RESEARCH ARTICLE

Impact of trace element addition on biogas production from food industrial waste ^ linking process to microbial communities Xin Mei Feng1,2, Anna Karlsson1, Bo H. Svensson1 & Stefan Bertilsson2 1

¨ ¨ Department of Water and Environmental Studies, Linkoping University, Linkoping, Sweden; and 2Limnology/Department of Ecology and Evolution, Uppsala University, Uppsala, Sweden

Received 25 November 2009; revised 10 May 2010; accepted 9 June 2010. Final version published online 14 July 2010. DOI:10.1111/j.1574-6941.2010.00932.x

MICROBIOLOGY ECOLOGY

Editor: Alfons Stams Keywords anaerobic digestion; real-time T-RFLP; experimental design; microbial community composition; trace elements.

Abstract Laboratory-scale reactors treating food industry waste were used to investigate the effects of additions of cobalt (Co), nickel/molybdenum/boron (Ni/Mo/B) and selenium/tungsten (Se/W) on the biogas process and the associated microbial community. The highest methane production (predicted value: 860 mL g1 VS) was linked to high Se/W concentrations in combination with a low level of Co. A combination of quantitative real-time PCR of 16S rRNA genes, terminal restriction fragment length polymorphism (T-RFLP) and clone library sequencing was used for the community analysis. The T-RFLP data show a higher diversity for bacteria than for archaea in all the treatments. The most abundant bacterial population (31–55% of the total T-RFLP fragments’ intensity) was most closely related to Actinomyces europaeus (94% homology). Two dominant archaeal populations shared 98–99% sequence homology with Methanosarcina siciliae and Methanoculleus bourgensis, respectively. Only limited influence of the trace metal additions was found on the bacterial community composition, with two bacterial populations responding to the addition of a combination of Ni/Mo/B, while the dominant archaeal populations were influenced by the addition of Ni/Mo/B and/or Se/W. The maintenance of methanogenic activity was largely independent of archaeal community composition, suggesting a high degree of functional redundancy in the methanogens of the biogas reactors.

Introduction The basic techniques for biogas production are well established in many countries and it is known to be an efficient strategy for producing biofuel (B¨orjesson & Mattiasson, 2008). In 2008, Sweden had 18 full-scale codigestion biogas plants (not including the sewage treatment plants) treating waste from households and/or food industry (Held et al., 2008). The process has developed from using relatively simple substrates, such as manure, towards more complex mixtures that often have a higher biogas production potential. Thus, more methane is obtained per m3 reactor volume. However, these more energy-rich substrates also result in less stable process conditions, where foaming and suboptimal use of the organic material may occur. A balanced availability of nutrients for the growth of the microorganisms in biogas digesters is important for the process performance, i.e. stability and substrate utilization (Taka2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

shima & Speece, 1989a). Apart from a balance among the macronutrients (C, N, P, etc.), the availability of certain trace elements has also been shown to strongly impact the biogas production. Trace elements known to be crucial for the activity of enzymes in methanogenic systems are cobalt (Co), nickel (Ni), iron (Fe), zinc (Zn), molybdenum (Mo) and/or tungsten (W) (reviewed by Takashima & Speece, 1989a). Often, additions of a single or combinations of trace elements to laboratory-scale processes have improved the performance, with a faster turnover of substrate and lower levels of volatile fatty acids (VFA). Positive effects have been reported both on the overall degradation process (Murray & van den Berg, 1981; Jarvis et al., 1997; Kim et al., 2002; Noyola & Tinajero, 2005; Zandvoort et al., 2006b; Zhang et al., 2008; Pobeheim et al., 2010) and on specific substrates such as acetate, propionate or methanol (Sch¨onheit et al., 1979; Fathepure, 1987; Takashima & Speece, 1989b; Kida et al., 2001; Osuna et al., 2003; Worm et al., 2009). However, FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Correspondence: Anna Karlsson, Department of Water and Environmental ¨ Studies, Linkoping University, 58183 ¨ Linkoping, Sweden. Tel.: 1461 328 4089; fax: 1461 313 3630; e-mail: [email protected]

227

Biogas production from food industrial waste

FEMS Microbiol Ecol 74 (2010) 226–240

environmental factors and process performance. Hence, more research on biogas community composition is needed. Furthermore, to the best of our knowledge, there are so far no studies on the microbial communities’ composition in biogas reactors treating protein-rich food industry waste as the main substrate. In the present study, the effects of various concentrations of three trace elements/trace element mixtures on a tank reactor biogas process performance and bacterial and archaeal community composition were studied in reactors fed with food industrial wastes at a final organic load of 4.0 g VS L1 day1. The overall aim was to determine optimal concentrations and combinations of the trace elements to obtain stable biogas processes working at high loading rates and thus enhancing the methane production. A second goal was to investigate to what extent the trace element-affected biogas processes differed in the bacterial and archaeal community composition. In order to cover wide ranges for the trace element additions, a reduced factorial design was used. The trace elements added were boron (B), Mo and Ni (trace element mix, TEM1), Se and W (TEM2) and Co. The process was evaluated by monitoring methane production per gram added VS, VS reduction, pH and VFA levels over time. Bacterial and archaeal communities were analysed using domain-specific real-time PCR combined with terminal restriction fragment length polymorphism (real-time T-RFLP). T-RFLP is a useful tool for comparing complex microbial communities responding to different environmental gradients and experimental manipulations (Osborne et al., 2006). A combination of real-time PCR and T-RFLP can simultaneously determine the microbial community composition and the abundance of specified groups within a complex community (Yu et al., 2005). In addition, the T-RFs obtained in the T-RFLP were identified by cloning and sequencing.

Materials and methods Experimental design A central composite face-centred design with three input variables (Eriksson et al., 2000) was used for the optimization experiment. The three input variables TEM1, TEM2 and Co were assigned three values: high, middle and low (Table 1). The study included 18 reactors: eight corner points, six axial points and four replicated centre points (Table 2). The four replicated centre points (middle value of all input variables) yield the experimental variation (replicate error) of the set-up. The software MODDE 6 (Umetrics AB, Ume˚a, Sweden) was used for experimental design.

Trace elements -- which and how much? The decision to include Co and Ni in the study was based on earlier experiences (unpublished data). The other elements 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

combinations of several trace elements have shown both synergetic and antagonistic effects. Murray & van den Berg (1981) found that Co combined with Ni increased the turnover of acetate more than expected when compared with experiments where these metals were added individually. In addition, Mo was shown to enhance reactor performance only in combination with Co and Ni. In a study of methane production from sulphate-laden organics, Patidar & Tare (2006) observed synergetic effects of Fe and Co and antagonistic effects of Co and Ni on the total methanogenic activity. In addition, Kim et al. (2002) further reported that organic loading is an important factor when evaluating the effects of trace element additions on a specific process. These authors showed that the positive effect of the addition of calcium (Ca), Fe, Ni and Co to a biogas process treating dog food was most pronounced at high organic loading rates (OLRs) [10–20 g volatile solids (VS) L1 day1]. However, in spite of the relative large number of investigations within this area, there still remain large knowledge gaps regarding optimal trace element combinations and concentrations for many substrates commonly used in fullscale applications. Furthermore, the main elements studied in relation to methane production efficiency are Fe, Ni and Co, while reports on selenium (Se), W and other trace elements are scarce. In most cases, the effect of trace element addition has been assessed as process performance (methane production, volatile solids-reduction, levels of VFAs, etc.). These parameters have, in some studies, been linked to total methanogenic activity tests (Patidar & Tare, 2006), specific methanogenic activity (Jarvis et al., 1997; Osuna et al., 2003; Zandvoort et al., 2006a) and maximum potential acetate utilization rate and maximum potential propionate utilization rate (Zitomer et al., 2008). However, to the best of our knowledge, only one study has linked such process parameters to microbial community compositions (Fermoso et al., 2008). These authors investigated the effect of Co deprivation on the microbial community of a methanol-fed upflow anaerobic sludge blanket (UASB) reactor using specific methanogenic activity and FISH targeting the kingdom Euryarchaeota and the genus Methanosarcina. A number of analyses targeting rRNA or protein-coding genes have been used with the purpose of studying the microbial communities’ composition of biogas processes (Godon et al., 1997; Fern´andez et al., 2000; McHugh et al., 2003; McMahon et al., 2004; Scully et al., 2005; Klocke et al., 2007; Nettmann et al., 2008). The emerging consensus from these studies is that bacterial diversity considerably exceeds archaeal diversity in the biogas process systems (Godon et al., 1997; Zumstein et al., 2000; Klocke et al., 2007). However, many different species from both domains are present and it is difficult to conclude on any general trends in the compositions among the combined collection of studies or on links between community composition,

228

X.M. Feng et al.

were included based on a comparison between their levels in the reactor liquid of the full-scale plant, from which both the starting material and the substrate used in the laboratory study were obtained, and that of an anaerobic growth medium (AGM; Wolin et al., 1963) modified as described by Karlsson et al. (1999). An elemental analysis (Eurofins, Lidk¨oping, Sweden) on a liquid sample from the full-scale process yielded the following concentrations (mg L1; only the trace elements present in the AGM are listed): Al, 39; B,  1.7; Cu, 2.6; Mg, 200; Mn, 7.9; Mo, 0.25; Se,  0.06; S, 1100; W,  0.07 and Zn, 19. Elements below the detection limit (B, Se and W) or present at o 10 times the concentration in the AGM (Mo) were selected for experimental manipulations. For TEM1 and Co, the concentrations supplied in the set-up were fivefold (low), 50-fold (middle) or 500-fold (high) higher than the AGM levels, while for

Variables

Trace element

High (H)

Middle (M)

Low (L)

TEM1

Boron Molybdenum Nickel Selenium Tungsten Cobalt

4.50 13.5 10.0 0.80 1.80 6.00

0.45 1.35 1.00 0.08 0.18 0.60

0.045 0.135 0.100 0.008 0.018 0.060

TEM2 Cobalt

Reactor startup and sampling Five-litre glass reactors with two openings at the top were used for the incubations. Tubes for the gas outlet and for the mixer were fitted into a gas-tight rubber stopper in the central reactor opening. The propellers (four blades; ø 50 mm) used for mixing were mounted on motors from IKA WERZE (Staufen, Germany). Mixing (300 r.p.m.) was performed for 30 min each day in connection to the sampling/feeding. A smaller opening was used for the withdrawal of reactor material and addition of substrate using a 100-mL syringe once a day. The 18 laboratory-scale reactors were run in two batches: reactors 1–9 were started on 19 April 2006 (set1) and reactors 10–18 were started on 8 September 2006 (set2). The incubation temperature was 37 1C. In both cases, material from the same full-scale biogas plant was used to start the laboratory reactors. The full-scale biogas plant treats food industrial wastes from a local slaughterhouse (62%), other food industries (21%), mycelium (10%) and stillage from ethanol production (7%). The protein content of the substrate was 70%. At the time of sampling, the plant was run at an OLR of 2.5–3.0 g VS L1 day1 with a hydraulic retention time (HRT) of 40–50 days. Both sets of

Table 2. Experimental design set-up and mean methane production (mL g1 added VS day1) and VS reduction (% of added VS) values from days 60 to 90 Reactor No.

TEM1/ TEM2/Co

Setup

CH4 (  SD) n 4 20

VS-red (  SD) n=5

Final VFA (mM)

Bacteria (copies mL1)

Archaea (copies mL1)

Archaea/ Bacteria (%)

1w 2w 3 4 5w 6w 7w 8w 9w 10 11w 12w 13w 14 15w 16 17 18w

M/M/M L/H/L M/L/M M/H/M H/H/H M/M/M L/L/L L/L/H H/L/H H/L/M H/L/L H/H/L L/H/H M/M/H M/M/M M/M/L L/M/M M/M/M

Cep Cop Axp Axp Cop Cep Cop Cop Cop Axp Cop Cop Cop Axp Cep Axp Axp Cep

786 (  89) 882 (  126) 704 (  55) 775 (  51) 760 (  144) 804 (  100) 772 (  66) 583 (  71) 793 (  129) 723 (  51) 739 (  82) 831 (  53) 781 (  66) 796 (  117) 796 (  64) 846 (  58) 712 (  64) 750 (  57)

70 (  2) 67 (  3) 69 (  1) 69 (  3) 69 (  3) 70 (  3) 69 (  6) 77 (  2) 69 (  3) 68 (  1) 67 (  3) 66 (  2) 64 (  3) 67 (  1) 67 (  2) 66 (  2) 67 (  2) 65 (  3)

1.7 2.4 1.9 4.0 1.6 1.4 7.2 9.5 1.6 2.1 2.1 1.6 3.0 1.6 1.6 1.9 1.5 3.3

23018 11090

8199 193

36 2

15963 23131 10148 4394 9025

1375 4690 60 49 532

9 20 1 1 6

7751 12463 17975

2365 1568 536

31 13 3

34471

493

1

42000

855

2

‘Final VFA’ is the sum of all VFAs (mM) at the final sampling point. Bacterial and archaeal copies and proportion are values from the final sampling of the reactor material. The 12 reactors on which microbial community analyses were performed are marked with w. H, high concentration; L, low concentration; M, middle concentration; Cop, corner point; Cep, centre point; Axp, axial point; VS-red, volatile solids reduction.

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Table 1. Trace elements added; high, middle and low concentrations in the reactors after addition are given (mg L1)

TEM2 they were 1-, 10- and 100-fold higher than the AGM levels, respectively.

229

Biogas production from food industrial waste

Genomic DNA extraction Thawed reactor liquid aliquots of 200 mL were centrifuged for 10 min at 14 000 g, followed by removal of the supernatants. The pellets were used to extract DNA using the FastDNAs Spin Kit for soil (Qbiogene, Carlsbad, CA) according to the manufacturer’s instructions with the modification of beadbeating for 30 s, followed by cooling on ice for 1 min and beating for an additional 30 s at 4000 r.p.m. in a Mini-Bead beater (Biospec Products, Bartlesville, OK). The size of extracted DNA ranged from 4 to 20 kb, revealed by electrophoresis on a 1% agarose gel (Eiler & Bertilsson, 2007). Real-time PCR analysis Quantitative real-time PCR was performed in triplicate using bacterial forward primer 27f labelled with hexafluorescein (hex; MWG Biotech, Germany) and universal reverse primer 519r as described previously (Eiler & Bertilsson, 2007). The hex labelling enabled downstream fluorescence detection of T-RFs. A Chromo4 real-time PCR system (BioRad) was used for the reactions, with an initial denaturation at 95 1C for 7 min, followed by 25 cycles of 30 s at 95 1C, 30 s at 50 1C and 30 s at 72 1C, and a final extension at 72 1C for 7 min. Each 20-mL reaction included 4–6 ng of template FEMS Microbiol Ecol 74 (2010) 226–240

DNA, 10 mL of 2  DyNAmoTM Flash SYBRs Green qPCR Master Mix (Finnzymes), 250 nM of each primer and 250 ng mL1 of BSA (New England Biolabs, MA). In parallel, quantitative real-time PCR for archaea was performed with archaeal primers ARC344f (5 0 -ACGGGGY GCAGCAGGCGCGA-3 0 ) and ARC915r-hex (5 0 -GTGCTCC CCCGCCAATTCCT-3 0 ; Casamayor et al., 2002). The initial real-time PCR results indicated that the concentration of archaea differed substantially among the reactors. Therefore, the amount of template DNA for archaeal PCRs was adjusted to levels from 2 to 235 ng per reaction. The reaction condition was the same as for the bacterial real-time PCR, except that 28 cycles were run with the annealing temperature set to 61 1C. Quantification of bacterial target DNA was performed by interpolating the threshold cycle (Ct) values of the reactor DNA samples with the Ct values of a standard regression curve from known concentrations of purified 10-fold dilutions of genomic DNA from a Flavobacterium isolate at 10–10 000 rRNA copies mL1 (Eiler & Bertilsson, 2007). Quantification of archaea was performed analogously using genomic DNA from Methanospirillum hungateii. The resulting standard curves for bacteria had slopes between 3.3 and 3.5, R2 between 0.95 and 0.99 and a PCR efficiency between 91% and 100%. For archaea, the slopes ranged from 3.1 to 3.5, R2 was above 0.99 and efficiency between 94% and 110%. The results are expressed as 16S rRNA gene copy numbers, and the relative copy number of archaea to bacteria was calculated for each reactor. T-RFLP analysis Hex-labelled real-time PCR products were purified and concentrated using the QIAquick PCR purification kit (Qiagen, Hilden, Germany). Twenty nanograms of purified PCR product aliquots were subject to restriction enzyme digestion (Eiler & Bertilsson, 2007) with enzymes HhaI, HaeIII and RsaI. Fluorescently labelled T-RFs were sized on an ABI 3700 96-capillary sequencer running in the GeneScan mode (Applied Biosystems) and analysed using GENEMARKER 1.6 (SoftGenetics LLC). A baseline threshold of 100 fluorescence units was used to separate peaks from background noise. Peaks o 0.5 bases apart were binned. T-RFs o 50 and 4 600 bp were removed from subsequent analyses. To account for uncontrolled differences in the quantity of DNA between samples, the raw peak height of individual T-RFs was divided by the total T-RF peak height of the respective sample (Culman et al., 2008). The reproducibility of T-RFLP was tested using results from triplicate PCR products originating from the same DNA extraction where binary T-RFLP patterns were 100% similar among triplicates. Both T-RFLP electropherograms and in silico enzyme cutting showed that HaeIII (bacteria) and HhaI (archaea) 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

laboratory-scale reactors were started with an OLR of 2.5 g VS L1 day1, which was increased in intervals of 0.5 g up to 4.0 g VS L1 day1 in order to enforce process operational limits. The final OLR was reached after 50 days of incubation and then sustained for another 48 days. The HRT of the reactors was 25 days. Trace elements were added with the daily feed to reach the target concentrations in Table 1. The additions were made from stock solutions of CoCl2  H2O; NiCl2  H2O, HB, (NH4)6Mo7O24  4H2O (TEM1) and Na2SeO3  5H2O, NaWO4  5H2O (TEM2). All reactors also received Fe with the daily feed at a concentration of 0.5 g L1 in the form of FeCl2. The total gas production from the reactors was measured using gas metres working on the basis of water displacement. The methane content was determined twice a week using a GFM 400 Gas Analyser (Gas Data Ltd Whitley, UK). pH (PHM93; Radiometer, Copenhagen, Denmark), VFA concentrations (acetate, propionate, butyrate, isobutyrate, valeriate, isovaleriate, capronate and isocapronate; Jonsson & Bor´en, 2002) and total and volatile solids (TS and VS; Swedish standard SS-028113;25) of the reactor liquid were determined once a week. The samples from the eight cornerand four centre-point reactors (Table 2) were analysed for the microbial community composition. Aliquots of 1 mL liquid were withdrawn from each reactor once a week and transferred to 1.5-mL Eppendorf tubes and stored at  80 1C until DNA extraction.

230

generated a large number of fragments of the desired size (50–500 bp). Therefore, most of the data analysis on microbial community change was based on T-RFs from HaeIII (bacteria) or HhaI (archaea), while T-RF data from all three restriction enzymes were included in the nonmetric multidimensional scaling (NMS) analysis. Analytical duplicates from each individual reactor were normalized for each enzyme, applying a specific threshold for peak scoring, where there was no correlation between the remaining peak numbers and the original total cumulative peak height (Osborne et al., 2006).

Cloning and sequencing

evaluated in relation to the three trace element input variables. A quadratic polynomial model was fitted to the methane production data using the partial least square (PLS) method available in the software MODDE 6 (Umetrics AB) and goodness of fit (i.e. squared correlation coefficient, R2; 0–1) and goodness of prediction (predictive power, Q2;  1–1) were calculated for model evaluation. The models were also evaluated using ANOVA tests (MODDE 6); both ‘regression model significant’ tests and ‘lack of fit’ tests were performed (cf. Eriksson et al., 2000). The PLS was used to identify the relative influence of each input variable as well as the effect of their settings (low, middle or high). Also, data on VFA and acetate concentrations (mM; days 88 and 89 for sets 1 and 2, respectively), VS reduction (mean values days 60–90) and pH from all reactors as well as the microbial community composition of the 12 centre- and corner-point reactors were tested for correlations with the trace elements input (TEM1, TEM2 and Co) using PLS. The 11 most frequent bacterial T-RFLP HaeIII fragments and the three most frequent archaeal HhaI fragments were used in the community composition analyses. The community fingerprint data were explored by NMS analysis (Clarke, 1999) using the software PC-ORD 5.10 (MjM Software, OR). The NMS method is considered to be appropriate for most complex datasets (Rees et al., 2005; Culman et al., 2008). For NMS ordinations, data matrices of community fingerprints were log(X11) transformed and the Bray–Curtis correlation coefficients were estimated based on the presence of shared T-RFs in the community profiles. The NMS was run with a random initial configuration, a maximum of 200 iterations and an instability criterion of 0.00001 and performed on 50 runs of the real data and 50 runs of the randomized data to test the null hypothesis of no treatment effects. For a two-dimensional solution with the lowest possible stress value ( o 10, a good ordination with no real risk of drawing false inferences), a final run using the best starting configuration from the first run was performed (McCune & Grace, 2002). The significance of the relationships explored by NMS was confirmed by the null hypothesis tested by Monte Carlo simulation with 999 randomizations. The significance of the relationships between bacterial and archaeal community compositions was tested by comparing the similarity matrices of the fingerprint data using Mantel’s test (Mantel, 1967) in XLSTAT 2007 (Pearson’s correlation with 1000 permutations).

Results Reactor performance

Data handling and statistical analysis The amount of methane produced from each reactor (mL g1 VS day1; mean value from days 60 to 90) was 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

The average amount of methane produced in the 18 reactors during days 60–90 varied between c. 700 and 900 mL g1 VS day1 (Table 2), with the exception of reactor FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Based on T-RFLP profiles, bacterial 16S rRNA gene clone libraries were constructed from PCR products amplified from the first batch startup sample (RK1) and the reactor 13 at day 97. Archaeal 16S rRNA gene clone libraries were constructed from PCR products amplified from the startup material of both batches (RK1 and RK2). PCR amplification for cloning was performed as described above, but using DyNAzymeTM II DNA Polymerase (Finnzymes) and fewer PCR cycles (20 for bacteria and 28 for archaea). Cloning was performed using the TOPO-TA cloning kit for sequencing as described by the manufacturer (Invitrogen, CA). For each clone library, up to 95 positive clones were screened with restriction enzymes HaeIII for bacteria and HhaI for archaea and subsequently classified into operational taxonomical units (OTUs; Eiler & Bertilsson, 2004). Duplicate clones from each OTU were sequenced on an ABI 3700 96-capillary sequencer (Applied Biosystems) using the primer M13f. Sequence scanner v1.0 was used to view and edit sequences. Chimeric sequences were identified using the RDPII CHIMERA CHECK program. One bacterial and two archaeal sequences were removed as likely chimeras. The remaining sequences were aligned against GenBank database entries using standard nucleotide BLAST at NCBI. For sequence nomenclature from startup materials (e.g. BRK1_1A), the first letter corresponds to bacteria (B) or archaea (A), the RK refers to startup material and the number after RK refers to startup sets (1 is set1 and 2 is set2). The number after the dash is the clone number. For sequence nomenclature from the reactor samples (e.g. B13_97_3H), the first letter refers to bacteria (B) or archaea (A), the number 13 refers to reactor 13, the number 97 corresponds to the sampling day and 3H is the clone number. The archaeal and bacterial 16S rRNA gene sequences obtained were deposited in GenBank under the accession numbers GQ501011–GQ501015 and GQ501016–GQ501042, respectively.

X.M. Feng et al.

231

Biogas production from food industrial waste

FEMS Microbiol Ecol 74 (2010) 226–240

Methane production (mL

1.8

1.5

1.2

g VS)

0.3 0 0.9 0.6

0.9 1.2 1.5 1.8 2.1 2.4 860

860

840

840

820

820

800

800

780

780

760

760

740 0 Co 10 log 0.3 0.6

740 TEM2 10 log

0.9 1.2 1.5

0.9 1.8

1.2 2.1

1.5 2.4

1.8

Fig. 1. The effect of the trace element additions on the CH4 production (mL g1 added VS day1) given by the best regression model obtained with MODDE 6. The values for TEM2 and Co given on the axis are 10 log values of the multiples from the AGM; see Table 1 for the actual values. The CH4 production values used are the means of days 60–90. The model is based on values from 17 of the 18 reactors as reactor 8 was regarded as an outlier.

(the sum of all analysed species in the final samples; R2 = 0.075 and Q2 = 0), the final acetate concentration data (R2 = 0.13 and Q2 = 0) or the final pH (R2 = 0.23 and Q2 = 0) and the trace element concentrations in the analyses.

Microbial community composition The estimates of the total amounts of microorganisms (bacteria and archaea 16S rRNA gene copy numbers observed by real-time PCR) were the highest in the centrepoint reactors (1, 6, 15 and 18; Table 2). No correlation was found between the total amounts of microorganisms and the amount of biogas produced. The ratio of archaea to bacteria differed substantially among the reactors (Table 2). A paired t-test was applied to investigate the correlation between these ratios and (1) the relative abundance of archaeal HhaI-92 (possibly aceticlastic methanogens) and (2) the relative abundance of archaeal HhaI-531HhaI-113 (hydrogenotrophic methanogens) for each reactor. A positive correlation was obtained between the archaea : bacteria ratio and HhaI-92 (0.71; P o 0.05, d.f. = 21), while a negative correlation was found with HhaI-531HhaI-113 (  0.63; P o 0.05, d.f. = 21). This suggests that an increase in archaea is linked to an increase in the amount of aceticlastic methanogens and a parallel decrease in the number of hydrogenotrophic methanogens. The number of detected bacterial T-RFs for individual samples was 11–19 for HaeIII, 8–16 for HhaI and 6–17 for 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

8 (around 600 mL g1 VS day1). This reactor was overflooded due to foaming on day 49, which resulted in the loss of about 2 L of reactor liquid. The reactor did not recover and was thus excluded from further analysis. Acetate was the most abundant volatile fatty acid in all reactors, with initial concentrations of 10–15 mM for reactors 1–9 and 25–35 mM for reactor 10–18, respectively. After day 30, the concentrations decreased to below 15 mM in all the reactors and the final levels were all below 7 mM (data not shown). The propionate concentration showed two peaks between 5 and 15 mM for reactors 1–9 at days 0–20 and days 40–60, respectively, while reactor 10–18 displayed a single peak (5–12 mM) between days 20 and 30. The final propionate concentrations were all below 2 mM (data not shown). The final VFA concentrations (the sum of all VFAs measured in the final sample) were between 1.4 and 9.5 mM (Table 2). The mean VS reduction between days 60 and 90 ranged from 64% to 70% in all the reactors, except reactor 8 (77%; Table 2). The final pH was between 7.7 and 8.0 (data not shown). Occasionally, all reactors (especially during the last 50 days of the experiment) developed foam layers that were not possible to break by stirring and some minor overflows occurred. PLS was applied on the methane (CH4) production data (mL g1 VS day1) from all reactors, except reactor 8. The best model obtained included the four coefficients ‘TEM2’ (P = 0.048), ‘Co’ (P = 0.086), ‘Co2’ (P = 0.021) and ‘TEM2  Co’ (P = 0.037). The model obtained had an R2 value of 0.63 and Q2 of 0.48. The ‘regression model significant’ and ‘lack of fit’ tests, included in the ANOVA analysis of the model data (for ANOVA details, see Supporting Information, Table S1), showed that the model was significant (P = 0.012) and showed no lack of fit (P = 0.284). According to the model, the highest methane production rates were linked to a low concentration of Co and high concentrations of TEM2 (Se/W). The graphic presentation of the model (Fig. 1) showed that this combination (0.8, 1.8 and 0.06 mg L1 of W, Se and Co, respectively) yielded methane production values above 860 mL g1 VS day1 at an OLR of 4 g VS L1 day1. The lowest daily methane production (around 730 mL g1 VS) was predicted by low levels of TEM2 (0.008 and 0.018 mg L1 of Se and W, respectively) in combination with the middle concentration of Co (0.6 mg L1; Fig. 1), while low concentrations of TEM2 together with high Co levels yielded higher methane production values of approximately 800 mL day1 (Fig. 1). The influence of Co concentration alone in the model was, however, uncertain due to the high P value for this coefficient (0.086). The variation in the levels of TEM1 (B, Mo, Ni) did not have a significant effect on the methane production rate according to the model used. No correlation was observed between VS reduction (R2 = 0.47 and Q2 = 0.24, but not significant), the final VFA

232

X.M. Feng et al.

(a) 100%

80%

60%

40%

20%

0% RK1 RK2

1

2

5

6

7

8

9

11

12

13

15

18

100%

80%

60%

40%

20%

0% RK1 RK2

1

2

5

6

7 8 Reactors

9

11

12

RsaI. The corresponding numbers of detected T-RFs from Archaea were 2–13 for HaeIII, 2–7 for HhaI and 3–14 for RsaI. The bacterial community compositions in the two startup materials differed considerably from those of the 12 studied reactors (Fig. 2a). The bacterial communities in the startup materials were dominated by fragment HaeIII282, while HaeIII-237 was dominant in the 12 reactors (Fig. 2a). Bacterial 16S rRNA gene sequences obtained in the cloning were affiliated with Bacteroidetes, Firmicutes and Actinobacteria. Fragment HaeIII-282 (clones GQ501028, GQ501039, GQ501022 and GQ501041–042) was identified as a Clostridiaceae, but with a mere 85% similarity to an uncultured clone EF559144 from anaerobic municipal solid waste incubated with glucose (Li et al., 2009). Fragment HaeIII-237 (clones GQ501020, GQ501021; GQ501030, GQ501032 and GQ501036) was identified as an Actinobacteria with 94% similarity to the potential pathogen Actinomyces europaeus. The less abundant fragment HaeIII-239 (0–8%) (clones GQ501033 and GQ501040) was identified as an 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

13

15

18

Fig. 2. Relative abundance of terminal restriction fragments for the bacterial (a) and archaeal (b) community in startup materials (RK1 and RK2 for set1 and set2, respectively) and the 12 analysed reactors. The patterns were generated using general bacterial and archaeal primers and restriction enzyme HaeIII for bacteria and HhaI for archaea.

Firmicutes bacterium sharing 99% homology with an uncultured clone EF559138 from anaerobic municipal solid waste incubated with glucose (Li et al., 2009). Three dominant archaeal T-RFs, HhaI-53, HhaI-92 and HhaI-113 were detected in all reactors, but at different abundances (Fig. 2b). The HhaI53 (clones GQ501011 and GQ501015) shared 99% homology with an uncultured archaeon clone and 97–98% homology with a cultured Methanoculleus bourgensis, while fragment HhaI-113 (Clones GQ501012–13) shared 99% homology with the same cultured M. bourgensis (Asakawa & Nagaoka, 2003). Fragment HhaI-92 (GQ501014) shared 99% similarity with an uncultured Methanosarcina sp. and 98% with a cultured Methanosarcina siciliae isolated from a marine sediment (Elberson & Sowers, 1997).

Impacts of trace element additions The NMS analysis revealed that the largest differences in the bacterial communities were between the two startup sets FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

(b)

233

Biogas production from food industrial waste

(a)

(b) 18

2

11 7 15

13

Axis 2

12

13 15

12

6

9

11

18

1

1

9 6

5

5

S = 6.7

S = 1.4

Axis 1 Fig. 3. Ordination plot using NMS based on Bray–Curtis similarities of bacterial (a) and archaeal (b) T-RFLP data consisting of log(X11)-transferred mean relative abundance values of two replicates from three enzyme restrictions among the bacterial and archaeal assemblages in the 12 analysed reactors. Clarke’s stress value of the plot for Fig. 3a is S = 6.7, and for Fig. 3b it is S = 1.4.

(reactors 1–9 and 10–18, respectively; Fig. 3a), while no general trends linking the bacterial community data to the trace element additions could be seen. This conclusion was supported by the low final stress value of 6.7 in the NMS model, and the rejected null hypothesis of no treatment effects (the Monte Carlo test; P o 0.02, 50 permutations). The NMS analysis of archaeal community structure (Fig. 3b) showed a gradual change along axis 1 from Methanoculleus to Methanosarcina being the dominant organism (Figs 2b and b). Reactor 13 was separated from the other reactors in the analysis (Fig. 3b), which might be linked to the absence of detectable Methanosarcina in this reactor (Fig. 2b). The archaeal community structures in reactors 2 and 7 differed from each other and they were also separated from the other reactors. Reactor 2 was dominated by Methanoculleus (especially HhaI-53), while reactor 7 harboured almost equal amounts of Methanoculleus and Methanosarcina (Fig. 2b). The ordination of the archaeal bi-plot was supported by the low final stress value of 1.4, and the Monte Carlo test showing that the null hypothesis of no treatment effects could be rejected (P o 0.02, 50 permutations). A mantel test revealed that the variation in the composition of bacterial and archaeal communities among the reactors was not correlated, as indicated by the low standardized Mantel statistic (rMantel = 0.018) and the high P value (P 4 0.05). Microbial communities were also tested for correlations with the trace element additions using PLS. The 11 most frequent bacterial T-RFLP HaeIII fragments (Fig. 2a) and FEMS Microbiol Ecol 74 (2010) 226–240

the three most frequent archaeal HhaI fragments (Fig. 2b) were used in the analyses. Correlations were found with bacterial fragments HaeIII-239 (uncultured Firmicutes bacterium) and HaeIII-282 (group within the family Clostridiaceae) and with all the archaeal fragments tested. Significant models were constructed for these fragments. The models revealed that both bacterial fragments were positively correlated to TEM1 (Fig. 4a and b and Table 3) and negatively to TEM1  Co (Table 3). However, the relative abundance of the bacterial fragment HaeIII-237, which was the dominant bacterial population in all reactors (Fig. 2a), was not significantly correlated to the input variables. For archaea, fragment HhaI-53 was positively correlated to TEM2 and fragment HhaI-92 was positively correlated to TEM1, whereas fragment HhaI-113 was negatively correlated to the latter trace metal addition (Fig. 5a–c; Table 3). Furthermore, fragment HhaI-53 was negatively correlated to the term TEM1  TEM2 (Table 3). ANOVA tables for all the models are given (Tables S2–S6).

Discussion Process performance All reactors (except reactor 8, which was excluded for the reasons stated above) had well-functioning biogas processes as demonstrated by the methane production values and the low final levels of VFAs (Table 2). The methane production 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

7 2

234

X.M. Feng et al.

(a)

(a)

Haelll 239 (%) 1.2 0.9 2.4 2.1 1.8 1.5

Hha l 53 (%) 1.8 1.5 1.2 0.9 0.6 0.3 0

0.9 1.2 1.5 1.8 2.1 2.4

7

7

0.9 1.2 1.5 1.8 2.1 2.4

30 30

6 25

6

25

5 20

5

4

4

20 15

3

15 10

3

2

2

1

1

0.9 Co 10 log 1.2 1.5 1.8 2.1 2.4

5 5 0

0.9

TEM1 10 log

10

1.2

1.5

1.8 2.1 2.4

0.9 1.2 1.5 1.8

0.9

1.8

1.2

2.1

1.5

2.4

(b)

Haelll 282 (%) 1.51.20.9 2.4 2.11.8

1.5

1.8

Hhal 92 (%) 0.6 0.3 0 1.8 1.5 1.2 0.9

2.1 2.4 14

0.9 1.2 1.5 1.8 2.1 2.4 90

90

80

14 80

12

70

70

12

10

10

8 6

8

4 0.2 Co 10 log 1.2 1.5 1.8 2.1 2.4

6

TEM1 10 log

0.9

1.2

1.5

1.8

2.1

60

50

50

40

40

30

30

20 10 0.3 TEM1 10 log 0.6 0.9 1.2 1.5 1.8

20 10 TEM2 10 log 0.9 1.2

1.5

1.8

2.1 2.4

2.4

Fig. 4. The effect of the trace element additions on bacterial T-RFs (a) HaeIII 239 and (b) HaeIII 282 (% of the total abundance in the final samples) given by the best regression model obtained in MODDE 6. The values for TEM1 and Co given on the axis are 10 log values of the multiples from the AGM; see Table 1 for the actual values. Table 3. Significant models of the relative abundance of restriction fragments for bacteria (HaeIII) and archaea (HhaI) correlating to the trace element additions T-RFLP fm. R2

Q2

Significant coeffs.

M sign. LoF

HaeIII-239 HaeIII-282 HhaI-53 HhaI-92 HhaI-113

0.62 0.67 0.41 0.68 0.61

TEM11, TEM1  Co  TEM11, TEM1  Co  TEM21, TEM1  TEM2  TEM11 TEM1 

0.009 0.025 0.031 0.005 0.002

0.79 0.72 0.87 0.73 0.66

60

0.622 0.976 0.615 0.900 0.794

(c)

Hhal 113 (%) 1.2 2.4 2.1 1.8 1.5

0.9

0.9 1.2 1.5 1.8 2.1 2.4 35

35

30

30

25

25

20

20

15

15

10 5

10

0

5 0 TEM1 10 log

0.9

1.2

1.5

1.8

2.1 2.4

0.9 Co 10 log 1.2 1.5 1.8 2.1 2.4

Model significance (M sign.) and lack of fit (LoF) test values are given. Model significance is obtained at P values below 0.05 and ‘no lack of fit’ is obtained if the model has P 4 0.05 for the LoF test. 1 or , positive or negative influence on the abundance of the fragment, respectively.

Fig. 5. The effect of the trace element additions on archaeal T-RFs (a) HhaI 53 and (b) HhaI 92 and (c) HhaI 113 (% of the total abundance in the final samples) given by the best regression model obtained in MODDE 6. The values for TEM1, TEM2 and Co given on the axis are 10 log values of the multiples from the AGM; see Table 1 for the actual values.

values obtained are in the range of the methane production potential for the organic fractions of municipal solid waste in codigestion with fat, restaurant waste or grease trap

material (400–800 mL CH4 g1 VS; Luostarinen et al., 2008; Neves et al., 2009). Hence, the methanogenic community was functioning well, with a minor accumulation of

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

(b)

TEM2 10 log 0.91.2

0 0.3 Co 10 log 0.6

235

Biogas production from food industrial waste

intermediate fermentation products. However, due to foaming, the maximum loading rate was limited to 4 g VS 1 day1. The lack of correlation between methane formation and VS reduction (PLS analysis) was likely due to the occurrence of foaming that resulted in an underestimation of the TS content of the reactor liquids, which in turn affects the VS determination. In later studies, the foaming layers have been shown to have a TS content of up to 14%, while it was only 4% in the liquid (unpublished data).

systems is unknown, but the drastic increase in abundance compared with the occurrence in the starting material (Fig. 2a; 2–4%) may be related to the difference in HRT between the full-scale biogas process and our laboratoryscale reactors (40–50 days compared with 25 days). The PLS analysis revealed that the two less abundant fragments HaeIII-239 (uncultured Firmicutes bacterium) and HaeIII-282 (groups within Clostridiaceae) were positively correlated to TEM1 (Fig. 4; Table 3). However, the resolution does not allow for a deduction of their functions in the reactors.

Methane production vs. trace element additions

Microbial community composition The bacterial community The trace element additions did not seem to affect the composition of the combined bacterial community in the reactors; instead, a difference between the two sets of starting materials seems to have governed the distribution (Fig. 3a). This indicates that there is a shift in the microbial community of the full-scale process over time. This is in line with earlier observations showing that biogas reactor microbial communities are very dynamic even in well-functioning processes (Ferna´ ndez et al., 1999 and discussed by Briones & Raskin 2003). The low effect of the trace element levels added was further demonstrated by the divergent community composition in the replicate central-point reactors 1 and 6 from set1, and reactors 15 and 18 from set2 (Fig. 3a). Thus, the change in the bacterial community composition over time in the full-scale reactor seems to override the effect of the trace element supplies. According to the individual fragments, the bacterial communities in the lab-scale reactors were dominated by HaeIII-237, contributing 32–55% of the total bacterial 16S rRNA gene amplicons (Fig. 2a), but the representation of this population was not correlated to the trace element additions according to the PLS analysis. Clones representing this fragment were most closely related to A. europaeus (94% similarity). Actinomyces sp. have frequently been isolated from humans and other warm-blooded animals (Funke et al., 1997), but they also occur in anaerobic sludge (Oh et al., 2003). The reason for its dominance in the present FEMS Microbiol Ecol 74 (2010) 226–240

The archaeal community The maintenance of the methanogenic activity was largely independent of archaeal community composition, pointing to functional redundancy in the community, with alternate archaeal populations being able to maintain the methane production rates. Species of the genus Methanosarcina are common in biogas processes (Godon et al., 1997; Fern´andez et al., 2000; McMahon et al., 2001; Mladenovska et al., 2003; Shigematsu et al., 2003; McHugh et al., 2004). Methanosarcina siciliae was reported to dominate in a mesophilic process treating cattle manure and glycerol trioleate (Mladenovska et al., 2003). The growth of M. siciliae was originally described as being restricted to methanol, methylamine and dimethylsulphide as substrates (Ni et al., 1994), but strains of this species have later also been shown to utilize acetate for growth (Elberson & Sowers, 1997). Hence, we suggest that the Methanosarcina population identified here is responsible for the acetatoclastic methanogenesis in the bioreactors studied. Methanoculleus bourgensis has also been previously identified in biogas processes (Hori et al., 2006; Klocke et al., 2008; Krause et al., 2008; Schl¨uter et al., 2008; Kr¨ober et al., 2009). Methanoculleus has been shown to be dominant in a biogas process containing high levels of ammonium and VFA, where acetate is degraded via syntrophic acetate oxidation (SAO; Schn¨urer et al., 1999). Methanoculleus bourgensis utilizes H2/ CO2 or formate for growth (Maestrojuan et al., 1990). In the present study, the ratio of archaea to bacteria was shown to be positively correlated to the abundance of HhaI92 (M. siciliae) and negatively correlated to the abundance of HhaI-531HhaI-113 (M. bourgensis). The positive correlation implies that archaea are quantitatively more significant, when there is a high abundance of Methanosarcina. This can be explained by the shift in methane production routes, i.e. under aceticlastic conditions, the main part of the carbon metabolized in a reactor finally ends up as acetate and will support the growth of acetoclastic methanogens such as Methanosarcina. During SAO, hydrogen is the main substrate for the methanogens. This most likely means a smaller archaeal population. A situation with SAO being more active 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

The highest predicted methane production occurred at high levels of TEM2 (Se/W) combined with a low level of Co. However, the model also showed that the maximum methane production from the substrate was not reached within the concentration interval of the trace elements. Thus, a higher supply of TEM2 (Se/W) may further improve the methane production (Fig. 1). The possible reasons for the observed effects of the trace element additions are further discussed in connection to the bacterial community analyses below.

236

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

nate degradation have remarkably high formate-oxidation and CO2-reduction rates as well as low Km values for formate. This implies that the W addition to our reactors may have stimulated formate turnover, which in turn would affect the degradation efficiencies of more complex substrates (Boone et al., 1989). Se is essential for enzymes active in the anaerobic degradation chain (e.g. formate dehydrogenase; cf. Andreesen & Makdessi, 2008), glycine reductase (Clostridium; reviewed by Stadtman, 2002) and a Ni-Fe-Se-hydrogenase (reviewed by Stolz et al., 2006). The effects of Se and Won microbial activity in biogas reactor systems have scarcely been reported; three studies showed a positive influence of Se or W additions as parts of trace element cocktails, but the role of Se or W per se was not evaluated (Wilkie et al., 1986; Osuna et al., 2003; Gung¨or-Demirci & Demirer 2004). Worm et al. (2009) studied the effect of deprivation of Mo, Se and W on a UASB-microbial community feed with propionate and observed a decrease in methanogenic activity over time. These authors suggested that a shift in propionate-degrading flora was due to low levels of Mo and W. The positive correlation between TEM1 and the abundance of HhaI-92 (M. siciliae; Fig. 5b and c; Table 3) shown by the PLS analysis is in accordance with the roles of the trace elements in this mixture (B, Mo, Ni). Ni is essential for the activity of carbon monoxide dehydrogenase (central in the Wood–Ljungdahl pathway as reviewed by Ragsdale & Pierce, 2008), and Mo has been identified as present in enzymes and cofactors of Methanosarcina acetovorans (Li et al., 2007). Thus, the availability of Mo will likely also affect the growth of other Methanosarcina sp. We have no explanation for the negative effects by TEM1 alone or in combination with TEM2 on HhaI-53 and HhaI-113. The lack of a response in methane production or the microbial composition to the supply of Co was surprising as the metal is essential in key enzymes of the methanogenic pathways (Roth et al., 1996; Zandvoort et al., 2006c). Earlier studies on the effects of Co on methanogenic systems have shown an increase in methane production over a broad concentration range (0.03–25 mg Co L1; Jarvis et al., 1997; Kida et al., 2001; Patidar & Tare, 2006; Zandvoort et al., 2006a; Zitomer et al., 2008, among others). The fact that additions as low as 0.03 mg Co L1 may enhance the process performance suggests that even the lowest addition of Co in the present study (0.06 mg L1) could potentially have a positive influence on methanogenesis.

Conclusions We observed moderate, but significant effects of trace element additions on the methane production efficiency, i.e. 7–15% increase as a result of the presence of a Se/W supplement. The main bacterial fragment found in the laboratory-scale study (HaeIII-237, Actinomyces) did not FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

in systems with a low archaea to bacteria ratio is further supported by the low archaea to bacteria ratio of the bioreactor starting materials (0.5% and 0.3% for sets 1 and 2, respectively) combined with low HhaI-92 values (22% of the total archaea) as the full-scale process is dominated by SOA (Schn¨urer & Nordberg, 2008). The primary role of SAO as the main acetate degradation route in the full-scale biogas process, from which the starting material and substrate for the laboratory-scale reactors were obtained, is likely explained by a combination of high ammonia levels and long HRT (40–50 days). High ammonia levels have repeatedly been shown to be toxic for the acetate-utilizing methanogens (Koster & Lettinga, 1984; Angelidaki & Ahring, 1994; Gallert et al., 1998; Hansen et al., 1998). The full-scale biogas reactor 1 (our sampled for this study had NH1 4 -N levels of 5 g L own data and reported by Schn¨urer & Nordberg, 2008), while the final concentrations in our laboratory study ranged from 2.3 to 3.0 g L1 (data not shown). The lower levels in the laboratory study are probably at least partly related to the higher dilution rate in these reactors. Nevertheless, the dominance shift from Methanosarcina to Methanoculleus also correlated with the trace element additions, for example Methanoculleus dominance (HhaI-53 or HhaI-113) correlated negatively to TEM1 or a combination of TEM1 and TEM2, while Methanosarcina (HhaI-92) was positively influenced by TEM1 (Fig. 2b; Table 3). The influence of trace elements was also seen in the NMS analysis; archaeal communities in reactors 13, 2 and 7 were clearly different from the rest of the reactors and had the common denominator TEM1 that was in low supply. The fact that they do not cluster together might be due to the combination effects of other trace element additions (TEM2 and Co) provided in different amounts for these three reactors (Table 2). The PLS analysis also suggests that TEM2 additions had a positive influence on the archaeal population represented by T-RF HhaI-53, while a two-factor interaction was shown between TEM1 and TEM2 (Table 3); thus, HhaI-53 was more frequent at high levels of TEM2 when the supply of TEM1 was low. Although both HhaI-53 and HhaI-113 were affiliated with the hydrogenotrophic methanogen M. bourgensis, it should be noted that the trace element addition seemed to affect the two fragments differently (Table 3), indicating differences in the enzyme systems. A possible explanation for the observed positive effect of TEM2 on HhaI-53 and on the methane production (Fig. 1) is the role of W in for example dehydrogenases. W is a part of the active centre in formylmethanofuran-dehydrogenase, which performs the first step in the autotrophic fixation of CO2 in methanogens (reviewed by Zandvoort et al., 2006c). The metal can also play a role in syntrophic propionate degradation, replacing Mo in formate dehydrogenases (De Bok et al., 2003; Ragsdale & Pierce 2008). De Bok et al. (2003) showed that two formate dehydrogenases involved in propio-

X.M. Feng et al.

237

Biogas production from food industrial waste

show any correlation to the trace element additions, but for fragments HaeIII-239 and HaeIII-282 (Clostridiaceae), a positive effect of TEM1 was evident (Table 3 and Fig. 4). For archaea, all three dominant T-RFs were significantly correlated to the trace element additions (Table 3 and Figs 3b and 5). This suggests that archaea in the bioreactors are more sensitive to trace element levels compared with the bacterial community members. Overall, our study indicates that there are effects of trace element additions on both methane production and microbial community compositions, although no direct correlation between methane production and microbial communities’ compositions was found, indicating a high degree of functional redundancy with regard to this specific process.

We thank Eva Linstr¨om and Janet Andert for statistical advice. This study was supported by the Swedish Energy Agency, the Swedish Research Council FORMAS and the Swedish Research Council (grant to S.B.).

Authors’contribution X.M.F. and A.K. have contributed equally to this work.

References Andreesen JR & Makdessi K (2008) Tungsten, the surprisingly positively acting heavy metal element for prokaryotes. Ann NY Acad Sci 1125: 215–229. Angelidaki I & Ahring BK (1994) Anaerobic thermophilic digestion of manure at different ammonia loads: effect of temperature. Water Res 28: 727–731. Asakawa S & Nagaoka K (2003) Methanoculleus bourgensis, Methanoculleus olentangyi and Methanoculleus oldenburgensis are subjective synonyms. Int J Syst Evol Micr 53: 1551–1552. Boone DR, Johnson RL & Liu Y (1989) Diffusion of the interspecies electron carriers H2 and formate in methanogenic ecosystems and its implication in the measurement of Km for H2 or formate uptake. Appl Environ Microb 55: 1735–1741. B¨orjesson P & Mattiasson B (2008) Biogas as a resource-efficient vehicle fuel. Trends Biotechnol 26: 7–13. Briones A & Raskin L (2003) Diversity and dynamics of microbial communities in engineered environments and their implications for process stability. Curr Opin Biotech 14: 270–276. Casamayor EO, Massana R, Benlloch S, Øver˚as L, D´ıez B, ´ Goddard VJ, Gasol JM, Joint I, Rodr´ıguez-Valera F & PedrosAlio´ C (2002) Changes in archaeal, bacterial and eukaryal assemblages along a salinity gradient by comparison of genetic fingerprinting methods in a multipond solar saltern. Environ Microbiol 4: 338–348.

FEMS Microbiol Ecol 74 (2010) 226–240

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Acknowledgements

Clarke KR (1999) Nonmetric multivariate analysis in community-level ecotoxicology. Environ Toxicol Chem 18: 118–127. Culman SW, Gauch HG, Blackwood CB & Thies JE (2008) Analysis of T-RFLP data using analysis of variance and ordination methods: a comparative study. J Microbiol Meth 75: 55–63. De Bok FAM, Hagedoorn P-L, Silva PJ, Hagen WR, Schiltz E, Fritsche K & Stams AJM (2003) Two W-containing formate dehydrogenases (CO2-reductases) involved in syntrophic propionate oxidation by Syntrophobacter fumaroxidans. Eur J Biochem 270: 2476–2485. Eiler A & Bertilsson S (2004) Composition of freshwater bacterial communities associated with cyanobacterial blooms in four Swedish lakes. Environ Microbiol 6: 1228–1243. Eiler A & Bertilsson S (2007) Flavobacteria blooms in four eutrophic lakes: linking population dynamics of freshwater bacterioplankton to resource availability. Appl Environ Microb 73: 3511–3518. Elberson MA & Sowers KR (1997) Isolation of an aceticlastic strain of Methanosarcina siciliae from marine canyon sediments and emendation of the species description for Methanosarcina siciliae. Int J Syst Bacteriol 47: 1258–1261. Eriksson L, Johansson N, Kettaneh-Wold N, Wikstr¨om C & Wold S (2000) Design of Experiments – Principles and Applications. Learnways AB, Ume˚a. Fathepure BZ (1987) Factors affecting the methanogenic activity of Methanothrix soehngenii VNBF. Appl Environ Microb 53: 2978–2982. Fermoso G, Collins G, Bartacek J, O’Flaherty V & Lens P (2008) Acidification of methanol-fed anaerobic granular sludge bioreactors by cobalt deprivation: induction and microbial community dynamics. Biotechnol Bioeng 99: 49–58. Fern´andez A, Huang S, Seston S, Xing J, Hickey R, Criddle C & Tiedje JM (1999) How stable is stable? Function versus community composition. Appl Environ Microb 65: 3697–3704. Fern´andez AS, Hashsham SA, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey RF, Criddel CS & Tiedje JM (2000) Flexible community structure correlates with stable community structure function in methanogenic bioreactor communities perturbed by glucose. Appl Environ Microb 66: 4058–4067. Funke G, Alvarez N, Pascual C, Falsen E, Akervall E, Sabbe L, Schouls L, Weiss N & Collins MD (1997) Actinomyces europaeus sp. nov., isolated from human clinical specimens. Int J Syst Bacteriol 47: 687–692. Gallert C, Bauer S & Winter J (1998) Effect of ammonia on the anaerobic degradation of protein by a mesophilic and thermophilic biowaste population. Appl Microbiol Biot 50: 495–501. Godon J-J, Zumstein E, Dabert P, Habouzit F & Moletta R (1997) Molecular microbial diversity of an anaerobic digester as determined by small-subunit rDNA sequence analysis. Appl Environ Microb 63: 2802–2813.

238

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

microarray analysis of the archaeon Methanosarcina acetivorans grown with acetate versus methanol. J Proteome Res 6: 759–771. Li T, Maz´eras L, Sghir A, Leblon G & Bouchez T (2009) Insights into networks of functional microbes catalysing methanization of cellulose under mesophilic conditions. Environ Microbiol 11: 889–904. Luostarinen S, Luste S & Sillanp¨aa¨ M (2008) Increased biogas production at wastewater treatment plants through codigestion of sewage sludge with grease trap sludge from a meat processing plant. Bioresource Technol 100: 79–85. Maestrojuan GM, Boone DR, Xun L, Mah RA & Zhang L (1990) Transfer of Methanogenium bourgense, Methanogenium marisnigri, Methanogenium olentangyi, and Methanogenium thermophilicum to the genus Methanoculleus gen. nov., emendation of Methanoculleus marisnigri and Methanogenium, and description of new strains of Methanoculleus bourgense and Methanoculleus marisnigri. Int J Syst Bacteriol 40: 117–122. Mantel N (1967) The detection of disease clustering and a generalised regression approach. Cancer Res 27: 209–220. McCune B & Grace JB (2002) Nonmetric multidimensional scaling. Analysis of Ecological Communities, pp. 125–142. Gleneden Beach, Oregon. McHugh S, Carton M, Mahony T & O’Flaherty V (2003) Methanogenic population structure in a variety of anaerobic bioreactors. FEMS Microbiol Lett 219: 297–304. McHugh S, Carton M, Collins G & O’Flaherty V (2004) Reactor performance and microbial community dynamics during anaerobic biological treatment of wastewaters at 16–37 1C. FEMS Microbiol Ecol 48: 369–378. McMahon KD, Stroot PG, Mackie RI & Raskin L (2001) Anaerobic codigestion of municipal waste and biosolids under various mixing conditions – II. Microbial population dynamics. Water Res 35: 1817–1827. McMahon KD, Zheng D, Stams AJM, Mackie RI & Raskin L (2004) Microbial population dynamics during start-up and overload conditions of anaerobic digesters treating municipal solid waste and sewage sludge. Biotechnol Bioeng 87: 823–834. Mladenovska Z, Dabrowski S & Ahring BK (2003) Anaerobic digestion of manure and mixture of manure with lipids: biogas reactor performance and microbial community analysis. Water Sci Technol 48: 271–278. Murray WD & van den Berg L (1981) Effects of nickel, cobalt, and molybdenum on performance of methanogenic fixed-film reactors. Appl Environ Microb 42: 502–505. Nettmann E, Bergmann I, Mundt K, Linke B & Klocke M (2008) Archaea diversity within a commercial biogas plant utilizing herbal biomass determined by 16S rDNA and mcrA analysis. J Appl Microbiol 105: 1835–1850. Neves L, Pereira MA, Mota M & Alves MM (2009) Detection and quantification of long chain fatty acids in liquid and solid samples and its relevance to understand anaerobic digestion of lipids. Bioresource Technol 100: 91–96. Ni S, Woese CR, Aldrich HC & Boone DR (1994) Transfer of Methanolobus siciliae to the genus Methanosarcina, naming it

FEMS Microbiol Ecol 74 (2010) 226–240

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Gung¨or-Demirci G & Demirer GN (2004) Effect of initial COD concentration, nutrient addition, temperature and microbial acclimation on anaerobic treatability of broiler and cattle manure. Bioresource Technol 93: 109–117. Hansen KH, Angelidaki I & Ahring BK (1998) Anaerobic digestion of swine manure: inhibition by ammonia. Water Res 32: 5–12. Held J, Mathiasson A & Nylander A (2008) Biogas from Manure and Waste – Swedish Case Studies. The Swedish Biogas Association. Available at http://www.sbgf.info/filer/ Biogasinfo_english.pdf. Hori T, Haruta S, Ueno Y, Ishii M & Igarashi Y (2006) Dynamic transition of a methanogenic population in response to the concentration of volatile fatty acids in a thermophilic anaerobic digester. Appl Environ Microb 72: 1623–1630. Jarvis A˚, Nordberg A˚, Jarlsvik T, Mathisen B & Svensson BH (1997) Improvement of a grass-clover silage-fed biogas process by the addition of cobalt. Biomass Bioenerg 12: 453–460. Jonsson S & Bor´en H (2002) Analysis of mono- and diesters of ophthalic acid by solid-phase extractions with polystyrenedivinylbenzene-based polymers. J Chromatogr A 963: 393–400. Karlsson A, Ejlertsson J, Nezirevic D & Svensson BH (1999) Degradation of phenol under meso- and thermophilic, anaerobic conditions. Anaerobe 5: 25–35. Kida K, Shigematsu T, Kijima J, Numaguchi M, Mochinaga Y, Abe N & Morimura S (2001) Influence of Ni21 and Co21 on methanogenic activity and the amounts of coenzymes involved in methanogenesis. J Biosci Bioeng 91: 590–595. Kim M, Ahn Y-H & Speece RE (2002) Comparative process stability and efficiency of anaerobic digestion; mesophilic vs. thermophilic. Water Res 36: 4369–4385. Klocke M, M¨ahnert P, Mundt K, Souidi K & Linke B (2007) Microbial community analysis of a biogas-producing completely stirred tank reactor fed continuously with fodder beet silage as mono-substrate. Syst Appl Microbiol 30: 139–151. Klocke M, Nettmann E, Bergmann I, Mundt K, Souidi K, Mumme J & Linke B (2008) Characterization of the methanogenic archaea within two-phase biogas reactor systems operated with plant biomass. Syst Appl Microbiol 31: 190–205. Koster IW & Lettinga G (1984) The influence of ammonianitrogen on the specific activity of pelletized methanogenic sludge. Agr Wastes 9: 205–216. Krause L, Diaz NN, Edwards RA et al. (2008) Taxonomic composition and gene content of a methane-producing microbial community isolated from a biogas reactor. J Biotechnol 136: 91–101. Kr¨ober M, Bekel T, Diaz NN et al. (2009) Phylogenetic characterization of a biogas plant microbial community integrating clone library 16S rDNA sequences and metagenome sequence data obtained by 454-pyrosequencing. J Biotechnol 142: 38–49. Li L, Li Q, Rohlin L, Kim U, Salmon K, Rejtar T, Gunsalus RP, Karger BL & Ferry JG (2007) Quantitative proteomic and

X.M. Feng et al.

239

Biogas production from food industrial waste

FEMS Microbiol Ecol 74 (2010) 226–240

on structure of a mesophilic acetate-degrading methanogenic community during continuous cultivation. J Biosci Bioeng 96: 547–558. Stadtman TC (2002) Discoveries of vitamin B12 and selenium enzymes. Annu Rev Biochem 71: 1–16. Stolz JF, Basu P, Santini JM & Oremland RS (2006) Arsenic and selenium in microbial metabolism. Annu Rev Microbiol 60: 107–130. Takashima M & Speece RE (1989a) Mineral requirements for methane fermentation. Crit Rev Env Contr 19: 465–479. Takashima M & Speece RE (1989b) Mineral nutrient requirements for high-rate methane fermentation of acetate at low SRT. Res J Water Pollut C 61: 1645–1650. Wilkie A, Goto M, Bordeaux FM & Smith PH (1986) Enhancement of anaerobic methanogenesis from napiergrass by addition of micronutrients. Biomass 11: 135–146. Wolin EA, Wolin MJ & Wolfe RS (1963) Formation of methane by bacterial extracts. J Biol Chem 238: 2882–2886. Worm P, Fermoso FG, Lens PNL & Plugge CM (2009) Decreased activity of a propionate degrading community in a UASB reactor fed with synthetic medium without molybdenum, tungsten and selenium. Enzyme Microb Tech 45: 139–145. Yu C-P, Ahuja R, Sayler G & Chu K-H (2005) Quantitative molecular assay for fingerprinting microbial communities of wastewater and estrogen-degrading consortia. Appl Environ Microb 71: 1433–1444. Zandvoort MH, Hullebusch ED, Gieteling J & Lens PNL (2006a) Granular sludge in full-scale anaerobic bioreactors: trace element content and deficiencies. Enzyme Microb Tech 39: 337–346. Zandvoort MH, Hullebusch ED, Golubnic S, Gieteling J & Lens PNL (2006b) Induction of cobalt limitation in methanol feed UASB reactors. J Chem Technol Biot 81: 1486–1495. Zandvoort MH, van Hullebusch ED, Fermoso FG & Lens PNL (2006c) Trace metals in anaerobic granular sludge reactors: bioavailability and dosing strategies. Eng Life Sci 6: 293–301. Zhang Y, Yyan L, Chi L, Long X, Mei Z & Zhang Z (2008) Startup and operation of anaerobic EGSB reactor treating palm oil mill effluent. J Environ Sci 20: 658–663. Zitomer DH, Johnson CC & Speece RE (2008) Metal stimulation and municipal digester. Thermophilic/mesophilic activity. J Environ Eng 134: 42–47. Zumstein E, Moletta R & Godon J-J (2000) Examination of two years of community dynamics in an anaerobic bioreactor using fluorescence polymerase chain reaction (PCR) singlestrand conformation polymorphism analysis. Environ Microbiol 2: 69–78.

Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. Results from the ANOVA of the quadratic polynomial regression model for the methane production data. 2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

Methanosarcina siciliae, and emendation of the genus Methanosarcina. Int J Syst Bacteriol 44: 357–359. Noyola A & Tinajero A (2005) Effect of biological additives and micronutrients on the anaerobic digestion of physicochemical sludge. Water Sci Technol 55: 275–281. Oh S-E, van Ginkel S & Logan BE (2003) The relative effectiveness of pH control and heat treatment for enhancing biohydrogen gas production. Environ Sci Technol 37: 5186–5190. Osborne CA, Rees GN, Bernstein Y & Janssen PH (2006) New threshold and confidence estimates for terminal restriction fragment length polymorphism analysis of complex bacterial communities. Appl Environ Microb 72: 1270–1278. Osuna MB, Zandvoort MH, Iza JM, Lettinga G & Lens PNL (2003) Effects of trace element addition on volatile fatty acid conversions in anaerobic granular sludge reactors. Environ Technol 24: 573–587. Patidar SK & Tare V (2006) Effect of nutrients on biomass activity in degradation of sulfate laden organics. Process Biochem 41: 489–495. Pobeheim H, Munk B, Johansson J & Guebitz GM (2010) Influence of trace elements on methane formation from a synthetic model substrate for maize silage. Bioresource Technol 101: 836–839. Ragsdale SW & Pierce E (2008) Acetogenesis and the Wood–Ljungdahl pathway of CO2 fixation. BBA-Proteins Proteom 1784: 1873–1898. Rees GN, Baldwin DS, Watson GO, Perryman S & Nielsen DL (2005) Ordination and significance testing of microbial community composition derived from terminal restriction fragment length polymorphisms: application of multivariate statistics. Anton Leeuw Int J G 86: 339–347. Roth JR, Lawrence JG & Bobik TA (1996) Cobalamin (coenzyme B12): synthesis and biological significance. Annu Rev Microbiol 50: 137–181. Schl¨uter A, Bekel T, Diaz NN et al. (2008) The metagenome of a biogas-producing microbial community of a production-scale biogas plant fermenter analysed by the 454-pyrosequencing technology. J Biotechnol 136: 77–90. Schn¨urer A & Nordberg A˚ (2008) Ammonia, a selective agent for methane production by syntrophic acetate oxidation at mesophilic temperature. Water Sci Technol 57: 735–740. Schn¨urer A, Zeller G & Svensson BH (1999) Mesophilic syntrophic acetate oxidation during methane formation in biogas reactors. FEMS Microbiol Ecol 29: 249–261. Sch¨onheit P, Moll J & Thauer RK (1979) Nickel, cobalt and molybdenum requirement for growth of Methanobacterium thermoautotrophicum. Arch Microbiol 123: 105–107. Scully C, Collins G & O’Flaherty V (2005) Assessment of anaerobic wastewater treatment failure using terminal restriction fragment length polymorphism analysis. J Appl Microbiol 99: 1463–1471. Shigematsu T, Tang Y, Kawaguchi H, Ninomiya K, Kijima J, Kobayashi T, Morimura S & da K (2003) Effect of dilution rate

240

Table S2. Results from the ANOVA of the model for archaeal T-RF HhaI 53. Table S3. Results from the ANOVA of the model for archaeal T-RF HhaI 92. Table S4. Results from the ANOVA of the model for archaeal T-RF HhaI 113. Table S5. Results from the ANOVA of the model for bacterial T-RF HaeIII 239.

X.M. Feng et al.

Table S6. Results from the ANOVA of the model for bacterial T-RF HaeIII 282. Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Downloaded from http://femsec.oxfordjournals.org/ by guest on January 15, 2017

2010 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

 c

FEMS Microbiol Ecol 74 (2010) 226–240