Deep-Sea Research II 61-64 (2012) 114–126

Contents lists available at SciVerse ScienceDirect

Deep-Sea Research II

0967-06

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/dsr2

Variability in North Pacific intermediate and deep water ventilation duringHeinrich events in two coupled climate models

Megumi O. Chikamoto a,n, Laurie Menviel b, Ayako Abe-Ouchi a,c, Rumi Ohgaito a, Axel Timmermann d,Yusuke Okazaki a, Naomi Harada a, Akira Oka c, Anne Mouchet e

a Research Institute for Global Change, Japan Agency for Marine-Earth Science and Technology, Japanb Oeschger Centre for Climate Change Research, Climate and Environmental Physics, University of Bern, Switzerlandc Atmosphere and Ocean Research Institute, University of Tokyo, Japand International Pacific Research Center, University of Hawai’i, USAe Departement AGO, Univerisite de Liege, Belgium

a r t i c l e i n f o

Available online 30 December 2011

Keywords:

Meridional overturning circulation

Coupled climate model

North Pacific

Marine carbon cycle

45/$ - see front matter & 2012 Elsevier Ltd. A

016/j.dsr2.2011.12.002

esponding author.

ail address: megchika@jamstec.go.jp (M.O. Ch

a b s t r a c t

The responses of North Pacific intermediate and deep water ventilation and ocean biogeochemical

properties to northern North Atlantic glacial freshwater perturbations are evaluated with a coupled

atmosphere–ocean general circulation model MIROC and an earth system model of intermediate

complexity LOVECLIM. When the Atlantic meridional overturning circulation (AMOC) is weakened as a

result of the North Atlantic freshwater discharge, both models simulate subthermocline and inter-

mediate water warming in the Pacific Ocean. The sensitivities of the Pacific meridional overturning

circulation (PMOC) to AMOC weakening differ significantly between the two models. MIROC simulates

a small enhancement of the deep sinking branch of the PMOC in the North Pacific. On the contrary, the

LOVECLIM freshwater experiment exhibits intensified deep water formation in the North Pacific,

associated with a maximum transport change of 19 Sv. Despite the significant differences in ocean

circulation response, both models successfully reproduce high-oxygen and low-nutrient conditions of

intermediate and deep waters, in accordance with sediment core based paleoproxy reconstructions

from the North Pacific and Bering Sea during Heinrich event 1. Emergence of younger intermediate and

deep water in the North Pacific can be partly attributed to an overall enhanced mixing as well as

intensified overturning circulation of the subpolar North Pacific. Our models simulate broad features

observed in several paleoproxy data of the Pacific Ocean: biological production decrease in northern

Japan, cooling in the western North Pacific Ocean, and the southward shift of the Pacific intertropical

convergence zone.

& 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Ice core records from Greenland reveal a high degree of millen-nial-scale variability during the last glacial period (Dansgaard et al.,1993). Some of the associated warm–cold transitions that occurredduring this period were accompanied by huge iceberg dischargesinto the northern Atlantic region; these are known as Heinrichevents (Bond and Lotti, 1995; Heinrich, 1988). It has been arguedthat freshwater discharge caused by melting icebergs reduces thesurface water density in the North Atlantic Ocean, thereby weak-ening and shoaling the Atlantic meridional overturning circulation(AMOC) (Sarnthein et al., 1994). Evidence for an AMOC collapse thatoccurred at the beginning of the last glacial termination � 17;500

ll rights reserved.

ikamoto).

years ago and coincided with Heinrich event 1 (H1) is suggested byPa/Th data from the subtropical North Atlantic Ocean (McManuset al., 2004). Numerous paleoclimatic records from both hemi-spheres have shown that Heinrich events in conjunction with AMOCcessation exerted a powerful influence on the global climate duringglacial times.

The influence of Heinrich events on the Pacific climate is wellestablished. Spatio-temporal reconstructions of sea surface tem-perature (SST) in the Pacific Ocean during H1 (Kiefer and Kienast,2005) demonstrate a high degree of coherence between the Atlanticand Pacific oceans. Prominent decrease in SST during H1 is found inplanktonic foraminiferal Mg/Ca values in the western North Pacific(Sagawa and Ikehara, 2008) and alkenone temperature reconstruc-tions from the Okhotsk Sea (Harada et al., 2006, 2012). During thisperiod, benthic and planktonic radiocarbon age differences indicatethe existence of younger intermediate and deep waters in thewestern North Pacific (Ahagon et al., 2003; Okazaki et al., 2010;

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 115

Sagawa and Ikehara, 2008) and older waters in the eastern NorthPacific at intermediate depths (Marchitto et al., 2007; Mix et al.,1999; Stott et al., 2009). Decreasing ventilation ages at intermediatedepths in the western North Pacific Ocean suggest an increased levelof mixing and water mass formation. Greater oxygen concentrationsduring H1 in a Bering Sea sediment core from a depth of about2000 m further corroborate this conclusion (Okazaki et al., 2005). Inaddition, Kienast et al. (2006) showed that biological productionincreases in the eastern equatorial Pacific Ocean during Heinrichevents, which is attributable to the upwelling of abundant nutrient-rich deep water.

Using an idealized earth system model by forcing a NorthAtlantic freshwater perturbation, Saenko et al. (2004) showedthat a Pacific meridional overturning circulation (PMOC) can beestablished with AMOC weakening. They found AMOC weakeningby creating an artificial PMOC through freshwater extraction inthe North Pacific. The actual existence of such pan-oceanic inter-play is an interesting topic that requires further investigation. Astrong PMOC was also obtained in waterhosing experimentsdescribed in Timmermann et al. (2005). While these studiesdescribe idealized modeling solutions, we attempt to further linkthe existing paleo-proxy evidence for major reorganizations ofNorth Pacific flow with climate model sensitivity experimentsconducted using two climate–carbon cycle models.

In our modern continental configuration, freshwater is exportedfrom the Bering Sea into the Arctic Ocean. A major freshwaterperturbation in the North Atlantic, often used in idealized water-hosing experiments, raises the sea-level in the Arctic and leads to thereversal of Bering Strait throughflow. The freshwater perturbationfrom the North Atlantic/Arctic Ocean then spills into the NorthPacific, thereby preventing the formation of deep water. Underglacial conditions, however, the Bering Strait was closed (Dykeet al., 1996) and the dilution of North Pacific waters by additionalNorth Atlantic freshwater discharge was prevented (Hu et al., 2007;Okumura et al., 2009). Such conditions may have preconditioned theNorth Pacific for intensified intermediate and deep water formation.

In this study, we describe the results of a North Atlanticfreshwater perturbation experiment using two coupled climatemodels under glacial conditions (including a closed Bering Strait)to investigate the oceanic and biogeochemical response of theNorth Pacific to an AMOC shutdown. We describe the globalclimate response to an AMOC shutdown in both models, and thenfocus our research more specifically on climate and carbon cycleresponse in the North Pacific. Finally, we compare the results ofboth models with paleo proxies.

2. Models and experiments

2.1. Two coupled models

In this study, we use a coupled atmosphere–ocean generalcirculation model (CGCM) MIROC version 3.2 (K-1 ModelDevelopers, 2004), and an earth system model of intermediatecomplexity LOVECLIM (Goosse et al., 2010; Menviel et al., 2008).Because this CGCM predicts climate and ocean fields withoutmarine carbon cycle, we additionally employ an off-line oceanbiogeochemical model forced by the ocean circulation and diffu-sion fields obtained from MIROC. In LOVECLIM, the climate andcarbon systems are fully coupled.

MIROC includes an atmospheric state-of-the-art GCM with T42resolution (approximately 2.81) and 20 hybrid vertical levels inaddition to an ocean GCM with 1.41 longitude �0.91 latitude and44 vertical levels. The marine carbon cycle is simulated by amarine biogeochemical model with an offline tracer advectionscheme and a one-layer atmosphere (Oka et al., 2008).

In the tracer calculations, monthly ocean model fields areinterpolated to 2.81 latitude �2.81 longitude grids for the bio-geochemical model from the spatial resolution of the ocean GCMto reduce the computational cost (Chikamoto et al., 2011; Okaet al., 2011). By this method, the water-column structure derivedfrom MIROC is approximately maintained, although the linearinterpolation reduces the spatial resolution of the ocean biogeo-chemical model.

Tracer advection is prescribed using monthly data. Our modeltherefore allows access to the transient response of the marinecarbon cycle through climate oscillations. The biological model isbased on a simplified ecosystem model (Palmer and Totterdell,2001), and the biogeochemical component is composed of twoplankton function groups, phytoplankton and zooplankton; sus-pended and sinking particulate detritus; four dissolved inorganiccomponents of alkalinity, dissolved inorganic carbon (DIC),nitrate, and oxygen; and two carbon isotopes, 13C and 14C. Theatmospheric d13C and D14C are set to �6.5 and 0%, respectively.This model enables estimation of conventional radiocarbon age ofwater mass.

LOVECLIM consists of an atmospheric spectral T21 modelECBILT with three levels and a free-surface primitive equationocean model with 31�31 resolution coupled to a thermody-namic–dynamic sea ice model CLIO. ECBILT dynamic core is basedon the quasi-geostrophic vorticity equation, and is extended byestimates of ageostrophic terms. The atmosphere–ocean–sea icemodel is coupled to an ocean carbon cycle model (LOCH) and adynamic vegetation model (VECODE), and explicitly simulates thefluctuation of carbon inventory in the ocean, atmosphere, andland biosphere. The biogeochemical components in the oceancarbon cycle model are DIC, total alkalinity, phosphate, organicmatters, dissolved oxygen and dissolved silica (Menviel et al.,2008).

To compare the MIROC and LOVECLIM results for nutrients, weestimated the nitrate concentration from the phosphate simu-lated by LOVECLIM assuming a constant stoichiometric ratio ofN:P (¼16:1; Redfield et al., 1963). The nitrate distributions thatestimated in LOVECLIM and that simulated by MIROC can bedirectly compared because MIROC neglects nitrate fixationand denitrification processes that in actuality affect the nitrateinventory.

2.2. Model setup and experimental design

We conduct a last glacial maximum (LGM) control run and aNorth Atlantic waterhosing experiment under LGM backgroundconditions using MIROC and LOVECLIM. The LGM control runuses the experimental protocol of the Paleoclimate ModellingIntercomparison Project (Braconnot et al., 2007), with respect tocontinental ice sheet conditions (Peltier, 2004), greenhouse gases,and Earth’s orbital parameters (Berger, 1978). The topographyis similar to previous LGM experiments (Chikamoto et al., 2011;Menviel et al., 2008), with the exception of the bathymetricconfiguration that accounts for a closed Bering Strait. In theLGM hosing experiment, an additional freshwater input of3.15�106 km3 is added into the northern North Atlantic (501N–701N) (Fig. 1A). The anomalous forcing is applied at a rate of 1 Sv(1 Sv¼106 m3 s�1) for 100 years (Fig. 1B). This forcing mimicsa strong Heinrich event such as H1 although the durationof the applied freshwater pulse may be shorter than that ofactual conditions. The integration is continued without the fresh-water hosing (Fig. 1B). The biogeochemical tracers of the LGMcontrol experiment were used as a reference for each model.Model results averaged over 100–300 years are analyzed in thisstudy.

Fig. 2. SST in the LGM control experiments in (A) MIROC and (B) LOVECLIM and SST reconstructions (circles) by paleoproxy (MARGO Project Members, 2009).

Fig. 1. (A) Area of freshwater addition (gray shaded area), (B) periods of freshwater addition (contour, Sv) and analysis in this study (gray shaded area), and (C) time series

of maximum meridional overturning circulation (Sv). Black and gray thick lines are the results of MIROC and LOVECLIM, respectively, in the Atlantic Ocean, and black and

gray thin lines are those of MIROC and LOVECLIM, respectively, in the Pacific Ocean.

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126116

3. Results

3.1. Glacial climate and carbon cycle simulations

In the LGM control runs, the global mean surface air tempera-tures are 5.0 and 4.5 1C lower in MIROC and LOVECLIM, respectively.The maximum North Atlantic Deep Water (NADW) flow is 26.3 and27.5 Sv, respectively. The NADW flow may have been partly affectedby the Bering Strait closure, which prevents freshwater transportfrom the North Pacific to the North Atlantic (Hu et al., 2010). Theresulting changes in the poleward heat transport affect the SST(Fig. 2) and sea ice concentration in the polar regions. In thenorthern North Atlantic, the surface air temperature decreasessignificantly. In addition, the Greenland, Weddell, and Ross seasare characterized by near-freezing water temperatures. The mod-eled temperatures in the northern North Atlantic are lower than theannual-mean SST reconstructions (Fig. 2). In the North Pacific Ocean,the equator–subpolar and west–east SST gradients are in goodagreement with the proxy records in both models. On the contrary,

the equatorial Pacific temperatures simulated by the two modelsdiffer slightly from the SST-reconstructions; MIROC underestimatesSST, while LOVECLIM overestimates.

Precipitation as a result of proximity to the IntertropicalConvergence Zone (ITCZ) is significant in the equatorial PacificOcean in MIROC, which is along a temperature contrast betweenequator and subtropics. LOVECLIM simulates lower precipitationover the equatorial Pacific and western North Pacific Ocean thanthat simulated by MIROC under LGM conditions. This discrepancyin the precipitation field leads to greater salinity in these areas byLOVECLIM compared to that by MIROC.

Export flux of particulate organic carbon (POC) decreases by1.1 GtC year�1 in MIROC. In LOVECLIM, export production remainsconstant in the modern and LGM simulations. In both models,biological production in the polar region is reduced in the LGMsimulation because the sea ice coverage limits light in the ocean,thus preventing biological productivity. This response is consistentwith proxy records of export production (Kohfeld et al., 2005).Biological production in the North Pacific also agrees well with this

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 117

data (Kohfeld et al., 2005), although the data from the Pacific Oceanare sparse. Chikamoto et al. (2011) and Menviel (2008) describe thedetails and reproducibility of the climatological and biogeochemicalfields under LGM simulations but with an opened Bering Strait.

3.2. Climate response to North Atlantic freshwater discharge

The freshwater pulse in the northern North Atlantic causes asignificant AMOC weakening in both models (Fig. 1C). The max-imum AMOC rapidly weakens during the first 10 years after thehosing, and then recovers 670 and 500 years later in MIROC andLOVECLIM, respectively.

The sensitivity of the PMOC to the freshwater discharge differssignificantly between the two models (Fig. 3). In MIROC, the PMOCchanges marginally from the LGM control run. The cyclonic circula-tion cell in the North Pacific Ocean extends further south. Conversely,LOVECLIM simulates a substantial intensification of the PMOC. Thestrength of the meridional streamfunction in the North Pacific Oceanincreases to 5 Sv in MIROC, whereas it increases to 19 Sv in LOVE-CLIM (Fig. 1C). Antarctic Bottom Water (AABW) formation in theNorth Pacific decreases from 12 to 8 Sv in MIROC and from 10 to 8 Svin LOVECLIM. The difference in PMOC sensitivity between the twomodels relative to the AMOC shutdown is discussed in Section 4.2.

Fig. 3. Pacific meridional overturning circulation (Sv). Results of LGM control in (A) MI

The freshwater inflow and associated AMOC shutdown affect theglobal ocean dynamics, as summarized in Table 1. In both models, theAMOC weakening causes the characteristic meridional dipole patternof SST in the Atlantic Ocean during the first 300 years (Fig. 4A and B),as simulated by other CGCMs (Stouffer et al., 2006). A large coolingtrend appears in the western North Pacific in association with severecooling of the overlying atmosphere in the Northern Hemisphere.This response intensifies the Aleutian low, as shown by Okumuraet al. (2009).

In the Atlantic Ocean, we also observe a strong north–southcontrast in sea surface salinity (SSS) (Fig. 4C and D) with ananomalously low SSS spread south of the equatorial Atlantic Ocean.In the South Atlantic, the SSS increase is caused by redistribution ofsurface water between low and high latitudes. In MIROC, precipita-tion increases in the South Atlantic and South Pacific oceans(Fig. 4C). This implies a southward shift of Atlantic and Pacific ITCZ.In LOVECLIM, the ITCZ initially shifts southwards in both theAtlantic and Pacific oceans. However, once the PMOC is established,the Pacific ITCZ reverts to its initial position.

Significant difference exist in the North Pacific climate responsebetween MIROC and LOVECLIM. Salinity in the Atlantic equatorialregion is lower in MIROC than in LOVECLIM. This variation indicatesthat fresh water remains in the equatorial Atlantic region in MIROC,

ROC and (B) LOVECLIM, and those of LGM hosing in (C) MIROC and (D) LOVECLIM.

Table 1Summarized results in MIROC and LOVECLIM and reconstruction of proxy data.

Factors Area (ocean dep.) LOVECLIM L Proxy

Climatology

Commonalities NADW NA � � �(1)

AABW SO þ þ þ(2)

PMOC NP (þ) þ þ(3)

SST NA � � �(4)

SST SA þ þ þ(5)

SST Western NP � � �(6,7,8)

SST SP þ þ þ(9)

SSS NA � � �(10,11)

Precip. SA þ þ þ(12,13)

Water temp. NP (500–1500 m) þ þ

Water density NP (500–1500 m) � �

Differences SST Eastern NP � þ �(14), þ(15)

SST Eastern tropical Pac. þ (�) �(16)

SST Western tropical Pac. þ � þ(9)

SST Ind. þ � �(17)

SST SO � þ

Precip. Eastern tropical Pac. � 0 �(18)

Precip. Western tropical Pac. � þ �(19)

SSS Eastern NP � þ

Salinity NP (500–1500 m) � þ

Density NP (500–1500 m) � þ

Biogeochemistry

Commonalities O2 NP (500–2000 m) þ þ þ(20,21)

O2 SP (42000 m) � �

Nutrient NP (500–2000 m) � � �(22,23)

Nutrient NP (4m) þ þ

Export POC Western NP � � �(24)

Differences O2 NP (43000 m) � þ

Nutrient Western NP (0 m) � þ þ(25)

Export POC Eastern tropical Pacific � þ þ(15)

þ , positive; � , negative; 0, not change. References are numbered as follows: 1. McManus et al. (2004); 2. Rickaby and Elderfield (2005); 3. Okazaki

et al. (2010); 4. Vidal et al. (1997); 5. Vidal et al. (1999); 6. Sagawa and Ikehara (2008); 7. Harada et al. (2012); 8. Ahagon et al. (2003);

9. Tachikawa et al. (2009); 10. Rosell-Mele et al. (1997); 11. Bard et al. (2000); 12. Peterson et al. (2000); 13. Wang et al. (2004); 14. Kennett et al.

(2000); 15. Seki et al. (2002); 16. Kienast et al. (2006); 17. Bard et al. (1997); 18. Leduc et al. (2007); 19. Partin et al. (2007); 20. Okazaki et al.

(2005); 21. Behl and Kennett (1996); 22. Hoshiba et al. (2006); 23. Robinson et al. (2005); 24. Kienast et al. (2004); and 25. Brunelle et al. (2010).

Fig. 4. Anomalies of annual (A and B) SST (shaded) and surface wind stress (vectors, N m�2) and (C and D) SSS (shaded) in the LGM hosing experiment from the LGM

control experiment. Contours in Fig. 4C and D are the annual precipitation change. Contour levels are �0.3 and �1.0 mm day�1 for red and 0.3 and 1.0 mm day�1 for blue.

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126118

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 119

which is compatible with the longer timescale of the AMOC recovery(Fig. 1C).

In the Pacific Ocean, except for the Okhotsk Sea, SSS is slightlyreduced north of 401N in MIROC. In contrast, LOVECLIM simulates alarger SSS increase in the entire North Pacific Ocean. This responsecan be explained by the positive Stommel feedback (Saenko et al.,2003; Stommel, 1961), whereby an intensified PMOC leads to anenhanced poleward salinity transport from the subtropics to theextratropics. With a closed Bering Strait, the additional fresh waterfrom the North Atlantic/Arctic Ocean cannot escape in the NorthPacific. Therefore, salinity change in the North Pacific Oceanrepresents the combined effects of anomalous precipitation minusevaporation, meridional salinity gradient in the Pacific Ocean, andsalinity transport from the equatorial regions. In LOVECLIM, thetransport through the Indonesian Throughflow is strongly reduced,which leads to a greater transport of warm and saline equatorialwaters to the North Pacific.

The responses of North Pacific temperature also differ betweenthe two models. In MIROC, the negative SST anomaly extendseastward. In the tropical Pacific, the surface water warms, and thenortheasterly trade winds are slightly enhanced. In LOVECLIM, thewarming begins in the eastern North Pacific because poleward heattransport is intensified by the establishment of a strong PMOC.

In both models, the winter mixed layer depth deepens in theNorth Pacific Ocean (Fig. 5C and D). In MIROC, the mixed layerdeepens near the Kamchatka region and in the western boundarycurrents (Fig. 5C), corresponding to the strong mixing of surfaceand subsurface waters. In LOVECLIM, the mixed layer depthdeepens by 67 m, and the mixing area extends over the subpolarregion (Fig. 5D). In addition, deep convection increases by 600 min the western North Pacific (not shown). In both hosing experi-ments, intense mixing occurs in areas with weak stratificationunder LGM conditions (Fig. 5A and B).

3.3. Response of North Pacific intermediate and deep water to the

freshwater perturbation

To compare the model representation of Heinrich events withmultiple paleoproxy data in the North Pacific Ocean, we analyzethe water properties in the ocean interior and the biogeochemical

Fig. 5. (A and B) Winter mixed layer depth (m) and (C and D) anomalies of the winter m

response to the AMOC shutdown. The AMOC weakening leads tosignificant changes in North Pacific intermediate and deep watertemperatures in both models (Fig. 6A and B). In MIROC, the NorthPacific subsurface water becomes colder and fresher (Fig. 6A andC), which indicates that the surface water sinks to deeper levels athigh latitudes in the North Pacific Ocean. The overall intermediatewater warms, which reflects reduced return flow of Pacific deep-water at low latitudes that are probably linked to the PMOC fieldsand the enhanced mixing and ventilation at mid latitudes, asshown by a slightly smaller vertical density gradient (Fig. 6E). InLOVECLIM, positive temperature and salinity anomalies at depthsgreater than 1000 m in the North Pacific Ocean exist (Fig. 6B andD), thus suggesting dense water penetration to intermediatedepths in the North Pacific Ocean in association with the strongPMOC. Both models simulate a warming effect to a depth of1200 m in the Pacific basin between 301N and 301S, which couldlead to changes in ocean stratification and ventilation.

The prominent changes in the subthermocline and intermediatewaters contribute to greater oxygen concentration in the watercolumn because of such reduced stratification and enhanced venti-lation. In MIROC, the oxygenated layer extends to 2000 m depth(Fig. 7A). The oxygen concentration increases at intermediate depthsnear the oxygen minimum zone, which is controlled by the massbalance of oxygen-deficient deep water and oxygen-rich surfacewater. The model also represents highly oxygenated water near3000 m along the northern continental margin, which would be aresult of stronger PMOC and convection. The oxygen concentrationdecreases in the abyssal ocean, probably because of the weakenedoverturning cells of AABW (Fig. 3C). In LOVECLIM, the strong PMOCleads to a greater dissolved oxygenation content from the surface todepths greater than 3000 m (Fig. 7B). Higher oxygen concentrationat intermediate depths and lower concentration in the deep oceanare consistent with previous freshwater experiments (Schmittneret al., 2007).

The existence of younger water in the North Pacific Ocean isalso verified by radiocarbon age simulated by MIROC. Conventionalradiocarbon ages of isopycnal surfaces (potential density ¼ 27.5)in the LGM control experiment are shown in Fig. 8A. The waterdecreases in age near the coast of Kamchatka, where the surfacewater is subducted in this model, as evidenced by isopycnaldoming. The young water is distributed advectively along the

ixed layer depth in the LGM hosing experiment from the LGM control experiment.

Fig. 6. Anomalies of annual (A and B) temperature (1C), (C and D) salinity (psu), and (E and F) potential density (kg m�3) in the Pacific Ocean in the LGM hosing experiment

from the LGM control experiment (shaded). Contours are (A and B) temperature, (C and D) salinity, and (E and F) potential density (sy426:5) in the LGM control

experiment.

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126120

Fig. 7. Zonal mean oxygen concentration change in the LGM hosing experiment from the LGM control experiment (shaded) and the oxygen concentration of the LGM

control experiment (contours) in the Pacific Ocean (mmol m�3) in MIROC and LOVECLIM.

Fig. 8. (A) Conventional radiocarbon age (shaded, years) as simulated by MIROC in the LGM control experiment along the 27.5-isopycnal surface and isopycnal depth

(contours, m) of the LGM control run. (B) Radiocarbon age difference (shaded, years) as simulated by MIROC between the LGM hosing and LGM control experiments and

the isopycnal depth of the LGM hosing experiment (contours, m).

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 121

isopycnal surface. Changes in radiocarbon age and isopycnal depthare shown in Fig. 8B. In the water hosing experiment, the depth of27.5-isopycnal surface increases in large parts of the North Pacific.On z-levels, this change translates into a decrease in the watermass age in the western North Pacific, indicating ventilation insurface waters deeper than 900–1200 m. The intermediate waterextends to 151N through advective processes and high diapycnalmixing within the subarctic gyre. In our simulation, ventilationincreases up to a depth of 1100 m in the eastern North Pacific.Older waters appear in the tropical Pacific, which may be explained

by simulated changes in deepwater upwelling or interbasin watertransport via the Indonesian Throughflow.

To determine the effects of altered ocean circulation on nutrientdistribution and biological productivity in the North Pacific Ocean,we focus on a biogeochemical response to AMOC weakening. Fig. 9shows nutrient concentration averaged over the euphotic zone(0–100 m depth) and zonal mean nitrate concentrations over thePacific Ocean. In MIROC, the nitrate concentration decreases in theNorth Pacific subarctic gyre and eastern equatorial Pacific, andslightly increases in the North Pacific subtropical gyre (Fig. 9A).

Fig. 9. Nitrate concentration changes in the euphotic zone of (A) MIROC and (B) LOVECLIM (shaded, mmol m�3) and the sea ice thickness in February (black contours) in

the LGM hosing experiments (0.15 and 0.3 m). Zonal mean nitrate concentration changes in the Pacific Ocean of (C) MIROC and (D) LOVECLIM (shaded, mmol m�3). Grey

contours in Fig. 9C and D are the nitrate concentration changes at 2.0 mmol m�3 intervals.

Fig. 10. Change in export flux of POC (shaded, molC m�2 year�1) in the North Pacific region and the sea ice thickness in February (black contours) in the LGM hosing

experiment (0.15 and 0.3 m) of (A) MIROC and (B) LOVECLIM.

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126122

The zonal mean nitrate concentration is significantly decreased atintermediate depths in the North Pacific (Fig. 9C). Nitrate depletion iscaused by increased ventilation at deeper levels, which is consistentwith the simulated oxygen and radiocarbon age (Figs. 7A and 8B).Nutrient-deficient waters also appear at the subsurface in the SouthPacific Ocean. This is possibly due to changes in the characteristics ofwater originating from the Southern Ocean. Lowered surface nutrientconcentration in the Pacific Ocean could have been caused by changein basin-scale nutrient supply from the Antarctic circumpolar cur-rent. In LOVECLIM, the nutrient depletion at intermediate and deepdepths of the North Pacific is also apparent (Fig. 9D), in agreementwith MIROC results and the existence of low-nitrate and oxygen-richwaters at intermediate depths.

A significant difference is evident in the surface nutrientresponse between MIROC and LOVECLIM (Fig. 9A and B). In MIROC,an increase in the North Pacific subpolar gyre tends to removenitrate from subpolar regions and transport it to subtropical regions(not shown), thus resulting in surface nitrate concentration decreasein subpolar regions and slightly increase in subtropical regions. InLOVECLIM, large increase in surface nitrate is related to the stronger

PMOC state. The PMOC intensifies mixing of nutrient-rich waters tothe surface layers in subpolar regions.

The redistribution of nitrate in the euphotic zone contributesto biological production (Fig. 10). In MIROC, the change in thesupply of nutrient-rich deep water leads to a significant reductionin export flux of POC in the entire North Pacific Ocean and to aminor increase in the eastern equatorial Pacific (Fig. 10A). Incontrast, LOVECLIM shows a significant increase in export flux ofPOC in high-nutrient waters of the North Pacific surface (Fig. 10B).Both models show a decrease in POC export over the westernNorth Pacific sea ice area. In this study, limitation of light becauseof expanded sea ice reduces biological productivity.

4. Discussion

4.1. Paleoproxy reconstruction of Heinrich 1 meltwater event

We compare the model results with multiple proxy dataobtained for H1 (Table 1). In the hosing experiment, both models

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 123

simulate a cooling effect in the North Atlantic and western NorthPacific, which agrees well with proxy data based on foraminiferald18 O in the North Atlantic (Vidal et al., 1997) and Mg/Ca andalkenone temperatures in the western North Pacific and OkhotskSea (Harada et al., 2008, 2012; Sagawa and Ikehara, 2008). TheSouth Atlantic and South Pacific warming in the two models isalso inferred from warm water species and foraminiferal d18 Orecords (Tachikawa et al., 2009; Vidal et al., 1999). A southwardshift of the ITCZ in the Atlantic (Fig. 4C and D) is supported bysediment fluctuation in terrigenous iron deposition (Petersonet al., 2000) and color reflectance (Wang et al., 2004) in thetropical Atlantic.

In the eastern North Pacific, however, MIROC simulates coolingunlike LOVECLIM, which simulates warming. The eastern NorthPacific paleo-reconstructions do not offer clear evidence for SSTchanges in the region. Kennett et al. (2000) indicate a cooling in theeastern North Pacific during H1 from the planktonic oxygen isotoperecord, whereas Seki et al. (2002) suggest a warming from thealkenone-derived SSTs. MIROC simulates precipitation decreases inthe western tropical Pacific (Fig. 4C), supporting the evidence of aweakened east Asian monsoon (Partin et al., 2007). Several monsoonrecords suggest that the precipitation decreased during H1 ineastern China (Wang et al., 2001), the western tropical Pacific(Partin et al., 2007), and the Arabian Sea (Altabet et al., 2002).

Both results of MIROC and LOVECLIM are consistent with thehigh oxygen conditions at intermediate depths in the Bering Seaduring H1 (Okazaki et al., 2005). Benthic–planktonic 14C agereconstructions from the North Pacific (Okazaki et al., 2010)indicate higher ventilation during H1.

Uncertainty remains the amplitude of water age fluctuation inthe western North Pacific. The simulated age changes are lessthan 200 years, while the compiled data of 14C ventilation agesprovide ages 800 years younger during H1 than LGM reconstruc-tion (Okazaki et al., 2010). This finding indicates that our simula-tion with MIROC does not reproduce the amplitude of theventilation change recorded by paleoproxy data. The appliedfreshwater pulse may not have been sufficiently long to repro-duce the overall changes in the North Pacific Ocean that occurredduring H1. Furthermore, the underestimation could be caused bythe minor change in PMOC strength in our simulation. In fact,when the PMOC intensity increases up to 10 Sv in an idealizedexperiment by LOVECLIM, intermediate and deep waters displayyounger ventilation ages by approximately 1000 years (Okazakiet al., 2010). Understanding of the extent of deepwater formationand its migration trend eastward is needed to comprehend thedynamics of North Pacific intermediate water during Heinrichevents.

The low nutrient signal in the subpolar North Pacific (Fig. 9Aand B) is well documented by the isotopic change of benthicforaminifera at 1366 m depth in the northwestern Pacific sedi-ment cores obtained during H1 (Hoshiba et al., 2006). In addition,sea ice capping weakens biological production (Fig. 10A and B),which is consistent with the record of sediment opal contentsfrom northern Japan (Kienast et al., 2004 and reference therein).Brunelle et al. (2007) suggest that surface nutrient was relativelyabundant in the North Pacific, but biological production remainedstable. Our experiment results lack these responses. Easternequatorial Pacific sediment cores show enhanced export POC(Kienast et al., 2006), which may be compatible with theenhanced biological productivity in LOVECLIM (Fig. 10B).

4.2. Combined effects of PMOC intensity and mixing on water

properties

In the water hosing experiment, MIROC simulates only aslightly enhanced PMOC, whereas LOVECLIM simulates a largely

intensified PMOC state (Fig. 3). Although the PMOC states differbetween the two models, both obtain a similar pattern ofoxygenated and low-nutrient water propagation in the inter-mediate and deep waters of the North Pacific. The intermediatewater oxygenation would be affected by three factors: PMOCintensity, mixing, and biological production. In MIROC, the signalof oxygenated water extends in the upper 2000 m of the entireNorth Pacific and deeper than 2000 m north of 501N (Fig. 7A),apparently due to increasing intermediate water ventilationthrough PMOC intensity, vertical mixing, and convection. Withregard to PMOC intensification, the return flow of Pacific deepwater and upwelling decreases, which prevents the supply of lowoxygen content deep water to intermediate depths in the tropics.Indeed, the reduction of upwelling from the abyssal ocean wouldcontribute to the warming of the tropical intermediate waters.The intermediate water migrates eastward along the isopycnalhorizon, extending from the northern North Pacific (Fig. 8B). Thissuggests that mixing across isopycnal horizons supplies oxygen atintermediate depths. In addition, because cold water enhancesoxygen solubility, physical processes in colder surface waters candeliver more oxygenated waters to the subsurface. Weakening ofexport production may have also changed the oxygen conditionbelow the euphotic zone. Because decrease in export productionreduces remineralization that consumes oxygen within the water,significant reduction in export POC would contribute to theoxygen increase in the tropical Pacific Ocean. In MIROC, changesin PMOC strength, mixing, and biological processes play a role inincreasing oxygen of the subsurface and intermediate waters. InLOVECLIM, on the contrary, the PMOC state is a dominantcontrolling mechanism for the oxygenated and low-nutrientwater transport into the high-latitude deep ocean. Indeed, theoxygenated water extends to depths of 3000 m (Fig. 7B). Furthersensitivity studies and analysis of proxy data at depths greaterthan 2000 m will reveal the quantitative association of the deep-water formation and mixing strengths with intermediate anddeep water properties in the North Pacific Ocean.

The difference in PMOC structures generated by MIROC andLOVECLIM in the hosing experiment may be affected by thedifference in background climate states in the LGM control run.The simulated glacial tropical SSTs are 2.8 and 1.4 1C lower inMIROC and LOVECLIM, respectively, compared with the prein-dustrial climate state. This cooling respectively represents theupper and lower limits of the proxy record reconstruction (1.7–2.7 1C) (Ballantyne et al., 2005; MARGO Project Members, 2009).The North Pacific SSTs are lowered by 2.6 1C in MIROC and 2.2 1Cin LOVECLIM. The glacial mean North Pacific cooling is stronger inMIROC compared with that in LOVECLIM. In addition, the glacialsalinity field and corresponding density stratification (Fig. 5A andB) may have contributed to PMOC sensitivity. In MIROC, reducedevaporation north of 401 N (not shown) contributes to the salinitydecrease at the sea surface. In contrast, in LOVECLIM, theprecipitation minus evaporation slightly decreases (not shown),thereby contributing to the salinity increase over the easternNorth Pacific. Moreover, it should be noted that LOVECLIM due toits relatively weak trade winds underestimates the freshwatertransport from the Atlantic to the Pacific under present-dayconditions. To correct this effect a weak freshwater redistributionis applied. The effect of this present-day correction on thegeneration of high North Pacific salinity under glacial freshwaterhosing condition needs to be further explored. However, theamplitude of change in precipitation minus evaporation is smallerin LOVECLIM than that in MIROC, which suggests that the changeis insufficient to cause a larger North Pacific salinity change inLOVECLIM compared to MIROC. The meridional gradients insurface salinity may also contribute to the change in surfacesalinity (Fig. 4D). Positive feedback between the saline water and

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126124

the strong PMOC state is possible (Saenko et al., 2003). Salinityincrease at high latitudes produces water of greater density at thesurface, allowing stratification to collapse. This response wouldamplify the PMOC. Moreover, the intensified PMOC supplies thesubtropical saline water to high latitudes. In this case, thestronger PMOC would be maintained as a positive feedback, inaccordance with the Stommel feedback scenario (Stommel, 1961).In contrast, in MIROC, the salinity in the North Pacific and thePMOC state would be more stable. Schmittner et al. (2007) arguedthat the PMOC response to the AMOC shutdown would be alteredby the basic North Pacific stratification conditions.

We find that the PMOC is established by changes in oceanicheat and salt transport in response to the AMOC shutdown inaddition to changes in atmospheric heat and moisture transport.Our interpretation of the pan-oceanic interplay is in agreementwith Menviel et al. (2012) who discuss the efficiency of theStommel feedback.

4.3. Global responses of carbon and nitrate to AMOC weakening

Ice core records from Antarctica reveal that the atmosphericpCO2 increased gradually by about 30 ppm during H1 (Monninet al., 2001). Previous studies highlighted the contribution ofvegetation change to the atmospheric pCO2 increase (Obata,2007) and its sensitivity to the background climate state(Menviel et al., 2008). In this study, the ocean acts as a carbonsink in both MIROC and LOVECLIM. In MIROC atmospheric pCO2

decreases by 15.1 ppmv, neglecting changes in the terrestrialcarbon cycle. In LOVECLIM, the terrestrial carbon stock decreases,leading to a net atmospheric pCO2 decrease of 4.4 ppmv.

Schmittner and Galbraith (2008) obtained an atmosphericpCO2 increase over a thousand-year time scale of AMOC shut-down using an earth system climate model. In their experiment,the atmospheric pCO2 increase was caused by a decrease in themarine carbon reservoir as a result of reduced biological produc-tivity and oceanic carbon sequestration. The ocean biogeochem-ical contribution to atmospheric pCO2 increase was also shown bya long-term simulation using an earth system model; however,the results were insufficient to explain the full pCO2 increase(Chikamoto et al., 2008).

5. Conclusion

The responses of North Pacific intermediate and deep waterventilation and biogeochemical fields to North Atlantic fresh-water perturbations were evaluated using the MIROC and LOVE-CLIM models. In our study, the North Pacific Ocean circulation andcorresponding water properties were sensitive to the AMOCshutdown. Both models succeeded in reproducing the deeperpenetration of high-oxygen and low-nutrient younger water atintermediate depths in the North Pacific Ocean. The existence ofoxygenated water was governed by the transport of oxygenatedwater due to the PMOC, as suggested by Okazaki et al. (2010).However, the PMOC was significantly weaker in MIROC than inLOVECLIM. Therefore, the North Pacific water chemistry in MIROCwas additionally affected by three factors: (1) weakening ofPacific deep water return flow linked to the PMOC establishmentreduced delivery of oxygen-deficient deep water to the inter-mediate depth in the tropics; (2) mixing transported oxygen-richsurface water to the subsurface and intermediate layers; and(3) significant reduction in biological production may have con-tributed to the decrease in oxygen consumption through remi-neralization, resulting in increased oxygen concentrations belowthe euphotic zone. The combined effects of PMOC, mixing, and

biological processes altered the characteristics of the subsurfaceand intermediate waters.

The difference in PMOC states between MIROC and LOVECLIMmay be explained by different initial stratifications in the sub-polar North Pacific. In MIROC, the winter mixed layer depth wasdeeper near Kamchatka margin than the control condition. Con-versely, in LOVECLIM, significant deepening of mixed layer andconvection depth occurred in the entire subpolar North Pacificwhere the mixed layer originally developed in the controlexperiment. This mechanism may reveal that the simulation withweaker stratification allowed for a collapse of stratification andassociated deepwater formation in the subpolar North Pacific. Inaddition, the representation of density stratification based onglacial salinity and temperature fields may have been instru-mental for abrupt climate transitions in a coupled atmosphere–ocean model.

Limited paleoproxy record from the North Pacific and datingissues make it impossible to conclude on which model can bestrepresent Heinrich events. In addition, due to coarseness of thesemodels, locations of deep-water formation might be very modeldependent.

Water properties in the North Pacific were simulated usingglacial boundary conditions with a closed Bering Strait. If the BeringStrait was opened, the anomalous freshwater transport from theArctic to the North Pacific Ocean would cause a significant salinitydecrease in the North Pacific Ocean (Okumura et al., 2009). Ourresults emphasize that the glacial climate fields and conditions areimportant for understanding ocean stratification, ventilation, andassociated biogeochemical fields in the North Pacific Ocean duringHeinrich events.

Acknowledgments

We would like to thank two anonymous external reviewers andthe editor, K. Takahashi, for helpful comments. The authors are alsograteful to H. Tatebe for his stimulating discussion. This researchwas conducted by JAMSTEC-IPRC Initiative (JII) project. The numer-ical simulations were preformed on the Earth Simulator at JAMSTECand HITACHI SR11000 at University of Tokyo. AT was supportedthrough the National Science Foundation grant AGS 1010869 andJAMSTEC through its co-sponsorship of the IPRC.

References

Ahagon, N., Ohkushi, K., Uchida, M., Mishima, T., 2003. Mid-depth circulation inthe northwest Pacific during the last deglaciation: evidence from foraminiferaradiocarbon ages. Geophys. Res. Lett. 30, 2097. doi:10.1029/2003GL018287.

Altabet, M.A., Higginson, M.J., Murray, D.W., 2002. The effect of millennial-scalechanges in Arabian Sea denitrification on atmospheric CO2. Nature 415, 159–162.

Ballantyne, A.P., Lavine, M., Crowley, T.J., Liu, J., Barker, P.B., 2005. Meta-analysis oftropical surface temperatures during the Last Glacial Maximum. Geophys. Res.Lett. 32, L05712. doi:10.1029/2004GL021217.

Bard, E., Rostek, F., Sonzogni, C., 1997. Interhemispheric synchrony of the lastdeglaciation inferred from alkenone paleothermometry. Nature 385, 707–710.

Bard, E., Rostek, F., Turon, J.L., Gendreau, S., 2000. Hydrological impact of Heinrichevents in the subtropical northeast Atlantic. Science 289, 1321–1324.

Behl, R.J., Kennett, J.P., 1996. Brief interstadial events in the Santa Barbara basin,NE Pacific, during the past 60 kyr. Nature 379, 243–246.

Berger, A., 1978. Long-term variations of caloric insolation resulting from theearth’s orbital elements. Quatern. Res. 9, 139–167.

Bond, G., Lotti, R., 1995. Iceberg discharges into the North Atlantic on millennialtime scales during the last glaciation. Science 267, 1005–1010.

Braconnot, P., Otto-Bliesner, B., Harrison, S., Joussaume, S., Peterchmitt, J.-Y., Abe-Ouchi, A., Crucifix, M., Driesschaert, E., Fichefet, T., Hewitt, C.D., Kagetama, M.,Kitoh, A., Laine, A., Loutre, M.-F., Marti, O., Merkel, U., Ramstein, G., Valdes, P.,Weber, S.L., Yu, Y., Zhao, Y., 2007. Results of PMIP2 coupled simulations of theMid-Holocene and Last Glacial Maximum. Part 1. Experiments and large-scalefeatures. Clim. Past 3, 261–277.

Brunelle, B.G., Sigman, D.M., Cook, M.S., Keigwin, L.D., Haug, G.H., Plessen, B.,Schettler, G., Jaccard, S.L., 2007. Evidence from diatom-bound nitrogenisotopes for subarctic Pacific stratification during the last ice age and a link

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126 125

to North Pacific denitrification changes. Paleoceanography 22, PA1215. doi:10.1029/2005PA001205.

Brunelle, B.G., Sigman, D.M., Jaccard, S.L., Keigwin, L.D., Plessen, B., Schettler, G.,Cook, M.S., Haug, G.H., 2010. Glacial/interglacial changes in nutrient supplyand stratification in the western subarctic North Pacific since the penultimateglacial maximum. Quatern. Sci. Rev. 29, 2579–2590.

Chikamoto, M.O., Matsumoto, K., Ridgwell, A., 2008. Response of deep-sea CaCO3

sedimentation to Atlantic meridional overturning circulation shutdown.J. Geophys. Res. 113, G03017. doi:10.1029/2007JG000669.

Chikamoto, M.O., Abe-Ouchi, A., Oka, A., Ohgaito, R., Timmermann, A., 2011.Glacial marine carbon cycle sensitivities to Atlantic ocean circulation reorga-nization by coupled climate model simulations. Clim. Past Discuss. 7,1261–1299. doi:10.5194/cpd-7-1261-2011.

Dansgaard, W., Johnsen, S.J., Clausen, H.B., Dahl-Jensen, D., Gundestrup, N.S.,Hammer, C.U., Hvidberg, C.S., Steffensen, J.P., Sveinbjornsdottir, A.E., Jouzel,J., Bond, G., 1993. Evidence for general instability of past climate from a 250-kyr ice-core record. Nature 364, 218–220.

Dyke, S.A., Dale, J.E., McNeely, R.N., 1996. Marine molluscs as indicators ofenvironmental change in glaciated North America and Greenland during thelast 18 000 years. Geogr. Phys. Quatern. 50, 125–184.

Goosse, H., Brovkin, V., Fichefet, T., Haarsma, R., Huybrechts, P., Jongma, J.,Mouchet, A., Selten, Y., Campin, J.-M., Deleersnijder, E., Driesschaert, E.,Goelzer, H., Janssens, F., Morales Maqueda, M.A., Opsteegh, T., Mathieu, P.-P.,Munhoven, G., Pettersson, E.J., Renssen, H., Roche, D.M., Schaeffer, M., Tartin-ville, B., Timmermann, A., Weber, S.L., 2010. Description of the Earth systemmodel of intermediate complexity LOVECLIM version 1.2. Geosci. Model Dev.3, 603–633. doi:10.5194/gmd-3-603-2010.

Harada, N., Ahagon, N., Sakamoto, T., Uchida, M., Ikehara, M., Shibata, Y., 2006.Rapid fluctuation of alkenone temperature in the southwestern Okhotsk Seaduring the past 120 kyr. Global Planet. Change 53, 29–46.

Harada, N., Sato, M., Sakamoto, T., 2008. Freshwater impacts recorded in tetra-unsaturated alkenones and alkenone sea surface temperatures from theOkhotsk Sea across millennial-scale cycles. Paleoceanography 23, PA3201.doi:10.1029/2006PA001410.

Harada, N., Sato, M., Seki, O., Timmermann, A., Moossen, H., Bendle, J., Nakamura,Y., Kimoto, K., Okazaki, Y., Nagashima, K., Gorbarenko, S.A., Ijiri, A., Nakatsuka,T., Menviel, L., Chikamoto, M.O., Abe-Ouchi, A., Schouten, S., 2012. Sea surfaceand subsurface temperature changes in the Okhotsk Sea and adjacent NorthPacific during the last glacial maximum and deglaciation. Deep-Sea Res. II61–64, 93–105.

Heinrich, H., 1988. Origin and consequences of cyclic ice rafting in the NortheastAtlantic Ocean during the past 130,000 years. Quatern. Res. 29, 142–152.

Hoshiba, M., Ahagon, N., Ohkushi, K., Uchida, M., Motoyama, I., Nishimura, A.,2006. Foraminiferal oxygen and carbon isotopes during the last 34 kyr offnorthern Japan, northwestern Pacific. Mar. Micropaleontol. 61, 196–208.

Hu, A., Meehl, G.A., Han, W., 2007. Role of the Bering Strait in the thermohalinecirculation and abrupt climate change. Geophys. Res. Lett. 34, L05704.doi:10.1029/2006GL028906.

Hu, A., Meehl, G.A., Otto-Bliesner, B.L., Waelbroeck, C., Han, W., Loutre, M.-F.,Lambeck, K., Mitrovica, J.X., Rosenbloom, N., 2010. Influence of Bering Straitflow and North Atlantic circulation on glacial sea-level changes. Nat. Geosci. 3,118–121. doi:10.1038/NGEO729.

K-1 Model Developers 2004. In: Hasumi, H., Emori, S. (Ed.). K-1 Coupled GCM(MIROC) Description. K-1 Technical Report No. 1. 34 pp.

Kennett, J.P., Roark, E.B., Cannariato, K.G., Ingram, L., Tada, R., 2000. Latestquaternary paleoclimatic and radiocarbon chronology, Hole 1017E, southernCalifornia margin. Proc. Ocean Drilling Program Sci. Results 167, 249–254.

Kiefer, T., Kienast, M., 2005. Patterns of deglacial warming in the Pacific Ocean:a review with emphasis on the time interval of Heinrich event 1. Quatern. Sci.Rev. 24, 1063–1081.

Kienast, M., Kienast, S.S., Calvert, S.E., Eglinton, T.I., Mollenhauer, G., Francois, R.,Mix, A.C., 2006. Eastern Pacific cooling and Atlantic overturning circulationduring the last deglaciation. Nature 443, 846–849.

Kienast, S.S., Hendy, I.L., Crusius, J., Pedersen, T.F., Calvert, S.E., 2004. Exportproduction in the subarctic North Pacific over the last 800 kyrs: no evidencefor iron fertilization? J. Oceanogr. 60, 189–203.

Kohfeld, K.E., Le Quere, C., Harrison, S.P., Anderson, R.F., 2005. Role of marinebiology in glacial–interglacial CO2 cycles. Science 308, 74–78.

Leduc, G., Vidal, L., Tachikawa, K., Rostek, F., Sonzogni, C., Beaufort, L., Bard, E.,2007. Moisture transport across Central America as a positive feedback onabrupt climatic changes. Nature 445, 908–911.

Marchitto, T.M., Lehman, S.J., Ortiz, J.D., Fluckiger, J., Van Geen, A., 2007. Marineradiocarbon evidence for the mechanism of deglacial atmospheric CO2 rise.Science 316, 1456–1459.

MARGO Project Members, 2009. Constraints on the magnitude and patterns ofocean cooling at the Last Glacial Maximum. Nat. Geosci. 18, 127–132.

McManus, J.F., Francois, R., Gherardi, J.-M., Keigwin, L.D., Brown-Leger, S., 2004.Collapse and rapid resumption of Atlantic meridional circulation linked todeglacial climate changes. Nature 428, 834–837.

Menviel, L., 2008. Climate–Carbon Interactions on Millennial to Glacial Timescalesas Simulated by a Model of Intermediate Complexity, LOVECLIM. PhD Thesis.Univ. of Hawaii, USA. p. 179.

Menviel, L., Timmermann, A., Mouchet, A., Timm, O., 2008. Meridional reorganiza-tions of marine and terrestrial productivity during Heinrich events. Paleocea-nography 23, PA1203. doi:10.1029/2007PA001445.

Menviel, L., Timmermann, A., Timm, O., Mouchet, A., AbeOuchi, A., Chikamoto,M.O., Harada, N., Ohgaito, R., Okazaki, Y., 2012. Removing the North Pacifichalocline: effects on global climate, ocean circulation and the carbon cycle.Deep-Sea Res. II 61–64, 106–113.

Mix, A.C., Lund, D.C., Pisias, N.G., Boden, P., Bornmalm, L., Lyle, M., Pike, J., 1999.Rapid climate oscillations in the northeast Pacific during the last deglaciationreflect northern and southern hemisphere sources. Mechanisms of globalclimate change at millennial time scales. Geophys. Monogr. 112, 127–148.

Monnin, E., Indermuhle, A., Dallenbach, A., Fluckiger, J., Stauffer, B., Stocker, T.F.,Raynaud, D., Barnola, J.-M., 2001. Atmospheric CO2 concentrations over thelast glacial termination. Science 291, 112–114.

Obata, A., 2007. Climate–carbon cycle model response to freshwater discharge intothe North Atlantic. J. Clim. 20, 5962–5976.

Oka, A., Kato, S., Hasumi, H., 2008. Evaluating effect of ballast mineral on deep-ocean nutrient concentration by using an ocean general circulation model.Global Biogeochem. Cycles 22, GB3004. doi:10.1029/2007GB003067.

Oka, A., Abe-Ouchi, A., Chikamoto, M.O., Ide, T., 2011. Mechanisms controllingexport production at the LGM: effects of changes in oceanic physical field andatmospheric dust deposition. Global Biogeochem. Cycles 25, GB2009.doi:10.1029/2009GB003628.

Okazaki, Y., Takahashi, K., Asahi, H., Katsuki, K., Hori, J., Yasuda, H., Sagawa, Y.,Tokuyama, H., 2005. Productivity changes in the Bering Sea during the lateQuaternary. Deep-Sea Res. II 52, 2150–2162.

Okazaki, Y., Timmermann, A., Menviel, L., Harada, N., Abe-Ouchi, A., Chikamoto,M.O., Mouchet, A., Asashi, H., 2010. Deepwater formation in the North Pacificduring the last glacial termination. Science 329, 200–204.

Okumura, Y.M., Deser, C., Hu, A., Timmermann, A., Xie, S.-P., 2009. North Pacificclimate response to freshwater forcing in the subarctic North Atlantic: oceanicand atmospheric pathways. J. Clim. 22, 1424–1445.

Palmer, J.R., Totterdell, I.J., 2001. Production and export in a global oceanecosystem model. Deep-Sea Res. I 48, 1169–1198.

Partin, J.W., Cobb, K.M., Adkins, J.F., Clark, B., Fernandez, D.P., 2007. Millennial-scale trends in west Pacific warm pool hydrology since the last glacialmaximum. Nature 449, 452–455.

Peltier, W.R., 2004. Global glacial isostasy and the surface of the ice-age Earth: theICE-5G (VM2) Model and GRACE. Annu. Rev. Earth Planet. Sci. 32, 111–149.

Peterson, L.C., Haug, G.H., Hughen, K.A., Rohl, U., 2000. Rapid changes in the hydrologiccycle of the tropical Atlantic during the last glacial. Science 290, 1947–1951.

Redfield, A.C., Ketchum, B.H., Richards, F.A., 1963. The influence of organisms onthe composition of sea-water. In: Hill, M.N. (Ed.), The Sea, vol. 2. Interscience,New York, pp. 26–77.

Rickaby, R.E.M., Elderfield, H., 2005. Evidence from the high-latitude North Atlanticfor variations in Antarctic Intermediate water flow during the last deglaciation.Geochem. Geophys. Geosyst. 6, Q05001. doi:10.1029/2004GC000858.

Robinson, R.S., Sigman, D.M., DiFiore, P.J., Rohde, M.M., Mashiotta, T.A., Lea, D.W.,2005. Diatom-bound 15N/14N: new support for nutrient consumption in the iceage subantarctic. Paleoceanography 20, PA3003. doi:10.1029/2004PA001114.

Rosell-Mele, A., Maslin, M.A., Maxwell, J.R., Schaeffer, P., 1997. Biomarker evidencefor ‘‘Heinrich’’ events. Geochim. Cosmochim. Acta 61, 1671–1678.

Saenko, O.A., Weaver, A.J., Gregory, J.M., 2003. On the link between the two modesof the ocean thermohaline circulation and the formation of global-scale watermasses. J. Clim. 16, 2797–2801.

Saenko, O.A., Schmittner, A., Weaver, A.J., 2004. The Atlantic–Pacific seesaw.J. Clim. 17, 2033–2038.

Sagawa, T., Ikehara, K., 2008. Intermediate water ventilation change in thesubarctic northwest Pacific during the last deglaciation. Geophys. Res. Lett.35, L24702. doi:10.1029/2008GL035133.

Sarnthein, M., Winn, K., Jung, S.J.A., Duplessy, J., Labeyrie, L., Erlenkeuser, H., Ganssen,G., 1994. Changes in east Atlantic deepwater circulation over thelast 30,000 years: eight time slice reconstructions. Paleoceanography 9, 209–267.

Schmittner, A., Galbraith, E.D., 2008. Glacial greenhouse-gas fluctuations con-trolled by ocean circulation changes. Nature 456, 373–376.

Schmittner, A., Galbraith, E.D., Hostetler, S.W., Pedersen, T.F., Zhang, R., 2007. Largefluctuations of dissolved oxygen in the Indian and Pacific oceans duringDansgaard-Oeschger oscillations caused by variations of North Atlantic DeepWater subduction. Paleoceanography 22, PA3207. doi:10.1029/2006PA001384.

Seki, O., Ishiwatari, R., Matsumoto, K., 2002. Millennial climate oscillations in NEPacific surface waters over the last 82 kyr: new evidence from alkenones.Geophys. Res. Lett. 29, 2144. doi:10.1029/2002GL015200.

Stommel, H., 1961. Thermohaline convection with two stable regimes of flow.Tellus 13, 224–230.

Stott, L., Southon, J., Timmermann, A., Koutavas, A., 2009. Radiocarbon ageanomaly at intermediate water depth in the Pacific Ocean during the lastdeglaciation. Paleoceanography 24, PA2223. doi:10.1029/2008PA001690.

Stouffer, R.J., Yin, J., Gregory, J.M., Dixon, K.W., Spelman, M.J., Hurlin, W., Weaver,A.J., Eby, M., Flato, G.M., Hasumi, H., Hu, A., Jungclaus, H., Kamenkovich, I.V.,Levermann, A., Montoya, M., Murakami, S., Nawrath, S., Oka, A., Peltier, W.R.,Robitaille, D.Y., Sokolov, A., Vettoretti, G., Weber, S.L., 2006. Investigating thecauses of the response of the thermohaline circulation to past and futureclimate changes. J. Clim. 19, 1365–1387.

Tachikawa, K., Vidal, L., Sonzogni, C., Bard, E., 2009. Glacial/interglacial sea surfacetemperature changes in the Southwest Pacific ocean over the past 360 ka.Quatern. Sci. Rev. 28, 1160–1170.

Timmermann, A., Krebs, U., Justino, F., Goosse, H., Ivanochko, T., 2005. Mechanismsfor millennial-scale global synchronization during the last glacial period.Paleoceanography 20, PA4008. doi:10.1029/2004PA001090.

M.O. Chikamoto et al. / Deep-Sea Research II 61-64 (2012) 114–126126

Vidal, L., Labeyrie, L., Cortijo, E., Arnold, M., Duplessy, J.D., Michel, E., Becque, S., vanWeering, T.C.E., 1997. Evidence for changes in the North Atlantic Deep Waterlinked to meltwater surges during the Heinrich events. Earth Planet. Sci. Lett. 146,13–27.

Vidal, L., Schneider, R.R., Marchal, O., Bickert, T., Stocker, T.F., Wefer, G., 1999. Linkbetween the North and South Atlantic during the Heinrich events of the lastglacial period. Clim. Dyn. 15, 909–919.

Wang, X., Auler, A.S., Edwards, R.L., Cheng, H., Cristalli, P.S., Smart, P.L., Richards,D.A., Shen, C.-C., 2004. Wet periods in northeastern Brazil over the past210 kyr linked to distant climate anomalies. Nature 432, 740–743.

Wang, Y.J., Cheng, H., Edwards, R.L., An, Z.S., Wu, C., Dorale, J.A., 2001. A high-resolution absolute-dated late Pleistocene monsoon record from Hulu Cavebb

China. Science 294, 2345–2348.