THE NO3-/-O2 RESPIRATION RATIO OF THE DEEP SEDIMENTARY BIOSPHERE IN THE PACIFIC GYRES

Organic matter produced in the ocean has an average C/N ratio of 106:16 (the Redfield ratio). However, during transport to the seafloor, N is preferentially respired. This results in depletion of the N relative to C in the organic matter that fuels subseafloor microbial communities. It has also been argued that preferential depletion of organic N occurs in sediment. This depletion may force sedimentary microorganisms to reduce nitrate or fix dinitrogen, which requires the expenditure of a significant amount of additional energy in oxic sediment. Thus the availability of organic nitrogen may place a fundamental control on energy budget and ecosystem of the sedimentary microbial life. We test this possibility by determining the NO3/-O2 respiration ratio at three sites in the Pacific gyres. We created a diffusion-reaction model and used existing dissolved oxygen and nitrate profiles from interstitial water to determine the best-fit NO3/-O2 respiration ratio. This model has an advantage over linear correlation analysis, because it explicitly considers respiration as a function of depth, and uses the curvature of the concentration profiles to determine the microbial reactions. The down-core profiles of dissolved oxygen and nitrate reflect the net production of nitrate and consumption of oxygen due to microbial aerobic respiration. The curvature in the profiles reveals that organic nitrogen is not depleted in sedimentary organic matter respired. Dissolved nitrate and oxygen concentrations from all sites are linearly correlated with a NO3/O2 ratio of -0.098 ± 0.005. The bestfit NO3/-O2 respiration ratios calculated by the one-zone model, assuming a constant respiration ratio in the entire sediment column at each site, are between 0.089 to 0.100.

x  Table 2 The best-fit NO 3 -/-O 2 respiration ratio and the referred C/N ratio from the onezone model at each site, using both Conc. BC and Flux BC. The uncertainty of NO 3 -/-O 2 respiration ratio is of 95% confidence level. The referred C/N is calculated based on the NO 3 -/-O 2 respiration ratio and the average carbon oxidation state of 0 (R. K.R, 1963), shown in italics, and -0.7 (Hedges et al., 2002),  Table 3 The best-fit NO 3 -/-O 2 respiration ratio from the two-zone model at each site.

LIST OF TABLES
The boundary between the two zones is placed at the depth where the change of The manuscript is currently being reviewed by co-authors and is to be submitted for publication in Limnology and Oceanography.
_____________________________________________________________________ a Graduate School of Oceanography, University of Rhode Island, Narragansett, RI 02882, USA.

INTRODUCTION
Microbial life has been identified in various natural environments, including ancient deep subseafloor sediments (Parkes et al., 2000;D'Hondt et al., 2004;Røy et al., 2012). Biogenic detrital organic matter in the sediment is the primary nutrient and energy source for most deep subseafloor microbial communities (D'Hondt et al., 2004). In oxic sediment, microbial aerobic respiration occurs. The chemical composition of the organic matter strongly impacts microbial respiration and energy utilization, which further profoundly influence the structure of deep subseafloor microbial communities. However, little is known about the impact of the sedimentary organic matter composition on microbial respiration reactions. In this study, we investigate the nitrogen content of the organic matter respired during aerobic respiration and its influence on microbial bioenergetics in deep-sea sediment.
The primary source of the organic matter in the pelagic sediment is phytoplankton debris from the upper ocean (Burdige, 2006). In 1934, Redfield found that the C/N/P ratio of planktons is 106:16:1 and is spatially constant in the ocean.
This ratio is referred as the Redfield ratio. The aerobic respiration of Redfieldian organic matter can be simply expressed by the traditional Redfield-Ketchum-Richards (R.K.R) equation (Redfield et al., 1963): This equation gives the NO 3 -/-O 2 respiration ratio, which is defined as the ratio of NO 3 production to O 2 consumption. The NO 3 -/-O 2 respiration ratio depends on the C/N ratio and the oxidation state of carbon of the organic matter respired.
During transport to the seafloor, organic matter undergoes strong preferential degradation of N-containing compounds (Bishop et al., 1977;Knauer et al., 1979;Honjo, 1980). This means that proteins and nucleic acids, the primary N-containing compounds in biomolecules, are preferentially degraded relative to carbohydrates and lipids, which are major N-deficient compounds in marine organic matter. As a result, the C/N ratio of sinking organic matter increases with water depth, implying that the organic matter supplied to sediment is depleted in N relative to the Redfieldian organic matter. The C/N ratio of sedimentary organic matter in anoxic sediment has also been reported to increase with sediment depth, indicating that preferential degradation of Nrich compounds also occurs in subseafloor sediment (Arrhenius, 1953;Barreto et al., 1975;Da Rocha et al., 1975).
With continuous preferential degradation, as in the conventional view, organic nitrogen will be depleted faster than carbon in sedimentary organic matter. Such preferential depletion of organic nitrogen would strongly impact microbial energy budgets, as microorganisms might have to reduce inorganic N as their nitrogen source, 4 which is energetically costly. This may be of particular importance in deep subseafloor sediment, where energy limitation is extreme.
As in the water column, the sedimentary NO 3 -/-O 2 respiration ratio reflects the C/N ratio of organic matter consumed during aerobic respiration. It can thus be used to indicate the availability of the nitrogen in sedimentary organic matter and whether microorganisms must pay a N-reduction tax. Grundmanis and Murray (1982) analyzed dissolved nitrate and oxygen in interstitial water in shallow Equatorial Pacific sediments and found an average ratio of dissolved nitrate to dissolved oxygen of 12:130 (0.092), and C/N ratio of decomposing organic matter of 8.5 ± 1.6, which is similar to the classic Redfield ratio of 6.6. In the northeast Pacific, Murray & Kuivila (1990)  Here, we analyze dissolved oxygen and nitrate profiles in interstitial water of long sediment cores collected in the North and South Pacific gyres, and infer the availability of reduced nitrogen in the organic matter consumed during aerobic respiration. We use a diffusion-reaction model to calculate the best-fit NO 3 -/-O 2 respiration ratio and examine its variation with sediment depth in deep subseafloor sediments.

Site locations & description
All data are from three sites in the Pacific gyres ( Site 10 and the estimated basement age is about 68.5 Ma (Becker et al., 2009). At Site 11, the estimated basement depth is 40 to 100 mbsf and its age is estimated to be about 88.7 Ma (Becker et al., 2009). Site 11 is at the same location as a well-studied previously retrieved core, GPC-3. The Cretaceous/Paleogene boundary (65.5 Ma) is at 20 meters below seafloor in GPC-3, which provides an accurate and precise age estimate for Site 11 (Kyte & Wasson, 1986

Sediment core selection
Multiple types of sediment cores were collected at Site 10 and Site 11, including gravity core, multi-core and long piston core. Sediment depth adjustments were performed before numerical model analysis, if necessary, to fix the depth offsets.
Oxygen and nitrate concentrations may not be measured on the same core. Taking this into account and based on the recovery depth and the quality of the measurements of each core, gravity core GC-2 and long piston core LC-2 at Site 10 are not included in the numerical model analysis. As a result, at this site, we use gravity core GC-1, long piston core LC-1 and multicore MC in our analysis (Figure 3.1). Similarly, only gravity core GC-1, long core LC-1 and multicore MC were selected for model analysis for Site 11 (Figure 3.2).
For IODP Site U1370, only piston cores were collected, however, multiple holes were drilled ( of Hole E by adding 2 meters to the measured depths, which results in a good matching between the adjusted oxygen profile of Hole E and the oxygen profiles from other holes. However, after adjustment, the nitrate profile of Hole E was not consistent with that of Hole B between 1 mbsf to 7.5 mbsf, as shown in Figure 3.3. We use both depth scales (unadjusted depth and adjusted depth for Hole E) for Site U1370 in the model analysis.

Profile curvature evaluation
In general, the nitrate profiles looks like the mirror images of the oxygen profiles at each site: oxygen decreases with depth while nitrate increases. The changes of oxygen and nitrate concentration with depth may be the effect of only diffusion or both diffusion and microbial reactions. To determine the extent of oxygen consumption and nitrate production in the sediment, we evaluated the curvature of each concentration profile in the following manner.
The curvature of both oxygen and nitrate profiles is maximal at shallower sediments and decreases with depth, becoming undetectable at greater depths. To quantitatively evaluate the curvature, we applied an F-test to both oxygen and nitrate profiles. From the end of the profile, we chose n data points. We used a linear regression and a quadratic regression to fit these n data points, and calculated the sum of square error (SSE) of each regression. We then determined the F value and compared it to the standard F distribution table (α=0.05, df = n-3). Our null hypothesis is that a quadratic regression is not significantly better than a linear regression to fit these n data points. If the F value is smaller than the standard F value, the null hypothesis is accepted. Then we added the data point right above those n points and selected a total of n+1 data points. We then repeated the regressions and F-test (α=0.05, df = n-2) on these n+1 points. We repeated this process until the calculated F value of the chosen data was bigger than the standard F value. At this endpoint, the quadratic regression fitted the data significantly better than the linear regression.
Significant curvature occurs above the depth where the null hypothesis was not true.
This approach limits the depth of our analysis to depths where the change in oxygen concentration is large enough to cause a measurable change in nitrate concentration that is measurable.

Overview of the method
The aim of our diffusion-reaction model analysis is to find the NO 3 -/-O 2 respiration ratio that best fits the measured data. Toward this end, we calculate O 2 consumption rates as a function of depth directly from the dissolved oxygen profile.
We then predicted the nitrate production rates and nitrate concentrations at different depths from the O 2 consumption rates, in which the NO 3 -/-O 2 respiration ratio is an adjustable parameter connecting the oxygen consumption and nitrate production during microbial aerobic respiration. We determined the best-fit NO 3 -/-O 2 respiration ratio by minimizing the difference between measured and predicted nitrate values. We then applied a Monte Carlo method to estimate the uncertainty of NO 3 -/-O 2 respiration ratio due to chemical analytical/sampling errors (Wang et al., 2008).

Removal of outliers
In general, the dissolved oxygen and nitrate concentrations measured at all of the three sites are of high spatial resolution throughout the sediment column (Knorr 195(III) Shipboard Scientists, 2009;Expedition 329 Scientists, 2011a). To eliminate the influence of outliers, we smoothed the data using the LOWESS function (locally weighted scatter plot smoothing, 50%). We then determined the standard deviation of the absolute difference between measured data and removed the significant outliers, which are two or more standard deviations different from the smoothed profile. This approach insures that all the data points included in this study are within reliable analytical uncertainties.

Numerical model
In sedimentary interstitial water, at steady state, the mass balance of dissolved oxygen or nitrate in one dimension (depth) can be expressed mathematically as where C(z) indicates either oxygen or nitrate concentration (µmol L -1 ) as a function of sediment depth z (m, positive downward), D is the molecular diffusion coefficient (m 2 yr -1 ), ∅ is the sediment porosity (unitless), ! is the sediment tortuousity (unitless), and R(z) represents the rate of biological reactions (µmol·L -1 yr -1 ). In equation (1), no advection is considered, because the advection is much smaller than diffusion at these sites.
Based on Equation (2), the mass balance equation for oxygen at steady state is as follows Rearranging Equation (3), the oxygen consumption rate can be expressed as We interpolated measured oxygen concentrations to constant depth intervals (0.025 m).
At each even depth, we calculated oxygen consumption rate ! ! ( ) by finite difference approximation of equation (4) assuming D, ∅, ! are constant with sediment depth. At Sites 10,11 and U1370, the temperature variation with depth is very small and D, ∅, ! are nearly constant over the depth we considered (Lado Insua, 2013). In this model analysis, we use ! ! = 6.0×10 -6 cm 2 s -1 , !" ! ! = 4.92×10 -6 cm 2 s -1 as the diffusion coefficient for dissolved oxygen and nitrate, respectively (Murray & Grundmanis, 1980, Schulz & Zabel, 2006. Based on the calculated oxygen consumption rate from equation (4), the predicted nitrate production rate is where r is the NO 3 -/-O 2 respiration ratio, which is an adjustable parameter for every depth. Rearranging the above equation We then calculated nitrate concentrations and interpolated them to depths of the nitrate measurements. We estimated the sum of square error (SSE) between ! ! !"#$%&"' and ! ! !"#!$#"%&' and minimized it using a least-square method.
By adjusting the NO 3 -/-O 2 respiration ratio to minimize the SSE, we found the best-fit NO 3 -/-O 2 respiration ratio.
At the same time, the model also predicts an oxygen profile. The predicted oxygen profile should fit with the measured profile perfectly if the model works well. This is because the predicted oxygen profile is calculated based on the oxygen consumption rate, which is the curvature of the measured oxygen profile. We used the predicted oxygen profile to test the internal consistency of the model.

Consideration of model settings
We considered two model cases to examine the variation of the NO 3 -/-O 2 respiration ratio with depth. We refer to these as the one-zone model and the two-zone model. In the one-zone model, we assumed that the NO 3 -/-O 2 respiration ratio is constant with sediment depth in the entire sediment interval under consideration. If the respiration ration varies significantly with depth, we expect a poor fit to the measured data. In the two-zone model, we separated the sediment core into two zones, each zone with a constant and independent respiration ratio. The two-zone model is aimed to evaluate the range of the variation of the NO 3 -/-O 2 respiration ratio with sediment depth.
In the two-zone model, we used the depth where the oxygen concentration depletion is half of the total change as the boundary between two zones. This boundary is chosen so that each zone has approximately the same signal/noise ratio.
Whether or not the bottom water concentration is included when calculating the total oxygen change can potentially create an artifact, in particular at Site U1370, due to the large concentration difference between the bottom water concentration and the shallowest sediment sample measured. This difference is most likely due to loss of sediment at the core top during the coring process or sediment recovery. To address this problem, we ran the analysis two different ways; we placed the boundary at the depth where the change of oxygen concentration is calculated including (denoted as boundary depth I) and excluding (denoted as boundary depth II) the bottom water value.

Consideration of boundary condition
In the one-zone model, we considered two cases for the boundary conditions in the numerical model analysis. In the first case, we used concentrations of oxygen and nitrate in bottom water (WOCE Atlas Volume 2: Pacific Ocean) at the sample site as the top boundary condition, and the concentrations of the deepest sample from the 14 LOWESS smooth curve as the bottom boundary condition. We refer to this case as concentration boundary condition (Conc. BC). In the second case, we used the same top concentration boundary condition as in the Conc. BC, but for the bottom boundary, instead of using concentration, we used the diffusive flux as boundary condition. We refer to this case as flux boundary condition (Flux BC).
The numerical model for each boundary condition case is: Case 1: Conc. BC: Equation (6) is formulated in matrix notation as Case 2: Flux BC: Equation (6) is formulated in matrix notation as In the two-zone model, we only used the Conc. BC for the upper zone but considered two boundary conditions, the Conc. BC and the Flux BC, for the deeper zone. We only used the Conc. BC for the upper zone, instead of Flux BC, because the concentration boundary condition makes the predicted profile continuous between two zones ( Figure 4).

Determination of uncertainty
We separately considered (i) uncertainty due to sampling and chemical analysis (data uncertainty) and (ii) the uncertainty due to the choice of the boundary conditions of the model (BC uncertainty).
To evaluate the data uncertainty of estimated NO 3 -/-O 2 respiration ratio, we applied a Monte Carlo method. We calculated a relative standard deviation based on the difference between measured concentrations and the smoothed concentration profile (see above). This relative standard deviation represents the overall method error. The assumption is that deviations from a smooth profile allow the estimation of data uncertainty because the data should smoothly vary in situ due to diffusion.
To generate a representative randomized profile for the Monte Carlo analysis, we first randomly generated a group of numbers from a standard normal distribution N (0,1) and then used it to generate the randomized profiles where is the relative deviation between measured data and smoothed data, is the random number, ! is the randomized data and ! are the LOWESS smoothed data.
Repeating the same procedures, each set of randomized concentration profiles is applied in the model calculation, and used to find a best-fit NO 3 -/-O 2 respiration ratio. For this study, we (i) ran one hundred Monte Carlo simulations at each sample site using each boundary condition case and (ii) then estimated the standard deviation of the NO 3 -/-O 2 respiration ratios from each run. We take this as the uncertainty due to the sampling/analytical error on the estimated NO 3 -/-O 2 respiration ratio.
To evaluate the BC uncertainty, we compared the difference between the estimated NO 3 -/-O 2 respiration ratios using different boundary conditions in the same model. In two-zone case, moreover, we also estimated the uncertainty caused by the boundary depth between two zones. We used the difference between the calculated ratios using different boundary depth for this uncertainty.

Dissolved NO 3 vs. O 2
Nitrate and oxygen concentrations are generally linearly correlated for all sites.
However, at low oxygen concentrations, nitrate negatively deviates from the linear trend ( Figure 5). This deviation may be due to a diffusive flux into the basement or to sediment denitrification. In the linear section, the data trend are similar to the Redfield ratio derived from the water column, as given in the red line in Figure 5 ( Takahashi et, 1985, Anderson andSarmiento, 1994). Curvature in the profiles solely reflects microbial nitrate production and oxygen consumption (assuming the tortuosity and diffusion coefficient do not significantly vary with sediment depth). We use the concentration profile curvature in the intervals where there is no denitrification (based on oxygen concentrations) to identify in-situ reaction rates and the associated respiration ratio.

Evaluation of profile curvature
At Site U1370, below the depth of 50 mbsf, the oxygen concentration slightly decreases with depth to below 8 µM, which is the oxygen level generally regarded below which denitrification may occur (Devol, 1978). Therefore, we only analyze the sediment above 50 mbsf at this site.
The oxygen profiles of all three sites present a negative and concave-upward curvature, which indicates consumption of oxygen. In contrast, the positive curvature of the nitrate profiles indicates production of nitrate in the sediment. Both oxygen and nitrate profiles exhibit decreasing curvature with sediment depth, indicating that the rates of aerobic respiration and nitrate production decrease with increasing depth. We used an F-test to evaluate the curvature of the profile and limit the depth of the  (Kyte & Wasson, 1986).

Diffusion-reaction model results
The one-zone model The model-fit profiles, shown as red curves in Figure 6 (for Conc. BC) and The calculated best-fit NO 3 -/-O 2 respiration ratio, for the one-zone model at each site is given in no matter which boundary condition is used.
As mentioned in a previous subsection, two groups of data for Site U1370, the depth-unadjusted profiles (unadjusted profiles) and the depth-adjusted profiles (adjusted profiles) of oxygen and nitrate in Hole E, are used in the one-zone model.
When using unadjusted profiles, the best-fit ratio is -0.098 using Conc. BC and -0.090 using Flux BC. When using the adjusted profiles, the best-fit ratio is -0.090 using Conc. BC and -0.087 using Flux BC ( Figure 6.2 & 7.2). The difference between the NO 3 -/-O 2 respiration ratios using the two different groups of data is relatively small (about 9% in the case of Conc. BC and 3% in the case of Flux BC). We therefore use the unadjusted profile of Hole E for the remaining analysis.
The average oxidation state of carbon in marine organic matter is generally thought to be either zero, as in the R.K.R equation, or between -0.3 and -0.7 (Hedges et al., 2002). We used the range of 0 to -0.7 for the carbon oxidation state to infer the C/N ratio of the organic matter respired. The calculated C/N ratios at all sites are within the range of 7 to 9, which is very similar to ratios in the literature (Table 3; Grundmanis and Murray, 1982;Takahashi et al., 1985;Anderson and Sarmiento, 1994).

20
The calculated NO 3 -/-O 2 respiration ratios from our model analysis are shown in Figure 8. The error bars indicate the data uncertainty due to the chemical analysis and sampling process, which is estimated by the Monte Carlo simulations. This data uncertainty is less than 5% of the calculated NO 3 -/-O 2 respiration ratio using Conc. BC, and 3% using Flux BC. These small uncertainties indicate that the errors associated with the chemical analysis and sampling methods are almost negligible.
We also estimated the uncertainty of the NO 3 -/-O 2 respiration ratio due to application of different boundary conditions in the model (BC uncertainty). We base this estimate on the difference in the best-fit ratio between the two cases. The uncertainty is 5%, 8% and 9% at Site 10, Site 11 and Site U1370, respectively. The data uncertainty is almost the same magnitude as the BC uncertainty at Site 10.
However, at Site 11 and Site U1370, the BC uncertainties are about two-fold larger.
This indicates that the BC uncertainty has a larger effect than the data uncertainty.
However, all of the uncertainties are within 10%.
Within 10%, no matter which boundary condition is used, the NO 3 -/-O 2 respiration ratios at the different sites are very similar and cannot be distinguished.
The modeled NO 3 -/-O 2 respiration ratios are very close to the linear slope, -0.098, of the N/O 2 correlation ( Figure 5).

The two-zone model
We also used the two-zone model to evaluate the range of the variation of the NO 3 -/-O 2 respiration ratio at each site (Table 3; Figure 9 & 10). As previously described for the one-zone model, both Conc. BC and Flux BC are considered for the deeper zone while only the Conc. BC is used in the upper zone ( Figure 4).

Site 10
The best-fit NO 3 -/-O 2 respiration ratio for the upper zone is 0.0855 ± 0.0005, where the uncertainty covers the range of the respiration ratio due to the choice of the boundary depth between the two zones (boundary depth I or II). The calculated respiration ratios for the deeper zone are 0.119 ± 0.002 using the Conc. BC, and 0.111 ± 0.003 using the Flux BC. At this site, for both zones, the calculated respiration ratios are not sensitive to the choice of boundary depth: the relative uncertainty range is less than 6 %. The choice of different boundary conditions (Conc. BC or Flux BC) in the deeper zone produces a relative uncertainty range of ~10 %.
The best-fit NO 3 -/-O 2 respiration ratio for the one-zone model at this site, 0.089 ± 0.004 using Conc. BC and 0.094 ± 0.002 using Flux BC, is bracketed by the ratios calculated in the two-zone model, as is expected. The calculated respiration ratio in the upper zone of the two-zone model is very similar to the ratio in the one-zone model and the Redfield ratio (~10% difference). However, the respiration ratio calculated in the deeper zone is ~ 30% higher than the ratio estimated by the one-zone model.

Site 11
The best-fit respiration ratio in the upper zone is 0.148 ± 0.003, which is almost 50% higher than the ratio calculated in the one-zone model. The reason for such a large difference is the limited spatial resolution of the data. As shown in Figure   9.2 & 10.2, there are two data gaps in the upper zone of the oxygen profile, due to a combination of sampling interval and removal of bad measurements. The interpolations across these gaps are nearly linear, leading to loss of curvature and 22 reduction in the calculated oxygen consumption rate. This strongly impacts the calculated NO 3 -/-O 2 respiration ratio, which depends on the calculation of oxygen consumption rate with depth. For this reason, the calculated ratios for this zone are unreliable and we will not consider them further.
For the deeper zone, the calculated ratio is 0.0755 ± 0.0005 using Conc. BC, while using Flux BC, the calculated ratio is 0.0665 ± 0.0005. The uncertainty due to the boundary depth between the two zones, no matter which boundary condition is chosen, is very small (~1 %). The uncertainty due to the choice of the boundary condition (Conc. BC or Flux BC) in the deeper zone is ~10 %.
The best-fit ratio, in the one-zone model, is 0.100 ± 0.003 using the Conc. BC and 0.092 ± 0.001 using the Flux BC. The calculated ratio for the upper zone, in the two-zone model, is not considered due to large errors, and the ratio for the deeper zone is ~30 % lower than the ratio from one-zone case.

Site U1370
The calculated NO 3 -/-O 2 respiration ratio in the upper zone is 0.097 ± 0.007.
The uncertainty caused by the choice of the boundary depth between the two zones is ~ 15%. For the deeper zone, the calculated ratios are 0.106 ± 0.030 using Conc. BC

Profile curvatures and microbial bioenergetics
In our analysis, we assume that the interstitial water chemical profiles of nitrogen and oxygen are in diffusive steady state. The characteristic time to establish steady state in a sediment column is L 2 /2D*, where L is the sediment column thickness and D* is the diffusion coefficient corrected for the sediment tortuosity  There are two potential sources of organic matter, detrital organic matter and subseafloor microbial biomass, supplying the nitrate production and fueling the oxygen consumption. Here we show that microbial biomass is not a significant source compared to detrital organic matter by using the cell abundance profiles and sedimentation rates. The magnitudes of the carbon flux produced by decomposing cells, at Site 10, 11 and U1370, are 10 -20 , 10 -21 and 10 -22 g/cm 2 s, respectively, which are approximately 5 to 6 magnitudes lower than the carbon flux inferred from the oxygen flux into the interstitial water (based on the oxygen gradient at the sediment column-water boundary). Similarly, the magnitudes of the nitrogen fluxes produced by decomposing cells at each site approximate 10 -21 g/cm 2 s, which is also 5 to 7 magnitudes less than total diffusive nitrogen fluxes. These comparisons indicate that the source of the organic matter respired by sedimentary microbes is not simply decomposition of in-situ microbial cells but detrital organic matter that was deposited from the overlying ocean.
The availability of detrital reduced nitrogen compounds potentially has a large impact on microbial bioenergetics. The amino acids in sedimentary microbial biomass turn over and must be replaced to sustain a steady-state community. However, the observation that N-rich compounds are preferentially degraded suggests that the C/N ratio of the organic matter that can be respired will increase with depth until available N is depleted.
If organic nitrogen were depleted, in oxic sediment, microorganisms would have to reduce NO 3 or fix N 2 to N (-III). Reduction of nitrate, however, is highly energetically unfavorable in oxic sediment. McCollom & Amend (2005) have investigated the energy requirement for biomass synthesis by chemolithoautotrophic microorganisms in oxic environment by the following reaction: The Gibbs energy of this reaction is 301 KJ•(mol NH ! ! ) -1 (at 25℃, 1 bar). This means that approximately 3000 J (g cell) -1 of additional metabolic energy is required to use nitrate rather than organic nitrogen as the nitrogen source for biomass fixation.
In environments with little exogenous organic matter input and limited energy, Our analysis demonstrates that reduced organic nitrogen is still available in organic-poor sediments that are millions of years old. Consequently, microorganisms in these sediments do not pay the energetic cost of reducing nitrate to synthesize organic matter.

Dissolved nitrate vs. dissolved oxygen
The aerobic respiration of microbes in subseafloor sediment can be expressed simply by a general equation (revised from the equation of Redfield et al. (1963)), assuming the C/N ratio of the organic matter is x/y: The oxygen demand to completely oxidize the organic matter, w in equation (11), is a function of C/N ratio (x/y), and the oxidation state of carbon (ox) in the organic matter: The correlation between nitrate production and oxygen consumption therefore reflects microbial aerobic respiration and the C/N stoichiometry of organic matter consumed during this process.
Dissolved oxygen is generally linearly correlated to dissolved nitrate at all three sites (Figure 5), except for a negative deviation when oxygen falls to less than 7-8 µmol L -1 at Site U1370. This deviation is likely due to a diffusive flux into the basement or sedimentary denitrification, since 7-8 µmol L -1 is generally thought to be the oxygen level below which denitrification occurs (Burdige, 2006). If we eliminate the data with oxygen concentrations that are below 7-8 µmol L -1 , in other words, we do not consider the sediment samples collected below the depth where denitrification might occur, dissolved oxygen and nitrate are linearly correlated and the slope is -0.098 ± 0.005 (95% confidence level).
This slope of the linear correlation between dissolved oxygen and nitrate is very similar to the Redfield NO 3 -/-O 2 ratio in much of the ocean, 0.093 (Takahashi et al., 1985) and 0.094 (Anderson & Sarmiento, 1994). Additionally, in a very short pelagic core (0.5 mbsf) from the equatorial Pacific, Grundmanis & Murray (1982) measured the dissolved NO 3 and O 2 in interstitial water and also reported a linear correlation between dissolved oxygen and nitrate of -0.099 ± 0.015, and the C/N ratio is 8.1 ± 2.2 (Table 2). Our data extend this result to a much greater depth (the maximum depth is about 40 mbsf at Site U1370) and much older sediments. However, the dissolved NO 3 -/O 2 ratio from our samples is indistinguishable with their results.

Diffusion-Reaction model analysis
The one-zone model We estimate the best-fit NO 3 -/-O 2 respiration ratio based on the curvature of oxygen and nitrate profiles. The calculated ratios are given in Table 2. Based on the assumption of steady state, the calculated NO 3 -/-O 2 respiration ratios and the average oxidation state of carbon, we determined the C/N ratio of the organic matter respired during the aerobic respiration. We use -0.7 to 0 for the carbon oxidation state range (Redfield et al., 1963;Hedges et al., 2002). The calculated C/N ratios at all of the three sites, are within the range of 8.0 to 9.2 for carbon with oxidation state of 0, and 6.8 to 7.9 for carbon with oxidation state of -0.7.  (Takahashi et al., 1985), and 117:16:170, i.e. 0.094 for N/-O 2 ratio and 7.3 for C/N ratio (Anderson & Sarmiento, 1994).

The two-zone model
Within the uncertainties of the model due to the choice of boundary conditions and boundary depths between the two zones, there is no clear indication that the 30 respiration ratio varies with depth. The uncertainties due to the boundary conditions and boundary depth between the two zones, for all the three sites, are ~ 10% to 20%.
Within this range of uncertainty, we are not able to determine whether there is real variation of the respiration with depth or it is only due to the assumptions of the model settings. As mentioned above, the variation of the respiration ratio for the upper zone is within 10% and for the deeper zone is within 30%. These variations are not significant since they are almost at the same magnitude of the uncertainty due to model assumptions.
If the NO 3 -/-O 2 respiration ratio is constant through the analyzed depth interval, the chemical stoichiometry of organic matter consumed by microbes beneath the seafloor, at these three sampled sites in the Pacific gyres, is similar to the C/N ratio of organic matter in the modern ocean (the Redfield C/N ratio). We consider two possible explanations for this nearly constant respiration ratio: 1) The chemical composition of organic matter consumed is similar to that in the modern ocean and has not changed for ~ 65 million years. There is no fractionation of C/N ratio during respiration.
2) Sedimentary microbes respire organic matter according to the Redfield ratio, even if the chemical composition of the bulk sedimentary organic matter has a different C/N ratio due to either change with time in the oceanic Redfield Ratio or fractionation in the water column or during sedimentation.
The first explanation, however, is inconsistent with the observation of preferential N degradation. The C/N ratio of the organic matter in the water column and sediment are observed to increase with depth (Arrhenius, 1953;Bishop et al., 31 1977;Knauer et al., 1979). In contrast, the second explanation is more consistent.
Even though the C/N ratio of the bulk sedimentary organic matter may differ from the Redfield ratio, at these sites, microbes only respire organic matter with the Redfield ratio. The respiration of organic matter with this nearly constant ratio may be controlled by the nutrient requirement of microbial communities and selection pressure to maximize ecological efficiency.

SUMMARY
In summary, we used a diffusion-reaction model to analyze the NO 3      Takahashi et al., (1985). The cyan solid line represents the NO 3 -/-O 2 Redfield ratio measured by Anderson & Sarmiento (1994). A B Figure 10.3 The best-fit oxygen and nitrate curves from the two-zone model at Site U1370. Measured depth of Hole E is used. The solid horizontal lines indicate the depth where the sediment column is separated. Flux BC is used for the deeper zone.  Table 2 The best-fit NO 3 -/-O 2 respiration ratio and the referred C/N ratio from the onezone model at each site, using both Conc. BC and Flux BC. The uncertainty of NO 3 -/-O 2 respiration ratio is of 95% confidence level. The referred C/N is calculated based on the NO 3 -/-O 2 respiration ratio and the average carbon oxidation state of 0 (R.K.R, 1963), shown in italics, and -0.7 (Hedges et al., 2002), shown in parenthesis. The C/N/-O 2 ratios from pervious studies are also listed for comparison.