Surface layer balance of the southern Antarctic Circumpolar Current (prime meridian) used to derive carbon and silicate consumptions and annual air-sea exchange for CO2 and oxygen

A simple model, using concentrations of nitrate and phosphate in austral winter 1992, reveals that the Antarctic Surface Water (AASW) of the southernmost Antarctic Circumpolar Current (ACC) between the Southern ACC Front and the Weddell Front is made up of about 90% Upper Circumpolar Deep Water (UCDW) and 10% northward-flowing AASW from the Weddell Gyre. With a typical time scale of about 1 year, the upwelling velocity was calculated to be as high as 60-100 m y-1. Knowing the composition of the surface water with respect to its sources, changes due to several processes in the surface layer were deduced for carbon dioxide, oxygen and silicate. As the time scale of changes in the surface layer of the southern ACC is about 1 year, this allows us to calculate changes on an annual basis without interference of short-term variations. Balancing the contributions by upwelling, biological activity and air-sea exchange to the concentrations in the surface layer, the area was found to be a large sink for atmospheric oxygen of 6.0 mol m-2 y-1 (53 µmol kg-1) and a small sink for atmospheric carbon dioxide of 1.0 mol m-2 y-1 (9 µmol kg-1). The most important cause for the oxygen sink is the upwelling of oxygen-poor UCDW, which surpasses the oxygen-elevating effect of primary productivity. This large oxygen sink, in between areas to the north and south which are only a small sink or even a source, conforms with the latitudinal distribution of atmospheric oxygen. The small CO2 sink is largely brought about by biological activity. The annual carbon utilization amounts to 76 ± 22 g C m-2 y-1, which is relatively high for an open ocean region in the Antarctic. However, it supports recent estimates of primary production of the Antarctic Ocean that are higher than early published values. The annual silicate consumption was calculated to be 126 ± 19 g Si m-2 y-1. This is considerably higher than the Southern Ocean mean in current estimates. Although the southernmost ACC may be atypical for the Southern Ocean, the current estimate for Southern Ocean silica production may well be an underestimation. The silicate to carbon utilization ratio derived here is 0.53 which aligns with investigations on Antarctic phytoplankton and thus underscores the consistency of our results.


Introduction
The Antarctic Circumpolar Current (ACC) is the largest coherent current structure in the world ocean. It connects all major oceans and acts as the prime water distributor for the ocean basins. The principal water mass of the ACC is the Circumpolar Deep Water (CDW) which originates from the where biogenic silica (opal) accumulation occurs [DeMaster, 1981]. Most of the ACC is thought to be less productive [Jacques, 1989;Banse, 1996], although it is persistently repleted in major nutrients. This renders the ACC one of the most prominent High Nutrient Low Chlorophyll (HNLC) regions. Estimates of the total production of organic matter in the Antarctic are accordingly low, which would imply a rather low drawdown of carbon dioxide by phytoplankton production in this region. On the other hand, the frontal systems, especially the Polar Front [Lutjeharms et al., 1985;De Baar et aL, 1995] and also the southern boundary [Tynan, 1998], have been identified as areas of enhanced phytoplankton productivity.
Many investigations have emphasized the highly variable characteristics of the carbon dioxide distribution in the surface layer of the ACC [e.g., Poisson et al., 1993;Bellerby et aL, 1995;Bakker et al., 1997]. The difference in CO 2 partial pressure (pCO2) between the atmosphere and the surface ocean, which governs the air-sea exchange, is dependent on many physical and biological factors such as sea surface temperature, wind action, and phytoplankton growth. A variable physical and biological environment in the ACC causes a large variability of surface water CO2. As long as the extent and rates of these processes and the factors that influence them are not sufficiently known, it will be impossible to satisfactorily assess the air-sea flux of CO2 on an annual basis from the few pCO2 measurements. In this study we apply a different, indirect approach for obtaining the CO2 flux between the ocean and the atmosphere. All possible CO2 changes in the surface layer of the southern ACC were balanced to give the annual fluxes of the CO2 air-sea exchange. Previously, this method has been applied to the Weddell Sea . This approach also yields an estimate of the annual carbon utilization (primary productivity). We also apply the concept to oxygen and silicate data, which yields estimates of the annual air-sea flux of oxygen and the annual silicate utilization.

Sampling and Methods
Data presented are from cruise ANT X/4 of the German research vessel FS Polarstern during early austral winter in June 1992 [Lemke, 1994]. All stations in the present study are situated near or on the prime meridian between Africa and Antarctica (0 ø longitude; Figure 1). Water for the determination of oxygen, total CO2, silicate, nitrate, and phosphate was taken from a rosette system with 24 Niskin samplers (12 L) coupled to a conductivity-temperature-depth (CTD) .

Results
In Figure 2 vertical sections of the prime meridian extending to 1500 m depth are shown for a suite of parameters. To the north, this section is bordered by the Subantarctic Front. The section crosses the Polar Front, the Southern ACC Front, and the Weddell Front (ACC-Weddell boundary) as described by Orsi et al. [1995]. All water of this section flows eastward as part of the ACC (see Spiridonov et al. [1996], in which some hydrographic data of cruise ANT X/4 are shown). The surface layer consists of Antarctic Surface Water. It has relatively low concentrations of TCO2 and nutrients and high concentration of oxygen. The properties of the Antarctic Surface Water (AASW) along this transect have a wide range of characteristics, with clear gradients between the northern part (ACC) and the southern part (Weddell Sea) for all parameters (Figure 2). Northward of the Polar Front (PF), the isolines that describe the surface water of the south bend downward. This illustrates the formation of Antarctic Intermediate Water (AAIW). At the northern end of our section (Figures 2c and 2g) Table 1

Modification of Surface Layer Properties
Since the fractions of source water masses of the AASW of the southern ACC are known, they can be used to estimate the conservative part of the concentrations of other properties. By difference with the actually measured concentrations then (Table 1), the nonconservative part is obtained. The nonconservative contribution of silicate, 02 and TCO2 is due to biological changes, whereas for the latter two gases, exchange between ocean and atmosphere contributes additionally. When silicate concentrations from Table 1  where the terms denote the same as in equations (1) Thus, through the combined action of biological processes and air-sea exchange the ACC surface layer lost 47 gmol kg-1 and gained 144 gmol kg-1 of TCO2 and 02, respectively.

A(TCO2)bio and A(O2)bio can be computed from A(NO3)bi o and A(PO4)bio as obtained from equations (1)-(3), using the appropriate Redfield ratios of conversion.
It is not feasible to derive C:N and C:P ratios from the in situ water masses (like for the N:P ratio, Figure 4a) because air-sea exchange of CO2 and 02 and mixing interfere with this method. This is illustrated in Figure 4b by a plot of 02 versus nitrate for this area, which is highly scattered.

Sensitivity of Calculations
Especially since surface waters are involved, the question as to the validity of the calculations is justified. The resulting value for FUCDW from equations (1) and (2) is crucial for all further calculations. Clearly, FUCDW is dependent on the concentrations of the end-members as well as on the ratio of biological changes of nitrate to phosphate (equation (3)). Possible errors due to dilution effects of the surface water by precipitation or melting of ice were removed by normalizing the concentrations to salinity 35. The nitrate to phosphate ratio in our study (15.1) is lower than the classical Redfield ratio of 16, but it is within the error range given by Anderson and Sarmiento [1994], and ocean wide values by Fanning [1992] and Tyrrell and Law [1997] are also closer to 15 than to 16. A similar lower nitrate to phosphate ratio of 15.2 was found in the surface water of the nearby western Weddell Sea This shows that the derived atmospheric exchanges are little sensitive to errors in the end-members of mixing. Hence even using the lowest probable FUCDW generates values for A(TCO2)atm and A(O2)atm that are consistent with the above result that the southern ACC is a small sink for atmospheric CO2 and a relatively large sink for atmospheric 02. This strongly hints that the actual errors in our estimates for A(TCO2)atm and A(O2)atm are much smaller than the formal errors. Below we give more indications for this contention.

Timescale and Upwelling Velocity of UCDW In the above balance calculations, changes in the Antarctic
Surface Water of the southern ACC were identified which are caused by mixing as well as by biological and physical (airsea exchange) processes. Assigning a time frame during which these processes take place would allow one to derive process rates, such as the upwelling velocity of UCDW. This is, however, somewhat problematic, but in the following it is shown that we can provide a fair estimate. The fact that our analysis comes to a result that compares well with other independent methods for determining the upwelling velocity puts much confidence in the method we used and the assumptions we made. This then not only holds for our calculated upwelling velocity but also for other results that derive from our analysis because upwelling and changes due to biological activity are closely intertwined and mutually dependent. It is evident that a large extent of upwelling, tending to raise the concentrations of the nutrients and TCO2 in the surface layer, should be compensated by intense biological activity, oppositely drawing down these species.

Balance background
At this point, some remarks as to the background of the balance calculation and its results are in order. We calculated the fractions of the sources of the AASW of the southern ACC and derived a typical timescale of 1 year for the processes determining the composition of this AASW. The derived concentration changes of nutrients, TCO2, and oxygen due to the different causes (section 4.2) are the net changes on an annual basis. It is important to realize that reliable net changes can only be obtained when the processes considered are steady, i.e., not active. In particular, biological activity should be resting, i.e., at a minimum. If this were not the case the concentrations of the biologically mediated nutrients would for a smaller or larger portion be caused by seasonally varying primary productivity or remineralization. The calculated change due to biological activity would in that case be an unknown composite of the net annual change and the actual change due to active biology.
Our early austral winter data are perfectly suited in the light of such deliberations since the level of biological activity is close to its annual minimum [Spiridonov et al., 1996]. Another asset of our analysis method is that knowledge of seasonal variations in the southern ACC is not required because the surface layer of the southern ACC acts as a memory for the net changes of the previous year. This holds both for the surface layer concentrations as well as for the mixed layer depth. An exception is the seasonal behavior of the nutrient concentrations of the inflowing AASW from the Weddell Sea: Such inflow occurs throughout the year, but in summer the nutrient concentrations are lower than in winter. In section 4.3 the effect of using summerlike nutrient concentrations on the budget calculation was found to be small.

CO2 and Oxygen Sinks
This study found the southernmost ACC to be a small sink for atmospheric CO2 and a very large sink for atmospheric oxygen on an annual basis. Here these results are discussed in terms of physical and biological processes. In addition, we substantiate our results by comparing them with studies using different methods.
Upwelling of UCDW tends to increase the CO2 content of the surface layer and thus also the partial pressure of CO2 (pCO2), which may lead to supersaturation with respect to the atmospheric pCO2. We estimate the pCO2 for UCDW to be -500 laatm, using TCO2 of 2260 I. tmol kg-1 (Table 1) and alkalinity derived from the relationship alk,(35/salinity) = 2386 gmol kg-1, as given by Poisson and Chen [1987] for this region. This would lead to outgassing of CO2 to the atmosphere. However, as we found the southern ACC to be a sink for atmospheric CO2, an annual mean undersaturation must be prevalent in the surface layer. Two mechanisms promote the reversal of the trend toward supersaturation in the surface layer. First, upon upwelling the temperature and salinity decrease by >3øC and -•0.7 salinity units, respectively. On the basis of these changes only we calculate the reduction of pCO2 to be 70-90 gatm. Second, the decrease of TCO2 due to biological activity, amounting to 56 I. tmol kg-1 (section 4.2), would reduce the pCO2 of the upwelled UCDW by -•165 laatm, which in itself is enough to cause undersaturation. This suggests that for establishing the CO2 sink in the southern ACC, biological drawdown of CO2 is more important than physical processes.
Can we find any indications from other studies that may corroborate our results? From pCO2 measurements a qualitative picture may be obtained, which could illustrate the relative importance of certain processes in different seasons.
Unfortunately, the temporal and spatial coverage of pCO2 data in this region is so scarce that calculated annual air-sea fluxes are not reliable. In early austral spring, pCO2 in the southern ACC at 6øW was close to full or slightly below saturation [Bakker et al., 1997]. This reflects the onset of photosynthesis. In summer a substantial undersaturation of CO2 is found for the southern ACC in most of the Atlantic sector of the Southern Ocean [Takahashi et al., 1993]. This is obviously the time of year that CO2 is being incorporated into organic matter. In austral autumn a small supersaturation of CO2 was observed in our area of investigation [Hoppema et al., 2000], which can be explained by a temporal preponderance of upwelling of TCO2-rich UCDW, possibly accompanied by some remineralization activity. Winter data from the South Atlantic [Takahashi et al., 1993] reveal both supersaturation and undersaturation of CO2. In summary, pCO2 data suggest that CO2 uptake from the atmosphere probably occurs mainly in summer. This region has the potential to take up substantial amounts of CO2 in summer because even in summer the wind speeds are relatively high, wind action promoting the CO2 exchange between the air-sea interface.
As regards 02, a similar calculation can be performed. UCDW has a low 02 concentration (Table 1), the percentage of saturation being only 53-54%. Consequently, upwelling produces a marked undersaturation of oxygen in the surface layer. This undersaturation is even reinforced by cooling and freshening of the upwelled UCDW, which would bring down the saturation level to -48%. These processes are counteracted by biological activity, which increases the 02 concentration by 91 •mol kg-1 (section 4.2). This 02 increase would bring the upwelled UCDW to -80% saturation (without cooling/ freshening) or 75% (including cooling/freshening). Appar-ently, the O2-reducing effects of upwelling and cooling/ freshening far outdo the 02increasing effect of primary production in the southern ACC. The large 02 undersaturation of the UCDW is the most important factor for the large 02 sink of the southern ACC.
The 02 uptake in the southern ACC calculated in this study is much larger than that in the Weddell Sea [Hoppema et al., 1995]. The main reason is the much larger transfer of 02poor water into the surface layel' in the southern ACC than in the Weddell Sea owing to the much higher upwelling velocity in the southern ACC. North of our area of investigation (in particular, north of the Polar Front), upwelling of deep waters into the local surface waters does not occur. As the upwelling of O2-poor deep water is a major trigger for the uptake of large amounts of atmospheric 02 by the ocean, we surmise that the ocean in this northern region is a source of 02 due to biological activity. The composite picture emerges of a regional distribution of air-sea exchange of 02 with a precipitous peak of 02 uptake by the southern ACC in between two areas where the 02 uptake is moderate or absent. This picture is perfectly in line with the regional distribution of the atmospheric 02 concentration between the South Pole and Tasmania [Stephens et al., 1998] year is typically the same timescale on which the biological cycle in the surface layer takes place. This leads to interference between these two causes of nutrient changes in the surface layer, with the consequence that the our annual carbon utilization does no longer represent the pure new production. Instead, it represents a lower boundary for the total annual primary production. Our annual carbon utilization, which represents an open ocean value, is much higher than traditional, incubation-based estimates of the primary production in open ocean Antarctic waters, which are <20 g C m-2 yr-1 [e.g., Jacques, 1989]. However, also previously elevated primary production was reported for the southern ACC [Darner and Mordasova, 1994;Tyrian, 1998]. In addition, on the basis of a bio-optical algorithm using satellite (Coastal Zone Color Scanner) observations, a Southern Ocean mean primary production of >100 g C m-2 yr-1 was reported [Arrigo et al., 1998]. The latter value compares fairly well with our estimate, and thus our study supports the notion of a moderately high mean primary production of the Southern Ocean, rather than an oligotrophic region with very low productivity. Further supporting this view is a modeling study indicating the Antarctic Ocean as a significant sink of atmospheric CO2 mainly due to the action of the biological pump [Louanchi et al., 1999].
Given the current debate around the mean level of primary production in the Southern Ocean, it is difficult to decide whether the southern ACC represents an area of elevated productivity or just an average region within the Southern Ocean. We may, though, try to explain the primary production in the southern ACC in the light of prevalent biotic and abiotic conditions. Biological studies suggest that the southern ACC is a region with low productivity, characterized by the absence of phytoplankton blooms [Bathmann et al., 1997]. However, blooms have been reported in the region, even in winter [Dieckmann, 1987], and according to Tyrian [1998], the southern ACC is characterized by high productivity. Also during our early winter cruise ANT X/4, horizontal chlorophyll a and biogenic silica maxima were observed in the southernmost ACC near the Weddell Front [Spiridonov et al., 1996] The silicate to carbon utilization as derived in the present study amounts to Si:C -0.71 mol mol-1, and accounting for the fact that diatoms constitute only 75% of the total phytoplankton production [Tr•guer et al., 1995], this is reduced to Si:C -0.53 mol mol-1. This is much higher than the 0.13 mol mol-1 found for diatoms in nutrient-replete conditions [Brzezinski, 1985]. However, our elevated ratio is in keeping with evidence that diatoms produce 2-4 times as much silica per unit carbon in the Southern Ocean as in other ocean regions [Nelson et al., 1989;Leynart et al., 1991], which is due to iron limitation [Takeda, 1998]. This consistency further increases the reliability of our results. Previous investigations on silicate have shown that the southernmost region of the ACC is an outstanding region [Van Bennekorn et al., 1988]. The sediments in this region are rich in biogenic opal [DeMaster, 1981 ], in contrast with areas farther south such as the Weddell Sea. A high production of biogenic silica (opal) by diatoms in the surface layer in general terms favors the transfer of opal tests to the seafloor. Thus the high silica production in this area could be a crucial factor for the occurrence of extensive silica-rich sediments.

Conclusions
We applied a new approach using nitrate and phosphate and their Redfield ratio to derive the composition of the Antarctic Surface Water of the southern ACC. For calculations involving surface waters these nutrients are very useful indeed because they have a negligible atmospheric component and a stable Redfield ratio. It is essential, though, that winter concentrations are used to exclude spario-temporal effects. This may be a disadvantage since winter data are generally hard to obtain. The UCDW fraction of 90% for the AASW of the southern ACC points to a large upwelling velocity. Since intensive upwelling transports large amounts of nutrients and CO2 into the surface layer, the biological utilization of these elements is accordingly high. The contribution of northward flowing AASW from the Weddell Gyre is relatively small with 10%. This suggests that exchange of surface water across the southern boundary of the ACC is restricted.
We found a large oxygen sink for the southern ACC, but the formal error in this estimate is large. There are strong indications (section 5.3) that this sink is real, and thus the actual error is smaller. The significant oxygen sink conforms with the latitudinal distribution of atmospheric oxygen. Also, the mean saturation level of the surface water after accounting for upwelling and cooling/freshening is definitely below 100%. The major cause of the oxygen sink is upwelling of strongly undersaturated UCDW. A small CO2 sink with a large error was derived for the southern ACC. Also in this case the real error is much smaller, and we assess that the southern ACC is neutral or a slight CO2 sink. UCDW upwelling tends to raise CO2 in the surface layer, and this is mainly counteracted by biological activity.
Annual biological utilizations of carbon and silicate as derived here are significantly higher than most Southern Ocean estimates in the literature. Our carbon utilization supports recent elevated mean primary production estimates for the Southern Ocean. Our silicate to carbon utilization ratio agrees well with data from phytoplankton investigations showing much higher values in the Southern Ocean than in other oceanic regions. It would suggest that the silica production in the Southern Ocean is substantially underestimated by Tr•guer et al. [1995]. In fact, raising the Southern Ocean primary production would indeed also increase the latter estimation of silica production.