Nutrient variability in Subantarctic Mode Waters forced by the Southern Annular Mode and ENSO


  • Jennifer M. Ayers,

    Corresponding author
    1. Institute for Marine and Antarctic Studies, University of Tasmania, Hobart, Tasmania, Australia
    2. Australian Research Council Centre of Excellence for Climate System Science
    • Corresponding author: J. M. Ayers, Institute for Marine and Antarctic Studies, University of Tasmania, Private Bag 129, Hobart, Tas 7001, Australia. (

    Search for more papers by this author
  • Peter G. Strutton

    1. Institute for Marine and Antarctic Studies, University of Tasmania, Hobart, Tasmania, Australia
    2. Australian Research Council Centre of Excellence for Climate System Science
    Search for more papers by this author


[1] As the primary source of nutrients to the global thermocline, Subantarctic Mode Waters (SAMWs) play a key role in primary production and climate. Here we use repeat hydrographic World Ocean Circulation Experiment/Climate Variability and Predictability data to quantify interannual SAMW nutrient variability and its forcing. Pacific sector SAMW nutrients were significantly correlated with the Southern Annular Mode (SAM) and wind stress curl anomalies associated with a faster meridional overturning circulation (MOC). A stronger MOC results in greater upwelling of nutrients at high latitudes, increased Ekman transport of nutrients equatorward, and subduction of higher preformed nutrient loads in SAMWs. Australian sector SAMWs were significantly correlated with El Niño–Southern Oscillation (ENSO), likely due to its modulation of transport in the East Australian Current extension. Interannual variability in SAMW nutrients impacts downstream tropical export production by as much as 5–12% of the annual mean.

1 Introduction

[2] Subantarctic Mode Waters (SAMWs) form part of the southern limb of the global meridional overturning circulation (MOC), the dominant pathway by which deep water nutrients return to the surface [Sarmiento et al., 2004]. Nutrient-rich deep waters upwell near the Antarctic continent and in the Antarctic Circumpolar Current (ACC) and are subsequently advected equatorward by the wind-driven Ekman flow. While at the surface, they support an inefficient, iron-limited biological pump which leaves behind relatively high nutrient concentrations. SAMWs then subduct on the equatorward flank of the ACC, carrying their moderate-to-high preformed nutrient loads northward at mid-depths throughout much of the world oceans [Palter et al., 2010]. SAMWs are therefore thought to be the primary source of nutrients to the global thermocline [Sarmiento et al., 2004] and support up to 75% of tropical export production [Palter et al., 2010].

[3] Despite the importance of SAMW nutrients to downstream primary production and subsequent carbon drawdown, few studies have addressed their variability. Pollard et al. [2006] used data from 11 World Ocean Circulation Experiment (WOCE) sections around Antarctica to calculate the climatological supply of nitrate and silicate from depth to the surface waters of the high-latitude Southern Ocean, but did not attempt to address temporal variability with the data available at that time. Álvarez et al. [2011] looked at nutrient variability along a 32°S repeat section in the Indian Ocean, finding spatial variability between western and eastern Indian Ocean mode waters, as well as temporal decoupling of these locations. Despite its relevance, SAMW nutrient variability remains largely uncharacterized on a circumpolar scale.

[4] The unresolved variability and importance of SAMW nutrients to global export production and climate motivate our research question: How much do SAMW nutrients vary spatially and temporally, and what processes control this? Herein we use the available WOCE and CLIVAR (Climate Variability and Predictability) hydrographic data to characterize the spatial and temporal variability of nitrate, phosphate, and silicate in SAMWs around Antarctica. We then consider these nutrient data in relation to modes of climate variability and the wind-driven circulation in the region, and discuss implications of SAMW nutrient variability for downstream primary production.

2 Observed Variability in Subantarctic Mode Water Nutrients

[5] WOCE and CLIVAR hydrographic sections that sampled nutrients across significant SAMW formation regions were used to characterize spatial and temporal variability in SAMW nutrient concentrations (Figure 1). Nutrient data were adjusted as per the Pacific Ocean Interior Carbon (PACIFICA) Data Synthesis Project ( Bottle data were interpolated vertically onto neutral density surfaces [Jackett and McDougall, 1997] every 0.02 kg m−3 from 25.9 to 28.2 kg m−3, then interpolated every 0.25° meridionally or zonally, depending on the orientation of the section. Mode waters were identified using a potential vorticity minimum criterion (PV < = 1.5 × 10−9 m−1s−1) [Herraiz-Borreguero and Rintoul, 2011], and the mean and standard deviations of temperature and nutrient concentrations in the mode waters were calculated. Figure 2 shows the seasonal time series of these data, spanning the years 1991–2009 and colored by the Southern Ocean sector.

Figure 1.

Southern Ocean hydrographic sections used in this study, colored by sector: Pacific in black, Australian in red, and Indian in blue. Climatological Polar Front (PF) and Subantarctic Front (SAF) locations are overlain. SAMW formation regions are indicated by mixed layer depths (MLDs) > 400 m. MLD data are from Dong et al. [2008].

Figure 2.

Time series of nutrient and temperature variability in SAMWs from 1991 to 2009, colored to match Figure 1. Error bars indicate one standard deviation of the mean. X axis indicates hydrographic sections sampled in the Pacific (black) and Indian (blue) Basins, with a star distinguishing multiple sections in the same season. For readability, Australian sections (red) are not labeled, as all are SR03 except for P11A, which is starred.

[6] The data show significant spatial as well as temporal variability. Pacific sector SAMWs are consistently colder and higher in nutrients than Indian and Australian sector mode waters. Indian sector mode waters appear to have similar characteristics to those downstream, south of Tasmania. Within any given region, SAMWs show significant interannual variability. In the Pacific sector, the mean nitrate concentration was 21.5 µmol kg−1, with an interannual range of 3.1 µmol kg−1, equal to about 15% of the mean. Phosphate varied similarly, with an interannual range of about 17% of the mean of 1.5 µmol kg−1. Not all nutrients varied proportionally, however. In the same region, mean mode water silicate was 8.7 µmol kg−1, with an interannual range of ±50%. In the Australian sector, the interannual range of nitrate was ~16%, phosphate ~25%, and silicate ~28% of the mean.

3 Potential Drivers of Variability

[7] The dominant mode of atmospheric and oceanic variability in the Southern Hemisphere is the Southern Annular Mode (SAM), characterized by large-scale fluctuations of the circumpolar wind fields [Thompson and Solomon, 2002]. Since the mid-1960s, the SAM has trended positive [Marshall, 2003], attributable to ozone depletion and anthropogenic carbon emissions, and resulting in broad changes in the Antarctic climate. Much attention has been paid to the expectation that the positive trend in the SAM will continue, though more recent studies suggest that the trend may slow, stop, or reverse in the 21st century as ozone levels recover [McLandress et al., 2011; Polvani et al., 2011]. Regardless of any future trend in the SAM, its dominant time scale of variability is interannual, and it drives large-scale changes in the Southern Ocean on this time scale.

[8] The response of the Southern Ocean circulation to the SAM and associated changes in wind forcing has received much attention [Hall and Visbeck, 2002; Oke and England, 2004; Sen Gupta and England, 2006]. During the positive phase of the SAM, the westerly winds strengthen and shift poleward (~55°S–60°S), leaving anomalously weak westerlies farther north [Hall and Visbeck, 2002]. Intensified westerlies result in negative wind stress curl (WSC) anomalies south of the maximum wind stress, greater northward Ekman transport throughout the region of the westerlies, and positive WSC anomalies north of the maximum wind stress [Hall and Visbeck, 2002; Saenko et al., 2005]. Through mass continuity, resulting anomalous surface divergences and convergences drive increased upwelling south of and in the ACC and increased downwelling to the north. The overall result is a strengthening and speeding up of the southern limb of the MOC [Saenko et al., 2005]. Though increased winds also increase the eddy return flow, its response is not likely to be large enough to compensate [Saenko et al., 2005; Palter et al., 2010].

[9] Figure 3 summarizes these main effects of a positive-phase SAM on the Southern Ocean circulation, as well as our hypotheses for its consequences for SAMW nutrients (following Lovenduski and Gruber [2005], Saenko et al. [2005], and Palter et al. [2010]). Increased wind-driven divergence forces increased upwelling in the Antarctic Zone (AZ) and Polar Frontal Zone (PFZ), bringing macronutrients and iron to the surface at a faster rate. Stronger westerlies drive increased Ekman transport equatorward, carrying these nutrients across the ACC. Though small relative to mean flow, Ekman flow has been shown to transport nutrients across strong jets when acting cross gradient [Ayers and Lozier, 2010]. Finally, enhanced convergence results in greater downwelling in the Subantarctic Zone (SAZ) and more rapid formation of SAMWs, which subduct with higher preformed nutrients. This faster rate of SAMW production is seen observationally as a decrease in SAMW age from the early 1990s to the late 2000s [Waugh et al., 2013], as well as in modeling studies [Oke and England, 2004].

Figure 3.

Schematic showing the southern limb of the meridional overturning circulation (black), its response to the westerly wind anomalies associated with a positive phase Southern Annular Mode (blue), and the expected impact on nutrients (red) (based on Lovenduski and Gruber [2005] and Sarmiento et al. [2004]).

[10] While at the surface, the extra nutrients may be used by biological production, which could mask the higher nutrient signal expected in SAMWs. However, any biological response is likely to be small due to light and iron limitation, as well as a shorter window of opportunity for nutrient utilization with a faster MOC.

[11] We therefore expect years of positive SAM and stronger MOC to be associated with higher preformed nutrient concentrations in SAMWs. We test for this association by linearly detrending all data and then correlating SAMW nutrient concentrations with the following wind-related parameters:

  1. [12] SAM index [Marshall, 2003], as the dominant driver of Southern Hemisphere variability;

  2. [13] WSC in the AZ, where a negative WSC anomaly indicates stronger upwelling;

  3. [14] WSC in the PFZ, also as an indication of upwelling strength;

  4. [15] Wind speed in the PFZ, as an indication of Ekman transport strength;

  5. [16] WSC in the SAZ, where a positive WSC anomaly indicates stronger downwelling; and

  6. [17] El Niño–Southern Oscillation (ENSO), as the secondary driver of Southern Hemisphere variability [Sallée et al., 2008].

[18] Though WSC is not a measure of the meridional overturning, for our discussion we use it as an approximation, based on the dynamics described above [Hall and Visbeck, 2002; Saenko et al., 2005]. We evaluated variability in the Pacific and Australian sectors separately due to the geographic differences in mode water properties. We did not consider the Indian sector due to sparse data. Significant data correlations discussed in the text are shown in Table S1 and Figure S1 in the supporting information.

4 SAMW Nutrient Response to Forcing

4.1 Pacific Sector

[19] Pacific Basin data support the hypothesis that a positive SAM and stronger MOC force higher nutrient concentrations in SAMWs. A positive SAM was significantly correlated with mode water nutrients at a lag time of three seasons, likely attributable to the time needed for the ocean to respond to the winds. For a one standard deviation increase in the SAM index, nitrate and phosphate increased by 3% (R = 0.66 and 0.83, p < 0.07 and 0.01), while temperature decreased by ~0.5°C (R = −0.90, p < 0.00). Silicate, however, showed no significant correlations. Though correlations between SAMW properties and the SAM index are in a direction to support the hypothesis, a closer relationship is expected with the wind fields that are associated with the SAM but more directly impact circulation.

[20] All SAMW nutrients showed consistent, significant relationships with changes in WSC throughout the region (Figure 4). WSC was calculated from cross-calibrated multiplatform (CCMP) winds [Atlas et al., 2011] and averaged across the Pacific Basin from 160°E to 80°W, within the following zones: the AZ (between the continent and the Polar Front), the PFZ (between the Polar Front and Subantarctic Front), and SAZ (between the Subantarctic Front and the Subtropical Front). Front data were from Sokolov and Rintoul [2009].

Figure 4.

Change in SAMW nutrients (in % concentration, left axis) and temperature (in °C, right axis) per one standard deviation change in negative wind stress curl (−WSC) in the three Antarctic zones. Increased upwelling (more negative WSC) in the AZ and PFZ is associated with higher nutrients and lower temperatures in SAMWs (left and middle). Increased downwelling in the SAZ (more positive WSC) is associated with higher nutrients and decreased temperatures in SAMWs (right). Correlations are significant at p < 0.05, except those marked with an asterisk, which are significant at p < 0.10.

[21] In the AZ, increased upwelling (more negative WSC) was significantly correlated with an increase in SAMW nutrients and a decrease in SAMW temperature (Figure 4, left). For a one standard deviation decrease in WSC (increase in upwelling), SAMW nitrate and phosphate increased by 4–5% (R = −0.64 and −0.76, p < 0.09 and 0.03), while silicate increased by ~14% (R = −0.77, p < 0.03). We discuss the greater change in silicate relative to nitrate and phosphate in section 5. Moving away from the continent into the PFZ and ACC, increased upwelling (more negative WSC) was again significantly correlated with an increase in SAMW nutrients and a decrease in SAMW temperature (Figure 4, middle). We found less consistent relationships between SAMW properties and wind speed in the PFZ, but the correlations found (with silicate and temperature) were consistent with the hypothesis. Finally, in the SAZ, greater downwelling (more positive WSC) was significantly correlated with increased nutrients in SAMWs (Figure 4, right). Lag times decreased from up to ~1 year in the AZ to two to three seasons in the SAZ (Table S1), likely reflecting proximity to the mode water formation region as well as the ocean's response time to the wind forcing. As cruises that overlapped multiple seasons were analyzed as occurring in the single season in which most time was spent, lag times within ±1 season may not be distinct.

[22] Finally, though ENSO events are known to affect the Pacific sector of the Southern Ocean more strongly than other regions [Sallée et al., 2008], we found no significant relationship between SAMW nutrients in this region and ENSO events, either as determined by the multivariate ENSO index (MEI) [Wolter and Timlin, 1998] or the Southern Oscillation index [Ropelewski and Jones, 1987].

4.2 Australian Sector

[23] In contrast to the Pacific, our analysis in the Australian sector did not show that changes in WSC associated with changes in meridional overturning forced SAMW nutrient concentrations. We found no consistent correlations between SAMW nutrients and either the SAM or regional WSC. WSC was averaged meridionally in the same zones as used for the Pacific (AZ, PFZ, SAZ) and zonally over several potential regions of influence: (1) south of Tasmania, between 130°E and 160°E; (2) over the eastern Indian Ocean, from 80°E to 130°E, upstream of the mode water formation region; and (3) over the southern Indian Ocean, from 30°E to 130°E. The lack of a relationship between nutrients and the wind-driven circulation could be due to the influence of meanders and eddies shed from the ACC and the East Australian Current (EAC) [Herraiz-Borreguero and Rintoul, 2010]; however, removing the eddy-influenced November 1991 and July 1995 sections they identified did not improve the correlations.

[24] In contrast, ENSO was significantly correlated with Australian sector SAMW nutrients. El Niño events (+MEI) were found to be followed by a decrease in SAMW nutrients and an increase in SAMW temperatures at a lag time of two to three seasons. Though the correlations are strong (nitrate: R = −0.66, phosphate: R = −0.65, silicate: R = −0.58, temperature: R = 0.75, all significant at p < 0.05), indicating that ~34–56% of variability is associated with ENSO, the changes are not large. Per one standard deviation increase in the MEI index, nitrate decreased ~2%, phosphate ~3%, silicate ~4%, and temperature increased ~0.1°C.

[25] Though observations remain limited, evidence is mounting that ENSO influences the EAC and its extension. The EAC extension, in turn, influences Australian sector mode water properties [Herraiz-Borreguero and Rintoul, 2011]. Peterson and White [1998] suggested that ENSO-associated anomalies in the equatorial western Pacific communicate with the Southern Ocean via the EAC. El Niño events have been linked to WSC anomalies that strengthen the South Pacific subtropical gyre, which includes the EAC as its western boundary current, at lag times of ~3 years [Hill et al., 2011; Holbrook et al., 2011]. A strengthened EAC would decrease SAMW nutrients by bringing more nutrient-poor water into the formation region. Finally, modeling studies have shown that on interannual to decadal time scales, ENSO-forced westward propagating Rossby waves also force EAC transport variability [Sasaki et al., 2008; Holbrook et al., 2011].

[26] We suggest that ENSO forces Australian sector SAMW nutrient concentrations via its control on the strength of the relatively nutrient-poor EAC extension, though the mechanisms responsible for the two to three season lag time remain unclear. Likewise, we attribute the lack of a relationship between SAMW nutrients and the wind-driven circulation to the greater regional influence of the EAC.

5 Why the Greatest Change in Silicate?

[27] In the Pacific sector, accelerated wind-driven overturning was associated with a greater increase in SAMW silicate than nitrate and phosphate (Figure 4). We propose that the decoupling of silicate from nitrate and phosphate is due to the effect of iron limitation on diatoms. Under the mean, iron-limited conditions of the Southern Ocean, Si(OH)4:NO3 uptake ratios are ~5:1 [Franck et al., 2000; Brzezinski et al., 2003], resulting in waters depleted in silicate relative to nitrate. This leads to characteristically high-nitrate, low-silicate SAMWs [Sarmiento et al., 2004]. In the scenario of a faster MOC, the flux of macronutrients and iron to the surface increases. Laboratory experiments show that under Fe-replete conditions, the Si(OH)4:NO3 uptake ratio is closer to 1:1 [Franck et al., 2000]. Though the Southern Ocean remains far from being Fe-replete in times of faster overturning, any relaxation of Fe limitation would lower the Si(OH)4:NO3 uptake ratio, leaving more silicate in the water. This excess silicate relative to nitrate would be reflected in SAMWs and could explain our results. This change in SAMW nutrient ratios may have consequences for downstream phytoplankton community structure, potentially favoring diatoms in the iron and silicate co-limited equatorial Pacific [Brzezinski et al., 2011].

6 Implications for Low-Latitude Productivity

[28] How significantly does variability in SAMW nutrients affect downstream production and carbon drawdown? SAMW nutrients fuel 33–75% [Palter et al., 2010] of the ~3 Pg C yr−1 [Dunne et al., 2007] of tropical export production (30°N–30°S), or approximately 1.1–2.5 Pg C yr−1. For scale, this is larger than the equatorial Pacific (15°S–15°N) export production of 1.09 Pg C yr−1 [Dunne et al., 2007] and comparable to half to all of global anthropogenic carbon emissions in 2011 (2.58 Pg C yr−1; Given the magnitude of tropical export production attributable to SAMW nutrients, even fractional changes in the nutrient supply can be significant. We estimate that a 14% change in SAMW silicate concentrations (associated with a one standard deviation change in WSC) would effect a change of ±0.15 to 0.35 Pg C yr–1 in tropical export production, which is equal to ~5–12% of the mean. Though small compared to the global particulate organic carbon export (~10 Pg C yr−1) [Dunne et al., 2007], the variability is still notable, as large as half of China's 2011 anthropogenic carbon emissions ((677 Tg C yr−1); from Carbon Dioxide Information Analysis Center, as above).

7 Summary and Conclusions

[29] This study used observational data from WOCE/CLIVAR hydrographic sections to characterize and investigate drivers of interannual variability in SAMW nutrients. In the Pacific sector, a positive phase SAM and WSC anomalies associated with faster meridional overturning were correlated with higher preformed nutrients in SAMWs. In the Australian sector, SAMW nutrients were significantly correlated with ENSO, likely attributable to its impact on the East Australian Current extension. As SAMW nutrients support a substantial amount of downstream low-latitude primary production and carbon drawdown, our contribution toward quantifying their interannual variability and understanding their controls is a step toward understanding future climate variability.


[30] This work was supported by the Australian Research Council Centre of Excellence for Climate System Science.

[31] The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.