Remineralization ratios of carbon, nutrients, and oxygen in the North Atlantic Ocean: A field databased assessment



[1] Remineralization ratios of carbon, nutrients, and oxygen have been assessed in the North Atlantic Ocean along the WOCE 1A/E section. The study is based on an extensive field data set comprising dissolved inorganic carbon (DIC), nitrate and nitrite (NO3/2), phosphate (PO4), and oxygen (O2) data as well as hydrographic data. A procedure has been introduced which normalizes DIC data to constant salinity and temperature and corrects for the contamination from anthropogenic CO2. The remaining variability on the normalized DIC values (DICbio) can be attributed to the remineralization of organic matter. DICbio can thus be seen as carbon-analogy to the apparent oxygen utilization (AOU). The consecutive evaluation of the remineralization ratios obtains two different regimes separated at the density level ρ = 1027.7 kg m−3. In the shallower level the ratios (C/N = 4.5; C/P = 67, AOU/C = 2.0; N:P = 15; AOU:P = 134; AOU/N = 9.0) are shifted toward relatively higher nutrient release and higher oxygen consumption with respect to the Redfield ratios of particulate organic matter (POM). In contrast, in the deeper levels the ratios are shifted toward relatively higher carbon release and lower oxygen demand (C/N = 11; C/P = 152, AOU/C = 0.86; N:P = 13.9; AOU:P = 130; AOU/N = 9.4). The depth integrated inventories of the remineralization products (DIC, NO3/2, PO4, and AOU) provide water column averaged ratios for the investigation area (C/N = 8.8; C/P = 124, AOU/C = 1.1; N:P = 14.2; AOU:P = 131; AOU/N = 9.3) which imply a higher efficiency of the biological carbon pump in the North Atlantic Ocean than predicted with respect to the elemental composition of POM.

1. Introduction

[2] The uptake of atmospheric CO2 by the oceans via the “biological carbon pump” is driven by primary production in the euphotic zone. Dissolved inorganic carbon (DIC) and nutrients are converted to particulate organic matter (POM) during photosynthesis. Only a minor part of the POM escapes remineralization in the surface layer by settling down into the deeper water column [Eppley and Peterson, 1979]. The export production in turn is either remineralized in the deeper water column causing DIC and nutrient release, or it is eventually buried in the sediments over geological timescales. The DIC thus exported from the surface layer is replenished by CO2 from the atmosphere and therefore represents the oceanic uptake of atmospheric CO2 via the “biological carbon pump.” The DIC and nutrients which are released in the deeper water column are then transported by the thermohaline circulation to the upwelling areas, where finally CO2 is given back to the atmosphere.

[3] The concept of Redfield et al. [1963] has been applied in order to assess the DIC released by remineralization of POM in the deeper water column. This biogeochemical key concept is based on the assumption that the elemental ratios of carbon and nutrients in freshly produced POM are equal to the corresponding release ratios of DIC and nutrients during remineralization of POM in the deeper water column. Redfield's classical ratios of carbon (C), nitrogen (N), phosphorus (P), and oxygen (O2) were given to C:N:P: − O2 = 106:16:1:138 with the oxygen ratio being inferred from stoichiometric considerations. Accordingly, changes in nutrient concentrations in the deeper water column can be converted to the associated carbon units with reference to carbon/nutrient relationships observed in freshly produced POM. Alternatively, Anderson and Sarmiento [1994] suggested that the remineralized DIC can be derived using the oxygen/carbon ratio observed during photosynthesis [Laws, 1991]. Several methods have followed one of these routes in order to assess the biological carbon pump.

[4] The Redfield concept with its assumption of fixed conversion factors has been a very useful tool for first estimates of the biological carbon pump. However, for more detailed analysis, for example assessing the penetration of anthropogenic CO2 using separation concepts introduced by Brewer [1978] and Chen and Millero [1979], this assumption has to be carefully reconsidered, since the separation concepts strongly depend on the Redfield ratios [e.g., Wanninkhof et al., 1999]. It has become evident that either during primary production or during remineralization of POM this assumption does not necessarily hold true. For example, a wide systematic change of carbon/nutrient ratios has been observed during photosynthesis [e.g., Thomas et al., 1999, and references therein; Copin-Montegut, 2000; Osterroht and Thomas, 2000; Körtzinger et al., 2001]. The elemental composition of particles trapped in the deeper water column may also show a systematic change in the elemental composition with depth [e.g., Olesen and Lundsgaard, 1995; Knauer et al., 1979; Wakeham et al., 1984; Honjo and Manganini, 1993]. Moreover, several studies have analyzed the related concentration changes of DIC, nitrate (NO3/2), phosphate (PO4), and oxygen in the deeper water column arguing in favor of changes of the release ratios [e.g., Minster and Boulahdid, 1987; Shaffer, 1996; Hupe and Karstensen, 2000; Hupe et al., 2001].

[5] The North Atlantic Ocean as a high latitude regime plays a key role in taking up atmospheric CO2 via the solubility pump [e.g., Thomas et al., 2001]. Moreover, it is characterized by high primary productivity which causes high CO2 uptake via the biological pump [e.g., Falkowski et al., 1998]. Despite this relevance, there is still only sparse information available about the remineralization ratios of POM in the deeper water column of the North Atlantic Ocean. Due to the lack of DIC data most earlier studies were confined to nutrient/oxygen ratios [Takahashi et al., 1985; Minster and Boulahdid, 1987; Fanning, 1992]. Other studies excluded the high latitude regions of the Atlantic Ocean, notably the North Atlantic [e.g., Anderson and Sarmiento, 1994] or focused on global rather than on regional scales [Shaffer, 1996].

[6] Here, a new method is proposed to assess the remineralization ratios of POM in the deeper water column of the North Atlantic Ocean referring to highly accurate DIC data determined according to Johnson et al. [1993] and WOCE nutrient data. The key idea is to develop a procedure which corrects the observed DIC data for the variability of the background and the anthropogenic CO2 signal. The remaining signal is then only affected by the remineralization of organic matter which thus enables the assessment of the remineralization ratios including carbon-related ratios. The corrected DIC value would be analogous to the apparent oxygen utilization (AOU).

2. Data

[7] The data were collected during the German WOCE cruise Meteor 30/3 from 24 November 1994 through 15 December 1994 covering the WOCE A1/E section (Figure 1). Dissolved oxygen (O2) and the nutrients nitrate (including nitrite, NO3/2) and phosphate (PO4) were determined in 920 samples from 43 stations according to the WOCE operations manual [World Ocean Circulation Experiment, 1994; Koltermann et al., 1996]. The overall errors were lower than 0.2% for O2, 1.38% for NO3/2, and 1.27% for PO4. The 360 samples from 42 stations were analyzed for dissolved inorganic carbon (DIC) immediately after sampling using the coulometric SOMMA system described by Johnson et al. [1993]. DIC reference samples (CRM, produced by A. Dickson, Scripps Institution of Oceanography) were used to calibrate the measurements for each cell individually. The overall error of the system was determined to ±1.5 μmol kg−1 (<0.1%) during onboard operation. The calculations of the equilibrium reactions of the carbonate system were performed using the program CO2SYS, version 1.04 [Lewis and Wallace, 1998].

Figure 1.

Cruise track and stations occupied along the WOCE A1 line during Meteor cruise 30/3. Stations east of Greenland (WOCE line A1/E) are subjects of investigation. Figure 1 is reprinted from Thomas and Ittekot [2001] with permission from Elsevier Science.

3. Observations

[8] Detailed discussion of the O2 and DIC distributions as well as of hydrographic topics relevant for the WOCE A1/E section (Figure 1) are given by Thomas and Ittekkot [2001, and references therein] or, for example, by Stoll et al. [1996]. The discussion here is thus confined to a brief description of the distributions of DIC, O2, nutrients, and density. An overview of the concentrations observed in the relevant water masses is given in Table 1.

Table 1. Major Water Masses and Corresponding DIC, Nutrient, and Oxygen Concentrationsa
Water Masses Approximate Depth, mApproximate DIC, μmol kg−1Approximate O2, μmol kg−1Approximate AOU, μmol kg−1Approximate NO3/2, μmol kg−1Approximate PO4, μmol kg−1
Surface waters  2080–2130260–3000–208–130.5–0.8
SPMWSubpolar Mode Water250–9002120–2150230–25020–7011–170.8–1.1
IWIntermediate Water900–12002170230–2608519–201.2
LSW, Irminger SeaLabrador Seawater>5002148280–30030171.1
DSOWDenmark Strait Overflow Water2500–3000211031030151.0
LSW, Iceland BasinLabrador Seawater1800215028045191.2
NADWNorth Atlantic Deep Water2500–30002190260–2808019–221.2–1.4
AABWAntarctic Bottom Water>40002200<25090>23>1.5

[9] The surface concentrations of PO4 (Figure 2a) increase slightly in a westerly direction from approximately 0.5 to 0.8 μmol kg−1. The Irminger Sea, characterized by younger Labrador Seawater (LSW) [e.g., Sy et al., 1997], reveals a rather homogenous level of ∼1.1 μmol kg−1 PO4, which decreases slightly in the Denmark Strait Overflow Water (DSOW) above the bottom (1.0 μmol kg−1). East of the Mid-Atlantic Ridge (MAR) the Subpolar Mode Water (SPMW), separating the surface waters from the Intermediate Water (IW), shows PO4 concentrations between 0.8 and 1.1 μmol kg−1. The IW can be identified by the intrusion of nutrient-rich waters from the south (1.2 μmol kg−1 PO4), while the older LSW below the IW shows again lower concentrations of approximately 1.1 μmol kg−1 PO4. The deeper water masses, North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW), show increasing PO4 levels of 1.2–1.4 and >1.5 μmol kg−1, respectively. In general, NO3/2 shows the same features as PO4 and the values are depicted in Table 1 (Figure 2b). The pattern of the oxygen concentrations (Figure 2c) can be seen as mirror image of the nutrient concentrations, although this image is disturbed by strong solubility effects. Accordingly, highest values can be observed in the surface waters. In the subsurface waters of the Irminger Sea (younger LSW), O2 concentrations of approximately 280–300 μmol kg−1 are found increasing in the deeper DSOW because of its very low temperatures (∼310 μmol kg−1). Within the eastern basin, SPMW and LSW show lower oxygen concentrations (230–260 μmol kg−1) which increase again in the LSW (280 μmol kg−1). As expected, the DIC concentrations (Figure 2d) follow the nutrient pattern with lowest concentrations in the surface waters. The LSW in the western basin is characterized by concentrations of about 2148 μmol kg−1 which decrease in the DSOW to approximately 2110 μmol kg−1 (not visible in Figure 2d because of the lower spatial resolution of the DIC samples). In the eastern basin the SPMW reveals concentrations between 2120 and 2150 μmol kg−1 increasing to approximately 2170 μmol kg−1 in the IW. A relative DIC minimum is observed in the LSW of 2150 μmol kg−1, while in the deeper NADW and AABW the concentrations increase to 2190 and 2200 μmol kg−1, respectively. The density level of ρ = 1027.7 kg m−3 (Figure 2e) ascends in westerly direction from ∼1600 m depth to less than 200 m at around 39°E. At the Greenland shelf this level is observed at a depth of approximately 800 m. As discussed later, this density level indicates the separation line of the upper and lower water column regarding the remineralization ratios of POM.

Figure 2.

Distributions of (a) PO4, (b) NO3/2, (c) O2, (d) DIC, and (e) density observed along the WOCE A1/E section. The density level ρ = 1027.7 kg m−3 (dash-dotted line) in Figure 2e represents the separation line of the water column into a shallower and deeper part. See text for details. Figure 2d is reprinted from Thomas and Ittekot [2001] with permission from Elsevier Science.

4. DIC Changes in the Deeper Water Column Due to Remineralization of Organic Matter

[10] In order to assess carbon export from the surface to the deeper waters via the biological carbon pump, the remineralization ratios of POM are usually referred to. Briefly, these ratios can be obtained employing “preformed conditions“ of a water mass and assigning the differences between preformed conditions and observations to the remineralization of organic matter. The preformed conditions describe the state of a water mass at the sea surface before a water mass is subducted. The relevant parameters to describe the preformed conditions regarding DIC are temperature (T), salinity (S), Alkalinity (AT) and the partial pressure of CO2 (pCO2). In order to define the preformed conditions of DIC an absolute value of AT is thus required. This might be obtained exploiting linear relationships between salinity and alkalinity, which, however, reveal regional variability [e.g., Millero et al., 1998]. The description of the preformed condition of an individual sample thus requires reliable knowledge of both the origin of the water masses constituting this sample and their mixing history.

[11] In order to avoid the above indicated shortcomings of describing preformed conditions a two-step procedure is employed to extract the biological variability from the observed DIC concentrations. This procedure (1) normalizes observed DIC concentrations to standard T, S, and AT conditions and (2) corrects for the anthropogenic CO2 contamination. Since calcification is of minor occurrence in the North Atlantic Ocean [e.g., Körtzinger et al., 2001], it can be neglected here. The parameter describing biological variability in DIC (DICbio) is thus computed as

equation image

considering the contributions of solubility effects compared to standard conditions (ΔDICsol) and anthropogenic CO2 (ΔDICant) to the observed DIC concentrations (DICobs).

4.1. Normalization of DIC Concentrations for Solubility Effects

[12] The solubility of CO2 in seawater is determined by its T, S, and AT conditions, which cause part of the observed variability in the DIC concentrations. The suggested normalization procedure to standard T, S, and AT conditions generates a homogenous DIC background; in front of that solely remineralization processes and ΔDICant cause the variability in DIC. Since the procedure accounts for differences between observed and standard conditions, the influence of the absolute value of AT on those differences is negligible and the generalized description of AT according to Millero et al. [1998] [AT = 520.1 + 51.24S] is referred to. A temperature of T = 6°C, a salinity of S = 35 and a corresponding AT have been chosen as standard conditions, since T = 6°C and S = 35 represent approximately the mean values observed during the cruise. Keeping in mind the above AT/S relationship a set of standard DIC values (DICstd) is defined for Tn = 6°C and Sn = 35 and given pCO2 values between 280 and 370 μatm. In order to describe the solubility effect (ΔDICsol), i.e., the impact of different T, S, and AT conditions on the DIC concentrations, the difference between DICstd and the DIC calculated for different temperatures between 0° and 20°C and different salinities between 34 and 37 at different pCO2 conditions DIC[34–37, 0–20°C, pCO2]:

equation image

The multiparameter analysis of the results of this exercise provides a linear equation which describes ΔDICsol just as a function of S, T [in °C], and pCO2 [in ppm]:

equation image

[13] Note that this equation does not yet account for the correction for anthropogenic CO2 contamination; it only describes the normalization of the DIC at different pCO2 conditions. Figure 3 shows the correction for a given pCO2 of 350 μatm. For the present study the pCO2 of the individual samples has been estimated following Thomas and Ittekkot [2001]. Accordingly, the pCO2 can be estimated with regard to the ventilation ages of water masses and the time history of the atmospheric CO2. The ventilation ages of the key water masses observed on the WOCE A1/E section are adopted and interpolated for all samples between the key water masses. Referring to the time history of the atmospheric CO2, the corresponding pCO2 is then obtained. Because of the low pCO2 coefficient, any pCO2 disequilibrium can be neglected here. An alternative to determining the pCO2 for each sample individually is to assume a fixed (mean) pCO2 value instead of a specific one.

Figure 3.

Plot of the function (equation (3)) to normalize observed DIC values to a constant temperature (6°C) and salinity (35) shown for a pCO2 of 350 μatm. The cross indicates the zero point with respect to Tn = 6°C and Sn = 35.

4.2. Correction for Anthropogenic CO2 Contamination

[14] Most of the methods assessing the penetration of anthropogenic CO2 in the oceans follow the separation concept introduced by Brewer [1978]. This concept, which has been modified and improved by several authors [e.g., Chen and Millero, 1979; Gruber et al., 1996; Körtzinger et al., 1998], refers to remineralization ratios of organic matter in order to separate out the biological component of DIC. Applying this concept to the present study would therefore not allow an independent determination of the remineralization ratios of POM, since these then would be part of the initial conditions. Recently, an alternative approach of assessing anthropogenic CO2 in the oceans has been suggested which refers to ventilation ages of water masses and CO2 equilibrium chemistry of seawater and is independent from remineralization ratios of POM. This approach has been applied on a global scale [Thomas et al., 2001] and also on the current data set of North Atlantic Ocean [Thomas and Ittekkot, 2001]. The latter data have been used to quantify ΔDICant in the present study (equation (1)).

4.3. Biological Variability of DIC (DICbio)

[15] This two-step procedure has been applied to the observed DIC concentrations and the associated T and S values. The cumulative profiles (Figures 4a and 4b) indicate that the values are decreased by the normalization in the deeper waters mainly as a consequence of the low temperatures whereas in the upper layers the values will be increased compared to the observations. Since the gradients in salinity are rather small in the investigation area, the normalization of the S (and AT) is of minor relevance compared to the temperature correction. As expected, the correction for anthropogenic CO2 contamination decreases with depth. The variability of DICbio provides a measure of the carbon which has been released by remineralization of organic matter from a homogeneous background. DICbio thus enables the assessment of the remineralization ratios with respect to oxygen and nutrients and can been seen as the carbon analogy to the apparent oxygen utilization (AOU). An absolute value of remineralized DIC as provided by the AOU for oxygen is not given at this stage, since the absolute value of DICbio depends on the standard conditions. Therefore, the obtained relationships of AOU to DIC, NO3/2, and PO4 will be exploited later to quantify water column inventories of remineralized DIC, NO3/2, and PO4.

Figure 4.

(a) The impact of the normalization procedure is shown for the cumulated DIC profiles. The diamonds indicate the observed values, whereas the dots represent the T- and S-normalized values and the crosses represent the values which have been additionally corrected for the contamination of anthropogenic CO2 (ΔDICant). (b) The corrections are shown for the individual samples. The triangles show the correction for ΔDICant, the dots show the normalization for T and S, and the crosses show the total correction. Since the observed variability in S is rather low (Smin ≈ 34 and Smax ≈ 35.5), the changes of the T/S normalization (dots) from negative to positive values can mainly be attributed to decreasing temperatures.

5. Release Ratios of DIC and Nutrients Due to the Remineralization of POM

[16] The remineralization ratios of POM will be discussed for DIC and the nutrients NO3/2, and PO4 and AOU. Already, the qualitative view of the property/property plots indicates two different biogeochemical regimes as exemplified by the DICbio/AOU plot (Figure 5). Apart from the surface layer samples (triangles), two different groups of samples can be identified, one characterized by a less steep slope (black dots) and one by a steeper slope (diamonds). The detailed analysis of this feature leads to the separation of the water column along the density horizon ρ = 1027.7 kg m−3 into an upper and deeper part (see also Figure 2e). The surface waters, the SPMW, and the IW constitute upper density range (ρ < 1027.7 kg m−3), and the layers below (ρ > 1027.7 kg m−3) are mainly characterized by both types of LSW and the deeper NADW and AABW.

Figure 5.

Plot of AOU versus normalized DIC values (DICbio). The separation of the water column is denoted by different symbols. The triangles mark the surface waters corresponding to AOU < 20 μmol kg−1 which have been excluded from further discussion. The diamonds mark the upper water column characterized by ρ < 1027.7 kg m−3, and the dots mark the deeper part below ρ = 1027.7 kg m−3. See Figure 2e for location of the density level ρ = 1027.7 kg m−3.

[17] Surface layer samples (associated with AOU < 20 μmol kg−1) have not been considered for further discussion, since both AOU and DIC are affected by air-sea exchange of O2 and CO2, respectively. Geometric mean functional regression analysis [Sprent and Dolby, 1980] has been performed on the property/property relationships. As indicated by Figure 5, the water column can be separated into two regimes with different biogeochemical characteristics. Accordingly, the remineralization ratios of carbon, nutrients and oxygen (Table 2) are shown for the water column above and below ρ = 1027.7 kg m−3 in Figures 6 and 7, respectively.

Figure 6.

Remineralization ratios of DIC nutrients and oxygen obtained for the upper water column above ρ = 1027.7 kg m−3. The plots are shown with the same scale as Figure 7 to enable better comparison.

Figure 7.

Remineralization ratios of DIC nutrients and oxygen obtained for the deeper water column below ρ = 1027.7 kg m−3. The plots are shown with the same scale as Figure 6 to enable better comparison.

Table 2. Remineralization Ratios Obtained for the WOCE A1/E Section in the North Atlantic Ocean
ρ < 1027.7 kg m−34.5 ± 0.367 ± 52.0 ± 0.115 ± 0.3134 ± 99.0 ± 0.6
ρ > 1027.7 kg m−311.0 ± 0.5152 ± 60.86 ± 0.0413.9 ± 0.3130 ± 69.4 ± 0.5

5.1. C:N Ratios

[18] The present study obtains for the upper layers of the North Atlantic Ocean above ρ = 1027.7 kg m−3 a carbon to nitrogen remineralization ratio of C:N = 4.5 ± 0.3 (Figure 6a) whereas for the deeper layer a C:N ratio of 11.0±0.5 is obtained (Figure 7a). The shift of the ratios implies relatively high nitrogen release by particle remineralization compared to carbon, whereas in the deeper layers the remaining carbon enriched tissue is remineralized. This general trend has been found in several studies which include carbon-related ratios. Good agreement is found with the gradients reported by Shaffer [1996] which increase from C:N = 5.0 in the upper layers to 9.3 in the deeper layers. Shaffer [1996] applied a modeling technique, and the corresponding C:N ratios have been converted from the given C:P and N:P ratios. The results of the present study are also in agreement with data from particle trap studies in the North Atlantic Ocean by Honjo and Manganini [1993]. The authors found the same gradient of increasing C:N ratios with depth which were found to be somewhat higher (6–8) for in the shallower layers but similar for the deeper layers (8–10), respectively. The low C:N ratio in the shallower layers presented here is also in reasonable agreement with data reported by Takahashi et al. [1985] for the isopycnal surfaces ρ = 1027.0 kg m−3 and ρ = 1027.2 kg m−3.

5.2. C:P Ratios

[19] For the shallower layer above ρ = 1027.7 kg m−3, a carbon to phosphorus release ratio of C:P = 67 ± 5 has been obtained, whereas it increases in the deeper layers to 152 ± 6 (Figures 6b and 7b). Both the shallower and the deeper C:P ratios are significantly different from the Redfield C:P ratio of 106:1. As already observed regarding the C:N ratios, the low C:P ratio in the shallower layers might be a hint to a preferential release of nutrients, whereas the carbon tissue is exported to the deeper layers causing high C:N and C:P ratios [e.g., Thomas et al., 1999]. The deviation from the strictly Redfieldian behavior is confirmed by the sediment trap data [Honjo and Manganini, 1993] which show also similar C:P ratios in both the shallower and deeper layers. Moreover, the C:P ratios by Takahashi et al. [1985] for the isopycnal surfaces ρ = 1027.0 kg m−3 and ρ = 1027.2 kg m−3 as well as the modeled C:P ratios and their gradient by Shaffer [1996] can be confirmed in general. The C:N and C:P ratios for the North Atlantic Ocean obtained in this study cannot support results of depth-invariant ratios as reported by earlier studies for different oceanic regimes.

5.3. AOU:C Ratios

[20] The AOU:C ratio changes from 2.0 ± 0.1 in the shallower layer to 0.86 ± 0.04 (Figures 6c and 7c) in the deeper layer as already indicated by Figure 5. The results clearly imply two regimes with different biogeochemical characteristics. Similar gradients of decreasing AOU:C have been reported, for example, by Hupe and Karstensen [2000] for the Arabian Sea (AOU:C = 1.56 in 500 m and AOU:C = 1.25 in the deepest layers) as well as by Shaffer [1996] (AOU:C = 1.5 in 100 m and AOU:C = 1.28 in 3000 m), although it has to be noted that the differences in the extreme values of the latter studies are smaller than in the present study. Taking into account stoichiometric considerations, the apparently high AOU:C ratio of the shallower layers is in agreement with the above low C:N release ratio (4.5:1), which might be justified by a high oxygen demand during combustion of organic matter as indicated by the idealized equation for amino acid combustion with short carbon chains (equations (4a) and (4b) with 4 and 5 carbon atoms):

equation image
equation image
equation image

Furthermore, high AOU:C ratios have been reported for the North Atlantic Ocean by Takahashi et al. [1985] (AOU:C = 1.7 and AOU:C = 1.96 at ρ = 1027.0 kg m−3 and ρ = 1027.2 kg m−3, respectively) and Körtzinger [1995] (1.97 and 1.98, respectively). The low AOU:C ratio found for the deeper layers might be interpreted as hint for the combustion of already high oxygenated organic matter. For example, the combustion of carbohydrates would equal an AOU:C ratio of 1:1 (equation (4c)).

5.4. N:P Ratios

[21] The N:P-remineralization ratio obtained to be 15 ± 0.3:1 for the shallower layer (Figure 6d and Table 2), which is slightly lower than the Redfieldian value of POM (C:P = 16:1). The ratio decreases with depth to N:P = 13.9 ± 0.3:1 (Figure 7d and Table 2). The general trend of these results is in good agreement with data reported by Minster and Boulahdid [1987], although their data show higher values in the shallower layers (16.73 to 15.5:1 at ρ = 1027.0 to 1027.4 kg m−3). The N:P ratios of POM reported by Honjo and Manganini [1993] reveal some variability and thus show only a weak agreement with the data presented here. The ratios of this study might not confirm with depth constant N:P ratios; however, since the difference between the shallower and deeper N:P ratios is rather small, a note of caution should be taken.

5.5. AOU:P Ratios

[22] The AOU:P remineralization ratios presented here of AOU:P = 134 ± 9:1 for the shallower and 130 ± 6:1 for the deeper layers (Figures 6e and 7e and Table 2) appear to be rather constant, taking into account the uncertainties. The values confirm the range of Redfield et al.'s [1963] AOU:P ratio which was inferred from stoichiometric considerations. The AOU:P ratios reported by Minster and Boulahdid [1987] for layers at ρ = 1027.4 and 1027.8 kg m−3 (142:1 and 127:1, respectively) are in agreement with the findings of this study. It is worth noting here that during the remineralization of organic matter, the relationship between oxygen and phosphorus is different from that between for example nitrogen and oxygen. The released phosphate water column can been seen as product of the hydrolysis of organic phosphoric acid esters, a process which does not consume oxygen. In contrast the release of NO3/2 to the water column points to the oxidation of amino acids or other reduced nitrogen compounds (see also equations (4a) and (4b)). Thus, the AOU:N ratio (to be discussed below) describes a joint process of oxygen and nitrogen, whereas the AOU:P ratio rather describes parallel processes.

5.6. AOU:N Ratios

[23] The AOU:N remineralization ratios have been obtained to AOU:N = 9.0 ± 0.6:1 for the shallower and AOU:N = 9.4 ± 0.5:1 for the deeper layers, respectively (Figures 6f and 7f and Table 2). In the given range of uncertainty, both ratios appear to be similar. The value of the AOU:N ratios is in agreement with the results by Minster and Boulahdid [1987] ranging between 9.1:1 in the shallower layers and 8.7:1 at ρ = 1027.8 kg m−3. The Redfield AOU:N ratio of 8.6:1 [Redfield et al., 1963] is also close to the present findings.

6. Water Column Inventories of Remineralized DIC, NO3/2, and PO4

[24] In order to assess the contributions of remineralization of organic matter to the observed concentrations of DIC, NO3/2, and PO4, an absolute measure is required, since the above ratios describe only relative changes. This absolute measure for the remineralization processes is provided by AOU, which in turn can be used to assess the required absolute changes [e.g., Hupe et al., 2001]. Thus, AOU concentrations have been related to the above remineralization ratios in order to obtain the remineralization inventories. The remineralized contributions of all four parameters (PO4,rem, NO3/2,rem, AOU, and DICrem; Figures 8a–8d) clearly reflect the separation between the western and eastern basins. Lower and rather homogeneous values are obtained for the western basin due to its younger age and thus shorter “biological history.” The concentrations in the western basin vary between 0.2 and 0.3 μmol kg−1 PO4,rem, 3 and 4 μmol kg−1 NO3/2,rem, 25 and 40 μmol kg−1 AOU, and 30 and 40 μmol kg−1 DICrem. In the eastern basin the IW originating from the subsurface waters of the equatorial Atlantic Ocean clearly can be identified by high concentrations of the remineralization products PO4,rem, NO3/2,rem, and AOU (0.4–0.6 μmol kg−1, 6–9 μmol kg−1, and 60–80 μmol kg−1, respectively), whereas a clear fingerprint of DICrem cannot be obtained due to the low carbon-related ratios in the shallower layers (ρ < 1027.7 kg m−3). As indicated above, the “missing carbon” in the shallower layers can be interpreted as a consequence of the combustion of N-and P-rich material under high oxygen demand. The intrusion of the LSW into the intermediate layers of the water column of the eastern basin is shown by lower concentrations of PO4,rem, NO3/2,rem, and AOU. In the deeper layers the concentrations of all parameters show an increase to values higher than 0.6 μmol kg−1PO4,rem, 9 μmol kg−1NO3/2,rem, 80 μmol kg−1AOU, and 100 μmol kg−1DICrem, respectively.

Figure 8.

(bottom) Distributions and (top) water column inventories of remineralized (a) PO4, (b) NO3/2, (c) AOU, and (d) DIC observed along the WOCE 1A/E section obtained as function of AOU and the corresponding remineralization ratios. (e) For comparison, the distribution of ΔDICant is given. Figures 8c and 8e (redrawn here) are reprinted from Thomas and Ittekot [2001] with permission from Elsevier Science.

[25] The water column inventories (given in the top of Figures 8a–8d) show as a general trend an increase from west to east, which can be seen as a function of water column depth, but most notably as a consequence of the longer “biological history” of the waters in the eastern basin compared to the younger waters in the western basin. Here, the PO4,rem inventory is slightly below 1 mol m−2 PO4,rem, while for the eastern basin, approximately 2 mol m−2 PO4,rem are calculated. The same patterns are obtained for NO3/2,rem, AOU, and DICrem revealing inventories of 10 mol m−2 NO3/2,rem, 100 mol m−2 AOU, and 100 mol m−2 DICrem, in the western basin and 30 mol m−2 NO3/2,rem, 280 mol m−2 AOU, and 280 mol m−2 DICrem, in the eastern basin, respectively. For convenience, the distribution of ΔDICant [Thomas and Ittekkot, 2001] is shown in Figure 8e. While the remineralized compounds show a general increase with depth, the ΔDICant concentrations are highest at the surface. These expected different features are caused by the different input ways: ΔDICant enters the ocean via the air-sea interface, whereas the remineralized compounds are released within the ocean and show the highest concentrations in those parts of the water column which are exposed for long time to the remineralization of POM. Expect for the IW, those are generally the deeper parts of the water column. In contrast to the compounds released by organic mater remineralization, ΔDICant shows similar inventories in both basins. The western basin has been ventilated more recently and thus is characterized by higher ΔDICant concentrations throughout the water column. These higher concentrations cause the comparable inventories, although the eastern basin is deeper.

[26] In order to obtain remineralization ratios for the entire investigation area, the concentrations of PO4,rem, NO3/2,rem, AOU, and DICrem are integrated over the water column and linearly interpolated along the sampling section. The averaged values are given in Table 3. The variability of the individual ratios is caused by the varying water column depth along the section and the subsequent application of the two different ratios according to the two density layers. Consequently, the AOU:P and AOU:N ratios show rather small variability, since the ratios are almost constant over the entire water column. In contrast, the carbon-related ratios show greater variability since the water column depth and thus the applied ratios vary along the section. Even the integrated C:N, C:P, and AOU:C ratios show that the carbon export to the water column is higher than predicted by the observed Redfield ratios of POM. The trend might also hold true for the export of P compared to N as indicated by the N:P, AOU:N, and AOU:P ratios.

Table 3. Integrated Water Column Remineralization Ratios Obtained for the WOCE A1/E Section in the North Atlantic Ocean
Water column average8.8 ± 1.7124 ± 221.1 ± 0.314.2 ± 0.3131 ± 19.3 ± 0.1
Redfield ratio6.61061.3161388.6

7. Discussion

[27] The present study in the North Atlantic Ocean confirms the picture of variable remineralization ratios. This has been shown for other regions of the world ocean [e.g., Hupe and Karstensen, 2000], but also for the North Atlantic Ocean restricted to N, P and AOU. Notably the carbon-related ratios obtained above, which play a key role in assessing the biological pump, now complete this picture and thus enable a comprehensive description of the relationships between carbon, nitrogen, phosphorus, and oxygen in the water column of the Atlantic Ocean. Attempts have been made to provide explanations for the variability as reviewed, for example, by Azam [1998] or Thomas et al. [1999]. The findings of lower C:N and C:P ratios in the shallower levels might thus be explained by higher N and P demand finally constituting a preferential recycling of nutrients. The high AOU to carbon ratios also fit into the picture of the early remineralization of organic nitrogen and phosphorus compounds like amino acids or phospholipids. The ratios integrated over the water column show that, on average, the export of carbon is higher than predicted by the elemental composition of POM. When applying these ratios in dynamic modeling experiments, more emphasis would be laid on the differences between the different depth levels. This means that carbon export relative to the nutrients along the deep currents would be described by the deeper water column ratio rather than by the water column average. The remineralization obtained here ratios might also have strong impact on the determination of anthropogenic CO2 following the above separation concepts, since the quantification of the biological component of the total DIC would be affected.

8. Conclusions

[28] The proposed normalization of DIC with respect to hydrographic and atmospheric conditions provides a corrected term DICbio which reflects biological changes of DIC and thus can be seen as carbon analogy to the apparent oxygen utilization (AOU). DICbio enables the description of carbon-related remineralization ratios of POM in the water column and thus provides a direct tool for assessing the “biological CO2 pump.” Two density levels can be identified, characterized by different remineralization ratios indicating different microbial key processes. Notably, the obtained carbon-related remineralization ratios are on average higher (for oxygen: lower) than the corresponding composition ratios of POM and thus imply a higher carbon drawdown of the “biological CO2 pump” than predicted by the Redfield ratios of POM.


[29] The excellent cooperation of the crew and the scientific staff during the METEOR cruise M30/3 under wintry conditions is gratefully acknowledged. Special thanks to Malte Möbius for performing DIC analysis during the cruise. I express my thanks to A. Sy for providing nutrient, oxygen, and hydrographic data. The comments of Peter Croot and two anonymous reviewers greatly helped improve an earlier manuscript. This work was supported by the German Ministry of Education, Science, Research and Technology (BMBF, 03F0108F) and the German Research Foundation (DFG, IT6/9-1). This is NIOZ publication 3692.