The sampling took place from 25 January to 4 February 2008, as part of the PNRA oceanographic survey of the Ross Sea. A total of 60 seawater samples were collected for measurements of pH and A_T. Stations A, H1, L, D and B are located on the continental shelf, stations H2, H3 and G over the slope, and station P offshore (Fig. 1). Information on the stations sampled is in Table 1.

Conductivity–temperature–depth (CTD) casts were performed using an SBE 911plus probe with double temperature, conductivity and oxygen sensors coupled to a SBE 32 Carousel Water Sampler (Sea-Bird Electronics, Bellevue, WA). The CTD was flushed at a constant rate, always below 1m s⁻¹.

Calibration of temperature and conductivity sensors was performed before and after the cruise. Hydrological data were corrected and processed according to international procedures (UNESCO 1988). Standard algorithms (Fofonoff & Millard 1983) were used to compute derived quantities such as potential temperature, salinity and potential density anomaly.

The water samples that were used for the pH and the A_T analyses were collected from 12 L Niskin bottles and put into 500-mL borosilicate glass bottles, using the standard operating procedures described in the DOE handbook (DOE 2007a). The water samples were poisoned with saturated HgCl₂ solution to stop biological activity from altering the carbon distributions in the sample containers before analysis. Samples were stored in dark, cold (+4°C) conditions and analysed at the Department of Chemistry and Industrial Chemistry at the University of Genoa. The analyses were performed within four months of sampling.

The A_T (µmol kg⁻¹ seawater) was measured using a potentiometric titration system with an 806 Exchange Unit (Metrohm, Herisau, Switzerland) and a 702 SM Titrino Dosimat (Metrohm, Herisau, Switzerland). We followed the methods described in the DOE handbook (DOE 2007a), which was adapted to our instrumentation (Rivaro, Messa et al. 2010).

Placed in an open cell, 165 g of seawater was titrated with a solution of hydrochloric acid (HCl) in a two-stage titration. The sample, maintained at constant temperature (25±0.1°C) using a thermostatic water bath, was first acidified to a pH of between 3.5 and 4.0 with a single aliquot of titrant (3.7 mL of HCl 0.1 M), then stirred for 10 min to allow for the liberation of the CO₂ that was evolved. The titration continued until a pH of about 3.0 was reached. The ionic strength of the HCl solution was adjusted with NaCl in order to be similar to that of the samples (salinity range 33.5–34.8). The progress of the titration was monitored using a pH glass electrode/reference electrode cell, and the A_T was computed from the titrant volume and potential difference measurements using a non-linear least-squares approach similar to that proposed in the DOE handbook (DOE 1994). The precision was computed from three replicate analyses of one sample at least daily and was determined to ±3 µmol kg⁻¹ (±0.1%). Routine analyses of Certified Reference Materials (batch no. 79, provided by A. G. Dickson, Scripps Institution of Oceanography) ensured the accuracy of the measurements, which was ±3 µmol kg⁻¹.

The pH was expressed in the pH total scale, (i.e., [H⁺] as moles per kilogram of seawater, pH_T) and it was determined using a potentiometric method that employed a combination glass/reference electrode with a negative temperature coefficient thermistor. The Tris(hydroxymethyl)aminomethane (TRIS) buffer used to standardize the pH electrode was prepared according to the DOE handbook (DOE 2007b). The salinity of the TRIS buffer was 35. Both the TRIS buffer and the seawater samples were brought to the same temperature (25±0.1°C) using a thermostatic water bath, before the measurements were conducted. The accuracy of the pH measure is strongly dependent on the composition of the TRIS buffer solution, but no certified material is currently available. Consequently, this figure of merit was not determined. The precision of the pH measurement was ±0.007 units and was evaluated by repeated analysis of the A_T certified material. The pH_T measures at 25°C were then recalculated at in situ temperature and pressure conditions (pH_{in situ}).

A_T, pH_T, salinity and temperature were used as input parameters in the CO₂-chemical speciation model to calculate pCO₂ (CO₂SYS program [Pierrot et al. 2006]). We used the CO₂-system dissociation constants (K*1 and K*2) estimated by Roy et al. (1993), as a recent paper showed these constants to be the most suitable for cold and fresher surface waters (Chierici & Fransson 2009). We used the Dickson constant for the HSO4- (1990). Following Millero (2007), the estimated probable errors in the calculated parameters of the carbonate system using A_T and pH_T data as input measurements are ±4 µmol kg⁻¹ and ±2 µatm, for both C_T and pCO₂, respectively.

The most direct method to estimate the uptake of atmospheric CO₂ by the oceans is by determining the air–sea flux from values of the pCO₂ in the sea surface and the overlying atmosphere. The net sea–air CO₂ exchange flux (F, mmol m⁻² d⁻¹) is calculated using the formula: 1 F=ks(ΔpCO2) where k is the CO₂ gas transfer velocity (cm h⁻¹), s is the solubility of CO₂ (mol kg⁻¹ atm⁻¹) and ΔpCO₂ is the difference between the pCO₂ of the seawater (pCO_2SW) and the air (pCO_2atm). Negative ΔpCO₂ values indicate absorption of the gas by the sea surface, while positive values correspond to the release of the gas into the atmosphere. Therefore, the sea–air flux of CO₂ is dependent on the difference of CO₂ concentration between sea surface and atmosphere (ΔpCO₂, µatm) and proportional to the wind speed.

The ΔpCO₂ values were calculated as the difference between the pCO₂ in seawater, calculated from our A_T and pH measurements, and the atmospheric CO₂ mean value of 382 µatm available from the South Pole Observatory (http://www.esrl.noaa.gov/gmd/obop/spo/).

The atmospheric pCO₂ was corrected to 100% humidity at in situ sea-surface temperature (SST) and salinity (SSS). Correction for the water vapour pressure was carried out using the equations of Weiss & Price (1980). The water vapour pressure values were then used to correct p(CO₂) dry to 100% humidity using the standard formula as given in DOE (1994) as Standard Operational Procedure 5.

The solubility coefficient s was calculated in each sample as a function of temperature (t) and salinity (S), using Weiss (1974). The gas transfer velocity k, depends on the wind speed u. We applied the Wanninkhof formulation (1992), where k_CO2 =0.31u ² (660/Sc_CO2 )^0.5 for instant wind speed measures or k_CO2 =0.39u ² (660/Sc_CO2 )^0.5 for averaged wind speed. We also applied the Wanninkhof & McGillis (1999) formulation where k_CO2 =0.0283u ³ (660/Sc_CO2 )^0.5. Sc_CO2 is the Schmidt number for CO₂. We used our ship-based wind speed data at 10m height and monthly data provided by the European Centre for Medium-Range Weather Forecast (ECMWF) with a spatial resolution of 1°×1° for the grid points. The wind speeds used in this study are reported in Table 2.

Different water masses were identified in the sampled stations using potential temperature (θ), salinity (S) and neutral density (γⁿ) values (Jackett & McDougall 1997) (Fig. 2; Table 3). Following Whitworth et al. (1998), Orsi et al. (1999) and Orsi & Wiederwohl (2009) we used the γⁿ 28.00 kg m⁻³ and 28.27 kg m⁻³ values to separate the CDW from the AASW and from the densest water masses below, respectively. The six water masses that were identified are described below.

The AASW is characterized by a γⁿ lower than 27.70 kg m⁻³ and by a wide range of temperatures (>−1.0°C) and salinities (<34.20) due to its interaction with atmosphere and ice production and melting. Following the approach introduced by Mitchell & Holm-Hansen (1991) an increase of in situ density >0.05 kg m⁻³ over5 m depth interval was the criterion used to establish the depth of UML in the AAWS. A thin UML was found in TNBp (10–15 m), whereas it became thicker in the other stations, ranging from 23 m (station P) to 34 m (station H3). Our values are in agreement with those reported for the same areas by Massolo et al. (2009).

The CDW is characterized by a γⁿ within the range of 28.00–28.27 kg m⁻³, with a sharp core showing the temperature maximum (>1.0°C).

The Modified Circumpolar Deep Water (MCDW), which is the CDW entering in the Ross Sea and isopycnally mixed with local water, is characterized by the same density range as CDW, but with lower temperatures.

The Deep Ice Shelf Water (DISW) is characterized by potential temperature below the surface freezing point (sfp) and present in front of the Ross Ice Shelf between 160°W and 180° (Orsi & Whiederwohl 2009).

The HSSW is characterized by a salinity maximum greater than 34.70, temperature near the sfp and by a γⁿ greater than 28.60 kg m⁻³.

The AABW is characterized always by a γⁿ greater than 28.27 kg m⁻³ and with potential temperature below 0°C, which is found offshore the 700 m on the slope.

The A_T measured in our samples ranged between 2275 and 2374 µmol kg⁻¹ and had good agreement with previous data collected in the Ross Sea (Sandrini et al. 2007).

Figure 3 shows the A_T as a function of depth at all of the stations. Generally, the A_T profiles in the continental shelf stations had the lowest values in surface water (from 2275 to 2346 µmol kg⁻¹), an increasing gradient reaching the maximum values (from 2343 to 2374 µmol kg⁻¹) at ca.500–700 m, and then lower values at deeper depths. The station P, sampled offshore, showed an increase in A_T values from ca. 1700 m to the deepest sampled depth (ca. 2300 m). At the surface, the highest values were measured at stations L, D and A (2328, 2334 and 2346 µmol kg⁻¹, respectively).

Pair-wise Pearson correlation coefficients were computed to examine the relationships between inorganic C parameters and hydrographic measurements. The A_T and the salinity were significantly and positively correlated (Pearson's r=0.86, number of samples=57), according to the following relationship: 2 AT=68-5 μmol kg-1

No significant correlation were found between A_T and θ when the whole data set was considered. The relationship becomes significantly positive (r=0.62, number of samples 16) when surface and subsurface samples (i.e., 1–10 m depth) are considered, suggesting that the temperature could explain some of the observed A_T variability in the AASW.

pH_T values recalculated at in situ temperature and pressure conditions (pH_{in situ}) ranged between 8.274 and 7.926. pH_{in situ} was significantly and negatively correlated to S (r=−0.71) when the whole data set was considered. There was no correlation between pH_{in situ} and either temperature or salinity when the 1–10 m depth samples were considered.

The pH_{in situ} vertical distribution is shown in Fig. 4. The pH_{in situ} has high and variable values in the surface layer (from 7.890 to 8.033 between 0 and ca.100 m), then an abrupt drop with depth to approximately 100–400 m, followed by a slight increase and nearly constant values. At the surface, the lowest values were measured at stations H2, H3, G, B and P (8.108±0.023), while stations D, L, A and H1 showed a higher variability and the highest pH values (8.186±0.088). The same trend was observed in the subsurface layer too. An increasing gradient of the pH_{in situ} can be observed from the offshore to the Ross Sea shelf stations, particularly at intermediate depths around 300–500 m. The left-most profile in Fig. 5 refers to station H3 sampled in the Glomar Challenger Trough (GCT; Fig. 1), where the absolute pH_{in situ} minimum (7.950) was found in correspondence to the CDW intrusion. Comparable values were measured at station H2, sampled as well as station H3 in the GCT and at station P offshore the Drygalski Trough (DT; Fig. 1), the other well-known intruding area of the CDW in the Ross Sea.

The A_T, pH_T and the hydrographic data were used to calculate C_T and pCO₂. The C_T ranged from 2082 to 2263 µmol kg⁻¹; generally the lowest values were measured at the surface, with the absolute minimum at station H1, while the highest were measured at intermediate depths (ca.500 m) at stations D, P, G, H3 and B. The C_T standard deviation in the AASW was higher than that of A_T (39 vs. 23 µmol kg⁻¹), whereas in the intermediate and deep layers the standard deviations were comparable (6 vs. 5 µmol kg⁻¹, as mean values). The computed pCO₂ ranged from 207 to 479 µatm; the top of AASW (2–20 m) was undersaturated in CO₂ compared to calculated atmospheric values, particularly at stations L, D, A and H1, with pCO₂ varying between 207 and 338 µatm. The maxima were recorded at intermediate depths (ca. 500 m) at stations P and H3. The range of both parameters is in agreement with those reported for the Ross Sea waters (Sweeney 2003; Sandrini et al. 2007).

The general vertical distribution of the inorganic carbon parameters (increasing for the A_T, the C_T and the pCO₂ and decreasing for pH) is mainly due to the biological activities (primary production and respiration, biogenic carbonate precipitation and dissolution), and to the physical properties of the water masses together with the circulation pattern along the whole water column. The AASW occupies the top of the water column. The large standard deviation of the physical and chemical properties of AASW compared to the other water masses was consistent with the high variability of the upper layer of the water column, from the surface down to a depth of more than 100 m, which is strongly influenced by sea–ice–atmosphere interactions and by biological activity particularly in the summer season (when the sampling was performed).

Processes such as melting and formation of sea ice can play an important role in controlling the AASW physical features and inorganic carbon chemistry. In 2008, during the cruise, the sea-ice coverage was minimal and in the Ross Sea wide open sea areas were present (see satellite images at http://www.usap.gov/videoClipsAndMaps). Solar heating and decreased salinity, due to pack-ice melting, were responsible for the formation of the shallow UML in the AASW. To evaluate any effect of the sea-ice melting on the A_T and C_T distribution at the surface, following Rivaro et al. (2011), we calculated the meltwater percentage (MW%), considering the difference between the salinity measured at the surface (S_meas) and the salinity measured in the same station at greater depth (S_deep), which was not influenced by sea-ice dilution, assuming an average sea-ice salinity of 6 (Ackley et al. 1979): 3 MW%=(1-Smeas-6Sdeep-6)*100

In our study, S_deep depths were chosen on the basis of their γⁿ values; that is, we chose samples collected below the γⁿ 28.00 kg m⁻³, which is the lower boundary of the AASW (Table 4).

MW% ranged from 1.1 (station P) to 4.1 (station A); the stations sampled in the GCT (H1, H2, H3) had ca. 3% MW, while stations L and D, sampled at TNBp had 1.2 and 1.8 MW%, respectively. The 2008 MW% fell into the range of values observed in 2006, which was considered as a standard year for sea-ice coverage in the Ross Sea shelf area (Massolo et al. 2009). The highest MW% calculated for stations H1, H2, H3 was consistent with the sea-ice coverage shown by satellite maps before the sampling period, showing a higher pack-ice extension in the GCT than in the polynya area. The MW% co-varied significantly and negatively with A_T (Pearson's r=−0.79, P=0.05) and C_T (Pearson's r=−0.62, P=0.05). This is consistent with the fact that the distribution of surface A_T and C_T is controlled by the factors that govern salinity. Therefore, the addition of sea-ice meltwater, which has low salinity, results in a dilution of the A_T and C_T. Unlike C_T, pH and pCO₂, photosynthesis and oxidation of organic matter have little direct influence on A_T (Redfield et al. 1963). Calcifying organisms can influence A_T and C_T concentration, but it is well known that the phytoplankton in the southern Ross Sea is dominated by two taxa that are spatially distinct: Fragilariopsis curta, a pennate diatom, dominating the coastal region, and P. antarctica, a colonial prymnesiophyte, dominating the south central region (Smith & Nelson 1985; Smith et al. 1996; Arrigo et al. 1999). Both are not calcifying organisms and therefore they are not able to influence A_T and C_T. Gordon et al. (2000) also reported that A_T does not vary during the summer in the Ross Sea, suggesting that phytoplanktonic calcifying organisms are unimportant in the carbon cycle in this region.

A correlation between the surface A_T and temperature values was found in our study, suggesting that the temperature could explain some of the A_T variability, consistent with the dependence found for surface waters of the world ocean. In particular, Lee et al. (2006) proposed several relationships to estimate A_T from salinity and temperature surface measurements. We applied the relationship proposed for the Southern Ocean to our data set: 4 AT=2305+52.48(SSS-35)+2.85(SSS-35-20 where SSS and SST are sea-surface salinity and temperature, respectively.

This relationship gave poor results, because the analysis of the residual, which is a good measure of the accuracy of the predictivity of the curve fitting, was about 11 µmol kg⁻¹, higher than the accuracy of our measurements (3 µmol kg⁻¹). Afterwards, a new similar relationship that provided more accurate estimates of A_T from the measured SSS and SST for the surface waters of the Ross Sea was found. To find the new relationship we used the surface samples collected during the 2003 Climatic Long Term Interaction for the Mass-balance in Antarctica (CLIMA) cruise (n=82; Sandrini et al. 2007). The results were: 5 AT=2308+108.98(SSS-34)+60.70(SSS-34 Equation 5 was then applied to our samples; the analysis of the residual was 4 µmol kg⁻¹, confirming its validity for the Ross Sea and showing that we could estimate the A_T from SSS and SST in the Ross Sea AASW with a precision comparable to our measurements.

The variability of the pH_{in situ} in the surface layer can be ascribed primarily to biological production and vertical mixing. Primary production increases the pH of seawater as a result of the displacement of the carbonate equilibrium related to CO₂ consumption, while the vertical mixing of the surface water with deeper water causes a decrease of the pH. The correlation between pH_{in situ} and S or θ was not significant in the AASW samples, suggesting that physical variables do not have a strong direct influence on the distribution of this parameter at the surface, which is therefore mainly governed by biological processes. The pH_{in situ} was negatively and significantly correlated to C_T, while it was not correlated with A_T. The highest values were observed in stations D, H1 and A, which coincided with low C_T values (2161, 2082 and 2158 µmol kg⁻¹, respectively) and a slightly higher A_T:C_T ratio (1.09), suggesting that C_T plays a more significant role in regulating pH than A_T. The exchanges of CO₂ with the atmosphere can result in a modification of pH as well, but this effect is weak over the short time scale that we investigated.

All pCO₂ values of the surface and subsurface waters (2–20m) were below atmospheric levels, particularly at stations L, D, A and H1, with pCO₂ varying between 207 and 338 µatm. Considerable variation was, however, observed spatially (Table 5). Consequently, the ΔpCO₂ (pCO_2SW–pCO_2air) was negative (from −44 to −175 µatm), outlining that the Ross Sea was a potential sink for CO₂ during the sampling (Fig. 5). In surface seawater the pCO₂ depends on SST, SSS, C_T and A_T concentrations. These determinants are influenced by several processes such as warming/cooling and biological activity (Bellerby et al. 2004). Even though the pCO₂ is influenced by the SST and SSS, no relationship between pCO₂ and physical variables was found in our samples collected in the surface waters. This agrees with the results obtained by Poisson et al. (1993) and Metzl et al. (1999), who did not find a large-scale correlation between fCO₂ and SST in the Southern Ocean during the austral summer. Laika et al. (2009) confirmed that biological effects dominate the temperature effect on the pCO₂ distribution, especially in the areas south of the Polar Frontal Zone, from the Permanent Open Ocean Zone to the Antarctica shelf areas. The Ross Sea is one of the most productive areas of the Southern Ocean. Marginal ice zones and polynyas are often associated with high primary productivity, which is the result of higher light availability (due to the reduced sea-ice cover) and higher micronutrient availability (in particular, Fe; Tagliabue & Arrigo 2006). The RSp (Fig. 1) is the most productive polynya, but the TNBp, which is smaller than the RSp, is also important in terms of productivity (Mangoni et al. 2004). The phytoplankton bloom develops later in the year at the TNBp compared to RSp and it is dominated by diatoms. Phytoplankton abundance is maximal in late December, declines thereafter, but a secondary peak appears in mid-February (Tremblay & Smith 2007). In our survey, the highest pCO₂ drawdown was calculated in the TNBp and RSp and in station H1, which was sampled in a marginal ice area. Unfortunately, we could not directly confirm the phytoplanktonic drawdown on the pCO₂, distribution, because neither chlorophyll-a sampling nor fluorescence measurements were performed. Nevertheless, primary productivity calculated from Sea WiFS data referred to the investigated period (available at http://giovanni.gsfc.nasa.gov) and the comparison with data collected in previous surveys allowed us to hypothesize that the phytoplanktonic drawdown made an important contribution in determining the pCO₂ in the upper AAWS (Bates et al. 1998; Gordon et al. 2000; Sweeney et al. 2000; Takahashi et al. 2002; Barbini et al. 2003; Sweeney 2003; Arrigo & Van Dijken 2007). In particular, Barbini et al. (2003) correlated the phytoplanktonic bloom events with the high chlorophyll-a concentrations and low pCO₂ values measured in selected sites during the XVI Italian Expedition (austral summer 2000–01) such as TNBp (corresponding with stations D and L), the RSp near Ross Island (corresponding with station A), and in the central basin area (corresponding with station H) (Fig. 1). Conversely, high values of surface pCO₂ were encountered towards the northern margin of the Ross Sea shelf area above 73°S latitude, in the offshore area near Cape Adare and Cape Hallet where stations G and P were sampled in our survey (Fig. 1). Other evidence corroborating the hypothesis that primary production was playing an important role in regulating CO₂ distribution in the Ross Sea during austral summer 2008, could come from our O₂ measurements (data not shown). The O₂ concentration was compared with oxygen saturation (calculated from S and T) and a disequilibrium in O₂ was observed in TNBp and RSp. The undersaturation in CO₂ coincided with a supersaturation in O₂, which could not be explained by a change in the hydrography. Finally, the pCO₂ and the pH_{in situ} correlated negatively and significantly in the AASW, consisting with the increase of the pH due to primary production as a result of the displacement of the carbonate equilibrium related to CO₂ consumption.

The sea–air fluxes of CO₂ were calculated (Table 5). Wind speed is the main driver of the air—sea flux, together with the ΔpCO₂. Both wind speed data (shipboard and ECMWF data) used to calculate the kCO₂ showed a marked spatial variation (Table 2), which greatly influenced the calculated CO₂ fluxes. Since the relationship between u and kCO₂ is still a source of uncertainty, we have applied different formulations. As a consequence of the negative ΔpCO₂ values, the air—sea CO₂ fluxes were directed towards the water column and no upward CO₂ flux was calculated. Significant variability in the CO₂ sea–air flux was observed, with mean values ranging from −0.3 to −15.4 mmol m⁻² d⁻¹. Although the differences, resulting from the choice of exchange coefficient and the chosen non-linear relationships between wind speed and gas transfer velocity were large, our range is comparable to those calculated by Bates et al. (1998) and Sweeney et al. (2000), who used long-term average wind speeds. The large flux uncertainty comes from the error in ΔpCO₂, the errors in the scaling factor for the gas transfer rate formula and the wind speed variability. Using shipboard wind data, the highest mean flux was calculated at station D, where high drawdown of CO₂ by biological activity was supposed. Using ECMWF wind data (lower than shipboard data; Table 2), the estimated flux resulted in being three times lower. When fluxes were calculated using the cubic relationship between wind speed and CO₂ flux, the sink observed at TNBp became even weaker. In any case, it is important to note that our estimates are a snapshot only, describing the situation that we found during our cruise so that the mean fluxes should only be taken as indicative; therefore, any extrapolation to a seasonal flux should be treated carefully.

On the basis of the thermohaline properties and γⁿ values, several intermediate and deep-water masses were identified in this study. Afterwards, the values of the chemical parameters were constrained by the physical values used to differentiate the local water masses. Differently from AASW, the physical and chemical properties varied within a relatively narrow range.

CDW is a relatively warm and salty water mass that originates in the Antarctic Circumpolar Current (ACC) and episodically intrudes into the Ross Sea on the shelf, forced by the curvature of the bathymetry (Dinniman et al. 2003). On the shelf, isopycnal mixing with cold and low-salinity AASW rapidly attenuates the characteristics of the CDW offshore, producing MCDW (Jacobs & Giulivi 1999). CDW or MCDW is thought to be important to many processes, such as the formation of AABW (Orsi et al. 1999; Gordon, Orsi et al. 2009; Orsi & Wiederwohl 2009), the dynamics of the RSp (Reddy et al. 2007), and supplying much of the heat involved in the basal melting of the Ross Ice Shelf (Smethie & Jacobs 2005). CDW contains low levels of O₂ and high levels of nutrients (nitrate, NO₃, phosphate PO₄ and silicate Si(OH)₄) and of C_T, as it has an older ventilation age compared to the recently formed SW (Rivaro et al. 2004; Sandrini et al. 2007; Rivaro, Massolo et al. 2010).

The mid-depth (ca.500 m) intrusion of the CDW and its modification in MCDW was traced over the shelf by the high θ, C_T and pCO₂ values (>−0.8°C, 2251±7 µmol kg⁻¹ and 463±14 µatm, respectively) and by the low pH_{in situ} (7.969±0.025), particularly at stations P and H3 (Fig. 5). These stations were located along the slope in the two well documented sites of CDW intrusion in the Ross Sea shelf area, the DT and the GCT, respectively (Orsi & Whiederwool 2009). The re-mineralization of organic matter consumes O₂, further increasing the nutrient, C_T and pCO₂ concentrations. The pH decreases due to the addition of CO₂ from the oxidation of organic carbon. The high C_T and the low pH_{in situ} characterizing the CDW are consistent with the old ventilation time, the low O₂ and the high nutrient values. For this last feature, the intrusion of CDW on the shelf could represent a source of nutrients for biological production.

The HSSW is defined by γⁿ >28.60 kg m⁻³, S >34.70 and temperatures close to the sfp; it is formed along Victoria Land and mainly in the TNBp area by brine rejection during surface water freezing and the strong katabatic winds flowing down to the Antarctic plateau (Budillon & Spezie 2000). The HSSW spreads northward along the DT, where it reaches the shelf break and ventilates the deep ocean producing AABW. Another branch moves southward beneath the Ross Ice Shelf taking part in the production of the DISW (Jacobs et al. 1985). DISW is defined by γⁿ=28.40 kg m⁻³, salinity higher than 34.45 and temperatures close to the sfp. The DISW core moves northward along the GCT, reaching the shelf break near 75°S close to the dateline meridian where it also mixes with the intruding CDW taking part in AABW production (Budillon et al. 2003; Rubino et al. 2003; Smethie & Jacobs 2005; Budillon et al. 2006). Both HSSW and DISW are characterized by high O₂ concentrations (about 295 µM), reflecting the origin of AASW in the case of HSSW and the contribution of glacial meltwater in the case of DISW. The HSSW was characterized by the A_T maxima (2367±6 µmol kg⁻¹), measured in the entire water column below 100–200 m in the D and L stations in the layers deeper than 500 m in the A, H1, B and G stations, which were sampled in correspondence to the preferential pathways of the spreading HSSW. The high correlation between A_T and S and the conservative behaviour justify the fact that the highest A_T values are found in the deeper layers of the water column measured at stations L and D sampled at TNBp, which is the main site of HSSW production in the Ross Sea.

The DISW was identified by θ and S values at intermediate depths (between 400 and 500 m) in H1, and it was characterized by lower A_T than the HSSW (2341±4 µmol kg⁻¹). This difference is due to the dilution effect produced by glacial meltwater addition in the DISW production. Therefore, the A_T adds some information to the chemical characterization of the water masses, being able to chemically trace the two water masses. The same feature is shown by C_T too; that is, HSSW has a C_T higher than DISW (2251±8 and 2228±3, respectively). Taking into account that both SWs take part in the Ross Sea AABW production, and the role played by AABW as a CO₂ sink we can suppose that the plume produced in the DT injects a higher C_T into the deep ocean than the plume leaving the GCT.

According to the definition by Orsi & Wiederwohl (2009), the AABW lies on the slope at depth >700 m, it is always characterized by a γⁿ greater than 28.27 kg m⁻³ and by θ<0°C. In our data set these characteristics were shown by the deepest layers (depth>1850 m) at station P only, sampled in the DT mouth. No AABW was identified in the samples collected in the GCT, although this area is a well-known AABW production site as documented by several CTD and moored station measurements (Smethie & Jacobs 2005; Budillon et al. 2011). In fact the deepest layers of station H3, (depth>700 m) do not have θ<0°C and γⁿ >28.27 kg m⁻³ (Fig. 2, Table 3). For this reason, we cannot evaluate whether the AABW produced in the two regions of the Ross Sea contains different amounts of C_T and as a consequence whether the plume produced in the DT injects a higher C_T into the deep ocean than the plume leaving the GCT.

The precipitation or formation of solid CaCO₃ in surface waters and the dissolution of solid CaCO₃ in deep waters can be important in transferring CO₂ from surface to deep waters. The uptake of CO₂ by surface layers increases the concentration of H⁺, which lowers pH and, in changing the chemical equilibrium of the inorganic carbon system, reduces the concentration of CO₃ ²⁻. Carbonate ions are required by marine calcifying organisms such as plankton, shellfish and fish to produce calcium carbonate shells and skeletons.

The saturation state of the seawater with respect to CaCO₃ is expressed by the ratio 6 Ω=[Ca2+][CO32-]/Ksp'

where [Ca²⁺][CO₃ ²⁻] is the ion product of the concentration of calcium and carbonate ions, and K′_sp is the solubility product depending on in situ temperature, salinity and pressure and on the particular mineral phase (aragonite is approximately 50% more soluble than calcite [Mucci 1983]). The values of [CO₃ ²⁻] and of Ω have been determined from A_T and pH, using the CO₂SYS program. All of the samples were oversaturated with respect both to the calcite and to the aragonite, with the exception of the deepest layers (below 1600 m) of P station, which were undersaturated with respect to the aragonite. Near corrosive levels of aragonite saturation state (Ω_arag) (Ω ca.1.1–1.2) were found in correspondence to the entrainment of CDW both in the DT and in the GCT. Our data are in agreement with Ω_arag and pH values and variability in the Ross Sea calculated by McNeil et al. (2010). It has been predicted that the Southern Ocean will become undersaturated with respect to aragonite by 2030 (McNeil & Matear 2008). The onshore transport of CDW, whether sporadic and occurring only at specific locations, could accelerate the Ross Sea acidification process, with respect to these predictions. Aragonite undersaturation is of particular concern for the zooplankton species comprising pteropods, which form aragonite shells. In fact, pteropods comprise up to one-quarter of the total zooplankton biomass in the Ross Sea (Hopkins 1987). Recently, Manno et al. (2007) described the occurrence of initial dissolution of the aragonitic shells of the pteropod Limacina helicina above the saturation depth in the TNBp area, while Honjo (2004) reported pteropod shell degradation at a depth of 1 km and below south of the Polar Front Zone. Bednaršek et al. (2012) comparing the shell structure of specimens L. helicina, that were extracted live from the Southern Ocean, with samples from aragonite-supersaturated regions, found severe levels of shell dissolution in the Southern Ocean organisms. All of these results suggest that changes in carbonate chemistry in the water column are already occurring and that they could possibly have significant effects throughout the Ross Sea marine food web.