Article | Open | Published:

Climatic modulation of surface acidification rates through summertime wind forcing in the Southern Ocean

Nature Communicationsvolume 9, Article number: 3240 (2018) | Download Citation

Abstract

While the effects of the Southern Annular Mode (SAM), a dominant climate variability mode in the Southern Ocean, on ocean acidification have been examined using models, no consensus has been reached. Using observational data from south of Tasmania, we show that during a period with positive SAM trends, surface water pH and aragonite saturation state at 60°–55° S (Antarctic Zone) decrease in austral summer at rates faster than those predicted from atmospheric CO2 increase alone, whereas an opposite pattern is observed at 50°–45° S (Subantarctic Zone). Together with other processes, the enhanced acidification at 60°–55° S may be attributed to increased westerly winds that bring in more “acidified” waters from the higher latitudes via enhanced meridional Ekman transport and from the subsurface via increased vertical mixing. Our observations support climatic modulation of ocean acidification superimposed on the effect of increasing atmospheric CO2.

Introduction

The Southern Ocean has naturally low pH and saturation states of calcium carbonate (CaCO3) due to cold temperatures and upwelling of CO2-enriched deep waters, and it is vulnerable to ocean acidification (OA) caused by increasing atmospheric CO2 levels1,2,3,4. Surface waters of the Southern Ocean are predicted to become undersaturated with respect to aragonite (a more soluble form of CaCO3 relative to calcite) as early as year 2030 if sea surface CO2 increases in concert with atmospheric CO2 (ref. 3). OA, defined as declining pH or CaCO3 saturation states over decades or longer timescales5, affects many marine organisms and especially fragile Southern Ocean ecosystems6,7,8. Although global OA is due primarily to increasing atmospheric CO2 by fossil fuel combustion and land use changes since the Industrial Revolution2,9, it may be enhanced by other processes such as upwelling, eutrophication, sea ice melt, and anomalous ocean circulation10,11,12,13,14,15,16,17. Such rapid acidification challenges the evolutionary adaptation capacity of organisms18. Therefore, understanding the processes or factors that modulate OA is important for projecting impacts on marine organisms and ecosystems.

Climatically, the Southern Ocean is sensitive, particularly during austral summer, to the Southern Annular Mode (SAM) that is the dominant mode of climate variability in the extratropical Southern Hemisphere19,20. This mode is quantified by the SAM index as the difference in normalized mean sea level pressure between 40° and 65° S (ref. 20). In January, there was a positive SAM trend towards a high-index particularly since the 1980s, but this trend changed around 2000: the following decade exhibited decreased or no significant SAM trends (Fig. 1a). A positive SAM trend is associated with increasing westerly winds at high-latitudes (south of 55° S, Fig. 1b) resulting in increased equatorward Ekman transport and vertical mixing. Thus, “acidified” waters with lower pH and aragonite saturation state (Ωarag) from the south and from deeper depths are likely to be transported to the surface further north. Therefore, enhanced surface OA in excess of the effect of increasing atmospheric CO2 may be expected at high-latitudes during a period with positive SAM trends. Here we define enhanced OA as evidenced by declining rates of pH or Ωarag that are faster than rates predicted from increasing atmospheric CO2 alone.

Fig. 1
Fig. 1

SAM index and change rates of zonal wind speed in the Southern Ocean. a SAM index in January 1983–2017 calculated by Marshall20. b, c Change rates of zonal wind speed in January 1991–2000 and January 2001–2011. In a, change rates of the SAM index (slope values) during the periods 1991–2000 (blue) and 2001–2011 (black) were determined using an ordinary least squares linear regression; slopes (s) and p-values of the regression analyses are also shown (differentiated with blue and black colors for the two periods). The gray line shows the weighted three-year running mean of the SAM index, which splits the data into two decades. In b, c, the red and blue lines show the mean positions of the subantarctic front (SAF) and the polar front (PF)58, respectively; the black rectangle delineates the study area south of Tasmania. Change rates of zonal wind speeds, which are based on the CCMP wind product, were calculated using an ordinary least squares linear regression in each grid (0.25° × 0.25°)

However, due partly to lack of observational data, previous studies on the effects of SAM on OA in the Southern Ocean use models which yielded different and even opposite conclusions18,21,22,23. Therefore, it is necessary and important to further investigate the mechanistic role of the SAM on sea surface carbonate chemistry and OA. Also, it is important to ascertain whether OA responds to the SAM differently for different latitudinal zones as was shown for circulation and biology24, since the SAM measures a seesaw of atmospheric mass between the high-latitudes and mid-latitudes of the Southern Hemisphere20.

Given that the region south of Tasmania is perhaps the only region where there is continuous observational CO2 data since 1991 (Supplementary Fig. 1), we use observations from this area spanning two decades during 1991–2011, with contrasting SAM trends before and after 2000 (Fig. 1a and Supplementary Figs. 23) and show how changing wind patterns related to the SAM affect the rate of surface OA. We find that the SAM appears to have significant modulating effects on OA rates over different latitudinal zones. To account for the SAM modulation of OA rates, we examine mechanisms associated with wind-driven meridional Ekman transport and vertical mixing during austral summer when the upper ocean layers are stratified. Our work helps improve understanding of the mechanisms of OA in the Southern Ocean, thus providing observational constraints for the improvements of prediction models for ocean uptake of atmospheric CO2 and impacts on the marine ecosystem.

Results

Changes of carbonate chemistry with time

Using observed sea surface CO2 fugacity (fCO2), temperature (SST) and salinity (SSS) from the Surface Ocean CO2 Atlas (SOCAT version 2)25, and estimated total alkalinity (TA) from SSS, SST, and latitude (Fig. 2), we calculated dissolved inorganic carbon (DIC), pH, and Ωarag over the two contrasting decades, 1991–2000 and 2001–2011 (see 'Methods'). The estimated values of TA and DIC agree well with measured data ('Methods' and Supplementary Fig. 4), giving high confidence in the calculated pH and Ωarag. To achieve a better spatial representation, prior to these calculations, the surface fCO2, SST, and SSS data were binned and averaged within 0.02° latitudinal bands. Then averages were taken for the 5° latitudinal bands of 60°–55° S (high-latitudes or Antarctic Zone), 55°–50° S (transition zone or Polar Frontal Zone) and 50°–45° S (mid-latitudes or Subantarctic Zone)26. Finally, these data were adjusted to January values using the climatological seasonal variations described by Takahashi et al.27 (see 'Methods'). While trends in SST, SSS, and TA were often not statistically significant, the relative rate of fCO2 increase in surface water vs. that in the atmosphere was clear over the three regions and both time periods. A faster fCO2 increase occurred during the pre-2000 positive SAM trend period in the high-latitude zone (60°–55° S), and a slower (or zero) increase in the mid-latitude zone (50°–45° S) compared to the atmospheric increase (Fig. 2).

Fig. 2
Fig. 2

Temporal variability in sea surface temperature, salinity, and carbonate parameters in January in three latitudinal bands. ae Sea surface temperature (a, SST), salinity (b, SSS), sea surface CO2 fugacity (c, fCO2), estimated total alkalinity (d, TA) and calculated dissolved inorganic carbon (e, DIC) at 60°–55° S. fj Show the same parameters but at 55°–50° S; ko also show the same parameters but at 50°–45° S (see 'Methods'). The vertical bars show one standard deviation, which reflects the spatial variability within each latitudinal band. Linear regression analyses were performed for the periods 1991–2000 (blue) and 2001–2011 (black). Slopes (s) and p-values of the regression analyses are also shown (differentiated with blue and black colors for the two periods). A trend of p-value < 0.1 is regarded as statistically significant (90% confidence interval) due to the small sample numbers (<10). Also, the atmospheric CO2 data (shown as fCO2) observed at the GCO (Cape Grim, Tasmania) atmospheric CO2 measurement station (ftp://aftp.cmdl.noaa.gov/data/trace_gases/co2/flask/) and the DIC values computed due solely to the atmospheric CO2 increase (see 'Methods') are indicated with open gray circles in Fig. 2c, e

Figure 3 shows that the rates of pH and Ωarag change (i.e., rate of acidification) correlate with the SAM trends (Fig. 1a). At high-latitudes (60°–55° S), pH at in situ temperature (pH@in situ) decreased faster (0.0035 yr–1) during the pre-2000 positive SAM trend than the pH decrease expected from atmospheric CO2 increase alone (0.0020 yr–1, gray dashed line, Fig. 3a). Correspondingly, Ωarag at the in situ temperature (Ωarag@in situ) decreased at a rate of 0.018 yr–1, which is more than twice the rate of 0.007 yr–1 due to atmospheric CO2 alone (Fig. 3b). During the subsequent decade (2001–2011) when there was no significant SAM trend, pH@in situ and Ωarag@in situ decreased at rates in accord with those predicted from atmospheric CO2 (Fig. 3a, b).

Fig. 3
Fig. 3

Temporal variability in sea surface pH, Ωarag, zonal wind speed and mixed layer depth (MLD) in January in three latitudinal bands. ad Sea surface pH at in situ temperature (a, pH@in situ), sea surface aragonite saturation state at in situ temperature (b, Ωarag@in situ), zonal wind speed (c) and mixed layer depth (d, MLD) at 60°–55° S. eh show the same parameters but at 55°–50° S; il shows the same parameters but at 50°–45° S. The vertical bars show one standard deviation, which reflects the spatial variability within each latitudinal band. Linear regression analyses were performed for the periods 1991–2000 (blue) and 2001–2011 (black). Slopes (s), and p-values of the regression analyses are also shown (differentiated with blue and black colors for the two periods). Trends of p-value < 0.1 are regarded as significant statistically (90% confidence interval) due to the small sample numbers (<10). The open gray circles indicate the values computed due solely to the atmospheric CO2 increase shown in Fig. 2c (see 'Methods'). Zonal wind speed and MLD are the mean values within 140°–148° E in the three latitudinal bands, respectively. Note by definition the trend of meridional Ekman transport is the same as that of zonal wind

In contrast, at mid-latitudes (50°–45° S), patterns opposite to those seen in the high-latitude band were observed (Fig. 3). During the decade of positive SAM trend (1991–2000), pH@in situ decreased much slower than would be expected from atmospheric CO2, and Ωarag@in situ even increased, although neither trend was statistically significant. During the subsequent decade (2001–2011) when there was no significant SAM trend, pH@in situ and Ωarag@in situ both showed enhanced rates of decrease relative to the atmospheric CO2 based prediction (Fig. 3i, j). For the transitional band (55°–50° S), the decrease in surface pH@in situ during the two SAM periods was not statistically distinguishable from that predicted from atmospheric CO2 and there were no significant changes in Ωarag@in situ (Fig. 3e, f). Overall, acidification rates differ during different SAM-trend periods and within different latitudinal bands, similar to the responses of circulation and biology to SAM24, suggesting that the influence of SAM on the acidification rates was likely associated with SAM-sensitive physical and/or biological factors.

Correlation between wind trend and OA rates

Our results display a consistently negative correlation between pH@in situ (or Ωarag@in situ) and wind speed, despite varying latitudinal responses of wind speed to the SAM trend (Fig. 3). In the high-latitude 60°–55° S band, wind speed increased significantly during the 1991–2000 positive SAM trend (Fig. 3c), when pH@in situ and Ωarag@in situ decreased faster than expected from the atmospheric CO2 increase (Fig. 3a, b). During a period with an insignificant change in SAM trends in 2001–2011 when wind speed decreased or did not change significantly, pH@in situ and Ωarag@in situ declined at rates similar to those expected from the atmospheric CO2 increase. In contrast, in the mid-latitude 50°–45° S band, during the period of positive SAM trends when winds decreased (Fig. 3k), pH@in situ only decreased slightly and Ωarag@in situ increased somewhat (though not significantly, Fig. 3i, j), whereas during a period with an insignificant change in SAM trends when winds increased (Fig. 3k), pH@in situ and Ωarag@in situ decreased evidently (Fig. 3i, j). For the transitional 55°–50° S band, there were no apparent changes in wind speed and, correspondingly, there was no enhanced acidification during the two periods of 1991–2000 and 2001–2011 (Fig. 3e–g).

We see more clearly the complex effects of wind on rates of pH and Ωarag change by subtracting their rates of decrease due solely to atmospheric CO2 increase from the observed rates of pH and Ωarag change (see 'Methods'). It is clear, after removing the effects of atmospheric CO2 increase, that the rates of pH and Ωarag change are negatively correlated with change rates of zonal wind speed over the two periods and the three latitudinal bands (Fig. 4). That is, increasing winds enhance acidification.

Fig. 4
Fig. 4

Impacts of SAM associated winds on acidification rates. a, b rates of surface pH (a) and Ωarag change (b) without atmospheric CO2 increase versus change rates of January zonal wind speed during two periods of 1991–2000 (blue) and 2001–2011 (black) in the three latitudinal bands of 60°–55° S, 55°–50° S and 50°–45° S. Rates of pH and Ωarag change without atmospheric CO2 increase are highly negatively correlated with change rates of zonal wind speed with a correlation coefficient of 0.92 and 0.89, respectively. Rates of pH and Ωarag change without atmospheric CO2 increase were the observed rates of pH@in situ and Ωarag@in situ change, subtracting their rates predicted from atmospheric CO2 increase alone (see 'Methods'). In this figure, negative change rates of pH or Ωarag denote enhanced acidification compared to that predicted from atmospheric CO2 increase alone. The bars show one standard deviation of change rates as shown in Fig. 3

Modulations of Ekman transport and vertical mixing on OA

Considering the correlation between enhanced pH and Ωarag decreases and zonal wind speed changes (Fig. 4), and the lateral and vertical distributions of pH and Ωarag in the Southern Ocean (Fig. 5), we explore the impacts on surface acidification from lateral transport and vertical mixing, both of which are influenced by wind speeds. Note in this section we used values at the regional mean temperature of 7.45 °C or pH@7.45 and Ωarag@7.45 to examine the non-thermal influences of pH and Ωarag although temperature influence (thermal influences) on these parameters was relatively minor (see 'Methods'). We considered that, among various drivers listed in Table 1, wind-driven lateral or Ekman transport was one of the important contributors to the trend in pH and Ωarag changes relative to the atmospheric CO2 increase. As shown in Fig. 5a, b, surface pH@7.45 and Ωarag@7.45 in the Southern Ocean decreased poleward. At high-latitudes (60°–55° S) during the positive SAM trend of 1991–2000, the increase in westerly winds (westerly anomaly) enhances equatorward Ekman transport (Supplementary Fig. 5a), causing more waters with low pH@7.45 and Ωarag@7.45 (“acidified” waters) from further south to be transported to this zone (Fig. 5a, b). This should result in further decreases in pH and Ωarag (enhanced acidification, Fig. 3a, b; Supplementary Fig. 6a, b) in addition to those due to atmospheric CO2 increase. In contrast, in the mid-latitude band (50°–45° S), westerly winds decreased (easterly anomaly) during a positive SAM trend, resulting in decrease in equatorward Ekman transport (i.e., anomalous poleward Ekman transport, Fig. 5; Supplementary Fig. 5a) and hence a slight increase in pH@7.45 and Ωarag@7.45 (Supplementary Fig. 6k, l). This should counteract the acidification by increasing atmospheric CO2, thus leading to no clear trends in pH@in situ and Ωarag@in situ (suppressed acidification, Fig. 3i, j). Similarly, during the subsequent decade of 2001–2011 changes in pH@7.45 and Ωarag@7.45 at high-latitudes and mid-latitudes (Fig. 3; Supplementary Fig. 6) were also consistent with those expected by wind-driven Ekman transport (Supplementary Fig. 5b).

Fig. 5
Fig. 5

A schematic of Ekman transport and vertical mixing modulation of acidification rates as well as climatological distribution of surface pH@7.45 and Ωarag@7.45. a, b Depict changes of wind-driven Ekman transport and mixed layer depth (MLD), and their influences on surface pH@7.45 (a) and Ωarag@7.45 (b) in the region south of Tasmania during a positive SAM trend. There will be an anomalous equatorward Ekman transport when westerly winds increase (westerly anomaly), while there will be an anomalous poleward Ekman transport when westerly winds decrease (easterly anomaly). MLD is used for showing the changes in vertical mixing in the upper ocean. Black dashed (normal) and solid (+SAM) lines denote MLD before and during a positive SAM trend, respectively. In a, b, SAF and PF denote the mean positions of the subantarctic front (SAF) and the polar front (PF)58. In the water column, neutral density contours with an interval of 0.25 kg m−3 (γn, white dashed line), and pH@7.45 (a) and Ωarag@7.45 (b) distribution (shaded) observed along Transect SR03 during December 1994–January 1995 are shown. Also in a, b Points S, N, and D are shown, details about which can be found in Supplementary Table 5. In c, d, climatological distribution of surface pH@7.45 (c) and Ωarag@7.45 (d) in January calculated from the TA and DIC data of Takahashi et al.9 is shown. Note a, c use the same color bar, and b, d use the same color bar. Figure 5 is plotted using Ocean Data View (odv_4.7.10_w64 version)59

Table 1 Observed trend change of OA, wind speed, Ekman transport, Ekman pumping velocity, mixed layer depth (MLD), sea surface temperature (SST), salinity (SSS), total alkalinity (TA), dissolved inorganic carbon gains due to air-sea CO2 exchange (ΔDICa-s) and chlorophyll a in the three latitudinal bands of 60°–55° S, 55°–50° S, and 50°–45° S during the 1991–2000 positive SAM period

Since pH and Ωarag decreased with depth (Fig. 5a, b), enhanced vertical mixing should also lead to an enhanced acidification. To determine whether changes in vertical mixing in the upper ocean can be a major contributor to a change of acidification rates, we examined changes in mixed layer depth (MLD, Fig. 3d, h, l). During a period with positive SAM trends at high-latitudes (60°–55° S), MLD showed an increasing trend (Fig. 3d), suggesting an increase in vertical mixing that will entrain more subsurface waters with low pH and Ωarag into the mixed layer (Fig. 5a, b), which enhances acidification rates. In contrast, in the mid-latitude band (50°–45° S), MLD showed a decreasing trend during a positive SAM trend (Fig. 3l), suggesting a decrease in vertical mixing that will entrain less subsurface waters into the mixed layer (Fig. 5a, b), which suppresses acidification rates. However, there were almost no changes in MLD during 2001–2011 at high-latitudes and mid-latitudes or in either period in the transition zone (Fig. 3), revealing that mixing in the upper ocean has no obvious changes during these periods.

It seems that Ekman transport brings more water from the higher latitudes than from the subsurface water to the surface Antarctic Zone, since observed changes of SST and SSS at the three latitudinal zones (Fig. 2) are consistent with changes expected due to Ekman transport (Supplementary Fig. 5a; Table 1). For example, at high-latitudes (60°–55° S) during a positive SAM trend, increased equatorward Ekman transport should induce a drop in SSS (Supplementary Fig. 7), whereas increased vertical mixing should cause a rise in SSS (Supplementary Fig. 7), but in fact we observed a decrease in SSS (Fig. 2b). However, vertical mixing may still play an important role in modulating OA due to the stronger gradients of pH and Ωarag in the vertical direction than in the lateral direction (Fig. 5a, b; Supplementary Table 5). For instance, vertically from Point N to Point D, salinity increased by 0.08, and pH@7.45 and Ωarag@7.45 decreased by 0.06 and 0.25 units, respectively, with pH@7.45 change per unit salinity of −0.72 and Ωarag@7.45 change per unit salinity of −3.01. Laterally, from Point S to Point N, salinity, pH@7.45 and Ωarag@7.45 increased by 0.42, 0.07 and 0.26 units, respectively, with pH@7.45 change per unit salinity of 0.17 and Ωarag@7.45 change per unit salinity of 0.62. Therefore, vertical mixing could also play an important role in modulating OA. This is further supported by a mass balance model calculation (see 'Methods' and Supplementary Table 6). Overall, given the covariation of Ekman transport and vertical mixing with SAM associated winds (Table 1) and their consistent effects on pH and Ωarag, they both synergistically modulated the OA rates caused by increasing atmospheric CO2.

Note in our paper we use vertical mixing rather broadly and mean to include convergence (i.e., downwelling) or divergence (i.e., upwelling), trend changes of which are quantified by Ekman pumping velocity (Supplementary Fig. 5c, d). We find there was a trend of decrease (increase) in Ekman pumping at the three latitudinal zones during 1991–2000 (2001–2011) (Supplementary Fig. 5c, d), but it appears that their influences on changes of pH and Ωarag are minor or not observed. This can be seen, for example, from the transition zone (55°–50° S) where there were no apparent changes in wind speed, MLD or Ekman transport in either period (Fig. 3g, h; Supplementary Fig. 5a, b). During 1991–2000 when there was a tendency toward anomalous convergence (i.e., decreasing Ekman pumping, Supplementary Fig. 5c), waters with relatively low pH and Ωarag from the high-latitude zone and waters with relatively high pH and Ωarag from the mid-latitude zone may have been simultaneously transported to the transition zone (Fig. 5a, b), resulting in the cancellation of these effects and no net enhanced acidification during this period (Fig. 3e, f). While there was a tendency toward increasing upwelling during the period 2001–2011 (i.e., increasing Ekman pumping, Supplementary Fig. 5d), the influence of upwelling on SST, SSS, pH, and Ωarag was not observed in the transition zone (Figs. 2 and 3). It indicates that the influence of upwelling on upper mixed layer was probably small and was susceptible to other processes, which needs further studies in this region.

Several processes could contribute to enhanced OA relative to atmospheric CO2 increase. These could include an increase in equatorward transport of high-latitude water, an increase in vertical mixing/upwelling, a decrease in biological production, and an increased gas exchange rate due to increasing wind speed. While we cannot discount the importance of changes in local air-sea CO2 flux and biological activity, the mass balance analysis suggests that they both only play a relatively minor role in modulating acidification induced by increasing atmospheric CO2 (see 'Methods'). Instead, the trend in TA (Fig. 2) and the mass balance analysis (see 'Methods') confirm that lateral and vertical transport are the dominant processes in modulating OA rates, since TA is immune to air-sea CO2 fluxes and is just weakly influenced by biology28. By these analyses and by comparing observed changes of quantities with expected influences of SAM-associated winds (Table 1), we therefore conclude that the observed enhanced acidification in the high-latitudes and suppressed acidification in the mid-latitudes are primarily attributable to wind-driven Ekman transport and vertical mixing.

Discussion

To put our findings in a broad context, we explore the possible influence of the SAM on pH and Ωarag changes in other regions of the Southern Ocean by examining CCMP (Cross-Calibrated Multi-Platform) January zonal wind trends in the Southern Ocean basin (Fig. 1b, c). We find that during positive SAM trends (1991–2000), January wind speed increased at high-latitudes (poleward of the subantarctic front, SAF), and decreased at mid-latitudes (equatorward of SAF) except in parts of the Pacific sector, similar to the patterns shown by Lovenduski and Gruber24. In contrast, during a period with insignificant SAM trends (2001–2011), January zonal winds generally exhibited an opposite spatial pattern—decreasing at high-latitudes and increasing at most mid-latitudes. In January, across the entire Southern Ocean wind speed trends were largely consistent with those in our study region, though there were regional differences in the location of the transitional zone between high-latitude and mid-latitude patterns (Fig. 1b, c). This suggests that our study area can represent the meridional feature of the wind change in the Southern Ocean, and that the SAM modulation of OA rates south of Tasmania is part of a SAM-regulated change spanning the whole Southern Ocean. Given the spatial distribution of pH@7.45 and Ωarag@7.45 in the entire Southern Ocean (Fig. 5), it may be inferred that during austral summer, the SAM would have substantial impacts on acidification rates in the whole Southern Ocean via wind forcing, although some heterogeneities do exist among the different Southern Ocean sectors. For instance, during 2001–2011, January zonal wind speed showed a decreasing trend in the 60°–55° S band south of Tasmania and in most other sectors in the Southern Ocean, but a weak increasing trend in the Drake Passage (68°–56° W in Fig. 1c). Accordingly, the rate of Ωarag decrease during this period appeared to be somewhat greater in the Drake Passage (0.013 ± 0.003 yr–1 for austral summer during 2002–2015, see Munro et al.29) than in the high-latitude band in our study area (0.007 ± 0.005 yr–1). This case further supports the mechanism of wind-driven modulation of OA, although it still needs to be verified in other sectors since the pattern of MLD change south of Tasmania is not fully consistent with that in other parts of the Southern Ocean (Supplementary Fig. 5e, f).

Our study based on observational data further supports the idea that the SAM has modulating effects on the Southern Ocean CO2 system, mainly via wind-driven Ekman transport and vertical mixing during stratified austral summer (Supplementary Fig. 7). Studies based on ocean circulation models30,31 and atmospheric CO2 inverse methods23 suggest modulation of the SAM on Southern Ocean carbon uptake. This conclusion is also supported by fCO2 observations32,33,34. In contrast, using data derived by a neural network technique, Landschützer et al.35 find that, on an annual basis, the reinvigoration of Southern Ocean CO2 uptake after the early 2000s cannot be explained by the SAM-associated wind trends, because there are almost no changes in annual trends of wind between the 1990s and the 2000s (Supplementary Fig. 8). Instead, they propose a mechanism associated with a more zonally asymmetric atmospheric circulation35. More recently, however, using ocean circulation models, DeVries et al.36 find that increased Southern Ocean CO2 uptake in the 2000s compared to the 1990s was due to reduced upwelling (a weakened upper-ocean overturning circulation, similar to our mechanism).

There are at least two possible reasons that can explain why our results differ from those of Landschützer et al.35. One important reason is that there would be a large difference between seasonal trends (e.g., summer) and annual trends. To validate this, we chose the CCMP wind data37,38 to examine the differences between January and annual trends. We find that there are substantial differences between January and annual trends (Fig. 1 and Supplementary Fig. 8). Another reason is that the SAM and its effects have a strong seasonality, with the most pronounced influence during austral summer19,20. Additionally, we recognize that changes in water properties of previous winter and spring seasons may also affect summer water properties, which is not discussed in our work due to data limitations and should be concerned in future observational and modeling efforts. Therefore, our study supports SAM modulation of acidification rates during austral summer (January), but sufficient observational data are not available27 to elucidate the full annual influence of SAM on acidification. To resolve these issues, more observations are needed in this remote ocean for all seasons, especially during the poorly sampled austral winter.

Our work clarifies the discrepancy regarding the influence of the SAM on Southern Ocean acidification. Previous studies in this aspect are all based on models that yielded different understandings of SAM impacts. Some of these studies indicate that a positive SAM trend would not substantially affect OA or that the role of climate-driven physical changes would be minor18,21. While other studies argue that OA in the entire Southern Ocean would be enhanced during a positive SAM trend22,23 in agreement with our viewpoint, they did not reveal latitudinal differences in their studies. Therefore, our work helps improve understanding of the mechanisms of OA in the Southern Ocean, which is important for modeling atmospheric CO2 uptake and ecosystem responses. Overall, our work provides observational support for climatic modulation of OA, which should be taken into account in future predictions of acidification. It is most likely that climate change and variability have already been affecting the advance of OA in the global ocean via wind forcing14, which requires further observations.

Methods

Data collection and processing

Sea surface fCO2, SST, and SSS data from 17 cruises south of Tasmania (along or near Transect SR03, 60°–45° S) during 1991–2011 are used in this study (Supplementary Figs. 23). These data were extracted from the Surface Ocean CO2 Atlas (SOCAT version 2)25 (http://www.socat.info/) and were collected by the groups of Bronte Tilbrook, Hisayuki Y. Inoue, Nicolas Metzl, Rik Wanninkhof, and Taro Takahashi32,39,40,41. TA and DIC data from discrete seawater samples collected along Transect SR03 during December 1994–January 1995, September 1996, March 1998, October–November 2001 and April 2008 in combination with salinity, temperature, and latitude data were used to derive the TA relationship (see 'TA estimation'). Also, TA and DIC data during December 1994–January 1995 were used to calculate pH@7.45 and Ωarag@7.45 as shown in Fig. 5a, b. These data were obtained from the Global Ocean Data Analysis Project, Version 2 (GLODAPv2)42.

SAM index used in this study is observation-based and developed by Marshall20, and is available at http://www.antarctica.ac.uk/met/gjma/sam.html. CCMP wind data37 were chosen for examining changes in zonal wind speed because of their good data quality38. This product has a resolution of 0.25° and is available at http://podaac.jpl.nasa.gov/datasetlist?search=ccmp. Monthly mean MLD data with a resolution of 0.5° determined by temperature criteria from SODA v3.3.1 (Simple Ocean Data Assimilation, available at http://apdrc.soest.hawaii.edu/dods/public_data/SODA/soda_3.3.1/) were used to examine the variability of MLD during January 1991–2011.

Before calculating carbonate system parameters, we binned, averaged, and deseasonalized the surface underway data, including sea surface fCO2, SST, and SSS. For each parameter, we first binned all data points into 0.02° latitudinal bands to overcome different sampling frequencies among the cruises, calculated the average for each band, and finally took average values for the latitudinal bands of 60°–55° S, 55°–50° S and 50°–45° S, respectively as in Xue et al.34. For deseasonalization, we adjusted the averaged fCO2, SST, and SSS data to January values using the long-term averaged seasonal cycle obtained by Takahashi et al.27 (Supplementary Tables 13) as done by e.g., Lauvset and Gruber43. January was chosen because it is the month during which most data are collected (Supplementary Figs. 23) and also because the influence of the Antarctic ozone hole on surface climate is most pronounced during austral summer19. In addition, there is an evident SAM trend during this month (Fig. 1a). The averaged and deseasonalized fCO2, SST, and SSS values at 60°–55° S, 55°–50° S, and 50°–45° S bands between 142.5° and 147.5° E are shown in Fig. 2.

TA estimation

Surface water TA within the study area was estimated using SSS (PSS), SST (°C) and latitude (Lat, in decimal degrees, negative for South latitudes) via Eq. (1):

$${\mathrm{TA}}\left( {{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - 1}} \right) = 35.94 \times {\mathrm{SSS}} + 0.49 \times {\mathrm{SST}} - 1.65 \times {\mathrm{Lat}} + 964.97$$
(1)

(r2 = 0.92, n = 346)

Equation (1) was determined via multiple linear regression using the measured data in the upper 60 dbar along Transect SR03 collected during December 1994–January 1995, September 1996, March 1998, October–November 2001 and April 2008 (see 'Data collection and processing'). A comparison between estimated TA and the measured TA yielded a root mean square error (RMSE) of ±3.5 μmol  kg–1 (Supplementary Fig. 4a). This is better than what was initially derived from the global equation of Lee44, which generates a RMSE of ±6.4 μmol kg–1.

To examine uncertainty associated with the TA estimation, DIC, pH, and Ωarag were calculated from fCO2 derived from the observed TA and DIC and the estimated TA, using the CO2SYS program45 and the apparent carbonic acid dissociation constants of Mehrbach et al.46 as refit by Dickson and Millero47. The resulting RMSEs for DIC, pH, and Ωarag were ±3.0 μmol kg–1, ±0.0010 and ±0.005, respectively, when compared with measured DIC, and calculated pH and Ωarag from a measured DIC and TA pair (Supplementary Fig. 4b–d). According to the error-calculation method of Lauvset and Gruber43, the calculation errors of pH and Ωarag are estimated to be 0.0022 and 0.010, respectively. Given the uncertainties of ±(0.005–0.01) for spectrophotometrically measured pH (refs. 48,49,50) and ±0.18 for Ωarag calculated from paired measurements of carbonate parameters51,52, we conclude that error associated with the estimation of TA will not affect our results or conclusions.

Calculation of pH and Ωarag

Surface pH on the total H+ concentration scale at in situ temperature and at the regional mean temperature of 7.45 °C was calculated using the CO2SYS program45, with inputs of measured surface fCO2 and estimated TA (Fig. 2) and climatological phosphate and silicate concentrations (Supplementary Table 4, though nutrient effects on pH and Ωarag are small). The apparent carbonic acid dissociation constants of Mehrbach et al.46 as refit by Dickson and Millero47 were used, as recommended by Chen et al.53 for polar ocean waters. For calculating Ωarag (=[CO32−] × [Ca2+] / Ksparagonite), carbonate ion concentration ([CO32−]) was also calculated using the CO2SYS program45. The calcium ion concentration ([Ca2+]) was calculated from salinity (0.01026 / 35 × salinity [mol kg–1]) based on the conservative behavior of [Ca2+] to salinity54, and the apparent solubility product of aragonite (Ksparagonite) was calculated after Mucci52. Also, pH@7.45 and Ωarag@7.45 in the upper 400 dbar along Transect SR03 during December 1994–January 1995 (Fig. 5a, b) and climatological values of surface pH@7.45 and Ωarag@7.45 in January (Fig. 5c, d)9 were calculated from the TA and DIC data, respectively.

Quantification of ocean acidification rates

First, we calculated a weighted three-year running mean (1:2:1) for the SAM time series and found that the January SAM trend showed a clear shift in 2000 (gray line in Fig. 1a), splitting the data into two decades (i.e., 1991–2000 vs. 2001–2011). Then, following the definition of OA5, rates of OA are characterized by rates of pH and Ωarag change with time (i.e., slopes in Fig. 3), which were obtained using ordinary least squares linear regressions over each of the two decades of interest. To obtain the rates of decrease of pH and Ωarag due solely to atmospheric CO2 increase, we used the CO2SYS program45 to calculate pH and Ωarag from constant TA and increasing fCO2, i.e., we held TA, SSS, and SST constant at their 1991 values while allowing surface water fCO2 to increase at the same rate as atmospheric CO2 observed at the GCO (Cape Grim, Tasmania) atmospheric CO2 measurement station (ftp://aftp.cmdl.noaa.gov/data/trace_gases/co2/flask/; Fig. 2c). Similarly, changes in DIC and the difference between TA and DIC ([TA−DIC]) due solely to the increase in atmospheric CO2 were calculated. Given that air-sea CO2 exchange affects DIC but not TA, changes in [TA−DIC] that are due solely to changes in atmospheric CO2 will have the same amplitude as the changes in DIC but will be of opposite sign (i.e., different direction of change). Rates of pH and Ωarag change without atmospheric CO2 increase (i.e., excluding the effects of increasing atmospheric CO2, Fig. 4) were the observed rates of pH@in situ and Ωarag@in situ change subtracting their rates predicted from atmospheric CO2 increase alone (shown by gray dashed lines in Fig. 3).

Thermal influences on OA rates

The thermal influences on changes of pH and Ωarag due to temperature changes are relatively minor. This is because there were no substantial changes in SST during the study period, although during a positive SAM trend (1991–2000) SST showed a decrease trend at 60°–55° S band, and an increase trend at 50°–45° S band (Fig. 2a, f, k). During 1991–2000 for the 60°–55° S band, although there was a decrease trend in SST of 0.08 °C yr–1 (Fig. 2a), thus tending to increase pH (ref. 55), the observed rate of pH@in situ decrease was still faster than the rate attributable solely to atmospheric CO2 increase (gray dashed line, Fig. 3a). Comparing the rates of decrease for pH@in situ (Fig. 3a) and pH@7.45 (Supplementary Fig. 6a) during this period shows that the effect of decreasing SST only partly counteracted the pH decreases. During 1991–2000 for the 50°–45° S band, the difference between the pH@in situ and pH@7.45 trends is not statistically significant (Fig. 3i and Supplementary Fig. 6k). Compared to pH, Ωarag is relatively insensitive to temperature changes, with surface Ωarag@in situ and Ωarag@7.45 showing almost the same variability and rates of change throughout (Fig. 3 and Supplementary Fig. 6).

Mass balance analysis of Ekman transport and vertical mixing

For convenience and effectiveness of discussion, we introduce a combined property, the difference between the concentrations of TA and DIC, i.e., [TA−DIC]. Unlike pH and Ωarag, [TA−DIC] is a conservative quantity composed of two conservative parameters, and hence is suited for analysis of water mass mixing. The [TA−DIC] approximates closely the concentration of carbonate ions ([CO32–]) by definition56 and can be used as a proxy for pH and Ωarag@in situ57. In our dataset, [TA−DIC] correlates well with pH@7.45 and with Ωarag@in situ, with a correlation coefficient of r of nearly one (Supplementary Fig. 9). Thus, [TA−DIC] can be used to assess the changing effect of a process on OA rate.

To examine the relative contribution of Ekman transport vs. vertical mixing on OA rates, we use salinity and TA as conservative tracers to resolve changes in TA and DIC due to changes in these two processes. In the following, we take the case at high-latitudes during the 1991–2000 positive SAM trend as an example, and estimate the relative contribution of these two processes. We selected three points along transect SR03 to be compared including surface point S (South) and point N (North) and deep point D under N (Fig. 5 and Supplementary Table 5). To derive quantitatively the amounts of TA, DIC, and [TA−DIC] changes caused by changes in lateral transport and vertical mixing during a positive SAM period, we consider the changes of salinity (S) and TA in surface waters at point N:

$$\Delta {S}_{\mathrm{E}} + \Delta {S}_{\mathrm{V}} = \Delta {S}$$
(2)
$$\Delta {\mathrm{TA}}_{\mathrm{E}} + \Delta {\mathrm{TA}}_{\mathrm{V}} = \Delta {\mathrm{TA}}$$
(3)

where the sign “Δ” denotes changes of a parameter; and subscripts “E” and “V” denote Ekman transport and vertical mixing, respectively. Note that the mass balances are built upon the changes of salinity and TA during the period but not on the absolute amount due to lateral and vertical transports. Also, we neglect the influence of change in precipitation–evaporation balance as its influence on TA and DIC is small and similar, and thus its influence on [TA−DIC] is negligibly small.

Based on gradients per salinity change between Points S and N (lateral), and Points D and N (vertical) (Supplementary Table 5), we

$$\Delta {\mathrm{TA}}_{\mathrm{E}} = 50.95 \times \Delta {S}_{\mathrm{E}}$$
(4)
$$\Delta {\mathrm{TA}}_{\mathrm{V}} = 84.34 \times \Delta {S}_{\mathrm{V}}$$
(5)

Thus, Eq. (3) can be rewritten as

$$50.95 \times \Delta {S}_{\mathrm{E}} + 84.34 \times {\Delta S}_{\mathrm{V}} = {\mathrm{\Delta TA}}$$
(6)

Since ΔS and ΔTA are known during the positive SAM period (slope values in Fig. 2b, d), through Eq. (2) and (6), we obtain ΔSE = −0.023 yr−1 and ΔSV = 0.006 yr−1 during the positive SAM period. Thus, based on the gradients of TA and DIC shown in Supplementary Table 5, the respective contribution of Ekman transport and vertical mixing on TA, DIC, and [TA−DIC] can be calculated (Supplementary Table 6

$$\Delta {\mathrm{TA}}_{\mathrm{E}} = -0.023 \times 50.95 = - 1.17\;{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - {\mathrm{1}}}\;{\mathrm{yr}}^{ - {\mathrm{1}}},$$$$\Delta {\mathrm{DIC}}_{\mathrm{E}} = -0.023 \times \left( { - 18.96} \right) = 0.44\;{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - 1}\;{\mathrm{yr}}^{ - 1},$$$$\Delta \left[ {{\mathrm{TA{-DIC}}}} \right]_{\mathrm{E}} = - 1.61\;{\mathrm{\mu mol}}\; \mathrm{kg}^{ - 1}\;{\mathrm{yr}}^{ - 1};$$$$\Delta {\mathrm{TA}}_{\mathrm{V}} = 0.006 \times 84.34 = 0.51\;{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - {\mathrm{1}}}\;{\mathrm{yr}}^{ - {\mathrm{1}}},$$$$\Delta {\mathrm{DIC}}_{\mathrm{V}} = 0.006 \times 366.27 = 2.20\;{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - {\mathrm{1}}}\;{\mathrm{yr}}^{ - {\mathrm{1}}},$$$$\Delta [{\mathrm{TA{-DIC}}} ]_{\mathrm{V}} = - 1.69\;{\mathrm{\mu mol}}\;{\mathrm{kg}}^{ - {\mathrm{1}}}\;{\mathrm{yr}}^{ - {\mathrm{1}}}$$

Our calculations show that at high-latitudes during the 1991–2000 positive SAM trend the contribution of Ekman transport and vertical mixing on OA rates (as Δ[TA–DIC]) are likely on the same order of magnitudes (Supplementary Table 6).

Impacts of air-sea CO2 flux and biological activity on OA

Similar to salinity and TA, we have the mass balance of DIC

$$\Delta {\mathrm{DIC}}_{\mathrm{E}} + \Delta {\mathrm{DIC}}_{\mathrm{V}} + \Delta {\mathrm{DIC}}_{\mathrm{A}} + \Delta {\mathrm{DIC}}_{\mathrm{B}} = \Delta {\mathrm{DIC}}$$
(7)

Here “A” and “B” denote changes in air-sea exchange and biological activity, respectively.

Based on this mass balance and the observed ΔDIC and calculated ΔDICE and ΔDICV, the total contribution to DIC change and thus [TA−DIC] change from air-sea gas exchange and biology is obtained (Δ[TA−DIC]A + Δ[TA−DIC]B = 1.30 µmol kg−1 yr−1), which is less than the effect by physical transports associated with changes in wind speed (Δ[TA−DIC]E + Δ[TA−DIC]V = −3.3 µmol kg−1 yr−1) (Supplementary Table 6). This result is consistent with that obtained by analyzing the trend changes of each process (Table 1). For example, during 1991–2000 at high-latitudes both decreasing air-sea CO2 flux and increasing biological production24 should result in a decrease in DIC (Supplementary Fig. 6; Table 1) and thus an increase in [TA−DIC], which can partly cancel out OA. In contrast, during this period increasing Ekman transport and vertical mixing (Table 1) should enhance OA. Comparing with observed enhancement of OA rates (Fig. 3a, b) indicates that Ekman transport and vertical mixing play a dominant role in modulating OA rates.

Further, Δ[TA−DIC]A and Δ[TA−DIC]B during 1991–2000 at high-latitudes can be calculated (Supplementary Table 6). When the time of 30 (or 100) days is considered for CO2 uptake each summer, ΔDICA and Δ[TA−DIC]A would be −0.29 (or −0.97) μmol kg−1 yr−1 and 0.29 (or 0.97) μmol kg−1 yr−1, respectively (slope values in Supplementary Fig. 6e), and thus ΔDICB and Δ[TA−DIC]B would be −1.01 (or −0.33) μmol kg−1 yr−1 and 1.01 (or 0.33) μmol kg−1 yr−1, respectively (Supplementary Table 6). Despite the fact that we cannot fully constrain the relative contribution between air-sea CO2 flux and biology, during the 1991–2000 positive SAM period at high-latitudes, biological carbon uptake induced an increase in [TA−DIC], reducing OA, which is consistent with the increasing biological production reported previously24. Note that in this paper we discuss the decadal changes of parameters or processes rather than seasonal changes of them. For example during 1991–2000 at high-latitudes the decrease in air-sea CO2 flux (Supplementary Fig. 6d) and the increase in biological production24 (characterized by chlorophyll) both should decrease DIC and thus increase [TA−DIC], reducing OA, although on seasonal timescale, for example, during January air-sea CO2 flux (absorbing CO2) will increase DIC and reduce pH and Ωarag, and biological carbon uptake will decrease DIC and increase pH and Ωarag.

Data availability

Sea surface fCO2 data can be obtained from the Surface Ocean CO2 Atlas (SOCAT version 2) (http://www.socat.info/) and the data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Change history

  • 05 September 2018

    This Article was originally published without the accompanying Peer Review File. This file is now available in the HTML version of the Article; the PDF was correct from the time of publication.

References

  1. 1.

    Fabry, V. J., McClintock, J. B., Mathis, J. T. & Grebmeier, J. M. Ocean acidification at high latitudes: the bellweather. Oceanography 22, 160–171 (2009).

  2. 2.

    Orr, J. C. et al. Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms. Nature 437, 681–686 (2005).

  3. 3.

    McNeil, B. I. & Matear, R. J. Southern Ocean acidification: a tipping point at 450-ppm atmospheric CO2. Proc. Natl Acad. Sci. 105, 18860–18864 (2008).

  4. 4.

    Midorikawa, T. et al. Decreasing pH trend estimated from 35-year time series of carbonate parameters in the Pacific sector of the Southern Ocean in summer. Deep Sea Res. Part I 61, 131–139 (2012).

  5. 5.

    Cooley, S. R., Mathis, J. T., Yates, K. K. & Turley, C. Frequently asked questions about ocean acidification. U.S. Ocean Carbon and Biogeochemistry Program and the UK Ocean Acidification Research Programme. Version 2. 24 September 2012. www.whoi.edu/OCB-OA/FAQs (2012).

  6. 6.

    Doney, S. C., Fabry, V. J., Feely, R. A. & Kleypas, J. A. Ocean acidification: the other CO2 problem. Annu. Rev. Mar. Sci. 1, 169–192 (2009).

  7. 7.

    Bednaršek, N. et al. Extensive dissolution of live pteropods in the Southern Ocean. Nat. Geosci. 5, 881–885 (2012).

  8. 8.

    Moy, A. D., Howard, W. R., Bray, S. G. & Trull, T. W. Reduced calcification in modern Southern Ocean planktonic foraminifera. Nat. Geosci. 2, 276–280 (2009).

  9. 9.

    Takahashi, T., Sutherland, S. C., Chipman, D. W., Goddard, J. G. & Ho, C. Climatological distributions of pH, pCO2, total CO2, alkalinity, and CaCO3 saturation in the global surface ocean, and temporal changes at selected locations. Mar. Chem. 164, 95–125 (2014).

  10. 10.

    Cai, W.-J. et al. Acidification of subsurface coastal waters enhanced by eutrophication. Nat. Geosci. 4, 766–770 (2011).

  11. 11.

    Dore, J. E., Lukas, R., Sadler, D. W., Church, M. J. & Karl, D. M. Physical and biogeochemical modulation of ocean acidification in the central North Pacific. Proc. Natl Acad. Sci. 106, 12235–12240 (2009).

  12. 12.

    Feely, R. A., Sabine, C. L., Hernandez-Ayon, J. M., Ianson, D. & Hales, B. Evidence for upwelling of corrosive “acidified” water onto the continental shelf. Science 320, 1490–1492 (2008).

  13. 13.

    Gruber, N. et al. Rapid progression of ocean acidification in the California current system. Science 337, 220–223 (2012).

  14. 14.

    Turi, G., Lachkar, Z., Gruber, N. & Münnich, M. Climatic modulation of recent trends in ocean acidification in the California Current System. Environ. Res. Lett. 11, 014007 (2016).

  15. 15.

    Yamamoto-Kawai, M., McLaughlin, F. A., Carmack, E. C., Nishino, S. & Shimada, K. Aragonite undersaturation in the Arctic Ocean: effects of ocean acidification and sea ice melt. Science 326, 1098–1100 (2009).

  16. 16.

    Brewer, P. G. A changing ocean seen with clarity. Proc. Natl Acad. Sci. 106, 12213–12214 (2009).

  17. 17.

    Qi, D. et al. Increase in acidifying water in the western Arctic Ocean. Nat. Clim. Change 7, 195–199 (2017).

  18. 18.

    Hauri, C., Friedrich, T. & Timmermann, A. Abrupt onset and prolongation of aragonite undersaturation events in the Southern Ocean. Nat. Clim. Change 6, 172–176 (2016).

  19. 19.

    Thompson, D. W. J. et al. Signatures of the Antarctic ozone hole in Southern Hemisphere surface climate change. Nat. Geosci. 4, 741–749 (2011).

  20. 20.

    Marshall, G. J. Trends in the southern annular mode from observations and reanalyses. J. Clim. 16, 4134–4143 (2003).

  21. 21.

    Conrad, C. J. & Lovenduski, N. S. Climate driven variability in the Southern Ocean carbonate system. J. Clim. 28, 5335–5350 (2015).

  22. 22.

    Lenton, A. et al. Stratospheric ozone depletion reduces ocean carbon uptake and enhances ocean acidification. Geophys. Res. Lett. 36, L12606 (2009).

  23. 23.

    Le Quéré, C. et al. Saturation of the Southern Ocean CO2 sink due to recent climate change. Science 316, 1735–1738 (2007).

  24. 24.

    Lovenduski, N. S. & Gruber, N. Impact of the Southern Annular Mode on Southern Ocean circulation and biology. Geophys. Res. Lett. 32, L11603 (2005).

  25. 25.

    Bakker, D. et al. An update to the Surface Ocean CO2 Atlas (SOCAT version 2). Earth Syst. Sci. Data 6, 69–90 (2014).

  26. 26.

    Sokolov, S. & Rintoul, S. R. Structure of Southern Ocean fronts at 140 E. J. Mar. Syst. 37, 151–184 (2002).

  27. 27.

    Takahashi, T. et al. Climatological mean and decadal change in surface ocean pCO2, and net sea-air CO2 flux over the global oceans. Deep Sea Res. Part II 56, 554–577 (2009).

  28. 28.

    Xue, L. et al. Sea surface carbon dioxide at the Georgia time series site (2006-2007): air-sea flux and controlling processes. Prog. Oceanogr. 140, 14–26 (2016).

  29. 29.

    Munro, D. R. et al. Recent evidence for a strengthening CO2 sink in the Southern Ocean from carbonate system measurements in the Drake Passage (2002–2015). Geophys. Res. Lett. 42, 7623–7630 (2015).

  30. 30.

    Lenton, A. & Matear, R. J. Role of the Southern Annular Mode (SAM) in southern ocean CO2 uptake. Glob. Biogeochem. Cycles 21, GB2016 (2007).

  31. 31.

    Lovenduski, N. S., Gruber, N., Doney, S. C. & Lima, I. D. Enhanced CO2 outgassing in the Southern Ocean from a positive phase of the Southern Annular Mode. Glob. Biogeochem. Cycles 21, GB2026 (2007).

  32. 32.

    Metzl, N. Decadal increase of oceanic carbon dioxide in Southern Indian Ocean surface waters (1991–2007). Deep Sea Res. Part II 56, 607–619 (2009).

  33. 33.

    Fay, A. R. & McKinley, G. A. Global trends in surface ocean pCO2 from in situ data. Glob. Biogeochem. Cycles 27, 541–557 (2013).

  34. 34.

    Xue, L., Gao, L., Cai, W.-J., Yu, W. & Wei, M. Response of sea surface fugacity of CO2 to the SAM shift south of Tasmania: regional differences. Geophys. Res. Lett. 42, 3973–3979 (2015).

  35. 35.

    Landschützer, P. et al. The reinvigoration of the Southern Ocean carbon sink. Science 349, 1221–1224 (2015).

  36. 36.

    DeVries, T., Holzer, M. & Primeau, F. Recent increase in oceanic carbon uptake driven by weaker upper-ocean overturning. Nature 542, 215–218 (2017).

  37. 37.

    Atlas, R. et al. A cross-calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications. Bull. Am. Meteorol. Soc. 92, 157–174 (2011).

  38. 38.

    Wanninkhof, R. Relationship between wind speed and gas exchange over the ocean revisited. Limnol. Oceanogr.: Methods 12, 351–362 (2014).

  39. 39.

    Brévière, E., Metzl, N., Poisson, A. & Tilbrook, B. Changes of the oceanic CO2 sink in the Eastern Indian sector of the Southern Ocean. Tellus B 58, 438–446 (2006).

  40. 40.

    Borges, A. V., Tilbrook, B., Metzl, N., Lenton, A. & Delill, B. Inter-annual variability of the carbon dioxide oceanic sink south of Tasmania. Biogeosciences 5, 141–155 (2008).

  41. 41.

    Yoshikawa-Inoue, H. & Ishii, M. Variations and trends of CO2 in the surface seawater in the Southern Ocean south of Australia between 1969 and 2002. Tellus B 57, 58–69 (2005).

  42. 42.

    Olsen, A. et al. The Global Ocean Data Analysis Project version 2 (GLODAPv2) – an internally consistent data product for the world ocean. Earth Syst. Sci. Data 8, 297–323 (2016).

  43. 43.

    Lauvset, S. K. & Gruber, N. Long-term trends in surface ocean pH in the North Atlantic. Mar. Chem. 162, 71–76 (2014).

  44. 44.

    Lee, K. et al. Global relationships of total alkalinity with salinity and temperature in surface waters of the world’s oceans. Geophys. Res. Lett. 33, L19605 (2006).

  45. 45.

    Lewis, E. & Wallace, D. W. R. Program Developed for CO 2 System Calculations. Report No. ORNL/CDIAC 105 (Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, 1998).

  46. 46.

    Mehrbach, C., Culberson, C. H., Hawley, J. E. & Pytkowicz, R. M. Measurement of the apparent dissociation constants of carbonic acid in seawater at atmospheric pressure. Limnol. Oceanogr. 18, 897–907 (1973).

  47. 47.

    Dickson, A. G. & Millero, F. J. A comparison of the equilibrium constants for the dissociation of carbonic acid in seawater media. Deep Sea Res. Part A. Oceanogr. Res. Pap. 34, 1733–1743 (1987).

  48. 48.

    Friis, K., Kortzinger, A. & Wallace, D. W. R. Spectrophotometric pH measurement in the ocean: requirements, design, and testing of an autonomous charge-coupled device detector system. Limnol. Oceanogr.: Methods 2, 126–136 (2004).

  49. 49.

    DeGrandpre, M. D. et al. Considerations for the measurement of spectrophotometric pH for ocean acidification and other studies. Limnol. Oceanogr.: Methods 12, 830–839 (2014).

  50. 50.

    Bockmon, E. E. & Dickson, A. G. An inter-laboratory comparison assessing the quality of seawater carbon dioxide measurements. Mar. Chem. 171, 36–43 (2015).

  51. 51.

    Sutton, A. J. et al. Using present-day observations to detect when anthropogenic change forces surface ocean carbonate chemistry outside preindustrial bounds. Biogeosciences 13, 5065–5083 (2016).

  52. 52.

    Mucci, A. The solubility of calcite and aragonite in seawater at various salinities, temperatures, and one atmosphere total pressure. Am. J. Sci. 283, 780–799 (1983).

  53. 53.

    Chen, B., Cai, W.-J. & Chen, L. The marine carbonate system of the Arctic Ocean: assessment of internal consistency and sampling considerations, summer 2010.Marine Chem. 176, 174–188 (2015).

  54. 54.

    Riley, J. P. & Tongudai, M. The major cation/chlorinity ratios in sea water. Chem. Geol. 2, 263–269 (1967).

  55. 55.

    Gieskes, J. M. Effect of Temperature on the pH of Seawater. Limnol. Oceanogr. 14, 679–685 (1969).

  56. 56.

    Broecker, W. S. & Peng, T. H. Tracers in the Sea (The Lamont-Doherty Geological Observatory, Palisades, 1982).

  57. 57.

    Xue, L., Cai, W.-J., Sutton, A. J. & Sabine, C. Sea surface aragonite saturation state variations and control mechanisms at the Gray’s Reef time-series site off Georgia, USA (2006–2007). Marine Chem. 195, 27–40 (2017).

  58. 58.

    Orsi, A. H., Whitworth III, T. & Nowlin Jr, W. D. On the meridional extent and fronts of the Antarctic Circumpolar Current. Deep Sea Res. Part I: Oceanogr. Res. Pap. 42, 641–673 (1995).

  59. 59.

    Schlitzer, R. Ocean Data View v.4.7.10. http://odv.awi.de (2017).

  60. 60.

    Palter, J. B. & Galbraith, E. D. Cessation of deep convection in the open Southern Ocean under anthropogenic climate change. Nat. Clim. Change 4, 278–282 (2014).

Download references

Acknowledgements

This work was supported by the National Natural Science Foundation of China (41506099), the Basic Scientific Fund for National Public Research Institutes of China (2016Q01, 2018S02), the National Natural Science Foundation of China-Shandong Joint Fund (U1606405), the National Key R&D Program of China (2018YFA0605701), Chinese Projects for Investigations and Assessments of the Arctic and Antarctic (CHINARE2017–01–01, CHINARE2017–04–01, CHINARE2017–04–04) and Shandong Provincial Natural Science Foundation, China (No. ZR2016DQ02). We thank Drs. Gareth J. Marshall, Baochao Liu, Yongliang Duan, and Qi Shu for useful discussions. We acknowledge the efforts of the Surface Ocean CO2 Atlas (SOCAT) in synthesis of the global surface ocean CO2 data. SOCAT is an international effort, endorsed by the International Ocean Carbon Coordination Project (IOCCP), the Surface Ocean Lower Atmosphere Study (SOLAS) and the Integrated Marine Biogeochemistry and Ecosystem Research program (IMBER), to deliver a uniform quality-controlled surface ocean CO2 database. The many researchers and funding agencies responsible for the collection of data and quality control are thanked for their contributions to SOCAT. Special thanks are given to Drs. Bronte Tilbrook, Claire Lo Monaco, Hisayuki Y Inoue, and Nicolas Metzl, who collected some of these data. W.-J.C. would like to thank the U.S. National Science Foundation (NSF), NASA, NOAA, and the University of Delaware (internal funds) for supporting his ocean carbon cycling research. T.T. was supported by a grant from the Climate Program Office of the National Oceanic and Atmospheric Administration (NOAA). R.W. was supported by NOAA’s Office of Oceanic and Atmospheric Research (OAR) including the Ocean Observation and Monitoring Division of the Climate Program Office (fund reference100007298)

Author information

Affiliations

  1. First Institute of Oceanography, State Oceanic Administration, Qingdao, 266061, China

    • Liang Xue
    • , Libao Gao
    • , Meng Wei
    • , Kuiping Li
    • , Lin Feng
    •  & Weidong Yu
  2. Laboratory for Regional Oceanography and Numerical Modeling, Qingdao National Laboratory for Marine Science and Technology, Qingdao, 266237, China

    • Liang Xue
    • , Libao Gao
    • , Meng Wei
    • , Kuiping Li
    • , Lin Feng
    •  & Weidong Yu
  3. School of Marine Science and Policy, University of Delaware, Newark, DE, 19716, USA

    • Wei-Jun Cai
  4. Lamont–Doherty Earth Observatory of Columbia University, Palisades, NY, 10964, USA

    • Taro Takahashi
  5. NOAA Atlantic Oceanographic and Meteorological Laboratory, Miami, FL, 33149, USA

    • Rik Wanninkhof

Authors

  1. Search for Liang Xue in:

  2. Search for Wei-Jun Cai in:

  3. Search for Taro Takahashi in:

  4. Search for Libao Gao in:

  5. Search for Rik Wanninkhof in:

  6. Search for Meng Wei in:

  7. Search for Kuiping Li in:

  8. Search for Lin Feng in:

  9. Search for Weidong Yu in:

Contributions

L.X. and W.-J.C. designed this study and prepared the paper. T.T. and R.W. contributed parts of the data and writing. L.G., M.W., K.L., L.F., and W.Y. analyzed the data. All authors contributed to discussion and revision of the paper.

Competing interests

The authors declare no competing interests.

Corresponding authors

Correspondence to Liang Xue or Wei-Jun Cai.

Electronic supplementary material

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41467-018-05443-7

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.