SST phases in the open-ocean and margins of the tropical Pacific ; implication on tropical climate dynamics

Introduction Conclusions References


Introduction
The tropical Pacific plays an important role in the global climate system.Solar radiation incident on the large area of the equatorial Pacific triggers atmosphere and ocean circulation that results in the transport of heat and moisture from the low to high latitudes (e.g., Hadley Circulation, the Intertropical Convergence Zone (ITCZ) migration, Kuroshio and other currents on the tropical Pacific (Fig. 1a).The trade wind system in tropical Pacific results in stronger wind-driven upwelling and cooler sea surface temperature (SST) (< 26 • C) in the central and eastern equatorial Pacific (EEP) named the "cold tongue".In the western tropical Pacific, the Western Pacific warm pool (WPWP) around the New Guinea and Maritime Continent areas is characterized by warmer SST (> 28 • C).The eastern to western SST gradient in the tropical Pacific drives the longitudinal Walker Circulation associated with El Niño and La Niña events that affect the modern climate deeply.In addition, the western tropical Pacific is also close to the Asian and Australian continent, playing a role in the East Asian-Australian monsoon systems.Based on the large amount ocean research and satellites data, the tropical Pacific is thought to exert a major effect on the global climate system (e.g.Yan et al., 1992).Similarly, modeling investigations also suggest that long-term changes in tropical Pacific SST patterns might have driven large extra-tropical climate change (Pierrehumbert, 2000).
A number of tropical Pacific paleo-SST records have been published (e.g. de Garidel-Figures

Back Close
Full domain after the Middle Pleistocene.Recently, Tachikawa et al. (2009) found that the CO 2 rise in Termination I lagged the SST rise in the Southern and tropical Pacific suggesting that early tropical warming could provide heat and moisture to the northern high latitudes and influence the onset of the last deglaciation.Other works suggest that WPWP SST varies similarly to Antarctica temperature changes on the millennial time scale (Rosenthal et al., 2003;Stott et al., 2007) indicating a link to the southern high latitudes through atmosphere or deep water circulation.
In addition to proxy data, there are model studies supporting the idea of a tropical Pacific influence on global climate.Clement et al. (1999) found tropical Pacific SST responses to insolation variability with an obvious precession period as well as ENSOlike variability.Lee and Poulsen (2005) also used a coupled ocean-atmosphere model to reveal the influence of obliquity forcing on climate via changes in meridional insolation gradients.FOAM model results also indicate the geographic distributions of phases between surface air temperature and insolation in the tropical Pacific (Kutzbach et al., 2008;Chen et al., 2011).
In order to better understand tropical Pacific climate dynamics, we recovered a CA-LYPSO core from the southeastern offshore of PNG, Northern Coral Sea (Fig. 1a), MD052928 ( 11• 17.26 S, 148 • 51.60 E, 2250 m).The sediments are mainly composed of a mixture of carbonate and siliciclastic detritus (Beaufort et al., 2005).The site location is affected by two surface currents, the South Equatorial Current (SEC) and Coral Sea Coastal Current (CSCC) (Wolanski et al., 1995), as well as by Antarctic Intermediate Water (AAIW), Circumpolar Deep Water (CPDW) and Antarctic Bottom Water (AABW) (Bostock et al., 2004;Wyrtki, 1961) connecting the regional oceanography to the southern high latitudes (Fig. 1a).The local climate has two major seasons, the northwestern monsoon during the austral summer and southeastern trades during the austral winter.The PNG area is characterized high annual precipitation, however, higher rainfall in the lowland occurs during the austral summer monsoon period.During El Niño's PNG experiences drought (Ogston et al., 2008).Introduction

Conclusions References
Tables Figures

Back Close
Full We compare our 320 kyr long high resolution U K 37 -SST record from core MD052928, to other SST records from the tropical Pacific including the WPWP, the EEP upwelling area and the South China Sea (SCS) (Fig. 1a).We document the timing of SST from these locations relative to local insolation, Antarctica temperature, and greenhouse gases (GHGs) by using time series methods in order to reveal underlying mechanisms driving surface temperature change.

Age model
We used 10 AMS-14 C dates of planktonic foraminifera Globigerinoides ruber and Globigerinoides sacculifer and compared oxygen isotopes of benthic foraminifera Uvigerina spp. with the LR04 stack (Lisiecki and Raymo, 2005) to establish the age model of core MD052928.The detailed AMS-14 C and oxygen isotope measurements and the age model had been described in Liao (2008) and Shiau et al. (2012).

U K 37 paleothermometer
The SST biomarker we used is the alkenone-derived U K 37 index (Prahl and Wakeham, 1987).In this study, we analyzed the whole 24 m of core MD052928.The top 6 m was sampled at 5 cm intervals, below 6 m, we sample at 10 cm intervals except termination II (1170-1270 cm) where we sampled at 5 cm for a total of 300 samples.We calibrated the U K 37 -SST by using the nonlinear calibration of Conte et al. (2006) which is based on alkenones from surface water particulates collected from the world ocean, the estimated SST error is 1.2 • C (Conte et al., 2006).The modern sediment trap observation on the western equatorial Pacific indicates the C 37 alkenones flux is with three maxima at February, June, and November (Harada et al., 2001).The mean U K 37 -SST of core MD052928 (Fig. 2) during the Holocene is about 28.3 • C that is correspondent to the Introduction

Conclusions References
Tables Figures

Back Close
Full SST average of these high C 37 alkenones flux periods, and close to the modern annual mean SST (Fig. 1).For the analytical procedure, we extracted the lipids with an Accelerated Solvent Extractor (DIONEX ASE 200) using a mixture of dichloromethane: methanol (6 : 4 v/v) at a temperature of 100 • C and a pressure of 7.6 × 10 6 Pa from the freeze-dried sediment samples.After the extraction, we used silica gel column chromatography to separate alkenones from the extracts.We eluted the alkenones using toluene.Then we analyzed the alkenones by gas chromatography with flame ionization detector.The alkenone analysis procedures are described in Yamamoto et al. (2000).

The SST records of MD052928
The SST of MD052928 varies between 26.5-29 • C, indicating obvious glacialinterglacial variation (Fig. 2).The warmest periods are the Holocene and MIS 5.5.During these two intervals temperatures exceed the upper limit of the C 37 alkenone calibration and are truncated at 28.7 • C (Conte et al., 2006).Other interglacial periods, such as MIS 5.1, MIS 5.3, MIS 7.1, MIS 7.3, MIS 7.5 and MIS 9, SST are cooler than in the Holocene and MIS 5.5, ranging between 28 and 28.5 • C.
During the glacial periods, the coldest SST occurred at the LGM (19-23 kyr BP) and MIS 8.2 (∼ 260 kyr BP), the lowest SST is ∼ 26.5 • C. Within other glacial periods and some cold stages in interglacials (e.g.MIS 7.4) the SST is between 27 and 28 • C except for MIS 6.5 and MIS 8.5 that exceed 28 • C and are warmer than other glacials.
We also note that SST changes lead ice volume at the three terminations, especially at Termination II where the SST leads by ∼ 4 kyr.These SST findings are consistent with other tropical Pacific records (Fig. 3).In addition, the SST of core MD052928 shows an obvious 23-kyr period.Introduction

Conclusions References
Tables Figures

Back Close
Full

Comparison of SST records across the tropical Pacific
In this study, we compare the MD052928 SST record with other SST records from the tropical Pacific region (Fig. 1a).The modern annual SST cycles for these sites are shown in Fig. 1b.Seasonal amplitudes are small for sites in the core of the warm pool and larger for cold tongue and continental margin sites.Two sites stand out with regard to seasonal amplitude.ODP 1146 has a 6 the past 320 kyr (Fig. 3).Prior to intercomparisons of SST records, we have assessed the various age models in order to compare them with confidence.Where available, we prefer to use age models constructed on the basis of benthic foraminiferal oxygen isotopes because it is less affected by surface hydrologic conditions.The age models of ODP 846 (Liu and Herbert, 2004) and ODP 1146 (Clemens et al., 2008) were constructed using benthic oxygen isotopes.However, ODP 806 (Medina-Elizalde and Lea, 2005) used planktic oxygen isotopes.In this case, we used the age model of Bickert et al. (1993) which is based on the benthic oxygen isotopes although cross spectral comparison of the benthic and planktic-based age models in this case indicates negligible phase differences.MD972140 and MD972142 have no benthic oxygen isotope records; for these cores we use the planktic records.cold water from the Peru-Chile margin during the SE trade prevails at the boreal summer (Liu and Herbert, 2004).Comparing SSTs variations on the terminations, we find that the SCS SST rises concurrently with the ice volume decrease derived from the δ 18 O.At other sites, SSTs increase leads ice volume terminations, especially at termination II.In addition, we did cross-spectral analysis of SST relative to orbital eccentricity (100-kyr), obliquity (41-kyr), and precession (23-kyr) (ETP) (Fig. 4 and Fig. 5).All spectra indicate the strongest concentration of variance at the eccentricity band indicating a strong link between equatorial SST and high latitude ice volume.All records indicate statistically significant coherence (> 80 %) with all orbital parameters (ETP) with the exception of MD972140 at the obliquity band and ODP 846 at the precession band.
In order to discuss the phases of tropical Pacific SSTs we plot all results reactive to δ 18 O from the same core, assigning δ 18 O to the SPECMAP (Imbrie et al., 1984) phases reactive to ETP.Details are described in the Supplement.Briefly, we applied the cross correlation function to adjust the lags of δ 18 O records in order to get better correlation to SPECMAP.After this adjustment, these six δ 18 O records are in phase with SPECMAP on the three orbital bands suggesting we have a reference time frame to compare tropical Pacific SST phases on the orbital time scale.These phase results are summarized and shown in the Supplement, Table 1, and illustrated in Fig. 5.
The SST records can be separated into two groups when considering SST phase relative to changes in orbital forcing, ice volume, air temperature and GHGs recorded in the Antarctica ice core (Fig. 5a-c).SST maxima at open-ocean sites within primary equatorial current systems (MD972140, ODP 806 and ODP 846) slightly lag obliquity maxima, and leading CH 4 maxima and ice volume minima at the obliquity band.These sites lead ice volume minima and CH 4 maximum at the eccentricity band as well.In contrast, SST maxima at sites proximal to land masses (ODP 1146, MD972142, and MD052928) are in phase with ice minima at the eccentricity and obliquity bands, indirectly influenced by the slow response of continental ice sheets (Supplement Fig. S6a  and b).At the precession band ODP 806, and MD052928 SST maxima slightly lag precession minima and are in phase with CH 4 maxima.All others are in phase with Introduction

Conclusions References
Tables Figures

Back Close
Full or slightly lag ice volume minima (Supplement Fig. S6c).We interpret these phase relationships in Sect.4.2.

The comparison of tropical Pacific SSTs to GHGs and Antarctica temperature
GHGs are be distributed globally in a short time and can be use to correlate the age of ice cores between the Northern and Southern Hemisphere (EPICA community members, 2004) and may act as a trigger mechanism of tropical SST (Lea, 2004).In addition, previous studies indicate the Antarctica temperature variability affects the tropical Pacific SST variability by AAIW or sub-Antarctic Mode water indicating the connection between tropical Pacific and Antarctica climate changes (Toggweiler et al., 1991;Stott et al., 2007;Tachikawa et al., 2009).These indicate a clear relationship among GHGs, southern high latitude temperature and tropical Pacific SST.In this study, we also compare tropical Pacific SSTs to the CH 4 concentration (Loulergue et al., 2008) and deuterium record (δD, Antarctica temperature proxy) (Jouzel et al., 2007) from EPICA Dome C using the EDC3 age model (Parrenin et al., 2007) as well as a composite CO 2 record from Antarctica ice cores which is also based on the EDC3 age model (Lüthi et al., 2008).Figure 3 also illustrates the comparison among GHGs, δD and tropical Pacific SSTs.Both CO 2 and CH 4 show higher concentration in interglacials, but there are some differences between these two GHGs.For instances, higher CO 2 concentration extends from MIS 5.5 to MIS 5.4, but CH 4 decreased rapidly after MIS 5.5.CH 4 also indicates an abrupt change around the Younger Dryas whereas CO 2 does not.Open ocean cores ODP 846, ODP 806 and MD972140 as well as MD052928, indicate slight leads or in-phase relationships between SST and GHGs at terminations whereas those in the SCS clearly lag.
Cross-spectral results show that CH 4 is in phase with or slightly lags the earlyresponse SST records and clearly leads the late response records (Fig. 5a-c).Among the CH 4 , CO 2 and δD, CH 4 reaches the maxima prior to δD and CO 2 at obliquity and Introduction

Conclusions References
Tables Figures

Back Close
Full At the eccentricity band, GHGs maxima fall between SST maxima of the early and late response groups.At the precession band, CH 4 is in phase with the early group SSTs and leads δD, CO 2 and the late SST group (Fig. 5).The phase estimations of GHGs and δD are showed in Table 1.

SST variability of the tropical Pacific
The SST variability of the tropical Pacific on annual and orbital time scales is strongly linked to geographic location.MD052928 located off PNG, on the southern margin of the WPWP, has a seasonal SST variation of ∼ 3 • C which is associated with seasonal meridional solar insolation.During the austral summer, the local SST reaches 29 • C due to the WPWP southward expansion.In contrast, the SST decreases to 26 • C when the WPWP contracts northward during the austral winter (Fig. 1b).The SST range at MD052928 on glacial-interglacial time scales is ∼ 1.5-2.5 • C (Fig. 2).This range is larger than that estimated by U K 37 method for the central WPWP region which indicates ∼ 1 • C variation during glacial-interglacial cycles (McClymont and Rosell-Melé, 2005;Ohkouchi et al., 1994).Thus, MD052928 is ideally located to monitor changes in the meridional extent of the WPWP on orbital time scales.In addition, we find the U Furthermore, we found our downcore U K 37 -SST record shows an obvious signal at the 23-kyr period but weak at 41-kyr (Fig. 4f).We will discuss this further in Sect.4.2.
Comparing MD052928 SST to local insolation indicates a strong correlation to June insolation in the Southern Hemisphere (SH), especially at the 23-kyr period (Fig. 6).This result indicates MD052928 SST on the orbital time scale is associated with austral winter insolation when the boreal summer solstice is at the perihelion.These relationships are also illustrated on the precession phase wheel (Supplement Fig. S6c).The central WPWP (ODP 806 and MD972140) is characterized by low amplitude seasonal SST variability (Fig. 1b).However, SST at ODP 806 and MD972140 show a 3 • C SST decrease at the LGM (Fig. 3) indicating significant WPWP cooling in glacials.The central WPWP may be less affected by land-sea interaction in glacials and likely reflects ocean conditions more directly.The SST differences between glacials and interglacials are lower than in the SCS and eastern Pacific upwelling region.In former studies, using foraminiferal transform functions such as CLIMAP or MARGO, the WPWP area SST decreases ∼ 2 • C in LGM (CLIMAP, 1981;MARGO Project Members, 2009).Previous U K 37 -SSTs studies also indicate ∼ 1 • C variations during glacial-interglacial cycles in the central WPWP (McClymont and Rosell-Melé, 2005;Ohkouchi et al., 1994;de Garidel-Thoron et al., 2007).These SST differences may result from different proxy responses but not affect the result that SST decreased in glacials.In addition, we note the MD052928 U K 37 -SST amplitude is lower than ODP 806 and MD972140 located on the central WPWP (Fig. 3).MD052928 located on the southern Margin of WPWP where should be more sensible to SST variations.Although our observation may be attributed to different proxies; however, some former SST studies in the Coral Sea, such as ODP 820 (16 • 38 S, 146 • 18 E, U K 37 -SST) (Lawrence and Herbert, 2005), MD972125 (22 • 34 S, 161 • 44 E, U K 37 and Mg/Ca-SST) (Tachikawa et al., 2009), and a latitudinal transect along the Coral Sea (foraminiferal transform function) (Anderson et al., 1989) all indicating small glacial-interglacial SST amplitudes similar to MD052928 in Introduction

Conclusions References
Tables Figures

Back Close
Full the North of ∼ 20 • S.This indicates the small SST amplitude may be the local character of the Coral Sea.
The SCS is a marginal sea of the Pacific and its modern annual SST changes are under the influence of the East Asian Monsoon (Wang, 1999).The modern East Asian winter monsoon results in a large north-south SST gradient whereas in summer, East Asian summer monsoon yields a homogenous high SST throughout the whole SCS (Wang, 1999).The seasonal SST variations are ∼ 2 • C (MD972142, in the southeastern SCS) ∼ 5 • C (ODP 1146, in the northern SCS) (Fig. 1b) suggesting that paleo variations in SST might be driven by winter dynamics.The glacial-interglacial U K 37 -SST variations are ∼ 2-4.5 • C (Fig. 3).The exposed Sunda Shelf would result in a semiclosed basin that prevented the warm water from the Indo-Pacific warm pool flowing into the SCS, and combine an inflow of cold surface water via the Luzon Strait resulted from the stronger East Asian winter monsoon that results in larger SST decline in glacials (Wang, 1999;Steinke et al., 2008).In addition, we note the SSTs in the SCS are in phase with the oxygen isotope records (Fig. 3) that may indicate the SSTs in the SCS are associated with the ice volume changes, through the strengthened winter monsoon forcing and descending sea level.ODP 846 is located in the upwelling region of the EEP.The seasonal SST reaches a minimum in boreal summer due to the stronger SE trade resulting in the stronger SEC and EUC as well as increased strength of trade-induced upwelling (Wyrtki, 1981).
SST in the east is lower than in the western tropical Pacific and with a larger seasonal SST change (Fig. 1b).The annual SST variation of the EEP is ∼ 6 • C, higher in boreal spring and lower in boreal summer and fall.On the glacial-interglacial time scale, the U K Introduction

Conclusions References
Tables Figures

Back Close
Full high latitude feedbacks, either related to atmospheric and oceanic convection or ocean thermohaline circulation (Liu and Herbert, 2004;Martinez-Garcia et al., 2010).These SST records indicate that SSTs are not simple responses to local insolation but, rather, can be strongly impacted by local geography (such as continental proximity), upwelling, and ocean and atmospheric circulation.Factors leading to the SST phase grouping on the orbital timescale we will discuss in the next section.

Obliquity and eccentricity bands
At the obliquity and eccentricity bands, two clear groupings exist.Open ocean sites (MD972140, ODP 806, and ODP 846) proximal to the equator (Fig. 4a) comprise a tightly clustered early-response group that slightly lags maximum obliquity and significantly leads CH 4 and ice volume minima (Figs. 5a,S6a).Between the latitudes of ∼ 14 • N and • S, increased tilt leads to decreased insolation (Kutzbach et al., 2008).
Thus, warm SST at these sites cannot be related to local insolation forcing at the obliquity band.
The early group leads the ice volume change which has been discovered in other studies on the Equatorial pacific (Lea et al., 2000;Liu and Herbert, 2004;de Garidel-Thoron et al., 2005;Medina-Elizade and Lea, 2005) indicating that ice volume change is not the main factor driving tropical, open ocean SST.ODP 846, ODP 806 and MD972140 show clear power density at the obliquity band as discussed in previous studies (e.g., de Garidel-Thoron et al., 2005;Liu and Herbert, 2004;Medina-Elizade and Lea, 2005).In addition, these three does have an obvious signal and appears in phase relationship at the obliquity band (Fig. S6a, Table S1).Raymo and Nisancioglu (2003) suggested that the insolation gradient between high and low latitudes is controlled by obliquity forcing; the heat difference between high and low latitudes is driven by obliquity which controls the fluxes of meridional sensible and latent heat transport.We found similar results when comparing MD972140, ODP 806 Introduction

Conclusions References
Tables Figures

Back Close
Full and ODP 846 with the summer half year (21 March to 21 September) insolation gradient between the equator and 65 • N, revealing an in-phase relationship at the obliquity band (Fig. 7).This result suggests the entire tropical Pacific SST at the obliquity band is strongly affected by high-low latitudes heat differences through atmospheric heat transport.In addition, the increasing high-low latitudes insolation gradient could trigger stronger latitudinal atmospheric circulation (e.g.Hadley circulation).Based on Raymo and Nisancioglu's hypothesis, the highest insolation gradient happens when the obliquity reaches a minimum.Our results show the maximum SST of early group is slightly lags obliquity maximum, this may indicate that weaker latitudinal circulation leads to decreasing strength of equatorial upwelling system causing the whole equatorial Pacific SST to become warmer.The late group (MD972142, ODP 1146, and MD052928) comprise sites proximal to land masses and lags the insolation gradient minimum and is in phase with or slightly lags ice volume minimum at the obliquity band (Figs.5a and 7).The late group, consisting of sites proximal to land, appears to be affected by processes linked to high latitude ice volume.MD972142 and ODP 1146 located in the SCS where is surrounded by the Asian continent, Philippine archipelago, and exposed Sunda Shelf during the glacial that results in a lowstand condition.In glacial periods, the warm water from the Indo-Pacific warm pool could not flow through the exposed Sunda Shelf.In addition, the stronger winter monsoon also results in cooler SST in the SCS during glacial as we discussed in Sect.4.1.Therefore, SST phases in the SCS might be attributed to the warm water inhibited by the exposed landmass and monsoon system at the obliquity and eccentricity bands.MD052928, located on the northern Coral Sea in which is a relatively opener basin than the SCS.However, MD052928 was close to the exposed Sahul Shelf during the lowstand period.The exposed Sahul Shelf and the New Guinea island might be a barrier to reduce the warm water inflow from the warm pool to Coral Sea during lowstand periods.Thus, MD052928 SST record indicates the in-phase relationship to the ice volume minimum on the obliquity and eccentricity bands (Fig. S6a  and b) even through the site does not located on the semi-closed basin like the SCS.Introduction

Conclusions References
Tables Figures

Back Close
Full Furthermore, we also note the late group lags to the CH 4 , Antarctica δD record, and in phase with the CO 2 (Fig. 5a) which may indicate the maximum SSTs of the late group could be related to high latitude forcing.At the obliquity band, the phase relationship among the early and late groups, GHGs and Antarctica δD record may reveal the warmer tropical SST resulting in more CH 4 released and the warmer high latitude area, then resulting in the continental ice volume minimum and the rising SSTs proximal to land masses.
Early studies of insolation at the eccentricity band indicate high eccentricity at perihelion results in more heat flux in the tropical latitude due to the effects of the 2 times overhead passages of the sun during each year (Ashkenazy and Gildor, 2008;Berger, 2006;Short et al., 1991).Short et al. (1991) used an energy balance model (EBM) to stimulate the thermal responses of different regions finding the equatorial area has a stronger response at the eccentricity band, whereas the tropical land masses have enhanced thermal responses at all the orbital bands, especially on the eccentricity and precession bands.However, our early group SST phase leads the eccentricity maxima, the orbital configuration might not be the major control factor at this band.Shackleton (2000) studied the phase relationships of deep-Pacific temperature, Antarctica air temperature and CO 2 concentration from the Vostok ice core to the ice volume and found the former three proxies all lead the ice volume change and ETP at the eccentricity band.He suggested the eccentricity signal resulted from changes of global carbon cycle caused by the atmospheric CO 2 concentration.In recent, a model study also suggests different CO 2 concentrations play an important role on the amplification on the eccentricity cycle (Ganopolski and Calvo, 2011).SST phases of early group are in phase with the CO 2 concentration at the eccentricity band may support this hypothesis.Nevertheless, we still need longer records and more evidences to discuss the role of the global carbon cycle on the eccentricity band.
The obliquity-band responses of SSTs indicate their phases are associated with the high-low latitudes insolation gradient (Fig. 7).When the gradient reaches a minimum, the weakened equatorial upwelling system resulting in the open-ocean SSTs reach

Conclusions References
Tables Figures

Back Close
Full maxima.Warmer equatorial Pacific SST heads to increased atmospheric moisture and tropical rainfall leading to expand tropical wetlands and increase CH 4 production.Increased CH 4 promotes increased deglaciation and CO 2 leading to increase continental and continental margin surface temperatures.At the eccentricity band, although the SSTs, GHGs and Antarctica δD also indicate similar phase relationship as the obliquity band, however, the phase lead of the early group SSTs is still not well explained.

Precession band
At the precession band, all SST records have obvious 23-kyr power (Fig. 4), and the phase can also be separated into 2 groups (Fig. 5c).However, unlike the eccentricity and obliquity bands these groupings do not reflect open-ocean and continental proximity simply.The early group includes MD052928 and ODP 806 located in the WPWP.
As in the eccentricity and obliquity bands, the early group slightly leads CH 4 and significantly also leads the ice volume minimum.The late group includes MD972142 and ODP 1146 located in the SCS and MD972140 located on the western margin of WPWP.
The late group lags ice volume minimum to a slightly greater extent than at obliquity and eccentricity bands.ODP 846 in the EEP has no significant coherence precession band so we don't discuss its phase variation.
The early group phases are close to the boreal summer solstice at the perihelion, when the solar radiation reaches the maximum in the Northern Hemisphere (NH) at the boreal summer (Fig. 6).However, in this case, the phase of the SH site (MD052928) indicates a link to winter dynamics; SST maxima at the time of warmest SH winter.The modern observation indicates the WPWP moves northward and the SST decreases in the MD052928 site during the boreal summer (austral winter).However, our result reveals the SSTs of the early group are in phase with the local insolation during the boreal summer solstice at the perihelion at the precession band indicating the WPWP SSTs are closely linked to the local insolation during this period (Fig. 6).The SSTs of the late group all show very large lags relative to direct local insolation forcing (Fig. 6)

Conclusions References
Tables Figures

Back Close
Full indicating there are some internal mechanisms acting on the late group at the precession band.The late group includes three SST records, MD972142 and ODP 1146 from the SCS and MD972140 from the western part of WPWP.It shows very large lags relative to direct local insolation forcing during the boreal summer and the ice volume minima indicating there are some internal mechanisms acting on the late group at the precession band (Figs.5c and 6).In addition, the SCS is a relatively closed basin compared to the open ocean and surrounded by the Asian continent and the Philippine and Indonesian archipelagos, and the shallow Sunda Shelf located on the southern SCS.These two records reveal a phase similar to the previous SST records from Asian summer monsoon regions at the precession band (Chen et al., 2003;Clemens and Prell, 2003).We infer that the SCS records may be caused by the timing of increased EASM winds, driving clockwise circulation in the SCS, bringing warm waters to these sites (Wang, 1999).If so, the exposed or submerged Sunda Shelf controlled by the ice volume change might affect the warm surface water flow into the SCS from the tropical Eastern Indian Ocean and the WPWP during the summer.MD972140, however, its SST phase lags ODP 806 and MD052928 on the precession band whereas they located on the WPWP region (Supplement Fig. S6c).This might be associated with its location.The New Guinea and Indonesian archipelago are under the Australian monsoon system that is associated with the East Asian Monsoon in the modern world (Suppiah and Wu, 1998).Previous studies also suggested that East Asian-Australian monsoon system influences climate variations on this region at the precession band (e.g.Tachikawa et al., 2011;Shiau et al., 2012).MD972140 located on the western part of WPWP, between the central WPWP and the Asian continent that may be affected by the East Asian-Australian monsoon system.Its phase also indicates slightly lead than sites in the SCS on the precession band (Fig. S6c) that may suggest the influence from the monsoon system except for the local seasonal insolation.Recently, Qu et al. (2005)  works.This theory suggests the relatively cold, salty water from the Pacific enters the SCS through the Luzon Strait, through the monsoon and upwelling dynamics to form the warmer and fresher SCSTF in the SCS basin flowing southward then exits the SCS through the Karimata Strait with a shallow water depth (∼ 40 m) into the Indonesian maritime continent.The SCSTF is suggested as a conveyor belt transferring up to 0.2 PW of heat and 0.1 Sv of freshwater from the SCS into the Indonesian maritime continent (Qu et al., 2006).The relatively warmer and fresher SCSTF flows into the Makassar Strait and forms a northward surface current then returns into the tropical Pacific (Qu et al., 2005).Furthermore, the model study suggested that the SCSTF would promote the northward heat and water transport in the Makassar Strait during the boreal winter that could enhance the increased SST in the tropical Pacific (Tozuka et al., 2007).MD972140 located on the western WPWP, whereas its SST is in phase with the ice volume minimum and belongs to the late group on the precession band (Fig. S6c) suggesting not only the local seasonal insolation as the early group but also other mechanisms.MD972140 SST might be possibly affected by the SCSTF during the interglacial periods.The SCSTF could exit the SCS through the Karimata Strait and flow into the Indonesian maritime continent easily during the highstand condition in interglacials.MD972140 might be under complex interactions of East Asian-Australian monsoon system, local insolation and the SCSTF resulting in its SST phase lag to other WPWP sites on the precession bands.Although the SST records of the late group are located in the tropics, however, their SSTs are not controlled by the local insolation directly that may be also associated with other internal mechanisms, such as the monsoon, land-sea configuration, and local ocean and atmospheric circulations.
Previous studies indicate significant effects of precession on the tropical climate (Beaufort et al., 2001;Clement et al., 1999;Short and Mengel, 1986).In this study, the early SST group is consistent with this view.The late group, however, responses to the East Asian-Australian monsoon wind patterns.The phase result at the precession band also indicates the influence of the land-sea distribution and local ocean and atmospheric circulations.Introduction

Conclusions References
Tables Figures

Back Close
Full At the precession band, SSTs, GHGs and δD show the similar phase relationship as the obliquity band (Fig. 5a and c) and indicate the same scenario as the obliquity band.Higher local insolation leads the warmer WPWP resulting in higher moisture supply and stronger tropical rainfall and the expansion of tropical wetlands.Then increased CH 4 production promotes increased deglaciation and CO 2 leading to increased continental and continental margin surface temperatures.The difference is that the early group SSTs is associated with the local insolation during the boreal summer directly and distributed on the WPWP, which show the WPWP may play an important role on the global climate system at the precession band.

Antarctic temperature on the orbital time scale
CH 4 shows an earlier phase relative to CO 2 at the obliquity and precession bands, and a lag relative to the early group.CH 4 is a stronger GHGs with the second radiative forcing after CO 2 (Forster et al., 2007) and has a variety of high-and low-latitude wetlands and animals sources (Chappellaz et al., 1993a), marine clathrates (Kennett et al., 2000), and thermokarst lakes (Walter et al., 2007) in nature and other anthropogenic sources (e.g.enhanced biomass burning, and rice agriculture) (Forster et al., 2007).Chappellaz et al. (1993a) estimated the annual atmospheric CH 4 nature budgets in the LGM and pre-industrial Holocene are 120 and 180 Tg yr −1 , respectively, which indicates that the difference of atmospheric CH 4 between LGM and pre-industrial Holocene is mostly contributed by the expanded wetlands in tropical and NH high latitude.Evidence from measuring δD of CH 4 (Sower, 2006) and cosmogenic 14 C of CH 4 (Petrenko et al., 2009(Petrenko et al., , 2013) ) in ice cores suggest that the marine clathrate is not a major driver for the abrupt climate change such as Younger Dryas or Heinrich events.The rising sea level also results in some wetlands on the shelf in deglacials.The original vegetation on the shelf buried and decomposed in the wetland when the sea rise in deglacials that could also contributes a part of CH 4 into atmosphere (Ridgwell et al., 2012).These results suggest the expanded wetlands are the major source of CH 4 since the last glacial.In Introduction

Conclusions References
Tables Figures

Back Close
Full addition, one of the major atmospheric CH 4 sinks is through the oxidation in the troposphere by reaction with hydroxyl radicals (Forster et al., 2007), which appears to bring minimal effects on the variations of CH 4 concentration in the glacial-interglacial cycles (Chappellaz et al., 1993a;Thompson et al., 1993, andSingarayer et al., 2011).In this research, phases of tropical SSTs maxima all lead the CH 4 maxima at the three orbital bands (Fig. 5).Higher tropical SST could result in more evaporation over the tropical ocean that may induce more rainfall on the tropical land and expanded wetlands.Chappellaz et al. (1993b) suggested the low latitude moisture fluctuations may play an important role in changes of atmospheric CH 4 during the deglaciation.From the LGM to the pre-industrial Holocene, the variation of annual CH 4 emission flux change is 60 Tg yr −1 .Approximately 50 % of the CH 4 emission flux was contributed from the tropics and the remaining were mainly sourced from wetlands in boreal high latitudes (Chappellaz et al., 1993a).Loulergue et al. (2008) studied the CH 4 in the Dome C ice core and found the strength of tropical CH 4 budgets influenced by monsoon systems and with an additional source input during major terminations as the retreat of the northern ice sheet allowed higher CH 4 emissions from boreal periglacial wetlands.Guo et al. (2012) analyzed the Dome C CH 4 concentration and Chinese loess weathering data pointing out the global monsoon changes have been a constant driver of atmospheric CH 4 , and the tropical insolation play an important role, especially on the precession and semi-precession bands.Recently, a model study suggested the rising atmospheric CH 4 concentration during the Holocene and the end of the last interglacial period is associated with the enhanced precipitation and expanded wetlands in the tropics linked to the nature changes in the Earth's orbital configuration, especially on the precession band (Singarayer et al., 2011).In this scenario, the increasing tropical insolation and warmer tropical SST would lead to variations in the hydrological cycles at low latitudes enhancing CH 4 emissions.The phase relationship of the early group SSTs and CH 4 maxima at the precession band shown in this study (Fig. 5c) supports these model results.On the obliquity band, however, the CH 4 maximum slightly lags the early group SST by ∼ 3 kyr (Fig. 5a, Table 1).The warmest equatorial SSTs could Introduction

Conclusions References
Tables Figures

Back Close
Full supply more moisture on the tropical land that might enhance the precipitation over the tropical lands and tropical wetland expansion in this band.In addition, the obliquity maximum could result in warmer climate in the high latitudes, especially on the NH.The warmer climate on the northern high latitudes would result in ice melting and more wetland that also promotes CH 4 production.On the eccentricity band, the early group of SSTs is also in phase with CH 4 maxima (Fig. 5b), however, the SSTs lead the eccentricity maxima which indicate that other internal forcings, such as the global carbon cycle (Shackleton, 2000) may play important roles at this band.All these observations suggest the early warming of tropical SST resulted from tropical insolation changes is primarily responsible for the expansion of tropical wetlands and therefore the increases of CH 4 emission in glacial-interglacial cycles.A modern observation indicates the imhomogenous warming SST over the WPWP results in anomalous meridional circulation induced the rising motion and increasing precipitation trend on tropical SH (Feng et al., 2013).This anomalous ascending south of the equator is due to the larger warming trend of WPWP SST in the SH.In our SST records, MD052928 and ODP 806 located on the tropical SH and equator, respectively, slightly leading MD972140 located on the north of equator during the deglaciation (Fig. 3), which might result in difference warming trend over the WPWP on the orbital time scale and enhance higher precipitation and CH 4 emission in the tropical SH.Furthermore, the WPWP was also close to exposed Sunda and Sahul shelves located on the tropical SH in glacials that could result in the large expanded wetland during glacials to interglacials until the sea level rose to overlap the shelves.Wetlands resulted from gradually flooding on these shelves also enhances a part of atmospheric CH 4 besides the more precipitation (Rigdwell et al., 2012).In addition, orbital configurations of obliquity maxima and precession minima forcing would enhance the ice sheet melting on high latitudes that may result in more periglacial wetlands and more additional CH 4 emission, though the lifetime of CH 4 in atmosphere is very short (∼ 12 yr) (Forster et al., 2007).
The EPICA dome C δD variability was associated with the high latitude insolation resulted from obliquity and precession forcings at the orbital time scale (Jouzel et al., Introduction Conclusions References Tables Figures

Back Close
Full 2007).The δD variability lags relative to the early SST group at obliquity and precession bands, and is in phase at the eccentricity band (Fig. 5).The early group SSTs lead at obliquity and precession bands indicates that tropical Pacific SSTs of the open ocean and WPWP are not influenced by the southern high latitudes directly as previous studies (e.g.Stott et al., 2007) but are mainly affected by low latitude insolation forcing.With satellite observation indicating that heat transport in the Pacific is from the low to high latitudes (Hasterath, 1982), this result suggests that the early warming of the tropical Pacific is driven by local insolation changes that lead southern high latitude warming.
In addition, the variations of δD in ice cores are in-phase with CH 4 at all orbital bands (Fig. 5) that indicate the increases of atmospheric CH 4 concentrations provide a positive feedback that contributes the southern high latitudes warming.Thus, the latitudinal heat transport in the Pacific in conjunction with the warming effect from CH 4 during the increases of high latitude insolation would result in the warming of southern high latitudes, the retreat of sea ice, and the ice sheet melting on the Antarctica during the deglaciation.
The warming of southern high latitudes results in the melting of sea ice and Antarctica ice sheet that would change the ocean currents around the Antarctic and promote the CO 2 rising in atmosphere (Stott et al., 2007).CO 2 is a long-lived GHG with strong radiative forcing (Forster et al., 2007).The ocean processes are proposed as the cause of CO 2 changes during the glacial-interglacial cycle (Jansen et al., 2007).The ocean is the largest of the relatively fast-exchanging (< 1 kyr) carbon reservoir, and the terrestrial change only is not sufficient to explain the low CO 2 during glacial (Sigman and Boyle, 2000;Jansen et al., 2007).There are several mechanisms including ocean circulation, marine biological activity, ocean-sediment interactions, seawater carbonate chemistry and air-sea exchange have been proposed to explain the atmospheric pCO 2 changes in glacial-interglacial time scales (Jansen et al., 2007).Though all these mechanisms are complicated, the Southern Ocean is thought to be an important driver on glacialinterglacial atmospheric pCO 2 changes (Sigman et al., 2010).The change in Southern Ocean circulation results in reducing the water exchange between the Antarctic

Conclusions References
Tables Figures

Back Close
Full surface water and the underlying deep water that have closed the valve of CO 2 releasing to atmosphere during glacials (Sigman et al., 2010).Paleceanographic records (Stott et al., 2007) based on deep-sea temperature suggest that the increase of austral spring insolation would result in the retreat of sea ice and the decrease of stratification by enhanced Ekman transport that would have led to stronger air-sea exchange and rising CO 2 during the deglaciation.Opal flux reconstructions in the Southern Ocean (Anderson et al., 2009) also suggest that the strengthened upwelling would enhance the atmospheric pCO 2 .SST reconstructions of the tropical and southern Pacific show an early warming than the CO 2 rise during the Termination I (Tachikawa et al., 2009), providing another support to this mechanism.Other physical parameters may affect the CO 2 change during glacial-interglacial cycles.For example, the ocean SST and salinity changes may possibly affect the pCO 2 variability during the glacial-interglacial cycle.
In a calculation for the pCO 2 variability due to the tropical SST and salinity (Sigman and Boyle, 2000), 5 • C variability in tropical SST and 3 % salinity changes in the ocean during the glacial-interglacial cycle only account for pCO 2 variability of 23.5 ppmv. in the total glacial-interglacial variation of ∼ 80 ppmv.
Our results indicate that the CO 2 maxima are in-phase with the early group of SSTs at the eccentricity band, but lag to the early group at obliquity and precession bands (Fig. 5).The phase of CO 2 maxima at the eccentricity band might reflect the influence of global carbon cycle at eccentricity frequency (Shackleton, 2000) and the effects on tropical SSTs (e.g.Lea, 2004;Medina-Elizadle and Lea, 2005).The early group of tropical Pacific SSTs in phase with Antarctic CO 2 at the eccentricity band shown in this study supports the CO 2 trigger mechanism at this band (Fig. 5b).On the other hand, the early groups of SSTs lead Antarctic CO 2 record at obliquity and precession bands, and the late groups are in phase with CO 2 at the obliquity band and lag to CO 2 at the precession band (Fig. 5a, c).These indicate that CO 2 may not be the major controlling factor on the tropical SST on obliquity and precession cycles.A box core study (ERDC-92) of the western equatorial Pacific, which is one of the CO 2 source regions during the deglaciation and early Holocene, reveals that the increase of Mg/Ca SST leads Boron

Conclusions References
Tables Figures

Back Close
Full isotope estimated pCO 2 maxima (Palmer and Pearson, 2003).If the early warming on tropical ocean could act efficiently to supply more moisture to make an expansion of tropical wetlands and more CH 4 emissions, then the increases of CH 4 would accelerate the warming in southern high latitudes and enhance the retreat of sea ice and ice sheet melting in Antarctica.In summary, our phase results of tropical SSTs, CH 4 , δD and CO 2 reveals the following scenario: the insolation at low latitudes results in the early warming of the tropical open ocean, and then trigger the atmospheric CH 4 concentration which in turn forms a positive feedback that may accelerate the deglaciation, then changing the ocean circulation frame in the southern high latitudes and rising of pCO 2 , and the increase of tropical SST in the marginal seas.The GHGs and high latitude warming may not be the major factors controlling the tropical climate.In contrast, the early warming of tropical open ocean and WPWP resulted from the insolation that may play an important role on the deglaciation during the glacial-interglacial cycle.

Conclusions
Core MD052928 from the southern Papua New Guinea provides a high resolution U K 37 -SST record revealing the variability of the southern WPWP margin and is ideally located to monitor the meridional extent of the WPWP.The downcore U K 37 -SST of core MD052928 shows the ∼ 1.5-2.5 • C variations during the glacial-interglacial variations and an obvious precession period associated to the local austral winter dynamics.However, the lower SST variation amplitude than in the central WPWP that may be attributed to the local character of the Coral Sea.All tropical Pacific SST records from the three different areas: the WPWP, SCS, and cold tongue in the EEP.Comparisons of the SST records from these three areas indicate that SSTs are not simple responses to their local insolation but, rather can be strongly impacted by local geography, such as continental proximity, upwelling, and ocean and atmospheric circulation.

CPD Introduction
Full These results suggest the GHGs and early warming in the southern high latitude may be not the major factors controlling the tropical Pacific SSTs variability at the orbital time scale.The tropical climate dynamics in the open ocean is dominated by the low latitude insolation variation, and influences high latitudes through atmosphere and ocean circulations; however, the land-sea distribution, feedbacks of continental ice sheets and GHGs could result in the more complex interactions that may affect the heat transport from the low to high latitudes, and the global climate system.Ice Volume minimum phases (kyr) −3.3 ± 2.8 −8.9 ± 2.1 −5.5 ± 0.5 Mean Phases (kyr) Early Group 5.6 ± 3.1 −2.7 ± 1.4 −1.5 ± 0.8 Late Group −3.9 ± 1.7 −9.0 ± 1.1 −6.1 ± 0.5 Antarctic Ice Core Records (kyr) EPICA Dome C δD 6.9 ± 3.3 −5.5 ± 1.5 −3.5 ± 0.6 EPICA Dome C CH 4 0.3 ± 3.6 −5.5 ± 2.4 −1.9 ± 1.0 Antarctic Composite CO 2 3.6 ± 3.6 −9.3 ± 2.3 −4.3 ± 1.3 Introduction
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | SST time-series comparisons are illustrated in Fig. 3. Glacial-interglacial SSTs variations are 2-3 • C with the exception of sites ODP 1146, ODP 846, and MD972142 which have ranges of ∼ 4 • C. All these three records show larger modern seasonal temperature ranges as well (Fig. 1b).On this basis, one might interpret low temperatures at SCS sites ODP 1146 and MD972142 as reflecting colder winter monsoon seasons.Low temperatures at ODP 846 are interpreted as reflecting increased upwelling resulted from the shoaling of the Equatorial Undercurrent (EUC) and the advection of Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | precession bands.CH 4 and δD lead ice volume minima at all orbital bands.At the obliquity band, CH 4 and δD lag SSTs from the open ocean and lead CO 2 and SSTs in the SCS and Coral Sea.
of core MD052928 rose earlier than ice volume change during Terminations II and III, leading ice volume changes by ∼ 4 kyrs.The early warming of SST indicates that ice volume change may not the major factor controlling the SST of the southern margin of the WPWP.At termination I, SST and ice volume are closer to being in phase, as reflected in Fig. 5b where the eccentricity phase wheel represents the phase relationship between SST and ice volume (growth and decay) averaged over all cycles.Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | hypothesized a heat and freshwater conveyor: South China Sea throughflow (SCSTF) based on modern observations and modeling Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Cross-Spectral results of tropical Pacific SSTs indicate the SSTs could be separated into two groups at all orbital periods.At the obliquity and eccentricity bands, the early group includes ODP 846, ODP 806 and MD972140 which located on the open ocean and proximal to the equator.The late group consists of ODP 1146, MD972142 and MD052928 which are proximal to the Asian and Australian continents.At the precession band, the two groups are different from the obliquity and eccentricity bands.The early group consists of ODP 806 and MD052928 which located on the WPWP, and the late group includes ODP 1146 and MD972142 located on the SCS and MD972140 located on the western part of WPWP.These results indicate that the changes in highlow latitudes insolation gradient and tropical local insolation during boreal summer resulting in the warmer SST in the equatorial Pacific and WPWP.Warmer tropical SST leads to more rainfall and the expansion of wetlands in tropical regions resulting in increasing the atmospheric CH 4 concentration.Increased CH 4 promotes the increased deglaciation and CO 2 leading to increased continents and continental margin surface temperatures.
Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 5 .
Figure 5 • C seasonal amplitude driven by cold winter monsoon winds off the Asian continent.ODP 846 has a 6 • C amplitude driven by seasonally strengthened upwelling during the prevalent SE trade period at boreal summer in the EEP cold tongue.These modern seasonal amplitudes carry through the paleo record, with ODP 1146 and ODP 846 having the highest amplitude SST records over

Table 1 .
Tropical Pacific SSTs, and Antarctic δD, GHGs phases v.s.ETP in this study.The ice volume minimum phases are based on the SPECMAP to ETP during 4-318 ka.