Using Molecular Tools to Elucidate Controls on Microbes Driving the Nitrogen Cycle in Marine Sediments

Marine sediments harbor metabolically versatile bac teria whose activities can influence the cycle of nutrients on global scales. Microbial communities driving nitrogen (N) cycling are extremely diverse, thus ma king it difficult to identify the functional groups and elucidate controls on their a ctivity. Denitrifiers in sediments remove significant amounts of N from the coastal oc ean, while diazotrophs are typically considered inconsequential. Recently, N f ixation has been shown to be a potentially important source of N in coastal sedime nts, however, the environmental drivers controlling this process are poorly underst ood. The goal of this dissertation was to identify and target the likely active denitr ifie s and N fixers in coastal marine sediments through the analysis of genes expressed f or proteins essential for denitrification, a nitrite reductase ( nirS) and nitrogen fixation, a nitrogenase subunit (nifH). Subsequently, quantitative PCR and RT-PCR were u sed to follow the changes in abundance, distribution and ifH expression of the dominant diazotrophic groups in response to environmental conditions. Two groups of diazotrophs related to anaerobic sulfur/iron reducers and sulfate reducers dominated nifH expression in Narragansett Bay (RI, USA) sediments. Increased seawater tempera tur nd severe hypoxia appear to be influencing the proliferation and activity of these two bacterial groups. Oxygen depletion also affects sediment porewater nutrients , indicating a shift in benthic microbial processes. In offshore sediments, ifH expression was related to UCYN-A, a unicellular cyanobacterium. These findings suggest tha UCYN-A, a known tropical and subtropical open ocean symbiont, has a broader thermal tolerance than previously assumed and can survive in the benthos after the li fespan of its eukaryotic host. Diazotrophic activity by these microbial communitie s in marine sediments is an unanticipated contribution of fixed N to coastal sy stems. Climate change may exacerbate the environmental conditions in which th ese microbes become active, consequently altering the global marine nitrogen cy cle in unprecedented ways.


INTRODUCTION Benthic Nitrogen Fixation in Coastal Systems
Benthic sediments in temperate coastal and estuarine ecosystems are generally considered areas of net nitrogen (N) loss through the removal of fixed N by denitrification (Nixon et al. 1996;. In these regions, denitrification typically exceeds rates of N fixation, the conversion of N 2 gas into ammonia. The result is a depletion of available N from the system .
Therefore, inputs of N through N fixation are commonly regarded as an inconsequential contribution to most estuarine N budgets .
Nearshore marine environments are characterized by high primary production of phytoplankton and rooted macrophytes, including seagrasses and salt marsh plants (Nixon et al. 1983). The availability of N in coastal marine systems limits primary production (Howarth 1988;Vitousek et al. 1991). To sustain high levels of growth, macrophytes require substantial sources of inorganic N (Patriquin 1972). While macrophytes largely recycle organic N entrained in the sediment (Iizumi et al. 1982;Dennison 1987;Caffrey et al. 1992;White et al. 1994), to combat the significant losses of N removed or denitrified, sources of new N are also essential to support their nutrient demands (Capone 1988). Consequently, resolving the importance N fixation in coastal benthic sediments has been of great interest to the scientific community.
Benthic N fixation has been extensively studied in areas associated with the photic zone, including sediments vegetated by macrophytes Capone 1988; and photosynthetic microbial mats . Nitrogen fixation in non-vegetated, unlit sediments is just recently gaining more attention ). Many of the early studies documented N fixation by utilizing the acetylene reduction assay (ARA) to measure activity of the nitrogenase enzyme, the conserved protein complex catalyzing N fixation (eg. Welsh et al. 1997)). More recent studies have used N 2 /Ar flux measurements to document N fixation by observing a shift in the net balance of N fluxing into the sediment from N fixation as opposed to out of the sediment from denitrification ).
However, these methods provide no information regarding the identity of the potentially active microbes driving N fixation in the environment, their diversity or the functional redundancy of this process in unvegetated sediments. Phylogenetic analysis of nifH, the gene encoding the iron protein component of the nitrogenase enzyme, has been widely used to study diazotrophs in a variety of environments reviewed in . Coupling molecular approaches and biogeochemical measurements can provide useful information regarding the ecology and functioning of benthic habitats.

Diazotrophic Activity By Sulfate Reducers Supports Macrophyte Growth
Intertidal zones of temperate coastal systems are vegetated by extensive meadows of rooted macrophytes including Spartina alterniflora, Zostera marina, and Z. noltii. The productivity of these meadows is largely supported by external inputs and by the recycling of remineralized nitrogen (Iizumi et al. 1982;Dennison 1987;Caffrey et al. 1992;White et al. 1994). Several studies have shown that the concentrations of inorganic N in the sediment porewater are insufficient to meet the growth requirements of the macrophytes (Patriquin 1972;Short 1983;Moriarty et al. 1985). Heterotrophic N fixation in the rhizosphere provides a significant source of new N for plant growth, overcoming the limitation of N availability and, in turn, influences primary productivity of these ecosystems (Capone 1988).
High rates of N fixation have been recorded in sediments colonized by rooted macrophytes (Table 1). Rates ranged from 0.1 mg N m -2 d -1 to 7.3 mg N m -2 d -1 and were dependent on species, location and seasonality (Table 1). In the Bassion d'Arachon, France, N fixation contributed ~6-12% of the nitrogen requirement of the seagrass, Z. noltii . Similar values were reported for Z. marina in Great South Bay, New York . Both studies also measured N fixation after the addition of sodium molybdate, a specific inhibitor of sulfate respiration. Severely reduced nitrogenase activity was detected in the rhizosphere, indicating sulfate reducing bacteria were important contributors to N fixation in the vegetated sediments .
To gain a better understanding on the association between the nitrogen fixing bacteria and the macrophtes, several studies assessed the location of and controls on nitrogenase activity in the rhizosphere. Roots and rhizomes accounted for about 31% and 91% of rhizosphere nitrogenase activity in Z. noltii and S. maritime, respectively (Table 1) (Nielsen et al. 2001). In Z. marina, about 39% of the rhizosphere N fixation was associated with roots, while only 4% was associated with rhizomes . The roots and rhizomes of the Zostera seagrasses comprise a small portion of the rhizosphere biomass and, therefore, N fixation from the bacteria colonizing the sediment is still significant (Nielsen et al. 2001). The combination of carbon availability and concentrations of ammonium appear to control N fixation in macrophyte-associated bacteria. The supply of organic substrate excreted from the plant can support heterotrophs, such as sulfate reducers, and has been shown to stimulate nitrogenase activity Nielsen et al. 2001). Several studies have debated the impact of sediment porewater ammonium on N fixation activity. In a Z. noltii meadow, about 30% of the nitrogenase activity remained after addition of 1mM ammonium chloride (Welsh et al. 1997), which contrasted results from a fjord in Denmark, in which the rhizosphere of Z. marina beds were not sensitive to ammonium additions ).
Hundreds of diazotrophic bacteria were isolated from salt marsh rhizospheres and characterized both morphologically and physiologically (Bagwell et al. 1998).

Representative species of Enterobacter, Vibrio, Azotobacter, Spirilla, Pseudomonas
and Rhizobia, as well as strains with no clear taxonomic affiliations, were recovered (Bagwell et al. 1998). Culturing techniques, however, can be biased, so the natural diazotrophic communities inhabiting the rhizosphere of S. alterniflora were analyzed using a nifH fingerprinting method (Table 2) (Lovell et al. 2000). The diversity of N fixers was dominated by nifH sequences closely related to γ-Proteobacteria, including Azotobacter chroococcum and Pseudomonas stutzeri, and sulfate reducing anaerobes, such as Desulfonema limicola and Desulfovibrio gigas (Lovell et al. 2000). The same method was performed concurrently with measuring nitrogenase activity in sediments containing tall and short S. alterniflora stands (Piceno et al. 1999). Nitrogenase activity differed between these sites, but the composition of the diazotrophs remained highly stable both spatially and temporally (Piceno et al. 1999). Conversely, another study of the same macrophyte detected seasonal variability of the rhizosphere diazotroph assemblages, in which several members of the anaerobic population were only detected during the winter months (Table 2) (Gamble et al. 2010). Only a small subset of the N fixing microbial community was identified to be actively expressing nifH in S. alterniflora rhizospheres (Table 2) . The diazotrophs were closely related to Pseudomonas stutzeri, Vibrio diazotrophicus, Desulfovibrio africanus and D. gigas . N fixation by macrophyte-associated bacteria helps contribute to the overall productivity of the plant communities by providing a significant source of new N to coastal ecosystems (Capone 1988;.

Mats
Marine benthic microbial mats frequently develop in the intertidal regions of estuaries and coastal embayments (Cohen et al. 1984;Cohen et al. 1989). The mats host metabolically diverse groups of microorganisms that contribute to critical steps in biogeochemical cycles (Cohen et al. 1984;Van Gemerden 1993). Temperate estuaries and coastal regions are typically considered N depleted (Vitousek et al. 1991), which in turn limits the primary productivity and growth of mat microbial communities (Paerl et al. 2000). N fixation by bacteria inhabitants provides a source of new N to aid in the development and persistence of the mat (Stal et al. 1985;Paerl et al. 2000).
Nitrogenase activity was generally attributed to diazotrophic cyanobacteria, which are visually prominent and dominate the biomass and energy production of most phototrophic mats Bebout et al. 1993). However, with the advent of molecular techniques, came the discovery that in addition to cyanobacteria, other microbes may be significant contributors of fixed N to the mats.
Microbial mats are composed of a rich assemblage of bacteria possessing the ability to fix N (Paerl 1990). By targeting the nifH gene, a high diversity of heterotrophic diazotrophs, many closely related to anaerobes including Chromatium (purple sulfur bacteria), Desulfovibrio (sulfate reducers) or Clostridium (strict anaerobes) spp., were detected in a North Carolina cyanobacterial mat (Table 2) (Zehr et al. 1995). The genetic potential to fix N shifted from diazotrophic cyanobacteria dominating the mat in the summer to heterotrophic N fixers in the fall and winter (Zehr et al. 1995 (Moisander et al. 2006). Based on these molecular studies, the key diazotrophs potentially providing a source of N to the microbial mat are related to cyanobacteria and sulfate reducers.
High rates of N fixation in microbial mats have been reported for several temperate coastal systems (Table 1). Joye et al. (1994) reported rates as high as 79 mg N m -2 d -1 . This fixed N by the mat microbial communities is locally essential to the development and high productivity of the mat. However, due to the restricted areal distribution of microbial mats and high rates of denitrification, their contribution to the total N budget in coastal marine ecosystems is minor (Joye et al. 1994;.

Reducers
The benthos of most estuarine and coastal ecosystems consists primarily of non-vegetated sediments. These bare sediments are considered net sinks for N due to high rates of denitrification (Nixon et al. 1996;. In fact, during the 1970s, N fixation in the benthic sediments of Narragansett Bay, Rhode Island and Rhode River Estuary, Maryland accounted for <1% and <5% of the total annual influx of N into the systems, respectively (Marsho et al. 1975;Seitzinger 1987). Reported rates of N fixation during the 1970s and 1980s were usually less than 1 mg N m -2 d -1 (Table 1), with higher values detected in organically rich sediments such as those found in Waccasassa Estuary, Florida (Brooks et al. 1971). For temperate estuarine sediments,  calculated an annual N fixation rate of 0.4 ± 0.07 g N m -2 year -1 . Due to the low rates of N fixation in these benthic habitats, very few studies focused on N fixation as a source of N in unvegetated sediments.
During the summer of 2006, several sites in upper Narragansett Bay exhibited high rates of net N fixation ). The dramatic reversal in net sediment N flux was attributed to climate change induced oligotrophication of the bay ).  suggested N fixation in the estuarine sediments during those summer months added a net (56-154) x 10 6 g of N to the bay over the annual cycle. To further investigate the impact of shifting phytoplankton bloom phenology, a biogeochemical-molecular approach was used to analyze the timing of organic matter deposition to mesocosms containing Narragansett Bay sediments . Increased nifH expression occurred concomitantly with lower rates of denitrification in sediments starved of organic matter . nifH mRNA transcripts were limited to two phylogenetic groups related to Pelobacter carbinolicus and Desulfovibrio vulgaris, anaerobes that can reduce sulfur and sulfate compounds, respectively (Table 2) .
Several studies in other coastal ecosystems have also suggested sulfate reducing bacteria are responsible for the measured nitrogen fixation activity. N fixation has been documented as an important mechanism for adding N (10.8 ± 8.5 mg N m -2 d -1 ) to several Texas estuaries . In Chesapeake Bay sediments, many nifH DNA sequences phylogenetically grouped with sulfate reducers, such as Desulfovibrio salexigens and Desulfobacter curvatus (Table 2) . Additionally, nitrogenase activity decreased substantially when sulfate reduction was inhibited .

Thesis Motivation and Outline
Unvegetated marine sediments are typically regarded as net sinks for N (Nixon et al. 1996; and N fixation is thought to be a negligible process due to the high input of N into the system ) and the process being repressed by combined N . Recently, benthic unvegetated sediments in several coastal ecosystems exhibited high rates of N fixation, challenging the denitrification-dominated paradigm (Table 1) . The environmental factors driving this microbially mediated switch are not fully understood. The primary objective of this dissertation research is to identify and target the potentially active bacterial groups driving nitrogen fixation and denitrification in estuarine sediments and subsequently follow the changes in community composition and activity of these important populations of microorganisms under different environmental conditions. The response of these microbes may provide insight into the environmental controls driving these N cycling processes and help us better understand the bacteria's influence on the ecosystem.
The capability to fix nitrogen or denitrify is widespread among diverse prokaryotic taxa Zumft 1997). Analysis of expressed functional genes associated with nitrogen fixation (nifH) and denitrification (nirS) can used to identify the most potentially active microbes driving these processes in the environment. Very few studies have used this method to study the biodiversity of likely active N cycling microbes in marine sediments. In Chapter 1, the expression of nifH and nirS were analyzed along the estuarine gradient of Narragansett Bay to an offshore continental shelf site over a temporal cycle. The dominant bacterial groups expressing nifH were related to anaerobic sulfur/iron reducers and sulfate reducers. The highest abundance and nifH expression of both groups was detected in upper Narragansett Bay sediments, which may be experiencing enhanced environmental disturbance due to warming seawater temperatures, seasonal hypoxia and the iron gradient exhibited in the upper bay.
At the offshore sites, nifH expression was closely related to a unicellular cyanobacterium, UCYN-A, recently renamed Canditatus Atelocyanobacterium thalassa . UCYN-A is a known tropical and subtropical open ocean diazotrophic cyanobacterium  that has a symbiotic association with a unicellular prymnesiophyte  suggesting nitrification (an oxygen requiring process) was stimulated. The greatest bulk nifH expression was detected in sediments usually impacted by hypoxia. Not surprisingly, the dominant groups expressing nifH in the sediment were related to iron/sulfur and sulfate reducing bacteria (previously identified in Chapter 1). The abundance and nifH expression of the dominant N fixing groups did not appear to respond to low oxygen conditions (2-3 mg/L DO). Perhaps closer to anoxic conditions are necessary to promote nitrogenase activity in these diazotropic groups.
The work from this dissertation demonstrates that anaerobes related to sulfate reducers and iron/sulfur respiring bacteria are the organisms that can add fixed N to non-vegetated marine sediments. In other coastal benthic habitats, N fixation provides a critical source of nutrients to support the growth and productivity of marine seagrasses, salt marsh plants and photosynthetic microbial mats (Capone 1988). Based on sulfate respiration inhibition assays and phylogenetic analyses, sulfate reducing bacteria have been recognized as important active N fixers in these benthic environments, supporting our findings (Table 2). These anaerobic bacteria appear to be significant contributors to the fixed N pool in marine benthic ecosystems. However, very few studies have investigated the controls of N fixation in anaerobic bacteria, including sensitivity to combined N and oxygen tolerance. Regulation of anaerobic N fixation studies are needed to better understand the controls of this process in benthic habitats and to predict how coastal ecosystems will respond to future environmental changes. Moriarty, D., Boon, P., Hansen, J., Hunt, W., Poiner, I., Pollard, P., et al. (1985). Microbial biomass and productivity in seagrass beds. Geomicrobiology Journal 4(1): 21-51.  concentrations of iron at site PRE, may influence the abundance and nifH expression of these two bacterial groups.

Introduction:
Estuaries and continental shelves are dynamic ecosystems that receive and process large inputs of anthropogenic added nutrients from human activity (Pinckney et al. 2001;Liu et al. 2010). Most of the nitrogen (N) is removed by denitrification in sediments in these coastal regions (Nixon et al. 1996;. Denitrification is an anaerobic microbially mediated process in which oxidized forms of nitrogen are sequentially reduced to N 2 gas. This pathway is responsible for the major loss of fixed nitrogen in coastal margins, which in turn drives the global N deficit ).
Biological nitrogen fixation, the conversion of N 2 gas into ammonia, is usually regarded as an inconsequential component in most estuarine nitrogen budgets . However, N fixation is being considered increasingly important in specific benthic habitats, particularly in areas associated with the photic zone including photosynthetic microbial mats  and sediments vegetated by sea grasses  and salt marsh plants . Due to the high rates of denitrification reported in estuarine systems Nixon et al. 1996), little attention has been given to whether nitrogen fixation occurs in non-vegetated sediments. Recently, benthic sediments from the temperate estuary Narragansett Bay (RI, USA) were shown to exhibit a seasonal switch in nitrogen cycling with high rates of net N 2 fixation, challenging the denitrification-dominated paradigm ).
Functional genes encoding cellular proteins that mediate biogeochemical transformations not only provide insight into the ecology of a system, but also can be used to investigate the diversity of specific groups of microorganisms (eg. denitrifiers and nitrogen fixers) in the environment. The key intermediate step in the denitrification pathway, reduction of nitrite to nitric oxide, is catalyzed by the NirS and NirK proteins, two known forms of dissimilatory nitrite reductase. Bacteria harbor copies of either nirS or nirK genes and both have been used as gene markers used for ecological studies to follow denitrifier community composition (Braker et al. 2001;Avrahami et al. 2003). nirS was targeted for this study as the gene is preferentially found in marine sediments through PCR-based methods, while nirK is detected more readily in soil (Braker et al. 2000). The study of diazotroph diversity has been largely based on the phylogenetic analysis of nifH , the gene encoding the nitrogenase iron protein component of the conserved nitrogenase protein complex, an enzyme catalyzing nitrogen fixation in these microbes .
The capability to denitrify or fix nitrogen is distributed through diverse prokaryotic taxa throughout the bacteria and archaea Zumft 1997).
Several studies in Chesapeake Bay (MD and VA, USA) have sought to understand the mechanisms driving the distribution of denitrifiers (Bulow et al. 2008) and diazotrophs Jenkins et al. 2004;Short et al. 2004;Steward et al. 2004;Moisander et al. 2007 Bulow et al. 2008) and to our knowledge, there has been only one report of nifH expression in non-vegetated benthic sediments ). Our purpose is to go beyond DNA diversity studies, targeting mRNA from sediments to understand the diversity of the assemblages of denitrifiers and nitrogen fixers expressing the nirS and nifH genes, respectively.
In this study, we examined the active nirS-and nifH-transcribing microbial populations to determine the likely functional denitrifers and diazotrophs in benthic sediment samples collected along the estuarine gradient from the head of Narragansett Bay to an offshore continental shelf site. One of our aims was to determine if the expressed nirS and nifH sequence diversity patterns resembled the unique distribution of denitrifiers and nitrogen fixers in Chesapeake Bay, in which the diversity of nirS and nifH decreased along the estuarine gradient from the freshwater end to the more saline mouth (Moisander et al. 2007;Bulow et al. 2008). Nitrogen fixation in bare estuarine sediments is recently becoming recognized as an important process occurring in coastal systems Bertics et al. 2012a;Bertics et al. 2012b;, so for the remainder of the study we focused on quantifying the transcriptional activity of bacterial populations actively expressing nifH. Predominant expressed nifH sequences were used to develop primers and probes for quantitative PCR to follow the changes in abundance, distribution and nifH expression of the microbial groups along the estuarine gradient over an annual temporal cycle. To investigate how the environment impacts the biodiversity of genetically active diazotrophs, we also examined potential mechanisms (eg. oxygen, temperature and salinity) driving shifts in these diazotroph communities in the benthic sediments.

Results:
Expression of functional genes associated with nitrogen fixation (nifH) and denitrification (nirS) were analyzed at four stations (PRE,MNB,RIS2 and MP1) along the estuarine gradient of Narragansett Bay to an offshore continental shelf site over a temporal cycle (Fig. 1, Table S1). (Refer to Experimental Procedures for more in-depth collection and site description.)

Phylogenetic Relationships of Expressed nirS and nifH Sequences
Expression of nifH was detected at all four sites throughout the temporal cycle ( Fig. 2). The spatial distribution of nifH mRNA transcripts was variable along the sediment depth gradient and did not appear to be impacted by location or season of collection ( Fig. 2). nirS expression was also observed at all four stations, however it was usually detected in the warmer sampling months (May through October) (Fig. 2).
nirS expression was localized to the top 2 cm except at station MP1 (Fig. 2). The expression of nirS was rarely detected without concurrent nifH expression (Fig. 2).
Phylogenetic analysis of nirS mRNA transcript sequences shows that they are distributed throughout several diverse groups amongst nirS phylogeny (Fig. S1, Table   S2). The majority of expressed nirS sequences group close to Azoarcus tolulyticus, a bacterium notable for its ability to both denitrify and fix nitrogen (Zhou et al. 1995) ( Fig S1). Both spatially and seasonally, a major shift was not detected in the distribution of denitrifiers expressing nirS.
Phylogenetic analysis of the expressed nifH sequences from these sites shows that they are restricted to two main nifH phylogenetic groups (nifH Clusters I and III, as previously defined ) and group with known sulfate, sulfur and iron reducing bacteria (Fig. 3, Table S3). The majority of expressed nifH sequences (61) are within group NB3 which has as its most closely related cultivated species Pelobacter carbinolicus, an anaerobe known to reduce sulfur and iron compounds (Lovley et al. 1995) (Fig. 3). The second largest group of expressed sequences, group NB7, contains 22 expressed nifH sequences with the most closely related cultivated bacteria being the sulfate reducers Desulfovibiro salexigens and Desulfovibrio vulgaris ( Fig. 3). Even though the expressed nifH and nirS sequences we recovered are constrained to a few broad taxonomic groups, there is microdiversity detected among the different sites.

Diazotroph Diversity Shifts Along the Estuarine Gradient
The diversity of microbes expressing nifH in the sediment decreases along the estuarine gradient, from 10 groups identified at the head of Narragansett Bay (PRE) to 3 groups at RIS2 (Fig. 4). Even though we only sampled site MP1 once, the trend continues and 3 groups were detected to be expressing nifH at the most offshore station (  established highest abundances at site MNB, expression of nifH by these groups follows the estuarine gradient, with maximum levels observed at site PRE decreasing out to the continental shelf station, MP1 (Fig. 7). The expression of both microbial groups is statistically significantly higher at site PRE compared to the other three sites (Tables S4 and S5). No statistical differences were detected in abundance or expression of groups NB3 and NB7 over the seasonal cycle or along the depth gradient at the sampling locations.

Discussion:
Recently, benthic sediments from several locations in upper Narragansett Bay, including sites PRE and MNB, were shown to exhibit a seasonal switch in nitrogen cycling with high rates of net nitrogen fixation (up to -650 µmol N 2 -N m -2 h -1 ) during the summer months ). These findings challenge the denitrification-dominated paradigm (Christensen et al. 1987;Hulth et al. 2005), in which nitrogen fixation was thought to be a negligible process occurring in coastal systems due to the high input of N into the system ) and the process being repressed by combined N . Numerous studies have reported the importance of nitrogen fixation in coastal habitats, specifically in photosynthetic microbial mats and in sediments vegetated by salt marsh plants and seagrass (e.g. . However, estuaries are still considered a net sink for nitrogen. More recently, nitrogen fixation Bertics et al. 2012a) as well as nifH expression ) has been detected in nonvegetated, bare sediments. We present gene expression data identifying and targeting the likely active microbes driving N fixation in a background of diverse genetic potential in marine sediments. Changing environmental conditions including elevated water temperatures due to climate change (Scavia et al. 2002) and increases in eutrophication-induced hypoxia Zhang et al. 2010) may be driving active heterotrophic N fixation in coastal and shelf sediments.

Variable nifH and nirS Expression
We detected nifH expression at all sites and time points, however the expression varied throughout the depth profile with no apparent seasonal correlation.
These results were not surprising as variable nifH expression has been recently reported in mesocosm experiments from sediments collected at site MNB ) and nitrogen fixation has been detected in sediments at depths >5cm . nirS expression was likely only detected in the top 2 cm because in coastal sediments denitrification is typically coupled to nitrification (Seitzinger et al. 1984;Nowicki 1994), an oxygen requiring process and needs to occur in the surface sediments. nirS expression was detected in deeper sediments at Site MP1, an area that has not been well studied, which may be attributed to slightly increased rates of direct denitrification occurring on the continental shelf. For example, on the mid-Atlantic Bight, 9% of nitrogen removed was accounted for by direct denitrification (Laursen et al. 2002).

Fluctuating Environmental Conditions Promote Diazotroph Diversity
The diversity of major nitrogen cycling organisms, including denitrifiers, nitrifiers and nitrogen fixers, have been well studied in Chesapeake Bay by detecting the functional genes nirS (Bulow et al. 2008), amoA (Francis et al. 2003;Ward et al. 2007) and nifH (Moisander et al. 2007), respectively. The diversity pattern was similar for all genes studied, with the greatest diversity observed at the freshwater head of Chesapeake Bay decreasing to the mouth. For this study, we focused on how the changing environmental variables along the down-bay gradient impacted the active microbes driving nitrogen fixation in Narragansett Bay sediments. We see a similar pattern with the potentially active nitrogen fixers, in which the highest diversity of microbes expressing nifH is detected at site PRE, near the head of the Bay and decreasing out to site MP1 on the continental shelf. Narragansett Bay, like Chesapeake Bay, exhibits an estuarine gradient with respect to temperature, salinity and nutrients (Kremer et al. 1978;. The northernmost area, site PRE, is a dynamic region with large fluctuations in temperature, salinity, oxygen and nutrients over a temporal cycle   (Connell 1978;Huston 1979). At low levels of disturbance, more competitive organisms will dominate the ecosystem while at high levels of disturbance, organisms may not be able to adapt to their surroundings.
IDH was originally developed for tropical rainforests and coral reefs (Connell 1978), and has recently been applied to plankton communities (Floder et al. 1999 nifH sequence types were identified that were related to various sulfur and sulfate reducing bacteria, including Desulfovibrio spp. and Desulfobacter spp., which are microbes that have been shown to fix nitrogen in culture ). Based on acetylene reduction and sulfate reduction inhibition assays, Bertics et al. attributed the nitrogen fixation rates to sulfur and sulfate reducing bacteria , corroborating our findings that these microbes are likely to be driving nitrogen fixation in these sediments. Our expressed nifH sequences also group with nifH RNA sequences recently reported from mesocosm experiments with sediment collected at site MNB ). Both microbial groups NB3 and NB7 are highly abundant in the sediment at sites PRE and MNB in upper Narragansett Bay (Figs 5c and 6c). Nitrogen fixation by these microbes could provide unanticipated inputs of nitrogen into ecosystems already stressed by eutrophication, including Narragansett Bay. Denitrification may not balance the anthropogenic inputs of N to the extent previously believed, and the sediments could instead become a net source of N exacerbating the nutrient loading into the system.
Temperature and dissolved oxygen concentrations are potentially key drivers of growth and activity of NB3 and NB7 as these groups are related to mesophilic anaerobes. We detected the highest nifH expression during the summer months when the water temperature was the warmest (18 ˚C) at sites PRE and MNB. The water temperatures in Narragansett Bay can reach up to 24 ˚C during the summer (Kremer et al. 1978). Bottom water temperature at site RIS2 during July 2010 was 13 ˚C and at site MP1 during August 2011 only reached 8˚C. We detect the lowest abundance and nifH expression of NB3 and NB7 at sites RIS2 and MP1, so the offshore regions may not provide an optimal temperature range for these microbial populations.
Low oxygen events may also be promoting these anaerobic diazotrophs to fix nitrogen. For the last several decades, episodic hypoxia has been documented in Narragansett Bay Bergondo et al. 2005;Melrose et al. 2007;). The severity of hypoxia generally decreases in intensity with distance from site PRE, in the Providence River Estuary, following the north-south gradient of nutrients, phytoplankton and freshwater influence Prell et al. 2004;Melrose et al. 2007;). During the same summer when high rates of net nitrogen fixation were recorded at sites PRE and MNB ), there were several bouts of widespread hypoxia that severely impacted regions of upper Narragansett Bay ). The occurrence of these low oxygen events may be stimulating the growth and activity of groups NB3 and NB7. We observed the highest expression of nifH at site PRE, which is an area that usually experiences severe hypoxia during the summer months (Saarman 2002). In some years, hypoxia can reach as far south as site MNB Melrose et al. 2007). One possible explanation for the increase in abundance and expression at these northern sites is that hypoxic conditions caused by high rates of microbial respiration in the sediments may disrupt the link between coupled nitrification-denitrification, as the former is an oxygen requiring process. A shrinking oxic sediment layer and inhibition of the coupled N-removal pathway could thus expand the niche for sulfur and sulfate reducers to thrive.
In addition to elevated water temperatures and hypoxia, regions of upper Narragansett Bay are also exposed to high levels of metal contaminants and organic carbon due to anthropogenic input Nixon 1995, Murray 2007(Nixon 1995Rincón 2006;Murray et al. 2007). Concentrations of iron have been shown to match up with the mapped extent of hypoxia in the Bay (Prell et al. 2004;Rincón 2006). Both iron and organic carbon concentrations in the sediment decrease along the estuarine gradient (King et al. 1995;Murray et al. 2007). Several of the potentially active nitrogen fixers are related to heterotrophic anaerobes that have the ability to also reduce iron, including Pelobacter carbinolicus (Nealson et al. 1994;Lovley et al. 1995). The energy gained from respiring iron, a fairly energy yielding electron acceptor, and consuming organic carbon may be promoting the growth and activity of these diazotrophs in upper Narragansett Bay sediments.
In comparison to the sites PRE and MNB in Narragansett Bay, the offshore locations are more uniform in terms of seasonal temperature differences and in similar rates of denitrification. We detected the lowest abundance and nifH expression of groups NB3 and NB7 at the offshore sites RIS2 and MP1. Environmental conditions, including temperature and dissolved oxygen concentrations, at these offshore sites may not be optimal for these potentially active diazotrophs to thrive in these regions.
Bottom water temperatures are unlikely to rise to the optimal threshold for mesophiles, possibly impeding the growth and activity of these groups of diazotrophs.  (Devol 1991;Devol et al. 1997).

Conditions in upper
Narragansett Bay may be more optimal for growth and activity of groups NB3 and NB7, while offshore sediments may not provide an appropriate niche for these microbes to thrive.

Conclusion:
Benthic nitrogen cycling processes are influenced by changes in environmental conditions, including temperature, dissolved oxygen concentrations, salinity and organic matter loading. Climate change is predicted to increase seawater temperatures (Scavia et al. 2002) and exacerbate eutrophication-driven hypoxia Zhang et al. 2010). Since the 1960s, the number of hypoxic zones has approximately doubled each decade ) and these expanding low oxygen events have the potential to perturb the functioning of the nitrogen cycle in estuarine ecosystems.
Narragansett Bay, like many coastal ecosystems, is exposed to elevated water temperatures and exhibits seasonal hypoxia Zhang et al. 2010).
Microbes related to iron/sulfur and sulfate reducers that express nifH are highly abundant and have increased levels of nifH mRNA transcripts in Narragansett Bay sediments. These diazotroph communities may proliferate and increase activity in response to elevated water temperatures, episodic hypoxic events or high concentrations of organic carbon in the bay. Consequently, nitrogen fixation by these microorganisms in coastal sediments could provide unanticipated inputs of nitrogen into environments already stressed by eutrophication, significantly altering the nitrogen cycle in unprecedented ways.

Study Sites
We sampled for sediment at four sites in southern New England coastal waters  Table S1).

Field Methods
Intact sediment cores (10 cm inner diameter and 30.5 cm long) were collected at site PRE using a 5 m pull corer and at site MNB by SCUBA divers. At sites RIS2 and MP1, sediment cores were collected using a box corer (0.25 m 2 ) and pre-mounted PVC cores. All cores were transported to and stored in the dark at field bottom-water temperature in a walk-in environmental chamber at the Graduate School of Oceanography at the University of Rhode Island. The cores were left uncapped with air gently bubbling through the overlying water for about 8-12 h prior to net N 2 flux incubations. For details regarding N 2 flux methods and offshore flux results, refer to Heiss et al. 2012.

Sub-sampling and Nucleic Acid Extractions
After the net N 2 flux incubations were completed, the cores were sub-sampled using a 60 mL syringe. The sub-cores were flash frozen in liquid N 2 and sectioned into 1 cm segments from the sediment water interface to 6 cm in depth. The frozen sediment cross-sections were cut up to yield 0.25 g and 0. µL of 2 µM outer reverse primers for both our genes of interest, nifH3 and nirS6R (Tables S6 and S7). After the reverse transcriptase was added, the mixture was incubated at 50°C for 50 min. All the other steps followed the instructions of the manufacturer. For every sample, we also included controls that did not contain reverse transcriptase to confirm there was no DNA contamination in the subsequent PCR amplification.

Functional Gene Sequence Analysis
The nifH gene from environmental cDNA was isolated using nested PCR with degenerate outer primers nifH4-nifH3 and inner primers nifH1-nifH2 (Table S6). Both rounds of PCR consisted of an initial denaturation step of 2 min at 94°C, cycling steps that included: a denaturation step of 30 s at 94°C, an annealing step of 30 s at 50°C, and an extension step of 1 min at 72°C. All reactions had a final extension step of 7 min at 72°C. First round reactions had 25 cycles and the second round reactions had 30 cycles . nirS was amplified using the primer pair nirS1F-nirS6R (Table S7). After a 2 min initial denaturation step 94°C, a touchdown PCR was performed that consisted of a denaturation step of 30 s at 94°C, an annealing step of 30 s, and an extension step of 1 min at 72°C. During the first 11 cycles, the annealing temperature decreased 0.5°C every cycle starting at 56°C. For the last 25 cycles the annealing temperature was 54°C. A final extension step was performed for 7 min at 72°C (Braker et al. 1998;Braker et al. 2000).
After amplification, the PCR products were loaded on to a 1% agarose (wt/vol) TAE gel. Bands of the correct size were purified using the QIAquick Gel Extraction

Statistical Analyses
One-way analysis of variance (ANOVA) tests were conducted using JMP 10.0.2 to determine statistically significant differences among samples. If the p value was deemed significant (< 0.05), a Tukey-Kramer HSD post-hoc test was performed to distinguish statistical significance between samples compared.

Acknowledgements:
The work presented in this manuscript was supported by funding from the

References:
Avrahami, S., Conrad, R. and Braker, G. (2003). Effect of ammonium concentration on N2O release and on the community structure of ammonia oxidizers and denitrifiers Appl Environ Microb 69 (5)     . Groups NB1-NB7 were previously described , while groups NB9-NB11 are novel to this study. The number inside the group indicates the total number of sequences within the grouping.       Figure 3: Percent of total expressed nifH sequences per site as a function of depth and increasing distance from site PRE. Each color represents a cultivated species our environmental expressed sequences are related to as depicted in the nifH Maximum Likelihood tree (Fig. 3). X-axes have different scales between sites.
Supplementary         (Fig. 3). The quantitative PCR cycling conditions for both target groups included an initial 10 minute denaturation at 95°C followed by 45 cycles of 95°C for 30 seconds and 60°C for 1 minute.

Summary
Nitrogen (N) is often a limiting element for biological productivity in ocean ecosystems (Ryther et al. 1971). Diazotrophic cyanobacteria provide a source of fixed N to lit marine waters, alleviating the N deficit (Stal et al. 2008). A symbiotic unicellular cyanobacterium, UCYN-A, newly renamed Canditatus Atelocyanobacterium thalassa , has recently been recognized as an important N fixer in global oligotrophic oceans ). Although an increase in discoveries of fixed N sources in marine ecosystems, including contributions from unicellular N-fixing cyanobacteria , N fixation in coastal heterotrophic sediments ) and cold seeps (Dekas et al. 2009), marine N fixation is still grossly underestimated, and therefore balancing the N budget has proved challenging (Brandes et al. 2002;Codispoti 2006

Results and Discussion
Constraints on oceanic N fixation to oligotrophic tropical and subtropical waters was largely based on the temperature and growth requirements of the marine filamentous cyanobacterium Trichodesmium (Capone et al. 1997;White et al. 2007), which was believed to be the most abundant and active oceanic diazotroph (Capone et al. 1997) until the discovery of two unicellular cyanobacteria, Canditatus Atelocyanobacterium thalassa, also designated UCYN-A, and Crocosphaera watsonii shown that UCYN-A is found in abundance at colder temperatures, including higher latitudes and deeper in subsurface ocean waters . Their domain has also expanded to include coastal regions (Mulholland et al. 2012) suggesting that oceanic N fixation may be more widespread than previously believed.
Unlike other cyanobacteria, UCYN-A lacks photosystem II, RuBisCo, and the tricarboxylic acid cycle among other crucial metabolic pathways . To support its carbon requirements in oligotrophic oceans, it has been proposed that UCYN-A is a symbiont of carbon fixing hosts  and has a loose symbiotic association with a unicellular prymnesiophyte ). In exchange for fixed carbon for energy and biosynthesis, UCYN-A provides the host with fixed N ( ). Therefore, UCYN-A is a significant contributor of biologically fixed N to the environment, and could also play a central role in the vertical downward flux of organic matter to the deep ocean if it is exported with its associated host.
Estuarine, continental shelf and other coastal margin sediments are responsible for 83% of biogeochemical cycling in the benthos, however these regions only make up only 9% of the total area of the seafloor (Jorgensen 1983). Denitrification, the sequential reduction of oxidized forms of nitrogen to N 2 gas by anaerobic bacteria, is responsible for the major loss of fixed N in coastal margins, therefore contributing to the unbalanced global N budget ). Due to the high inputs of anthropogenic added nutrients, estuaries and continental shelves are not typically considered N limited. Therefore, N fixation was considered to be an insignificant part of the benthic N cycle, unless associated with photosynthetic microbial mats , seagrass ) and salt marsh plant ) vegetation in the photic zone. Only recently has N fixation by anaerobic microorganisms become recognized as an important process occurring in non-vegetated benthic sediments ).
Our goal was to use gene expression to determine if diazotroph activity was an underestimated component of the N cycle in marine heterotrophic sediments. N fixation is catalyzed by the conserved nitrogenase protein complex . nifH, the gene encoding the nitrogenase iron protein component, is a common marker for the phylogenetic analysis of diazotroph diversity ).
UCYN-A nitrogenase expression has been linked to measured rates of N fixation ). Expression of the functional gene nifH was analyzed in sediments collected at four sites (PRE, MNB, RIS2 and MP1) along the estuarine gradient of Narragansett Bay (RI, USA) to an offshore continental shelf site (Fig 1). Sites PRE, MNB and RIS2 were sampled over a seasonal cycle, while at site MP1, sediments were only collected in August 2011. Unexpectedly, nifH mRNA sequences related to UCYN-A were recovered in sediments at sites MNB, RIS2 and MP1 (Fig 2). At site MNB, expressed sequences were detected in the surface sediments (1-2 cm in depth), while the offshore sites contained expressed sequences up to 6 cm, the deepest subsection collected.
The UCYN-A related nifH sequences cluster into two sub-groups, designated A1 and A2 (Fig 2). Sub-group A1 contains mRNA sequences from all three sites and are identical to sequences collected from tropical and sub-tropical waters including station ALOHA, the South China Sea and the North Pacific sub-tropical gyre eddy (Fig 2). Sub-group A2 contains mRNA sequences from site MNB and RIS2 and are closely related to nifH sequences detected in temperate waters including the Western English Channel and the Mediterranean Sea (Fig 2). Although UCYN-A are globally distributed, unlike other marine bacteria including Pelagibacter and Prochlorococus, populations of UCYN-A are homogenous . Not surprisingly, subgroups A1 and A2 shared a >98% pairwise sequence identity and with such low sequence diversity both UCYN-A sub-groups can be assessed with quantitative PCR.
Abundance, distribution and levels of nifH expression of UCYN-A in the sediments were determined by quantitative PCR targeting the nifH gene. The UCYN-A genome contains one copy of nifH and can be used as a proxy for abundance. The abundance of UCYN-A increased along the estuarine gradient of the Bay, with the greatest number of gene copies detected at the most offshore site, MP1 (Fig 3a). At sites PRE, MNB and RIS2, the highest abundance occurred during the winter in January (Fig 3a). During July, UCYN-A was undetectable at site PRE and at very low levels at site MNB (Fig 3a). Diazotroph abundance and N fixation rates in surface North Atlantic coastal waters were measured between Cape Hatteras and Georges Bank. UCYN-A were the most abundant diazotroph measured along the coast and among the highest abundances ever recorded for the cyanobacterium (Carpenter et al.

2008; Moisander et al. 2010). At several regions just west of Georges Bank and in
Block Island Sound, high rates of N fixation with concurrent high abundances of UCYN-A were measured (Mulholland et al. 2012), indicating that these microbes are prevalent and active in the water column near our offshore sites, RIS2 and MP1.
About 53 km to the northeast of our sampling site MP1 (Fig 1), the highest areal rate of N fixation (873.9 umol N m -2 d -1 ) was documented for North American coastal sites (Mulholland et al. 2012). Corroborating our results, the lowest abundance of UCYN-A was also measured in an estuary, at the mouth of Chesapeake Bay (Mulholland et al. 2012).
UCYN-A nifH expression followed a similar trend to abundance, with nifH transcript copies increasing from the head of the Bay to the offshore sites (Fig 3b). At site PRE, nifH expression was undetectable during all sampling time points and very low transcript copies were detected in October and January at site MNB (Fig 3b).
Previous studies have observed nifH transcripts of UCYN-A at water temperatures as low as 12º to 19ºC (Needoba et al. 2007;Short et al. 2007). At site MNB, although low levels, UCYN-A transcripts were detected in January when the bottom water temperature was 2ºC. The highest levels of nifH expression were detected at site MP1 in August and site RIS2 in January with bottom water temperatures at 8º and 5 ºC, respectively (Fig 3b). Although the lowest recorded abundance at site RIS2, UCYN-A were most active in October with a gene expression to abundance ratio of 54.5 when the bottom water temperature was 14ºC (Fig 3c). Since UCYN-A remains uncultivated, the boundaries of its temperature limit are not well understood, and may be lower than previously predicted. Regardless of season, the highest abundance and nifH expression of UCYN-A was detected at the offshore site RIS2 (Fig 4). Peak abundance and expression occurred at site MP1 (Fig 3a and b), but this location was only sampled in August 2011, so was disregarded in analysis. Due to UCYN-A's symbiotic lifestyle, the bacterium may be constrained by the geographical range of its eukaryotic partner.
UCYN-A has a loose extracellular association with a single-celled eukaryotic alga, identified as a prymnesiophyte closely related to Braarudosphaera bigelowii and Chrysochromulina parkeae . Both eukaryotes, although morphologically distinct, appear to contain calcified scales (Gran andBraarud 1935, Saez 2004), which may help stabilize the fragile association. The distribution of B.
bigelowii is restricted to nearshore regions (Hagino et al. 2009), typically in cold, low salinity waters (Bukry 1974). C. parkeae was first identified near South West England and Norway (Green et al. 1972); however, its range is not well-characterized. The global and widespread distribution of UCYN-A (from oligotrophic tropical oceans to temperate coastal regions), along with its loose partner association indicates that UCYN-A may have multiple prymnesiophte hosts. During a two-year temporal study of the Northeastern Atlantic Outer Continental Shelf, the abundance of coccoliths was greatest during the winter months and decreased in the summer and were replaced by diatoms southwest of Georges Bank (Aaron, J. 1980). These findings support our observations, in which we detect the highest abundance and expression of UCYN-A during January at site RIS2. UCYN-A's symbiotic partnership with potentially multiple calcifying prymnesiophytes has important implications for the broad thermal tolerance of the N fixing unicellular cyanobacteria and the global export of carbon and nitrogen into the deep ocean.
When the prymnesiophyte partner sinks to the benthos, presumably the loosely attached UCYN-A gets vertically transported to the sediments as well. UCYN-A lacks key genes for photosynthesis and carbon fixation , and in the water column, obtains organic carbon from its host ). If UCYN-A is no longer associated with its partner in the sediment, its genome does contain genetic machinery to acquire essential nutrients from the environment, including multiple nonspecific sugar transporters ). UCYN-A has a complete suite of enzymes for upper glycolysis and the pentose phosphate pathway to metabolize the sugars for energy and biosynthetic purposes . The genome also encodes for trace metal transporters for iron, molybdenum and nickel which are essential for certain enzymatic function, including N fixation ).
UCYN-A would be capable of surviving in the benthos without its partner due to the sources of organic matter available in the sediment (Rowe et al. 1988). nifH transcripts detected up to 6 cm in depth suggests that UCYN-A remains active in the sediment after the lifespan of its host. UCYN-A diazotrophic activity in coastal sediments strengthens benthic-pelagic coupling and is an unprecedented source of fixed N to the marine system.

Methods Summary:
Sediment cores were collected at four sites (PRE, MNB, RIS2 and MP1) in southern New England coastal waters from May 2010 to August 2011 with bottom water temperatures ranging from 2 to 18°C. Sediment cores were sub-sampled with 60 mL syringes, flash frozen in liquid N 2 and sectioned into 1cm intervals to 6cm in depth. Total DNA and RNA were extracted from sediment samples as previously described .
RNA was transcribed to cDNA and nifH mRNA transcripts were amplified from the cDNA using nested PCR  The cores were left uncapped with air gently bubbling through the overlying water.

Sub-sampling and Nucleic Acid Extractions
The cores were sub-sampled using a 60 mL syringe and were subsequently flash frozen in liquid N 2 . The sub-cores were sectioned into 1 cm segments from the sediment water interface to 6 cm in depth. The frozen sediment cross-sections were cut up to yield 0.25 g and 0.5 g of wet sediment for DNA and RNA isolation,

Functional Gene Sequence Analysis
nifH mRNA transcripts were amplified from environmental cDNA using nested PCR with degenerate outer primers nifH4-nifH3 and inner primers nifH1-nifH2 . Both rounds of PCR consisted of an initial denaturation step of 2 min at 94°C, cycling steps that included: a denaturation step of 30 s at 94°C, an annealing step of 30 s at 50°C, and an extension step of 1 min at 72°C. All reactions had a final extension step of 7 min at 72°C. First round reactions had 25 cycles and the second round reactions had 30 cycles .

Abstract:
Episodic hypoxia is becoming a common occurrence in coastal systems globally, yet it remains unknown how water column oxygen depletion influences nitrogen (N) cycling processes in the sediment. Estuarine sediments are generally considered areas of net N loss through the coupled nitrification-denitrification.
Conversely, inputs of N through N fixation are typically considered negligible.
Recently, anaerobic diazotrophs have been shown to be a potentially important source of N in coastal benthic habitats. Sediments were collected at several sites prior to and after hypoxia in Narragansett Bay (RI, USA). Sets of pre-hypoxic cores were incubated under oxic and hypoxic conditions. Benthic sediment characteristics (e.g. C/N ratio, oxygen penetration and porewater N) were analyzed to determine the influence of oxygen depletion on sediment N cycling microbial activity. Concurrently, nitrogenase subunit (nifH) gene expression was used to identify the likely active nitrogen fixers in the surface sediments. The greatest bulk nifH expression was detected at the site most heavily impacted by hypoxia with anaerobic sulfur/iron and sulfate reducers dominating expression. After exposure to severe hypoxia, a disappearance of porewater nitrate was detected in sediments, indicating coupled nitrification-denitrification was possibly repressed. Oxygen depletion in the water column potentially influences natural sediment redox dependent N transformations and while also stimulating anaerobic N fixation.

Introduction:
Anthropogenic fertilization of coastal marine systems by excess nitrogen (N) can lead to undesirable impacts including eutrophication-induced hypoxia Rabalais et al. 2010;Zhang et al. 2010). Hypoxic zones form when river freshwater and excess nutrients from natural and anthropogenic sources enter coastal waters promoting phytoplankton blooms (Bricker et al. 2008). Eutrophication enhances deposition of organic matter to the benthos, which in turn promotes microbial decomposition and respiration (Meyer-Reil et al. 2000). Consequently, the increased demand for oxygen in the benthos coupled with water column stratification can deplete the system of dissolved oxygen . Hypoxia and anoxia are one of the most deleterious anthropogenic influences facing our marine environments Zhang et al. 2010).
Hypoxia is expanding in coastal regions globally ), yet the impact of oxygen-depletion on benthic N cycling is still poorly understood. In estuarine sediments, nitrogen is removed from the system through coupled nitrification-denitrification . Ammonia is first oxidized into nitrate by specialized groups of nitrifiers. Subsequently, denitrifiers anaerobically reduce the oxidized forms of nitrogen to N 2 gas. Hypoxia can indirectly inhibit denitrification through the suppression of nitrification, an oxygen requiring pathway (Childs et al. 2002), potentially repressing the loss of fixed N from coastal systems.
Benthic nitrogen fixation, the reduction of N 2 gas to biologically available ammonia, is typically considered an insignificant component in most estuarine nitrogen budgets ). However, biological N fixation is a critical source of nitrogen in specific benthic habitats, including sediments vegetated by rooted macrophytes  and photosynthetic microbial mats . More recently, unvegetated coastal sediments have also been recognized as a periodic source of fixed N .
Biological N fixation is catalyzed by the nitrogenase enzyme, which is highly conserved and distributed throughout diverse prokaryotic taxa . nifH, the gene encoding the nitrogenase iron protein subunit, is the gene marker largely used to study diazotroph diversity in various environments . Phylogenetic analysis of nifH DNA in the aforementioned benthic habitats, all revealed that anaerobes related to sulfate reducing bacteria were likely significant contributors of fixed N to the system (Lovell et al. 2000;. Microbes respiring iron and sulfur compounds, along with sulfate reducers dominated active nifH gene expression in unvegetated sediments in Narragansett Bay, a temperate estuary in southern New England Brown et al. submitted). Because of the abundance and activity of anaerobic diazotrophs recovered from unvegetated sediments in Narragansett Bay, we hypothesize that a reduction in oxygen supply to the overlying water and surface sediments during hypoxic and anoxic events will stimulate N fixation by these anaerobic organisms. Hypoxia may perturb the functioning of coastal ecosystems by potentially inhibiting denitrification while also stimulating anaerobic nitrogen fixation.
Exploring areas impacted by hypoxia may help resolve the global marine N budget discrepancy.
For the last several decades, episodic hypoxic events have been documented in Narragansett Bay Bergondo et al. 2005;Melrose et al. 2007;. Over the course of the 2013 summer season, water column oxygen concentrations, salinity and temperature were monitored at several sites in upper Narragansett Bay. Sediments were harvested at the end of May, prior to hypoxia and at the beginning of August, after a hypoxic event.
Sets of pre-hypoxic sediment cores collected at the end of May were incubated in oxic or hypoxic conditions. To investigate the source and processing of organic matter deposited to the benthos, we measured total organic carbon and nitrogen in the sediments. Analysis of other benthic characteristics including oxygen penetration and porewater nutrients can provide insight into the microbial processes occurring in the surface sediments. Dominant nifH-transcribing microbial populations were identified and quantitative PCR was used to follow changes in abundance and nifH expression of these bacterial groups in response to fluctuating bottom water oxygen concentrations.
The overall goal of this study was to determine (1) how hypoxia influences benthic microbial activities through analysis of sediment characteristics and (2) if changes in sediment characteristics reflects the activity of diazotrophs.

Results:
Water quality parameters, sediment characteristics and expression of nifH, the functional gene associated with nitrogen fixation, were analyzed during the 2013 summer season at three sites, Bullock Reach (BR), Sally Rock (SR) and Hope Island (HI) in Narragansett Bay, RI (Fig 1, Table 1). (Refer to Experimental Procedures for more in-depth collection and site description.)

Bottom Water Dissolved Oxygen
Sediment cores were harvested prior to hypoxia at the end of May when bottom water dissolved oxygen (DO) concentrations ranged from 7.39-9.09 mg/L (Fig   2, Table 2). Throughout much of July, sites BR and SR exhibited severe hypoxia (Fig   2). Near anoxic conditions (0.0 mg/L DO) were observed at site SR (Fig 2). Sediment cores were collected in early August after hypoxia (Fig 2). At site BR, the bottom water DO was 3.49 mg/L while the concentrations were a bit higher at sites SR and HI, 6.25 mg/L and 5.43 mg/L, respectively (Fig 2, Table 2).

Sediment Characteristics
Molar carbon to nitrogen (C/N) ratios of the sediment did not vary more than ±0.33 from the average along depth profiles or between months sampled at each site (Fig 3). The average C/N ratio was 11.4 and 11.3 at sites BR and HI, respectively, with a slightly lower average C/N ratio of 10.4 at site SR. Weight percent total organic carbon (TOC) and weight percent total nitrogen (TN) of the sediment were much lower at site HI, with an average of 1.91% TOC and 0.20% TN compared with sites BR (3.62% TOC and 0.37% TN) and SR (3.34% TOC and 0.38% TN) (Fig 4).
Sediment downcore oxygen profiles differed across sites and months sampled (Fig 5). At all three sites in May and August, DO was depleted in the sediment by 0.5 cm in depth (Fig 5). The oxygen profiles for sediment collected in May and August at site BR were very similar (Fig 5a). Conversely, at site SR the oxygen percent saturation in the surface sediment in August was much lower and depleted more quickly compared to sediment collected prior to hypoxia in May (Fig 5b). At site HI, the sediment oxygen profiles in May and August follow a similar trend with oxygen penetrating slightly deeper (~0.5 mm) into the sediment in May (Fig 5c).
Nitrite plus nitrate in the sediment porewater varied between sites and months sampled (Figs 6a, 7a and 8a). The highest levels of nitrite plus nitrate were detected in the porewater of site BR (Fig 6). At the sediment water interface in May, the porewater nitrite plus nitrate concentration ranged between an average 0.6 µM to 5.6 µM depending on the site (Figs 6a, 7a, and 8a). In August, the concentration of nitrite plus nitrate was nearly undetectable along the entire sediment depth profile of all three sites (Figs 6a, 7a and 8a). Sediments from sites BR and SR exhibited similar porewater ammonium concentration profiles (Figs 6c and 7c), while slightly lower concentrations of ammonium were detected at site HI (Fig 8c). Between May and August at each site, the concentration of ammonium remained relatively stable (Fig   6c, 7c and 8c).

Bottom Water Dissolved Oxygen
Sets of pre-hypoxia cores collected in May were incubated under oxic and hypoxic conditions. The overlying bottom water DO concentrations of the oxic treatment cores were between 8.46-8.83 mg/L for all three sites (Fig 2, Table 2). For the hypoxic incubated cores, the DO concentrations of the water at the sediment water interface ranged from 1.5-3.3 mg/L (Fig 2, Table 2).

Sediment Characteristics
Sediment C/N ratios and weight percent TOC and TN of the incubated cores did not vary from the field cores with the exception of site SR. A slightly lower average C/N ratio of 10.1 at site SR was observed in the surface sediments (0-2 cm) of the oxic and hypoxic treatment cores (Fig 3). Also, an increase in weight percent TOC and TN (3.51% and 0.41%, respectively) was detected in the surface sediments of site SR incubated cores compared sediment collected in May (Fig 4).
Downcore oxygen profiles of the incubated sediment cores varied depending on oxygen treatment. In all cores incubated under oxic conditions, the oxygen saturation at the sediment-water interface increased to ~100% and oxygen penetrated deeper into the sediments compared to the control cores collected in May (Fig 5). At site BR, the oxygen decreased more rapidly in the hypoxic incubated cores compared to sediment collected in May and August (Fig 5a). At site SR, similar oxygen profiles were observed in hypoxic treated sediment and sediment collected in August after hypoxia (Fig 5b). The oxygen profiles for sediment collected in May, August and exposed to hypoxic conditions were very similar for site HI (Fig 5c).
Sediment porewater nitrite plus nitrate from sites BR and SR differed between the incubated cores and sediments collected in May (Fig 6a,b and 7a,b). For site HI, little variation in nitrite plus nitrate concentrations was detected between May, oxic treatment and hypoxic treatment cores (Fig 8a,b). At both sites BR and SR, a spike in nitrite plus nitrate (up to 22 µM and 18 µM, respectively) was detected in the surface sediments of one set of oxic treatment cores (Figs 6b and 7b). At all three sites, porewater ammonium concentrations increased in both the oxic and hypoxic incubated cores compared to the May control cores (Figs 6d, 7d and 8d). The overall highest ammonium concentrations were detected in the hypoxic treatment cores (Fig 6d, 7d and 8d).

Phylogenetic Relationships of Expressed nifH Sequences
Expression of the nifH gene was observed at all three sites and was localized to the top 3 cm of sediment (Fig 9). The spatial distribution of nifH mRNA transcripts was variable between months sampled and treatments (Fig 9). The greatest nifH bulk expression was detected at site SR, particularly in the hypoxic treatment sediment (Fig   9).
Phylogenetic analysis of expressed nifH mRNA transcript sequences from the sediments revealed that they are limited to nifH Clusters I and III, as previously defined . Many expressed nifH sequences group with known sulfate, sulfur and iron reducing bacteria (Fig 10). Several of these phylogenetic groups (NB2, NB3, NB5, NB7 and NB10) have been previously reported from unvegetated sediments in the same estuary Brown et al. submitted). Three novel groups (NB12-NB14) were recovered in this study. Group NB3 contains the majority of expressed nifH sequences (41) and is most closely related to a sulfur and iron reducing anaerobe, Pelobacter carbinolicus (Lovley et al. 1995) (Fig 10). The second largest group of expressed nifH sequences (13), group NB7 also contains the sulfate reducers Desulfovibrio salexigens and Desulfovibrio vulgaris (Fig 10). At site HI, only groups NB3 and NB7 nifH mRNA transcripts were detected in the sediments (Fig 11). A higher diversity of microbes expressing nifH was identified at sites BR and SR (Fig 11).

Targeting Specific Diazotroph Groups with QPCR
Quantitative PCR was used to follow changes in abundance and levels of nifH expression of the two dominant microbial groups expressing nifH, groups NB3 and NB7 (Fig 10). The greatest abundance of group NB3 was observed at site HI at 1 cm in depth in the hypoxic treatment sediments (Fig 12c). The distribution of group NB3 was similar at sites BR and SR, with a slight increase in abundance in the surface sediments of the oxic and hypoxic treatment cores at site SR (Fig 12a,b). Similar levels of NB3 nifH expression were detected at sites BR and SR (Fig 12d,e). Little to no nifH expression was observed between 3-5 cm in depth at site HI (Fig 12f). The overall abundance of group NB7 was greater than NB3, however, a difference between sites was not observed (Figs 12a-c and 13a-c). The lowest NB7 abundance was detected along the entire depth profile at site SR in May prior to hypoxia (Fig   13b). The expression of nifH by group NB7 followed a similar distribution to group NB3 (Fig 12d-e and 13d-e). No significant difference in NB3 and NB7 abundance and nifH expression was observed between months sampled and treatment.

Discussion:
Summer hypoxic events have been the subject of increasing concern in Narragansett Bay. Episodic hypoxia has been recorded in the bay for the last several decades Bergondo et al. 2005;Melrose et al. 2007;, and the summer of this study was no different. Severe hypoxia was documented for several weeks during July in Providence River Estuary and Greenwich Bay, locations of site BR and SR, respectively (Fig 2) (BART 2013;NBFSMN 2013). Additionally, near-hypoxic conditions reached as far south as site HI for part of July (Fig 2) (NBFSMN 2013).
Interestingly, high rates of net nitrogen fixation were measured (up to -650 µmol N 2 -N m -2 h -1 ) at several sites in upper Narragansett Bay in 2006  during the same summer months that these regions were severely impacted by widespread hypoxia ). We present sediment TOC, TN, oxygen and porewater nutrient profiles with concurrent gene expression data from in situ and incubated sediments under varying water column oxygen conditions. After severe hypoxia in upper Narragansett Bay, the concentration of porewater nitrite plus nitrate has disappeared in the sediment. In oxic incubated cores, an increase in nitrite plus nitrate was observed in the surface sediments, while an accumulation of ammonium was detected in cores exposed to hypoxic conditions. Hypoxic and anoxic conditions in the water column may in fact be driving a change in N cycling in coastal marine sediments.

Organic Matter Sources and Processing
The ratio of C/N can be a useful indicator of the source of organic matter preserved in sediments. Organic matter produced by marine algae and phytoplankton have C/N ratios of 6-9 (Bordovsky 1965), whereas terrestrial organic matter tends to have C/N ratios greater than 20 (Meyers et al. 2001). The C/N ratio of estuarine sediments, including sites BR, SR and HI, reflects a mixture of both marine and terrestrial sources. Site SR has a slightly lower C/N ratio compared to the other sites, indicating that marine algae may be a slightly more important source of organic matter to those sediments.
Measurements of total organic carbon (TOC) and total nitrogen (TN) can also provide evidence for the long-term impacts of hypoxia. The concentration of total C and N are indicators of primary productivity and represent the proportion of organic matter that remains after sedimentary remineralization. Increased TOC and TN concentrations have been attributed to several factors including, higher water column productivity, greater terrestrial organic matter inputs or increased preservation in sediments due to hypoxic conditions (Gooday et al. 2009). In Chesapeake Bay sediments, several studies observed increases in percent TOC and TN corresponding to both higher primary productivity and periods of hypoxia (Cooper et al. 1991;Bratton et al. 2003). Higher concentrations of TOC and TN detected in sediments from sites BR and SR than in site HI sediments correlates with higher levels of primary productivity ) and more severe seasonal hypoxia (Melrose et al. 2007;) observed at sites BR and SR in comparison to site HI. Site HI may have lower concentrations of TOC and TN due to differences in sediment composition (King et al. 2008). We did not observe a shift in C/N ratio or concentrations of total C and N in the sediments at any of our sites during the course of the sampling, including sites SR and BR that experienced severe hypoxia. It may be that one summer of sampling was not enough to detect a shift in the C/N ratio. However, differences in total C and total N concentrations between sites suggests that long-term hypoxia may impact organic matter preserved in the sediment.

Hypoxic Conditions Alter Sediment Porewater Nutrient Profiles
Redox dependent nitrogen transformations in the sediment are potentially influenced by low oxygen events occurring in the water column. Monitoring concentrations of porewater oxygenated nitrogen species (nitrite plus nitrate) and reduced species (ammonium) can provide insight into microbial N cycling activities occurring in the sediment. In comparison to sediment collected in May, concentrations of nitrite plus nitrate disappear to nearly undetectable levels throughout the entire sediment depth profile after hypoxia in early August. The depletion of nitrite plus nitrate suggests nitrification has been repressed. Hydrogen sulfide, a respiratory poison (Wang et al. 1999) has been shown to specifically inhibit nitrification (Joye et al. 1995). Hypoxic waters in Chesapeake Bay have stimulated sulfate reduction and therefore heightened hydrogen sulfide levels (Kemp et al. 1990). Increased concentrations of hydrogen sulfide and low oxygen conditions associated with hypoxia are likely limiting nitrification.
Conversely, in the oxic incubated cores from sites BR and SR, increased oxygen penetration in the surface sediments appears to have stimulated nitrification activity as increased concentrations of nitrite plus nitrate were observed at the sediment water interface. Similar findings were detected in a Denmark estuary in which changes in oxygen penetration depth fueled nitrification (Rysgaard et al. 1995).
Unlike the sediments collected after hypoxia in August, we did not observe depletion in nitrite plus nitrate in the hypoxic treatment cores. Perhaps in this short incubation, the low oxygen concentrations repress nitrification but the hypoxic incubation was not long enough to allow denitrification to completely reduce the stock of oxidized forms of nitrogen available in the sediment.
High concentrations of ammonium were observed in all sediment depth profiles, and fall within concentrations reported for other estuarine sediments Rao et al. 2011). Higher porewater ammonium concentrations were detected at sites BR and SR in comparison with site HI, possibly correlating with strong down bay gradients in water column inorganic nutrients and primary production (e.g (Oviatt 1980;).
An accumulation of porewater ammonium was observed in both oxic and hypoxic treatment cores at all three sites, with the highest concentrations detected in sediment incubated under hypoxic conditions. An increase in porewater ammonium was not detected in the in situ cores. The dissimilatory nitrate reduction to ammonium (DNRA) pathway retains available N in the system in the form of ammonium. In estuarine sediments, several studies have proposed that under low oxygen conditions, DNRA is favored over denitrification (Tiedje et al. 1983;Rysgaard et al. 1996) resulting in an increase in ammonium, supporting our findings in the hypoxic incubated sediments. However, we also observed an increase in porewater ammonium in the oxic treatment cores, but generally the highest concentration was detected below oxygen penetration (0.5 cm). Perhaps an increase in nitrate due to oxygen-stimulated nitrification in the oxic cores, promoted competition for nitrate by microbes involved in the denitrification and DNRA pathways resulting in an increase in ammonium compared to the May control cores.
Low oxygen concentrations in overlying bottom waters appear to, at least in part, influence sediment N cycling. For example, oxygen depletion and increased sulfate reduction associated with hypoxic conditions potentially inhibit nitrification, which has likely consequences for decreasing subsequent rates of denitrification. The repression of coupled nitrification-denitrification may open a niche for other N cycling microbes to thrive and become active.

Sulfur and Sulfate Reducers are the Dominate Diazotrophs in Benthic Habitats
Due to the high input of N into coastal systems ) and N fixation being repressed by combined N , benthic sediments were not consider a source of fixed N. However, in some marine environments, N fixation may not be as sensitive to dissolved inorganic N, as previously believed (Knapp 2012 Bertics et al. 2012), indicating these potentially active diazotrophs are widely distributed in coastal marine sediments.
Various sulfur and sulfate reducing bacteria are genetically capable of fixing N  and have been shown to do so in culture , suggesting these microbes may be potential sources of fixed N in benthic sediments. Phylogenetic analysis of the nifH gene in benthic habitats including sediments vegetated by macrophytes, microbial mats and unvegetated sediments have identified sulfate reducers as being the dominant N fixer (Lovell et al. 2000;Brown et al. submitted). Based on acetylene reduction and sulfate respiration inhibition assays, several studies have confirmed that sulfate reducers are responsible for the N fixation rates detected in these benthic habitats .

Steady Abundance and nifH expression of the Dominant Diazotrophic Groups
We hypothesized low oxygen conditions associated with hypoxia would stimulate N fixation in the benthic sediments and we would detect this potential switch in N cycling with an increase in NB3 and NB7 nifH expression under low DO levels.
However, the abundance and nifH expression of groups NB3 and NB7 did not appear to respond to changing bottom water oxygen concentrations. At each site, a steady abundance and nifH expression of both microbial groups was observed along the entire depth profile, with slight variation. Perhaps we did not capture an increase in nifH expression by these microbial groups because bottom water oxygen concentrations were too high. All sediments analyzed had an overlying DO concentration of >2 mg/L. The acetylene reduction assay was used to determine the rates of N fixation at the sites in this study as well as Greenwich Cove, near a wastewater treatment plant. Higher rates of N fixation were measured at Greenwich Cove with DO concentrations of <1 mg/L compared to sediments from sites BR, SR, and HI with DO concentrations ranging from 2-9 mg/L (Rodrigue Spinette, unpublished). Based on these results, our assumption seems likely. If we sampled during anoxia in July or incubated the cores longer under lower oxygen conditions, we may have detected a greater change in nifH expression.
Alternatively, perhaps we are not following the most low oxygen responsive nifH expressing microbial group. The two phylogenetic groups (NB3 and NB7) targeted with qPCR are the most abundant and active nifH expressers under low oxygen (2-4 mg/L DO) and oxygenated conditions in the sediment. By cloning nifH transcripts in the surface of sediments exposed to near-anoxic conditions, we may identify novel groups with increased nifH expression under oxygen depletion.

Conclusions
Hypoxic regions are expanding and predicted to increase in the near future due to human activities and climate change ). However, little is known about how low oxygen waters influences nitrogen cycling dynamics in coastal marine sediments. The global N cycle is driven by microbial metabolism and oxygen depletion has the potential to perturb N transformations. In Narragansett Bay sediments, microbes expressing nifH are related to sulfur/iron and sulfate reducers.
Close relatives of these microbes have been shown to actively fix N in other benthic habitats . Episodic hypoxic events in the bay not only may stimulate N fixing activity by these anaerobes, but if low oxygen conditions persistent over time, may also select for these diazotrophic microbial communities. An increase in N fixation in the sediments could establish a positive feedback loop, exacerbating hypoxic conditions. Understanding the impacts of low oxygen on N transformations in coastal sediments is crucial for predicting how these expanding hypoxic events will influence the marine N cycle globally. buoys which record continuous measurements of dissolved oxygen (DO), chlorophyll, temperature and salinity (Fig 1, Table 1).
At each site, a YSI 6920 V2 was used to measure DO, temperature and salinity of the entire water column (Table 1). SCUBA divers harvested intact sediment cores (10 cm inner diameter and 30.5 cm long). In May, six cores were collected from each site. Two cores were sacrificed on day of collection and referred to as pre-hypoxic May cores. The four remaining cores were stored in the dark in an incubator at the University of Rhode Island at average in situ bottom water temperature (17°C). Two cores were capped to mimic hypoxic conditions (referred to as hypoxic treatment) and two cores were left uncapped with air gently bubbling through the overlying water (referred to as oxic treatment). Capped cores were monitored and when the DO dropped below ~3 mg/L of oxygen (7 days after collection), all four cores were sacrificed. In August, two cores were collected from each site and were sacrificed on day of collection and referred to as August control cores. Before sediment cores were sectioned, triplicate profiles of DO in the sediment surface of one core per treatment were measured with Unisense's oxygen microsensor. Calibration of the instrument and measurements were performed based on the manufacturer's protocol. Triplicate measurements were averaged for the sediment oxygen penetration profiles. Oxygen percent saturation was measured in 250 µm intervals for the May, oxic and hypoxic cores and 500 µm intervals for the August cores.

Sediment Core Sectioning and Sub-sampling
Sediment cores were sectioned in 0.5 cm increments for the top 2 cm and then in 1 cm increments to 5 cm in depth. For each section, duplicate 0.25 mL and 0.5 mL samples were flash frozen for DNA and RNA isolation, respectively. A ~5 mL subsample was saved for carbon and nitrogen analysis. The remaining sediment was transferred into a 50 mL conical tube and centrifuged at 4,000 rpm for 5 min. The overlying water was saved for porewater analysis.

Carbon and Nitrogen Analysis
Samples were analyzed in 1cm increments, so sediment from the surface water interface to 1 cm in depth were pooled and sediment from 1 cm to 2 cm in depth were pooled. Samples were dried at 60°C for 18-24 hrs. The sediment was then homogenized using a clean mortar and pestle. ~10-15 mg of sediment was weighed into an ultra-clean tin capsule on a microbalance and then placed into a nickel sleeve.
Sediment samples and acetanilide standards were analyzed with a Carlo Erba EA1108 CHN analyzer.

Porewater Nitrite plus Nitrate
Porewater nitrite and nitrate samples were reduced by a vanadium (III) solution to nitric oxide for chemiluminescence detection with a NO x box as previously described (Hendrix et al. 1995). Standards were prepared from dilutions of potassium nitrate.

Porewater Ammonium
Porewater ammonium concentrations were measured colormetrically using a µL of 2 µM outer reverse primer for our gene of interest, nifH3. After the reverse transcriptase was added, the mixture was incubated at 50°C for 50 min. All the other steps followed the instructions of the manufacturer. For every sample, we also included controls that did not contain reverse transcriptase to confirm there was no DNA contamination in the subsequent PCR amplification.

Functional Gene Sequence Analysis
The nifH gene from environmental cDNA was isolated using nested PCR with degenerate outer primers nifH4-nifH3 and inner primers nifH1-nifH2 . Both rounds of PCR consisted of an initial denaturation step of 2 min at 94°C, cycling steps that included: a denaturation step of 30 s at 94°C, an annealing step of 30 s at 50°C, and an extension step of 1 min at 72°C. All reactions had a final extension step of 7 min at 72°C. First round reactions had 25 cycles and the second round reactions had 30 cycles    Blank spaces indicate that no gene expression was detected. Sampling month or treatment denoted above each profile (A= May, B= August, C= oxic treatment and D= hypoxic treatment).  . Groups NB2, NB3, NB5, NB7 and NB10 were previously described (Fulweiler et al. 2013 and Brown et al. submitted), while groups NB12-14 are novel to this study. The number inside the group indicates the total number of sequences within the grouping. Bootstrap values (1,000 replicates) > 50% are shown at the respective nodes. Asterisk indicates groups targeted for quantitative PCR. Figure 11: Percent of total expressed nifH sequences per site. Each color represents a cultivated species our environmental expressed sequences are related to as depicted in the nifH maximum likelihood tree (Fig. 3). Organisms listed in parentheses are contained within the grouping.

CONCLUSION
Marine nitrogen cycling is complex and the transformations between different chemical forms of N are driven by extremely diverse microbial communities . Functional genes encoding cellular proteins that mediate these biogeochemical processes can provide insight into the functioning of an ecosystem.
Additionally, phylogenetic analysis of these genes can be used to investigate the diversity of specific groups of microorganisms (eg. denitrifiers and nitrogen fixers) in the environment, however the analysis does not provide information on what fraction of these microbes are metabolically active. This dissertation aimed to use a gene expression approach to identify and follow the likely active microbes responsible for the removal and input of N to coastal marine sediments under different environmental conditions.
During the summer of 2006, benthic sediments in Narragansett Bay exhibited a seasonal switch in N cycling with high rates of net N fixation ).
We proposed that timing of organic matter deposition to the benthos might be important in determining which process (denitrification vs. nitrogen fixation) dominates ). Using mesocosms with sediment from the same site, we analyzed expression of functional genes associated with nitrogen fixation (nifH) and denitrification (nirS) in response to varying the timing of organic matter deposition. We discovered that both processes were occurring simultaneously . Phylogenetic evidence suggested that sulfur/iron respiring microbes and sulfate reducers were responsible for the nitrogen fixation activity . These findings led us to investigate the spatial extent of heterotrophic sediment N fixation in temperate regions.
Expression of nifH and nirS were analyzed along the estuarine gradient of Narragansett Bay to an offshore continental shelf site over a temporal cycle. The biodiversity of N fixers expressing nifH decreased along the gradient of the bay. The freshwater head of the bay experiences fluctuating conditions (e.g. oxygen, temperature, salinity and nutrients) , whereas the offshore sites remain relatively stable throughout the year ). Exposure to varying environmental factors may be driving diversity in the upper bay sediments as microbes are competing to adapt to a continuously changing ecosystem, while at the more stable offshore sites, the benthos is dominated by the more competitive microbes. Again, the dominant likely active diazotrophs in upper bay sediments were related to anaerobic sulfur/iron and sulfate reducing bacteria. Warming seawater temperatures, low oxygen events and high organic carbon content in the sediment appear to stimulate growth and activity of these microbial groups as the highest abundance and nifH expression was detected in the upper bay sediments. Conversely, at the offshore sites, nifH expression was dominated by a unicellular cyanobacterium, Canditatus Atelocyanobacterium thalassa (UCYN-A) .
The genome of UCYN-A revealed that this cyanobacterium cannot fix carbon, but does have the capability to fix N . In fact, UCYN-A lacks critical metabolic pathways including photosystem II, RuBisCo, and the tricarboxylic acid cycle, yet contains the complete suite of nitrogenase genes ).
UCYN-A was recently recognized as being a symbiont of carbon fixing eukaryotic hosts and was reported to have a loose symbiotic association with a unicellular prymnesiophyte ). In return for fixed carbon, UCYN-A provides the host with fixed N . Recovering nifH mRNA transcripts related to UCYN-A in temperate coastal sediments was surprising because to date this microbe has only been documented to be an important N fixer in tropical and subtropical oligotrophic oceans Church et al. 2005;. We detected the highest UCYN-A abundance and nifH expression in sediments when bottom water temperatures were 5-8 ºC suggesting this microbe has a broader thermal tolerance and may be more widespread than previously believed. This microbe may also have a wider range of host associations than currently understood. Presumably, when the prymnesiophte partner sinks to the benthos, loosely attached UCYN-A gets exported to the sediments as well. UCYN-A's genome suggests that the microbe is physiologically capable of thriving in the sediments after the lifespan of its host. In fact, we detected mRNA transcripts from UCYN-A as deep as 6 cm, indicating UCYN-A remains active in sediments long after burial of any eukyarotic hosts it may be associated with. These findings from Chapter 2 have implications for benthic-pelagic coupling and N cycling in continental shelf sediments.
The results of Chapter 1, demonstrating that bacteria that live anaerobically are the major groups of active N fixers in estuarine sediments, led us to further investigate the influence of hypoxia on microbial activities in estuarine sediments. Hypoxia is increasingly becoming a common occurrence in Narragansett Bay ), yet little is known about how oxygen depletion in the water column impacts N cycling in the sediments. Based on analysis of porewater nutrients, it appears that an increase in oxygen in surface sediments stimulates nitrification at some sites, while hypoxic conditions repress nitrification and perhaps stimulate nitrogen fixation. The greatest bulk nifH expression was detected at sites with the highest organic carbon content in the sediment and most heavily impacted by hypoxia. Similarly to Chapter 1, the dominant diazotrophs expressing nifH were related to iron/sulfur and sulfate reducers.
The abundance and nifH expression of the dominant N fixing groups did not vary in response to fluctuating oxygen concentrations; however, at the times we sampled, the dissolved oxygen was near the EPA's hypoxia threshold of 2.3 mg/L (USEPA 2000)during sediment collection. Perhaps conditions closer to complete anoxia are needed to stimulate nitrogenase activity in these diazotrophic communities.
In conclusion, this thesis has provided evidence that anaerobes related to iron/sulfur respiring bacteria and sulfate reducers can be a source of fixed N in benthic coastal sediments. We propose that increased water temperatures, hypoxia and sediment organic carbon concentrations are key drivers promoting their activity.
Although these microbes are highly abundant in marine sediments, their contribution to the N budget was considered negligible. Therefore, little is known regarding controls on N fixation activity in these anaerobic bacteria. To better predict how an ecosystem, like Narragansett Bay, may respond to future environmental changes, regulating factors including sensitivity to combined N and oxygen tolerance need to be evaluated in these microbes, both in the laboratory and in the environment.
Determining how microbial community composition and activity respond to different