actinobacteria dominated in the manure samples, with

relative abundance of 7.7% and 16.4%, 78.9% and 35.2%, 12.1% and

6.4% in OCM and CCM respectively. The phylum Proteobacteria also

represented a major fraction, particularly in CCM with a relative

abundance of 40.2%. The bacterial community in the inoculum was

mainly composed of the phyla Bacteroidetes (61.6%), Firmicutes

(22.0%) and WWE1 (11.9%; Figure 2).

After 94 days of operation with the four different substrate mix-

tures, the bacterial communities in all reactors slightly shifted from

the original community of the inoculum to a similar pattern at phy-

lum level (Figure 2). The relative abundance of Bacteroidetes

decreased slightly from 61.6% to 53.2 � 5.2% (mean value of R1–

R4 on day 94), while that of Firmicutes increased on average from

22% to 36.5 � 3.5%. Moreover, the Actinobacteria also increased in

relative abundance, from 0.3% (inoculum) to 2.6 � 1.0%. At deeper

phylogenetic level, the four reactor samples still shared similar bacte-

rial communities, which were dominated by the classes Bacteroidia

and Clostridia, with relative abundance ranging from 46.1% (R394) to

57.6% (R1105) and from 17.6% (R3105) to 26.5% (R2105) respectively

(Figure 3). Only small differences were seen between the four sam-

ples. For example, the class Coriobacteria (belonging to the Acti-

nobacteria) was mainly detected in the samples taken from R1 and

R3 (operated with WCB) but still at low abundance (1.5% [R1105] to

3.7% [R394]) compared with R2 and R4 (<0.8%). In contrast, the class

Actinobacteria (phylum Actinobacteria) showed slightly higher

F IGURE 1 Phylogenic distance of the microbial community insamples as determined by unweighted UniFrac principal coordinate anal-ysis. R1–R4, reactors; OCM, organic cattle manure; CCM, conventional

cattle manure. (R2174 and R2221 overlapped with each other perfectly)[Colour figure can be viewed at wileyonlinelibrary.com]

F IGURE 2 Relative abundance of the Bacterial and Archaeal community (Phyla level) obtained by illumina sequencing. R1–R4, reactors; OCM,organic cattle manure; CCM, conventional cattle manure; 94–211, days of operation [Colour figure can be viewed at wileyonlinelibrary.com]

AHLBERG-ELIASSON ET AL. | 747

relative abundance in R2 and R4 (1.2% [R494] to 3.0% [R2105]) com-

pared with R1 and R3 (<0.6%; Figure 3).

When SW was added, an unidentified genus belonging to the

order Bacteroidales increased to different extents in all reactors (R1:

from 25.8% to 35.8%; R2: from 33.4% to 59.4%; R3: from 27.4% to

30.4%; R4: 23.0% to 43.0%; data not shown). In contrast, the genus

BF311 (order Bacteroidales, phylum Bacteroidetes) decreased in all

four reactors, especially in R3 and R4, from 10.6 to 1.3% and from

11.2 to 0.7% respectively. In addition, reactors receiving the same

crop, but operated with different manures, showed different

responses to SW addition. For example, the relative abundance of

the Bacteroidetes increased slightly, while the relative abundance of

the Firmicutes decreased in R2 and R4 (operating with GCS). How-

ever, in R1 and R3 (operating with WCB), this change was the oppo-

site. (Figure 3). In R2 and R4, the increase in the Bacteroidetes was

mainly represented by an increase in the class Bacteroidia from

48.5% to 59.3% (R2) and from 53.4% to 58.7% (R4). The decrease in

the Firmicutes was mainly caused by a decrease in the class Clostri-

dia from 33.1% to 19.6% (R2) and from 29.5% to 17.2% (R4; Fig-

ure 3). In addition, within the Actinobacteria, the relative abundance

of the classes Coriobacteria and Actinobacteria decreased and

increased, respectively, especially in R1 and R3 (Figure 3).

Within the domain Archaea, the phylum Euryarchaeota domi-

nated, but still at very low relative abundance (in total representing

~0.14%) compared with the bacterial community. On genus level,

Methanobacterium was mainly detected in the original inoculum,

while Methanobrevibacter dominated in the manure samples. After

operation with the different crops and manure mixtures, the genus

Methanosarcina became slightly more abundant in R1 and R2, while

the genus Methanobacterium was more abundant in R3 and R4.

Moreover, no clear difference in the reactors was seen based on SW

addition (Figure A3).

4 | DISCUSSION

4.1 | Effects of substrate origin on gas productionand degradation

The BMP values obtained for GCS in the batch test were in line with

previous studies. Giss�en et al. (2014) reported values of 271 and

327 ml/g VS for second and first cut of grass-clover forage, respec-

tively, and Lehtomaki (2006) showed a value of 306 ml/g VS for

grass silage. These similar BMP values might depend on that the for-

ages were harvested at similar maturity stages as the concentrations

of hemicellulose, cellulose and lignin increase while the concentra-

tion of protein decreases with advancing maturity stages (McEniry &

O’Kiely, 2013; Van Soest, 1994). For WCB, not many studies are

available but Herrmann, Idler, and Heiermann (2016) presented val-

ues of 355 ml/g VS for WCB at very early maturity stages before

heading. One possible explanation for the lower BMP value for

WCB in this study compared with the previous study is that the

WCB used in this study had high field losses of kernels during

F IGURE 3 Relative abundance of the bacterial community (Class level) obtained by illumina sequencing. R1–R4, reactors; OCM, organiccattle manure; CCM, conventional cattle manure; 94–211, days of operation [Colour figure can be viewed at wileyonlinelibrary.com]

748 | AHLBERG-ELIASSON ET AL.

mowing and baling at harvest, resulting in a lower starch content in

WCB compared with direct-cut barley silage (Nadeau, 2007). The

differences in BMP between our study compared with others can

also be related to differences in maturity stages at harvest of the

WCB as the changes in sugars, starch and NDF digestibility at

advancing maturity affect BMP (Rustas, Bertilsson, Martinsson, Elver-

stedt, & Nadeau, 2011). In this study, WCB silage resulted in faster

degradation compared with GCS at the early stage of the batch pro-

cess; however at the end of the BMP test, no differences between

the crops could be proven. Negative correlations exist between BMP

and the lignin content in the crop, and between the lignin content

and the degradation rate (Dandikas, Heuwinkel, Lichti, Drewes, &

Koch, 2015; Herrmann, Prochnow, Heiermann, & Idler, 2014; Triolo,

Sommer, Møller, Weisbjerg, & Jiang, 2011). In this study, WCB had a

higher content of rapidly degradable carbohydrates, such as starch

and sugar, and a lower content of lignified fibre, which most likely

explains the higher initial degradation rate compared with GCS in

the BMP tests (Mertens, 2007).

The methane potential for the manures used in this study was in

the upper part of the range reported in other studies (148–265 ml/g

VS; Masse, Jarret, Benchaar, & Saady, 2014; Moller, Sommer, & Ahr-

ing, 2004; Ruile, Schmitz, Moench-Tegeder, & Oechsner, 2015; Triolo

et al., 2011). Difference in values between different studies is proba-

bly a consequence of different feed composition and intake and

digestibility in the cow (Amon et al., 2007; Van Soest, 1994). Dairy

cows in organic and conventional production systems receive different

feedstuffs, which can have an impact on the final composition of the

manure, which in turn can affect biogas production (Masse et al.,

2014; Moller, Sarker, Hellwin, & Weisbjerg, 2012). In this study, CCM

showed a slightly higher final BMP value, that is 253 ml/g VS com-

pared with 222 ml/g VS for OCM. This difference is probably

explained by the higher content of protein in CCM and the higher ADL

content in OCM (Table 2). Comparing the feed rations fed to dairy

cows in these two production systems, conventional dairy cows are

fed higher contents of protein and crude fat, but lower contents of

NDF in the ration (Table A3). These differences in nutrient concentra-

tions of their rations between conventional and organic production

partly depend on differences in dietary forage DM proportions, where

dairy cows in conventional production have a lower forage proportion

in the ration (approx. 50%) compared with cows in organic production

(approx. 60%), resulting in a higher passage rate in the rumen of cows

in conventional production. Increased passage rate in the rumen short-

ens the retention time for the digesta in the rumen, resulting in more

digestible protein and fibre in the manure, which is in line with the

composition of CCM and OCM (Mgbeahuruike et al., 2016). These dif-

ferences in degradable compounds may partly explain the higher BMP

value for CCM compared with OCM. Comparing the BMP results of

the manure/crop mixtures with the values of the single substrates, it

appears to be no synergistic effects (just additive effects) of codiges-

tion in the batch tests (Table 3). This result is in line with previous

findings and illustrates that the effect of codigestion seems to be diffi-

cult to evaluate by batch tests (Belle, Lansing, Mulbry, & Weil, 2015;

Sondergaard et al., 2015). However, some synergistic effects have

been demonstrated previously for wheat straw and meadow grass in

different combinations with manure in thermophilic batch experiments

(Awais, Alvarado-Morales, Tsapekos, Gulfraz, & Angelidaki, 2016).

For the CSTR systems evaluating the combined substrate mixes

in R1–R4, the highest biogas yield and SMP were obtained in R1

and R3 receiving WCB in both experiments. This result can be attrib-

uted to the higher content of easily degradable starch and water-

soluble carbohydrates but lower content of lignified fibre in WCB

compared with GCS (Table 1). When the SW was added in Exp B,

the average daily gas production increased as a consequence of the

increase in SMP and OLRs. The SW contributed more digestible pro-

tein and starch compared with the substrates used in Exp A, explain-

ing the increase in SMP. In Exp B, total GP and SMP increased by

39%–53% and 1%–13%, respectively, compared with Exp A.

The observed degree of degradation (DD) was in line with our

previous findings for Swedish farm-scale biogas plants using codiges-

tion with manure and crops, that is approximately 50% (Ahlberg-

Eliasson, Nadeau, Lev�en, & Schn€urer, 2017). Degree of degradation

was highest in R3 and R4, receiving CCM. The explanation is proba-

bly the same as for the differences in BMP between OCM and

CCM, that is the feedstuff given to the cow and the passage rate in

the rumen affect the manure quality, that is the fibre content varies

for different manures (Amon et al., 2007; Mgbeahuruike et al., 2016;

Triolo et al., 2011). Interestingly, the DD increased, on average, by

6% in Exp B compared with Exp A for all reactors. These results

agree with previous findings of increased VS removal with increasing

OLRs during codigestion of manure and grass silage (Lehtomaki,

2006). In this study, the increased digestibility was probably caused

by a higher level of rapidly degradable compounds in SW, such as

protein and starch.

The final economic output from the farm-scale biogas system

depends on several factors, such as choice of technology and diges-

tate value, both depending on the substrate used (Lantz, 2012; Ruile

et al., 2015). Gas production depends to a great extent on the

amount of energy crops in the substrate mix, together with the pre-

viously discussed effect of chemical composition (Ebner, Labatut,

Lodge, Williamson, & Trabold, 2016; Herrmann et al., 2016; Seppala,

Pyykkonen, Vaisanen, & Rintala, 2013; Triolo et al., 2011). A fast

degradation rate will allow the process to reach the biogas potential

even at relatively short HRT, potentially allowing operation at high

organic load of the crop (Ruile et al., 2015). In general, it is important

to match the chemical composition the substrate with the technique

used at the biogas plant, in order to increase the degradation rate of

VS. If VS is well degraded, the risk of methane leakages from storage

will decrease, provided an accurate technique is used (Clemens,

Trimborn, Weiland, & Amon, 2006; Insam et al., 2015; Moller et al.,

2012; Ruile et al., 2015).

4.2 | Gas quality

The methane content in the biogas in Exp A was approximately

54%, a common proportion when energy crops are used as a main

substrate (Weiland, 2010). Herrmann et al. (2016) found methane

AHLBERG-ELIASSON ET AL. | 749

content to be approx. 55% for barley silage and approx. 59% for

lucerne silage in batch tests. In Exp B, the methane content

increased somewhat compared with Exp A, probably as a conse-

quence of the higher protein content in the substrate in Exp B. A

higher protein content in the substrate should theoretically increase

the level of methane in the biogas produced if the protein is degrad-

able in the process (Weiland, 2010). In this case, the digestibility of

protein increased, on average, by 25%, in Exp B compared with Exp

A (Figure A1), which may explain the tendency for higher methane

content in Exp B.

The H2S content was within the range reported previously (100–

3,000 ppm; Weiland, 2010). The H2S content was low in Exp A, but

increased slightly in Exp B, when the protein content and digestibil-

ity increased for all substrate mixes, which most likely explains the

higher H2S concentration. The highest concentrations of H2S were

seen in R3 and R4, receiving CCM, and this can be explained by the

higher level of S in CCM compared with OCM (Table 2). Any H2S in

the raw biogas is unwanted, as it can cause corrosion problems in

pipes and combustion devices in the biogas plant (Rasi, Lantela, &

Rintala, 2011). It can also inhibit microbes in the reactor, directly or

indirectly, by precipitation of metals important for enzymatic activity

(Chen, Cheng, & Creamer, 2008; Thanh, Ketheesan, Yan, & Stuckey,

2016). Iron is often added to the substrate mix or directly to the

reactor in order to reduce the S (Ryckebosch, Drouillon, & Ver-

vaeren, 2011). However in this study, no extra Fe was used. Com-

paring the concentrations of Fe and S in the digestate (Table 5), the

Fe content decreased and the S content increased between Exp A

and Exp B, reflecting the increased H2S in the raw gas during Exp B.

The Fe concentration was apparently not sufficient to trap the H2S

produced.

4.3 | Nutrient composition

Concentrations of nutrients (NPKS) in the digestate are highly impor-

tant for farmers and are a central part of the investment in a biogas

system, with the digestate being especially valuable in organic farm-

ing systems (Antonio Alburquerque et al., 2012; Novak & Fiorelli,

2010). In this regard, the content of nutrients in the digestate is of

specific interest in a more nutrient-efficient future agricultural sys-

tem (Vaneeckhaute, Meers, Michels, Buysse, & Tack, 2013). Several

studies on digestates as fertilizers have been carried out and

reported in recent years (reviewed in Insam et al., 2015). All in all,

these studies illustrate that digestates function well as fertilizers and

even can give similar yields as mineral fertilizers. Moreover, diges-

tates have been shown to improve soil structure through the addi-

tion of organic matter. The actual need of nutrients depends on the

plant but also the soil type (Insam et al., 2015; M€oller, 2015). The

mineralization of nitrogen to ammonium (N-NHþ4 ) resulted in

increased ammonium contents in the digestate for Exp B compared

with Exp A. The average ammonium content in the two experiments

was in line with earlier research evaluating different proportions of

manure and straw (range 1.4–2.0 kg N-NHþ4 /ton ww; Risberg, Sun,

Leven, Horn, & Schnurer, 2013), and from codigestion plants with

cattle manure and energy crops (2.2 kg N-NHþ4 /ton ww; Ahlberg-

Eliasson et al., 2017). In the substrate, the content of total N

increased between Exp A and Exp B by, on average, 1.4 kg/ton ww

(Table A1). This increase is explained by the increased content of

proteins brought about by supplementation with SW. In Exp B, the

digestibility of protein also increased in all reactors compared with

Exp A, as described in section 4.1. The increase in protein digestibil-

ity was larger in R1 and R3, compared with R2 and R4, which could

be explained by less concentration of crude protein in WCB com-

pared with GCS. Therefore, the added protein in SW resulted in a

greater availability in R1 and R3 compared with R2 and R4. This

increased the N-NHþ4 concentration by on average 0.8 kg/ton ww in

Exp B compared with Exp A (Table 5). High nitrogen concentration

is important for the value of the digestate as a fertilizer. However, a

high level of ammonium, in equilibrium with ammonia, also repre-

sents a risk of process inhibition, typically occurring at levels >2 kg

N-NHþ4 /ton ww, as well as ammonia emissions during storage and

handling of the digestate (Clemens et al., 2006; Rajagopal, Masse, &

Singh, 2013).

The K concentration in the digestate was in the upper part of

the range (19 to 43 g/kg DM) reported in a review by M€oller and

M€uller (2012). The K concentration was highest in the substrate and

in the digestate produced from reactors R1 and R2, receiving OCM.

The K concentration in the OCM (44.2 g/kg DM) was almost twice

that in the CCM (22.5 g/kg DM). We have no clear explanation for

this difference between manure types as a previous study has shown

only minor or no difference in potassium concentration between

legumes and grasses (Linse, Dahlin, Nadeau, Forkman, & €Oborn,

2015). Most of the other nutrients, except Fe and Na, increased in

the digestate between Exp A and Exp B, owing to the higher content

of nutrients in the SW. The P content in the samples was around 7–

8 g/kg DM, which was lower than found in German full-scale codi-

gestion plants (6–17 g/kg DM; M€oller & M€uller, 2012). Most of the

P was related to the composition of manure, but increased in Exp B

when SW was added to the substrate. For Swedish conditions, there

is a limitation of maximum annual application of 22 kg P/ha. The

digestate evaluated in this project had phosphorus concentrations

below this level (Swedish Board of Agriculture, 2017).

4.4 | Effects on the microbiological community

The communities in the two manures were similar, but clearly sepa-

rated from the reactors and the original inoculum, as illustrated in

the PCoA plot (Figure 2). The separation was mainly caused by the

lower relative abundance of the phylum Bacteroidetes in manure

communities compared with the reactor and inoculum communities.

Differences between the reactors and the manure illustrate that the

community present in the manure did not establish in the reactors

to any great extent. Similar results have been reported in a previous

study where cattle manure was used during codigestion with manure

(Sun, Pope, Eijsink, & Schn€urer, 2015). This suggests that variations

in community structure and performance between the reactors were

caused by differences in the chemical composition of the substrate,

750 | AHLBERG-ELIASSON ET AL.

rather than by a difference in the manure microbial community com-

position. Using WCB and GCS in codigestion with manure in the

Exp A had little impact on the development of the communities (i.e.

the communities were still similar to the original inoculum), and only

small differences were seen between the reactors. However, com-

pared with both the original manure and the inoculum, a clear

increase in microbial diversity was seen after operating the reactor

with the manure–crop mixture for 94 days. High microbial diversity

has been suggested to correlate with high functionality (De Francisci,

Kougias, Treu, Campanaro, & Angelidaki, 2015; M€uller et al., 2016).

Irrespective of manure/crop, dominance of the phyla Firmicutes and

Bacteroidetes was seen in all samples (Figure 3). These two phyla

are frequently reported to be highly abundant in different anaerobic

biogas reactors operating with various substrates (Azman, Khadem,

Lier, Zeeman, & Plugge, 2015; Sun, Liu, M€uller, & Schn€urer, 2016;

Sundberg et al., 2013). Their high abundance is most likely related to

the high metabolic capacity of members of these phyla. The relative

ratio of these two phyla has been suggested as a potential indicator

for the performance of biogas production processes, with higher

abundance of Firmicutes relative to Bacteroidetes being correlated

with higher methane yield (Chen et al., 2016). This was confirmed in

this study, where R1 and R3 produced more methane than R2 and

R4. Thus, the higher yield in reactors R1 and R3 might have both a

chemical and microbiological explanation. On class level, only small

differences were seen between the four reactors, with Coriobacteria

and Actinobacteria (both belonging to the phylum Actinobacteria)

being present at higher abundance in R1/R3 (WCB) and R2/R4

(GCS) respectively (Figure 3). This difference might be explained by

differences in the nutrient composition between the WCB and GCS

used. The higher ADF content in R2 and R4 (GCS) could explain the

generally higher relative abundance of the class Actinobacteria, as its

members are reported to be able to degrade cellulose and produce

lignin-degrading enzymes (Saini, Aggarwal, Sharma, & Yadav, 2015).

Addition of SW resulted in similar response in the different reac-

tors. The major change was an increase in the abundance of an

unidentified genus in the order Bacteroidales, as well as an increase

and decrease in the class Clostridia in reactors operated with WCB

(R1 and R3) and GCS (R2 and R4) respectively (Figure 3). Chemical

analysis of the digestate showed that the protein and NDF contents

of R2 and R4 were higher than that of R1 and R3 in Exp B, which

implies lower hydrolysis efficiency in R2 and R4 (Table 5). This might

be explained by the higher relative abundance changes in the class

Clostridia in R1 and R3, members of which can hydrolyse various

organic substrate such as lignocellulose, carbohydrate and proteins

(Carere, Sparling, Cicek, & Levin, 2008; Matsushita & Okabe, 2001).

Soya bean meal and wheat grain (SW) addition had no clear

effect on microbial diversity in the reactors, except for R2174, which

showed relatively low diversity compared with other reactor sam-

ples. Interestingly, R2 also showed the lowest degree of degradation

and total biogas production (Table 4) compared with other reactors,

which partly might suggest a relationship between microbial diversity

and DD.

The relative abundance of Archaea compared with Bacteria was

low in all the investigated reactors, a result in line with previous

studies of manure-based biogas processes (Liu, Sun, M€uller, &

Schn€urer, 2017; Sun et al., 2015). As also previously observed, the

genus Methanobrevibacter dominated the manure but not the biogas

processes (Sun et al., 2015) illustrating again that the community

present in the manure did not establish in the reactors to any great

extent. Still, different methanogenic genus dominated in reactors

operated with different manures, for example Methanosarcina (OCM)

or Methanobacterium (CCM), likely a consequence of different chemi-

cal composition of the manures. The genus Methanobacterium com-

prises hydrogenotrophic species while Methanosarcina is more

versatile and can use both hydrogen/carbon dioxide, acetate and

methanol as substrates for growth. Both genera are commonly found

in various biogas processes, including manure-based systems, but

Methanosarcina has been highlighted as a very robust methanogen

due to its ability to withstand common stress factors in biogas pro-

cesses (De Vrieze, Hennebel, Boon, & Verstraete, 2012).

5 | CONCLUSIONS

The chemical composition of the substrates used in this study, both

manure and forages, was shown to influence gas production, gas

quality, nutrient concentrations in the digestate and the microbial

community. Substrate combinations with a higher level of proteins

and degradable fibres resulted in faster degradation and higher final

methane yields. For the crops, the contents of water-soluble carbo-

hydrates, starch and digestible fibre had a clear effect on gas pro-

duction, resulting in higher gas production from WCB than from

GCS. Addition of more starch and easily degradable proteins (soya

bean-wheat) resulted in increased gas production and degree of

degradation. As expected based on feed rations, the manure origi-

nating from conventional dairy production contained a higher frac-

tion of easily available fibre and proteins, resulting in higher gas

yield for conventional compared with organic manure. Interestingly,

the reactors with the highest gas yields also showed the highest

ratio between Firmicutes and Bacteroidetes, which may be corre-

lated with high methane production. Moreover, an increase in the

OLRs by addition of SW improved the degree of degradation, but to

varying extents depending on the other substrates used in codiges-

tion. A drawback with the improved degradation and yield caused by

the higher protein content was an increase in H2S concentration in

the raw biogas. However, the results suggested that addition of a

crop with a high iron content could decrease concentration of H2S.

The content of ammonia in the digestate, which is of great impor-

tance for farmers investing in a biogas plant, was also shown to be

related to the protein content in the substrate and degradation of

proteins in the biogas process. In the long run, energy crops grown

with low life cycle costs have an advantage over more environmen-

tally demanding biomass production. For farm-scale biogas produc-

tion, using codigestion with manure and WCB or GCS can be one

AHLBERG-ELIASSON ET AL. | 751

promising way to improve the gas production as well as the nutrient

content in the digestate.

ACKNOWLEDGMENTS

This study was funded by the Swedish Energy Agency, as part of

the project “Effects of codigestion in biogas production—energy

crops and cattle manure” (project no. 36474-1), V€astra G€otaland

County Council and the Chinese Scholarship Council (CSC). Thanks

to the farmers Torbj€orn Holgersson, Jonas Holgersson and Carl

Trawniczek and to colleagues at the R�adde Experimental Farm; Ste-

fan Wallin, Caroline Dahr�en, Dan Claesson and Fredrik Andersson.

Colleagues at the biogas group at SLU, Uppsala; Li Sun, Bettina

M€uller, Simon Isaksson and Robin Hagblom, are acknowledged.

Thanks to Dr. Jan-Eric Englund at the Department of Biosystems

and Technology, SLU Alnarp, for statistical advice.

ORCID

K.Ahlberg-Eliasson http://orcid.org/0000-0002-8351-6349

REFERENCES

Ahlberg-Eliasson, K., Nadeau, E., Lev�en, L., & Schn€urer, A. (2017). Produc-

tion efficiency of Swedish farm-scale biogas plants. Biomass and

Bioenergy, 97, 27–37. https://doi.org/doi: 10.1016/j.biombioe.2016.

12.002

Aichinger, P., Wadhawan, T., Kuprian, M., Higgins, M., Ebner, C., Fimml,

C., & Wett, B. (2015). Synergistic co-digestion of solid-organic-waste

and municipal-sewage-sludge: 1 plus 1 equals more than 2 in terms

of biogas production and solids reduction. Water Research, 87, 416–

423. https://doi.org/10.1016/j.watres.2015.07.033�Akerlind, M., Weisbjerg, M. R., Eriksson, T., Togersen, R., Ud�en, P., �Olafs-

son, O. M., & Volden, H. (2011). Feed analyses and digestion meth-

ods. In H. Volden (Ed.), NorFor-The Nordic feed evaluation system:

EAAP Publication no 130 (pp. 41–54). Wageningen Academic Pub-

lisher, The Netherlands.

Aman, P., & Hesselman, K. (1984). Analysis of starch and other main con-

stituents of cereal. Swedish Journal of Agricultural Research, 14(3),

135–139.

Amon, T., Amon, B., Kryvoruchko, V., Zollitsch, W., Mayer, K., & Gruber,

L. (2007). Biogas production from maize and dairy cattle manure -

Influence of biomass composition on the methane yield. Agriculture

Ecosystems & Environment, 118(1–4), 173–182. https://doi.org/10.

1016/j.agee.2006.05.007

Antonio Alburquerque, J., de la Fuente, C., Ferrer-Costa, A., Carrasco, L.,

Cegarra, J., Abad, M., & Pilar Bernal, M. (2012). Assessment of the

fertiliser potential of digestates from farm and agroindustrial residues.

Biomass & Bioenergy, 40, 181–189. https://doi.org/10.1016/j.biomb

ioe.2012.02.018

Awais, M., Alvarado-Morales, M., Tsapekos, P., Gulfraz, M., & Angelidaki,

I. (2016). Methane production and kinetic modeling for co-digestion

of manure with lignocellulosic residues. Energy & Fuels, 30(12),

10516–10523. https://doi.org/10.1021/acs.energyfuels.6b02105

Azman, S., Khadem, A. F., Lier, J. B., Zeeman, G., & Plugge, C. M. (2015).

Presence and role of anaerobic hydrolytic microbes in conversion of

lignocellulosic biomass for biogas production. Critical Reviews in Envi-

ronmental Science and Technology, 45, https://doi.org/10.1080/

10643389.2015.1053727

Balsberg-P�ahlsson, A. (1990). F€orbehandling, uppslutning och extraktbered-

ning av v€axt-och f€ornaprov. Handledning i Kemiska Metoder vid

V€axtekologiska Arbeten (pp. 18–19). Lund, Sweden: Meddelanden fr�an

V€axtekologiska Avdelningen, Lunds Universitet.

Bauer, A., Leonhartsberger, C., Boesch, P., Amon, B., Friedl, A., & Amon,

T. (2010). Analysis of methane yields from energy crops and agricul-

tural by-products and estimation of energy potential from sustainable

crop rotation systems in EU-27. Clean Technologies and Environmental

Policy, 12(2), 153–161. https://doi.org/10.1007/s10098-009-0236-1

Belle, A. J., Lansing, S., Mulbry, W., & Weil, R. R. (2015). Methane and

hydrogen sulfide production during co-digestion of forage radish and

dairy manure. Biomass and Bioenergy, 80, 44–51. https://doi.org/10.

1016/j.biombioe.2015.04.029

Borjesson, P., Prade, T., Lantz, M., & Bjornsson, L. (2015). Energy crop-

based biogas as vehicle fuel-the impact of crop selection on energy

efficiency and greenhouse gas performance. Energies, 8(6), 6033–

6058. https://doi.org/10.3390/en8066033

Carere, C. R., Sparling, R., Cicek, N., & Levin, D. B. (2008). Third genera-

tion biofuels via direct cellulose fermentation. International Journal of

Molecular Sciences, 9(7), 1342–1360.

Chen, X., Chen, Z., Wang, X., Huo, C., Hu, Z., Xiao, B., & Hu, M. (2016).

Application of ADM1 for modeling of biogas production from anaero-

bic digestion of Hydrilla verticillata. Bioresource Technology, 211, 101–

107. https://doi.org/10.1016/j.biortech.2016.03.002

Chen, Y., Cheng, J. J., & Creamer, K. S. (2008). Inhibition of anaerobic

digestion process: A review. Bioresource Technology, 99, 4044–4064.

https://doi.org/10.1016/j.biortech.2007.01.057

Clemens, J., Trimborn, M., Weiland, P., & Amon, B. (2006). Mitigation of

greenhouse gas emissions by anaerobic digestion of cattle slurry.

Agriculture Ecosystems & Environment, 112(2–3), 171–177. https://doi.

org/10.1016/j.agee.2005.08.016

Dandikas, V., Heuwinkel, H., Lichti, F., Drewes, J. E., & Koch, K. (2015).

Correlation between biogas yield and chemical composition of grass-

land plant species. Energy & Fuels, 29(11), 7221–7229. https://doi.

org/10.1021/acs.energyfuels.5b01257

De Francisci, D., Kougias, P. G., Treu, L., Campanaro, S., & Angelidaki,

I. (2015). Microbial diversity and dynamicity of biogas reactors

due to radical changes of feedstock composition. Bioresource

Technology, 176, 56–64. https://doi.org/10.1016/j.biortech.2014.

10.126

De Vrieze, J., Hennebel, T., Boon, N., & Verstraete, W. (2012). Methano-

sarcina: The rediscovered methanogen for heavy duty biomethana-

tion. Bioresource Technology, 112, 1–9. https://doi.org/10.1016/j.b

iortech.2012.02.079

Ebner, J. H., Labatut, R. A., Lodge, J. S., Williamson, A. A., & Trabold, T.

A. (2016). Anaerobic co-digestion of commercial food waste and

dairy manure: Characterizing biochemical parameters and synergistic

effects. Waste Management, 52, 286–294. https://doi.org/10.1016/j.

wasman.2016.03.046

Ericson, B., & Andr�e, J. (2010). HPLC-applications for agricultural and ani-

mal science. Proceedings from the 1st.

European Comission (2016a). Clean Energy For All Europeans. Communi-

cation from the Commission to the European Parliament, the Council,

the European Economic and Social Committee, the Committee of the

Regions and the European Investment Bank. Retrieved from https://

ec.europa.eu/energy/en/news/commission-proposes-new-rules-consu

mer-centred-clean-energy-transition 2016-12-02: Brussels, 30.11.

2016.

European Comission (2016b). Optimal use of biogas from waste streams -

An assessment of the potential of biogas from digestion in the EU

beyond 2020. Retrieved from https://ec.europa.eu/energy/sites/ener/f

iles/documents/ce_delft_3g84_biogas_beyond_2020_final_report.pdf.

FAO (2014a). Building a Common Vision for Sustainable Food and Agricul-

ture. Principles and Approaches. Retrieved from http://www.ieep.eu/as

sets/1410/Circular_economy_scoping_study_-_Final_report.pdf.

752 | AHLBERG-ELIASSON ET AL.

FAO (2014b). The Water-Energy-Food Nexus: A new approach in sup-

port of food security and sustainable agriculture. Retrieved from

http://www.fao.org/documents/card/en/c/182bf36b-87fa-4ea5-b898-

06c89c88f241/:

Giss�en, C., Prade, T., Kreuger, E., Nges, I. A., Rosenqvist, H., Svensson, S.-

E., & Bj€ornsson, L. (2014). Comparing energy crops for biogas pro-

duction – Yields, energy input and costs in cultivation using digestate

and mineral fertilisation. Biomass and Bioenergy, 64, 199–210.

https://doi.org/doi: 10.1016/j.biombioe.2014.03.061

Heiermann, M., Ploechl, M., Linke, B., Schelle, H., & Herrmann, C. (2009).

Biogas crops - Part I: Specifications and suitability of field crops for

anaerobic digestion. Agricultural Engineering International, 11, Manu-

script 1087.

Heinrichs, J., & Kononoff, P. (2002). Evaluating particle size of forages

and TMRs using the new Penn State Forages Particle Separator.

Retrieved from http://www.das.psu.edu/dcn/catforg/particle

Herrmann, C., Idler, C., & Heiermann, M. (2016). Biogas crops grown in

energy crop rotations: Linking chemical composition and methane

production characteristics. Bioresource Technology, 206, 23–35.

https://doi.org/doi: 10.1016/j.biortech.2016.01.058

Herrmann, C., Prochnow, A., Heiermann, M., & Idler, C. (2014). Biomass

from landscape management of grassland used for biogas production:

Effects of harvest date and silage additives on feedstock quality and

methane yield. Grass and Forage Science, 69(4), 549–566. https://doi.

org/10.1111/gfs.12086

Holm-Nielsen, J. B., Al Seadi, T., & Oleskowicz-Popiel, P. (2009). The

future of anaerobic digestion and biogas utilization. Bioresource Tech-

nology, 100(22), 5478–5484. https://doi.org/10.1016/j.biortech.2008.

12.046

Hugerth, L. W., Wefer, H. A., Lundin, S., Jakobsson, H. E., Lindberg, M.,

Rodin, S., & Andersson, A. F. (2014). DegePrime, a program for

degenerate primer design for broad-taxonomic-range PCR in micro-

bial ecology studies. Applied and Environmental Microbiology, 80(16),

5116–5123. https://doi.org/10.1128/AEM.01403-14

Insam, H., G�omez-Brand�on, M., & Ascher, J. (2015). Manure-based biogas

fermentation residues – Friend or foe of soil fertility? Soil Biology and

Biochemistry, 84, 1–14. https://doi.org/10.1016/j.soilbio.2015.02.

006

Insam, H., & Markt, R. (2016). Comment on “Synergistic co-digestion of

solid-organic-waste and municipal-sewage-sludge: 1 plus 1 equals

more than 2 in terms of biogas production and solids reduction”.Water Research, 87, 416–423. https://doi.org/10.1016/j.watres.2016.

02.053

Lantz, M. (2012). The economic performance of combined heat and

power from biogas produced from manure in Sweden - A comparison

of different CHP technologies. Applied Energy, 98, 502–511. https://d

oi.org/10.1016/j.apenergy.2012.04.015

Larsson, K., & Bengtsson, S. (1983). Determination of Nonstructural Car-

bohydrates in Plant Material, Method Description No. 22. National

Laboratory of Agricultural Chemistry, Uppsala, Sweden.

Lehtomaki, A. (2006). Biogas production from energy crops and crops resi-

dues. Jyv€askyl€a, Finland: University of Jyv€askyl€a.

Lehtomaki, A., Huttunen, S., & Rintala, J. A. (2007). Laboratory investiga-

tions on co-digestion of energy crops and crop residues with cow

manure for methane production: Effect of crop to manure ratio.

Resources Conservation and Recycling, 51(3), 591–609. https://doi.org/

10.1016/j.resonrec.2006.11.004

Lindgren, E. (1979). The nutritional value of roughages determined

in vivo and by laboratory methods. Report, 45, 63.

Lindgren, E. (1983). Nykalibrering av VOS-metoden f€or best€amning av ener-

giv€arde hos vallfoder. Uppsala, Sweden: Department of Animal Nutri-

tion and Management, Swedish University of Agricultural Sciences.

Linse, L., Dahlin, A. S., Nadeau, E., Forkman, J., & €Oborn, I. (2015). Boron

fertilisation of organically managed grass-clover swards on coarse-

textured soils: Effects on botanical and element composition. Agricul-

tural and Food Science, 24(4), 261–272.

Liu, T., Sun, L., M€uller, B., & Schn€urer, A. (2017). Importance of inoculum

source and initial community structure for biogas production from

agricultural substrates. Bioresource Technology, 245(Part A), 768–777.

https://doi.org/10.1016/j.biortech.2017.08.213

Luscher, A., Mueller-Harvey, I., Soussana, J. F., Rees, R. M., & Peyraud, J.

L. (2014). Potential of legume-based grassland-livestock systems in

Europe: A review. Grass and Forage Science, 69(2), 206–228. https://d

oi.org/10.1111/gfs.12124

Masse, D. I., Jarret, G., Benchaar, C., & Saady, N. M. C. (2014). Effect of

corn dried distiller grains with solubles (DDGS) in dairy cow diets on

manure bioenergy production potential. Animals, 4(1), 82–92.

Mata-Alvarez, J., Dosta, J., Romero-Gueiza, M. S., Fonoll, X., Peces, M., &

Astals, S. (2014). A critical review on anaerobic co-digestion achieve-

ments between 2010 and 2013. Renewable & Sustainable Energy

Reviews, 36, 412–427. https://doi.org/10.1016/j.rser.2014.04.039

Matsushita, O., & Okabe, A. (2001). Clostridial hydrolytic enzymes

degrading extracellular components. Toxicon, 39(11), 1769–1780.

McEniry, J., & O’Kiely, P. (2013). Anaerobic methane production from

five common grassland species at sequential stages of maturity. Biore-

source Technology, 127, 143–150. https://doi.org/doi: 10.1016/j.bior

tech.2012.09.084

Mertens, D. R. (2007). Digestibility and intake. In R. F. Barnes, C. J. Nel-

son, K. J. Moore & M. Collins (Eds.), Forages, the science of grassland

agriculture volume II (6th ed., pp. 487–507). Ames, IA: Blackwell Pub-

lishing Professional.

Mgbeahuruike, A. C., Norgaard, P., Eriksson, T., Nordqvist, M., & Nadeau,

E. (2016). Faecal characteristics and milk production of dairy cows in

early-lactation fed diets differing in forage types in commercial herds.

Acta Agriculturae Scandinavica Section A-Animal Science, 66(1), 8–16.

https://doi.org/10.1080/09064702.2016.1193216

M€oller, K. (2015). Effects of anaerobic digestion on soil carbon and nitro-

gen turnover, N emissions, and soil biological activity. A review.

Agronomy for Sustainable Development, 35(3), 1021–1041.

M€oller, K., & M€uller, T. (2012). Effects of anaerobic digestion on diges-

tate nutrient availability and crop growth: A review. Engineering in

Life Sciences, 12(3), 242–257. https://doi.org/10.1002/elsc.

201100085

Moller, H. B., Nielsen, A. M., Murto, M., Christensson, K., Rintala, J.,

Svensson, M., . . . Kaparaju, P. (2008). Manure and energy crops for bio-

gas production -Status and barriers. Nordic Council of Ministers,

Copenhagen.

Moller, H. B., Sarker, S., Hellwin, A. L. F., & Weisbjerg, M. R. (2012).

Quantification of methane production and emission from anaerobic

digestion of cattle manure derived from different feeding. Energy, bio-

mass and biological residues. International Conference of Agricultural

Engineering - CIGR-AgEng 2012: Agriculture and Engineering for a

Healthier Life, Valencia, Spain, 8-12 July 2012, C-0327.

Moller, H. B., Sommer, S. G., & Ahring, B. (2004). Methane productivity

of manure, straw and solid fractions of manure. Biomass & Bioenergy,

26(5), 485–495. https://doi.org/10.1016/j.biombioe.2003.08.008

M€uller, B., Sun, L., Westerholm, M., & Schn€urer, A. (2016). Bacterial com-

munity composition and fhs profiles of low-and high-ammonia biogas

digesters reveal novel syntrophic acetate-oxidising bacteria. Biotech-

nology for Biofuels, 9(1), 1.

Nadeau, E. (2007). Effects of plant species, stage of maturity and additive

on the feeding value of whole-crop cereal silage. Journal of the

Science of Food and Agriculture, 87(5), 789–801. https://doi.org/10.

1002/jsfa.2773

Novak, S. M., & Fiorelli, J. L. (2010). Greenhouse gases and ammonia

emissions from organic mixed crop-dairy systems: A critical review of

mitigation options. Agronomy for Sustainable Development, 30(2), 215–

236. https://doi.org/10.1051/agro/2009031

AHLBERG-ELIASSON ET AL. | 753

Pucker, J., Jungmeier, G., Siegl, S., & Poetsch, E. M. (2013). Anaerobic

digestion of agricultural and other substrates - implications for green-

house gas emissions. Animal, 7, 283–291. https://doi.org/10.1017/

s1751731113000840

Rajagopal, R., Masse, D. I., & Singh, G. (2013). A critical review on inhibi-

tion of anaerobic digestion process by excess ammonia. Bioresource

Technology, 143, 632–641. https://doi.org/10.1016/j.biortech.2013.

06.030

Rasi, S., Lantela, J., & Rintala, J. (2011). Trace compounds affecting biogas

energy utilisation - A review. Energy Conversion and Management, 52

(12), 3369–3375. https://doi.org/10.1016/j.enconman.2011.07.005

Risberg, K., Sun, L., Leven, L., Horn, S. J., & Schnurer, A. (2013). Biogas

production from wheat straw and manure - Impact of pretreatment

and process operating parameters. Bioresource Technology, 149, 232–

237. https://doi.org/10.1016/j.biortech.2013.09.054

Rockstrom, J., Williams, J., Daily, G., Noble, A., Matthews, N., Gordon, L.,

& Smith, J. (2017). Sustainable intensification of agriculture for

human prosperity and global sustainability. Ambio, 46(1), 4–17.

https://doi.org/10.1007/s13280-016-0793-6

Ruile, S., Schmitz, S., Moench-Tegeder, M., & Oechsner, H. (2015).

Degradation efficiency of agricultural biogas plants - A full-scale

study. Bioresource Technology, 178, 341–349. https://doi.org/10.

1016/j.biortech.2014.10.053

Rustas, B. O., Bertilsson, J., Martinsson, K., Elverstedt, T., & Nadeau, E.

(2011). Intake and digestion of whole-crop barley and wheat silages

by dairy heifers. Journal of Animal Science, 89(12), 4134–4141.

https://doi.org/10.2527/jas.2010-3585

Ryckebosch, E., Drouillon, M., & Vervaeren, H. (2011). Techniques for

transformation of biogas to biomethane. Biomass and Bioenergy, 35

(5), 1633–1645. https://doi.org/doi: 10.1016/j.biombioe.2011.02.033

Saini, A., Aggarwal, N. K., Sharma, A., & Yadav, A. (2015). Actinomycetes:

A source of lignocellulolytic enzymes. Enzyme Research, 2015, 1–15.

Schn€urer, A. (2016). Biogas production: Microbiology and technology. In

R. Hatti-Kaul, G. Mamo, & B. Mattiasson (Eds.), Anaerobes in Biotech-

nology (pp. 195–234). Cham, Switzerland: Springer International Pub-

lishing.

Schn€urer, A., Bohn, I., & Moestedt, J. (2017). Protocol for start-up and

operation of CSTR biogas processes. In: McGenity T., Timmis K.,

Nogales B. (eds) Hydrocarbon and Lipid Microbiology Protocols.

Springer Protocols Handbooks. Springer, Berlin, Heidelberg 171–200.

Seppala, M., Pyykkonen, V., Vaisanen, A., & Rintala, J. (2013). Biomethane

production from maize and liquid cow manure - Effect of share of

maize, post-methanation potential and digestate characteristics. Fuel,

107, 209–216. https://doi.org/10.1016/j.fuel.2012.12.069

Sondergaard, M. M., Fotidis, I. A., Kovalovszki, A., & Angelidaki, I. (2015).

Anaerobic co-digestion of agricultural byproducts with manure for

enhanced biogas production. Energy & Fuels, 29(12), 8088–8094.

Sun, L., Liu, T., M€uller, B., & Schn€urer, A. (2016). The microbial commu-

nity structure in industrial biogas plants influences the degradation

rate of straw and cellulose in batch tests. Biotechnology for Biofuels, 9

(1), 1–20. https://doi.org/10.1186/s13068-016-0543-9

Sun, L., Pope, P. B., Eijsink, V. G. H., & Schn€urer, A. (2015). Characteriza-

tion of microbial community structure during continuous anaerobic

digestion of straw and cow manure. Microbial Biotechnology, 8(5),

815–827. https://doi.org/10.1111/1751-7915.12298

Sundberg, C., Al-Soud, W. A., Larsson, M., Alm, E., Yekta, S. S., Svensson,

B. H., & Karlsson, A. (2013). 454 pyrosequencing analyses of bacterial

and archaeal richness in 21 full-scale biogas digesters. Fems Microbiol-

ogy Ecology, 85(3), 612–626.

Swedish Board of Agriculture (2017). Recommendations for fertilization

and liming (Rekommendationer f€or G€odsling och kalkning). Retrieved

from http://webbutiken.jordbruksverket.se/sv/artiklar/rekommen

dationer-for-godsling-och-kalkning.html.

Thanh, P. M., Ketheesan, B., Yan, Z., & Stuckey, D. C. (2016). Trace metal

speciation and bioavailability in anaerobic digestion: A review.

Biotechnology Advances, 34(2), 122–136. https://doi.org/10.1016/j.b

iotechadv.2015.12.006

Triolo, J. M., Sommer, S. G., Møller, H. B., Weisbjerg, M. R., & Jiang, X. Y.

(2011). A new algorithm to characterize biodegradability of biomass

during anaerobic digestion: Influence of lignin concentration on

methane production potential. Bioresource Technology, 102(20), 9395–

9402. https://doi.org/doi: 10.1016/j.biortech.2011.07.026

Tufaner, F., & Avs�ar, Y. (2016). Effects of co-substrate on biogas produc-

tion from cattle manure: A review. International Journal of Environ-

mental Science and Technology, 13(9), 2303–2312. https://doi.org/10.

1007/s13762-016-1069-1

Van Soest, P. J. (1994). Nutritional ecology of the ruminants (2nd ed.).

Cornell Univesity Press.

Van Soest, P. J., Robertson, J. B., & Lewis, B. A. (1991). Methods for diet-

ary fiber, neutral detergent fiber, and nonstarch polysaccharides in

relation to animal nutrition. Journal of Dairy Science, 74(10), 3583–

3597. https://doi.org/doi: 10.3168/jds.S0022-0302(91)78551-2

Vaneeckhaute, C., Meers, E., Michels, E., Buysse, J., & Tack, F. M. G.

(2013). Ecological and economic benefits of the application of bio-

based mineral fertilizers in modern agriculture. Biomass & Bioenergy,

49, 239–248. https://doi.org/10.1016/j.biombioe.2012.12.036

Wall, D. M., Allen, E., Straccialini, B., O’Kiely, P., & Murphy, J. D. (2014).

Optimisation of digester performance with increasing organic loading

rate for mono- and co-digestion of grass silage and dairy slurry. Biore-

source Technology, 173, 422–428. https://doi.org/doi: 10.1016/j.bior

tech.2014.09.126

Weiland, P. (2010). Biogas production: Current state and perspectives.

Applied Microbiology and Biotechnology, 85(4), 849–860. https://doi.

org/10.1007/s00253-009-2246-7

Westerholm, M., Roos, S., & Schnurer, A. (2010). Syntrophaceticus schin-

kii gen. nov., sp. nov., an anaerobic, syntrophic acetate-oxidizing bac-

terium isolated from a mesophilic anaerobic filter. FEMS Microbiology

Letters, 309(1), 100–104.

How to cite this article: Ahlberg-Eliasson K, Liu T, Nadeau E,

Schn€urer A. Forage types and origin of manure in codigestion

affect methane yield and microbial community structure.

Grass Forage Sci. 2018;73:740–757. https://doi.org/10.1111/

gfs.12358

754 | AHLBERG-ELIASSON ET AL.

APPENDIX 1

TABLE A1 Contents of nutrients, minerals and volatile fatty acids (VFA) for the different substrate mixes used in the continuous stirredtank reactors

Parameters Unit

Substrate (mix) Exp A Substrate (mix) Exp B

R1 R2 R3 R4 R1 R2 R3 R4

DM % of ww 10 10 10 9 13 12 12 12

VS % of ww 9 8 9 8 11 11 11 11

Protein g/kg DM 99 138 107 148 149 180 157 190

Starch g/kg DM 34 11 34 10 99 83 101 84

NDF g/kg DM 480 450 512 481 402 376 423 398

ADF g/kg DM 334 346 329 341 257 265 250 258

ADL g/kg DM 90 102 77 89 70 78 59 68

Crude Fat g/kg DM 23 29 22 29 25 30 25 30

WSC g/kg DM 48 13 50 14 51 25 53 26

VOS % of VS 53 47 54 56 36 35 41 41

Total N % of DM 27 34 24 31 32 38 30 36

Am-N g/kg DM 10 11 6 7 8 9 5 5

Ca g/kg DM 13 17 7 12 10 14 6 9

K g/kg DM 29 36 17 24 26 31 16 21

P g/kg DM 4.2 4.4 4.1 4.3 4.3 4.5 4.3 4.4

Mg g/kg DM 4.1 4.8 4.5 5.3 3.7 4.2 4.0 4.6

Na g/kg DM 3.5 3.5 2.2 2.1 2.7 2.7 1.7 1.6

S g/kg DM 2.3 2.7 2.9 3.3 2.4 2.7 2.8 3.1

Cu mg/kg DM 21 22 19 20 18 19 16 17

Fe mg/kg DM 952 761 943 743 756 606 742 586

Mn mg/kg DM 166 182 121 136 135 146 99 110

Zn mg/kg DM 76 80 95 100 68 71 83 86

Lactate g/kg DM 13 23 14 24 10 17 11 18

Acetate g/kg DM 34 34 27 27 26 26 21 20

Propionate g/kg DM 1.9 2.3 1.7 2.1 1.5 1.8 1.3 1.6

Butyrate g/kg DM 0.5 0.5 0.6 0.6 0.4 0.4 0.4 0.4

Ethanol g/kg DM 3.5 1.1 3.8 1.2 2.7 0.9 2.9 0.9

Formic acid g/kg DM 0.8 0.5 0.9 0.6 0.6 0.4 0.7 0.4

2.3 Butanediol g/kg DM 1.2 2.1 1.3 1.0 0.9 1.6 1.0 0.7

DM, dry matter; VS, volatile solids; NDF, neutral detergent fibre; ADF, acid detergent fibre; ADL, acid detergent lignin; WSC, water-soluble carbohy-

drates (i.e. sugar); VOS, organic matter digestibility in vitro. Substrate mix: R1, OCM:WCB; R2, OCM:GCS; R3, CCM:WCB; R4, CCM:GCS.

In Exp B, SW was added to all substrates.

AHLBERG-ELIASSON ET AL. | 755

TABLE A2 Summary of observed operational taxonomic units (OTUs). Chao1, Shannon and Simpson index

Samples Chao1† Observed OTUs Sequencing coverage (%) Shannon‡ Simpson‡

CCM 864 650 75.2 6.343 0.932

OCM 841 615 73.1 7.419 0.987

INOCULUM 1,474 1,087 73.7 5.059 0.835

R194 1,887 1,673 88.7 6.119 0.909

R1105 1,937 1,741 89.9 6.603 0.95

R1174 1,986 1,741 87.7 6.249 0.912

R1211 1,842 1,612 87.5 6.237 0.941

R294 2,004 1,856 92.6 7.402 0.972

R2105 2,119 1,929 91 7.345 0.965

R2174 1,932 1,604 83 5.121 0.753

R2211 1,833 1,562 85.2 6.196 0.888

R394 1,909 1,686 88.3 7.098 0.963

R3105 1,834 1,655 90.2 6.81 0.949

R3174 1,952 1,646 84.3 6.686 0.95

R3211 2,030 1,777 87.5 6.541 0.923

R494 2,103 1,929 91.7 7.424 0.973

R4105 2,154 1,950 90.5 7.253 0.96

R4174 1,879 1,664 88.6 6.111 0.876

R4211 1,935 1,692 87.4 6.514 0.925

†Chao1 represented the community richness.‡Shannon index and Simpson index represent the community diversity and evenness. Higher value indicates higher richness and diversity and evenness.

TABLE A3 Contents of nutrients in the ration fed to dairy cattle herds used for collection of manures (CCM and OCM) used as substrate inthe present study

Parameters UnitConventionaldairy herd Organic dairy herd

Crude protein g/kg DM 194 170

NDF g/kg DM 330 393

Crude fat g/kg DM 52 15

F IGURE A1 Calculated digestibility for protein and neutral detergent fibre (NDF) at the end point of Exp A and Exp B

756 | AHLBERG-ELIASSON ET AL.

F IGURE A2 Number of observed species in the digestate from reactors R1-R4, conventional (CCM) and organic (OCM) manure samples

and inoculum sample

F IGURE A3 Relative abundance of the archaeal community (genus level) obtained by illumina sequencing. R1–R4, reactors; OCM, organiccattle manure; CCM, conventional cattle manure; 94–211, days of operation

AHLBERG-ELIASSON ET AL. | 757