University of Southern California Dissertations and Theses

New insights into glacial-interglacial carbon cycle: multi-proxy and numerical modeling

(USC Thesis Other)

New insights into glacial-interglacial carbon cycle: multi-proxy and numerical modeling

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
NEW INSIGHTS INTO GLACIAL-INTERGLACIAL CARBON
CYCLE:
MULTI-PROXY AND NUMERICAL MODELING
by
Jun Shao

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
DOCTOR OF PHILOSOPHY
(GEOLOGICAL SCIENCES)

December 2020

Copyright 2020 Jun Shao

ii
Acknowledgements
Thanks everybody.

iii

Table of Contents
Acknowledgements ................................................................................................................ ii
List of Figures ......................................................................................................................... v
List of Tables .......................................................................................................................... xi
Abstract ................................................................................................................................. xii
1 Chapter 1 ........................................................................................................................ 1
1.1 Introduction ............................................................................................................................... 1
1.2 Outline of the Dissertation Chapters .......................................................................................... 3
2 Chapter 2 ........................................................................................................................ 7
2.1 Introduction ............................................................................................................................... 8
2.2 Methods: ................................................................................................................................. 13
2.2.1 Oceanographic Settings for Cores PC75-2 and PC83-1 ........................................................................... 13
2.2.2 Age Model for PC75-2 and PC83-1 ........................................................................................................... 15
2.2.3 Trace Element and δ
11
B Analyses ............................................................................................................. 19
2.2.4 Salinity estimate ....................................................................................................................................... 20
2.2.5 Paired SST/pH and pCO 2 estimate ............................................................................................................ 21
2.2.6 Composite pH and Seawater pCO 2 ........................................................................................................... 23
2.2.7 Modeling the GMSSpH with cGENIE Earth System Model ....................................................................... 27
2.3 Results ..................................................................................................................................... 28
2.3.1 Two New Boron Isotope-Based pH Reconstructions from the SW Pacific ............................................... 28
2.3.2 pH and Seawater pCO 2 Composite vs the Simulated GMSSpH ................................................................ 29
2.4. Discussion ............................................................................................................................... 36
2.4.1 Chatham Rise as a Source of Carbon to the Atmosphere over the Last 19 kyr. ....................................... 36
2.4.2 Positive Air-Sea ∆pCO 2 Anomalies during the deglaciation and early Holocene ...................................... 44
3 Chapter 3 ...................................................................................................................... 49
3.1 Introduction ............................................................................................................................. 49
3.2 Methods .................................................................................................................................. 52
3.3 Results ..................................................................................................................................... 56
3.4 Atmospheric δ
13
C Bridge .......................................................................................................... 65
3.5 Revisiting EEP Thermocline δ
13
C ............................................................................................... 67
3.6 How deep can the low preformed δ
13
C signal reach during the early deglaciation? ................... 69
4 Chapter 4 ...................................................................................................................... 73
4.1 Introduction ............................................................................................................................. 74
4.2 Methods .................................................................................................................................. 77

iv
4.3 Correlation between air-sea CO2 flux anomaly and surface δ
13
C anomaly in sensitivity
experiments ................................................................................................................................... 79
4.4 Spatial and temporal surface δ
13
C and air-sea CO2 flux pattern in transient simulations ........... 80
4.5 Deglacial surface SO δ
13
C, a model-data comparison ................................................................ 85
5 Chapter 5 ...................................................................................................................... 89
5.1 Introduction ............................................................................................................................. 90
5.2 Methods .................................................................................................................................. 94
5.3 transient evolution of
14
Cpro, atmospheric ∆
14
C and bulk ocean ∆
14
C in cGENIE experiments ... 95
5.4 Magnitude of deglacial geological flux compared to modern geological flux ............................. 97
5.5 Implications to ocean carbonate chemistry and atmospheric pCO2 ......................................... 100
6 Chapter 6 .................................................................................................................... 103
6.1 Conclusions ............................................................................................................................ 103
References .......................................................................................................................... 106

v
List of Figures
Figure 2.1. a) Map of mean annual ∆pCO2 sea-air. Oceanic sinks and sources are presented by
negative and positive ∆pCO2 sea-air values (Takahashi et al., 2009), respectively. The green
triangle shows the location of the studied sediment cores PC75-2 and PC83-1. The Circles show
the locations of the cores with published boron isotope data that span the last deglaciation. b)
Salinity of the P15 transect from the World Ocean Circulation Experiment. The yellow star shows
the location of corals (Hines et al., 2015); the black circles show the locations of other sediment
cores (Allen et al., 2015; Ronge et al., 2016) discussed in this study. The red-dotted line marks the
boundary between Antarctic Intermediate Water (AAIW) and upper circumpolar Deep Water
(UCDW). ....................................................................................................................................... 10

Figure 2.2. 12 published boron records span the last deglaciation grouped into 4 species. (a) δ
11
B
measured in T. sacculifer, ODP1238 (red) (Martínez-Botí et al., 2015), EDRC92 (brown) (Palmer
& Pearson, 2003)and NIOP464 (pink) (Palmer et al., 2010). (b) δ
11
B measured in N. pachyderma,
RAPiD-15-4p (yellow) (Yu et al., 2013), MD01-2416 (blue) (Gray et al., 2018a) and JM11-PI-
19PC (dark purple) (Ezat et al., 2017). (c) δ
11
B measured in G. ruber, GeoB1105 (orange) (Foster
& Sexton, 2014), ODP999 (light purple) (Foster, 2008), GeoB 1523 (dark green) (Henehan et al.,
2013) and AA2.16 (gray) (Naik et al., 2015) . (d) δ
11
B measured in G. bulloides (magenta)
(Martínez-Botí et al., 2015). Records generated on a MC-ICPMS are marked with circles and
records generated on a NTIMS are marked with diamonds. circles and record ........................... 11

Figure 2.3. Meridional sections of ∆pCO2 and CO2 flux reanalysis data at at 177°E (year 2008-
2012) from biogeochemical-sea ice-ocean state estimate (B-SOSE) as part of the SOCCOM
project (Verdy & Mazloff, 2017). The black (red) lines are 5-year ‘climatology’ of ∆pCO2 and
CO2 flux between September to February (March to August) ..................................................... 14

Figure 2.4. left panel: PC75-1 and PC75-2 raw planktic radiocarbon dates versus depth; right panel:
PC75-1 and PC75-2 G. inflata oxygen isotope versus depth ........................................................ 16

Figure 2.5. PC75-2 and PC83-1 G. inflata stable oxygen and carbon isotope. Based on the planktic
stable isotope stratigraphy, we aligned the age of 88cm of PC83-1 to 88cm of PC75-2, the age of
which is 11850 yrBP (the vertical black lines). ............................................................................ 18

Figure 2.6. Mg/Ca and boron isotope data plotted with Al/Ca. ................................................... 20

Figure 2.7. Deglacial changes in seawater carbonate chemistry and sea surface temperature of
Chatham Rise. a-d, PC75-2; e-h, PC83-1 (a, e) G. bulloides d
11
B (versus boric acid standard NIST
951) with analytical uncertainties (2 SD). (b, f) G. bulloides Mg/Ca-based SST with uncertainties
(1 SD) (c, g) d
11
B-based pH. (d, h) d
11
B-derived pCO2. The dotted envelopes represent uncertainty
of sea surface pH and pCO2 (1 SD). Triangle marks represent intervals with radiocarbon age
constraints. .................................................................................................................................... 29

Figure 2.8. Individual pH reconstructions (corrected for ‘location bias’, i.e. pHfinal = pHboron -
pHboron_Holocene + pHglobal_preind) plotted against the composite pH. ................................................ 31

vi
Figure 2.9. Individual uncorrected pH reconstructions (i.e. pHboron) plotted against the composite
pH built upon pHfinal. The red dots represent pHsite_preind calculated from GLODAP V2 2016b
(Lauvset et al., 2016). ................................................................................................................... 32

Figure 2.10. The climatology of seasonal and annual mean surface pCO2 (year 2001-2015)
(Landschützer et al., 2017) for each core site. The data is averaged over a 10° by 10° box around
the core site. The black bars represent 1 standard error over the 15-year time period. Among these
cores, JM11, Rapid 15 and MD01-2416 from high latitude oceans are characterized with strong
seasonality. Rapid 15 is not included in our composite because it does not cover the LGM. For
JM11, sediment trap data (Jonkers & Kučera, 2015) near the Norwegian Sea suggest that these is
no N. pachyderma flux during winter, while for the rest of the year, the N. pachyderma flux is
evenly distributed. Thus the geochemical signature of N. pachyderma would be an underestimate
of the surface pCO2. For MD01-2416, sediment trap data (Jonkers & Kučera, 2015) in the modern
northwest Pacific display two abundance peaks during the spring (with high pCO2) and autumn
(with low pCO2), the two seasonal fluxes are roughly equal and the average pCO2 during the two
seasons are close to the annual mean. ........................................................................................... 33

Figure 2.11. pCO2 gradient between surface and the average of the first 50m at the core sites, the
station data from GLODAP V2 (Olsen et al., 2016) is averaged over a 15° by 15°box around the
core sites. For most of the cores in our composite, the depth difference is minimal (i.e. even though
the foraminifera may not calcify not strictly at the surface, but the seawater carbonate chemistry
state they record may still represent the surface condition). At the site of NIOP464, the surface-
subsurface difference is up to 50 ppm on average, however the station data were all taken in a
single cruise in Aug 1995, thus it is not clear of those station data are truly representative to the
long term condition. The site of ODP1238 is affected by El-Niño, the station data were taken
between year 1992-1994 when El-Niño was in the positive phase (i.e. the eastern equatorial Pacific
is more stratified). Thus the 100~150 ppm of surface-subsurface ΔpCO2 should be taken as a
maximum. ..................................................................................................................................... 34

Figure 2.12. composite pH and pCO2. a) The solid blue line denotes the median of the 10,000
simulations of the composite pH values. The dotted blue line denotes the 95% quantiles of the
10,000 simulations of the composite pH values. The grey dots represent individual boron isotope
derived pH (n=219) after corrections (i.e. pHfinal). b) composite seawater pCO2 versus the ice core
record of atmospheric CO2. The solid blue line denotes the median of the 10,000 composite pCO2.
The dot blue line denotes the 95% quantile of the 10,000 composite pCO2. The solid black line
represents the ice core record of atmospheric pCO2. c) composite pH versus cGENIE inverted pH.
The blue lines are as in panel a); The solid brown, red, and magenta lines denote the cGENIE
simulated GMSSpH evolution where the initial conditions are achieved through ‘ALK addition’,
‘CaCO3 addition’ and ‘DIC removal’, respectively. ..................................................................... 35

Figure 2.13. a) PC75-2 and PC83-1 ∆pCO2 plotted with an opal flux record from the SW Pacific
sector of the Southern Ocean (Chase et al., 2003). The dotted envelopes represent uncertainty of
∆pCO2 (the 14% and 86% quantile). The red arrow marks the modern annual ∆pCO2 at our site
(Verdy & Mazloff 2017) b) JPC83 benthic ∆[CO3]
2-
from the Bay of Plenty (Allen et al., 2015),
plotted with PC75-1 (this study) benthic δ
13
C records. The raw data are plotted as dots; the thick
lines are 3-point running averages. c) Three records of benthic - atmospheric ∆
14
C offset (∆∆
14
C)

vii
from Chatham Rise. PC75-1 and PC75-2 radiocarbon data are combined together but are labeled
with different color coding. d) Atmospheric pCO2 (Bereiter et al., 2015). The 16.5–14 kyr BP event
is highlighted in yellow for all panels. .......................................................................................... 38

Figure 2.14. PC75 benthic - atmospheric ∆
14
C offset (∆∆
14
C) and Tasman Sea coral (43°S-47°S
144°E-152°E) – atmospheric ∆∆
14
C (Hines et al., 2015). The dotted blue line assumes a constant
reservoir age (~ 400 yrs) over the entire PC75 record. ................................................................. 40

Figure 2.15. Conceptual diagrams of SW Pacific circulation and radiocarbon distribution. (a)
Modern pattern. Site PC75-2 and site PC83-1 are bathed under well ventilated AAIW, with no
influence of geological carbon. The color shading is based on interpolating a transect of ∆
14
C data
between 170°E to 180°E from the GLODAP database to a 10° in latitude by 500m in depth grid.
(b) LGM pattern. Site PC75-2 and site PC83-1 are bathed under poorly ventilated, respired carbon-
rich UCDW. Episodic release of
14
C-dead geological carbon makes the deep/intermediate waters
as well as the surface of Southern Ocean appear to be ‘old’. The color shading is based on multiple
published ∆
14
C records from sediment cores (not labelled in the panels) that represent glacial ∆
14
C
signature of PDW, AAIW, SAMW in the Southern Hemisphere. The high latitude Southern Ocean
is labelled with question marks due to a lack of data. Panels were generated using ODV 4.7.10.
....................................................................................................................................................... 41

Figure 2.16. a): benthic δ
13
C records from the South Pacific that show first rapid increase and then
decline between 16.5-14 kyrBP ( this study; Pahnke & Zahn, 2005; Rose et al., 2010; Siani et al.,
2013). b): benthic δ
13
C records from the South Pacific that show a general increase between 16.5-
14 kyrBP (Ronge et al., 2015; Sikes et al., 2016). The lines are 3-point running average. .......... 42

Figure 2.17. a) atmospheric pCO2 (Bereiter et al., 2015); b) δ
13
C-CO2 from Schmitt et al., 2012
(green) and Bauska et al., 2016 (Olive); c) relative sea level changes (Spratt & Lisiecki, 2016); d)
the median of 10,000 composite pH from 12 cores (thick blue line), the dotted lines represent 2.5%
and 97.5% quantile, the black line with triangle markers represent the median of 10,000 composite
pH without our two new records; e) the median of 10,000 composite ∆pCO2 from 12 cores (thick
magenta line), the dotted lines represent 2.5% and 97.5% quantile; f) number of data points from
all (red bars)/upwelling (green bars) sites that go into each 1000-yr bin. Upwelling sites include
ODP1238 from the EEP, PS2498 from the sub-Antarctic Atlantic, MD01-2416 from the subarctic
Pacific, NIOP464 and AAS9 from the Northern Indian Ocean. ................................................... 46

Figure 2.18. ∆pCO2 composite with all 12 cores included (magenta), and with PC75-1 and PC83-
2 from this study excluded (black). ............................................................................................... 46

Figure 3.1. Timeseries from the LOVECLIM transient experiment (Menviel et al., 2018). a)
Freshwater input into the North Atlantic and the Southern Ocean; b) Southern Hemisphere
westerly wind forcing; c) simulated NADW, AABW, AAIW and NPIW maximum stream function
in LOVECLIM. 21-year moving averages are shown for the maximum stream function to filter
the high-frequency variability; d) ice core record of atmospheric CO2 (blue) and LOVECLIM
simulated atmospheric CO2 (red); e) ice core record of atmospheric δ
13
C (black and olive) and
LOVCLIM simulated atmospheric δ
13
C (magenta). ..................................................................... 57

viii
Figure 3.2. a) LOVECLIM simulated sea surface δ
13
C anomaly (15-17.2 ka) b) sea surface δ
13
C
anomaly due to thermodynamic fractionation (air-sea exchange + SST effect) c) residual sea
surface δ
13
C anomaly that are not attributed to thermodynamic fractionation. ............................ 59

Figure 3.3. simulated primary productivity anomaly (15-17.2ka) in LOVECLIM ..................... 60

Figure 3.4. Pacific zonal averaged (160°E-140°W) a) δ
13
C b) δ
13
Creg c) δ
13
Cpref d) PO4 anomaly
(15ka minus 17.2ka) simulated by LOVECLIM. The magenta circle marks the GeoB17402 site.
....................................................................................................................................................... 60

Figure 3.5. a): Atmospheric δ
13
C records (Bauska et al., 2016; Schmitt et al., 2012) b): C.
mundulus δ
13
C record for upper intermediate and mode waters in the western equatorial Pacific.
The negative δ
13
C excursions in the atmospheric and our benthic record are highlighted in a grey
bar. ................................................................................................................................................ 61

Figure 3.6. changes in air-sea surface pCO2 gradient (15-17.2 ka) ............................................. 63

Figure 3.7. cGENIE simulated a) Δδ
13
Creg b) Δδ
13
Cpref in experiment ‘free’ and simulated c)
Δδ
13
Creg d) Δδ
13
Cpref in experiment ‘free’. Anomaly are calculated as the difference between model
year 2000 and 0. ............................................................................................................................ 64

Figure 3.8. LOVECLIM simulated Δδ
13
C in thermocline EEP (90-82°W, 5°S-5°N, 77-105m),
South Pacific subtropical gyre (STGSP, 160°E- 100°W, 40-22°S, 187-400m), North Pacific
subtropical gyre (STGNP, 110°E- 140°W, 22-40°N, 187-400m), NPIW (167-170°E, 54-57°N,
660m. The average of 23.8-20 ka (i.e. LGM) is used as a reference level for the Δδ
13
C calculations.
The 16.2-15.8 ka excursion is highlighted with a grey bar. .......................................................... 66

Figure 3.9. a): Atmospheric δ
13
C records (Bauska et al., 2016; Schmitt et al., 2012), simulated
atmospheric δ
13
C (21-year running average) in LOVECLIM (Menviel et al., 2018) b):
Neogloboquadrina. dutertrei (N. dutertrei, a shallow thermocline species) δ
13
C data from ODP
1238 (Martínez-Botí et al., 2015), GGC17/JPC30 (Zhao and Keigwin, 2018), and LOVECLIM
simulated δ
13
C of DIC at 100m (average of 82-90°W, 5°S-5°N). The N. dutertrei data are corrected
by -0.5‰ to normalize to δ
13
C of DIC (Spero et al., 2003). The grey shadow bars highlight the
time period we focus in this study. ............................................................................................... 69

Figure 3.10. Observed and simulated δ
13
C anomaly at the Brazil Margin. ................................. 70

Figure 3.11. same as Figure 3.4, but for the Atlantic zonal averaged (60°W-10°W) anomaly. The
location of 78GGC and 33GGC (Lund et al., 2015) are marked as magenta circles. ................... 71

Figure 3.12. same as Figure 3.11, but for 17.2-19ka anomaly in the Atlantic ............................. 71

Figure 4.1. surface δ
13
C anomaly versus air-sea CO2 flux anomaly in four sensitivity experiments.
Anomalies are calculated as model year 3000-0. Each dot represents a surface grid box in cGENIE.
The black dash lines in each panel represent atmospheric δ
13
C anomaly in that particular sensitivity
experiment. .................................................................................................................................... 79

ix
Figure 4.2. Time series of applied forcing are shown in panel a, b and c. The simulated
atmospheric CO2 and δ
13
C are plotted against ice core data in panel d and e, respectively. ........ 81

Figure 4.3. δ
13
C and CO3 model-data comparison at intermediate depth of the South Pacific (Allen
et al., 2020) and South Atlantic (Lacerra et al., 2019). ................................................................. 82

Figure 4.4. a) simulated surface δ
13
C anomaly (15-18.4ka). b) simulated air-sea CO2 flux anomaly.
The purple rectangles mark the boundaries of CO2 outgassing band. The black lines represent 50%
annual sea ice coverage. ................................................................................................................ 83

Figure 4.5. Simulated air-sea CO2 flux at LGM in cGENIE. The purple rectangles are the same as
Figure 4.4. ..................................................................................................................................... 84

Figure 4.6. Simulated δ
13
C (a) and air-sea CO2 flux (b) time series at three locations with different
oceanographic settings. ................................................................................................................. 85

Figure 4.7. Simulated surface temperature (a), CO3 (b) anomaly in the Southern Hemisphere. . 86

Figure 4.8. Simulated surface temperature (a), CO3 effect (b) on foraminifera δ
13
C. Panel c is
identical as figure 4.4a. Panel d shows model-predicted δ
13
C anomaly potentially recorded by
foraminifera with all three effects taken into account. Panel c and d are plotted with compiled
planktic δ
13
C data (HS1-LGM) ..................................................................................................... 87

Figure 5.1. a) observation-based atmospheric
14
C production rate reconstructions (Adolphi et al.,
2018; Channell et al., 2018; Laj et al., 2000; Laj et al., 2004; Nowaczyk et al., 2013) b) atmospheric
∆
14
C records and c) bulk ocean ∆
14
C estimate. ............................................................................. 92

Figure 5.2. a) prescribed
14
Cpro in the two experiments plotted with reconstructions (grey), b)
simulated ocean DIC inventory, c) simulated atmospheric ∆
14
C and d) simulated ocean bulk ∆
14
C
plotted with reconstructions. ......................................................................................................... 96

Figure 5.3. Modern Geological a) carbon (Wallmann & Aloisi 2012) and b) ALK flux
(Middelburg et al., 2020). The flux of reverse weathering follows (Isson et al., 2020). .............. 98

Figure 5.4. Five locations where benthic ∆
14
C records document a negative excursion that is
larger than the bulk ocean ∆
14
C trend at the glacial termination. Multiple records over a range of
depth were averaged to represent a local ∆
14
C anomaly. Negative ∆
14
C excursion at these locations
occurred at different stages of the last glacial termination. Note the magnitude of local ∆
14
C
anomaly does not necessarily scale with the local flux of geological carbon input. .................... 99

Figure 5.5. Carbon and ALK sink flux due to CaCO3 (Cartapanis et al., 2018) and POC burial
(Cartapanis et al., 2016, Cartapanis et al., 2018). ....................................................................... 100

Figure 5.6. Simulated atmospheric pCO2 in experiment “Inventory” (solid purple, injection rate =
0.5 PgC/yr) versus ice core record of pCO2 (black). Also plotted is simulated atmospheric pCO2

x
in an additional experiment where the DIC:ALK injection ratio switches from 1:1.25 to 1:1 at 18ka
(dash purple). .............................................................................................................................. 101

xi
List of Tables

Table 2.1. Calibration curves used for each individual record; regression in the form of δ
11
BCaCO3
= m* δ
11
Bborate +c .......................................................................................................................... 25

Table 2.2. Scenarios to achieve atmospheric pCO2~180ppm through surface inversion in
cGENIE. ........................................................................................................................................ 28

Table 3.1. Prescribed Forcings in cGENIE Experiments ............................................................. 54

Table 3.2. δ
13
C Response in cGENIE Experiments. Δδ
13
C is defined as the difference between
model year 2000 and 0. ................................................................................................................. 67

xii

Abstract

Over the past few decades, Earth Scientists have begun to realize that Earth’s components
- its solid part (crust, mantle and core), the oceans, the atmosphere and diverse ecosystems must
be studied as a whole in order to tackle problems like global change in the past, present and future.
The holistic philosophy of this emerging Earth System Science encourages interdisciplinary
thinking and an exploration of the processes at the boundaries of the Earth’s components. Key
concepts in System Sciences involve ‘reservoir’, ‘flux’, ‘timescale’, ‘feedback’, etc. Applications
of these concepts in studies of Earth’s biogeochemical cycle have provided insights into the natural
variability within the Earth system. For my Ph.D., I focused on the carbon cycle dynamics between
warm interglacial and cold glacial climates of the late Pleistocene epoch. Of particular interest is
how carbon flux at the boundaries of Earth’s reservoirs might have changed in response to physical,
biogeochemical and geological processes. Four distinct projects that span atmosphere-ocean
carbon exchange to geological carbon cycle imbalance induced ocean carbon inventory change are
presented in this thesis.
Chapters 2-4 present the effort to improve our understanding of what happened at the ocean
and atmosphere boundary since the last glacial period. Chapters 2 & 4 are devoted to better
understanding air-sea CO2 gas exchange at the last glacial termination by using boron isotope and
stable carbon isotope reconstructions as well as numerical modeling; Chapter 3 investigates how
air-sea CO2 gas exchange could influence marine stable carbon isotope (δ
13
C) distributions. By
investigating the influence of gas exchange on the δ
13
C distribution in the ocean it provides better
constraints on interpretations of biogenic carbonate δ
13
C obtained from sediment records collected

xiii
from the upper ocean. Chapter 5 explores carbon exchange between the solid Earth and the surface
environment. Using radiocarbon constraint, I look at the glacial-interglacial carbon cycle from a
different perspective by asking the questions: has the total carbon inventory in the atmosphere-
ocean-terrestrial system remained constant on the glacial-interglacial timescales? And, could a
changing total carbon inventory in the surface environment be an important missing piece in the
glacial-interglacial atmospheric pCO2 problem?

1

1 Chapter 1
1.1 Introduction
Over the past 800 kyr the partial pressure of atmospheric CO2 (pCO2) fluctuated between
190 and 280ppm in association with the waxing and waning of large continental ice sheets (Lüthi
et al., 2008; Monnin et al., 2001; Petit et al., 1999). After more than 4 decades of research it remains
unclear what mechanisms caused atmospheric pCO2 to decrease during long, 100kyr glaciations
and then to increase more rapidly during subsequent deglaciations. It has long been recognized
that the Ocean must have played a central role in regulating atmospheric pCO2 on glacial-
interglacial timescales as the Ocean holds more than 95% of ‘active’ carbon in the atmosphere-
ocean-terrestrial system. Early studies focused on identifying oceanic processes to explain the low
atmospheric pCO2 at the last glacial maximum (LGM) between 23 to 19 ka (Sigman & Boyle,
2000). The main outcome of these studies is that no single process can account for the full 90 ppm
glacial-interglacial pCO2 variability (Kohfeld & Ridgwell, 2009). Since then, modeling efforts
have explored how ocean physical and biological processes could have worked in concert to
influence excess carbon storage in the glacial ocean and its release back to the atmosphere during
deglaciations (Brovkin et al., 2012; Hain et al., 2010; Menviel et al., 2012). Sedimentary records
containing imprints of physical, chemical and biological properties of the past ocean provide useful
constraints and have helped to evaluate potential mechanisms (Gottschalk et al., 2016; Hu &
Piotrowski, 2018; Jaccard et al., 2016; Yu et al., 2013).

2
However, the process of air-sea gas exchange has been largely overlooked in these model-
data iterative explorations. It is rather ironic that air-sea gas exchange has received so little
attention since any effect of ocean processes on atmospheric pCO2 must occur through air-sea CO2
exchange. The flux of carbon entering and leaving the ocean is directly relevant to atmospheric
CO2 variability. Moreover, recognizing that air-sea gas exchange can affect interpretations of
marine records has begun to change our view of the Ocean’s role in regulating atmospheric pCO2.
For example, a more sluggish glacial ocean could have stored more respired organic carbon
exported from surface waters and thereby played a major role in lowering atmospheric pCO2
during glaciations (Menviel et al., 2017). This scenario is largely based on older radiocarbon ages
in the glacial ocean (Skinner et al., 2017). However, recent modeling studies suggest this glacial
‘ageing’ can be accounted for by reduced air-sea CO2 (and thus radiocarbon) exchange due to
extent glacial sea ice at high latitudes where most of the deep waters form (Galbraith & de
Lavergne, 2019; Khatiwala et al., 2019). In fact, the global overturning circulation rate during
glacial periods could have been similar to (Galbraith & Skinner, 2020) or even faster than (Hu &
Piotrowski, 2018) that in the modern time.
Observations also suggest carbon transfer between Earth’s surface environment and
Earth’s interior have changed over glacial-interglacial cycles (Hasenclever et al., 2017; Huybers
& Langmuir, 2009; Kölling et al., 2019; Lund et al., 2016). However, Earth System modeling
efforts to mechanistically understand carbon cycles on these timescales have not taken the
interactions between Earth’s surface and deep carbon reservoirs into account. From a geological
point of view, atmospheric pCO2 variation over multiple glacial cycles during the past 1 million
years has been considered as ‘well-bounded’ and ‘stable’. This is justified by the reasoning that
any significant imbalance in carbon flux between the surface and deep reservoirs would have

3
caused extremely high or low atmospheric pCO2 values that are not supported by the ice core
records (Zeebe & Caldeira, 2008). However, this argument would no longer be valid if there was
an associated alkalinity imbalance, a possibility has not been considered.
In summary, recognizing the important role of ocean-atmosphere CO2 exchange and
carbon flux between the surface and deep reservoir opens exciting new research directions in the
search for a complete solution to the question of glacial-interglacial atmospheric CO2 variability.
Hopefully, a better understanding of important feedbacks and processes within the Earth’s carbon
cycle will improve the fidelity of our future atmospheric CO2 and climate projections.
1.2 Outline of the Dissertation Chapters
Chapter 2
Although it is widely accepted that the Ocean was a source of carbon to the atmosphere
during the last deglaciation, direct empirical evidence that document air-sea CO2 flux has been
lacking. The Boron isotope-pH proxy (Foster & Rae, 2016; Rae, 2018) provides an opportunity to
fill this gap; past surface seawater pCO2 can be reconstructed by measuring the boron isotope
composition of planktic foraminiferal calcite plus some assumptions about carbonate chemistry
such as the past alkalinity. In this chapter, I present a first global sea surface pH/pCO2 composite
developed from 12 boron isotope records (2 new records generated in this study + 10 published
records). Although the spatial coverage of the data is still very limited, the composite suggests the
Ocean was a source of CO2 for the atmosphere over the last 25 thousand years, with the strongest
outgassing occurring during the deglaciation when the atmospheric pCO2 increased. The results
are robust under a wide range of assumptions about the surface ocean carbonate chemistry.
Individual records show complex spatial and temporal patterns of CO2 outgassing, suggesting
carbon was emitted from the ocean at various locations during the last 25 ka. Two new records are

4
presented in this study; the sites are on the Chatham Rise off the coast of New Zealand, which
today is a CO2 sink region. This site also exhibits some of the largest CO2 outgassing during the
deglaciation of the 12 compiled records. This likely reflects a release of both respired carbon and
geological sources of carbon from depth. Chatham Rise serves as an interesting example that the
source of carbon supplying deglacial oceanic CO2 outgassing may not have come from seawater
itself.
Chapter 3:
In addition to the boron isotope-pH proxy, another commonly used proxy to infer oceanic
CO2 outgassing during the last deglaciation is δ
13
C measured from planktic foraminiferal calcite.
The rationale is if the deglacial atmospheric CO2 rise involved mainly a transport of δ
13
C-depleted
carbon (i.e. remineralized organic carbon) from the deep ocean, upwelling of DIC-enriched and
δ
13
C-depleted deep water would result in lower δ
13
C and higher seawater pCO2 (leading to CO2
outgassing to the atmosphere) from the upper ocean. Therefore, lower planktic δ
13
C might be
linked to carbon release from the deep ocean. I designate this process as “bottom up δ
13
C transport”.
On the other hand, ice core records of atmospheric δ
13
C reveal distinct negative excursions during
the last deglaciation, which could imply that a source of δ
13
C-depleted carbon entered the
atmosphere and then equilibrated with the upper ocean. In this scenario the lower planktic δ
13
C
values during the deglaciation could simply be the result of ocean-atmosphere δ
13
C equilibrium
through air-sea gas exchange. I term this process “top down δ
13
C transport”. In this chapter, a
deglacial transient simulation conducted using an Earth system model of intermediate complexity
(EMIC)– LOVECLIM and additional sensitivity experiments conducted with another EMIC –
cGENIE were used to investigate these two scenarios. The results suggest the atmosphere acted as
a bridge to transmit a δ
13
C signal (a negative δ
13
C excursion) to the global upper ocean, specifically,

5
a preformed signal. In doing so, these results support a ‘top down’ rather than a ‘bottom up’
mechanism for δ
13
C transport. The results imply that planktic δ
13
C data at any particular location
cannot be used to infer deep water ventilation of carbon without additional constraints.
Chapter 4:
Southern Ocean physical and biological processes have strong controls on atmospheric
pCO2. In numerical modeling studies, prescribed deglacial forcing often leads to major CO2
outgassing in the Southern Ocean (e.g. Chapter 3). Testing and validating this with observations
will improve our understanding of how and where carbon left the ocean at the glacial terminations
as atmospheric pCO2 rose. Three boron isotope records from the northern edge of the Southern
Ocean show complex patterns and do not unambiguously support stronger CO2 outgassing from
the broad Southern Ocean during the last deglaciation. This chapter explores whether surface
ocean δ
13
C observations (planktic foraminiferal δ
13
C) are a reliable indicator of air-sea CO2 flux
from the Southern Ocean. Using numerical model simulations, winds, biological activity and deep
overturning circulation are perturbed to investigate how these processes influence the surface δ
13
C.
If in the model simulations there is a strong correlation between the δ
13
C response and these
discrete forcings, then an array of planktic δ
13
C records from a wide region of the Southern Ocean
could supplement the limited boron isotope records. The results of this investigation reveal that
surface δ
13
C correlates poorly with air-sea CO2 flux at a local scale. On the other hand, both
observational data and model simulations reveal a regional-scale δ
13
C depletion in the sub-
Antarctic Southern Ocean that is consistent with enhanced deglacial CO2 outgassing. Overall, I
conclude that δ
13
C records from planktic foraminifera provide only minor constraint on past air-
sea CO2 exchange and thus, the boron isotope proxy remains the most valuable approach to studies
of Southern Ocean CO2 flux.

6
Chapter 5：
It is generally assumed that ocean inventory of carbon and alkalinity (ALK) do not vary
significantly during glacial/interglacial cycles. In this chapter, I present a compiled bulk ocean
radiocarbon record that reveals a 15-20% decrease in seawater
14
C/
12
C during the last glacial
termination. Numerical simulations conducted with the EMIC-cGENIE suggest this can only be
explained by a ~20% decrease in
14
C production rate in the atmosphere or, an increase in the
Ocean’s dissolved inorganic carbon (DIC) inventory by more than 5000 GtC. The five existing
observation-based
14
C production rate reconstructions exhibit wide variability and there is no clear
trend among the records that would account for the deglacial decline in oceanic
14
C/
12
C. At the
same time, multiple lines of evidence suggest that both the sinks and the sources of the geological
carbon and alkalinity (ALK) have varied over this time period and perhaps significantly enough
to have altered the ocean DIC and ALK inventory by 15-20%. Based on these results, I propose a
novel mechanism to partially explain the atmospheric pCO2 variability over glacial cycles –
varying geological DIC:ALK fluxes into the ocean. This result has potentially profound
implications for carbon cycle studies and will require a concerted effort to test. But the potential
implications justify this effort.

7

2 Chapter 2
Abstract
Identifying processes within the Earth System that have modulated atmospheric pCO2
during each glacial cycle of the late Pleistocene stands as one of the grand challenges in climate
science. The growing array of surface ocean pH estimates from the boron isotope proxy across the
last glacial termination may reveal regions of the ocean that influenced the timing and magnitude
of pCO2 rise. Here we present two new boron isotope records from the subtropical-subpolar
transition zone of the Southwest Pacific that span the last 20 kyr, as well as new radiocarbon data
from the same cores. The new data suggest this region was a source of carbon to the atmosphere
rather than a moderate sink as it is today. Significantly higher outgassing is observed between
~16.5-14 kyrBP, associated with increasing δ
13
C and [CO3]
2-
at depth, suggesting loss of carbon
from the intermediate ocean to the atmosphere. We use these new boron isotope records together
with existing records to build a composite pH/pCO2 curve for the surface oceans. pH
disequilibrium/CO 2 outgassing is widespread throughout the last deglaciation, likely explained by
upwelling of CO 2 from the deep/intermediate ocean. During the Holocene, a smaller outgassing
peak is observed at a time of relatively stable atmospheric CO 2, which may be explained by
regrowth of the terrestrial biosphere countering ocean CO 2 release. Our stack is likely biased
toward upwelling/CO2 source regions. Nevertheless, the composite pCO2 curve provides robust
evidence that various parts of the ocean were releasing CO2 to the atmosphere over the last 25 kyr.

8

2.1 Introduction
Growth and retreat of ice sheets in the late Pleistocene was accompanied by variations in
atmospheric pCO2 from ~180 ppm during the glacial maxima to 280 ppm during peak preindustrial
interglacials (Petit et al., 1999). Most explanations for this CO 2 change have focused on a more
efficient polar-ocean biological pump, driven by a combination of increased stratification (Basak
et al., 2018; Ferrari et al., 2014; Francois et al., 1997) and stronger iron fertilization in the Southern
Ocean (Martinez-Garcia et al., 2014). Reduced hydrothermal flux of CO2 has also recently been
suggested to play a role (Lund et al., 2016; Stott & Timmermann, 2011). However, the relative
importance and timing of the mechanisms that contributed to glacial CO2 change is yet to be fully
resolved.
To improve constraints on the role of ocean-atmosphere CO2 exchange between the ocean
and the atmosphere in glacial-interglacial changes in atmospheric pCO2, several recent studies
have examined when and where CO2 was entering and leaving the ocean using sea surface
carbonate chemistry reconstructions. Changes in surface CO 2 chemistry may be driven by changes
in upwelling or stratification, and changes in the biological soft tissue and CaCO3 pumps. In turn
these influence CO2 gas exchange, which is determined by the CO2 partial pressure gradient
between the surface ocean and the atmosphere (∆pCO2) and the gas exchange coefficient, which
is affected by wind speed. Tracking past air-sea CO2 exchange may help test hypotheses that have
been put forward to explain glacial/interglacial atmospheric pCO2 variability.

9
The boron isotope composition (δ
11
B) of foraminifera is a proxy for oceanic pH (Foster &
Rae, 2016; Hemming & Hanson, 1992; Hönisch & Hemming, 2005). This proxy is based on the
pH-dependency of boron speciation in seawater between boric acid (B(OH)3) and borate ion
(B(OH)4
-
) (Dickson, 1990). At low pH (< ~8.6), seawater boron primarily exists in the form of
boric acid, whereas at high pH (> ~8.6), seawater boron is dominantly present in the form of borate.
Because there is a constant equilibrium fractionation of 27.2‰ associated with this speciation
(Klochko et al., 2006), the boron isotopic composition of each species is also pH dependent. A
positive correlation between the boron isotope composition of planktic foraminifera and pH has
been established through culturing experiments (Henehan et al., 2013; Sanyal et al., 1996, 2001).
However, the δ
11
B in calcite from various foraminiferal species deviates from that of the borate in
seawater to varying degrees. This may arise from biological processes, including respiration and
photosynthesis that are collectively referred to as “vital effects” (Hönisch et al., 2003; Rae, 2018;
Zeebe et al., 2003). As a result, species-specific calibrations have been developed from culturing
experiments, sediment core-tops, sediment traps and plankton tows (Henehan et al., 2013, 2016;
Sanyal et al., 1996).
By taking advantage of the calibrated boron-pH proxy, a number of studies have
investigated the history of ocean-atmosphere CO2 exchange across the last deglaciation from sites
in the tropical Pacific (Martínez-Botí et al., 2015; Palmer & Pearson, 2003), the North Pacific
(Gray et al., 2018a), the North Indian Ocean (Naik et al., 2015; Palmer et al., 2010), the tropical
Atlantic (Foster, 2008; Foster & Sexton, 2014; Henehan et al 2013), the North Atlantic (Ezat et al.,
2017; Yu et al., 2013) and the South Atlantic (Martínez-Botí et al., 2015) (Figure 2.1a).
Collectively, the available records reveal millennial-scale δ
11
B variations that reflect substantial
changes in air-sea ∆pCO2 over the deglaciation (Figure 2.2). These variations have been

10
interpreted to reflect local or regional changes in upwelling intensity, the CO2 content of
subsurface waters, and the efficiency of the biological pump. While each of these proposed
mechanisms can be used to explain the individual δ
11
B records, the collective influence of these
changes on air-sea ∆pCO2 has not been evaluated. Here, we consider the implication of the
combined reconstructed anomalies on the evolution of atmospheric pCO2.

Figure 2.1. a) Map of mean annual ∆pCO2 sea-air. Oceanic sinks and sources are presented by
negative and positive ∆pCO2 sea-air values (Takahashi et al., 2009), respectively. The green

11
triangle shows the location of the studied sediment cores PC75-2 and PC83-1. The Circles show
the locations of the cores with published boron isotope data that span the last deglaciation. b)
Salinity of the P15 transect from the World Ocean Circulation Experiment. The yellow star shows
the location of corals (Hines et al., 2015); the black circles show the locations of other sediment
cores (Allen et al., 2015; Ronge et al., 2016) discussed in this study. The red-dotted line marks the
boundary between Antarctic Intermediate Water (AAIW) and upper circumpolar Deep Water
(UCDW).

Figure 2.2. 12 published boron records span the last deglaciation grouped into 4 species. (a)
δ
11
B measured in T. sacculifer, ODP1238 (red) (Martínez-Botí et al., 2015), EDRC92 (brown)
(Palmer & Pearson, 2003)and NIOP464 (pink) (Palmer et al., 2010). (b) δ
11
B measured in N.
pachyderma, RAPiD-15-4p (yellow) (Yu et al., 2013), MD01-2416 (blue) (Gray et al., 2018a)
and JM11-PI-19PC (dark purple) (Ezat et al., 2017). (c) δ
11
B measured in G. ruber, GeoB1105
(orange) (Foster & Sexton, 2014), ODP999 (light purple) (Foster, 2008), GeoB 1523 (dark
green) (Henehan et al., 2013) and AA2.16 (gray) (Naik et al., 2015) . (d) δ
11
B measured in G.
bulloides (magenta) (Martínez-Botí et al., 2015). Records generated on a MC-ICPMS are
marked with circles and records generated on a NTIMS are marked with diamonds. circles and
record

12
As the number of boron isotope records has expanded to various parts of the ocean, it may
now be possible to evaluate how the records reflect global as well as local influences. However,
the subtropical and subpolar South Pacific remain poorly constrained (Figure 2.1a). Hence, a
global picture of the air-sea exchange of CO2 is far from complete without additional
reconstructions from these areas. To address this, we present two new boron isotope
reconstructions from cores recovered from Chatham Rise in the subtropical-subpolar transition
zone of the SW Pacific. Although located in a moderate CO2 sink region today, this site may also
be influenced by changes in ventilation and biological pump efficiency in the Southern Ocean
(Allen et al., 2015; McCave et al., 2008; Studer et al., 2015), migration of the subtropical front
(Bostock et al., 2015), and/or localized inputs of radiocarbon-dead carbon (Ronge et al., 2016) on
glacial-interglacial timescales. We also generate new benthic-planktic radiocarbon data to trace
how reconstructed changes in CO2 outgassing relate to changes in ventilation at depth. We
combine our new boron isotope records with available records to build an initial composite sea
surface pH reconstruction that spans the last 25 kyr and use this composite to assess whether the
available data provide a realistic representation of global surface ocean surface CO2 exchange. Our
null hypothesis is that a composite pH/pCO2 record that includes both source and sink regions
constitutes a first-order representation of the equilibrium exchange of CO2 between the surface
ocean and atmosphere over the last 25 kyr. The null hypothesis predicts: First, during times of
relatively stable atmospheric CO2, the composite ocean pCO2 stack should be roughly in
equilibrium with the atmosphere; second, during intervals when the oceans released CO2 to the
atmosphere, the composite pCO2 should be higher than the contemporary atmospheric pCO2, with
larger offsets occurring when atmospheric pCO2 rose rapidly. Alternatively, the available records
may be biased because of their locations, such that the regional ocean/atmosphere dynamics or

13
extra carbon input from a local source overshadow the history of equilibrium exchange between
the ocean and atmosphere. We test the null hypothesis in both the pH and the pCO2 space. The
boron isotope pH composite is evaluated against an independent estimation of the global mean
equilibrium sea surface pH (GMSSpH), obtained by forcing the Earth System model cGENIE
(Ridgwell et al., 2007; Cao et al., 2009) with the ice core record of atmospheric pCO2 over the last
25 kyr (Bereiter et al., 2015; Monnin et al., 2001). In the model simulation, atmospheric pCO2 data
from ice cores are treated as a constraint instead of a problem to solve. We also compare the boron
isotope-derived pCO2 composite directly with the atmospheric pCO2 record.
2.2 Methods:
2.2.1 Oceanographic Settings for Cores PC75-2 and PC83-1
In the SW Pacific today, the subtropical gyre between 20°S to 45°S is a carbon sink, while
the surface waters south of the subtropical front (STF) between 45°S to 55°S are in near
equilibrium with the atmosphere (Figure 2.1a). Cores PC75-2 (177°8.97’ E, 44°14, 39’ S; 967m
water depth) and PC83-1 (177°2,49’ E, 44°18,38’ S; 1010m water depth) were retrieved by the
RV Sonne from the southern flank of the Chatham Rise east of New Zealand. The core locations
are close to the modern position of the Subtropical Front (STF), which separates cold, less salty
sub-Antarctic waters from warm, more saline subtropical waters (Coffin et al., 2013), and are
bathed by Antarctic Intermediate Water (AAIW). Our core sites are located at the south edge of
the subtropical CO2 sink, with an annual mean modern ΔpCO2= ~ -20ppm and a modest seasonal
cycle of ~30 ppm (Figure 2.1b, Figure 2.3). Direct comparison between our paleo-reconstruction
and modern ΔpCO2 at the core site might be complicated by the following three factors (Figure
2.3): 1) our site is potentially sensitive to front shifts during the last deglaciation: everything else
being equal, a northward shift of the STF during the Last Glacial Maximum (LGM) and early

14
deglaciation (Bostock et al., 2015) would turn our core site into a weaker sink or even a minor
source; 2) a sediment trap study suggests that flux of Globigerina bulloides is much higher near
our site between September to February, when the CO2 sink is weaker than rest of the year (Jonkers
& Kučera, 2015); 3) strictly speaking, modern ΔpCO2 may not be under a steady state (e.g.
Holocene) due to anthropogenic CO2 emissions.

Figure 2.3. Meridional sections of ∆pCO2 and CO2 flux reanalysis data at at 177°E (year 2008-
2012) from biogeochemical-sea ice-ocean state estimate (B-SOSE) as part of the SOCCOM
project (Verdy & Mazloff, 2017). The black (red) lines are 5-year ‘climatology’ of ∆pCO2 and CO2
flux between September to February (March to August)

15
2.2.2 Age Model for PC75-2 and PC83-1
Radiocarbon Measurements:
For reconstruction of radiocarbon activities, mixed planktic and benthic foraminifera from
PC75-1; Globorotalia inflata (G. inflata), G. bulloides and mixed benthic foraminifera from PC75-
2; G. inflata from PC83-1 were picked. For PC75-1, radiocarbon was measured at the Rafter
Radiocarbon Lab at the GNS Science National Isotope Centre. For PC75-2 and PC83-1,
14
C dating
was carried out at the Keck Carbon Cycle Accelerator Mass Spectrometry Laboratory at University
of California, Irvine. Samples were leached in 10% dilute HCL immediately prior to hydrolysis.
Stable Isotope Measurements:
Approximately 20-30 G. inflata were picked from PC75-2 and PC83-1 at each interval for
δ
18
O and δ
13
C measurements. These were analyzed on a dual inlet VG Micromass Isoprime stable
isotope ratio mass spectrometer equipped with an autocarbonate system at University of Southern
California. The precision of in-house calcite standards run in conjunction with the foraminiferal
samples averaged ~0.15‰ for δ
18
O and ~0.06‰ for δ
13
C (2SD), over the course of the present
study.
PC75-1 G. inflata and Cibicides δ
18
O and δ
13
C were analyzed by a Finnigan MAT 252
mass spectrometer at National Institute of Water and Atmospheric Research’s (NIWA) stable
isotope lab. Concurrently-run carbonate standards (NBS-19) had an internal precision of 0.08‰
for δ
18
O and ~0.04‰ for δ
13
C (2SD).
Age model:
To develop an age model for the PC75 core location, we combined 6 new AMS
14
C dates
measured on calcite tests of the planktic foraminiferal species G. inflata and G. bulloides from

16
PC75-2 and 4 AMS
14
C dates from mixed planktic foraminifera from the co-located core PC75-1
core. These two cores were taken from the same location but for different purposes. There was not
enough material left from the PC75-1 core for both boron isotope analysis and radiometric dating.
Therefore, samples from PC75-2 were used. The two cores have almost identical stratigraphies
(Figure 2.4).

Figure 2.4. left panel: PC75-1 and PC75-2 raw planktic radiocarbon dates versus depth; right
panel: PC75-1 and PC75-2 G. inflata oxygen isotope versus depth

Each planktic
14
C age was converted to calendar years using the BChron Bayesian
chronology package (Haslett & Parnell, 2008), with the marine calibration dataset – MARINE13
(Reimer et al., 2013). This dating technique requires some knowledge of surface reservoir ages.
Surface reservoir ages of the SW Pacific Ocean at various times during the last 25 kyr has been
constrained in previous studies by pairing radiocarbon measurements of marine carbonates
deposited above and below volcanic tephra layers. The chronologic ages of the tephra have been
constrained by radiocarbon dating terrestrial organic material from land-based deposits that
contain the same tephra (Sikes et al., 2000; Sikes & Guilderson, 2016; Skinner et al., 2015). By

17
applying the up-to-date global calibration curve IntCal 13 (Reimer et al., 2013), Sikes and
Guilderson (2016) concluded that surface reservoir ages during the late deglaciation and early
Holocene were not much different from modern values; we therefore applied a reservoir age of
400±100 yrs for each radiocarbon date from the core top to 110 cm, the oldest late deglacial
interval where radiocarbon dates are available. For the last glacial maximum and early deglacial
intervals, reservoir ages at the SW Pacific have been determined to be ~1100-1500 yrs, by using
tephra as independent age constraints (Sikes & Guilderson, 2016; Skinner et al., 2015); we
therefore applied reservoir ages of ~1300±200 yrs to the late glacial and early deglacial samples.
However, surface reservoir ages in the SW Pacific during those intervals could be spatially variable
(Sikes & Guilderson, 2016). To provide further age constraints, we utilized two chronologic
datums: 1) a tephra layer (‘Kawakawa’) identified at 346cm in core PC75-1 with a calendar age of
25,650±40 yr BP (Sikes & Guilderson, 2016); 2) the beginning of the deglacial transition in the
benthic δ
18
O isotope stratigraphy of PC75-1, which occurs at 165cm. The regional deglacial
transition at intermediate water depths of the SW Pacific has been dated at ~16.2-16.9 kyrBP (Stern
& Lisiecki, 2014). We therefore, assigned an age of 16.5±0.15 kyrBP to the 165cm horizon. The
calibrated ages from radiocarbon measurements and the two extra age control points (tephra layer
and δ
18
O transition) were used as tie points to develop the age model for PC75-2.
The age model for PC83-1 is based on 3 AMS
14
C dates, and the reservoir ages are same
as those applied to the PC75 record since the two sites are very close to each other. An additional
age constraint comes from aligning the stable isotope stratigraphy of PC75-2 and PC83-1 (Figure
2.5). For the intervals where there are no planktic
14
C dates or other age control, a linear
interpolation has been used to estimate ages. We note that ages in PC83 are relatively uncertain,

18
given the low resolution of our age control points, and this may be a source of uncertainty when
comparing deglacial δ
11
B data from these different sites.

Figure 2.5. PC75-2 and PC83-1 G. inflata stable oxygen and carbon isotope. Based on the planktic stable
isotope stratigraphy, we aligned the age of 88cm of PC83-1 to 88cm of PC75-2, the age of which is 11850
yrBP (the vertical black lines).

19
2.2.3 Trace Element and δ
11
B Analyses
Approximately 100-200 tests of G. bulloides were picked from the 250-350 μm size
fraction for trace element and δ
11
B analyses. Sample preparations were carried out in a low-boron
clean lab at the University of St Andrews. Samples were cleaned based on the “Mg-cleaning”
oxidation procedure (Barker et al., 2003; Rae et al., 2011). An aliquot (~3% of the total sample)
was taken for trace element analyses, which were performed on an Agilent 7500 ICP-MS at the
University of St Andrews using matrix-matched standards. Long-term reproducibility of Mg/Ca
and B/Ca using this method is 1.2% and 2.3% (2 SD), respectively. Boron was separated from the
sample matrix using Amberlite IRA-743 boron specific anion exchange resin, following the
protocols of Foster (2008) and Foster et al., (2013). Δ
11
B was measured on a Thermo Scientific
Neptune MC-ICP-MS at the University of St Andrews, based on protocols described in Foster
(2008), Rae et al., (2011), and Rae et al., (2018), but with the addition of high ohmage (10
13
W)
resistors and triplicate sample analyses. Samples were corrected for total procedural blank, which
averaged 33 pg in this session. Rae et al. (2011) previously reported uncertainties of ±0.23 ‰ (2
SD) for samples of ~20 ng, increasing at smaller sample sizes. This has been improved given the
developments in analyses described above. For example a boric acid standard run during these
sessions at the same concentration and under the same conditions as these samples (15 ppb) gave
δ
11
B = 19.59 ±0.14 ‰ (2SD, n = 8); and replicate purifications and measurements of a dissolved
carbonate standard with a composition mimicking planktic foraminifera, and run under these
conditions at a concentration of 7 ppb, gave 2SD of 0.20 ‰ (n = 12). However as full description
and quantification of this updated method is beyond the scope of the current study, and as the
signals found here are large compared to analytical uncertainty, we assign conservative analytical
uncertainties following the relationships in Rae et al. (2011). Prior to δ
11
B analysis, samples were

20
screened for potential contamination by checking various elemental ratios (B/Ca, Mg/Ca and
Al/Ca). A few samples had elevated Al/Ca ratios (up to about 200-300 μmol/mol) but showed no
correlation with either δ
11
B or Mg/Ca (Figure 2.6).

Figure 2.6. Mg/Ca and boron isotope data plotted with Al/Ca.
2.2.4 Salinity estimate
To estimate sea surface salinity (SSS) changes some authors have previously combined
Mg/Ca and δ
18
O of planktic foraminifera under the assumption that the amount of δ
18
O change

21
that cannot be accounted for by the Mg/Ca-derived temperature change must reflect a change in
the oxygen isotopic composition of sea water, which is affected by salinity (Ezat et al., 2017; Naik
et al., 2015). However, this approach assumes a constant relationship between salinity and the
δ
18
O of seawater; this assumption is unlikely to hold over G-IG timescales and can therefore result
in large biases within the resulting salinity reconstruction. Given modelled regional changes in
salinity are relatively small throughout most of the ocean during the LGM (Gray and Evans, 2019),
we followed the approach outlined in Gray and Evans (2019) and calculate salinity as the modern
salinity at our site plus the modelled mean surface ocean glacial salinity increase (0.7 PSU) scaled
to global sea level (using the curve of Spratt & Lisiecki 2016), with an uncertainty of ±1 PSU
(2SD). For any individual core, the two methods of salinity estimate would only introduce minor
differences in reconstructed pH/pCO2 values (Hönisch & Hemming, 2005). When developing a
composite, one advantage of our approach is that removal of the long-term sea level influence on
SSS, residual local SSS variations between sites are likely uncorrelated and would not
systematically impact reconstruct pH/pCO2 between sites.
2.2.5 Paired SST/pH and pCO2 estimate
pH influences the Mg/Ca of planktic foraminifera (Evans et al., 2016; Gray et al., 2018b;
Kısakürek et al., 2008; Lea et al., 1999; Russell et al., 2004). Temperature estimates from Mg/Ca
are in turn required to calculate the boric acid and carbonate system dissociation constants (e.g.
Dickson, 1990), and thus pH and pCO2. Here, we apply a recently developed algorithm (Gray &
Evans, 2019) that iteratively solves Mg/Ca and δ
11
Bborate for SST and pH, overcoming the
covariance induced between pH and temperature due to the thermal control on the carbonate
system and boric acid disassociation constants. Mg/Ca based SSTs estimated from this method are
more consistent with alkenone based SST estimates over deglaciation (Gray & Evans, 2019), thus

22
carbonate system and boric acid dissociation constant estimates (and therefore pH and pCO2) will
also be more accurate. We use the G. bulloides Mg/Ca calibration given in Gray & Evans (2019),
based on foraminifera grown in laboratory culture. Partial dissolution of planktic foraminiferal
tests lowers Mg/Ca (Regenberg et al., 2014); we note, it does not affect δ
11
B (Edgar et al., 2015).
Here, we calculate temperature downcore as the temperature anomaly from the mean Holocene (0-
10 ka) temperature, assuming the mean annual climatological temperature for the Holocene; this
approach assumes the effects of dissolution have remained ~constant through time.
The ability to derive estimates of pH from the δ
11
B of foraminiferal calcite stems from the
fact that the borate ion is the dominant species incorporated into CaCO3 (Hemming & Hanson,
1992; Rae, 2018; Rae et al., 2011). The methodology of transforming the boron isotopic
composition of foraminifera to an estimate of ocean pH has recently been reviewed by Foster and
Rae (2016) and Rae (2018). In the present study, δ
11
Bborate is derived using the species-specific
calibration for G. bulloides determined by a Monte Carlo/wild bootstrap approach (Henehan et al.,
2016) where, δ
11
Bborate = (δ
11
Bcaco3 + 3.58 ± 11.77)/1.09 ± 0.65 (Raitzsch et al., 2018), which has
a wider range of uncertainty than used by Martínez-Botí et al. (2015). The δ
11
Bborate value is then
used to estimate pH using the accurate formula from Rae (2018).
The estimate of local seawater pCO2 from the pH value requires an assumption about a
second parameter of the carbonate system, typically the local sea surface alkalinity (ALK), which
is poorly constrained for the glacial and deglacial ocean. To estimate sea surface ALK some
authors have applied a modern local SSS-ALK regression relationship, using estimates of paleo-
SSS (Ezat et al., 2017; Martínez-Botí et al., 2015; Naik et al., 2015). This assumes that most of the
variability in ALK and SSS are dominated by evaporation and precipitation and/or sea level
changes. However, SSS and ALK could be decoupled due to changes in riverine input of ALK,

23
nutrient uptake, and remineralization, and production and export of CaCO3 (Fry et al., 2015). Thus,
applying the modern local SSS-ALK regression could introduce significant bias. In a previous
study, Martínez-Botí et al., (2015) randomly varied ALK between the modern value at the site
today +125 µmol/kg (Hain et al., 2010; Toggweiler, 1999) to -25 µmol/kg with a flat distribution
given that knowledge of the secular evolution of ALK during the last deglaciation is currently
lacking. However it is possible that, even with higher whole ocean alkalinity, sea surface ALK was
not much different from modern values due to stronger ocean stratification and the deep
remineralization profile of CaCO3. We therefore randomly vary ALK between ‘modern plus 75
μmol/kg’ to ‘modern minus 75 μmol/kg’ in each pCO2 calculation. Note that this uncertainty range
is large, encompassing much of the variability of the modern ocean. Also resulting uncertainty on
pCO2 remains dominated by the uncertainty on δ
11
B-derived pH, given the close coupling of pH
and CO2 within the carbonate system (Rae 2018).
pH and seawater pCO2 for each sample horizon is calculated using the ‘seacarb’ package
in R (https://CRAN.R-project.org/package=seacarb).To fully propagate the uncertainties
associated with pH/pCO2 estimations, we ran 10,000 Monte Carlo simulations that included the
following uncertainties (2SD): analytical uncertainty on δ
11
B, salinity ± 1 PSU, Mg/Ca calibration
uncertainty, δ
11
Bcc- δ
11
Bborate calibration uncertainty.
2.2.6 Composite pH and Seawater pCO2
We compiled 10 previously published and 2 new (this study) boron-based pH/seawater
pCO2 reconstructions with millennial-scale resolution spanning the last 25 kyr. RAPiD-15-4P (Yu
et al., 2013) is not included in the composite because both Holocene and LGM intervals are missing
in this record (Figure 2.2). We recalculated pH and seawater pCO2 from each published boron
isotope value with a self-consistent framework. This includes modifying published estimations of

24
1) SST; 2) SSS; and 3) the carbonate δ
11
B to δ
11
Bborate in seawater. The SSS and SST changes for
each record are recalculated in the manner described in section 2.2.3 and 2.2.4, with species-
specific Mg/Ca calibrations from Gray & Evans (2019). Calculating seawater δ
11
Bborate from the
foraminiferal δ
11
B under a self-consistent framework is not a trivial task since the records were
generated using different machines and/or techniques. The pioneering work on the species-specific
δ
11
B calibration was conducted on negative Thermal Ionization Mass Spectrometry (N-TIMS)
(Sanyal et al., 1996, 2001). More recently, Multicollector-Inductively Coupled Plasma Mass
Spectrometry (MC-ICPMS) has been used for boron isotope analyses of marine carbonates (Foster,
2008). However, measured δ
11
B from the same planktic species (see Figure 2a for example) or
even homogenized calcite samples (Farmer et al., 2016) are lower on ICPMS than NTIMS by up
to several permil in some cases. Currently, this technique offset is not easily explained (see Farmer
et al., 2016), but may be possible to correct. Logically, the same slopes ‘m’ (i.e. pH sensitivities)
should be applied to all records using the same species, no matter which technique was used to
generate the boron isotope data. Indeed, studies suggest that the calibration slope may be
transferable between different measurement techniques (Farmer et al., 2016; Foster et al., 2013).
Therefore, the correction involves applying a constant offset to the established species-specific
calibrations.
The slope (‘m’) of the foraminiferal δ
11
Bcalcite vs δ
11
Bborate in seawater has been determined
through culturing, core-top and sediment trap data sampling using both techniques (Henehan et al.,
2013; Martínez-Botí et al., 2015; Sanyal et al., 2001). Our strategy is to apply the latest species-
specific calibration curves that include newer data and more robust statistical techniques. The
calibration curve for Globigerinoides ruber and G. bulloides comes from Raitzsch et al., (2018).
For Trilobatus sacculifer, the intercept presented in Raitzsch et al., (2018) is offset from that of

25
Henehan et al., (2016) by ~0.8‰ with essentially the same slope. This is because one culture data
point from Sanyal et al., (2001) was misplaced (i.e. one δ
11
BCaCO3 data should be 18.49‰ not
18.9‰) by Raitzsch et al., (2018) and more coretop data were included in that study. To let our
recalculated pH/pCO2 be close to original published records, we chose the calibration curve from
Henehan et al., 2016 for T. sacculifer. The calibration for Neogloboquadrina pachyderma (N.
pachyderma) we use is δ
11
Bborate = δ
11
BCaCO3 +3.38±0.72‰ (first calibrated by Yu et al., 2013,
later confirmed by Gray et al., 2018a over a wider range of pH). For the N. pachyderma JM11
record generated by NTIMS (Ezat et al., 2017), we used the original intercept of 2.053‰ (Ezat et
al., 2017, NTIMS) instead of 3.38‰ (Gray et al., 2018a, MC-ICP-MS) to account for
technique/machine offsets. Applying the T. sacculifer calibration curve to the Holocene δ
11
B of
NIOP464 (Palmer et al., 2010, NTIMS) and EDRC92 (Palmer & Pearson, 2003, NTIMS) resulted
in very different pH from the original publication, due to technique/machine offsets. Thus, we
modified the intercept such that the average Holocene pH from each record equals the pre-
industrial sea surface pH at the core site (derived from the GLODAP V2 2016b dataset).
Table 2.1. Calibration curves used for each individual record; regression in the form of
δ
11
BCaCO3 = m* δ
11
Bborate +c
Core
name
Species Slope
('m')
Intercept
('c')
Reference Calibration Technique
ODP1238 T. sacculifer 0.82 3.94 Martínez-Botí
et al., 2015
Henehan et al.,
2016
MC-ICP-MS
NIOP464 T. sacculifer 0.82 -0.66 Palmer et al.,
2010
Henehan et al.,
2016 ‘c’ modified
by this study
NTIMS
EDRC92 T. sacculifer 0.82 -1.66 Palmer &
Pearson, 2003
Henehan et al.,
2016 ‘c’ modified
by this study
NTIMS
MD01-
2416
N.
pachyderma
1 -3.38 Gray et al.,
2018a
Gray et al., 2018a MC-ICP-MS
JM11-PI-
19PC
N.
pachyderma
1 -2.053 Ezat et al.,
2017
Gray et al., 2018a,
‘c’ modified by
Ezat et al., 2017
NTIMS
ODP999 G. ruber 0.55 9.82 Foster, 2008 Raitzsch et al.,2018 MC-ICP-MS

26
GeoB1523 G. ruber 0.55 9.82 Henehan et al.,
2013
Raitzsch et al.,2018 MC-ICP-MS
GeoB1105 G. ruber 0.55 9.82 Foster &
Sexton, 2014
Raitzsch et al.,2018 MC-ICP-MS
AAS9 G. ruber 0.55 9.82 Naik et al.,
2015
Raitzsch et al.,2018 MC-ICP-MS
PS2498-1 G. bulloides 1.09 -3.58 Martínez-Botí
et al., 2015
Raitzsch et al.,2018 MC-ICP-MS
PC75-2 G. bulloides 1.09 -3.58 this study Raitzsch et al.,2018 MC-ICP-MS
PC83-1 G. bulloides 1.09 -3.58 this study Raitzsch et al.,2018 MC-ICP-MS

Table 2.1 summarizes the regression applied for each individual record. Where possible
we use reported analytical uncertainty of δ
11
B; Palmer & Pearson (2003) and Palmer et al., (2010)
did not report analytical uncertainties, so we use ±0.3‰ (2 SD). The procedure described in section
2.2.5 is then applied to calculate pH and pCO2 for each data point. We term the results as ‘pHboron
and ‘pCO2boron. When we developed the pH and pCO2 composite, we treated each ‘pHboronand
‘pCO2boron data point as a sample of the global mean sea surface pH and pCO2. However, δ
11
B-
based pH reconstruction may carry a bias relative to the true mean pH at a specific sample site
(‘reconstruction bias’) and true mean pH at each site carries a bias relative to mean global pH
(‘location bias’). For each individual recored, the ‘reconstruction bias’ is estimated by the
difference between the average pHboron over the Holocene (pHboron) and an estimate pre-industrial
pH from GLODAP at that site (pHsite_preind); the ‘location bias’ is estimated by the difference
between pHsite_preind and the global pre-industrial pH from the same GLODAP product
(‘pHglobal_preind’). The expression for the final pH data is then: pHfinal = pHboron – (pHboron_Holocene –
pHsite_preind) – (pHsite_preind – pHglobal_preind) = pHboron - pHboron_Holocene + pHglobal_preind. The uncertainty
of ‘pHboron_Holocene’ and ‘pHglobal_preind’ are then propagated into using 10,000 Monte Carlo
simulations, along with a ± 700-year age uncertainty (2 SD) for each sample age. The same
procedure is applied to pCO2: pCO2final = pCO2boron – (pCO2boron_Holocene – pCO2site_preind) –
(pCO2site_preind – pCO2global_preind) = pCO2 boron - pCO2boron_Holocene + pCO2global_preind. The final

27
pH/pCO2 data from each of the cores were then binned into 1000-year intervals, and the average
pH/pCO2 for each bin was calculated.
2.2.7 Modeling the GMSSpH with cGENIE Earth System Model
The Earth System model cGENIE was used to develop a simulated GMSSpH over the last
25 kyr to compare against the composite pH curve. The cGENIE model includes a 3-D dynamical
ocean model coupled to the 2-D energy-moisture balance atmospheric model (Edwards & Marsh,
2005). The ocean model is based on a 36x36 horizontal grid with 16 vertical layers. cGENIE has
a dynamic and thermodynamic component of sea ice. The model also incorporates a marine
biogeochemical cycling of carbon and other tracers (Ridgwell et al., 2007). All simulations in the
present study used a pre-industrial configuration (Cao et al., 2009) and were spun up as a closed
system for 20 kyr.
To estimate the GMSSpH over the last 25 kyr, an ‘LGM’ like sea surface carbonate
chemistry state is a necessary initial condition for the model. Since an investigation of carbonate
system feedbacks is not the purpose of this study, interactive sediments (i.e. open-system
configurations) are not used in the following simulation. We achieved a peak glacial atmospheric
pCO2 value through an inverse approach with three end-member scenarios. Specifically, in
scenario 1 ALK is added to the surface ocean; in scenario 2 ALK and DIC are added to the surface
ocean in a 2:1 ratio (i.e. CaCO3 addition); in scenario 3 DIC is subtracted from the surface ocean.
We performed the inversion under a modern ocean circulation for 10,000 years. The resulting
‘LGM’ atmospheric pCO2 and sea water chemistry responses at the end of the inversion are
summarized in Table 2. Since the GMSSpH is strongly coupled to the atmospheric pCO2, we
would get essentially the same answer even if we apply an ‘LGM’ like circulation scenario (not

28
shown). Then, to estimate the GMSSpH over the last 25 kyr, a transient carbon flux was added or
taken out of the atmosphere so that the atmospheric pCO2 follows the ice core record.
Table 2.2. Scenarios to achieve atmospheric pCO2~180ppm through surface inversion in
cGENIE.
Experiment name spin ALK addition CaCO3 addition DIC removal
atm pCO2
(ppm)
278 184 184 181
Surface pH
(total scale)
8.15 8.31 8.34 8.30
Surface ALK
(μmol/kg)
2268 2350 2542 2264
Surface DIC
(μmol/kg)
1953 1937 2084 1871

2.3 Results
2.3.1 Two New Boron Isotope-Based pH Reconstructions from the
SW Pacific
The new δ
11
B record from subtropical South Pacific core PC75-2 is shown in Figure 2.7a.
This core documents ~1.5‰ lower values during the early deglacial relative to the glacial and late
deglacial samples and shows a pronounced excursion to low δ
11
B values around 16.5-14 kyrBP.
The record from core PC83-1 spans the last 16 kyr (Figure 2.7e). In this core, the δ
11
B values in
both the early deglacial section and the Holocene section are close to 15-15.5‰, while values from
the late deglacial section and the core top are characterized by values of ~16‰. Once converted to
pH (Figure 2.7c, g) these records demonstrate the surface/shallow subsurface pH in the SW Pacific
was ~8.2 during the LGM and early deglaciation, before being punctuated by low pH values
between ~16.5-14 kyrBP (Figure 2.7c). pH values are around ~8.1 in the late deglaciation and
early Holocene and rise to 8.2 at the core top (~2.3 kyrBP). The pCO2 reconstruction at our site

29
suggests on Chatham Rise, sea surface pCO2 were up to ~300 to 450 μatm during the last
deglaciation and early Holocene.

Figure 2.7. Deglacial changes in seawater carbonate chemistry and sea surface temperature of
Chatham Rise. a-d, PC75-2; e-h, PC83-1 (a, e) G. bulloides d
11
B (versus boric acid standard NIST
951) with analytical uncertainties (2 SD). (b, f) G. bulloides Mg/Ca-based SST with uncertainties
(1 SD) (c, g) d
11
B-based pH. (d, h) d
11
B-derived pCO2. The dotted envelopes represent uncertainty
of sea surface pH and pCO2 (1 SD). Triangle marks represent intervals with radiocarbon age
constraints.

2.3.2 pH and Seawater pCO2 Composite vs the Simulated GMSSpH
Based on estimates of pHboron_Holocene, pHsite_preind, pCO2boron_Holocene and pCO2site_preind for 12
individual records, the ‘location bias’ and the ‘reconstruction bias’ are corrected. Individual pH

30
records are shown against the composite pH in Figure 2.8. ODP999 closely follows the mean,
while in most other cases, an individual record generally fluctuates around the composite, with a
few data points falling outside of the 95% envelop of the composite. Some recognizable patterns
are: 1) During the LGM and early deglaciation (25-15ka), the ODP1238 record consistently shows
higher pH, while the GeoB1105 record shows lower pH than the composite; 2) AA2.16, NIOP464,
PC75, EDRC92 and MD01-2416 all show anomalously low pH values between 15-14ka; 3)
During the late deglaciation (15-10 ka), although some data points from GeoB1523, GeoB1105
and PC83 are close to the composite, other pH values are higher than the composite mean. We note
that the above description is based on corrected pH values (i.e. pHfinal) in this study, which should
not be confused with conclusions drew from uncorrected pH values. For comparison, individual
uncorrected pH records (i.e. pHboron) are plotted with the composite in Figure 2.9.

31

Figure 2.8. Individual pH reconstructions (corrected for ‘location bias’, i.e. pHfinal = pHboron -
pHboron_Holocene + pHglobal_preind) plotted against the composite pH.

32

Figure 2.9. Individual uncorrected pH reconstructions (i.e. pHboron) plotted against the composite
pH built upon pHfinal. The red dots represent pHsite_preind calculated from GLODAP V2 2016b
(Lauvset et al., 2016).

The composites are developed from 12 cores that come from a wide range of oceanographic
settings. At sites where the carbonate system is strongly affected by seasonality and the fluxes of
planktic foraminiferal tests have distinct seasonal patterns, the pH and thus pCO2 estimates will
likely be biased towards the season of maximum production. Also, since foraminifers calcify over

33
a range of depths instead of strictly at the surface, the composite may be biased towards higher
pCO2 because sea-water pCO2 increases with depth within the top few hundreds of meters (Ezat
et al., 2017; Raitzsch et al., 2018; Taylor et al., 2018, Yu et al., 2013). However, we found that
these complexities would only have minor effects on our composite pCO2 (Figure 2.10, 2.11),
probably because the calibrations that are largely based on core top measurements that already
account for some of these effects.

Figure 2.10. The climatology of seasonal and annual mean surface pCO2 (year 2001-2015) (Landschützer
et al., 2017) for each core site. The data is averaged over a 10° by 10° box around the core site. The black
bars represent 1 standard error over the 15-year time period. Among these cores, JM11, Rapid 15 and
MD01-2416 from high latitude oceans are characterized with strong seasonality. Rapid 15 is not included
in our composite because it does not cover the LGM. For JM11, sediment trap data (Jonkers & Kučera,
2015) near the Norwegian Sea suggest that these is no N. pachyderma flux during winter, while for the rest
of the year, the N. pachyderma flux is evenly distributed. Thus the geochemical signature of N. pachyderma
would be an underestimate of the surface pCO2. For MD01-2416, sediment trap data (Jonkers & Kučera,
2015) in the modern northwest Pacific display two abundance peaks during the spring (with high pCO2)
and autumn (with low pCO2), the two seasonal fluxes are roughly equal and the average pCO2 during the
two seasons are close to the annual mean.

34

Figure 2.11. pCO2 gradient between surface and the average of the first 50m at the core sites, the
station data from GLODAP V2 (Olsen et al., 2016) is averaged over a 15° by 15°box around the
core sites. For most of the cores in our composite, the depth difference is minimal (i.e. even though
the foraminifera may not calcify not strictly at the surface, but the seawater carbonate chemistry
state they record may still represent the surface condition). At the site of NIOP464, the surface-
subsurface difference is up to 50 ppm on average, however the station data were all taken in a
single cruise in Aug 1995, thus it is not clear of those station data are truly representative to the
long term condition. The site of ODP1238 is affected by El-Niño, the station data were taken
between year 1992-1994 when El-Niño was in the positive phase (i.e. the eastern equatorial
Pacific is more stratified). Thus the 100~150 ppm of surface-subsurface ΔpCO2 should be taken
as a maximum.

The pH and pCO2 composite built on 219 boron isotope data from 12 cores sites are shown
in Figure 2.12a and 2.12b. Three scenarios of simulated GMSSpH over the last 25 kyr are
presented in Figure 2.12c; these curves follow the same deglacial structure, due to the close
coupling between pCO2 and surface ocean pH, but have different absolute values, due to their
different alkalinity and DIC (Table 2). For a given pCO2, DIC removal results in slightly lower
pH than alkalinity addition, while addition of alkalinity and DIC in a 2:1 ratio (CaCO3 addition)
gives higher pH with a larger offset, due to the larger change in ALK-DIC chemistry in this
scenario. Our three scenarios provide different end member GMSSpH solutions for glacial pCO2,

35
and the real sea surface carbonate chemistry condition at the LGM and thus the real deglacial
GMSSpH evolution likely falls in the range of the three curves in Figure 2.12c. The
11
B-derived
pH and pCO2 composites match the overall trend of the GMSSpH simulated by cGENIE and the
atmospheric record of pCO2, respectively (Figure 2.12b). However during the last deglaciation,
the composite pH is generally lower than all three simulated GMSSpH, and, correspondingly, the
composite pCO2 is generally higher than the contemporary atmosphere. pCO2 disequilibrium in
the deglaciation is quite robust, given that this interval has the best data coverage, and the lower
2.5% quantile also falls above the atmospheric pCO2 record (Figure 2.12b).

Figure 2.12. composite pH and pCO2. a) The solid blue line denotes the median of the 10,000
simulations of the composite pH values. The dotted blue line denotes the 95% quantiles of the
10,000 simulations of the composite pH values. The grey dots represent individual boron isotope
derived pH (n=219) after corrections (i.e. pHfinal). b) composite seawater pCO2 versus the ice core
record of atmospheric CO2. The solid blue line denotes the median of the 10,000 composite pCO2.
The dot blue line denotes the 95% quantile of the 10,000 composite pCO2. The solid black line
represents the ice core record of atmospheric pCO2. c) composite pH versus cGENIE inverted pH.
The blue lines are as in panel a); The solid brown, red, and magenta lines denote the cGENIE
simulated GMSSpH evolution where the initial conditions are achieved through ‘ALK addition’,
‘CaCO3 addition’ and ‘DIC removal’, respectively.
5 10 15 20 25
7.95
8
8.05
8.1
8.15
8.2
8.25
8.3
8.35
8.4
8.45
a a
5 10 15 20 25
7.95
8
8.05
8.1
8.15
8.2
8.25
8.3
8.35
8.4
8.45
c c
5 10 15 20 25
150
200
250
300
350
b b

36
2.4. Discussion
2.4.1 Chatham Rise as a Source of Carbon to the Atmosphere over
the Last 19 kyr.
Results presented in this study indicate that the subtropical-subpolar transition zone of the
SW Pacific was a site of carbon ventilation during the last deglaciation (Figure 2.13a), whereas in
the modern ocean this region is a modest carbon sink (Figure 2.1a, Figure 2.3). A similar result
was observed at a South Atlantic site (Martínez-Botí et al., 2015) that is also a sink region in the
modern ocean. Together, these results suggest that both the Atlantic and Pacific sectors of the
Southern Ocean became sources of CO2 to the atmosphere during the deglaciation.

37

38
Figure 2.13. a) PC75-2 and PC83-1 ∆pCO2 plotted with an opal flux record from the SW Pacific
sector of the Southern Ocean (Chase et al., 2003). The dotted envelopes represent uncertainty of
∆pCO2 (the 14% and 86% quantile). The red arrow marks the modern annual ∆pCO2 at our site
(Verdy & Mazloff 2017) b) JPC83 benthic ∆[CO3]
2-
from the Bay of Plenty (Allen et al., 2015),
plotted with PC75-1 (this study) benthic δ
13
C records. The raw data are plotted as dots; the thick
lines are 3-point running averages. c) Three records of benthic - atmospheric ∆
14
C offset (∆∆
14
C)
from Chatham Rise. PC75-1 and PC75-2 radiocarbon data are combined together but are labeled
with different color coding. d) Atmospheric pCO2 (Bereiter et al., 2015). The 16.5–14 kyr BP event
is highlighted in yellow for all panels.
We acknowledge that our cores are of low resolution, therefore the two records may not
resolve the complete history of surface ΔpCO2 in this region. We note that for many time intervals
we do not have overlapping data from both cores, thus it is not possible to rule out that the offsets
seen between the two cores may represent higher-frequency variability than is captured by the
resolution of each core. There are also mismatches in reconstructed pH/ΔpCO2 at apparently
overlapping time intervals between the two records, but due to age model uncertainties (note there
are only 3 radiocarbon dates to constrain the age model of PC83-1, see the magenta triangle
markers in Figure 2.7) and low resolution it is possible that these intervals of apparent mismatch
do not truly represent overlapping time intervals, as evident by some notable differences in Mg/Ca
based SST between these records too. With the above caveats in mind, we now discuss the main
signals in the boron isotope data in context with other paleoceanographic proxies in this region.
Possible geological carbon influence on our site on Chatham Rise
Two cores from Chatham Rise have documented extreme negative Δ
14
C values between
2000-3000m (Ronge et al., 2016). Ronge et al., (2016) hypothesized that their sites were recording
the influence of radiocarbon-dead hydrothermal CO2. However, there are no known hydrothermal
systems near Chatham Rise and it is unlikely that such large Δ
14
C anomalies from volcanic
degassing of CO2 from the East Pacific Rise would reach Chatham Rise without losing their highly
14
C-depleted radiocarbon signature (Ronge et al., 2016).

39
Large pockmarks have been identified across the southern edge of the Chatham Rise in
seismic sections (Davy et al., 2010), close to the cores that document large negative ∆∆
14
C
anomalies (this study and Ronge et al., 2015). Davy et al., (2010) previously hypothesized that the
pockmarks formed at glacial terminations during the Pleistocene and argued that they came about
in response to destabilization of methane hydrates. A subsequent research cruise was conducted in
2013 to explore for methane hydrates but no data support current or past presence of methane in
sediment (Coffin et al., 2013), which suggests the pockmarks formed in response to release of
another source geologic source of CO2 rather than CH4 (Stott et al., 2019). It appears from the
seismic profiles that there have been recurrent episodes of geologic carbon release through these
pockmarks on the Chatham Rise during earlier glacial cycles. Our PC75 core was taken in close
proximity to one of the large pockmarks as was the core studied by Ronge et al., (2016).
Today, our PC75 core site is bathed by well-ventilated, relatively ‘young’ AAIW. However,
between 25-16 kyrBP, the benthic radiocarbon values were much more depleted compared to the
contemporaneous atmosphere (∆∆
14
C of ~300-400‰; Figure 2.13c). This atmosphere-benthic
∆
14
C offset is much larger than values from corals collected at slightly deeper water depths in the
Tasman Sea (Hines et al., 2015) (regardless of our assumptions about surface reservoir ages, see
Figure 2.14). These data, together with the results of Ronge et al., (2016), indicate there were local
sources of much older carbon at the Chatham Rise between 25-16 kyrBP.

40

Figure 2.14. PC75 benthic - atmospheric ∆
14
C offset (∆∆
14
C) and Tasman Sea coral (43°S-47°S 144°E-
152°E) – atmospheric ∆∆
14
C (Hines et al., 2015). The dotted blue line assumes a constant reservoir age
(~ 400 yrs) over the entire PC75 record.

Our boron isotope data only extends to 20 kyr. Over this period ΔpCO2 values range
between 0-80ppm in the PC75-2 record. It is possible that some of the old carbon that was
responsible for the benthic ∆
14
C excursions reached surface waters, which could account for the
elevated surface ocean pCO2 and the anomalously old surface reservoir ages (Sikes & Guilderson,
2016; Skinner et al., 2015) (see a conceptual diagram for the LGM, Figure 2.15b). This idea will
require additional evaluation but the fact that the pockmarks, the ∆
14
C anomalies, including those
documented by Ronge et al., (2016), and elevated pCO2 values coincided suggest they may have
been mechanistically linked. And the fact that the atmosphere-benthic ∆∆
14
C values remained low
for several thousand years further suggests there was a substantial and relatively continuous supply
of old carbon and perhaps reduced ventilation between 900-2500m of the SW Pacific that
prolonged the ∆∆
14
C anomaly during the early deglaciation.

41

Figure 2.15. Conceptual diagrams of SW Pacific circulation and radiocarbon distribution. (a)
Modern pattern. Site PC75-2 and site PC83-1 are bathed under well ventilated AAIW, with no
influence of geological carbon. The color shading is based on interpolating a transect of ∆
14
C
data between 170°E to 180°E from the GLODAP database to a 10° in latitude by 500m in depth
grid. (b) LGM pattern. Site PC75-2 and site PC83-1 are bathed under poorly ventilated, respired
carbon-rich UCDW. Episodic release of
14
C-dead geological carbon makes the deep/intermediate
waters as well as the surface of Southern Ocean appear to be ‘old’. The color shading is based on
multiple published ∆
14
C records from sediment cores (not labelled in the panels) that represent
glacial ∆
14
C signature of PDW, AAIW, SAMW in the Southern Hemisphere. The high latitude
Southern Ocean is labelled with question marks due to a lack of data. Panels were generated using
ODV 4.7.10.
Surface Ocean pCO2, 16.5-14 kyr
The interval from 16.5-14 kyrBP is characterized by the most prominent CO2 outgassing
event in our reconstruction (Figure 2.13a, shaded in yellow). There are five data points from the
two records that suggest the surface ΔpCO2 was up to +100ppm during this period. Notably, the

42
peak outgassing coincided with periods of rapidly rising atmospheric CO2 (~16.3 and 14.8 kyrBP)
(Marcott et al., 2014). This outgassing event at the surface is also closely associated with a rapid
rise in carbonate ion saturation at mid depths of the Southwest Pacific (Allen et al., 2015) (Figure
2.13b) and thus, an apparent loss of carbon from mid-depth deep waters (assuming that alkalinity
is unlikely to change this dramatically). At the same time, benthic ∆∆
14
C values (this study, Ronge
et al., 2016) returned to younger (Figure 2.13c), Holocene-like values, suggesting enhanced
ventilation of the intermediate depths of the Southwest Pacific.

Figure 2.16. a): benthic δ
13
C records from the South Pacific that show first rapid increase and
then decline between 16.5-14 kyrBP ( this study; Pahnke & Zahn, 2005; Rose et al., 2010; Siani
et al., 2013). b): benthic δ
13
C records from the South Pacific that show a general increase between
16.5-14 kyrBP (Ronge et al., 2015; Sikes et al., 2016). The lines are 3-point running average.

43
Between 16.5-14 kyrBP, there was a rapid 0.5‰ increase in benthic δ
13
C values, followed
by a 0.3‰ decline at our site. This pattern is also observed in the MD97-2120 and 79JPC records
from the SW Pacific and the MD07-3088 record from the Southeast Pacific, although the
magnitude of δ
13
C drop differs among these records (Figure 2.16a) and in some other cores is
much less pronounced (Figure 2.16b). Given the rapidity of this transient benthic δ
13
C event,
previous authors attributed this pattern changes in ventilation and/or water mass structures driven
by changes in winds or buoyancy forcing (Ronge et al., 2015; Sikes et al., 2016) in the SW Pacific.
More specifically, during late HS1 when the Southern Hemisphere warmed, enhanced upwelling
(Anderson et al., 2009) and a southward shift in the westerlies (Lamy et al., 2010; Putnam et al.,
2010) and the sea ice edge (Ferrari et al., 2014; Rae et al., 2018) would have flushed respired
carbon from these mid-depth waters (Jaccard et al., 2016; Rae et al., 2018) and/or supplied positive
δ
13
C to these layers, due to enhanced air-sea gas exchange in the Southern Ocean. The reverse (i.e.
reduced ventilation of intermediate waters) would have occurred during the Antarctic Cold
Reversal (ACR), a period of SH cooling. A deepening (shoaling) of the AAIW/CDW boundary
during late HS1 (ACR) may also contribute to this pattern (Figure 2.16).
However there are also some differences in the nature of the deglacial benthic d
13
C signal
from this region that may point to some added complexities (Figure 2.16b). For instance, some
records are characterized by a more gradual increase in δ
13
C, ranging from 0.3-0.6‰ between
16.5-14 kyrBP, without any significant drop after HS1. It is possible that the temporal resolutions
of some records (e.g. SO136-003 and SO213-84) are not high enough to capture the excursion.
The potential influence of localized geological sources of CO2 may also influence some of the
records and introduce additional complexity to interpretation of benthic δ
13
C records in this region.
Direct observations of the δ
13
C signature of the carbon flux coming out of such geological systems

44
is crucial to improve our understanding of the benthic δ
13
C from this region. A full investigation
of benthic δ
13
C records from intermediate depths is beyond the scope of the current study, but has
the potential to shed further light on these processes.
Notwithstanding the complexities described above, the large outgassing event between
16.5-14 kyrBP, documented by our new boron isotope data, was associated with a rapid increase
in bottom water [CO3
2-
], benthic δ
13
C and ∆∆
14
C increase, as well as high opal fluxes in the
Southern Ocean, that would have accompanied release of respired - and possibly also geological -
carbon from intermediate depths of the South Pacific, probably through a general enhancement of
ocean ventilation relative to the glacial period. Taking our reconstructed ΔpCO2 at face value, the
data imply substantial carbon transfer from intermediate depths of the South Pacific to the upper
ocean and eventually to the atmosphere between ~16.5-14 kyrBP.
2.4.2 Positive Air-Sea ∆pCO2 Anomalies during the deglaciation and
early Holocene
The median composite ΔpCO2 curve shows a positive excursion of up to 50 ppm during
the last deglaciation, and subtler positive excursions during the early Holocene (Figure 2.17e).
Given the potential caveats of our new data, we also present a second composite excluding the two
records from Chatham Rise (Figure 2.18). Between 14-15 kyrBP and 5-10 kyrBP, ΔpCO2 from the
second composite is lower than the first one by ~10ppm. However, excluding the two records from
Chatham Rise in the composite does not change the first order picture - the composite ΔpCO2
shows a positive anomaly over the last deglaciation and in the early Holocene, providing direct
evidence that the ocean released carbon to the atmosphere at these times.

45

46
Figure 2.17. a) atmospheric pCO2 (Bereiter et al., 2015); b) δ
13
C-CO2 from Schmitt et al., 2012
(green) and Bauska et al., 2016 (Olive); c) relative sea level changes (Spratt & Lisiecki, 2016);
d) the median of 10,000 composite pH from 12 cores (thick blue line), the dotted lines represent
2.5% and 97.5% quantile, the black line with triangle markers represent the median of 10,000
composite pH without our two new records; e) the median of 10,000 composite ∆pCO2 from 12
cores (thick magenta line), the dotted lines represent 2.5% and 97.5% quantile; f) number of
data points from all (red bars)/upwelling (green bars) sites that go into each 1000-yr bin.
Upwelling sites include ODP1238 from the EEP, PS2498 from the sub-Antarctic Atlantic,
MD01-2416 from the subarctic Pacific, NIOP464 and AAS9 from the Northern Indian Ocean.

Figure 2.18. ∆pCO2 composite with all 12 cores included (magenta), and with PC75-1 and PC83-2
from this study excluded (black).

The deglacial ΔpCO2 pulse is the most statistically significant, with the lower 2.5%
quantile being above the equilibrium line (Figure 2.17e). This highlights the importance of ocean
CO2 outgassing for atmospheric pCO2 rise during the last deglaciation. Although it is difficult to
constrain how much area of the ocean each individual core represents, it appears that the deglacial
ΔpCO2 and presumably oceanic CO2 outgassing flux encompassed a broader part of the global

47
surface ocean than the just the areas typically thought of as key for CO 2 outgassing. However, note
that for each 1000-year bin, there are roughly equal amounts of data from upwelling sites and non-
upwelling sites (Figure 2.17f), whereas upwelling regions only occupy a relatively small amount
of the global surface ocean area. These areas are connected to carbon in the deep and abyssal ocean
and therefore were more likely to release extra carbon during the deglaciation. Interestingly, the
site of PC75/83 and PS2498-1, both located in a moderate sink region today, appear to be a source
of carbon over the last deglaciation, perhaps due to frontal shifts (Barker et al., 2009; Bostock et
al., 2015) and/or stronger upwelling in the Southern Ocean (Anderson et al., 2009). Such
transitions are not taken into account by correcting the ‘location bias’, but importantly, they are
likely to reveal real signals of oceanic carbon cycle perturbations. Our results highlight a need to
constrain the oceanic sink regions to better understand the complete history of air-sea CO2
exchange over the last deglaciation.
ΔpCO2 values in the late deglacial and early Holocene are also elevated, notably at a time
of little change in atmospheric pCO2. This may be explained if ocean CO2 outgassing at this time
is partially compensated by expansion of carbon sinks. The most likely candidate is regrowth of
the terrestrial biosphere, which would also contribute to the increase in δ
13
C in the deep ocean and
the atmosphere at this time (Figure 2.17b). Thus, a substantial portion of the CO2 released from
the ocean during the late deglacial and early Holocene was probably being absorbed by terrestrial
vegetation instead of remaining in the atmosphere, as previously suggested based on an increase
in deep ocean carbonate ion concentration (Yu et al., 2010). There is also a significant peak in
ΔpCO2 in the mid Holocene, at a time of relatively stable δ
13
C and gradually rising CO2. This may
be driven in part by CO2 released from renewed shallow water CaCO3 deposition (Ridgwell et al.,

48
2003) as sea levels rose (Spratt & Lisiecki, 2016) (Figure 2.17c), alongside carbonate
compensation at depth (Broecker & Clark, 2007).
The interpretations presented here remain limited by a sparse sampling of the global ocean
and age model uncertainties. Another caveat is the net exchange of carbon between the global
ocean and the atmosphere depends not only on global mean ΔpCO2, but also on the local CO2 gas
exchange rates governed by wind speed, which has not been quantitatively constrained here. Future
work to increase the spatial and temporal resolution of planktic δ
11
B-based pH and CO2
reconstructions, particularly in sink regions, and a better understanding of mechanisms responsible
for the complex pattern of air-sea CO2 exchange in different parts of the global Ocean, is essential
to advancing our understanding of glacial/interglacial CO2 variability.

49

3 Chapter 3
Abstract
During the early last glacial termination (17.2-15 ka) atmospheric δ
13
C declined as atmospheric
pCO2 rose. A declined δ
13
C excursion has been documented in marine proxy records from both
surface and thermocline-dwelling planktic foraminifera. The foraminiferal δ
13
C excursion has
been attributed to a flux of respired carbon from the deep ocean that was subsequently transported
within the upper ocean (i.e. ‘bottom up’ transport) to sites where the excursion is recorded. Here,
we provide modeling evidence that the marine δ
13
C excursion can instead be driven by the
contemporaneous atmospheric δ
13
C decline through air-sea exchange (i.e. top down transport).
Due to this efficient ‘atmospheric bridge’, the pathway of δ
13
C transport was likely to be different
from nutrient transport during the early deglaciation. The widespread upper ocean δ
13
C excursion
limits the usage of planktic δ
13
C records for identifying the carbon source(s) responsible for the
atmospheric pCO2 rise during the early deglaciation. The model results also suggest that
thermocline waters in upwelling systems like the eastern equatorial Pacific, and even upper deep
waters above 2000m, can be affected by this atmospheric bridge during the early deglaciation. Our
results imply that caution must be applied when interpreting early deglacial marine δ
13
C records
from depths affected by the atmospheric δ
13
C forcing.
3.1 Introduction
Atmospheric pCO2 increased by 80-100ppm from the last glacial maximum (LGM) to the
Holocene (Marcott et al., 2014; Monnin et al., 2001). The mechanisms and the chain of events that

50
were responsible for this pCO2 rise are not well understood. High resolution ice core CO2 and δ
13
C
records provide valuable constraints on the timing and magnitude of this deglacial history. During
the initial ~35ppm rise in pCO2 rise between 17.2 to 15 ka, ice core records have documented a
0.3‰ decrease in atmospheric δ
13
C (Bauska et al., 2016; Schmitt et al., 2012). This millennial-
scale trend was punctuated by a rapid 12ppm pCO2 increase between 16.3-16.1 ka (Marcott et al.,
2014) and a -0.2‰ decrease in atmospheric δ
13
C (Bauska et al., 2016). A contemporaneous δ
13
C
drop is recorded by both surface and thermocline dwelling foraminifers (e.g. Hertzberg et al., 2016;
Lund et al., 2019; Spero & Lea, 2002) as well as by benthic foraminifera from shallow depths (e.g.
Lynch-Stieglitz et al., 2019; Romahn et al., 2014; Stott et al., 2019; this study) around the global
ocean. The planktic δ
13
C records have been previously interpreted to reflect a spread of high
nutrient, low δ
13
C waters that upwelled in the Southern Ocean, and then transported throughout
the upper ocean via a so-called intermediate water teleconnection (Martínez-Botí et al., 2015; Pena
et al., 2013; Spero & Lea, 2002). According to this hypothesis, formally sequestered carbon from
deep waters were upwelled in the Southern Ocean (Anderson et al., 2009) in response to a
breakdown of deep ocean stratification (Basak et al., 2018) and this carbon was then carried by
Antarctic Intermediate Water (AAIW) and Southern Ocean Mode Water (SAMW) to low latitudes
where it then outgassed to the atmosphere in upwelling regions like the eastern equatorial Pacific
(EEP). This hypothesis implies that the upper ocean at lower latitudes acts as a conduit for
13
C-
depleted carbon to the atmosphere. Here we term this scenario ‘bottom up’ transport.
Recently, it has been suggested that a negative δ
13
C excursion occurring initially in the
atmosphere (i.e. as recorded by the ice cores) may have left an isotopic imprint on the global
surface ocean through air-sea exchange (Lynch-Stieglitz et al., 2019) . That signal could then be

51
transmitted to thermocline depths and maybe even reach intermediate depths. Here we term this
alternative scenario ‘top down’ transport.
The two scenarios have different implications. In the ‘bottom up’ transport scenario, a
significant portion of
13
C-depleted carbon from the deep ocean was outgassed at low latitudes and
contributed to the atmospheric pCO2 rise, while in the ‘top down’ transport,
13
C-depleted carbon
from the ocean could be outgassed to the atmosphere from anywhere (e.g. the high latitude
Southern Ocean and low latitude upwelling regions), and the subsequent δ
13
C decline in the global
upper ocean does not necessarily contribute to atmospheric pCO2 variability.
Here, we analyze a transient deglacial simulation conducted with the Earth system model
LOVECLIM (Menviel et al., 2018). In this experiment, sequestered respired carbon in the deep
and intermediate waters are ventilated through the Southern Ocean and this leads to a sharp decline
in atmospheric δ
13
C, consistent with ice core records. We evaluate the two δ
13
C transport scenarios
by partitioning the simulated carbon pool and its stable isotope signature into a preformed and a
remineralized component (Methods). The preformed component is dominated by surface
processes such as air-sea thermodynamic equilibrium and primary productivity. The remineralized
component reflects gain/loss of respired carbon due to changes in residence time and export
productivity, or a transient input of respired carbon from the deep ocean. If the ‘top down’
transport scenario was the mechanism responsible for the δ
13
C excursion seen in marine proxy
records, the preformed signal should dominate in the upper 1000m, while a remineralized signal
would dominate the ‘bottom up’ scenario. Our approach requires an accurate representation of the
preformed and remineralized component. The LOVECLIM transient experiment does not simulate
preformed or remineralized carbon explicitly and our offline calculation (see Methods) is likely
subject to errors. However, the LOVECLIM results are complemented by sensitivity experiments

52
performed with another Earth System model (cGENIE), wherein the preformed tracers are
explicitly simulated.
The partitioning framework is not new, previous studies have used this framework to study
the mechanisms that lead to lower glacial atmospheric CO2 (Ito & Follows, 2005; Khatiwala et al.,
2019) and processes that control atmospheric and marine δ
13
C (Menviel et al., 2015; Schmittner
et al., 2013). This diagnostic framework has also been applied to study the carbon cycle
perturbation in response to a weaker AMOC (Schmittner & Lund, 2015), however in that study
the experiments were performed under constant pre-industrial conditions. To our knowledge, the
nature of δ
13
C perturbation during the early deglaciation has not been established yet.
We also present a new benthic δ
13
C record from the upper western equatorial Pacific (WEP)
to expand the deglacial benthic δ
13
C dataset of the upper ocean isotope excursion (e.g. Lynch-
Stieglitz et al., 2019).
3.2 Methods
LOVECLIM deglacial transient simulation
The LOVECLIM model (Goosse et al., 2010) consists of a free-surface primitive
equation ocean model (3° × 3°, 20 vertical levels), a dynamic–thermodynamic sea ice model, an
atmospheric model based on quasi-geostrophic equations of motion (T21, three vertical levels), a
land surface scheme, a dynamic global vegetation model (Brovkin et al., 1997)

and a marine
carbon cycle model (Menviel et al., 2015). To study the sensitivity of the carbon cycle to
different changes in oceanic circulation, a series of transient simulations of the early part of the
last deglaciation (19-15ka) was performed by forcing LOVECLIM with changes in orbital
parameters (Berger, 1978) as well as Northern Hemispheric ice-sheet geometry and albedo (Abe-
Ouchi et al., 2007), and starting from a LGM simulation that best fitted oceanic carbon isotopic

53
(
13
C and
14
C) records (Menviel et al., 2017). The simulation we analyzed for this study is “LH1-
SO-SHW” from Menviel et al, (2018). We briefly describe the relevant deglacial forcing here.
Firstly, a freshwater flux of 0.07 Sv is added into the North Atlantic between 17.6 ka and 16.2
ka, resulting in an AMOC shut down. Secondly, a salt flux is added into the Southern Ocean
between 17.2 ka and 16.0 ka to enhance Antarctic Bottom Water (AABW) formation. Enhanced
AABW could have occurred due to changes in buoyancy forcing at the surface of the Southern
Ocean, or opening of polynyas, processes which could be mis-represented in the model due to its
relatively coarse resolution. Lastly, two stages of enhanced Southern Ocean westerlies are
prescribed in the simulation at 17.2 ka and at 16.2 ka; this timing generally corresponds to
Southern Ocean warming associated with two phases of NADW weakening during Heinrich
Stadial 1 (Hodell et al., 2017). For more detail about this experiment, see Menviel et al., (2018).
cGENIE Sensitivity experiments
The cGENIE model is also based on a 3-D dynamical ocean model plus dynamic and
thermodynamic sea ice components but run at a lower resolution (36x36 horizontal grid with 16
vertical layers.) than LOVECLIM. In addition, cGENIE lacks a dynamical atmosphere and climate
feedback is instead provided by a 2-D energy-moisture balance atmosphere (Edwards and Marsh,
2005), making it much less computationally expensive than LOVECLIM. cGENIE includes a
representation of marine biogeochemical cycling (Ridgwell et al., 2007). Preformed dissolved
inorganic carbon (DICpref) and δ
13
C (δ
13
Cpref) tracers are explicitly simulated in cGENIE (Ödalen
et al., 2018). These are created by resetting the preformed tracer value at the ocean surface to the
equivalent full tracer value at each model time-step, and then allowing ocean circulation to
transport the preformed tracers conservatively – i.e. no remineralization or other interior ocean
geochemical processes are allowed to modify the preformed tracer value.

54

Table 3.1. Prescribed Forcings in cGENIE Experiments
spin free fix
NA freshwater flux +0.1Sv +0.1Sv +0.1Sv
SO freshwater flux 0 -0.3 Sv -0.3Sv
atmosphere pCO2=278ppm;
d
13
Catm= -6.5‰
freely evolve pCO2=278ppm;
d
13
Catm= -6.5‰

A series of cGENIE simulations were run based on the pre-industrial configuration of Cao
et al., (2009) (Table 3.1). The spin-up includes two stages: the first stage was run for 10,000yrs,
with prescribed atmospheric pCO2 = 278ppm, δ
13
C = -6.5‰; the second stage was run for 3,000yrs,
with a 0.1Sv of freshwater input into the North Atlantic to weaken the Atlantic Meridional
Overturning Circulation (AMOC). The atmosphere is fixed at the first stage. We then performed
2 idealized simulations to investigate the role of air-sea exchange in upper ocean δ
13
C, each
integrated for 2000 years. In these sensitivity experiments, in addition to a continuous freshwater
flux into the North Atlantic, we applied a -0.3Sv freshwater flux (i.e. salt flux) into the Pacific
sector of the surface Southern Ocean to enhance AABW formation and ventilation rate. In the
experiment ‘fix’, the atmosphere is held constant as in the spin up. In the experiment ‘free’, the
atmosphere is allowed to evolve freely. The radiative forcing for both experiments is fixed at the
pre-industrial level so that the two experiments have an identical climate even though the CO2

55
concentrations will be different. Note also that by holding the atmospheric CO2 and δ
13
C constant
in the experiment ‘fix’, the atmosphere-ocean inventories of carbon and δ
13
C are slightly different
from those in the experiment ‘free’.
Carbon pool partitioning
To diagnose the mechanisms responsible for the carbon isotope perturbation, the DIC pool
is separated into remineralized and preformed components: DIC = DICreg+ DICpref (Ito and
Follows, 2005). DICreg consists of remineralized organic matter (DICorg) and remineralized
calcium carbonate (DICCaCO3). In LOVECLIM, DICorg is estimated by the Apparent Oxygen
Utilization (AOU); AOU is the difference between saturation concentration for oxygen at the
ambient temperature and salinity and in situ oxygen. DICorg = -RC/O * AOU, where RC/O = 117:-
170 (Anderson & Sarmiento, 1994). DICorg is highly depleted in δ
13
C and changes in DICorg has a
strong impact on the δ
13
C budget. DICCaCO3, on the other hand, has an isotopic composition close
to seawater and does not have a strong influence on oceanic δ
13
C variability. DICCaCO3 and its
impact on δ
13
C is therefore not considered in this study. Changes in δ
13
C (Δδ
13
C) due to
remineralization of organic matter is estimated by Δδ
13
Creg = δ
13
Corg * Δ(
12
Corg /
12
C) = δ
13
Corg *
Δ(
12
Corg/DIC); Δδ
13
C in response to perturbations of the preformed pool (Δδ
13
Cpref) is estimated
by Δδ
13
C - Δδ
13
Creg.
In cGENIE, δ
13
Cpref is explicitly simulated and the remineralization effect is estimated by
Δδ
13
Creg= Δδ
13
C- Δδ
13
Cpref.
Stable Isotope Analyses and Age Model for Piston Core GeoB17402
The WEP piston core GeoB17402 (8°N, 126°34’E, 556m) was recovered from the
expedition SO-228. The planktic foraminiferal samples for
14
C age dating were typically between

56
2 and 5mg. All new radiocarbon ages were measured at the University of California Irvine
Accelerator laboratory. An age model was developed for this core by converting the Trilobatus
sacculifer (T. sacculifer)
14
C ages to calendar age using BChron and the Marine13 calibration
database. We have no constraint on variable surface reservoir ages at this location and therefore
did not apply any adjustment to Delta R. Once the calendar ages were established the results were
plotted vs depth. Between each adjacent T. sacculifer age, linear interpolation was applied to
develop an age model for the core. For benthic foraminiferal d
18
O and d
13
C measurements
approximately 4-8 Cibicidoides mundulus (C. mundulus) were picked. These samples were
cleaned by first cracking the tests open and then sonicating them in deionized water and then dried
at low temperature. The isotope measurements were conducted at the University of Southern
California on a GV Instruments Isoprime mass spectrometer equipped with an autocarb device.
An in-house calcite standard (ultissima marble) was run in conjunction with foraminiferal samples
to monitor analytical precision. The 1s standard deviation for standards measured during the study
was less than 0.1‰ for both d
18
O and d
13
C. The stable isotope data are reported in per mil with
respect to VPDB and will be archived on Pangaea.
3.3 Results
In the LOVECLIM transient simulation, freshwater input into the North Atlantic leads to
reduced North Atlantic Deep Water (NADW) formation. Atlantic Meridional Overturning
Circulation (AMOC) is significantly weaker than its glacial condition by ~18 ka (Figure 3.1a), but
this only has a minor effect on the atmospheric CO2 (Figure 3.1b) and δ
13
C (Figure 3.1c). On the
other hand, enhanced ventilation of Antarctic bottom water (AABW) and Antarctic intermediate
water (AAIW) between 17.2-15 ka leads to an atmospheric CO2 increase of ~25ppm and δ
13
C

57
decline of -0.35‰ (Figure 3.1b, 1c). Below, our analyses are centered around the Pacific basin
between 17.2 - 15.0 ka. The Atlantic basin will be covered in the discussion section.

Figure 3.1. Timeseries from the LOVECLIM transient experiment (Menviel et al., 2018). a)
Freshwater input into the North Atlantic and the Southern Ocean; b) Southern Hemisphere
westerly wind forcing; c) simulated NADW, AABW, AAIW and NPIW maximum stream function
in LOVECLIM. 21-year moving averages are shown for the maximum stream function to filter the
high-frequency variability; d) ice core record of atmospheric CO2 (blue) and LOVECLIM
simulated atmospheric CO2 (red); e) ice core record of atmospheric δ
13
C (black and olive) and
LOVCLIM simulated atmospheric δ
13
C (magenta).

58
The model simulates a global negative sea surface δ
13
C anomaly (here defined as 15-17.2ka)
(Figure 3.2a). The strongest negative anomaly occurs at the surface of the Southern Ocean and
North Atlantic. We separate the simulated sea surface δ
13
C signal (Figure 3.2a) into 2 components:
1) air-sea thermodynamic component due to atmospheric δ
13
C and sea surface temperature (SST)
changes (Δδ
13
Cthermo, Figure 3.2b) (Zhang et al., 1995) and 2) the residual component (Δδ
13
Cres)
(Figure 3.2c) that mainly reflects enhanced primary productivity in response to increased nutrient
supply upwelled from the deep ocean (Figure 3.3).

59

Figure 3.2. a) LOVECLIM simulated sea surface δ
13
C anomaly (15-17.2 ka) b) sea surface δ
13
C
anomaly due to thermodynamic fractionation (air-sea exchange + SST effect) c) residual sea
surface δ
13
C anomaly that are not attributed to thermodynamic fractionation.

60

Figure 3.3. simulated primary productivity anomaly (15-17.2ka) in LOVECLIM
The surface Δδ
13
C (Figure 3.2a) is then propagated into the ocean interior as a preformed
signal when surface waters subduct. To the North of 50°S in the Pacific, δ
13
C in the upper 1000m
(Figure 3.4a) is dominated by a preformed signal of ~-0.3‰ (Figure 3.4b), with minor contribution
from respired carbon transport within the ocean interior (Figure 3.4c).

Figure 3.4. Pacific zonal averaged (160°E-140°W) a) δ
13
C b) δ
13
Creg c) δ
13
Cpref d) PO4 anomaly
(15ka minus 17.2ka) simulated by LOVECLIM. The magenta circle marks the GeoB17402 site.
In the Pacific, Δδ
13
Cpref of -0.2 to -0.3‰ can be traced to ~1000m in the LOVECLIM
simulation (Figure 3.4c). Since there is a minor change in Δδ
13
Creg above 1000m in the tropical
Pacific, this preformed signal would have been recorded by benthic foraminifera from this region.

61
Indeed, our new record from 556m depth in the WEP (Figure 3.5b) documents a -0.3 to -0.4‰
excursion during the early deglaciation (Figure 3.5b), consistent with records from the EEP at
similar depths (Stott et al., 2019). We acknowledge that the magnitude of positive δ
13
C excursion
between 15-13ka is much larger in our benthic δ
13
C record, which might be caused by other factors.
Nonetheless, our new shallow benthic δ
13
C records share a similar ‘W’ shape as the atmospheric
record over the last 20ka, suggesting a sustained influence from the atmosphere.

Figure 3.5. a): Atmospheric δ
13
C records (Bauska et al., 2016; Schmitt et al., 2012) b): C.
mundulus δ
13
C record for upper intermediate and mode waters in the western equatorial Pacific.
The negative δ
13
C excursions in the atmospheric and our benthic record are highlighted in a grey
bar.
Our results support the ‘top down’ transport scenario and challenges the ‘bottom up’
transport scenario as the primary mechanism for the negative δ
13
C excursion seen in upper ocean

62
proxy reconstructions. The ‘top down’ scenario is also compatible with the idea of a nutrient
teleconnection between the Southern Ocean and low latitudes (Palter et al., 2010; Pasquier and
Holzer, 2016; Sarmiento et al., 2004). Figure 3.4d illustrates that stronger upwelling brings excess
nutrients to the surface of the Southern Ocean. Unused nutrients are then transported to low
latitudes within the upper ocean circulation (e.g. through mode waters and thermocline waters).
However, a nutrient teleconnection does not, in itself, reflect enhanced flux of δ
13
C-depleted DIC
from the deep ocean to low latitudes in a ‘tunnel-like’ fashion. The δ
13
C signal that is transported
in the upper ocean has been strongly affected by air-sea gas exchange at the surface of Southern
Ocean and therefore, its pathway is different from the nutrient signal in the LOVECLIM simulation.
To be clear, the stronger negative Δδ
13
Cpref compared to Δδ
13
Creg in the upper ocean does
not mean respired carbon is not important in the simulation. In fact, the ultimate δ
13
C-depleted
carbon source in LOVECLIM is the simulated respired carbon that accumulated in the deep and
intermediate waters during the glacial period as a consequence of the imposed weakened deep
water formation that fits the LGM benthic δ
13
C data (Menviel et al., 2017). The LOVECLIM
transient simulation is used as a tool to investigate the pathway of low δ
13
C signal transport under
a prevailing deglacial scenario that involves Southern Ocean processes. Our results suggest that
when deep ocean stratification breaks down in the model, the δ
13
C-depleted deep waters upwell;
the signal is transmitted to the atmosphere through strong outgassing in the Southern Ocean
(Figure 3.6). δ
13
C in the rest of the upper ocean, particularly in the Indian and Pacific Ocean,
mainly reflect equilibrium thermodynamic exchange with a lower atmospheric δ
13
C.

63

Figure 3.6. changes in air-sea surface pCO2 gradient (15-17.2 ka)
Our approach to partitioning carbon in LOVECLIM is not perfect and subject to errors. For
example, AOU likely over estimates the true Oxygen Utilization, and thus DICorg, particularly in
water masses formed in regions with sea ice (Bernardello et al., 2014; Ito et al., 2004). To confirm
the results obtained from the LOVECLIM simulation are robust, we conducted experiments with
another Earth System model, cGENIE with a freshwater input to the North Atlantic and a salt flux
to the Southern Ocean as in the LOVECLIM simulation. Unlike LOVECLIM, in cGENIE carbon
partitioning can be performed in a traceable and more accurate manner (see Methods). In the
cGENIE ‘free’ experiment, enhanced deep ocean ventilation is associated with a rapid decrease in
upper ocean δ
13
C during the first 1000 years and then stabilizes between model year 1000 and
2000 (not shown). The δ
13
C decline in the upper 1000m is also dominated by the preformed signal
(Figure 3.7), consistent with the LOVECLIM simulation. When atmospheric pCO2 and
atmospheric δ
13
C are held constant in experiment ‘fix’, the Δδ
13
Cpref in the upper 500m is very
small (Figure 3.7d). These sensitivity experiments with cGENIE further reinforce the fact that it
is the atmospheric signal that dominates the upper ocean δ
13
C response.

64

Figure 3.7. cGENIE simulated a) Δδ
13
Creg b) Δδ
13
Cpref in experiment ‘free’ and simulated c)
Δδ
13
Creg d) Δδ
13
Cpref in experiment ‘free’. Anomaly are calculated as the difference between model
year 2000 and 0.
We acknowledge that LOVECLIM and cGENIE are two distinct models and initial
conditions, boundary conditions, prescribed forcing in the simulations are all quite different. For
example, the LOVECLIM simulation starts with a glacial like climate and a strong stratified deep
ocean; atmospheric pCO2 increases from 190 ppm to 207 ppm between 17.2 ka and 15 ka. The
cGENIE simulations begin with a pre-industrial like climate; atmospheric pCO2 increases from
278 ppm to 310 ppm. Thus, direct comparisons between the model simulations are not our focus.

65
Nonetheless, both models suggest upper ocean δ
13
C cannot be used as a direct tracer of respired
carbon transport during the deglaciation, due to the influence of air-sea exchange.
3.4 Atmospheric δ
13
C Bridge
Our simulations imply that the wide-spread deglacial δ
13
C minimum observed in marine
planktic records can be explained, to the first order, by air-sea gas exchange (Lynch-Stieglitz et
al., 2019) originating from the atmosphere. The atmosphere seems to act as a bridge in transmitting
a lower δ
13
C signal from sites where δ
13
C-depleted carbon is released to the atmosphere (i.e. high
latitude Southern Ocean in the LOVECLIM and cGENIE simulation) into the global surface and
subsurface ocean. A good example is the simulated transient δ
13
C minimum event between 16.2 -
15.8 ka in LOVECLIM (Figure 3.1c), which clearly originates from the Southern Hemisphere and
particularly from enhanced ventilation of AAIW (Figure 3.1a). For the upper Pacific, if the ‘top
down’ scenario is true the upper ocean, away from the Southern Hemisphere, would show similar
magnitude of δ
13
C changes as the atmosphere, while water masses in the mid or high latitude
Southern Hemisphere may show different δ
13
C responses due to dynamical circulation and
productivity changes induced by Southern Ocean processes. On the other hand, if the ‘bottom up’
scenario is true, a large depleted δ
13
C signal should first appear in the South Pacific subtropical
gyre (STGSP), then progressively spread to the tropics and finally reach the North Pacific; the
depleted δ
13
C anomaly is also likely to be diluted along its pathway from the South Pacific to the
North Pacific. In the LOVECLIM simulation, there is no δ
13
C minimum in the upstream STGSP,
while the atmosphere-like negative δ
13
C excursion appears in the EEP thermocline, the North
Pacific subtropical gyre (STGNP) and North Pacific Intermediate Water (NPIW) simultaneously
(Figure 3.8). In addition, millennial-scale δ
13
C evolution in these upper ocean water masses to the
north of the equator all show pattern similar to the atmosphere. The synchronized δ
13
C anomaly

66
clearly points to the dominant role of atmospheric communication rather than time-progressive
oceanic transport of a low δ
13
C signal in LOVECLIM.

Figure 3.8. LOVECLIM simulated Δδ
13
C in thermocline EEP (90-82°W, 5°S-5°N, 77-105m),
South Pacific subtropical gyre (STGSP, 160°E- 100°W, 40-22°S, 187-400m), North Pacific
subtropical gyre (STGNP, 110°E- 140°W, 22-40°N, 187-400m), NPIW (167-170°E, 54-57°N,
660m. The average of 23.8-20 ka (i.e. LGM) is used as a reference level for the Δδ
13
C calculations.
The 16.2-15.8 ka excursion is highlighted with a grey bar.
In the LOVECLIM simulation, both millennial- and centennial-scale atmospheric δ
13
C
decline are the result of enhanced deep ocean and/or intermediate ocean ventilation through the
Southern Ocean. Using the UVic Earth-System model, Schmittner and Lund (2015) showed that a
slow-down of AMOC alone is able to weaken the global biological pump and lead to light carbon
accumulation in the upper ocean and the atmosphere, without invoking any Southern Ocean
processes. Despite the different prescribed forcing, Δδ
13
Cpref also dominates the total Δδ
13
C in the

67
upper 1000m of the global ocean in the UVic experiment (See fig. 6 in Schmittner & Lund, 2015).
Taken together, these simulations suggest that any process that lowers the atmospheric δ
13
C would
have an influence on the global upper ocean δ
13
C. This limits the use of planktic δ
13
C records alone
for identifying source(s) and locations of light carbon released to the atmosphere during the last
deglaciation, consistent with what Lynch-Stieglitz et al., (2019) proposed.
3.5 Revisiting EEP Thermocline δ
13
C
Waters at eastern equatorial Pacific (EEP) thermocline depths are thought to be connected to the
deep ocean through AAIW from the south and NPIW from the north. The EEP is therefore a
potential conduit for deep ocean carbon release to the atmosphere. As illustrated in the result
section, the LOVECLIM simulated δ
13
C changes in the thermocline of the EEP mainly reflects a
preformed signal. In the cGENIE experiment ‘free’, Δδ
13
Cpref also plays a dominant role in the
total Δδ
13
C signal in the thermocline EEP. In contrast, in the experiment ‘fix’, there is almost no
change in δ
13
C (Table 3.2). Thus, the release of isotopically-light carbon through the surface
Southern Ocean (in both exp ‘fix’ and ‘free’) and the consequential air-sea re-equilibration of δ
13
C
through gas exchange (in exp ‘free’ only) are both necessary to obtain the observed magnitude of
δ
13
C decline within the EEP thermocline.
Table 3.2. δ
13
C Response in cGENIE Experiments. Δδ
13
C is defined as the difference between
model year 2000 and 0.
Variable ‘free’ ‘fix’
Δδ
13
Catm -0.39‰ 0‰
EEP Δδ
13
C -0.41‰ -0.04‰
EEP Δδ
13
Creg -0.11‰ -0.09‰
EEP Δδ
13
Cpref -0.3‰ +0.05‰

68
The EEP is a dynamical region and observed δ
13
C variability in its upper waters likely
reflects local processes that are not accounted for by the LOVECLIM simulation. However, we
would like to highlight two plantkic δ
13
C records that show strikingly similar evolution to the
model simulation (Figure 3.9). Site GGC17/JPC30 is close to the coast and the wood-constrained
constant surface reservoir ages over the last 20ka suggest this site was not influenced by old
respired carbon from high latitudes (Zhao and Keigwin, 2018). Site ODP1238 is located in the
main upwelling region where strengthened CO2 outgassing inferred from boron isotope data has
been interpreted to reflect respired carbon transported from the Southern Ocean (Martínez-Botí et
al., 2015). If the deglacial history of subsurface influence was indeed distinctively different at the
two sites, the remarkably similar planktic δ
13
C evolution provides strong evidence that thermocline
δ
13
C in the EEP is dominantly controlled by the ‘top down’ mechanism rather than the ‘bottom up’
mechanism as previously suggested (Martínez-Botí et al., 2015; Spero & Lea, 2002), consistent
with the LOVECLIM simulation. At the same time, if there was significant outgassing in the EEP
as expressed in the Boron isotope results, the EEP itself could have partially contributed to the
overall deglacial atmospheric δ
13
C evolution that dominates upper ocean δ
13
C response.
Collectively, our results show that even in strong upwelling regions, where atmospheric
signal δ
13
C from above are likely to be erased by outcropping subsurface waters from below,
thermocline δ
13
C is still subjected to strong atmosphere influences. This implies that the upper few
hundred meters of the water column can be influenced by the atmosphere, consistent with our
interpretation of the new benthic δ
13
C record presented in this study.

69

Figure 3.9. a): Atmospheric δ
13
C records (Bauska et al., 2016; Schmitt et al., 2012), simulated
atmospheric δ
13
C (21-year running average) in LOVECLIM (Menviel et al., 2018) b):
Neogloboquadrina. dutertrei (N. dutertrei, a shallow thermocline species) δ
13
C data from ODP
1238 (Martínez-Botí et al., 2015), GGC17/JPC30 (Zhao and Keigwin, 2018), and LOVECLIM
simulated δ
13
C of DIC at 100m (average of 82-90°W, 5°S-5°N). The N. dutertrei data are corrected
by -0.5‰ to normalize to δ
13
C of DIC (Spero et al., 2003). The grey shadow bars highlight the
time period we focus in this study.
3.6 How deep can the low preformed δ
13
C signal reach during the
early deglaciation?
We have shown that given the dominant negative δ
13
Cpref anomaly in the upper ocean, some
interpretation of planktic δ
13
C records might need to be re-evaluated. Our simulations also reveal

70
that an atmospheric influence can reach much deeper depths, which is supported by the fact that
some benthic δ
13
C records from upper 1000m resemble the atmospheric δ
13
C evolution. It is
plausible that below 1000m, a Δδ
13
Cpref signal from the atmosphere still exists, but no longer
dominates the total Δδ
13
C as Δδ
13
Creg becomes increasingly important at depth.
It has been suggested that deglacial δ
13
C variability in the waters above 2000m depth in
the Atlantic could be driven by air-sea exchange (Lynch-Stieglitz et al., 2019). However, mid-
depth (1800-2100m) benthic δ
13
C records from the Brazil margin (~27°S) document a δ
13
C decline
of -0.4‰ between 18.3 and 17 ka, earlier than the atmospheric δ
13
C signal, which decreases
between 17 and 15 ka (Lund et al., 2019). Lund et al., (2019) suggest these observations seem at
odds with the idea that δ
13
Cpref contributed to δ
13
C variability at their site.

Figure 3.10. Observed and simulated δ
13
C anomaly at the Brazil Margin.
The observed benthic δ
13
C anomaly at these Brazil margin sites are well simulated by
LOVECLIM (Figure 3.10). Interestingly, LOVECLIM also reveals a strong negative Δδ
13
Cpref
signal between 17-15ka when the atmospheric δ
13
C declines (Figure 3.11c). But a positive Δδ
13
Creg
(Figure 3.11b) signal originating from a loss of respired carbon due to enhanced ventilation at

71
those depths completely compensates for the negative Δδ
13
Cpref. As a result, there is only a minor
Δδ
13
C (Figure 3.11a) signal, which is consistent with the proxy observations. Prior to 17.2 ka, δ
13
C
variability at ~2000m depth at the Brazil Margin in the LOVECLIM simulation is dominantly
controlled by accumulation of respired carbon (Figure 3.12b) and there is only a minor contribution
from δ
13
Cpref (Figure 3.12c), as previous studies suggested (Lacerra et al., 2017; Lund et al., 2019;
Schmittner and Lund, 2015).

Figure 3.11. same as Figure 3.4, but for the Atlantic zonal averaged (60°W-10°W) anomaly. The
location of 78GGC and 33GGC (Lund et al., 2015) are marked as magenta circles.

Figure 3.12. same as Figure 3.11, but for 17.2-19ka anomaly in the Atlantic

72
It is not clear why the simulated Δδ
13
Cpref values are more negative in the Atlantic than in
the Pacific in LOVECLIM simulation (compare Figure 3.4c and Figure 3.11c). It could be due to
different circulation and/or subduction of different water masses in the two basins. Nonetheless,
these results suggest that, between 17.2 and 15ka, a negative preformed δ
13
C signal from the
atmosphere needs to be considered when interpreting benthic δ
13
C records shallower than 2000m
depth.

73

4 Chapter 4

Abstract
The Southern Ocean (SO) connects the deep ocean to the upper ocean and the atmosphere
as deep waters outcrop in the SO. Stronger CO2 outgassing in the SO could have released carbon
from the deep ocean and thus contributed to atmospheric pCO2 rise during the last glacial
termination. Surface δ
13
C might be a useful tracer to track past air-sea CO2 exchange and constrain
the SO’s role in releasing carbon from the deep ocean. Here we use a numerical modeling approach
to systematically explore the response of surface δ
13
C and air-sea CO2 flux to a variety of SO
physical and biological processes. We find a large positive air-sea CO2 flux is indeed associated
with large negative surface δ
13
C anomaly in the SO. However, outside of the outgassing region,
equally large δ
13
C decreases are simulated. This implies surface δ
13
C is a poor indicator of air-sea
CO2 flux at a local scale. Nonetheless, large-scale depleted surface δ
13
C in the SO is still likely
linked to excess carbon escaped from the deep ocean. A transient simulation and a compilation of
planktic foraminifera δ
13
C data both indicate depleted δ
13
C in the sub-Antarctic SO that might be
associated with enhanced regional CO2 outgassing during the deglaciation. However, in the
Antarctic zone, a positive deglacial δ
13
C anomaly is not captured by the model, which may indicate
the model misses key interactions among sea ice dynamics, biological activity and air-sea CO2
exchange. Boron isotope reconstructions in seasonal sea ice region might shed light on the SO
dynamics and its role in deglacial CO2 outgassing.

74

4.1 Introduction
The Southern Ocean (SO) is believed to play a vital role in regulating glacial-interglacial
variability of atmospheric CO2 because carbon-rich deep waters outcrop in the SO and exchange
carbon with the atmosphere. Carbon-rich deep water upwells to the south of Antarctic Circumpolar
Current (ACC); the degree to which deep-water carbon is able to escape to the atmosphere depends
on the physical and biological processes in this region. Variations in dissolved Fe-driven biological
carbon fixation (Martin, 1990; Martinez-Garcia et al., 2014) and deep ocean ventilation regulated
by wind stress forcing (Toggweiler et al., 2006), exposure time at the surface and sea-ice related
buoyancy forcing (Ferrari et al., 2014; Stein et al., 2020), determine the strength and the sign of
net air-sea CO2 flux in the SO.
At the end of last ice age, changing conditions in the SO could have enhanced the flux of
CO2 from the ocean to the atmosphere. For example, decreasing aeolian dust flux likely reduced
overall biological carbon export from the surface (Martinez-Garcia et al., 2014). A weakened
Atlantic Meridional Overturning Circulation (AMOC) (McManus et al., 2004) during the early
deglacial interval would have caused a southward shift of the inter-tropical convergence zone
(ITCZ) (Denton et al., 2010), which in turn would have shifted the eddy-driven jet poleward and
intensified the Southern Hemisphere westerlies (Ceppi et al., 2013; Lee et al., 2011) and enhanced
upwelling within the SO (Anderson et al., 2009). Rising atmospheric pCO2 and enhanced
southward heat transport (Stephen Barker et al., 2009) warmed the SO and reduced sea ice
coverage, promoting air-sea gas exchange. Sea ice retreat also reduced freshwater input during the
melting season in the Antarctic Zone compared to the last glacial maximum (LGM), which
introduced negative buoyancy forcing that weakened abyssal stratification (Basak et al., 2018).

75
Consequently, more CO2 could have escaped from a deep ocean reservoir. Decreasing nutrient
utilization at the surface SO inferred from nitrogen isotope data provides further support that these
processes may have transferred nutrient and carbon from the deep ocean to the surface SO (Wang
et al., 2017) and thus would have been a dominant driver of the early deglacial CO2 rise of 35-
40ppm (Monnin et al., 2001) and atmospheric δ
13
C decline (Bauska et al., 2016; Schmitt et al.,
2012).
Although the above mechanisms appear to be compelling, direct evidence that documents
CO2 outgassing in the SO is still lacking. Up to date, there are only a few boron-isotope based
records at ~45°S in the South Pacific (Moy et al., 2019; Shao et al., 2019) and South Atlantic
(Martínez-Botí et al., 2015), each showing distinctively different patterns. Indirect evidence of
CO2 outgassing, has included lower δ
13
C values in some planktic foraminiferal records from the
sub-Antarctic region (Gottschalk et al., 2015; Hu et al., 2020; Martínez-Botí et al., 2015; Tapia et
al., 2019; Ziegler et al., 2013). However, planktic δ
13
C is known to be complicated by changes in
temperature (Bemis et al., 2000), carbonate ion concentration (Howard J. Spero et al., 1997). Thus,
planktic δ
13
C records do not simply reflect δ
13
C of seawater. Correcting all of these complicating
factors and revealing a pure signal of δ
13
C of seawater is not yet practical using proxy
reconstructions. More importantly, to what extent δ
13
C of surface water in the SO could reflect
changes in CO2 outgassing has never been quantitatively investigated.
Here, we present model simulations conducted with an Earth System Model (EMIC) -
cGENIE to explore how surface δ
13
C and air-sea CO2 exchange within the SO respond to varying
physical and biological processes as described above. More specifically, we individually perturb
aeolian dust flux, wind stress and buoyancy flux in the SO to investigate whether surface δ
13
C
anomalies are an indicator of air-sea CO2 flux changes on millennial time scales.

76

Two deglacial modeling experiments were designed to fit ice core records of atmospheric
δ
13
C (Bauska et al., 2016) by transiently varying each of the parameters discussed above. Similar
transient experiments have been conducted with a different model, where processes leading to the
early deglacial CO2 increase were explored (Menviel et al., 2018). In that study, δ
13
C and
radiocarbon time series from selected cores from intermediate and deep waters were used to justify
prescribed salt flux and wind stress forcings in the SO in the LOVECLIM model that would
simulate increasing deep water δ
13
C and reduced ocean ventilation ages over 70% of the ocean’s
volume (i.e. the Indo-Pacific) during the first four thousand years of the deglaciation (Menviel et
al., 2018). We did not use the LOVECLIM output for several reasons: Firstly, reduced iron
fertilization is believed to play an important role during glacial/interglacial transitions; in particular,
the carbon cycle in the SO is sensitive to iron availability. However, biological productivity is not
limited by iron in LOVECLIM. Secondly, enhanced deep-ocean ventilation in LOVECLIM leads
to higher carbonate ion concentration at most of the surface ocean, which contradicts some
available shell-weight observations (Moy et al., 2019). As discussed above, planktic δ
13
C records
are affected by carbonate ion changes. By simulating changes in carbonate ion at least in the right
direction is an important aspect of this study. Lastly, our main interest is, to what extent, changes
in surface δ
13
C in the SO correlates with air-sea CO2 flux given a plausible deglacial forcing
scenario. But air-sea CO2 fluxes are not readily available from the LOVECLIM simulations.
We emphasize that the forcings chosen for this study are specifically intended to evaluate
one question, how well planktic δ
13
C data reflect changing air-sea gas exchange. For this reason,
the forcings applied are not be taken as ‘known’ solutions to the deglacial carbon cycle dynamics.
Millennial-scale atmospheric CO2 is likely to be controlled by multiple processes and most of the

77
proposed mechanisms are not well constrained (Gottschalk et al., 2019). It is likely that the model
simulation achieves a ‘correct’ answer (in the sense of being consistent with the few available
observations) but for the wrong reasons. Nonetheless, the transient deglacial simulation allows us
to investigate to what extent planktic δ
13
C records from mid and high latitude Southern
Hemisphere are useful for studying changing CO2 outgassing.
4.2 Methods
cGENIE modeling
The spin-up is a 20ka run under glacial boundary conditions of albedo, aeolian iron flux
and orbital configurations (Odalen et al., 2020, in prep). The configuration includes an iron cycle
where biological productivity is limited by both PO4 and Fe. For each sensitivity experiment, the
first 1000 yr is forced by applying a +0.05SV freshwater forcing to the North Atlantic and -0.05SV
freshwater forcing to the North Pacific relative to the LGM spin-up. After the first 1000 years the
following forcings were applied (each experiment is labeled):
Exp1 ‘FwF’: 2000 yrs with the same freshwater forcing as in the first 1000 yrs.
Exp2 ‘FwF_Fe’: same as ‘FwF’. Additionally, dust forcing is switched from LGM-like condition
to a Holocene-like condition (Albani et al., 2016).
Exp3 ‘FwF_SO0.15Sv’: same as ‘FwF’. Additionally, a -0.15SV freshwater forcing is applied to
the surface SO to enhance SO ventilation.
Exp4 ‘FwF_SHW50’: same as ‘FwF’. Additionally, wind stress is enhanced by 50% in the SO
between 60 to 40°S.
Two transient experiments ‘fit Taylor’ and ‘no centennial’ were run to compare with
observations. These runs were forced by transient variation of Aeolian dust flux, SO wind stress
and salt flux. Detail descriptions are provided in the result section.

78

Evaluating temperature and carbonate ion effects on planktic foraminiferal calcite δ
13
C:
To the south of 30°S, Globigerina bulloides and Neogloboquadrina pachyderma are the
most common species used to generate planktic stable isotope records. The carbon isotope
composition from the two species appears to be offset from sea water δ
13
C. The offset is not a
constant and is affected by sea surface temperature and carbonate ion concentration. We follow
the calibration established by Kohfeld et al., (2000) to correct the temperature and carbonate ion
effect:
∆δ
13
Cforam=∆δ
13
Cseawater -0.13*∆T-0.013*∆CO3

79

Figure 4.1. surface δ
13
C anomaly versus air-sea CO2 flux anomaly in four sensitivity experiments.
Anomalies are calculated as model year 3000-0. Each dot represents a surface grid box in cGENIE.
The black dash lines in each panel represent atmospheric δ
13
C anomaly in that particular
sensitivity experiment.

4.3 Correlation between air-sea CO2 flux anomaly and surface δ
13
C
anomaly in sensitivity experiments

80
Figure 4.1 shows that although a slowdown of AMOC (exp ‘FwF’) generates some perturbation
in surface δ
13
C and air-sea CO2 flux in the global ocean, no CO2 outgassing is simulated in mid-
or high-latitude Southern Hemisphere (orange dots in Figure 4.1a). On the other hand, in all of
other three experiments where individual SO processes are perturbed, positive air-sea CO2 flux
anomalies and negative surface δ
13
C anomalies in the SO are simulated (orange dots in Figure 4.1b,
c, d). Enhanced SO wind stress and salt flux input seem to generate larger magnitude of anomaly
than switching aeolian iron flux from a LGM setting to a Holocene setting. The cross plot also
suggests that given a similar millennial-scale surface δ
13
C anomaly, air-sea flux anomalies at
individual surface grid points in the model can be vastly different. This is consistent among all
three distinct SO mechanisms. Outside of the SO (blue dots in Figure 4.1), equally large negative
surface δ
13
C anomaly are associated with no changes or even negative air-sea CO2 flux anomaly.
4.4 Spatial and temporal surface δ
13
C and air-sea CO2 flux pattern in
transient simulations
Now we turn to the two transient deglacial simulations that include variations in all three
SO mechanisms. In ‘Fit_Taylor’, extreme salt flux into the SO and stronger westerlies were
prescribed in an effort to simulate the centennial atmosphere CO2 rise and δ
13
C drop event
documented by high-resolution ice cores between 16.4-16.1 ka (Figure 4.2). It is possible that
during this period carbon was released from a terrestrial source not from the ocean (Nielsen et al.,
2019) or from marine geologic sources that is not accounted for by our model. If this atmospheric
excursion had a marine source, the available marine records have yet to resolve it. Thus, a second
transient experiment ‘without centennial’ was run to only simulate the slow millennial event.

81

Figure 4.2. Time series of applied forcing are shown in panel a, b and c. The simulated
atmospheric CO2 and δ
13
C are plotted against ice core data in panel d and e, respectively.
Simulated δ
13
C and CO3 time series at intermediate Pacific and Atlantic plotted against
observations are shown in Figure 4.3 to justify our choices of forcing. The two sites are chosen
because 1) they are sensitive to circulation changes at both deep and intermediate depths in the
real world and are sensitive to the parameters we perturbed in the model. 2) Paired carbon isotope
and carbonate chemistry constraints give us more confidence that the model captures the essential

82
carbon cycle dynamics at these key sites. We note that it is difficult to assess whether simulated
CO3 and δ
13
C at the two sites in ‘Fit Taylor’ or ‘without centennial’ matches the observations
better. Nonetheless, we focus on ‘Fit Taylor’ in the following discussions.

Figure 4.3. δ
13
C and CO3 model-data comparison at intermediate depth of the South Pacific (Allen
et al., 2020) and South Atlantic (Lacerra et al., 2019).

The simulated CO2 outgassing occurs at ~45-55°S in the Atlantic and west Indian sector of
the SO and ~52-58°S in the east Pacific sector of the SO (Figure 4.4b), to the north of the Antarctic

83
seasonal sea ice region in the model. These regions are also characterized with the strongest
westerlies and are the largest sink in the model at the LGM spin-up (Figure 4.5).

Figure 4.4. a) simulated surface δ
13
C anomaly (15-18.4ka). b) simulated air-sea CO2 flux anomaly.
The purple rectangles mark the boundaries of CO2 outgassing band. The black lines represent 50%
annual sea ice coverage.

84

Figure 4.5. Simulated air-sea CO2 flux at LGM in cGENIE. The purple rectangles are the same
as Figure 4.4.
Within CO2 outgassing hotspots (i.e. within the purple rectangles in Figure 4.4a, 4.4b),
local maximum negative δ
13
C anomalies correspond to local maximum CO2 outgassing anomalies
quite well. However, in the South Atlantic, North Pacific, and the broad tropical region, equally
large negative δ
13
C anomalies (Figure 4.6a) are associated with CO2 absorption (Figure 4.6b). To
further illustrate that surface δ
13
C provides little constraint on air-sea CO2 flux, particularly outside
of the SO, we show δ
13
C and air-sea CO2 flux time series at four locations where the oceanographic
settings are completely different (Figure 4.6). Surface δ
13
C evolution at these sites share very
similar histories (Figure 4.6a), while only the site at high latitude SO shows increasing air-sea CO2
flux and decreasing surface δ
13
C dynamically in response to the applied forcing (Figure 4.6b,
black).

85

Figure 4.6. Simulated δ
13
C (a) and air-sea CO2 flux (b) time series at three locations with different
oceanographic settings.
4.5 Deglacial surface SO δ
13
C, a model-data comparison
We now present data-model comparison that takes the temperature and carbonate ion
effects on planktic δ
13
C into account. In response to strong upwelling of relatively warm deep
water as well as increased heat transport to the South Atlantic due to a slowdown of AMOC, the
Southern Hemisphere warms (Figure 4.7a). Extra DIC from the upwelled deep water decreases
carbonate ion concentration at the surface SO (Figure 4.7b). Higher SSTs and lower carbonate ion
concentration shift δ
13
C of foraminifera shell (δ
13
Cforam) towards lower and higher values,

86
respectively (Figure 4.8a, 4.8b). The two effects compensate each other in the SO and the resulting
δ
13
Cforam anomaly is essentially the same as δ
13
Cseawater anomaly (compare Figure 4.8c and 4.8d).
One exception is the South Atlantic, where higher SSTs and higher carbonate ion concentration
both shift δ
13
Cforam towards lower values (Figure 4.8a, 4.8b), which leads to the strongest negative
δ
13
Cforam anomaly in this region compared to the South Pacific and the South Indian Ocean (Figure
4.8d). Nonetheless, the magnitude and pattern of simulated negative excursion in planktic δ
13
C
from the outgassing band is generally consistent with observations (Figure 4.8c, 4.8d).

Figure 4.7. Simulated surface temperature (a), CO3 (b) anomaly in the Southern Hemisphere.

87

Figure 4.8. Simulated surface temperature (a), CO3 effect (b) on foraminifera δ
13
C. Panel c is
identical as figure 4.4a. Panel d shows model-predicted δ
13
C anomaly potentially recorded by
foraminifera with all three effects taken into account. Panel c and d are plotted with compiled
planktic δ
13
C data (HS1-LGM)
Near the Southern edge of the outgassing band or farther south, planktic δ
13
C data
document positive anomalies. One proposed explanation for these positive anomaly is that reduced
sea ice coverage due to deglacial warming could have promoted air-sea gas exchange and increases
productivity; both mechanisms would have increased surface δ
13
C, particularly during the austral
summer growing season (Gottschalk et al., 2015). However, the simulated δ
13
C anomaly in those
regions are still negative in the model, although the signal is much weaker (Figure 4.8c, 4.8d).
Seasonal output from the transient run and sensitivity experiments also do not show evidence of

88
positive δ
13
C anomaly during the austral summer (not shown). It is possible that the
thermodynamic sea ice module in cGENIE is too crude to simulate dynamical sea ice response to
ocean circulation and wind forcing, which may prohibit the model from simulating surface δ
13
C
correctly in regions characterized with seasonal sea ice coverage.

89

5 Chapter 5

Abstract
It is thought that the glacial/interglacial carbon cycle dynamics involved a repartitioning of
carbon between the atmosphere and ocean carbon reservoirs. Inherent in this hypothesis is the
notion that the flux of geologic carbon and alkalinity to/from the oceans remains constant on
glacial-interglacial time scales. However, emerging evidence now suggests that the flux of
geological carbon to/from the surface environment varied on these timescales during the
Pleistocene. Here we explore potential changes in ocean dissolved inorganic carbon (DIC)
inventory by studying the ocean’s marine radiocarbon budget. The ocean’s total carbon inventory
at any point in time can be calculated if the bulk ocean
14
C/
12
C and
14
C production rate in the
atmosphere (
14
Cpro) are both known. We present a bulk ocean
14
C/
12
C record from a compiled
dataset (Zhao et al., 2018), which shows 15-20% decrease between 25-14 ka. Unfortunately,
observation-based
14
Cpro estimates differ from each other and do not provide strong constraints
during this period. We use an Earth System Model (cGENIE) to evaluate the required
14
Cpro and
ocean carbon inventory changes that would be required to reproduce the record of bulk ocean
14
C/
12
C between 25-14 ka. This is achieved by :1)
14
Cpro drops from 2.17 atoms cm
-2
s
-1
to 1.75
atoms cm
-2
s
-1
with a fixed modern DIC inventory (39000 GtC) and 2) the ocean DIC inventory
increases from 35000 GtC to 40000 GtC with fixed
14
Cpro at 2 atoms cm
-2
s
-1
. Our findings, along
with other lines of evidence, suggest that geological DIC and ALK flux into and out of the surface
environment has significantly varied over glacial-interglacial cycles and this would require a
fundamental change to studies that attempt to explain the history of atmospheric pCO2.

90

5.1 Introduction
Radiocarbon is produced in the atmosphere by cosmic ray interactions with Nitrogen atoms.
14
Cpro is modulated by the solar flux and the Earth’s magnetic field. Once
14
C atoms are produced
in the atmosphere they enter the Earth’s surface carbon cycle and are incorporated into one of the
major carbon reservoirs, including the ocean and terrestrial carbon pools. With a half-life of 5730
years,
14
C can be used as a tracer in carbon cycle studies back to ~50 ka. Records of Marine
radiocarbon have been obtained by measuring radiocarbon content of fossil carbonates produced
by protists (foraminifera) and corals. These records document the history of atmosphere-ocean
radiocarbon offset (i.e. ocean radiocarbon age) and the evolution of bulk ocean radiocarbon over
the last ~30 ka. A majority of studies have sought to learn how changes in the rate of ocean
overturning circulation affected the partitioning of carbon between the atmosphere and the ocean
and by doing so, learn how the ocean may have modulated the atmospheric CO2 concentration (e.g.
Skinner et al., 2017). Only recently, have some researchers begun to consider the possibility that
the ocean-atmosphere carbon pool is an open system on glacial/interglacial timescales and that the
total inventory of carbon in the surface environment changed in response to variations in the flux
of geologic carbon (Cartapanis et al., 2018; Galbraith & Skinner, 2020; Huybers & Langmuir,
2017; Kölling et al., 2019; Lund & Asimow, 2011; Rafter et al., 2019; Stott & Timmermann, 2011;
Torres et al., 2017). Close to a geological carbon sources, some of the ocean radiocarbon ages
obtained from marine carbonates are essentially decoupled from true ocean ventilation ages.
Moreover, if the geological flux is large enough to significantly increase the total DIC inventory
(i.e.
12
C), bulk ocean
14
C/
12
C (denoted as ∆
14
C) should decrease when
14
Cpro remains constant.
Indeed, a recent compilation has shown that ∆
14
C declined in all ocean basins over the last glacial

91
termination (Zhao et al., 2018). On the other hand, it is also possible that the production rate of
14
C
decreased over the glacial termination that led to a lowering of bulk ocean ∆
14
C. We first review
previous efforts to reconstruct
14
Cpro. We then use the bulk ocean ∆
14
C as a constraint to test what
change in
14
Cpro would be required to reproduce the history of bulk ocean ∆
14
C decline. These
results of that experiment guide a further modeling experiment that explores how variations in the
ocean carbon inventory influences the ∆
14
C when
14
Cpro does not change.

92

Figure 5.1. a) observation-based atmospheric
14
C production rate reconstructions (Adolphi et al.,
2018; Channell et al., 2018; Laj et al., 2000; Laj et al., 2004; Nowaczyk et al., 2013) b)
atmospheric ∆
14
C records and c) bulk ocean ∆
14
C estimate.

Studies that have attempted to reconstruct the history of
14
Cpro indicate the highest
14
Cpro

occurred during the Laschamp event between 42.3-39.7ka (Lascu et al., 2016) (Figure 5.1a). The
large positive excursion in
14
Cpro is replicated in four independent paleomagnetic (Channell et al.,
2018; Laj et al., 2000; Laj et al., 2004; Nowaczyk et al., 2013) and one
10
Be-based estimates of

93
14
Cpro (Figure 5.1a). The Laschamp excursion is also associated with the highest atmospheric
∆
14
C values of the last 50ka, as depicted in the new high resolution Hulu Cave record (Cheng et
al., 2018) (Figure 5.1b). After the Laschamp excursion, the average
14
Cpro (based on five estimates)
was about 2 atoms cm
-2
s
-1
between 38 and 14 ka, with large variations among the reconstructions.
In a recent study (Dinauer et al., 2020) used each of the 5 observation-based
14
Cpro curves
to force an Earth system model under pre-industrial boundary conditions in an effort to estimate
how atmospheric ∆
14
C would respond to these reconstructed changes in
14
Cpro. Their simulated
atmospheric ∆
14
C response consistently falls below the Intcal 13 and Hulu Cave records. Dinauer
et al., (2020) then attempted to independently reconstruct
14
Cpro by forcing their carbon cycle
model with the Intcal13 atmospheric ∆
14
C record. The result depends critically on the strength of
air-sea exchange of
14
C in the model as the atmosphere only holds less than 2% of the total
14
C in
the atmosphere-ocean system. For example, a stronger air-sea exchange due to enhanced ocean
overturning circulation and/or stronger winds leaves less
14
C in the atmosphere. In such a scenario,
the model must then diagnose a higher
14
Cpro to maintain a higher atmospheric ∆
14
C. In their study
Dinauer et al., (2020) varied the strength of overturning circulation and winds to generate a suite
of model-based
14
Cpro; the simulated
14
Cpro are consistently higher than the paleomagnetic and
10
Be-based estimates. They concluded that current observations are likely to systematically
underestimate
14
Cpro during the last glacial period. However, if the real glacial
14
Cpro fell within
the range of observed glacial
14
Cpro values (Figure 5.1a), the higher glacial atmospheric ∆
14
C
could be explained by a smaller glacial atmosphere-ocean carbon inventory, an idea that Dinauer
et al., (2020) did not consider in their study.
For the purpose of constraining atmospheric
14
Cpro and the atmosphere-ocean
14
C budget,
the bulk ocean ∆
14
C is a better target as most of the
14
C resides in the Ocean. A recent compilation

94
of ocean ∆
14
C records provides an opportunity for us to calculate bulk ocean ∆
14
C

that spans the
last 25 ka (Zhao et al., 2018). This bulk ∆
14
C

record documents high glacial ∆
14
C

values of
approximately 200‰ followed by a 200‰ decrease during the last glacial termination between
25-14ka (Figure 5.1c). For the reasons outlined above, variations in the bulk ocean ∆
14
C
(approximately bulk ∆
14
C in the atmosphere-ocean system)

are dominantly controlled by changes
in
14
Cpro and/or DIC inventory, while ocean physics plays a minor role. In this study we fit a
numerical Earth System model - cGENIE to this bulk ocean ∆
14
record to evaluate what changes
in
14
Cpro or DIC inventory are required to explain the history of bulk ocean ∆
14
C between 25-
14ka. We argue that with our current knowledge of past bulk ocean ∆
14
C (relatively well
constrained) and
14
Cpro (relatively poorly constrained), ocean DIC and ALK inventories could
have each varied by as much as 15% on the glacial-interglacial timescale.
5.2 Methods
We conducted two transient model experiments to simulate the evolution of bulk ocean
∆
14
C

between 25 and 14 ka by 1) altering
14
C production rate (EXP “
14
Cpro”) and 2) changing the
global ocean DIC inventory (EXP “Inventory”).

In EXP “
14
Cpro”, the cGENIE model was spun-up under glacial boundary conditions, but
atmospheric ∆
14
C was set to 673‰, a bit higher than Intcal13. The resulting
14
Cpro at equilibrium
is =2.18 atoms cm
-2
s
-1
. The benefit of prescribing higher atmospheric ∆
14
C is to allow the
simulated initial bulk ocean ∆
14
C to match the observed values between 26-23 ka.
In EXP “Inventory”, the cGENIE model was spin-up under glacial boundary conditions,
except that the atmospheric ∆
14
C was set to 673‰ as in EXP “
14
Cpro”. In addition, the initial DIC
and ALK inventory are reduced by ~15%. At the end of the spin-up,
14
Cpro held constant at ~2

95
atoms cm
-2
s
-1
. The bulk ocean ∆
14
C is similar to EXP “
14
Cpro” at the end of the spin-up. The
reason that a lower
14
Cpro rate in EXP “Inventory” generates a similar bulk ocean ∆
14
C as EXP
“
14
Cpro” is because the initial ocean DIC inventory is smaller (Figure 5.2b).
5.3 transient evolution of
14
Cpro, atmospheric ∆
14
C and bulk ocean
∆
14
C in cGENIE experiments
Figure 5.2a illustrates the transient evolution of
14
Cpro. In EXP “
14
Cpro”, a decrease in
14
Cpro from 2.17 atoms cm
-2
s
-1
to 1.97 atoms cm
-2
s
-1
(i.e. 10% reduction compared to the spin up)
during the first 4000 yrs is required to match the observed bulk ocean ∆
14
C. Over the next 7000
yrs,
14
Cpro is further decreased to 1.75 atoms cm
-2
s
-1
(i.e. 20% decrease compared to the spin up).
In EXP “Inventory”, after the spin-up,
14
Cpro is held constant at 2 atoms cm
-2
s
-1
. Figure 5.2b
illustrates ocean DIC inventory changes. In EXP “Inventory”,
14
C-dead DIC was injected into the
ocean at 41.67 Tmol/yr (i.e. 0.5 PgC/yr) between 25-14 ka. As a result, DIC inventory increases
by 5500 GtC and ends up with a slightly higher value at 14ka than the standard LGM configuration
in EXP “
14
Cpro”. For each one mole of DIC injected, there is an associated 1.25 mol of Alkalinity
(ALK) flux injected into the ocean as well. The ALK flux is required to maintain a constant
atmospheric CO2 between 25-14 ka as reflected in ice core records. In section 5.6, we demonstrate
the implication of varying geological DIC:ALK flux ratio on atmospheric CO2.
We note that these transient forcings serve as a first-order estimate rather than a precise
reconstruction of
14
Cpro and/or ocean DIC/ALK inventory, that allows our model to match the
reconstructed bulk ocean ∆
14
C (Figure 5.2d). No attempt has been made to match the simulated
atmospheric ∆
14
C to Intcal13 (Figure 5.2c).

96

Figure 5.2. a) prescribed
14
Cpro in the two experiments plotted with reconstructions (grey), b)
simulated ocean DIC inventory, c) simulated atmospheric ∆
14
C and d) simulated ocean bulk ∆
14
C
plotted with reconstructions.

97
5.4 Magnitude of deglacial geological flux compared to modern
geological flux
If the observed bulk ocean ∆
14
C evolution (Figure 5.2d) is best explained by changes in
14
Cpro, our results indicate 1) during the glacial period, a
14
Cpro value at the high end of current
reconstructed
14
Cpro estimates would be required to maintain the high bulk ocean ∆
14
C values. 2)
the decreasing bulk ocean ∆
14
C trend between 25-14ka (the last deglaciation) requires up to a 20%
reduction in
14
Cpro. None of the current
14
Cpro reconstructions unambiguously documents such a
decreasing trend in production rate (Figure 5.1a).
The alternative to the
14
C production forcing requires a change in total carbon invitatory.
In this case the bulk ocean ∆
14
C evolution is best explained by a smaller ocean DIC inventory
during the last glacial period and a gradual expanding DIC inventory during the deglaciation. To
accommodate this scenario and the progressively decrease ∆
14
C from 25 ka to 14 ka (Figure 5.2d),
the total flux of carbon and alkalinity to the ocean from geologic sources must have increased.
The geological DIC and ALK flux imbalance would have been on the order of 40Tmol/yr. To put
these values into context, the modern geological carbon (Wallmann & Aloisi, 2012) and ALK flux
(Middelburg et al., 2020) are listed in Figure 5.3. These flux terms should not be viewed as a
complete list that fully describes the modern geological carbon and ALK cycle as recent studies
continue to find new processes that are significant enough to affect the DIC and ALK budgets. For
example, HCO3 flux from freshwater ground discharge could be 80%-237% of the global riverine
HCO3 flux (S. Zhang & Planavsky, 2020). Anaerobic oxidation of methane in shallow marine
sediment converts methane to DIC in the form of HCO3 (DIC: ALK= ~1:1), which mostly flows
into the local water column. The total outflux is estimated to be ∼20% of global riverine flux to
oceans (Akam et al., 2020). Nonetheless, a large positive geological carbon and ALK flux

98
imbalance is required to explain the deglacial negative trend in bulk ∆
14
C. The magnitude of the
imbalance would be ~50-100% of the total modern flux.

Figure 5.3. a) Modern Geological a) carbon (Wallmann & Aloisi 2012) b) ALK flux (Middelburg
et al., 2020). The flux of reverse weathering follows (Isson et al., 2020).
Large local negative ∆
14
C excursions during the last deglaciation have been observed on
benthic foraminifera from various parts of the ocean that indicate local geological carbon input
(Bova et al., 2018; Bryan et al., 2010; Marchitto et al., 2007; Rafter et al., 2018; T. A. Ronge et
al., 2016; L. Stott et al., 2009, 2019) (Figure 5.4).
14
C–dead carbon injection from these locations
likely contributed to the deglacial trend in bulk ocean ∆
14
C. However, it is challenging to estimate

99
what the local geological carbon flux has been from each site because neither the spatial extent of
∆
14
C excursions nor the ocean circulation variability that would have diluted the ∆
14
C signals are
well constrained. One exception perhaps is the Gulf of California, where the semi-enclosed
geometry and an array of benthic ∆
14
C measurements from multiple sites allowed the authors to
estimate the local geological carbon flux to be 0.76-1.56 Tmol/yr during the deglaciation (Rafter
et al., 2019). Interestingly, the geological carbon source at the Gulf of California was proposed to
be from sediment pyrolysis and anaerobic oxidation of methane, which is also associated with
same order of ALK outflux. This is consistent to our model results.

Figure 5.4. Five locations where benthic ∆
14
C records document a negative excursion that is
larger than the bulk ocean ∆
14
C trend at the glacial termination. Multiple records over a range of
depth were averaged to represent a local ∆
14
C anomaly. Negative ∆
14
C excursion at these
locations occurred at different stages of the last glacial termination. Note the magnitude of local
∆
14
C anomaly does not necessarily scale with the local flux of geological carbon input.
Our results provide a first-order constraint on the total geological carbon and ALK fluxes
to the ocean that would be required at the end of last glacial termination if the total carbon
inventory was responsible for the observed history of
14
C. Of course, if more reliable
14
Cpro rate
estimates turn out to capture a decreasing trend over this period, the estimated geological carbon

100
flux to explain the depleting bulk ∆
14
C would be less. Hence, refining the uncertainties in the
history of
14
Cpro evolution are critically important.
5.5 Implications to ocean carbonate chemistry and atmospheric pCO2
It has long been considered that geological carbon and ALK cycles are stable over one full
glacial cycle. However, new evidence is suggesting that the largest sink terms in the geological
carbon and ALK cycle have in fact, changed over the last glacial cycle. Reconstructions of global
CaCO3 and POC burial in the open ocean/shallow water environment over the last 150 ka
(Cartapanis et al., 2016, 2018) show weaker and stronger sinks during the glacial and interglacial
period, respectively. The magnitude of glacial-interglacial variability of these sink terms are up to
20 Tmol/yr (Figure 5.5).

Figure 5.5. Carbon and ALK sink flux due to CaCO3 (Cartapanis et al., 2018) and POC burial
(Cartapanis et al., 2016, Cartapanis et al., 2018).

The reconstructed variations in the sink terms as well as bulk ocean ∆
14
C constraints presented in
this study indicate that ‘short-term’ imbalances on time scales of tens of thousands of years could

101
shrink/expand ocean DIC/ALK inventory by as much as 15%. More precise reconstructions of
14
Cpro and bulk ocean ∆
14
C record beyond 25 ka would provide further constraints on this estimate.

Figure 5.6. Simulated atmospheric pCO2 in experiment “Inventory” (solid purple, injection rate
= 0.5 PgC/yr) versus ice core record of pCO2 (black). Also plotted is simulated atmospheric pCO2
in an additional experiment where the DIC:ALK injection ratio switches from 1:1.25 to 1:1 at
18ka (dash purple).
Although simulating the deglacial pCO2 rise was not the focus of this study, we emphasize
that if ocean DIC/ALK inventory did vary significantly on glacial-interglacial timescale due to
variations in geological flux, it would be a potential new mechanism to explain some part of the
glacial/interglacial atmospheric pCO2 variability. More specifically, in EXP “Inventory”, DIC:
ALK injection ratio was kept at 1:1.25 such that the simulated atmospheric CO2 remains stable
(Figure 5.6, solid purple). However, if this ratio is 1:1 at the beginning of the deglaciation (~18
ka), the geological flux would have contributed more than 20ppm to the atmospheric pCO2 rise
(Figure 5.6, dash purple).
One way to test the proposed dynamical ocean DIC/ALK inventory on the
glacial/interglacial timescale could be a careful data-model comparison in carbonate ion
concentration. Everything else being equal, a smaller glacial carbon inventory would have been
associated with a lower global average carbonate ion content, although it is not clear how big the

102
signal might be. Models with an open CaCO3 cycle configuration could be used in future research
to estimate the magnitude of such a signal. On the data side, an array of B/Ca based carbonate ion
reconstructions from the global ocean over 25-18 ka would be ideal as atmospheric CO2, sea level,
ocean circulation and biological pump were all at semi-equilibrium state during this time interval
and therefore a ‘reservoir effect’ could potentially be teased out.

103

6 Chapter 6
6.1 Conclusions
In chapter 2, I added two new boron isotope records to the existing database, partly filling
an important spatial data gap in the subtropical-subpolar transition zone of the SW Pacific. The
results suggest that the Chatham Rise was a source of carbon to the atmosphere; some of the carbon
was from the Earth’s interior. Composite records of pH and pCO2 derived from 12 boron isotope
records throughout the global ocean captures the first-order deglacial trend and also reveals
pronounced ocean outgassing to the atmosphere from a broad part of the global surface ocean
during the last deglaciation and early Holocene.
In chapter 3, I have shown that ocean-atmosphere gas exchange likely dominates the
negative δ
13
C excursions documented in surface and subsurface proxy reconstructions between
17.2ka and 15ka. Numerical simulations performed with LOVECLIM and cGENIE suggest that
enhanced Southern Ocean upwelling leads to a transfer of δ
13
C-depleted respired carbon from the
Southern Ocean to the atmosphere. Consequently, atmospheric δ
13
C declines, which leaves its
imprint on the global upper ocean through air-sea equilibration. This preformed signal could
account for a negative marine δ
13
C excursion up to 0.3-0.4‰ at the surface and down to 2000m
depth during the early deglaciation. At the same time, it is also possible that there were other sites
where excess carbon was ventilated to the atmosphere during the deglaciation (e.g. boron isotope
records presented in Chapter 2), which would have also affected atmospheric δ
13
C. Our findings
therefore imply that planktic δ
13
C records do not provide strong constraints on the site or the
mechanisms through which CO2 was released from the ocean to the atmosphere on a global scale.

104
Interpretation of early deglacial benthic δ
13
C records shallower than 2000m depth needs to take
into account an atmospheric influence.
In chapter 4, I explored how SO processes such as reduced iron fertilization, enhanced deep
ocean ventilation and stronger deep ocean upwelling are linked to CO2 outgassing in the SO and
whether the strength of CO2 outgassing anomaly correlates with the magnitude of negative surface
δ
13
C anomaly. In response to these perturbations, the sub-Antarctic SO switches from a strong
CO2 sink during the LGM to a weak CO2 sink during the deglaciation, which leads to the largest
CO2 outgassing anomaly in our model simulation. Within the outgassing band, there is indeed a
correlation between the CO2 outgassing and surface δ
13
C anomaly. However, the model also
simulates equally large surface δ
13
C anomalies in other parts of the ocean where no anomalously
high CO2 outgassing occurs. This also includes the Chilean Margin and the Tasmania Sea that are
located at similar latitudes as the simulated outgassing band in the SO. We conclude that even if
the SO was responsible for most of the oceanic CO2 outgassing during the deglaciation, planktic
δ
13
C is not an ideal proxy to identify whether a particular site in the SO is located within the actual
outgassing band or not. However, a regional δ
13
C depletion could be an indication of enhanced
deep water upwelling that promotes CO2 outgassing. A compilation of plantkic δ
13
C data shows
negative and positive δ
13
C trend depletion in the sub-Antarctic and Antarctic zone during the
deglaciation, respectively. This is consistent with a regionalized CO2 outgassing in the sub-
Antarctic zone, although the exact outgassing hotspot cannot be constrained by planktic δ
13
C. In
the model outgassing hotspots are located where rapid seasonal sea ice melting and strongest
westerlies occurs. The positive δ
13
C anomaly in the Antarctic zone is not captured by the model,
suggesting the model misses key feedbacks probably associated with sea ice. Boron isotope-based

105
seawater pCO2 reconstructions at sites affected by seasonal sea ice would be helpful in evaluating
the strength of CO2 outgassing in the SO.
In chapter 5, a global ocean benthic ∆
14
C dataset allows us to estimate the radiocarbon
budget of the atmosphere-ocean system. If the carbon inventory in the atmosphere-ocean system
remained constant between 25-14 ka as generally thought, our simulation suggests the secular
evolution in bulk ∆
14
C could be explained by a relatively high
14
Cpro during the glacial period,
followed by a ~20% reduction during the deglaciation. However, significant differences among
the available
14
Cpro reconstructions leave the production rate scenario not well-constrained.
Alternatively, if
14
Cpro remained effectively constant, the secular evolution in bulk ∆
14
C would
require a shrinking ocean DIC inventory (compare to modern) during the glacial and a gradual
expanding DIC inventory during the deglaciation. Our estimated imbalanced carbon flux is ~0.5
PgC/yr, about 50-100% of the modern total carbon source flux. The bulk ∆
14
C reconstruction
suggests that the imbalance began during the LGM (25-19 ka) when atmospheric pCO2 was low
and stable. Thus the enhanced geological carbon input scenario requires the same order of ALK
flux imbalance to the ocean during the LGM. There is no evidence-based hypothesis to explain
how or why the geological flux changes that produces a large DIC and ALK imbalance; no one
knows how the DIC:ALK ratio might have evolved over time that contributes to atmospheric pCO2
variation. Nonetheless, I hope this study stimulates a new direction for research as emerging
evidence suggests glacial-interglacial atmospheric pCO2 as well as ocean carbonate chemistry
variability might be more of an open-system problem that involves variable geological flux rather
than a close-system problem that only involved carbon redistribution between the atmosphere and
the deep ocean.

106
References

Adolphi, F., Bronk Ramsey, C., Erhardt, T., Edwards, R. L., Cheng, H., Turney, C. S. M., et al.
(2018). Connecting the Greenland ice-core and U⁄Th timescales via cosmogenic
radionuclides: testing the synchroneity of Dansgaard–Oeschger events. Climate of the
Past, 14(11), 1755–1781. https://doi.org/10.5194/cp-14-1755-2018
Akam, S. A., Coffin, R. B., Abdulla, H. A. N., & Lyons, T. W. (2020). Dissolved Inorganic
Carbon Pump in Methane-Charged Shallow Marine Sediments: State of the Art and New
Model Perspectives. Frontiers in Marine Science, 7.
https://doi.org/10.3389/fmars.2020.00206
Albani, S., Mahowald, N. M., Murphy, L. N., Raiswell, R., Moore, J. K., Anderson, R. F., et al.
(2016). Paleodust variability since the Last Glacial Maximum and implications for iron
inputs to the ocean: PALEODUST VARIABILITY AND IRON. Geophysical Research
Letters, 43(8), 3944–3954. https://doi.org/10.1002/2016GL067911
Allen, K. A., Sikes, E. L., Hönisch, B., Elmore, A. C., Guilderson, T. P., Rosenthal, Y., &
Anderson, R. F. (2015). Southwest Pacific deep water carbonate chemistry linked to high
southern latitude climate and atmospheric CO2 during the Last Glacial Termination.
Quaternary Science Reviews, 122, 180–191.
https://doi.org/10.1016/j.quascirev.2015.05.007
Allen, K. A., Sikes, E. L., Anderson, R. F., & Rosenthal, Y. (2020). Rapid Loss of CO 2 From
the South Pacific Ocean During the Last Glacial Termination. Paleoceanography and
Paleoclimatology, 35(2). https://doi.org/10.1029/2019PA003766

107
Anderson, L. A., & Sarmiento, J. L. (1994). Redfield ratios of remineralization determined by
nutrient data analysis. Global Biogeochemical Cycles, 8(1), 65–80.
https://doi.org/10.1029/93GB03318
Anderson, R. F., Ali, S., Bradtmiller, L. I., Nielsen, S. H. H., Fleisher, M. Q., Anderson, B. E., &
Burckle, L. H. (2009). Wind-Driven Upwelling in the Southern Ocean and the Deglacial
Rise in Atmospheric CO2. Science, 323(5920), 1443–1448.
https://doi.org/10.1126/science.1167441
Barker, S., Greaves, M., & Elderfield, H. (2003). A study of cleaning procedures used for
foraminiferal Mg/Ca paleothermometry. Geochemistry, Geophysics, Geosystems, 4(9).
https://doi.org/10.1029/2003GC000559
Barker, Stephen, Diz, P., Vautravers, M. J., Pike, J., Knorr, G., Hall, I. R., & Broecker, W. S.
(2009). Interhemispheric Atlantic seesaw response during the last deglaciation. Nature,
457(7233), 1097–1102. https://doi.org/10.1038/nature07770
Basak, C., Fröllje, H., Lamy, F., Gersonde, R., Benz, V., Anderson, R. F., et al. (2018). Breakup
of last glacial deep stratification in the South Pacific. Science, 359(6378), 900–904.
https://doi.org/10.1126/science.aao2473
Bauska, T. K., Baggenstos, D., Brook, E. J., Mix, A. C., Marcott, S. A., Petrenko, V. V., et al.
(2016). Carbon isotopes characterize rapid changes in atmospheric carbon dioxide during
the last deglaciation. Proceedings of the National Academy of Sciences, 113(13), 3465–
3470. https://doi.org/10.1073/pnas.1513868113
Bemis, B. E., Spero, H. J., Lea, D. W., & Bijma, J. (2000). Temperature influence on the carbon
isotopic composition of Globigerina bulloides and Orbulina universa (planktonic

108
foraminifera). Marine Micropaleontology, 38(3–4), 213–228.
https://doi.org/10.1016/S0377-8398(00)00006-2
Bereiter, B., Eggleston, S., Schmitt, J., Nehrbass-Ahles, C., Stocker, T. F., Fischer, H., et al.
(2015). Revision of the EPICA Dome C CO2 record from 800 to 600 kyr before present:
Analytical bias in the EDC CO2 record. Geophysical Research Letters, 42(2), 542–549.
https://doi.org/10.1002/2014GL061957
Bernardello, R., Marinov, I., Palter, J. B., Sarmiento, J. L., Galbraith, E. D., & Slater, R. D.
(2014). Response of the Ocean Natural Carbon Storage to Projected Twenty-First-
Century Climate Change. Journal of Climate, 27(5), 2033–2053.
https://doi.org/10.1175/JCLI-D-13-00343.1
Bostock, H. C., Hayward, B. W., Neil, H. L., Sabaa, A. T., & Scott, G. H. (2015). Changes in the
position of the Subtropical Front south of New Zealand since the last glacial period.
Paleoceanography, 30(7), 824–844. https://doi.org/10.1002/2014PA002652
Bova, S. C., Herbert, T. D., & Altabet, M. A. (2018). Ventilation of Northern and Southern
Sources of Aged Carbon in the Eastern Equatorial Pacific During the Younger Dryas
Rise in Atmospheric CO 2. Paleoceanography and Paleoclimatology, 33(11), 1151–1168.
https://doi.org/10.1029/2018PA003386
Broecker, W., & Clark, E. (2007). Is the magnitude of the carbonate ion decrease in the abyssal
ocean over the last 8 kyr consistent with the 20 ppm rise in atmospheric CO 2 content?:
HOLOCENE DECREASE IN DEEP-SEA CARBONATE. Paleoceanography, 22(1),
n/a-n/a. https://doi.org/10.1029/2006PA001311

109
Brovkin, V., Ganopolski, A., Archer, D., & Munhoven, G. (2012). Glacial CO 2 cycle as a
succession of key physical and biogeochemical processes. Climate of the Past, 8(1), 251–
264. https://doi.org/10.5194/cp-8-251-2012
Brovkin, Victor, Ganopolski, A., & Svirezhev, Y. (1997). A continuous climate-vegetation
classification for use in climate-biosphere studies. Ecological Modelling, 101(2–3), 251–
261. https://doi.org/10.1016/S0304-3800(97)00049-5
Bryan, S. P., Marchitto, T. M., & Lehman, S. J. (2010). The release of 14C-depleted carbon from
the deep ocean during the last deglaciation: Evidence from the Arabian Sea. Earth and
Planetary Science Letters, 298(1–2), 244–254. https://doi.org/10.1016/j.epsl.2010.08.025
Cao, L., Eby, M., Ridgwell, A., Caldeira, K., Archer, D., Ishida, A., et al. (2009). The role of
ocean transport in the uptake of anthropogenic CO2. Biogeosciences, 6(3), 375–390.
https://doi.org/10.5194/bg-6-375-2009
Cartapanis, O., Bianchi, D., Jaccard, S. L., & Galbraith, E. D. (2016). Global pulses of organic
carbon burial in deep-sea sediments during glacial maxima. Nature Communications,
7(1). https://doi.org/10.1038/ncomms10796
Cartapanis, O., Galbraith, E. D., Bianchi, D., & Jaccard, S. L. (2018). Carbon burial in deep-sea
sediment and implications for oceanic inventories of carbon and alkalinity over the last
glacial cycle. Climate of the Past, 14(11), 1819–1850. https://doi.org/10.5194/cp-14-
1819-2018
Ceppi, P., Hwang, Y.-T., Liu, X., Frierson, D. M. W., & Hartmann, D. L. (2013). The
relationship between the ITCZ and the Southern Hemispheric eddy-driven jet: ITCZ
AND JET LATITUDE. Journal of Geophysical Research: Atmospheres, 118(11), 5136–
5146. https://doi.org/10.1002/jgrd.50461

110
Channell, J. E. T., Hodell, D. A., Crowhurst, S. J., Skinner, L. C., & Muscheler, R. (2018).
Relative paleointensity (RPI) in the latest Pleistocene (10–45 ka) and implications for
deglacial atmospheric radiocarbon. Quaternary Science Reviews, 191, 57–72.
https://doi.org/10.1016/j.quascirev.2018.05.007
Chase, Z., Anderson, R. F., Fleisher, M. Q., & Kubik, P. W. (2003). Accumulation of biogenic
and lithogenic material in the Pacific sector of the Southern Ocean during the past 40,000
years. Deep Sea Research Part II: Topical Studies in Oceanography, 50(3–4), 799–832.
https://doi.org/10.1016/S0967-0645(02)00595-7
Cheng, H., Edwards, R. L., Southon, J., Matsumoto, K., Feinberg, J. M., Sinha, A., et al. (2018).
Atmospheric
14
C/
12
C changes during the last glacial period from Hulu Cave. Science,
362(6420), 1293–1297. https://doi.org/10.1126/science.aau0747
Coffin, R. B., Boyd, T. J., Rose, P. S., Yoza, B., Millholland, L. C., Downer, R., & Woods, S.
(2013). Geochemical Cruise Report SO226/2 RV Sonne Chatham Rise Expedition: Fort
Belvoir, VA: Defense Technical Information Center.
https://doi.org/10.21236/ADA587416
Davy, B., Pecher, I., Wood, R., Carter, L., & Gohl, K. (2010). Gas escape features off New
Zealand: Evidence of massive release of methane from hydrates: MASSIVE NZ GAS
ESCAPE FEATURES. Geophysical Research Letters, 37(21).
https://doi.org/10.1029/2010GL045184
Denton, G. H., Anderson, R. F., Toggweiler, J. R., Edwards, R. L., Schaefer, J. M., & Putnam,
A. E. (2010). The Last Glacial Termination. Science, 328(5986), 1652–1656.
https://doi.org/10.1126/science.1184119

111
Dickson, A. G. (1990). Thermodynamics of the dissociation of boric acid in synthetic seawater
from 273.15 to 318.15 K. Deep Sea Research Part A. Oceanographic Research Papers,
37(5), 755–766. https://doi.org/10.1016/0198-0149(90)90004-F
Dinauer, A., Adolphi, F., & Joos, F. (2020). Mysteriously high
Δ14C of the glacial atmosphere: Influence of
14C production and carbon cycle changes (preprint). Carbon
Cycle/Modelling only/Millenial/D-O. https://doi.org/10.5194/cp-2019-159
Edgar, K. M., Anagnostou, E., Pearson, P. N., & Foster, G. L. (2015). Assessing the impact of
diagenesis on δ 11 B, δ 13 C, δ 18 O, Sr/Ca and B/Ca values in fossil planktic
foraminiferal calcite. Geochimica et Cosmochimica Acta, 166, 189–209.
https://doi.org/10.1016/j.gca.2015.06.018
Edwards, N. R., & Marsh, R. (2005). Uncertainties due to transport-parameter sensitivity in an
efficient 3-D ocean-climate model. Climate Dynamics, 24(4), 415–433.
https://doi.org/10.1007/s00382-004-0508-8
Evans, D., Wade, B. S., Henehan, M., Erez, J., & Müller, W. (2016). Revisiting carbonate
chemistry controls on planktic foraminifera Mg/Ca: implications for sea surface
temperature and hydrology shifts over the Paleocene–Eocene Thermal Maximum and
Eocene–Oligocene transition. Climate of the Past, 12(4), 819–835.
https://doi.org/10.5194/cp-12-819-2016
Ezat, M. M., Rasmussen, T. L., Hönisch, B., Groeneveld, J., & deMenocal, P. (2017). Episodic
release of CO2 from the high-latitude North Atlantic Ocean during the last 135 kyr.
Nature Communications, 8, 14498. https://doi.org/10.1038/ncomms14498

112
Farmer, J. R., Hönisch, B., & Uchikawa, J. (2016). Single laboratory comparison of MC-ICP-MS
and N-TIMS boron isotope analyses in marine carbonates. Chemical Geology, 447, 173–
182. https://doi.org/10.1016/j.chemgeo.2016.11.008
Ferrari, R., Jansen, M. F., Adkins, J. F., Burke, A., Stewart, A. L., & Thompson, A. F. (2014).
Antarctic sea ice control on ocean circulation in present and glacial climates. Proceedings
of the National Academy of Sciences, 111(24), 8753–8758.
https://doi.org/10.1073/pnas.1323922111
Foster, Gavin L., & Rae, J. W. B. (2016). Reconstructing Ocean pH with Boron Isotopes in
Foraminifera. Annual Review of Earth and Planetary Sciences, 44(1), 207–237.
https://doi.org/10.1146/annurev-earth-060115-012226
Foster, Gavin L., Hönisch, B., Paris, G., Dwyer, G. S., Rae, J. W. B., Elliott, T., et al. (2013).
Interlaboratory comparison of boron isotope analyses of boric acid, seawater and marine
CaCO3 by MC-ICPMS and NTIMS. Chemical Geology, 358, 1–14.
https://doi.org/10.1016/j.chemgeo.2013.08.027
Foster, G.L. (2008). Seawater pH, pCO2 and [CO
2-
3] variations in the Caribbean Sea over the last
130 kyr: A boron isotope and B/Ca study of planktic foraminifera. Earth and Planetary
Science Letters, 271(1–4), 254–266. https://doi.org/10.1016/j.epsl.2008.04.015
Foster, G.L., & Sexton, P. F. (2014). Enhanced carbon dioxide outgassing from the eastern
equatorial Atlantic during the last glacial. Geology, 42(11), 1003–1006.
https://doi.org/10.1130/G35806.1
Franois, R., Altabet, M. A., Yu, E.-F., Sigman, D. M., Bacon, M. P., Frank, M., et al. (1997).
Contribution of Southern Ocean surface-water stratification to low atmospheric CO2

113
concentrations during the last glacial period. Nature, 389(6654), 929–935.
https://doi.org/10.1038/40073
Fry, C. H., Tyrrell, T., Hain, M. P., Bates, N. R., & Achterberg, E. P. (2015). Analysis of global
surface ocean alkalinity to determine controlling processes. Marine Chemistry, 174, 46–
57. https://doi.org/10.1016/j.marchem.2015.05.003
Galbraith, E., & de Lavergne, C. (2019). Response of a comprehensive climate model to a broad
range of external forcings: relevance for deep ocean ventilation and the development of
late Cenozoic ice ages. Climate Dynamics, 52(1–2), 653–679.
https://doi.org/10.1007/s00382-018-4157-8
Galbraith, E. D., & Skinner, L. C. (2020). The Biological Pump During the Last Glacial
Maximum. Annual Review of Marine Science, 12(1), 559–586.
https://doi.org/10.1146/annurev-marine-010419-010906
Goosse, H., Brovkin, V., Fichefet, T., Haarsma, R., Huybrechts, P., Jongma, J., et al. (2010).
Description of the Earth system model of intermediate complexity LOVECLIM version
1.2. Geoscientific Model Development, 3(2), 603–633. https://doi.org/10.5194/gmd-3-
603-2010
Gottschalk, J., Skinner, L. C., & Waelbroeck, C. (2015). Contribution of seasonal sub-Antarctic
surface water variability to millennial-scale changes in atmospheric CO2 over the last
deglaciation and Marine Isotope Stage 3. Earth and Planetary Science Letters, 411, 87–
99. https://doi.org/10.1016/j.epsl.2014.11.051
Gottschalk, J., Skinner, L. C., Lippold, J., Vogel, H., Frank, N., Jaccard, S. L., & Waelbroeck, C.
(2016). Biological and physical controls in the Southern Ocean on past millennial-scale

114
atmospheric CO2 changes. Nature Communications, 7(1).
https://doi.org/10.1038/ncomms11539
Gottschalk, J., Battaglia, G., Fischer, H., Frölicher, T. L., Jaccard, S. L., Jeltsch-Thömmes, A., et
al. (2019). Mechanisms of millennial-scale atmospheric CO2 change in numerical model
simulations. Quaternary Science Reviews, 220, 30–74.
https://doi.org/10.1016/j.quascirev.2019.05.013
Gray, W. R., & Evans, D. (2019). Non-thermal influences on Mg/Ca in planktonic foraminifera:
A review of culture studies and application to the last glacial maximum.
Paleoceanography and Paleoclimatology. https://doi.org/10.1029/2018PA003517
Gray, W. R., Rae, J. W. B., Wills, R. C. J., Shevenell, A. E., Taylor, B., Burke, A., et al. (2018).
Deglacial upwelling, productivity and CO2 outgassing in the North Pacific Ocean. Nature
Geoscience, 11(5), 340–344. https://doi.org/10.1038/s41561-018-0108-6
Gray, W. R., Weldeab, S., Lea, D. W., Rosenthal, Y., Gruber, N., Donner, B., & Fischer, G.
(2018). The effects of temperature, salinity, and the carbonate system on Mg/Ca in
Globigerinoides ruber (white): A global sediment trap calibration. Earth and Planetary
Science Letters, 482, 607–620. https://doi.org/10.1016/j.epsl.2017.11.026
Hain, M. P., Sigman, D. M., & Haug, G. H. (2010). Carbon dioxide effects of Antarctic
stratification, North Atlantic Intermediate Water formation, and subantarctic nutrient
drawdown during the last ice age: Diagnosis and synthesis in a geochemical box model:
ATMOSPHERIC CO 2 DURING THE LAST ICE AGE. Global Biogeochemical Cycles,
24(4), n/a-n/a. https://doi.org/10.1029/2010GB003790

115
Hasenclever, J., Knorr, G., Rüpke, L. H., Köhler, P., Morgan, J., Garofalo, K., et al. (2017). Sea
level fall during glaciation stabilized atmospheric CO2 by enhanced volcanic degassing.
Nature Communications, 8, 15867. https://doi.org/10.1038/ncomms15867
Haslett, J., & Parnell, A. (2008). A simple monotone process with application to radiocarbon-
dated depth chronologies. Journal of the Royal Statistical Society: Series C (Applied
Statistics), 57(4), 399–418. https://doi.org/10.1111/j.1467-9876.2008.00623.x
Hemming, N. G., & Hanson, G. N. (1992). Boron isotopic composition and concentration in
modern marine carbonates. Geochimica et Cosmochimica Acta, 56(1), 537–543.
https://doi.org/10.1016/0016-7037(92)90151-8
Henehan, M. J., Rae, J. W. B., Foster, G. L., Erez, J., Prentice, K. C., Kucera, M., et al. (2013).
Calibration of the boron isotope proxy in the planktonic foraminifera Globigerinoides
ruber for use in palaeo-CO2 reconstruction. Earth and Planetary Science Letters, 364,
111–122. https://doi.org/10.1016/j.epsl.2012.12.029
Henehan, M. J., Foster, G. L., Bostock, H. C., Greenop, R., Marshall, B. J., & Wilson, P. A.
(2016). A new boron isotope-pH calibration for Orbulina universa, with implications for
understanding and accounting for ‘vital effects.’ Earth and Planetary Science Letters,
454, 282–292. https://doi.org/10.1016/j.epsl.2016.09.024
Hertzberg, J. E., Lund, D. C., Schmittner, A., & Skrivanek, A. L. (2016). Evidence for a
biological pump driver of atmospheric CO 2 rise during Heinrich Stadial 1: Bio Pump and
CO 2 Rise During HS1. Geophysical Research Letters, 43(23), 12,242-12,251.
https://doi.org/10.1002/2016GL070723

116
Hines, S. K. V., Southon, J. R., & Adkins, J. F. (2015). A high-resolution record of Southern
Ocean intermediate water radiocarbon over the past 30,000 years. Earth and Planetary
Science Letters, 432, 46–58. https://doi.org/10.1016/j.epsl.2015.09.038
Hodell, D. A., Nicholl, J. A., Bontognali, T. R. R., Danino, S., Dorador, J., Dowdeswell, J. A., et
al. (2017). Anatomy of Heinrich Layer 1 and its role in the last deglaciation: HEINRICH
EVENT 1. Paleoceanography, 32(3), 284–303. https://doi.org/10.1002/2016PA003028
Hönisch, B., & Hemming, N. G. (2005). Surface ocean pH response to variations in pCO2
through two full glacial cycles. Earth and Planetary Science Letters, 236(1–2), 305–314.
https://doi.org/10.1016/j.epsl.2005.04.027
Hönisch, B., Bijma, J., Russell, A. D., Spero, H. J., Palmer, M. R., Zeebe, R. E., & Eisenhauer,
A. (2003). The influence of symbiont photosynthesis on the boron isotopic composition
of foraminifera shells. Marine Micropaleontology, 49(1–2), 87–96.
https://doi.org/10.1016/S0377-8398(03)00030-6
Hu, R., & Piotrowski, A. M. (2018). Neodymium isotope evidence for glacial-interglacial
variability of deepwater transit time in the Pacific Ocean. Nature Communications, 9(1).
https://doi.org/10.1038/s41467-018-07079-z
Hu, R., Bostock, H. C., Greaves, M., Piotrowski, A. M., & McCave, I. N. (2020). Coupled
evolution of stable carbon isotopes between the Southern Ocean and the atmosphere over
the last 260 ka. Earth and Planetary Science Letters, 538, 116215.
https://doi.org/10.1016/j.epsl.2020.116215
Huybers, P., & Langmuir, C. (2009). Feedback between deglaciation, volcanism, and
atmospheric CO2. Earth and Planetary Science Letters, 286(3–4), 479–491.
https://doi.org/10.1016/j.epsl.2009.07.014

117
Huybers, P., & Langmuir, C. H. (2017). Delayed CO 2 emissions from mid-ocean ridge
volcanism as a possible cause of late-Pleistocene glacial cycles. Earth and Planetary
Science Letters, 457, 238–249. https://doi.org/10.1016/j.epsl.2016.09.021
Isson, T. T., Planavsky, N. J., Coogan, L. A., Stewart, E. M., Ague, J. J., Bolton, E. W., et al.
(2020). Evolution of the Global Carbon Cycle and Climate Regulation on Earth. Global
Biogeochemical Cycles, 34(2). https://doi.org/10.1029/2018GB006061
Ito, T., Follows, M. J., & Boyle, E. A. (2004). Is AOU a good measure of respiration in the
oceans?: AOU AND RESPIRATION. Geophysical Research Letters, 31(17), n/a-n/a.
https://doi.org/10.1029/2004GL020900
Ito, Takamitsu, & Follows, M. J. (2005). Preformed phosphate, soft tissue pump and atmospheric
CO₂. Journal of Marine Research, 63(4), 813–839.
https://doi.org/10.1357/0022240054663231
Jaccard, S. L., Galbraith, E. D., Martínez-García, A., & Anderson, R. F. (2016). Covariation of
deep Southern Ocean oxygenation and atmospheric CO2 through the last ice age. Nature,
530(7589), 207–210. https://doi.org/10.1038/nature16514
Jonkers, L., & Kučera, M. (2015). Global analysis of seasonality in the shell flux of extant
planktonic Foraminifera. Biogeosciences, 12(7), 2207–2226. https://doi.org/10.5194/bg-
12-2207-2015
Khatiwala, S., Schmittner, A., & Muglia, J. (2019). Air-sea disequilibrium enhances ocean
carbon storage during glacial periods. Science Advances, 5(6), eaaw4981.
https://doi.org/10.1126/sciadv.aaw4981
Kısakürek, B., Eisenhauer, A., Böhm, F., Garbe-Schönberg, D., & Erez, J. (2008). Controls on
shell Mg/Ca and Sr/Ca in cultured planktonic foraminiferan, Globigerinoides ruber

118
(white). Earth and Planetary Science Letters, 273(3–4), 260–269.
https://doi.org/10.1016/j.epsl.2008.06.026
Klochko, K., Kaufman, A. J., Yao, W., Byrne, R. H., & Tossell, J. A. (2006). Experimental
measurement of boron isotope fractionation in seawater. Earth and Planetary Science
Letters, 248(1–2), 276–285. https://doi.org/10.1016/j.epsl.2006.05.034
Kohfeld, K. E., & Ridgwell, A. (2009). Glacial-interglacial variability in atmospheric CO2. In C.
Le Quéré & E. S. Saltzman (Eds.), Geophysical Monograph Series (Vol. 187, pp. 251–
286). Washington, D. C.: American Geophysical Union.
https://doi.org/10.1029/2008GM000845
Kölling, M., Bouimetarhan, I., Bowles, M. W., Felis, T., Goldhammer, T., Hinrichs, K.-U., et al.
(2019). Consistent CO2 release by pyrite oxidation on continental shelves prior to glacial
terminations. Nature Geoscience, 12(11), 929–934. https://doi.org/10.1038/s41561-019-
0465-9
La j, C., Kissel, C., Mazaud, A., Channell, J. E. T., & Beer, J. (2000). North Atlantic
palaeointensity stack since 75ka (NAPIS–75) and the duration of the Laschamp event.
Philosophical Transactions of the Royal Society of London. Series A: Mathematical,
Physical and Engineering Sciences, 358(1768), 1009–1025.
https://doi.org/10.1098/rsta.2000.0571
Lacerra, M., Lund, D., Yu, J., & Schmittner, A. (2017). Carbon storage in the mid-depth Atlantic
during millennial-scale climate events: Mid-depth Atlantic Carbon Storage.
Paleoceanography, 32(8), 780–795. https://doi.org/10.1002/2016PA003081
Lacerra, M., Lund, D. C., Gebbie, G., Oppo, D. W., Yu, J., Schmittner, A., & Umling, N. E.
(2019). Less Remineralized Carbon in the Intermediate‐Depth South Atlantic During

119
Heinrich Stadial 1. Paleoceanography and Paleoclimatology, 34(7), 1218–1233.
https://doi.org/10.1029/2018PA003537
Laj, C., Kissel, C., & Beer, J. (2004). High Resolution Global Paleointensity Stack Since 75 kyr
(GLOPIS-75) Calibrated to Absolute Values. In J. E. T. Channell, D. V. Kent, W.
Lowrie, & J. G. Meert (Eds.), Geophysical Monograph Series (pp. 255–265).
Washington, D. C.: American Geophysical Union. https://doi.org/10.1029/145GM19
Lambeck, K., Rouby, H., Purcell, A., Sun, Y., & Sambridge, M. (2014). Sea level and global ice
volumes from the Last Glacial Maximum to the Holocene. Proceedings of the National
Academy of Sciences, 111(43), 15296–15303. https://doi.org/10.1073/pnas.1411762111
Lamy, F., Kilian, R., Arz, H. W., Francois, J.-P., Kaiser, J., Prange, M., & Steinke, T. (2010).
Holocene changes in the position and intensity of the southern westerly wind belt. Nature
Geoscience, 3(10), 695–699. https://doi.org/10.1038/ngeo959
Lascu, I., Feinberg, J. M., Dorale, J. A., Cheng, H., & Edwards, R. L. (2016). Age of the
Laschamp excursion determined by U-Th dating of a speleothem geomagnetic record
from North America. Geology, 44(2), 139–142. https://doi.org/10.1130/G37490.1
Lauvset, S. K., Key, R. M., Olsen, A., van Heuven, S., Velo, A., Lin, X., et al. (2016). A new
global interior ocean mapped climatology: the 1◦ × 1◦ GLODAP version 2, 16.
Lea, D. W., Mashiotta, T. A., & Spero, H. J. (1999). Controls on magnesium and strontium
uptake in planktonic foraminifera determined by live culturing. Geochimica et
Cosmochimica Acta, 63(16), 2369–2379. https://doi.org/10.1016/S0016-7037(99)00197-
0
Lee, S.-Y., Chiang, J. C. H., Matsumoto, K., & Tokos, K. S. (2011). Southern Ocean wind
response to North Atlantic cooling and the rise in atmospheric CO 2 : Modeling

120
perspective and paleoceanographic implications. Paleoceanography, 26(1).
https://doi.org/10.1029/2010PA002004
Lund, D., Hertzberg, J., & Lacerra, M. (2019). Carbon isotope minima in the South Atlantic
during the last deglaciation: evaluating the influence of air-sea gas exchange.
Environmental Research Letters, 14(5), 055004. https://doi.org/10.1088/1748-
9326/ab126f
Lund, D. C., Tessin, A. C., Hoffman, J. L., & Schmittner, A. (2015). Southwest Atlantic water
mass evolution during the last deglaciation: DEGLACIAL SOUTHWEST ATLANTIC
CIRCULATION. Paleoceanography, 30(5), 477–494.
https://doi.org/10.1002/2014PA002657
Lund, D. C., Asimow, P. D., Farley, K. A., Rooney, T. O., Seeley, E., Jackson, E. W., &
Durham, Z. M. (2016). Enhanced East Pacific Rise hydrothermal activity during the last
two glacial terminations. Science, 351(6272), 478–482.
https://doi.org/10.1126/science.aad4296
Lund, David C., & Asimow, P. D. (2011). Does sea level influence mid-ocean ridge magmatism
on Milankovitch timescales?: SEA LEVEL AND MID-OCEAN RIDGE MAGMATISM.
Geochemistry, Geophysics, Geosystems, 12(12), n/a-n/a.
https://doi.org/10.1029/2011GC003693
Lüthi, D., Le Floch, M., Bereiter, B., Blunier, T., Barnola, J.-M., Siegenthaler, U., et al. (2008).
High-resolution carbon dioxide concentration record 650,000–800,000 years before
present. Nature, 453(7193), 379–382. https://doi.org/10.1038/nature06949
Lynch-Stieglitz, J., Valley, S. G., & Schmidt, M. W. (2019). Temperature-dependent ocean–
atmosphere equilibration of carbon isotopes in surface and intermediate waters over the

121
deglaciation. Earth and Planetary Science Letters, 506, 466–475.
https://doi.org/10.1016/j.epsl.2018.11.024
Marchitto, T. M., Lehman, S. J., Ortiz, J. D., Fluckiger, J., & van Geen, A. (2007). Marine
Radiocarbon Evidence for the Mechanism of Deglacial Atmospheric CO2 Rise. Science,
316(5830), 1456–1459. https://doi.org/10.1126/science.1138679
Marcott, S. A., Bauska, T. K., Buizert, C., Steig, E. J., Rosen, J. L., Cuffey, K. M., et al. (2014).
Centennial-scale changes in the global carbon cycle during the last deglaciation. Nature,
514(7524), 616–619. https://doi.org/10.1038/nature13799
Martin, J. H. (1990). Glacial-interglacial CO 2 change: The Iron Hypothesis. Paleoceanography,
5(1), 1–13. https://doi.org/10.1029/PA005i001p00001
Martínez-Botí, M. A., Marino, G., Foster, G. L., Ziveri, P., Henehan, M. J., Rae, J. W. B., et al.
(2015). Boron isotope evidence for oceanic carbon dioxide leakage during the last
deglaciation. Nature, 518(7538), 219–222. https://doi.org/10.1038/nature14155
Martinez-Garcia, A., Sigman, D. M., Ren, H., Anderson, R. F., Straub, M., Hodell, D. A., et al.
(2014). Iron Fertilization of the Subantarctic Ocean During the Last Ice Age. Science,
343(6177), 1347–1350. https://doi.org/10.1126/science.1246848
McCave, I. N., Carter, L., & Hall, I. R. (2008). Glacial–interglacial changes in water mass
structure and flow in the SW Pacific Ocean. Quaternary Science Reviews, 27(19–20),
1886–1908. https://doi.org/10.1016/j.quascirev.2008.07.010
McManus, J. F., Francois, R., Gherardi, J.-M., Keigwin, L. D., & Brown-Leger, S. (2004).
Collapse and rapid resumption of Atlantic meridional circulation linked to deglacial
climate changes. Nature, 428(6985), 834–837. https://doi.org/10.1038/nature02494

122
Menviel, L., Joos, F., & Ritz, S. P. (2012). Simulating atmospheric CO2, 13C and the marine
carbon cycle during the Last Glacial–Interglacial cycle: possible role for a deepening of
the mean remineralization depth and an increase in the oceanic nutrient inventory.
Quaternary Science Reviews, 56, 46–68. https://doi.org/10.1016/j.quascirev.2012.09.012
Menviel, L., Mouchet, A., Meissner, K. J., Joos, F., & England, M. H. (2015). Impact of oceanic
circulation changes on atmospheric δ
13
CO 2: δ
13
CO 2. Global Biogeochemical Cycles,
29(11), 1944–1961. https://doi.org/10.1002/2015GB005207
Menviel, L., Yu, J., Joos, F., Mouchet, A., Meissner, K. J., & England, M. H. (2017). Poorly
ventilated deep ocean at the Last Glacial Maximum inferred from carbon isotopes: A
data-model comparison study: LGM δ
13
C. Paleoceanography, 32(1), 2–17.
https://doi.org/10.1002/2016PA003024
Menviel, L., Spence, P., Yu, J., Chamberlain, M. A., Matear, R. J., Meissner, K. J., & England,
M. H. (2018). Southern Hemisphere westerlies as a driver of the early deglacial
atmospheric CO2 rise. Nature Communications, 9(1). https://doi.org/10.1038/s41467-
018-04876-4
Middelburg, J., Soetaert, K., & Hagens, M. (2020). Ocean alkalinity, buffering and
biogeochemical processes (preprint). Oceanography.
https://doi.org/10.1002/essoar.10501337.2
Monnin, E., Indermuhle, A., Dallenbach, A., Fluckiger, J., Stauffer, B., Stocker, T. F., et al.
(2001). Atmospheric CO2 Concentrations over the Last Glacial Termination. Science,
291(5501), 112–114. https://doi.org/10.1126/science.291.5501.112

123
Moy, A. D., Palmer, M. R., Howard, W. R., Bijma, J., Cooper, M. J., Calvo, E., et al. (2019).
Varied contribution of the Southern Ocean to deglacial atmospheric CO2 rise. Nature
Geoscience, 12(12), 1006–1011. https://doi.org/10.1038/s41561-019-0473-9
Naik, S. S., Divakar Naidu, P., Foster, G. L., & Martínez-Botí, M. A. (2015). Tracing the
strength of the southwest monsoon using boron isotopes in the eastern Arabian Sea.
Geophysical Research Letters, 42(5), 1450–1458. https://doi.org/10.1002/2015GL063089
Nielsen, S. B., Jochum, M., Pedro, J. B., Eden, C., & Nuterman, R. (2019). Two‐Timescale
Carbon Cycle Response to an AMOC Collapse. Paleoceanography and
Paleoclimatology, 34(4), 511–523. https://doi.org/10.1029/2018PA003481
Nowaczyk, N. R., Frank, U., Kind, J., & Arz, H. W. (2013). A high-resolution paleointensity
stack of the past 14 to 68 ka from Black Sea sediments. Earth and Planetary Science
Letters, 384, 1–16. https://doi.org/10.1016/j.epsl.2013.09.028
Ödalen, M., Nycander, J., Oliver, K. I. C., Brodeau, L., & Ridgwell, A. (2018). The influence of
the ocean circulation state on ocean carbon storage and CO2
drawdown potential in an Earth system model. Biogeosciences, 15(5), 1367–1393.
https://doi.org/10.5194/bg-15-1367-2018
Olsen, A., Key, R. M., Van Heuven, S., Lauvset, S. K., Velo, A., Lin, X., et al. (2016). The
global ocean data analysis project version 2 (GLODAPv2) - An internally consistent data
product for the world ocean. Earth System Science Data, 8(2), 297–323.
https://doi.org/10.5194/essd-8-297-2016
Pahnke, K., & Zahn, R. (2005). Southern Hemisphere Water Mass Conversion Linked with
North Atlantic Climate Variability. Science, 307(5716), 1741–1746.
https://doi.org/10.1126/science.1102163

124
Palmer, M. R., & Pearson, P. N. (2003). A 23,000-Year Record of Surface Water pH and pCO2
in the Western Equatorial Pacific Ocean. Science, 300(5618), 480–482.
https://doi.org/10.1126/science.1080796
Palmer, M. R., Brummer, G. J., Cooper, M. J., Elderfield, H., Greaves, M. J., Reichart, G. J., et
al. (2010). Multi-proxy reconstruction of surface water pCO2 in the northern Arabian Sea
since 29ka. Earth and Planetary Science Letters, 295(1–2), 49–57.
https://doi.org/10.1016/j.epsl.2010.03.023
Palter, J. B., Sarmiento, J. L., Gnanadesikan, A., Simeon, J., & Slater, R. D. (2010). Fueling
export production: nutrient return pathways from the deep ocean and their dependence on
the Meridional Overturning Circulation. Biogeosciences, 7(11), 3549–3568.
https://doi.org/10.5194/bg-7-3549-2010
Pasquier, B., & Holzer, M. (2016). The plumbing of the global biological pump: Efficiency
control through leaks, pathways, and time scales: PLUMBING OF THE GLOBAL
BIOLOGICAL PUMP. Journal of Geophysical Research: Oceans, 121(8), 6367–6388.
https://doi.org/10.1002/2016JC011821
Pena, L. D., Goldstein, S. L., Hemming, S. R., Jones, K. M., Calvo, E., Pelejero, C., & Cacho, I.
(2013). Rapid changes in meridional advection of Southern Ocean intermediate waters to
the tropical Pacific during the last 30kyr. Earth and Planetary Science Letters, 368, 20–
32. https://doi.org/10.1016/j.epsl.2013.02.028
Petit, J. R., Jouzel, J., Raynaud, D., Barkov, N. I., Delaygue, G., Delmotte, M., et al. (1999).
Climate and atmospheric history of the past 420,000 years from the Vostok ice core,
Antarctica, 399, 8.

125
Putnam, A. E., Denton, G. H., Schaefer, J. M., Barrell, D. J. A., Andersen, B. G., Finkel, R. C., et
al. (2010). Glacier advance in southern middle-latitudes during the Antarctic Cold
Reversal. Nature Geoscience, 3(10), 700–704. https://doi.org/10.1038/ngeo962
Rae, J. W. B., Burke, A., Robinson, L. F., Adkins, J. F., Chen, T., Cole, C., et al. (2018). CO2
storage and release in the deep Southern Ocean on millennial to centennial timescales.
Nature, 562(7728), 569–573. https://doi.org/10.1038/s41586-018-0614-0
Rae, James W. B. (2018). Boron Isotopes in Foraminifera: Systematics, Biomineralisation, and
CO2 Reconstruction. In H. Marschall & G. Foster (Eds.), Boron Isotopes: The Fifth
Element (pp. 107–143). Cham: Springer International Publishing.
https://doi.org/10.1007/978-3-319-64666-4_5
Rae, James W.B., Foster, G. L., Schmidt, D. N., & Elliott, T. (2011). Boron isotopes and B/Ca in
benthic foraminifera: Proxies for the deep ocean carbonate system. Earth and Planetary
Science Letters, 302(3–4), 403–413. https://doi.org/10.1016/j.epsl.2010.12.034
Rafter, P. A., Herguera, J.-C., & Southon, J. R. (2018). Extreme lowering of deglacial seawater
radiocarbon recorded by both epifaunal and infaunal benthic foraminifera in a wood-
dated sediment core. Climate of the Past, 14(12), 1977–1989. https://doi.org/10.5194/cp-
14-1977-2018
Rafter, P. A., Carriquiry, J. D., Herguera, J., Hain, M. P., Solomon, E. A., & Southon, J. R.
(2019). Anomalous > 2000‐Year‐Old Surface Ocean Radiocarbon Age as Evidence for
Deglacial Geologic Carbon Release. Geophysical Research Letters, 46(23), 13950–
13960. https://doi.org/10.1029/2019GL085102
Raitzsch, M., Bijma, J., Benthien, A., Richter, K.-U., Steinhoefel, G., & Kučera, M. (2018).
Boron isotope-based seasonal paleo-pH reconstruction for the Southeast Atlantic – A

126
multispecies approach using habitat preference of planktonic foraminifera. Earth and
Planetary Science Letters, 487, 138–150. https://doi.org/10.1016/j.epsl.2018.02.002
Regenberg, M., Regenberg, A., Garbe-Schönberg, D., & Lea, D. W. (2014). Global dissolution
effects on planktonic foraminiferal Mg/Ca ratios controlled by the calcite-saturation state
of bottom waters. Paleoceanography, 29(3), 127–142.
https://doi.org/10.1002/2013PA002492
Reimer, P. J., Bard, E., Bayliss, A., Beck, J. W., Blackwell, P. G., Ramsey, C. B., et al. (2013).
IntCal13 and Maine13 radiocarbon age calibration curves 0-50000 years cal BP.
Radiocarbon, 55(4), 1869–1887. https://doi.org/10.2458/azu_js_rc.55.16947
Ridgwell, A., Hargreaves, J. C., Edwards, N. R., Annan, J. D., Lenton, T. M., Marsh, R., et al.
(2007). Marine geochemical data assimilation in an efficient Earth System Model of
global biogeochemical cycling. Biogeosciences, 4(1), 87–104.
Ridgwell, A. J., Watson, A. J., Maslin, M. A., & Kaplan, J. O. (2003). Implications of coral reef
buildup for the controls on atmospheric CO2 since the Last Glacial Maximum.
Paleoceanography, 18(4), 1083. https://doi.org/10.1029/2003PA000893
Romahn, S., Mackensen, A., Groeneveld, J., & Pätzold, J. (2014). Deglacial intermediate water
reorganization: new evidence from the Indian Ocean. Climate of the Past, 10(1), 293–
303. https://doi.org/10.5194/cp-10-293-2014
Ronge, T. A., Tiedemann, R., Lamy, F., Köhler, P., Alloway, B. V., De Pol-Holz, R., et al.
(2016). Radiocarbon constraints on the extent and evolution of the South Pacific glacial
carbon pool. Nature Communications, 7, 11487. https://doi.org/10.1038/ncomms11487
Ronge, Thomas A., Steph, S., Tiedemann, R., Prange, M., Merkel, U., Nürnberg, D., & Kuhn, G.
(2015). Pushing the boundaries: Glacial/interglacial variability of intermediate and deep

127
waters in the southwest Pacific over the last 350,000 years: Variability in SW-Pacific
AAIW and UCDW. Paleoceanography, 30(2), 23–38.
https://doi.org/10.1002/2014PA002727
Rose, K. A., Sikes, E. L., Guilderson, T. P., Shane, P., Hill, T. M., Zahn, R., & Spero, H. J.
(2010). Upper-ocean-to-atmosphere radiocarbon offsets imply fast deglacial carbon
dioxide release. Nature, 466(7310), 1093–1097. https://doi.org/10.1038/nature09288
Russell, A. D., Hönisch, B., Spero, H. J., & Lea, D. W. (2004). Effects of seawater carbonate ion
concentration and temperature on shell U, Mg, and Sr in cultured planktonic
foraminifera. Geochimica et Cosmochimica Acta, 68(21), 4347–4361.
https://doi.org/10.1016/j.gca.2004.03.013
Sanyal, Hemming, N. G., Broecker, W. S., Lea, D. W., Spero, H. J., & Hanson, G. N. (1996).
Oceanic pH control on the boron isotopic composition of foraminifera: Evidence from
culture experiments. Paleoceanography, 11(5), 513–517.
https://doi.org/10.1029/96PA01858
Sanyal, Bijma, J., Spero, H., & Lea, D. W. (2001). Empirical relationship between pH and the
boron isotopic composition of Globigerinoides sacculifer: Implications for the boron
isotope paleo-pH proxy. Paleoceanography, 16(5), 515–519.
https://doi.org/10.1029/2000PA000547
Sarmiento, J. L., Gruber, N., Brzezinski, M. A., & Dunne, J. P. (2004). High-latitude controls of
thermocline nutrients and low latitude biological productivity. Nature, 427(6969), 56–60.
https://doi.org/10.1038/nature02127

128
Schmitt, J., Schneider, R., Elsig, J., Leuenberger, D., Lourantou, A., Chappellaz, J., et al. (2012).
Carbon Isotope Constraints on the Deglacial CO2 Rise from Ice Cores. Science,
336(6082), 711–714. https://doi.org/10.1126/science.1217161
Schmittner, A., & Lund, D. C. (2015). Early deglacial Atlantic overturning decline and its role in
atmospheric CO 2 rise inferred from carbon isotopes (δ
13
C). Climate of the Past, 11(2),
135–152. https://doi.org/10.5194/cp-11-135-2015
Schmittner, A., Gruber, N., Mix, A. C., Key, R. M., Tagliabue, A., & Westberry, T. K. (2013).
carbon isotope ratios (δ13C) in the ocean, 24.
Shao, J., Stott, L. D., Gray, W. R., Greenop, R., Pecher, I., Neil, H. L., et al. (2019).
Atmosphere‐Ocean CO 2 Exchange Across the Last Deglaciation From the Boron Isotope
Proxy. Paleoceanography and Paleoclimatology, 34(10), 1650–1670.
https://doi.org/10.1029/2018PA003498
Siani, G., Michel, E., De Pol-Holz, R., DeVries, T., Lamy, F., Carel, M., et al. (2013). Carbon
isotope records reveal precise timing of enhanced Southern Ocean upwelling during the
last deglaciation. Nature Communications, 4(1). https://doi.org/10.1038/ncomms3758
Sigman, D. M., & Boyle, E. A. (2000). Glacial/interglacial variations in atmospheric carbon
dioxide. Nature, 407(6806), 859–869. https://doi.org/10.1038/35038000
Sikes, E. L., & Guilderson, T. P. (2016). Southwest Pacific Ocean surface reservoir ages since
the last glaciation: Circulation insights from multiple-core studies. Paleoceanography,
31(2), 298–310. https://doi.org/10.1002/2015PA002855
Sikes, E. L., Samson, C. R., Guilderson, T. P., & Howard, W. R. (2000). Old radiocarbon ages in
the southwest Pacific Ocean during the last glacial period and deglaciation, 405, 5.

129
Sikes, E. L., Elmore, A. C., Allen, K. A., Cook, M. S., & Guilderson, T. P. (2016). Glacial water
mass structure and rapid δ
18
O and δ
13
C changes during the last glacial termination in the
Southwest Pacific. Earth and Planetary Science Letters, 456, 87–97.
https://doi.org/10.1016/j.epsl.2016.09.043
Skinner, L., McCave, I. N., Carter, L., Fallon, S., Scrivner, A. E., & Primeau, F. (2015). Reduced
ventilation and enhanced magnitude of the deep Pacific carbon pool during the last
glacial period. Earth and Planetary Science Letters, 411, 45–52.
https://doi.org/10.1016/j.epsl.2014.11.024
Skinner, L. C., Primeau, F., Freeman, E., de la Fuente, M., Goodwin, P. A., Gottschalk, J., et al.
(2017). Radiocarbon constraints on the glacial ocean circulation and its impact on
atmospheric CO2. Nature Communications, 8(1). https://doi.org/10.1038/ncomms16010
Spero, H. J. (2002). The Cause of Carbon Isotope Minimum Events on Glacial Terminations.
Science, 296(5567), 522–525. https://doi.org/10.1126/science.1069401
Spero, Howard J., Bijma, J., Lea, D. W., & Bemis, B. E. (1997). Effect of seawater carbonate
concentration on foraminiferal carbon and oxygen isotopes. Nature, 390(6659), 497–500.
https://doi.org/10.1038/37333
Spero, Howard J., Mielke, K. M., Kalve, E. M., Lea, D. W., & Pak, D. K. (2003). Multispecies
approach to reconstructing eastern equatorial Pacific thermocline hydrography during the
past 360 kyr: PAST EASTERN EQUATORIAL PACIFIC HYDROGRAPHY.
Paleoceanography, 18(1), n/a-n/a. https://doi.org/10.1029/2002PA000814
Spratt, R. M., & Lisiecki, L. E. (2016). A Late Pleistocene sea level stack. Climate of the Past,
12(4), 1079–1092. https://doi.org/10.5194/cp-12-1079-2016

130
Stein, K., Timmermann, A., Kwon, E. Y., & Friedrich, T. (2020). Timing and magnitude of
Southern Ocean sea ice/carbon cycle feedbacks. Proceedings of the National Academy of
Sciences, 117(9), 4498–4504. https://doi.org/10.1073/pnas.1908670117
Stern, J. V., & Lisiecki, L. E. (2014). Termination 1 timing in radiocarbon-dated regional benthic
δ
18
O stacks. Paleoceanography, 29(12), 1127–1142.
https://doi.org/10.1002/2014PA002700
Stott, L., & Timmermann, A. (2011). Hypothesized Link Between Glacial/Interglacial
Atmospheric CO2 Cycles and Storage/Release of CO2-Rich Fluids From Deep-Sea
Sediments. In H. Rashid, L. Polyak, & E. Mosley-Thompson (Eds.), Geophysical
Monograph Series (Vol. 193, pp. 123–138). Washington, D. C.: American Geophysical
Union. https://doi.org/10.1029/2010GM001052
Stott, L., Southon, J., Timmermann, A., & Koutavas, A. (2009). Radiocarbon age anomaly at
intermediate water depth in the Pacific Ocean during the last deglaciation:
RADIOCARBON AGE ANOMALY. Paleoceanography, 24(2), n/a-n/a.
https://doi.org/10.1029/2008PA001690
Stott, L., Davy, B., Shao, J., Coffin, R., Pecher, I., Neil, H., et al. (2019). CO 2 Release From
Pockmarks on the Chatham Rise‐Bounty Trough at the Glacial Termination.
Paleoceanography and Paleoclimatology, 34(11), 1726–1743.
https://doi.org/10.1029/2019PA003674
Stott, L. D., Harazin, K. M., & Quintana Krupinski, N. B. (2019). Hydrothermal carbon release
to the ocean and atmosphere from the eastern equatorial Pacific during the last glacial
termination. Environmental Research Letters, 14(2), 025007.
https://doi.org/10.1088/1748-9326/aafe28

131
Studer, A. S., Sigman, D. M., Martínez-García, A., Benz, V., Winckler, G., Kuhn, G., et al.
(2015). Antarctic Zone nutrient conditions during the last two glacial cycles:
ANTARCTIC ZONE NUTRIENT CONDITIONS. Paleoceanography, 30(7), 845–862.
https://doi.org/10.1002/2014PA002745
Tapia, R., Nürnberg, D., Ho, S. L., Lamy, F., Ullermann, J., Gersonde, R., & Tiedemann, R.
(2019). Glacial differences of Southern Ocean Intermediate Waters in the Central South
Pacific. Quaternary Science Reviews, 208, 105–117.
https://doi.org/10.1016/j.quascirev.2019.01.016
Toggweiler, J. R. (1999). Variation of atmospheric CO 2 by ventilation of the ocean’s deepest
water. Paleoceanography, 14(5), 571–588. https://doi.org/10.1029/1999PA900033
Toggweiler, J. R., Russell, J. L., & Carson, S. R. (2006). Midlatitude westerlies, atmospheric CO
2 , and climate change during the ice ages: WESTERLIES AND CO 2 DURING THE
ICE AGES. Paleoceanography, 21(2), n/a-n/a. https://doi.org/10.1029/2005PA001154
Torres, M. A., Moosdorf, N., Hartmann, J., Adkins, J. F., & West, A. J. (2017). Glacial
weathering, sulfide oxidation, and global carbon cycle feedbacks. Proceedings of the
National Academy of Sciences, 114(33), 8716–8721.
https://doi.org/10.1073/pnas.1702953114
Verdy, A., & Mazloff, M. R. (2017). A data assimilating model for estimating Southern Ocean
biogeochemistry: SOUTHERN OCEAN BIOGEOCHEMISTRY ESTIMATE. Journal of
Geophysical Research: Oceans, 122(9), 6968–6988.
https://doi.org/10.1002/2016JC012650
Wallmann, K., & Aloisi, G. (2012). The Global Carbon Cycle: Geological Processes. In A. H.
Knoll, D. E. Canfield, & K. O. Konhauser (Eds.), Fundamentals of Geobiology (pp. 20–

132
35). Chichester, UK: John Wiley & Sons, Ltd.
https://doi.org/10.1002/9781118280874.ch3
Wang, X. T., Sigman, D. M., Prokopenko, M. G., Adkins, J. F., Robinson, L. F., Hines, S. K., et
al. (2017). Deep-sea coral evidence for lower Southern Ocean surface nitrate
concentrations during the last ice age. Proceedings of the National Academy of Sciences,
114(13), 3352–3357. https://doi.org/10.1073/pnas.1615718114
Yu, J., Thornalley, D. J. R., Rae, J. W. B., & McCave, N. I. (2013). Calibration and application
of B/Ca, Cd/Ca, and δ
11
B in Neogloboquadrina pachyderma (sinistral) to constrain CO2
uptake in the subpolar North Atlantic during the last deglaciation. Paleoceanography,
28(2), 237–252. https://doi.org/10.1002/palo.20024
Yu, J., Anderson, R. F., Jin, Z., Rae, J. W. B., Opdyke, B. N., & Eggins, S. M. (2013). Responses
of the deep ocean carbonate system to carbon reorganization during the Last Glacial–
interglacial cycle. Quaternary Science Reviews, 76, 39–52.
https://doi.org/10.1016/j.quascirev.2013.06.020
Zeebe, R. E., & Caldeira, K. (2008). Close mass balance of long-term carbon fluxes from ice-
core CO2 and ocean chemistry records. Nature Geoscience, 1(5), 312–315.
https://doi.org/10.1038/ngeo185
Zeebe, R. E., Wolf-Gladrow, D. A., Bijma, J., & Hönisch, B. (2003). Vital effects in
foraminifera do not compromise the use of δ
11
B as a paleo- p H indicator: Evidence
from modeling: CRETACEOUS MARINE TEMPERATURE EVOLUTION.
Paleoceanography, 18(2), n/a-n/a. https://doi.org/10.1029/2003PA000881
Zhang, J., & Quay, P. D. (n.d.). Carbon isotope fractionation during gas-water exchange and
dissolution of CO, 8.

133
Zhang, S., & Planavsky, N. J. (2020). Revisiting groundwater carbon fluxes to the ocean with
implications for the carbon cycle. Geology, 48(1), 67–71.
https://doi.org/10.1130/G46408.1
Zhao, N., & Keigwin, L. D. (2018). An atmospheric chronology for the glacial-deglacial Eastern
Equatorial Pacific. Nature Communications, 9(1). https://doi.org/10.1038/s41467-018-
05574-x
Zhao, N., Marchal, O., Keigwin, L., Amrhein, D., & Gebbie, G. (2018). A Synthesis of Deglacial
Deep-Sea Radiocarbon Records and Their (In)Consistency With Modern Ocean
Ventilation. Paleoceanography and Paleoclimatology, 33(2), 128–151.
https://doi.org/10.1002/2017PA003174
Ziegler, M., Diz, P., Hall, I. R., & Zahn, R. (2013). Millennial-scale changes in atmospheric CO2
levels linked to the Southern Ocean carbon isotope gradient and dust flux. Nature
Geoscience, 6(6), 457–461. https://doi.org/10.1038/ngeo1782

Abstract (if available)

Abstract Over the past few decades, Earth Scientists have begun to realize that Earth’s components—its solid part (crust, mantle and core), the oceans, the atmosphere and diverse ecosystems must be studied as a whole in order to tackle problems like global change in the past, present and future. The holistic philosophy of this emerging Earth System Science encourages interdisciplinary thinking and an exploration of the processes at the boundaries of the Earth’s components. Key concepts in System Sciences involve ‘reservoir’, ‘flux’, ‘timescale’, ‘feedback’, etc. Applications of these concepts in studies of Earth’s biogeochemical cycle have provided insights into the natural variability within the Earth system. For my Ph.D., I focused on the carbon cycle dynamics between warm interglacial and cold glacial climates of the late Pleistocene epoch. Of particular interest is how carbon flux at the boundaries of Earth’s reservoirs might have changed in response to physical, biogeochemical and geological processes. Four distinct projects that span atmosphere-ocean carbon exchange to geological carbon cycle imbalance induced ocean carbon inventory change are presented in this thesis. ❧ Chapters 2-4 present the effort to improve our understanding of what happened at the ocean and atmosphere boundary since the last glacial period. Chapters 2 & 4 are devoted to better understanding air-sea CO₂ gas exchange at the last glacial termination by using boron isotope and stable carbon isotope reconstructions as well as numerical modeling

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A trace metal investigation of the radiocarbon anomaly in the eastern equatorial Pacific during the last glacial termination

PDF

Abrupt climate variability during the last glacial and the late Holocene from the western tropical Pacific perspective

PDF

Lithium isotopes as a carbon cycle proxy: examining the effects of high-temperature weathering, understanding brachiopod archives, and validating Cenozoic records

$Isotopic fractionations in plant biomarker molecules with application to paleoclimate$

PDF

Isotopic fractionations in plant biomarker molecules with application to paleoclimate

PDF

Anaerobic iron cycling in an oxygen deficient zone

PDF

Recent variability in the hyrdological cycle of tropical Asia from oxygen isotopes of tree celulose

PDF

On protracted Laurentian rifting, continental freeboard, and cap carbonates in Neoproterozoic time

PDF

Carbonate dissolution at the seafloor: fluxes and drivers from a novel in situ porewater sampler

PDF

Carbonate geochemistry in primary, diagenetic and biological systems

PDF

Great Salt Lake ooids: insights into rate of formation, potential as paleoenvironmental archives, and biogenicity

PDF

Evaluation of the mechanism of variability of the delta18O in alpha-cellulose of tropical deciduous tree pinus kesiya: conventional models and new thinking

PDF

How open ocean calcifiers broke the link between large igneous provinces and mass extinctions

$Germanium-silicon fractionation in a continental shelf environment: insights from the northern Gulf of Mexico$

PDF

Germanium-silicon fractionation in a continental shelf environment: insights from the northern Gulf of Mexico

PDF

Concentration and size partitioning of trace metals in surface waters of the global ocean and storm runoff

PDF

The connection of the phosphorus cycle to diazotrophs and nitrogen fixation

PDF

Flowstone ideograms: deciphering the climate messages of Asian speleothems

PDF

Investigations on marine metal cycling through a global expedition, a wildfire survey, and a viral infection

PDF

Tracking fluctuations in the eastern tropical north Pacific oxygen minimum zone: a high-resolution geochemical evaluation of laminated sediments along western North America

PDF

Sulfur isotope geochemistry and the end Permian mass extinction

PDF

Calcite and aragonite dissolution in seawater: kinetics, mechanisms, and fluxes in the North Pacific

Asset Metadata

Creator Shao, Jun (author)

Core Title New insights into glacial-interglacial carbon cycle: multi-proxy and numerical modeling

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Geological Sciences

Publication Date 11/18/2020

Defense Date 07/09/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag carbon cycle,glacial-interglacial cycle,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Stott, Lowell (committee chair), Capone, Douglas (committee member), John, Seth (committee member)

Creator Email junshao@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-394620

Unique identifier UC11666150

Identifier etd-ShaoJun-9129.pdf (filename),usctheses-c89-394620 (legacy record id)

Legacy Identifier etd-ShaoJun-9129.pdf

Dmrecord 394620

Document Type Dissertation

Rights Shao, Jun

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA