Microb Ecol DOI 10.1007/s00248-015-0605-8
ENVIRONMENTAL MICROBIOLOGY
Seasonal Patterns in Microbial Community Composition in Denitrifying Bioreactors Treating Subsurface Agricultural Drainage Matthew D. Porter 2,4 & J. Malia Andrus 1,5 & Nicholas A. Bartolerio 2,6 & Luis F. Rodriguez 1 & Yuanhui Zhang 1 & Julie L. Zilles 2 & Angela D. Kent 3
Received: 17 November 2014 / Accepted: 23 March 2015 # Springer Science+Business Media New York 2015
Abstract Denitrifying bioreactors, consisting of water flow control structures and a woodchip-filled trench, are a promising approach for removing nitrate from agricultural subsurface or tile drainage systems. To better understand the seasonal dynamics and the ecological drivers of the microbial communities responsible for denitrification in these bioreactors, we employed microbial community Bfingerprinting^ techniques in a time-series examination of three denitrifying bioreactors over 2 years, looking at bacteria, fungi, and the denitrifier functional group responsible for the final step of complete denitrification. Our analysis revealed that microbial community composition responds to depth and seasonal variation in moisture content and inundation of the bioreactor media, as well as temperature. Using a geostatistical analysis approach, we observed recurring temporal patterns in bacterial and Electronic supplementary material The online version of this article (doi:10.1007/s00248-015-0605-8) contains supplementary material, which is available to authorized users. * Angela D. Kent [email protected] 1
Department of Agricultural and Biological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA
2
Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA
3
Department of Natural Resources and Environmental Sciences, University of Illinois at Urbana-Champaign, Urbana, IL, USA
4
Present address: Environmental Resources Management, 1701 Golf Road, Suite 1-700, Rolling Meadows, IL 60008, USA
5
Present address: Waterborne Environmental, Inc., 2001 S First Street, Suite 109, Champaign, IL 61820, USA
6
Present address: Strand Associates, Inc., 910 West Wingra Drive, Madison, WI 53715, USA
denitrifying bacterial community composition in these bioreactors, consistent with annual cycling. The fungal communities were more stable, having longer temporal autocorrelations, and did not show significant annual cycling. These results suggest a recurring seasonal cycle in the denitrifying bioreactor microbial community, likely due to seasonal variation in moisture content. Keywords Denitrification . Geostatistics . Nitrate removal . Seasonal variation . Subsurface drainage
Introduction Denitrifying bioreactors, consisting of water flow control structures and a woodchip-filled trench placed in-line with drainage tile, are a promising approach for removing nitrate from tile drain effluent (reviewed in [1]). Due to the widespread use of tile drains in agricultural fields in the Midwestern USA, where they drain an estimated 20 million hectares (ha) of agricultural land [2], and to the role of tile drains as a direct conduit for nitrate transport between agricultural fields and receiving surface waters [2, 3], widespread installation and effective performance of these bioreactors could result in a substantial reduction in agricultural nitrate pollution. Although many field studies have demonstrated good performance of these denitrifying bioreactors, concerns about performance and potential undesirable effects such as leaching of organic matter and nitrous oxide production remain [1]. Previous bioreactor studies have focused primarily on design aspects, including carbon substrate selection (e.g., [4, 5]), hydrology (e.g., [6, 7]), and longevity (e.g., [8–10]), with limited investigation of the microorganisms responsible for nitrate removal in these engineered environments. There is, however, strong evidence that nitrate removal in the
M. D. Porter et al.
bioreactors is primarily due to denitrification, based on stable isotope experiments [5] and measured ammonium concentrations that are too low for anammox and dissimilatory nitrate reduction to account for the observed nitrate removal (e.g., [11, 12]). One study of laboratory-scale denitrifying bioreactors employing molecular techniques found a correlation between nitrate removal and the abundance of genes involved in denitrification [12]. In the same study, the ratio between nir and nosZ gene abundance was higher in samples with higher nitrous oxide production. In the field, the bacterial community composition exhibited spatial patterns along the flow path and between depths of a bioreactor, but no spatial patterns were discerned for the denitrifying bacteria [13]. To our knowledge, the temporal dynamics of denitrifying bioreactor microbial communities have not been investigated, despite the strong seasonal patterns of tile drain hydrology. Due to precipitation and uptake by crops, in a typical year, tile drain flow begins in late fall and ceases in early summer [6, 14, 15]. Although little is known about the microbial ecology of denitrifying bioreactors, denitrification has been widely studied in other environments (reviewed in [16, 17]). Environmental parameters known to affect denitrification and the denitrifying community include carbon, moisture, and temperature (reviewed in [16, 17]). However, predictive relationships have yet to be established among these environmental parameters and denitrifier abundance and composition. Seasonal dynamics of denitrifiers in agricultural soils have been linked to freeze-thaw cycles, likely due to changes in temperature and soil moisture [18]. In comparison to agricultural fields, the bioreactors provide an interesting decoupling of season, temperature, and moisture content, as much of the bioreactor volume is typically below the frost line and thus should be buffered from temperature extremes by the surrounding soil, and the winter/spring tile water flow has a different seasonal pattern than precipitation, which is typically heaviest in April– June in Illinois. Furthermore, increasing evidence of substantial contributions from fungal denitrifiers [19, 20] and bacteria containing the atypical or clade II nitrous oxide reductase [21, 22] in some environments raises the possibility that such relationships have not been identified due to incomplete coverage of the commonly used molecular methods. For the denitrifying bioreactors, inhibitor experiments suggest that the denitrification is primarily bacterial, but inhibition of fungi also affected denitrifying enzyme activity, possibly via organic carbon availability [23]. Seasonal and annual patterns in microbial processes have been observed previously, particularly in aquatic and marine systems [24–29], where they were attributed to seasonal changes in temperature, as well as redox conditions, resource availability, and food web interactions. Seasonal patterns have also been reported in terrestrial systems [30], but the few studies that have examined microbial community structure on multiyear time scales in terrestrial systems do not report
predictable annual patterns in community composition [31, 32]. Although temporal studies such as these can be done with either fingerprinting or high-throughput sequencing methods, due to the large number of samples required, the sequencing approach is rare. Of the above references, only one used this approach [30]. More importantly, in studies comparing the two approaches, similar conclusions about patterns of biodiversity were reached with both methods [33, 34]. In the current study, we therefore opted to use fingerprinting methods, allowing analysis of a greater number of samples and finer temporal resolution. The objective of this study was to investigate temporal dynamics in the composition and abundance of the microbial communities responsible for bioreactor function in field-scale denitrifying bioreactors and to relate those dynamics to a suite of environmental parameters. This was accomplished through characterization of the bacteria, the fungi, and the denitrifier functional group in samples taken over 2 years from three denitrifying biofilters. The insight gained from this study will ultimately enhance our understanding of the microbial ecology of denitrification and our ability to manage such bioreactors for effective and reliable performance.
Materials and Methods Study Sites Samples (water and woodchips) were collected from three active denitrifying bioreactors located in central Illinois (Table 1). Woodchip sampling ports were installed in each bioreactor to allow for woodchip samples to be repeatedly collected from defined locations and depths. Sampling ports were constructed from PVC and consisted of a perforated 10.16-cm outer diameter (OD) casing installed vertically in the bioreactor in a fixed location and a perforated 7.62-cm OD woodchip-containing removable inner sampling port [35]. Sampling ports were installed at two depths, 0.76 and 1.52 m (2.5 and 5 ft, respectively) below ground surface, corresponding to approximately half and full depth of the bioreactor, respectively. Sampling ports were installed at 6.1m increments along the water flow path—spacing previously determined to provide optimal spatial resolution of microbial community structure [13]. Sample Collection Woodchip and water samples were collected from January 2009 through March 2010 (DE01 bioreactor) or December 2010 (FP07 and FP03 bioreactors). The initial sampling frequency was weekly, while later sampling dates were 2– 4 weeks apart. A total of 33 sample dates were included for the DE01 bioreactor, 38 for the FP03 bioreactor, and 37 for the
Seasonal Patterns in Denitrifying Bioreactors Table 1 Bioreactor
Study site location, design, size, and sampling information Location
Installation date
Shape
Drainage area (ha)
Bioreactor dimensions Width (m)
FP07
Decatur, IL
Summer 2006
Length (m)
Dates sampled Depth (m)
Square
2.0
6.1
6.1
1.5
Jan 2009–Dec 2010
FP03
Decatur, IL
Summer 2005
L-shape
5.0
1.5
30.5
1.8
Jan 2009–Dec 2010
DE01
De Land, IL
Fall 2007
Rectangle
6.1
3.1
12.4
2.2
Jan 2009–Mar 2010
FP07 bioreactor. For sampling ports containing water, water samples were collected for nitrate measurements in acidwashed 100-mL Nalgene bottles (Thermo Fisher Scientific, Rochester, NY). Water samples were transported to the lab on ice, preserved with sulfuric acid (0.047 mol/L), and stored at 4 °C until analysis. Woodchip samples were collected from the bottom of each inner PVC sampling port and placed into sterile 250-mL Nalgene bottles (Thermo Fisher Scientific, Rochester, NY). After sampling, surficial woodchips from the area immediately adjacent to the sampling ports were added to the top of the inner sampling column to replace the volume of woodchips removed during sample collection. All samples were stored on ice during transport. Upon returning from the field, 110 mL of Ringer’s solution (Oxoid, Cambridge, UK) was added to 30 g of woodchip sample. Samples were shaken at a moderate speed for 8–12 h at 30 °C. The woodchips were removed, and the woodchip wash for each sample was centrifuged at 5000×g for 3 min to concentrate microorganisms. Phosphatebuffered saline (PBS) (2.5 mL, Fisher Bioreagents #BP665-1) and five autoclaved 5-mm glass beads were added to each pellet, vortexed for 2 min, and centrifuged at 750×g for 5 min to remove debris. The supernatant was collected and stored at −20 °C. Environmental Variables Meteorological data included temperature and precipitation values for the day of sampling, as well as averaged values for the week preceding sampling and were obtained from the online weather database Weather Underground (http://www. wunderground.com/). Water flow data were measured using MULTI MINI-SAT™ field stations (Automata Inc. Nevada City, CA) with V-notch weirs, pressure transducers, and data loggers placed at both the inlet and outlet flow control structures of each bioreactor. Water flow values were reported hourly and were accessible via an online database managed by the Agri Drain Corporation (Agri Drain Corporation, Adair, IA). Instantaneous flow data were extrapolated for the hour reported and then summed for a given calendar day, producing a daily flow value in liters per day. Rolling averages were then used to calculate 5-day average flow, 7-day average flow, and 14-day average flow.
Average hydraulic retention time (HRT) was calculated by multiplying the volume of the bioreactor by the average woodchip porosity (n=0.57), divided by the average sum of daily influent flow for a given seasonal classification (Jan–Jun or Jul–Dec); days of no flow were included in the average sum of daily flow with a value of 0. Grab samples of influent and effluent water were collected from each bioreactor, and aqueous nitrate concentrations were measured by the Agricultural and Biological Engineering Water Quality Laboratory (University of Illinois, Urbana, IL) using US EPA method 353.1. Woodchip moisture content was measured by weighing woodchip samples before and after drying overnight at 105 °C. Molecular Analyses DNA was extracted from woodchip sample washes using a FastDNA Spin Kit (MP Biomedicals, Solon, OH) following the manufacturer’s instructions. Due to humic acid contamination, extracted DNA was further purified using a cetyltrimethylammonium bromide (CTAB) cleanup method [36]. Extracted DNA was quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA) and standardized to a final concentration of 10 ng/μL. Molecular community analyses included automated ribosomal intergenic spacer analysis (ARISA) for the bacterial community [37], fungal automated ribosomal intergenic spacer analysis (FARISA) [38] for the fungal community, and terminal restriction fragment length polymorphism (T-RFLP) of the functional gene for nitrous oxide reductase (nosZ) to assess the community structure of the bacteria responsible for the final step of denitrification [36, 39]. We focus here on the typical nosZ rather than the recently described atypical or clade II nosZ [21, 22] because qPCR comparing the abundance of typical vs. atypical nosZ revealed very low abundance (
ENVIRONMENTAL MICROBIOLOGY
Seasonal Patterns in Microbial Community Composition in Denitrifying Bioreactors Treating Subsurface Agricultural Drainage Matthew D. Porter 2,4 & J. Malia Andrus 1,5 & Nicholas A. Bartolerio 2,6 & Luis F. Rodriguez 1 & Yuanhui Zhang 1 & Julie L. Zilles 2 & Angela D. Kent 3
Received: 17 November 2014 / Accepted: 23 March 2015 # Springer Science+Business Media New York 2015
Abstract Denitrifying bioreactors, consisting of water flow control structures and a woodchip-filled trench, are a promising approach for removing nitrate from agricultural subsurface or tile drainage systems. To better understand the seasonal dynamics and the ecological drivers of the microbial communities responsible for denitrification in these bioreactors, we employed microbial community Bfingerprinting^ techniques in a time-series examination of three denitrifying bioreactors over 2 years, looking at bacteria, fungi, and the denitrifier functional group responsible for the final step of complete denitrification. Our analysis revealed that microbial community composition responds to depth and seasonal variation in moisture content and inundation of the bioreactor media, as well as temperature. Using a geostatistical analysis approach, we observed recurring temporal patterns in bacterial and Electronic supplementary material The online version of this article (doi:10.1007/s00248-015-0605-8) contains supplementary material, which is available to authorized users. * Angela D. Kent [email protected] 1
Department of Agricultural and Biological Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA
2
Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA
3
Department of Natural Resources and Environmental Sciences, University of Illinois at Urbana-Champaign, Urbana, IL, USA
4
Present address: Environmental Resources Management, 1701 Golf Road, Suite 1-700, Rolling Meadows, IL 60008, USA
5
Present address: Waterborne Environmental, Inc., 2001 S First Street, Suite 109, Champaign, IL 61820, USA
6
Present address: Strand Associates, Inc., 910 West Wingra Drive, Madison, WI 53715, USA
denitrifying bacterial community composition in these bioreactors, consistent with annual cycling. The fungal communities were more stable, having longer temporal autocorrelations, and did not show significant annual cycling. These results suggest a recurring seasonal cycle in the denitrifying bioreactor microbial community, likely due to seasonal variation in moisture content. Keywords Denitrification . Geostatistics . Nitrate removal . Seasonal variation . Subsurface drainage
Introduction Denitrifying bioreactors, consisting of water flow control structures and a woodchip-filled trench placed in-line with drainage tile, are a promising approach for removing nitrate from tile drain effluent (reviewed in [1]). Due to the widespread use of tile drains in agricultural fields in the Midwestern USA, where they drain an estimated 20 million hectares (ha) of agricultural land [2], and to the role of tile drains as a direct conduit for nitrate transport between agricultural fields and receiving surface waters [2, 3], widespread installation and effective performance of these bioreactors could result in a substantial reduction in agricultural nitrate pollution. Although many field studies have demonstrated good performance of these denitrifying bioreactors, concerns about performance and potential undesirable effects such as leaching of organic matter and nitrous oxide production remain [1]. Previous bioreactor studies have focused primarily on design aspects, including carbon substrate selection (e.g., [4, 5]), hydrology (e.g., [6, 7]), and longevity (e.g., [8–10]), with limited investigation of the microorganisms responsible for nitrate removal in these engineered environments. There is, however, strong evidence that nitrate removal in the
M. D. Porter et al.
bioreactors is primarily due to denitrification, based on stable isotope experiments [5] and measured ammonium concentrations that are too low for anammox and dissimilatory nitrate reduction to account for the observed nitrate removal (e.g., [11, 12]). One study of laboratory-scale denitrifying bioreactors employing molecular techniques found a correlation between nitrate removal and the abundance of genes involved in denitrification [12]. In the same study, the ratio between nir and nosZ gene abundance was higher in samples with higher nitrous oxide production. In the field, the bacterial community composition exhibited spatial patterns along the flow path and between depths of a bioreactor, but no spatial patterns were discerned for the denitrifying bacteria [13]. To our knowledge, the temporal dynamics of denitrifying bioreactor microbial communities have not been investigated, despite the strong seasonal patterns of tile drain hydrology. Due to precipitation and uptake by crops, in a typical year, tile drain flow begins in late fall and ceases in early summer [6, 14, 15]. Although little is known about the microbial ecology of denitrifying bioreactors, denitrification has been widely studied in other environments (reviewed in [16, 17]). Environmental parameters known to affect denitrification and the denitrifying community include carbon, moisture, and temperature (reviewed in [16, 17]). However, predictive relationships have yet to be established among these environmental parameters and denitrifier abundance and composition. Seasonal dynamics of denitrifiers in agricultural soils have been linked to freeze-thaw cycles, likely due to changes in temperature and soil moisture [18]. In comparison to agricultural fields, the bioreactors provide an interesting decoupling of season, temperature, and moisture content, as much of the bioreactor volume is typically below the frost line and thus should be buffered from temperature extremes by the surrounding soil, and the winter/spring tile water flow has a different seasonal pattern than precipitation, which is typically heaviest in April– June in Illinois. Furthermore, increasing evidence of substantial contributions from fungal denitrifiers [19, 20] and bacteria containing the atypical or clade II nitrous oxide reductase [21, 22] in some environments raises the possibility that such relationships have not been identified due to incomplete coverage of the commonly used molecular methods. For the denitrifying bioreactors, inhibitor experiments suggest that the denitrification is primarily bacterial, but inhibition of fungi also affected denitrifying enzyme activity, possibly via organic carbon availability [23]. Seasonal and annual patterns in microbial processes have been observed previously, particularly in aquatic and marine systems [24–29], where they were attributed to seasonal changes in temperature, as well as redox conditions, resource availability, and food web interactions. Seasonal patterns have also been reported in terrestrial systems [30], but the few studies that have examined microbial community structure on multiyear time scales in terrestrial systems do not report
predictable annual patterns in community composition [31, 32]. Although temporal studies such as these can be done with either fingerprinting or high-throughput sequencing methods, due to the large number of samples required, the sequencing approach is rare. Of the above references, only one used this approach [30]. More importantly, in studies comparing the two approaches, similar conclusions about patterns of biodiversity were reached with both methods [33, 34]. In the current study, we therefore opted to use fingerprinting methods, allowing analysis of a greater number of samples and finer temporal resolution. The objective of this study was to investigate temporal dynamics in the composition and abundance of the microbial communities responsible for bioreactor function in field-scale denitrifying bioreactors and to relate those dynamics to a suite of environmental parameters. This was accomplished through characterization of the bacteria, the fungi, and the denitrifier functional group in samples taken over 2 years from three denitrifying biofilters. The insight gained from this study will ultimately enhance our understanding of the microbial ecology of denitrification and our ability to manage such bioreactors for effective and reliable performance.
Materials and Methods Study Sites Samples (water and woodchips) were collected from three active denitrifying bioreactors located in central Illinois (Table 1). Woodchip sampling ports were installed in each bioreactor to allow for woodchip samples to be repeatedly collected from defined locations and depths. Sampling ports were constructed from PVC and consisted of a perforated 10.16-cm outer diameter (OD) casing installed vertically in the bioreactor in a fixed location and a perforated 7.62-cm OD woodchip-containing removable inner sampling port [35]. Sampling ports were installed at two depths, 0.76 and 1.52 m (2.5 and 5 ft, respectively) below ground surface, corresponding to approximately half and full depth of the bioreactor, respectively. Sampling ports were installed at 6.1m increments along the water flow path—spacing previously determined to provide optimal spatial resolution of microbial community structure [13]. Sample Collection Woodchip and water samples were collected from January 2009 through March 2010 (DE01 bioreactor) or December 2010 (FP07 and FP03 bioreactors). The initial sampling frequency was weekly, while later sampling dates were 2– 4 weeks apart. A total of 33 sample dates were included for the DE01 bioreactor, 38 for the FP03 bioreactor, and 37 for the
Seasonal Patterns in Denitrifying Bioreactors Table 1 Bioreactor
Study site location, design, size, and sampling information Location
Installation date
Shape
Drainage area (ha)
Bioreactor dimensions Width (m)
FP07
Decatur, IL
Summer 2006
Length (m)
Dates sampled Depth (m)
Square
2.0
6.1
6.1
1.5
Jan 2009–Dec 2010
FP03
Decatur, IL
Summer 2005
L-shape
5.0
1.5
30.5
1.8
Jan 2009–Dec 2010
DE01
De Land, IL
Fall 2007
Rectangle
6.1
3.1
12.4
2.2
Jan 2009–Mar 2010
FP07 bioreactor. For sampling ports containing water, water samples were collected for nitrate measurements in acidwashed 100-mL Nalgene bottles (Thermo Fisher Scientific, Rochester, NY). Water samples were transported to the lab on ice, preserved with sulfuric acid (0.047 mol/L), and stored at 4 °C until analysis. Woodchip samples were collected from the bottom of each inner PVC sampling port and placed into sterile 250-mL Nalgene bottles (Thermo Fisher Scientific, Rochester, NY). After sampling, surficial woodchips from the area immediately adjacent to the sampling ports were added to the top of the inner sampling column to replace the volume of woodchips removed during sample collection. All samples were stored on ice during transport. Upon returning from the field, 110 mL of Ringer’s solution (Oxoid, Cambridge, UK) was added to 30 g of woodchip sample. Samples were shaken at a moderate speed for 8–12 h at 30 °C. The woodchips were removed, and the woodchip wash for each sample was centrifuged at 5000×g for 3 min to concentrate microorganisms. Phosphatebuffered saline (PBS) (2.5 mL, Fisher Bioreagents #BP665-1) and five autoclaved 5-mm glass beads were added to each pellet, vortexed for 2 min, and centrifuged at 750×g for 5 min to remove debris. The supernatant was collected and stored at −20 °C. Environmental Variables Meteorological data included temperature and precipitation values for the day of sampling, as well as averaged values for the week preceding sampling and were obtained from the online weather database Weather Underground (http://www. wunderground.com/). Water flow data were measured using MULTI MINI-SAT™ field stations (Automata Inc. Nevada City, CA) with V-notch weirs, pressure transducers, and data loggers placed at both the inlet and outlet flow control structures of each bioreactor. Water flow values were reported hourly and were accessible via an online database managed by the Agri Drain Corporation (Agri Drain Corporation, Adair, IA). Instantaneous flow data were extrapolated for the hour reported and then summed for a given calendar day, producing a daily flow value in liters per day. Rolling averages were then used to calculate 5-day average flow, 7-day average flow, and 14-day average flow.
Average hydraulic retention time (HRT) was calculated by multiplying the volume of the bioreactor by the average woodchip porosity (n=0.57), divided by the average sum of daily influent flow for a given seasonal classification (Jan–Jun or Jul–Dec); days of no flow were included in the average sum of daily flow with a value of 0. Grab samples of influent and effluent water were collected from each bioreactor, and aqueous nitrate concentrations were measured by the Agricultural and Biological Engineering Water Quality Laboratory (University of Illinois, Urbana, IL) using US EPA method 353.1. Woodchip moisture content was measured by weighing woodchip samples before and after drying overnight at 105 °C. Molecular Analyses DNA was extracted from woodchip sample washes using a FastDNA Spin Kit (MP Biomedicals, Solon, OH) following the manufacturer’s instructions. Due to humic acid contamination, extracted DNA was further purified using a cetyltrimethylammonium bromide (CTAB) cleanup method [36]. Extracted DNA was quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA) and standardized to a final concentration of 10 ng/μL. Molecular community analyses included automated ribosomal intergenic spacer analysis (ARISA) for the bacterial community [37], fungal automated ribosomal intergenic spacer analysis (FARISA) [38] for the fungal community, and terminal restriction fragment length polymorphism (T-RFLP) of the functional gene for nitrous oxide reductase (nosZ) to assess the community structure of the bacteria responsible for the final step of denitrification [36, 39]. We focus here on the typical nosZ rather than the recently described atypical or clade II nosZ [21, 22] because qPCR comparing the abundance of typical vs. atypical nosZ revealed very low abundance (
