Article pubs.acs.org/est
Structure and Function of Assemblages of Bacteria and Archaea in Model Anaerobic Aquifer Columns: Can Functional Instability Be Practically Beneficial? Denice K. Nelson,† Timothy M. LaPara, and Paige J. Novak* Department of Civil Engineering, University of Minnesota, 500 Pillsbury Drive SE, Minneapolis, Minnesota 55455, United States S Supporting Information *
ABSTRACT: Biodegradable organic carbon is often added to aquifers to stimulate microbial reduction of oxidized contaminants. This carbon also stimulates fermenters, which generate important metabolites that can fuel contaminant reduction and may enhance dissolution of hydrophobic compounds. Therefore, understanding how different methods of carbon addition affect the fermentative community will enable design of more effective remediation strategies. Our research objective was to evaluate the microbial communities that developed in model aquifer columns in response to pulsed or continuous molasses input. Results indicated that the continuously fed column produced relatively low concentrations of metabolic intermediates and had a greater proportion of Bacteria and methanogens, as measured by quantitative polymerase chain reaction, near the column inlet. In contrast, the pulsed-fed column generated periodic high concentrations of metabolic intermediates, with Bacteria and methanogens distributed throughout the length of the column. The community structures of Bacteria and Archaea, measured via automated ribosomal intergenic spacer analysis, in the pulsed-fed column were significantly different from those in the control column (not fed). The microbial community composition of the continuously fed column, however, became increasingly similar to the control column along the column length. These results demonstrate that a strategy of pulsed carbon addition leads to activity that is associated with functional instability, in terms of the production of periodic pulses of fermentation products and changing carbon concentration, and may be advantageous for remediation by producing large quantities of beneficial intermediates and resulting in more homogenously distributed biomass.
■
profile.11 Although studies have been performed on how the method of substrate addition impacts the community structure and function of aerobic trichloroethene-degraders12 and an anaerobic digester community,13 the effect of the carbon delivery strategy on the development of an entire anaerobic microbial community has not been investigated. The objective of this study was to determine the effect of carbon addition via two commonly used carbon application strategies (low-dose, continuously fed carbon versus high-dose, pulsed-fed carbon) on microbial community structure and function in anaerobic soil communities. We hypothesized that the method of carbon addition would affect the type and concentration of fermentation products generated, which would then affect the composition and distribution pattern of methanogenic Archaea. A better understanding of the interrelationships among carbon addition, the production of fermentation products, and the microbial community structure of both the Bacteria and Archaea (and methanogens in
INTRODUCTION Biodegradable organic carbon is commonly added to groundwater to anaerobically treat plumes of contaminants, such as chlorinated solvents or heavy metals.1 This process of biostimulation works by promoting biologically mediated contaminant reduction using indigenous organisms. Substantial research has been conducted on optimizing the growth of these contaminant-reducing organisms (e.g., 2−4) with a particular emphasis on the different types of carbon substrates that can be used to most effectively stimulate their growth (e.g., 5−7). In contrast, the structure and function of the entire anaerobic microbial community that develops in response to carbon inputs during biostimulation applications, particularly those organisms that ferment this carbon, is poorly understood. Because the success of in situ biostimulation relies on the distribution of injected substrate to the targeted treatment zone, research has also been conducted on the comparative advantages of different carbon delivery strategies. Two different carbon delivery strategies are used for in situ treatment: continuous or pulsed delivery.8,9 Studies evaluating continuous versus pulsed inputs of nutrients (i.e., carbon and nitrogen) have suggested that pulsed amendments can achieve a larger radius of influence10 and a more evenly distributed biomass © 2012 American Chemical Society
Received: Revised: Accepted: Published: 10137
February 23, 2012 July 30, 2012 August 6, 2012 August 8, 2012 dx.doi.org/10.1021/es300652z | Environ. Sci. Technol. 2012, 46, 10137−10144
Environmental Science & Technology
Article
Figure 1. Schematic of the experimental columns used in this study.
columns was changed every 3−4 days to minimize the degradation of molasses and/or media components in the feed prior to column addition. The pulsed addition of molasses occurred approximately every 30 days (once COD concentrations fell below 500 mg/L), consisting of a 3-day step input of 10% molasses (by volume, in minimal groundwater media) followed by approximately 27 days of minimal groundwater media addition. The minimal groundwater media is described in the SI. The continuously fed column had a total carbon mass input of 504 g of COD, while the pulsed-fed column had a total carbon mass input of 1460 g of COD over the 145-day experiment. Although the total mass of carbon differed by a factor of about 2.9 in the two columns, this operation was consistent with practice, where a range of carbon dosing is employed to stimulate microbial communities either through continuous application of low concentrations of carbon (via injections or emplacement of an insoluble/slow release substrate) or periodic introduction of concentrated carbon.14 Operating over this range of input concentrations allowed a better understanding of what might be expected in the field under these conditions. Bromide tracer studies were performed on the columns before and after the experiment (data not shown) and showed linear velocities of between 2.5 and 4.3 cm/day in the columns prior to the beginning of the experiment and 3.3 and 6.1 cm/day in the columns at the end of the experiment. The increase in linear velocity is a result of biomass growth and some short-circuiting in the columns. Column influent lines were disinfected twice weekly using a 10% (by volume) bleach solution to remove biomass growth within the tubing and to prevent clogging of the feed lines. Periodic flushing of the column inlets was also performed, with the frequency based on individual column performance. Aqueous samples were collected weekly from each sampling port for chemical analysis. Samples were withdrawn using syringes equipped with 5-cm long 22-gauge needles. All samples were filtered through a 0.2-μm polyethersulfone
particular) should be useful in optimizing the bioremediation of numerous contaminated sites based on a given site’s specific requirements (e.g., optimize halorespiration, biomass distribution, or minimize the production of fermentation intermediates).
■
MATERIALS AND METHODS
Soil Columns. Three soil columns (Figure 1) were continuously operated for 145 days. Columns were initially aerobic to simulate field conditions prior to carbon addition. All columns were packed with an aquifer material mixture (described in detail in the Supporting Information (SI)). One column was supplied with a continuous dilute molasses feed (0.4% by volume in a minimal groundwater media), while the second column was fed periodic pulses of high strength molasses (10% by volume in minimal groundwater medium). A third column served as an experimental control and therefore received no carbon input (continuous feed of minimal groundwater media only). Column construction is described in detail in the SI. Briefly, columns were constructed using 1.5-m long, 0.1-m inner diameter polyvinyl chloride (PVC) pipes. Six sample ports (labeled A−F) were tapped along each column length. Columns were operated horizontally to simulate groundwater flow in an aquifer setting. In addition to sample ports, several vertical 6-mm diameter glass tubes were connected to the top of the column to allow for off-gassing (e.g., methane) during column operation. This was important to prevent the column from clogging and also to prevent dangerous pressure build-up due to biogenic gas formation. Wholesome Sweeteners dark (blackstrap) molasses (unsulphured) was used as the carbon source for the continuously and pulsed-fed columns. The feed solution was pumped into each of the columns at a rate of 0.06 mL/min (empty bed velocity ≈1.1 cm/day). The pH of the influent was adjusted to 8 throughout the experiment using 10% HCl. The feed to the 10138
dx.doi.org/10.1021/es300652z | Environ. Sci. Technol. 2012, 46, 10137−10144
Environmental Science & Technology
Article
Figure 2. Chemical oxygen demand concentrations versus time in Port A of the three columns. The introduction of molasses began at time = 0.
syringe filter (Nalgene, Rochester, NY) prior to analysis. A total of 8 mL was collected from each port during each sampling event. Soil Sample Collection. Soil columns were sacrificed after the final liquid sampling event 145 days following the introduction of molasses. The columns were disconnected from the feed solution and drained of excess liquid. Soil representing the microbial community at each sampling port and the column influent was collected, homogenized, and immediately frozen at −20 °C until further analysis. The methodology used for soil collection is described in the SI. Genomic DNA Extraction. Genomic DNA was extracted from column samples (∼ 500 mg of wet weight per sample) using a bead beater to lyse cells. Genomic DNA was then extracted and purified from sediment samples using a PowerSoil DNA Kit (MoBio Laboratories, Inc., Carlsbad, CA). Genomic DNA extractions were performed on three separate soil samples from each location and stored at −20 °C until needed. Community Analysis. Fingerprints of the assemblages of Bacteria and Archaea in the experimental columns were generated by automated ribosomal intergenic spacer analysis (ARISA). ARISA was performed on triplicate soil samples collected at the inlet and sampling ports of each column. The ribosomal intergenic spacer (ITS) regions of Bacteria were amplified using primers ITSF and ITSReub.15 The ribosomal ITS regions of Archaea were amplified using primers 1389F and 71R.16 The ribosomal ITS regions of fungi were amplified using primers 2234C and 3126T.17 Fungi were never detected in any of the column samples (data not shown). Amplified products were resolved by capillary electrophoresis using an ABI 3130xl capillary instrument (Applied Biosystems Inc., Foster City, CA). Fragment peak areas were analyzed using Gene Profiler software. The ARISA methodology is described in detail in the SI. Quantitative PCR. Members of the domain Bacteria and methanogens (in the domain of Archaea) were quantified by quantitative real-time PCR (qPCR) using a Realplex2 Mastercycler (Eppendorf) thermocycler. qPCR was performed on
DNA from each of the triplicate soil samples from each sampling port location. Primers 338F and 518R18 were used to enumerate the Bacteria, whereas primers Met630F and Met803R19 were used to enumerate the methanogens. Melting curve analysis was performed following each qPCR run to ensure that nonspecific PCR products were not generated. Standards (described in the SI) were used to quantify the number of gene copies present in each sample. Detection limits were 3 × 104 gene copies and 6 × 102 gene copies per gram of soil (dry weight) for Bacteria and methanogens, respectively. The qPCR methodology is described in detail in the SI. Analytical Methods. Ketones (acetone, methyl ethyl ketone, methyl isobutyl ketone), alcohols (methanol, ethanol, butanol), and methane were analyzed by gas chromatography coupled to a flame ionization detector. Detection limits were 11 mg/L for methanol, 2 mg/L for ethanol, 2 mg/L for acetone, 1 mg/L for methyl ethyl ketone, 1 mg/L for butanol, 1 mg/L for methyl isobutyl ketone, and 0.5 mg/L for methane. The methodology is described in detail in the SI. Volatile fatty acid analysis was performed using an HPLC equipped with a diode-array detector (210 nm wavelength). Detection limits were 18 mg/L for lactate, 6 mg/L for formate, 6 mg/L for acetate, 8 mg/L for propionate, 12 mg/L for valerate, and 16 mg/L for butyrate. The methodology is described in detail in the SI. Chemical oxygen demand (COD) was analyzed using an Accu-TEST (range 20−900 mg/L) mercury-free Micro-COD system (Bioscience Inc., Bethlehem, PA) following the manufacturer’s protocol. COD standards were prepared using a potassium hydrogen phthalate standard. A Beckman 32 digital pH meter equipped with a combination electrode was used for pH measurement. Data Analysis. ARISA data were analyzed using the following protocol: (1) fragment lengths 0.5% of the total peak area were considered, followed by (3) manual alignment of the remaining fragment lengths between triplicate samples, 10139
dx.doi.org/10.1021/es300652z | Environ. Sci. Technol. 2012, 46, 10137−10144
Environmental Science & Technology
Article
Figure 3. Concentration of fermentation products detected in the columns over time. Panels show (a) Port A in the pulsed-fed column, (b) Port C in the pulsed-fed column, (c) Port F in the pulsed-fed column, (d) Port A in the continuously fed column, (e), Port C in the continuously fed column, and (f) Port F in the continuously fed column. Symbols are ethanol (⧫), acetate (■), butyrate (▲), butanol ( × ), and propionate (●). Time = 0 represents the start of molasses introduction in the continuously fed column. Vertical dashed lines represent the times of molasses addition in the pulsed-fed column.
A two-sided student t test performed at the 95% confidence interval was used to determine statistical differences between values obtained using qPCR.
(4) manual alignment of fragment lengths between sampling port locations of the same column, and (5) manual alignment of fragment lengths between columns. Each manual alignment considered fragments within one nucleotide of length to be the same fragment. Ordination statistical analyses of the ARISA data were performed using R.20 The package vegan was used to analyze ARISA community structure using nonmetric multidimensional scaling (NMDS). The function metaMDS was used to perform the NMDS analysis with the dissimilarity matrix calculated for both species and sample locations using Bray−Curtis distance measure (function vegdist).21 The ordination was overlain by environmental factors (COD, volatile fatty acids, ketones, and 16S rRNA genes of Bacteria or of methanogens) using the function envfit to determine the correlation between these parameters and the ordination of operational taxonomic units.21,22
■
RESULTS As expected, different COD patterns over time were observed in Port A as a result of the different molasses feeding methods (Figure 2). In the continuously fed column, an initial increase in COD concentration was followed by a steady decline until a low quasi-steady state COD concentration of 751 ± 147 mg/L (about 4% of what was fed) was reached at Port A around Day 90. In the pulsed-fed column, the periodic inputs of molasses resulted in regular and repeated spikes of COD (about 2−6% of what was fed). Substantial differences were observed with respect to the generation of several fermentation products (ethanol, butanol, butyrate, acetate, and propionate) between the columns (Figure 3). In the pulsed-fed column, acetate initially dominated in all sampling ports (ranging up to 4000 mg/L), followed by a shift 10140
dx.doi.org/10.1021/es300652z | Environ. Sci. Technol. 2012, 46, 10137−10144
Environmental Science & Technology
Article
103 copies/g dry soil). In contrast, greater numbers of methanogens were detected along the entire length (Ports B−F) of the pulsed-fed column, ranging from 4.0 × 106 to 6.0 × 107 gene copies/g dry soil. Indeed, when all the data were pooled, the number of methanogenic gene copies in the pulsedfed column was statistically greater (p = 0.016) than that in the continuously fed column. Nevertheless, the numbers of methanogen gene copies at each sampling port were not statistically different in the two columns (p = 0.552, 0.358, 0.249, 0.103, 0.260, 0.256 for Ports A−F, respectively). Of particular interest was the fact that the pulsed feeding regime changed the distribution of organisms. When all of the data from the sampling ports (A−F) were pooled, the fraction of methanogens comprising the analyzed prokaryotic community (methanogens plus Bacteria) was statistically greater (p = 0.002) in the pulsed-fed column when compared to the continuously fed column. Although the mass of carbon fed to the two columns was different and was expected to result in different total numbers of organisms, this dramatic change in the percent of the community made up of methanogens was unexpected and most likely a result of the greater penetration of acetate and hydrogen- and acetate-producing fermentation intermediates such as ethanol and butyrate. As mentioned above, the numbers of Bacteria were statistically the same along the length of the continuously fed and pulsed-fed columns; their composition, however, differed significantly at the 95% confidence level (Figure 5). Bacteria community composition transitioned significantly along the length of both the continuously fed and pulsed-fed columns, where notably, the composition at nearly every port in the
to butyrate, which was produced in excess of 7000 mg/L in Port A on Day 70. This shift may be important, as butyrate has been shown to be an excellent hydrogen donor for dechlorinators, producing low concentrations of hydrogen and acetate during its fermentation (e.g., 2). Propionate was detected in greater quantities toward the end of the study (Day 140) in this column. In contrast, acetate, butyrate, and propionate were only present through Day 125 at levels
