Interactive comment on “ Millennial meridional dynamics of Indo-Pacific Warm Pool during the last termination ” by L

Abstract. To develop an in-depth understanding of the natural dynamics of the Indo-Pacific Warm Pool (IPWP) during the last deglaciation, stacked north- (N-) and south-IPWP (S-IPWP) thermal and hydrological records over the past 23–10.5 ka were built using planktonic foraminiferal geochemistry data from a new core, MD05-2925 (9.3° S, 151.5° E water depth 1661 m) in the Solomon Sea and eleven previous sites. Ice-volume-corrected seawater δ18O (δ18OSW-IVC) stacks show that S-IPWP δ18OSW-IVC values are indistinguishable from their northern counterparts through glacial time. The N-IPWP SST (sea surface temperature) stacked record features an increasing trend of 0.5 °C ka−1 since 18 ka. Its S-IPWP counterpart shows an earlier onset of temperature increase at 19 ka and a strong teleconnection to high-latitude climate in the Southern Hemisphere. Meridional SST gradients between the N- and S-IPWP were 1–1.5 °C during the Bolling/Allerod period and 1 °C during both Heinrich event 1 and the Younger Dryas, due to a warmer S-IPWP. A warm S-IPWP during the cold events could weaken the southern hemispheric branch of the Hadley cell and reduce precipitation in the Asian monsoon region.


Introduction
The Indo-Pacific Warm Pool (IPWP) is the largest warm water mass in the world, with an annual average sea surface temperature (SST) greater than 28 • C (Yan et al., 1992).Vigorous regional atmospheric circulation transports latent heat and water moisture from the IPWP to the middle and high latitudes (Yan et al., 1992).For the past 5 decades, the IPWP has experienced surface water freshening and a westward shift in precipitation, resulting in regional drought in eastern Africa and storm track changes in eastern Australia (Cravatte et al., 2009;Williams and Funk, 2011).Since the early 2000s, intensive paleoclimatological studies have been conducted in this region to understand long-term thermal and hydrological changes in the IPWP.These studies have shed light on the influence of glacial/interglacial (G/IG) cycles and have placed constraints on the relationships between (1) warm pool thermal and hydrological fluctuations, (2) high-latitude ice sheets, and (3) greenhouse gas concentrations during the late Pleistocene (e.g., Lea et al., 2000;Stott et al., 2002Stott et al., , 2004;;Visser et al., 2003;Rosenthal et al., 2003;de Garidel-Thoron et al., 2005;Steinke et al., 2006;Levi et al., 2007;Xu et al., 2008;Linsley et al., 2010;Bolliet et al., 2011;Mohtadi et al., 2014).
Stacked IPWP SST and seawater oxygen isotope (δ 18 O SW ) records from the last glacial to the Holocene show a close link among the IPWP SST, the Asian-Australian  1950-2004(Reynolds et al., 2002)).Upper panels are June-July-August (JJA), and lower panels are December-January-February (DJF) averages of (a, c) SSTs and (b, d) precipitation distribution maps.SST and precipitation are at 0.5 • C and 2 mm day −1 intervals.Our study site MD05-2925 is shown as the green star.Orange and green dots denote previous study sites in the IPWP region (Table 2) for reconstruction of meridional thermal and precipitation variations during the glacial/interglacial change.
In particular, little is known about the meridional thermalhydrological dynamics between the N (north)-IPWP and S (south)-IPWP during the last termination.
Here we present new oceanic proxy-inferred SST and ice-volume-corrected surface seawater oxygen isotope δ 18 O (δ 18 O SW−IVC ) records from the Solomon Sea, Papua New Guinea (PNG), for the past 23-10.5 ka (before AD 1950, hereafter).New SST and δ 18 O SW−IVC stacked records since the last termination are built for both the N-and S-IPWP to understand regional thermal-hydrological changes and interhemispheric teleconnections.
For oxygen stable isotope analysis, 7-10 individuals were immersed in methanol, ultrasonicated for 10 s, and then rinsed with deionized water 5 times.Samples were immersed afterward in sodium hyperchlorite (NaOCl) for 24 h, and then analyzed with an isotopic ratio mass spectrometer (IRMS), Micromass IsoPrime, at the National Taiwan Normal University.The long-term 1 σ precision of this instrument is better than ±0.05 ‰ (N = 701, Lo et al., 2013)   To extract seawater δ 18 O (δ 18 O SW ) values, we used a cultural-based equation, SST = 16.5 − 4.8 × (δ 18 O C − δ 18 O SW ) (Bemis et al., 1998) and a constant offset of 0.27 ‰ between carbonate VPDB and Vienna Standard Ocean Water (VSMOW) scales.Ice volume-corrected δ 18 O SW (δ 18 O SW−IVC ) was calculated using the method proposed by Waelbroeck et al. (2002).
The empirical orthogonal function (EOF) analysis of a modern SST data set (AD 1950(AD -2004;;Reynolds et al., 2002) for a sector from 20 • S to 20 • N, and 100 to 180 • E was conducted (Fig. 2) to determine the boundary between N-IPWP and S-IPWP.With an equatorial border, the EOF1 factor (83.4 %) effectively resolved SST variation groups.The EOF2 factor shows minor (9.7 %) but significant interannual zonal (ENSO) control on the SST patterns.EOF results show that the geographic equator is also the thermal equator between N-IPWP and S-IPWP (Fig. 2).
To build a stacked N-and S-IPWP record, we followed the suggestions of Leduc et al. (2010) and considered three criteria for this data set: (1) sites with locations from 12 • N to 15 • S, which is the main IPWP range (Yan et al., 1992;Gagan et al., 2004), and (2) usage of specific proxies, Mg / Ca-derived SST and δ 18 O C records of planktonic foraminifera, G. ruber (white, s.s.).Records from 12 sites were selected, including this study (Table 2).We adopted the published age model for sites ODP806, MD97-2140, MD97-2141, MD98-2162, MD98-2170, MD98-2176, and MD98-2181.For records with available original radiocarbon The sea level change effect on δ 18 O SW was also corrected.We divided the total data into 400-year windows and calculated the mean and standard error of the mean for each time window.

Solomon SST and δ 18 O SW−IVC records during the last termination
Mg / Ca SST records of the planktonic foraminifera G. ruber reveal stable glacial thermal conditions during the period 23.0-18.5 ka, with a variation < 1 • C and a glacialinterglacial difference of ∼ 3 • C between the last glacial maximum (LGM) and the end of the Younger Dryas (YD) in the Solomon Sea (Fig. 4a).This record is characterized by (i) the end of glacial conditions at 18.5 ka, and (ii) rapid SST increases of 1-2 • C at 18.5-18.0,17.0-16.0,15.0-14.5, and 13.0-12.5ka.The onset of deglacial SST increases in this region is consistent with the timing of thermal changes in the Southern Ocean as inferred from Antarctic ice core δD records (Stenni et al., 2003) (Fig. 4a).This agreement indicates a strong cli-  1).matic teleconnection between low-and high-latitude realms in the Southern Hemisphere (SH), as well as change of greenhouse gas concentrations (Mohtadi et al., 2014).There are significant SST increases of 1-2 • C during Heinrich event 1 (H1) and the YD.Previous studies from the eastern equatorial and South Pacific reveal a mechanism characterized by early warming of South Pacific subtropical mode water (Pahnke et al., 2003;Lamy et al., 2004;Pena et al., 2008).This warm signal is transported along a gyre to the east equatorial Pacific (EEP) and eventually to the west Pacific through ocean tunneling (Pena et al., 2008;Qu et al., 2013, Fig. 4a).Our new SST record is similar to those in the EEP (Pena et al., 2008) and eastern Indian Ocean records (Xu et al., 2008;Mohtadi et al., 2014) for both termination timing (within dating error) and significant warming during the H1 and YD events.There is a slight warming (< 1 • C) interval at 14.5-13.5 ka during the B/A (Bølling/Allerød) period (Fig. 4a).The warming could be attributed to possible mixing with warm surface waters of the N-IPWP.
The Solomon Sea δ 18 O SW−IVC record is given in Fig.   (Stenni et al., 2003), and the yellow line is the Greenland ice core NGRIP (Northern Greenland Ice Core Project Members, 2004) oxygen isotope record.The superimposed dark cyan and dark yellow lines are the 200 a smoothed records, respectively.Black triangles are AMS 14 C dates (Table 1).Vertical bars denote the H1 and YD periods.
The dramatic δ 18 O SW−IVC increases during H1 and the YD likely resulted from a weakening and/or southward shift of the ITCZ (Chiang and Bitz, 2005;Broccoli et al., 2006), and local evaporation may also play a role.Agreement between the δ 18 O sequences of the Greenland NGRIP (Northern Greenland Ice Core Project) ice core and the Solomon Sea δ 18 O SW−IVC indicates an imprint from the high-latitude Northern Hemisphere (NH) during the last termination period (Shakun and Carlson, 2010) (Fig. 4b).

Millennial timescale variations of N-and S-IPWP SST stacks
Both N-and S-IPWP stacked SSTs show the same difference of ∼ 3 • C between the last glacial and interglacial states (Fig. 5a).N-IPWP stacked SST values increased steadily since 18 ka through the termination at a rate of 0.5 • C ka −1 .Millennial timescale variability is absent in this record, which is similar to Linsley et al. (2010) and Stott et al. (2002).Although the resolutions of ODP806 and MD97-2140 are insufficient to resolve the millennial-timescale event, there is no significant difference with/without their records in our N-IPWP stacks (not shown).
The onset of the termination at ∼ 19 ka in the S-IPWP stack is consistent with temperature increases in Antarctica (Stenni et al., 2003), and occur about 1 kyr earlier than in the N-IPWP stack (Fig. 5a).This timing is synchronous with EEP (Pena et al., 2008) and SST records of the nonupwelling  2) were calculated with the nonoverlapping binned methods (Oppo et al., 2009;Linsley et al., 2010).All dashed lines represent 1 σ uncertainty ranges.Gray bars show the H1 and YD events.region of the eastern Indian Ocean (Xu et al., 2008;Mohtadi et al., 2014).Thus, our MD05-2925 and S-IPWP stacked SST does not appear to be severely affected by equatorial upwelling.Instead, the S-IPWP stacked SST record represents the broad SH equatorial region thermal conditions applicable to upwelling and nonupwelling E-W equatorial environments of both the Indian and Pacific oceans.Records from the tropical South China Sea show interproxy (U K 37 and Mg / Ca) differences during H1 and the YD (Zhao et al., 2006;Steinke et al., 2008), probably due to intrinsic limitations of the different proxies, such as seasonality and upwelling intensity.The S-IPWP stacked SST record is characterized by a warming trend during H1 and the YD, similar to Antarctic ice core temperature records (Stenni et al., 2003), and a steady thermal condition at ∼ 27 • C during the B/A, corresponding to the Antarctic cold reversal (ACR) (Fig. 5a).
The thermal gradient between the N-and S-IPWP is around 1 • C from 23 to 19 ka.Due to the earlier S-IPWP warming, the thermal gradient dropped from 1 to 0.5 • C around 19-18 ka, and persisted to the end of the H1 event.
The largest observed thermal gradient (1.5-2.0 • C) occurred during the B/A period, and was followed by a 1 • C drop during the YD.The meridional SST gradient between N-and S-IPWP over the last termination is attributed to the large thermal variability in the S-IPWP (Fig. 5a).Asynchroneity between persistent N-IPWP and fluctuating S-IPWP SST sequences (Fig. 5a) indicates a meridionally dynamic IPWP through the last termination period.This N-S SST gradient variability would also affect interhemispheric air flow and heat transport (Gibbons et al., 2014;McGee et al., 2014), providing a mechanism to explain heat transport variability between the hemispheres on a millennial timescale.

N-and S-IPWP δ 18 O SW−IVC records
Both the N-and S-IPWP δ 18 O SW−IVC records feature (1) low values of −0.3-0.0 ‰ during glacial times, and (2) increasing trends after 19 ka (Fig. 5c).The gradient between N-and S-IPWP gradually increased from 0 to 0.2 ‰ through the termination (Fig. 5d).A similar pattern of δ 18 O SW−IVC between the N-and S-IPWP suggests that hydrological conditions in the two regions were governed by the same factor(s), probably related to Northern Atlantic cold perturbations (Shakun and Carlson, 2010).It has also been suggested that a major δ 18 O SW−IVC increase during the H1 and YD periods in the IPWP region likely resulted from reduced precipitation and oceanic advection in both the N-IPWP and S-IPWP regions (Gibbons et al., 2014;McGee et al., 2014).

Meridional IPWP SST gradient and a southward-shift of the ITCZ precipitation boundary
A striking feature of the stacked SST records is the warming in the S-IPWP during the H1 and YD periods (Fig. 5a).
Observations over the past 6 decades (Fig. 12 in Feng et al., 2013) show that an equatorward shift of the NH convection branch of the Hadley cell (HC) could result from an oceanic warming at ∼ 10 • S.This equatorward shift could induce a southward ITCZ shift of about 10 • (Feng et al., 2013).Model simulations (Chiang and Bitz, 2005;Broccoli et al., 2006;Lee et al., 2011) suggest that this altered circulation provides    a powerful teleconnection between the NH and SH climate systems through a coupled tropical ocean-atmosphere pathway, and is supported by marine and terrestrial hydrological proxy data (Y.J. Wang et al., 2001;Lea et al., 2003;X. Wang et al., 2007;Griffiths et al., 2009;Shakun and Carlson, 2010;Mohtadi et al., 2011;Meckler et al., 2012;Ayliffe et al., 2013;Carolin et al., 2013;Gibbons et al., 2014;McGee et al., 2014, Fig. 6).Distinctly different precipitation conditions across 8-10 • S in the IPWP during the H1 and YD events are illustrated in Fig. 6.For example, enhanced terrestrial sediment flux into the Coral Sea is suggested by a marine sediment thorium isotopic proxy record at 11 • S (Shiau et al. 2011).Lynch's crater records from northeastern Australia at 17 • S (Muller et al., 2008) show strong Australian summer monsoonal conditions.Stalagmite δ 18 O records at Flores Island (8 • S) also feature intense precipitation during H1 and the YD (Griffiths et al., 2009;Ayliffe et al., 2013).However, marine and stalagmite δ 18 O evidence reveal conditions of reduced precipitation and increased salinity in the northern IPWP, north of 8-10 • S, including the South China Sea (12 • N; Steinke et al., 2006), Sulu Sea (8 • N; Rosenthal et al., 2003), Philippine Sea (6 • N; Stott et al., 2002;Boillet et al., 2011), Java Island (8 • S; Mohtadi et al., 2011), Solomon Sea (9 • S; this study), and Borneo island (4 • N; Meckler et al., 2012;Carolin et al., 2013) (Fig. 6).On the basis of previous terrestrial and marine hydrological records and our new data, as well as modern (Feng et al., 2013) and simulated (Chiang and Bitz, 2005;Broccoli et al., 2006) data, we speculate a sharp precipitation boundary between the maritime continents and Australia at about 8-10 • S, extending from the Solomon Sea, Arafura Sea and Timor Sea, to the eastern Indian Ocean during H1 and the YD (Fig. 6).We propose that the west and east boundaries are between the Java-Flores islands (Griffiths et al., 2009;Mohtadi et al., 2011), and Solomon-Coral seas, respectively (Shiau et al., 2011;this study).A geographical mismatch between N-and S-IPWP thermal and precipitation patterns could be associated with the complex island mountain range configurations and sea level changes (Linsley et al., 2010).
To sum up our geochemical and composite data set in the IPWP region during the last terminations, we propose that the enlarged IPWP meridional SST gradient could result in an altered HC and reduced (increased) precipitation for the east Asian (Australia) monsoon territories during the H1 and YD periods (McGee et al., 2014).We also propose that variations in the meridional IPWP SST gradient during the termination period were mainly caused by the S-IPWP, which is closely linked to high-latitude climate systems.

Conclusions
Our new MD05-2925 marine geochemical records and stacked SSTs suggest that the meridional IPWP thermal conditions are strongly linked to interhemispheric high-latitude climate during the last deglaciation.Ice-volume-corrected δ 18 O SW stacked records show an increasing salinity gradient between N-and S-IPWP over the last termination.However, the δ 18 O SW−IVC could be affected by complex mountain range configurations in the IPWP region, and sea-levelcontrolled openings/connections among semiclosed seas.
Here we propose a new process of the thermal evolution of the IPWP region, where meridional differences in the thermal gradient could amplify the signal from high-latitude Northern Hemisphere climate events (e.g., H1, B/A and the YD), and radiative forcing from greenhouse gases.A hypothetical precipitation boundary around 8-10 • S during H1 and the YD has also been proposed, which is most likely caused by the meridional IPWP SST gradient and HC anomalies.We suggest that more advanced high-resolution regional model simulations are required to clarify (1) local precipitation variations in response to the complicated sea level and convection changes, (2) the role of the IPWP meridional thermalhydrological gradient to an altered HC, and (3) its relationship with regional and global climate systems during global climate perturbation events.

Figure 1 .
Figure 1.Climatological map of the IPWP SST (left) and precipitation (right) during AD1950 -2004(Reynolds et al., 2002)).Upper panels are June-July-August (JJA), and lower panels are December-January-February (DJF) averages of (a, c) SSTs and (b, d) precipitation distribution maps.SST and precipitation are at 0.5 • C and 2 mm day −1 intervals.Our study site MD05-2925 is shown as the green star.Orange and green dots denote previous study sites in the IPWP region (Table2) for reconstruction of meridional thermal and precipitation variations during the glacial/interglacial change.
Figure2.EOF analysis on SST (data set fromReynolds et al., 2002) and selected sites (Table2) used for stacked N-and S-IPWP records.(a) EOF1 explains 83.4 % of the total variance, which mainly represents intra-annual seasonality.(b) EOF2 shows a clear zonal pattern.Orange circles represent selected sites for the N-IPWP group and green ones for the S-IPWP group.The green star denotes the MD05-2925 site used in this study.

Figure 4 .
Figure 4. Geochemical proxy records of MD05-2925.(a) SST (red circles and line) and (b) δ 18 O SW−IVC (blue line) were reconstructed with G. ruber Mg / Ca ratios and δ 18 O C .The cyan line denotes the Antarctica EPICA (European Project for Ice Coring in Antarctica) deuterium isotope record(Stenni et al., 2003), and the yellow line is the Greenland ice core NGRIP (Northern Greenland Ice Core Project Members, 2004) oxygen isotope record.The superimposed dark cyan and dark yellow lines are the 200 a smoothed records, respectively.Black triangles are AMS 14 C dates (Table1).Vertical bars denote the H1 and YD periods.

Figure 5 .
Figure 5. Four-hundred-year nonoverlapping binned (a) SST and (c) δ 18 O SW−IVC of N-(orange solid line) and S-IPWP (green solid line).Lower panels show the differences in (b) SST and (d) δ 18 O SW−IVC between N-and S-IPWP.The compilations of N-and S-IPWP surface water thermal and hydrological records (Table2) were calculated with the nonoverlapping binned methods(Oppo et al., 2009;Linsley et al., 2010).All dashed lines represent 1 σ uncertainty ranges.Gray bars show the H1 and YD events.

Figure 1 |
Figure 1 | Sea surface temperature and salinity in the WPWP and Makassar Strait.a, Annual average SST in the WPWP.The locations of the sediment cores discussed in this study are indicated.SST data from ref. 46.b,c, Sea surface salinity during the boreal winter (northwest monsoon) (January-March) (b) and boreal summer (southwest monsoon) (July-September) (c).Note the location of core sites in the southern Makassar Strait (13GGC, 70GGC and MD62).The grey arrows show generalized flow of the ITF.The dark green arrows show the flow of relatively fresh water from the South China and Java seas into the Makassar Strait during the boreal winter monsoon.Salinity data from ref. 47.

2Figure 6 .
Figure6.Hypothetical proxy-inferred precipitation boundary during the H1 and YD events (modified from theLinsley et al., 2010).Blue dots represent a relatively increasing precipitation / δ 18 O SW lighter condition, and brown dots a decreasing precipitation / δ 18 O SW heavier condition.The segment between the Java and Flores islands of this sharp boundary (red dashed line) was proposed byMohtadi et al. (2011), and the one between the Solomon and Coral seas by this study.Black contours represent SST.

Table 2 .
Selected sites for stacked N-and S-IPWP records.

Table 1 | Sediment cores used in this study. Core ID, location Core ID in text Lat./Long. Water depth (m) Sample interval (yr)
SSTa calculated relative to average of SST from 4,300-4,900 yr BP, as sediments younger than 4,300 yr BP are missing from this core.§ Not used in SST composite reconstruction owing to geographic location. ‡