Effects of glucose overloading on microbial community structure and biogas production in a laboratory-scale anaerobic digester

Download Effects of glucose overloading on microbial community structure and biogas production in a laboratory-scale anaerobic digester

Post on 05-Jul-2016




3 download

Embed Size (px)


<ul><li><p>Eects of glucose overloading on microbial community structureand biogas production in a laboratory-scale anaerobic digester</p><p>Ingvar Sundh a,*, Helena Carlsson a, AAke Nordberg b, Mikael Hansson b, Berit Mathisen b</p><p>a Department of Microbiology, Swedish University of Agricultural Sciences, P.O. Box 7025, SE-750 07 Uppsala, Swedenb Swedish Institute of Agricultural and Environmental Engineering, P.O. Box 7033, SE-750 07 Uppsala, Sweden</p><p>Received 8 October 2002; received in revised form 10 February 2003; accepted 25 February 2003</p><p>Abstract</p><p>This study characterizes the response of the microbial communities of a laboratory-scale mesophilic biogas process, fed with a</p><p>synthetic substrate based on cellulose and egg albumin, to single pulses of glucose overloading (15 or 25 times the daily feed based on</p><p>VS). The microbial biomass and community structure were determined from analyses of membrane phospholipids. The ratio be-</p><p>tween phospholipid fatty acids (PLFAs; eubacteria and eucaryotes) and di-ethers (PLEL; archaea) suggested that methanogens</p><p>constituted 48% of the microbial biomass. The glucose addition resulted in transient increases in the total biomass of eubacteria</p><p>while there were only small changes in community structure. The total gas production rate increased, while the relative methane</p><p>content of the biogas and the alkalinity decreased. However, the biomass of methanogens was not aected by the glucose addition.</p><p>The results show that the microbial communities of biogas processes can respond quickly to changes in the feeding rate. The glucose</p><p>overload resulted in a transient general stimulation of degradation rates and almost a doubling of eubacterial biomass, although the</p><p>biomass increase corresponded to only 7% of the glucose C added.</p><p> 2003 Elsevier Science Ltd. All rights reserved.</p><p>Keywords: Biogas; Microbial community structure; PLFA; Substrate overload; Methanogenesis; Di-ether lipid</p><p>1. Introduction</p><p>In biogas production from organic materials, the in-</p><p>tegrated action of several types of microorganisms,</p><p>which perform dierent degradation steps, results in the</p><p>sequential degradation of polymeric carbohydrates,</p><p>proteins and fats (Gujer and Zehnder, 1983; Schink,</p><p>1988). Functionally, the organisms can be broadly</p><p>classied into hydrolytic, fermentative organic acid-producing, acetate-producing and, in the terminal step,</p><p>methane-producing, organisms.</p><p>Due to low energy yields, many anaerobic microor-</p><p>ganisms grow slowly, the methanogens in particular.</p><p>Therefore, in completely mixed systems with relatively</p><p>short retention times, disturbance due to, for example,</p><p>poor composition or overloading of the substrate, ac-</p><p>cumulation of inhibitory substances, or presence of an-thropogenic contaminants, may inhibit the active</p><p>organisms and result in serious malfunctioning of the</p><p>process. However, knowledge of the response of themicrobial communities to dierent kinds of disturbance</p><p>is still meagre, particularly regarding the communitiesstructural characteristics.</p><p>In the present study, we investigated the eects of</p><p>substrate overload, with a single dose of glucose, on the</p><p>size and structure of the microbial community and on</p><p>the biogas production in a laboratory scale biogas pro-</p><p>cess fed with a synthetic substrate. The microbial bio-mass and community structure were determined from</p><p>analysis of membrane phospholipids. In addition, the</p><p>fate of the added glucose was determined by construc-</p><p>tion of a C budget.</p><p>2. Methods</p><p>2.1. Reactor system</p><p>The reactor was a continuously mixed glass tank of8 l active volume, operated under mesophilic condi-</p><p>tions (37 C). It was started with digester contents froma larger reactor which had operated with the same</p><p>*Corresponding author. Tel.: +46-18-673210; fax: +46-18-673392.</p><p>E-mail address: ingvar.sundh@mikrob.slu.se (I. Sundh).</p><p>0960-8524/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.doi:10.1016/S0960-8524(03)00075-0</p><p>Bioresource Technology 89 (2003) 237243</p></li><li><p>substrate for three years. The substrate consisted of a</p><p>mixture of microcrystalline cellulose, egg albumin and</p><p>vitamins in a mineral medium (Nordberg et al., 2000).</p><p>The experimental reactor was automatically fed every</p><p>12th h and the hydraulic retention time was 20 d. The</p><p>regular feeding corresponded to 1.0 gVS l1 d1.Two experiments with glucose additions were carried</p><p>out, separated by ve months, where extra glucose wasadded representing approximately 15 (Experiment B)</p><p>and 25 (Experiment C) times (in gVS l1 d1) the normaldaily feeding rate, with cellulose and albumin. The</p><p>glucose was added as a single dose, which replaced one</p><p>regular feeding, after which normal feeding rates were</p><p>resumed. A second reactor operated in an identical</p><p>manner served as a control and did not receive glucose</p><p>at any stage.</p><p>2.2. Microbial biomass and community structure</p><p>The microbial communities were characterized fromanalyses of normal phospholipid fatty acids (PLFAs)</p><p>from eucaryotes and bacteria, and of phospholipid die-</p><p>ther lipids (PLELs) from methanogenic archae. Samples</p><p>for lipid analysis were collected immediately before and</p><p>after the glucose addition, after 4 h, and then at intervals</p><p>during a 20 day period. 2-ml aliquots of the reactor</p><p>contents were immediately frozen in extraction tubes</p><p>and stored at )20 C until analysis. Detailed descri-ptions of the lipid extractions have been given previously</p><p>(Sundh et al., 1997). Briey, the total lipid fraction of</p><p>the reactor contents was extracted in a one-phase chlo-</p><p>roform/methanol/water extraction. The lipids were then</p><p>fractionated with silicic acid chromatography. The polar</p><p>fraction (containing the phospholipids) was subjected to</p><p>an alkaline methanolysis, after which the normal fatty</p><p>acids were recovered as methyl esters. The subsequentderivatizations of the intact diether lipids with</p><p>bis(trimethylsilyl)triuoroacetamide (BSTFA), quanti-</p><p>cations of the lipid components with gas chromato-</p><p>graphy (GC) with a ame ionization detector, and</p><p>identications with a GC with a mass selective detector</p><p>were made according to Virtue et al. (1996). However, in</p><p>contrast to them, we used helium as the carrier gas and a</p><p>30 m HP1-MS column. The quantications were madeby using the methyl ester of the fatty acid 19:0 as in-</p><p>ternal standard and identications were facilitated by</p><p>comparisons with standard mixtures (standards were</p><p>obtained from Larodan Fine Chemicals AB, Lim-</p><p>hamnsgaardens allee 9, SE-216 16 Malmoo, Sweden). Therelative amounts of cis and trans isomers of monoun-</p><p>saturated PLFAs were determined after dimethyldisul-</p><p>phide (DMDS) derivatizations followed by GC-MSanalysis (Nichols et al., 1986).</p><p>The coecient of variation (CV) for PLFA determi-</p><p>nations in replicate samples from the reactor was 19%,</p><p>both for the total concentration and the mean for indi-</p><p>vidual PLFAs. For the diether PLEL the CV was 42%.</p><p>2.3. Analytical methods for gas and glucose</p><p>The total gas ow from the reactor was continously</p><p>registered using a gas meter with a volume of 50 ml percycle, calibrated against a wet gas meter (Schlumberger</p><p>Industries, Meterfabriek B.V.). Methane and carbon</p><p>dioxide concentrations in the reactor gas were measured</p><p>with gas chromatography, and as the gas volume ab-</p><p>sorbed in 7 M NaOH, respectively (Jarvis et al., 1995).</p><p>The glucose concentration was measured with HPLC</p><p>(Schnuurer et al., 1996) and alkalinity with acid titration(APHA, 1985).</p><p>2.4. Fate of added glucose</p><p>A carbon budget for the added glucose was con-</p><p>structed for the total duration of Experiment B, i.e. one</p><p>hydraulic retention time. The following sinks for the</p><p>added glucose were considered: 1. Total amounts ofmethane and carbon dioxide leaving the reactor (also</p><p>corrected for gas produced from the regular substrate).</p><p>2. Glucose passing the reactor without being metabo-</p><p>lized. 3. Increase in microbial biomass in the reactor</p><p>contents. 4. Microbial biomass washed out of reactor.</p><p>5. Propionate and acetate washed out of reactor.</p><p>We did not correct the ow of carbon dioxide for</p><p>losses of bicarbonate from the reactor uid due to dropsin pH and alkalinity, since both parameters returned to</p><p>initial values well before the end of the experiment.</p><p>2.5. Data analysis</p><p>In order to detect general, systematic variation in thefatty acid composition, the PLFA data were treated with</p><p>principal component analysis (PCA), performed with</p><p>the JMP software. Standardized principal compo-</p><p>nents calculations were used, where the principal</p><p>component scores were scaled to have unit instead of</p><p>canonical variance. By investigating two-dimensional</p><p>plots of the principal components (PCs), general dier-</p><p>ences among samples could be detected.</p><p>3. Results</p><p>3.1. Eects on microbial communities</p><p>During the two sampling periods, a total of 22 die-</p><p>rent PLFAs and one PLEL were quantied in the re-</p><p>actor contents (Table 1). There were traces of otherdiethers that could not be quantied. Nineteen of the</p><p>normal fatty acids were common to both experiments.</p><p>Branched-chain and saturated PLFAs with uneven</p><p>238 I. Sundh et al. / Bioresource Technology 89 (2003) 237243</p></li><li><p>numbers of carbons were most prominent, showing that</p><p>gram-positive and anaerobic gram-negative eubacteria</p><p>dominated the community. With the exception of small</p><p>amounts of 18:2, polyunsaturated fatty acids were</p><p>missing, demonstrating a lack of eucaryotic microor-</p><p>ganisms. The ratio of PLEL to PLFAs is an estimate ofthe relative contribution of methanogens to the micro-</p><p>bial biomass. With respect to complete phospholipid</p><p>molecules, the average amount of PLEL was 3.5% and</p><p>8.3% of the PLFAs in the two experiments.</p><p>In both experiments, the total PLFA concentration</p><p>roughly doubled in response to glucose addition,</p><p>showing that the eubacterial biomass increased sub-</p><p>stantially (Fig. 1). On the other hand, there was noobvious change in PLELs (Fig. 1), and hence not in</p><p>methanogenic biomass. However, the latter diered by a</p><p>factor of two between the two overloading experiments.</p><p>Total PLFA concentration peaked after 34 days and</p><p>then gradually declined, returning to its initial level</p><p>approximately three weeks after glucose addition.</p><p>The glucose overloadings resulted in time-dependent</p><p>changes in the PLFA composition in the reactor. Whenthe PLFA data from the two experiments were separately</p><p>treated by PCA, regular changes with time were evident</p><p>(Fig. 2).Most of the quantitatively dominating fatty acids</p><p>were high in PC1, i.e. they had a peak in concentration 2</p><p>4 days after the glucose addition. Most of the fatty acids</p><p>responsible for the separation of early and late samples in</p><p>the PCA (e.g. 16:1x7, 18:1x9, 18:0) had overall lowconcentrations (Table 1). Thus, the time-dependent</p><p>changes demonstrated by the PCA are caused both by an</p><p>increase in total PLFA concentration and by changes in</p><p>the relative contributions of individual fatty acids.</p><p>In a simultaneous PCA with all samples, the generalpattern of movement with time in the score plot, was</p><p>similar in both experiments, showing that the eect of</p><p>the glucose on the microbial community structure was</p><p>similar. However, the two overloading experiments were</p><p>clearly separated in the score plot, showing that they</p><p>had slightly dierent PLFA composition overall. The</p><p>main causes for this separation were the three fatty acids</p><p>occurring in only one of the experiments (i17:1, 20:0,cy21:0), and the overall dierent concentrations of 18:2,</p><p>i16:0 and PLEL.</p><p>The abundances of the trans isomers of the mono-</p><p>unsaturated PLFAs were determined in Experiment C.</p><p>The average trans/cis ratio of 16:1x7, 18:1x7 and18:1x9 was generally low (0.61.6%) and did not changesignicantly during the course of the experiment. These</p><p>low ratios show that the nutrient supply to the microbialcommunity was generally high.</p><p>3.2. Process parameters</p><p>During normal feeding rates with cellulose and al-</p><p>bumin, glucose was below the detection limit (0.2 g l1).</p><p>Table 1</p><p>Average concentrations of normal phospholipid fatty acids (PLFAs; in mol % for the individual fatty acids) and archaeal diether lipids (PLEL; in</p><p>nmolml1) in a mesophilic biogas reactor fed a synthetic substrate mixture and receiving instant overloadings with glucose</p><p>PLFA 15 Times overloading (B) 25 Times overloading (C)</p><p>Normal After overload Normal After overload</p><p>i13:0 0.81 (0.010) 1.05 (0.35) 1.47 (0.11) 1.33 (0.19)</p><p>a13:0 0.51 (0.027) 0.74 (0.13) 0.70 (0.093) 0.90 (0.12)</p><p>i14:0 4.9 (0.069) 5.5 (0.61) 4.4 (0.16) 5.5 (0.77)</p><p>14:0 2.9 (0.11) 2.3 (0.61) 2.2 (0.097) 3.2 (1.1)</p><p>i15:0 20.6 (0.17) 22.9 (4.1) 25.4 (0.29) 26.4 (3.9)</p><p>a15:0 28.3 (0.059) 30.5 (2.0) 25.8 (0.30) 24.6 (1.5)</p><p>15:0 2.9 (0.072) 7.2 (2.2) 6.6 (0.46) 11.1 (2.6)</p><p>i16:0 11.1 (0.064) 7.4 (2.4) 4.5 (0.075) 3.6 (0.60)</p><p>16:1x7 0.81 (0.005) 0.67 (0.17) 1.1 (0.24) 0.99 (0.32)16:0 5.9 (0.051) 4.4 (1.2) 5.9 (0.13) 4.9 (0.91)</p><p>i17:1 0.81 (0.26) 0.51 (0.18)</p><p>i17:0 5.4 (0.011) 4.0 (0.64) 5.8 (0.11) 4.8 (1.5)</p><p>a17:0 5.0 (0.004) 3.9 (0.62) 5.1 (0.12) 3.7 (0.68)</p><p>17:0 1.6 (0.011) 2.6 (1.09) 1.5 (0.050) 2.0 (0.58)</p><p>18:2 2.4 (0.055) 1.6 (0.43) 1.4 (0.31) 0.84 (0.23)</p><p>18:1x9 2.8 (0.031) 2.0 (0.30) 4.4 (0.16) 2.7 (0.54)18:1x7 0.54 (0.001) 0.42 (0.11) 0.47 (0.026) 0.39 (0.056)18:0 2.1 (0.072) 1.5 (0.31) 2.6 (0.12) 1.6 (0.26)</p><p>10Me18:0 0.38 (0.085) 0.23 (0.029) 0.24 (0.12)</p><p>20:1 0.62 (0.010) 0.46 (0.082) 0.50 (0.051) 0.47 (0.079)</p><p>20:0 0.37 (0.042) 0.33 (0.17)</p><p>cy21:0 0.21 (0.11)</p><p>Total PLFAs (nmolml1) 160 (2.7) 212 (39.9) 121 (2.7) 182 (40.3)PLEL (nmolml1) 3.86 (0.98) 3.42 (0.98) 7.33 (0.26) 7.62 (0.73)</p><p>Data are given separately for normal, pre-overload, conditions and for a three-week period following overload. SD is shown within parenthesis.</p><p>I. Sundh et al. / Bioresource Technology 89 (2003) 237243 239</p></li><li><p>The added glucose was rapidly consumed and the con-</p><p>centration had returned to non-detectable levels after 2</p><p>3 days (Fig. 3). Simultaneously, the alkalinity droppedfrom roughly 7, to 5 and 4 g CaCO3 l1, after 15 and 25times overloading, respectively. The alkalinity gradually</p><p>returned to initial values after the glucose was consumed</p><p>(Fig. 3).</p><p>The ow of biogas increased substantially as an im-</p><p>mediate result of the glucose addition. The ow of car-</p><p>bon dioxide returned to pre-glucose rates sooner than</p><p>the ow of methane. As long as free glucose was avail-able, carbon dioxide ow exceeded methane production.</p><p>Thereafter, the biogas composition changed back to</p><p>dominance by methane. The regular feeding occasions</p><p>were evident as regular cycles of increased production of</p><p>methane and carbon dioxide.</p><p>The sum of the glucose sinks considered amounted to</p><p>3.17 mol C (Table 2). This is 56% of the glucose carbon</p><p>in the reactor immediately after the addition (the rstsamples were taken approximately 5 min after addition).</p><p>The calculation assumes that the degradation of the</p><p>regular substrate was not aected by the addition. This</p><p>is not likely, and on the converse assumption, that there</p><p>was no degradation of regular feed after the addition,</p><p>the total recovery was 6.90 mol C, equivalent to 121% of</p><p>the glucose carbon (Table 2).</p><p>4. Discussion</p><p>Comparatively short and saturated, branched- or</p><p>straight-chain fatty acids with...</p></li></ul>