Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 A possible imprint on equatorial Pacific sediments of a deglacial reinvigoration of the Southern Ocean overturning is increased opal accumulation rate. This would arise from the transmission of silica-rich deep water to the equatorial thermocline via Subantarctic Mode Water and an associated increase in diatom productivity. In search of this imprint, sediment cores from the central (TT013-PC72) and eastern (V19-30) equatorial Pacific have been analyzed for 230Th-normalized opal accumulation rates over the past five and three glacial terminations, respectively. Equatorial opal accumulation rates sustained relatively low values over much of the records and were punctuated by large increases centered on some terminations, but not all. Furthermore, two periods of increased opal flux were observed that do not coincide with terminations. Sources other than the Southern Ocean may need to be considered in the silica budget of the equatorial Pacific, but the δ13C of Neogloboquadrina dutertrei can be used to support the presence of a deepwater nutrient signal in each case. Although a common deglacial mechanism, or a common imprint thereof, for each of the late Pleistocene glaciations remains elusive, the combination of opal flux and δ13C of N. dutertrei provides a diagnostic for past injection of deepwater nutrients into the Equatorial Undercurrent.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Recent concern over Earth's changing climate has motivated research of Earth's past; knowledge of the mechanics and consequences of climate change through study of past events provides clues to the future behavior of a very large and complex system. The relationship through the late Pleistocene between atmospheric CO2 and global climate indicators such as ice volume and deep sea temperature as revealed by ice cores [Petit et al., 1999] and deep sea sediments [Shackleton, 2000] signaled CO2 as a key player in Earth's climate system. Anderson et al.  put that key player to work in the scene of a particular mechanism for the end of the last ice age. Southward shifted and intensified southern hemisphere westerly winds have been inferred from their effect on the diatom community in the Southern Ocean during the last deglaciation (beginning around 18,000 years ago). This change in the winds ventilated the CO2-rich deep ocean, causing a rise in atmospheric CO2, ultimately contributing to warming and interglacial conditions [Denton et al., 2010]. Could a similar process be responsible for the end of previous glacial periods over the past 500,000 years?
 The hydrographic connection between the Southern Ocean and the equatorial upwelling areas of the Pacific affords a testable prediction about the above mentioned deglacial mechanism. The abyssal carbon dioxide-rich waters upwelled in the Southern Ocean by the westerlies would also be rich in dissolved silicon (also referred to as silica, silicate, Si(OH)4, or dissolved Si). If some of that high-silica signal made it to the equatorial upwelling zones, an associated increase in the production of diatom frustules and therefore increased opal (biogenic silica) flux to the sediments would be expected on glacial terminations. This study presents two records of 230Th-normalized opal accumulation rate, covering 250,000 years in the eastern equatorial Pacific (EEP) and 475,000 years in the central equatorial Pacific (CEP). The data support the Southern Ocean upwelling deglacial scenario in some but not all of the recorded terminations. Either the processes resulting in the end of each of the late Pleistocene ice ages were not identical or the record of such processes is not always faithfully recorded in equatorial sediments. The interpretation of paired opal flux and δ13C of N. dutertrei (δ13CN. dutertrei) records offers insight into the source of nutrients which cause the observed diatom productivity events.
 Our analysis takes a simplified view of the relevant circulation and hydrography. As southward flowing circumpolar deep water rises to the surface around Antarctica, some of this water moves northward by Ekman transport where it can be added to the deep winter mixed layers [Dong et al., 2008] which entrain waters from below and eventually form the northward flowing Antarctic Intermediate Water (AAIW) and Subantarctic Mode Water (SAMW) near the subantarctic front [Speer et al., 2000; Sloyan and Rintoul, 2001; Sloyan et al., 2010]. SAMW spreads northward to feed into the eastward traveling Pacific Equatorial Undercurrent (EUC) [Goodman et al., 2005], a thermocline current [Cromwell et al., 1954] whose core is at a water depth of 100–140 m [Wyrtki and Kilonsky, 1984], to finally surface in the east [Toggweiler et al., 1991]. Figure 1 displays the replenishing effect of this upwelling on surface water silicate, out of which diatoms build their opaline frustules.
 The production of biogenic opal by diatoms in the equatorial Pacific today is generally limited by the availability of dissolved silica [Dugdale and Wilkerson, 1998], although the growth of individual diatoms and Si uptake rates are colimited by iron [Moore et al., 2004; Brzezinski et al., 2008]. The status of iron conditions may have been different in the past ocean [Winckler et al., 2008]. Nonetheless, iron-fertilized ocean productivity during the last glacial period was not pervasive in the Southern Ocean or in the equatorial Pacific [Kohfeld et al., 2005]. With regard to deglaciations, iron supply to the Southern Ocean from atmospheric dust [Wolff et al., 2006] began declining rapidly early during terminations, while enhanced upwelling may have increased the supply of dissolved iron to surface waters [Watson et al., 2000]. Given these uncertainties, we refrain from commenting on changes in Fe input and assume that the addition of silicate to the equatorial Pacific euphotic zone from the Southern Ocean via SAMW-routing is expected to enhance diatom productivity and subsequently increase opal flux to the sediments.
 This routing, or “oceanic tunneling” between the Antarctic and equatorial Pacific has been traced on glacial-interglacial timescales with carbon isotopes [Spero and Lea, 2002; Pena et al., 2008]. Additional evidence for the transfer of climate signals from the Antarctic to low latitudes, routed by SAMW, was found in the Indian Ocean [Kiefer et al., 2006]. The chemical signature of upwelling deep water should be high CO2, high Si(OH)4, and low δ13C of dissolved inorganic carbon (DIC, δ13CDIC). Spero and Lea  explained the deglacial carbon isotope minimum found in foraminfera records throughout the Southern Hemisphere and in the EEP as a result of the deglacial reinvigoration of the Southern Ocean overturning, the mechanism later invoked by Anderson et al.  to explain the deglacial peak in opal flux in Southern Ocean sediments. The deep-dwelling planktic foraminifera N. dutertrei is especially fit to record a SAMW-routed signal as it tends to live in a depth zone (60–140 m) that intersects the EUC [Fairbanks et al., 1982]. The δ13C of planktic foraminferal calcite is not always easily interpreted strictly as a nutrient proxy because it can be influenced by, among other mitigating factors, isotope fractionation during air-sea gas exchange [Lynch-Stieglitz et al., 1995] and the microenvironment of calcification [Spero et al., 1997]. Thus to evaluate the events of past terminations, we pair these two proxies to trace the Southern Ocean signal: δ13C and opal accumulation rate. Note that opal accumulation rates are not without their own limitations as discussed in section 4.3.
 The modern equatorial Pacific thermocline is ventilated by mode/intermediate waters from northern and southern subpolar/subtropical regions. Of the waters ending up in the EUC [Goodman et al., 2005; Khatiwala, 2007], southern subpolar/subtropical waters contribute about 70% of the volume while the counterpart northern sources, including North Pacific Intermediate Water (NPIW) contribute the remaining 30%. In terms of nutrient supply, Dugdale et al.  concluded that Northern sources constitute 70% (Southern sources constituting 30%) of the silicate supply to the EUC. Using the names NPIW and SAMW to represent northern and southern nutrient sources, NPIW is at least 5 times more concentrated in silicate than SAMW, fairly consistent with observations [Garcia et al., 2006]. Additionally, the northern Pacific waters which form NPIW have a lower δ13CDIC than those in the south forming SAMW, about 1‰ lower [Kroopnick, 1985; Quay et al., 2003]. Yet, our working assumption is that deglacial upwelling in the Southern Ocean would drastically increase the concentration of silicate and decrease δ13CDIC in SAMW. Put another way, we assume that deglacial SAMW was the largest source of silica to equatorial diatom communities and was almost entirely responsible for the δ13C excursion seen in equatorial N. dutertrei communities. This assumption is supported by the Nd isotopic composition of EUC waters, reconstructed using N. dutertrei, which indicate increased advection of southern-sourced waters during the last deglaciation [Pena et al., 2010]. Although variability in the production of NPIW over glacial-interglacial scales has been documented [Hendy and Kennett, 2003; Hendy and Pederson, 2006], we proceed here in the absence of detailed information about how Pleistocene NPIW has varied in its silicate and 13C contents. The only connection between modern NPIW and deep water is through slow diapycnal mixing [Talley, 1993]. If past NPIW were similar, signs of deep water (low δ13CN. dutertrei and high opal flux) are best ascribed to the Southern Ocean, where much stronger vertical mixing connects surface and deep water in the modern ocean.
 From the work of Bradtmiller et al. , a strong peak in opal accumulation rate is centered on Termination I for both cores in this study, coeval with a minimum in the δ13C record from V19-30. We now look further back in time at Terminations II and III for V19-30 and Terminations II-V in TT013-PC72.
3. Materials and Methods
 Core V19-30 (3.383°S, 83.517°W) is from a water depth of 3091 m and core TT013-PC72 (0.11°N, 139.4°W) from 4298 m. Both cores have excellent benthic δ18O records to define their stratigraphies and develop their age models (see Figure 2 caption for references). V19-30 has an average sedimentation rate of 7 cm/kyr; it was analyzed at ∼0.5 kyr resolution in the 22 ka to 60 ka section and 2 kyr resolution in the section older than 60 ka. TT013-PC72 has an average sedimentation rate of 2 cm/kyr; it was analyzed at a resolution of ∼1.3 kyr in the 22 ka to 40 ka section and at 3.5 kyr resolution in samples older than 40 ka.
 The 230Th-normalized opal fluxes were calculated as % biogenic opal (dry weight), measured by the alkaline extraction and molybdate-blue spectrophotometry of Mortlock and Froelich , multiplied by the 230Th-normalized mass accumulation rate. Concentrations of U and Th isotopes, as well as 231Pa, were measured by inductively coupled plasma mass spectrometry after strong acid sediment digestion and chromatographic separation as described by Fleisher and Anderson . The concept of excess, or unsupported, 230Th as a constant-flux proxy and the calculations used for determining 230Th-normalized fluxes are described in detail by Francois et al. . Most importantly, normalizing to 230Th corrects accumulation rates for lateral redistributions of sediments, or sediment focusing. This procedure also allows calculation of the initial excess 231Pa:230Th ratio, or (231Pa/230Th)xs,o. This ratio is elevated in the opal phase as it falls through the water column [Chase et al., 2002] and is impervious to dissolution at the seafloor [Kumar et al., 1993]. It is thus used here to discern possible preservation biases in the derived opal fluxes [Bradtmiller et al., 2006]. In V19-30, the excess 230Th data were collected in samples older than 250 ka; however they are not reported here because uncertainties become too large due to a relatively large ingrowth correction for production of 230Th by decay of authigenic uranium.
4. Results and Discussion
 The 230Th-normalized opal fluxes, or mass accumulation rates, are displayed in Figure 2 along with the benthic oxygen isotopic records for both cores and the carbon isotopic record of N. dutertrei from V19-30. The oxygen records define the glacial terminations as periods of rapid decrease in δ18O (highlighted in light green bars in Figure 2). The δ13C record from the eastern site helps confirm a deepwater signal, likely from the Southern Ocean (when minima in the δ13C and maximum in the opal flux occur simultaneously). As a caution, this interpretation of the δ13C record may need a caveat related to the effects of local wind-driven upwelling in the equatorial ocean as discussed in section 4.5. Results reported here that have not been published previously will be archived at the U.S. National Climate Data Center.
 Termination II (T-II) shows increased opal flux at both sites, mirroring a very strong minimum in the δ13C (Figure 2) [Spero and Lea, 2002; Pena et al., 2008]. However, in TT013-PC72, the opal flux peak is broad and appears to begin well before the deglaciation. It is possible that the low resolution of this core obscures two separate events: one deglacial peak and a separate peak prior to T-II (similar to the peak in opal flux prior to T-III; see section 4.2).
 Termination III is the first in a two-step transition from glacial to interglacial conditions, seen in the δ18O records (Figure 2) as a series of two moderate decreases from higher to lower values. Cheng et al.  suggested that the second decrease event, one precession cycle after T-III, deserved terminology as T-IIIA. During T-III, V19-30 displays a peak in opal flux which is also well-correlated with a minimum in the δ13C. Another opal flux peak is apparent in V19-30 at 231–232 ka, slightly earlier than the δ13C minimum at around 225 ka which could tentatively be assigned to T-IIIA. The opal fluxes in TT013-PC72, however, remain low throughout T-III and T-IIIA. The resolution in this record may have been too low to reveal any smaller, more rapid events during this period (250–215 ka).
 Opal fluxes for Terminations IV and V (T-IV and T-V) are confidently recorded only by TT013-PC72. The δ13C extended to the very end of the V19-30 core truncates what is likely another strong minimum centered on T-IV, and TT013-PC72 shows elevated opal fluxes centered on T-IV. Termination V, the strongest termination over this time period on the basis of the change in δ18O, paradoxically shows little significant change in the opal flux, merely a small peak at the end of this termination. To summarize, T-I, T-II and T-IV show evidence of opal flux maxima and δ13C minima centered on the terminations. These findings are consistent with a deglacial mechanism invoking a reinvigorated Southern Ocean overturning that injects nutrient-rich deep water into the thermocline of the Southern Hemisphere, increasing the supply of silica and 13C-depleted DIC into the EUC and eventually into equatorial Pacific upwelling zones. Inconsistent with this mechanism, T-III shows opal flux maxima at only one of the two study sites and T-V shows no opal flux maxima at the one site covering this interval.
4.2. Opal Flux Maxima During Glacial Periods
 Our opal flux records also show interesting features unrelated to the terminations (∼50 ka and ∼255 ka; highlighted in pink bars in Figure 2). The largest opal fluxes throughout the entire history of V19-30, in fact, are associated with a large peak during the last glacial period at ∼50 ka, preceded by a smaller maximum at ∼60 ka. Opal flux maxima at ∼50 ka (Figure 2c) and at ∼255 ka (Figure 2d) are each accompanied by a minimum in δ13CN. dutertrei (Figure 2b), supporting the view that increased diatom productivity and opal flux resulted from increased supply of nutrient-rich deep water into the EUC. This is analogous to the relationship proposed for terminations, although the processes and conditions that increased the supply of nutrient-rich deep water clearly must have been different.
S. S. Kienast et al.  reported a similar feature at ∼50 ka in the 230Th-normalized opal fluxes of core TR163-31 from a site near V19-30. That study explained this peak, also seen in other sites from the EEP, as a consequence of silica leakage, whereby extended sea ice and low dust flux limited diatom productivity in the Southern Ocean. Limited diatom productivity allowed dissolved silica to escape northward toward the equator by the same SAMW routing invoked for terminations, but not related to changes in the rate of deep water upwelling.
 A contrasting interpretation was offered by Dubois et al. , who attributed the opal flux maximum at ∼50 ka to enhanced opal preservation. The primary evidence in support of this interpretation is the absence of an increase in (231Pa/230Th)xs,o during the opal flux maximum in most of the cores studied. We disagree with this interpretation for two reasons. First, (231Pa/230Th)xs,o is well correlated with opal flux across the 50 ka event in V19-30 (Figure 3a). Furthermore, there is no reason to expect increased opal preservation to be accompanied by lower δ13C values of N. dutertrei, a relationship that holds up throughout each opal flux peak (Figure 2). The consistent relationship between opal flux maxima and lower δ13C values of N. dutertrei suggests that each event was caused by increased diatom productivity fueled by a greater supply of nutrients via the EUC.
 Why the 50 ka event does not appear in the CEP (TT013-PC72) is unclear. Assuming that a higher silica signal originated in the eastern surface waters during this time period, the distance toward the west that this signal could travel is dependent of the strength of equatorial upwelling which, in turn, can vary with the position of the intertropical convergence zone (ITCZ). The presence of the opal flux peak at 50 ka in the EEP and its absence in the CEP may indicate enhanced Si supply in the EEP without enhanced equatorial upwelling. Correspondingly, this would imply an increase in Si concentration carried by the EUC. These ideas are worthy of further work.
 Can different conditions in the Southern Ocean (e.g., increased upwelling on terminations versus periods of low silica utilization during glacial periods) lead to similar responses in the equatorial Pacific (i.e., increased diatom productivity and opal burial)? Two lines of reasoning may help distinguish these processes based on (1) reconstructions of conditions in the glacial Southern Ocean and (2) the consistent relationship between opal flux maxima and the δ13C minima observed throughout our records (Figure 2).
 First, existing records of Southern Ocean Th-normalized opal fluxes covering Marine Isotope Stage 3 (MIS3) [Frank et al., 2000; Anderson et al., 1998] display no peak in opal flux at 50 ka either north or south of the Antarctic Polar Front (APF). Therefore, in our interpretation, no enhancement of Southern Ocean upwelling occurred at this time. Records north of the APF reported by Anderson et al.  (RC13-254, VM22-108) and by Frank et al.  (PS1754, PS1756, PS2082) show that opal fluxes at 50 ka are higher than during the Holocene, but they are no greater than during MIS2. Thus upwelling was no greater during MIS3 than during the MIS2. The 50 ka peak in opal flux in the EEP must be related to factors other than changes in the rate of deep water upwelling in the Southern Ocean.
 Second, it is not clear that low silica utilization in the Southern Ocean would coincide with the transmission of very low δ13CDIC water to the EUC. The interpretation of paired δ13C-opal flux records is discussed further in section 4.5 but for now it is worth noting how nutrient utilization by diatoms in the Antarctic Zone is related to the formation of SAMW. Today Antarctic diatoms consume all of the available silica during summer and the associated utilization of DIC does increase the residual δ13CDIC of Antarctic surface waters, as shown by seasonal time series data from the JGOFS/AESOPS program (http://usjgofs.whoi.edu/jg/dir/jgofs/southern/). Therefore, relaxing Antarctic silica utilization would cause a decrease in the δ13CDIC of Antarctic surface waters unless other taxa compensated for reduced DIC utilization by diatoms. However, δ13CDIC of SAMW is influenced by two other factors in addition to nutrient utilization in surface waters. First, the Antarctic surface waters that contribute to SAMW are subject to isotope fractionation due to low-temperature air-sea exchange (which increases δ13CDIC) [see Lynch-Stieglitz et al., 1995, Figure 5]. This effect opposes the reduction of δ13CDIC in Antarctic surface waters due to a relaxation of silica utilization. Second, formation of SAMW involves entrainment of thermocline-depth water as well as subduction of Antarctic surface waters [Rintoul and England, 2002; Sallée et al., 2010]. During winter, when SAMW forms, deep mixing combines a chemical signature from Antarctic surface water and from deeper water entrained by mixing. Therefore, the composition of SAMW will be influenced by the Si concentration and the δ13C-character of the deeper water (dependent of the rate of circumpolar deep water upwelling) being entrained into the winter mixed layers in the Subantarctic Zone. New seasonal time series studies from the Antarctic and Subantarctic zones will help to quantify the extent to which Antarctic surface water signals survive to be incorporated into SAMW.
 From this we conclude that the deglacial carbon isotope minima are more likely the chemical signatures of deepwater entrainment rather than of changes in surface water nutrient utilization. The glacial (∼50 and ∼255 ka) diatom productivity events in the equatorial Pacific do not discount the Southern Ocean upwelling deglacial mechanism but offer a separate and interesting problem of how deep water geochemical signatures can be transported from the deep Southern Ocean to the tropical thermocline without producing productivity events at the site of deep water upwelling.
4.3. Fidelity of the Opal Flux Records
 The flux of opal to the sediments in general is an attenuated signal. Much of the opal produced in surface waters is dissolved before burial [Nelson et al., 1995]. Therefore, it is necessary to provide additional information about opal fluxes to determine that they reflect changes in opal production rather than changes in preservation. Here we use two additional pieces of information to confirm that the opal fluxes reported here faithfully record changes in opal production: (231Pa/230Th)xs,o and the total mass accumulation rate (MAR).
 The reconstruction of (231Pa/230Th)xs,o is limited primarily by the half-life of 231Pa (32.5 kyr). As far back as it has been reconstructed in both cores (60 ka in V19-30 and 150 ka in TT013-PC72), the (231Pa/230Th)xs,o ratio correlates well with the 230Th-normalized opal flux (Figure 3). To the degree that the sedimentary (231Pa/230Th)xs,o ratio can be viewed as proxy for opal flux [Chase et al., 2003], for at least these time periods, no large preservation biases are suggested. In V19-30 (Figure 3a), (231Pa/230Th)xs,o peaks to the same value (activity ratio of ∼0.225) during T-I and during the 50 ka opal flux peak while the opal flux peak is smaller for T-I relative to that of the 50 ka event. Although we conclude that the opal flux peak at ∼50 ka was primarily caused by enhanced diatom productivity (section 4.2), we cannot rule out that opal dissolution was somewhat stronger for the T-I peak [Dubois et al., 2010].
 Biogenic opal, being undersaturated throughout the water column in the global ocean [Hurd, 1972; Dixit et al., 2001], has the potential to be preferentially preserved at sites of high MAR or preferentially dissolved at sites of low MAR [Broecker and Peng, 1982]. V19-30 is a site of relatively high MAR and opal makes up between 10 and 40% of the total flux so it is unlikely this core is subject to preservation biases. Additionally, Warnock et al.  determine from a dissolution index involving visual inspection of fossil diatoms that at least for the time interval spanning the last glacial to the Holocene, production, not preservation, controlled opal accumulation rates in the EEP. The MAR at TT013-PC72 is relatively low and according to Murray et al.  all semilabile elements may be subject to diagenetic remobilization at times of low sediment accumulation in the CEP. To test for an influence over opal preservation, we compared the 230Th-normalized opal fluxes to two reconstructions of the total MAR (Figure 4): one by 230Th normalization, the other by the “stratigraphic” technique (calculated by Murray et al. ), whereby the apparent sedimentation rate, derived from age control points, is multiplied by the dry bulk density of the sediment to calculate MAR.
 The age model (as reported by Murray et al. [1995, 2000]) for TT013-PC72 is fairly unique and deserves a few words of explanation. Using an inverse technique for signal correlation [Martinson et al., 1982], the benthic δ18O record of TT013-PC72 was tuned (at 1 kyr intervals) to the SPECMAP benthic stack [Imbrie et al., 1984]. Linear sedimentation rates were iteratively calculated at high resolution in order to maximize coherence between the TT013-PC72 record and the SPECMAP stack. Anderson et al.  showed that there are no offsets in the TT013-PC72 age model when compared to the updated benthic stack provided by Lisiecki and Raymo  for the time periods covered here. The method may obscure random high-frequency variability in sediment accumulation rate. However, the average MAR values calculated by this method integrated over the time scales of the features in the opal flux record should smooth out any short-term errors.
 The “stratigraphic” technique for calculating MAR does not correct for sediment focusing. Therefore the periods when the stratigraphic MAR is larger than the Th-normalized MAR are periods of syn-depositional lateral redistribution of sediment in bottom waters which could have an additional preservation effect on biogenic opal.
 A preservation bias related to MAR for our opal flux records is not apparent. For instance, periods of increased total MAR are not consistently correlated with periods of increased opal MAR. Consequently, while we cannot rule out a secondary influence on the opal flux records caused by variable preservation, we conclude that variable production of opal by diatoms is the primary factor regulating the opal accumulation rates at our study sites. Given that diatom productivity is limited by supply of dissolved silica in this region [Dugdale and Wilkerson, 1998], we further conclude that variability in opal accumulation rate (Figure 2) records past changes in the supply of dissolved silica to surface waters of the equatorial Pacific Ocean.
4.4. Equatorial Pacific Silicate Budgeting
 When interpreting paleoceanographic changes in this basin-wide context, two important constraints on the problem are missing: (1) the relative volumes of northern and southern contributions to EUC waters and (2) the silicate:nitrate ratios in either water mass back through time. Sarmiento et al.  noted that modern SAMW is very depleted in Si relative to NO3− (Si* values below −10 μmol/kg, where Si* = [Si(OH)4] − [NO3−]), while NPIW is very enriched (Si* values approximately 20 μmol/kg).
 With respect to the first problem, budgeting northern versus southern contributions to nutrients carried by the EUC becomes complicated if the relative volumes of source waters have varied over the Pleistocene. From a study in the Santa Barbara Basin, Hendy and Kennett  postulated that production of NPIW consistently varied in concert with the Dansgaard-Oescher events of the Greenland ice core. Furthermore, Hendy and Pederson  showed evidence for increased northward penetration of low-O2 AAIW/SAMW meeting at a physical boundary with NPIW around 20°N during the last deglaciation. For additional control on this variation through time we can use the fact that the isotopic composition of Nd, that is, ɛNd (a quasi-conservative water mass tracer), of planktic foraminifera has been shown to reflect the isotopic composition of seawater [Vance et al., 2004]. The ɛNd of N. dutertrei in the EEP therefore should reflect the relative volume contributions of NPIW (high ɛNd) and SAMW (low ɛNd) to the EUC. Although quantitative estimates of the relative volumes cannot be made due to uncertainty in the constancy of the ɛNd end-members, recent results showed a pronounced influence of SAMW in the EUC during the last deglaciation [Pena et al., 2010]. At the very least, the Nd results support the view that large nutrient supplies were not coming from NPIW during the last termination.
 We address the second problem as it pertains to the Southern Ocean. Today, utilization of Si by diatoms in the Subantarctic is nearly complete, causing a silicate-depleted (though not nitrate-depleted) SAMW source to the EUC. In order for a high-silicate (and low δ13C) pulse to be sent from the deep Southern Ocean to the equatorial thermocline, one of three conditions must have existed: (1) increased supply of iron may have reduced the biological utilization of Si relative to that of N, leaving unused Si to be exported into SAMW (the original SALH hypothesis [Matsumoto et al., 2002; Brzezinski et al., 2002]); (2) a change in ocean circulation permitted deep water upwelled south of the APF during winter to be incorporated into SAMW in a single season, before summer growth of diatoms had a chance to deplete surface waters of their Si; or (3) an increase in deep winter mixing in the region of SAMW formation may have altered the relative proportions of Antarctic surface water and Subantarctic thermocline-depth water combined to form SAMW. The third condition comes with the caveat that this paleo-SAMW mixture must have somehow maintained suitable temperature and salinity characteristics such that it did not become too dense to upwell at the equator. We believe each of these conditions is a plausible explanation of the results and it may be that a different condition applies for the glacial versus deglacial equatorial Pacific opal flux events (and δ13C minima). We focus on the third condition because the first two rely more heavily on the incorporation of Antarctic surface water which, as discussed in section 4.2, is less likely to carry simultaneously a high-silicate and low-δ13CDIC nutrient signal.
 Despite missing constraints, we can try some basic calculations in order to get a feeling for the implications of our working hypothesis. Exercises in mixing Pacific water masses in terms of their geochemical tracer content are presented, following to some degree the work of Bostock et al. , although that study only considers waters of intermediate depth (below the EUC). Suppose the present-day volume split were to remain unchanged (30% NPIW, 70% SAMW). NPIW and SAMW in a strict sense are not supplying the full 30% and 70% volumes to the EUC. The volume split includes contributions from subtropical as well as subpolar waters. The water mass names are used here as shorthand designators for northern and southern sources, respectively. Suppose further that the silicate and 13C properties of NPIW did not significantly change. The chemical composition of deglacial SAMW would then be entirely responsible for the chemical variations observed in the EUC (increased silica supply and negative excursions of ∼0.5‰ in δ13C). We would need SAMW-silicate to go from being 5 times less concentrated than NPIW, as it is today, to approximately 3 times more concentrated than NPIW in order to raise the percentage of southern-sourced silicate supply to the EUC up to approximately 90%. Under this condition, nearly all the silica in the EUC is coming from southern sources and therefore this gives a maximum estimate for the hypothetical enrichment of silica in SAMW. The observed 0.5‰ excursion of equatorial Pacific δ13CN. dutertrei, divided by 0.7, the volume fraction of SAMW, implies about a 0.7‰ drop in δ13CDIC for SAMW. Including a modest (∼5%) change in the relative contributions of DIC by NPIW and SAMW does not significantly affect this estimate. This was considered because SAMW and NPIW have very different δ13CDIC values.
 In this calculation, deglacial SAMW should have been 15 times more concentrated in silicate and about 0.7‰ lower in δ13CDIC than modern SAMW. The drop in δ13CDIC is plausible insofar as it is not larger than observed global variations in modern surface water [Quay et al., 2003] nor is it larger than the surface to deep gradient in δ13CDIC in the subantarctic Pacific (see Figure 5b) [Lamb et al., 1997]. The significance of the silicate increase depends on the value chosen for modern SAMW (estimated from World Ocean Atlas 2005 this could range from 1 to 5 μmol/kg [Garcia et al., 2006]). While an increase from 1 to 15 μmol/kg is plausible, as concentrations in this range are observed in surface waters today and small compared to the surface to deep gradient in the subantarctic Pacific of 140 μmol/kg (see Figure 5b), an increase from 5 to 75 μmol/kg is less plausible as 75 μmol silicate/kg is usually observed only in waters of intermediate depth or deeper. In our calculation we chose a modern value of 2 μmol/kg and correspondingly a deglacial value of 30 μmol/kg. For reference, in North Pacific Deep Water (NPDW) silicate concentrations are about 155 μmol/kg while Upper Circumpolar Deep Water (UCPW) has silicate concentrations near 100 μmol/kg. Of course, any increase in the volume fraction of SAMW would reduce the necessary concentration increase and vice versa. This calculation is admittedly simplistic; nonetheless, Table 1 lays out the calculated effect on the EUC of mixing an unchanged NPIW with the hypothesized deglacial SAMW. As calculated, the change from modern to hypothetical deglacial results in a 0.48‰ carbon isotope excursion, as prescribed, and a fivefold increase in dissolved silica, roughly equivalent to the relative magnitude of changes seen in the N. dutertrei and opal flux records. This simple mixing calculation demonstrates that with a constant mixing fraction of SAMW and NPIW in the EUC, a very large but not unrealistic change in the SAMW end-member silicate and δ13CDIC could be responsible for the sedimentary features highlighted in Figure 2.
Table 1. Approximate Water Mass Mixing Calculations of Northern- and Southern-Sourced Waters in the Equatorial Undercurrenta
4.5. Equatorial Upwelling, Southern Ocean Upwelling, and the Role of the Winds
 Increased incorporation of nutrients into SAMW (section 4.4) can provide the equatorial thermocline with a greater nutrient supply; however, if the equatorial upwelling which returns those nutrients to the surface is reduced, equatorial primary producers will not benefit. The degree of equatorial upwelling could either amplify or diminish a Southern Ocean nutrient signal.
 The data to test for enhanced upwelling in the Southern Ocean during deglaciation for terminations prior to T-II (i.e., Th-normalized opal fluxes from sediment cores south of the polar front) do not yet exist. The record of core PS-1768 given by Frank et al.  shows a peak in Th-normalized opal flux centered on Termination II, albeit at fairly low resolution, signifying deep water upwelling during that termination as well. For terminations T-I through T-IV, the record of the Asian Monsoon [Cheng et al., 2009] does lend support to a deglacial mechanism consistent with enhanced upwelling in the Southern Ocean. The last four terminations were characterized by a period of weak Asian Monsoon likely associated with extreme cold events in the North Atlantic. These cold events or stadial conditions could act as a trigger via atmospheric teleconnections to eventually push the Southern Hemisphere westerlies southward and enhance the wind-driven upwelling that contributes to deglaciation [Denton et al., 2010]. This teleconnection is consistent with the high-resolution δ13C record from benthic foraminifera on the Chatham Rise, which exhibits a shift toward more positive values at a depth of 1210 m during each major Northern Hemisphere stadial, including those coinciding with terminations [Pahnke and Zahn, 2005]. Such a relationship is expected if a southward shift in the position of the westerlies moves the boundary between AAIW and Upper Circumpolar Deep Water (UCDW) either southward or to greater depths, and vice versa.
 Reinforcing the effect of increased upwelling in the Southern Ocean on low-latitude diatoms during terminations, the alkenone-based SST record from the EEP of M. Kienast et al.  suggests that there was increased equatorial upwelling during the last deglaciation as well. More specifically, they reported two intervals of surface cooling: one during Heinrich Event 1 and again at the Younger Dryas, the same doublet structure seen in Southern Ocean opal fluxes [Anderson et al., 2009].
Pena et al.  interpreted saltier thermocline waters (inferred from reconstructed δ18O of EUC waters) in the EEP as evidence for dominant La Niña-like conditions during terminations I-III. This corroborates the presence of enhanced equatorial upwelling during deglaciation as indicated by deglacial SST coolings [M. Kienast et al., 2006]. Additionally the SST cooling in the area where the EUC upwells today is considered to result almost entirely from the upwelling of EUC waters, rather than advective input of cold waters from the Peruvian coastal upwelling system [Kessler, 2006]. Therefore it is likely that equatorial upwelling was enhanced during deglaciation, such that a nutrient pulse from the Southern Ocean would be available to equatorial plankton.
 Moreover, new biomarker results from the EEP [Calvo et al., 2010] show that the carbon isotope minimum recorded for the last deglaciation coincided with an increase in the diatom to coccolithiphore abundance ratio. This floral change is most likely explained by an increase in the Si:N ratio of the upwelling EUC water. When linked with the Nd isotope-based conclusion that nutrient supplies in the EUC during deglaciation were more likely of southern rather than northern origin, the elevated Si:N ratio implies an increase in the Si content of SAMW. In summary, there is strong evidence for increased upwelling of EUC water (based on SSTs and δ18O of seawater) and for an enhanced Southern Ocean contribution to the EUC which was enriched in Si and depleted in δ13CDIC relative to the Holocene (based on the Si:N ratio and ɛNd of paleo-EUC water) during T-I.
 The deglacial nature of the peaks in equatorial Pacific opal flux is consistent with the idea that increased upwelling in the Southern Ocean increased the end-member (or preformed) Si content of deglacial SAMW. After the last deglaciation and during the Holocene, the supply of Si by upwelling in the Southern Ocean declined [Anderson et al., 2009] most likely due to a northward retreat of the position and/or a decrease in the intensity of the Southern Hemisphere westerly winds [Denton et al., 2010]. With less nutrients upwelled in the Southern Ocean, the end-member Si of SAMW would have returned to a lower concentration and therefore EUC-fed diatom communities would have also declined into the Holocene, as observed. A similar transition (but reverse, δ13CDIC increases into the Holocene) applies to the end-member δ13CDIC of SAMW. The decline of equatorial upwelling through the Holocene [Pena et al., 2008; M. Kienast et al., 2006] could also have contributed to the Holocene decline in equatorial opal flux and the Holocene rise of equatorial δ13CN. dutertrei.
 As a speculative note, the water column profiles of δ13CDIC and Si(OH)4 may offer further insight into the events recorded by paired records of δ13CN. dutertrei and opal flux. The depth of maximum regeneration for DIC (and correspondingly the depth of greatest depletion in δ13CDIC) is always shallower than the depth of maximum regeneration for Si(OH)4. This fact may be of use in distinguishing between changes brought on by increased wind-driven equatorial upwelling and changes brought on by deep water upwelling in the Southern Ocean propagated to the equator via SAMW. As is laid out in Tables 2a and 2b, using the WOCE profiles [Lamb et al., 1997; Rubin et al., 1998] seen in Figure 5, the mixture of surface and thermocline waters (to simulate an exaggerated effect of equatorial upwelling) in the equatorial Pacific and the mixture of SAMW and UCDW in the subantarctic [Macdonald et al., 2009] (to simulate the effect of deep water upwelling in the Southern Ocean) cause a decrease in δ13CDIC (compared to the original surface water in the first case and to unaltered SAMW in the second case) that is similar in magnitude for both mixtures. In contrast, the resultant enrichment in Si(OH)4 concentration is much greater for the Southern Ocean upwelling scenario. Simple as it is, this exercise neglects the effect of increased isotope fractionation due to air-sea exchange (making δ13C less depleted) which would presumably be associated with these mixing scenarios. For 0.6‰ δ13CDIC depletion in both cases, silicate increases by a factor of 3 for the equatorial upwelling scenario while the increase is a factor of 13 for the Southern Ocean upwelling scenario.
Table 2a. Scenario 1: Equatorial Upwelling Dominates Nutrient Status of Equatorial Euphotic Zone, Derived With Data From Equatorial Pacific WOCE Station (Figure 5a)
 Is this finding applicable to the sedimentary record? The structure of deep water masses and their nutrient concentrations may have been very different during different stages of the Pleistocene compared to today. Also, again the assumptions must be made that N. dutertrei reliably recorded δ13CDIC and that silica-limitation of equatorial diatoms persisted so that increased delivery of Si(OH)4 resulted in increased productivity and ultimately increased opal burial. Those considerations aside, using these simplistic calculations, one might expect that for an event characterized by a given depletion in δ13CN. dutertrei, the concomitant increase in opal flux will be smaller for the equatorial upwelling case than the increase expected for the Southern Ocean upwelling case.
 All of the opal flux events displayed in Figure 2 (during terminations or glacial periods) occur with a δ13C excursion of approximately 0.5‰. Strikingly, the proportional increase in opal flux during the 50 ka event in V19-30 is greater than the other events. Could it be that the large opal flux events in glacial periods (∼50 and 255 ka) are the result of a relatively greater injection of deep water into SAMW within the Southern Ocean whereas the smaller opal flux events during terminations are influenced more strongly by the amplitude of the increase in equatorial upwelling? Undoubtedly, the two scenarios could be inextricably mixed and this obfuscates the contrasting effects. Nonetheless, in V19-30, does the fact that T-I and T-II have different opal flux responses imply contrasting mechanisms for those two terminations? The conceptual Si-δ13C pairing strategy is certainly more complex than presented here and is worth future attention. Finally, while it may be possible to explain an excursion of δ13CDIC in the EUC by local wind-driven vertical mixing, in the vein of Spero and Lea , this mechanism alone does not explain the widespread nature of the deglacial carbon isotope excursion throughout the Southern hemisphere ocean at thermocline and intermediate depths.
 The late Pleistocene 230Th-normalized opal flux records presented here do not entirely support nor refute the mechanism in which wind-driven intensified Southern Ocean upwelling accompanied ice age terminations. That most of the recorded terminations occur with a peak in opal flux (T-I, T-II, T-IV), that one prominently does not (T-V), and that an eastern and central equatorial Pacific site disagree for another (T-III) leads to one of two conclusions: either glacial terminations did not occur through identical processes or some combination of factors caused a poor sedimentary record of the proposed events. Two large opal accumulation events appear in the equatorial Pacific during glacial periods (50 ka and 255 ka) and may be explained by enhanced deep water entrainment into SAMW. More detailed constraints on the behavior of NPIW and SAMW, regarding both their distribution and silicate:nitrate content over the late Pleistocene, will elucidate conclusions about the nature and significance of equatorial Pacific opal fluxes during glacial terminations.
 This manuscript benefited from comments made during a discussion with Jerry McManus and Samar Khatiwala as well as from thought provoking reviews by the editor (Rainer Zahn) and three anonymous reviewers. Thanks to Patricia Malone for opal analyses. This work was supported by the U.S. National Science Foundation through award 0526522. Samples from TT013-PC72 and from V19-30 were provided by repositories at the University of Rhode Island and at the Lamont-Doherty Earth Observatory, respectively. This is LDEO Contribution 7429.