Journal of Hazardous Materials 271 (2014) 292–301
Contents lists available at ScienceDirect
Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat
Moisture effects on gas-phase biofilter ammonia removal efficiency, nitrous oxide generation, and microbial communities Liangcheng Yang a,∗ , Angela D. Kent b , Xinlei Wang a , Ted L. Funk a , Richard S. Gates a , Yuanhui Zhang a a b
Department of Agricultural and Biological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA Department of Natural Resources and Environmental Sciences, University of Illinois at Urbana-Champaign, Urbana, IL, USA
h i g h l i g h t s • • • • •
Increase moisture content from 35 to 55% greatly improved NH3 removal. Increase moisture content from 55 to 63% triggered N2 O generation. Ammonia oxidizers (amoA communities) resisted to moisture disturbance. Denitrifiers (nosZ communities) were resilient to moisture disturbance. Decrease of nosZ abundance caused N2 O generation at high MC.
a r t i c l e
i n f o
Article history: Received 29 October 2013 Received in revised form 23 January 2014 Accepted 25 January 2014 Available online 12 February 2014 Keywords: Biofilter Ammonia Nitrous oxide Ammonia oxidizer Denitrifier Moisture
a b s t r a c t We established a four-biofilter setup to examine the effects of moisture content (MC) on biofilter performance, including NH3 removal and N2 O generation. We hypothesized that MC increase can improve NH3 removal, stimulate N2 O generation and alter the composition and function of microbial communities. We found that NH3 removal efficiency was greatly improved when MC increased from 35 to 55%, but further increasing MC to 63% did not help much; while N2 O concentration was low at 35–55% MC, but dramatically increased at 63% MC. Decreasing MC from 63 to 55% restored N2 O concentration. Examination of amoA communities using T-RFLP and real-time qPCR showed that the composition and abundance of ammonia oxidizers were not significantly changed in a “moisture disturbance-disturbance relief” process in which MC was increased from 55 to 63% and then reduced to 55%. This observation supported the changes of NH3 removal efficiency. The composition of nosZ community was altered at 63% MC and then was recovered at 55% MC, which indicates resilience to moisture disturbance. The abundance of nosZ community was negatively correlated with moisture content in this process, and the decreased nosZ abundance at 63% MC explained the observation of increased N2 O concentration at that condition. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Over 70% ammonia emissions originate from livestock operations [1], and through reactions with sulfuric and nitric acid aerosols, these emissions generate 5–11% of the total PM2.5 (particulate matter with aerodynamic equivalent diameter of 2.5 l or less) in the United States [2,3]. Biofilters are commonly used to mitigate livestock emissions [4,5]. These engineered systems are bioreactors filled with organic packing materials such as wet
∗ Corresponding author at: 310B AESB, 1304 W. Pennsylvania Avenue, Urbana, IL 61801. Tel.: +1 217 333 9415; fax: +1 217 244 0323. E-mail address: [email protected] (L. Yang). http://dx.doi.org/10.1016/j.jhazmat.2014.01.058 0304-3894/© 2014 Elsevier B.V. All rights reserved.
woodchips or composts that remove ammonia from contaminated air streams through absorption, and then oxidize it into nitrite and nitrate using microorganisms growing on the surface of packing media [6]. Most studies have focused on the value and importance of improving biofilter ammonia capture ability [7–9], however, recent reports about the generation of nitrous oxide from biofilters have prompted designers to consider the consequences of greenhouse gas (GHG) emissions [10–12]. Nitrous oxide can be produced by both ammonia oxidizers and denitrifiers within a biofilter [13]. A very small group of ammonia oxidizers are able to directly reduce NO2 − under microaerophilic conditions through the activities of denitrifying enzymes [14,15], but the majority of nitrous oxide is believed to come from denitrification [5,16]. Within a large scale biofilter, given the compaction of small
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
Nomenclature ANOSIM Analysis of similarity ANOVA Analysis of variance ARISA Automated ribosomal intergenic spacer analysis BF Biofilter CTAB Cetyl trimethyl ammonium bromide GHG Greenhouse gas ID Inner diameter LSD Least significant difference MC Moisture content NMDS Non-metric multidimensional scaling OD Outer diameter PCR Polymerase chain reaction PM Particulate matter PerMANOVA Permutational multivariate analysis of variance qPCR Quantitative polymerase chain reaction RE Removal efficiencies RH Relative humidity T-RFLP Terminal restriction fragment length polymorphism VOC Volatile organic compound
media [17], there exist anaerobic zones which favor denitrification [18]. Moisture content is a major determinant of biofilter performance [19]. Too little moisture slows absorption of gaseous ammonia and microbial activities [20]; while too much water fills biofilter media pores and retards the transport of NH4 + , O2 , and nutrients across the water film coating the microbes. Despite the explicit role of moisture for determining ammonia removal [21], the effects of moisture on nitrous oxide generation are rarely specifically considered or tested. A primary reason for this gap is the lack of awareness about the extent of localized oxygen limitation in media due to high moisture contents [11], because a biofilter is generally considered to be an aerobic environment since air continuously goes through it. Moreover, the adsorption of nitrous oxide to media surfaces complicates observations as several studies reported mitigation of nitrous oxide through biofilters [22,23]. However, water competes with nitrous oxide for adsorption spots, especially when the media is decorated with hydrophilic functional groups such as OH or COOH [24,25]. Therefore, at high moisture content, less nitrous oxide adsorption and higher nitrous oxide generation rate are expected [26,27]. Maia et al. [10] reported that 0.6–2.0 ppm nitrous oxide was produced in a start-up biofilter with about 60% media moisture content. Besides the influence of moisture content on the physicochemical interactions of ammonia and nitrous oxide with biofilter media, the response of biofilter microorganisms to moisture changes is another key issue, particularly given the lack of effective moisture managements in many agricultural biofilters. The relationship between community dynamics and system functional stability has been debated, and different phenomena were observed from case to case [28–32]. The concepts of resistance, resilience and redundancy have been applied to describe the relationships between microbial community structure and biofilter performance [33]. Long-term stable function has been sustained in a biofilter treating ammonia and VOCs emissions, although the microbial communities were highly dynamic [34]. Li et al. also observed functional redundancy when phosphorus was added to nitrate removal bioreactors [35]. Another study carried out by Gentile et al. [36] demonstrated resilience of microbial communities to environmental disturbances in a denitrifying reactor; and studies in soil science showed that resilience depends on both soil structure and soil
293
microbial community composition [37,38]. Resistance of disturbances was observed in several natural systems [39,40]. Given these observations in other systems, it seems the design of reliable biofiltration systems will require better understanding of the relationship among microbial community structure, moisture fluctuations (disturbance), and biofilter performance in an engineered, well-controlled system. Unlike contaminant gradients [34] or toxic chemical stress [41,42], moisture variation not only influences the microbial communities which eventually determine biofilter performance, but also directly affects absorption of ammonia into biofilters [43]. In other words, moisture may affect biofilter performance in both biological and chemical ways. In this study, moisture content was manipulated for a wide range of 35–63% (wet basis) to discover its roles. Particular microbial communities such as ammonia oxidizers and denitrifiers are responsible for ammonia oxidation and nitrous oxide production. Thus, their communities were examined in this study. In this investigation, four bench-scale biofilters were constructed to evaluate the effects of moisture on biofilter performance and microbial community structure. The objectives were: (i) to test the influence of moisture content on biofilter ammonia removal and nitrous oxide generation, and to determine the appropriate moisture range for desirable performance; (ii) to explore the response of bacterial communities, ammonia oxidizers and denitrifiers to moisture variance, including community structure shifting and abundance change; and (iii) to link the microbial communities to biofilter function stability.
2. Materials and methods 2.1. Reactor design and operation The system includes a gas preparation unit, a gas analysis unit, a control unit, and four biofilters (Fig. 1). Anhydrous ammonia (99.99%, S. J. Smith Co., Urbana, IL) regulated by a mass flow controller was mixed with pre-humidified air to provide 70 liter per minute (lpm), 40 ppm ammonia gas for each biofilter (BF). The loading rate was 5.24 g-NH3 h−1 m−3 . Air from the blower was 2–3 ◦ C higher than normal room conditions. Therefore, a cooling coil was used to bring the gas closer to room temperature. Four cylindrical biofilters (column ID = 0.45 m, column H = 0.50 m) were made of transparent plastics. A layer of 0.25 m media, 1:1 volume mixture of compost and woodchip, was supported 0.10 m above the bottom of the biofilter tank by a perforated plate. For each biofilter, there were four sampling ports (4 × 90◦ ) located in both upper (0.05 m bellow top surface) and lower (0.05 m above bottom surface) layer. Water was pumped through a coiled micro soaker hose (OD = 0.0064 m) onto the top of the media. Inlet gas was treated by biofilters and then the purified gas left via an outlet port. Empty bed retention time was 34 s. The retention time is high comparing to some other studies, mainly to reduce pressure drop. A control and data acquisition system (National Instruments Co., Austin, TX) was used to record data and control the whole system including the water pump (170DM5, Stenner Pumps & Parts, Indianapolis, IN) and solenoid valves. No extra nutrients or inoculum cultures were added, and a 10-day start-up step was allowed. The test was composed of four steps. Each step took 22–35 days, depending on operation conditions. To increase moisture content between steps, a certain amount of water was added based on calculation and moisture content measurement; to reduce moisture content between steps 3 and 4, the biofilters were air-dried for several days, during which air stream (no ammonia input) remained unchanged but water addition was stopped; and to maintain moisture content during each step, water was added regularly based
294
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
Fig. 1. Ammonia removal and greenhouse gas generation experimental setup.
on moisture content measurement. Targeted moisture contents in each biofilter are shown in Table 1. BF1 and BF2 were designed as replicate treatments, while BF3 and BF4 were considered as controls. For treatment biofilters, steps 2–4 can be considered as a “moisture disturbance-disturbance relief” process [33]. The initial media moisture content of mixed media was 33.2% (considered as step 0). Moisture content was measured by oven drying at 105 ◦ C for 24 h, and media size distribution was analyzed using a Penn State Forage Particle Separator (NASCO, product #: C24682N). Packing media was obtained from the Urbana Recycling Center (Urbana, IL). According to the provider’s description, the woodchip was composed of brush and log materials, and was aged for several months before finally shredding with a tub grinder into shredded hardwood mulch; while the compost was produced by composting leaf in the fall of the year.
2.2. Gas and media sampling Each of the five gas sources (inlet gas, and outlet gases from biofilters 1–4) was analyzed for 4.8 h every day in a rotation, using one analyzer (INNOVA1412, California Analytical, Inc., Orange, CA. NH3 filter #: 0974, detection limit 0.2 ppm; N2 O filter #: 0985, detection limit 0.03 ppm), by means of the control system via the solenoid valves. The analyzer was calibrated before experiment initiation, and then was checked using certified gases (80 ppm NH3 and 5 ppm N2 O, within ±1% error, Airgas, Inc.) every two weeks. Temperature and relative humidity (RH) were measured using a probe (HMP155, Vaisala Inc., Woburn, MA). The daily variance is negligible since the temperature and flowrate of the inlet gas were stable. The system recorded data every twenty seconds. Data generated in the first 30 min was discarded since a gas transition normally took 3–5 min. At the end of each testing step, 10 g of randomly selected media particles was collected into a 15 ml tube from each sampling port for microbial analyses (4 biofilter × 8 sampling ports × 4 steps = 128 samples). Eight original media samples (considered as step 0) were
Table 1 Target moisture content and average of measured moisture content. Target MC
Average of measured MC
MC, (%)
BF 1, 2
BF 3, 4
BF1
BF2
BF3
BF4
Step 1 Step 2 Step 3 Step 4
35 55 63 55
35 55 55 55
34.1 53.1 62.4 56.8
32.5 52.6 62.5 54.2
34.6 55.4 55.8 53.7
33.3 55.9 55.6 54.2
MC – moisture content, BF – biofilter. BF 1, 2 are considered as replicate treatment, BF 3, 4 are considered as replicate controls.
collected as well. The samples were stored at −80 ◦ C before analyses. In total, 128 + 8 = 136 samples were collected. 2.3. DNA extraction and purification Biofilter media samples were lyophilized. Large woodchips were physically broken into small ones, and then were manually homogenized. DNA was extracted from 0.15 g dry media using a FastDNA SPIN Kit for Soil (MP Biomedicals, Solon, OH) according to the manufacturer’s protocol. Extracted DNA was then purified using cetyl trimethyl ammonium bromide (CTAB) extraction [44]. DNA concentration was adjusted to standard concentrations of 10 ng/l prior to analyses. 2.4. Microbial community analyses Bacterial community composition in biofilter media samples was assessed using automated ribosomal intergenic spacer analysis (ARISA) [45]. The polymerase chain reaction (PCR) was carried out in an Eppendorf MasterCycler Gradient (Eppendorf AG, Hamburg, Germany) and the reagents included 1× Tris buffer, 0.25 mg/ml BSA, 2.5 mM MgCl2 , 0.4 M of 1406 F-primer (5 -TGYACACACCGCCCGT3 ) and 23S R-primer (5 -GGGTTBCCCCATTCRG-3 ), 0.025 U/l Promega GoTaq and 2 l purified DNA (10 ng/l). PCR cycle was composed of: initial denaturation at 94 ◦ C for 120 s, 26 cycles of 94 ◦ C for 35 s, 55 ◦ C for 45 s, and 72 ◦ C for 2 min, followed by a final extension at 72 ◦ C for 120 s. DNA fragments generated from ARISA were analyzed by denaturing capillary electrophoresis using an ABI 3730XL Genetic Analyzer (PE Biosystems, Foster City, CA). Details about this method were described previously by Kent et al. [46]. Ammonia oxidizer community composition was analyzed using terminal restriction fragment length polymorphism (T-RFLP) analyses of the amoA gene encoding the catalytic ␣-subunit of archaeal ammonia monooxygenase [47]. A pre-test using real-time qPCR method showed that the abundance of archaeal amoA in the sampled media was about six times of bacterial amoA; therefore, only archaeal amoA community results were reported. PCR reagents contained 50 mM Tris buffer (pH 8.3), 100 g of BSA per ml, 1.5 mM MgCl2 , 200 M of each dNTP, 0.4 M of each primer (amoA-F: 5 -STAATGGTCTGGCTTAGACG-3 and amoA-R: 5 GCGGCCATCCATCTGTATGT-3 ) [48,49], 2.5 U of Taq polymerase, and 50 ng DNA in a final volume of 50 l. The amoA F-primer and R-primer was labeled with the fluorescent dye HEX and NED, respectively. PCR cycle was composed of: initial denaturation at 94 ◦ C for 300 s, 30 cycles of 94 ◦ C for 45 s, 53 ◦ C for 60 s, and 72 ◦ C for 60 s, then followed by a final extension at 72 ◦ C for 900 s. PCR products were purified using the MinElute PCR purification kit (Qiagen, Valencia, CA,), and then were digested with RsaI (New England Biolabs Inc., Ipswich, MA). The terminal restriction fragments labeled
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
295
Table 2 percentages of chemical components and pH conditions in media.
Before testa After testb p-value
Total C
Organic matter
Total-N
Organic-N
NH3 -N
pH
45.3 ± 1.8 32.0 ± 2.6 ***
85.6 ± 6.6 66.2 ± 12.5 0.017*
1.3 ± 0.2 2.1 ± 0.7 0.045*
1.2 ± 0.2 1.9 ± 0.5 0.042*
0.03 ± 0.03 0.2 ± 0.2 0.112
8.1 ± 0.5 6.1 ± 0.4 ***
Data analyzed in the Midwest Laboratories, Omaha, NE. Significance level: ***: p < 0.001, *: p < 0.05. a Four replicates. b One sample from each layer of four biofilters, eight replicates total.
with fluorescence were analyzed by denaturing capillary electrophoresis using an ABI 3730XL Genetic Analyzer (PE Biosystems, Foster City, CA). Denitrifier community composition was also analyzed using TRFLP analyses. The nitrous oxide reductase gene (nosZ) are involved in the transformation of nitrous oxide to dinitrogen gas [50]. The PCR reaction mixtures were the same as above, except primers. The nosZ-F-1181 (5 -CGCTGTTCITCGACAGYCAG-3 ) and nosZ-R-1880 (5 -ATGTGCAKIGCRTGGCAGAA-3 ) were used. The nosZ F-primer was labeled with the phosphoramidite dye 6-FAM. Reaction cycle was composed of: initial denaturation at 94 ◦ C for 180 s, 25 cycles of 94 ◦ C for 45 s, 56 ◦ C for 60 s, and 72 ◦ C for 120 s, then followed by a final extension carried out at 72 ◦ C for 420 s. Similarly, the nosZ PCR products were purified, and then were digested in single-enzyme restriction digests containing either AluI or HhaI (New England Biolabs Inc., Ipswich, MA). The results obtained from AluI and HhaI digested fragment were combined together for community analyses. 2.5. Determination of functional gene abundance The amoA gene was amplified for quantitative PCR (qPCR) using primers Arch-amoAF (5 -STAATGGTCTGGCTTAGACG-3 ) and Arch-amoAR (5 -GCGGCCATCCATCTGTATGT-3 ) [51]; and the nosZ gene was amplified for qPCR using nosZ1527F, 5 -CGCTGTTCA CTTCGACAGCTCA-3 and nosZ1622R, 5 - CGCGAACGGGCAAGCAA GGTGCCG-3 [52]. PCR reactions (for both amoA and nosZ) contained 1× 5 l SYBR green master mix (Applied Biosystems Inc., Foster City, CA), 0.4 mM of each primer, 0.5 mg/ml BSA, and 10 ng of DNA in a final volume of 10 l. A same PCR amplification procedure was applied for both amoA and nosZ, which included an initial denaturation step at 95 ◦ C for 300 s, followed by 40 cycles of 95 ◦ C for 45 s, 56 ◦ C for 60 s, and 72 ◦ C for 60 s. Triplicates of each sample were analyzed. Standard curves of both amoA and nosZ amplicons were obtained based on serial dilutions of mixed PCR products of biofilter samples. Reactions were analyzed on a 384-well Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems Inc., Foster City, CA). 2.6. Statistical analyses Analysis of variance (ANOVA) was used to compare the physical and chemical properties. Abundance of nosZ genes was also compared using ANOVA. LSD Post hoc test was applied to evaluate the difference in ammonia removal efficiency and nitrous oxide concentration in the four operation steps. Bacterial and functional gene communities were analyzed using multivariate analyses. ARISA and T-RFLP data were Hellingertransformed in order to convert the raw density of each peak to a relative intensity [53]. Non-metric multidimensional scaling (NMDS) was applied to visualize the results [54,55]. The Bray–Curtis distance matrices were taken as NMDS inputs [54], and the centroid and standard deviation of each group are displayed in the ordinations [56]. PerMANOVA tests were carried out to examine
how moisture content affected the bacterial, amoA and nosZ communities [57]. To statistically compare two sample groups, pairwise analysis of similarity (ANOSIM) was conducted [58]. The analysis of multivariate homogeneity of group dispersions (using the PERMDISP2 procedure) was carried out to assess the homogeneity of microbial communities within a group of samples [59,60]. It generated a distance between 0 and 1, with longer distance showing lower homogeneity. Calculations of ANOSIM and NMDS were carried out using PRIMER 6 (Primer-E Ltd., Plymouth, United Kingdom); calculation of ANOVA, post hoc and PerMANOVA significance levels (function: adonis), and the group dispersion distances (function: betadisper) were performed in the R statistical environment [61] using functions found in the “MASS” and “vegan” packages. 3. Results 3.1. Physical and chemical properties of media In the tested four months, physical and chemical properties of media changed dramatically, mainly due to intensive chemical and biological processes (Table 2). Total-C and organic matter were consumed and significantly decreased. In contrast, total-N and organic-N concentrations increased significantly as a result of ammonia absorption and immobilization. The pH values decreased over time, most likely due to microbial activities. In general, nitrification process releases H+ and reduces pH [62], while NH3 absorption consumes H+ . Therefore, the balance of proton depends on the extent of nitrification. In this test, more H+ was released than consumed. Media became relatively smaller (Table 3). Microbial activities and aging process might be responsible for the size reduction. 3.2. Moisture management Since water was added onto the top surface of media, upper layers were wetter than lower layers. Moisture contents of the upper layers were relatively constant during steps 2–4, while moisture contents of the lower layers were managed to be different. In step 3, lower layers of BF1 and BF2 pair (59.7 ± 1.6%) were significantly wetter (p < 0.001) than the controls: BF3 and BF4 pair (48.8 ± 1.1%). Averaged moisture contents of the upper and lower layers of the four biofilters are listed in Table 1, and were quite close to the target moisture contents (Table 1). No leachate was observed during the operation.
Table 3 Percent of media size distribution in each size class, determined for 3 replicates.
before test After test p-value
>1.9 cm
0.8–1.9 cm
0.2–0.8 cm
Contents lists available at ScienceDirect
Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat
Moisture effects on gas-phase biofilter ammonia removal efficiency, nitrous oxide generation, and microbial communities Liangcheng Yang a,∗ , Angela D. Kent b , Xinlei Wang a , Ted L. Funk a , Richard S. Gates a , Yuanhui Zhang a a b
Department of Agricultural and Biological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA Department of Natural Resources and Environmental Sciences, University of Illinois at Urbana-Champaign, Urbana, IL, USA
h i g h l i g h t s • • • • •
Increase moisture content from 35 to 55% greatly improved NH3 removal. Increase moisture content from 55 to 63% triggered N2 O generation. Ammonia oxidizers (amoA communities) resisted to moisture disturbance. Denitrifiers (nosZ communities) were resilient to moisture disturbance. Decrease of nosZ abundance caused N2 O generation at high MC.
a r t i c l e
i n f o
Article history: Received 29 October 2013 Received in revised form 23 January 2014 Accepted 25 January 2014 Available online 12 February 2014 Keywords: Biofilter Ammonia Nitrous oxide Ammonia oxidizer Denitrifier Moisture
a b s t r a c t We established a four-biofilter setup to examine the effects of moisture content (MC) on biofilter performance, including NH3 removal and N2 O generation. We hypothesized that MC increase can improve NH3 removal, stimulate N2 O generation and alter the composition and function of microbial communities. We found that NH3 removal efficiency was greatly improved when MC increased from 35 to 55%, but further increasing MC to 63% did not help much; while N2 O concentration was low at 35–55% MC, but dramatically increased at 63% MC. Decreasing MC from 63 to 55% restored N2 O concentration. Examination of amoA communities using T-RFLP and real-time qPCR showed that the composition and abundance of ammonia oxidizers were not significantly changed in a “moisture disturbance-disturbance relief” process in which MC was increased from 55 to 63% and then reduced to 55%. This observation supported the changes of NH3 removal efficiency. The composition of nosZ community was altered at 63% MC and then was recovered at 55% MC, which indicates resilience to moisture disturbance. The abundance of nosZ community was negatively correlated with moisture content in this process, and the decreased nosZ abundance at 63% MC explained the observation of increased N2 O concentration at that condition. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Over 70% ammonia emissions originate from livestock operations [1], and through reactions with sulfuric and nitric acid aerosols, these emissions generate 5–11% of the total PM2.5 (particulate matter with aerodynamic equivalent diameter of 2.5 l or less) in the United States [2,3]. Biofilters are commonly used to mitigate livestock emissions [4,5]. These engineered systems are bioreactors filled with organic packing materials such as wet
∗ Corresponding author at: 310B AESB, 1304 W. Pennsylvania Avenue, Urbana, IL 61801. Tel.: +1 217 333 9415; fax: +1 217 244 0323. E-mail address: [email protected] (L. Yang). http://dx.doi.org/10.1016/j.jhazmat.2014.01.058 0304-3894/© 2014 Elsevier B.V. All rights reserved.
woodchips or composts that remove ammonia from contaminated air streams through absorption, and then oxidize it into nitrite and nitrate using microorganisms growing on the surface of packing media [6]. Most studies have focused on the value and importance of improving biofilter ammonia capture ability [7–9], however, recent reports about the generation of nitrous oxide from biofilters have prompted designers to consider the consequences of greenhouse gas (GHG) emissions [10–12]. Nitrous oxide can be produced by both ammonia oxidizers and denitrifiers within a biofilter [13]. A very small group of ammonia oxidizers are able to directly reduce NO2 − under microaerophilic conditions through the activities of denitrifying enzymes [14,15], but the majority of nitrous oxide is believed to come from denitrification [5,16]. Within a large scale biofilter, given the compaction of small
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
Nomenclature ANOSIM Analysis of similarity ANOVA Analysis of variance ARISA Automated ribosomal intergenic spacer analysis BF Biofilter CTAB Cetyl trimethyl ammonium bromide GHG Greenhouse gas ID Inner diameter LSD Least significant difference MC Moisture content NMDS Non-metric multidimensional scaling OD Outer diameter PCR Polymerase chain reaction PM Particulate matter PerMANOVA Permutational multivariate analysis of variance qPCR Quantitative polymerase chain reaction RE Removal efficiencies RH Relative humidity T-RFLP Terminal restriction fragment length polymorphism VOC Volatile organic compound
media [17], there exist anaerobic zones which favor denitrification [18]. Moisture content is a major determinant of biofilter performance [19]. Too little moisture slows absorption of gaseous ammonia and microbial activities [20]; while too much water fills biofilter media pores and retards the transport of NH4 + , O2 , and nutrients across the water film coating the microbes. Despite the explicit role of moisture for determining ammonia removal [21], the effects of moisture on nitrous oxide generation are rarely specifically considered or tested. A primary reason for this gap is the lack of awareness about the extent of localized oxygen limitation in media due to high moisture contents [11], because a biofilter is generally considered to be an aerobic environment since air continuously goes through it. Moreover, the adsorption of nitrous oxide to media surfaces complicates observations as several studies reported mitigation of nitrous oxide through biofilters [22,23]. However, water competes with nitrous oxide for adsorption spots, especially when the media is decorated with hydrophilic functional groups such as OH or COOH [24,25]. Therefore, at high moisture content, less nitrous oxide adsorption and higher nitrous oxide generation rate are expected [26,27]. Maia et al. [10] reported that 0.6–2.0 ppm nitrous oxide was produced in a start-up biofilter with about 60% media moisture content. Besides the influence of moisture content on the physicochemical interactions of ammonia and nitrous oxide with biofilter media, the response of biofilter microorganisms to moisture changes is another key issue, particularly given the lack of effective moisture managements in many agricultural biofilters. The relationship between community dynamics and system functional stability has been debated, and different phenomena were observed from case to case [28–32]. The concepts of resistance, resilience and redundancy have been applied to describe the relationships between microbial community structure and biofilter performance [33]. Long-term stable function has been sustained in a biofilter treating ammonia and VOCs emissions, although the microbial communities were highly dynamic [34]. Li et al. also observed functional redundancy when phosphorus was added to nitrate removal bioreactors [35]. Another study carried out by Gentile et al. [36] demonstrated resilience of microbial communities to environmental disturbances in a denitrifying reactor; and studies in soil science showed that resilience depends on both soil structure and soil
293
microbial community composition [37,38]. Resistance of disturbances was observed in several natural systems [39,40]. Given these observations in other systems, it seems the design of reliable biofiltration systems will require better understanding of the relationship among microbial community structure, moisture fluctuations (disturbance), and biofilter performance in an engineered, well-controlled system. Unlike contaminant gradients [34] or toxic chemical stress [41,42], moisture variation not only influences the microbial communities which eventually determine biofilter performance, but also directly affects absorption of ammonia into biofilters [43]. In other words, moisture may affect biofilter performance in both biological and chemical ways. In this study, moisture content was manipulated for a wide range of 35–63% (wet basis) to discover its roles. Particular microbial communities such as ammonia oxidizers and denitrifiers are responsible for ammonia oxidation and nitrous oxide production. Thus, their communities were examined in this study. In this investigation, four bench-scale biofilters were constructed to evaluate the effects of moisture on biofilter performance and microbial community structure. The objectives were: (i) to test the influence of moisture content on biofilter ammonia removal and nitrous oxide generation, and to determine the appropriate moisture range for desirable performance; (ii) to explore the response of bacterial communities, ammonia oxidizers and denitrifiers to moisture variance, including community structure shifting and abundance change; and (iii) to link the microbial communities to biofilter function stability.
2. Materials and methods 2.1. Reactor design and operation The system includes a gas preparation unit, a gas analysis unit, a control unit, and four biofilters (Fig. 1). Anhydrous ammonia (99.99%, S. J. Smith Co., Urbana, IL) regulated by a mass flow controller was mixed with pre-humidified air to provide 70 liter per minute (lpm), 40 ppm ammonia gas for each biofilter (BF). The loading rate was 5.24 g-NH3 h−1 m−3 . Air from the blower was 2–3 ◦ C higher than normal room conditions. Therefore, a cooling coil was used to bring the gas closer to room temperature. Four cylindrical biofilters (column ID = 0.45 m, column H = 0.50 m) were made of transparent plastics. A layer of 0.25 m media, 1:1 volume mixture of compost and woodchip, was supported 0.10 m above the bottom of the biofilter tank by a perforated plate. For each biofilter, there were four sampling ports (4 × 90◦ ) located in both upper (0.05 m bellow top surface) and lower (0.05 m above bottom surface) layer. Water was pumped through a coiled micro soaker hose (OD = 0.0064 m) onto the top of the media. Inlet gas was treated by biofilters and then the purified gas left via an outlet port. Empty bed retention time was 34 s. The retention time is high comparing to some other studies, mainly to reduce pressure drop. A control and data acquisition system (National Instruments Co., Austin, TX) was used to record data and control the whole system including the water pump (170DM5, Stenner Pumps & Parts, Indianapolis, IN) and solenoid valves. No extra nutrients or inoculum cultures were added, and a 10-day start-up step was allowed. The test was composed of four steps. Each step took 22–35 days, depending on operation conditions. To increase moisture content between steps, a certain amount of water was added based on calculation and moisture content measurement; to reduce moisture content between steps 3 and 4, the biofilters were air-dried for several days, during which air stream (no ammonia input) remained unchanged but water addition was stopped; and to maintain moisture content during each step, water was added regularly based
294
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
Fig. 1. Ammonia removal and greenhouse gas generation experimental setup.
on moisture content measurement. Targeted moisture contents in each biofilter are shown in Table 1. BF1 and BF2 were designed as replicate treatments, while BF3 and BF4 were considered as controls. For treatment biofilters, steps 2–4 can be considered as a “moisture disturbance-disturbance relief” process [33]. The initial media moisture content of mixed media was 33.2% (considered as step 0). Moisture content was measured by oven drying at 105 ◦ C for 24 h, and media size distribution was analyzed using a Penn State Forage Particle Separator (NASCO, product #: C24682N). Packing media was obtained from the Urbana Recycling Center (Urbana, IL). According to the provider’s description, the woodchip was composed of brush and log materials, and was aged for several months before finally shredding with a tub grinder into shredded hardwood mulch; while the compost was produced by composting leaf in the fall of the year.
2.2. Gas and media sampling Each of the five gas sources (inlet gas, and outlet gases from biofilters 1–4) was analyzed for 4.8 h every day in a rotation, using one analyzer (INNOVA1412, California Analytical, Inc., Orange, CA. NH3 filter #: 0974, detection limit 0.2 ppm; N2 O filter #: 0985, detection limit 0.03 ppm), by means of the control system via the solenoid valves. The analyzer was calibrated before experiment initiation, and then was checked using certified gases (80 ppm NH3 and 5 ppm N2 O, within ±1% error, Airgas, Inc.) every two weeks. Temperature and relative humidity (RH) were measured using a probe (HMP155, Vaisala Inc., Woburn, MA). The daily variance is negligible since the temperature and flowrate of the inlet gas were stable. The system recorded data every twenty seconds. Data generated in the first 30 min was discarded since a gas transition normally took 3–5 min. At the end of each testing step, 10 g of randomly selected media particles was collected into a 15 ml tube from each sampling port for microbial analyses (4 biofilter × 8 sampling ports × 4 steps = 128 samples). Eight original media samples (considered as step 0) were
Table 1 Target moisture content and average of measured moisture content. Target MC
Average of measured MC
MC, (%)
BF 1, 2
BF 3, 4
BF1
BF2
BF3
BF4
Step 1 Step 2 Step 3 Step 4
35 55 63 55
35 55 55 55
34.1 53.1 62.4 56.8
32.5 52.6 62.5 54.2
34.6 55.4 55.8 53.7
33.3 55.9 55.6 54.2
MC – moisture content, BF – biofilter. BF 1, 2 are considered as replicate treatment, BF 3, 4 are considered as replicate controls.
collected as well. The samples were stored at −80 ◦ C before analyses. In total, 128 + 8 = 136 samples were collected. 2.3. DNA extraction and purification Biofilter media samples were lyophilized. Large woodchips were physically broken into small ones, and then were manually homogenized. DNA was extracted from 0.15 g dry media using a FastDNA SPIN Kit for Soil (MP Biomedicals, Solon, OH) according to the manufacturer’s protocol. Extracted DNA was then purified using cetyl trimethyl ammonium bromide (CTAB) extraction [44]. DNA concentration was adjusted to standard concentrations of 10 ng/l prior to analyses. 2.4. Microbial community analyses Bacterial community composition in biofilter media samples was assessed using automated ribosomal intergenic spacer analysis (ARISA) [45]. The polymerase chain reaction (PCR) was carried out in an Eppendorf MasterCycler Gradient (Eppendorf AG, Hamburg, Germany) and the reagents included 1× Tris buffer, 0.25 mg/ml BSA, 2.5 mM MgCl2 , 0.4 M of 1406 F-primer (5 -TGYACACACCGCCCGT3 ) and 23S R-primer (5 -GGGTTBCCCCATTCRG-3 ), 0.025 U/l Promega GoTaq and 2 l purified DNA (10 ng/l). PCR cycle was composed of: initial denaturation at 94 ◦ C for 120 s, 26 cycles of 94 ◦ C for 35 s, 55 ◦ C for 45 s, and 72 ◦ C for 2 min, followed by a final extension at 72 ◦ C for 120 s. DNA fragments generated from ARISA were analyzed by denaturing capillary electrophoresis using an ABI 3730XL Genetic Analyzer (PE Biosystems, Foster City, CA). Details about this method were described previously by Kent et al. [46]. Ammonia oxidizer community composition was analyzed using terminal restriction fragment length polymorphism (T-RFLP) analyses of the amoA gene encoding the catalytic ␣-subunit of archaeal ammonia monooxygenase [47]. A pre-test using real-time qPCR method showed that the abundance of archaeal amoA in the sampled media was about six times of bacterial amoA; therefore, only archaeal amoA community results were reported. PCR reagents contained 50 mM Tris buffer (pH 8.3), 100 g of BSA per ml, 1.5 mM MgCl2 , 200 M of each dNTP, 0.4 M of each primer (amoA-F: 5 -STAATGGTCTGGCTTAGACG-3 and amoA-R: 5 GCGGCCATCCATCTGTATGT-3 ) [48,49], 2.5 U of Taq polymerase, and 50 ng DNA in a final volume of 50 l. The amoA F-primer and R-primer was labeled with the fluorescent dye HEX and NED, respectively. PCR cycle was composed of: initial denaturation at 94 ◦ C for 300 s, 30 cycles of 94 ◦ C for 45 s, 53 ◦ C for 60 s, and 72 ◦ C for 60 s, then followed by a final extension at 72 ◦ C for 900 s. PCR products were purified using the MinElute PCR purification kit (Qiagen, Valencia, CA,), and then were digested with RsaI (New England Biolabs Inc., Ipswich, MA). The terminal restriction fragments labeled
L. Yang et al. / Journal of Hazardous Materials 271 (2014) 292–301
295
Table 2 percentages of chemical components and pH conditions in media.
Before testa After testb p-value
Total C
Organic matter
Total-N
Organic-N
NH3 -N
pH
45.3 ± 1.8 32.0 ± 2.6 ***
85.6 ± 6.6 66.2 ± 12.5 0.017*
1.3 ± 0.2 2.1 ± 0.7 0.045*
1.2 ± 0.2 1.9 ± 0.5 0.042*
0.03 ± 0.03 0.2 ± 0.2 0.112
8.1 ± 0.5 6.1 ± 0.4 ***
Data analyzed in the Midwest Laboratories, Omaha, NE. Significance level: ***: p < 0.001, *: p < 0.05. a Four replicates. b One sample from each layer of four biofilters, eight replicates total.
with fluorescence were analyzed by denaturing capillary electrophoresis using an ABI 3730XL Genetic Analyzer (PE Biosystems, Foster City, CA). Denitrifier community composition was also analyzed using TRFLP analyses. The nitrous oxide reductase gene (nosZ) are involved in the transformation of nitrous oxide to dinitrogen gas [50]. The PCR reaction mixtures were the same as above, except primers. The nosZ-F-1181 (5 -CGCTGTTCITCGACAGYCAG-3 ) and nosZ-R-1880 (5 -ATGTGCAKIGCRTGGCAGAA-3 ) were used. The nosZ F-primer was labeled with the phosphoramidite dye 6-FAM. Reaction cycle was composed of: initial denaturation at 94 ◦ C for 180 s, 25 cycles of 94 ◦ C for 45 s, 56 ◦ C for 60 s, and 72 ◦ C for 120 s, then followed by a final extension carried out at 72 ◦ C for 420 s. Similarly, the nosZ PCR products were purified, and then were digested in single-enzyme restriction digests containing either AluI or HhaI (New England Biolabs Inc., Ipswich, MA). The results obtained from AluI and HhaI digested fragment were combined together for community analyses. 2.5. Determination of functional gene abundance The amoA gene was amplified for quantitative PCR (qPCR) using primers Arch-amoAF (5 -STAATGGTCTGGCTTAGACG-3 ) and Arch-amoAR (5 -GCGGCCATCCATCTGTATGT-3 ) [51]; and the nosZ gene was amplified for qPCR using nosZ1527F, 5 -CGCTGTTCA CTTCGACAGCTCA-3 and nosZ1622R, 5 - CGCGAACGGGCAAGCAA GGTGCCG-3 [52]. PCR reactions (for both amoA and nosZ) contained 1× 5 l SYBR green master mix (Applied Biosystems Inc., Foster City, CA), 0.4 mM of each primer, 0.5 mg/ml BSA, and 10 ng of DNA in a final volume of 10 l. A same PCR amplification procedure was applied for both amoA and nosZ, which included an initial denaturation step at 95 ◦ C for 300 s, followed by 40 cycles of 95 ◦ C for 45 s, 56 ◦ C for 60 s, and 72 ◦ C for 60 s. Triplicates of each sample were analyzed. Standard curves of both amoA and nosZ amplicons were obtained based on serial dilutions of mixed PCR products of biofilter samples. Reactions were analyzed on a 384-well Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems Inc., Foster City, CA). 2.6. Statistical analyses Analysis of variance (ANOVA) was used to compare the physical and chemical properties. Abundance of nosZ genes was also compared using ANOVA. LSD Post hoc test was applied to evaluate the difference in ammonia removal efficiency and nitrous oxide concentration in the four operation steps. Bacterial and functional gene communities were analyzed using multivariate analyses. ARISA and T-RFLP data were Hellingertransformed in order to convert the raw density of each peak to a relative intensity [53]. Non-metric multidimensional scaling (NMDS) was applied to visualize the results [54,55]. The Bray–Curtis distance matrices were taken as NMDS inputs [54], and the centroid and standard deviation of each group are displayed in the ordinations [56]. PerMANOVA tests were carried out to examine
how moisture content affected the bacterial, amoA and nosZ communities [57]. To statistically compare two sample groups, pairwise analysis of similarity (ANOSIM) was conducted [58]. The analysis of multivariate homogeneity of group dispersions (using the PERMDISP2 procedure) was carried out to assess the homogeneity of microbial communities within a group of samples [59,60]. It generated a distance between 0 and 1, with longer distance showing lower homogeneity. Calculations of ANOSIM and NMDS were carried out using PRIMER 6 (Primer-E Ltd., Plymouth, United Kingdom); calculation of ANOVA, post hoc and PerMANOVA significance levels (function: adonis), and the group dispersion distances (function: betadisper) were performed in the R statistical environment [61] using functions found in the “MASS” and “vegan” packages. 3. Results 3.1. Physical and chemical properties of media In the tested four months, physical and chemical properties of media changed dramatically, mainly due to intensive chemical and biological processes (Table 2). Total-C and organic matter were consumed and significantly decreased. In contrast, total-N and organic-N concentrations increased significantly as a result of ammonia absorption and immobilization. The pH values decreased over time, most likely due to microbial activities. In general, nitrification process releases H+ and reduces pH [62], while NH3 absorption consumes H+ . Therefore, the balance of proton depends on the extent of nitrification. In this test, more H+ was released than consumed. Media became relatively smaller (Table 3). Microbial activities and aging process might be responsible for the size reduction. 3.2. Moisture management Since water was added onto the top surface of media, upper layers were wetter than lower layers. Moisture contents of the upper layers were relatively constant during steps 2–4, while moisture contents of the lower layers were managed to be different. In step 3, lower layers of BF1 and BF2 pair (59.7 ± 1.6%) were significantly wetter (p < 0.001) than the controls: BF3 and BF4 pair (48.8 ± 1.1%). Averaged moisture contents of the upper and lower layers of the four biofilters are listed in Table 1, and were quite close to the target moisture contents (Table 1). No leachate was observed during the operation.
Table 3 Percent of media size distribution in each size class, determined for 3 replicates.
before test After test p-value
>1.9 cm
0.8–1.9 cm
0.2–0.8 cm
