Environ. Sci. Technol. 2009, 43, 4273–4279
Spatial Heterogeneity of Denitrification Genes in a Highly Homogenous Urban Stream †,‡,|
CHARLES W. KNAPP, WALTER K. DODDS,§ K Y M B E R L Y C . W I L S O N , §,⊥ J O N A T H A N M . O ’ B R I E N , §,# A N D D A V I D W . G R A H A M * ,†,‡ Department of Civil, Environmental, and Architectural Engineering, University of Kansas, Lawrence, Kansas 66045, School of Civil Engineering and Geosciences, Newcastle University, Newcastle upon Tyne NE1 7RU, United Kingdom, and Division of Biology, Kansas State University, Manhattan, Kansas 66506
Received January 15, 2009. Revised manuscript received April 9, 2009. Accepted April 16, 2009.
Human modification of natural streams by urbanization has led to more homogeneous channel surfaces; however, the influence of channel simplification on in situ microbial distribution and function is poorly characterized. For example, denitrification, a microbial process that reduces soluble nitrogen (N) levels, requires peripheral anoxic zones that might be lost in artificial channels such as those with a concrete lining. To examine how microbial function might be influenced by channel simplification, we quantified denitrification rates and conditions in microbial mats within an urban concrete channel. We quantified spatial and diurnal patterns of nitrate uptake, diurnal dissolved oxygen (DO) levels, and nutrient conditions, along with the spatial distribution of DO, solids, chlorophyll a, and genes associated with denitrification (nirS and nirK), ammoniaoxidizing bacteria (AOB), cyanobacteria, and algal chloroplasts. Despite the channel being superficially homogeneous, nir genes were distributed in a patchy manner. Two types of gene patches were observed: one associated with nirK, which had diurnally variable DO levels and high nocturnal nitrate uptake rates, and the other associated with nirS, which had elevated AOB genes, thicker layers of mud, and an apparent 24 h nitrate uptake. All active nir patches had elevated microbial photosynthetic genes. Results imply that even artificial channels, with reduced macroscale heterogeneity, can sustain significant rates of denitrification, although the responsible communities vary with space and time. This patchiness has significant
* Corresponding author phone: +44-(0)-191-222-7930; fax: +44(0)-191-222-6502; e-mail:
[email protected]. † University of Kansas. ‡ Newcastle University. § Kansas State University. | Current address: David Livingstone Centre for Sustainability, Department of Civil Engineering, University of Strathclyde, Glasgow, United Kingdom G1 1XN. ⊥ Current address: Arizona Department of Water Resources, 3550 N. Central Ave., Phoenix, AZ 85012. # Current address: W.K. Kellogg Biological Station, Michigan State University, 3700 East Gull Lake Drive, Hickory Corners, MI 49060. 10.1021/es9001407 CCC: $40.75
Published on Web 05/07/2009
2009 American Chemical Society
implications to extending local data to landscape level predictions and field sampling strategies but also suggests alternate channel designs to increase N retention rates.
Introduction Two major and potentially interacting anthropogenic influences on streams are the alteration of natural stream channel integrity and nutrient pollution. For example, urbanization often leads to reduced stream channel heterogeneity by replacing natural surfaces with synthetic surfaces, like concrete, which tend to increase nitrogen (N) transport downstream because of reduced N retention capacity (1-4) and reduced biotic integrity (5). Although increased N transport also results from generally higher N inputs due to changing human activities, lower in-channel N processing rates are largely a consequence of the loss of stream habitat that promotes N retention reactions (6, 7) such as denitrification [i.e., anoxic conversion of nitrate (NO3-) to nitrogen gas (N2) that is released into the atmosphere (8)]. Specifically, simplified channels tend to have lower sediment-to-water ratios and fewer proximal areas of anoxia than natural channels (3, 4, 9) and theoretically sustain lower levels of in situ denitrification. The use of flow barriers has been shown to successfully increase water contact with benthos and has enhanced N removal (10, 11). However, such approaches are not often used, and the modification of natural streams to simplified channels via urban development continues to result in reduced denitrification rates and impaired stream N retention characteristics. This experiment was part of a project studying nitrification and denitrification processing rates in nine low-order, lowdischarge streams near Manhattan, Kansas (12) The nine streams in the larger study had a broad range of NO3- levels (from 0.9 to 21000 µg N/L) and very different physical characteristics and land uses, ranging from pristine prairie to agricultural to urban catchments. Overall, nitrification and denitrification rates were greater when soluble inorganic N levels were high. However, contrary to expectations, the stream with the highest N processing rates was the least natural channel in the study, a 3 m wide concrete-lined ditch (called the “Ditch”) near an urban commercial development (12). Furthermore, screening results showed Ditch denitrification rates were high, even when compared with those of 61 other streams surveyed across North America (7). These results were amazing given that the Ditch had no apparent habitat for denitrification (i.e., no deep sediment zones, barriers, or obvious regions of anoxia), although it did have patches of caked mud, green biofilms, and solid debris scattered along its bottom. Given the high denitrification rates that were contrary to expectations (1-4) and the patchy distribution of bottom debris in the Ditch, we decided to quantify the spatial distribution of genes in the channel related to denitrification to assess why such a channel might sustain elevated denitrification. As such, samples were collected and preserved for molecular characterization from 20 locations along and across the Ditch during a 15N tracer study to characterize microbial community conditions associated with diurnal NO-3 uptake in the channel. Quantitative real-time PCR (qPCR) was employed to quantify two different nitrite reductase genes associated with denitrification (i.e., copper-associated nirK and cytochrome-containing nirS) and 16S-rRNA gene sequences for cyanobacteria, plastids, ammonia-oxidizing bacteria (AOB), and “total” eubacteria. These data were then compared with each other, local habitat conditions, NO3VOL. 43, NO. 12, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4273
FIGURE 1. The Ditch, which is located below the parking lot of a commercial shopping area in Manhattan, Kansas. The channel is fed by urban stormwater drains upstream of the distant bridge and flow proceeds toward the camera, which is pointing north. The 15NO3-isotope addition point was approximately 3 m downstream of the bridge, and the picture was taken about 100 m downstream. The Ditch had a flow width of ∼3 m at the time of sampling. levels, and uptake rates to determine how the distribution of genes relate to in situ denitrification in the Ditch. Ultimately, the goals were to develop a deeper understanding how urbanization impacts microbial communities responsible for in-channel N retention and to translate that greater understanding into alternate channel designs that might enhance N retention in urban channels.
Materials and Methods Channel Conditions and Sampling Layout. The experiment site was an urban concrete ditch built to replace a historical channel of the Big Blue River located in an urban commercial zone in Manhattan, KS. The Ditch primarily receives Manhattan storm drainage, with flow rates ranging dramatically according to local rainfall events. Figure 1 presents the sampling reach, which extended from just below the visible bridge to a sample site 300 m downstream. This reach was segmented into six discrete sections, bracketing sampling sites at 0, 20, 40, 80, 150, 240, and 300 m downstream from the point of 15N isotope addition (mentioned later). Each of the seven stations was sampled at three cross-sectional points for molecular microbial and other analyses of the system. Two of the cross-sectional points were about 50 cm from each edge, and one point was at midstream. Specific samples were obtained in the morning of the day after isotope addition, resulting in 20 discrete points for analysis of gene abundances, chlorophyll a, and sediment solids (only two cross-sectional sites were sampled at 0 m). No rain events had occurred within three days of sampling and water levels in the channel were consistently low, although flow covered the entire channel bottom throughout the experiment. Water Chemistry and 15N Isotope Assays. The experiment was initiated on the morning of the first day by the addition of the 15NO3- isotope used for tracking 15N incorporation into different compartments in the stream (12). Background samples had been collected for all analyses on the previous day, and the 15NO3- isotope was added at noon on the first day at 0 m at a steady rate for 24 h (at 20 mL min-1 as K15NO3 to a target enrichment of 20000‰ 15NO3-) (7, 13). Samples were then collected at 1:00 a.m. and 11:00 a.m. the following days, while the isotope was still being added and preserved for nitrate concentration and isotopic analysis (i.e., NO3-, 15 15NO3 , dissolved N2, and 15N2O gases) (7, 12). Typically, samples for water column15NO3- determination were col4274
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 12, 2009
lected along the stream reach and then filtered (Whatman GF/F) and analyzed using an adapted version of the ammonia diffusion method following nitrate reduction (14) and mass spectrometry detection (ThermoFinnigan Delta Plus). Gaseous N samples were collected using plastic syringes and were allowed to equilibrate with 20 mL ultrapure helium for 5 min. The dissolved gases and helium were then collected in 12 mL exetainer vials, and levels of 15N2 and 15N2O were determined by mass spectrometry. Actual rates of denitrification were calculated using the longitudinal flux of the tracer 15N2 model as previously described (13). Stream water N levels were determined using the cadmium reduction method for NO3-, the indophenol method for NH4+, and the molybdate reduction method for soluble reactive phosphorus (SRP) on a Technicon autoanalyzer (15). Dissolved organic nitrogen (DON) was determined using persulfate digestion followed by NO3- determination. Benthic biofilms were analyzed for chlorophyll a using hot ethanol extraction (16), followed by quantification with an optimized Turner Model 112 fluorometer (Turner Designs, Inc., Sunnyvale, CA) (17). Sampling for Molecular and Other Characterization of Microbial Mats. The Ditch contained microbial mats, defined as visible biofilms, of varying thickness (usually approximately 1 cm or less). Samples for molecular characterization were collected at midday on the second day because it was thought communities present at that time would be best represent dentrification activity occurring during the 15NO3- addition window. As such, molecular and related analyses provided a “snapshot” of ambient microbial conditions during the isotope study, which were then compared with water quality, isotope, and other data measured around the time of sampling. Water column samples for molecular microbial analyses were collected aseptically into clean, presterilized amber glass bottles. Collection of the substrate microbial mat and sediment solids involved inserting a washed and ethanolsterilized PVC tube (15 cm diameter) into the stream to isolate a portion of the water column and sediment. This approach caused the mat and other solids to suspend in the sampler, which created a slurry sample reflecting the sediments and mat at each site. Upon return to the laboratory, replicate water column and slurry samples were centrifuged (10000 × g, 10 min; Fisher Scientific) and the pellets stored at -80 °C. Quantitative PCR Gene Detection. qPCR was used to quantify genes associated with different microbial guilds or functions in microbial mats in the channel. As such, a series of primers and probes were either constructed or adapted to quantify key microbial groups potentially pertinent to denitrification within the system, which are summarized in Table 1. Primers and probes (Sigma-Genosys, The Woodlands, TX) were employed that target semiconserved regions of 16S-rRNA genes specific to clades of ammonia-oxidizing bacteria (18, 19), cyanobacteria, plastids (which closely resemble cyanobacteria) (20, 21), and total eubacteria (22-24), whereas primers for the nitrite reductase genes, nirS and nirK, were used to determine denitrification potential (25). Although slightly more comprehensive nirS and nirK primer combinations have since been developed (26), they were not available at the time of analysis on this project. Fortunately, the earlier probes have been used successfully elsewhere and were considered adequate to observe general trends, which was the goal of this study. DNA was extracted using the MoBio UltraClean Soil DNA kit (Carlsbad, CA) with some modifications. Filters, beads, and extraction buffer were combined and agitated using a FastPrep (Qbiogene, Irvine, CA) cell disruptor for 30 s (5 speed) to lyse cells. Samples were then incubated at 70 °C for 10 min to further enhance cell lysis and reagitated for 15 s (4.5 speed). The remaining purification steps followed the manufacturer protocols. Reactions were performed using a
TABLE 1. PCR Reaction Conditions and Primer/Probe Sequences Used target nirK nitrite reductase nirS nitrite reductase eubacterial 16S-rRNA gene
primer/probe (concentration) nirK1F (700 nM) nirK5R (700 nM) nirS1F (700 nM) nirS6R (700 nM) 1055f (600 nM) 1039r (600 nM) 16STaq1115 (250 nM)
sequence (5′ s 3′)a
annealing elongation references conditionsb conditionsb
GGMATGGTKCCSTGGCA 49 °C/40s 72 °C/20s GCCTCGATCAGRTTRTGG CCTAYTGGCCGCCRCART 49 °C/40s 72 °C/20s CGTTGAACTTRCCGGT ATGGCTGTCGTCAGCT 50 °C/60s 72 °C/20s ACGGGCGGTGTGTAC FAM-CAACGAGCGCAACCC-TAMRAc
ammonia-oxidizing bacterial CTO 189f A/B (300 nM) GGAGRAAAGCAGGGGATCG 16S-rRNA gene CTO 189f C (300 nM) GGAGGAAAGTAGGGGATCG RT1 (300 nM) CGTCCTCTCAGACCARCTACTG FAM-CAACTAGCTAATCAGRCATCR TMP1 (Taq) (125 nM) GCCGCTC-TAMRAc cyanobacterial/plastid Cya359F (500 nM) GGGGAATYTTCCGCAATGGG 16S-rRNA gene Cya781R(a) (500 nM)d GACTACTGGGGTATCTAATCCCATT Cya781R(b) (500 nM)e GACTACAGGGGTATCTAATCCCTTT
60 °C/60s
25
22, 23 23, 24 23 18, 19 18
58 °C/60s 78 °C/20s
20, 21
a Nucleotide degeneracies: M ) (A or C), R ) (A or G), W ) (A or T), S ) (C or G), Y ) (C or T), and K ) (G or T). b Many methods were reoptimized for the BioRad iCycler. c FAM ) 6-carboxyfluorescein 5′-fluorophore. TAMRA ) carboxyetramethylrhodamine 3′-quencher dye. d CyanA targets Nostoc spp., Oscillatoria spp. and plastid 16S-rRNA. e CyanB targets other cyanobacterial 16S-rRNA, specifically those not targeted with Cya781R(a).
BioRad iCycler (Hercules, CA) with iCycler fluorescence detector and software 2.3 (BioRad). Each 25 µL reaction mixture combined 2 µL of template DNA, primers, and Taqman probe (if used) with the iQ Supermix PCR reagent (BioRad). Reaction conditions involved initial DNA denaturation for 10 min at 95 °C and then 40 cycles of denaturation (20-30s) at 94 °C, primer annealing (50-60 °C), elongation (60-72 °C), and fluorescence detection. Reaction specifics are provided in Table 1. When provided, Taqman probes were used; otherwise, reactions were monitored using SYBR Green I. Purified plasmid DNA (High Pure Plasmid Isolation Kit, Roche Dignostics, Indianapolis, IN), containing cloned targeted gene fragments (TOPO-TA, Invitrogen, Carlsbad, CA), were diluted by 108 to 101 copies per microliter to provide quantitative DNA standards. Targets were verified by gel electrophoresis and DNA sequencing. The presence of inhibitory substances in the sample matrix were checked by spiking the samples with known amounts of template DNA and comparing differences in concentration threshold values (CT) between the matrix and controls (differences were always less than one cycle, implying no major inhibition). PCR efficiencies were further examined by comparing serial dilutions of selected samples (those with high levels of DNA) and plasmid controls, typically between 80-105%. Correlation coefficients were always >0.98 for calibration curves, and log gene abundances were within the linear range of the calibration curve. Sediment Solids. A plastic column sleeve was placed over the sediment (similar to mat sampling), and the sediment was mixed into the water column. A known volume of each sample was filtered onto preweighed, precombusted glass fiber filters and weighed after drying overnight at 105 °C for determination of dry matter (DM) and then ignited at 550 °C for determination of ash-free dry matter (AFDM). Dissolved Oxygen Profiles. Water column O2 levels (DO) were monitored using a YSI Multiparameter sonde. Detailed DO profiles were obtained with a needle-encased cathodetype microelectrode with a 5 µm diameter sensing tip that was not sensitive to water velocity (9). About 20 profiles per cross section were collected at the 80, 150, 240, and 300 m sampling sites. Data Analysis for Mapping. Data mapping was performed using the 3D Smoothing algorithm in SigmaPlot (version 10;
Systat Software, Inc., San Jose, CA) based on untransformed gene abundances, chlorophyll a, and solid data for the 20 sites. Gene abundances in the microbial mat were determined by taking the difference between the total resuspended and water column sample concentrations. The surface density of genes was then calculated based on the liquid concentrations of genes, volume of the contained water column, and measured water depth. The data for the individual sites were then combined and “pseudo-topographic” maps were developed to display the spatial distribution of key genes and parameters relative to the channel surface.
Results and Discussion Site Characterization. Figure 1 shows that the channel bottom was uniform concrete, and water flow was bankto-bank, although the extent and type of surface “debris” differed locally. On the basis of the sampling along and across the reach, Figure 2 was generated, which “topographically” presents the spatial distribution of solids, chlorophyll a, and gene abundances in the channel. Figure 2 shows greater variability in surface coverage from a physical and biological perspective than was initially apparent but was generally consistent with physical field observations. For example, the 20 m sampling site had >1 cm thick caked mud that was fringed with green algae on its east bank, whereas the 40 m site had little mud but was largely covered by a ∼1 cm thick green microbial mat. Furthermore, the 80 m station was covered by an ∼1.0 cm thick layer of mud interspersed with filamentous cyanobacteria, and stations at 150 and 240 m had thin (∼2-3 mm) mat coatings across their sections. The downstream station at 300 m had various debris across its section. Nitrogen Dynamics along the Stream Reach. Water column nutrient conditions are summarized in Figure 3 and Table S1 of the Supporting Information. Concentrations of NO3- at the top of the reach were high, ranging from 430 to 610 µg N/L at the 20 m station (Figure 3A) and were quite consistent over the experiment due to steady input from urban runoff above the reach. However, NO3- levels declined rapidly with distance downstream as indicated by both daytime and nighttime measurements. Observed net NO3uptake rates were highest at night between 80 and 150 m stations along the reach (Figure 3B). Furthermore, the gross NO3- uptake rate (i.e., net NO3- uptake, corrected for NO3VOL. 43, NO. 12, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4275
FIGURE 2. Topographic maps of gene abundances, chlorophyll a, and sediment solid patterns in the 300 m study reach. The maps were generated using SigmaPlot (version 10; Systat Software, Inc., San Jose, CA), using untransformed gene and other data. Relative abundances are noted by color, with orange typically indicating locally high levels, which are indicated in the legend below each map. North arrow indicates the orientation of the relative stream.
FIGURE 3. Typical ambient day and night nitrate (NO3-) concentrations and uptake rates over the isotope addition experiment, including (A) NO3- levels, (B) net and gross NO3- uptake rates. (C) 15NO3-, (D) 15N2O, and (E) 15N2 concentrations are provided. Replicate measurements are shown in each graph. produced by nitrification during the measurements estimated from 15NO3- data) was also elevated above 150 m, especially 4276
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 12, 2009
between 40 and 80 m. Elevated nocturnal denitrification is clearly apparent in panels C-E of Figure 3, which show high
TABLE 2. Correlation Coefficients between Measured Genes, Solids, and Other Water Quality Parameters, and Daytime and Nighttime Net NO3- Uptake Rates day
night genes, solids, and chlorophyll aa
nirS/16S (0.555)** cyanB/16S (0.475)** cyanA/16S (0.467)* DM (0.450)* 16S (-0.441)*
chlorophyll a (0.801)** cyanB (0.578)** AFDM (0.490)** nirK (0.445)*
nutrientsb nonec
NH4+ (0.951)** NO3- (0.921)** SRP (0.866)* DON (-0.840)* DOC (-0.835)*
a Data from bivariate correlation analysis between day and night net NO3- uptake rates, respectively, and gene and solid levels (both absolute values and values normalized to 16S-rRNA gene abundance at each point) for the 20 sampling locations in the grid. ** and * indicate p values
