4.1. Contributions of Benthic Foraminifera to Sedimentary Nitrate Storage
 Nitrate concentrations (12–217 mM) in all four foraminiferal taxa analyzed here were elevated relative to bottom water, although not all specimens of any species had high nitrate concentrations and the data for all species varied considerably. This latter observation is consistent with earlier publications regarding nitrate content and its variability in foraminifera [Høgslund et al., 2008; Koho et al., 2011; Piña-Ochoa et al., 2010; Risgaard-Petersen et al., 2006]. Intracellular concentrations of nitrate in our foraminifera were comparable to those reported by Risgaard-Petersen et al.  (35–180 mM), even though the species analyzed were different, aside from N. stella. Reports of nitrate concentrations in other Bolivina species average 153 ± 49 mM [Piña-Ochoa et al., 2010], which are slightly lower than nitrate concentrations in our B. argentea (195 ± 160 mM). Although Buliminella was not analyzed by other authors, species of the closely related genus Bulimina is documented to have nitrate concentrations of 116 ± 43 mM [Piña-Ochoa et al., 2010], compared to our Buliminella tenuata, which had 217 ± 150 mM nitrate. The B. tenuata concentrations were the highest we obtained, but these were lower than the highest values reported by Piña-Ochoa et al.  for a species from the same superfamily (Globobulimina cf. ovula, 375 ± 174 mM).
 The SBB sediments are known to have high subsurface nitrate + nitrite peaks [Bernhard and Reimers, 1991; Reimers et al., 1996] but, at the time of those reports, the source of this peak was unclear. In the past, it was thought that the peak could be due to the large populations of the sulfur-oxidizing denitrifying bacteriumBeggiatoa, which can contain high concentrations of nitrate [McHatton et al., 1996] that could have been released during pore water sampling. Our present data suggest that nitrate storage in foraminiferal vacuoles may have also contributed to the peak in SBB sediment nitrate + nitrite reported by Bernhard and Reimers . However, a more quantitative assessment of foraminiferal abundances and distributions, with respect to sediment porosity, are required to answer this question completely.
4.2. Bacterial Associations With Benthic Foraminifera
 High intracellular nitrate contents suggest that these SBB foraminifera are either intentionally transporting nitrate inside their cells or producing it intracellularly. Because these organisms inhabit oxygen-deficient sediments, and nitrate production by nitrification requires oxygen, we discount intracellular production in our populations. Furthermore, our data fromB. tenuata did not include any known nitrifying genera among the 16S rRNA genes obtained. While future studies may more directly address the question of nitrifying symbionts through functional gene (ammonia monooxygenase) analysis, it would be difficult to sustain high rates of nitrification in sulfidic sediments, and we find the most plausible explanation to be active transport of NO3− into the foraminifer from surrounding pore waters. Molecular characterization and/or localization of nitrate transport proteins in these foraminifera may help shed light on this issue.
 Most of the small subunit rRNA sequences obtained from endobionts of B. tenuatafrom oxygen-depleted sites most closely affiliate with relatives ofMarinobacter, which is a denitrifying rod-shaped bacterial genus within the gammaproteobacteria found commonly in marine sediments [e.g.,Nakano et al., 2010]. Thus, these sequencing results are consistent with morphologic results obtained with TEM (i.e., rod-shaped bacteria;Figure 1h) [see also Bernhard et al., 2000; Buck and Bernhard, 2001]. There are many representatives of denitrifying bacteria within the gammaproteobacteria, and most denitrifiers are facultative, and can grow by aerobic respiration in the presence of oxygen [Kuenen et al., 1991]. Some sulfide-oxidizing bacteria will also denitrify [Kuenen et al., 1991], an ability that may serve an additional role in detoxification for foraminiferal hosts in anoxic sulfidic sediments. A nirK gene was detected in DNA extracts from B. tenuatafrom oxygen-depleted sediments, but it has not yet been localized to the endobionts, attached bacteria, or the foraminifer host. The bacteria associated withB. tenuata from more aerated sediments most closely affiliate with a clade of gammaproteobacterial sequences that include 16S rRNA gene sequences from uncultured bacteria and some Enterobactriales. The localization and identity of the gammaproteobacterial associates of the B. tenuata from different habitats needs to be confirmed by positive FISH using probes specific to each sequence type, but symbiosis plasticity is not too surprising given some B. tenuata from other habitats lack endobionts altogether [Bernhard et al., 2001]. In this species, the potential for and mediator of denitrification remain equivocal.
 The other species investigated in this study contained no endobionts, although N. stella is known to sequester plastids that contain nitrate reductase [Grzymski et al., 2002]. The fact that a gene encoding dissimilatory nitrite reductase was obtained from each of the foraminiferal species analyzed indicates that these eukaryotes and/or their external bacterial community can perform a critical step of the denitrification process. We did not survey for genes involved in the remainder of the denitrification process (i.e., norB and norQ for nitric oxide reduction and nosZ for nitrous oxide reduction); those analyses remain for future investigations.
4.3. Contributions of Benthic Foraminifera to Sedimentary Denitrification
 Interestingly, the negligible amounts of nitrite present in our foraminifera, as evidenced by differences between paired sulfamic and non-sulfamic-treated samples, suggests that the turnover of nitrite in the SBB foraminifera must be tightly coupled between nitrate reduction and nitrite reduction. Furthermore, our measured denitrification rates (1976 ± 1103 pmol/B. argentea/day; 1386 ± 320 pmol/F. cornuta/day), while containing a high degree of uncertainty, are higher than those reported for other species (88 pmol/N. stella/day and 565 pmol/G. pseudospinescens/day; [Høgslund et al., 2008; Risgaard-Petersen et al., 2006]). While nitrate contents were not determined for the incubated specimens, average nitrate contents for these species indicate nitrate turnover times of 0.5 to 0.7 days, which is similar to the estimate of 2.5 days for N. stella [Høgslund et al., 2008] and much faster than the estimate of over a month for G. pseudospinescens [Risgaard-Petersen et al., 2006]. Our data indicate that, when these SBB foraminiferal populations are actively denitrifying, they need to replace their cellular quotient of nitrate at least once every day. These rates are supported by the observation that intracellular nitrate contents in B. argenteacells dropped dramatically over a 7-day incubation under anoxic conditions (Figure 6a).
 SBB sedimentary denitrification rates have been estimated to be quite high (∼4.5 mM N m−2 day−1) relative to surrounding basins, although the cause of such high rates is unclear [Sigman et al., 2003]. A typical density of living (as determined by adenosine triphosphate analysis) benthic foraminifera in SBB surface sediments is ∼200 specimens/cm3 [Bernhard and Reimers, 1991]. Using our denitrification rates for B. argentea and F. cornuta (∼1500 pmol/cell/day), foraminifera could account for ∼3 mM N m−2 day−1, or 67% of the estimated sedimentary denitrification [Sigman et al., 2003]. Other cases where benthic foraminifera account for two thirds of sedimentary denitrification are known [Piña-Ochoa et al., 2010], while considerably lower proportions have also been reported [Glud et al., 2009].
 The δ15N of the foraminiferal nitrate provides additional insights about nitrogen cycling in these protists. To begin with, when sufficient nitrate was present for analysis, the observed δ15NNO3 values for fresh specimens were generally elevated, as were those reported for another SBB foraminifer (∼18‰) [Bernhard et al., 2012]. Intracellular δ15NNO3values in endobiont-freeB. argentea and N. stella were often higher than the δ15NNO3 values of pore waters bathing the foraminifera, suggesting that the foraminifera may have actively consumed nitrate in their natural environment. N. stella has previously been shown to denitrify [Risgaard-Petersen et al., 2006], while B. argentea had not. Foraminiferal δ15NNO3values that lie between pore water and overlying-waterδ15NNO3 (F. cornuta, B. tenuata, and some B. argentea) suggest that these specimens may have been exposed to overlying water in the surface sediments, that our pore water sampling scale was too coarse to resolve δ15NNO3 gradients at microhabitat scales relevant to these microbes, and/or that these species or individuals had lower nitrate reducing activity.
 In our experiments with B. argentea, the magnitude of decrease in nitrate concentrations in individuals from the aerated experimental treatments was slightly less than the decrease in nitrate concentrations from specimens from either of the oxygen-depleted treatments. Although the difference in the mean nitrate concentrations was not significant, there were more individuals with measurable nitrate concentrations in aerated experiments, and the range of nitrate concentrations was greater in cells from oxygenated incubations. These results suggest that the presence of oxygen may have partially inhibited or delayed the onset of nitrate respiration. This species inhabits a more oxygenated habitat and is thought to be less tolerant of low oxygen than other species occurring in the vicinity of SBB [Bernhard et al., 1997; Stott et al., 2002], but elevated δ15NNO3 values in pore waters clearly indicate that these sediments support denitrification. Prior research has shown that benthic foraminifera appear to be highly plastic with regards to metabolism, apparently having the ability to switch from aerobic respiration to denitrification as necessary [e.g., Piña-Ochoa et al., 2010]. Indeed, certain benthic foraminifera have been known to survive considerable periods of anoxia [e.g., Bernhard, 1993; Bernhard and Reimers, 1991; Moodley and Hess, 1992; Moodley et al., 1998] and some must have an as-yet-unknown alternative oxidative pathway, as evidenced by experiments done with electron transport inhibitors [Travis and Bowser, 1986].
 As observed for matched field specimens and pore water data, the disappearance of nitrate associated with increases in δ15NNO3 and δ18ONO3 in our experiments provides evidence for reductive nitrate removal, rather than nitrate leakage from the foraminifera. In cases where nitrate disappearance was associated with no change, or even a decrease, in δ15NNO3 and δ18ONO3, the fate of intracellular nitrate is uncertain. Specimens may have died yet not fully lysed, thereby continuing to have concentrated nitrate in their cytoplasm. Dormancy may also have occurred [e.g., Alve and Goldstein, 2010; Bernhard and Alve, 1996]. The ratio of increases in δ18ONO3 to δ15NNO3 (0.83–0.85) was lower than expected for most bacterial denitrifiers (slope of 1) [Granger et al., 2008]. This may be indicative of activity from a periplasmic nitrate reductase, rather than a more typical respiratory nitrate reductase. This possibility will have to be pursued in future studies.
 The lack of a significant difference in nitrate removal in the presence and absence of the antibiotic cocktail during our second experiment with B. argentea is suggestive of nitrate reduction by foraminifera themselves. It is also possible that a subset of bacteria was not affected, or was incompletely inhibited, by the cocktail composed of five antibiotics, and thus, could continue to denitrify throughout the experiment. For example, chloramphenicol inhibits RNA transcription [Jardetzky, 1963; Wolfe and Hahn, 1965] but would not inactivate existing enzymes. However, the combination of chloramphenicol, streptomycin, neomycin, tetracycline and penicillin is expected to inhibit most if not all prokaryotic activity. Given that the foraminiferal species used in both experiments (B. argentea) lacks symbionts, if bacterial denitrification was occurring, the bacteria would have to be contaminants either on the test exterior or as parasites in foraminiferal endoplasm. Bacterial parasites have recently been documented in foraminifera from deep-sea sediments in and adjacent to hydrocarbon seeps [Bernhard et al., 2010]. However, it is estimated that 6,000–23,000 denitrifying bacteria per foraminifera would be required to account for denitrification rates reported for foraminifera in Risgaard-Petersen et al. ; this magnitude of cytoplasmic parasites have never been observed in our electron microscopic imagery of B. argentea although serial sections including entire specimens have not received dedicated analysis in this context. To resolve this issue of whether bacteria or the foraminifer itself is responsible for denitrification, methods such as geneFISH [Moraru et al., 2010], which localizes the sites of denitrification genes to the eukaryote, symbionts and/or contaminants, will need to be employed, as has been done for an allogromiid foraminifer [Bernhard et al., 2012].
4.4. Synthesis and Evolutionary Implications
 Results of this study and Bernhard et al.  suggest that foraminifera have evolved at least two ways to perform the process of denitrification: (1) with symbionts and (2) by the eukaryote. The allogromiid foraminifer studied by Bernhard et al.  appears to host denitrifying symbionts. B. tenuata also contains symbionts, and although their identity and physiological capacity remain to be determined, initial data do not exclude denitrifiers. If the nirK gene obtained from B. tenuata DNA extracts can be localized to the endobionts, then this system may function similarly to the allogromiid. The other foraminifera studied here do not contain endobionts (N. stella husbands plastids), although they may have external bacterial associates. B. argentea was shown to produce N2 from NO3− and rapidly consume intracellular nitrate during both oxic and anoxic incubations. Isotopic fractionation during nitrate consumption indicates that it is being reduced intracellularly, rather than simply leaking out of the cell or being consumed by extracellular contaminants. A gene for nitrite reduction (nirS) was also detected in DNA extracts from B. argentea although it has not yet been localized. Incubations of F. cornuta were also found to produce N2 from NO3−. Further investigation is needed to determine whether this foraminifer is catalyzing the reaction itself, but its intracellular δ15NNO3 values were occasionally elevated relative to pore water nitrate, and no endobionts are known to exist in this species. N. stella has previously been shown to sequester chloroplasts containing nitrate reductase [Grzymski et al., 2002] and has previously been shown to denitrify [Piña-Ochoa et al., 2010; Risgaard-Petersen et al., 2006], presumably without endobionts. N. stella's elevated δ15NNO3 values suggest that this organism does participate in nitrate reduction in the field.
 It will be fascinating to determine if additional protists such as testate amoebae, ciliates, and/or flagellates can perform complete denitrification and, if they can, determine the mechanisms employed to do so. From a paleontologic perspective, both ciliates and flagellates likely evolved approximately 1.9 Billion years ago [Hedges et al., 2004; Wright and Lynn, 1997], a time in the Paleoproterozoic when a “whiff” of oxygen entered the atmosphere [e.g., Poulton et al., 2004]. However, at that time, atmospheric and oceanic oxygenation likely did not last long [Frei et al., 2009], so if these early evolving protists were to survive, an anaerobic metabolic pathway(s) was required. Indeed, it is argued that much of the Proterozoic, during and after the origin of ciliates and flagellates, was in a “nitrogen crisis” where oceanic oxygen concentrations remained very low to undetectable while denitrification was prevalent [Falkowski and Godfrey, 2008; Fennel et al., 2005]. Although nitrate may have been limited at this time [Falkowski and Godfrey, 2008], the ability to denitrify could have imparted a major ecological advantage to allow success of early evolving eukaryotic lineages. In cases where fossilizable protistan body parts exist, our understanding of protistan physiology with respect to the nitrogen cycle will help interpret the paleontologic record in terms of paleoecology.