University of Southern California Dissertations and Theses

Anaerobic iron cycling in an oxygen deficient zone

(USC Thesis Other)

Anaerobic iron cycling in an oxygen deficient zone

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Anaerobic iron cycling in an oxygen deficient zone
by
Kenneth M. Bolster
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Geological Sciences)
August 2021
Copyright 2021 Kenneth M. Bolster
Acknowledgements
I would like to acknowledge the partial financial support of the National Science Foundation, the
Wrigley Institute for Environmental Studies, and the University of Southern California.
I would like to thank my advisor Jim Moffett
I would like to thank my committee members Doug Hammond, Seth John, and Doug Capone.
I would like to thank all of my friends in the Earth Sciences and Marine Biology departments
at USC, as well as the friends I’ve made at other institutions during this journey.
Finally, I would like to thank my family for their support and encouragement.
ii
Table of Contents
Acknowledgements ii
List of Figures vi
Abstract ix
Introduction to Chapter 1 1
Chapter 1: Determination of iron(II) by chemiluminescence using masking ligands to
distinguish interferences 3
1.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Materials and procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.2 Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6 Comments and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Introduction to Chapter 2 24
Chapter 2: Iron and manganese accumulation within the Eastern Tropical North Pacific
oxygen deficient zone 26
2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3.1 Sampling sites and trace metal sample collection . . . . . . . . . . . . . . 30
2.3.2 Nutrient sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3.3 Principal component analysis . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4.1 Physics, nutrients, and oxygen . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4.2 Deep plumes in ETNP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5.1 Implications of correcting for interferences with DTPA . . . . . . . . . . . 39
2.5.2 Iron transport within the 13CW water mass . . . . . . . . . . . . . . . . . 39
iii
2.5.3 Comparison of ETNP and ETSP ODZs . . . . . . . . . . . . . . . . . . . 41
2.5.4 Origin and significance of deep iron features . . . . . . . . . . . . . . . . 43
2.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Introduction to Chapter 3 46
Chapter 3: Iron oxidation in the water column of the Eastern Tropical North Pacific
Oxygen Deficient Zone 48
3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.3.1 Comparison of shipboard and in-situ incubations . . . . . . . . . . . . . . 51
3.3.2 Distribution of iron oxidation rates . . . . . . . . . . . . . . . . . . . . . . 54
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.4.1 Particle dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.4.2 Quantification of ambient rates . . . . . . . . . . . . . . . . . . . . . . . . 57
3.4.3 Residence times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.4.4 Implications for iron transport . . . . . . . . . . . . . . . . . . . . . . . . 59
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.6 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.1 Sample sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.2 Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.3 Shipboard incubations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.6.4 In-situ incubations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.6.4.1 Isotope spikes . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.6.4.2 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Introduction to Chapter 4 64
Chapter 4: Iron reduction in a nitrate-replete oxygen deficient zone mesocosm 66
4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3.1 Mesocosm construction . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3.2 Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.7 Supplemental Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Introduction to Chapter 5 79
iv
Chapter 5: Modeling anaerobic iron oxidation in the Eastern Tropical Pacific 81
5.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
References 94
v
List of Figures
1.1 Iron(II) chemiluminescence signals (solid) are inhibited by both DTPA (dashes)
and EDTA (dotted). Error bars show 2s confidence intervals. . . . . . . . . . . . 12
1.2 Iron(II) from the ETNP ODZ calculated by correction with DTPA and EDTA
yields similar results. Error bars show 2s confidence intervals. . . . . . . . . . . 13
1.3 Effect of masking ligands on V(IV) signals. In contrast to iron(II), signals from
vanadium(IV) (solid) are not inhibited by DTPA (dashes), but are inhibited by
EDTA (dotted). Error bars show 2s confidence intervals. . . . . . . . . . . . . . . 15
1.4 Chemiluminescent signals from iron(II) alone (circles) and iron(II) plus 4 nM
vanadium(IV) (squares). Signals from iron(II) and vanadium(IV) are independent
and additive. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5 Iron(II) concentrations (circles) are correlated with sunlight (solid line), while total
dissolved iron (squares) increases during the day and decreases at night. Error bars
show 2s confidence intervals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.6 Iron(II) in Big Fisherman’s Cove oxidizes with a half-life of 3.1 ( 0.8) min. Error
bars show 2s confidence intervals. . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.7 Apparent iron(II) concentrations (dashed) compared with iron(II) concentrations
measured after a DTPA correction (solid), shown with the oxycline depth (dotted).
Iron(II) concentrations in the ETNP follow a similar pattern to previous reports,
but are approximately 10% lower. Error bars show 2s confidence intervals. . . . . 19
2.1 Station map. Stations occupied during the 2016 (NOAA ship Brown) cruise (cir-
cles), 2018 (R/V Revelle) (squares), and during a 2012 cruise (R/V Thompson)
(triangles). Major sites of known or suspected trace element inputs are marked as
well (stars), including the R´ ıo Balsas river mouth, near station 7, the Revillagigedo
archipelago, near station 16, and two known hydrothermal vent fields along the
East Pacific Rise (EPR). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Dissolved nitrate, nitrite, and oxygen in the alongshore transect in 2016. Dissolved
oxygen contour lines are shown in white. The dark black line shows the 26.5 kg/m
3
isopycnal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
vi
2.3 Dissolved oxygen, nitrate, and nitrite in the offshore transect in 2016. White con-
tour lines show dissolved oxygen concentrations, and the dark black line is the 26.5
kg/m
3
isopycnal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.4 Trace metal data in the alongshore transect in 2016. White contour lines show
dissolved oxygen concentrations, and the thick dark line shows the 26.5 kg/m
3
isopycnal. The R´ ıo Balsas outflow is located between stations 7 and 8. . . . . . . . 35
2.5 Trace metal data in the offshore transect in 2016. White contour lines show dis-
solved oxygen concentrations, and the dark line shows the 26.5 kg/m
3
isopycnal.
Note that the color scales are different from those in Figure 2.4 . . . . . . . . . . . 36
2.6 Iron(II) and dissolved oxygen profiles at (a) the San Blas Basin in 2018. The
dashed line shows the bottom depth of the station. Note that iron(II) concentrations
are plotted on a log scale. Iron(II) concentrations shallower than 80 m were below
the limit of detection. (b) The shelf station in 2012 near the mouth of the Rio Balsas. 37
2.7 Iron, manganese, iron(II), and oxygen concentrations at (a) station 6, (b) station
12, and (c) station 16. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.8 Temperature-salinity plots for all stations occupied during the 2016 Brown cruise
(grey), and all samples with>0.5 nM Fe(II) (black). Almost all elevated iron(II)
concentrations are associated with the 13CW endmember. . . . . . . . . . . . . . . 40
2.9 Dissolved sample measurements< 1000 m in the ETNP (a) in 2016 and the ETSP
(b) during the GEOTRACES GP16 cruise plotted on the first two coordinates of
principal component analyses. In the ETNP, dissolved manganese closely tracks
iron(II) and nitrite, while total dissolved iron has a more nutrient-like distribution.
In the ETSP, dissolved iron was closely coupled with nitrite and iron(II), but not
strongly related to manganese, which showed surface maxima. . . . . . . . . . . . 42
3.1 Map of the sampling sites in the Eastern Tropical North Pacific, overlying the
average dissolved oxygen in the region at 200 m. Data from the World Ocean
Atlas 2018 ([118]). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2 (Left) Iron oxidation rate constants for in-situ incubations at 120 and 180 m depth,
including incubations with suspended particles only (medium) and suspended +
sinking particles (dark). (Right) Rate constants for shipboard incubations at 180 m,
including filtered incubations (light) and suspended particle incubations (medium). 53
3.3 Nitrate (top left), nitrite (top right), dissolved iron(II) (bottom left), and oxidation
rate constants (bottom right), at station P2 in October 2019, including experiments
with only suspended particles (triangles) and with sinking material in a sediment
trap incubator (circles). Oxygen concentrations are beneath the limit of detection
below 110 m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
vii
3.4 Incubation data from P3. Iron(II) profile (left), shipboard incubation replicates
(center), and comparison between the shorter (shipboard) and longer (in-situ) in-
cubations (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.5 Oxidation rate constants in filtered (F1 and F2) and unfiltered (U1 and U2) repli-
cates of shipboard incubations at 150 m (top left), 180 m (top right), 250 m (bottom
left) and 300 m (bottom right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.6 Dissolved iron(II) concentration and iron(II) half-lives, plotted on a log-log scale. . 58
4.1 Nitrite (mM), ammonium (nM), and phosphate (mM) concentrations over time in
a mesocosm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.2 First mesocosm measurements of iron(II), including one mesocosm with unfiltered
seawater and one with 50 mm filtered seawater. Error bars are 2s confidence
intervals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3 Second mesocosm measurements of iron(II), including unfiltered and 50 mm fil-
tered seawater. Error bars are 2s confidence intervals. . . . . . . . . . . . . . . . . 73
4.4 Particulate iron concentrations at the start of each mesocosm, including the unfil-
tered and 50mm filtered treatments. Error bars are 2s confidence intervals. . . . . 74
4.5 Total dissolved iron over time in the first mesosocosm with size fractionation treat-
ment. Corresponds to Figure 4.2. Total dissolved iron was higher in the unfiltered
treatment, and peaked at the same time the iron(II) maximum was achieved. . . . . 77
4.6 Total dissolved iron over time in the second mesocosm with size fractionation.
Corresponds to Figure 4.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.1 A zonal transect of modeled iron(II) concentrations along 19

N. This model used
only sedimentary inputs along the 26.5 isopycnal, and first order oxidation kinetics
within the ODZ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.2 Modeled iron(II) concentrations compared to the observed concentrations in each
model box. The 1:1 line is shown in orange. The model significantly underesti-
mates the iron(II) concentrations in many samples, especially at the lower end. . . . 87
5.3 Model results along 19

N, from a model incorporating a remineralization source
of iron(II) and first order oxidation kinetics. . . . . . . . . . . . . . . . . . . . . . 88
5.4 Comparison of results from a model with remineralization and first order kinetics
to observed data. The model still underestimates iron(II) values at low concentrations. 89
5.5 Results from a model incorporating second order oxidation kinetics, along 19

N. . 90
5.6 Comparison of the model with second order oxidation kinetics with observation.
This model does a significantly better job at low concentrations, compared to Fig-
ures 5.2 and 5.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
viii
Abstract
This thesis contains several studies focused on iron cycling and transport in the Eastern Tropical
North Pacific (ETNP) oxygen deficient zone (ODZ). This thesis will argue that biogeochemical
cycling within the ODZ drives significant transport of iron away from the continental margins of
Mexico to the oligotrophic ocean.
Iron is generally abundant for land dwelling organisms, since it makes up a significant amount
of Earth’s crust. However, in the ocean, iron will typically oxidize and form either iron oxide
particles, or it will adsorb onto other large particles sinking through the ocean, leaving only trace
amounts at the ocean surface. Iron availability may have changed significantly over geological
time, and is a key component in the cycle of glacial and interglacial periods that Earth experiences
every few millennia, since it impacts the ability of the oceans to absorb carbon dioxide from the
atmosphere. From this perspective, there are strong motivations for marine scientists to study iron.
ODZs are open ocean regions in which the dissolved oxygen concentration drops to negligi-
ble levels, as a result of high productivity and slow circulation. These are anoxic settings, but
rather than being dominated by sulfate reduction processes like many enclosed basins, ODZs are
dominated by nitrogen redox processes.
In the past two decades, it was discovered that all three ODZs contain elevated concentrations
of iron in the +2 oxidation state. Because the scarcity of iron is strongly impacted by oxygen,
it seemed reasonable to assume that the processes of iron oxidation and scavenging would be
inhibited in ODZs, where oxygen is absent. However, there is a thermodynamic challenge to this
argument. Although oxygen is typically the oxidant of iron in the ocean, nitrate and nitrite both
have high enough redox potentials for them to make iron oxidation thermodynamically favorable.
ix
In sediments, for instance, iron(II) is not found in layers with detectable nitrate or nitrite, since
these reactions can occur rapidly under certain conditions. ODZs, although depleted in oxygen,
contain significant concentrations of nitrate and nitrite. Why would iron(II) then accumulate in
these environments?
Anaerobic iron oxidation would not necessarily inhibit iron transport in these regions. A pro-
cess known as the shelf to basin shuttle was believed to be active in other anoxic basins, in which a
cycle of iron oxidation in the water column and reduction in sediments could progressively move
iron from coastal margins towards the continental shelves, forming deep plumes that could trans-
port iron significant distances. Shortly before the work in this thesis began, such a plume was
observed off the coast of Peru, below the Eastern Tropical South Pacific ODZ.
This thesis attempts to answer five major questions about iron cycling in the ETNP, listed
below.
1. How can we accurately measure iron(II) in ODZs?
2. Is the shelf to basin shuttle active in the ETNP ODZ?
3. What is the rate of anaerobic iron oxidation in the ETNP ODZ?
4. Is it possible for iron(II) to accumulate in an ODZ independent of reducing sediments?
5. What is the overall rate of transport of iron through the ETNP ODZ?
Each of these questions is addressed in its own chapter. Chapter 1 describes a technique to
measure iron(II) in ODZs using luminol chemiluminescence and correcting for interfering com-
pounds. Chapter 2 describes the distribution of iron(II) (and other trace metals) throughout the
ETNP, and analyzes the controls on that distribution. Chapter 3 contains measurements of the rate
of anaerobic iron oxidation using an incubation approach, and evaluates possible mechanisms of
that oxidation. Chapter 4 describes a mesocosm which simulates the water column of an ODZ
x
while not containing sediments, allowing experiments on exclusively water cycle processes. Fi-
nally, Chapter 5 includes attempts to model the iron cycling studied in Chapters 2-4, and to use
that model to estimate the overall iron transport in the region.
xi
Introduction to Chapter 1
In all chapters of this thesis, iron(II) features in a prominent role. However, prior to working on
iron(II) cycling, it was necessary to validate a method to quantify iron(II) in seawater samples.
Prior to the work in this chapter, the preferred technique was luminol chemiluminescence. Unfor-
tunately, it had recently been shown that this method was systematically overestimating iron(II)
concentrations, especially in surface waters. This issue needed to be corrected before additional
work on iron(II) could be performed.
The luminol chemiluminescence method consists of mixing a seawater sample and a reagent
containing the compound luminol in a chamber connected to a photomultiplier. The presence of
iron(II) in the sample causes luminol to emit photons, which can be counted and used to calculate
a concentration. However, this process is not driven by a direct reaction between iron(II) and lumi-
nol. Instead, iron(II) oxidizes under the high pH conditions of the luminol reagent, and produces
reactive oxygen species which react with luminol. In theory, any compound in seawater which pro-
duces reactive oxygen species at high pH could artificially inflate the signal. A variety of organic
compounds, as well as the +4 oxidation state of vanadium, were known to have this effect.
In this chapter, diethylenetriaminepentaacetic acid (DTPA), was shown to completely inhibit
the iron(II) signal from this technique, while not blocking signals from vanadium(IV) or organic
compounds. Therefore, a seawater sample treated with DTPA could be used as a matrix blank in
order to correct for these interfering effects.
This method was validated in a study on diel cycling of iron(II) in Big Fisherman’s Cove, at
the Wrigley Institute for Environmental Studies on Santa Catalina Island. This method was used
to measure strong diel cycling in the surface waters of the cove, and to quantify oxidation kinetics
1
in that system. The results of these experiments were interesting in their own right, but are outside
the scope of this thesis.
The new method was then applied to a depth profile of iron(II) measured during a cruise to the
ETNP ODZ, showing that luminol chemiluminescence, without the DTPA correction, would sig-
nificantly overestimate the iron(II) concentrations. The shape of the profile was also significantly
different, which was later investigated in Chapters 2 and 3 of this thesis.
This chapter was published in 2018 by the journal Limnology and Oceanography: Methods.
2
Chapter 1
Determination of iron(II) by chemiluminescence using masking
ligands to distinguish interferences
1.1 Abstract
Iron(II) can catalyze the oxidation of luminol in seawater and this chemiluminescent reaction has
been widely used for iron(II) determination. The method is vulnerable to interferences from other
analytes that catalyze luminol oxidation. We have shown that addition of diethylenetriamine pen-
taacetic acid (DTPA) to a sample inhibits the reaction of iron(II) with luminol, while not affecting
other substances that also catalyze luminol oxidation under our experimental conditions. DTPA-
treated samples can therefore be used as sample blanks, with the difference between an untreated
seawater sample and a DTPA-treated seawater sample related to the concentration of dissolved
iron(II). The DTPA correction has been applied to measure diel variability of iron(II) due to pho-
toreduction in a coastal environment, and to measure vertical distributions of iron(II) in the eastern
tropical north Pacific oxygen deficient zone.
1.2 Introduction
Iron plays a number of key roles in marine environments, as a micronutrient, a terminal electron
acceptor, and a source of chemical energy (e.g., [1]). Its relevance to an ecosystem depends not
3
only on iron’s concentration, but also on its redox speciation. In the vast majority of modern marine
settings, the thermodynamically favorable form is the ferric (+3) oxidation state [2, 3]. The ferrous
(+2) form is less common since, under typical conditions of ocean pH and temperature, iron(II) will
react spontaneously with dissolved oxygen and reactive oxygen species (ROS; e.g. peroxide) to
form iron(III) complexes, with a half-life of minutes [4–7]. Iron(III) is more particle reactive than
iron(II), causing it to be removed more rapidly from the water column. In addition, iron(III) is more
strongly complexed by organic ligands, which may limit bioavailability but increase solubility [8–
11]. Because of these distinctions, processes which lead to iron(II) accumulation are important
for understanding the influence of iron cycling on marine environments. Under reducing/anoxic
conditions, iron(III) can be used as a terminal electron acceptor [12]. Iron(III)-ligand complexes
or minerals can release ferrous iron through photoreduction [13]. Iron(II) may also be released
during remineralization, or be reduced extracellularly as part of an uptake strategy by microorgan-
isms [14]. Hydrothermal vents have also been recognized as an important source of ferrous iron
[15]. In very reducing environments in which iron(II) accumulates to micromolar concentrations,
iron(II) concentrations can be measured colorimetrically [16]. However in oxygenated environ-
ments, or denitrifying oxygen deficient zones (ODZs), iron(II) concentrations can be in the nano-
or picomolar range, and more sensitive methods are required. One commonly used technique is a
flow-injection analysis using chemiluminescence for detection [17–31]. In this system, a seawater
sample is mixed with a reagent containing luminol (5-amino-2,- 3-dihydrophthalazine-1,4-dione)
buffered to a high pH. Under those conditions, iron(II) in the sample is rapidly oxidized by oxy-
gen, producing ROS, which then react with luminol and trigger the release of light [32]. However,
iron(II) is not the only compound capable of stimulating the release of light from luminol. In
general, any compound which oxidizes slowly at pH 8 but rapidly at pH 10 in a single-electron
oxidation giving rise to superoxide could also stimulate luminol oxidation and generate an inter-
fering signal. In a complex mixture such as seawater, other interfering species may be present,
which could lead to an inflated estimate of the iron(II) concentration. Several trace metal interfer-
ents are known to exist [24, 28]. It is likely that some organic compounds could create a similar
4
effect [33]. ROS are able to do so as well [5, 32, 34]. A number of recent studies have found
apparent iron(II) concentrations, as measured using flow-injection, which were higher than the
concentration of total dissolved iron in the same samples, which could indicate the presence of an
interfering compound [29]. For this reason, the iron(II) concentration inferred from flow injection
analyses is sometimes interpreted as an upper-bound [24]. However, Hopkinson and Barbeau [24]
investigated the potential for chemiluminescence interference by I

, Cr(III), Se(IV), NO

2
, Mn(II),
Cu(I), Co(II), Sb(III), As(III), V(III), V(IV), and Mo(V), all of the inorganic compounds found
in ODZs at concentrations of 100 pM or higher with redox pairs whose reduction potentials fell
within a plausible range of oxic to sulfidic conditions. They found that the only compounds ca-
pable of generating a detectable chemiluminescence signal at plausible concentrations in a natural
environment were the reduced vanadium species. Vanadium(III) is only formed in very reducing
conditions and is highly insoluble [35], so it can be ruled out as an interferent. Vanadium(IV),
while thermodynamically unstable in the nitrate-replete environment of an ODZ [35], could con-
ceivably accumulate, and must be considered a possible interferent. A number of studies which
attempted to measure iron(II) oxidation found much slower rates than would be predicted for in-
organic Fe2+, based on local temperature and pH conditions (e.g., [36]). While it is possible that
this could be the result of complexation by organic ligands, alternative evidence for strong iron(II)
binding ligands, whether by electrochemical detection (e.g., [8]) or by chromatographic separation
(e.g., [37]), is lacking. These measurements are at least consistent with the hypothesis that some
of the apparent iron(II) signal is in fact coming from an interferent with a longer half-life. If the
identity of an interfering compound or compounds were known, it would be desirable to develop
a masking reagent capable of suppressing the interfering signal. However, since predicting which
interferents are present in an environment is typically not possible, especially since not all interfer-
ents may have been identified, so attempting to eliminate the signal from interfering compounds is
not practical. An alternative approach, which we investigate here, is to develop a masking reagent
specific to iron(II), and to use a sample treated with that reagent as a new blank. In this way, any
signal coming from interfering compounds could be subtracted. We have investigated the chelators
5
diethylenetriamine pentaacetic acid (DTPA) and ethylenediamine tetraacetic acid (EDTA) for this
purpose. Additionally, as EDTA masks a signal from vanadium(IV), but DTPA does not, the dif-
ference between an EDTA and DTPA-treated sample can be used as an indicator for the presence
of reduced vanadium compounds. Calibration curves made with the luminol chemiluminescence
are typically nonlinear with a positive second derivative, and it is common to use power law (e.g.,
[32]) or quadratic (e.g., [29]) relationships in order to fit a standard curve. In other words, the
sensitivity of luminol chemiluminescence to iron(II) increases as the concentration of iron(II) in-
creases, as a result of changing the ROS produced [32]. It would be reasonable to ask whether
the same relationship would also hold for an interferent. If vanadium(IV) is present in a sample,
does the sensitivity of the luminol to iron(II) increase? Given the complexity of the mechanisms
by which metal compounds trigger the chemiluminescent oxidation of luminol, it is plausible that
different compounds will act synergistically, or alternatively that they will not interact at all, and
the observed signal will simply be additive. Establishing which of these possibilities is true is nec-
essary in order to determine the proper mathematical method to correct for interferences. In this
article, we report on a series of experiments to validate the method, and then apply the correction to
two natural environments in which interferents may be present. The first is a diel study to measure
photocycling of iron at Santa Catalina Island, where other photoproduced species could generate
false signals, as has been suggested previously [29]. In the second demonstration, vertical profiles
of dissolved iron(II) were collected in the eastern tropical north Pacific (ETNP) ODZ.
1.3 Materials and procedures
1.3.1 Materials
All reagents were prepared in acid-washed opaque low-density polyethylene (LDPE) bottles, to
prevent trace metal contamination or degradation of the reagents by light. The luminol reagent was
initially prepared as a concentrate, by mixing 0.796 g sodium luminol (Sigma), 250 mL Optima
6
grade ammonium hydroxide solution (Fisher), high-purity water (18.2 MW cm
1
), and approxi-
mately 45 mL Optima grade hydrochloric acid (Fisher) to reach a final volume of 1 L and a pH of
10.25. To produce the working reagent, the concentrated mixture was diluted in high-purity water
by a factor of four and heated at 50

C for 9–12 h in either an oven or a hot water bath. DTPA
was prepared in high-purity water with final concentrations of 50 mM DTPA (Sigma-Aldrich) and
200 mM NaOH (Aldrich, trace metals basis). EDTA was prepared in high-purity water with a
final concentration of 50 mM EDTA (Aldrich, trace metals basis) and 100 mM NaOH (Aldrich,
trace metals basis). The MOPS (3-(N-Morpholino)propanesulfonic acid) buffer was prepared with
a final concentration of 100 mM MOPS (Sigma) and 50 mM NaOH (Aldrich, trace metals ba-
sis) in high-purity water, so that the final pH was 7.2. All reagents were used within 1 month of
preparation. Because iron(II) oxidizes rapidly in seawater, standards were prepared individually
and analyzed immediately. A primary stock of 10
2
M Fe(II) was prepared by dissolving ammo-
nium iron(II) sulfate in high-purity water acidified to pH 2 using Optima grade hydrochloric acid
(Fisher). The low pH slows oxidation to negligible levels, but the standard was replaced monthly.
A working stock was prepared by diluting the primary stock by a factor of 10
4
in high purity, cre-
ating a 10
6
M solution at pH 6. This solution would oxidize to a significant extent within several
hours, so it was always prepared fresh immediately before a calibration curve or an experiment
using standards. This working stock was then added to iron(II) free seawater, acquired by filtering
seawater through a 0.2mm filter and storing in an opaque bottle at room temperature for at least 24
h. Vanadium(IV) calibration curves were prepared using a similar procedure. A 10
2
M primary
stock was made by dissolving vanadyl sulfate hydrate in pH 2 Optima grade HCl (Fisher). Vanadyl
sulfate can have varying degrees of hydration, so a sample of the salt was weighed and dissolved in
5% nitric acid and the vanadium concentration was measured by a Thermo Element 2 inductively
coupled plasma mass spectrometer (ICP-MS), using indium as an internal standard, using a certi-
fied vanadium standard (VWR) to determine the true weight fraction of vanadium in the vanadyl
sulfate hydrate. A 10
6
M working stock was prepared immediately before a calibration curve,
7
and spikes of the working stock were added to the same aged seawater as described above. Dis-
solved iron(II) concentrations were measured using a luminol chemiluminescence based FeLume
flow-injection system (Waterville Analytical), with data collection and integration controlled by
Waterville Analytical software. The hardware was similar to that used in a number of previous
open-ocean studies. A standard quartz flow cell was used in conjunction with a Hamamatsu HC135
photon counter. The flow rate through the peristaltic pump was 2 mL min
1
. The photon counter
integration time was 200 ms, with two measurements per second for up to 360 s. All samples were
introduced in acid-washed Teflon bottles. The DTPA and EDTA reagents used in these studies
have small reagent blanks, likely due to contamination by other trace elements. In addition to the
daily calibration measurements, the reagent blanks were also measured as the difference between
the aged seawater blanks and a sample of the aged seawater treated with the chelator at the same
concentration to be used in actual samples. The MOPS buffer did not have a detectable reagent
blank.
1.3.2 Procedures
In the FeLume system, a peristaltic pump mixes a sample to be analyzed with the luminol reagent
at a 1:1 ratio in the mixing chamber of the photon counter, which is then recorded by Waterville
Analytical software in continuous mode. An aged seawater sample was measured as a procedu-
ral blank immediately prior or immediately after each sample, and the signal from the blank was
subtracted, in order to correct for instrument drift. Measurements were continued until the signal
was steady for at least 30 s, typically within 2 min of the beginning of an analysis. At that point,
the operator of the FeLume applied a time stamp within the Waterville Analytical software, which
adds a series of five data points several orders of magnitude higher than the typical signal, and
analysis was stopped and the data saved to an individual file. After analysis, an R script [38] was
used to identify the locations of the time stamps in each file and record the average and standard
deviation of the proceeding 50 data points. Because the residence time of the sample and luminol
reagent in the mixing chamber is significantly longer than the time between measurements, the
8
raw data from the photon counter is heavily autocorrelated. Therefore, the effective degrees of
freedom of each measurement are considerably less than for a measurement of 50 independent
data points. The average of the lag-1 autocorrelation for the measurements reported in this arti-
cle is 0.779, with a standard deviation of 0.057. That average was used to compute the effective
degrees of freedom for a typical measurement, 6.1, and the standard error of the mean for each
measurement as the standard deviation divided by the square root of 6.1. That uncertainty was
then propagated through the conversion to a concentration using a calibration curve using standard
techniques. All error bars reported in this article are 2s. Samples for total dissolved iron were fil-
tered immediately after collection. Samples were collected in acid-washed LDPE bottles, acidified
to result in a pH of 1.7 with Optima grade HCl (Fisher), and stored at room temperature for several
months. The analysis was performed using isotope dilution on the samples in triplicate, using an
ESI seaFAST S2 to preconcentrate the trace metals in the sample, using a procedure adapted from
Rapp et al. [39]. Iron concentrations were then measured on a Thermo Element 2 ICP-MS in
medium resolution mode. Diel studies were performed in Big Fisherman’s Cove, Santa Catalina
Island (33

26’42.5”N, 118

29’4.2”W). Water depth at this location is approximately 10 m, with a
sandy bottom. Catalina Island is arid, with significant amounts of dust. Samples were collected
using a metal-free pole sampler off the end of the dock at the Wrigley Institute for Environmental
Studies, as far as possible from boats coming to and from the dock. Samples for the diel study
shown in this article were collected approximately 2 h apart for a single day, starting before dawn
and ending after dark. As it was a 5 min walk from the sampling location to the analytical lab,
it was necessary to treat samples with the MOPS buffer in order to stabilize the iron(II). The ox-
idation rate was much slower at the lower pH and negligible over this time period [29]. Prior to
sample collection, two 30 mL Teflon bottles were spiked with 200 mL of the MOPS buffer, and
a third identical bottle was spiked with 200 mL of the DTPA reagent. The radiant solar flux was
measured using a VWR Traceable Dual-Display Light Meter. The pole sampler was rinsed several
times with the sample water, and a 500 mL sample was collected from just under the water surface.
Approximately 60 mL of the sample was transferred to a clean syringe and used to rinse a filter.
9
A 20 mL aliquot was dispensed into the bottle containing the MOPS buffer, which was capped
and shaken to lower the pH and slow the rate of iron(II) oxidation in the sample. A second 20
mL aliquot was dispensed into the DTPA bottle, which was capped and shaken, then poured into
the final bottle containing the MOPS buffer. All samples, including the remaining unfiltered wa-
ter, were transported in sealed containers to the lab space at the Wrigley Institute where a laminar
flow bench had been set up in order to provide a trace metal clean workspace. The MOPS-treated
seawater and the DTPA + MOPS seawater were analyzed for luminol chemiluminescence, within
8 min of sample collection, and the remaining water was filtered via vacuum through a 0.2 mm
polycarbonate filter and stored for total dissolved metal analysis. To measure the oxidation rate of
iron(II) in Big Fisherman’s Cove, a similar procedure was used. Five bottles were prepared with
20 mmol of MOPS buffer, and one with 10 mmol of the DTPA reagent, so that the final concen-
tration after mixing with a 20 mL sample would be 1 mM MOPS and 500 mM DTPA. A sample
was collected shortly after midday, when diel studies had shown the iron(II) concentration was
at the maximum, and immediately transferred to a darkened box. The initial two samples were
treated with MOPS and DTPA + MOPS, with additional MOPS-treated samples collected over the
following several minutes and brought to the FeLume for analysis. The DTPA + MOPS signal was
assumed to be constant over the time frame of the experiment, since experience had shown that
the DTPA-resistant signal was significantly more stable than iron(II). In all samples treated with
MOPS, the DTPA treated counterpart was also treated with MOPS in order to ensure consistency,
although it does not affect DTPA masking. Depth profiles of iron(II) were collected in January
2017 in the ETNP ODZ, on the R/V Sikuliaq, at 20

N, 107.1

W, an oligotrophic offshore sta-
tion, though still the site of active denitrification and dissolved oxygen concentrations undetectable
with even the most sensitive oxygen electrodes available, which have detection limits between 10
nM and 100 nM, as measured by Garcia-Robledo et al. [40]. All samples were collected in 5
L acid-cleaned Teflon-coated external-spring “Niskin-type” bottles (Ocean Test Equipment) on a
powder-coated trace metal clean rosette (Sea-Bird Electronics). Prior to sampling, the rosette was
lowered to approximately 400 m and allowed to de-gas for 30 min, in order to avoid contamination
10
by oxygen adsorbed to the Teflon. After sample collection, the Niskin-like bottles were taken to a
laminar-flow clean van, pressurized with argon, and filtered through cleaned 0.2 mm Supor mem-
brane filters. Three 30 mL acid-cleaned Teflon bottles were rinsed three times with the sample.
One bottle was then filled and analyzed via luminol chemiluminescence. The other two bottles
were spiked with 200 mL DTPA and EDTA reagents, respectively, so that the final concentration
after dilution would be 500 mM of the chelator, and then 20 mL of the sample was dispensed into
them before they were analyzed immediately. As the samples were isolated from oxygen until im-
mediately prior to analysis, no treatment with the MOPS buffer was necessary. Samples for total
dissolved metals were collected in 500 mL acid-cleaned LDPE bottles, which were acidified and
analyzed as described above.
1.4 Assessment
Standard curves for flow-injection analysis typically show a strong, nonlinear relationship between
added Fe2+ and the chemiluminescence signal. However, when DTPA is added to an iron(II) stan-
dard and diluted to a final concentration of 500mM, the iron(II)-dependent signal no longer exists,
generating a flat line. A similar effect is observed for addition of EDTA with a final concentration
of 500 mM. Standard curves prepared with no chelators, +DTPA, and + EDTA are shown in Fig.
1.1. Both DTPA and EDTA carry with them reagent blanks, but the magnitude of these blanks is
typically small. The uncertainty associated with a particular measurement, taking the difference
between an untreated and a DTPA-treated sample and adding the DTPA reagent blank, is there-
fore a function of the uncertainty in four measurements. Propagation of the uncertainty through a
quadratic calibration curve results in a limit of detection (three times the standard error) of 0.026
nM, which is fairly typical for this study.
Fig. 1.2 shows collected data from two stations occupied by the Sikuliaq, in which the effect
of both DTPA and EDTA treatment were compared on all samples. The fraction of the chemi-
luminescence signal which was resistant to addition of the interferences was subtracted and the
11
Figure 1.1: Iron(II) chemiluminescence signals (solid) are inhibited by both DTPA (dashes) and
EDTA (dotted). Error bars show 2s confidence intervals.
12
Figure 1.2: Iron(II) from the ETNP ODZ calculated by correction with DTPA and EDTA yields
similar results. Error bars show 2s confidence intervals.
13
difference was used to calculate iron(II) concentrations. These concentrations tended to be simi-
lar, whether a correction from DTPA or EDTA was applied. This was not surprising, since both
chelators bind Fe(II) and Fe(III) strongly. However, there was a notable discrepancy in samples
where the DTPA-derived values for Fe(II) were zero (i.e., no difference between the signal with
and without DTPA) while the EDTA derived numbers were quite high. These samples all came
from oxygenated waters below the euphotic zone where we would expect to find little or no Fe(II).
This suggests that there may be an effect of EDTA on non-Fe(II) species that we have not fully
characterized, at least in some samples. One further consideration is that the rate of complexation
of Fe by EDTA is hindered by side reactions with calcium and magnesium ions in seawater. While
that is also the case for DTPA, it nevertheless reacts more rapidly with Fe(II) than EDTA. However,
that could only account for cases where the DTPA-derived Fe(II) is higher than the EDTA-derived
value, and these discrepancies are fewer and less pronounced (Fig. 1.2). Additionally, while the
reagent blank for DTPA tends to be consistent and low, the reagent blank for EDTA is more vari-
able and typically higher (an example is visible in Fig. 1.1). For those reasons, we recommend
using DTPA as a masking ligand to correct for interferences.
While we argue that DTPA is more useful than EDTA as a masking ligand, a comparative
study of the two in samples collected within ODZs may offer some useful insights. Hopkinson and
Barbeau [24] showed that of all the redoxreactive elements that might catalyze luminol oxidation
and create an interference with Fe(II), only V(IV) had sufficient sensitivity and concentration to
be important. Moreover, they showed that its oxidation kinetics are very similar to Fe(II), so
decay characteristics alone cannot distinguish the two. However, DTPA and EDTA affect V(IV)
in very different ways. Similar to iron(II), addition of EDTA to a vanadium(IV) standard inhibits
the chemiluminescent signal. However, addition of DTPA does not affect the V(IV)-dependent
signal. Fig. 1.3 shows vanadium standard curves prepared under the same conditions as the iron(II)
standard curves in Fig. 1.1. The difference in behavior shown by vanadium(IV) and iron(II) creates
a possible test for the presence of reduced vanadium species. If vanadium(IV) is present in a
14
Figure 1.3: Effect of masking ligands on V(IV) signals. In contrast to iron(II), signals from vana-
dium(IV) (solid) are not inhibited by DTPA (dashes), but are inhibited by EDTA (dotted). Error
bars show 2s confidence intervals.
sample, addition of EDTA should decrease the signal more than the signal is decreased by addition
of DTPA.
Fig. 1.4 shows the results of combining iron(II) and vanadium(IV) in the same standard. In
samples containing 4 nM V(IV), the sensitivity of the signal to iron(II) did not increase significantly
(p ¿ 0.1). This suggests that interferents and iron(II) generate signals independently of one another,
without synergistic interactions, and implies that the difference in signal between untreated and
DTPA-treated seawater should be used to calculate the concentration of iron(II) in a sample.
The results of a diel study at Big Fisherman’s Cove are shown in Fig. 1.5. Dissolved iron(II)
typically follows a bell curve correlated with incident sunlight (R = 0.65), peaking in the early
15
Figure 1.4: Chemiluminescent signals from iron(II) alone (circles) and iron(II) plus 4 nM vana-
dium(IV) (squares). Signals from iron(II) and vanadium(IV) are independent and additive.
16
Figure 1.5: Iron(II) concentrations (circles) are correlated with sunlight (solid line), while total
dissolved iron (squares) increases during the day and decreases at night. Error bars show 2s
confidence intervals.
afternoon and decreasing towards sunset. Total dissolved iron, however, increases throughout the
day, and decreases at night.
An iron(II) oxidation experiment is shown in Fig. 1.6, showing the decrease in the ambient
dissolved iron(II) concentration after a seawater sample collected in the same manner as a diel
sample was isolated from light. The decay in iron(II) concentrations follows first-order reaction
kinetics (R
2
= 0.97), with a reaction constant k = -0.22 0.06 min
1
, which corresponds to an
iron(II) half-life in this environment of 3.1 0.8 min (2s confidence interval).
17
Figure 1.6: Iron(II) in Big Fisherman’s Cove oxidizes with a half-life of 3.1 ( 0.8) min. Error
bars show 2s confidence intervals.
18
Figure 1.7: Apparent iron(II) concentrations (dashed) compared with iron(II) concentrations mea-
sured after a DTPA correction (solid), shown with the oxycline depth (dotted). Iron(II) concentra-
tions in the ETNP follow a similar pattern to previous reports, but are approximately 10% lower.
Error bars show 2s confidence intervals.
Fig. 1.7 shows the results of two depth profiles of iron(II), collected at 20

N, 107.1

W in the
ETNP, with and without DTPA correction. As has been previously reported, a peak in iron(II)
coincides with the secondary nitrite maximum, although the DTPA correction indicates that the
true concentration of iron(II) is 10% lower than previous estimates. The DTPA corrected profile
also indicates a sharper upper bound to the iron(II) feature, at least in this location. The dotted
horizontal line indicates the depth of the oxycline, determined by the method of Ohnemus et al.
[41]. Both the iron(II) and the interferent were only observed at depths with no detectable oxygen.
19
1.5 Discussion
Figs. 1.1 and 1.3 show that the addition of EDTA to seawater containing vanadium(IV) or iron(II)
inhibits the reaction with luminol, preventing iron(II) from generating a chemiluminescence signal.
The addition of DTPA, however, inhibits only the reaction of iron(II) with luminol, while leaving
the reaction of vanadium(IV) unchanged. DTPA is therefore more useful as a masking reagent
in environments in which reduced vanadium may be implicated, especially in sulfidic settings or
near strongly reducing sediments [42]. DTPA is also preferred over EDTA due to its stronger and
faster interactions with iron(II). Correcting for the presence of interferents using DTPA requires
two measurements and thus increases the uncertainty of the measurement. Moreover, a significant
chemiluminescent signal is observed coming from iron(II)-free seawater, due to a ROS formed
within the luminol reagent itself [32]. It is recommended to measure a blank in between every
sample in order to correct for drift in that blank. When a DTPA-correction is used, the noise in
an individual measurement depends on the noise from an untreated sample, a DTPA-treated sam-
ple, and from the DTPA reagent blank, which is the increase in signal caused by adding DTPA
to iron(II)-free seawater, as seen in Fig. 1.1. Uncertainty in the estimation of the DTPA blank
will also cause a systematic bias in the calculated iron(II) concentrations of all samples measured
within a certain period. For that reason, we recommend performing replicate measurements of the
DTPA reagent blank, in addition to using the highest-purity reagents available and performing all
handling with trace-metal clean plastic labware. The difference in the effect of DTPA and EDTA
on vanadium(IV) can be used as a test for the presence of reduced vanadium, since if vanadium(IV)
is present the signal after DTPA-treatment should be higher than that after EDTA-treatment. This
comparison suffers from additional noise relative to a typical measurement of iron(II), since it re-
quires estimation of reagent blanks for both DTPA and EDTA, and errors in those estimates will
affect multiple samples. In a number of samples, it was observed that the addition of EDTA de-
creased the signal more than the addition of DTPA, as shown in Fig. 1.2, which would be consistent
with the presence of vanadium(IV). However, this effect was not particularly consistent, and was
not, for instance, observed in samples near the maximum in iron(II), where vanadium(IV) might
20
be expected to accumulate, as a result of either in situ reduction or transport from shelf sediments.
The reagent blank associated with EDTA tends to fluctuate as well, possibly due to interactions
with metals and ROS. In many samples, the addition of EDTA did not eliminate all of the chemi-
luminescence signal (data not shown), so while we cannot rule out the presence of vanadium(IV)
in oxygen deficient waters, a large fraction of luminol interference is coming from a compound
which is not masked by either DTPA or EDTA. An alternative hypothesis which this data does not
rule out is that the interfering signal is the result of a naturally occurring iron(II)-ligand complex
which does not exchange with DTPA or EDTA on the time scale of our measurements [43]. The
effective formation constants (after correcting for the effects of Ca and Mg) of the iron(II)-DTPA
and iron(II)-EDTA complexes are 108.24 and 105.77, respectively [44]. As the concentrations of
DTPA and EDTA used are many orders of magnitude higher than the concentrations of naturally
occurring ligands, such naturally occurring ligands would have to be much stronger than typical
Fe(II) chelators (which are generally weaker than Fe(III) chelators) in order to compete with DTPA
and EDTA at thermodynamic equilibrium. For instance, DTPA at the concentrations we used could
rapidly remove Fe(II) from the colorimetric reagent ferrozine in seawater (data not shown). The
diel study in Fig. 1.5 shows the changing availability of iron in a coastal setting over a single day.
Photoreduction can produce significant diel variability in iron(II) concentrations during the day,
and can lead to accumulation of a large fraction of dissolved iron in the ferrous form, approaching
100% in at least one sample. Given the fast oxidation kinetics of iron(II) observed in Fig. 1.6, it
implies that there must be a significant pool of photoreducible iron(III) in the particulate phase.
Previous studies have found a correlation between labile particulate minerals and dissolved metal
concentrations following irradiation [45], which suggests that in Big Fisherman’s Cove, a signifi-
cant amount of iron minerals may be relatively fresh oxyhydroxides. The increase in total dissolved
iron concentrations throughout the day indicates that photoreduction and abiological processes are
the dominant controls on iron cycling, as opposed to biological uptake and remineralization. In
contrast a previous study conducted 1.4 km offshore from Big Fisherman’s Cove found decreasing
21
dissolved iron during the day [46], suggesting that the intense photodissolution seemingly occur-
ring in the cove may be limited to nearshore areas. Within the ODZ, DTPA correction indicates
that iron(II) concentrations there may be somewhat lower than previously estimated, although the
shape of the iron(II) feature is similar. It does appear at this location that the interferent may accu-
mulate at shallower depths than iron(II), since the upper boundary of the true iron(II) concentration
is considerably sharper, reflecting a combination of in situ cycling and transport from the continen-
tal shelf. This interferent is likely not vanadium(IV), as discussed above, and other potential metal
interferents are unlikely to accumulate in ODZs [24]. The extremely low oxygen concentrations
also indicate that ROS formation is unlikely. The most plausible explanation is that the interferent
is the result of an organic compound susceptible to oxidation by dissolved oxygen, and which can
therefore accumulate throughout the vertical extent of oxygen depleted waters.
1.6 Comments and Recommendations
Like all procedures to measure trace metals, all materials used need to be thoroughly cleaned.
Polyethylene and Teflon bottles can be cleaned by an overnight soak in a bath of 5% Citranox
(Alconox) followed by a 7 day soak in 10% hydrochloric acid. Thorough rinsing and soaking in
high-purity water or filtered, oligotrophic surface seawater is recommended in order to prevent
acid adsorbed to the bottles from affecting the pH. This is particularly important for Teflon bottles
which are used to contain seawater samples. Pipette tips should be rinsed with 5% trace metal
grade HCl and high purity water before being used. The trace metal cleanliness, and the reagent
blanks, of commercially available DTPA is generally better for the free acid, rather than salts with
calcium or sodium. It is recommended to dissolve the highest purity available free acid of DTPA
in high-purity water with the addition of trace metal grade sodium hydroxide. Blanks for DTPA
tend to decrease over time, presumably due to oxidation of whatever interfering impurities are
present, so it is sometimes helpful to prepare reagents several weeks in advance. The FeLume
itself requires care to ensure cleanliness. Flushing with pH 2 trace metal grade HCl can clear away
22
any contamination, or salts which have built up in the mixing chamber. This tends to drastically
increase procedural blanks for some time, unless the instrument is then thoroughly rinsed with
clean seawater. Iron(II) free seawater, for cleaning or for use as blanks, can be acquired by aging
filtered seawater. Surface seawater stored in darkness for several days works well. Trace metal
clean sampling should be used to prevent contamination by iron(II) or other interferents. Sampling
bottles should be acid-washed carefully and then soaked in clean seawater in order to avoid pH
changes. We recommend taking procedural blanks in between every sample, in order to detect
drift in the instrument or abrupt changes in the flow rate, such as those resulting from entraining
air bubbles in the mixing chamber or stretching in the peristaltic pump tubing, which need to be
identified as soon as possible. Reagent blanks are also important, as an erroneous measurement
can bias all samples measured in a run. Reagent blanks should be measured at least daily, by
measuring the difference between a blank of iron(II) free seawater and the same seawater with the
appropriate amount of reagent added to it. In measuring the reagent blanks for chelators to serve as
masking ligands, it is important that the aged seawater not be contaminated, since that will result
in an underestimate of the reagent blank.
23
Introduction to Chapter 2
After correcting the method for measuring iron(II), the next step was to apply that method to study
iron dynamics in an ODZ. The majority of data in this chapter was collected during a 2016 cruise
to the ETNP, in collaboration with the Mulholland lab at Old Dominion University.
One major goal of this study was to determine the differences in iron(II) values calculated using
the DTPA correction of Chapter 1 as opposed to the previous method, which previous ODZ studies
had used. However, this study also investigated an iron cycling process called the shelf to basin
shuttle, which was believed to be important in this region. Under this paradigm, iron(II) would
diffuse from reducing sediments on the continental shelf into the bottom waters there, and then be
transported offshore along isopycnal surfaces. That iron(II) would then be oxidized anaerobically
(the subject of Chapter 3) and scavenged to form a deep plume of iron off the continental slope.
This chapter evaluates that hypothesis in the context of the ETNP ODZ.
Using the DTPA correction significantly changed the iron(II) concentrations calculated. Con-
centrations were lower throughout the region, and the peaks in iron(II) in depth profiles were
sharper. However, the iron(II) features were still found at the same locations, suggesting that pre-
vious studies were still qualitatively reliable.
The cruise in 2016 did not visit the continental shelf, but bottom waters were sampled for
iron(II) on two other cruises, and those measurements were included in this chapter. Those sam-
ples, from stations near the Northern and Southern boundaries of the ODZ, both contained iron(II)
in concentrations much higher than anything seen offshore, consistent with the idea that the sedi-
ments were the iron(II) source.
24
The peaks in iron(II) were centered on the same isopycnal surface throughout the region, as
would be expected if it was transported through mixing. Manganese also had similar peaks in its
profiles along the same isopycnal. Manganese, like iron, can be released by reducing sediments,
and was likely controlled by similar processes. Manganese features similar to these had been
previously seen in the Arabian Sea, but are absent in the ETSP. This is likely the result of higher
fluvial inputs in Mexico compared to Peru, but it was an independent line of evidence that the
iron(II) feature was sourced from shelf sediments.
Several samples were also collected from deep waters below the ODZ. Despite the elevated
oxygen concentration at depth, iron(II) concentrations were much higher than would be expected
in the open ocean, suggesting that deep plumes were forming in these areas, and that the shelf to
basin shuttle in this region was fully active.
This chapter is currently under review in the journal Geochimica et Cosmochimica Acta.
25
Chapter 2
Iron and manganese accumulation within the Eastern Tropical
North Pacific oxygen deficient zone
2.1 Abstract
The Eastern Tropical North Pacific (ETNP) contains the largest oxygen deficient zone (ODZ) in
the modern ocean. We determined dissolved concentrations of iron, iron(II), and manganese from
three cruises in the region. Similar to other reported ODZs, iron(II) was highest in the region
associated with the secondary nitrite maximum. Its main source was likely lateral advection of
water overlying reducing shelf sediments within a narrow density range centered on the potential
density anomaly of 26.5 kg/m
3
. This density horizon is similar to the Eastern Tropical South
Pacific (ETSP) and reflects the intersection of the same density range with a large fraction of the
continental shelf bottom waters. We also observed subsurface maxima of dissolved manganese
in this density range, in contrast to the ETSP. Deep waters were enriched in Fe within the ETNP,
analogous to other eastern boundary upwelling systems as well as the Arabian Sea. We argue that
in these systems, reducing conditions on the shelf and overlying water column facilitate a robust
shelf to basin shuttle of iron, moving iron from the coastal margin to deep plumes. Manganese is
also transported offshore in the core of the ODZ, and the relationship between iron(II), manganese,
and nitrite is remarkably similar between the ETNP, ETSP, and Arabian Sea. The exception is
26
that manganese supply from the Peruvian shelf is less pronounced than in the other two ODZs,
potentially reflecting the absence of large rivers in the Peruvian system.
2.2 Introduction
Iron, an essential micronutrient to all living organisms, limits primary productivity in large frac-
tions of the modern ocean (e.g. [1]). The ferric (+3) oxidation state, the thermodynamically stable
form of iron in most of the ocean [2], is highly particle reactive and removed from the dissolved
phase relatively rapidly, although its solubility can be enhanced by complexation with organic lig-
ands. Iron(II), the other prevalent redox state observed in the ocean, is considerably less particle
reactive, although it may be more bioavailable [8–11]. However, under typical oceanic conditions
of temperature, dissolved oxygen, and pH, iron(II) oxidizes spontaneously and rapidly [4–7], so
modeling this process requires understanding both the sources of iron(II) and the ways iron(II) can
be stabilized.
Oxygen deficient zones (ODZs) are open ocean regions in which the concentration of dissolved
oxygen drops to negligible levels, and heterotrophic respiration is primarily carried out by denitri-
fication [47]. Oxygen depletion typically occurs due to a combination of sluggish circulation in the
shadow zones beneath eastern boundary currents, coupled with high surface productivity leading
to increased respiration at depth [48]. ODZs have also been recognized as a potential source of
both total iron and iron(II) to the oligotrophic ocean, since the absence of oxygen enables iron(II)
to be stabilized [3, 24, 29, 31, 49–53]. ODZ’s are areas where the redox chemistry of the oceans
are pronounced affecting the distribution and speciation of many chemical elements.
ODZs have long been recognized as important sinks for fixed nitrogen, due to denitrifiers con-
suming nitrate and producing dinitrogen (e.g. [54]). Our understanding of these processes has
increased significantly as additional nitrogen loss pathways have been identified (e.g., anammox,
autotrophic denitrification, dissimilatory nitrate reduction to ammonium) [55–59]. Other elements
27
have also been implicated in redox cycling in these regions, including iodine [60, 61] and sulfur
[62].
Iron(II) plumes have been observed in all three modern ODZs [24, 49, 50]. The plumes typi-
cally occur at the same depth as the secondary nitrite maximum, a location of high rates of denitri-
fication. Cutter et al. [31] also showed that the iron(II) feature coincided with a plume of iodide,
and was located on an isopycnal surface connected to bottom waters overlying reducing sediments
on the Peruvian shelf. The concentration of iron(II) in the plume decreases offshore more rapidly
than other tracers, suggesting an in-situ iron(II) loss pathway [31]. This hypothesis is supported
by measurements of particulate iron speciation along the same transect, suggesting iron oxidation
as a result of nitrogen cycle processes [52].
Most previous studies of iron(II) plumes in ODZs have employed luminol chemiluminescence
to measure iron(II). Recently it was recognized that this method is vulnerable to interference from
other chemicals in seawater, which can lead to an overestimation of iron(II) concentrations in some
samples [29]. This study is the first oceanographic survey to use a modified method [63] to correct
for this artifact.
In addition to plumes within the ODZ, plumes of both dissolved iron and iron(II) have been
observed beneath the ODZs in the Arabian Sea and Peru [15, 64]. These plumes occur at depths
greater than 1000 m, significantly deeper than the anoxic zone, where oxygen concentrations range
from 50 to 150 μM [65]. Deep iron plumes can be formed by hydrothermal inputs (e.g. [15]) or
through reversible scavenging and dissolution in continental slope sediments [64, 66, 67]. We can
distinguish different plume sources using manganese, since manganese is enriched in hydrothermal
plumes but low in slope-derived plumes [15, 64]. Deep iron features associated with the ETSP
ODZ appear to be slope-derived, while similar features in the Arabian Sea are influenced by both
processes [64]. Landing and Bruland [68] collected some samples in the ETNP above 2000 m,
but the deep ETNP has not otherwise been sampled. In this study we report the first trace metal
concentrations from waters collected> 2000 m below the ETNP ODZ, to investigate whether deep
plumes exist here, and to determine their sources.
28
The distribution of manganese in the ocean is significantly different from the distribution of
iron, and its comparative geochemistry with iron in ODZs can help to distinguish processes which
influence both. Mn(II) is highly soluble in seawater, like Fe(II), and Mn(IV) is more thermody-
namically stable and more particle reactive, like Fe(III), [69]. Unlike Fe, Mn can be found in an
intermediate oxidation state, Mn(III), which was suggested to be a significant component of the
dissolved manganese pool if organic ligands are available to stabilize it [70]. Mn oxidation (to
either the +3 or +4 state) in the presence of oxygen is considerably slower than iron oxidation
under most circumstances, and Mn oxidation does not take place abiotically [71]. Dissolved Mn
is generally high in the surface ocean, due to aeolian and riverine inputs [68], in addition to pho-
toreduction of manganese oxides [71–73], while older and deeper waters typically have very low
manganese concentrations. All currently known Mn oxidation processes in the ocean require oxy-
gen [69, 71, 74]. However, it is unclear whether the lack of oxygen leads to Mn accumulation in
ODZs. High Mn concentrations have been observed in the anoxic core of the Arabian Sea ODZ
[75, 76] as well as in the oxygen minimum zone off the southern California coast [77, 78], but
Mn concentrations in the ETSP are very low, comparable to oxygenated deep waters [79]. This is
likely due to differences in Mn supply to these regions, rather than being driven solely by oxygen.
The ETNP ODZ is the largest in the world, and the least studied from the perspective of trace
elements. While both the Peruvian and Arabian Sea ODZs have been the subject of GEOTRACES
surveys as well as additional trace metal projects, relatively few studies have examined the distri-
bution of trace metals in the ETNP. Nutrient cycling in the ETNP is significant for several reasons,
including that the region is home to some of the largest and most productive fisheries in the world
[80], both from an economic perspective as well as a source of protein to a large number of people.
The region is influenced by issues such as ocean warming, acidification, and deoxygenation [81],
but also faces regional challenges. Two large rivers discharge into the area, the Rio Balsas and the
Rio Grande de Santiago. The Balsas is the largest river that flows into the Pacific coast of North
America and is closest to our sampling stations. It will likely be subject to water diversion in the
future to provide a new fresh water source to Mexico City [82]. The extent to which the Rio Balsas
29
supplies nutrients, including trace elements, to the ETNP has not been well characterized, making
the ecological consequences unpredictable.
In this study, we analyzed the spatial and redox distribution of iron in the ETNP ODZ, and
its relationships with other key parameters, including connections to the redox cycling of oxygen,
nitrogen, and manganese, in order to identify the processes controlling iron cycling in the region,
and to compare this region to the ETSP. Additionally, this is the first large scale study to report
iron(II) values using diethylenetriamine pentaacetic acid (DTPA) as a masking ligand to distinguish
Fe(II) from non-metal luminol-reactive interferences in the method [63].
2.3 Methods
2.3.1 Sampling sites and trace metal sample collection
Samples were collected from 18 stations on three cruises to the ETNP ODZ. Samples were col-
lected from 16 stations during a cruise aboard the NOAA ship Ronald H. Brown in March-April
2016. Stations 1-9 were occupied during an alongshore transect starting west of Baja California
and traveling southeast, approximately parallel to the shoreline. The subsequent stations were then
collected on an offshore transect, angling northwest through the core of the ODZ, finishing near the
Revillagigedo archipelago. An additional station was occupied during a March 2018 cruise aboard
the R/V Roger Revelle, near the mouth of the Gulf of California over a wide stretch of reducing
continental shelf sediments. A coastal shelf station south of Manzanillo, Mexico was occupied in
March 2012 during a cruise aboard the R/V Thomas G. Thompson. Trace metal sampling for all
cruises was performed using the same methods. Station coordinates for all cruises, as well as some
regional points of interest, are shown in Figure 2.1.
Samples were collected in 5 L Teflon-coated external spring “Niskin-type” bottles (Ocean Test
Equipment) mounted on a powder-coated trace metal clean rosette (Sea-Bird Electronics) using
a pre-programmed Auto-Fire Module suspended from an AmSteel cable. Prior to sampling, the
rosette was allowed to degas for 30 minutes in the anoxic waters of the ODZ, in order to remove
30
Figure 2.1: Station map. Stations occupied during the 2016 (NOAA ship Brown) cruise (circles),
2018 (R/V Revelle) (squares), and during a 2012 cruise (R/V Thompson) (triangles). Major sites of
known or suspected trace element inputs are marked as well (stars), including the R´ ıo Balsas river
mouth, near station 7, the Revillagigedo archipelago, near station 16, and two known hydrothermal
vent fields along the East Pacific Rise (EPR).
31
oxygen adsorbed to the bottles which might affect the redox chemistry. After sample collection,
the bottles were taken to a laminar-flow clean van, pressurized with filtered nitrogen, and filtered
with 0.2 μm AcroPak cartridge filters.
Most samples for Fe(II) were measured using the modified luminol method described in Bolster
et al. [63], based on the method of King et al. [5]. The modified method uses DTPA as a masking
ligand for Fe(II), allowing the concentration to be determined by the difference between an un-
treated and a DTPA-treated sample. In this way, it is possible to correct for the influence of other
interfering compounds that also interact with luminol, which have been reported in the ODZ [63].
This DTPA correction was performed on all samples in this paper, with the exception of samples
from the 2012 cruise. Samples for dissolved trace metals were stored in low density polyethylene
bottles, acidified to pH 1.7 using Optima grade hydrochloric acid (Fisher), to a final acid concentra-
tion of 20 mM, and stored at room temperature for several months before analysis. Samples were
preconcentrated using a seaFAST system (ESI), as described in Rapp et al. [39]. Samples were
quantified using isotope dilution for iron and standard additions for manganese. Analyses were
performed on an inductively coupled mass spectrometer in medium resolution mode (Thermo El-
ement 2), using indium as an internal standard. The accuracy of the analysis was assessed by the
measurement of the GSC and GSP standard solutions. Measured concentrations in GSC were 1.97
± .11 nM Mn and 1.566 ± .017 nM Fe (n = 5), statistically indistinguishable from the consensus
values of 2.180 ± 0.075 nM Mn and 1.535 ± 0.115 nM Fe. For GSP, measured concentrations
were 0.724 ± .044 nM Mn and 0.149 ± .012 nM Fe (n = 5), statistically indistinguishable from the
consensus values of 0.778 ± 0.034 nM Mn and 0.155 ± 0.045 nM Fe.
2.3.2 Nutrient sampling
For hydrographic characterization of the sections on the NOAA ship Ronald H. Brown, as well as
for a principal component analysis (described below), samples were collected from a non-trace-
metal clean rosette. Temperature, salinity, and oxygen were determined from a Sea-Bird SBE
11plus CTD and a model 43 dissolved oxygen sensor, attached to a sampling rosette for dissolved
32
nutrient concentrations. Nitrate, nitrite, and phosphate were measured onboard as described in
Selden et al. [83].
2.3.3 Principal component analysis
In order to assess the relationships between the variables measured, a principal component analysis
was performed on the data collected aboard the Brown (stations 2-16). The analysis was performed
using the statistical programming language R [38]. Nutrient measurements were associated with
trace metal concentrations using a nearest-neighbor interpolation, removing depths at which there
was no nutrient measurement within 5 m. Data were standardized for a principal component anal-
ysis and visualizations generated using the factoextra package [84]. For comparison, an identical
analysis was performed on an ETSP data set from the GEOTRACES Intermediate Data Product
[65] with iron(II) data from Cutter et al. [31] and additional iron and manganese data from Resing
et al. [15]. In order to ensure an adequate comparison between the cruises, the analysis was re-
stricted to the upper 1000 m of the water column, since more samples were collected in the ETSP.
The range therefore includes a full gradient of oxygen concentrations both above and below the
ODZ, to visualize the trends across those water masses. The code to reproduce this analysis is
provided in the supplemental data.
2.4 Results
2.4.1 Physics, nutrients, and oxygen
Section plots of oxygen, nitrate, and nitrite for the alongshore and offshore transects are shown in
Figures 2.2 and 2.3. Throughout most of the study region, dissolved oxygen concentrations were
undetectable from the oxycline, found at approximately 70 m at most stations, to a deeper, less
well-defined boundary which ranges from 500 to 700 m. A secondary nitrite maximum, similar to
those seen in other ODZs, was located within the upper ODZ. Potential density contours showed
that the nitrite maximum occurred in a narrow density range centered around 26.5 kg/m
3
. This
33
Figure 2.2: Dissolved nitrate, nitrite, and oxygen in the alongshore transect in 2016. Dissolved
oxygen contour lines are shown in white. The dark black line shows the 26.5 kg/m
3
isopycnal.
corresponded to a local minima of nitrate in the water column, although nitrate was never fully
depleted at any station.
Trace metal data from both transects is presented in Figures 2.4 and 2.5. Total dissolved Fe
increased with depth at most stations, consistent with the nutrient-like distribution it displays else-
where. Distributions of Fe(II) and Mn were more complicated, and likely influenced directly by
the ODZ. Similar to the secondary nitrite maximum, the peak dissolved Fe(II) concentrations were
also centered around the potential density anomaly of 26.5 kg/m
3
, similar to what has been ob-
served in the ETSP [31].
At the northern end of the nearshore transect, outside the region of the oxygen deficient zone,
concentrations of Fe(II), dissolved Fe, and dissolved Mn were all fairly low, but concentrations
increased towards the southern part of the transect where the ODZ was most pronounced (Figure
2.4). At nearshore stations, manganese profiles had characteristic surface maxima, reflecting inputs
from dust and margin sources [68] or photoreduction of manganese oxides [71–73]. Dissolved
iron and manganese concentrations were highest at station 7, which was the station closest to
the Rio Balsas, presumably the primary source for manganese in the region. Along the offshore
transect, Fe(II) and Mn concentrations decrease from east to west within the ODZ, as shown in
34
Figure 2.3: Dissolved oxygen, nitrate, and nitrite in the offshore transect in 2016. White contour
lines show dissolved oxygen concentrations, and the dark black line is the 26.5 kg/m
3
isopycnal.
Figure 2.4: Trace metal data in the alongshore transect in 2016. White contour lines show dissolved
oxygen concentrations, and the thick dark line shows the 26.5 kg/m
3
isopycnal. The R´ ıo Balsas
outflow is located between stations 7 and 8.
35
Figure 2.5: Trace metal data in the offshore transect in 2016. White contour lines show dissolved
oxygen concentrations, and the dark line shows the 26.5 kg/m
3
isopycnal. Note that the color
scales are different from those in Figure 2.4
Fig. 2.5, as would be expected if they are sourced from shelf sediments and oxidized in the water
column. Stations further west, along the offshore transect had subsurface manganese maxima,
while stations closer to shore had higher concentrations of manganese in surface waters, suggesting
either manganese loss from surface waters or a change in the manganese source. In the ODZ at
those offshore stations, dissolved Fe(II) concentrations were typically between 10% and 50% of
the total dissolved Fe (median 21%), with the highest concentrations observed near the core of the
ODZ. At stations 15 and 16, near the Revillagigedo archipelago, iron concentrations were higher
than at other open ocean stations, potentially indicating sedimentary sources near those islands. At
the Gulf of California station occupied during March 2018, the iron(II) concentrations in bottom
waters were extremely high (Figure 2.6a). The continental shelf in this area is extremely broad,
similar to shelves in other regions have been shown to be important for supplying Fe to the surface
in nearshore environments [85–87]. At the station occupied during March 2012, (Fig. 2.6b),
Fe(II) concentrations were elevated, as would be expected if sediments are the source of the Fe(II)
observed in the core of the ODZ.
36
Figure 2.6: Iron(II) and dissolved oxygen profiles at (a) the San Blas Basin in 2018. The dashed
line shows the bottom depth of the station. Note that iron(II) concentrations are plotted on a log
scale. Iron(II) concentrations shallower than 80 m were below the limit of detection. (b) The shelf
station in 2012 near the mouth of the Rio Balsas.
2.4.2 Deep plumes in ETNP
Sampling below the ODZ was carried out only at stations 6, 12, and 16 (Fig. 2.7). In all three pro-
files, we see iron features deeper than the anoxic layer which indicate deep plumes. At station 6,
closest to the shelf, manganese concentrations were low, but iron(II) was found in oxygenated wa-
ters, similar to the plume observed on the Peru margin during the GEOTRACES GP16 expedition
[15]. In contrast, no iron(II) was observed below the ODZ at stations 12 and 16, and manganese
concentrations were elevated below 1700 m at station 12, consistent with a hydrothermal plume.
Hydrothermal vents are known to be present in the region, at 2000 and 2600 m in the East Pacific
Rise (EPR) at 21°N [88, 89], as well as at 2600 m at 18°N [89, 90] which could be the source of
those features. At 2000 m depth, overall flow in these regions is southward [91], which may help
explain the broad peak of manganese at those depths at station 12.
37
Figure 2.7: Iron, manganese, iron(II), and oxygen concentrations at (a) station 6, (b) station 12,
and (c) station 16.
38
2.5 Discussion
2.5.1 Implications of correcting for interferences with DTPA
After using DTPA to correct for interfering compounds, an iron(II) maxima coinciding with the
secondary nitrite maximum is still apparent in ODZ samples, which is qualitatively similar to
previous studies [24, 49]. However, the concentration of Fe(II) at that maximum in offshore sta-
tions is 20-40% lower than the apparent concentration without that correction (data not shown),
highlighting the importance of this correction for quantitative studies. Fe(II) features also appear
considerably sharper [63], indicating that whatever compounds cause the interference are more
broadly distributed within the anoxic layer.
2.5.2 Iron transport within the 13CW water mass
The impact of water masses on Fe(II) transport can be seen clearly in temperature-salinity space.
Figure 2.8 shows a temperature-salinity plot for the entire 2016 cruise, with the highest Fe(II) con-
centrations (>0.5 nM) shown as circles. All but one of these points cluster at a salinity maximum
with a temperature of 13°C. This corresponds to a water mass referred to as 13°C water (13CW)
[92], or Equatorial Subsurface Water (ESSW) [93], which originates in the north Tasman Sea and
spreads into both the ETNP and ETSP. This water mass has low dissolved oxygen concentrations
due to a long residence time in isolation from the surface as well as high salinity which inhibits
mixing with the oxygenated waters overlying it [92]. Moreover, the potential density range asso-
ciated with this water mass leads it to intersect a large fraction of the shelf [92]. As a result of this
combination, 13CW is important for both in-situ processes (denitrification and iodate reduction)
and lateral advection of materials from shelf sediments. Moreover, the 13CW is common to the
ETNP and the ETSP, explaining why the maxima in nitrite and Fe(II) are clustered over the same
potential density range in both regions, despite nitrite’s higher turnover rate and in-situ production
[31].
39
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 10
20
30
33.5 34.0 34.5
Salinity (PSU)
Temperature (deg C)
Figure 2.8: Temperature-salinity plots for all stations occupied during the 2016 Brown cruise
(grey), and all samples with>0.5 nM Fe(II) (black). Almost all elevated iron(II) concentrations
are associated with the 13CW endmember.
40
This density range also contains a maximum in excess iodine, highlighting the importance of
lateral advection in this water mass [92]. Excess iodine is the amount of iodine in excess of the
mean ocean concentration, which is fairly conservative. It has been argued that excess iodine in
the ETNP must come from a sedimentary source [92]. There is also a strong correlation between
excess iodine at Fe(II) in the ETSP [31]. Unfortunately, Fe(II) and iodine have not been measured
on common transects in the ETNP, but excess iodine is also associated with 13CW in this region
[61, 92], similar to the Fe(II) distribution reported here. While the water mass originates far west of
the ODZs, the high concentrations of Fe(II) and excess iodine probably arise from eddy-mediated
transport of benthic-derived material from the shelf [92].
2.5.3 Comparison of ETNP and ETSP ODZs
Both ODZs in the Eastern Tropical pacific share common features. Fe(II) maxima are coincident
with nitrite maxima. Moreover, dissolved Fe and Fe(II) concentrations are high on the shelf,
but decline past the shelf slope break. This probably reflects dynamic redox cycling and iron
trapping between the shelf sediments and water column. Beyond the shelf-slope break, there is net
oxidative scavenging of Fe(II) by nitrate or nitrite and net export of Fe beyond the shelf [94]. These
features have been identified as key components of the shelf to basin shuttle (SBS), a mechanism
that describes how reducing conditions on coastal shelves associated with ODZs lead to enhanced
transport of iron to the ocean interior [95]. This study, combined with earlier work [64, 95], suggest
that the SBS is common to both of these ODZs and the Arabian Sea as well. High benthic Fe(II)
fluxes have been measured off the Peruvian Shelf [96] and the Pakistan margin [97]. Presumably,
benthic Fe fluxes are also high in our study region to sustain the high shelf water column Fe(II)
concentrations (Figure 2.6). Iron flux measurements have not been made on shelf sediments near
our study site, but fluxes measured in the Guaymas Basin in the Gulf of California are comparable
to Peruvian fluxes [98]. A global model indicates high benthic Fe fluxes throughout the ETNP [99].
Interestingly, this model predicts higher fluxes in this region than off of Peru. The comparative
distribution of iron and manganese in ODZs also provides important information about the sources
41
●
●
● ● ●
● ●
● ●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
O x y g e n
N i t r i t e
P h o s p h a t e
N i t r a t e
M a n g a n e s e
I r o n
− 2
0
2
− 2 0 2 4
D i m 1 ( 4 5 . 2 % )
D i m 2 ( 2 3 . 6 % )
● ● ● ● ● ●
I r o n ( I I )
a .
● ●
●
●
●
●
●
●
●
●
●
● ●
● ●
● ●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
● ● ●
●
●
● ● ●
●
●
●
● ● ●
●
● ● ● ● ● ● ●
● ●
● ● ●
●
●
●
●
●
●
●
● ● ● ● ●
● ●
● ●
● ● ●
●
●
●
●
●
● ● ● ● ● ● ●
● ● ●
● ●
O x y g e n
N i t r i t e
P h o s p h a t e
N i t r a t e
I r o n ( I I )
M a n g a n e s e
I r o n
0 . 0
2 . 5
5 . 0
7 . 5
− 2 0 2
D i m 1 ( 4 1 . 8 % )
D i m 2 ( 3 4 . 6 % )
b .
Figure 2.9: Dissolved sample measurements< 1000 m in the ETNP (a) in 2016 and the ETSP (b)
during the GEOTRACES GP16 cruise plotted on the first two coordinates of principal component
analyses. In the ETNP, dissolved manganese closely tracks iron(II) and nitrite, while total dissolved
iron has a more nutrient-like distribution. In the ETSP, dissolved iron was closely coupled with
nitrite and iron(II), but not strongly related to manganese, which showed surface maxima.
of trace elements. The subsurface manganese maxima in the ETNP reported here is coincident with
Fe(II) and nitrite, which has also been observed in the Arabian Sea [76]. The close correspondence
between those three variables is apparent in the principal component analysis shown in Figure 2.9.
The distribution of Mn within the ETSP is very different from the ETNP. Subsurface Mn max-
ima were not observed off of Peru, with the highest concentrations always at the surface. Thus,
although Fe(II) and Mn are closely related in the ETNP, they are not correlated in the ETSP (Fig-
ure 2.9). Surprisingly, Mn concentrations on the Peruvian shelf are low (less than 2 nM; [15, 79])
compared with the Mexican shelf (up to 10 nM), especially in the Rio Balsas zone and near the
Rio Grande de Santiago. In Peru, rivers are smaller and discharge is more episodic [100]. Offshore
in the ETNP, Mn and Fe(II) are probably sourced from reducing shelf sediments and advected off-
shore, a process proposed to account for the excess iodine seen at the same depths in this region
[61, 92]. As argued in those studies, all of these species are enriched over a relatively narrow
density range that happens to span a large fraction of the Mexican continental shelf seafloor. Both
Mn and Fe likely accumulate at the sediment-water interface under oxidizing conditions and are
released to overlying waters during reducing conditions. In Peru, this redox cycling results in large
42
Fe fluxes, but the oxidation part of the cycle is driven by nitrate [51] which does not oxidize Mn(II).
Therefore Mn does not participate significantly in the shelf redox cycle. Shifts in the redox condi-
tions off Peru leading to oxygenation of the shelf water column are driven by El Ni˜ no and occur
infrequently [101]. One explanation for the appearance of subsurface Mn plumes off Mexico is
that high river discharge of dissolved and particulate Mn in Mexico (relative to Peru), coupled with
periods when the shelf water column is oxygenated, could concentrate Mn at the sediment-water
interface on the shelf. Subsequent reducing conditions could release this Mn along the same po-
tential density surface as Fe(II). Such processes are presumably less important in the persistently
reducing Peruvian shelf, and the fluvial sources there are smaller, so Mn plumes coincident with
Fe(II) are weaker. The main problem with this hypothesis is lack of data in the ETNP. We do not
know how much temporal variability there is in the redox state of the Mexican shelf, nor the Mn
flux from rivers in the area.
2.5.4 Origin and significance of deep iron features
In some deep samples, where dissolved manganese was elevated, as seen below 2000 m at station
12, hydrothermal contributions may be important. However, hydrothermal inputs cannot explain
many of the other deep samples with elevated iron concentration.
In the SBS, iron transported offshore from the shelf is removed by scavenging and settles on
shelf and abyssal sediments, where it is resuspended and forms deep plumes of Fe. Such plumes
have been observed off the Peruvian slope [15] and the Arabian Sea [64]. Deep plumes also arise
from productive eastern boundary regions even when they are not ODZs, including Namibian [102]
and Mauritanian [103] continental margins. In each case, manganese is decoupled from these SBS
plumes, presumably due to slower redox kinetics. This is a useful diagnostic feature, because
iron plumes derived from hydrothermal sources typically contain high Mn [64]. Deep water iron
concentrations beneath the ODZ (between 800 and 3000 m) at stations near the shelf (e.g. Station
12) are comparable with deep samples off Peru [15], indicating that the final component of the SBS
is operating in this region. These findings support the suggestion that it is a ubiquitous feature of
43
eastern boundary upwelling systems [64]. In the GEOTRACES GP16 section, the absence of Mn
in the Peruvian margin plume suggested that hydrothermal sources were unimportant, in contrast
to the East Pacific Rise plume, which was enriched in Fe and Mn. In the Arabian Sea, the situation
was more complicated, with evidence for hydrothermal and margin sources intermingled [64].
Although most deep samples in this dataset are low in Mn, there is some potential for hydrothermal
inputs at 2650 m at Station 12, and beneath 1000 m at Station 6.
2.6 Conclusions
This study significantly increases our understanding of redox cycling in the ETNP, an understudied
region relative to the ETSP and the Arabian Sea. Iron redox cycling within the core of the ODZ re-
sembles those two systems in many respects. In particular, it appears that the well-defined maxima
in Fe(II) is influenced at least in part by strong lateral advection processes from the adjacent mar-
gin. The shelf waters also show important similarities with the Peruvian shelf, with exceptionally
high concentrations of dissolved Fe. The deep waters underlying the ODZ have elevated concen-
trations of Fe, indicating the presence of a deep plume, similar to other eastern boundary systems,
potentially also influenced by hydrothermal fluxes. Fe(II) measurements reported here for the first
time using a DTPA correction to the luminol method is similar to previous data but significantly
lower, highlighting the importance of this correction. Most significantly, the iron(II) maxima is
restricted to a narrower density range than previously observed, placing an additional constraint
on how it is formed. The Mn distribution in this region was similar to that of Fe(II), similar to
the Arabian Sea but in contrast to the ETSP, where Mn is always highest at the surface. Internal
cycling likely plays a role in this, but it is also likely to be a result of higher riverine fluxes than in
the ETSP. Deep iron plumes from eastern boundary systems are probably an important source of
deep water iron, which can be transported significant distances. Despite the differences in redox
cycles between the ETSP and ETNP, the ETNP is also a significant pathway for Fe transport to the
Pacific Ocean, and should be incorporated into models for iron cycling on larger scales.
44
2.7 Acknowledgements
The authors would like to thank the captains and crews of the R/V Thomas G. Thompson, the
NOAA ship Ronald H. Brown, and the R/V Roger Revelle. We would also like to thank the chief
scientists, Al Devol, and Gabrielle Rocap. We would additionally like to acknowledge Alejandro
Arias, Peter Bernhardt, Alfonso Mac´ ıas Tapia, Nicole Travis, Rachel Kelly, Shun-Chung Yang,
Paulina Pinedo-Gonzalez, and Alexis Floback. Funding was provided by the University of South-
ern California, and the National Science Foundation [OCE-1459584, 1636332, DEB-1542240 to
JWM and 1356056 to MRM at ODU].
45
Introduction to Chapter 3
A critical step in the shelf to basin shuttle is the oxidation of iron(II) to iron(III) oxyhydroxides,
which can sink and form deep plumes. Measurements from the ETNP in Chapter 2 suggested that
this process was occurring, but quantifying the rate is difficult. Previous studies in the ETSP had
attempted to do so using modeling approaches, which measured the rates indirectly. The goal of
this chapter was to measure the oxidation rate directly using an incubation approach.
A secondary goal of this project was to test the mechanism of iron oxidation. There were
two proposed explanations for this process. One hypothesis was that anaerobic iron oxidation was
an abiotic process between iron(II) and nitrite, with the surface of mineral particles acting as a
catalyst. The other hypothesis was that the reaction was being catalyzed by microbes using nitrate
as a terminal electron acceptor in order to extract energy from the reaction. These hypotheses could
be tested by measuring spatial patterns in the iron oxidation rate, and by the impact of different
classes of particles.
The incubations were performed by adding a stable isotope tracer, iron-57, in the +2 oxida-
tion state, and then to measure the rate of that tracer’s incorporation into iron(III) oxyhydroxides.
Those mineral particles could be collected by filtration, and then a selective leaching process could
measure the concentrations.
The Keil group at the University of Washington had developed a system which could perform
incubations in-situ, while tethered to a free drifting buoy, that could be deployed inside the ODZ
for several days. Timers on board automated the sample collection process and the addition of
the tracer to the sample. The entire system stayed at depth for the entire length of the incubation
period, ensuring that there was no contamination by oxygen, temperature, or light. The incubators
46
each contained two chambers, one of which was connected to a sediment trap, allowing the impact
of large, sinking particles to be measured relative to that of smaller, suspended particles.
These incubators were capable of providing valuable data, but few of them exist and deploy-
ments are expensive and time consuming. Shipboard incubations could supplement these mea-
surements, and would also enable experiments on the impact of filtration. However, shipboard
incubations would have a high risk of oxygen contamination. There are validated methods for
anaerobic incubations, but all standard anaerobic techniques cause a significant amount of iron
contamination, which would bias these experiments.
Several novel approaches to solve this problem were tested during a cruise in 2018, with the
results compared to the rates measured by the in-situ incubators. One method was adopted, which
used glovebags filled with filtered nitrogen which could keep samples isolated from metal contam-
ination and oxygen simultaneously.
In 2019, that method was used to measure iron oxidation rates at 8 depths over 2 stations,
generating profiles of iron oxidation rates to compare with the dissolved iron(II) profiles. At a
representative station of the oligotrophic ODZ, the half-life of iron(II) with respect to oxidation was
between 43 and 132 days. In all shipboard experiments, 0.2mm filtration inhibited iron oxidation,
consistent with both the abiotic and biological hypotheses. However, in-situ incubations with
sinking particles did not show higher rates, as would be expected if the reaction was catalyzed by
mineral surfaces. In addition, the rate of iron oxidation was elevated in regions with higher iron(II)
supply, which would be expected if iron(II) controlled the activity of iron oxidizing microbes.
These data collectively supported the hypothesis that anaerobic iron oxidation was a biologically
driven process.
This chapter is being prepared for submission to the journal Proceedings of the National
Academy of Sciences of the United States of America.
47
Chapter 3
Iron oxidation in the water column of the Eastern Tropical
North Pacific Oxygen Deficient Zone
3.1 Abstract
Oxygen deficient zones (ODZs) contain elevated concentrations of dissolved iron(II) sourced from
shelf sediments. As that iron(II) is transported offshore, it is oxidized and scavenged to the con-
tinental slope, where it forms deep plumes of dissolved iron. These processes are not well con-
strained, making it challenging to evaluate their importance in the global iron cycle. In this study,
we use a stable isotope incubation approach to directly measure the rate of iron oxyhydroxide for-
mation at two stations in the Eastern Tropical North pacific ODZ, and to test the dependence of
that process on sinking and suspended particles, in order to investigate possible mechanisms of
oxidation. At a representative station of the offshore ODZ region, we calculate iron(II) half-lives
ranging from 43 to 132 days, generally decreasing with depth in the anoxic zone. At a second
station with lower iron(II) concentrations, we find considerably slower rates. Because the reac-
tion depends on the presence of suspended particles but is not systematically affected by sinking
particles, we conclude that iron oxidation in this region is likely a microbially driven process.
48
3.2 Introduction
Oxygen deficient zones (ODZs) facilitate iron transport from continental margins to the open ocean
[64]. As a part of this process, dissolved iron(II) is converted to a particulate form and scav-
enged from the water column [95]. This scavenged iron can then generate deep plumes capable
of transporting dissolved iron over significant distances [64, 104]. The process of oxidation and
scavenging is not well quantified.
There are three ODZs in the modern ocean, located in the Eastern Tropical North Pacific
(ETNP), the Eastern Tropical South Pacific (ETSP), and the Arabian Sea. These regions gen-
erally have oxygen concentrations<10 nM, sufficient to prevent aerobic respiration [105], due to
a combination of older, less-ventilated water masses in addition to high respiration rates of organic
matter. For this reason, these processes are typically associated with upwelling regions and eastern
boundary shadow zones. Organic carbon is still available for microbial consumption, but because
oxygen is depleted, respiration in ODZs is primarily driven by nitrate reduction, leading to the
accumulation of nitrite and nitrous oxide as intermediates, and a loss of fixed nitrogen from these
areas [106].
An iron(II) feature is typically observed in the core of ODZs, typically on the same depth
horizon as the secondary nitrite maximum [24, 107, 108]. This feature is located on an isopycnal
surface connected to reducing sediments on the shelf, which are a source of iron(II) to the water
column [109]. Despite the high solubility of iron(II) in seawater, dissolved iron(II) is scavenged to
the particulate phase in the water column, as several studies have shown [51, 110]. This scavenging
is a major component of the so-called ”shelf to basin shuttle” (SBS), an important mechanism for
the transport of iron to open ocean regions through a cycle of oxidation and reduction in low
oxygen or anoxic environments [95]. There are a number of potential mechanisms to explain this
scavenging process.
One potential mechanism for anaerobic iron removal would be precipitation of iron sulfides.
Sulfate is reduced in ODZs by the ”cryptic sulfur cycle” [62], and the sulfide product is either
49
reoxidized or scavenged by other compounds, including iron(II). Particulate sulfide is present in
the ODZ, but pyrite or iron sulfides have not been detected [41, 111].
Another possibility is that iron(II) could react with chromium present in the water. Chromium(VI)
can oxidize iron(II) to form chromium(III) and iron(III), both of which are relatively insoluble
[112, 113]. This reaction, and the resulting isotope fractionation of chromium, has been used as a
paleoredox proxy e.g. [114]. This reaction does not require any solid phase catalysts, but the rates
of reactions are not well constrained in modern ODZs.
Yet another potential mechanism is abiotic iron oxidation by nitrite. This process has been
observed in freshwater and sedimentary systems, and does not require microbial activity. However,
it requires a mineral surface to act as a catalyst [115]. This mechanism has been hypothesized
in the ETSP based on the presence of particulate iron oxyhydroxides in offshore samples and a
correlation between those particles and the secondary nitrite maximum [110].
Finally, iron(II) could be scavenged through biological oxidation by microbes utilizing nitrate
as an electron acceptor. The oxidation of iron(II) by nitrate is thermodynamically favorable in
most natural systems, but proceeds very slowly without a catalyst, making it a prime candidate for
chemoautotrophic metabolisms [116]. Such metabolisms are present in many environments in the
modern era, and this metabolism may have been an important control on nitrogen availability in the
Proterozoic, when reduced iron was abundant [12]. Although several microbes have been isolated
which utilize this pathway, the exact mechanism of anaerobic iron oxidation, or gene markers for
the process, has not yet been identified [117]. However, some bacteria closely related to known
iron oxidizers are present in the waters of the Peruvian shelf [51]. This hypothesis is also consistent
with observations of particulate iron oxyhydroxide formation in ODZs [110].
Several previous studies have attempted to constrain the rate of iron loss through modeling
approaches. One study [109], using data from the GEOTRACES GP16 cruise, modeled the dis-
tributions of iron(II) and excess iodide (a semi-conservative shelf tracer) along isopycnal surfaces
in the ETSP. The study confirmed that excess iodide behaved conservatively inside the ODZ, and
50
that iron(II) was being removed more rapidly than excess iodide. This study estimated the flux of
iron(II) offshore at several stations, but did not calculate a rate of iron loss.
Another study [51] measured iron distributions on the Peruvian shelf, where the bottom waters
were enriched in both sulfide and iron(II). At the interface between the denitrifying and euxinic
zones, the iron(II) concentrations decreased drastically, suggesting that iron oxidation was happen-
ing under denitrifying conditions, rather than exclusively above the oxycline. Advection-diffusion
models were used to calculate an oxidation rate, with a half-life of approximately 20 hours in shelf
bottom waters.
In this study, we measured the oxidation rate of iron(II) in the ETNP ODZ directly, using an
incubation method with a stable isotope tracer. We report the oxidation rates at two open ocean
stations off of Mexico, and several experiments investigating the mechanism of iron loss in these
areas.
3.3 Results
3.3.1 Comparison of shipboard andin-situ incubations
In 2018, we tested several incubation methods at P2. There were two successful deployments of
the in-situ incubators and one successful set of shipboard incubations with filtered treatments. Our
unfiltered shipboard incubation rates were consistent with rates from the in-situ incubators, which
gave us confidence that it was possible to achieve anoxic and trace metal clean conditions in a
shipboard incubation. We measured oxidation rates on the order of 0.01 d
1
using both methods,
as shown in Figure 3.2.
When we returned to P2 in 2019, we completed a profile of the iron oxidation rate at that site.
We performed four sets of shipboard incubations ranging in depth from 150 to 300 m, in addition
to an in-situ incubation at 170 m, all of which yielded fairly consistent measurements, as shown in
Figure 3.3.
51
P 3
P 2
O xyg e n [ μ M ] @ d e p t h = 2 0 0
5 0
4 0
3 0
2 0
1 0
0
1 1 5 ° W 1 1 0 ° W 1 0 5 ° W 1 0 0 ° W
1 5 ° N
2 0 ° N
2 5 ° N
3 0 ° N
Figure 3.1: Map of the sampling sites in the Eastern Tropical North Pacific, overlying the average
dissolved oxygen in the region at 200 m. Data from the World Ocean Atlas 2018 ([118]).
52
Figure 3.2: (Left) Iron oxidation rate constants for in-situ incubations at 120 and 180 m depth,
including incubations with suspended particles only (medium) and suspended + sinking particles
(dark). (Right) Rate constants for shipboard incubations at 180 m, including filtered incubations
(light) and suspended particle incubations (medium).
Figure 3.3: Nitrate (top left), nitrite (top right), dissolved iron(II) (bottom left), and oxidation rate
constants (bottom right), at station P2 in October 2019, including experiments with only suspended
particles (triangles) and with sinking material in a sediment trap incubator (circles). Oxygen con-
centrations are beneath the limit of detection below 110 m
53
Figure 3.4: Incubation data from P3. Iron(II) profile (left), shipboard incubation replicates (center),
and comparison between the shorter (shipboard) and longer (in-situ) incubations (right)
3.3.2 Distribution of iron oxidation rates
The distribution of iron(II) at P2 in 2019 (shown in Figure 3.3) was similar to previously reported
profiles in this area (e.g. [24]. At this site, the iron(II) maximum was observed at 180 m, the same
depth as the secondary nitrite maximum with a potential density anomaly of 26.5 kg m
3
, with
lower concentrations deeper in the ODZ.
Between the depths of 150 and 300 m, the oxidation rate ranged from 0.0163 d
1
to 0.0052
d
1
(half-lives of 43 to 132 days, respectively), generally decreasing with depth below the iron(II)
maximum, as shown in Figure 3.3.
We also performed a deeper in-situ incubation at 400 m. The rate of oxidation was much
higher there, averaging 0.054 d
1
(13 day half-life). This high oxidation rate corresponds to a
local minimum in the iron(II) concentrations, demonstrating the impact of oxidation kinetics on
the distribution of iron(II) in the area.
At station P3, the iron(II) concentrations were much lower than at P2, due to limited supply
from reducing sediments and/or proximity to the Northern edge of the ODZ. There was some de-
tectable iron(II) below the oxycline, all below 20 pM. This is close enough to the limit of detection
54
for the iron(II) method that the overall structure of the feature cannot be determined, beyond the
presence of small amounts, as shown in Figure 3.4.
Incubations at P3 were both done at 300 m. However, the in-situ incubations were run for much
longer than usual. The shipboard incubations, performed for the usual length of time (24 hours),
saw rates averaging 0.0014 d
1
, much slower than seen at P2. The in-situ incubations, however,
showed much higher rates, indicating a response to increased iron(II) availability on a time scale
of days, as shown in Figure 3.4.
3.4 Discussion
3.4.1 Particle dependence
As shown in Figure 3.4 and 3.5, in all shipboard incubations, filtered replicates had much lower
rates of iron oxidation, demonstrating the particle dependence of this process. However, the filtered
rates were not zero. This is likely not due to analytical limitations, since the concentrations of
tracer measured on those filters is well above the limit of detection. It could potentially be a result
of adsorption to the filter material, or oxidation due to trace oxygen introduced during the filtration
process. Alternatively, we could be seeing indications of a non-particle dependent process, such as
oxidation by chromium(VI), in addition to the dominant particle dependent process.
Although shipboard incubations clearly indicate the importance of particles for anaerobic iron
oxidation, we did not see any consistent impact of sinking particles on the oxidation rate. While ox-
idation was higher in some in-situ incubations with sediment trap material, it was lower in others,
and most commonly statistically indistinguishable. This suggests that the oxidation is primarily
driven by suspended particles in a smaller size fraction, consistent with the hypothesis that oxi-
dation is driven by microbes in the water column, and that oxidation is likely controlled by the
availability of mineral surfaces, since they would likely accumulate in a sediment trap.
While seeing higher oxidation rates in the presence of sinking particles would have been a
potential indicator of abiotic oxidation, seeing lower apparent oxidation rates would have been
55
Figure 3.5: Oxidation rate constants in filtered (F1 and F2) and unfiltered (U1 and U2) replicates
of shipboard incubations at 150 m (top left), 180 m (top right), 250 m (bottom left) and 300 m
(bottom right)
56
evidence suggesting reduction of iron within reducing microenvironments around sinking particles.
We did see such a pattern in one array of incubations, in 2018 (Figure 3.6), but not in any others.
It is possible that this process has a significant impact, but if so it is confined to a relatively narrow
depth range.
3.4.2 Quantification of ambient rates
In our calculations, we have generally assumed that the reaction rate is first order, and that the rate
which we are measuring has increased in proportion to the concentration of the added tracer. That
first order approximation yields reasonable results, but is a potential source of bias. If the true
kinetics were zero order, for instance, we would drastically overestimate the half-life of iron(II) in
these regions. This is presumably not the case, since a half-life that short would be too rapid to be
resupplied from sediments, requiring remineralization or in-situ reduction in order to maintain the
iron(II) plume at steady state.
A more likely possibility, if iron oxidation in these regions is driven by microbes, is that the
process would best be described using Michaelis-Menten kinetics. In that case, adding a significant
amount of tracer would increase the rate, but that increase might level off at some point, leading
us to underestimate the reaction constant. The amount of error in this case would depend on the
Michaelis-Menten parameters, which we have no way of estimating in this study.
In addition to changing the rates of iron oxidation, we might also be impacting a population
of iron oxidizers in these incubations. By adding a large amount of iron(II) as a tracer, we might
encourage the growth of microbes utilizing it. The P3 incubations suggest that exposure to high
iron(II) concentrations for a period of several days can increase the rates of oxidation, which could
indicate enrichment of iron oxidizing microbes, or metabolic switching if the iron oxidizers are
responding to the external stimulus.
The potential impact of these processes could be explored in future studies by conducting incu-
bations on the same water samples using different amounts of the tracer. Doing this with the stable
isotope used in this study might be difficult, but it would become more feasible to perform these
57
Figure 3.6: Dissolved iron(II) concentration and iron(II) half-lives, plotted on a log-log scale.
experiments using an iron radioisotope. That could yield greater sensitivity, allowing incubations
to proceed with smaller tracer additions, as well as multiple time points, since it would not be nec-
essary to filter the entire incubation in order to ensure a quantifiable signal. Radiotracer-derived
rates could also be measured while at sea, allowing real time adjustment of tracer concentrations
in order to determine Michaelis-Menten parameters.
3.4.3 Residence times
At P2, the shallow incubation samples, covering much of the depth range of the iron(II) plume,
have half-lives ranging from 43 to 132 days. Meanwhile, in the deeper samples, the oxidation rate
was much faster, averaging 13 days. AT P3, the slower oxidation rates indicate a half-life ranging
from 319 to 1061 days, which is slow enough that loss is presumably dominated by advection or
mixing.
Most of these differences are likely due to the iron supply to a given region. P3 is closer to
the continental shelf than P2, but not in a particularly reducing region (Bolster et al, in review).
The data presented here is also consistent with previous estimates of oxidation rates, as shown in
Figure 3.2, with a relationship approximated by a power law. The deeper However, the deeper
58
samples at P2 are associated with an area of low iron(II) concentrations. Those samples are also in
the depth range of the Northeast pacific intermediate water mass which contains trace amounts of
oxygen [119]. At those concentrations, the abiotic rate of oxidation would still be much lower than
what we’ve observed [120], but could be a result of microaerophilic iron oxidation. If this pattern
is commonplace then this could explain why the iron(II) feature in the ETNP is so concentrated
within the upper ODZ.
3.4.4 Implications for iron transport
The data presented here indicates that iron oxidizing microbes are abundant within the water col-
umn of the ETNP. There is a significant rate of anaerobic iron oxidation taking place in regions
with elevated iron(II), and the experiments at P3 suggest that these microbes can respond to fresh
inputs of iron(II) on a time scale of days. This means that if iron inputs to the region change, due
to factors such as climate change or riverine input changes, iron oxidizing microbes will tend to
dampen the effect.
3.5 Conclusions
In this study, iron(II) oxidation within the ETNP ODZ has been directly measured for the first time.
This iron oxidation process is likely the source of iron to the deep plume previously observed in
this region (Bolster et al. in review). The overall impact of that plume has not been explored,
but a better understanding of iron oxidation kinetics in that region will allow us to estimate the
magnitude of the shelf to basin shuttle in the ETNP, and its importance for iron cycling in the
Pacific and beyond.
59
3.6 Materials and methods
3.6.1 Sample sites
Incubations were performed at two Lagrangian stations, referred to as P2 and P3. The locations of
these stations are shown in Figure 3.1. These stations were occupied on two cruises, in April 2018
on the R/V Roger Revelle (RR-1805), and in October 2019 on the R/V Kilo Moana (KM-19-19).
Station P2, with a starting location of 16.5

N 106.5

W, is in a relatively oligotrophic area well off
the continental shelf. P3, starting at 22

N 110

W, is near the Northern boundary of the ODZ, very
close to where aerobic respiration becomes consistently feasible again.
Samples for trace metal analysis were collected using 5 L Teflon-coated external-spring ”Niskin-
type” bottles (Ocean Test Equipment) on a powder-coated trace metal clean rosette (Sea-Bird Elec-
tronics). The rosette was lowered into the ODZ and held at depth for 30 minutes before sampling,
in order for oxygen adsorbed on the bottles to degas. After sample collection, the bottles were
taken to a laminar-flow clean van and pressurized with filtered nitrogen gas for sampling. Filtered
samples were passed through 0.2mm PES membrane filters.
3.6.2 Analytical methods
Dissolved iron(II) measurements were performed as described in [63], using a flow-injection lu-
minol chemiluminescence method with DTPA as a masking ligand for interference correction.
Samples for total dissolved metals were collected in acid-washed LDPE bottles, and double
bagged for transport to an onshore laboratory. Once onshore, Teflon-distilled trace metal grade
hydrochloric acid was added to bring the pH to 1.7. After acidification, samples were stored
at room temperature for several months. Samples were prepared for analysis using a seaFAST
preconcentration system (ESI) and analyzed by ICP-MS (Thermo Element 2) in medium resolution
mode using isotope dilution for quantification. The accuracy of the analysis was assessed using the
GSC and GSP standard solutions. Measured concentrations of GSC were 1.566 ± .017 nM Fe (n =
5), statistically indistinguishable from the consensus value of 1.535 ± .115 nM. For GSP, measured
60
concentrations were 0.149 ± .012 nM Fe (n = 5), statistically indistinguishable from the consensus
value of 0.155 ± .045 nM.
Incubation samples for particulate trace metals were collected on 0.2 mm PES filters (Supor),
cleaned using procedures in the GEOTRACES cookbook ([121]). After sample collection, the
filters were stored in acid-washed trace metal free centrifuge tubes until analysis.
In order to measure the concentration of particulate iron in the form of oxyhydroxides, as
opposed to iron in biomass or more refractory mineral fractions, some of our filters were washed
with an oxalate/EDTA reagent designed to dissolve labile minerals without lysing cell membranes
([122]). We collected two filters from each incubation and washed one of them with the reagent,
and calculated the labile particulate concentration by the difference between filters.
To measure particulate concentrations, the material was digested in sealed Teflon containers
with refluxing 50% HNO
3
. Indium was added to the acid mixture before digestion as an internal
standard. This method should dissolve any biomass as well as all but the most refractory mineral
phases ([123]), which our tracer is unlikely to be incorporated into after a one-day incubation.
After digestion, concentrations of
56
Fe and
57
Fe were determined by standard addition, and the
isotope ratio of the spike was used to calculate the amount of labeled tracer incorporated into
particles. The rate of oxidation was calculated from that value and the initial spike concentration,
assuming pseudo-first-order kinetics.
3.6.3 Shipboard incubations
For the 2018 cruise, we attempted to keep shipboard incubations anoxic by continuously bubbling
a mixture of nitrogen and carbon dioxide gas for the duration of the incubation. However, the gas
manifolds used were prone to malfunctioning on a moving ship, and many of our incubations were
contaminated by oxygen. In preparation for the 2019 cruise, we switched to using sealed glovebags
in order to keep samples anaerobic. Incubations were performed in 2L acid-washed polycarbonate
bottles that were allowed to degas in the glovebag for several hours before filling. The sampling
line from the Niskin was led directly to the glovebag, with filters inside the glovebag for filtered
61
treatments. After filling the incubation samples, oxygen scrubbers (Anaerogen, Thermo Scientific)
were used inside the glovebag in order to keep out oxygen contamination through the bag after
sealing. This system was tested before the cruise using solutions of resazurin, an oxygen indicator,
in order to confirm that samples could be kept anoxic for several days.
The iron-57 isotope tracer (Isoflex) used for the incubations was originally in the +3 oxidation
state. Prior to use, the spike was bubbled with SO
2
gas in order to reduce the iron(III) to iron(II).
Quantitative reduction was confirmed by diluting the spike in seawater and analyzing it in the same
manner as dissolved iron(II) samples. The reduced solution was stored at pH 2 in order to prevent
reoxidation.
3.6.4 In-situ incubations
In-situ incubations were performed using surface float-tethered sediment trap incubators [124].
The incubators contained two chambers, one of which was connected to a cone shaped 50 mm
mesh with an outer diameter of 2 m. The other chamber was located in the shadow of the cone, so
that it would only collect suspended particles. Spring-loaded mechanisms could be used to close
doors on the incubation chambers and to inject syringes containing a tracer into the chambers
once closed. The traps were deployed with all chamber doors open, including the lower door
on the chamber connected to the cone. The entire sediment trap sat at depth for 8 hours in the
anoxic waters of the ODZ, in order to allow adsorbed oxygen to degas, before the lower chamber
door was closed and the collection of sinking particles began. After the collection period of 24
hours, the top door of the suspended + sinking particle chamber was closed, as well as both doors
on the suspended particle chamber, and the tracer injected to both chambers immediately after.
Incubations typically took place for 24 hours before the trap was recovered on deck and filtered.
62
3.6.4.1 Isotope spikes
In preparation for an incubator deployment, isotope spikes were bubbled with nitrogen to degas for
several minutes before loading onto syringes. The syringes were also degassed at the same time,
in order to prevent the spike from being a source of oxygen contamination.
3.6.4.2 Sampling
In order to sample incubator material without oxygen contamination, a clean length of PVC pip-
ing was attached to the bottom of the incubator chamber and connected to anaerobic glovebag.
Continuous nitrogen pressure was used to flush oxygen out of the pipe prior to sealing. Incubator
material was drained into sample bottles inside the glovebag, which were then taken to the clean
van and filtered in the same way as shipboard incubations.
3.7 Acknowledgements
The authors would like to thank the chief scientists, Gabrielle Rocap and Al Devol, as well as the
captains and crew of the R/V Roger Revelle and R/V Kilo Moana. We would also like to thank
Shun-Chun Yang, Rintaro Moriyasu, Megan Duffy, Khadijah Homolka, Emmet Bush, Anna Boyar,
Natalya Evans, Tianyi Huang, Morgan Raven, and Nick Hawco.
63
Introduction to Chapter 4
So far, this thesis has only considered a single source of iron(II) to the ODZ: reducing sediments
on the continental shelf. Although this conceptual model is widely used, it is not clear that it can
actually reproduce the observed distributions of iron(II), a topic which will be revisited in Chapter
5. The purpose of this chapter is to investigate whether other sources of iron(II) are possible.
The water column of an ODZ still contains significant amounts of nitrate, as well as nitrite in
some samples. By analogy to anaerobic sedimentary systems, we would not expect that iron(II)
would be found in these areas, much less be produced. Ferrous iron production in the water column
would run contrary to expectations based on the bulk thermodynamics of the area. There are some
potential mechanisms to reconcile this paradox, both biological and abiotic, which involve large
particles present in the water column, whose role is the research question of this chapter.
For this study, a mesocosm was constructed at the Wrigley Institute for Environmental Studies
on Santa Catalina Island. The goal was to create a simulated ”ODZ in a bottle” that would mimic
the biogeochemical properties of an ODZ water column, but which would not include reducing
sediments. In that way, water column processes could be studied independently of sedimentary
sources. In particular, we wanted to test the role of large particles in iron(II) accumulation.
The mesocosm was assembled in a 20 L carboy, kept cold and in the dark, with a mixture of
nitrogen and carbon dioxide continuously bubbling to maintain anoxia and pH. The system was
spiked with nitrate, phosphate, and succinic acid (as a carbon source). Measurements of nutrients
showed that a denitrifying community was rapidly established, with nitrite detected in the system
for over one week. Ammonium concentrations, initially quite high in the surface seawater used for
64
the mesocosm, dropped rapidly, indicating the presence of anammox as well, and suggesting that
the mesocosm was reproducing many of the ODZ biogeochemical features.
The next experiments focused on measuring iron(II). Mesocosms were established in parallel,
filled by either unfiltered or 50 mm filtered seawater. This experiment was replicated twice, using
different initial water samples. In both cases, the mesocosm containing unfiltered seawater (i.e.
large particles) accumulated dissolved iron(II), while the mesocosm without large particles did
not.
The amount of iron in the different particle fractions was measured as well, to check whether
this could be due to vastly different particulate iron concentrations. There was a significant amount
of iron in the> 50mm fraction, but only by a factor of about 30%. This suggests that the change
in iron(II) production that we see must result from chemical processes associated with the large
particles, rather than simply total concentration.
These experiments demonstrate that not only can iron(II) persist in the presence of nitrate and
nitrite for several days, but that it can actually be produced under those circumstances. This is
in contrast to predictions based on thermodynamic equilibria, as well as analogies to sedimentary
geochemistry. It also suggests that accurately modeling ODZ iron cycling may require incorpora-
tion of water column iron(II) sources, in addition to sediments.
This study is likely to be expanded before publication in a journal. In particular, frozen samples
for additional nutrient measurements, which were collected but not yet analyzed due to the COVID-
19 pandemic, will shed some light on potential mechanisms for iron(II) formation in this system.
In the absence of those data, the primary finding of this study has been reproduced in this chapter.
65
Chapter 4
Iron reduction in a nitrate-replete oxygen deficient zone
mesocosm
4.1 Abstract
A mesocosm designed to simulate oxygen deficient zone (ODZ) biogeochemistry was assembled
and used to investigate iron(II) production in the water column. Measurements of nitrite and am-
monia in the mesocosm indicate a realistic combination of denitrification and anammox driven
nitrogen cycling. Replicates of mesocosms using unfiltered or 50 mm filtered seawater showed
that iron(II) production in the presence of nitrate and nitrite was possible, but only if large particles
are present in the system. Analysis of the bulk particulate iron chemistry shows that this is not
a result of total iron concentrations, but due to specific processes associated with large particles.
These iron sources are relevant for accurately modeling iron cycling in ODZs.
4.2 Introduction
Iron is a scarce nutrient in the modern ocean, and limits primary productivity in large areas of
the ocean surface (e.g. [125]). Under most conditions in the ocean, the thermodynamically stable
form of iron is the ferric (+3) oxidation state, which is relatively insoluble and particle reactive [2,
126]. The ferrous (+2) form, although rapidly oxidized under most conditions, is much less particle
66
reactive [5, 120]. Therefore, processes which stabilize iron(II) can increase iron’s residence time
in the ocean.
Ferruginous or euxinic conditions, in which iron(II) is thermodynamically stable, are rare in
the ocean. However, iron(II) can be produced in reducing sediments and diffuse into the water
column above [51]. This process has been invoked to explain the presence of iron(II) in open
ocean oxygen deficient zones (ODZs) [51, 52]. However, iron(II) is found a significant distance
away from the continental shelf, despite the presence of nitrate and nitrite, oxidants strong enough
to oxidize iron(II) and limit its transport [51]. That oxidation process may be slow enough that
this transport still occurs, but it is also possible that this process needs to be balanced by a water
column source of iron(II).
There are several potential mechanisms for iron(II) production in the water column. One possi-
bility is that the iron(II) source is linked to remineralization. ODZs are typically found in upwelling
regions, with high surface productivity and large fluxes of sinking organic matter [127]. Iron is
exported as a component of this biomass, and can be released during remineralization. Many met-
alloenzymes store iron in the ferrous oxidation state [128]. Under oxygenated conditions, that iron
would likely oxidize as cell membranes are ruptured, but under anaerobic conditions it’s possible
that the remineralized iron could be released in a reduced state.
Alternatively, rather than iron(II) simply being released during remineralization, iron reduction
could be coupled to carbon oxidation in a dissimilatory manner. In environments where oxygen
and nitrate have been depleted, iron(III) is used as a terminal electron acceptor by microbes [129].
This is likely the source of iron(II) fluxes out of reducing sediments. In the bulk water column of
ODZs, nitrate is not fully depleted. However, it is possible that large sinking particles of organic
matter could host microenvironments, in which oxygen and nitrate are locally depleted, where iron
could become a viable terminal electron acceptor [130].
It is also possible that iron(II) could be produced independently from biology. In most cases,
particulate iron(II) in minerals would oxidize rapidly upon exposure to oxygenated seawater [5],
67
but in anaerobic environments it could dissolve as ferrous iron. Resuspended sediments or partic-
ulate terriginous material could therefore act as a source of iron(II) directly.
These processes are all theoretical, and have never been shown to occur. The purpose of this
study is to evaluate whether dissolved iron(II) can accumulate in an anoxic but nitrate replete
environment. We developed a mesocosm to simulate conditions of an ODZ water column, isolated
from reducing sediments, to determine whether water column processes alone could lead to iron(II)
production.
4.3 Methods
4.3.1 Mesocosm construction
To construct ODZ mesocosms, we began with seawater collected at the Wrigley Institute of Envi-
ronmental Studies, located on Santa Catalina Island in Southern California. Mesocosms were as-
sembled in 20 L polyethylene terephthalate carboys, acid-washed and conditioned for 24 hours us-
ing filtered seawater. The starting material, either unfiltered or passed through a 50mm polypropy-
lene filter (Evoqua Regard), was added to the carboys, and a mixture of 500 ppm CO
2
in nitrogen
gas was bubbled continuously through the carboys, to ensure that the system stayed anoxic and
to reproduce the lowered pH of an ODZ. The mesocosms were covered in order to avoid light
contamination, and chilled using a 13

C water bath. After several hours of bubbling, the meso-
cosm was amended with 25 mM sodium nitrate, 1.56 mM disodium hydrogen phosphate, and 5.8
mM succinic acid. These concentrations were chosen to maintain a 16:1 ratio between nitrate and
phosphate, and to ensure that added carbon would not be sufficient to drive the system into euxinia.
A length of acid washed tubing was inserted into the mesocosm, which could be connected
to a syringe for daily sampling. Samples were filtered using 0.2 mm polyethersulfone disc filters
(Fisher) and stored in acid washed low density polyethylene bottles.
68
4.3.2 Analytical methods
Samples for iron(II) were analyzed using flow injection luminol chemiluminescence using pentetic
acid for interference correction, as in the method of [63]. Total dissolved trace metal samples were
acidified to a final concentration of 20 mM Teflon distilled HCl and stored for several months.
Samples were then preconcentrated using the seaFAST-pico system [131] and then analyzed using
a Thermo Element 2 ICP-MS, using isotope dilution for quantification.
Nitrite and phosphate were measured spectrophotometrically [132, 133]. Ammonium was mea-
sured fluorometrically [134].
For each mesocosm, a filtered sample was collected on polyethersulfone disc filters (0.2 mm)
cleaned in accordance with protocols in the GEOTRACES cookbook [121]. Those filters were
digested using refluxing HNO
3
[123] and then analyzed in the same manner as the dissolved trace
metal samples.
4.4 Results
The first experiment focused specifically on characterizing the nutrient cycling of the system, in
order to determine if the mesocosm accurately reproduced the biogeochemical cycling of an ODZ.
For this experiment, 50mm filtered seawater was used. The results from this mesocosm are shown
in Figure 4.1. Within a day, nitrite concentrations began to increase, indicating that denitrification
was active within the system. The concentration of nitrite leveled off after a few days, until finally
decreasing back to undetectable levels on the tenth day. Ammonium concentrations, initially ele-
vated, dropped to a low level after the onset of anoxia, and stayed very low except for one spike in
the ammonium concentration after 9 days, immediately before the nitrite concentration dropped.
Phosphate concentrations stayed relatively constant throughout the experiment (1.4-1.7mM), until
dropping to approximately 1mM after the end of denitrification.
Having evaluated the overall cycling of nutrients in the mesocosm, the next experiments fo-
cused on iron chemistry. In order to evaluate the role of large particles in the formation of an
69
Figure 4.1: Nitrite (mM), ammonium (nM), and phosphate (mM) concentrations over time in a
mesocosm.
70
iron(II) feature, two mesocosms were constructed in parallel, using unfiltered and 50 mm filtered
seawater, and measured the iron(II) concentrations and nitrite over several days. The results are
shown in Figure 4.2. Both mesocosms developed denitrifying environments within a day of the
experiment start, but detectable iron(II) only appeared in the unfiltered mesocosm. Iron(II) con-
centrations reached their peak approximately two days after the start of the experiment, and then
began to decrease.
This experiment was replicated, with results shown in Figure 4.3. Analytical issues reduced the
precision of these measurements, but the results were similar overall. A small amount of iron(II)
was measured in the 50 mM filtered treatment on day 1, but concentrations were otherwise below
the limit of detection. Again, the unfiltered mesocosm accumulated a significant amount of iron(II),
peaking on day 4 before slowly decreasing.
For both of these experiments, particulate iron samples were collected before the mesocosm
began in both size fractions, to evaluate how much iron might be contributed to the system by
particles larger than 50 mm. Those large particles accounted for 24-37% of the particulate iron in
the system, as shown in Figure 4.4.
4.5 Discussion
In all experiments, nitrite was observed accumulating by day 1. The nitrite tending to level off after
a few days suggests that a relatively stable denitrifying community is established fairly rapidly.
That does not require denitrifiers to grow particularly rapidly, due to the location of the sampling
site. Although samples were collected from surface waters, they were collected at a coastal site
and likely contained significant numbers of resuspended sedimentary microbes. It’s significant that
ammonium concentrations decreased after anoxia, suggesting that anammox bacteria were likely
active in the community as well, further enhancing the comparison with ODZ communities [135].
Despite this significant biological activity, phosphate concentrations did not increase during the
experiment. This suggests that relatively little of the denitrification was driven by remineralization
71
● ● ● ● ● ● ● ● ● ● 0.0
0.5
1.0
0 25 50 75
Time (hours)
Iron(II) (nM)
● ● Unfiltered
50 µm filtered
Figure 4.2: First mesocosm measurements of iron(II), including one mesocosm with unfiltered
seawater and one with 50mm filtered seawater. Error bars are 2s confidence intervals.
72
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.5
0.0
0.5
1.0
0 50 100 150 200
Time (hours
Iron(II) (nM)
● ● Unfiltered
50 µm filtered
Figure 4.3: Second mesocosm measurements of iron(II), including unfiltered and 50 mm filtered
seawater. Error bars are 2s confidence intervals.
73
● ● ● ● 0
5
10
15
20
1 2
Mesocosm
Particulate iron (nM)
● ● Unfiltered
50 µm filtered
Figure 4.4: Particulate iron concentrations at the start of each mesocosm, including the unfiltered
and 50mm filtered treatments. Error bars are 2s confidence intervals.
74
of POM in the initial samples, rather than the succinic acid supplements. It also suggests that the
end of the nitrite signal at 10 days was a result of organic carbon depletion, rather than a transition
to sulfidic conditions after the depletion of nitrate. Samples to measure nitrate concentration were
collected and frozen, but have not been analyzed due to the COVID-19 pandemic.
The iron(II) concentrations observed in experiments including large particles was not particu-
larly stable. This likely results from anaerobic iron oxidation of some kind, as discussed in Chapter
3. However, we did on multiple occasions see iron(II) accumulation with nitrate and nitrite still
present in the system, in contrast to what is typically seen in sediments. Additionally, rather than
merely coexisting with those species, iron(II) concentrations actually increased for several days,
showing that iron(II) production is not inhibited by those compounds, and suggesting that the dy-
namics of these systems cannot be explained using thermodynamic equilibrium.
In mesocosms without large particles, iron(II) did not accumulate, despite the presence of
ODZ-like nutrient cycling, as in Figure 4.1. Although a significant fraction of particulate iron
was contained in those larger particles, a majority of the particulate iron was still present in the
filtered mesocosms (Figure 4.4). This is evidence against the idea that iron(II) in this system came
from mineral dissolution. It is possible that the larger particles contained more iron in a labile
form, which could be investigated through mineralogical studies, but the evidence presented here
supports the hypothesis that iron(II) production was linked to biological processes.
4.6 Conclusions
We have repeatedly observed the production of dissolved iron(II) under nitrate replete conditions in
a system isolated from reducing sediments. Regardless of the mechanism behind this phenomenon,
it does suggest that non-sedimentary sources of iron(II) need to be considered. The accumulation
of iron(II) in these environments suggests that analogies to sedimentary redox cycling of these
species may be misleading, and that modeling of iron dynamics in ODZs may require the inclusion
of in-situ reduction.
75
4.7 Supplemental Data
76
● ● ● ● ● ● ● ● ● ● 5
6
7
8
0 25 50 75
Time (hours)
Dissolved iron (nM)
● ● Unfiltered
50 µm filtered
Figure 4.5: Total dissolved iron over time in the first mesosocosm with size fractionation treatment.
Corresponds to Figure 4.2. Total dissolved iron was higher in the unfiltered treatment, and peaked
at the same time the iron(II) maximum was achieved.
77
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 6
7
0 50 100 150 200
Time (hours
Dissolved iron (nM)
● ● Unfiltered
50 µm filtered
Figure 4.6: Total dissolved iron over time in the second mesocosm with size fractionation. Corre-
sponds to Figure 4.3.
78
Introduction to Chapter 5
The measurements of iron(II) oxidation rates from Chapter 3 are a significant improvement in our
understanding of ODZ iron oxidation. However, the measurements reported in that chapter are
fairly sparse, from only two stations and a handful of depths. In order to evaluate this process in
the ETNP as a whole, a modeling approach is needed.
Several previous studies have applied advection-diffusion models to iron transport in the ETSP
(Scholz et al. 2016; Cutter et al. 2018). Both of those studies used one dimensional models, which
are easy to solve analytically although difficult to generalize. Neither study was able to estimate
an overall flux of iron in their regions of interest. In general, that type of calculation require three
dimensional circulation patterns.
In this chapter, I report the results of several modeling exercises using the AWESOME OCIM, a
matrix transport model which uses realistic circulation and first order kinetic processes to calculate
the global steady state distribution of a tracer very rapidly. The objectives of this study were to
evaluate the importance of the various processes discussed in this thesis, and to ultimately estimate
the total flux of oxidized iron out of the system.
The AWESOME OCIM comes with a number of ”built in” functions designed to simulate
various oceanographic processes, but none of them were well suited for the processes discussed
here. Five functions were written to represent them, and used in different combinations for each
model run. One source function modeled iron inputs from reducing sediments on the continental
shelf, injecting iron(II) into model boxes on a particular isopycnal surface, as discussed in Chapter
2. Two functions were written to represent anaerobic iron oxidation, as in Chapter 3, using first and
second order kinetics, respectively. Another source function added iron(II) to the water column
79
proportional to the remineralization of organic carbon, analogous to the processes discussed in
Chapter 4. Lastly, a sink function was written for iron(II) that escapes the ODZ, using the rate law
for iron oxidation by dissolved oxygen developed by Millero et al. (1987).
Models were fitted to the iron(II) data presented in Chapter 2, using a MATLAB optimization
routine to minimize the squared error between predicted iron(II) concentrations and direct mea-
surements. Comparison to that data enabled the importance of the various model processes to be
assessed.
Models using first order anaerobic iron oxidation kinetics consistently underestimated the
iron(II) concentrations at offshore stations. This pattern held true whether or not water column
sources were included to replenish iron(II) offshore. To some extent this is unsurprising, since it
matches patterns in measured oxidation rate kinetics (see Figure 6 in Chapter 3). In order to match
the relatively steep gradient in iron(II) close to the continental margin, a higher oxidation rate was
necessary, which was inconsistent with the relatively shallow gradient further offshore. Second
order oxidation kinetics, representing iron oxidation by microbes whose activity is controlled by
iron supply, captures this trend much better.
Finally, the model can be used to estimate the overall transport of iron(II) through the system.
Applying the second order oxidation rate law from those models to the optimized distributions,
we can calculate the supply of particulate iron oxyhydroxides to continental slope sediments, and
therefore put an upper bound on the flux of iron to deep plumes. This model suggests that shelf to
basin shuttling in the ETNP could transport up to 0.15 Gmol iron per year to the deep waters of
the Pacific.
80
Chapter 5
Modeling anaerobic iron oxidation in the Eastern Tropical
Pacific
5.1 Abstract
Anaerobic iron oxidation rates in ODZs constrain iron transport in these regions, but measure-
ments of those rates are sparse. A modeling approach is used to estimate iron oxidation rates in the
Eastern Tropical North Pacific (ETNP). Models using first order oxidation kinetics systematically
underestimate iron(II) supply to offshore stations, and second order kinetics are shown to be nec-
essary in order to accurately model iron cycling in the region. Optimized parameters for iron(II)
cycling are used to estimate the overall oxidation flux of iron(II) in the ETNP of 0.15 Gmol/year.
5.2 Introduction
Oxygen deficient zones (ODZs) are regions of the ocean in which denitrification is the primary
mode of respiration, and are implicated in the biogeochemical cycling of several important el-
ements. ODZs contain elevated concentrations of dissolved iron, particularly in the ferrous (+2)
oxidation state [3, 107, 136], and are important sources of iron to the open ocean through the “shelf
to basin shuttle” [137]. During this process, iron(II) from reducing shelf sediments is transported
offshore and oxidized anaerobically before being scavenged to the deep ocean. This scavenged
81
iron can form deep plumes and transport iron significant distances [138]. Understanding the rates
of these processes is critical in order to evaluate the importance of ODZs for global iron cycling.
Iron(II) is less thermodynamically stable than iron(III) in ODZs, due to significant concentra-
tions of nitrate and nitrite [107]. There is therefore potential for an anaerobic iron(II) oxidation
process, which converts the highly soluble ferrous form into the more particle reactive iron(III)
form. Mineralogical evidence of particulate iron supports this hypothesis [110]. Additionally,
experiments shown in Chapter 3 detected this process using isotope tracers, with typical reaction
rates on the order of 0.01 d
1
.
These incubation measurements are sparse, and widespread datasets of iron oxidation rates
are not likely to be available in the immediate future. However, several cruises have sampled for
dissolved iron(II) in ODZs over significant transects. Modeling studies have shown the impact
of an anaerobic scavenging process removing iron(II) as it is transported off the continental shelf
[31, 51]. These studies have used one dimensional advection-diffusion modeling frameworks, and
generalizing these results to the region as a whole is challenging.
The AWESOME OCIM is a global steady state model which can be used to investigate many
processes in marine biogeochemistry [139]. It uses a data-constrained transport matrix and first
order chemical kinetics to solve for a steady state distribution of a tracer. Unlike other global
circulation models, the AWESOME OCIM can be solved very rapidly, in a matter of seconds
on a typical desktop computer. This enables simple iterative approaches for model optimization,
allowing the model to tune parameters for a given set of processes in order to match observations.
In this study, we optimize the AWESOME OCIM in order to reproduce the iron(II) distributions
observed during a 2016 cruise on the NOAA Ship Ronald H. Brown. Several model frameworks
are compared, starting with shelf inputs and first order oxidation kinetics, and then expanding to
incorporate higher order oxidation kinetics and remineralization based sources of iron(II).
82
5.3 Methods
Data for this study was collected from March-April 2016 on the NOAA Ship Ronald H. Brown
(cruise number RB1603), as described in Bolster et al. (in review).
All modeling runs were performed using the original version of the AWESOME OCIM [139].
Each iron(II) measurement was assigned to its model box in the AWESOME OCIM. Observations
in the same model box were averaged in preparation for optimization. For all optimizations, least
squares error minimization was performed using the MATLAB function fminsearch.
Iron(II) models were constructed using four functions written for the AWESOME OCIM, two
sources (sedimentary input and remineralization) and two sinks (aerobic and anaerobic oxida-
tion). The sedimentary source function, used in all model runs, represents input of iron(II) to the
water column from reducing sediments. The AWESOME OCIM does not have a high enough
bathymetric resolution in order to model shelf sediments explicitly, but previous work has shown
that sedimentary inputs in the region correspond to an isopycnal surface with a potential density
anomaly of 26.5 (Bolster et al. in review). This function adds iron(II) to model boxes bordering
continents with a potential density anomaly in a defined range. The iron(II) flux is assumed to be
proportional to the particulate organic carbon flux through that model box, and is only added if
dissolved oxygen is below a certain threshold, building off the work of [140].
dFe
z
dt
= iPOC
z
(5.1)
For model boxes which meet this criterion, the input equation is shown in Equation 5.1, where
the input is approximated as being linearly related to the total amount of particulate organic carbon
in the model box.
An additional function was needed to model water column remineralization. The AWESOME
OCIM contains several functions which model remineralization processes, but all are designed
for nutrients such as phosphate which can be accumulated at the surface and exported to depth.
When attempting to specifically model iron(II) (as opposed to total iron), these functions are not
83
applicable, since iron(II) is not the predominant form of iron in oxygenated surface waters. In-
stead, iron(II) remineralization was parameterized as being proportional to organic matter rem-
ineralization within a model box, assuming a constant Redfield-like Fe(II):C ratio. This ratio is
not necessarily the same as typical Fe:C ratios, since not all particulate organic iron is in the +2
oxidation state, and non-biogenic iron could also be reduced during the remineralization process
(see Chapter 4). This relationship is shown in Equation 5.2, where POC
z
is equal to the POC at a
given depth, and POC
z+1
is the POC in the box above that depth.
dFe
z
dt
= j(POC
z+1
POC
z
) (5.2)
The primary parameter of interest for these models is the rate of iron(II) loss within the anoxic
zones of the ODZ. Two different frameworks were used to model this process. The most basic,
shown in Equation 5.3, is a first order oxidation rate, depending only on the concentration of
iron(II) in the model box. Our implementation of this is identical to the radioactive decay function
of the AWESOME OCIM, except that it was only implemented when oxygen concentrations are
below the detection limit of a typical CTD oxygen electrode. Another formulation, shown in
Equation 5.4, approximated second order oxidation kinetics using Newton’s method, an iterative
approach, in which the output of a previous model was used to calculate a loss rate for the next
model. For each model run, 100 iterations were used, which was sufficient in all cases for the
distributions to converge.
dFe
dt
=k[Fe] (5.3)
dFe
dt
=k[Fe]
2
(5.4)
Lastly, aerobic oxidation of iron(II) was also implemented in all models, to represent the abiotic
oxidation of iron(II) in oxygenated waters. This process is first order with respect to iron(II)
and dissolved oxygen, as shown in Equation 5.5, where the rate constant l is calculated from the
84
temperature and pH according to a previously published rate law [141], and implemented in model
boxes with oxygen concentrations above the limit of detection.
dFe
dt
=l[O
2
][Fe] (5.5)
Because of the linear structure of the Awesome OCIM, a common assumption in all these
parameterizations is that all iron(II) is equivalent and exchangeable. For instance, we assume that
there is not an inert phase of iron(II)-ligand complexes which is slower to react than other phases.
At present, there is no evidence for such phases, and kinetics experiments suggest that reaction
rates are similar across any chemical differences in iron(II) [3], but future work could lead to a
revision of that assumption.
5.4 Results
The first model used only the sedimentary input function and a first order oxidation process in
the ODZ, in addition to the aerobic oxidation process outside the ODZ. A zonal transect of these
model results is shown in Figure 5.1. Although this model converged in a straightforward man-
ner, the results significantly underestimated the concentration of iron(II) at positions further from
continental shelf sediments, as shown in Figure 5.2. In order to fit the relatively steep gradient
of iron(II) in nearshore samples, first order models require an oxidation rate which is too high to
permit elevated concentrations offshore. This suggests that either a high order rate law is required,
or that dissolved iron(II) must be replenished in some way offshore.
The results of a model incorporating a remineralization term are shown in Figure 5.3. Although
iron(II) concentrations increase throughout, the same pattern is observed as the model continues to
underestimate the concentration at samples farther from the shelves (Figure 5.4).
A model using second order oxidation kinetics significantly changes the distribution (Figure
5.5). It also significantly reduces this bias (Figure 5.6). Higher order rate laws could in theory
reproduce this pattern, although the exact mechanism of such a process is not clear.
85
Figure 5.1: A zonal transect of modeled iron(II) concentrations along 19

N. This model used only
sedimentary inputs along the 26.5 isopycnal, and first order oxidation kinetics within the ODZ.
86
Figure 5.2: Modeled iron(II) concentrations compared to the observed concentrations in each
model box. The 1:1 line is shown in orange. The model significantly underestimates the iron(II)
concentrations in many samples, especially at the lower end.
87
Figure 5.3: Model results along 19

N, from a model incorporating a remineralization source of
iron(II) and first order oxidation kinetics.
88
Figure 5.4: Comparison of results from a model with remineralization and first order kinetics to
observed data. The model still underestimates iron(II) values at low concentrations.
89
Figure 5.5: Results from a model incorporating second order oxidation kinetics, along 19

N.
90
Figure 5.6: Comparison of the model with second order oxidation kinetics with observation. This
model does a significantly better job at low concentrations, compared to Figures 5.2 and 5.4
91
5.5 Discussion
These optimization experiments show that higher order oxidation kinetics are critical for repro-
ducing the observed iron(II) distributions in the ETNP. Attempts to measure the oxidation rate
using incubation approaches are consistent with these suggestions, as is the hypothesis that iron
oxidation is microbially driven (see Chapter 3). In areas with high iron(II) supply closer to shore,
a higher population of active iron oxidizing microbes can be supported, leading to a higher overall
rate. The expanded form of the second order kinetic rate law is shown in Equations 5.6 - 5.8, where
[NDFO] is the concentration of nitrate dependent iron oxidizers, and mn= k from Equation 5.4.
This regional trend would be maintained even if the rate law for an individual microbe followed a
different pattern, such as Michaelis-Menten kinetics.
dFe
dt
=m[NDFO][Fe] (5.6)
[NDFO]= n[Fe] (5.7)
dFe
dt
=mn[Fe]
2
(5.8)
In addition, this type of rate law allows an estimate to be made of the total amount of iron
transported through these processes. Iron oxidation in these areas is the source of particulate iron
to continental slope sediments [64, 137, 138]. Some of the iron delivered will be buried, but the
oxidation rate therefore is an upper bound on the amount of iron which could supply the deep
plumes observed in these regions. Applying the second order rate law to the optimized distribution
in Figure 5.5 yields an overall flux of 0.15 Gmol/year for the ETNP region, as an approximation
for shelf to basin shuttling in this area.
92
5.6 Conclusions
Several model frameworks have been applied to reproduce the distribution of iron(II) in the ETNP.
Models employing first order iron oxidation kinetics consistently underestimate iron(II) concen-
trations offshore, even when remineralization based sources are included. Second order kinetics
perform much better, consistent with measurements of the oxidation rate. These kinetics can be
used to estimate a yearly iron(II) oxidation flux out of the ODZ of 0.15 Gmol.
93
References
1. Sunda, W. G. Feedback Interactions between Trace Metal Nutrients and Phytoplankton in
the Ocean. English. Frontiers in Microbiology 3, 204. doi:10.3389/fmicb.2012.00204
(2012).
2. Millero, F. J. Effect of Ionic Interactions on the Oxidation of Fe(II) and Cu(I) in Natural
Waters. en. Marine Chemistry 28, 1–18. doi:10.1016/0304-4203(89)90183-7 (1989).
3. Kondo, Y . & Moffett, J. W. Iron Redox Cycling and Subsurface Offshore Transport in the
Eastern Tropical South Pacific Oxygen Minimum Zone. Marine Chemistry 168, 95–103.
doi:10.1016/j.marchem.2014.11.007 (2015).
4. Millero, F. J., Sotolongo, S. & Izaguirre, M. The Oxidation Kinetics of Fe(II) in Seawater.
Geochimica et Cosmochimica Acta 51, 793–801. doi:10.1016/0016-7037(87)90093-7
(1987).
5. King, D. W., Lounsbury, H. A. & Millero, F. J. Rates and Mechanism of Fe(II) Oxidation at
Nanomolar Total Iron Concentrations. Environmental Science & Technology 29, 818–824.
doi:10.1021/es00003a033 (1995).
6. Santana-Casiano, J. M., Gonz´ alez-D´ avila, M. & Millero, F. J. Oxidation of Nanomolar Lev-
els of Fe(II) with Oxygen in Natural Waters. en. Environmental Science & Technology 39,
2073–2079. doi:10.1021/es049748y (2005).
7. Fujii, M., Rose, A. L., Waite, T. D. & Omura, T. Oxygen and Superoxide-Mediated Redox
Kinetics of Iron Complexed by Humic Substances in Coastal Seawater. en. Environmental
Science & Technology 44, 9337–9342. doi:10.1021/es102583c (2010).
8. Rue, E. L. & Bruland, K. W. Complexation of Iron(III) by Natural Organic Ligands in the
Central North Pacific as Determined by a New Competitive Ligand Equilibration/Adsorptive
Cathodic Stripping V oltammetric Method. Marine Chemistry. The Chemistry of Iron in Sea-
water and Its Interaction with Phytoplankton 50, 117–138. doi:10.1016/0304-4203(95)
00031-L (1995).
9. Shaked, Y ., Kustka, A. B. & Morel, F. M. M. A General Kinetic Model for Iron Acquisition
by Eukaryotic Phytoplankton. English. Limnology and Oceanography 50, 872–882. doi:10.
4319/lo.2005.50.3.0872 (2005).
10. Morel, F. M. M., Kustka, A. B. & Shaked, Y . The Role of Unchelated Fe in the Iron Nutrition
of Phytoplankton. English. Limnology and Oceanography 53, 400–404. doi:10.4319/lo.
2008.53.1.0400 (2008).
11. Shaked, Y . & Lis, H. Disassembling Iron Availability to Phytoplankton. Frontiers in Micro-
biology 3. doi:10.3389/fmicb.2012.00123 (2012).
12. Michiels, C. C., Darchambeau, F., Roland, F. A. E., Morana, C., Llir´ os, M., Garc´ ıa-Armisen,
T., et al. Iron-Dependent Nitrogen Cycling in a Ferruginous Lake and the Nutrient Status of
Proterozoic Oceans. Nature Geoscience. doi:10.1038/ngeo2886 (2017).
94
13. Barbeau, K. Photochemistry of Organic Iron(III) Complexing Ligands in Oceanic Systems.
en. Photochemistry and Photobiology 82, 1505. doi:10.1562/2006-06-16-IR-935
(2006).
14. Hogle, S. L., Thrash, J. C., Dupont, C. L. & Barbeau, K. A. Trace Metal Acquisition by
Marine Heterotrophic Bacterioplankton with Contrasting Trophic Strategies. en. Applied
and Environmental Microbiology 82, 1613–1624. doi:10.1128/AEM.03128-15 (2016).
15. Resing, J. A., Sedwick, P. N., German, C. R., Jenkins, W. J., Moffett, J. W., Sohst, B. M.,
et al. Basin-Scale Transport of Hydrothermal Dissolved Metals across the South Pacific
Ocean. en. Nature 523, 200–203. doi:10.1038/nature14577 (2015).
16. Stookey, L. Ferrozine - a New Spectrophotometric Reagent for Iron. English. Analytical
Chemistry 42, 779–&. doi:10.1021/ac60289a016 (1970).
17. O’Sullivan, D. W., Hanson, A. K. & Kester, D. R. Stopped Flow Luminol Chemilumines-
cence Determination of Fe(II) and Reducible Iron in Seawater at Subnanomolar Levels.
Marine Chemistry 49, 65–77. doi:10.1016/0304-4203(94)00046-G (1995).
18. Bowie, A. R., Achterberg, E. P., Mantoura, R. F. C. & Worsfold, P. J. Determination of
Sub-Nanomolar Levels of Iron in Seawater Using Flow Injection with Chemiluminescence
Detection. English. Analytica Chimica Acta 361, 189–200. doi:10.1016/S0003-2670(98)
00015-4 (1998).
19. Emmenegger, L., King, D. W., Sigg, L. & Sulzberger, B. Oxidation Kinetics of Fe(II) in
a Eutrophic Swiss Lake. English. Environmental Science & Technology 32, 2990–2996.
doi:10.1021/es980207g (1998).
20. Cannizzaro, V ., Bowie, A. R., Sax, A., Achterberg, E. P. & Worsfold, P. J. Determination of
Cobalt and Iron in Estuarine and Coastal Waters Using Flow Injection with Chemilumines-
cence Detection. English. Analyst 125, 51–57. doi:10.1039/a907651d (2000).
21. Bowie, A. R., Achterberg, E. P., Sedwick, P. N., Ussher, S. & Worsfold, P. J. Real-Time
Monitoring of Picomolar Concentrations of Iron(II) in Marine Waters Using Automated
Flow Injection-Chemiluminescence Instrumentation. English. Environmental Science & Tech-
nology 36, 4600–4607. doi:10.1021/es020045v (2002).
22. Croot, P. L. & Laan, P. Continuous Shipboard Determination of Fe(II) in Polar Waters Using
Flow Injection Analysis with Chemiluminescence Detection. en. Analytica Chimica Acta
466, 261–273. doi:10.1016/S0003-2670(02)00596-2 (2002).
23. Ussher, S. J., Yaqoob, M., Achterberg, E. P., Nabi, A. & Worsfold, P. J. Effect of Model
Ligands on Iron Redox Speciation in Natural Waters Using Flow Injection with Lumi-
nol Chemiluminescence Detection. English. Analytical Chemistry 77, 1971–1978. doi:10.
1021/ac048850a (2005).
24. Hopkinson, B. M. & Barbeau, K. A. Organic and Redox Speciation of Iron in the Eastern
Tropical North Pacific Suboxic Zone. Marine Chemistry. Special Issue: Dedicated to the
Memory of Professor Roland Wollast 106, 2–17. doi:10.1016/j.marchem.2006.02.008
(2007).
25. Shaked, Y . Iron Redox Dynamics in the Surface Waters of the Gulf of Aqaba, Red Sea.
English. Geochimica Et Cosmochimica Acta 72, 1540–1554. doi:10.1016/j.gca.2008.
01.005 (2008).
26. Hansard, S. P. & Landing, W. M. Determination of Iron(II) in Acidified Seawater Samples
by Luminol Chemiluminescence. en. Limnology and Oceanography: Methods 7, 222–234.
doi:10.4319/lom.2009.7.222 (2009).
95
27. Hansard, S. P., Landing, W. M., Measures, C. I. & V oelker, B. M. Dissolved Iron(II) in the
Pacific Ocean: Measurements from the PO2 and P16N CLIV AR/CO2 Repeat Hydrography
Expeditions. English. Deep-Sea Research Part I-Oceanographic Research Papers 56, 1117–
1129. doi:10.1016/j.dsr.2009.03.006 (2009).
28. Ussher, S. J., Milne, A., Landing, W. M., Attiq-ur-Rehman, K., Seguret, M. J. M., Hol-
land, T., et al. Investigation of Iron(III) Reduction and Trace Metal Interferences in the
Determination of Dissolved Iron in Seawater Using Flow Injection with Luminol Chemi-
luminescence Detection. English. Analytica Chimica Acta 652, 259–265. doi:10.1016/j.
aca.2009.06.011 (2009).
29. Kondo, Y . & Moffett, J. W. Dissolved Fe(II) in the Arabian Sea Oxygen Minimum Zone and
Western Tropical Indian Ocean during the Inter-Monsoon Period. en. Deep Sea Research
Part I: Oceanographic Research Papers 73, 73–83. doi:10.1016/j.dsr.2012.11.014
(2013).
30. Schallenberg, C., Davidson, A. B., Simpson, K. G., Miller, L. A. & Cullen, J. T. Iron(II)
Variability in the Northeast Subarctic Pacific Ocean. Marine Chemistry. Biogeochemistry of
Trace Elements and Their Isotopes 177, 33–44. doi:10.1016/j.marchem.2015.04.004
(2015).
31. Cutter, G. A., Moffett, J. W., Nielsd´ ottir, M. C. & Sanial, V . Multiple Oxidation State Trace
Elements in Suboxic Waters off Peru: In Situ Redox Processes and Advective/Diffusive
Horizontal Transport. Marine Chemistry. The U.S.GEOTRACES Eastern Tropical Pacific
Transect (GP16) 201, 77–89. doi:10.1016/j.marchem.2018.01.003 (2018).
32. Rose, A. L. & Waite, T. D. Chemiluminescence of Luminol in the Presence of Iron(II) and
Oxygen: Oxidation Mechanism and Implications for Its Analytical Use. English. Analytical
Chemistry 73, 5909–5920. doi:10.1021/ac015547q (2001).
33. Kubo, H. & Toriba, A. Chemiluminescence Flow Injection Analysis of Reducing Agents
Based on the Luminol Reaction. English. Analytica Chimica Acta 353, 345–349. doi:10.
1016/S0003-2670(97)87796-3 (1997).
34. Lu, C., Song, G. & Lin, J.-M. Reactive Oxygen Species and Their Chemiluminescence-
Detection Methods. English. Trac-Trends in Analytical Chemistry 25, 985–995. doi:10.
1016/j.trac.2006.07.007 (2006).
35. Wanty, R. B. & Goldhaber, M. B. Thermodynamics and Kinetics of Reactions Involving
Vanadium in Natural Systems: Accumulation of Vanadium in Sedimentary Rocks. Geochim-
ica et Cosmochimica Acta 56, 1471–1483. doi:10.1016/0016-7037(92)90217-7 (1992).
36. Roy, E. G., Wells, M. L. & King, D. W. Persistence of Iron(II) in Surface Waters of the
Western Subarctic Pacific. English. Limnology and Oceanography 53, 89–98. doi:10.4319/
lo.2008.53.1.0089 (2008).
37. Boiteau, R. M., Mende, D. R., Hawco, N. J., McIlvin, M. R., Fitzsimmons, J. N., Saito,
M. A., et al. Siderophore-Based Microbial Adaptations to Iron Scarcity across the Eastern
Pacific Ocean. en. Proceedings of the National Academy of Sciences 113, 14237–14242.
doi:10.1073/pnas.1608594113 (2016).
38. Team, R. C. R: A Language and Environment for Statistical Computing R Foundation for
Statistical Computing. Vienna, Austria, 2017.
39. Rapp, I., Schlosser, C., Rusiecka, D., Gledhill, M. & Achterberg, E. P. Automated Pre-
concentration of Fe, Zn, Cu, Ni, Cd, Pb, Co, and Mn in Seawater with Analysis Using
96
High-Resolution Sector Field Inductively-Coupled Plasma Mass Spectrometry. en. Analyt-
ica Chimica Acta 976, 1–13. doi:10.1016/j.aca.2017.05.008 (2017).
40. Garcia-Robledo, E., Padilla, C. C., Aldunate, M., Stewart, F. J., Ulloa, O., Paulmier, A.,
et al. Cryptic Oxygen Cycling in Anoxic Marine Zones. en. Proceedings of the National
Academy of Sciences 114, 8319–8324. doi:10.1073/pnas.1619844114 (2017).
41. Ohnemus, D. C., Rauschenberg, S., Cutter, G. A., Fitzsimmons, J. N., Sherrell, R. M. &
Twining, B. S. Elevated Trace Metal Content of Prokaryotic Communities Associated with
Marine Oxygen Deficient Zones: Elevated Trace Metals in ODZ Prokaryotes. en. Limnology
and Oceanography. doi:10.1002/lno.10363 (2016).
42. Ho, P., Lee, J.-M., Heller, M. I., Lam, P. J. & Shiller, A. M. The Distribution of Dissolved
and Particulate Mo and V along the U.S. GEOTRACES East Pacific Zonal Transect (GP16):
The Roles of Oxides and Biogenic Particles in Their Distributions in the Oxygen Deficient
Zone and the Hydrothermal Plume. Marine Chemistry. The U.S.GEOTRACES Eastern Trop-
ical Pacific Transect (GP16) 201, 242–255. doi:10.1016/j.marchem.2017.12.003
(2018).
43. Hering, J. G. & Morel, F. M. M. Kinetics of Trace Metal Complexation: Ligand-Exchange
Reactions. Environmental Science & Technology 24, 242–252. doi:10.1021/es00072a014
(1990).
44. Motekaitis, R. J. NIST Critically Selected Stability Constants of Metal Complexes Database
U.S. Department of Commerce. College Station, TX, 2004.
45. Skrabal, S. A., McBurney, A. M., Webb, L. A., Brooks Avery, G., Kieber, R. J. & Mead,
R. N. Photodissolution of Copper from Resuspended Coastal Marine Sediments. en. Lim-
nology and Oceanography, n/a–n/a. doi:10.1002/lno.10668 (2017).
46. Pinedo-Gonzalez, P., West, A. J., Rivera-Duarte, I. & Sa˜ nudo-Wilhelmy, S. A. Diel Changes
in Trace Metal Concentration and Distribution in Coastal Waters: Catalina Island As a Study
Case. en. Environmental Science & Technology 48, 7730–7737. doi:10.1021/es5019515
(2014).
47. Bange, H. W., Naqvi, S. W. A. & Codispoti, L. A. The Nitrogen Cycle in the Arabian
Sea. en. Progress in Oceanography. The Arabian Sea of the 1990s: New Biogeochemical
Understanding 65, 145–158. doi:10.1016/j.pocean.2005.03.002 (2005).
48. Karstensen, J., Stramma, L. & Visbeck, M. Oxygen Minimum Zones in the Eastern Tropical
Atlantic and Pacific Oceans. Progress in Oceanography. A New View of Water Masses After
WOCE. A Special Edition for Professor Matthias Tomczak 77, 331–350. doi:10.1016/j.
pocean.2007.05.009 (2008).
49. Moffett, J. W., Goeffert, T. J. & Naqvi, S. W. A. Reduced Iron Associated with Secondary
Nitrite Maxima in the Arabian Sea. English. Deep-Sea Research Part I-Oceanographic Re-
search Papers 54, 1341–1349. doi:10.1016/j.dsr.2007.04.004 (2007).
50. Vedamati, J., Goepfert, T. & Moffett, J. W. Iron Speciation in the Eastern Tropical South
Pacific Oxygen Minimum Zone off Peru. English. Limnology and Oceanography 59, 1945–
1957. doi:10.4319/lo.2014.59.6.1945 (2014).
51. Scholz, F., L¨ oscher, C. R., Fiskal, A., Sommer, S., Hensen, C., Lomnitz, U., et al. Nitrate-
Dependent Iron Oxidation Limits Iron Transport in Anoxic Ocean Regions. Earth and Plan-
etary Science Letters 454, 272–281. doi:10.1016/j.epsl.2016.09.025 (2016).
97
52. Heller, M. I., Lam, P. J., Moffett, J. W., Till, C. P., Lee, J.-M., Toner, B. M., et al. Ac-
cumulation of Fe Oxyhydroxides in the Peruvian Oxygen Deficient Zone Implies Non-
Oxygen Dependent Fe Oxidation. English. Geochimica Et Cosmochimica Acta 211, 174–
193. doi:10.1016/j.gca.2017.05.019 (2017).
53. Plass, A., Schlosser, C., Sommer, S., Dale, A. W., Achterberg, E. P. & Scholz, F. The Control
of Hydrogen Sulfide on Benthic Iron and Cadmium Fluxes in the Oxygen Minimum Zone
off Peru. English. Biogeosciences Discussions, 1–52. doi:10.5194/bg-2019-390 (2019).
54. Codispoti, L. A., Friederich, G. E., Packard, T. T., Glover, H. E., Kelly, P. J., Spinrad, R. W.,
et al. High Nitrite Levels off Northern Peru: A Signal of Instability in the Marine Deni-
trification Rate. en. Science 233, 1200–1202. doi:10.1126/science.233.4769.1200
(1986).
55. Dalsgaard, T., Canfield, D. E., Petersen, J., Thamdrup, B. & Acuna-Gonzalez, J. N-2 Pro-
duction by the Anammox Reaction in the Anoxic Water Column of Golfo Dulce, Costa
Rica. English. Nature 422, 606–608. doi:10.1038/nature01526 (2003).
56. Jensen, M. M., Lam, P., Revsbech, N. P., Nagel, B., Gaye, B., Jetten, M. S. M., et al. Inten-
sive Nitrogen Loss over the Omani Shelf Due to Anammox Coupled with Dissimilatory Ni-
trite Reduction to Ammonium. English. Isme Journal 5, 1660–1670. doi:10.1038/ismej.
2011.44 (2011).
57. Bonnet, S., Dekaezemacker, J., Turk-Kubo, K. A., Moutin, T., Hamersley, R. M., Grosso,
O., et al. Aphotic N-2 Fixation in the Eastern Tropical South Pacific Ocean. English. Plos
One 8, e81265. doi:10.1371/journal.pone.0081265 (2013).
58. Galan, A., Faundez, J., Thamdrup, B., Francisco Santibanez, J. & Farias, L. Temporal Dy-
namics of Nitrogen Loss in the Coastal Upwelling Ecosystem off Central Chile: Evidence of
Autotrophic Denitrification through Sulfide Oxidation. English. Limnology and Oceanog-
raphy 59, 1865–1878. doi:10.4319/lo.2014.59.6.1865 (2014).
59. Widner, B., Mordy, C. W. & Mulholland, M. R. Cyanate Distribution and Uptake above
and within the Eastern Tropical South Pacific Oxygen Deficient Zone: Cyanate in the ETSP
Oxygen Deficient Zone. en. Limnology and Oceanography 63, S177–S192. doi:10.1002/
lno.10730 (2018).
60. Babbin, A. R., Peters, B. D., Mordy, C. W., Widner, B., Casciotti, K. L. & Ward, B. B.
Multiple Metabolisms Constrain the Anaerobic Nitrite Budget in the Eastern Tropical South
Pacific. en. Global Biogeochemical Cycles, 2016GB005407. doi:10.1002/2016GB005407
(2017).
61. Moriyasu, R., Evans, Z. C., Bolster, K. M., Hardisty, D. S. & Moffett, J. W. The Distribution
and Redox Speciation of Iodine in the Eastern Tropical North Pacific Ocean. en. Global
Biogeochemical Cycles 34, e2019GB006302. doi:10.1029/2019GB006302 (2020).
62. Canfield, D. E., Stewart, F. J., Thamdrup, B., De Brabandere, L., Dalsgaard, T., Delong,
E. F., et al. A Cryptic Sulfur Cycle in Oxygen-Minimum-Zone Waters off the Chilean Coast.
English. Science 330, 1375–1378. doi:10.1126/science.1196889 (2010).
63. Bolster, K. M., Heller, M. I. & Moffett, J. W. Determination of Iron(II) by Chemilumines-
cence Using Masking Ligands to Distinguish Interferences. en. Limnology and Oceanogra-
phy: Methods 16, 750–759. doi:10.1002/lom3.10279 (2018).
64. Moffett, J. W. & German, C. R. Distribution of Iron in the Western Indian Ocean and the
Eastern Tropical South Pacific: An Inter-Basin Comparison. en. Chemical Geology 532,
119334. doi:10.1016/j.chemgeo.2019.119334 (2020).
98
65. Schlitzer, R., Anderson, R. F., Dodas, E. M., Lohan, M., Geibert, W., Tagliabue, A., et al.
The GEOTRACES Intermediate Data Product 2017. en. Chemical Geology 493, 210–223.
doi:10.1016/j.chemgeo.2018.05.040 (2018).
66. Scholz, F., McManus, J. & Sommer, S. The Manganese and Iron Shuttle in a Modern Eux-
inic Basin and Implications for Molybdenum Cycling at Euxinic Ocean Margins. en. Chem-
ical Geology 355, 56–68. doi:10.1016/j.chemgeo.2013.07.006 (2013).
67. Reed, D. C., Gustafsson, B. G. & Slomp, C. P. Shelf-to-Basin Iron Shuttling Enhances
Vivianite Formation in Deep Baltic Sea Sediments. en. Earth and Planetary Science Letters
434, 241–251. doi:10.1016/j.epsl.2015.11.033 (2016).
68. Landing, W. M. & Bruland, K. W. The Contrasting Biogeochemistry of Iron and Manganese
in the Pacific Ocean. en. Geochimica et Cosmochimica Acta 51, 29–43. doi:10.1016/0016-
7037(87)90004-4 (1987).
69. Stone, A. T. & Morgan, J. J. Reduction and Dissolution of Manganese(III) and Manganese(IV)
Oxides by Organics: 2. Survey of the Reactivity of Organics. Environmental Science & Tech-
nology 18, 617–624. doi:10.1021/es00126a010 (1984).
70. Oldham, V . E., Mucci, A., Tebo, B. M. & Luther, G. W. Soluble Mn(III)–L Complexes
Are Abundant in Oxygenated Waters and Stabilized by Humic Ligands. en. Geochimica et
Cosmochimica Acta 199, 238–246. doi:10.1016/j.gca.2016.11.043 (2017).
71. Sunda, W. G. & Huntsman, S. A. Effect of Sunlight on Redox Cycles of Manganese in
the Southwestern Sargasso Sea. en. Deep Sea Research Part A. Oceanographic Research
Papers 35, 1297–1317. doi:10.1016/0198-0149(88)90084-2 (1988).
72. Sunda, W. G., Huntsman, S. A. & Harvey, G. R. Photoreduction of Manganese Oxides
in Seawater and Its Geochemical and Biological Implications. en. Nature 301, 234–236.
doi:10.1038/301234a0 (1983).
73. Sunda, W. G. & Huntsman, S. A. Photoreduction of Manganese Oxides in Seawater. en.
Marine Chemistry. 12th International Symposium @’Chemistry of the Mediterranean@’
46, 133–152. doi:10.1016/0304-4203(94)90051-5 (1994).
74. Clement, B. G., Luther, G. W. & Tebo, B. M. Rapid, Oxygen-Dependent Microbial Mn(II)
Oxidation Kinetics at Sub-Micromolar Oxygen Concentrations in the Black Sea Suboxic
Zone. en. Geochimica et Cosmochimica Acta 73, 1878–1889. doi:10.1016/j.gca.2008.
12.023 (2009).
75. Saager, P. M., De Baar, H. J. W. & Burkill, P. H. Manganese and Iron in Indian Ocean
Waters. Geochimica et Cosmochimica Acta 53, 2259–2267. doi:10.1016/0016-7037(89)
90348-7 (1989).
76. Lewis, B. L. & Luther III, G. W. Processes Controlling the Distribution and Cycling of
Manganese in the Oxygen Minimum Zone of the Arabian Sea. Deep Sea Research Part II:
Topical Studies in Oceanography 47, 1541–1561. doi:10.1016/S0967-0645(99)00153-8
(2000).
77. Johnson, K. S., Berelson, W. M., Coale, K. H., Coley, T. L., Elrod, V . A., Fairey, W. R., et
al. Manganese Flux from Continental Margin Sediments in a Transect Through the Oxygen
Minimum. en. Science 257, 1242–1245. doi:10.1126/science.257.5074.1242 (1992).
78. Johnson, K. S., Coale, K. H., Berelson, W. M. & Gordon, R. M. On the Formation of the
Manganese Maximum in the Oxygen Minimum. English. Geochimica Et Cosmochimica
Acta 60, 1291–1299. doi:10.1016/0016-7037(96)00005-1 (1996).
99
79. Vedamati, J., Chan, C. & Moffett, J. W. Distribution of Dissolved Manganese in the Peruvian
Upwelling and Oxygen Minimum Zone. Geochimica et Cosmochimica Acta 156, 222–240.
doi:10.1016/j.gca.2014.10.026 (2015).
80. Mangin, T., Cisneros-Mata, M.
´
A., Bone, J., Costello, C., Gaines, S. D., McDonald, G.,
et al. The Cost of Management Delay: The Case for Reforming Mexican Fisheries Sooner
Rather than Later. en. Marine Policy 88, 1–10. doi:10.1016/j.marpol.2017.10.042
(2018).
81. Ito, T., Minobe, S., Long, M. C. & Deutsch, C. Upper Ocean O
2
Trends: 1958-2015. en.
Geophysical Research Letters 44, 4214–4223. doi:10.1002/2017GL073613 (2017).
82. Water: A Shared Responsibility (eds UNESCO & Nations), W. W. A. P. () United Nations
World Water Development Report 2 (United Nations Educational, Scientific and Cultural
Organization (UNESCO) ; Berghahn Books, Paris, France : New York, 2006).
83. Selden, C. R., Mulholland, M. R., Bernhardt, P. W., Widner, B., Mac´ ıas-Tapia, A., Ji, Q., et
al. Dinitrogen Fixation Across Physico-Chemical Gradients of the Eastern Tropical North
Pacific Oxygen Deficient Zone. en. Global Biogeochemical Cycles 33, 1187–1202. doi:10.
1029/2019GB006242 (2019).
84. Kassambara, A. & Mundt, F. Factoextra: Extract and Visualize the Results of Multivariate
Data Analyses (2019).
85. Bruland, K. W., Rue, E. L. & Smith, G. J. Iron and Macronutrients in California Coastal
Upwelling Regimes: Implications for Diatom Blooms. en. Limnology and Oceanography
46, 1661–1674. doi:10.4319/lo.2001.46.7.1661 (2001).
86. Bruland, K. W., Rue, E. L., Smith, G. J. & DiTullio, G. R. Iron, Macronutrients and Di-
atom Blooms in the Peru Upwelling Regime: Brown and Blue Waters of Peru. en. Marine
Chemistry 93, 81–103. doi:10.1016/j.marchem.2004.06.011 (2005).
87. Chase, Z., Strutton, P. G. & Hales, B. Iron Links River Runoff and Shelf Width to Phyto-
plankton Biomass along the U.S. West Coast. en. Geophysical Research Letters 34. doi:10.
1029/2006GL028069 (2007).
88. Spiess, F. N., Macdonald, K. C., Atwater, T., Ballard, R., Carranza, A., Cordoba, D., et al.
East Pacific Rise: Hot Springs and Geophysical Experiments. en. Science 207, 1421–1433.
doi:10.1126/science.207.4438.1421 (1980).
89. Beaulieu, S. & Szafranski, K. InterRidge Global Database of Active Submarine Hydrother-
mal Vent Fields, Version 3.4. World Wide Web Electronic Publication Available from http://vents-
data.interridge.org 2020.
90. Baker, E. T., Cormier, M.-H., Langmuir, C. H. & Zavala, K. Hydrothermal Plumes along
Segments of Contrasting Magmatic Influence, 15°20
0
–18°30
0
N, East Pacific Rise: Influence
of Axial Faulting. en. Geochemistry, Geophysics, Geosystems 2. doi:10.1029/2000GC000165
(2001).
91. Talley, L. D., Pickard, G. L., Emery, W. J. & Swift, J. H. en. in Descriptive Physical
Oceanography (Sixth Edition) (eds Talley, L. D., Pickard, G. L., Emery, W. J. & Swift, J. H.)
303–362 (Academic Press, Boston, 2011). doi:10.1016/B978-0-7506-4552-2.10010-1.
92. Evans, N., Schroeder, I. D., Buil, M. P., Jacox, M. G. & Bograd, S. J. Drivers of Subsur-
face Deoxygenation in the Southern California Current System. en. Geophysical Research
Letters 47, e2020GL089274. doi:10.1029/2020GL089274 (2020).
93. Peters, B. D., Jenkins, W. J., Swift, J. H., German, C. R., Moffett, J. W., Cutter, G. A., et al.
Water Mass Analysis of the 2013 US GEOTRACES Eastern Pacific Zonal Transect (GP16).
100
Marine Chemistry. The U.S.GEOTRACES Eastern Tropical Pacific Transect (GP16) 201,
6–19. doi:10.1016/j.marchem.2017.09.007 (2018).
94. Lam, P., Heller, M. I., Lerner, P. E., Moffett, J. W. & Buck, K. Unexpected Source and
Transport of Iron from the Deep Peru Margin. ACS Earth and Space Chemistry. doi:10.
1021/acsearthspacechem.0c00066 (2020).
95. Scholz, F., Severmann, S., McManus, J. & Hensen, C. Beyond the Black Sea Paradigm: The
Sedimentary Fingerprint of an Open-Marine Iron Shuttle. en. Geochimica et Cosmochimica
Acta 127, 368–380. doi:10.1016/j.gca.2013.11.041 (2014).
96. Noffke, A., Hensen, C., Sommer, S., Scholz, F., Bohlen, L., Mosch, T., et al. Benthic Iron
and Phosphorus Fluxes across the Peruvian Oxygen Minimum Zone. en. Limnology and
Oceanography 57, 851–867. doi:10.4319/lo.2012.57.3.0851 (2012).
97. Kraal, P., Slomp, C. P., Reed, D. C., Reichart, G.-J. & Poulton, S. W. Sedimentary Phos-
phorus and Iron Cycling in and below the Oxygen Minimum Zone of the Northern Arabian
Sea. English. Biogeosciences 9, 2603–2624. doi:10.5194/bg-9-2603-2012 (2012).
98. Scholz, F., Schmidt, M., Hensen, C., Eroglu, S., Geilert, S., Gutjahr, M., et al. Shelf-to-Basin
Iron Shuttle in the Guaymas Basin, Gulf of California. en. Geochimica et Cosmochimica
Acta 261, 76–92. doi:10.1016/j.gca.2019.07.006 (2019).
99. Dale, A. W., Nickelsen, L., Scholz, F., Hensen, C., Oschlies, A. & Wallmann, K. A Revised
Global Estimate of Dissolved Iron Fluxes from Marine Sediments. en. Global Biogeochem-
ical Cycles 29, 691–707. doi:10.1002/2014GB005017 (2015).
100. Scheidegger, K. F. & Krissek, L. A. Dispersal and Deposition of Eolian and Fluvial Sed-
iments off Peru and Northern Chile. en. GSA Bulletin 93, 150–162. doi:10.1130/0016-
7606(1982)93<150:DADOEA>2.0.CO;2 (1982).
101. Rapp, I., Schlosser, C., Browning, T. J., Wolf, F., Le Moigne, F. A. C., Gledhill, M., et al. El
Ni˜ no-Driven Oxygenation Impacts Peruvian Shelf Iron Supply to the South Pacific Ocean.
en. Geophysical Research Letters 47. doi:10.1029/2019GL086631 (2020).
102. Noble, A. E., Lamborg, C. H., Ohnemus, D. C., Lam, P. J., Goepfert, T. J., Measures, C. I., et
al. Basin-Scale Inputs of Cobalt, Iron, and Manganese from the Benguela-Angola Front to
the South Atlantic Ocean. en. Limnology and Oceanography 57, 989–1010. doi:10.4319/
lo.2012.57.4.0989 (2012).
103. Hatta, M., Measures, C. I., Wu, J., Roshan, S., Fitzsimmons, J. N., Sedwick, P., et al. An
Overview of Dissolved Fe and Mn Distributions during the 2010–2011 U.S. GEOTRACES
North Atlantic Cruises: GEOTRACES GA03. en. Deep Sea Research Part II: Topical Stud-
ies in Oceanography. GEOTRACES GA-03 - The U.S. GEOTRACES North Atlantic Transect
116, 117–129. doi:10.1016/j.dsr2.2014.07.005 (2015).
104. Resing, J. A., Sedwick, P. N., German, C. R., Jenkins, W. J., Moffett, J. W., Sohst, B. M.,
et al. Basin-Scale Transport of Hydrothermal Dissolved Metals across the South Pacific
Ocean. Nature 523, 200–203. doi:10.1038/nature14577 (2015).
105. Thamdrup, B., Dalsgaard, T. & Revsbech, N. P. Widespread Functional Anoxia in the Oxy-
gen Minimum Zone of the Eastern South Pacific. Deep Sea Research Part I: Oceanographic
Research Papers 65, 36–45. doi:10.1016/j.dsr.2012.03.001 (2012).
106. DeVries, T., Deutsch, C., Primeau, F., Chang, B. & Devol, A. Global Rates of Water-Column
Denitrification Derived from Nitrogen Gas Measurements. English. Nature Geoscience 5,
547–550. doi:10.1038/NGEO1515 (2012).
101
107. Moffett, J. W., Goepfert, T. J. & Naqvi, S. W. A. Reduced Iron Associated with Secondary
Nitrite Maxima in the Arabian Sea. Deep Sea Research Part I: Oceanographic Research
Papers 54, 1341–1349. doi:10.1016/j.dsr.2007.04.004 (2007).
108. Vedamati, J., Goepfert, T. & Moffett, J. W. Iron Speciation in the Eastern Tropical South
Pacific Oxygen Minimum Zone off Peru. en. Limnology and Oceanography 59, 1945–1957.
doi:10.4319/lo.2014.59.6.1945 (2014).
109. Cutter, G. A., Moffett, J. G., Nielsd´ ottir, M. C. & Sanial, V . Multiple Oxidation State Trace
Elements in Suboxic Waters off Peru: In Situ Redox Processes and Advective/Diffusive
Horizontal Transport. en. Marine Chemistry. doi:10.1016/j.marchem.2018.01.003
(2018).
110. Heller, M. I., Lam, P. J., Moffett, J. W., Till, C. P., Lee, J.-M., Toner, B. M., et al. Accumu-
lation of Fe Oxyhydroxides in the Peruvian Oxygen Deficient Zone Implies Non-Oxygen
Dependent Fe Oxidation. Geochimica et Cosmochimica Acta 211, 174–193. doi:10.1016/
j.gca.2017.05.019 (2017).
111. Raven, M. R., Keil, R. G. & Webb, S. M. Microbial Sulfate Reduction and Organic Sul-
fur Formation in Sinking Marine Particles. en. Science. doi:10.1126/science.abc6035
(2020).
112. Pettine, M., D’Ottone, L., Campanella, L., Millero, F. J. & Passino, R. The Reduction of
Chromium (VI) by Iron (II) in Aqueous Solutions. en. Geochimica et Cosmochimica Acta
62, 1509–1519. doi:10.1016/S0016-7037(98)00086-6 (1998).
113. Bruggmann, S., Scholz, F., Klaebe, R. M., Canfield, D. E. & Frei, R. Chromium Isotope Cy-
cling in the Water Column and Sediments of the Peruvian Continental Margin. Geochimica
et Cosmochimica Acta. doi:10.1016/j.gca.2019.05.001 (2019).
114. Frei, R., Gaucher, C., Poulton, S. W. & Canfield, D. E. Fluctuations in Precambrian Atmo-
spheric Oxygenation Recorded by Chromium Isotopes. en. Nature 461, 250–253. doi:10.
1038/nature08266 (2009).
115. Sørensen, J. & Thorling, L. Stimulation by Lepidocrocite (7-FeOOH) of Fe(II)-Dependent
Nitrite Reduction. en. Geochimica et Cosmochimica Acta 55, 1289–1294. doi:10.1016/
0016-7037(91)90307-Q (1991).
116. Melton, E. D., Swanner, E. D., Behrens, S., Schmidt, C. & Kappler, A. The Interplay of
Microbially Mediated and Abiotic Reactions in the Biogeochemical Fe Cycle. en. Nature
Reviews Microbiology 12, 797–808. doi:10.1038/nrmicro3347 (2014).
117. Ilbert, M. & Bonnefoy, V . Insight into the Evolution of the Iron Oxidation Pathways. en.
Biochimica et Biophysica Acta (BBA) - Bioenergetics. The Evolutionary Aspects of Bioen-
ergetic Systems 1827, 161–175. doi:10.1016/j.bbabio.2012.10.001 (2013).
118. H. E. Garcia, K. Weathers, C. R. Paver, I. Smolyar, T. P. Boyer, R. A. Locarnini, et al. World
Ocean Atlas 2018, Volume 3: Dissolved Oxygen, Apparent Oxygen Utilization, and Oxygen
Saturation (Ed. Mishonov, A.) 2019.
119. Larsen, M., Lehner, P., Borisov, S. M., Klimant, I., Fischer, J. P., Stewart, F. J., et al. In Situ
Quantification of Ultra-Low o
2
Concentrations in Oxygen Minimum Zones: Application
of Novel Optodes: In Situ Trace Sensing of o
2
Using Novel Optodes. en. Limnology and
Oceanography: Methods 14, 784–800. doi:10.1002/lom3.10126 (2016).
120. Millero, F. J., Sotolongo, S. & Izaguirre, M. The Oxidation Kinetics of Fe(II) in Seawater.
en. Geochimica et Cosmochimica Acta 51, 793–801. doi:10.1016/0016-7037(87)90093-
7 (1987).
102
121. Cutter, G., Casciotti, K., Croot, P., Geibert, W., Heimb¨ urger, L., Lohan, M., et al. Sampling
and Sample-Handling Protocols for GEOTRACES Cruises, Version 3 2017.
122. Tovar-Sanchez, A., Sa˜ nudo-Wilhelmy, S. A., Garcia-Vargas, M., Weaver, R. S., Popels, L. C.
& Hutchins, D. A. A Trace Metal Clean Reagent to Remove Surface-Bound Iron from
Marine Phytoplankton. en. Marine Chemistry 82, 91–99. doi:10.1016/S0304-4203(03)
00054-9 (2003).
123. Planquette, H. & Sherrell, R. M. Sampling for Particulate Trace Element Determination
Using Water Sampling Bottles: Methodology and Comparison to in Situ Pumps: Particulate
Trace Element Sampling. en. Limnology and Oceanography: Methods 10, 367–388. doi:10.
4319/lom.2012.10.367 (2012).
124. Mooy, B. A. S. V . & Keil, R. G. en. US9188512B2 (2015).
125. Sunda, W. G. Feedback Interactions between Trace Metal Nutrients and Phytoplankton in
the Ocean. Frontiers in Microbiology 3. doi:10.3389/fmicb.2012.00204 (2012).
126. Kondo, Y . & Moffett, J. W. Iron Redox Cycling and Subsurface Offshore Transport in the
Eastern Tropical South Pacific Oxygen Minimum Zone. en. Marine Chemistry 168, 95–103.
doi:10.1016/j.marchem.2014.11.007 (2015).
127. Loescher, C. R., Bange, H. W., Schmitz, R. A., Callbeck, C. M., Engel, A., Hauss, H., et al.
Water Column Biogeochemistry of Oxygen Minimum Zones in the Eastern Tropical North
Atlantic and Eastern Tropical South Pacific Oceans. English. Biogeosciences 13, 3585–
3606. doi:10.5194/bg-13-3585-2016 (2016).
128. Valentine, J. S., Sheridan, R. P., Allen, L. C. & Kahn, P. C. Coupling between Oxidation
State and Hydrogen Bond Conformation in Heme Proteins. Proceedings of the National
Academy of Sciences 76, 1009–1013. doi:10.1073/pnas.76.3.1009 (1979).
129. Rue, E. L., Smith, G. J., Cutter, G. A. & Bruland, K. W. The Response of Trace Element
Redox Couples to Suboxic Conditions in the Water Column. en. Deep Sea Research Part I:
Oceanographic Research Papers 44, 113–134. doi:10.1016/S0967-0637(96)00088-X
(1997).
130. Bianchi, D., Weber, T. S., Kiko, R. & Deutsch, C. Global Niche of Marine Anaerobic
Metabolisms Expanded by Particle Microenvironments. Nature Geoscience 11, 263–268.
doi:10.1038/s41561-018-0081-0 (4 2018).
131. Rapp, I., Schlosser, C., Rusiecka, D., Gledhill, M. & Achterberg, E. P. Automated Pre-
concentration of Fe, Zn, Cu, Ni, Cd, Pb, Co, and Mn in Seawater with Analysis Using
High-Resolution Sector Field Inductively-Coupled Plasma Mass Spectrometry. Analytica
Chimica Acta 976, 1–13. doi:10.1016/j.aca.2017.05.008 (2017).
132. Garc´ ıa-Robledo, E., Corzo, A. & Papaspyrou, S. A Fast and Direct Spectrophotometric
Method for the Sequential Determination of Nitrate and Nitrite at Low Concentrations in
Small V olumes. en. Marine Chemistry 162, 30–36. doi:10.1016/j.marchem.2014.03.
002 (2014).
133. Galhardo, C. X. & Masini, J. C. Spectrophotometric Determination of Phosphate and Sili-
cate by Sequential Injection Using Molybdenum Blue Chemistry. Analytica Chimica Acta
417, 191–200. doi:10.1016/S0003-2670(00)00933-8 (2000).
134. Holmes, R. M., Aminot, A., K´ erouel, R., Hooker, B. A. & Peterson, B. J. A Simple and Pre-
cise Method for Measuring Ammonium in Marine and Freshwater Ecosystems. Canadian
Journal of Fisheries and Aquatic Sciences 56, 1801–1808. doi:10.1139/f99-128 (1999).
103
135. Lam, P., Lavik, G., Jensen, M. M., van de V ossenberg, J., Schmid, M., Woebken, D., et al.
Revising the Nitrogen Cycle in the Peruvian Oxygen Minimum Zone. Proceedings of the
National Academy of Sciences of the United States of America 106, 4752–4757. doi:10.
1073/pnas.0812444106 (2009).
136. Hopkinson, B. M. & Barbeau, K. A. Organic and Redox Speciation of Iron in the Eastern
Tropical North Pacific Suboxic Zone. English. Marine Chemistry 106, 2–17. doi:10.1016/
j.marchem.2006.02.008 (2007).
137. Scholz, F., Severmann, S., McManus, J. & Hensen, C. Beyond the Black Sea Paradigm: The
Sedimentary Fingerprint of an Open-Marine Iron Shuttle. en. Geochimica et Cosmochimica
Acta 127, 368–380. doi:10.1016/j.gca.2013.11.041 (2014).
138. Lam, P. J., Heller, M. I., Lerner, P. E., Moffett, J. W. & Buck, K. N. Unexpected Source and
Transport of Iron from the Deep Peru Margin. ACS Earth and Space Chemistry 4, 977–992.
doi:10.1021/acsearthspacechem.0c00066 (2020).
139. John, S. G., Liang, H., Weber, T., DeVries, T., Primeau, F., Moore, K., et al. AWESOME
OCIM: A Simple, Flexible, and Powerful Tool for Modeling Elemental Cycling in the
Oceans. en. Chemical Geology 533, 119403. doi:10.1016/j.chemgeo.2019.119403
(2020).
140. Severmann, S., McManus, J., Berelson, W. M. & Hammond, D. E. The Continental Shelf
Benthic Iron Flux and Its Isotope Composition. English. Geochimica Et Cosmochimica Acta
74, 3984–4004. doi:10.1016/j.gca.2010.04.022 (2010).
141. Millero, F. J., Sotolongo, S. & Izaguirre, M. The Oxidation Kinetics of Fe(II) in Seawater.
Geochimica et Cosmochimica Acta 51, 793–801. doi:10.1016/0016-7037(87)90093-7
(1987).
104

Abstract (if available)

Abstract This thesis contains several studies focused on iron cycling and transport in the Eastern Tropical North Pacific (ETNP) oxygen deficient zone (ODZ). This thesis will argue that biogeochemical cycling within the ODZ drives significant transport of iron away from the continental margins of Mexico to the oligotrophic ocean. ❧ Iron is generally abundant for land dwelling organisms, since it makes up a significant amount of Earth's crust. However, in the ocean, iron will typically oxidize and form either iron oxide particles, or it will adsorb onto other large particles sinking through the ocean, leaving only trace amounts at the ocean surface. Iron availability may have changed significantly over geological time, and is a key component in the cycle of glacial and interglacial periods that Earth experiences every few millennia, since it impacts the ability of the oceans to absorb carbon dioxide from the atmosphere. From this perspective, there are strong motivations for marine scientists to study iron. ❧ ODZs are open ocean regions in which the dissolved oxygen concentration drops to negligible levels, as a result of high productivity and slow circulation. These are anoxic settings, but rather than being dominated by sulfate reduction processes like many enclosed basins, ODZs are dominated by nitrogen redox processes. ❧ In the past two decades, it was discovered that all three ODZs contain elevated concentrations of iron in the +2 oxidation state. Because the scarcity of iron is strongly impacted by oxygen, it seemed reasonable to assume that the processes of iron oxidation and scavenging would be inhibited in ODZs, where oxygen is absent. However, there is a thermodynamic challenge to this argument. Although oxygen is typically the oxidant of iron in the ocean, nitrate and nitrite both have high enough redox potentials for them to make iron oxidation thermodynamically favorable. In sediments, for instance, iron(II) is not found in layers with detectable nitrate or nitrite, since these reactions can occur rapidly under certain conditions. ODZs, although depleted in oxygen, contain significant concentrations of nitrate and nitrite. Why would iron(II) then accumulate in these environments? ❧ Anaerobic iron oxidation would not necessarily inhibit iron transport in these regions. A process known as the shelf to basin shuttle was believed to be active in other anoxic basins, in which a cycle of iron oxidation in the water column and reduction in sediments could progressively move iron from coastal margins towards the continental shelves, forming deep plumes that could transport iron significant distances. Shortly before the work in this thesis began, such a plume was observed off the coast of Peru, below the Eastern Tropical South Pacific ODZ. ❧ This thesis attempts to answer five major questions about iron cycling in the ETNP, listed below. ❧ 1. How can we accurately measure iron(II) in ODZs? ❧ 2. Is the shelf to basin shuttle active in the ETNP ODZ? ❧ 3. What is the rate of anaerobic iron oxidation in the ETNP ODZ? ❧ 4. Is it possible for iron(II) to accumulate in an ODZ independent of reducing sediments? ❧ 5. What is the overall rate of transport of iron through the ETNP ODZ? ❧ Each of these questions is addressed in its own chapter. Chapter 1 describes a technique to measure iron(II) in ODZs using luminol chemiluminescence and correcting for interfering compounds. Chapter 2 describes the distribution of iron(II) (and other trace metals) throughout the ETNP, and analyzes the controls on that distribution. Chapter 3 contains measurements of the rate of anaerobic iron oxidation using an incubation approach, and evaluates possible mechanisms of that oxidation. Chapter 4 describes a mesocosm which simulates the water column of an ODZ while not containing sediments, allowing experiments on exclusively water cycle processes. Finally, Chapter 5 includes attempts to model the iron cycling studied in Chapters 2-4, and to use that model to estimate the overall iron transport in the region.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Discerning local and long-range causes of deoxygenation and their impact on the accumulation of trace, reduced compounds

PDF

Comparative behavior and distribution of biologically relevant trace metals - iron, manganese, and copper in four representative oxygen deficient regimes of the world's oceans

PDF

The distributions and geochemistry of iodine and copper in the Pacific Ocean

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

Going with the flow: constraining the lateral advection of redox-active metals from continental margins under differing oxygen regimes

PDF

Oxygen uptake rates in the thermocline of coastal waters: assessing the role of carbon and inorganic nutrient inputs

PDF

Understanding the mechanism of oxygen reduction and oxygen evolution on transition metal oxide electrocatalysts and applications in iron-air rechargeable battery

PDF

Modeling deep ocean water and sediment dynamics in the eastern Pacific Ocean using actinium-227 and other naturally occurring radioisotopes

PDF

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

PDF

Investigating the global ocean biogeochemical cycling of alkalinity, barium, and copper using data-constrained inverse models

PDF

The marine biogeochemistry of nickel isotopes

PDF

Actinium-227 as a tracer for mixing in the Deep Northeast Pacific

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

PDF

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

PDF

Germanium and silicon isotope geochemistry in terrestrial and marine low-temperature environments

PDF

Microbial metabolism in deep subsurface sediments of Guaymas Basin (Gulf of California): methanogenesis, methylotrophy, and asgardarchaeota

PDF

Ages, origins and biogeochemical role of water across a tropical mountain to floodplain transition

PDF

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

PDF

A low detection limit sulfide measurement method in marine environments

PDF

Environmental controls on alkalinity generation from mineral dissolution: from the mineral surface to the global ocean

PDF

Diagenesis of C, N, and Si in marine sediments from the Western Tropical North Atlantic and Eastern Subtropical North Pacific: pore water models and sedimentary studies

Asset Metadata

Creator Bolster, Kenneth McCarthy (author)

Core Title Anaerobic iron cycling in an oxygen deficient zone

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Geological Sciences

Degree Conferral Date 2021-08

Publication Date 07/14/2021

Defense Date 05/12/2021

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

Tag anaerobic,geobiology,Iron,iron cycling,iron redox,OAI-PMH Harvest,oceanography,ODZ,OMZ,oxygen deficient zone,oxygen minimum zone,trace metal

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Moffett, James (committee chair), Capone, Douglas (committee member), Hammond, Douglas (committee member), John, Seth (committee member)

Creator Email kbolster@usc.edu,kennybolster@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15491155

Unique identifier UC15491155

Legacy Identifier etd-BolsterKen-9736

Document Type Dissertation

Format application/pdf (imt)

Rights Bolster, Kenneth McCarthy

Internet Media Type application/pdf

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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu