Citation
Community composition of nitrite reductase genes in an acid mine drainage environment

Material Information

Title:
Community composition of nitrite reductase genes in an acid mine drainage environment
Creator:
Wise, Ben ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (56 pages) : ;

Subjects

Subjects / Keywords:
Acid mine drainage ( lcsh )
Water quality ( lcsh )
Denitrification ( lcsh )
Acid mine drainage ( fast )
Denitrification ( fast )
Water quality ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
High elevation, mountainous regions have a high concentration of mining activities and resulting acid mine drainage (AMD) that is typically acidic and often contains elevated concentrations of metals. The impacts of AMD on denitrifying microbial communities is not well understood, despite these organisms' central role in the nitrogen cycle, contribution to greenhouse gas production, and potential to provide ecosystem services through the mitigation of nitrogen pollution. This study examined denitrifying microbes across four regions within the Iron Springs Mining District (13 sites over four time-points) located in Southwest Colorado at high elevation that receive AMD or naturally-occurring acid rock drainage (ARD). Denitrification functional gene sequences (nirS and nirK coding for nitrite reductase) had a high number of observed OTUs (260 for nirS and 253 for nirK) and were observed at sites with pH as low as 3.2, dissolved oxygen as low as 1.0 mg/L, and metals>10 mg/L (including aluminum, iron, manganese, and zinc). A majority of the nirK and nirS OTUs (>60%) were present in only one sampling region. Approximately 8% of the nirK and nirS OTUs had a more cosmopolitan distribution with presence in three or more regions. Phylogenetically related OTUs were found across sites with very different chemistry. The total nirS community structure was correlated to iron, conductivity, sodium, and calcium, which may suggest that these factors play an important role in shaping the nirS community. Overall, these findings improve upon our understanding of the potential for denitrification within an ecosystem impacted by AMD and provide a foundation for future research to understand the rates and physiology of these denitrifying organisms.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Ben Wise.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
on10049 ( NOTIS )
1004963989 ( OCLC )
on1004963989
Classification:
LD1193.L547 2017m W57 ( lcc )

Downloads

This item has the following downloads:


Full Text
COMMUNITY COMPOSITION OF NITRITE REDUCTASE GENES IN AN ACID
MINE DRAINAGE ENVIRONMENT by
BEN WISE
B.A., University of Colorado, 2013.
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Environmental Science Program
2017


2017 BEN WISE
ALL RIGHTS RESERVED
11


This thesis for the Master of Science degree by
Ben R. Wise
has been approved for the Environmental Science Program by
Annika C. Mosier, Chair Christy Briles Timberley M. Roane
Date: July 29, 2017
m


Wise, Ben (M.S. Environmental Science)
Community Composition of Nitrite Reductase Genes in an Acid Mine Drainage Environment Thesis directed by Assistant Professor Annika C. Mosier
ABSTRACT
High elevation, mountainous regions have a high concentration of mining activities and resulting acid mine drainage (AMD) that is typically acidic mid often contains elevated concentrations of metals. The impacts of AMD on denitrifying microbial communities is not well understood, despite these organisms central role in the nitrogen cycle, contribution to greenhouse gas production, and potential to provide ecosystem services through the mitigation of nitrogen pollution. This study examined denitrifying microbes across four regions within the Iron Springs Mining District (13 sites over four time-points) located in Southwest Colorado at high elevation that receive AMD or naturally-occurring acid rock drainage (ARD). Denitrification functional gene sequences (nirS mid nirK coding for nitrite reductase) had a high number of observed OTUs (260 for nirS and 253 for nirK) and were observed at sites with pH as low as 3.2, dissolved oxygen as low as 1.0 mg/L, and metals >10 mg/L (including aluminum, iron, manganese, and zinc). A majority of the nirK and nirS OTUs (>60%) were present in only one sampling region. Approximately 8% of the nirK and nirS OTUs had a more cosmopolitan distribution with presence in three or more regions. Phylogenetic ally related OTUs were found across sites with very different chemistry. The total nirS community structure was correlated to iron, conductivity, sodium, and calcium, which may suggest that these factors play an important role in shaping the nirS community. Overall, these findings improve upon our understanding of the potential for denitrification within mi ecosystem impacted by AMD and provide a foundation for future research to understand the rates and physiology of these denitrifying organisms.
IV


The form and content of this abstract are approved. I recommend its publication.
Approved: Annika C. Mosier
v


ACKNOWLEDGEMENTS
I cannot fully express my appreciation for the guidance, patience, and fortitude that Dr. Annika C. Mosier has provided throughout the duration of this project. I am tremendously thankful for the opportunity she has given me to work alongside her. I would also like to thank my thesis committee members: Dr. Christy Briles for being both a friend and mentor and Dr. Timberley M. Roane for her unaltering enthusiasm and willingness to help.
I am thankful to Robert Edgar, Adrienne Narrowe, Bhargavi Ramanathan, and Sladjana Subotic for support on bioinformatics analysis. I would like to thank Joshua D. Sackett and Ashley Joslin for sample collection. I would like to thank the following institutes for their role in biological and chemical analyses: United States Environmental Protection Agency Region 8 Laboratory, and the Roy J. Carver Biotechnology Center at the University of Illinois.
As with all of my accomplishments, past and future, I must thank my family. I owe my patience, inquisitiveness, mid work ethic to my parents, John mid Carol Wise. I owe my confidence and passion to my sister, Madeline Wise. I love you guys.
vi


TABLE OF CONTENTS
CHAPTER
I. BACKGROUND.........................................................1
II. INTRODUCTION.......................................................5
III. MATERIALS AND METHODS..............................................7
Site Description and Sample Collection.............................7
Environmental Parameters...........................................7
DNA Extraction and Amplicon Sequencing.............................8
Sequence Analyses................................................ 11
Phylogenetic Analyses............................................ 11
Statistical Analyses............................................. 12
IV. RESULTS............................................................14
Environmental Parameters......................................... 14
Community Composition and Phylogeny of nirK Gene Sequences....... 19
Community Composition and Phylogeny of nirS Gene Sequences........24
Relationship Between Environmental Parameters mid Gene Sequences..29
Seasonal Variations in Gene Sequences.............................31
V. DISCUSSION.........................................................32
Conclusions.......................................................36
REFERENCES................................................................37
APPENDIX
SUPPLEMENTAL TABLES AND FIGURES.......................................42
vii


LIST OF TABLES
Table 2.1: List of PCR primers used for amplicon sequencing, including primer sequence and expected region and size of amplification...............................................9
Table 3.1: Chemistry data from surface water for sample sites across June mid September 2013 in the Iron Springs Mining District................................................... 17
Table 3.2: Chemistry data from surface water for sample sites across June and September 2014 in the Iron Springs Mining District................................................... 18
vm


LIST OF FIGURES
Figure 1. Sequential reductive pathway of denitrification showing the location of enzymes relative to the cytoplasmic membrane. Nar, nitrate reductase; Nir, nitrite reductase; Nor, nitric oxide reductase; Nos, nitrous oxide reductase (Wallenstein et al., 2006).........................2
Figure 2. Map of sampling regions and individual sites within the Iron Springs Mining District located in Southwest Colorado. Base image modified from Google Maps (https://www.google.com/maps/)............................................................. 14
Figure 3. Hierarchical clustering map of environmental chemistry data using analytes present in at least 50% of samples. Analyte values were normalized to a sum of 1. Scale bar indicates the proportion of the normalized sum. Chemistry data clusters according to sampling region, as indicated by colored labels................................................................ 16
Figure 4. Maximum likelihood tree for nirK gene sequences across all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et al., 2010) in Geneious v8.1.8, with Jukes-Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by green circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity..................................................................................20
Figure 5. nirK hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least four times the average relative abundance across all sites. Groups color coded black contained OTUs present across multiple regions. nirK_OTU_29 (present across all regions) indicated by arrow................................23
Figure 6. Maximum likelihood tree for nirS gene sequences across all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et al., 2010) in Geneious v8.1.8, with Jukes-Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by blue circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity.............................................................................25
Figure 7. nirS hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least ten times the average relative abundance across all sites. OTUs spread across all regions and OTU of interest nirS_OTU_2 are indicated by arrows ...........................................................................................28
Figure 8. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirK gene sequences. DM (dissolved metal) strontium significantly (Bonferroni corrected p-value <0.05) correlated to nirK distribution across the Iron Springs Mining District.....................30
IX


Figure 9. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirS gene sequences (Panel A). Conductivity, TRW (total recoverable metals from surface water) Iron, TRW Sodium, mid TRW Calcium significantly (Bonferroni corrected p-value <0.05) correlated to nirS distribution across the Iron Springs Mining District. Samples within each region are indicated by a colored circle. Panel B contains OTUs found within individual HCL groups as shown in Figure 7. Panel C contains all OTUs that were not found within individual HCL groups shown in Figure 7.................................................................................30
x


CHAPTER 1
BACKGROUND
Denitrifying microbes perform an important ecosystem service in the conversion of nitrate to nitrogen gas (Eq. 1) (Seitzenger et al., 2006). Denitriflers can help to limit eutrophication and play an integral in the nitrogen cycle. At the global scale, denitrification controls most of the fixed nitrogen in the worlds oceans, which in turn regulates primary production and dissolved CO2 in the oceans and atmosphere (Altabet et al., 2002). The process of denitrification also affects global climate through the production of nitrous oxide (N2O); an important greenhouse gas.
Eq. 1: NCV NOf NO N20 N2
Denitrification involves four enzymatically-catalyzed steps (Fig. 1): nitrate reduction, nitrite reduction, nitric oxide reduction, and nitrous oxide reduction (Philippot et al. 2002). These four steps are catalyzed by separated enzymes (Nar, Nir, Nor, and Nos). Although denitrification is classified as a type of anaerobic respiration (typically coupled to the oxidation of organic matter in the absence of oxygen), denitrifying bacteria have been shown to reduce nitrate to nitrogen gases under low oxygen conditions, typically less than ~0.2 mg O2 per liter (Seitzinger et al., 2006). Thus, denitrification occurs under conditions when O2 supply as a respiratory electron acceptor is limited.
1


Figure 1. Sequential reductive pathway of denitrification showing the location of enzymes relative to the cytoplasmic membrane. Nar, nitrate reductase; Nir, nitrite reductase; Nor, nitric oxide reductase; Nos, nitrous oxide reductase (Wallenstein et al., 2006).
The nitrite reduction step of denitrification is unique among other forms of nitrate metabolism (Shapleigh, 2006). Nitrite reductase is an especially important enzyme in the denitrification process because it catalyzes the first committed step to a gaseous product through the reduction of nitrite to nitric oxide (Zumft, 1997). This protein occurs in two separate and evolutionary unrelated forms: NirK (containing copper) and NirS (containing iron). Characterizing communities of denitrifying microorganisms is most often accomplished through analysis of the nirS and nirK denitrification functional genes as opposed to 16S rRNA genes because denitrifiers are found within a wide range of phylogenetically unrelated groups from over 50 genera (Zumft, 1997). Nitrite reductase genes have been studied in a plethora of environments, not including those that have been affected by acid mine drainage (AMD), despite this being a w id e sp r ea d p ol lutant.
AMD is a well-documented type of freshwater pollutant. The waters that drain from active and abandoned mines are typically acidic and often contain elevated concentrations of metals. The acidic and metal-rich fluids that characterize AMD are generated by the chemical weathering of rocks that contain metal sulfides, such as pyrite (FeS2), arsenopyrite (FeAsS), chalcopyrite
2


(CuFeS2), sphalerite (ZnS), and marcasite (FeS2) (Baker and Banfleld, 2003). The generation of AMD begins with the oxidation of ferrous iron by oxygen (Eq. 1). Ferric iron is then reduced by a sulfide such as pyrite (Eq. 2). The overall reaction results in dissolved ferrous iron, sulfate, and hydrogen ions (Eq. 3). The oxidation of sulfide mineral may initially be abiotic but the rate of the reaction is increased by the presence of prokaryotes through the regeneration of ferric iron via ferrous iron oxidation (Johnson and Hallberg, 2005) (Eq. 4).
Fe2+ + 3.502 + 14H+ = 14Fe3+ + 7H20 FeS2+ 14Fe3+ + 8H2O = 15Fe2+ + 2S042~ + 16H+ FeS2 + 3.502 + H20 = Fe2+ + 2S042~ + 2H+
(4) 4Fe2+ + 02 + 4H+ 4Fe3+ +2H2Q
Pyrite-rich earth is often mined for other metals such as gold (Au), silver (Ag), copper (Cu), zinc (Zn), and lead (Pb), which are released during the oxidation of metal sulfide (Baker and Banfleld, 2005). Iron, in either ferrous or ferric forms is the dominant metal present in AMD (Johnson and Hallberg, 2005). Metals within AMD systems have widespread, nonspecific biological toxicities and their effects on ecosystems are poorly understood. Metals and metalloids can be divided into two categories: essential metals required in some unicellular metabolisms (e.g., Co, Cu, Mn, and Zn) and toxic metals with no known essential functions (e.g. As, Cd, Hg, Pb, and U). The negative effects of these metals include cell membrane disruption, disabled DNA replication, stunted growth in plants, and neurological impairment, cancer, and organ failure in animals (Roane and Lantz, 2016). Acidity (pH <6) often increases the quantity of dissolved metals in solution, while adsorption and precipitation reactions increase with pH >7. Organic matter also influences solubility by binding to metals and thus reducing the solubility.
Mining within the study area analyzed here, the Iron Springs Mining District, began in the mid 1870s. Mainly silver ore was packed over the Ophir Pass to a smelter at Silverton.
3


Development of the town of Ophir, CO (in the Southwestern comer of the state, approximately 5.5 miles South of Telluride) proceeded at a steady pace until 1890 when the railroad reached the town and electricity was made widely available through the constmction of a local power plant. Residents quickly flocked to Ophir in search of the silver mid low-grade gold that was abundant in the veins straddling either side of the town mid Howard Fork River. Records of the amount of ore removed are very rare. The only formal report is for 1883, when 18 mines within the district produced 96,500 oz of silver and 760,000 lbs of lead (Luedke, 1996). By 1920, the heavy assault on the mines left them devoid of valuable ore except for that existing deep into the vein systems, yet extracting this ore was outside the expertise of the local miners. By 1947, most of the mines and mills had fallen into disrepair (Neubert et ah, 2002). Documentation does not exist for most of the mines and mills within the Iron Springs Mining District (Luedke, 1996).
This research was undertaken to assess the overall community stmcture of nitrite reductase genes in an AMD environment and to look for relationships between community stmcture and environmental variables. Both nirK mid nirS groups had high numbers of observed OTUs (253 for nirK mid 260 for nirS) that were phylogenetic ally diverse. The nirS community had a significant relationship to iron, sodium, calcium, and conductivity. Both nirK mid nirS communities showed potential adaptability to very acidic conditions. Future work should continue to investigate the relationship between nitrite reductase genes and environmental variables at smaller spatial scales. For instance, within the New Dominion region, how might the CSIB community differ from other sites? Future work should also assess rates of denitrification in this environment. Potential Nir adaptability to high iron concentrations should be addressed in an incubation study. Ultimately, this research has applicability to industry; denitrifiers that are adapted to harsh conditions may be utilized to treat specialized industrial waste.
4


CHAPTER 2
INTRODUCTION
Nitrogen is an essential element required by all life on Earth, mainly for the synthesis of amino acids and nucleotides. Nitrogen availability is controlled by the balance between the microbiologically-driven processes of nitrogen fixation (atmospheric N2 NH4+), nitrification (NH4+ NO3'), and anammox/denitriflcation that recycles fixed nitrogen back into atmospheric
N2 (Canfield et al., 2010). Denitrification involves four enzymatically-catalyzed steps: nitrate reduction by Nar proteins, nitrite reduction by Mr proteins, nitric oxide reduction by Nor proteins, and nitrous oxide reduction by Nos proteins (Philippot et al., 2002). The nitrite reduction step of denitrification catalyzes the first committed step to a gaseous product through the reduction of nitrite to nitric oxide (Zumft, 1997). The Nir protein occurs in two separate and evolutionarily unrelated forms: NirK (utilizing copper) and NirS (utilizing iron).
Due to their abundance and ubiquity across natural environments, denitriflers have proven to be one of the most successful physiological groups of microorganisms (Shapleigh, 2013). Despite this, little is known about denitrification in systems impacted by acid mine drainage (AMD), which is characterized by acidic and metal-rich fluids generated by the weathering of rocks that contain metal sulfides. Sediment denitrification in several AMD-impacted streams in Colorado (pH 2.6-6.0) was readily induced in the presence of nitrate (Baesman et al., 2006). However, it is unknown whether this is a general phenomenon in other AMD-impacted systems, whether the denitrifler community may be influenced by specific environmental conditions that characterize AMD systems, or if different denitrifying organisms (e.g., nirK- and rarS-type denitriflers) respond differently within this harsh environment.
5


While the impacts of AMD on denitrification are largely unknown, other research has evaluated how specific factors associated with AMD influence denitrifier communities. For example, acidic pH has repeatedly been shown to be a limiting factor in both diversity and rates of denitrification (Mendez-Garcia et al., 2015; Wallenstein et al., 2006; Simek et al., 2002; Wiljer and Delwiche, 1954). Nonetheless, denitrifier communities within environments that are historically acidic may be well-adapted to these conditions mid maintain activity at low pH (Parkin et al., 1985; Di Capua et al., 2017). Previous studies have also shown that heavy metals are an important factor in shaping denitrifier community composition in soils contaminated with heavy metals (Kandeler et al., 1996; Holtan-Hartwig et al., 2002; Cao et al., 2008). As the concentration of heavy metals increases, the diversity of denitrifiers mid rates of denitrification generally decrease (Sobolev mid Begonia, 2008; Liu et al., 2016).
In the present study, we used high-throughput sequencing to examine the diversity and changes in relative abundance of nitrite reductase genes in AMD-impacted sediments in the Colorado Rocky Mountains. AMD poses a significant environmental threat in Colorado due to the ubiquity of abandoned mines (23,000), many of which continuously emit AMD into freshwater systems (Colorado Geological Survey, 1998). The objectives of this study were to assess the overall community structure of nirK and nirS gene sequences mid to determine if community structure corresponded to environmental variables. Approximately 8% of the nirK and nirS OTUs had a cosmopolitan distribution across sites with wide ranging pH mid metal concentrations, possibly suggesting that these organisms are tolerant of variable conditions. Iron and conductivity appeared to play a role in shaping the overall nirS community composition. Ultimately, gaining a better understanding of how denitrifying microbes respond to adverse environmental conditions may improve our ability to maximize their conversion of nitrate to nitrogen gasmi important ecosystem service.
6


CHAPTER 3
MATERIALS AND METHODS Site Description and Sample Collection
The Iron Springs Mining District located in Ophir, Colorado consists of several abandoned mines. From 1877-1960, the Iron Springs Mining District was predominantly mined for metals such as silver, gold and lead, and to some extent for iron and tungsten (Nash, 2002). AMD from these mines continues to drain directly into the Howard Fork River. The mining district was divided into four regions for sampling based on the proximity of individual sample sites within each region mid their unique environmental conditions: Caribbeau, New Dominion, Iron Bog, and Howard Fork River.
Sampling, measurement of environmental parameters, DNA extraction, and amp lie on sequencing were performed as described previously (Ramanathan, 2016; Sackett, 2015). Composite sediment samples (approximately two inches deep) were collected from 13 sampling sites during June mid August 2013 and from 11 sampling sites during June and September 2014 at Iron Springs. Samples were stored on dry ice in the field until permanent storage at -20C in the laboratory freezer. For total recoverable metal analysis (TRW), 500 mL of surface water sample was collected, acidified to pH <2 with concentrated nitric acid mid stored at 4C. For dissolved metal analysis (DM), 500 mL of surface water sample was collected, filtered with a cellulose nitrite membrane filter (Thermo Scientific, Waltham, MA), acidified to pH <2 with concentrated nitric acid mid stored at 4C.
Environmental Parameters
Temperature, pH, conductivity, mid dissolved oxygen were measured at the sediment surface using a Thermo Scientific Orion 5-Star Multiparameter Meter Kit (Thermo Fisher
7


Scientific, Inc., Waltham, MA) and an In-Situ Multiparameter meter. Total recoverable and dissolved metal concentration analysis of the Iron Springs water samples were done at the EPA Region 8 lab (Golden, CO) using Inductively-coupled Plasma Mass Spectrometry (ICP-MS) following EPA method 200.8, and Inductively-coup led Plasma Optical Emission Spectrometry (ICP-OES) following EPA method 200.7. The analytes measured for the total recoverable metal concentrations (TRW) and dissolved metal concentrations (DM) included aluminum (Al), antimony (Sb), arsenic (As), barium (Ba), cadmium (Cd), calcium (Ca), copper (Cu), iron (Fe), lead (Pb), magnesium (Mg), manganese (Mn), nickel (Ni), selenium (Se), silver (Ag), sodium (Na), strontium (Sr), thallium (Tl), vanadium (V) mid zinc (Zn).
DNA Extraction and Amplicon Sequencing The MO-BIO PowerMax Soil DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA) was used to isolate total DNA from subsamples (10 grams) of mechanically homogenized saturated composite sediment from each sample site. DNA extracts were quantified using the Qubit dsDNA HS Assay Kit with the Qubit 2.0 Fluorometer (Life Technologies Corporation, Carlsbad, CA). High quality samples (=38) based on gel electrophoresis, DNA quantification, and PCR amplification were selected for further analyses (excluded samples were Carib01Junl4, FenND03Junl3, and all samples from sites Opp03, NDGP, and NDCS02). DNA extracts from each sample were sent to the University of Illinois Roy J. Carver Biotechnology Center, Urbana, Illinois for amplicon sequencing on the Illumina MiSeq sequencing platform. Library preparation was completed with the Fluidigm 48.48 Access Array IFC platform (Fluidigm Corporation, South San Francisco, CA) to amplify the nirK and nirS genes using the PCR primer sets nirK876/nirK1040 (Henry et al., 2004) mid nirSCd3aF/nirSR3cd (Kandeler et al., 2006; Throback et al., 2004), respectively.
8


Table 2.1: List of PCR primers used for amplicon sequencing, including primer sequence and expected region and size of amplification.
Gene Forward Primer Reverse Primer Fragment Size Reference
Denitrification nirS nirSCd3aF (5'-AACGYSAAGG ARACSGG) nirSR3cd (5'-GASTTCGG RTGSGTCT TSAYGAA) 425 bp (Kandeler et al., 2006; Throback, Enwall, Jarvis, & Hallin, 2004)
Denitrification nirK nirK876 (5'-ATY GGCGGVC AYGGCGA) nirK 1040 (5'-GCCTCGAT CAGRTTRT GGTT) 165 bp (Henry et al., 2004)
Samples were diluted to a final concentration of 2 ng/pL. A mastermix was prepared with Roche (Basel, Switzerland) High Fidelity Fast Start Kit and 2Ox Access Array loading reagent according to Fluidigm protocols. Into each well of a PCR plate, 1 pL of each sample was mixed with 1 pL of Fluidigm Illumina linkers and unique barcodes mix, 0.5 pL of 10X FastStart Reaction Buffer without MgCh, 0.9 pL of 25 mM MgCh, 0.25 pL of DMSO, 0.1 pL of 10 mM PCR grade nucleotide mix, 0.05 pL of 5 U/pL FastStart High Fidelity enzyme blend, 0.25 pL of 20X Access Array Loading Reagent, mid 0.95 pL of water. In a separate plate, 20X primer solutions were prepared by adding 2 pL of each forward mid reverse primer (synthesized by IDT Corp, Coralville, IA), 5 pL of 20X Access Array Loading Reagent, and 91 pL of water.
Once the sample mixture was complete, 4 pL was loaded into the sample inlets and 4 pL of the primer solution was loaded into the primer inlets of a primed Fluidigm 48.48 Access Array IFC. The IFC was then placed in a Fluidigm AX controller for microfluidic mixing of each primer and sample combination before being loaded into the Fluidigm Biomark HD PCR machine. Amplicons were generated using the following Access Array cycling program without imaging: 50C for 2 minutes, 70C for 20 minutes, 95C for 10 minutes, 10 cycles of (95C for 15 seconds, 60C for 30 seconds, and 72C for 1 minute), 2 cycles of (95C for 15 seconds, 80C for 30
9


seconds, 60C for 30 seconds, and 72C for 1 minute), 8 cycles of (95C for 15 seconds, 60C for 30 seconds, mid 72C for 1 minute), 2 cycles of (95C for 15 seconds, 80C for 30 seconds, 60C for 30 seconds, and 72C for 1 minute), 8 cycles of (95C for 15 seconds, 60C for 30 seconds, and 72C for 1 minute) and 5 cycles of (95C for 15 seconds, 80C for 30 seconds, 60C for 30 seconds, and 72C for 1 minute).
After amplification, 2 juL of Fluidigm Harvest Buffer was added to each sample inlet, and the IFC loaded onto the AX controller to harvest all PCR products from each sample (e.g., all primer amplifications pooled together for each sample). The PCR products were quantified using Qubit mid stored at -20C. The samples were run on a Fragment Analyzer (Advanced Analytics, Ames, IA) to confirm the expected sizes of amplicons. All of the 48 samples (containing all primer amplifications pooled together) were then pooled together in equal DNA concentrations into one tube. The pooled product was size selected on a 2% E-gel (Life Technologies, Waltham, MA), then recovered based on expected fragment size with a Qiagen (Hilden, Germany) gel extraction kit. Cleaned, size-selected products were run on mi Agilent Bioanalyzer to confirm the expected profile and determine the average product size.
The size-selected pool was qPCR quantitated mid loaded onto one MiSeq flowcell using a MiSeq 600-cycle sequencing kit, version 3 for 300 bp paired-end sequencing using a MiSeq FGx system in RUO mode. After sequencing, read data was translated into FASTQ files using the Illumina bcl2fastq 1.8.4 software with mi ASCII offset of 33. PhiX DNA reads (used as a spike-in control) were removed by alignment to the PhiX genome. The Roy J. Carver Biotechnology Center used in-house scripts for sorting the reads (with two mismatches allowed in the 5 primer sequences) mid demultiplexing (with one mismatch allowed in the index sequence attached in library prep).
10


Sequence Analyses
UP ARSE (Edgar, 2013) was used to analyze the amplicon sequence data. Primers from forward and reverse reads were sorted and demultiplexed. The paired-ends were joined and quality filtered at Phred quality score of 20. The last 20 bp were removed from both ends. Sequences with minimum merge length < 80 bp were discarded. Sequencing reads were clustered into operational taxonomic units (OTUs) at 97% nucleotide sequence identity. Representative sequences from each OTU were compared to the NCBI database using BLAST to ensure sequence specificity and only nirK mid nirS sequences were retained for further analyses.
Diversity analyses were conducted using QIIME (Quantitative Insights into Microbial Ecology) (Caporaso et al., 2010). Rarefaction was performed at multiple depths between one and the rarified depth (set to the median number of sequences for each gene). For nirK, samples containing fewer than 13,000 total sequences were excluded from analysis based on rarefaction curves (Figure SI). For nirS, samples containing fewer than 1,000 total sequences were excluded from analysis based on rarefaction curves (Figure SI). The number of observed OTUs mid Chaol richness estimates for each gene were determined using alpha diversity analyses in QIIME.
Phylogenetic Analyses
Representative nucleotide sequences of the observed OTUs were aligned in Geneious v8.1.8 (Kearse et al., 2012) using the FFT-NS-2 algorithm within MAFFT v7.017 (Katoh et al., 2002) mid manually checked mid trimmed. The alignment length for nirK was 123 bp and nirS was 417 bp. Maximum likelihood trees were constructed for representative sequences of observed OTUs using FastTree v2.1.5 package (Price et al., 2010) in Geneious v8.1.8, with Jukes-Cantor Correction and 1,000 resamples without branch length reoptimization. Trees were visualized using the Interactive Tree of Life (iTOL) (Letunic mid Bork, 2016). The normalized average relative
11


abundance was plotted for each OTU: OTU relative abundance averaged by region, then divided by the sum of the averages for all regions.
Statistical Analyses
Hierarchical Clustering (HCL) was performed on normalized physicochemical parameters (with each parameter summing to one in order to compare scale across variable units) and on the relative abundance values for each OTU within each sample site. The clustering method used a centered Pearson correlation distance matrix and average linkage clustering (using Multiexperiment Viewer, MeV 4.8; www.tm4.org/mev/) (Saeed et al., 2003). For clustering analysis only, 0.000001 was added to counts of zero to avoid software adjustments of zero values. HCL for chemistry only included parameters measured in at least 50% of the samples.
Correlations between community composition and environmental parameters were analyzed by canonical correspondence analysis (CCA) using the program Canoco, version 5 (Ter Braak, 1985). CCA was used to determine if the denitrifier community structure was more strongly correlated to specific environmental variables than expected by chance. Relative abundance of sequences for each OTU (defined at 97%) was used as the species input and environmental parameters were used as possible explanatory variables. Environmental parameters were included in the analysis if they were measured (above detection limits) in 15 or more samples. All dissolved metals (DM) were excluded from the analysis as these values were covarying with total recoverable metals (TRW). For nirK and nirS, environmental parameters from surface sediments included in the analyses were: pH, temperature (C), conductivity (pS/cm) mid dissolved oxygen (mg/L). For nirK, total recoverable metals (TRW in surface water; pg/L) included in the CCA analyses were Ca, Fe, Mg, Mn, Na, Sr, mid Zn. For nirS, total recoverable metals (TRW in surface water; pg/L) included in the CCA analyses were Al, Ca, Cu, Fe, Pb, Mg, Mn, Na, Sr, and Zn.
12


Significant environmental parameters (p-values < 0.05) after Bonferroni correction were selected via forward selection mid included in analysis.
Spearman correlations between environmental parameters and the relative abundance of individual OTUs were performed in QIIME (observationmetadatacorrelation.py) with Bonferroni corrected /7-values < 0.05. Environmental parameters included in the analysis were the same as those used for CCA, with the addition of dissolved metals (DM in surface water; pg/L): Ca, Mg, Mn, Na, Sr, and Zn for nirKm, Fe, Mg, Mn, Na, Sr, and Zn for nirS.
13


CHAPTER 4
RESULTS
Environmental Parameters
The sampling area was divided into four separate regions (Caribbean New Dominion, Iron Bog, and Howard Fork) based on chemistry and mining history (Figure 2). Sites within Caribbeau and New Dominion were actively mined for approximately 75 years and still emit visible AMD today. The CSEB site was an EPA-remediated site that was receiving AMD from a different source than the other New Dominion sites and had unique environmental characteristics. Sites within Iron Bog are receiving acid rock drainage, a chemical equivalent to AMD, generated through natural groundwater. The Howard Fork samples were taken directly from the Howard Fork river (upstream of the other AMD regions). This site is not located near a mine but likely receives intermittent AMD runoff from small mines further up the drainage basin.
ft Carib03 Carib02 iaribOl
Iron Springs Mining District, Colorado
Caribbeau
Region
Howard Fork River
NDMD02 FenlB02
FenNDOl
Figure 2. Map of sampling regions and individual sites within the Iron Springs Mining District located in Southwest Colorado. Base image modified from Google Maps (http s: //www .google.com/maps/).
14


Each region contains its own, unique physical mid chemical characteristics, which is likely a repercussion of the complex geology within the Iron Springs Mining District that varies among the four regions. Across all individual samples, pH ranged from 3.2-8.3, temperature ranged from 6.6C-22.4C, dissolved oxygen levels ranged from 1.0-10.0 mg/L, mid conductivity ranged from 296-1608 pS/cm. Sediment pH averaged 7.3 within Caribbeau, 5.5 within New Dominion, 4.8 within Iron Bog, and 7.0 for Howard Fork. The pH within New Dominion varied across sites and time points (pH 3.2-8.1). Temperature averaged 8.4 C at Caribbeau, 13.3 C at New Dominion, 13.7 C at Iron Bog, mid 9.9 C at Howard Fork. Dissolved oxygen was highest at Caribbeau (average of 8.4 mg/F) and lowest at Iron Bog (average of 5.12 mg/F). Conductivity was highest at New Dominion (average of 1238.4 pS/cm) mid lowest at Howard Fork (315.5 pS/cm). A comparison of the sum of dissolved metals to conductivity at each site resulted in a Pearson correlation coefficient of 0.83 and an R2 value of 0.68 (Figure S5). The most abundant metals in the surface water (total recoverable metals) were aluminum, iron, manganese, and zinc (>10 mg/F). Strontium, barium, copper, cadmium, lead, mid nickel were found in lower amounts. Other metals commonly found in AMD, such as arsenic, were below detectable limits at most sites and time points. Hierarchical clustering (HCF) of chemistry data revealed clusters according to sampling regions and sites within, indicating that each region, as a whole, has distinct chemical characteristics (Figure 3).
15


0.02
>0.03
Dissolved Oxygen pH
Conductivity TRW Calcium DM Calcium DM Strontium TRW Strontium TRW Sodium DM Sodium TRW Magnesium DM Magnesium TRW Lead TRW Zinc DM Zinc
TRW Manganese DM Manganese DM Iron TRW Iron DM Aluminim TRW Aluminum TRW Copper Temperature
Figure 3. Hierarchical clustering map of environmental chemistry data using analytes present in at least 50% of samples. Analyte values were normalized to a sum of 1. Scale bar indicates the proportion of the normalized sum. Chemistry data clusters according to sampling region, as indicated by colored labels.
16


Table 3.1: Chemistry data from surface water for sample sites across June mid
September 2013 in the Iron Springs Mining District
Sample Region Sample Name Sample Date pH Cond. (pS/cm) Temp. Cc) DO (mg/L)
Caribbean Mine CaribO 1 Junl 3 6.25.13 7.3 1095 7.9 8.8
Caribbean Mine Carib01Augl3 8.06.13 7.1 829 7.8 7.7
Caribbean Mine CaribO 2 Junl 3 6.25.13 7.2 1034 8.4 8
Caribbean Mine Carib02Augl3 8.06.13 7.8 830 8 8.5
Caribbean Mine CaribO 3 Junl 3 6.25.13 6.6 1068 13.1 7.9
Caribbean Mine Carib03Augl3 8.06.13 7 814 7.9 10
Iron Bog/Fen FenIB01Junl3 6.25.13 4.5 945 13.4 1.4
Iron Bog/Fen FenIB01Augl3 8.05.13 4.8 745 10.4 1.1
Iron Bog/Fen FenIB02Junl3 6.25.13 4.1 827 18.5 6.4
Iron Bog/Fen FenIB02Augl3 8.05.13 4.3 533 19.8 3.5
Iron Bog/Fen FenIB03Junl3 6.25.13 6.1 700 17.4 6.3
Iron Bog/Fen FenIB03Augl3 8.05.13 6.4 479 14.6 7.5
New Dominion Mine FenNDOl Junl3 6.25.13 7.3 1380 11 8.1
New Dominion Mine FenND01Augl3 8.05.13 6.2 1051 13.3 6
New Dominion Mine FenND03Augl3 8.05.13 6.5 1040 18.7 5.9
New Dominion Mine NDMD02Junl3 6.25.13 6.8 1424 9.5 8.1
New Dominion Mine NDMD02Augl3 8.06.13 7.4 1072 8.7 8.9
New Dominion Mine CSIBJunl3 6.25.13 3.2 1529 14.9 5.1
New Dominion Mine C SIB Aug 13 8.05.13 4.2 947 21.2 6.6
Howard Fork River HF04Junl3 6.25.13 5.7 335 8.9 7.8
Howard Fork River HF04Augl3 8.05.13 8.3 296 10.9 7.9
17


Table 3.2: Chemistry data from surface water for sample sites across June mid
September 2014 in the Iron Springs Mining District
Sample Region Sample Name Sample Date pH Cond. (pS/cm) Temp. Cc) DO (mg/L)
Caribbean Mine Carib01Septl4 9.30.2014 7.3 765 8 8.5
Caribbean Mine Carib02Junel4 6.24.2014 7.3 1230 7.1 8.5
Caribbean Mine Carib02Septl4 9.30.2014 7.8 763 7.9 8.7
Caribbean Mine Carib03Junel4 6.24.2014 7.2 828.3 9.3 8.0
Caribbean Mine Carib03Septl4 9.30.2014 8.0 562 8.4 8.5
Iron Bog/Fen FenIBOl Junel4 6.24.2014 3.5 1149 13.9 7.1
Iron Bog/Fen FenIB01Septl4 9.30.2014 4.3 616 13.7 6.9
Iron Bog/Fen FenIB02Junel4 6.24.2014 3.5 808.1 15.3 5.7
Iron Bog/Fen FenIB03Junel4 6.24.2014 5.5 591.8 9.9 8.1
Iron Bog/Fen FenIB03Septl4 9.30.2014 5.4 450 8.7 5.5
New Dominion Mine FenND01Junel4 6.24.2014 5.7 1608 13.9 9.2
New Dominion Mine FenND01Septl4 9.30.2014 7.3 1575 8.4 5.4
New Dominion Mine FenND03Junel4 6.24.2014 7.0 980 13.3 6.9
New Dominion Mine NDMD02Junel4 6.24.2014 6.5 1602 7.7 8.2
New Dominion Mine NDMD02Septl4 9.30.2014 8.1 1017 8.5 8.5
New Dominion Mine CSIBJunel4 6.24.2014 3.2 1242 9.9 7.0
New Dominion Mine CSIBSeptl4 9.30.2014 3.8 890 10.6 8.4
18


Community Composition and Phytogeny of nirK Gene Sequences
nirK gene sequences were recovered from 9 out of 11 different sample sites spanning all four regions (25 individual samples in total). Amplicon sequencing resulted in a total of 940,279 paired-end reads for nirK across all samples, with 13,291 to 90,423 nirK reads per sample. From all sites combined, we recovered a total of 253 unique nirK OTUs defined at a 97% identity cutoff. Between 7-51 nirK OTUs were observed in each individual sample (Figure S2). Samples from the Caribbeau region contained the highest number of OTUs (30-51 OTUs), whereas the lowest number of OTUs was observed in the Iron Bog samples (7-22 OTUs). Chaol richness estimates ranged from 13-74 OTUs for each sample (Figure S2). The difference between the number of observed OTUs mid the estimated total number of OTUs (Chao 1 estimate) ranged from 0-10 across all sites except sites Carib02Sepl4, Carib03Junl3, and Carib03Junl4, and Carib02Augl3 that had a difference of 10.5-28. This indicates that the majority of the nirK genes were sequenced in each sample (with the exception of the four Caribbeau sites).
Phylogenetic analysis of nirK OTUs revealed large evolutionary distance among sequences from sites in close proximity within the Iron Springs Mining District (Figure 4). Thirteen OTUs from the Caribbeau region were phylogenetically similar to each other and clustered according to sampling region. Otherwise, most OTUs from the same region were phylogenetically disparate. For instance, OTUs predominantly found at Howard Fork (blue) or Iron Bog (purple) were distributed across the tree without site- or region-specific clustering. Phylogenetically similar OTUs were found at sites with very different chemistry (e.g., nirK_OTU_78 and nirK_OTU_408 with 89%ID were found at Howard Fork and Iron Bog with drastically different iron concentrations of 381 and 5960 pg/L, respectively).
19


Figure 4. Maximum likelihood tree for nirK. gene sequences across all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et ah, 2010) in Geneious v8.1.8, with Jukes-Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by green circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity.
20


Iron Springs nirK sequences were related (76%-100% nucleotide identity) to nirK sequences from other environments, including agricultural soils, freshwater environments, mid soils impacted by heavy metals (based on BLAST searches to the NCBI database). nirK_OTU_255 was identical to an OTU found in agricultural plots in Japan (NCBI accession DQ783709). nirK_OTU_231 was identical to an OTU found within an oligotrophic, alpine lake microbial mat in the Central Pyrenees, Spain (NCBI accession KF816401). Thirteen OTUs found within the Caribbeau and New Dominion regions were closely related (>90% nucleotide identity) to silver-tolerant OTUs from topsoil (pH of 7.8) in Alunda, Uppsala, Sweden (Throback et al., 2007).
A majority of the nirK OTUs were found in one site or region: 101 OTUs (40% of all nirK OTUs) were found at a single site mid 153 OTUs (60%) were found in just one region. Among these, 62 OTUs (24%) were only found in the Caribbeau region, 68 OTUs (27%) were only found in the New Dominion region, 15 OTUs (6%) were only found in the Iron Bog region, mid 8 OTUs (3%) were only found in the Howard Fork region. Nearly 30% of the OTUs were shared between two regions: A smaller percentage of OTUs (11%; 29 OTUs) were found in three different regions. Caribbeau mid New Dominion regions shared the most OTUs (75), while Iron Bog mid Howard Fork shared the fewest (5). Only one OTU was found in all four regions: nirK_OTU_29, which had a nearest BLAST hit (88% identity) to a sequence found in agricultural soil.
Hierarchical clustering (HCL) analysis of the relative abundance of each nirK OTU across
samples revealed that the Howard Fork, Iron Bog, and Caribbeau Mine samples clustered
independently from each other (Figure 5). The New Dominion samples clustered closely with Iron
Bog and Caribbeau Mine samples. HCL clustering revealed 16 distinct groups of nirK OTUs that
had significant changes in the relative abundance between sites/regions, defined as OTUs with an
average relative abundance at least four times higher than the average relative abundance across
21


all sites. A majority of the groups were specific to one region: four Caribbean Mine groups comprised of 67 total OTUs; one Iron Bog group comprised of three OTUs; seven New Dominion groups comprised of 62 total OTUs; and one Howard Fork group comprised of 24 OTUs. Groups 14-16 represent OTUs that were found in two regions (Iron Bog and New Dominion or Caribbeau Mine and New Dominion). OTUs within each group displayed great sequence diversity with low percent identities (75% on average) and high phylogenetic distances.
22


n >J ^ to o C C c O) C Q.
= = 3330)
->->-><-> (O
I* r; T- CM o O o o o o
2 2 o Q Q Q
S £ S i i i
Caribbeau
New Dominion
Iron Bog Howard Fork
Figure 5. nirK hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least four times the average relative abundance across all sites. Groups color coded black contained OTUs present across multiple regions. nirK_OTU_29 (present across all regions) indicated by arrow.
23


Community Composition and Phytogeny of nirS Gene Sequences
nirS sequences were recovered from all four regions and all 11 sample sites, totaling 26 individual samples. Amplicon sequencing resulted in a total of 277,979 paired-end reads for nirS across all samples. The number of nirS reads per sample ranged from 1,033 to 59,838. From all sites combined, we recovered a total of 260 unique nirS OTUs defined at a 97% identity cut-off. Both the lowest mid highest number of OTUs was observed in the Caribbeau region (5-80 OTUs) and Chaol richness estimates ranged from 5-87 OTUs for each sample (Figure S2). The difference between the number of observed OTUs and the estimated total number of OTUs (Chaol estimate) ranged from 0-7, indicating that the majority of nirS genes were sequenced in each sample.
Similar to nirK, phylogenetic analysis of nirS sequences within the Iron Springs Mining District revealed large evolutionary distance among sequences from sites in close proximity (Figure 6). Sequences that were similar in their distribution across the study area were phylogenetically disparate. For example, Howard Fork, which contained the highest percentage of endemic species, showed little phylogenetic clustering. Phylogenetically similar OTUs were found at sites with very different chemistry (e.g., nirS_OTU_104 and nirS_OTU_373 with 92% similarity found at Howard Fork mid Caribbeau sites with very different iron concentrations; 381 and 4930 pg/L, respectively).
24


Figure 6. Maximum likelihood tree for nirS gene sequences across all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et at., 2010) in Geneious v8.1.8, with Jukes-Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by blue circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity.
25


Iron Springs nirS sequences were related (78%-100% identity) to nirS sequences from other environments, including agricultural soils, fen soils, and freshwater environments (based on BLAST searches to the NCBI database) nirS_OTU_95 was closely related (98% identity) to an OTU found within pH-neutral fen soils in Denmark (Palmer mid Horn, 2015). nirS_OTU_205 was closely related (90% identity) to an OTU found in mercury-contaminated soil (NCBI accession JF261040). The Howard Fork region contained 13 OTUs that were closely related (>85% identity) to OTUs found in freshwater environments (NCBI accession AM419565, JN179246, GU393102, KT444058, GU322137, DQ337856, GU393043, AM419582, HG800325, AB937597, EF615473, JF966853, AM419564).
A majority of the nirS OTUs were found at a single site (111 OTUs; 43% of all nirS OTUs) or in a single region (166 OTUs; 64%). Among those, 62 OTUs (24%) were only found in Caribbeau, 68 (26%) were only found in New Dominion, 24 (9%) were only found in Iron Bog, and 10 (4%) were only found in Howard Fork. Approximately one-third of the OTUs were shared among more than one region, with Caribbeau and New Dominion sharing the most OTUs (59) mid Howard Fork/Caribbeau mid Howard Fork/New Dominion sharing the fewest OTUs (13). Five OTUs were found across all four regions (nirS_OTU_4, nirS_OTU_5, nirS_OTU_58, nirS_OTU_30, nirS_OTU_45). These five OTUs were genetically dissimilar from each other (68.6% average identity), but were similar (85-97% identity) to database sequences from freshwater river sediment (nirS_OTU_4), agricultural soil (nirS_OTU_5 and nirS_OTU_45), glacier foreland soil (nirS_OTU_58), and the biofilm of a freshwater lake (nirS_OTU_30). nirS_OTU_2 was found at 3 regions and made up >10% of sequences at nine individual sites, including 94% at Carib01Augl3 and 75% at Carib01Junl3.
Hierarchical clustering (HCL) analysis of the relative abundance of each nirS OTU across
samples revealed that the most of the Caribbeau Mine samples clustered independently from each
26


other (Figure 7). Four of the New Dominion samples clustered independently from all other samples. Howard Fork samples clustered together, but also with an Iron Bog sample. HCL clustering revealed 10 distinct groups of nirS OTUs that had significant changes in the relative abundance between sites/regions (defined as OTUs with an average relative abundance at least 10 times higher than the average relative abundance across all sites). All of the groups were specific to one region: one Caribbeau group comprised of six OTUs; five New Dominion groups comprised of 74 total OTUs; two Iron Bog groups comprised of 104 total OTUs; two Howard Fork groups comprised of 22 total OTUs. OTUs within each group displayed great sequence diversity with low percent identities (70% on average) and high phylogenetic distances.
27


O C C 13)
< T -^ <
§ S S §
z z z z
CCCCQ-0)0-CQ.C
3333 Siiillf
Caribbeau New Dominion Iron Bog Howard Fork
Figure 7. nirS hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least ten times the average relative abundance across all sites. OTUs spread across all regions and OTU of interest nirS_OTU_2 are indicated by arrows.
28


Relationship Between Environmental Parameters and Gene Sequences
Sediment pH was not correlated to nirK or nirS richness, overall community structure, or relative abundance of individual OTUs. There was a slightly positive trend between the number of observed nirK OTUs mid pH, though the relationship was not significant (R2=0.26; Figure S3). There was no trend observed between the number of observed nirS OTUs mid sediment pH (R2=0.04; Figure S3). Spearman rank correlations did not show a significant relationship between pH and the relative abundance of individual nirK or nirS OTUs. Canonical correspondence analysis (CCA) did not reveal a significant relationship between pH and the overall nirK or nirS community structure.
Spearman correlations revealed that the relative abundance of nirK_OTU_14 was significantly negatively correlated to temperature (rho = -0.71) mid nirS_OTU_2 was significantly positively correlated to dissolved manganese (rho = 0.86) and total recoverable copper (rho=0.87).
CCA showed the nirK community was correlated to DM Strontium (Figure 8) mid showed that the nirS community was correlated to conductivity, total recoverable calcium, total recoverable iron, and total recoverable sodium (Figure 9). When the nirS relative abundance HCL groups (Figure 7) were highlighted on the CCA diagram, it suggested that the OTUs that formed HCL groups were driving much of the overall community structure (Figure 9). For instance, the Howard Fork HCL groups 9 mid 10 were distinctly clustered on the CCA plot, based in part on the low conductivity and iron values at those sites. The New Dominion HCL groups all had relatively high conductivity values (>980 pS/cm), but were separated by iron levels. When analyzing individual regions on their own, the New Dominion nirS OTUs were significantly correlated to total recoverable iron (Figure S4), further suggesting that these sites were driving the relationship with iron seen in the analysis of the overall community structure (Figure 9).
29


90-
%
5lvi Strontium ^ afcP
-1.0 0.6
O Caribbeau O Howard Fork ^ Iron Bog ^ New Dominion
Figure 8. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirK gene sequences. DM (dissolved metal) strontium significantly (Bonferroni corrected p-value <0.05) correlated to nirK distribution across the Iron Springs Mining District.
Figure 9. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirS gene sequences (Panel A). Conductivity, TRW (total recoverable metals from surface water) Iron, TRW Sodium, and TRW Calcium significantly (Bonferroni corrected p-value <0.05) correlated to nirS distribution across the Iron Springs Mining District. Samples within each region are indicated by a colored circle. Panel B contains OTUs found within individual HCL groups as shown in Figure 7. Panel C contains all OTUs that were not found within individual HCL groups shown in Figure 7.
30


Seasonal Variations in Gene Sequences
Within both sampling years, 46% of nirK OTUs mid 45% of nirS OTUs were present. Within multiple sampling months, 50% of nirK OTUs mid 49% of nirS OTUs were present. Pearson and Spearman correlations did not reveal a significant relationship between seasonality and the number of observed OTUs at each site for either nirK or nirS. Changes in the relative abundance of individual OTUs across sites does not appear to be linked to seasonality as relative abundance-based hierarchical clustering did not reveal nodes segregated by sampling month (e.g., Caribbeau samples taken in June, August, and September all clustered together for both nirK and nirS). CCA did not show a significant relationship between sampling month mid the relative abundance of gene sequences for either group.
31


CHAPTER 5
DISCUSSION
AMD is a widespread environmental hazard that can severely damage aquatic ecosystems. Little is known about how nirK and nirS gene sequences are distributed within an AMD environment. In this study, we determined the diversity, phylogeny, and distribution of nirK mid nirS gene sequences within four distinct regions in AMD-impacted sediments at the Iron Springs Mining District in Southwestern Colorado. Although gene sequences are not a proxy for function, analyses of the separate gene groups suggest that each have the potential for denitrification as a high number of OTUs (253 for nirK and 260 for nirS) were observed across the entire sampling area with varying relative abundance patterns. The presence of 117 nirK mid 116 nirS OTUs shared across two sampling years may suggest that many of the nir OTUs permanently reside in the sediments, and are not merely washed in from the surrounding watershed.
Hierarchical clustering revealed that the relative abundance patterns of nirK and nirS OTUs were often similar among sites within a sampling region but more dissimilar between regions (e.g., most Caribbeau samples clustered together on the nirK mid nirS HCL plots, separate from other regions). Previous studies also found similar patterns of site/region specificity for nirK mid nirS (Santoro et al., 2006; Smith and Ogrrnn, 2008; Mosier and Francis, 2010). One nirK OTU mid five nirS OTUs displayed a cosmopolitan distribution across all four sampling regions, which may represent either nirK/nirS-containing organisms tolerant of a range of pH and metal conditions or organisms that have been transported between sites by dispersion.
Previous work has shown pH to be an important factor in controlling denitrifying communities in the environment. Diversity and productivity generally decreases with acidic pH (Baesman et al., 2006; Wallenstein et al., 2006; Saleh-Lakha, 2009). Evidence has shown the
32


optimum pH range for complete reduction of nitrate to nitrogen gas is pH 6-8, below which the proportion of intermediate products (nitric and nitrous oxides) increases (Knowles, 1982; Nagele and Conrad, 1990). However, some studies have shown denitrification to occur and even thrive at an acidic pH. The optimum pH for denitrification rates in historically acidic soils was 3.90 (Parkin et al., 1985). Denitrification rates were >99% of the optimum at a pH of 5.3-5.75 within synthetic wastewater (Di Capua et al., 2017). These results indicate that denitrifler communities are able to metabolize in acidic conditions, perhaps especially if they are found within a historically acidic environment and have been able to adapt. In Iron Springs, pH ranged from 3.2-8.3 across all sample sites, but did not correlate to the number of observed nir OTUs, the relative abundance of any specific nir OTU, or the overall community structure or phylogeny of either gene group. Although there is a general increase in the number of observed nirK OTUs as pH increases, substantial variation exists around this trend. Because a decrease in pH will increase metal solubility, it is likely that changes in pH have indirect effects on the denitrifler community. These results may indicate that the community is well adapted to a low pH environment, but incubation studies are needed to determine whether or not denitrification rates at these sites is impacted by pH.
Denitrification enzymes are located on or near the outer cell surfaces, which makes them especially vulnerable to chemical disruption. Metals have been shown to affect denitrifler diversity and inhibit denitrification in a way that differentially affects the steps in the reduction of nitrate to nitrogen gas, with nitrite reductase being the most sensitive step within the denitrification pathway (Bollag and Barabsz, 1979; Sakadevan, et al., 1999; Sobolev and Begonia, 2008, Liu et al., 2016). However, other studies have shown diversity to increase with the addition of specific metals (Throback et al., 2007; Sandaa et al., 2001). In the present study, metal concentrations (e.g., Al, Cu, Fe, and Zn) in the Iron Springs region exceed the allowable concentrations as determined by
the Colorado Department of Public Health and Environment (CDPHE) Regulation 3;
33


concentrations that have been shown to harm terrestrial and aquatic life (Colorado Water Quality Control Commission, 2012). Nonetheless, nirK mid nirS OTUs were observed at sites with very high metal concentrations (>10 mg/L), which may indicate that these organisms are tolerant of these harsh conditions. Iron, in particular, appeared to play a role in shaping the nirS community structure (based on CCA results; Figure 9). Rates of denitrification in an AMD environment have been shown to substantially decrease with the addition of ferric or ferrous iron above background, likely due to chemical disruption of the cell wall (Baesman et al., 2007). Iron can also strongly complex with organic matter and reduce the bioavailability of organic carbon to denitrifiers (McKnight and Bencala, 1990). Iron concentrations differentially affected nirS communities within the New Dominion region, indicating potential adaptability to high iron concentrations in some OTUs (Figure S4).
The total nirS community composition was strongly correlated to conductivity or conductivity-related ions (calcium and sodium; based on CCA analysis; Figure 9). Conductivity may be an indication of the total dissolved metal concentration (Figure S5). Dissolved metals are those that are bioavailable and thus more likely to have toxic effects on microbes. The denitrifying microbes within Iron Springs may be resistant to dissolved metals as a result of long term genetic modification, spread of resistance genes, or the replacement of metal-sensitive strains by strains more tolerant of dissolved metals (Holtan-Hartwig et al., 2002). Interestingly, prior work with these same samples showed that organisms involved in nitrification may have also been influenced by conductivity (overall community structure and relative abundance of individual OTUs were correlated to conductivity) (Ramanathan, 2016). Overall, these findings may implicate conductivity, mid possibly dissolved metal concentrations, as a driving factor in controlling nitrogen cycling in these sediments.
34


Other organisms not studied here may also contribute to denitrification in the Iron Springs region. For instance, other nirK mid nirS gene sequences that were not detected with the conventional primer sets used here may be present (Wei et al., 2015). Additionally, fungal denitrifiers have the potential to convert nitrate to nitrogen gas within acidic environments. Fungal denitrification has occurred in groundwater with a pH of 3.67 (Jasrotia et al., 2014). Certain species have shown evidence of denitrification in an AMD environment (at pH 1) and express transcripts for aerobic respiration and denitrification in order to adapt to fluctuating conditions (Mosier et al., 2016). The possibility for other denitrifiers that may contribute to nitrogen cycling in the Iron Springs region warrants further inquiry.
Understanding how denitrification functional genes relate to the harsh conditions of an AMD environment shows how this widespread pollutant may broadly affect the nitrogen cycle. This understanding may also allow us to better utilize the ecosystem services provided by denitrifiers: the conversion of nitrate to nitrogen gas and improvement of water quality. Nitrogen pollution can cause a suite of problems ranging from eutrophication mid extensive kills of aquatic species to methemoglobinemia and birth defects in humans (Camargo and Alonso, 2006). Denitrifier services are utilized in wastewater treatment facilities around the globe that seek to reduce nitrogen pollution and some facilities have recently begun to isolate strains of denitrifiers within critically polluted environments to treat special industrial wastes (Kim et al., 2014). The denitrifiers within the Iron Springs Mining District, and perhaps other environments affected by AMD, have the potential to serve a similar purpose and treat wastewater that is acidic mid contains high concentrations of metals.
35


Conclusions
Denitrifying microbes provide an important ecosystem service in the conversion of nitrate to nitrogen gas. Gaining a better understanding of how these microbes respond to adverse environmental conditions, like those that exist in an AMD environment, may improve our understanding of their adaptability and our own ability to utilize this service. This study has demonstrated broad community composition patterns for two denitrification functional genes, nirK and nirS, in an environment impacted by AMD. Neither gene group was significantly affected by pH, possibly indicating that these organisms are well adapted to acidic conditions. Both gene groups had high numbers of observed OTUs across all sampling sites (253 for nirK and 260 for nirS) but were differentially affected by environmental conditions. nirK community composition was correlated to strontium concentrations. nirS community composition was correlated to conductivity, sodium, calcium, mid iron concentrations, which resulted in distinct groups of OTUs segregated by sampling region or individual samples. These findings improve upon our understanding of the potential for denitrification within an ecosystem impacted by AMD and provide a foundation for future research into the rates mid physiology of denitrifying organisms.
36


REFERENCES
Altabet, M.A., Higginson, M.J., & Murray, D.W. (2002). The effect of millennial-scale changes in Arabian Sea denitrification on atmospheric C02. Nature 415: 159-162.
Baeseman, J.L., Smith, R.L., & Silverstein, J. (2006). Denitrification Potential in Stream Sediments Impacted by Acid Mine Drainage: Effects of pH, Various Electron Donors, and Iron. Microbial Ecology 51: 232-241.
Baker, B.J., & Banfield, J.F. (2003). Microbial communities in acid mine drainage. FEMS microbiology ecology 44: 139-152.
Bollag, J.M. & Barabasz, W. (1979). Effect of heavy metals on the denitrification process in
soil. Journal of Environmental Quality 8: 196-201
Camargo, J. A., & Alonso, A. (2006). Ecological and toxicological effects of inorganic nitrogen pollution in aquatic ecosystems: a global assessment. Environment international 32: 831-849.
Canfield, D.E., Alexander, N.G., & Falkowski, P.G. (2010) The Evolution and Future of Earth's Nitrogen Cycle. Science 330: 192-6.
Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E. K., & Huttley, G. A. (2010). QIIME allows analysis of high-throughput community sequencing data. Nature methods 5: 335-336.
Cao, Y., Green, P.G., & Holden, P.A. (2008). Microbial community composition and denitrifying enzyme activities in salt marsh sediments. Applied and environmental microbiology 74: 7585-7595.
Colorado Geological Survey (1998) Abandoned Mine Lands. Web. http://coloradogeologicalsurvev.org/mineral-resources/abandoned-mine-lmids/. Accessed 2/2/2016.
Colorado Water Quality Control Commission. (2012). Regulation No. 31 The Basic Standards mid Methodologies for Surface Water.
Di Capua, F., Lakaniemi, A.M., Puhakka, J.A., Lens, P.N., & Esposito, G. (2017). High-rate thiosulfate-driven denitrification at pH lower than 5 in fluidize d-bed reactor. Chemical Engineering Journal 310: 282-291.
Edgar, R.C. (2013). UPARSE: highly accurate OTU sequences from microbial amplicon reads. Nature methods 10: 996-998.
Einsle, O., Messerschmidt, A., Stach, P., Bourenkov, G.P., Bartunik, H.D., Huber, R., & Kroneck, P.M. (1999). Structure of cytochrome c nitrite reductase. Nature. 400: 476-480.
37


Fierer, N. & Jackson, R.B. (2006). The diversity and biogeography of soil bacterial communities. Proceedings of the National Academy of Sciences of the United States of America 103: 626-631.
Giller, K.E., Witter, E., & Mcgrath, S.P. (1998). Toxicity of heavy metals to microorganisms and microbial processes in agricultural soils: a review. Soil Biology and Biochemistry 30: 1389-1414.
Henry, S., Baudoin, E., Lopez-Gutierrez, J. C., Mart in-Laurent, F., Brauman, A., & Philippot, L. (2004). Quantification of denitrifying bacteria in soils by nirK gene targeted real-time
PCR. Journal of Microbiological Methods 59: 327-335.
Holtan-Hartwig, L., Bechmann, M., Hoyas, T.R., Linjordet, R., & Bakken, L.R. (2002). Heavy metals tolerance of soil denitrifying communities: N20 dynamics. Soil Biology and Biochemistry 34: 1181-1190.
Jasrotia, P., Green, S.J., Canion, A., Overholt, W.A., Prakash, O., Wafula, D., Hubbard, D., Watson, D.B., Schadt, C.W., Brooks, S.C., & Kostka, J.E. (2014). Watershed-scale fungal community characterization along a pH gradient in a subsurface environment cocontaminated with uranium and nitrate. Applied Environmental Microbiology. 80: 1810-1820
Johnson, D.B., and Hallberg, K.B. (2005) Acid Mine Drainage Remediation Options: A Review.
Science of the Total Environment 338: 3-14.
Kandeler, F., Kampichler, C., & Horak, O. (1996). Influence of heavy metals on the functional diversity of soil microbial communities. Biology and Fertility of Soils 23: 299-306.
Kandeler, E., Deiglmayr, K., Tscherko, D., & Philippot, L. (2006). Quantification of functional genes narG, nirK and nosZ of denitrifying bacteria across a glacier foreland by real time
PCR. Applied Environmental Microbiology 72: 5957-5962.
Katoh, K., Misawa, K., Kuma, K.I., & Miyata, T. (2002). MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic acids research 30: 3059-3066.
Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., & Thierer, T. (2012). Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28: 1647-1649.
Kim, I.S., Ekpeghere, K.I., Ha, S.Y., Kim, B.S., Song, B., Kim, J.T., Kim, H.G., Koh, S.C. (2014). Full-scale biological treatment of tannery wastewater using the novel microbial consortium BM-
S-l. J. Environ. Sci. Health Part A Tox. Hazard. Subst. Environ. Eng. 49: 355-364.
Knowles, R. (1982). Denitrification. Microbiological reviews 46: 43.
Letunic, I. & Bork, P. (2016). Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic mid other trees. Nucleic acids research 44: 242-245.
38


Liu, Y., Liu, Y., Zhou, H., Li, L., Zheng, J., Zhang, X., & Pan, G. (2016). Abundance, composition and activity of denitrifier communities in metal polluted paddy soils. Scientific reports 6.
Luedke, R.G. (1996) Geologic Map of the Ophir Quadrangle, San Juan, San Miguel, and Dolores Counties, Colorado. U.S. Geological Survey Geologic Quadrangle Series GQ-1760.
Martens-Habbena W., Berube P.M., Urakawa, H., de la Torre, J.R., & Stahl D.A. (2009). Ammonia oxidation kinetics determine niche separation of nitrifying Archaea and Bacteria. Nature 461: 976-979.
Mendez-Garcia, C., Pelaez, A.I., Mesa, V., Sanchez, J., Golyshina, O.V., & Ferrer, M. (2015). Microbial diversity and metabolic networks in acid mine drainage habitats. Frontiers in microbiology 6: 475.
McKnight, D.M. & Bencala, K.E. (1990). The Chemistry of Iron, Aluminum, and Dissolved Organic Material in Three Acidic, Metal-Enriched, Mountain Streams, as Controlled by Watershed and in-Stream Processes. Water Resources Research. 26: 3087-3100.
Mosier, A.C. & Francis, C.A. (2010). Denitrifier abundance and activity across the San Francisco Bay estuary. Environmental Microbiology Reports 2: 667-676.
Mosier, A.C., Miller, C.S., Frischkom, K.R., Ohm, R.A., Li, Z., LaButti, K., Lapidus, A., Lipzen, A., Chen, C., Johnson, J., Lindquist, E.A., Pan, C., Hettich, R.L., Grigoriev, I.V., Singer, S.W., & Banfield, J.F. (2016). Fungi contribute critical but spatially varying roles in nitrogen and carbon cycling in acid mine drainage. Frontiers in microbiology 7.
Nagele, W., & Conrad, R. (1990). Influence of pH on the release of NO and N20 from fertilized and unfertilized soil. Biology and Fertility of Soils 10: 139-144.
Nash, J. T. (2002). Hydro geo chemical investigations of historic mining districts, Central Western Slope of Colorado, including influence on surface-water quality. US Department of the Interior, US Geological Survey (USGS).
Neubert, J.T. (2002) History, Geology, and Environmental Setting of Selected Mines Near Ophir, Uncompahgre National Forest, San Miguel County, Colorado. Colorado Geological Survey.
Parkin, T.B., Sexstone, A.J., & Tiedje, J.M. (1985). Adaptation of denitrifying populations to low soil pH. Applied and Environmental Microbiology 49: 1053-1056.
Palmer, K., & Horn, M.A. (2015). Denitrification activity of a remarkably diverse fen denitrifier community in Finnish Lapland is N-oxide limited. PloS one 10: 123.
Philippot, L. (2002). Denitrifying genes in bacterial and archaeal genomes. Biochimica et biophysic a acta (BBA)-Gene structure and expression 3: 355-376.
39


Price, M.N., Dehal, P.S., & Arkin, A.P. (2010). FastTree 2-approximately maximum-likelihood trees for large alignments. PloS one 5: 9490.
Ramanathan, B. (2016). Abundance and diversity of nitrifying microbes in sediments impacted by acid mine drainage. University of Colorado at Denver.
Roane, T., & Lantz, M. (2016). Microbial Uses in the Remediation of Metal-Imp acted Soils.
In Manual of Environmental Microbiology, Fourth Edition (pp. 5-2). American Society of Microbiology.
Sackett, J. D. (2015). Comparative microbial ecology of sediment-associated microbial communities from anthropogenically and endogenously metal impacted systems. University of Colorado at Denver.
Saeed, A., Bhagabati, N., Braisted, J., Stum, A., & Quackenbush, J. (2003). TIGR MeV Multiexperiment Viewer. The Institute for Genomic Research.
Sakadevan, K., Zheng, H., & Bavor, H.J. (1999). Impact of heavy metals on denitrification in surface wetland sediments receiving wastewater. Water Science and Technology 40: 349-355.
Saleh-Lakha, S., Shannon, K.E., Henderson, S.L., Goyer, C., Trevors, J.T., Zebarth, B.J., & Burton, D.L. (2009). Effect of pH mid temperature on denitrification gene expression and activity in Pseudomonas mandelii. Applied and environmental microbiology 75: 3903-3911.
Sandaa, R.A., Torsvik, V., & Enger, 0. (2001). Influence of long-term heavy-metal contamination on microbial communities in soil. Soil Biology and Biochemistry 33: 287-295.
Santoro, A.E., Boehm, A.B., & Francis, C. A. (2006). Denitrifier community composition along a nitrate mid salinity gradient in a coastal aquifer. Applied and Environmental Microbiology 72: 2102-2109.
Seitzinger, S., Harrison, J. A., Bohlke, J. K., Bouwman, A. F., Lowrance, R., Peterson, B., Tobias, C., & Van Drecht, G. (2006). Denitrification across landscapes mid waterscapes: a synthesis. Ecological Applications 6: 2064-2090.
Shapleigh, J.P. (2013). Denitrifying prokaryotes. In The prokaryotes (pp. 405-425). Springer Berlin Heidelberg.
Simek, M., Jisova, L., and Hopkins, D.W. (2002) What is the so-Called Optimum pH for Denitrification in Soil?. Soil Biology and Biochemistry 34:1227-34.
Smith, J.M., & Ogrrnn, A. (2008). Genetic and functional variation in denitrifier populations along a short-term restoration chronosequence. Applied and environmental microbiology 74: 5615-5620.
40


Sobolev, D. & Begonia, M. (2008). Effects of heavy metal contamination upon soil microbes: lead-induced changes in general and denitrifying microbial communities as evidenced by molecular markers. International journal of environmental research and public health 5: 450-456.
Ter Braak, C. (1985). Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179.
Thamdrup, B. (2012) New Pathways and Processes in the Global Nitrogen Cycle. Annual Review of Ecology, Evolution, and Systematics 43: 407-28.
Throback, I.N., Enwall, K., Jarvis, A., & Hallin, S. (2004). Reassessing PCR primers targeting nirS, nirK mid nosZ genes for community surveys of denitrifying bacteria with DGGE. FEMS microbiology ecology 49: 401-417.
Throback, I.N., Johansson, M., Rosenquist, M., Pell, M., Hansson, M., & Hallin, S. (2007). Silver (Ag+) reduces denitrification and induces enrichment of novel nirK genotypes in soil. FEMS Microbiology Letters 270: 189-194.
Wallenstein, M.D., Myrold, D.D., Firestone, M., & Voytek, M. (2006). Environmental Controls on Denitrifying Communities and Denitrification Rates: Insights from Molecular Methods. Ecological Applications 16: 2143-2152.
Wei, W., Isobe, K., Nishizawa, T., Zhu, L., Shiratori, Y., Ohte, N., Koba, K., Otsuka, S., & Senoo, K. (2015). Higher diversity and abundance of denitrifying microorganisms in environments than considered previously. The ISME journal 9: 1954.
Wijler, J. & Delwiche, C.C. (1954) Investigations on the denitrifying process in soil. Plant and SoilS\ 155-169.
Zumft, W.G. (1997) Cell biology and molecular basis of denitrification. Microbiology and Molecular Biology Review 61: 533-616.
41


APPENDIX
SUPPLEMENTAL TABLES AND FIGURES
Sequences
Carib02Augl3
Carib02Junl3
a Carib02Junl4
Carib02Sepl4
Carib03Augl3 Carib03Junl3 Carib03Junl4 Carib03Sepl4
3 FenlB01iunl4
FenlB01Sepl4
FenlB03Junl4
FenlB03Sepl4 FenND01Augl3
FenND01Junl3 FenND01Junl4 FenND01Sepl4
a Fenl\ID03Augl3 FenND03Junl4 HF04Junl3
NDMD02Augl3
NDMD02Junl3
NDMD02Sepl4
Sequences
CSIBSepl4
Carib01Augl3 Carib02Augl3
Carib02Junl3
Carib02Junl4 Carib02Sepl4 Carib03Augl3 Carib03Junl3
Carib03Junl4
Carib03Sepl4
FenlB01Junl4 FenlB01Sepl4 FenlB03Junl3 FenlB03Junl4 FenND01Augl3 FenND01Junl3 FenND01Junl4 FenND01Sepl4 FenND03Augl3
FenND03Junl4
HF04Augl3
HF04Junl3
NDMD02Augl3
NDMD02Junl3
NDMD02Sepl4
Figure SI: Rarefaction curves for nirK gene sequences across all samples (Panel A); samples containing fewer than 135000 reads were excluded from analysis. Rarefaction curves for nirS gene sequences across all samples (Panel B); samples containing fewer than 1,000 reads were excluded from analysis.
42


80
A
Site
Figure S2: Number of observed OTUs and Chaol richness estimates for nirK (Panel A) and nirS (Panel B) gene sequences across all sample sites.
43


60
A
o
§
r-
5
in
o
50
40
30
20
10
Caribbeau 1 New Dominion 1 Iron Bog Howard Fork
R2 = 0.2611

PH
B
pH
Figure S3: Comparison between the number of observed nirK OTUs (panel A) and the number of observed nirS OTUs (panel B) to the sediment pH at each site. Samples within the four regions are indicated by separate colors.
44


o
NDMD^
Jun13
d
00
o
enND03Aug13
CSIBSep14
O
^FenND03Jun14 FfenNDtXI Jun14
(^enND01Jljnj3 TRWIron
NDMD02Sep14
s
F;enND01 Aug13
JDMD02Aug13
O FenND01Sep14
Environmental Variables
TRW = total recoverable metals
Samples
o
-0.4
1.0
Figure S4: Canonical Correspondence Analysis (CCA) of relative gene abundance of nirS gene sequences for individual samples within the New Dominion region. TRW Iron significantly correlated (Bonferroni corrected p-value <0.05) to the distribution of relative gene abundance.
45


4*.
ON
Figure S5. Comparison of conductivity to the sum of all dissolved metals at each site with a Pearson correlation coefficient of 0.83 and an R2 value of 0.68.
Dissolved Metal Aggregate (pig/L)
CaribOl Aug 13 Carib01Junl3 CaribOl Sep 14 Carib02Augl3 Carib02Junl3 Carib02Junl4 Carib02Sepl4 Carib03Angl3 Carib03Junl3 Carib03Junl4 Carib03Sepl4 NDMD02Augl3 NDMD02Junl3 NDMD02Junl4 NDMD02Sepl4 FenNDOl Augl3 FenND01Junl3 FenND01Junl4 FenNDOl Sep 14 FenND03Augl3 FenND03Junl4 CSIBAugl3 CSIBJunl3 CSIBJunl4 CSIBSepl4 FenIBOl Aug 13 FenIB01Junl3 FenIB01Junl4 FenIBOl Sep 14 FenIB02Augl3 FenIB02Junl3 FenIB02Junl4 FenIB03Augl3 FenIB03Junl3 FenIB03Junl4 FenIB03Sepl4 HF04Augl3 HF04Junl3
OOOOOtOJ^ChCO
ooooooooo o o o o o
Conductivity (pS/cm)


Full Text

PAGE 1

COMMUNITY COMPOSITION OF NITRITE REDUCTASE GENES IN AN ACID MINE DRAINAGE ENVIRONMENT by BEN WISE B.A. University of Colorado, 2013. A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Environmental Science Program 2017

PAGE 2

ii 2017 BEN WISE ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Science degree by Ben R. Wise has been approved for the Environmental Science Program b y Annika C. Mosier, Chair Christy Briles Timberley M. Roane Date: July 29, 2017

PAGE 4

iv Wise, Ben (M.S. Environmental Science) Community Composition of Nitrite Reductase Genes in an Acid Mine Drainage Environment The sis directed by Assistant Professor Annika C. Mosier ABSTRACT High elevation, mountainous regions have a high concentration of mining activities and resulting acid mine drainage (AMD) that is typically acidic and often contains elevated concentrations of metals. The impacts of AMD on denitrifying microbial communities is not well understood, despite these organism s central ro le in the nitrogen cycle, contribution to greenhouse gas production, and potential to provide ecosystem services through the mitigat ion of nitrogen pollution. This study examined denitrifying microbes across four regions within the Iron Springs Mining District (13 sites over four time points) located in Southwest Colorado at high elevation that receive AMD or naturally occurring acid r ock drainage (ARD) Denitrification functional gene sequences ( nirS and nirK coding for nitrite reductase) had a high number of observed OTUs (260 for nirS and 253 for nirK ) and were observed at sites with pH as low as 3.2, dissolved oxygen as low as 1.0 m g/L, and metals >10 mg/L (including aluminum, iron, manganese, and zinc). A majority of the nirK and nirS OTUs (>60%) were present in only one sampling region. Approximately 8% of the nirK and nirS OTUs had a more cosmopolitan distribution with presence in three or more regions. Phylogenetically related OTUs were found a cross sites with very different chemistry The t otal nirS community structure was correlated to iron, conductivity, sodium, and calcium, which may suggest that these factors play an importan t role in shaping the nirS community. Overall, these findings improve upon our understanding of the potential for denitrification within an ecosystem impacted by AMD and provide a foundation for future research to understand the rates and physiology of the se denitrifying organisms.

PAGE 5

v The form and content of this abstract are approved. I recommend its publication. Approved: Annika C. Mosier

PAGE 6

vi ACKNOWLEDGEMENTS I cannot fully express my appreciation for the guidance, patience, and fortitude th at Dr. Annika C. Mosier has provided throughout the duration of this project. I am tremendously thankful for the opportunity she has given me to work alongside her. I would also like to thank my thesis committee members: Dr. Christy Briles for being both a friend and mentor and Dr. Timberley M. Roane for her unaltering enthusiasm and willingness to help. I am thankful to Robert Edgar, Adrienne Narrowe, Bhargavi Ramanathan and Sladjana Subotic for support on bioinformatics analysis. I would like to thank Jo shua D. Sacke t t and Ashley Joslin for sample collection. I would like to thank the following institutes for their role in biological and chemical analyses: United States Environmental Protection Agency Region 8 Laboratory, and the Roy J. Carver Biotechnolog y Center at the University of Illinois. As with all of my accomplishments, past and future, I must thank my family. I owe my patience, inquisitiveness, and work ethic to my parents, John and Carol Wise. I owe my confidence and passion to my sister, Madelin e Wise. I love you guys.

PAGE 7

vii TABLE OF CONTENTS CHAPTER I. BACKGROUND ................................ ................................ ................................ ............. 1 II. INTRODUCTION ................................ ................................ ................................ ........... 5 III. MATERIALS AND METHODS ................................ ................................ .................... 7 Site Description and Sample Collection ................................ ................................ ....... 7 Envi ronmental Parameters ................................ ................................ ............................ 7 DNA Extraction and Amplicon Sequencing ................................ ................................ 8 Sequence Analyses ................................ ................................ ................................ ...... 11 Phylogenetic Analyses ................................ ................................ ................................ 11 Statistical Analyses ................................ ................................ ................................ ..... 12 IV. RESULTS ................................ ................................ ................................ ..................... 14 Environmental Parameters ................................ ................................ .......................... 14 Community Composition and Phylogeny of nirK Gene Sequences ........................... 19 Community Composition and Phylogeny of nirS Gene Sequences ............................ 24 Relationship Between Environmental Parameters and Gene Sequences .................... 29 Seasonal Variations in Gene Sequences ................................ ................................ ..... 31 V. DISCUSSION ................................ ................................ ................................ ............... 32 Conclusions ................................ ................................ ................................ ................. 36 REFERENCES ................................ ................................ ................................ ............................. 37 APPENDIX SUPPLEMENTAL TABLES AND FIGURES ................................ ................................ .... 42

PAGE 8

viii LIST OF TA BLES Table 2.1: List of PCR primers used for amplicon sequencing, including primer sequence and expected region and size of amplification. ................................ ................................ ..................... 9 Table 3.1: Chemistry data from surface water for sample sites across June and September 2013 in the Iron Springs Mining District ................................ ................................ ............................... 17 Table 3.2: Chemistry data from surface wate r for sample sites across June and September 2014 in the Iron Springs Mining District ................................ ................................ ............................... 18

PAGE 9

ix LIST OF FIGURES Figure 1. Sequential reductive pathway of denitrification showing the location of enzymes relative to the cytoplasmic membrane. Nar, nitrate reductase; Nir, nitrite reductase; Nor, nitric oxide reductase; Nos, nitrous oxide reductase (Wallenstein et al., 2006). ................................ ............... 2 Figure 2. Map of sampling regions and individual sites within the Iron Springs Mining District located in Southwest Colorado. Base image modified from Google Maps (https://www.google.com/maps/). ................................ ................................ ................................ 14 Figure 3. Hierarchical clustering map of environmental chemistry data using analytes present in at least 50% of samples. Analyte values were normalized to a sum of 1. Scale bar indicates the proportion of the normali zed sum. Chemistry data clusters according to sampling region, as indicated by colored labels. ................................ ................................ ................................ ........... 16 Figure 4. Maximum likelihood tree for nirK gene sequences across all regions within the Iron S prings Mining District using FastTree v2.1.5 package (Price et al., 2010) in Geneious v8.1.8, with Jukes Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by green circle. The relative abunda nce of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity ................................ ................................ ................................ ................................ ....... 20 Figure 5. nirK hierarchical clustering map using Pearson c orrelation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group wa s at least four times the average relative abundance across all sites. Groups color coded black contained OTUs present across multiple regions. nirK_OTU_29 (present across all regions) indicated by arrow. ................................ ................... 23 Figure 6. Maximum likelihood tree for nirS gene sequences across all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et al. 2010) in Geneious v8.1.8, with Jukes Cantor Correction and 1,000 resamples wit hout branch length reoptimization. Bootstrap values above 75.0% indicated by blue circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity. ................................ ................................ ................................ ............................. 25 Figure 7. nirS hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least ten times the average relative abundance across all sites. OTUs spread across all regions and OTU of interest nirS_OTU_2 are indicated by arrows ................................ ................................ ................................ ................................ ....................... 28 Figure 8. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirK gene sequences. DM (dissolved metal) strontium significantly (Bonferroni corrected p value <0. 05) correlated to nirK distribution across the Iron Springs Mining District. ................................ ...... 30

PAGE 10

x Figure 9. Canonical Correspondence Analysis (CCA) of relative gene abundance of nirS gene sequences (Panel A). Co nductivity, TRW (total recoverable metals from surface water) Iron, TRW Sodium, and TRW Calcium significantly (Bonferroni corrected p value <0.05) correlated to nirS distribution across the Iron Springs Mining District. Samples within each region are indic ated by a colored circle. Panel B contains OTUs found within individual HCL groups as shown in Figure 7. Panel C contains all OTUs that were not found within individual HCL groups shown in Figure 7. ................................ ................................ ................................ ................................ .................... 30

PAGE 11

1 CHAPTER 1 BAC KGROUND Denitrifying microbes perform an important ecosystem service in the conversion of nitr ate to nitrogen gas (Eq. 1 ) (Seitzenger et al., 2006) Denitrifiers can help to limit eutrophication and play an integral in the nitrogen cycle. At the global sca le, denitrification controls most of the fixed oceans, which in turn regulates primary production and dissolved CO 2 in the oceans and atmosphere (Altabet et al., 2002). The process of denitrification also affects global climate thro ugh the production of nitrous oxide (N 2 O); an important greenhouse gas. Eq. 1 : NO 3 NO 2 NO N 2 O N 2 Denitrification involves four enzymatically catalyzed steps (Fig. 1 ): nitrate reduction, nitrite reduction, nitric oxide reduction, and nitrous oxide reduction ( Philippot et al. 2002). These four steps are catalyzed by separated enzymes ( Nar, Nir, Nor, and Nos ) Although denitrification is classified as a type of anaerobic respiration (typically coupled to the oxidation of organic matter in the absence of oxygen) denitrifying bacteria have been shown to reduce nitrate to nitrogen gases under low oxygen conditions, typically less than ~0.2 mg O 2 per liter ( Seitzinger et al. 2006). Thus, denitrification occurs under conditions when O 2 supply as a respiratory electron acceptor is limited.

PAGE 12

2 Figure 1 Sequential reductive pathway of denitrification showing the location of enzymes relative to the cytoplasmic membrane. Nar, nitrate reductase; Nir, nitrite reductase; Nor, nitric oxide reductase; Nos, nitrous oxide reductase (Wallenstein et al., 2006) The nitrite reduction st ep of denitrification is unique among other forms of nitrate metabolism (Shapleigh, 2006). Nitrite reductase is an especially important enzyme in the denitrification process because it catalyzes the first committed step to a gaseous product through the red uction of nitrite to nitric oxide (Zumft, 1997). This protein occurs in two separate and evolutionarily unrelated forms: NirK (containing copper) and NirS (containing iron). Characterizing communities of denitrifying microorganisms is most often accomplish ed through analysis of the nirS and nirK denitrification functional genes as opposed to 16S rRNA genes because denitrifiers are found within a wide range of phylogenetically unrelated groups from over 50 genera (Zumft, 1997). Nitrite reductase genes have b een studied in a plethora of environments, not including those that have been affected by acid mine drainage (AMD), despite this being a widespread pollutant. AMD is a well documented type of freshwater pollutant The waters that drain from active and a bandoned mines are typically acidic and often contain ele vated concentrations of metals. The acidic and metal rich fluids that characterize AMD are generated by the chemical weathering of rocks that contain metal sulfide s such as pyrite (FeS 2 ), arsenopyri te (FeAsS), chalcopyrite

PAGE 13

3 (CuFeS 2 ), sphalerite (ZnS), and marcasite (FeS 2 ) (Baker and Banfield, 2003). The generation of AMD begins with the oxidation of ferrous iron by oxygen (Eq. 1) F erric iron is then reduced by a sulfide such as pyrite (Eq. 2). The o verall reaction results in dissolved ferrous iron, sulfate, and hydrogen ions (Eq. 3). The oxidation of sulfide mineral may initially be abiotic but the rate of the reaction is increased by the presence of prokaryotes through the regeneration of ferric iro n via ferrous iron oxida tion (Johnson and Hallberg, 2005 ) (Eq. 4). (1) Fe 2+ + 3.5O 2 + 14H + = 14Fe 3+ + 7H 2 O (2) FeS 2 + 14Fe 3+ + 8H 2 O = 15Fe 2+ + 2SO 4 + 16H + (3) FeS 2 + 3.5O 2 + H 2 O = Fe 2+ + 2SO 4 + 2H + (4) 4Fe 2+ + O 2 + 4H + 3+ +2H 2 O Pyrite rich earth is often mined f or other metals such as gold ( Au ) silver ( Ag ) copper ( Cu ) zinc ( Zn ) and lead ( Pb ) which are released during the oxidation of metal sul fide (Baker and Banfield, 2005 ). Iron, in either ferrous or ferric forms is the dominant metal present in AMD (Johnso n and Hallberg, 2005 ). Metals within AMD systems have widespread, nonspecific biological toxicities and their effects on ecosystems are poorly understood. Metals and metalloids can be divided into two categories: essential metals required in some unicellul ar metabolisms (e g. Co, Cu, Mn, and Zn) and toxic metals with no known essential functions (e g. As, Cd, Hg, Pb, and U). The negative effects of these metals include cell membrane disruption, disabled DNA replication, stunted growth in plants, and neurol ogical impairment, cancer, and organ failure in animals (Roane and Lantz, 2016) Acidity (pH <6) often increases the quantity of dissolved metals in solution while adsorption and precipitation reactions increase with pH >7. Organic matter also influences solubility by binding to metals and thus reducing the solubility. Mining within the study area analyzed here, the Iron Springs Mining District, began in the

PAGE 14

4 Development of the town of Ophir CO ( in the Southwestern corner of the state, approximately 5.5 miles South of Tellurid e ) proceeded at a steady pace until 1890 when the railroad reached the town and electricity was made widely available through the construction of a local power plant. Residents quickly flocked to Ophir in search of the silver and low grade gold that was abundant in the veins straddling either side of the town and Howard Fork River. Records of the amount of ore removed are very rare. The only formal r eport is for 1883, when 18 mines within the district produced 96,500 oz of silver and 760,000 lbs of lead (Luedke, 1996). By 1920 the heavy assault on the mines left them devoid of valuable ore except for that existing deep into the vein systems yet extr acting this ore was outside the expertise of the local miners. By 1947, most of the mines and mills had fallen into disrepair (Neubert et al., 2002). Documentation does not exist for most of the mines and mills within the Ir on Springs Mining District (Lued ke, 1996). This res earch was undertaken to assess the overall community structure of nitrite reductase genes in an AMD environment and to look for relationships between community structure and environmental variables. Both nirK and nirS groups had high numbers of observed OT Us ( 253 for nirK and 260 for nirS ) that were phylogenetically diverse. The nirS community had a significant relationship to iron, sodium, calcium, and conductivity. Both nirK and nirS communities showed potential adaptability to very acidic conditions. Fut ure work should continue to investigate the relationship between nitrite reductase genes and environmental variables at smaller spatial scales. For instance, within the New Dominion region, how might the CSIB community differ from other sites? Future work should also assess rates of denitrification in this environment. P otential Nir adaptability to high iron concentrations should be addressed in an incubation study. Ultimately, this research has applicability to in dustry ; denitrifiers that are adapted to harsh conditions may be utilized to treat specialized industrial waste.

PAGE 15

5 CHAPTER 2 INTRODUCTION Nitrogen is an essential element required by all life on Earth, mainly for the synthesis of amino acids and nucleotide s. Nitrogen availability is controlled by the balance between the microbiologically driven processes of nitrogen fixation (atmospheric N 2 NH 4 + ), nitrification (NH 4 + NO 3 ), and anammox/denitrification that recycle s fixed nitrogen back into atmospheric N 2 (Canfield et al., 2010). Denitrification involves fou r enzymatically catalyzed steps: nitrate reduction by Nar proteins nitrite reduction by Nir proteins nitric oxide reduction by Nor proteins and nitrous oxide reduction by Nos proteins ( Philippot et al. 2002). The nitrite reduction step of denitrification catalyzes the first committed step to a gaseous product through the reduction of nitrite to nitric oxide (Zumft, 1997). The Nir protein occurs in two separate and evolutionarily unrelated forms: N ir K (utilizing copper) and N irS (utilizing iron). Due to their abundance and ubiquity across natural environments, denitrifiers have proven to be one of the most successful physiological groups of microorganisms (Shapleigh, 2013) Despite this little is k nown about denitrification in systems impac ted by acid mine drainage (AMD), which is characterized by acidic and metal rich fluids generated by the weathering of rocks that contain metal sulfides Sediment denitrification in several AMD impacted streams in Colorado (pH 2.6 6.0) was readily induced in the presence of nitrate ( Baesman et al. 2006). However, it is unknown whether this is a general phenomenon in other AMD impacted systems, whether the denitrifier community may be influenced by specific environ mental conditions that characterize AMD systems or if different denitrifying organisms (e.g., nirK and nirS type denitrifiers) respond differently within this harsh environment.

PAGE 16

6 While the impacts of AMD on denitrification are largely unknown other rese arch has evaluated how specific factors associated with AMD influence denitrifier communities. For example, acidic pH has repeatedly been shown to be a limiting factor in both diversity and rates of denitrification (Mendez Garcia et al., 2015; W allenstein et al., 2006 ; Simek et al., 2002; Wiljer and Delwiche, 1954) Nonetheless, denitrifier communities within environments that are historically acidic may be well adapted to these conditions and maintain activity at low pH (Parkin et al., 1985; Di Capua et al ., 2017). Previous studies have also shown that heavy metals are an important factor in shaping denitrifier community composition in soils contaminated with heavy metals (Kandeler et al., 1996; Holtan Hartwig et al., 2002; Cao et al., 2008). As the concent ration of heavy metals increases, the diversity of denitrifiers and rates of denitrification generally decrease (Sobolev and Begonia, 2008; Liu et al., 2016). In the present study, we used high throughput sequencing to examine the diversity and changes in relative abundance of nitrite reductase genes in AMD impacted sediments in the Colorado Rocky Mountains AMD poses a significant environmental threat i n Colorado due to the ubiquity of abandoned mines (~2 3,000 ) many of which continuously emit AMD into fr eshwater systems (Colorado Geological Survey, 1998 ). The objectives of this study were to assess the overall community structure of nirK and nirS gene sequences and to determine if community structure corresponded to environmental variables. Approximately 8% of the nirK and nirS OTUs had a cosmopolitan distribution across sites with wide ranging pH and metal concentrations, possibl y suggesting that these organisms are tolerant of variable conditions Iron and conductivity appeared to play a role in shaping the overall nirS community composition. Ultimately, gaining a better understanding of how denitrifying microbes respond to adver se environmental conditions may improve our ability to maximize their conversion of nitrate to nitrogen gas an important ecosystem service.

PAGE 17

7 CHAPTER 3 MATERIALS AND METHODS Site Description and Sample C ollection The Iron Springs Mining District located in O phir, Colorado consists of sever al abandoned mines. From 1877 1960, the Iron Springs Mining District was predominantly mined for metals such as silver, gold and lead, and to some extent for iron and tungsten (Nash, 2002) AMD from these mines continues to drain directly into the Howard Fork River. The mining district was divided into four regions for sampling based on the proximity of individual sam ple sites within each region and their unique environmental conditions: Caribbeau New Dominion, Iron Bog, and Howard Fork River Sampling, measurement of environmental parameters, DNA extraction, and amplicon sequencing were performed as describ ed previo usly (Ramanathan 2016 ; Sacke t t, 2015 ). Composite sediment samples (approximately two inches deep) were collected from 13 sampling sites during June and August 2013 and from 11 sampling sites during June and Septembe r 2014 at Iron Springs Samples were stored o n dry ice in the field until permanent storage at 20 C in the laboratory freezer. For total recoverable metal analysis (TRW), 500 mL of surface water sample was collected, acidified to pH <2 with concentrated nitric acid and stored at 4 C. For dissolved metal analysis (DM), 500 mL of surface water sample was collected, filtered with a cellulose nitrite membrane filter (Thermo Scientific, Waltham, MA), acidified to pH <2 with concentrated nitric acid and stored at 4 C. Environmental P arameters Temperature pH, conductivity, and dissolved oxygen were measured at the sediment surface using a Thermo Scientific Orion 5 Star Multiparameter Meter Kit (Thermo Fisher

PAGE 18

8 Scientific, Inc., Waltham, MA) and an In Situ Multiparameter meter. Total recoverable and dissolve d metal concentration analysis of the Iron Springs water samples were done at the EPA Region 8 lab (Golden, CO) using Inductively coupled Plasma Mass Spectrometry (ICP MS) following EPA method 200.8, and Inductively coupled Plasma Optical Emission Spectrom etry (ICP OES) following EPA method 200.7. The analytes measured for the total recoverable metal concentrations (TRW) and dissolved metal concentrations (DM) included aluminum (Al), antimony (Sb), arsenic (As), barium (Ba), cadmium (Cd), calcium (Ca), copp er (Cu), iron (Fe), lead (Pb), magnesium (Mg), manganese (Mn), nickel (Ni), selenium (Se), silver (Ag), sodium (Na), strontium (Sr), thallium (Tl), vanadium (V) and zinc (Zn). DNA Extraction and Amplicon Sequencing The MO BIO PowerMax Soil DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA) was used to isolate total DNA from subsamples (~10 grams) of mechanically homogenized saturated composite sediment from each sample site. DNA extracts were quantified using the Qubit dsDNA HS Assay Kit with the Qubit 2.0 Fluorometer (Life Technologies Corporation, Carlsbad, CA). High quality samples ( n= 38) based on gel electrophoresis, DNA quantification, and PCR amplification were selected for further analyses (excluded samples were Carib01Jun14, FenND03Jun13, and all samples from sites Opp03, NDGP, and NDCS02). DNA extracts from each sample were sent to the University of Illinois Roy J. Carver Biotechnology Center, Urbana, Illinois for amplicon sequencing on the Illumina MiSeq sequencing platform. Library prepa ration was completed with the Fluidigm 48.48 Access Array IFC platform (Fluidigm Corporation, South San Francisco CA) to amplify the nirK and nirS genes using the PCR primer sets nirK876/nirK1040 (Henry et al., 2004) and nirSCd3aF/nirSR3cd (Kandeler et al ., 2006; Throbck et al., 2004 ) respectively.

PAGE 19

9 Table 2.1: List of PCR primers used for amplicon sequencing, including primer sequence and expected region and size of amplification. Gene Forward Primer Reverse Primer Fragment Size Reference Denitrifica tion nirS nirSCd3aF (5' AACGYSAAGG ARACSGG) nirSR3cd (5' GASTTCGG RTGSGTCT TSAYGAA) 425 bp (Kandeler et al., 2006; Throbck, Enwall, Jarvis, & Hallin, 2004) Denitrifi cation nirK nirK876 (5' ATYGGCGGVC AYGGCGA) nirK1040 (5' GCCTCGAT CAGRTTRT GGTT) 165 bp (Henry et a l., 2004) Samples were Roche (Basel, Switzerland) High Fidelity Fast Start Kit and 20x Access Array loading reagent according to Fluidigm protocols. Into each well of a PCR pla Buffer without MgCl 2 2 of the primer solution was loaded into the primer inlets of a primed Fluidigm 48.48 Access Array IFC. The IFC wa s then placed in a Fluidigm AX controller for microfluidic mixing of each primer and sample combination before being loaded into the Fluidigm Biomark HD PCR machine. Amplicons were generated using the following Access Array cycling program without imaging: 50C for 2 minutes, 70C for 20 minutes, 95C for 10 minutes, 10 cycles of (95C for 15 seconds, 60C for 30 seconds, and 72C for 1 minute), 2 cycles of (95C for 15 seconds, 80C for 30

PAGE 20

10 seconds, 60C for 30 seconds, and 72C for 1 minute), 8 cycles of ( 95C for 15 seconds, 60C for 30 seconds, and 72C for 1 minute), 2 cycles of (95C for 15 seconds, 80C for 30 seconds, 60C for 30 seconds, and 72C for 1 minute), 8 cycles of (95C for 15 seconds, 60C for 30 seconds, and 72C for 1 minute) and 5 cycles of (95C for 15 seconds, 80C for 30 seconds, 60C for 30 seconds, and 72C for 1 minute). the IFC loaded onto the AX controller to harvest all PCR products from each sample (e.g., all primer amplifications pooled together for each sample). The PCR products were quantified using Qubit and stored at 20C. The samples were run on a Fragment Analyzer (Advanced Analytics, Ames, IA) to confirm the expected sizes of amplico ns. All of the 48 samples (containing all primer amplifications pooled together) were then pooled together in equal DNA concentrations into one tube. The pooled product was size selected on a 2% E gel (Life Technologies, Waltham, MA), then recovered based on expected fragment size with a Qiagen (Hilden, Germany) gel extraction kit. Cleaned, size selected products were run on an Agilent Bioanalyzer to confirm the expected profile and determine the average product size. The size selected pool was qPCR quanti tated and loaded onto one MiSeq flowcell using a MiSeq 600 cycle sequencing kit, version 3 for 300 bp paired end sequencing using a MiSeq FGx system in RUO mode. After sequencing, read data was translated into FASTQ files using the Illumina bcl2fastq 1.8. 4 software with an ASCII offset of 33. PhiX DNA reads (used as a spike in control) were removed by alignment to the PhiX genome. The Roy J. Carver Biotechnology Center used in mer sequences) and demultiplexing (with one mismatch allowed in the index sequence attached in library prep).

PAGE 21

11 Sequence Analyses UPARSE (Edgar, 2013) was used to analyze the amplicon sequence data. Primers from forward and reverse reads were sorted and dem ultiplexed The paired ends were joined and quality filtered at Phred quality score of 20. The last 20 bp were removed from both ends. Sequence s with minimum merge length < 80 bp were discarded Sequencing reads were clustered into operational taxonomic units (OTUs) at 97% nucleotide sequence identity. Representative sequences from each OTU were compared to the NCBI database using BLAST to ensure sequence s pecificity and only nirK and nirS sequences were retained for further analyses. Diversity analyses were conducted using QIIME (Quantitative Insigh ts into Microbial Ecology) (Capo raso et al., 2010). Rarefaction was performed at multiple depths between one and the rarified depth (set to the median number of sequences for each gene). For nirK, samples containing fewer than 13,000 total sequences were excluded from analysis based on rarefaction curves ( Figure S1 ) For nirS samples containing fewer than 1,000 total sequences were excluded from analysis based on rarefaction curves (Figure S1 ) The number of observed OTUs and Chao1 richness estimates for each gene w ere determined using alpha diversity analyses in QIIME. Phylogenetic Analyses Representative nucl eotide sequences of the observed OTUs were aligned in Geneious v8.1.8 (Kearse et al. 2012) using the FFT NS 2 algorithm within MAFFT v7.017 (Katoh et al., 2002) and manually checked and trimmed The alignment length for nirK was 123 bp and nirS was 417 bp. Maximum likelihood trees were constructed for representative sequences of obs erved OTUs using FastTree v2.1.5 package (Price et al. 2 010) in Geneious v8.1.8, with Jukes Cantor Correction and 1,000 resamples without branch length reoptimization. Trees were visualized using the Interactive Tree of Life (iTOL) (Letunic and Bork, 2016) The normalized average relative

PAGE 22

12 abundance was plotted for each OTU: OTU relative abundance averaged by region, then divided by the sum of the averages for all regions. Statistical Analyses Hierarchical Clustering (HCL) was performed on normalized physicochemical parameters (with each parameter summing to one in order to compare scale across v ariable units) and on the relative abundance values for each OTU within each sample site. The clustering method used a centered Pearson correlation distance matrix and average linkage clustering (using Multi experiment Viewer, MeV 4.8 ; www.tm4.org/mev/ ) (Saeed et al., 2003). For clustering analysis only, 0.000001 was added to counts of zero to avoid software adjustments of zero values. HCL for chemistry only included parameters measured in at least 50% of the samples Correlations between community composition and environmental parameters were analyzed by canonical correspondence analysis (CCA) using the program Canoco, version 5 (T er Braak 1985 ). CCA was used to determine if the denitrifier community structure was more strongly correlated to specific environmental variables than expected by chance. Relative abundance of sequences for each OTU (defined at 97% ) was used as the species input and e nvironmental parameters were used as possible explanatory variables. Envi ronmental parameters were included in the analysis if they were measured (above detection limits) in 15 or more samples. A ll dissolved metals ( DM ) were excluded from the analysis as these values were covarying with total recoverable metals (TRW) For nirK and nirS environmental parameters from surface sediments included in the analy ) and dissolved oxygen (mg/L). For nirK, total recoverable metals (TRW in surface water ; ) included in the CCA analyses were Ca, Fe, Mg, Mn, Na, Sr, and Zn. For nirS, total recoverable metals (TRW in s urface water ; ) included in the CCA analyses were Al, Ca, Cu, Fe, Pb, Mg, Mn, Na, Sr, and Zn.

PAGE 23

13 Significant environmental parameters ( p values < 0 .05) after Bonferroni correction were selected via forward selection and included in analysis. Spearman cor relations between environmental parameters and the relative abundance of individual OTUs were performed in QIIME (observation_metadata_correlation.py) with Bonferroni corrected p values < 0 .05. Environmental parameters included in the analysis were the sam e as those used for CCA, with the addition of dissolved metals (DM in surface water ; ): Ca, Mg, Mn, Na Sr and Zn for nirK ; Fe, Mg, Mn, Na Sr, and Zn for nirS

PAGE 24

14 CHAPTER 4 RESULTS Environmental P arameters The sampling area was divided into four separate regions ( Caribbeau New Dominion, Iron Bog, and Howard Fork) bas ed on chemistry and mining history (Figure 2 ) Sites within Caribbeau and New Dominion were actively mined for approximately 75 years and still emit visible AMD today. The CSIB site was an EPA remediated site that was receiving AMD from a different source than the other New Dominion sites and had unique environmental characteristics. Sites within Iron Bog are receiving acid rock drainage a chemical equivalent to AMD generated through natural groundwater. The Howard Fork samples were taken directly from th e Howard Fork river (upstream of the other AMD regions). This site is not located near a mine but likely receive s intermittent AMD runoff from small mines further up the drainage basin. Figure 2. Map of sampling regions and individual sites within the I ron Springs Mining District located in Southwest Colorado. Base image modified from Google Maps ( https://www.google.com/maps/ ).

PAGE 25

15 Each region contains its own, unique physical and chemical characteristics whi ch is likely a repercussion of the complex geology within the Iron Springs Mining District that varies among the four regions. Across all individual samples, pH ranged from 3.2 8.3, temperature ranged from 6.6 C 22.4 C, dissolved oxygen levels ra nged from 1.0 10 .0 mg/L, and c onductivity ranged from 296 1608 Sediment pH averaged 7.3 within Caribbeau 5.5 within New Dominion, 4.8 within Iron Bog, and 7.0 for Howard Fork. The pH within New Dominion varied across sites and time points (pH 3.2 8.1). Temperature averaged 8.4 C at Caribbeau 13.3 C at New Do minion, 13.7 C at Iron Bog, and 9.9 C at Howard Fork D issolved oxygen was highest at Caribbeau (average of 8.4 mg/L ) and lowest at Iron Bog ( average of 5.12 mg/L ) Conductivity was highest at New Dominion (average of 1238.4 ) and lowest at Howard For k (315.5 ). A comparison of the sum of dissolved metals to conductivity at each site resulted in a Pearson correlation coefficient of 0.83 and an R 2 value of 0.68 (Figure S5). The most abundant metals in the surface water (total recoverable metals) we re aluminum, iron, manganese, and zinc ( > 1 0 mg/L). Strontium barium, copper, cadmium, lead, and nickel were found in lower amounts. Other metals commonly found in AMD, such as arsenic, were below detectable limits at most sites and time points. Hierarchic al clustering (HCL) of chemistry data revealed clusters according to sampling regions and sites within indicatin g that each region, as a whole, has distinct chemical characteristics (Figure 3 ).

PAGE 26

16 Figure 3 Hierarchical clustering map of environmental che mistry data using analytes present in at least 50% of samples. Analyte values were normalized to a sum o f 1. Scale bar indicates the proportion of the normalized sum. Chemistry data clusters according to sampling region, as indicated by colored labels.

PAGE 27

17 Table 3.1 : Chemistry data from surface water for sample sites across June and September 2013 in the Iron Springs Mining District Sample Region Sample Name Sample Date pH Cond. Temp. (C) DO (mg/L) Caribbeau Mine Carib01Jun13 6.25.13 7.3 1095 7.9 8.8 Caribbeau Mine Carib01Aug13 8.06.13 7.1 829 7.8 7.7 Caribbeau Mine Carib02Jun13 6.25.13 7.2 1034 8.4 8 Caribbeau Mine Carib02Aug13 8.06.13 7.8 830 8 8.5 Caribbeau Mine Carib03Jun13 6.25.13 6.6 1068 13.1 7.9 Caribbeau Mine Carib03Aug13 8.06.13 7 814 7.9 10 Iron Bog/Fen FenIB01Jun13 6.25.13 4.5 945 13.4 1.4 Iron Bog/Fen FenIB01Aug13 8.05.13 4.8 745 10.4 1.1 Iron Bog/Fen FenIB02Jun13 6.25.13 4.1 827 18.5 6.4 I ron Bog/Fen FenIB02Aug13 8.05.13 4.3 533 19.8 3.5 Iron Bog/Fen FenIB03Jun13 6.25.13 6.1 700 17.4 6.3 Iron Bog/Fen FenIB03Aug13 8.05.13 6.4 479 14.6 7.5 New Dominion Mine FenND01Jun13 6.25.13 7.3 1380 11 8.1 New Dominion Mine FenND01Aug13 8.05.13 6.2 1051 13.3 6 New Dominion Mine FenND03Aug13 8.05.13 6.5 1040 18.7 5.9 New Dominion Mine NDMD02Jun13 6.25.13 6.8 1424 9.5 8.1 New Dominion Mine NDMD02Aug13 8.06.13 7.4 1072 8.7 8.9 New Dominion Mine CSIBJun13 6.25.13 3.2 1529 14.9 5.1 New Dominion Mine CSIBAug13 8.05.13 4.2 947 21.2 6.6 Howard Fork River HF04Jun13 6.25.13 5.7 335 8.9 7.8 Howard Fork River HF04Aug13 8.05.13 8.3 296 10.9 7.9

PAGE 28

18 Table 3.2 : Chemistry data from surface water for sample sites across June and September 2014 in the Iron Springs Mining District Sample Region Sample Name Sample Date pH Cond. Temp. (C) DO (mg/L) Caribbeau Mine Carib01Sept14 9.30.2014 7.3 765 8 8.5 Caribbeau Mine Carib02June14 6.24.2014 7.3 1230 7.1 8.5 Caribbeau Mine Carib02Sept14 9.30.2014 7.8 763 7.9 8.7 Caribbeau Mine Carib03June14 6.24.2014 7.2 828.3 9.3 8.0 Caribbeau Mine Carib03Sept14 9.30.2014 8.0 562 8.4 8.5 Iron Bog/Fen FenIB01June14 6.24.2014 3.5 1149 13.9 7.1 Iron Bog/Fen FenIB01Sept14 9.30.2014 4.3 616 13.7 6.9 Iron Bog/Fen FenIB02Ju ne14 6.24.2014 3.5 808.1 15.3 5.7 Iron Bog/Fen FenIB03June14 6.24.2014 5.5 591.8 9.9 8.1 Iron Bog/Fen FenIB03Sept14 9.30.2014 5.4 450 8.7 5.5 New Dominion Mine FenND01June14 6.24.2014 5.7 1608 13.9 9.2 New Dominion Mine FenND01Sept14 9.30.2014 7.3 15 75 8.4 5.4 New Dominion Mine FenND03June14 6.24.2014 7.0 980 13.3 6.9 New Dominion Mine NDMD02June14 6.24.2014 6.5 1602 7.7 8.2 New Dominion Mine NDMD02Sept14 9.30.2014 8.1 1017 8.5 8.5 New Dominion Mine CSIBJune14 6.24.2014 3.2 1242 9.9 7.0 New Domin ion Mine CSIBSept14 9.30.2014 3.8 890 10.6 8.4

PAGE 29

19 Community Composition and Phylogeny of nirK Gene Sequences nirK gene sequences were recovered from 9 out of 11 different sample sites spanning all four regions (25 individual samples in total) Amplicon s equencing resulted in a total of 940,279 paired end reads for nirK across all samples, with 13,291 to 90,423 nirK reads per sample. From all sites combi ned, we recovered a total of 253 unique nirK OTUs defined at a 97 % identity cut off. Between 7 51 nirK O TUs were observed in each individual sample (Figure S2 ) Samples from the Caribbeau region contained the highest number of OTUs (30 51 OTUs), whereas the lowest number of OTUs was observed in the Iron Bog samples (7 22 OTUs). Chao1 richness estimates range d from 13 74 OTUs for each sample (Figure S 2). The difference between the number of observed OTUs and the estimated total number of OTUs (Chao 1 estimate) ranged from 0 10 across all sites except sites Carib02Sep14, Carib03Jun13, and Carib03Jun14, and Carib 02Aug13 that had a difference of 10.5 28 This indicates that the majority of the nirK genes were sequenced in each sample (with the exception of the four Caribbeau sites ) Phylogenetic analysis of nirK OTUs revealed large evolutionary distance among seq uences from sites in close proximity within the Iron S prings Mining District (Figure 4 ). Thirteen OTUs from the Caribbeau region were phylogenetically similar to each other and clustered according to sampling region. Otherwise, most OTUs from the same regi on were phylogenetically disparate. For instance, OTUs predominantly found at Howard Fork (blue) or Iron Bog (purple) were distributed across the tree without site or region specific clustering. Phylogenetically similar OTUs were found at sites with very different chemistry (e.g., nirK_OTU_78 and nirK_OTU_408 with 89 %ID were found at Howard Fork and Iron Bog with drastically different iron concentrations of 381 and 5960 respectively).

PAGE 30

20 Figure 4. Maximum likelihood tree for nirK gene sequences acros s all regions within the Iron Springs Mining District using FastTree v2.1.5 package (Price et al., 2010) in Geneious v8.1.8, with Jukes Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by green circle. The relative abundance of OTUs across region s (as indicated by colored bars) shows large evolutionary distance among sequences within close proximity

PAGE 31

21 Iron Springs nirK sequences were related (76 % 100 % nucleotide identity) to nirK sequences from other environments, including agricultural soil s, freshwater environments, and soils impacted by heavy metals (based on BLAST searches to the NCBI database ). nirK OTU 255 was identical to an OTU found in agricultural plots in Japan (NCBI accession DQ783709) nirK_OTU_231 was identical to an OTU found w ithin an oligotrophic, alpine lake microbial mat in the Central Pyrenees, Spain (NCBI accession KF816401). Thirteen OTUs found within the Caribbeau and New Dominion regions were closely related (>90% nucleotide identity) to silver tolerant OTUs from topsoi l (pH of 7.8) in Alunda, Uppsala, Sweden ( Throbck et al., 2007). A majority of the nirK OTUs were found in one site or region: 101 OTUs (40% of all nirK OTUs) we re found at a single site and 153 OTUs (60%) we re found in just one region Among these, 62 O TUs (24%) were only found in the Caribbeau region 68 OTUs (27%) were only found in the New Dominion region, 15 OTUs (6%) were only found in the Iron Bog region, and 8 OTUs (3%) were only found in the Howard Fork region. Nearly 30% of the OTUs were shared between two regions: A smaller percentage of OTUs (11%; 29 OTUs) were found in three different regions. Caribbeau and New Dominion regions shared the most OTUs (75), while Iron Bog and Howard Fork shared the fewest (5). O nly one OTU wa s found in all four r egions : nirK_OTU_29 which had a nearest BLAST hit (88% identity) to a sequ ence found in agricultural soil. Hierarchical clustering (HCL) analysis of the relative abundance of each nirK OTU across samples revealed that the Howard Fork, Iron Bog, and Car ibbeau Mine samples clustered independently from each other (Figure 5 ) The New Dominion samples clustered closely with Iron Bog and Caribbeau Mine samples. HCL clustering revealed 16 distinct groups of nirK OTUs that had significant changes in the relati ve abundance between sites/regions, defined as OTUs with an average relative abundance at least four times higher than the average relative abund ance across

PAGE 32

22 all sites A majority of the groups were specific to one region: four Caribbeau Mine groups compris ed of 67 total OTUs; one Iron Bog group comprised of three OTUs; seven New Dominion groups comprised of 62 total OTUs; and one Howard Fork group comprised of 24 OTUs. Groups 14 16 represent OTUs that were found in two regions (Iron Bog and New Dominion or Caribbeau Mine and New Dominion). OTUs within each group displayed great sequence diversity with low percent identities (75% on average) and high phylogenetic distances

PAGE 33

23 Figure 5. nirK hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least fou r times the average relative abundance across all sites. Groups color coded black contained OTUs present across multiple regions. nirK_OTU_29 (present across all regions) indicated by arrow.

PAGE 34

24 Community Composition and Phylogeny of nirS Gene Sequences nir S sequences were recovered from all four regions and all 11 sample sites, totaling 26 individual samples Amplicon sequencing resulted in a total of 277,979 paired end reads for nirS across all samples. The number of nirS reads per sample ranged from 1,033 to 59,838. From all sites combi ned, we recovered a total of 260 unique nir S OTUs defined at a 97 % identity cut off. Both the lowest and highest number of OTUs was observed in the Caribbeau region (5 80 OTUs) and Chao1 richness estimates ranged from 5 87 O TUs for each sample (Figure S2 ). The difference between the number of observed OTUs and the estimated total number of OTUs (Chao1 estimate) ranged from 0 7, indicating that the majority of nirS genes were sequenced in each sample. Similar to nirK, phyloge netic analysis of nirS sequences within the Iron Springs Mining District revealed large evolutionary distance among sequences from sites in c lose proximity (Figure 6 ). Sequences that were similar in their distribution across the study area were phylogeneti cally disparate. For example, Howard Fork, which contained the highest percentage of endemic species, showed little phylogenetic clustering. Phylogenetically similar OTUs were found at sites with very different chemistry (e.g., nirS_OTU_104 and nirS_OTU_37 3 with 92% similarity found at Howard Fork and Caribbeau sites with very different iron concentrations ; 381 and 4930 respectively).

PAGE 35

25 Figure 6 Maximum likelihood tree for nirS gene sequences across all regions within the Iron Springs Mini ng District using FastTree v2.1.5 package (Price et al. 2010) in Geneious v8.1.8, with Jukes Cantor Correction and 1,000 resamples without branch length reoptimization. Bootstrap values above 75.0% indicated by blue circle. The relative abundance of OTUs across regions (as indicated by colored bars) shows larg e evolutionary distance among sequences within close proximity.

PAGE 36

26 Iron Springs nirS sequences were related (78 % 100 % identity) to nir S sequences from other environments, including agricultural soils, fen soils, and freshwater environments (based on BLAST searches to the NCBI database ) nir S_ OTU 95 was closely related (98% identity) to an OTU found within pH neutral fen soils in Denmark (Palmer and Horn, 2015). nirS OTU 205 was closely related (90% identity) to an OTU found in mercury contaminated soil (NCBI accession JF261040). The Howard Fork region contained 13 OTUs that were closely related (>85% identity) to OTUs found in freshwater environments (NCBI accession AM419565 JN179246 GU393102 KT444058 GU322137 DQ337856 GU393043 AM419582 HG800325 AB93 7597 EF615473 JF966853 AM419564 ). A majority of the nirS OTUs we re found at a single site ( 11 1 OTUs; 4 3 % of all nirS OTUs ) or in a single region ( 1 66 OTUs; 64 %) Among those, 62 OTUs (24%) were only found in Caribbeau 68 (26%) were only found in New Dominion, 24 (9%) were only found in Iron Bog, and 10 (4%) were only found in Howard Fork. Approximately one third of the OTUs were shared among more than one region, with Caribbeau and New Dominion sharing the most OTUs (59) and Howard Fork/ Caribbeau and Howard Fork/New Dominion sharing the fewest OTUs (13). Five OTUs were found across all four regions (nirS_OTU_4, nirS_OTU_5, nirS_OTU_58, nirS_OTU_30, nirS_OTU_45). These five OTUs were genetically dissimilar from each other (68.6% average identity), but w ere similar ( 85 97% identity) to database sequences from freshwater river sediment (nirS_OTU_4), agricultural soil (nirS_OTU_5 and nirS_OTU_45), glacier foreland soil (nirS_OTU_58), and the biofilm of a freshwater lake (nirS_OTU_30). nirS OTU 2 was found a t 3 regions and made up >10% of sequences at nine individual sites, including 94% at Carib01Aug13 and 75% at Carib01Jun13. Hierarchical clustering (HCL) analysis of the relative abundance of each nirS OTU across samples revealed that the most of the Caribb eau Mine samples clustered independently from each

PAGE 37

27 other (Figure 7) Four of the New Dominion samples clustered independently from all other samples. Howard Fork samples clustered together, but also with an Iron Bog sample. HCL clustering revealed 1 0 disti nct groups of nirS OTUs that had significant changes in the relative abundance between sites/regions ( defined as OTUs with an average relative abundance at least 10 times higher than the average relative abundance across all sites ). A ll of the groups were specific to one region: one Caribbeau group comprised of six OTUs; five New Dominion groups comprised of 74 total OTUs; two Iron Bog group s comprised of 104 total OTUs; two Howard Fork group s comprised of 2 2 total OTUs. OTUs within each group displayed gre at sequence diversity with low percent identities ( 70% on average) and high phylogenetic distances

PAGE 38

28 Figure 7. nirS hierarchical clustering map using Pearson correlation of relative abundance of individual OTUs. Scale bar represents relative abundance from 0.00 to greater than 0.25. OTU groups (in boxes color coded by region) were defined as those where the average relative abundance of OTUs within each group was at least ten times the average relative abundance across all sites. OTUs spread across all regions and OTU of interest nirS_OTU_2 are indicated by arrows

PAGE 39

29 Relationship Between Environmental Parameters and Gene Sequences Sediment pH was not correlated to nirK or nirS richness, overall community structure, or relative abundance of individual O TUs. There was a slightly positive trend between the number of observed nirK OTUs and pH, though the relationship was not significant (R 2 =0.26; F igure S3 ). There was no trend observed between the number of observed nirS OTUs and sediment pH (R 2 =0.04; Figu re S3 ). Spearman rank correlations did not show a significant relationship between pH and the relative abundance of individual nirK or nirS OTUs. Canonical correspondence analysis (CCA) did not reveal a significant relationship between pH and the overall n irK or nirS community structure. Spearman correlations revealed that the relative abundance of nirK OTU 14 was significantly negatively correlated to temperature (rho = 0.71) and nirS_OTU_2 was significantly positively correlated to dissolved manganese (rho = 0.86) and total recoverable copper (rho=0.87). CCA showed the nirK community was correlated to DM Strontium (Figure 8) and showed that the nirS community was correlated to conductivity, total recoverable calcium, total recoverable iron, and tot al recoverable sodium (Figure 9 ). When the nirS relative a bundance HCL groups (Figure 7 ) were highlighted on the CCA diagram, it suggested that the OTUs that formed HCL groups were driving much of the overal l community structure (Figure 9 ). For instance, the Howard Fork HCL groups 9 and 10 were distinctly clustered on the CCA plot, based in part on the low conductivity and iron values at those sites. The New Dominion HCL groups all had relatively high conductivity were separated by iron levels When analyzing individual regions on their own, the New Dominion nirS OT Us were significantly correlated to t otal recoverable iron (Figure S4 ) further suggesting that these sites were driving the relationship with iron seen in the analysis of the overal l community structure (Figure 9 )

PAGE 40

30 Figure 8 Canonical Correspondence An alysis (CCA) of relative gene abundance of nirK gene sequences DM (dissolved metal ) strontium significantly (Bonferroni co rrected p value <0.05) correlated to nirK distribution across t he Iron Springs Mining District. Figure 9 Canonical Correspondenc e Analysis (CCA) of relative gene abundance of nirS gene sequences (Panel A) Conductivity, TRW (total recoverable metals from surface water) Iron, TRW Sodium, and TRW Calcium significantly (Bonferroni corrected p value <0.05) correlated to nirS distributi on across t he Iron Springs Mining District. Samples within e ach region are indicated by a colored circle. Panel B contains OTUs found within individual HCL groups as shown in Figure 7 Panel C contains all OTUs that were not found within individual HCL gro ups shown in Figure 7

PAGE 41

31 Seasonal Variations in Gene Sequences Within both sampling years, 46% of nirK OTUs and 45% of nirS OTUs were present Within multiple sampling months, 50% of nirK OTUs and 49% of nirS OTUs were present. Pearson and Spearman correl ations did not reveal a significant relationship between seasonality and the number of observed OTUs at each site for either nirK or nirS Changes in the relative abundance of individual OTUs across sites does not appear to be linked to seasonality as rela tive abundance based hierarchical clustering did not reveal nodes segregated by sampling month (e.g., Caribbeau samples taken in June, August, and September all clustered together for both nirK and nirS ). CCA did not show a significant relationship between sampling month and the relative abundance of gene sequences for either group.

PAGE 42

32 CHAPTER 5 DISCUSSION AMD is a widespread environmental hazard that can severely damage aquatic ecosystems. Little is known about how nirK and nirS gene sequences are distributed within an AMD environment. In this study, we determined the diversity, phylogeny, and distribution of nirK and nirS gene sequences within four distinct regions in AMD impacted sediments at the Iron Springs Mining District in Southwestern C olorado. Although gene sequences are not a proxy for function, analyses of the separate gene groups suggest that each have the potential for denitrification as a high number of OTUs ( 253 for nirK and 260 for nirS ) were observed across the entire sampling a rea with varying relative abundance patterns The presence of 117 nirK and 116 nirS OTUs shared across two sampling years may suggest that many of the nir OTUs permanently reside in the sediments, and are not merely washed in from the surrounding watershed Hierarchical clustering revealed that the relative abundance patterns of nirK and nirS OTUs were often similar among sites within a sampling region but more dissimilar between regions (e.g., most Caribbeau samples clustered together on the nirK and nir S HCL plots separate from other regions ) Previous studies also found similar patterns of site/region specificity for nirK and nirS (Santoro et al., 2006; Smith and Ogram, 2008; Mosier and Francis, 2010 ). One nirK OTU and five nirS OTUs displayed a cosmop olitan distribution across all four sampling regions, which may represent either nirK/nirS containing organisms tolerant of a range of pH and metal conditions or organisms that have been transported between s ites by dispersion Previous work has shown pH to be an important factor in controlling denitrifying communities in the environment. Diversity and productivity generally decreases with acidic pH (Baesman et al., 2006; Wallenstein et al., 2006 ; Saleh Lakha, 2009). Evidence has shown the

PAGE 43

33 optimum pH rang e for complete reduction of nitrate to nitrogen gas is pH 6 8, below which the proportion of intermediate products (nitric and nitrous oxides) increases (Knowles, 1982; Ngele and Conrad, 1990). However, some studies have shown denitrification to occur and even thrive at an acidic pH. T he optimum pH for denitrification rates in historically acidic soils was 3.90 ( Parkin et al. 1985) Denitrificat ion rates were >99% of the optimum at a pH of 5.3 5.75 within synthetic wastewater ( Di Capua et al. 2017 ) These results indicate that denitrifier communities are able to metabolize in acidic conditions, perhaps especially if they are found within a historically acidic environment and have been able to adapt. In Iron Springs, pH ranged from 3.2 8.3 across all sample sites but did not correlate to the number of observed nir OTUs, the relative abundance of any specific nir OTU, or the overall community structure or phylogeny of either gene group. Although there is a general increase in the number of observed nirK OTUs as pH increases, substantial variation exists around this trend. Because a decrease in pH will increase metal solubility, it is likely that changes in pH have indirect effects on the denitrifier community. These results may indicate that the community is well adapted to a low pH environment, but incubation studies are needed to determine whether or not denitrification rates at thes e sites is impacted by pH. Denitrification enzymes are located on or near the outer cell surfaces, which makes them especially vulnerable to chemical disruption. M etals have been shown to affect denitrifier diversity and inhibit denitrification in a way t hat differentially affect s the steps in the reduction of nitrate to nitrogen gas, with nitrite reductase being the most sensitive step within the denitrification pathway (Bollag and Barabsz, 1979; Sakadevan, et al., 1999; Sobolev and Begonia, 2008, Liu et al., 2016) However, other studies have shown diversity to increase with the addition of specific metals ( Throbck et al., 2007; Sandaa et al., 2001). In the present study, m etal concentrations ( e.g., Al, Cu, Fe, and Zn) in the Iron Springs region exceed t he allowable concentrations as determined by the Colorado Department of Public Health and Envi ronment (CDPHE) Regulation 3;

PAGE 44

34 concentrations that have been shown to harm terrestrial and aquatic life (Colorado Water Quality Control Commission, 2012). Nonethel ess, nirK and nirS OTUs were observed at sites with very high metal concentrations (>10 mg/L), which may indicate that these organisms are tolerant of these harsh conditions. Iron, in particular, appeared to play a role in shaping the nirS community struct ure (based on CCA results ; Figure 9 ). R ates of denitrification in an AMD environment have been shown to substantially decrease with the addition of ferric or ferrous iron above background, likely due to chemical disruption of the cell wall (Baesman et al., 2007). Iron can also strongly complex with organic matter and reduce the bioavailability of organic carbon to denitrifiers (McKnight and Bencala, 1990). Iron concentrations differentially affected nirS communities within the New Dominion region, indicatin g potential adaptability to high iron concentrations in some OTUs (Figure S4 ). The t otal nirS community composition was strongly correlated to conductivity or conductivity related ions (calcium and sodium; based on CCA analysis; Figure 9 ). Conductivity may be an indication of the total dissolved metal concentration (Figure S5 ). Dissolved metals are those that are bioavailable and thus more likely to have toxic effects on micr obes The d enitrifying microbes within Iron Springs may be resistant to dissolve d metals as a result of long term genetic modification, spread of resistance genes, or the replacement of metal sensitive strains by strains more tolerant of dissolved metals (Holtan Hartwig et al., 2002). Interestingly, prior work with these same samples showed that organisms involved in nitrification may have also been influenced by conductivity (overall community structure and relative abundance of individual OTUs were correlated to conductivity) (Ramanathan, 2016 ). Overall, these findings may implicate conductivity, and possibly dissolved metal concentrations, as a driving factor in controlling nitrogen cycling in these sediments.

PAGE 45

35 Other organisms not studied here may also contribute to denitrificat ion in the Iron Springs region. For instance, other n irK and nirS gene sequences that were not detected with the conventional primer sets used here may be present (Wei et al., 2015 ). Additionally, fungal denitrifiers have the potential to convert nitrate to nitrogen gas within acidic environments. Fungal den itrification has occurred in groundwater with a pH of 3.67 (Jasrotia et al., 2014). Certain species have shown evidence of denitrification in an AMD environment (at pH 1) and express transcripts for aerobic respiration and denitrification in order to adapt to fluctuating conditions (Mosier et al., 2016). The possibility for other denitrifiers that may contribute to nitrogen cycling in the Iron Springs region warrants further inquiry. Understanding how denitrification functional genes relate to the harsh conditions of an AMD environment shows how this widespread pollutant may broadly affect the nitrogen cycle. This understanding may also allow us to better utilize the ecosystem service s provided by denitrifi ers: the conversion of nitrate to nitrogen gas a nd improvement of water quality Nitrogen pollution can cause a suite of problems ranging from eutrophication and extensive kills of aquatic species to methemoglobinemia and birth defects in humans (Camargo and Alonso, 2006). Denitrifier services are utili zed in wastewater treatment facilities around the globe that s eek to reduce nitrogen pollution and some f acilities have recently begun to isolate strains of denitrifiers within critically polluted environments to treat s pecial industrial wastes ( Kim et al. 2014). The denitrifiers within the Iron Springs Mining District, and perhaps other environments affected by AMD, have the potential to serve a similar purpose and treat wastewater that is acidic and contains high concentrations of metals

PAGE 46

36 Conclusions Denitrifying microbes provide an important ecosystem service in the conversion of nitrate to nitrogen gas. Gaining a better understanding of how these microbes respond to adverse environmental conditions, like those that exist in an AMD environment, may im prove our understanding of their adaptability and our own ability to utilize this service. This study has demonstrated broad community composition patterns for two denitrification functional genes, nirK and nirS in an environment impacted by AMD Neither gene group was significantly affected by pH, possibly indicating that these organisms are well adapted to acidic conditions. Both gene groups had high numbers of observed OTU s across all sampling sites (253 for nirK and 260 for nirS ) but were differentiall y affected by environmental conditions. nirK community composition was correlated to strontium concentrations. nirS community composition was correlated to conductivity, sodium, calcium, and iron concentrations which resulted in distinct groups of OTUs se gregated by sam pling region or individual samples. These findings improve upon our understanding of the potential for denitrification within an ecosystem impacted by AMD and provide a foundation for future research into the rates and physiology of denitrif ying organism s.

PAGE 47

37 REFERENCES Altabet, M.A., Higginson, M.J., & Murray, D.W. (2002). The effect of millennial scale changes in Arabian Sea denitrification on atmospheric CO2. Nature 415 : 159 162. Baeseman, J.L., Smith, R.L., & Silverstein, J. (2006). Denit rification Potential in Stream Sediments Impacted by Acid Mine Drainage: Effects of pH, Various Electron Donors, and Iron. Microbial Ecology 51 : 232 241. Baker, B.J., & Banfield, J.F. (2003). Microbial communities in acid mine drainage. FEMS microbiology ecology 44 : 139 152. Bollag, J.M. & Barabasz, W. (1979). Effect of heavy metals on the denitrification process in soil. Journal of Environmental Quality 8: 196 201 Camargo, J.A., & Alonso, (2006). Ecological and toxicological effects of inorganic nitr ogen pollution in aquatic ecosystems: a global assessment. Environment international 32 : 831 849. Canfield, D.E., Alexander, N.G., & Falkowski, P.G. (2010) The Evolution and Future of Earth's Nitrogen Cycle. Science 330 : 192 6. Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bush man, F. D., Costello, E. K., & Huttley, G. A. (2010). QIIME allows analysis of high thr oughput community sequencing data. Nature methods 5 : 335 336 Cao, Y., Green, P.G., & Hold en, P.A. (2008). Microbial community composition and denitrifying enzyme activities in salt marsh sediments. Applied and environmental microbiology 74 : 7585 7595. Colorado Geological Survey (1998) Abandoned Mine Lands. Web. http://coloradogeologicalsurvey.org/mineral resources/abandoned mine lands/ Accessed 2/2/2016. Colorado Water Quality Control Commission. (2012). Regulation No. 31 The Basic Standards and Meth odologies for Surface Water. Di Capua, F., Lakaniemi, A.M., Puhakka, J.A., Lens, P.N., & Esposito, G. (2017). High rate thiosulfate driven denitrification at pH lower than 5 in fluidized bed reactor. Chemical Engineering Journal 310 : 282 291. Edgar, R. C. (2013). UPARSE: highly accurate OTU sequences from microbial amplicon reads. Nature methods 10 : 996 998. Einsle, O., Messerschmidt, A., Stach, P., Bourenkov, G.P., Bartunik, H.D., Huber, R., & Kroneck, P.M. (1999). Structure of cytochrome c nitrite reductase. Nature 400 : 476 480.

PAGE 48

38 Fierer, N. & Jackson, R.B. (2006). The diversity and biogeography of soil bacterial commu nities. Proceedings of the National Academy of Sciences of the United States of America 103 : 626 631. Giller, K.E., Witter, E., & Mcgrath, S.P. (1998). Toxicity of heavy metals to microorganisms and microbial processes in agricultural soils: a review. Soi l Biology and Biochemistry 30 : 1389 1414. Henry, S., Baudoin, E., Lpez Gutirrez, J. C., Martin Laurent, F., Brauman, A., & Philippot, L. (2004). Quantification of denitrifying bacteria in soils by nirK gene targeted real time PCR. Journal of Microbiolog ical Methods 59 : 327 335. Holtan Hartwig, L., Bechmann, M., Hys, T.R., Linjordet, R., & Bakken, L.R. (2002). Heavy metals tolerance of soil denitrifying communities: N2O dynamics. Soil Biology and Biochemistry 34: 1181 1190. Jasrotia, P., Green, S.J., Canion, A., Overholt, W.A., Prakash, O., Wafula, D., Hubbard, D., Watson, D.B., Schadt, C.W., Brooks, S.C., & Kostka, J.E. (2014). Watershed scale fungal community characterization along a pH gradient in a subsurface environment cocontaminated with uranium and nitrate. Applied Environmental Microbiology 80 : 1810 1820 Johnson, D.B., and Hallberg, K.B. (2005) Acid Mine Drainage Remediation Options: A Review. Science of the Total Environment 338 : 3 14. Kandeler, F., Kampichler, C., & Horak, O. (1996). Infl uence of heavy metals on the functional diversity of soil microbial communities. Biology and Fertility of Soils 23 : 299 306. Kandeler, E., Deiglmayr, K., Tscherko, D., & Philippot, L. (2006). Quantification of functional genes narG, nirK and nosZ of denit rifying bacteria across a glacier foreland by real time PCR. Applied Environmental Microbiology 72 : 5957 5962. Katoh, K., Misawa, K., Kuma, K. I., & Miyata, T. (2002). MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic acids research 30 : 3059 3066. Kearse, M., Moir, R., Wilson, A., Stones Havas, S., Cheung, M., Sturrock, S., & Thierer, T. (2012). Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28 : 1647 1649. Kim, I.S., Ekpeghere, K.I., Ha, S.Y., Kim, B.S., Song, B., Kim, J.T., Kim, H.G., Koh, S.C. (2014). Full scale biological treatment of tannery wast ewater using the novel microbial consortium BM S 1. J. Environ. Sci. Health Part A Tox. Hazard. Subst. Environ. Eng. 49 : 355 364. Knowles, R. (1982). Denitrification. Microbiological reviews 46 : 43. Letunic, I. & Bork, P. (2016). Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic acids research 44 : 242 245.

PAGE 49

39 Liu, Y., Liu, Y., Zhou, H., Li, L., Zheng, J., Zhang, X., & Pan, G. (2016). Abundance, composition and activity of denitrifier c ommunities in metal polluted paddy soils. Scientific reports 6 Luedke, R.G. (1996) Geologic Map of the Ophir Quadrangle, San Juan, San Miguel, and Dolores Counties, Colorado. U.S. Geological Survey Geologic Quadrangle Series GQ 1760. Martens Habbena W ., Berube P.M., Urakawa, H., de la Torre, J.R., & Stahl D.A. (2009). Ammonia oxidation kinetics determine niche separation of nitrifying Archaea and Bacteria. Nature 461 : 976 979. Mndez Garca, C., Pelez, A.I., Mesa, V., Snchez, J., Golyshina, O.V., & F errer, M. (2015). Microbial diversity and metabolic networks in acid mine drainage habitats. Frontiers in microbiology 6 : 475. McKnight, D.M. & Bencala, K.E. (1990). The Chemistry of Iron, Aluminum, and Dissolved Organic Material in Three Acidic, Metal Enriched, Mountain Streams, as Controlled by Watershed and in Stream Processes. Water Resources Research 26 : 3087 3100. Mosier, A.C. & Francis, C.A. (2010). Denitrifier abundance and activity across the San Francisco Bay estuary. Environmental Microbiol ogy Reports 2 : 667 676. Mosier, A.C., Miller, C.S., Frischkorn, K.R., Ohm, R.A., Li, Z., LaButti, K., Lapidus, A., Lipzen, A., Chen, C., Johnson, J., Lindquist, E.A., Pan, C., Hettich, R.L., Grigoriev, I.V., Singer, S.W., & Banfield, J.F. (2016). Fungi co ntribute critical but spatially varying roles in nitrogen and carbon cycling in acid mine drainage. Frontiers in microbiology 7 Ngele, W., & Conrad, R. (1990). Influence of pH on the release of NO and N2O from fertilized and unfertilized soil. Biology a nd Fertility of Soils 10 : 139 144. Nash, J. T. (2002). Hydrogeochemical investigations of historic mining districts, Central Western Slope of Colorado, including influence on surface water quality US Department of the Interio r, US Geological Survey (USGS ). Neubert, J.T. (2002) History, Geology, and Environmental Setting of Selected Mines Near Ophir, Uncompahgre National Forest, San Miguel County, Colorado. Colorado Geological Survey Parkin, T.B., Sexstone, A.J., & Tiedje, J.M. (1985). Adaptation of denitrifying populations to low soil pH. Applied and Environmental Microbiology 49 : 1053 1056. Palmer, K., & Horn, M.A. (2015). Denitrification activity of a remarkably diverse fen denitri fier community in Finnish Lapland is N oxide limited. PloS one 10 : 123. Philippot, L. (2002). Denitrifying genes in bacterial and archaeal genomes. Biochimica et biophysica acta (BBA) Gene structure and expression 3 : 355 376.

PAGE 50

40 Price, M.N., Dehal, P.S., & Arkin, A.P. (2010). FastTree 2 approximately maximum likelihood trees for large alignments. PloS one 5 : 9490. Ramanathan, B. (2016). Abundance and diversity of nitrifying microbes in sediments impacted by acid mine drainage University of Colorado at Den ver. Roane, T., & Lantz, M. (2016). Microbial Uses in the Remediation of Metal Impacted Soils. In Manual of Environmental Microbiology, Fourth Edition (pp. 5 2). American Society of Microbiology. Sackett, J. D. (2015). Comparative microbial ecology of se diment associated microbial communities from anthropogenically and endogenously metal impacted systems University of Colorado at Denver Saeed, A., Bhagabati, N., Braisted, J., Sturn, A., & Quackenbush, J. (2003). TIGR MeV Multiexperiment Viewer. The Ins titute for Genomic Research Sakadevan, K., Zheng, H., & Bavor, H.J. (1999). Impact of heavy metals on denitrification in surface wetland sediments receiving wastewater. Water Science and Technology 40 : 349 355. Saleh Lakha, S., Shannon, K.E., Henderson, S.L., Goyer, C., Trevors, J.T., Zebarth, B.J., & Burton, D.L. (2009). Effect of pH and temperature on denitrification gene expression and activity in Pseudomonas mandelii. Applied and environmental microbiology 75 : 3903 3911. Sandaa, R.A., Torsvik, V., & Enger, (2001). Influence of long term heavy metal contamination on microbial communities in soil. Soil Biology and Biochemistry 33 : 287 295. Santoro, A.E., Boehm, A.B., & Francis, C.A. (2006). Denitrifier community composition along a nitrate and sali nity gradient in a coastal aquifer. Applied and Environmental Microbiology 72 : 2102 2109. Seitzinger, S., Harrison, J. A., Bhlke, J. K., Bouwman, A. F., Lowrance, R., Peterson, B., Tobias, C., & Van Drecht, G. (2006). Denitrification across landscapes and waterscapes: a synthesis. Ecological Applications 6: 2064 2090. Shapleigh, J.P. (2013). Denitrifying prokaryotes. In The prokaryotes (pp. 405 425). Springer Berlin Heidelberg. Called Optimum pH for Denitrification in Soil?. Soil Biology and Biochemistry 34 :1227 34. Smith, J.M., & Ogram, A. (2008). Genetic and functional variation in denitrifier populations along a short term restoration chronosequence. Applied and environmental microbiology 74 : 5615 5620.

PAGE 51

41 Sobolev, D. & Begonia, M. (2008). Effects of heavy metal contamination upon soil microbes: lead induced changes in general and denitrifying microbial communities as evidenced by molecular markers. International journal of environmental research and public health 5 : 450 456. Ter Braak, C. (1985). Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67 : 116 7 1179. Thamdrup, B. (2012) New Pathways and Processes in the Global Nitrogen Cycle. Annual Review of Ecology, Evolution, and Systematics 43 : 407 28. Throbck, I.N., Enwall, K., Jarvis, ., & Hallin, S. (2004). Reassessing PCR primers targeting nirS, nir K and nosZ genes for community surveys of denitrifying bacteria with DGGE. FEMS microbiology ecology 49 : 401 417. Throbck, I.N., Johansson, M., Rosenquist, M., Pell, M., Hansson, M., & Hallin, S. (2007). Silver (Ag+) reduces denitrification and induces e nrichment of novel nirK genotypes in soil. FEMS Microbiology Letters 270 : 189 194. Wallenstein, M.D., Myrold, D.D., Firestone, M., & Voytek, M. (2006). Environmental Controls on Denitrifying Communities and Denitrification Rates: Insights from Molecular Me thods. Ecological Applications 16 : 2143 2152. Wei, W., Isobe K., Nishizawa, T., Z hu, L., Shiratori, Y., Ohte, N., Ko ba, K., Otsuka, S., & Senoo, K (2015). Higher diversity and abundance of denitrifying microorganisms in environments than considered previously. The ISME journal 9 : 1954. Wijler, J. & Delwiche, C.C. (1954) Investigations on the denitrifying process in soil. Plant and Soil 5 : 155 169. Zumft, W.G. (1997) Cell biology and molecular basis of denitrification. Microbiology and Molecu lar Biology Review 61 : 533 616.

PAGE 52

42 APPENDIX SUPPLEMENTAL TABLES AND FIGURES Figure S1: Rarefaction curves for nirK gene sequences across all samples (Panel A); s amples containing fewer than 13,000 rea ds were excluded from analy sis. Rarefaction curves for nirS gene sequences across all samples (Panel B); samples containing fewer than 1,000 reads were excluded from analysis.

PAGE 53

43 Figure S2 : Number of observed OTUs and Chao1 richness estimate s for nirK (Panel A) and nirS (Panel B) gene se quences across all sample sites.

PAGE 54

44 Figure S3 : Comparison between the number of observed nirK OTUs (panel A) and the number of observed nirS OTUs (panel B) to the sediment pH at each site. Samples within the four regions are indicated by separ ate colors.

PAGE 55

45 Figure S4 : Canonical Correspondence Analysis (CCA) of relative gene abundance of nirS gene sequences for individual samples within the New Dominion region. TRW Iron significantly correlated ( Bonferroni corrected p value <0.0 5 ) to the distribution of relative gene abundance.

PAGE 56

46 Figure S5 Comparison of conductivity to the sum of a ll dissolved metals at each site with a Pearson correlation coefficient of 0.83 and an R 2 value of 0.68. 0 200 400 600 800 1000 1200 1400 1600 1800 0 50000 100000 150000 200000 250000 300000 350000 400000 450000 Carib01Aug13 Carib01Jun13 Carib01Sep14 Carib02Aug13 Carib02Jun13 Carib02Jun14 Carib02Sep14 Carib03Aug13 Carib03Jun13 Carib03Jun14 Carib03Sep14 NDMD02Aug13 NDMD02Jun13 NDMD02Jun14 NDMD02Sep14 FenND01Aug13 FenND01Jun13 FenND01Jun14 FenND01Sep14 FenND03Aug13 FenND03Jun14 CSIBAug13 CSIBJun13 CSIBJun14 CSIBSep14 FenIB01Aug13 FenIB01Jun13 FenIB01Jun14 FenIB01Sep14 FenIB02Aug13 FenIB02Jun13 FenIB02Jun14 FenIB03Aug13 FenIB03Jun13 FenIB03Jun14 FenIB03Sep14 HF04Aug13 HF04Jun13 Conductivity ( S/cm) Dissolved Metal Aggregate ( g/L) Site Dissolved Metal Aggregate ( g/L) Conductivity ( S/cm)