of 51
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
1
On the use of
dissolved oxygen
isotopologues as biogeochemical tracers in the Pacific Ocean
1
Boda Li,
1
Huanting Hu,
1,2*
William M. Berelson,
3
Jess F. Adkins
,
4
and Laurence Y. Yeung,
1,5*
2
1
Department of Earth, Environmental and Planetary Sciences, Rice University, Houston, TX
3
77005
4
2
School of Oceanography, Shanghai Jiao Tong University, Shanghai 200240, China
5
3
Department of Earth Science, University of Southern California, Los Angeles, CA 90089
6
4
Division of Geological and Planeta
ry Sciences, California Institute of Technology, Pasadena,
7
CA 91125
8
5
Department of Chemistry, Rice University, Houston, TX 77005
9
*
Correspondence:
haunting.hu@sjtu.edu.cn, lyeung@rice.edu
10
11
Abstract:
12
The isotopic composition of dissolved oxygen offers a family of potentially unique
13
tracer
s of respiration and transport
in the subsurface ocean
. Uncertainties in transport parameters
14
and isotopic fractionation factors, however, have limited the strength of the constraints offered
15
by
18
O/
16
O and
17
O/
16
O ratios
in dissolved oxygen. In particular, puzzlingly low
17
O/
16
O ratios
16
observed for some low
-oxygen samples have been difficult to explain. To
improve our
17
understanding of oxygen cycling in the ocean
’s interior, we investigated the systematics of
18
oxygen isotopologues in the subsurface Pacific using new data and a 2
-D isotopologue
-enabled
19
isopycnal reaction
-transport model. We
measured
18
O/
16
O and
17
O/
16
O ratios
, as well as the
20
“clumped”
18
O
18
O isotopologue in
the northeast Pacific, and compared
the results to
previously
21
published data. W
e find that transport and respiration rates constrained by O
2
concentrations
in
22
the oligotrophic Pacific
yield good measurement
-model agreement across all O
2
isotopologues
23
only when using a recently reported set of respiratory isotopologue fractionation fa
ctors
that
24
differ from those most often used for oxygen cycling in the ocean. These fractionation factors
25
imply that an elevated proportion of
17
O compared to
18
O in dissolved oxygen―i.e.,
its
triple
-
26
oxygen isotope composition―do
es
not uniquely reflect gross primary productivity and mixing.
27
For all oxygen isotopologues, transport, respiration, and photosynthesis comprise important parts
28
of their respective budgets. M
echanism
s of oxygen removal in the subsurface ocean are
29
discussed
.
30
8450 words
(main text)
31
561 words (
figure captions)
32
9 Figures
33
2 Tables
34
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
2
Plain Language Summary
35
The marine biosphere produces and consumes oxygen, and in doing so, imparts
36
fingerprints of photosynthesis and respiration (as well as other oxygen-
consuming processes) on
37
dissolved oxygen. These fingerprints are characterized by patterns in the abundances
of istable
38
isotopes
16
O,
17
O, and
18
O
versions of oxygen atoms that differ only in their atomic mass and
39
do not decay over time. Dissolved oxygen contains two oxygen atoms, and thus has six different
40
isotopic variants (e.g.,
16
O
16
O,
16
O
17
O, and
16
O
18
O, among others). We report new measurements
41
of five of these isotopic variants of molecular oxygen in the deep northeast Pacific Ocean and
42
explain their patterns using a simplified model of oxygen transport and consumption. We find,
43
contrary to prior reports, that all the isotopic fingerprints in the Pacific Ocean can be explained
44
under a common framework without invoking unusual oxygen production, consumption, or
45
transport mechanisms in the ocean. The results have implications for the use of oxygen isotopes
46
as tracers of marine productivity, respiration, and transport, providing field evidence consistent
47
with recent laboratory and theoretical studies of these isotopic fingerprints. Overall, the results
48
suggest that revision of canonical isotopic fingerprints is warranted, affecting our understanding
49
of biosphere productivity both in the present and past.
50
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
3
1. Introduction
51
1.1 Overview
52
Respiration and transport play fundamental roles in the chemical budgets of the
53
subsurface ocean.
Yet the myriad physical and biological processes comprising these budge
ts are
54
challenging to resolve.
Aerobic respiration
remineralizes
nutrients contained in organic matter,
55
removes dissolved O
2
from seawater, and produces
CO
2
, while large
-scale advection
56
redistributes these constituents globally
. Eddy diffusion operates in concert,
decreasing
57
concentration gradient
s on a
smaller scale. Finally, ventilation
at the surface
drives
the
58
concentration of dissolved O
2
and CO
2
to
ward
solubility
equilibrium with the atmosphere
. In
59
principle, b
ioactive tracers like nutrients, dissolved inorganic carbon (DIC) and oxygen
60
concentration can track features of these processes, but they cannot fully decouple respiration
61
and transport in deep sea
: distributions of nutrients and DIC vary stoichiometrically with oxygen
62
concent
ration, and hence none of them provide independent constraints on respiration or
63
transport (Takahashi et al., 1985)
.
64
Direct measurements of
respiration rates in
the deep sea are challenging
because
rates are
65
slow
and vary with location. A
pparent oxygen utilization (AOU) can be combined with mean
66
water mass
ventilation ages to estimate oxygen utilization rate
s (OURs)
(Feely et al., 2004)
.
67
Mean water mass ages can be
obtained through ocean circulation modeling (
Riley, 1951; Craig,
68
1969; Haine & Hall, 2002)
, radiometric
dating (
e.g.
,
14
C of DIC
) (Matsumoto, 2007; Koeve et al.,
69
2015)
or via the evolution of recently incorporated chemical constituents
(e.g.,
70
chlorofluorocarbons
) (Sonnerup, 2001)
. Implicit in the
OUR
method, however, are
the
71
assumptions
that
ventilation results in solubility equilibrium for O
2
and diffusive mixing
72
influences AOU and age in the same way
. The former is violated at h
igh latitudes
(Ito et al.,
73
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
4
2004)
, while the latter depends on the characteristic timescales of O
2
consumption and tracer
74
decay
. F
or example, a water mass at 50% O
2
saturation may have formed via closed
-system
75
respiration (i.e., OUR = AOU/age) or through mixing of many partially respirated water masses
76
(i.e., OUR =
Σ
f
i
AOU
i
/age
i
, where the AOU, age
, and mixing fraction
f
of each constituent water
77
mass
i
is not known)
(Bender, 1990)
. Consequently, other
tracers are needed to characterize the
78
marine oxygen budget.
79
The distribution of O
2
isotopologues
―the
δ
18
O value of O
2
in particular
has been
used
80
to disentangle respiration from transport, as
closed
-system respiration and mixing
effects are
81
distinguishable
(Bender, 1990; Quay et al., 1993; Levine et al., 2009)
. Still, this single additional
82
constraint has proven non-
unique in part because reproducing
δ
18
O-O
2
relationships requires
83
independent knowledge of isotopic fractionation factors in the deep ocean. These fractionation
84
factors may
vary widely
depending on environmental
conditions such as diffusi
ve limitation
85
(Bender, 1990)
or temperature (Stolper et al., 2018). Oxygen has three stable isotopes and O
2
has
86
six
stable isotopologues, however, so measurements of multiple O
2
isotopologues in the same
87
sample of seawater may
alleviate some of these uncer
tainties. Previous work on the
triple
-
88
oxygen isotope composition of O
2
(i.e.,
17
Δ values, which are
derived from
δ
18
O
and δ
17
O values
;
89
see Methods
) has focused on estimating
productivity in the surface ocean
(Bender, 2000; Juranek
90
& Quay, 2013; Luz & Barkan, 2000)
, although a few studies report
data
in the deep ocean.
91
Hendricks et al., (2005)
reported the most extensive
triple
-oxygen dataset from
the
92
subsurface equatorial Pacific
, which
revealed some surprising and unexplained observations.
93
First, non
-mon
otonically varying
17
Δ values
in the aphotic zone were observed, peaking at
94
moderate (
50 - 80%
) O
2
saturation
, which were interpreted as
a combination of
photosynthesis in
95
waters below the 1% light level and
entrainment of productive waters
. Second, some
17
Δ values
96
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
5
at <50% O
2
saturation were unusually low and only explainable
as two
-component mixtures
97
between extreme endmembers (e.g., surface water and a ~5% O
2
saturation
, highly respired
98
water mass). Moreove
r, recent experimental and theoretical evidence for
variable isotopic
99
fractionation
factors in the
17
Δ system
have offered alternate explanations for these data
(Stolper
100
et al., 2018; Ash et al., 2020)
. The sensitivity of
17
Δ values to photosynthesis, and uncerta
inty in
101
the relevant isotopic fractionation factors complicates their use as tracers of respiration and
102
mixing in the ocean.
103
The
distribution
s of
the remaining resolvable O
2
isotopologues in the ocean,
17
O
18
O and
104
18
O
18
O, ha
ve not yet been investigated. Yeung et al., (2015)
and Ash et al., (
2020)
showed tha
t
105
they are affected by photosynthesis and respiration, but in a manner different from either of the
106
singly-
substituted isotopologues
16
O
17
O and
16
O
18
O. In addition, physical fractionation
107
mechanisms such as diffusion
lead to unique isotopologue patterns that can be distinguished
108
from those of photosynthesis and respiration
(Li et a
l., 2019)
. This complementary sensitivity to
109
biogeochemical
fractionation
offered by these “clumped” isotopes
, constrained by preliminary
110
isotopic fractionation factors,
may facilitate a unified description of O
2
consumption and
111
transport in deep ocean
.
112
Here we present new O
2
concentration and isotopologue data fr
om
northeast Pacific
113
ocean, from the surface into the aphotic zone
. Bulk
-
(i.e., δ
18
O and
17
Δ values) and clumped
-
114
isotope (i.e., Δ
36
values reflecting
18
O
18
O) compositions were measured in the same dissolved O
2
115
samples for the first time to explore the potential utility of this suite of tracers to constrain
116
respiration and transport in the ocean.
We further
develop a
2 -D isopycnal
reaction
-transport
117
model
to examine the effects of respiration, photosynthesis, and transport
on O
2
isotopologue
118
patterns in the Pacific
. Finally, we revisit previous interpretations of deep
-
sea δ
18
O and
17
Δ data
119
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
6
and discuss the potential role of the recen
tly proposed superoxide
O
2
consumption pathway
120
(Sutherland et al., 2020a
) on the O
2
budget of the
ocean
.
121
122
2. Method
s:
123
2.1 Isotope terminology and systematics
124
Dissolved O
2
isotopologue
ratios
are reported as
δ
18
O,
17
∆,
and
36
values
for
16
O
17
O,
125
16
O
18
O, and
18
O
18
O, respectively. The definitions of these terms are based on
126
isotope/isotopologue ratios
R
. The de
nominator
for
R
is the most abundant isotope
or
127
isotopologue
(i.e.,
16
O or
16
O
16
O)
, while the
numerator
is the rare isotope
or isotopologue of
128
interest. For
example, the
18
O
18
O isotopologue has a
R
value defined by
129
푅푅
36
=
O
18
O
18
O
16
O
16
(1)
130
and is equal to the mol
ar concentration of
18
O
18
O divided by that of
16
O
16
O. Similar definitions
131
are made for
18
R
(i.e., [
18
O]/[
16
O]) and
17
R
(i.e., [
17
O]/[
16
O])
. The
δ
18
O,
17
∆,
and
36
values
are
132
defined as
133
훿훿
18
푂푂
=
(
푅푅
18
푠푠푠푠푠푠푠푠푠푠푠푠
푅푅
18
푠푠푎푎푎푎
1)
(2)
134
17
=
푙푙푙푙
푅푅
푠푠푠푠푠푠푠푠푠푠푠푠
17
푅푅
푠푠푎푎푎푎
17
0.518
×
푙푙푙푙
푅푅
푠푠푠푠푠푠푠푠푠푠푠푠
18
푅푅
푠푠푎푎푎푎
18
(3)
135
36
=
(
푅푅
푠푠푠푠푠푠푠푠푠푠푠푠
36
푅푅
푠푠푠푠푠푠푠푠 ℎ푠푠푠푠푠푠 푎푎푠푠
36
1)
(4)
136
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
7
with
δ
18
O and
36
values
reported in per mil (‰) and
17
values reported in parts per million
137
(ppm). T
he denominator
s relevant to
δ
18
O and
17
values are
the
R
values for atmospheric O
2
,
138
which has been recently re-
determined. Measurements at Rice University are consistent with the
139
lab reporting
Vienna Standard Mean Ocean Water
-2 (VSMOW2)
as
δ
18
O = –23.481‰ and
140
17
∆ = 204
ppm relative to
atmospheric O
2
(Wostbrock et al., 2020)
, which we will use
141
subsequently in this manuscript
. T
he denominator relevant to
36
values
is defined by
142
푅푅
푠푠푠푠푠푠푠푠 ℎ푎푎푠푠푠푠푎푎푠푠
36
=
푅푅
2
18
(5)
143
and represents the
18
O
18
O/
16
O
16
O ratio for a stochastic (random) distribution of isotopes within
144
the
sample
being analyzed.
145
Biogeochemical cycling leads to a broad range of potential isotopic compositions in
146
dissolved O
2
.
Isotopic fractionation due to respiration increases δ
18
O and Δ
36
values in the
147
residual O
2
(Guy et al., 1993; Ash et al., 2020)
, but its effects on
17
Δ
values have recently been
148
questioned:
early
work had initially suggested that
17
Δ values do not
change
during respiration
149
(Luz & Barkan, 2000; Angert et al., 2003; Helman et al., 2005; Luz & Barkan, 2005)
, but more
150
recent work has argued that they may decrease
(Stolper et al., 2018)
or increase (Ash et al., 2020)
151
in the residue fraction
. The two-
gyre model employed in this study uses the re
spiratory
152
fractionation factors from Ash et al., (2020), as they were also measured in the Rice University
153
laboratory and supported by first
-principles calculations of enzymatic active-
site analogues (
see
154
Table 1 and
Section 2.3.2). M
ixing relationships for δ
18
O values are generally linear,
but mixing
155
relationships for
17
Δ and Δ
36
values are curved and may not be monotonic with mixing fraction
156
(Miller, 2002; Eiler, 2007; Yeung et al., 2012)
. Nevertheless, the addition of photosynthetic O
2
157
into a dissolved pool of O
2
tends to decrease δ
18
O and
Δ
36
values (Guy et al., 1993; Quay et al.,
158
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
8
1993; Yeung et al., 2015)
and increase
17
Δ values
(Luz & Barkan, 2000, 2011)
. In principle, t
he
159
combination of unique is
otopic fractionation factors and mixing relationships for each isotopic
160
system leads to a system of independent constraints on the history of a water parcel
in the ocean
,
161
provided the fractionation factors for each biological process
are known. In the deep ocean, the
162
predominant mechanisms are respiration and mixing, although imprints of photosynthetic O
2
163
addition inherited from the surface ocean may
also be present
.
164
165
2.2 Water s
ampling and m
easurements of dissolved oxygen isotopologues
166
Ninety
-four
samples of dissolved O
2
for multi-
isotopologue analysis
were collected
167
during a transect from Hawai’i
to Alaska on R/V
Kilo Moana
during the
Carbonate D
issolution
168
In S
itu K
inetics project in
August 2017 (CDISK IV; see Fig. 1
). Depth profiles were obtained at
169
six sampling stations from the Hawai’i
Ocean Time Series site (
22.75 ̊N, 158 ̊W
) up to the Gulf
170
of Alaska (
60 ̊N, 149.3 ̊W
). Sampling methods follo
wed
th
ose
used previously for triple
-oxygen
171
isotope
analysis of dissolved oxygen (Reuer et al., 2007)
. Briefly,
Niskin bottles
from a
172
conductivity
-temperature
-depth rosette (
CTD)
were sampled
into pre
-evacuated (<10
-3
mbar),
173
pre
-poisoned glass bottles (
1L, 2L, and 5L sizes depending on dissolved O
2
concentration with
174
final H
gCl
2
c oncentrations
of
>20 μg/mL
seawater
) that were each fitted with a Louwers
-
175
Hanique 9mm I.D. high-
vacuum valve
. During transport and storage before and after sampling,
176
the side arm of the valve on each bottle was filled with water
, with all visible bubbles removed,
177
to minimize
air
contamination
.
178
Gas extraction and analysis occurred at Rice University. H
eadspace gases were
first
179
collected onto silica gel fingers, with two U
-shaped traps held at -
196
°
C upstream of the gel
180
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
9
finger to remove
residual CO
2
and water vapor
. The gases
were then
purified according to
181
methods described previously, using an Agilent 7890B Gas Chromatograph (GC) held at
80
°
C
182
to separate O
2
from Ar
, N
2
, and other trace gases (Yeung et al., 2016)
. The O
2
/Ar
ratio was
183
calculated using calibrated GC peak integration of O
2
and Ar
, an approach that has a
precision of
184
±4‰
(1σ)
and shows good agreement with manometric checks performed in a calibrated volume
185
(Ash et al., 2020)
. This ratio, relative to solubility equilibrium, was used to quantify dissolved O
2
186
saturation
as reported herein
; this biological supersaturation
normalizes against physical
187
disequilibria and is likely within several percent of the true dissolved O
2
saturation.
It also allows
188
one to focus on biological fractionation, reducing the scope of uncertainties relevant to
189
measurement
-model comparison. The purified O
2
was then analyzed
for its isotopic composition
190
on a high-
resolution Nu
Instruments
Perspective
IS
isotope
ratio mass spectrometer in dual-
inlet
191
mode. The pooled standard deviations for replicates within the CDISK IV dataset
were
±
0.20
,
192
±
5 ppm, and
±
0.045‰
(1σ)
for
δ
18
O,
17
Δ, and Δ
36
values, respectively.
193
194
2.3 Two
-Gyre
Model
195
Our 2
-D advection diffusion reaction model
is similar to th
e two
-gyre models
described
196
in Levine et al.,
(2009)
and Glover et al., (
2011)
. We
treat
the Pacific ocean circulation as
197
isopycnal in two dimensions
because of the
relative
ly strong mixing behavior within
isopycnal
s
198
and relatively weak
diapycnal exchange
( Bauer & Siedler, 1988; Glover et al., 2011)
. Ventilation
199
occurs at high latitudes (~
40°), where
air
-water equilibration
resets dissolved gas concentrations
200
and the isotop
ic signatures of dissolve
d O
2
. There, w
e apply an equilibrium solubility and
201
18
O/
16
O isotopic fractionation endmember corresponding to a sea surface
temperature
of
14 ̊C
202
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
10
(Benson & Krause, 1984; Li et al., 2019)
, which yields O
2
concentrations consistent with those
203
observed in the ventilation region (see Table 1)
. The isotopic results are not sensitive to th
e
204
particular choice of endmember, however, given the small overall isotopic fractionation at the
205
air
-sea interface.
206
2.3.1 Streamfunction
207
The 2
-D isopycnal slab model contains two gyres: a cyclonic gyre to the south and an
208
anticyclonic gyre to the north (
Fig. 2
and S1). The model’s advective geometry mimics
the
209
circulation observe
d in the North Pacific gyre between 0 ̊ and 40 ̊N
that is
composed of the
210
Kuroshio Current, North Equatorial Current, North Pacific Current, and California Current
, and
211
the circulation observed in the
South Pacific gyre between 0 ̊ and 40 ̊S
that is composed of the
212
Peru Current, South Equatorial Current and Antarctic Circumpolar Current.
The streamfunction
213
amplitude
,
A
, which controls the absolute strength of advection in the isopycnal, was set to be
214
consistent with the speed of the Kuroshio Extension and tuned using a cost function grid search
215
(Section 2.3.3)
. The streamfunction asymmetry (i.e., driving stronger and narrower western
216
boundary currents) was tuned to match the relative width of
the Kuroshio current. Additional
217
details, including the governing equations, can be found in Text S1.
218
219
2.3.2 Advection-
Diffusion
-Reaction
Equations
220
Isopycnal
advection
, diffusion
, and respiration
can be
generalized by eq. 6
:
221
휕휕휕휕
휕휕푠푠
=
2
(
퐾퐾퐾퐾
)
−∇∙
(
풖풖퐾퐾
)
−퐽퐽
(6)
222
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
11
where
C
is the concentration of the chemical species,
K
is the eddy diffusivity, and
u
i s the
223
advective velocity (
i.e.,
a vector
[
u, v
]), and
J
i s the
respiration rate
. At each time step, eq
. 6 was
224
solved for a grid representing the subtropical gyres of the Pacific
(e.g., 667 × 572 for
σ
θ
= 25.8 –
225
26.2) using inputs of
A
,
K
, and
J
and the
second upwind differencing method (Glover et al.,
226
2011)
.
227
In practice, t
he effects of transport and respiration are computed separately for
each time
228
step
, with the respiration term computed after transport for each isotopologue
. R
espiration rates
229
for the rare O
2
isotopologues a
re computed
relative to
that of
16
O
16
O using
the
ir fractionation
230
factor
s according to the equation below:
231
퐾퐾
푟푟푟푟푠푠
푠푠+1
,
푟푟푎푎 푟푟푟푟
=
퐾퐾
푚푚푎푎 푚푚
,
푛푛푠푠
푟푟푟푟푠푠
푠푠+1
,
푟푟푎푎 푟푟푟푟
−퐽퐽∗ 훼훼∗
푟푟푎푎 푟푟푟푟
퐾퐾
푚푚푎푎 푚푚
,
푛푛푠푠
푟푟푟푟푠푠
푠푠+1
,
푟푟푎푎 푟푟푟푟
/
퐾퐾
푚푚푎푎 푚푚
,
푛푛푠푠
푟푟푟푟푠푠
푠푠+1
,
푏푏푏푏푏푏푏푏
(7)
232
Here
,
α
is the isotopologue
-specific respiration
fractionation factor, i.e., the relative rate of
233
consumption compared to that for
16
O
16
O ( Table 1)
. The
α
values for
16
O
17
O/
16
O
16
O and
234
18
O
18
O/
16
O
16
O fractionation
are calculated from
the mass
-dependent exponents
θ
17/18,resp
and
235
θ
36/18,resp
of Ash et al., (2020) also shown in Table 1, using the equation:
236
훼훼
푚푚
=
� 훼훼
18
휃휃
( 8)
237
where
x
= 17 or 36. The
subscripts
no res
and
res
in eq. 7 denote the concentrations before and
238
after the
respiration
step
, respectively.
239
After updating the concentrations
of the rare
isotopologues
, the model
then update
s the
240
total O
2
concentration based on the total change in concentration of all isotopologues
. If
an O
2
241
isotopologue concentration is negative, it is set to zero
, and i
f the
16
O
16
O isotopologue
242
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
12
concentration
is zero, all rare O
2
isotopologues are also set
to zero. This
approach avoid
s
243
potential numerical instabilities associated with negative concentrations
.
244
2.3.3 Model Parameter Initialization
245
The
range of
parameters
considered for the isopycnal model are similar to those used for
246
the south subtropic
al Atlantic
(Levine et al., 2009)
. T
he
σ
θ
=
25.8 – 26.2 and 26.5 – 26.9
247
isopycnal layer
s w ere simulated
because the former
is near the median value for the
measured
248
samples, whereas the latter includes areas of lower O
2
saturation (i.e., 20 –
50%). Figure 3 shows
249
the annual
-mean depth and
O
2
concentration
for th
e
σ
θ
= 25.8 – 26.2 isopycnal
derived from the
250
World O
cean A
tlas
2013 (WOA 2013; Locarnini et al., 2013,
Zweng et al., 2013, Garcia et al.,
251
2014). The ventilation region for the
North Pacific subtropical gyre was set
by approximating the
252
areas o
f the isopycnal
that lie within
the
mixed layer
in the wintertime
or 50m below
; this
253
seasonal variation is significant
and has a strong influence on implied North Pacific respiration
254
rates. T
he ventilation region in the Sout
h Pacific subtropical gyre
was set
by selecting the areas
255
with depth <50m and >90% O
2
saturation
in the annual mean
because the seasonal variation has
256
a ne
gligible effect on the oxygen budget in the model North Pacific
. T
hese areas are shown
as
257
yellow rectangle
s i n Fig. 2 for the
σ
θ
= 25.8 – 26.2 surface and Fig. S
1 for the
σ
θ
= 2
6. 5 – 26.9
258
surface.
259
Mass exchange between the northern and southern gyre
s is
relatively small because the
260
stream function is equal to
zero
at the
ir boundary
. Therefore
,
K
at the boundary was set to be
261
larger, particularly at the eastern and western edges
of the
tropical Pacific (
see Table
s 2 & 3)
.
262
According to Cole et al., (
2015)
, the horizontal eddy diffusivity is elevated near the equator, with
263
the westernmost third having a horizontal eddy diffusivity
of 10
3.8
m
2
/s at the surface.
This eddy
264
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
13
diffusivity decreases with depth
, and
the data in
Cole et al., (
2015)
imply
a scaling factor of 0.8
265
for the
σ
θ
= 25.8 – 26.2 surface and 0.5 for the
σ
θ
= 26.5 – 26.9 surface
, which yield
K
= 5040
266
m
2
/s and 3150 m
2
/s, respectively,
for the westernmost region. The elevated equatorial eddy
267
diffusivities for the central and easternmost third were calculated similarly
and shown in Table 2.
268
A grid search was employed to optimize the stream function amplitude
A
, the isotropic eddy
269
diffusion coefficient
K
,
and the respiration rate
J
(see Text S1)
.
270
Due of the unique biogeochemistry of the Pacific, with upwelling and high productivity
271
near the equator and low productivity in the subtropical gyres, region
-specific
J
values
were
272
utilized (
J
equator
and
J
resp
for respiration within and outside the equatorial region, respectively)
.
273
The equatorial upwelling region was delineated as the easternmost two
-thirds of the area
274
between
-1000
km and 1000
km on the Y
-axis of Fig. 5,
which resembles
the
Pacific
cold tongue
.
275
The oligotrophic regions compris
ed t he rest of the model domain (±1000 km to ±4500 km on the
276
Y-axis of Fig. 5). The stream function amplitude was varied within a range of flow velocities
277
consistent with that observed for the Kuroshiro extension
in the subsurface (Hall, 1989)
. The
278
resulting optimized para
meter set is
shown in Table 2. The
best
-fit
J
resp
value for the
σ
θ
= 25.8 –
279
26.2 surface is 3.0 μ
mol/kg/yr
, similar to the modeled value of 2.9 μ
mol/kg/yr for the
σ
θ
= 26.9 –
280
27.4 surface in the Atlantic reported in
Levine et al., (2009)
and the estimated value of 3
281
μmol/kg/yr reported in
Feely et al.,
(2004)
that was based on
remineralization rates
. The
best
-fit
282
J
resp
value for th
e
σ
θ
= 26
.5 – 26.9 surface is 1.6 μ
mol/kg/yr, although there is significant
283
uncertainty in this value because of both a
strong dependence on
size of the
exposure area
in the
284
northwest Pacific
and the
larger
depth range of the isopycna
l (~800m)
. Nevertheless,
the
J
resp
285
value is within the range
prior estimates for these depths
(Feely et al., 2004)
.
286
287
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
14
2.3.4 Model with photosynthesis
288
To simulate the effects of photosynthesis in the photic zone, we appl
ied
a photosynthe
tic
289
flux
signal
to a 10 × 20 box region at the northwestern boundary of th
e model for the
σ
θ
= 25.8 –
290
26.2
surface
. The
flux is of pure O
2
with a composition of
δ
18
O
= -20.172
and
δ
17
O
= -10.275
291
relative to air, with
36
= - 0.4‰, resulting in an admixture of photosynthetic and respired O
2
in
292
the photic zone. We thus call this the “explicit addition” method. Th
e photosynthetic endmember
293
wa
s calculated by first computing the
18
O/
16
O and
17
O/
16
O fractionation relative to VSMOW2 for
294
“average phytoplankton” reported in (Luz & Barkan, 2011)
i.e.,
18
α
= 1.003389 and
17
α
=
295
1.001778. These fractionation factors were then applied to V
SMOW
2 as the source water [
δ
18
O
296
= - 23.481‰
and
δ
17
O =
-12.031‰ relative to air
(Wostbrock et al., 2020)
]. The goal of this
297
scheme is to use the photosynthetic isotope fractionation
from
(Luz & Barkan, 2011)
, but
to
298
scale
the isotopic composition of O
2
to be consistent with
17
Δ measurements made in our lab
299
(Yeung et al., 2018; Pack et al., 2016; Wostbrock et al., 2020)
. The
36
value of photosynthetic
300
O
2
was estimated
from
preliminary
measurements (Yeung et al., 2015)
, but the results are not
301
sensitive to
its precise value near
36
= 0
. The maximum
amount of photosynthetic O
2
added
into
302
the system
wa
s equivalent to +
60% saturation,
which, while
not
typically present in the ocean
,
303
were used to evaluate the range of
the possible
admixtures of
photosynthe
tic and respired O
2
. We
304
note that t
his implementation does not include explicit biogeochemical cycling of photosynthe
tic
305
O
2
within the photic zone
, which
results in an accumulated triple
-oxygen isotope signature from
306
photosynthetic O
2
addition and partial respiration
of the admixture
. Instead, those effects are
307
represented schematically alongside the model results.
308
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
15
Using the same approach for
the
smaller ventilation region of the
σ
θ
= 26
.5 – 26.9 surface
309
yielded negligible changes to the isotopologue patterns, so an
second
approach was
also used to
310
simulate the effects of photosynthesis, following that used in an earlier 3D model simulation
311
(Nicholson et al., 2014)
. Rather than adding in photosynthetic O
2
explicitly, photosynthetic O
2
312
was added implicitly by changing the ventilation boundary condition to be representative of the
313
mixed
-layer isotopologue compositions measured during CDISK4,
namely, δ
18
O = 0.36‰,
17
Δ =
314
30 ppm, and Δ
36
= 1.90‰. The largest effect of this “implicit addition”
method
is to elevate
17
Δ
315
values
from equilibrium (i.e., above 8 ppm)
, although overall the effects remain subtle
and
316
sufficient for illustrative purposes
.
317
318
3. Results:
319
3.1 Isotopic measurements
320
The CDISK
4 data span
the oligotrophic and subarctic North
east Pacific (
Fig. 1) at depths
321
ranging from the surface to 3000 m (
σ
θ
= 20.7 –
27.8)
, and show consistent patterns associated
322
with biogeochemical processing. At the Hawaii Ocean Time Series site (CDISK
4- S1),
for
323
example, dissolved O
2
saturation generally decreases,
while δ
18
O and
Δ
36
values generally
324
increase with
increasing depth except
towards the base of the oxygen minimum zone (
1486m
325
sample; Fig. 4). These trends are associated with respiratory isotopic fractionation, which
326
increases δ
18
O and
Δ
36
values in the residual O
2
(Guy et al., 1993; Ash et al., 2020)
. The
17
Δ and
327
Δ
36
values also show prominent photosynthetic signals
:
17
Δ values are elevated at the top of the
328
thermocline, while Δ
36
values are lowered, owing to the accumulation of photosynthetic O
2
that
329
has been partially res
pired (Luz & Barkan, 2000, 2009; Yeung et
al., 2015)
. In the mixed layer,
330
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
16
the
17
Δ and Δ
36
values of O
2
approach atmospheric
values, consistent with
gas exchange driv
ing
331
the isotopic composition of O
2
toward solubility equilibrium with the atmosphere (Knox et al.,
332
1992; Li et al., 2019)
.
333
T
he δ
18
O and Δ
36
data increase as dissolved O
2
concentrations decrease, from values
334
below solubility equilibrium (
-0.6‰ and 1.5‰ with some variability
, respectively) to values
335
much higher than those in air (18‰ and 3.1‰, respectively), with little variance about their
336
cur
vilinear trends. The
17
Δ data, however, show more variable behavior: near the surface,
17
Δ
337
values range from 21 – 118 ppm, whereas at low O
2
concentrations (< 40% saturation) they
338
range from 50
– 100 ppm. In effect, the
17
Δ data envelope
appears to narrow with decreasing O
2
339
concentrations
, w
ith a pronounced increase in minimum values
, although the narrowing may
340
simply reflect the locations sampled
. The Δ
36
data show no discernable trend toward local
341
isotopic equilibrium, which would range from 1.77‰ (2°C) to 1.49‰ (27°C) in these waters.
342
These data are plotted and compared with the two
-gyre model results in Section 4.1.
343
344
3.2 Two
-Gyre Model
345
The model domain is generally ventilated near
the Northwest corner and along the entire
346
Southern edge, with some interhemispheric mixing, leading to different systematics in the
347
Northern and Southern gyres. We illustrate the
general features of these advection
-diffusion-
348
respiration trajectories
on th
e
σ
θ
= 25.8 – 26.2 surface below
. T
hree specific regions are
349
highlighted in Fig. 5;
their advection-
diffusion
-respiration arrays
are shown in Fig. 6.
350
In the northwest
corner
, the advecti
ve direction is clockwise
, wh
ereas the direction of
351
eddy-
diffusi
ve transport of O
2
is primarily counter
-clockwise
. Oxygen concentrations decrease
352
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
17
from the
northeast
to the
southwest
, driving net diffusive transport along this gradient.
Advection
353
and diffusion thus drive O
2
transport
in opposi
ng
directions. At
steady state, the O
2
flux
es
F
in
354
this region satisfy th
e following relationship:
355
|
퐹퐹
퐾퐾
| =
|
퐹퐹
퐴퐴
| +
|
퐹퐹
푅푅
|
(
9)
356
Here, the subscripts refer to the contributions from
eddy diffusion (
K
), advection
(
A
), and
357
respiration
(
R
). Because
F
R
is small (
3.0
μ
mol
O
2
/kg seawater
/yr) and the O
2
gradient is
358
relatively large,
the O
2
budget
has a large contribution from
diffusive
-advective mixing between
359
high-
and low
-O
2
waters. The
effects on the isotopic composition of O
2
therefore trend toward
360
that of two
-end
member
mixing
between high-
and low
-saturation waters
; in Fig. 6, the isotopic
361
trends
in the northwest region resemble those predicted
for mixing between surface waters and
362
Rayleigh-
fractionated waters at ~
30% O
2
saturation
, as implied by Fig. 5.
363
In the northeast
, the O
2
flow pattern changes
: a dvection and eddy diffusion drive
O
2
364
transport
in a similar direction
(counterclockwise
). At steady state, the O
2
fluxes th
erefore
satisfy
365
the relationship:
366
|
퐹퐹
푅푅
| =
|
퐹퐹
퐴퐴
| +
|
퐹퐹
퐾퐾
|
(10)
367
Both the advecti
ve and eddy-
diffusi
ve fluxes
(cf. Fig. 2 and the concentration gradient in Fig. 5)
368
are smaller
in this region, balancing the
small respiration
flux
. Changes in O
2
saturation in this
369
region are therefore more strongly affected by
respiratory consumption, resulting in a
n isotopic
370
pattern that trend
s closer to
Rayleigh fractionation
, at least compared to the Northwest Pacific
:
371
the relatively weak
advection and diffusion makes this region show more
closed
-system
behavior
.
372
In the
southern gyre
, advection is counterclockwise, and the concentration gradients are
373
weaker owing to a larger exposure surface
for the isopycnal
. The mixing pattern
in the southeast
374
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
18
is similar to that of the n
ortheast
, with advection and diffusion in the same direction
, resulting in
375
a Rayleigh
-like isotopic fractionation pattern
; however, the O
2
saturation range is smaller than in
376
the northeast
, so that portion of the array does not appear prominently in Fig. 6. In the southwest,
377
the mixing patten is similar to
the northwest, where the large O
2
concentration gradient lead
s to
378
isotopic trends
closer to
those for
two
-endmember mixing.
379
These basic systematics suggest that t
he model results can be
understood
as a continuum
380
of water parcels lying somewhere between
the trends expected from
closed
-system Rayleigh
381
fractionation and two-
endmember mixing between low
- and a high-
O
2
endmember
s, with region-
382
specific patterns reflecting the local
budget
. For example, the
σ
θ
= 26.5 –
26.9 surface reflects a
383
more compl
ex mixture likely involving more than two mixing endmembers (cf. Fig. S
3). The
384
range in the observational data should
nevertheless constrain the range of low
-O
2
endmembers
385
that contribute to the subsurface Pacific O
2
budget
.
386
Ultimately
, the simulated
δ
18
O,
17
∆,
and
Δ
36
value
s are all anticorrelated with
O
2
387
saturation because of the tendency for respira
tion to consume light oxygen isotopologues
. For
388
17
values, the
trend
with O
2
saturation
is controlled by the particular mass
-dependent
389
fractionation slope for respiration used
(see Table 1)
: t he
θ
17/18
,resp
value of
0.520 used in these
390
simulations
is larger than the reference slope of
λ =
0.518 used to defin
e
17
, resulting in an
391
increase in
17
Δ values as O
2
saturation decreases.
Similarly, the mass
-dependent fractionation
392
slope for
18
O
18
O relative to
16
O
18
O,
θ
36/18,resp
= 2.048, is larger than the slope that would preserve
393
36
values upon Rayleigh fractionation (i.e.,
θ
36/18
= 2.000)
, resulting in an increase in
36
values
394
as O
2
saturation decreases.
395
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.
Manuscri
pt under consideration at
Jo
urnal of Geophysical Research -
Oceans
19
Modeled
17
Δ
-
δ
18
O and
Δ
36
-
δ
18
O correlations are generally positive with modest
396
variability about overall curvilinear trends (Fig. 6 D
-F). Compared to isotopologue
-O
2
saturation
397
plots (Fig. 6 A
-C), t
he model results for the northwest region in these isotopologue cross
-plots
398
adhere close
r to the expected trajector
ies
for two
-endmember mixing. Th
is observation indicates
399
that the de
partures
from the mixing trajectories in Fig. 6 A
-C are mainly du
e to
minor
400
contributions from
low
er-O
2
waters.
401
T he apparent mass dependence
resulting from
isopycnal
transport
, mixing, and
402
respiration
resembles Rayleigh fractionation with
θ
17/18,resp
< 0.520 and
θ
36/
18,resp
> 2.0
48
―i.e.,
403
different from the process
-
level values―
mainly because curvilinear mixing trajectories for
17
Δ
404
and Δ
36
values (Miller, 2002; Eiler, 2007; Yeung et al., 2012)
draw the model array away from
405
pure Rayleigh-
like trends. These deviations from the
process-
level
mass
-dependent fractionation
406
slopes for respiration
resemble
the
deviations observed in
δ
18
O data
reported here and in
407
previous work (Bender, 1990; Levine et al., 2009)
. For example, a respiratory fractionation
408
factor of
18
α
= 0.982 yiel
ded model results having an apparent Rayleigh fractionation factor of
409
18
α
= 0.990 in the subtropical Atlantic (Levine e
t al., 2009)
. Notably, however, Rayleigh
410
fractionation and two
-
endmember mixing yield nearly coincident trajectories for Δ
36
values when
411
plotted against O
2
saturation (Fig. 6C), leading to minimal spread in the model results compared
412
to the other isotopologue tracers.
413
The addition of photosynthe
tic O
2
to a small
surface-
outcropping region
in the northwest
414
corner
of the
σ
θ
= 25.8 – 26.2 surface
results in higher O
2
concentrations throughout the
415
isopycnal and a
larger range of predicted isotopic compositions
compared to
the respiration
-only
416
scheme. T
he large
st spread i
n isotopic
composition occurs at
high O
2
saturation
; however, these
417
compositions converge
toward the respiration
-only results as
O
2
saturation decreases
because of
418
ESSOAr | https://doi.org/10.1002/essoar.10510808.2 | CC_BY_4.0 | First posted online: Fri, 11 Mar 2022 10:11:21 | This content has not been peer reviewed.