of 37
Shallow Calcium Carbonate Cycling in the
North Pacific
Ocean
Adam V. Subhas
1
, Sijia Dong
2
, John
D. Naviaux
3
, Nick E. Rollins
4
, Patrizia Ziveri
5
,6
, William Gray
7
,
James W.B. Rae
8
, Xuewu Liu
9
, Robert H. Byrne
9
, Sang Chen
10
, Christopher Moore
9
, Loraine Martell
-
Bonet
9
, Zvi Steiner
1
1
, Gilad Antler
1
2
, Huanting Hu
10
, Abby Lunstrum
4
, Yi Hou
1
3
, Nathaniel
Kemnitz
4
,
Johnny Stutsman
1
4
, Sven Pallacks
5
, Mathilde Dugenne
1
5
, Paul D. Quay
1
4
, William M. Berelson
4
, and Jess
F. Adkins
2
1
Dep
artment
of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution,
Woods Hole, MA, USA.
2
Division of Geological
and Planetary Sciences, California Institute of
Technology, Pasadena, CA, USA.
3
Four Twenty Seven Inc., Sunnyvale, CA, USA
.
4
Department
of Earth Sciences, University of Southern California, Los Angeles, CA, USA.
5
Institut de
Ciencia I Tecnologia Ambientals
, Universitat Autonoma de Barcelona, Barcelona, Spain.
5
Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain
7
Laboratorie des Sciences du Climat et de l’Environnement,
(LSCE/IPSL)
,
Universite Paris
-
Saclay,
Gif
-
sur
-
Yvette,
France
8
School of Earth and Environmental Sciences, University of St
Andrews, St Andrews, Scotland
9
College of Marine Science, University of South Florida, St.
Petersburg, FL, USA
10
School of Oceanography, Shanghai Jiao Tong University, Shanghai,
China
1
1
GEOMAR,
Helmholtz Centre for Ocean Research Kiel, Kiel,
Germany
1
2
Department of
Earth and Environmental Sciences, Ben
-
Gurion University, Israel
1
3
Department of Earth,
Environmental, and Planetary Sciences, Rice University, Houston, TX, USA
1
4
School of
Ocea
nography, University of Washington, Seattle, WA, USA
1
5
School of Ocean and Earth
Science and Technology, University of Hawaii, HI, USA
Correspond
ing
author:
Adam V. Subhas
(
asubhas@whoi.edu)
Key Points:
High resolu
tion carbonate chemistry,
δ
13
C
-
DIC, and particle flux measurements in the
NE Pacific sheds light on the upper ocean
calcium carbonate and
alkalinity cycle
s
.
Based on this
sampling c
ampaign, there is
evidence for
substantial
CaCO
3
dissolution
in
the
mesopelagic zone above the saturation horizon
.
Dissolution experiment
s
, o
bservations
,
and modeling suggest that shallow
CaCO
3
dissolution
is coupled to the consumption of organic carbon, through
a combination of
zooplankton grazing and oxic respiration wit
hin
particle
microenvironments
.
Accepted
Article
This article has been accepted for publication and undergone full peer review but has not been through
the copyediting, typesetting, pagination and proofreading process, which may lead to differences between
this version and the
Version of Record
. Please cite this article as
doi: 10.1029/2022GB007388
.
This article is protected by copyright. All rights reserved.
Accepted
Article
This article is protected by copyright. All rights reserved.
Abstract
The cycling of biologically produced calcium carbonate (CaCO
3
) in the ocean is a
fundamental component of the global carbon cycle
.
Here, w
e present
experimental
determinations of
in situ
coccolith and foraminiferal calcite dissolution rates.
We combine these
rates with
solid phase fluxes
,
dissolved tracers
, and historical data to constrain the alkalinity
cycle in the
shallow
North Pacific Ocean
.
The
in situ
d
issolution rates of coccolitho
phores
demonstrate a nonlinear dependence on saturation state
. Dissolution
rates
of all three major
calcifying groups (coccoliths, foraminifera, and
aragonitic
pteropods)
are
too slow to explain
the
patterns
of
both
CaCO
3
sinking
flux
and alkalinity regeneration
in the N
orth
Pacific.
Using
a
combination of dissolved and solid
-
phase tracers,
we
document
a significant dissolution signal in
seawater supersaturated for calcite
.
Driving
CaCO
3
dissolution
with a
combination of ambient
saturation state and oxygen consumption simultaneously explain
s
solid
-
phase CaCO
3
flux
profiles and patterns of alkalinity regeneration across the entire N. Pacific basin.
We
do not
need
to
invoke
the presence of
carbonate phases wi
th
higher
solubilities
.
Instead, b
iomineralization
and
metabolic processes
intimately associat
e
the acid (CO
2
) and the base (CaCO
3
) in the same
particles
,
driving the coupled shallow remineralization of
organic carbon
and CaCO
3
.
The
linkage of these processes likely occurs through a combination of dissolution due to zooplankton
grazing
and microbial aerobic respiration within
degrading particle aggregates.
Th
e coupling of
these
cycle
s
acts as a major filter on the export of both organic and inorganic carbon to the deep
ocean.
Plain Language Summary
The marine carbon cycle is made of organic carbon and calcium carbonate
(CaCO
3
)
components
.
While
the organic carbon cycle has received
much attention, the CaCO
3
cycle is relatively
understudied.
Through a dedicated research expedition to the North Pacific Ocean, w
e
demonstrate here a coupling of these two cycles, stemming from the fact that
all organisms that
produce
CaCO
3
also produce
in
timately associated
organic carbon
. We suggest that the
mechanisms responsible for the formation and sinking of
organic carbon
particles in the ocean
are likely as important for
CaCO
3
export
, and that the
respiration of
organic carbon
is responsible
for th
e dissolution of
a substantial portion of
CaCO
3
in the upper ocean.
1
.1
Introduction
The marine calcium carbonate (CaCO
3
) cycle is integral to the global carbon cycle. The
production of biogenic
CaCO
3
tends to raise atmospheric CO
2
due to consumption of
surface
ocean alkalinity, while the ballasting of organic matter and export into the deep ocean provided
by this material tends to lower CO
2
(De La Rocha et al., 2008, Passow et al., 2006, Klaas and
Archer, 2002)
. In addition to these roles, solid CaCO
3
i
s crucial to the neutralization of CO
2
through its dissolution and associated production of ocean alkalinity (
Archer et al., 1998
).
Despite consensus on the general dynamics of marine CaCO
3
cycling,
rates
of CaCO
3
production
and dissolution are
poorly constrained
(Dunne et al., 2012
,
Berelson et al., 2007, Battaglia et al.,
2016). If a substantial amount of alkalinity is regenerated in the shallow ocean
where
precipitation is thermodynamically favored
, then the
traditional relationship between
w
ater
column CaCO
3
dissolution
rates and
mineral
saturation state
must be reexamined.
T
he formation and dissolution of calcium carbonate minerals is
canonically
described as
a
function of seawater saturation state
(Ω = [Ca
2+
][CO
3
2
-
]/K’
sp
), where K’
sp
is the
in situ apparent
solubility product of the CaCO
3
mineral of interest (e.g. calcite or aragonite
). The dissolution
Accepted
Article
This article is protected by copyright. All rights reserved.
rate
R
is empirically related to Ω
using the equation
R = k(1
-
Ω)
n
,
where
k
is the specific
dissolution rate constant
,
and
n
is a fit parameter describing the nonlinearity of dissolution rate as
a function of
undersaturation
(Keir, 1980). In brief, calcite dissolution is a highly nonlinear
function of seawater undersaturation
across a wide range of saturation states
, a
n observa
tion that
dates back to early experiments by Peterson (1966) and
Berner and Morse (1974).
This high degree of nonlinearity can be explained by applying kinetic crystal growth
models to ca
lcite dissolution, as summarized in Adkins et al. (2021)
for inorganic calcite
.
As
saturation state decreases,
inorganic calcite
dissolution goes through
distinct
rate transitions at
“critical” Omega values (Ω
crit
,
Naviaux et al., 2019b
, Dong et al., 202
0a
).
At colder
temperatures
,
a single break in slope in log(rate) versus log(1
-
Ω)
is observed
at an Ω
crit
~0.75
-
Figure 1:
A schematic of
surface area
-
normalized
log(rate) as a function of log(1
-
Ω)
at 5°C,
for
inorganic calcite (
black
) and aragonite and
biogenic calcites (
red
), adapted from Naviaux et al.
(2019a,b), Subhas et al. (2018), and Dong et al.
(2019). The dashed line denotes Ω
crit
, the value of
Ω at which the dissolution mechanism shifts from
near
-
equilibrium step
-
edge retreat to far
-
from
-
equilibrium 2
-
dimensional
, homogenous
dissolution.
All
biogenic calcites, and aragonite,
demonstrate a rate transition at the same Ω
crit
as
inorganic calcit
e, but
demonstrate a shallower far
-
from
-
equilibrium slope compared to inorganic
calcite
.
The larger
red
envelope is intended
to
illustrate
the larger range in
dissolution
rates of
biogenic materials.
Logarithmically spaced ticks
for 1
-
Ω are shown at the top of the plot for clarity.
Accepted
Article
This article is protected by copyright. All rights reserved.
0.8
, marking the transition between dissolution at step
-
edges and dissolution on the 2D mineral
surface
(
Figure 1
black
envelope
,
Peterson, 1966,
Naviaux et al., 2019a,b, Dong et al., 2019).
Practi
c
ally,
two dissolution regimes mean
that
calcite dissolution in seawater is best described by
two rate equations
,
uniquely
defined between 1>Ω> Ω
crit
and Ω
crit
>Ω>0
,
each
with distinct
k
-
n
pairs
. Near equil
ibrium,
dissolution rate is slow and relatively
insensitive to saturation state
with
a small
k
and n<1.
F
ar from equilibrium
,
dissolution rapidly increases as a function of
undersaturation
, leading to a large
k
and
n
>>1
.
S
everal
in situ
water column
dissolution
experiments
with inorganic calcite have documented this sharp increase in dissolution rate
below
Ω
crit
(
Peterson, 1966,
Honjo and Erez, 1978,
Naviaux et al., 2019a).
Like inorganic calcite, biogenic
calcites and aragonite demonstrate
a
rate tra
nsition at the
same Ω
crit
as inorganic calcite (Subhas et al., 2018, Dong et al., 2019).
B
iogenic
calcite
dissolution rates
demonstrate
k
and
n
values that
are distinct
from the inorganic calcite curve
(
Figure 1
,
red
envelope
).
Coccolith
dissolution rates
appear to
be
have similarly to calcite near
equilibrium, but are
less sensitive to saturation state
far
-
from
-
equilibrium
(
n
=
2
.2
,
Subhas et al.,
2018)
.
Foraminifera
also demonstrate a far
-
from
-
equilibrium
n
~2
. They
appear to dissolve faster
than coccoliths n
ormalized by both mass and by surface area,
suggesting that foraminife
ra
require their own specific
k
values
(
Subhas et al., 2018, Subhas et al., 2019a,
Gehlen et al.,
2005
). Aragonite dissolves more slowly than biogenic calcite, and also has the lowest
n
~1.8
(Dong et al., 2019).
Compared to inorganic calcite, t
he lower sensitivity of biogenic calcite
dissolution to saturation state
, particularly farther from equilibrium,
is potentially
due to the
presence of organic
templates and
matrices within the calci
te structure (
e.g.
Subhas et al., 2018
,
Walker and Langer, 2021
)
.
Differences in specific
k
and
n
values aside
, it appears that
all
calcium carbonates
dissolve slowly
near equilibrium, and
only
rapidly increase their dissolution
rates
once
Ω drops below Ω
crit
~0.8
for the colder temperatures of the modern water column
.
Contrary to
these experiments, most
oceanographic
observations of
CaCO
3
dissolution
argue for a
large dissolution flux in the shallow ocean
, where waters are typically supersaturated,
follow
ed by relatively little dissolution in the deep ocean
, where waters are more deeply
undersaturated
.
The mechanisms driving this large shallow dissolution flux are confusing,
because they cannot be explained by
rate relationships relating saturation state t
o dissolution rate
.
This conundrum was posed explicitly by
Milliman et al. (1999)
, who
postulated
that balancing
global
CaCO
3
production and burial
required
a large shallow dissolution flux
.
This shallow
dissolution flux was then matched to the appearance of excess alkalinity in the Pacific (Feely et
al., 2002) and Atlantic basins (Chung et al., 2003).
Later, some
researchers
argued
that the
appearance of excess alkalinity
above the calcite s
aturation horizon
is complicated by water mass
transport and mixing processes (Friis et al., 2006
, Battaglia et al., 2016, Carter et al., 2020
).
Recen
t
modeling efforts have shown that
a supersaturated
alkalinity signal is due to a
combination of circulation and dissolution
, lending support to the
presence of a
shallow
CaCO
3
cycle
(
Carter et al., 2020, Sulpis et al., 2021)
.
N
umerous observations of
particulate
CaCO
3
loss in the upper water column serve
as a
complement to the appearance of dissolved alkalinity
(Bishop et al., 1980, Bishop et al., 1986,
Troy et al., 1997, Milliman et al., 1999, Wong et al., 1999, Bishop and Wood, 2008, Timothy et
al., 2013, Barrett et al., 2014, Dong et al., 2019).
Some a
uthors have
attempted to explain these
observations
through
the production and dissolution of more soluble CaCO
3
polymorphs such as
aragonite
(Feely et al., 2002, Buitenhuis et al., 2019)
,
and
pelagic fish
-
produced Mg
-
calcite, and
amorphous carbonates
(Wil
son et al., 2009, Woosley et al., 2012).
The
total production of these
Accepted
Article
This article is protected by copyright. All rights reserved.
“missing”
highly soluble
phases
remains
poorly constrained
,
and therefore their contribution to
the shallow CaCO
3
cycle remains
an open question
.
An alternate hypothesis is that
t
he
aggregation of
marine snow
and
the
grazing activity of
zooplankton
create hotspots of calcium carbonate dissolution
(Milliman et al., 1999, Bishop,
1980). The interiors of marine snow aggregates
and zooplankton guts
both exhibit
lower pH
, and
therefore lo
wer
Ω,
than ambient seawater
, due to the production of
respiratory CO
2
and digestive
acids
(Alldredge and Cohen, 1987, Pond
et al.
, 199
5)
.
T
hese grazing and aggregation
processe
s
are
recognized
as fundamental
to total
carbon export
from the
euphotic zone
through the
mesopelagic
(Henson et al., 2019, Boyd et al., 2019, Grabowski et al., 2019
)
, and
they
would
allow more abundant phases like coccolith and foraminiferal calcite to play a role in shallow
water column dissolution.
Here
, we present new
results and analyses from the CDisK
-
IV research expedition,
occupying five stations in the North Pa
cific from Hawaii to Alaska in the summer of 2017.
We
present new
in situ
dissolution
rate measurements of coccolith and foraminiferal calcite using a
13
C
-
tracer approach (
Subhas et al., 2015,
Naviaux et al., 2019
)
. The quantification of
in situ
dissolution rates across a wide range of saturation states allows us to
estimate
the magnitude
of
ambient
Ω
-
driven
water column
CaCO
3
dissolution
, with
explici
t
contributions from
the
th
ree
main
calcifier groups (coccoliths, for
a
minifera, and pteropods) t
o the total CaCO
3
dissolution
flux.
In order to assess the significance of dissolution
above the saturation horizon
, we present
new analyses of dissolved water column carbonate chemistry and oxygen data
that
complement
the canonical excess alkalinity
approach
.
Finally, we attempt to reconcile these
in situ
dissolution
rate measurements
and our own analyses
with
historical
observations of a large, shallow
dissolution cycle in the North Pacific ocean.
2 Materials and Methods
2.1 Study Area
Table 1
: Station locations and flux data for the CDisK
-
IV cruise.
Floating sediment trap
PIC
fluxes
(F)
are in units of mmol m
-
2
d
-
1
, subscript is trap depth.
PIC fluxes, and the fraction of
total calcium carbonate as aragonite
(
f
arag
)
,
are from Dong et al. (201
9).
The fraction of calcite
is calculated as 1
-
f
arag
.
The
mixed layer
calcite inventory is further split into coccolith and
foraminifer
al fractions
, with
f
cocco
representing the coccolith
calcite
fraction (
Ziveri et al.
,
in
revision).
Station 1 was located at Station ALOHA (22 45’N, 158’ W).
Station 5 was located
near Ocean Station Papa (OSP; 50 06’N, 144 54’ W).
Accepted
Article
This article is protected by copyright. All rights reserved.
The North Pacific is a key region for constraining CaCO
3
cycling, due
to the large portion
of the water column that is undersaturated for calcite and aragonite (Figure 2A), and the
foundational studies on changes in saturation horizon and degree of carbonate dissolution
performed there (Peterson, 1966, Feely et al., 2002, Fr
iis et al., 2006, Sabine et al., 2004).
We
conducted the CDisK
-
IV cruise on the R/V Kilo Moana in the North Pacific Ocean during
August 2017, on a transect from Honolulu, HI to Seward, AK.
At five survey stations (Fig. 2
gray points, Table 1) we measured
an extensive range in saturation states (Ω, defined as
[Ca
2+
][CO
3
2
-
]/K’
sp
, where K’
sp
is the apparent solubility of the carbonate mineral; Fig. 2A), and
established a spatial gradient in CaCO
3
production ecology. Quantitative CaCO
3
mineralogy,
ecology, sta
nding stock, production, and export flux measurements along this transect are
presented elsewhere
(Dong et al., 2019, Ziveri et al., in revision).
We complemented these solid
-
phase measurements by sampling the full water column for its dissolved inorganic
carbon (DIC),
total alkalinity (Alk), pH, and δ
13
C of DIC; and deployed custom
-
built
in situ
incubators to
measure biogenic calcite dissolution rates as a function of water column chemistry. The
dissolution rates of aragonite and inorganic calcite are pres
ented in detail elsewhere (Dong et al.,
2019, Naviaux et al., 2019a).
2.1 Sediment Traps
Figure
2
Carbonate chemistry of the upper 1000m of the North Pacific Ocean.
A
) Depth
-
latitude zonal section of in situ Ω calculated using Alk and DIC data from the
GLODAPv2.2019 database. Color bar indicates Ω
calcite
. Red and black dashed lines denote the
aragonit
e and calcite saturation horizons (defined as Ω=1), respectively. Gray points indicate
sampling locations during our 2017 research expedition (Table 1). Yellow stars indicate the
location of NPIW (main text for details).
B
) a calculation of
Ω
met
for calcite
along the section in
A
.
The 0.75 calcite “isosat” is shown to demonstrate the potential for deep undersaturation
within
confined environments
in the upper water column.
Accepted
Article
This article is protected by copyright. All rights reserved.
Sediment trap protocols can be found in companion papers by Subhas et al. (2019a) and
Dong et al. (2019)
and are outlined briefly here. Two sediment traps positioned at 100 and 200
meters on a single line were deployed at each station as a floating array. Each trap consisted of
12 polycarbonate particle interceptor tubes (70 cm long, 10 cm diameter) mounted
on a circular
frame
, with
a baffle grid
to screen for macrofauna
(
Haskell et al., 2016,
Fig
S
1
for picture)
.
Sinking material was funneled into a 50 mL falcon tube at the base of each tube. Falcon tubes
were filled with 30 mL of buffer solution, prepared
following US JGOFS protocol. Six random
tubes were combined, and “swimmers'' (mostly large copepods, amphipods and larvae) were
picked out. The samples were filtered onto a pre
-
weighed glass fiber filter (Whatman glass GF/F,
1825
-
047), which was then used
to calculate mass flux and mineralogy via X
-
ray diffraction.
Particulate inorganic carbon (PIC) and total carbon fluxes were calculated using established
methods (Dong et al., 2019).
2.2 Dissolved Carbonate System and Nutrient Measurements
Methods for
total
a
lkalinity, pH, δ
13
C
DIC
, silica, and nutrient measurements are well
established, and have been reported previously (Naviaux et al., 2019a, Dong et al., 2019). DIC
and secondary measurements of δ
13
C were measured on Niskin rosette samples using a cus
tom
Picarro
-
based system developed at the University of South Florida (USF)
and at
University of
Southern California (
USC
)
. Samples were collected in 300 ml biological oxygen demand (BOD)
bottles in a manner similar to that used for alkalinity samples, wit
h the exception that the bottle
had a 5 ml headspace and was poisoned using a saturated HgCl
2
solution. The DIC instrument
was housed in a 15 gallon insulated cooler that was capable of accommodating 10 samples
including a certified reference material (CRM
). After a brief tubing rinse, the sample was
pumped from the BOD bottle to a 20 ml glass bulb that was submerged in a thermostat
t
ed water
bath. An acid pump injected 3 ml of 17% H
3
PO
4
into the top of the glass bulb. A three
-
way valve
was then switched, al
lowing carrier gas to push the acidified sample into the purging tube, and
the CO
2
gas evolved was measured on a Picarro 2131
-
i instrument. The software first recorded a
baseline reading while N
2
was flowing in the system, and a threshold signal of 70 ppm triggered
the peak detection. Once a signal was detected, the peak integration continued until the signal
returned to the baseline. The DIC purging time was set to be longer than the required int
egration
time to ensure all CO
2
was detected. Total counts of both
12
C and
13
C were obtained. The
13
C
/
12
C ratio was calculated based on total peak integrations. Since ratios are based on total
12
C and
13
C, potential fractionation of carbon during purging
is avoided. Sample
13
C /
12
C ratios and total
DIC concentrations were linked to measured CRM solutions, and agree within error to Niskin
δ
13
C values measured by the CalTech/USC group (Naviaux et al., 2019a). We use the
Caltech/USC δ
13
C and USF DIC values h
ere.
2.3 Biogenic CaCO
3
dissolution experiments
Dissolution experiments were conducted
in situ
using modified Niskin incubators,
described in detail by
Naviaux et al. (2019a)
and
Dong et al. (2019)
. In this study, we
report
dissolution rates of
bleached,
13
C
-
labeled
E. huxleyi
liths.
A total of 20 coccolith dissolution
experiments were conducted at depths between 240
-
1000m at Stations 2
-
5 with temperatures
ranging from 2.4
-
4.8 C and Ω
calcite
from 0.96 to 0.67.
We conducted one experiment with a
planktic foraminiferal assemblage, cultured and
13
C
-
labeled as described in
Subhas et al. (2018)
.
Roughly
0.5
-
1.5 mg of labeled biogenic calcite was sealed in between 47mm
diameter
“Nuclepore” polycarbonate membrane filters (0.8 um pore size).
This
preparation
has no effect
Accepted
Article
This article is protected by copyright. All rights reserved.
on net dissolution rates, as documented by Naviaux et al. (2019a).
These packets were then
mounted inside the Niskin incubators. The incubators were hung on a hydrowire, sent down to
depth, and triggered closed.
As a closed system
, a pump
continuously circulated water
from the
housing holding the packets into the Niskin.
The Niskin reactors remained closed at depth for
24
58 h and were sampled for silica,
soluble reactive phosphorus (SRP)
, nitrate, alkalinity, pH,
and
δ
13
C
-
DIC upo
n recovery. Niskin data were quality checked by comparing SRP, silica, and
nitrate to ambient water
-
column values obtained via CTD/rosette deployments on the same
cruise.
Saturation states in the Niskin reactors were determined from Alk
-
pH pairs, input in
to
CO2SYS along with the temperature, salinity, depth, SRP, and silica concentrations at which the
reactor was deployed. Dissolution rates were calculated by taking the difference between the
final incubator and ambient water column
13
C/
12
C ratios, multipl
ied by the [DIC] and the mass of
seawater (1.126 kg) inside the incubators, and divided by the incubation time
(Naviaux et al.,
2019a)
. Biogenic materials were not 100% labeled
and
rates were scaled for the extent of isotope
labeling following Subhas et al
. (2018). Briefly, when the amount of
13
C in the dissolving
material is
greatly
enriched above natural abundance (
13
C/
12
C~0.01), isotope ratio differences
can be multiplied by a correction factor of (
R
s
+1)/
R
s
, where
R
s
is the
13
C/
12
C ratio of the
dissolving solid (i.e. a reduced form of Eq. 3 from Subhas et al.
,
2018). One batch of
13
C
-
labeled
E. huxleyi
was used for
all
dissolution rates shown here (
R
s
= 0.928)
, except for o
ne experiment
that
used
an older
batch (
R
s
= 20, Subhas et al. 2018).
The
planktic foraminifera assemblage
from Subhas et al. (2018) was used (
R
s
= 1.6).
Adjusting the rates
using these
R
s
values
gives
mass
-
normalized rates in units of g CaCO
3
g
-
1
d
-
1
.
Mass
-
normalized d
issolution rates were
further
divided
by the specific surface areas of
E. huxlyei
liths (105
,000
c
m
2
g
-
1
) and planktic
foraminifera (
4
3
,000 cm
2
g
-
1
, Subhas et al., 2018)
and the molar mass of calcium carbonate
(100
g mol
-
1
)
to generate
specific dissolution
rates
for each calcite type
in units of
mol CaCO
3
cm
-
2
d
-
1
.
2.4 1
-
D CaCO
3
flux and dissolution calculations
We constructed a
one
-
dimensional
model of particulate CaCO
3
sinking and dissolution,
combining our measurements of CaCO
3
export fluxes and export mineralogy (Table 1); ambient
seawater Ω; and aragonite, foraminiferal, and coccolith dissolution kinetics
(Table 2)
.
This
model
extends
the
analysis
of Dong et al. (2019)
to
coccolith, foraminifera, and aragonite
dissolution
at a
ll five survey stations
.
The model was initiated with our measured
100m
sinking
PIC
fluxes
,
partitioned into aragonite and calcite:
푎푟푎푔
=
푡표푡
푎푟푎푔
;
푐푎푙푐
=
푡표푡
(
1
푎푟푎푔
)
,
where F
tot
is the total PIC export flux at 100 meters, and
f
arag
is the fraction of total CaCO
3
as
aragonite measured in the sinking flux (Table 1,
Dong et al., 2019
). The dissolution flux
for each
phase
was then calculated at all depths for which we acquired water column dissolution data (at
least 24 vertical poi
nts, usually more) in a similar way to Dong et al. (2019):
푧푖
=
푧푖
1
(
1
푑푖푠푠
(
(
Ω
푧푖
1
)
)
1
)
,
where
F
zi
is the mineral flux at depth
z
i
,
R
diss
is the dissolution rate of the mineral and is a
function of the measured mineral saturation state at depth
z
i
-
1
, and w is the particle sinking rate.
Accepted
Article
This article is protected by copyright. All rights reserved.
For aragonite
fluxes (
F
arag
) and
dissolution (
R
arag
), w
e use the aragonite dissolution kinetic rate
law ge
nerated from in situ data on the same cruise (Dong et al., 2019, Table 2).
For calcite fluxes
(
F
calc
), w
e assume that the calcite rain is exclusively foraminifera and coccoliths (
f
cocco
+ f
foram
=
1
, Table 1
)
.
The dissolution rate of coccoliths,
R
cocco
, uses the parameters derived from the data
in Figure
3
, described in Table 2. The dissolution rate of foraminifera,
R
foram
, is assumed to
follow the same functional form as the dissolution rate of coccoliths,
R
cocco
, but dissolves a factor
of 1.6 faster,
when normalized by mass (
R
foram
= 1.6R
cocco
; Figure
3
, Table 2
, Section 3.
2
). This
assumption yields a calcite dissolution rate of:
푐푎푙푐
=
푓표푟푎푚
푓표푟푎푚
+
푐표푐푐표
푐표푐푐표
;
=
(
1
푐표푐푐표
)
1
.
6
푐표푐푐표
+
푐표푐푐표
푐표푐푐표
;
=
(
1
.
6
0
.
6
푐표푐푐표
)
푐표푐푐표
.
Errors in regression parameters were propagated through the dissolution model (shaded
envelopes in Figure 3
A
) by using the linearized log
-
log fits as “linear model” varia
bles in
MATLAB and the “predict” function.
Aragonite and calcite fluxes were then summed together to
calculate the total PIC flux at each depth.
The appearance of regenerated alkalinity, normalized to the mass of seawater (e.g. Feely
et al., 2002, Battagl
ia et al., 2016, Carter et al., 2014, Carter et al., 2020
)
is linked to the
dissolution
of particulate CaCO
3
.
Although Feely et al. (2002) divided their TA* tracer by 2 to
convert into moles of CaCO
3
dissolved, we leave all tracer results in units of alkalinity
equivalents per kg of seawater to distinguish between the solid and dissolved sides of the
process.
It follows that t
he
alkalinity regeneration
rate at each depth z,
R
Alk,zi
, in units of μmol k
g
-
1
yr
-
1
, is calculated as:
퐴푙푘
,
푧푖
=
2
푧푖
푑푖푠푠
(
(
Ω
푧푖
)
)
휌푤
,
W
here ρ is the density of seawater.
To account for along
-
isopycnal mixing, we
averag
ed
the
R
Alk
profiles
from
each of the five stations,
binned by potential density (σ
θ
) in
0.1 kg m
-
3
bins
.
This construction yields an approximation of the basin
-
wide alkalinity regeneration rate profile.
We
compared our
alkalinity regeneration
model to the water column tracer Alk*
-
derived rates (e.g. Feely et al., 2002).
We used
GLODAP v2.2
019 water chemistry data to
calculate both
TA* (Feely et al., 2002) and Alk* (Carter et al., 2014)
, and regressed these
quantities
against T
ime
T
ransit
D
istribution
ages
(Jeansson et al., 2021,
Supplemental
Information for calculation details). These TTD a
ges are thought to provide more accurate water
mass age estimates than apparent CFC ages, because they take into account a distribution of
mixing and other transport processes that can lead to artificially young apparent ages (Waugh et
al., 2006, Olsen et
al., 2021
, Sulpis et al., 2021
).
2.5 Calculation of a respiratory effect on Ω
To illustrate the potential effect of locally confined metabolic acidification on saturation
state, we assume that organic carbon and CaCO
3
are
closely associated and packaged
together in
marine particulate material. It follows that the consumption and degradation of this CaCO3
-
associated organic matter will drive a localized decrease in saturation state, providing an
increased driving force for dissolution. We further assume th
at this metabolic activity is aerobic,
and therefore limited by ambient oxygen concentrations.
We make a first attempt at constraining
the
potential for aerobic metabolism to drive undersaturation and dissolution
by “
metabolizing
Accepted
Article
This article is protected by copyright. All rights reserved.
all ambient oxygen
and pr
oportionally
modifying ambient
DIC and alkalinity values
in close
proximity to particulate CaCO
3
:
DIC
met
= DIC
m
eas
+ [O
2
]
m
eas
*R
CO
;
(1a)
Alk
met
= Alk
m
eas
-
[O
2
]
m
eas
*R
NO
;
(1b)
PO
4r
met
= PO
4m
eas
+ [O
2
]
m
eas
*R
PO
,
(1c)
where the subscripted “m
eas
quantities are the reported measurements in the GLODAP
v2.2019 database for the North Pacific basin (Olsen et al., 2019). R
CO
, R
NO
, and R
PO
are the
Redfield ratios of carbon to oxygen (0.688), nitrate to oxygen (0.0941), and phosphate to oxygen
(0.0059), r
espectively (Anderson and Sarmiento, 1994).
We
then
use DIC
met
, Alk
met
, and PO
4met
to
reca
l
culate
the Ω influenced by
in situ
metabolism, Ω
met
(Fig
2
B).
3
Results
and Discussion
3.1
Water Column Setting
All water column
, sediment trap, and dissolution rate
data
can be found in the BCO
-
DMO Data Repository for the CDisK
-
IV cruise (
https://www.bco
-
dmo.org/project/824992
).
Figure
2
A shows the
zonally averaged Pacific water
column saturation state, calculate
d
using
GLODAP v2.2019
DIC and
total
alkalinity data
.
F
or our CDisK
-
IV data, we use alkalinity and
pH pairs to calculate saturation states that result in systematically lower Ω values throughout the
water column
(
see
Naviaux et al., 2019a
for a thorough di
scussion of this topic
)
.
Nevertheless,
the trends that we observe are consistent with this zonal mean: a shoaling saturation horizon
moving northward,
along with
a collapsing of the distance between the aragonite and calcite
saturation horizons.
Figure
2
B
shows
the potential metabolic effect
on
Ω, demonstrating that
Ω
met
is
always lower than ambient Ω (Fig.
2
A)
due to the confined consumption of oxygen and
associated production of respiratory CO
2
and/or digestive acids
.
In general,
Ω
met
is about
the
same or
slightly
lower than the ambient aragonite saturation state, i.e. the black line in Fig
2
B
sits at or above the red line in Fig.
2
A
.
This calculation is independent of particle size or sinking
speed, and metabolism will influence the local saturation state
long as metabolic products (e.g.
respired CO
2
or digestive acids) are confined in close proximity to particulate CaCO
3
.
We further
investigate the
relevance
of
Ω
met
in section
3.
5
.
3.2 Dissolution rates
The
E. huxleyi
lith data (
purple
circle) and foraminiferal calcite data (
yellow
symbols) are
presented in
Figure 3
, and we interpret these data using the framework presented in Figure 1
.
Panels A and B
show
the same underlying rate data with different normalizations.
T
he surface
area
-
normalized rates
are included for
those readers most interested in the mechanistic aspects of
calcite dissolution
(Fig. 3A). C
hemical oceanographers may find the mass
-
normalized rate
s
most
useful
(Fig. 3B)
.
The c
occolith and foraminifera data are
mode
led
independently
, as they are
distinct forms of biogenic calcite with
well
-
documented
differences in
their
mineralogy
, surface
area,
and
dissolution
rate (Honjo and Erez, 1978, Keir, 1980, Subhas et al., 2018)
.
We
observe
a
Accepted
Article
This article is protected by copyright. All rights reserved.
break in slope in the coccolith data at
Ω
critt
=0.78 consistent with previous mechanistic
determinations
and interpretations
of Ω
crit
(Naviaux et al., 2019a,b,
Dong et al. 2019, Subhas et
al., 2018)
.
We fit
E. huxleyi
dissolution rates
in each regime (1>Ω>0.78 and 0.78>Ω>0)
to a rate
law of the form:
R = k(1
-
Ω)
n
.
These fits result in a
unique
set of of
k
-
n
values for eac
h regime.
Regression parameters for
E. huxleyi
are shown in Table 2
, w
ith
k
values normalized by both
mass
(
k
mass
)
and surface area
(
k
sa
)
.
The calculated
E. huxleyi
n
val
ues for both near
-
and far
-
from
-
equilibrium regimes are
within error
of laboratory values
(Table 2)
.
The
agreement between laboratory and
in situ n
values is consistent with previous comparisons between lab and
in situ
dissolution rates for
inorganic calcite and aragonite (Naviaux et al., 2019a, Dong et al., 2019).
Although
the
E.
huxleyi
n
values
are consistent between laboratory and
in situ
experiments, the
absolute
E.
huxleyi
in situ
dissolution
rates are
~10x
slower than those measured in the laboratory,
leading to
k
values that are about an order of magnitude
lower
than
lab
-
determined
k
values
(Table 2)
.
There are two
potential
drivers of
slow
in situ
rates
. First, low
in situ
temperature leads to
slower dissolution kinetics (Naviaux et al., 2019b), so we should expect
our
in situ
rates
(
determined
at
2.4
-
4.8°C) to be slower than laboratory rates (
determined at
21°C). Second,
ads
orption of
naturally occurring dissolved organic carbon (
DOC
)
onto the calcite surface
was
shown to
inhibit calcite dissolution in the ocean by a factor of ~4 (Naviaux et al., 2019a).
Combined, the
temperature
and DOC effects
explain the factor of 10 decre
ased
dissolution
rate
across the entire saturation range measured here (Fig. 3, Table 2).
The single previous
E. huxleyi
dissolution rate measurement from the Atlantic basin fits well within our more complete dataset,
suggesting that these rate parameters are broadly applicable to
E. huxleyi
dissolution rates in the
ocean (Figure 3 light green circle, Honjo and Erez, 1978).
Our si
n
gle planktic foraminiferal assemblage dissolution rate is broadly consistent with
the sparse and
foraminiferal
data from the North Atlantic
and the Pacific (Honjo and Erez, 1978,
Figure
3
:
Dissolution rates determined using our custom
-
built
in situ
reactors for
E. huxleyi
liths
(
purple
circles), and a
planktic foraminifera assemblage
(
yellow
5
-
pointed star).
A
Surface
-
area normalized dissolution rates plotted
in
logarithmic space
.
Previous
study data are
presented
for comparison (
H&E ’78 =
Honjo and Erez, 1978;
F ’08
=
Fukuhara et al., 2008
)
.
B
shows the same data as
A
plotted
in
linear space, normalized by mass
only
.
Regression
parameters for the fits are shown in Table 2.
The
yellow
curve
shows the coccolith regression
with
k
mas
s
scaled by a factor of 1.6
.