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

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INVESTIGATIONS ON MARINE METAL CYCLING THROUGH A GLOBAL EXPEDITION, A
WILDFIRE SURVEY, AND A VIRAL INFECTION

by

Rachel Lauren Kelly

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

MAY 2022

Copyright 2022 Rachel Lauren Kelly
ii
Acknowledgements
The work in this thesis would not have been possible without the financial support
of the John Lab by the Simons Foundation (SCOPE Award 329108; throughout my Ph.D.),
the University of Southern California (throughout my Ph.D.), the Tara Ocean Foundation
(Chapter 2 project), the National Science Foundation (NSF-OCE 1736896; Chapter 3
project), and the Gordon and Betty Moore Foundation (Chapter 4 project).
I would first like to thank Seth John for being an incredible advisor, not only
throughout my Ph.D., but also during my time as an undergraduate student. If I had not
taken Seth’s Chemical Oceanography class and applied to work in his lab in undergrad, I
am fairly certain I would not be here writing this thesis. I will forever be grateful to Seth for
fostering me throughout my early research career and for his constant enthusiasm and
patience over the past 8 years.
Thank you to my thesis committee: Jim Moffett, Dave Hutchins, and Josh West, for
their guidance and mentorship. Jim showed me the ins-and-outs of fieldwork by taking me
on my first research cruise and has continually pushed me to become a better scientist. I
learned so much about trace metal clean-culturing through Dave and it’s been an honor to
learn from the best of the best. I am very grateful to Josh for conceptualizing the Thomas
Fire study and for his help throughout the project.
Special thanks to all the members of the John Lab, I truly could not have survived 6
years at sea or in the lab without them. Nick Hawco and Paulina Pinedo-González helped
me find my footing in the lab during the first years of my Ph.D. I aspire to be a great
scientist and mentor like each of them someday. Xiaopeng Bian and Shun-Chung Yang
iii
miraculously filled the void Nick and Paulina left and are some of the most generous
people I know. I am indebted to them after the endless hours they have spent helping me
on the Element. I could not have asked for a better office mate than Hengdi Liang who is
always so upbeat and kind. Emily Seelen, Emily Townsend, and Phil Kong have brought so
much fun and joy to the John Lab and the last few years of my Ph.D. have been a blast
thanks to the three of them.
A huge shoutout is in order to all of my collaborators-turned-friends which,
unfortunately, there are too many to list here. The Tara Pacific project would not have been
as enjoyable without the excitement and assistance provided by both Fabien Lombard and
Natalie Cohen. During the Thomas Fire project, Abra Atwood quite literally saved my back
by helping me collect some of the river samples and has metaphorically had my back
throughout grad school. On top of being amazing scientists, Lydia Babcock-Adams,
Marianne Acker, Daniel Muratore, and Jenn Beatty are caring friends who have made
many sleepless nights on the R/V Kilo Moana memorable instead of miserable.
Thank you to all of those who have supported me outside of my science and have
kept me sane over the past 6 years, in particular: Natalie Sajovec who edited all of my grad
school apps. Mary Malloy, the best big sister I never had, who coached me through the
Ph.D. process and Loretta Malloy, my other “sister”, who has been there with me through
thick and thin for the past 23 years. Jaclyn Pittman, Naomi Rodgers, and Mathilde Wimez
who were always there to be my sounding board and are my family at USC. Kate
Weinheimer who has made life in Los Angeles so much more fun and Abby Bussard for
being my first best friend in LA. JP who is the best partner I could ask for and my safe
iv
place. My family who are my ultimate support group, especially my brothers James, John,
and Tom who have been amazing role models throughout my life. Last but definitely not
least, my parents, Tom and Yon, who have always encouraged me to pursue whatever I
wanted to do in life and are the most selfless people I know. None of this would have been
possible without them.

v
Table of Contents Page
Acknowledgments ii
List of Tables ix
List of Figures x
Abstract xii

Chapter 1 – Introduction 1
1. A history of trace metal marine chemistry 1
2. Global patterns of surface ocean metals 3
3. Wildfires as a potentially important marine metal source 4
4. Iron cycling through viral infection 5
5. The work presented in this thesis 6

Chapter 2 – High resolution surface metal sampling of the North Atlantic, North Pacific,
and South Pacific during the Tara Pacific expedition 9
Abstract 9
1. Introduction 10
2. Methods 14
2.1 Sampling 14
2.2 Trace metal analyses 15
2.3 Data quality 16
2.4 Data management 17
2.5 Machine learning techniques 17
2.5.1 Subsetting GEOTRACES data 17
2.5.2 Regression tree algorithm 18
3. Results and Discussion 19
3.1 Reliability of pole sampling 19
3.2 The importance of metal sampling 22
3.3 Patterns between environmental and metal data 23
3.3.1 Major nutrients 24
3.3.2 Salinity 25
3.3.3 Location 26
3.4 Observational comparison to GEOTRACES 27
3.5 Biogeochemical observations 28
3.5.1 Coastal biome 28
3.5.1.1 Upwelling of Cd 29
3.5.1.2 Remobilized Cu 31
3.5.1.3 Coastal Ni and Pb 32
3.5.2 Westerly biome 33
3.5.2.1 Fe, Mn, Co, and Saharan dust 34
3.5.3 Trade biome 37
3.5.3.1 Pb in Asian aerosols 38
vi
3.5.3.2 Fe and Mn in the South Pacific Islands 38
4. Conclusion 39
Tables 41
Figures 45
Supplementary tables and figures 52

Chapter 3 – Delivery of metals and dissolved black carbon to the southern California
coastal ocean via aerosols and floodwaters following the 2017 Thomas
Fire 67
Abstract 67
1. Introduction 68
2. Materials and Methods 71
2.1 Field sampling 71
2.2 Organic analyses 73
2.2.1 Aerosol samples 73
2.2.1.1 Particulate mass (PM2.5) 73
2.2.1.2 Levoglucosan (in TSP) 74
2.2.1.3 Black carbon (in PM2.5 and TSP) 74
2.2.2 Surface seawater samples 75
2.2.2.1 Carbon isotopes 75
2.2.2.2 Chlorophyll a 76
2.2.2.3 Dissolved organic carbon and dissolved black carbon 76
2.2.3 River samples 77
2.2.3.1 Dissolved organic carbon and dissolved black carbon 77
2.3 Trace metal analyses 77
2.3.1 Aerosol samples 77
2.3.1.1 Instantaneous leach (soluble concentrations) 78
2.3.1.2 Total digestion (total concentrations) 78
2.3.2 Surface seawater samples 80
2.3.2.1 Dissolved concentrations 80
2.3.3 River samples 80
2.3.3.1 Dissolved and soluble concentrations 80
2.4 Other analyses 82
2.4.1 Nutrients 82
2.4.1.1 Dissolved concentrations 82
3. Results 82
3.1 Organic analyses 82
3.1.1 Aerosol samples 82
3.1.2 Surface seawater samples 83
3.1.3 River samples 84
3.2 Trace metal analyses 84
3.2.1 Aerosol samples 84
3.2.2 Surface seawater samples 84
3.2.3 River samples 85
vii
3.3 Other analyses 86
3.3.1 Nutrients 86
4. Discussion 86
4.1 Signatures of biomass burning in aerosols and delivery of BC to the
the surface ocean
86
4.2 Atmospheric deposition of trace metals into the Santa Barbara Basin 89
4.3 Impacts of BC and metals from pyrogenic aerosols on organisms in
the basin
91
4.4 Increased loading of dissolved metals in river water after the Thomas
Fire
94
4.5 Hydrologic control on fluvial mobilization of trace metals, DBC, and
DOC
95
4.6 Fluvial mobilization of dissolved trace metals and DBC into the Santa
Barbara Basin
98
5. Conclusion 101
Tables 104
Figures 106
Supplementary tables and figures 115

Chapter 4 – The effect of host iron limitation on viral infection dynamics between
Vibrio and vibriophage 121
Abstract 121
1. Introduction 122
2. Materials and Methods 126
2.1 Iron one-step phage growth experiments 126
2.1.1 Host 126
2.1.2 Phage 127
2.1.3 Plating 128
2.1.4 Quantifying concentrations of Vibrio and vibriophage in stock
solutions
129
2.1.5 Host cell growth curves 130
2.1.6 One-step vibriophage growth experiment 130
2.2 Cell size measurements 132
2.3 Temperature one-step phage growth experiment 133
3. Results 133
3.1 Iron one-step phage growth experiments 133
3.2 Cell size measurements 134
3.3 Temperature one-step phage growth experiment 135
4. Discussion 136
4.1 Effect of Fe limitation on viral infection dynamics 136
4.2 Latent period in relation to Fe-limited host growth 137
4.3 Effects on microbial diversity 138
5. Conclusion 141
Tables 143
viii
Figures 144

Chapter 5 – Conclusions 147
1. Contributions to metal distributions through simple pole sampling 147
2. Metal transport through increasing wildfires 148
3. Utilization of Fe in viral infection 150
4. Thesis summary 152

References 153

Appendix – Iron recycling through the microbial loop in the North Pacific Subtropical
Gyre 189
Abstract 189
1. Introduction 190
2. Methods 194
2.1 Labeling the iron complexes 194
2.1.1 ALOHA media 194
2.1.2 Prochlorococcus and grazer cultures 194
2.1.3 Labeling viral lysates 195
2.1.4 Labeling grazing byproducts 196
2.1.5 Labeling siderophores 196
2.1.6 Measuring
54
Fe,
57
Fe, and
58
Fe in stocks 197
2.2 Fe addition incubations 197
2.2.1 Incubations in the field 197
2.2.2 Sample digestions 199
3. Results and Discussion 199
4. Conclusion 204
Tables 205
Figures 208

ix
List of Tables Page

Chapter 2 – High resolution surface metal sampling of the North Atlantic, North
Pacific, and South Pacific during the Tara Pacific expedition
Table 2-1 R
2
and RMSE from GT-A regression tree model outputs 41
Table 2-2 R
2
and RMSE from GT-A and Tara regression tree
models 42
Table 2-3 List of Longhurst biomes and provinces sampled 43
Table 2-4 Average metal concentrations from each province 44

Chapter 3 – Delivery of metals and dissolved black carbon to the southern California
coastal ocean via aerosols and floodwaters following the 2017 Thomas Fire
Table 3-1 Samples taken and analyses done during both sampling
campaign of the Thomas Fire in the Santa Barbara Basin and
Ventura River 104
Table 3-2 Dissolved metal concentration data for 8 surface seawater
samples collected in the Santa Barbara Basin during the 2017
Thomas Fire 105

Chapter 4 – The effect of host iron limitation on viral infection dynamics between
Vibrio and vibriophage
Table 4-1 Cell size measurements of low Fe and high Fe Vibrio cells 143

Appendix – Iron recycling through the microbial loop in the North Pacific
Subtropical Gyre
Table A-1 Detailed information on Fe-labeled viral lysate and grazing
byproduct stocks 205
Table A-2 Results from Fe addition Experiment 1, 25 m incubations 206
Table A-3 Results from Fe addition Experiment 2, 200 m incubations 207

x
List of Figures Page

Chapter 2 – High resolution surface metal sampling of the North Atlantic, North
Pacific, and South Pacific during the Tara Pacific expedition
Figure 2-1 Map of sampling locations during the Tara Pacific expedition 45
Figure 2-2 Sea surface Fe concentrations 46
Figure 2-3 Predicted vs true metal concentrations 47
Figure 2-4 Estimates of predictor importance values 48
Figure 2-5 GT-A’s Cd optimizable regression tree 49
Figure 2-6 Metal concentrations for the GEOTRACES and Tara
samples in each ocean biome 50
Figure 2-S1a GT-A’s Cu optimizable regression tree 52
Figure 2-S1b GT-A’s Fe optimizable regression tree 52
Figure 2-S1c GT-A’s Mn optimizable regression tree 53
Figure 2-S1d GT-A’s Ni optimizable regression tree 53
Figure 2-S1e GT-A’s Pb optimizable regression tree 54
Figure 2-S1f GT-A’s Zn optimizable regression tree 54
Figure 2-S2a Sea surface Cu concentrations 56
Figure 2-S2b Sea surface Mn concentrations 56
Figure 2-S2c Sea surface Ni concentrations 57
Figure 2-S2d Sea surface Pb concentrations 57
Figure 2-S2e Sea surface Zn concentrations 58
Figure 2-S2f Sea surface Co concentrations 58
Figure 2-S2g Sea surface Cd concentrations 59
Figure 2-S3a Sea surface Salinity concentrations 60
Figure 2-S3b Sea surface PO4 concentrations 60
Figure 2-S3c Sea surface temperature 61
Figure 2-S3d Sea surface SiO4 concentrations 61
Figure 2-S3e Sea surface O2 concentrations 62
Figure 2-S3f Sea surface NO3 concentrations 62
Figure 2-S3g Sea surface NO2 concentrations 63
Figure 2-S3h Aerosol concentrations 63
Figure 2-S4a Sea surface Net primary productivity sea surface 64
concentrations
Figure 2-S4b Sea surface Chlorophyll a concentrations 64
Figure 2-S5a Sea surface temperature off the northwestern coast of the
United States 65
Figure 2-S5b Sea surface temperature off the northwestern coast of the
South America 65
Figure 2-S6 Total dust deposition, dry and wet, in the North Atlantic
Ocean 66

xi
Chapter 3 – Delivery of metals and dissolved black carbon to the southern California
coastal ocean via aerosols and floodwaters following the 2017 Thomas Fire
Figure 3-1 Map of smoke trajectory, sampling locations, satellite image,
and photo taken during the cruise 106
Figure 3-2 Plots of PM2.5, BCPM2.5, levoglucosan, DBC, BCTSP, δ
13
POM,
and chlorophyll a 107
Figure 3-3 Solubilities of trace elements measured in aerosols 109
Figure 3-4 Trace element load associated with atmospheric aerosols 110
Figure 3-5 Photo of river samples, stage height plot, and DOC and DBC
vs time 111
Figure 3-6 Total dissolved metal concentrations in river samples 112
Figure 3-7 DBC and DBC flux of past studies and present study vs
Discharge 113
Figure 3-8 Estimates of the total mass and total soluble mass of metals
released as aerosols and in river flux 114
Figure 3-S1 Total aerosol metal concentrations as a fraction of the aerosol
Mass 115
Figure 3-S2 Surfaces seawater metal, nutrient, and organic concentrations 116
Figure 3-S3 River metal concentrations from control vs burned river basins 118
Figure 3-S4 Concentration-discharge relationships of metals and organics
in river water samples 119
Figure 3-S5 Average colloidal fraction of dissolved metals in river water
samples, and concentration-discharge relationships of metals
vs the average colloidal fraction of dissolved river water metal
concentrations 120

Chapter 4 – The effect of host iron limitation on viral infection dynamics between
Vibrio and vibriophage
Figure 4-1 Iron one-step phage growth curves for V40 144
Figure 4-2 Iron one-step phage growth curves for V36 145
Figure 4-3 Temperature one-step phage growth curves for V40 146

Appendix – Iron recycling through the microbial loop in the North Pacific
Subtropical Gyre
Figure A-1 Schematic of labeling the Fe ligands 208
Figure A-2 Schematic of Fe addition incubations 209
Figure A-3 Data from Fe addition Experiment 1, 25 m incubation 210
Figure A-4 Data from Fe addition Experiment 2, 200 m incubation 211

xii
Thesis Abstract
Since the first accurate marine metal measurements in the 1970s, trace metals in
the ocean have been identified as critical components in major biogeochemical cycles,
because most metals serve as essential nutrients for phytoplankton. Even though the field
of trace metal marine chemistry has advanced considerably over the last 50 years, simple
sampling methods are needed to make metal analyses more routine so that the effects of
metals on the marine ecosystem can continue to be explored. With frequent metal
sampling and a better understanding of the various roles metals play in marine biology we
can better predict how life in the global ocean will look like under the future climate. The
work presented in this thesis first investigates if simple pole sampling can produce reliable
metal data that captures the nuances in surface metal concentrations throughout the
world’s oceans. Subsequent chapters then look at the transport of metals by wildfires,
which are expected to intensify worldwide due to climate change, and observe how
changes in metal concentrations, particularly Fe, can affect both bottom-up and top-down
controls on the marine ecosystem.
Using a simple pole sampler, 242 surface metal samples collected from the North
Atlantic, North Pacific, and South Pacific Oceans produced pristine trace metal
measurements, that were comparable to previous measurements made by the
GEOTRACES program. Machine learning algorithms and environmental data were then
used to identify the likely sources of metals to the sampling regions, such as upwelling in
the South Pacific and anthropogenic aerosol deposition in the North Pacific. In another
study, a wildfire was directly surveyed through aerosol, floodwater, and seawater samples.
xiii
Aerosol and floodwater samples collected during the wildfire and subsequent flash flood
event indicated that metals, along with black carbon, were mobilized by the fire, but
seawater samples showed that fire-associated metals did not greatly affect the overall
coastal metal inventory. The fire also did not appear to immediately impact organic matter
cycling in the coastal ocean. However it is possible that the marine microbial communities
could be impacted as fires transport metals to the open ocean. Finally, I show how
changes in Fe concentration in laboratory cultures have varying effects on the viral infection
of marine heterotrophs, specifically Vibrio bacteria. While higher Fe was found to increase
the growth rates of both strains of Vibrio bacteria, higher Fe only increased the viral
progeny of one infected strain of Vibrio and not the other. If Fe also has the same impacts
on viral dynamics in natural ocean communities, an increase in viral progeny due to
increases in Fe could impact diversity and nutrient cycling within microbial communities.

1
Chapter 1
Introduction

1. A history of trace metal marine chemistry
Research on the trace metal chemistry in marine environments started to gain
prevalence in the 1970s and 1980s. Beginning in the early 1970s, advances in sampling
and analytical techniques made it possible for chemists to accurately measure nanomolar
(nM) to picomolar (pM) metal concentrations in seawater (Bruland et al., 2013; Crompton,
2006). With the advent of reliable trace metal measurements, biologists in the 1980s
began to realize the importance of these transition metals as micronutrients for
phytoplankton (Brand et al., 1983; Martin & Fitzwater, 1988; Sunda, 1989). For example,
incubations in the subarctic region of the northeast Pacific Ocean found that additions of
Fe at nM concentrations resulted in the complete drawdown of excess NO3 and as a result
phytoplankton-produced chlorophyll increased proportionally to the amount of Fe added
(Martin & Fitzwater, 1988). Once the subject was recognized as an important area of
marine research, scientists began dedicating their careers to characterizing the metal
distributions throughout the world’s ocean and studying the effects of metals on marine
phytoplankton growth.
Metals found in seawater originate primarily from the interaction between fresh
water and igneous rocks (Goldberg, 1958; Whitfield & Turner, 1979). The production of
seawater, which includes the incorporation of metals, can be simply presented as:

2
igneous rocks + acid volatiles + water à seawater + sediment + atmosphere
(Whitfield & Turner, 1979)
While conceptually useful, this equation is oversimplified and the dissolution of metals into
seawater can be better described as:
“The quantity of the different elements in seawater is not proportional to the quantity
of the different elements which river water pours into the sea, but inversely
proportional to the facility with which elements in seawater are made insoluble by
general chemical or organochemical reactions in the sea” (Forchhammer, 1865)

This statement not only applies to the input of metals through fluvial deposition, but to all
inputs of metals to the world’s oceans including aerosol deposition, hydrothermal venting,
volcanic activity, and input from reductive coastal sediments (Boyd & Ellwood, 2010;
Bruland et al., 2013; Charette et al., 2016; Jickells et al., 2005; Resing et al., 2015).
Common metals like Fe, Cu, Mn, and Co, occur in different oxidation states, which affect
the metal’s solubility, binding strengths with ligands, ligand exchange kinetics, and
biological uptake (Sunda, 2012). The numerous inputs of metals paired with the various
factors impacting metal solubility in seawater creates a dynamic global metal distribution.
Once transported into the ocean, dissolved metals take on an important biological
role in major phytoplankton processes (Morel et al., 2003; Twining & Baines, 2013). For
instance, metals such as Fe, Mn, and Cu are needed for photosynthesis (Fe – ferredoxin,
Schoffman et al., 2016; Mn – O2 evolving enzyme, Sunda, 1989; Cu – plastocyanin, Peers
& Price, 2006) and are also used as co-factors in superoxide dismutase along with Zn and
Ni (Chadd et al., 1996; Twining & Baines, 2013). Several of these trace metals have been
found to limit the growth of marine phytoplankton due to a mismatch between their
3
biological demand and the relatively low nM to pM concentrations available for uptake
(Beardall et al., 2001; Moore et al., 2013).
In the past 50 years, we have learned a tremendous amount about the transport,
distribution, and biological utilization of trace metals in the world ocean. Trace metal
chemists are now able to observe depth profiles of metals down to thousands of meters
(Aparicio-González et al., 2012) and measure the isotopes of these metals (Conway et al.,
2021; Sohrin & Bruland, 2011). Additionally, isotopic (Hurst & Bruland, 2007; John et al.,
2007) and genetic (Hogle et al., 2016; Manck et al., 2020) analyses of incubations have
produced a more comprehensive picture of the biological uptake and utilization of various
trace metals. The research presented in this thesis builds upon these advancements in the
trace metal field by contributing data on the global surface ocean metal distribution,
examining the marine impacts of metals emitted from wildfires, and observing how marine
viruses contribute to the recycling of metals.

2. Global patterns of surface ocean metals
Since the advancements in trace metal sampling and analyses in the 1970s, several
research programs (e.g. GEOTRACES, Conway et al., 2021; Sohrin & Bruland, 2011) and
stand-alone expeditions (e.g. Pinedo-González et al., 2020; Pinedo-González et al., 2014;
Vink et al., 2000) have conducted high-resolution sampling in an effort to characterize the
global ocean metal distribution. Survey studies looking at marine metals have found that
metal concentrations tend to increase in coastal regions and decrease further into the
oligotrophic open ocean (Charette et al., 2016; Pinedo-González et al., 2014). Exceptions
4
to this generalized spatial trend in metal distributions include increases in surface metal
concentrations in the middle of the North Atlantic and North Pacific Oceans due to
Saharan dust (Conway & John, 2014; Shelley et al., 2015) and Asian anthropogenic
aerosol input (Gallon et al., 2011; Pinedo-González et al., 2020), respectively. Metal-
focused studies have also observed temporal changes in surface metal concentrations
due to the effects of seasonal wind patterns on upwelling (Biller & Bruland, 2013; Bruland
et al., 2004) and dry versus wet seasons on dust deposition (Hsu et al., 2010; Powell et
al., 2015; Winton et al., 2016). Increased surveillance of the global ocean is important as
changes in surface metal concentrations directly affect marine primary productivity, which
produces up to 80% of the world’s O2 (Pomeroy et al., 2007; Simon et al., 2009) and
draws down about one third of the anthropogenic CO2 in the atmosphere (Basu & Mackey,
2018; Häder et al., 2014). In order to increase the range and frequency of trace metal
sampling, simpler methods need to be tested and implemented.

3. Wildfires as a potentially important marine metal source
Currently, the major sources of trace metals to the ocean have been identified as
aerosols (Jickells, 1995; Mahowald et al., 2018), hydrothermal venting (Gartman & Findlay,
2020; Sander & Koschinsky, 2011), and terrigenous input (Moore, 2010; Tovar-Sanchez et
al., 2006). Many studies have tried to quantify these metal fluxes (e.g. German & Angel,
1995; Samanta & Dalai, 2018) and observe the subsequent effects of metal deposition on
marine ecosystems (e.g. Paytan et al., 2009; Wilson et al., 2019). Recent reviews (e.g.
Hutchins & Boyd, 2016; Jickells et al., 2005) have highlighted that the fluxes and
5
bioavailability of metals from various sources will most likely change with the rise of
atmospheric CO2 and fluctuations in Earth’s climate. For example, studies have shown that
aerosols from wildfires currently emit twice as much Fe in comparison to industrial
combustion, but projections show that this difference in Fe emissions could increase up to
an order of magnitude due to the higher frequency of wildfires and lower industrial
emissions in the future (Mahowald et al., 2018, and references therein). Based off
projections by Hoffmann et al. (2012) and Millero et al. (2009), the current solubilities of Fe
produced by wildfires, 0.5 to 46%, may also increase as the oceans and atmosphere
become more acidic (Desboeufs et al., 2001; Mahowald et al., 2009). Continual research
on the sources of metals to the marine environment is important for understanding how
the distribution and bioavailability of these metals might look like in the future. Metal
sources that may seem insignificant to today’s oceans, like wildfires, could be important to
future metal inventories and therefore should be included in these source studies.

4. Iron cycling through viral infection
Studies have found that several of the trace metals, such as Fe, Zn, Mn, Ni, Cd, Co,
and Cu, are essential for phytoplankton growth, due to their roles as either active centers
or structural factors in crucial enzymes (Sunda, 2012; Twining & Baines, 2013). Much of
the research focusing on metals in marine biology has focused on Fe due to the balance
between its high biological demand and its low, growth-limiting concentrations (Boyd &
Ellwood, 2010; Price & Morel, 1998). In fact, estimates show that Fe limits at least 30% of
the primary productivity in the surface ocean (Boyd et al., 2007; Moore et al., 2013) and
6
much of the dissolved Fe in seawater is bound to organic ligands (Gledhill & van den Berg,
1994; Rue & Bruland, 1995). Some microbes produce siderophores in order to access
these scarce amounts of ligand-bound Fe (Hopkinson & Morel, 2009; Manck et al., 2022).
In a collective effort to satiate the biological iron demand, marine microbes have also been
found to recycle Fe amongst themselves (Rafter et al., 2017; Strzepek et al., 2005). Some
of these recycling mechanisms includes the biological uptake of Fe bound to cell debris,
which can be products of grazing or viral lysis (Hutchins & Bruland, 1994; Hutchins et al.,
1993; Poorvin et al., 2004; Wilhelm & Suttle, 1999). Many studies have looked at how
grazing and viral lysis contribute to the Fe cycle, but only a few studies have looked at how
Fe affects grazing and viral lysis (e.g. Bonnain et al., 2016; Kranzler et al., 2021, 2019).
This area of research is important as the effects of metals on top-down controls is an
integral part of how these metals are recycled through the microbial loop.

5. The work presented in this thesis
Chapter 2 investigates the biogeochemical factors shaping the surface metal
distributions in the North Atlantic, North Pacific, and South Pacific Oceans. In Chapter 2,
the total dissolvable Fe, Zn, Mn, Ni, Cd, Co, Cu, and Pb concentrations of 242 surface
seawater samples collected on the Tara Pacific expedition are presented. Using a machine
learning algorithm, I examined if the simple pole sampling method used in this study was
able to produce pristine trace metal samples that captured the nuances in the global
surface metal distribution. Regression tree models, past metal data, and paired
environmental data were then used to determine which biogeochemical drivers (e.g.
7
biological uptake, upwelled waters, dust deposition) had the greatest impact on surface
metal concentrations.
Chapter 3 examines the transportation of metals to the surface ocean during two
extreme weather events, the 2017 Thomas Fire and subsequent flash flood event, which
impacted the coastal ocean near Santa Barbara, California. Aerosol, river water, and
seawater samples were collected to measure the paired fluxes of organics and trace
metals associated with fires and subsequent flooding events to the coastal ocean. This
study was one of the first of its kind and is important as wildfires are projected to increase
with warming in the western United States (Abatzoglou & Williams, 2016). If the frequency
of fires does continue to increase in the future, then the metals associated with biomass
burning may become an important source of marine metals.
Chapters 3 and 4, along with the Appendix, explore the biological utilization of
surface metals by the marine microbial community. In Chapter 3, samples for δ
13
C,
chlorophyll a, and dissolved organic carbon analyses were collected in the coastal waters
underneath the smoke plume produced by the Thomas Fire. These coastal water samples
informed us of the general impact the wildfire-associated metals had on the organic matter
cycling in the Santa Barbara Basin. Chapter 4 delves deeper into the interactions between
metals and the microbial community, more specifically, it looks at the effects of Fe on the
viral infection of heterotrophic marine bacteria. The lytic cycles of two strains of Vibrio and
vibriophage were observed under varying Fe conditions to see if the changes in Fe
concentration affected the viral dynamics of the phage. These studies are some of the first
to observe how metal bioavailability affects the top-down controls of marine
8
phytoplankton. Finally, the Appendix presents experimental incubations that used three
stable Fe isotope tracers (
54
Fe,
57
Fe,
58
Fe) to quantify the relative bioavailability of three
sources of recycled Fe (viral lysates, grazing byproducts, siderophores) to the natural
microbial community in the North Pacific Subtropical Gyre. While the incubations were
deemed unsuccessful, we present useful information and recommendations for similar
future experiments.

9
Chapter 2
High resolution surface metal sampling of the North Atlantic, North
Pacific, and South Pacific during the Tara Pacific expedition

Author Contributions
Seth John designed the metal study for the Tara Pacific expedition and provided input on
the manuscript. Fabien Lombard and Guillaume Bourdin directed the field work and
collected samples at sea. Rachel Kelly led the metal analysis, data processing, and
manuscript writing. Natalie Cohen, Pauline Pinedo-González, and Nicholas Hawco helped
with metal analysis and data processing.

Abstract
Many trace metals are needed as either active centers or structural factors in
marine microbial enzymes, trace metals dissolved in surface ocean seawater are therefore
vital to the marine ecosystem. During the Tara Pacific expedition, samples for trace metals,
aerosols, and other environmental and biological analyses were collected in parallel across
surface waters of the North Atlantic, North Pacific, and South Pacific Oceans with a simple
hand-held pole sampler. Here we present total dissolvable trace metal (Fe, Zn, Mn, Ni, Cd,
Co, Cu, and Pb) concentrations of 242 surface water samples from the Tara Pacific
expedition. A machine learning framework is used to compare these data to past
GEOTRACES data in order to demonstrate that pole sampling was successful in collecting
pristine trace metal samples. We also assess whether machine learning algorithms have
sufficient skill to eventually replace physical sampling and analysis, finding that machine
learning can accurately reproduce broad-scale patterns in metal distribution but can miss
local-scale variability. Decision tree algorithms are further used to determine which
10
biological and physical factors are most predictive of trace metal concentrations. Trace
metal distributions across the coastal and open ocean gradients were influenced by
various biological, geochemical, and physical processes. For example, high concentrations
of iron (Fe) and manganese (Mn) were observed in several regions, including in the North
Atlantic Ocean and near the South Pacific islands, possibly due to Saharan dust and
hydrothermal vent input, respectively. Elevated lead (Pb) was found in the North Pacific
near southeast Asia, where anthropogenic sources may contribute. We also observe inter-
basin differences in concentrations for most of the metals, such as cobalt (Co), which is
relatively high in the North Atlantic in comparison to the Pacific, perhaps due to dust
deposition or continental weathering. There are also intrabasin differences in metal
concentrations between oligotrophic and upwelling regions, exemplified by the higher
cadmium (Cd) concentrations near the Peruvian coast, likely due to upwelling. This study
displays how a simple sampling method, like pole sampling, can capture high resolution
trace metal data that depicts the nuances in the metal distribution of the global ocean.

1. Introduction
Long after the time when concentrations of macronutrients such as N, P, and Si
were measured throughout the world ocean, trace metal concentrations in seawater were
barely understood due to the difficulties of avoiding contamination when sampling, and the
absence of analytical techniques to measure the nanomolar (nM) to picomolar (pM) range
concentrations of these elements (Morel et al., 2013). In the early 1970s advances in highly
sensitive analytical techniques made it possible for chemists to accurately measure these
11
minute metal concentrations (Bruland & Lohan, 2013; Crompton, 2006). With new
instrumentation that could detect trace metal concentrations, it was found that
phytoplankton were dependent on trace metals including iron (Fe), zinc (Zn), manganese
(Mn), nickel (Ni), cadmium (Cd), cobalt (Co), and copper (Cu), despite their extremely low
concentrations in seawater (Brand et al., 1983, 1986; Martin & Fitzwater, 1988). Microbes
depend on these trace elements due to their roles as either active centers (e.g. Mn –
oxygen evolving complex of PSII; Fe – ferredoxin) or structural factors (e.g. Ni, Cu –
superoxide dismutase; Co, Zn, Cd – carbonic anhydrase) in crucial enzymes (Morel et al.,
2013, Twining & Baines, 2013). Once oceanographers understood the importance of trace
elements to biological productivity, they prioritized engineering devices to collect trace
metal clean seawater samples, as well as developing new analytical methods to analyze
seawater metal concentrations.
Traditional seawater sampling devices, such as metal-containing Niskin bottles, and
the metal research vessels from which they are deployed can both lead to extreme levels
of metal contamination (Bruland & Lohan, 2013; Spencer et al., 1982). Thus, trace metal
chemists over the years have designed specialized, minimal-metal sampling devices, along
with deployment methods which ensure that the water being sampled has not come in
contact with the large metal research vessels (Cutter & Bruland, 2012; Measures et al.,
2008). Minimal-metal sampling devices used to collect surface seawater can be as simple
as a hand-held pole sampler (Boyle et al., 2005), or as complex as a purpose-built metal
free Niskin sampling system (Cutter & Bruland, 2012) or an underway tow-fish
continuously pumping water onboard (Hawco et al., 2020). Due to the difficulties of
12
collecting trace metal-clean samples, trace metal analyses are not typically included
among the suite of biogeochemical parameters analyzed on oceanographic cruises.
Instead, a special effort must be made to deploy trace metal clean sampling systems, as
has been done on a handful of programs including the Climate Variability and Predictability
(CLIVAR) program, the annual Atlantic Meridional Transect (AMT), the GEOTRACES
program, and now the Tara Oceans program discussed here.
From May 2016 to October 2018 the Tara Ocean Foundation’s Pacific expedition
(Tara Pacific) completed an approximately 110,000 km route in both the Atlantic and
Pacific oceans. The three main objectives of the Tara Pacific expedition were to (1) study
Pacific reef ecosystems, (2) observe community structure in surface waters both near the
Pacific Islands and in the open ocean, and (3) to examine the ocean and atmosphere
interface (Gorsky et al., 2019). Both high-resolution underway sampling and daily discrete
sampling occurred aboard the 110-foot research schooner Tara throughout the
expedition. Underway sea surface measurements were taken to monitor chlorophyll, sea
surface salinity, temperature, and net community production. Discrete sea surface
samples were taken for a multitude of chemical (e.g. carbonates, micro- and
macronutrient) and biological (e.g. phytoplankton pigment, taxonomic and genomic)
analyses. Continuous and discrete aerosol particulate sampling also occurred throughout
the expedition (Flores et al., 2020). The thousands of samples and measurements
collected were analyzed and processed by an international scientific consortium, as
described by Gorsky et al. (2019).
13
A key novelty that set the Tara Pacific Expedition apart from the ten previous Tara
Ocean Foundation’s expeditions was the inclusion of trace metal sampling. Trace metal
analyses were performed with the goals of characterizing the surface seawater trace metal
distribution across the open ocean and coastal regions in both the Atlantic and Pacific,
and exploring metal-dependent ecosystem structure and metabolism (Gorsky et al., 2019).
At 36 m length x 10 m beam, the Tara schooner was able to collect high-resolution trace
metal-clean samples using a simple all-plastic pole sampler deployed by hand. 242 trace
metal surface seawater samples were collected during the Tara Pacific expedition and
analyzed for total dissolvable (i.e. unfiltered whole seawater) Fe, Zn, Mn, Ni, Cd, Co, Cu,
and lead (Pb) concentrations.
In this study, we characterize the surface trace metal distribution across the Atlantic
and Pacific oceans, and use coupled environmental and biological data to examine the
biogeochemical processes controlling surface metal concentrations. A decision tree
regression algorithm was used to predict metal concentrations from both the
GEOTRACES and Tara datasets based on various physical parameters and macronutrient
concentrations. These analyses were used first to determine whether or not pole sampling
aboard the Tara schooner successfully collected pristine trace metal seawater samples.
We then studied whether or not the decision tree algorithms had sufficient skill to replace
collection of metal samples. Finally, the decision trees were explored to determine which
parameters were most predictive of the concentrations of the various trace metals. The
decision tree machine learning approach was followed by a more qualitative analyses of
metal concentrations in various biogeochemical provinces. Surface measurements from
14
the GEOTRACES 2021 Intermediate Data Product and Tara Pacific datasets were divided
into 21 biogeographical marine provinces, which make up 3 of the 4 major ocean biomes
(coastal, westerly, trade, and polar biomes) designated by Longhurst (1998). The potential
biogeochemical factors influencing the metal distributions observed on the Tara Pacific
expedition were then identified by comparing Tara data to the GEOTRACES data in each
biome, as well as other surface ocean metal concentrations reported in the literature.
The trace metal concentrations presented here will complement data already in the
Tara Oceans database, while filling in some sampling gaps of other global trace metal-
focused expeditions. Once paired with additional biological data collected at the same
stations during the Tara Pacific expedition, including transcriptomic, taxonomic, and
genomic data, we will further explore how these metals are regulating plankton community
composition, metabolic functionality, and ecological interactions in both the Atlantic and
Pacific Oceans (Gorsky et al., 2019).

2. Methods
2.1 Sampling
Two hundred and forty two seawater samples were collected at about a 0.3 m
depth in the North Atlantic and North and South Pacific Oceans during the Tara Pacific
expedition (May 28, 2016, to October 27, 2018; Figure 2-1). During the expedition special
attention was paid to the coral islands in the equatorial Pacific, the western Pacific, and the
north-eastern Pacific where the “Great Pacific Garbage Patch” is located (centered on
35°N, 140°W; Gorsky et al., 2019). Trace metal-clean surface seawater samples were
15
collected using an extendable carbon-fiber pole with a polyvinyl chloride (PVC) bottle
holder attached to the end of it. The pole was held off the bow of the ship with the pole
oriented toward the wind while sailing at up to 9 knots in order to minimize metal
contamination from the ship. During sample collection an uncapped 125 mL low-density
polyethylene (LDPE) bottle was attached to the pole and was rinsed 4 to 5 times with
surface seawater before being filled, recapped, and double-bagged in zip-lock bags.
Samples were then stored at room temperature in the dark for the remainder of the
voyage. All LDPE bottles were acid-washed prior to sample collection and trace metal-
clean practices were used during sample collection, for example, handling samples with
polyethylene gloves.
2.2 Trace metal analyses
Trace metal-clean surface seawater samples were analyzed in a class 100 clean
room at the University of Southern California (USC). Upon arrival to USC, samples were
acidified with 0.1% 12 N HCl and stored at room temperature in the dark for 4 to 6 months
in order to redissolve the total dissolvable metals (Sedwick et al., 2011). This method
dissolves the “acid labile” particulate metals and metals which may have precipitated
during storage but does not dissolve metals from refractory particles (Sedwick et al.,
2011).
After storage under acidic conditions, 15 mL subsamples were collected and 50 μL
of an isotope spike (which includes:
57
Fe,
62
Ni,
65
Cu,
67
Zn,
207
Pb,
110
Cd) was added to each
of the 15 mL tubes. Spiked samples were then placed into the seaFAST (manufactured by
Elemental Scientific; Lagerström et al., 2013) in order to concentrate the samples and
16
purify them from major seawater salts. The seaFAST instrument pushed each of the
samples through a column with Nobias resin PA-1 (Sohrin et al., 2008) at pH ~6 with an
ammonium acetate, acetic acid buffer, and the preconcentrated sample was eluted into
0.5 mL 1 M HNO3 which included 1 ppb In. Metal concentrations in the preconcentrated
samples were then measured on an Element 2 inductively coupled plasma mass
spectrometer (ICP-MS, Thermo Fisher Scientific). Concentrations for all elements except
Mn and Co were derived by using an isotope dilution method (Lee et al., 2011). Mn and
Co concentrations were quantified relative to a 10 ppb multielement standard. Metal
concentration analysis techniques were identical to those used in Hawco et al. (2020). Co
concentrations do not include the fraction bound by strong organic ligands because
samples were not UV oxidized and therefore are assumed to be “labile Co”.
2.3 Data quality
Due to the abundance of samples and the staggered sample delivery, samples
were analyzed at different times over a span of several months. Triplicate analyses were
performed on most samples, therefore average concentrations are reported here. A
GEOTRACES standard (coastal surface seawater standard) was included in each sample
run as a control. If the concentrations of the GEOTRACES standard were outside the
community consensus values, the sample run was rejected. All of the values reported here
are from sample runs were the GEOTRACES standard value was within community
consensus range.
A final data quality check was done where finalized trace metal concentrations were
first grouped into three regions: (1) Pacific Open Ocean Stations, (2) North Atlantic and
17
Pacific Coastal Stations, (3) North Atlantic Open Ocean Stations. The collated
concentrations for each metal in each region were then compared to corresponding
GEOTRACES standards: (1) Pacific surface seawater standard, (2) coastal surface
seawater standard, (3) Atlantic surface seawater standard, respectively. The Tara Pacific
expedition trace metal concentrations agreed well with each of the corresponding
GEOTRACES standards. Visuals of this final data quality check can be found in the
Supplemental Information of Gorsky et al. (2019).
2.4 Data management
Each sample was assigned a unique code in order to link the trace metal
concentrations with corresponding metadata. All environmental data can be found in the
free, open access database PANGAEA.
2.5 Machine learning techniques
2.5.1 Subsetting GEOTRACES data
Trace metal (dissolved Fe, Zn, Mn, Ni, Cd, Cu, and Pb) and environmental
(longitude, latitude, depth, temperature, salinity, dissolved O2, dissolved PO4, dissolved
SiO4) data from the GEOTRACES Intermediate Data Product 2021 was extracted from the
GEOTRACES webODV platform (GEOTRACES Intermediate Data Product Group, 2021).
Co concentrations were excluded as they were scarce amongst the GEOTRACES dataset.
From the full dataset, a subset of surface samples were chosen for depths shallower than
25 m, and for which there was reported data for each of the 16 metal and environmental
variables used in our regression tree analysis. The final dataset included samples from the
transects GA10, GI04, Glpr05, GP03, GP13, GP16, GP18, GP19, and GS01. In order to
18
test the skill of the regression tree algorithm, the data was further split into dataset A (“GT-
A”), which includes the surface data from the GA10, Glpr05, GP16, GP18, and GP19
cruises, and B (“GT-B”) which includes the surface data from the GI04, GP03, GP13, and
GS01 cruises. The GEOTRACES data was split so that each major ocean basin, the
Atlantic, the Pacific, and the Indian Ocean, was represented in both dataset A and B.
2.5.2 Regression tree algorithm
Regression tree models were used to predict trace metal concentrations in the GT-
A (n = 234), GT-B (n = 129), and Tara (n = 242) datasets. MATLAB’s Regression Learner
App was used to train optimizable regression tree models to predict the true metal
concentrations (Nadler, 2019). Eight predictive variables were used for the optimizable
regression tree models, including (1) latitude, (2) longitude, (3) depth, (4) salinity, (5)
temperature, (6) O2, (7) PO4, and (8) SiO4. A regression tree was constructed for each of
the metals, except for GT-A Co, and trained using the GT-A and Tara Pacific datasets. The
minimum leaf value (the number of prediction values associated with a leaf at which a node
stops splitting) and total number of nodes (sum of branches and leaves) varied for each
model due to the optimization process. The minimal leaf values for the GT-A models are
Cd (11), Cu (10), Fe (2), Mn (1), Ni (3), Pb (3), Zn (18), and the number of nodes are Cd
(35), Cu (33), Fe (87), Mn (95), Ni (79), Pb (85), Zn (17). For the Tara Pacific models the
minimal leaf values are Cd (8), Cu (1), Fe (102), Mn (2), Ni (1), Pb (8), Zn (65), Co (4), and
the number of nodes are Cd (43), Cu (91), Fe (3), Mn (97), Ni (97), Pb (47), Zn (5), Co (85).
The error associated with regression tree predictions was derived from the R-
squared (R
2
) between observed and predicted values, and the root mean square error
19
(RMSE). Also, estimates of predictor importance were calculated for each model.
Estimates of predictor importance were calculated for each of the predictive variables in all
of the regression tree models by first calculating the node risks for each node in the
regression trees. For example, the node risk (Ri) of node i, which is the error of a node
weighted by the node probability, is calculated using the equation:

Ri = PiEi (1)

where Pi is the node probability of node i, and Ei is the mean squared error of node i. The
sum of the changes in the node risk due to splits per predictor is then divided by the total
number of branch nodes (Nbranch) to give you the estimate of predictor importance:

estimate of predictor importance =
Ri1- Ri2 - Ri3 -...- Rix
Nbranch
(2)

The greater the estimated value the more important that predictor variable (i.e.
environmental parameter) is at predicting the response values (i.e. metal concentrations).

3. Results and Discussion
3.1 Reliability of pole sampling
A machine learning algorithm was first used to confirm if pole sampling during the
Tara Pacific expedition collected pristine (uncontaminated) trace metal samples. The
predicted metal concentrations for the three datasets were compared to true metal
20
concentrations (Figure 2-3) in order to determine whether there were systematic offsets
between Tara samples and GEOTRACES samples, which could tentatively be attributed
either to the different size fractions measured (dissolved in GEOTRACES compared to
total dissolvable in Tara) or possible contamination due to pole sampling during the Tara
Pacific expedition.
Generally, the ranges in predicted and observed metal concentrations overlapped
for the Tara and GEOTRACES datasets, with the exception of Fe (Figure 2-3). Measured
Fe concentrations included in the GEOTRACES datasets had generally lower
concentrations (0.32 ± 1.20 nM) compared to those measured for Tara samples (3.90 ±
31.30 nM), so it is perhaps not surprising that the regression tree model, which was
trained on GT-A, is not able to explain predict Tara Fe concentrations with any skill (R
2
= 0)
and underestimates Tara’s Fe concentrations (RMSE = 31.40, Table 2-2). These
discrepancies between the true Fe concentrations in the Tara and GEOTRACES datasets
can be explained by both the differences in size fractions measured and sampling
locations. Tara Pacific samples represent the total dissolvable fraction (tdmetal; whole
seawater), which includes soluble (<0.02 µm), colloidal (0.02-0.2 µm), and particulate (>0.2
µm) metals that are solubilized after extended storage (>6 months) at pH ~2 (Sedwick et
al., 2005). The GEOTRACES samples represent the dissolved fraction (dmetal; <0.2 µm
filtered seawater), which includes soluble (<0.02 µm) and colloidal (0.02-0.2 µm) metals.
The differences in Fe concentrations due to size fractions observed between the Tara and
GEOTRACES data are unsurprising and have been reported previously (Boyd & Ellwood,
2010; Sedwick et al., 2005). Fe has an extremely low solubility in oxic environments (i.e.
21
the surface ocean) and is very particle reactive (Bowie et al., 2002; Tagliabue et al., 2017)
thus dFe surface concentrations are often much lower than tdFe. Measured Tara Fe
concentrations are also higher than those from the GEOTRACES datasets because the
Tara samples were collected very close to coastlines, sometimes as close as 3 km. In fact,
18 out of the 242 Tara Fe samples have concentrations that vary between 5 and 474 nM
tdFe, most of which represent coastal samples, whereas the rest of the 224 samples have
an average Fe concentration of 0.50 ± 0.71 nM tdFe (Figure 2-2). It is not surprising that
the model is unable to accurately predict the 18 highest Fe concentrations since the model
was trained on data in which the highest Fe concentration was 21.93 nM.
For the rest of the predicted metal data, Tara and GEOTRACES are within the same
concentration ranges and have similar RMSE values. Tara Cd, Ni, and Zn fall on the lower
end of their concentration ranges (Figure 2-3) which is reflected in the Tara’s low R
2
values
(Cd – 0.01, Ni – 0.18, Zn – 0) compared to both GT-A’s (Cd – 0.95, Ni – 0.96, Zn – 0.45)
and B_GT_2021’s (Cd – 0.76, Ni – 0.81, Zn – 0.35) R
2
values (Table 2-2). If there was any
metal contamination from the pole sampling, we would expect to see the Tara metal data
at the higher end of the concentration ranges, which we did not see. With the exception of
Fe, the similarities in RMSE values for each of the metals, especially between GT-B and
Tara datasets, which the models had never seen before, further prove that the metal
concentrations for both datasets were comparable.
Our results demonstrate that surface metal data collected by a pole sampler
operated by scientists without previous training in trace metal clean techniques is similar to
observations for surface samples collected on large research vessels with an extensive
22
trace metal sampling setup. The easy deployment and success of the pole sampling
during the Tara Pacific expedition demonstrates that trace metal sampling could become a
more routine measurement, extending the opportunities for collection of trace metal
samples on a much wider variety of research expeditions or perhaps even by the general
sea-going public.
3.2 The importance of metal sampling
The regression tree models were also used to explore whether surface metal
distributions could be predicted using other parameters which are easier to sample and
analyze such as location, salinity, temperature, oxygen, and macronutrients, and, if so,
might replace the need for trace metal sampling. Naturally, the R
2
values for the GT-A data
predictions were the highest on average amongst the three datasets that were tested
since the models were trained on that same dataset (Table 2-2). When the regression tree
models were asked to predict the GT-B metal concentrations R
2
values slightly decreased
for Cd, Cu, Ni, and Zn (8 to 19%) and drastically decreased for Fe, Mn, and Pb (56 to
88%) in comparison to the GT-A R
2
values. The Tara prediction R
2
values decreased from
26 to 94% for all of the metals in comparison to the GT-A R
2
values.
The decreases in R
2
values between the GT-A and GT-B predictions alone show
that the optimizable regression tree models are unable to accurately predict metal
concentrations using the seven predictor variables used in this study. In order to say that
machine learning algorithms could replace metal sampling we needed to show that the
regression tree models could at least predict GT-B metal values with decent accuracy
because of its inherent similarity to the GT-A training dataset. Discouragingly, the algorithm
23
was unable to explain any of the variation in an important metal like Fe in GT-B’s dataset,
for example (R
2
= 0; Table 2-2). Using a regression model trained on the GT-A dataset to
predict observations from the Tara dataset yielded even lower R
2
values (R
2
= 0 to 0.33)
due to its inability to accurately predict the smaller changes in metal concentrations found
amongst the Tara dataset, demonstrating further that surface trace metal sampling is
necessary to produce accurate metal concentrations. While the RMSE values may show
that the models are predicting the metal concentrations in the right range, the R
2
values
are proving that the models are unable to predict accurate concentrations. Thus, our
results suggest that collecting real trace metal samples is necessary in order to obtain
accurate data, a task which will hopefully be made easier now that we have demonstrated
the ease and reliability of pole sampling.
3.3 Patterns between environmental and metal data
Estimates of predictor importance were calculated for each of the predictor
variables (environmental parameters) used in the regression tree models for both the Tara
Pacific and GT-A metal data (Figure 2-4). These estimates of predictor importance tell us
how important each environmental parameter (latitude, longitude, salinity, temperature, O2,
PO4, SiO4) was in determining the predicted metal concentrations. For instance, PO4 is the
top predictor for both the GT-A and Tara Cd regression trees according to the estimates
of predictor importance. When looking at GT-A’s Cd regression tree (Figure 2-5) the top
three branches are PO4 values which clearly shows how important PO4 is in predicting Cd
concentrations. This is unsurprising as it has been largely accepted that Cd and PO4 have
very similar nutrient-like distributions in the oceans (Boyle et al., 2005; Bruland, 1980). One
24
particularly interesting branch in GT-A’s Cd regression tree is one which has a PO4 value of
0.88 µM (Figure 2-5). Previously published plots of global Cd against PO4 show a visible
“kink” in the otherwise linear relationship at about 0.8 µM PO4 or 70 pM Cd (Roshan &
DeVries, 2021). Roshan & DeVries (2021) found that this globally optimum kink is located
where a “low slope” regime with a Cd:PO4 ratio of 0.1 and a “high slope” regime with a
Cd:PO4 ratio of 0.4 meet. We see that in the GT-A’s Cd regression tree all of the leaves
(predicted dCd concentrations) to the left of the 0.88 µM PO4 branch, <0.88 µM PO4, have
dCd concentrations <70 pM, whereas to the right of the 0.88 µM PO4 branch, ≥0.88 µM
PO4, dCd concentrations are >70 pM (Figure 2-5). These consistencies in the relationship
between Cd and PO4 in the regression tree models shows that the machine learning
algorithms can reproduce previously observed patterns, and that there is valuable
information in the estimates of predictor importance.
3.3.1 Major nutrients
PO4 was identified as the most important predictor for GT-A’s Cd, Fe, and Zn
concentrations and SiO4 was a top predictor for Tara’s Fe, Mn, and Co concentrations
(Figure 2-4). All of these metals (Cd, Fe, Zn, Mn, and Co) associated with the two
macronutrients, PO4 and SiO4, are needed for major phytoplankton processes (Twining &
Baines, 2013) thus these connections could reflect the biological uptake of these metals
alongside PO4 and SiO4. However, when looking at the R
2
values for each of these
regression tree models we see that only GT-A’s Cd R
2
value is above a high correlation
threshold of 0.7 (Cd R
2
= 0.92, Table 2-2) therefore only the relationship between GT-A’s
Cd and PO4 can be considered more than coincidental. PO4 and SiO4 may be used to
25
predict Fe, Zn, Mn, and Co concentrations simply because both macronutrient and metal
concentrations tend to increase near the continental-marine interface. Verifying if biological
uptake is the reason why PO4 and SiO4 are used to predict these metal concentrations is
beyond the scope of this study but can be answered by future processing of metagenomic
data collected alongside the metal samples on the Tara Pacific expedition (Gorsky et al.,
2019).
3.3.2 Salinity
Salinity is the most important predictor for GT-A’s Cu and Mn models and Tara’s Ni
and Zn models (Figure 2-4). Past studies have found that high metal concentrations
alongside low salinities are a result of metal input from fresh water sources (Vance et al.,
2008; Wang & Liu, 2003; Bewers & Yeats, 1989). In GT-A Cu’s regression tree plot the
highest predicted Cu concentrations (1.18 to 2.40 nM dCu) are found at salinities <34.09
PSU (Figure 2-S1a), we also see this in GT-A Mn’s regression tree plot where the model
predicts the highest Mn concentration, 8.04 nM, at a salinity <31.62 PSU (Figure 2-S1c).
During the Tara Pacific expedition low salinities were observed off the coast of Oregon
(29.20 PSU, 160 km south of the mouth of the Columbia River), Costa Rica (29.99-30.57
PSU, 95 km southwest of the Pavon Bay), Panama (30.13 PSU, in the Gulf of Parita), and
Nova Scotia (31.63 PSU, 110 km from the coast)(Figure 2-S3a). Cu concentrations
increased at each of these locations ranging from 1.49 to 3.53 nM (Figure 2-S2a). High Mn
and Ni concentrations were only found near Oregon (9.66 nM Mn, 4.03 nM Ni) and Nova
Scotia (7.91 nM Mn, 4.69 nM Ni)(Figure 2-S2b&c), and Zn only increased near Costa Rica
26
(9.77 nM Zn) and Nova Scotia (2.49 nM Zn)(Figure 2-S2e). The high metal concentrations
found alongside salinities <32 PSU could be a result of a fresh water source of metals.
3.3.3 Location
Cu and Pb distributions in the surface ocean are known to be influenced by
atmospheric deposition (Maring & Duce, 1989; Jacquot & Moffett, 2015; Echegoyen et al.,
2014; Gallon et al., 2011) which is often times location specific, for example, Saharan dust
deposition is common in the North Atlantic (Jickells et al., 2005; Shelley et al., 2015) and
Asian anthropogenic aerosols are found in the North Pacific (Pinedo-González et al., 2020;
Rusiecka et al., 2018; Zurbrick et al., 2017). The second top predictor in both the GT-A
and Tara Cu models is longitude, and latitude was the top predictor for both GT-A’s and
Tara’s Pb regression tree models (Figure 2-4). In GT-A Cu’s regression tree plot, higher
predicted Cu concentrations (0.93 to 1.55 nM dCu) are found in the North Atlantic Ocean
between 105.26°W and 19.77°W (Figure 2-S1a), which again aligns well with the
observational tdCu data from the Tara Pacific expedition (Figure 2-S2a). High aerosol
concentrations were also found in the same region in the North Atlantic on the Tara Pacific
expedition (Flores et al., 2020). In GT-A’s Pb regression tree plot (Figure 2-S1e), the higher
predicted Pb concentrations are found north of 38.09°S and the highest Pb value (53.66
pM dPb) is found north of 20°N (Figure 2-S1e). Even though most of the Tara Pacific data
already lies north of 38.09°S we do see elevated tdPb concentrations north of 20°N
together with high aerosol concentrations (Figure 2-S2d). The estimates of importance
factors highlight the noteworthy increases in Cu, Pb in regions with high aerosol
concentrations.
27
3.4 Observational comparison to GEOTRACES
Following our quantitative comparison of GEOTRACES and Tara surface metal
concentrations using machine learning techniques, we have compared these data in a
more qualitative fashion by binning the data into various ocean biomes. We first
categorized the Tara Pacific data and the GEOTRACES data into three ocean biomes
(coastal, trade, and westerly) that were established by Longhurst (1998) using
climatological data (monthly mixed layer depth, solar irradiance penetration, and
chlorophyll concentrations). Much of the variance between the Tara Pacific and
GEOTRACES datasets can be explained by near shore sampling during the Tara
expedition and the differences in the size fractions measured, as previously discussed.
Consequently the average concentrations in Tara tdFe samples were much higher than
those found in the GEOTRACES dFe samples (Figure 2-6, Table 2-3). Metal contamination
was ruled out since both Tara Pacific and GEOTRACES Zn concentrations were very
similar and Zn is considered to be extremely contamination prone (Figure 2-S2e, Croot et
al., 2011; Cutter & Bruland, 2012; Spencer et al., 1982). Due to its biological importance,
Zn distributions in the oceans are nutrient-like with depleted concentrations at the surface
due to active biological uptake and scavenging onto particles (John & Conway, 2014). Low
Zn values persisted throughout much the route (all Tara Pacific average: 0.36 ± 0.80 nM)
with the exception of a few coastal samples (coastal biome average: 0.58 ± 0.89 nM,
Figure 2-S2e, Table 2-1). Other notable differences between the Tara and GEOTRACES
datasets were coastal and westerly Cd (Figure 2-6). Higher coastal GEOTRACES Cd was
sampled during a period of upwelling on the Peruvian coast (GP16; Hawco et al., 2016;
28
John et al., 2018; Moffett & German, 2018), and high westerly GEOTRACES Cd samples
are from the Southern Ocean (Glpr05 & GS01) where the upwelling of Cd enriched water
and diatom Cd uptake and regeneration occurs (Cloete et al., 2021). Overall, the predicted
and true metal concentrations from the Tara and GEOTRACES datasets are similar and
show that the pole sampling on the Tara Pacific expedition was successful.
3.5 Biogeochemical observations
3.5.1 Coastal biome
The highest average concentrations for all of the metals, except for Pb, during the
Tara Pacific expedition were found in the coastal biome (Figure 2-6, Table 2-3). Metal
concentrations are generally elevated near the continental-marine interface where multiple
external (e.g. fluvial deposition, aeolian dust; Martin & Whitfield, 1983) and internal (e.g.
remobilization of metals from continental shelf sediments and upwelled waters; Elrod et al.,
2004; Laès et al., 2007) sources of trace metals affect surface ocean chemistry. Moderate
levels of productivity (12.25 ± 15.37 g/m
3
d
1
; Figure 2-S4a) and chlorophyll a (0.34 ± 0.35
mg/m
3
; Figure 2-S4b), varied macronutrient concentrations (0.25 ± 0.57 µM NO3, 0.04 ±
0.06 µM NO2, 0.12 ± 0.09 µM PO4, 1.41 ± 0.71 µM SiO3; Figures 2-S3f, 2-S3e, 2-S3b, &
2-S3d, respectively), and moderate aerosol concentrations (D>0.25 µm; 43.62 ± 45.66
particles/cm
3
; Figure 2-S3h) were also found together with the high metal concentrations.
Due to the size of the Tara schooner and the focus on surface sampling we were able to
sample closer to the coastal margins. For example, the distance between the coastal
biome stations and the closest coastlines ranged from approximately 3 to 550 km. Our
29
high resolution spatial sampling allowed us to observe the gradient between coastal and
open ocean metal concentrations.
3.5.1.1 Upwelling of Cd
The average Cd concentration in the coastal biome is over four times higher than
the averages for the westerly and trade biomes (Figure 2-6, Table 2-3). Past studies have
shown that high Cd concentrations (>200 pM), along with increased concentrations of
major nutrients, in surface waters are indicative of upwelling (Segovia-Zavala et al., 1998;
Van Geen & Husby, 1996). Cd is considered a tracer of upwelling because its nutrient-type
distributions are generally not influenced by external enrichment and are instead
dominated by internal biogeochemical cycling and basin-scale circulation (Biller & Bruland,
2013).
Two main coastal upwelling systems associated with eastern boundary currents
were sampled during the Tara Pacific expedition: the California and Peru upwelling
ecosystems (CCAL and PEQD provinces, respectively; Capone & Hutchins, 2013). The
CCAL and PEQD provinces were two of the three provinces with the highest average Cd
concentrations (PEQD – 82.81 ± 67.71 pM, NECS – 80.93 ± 58.06 pM, and CCAL –
67.71 ± 48.49 pM; Figure 2-S2g, Table 2-3). Even though the Pacific Equatorial
Divergence province (PEQD) is officially in an open ocean biome, the three samples from
this province bordered the coastal Chili-Peru Current Coastal province (CHIL). Additionally,
two of the three PEQD samples had the highest Cd concentrations and thus will be
included in this discussion. By comparing corresponding biological (chlorophyll a and net
primary productivity), environmental (temperature), and metal data we were able to
30
determine whether or not our sampling in these eastern boundary current coastal regions
occurred during periods of upwelling.
Increased Cd concentrations were found in the California upwelling region in July of
2018 (CCAL, Figure 2-S2g), but were not characteristic of a strong upwelling period. The
California upwelling ecosystem is formed in the spring and goes through the summer when
strong equatorward winds form along the western coast of North America causes the
upwelling of cold, salty, nutrient rich water to the surface (Biller & Bruland, 2013). The
strongest upwelling occurs between 34°N and 44°N (Chavez & Messié, 2009). Biller &
Bruland (2013) sampled two upwelling events in the California Coastal System (May 2010
and August 2011) and found dCd concentrations that ranged from 322 to 870 pM. Total
dissolvable Cd concentrations from our samples taken in the California Coastal System
were much lower and ranged from 15 to 174 pM. Using sea surface temperature as a
tracer for upwelling, we see colder temperatures (<12°C) in the months leading up to our
sampling period (May and June 2018), but in July 2018 we only see <12°C temperatures
very close to the coastline outside of our sampling region (Figure 2-S5). In July 2018
slightly elevated chlorophyll a concentrations (0.08 to 0.63 mg/m
3
; Figure 2-S4b) and net
primary productivity rates (1.21 to 62.36 g/m
3
d
1
; Figure 2-S4a) were found in the CCAL
province, which may be an artifact of the stronger upwelling in the preceding months.
Also sampled during the Tara Pacific expedition was the Peru upwelling ecosystem
which extends from 4°S to 16°S and occurs year-round (Chavez & Messié, 2009; Mackas
et al., 2006). High Cd concentrations were found in samples off the coast of Peru (131-
153 pM; Figure 2-S2g) in colder surface waters (~22°C; Figure 2-S6). Unlike the CCAL
31
samples, the high Cd concentrations in the PEQD were accompanied by higher
macronutrient concentrations (PEQD: 6.8-11.82 µM NO3 and 0.76-1.07 µM PO4; CCAL:
0.04-1 µM NO3 and 0.08-0.36 µM PO4) which is indicative of upwelling (Figures 2-S3f & 2-
S3b). Chlorophyll a concentrations (0.18 to 0.26 mg/m
3
; Figure 2-S4b) and net primary
productivity rates (3.32 to 29.11 g/m
3
d
1
; Figure 2-S4a) however were low possibly due to
Fe-limitation which has been previously observed in this region (Bruland et al., 2004;
Hutchins et al., 2002). Previous studies found Fe-limitation in this upwelling region during
the austral winter (June-September) when surface dFe concentrations were ~0.1 nM and
apparent half-saturation constants for growth ranged from 0.17 to 0.26 nM Fe (Hutchins et
al., 2002). We sampled the PEQD also during the austral winter in August 2016 and found
Fe-limiting tdFe concentrations between 0.14 and 0.29 nM which were close to and below
the apparent half-saturation constants for phytoplankton growth.
Higher Cd concentrations off the coasts of California and Peru during the Tara
Pacific expedition hinted at the presence of upwelled waters. Based on environmental data
we suspect that we caught the tail-end of an upwelling event off the California coast and
sampled Fe-limited upwelled waters off the Peruvian coast. Additionally, the predicted Cd
concentrations from the GEOTRACES and Tara regression tree models showed a
correlation between Cd and the major nutrients PO4 and SiO4, which may have been a
result of upwelling.
3.5.1.2 Remobilized Cu
Average tdCu concentrations were the highest in the coastal biome and agree well
with dCu from pervious GEOTRACES cruises (Figure 2-6, Table 2-3). Cu has been found
32
to be significantly enriched in coastal waters due to the remobilization of Cu from
continental shelf sediments (Boyle et al., 1981; Skrabal et al., 1997) and Cu contamination
by antifouling marine paints (Bruland, 1980). One of the highest Cu concentrations found in
our study was sampled near the Panama Canal (3.53 nM tdCu; Figure 2-S2a) and agrees
well with past values measured in the Panama region (3-4 nM dCu; Boyle et al., 1981),
which is impacted by Cu contamination from ship traffic and near-shore sediments.
Though it can be considered toxic (Moffett & Brand, 1996), especially at high labile
concentrations, Cu is biologically essential as it is utilized in multiple electron transporters
(plastocyanin, Peers & Price, 2006; cytochrome oxidase, ascorbate peroxidase,
Contreras-Porcia et al., 2011), an Fe transporter (multicopper ferroxidase, Annett et al.,
2008), and can be used as the metal cofactor in superoxide dismutase (Chadd et al.,
1996). The importance of salinity in the GEOTRACES and Tara regression tree models also
suggested that Cu was influenced by fluvial inputs, thus we suspect Tara surface Cu
distributions were mainly controlled by coastal proximity.
3.5.1.3 Coastal Ni and Pb
The average coastal biome metal concentrations for all of the metals, except Ni and
Pb, were approximately two to four times greater than the average concentrations found in
the open ocean biomes (Figure 2-6, Table 2-3). Ni concentrations during the Tara Pacific
expedition were not significantly higher near the coast (2.95 ± 0.78 nM, Figure 2-S2c)
most likely due to the comparative minimal drawdown of Ni in the open ocean. Ni
concentrations in oligotrophic gyres do not decrease much below 2 nM despite its
biological demand (Boyle et al., 1981; Middag et al., 2020). This 2 nM Ni threshold is
33
thought to be a result of macronutrient limitation (John et al., in review). Though the coastal
biome’s average tdNi concentration is not drastically higher than the other biomes
(westerly – 2.49 ± 0.42 nM, trade – 2.34 ± 0.78 nM), we do see a general trend of
increases in tdNi closer to the coast (Figure 2-S2c).
Pb concentrations in the open ocean biomes (westerly – 24.00 ± 13.18 pM, trade –
16.53 ± 13.81 pM) are similar to the concentrations found in the coastal biome (20.93 ±
20.51 pM) due to the transport of Pb enriched aerosols to the open ocean (Figure 2-S2d).
Since the rise of industrialization in the mid-eighteenth century delivery of Pb to the oceans
has increased due to emissions from coal burning and combustion of leaded gasoline
(Boyle et al., 2005; Schaule & Patterson, 1981). Previous atmospheric observations have
found that anthropogenic are largely confined to the Northern Hemisphere where most of
the world’s population and industrialization resides (Cai et al., 2006; Xie et al., 2013).
Anthropogenic Pb from Asian industrial sources has been found as far as ~8,000 km east
of Asia in the middle of the North Pacific Subtropical Gyre at approximately 158°W
(Pinedo-González et al., 2020). Some of the highest tdPb concentrations were measured
in open ocean provinces, such as the NPSW, NPTG, and NASW provinces (Figure 2-S2d,
Table 2-3), therefore increases in Pb concentrations near the coast are not exceptionally
high in comparison to the entire Tara Pb dataset.
3.5.2 Westerly biome
The average metal concentrations for the five provinces sampled within the westerly
biome were moderate in comparison to the coastal and trade biomes (Figure 2-6). The
westerly biome is characterized as an open ocean regime and so the moderate metal
34
concentrations indicate there must be external sources of metals to this region. Most of
the sampling in the westerly biome was done in the North Atlantic (NASW, NASE, and
NADR provinces, Figure 2-1) where the surface metal inventory is known to be affected by
Saharan dust deposition (Jickells et al., 2005; Shelley et al., 2015). The Tara Pacific
expedition transected through the North Atlantic in the NASE and NASW provinces from
May to July 2016 (between approximately 25-42°N) when the maximum of the Saharan
dust plume is usually around 20°N (Bowie et al., 2002; Moulin et al., 1997). 48 hour back
trajectories of the aerosol masses sampled on the Tara Pacific expedition made by Flores
et al. (2020) show that the aerosols sampled between 25-30°N originated west of the
sampling locations, near the northwestern coast of Africa. Flores et al. (2020) also found
high aerosol concentrations in the transect through the NASE and NASW provinces with
concentrations ranging from 9.28 to 244.21 cm
-3
(D>0.25 µm). A monthly average of the
total dust deposition (both dry and wet) from May to July 2016 shows that the stations
closest to 25°N experienced the largest amount of dust deposition (Figure 2-S7).
3.5.2.1 Fe, Mn, Co, and Saharan dust
In the North Atlantic transect of the Tara Pacific expedition we measured high Fe,
Mn, and Co concentrations near 25°N where the largest amount of dust deposition was
observed (1.66-13.55 nM tdFe, 2.19-5.36 nM tdMn, 16-62.86 pM dCo; Figures 2-2, 2-
S2b, & 2-S2f). Past studies have shown that the aerosol sources for Fe, Mn, and Co are
primarily desert dust (Desboeufs et all., 2005; Mahowald et al., 2018; Shelley et al., 2015).
For example, a GEOTRACES GA03 zonal transect cruise through the North Atlantic (17-
39°N) used Fe isotope measurements to determine that 71-87% of the dFe measured was
35
from Saharan dust aerosols (Conway & John, 2014). In the NASW province, surface dFe
concentrations measured during GA03 ranged from 0.48 to 1.8 nM (Conway & John,
2014; Hatta et al., 2015). Surface Tara tdFe was higher in the NASW province (0.15 to
8.96 nM tdFe; Figure 2-2) than the dFe concentrations observed on GA03, but similar to
previous tdFe measurements from 20°N in the North Atlantic (7.4 nM tdFe, Bowie et al.,
2002).
During GA03, dMn ranged from 1.70 to 3.10 nM (Wu et al., 2014) which was within
the range of tdMn found during Tara Pacific’s NASW transect (0.96 to 5.36 nM; Figure 2-
S2b). Previous studies have measured high dMn up to at least 30°N in the North Atlantic
due to increased dust deposition from the Sahara desert and subsequent photoreduction
(de Jong et al., 2007; Van Hulten et al., 2017; Wu et al., 2014). Even though elevated
surface Mn in the open ocean is often attributed to atmospheric sources, surface Mn can
also be a result of the lateral transport of reducing sediments from the continental margin.
The influence of reducing sediments on Mn inventories has been connected to a strong
inverse relationship between Mn concentrations and salinity (Wu et al., 2014). Input of
tdMn by reducing sediments was ruled out for the samples taken between 25-30°N during
the Tara Pacific expedition since there was a weak inverse correlation between tdMn and
salinity (R
2
= 0.23). Consequently, we suspect the elevated tdMn concentrations observed
in the North Atlantic are mainly a product of Saharan dust deposition. Wu et al. (2014)
similarly concluded that the Mn measured at the surface during the GA03 cruise was
controlled primarily by dust deposition and photochemical reduction of MnO2.
36
Like Mn, the main external sources of Co are atmospheric deposition and
continental weathering (Wong et al., 1995), therefore Mn and Co distributions tend to
mimic one another (Knauer et al., 1982). Co is an essential micronutrient as it is used in
vitamin B12 (Bertrand et al., 2007) and carbonic anhydrase (Morel et al., 1994). We found a
moderate correlation between the tdCo and salinity values measured between 25-30°N (R
2
= 0.45). This inverse correlation suggests a continental weathering source of Co to this
region (Knauer et al., 1982; Saito & Moffett, 2001), but does not rule out the possibility of
an additional atmospheric source of Co.
Net primary productivity (0.36-12.54 g/m
3
d
1
; Figure 2-S4a) and chlorophyll a
(0.012-0.40 mg/m
3
; Figure 2-S4b) levels were not elevated in the NASW and NASE
provinces possibly due to macronutrient limited phytoplankton growth or the inherently low
solubility of nutrients associated with dust (0.01-0.75 µM NO3, 0.01-0.02 µM NO2, 0.01-
0.04 µM PO4, 0.53-1.21 µM SiO3; Figures 2-S3f, 2-S3g, 2-S3b, & 2-S3d, respectively).
The solubility of nutrients in dust particles has been found to be much lower than that of
nutrients from other aerosol sources (Baker et al., 2006; Baker et al., 2016; Mahowald et
al., 2018). For example, Baker et al. (2006) found that the solubilities of nutrients in
Saharan dust samples (Fe – 1.2%, P – 8.0%, Si – 0.12%, Al – 3%, Mn – 51%) were
significantly lower than remote South Atlantic marine air (Fe – 2.7%, P – 17%, Si – 0.29%,
Al – 7.6%, Mn – 57%). Other studies have shown that even if the dust-associated nutrients
are soluble and bioavailable, that the atmospheric deposition of more N and Fe relative to
P can sometimes induce further P limitation therefore decreasing primary productivity in
37
the North Atlantic (Boyd et al., 2010; Chien et al., 2016) hence possibly explaining the lack
of productivity observed during the Tara Pacific expedition.
The increases in Fe, Mn, and Co concentrations observed during Tara’s southern
transect of the North Atlantic are most likely a product of Saharan dust deposition.
Comparisons to satellite data and past studies provide further evidence that dust
deposition controls most of the metal distributions in the North Atlantic open ocean.
However, the inverse relationship between Mn or Co and salinity in both the observed and
predicted data shows that the lateral transport of metals from continental sources may
also contribute to the North Atlantic Mn and Co inventories.
3.5.3 Trade biome
The metal concentrations found in the eight sampled trade provinces varied
considerably due to the close proximity of some stations to the Pacific Islands and
anthropogenic aerosol deposition from southeast Asia despite the classification of the
trade biome as an open ocean regime. More than half (68%) of the samples from the Tara
Pacific expedition were collected in the trade biome and all but three of the trade biome
samples were from the Pacific Ocean (Figure 2-1). For 5 of the 8 metals measured in this
study (Co, Cu, Pb, Mn, Ni), the average metal concentrations for the trade biome was the
lowest of the three sampled biomes (Figure 2-6, Table 2-3). Along with low metal
concentrations, the trade biome had on average low productivity (5.40 ± 7.55 g/m
3
d
1
;
Figure S4a), chlorophyll a concentrations (0.09 ± 0.07 mg/m
3
; Figure S4b), and aerosol
concentrations (D>0.25 µm; 20.27 ± 27.35 particles/cm
3
; Figure S3h).

38
3.5.3.1 Pb in Asian aerosols
Some of the highest tdPb concentrations were found in the North Pacific provinces
(NPSW – 34.14 ± 16.36 pM and NPTG – 32.88 ± 17.09 pM, Figure 2-S2d) alongside high
aerosol concentrations (4.32-63.53 particles/cm
3
, Figure 2-S3h; Flores et al., 2020) even
though the average Pb concentration for the overall trade biome was the lowest (Table 2-
3). Elevated Pb concentrations have previously been measured in the North Pacific and
linked to anthropogenic Asian aerosol deposition (Gallon et al., 2011; Pinedo-González et
al., 2020). For instance, in May of 2017 Pinedo-González et al. (2020) found high Fe
concentrations (0.30-0.52 nM dFe) between 35-40°N at 158°W which coincided with high
Pb (35.5-65 pM dPb) and Pb isotope ratios matching Asian anthropogenic sources.
During the Tara Pacific expedition we found a similar increase in Fe (0.09-1.43 nM tdFe)
and Pb (27.82-60.60 pM tdPb) concentrations between 28-35°N during the summer
months of 2018 (Figures 2-2 & 2-S2d). A strong relationship between Pb and northern
latitudes was also observed in the Pb regression tree models for both the GEOTRACES
and Tara datasets (Figure 2-4), providing further evidence that aerosol deposition is likely
driving Pb distributions.
3.5.3.2 Fe and Mn in the South Pacific Islands
Elevated Fe and Mn concentrations were found in the western portion of the South
Pacific Subtropical Gyre (SPSG) and the Archipelagic Deep Basins (ARCH) provinces
during the Tara Pacific expedition (Figures 2-2 & 2-S2b). The similarities between the Fe
and Mn distributions are unsurprising as both of these metals have similar sources and
sinks (Saager et al., 1989). The high Fe and Mn concentrations in the SPSG province were
39
found in the coastal waters near Fiji (473.55 nM tdFe, 27.92 nM tdMn) and Samoa (66.52
nM tdFe, 5.18 nM tdMn). Possible sources of Fe and Mn in the western region of the
SPSG include near-shore sediments and shallow hydrothermal inputs from the Lau Ridge
and Tonga Arc (Guieu et al., 2018). High Fe (0.16-18.18 nM tdFe) and Mn (0.95-7.13 nM
tdMn) was also found in the ARCH province (Figures 2-2 & 2-S2b). The increased Fe and
Mn concentrations in this region overlap with previous Fe and Mn measurements from the
Bismarck, Solomon, and Coral Seas, which are also located in the ARCH province (0.20-
1.90 nM tdFe & 1.50-2.50 nM tdMn – Mackey et al., 2002; 0.79-1.61 nM tdMn – Michael
et al., 2021; <0.05 - 0.60 nM tdFe & 0.60-3.00 nM tdMn – Obata et al., 2008). The ARCH
province consists of minor deep seas partially enclosed, or entirely enclosed, by the island
chains and archipelagos between the Indian and Pacific Oceans (Longhurst, 1998). Waters
surrounding the Pacific Islands in the ARCH province come into contact with many
potential sources of Fe and Mn such as near-shore sediments, fluvial runoff, mine tailings
(Morrison et al., 2013), and hydrothermal or volcanic activity (Guieu et al., 2018; Michael et
al., 2021).

4. Conclusion
In this study 242 trace metal surface seawater samples were measured and
analyzed to characterize the surface metal distributions across the North Atlantic, North
Pacific, and South Pacific Oceans. An analysis of the data by regression tree analysis
suggests that the simple pole sampler successfully collected uncontaminated, high spatial
resolution trace metal samples. We find that the machine-learning techniques are generally
40
not able to produce data which matches the accuracy of observations, however,
motivating future efforts to collect samples for trace metal analysis. Using previous studies
we assessed which sources of surface metals (e.g. upwelled waters, aerosols, coastal
proximity, hydrothermal vents) were contributing to the metal inventory in 21 marine
provinces and 3 major ocean biomes. For example, decreases in sea surface temperature,
increases in tdCd, and the importance of PO4 and SiO4 as predictors of Cd concentrations
suggested there was upwelling occurring in the PEQD province. Additionally, the inverse
relationship between Co and salinity in both the observational data and the regression tree
models implied riverine input was affecting Co distributions in the NASW province. Using
coupled aerosol, macronutrient, chlorophyll a concentrations, and net primary productivity
rates we were able to further examine which biogeochemical processes were controlling
the bioavailability (i.e. solubility) of these metals.
The trace metal data presented in this study complement other biological and
environmental data in the larger Tara Oceans database. By comparing the metal
concentrations reported here with future biological datasets we can explore how the
metals are interacting with the marine microbial communities throughout the North Atlantic
and Pacific Oceans.

41

Table 2-1
R-squared (R
2
) and root mean square deviation (RMSE) values for outputs of optimizable
regression tree models tested using GT-A, GT-B, and Tara environmental data. Each
model was previously trained using GT-A data.

Metal
R
2
RMSE
GT-A GT-B Tara GT-A GT-B Tara
Cd 0.95 0.76 0.01 43.22 98.05 56.94
Cu 0.59 0.51 0.33 0.40 0.61 0.60
Fe 0.88 0 0 0.15 1.94 31.40
Mn 0.94 0.07 0.14 0.37 1.51 2.71
Ni 0.96 0.81 0.18 0.26 0.93 0.81
Pb 0.84 0.28 0.24 3.54 16.07 22.15
Zn 0.45 0.35 0 0.61 0.76 0.92

42
Table 2-2
R-squared (R
2
) and root mean square deviation (RMSE) values for outputs of optimizable
regression tree models tested using GT-A and Tara environmental data. Each model was
previously trained using each respective dataset.

Metal
R
2
RMSE
GT-A Tara GT-A Tara
Cd 0.92 0.16 54.73 47.09
Cu 0.49 0.34 0.44 0.46
Fe 0.68 0 0.26 31.34
Mn 0.64 0.12 0.93 2.33
Ni 0.87 0.28 0.46 0.65
Pb 0.51 0.47 6.12 11.09
Zn 0.31 0.02 0.69 0.84
Co N/A 0.28 N/A 26.18

43
Table 2-3
List of Longhurst biomes and provinces sampled in this study

Biome Province ID Province
Coastal AUSE East Australian Coastal
CAMR Central American Coastal
CCAL California Upwelling Coastal
CHIL Chile-Peru Current Coastal
CHIN China Sea Coastal
NECS NE Atlantic Shelves
NWCS NW Atlantic Shelves
SUND Sunda-Arafura Shelves
Westerly KURO Kuroshio Current
NADR North Atlantic Drift
NASE North Atlantic Subtropical Gyre – East
NASW North Atlantic Subtropical Gyre – West
NPPF North Pacific Polar Front
Trade ARCH Archipelagic Deep Basins
CARB Caribbean
NPSW North Pacific Subtropical Gyre – West
NPTG North Pacific Tropical Gyre
PEQD Pacific Equatorial Divergence
PNEC North Pacific Equatorial Countercurrent
SPSG South Pacific Subtropical Gyre
WARM West Pacific Warm Pool

44
Table 2-4
Average metal concentrations from each province and biome. C_GT, W_GT, and T_GT
refer to the average metal concentrations for the surface (<25 m) GEOTRACES data in the
coastal, westerly, and trade biomes, respectively.

Provinces Fe (nM) Zn (nM) Mn (nM) Ni (nM) Cd (pM) Co (pM) Cu (nM) Pb (pM)
AUSE 9.57±21.62 0.27±0.26 1.39±0.55 2.17±0.08 3.16±2.86 7.81±5.39 0.59±0.08 12.52±1.27
CAMR 6.05±7.28 0.53±0.38 3.11±1.24 2.43±0.38 26.30±32.92 28.08±19.91 1.81±1.23 11.10±6.07
CCAL 0.61±0.72 0.13±0.09 3.82±3.73 3.43±0.27 67.71±48.49 47.85±47.48 1.19±0.37 14.43±3.61
CHIL 0.22±0.11 0.45±0.25 2.2±0.75 3.06±0.90 92.94±85.12 26.74±8.92 1.03±0.07 17.99±3.67
CHIN 0.75±0.31 0.15±0.03 2.66±0.13 2.61±0.03 20.59±2.61 19.44±3.96 1.08±0.10 36.84±4.47
NECS 33.06±32.42 2.49±1.80 4.20±3.34 3.13±0.39 80.93±58.06 44.27±34.81 1.91±0.92 73.41±46.69
NWCS 2.06±3.08 0.96±1.03 5.14±4.67 3.67±1.16 70.88±68.99 65.44±60.78 2.23±1.11 13.84±3.89
SUND 1.58±2.24 1.10±1.27 4.23±3.49 3.13±0.99 52.35±59.96 47.32±46.36 1.74±1.01 29.52±30.30
COAST 5.41±14.10 0.58±0.89 3.39±3.00 2.95±0.78 46.50±52.55 36.92±40.02 1.36±0.85 20.93±20.51
C_GT 0.32±0.31 0.33±0.50 2.34±1.90 3.13±0.82 100.84±192.45 N/A 1.05±1.03 27.79±15.57
KURO 1.02±0.45 0.16±0.08 1.61±.15 2.23±0.05 17.12±8.48 9.60±4.23 0.66±0.02 47.08±15.95
NADR 0.31±0.42 0.57±0.25 0.95±0.43 2.72±0.29 29.98±10.49 28.94±12.42 1.22±0.20 29.17±19.15
NASE 0.61±0.81 0.32±0.25 0.74±0.20 2.45±0.12 4.90±1.90 21.35±5.70 1.04±0.04 21.46±4.12
NASW 3.31±2.92 0.20±0.20 2.79±1.27 2.21±0.10 1.26±0.69 29.62±13.07 0.96±0.30 19.23±7.28
NPPF 0.20±0.17 0.05±0.02 0.65±0.08 3.22±0.30 9.18±10.19 7.01±2.96 0.94±0.11 23.37±13.33
WEST 1.80±2.48 0.25±0.25 1.80±1.32 2.49±0.42 9.09±12.14 23.15±13.86 0.98±0.26 24.00±13.18
W_GT 0.37±0.52 0.58±0.61 0.81±0.59 4.35±1.34 195.65±221.33 N/A 0.95±0.49 14.33±12.89
ARCH 2.87±5.19 0.52±0.64 2.30±1.51 2.40±0.85 2.42±2.34 20.29±44.30 0.83±0.89 16.78±15.08
CARB 1.36±0.28 0.24±0.09 4.70±1.34 2.21±0.24 3.62±4.59 34.87±19.28 1.31±0.28 10.08±3.77
NPSW 0.37±0.38 0.26±0.22 1.38±0.39 2.25±0.10 6.08±8.336 7.12±2.65 0.64±0.05 34.14±16.36
NPTG 0.29±0.22 0.23±0.27 0.98±0.26 2.32±0.06 2.10±1.16 3.80±2.06 0.57±0.05 32.88±17.09
PEQD 0.23±0.08 0.21±0.17 2.52±0.17 3.19±0.22 81.81±67.71 21.63±0.94 0.87±0.21 12.64±9.37
PNEC 0.44±0.40 0.98±2.77 2.73±0.88 2.45±0.40 18.59±15.61 10.82±6.08 0.85±0.28 13.83±10.10
SPSG 12.32±70.24 0.29±0.73 1.78±4.22 2.46±1.30 1.22±2.681 11.61±37.96 0.47±0.39 8.26±4.97
WARM 0.26±0.56 0.28±0.57 1.27±0.72 2.18±0.22 17.94±103.24 11.05±22.64 0.60±0.19 12.92±3.56
TRADE 4.00±47.10 0.35±0.92 1.74±2.43 2.34±0.78 8.69±53.61 11.94±28.64 0.63±0.43 16.53±13.81
T_GT 0.19±0.22 0.31±0.74 1.52±1.50 2.66±0.71 15.23±30.29 N/A 0.74±0.64 15.67±9.44
45

Figure 2-1
Sampling locations during the Tara Pacific expedition from May 2016 to October 2018.
The stars indicate where aerosol concentrations (diameter >2.5 µm) were >50 cm
-3
.
Arrows and dates show the path of the Tara Pacific transect. Provinces are indicated by
grey boundaries with the province ID labeled in the middle of each region. The coastal,
westerly, and trade biomes are indicated by the blue, green, and yellow shading,
respectively.

BPLR
PSAW PSAE
NPPF KURO
NPSW
NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
SSTC
SANT
SATL
WTRA
FKLD
ETRA
GUIN
ARCT
SARC
ANTA
CHIL
BRAZ
GUIA
CAMR
NATR
NASW
GFST
NADR
NASE
NECS
CNRY
ALSK
BERS
TASM
NEWZ
AUSW
Coastal
Westerly
Trade
May
2016
July 2016
Aug
2016
Oct 2016
CARB
Dec 2016
Feb 2017
May 2017
June 2017
Oct
2017
Dec 2017
Mar
2018
May 2018
July 2018
Sept 2018
Oct 2018 NWCS
46

Figure 2-2
Sea surface total dissolvable iron concentrations measured during the Tara Pacific
expedition. Grey outlined circle represent the stations where aerosol concentration was
>50 cm
-3
(diameter >2.5 µm). Note changes in ranges of concentrations and changes of
units on the z-axis of each plot. Sample concentrations that are not represented on the z-
axis are labeled accordingly on figure.

474
66
17
16
14
34
65
54
18
NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Fe (nM)
47

Figure 2-3
Predicted versus true metal concentrations for outputs of optimizable regression tree
models trained on GT-A data and tested using GT-A, GT-B, and Tara environmental data.
Values for GT-A, GT-B, and Tara data are represented by the dark grey, light grey, and red
circles, respectively. Note changes in ranges of concentrations and changes of units on
the x-axis and y-axis of each plot.

GT-A GT-B Tara
0
20
40
60
80
100
0 20 40 60 80 100
True Fe (nM)
Fe
0
2
4
6
8
10
0 2 4 6 8 10
Predicted Zn(nM)
True Zn (nM)
Zn
0
20
40
60
80
100
0 20 40 60 80 100
Predicted Pb (pM)
True Pb (pM)
Pb
0
2
4
6
8
10
0 2 4 6 8 10
Predicted Ni (nM)
True Ni (nM)
Ni
0
5
10
15
20
0 5 10 15 20
Predicted Mn (nM)
True Mn (nM)
Mn
0
2
4
6
8
10
0 2 4 6 8 10
Predicted Fe (nM)
True Fe (nM)
Fe
0
1
2
3
4
5
0 1 2 3 4 5
Predicted Cu (nM)
True Cu (nM)
Cu
0
200
400
600
800
1000
0 200 400 600 800 1000
True Cd (pM)
Cd
Predicted Cd (pM)
True Cd (pM) True Mn (nM) True Pb (pM) True Fe (nM)
True Cu (nM) True Ni (nM) True Zn (nM) True Fe (nM)
Predicted Mn (nM)
Predicted Pb (pM)
Predicted Fe (nM)
Predicted Cu (nM)
Predicted Ni (nM)
Predicted Zn (nM)
Predicted Fe (nM)
48

Figure 2-4
Estimates of predictor importance for each of the environmental parameters used to train
the optimizable regression tree models for the GT-A and Tara datasets. The first, second,
and third highest estimates of predictor importance values are highlighted in blue, green,
and yellow, respectively.

Metal Data Set Latitude Longitude Salinity Temperature O
2
PO
4
SiO
4
Cd
GT-A 0.7413 11.33 18.46 0 0.1417 2126 0.8489
Tara Pacific 0.0141 0.0030 1.087 13.20 0 17.23 14.86
Cu
GT-A 0.0010 0.0022 0.0096 0.0002 0 0.0007 0.0001
Tara Pacific 0.0001 0.0026 0.0003 0.0030 0 0.0001 0.0003
Fe
GT-A 0.0005 0.0004 0.0001 0 0.0002 0.0030 0
Tara Pacific 0 0 0 0 0 0 11.88
Mn
GT-A 0.0086 0.0122 0.0131 0.0048 0 0.0015 0.0068
Tara Pacific 0.0027 0.0021 0.0001 0.0422 0.0056 0.0017 0.0466
Ni
GT-A 0.0007 0.0012 0.0019 0.0328 0.0005 0.0019 0.0004
Tara Pacific 0.0004 0.0019 0.0073 0.0013 0 0.0005 0.0001
Pb
GT-A 1.070 0.0542 0.0048 0.1810 0.0357 0.0079 0.1715
Tara Pacific 2.857 2.697 0.0065 0.0131 1.332 0.1909 0.0295
Zn
GT-A 0 0.0041 0.0002 0.0024 0 0.0312 0.0004
Tara Pacific 0.0048 0 0.0243 0 0 0 0
Co
Tara Pacific 0.5958 0.5488 2.7906 0.0434 5.064 0.6977 5.903
49

Figure 2-5
GT-A’s Cd optimizable regression tree. Each of the nodes represents a “decision”. The
samples with a value less than or to the west and south, for latitude and longitude,
respectively, are binned to the left of the node. The samples with a value higher than or to
the east and north, for latitude and longitude, respectively, are binned to the right of the
node. Each of the leaves represent a predicted Cd concentration.

PO
4
1.20 µM
Latitude
39.67°S
Longitude
18.98°W
Latitude
39.99°S
Longitude
12.95°W
Longitude
122.51°W
Longitude
73.44°E
PO
4
1.71 µM
PO
4
0.89 µM
SiO
4
2.69 µM
O
2
240.53 µM
PO
4
0.10 µM
PO
4
0.46 µM
O
2
235.43 µM
O
2
253.36 µM
8.81 pM
12.46 pM 11.68 pM 8.98 pM 12.08 pM
14.58 pM 11.31 pM
17.35 pM 14.12 pM
18.70 pM
1.16 pM 4.53 pM
27.51 pM
Latitude
33.27°S
131.70 pM
416.93 pM 308.44 pM 591.20 pM 509.69 pM
Salinity
33.86
Cd
50

GEOTRACES TARA
0
100
200
300
400
500
600
700
800
900
1000
Cd (pM) Pb (pM)
GEOTRACES TARA
0
1
2
3
4
5
6
7
8
9
10
Fe (nM) Cu (nM) Mn (nM) Ni (nM) Zn (nM)
GEOTRACES TARA
0
100
200
300
400
500
600
700
800
900
1000
GEOTRACES TARA
0
1
2
3
4
5
6
7
8
9
10
GEOTRACES TARA
0
100
200
300
400
500
600
700
800
900
1000
GEOTRACES TARA
0
1
2
3
4
5
6
7
8
9
10
v
v
v
v
v
v
v
v
v
Coastal Coastal
Westerlies Westerlies
Trades Trades
51
Figure 2-6
The metal concentrations for the GEOTRACES and Tara samples in each of the sampled
ocean biomes (coastal, westerly, and trade). The lines inside the boxes represent the
median (middle value), the lower and upper bounds of the boxes represent the lower and
upper quartiles (middle values of the lower and upper halves, respectively), the whiskers
represent the minimum and maximum values, and the dots represent outliers. Y-axis units
are metal dependent and signified in the legend.

52

PO
4
0.37 µM
Latitude
38.29°S
Longitude
105.26°W
Temperature
18.18°C
Longitude
19.77°W
PO
4
1.05 µM
SiO
4
2.25 µM
SiO
4
1.43 µM
Temperature
13.99°C
Longitude
22.26°W
Latitude
40°S
Salinity
34.67
O
2
320.34 µM
PO
4
1.69 µM
Salinity
33.72 PSU
Salinity
34.09 PSU
2.40 nM
1.18 nM
1.33 nM 1.44 nM
0.52 nM 0.45 nM 0.55 nM
0.64 nM
0.50 nM
0.60 nM
0.64 nM
1.32 nM 1.55 nM
0.93 nM
0.82 nM 0.74 nM
0.79 nM
PO4 0.14 µM
Latitude 40.001°S
Latitude 40°S
Longitude 9.41°W
Longitude 6.38°W
Longitude 165°W
Latitude 39.9995°S
Latitude 39.9993°S
Latitude 39.97°S
Longitude 11.41°E
SiO4 2.97 µM
Longitude 33.02°W
O2
237.55 µM
O2
226.39 µM
Longitude 16.42°W
Longitude
87.76°W
Temperature
27.91°C
Latitude
14.39°S
Latitude
15.22°S
Latitude
27.50°S
Latitude
33.75°S
Salinity
33.77 PSU
O2
290.98 µM
PO4
1.71 µM
PO4
1.86 µM
Latitude
52.88°S
Latitude
53.05°S
Salinity
33.87 PSU
Salinity 33.88 PSU
O2
318.30 µM
0.06 nM
0.07 nM 0.11 nM
0.04 nM
0.10 nM
0.05 nM 0.07 nM
0.12 nM
0.31 nM
0.02 nM
1.08 nM 0.28 nM
0.17 nM
0.59 nM 0.29 nM
0.20 nM 0.13 nM
0.44 nM
0.42 nM
0.20 nM 0.08 nM
0.18 nM
0.11 nM 0.19 nM
0.15 nM
0.07 nM
0.16 nM
0.08 nM
0.13 nM
0.08 nM
0.12 nM
0.09 nM 0.11 nM
0.13 nM
1.46 nM 1.35 nM
1.52 nM
1.72 nM
2.09 nM
0.69 nM
0.50 nM
0.67 nM
O2
206.33 µM
O2
250.45 µM
0.08 nM
O2
217.17 µM
PO4
0.08 µM
Temperature
18.23°C O2
233 µM
Temperature
18.59°C
Longitude
17.26°W
SiO4
0.82 µM
O2 214.87 µM
PO4
1.50 µM
SiO4
1.17 µM
0.31 nM
PO4 0.1 µM
Cu
Fe
(a)
(b)
53

0.99 nM
0.60 nM
0.62 nM
0.50 nM
0.38 nM
0.61 nM
0.47 nM
0.30 nM
0.38 nM
0.30 nM
0.34 nM 0.37 nM
0.20 nM
0.59 nM
0.97 nM
1.52 nM
1.44 nM
1.79 nM
1.29 nM
0.94 nM
0.67 nM
0.99 nM
0.73 nM
0.63 nM
1.44 nM
11.18 nM
6.47 nM
1.54 nM
2.66 nM 0.99 nM
4.19 nM
1.96 nM
2.33 nM
2.06 nM
2.98 nM
8.04 nM
PO4
0.38 µM
Temperature
17.46°C
PO4
1.71 µM
PO4
1.78 µM
Longitude
44.30°W
Salinity
33.87 PSU
1.98 nM
Longitude
37.80°W
Longitude
107.86°W
Longitude
102.76°W
Longitude
19.77°W
Temperature
23.82°C
Temperature
23.26°C
Salinity
35.64 PSU
Salinity
35.07 PSU
Latitude
13.28°S
Longitude
92.76°W
Temperature
18.93°C
Latitude
38.29°S
Salinity
31.62 PSU
SiO4
1.90 µM
Longitude
72.59°E
Temperature
15.50°C
PO4
0.19 µM
O2 234 µM
Latitude
40°S
Temperature 10.91°C
Latitude 39.14°S
Longitude 6.34°E
Latitude 40°S
O2
321.29 µM
Salinity
33.88 PSU
1.52 nM
Latitude 40°S
0.72 nM
PO4
0.31 µM
Longitude
10.12°W
0.48 nM
Longitude
18.98°W
O2
285.19 µM
0.53 nM
Longitude 6.38°W
Temperature
3.03°C
0.11 nM 0.18 nM
1.86 nM
1.73 nM
O2
321.9 µM
Longitude
74.19°E
Latitude
31.25°S
Temperature
26.62°C
0.87 nM
Temperature
24.30°C
3.41 nM
SiO4
20.33 µM
Longitude
16.86°E
SiO4
1.67 µM Temperature
18°C
1.26 nM
6.25 nM 6.40 nM 6.31 nM
6.51 nM
6.59 nM
6.07 nM
5.82 nM 6.19 nM
4.06 nM
5.50 nM
2.68 nM
4.24 nM
3.29 nM
3.54 nM
3.86 nM 3.99 nM
3.77 nM
3.64 nM
3.41 nM 3.28 nM
3.06 nM
3.39 nM
3.12 nM
2.65 nM
3.47 nM 3.27 nM
3.64 nM 3.47 nM
3.85 nM
3.25 nM
2.51 nM
3.02 nM
2.67 nM
2.80 nM 2.73 nM
2.28 nM
2.38 nM
3.26 nM 2.83 nM
3.43 nM
Temperature
9.91°C
PO4
1.25 µM
Latitude
50.74°S
Longitude
38.81°W
O2
320.84 µM
O2
320.34 µM
Temperature
3.22°C
Temperature
2.91°C
Temperature
2.78°C
Salinity
33.88 PSU
SiO4
12.64 µM
Longitude
170°W
SiO4
8.56 µM
O2
235.41. µM
Longitude
51.05°W
Salinity
35.07 PSU
Salinity
35.24 PSU
PO4
0.44 µM
O2
265.75 µM
SiO4
0.55 µM
Latitude
40.04°S
PO4
1.05 µM
SiO4
2.83 µM
O2
240.73 µM
O2
206.51 µM
PO4
0.09 µM
O2
221.61 µM
Latitude
36.98°S
Salinity
35.39 PSU
Longitude
14.59°E
Temperature
21.13°C
Longitude
48.87°W
Salinity
35.14 PSU
Salinity
34.98 PSU
Temperature
16.32°C
Longitude
18.59°W
Longitude
31.27°W
Latitude 40°S
Latitude 40.01°S
Mn
Ni
(c)
(d)
54

Temperature
11.53°C
Longitude 27.97°W
SiO4 0.52 µM
Salinity 35.17 PSU
PO4 0.22 µM
O2
233.20 µM
Latitude 40.05°S
Latitude 40°S
SiO4 1.07 µM
Latitude 39.99°S
Latitude 39.70°S
PO4 0.11 µM
SiO4
2.17 µM
Latitude
38.99°S
Latitude
43.14°S
Temperature
4.18°C
O2
228.8 µM
Latitude 39.99°S
Longitude
72.83°E
Longitude 17.59°E
Latitude 12°S
PO4 0.19 µM
O2
216.58 µM
Salinity 35.92 PSU
SiO4
1.76 µM
Salinity 35.84 PSU
Longitude 51.20°W
SiO4
10.07 µM
Temperature 19.07°C
Latitude
20°N
Latitude
38.09°S
Latitude
22.50°S
11.67 pM
10.77 pM
11.17 pM
10.46 pM
10.27 pM
9.94 pM
9.79 pM
11.57 pM
11.77 pM
9.59 pM
9.56 pM 10.07 pM
9.19 pM
11.97 pM
20.06 pM
22.47 pM
23.04 pM
25.10 pM
19.20 pM
13.81 pM
10.01 pM
9.00 pM
13.81 pM
16.25 pM
16.97 pM
17.54 pM
13.41 pM
14.39 pM
20.08 pM
40.59 pM
14.24 pM
53.66 pM
8.90 pM
20.07 pM
10.54 pM
12.74 pM
4.54 pM
5.55 pM
2.91 pM
8.84 pM
6.01 pM
9.30 pM
13.17 pM
Longitude
14.79°E
O2 234.35 µM Longitude
52.46°W
Latitude
10.79°S
SiO4 1.46 µM
Latitude
53.03°S
PO4
1.79 µM
O2 324.17 µM
Latitude
52.98°S
Longitude 37.80°W
Longitude
8.04°E
Temperature
18.92°C
SiO
4
0.07 µM
0.37 nM
Salinity
35.88 PSU
SiO
4
1.07 µM
0.10 nM
Longitude
17.26°W
0.18 nM 0.22 nM
0.33 nM
0.08 nM
0.75 nM
PO
4
1.41 µM
Temperature
3.10°C
1.89 nM 1.28 nM
Pb
Zn
(e)
(f)
55
Figure 2-S1
GT-A’s optimizable regression trees for (a) Cu, (b) Fe, (c) Mn, (d) Ni, (e) Pb, and
(f) Zn. Each of the nodes represents a “decision”. The samples with a value less than or to
the west and south, for latitude and longitude, respectively, are binned to the left of the
node. The samples with a value higher than or to the east and north, for latitude and
longitude, respectively, are binned to the right of the node. Each of the leaves represent a
predicted metal concentration for each respective tree.

56

NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Cu (nM)
12
14
28
NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Mn (nM)
(b)
(a)
57

6
NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Ni (nM)
Pb (pM) (d)
(c)
58

10
NPPF KURO
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
Zn (nM)
183
138
129
116
238
151
206
NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Co (pM) (f)
(e)
59

Figure 2-S2
Sea surface total dissolvable (a) copper, (b) manganese, (c) nickel, (d) lead, (e) zinc, (f)
cobalt, and (g) cadmium concentrations measured during the Tara Pacific expedition. Grey
outlined circle represent the stations where aerosol concentration was >50 cm
-3
(diameter
>2.5 µm). Note changes in ranges of concentrations and changes of units on the z-axis of
each plot. Sample concentrations that are not represented on the z-axis are labeled
accordingly on figure.

174
153
131
679
169
146
106
122
106 NPPF
NPSW NPTG
CCAL
PNEC
PEQD
SPSG
ARCH
AUSE
SUND
WARM
CHIN
CHIL
CAMR
CARB
NASW
NWCS
NADR
NASE
NECS
KURO
Cd (pM)
(g)
60

Salinity (PSU)
PO
4
(µM)
(b)
(a)
61

Temperature (°C)
SiO
4
(µM)
(d)
(c)
62

O
2
(µM)
NO
3
(µM)
(f)
(e)
63

Figure 2-S3
Sea surface (a) salinity, (b) dissolved phosphate, (c) temperature, (d) dissolved silicate, (e)
dissolved oxygen, (f) dissolved nitrate, (g) dissolved nitrite, and (h) aerosol concentrations
measured during the Tara Pacific expedition. Grey outlined circle represent the stations
where aerosol concentration was >50 cm
-3
(diameter >2.5 µm). Note changes in ranges of
concentrations and changes of units on the z-axis of each plot.
NO
2
(µM)
Aerosol, Diameter >0.25 µm (particles/cm
-3
)
(h)
(g)
64

Figure 2-S4
Sea surface (a) net primary productivity and (b) chlorophyll a measured during the Tara
Pacific expedition. Grey outlined circle represent the stations where aerosol concentration
was >50 cm
-3
(diameter >2.5 µm). Note changes in ranges of concentrations and changes
of units on the z-axis of each plot.

Net Primary Production (g m
-3
d
-1
)
Chlorophyll a (mg m
-3
)
(b)
(a)
65

Figure 2-S5
Sea surface temperature in degrees Celsius off the northwestern coast of (a) the United
States and (b) South America. Black dots represent the Tara Pacific expedition stations.
Images retrieved from NOAA NMFS SWFSC ERD (NOAA, 2022).

(a)
(b)
66

Figure 2-S6
Total dust deposition, both dry and wet PM2.5 deposition, over the North Atlantic Ocean
from May through July 2016. Tara Pacific expedition stations plotted in black dots, with
stations that experienced >50 cm
-3
(diameter >2.5 µm) highlighted in red triangles. Satellite
images retrieved from NASA worldview (NASA, 2022).

67
Chapter 3
Delivery of metals and dissolved black carbon to the southern
California coastal ocean via aerosols and floodwaters following the
2017 Thomas Fire

This chapter was published open access in the Journal of Geophysical Research:
Biogeosciences by John Wiley & Sons, Inc. in 2021 and is reprinted here.

Kelly, R. L., Bian, X., Feakins, S. J., Fornace, K. L., Gunderson, T., Hawco, N. J., … John
S. G. (2021). Delivery of metals and dissolved black carbon to the southern California
coastal ocean via aerosols and floodwaters following the 2017 Thomas Fire. Journal of
Geophysical Research: Biogeosciences,126, e2020JG006117.
https://doi.org/10.1029/2020JG006117

Author Contributions
Joshua West, Seth John, and Sarah Feakins designed the overall study. Rachel Kelly,
Xiaopeng Bian, Sarah Feakins, Nicholas Hawco, Hengdi Liang, Suzanne Paulson, Paulina
Pinedo-González, Shun-Chung Yang, and Seth John collected samples during the Santa
Barbara cruise. Rachel Kelly and Joshua West collected samples from the Ventura River.
Seth John led the trace metal sample collection and analyses. Rachel Kelly, Xiaopeng
Bian, Nicholas Hawco, Paulina Pinedo-González, and Shun-Chung Yang processed and
analyzed all of the trace metal samples. Sarah Feakins and Troy Gunderson produced the
carbon isotope data. Kyrstin Fornace measured levoglucosan in the aerosol samples. Jutta
Niggemann led the dissolved organic carbon and dissolved black carbon analyses.
Suzanne Paulson was in charge of aerosol collection along with particulate mass and
black carbon aerosol analyses. Rachel Kelly measured chlorophyll a and organized the
manuscript. All authors contributed significantly to the manuscript.

Abstract
The Thomas Fire began on December 4, 2017 and burned 281,893 acres over a
40-day period in Ventura and Santa Barbara Counties, making it one of California’s most
destructive wildfires to date. A major rainstorm then caused a flash flood event, which led
to the containment of the fire. Both airborne ash from the fire and the runoff from the flash
flood entered into the Santa Barbara Basin (SBB). Here, we present the results from
68
aerosol, river, and seawater studies of black carbon and metal delivery to the SBB
associated with the fire and subsequent flash flood. On day 11 of the Thomas Fire,
aerosols sampled under the smoke plume were associated with high levels of PM2.5,
levoglucosan, and black carbon (average: 49 μg/m
3
, 1.05 μg/m
3
, and 14.93 μg/m
3
,
respectively) and aerosol metal concentrations were consistent with a forest fire signature.
Metal concentrations in SBB surface seawater were higher closer to the coastal perimeter
of the fire (including 2.22 nM Fe) than further off the coast, suggesting a dependence
on continental proximity rather than fire inputs. On days 37–40 of the fire, before, during,
and after the flash flood in the Ventura River, dissolved organic carbon, dissolved black
carbon, and dissolved metal concentrations were positively correlated with discharge
allowing us to estimate the input of fire products into the coastal ocean. We estimated
rapid aerosol delivery during the fire event to be the larger share of fire-derived metal
transport compared to runoff from the Ventura River during the flood event.

1. Introduction
Wildfires have increased in California in recent decades, linked to rising
temperatures coinciding with high offshore winds and later onsets of the rainy season
(Goss et al., 2020). Extreme fire-inducing conditions in late 2017 led the 2017–2018
Thomas Fire to become the largest wildfire in California’s history at the time (Cal Fire,
2018). The Thomas Fire began on December 4, 2017 when two separate fires ignited
south of Thomas Aquinas College in Ventura County. Several factors including dry
vegetation, unusually strong Santa Ana winds, and low humidity in the winter of 2017
69
contributed to the spread of the fire. In 2017, California experienced its seventh-wettest
year since 1901 (October 1, 2016 through September 30, 2017; USGS, 2020), which
promoted the growth of thick grasses and shrubs (Montecito Fire Protection District,
2018). This new vegetation dried out when record-breaking high temperatures were
reached in the summer and fall of 2017 (NOAA, 2020; Swain, 2017). The dry vegetation
then acted as a fuel for the fire that was spread rapidly by the Santa Ana winds (Montecito
Fire Protection District, 2018). Models and observations suggest that the Santa Ana winds
were as fast as 32 meters per second at the time of the first ignition, with speeds
continuing to increase in the days following (up to 35.3 meters per second; Fovell &
Gallagher, 2018). When winds weakened, much of the smoke hung in the air of the
affected areas causing the adjacent Santa Barbara County to issue an air quality warning
from December 5th through the 29th as the air quality reached unhealthy to hazardous
levels (PM2.5 concentrations >52.5 μg/m
3
; Santa Barbara County Public Health
Department, 2017). The fire lasted for 40 days and burned ∼281,893 acres (1,141 km
2
)
before being 100% contained on January 12, 2018.
A major rainstorm hit Ventura County, Santa Barbara County, and parts of Los
Angeles County, starting January 8, 2018, which helped to extinguish the fire. On the
morning of January 9 a half-inch of rain fell in 5 min in parts of the Santa Ynez Mountains
that had burned during the fire. This intense rain was sufficient to initiate debris flows in the
steep-sided fire scars and stream channels (Kean et al., 2019). Enhanced erosion
is common in steep landscapes post-fire due to the loss of vegetation that holds topsoil
and increases in the soil water repellency (Moody et al., 2013). The debris flows following
70
the Thomas Fire, which occurred mostly near the community of Montecito in Santa
Barbara County, damaged 408 residential buildings and caused 23 fatalities (Kean et al.,
2019).
The effects from fire-flood sequences, similar to the one associated with the
Thomas Fire, extend beyond the initial damage. As both smoke and flood waters from the
burned land make their way into the coastal ocean, they carry with them substances
released from the burned biomass. These substances include black carbon (BC; e.g.,
Hunsinger et al., 2008; Olivella et al., 2006; Simoneit & Elias, 2000; Wagner et al., 2018)
and trace metals (Burton et al., 2016; Pinedo-González et al., 2017; Stein et al., 2012;
Young & Jan, 1977 all in Southern California; see Abraham et al., 2017 for a global review).
Several studies have explored the ecological and carbon cycle consequences of the
pyrogenic chemical influx into terrestrial and freshwater ecosystems (e.g., Charette &
Prepas, 2003; Earl & Blinn, 2003; Oliveira-Filho et al., 2018). Fewer studies have focused
on the paired fluxes of organics and trace metals to the coastal ocean associated with
fires. Trace metal and black carbon fluxes are important, as they can either serve as a
nutrient or be toxic to marine organisms. For example, the delivery of a limiting nutrient
such as Fe can promote growth (Boyd et al., 2007; Moore et al., 2013), whereas the
delivery of Cu (Brand et al., 1983; Paytan et al., 2009), polycyclic aromatic hydrocarbons
(PAHs) associated with BC (Campos et al., 2012), or PAHs alongside metals (Brito et al.,
2017) can be toxic to marine organisms. As wildfires are projected to increase with
warming in the western United States (Abatzoglou & Williams, 2016) understanding their
effects on the coastal marine ecosystems will be increasingly important.
71
A starting point to understanding the effect of wildfires on marine ecosystems is
more complete characterization of the transport of black carbon (BC) and metals from fires
to the coastal ocean, and documentation of the marine response. Few, if any, studies have
considered paired BC and metal loadings, or have evaluated and quantitatively compared
both atmospheric and fluvial transport of these substances. With occasional exceptions
(e.g., Young & Jan, 1977), little work has connected atmospheric and fluvial
measurements with observations of changes in surface seawater composition and the
biological response. This study examines aerosols and river runoff associated with the
2017 Thomas Fire to test if fires and related flash flood events significantly affected black
carbon and metal delivery to the waters of the Santa Barbara Basin.

2. Materials and Methods
2.1 Field sampling
On December 14, 2017, day 10 of the Thomas Fire, aerosol and seawater samples
were collected from the Santa Barbara Basin aboard a small vessel that departed from
Oxnard, California and transected the southern portion of the basin (Figure 3-1c; Table 3-
1). The Santa Barbara Basin is the northern section of the Southern California Current
System (King & Barbeau, 2007; Thunell, 1998), with upwelling of cold salty waters bringing
macronutrients to the surface, and relatively high concentrations of chlorophyll a and
metals (Hayward & Venrick, 1998; Lynn & Simpson, 2008). Primary sources of trace
metals to the Santa Barbara Basin include atmospheric deposition, fluvial deposition,
upwelling (i.e., the continental shelf and bottom boundary layer), and the equatorward
72
transport of water masses from central California by the California Current system (King &
Barbeau, 2011; Thunell, 1998; Warrick et al., 2007). The cruise started ∼30 km south of
the Thomas Fire at the coast (Station 1) and proceeded northeast to a location directly
underneath the smoke plume on the day of sampling (Station 2). The transect then
approached the coast closer to the origin of the smoke plume stopping at about 15 and 7
km from the perimeter of the fire (Stations 3 and 4, respectively) and finally returned
southeast along the coast toward Oxnard (Stations 5 through 8).
For the aerosol samples, both total suspended particulate (TSP) and particles
smaller than 2.5 μm (PM2.5) were collected between stations (Table 3-1). TSP samples
were collected on pre-baked quartz fiber filters (QFFs) using two URG© cyclone inlets, and
the flow rates were controlled at 33.4 ± 0.1 liters per minute by rotameters (TSI
®
mass flow
meter) (Paulson et al., 2019). PM2.5 samples were collected on acid washed, preweighed
Teflon filters (PALL, 47 mm, 2.5 μm pore size) at 90 ± 2 liters per minute. Samples were
collected on average for ∼1 h per sample. Two full field blanks were created using the
same handling procedures as samples, but with the pump turned on for only 5 s. After
sampling, filters were stored in individual petri dishes in a freezer until analysis.
Surface seawater samples (∼2 m depth) were collected using a
polytetrafluorethylene (PTFE) Saint Gobain Furon™ bellows pump with low-density
polyethylene (LDPE) tubing (Table 3-1). The inlet tubing was attached to a PVC pole that
was positioned in front of the ship. Water was only collected as the vessel steamed into
the station to avoid collecting any water that had come into contact with the hull. Samples
were filtered in-line with a detachable 0.2 μm polyethersulfone membrane (Pall Acropak™
73
1500 capsule filter) and collected in either 50 mL centrifuge tubes or 20 L cubitainers. All
sampling materials were soaked in 10% hydrochloric acid and washed extensively with
clean water prior to the cruise. Samples were collected following trace metal clean
protocols at sea.
Flood water samples were taken from the Ventura River in Foster Park, Ventura
from January 8–12, 2018 (Table 3-1). Sampling started at 5:30 p.m. (PST) on January 8th
during a period of low flow before the flood. Flash flooding began at 8:00 a.m. (PST) on
January 9th, and sampling proceeded through flooding until January 12th. River samples
were collected with a plastic-coated rope and an all-plastic bucket from the center of the
stream. The bucket was thrown into the surface of the river, retrieved, and then the
contents were poured into both 250 mL bottles (high density polyethylene) and 10 L bags
(ethylene vinyl alcohol, food-grade).
2.2 Organic analyses

2.2.1 Aerosol samples
Three parameters for organics were measured on the aerosol filters: particulate
mass (PM2.5), levoglucosan, and black carbon (BC).
2.2.1.1 Particulate mass (PM2.5 )
PM2.5 is atmospheric particulate matter that has a diameter of <2.5 μm and is used
as a proxy for air pollution (Tian et al., 2009). Particulate mass was determined by using a
microbalance (1 μg precision, ME 5, Sartorius or similar) to weigh the Teflon filters (11
samples total). Prior to being weighed, the filters were equilibrated at 22-24
º
C and at 42-
74
44% relative humidity for one hour and were charge neutralized. The blanks had negligible
additional mass.
2.2.1.2 Levoglucosan (in TSP)
Levoglucosan (1,6-anhydro-β-D-glucopyranose) is a product of cellulose pyrolysis
and a tracer of biomass burning (Simoneit et al., 1999). Six samples for levoglucosan
analyses were collected on QFF filters. Levoglucosan was measured with a method
adapted from the California Air Resources Board (2014). A portion of each filter was spiked
with
13
C6-levoglucosan (Cambridge Isotope Laboratories) and extracted by ultrasonication
in acetonitrile (40
º
C, 60 min). Extracts were filtered to remove any particles, and an aliquot
of each extract was derivatized with Tri-Sil TBT (TMSI:BSA:TMCS, Thermo Fisher
Scientific) at 70
º
C for 60 min. Derivatized extracts were analyzed by GC-MS (Thermo Trace
1310/ISQ LT) using a simultaneous selective ion monitoring (SIM)/full scan method.
Levoglucosan was quantified by calibrating mass spectrometry analyses against
authenticated levoglucosan standards (Sigma Aldrich, Carbosynth Limited) derivatized in
parallel to samples. A method blank (an unexposed filter spiked with
13
C-levoglucosan)
was also extracted and analyzed to assess any blank contribution to reported results.
2.2.1.3 Black carbon (in PM2.5 and TSP)
Black carbon (BC) is an organic byproduct of the incomplete combustion of organic
matter (e.g. fossil fuels, biofuel, and biomass; Ni et al., 2014). In order to measure the
amount of black carbon, aerosol filters (11 PM2.5 samples and 6 TSP samples) optical
absorption at 880 nm was measured using an OT21 dual-wavelength optical
transmissometer (Magee Scientific Corporation). Each filter was backed with a quartz
75
diffuser backing (Pallflex Fiberfilm) in order to provide an even distribution of light to the
detector (Kuang et al., 2017). Light absorption at 880 nm is proportional to the
concentration of elemental carbon, as described by equations presented in Hansen et al.
(1983). The term “black carbon” is usually used for material measured optically; elemental
carbon is measured chemically, although both types of measurements are operationally
defined and should be regarded as estimates. The instrument measures absorption for a
reference and sample filter simultaneously and subtracts the reference from the sample.
Here, one of the field blank filters was used as the reference.
2.2.2 Surface seawater samples
Four parameters for organics were measured on the surface seawater samples:
carbon isotopes (δ
13
C), chlorophyll a, dissolved organic carbon (DOC) and dissolved black
carbon (DBC).
2.2.2.1 Carbon isotopes
δ
13
C was analyzed to determine the main sources of carbon found in the coastal
waters of the Santa Barbara Basin at the time of sampling. Surface seawater particulate
samples (8 total) were collected on QFFs for carbon isotope analysis using the identical
sample collection and analytical procedures used in Knapp et al. (2016) for δ
15
N. The QFFs
were dried at 60
º
C and then pelletized in tin capsules before analysis of stable isotopic
composition (δ
13
C) by continuous-flow isotope ratio mass spectrometry. A Costech ECS
4010 elemental analyzer interfaced to a Micromass Isoprime mass spectrometer was used
to measure δ
13
C. Due to varying amounts of material on each filter, 1/8 of a filter was
analyzed for each of the filter samples for Stations 1-6, and 1/4 of a filter for Stations 7 and
76
8. Duplicate filter fractions yielded an average standard deviation of ± 0.20‰. Standards
were routinely analyzed during sample runs, which included acetanilide for N and C
elemental mass and glycine for δ
13
C.
2.2.2.2 Chlorophyll a
Chlorophyll a was measured to estimate phytoplankton biomass in the surface
seawater samples. Surface seawater was collected in 50 mL tubes for chlorophyll a
analysis. Upon returning to lab, the 8 seawater samples were filtered onto GF/F filters,
which were extracted with 90% (V/V) acetone and kept in the dark at -4
º
C for 24 hours
(Andersen, 2005). Chlorophyll a concentrations were then measured by fluorometry
calibrated with a pure chlorophyll a standard (Andersen, 2005).
2.2.2.3 Dissolved organic carbon and dissolved black carbon
Dissolved organic carbon (DOC) concentrations were quantified on 8 frozen filtrate
samples (0.7 µm, GFF; filtrate stored in amber glass bottles) after thawing and acidification
to pH 2 (HCl, p.a.). Analysis was done using high-temperature combustion on a Shimadzu
TOC-VCPH total organic carbon analyzer. Accuracy was monitored by replicate analyses
of Deep Atlantic Seawater Reference material (DSR, D.A. Hansell, University of Miami,
Miami, FL), which deviated on average <5%. Deviation of analytical triplicates of samples
was 4 ± 3% on average (range 0%–10%).
Dissolved black carbon (DBC) concentrations were determined following the
benzene polycarboxylic acids (BPCA) method (Dittmar et al., 2008), with the modifications
described in Stubbins et al. (2012). In brief, aliquots of methanolic solid-phase extracts
(Dittmar et al., 2008) were dried, re-dissolved in nitric acid (65%, p.a.) and kept at 170
º
C
77
for 9 hours. After cooling and evaporation of the acid, the 8 samples were dissolved in a
phosphate buffer for separation and quantification of benzene polycarboxylic acids on a
Waters Acquity UPLC (ultrahigh‐performance liquid chromatography) equipped with a
photodiode array detector. Concentrations of DBC were calculated from the
concentrations of benzene penta- and hexacarboxylic acids based on their power‐function
relationship (Stubbins et al., 2012). Procedural blanks did not yield any detectable DBC.
Half of the samples had enough material to prepare duplicates, three samples were
analyzed in quadruplicates and on average replicate analyses deviated 3 ± 2% (range 0%–
6%).
2.2.3 River samples
2.2.3.1 Dissolved organic carbon and dissolved black carbon
The analytical methods used for the 9 river water samples were the same as applied
for “Surface Seawater Samples (Dissolved Organic Carbon and Dissolved Black Carbon)”.
2.3 Trace metal analyses
All trace metal analyses were performed at the University of Southern California in a
class-100 clean room. After preparation of different sample types as described below, final
concentrations in all samples were measured on an Element 2 inductively coupled plasma
mass spectrometer (ICP-MS, Thermo Fisher Scientific).
2.3.1 Aerosol samples
Two separate trace metal analyses were done using the Teflon filters: an
instantaneous leach and a total digestion. Each analysis was done on a total of 11 aerosol
samples.
78
2.3.1.1 Instantaneous leach (soluble concentrations)
In order to understand the magnitude of aerosol deposition of metals into the ocean
and the subsequent dissolution of these metals into the surface waters, we performed
instantaneous leaching experiments (Buck et al., 2006). Teflon filters were exposed to
ultrapure deionized water for 1 minute. The water was collected after exposure and metal
concentrations were measured. These metal concentrations were interpreted as the
“instantaneous” soluble (i.e., bioavailable) fraction. Each Teflon filter was placed on a 47
mm diameter, Millipore Sterifil Aseptic Filtration System attached to a vacuum pump. 100
mL of Milli-Q water was poured into the filter holder and the filtrate was collected. The
filtrate was stored in two 50 mL acid-cleaned centrifuge tubes, which were then acidified
to a pH of 2 with distilled HCl and stored for at least one month. The filtration rig was
rinsed 3 times with Milli-Q between samples in order to prevent cross-contamination.
436 μL subsamples of the filtrate from the instantaneous leaching experiment were
collected and an In standard solution was added to each subsample to achieve a final
concentration of 1 ppb In as an internal standard. The samples were then amended with
distilled concentrated HNO3 to achieve a final sample matrix containing 2% HNO3, followed
by ICPMS analysis. Concentrations for all elements were qualified relative to a 10 ppb
multielement standard.
2.3.1.2 Total digestion (total concentrations)
Total digestions of the remaining samples were done to measure the total metal
concentrations. After the instantaneous leaching, particles on each of the Teflon filters
were totally digested in perfluoroalkoxy (PFA) vials using the Piranha digestion method
79
(Ohnemus et al., 2014), except that instead of a 3 parts concentrated sulfuric acid to 1
part concentrated hydrogen peroxide solution a 2 parts sulfuric acid to 1 part hydrogen
peroxide solution was used. Digested samples were dried down and resuspended with 1
mL 2% HNO3. 0.1 mL subsamples of the digest were diluted to a final volume of 1 mL with
2% HNO3 which included 1 ppb In as an internal standard, followed by ICPMS analysis.
Concentrations for all elements were quantified relative to a 10 ppb multielement standard.
Total metal concentrations were used to calculate the total mass of metals
mobilized through atmospheric transport during the Thomas Fire. First, the amount of
burnt biomass produced by the fire was calculated using the following equation:

281,893 acres burnt *
1 hectare
2.47 acre
*
27 tons of fuel
1 hectare
= 3.1x10
6
tons of burnt biomass (1)

The number of acres burnt and the amount of fuel burned per hectare were values found
in Cal Fire (2018) and Van Leeuwen et al. (2014), respectively. Then the amount of
pyrogenic aerosols produced was calculated:

3.1x10
6
tons of burnt biomass *
1,000 kg
1 ton
*
5.46 g of PM2.5 aerosol
1 kg burnt biomass
= 1.68x10
10
g aerosol (2)

The amount of PM2.5 aerosol per amount of burnt biomass is from Hosseini et al. (2013).
Finally, the total mass of atmospheric metals was calculated:

1.68x10
10
g aerosol *
1 metric ton
1,000,000 g
*
x TE ppm
1,000,000
= y tons of TE in aerosol (3)
80

where x represents the average concentration of each metal measured in the aerosol
samples which is used to calculate y, the total amount of each metal produced by the
Thomas Fire.
2.3.2 Surface seawater samples
2.3.2.1 Dissolved concentrations
Filtered (<0.2 µm) surface seawater samples (8 total) were acidified to pH 2 using
HCl directly after the cruise and stored for ~2 months. Metal concentration analyses were
identical to those used in Hawco et al. (2020). 15 mL subsamples of the filtered seawater
were collected and 50 μL of an isotope spike (which includes the following:
57
Fe,
62
Ni,
65
Cu,
67
Zn,
207
Pb,
110
Cd) was added to each of the 15 mL tubes. Spiked samples were then
extracted using a seaFAST preconcentration system (manufactured by Elemental
Scientific; Lagerström et al., 2013), the instrument pushed each of the samples through a
column with Nobias resin PA-1 (Sohrin et al., 2008), and the preconcentrated sample was
eluted into 0.5 mL 1 M HNO3 which included 1 ppb In, followed by ICPMS analysis.
Concentrations for all elements except for Mn and Co were derived by using an isotope
dilution method (Lee et al., 2011). Mn and Co concentrations were quantified relative to a
10 ppb multielement standard.
2.3.3. River samples
2.3.3.1 Dissolved and soluble concentrations
13 river water samples were first filtered through a 0.2 μm Acropak, subsampled,
and then filtered through 0.02 μm Anopore
®
membrane filters to get the soluble
81
concentrations. 0.1 mL subsamples of the 0.2 μm and 0.02 μm filtrate were collected and
diluted to a final volume of 1 mL with 2% HNO3 which included 1 ppb In, followed by
ICPMS analysis. Concentrations for all elements were quantified relative to a 10 ppb
multielement standard.
Both total dissolved (<0.2 μm) and soluble (<0.02 μm) fractions were measured.
The colloidal concentrations (0.02-0.2 μm) were calculated by subtracting soluble
concentrations from total dissolved concentrations in each sample. Percent soluble and
colloidal fractions were then calculated by dividing the respective concentrations by the
total dissolved concentrations.
Total dissolved metal concentrations were used to calculate the total mass of
metals mobilized through fluvial transport during the January 2018 flash flood event. First,
fluvial metal concentrations for each day of the flash flood event were calculated using the
equation for the power trendline of the concentration-discharge (C-Q) relationship of each
metal in log-log space:

y = c * x
b
(4)

where y is the metal concentration (mol/L), ln(c) is the y-intercept, x is discharge (m
3
/s),
and b is the slope of the line. Then using the calculated metal concentration for each day
(y) the metal flux was calculated using the following equation:

y * Q * ma = jd (5)
82

where Q is the measured discharge (L/day) for each day from USGS (2020b), ma is the
atomic mass of each metal, and jd is the daily metal flux (mol/day). The total mass of each
metal (metric tons) transported by runoff from January 8 to 12, 2018 was calculated by
summing the calculated daily fluxes for each of the 5-days.
2.4 Other analyses
2.4.1 Nutrients
2.4.1.1 Dissolved concentrations
About 40 mL of each 0.2 μm filtered, surface seawater sample (8 total) was frozen
after collection and sent to the Marine Science Institute at the University of California,
Santa Barbara (UCSB) for flow injection analysis. Measurements for phosphate, silicate,
ammonia, and nitrite plus nitrate were taken all simultaneously on a QuikChem 8500 Series
2 flow injection analyzer (manufactured by Lachat Instruments). Flow injection analysis (FIA)
consisted of a continuously flowing reagent stream, reaction manifolds, and flow-through
detectors (Worsfold et al., 2013). Nutrient concentrations were all reported in the
micromolar range with a precision of ± 5%.

3. Results
3.1 Organic analyses
3.1.1 Aerosol samples
PM2.5, levoglucosan, and black carbon concentrations were measured in the
aerosol samples. The measured PM2.5 values ranged from 33 to 65 μg/m
3
(Figure 3-2a).
83
The PM2.5 concentrations agree well with the offshore projections by the Copernicus
Atmosphere Monitoring Service’s air quality forecast for December 14, 2017 (the day of
the cruise), which ranged from 40 to 70 μg/m
3
(Copernicus Atmosphere Monitoring
Service, 2020). The levoglucosan concentrations in the TSP samples ranged from 0.33 to
1.55 µg/m
3
(Figure 3-2b). The concentrations of BC in the PM2.5 (BCPM2.5) samples ranged
from 5.57 to 16.44 μg/m
3
(Figure 3-2a) and BC in the TSP (BCTSP) samples ranged from
10.36 to 19.84 μg/m
3
(Figure 3-2b). PM2.5 concentrations generally increased at stations
closer to the perimeter of the fire, whereas BCPM2.5 concentrations showed no obvious
pattern (Figure 3-2a). In contrast, both levoglucosan (in TSP) and BCTSP concentrations
displayed strong trends as both concentrations steadily increased as the cruise
approached the coast (Figure 3-2b). Overall, the highest concentrations for PM2.5,
levoglucosan, and BCTSP were all measured near Station 4, the station closest to the
perimeter of the fire.
3.1.2 Surface seawater samples
The 0.2 μm filtered surface seawater samples were analyzed for both dissolved
organic carbon (DOC) and dissolved black carbon (DBC). At the surface of the Santa
Barbara Basin, DOC concentrations ranged from 0.392 to 0.846 mM and DBC
concentrations ranged from 0.2 to 0.7 μM (Figure 3-S2c). The highest DOC concentration
was found at Station 4 (Figure 3-S2c), but no obvious trend was seen in the
concentrations at the other 7 stations. Like levoglucosan and BCTSP, DBC concentrations
in surface seawater steadily increased as the cruise approached the perimeter of the fire
(Figure 3-2b).
84
Bulk surface seawater samples were filtered and the particulates were analyzed for
both δ
13
C and chlorophyll a. In the seawater particulates, δ
13
C values ranged from -24.50
to -21.14‰ (Figure 3-2c). Concentrations of chlorophyll a ranged from 0.16 to 1.05
µgChla/L (Figure 3-2c). Both Chl a and δ
13
C values decreased in the more southern
stations (Stations 1, 2, and 8) and increased farther north (Station 4) following the same
trend seen in the PM2.5, levoglucosan, BCTSP, and DBC concentrations.
3.1.3 River samples
River water samples from the Ventura River were analyzed for dissolved organic
carbon (DOC) and dissolved black carbon (DBC). DOC concentrations ranged from 181 to
7608 µM and DBC concentrations ranged from 7 to 183.1 µM (Figure 3-5c).
Concentrations of both DOC and DBC increased along with rising water stage (Figure 3-
5c) and had positive relationships with discharge (Figure 3-S4).
3.2 Trace metal analyses
3.2.1 Aerosol samples
In the aerosol samples, both soluble and total trace metal concentrations were
measured. The trace metals analyzed included Fe, Zn, Cd, Ni, Cu, Pb, Mn, and Co.
Solubilities decreased in the following order, Zn ≈ Cd > Pb > Mn > Cu > Co > Ni > Fe,
which follows a trend found in previous similar studies (Mahowald et al., 2018; Figure 3-3).
Total aerosol metal concentrations are reported both as a fraction of the aerosol mass (in
ppm; Figure 3-S1), and were converted into concentrations in the atmosphere (pmol/m
3
)
by normalizing by the total volume of air sampled (Figure 3-4).
3.2.2 Surface Seawater Samples
85
Filtered (<0.2 µm) surface seawater samples were measured for dissolved metal
concentrations (Fe, Zn, Cd, Ni, Cu, Pb, Mn, and Co; Figure 3-S2a & Table 3-2). The
average metal concentrations of samples taken from all 8 stations were 2.42 ± 1.94 nM
Fe, 0.88 ± 0.62 nM Zn, 45.90 ± 17.46 pM Cd, 3.07 ± 0.11 nM Ni, 1.16 ± 0.13 nM Cu,
25.35 ± 1.86 pM Pb, 3.06 ± 0.98 nM Mn, and 46.69 ± 13.49 pM Co (Table 3-2). The
lowest concentrations for all of the metals were found at Stations 1 and 8, except for Pb.
Surface seawater samples were collected on the 11th day of the fire, according to NASA
worldview satellite images Stations 1 and 8 seem to be the least affected by the smoke
produced from the fire in 9 of those previous 11 days (Figure 3-S2). The highest
concentrations for all of the metals were found at Stations 3 through 7. Trace metal
concentrations follow the same trend found in the organic analyses, where the
concentrations increase as the cruise approaches the perimeter of the fire on the coast.
3.2.3 River samples
In the Ventura River, water samples were taken before, during, and after the
January 2018 flash flood event and were analyzed for trace metals (Figure 3-6). During the
flooding, river water dissolved Mn concentrations increased by over two orders of
magnitude in comparison to its concentrations before the rainstorm. Fe, Pb, and Co
concentrations increased at least one order magnitude. Zn and Ni concentrations
approximately doubled during flooding. Cu concentrations did not increase by much
during flooding, but doubled right after flooding stopped. The metal concentrations in the
river water samples increased with rising water stage (Figure 3-6) and, except for Cd, all
86
had positive relationships with discharge (Figure 3-S4), similar to DOC and DBC
concentrations.
3.3 Other analyses
3.3.1 Nutrients
Nutrient concentrations were measured in the dissolved (<0.2 µM) surface seawater
samples (Figure 3-S2b). The average nutrient concentrations in the samples were 0.14 ±
0.06 µM phosphate, 1.49 ± 0.43 µM silicate, 0.69 ± 0.56 µM ammonia, and 0.22 ± 0.24
µM nitrate. These nutrient concentrations are close to the average macronutrient
concentrations found in the Santa Barbara Basin during the winter: 0.33 ± 0.12 μM
phosphate, 1.75 μM ± 0.7 silicate, 0.19 ± 0.22 μM ammonia, and 0.21 ± 0.18 μM nitrate
(data from 2014 to 2018; CalCOFI, 2018). No obvious trend in concentration distribution
was seen for any of the nutrients, unlike the organic and metal distributions.

4. Discussion
4.1 Signatures of biomass burning in aerosols and delivery of BC to the surface ocean
In the aerosol samples we determined the concentrations of multiple parameters
which are commonly used as proxies for biomass burning, including PM2.5, levoglucosan,
and black carbon (Ni et al., 2014; Simoneit et al., 1999, Tian et al., 2009). From 2015
through 2018, the average PM2.5 concentration for Ventura and Santa Barbara Counties
was 8.4 ± 1.4 μg/m
3
(California Air Resources Board, 2020). During the Thomas Fire, the
PM2.5 concentrations (33 to 65 μg/m
3
) above the Santa Barbara Basin were 4 to 8 times
higher than the 2015-2018 average and were classified as moderate to unhealthy
87
according to the United States air quality index (AQI; Figure 3-2a; EPA, 2012). Across the
transect, levoglucosan concentrations (0.33 to 1.55 µg/m
3
; Figure 3-2b) were three orders
of magnitude higher than normal, non-fire conditions (0.0012 µg/m
3
in TSP from the
California Coast, Fu et al., 2011) (Figure 3-2b). Similarly, reports from other wildfires found
high levoglucosan concentrations (2.70 ± 1.6 µg/m
3
) compared to their background
(during non-fire conditions) reports of 0.06 ± 0.05 µg/m
3
(de Oliveira Alves et al., 2015; Pio
et al., 2008; Ward et al., 2006). Additionally, levoglucosan to PM2.5 ratios have been shown
to vary substantially with fuel types (Hosseini et al., 2013). The average levoglucosan to
PM2.5 ratio (2.2 ± 1.1 LG/PM2.5%) measured on our cruise is close to the results from
chaparral test burns (1.64 ± 1.18 LG/PM2.5%; Hosseini et al., 2013). Here, our closest
sampling station to the perimeter of the fire was ~7 km west of the fire (Station 4, Figure 3-
2) and the plume was visibly higher in the atmosphere than the sea level aerosol sampling
(Figure 3-1d). As we approached the fire, BC concentrations in atmospheric TSP steadily
increased together with levoglucosan (from Stations 1 to 4, Figure 3-2b), which is
consistent with wildfires being known as one of the major sources of black carbon to the
environment (Kang et al., 2014; Santoso et al., 2013). Atmospheric PM2.5 BC sampled
during the cruise (5.57 to 16.44 μg/m
3
; Figure 3-2a) was over three times greater than the
annual-average atmospheric PM2.5 BC found in the Los Angeles Basin from 2003 to 2011
(1.62 μg/m
3
, McDonald et al., 2015). The PM2.5 BC from this study also exceeded the
aerosol BC concentrations reported from three other forest fires, which ranged from 2.4 to
6.04 µg/m
3
(Kang et al., 2014 – Canada ; Santoso et al., 2013 – Indonesia). Overall, we
88
found that the aerosol concentrations of PM2.5, levoglucosan, and black carbon increased
with proximity to the fire under the instantaneous position of the smoke plume.
In the surface waters of the Santa Barbara Basin, dissolved black carbon
concentrations also increased in waters sampled below the smoke plume (Stations 2
through 7). The slightly elevated DBC in seawater at the stations beneath where aerosol
PM2.5, levoglucosan, and BC were highest (Figure 3-2) suggest atmospheric deposition of
soluble BC into the basin due to the Thomas Fire. The DBC concentrations in our study
(0.55-0.70 μM) were higher than the usual values found in the Pacific open ocean (0.08-
0.33 μM; Coppola & Druffel, 2016; Ziolkowski & Druffel, 2010), but were much lower than
the minimum values found in other coastal waters, for example on the coast of the East
China Sea (2.58 μM; Wang et al., 2016). Freshly produced BC is thought to be largely
insoluble (Wagner et al., 2017), which may explain why the DBC concentrations at the
surface of the Santa Barbara Basin were only slightly elevated. Similarly, a study in Halong
Bay, Vietnam found that the atmospheric deposition of BC into the coastal waters did not
result in an increase of DBC (Mari et al., 2017). Atmospheric BC becomes soluble through
oxidative processes (e.g., weathering, microbial oxidation, photooxidation) that introduce
O-containing functional groups (e.g., carboxyl groups), making the molecules more polar
(i.e., soluble) (Wagner et al., 2017). Therefore, it is thought that “older” BC is more soluble
(Wagner et al., 2017). Since sampling occurred on day 11 of the fire, it is also possible that
in the days before sampling any fire-derived DBC in the surface waters may have been
removed by adsorption onto sinking particles (Coppola et al., 2014) and/or by
photooxidation (Stubbins et al., 2012). Overall, the samples from our study suggest the
89
smoke contributed minimally to the surface seawater DBC inventory at this site.
Regardless, we expected that we might see larger atmospheric deposition of soluble
metals to these waters from the fire-derived particulates.
4.2 Atmospheric deposition of trace metals into the Santa Barbara Basin
Our sampled aerosol metal concentrations were normalized to the volume of air
collected and compared to the average metal concentrations for both North Pacific air
(11°N and 162°E; Duce et al., 1983) and Los Angeles air (Saffari et al., 2013) (Figure 3-4).
Except for Pb and Co, the median metal concentrations of the air was at least one order of
magnitude higher than the average metal concentrations for the North Pacific and are
similar to the average metal concentrations for Los Angeles. Thus, our observations
suggest that the Thomas Fire was responsible for elevated metal concentrations observed
in the air above the Santa Barbara Basin.
Trace-metal concentrations in our aerosols from the Thomas Fire were compared
to those found in wood (Butkus & Baltrėnaitė, 2007; Nicewicz & Szczepkowski, 2008;
Queirolo et al., 1990), fly ash from wood burning (Koukouzas et al., 2007; Pitman, 2006;
Steenari et al., 1999; Świetlik et al., 2013), ash from two previous California wildfires
(Odigie & Flegal, 2011, 2014), and Los Angeles aerosols (Saffari et al., 2013) (Figure 3-S1).
The degree of elemental enrichment during burning depends on many factors including the
volatility of the elements and the heat of the fire (Bodí et al., 2014). Still, most metal
concentrations in aerosols measured here are similar to previously-reported fly ash values;
however Fe, Mn, and Pb concentrations were lower than reported fly-ash values, and
more similar to concentrations reported for wood.
90
The relative concentrations and solubility of different trace metals in aerosols is
often dependent on the source of the aerosols (i.e., mineral, volcanic, combustion).
Aerosols from mineral sources (i.e., dust) are typically enriched with Fe, Mn, Al, and Ti,
volcanic aerosols are enriched with Cd, and combustion aerosols (i.e., fires) have high
concentrations of Zn, Pb, and Cu (Desboeufs et al., 2005; Mahowald et al., 2018). It is
therefore not surprising that out of all the metals measured, Zn concentrations in the air
were the most elevated compared to samples from non-smoky environments (331 and 8
times greater than North Pacific Ocean and Los Angeles values, respectively) (Figure 3-4).
Metals in aerosols produced from combustion tend to be more soluble than metals
associated with dust (Desboeufs et al., 2005; Mahowald et al., 2018; Sedwick et al.,
2007). For example, the concentrations of Fe in smoke are 1–2 orders of magnitude lower
than in dust (Mahowald et al., 2018), but Fe associated with biomass burning may be up
to 10% more bioavailable than Fe associated with dust (Luo et al., 2008). Therefore, we
expected that the Fe and other metals associated with the smoke from the Thomas Fire
likely would have a more notable effect on the marine ecosystem than typical dust aerosol
deposition.
Evidence of metal deposition into basin waters from smoke aerosols was sought by
comparing the trace metal concentrations of the Santa Barbara Basin surface seawater
collected from our cruise during the Thomas Fire with a surface seawater sample (GSC
reference sample) taken from the basin in 2009 when there was no fire. We found that the
metal concentrations during the fire were similar to the background concentrations
measured in 2009 (Figure 3-S2a, Table 3-2). Only Fe and Mn showed higher
91
concentrations during our cruise than in 2009. Average Cd concentrations were almost a
whole order of magnitude lower during our cruise than in 2009. However, it is important to
note that the GSC reference sample was likely collected during an upwelling event,
explaining why the Cd concentration was so high (Segovia-Zavala et al., 1998; Van Geen &
Husby, 1996). In any case, the similarity of our data to measurements from 2009 suggests
that metals associated with the smoke were not delivered to the surface waters in the
Santa Barbara Basin in measurable amounts, or at least that this delivery did not exceed
what would be expected from an upwelling event. It is possible that at the time of sampling
most of the metals from the Thomas Fire in the surface seawater were still in the
particulate size fraction (>0.2 μm), which was not sampled for in this study. Additionally,
aerosol flux to the ocean is largely controlled by wet deposition (particularly for finer
combustion aerosols; Hand et al., 2004; Zhang et al., 2007) and the sampling in this study
took place during a dry period; therefore we might have expected to see a pulse of aerosol
deposition when the rain first began in January 2018. The absence of a signal of fire-
derived inputs in the dissolved seawater metals is consistent with the lack of dissolved
black carbon beneath the smoke plume and with conclusions from prior work on fallout of
metals to the oceans after fires in southern California (Young & Jan, 1977).
4.3 Impacts of BC and metals from pyrogenic aerosols on organisms in the basin
The Thomas Fire did not appear to have an immediate large impact on organic
matter cycling in the Santa Barbara Basin based on δ
13
C, chlorophyll a, or dissolved
organic carbon (DOC) concentrations in surface seawater (Figure 3-2). The highest Chl a
concentrations were found nearest to the coast, and the measured δ
13
C values of
92
seawater particulates were as high as -21.14‰, more consistent with expected values for
marine phytoplankton than for terrestrial carbon (Degens, 1969). Away from the fire
(Stations 1, 2, and 8; Figure 3-2c) marine productivity (Chl a) declined as δ
13
C values
decreased to as low as -24.50‰, such that an approximately 50:50 mixture of marine
phytoplankton and terrestrial C3 plant inputs (perhaps from the smoke or terrestrial
outflow) can be inferred. With the expected increase in trace metal and black carbon fluxes
to the coastal ocean due to the fire we expected marine primary productivity to be
affected, but the chlorophyll a values measured in our samples (0.16 to 1.05 µg/L) were
not significantly different from average Chl a values for the Santa Barbara Basin (~1 µg/L;
CalCOFI, 2018). Thus, we do not see evidence of a fire-response bloom or bust, and
instead the patterns are consistent with normal coastal productivity. In the dissolved
organic component, we found no clear relationship between DOC and proximity to the fire,
though the DOC concentrations were about six times higher (average: 564 ± 154 μmol/L)
than the average values found in the Santa Barbara Basin (~85 μmol/L; Alldredge, 2000;
Wear et al., 2015; Figure 3-S2c). These elevated concentrations of DOC might be
attributed to atmospheric smoke deposition from the Thomas Fire, perhaps integrated
over the eleven previous days of burning, however the elevated DOC cannot be definitively
attributed to the Thomas Fire in the absence of elevated DBC, levoglucosan, or other
tracers which are diagnostic of biomass burning. Similarly, we cannot attribute the
elevated DOC to increased biological productivity stimulated by the Thomas Fire in the
absence of elevated macronutrient or trace-metal concentrations. Again, it is possible that
the flux of aerosols to the surface waters did not increase until the rainstorm that occurred
93
after sampling in January 2018, in which case we would not have seen an immediate
impact on the organic matter in the Santa Barbara Basin due to the Thomas Fire since our
sampling took place before these rains.
Although at the time of sampling we found no detectable evidence that the metals
associated with the smoke from the Thomas Fire directly impacted the metal inventory of
the Santa Barbara Basin, it is possible that the metals in the smoke may have impacted
the waters southwest of the basin (beyond where we collected samples). Based on a
satellite image of the smoke plume taken on December 5, 2017 and a HYSPLIT model
showing the trajectory of the wildfire smoke from December 5–14, 2017, the smoke from
the fire extended at least ~197 km (Figure 3-1b) and possibly as much as ~1,000 km
(Figure 3-1a) southwest. Thus the smoke plume extended past the coast through the
transition zone of the southern California Current System and into the offshore zone
(easternmost reach of the central north Pacific gyre) (King & Barbeau, 2011). The transition
zone is characterized as having lower iron concentrations (0.62 ±0.48 nM Fe) and medium
macronutrient concentrations, whereas the offshore zone is characterized as having both
low iron (0.18 ±0.08 nM Fe) and low macronutrient concentrations (King & Barbeau, 2011).
If there was any atmospheric deposition of metals from the smoke plume into these
waters, the input of soluble metals to both the transition and offshore zones could possibly
have stimulated primary production, with those effects being more likely felt in the distant
regions beyond the reach of our sampling.

94
4.4 Increased loading of dissolved metals in river water after the Thomas Fire
Rivers are another means for transport from burned areas to the oceans.
Substantial increases in metal loading to surface waters have been documented after
wildfires in southern California (Burton et al., 2016; Stein et al., 2012) and elsewhere
(Abraham et al., 2017). These increases have typically been quantified by comparing
adjacent catchments, one affected by recent fire and the other not (Burke et al., 2013;
Gallaher & Koch, 2004; Pinedo-González et al., 2017; Yoon & Stein, 2008). An unburned
catchment was not available in this study. Instead, the Ventura River baseflow samples
collected before the flash flood event are used to reflect the chemical signature of the river
water prior to the fire. This approach assumes there was minimal delivery of fire-associated
material to the river prior to rainstorm, which is reasonable if most of the baseflow was
from groundwater discharge of “old” water (Hornberger et al., 1998).
Dissolved trace metal concentrations from the Ventura River samples collected
before the flooding event (control) and during flooding (burned) were compared to
previously-measured metal concentrations in paired catchment studies that looked at
runoff from control versus burned landscapes in southern California (Figure 3-S3). The
dissolved concentrations of all trace metals measured in this study except Ni were
substantially lower compared to those found in other work in southern California (Burke et
al., 2013; Gallaher & Koch, 2004; Pinedo-González et al., 2017; Yoon & Stein, 2008). This
difference may be explained by the location of the Ventura River catchment farther from
sources of anthropogenic pollution such as the Los Angeles Basin. Lead concentrations in
particular were much lower, which is expected since the primary source of this metal to
95
the ecosystems in this region is anthropogenic aerosols (Odigie & Flegal, 2014).
Nonetheless, the data from this study did show increases in river water metal
concentrations after the fire-flood event, including for Pb. These findings are consistent
with prior studies in Southern California which showed varying degrees of metal
enrichment after fires and may be a product of the large increase in river water discharge
(pre-flood average: 0.12 ft
3
/s vs. flood average: 1,978 ft
3
/s) during the rainstorm that
followed the Thomas Fire (Figure 3-S3).
The partitioning of the total dissolved trace metal concentrations showed that the
colloidal fraction for most of the metals (except Fe and Cu) decreased in the fire-affected
conditions (“Flood” and “Post-Flood”; Figure 3-S5a), and conversely that the soluble
fraction of these metals increased. These results support the observations made by
Pinedo-González et al. (2017) in the San Gabriel Mountains that metals in fire-affected
runoff may be more soluble (i.e., bioavailable) than metals in non-fire associated runoff. The
mechanisms that cause the potential size partitioning of metals due to fires are still
unknown, but one proposed mechanism is that the high temperatures during combustion
break down organic ligands which could bind metals, resulting in an increase in metal
solubility (Gallaher & Koch, 2004).
4.5 Hydrologic control on fluvial mobilization of trace metals, DBC, and DOC
Changes in dissolved concentrations as a function of discharge in rivers
(concentration-discharge, or C-Q, relationships) can reflect sources and transformation of
solutes (e.g., Godsey et al., 2009), as well as revealing how hydrologic conditions influence
fluxes to the oceans. In the Ventura River data from this study, the C-Q relationships for
96
DOC and DBC were both positive in log-log space with slopes of 0.29 and 0.32,
respectively (Figure 3-S4). These increases in concentration with discharge are consistent
with leaching from surface soil horizons and may reflect increased flow through these
horizons at higher discharge (Lohse et al., 2009; Marques et al., 2017; Perdrial et al.,
2014). The positive C-Q relationships seem to reflect a strong hydrological control on the
mobilization of DBC and DOC following wildfires: since the flux is a product of water flux
multiplied by concentration, positive C-Q relationships magnify the effect of discharge on
total export. This effect is well known for DOC (e.g., Raymond et al., 2016). Other studies
have shown that most DBC transport also occurs at high discharge, including in tropical
forests (Dittmar et al., 2012; Marques et al., 2017), in the arctic (Myers-Pigg et al., 2017;
Stubbins et al., 2015), and in the Rocky Mountains of Colorado (Wagner et al., 2015), with
a somewhat less pronounced relationship in the southeastern US (Roebuck et al., 2018)
and southeast China (Bao et al., 2019) (Figure 3-7). Here we document this phenomenon
in the Mediterranean climate of southern California directly after a wildfire, emphasizing that
across a wide variety of climatic regimes, hydrologic conditions are important in
determining how much DBC is mobilized from soils and exported in rivers following fires.
However, we cannot be certain as to what extent the positive C-Q relationship of DBC and
DOC in our study is affected by wildfires due to the lack of DBC and DOC data from the
Ventura River in non-fire conditions.
In more detail, when compared to the C-Q relationships for DBC from other
studies, the Ventura River samples yield a steeper slope than observed in other regions in
the US, but a slightly shallower slope than found for Arctic rivers (Figure 3-7) — suggesting
97
a relatively strong hydrological control on DBC transport in the Mediterranean climate of
southern California compared to other regions. In addition, for a given discharge, DBC
concentrations are notably higher in the post-fire Ventura River when compared to other
studies. Although this difference may not be surprising since the Ventura River samples
were collected immediately after an intense fire in a fire-prone region, the higher
concentrations suggest that water-soluble (benzene polycarboxylic acids (BPCA)-derived)
DBC can be rapidly eroded in the first rain event after a fire. In contrast, prior studies have
found little relationship between the frequency of recent fires and DBC concentrations
(Ding et al., 2013; Wagner et al., 2015), perhaps because those studies did not include the
initial rapid erosion of fire products, captured here by our rapid-response event sampling.
Many of the metals also showed positive C-Q relationships in the Ventura River,
indicating hydrological control on the export of these constituents, too. Again, it is
important to note that in this study only one sample was taken post-fire, pre-flood. The
positive C-Q slopes for Ni (0.22) and Co (0.26) are similar to those for DOC and DBC,
perhaps indicating complexation of these metals with dissolved organic matter, or at least
their leaching from similar sources. The log-log C-Q slopes for Cd, Cu, and Zn were all
near-zero for the Ventura River, representing “chemostatic behavior” that may reflect
minimal leaching from the surface soil layer affected by burning. These elements also
showed less enrichment in post-flood versus pre-flood samples compared to the other
metals, with especially little enrichment for Cd and Zn, which complements the
interpretation of their C-Q relationships that there is less mobilization of these metals from
the fire-affected soils. At first glance, the steeper C-Q slopes for Fe (0.44), Pb (0.46), and
98
Mn (0.55) are consistent with their association to colloidal material, which may also be
sourced from surface soils and may exhibit steeper C-Q slopes compared to dissolved
organics (Trostle et al., 2016). This interpretation is consistent with a study of river water
following fire in the San Gabriel Mountains, which found that Fe and Pb were the metals
most strongly associated with colloids (Pinedo-González et al., 2017; Mn was not reported
in that study). However, when considering the colloidal data, there is a rough negative
relationship between the C-Q slopes and percent colloidal during the flooding event (Figure
3-S5b). This contradicts the observations made by Trostle et al. (2016), which suggested
steeper C-Q slopes are due to greater colloidal association. Our findings imply that size
partitioning of trace metals may not reflect organic versus non-organic partitioning.
4.6 Fluvial mobilization of dissolved trace metals and DBC into the Santa Barbara Basin
This study provides data collected over 5 days of the 40-day long Thomas Fire;
therefore, in an effort to gauge the full impact of the fire on the Santa Barbara Basin we
estimated and compared the total masses of metals brought to the basin and coastal
waters further offshore via atmospheric and fluvial transport (Figure 3-8). The total mass of
metals mobilized through atmospheric transport was calculated by multiplying the total
amount of burned area from the Thomas Fire (281,893 acres; Cal Fire, 2018) by how
much biomass is typically burned for an acre of California chaparral (27 ± 19 tons of fuel
per hectare of burnt land; Van Leeuwen et al., 2014) — yielding a total of 3.1x10
6
tons of
burnt biomass (Equation 1). Next, the amount of burnt biomass was converted into the
amount of particulate matter (PM2.5) released, using the amount of PM2.5 aerosol per
amount of burnt biomass for chaparral fires (5.46 ± 1.31 g PM2.5 aerosol per kg of burnt
99
biomass; Hosseini et al., 2013) (Equation 2). This calculation yields a total amount of
particulate matter released of 1.68 x 10
10
g PM2.5 aerosol or 16,825 tons of PM2.5 aerosol.
The total amount of particulate matter released was then multiplied by the average
concentration of each metal measured in aerosols in our samples, to calculate the total
amount of each metal produced by the Thomas Fire (Equation 3). We then assume that
100% of the fire-derived PM2.5 is deposited into the coastal ocean.
In order to calculate the total mass of metals mobilized through fluvial transport we
applied the C-Q relationships to the daily mean discharge values (Q) for the Ventura River
(Site 11118500; USGS, 2020b) from January 8, 2018 (start of the rain event) to January
12, 2018. We calculated a metal concentration (C) for each day based on the C-Q
relationship and known Q value (Equation 4), then multiplied the calculated C value by Q to
get the metal flux for each day (Equation 5). The daily fluxes for each metal were summed
over the 5 days to calculate the total mass of each metal transported by runoff in the
Ventura River during the rain event following the Thomas Fire. We then took into account
the fluxes from the larger Santa Clara River (which included some area burned during the
Thomas Fire) as well as the smaller mountainous rivers draining along the Santa Barbara
coast from burned areas of the Santa Ynez mountains. To account for these areas, we
scaled the Ventura River basin fluxes (from a catchment area of 226 square miles) to the
entire area burned during the Thomas Fire (440 square miles; yielding a scaling of 1.95 x).
This calculation provides an upper estimate of fluvial fluxes since not all of the Thomas Fire
area drains directly to the Santa Barbara Basin.
100
For Fe, Zn, Cu, Ni, and Cd, a larger amount of metal was mobilized by atmospheric
transport during the 40-day fire versus 5 days of fluvial transport during the rain event in
January 2018. For Mn, Pb, and Co, the atmospheric and fluvial fluxes were not different
outside of uncertainty. Only Mn (and perhaps Pb) indicate fluvial fluxes that were higher
than atmospheric. In general, then, we expect that the greatest short-term, metal-related
impact on the coastal system from the Thomas Fire would have come via the atmospheric
pathway. Logistics of coastal sampling during storms meant we were not able to capture
coastal seawater composition during the time of the flood runoff, but based on the amount
of metals delivered, we have no reason to expect that the effects of the riverine delivery
would have been more pronounced than those we observed under the smoke plume
during the fire itself. As discussed above, we found those effects to be essentially un-
identifiable, except in DBC. In contrast, the atmospheric transport is limited to the fire
event itself, whereas rivers may continue eroding for decades. Previous studies have
shown extended increases in sediment/nutrient discharge for years following a fire event
(Meixner et al., 2006; Moody et al., 2013; Warrick et al., 2012) due to the lack of terrestrial
plants that usually either ground the soil or take up nutrients (Minshall et al., 1989). The
long-term flux from rivers would be interesting to ascertain, but is beyond the scope of this
sampling campaign.
Of all the metals studied, the greatest delivery flux was of atmospheric Fe, with ~65
tons delivered to the Santa Barbara Basin and eastern North Pacific as a result of the
Thomas Fire. While this input did not appreciably affect Fe concentrations in the Santa
Barbara Basin, and in any case the Santa Barbara Basin is not Fe-limited, such a large
101
quantity of Fe could have important biogeochemical consequences if delivered further
afield. Applying our measured 5.5% solubility for aerosol Fe suggests 3.5 tons of soluble
Fe released from the Thomas Fire. Considering the Fe:C ratios of typical Fe-limited
diatoms (~3 µmol Fe per mol C; Marchetti et al., 2006), this Fe could stimulate the
production of roughly 250,000 tons of organic carbon. This amount is small compared to
total global productivity in HNLC (high-nutrient, low-chlorophyll) regions, and the true
stimulation of productivity is likely to be lower because this Fe will not all be delivered to
Fe-limited HNLC waters. None the less, this work highlights the potential importance of
biomass burning as a globally important source of Fe to the oceans (e.g. Guieu et al.,
2005; Matsui et al., 2018). As a fractional share of global aerosol input to the oceans
(Jickells, 1995), zinc and copper in the Thomas Fire aerosols seem to have the largest
impact representing 0.31% and 0.26% of the annual global input, respectively. Zinc was
also the most soluble metal in the Thomas Fire aerosols followed by Cd > Pb > Mn > Cu >
Co > Ni > Fe (Figure 3-3). Iron associated with the Thomas Fire aerosols had a negligible
impact on the annual global aerosol input to the oceans, but represents 0.40% of the
annual aerosol input into the North Pacific (Buck et al., 2013; Eakins & Sharman, 2010).

5. Conclusion
Black carbon and metals were released by the Thomas Fire and transported by
both atmospheric and fluvial pathways to the coastal ocean. For most metals, we estimate
that the atmospheric delivery to surface waters was larger than the river transport during
the rainstorm after the fire. Although we estimate that substantial amounts of metals were
102
released by the fire, our field measurements of coastal waters from beneath the smoke
plume did not reveal metal enrichment which was distinguishable from background
concentrations. These results are in accord with prior studies concluding that the direct
fallout of metals from the atmosphere after wildfires does not significantly alter southern
California coastal ocean metal cycling (Young & Jan, 1977). However, we did observe the
signature of black carbon delivery from the smoke plume to the coastal ocean, in terms of
elevated dissolved black carbon concentrations in surface waters, but no obvious effect on
marine productivity. Our results suggest that the immediate (5-day) fluvial delivery of metals
was of a similar or lower magnitude to atmospheric metal delivery, and therefore also not
likely to strongly affect the metal budgets of the coastal system. Nevertheless, we
anticipate that fluvial transport may have a longer-term effect on the Santa Barbara Basin
as rainfall continues to bring excess sediment and nutrients, potentially along with BC and
metals, from the burned coastal watershed many years after the time of the fire (Meixner et
al., 2006; Minshall et al., 1989; Moody et al., 2013; Warrick et al., 2012). The extended
erosion of sediments and nutrients in watersheds post-wildfire is important because fluvial
deposition is one of the major sources of metals to the Santa Barbara Basin, where the
residence time of water is <5 years (Emery, 1960; Krishnaswami et al., 1973; Sholkovitz &
Gieskes, 1971). The extended-release of sediment into the Santa Barbara Basin means
that nutrient and metal inventory of the basin may be affected by the Thomas Fire for
years. With the increase in size and frequency of wildfires on the western coast of the
United States, but also globally, the importance of studies of the effects of wildfires on
nearby ecosystems grows. Our study suggests that research to understand these effects
103
may be most effectively targeted at understanding the long-term fluvial fluxes and their
effects on the coastal system, and by studying the far-field effects of metals carried into
the open ocean with smoke aerosols.

104
Table 3-1
Samples taken and analyses done during both sampling campaigns of the Thomas Fire in
the Santa Barbara Basin and Ventura River

Type System Analysis Materials Lab
Organic
Aerosol
Levoglucosan QFF filter SCAQMD
Particulate Mass (PM2.5) Teflon filter UCLA
Black Carbon Teflon filter UCLA
Marine
Carbon Isotopes QFF filter USC
Chlorophyll a QFF filter USC
Marine, River
Dissolved Organic Carbon SPE-DOC Cartridge Oldenburg
Dissolved Black Carbon (BPCA) SPE-DOC Cartridge Oldenburg
Trace Metal
Aerosol
Instantaneous Leach Teflon filter USC
Total Digestion Teflon filter USC
Marine, River Dissolved Concentrations 0.2 µm filtered water USC
River Soluble Concentrations 0.02 µm filtered water USC
Other Marine Nutrients 0.2 µm filtered water UCSB

105
Table 3-2
Dissolved metal concentration data for 8 surface seawater samples collected in the Santa
Barbara Basin during the 2017 Thomas Fire

Sample ID Fe (nM) Zn (nM) Cd (pM) Ni (nM) Cu (nM) Pb (pM) Mn (nM) Co**(pM)
GSC* 1.45 1.36 413.76 4.19 1.32 26.25 1.49 58.18
Average during
the Thomas Fire
2.42±1.94 0.88±0.62 45.90±17.46 3.07±0.11 1.16±0.13 25.35±1.86 3.06±0.98 46.69±13.49
Station 1 0.38 0.15 27.08 2.92 0.97 23.41 2.14 29.55
Station 2 0.43 1.49 30.72 2.97 1.04 24.84 2.17 32.49
Station 3 2.75 1.55 71.49 2.96 1.14 28.32 3.22 50.65
Station 4 2.22 1.75 65.06 3.16 1.30 23.30 5.01 59.42
Station 5 2.61 0.73 61.93 3.19 1.30 27.41 3.64 56.05
Station 6 3.38 0.40 41.45 3.08 1.12 24.25 3.13 45.15
Station 7 6.36 0.51 34.98 3.17 1.31 26.33 3.04 65.72
Station 8 1.23 0.43 34.46 3.11 1.08 25.05 2.11 34.51

*GEOTRACES Surface Coastal (GSC) reference sample collected in the Santa Barbara Basin in 2009 during
a period of upwelling. Average of certified values from inter-calibration with other labs.

**Because samples were not UV oxidized, seawater cobalt concentrations do not include the fraction bound
by strong organic ligands and therefore resemble ‘labile cobalt’.

106

Figure 3-1
(a) A HYSPLIT model showing the trajectory of the wildfire smoke from December 5–14,
2017 (Rolph et al., 2017; Stein et al., 2015). Red box indicates the area shown in Figure
1B. (b) NASA Earth Observatory image by Joshua Stevens taken on December 5, 2017,
the day after the Thomas Fire started. The photo was taken using MODIS data from
LANCE/EOSDIS Rapid Response and modified Copernicus Sentinel data processed by
the European Space Agency (NASA, 2017b). Red box indicates the area shown in Figure
1C. (c) A satellite image of the Santa Barbara Basin on December 14, 2017, retrieved from
NASA worldview, is laid over a map of the basin (NASA, 2017a). Numbered markers
represent the stations from the Thomas Fire Cruise conducted as part of this study. The
red line shows the perimeter of the Thomas Fire on December 14, 2017 (County of Santa
Barbara, 2018). The yellow star is the location where the Ventura River sampling occurred.
(d) A photo taken by S. Feakins during the December 14th Thomas Fire Cruise of the
smoke plume over the Santa Barbara Basin.

Oxnard
Santa Cruz Island
Santa Barbara
33.90
34.10
34.30
34.50
-119.90 -119.60 -119.30 -119.00
34.50
o
N
-119.00
o
W
December 14, 2017
a.
b.
c.
12/05
12/06
12/07
12/08
12/09
12/10
12/11
12/12
12/13
12/14
Source at 34.43
o
N 119.09
o
W
Height 500 m
Ventura
Los Angeles
Clouds
Smoke
Thomas Fire
50 km
Rye Fire
Creek Fire
December 5, 2017
d.
107

Figure 3-2
(a) PM2.5 concentrations (μg/m
3
) in aerosols are plotted as black rectangles. Black carbon
(BC) concentrations (μg/m
3
) in PM2.5 are plotted as gray dashed rectangles. Rectangular
shaped data points show how long each aerosol sample was collected for. Four
categories (good, moderate, unhealthy for sensitive groups, unhealthy) of the United
States air quality index (AQI) are shown on the right. (b) Levoglucosan concentrations
(μg/m
3
) in aerosols are plotted as teal rectangles. The average levoglucosan concentration
0
5
10
15
20
25
in Aerosols (µg/m
3
)
Station
in Aerosols (µg/m
3
)
Levoglucosan d
13
C POM
in Seawater (µg/m
3
)
in Aerosols (µg/m
3
) ‰ in Seawater
Chlorophyll a
in Seawater (µg/L)
BC
TSP
DBC
Time (PST)
0
5
10
15
20
Unhealthy for
Sensitive Groups
in Aerosols (µg/m
3
)
BC
PM2.5
Good
Unhealthy
Moderate
1.2
1
0.8
0.6
0.4
0.2
0
1 2 3 4 5 6 7 8
PM
2.5
0
20
40
60
80
0.0
0.4
0.8
1.2
1.6
2.0
5:00 7:00 9:00 11:00 13:00 15:00
5:00 7:00 9:00 11:00 13:00 15:00
-25
-24
-23
-22
-21
-20
a.
b.
c.
Common LA Pollution
0
2000
4000
6000
8000
10000
108
found in Central Los Angeles in December 2009 and 2012 is indicated by the dashed line
(0.14 ± 0.04 μg/m
3
; Schauer & Sioutas, 2012; Shirmohammadi et al., 2016). Black carbon
(BC) concentrations (μg/m
3
) in total suspended particles (TSP) are plotted as black dashed
rectangles. Dissolved black carbon (DBC) concentrations (μg/m
3
) in surface seawater are
plotted as gray circles. (c) δ
13
C values (‰) for particulate organic matter (POM) in surface
seawater are plotted as navy blue diamonds. Chlorophyll a concentrations (μg/L) in surface
seawater are plotted as teal circles.

109

Figure 3-3
The y-axis shows percentages of the amount of each trace element that was found to be
soluble from aerosols collected offshore during the Thomas Fire, following an
instantaneous leach experiment (see Methods). The solubility values from this study are
compared to values from an aerosol trace metal leaching review by Mahowald et al.
(2018). The box plots represent percent soluble values for all 11 aerosol samples for each
metal. The lines inside the boxes represent the median (middle value), the lower and upper
bounds of the boxes represent the lower and upper quartiles (middle values of the lower
and upper halves, respectively), the whiskers represent the minimum and maximum
values, and the dots represent outliers.

0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Zn
60.8%
40%
Pb
42.1%
35%
Mn
37.7%
42%
Cd
56.8%
70%
Co
13.6%
20%
Ni
7.2%
Cu
27.9%
35%
Fe
5.5%
3.5%
Percent Soluble
This Study (mean values):
Mahowald et al. (2018):
110

Figure 3-4
Trace element (TE) load associated with atmospheric aerosols, where [TE]air,PM2.5 = (total
PM2.5 metal quantity)/(total air volume sampled); units are in pmol/m
3
. The navy-blue line
represents the median (MED) of the concentrations found in this study for aerosols. The
green line represents average concentrations for North Pacific aerosols (NP; 11°N and
162°E; Duce et al., 1983). The teal line represents the average for Los Angeles aerosols
(LA; Saffari et al., 2013).
1 2 3 4 5 6 7 8
Time (PST)
Station Station
Time (PST)
1 2 3 4 5 6 7 8
0
4000
8000
12000
LA: 1988
0
2
4
6
5:00 7:00 9:00 11:00 13:00 15:00
0
100
200
300
5:00 7:00 9:00 11:00 13:00 15:00
0
300
600
900
0
4
8
12
0
5
10
15
0
40
80
120
0
1000
2000
3000
Fe (pmol/m
3
) Zn (pmol/m
3
) Cd (pmol/m
3
) Ni (pmol/m
3
)
Cu (pmol/m
3
) Pb (pmol/m
3
) Mn (pmol/m
3
) Co (pmol/m
3
)
NP: 322
NP: 0.7
MED: 2020
LA: 142
MED: 155
NP: 3
NP: 1
NP: 5
LA: 123
MED: 8
MED: 52
LA: 32
MED: 994
NP: 0.3
NP: 0.03
MED: 1.9
LA: 24
LA: 1
LA: 1.8
MED: 67
MED: 3
LA: 9
111

Figure 3-5
(a) A photo of the trace metal samples taken during river water sampling. (b) The stage
height for the Ventura River water before (yellow), during (red), and after (orange) the
January 2018 flash flood event (Site 11118500;USGS, 2020b). (c) Dissolved organic
carbon concentrations (DOC; μM) in river waters are plotted as navy blue triangles.
Dissolved black carbon concentrations (DBC; μM as BPCAs) in river waters are plotted as
gray circles. T0 is the time at which the first sample was taken.

Ventura River January 2018
Hours Since T
0
DOC (µM)
DBC (µM)
Stage Height (ft)
01/08/18 01/10/18 01/12/18
0
3000
6000
9000
0 20 40 60 80 100
0
50
100
150
200
a.
b.
c.
6
10
14
18
112

Figure 3-6
Total dissolved metal concentrations in river samples (<0.2 μm filtrate). Note the changes
in ranges of concentrations and changes units on the y-axis of each plot. T0 is the time at
which the first sample was taken.

0
1000
2000
3000
4000
Fe (nM)
0
10
20
30
40
Zn (nM)
0
250
500
750
1000
Cd (pM)
0
25
50
75
100
0 20 40 60 80 100
Ni (nM)
0
10
20
30
40
Cu (nM)
0
100
200
300
400
0
2500
5000
7500
10000
Pb (nM) Mn (nM)
0
10
20
30
40
0 20 40 60 80 100
Hours Since T
0
Co (nM)
Hours Since T
0
Ventura River January 2018 Ventura River January 2018
6
10
14
18
Stage Height (ft)
01/08/18 01/10/18 01/12/18
6
10
14
18
Stage Height (ft)
01/08/18 01/10/18 01/12/18
113

Figure 3-7
(a) The concentration of dissolved black carbon (DBC) versus discharge (m
3
/s) found in
this study (Ventura River), the arctic (Myers-Pigg et al., 2017; Stubbins et al., 2012), the
Rocky Mountains of Colorado (Wagner et al., 2015), the tropical forests of Brazil (Dittmar et
al., 2012), the southeastern US (Roebuck et al., 2018), and southeast China (Bao et al.,
2019). (b) Dissolved black carbon (DBC) flux (tons/day) versus discharge (m
3
/s).
0.00
0.01
0.10
1.00
10.00
Ventura River
Arctic Rivers
Colorado
Brazil
Southeast US
Southeast China
0.00001
0.001
0.1
10
1000
100000
0.001 1 1000 1000000
Ventura River
Arctic Rivers
Colorado
Brazil
Southeast US
Southeast China
DBC (mg/L) DBC Flux (tons/day)
Discharge (m
3
/s)
a.
b.
114

Figure 3-8
Estimates of the (a) total mass and (b) total soluble mass of each trace metal that was
released as aerosols due to the Thomas Fire, and in river flux for the January 8–12, 2018
flash flood event. Some of the Cd estimates not visible due to scale of y-axis.

0.01
0.1
1
10
100
1000
Fe Zn Cu Ni Mn Pb Cd Co
Trace Element Mass (tons)
aerosols
river flux
0.01
0.1
1
10
100
1000
Fe Zn Cu Ni Mn Pb Cd Co
Trace Element Mass (tons)
soluble aerosols
soluble river flux
a.
b.
115

Figure 3-S1
Trace element (TE) content in PM2.5 aerosols, where [TE]aerosols;PM2.5 = (total TE PM2.5 mass)
÷ (total PM2.5 mass); units are in ppm (μg/g). Trace element concentrations are compared
to values found in previous publications for: wood (total particulate samples: Butkus &
Baltrėnaitė, 2007; Nicewicz & Szczepkowski, 2008; Queirolo et al., 1990), Los Angeles
aerosols (LA; <0.25 μm particulate samples: Saffari et al., 2013), fly ash (FA; <300 μm
particulate samples: Koukouzas et al., 2007; Pitman, 2006; Steenari et al., 1999; Świetlik
et al., 2013), and ash from two California wildfires (WA; total particulate samples: Odigie &
Flegal, 2011, 2014).
0
2000
4000
6000
0
100
200
300
5:00 7:00 9:00 11:00 13:00 15:00
1 2 3 4 5 6 7 8
Time (PST)
Station
Wood Wood
LA
LA
0
10
20
30
Station
0
400
800
1200
0
50
100
150
0
2
4
6
5:00 7:00 9:00 11:00 13:00 15:00
Time (PST)
0
5000
10000
15000
23600-88900
FA
to 16834
Fe (ppm) Zn (ppm) Cd (ppm) Ni (ppm)
Cu (ppm) Pb (ppm) Mn (ppm) Co (ppm)
0
30
60
90
1 2 3 4 5 6 7 8
Wood
FA
LA
Wood
FA
LA
Wood
FA
LA
Wood
FA to 1000
109-227
Wood
LA
LA
to 393
LA
WA to 24338 WA WA
FA
195-344
WA WA
FA to 300
WA to 7
to 12 3092-17300
to 140 FA
116

Trace Metals
34.6
o
N
34.4
o
N
34.2
o
N
34
o
N
119.8
o
W 119.4
o
W 119
o
W
Fe [nM] Zn [nM]
Pb [pM] Cu [nM]
Cd [pM] Ni [nM]
Co [pM] Mn [nM]
a.
117

Figure 3-S2
The concentrations of a) trace metals, b) nutrients, and c) organics in the surface seawater
samples are labeled in the z-axis (color scale). The range of concentrations for each
metal/nutrient/organic varies; note the z-axis and the units next to each of the
metal/nutrient/organic varies. The brown shaded area shows where the smoke plumes
were located on December 5-14th, 2017 (retrieved from NASA worldview; excluding the
9th and 10th due to cloud coverage in satellite images). Gray bars on the z-axis references
background metal/nutrient/organic concentrations in the Santa Barbara Basin.
Nutrients
Organics
34.6
o
N
34.4
o
N
34.2
o
N
34
o
N
119.8
o
W 119.4
o
W 119
o
W
Phosphate [µM] Silicate [µM]
Nitrite + Nitrate [µM] Ammonia [µM]
DOC [mM] DBC [µM]
b.
c.
118

Figure 3-S3
Comparisons of metal concentrations in water samples from control versus burned river
basins in southern California. Previous studies were done on paired catchments (Burke et
al., 2013 – total metal concentrations; Gallaher & Koch, 2004– dissolved metal
concentrations; Pinedo-González et al., 2017– dissolved metal concentrations; Yoon &
Stein, 2008 – total metal concentrations). This study compared one catchment over time
(see text – dissolved metal concentrations).
0
1
2
3
0
100
200
300
0
20
40
60
0
40000
80000
120000
0
100
200
300
0
200
400
600
Cd (pM)
Pb (nM) Ni (nM) Cu (nM)
Zn (nM)
34x
1.1x
316x
58x
27x
4.3x
1.4x
7.9x
9.6x
4.6x
2.4x
7.5x
3.1x
1.6x
Fe (nM)
100x
10x
61x
Control
Burned
2.7x
1.9x
2.5x
3.6x
1.5x
Y+G B P This Study
Y
G
B
P
Yoon and Stein, 2008
Gallaher and Koch, 2004
Burke et al., 2013
Pinedo-Gonzalez et al., 2017
This study
Y+G B P This Study
119

Figure 3-S4
Concentration-discharge (C-Q) relationships of different constituents in the river water
collected during the flooding event in the Ventura River. Both discharge and
concentrations are on a log-scale. The line of best-fit is plotted for each of the C-Q
relationships, assuming a power-law relationship, and the slope of each line in log-log
space is reported in the bottom right corner of each plot.
0.1
1
10
100
1000
10000
0.1
1
10
100
1000
10000
0.1
1
10
100
1000
10000
0.1
1
10
100
1000
10000
+0.44
+0.08
+0.05
+0.46
Fe (nM) Zn (nM)
Cu (nM) Pb (nM)
0.1
1
10
100
1000
10000
0.1
1
10
100
1000
10000
0.001 0.1 10 1000
0.1
1
10
100
1000
10000
0.001 0.1 10 1000
0.1
1
10
100
1000
10000
-0.07
+0.22
+0.55
+0.26
Cd (pM)
Mn (nM)
Ni (nM)
Co (nM)
0.1
1
10
100
1000
10000
0.1
1
10
100
1000
10000
+0.29 +0.32
DOC (µM)
DBC (µM)
Discharge (m
3
/s)
Discharge (m
3
/s)
120

Figure 3-S5
a) The average colloidal fraction (in %) of dissolved trace metals in Ventura River water
samples grouped by the period in which the samples were taken: “Pre-Flood”, “Flood”,
and “Post-Flood”. Ni, Cu, and Co “Pre-Flood” values missing due to colloidal
concentrations being below detection limits. b) Concentration-discharge (C-Q)
relationships of the metals versus the average colloidal fraction (in %) of dissolved trace
metal concentrations, both in the Ventura River water collected during the flood period.
Fe
Zn
Cd
Ni
Cu
Pb
Mn
Co
-0.2
0
0.2
0.4
0.6
0.8
0 20 40 60 80 100
C-Q Relationship
% Colloidal During Flood Event
0
20
40
60
80
100
Pre-Flood Flood Post-Flood
% Colloidal
Fe
Zn
Cd
Ni
Cu
Pb
Mn
Co
a.
b.
121
Chapter 4
The effect of host iron limitation on viral infection dynamics between
Vibrio and vibriophage

Author Contributions
Seth John designed the study. Kathryn Kauffman provided both the Vibrio host cultures
and the paired phage stocks. Rachel Kelly led and performed the experiments and
organized the manuscript.

Abstract
While microbes may have the largest biomass throughout the world oceans, viruses
are the most abundant. Viral infection is a main cause of microbial mortality and therefore
has major impacts on community structure and nutrient cycling (Fuhrman, 1999; Suttle,
2007; Zimmerman et al., 2020). Iron limitation of microbial growth is common in the ocean
(Moore et al., 2013; Schoffman et al., 2016; Tagliabue et al., 2017), in this study we
examine whether iron limitation of bacterial growth affects the viral dynamics of their
bacteriophage. More specifically, we observe how iron affects the latent period, the time in
which it takes a virus to complete a single lytic cycle, and burst size, the number of times a
single viral particle replicates in a host cell, of marine bacteriophage.
Our experiments studied multiple strains of coastal marine Vibrio splendidus
bacteria and their paired vibriophage. Growth in a low-iron medium successfully limited
growth of the host Vibrio by up to 40%. Iron replete (0.05 nM Fe’) and deplete (no added
Fe and 0.01 nM Fe’) host cultures were then infected with their paired vibriophage and
monitored for phage abundance to observe the latent period and burst sizes after one
cycle of lysis (one-step phage growth). Our results revealed that for one host strain (Vibrio
122
splendidus 10N.261.52.C3 infected by myovirus 1.170.O) post-infection burst sizes of iron
replete cultures (48-89 PFU/cell) were greater than those for the deplete cultures (6-16
PFU/cell), but that latent period was unaffected by the varying Fe concentrations in all
experiments. For the other host strain (Vibrio splendidus 10N.222.46.C2 infected by
podovirus 2.130.O) both burst sizes and latent periods were unaffected by the differences
in Fe concentrations. Cell size of iron replete and deplete host cultures were measured and
no increases in cell sizes were found due to higher iron concentrations. Finally, we studied
the viral dynamics of infected Vibrio splendidus at different temperatures and found that
the treatment which resulted in the largest burst sizes (at 23°C, 24 ± 9.9 PFU/cell) did not
coincide with the optimal growth rate temperature (39.5°C). Our findings are consistent
with previous work suggesting that cell size is a primary control on latent period, thus
explaining why we observed no strong effect of Fe on latency. Similarly, our work is
consistent with prior suggestions that the efficiency of the host’s DNA replication
machinery regulates burst sizes.

1. Introduction
For almost a century now it has been accepted that marine life is dependent on
three major nutrients: carbon (C), nitrogen (N), and phosphorus (P) (Redfield, 1934), but
not until the mid-1980’s did oceanographers realize phytoplankton growth was limited by
the micronutrient iron in regions where C, N, and P were replete (Brand et al., 1983; Martin
& Gordon, 1988). Iron (Fe) is needed by all organisms because of the crucial role it plays in
enzymes responsible for electron transfer, oxygen metabolism, and DNA replication (Fuss
123
et al., 2015; Morel et al., 2003; Morrissey & Bowler, 2012). Even though iron is abundant in
the Earth’s crust (5% by weight), it is insoluble in oxic seawater unless bound by organic
ligands (Moffett & Zika, 1987); therefore, iron is found at very low dissolved concentrations
of 0.02 to 2 nM in most of the world’s oceans (Mawji et al., 2014). The low concentrations
of Fe in seawater combined with the biological iron requirements by almost all marine life
makes it the most common limiting micronutrient in the world’s oceans (Moore et al.,
2013). Laboratory-culture studies that included many of the major marine bacterial and
cyanobacterial species (Webb et al., 2001; Wilhelm et al., 1998), along with shipboard in
situ incubations, have shown that iron limits up to one third of marine biological production
(Boyd et al., 2007).
Marine microbes, which include bacteria and protists, make up 70% of all living
biomass in the oceans (Bar-on et al., 2018), but even more abundant than microbes are
viral particles, with an estimated concentration of 10
7
– 10
9
viral particles per milliliter of
surface seawater (Fuhrman, 1999; Maranger & Bird, 1995; Suttle, 2005). Viral infection is
an important top-down control on marine ecosystems where infection and lysis is
estimated to account for 8-43% of bacterial mortality (Fuhrman, 1999; Suttle, 2005;
Wilhelm & Suttle, 1999). Examining the factors which control viral infection is therefore
critical to understanding the marine microbial community as a whole. While there have
been many studies on how iron affects bacterial cell growth, there have been little to no
studies on how this key limiting micronutrient affects viral infection, a crucial part of the
microbial loop.
124
Viral infection dynamics are often studied in the laboratory with ‘one-step’ phage
growth experiments, in which a single cycle of viral infection is tracked (Ellis & Delbruck,
1939). The two important pieces of information which come from these experiments are
the latent period and the burst size. The latent period refers to the time from when a viral
particle absorbs onto a host cell until that virus’ progeny lyse/burst from the host cell
(Weitz, 2016). The burst size measures the number of times a single viral particle replicates
within one host cell (Weitz, 2016). To date, the effect of iron on marine viral infection
dynamics is limited to one unpublished one-step phage growth experiment looking at the
viral infection of Fe-limited Synechococcus (Cunningham, 2015). Despite the large
changes in growth rates for iron replete and deplete Synechococcus, no difference in
latent period or burst size was observed between the two treatments in Cunningham's
(2015) study. Other studies that focus on the role of iron in viral replication have come from
the biomedical field. Previous biomedical research has looked at how varying Fe-levels
affect the replication of the hepatitis C virus (Drakesmith & Prentice, 2008; Mcdermid et al.,
2007). In contrast to Cunningham's (2015) findings, research on the hepatitis C virus
showed that infected patients who had higher iron levels had a lower probability of survival
due to an increase in hepatitis C viral replication (Mcdermid et al., 2007).
Viruses are very complex and varying changes to viral dynamics due to differences
in growth conditions, host cell species, or phage strains is common. Gledhill et al. (2012),
for example, showed that the lysis rates of the Emiliana huxleyi – Ehv86, host – virus
system were reduced at high copper concentrations and was unaffected by increases in
zinc, cadmium, and cobalt concentrations. Additionally, previous studies have shown that
125
the viruses which infect different marine eukaryotes had various responses to phosphorus
limited growth. Both phosphorus limited Emiliana huxleyi and Micromonas pusilla resulted
in lower viral burst sizes when compared to cultures grown in phosphorus replete
conditions (Bratbak et al., 1993; Maat et al., 2014), whereas Phaeocystis pouchetii’s viral
burst sizes were unaffected by changes in phosphorus concentrations (Bratbak et al.,
1998). Furthermore, Wilson et al. (1996) looked at the burst size of nitrate deplete versus
replete infected Synechococcus sp. WH7803 cultures and found that one strain of phage
(S-PM2) was unaffected by the differences in nitrate concentration, while another strain of
phage (S-WHM1) showed a 25% reduction in titer. The relationships between nutrient
availability, health of the host cultures, and viral infection thus does not seem to follow a
simple and predictable pattern.
In addition to top-down controls, viral lysis of microbes results in an important
bottom-up control on marine ecosystems as viral lysis leads to the release of nutrients
from host cells which can support the growth of new microbes. As host cells are lysed
during viral infection it has been shown that important nutrients such as C, N, P, and Fe
are released with cell debris (lysates) in the dissolved phase, therefore making them
bioavailable to other microbial cells (Poorvin et al., 2004; Strzepek et al., 2005; Wilhelm &
Suttle, 1999). This recycling of nutrients has been coined “the viral shunt” (Wilhelm &
Suttle, 1999). This concept has large implications on microbial ecology and diversity. The
recycling of these essential nutrients promotes microbial growth, which increases
adsorption rates of viral particles to host cells, the cells are lysed, and the nutrients are
recycled again. A high percentage of viral infection may promote microbial diversity by
126
killing the most abundant species, therefore providing an opportunity for other populations
to grow. The viral shunt also plays a significant role in the carbon cycle, and therefore
global climate. The result is less organic matter leaves the surface ocean, which in turn
changes the sinking rates of carbon to the deep ocean.
In this study we aimed to expand on the early work looking at the effects of host
cell Fe levels on the viral dynamics of marine bacteria. We conducted four one-step phage
growth experiments on two different strains of Vibrio splendidus grown on three different
medias with no added Fe, 0.01 nM Fe’, and 0.05 nM Fe’ to look at the effects of varying
Fe concentrations on viral infection (free Fe concentrations, Fe’, calculated using equations
found in Sunda et al., 2005). We then conducted one-step phage growth experiments at
three different temperatures (7°C, 23°C, and 39.5°C) to test how varying temperatures
affects vibriophage infection. The quantitative burst size and latent period information on
viral dynamics gained from these experiments may help further our understanding of top-
down and bottom-up controls on the oceanic microbial ecosystem – the root of marine
life.

2. Materials and Methods
2.1 Iron one-step phage growth experiments
2.1.1 Host
Two strains of Vibrio splendidus were used in these experiments: Vibrio splendidus
10N.222.46.C2 (referred to as “V36”) and Vibrio splendidus 10N.261.52.C3 (referred to as
“V40”). Both Vibrio hosts were isolated off the coast of Nahant, Massachusetts, USA
127
(Kauffman et al., 2018). Between experiments, Vibrio stocks were saved by adding 500 µL
of culture grown in 0.01 nM Fe’ Vib-FeL to 500 µL of 50% glycerol and were stored at
-80°C.
The hosts were grown in a modified version of the Vib-FeL medium used in
Westrich et al. (2016) which included a mixture of salts and nutrients (420 mM NaCl, 2.5
mM MgSO4, 0.1 mM K2HPO4, 0.1 mM CaCl2, 18 mM NH4Cl, and 42 mM Na2PO4) and a
carbon source of sucrose (Sigma, 0.4% wt/vol) dissolved in Milli-Q water. The one
difference between the Westrich et al. (2016) media and the media used in our
experiments was the trace metal stock (trace metal stock used in this experiment: 100 μM
EDTA-buffered solution of 79.7 nM Zn, 121 nM Mn, 50.3 nM Co, 100 nM Mo, 19.6 nM Cu,
and 10 nM Se; Westrich et al., 2016 trace metal stock: 50 μM EDTA-buffered solution of
39 nM Zn, 230 nM Mn, 25 nM Co, 110 nM Mo, and 9.8 nM Cu). The only trace metal
which was tested for its impact on viral phage dynamics was Fe. For each experiment,
three different amounts of FeCl3·6H2O were used including no added Fe, 1 μM added Fe,
and 5 μM added Fe. Free iron concentrations (Fe’) in each of the treatments were
calculated using equations found in Sunda et al. (2005) as 1 μM added Fe = 0.01 nM Fe’,
and 5 μM added Fe = 0.05 nM Fe’.
2.1.2 Phage
The vibriophage used in these experiments included the podovirus phage
2.130.O_10N.222.46.C2 and the myovirus phage 1.170.O_10N.261.52.C3, which
infected host V36 and host V40, respectively. New phage stocks were prepared by adding
10 μL of a previously made phage stock to 5 mL of host culture grown in 0.01 nM Fe’ Vib-
128
FeL. The infected host culture was then incubated at room temperature (23°C) for at least
4 hours. After 4 hours the infected culture was filtered through a 0.2 μm acid washed
Supor
®
filter and the resulting filtrate (the new phage stock) was stored in a Protein LoBind
Eppendorf Tube
®
at 7°C.
2.1.3 Plating
A separate top agar and bottom agar were prepared for plating. Top agar was
made to be less viscous than bottom agar to allow for easy diffusion of bacteriophage
plaques. The top agar included 39 g of HiMedia
®
Zobell Marine Broth 2216, 52 mL of
glycerol, and 3 g of Bacto™ Agar per 1000 mL of Milli-Q water (18.2 MΩ × cm), while the
bottom agar included 58 g of HiMedia
®
Zobell Marine Agar 2216 and 52 mL of glycerol per
1000 mL of Milli-Q water. Both types of agars were first autoclaved and then kept in a
50°C water bath before being used to make plates. Bottom agar plates were made by
pouring approximately 5 mL of bottom agar into 100 mm-diameter, polystyrene VWR
®

petri dishes and then plates were left to cool/solidify for at least 30 minutes.
Different plating procedures were used depending on if the vibriophage or the Vibrio
hosts needed to be counted. In order to prepare plates for counting phage abundance the
phage stock or infected culture (various volumes), an axenic host culture (100 μL; added to
create a host lawn), and liquid top agar (2.5 mL) were all pipetted onto a premade bottom
agar plate. Immediately after pipetting, the plate was swirled until the top liquid mixture
was evenly mixed and distributed across the bottom agar base. The finished plate was
then covered and left to incubate at 23°C for about 24 hours. When plating, in order to
count Vibrio abundance, only the Vibrio culture (volume varied) and liquid top agar (2.5 mL)
129
were pipetted onto the bottom agar. The top liquid mixture was then mixed, and plates
were left to incubate.
2.1.4 Quantifying concentrations of Vibrio and vibriophage in stock solutions
At least one day before performing a one-step phage growth experiment, the
concentration of Vibrio and vibriophage to be used was quantified by performing serial
dilutions. Serial dilutions of the host culture were performed alongside measurements of
absorbance at 595 nm in order to establish the relationship between absorbance and cell
density. A subsample of the host culture was placed into a sterile semi-micro polystyrene
cuvette and absorbance was measured at 595 nm using a Thermo Scientific™
GENESYS™ 20 Visible Spectrophotometer. An absorbance blank prepared with sterile
Vib-FeL media was measured prior to each measurement and subtracted from all sample
measurements.
Immediately after measuring absorbance, the same liquid host culture was then
diluted several times in sterile Vib-FeL media, serially, and each dilution was plated onto an
agar plate. After a 24 hour incubation, the number of colonies on the plates were counted,
extrapolated to determine Vibrio concentration in the original stock, and related to
absorbance measurements.
Similarly, serial dilutions of the phage stock were performed in order to determine
the concentration of infective phage particles in the stock solution. The phage stock was
diluted several times in sterile Vib-FeL media, and each dilution was added to agar plates
with axenic host culture. After being incubated for 24 hours, plates were examined for
small circular clearings on the host lawn, with each clearing, or plaque forming unit (PFU),
130
representing a single infective phage particle. The number of plaque forming units was
counted to determine the original concentration of phage in the phage stock.
2.1.5 Host cell growth curves
In the vibriophage experiments the growth of the two Vibrio hosts each cultured in
three different media with varying Fe concentrations was monitored before and during the
one-step growth experiments. The growth of each treatment was monitored using a
spectrophotometer set to a wavelength of 595 nm. Doubling time (Td) was calculated by
using the following equation: N(t) = N02
t/Td
. In the equation N(t) is the concentration of host
cells (units: absorbance units) at time t (units: hours) and N0 is the initial concentration of
host cells. The growth rate (μ) was calculated by first determining which four-hour period of
the growth curve was the host culture in linear phase, plotting the natural log of the
measured absorbance units during the linear phase, and calculating the slope of the linear
regression which is equal to μ.
2.1.6 One-step vibriophage growth experiment
The one-step vibriophage growth experiments were initiated by inoculating cultures
of the Vibrio host, introducing phage once the host had achieved exponential growth, and
plating the infected host culture to track a single step of the viral infection. At least one day
before a one-step experiment, a single host culture (either V36 or V40) grown in the Vib-
FeL with 1 μM added Fe was streaked onto a bottom agar plate and incubated at 23°C.
Using a sterile polystyrene inoculation loop, host colonies of similar size were inoculated
from the same agar plate into three 14 mL polystyrene culturing tubes (one tube for each
treatment) which contained 5 mL of the appropriate Vib-FeL media (no added Fe, 0.01 nM
131
Fe’, and 0.05 nM Fe’). The inoculated liquid cultures were then vortexed, and the initial
absorbance value was measured using a spectrophotometer. The liquid cultures were
then incubated at 23°C on a shaker table.
The growth of host cultures from each treatment was monitored throughout the
experiment and triplicate subsamples of each treatment were infected with the appropriate
phage once the cultures had reached the exponential phase of growth. The growth of the
host cultures was monitored every 30 minutes by absorbance at 595 nm for approximately
3 to 4 hours, the average time it takes Vibrio host cultures to reach exponential phase.
Once the cultures had been identified as being in exponential phase the amount of host
culture and phage stock needed for the start of the one-step phage growth experiment
was determined. Three main factors went into determining the amount of host culture and
phage stock that was needed to start the experiment: 1) the ideal multiplicity of infection
should be approximately 0.1 (1 phage for every 10 host cells), 2) the total inoculum (known
as the adsorption mixture) should be under 1 mL in volume in order to increase the host to
phage interaction, 3) the volume of the host culture and volume of the phage stock added
to the adsorption mixture should be the same for all three treatments (no added Fe, 0.01
nM Fe’, and 0.05 nM Fe’). In order to satisfy all three criteria, either the host culture or the
phage stock needed to occasionally be diluted in sterilized media prior to being added to
the adsorption mixture. Once the volume of both the host culture and phage stock needed
was determined, the two appropriate volumes were pipetted into a polystyrene
culturing tube. The adsorption mixture was left in the tube for 15 minutes and was gently
swirled throughout the waiting period in an effort to increase host cell and phage
132
interaction. After the 15 minutes, the adsorption mixture was diluted several times (final
dilution ≈ 200 phage/mL) to decrease the chance of new phage progeny interacting and
infecting a new host cell after lysis.
Once the adsorption mixture had been diluted, the one-step phage growth was
considered to officially have begun and the plating portion of the experiment began. A
subsample from the last dilution of the adsorption mixture was pipetted onto a bottom
agar plate (100 μL ≈ 20 phage particles) and mixed with an axenic host culture and top
agar. A subsample from the last dilution was pipetted onto a new agar plate every 20
minutes for a total of 3 hours, which is slightly longer than the typical time it takes for a
single cycle of infection by vibriophage. Two subsamples at T0 were taken. One T0
subsample was plated for total phage counts and the other was gravity filtered through a
0.2 μm Supor
®
filter then plated for external phage counts.
The latent period of each one-step phage growth curve was visually determined by
identifying the first time point at which an obvious increase in phage abundance was
observed. The burst size of each experiment was calculated by first averaging the number
of PFUs during the latent period and subtracting that from the average number of PFUs
after the latent period, then dividing that value by the number of total PFUs minus the
external PFUs counted on the T0 samples as follows:
burst size = (avg. PFU after burst – avg. PFU before burst) ÷ (total PFU – external PFU)
2.2 Cell size measurements
Iron replete and deplete cells were measured using a Zeiss AxioStar Plus
Microscope with a digital AxioCam equipped for phase contrast and epifluorescence
133
microscopy. V36 and V40 host cultures grown in no added Fe and 0.05 nM Fe’ Vib-FeL
media were subsampled at exponential growth (at 3-4 hours after inoculation) and placed
on a wet mount. Photos were taken and stored using Zeiss AxioVision 4.5 software. Sizing
was done via calibration with a slide micrometer and measurements were done using the
ImageJ image processing software.
2.3 Temperature one-step phage growth experiment
Temperature controlled one-step phage growth experiments were performed in the
same fashion as the varying iron experiments. All of the cultures in the temperature
controlled experiments were grown in Vib-FeL media with 0.05 nM Fe’. For each
temperature treatment the inoculated experimental host culture and the adsorption mixture
was incubated at temperatures including 7°C, 23°C, and 39.5°C, with the 7°C treatment
performed in a refrigerator, the 23°C treatment grown at room temperature, and the
39.5°C treatments performed in an incubator. Temperatures for all treatments were
monitored throughout the experiment.

3. Results
3.1 Iron one-step phage growth experiments
Two different strains of Vibrio splendidus were used in the iron one-step phage
growth experiments: Vibrio splendidus 10N.261.52.C3 (V40, Figure 4-1) and
10N.222.46.C2 (V36, Figure 4-2). Duplicate identical experiments were performed with
each strain. For the first experiment with V36, growth rates for the V36 in the no Fe and
0.01 nM Fe’ treatments were ~40% lower than the growth rate in the 0.05 nM Fe’
134
treatment, as determined from dividing the linear growth rate of the specified low Fe
treatment by the linear growth rate of the 0.05 nM Fe’ treatment (μ/μmax). However, there
was no significant change in latent period and burst size in all three treatments. In the
second experiment using V36 the growth rates did not vary as much compared to the first
experiment. The growth rates for the 0.01 nM Fe’ and 0.05 nM Fe’ treatments were about
the same and the no Fe treatment was ~20% lower than the 0.05 nM Fe’ treatment. As
with the first experiment, we found no significant difference in latent period or burst size
between all three treatments. The latent periods in both experiments was 120 minutes,
and average burst sizes varied between 8-34 PFU/cell.
Two duplicate identical experiment were also performed with V40 and its phage.
Growth rates diminished by a similar amount in Fe-free and low-Fe treatments in both
experiments, with growth in the no Fe treatment about 30-40% lower than the 0.05 nM Fe’
treatment and growth rates about 12% lower in the 0.01 nM Fe’ treatments. Latent
periods for all of the Fe treatments in both experiments were 80 minutes long. The burst
sizes for the no Fe and 0.01 nM Fe’ treatment were not significantly different and were on
average between 6-16 PFU/cell, whereas burst sizes were significantly larger in the 0.05
nM Fe’ treatments. In the first experiment the burst size for the 0.05 nM Fe’ treatment was
48 ± 8.2 PFU/cell (1σ SD)and in the second experiment it was 89 ± 67.5 PFU/cell (1σ SD).
3.2 Cell size measurements
Cell size during exponential growth of Vibrio V36 and V40 was measured for both
Fe-replete and Fe-deplete condition, with 10 cells measured under each condition (Table
4-1). The average width, height, and area of the iron deplete V40 cells were 0.89 μm, 2.45
135
μm, and 8.21 μm
2
, respectively. For iron replete V40 cells the average width, height, and
area were 0.96 μm, 2.54 μm, and 9.03 μm
2
, respectively. The cell sizes of iron replete and
deplete V40 cells were not significantly different (p-value = 0.4525). The cell sizes of V36
iron deplete cells were similar to V40 cells. Average width, height, and area of iron deplete
V36 cells were 1.08 μm, 2.06 μm, 9.24 μm
2
, respectively. The cell sizes of V36 iron replete
cells were smaller than the iron deplete cells (averages: 0.92 μm width, 1.54 μm height,
5.89 μm
2
area). V36 iron replete cells were significantly smaller than V36 iron deplete cells
(p-value = 0.0459).
3.3 Temperature one-step phage growth experiment
Vibrio splendidus 10N.261.52.C3 (V40) cultures were grown at three different
temperatures (7°C, 23°C, and 39.5°C) and then infected with their paired phage (myovirus
1.170.O; Figure 4-3). The V40 host culture grew the fastest at 39.5°C (μ = 0.13; μ/μmax =
100%), second fastest at 7°C (μ = 0.10; μ/μmax = 78%), and slowest at 23°C (μ = 0.09;
μ/μmax = 67%). The observed latent periods for the two cultures that burst, at 7°C and
23°C, were the same and were approximately 80 minutes long, which is equivalent to the
latent period for V40 and its phage in the Fe manipulation experiments. The largest burst
size was observed in the 23°C treatment (room temperature, average: 24 ± 9.9 PFU/cell,
1σ SD), which is the temperature at which all of the iron one-step phage growth
experiments presented here were conducted. Burst sizes were 70% lower at 7°C
(average: 7 ± 2.9 PFU/cell, 1σ SD). Though the host growth rate was fastest at 39.5°C, the
addition of phage did not lead to viral infection and lysis, and no burst was observed over
the 3.5 hour experimental timeframe.
136
4. Discussion
4.1 Effect of Fe limitation on viral infection dynamics
One-step phage growth experiments with V40 showed that, the latent period was
not affected by Fe limitation of the host, but differences in burst sizes were observed
(Figure 4-1). For both duplicate experiments which were performed, the two lower iron
treatments (no Fe and 0.01 nM Fe’) had lower burst sizes than the high iron treatment
(0.05 nM Fe’, Figure 4-1). The effect of Fe-limited host growth on viral burst sizes may be a
result of low Fe levels limiting certain metabolic processes within the host cell, specifically
those related to DNA replication. Iron is an essential co-factor in enzymes used in many
important cellular operations including electron transfer (cytochromes and ferredoxin;
Morrissey & Bowler, 2012; Saito et al., 2011), oxygen metabolism (peroxidase and
superoxide dismutase; Chadd et al., 1996; Hogle et al., 2014), and DNA replication
(ribonucleotide reductase, pirin, DNA primase, and DNA helicases; Drakesmith & Prentice,
2008; Fuss et al., 2015). Viral replication utilizes the host’s DNA replication machinery, and
thus viral infection may have a special dependence on host enzymes involved in DNA
synthesis, replication, repair, and transcription (Drakesmith & Prentice, 2008).
Ferrochelatase, an Fe-S containing enzyme that inserts ferrous iron into heme, and
ribonucleotide reductases, which uses ATP to generate DNA, are thought to be among the
first enzymes to be inhibited during the onset of iron deficiency due to their roles at the
beginning of ATP and DNA synthesis and their sensitivity to iron levels (Drakesmith &
Prentice, 2008; Nyholm et al., 1993; Umbreit, 2005). If ribonucleotide reductase is indeed
one of the first enzymes to be inhibited by a limited iron supply, then viral DNA production
137
should be directly affected by the low iron levels. Decreased viral DNA production within
the same latent period due to the preferential inhibition of DNA replication enzymes would
lead to lower burst sizes, matching observations from our experiments with V40.
Previous studies looking at the effects of varying sources and concentrations of
macronutrients on viral dynamics hypothesized that optimal growth increased host cell
sizes which then increased viral burst sizes. For example, Hadas et al. (1997) infected
Escherichia coli cultures grown on low penicillin concentrations, which is known to block
cell division and in turn cause cells to grow larger (Hadas et al., 1995). In the larger cells
Hadas et al. (1997) found more protein-synthesizing systems per cell which they believed
were used to produce more viral progeny per infected cell (i.e. larger burst sizes). In our
study we looked at the cell sizes of V40 during exponential growth in Fe-limited (no Fe
added) versus Fe-replete (0.05 nM Fe’) media. We found no significant differences in cell
size (Table 4-1) even though we observed differences in burst sizes between the two Fe
treatments. Hence our findings contradict Hadas et al. (1997)’s conclusions and, at least in
the system studied here, we reject the hypothesis that cell size affects the burst sizes of
phage. Instead we hypothesize that the differences in burst sizes between the low Fe and
high Fe treatments was caused by the changes in efficiency of the host’s DNA replication
machinery.
4.2 Latent period in relation to Fe-limited host growth
Even though we observed differences in growth rates and burst sizes depending on
the iron levels of the V40 host cultures we did not see any differences in latent period
(Figure 4-1). Like our V40 experiments, Wilson et al. (1996) concluded that growth rates of
138
host cultures did not affect the timing of cell lysis. To show that growth rate did not affect
latent periods, Wilson et al. (1996) infected Synechococcus grown in the same media with
two different growth rates and found that both cultures lysed at the same time. Latent
period is dependent on endolysins/lysozyme (enzymes that break down host cell walls
during the final stage of the lytic cycle) synthesis and concentration, which is dictated by
the dimensions (volume and surface area) of the host cells (Hadas et al., 1997; Young et
al., 2000). Since we observed no change in cell size between the Fe deplete and replete
V40 cultures it is reasonable that the latent periods in our one-step phage growth
experiments were the same for each of the iron treatments. This may have also been the
case in Wilson et al. (1996) though they did not measure cell size in their study.
4.3 Effects on microbial diversity
The one-step phage growth experiments presented in this study indicate that
differences in burst sizes at the strain level due to changes in iron supply could have
considerable consequences on the diversity of the global marine microbial community. For
example, both Vibrio splendidus strains used in this study (V40 and V36) and their paired
phage were isolated from the same environment off the coast of Nahant, Massachusetts
by Kauffman et al. (2018). If a flux of iron (from dust, fluvial deposition, anthropogenic
sources) were to be introduced to the coastal Nahant area then larger burst sizes could
result from V40’s myovirus, therefore suppressing V40’s population. Other populations
whose burst sizes are not affected, such as V36 (Figure 4-2), could then benefit from the
increased iron and become the dominant Vibrio splendidus strain. Without an excess of
iron the burst sizes of V40 and V36 are fairly similar, therefore if adsorption rates and
139
population sizes are comparable the two strains would be competitive. If we were to
assume our findings reflect what is happening in the natural environment, we might expect
a shift in the diversity of a microbial community as a result of an increase in bioavailable Fe
not only due to bottom-up controls, but also because of top-down controls. These varying
dynamics between host cell growth and viral replication at the strain level caused by
changes in iron concentrations has particularly interesting ecological implications for most
of the world’s open oceans where dissolved Fe concentrations are on average near 0.2
nM (Hutchins & Boyd, 2016).
Generally the iron limitation of heterotrophic bacteria, like Vibrio, has only been
found in high-nutrient low-chlorophyll regions (Church et al., 2000; Tortell et al., 1996) and
not in oligotrophic waters (Kirchman et al., 2000; Zhang et al., 2019). The location of these
iron limiting regimes may shift though with the future climate as iron associated with glacial
melt (Gerringa et al., 2012; Lannuzel et al., 2007) and smoke from wildfires continues to
enter the ocean (Abatzoglou & Williams, 2016; Kelly et al., 2021), for example. Resilience
to Fe-limitation can vary amongst phytoplankton within a single environment due to the
differences in biological Fe requirements (Ho et al., 2003; Twining & Baines, 2013) and
modes of Fe uptake (Hutchins et al., 1999; Tortell et al., 1999). One adaptation to low Fe
availability in heterotrophs is the production of Fe-binding siderophores. For example,
studies have shown that Vibrio cholerea can start producing siderophores as soon as 3.5
hours after being transferred into low iron media (Sigel & Payne, 1982) and that Vibrio
splendidus can also produce siderophores under iron stress (Boiteau et al., 2016; Song et
al., 2018). Viruses have also been shown to manipulate host genes that are directly
140
involved with iron metabolism so that the host cells can acquire more iron and synthesize
nucleic acids and proteins efficiently (Drakesmith & Prentice, 2008). A similar manipulation
of the host cell machinery was seen in P-depletion experiments performed by Wilson et al.
(1996), where they detected the presence of a polypeptide in their phosphate deplete
cultures that is assumed to be a part of a high-affinity phosphate uptake system. Changes
in future Fe supply to the oceans due to changes in climate, in combination with variability
in the efficiencies of cellular Fe acquisition, due to host specific Fe-uptake mechanisms or
infection status, could alleviate or exacerbate some of the iron stress in both the host and
virus, further complicating the effects of viral infection on microbial diversity.
Changes in viral dynamics due to fluctuations in temperature were also explored in
this study through an additional one-step phage growth experiment using the V40 host
and myovirus (Figure 4-3). Even though we see the fastest growth rate at 39.5°C, the cells
at that temperature did not lyse during the 3.5 hour experiment. This observation is
somewhat contradictory to the leading hypothesis that larger burst sizes are associated
with optimal cell growth. The myoviruses thus appear to have become inactive in the
39.5°C treatment and were not able to successfully replicate and lyse the host cell. This
explanation is supported by studies which have shown that marine viruses become
inactive and start to decay at high temperatures, usually above 20°C (Finke et al., 2017;
Wei et al., 2018). Coincidentally, the average sea surface temperature of the world’s
oceans is ~20°C (Merchant et al., 2019). With projections of sea surface temperature
increasing by <1.5°C by 2050 (Freer et al., 2018; Overland & Wang, 2007) we may expect
to see more viral decay and less viral infection in the surface oceans. The offset between
141
optimal host cell growth and viral burst sizes at various temperatures further complicates
predictions of how viral dynamics will change in future climates.

5. Conclusion
The goal of this study was to better understand how metal micronutrients, such as
iron, may affect bacteriophage infection dynamics (i.e. latent period and burst size) and
how this may alter the microbial community. The results show that varying iron
concentrations can affect the burst size of some Vibrio splendidus strains, but that latent
period does not change. After comparing our results with those of past studies that looked
at changes in nutrient supplies in one-step phage growth experiments, we hypothesize
that cell sizes control latent periods whereas the efficiency of the host’s DNA replication
machinery regulates burst sizes. Iron’s direct role in DNA replication through its inclusion in
ribonucleotide reductase, pirin, DNA primase, and DNA helicases may explain why viral
burst sizes are affected by changes in cellular iron. Increased viral progeny due to
increases in iron presents large consequences for diversity and nutrient cycling within the
microbial community. Our study has highlighted how the opposing responses to iron levels
by two closely related bacteriophage suggest that many different strains of marine
heterotrophs could respond differently to Fe limitation, or relief from Fe limitation, with
implications for the diversity of the microbial community through top-down controls (viral
lysis). Resulting bottom-up controls (recycling of nutrients) post cell lysis were not
examined in this study but have large effects on global climate considering that a quarter
142
of organic carbon in the ocean is thought to pass through the “viral shunt” (Wilhelm &
Suttle, 1999).
Predictions show that the availability of iron is changing rapidly due to accelerated
acidification, stratification, warming, and deoxygenation (Hutchins & Boyd, 2016) therefore
it is crucial that we continue to study how these changes in iron will affect marine microbial
ecosystems. Future work could expand the range of hosts and viruses tested for the
impact of Fe limitation. Additional analyses such as metatranscriptomic analysis of Fe
replete and deplete infected cultures may also help to elucidate the mechanisms by which
Fe impacts viral production. Finally, most of the current ecological models apply constant
rates of viral infection across different species (which are calculated from the latent period
and burst size) due to the limited information on viral dynamics. We can improve our
understanding of how nutrients, such as iron, may affect marine viral dynamics by pairing
modeling with experimental work, making more accurate viral mortality estimations, and
constraining recycling rates of nutrients due to viral lysates.

143
Table 4-1
Cell size measurements for Vibrio host cells V40 and V36 grown in low and high iron
media. “Low Fe” signifies cultures grown in media with no added Fe, and “high Fe”
signifies cultures grown in 0.05 nM Fe’ media. Averages (μ) and standard deviations (σ) of
each measurement under each treatment are shown in the last two rows.

V40 V36
Low Fe High Fe Low Fe High Fe
Width
(µm)
Height
(µm)
Area
(µm
2
)
Width
(µm)
Height
(µm)
Area
(µm
2
)
Width
(µm)
Height
(µm)
Area
(µm
2
)
Width
(µm)
Height
(µm)
Area
(µm
2
)
0.73 2.11 5.68 0.77 2.53 7.05 0.98 3.06 10.93 1.01 2.24 8.71
0.85 2.08 6.69 0.87 3.61 11.06 1.19 1.47 7.72 0.88 1.53 5.45
0.85 1.98 6.42 1.08 2.31 9.67 1.29 2.12 11.21 1.22 1.6 8.47
0.9 2.97 9.67 1.22 2.64 12.46 1.36 2.52 13.67 0.66 1.36 3.50
0.73 1.67 4.67 0.85 2.21 7.04 0.7 1.4 3.85 0.8 1.33 4.35
0.86 2.47 7.84 0.90 3.64 11.56 1.11 1.65 7.69 0.78 1.33 4.21
0.92 2.1 7.40 1.04 2.09 8.53 1.21 2.6 12.18 1.15 1.95 9.12
0.97 3.31 11.56 0.99 2.78 10.19 1.35 3.21 16.48 0.9 1.16 4.55
0.89 3.08 9.86 1.00 1.52 6.35 0.82 1.4 4.66 1.03 1.19 5.52
1.18 2.73 12.31 0.83 2.06 6.45 0.83 1.14 4.05 0.77 1.69 5.02
µ = 0.89 µ = 2.45 µ = 8.21 µ = 0.96 µ = 2.54 µ = 9.03 µ = 1.08 µ = 2.06 µ = 9.24 µ = 0.92 µ = 1.54 µ = 5.89
σ = 0.13 σ = 0.55 σ = 2.54 σ = 0.14 σ = 0.67 σ = 2.26 σ = 0.24 σ = 0.75 σ = 4.33 σ = 0.18 σ = 0.35 σ = 2.08

144

Figure 4-1
Iron one-step phage growth curve for V40 (Vibrio splendidus 10N.261.52.C3) infected by
myovirus 1.170.O. Host cultures were grown in three different iron medias, no added Fe,
0.01 nM Fe’, and 0.06 nM Fe’ which are signified by the grey circles, yellow triangles, and
red diamonds, respectively. Latent period is signified in the left panels. Growth rate, relative
growth rate, final plaque forming units (PFU), and burst size for each treatment are all listed
in the right panels.

0
20
40
60
80
100
0 0.5 1 1.5 2 2.5 3 3.5
PFU/mL (thousands)
Time (hours)
No Added Fe
0.01 nM Fe'
0.05 nM Fe'
0
20
40
60
80
100
0 0.5 1 1.5 2 2.5 3 3.5
PFU/mL (thousands)
Time (hours)
No Added Fe
0.01 nM Fe'
0.05 nM Fe'
No Added Fe 0.01 nM Fe’ 0.05 nM Fe’
Growth Rate
(µ)
0.42 0.54 0.61
Relative
Growth Rate
(µ/µ
max
)
69% 88% 100%
Final PFU
(thousands)
6.3 ± 0.4 4.1 ± 4.4 70.6 ± 50.3
Burst Size
(PFU/cell)
11 ± 0.1 6 ± 7.8 89 ± 67.5
No Added Fe 0.01 nM Fe’ 0.05 nM Fe’
Growth Rate
(µ)
0.39 0.56 0.64
Relative
Growth Rate
(µ/µ
max
)
60% 88% 100%
Final PFU
(thousands)
4.8 ± 1.8 8.1 ± 1.2 19.7 ± 3.9
Burst Size
(PFU/cell)
16 ± 10.4 12 ± 3.9 48 ± 8.2
V40 – Vibrio splendidus 10N.261.52.C3 & myovirus 1.170.O
Latent Period = 80 min
Latent Period = 80 min
145

Figure 4-2
Iron one-step phage growth curve for V36 (Vibrio splendidus 10N.222.46.C2) infected by
podovirus 2.130.O. Host cultures were grown in three different iron medias, no added Fe,
0.01 nM Fe’, and 0.06 nM Fe’ which are signified by the grey circles, yellow triangles, and
red diamonds, respectively. Latent period is signified in the left panels. Growth rate, relative
growth rate, final plaque forming units (PFU), and burst size for each treatment are all listed
in the right panels.

0
5
10
15
20
25
0 0.5 1 1.5 2 2.5 3 3.5
PFU/mL (thousands)
Time (hours)
No Added Fe
0.01 nM Fe'
0.05 nM Fe'
0
5
10
15
20
25
0 0.5 1 1.5 2 2.5 3 3.5
PFU/mL (thousands)
Time (hours)
No Added Fe
0.01 nM Fe'
0.05 nM Fe'
No Added Fe 0.01 nM Fe’ 0.05 nM Fe’
Growth Rate
(µ)
0.20 0.25 0.25
Relative
Growth Rate
(µ/µ
max
)
82% 99% 100%
Final PFU
(thousands)
13.4 ± 4.2 7.2 ± 0.8 6 ± 0.6
Burst Size
(PFU/cell)
30 ± 18.9 12 ± 2.6 10 ± 2.6
No Added Fe 0.01 nM Fe’ 0.05 nM Fe’
Growth Rate
(µ)
0.25 0.24 0.37
Relative
Growth Rate
(µ/µ
max
)
66% 63% 100%
Final PFU
(thousands)
12.5 ± 0 15.8 ± 1 14.7 ± 0.6
Burst Size
(PFU/cell)
8 ± 1.9 14 ± 3.3 34 ± 28.5
V36 – Vibrio splendidus 10N.222.46.C2 & podovirus 2.130.O
Latent Period = 120 min
Latent Period = 120 min
146

Figure 4-3
Temperature one-step phage growth curve for V40 (Vibrio splendidus 10N.261.52.C3)
infected by myovirus 1.170.O. Host cultures were grown at three different temperatures,
7°C, 23°C and 39.5°C which are signified by the navy circles, blue triangles, and green
diamonds, respectively. Latent period is signified in the left panels. Growth rate, relative
growth rate, final plaque forming units (PFU), and burst size for each treatment are all listed
in the right panels.

0
20
40
60
80
100
0 0.5 1 1.5 2 2.5 3 3.5
PFU/mL (thousands)
Time (hours)
7°C
23°C
39.5°C
7°C 23°C 39.5°C
Growth Rate
(µ)
0.1 0.09 0.13
Relative
Growth Rate
(µ/µ
max
)
78% 67% 100%
Final PFU
(thousands)
20.5 ± 4.9 55.9 ± 20.2 0.8 ± 0.3
Burst Size
(PFU/cell)
7 ± 2.9 24 ± 9.9 0 ± 0.1
V40 – Vibrio splendidus 10N.261.52.C3 & myovirus 1.170.O
Latent Period = ~80 min
147
Chapter 5
Conclusions

1. Contributions to metal distributions through simple pole sampling
The increase in trace metal measurements since the 1970s has showcased that the
ocean’s global metal distribution is dynamic, even more so now due to the anthropogenic
impacts on Earth’s climate such as ocean acidification and deoxygenation. In order to
properly monitor these changes in metal distributions, trace metal sampling must become
even more routine on marine research expeditions. In Chapter 2, I analyzed 242 pristine
trace metal seawater samples which were collected by a non-trace metal chemist using a
simple pole sampler on the Tara Pacific expedition. Using machine-learning techniques I
show that Tara metal concentrations are generally similar to metal concentrations
measured by the GEOTRACES program. In particular, similarities in contamination-prone
Zn (0.37 ± 0.85 nM Tara, 0.56 ± 0.86 nM GEOTRACES) suggested that Tara metal data
was reliable. Predicted metal concentrations for the Tara and GEOTRACES datasets were
also within the same concentration ranges and had similar RMSE values. The study
presented in Chapter 2 highlights how a simple sampling method can capture the nuances
in the global surface metal distribution and, thus, can be implemented on future cruises
with simple pole sampling techniques which can be utilized even when a more
sophisticated GEOTRACES-style trace metal sampling infrastructure is not available.
The Tara Pacific metal data in Chapter 2 was used to identify which major
biogeochemical factors were most highly correlated for surface marine ecosystems in 21
148
marine provinces in 3 major ocean biomes. Environmental and biological data collected on
the Tara Pacific expedition and data from previous metal-focused expeditions, like the
GEOTRACES cruises, were used to distinguish and verify the trends in metal distributions
observed during the Tara cruise. Using regression tree models and previously published
studies, I identified inter- and intra-basin differences in metal concentrations amongst the
North Atlantic, North Pacific, and South Pacific Oceans. For instance, in the North Atlantic
Fe and Mn concentrations appear to be mostly impacted by Saharan dust deposition,
whereas an inverse relationship between Cu and salinity suggested North Atlantic Cu
distributions were heavily influenced by coastal proximity. In the Pacific, increased Cd off
the coast of Peru was linked to the upwelling of low-Fe waters and higher Pb
concentrations in the northern basin were connected to Asian anthropogenic aerosols.
Intra-basin differences in Co were tentatively attributed to the presence of oxygen
minimum zones in the South Pacific and increased continental weathering in the North
Atlantic. The results in Chapter 2 demonstrate how surface metal samples from a simple
pole sampler and machine learning algorithms can be used to identify the biogeochemical
factors driving global metal distributions.

2. Metal transport through increasing wildfires
The current fluxes of the major metal sources to the ocean and the resulting
solubilities of the metals associated with each source are likely to change with variabilities
in Earth’s climate and with the rise of atmospheric CO2 (Hutchins & Boyd, 2016; Jickells et
al., 2005). Wildfires are presently acknowledged as a minor source of metals to the
149
oceans, but forecasts of more wildfires and acidic oceans suggest that fire-derived metals
could have a larger impact on the future global ocean metal distribution (Millero et al.,
2009; Hoffmann et al., 2012). Chapter 3 presented one of the first studies to quantify the
fluxes of fire-derived metals, observe the subsequent changes in surface seawater
composition, and measure the biological response in the coastal ocean. Such studies, like
the Thomas Fire study presented in this thesis, are needed to improve our understanding
on what the transportation of metals will look like in the future global ocean.
The results in Chapter 3 revealed that metals, along with black carbon, were
released by the 2017 Thomas Fire and transported by atmospheric and fluvial deposition
to the coastal ocean. Field measurements showed that the fire-derived metals transported,
mainly through atmospheric deposition, to the coastal ocean had little to no impact on the
total metal inventories in the Santa Barbara Basin. Conversely, the dissolved black carbon
concentrations in the coastal ocean did increase due to the Thomas Fire. Even though we
did not observe an immediate effect on the Santa Barbara Basin, previous studies indicate
that fluvial transport can continue to bring black carbon and metals from a burned
watershed into the coastal ocean for many years after a fire (Meixner et al., 2006; Minshall
et al., 1989; Moody et al., 2013; Warrick et al., 2012). If the projections of a more acidic
ocean materialize, we may expect to see an increased solubility and bioavailability of these
fire-derived metals due to the greater dissolution of metals from fire aerosols. The
increases in soluble metals would then affect marine productivity as elevated metals can
serve as nutrients or be toxic to marine organisms. With the potential increase in size and
frequency of wildfires in the future, studies such as the work I have presented in Chapter 3
150
are needed to understand how these fires will impact nearby ecosystems, such as the
coastal ocean.

3. Utilization of Fe in viral infection
Most of the literature that focuses on the impact of metals, such as Fe, on marine
ecosystems emphasize how metals are essential to phytoplankton as micronutrients and
therefore metals act as bottom-up controls (Boyd & Ellwood, 2010; Price & Morel, 1998).
The role of metals as top-down controls has only recently been explored by a few studies
looking at the effects of Fe, in particular, on grazing and viral infection (Bonnain et al.,
2016; Kranzler et al., 2021, 2019). Chapter 4 featured some of the first experiments to
look at the effects of Fe on the viral infection of marine heterotrophs. The Appendix then
focused on the comparative bioavailability of Fe associated with the resulting cell-debris
produced by viral lysis and grazing versus siderophore-bound Fe.
In Chapter 4, I showed that Fe had different effects on the viral lysis of two closely
related heterotrophic marine bacteria. High Fe levels (0.05 nM Fe’) in a culture of one Vibrio
splendidus strain (10N.261.52.C3) caused its paired myovirus (1.170.O) to produce more
viral progeny or, in other words, increase in burst size. In contrast, the burst size of a
podovirus (2.130.O) that infected a separate strain of Vibrio splendidus (10N.222.46.C2)
was unaffected by Fe concentrations. Previous work in the biomedical field suggest that
Fe affects the efficiency of the host cell’s DNA replication machinery, which in turn can
impact burst sizes (Drakesmith & Prentice, 2008; Fuss et al., 2015). Thus, Fe’s role in DNA
replication may explain why myovirus 1.170.O’s burst sizes increased with cellular Fe.
151
Results from the Chapter 4 experiments also showed that the time in which the viral
progeny lysed from the host cells, otherwise known as the latent period, was unaltered by
the changes in Fe for both host-phage pairs. Simultaneously, cell sizes of both host
cultures also did not change with Fe. The lack of fluctuations in latent period and cell size
suggests that the cell size of an infected cell controls the latent period of the virus. The
opposing responses to Fe availability in the two host-phage pairs indicates that marine
viruses have diverse responses to Fe limitation or additions. These various viral responses
to Fe imply that Fe availability affects the diversity of the microbial community not only
through bottom-up controls, but also top-down controls. The results in Chapter 4 show
that the effects of nutrients on top-down controls should be studied further especially
since top-down controls inevitably affect bottom-up controls through the production of
nutrient-rich cell debris.
The relative bioavailability of Fe produced by cell lysis, grazing, and bound to
siderophores was explored in incubation experiments described in the Appendix. Stable
iron isotope tracers (
54
Fe,
57
Fe,
58
Fe) were bound to lysates, byproducts of grazing, and
siderophores and then incubated with the natural phytoplankton community from the
North Pacific Subtropical Gyre. The samples from these incubation experiments did not
show any clear signs of biological uptake of the labeled Fe. The Appendix highlights weak
points in the experimental setup and provides recommendations for future similar
experiments. Some suggestions include using a naturally produced L2-type siderophore
and concentrating the labeled Fe stocks before adding them to the incubations to
minimize the transfer of labeled Fe amongst ligands, as well as increasing incubation
152
biomass. Future similar experiments should also consider performing preliminary
incubations with laboratory cultures or natural communities from productive waters (e.g.
coastal waters) and incubating for longer than 24 hours to promote the biological uptake
of Fe. Such experiments are potentially valuable because the recycling of Fe amongst
microorganisms is thought to be crucial for the persistence of productivity in oligotrophic
or Fe-limited regions (Hutchins et al., 1993; Landry et al., 1997; Rafter et al., 2017).

4. Thesis summary
The work presented in this thesis builds upon the findings from the past 50 years of
trace metal research and explores new connections between marine metals and
phytoplankton. The dataset presented in Chapter 2 fills gaps in the global surface metal
distributions and uses novel machine-learning tools demonstrate the abundance of
information on surface metal cycling that can be produced through simple pole sampling.
Chapter 3 presents a first of its kind study on a potentially large source of marine trace
metals, a fire-flood sequence, to the future ocean. Chapter 4 reveals how fluctuations in
metal concentrations, specifically for Fe, not only affect bottom-up controls, but also top-
down controls such as viral lysis. Finally, the Appendix describes a novel approach to
measure the relative bioavailability of various recycled Fe sources to a natural marine
community. This thesis highlights how far research on trace metal chemistry in the marine
environments has come since the 1970s, while also showing how much remains to be
discovered. Metals in the ocean may have low concentrations, but they have a large
impact on Earth’s major biogeochemical cycles.
153
References

Abatzoglou, J. T., & Williams, A. P. (2016). Impact of anthropogenic climate change on
wildfire across western US forests. Proceedings of the National Academy of
Sciences, 113(42), 11770–11775. https://doi.org/10.1073/pnas.1607171113

Abraham, J., Dowling, K., & Florentine, S. (2017). Risk of post-fire metal mobilization into
surface water resources: A review. Science of the Total Environment, 599–600,
1740–1755. https://doi.org/10.1016/j.scitotenv.2017.05.096

Albrecht-Gary, A. M., & Crumbliss, A. L. (1998). Coordination chemistry of siderophores:
thermodynamics and kinetics of iron chelation and release. In Metal ions in
biological systems (pp. 239–327).

Alldredge, A. L. (2000). Interstitial dissolved organic carbon (DOC) concentrations within
sinking marine aggregates and their potential contribution to carbon flux. Limnology
and Oceanography, 45(6), 1245–1253. https://doi.org/10.4319/lo.2000.45.6.1245

Andersen, R. A. (Ed.). (2005). Algal Culturing Techniques. Elsevier Academic Press.
Elsevier.

Annett, A. L., Lapi, S., Ruth, T. J., & Maldonado, M. T. (2008). The effects of Cu and Fe
availability on the growth and Cu:C ratios of marine diatoms. Limnology and
Oceanography, 53(6), 2451–2461. https://doi.org/10.4319/lo.2008.53.6.2451

Aparicio-González, A., Duarte, C. M., & Tovar-Sánchez, A. (2012). Trace metals in deep
cean waters: A review. Journal of Marine Systems, 100, 26–33.
https://doi.org/10.1016/j.jmarsys.2012.03.008

Azam, F., Fenchel, T., Field, J. G., Gray, J. S., Meyer-Reil, L. A., & Thingstad, F. (1983).
The Ecological Role of Water-Column Microbes in the Sea. Marine Ecology
Progress Series, 257–263. https://doi.org/10.3354/meps010257

Baker, A. R., Jickells, T. D., Witt, M., & Linge, K. L. (2006). Trends in the solubility of iron,
aluminium, manganese and phosphorus in aerosol collected over the Atlantic
Ocean. Marine Chemistry, 98(1), 43–58.
https://doi.org/10.1016/j.marchem.2005.06.004

Baker, A. R., Landing, W. M., Bucciarelli, E., Cheize, M., Fietz, S., Hayes, C. T., … Van
Hulten, M. M. P. (2016). Trace element and isotope deposition across the air-sea
interface: Progress and research needs. Philosophical Transactions of the Royal
Society A: Mathematical, Physical and Engineering Sciences, 374(2081),
20160190. https://doi.org/10.1098/rsta.2016.0190
154

Bao, H., Niggemann, J., Huang, D., Dittmar, T., & Kao, S. J. (2019). Different Responses
of Dissolved Black Carbon and Dissolved Lignin to Seasonal Hydrological Changes
and an Extreme Rain Event. Journal of Geophysical Research: Biogeosciences,
124(3), 479–493. https://doi.org/10.1029/2018JG004822

Bar-on, Y. M., Phillips, R., & Milo, R. (2018). The biomass distribution on Earth.
Proceedings of the National Academy of Sciences, 115(25), 6506–6511.
https://doi.org/10.1073/pnas.1711842115

Barbeau, K., Moffett, J. W., Caron, D. A., Croot, P. L., & Erdner, D. L. (1996). Role of
protozoan grazing in relieving iron limitation of phytoplankton. Nature, 380, 61-64.
https://doi.org/10.1038/380061a0

Barbeau, K., Rue, E. L., Trick, C. G., Bruland, K. W., & Butler, A. (2003). Photochemical
reactivity of siderophores produced by marine heterotrophic bacteria and
cyanobacteria based on characteristic Fe(III) binding groups. Limnology and
Oceanography, 48(3), 1069–1078. https://doi.org/10.4319/lo.2003.48.3.1069

Basu, S., & Mackey, K. R. M. (2018). Phytoplankton as key mediators of the biological
carbon pump: Their responses to a changing climate. Sustainability, 10(3), 869.
https://doi.org/10.3390/su10030869

Beardall, J., Young, E., & Roberts, S. (2001). Approaches for determining phytoplankton
nutrient limitation. Aquatic Sciences, 63(1), 44–69.
https://doi.org/10.1007/PL00001344

Bertrand, E. M., Saito, M. A., Rose, J. M., Riesselman, C. R., Lohan, M. C., Noble, A. E.,
… DiTullio, G. R. (2007). Vitamin B12 and iron colimitation of phytoplankton growth
in the Ross Sea. Limnology and Oceanography, 52(3), 1079–1093.
https://doi.org/10.4319/lo.2007.52.3.1079

Bewers, J. M., & Yeats, P. A. (1989). Transport of river-derived trace metals through the
coastal zone. Netherlands Journal of Sea Research, 23(4), 359-368.
https://doi.org/10.1016/0077-7579(89)90020-3

Biller, D. V., & Bruland, K. W. (2013). Sources and distributions of Mn, Fe, Co, Ni, Cu, Zn,
and Cd relative to macronutrients along the central California coast during the
spring and summer upwelling season. Marine Chemistry, 155, 50–70.
https://doi.org/10.1016/j.marchem.2013.06.003

155
Bodí, M. B., Martin, D. A., Balfour, V. N., Santín, C., Doerr, S. H., Pereira, P., ... Mataix
Solera, J. (2014). Wildland fire ash: production, composition and eco-hydro
geomorphic effects. Earth-Science Reviews, 130, 103-127.
http://dx.doi.org/10.1016/j.earscirev.2013.12.007

Boiteau, R. ., Mende, D. ., Hawco, N. J., Mcilvin, M. R., & Fitzsimmons, J. . (2016).
Siderophore-based microbial adaptations to iron scarcity across the eastern Pacific
Ocean. Proceedings of the National Academy of Sciences, 113(50), 14237–14242.
https://doi.org/10.1073/pnas.1608594113

Bonnain, C., Breitbart, M., & Buck, K. N. (2016). The Ferrojan horse hypothesis: Iron-virus
interactions in the ocean. Frontiers in Marine Science, 3(82), 1–11.
https://doi.org/10.3389/fmars.2016.00082

Bowie, A. R., Whitworth, D. J., Achterberg, E. P., Mantoura, R. F. C., & Worsfold, P. J.
(2002). Biogeochemistry of Fe and other trace elements (Al, Co, Ni) in the upper
Atlantic Ocean. Deep-Sea Research Part I: Oceanographic Research Papers, 49(4),
605–636. https://doi.org/10.1016/S0967-0637(01)00061-9

Boyd, P. W., & Ellwood, M. J. (2010). The biogeochemical cycle of iron in the ocean.
Nature Geoscience, 3(10), 675–682. https://doi.org/10.1038/ngeo964

Boyd, P. W., Jickells, T., Law, C. S., Blain, S., Boyle, E. A., Buesseler, K. O., … Watson,
A. J. (2007). Mesoscale iron enrichment experiments 1993-2005: Synthesis and
future directions. Science, 315(5812), 612–617.
https://doi.org/10.1126/science.1131669

Boyd, P. W., Mackie, D. S., & Hunter, K. A. (2010). Aerosol iron deposition to the surface
ocean - Modes of iron supply and biological responses. Marine Chemistry, 120(1
4), 128–143. https://doi.org/10.1016/j.marchem.2009.01.008

Boyle, E. A., Bergquist, B. A., Kayser, R. A., & Mahowald, N. (2005). Iron, manganese, and
lead at Hawaii Ocean Time-series station ALOHA: Temporal variability and an
intermediate water hydrothermal plume. Geochimica et Cosmochimica Acta, 69(4),
933–952. https://doi.org/10.1016/j.gca.2004.07.034

Boyle, E. A., Huested, S. S., & Jones, S. P. (1981). On the distribution of copper, nickel,
and cadmium in the surface waters of the North Atlantic and North Pacific Ocean.
Journal of Geophysical Research, 86, 8048–8066.
https://doi.org/10.1029/jc086ic09p08048

Boyle, E. A., Sclater, F. R., & Edmond, J. M. (1977). The distribution of dissolved copper in
the Pacific. Earth and Planetary Science Letters, 37(1), 38–54.
https://doi.org/10.1016/0012-821X(77)90144-3
156

Brand, L. E., Sunda, W. G., & Guillard, R. R. L. (1983). Limitation of marine phytoplankton
reproductive rates by zinc, manganese, and iron. Limnology and Oceanography,
28(6), 1182–1198. https://doi.org/10.4319/lo.1983.28.6.1182

Brand, L. E., Sunda, W. G., & Guillard, R. R. L. (1986). Reduction of marine phytoplankton
reproduction rates by copper and cadmium. Journal of Experimental Marine Biology
and Ecology, 96(3), 225–250. https://doi.org/10.1016/0022-0981(86)90205-4

Bratbak, G., Egge, J. K., & Heldal, M. (1993). Viral mortality of the marine alga Emiliania
huxleyi (Haptophyceae) and termination of algal blooms. Marine Ecology Progress
Series, 93, 39–48. https://www.jstor.org/stable/24832806

Bratbak, G., Jacobsen, A., Heldal, M., Nagasaki, K., & Thingstad, F. (1998). Virus
production in Phaeocystis pouchetii and its relation to host cell growth and nutrition.
Aquatic Microbial Ecology, 16(1), 1–9. https://doi.org/10.3354/ame016001

Brito, D. Q., Passos, C. J. S., Muniz, D. H. F., & Oliveira-Filho, E. C. (2017). Aquatic
ecotoxicity of ashes from Brazilian savanna wildfires. Environmental Science and
Pollution Research, 24(24), 19671–19682.
https://doi.org/10.1007/s11356-0179578-0

Bruland, K. W., & Lohan, M. C. (2013). Controls of Trace Metals in Seawater. In Treatise
on Geochemistry: Second Edition (pp. 23–47).
https://doi.org/10.1016/B978-0-08-095975-7.00602-1

Bruland, K. W. (1980). Oceanographic distributions of cadmium, zinc, nickel, and
copper in the North Pacific. Earth and Planetary Science Letters, 47(2), 176–198.
https://doi.org/10.1016/0012-821X(80)90035-7

Bruland, K. W., Rue, E. L., Smith, G. J., & DiTullio, G. R. (2004). Iron, macronutrients
and diatom blooms in the Peru upwelling regime: Brown and blue waters of Peru.
Marine Chemistry, 93(2–4), 81–103.
https://doi.org/10.1016/j.marchem.2004.06.011

Buck, C. S., Landing, W. M., & Resing, J. (2013). Pacific Ocean aerosols: Deposition and
solubility of iron, aluminum, and other trace elements. Marine Chemistry, 157, 117
130. https://doi.org/10.1016/j.marchem.2013.09.005

Buck, C. S., Landing, W. M., Resing, J. A., & Lebon, G. T. (2006). Aerosol iron and
aluminum solubility in the northwest Pacific Ocean: Results from the 2002 IOC
cruise. Geochemistry, Geophyics, Geosystems, 7(4), 1–21.
https://doi.org/10.1029/2005GC000977

157
Buck, K. N., Selph, K. E., & Barbeau, K. A. (2010). Iron-binding ligand production and
copper speciation in an incubation experiment of Antarctic Peninsula shelf waters
from the Bransfield Strait, Southern Ocean. Marine Chemistry, 122(1–4), 148–159.
https://doi.org/10.1016/j.marchem.2010.06.002

Bundy, R. M., Boiteau, R. M., McLean, C., Turk-Kubo, K. A., McIlvin, M. R., Saito, M. A.,
… Repeta, D. J. (2018). Distinct Siderophores Contribute to Iron Cycling in the
Mesopelagic at Station ALOHA. Frontiers in Marine Science, 5(61), 1–15.
https://doi.org/10.3389/fmars.2018.00061

Burke, M. P., Hogue, T. S., Kinoshita, A. M., Barco, J., Wessel, C., & Stein, E. D. (2013).
Pre and post-fire pollutant loads in an urban fringe watershed in Southern
California. Environmental Monitoring Assessment, 185(12), 10131–10145.
https://doi.org/10.1007/s10661-013-3318-9

Burton, C. A., Hoefen, T. M., Plumlee, G. S., Baumberger, K. L., Backlin, A. R., Gallegos,
E., & Fisher, R. N. (2016). Trace Elements in Stormflow, Ash, and Burned Soil
following the 2009 Station Fire in Southern California. PLoS ONE, 11(5), 1–26.
https://doi.org/10.1371/journal.pone.0153372

Butkus, D., & Baltrėnaitė, E. (2007). Transport of heavy metals from soil to Pinus sylvestris
L. and Betula pendula trees. Ekologija, 53(1), 29–36.

Cai, W., Bi, D., Church, J., Cowan, T., Dix, M., & Rotstayn, L. (2006). Pan-oceanic
response to increasing anthropogenic aerosols: Impacts on the Southern
Hemisphere oceanic circulation. Geophysical Research Letters, 33(21), 1–5.
https://doi.org/10.1029/2006GL027513

Cal Fire. (2018). Top 20 Most Destructive California Wildfires, 2019. Retrieved from
http://www.fire.ca.gov/communications/downloads/fact_sheets/Top20_Destructio
.pdf

CalCOFI. (2018). Hydrographic Data Reports. Retrieved from
https://www.calcofi.org/data/67-hydrographic/450-hydrographic-data
report.html?&modjemcal_id=292&modjemcal_month=5&modjemcal_year=2018

California Air Resources Board. (2014). Standard Operating Procedure for the Analysis of
Levoglucosan, Mannosan, and Galactosan in Ambient Air Samples Using Gas
Chromatography/Mass Spectrometry (SOP MLD073). Retrieved from
https://ww2.arb.ca.gov/sites/default/files/2020-01/mld073_1.pdf

California Air Resources Board. (2020). iADAM: Air Quality Data Statistics. Retrieved from
https://www.arb.ca.gov/adam/index.html

158
Campbell, L., & Vaulot, D. (1993). Photosynthetic picoplankton community structure in the
subtropical North Pacific Ocean near Hawaii (station ALOHA). Deep-Sea Research
Part I, 40(10), 2043–2060. https://doi.org/10.1016/0967-0637(93)90044-4

Campos, I., Abrantes, N., Vidal, T., Bastos, A. C., Gonçalves, F., & Keizer, J. J. (2012).
Assessment of the toxicity of ash-loaded runoff from a recently burnt eucalypt
plantation. European Journal of Forest Research, 131(6), 1889–1903.
https://doi.org/10.1007/s10342-012-0640-7

Capone, D. G., & Hutchins, D. A. (2013). Microbial biogeochemistry of coastal upwelling
regimes in a changing ocean. Nature Geoscience, 6(9), 711–717.
https://doi.org/10.1038/ngeo1916

Chadd, H. E., Newman, J., Mann, N. H., & Carr, N. G. (1996). Identification of iron
superoxide dismutase and a copper/zinc superoxide dismutase enzyme activity
within the marine cyanobacterium Synechococcus sp. WH 7803. FEMS
Microbiology Letters, 138(2–3), 161–165.
https://doi.org/10.1016/0378-1097(96)00096-1

Charette, M. A., Lam, P. J., Lohan, M. C., Kwon, E. Y., Hatje, V., Jeandel, C., … Garcia
Orellana, J. (2016). Coastal ocean and shelf-sea biogeochemical cycling of trace
elements and isotopes: Lessons learned from GEOTRACES. Philosophical
Transactions of the Royal Society A: Mathematical, Physical and Engineering
Sciences, 374(2081). https://doi.org/10.1098/rsta.2016.0076

Charette, T., & Prepas, E. E. (2003). Wildfire impacts on phytoplanktoon communities of
three small lakes on the Boreal Plain, Alberta, Canada: a paleolimnological study.
Canadian Journal of Fisheries and Aquatic Sciences, 60, 584–593.
https://doi.org/10.1139/F03-049

Chavez, F. P., & Messié, M. (2009). A comparison of Eastern Boundary Upwelling
Ecosystems. Progress in Oceanography, 83(1–4), 80–96.
https://doi.org/10.1016/j.pocean.2009.07.032

Chien, C., Mackey, K. R. M., Dutkiewicz, S., Mahowald, N., Prospero, J. M., & Paytan, A.
(2016). Effects of African dust deposition on phytoplankton in the western tropical
Atlantic Ocean off Barbados. Global Biogeochemical Cycles, 30, 716–734.
https://doi.org/10.1002/2015GB005334

Church, M. J., Hutchins, D. A., & Ducklow, H. W. (2000). Limitation of bacterial growth by
dissolved organic matter and iron in the Southern Ocean. Applied and
Environmental Microbiology, 66(2), 455–466.
https://doi.org/10.1128/AEM.66.2.455-466.2000

159
Cloete, R., Loock, J. C., van Horsten, N. R., Fietz, S., Mtshali, T. N., Planquette, H., &
Roychoudhury, A. N. (2021). Winter Biogeochemical Cycling of Dissolved and
Particulate Cadmium in the Indian Sector of the Southern Ocean (GEOTRACES
GIpr07 Transect). Frontiers in Marine Science, 8.
https://doi.org/10.3389/fmars.2021.656321

Contreras-Porcia, L., Dennett, G., González, A., Vergara, E., Medina, C., Correa, J. A., &
Moenne, A. (2011). Identification of Copper-Induced Genes in the Marine Alga Ulva
compressa (Chlorophyta). Marine Biotechnology, 13(3), 544–556.
https://doi.org/10.1007/s10126-010-9325-8

Conway, T. M., Horner, T. J., Plancherel, Y., & González, A. G. (2021). A decade of
progress in understanding cycles of trace elements and their isotopes in the
oceans. Chemical Geology, 580.
https://doi.org/10.1016/j.chemgeo.2021.120381

Conway, T. M., & John, S. G. (2014). Quantification of dissolved iron sources to the North
Atlantic Ocean. Nature, 511(7508), 212–215.
https://doi.org/10.1038/nature13482

Copernicus Atmosphere Monitoring Service. (2020). Particulate matter forecasts. Retrieved
from https://atmosphere.copernicus.eu/charts/cams/particulate-matter
forecasts?facets=undefined&time=2020081000,3,2020081003&projection=classi
al_global&layer_name=composition_pm2p5

Coppola, A. I., & Druffel, E. R. M. (2016). Cycling of black carbon in the ocean.
Geophysical Research Letters, 43(9), 4477–4482.
https://doi.org/10.1002/2016GL068574

Coppola, A. I., Ziolkowski, L. A., Masiello, C. A., & Druffel, E. R. (2014). Aged black carbon
in marine sediments and sinking particles. Geophysical Research Letters, 41(7),
2427-2433. https://doi.org/10.1002/ 2013GL059068

County of Santa Barbara. (2018). Thomas Fire Information and Updates. Retrieved from
https://www.countyofsb.org/thomasfire.sbc

Crompton, T. R. (2006). Analysis of seawater: A guide for the analytical and environmental
chemist. Analysis of Seawater. Springer Science & Business Media.
https://doi.org/10.1007/b138036

Croot, P. L., Baars, O., & Streu, P. (2011). The distribution of dissolved zinc in the Atlantic
sector of the Southern Ocean. Deep-Sea Research Part II: Topical Studies in
Oceanography, 58(25–26), 2707–2719. https://doi.org/10.1016/j.dsr2.2010.10.041

160
Cunningham, B. R. (2015). Viruses and metals in ocean food webs: top-down and
bottom-up effects of marine viruses and trace elements on marine
picophytoplankton. ProQuest LLC. University of South Carolina.

Cunningham, B. R., & John, S. G. (2017). The effect of iron limitation on cyanobacteria
major nutrient and trace element stoichiometry. Limnology and Oceanography,
62(2), 846–858. https://doi.org/10.1002/lno.10484

Cutter, G. A., & Bruland, K. W. (2012). Rapid and noncontaminating sampling system for
trace elements in global ocean surveys. Limnology and Oceanography: Methods,
10, 425–436. https://doi.org/10.4319/lom.2012.10.425

de Jong, J. T. M., Boyé, M., Gelado-Caballero, M. D., Timmermans, K. R., Veldhuis, M. J.
W., Nolting, R. F., … de Baar, H. J. W. (2007). Inputs of iron, manganese and
aluminium to surface waters of the Northeast Atlantic Ocean and the European
continental shelf. Marine Chemistry, 107(2), 120–142.
https://doi.org/10.1016/j.marchem.2007.05.007

de Oliveira Alves, N., Brito, J., Caumo, S., Arana, A., de Souza Hacon, S., Artaxo, P., …
de Castro Vasconcellos, P. (2015). Biomass burning in the Amazon region: Aerosol
source apportionment and associated health risk assessment. Atmospheric
Environment, 120, 277–285. https://doi.org/10.1016/j.atmosenv.2015.08.059

Desboeufs, K. V., Losno, R., & Colin, J. L. (2001). Factors influencing aerosol solubility
during cloud processes. Atmospheric Environment, 35(20), 3529–3537.
https://doi.org/10.1016/S1352-2310(00)00472-6

Degens, E. T. (1969). Biogeochemistry of Stable Isotopes. In G. Eglinton & M. T. J. Murphy
(Eds.), Organic Geochemistry (1st ed., pp. 304–328). Springer-Verlag Berlin
Heidelberg. https://doi.org/10.1038/227634c0

Desboeufs, K. V., Sofikitis, A., Losno, R., Colin, J. L., & Ausset, P. (2005). Dissolution and
solubility of trace metals from natural and anthropogenic aerosol particulate matter.
Chemosphere, 58(2), 195–203.
https://doi.org/10.1016/j.chemosphere.2004.02.025

Ding, Y., Yamashita, Y., Dodds, W. K., & Jaffé, R. (2013). Dissolved black carbon in
grassland streams: Is there an effect of recent fire history? Chemosphere, 90(10),
2557–2562. https://doi.org/10.1016/j.chemosphere.2012.10.098

Dittmar, T., De Rezende, C. E., Manecki, M., Niggemann, J., Coelho Ovalle, A. R.,
Stubbins, A., & Bernardes, M. C. (2012). Continuous flux of dissolved black carbon
from a vanished tropical forest biome. Nature Geoscience, 5(9), 618–622.
https://doi.org/10.1038/ngeo1541
161

Dittmar, T., Koch, B., Hertkorn, N., & Kattner, G. (2008). A simple and efficient method for
the solid-phase extraction of dissolved organic matter (SPE-DOM) from seawater.
Limnology and Oceanography, 6(6), 230–235.
https://doi.org/10.4319/lom.2008.6.230

Drakesmith, H., & Prentice, A. (2008). Viral infection and iron metabolism. Nature Reviews
Microbiology, 6(7), 541–552. https://doi.org/10.1038/nrmicro1930

Duce, R. A., Arimoto, R., Ray, B. J., Unni, C. K., & Harder, P. J. (1983). Atmospheric Trace
Elements at Enewetak Atoll: 1, Concentrations, Sources, and Temporal Variability.
Journal of Geophysical Research, 88(C9), 5321–5342.

Eakins, B. W., & Sharman, G. F. (2010). Volumes of the World’s Oceans from ETOPO1.

Earl, S. R., & Blinn, D. W. (2003). Effects of wildfire ash on water chemistry and biota in
south-western U.S.A. streams. Freshwater Biology, 48(6), 1015–1030.
https://doi.org/10.1046/j.1365-2427.2003.01066.x

Echegoyen, Y., Boyle, E. A., Lee, J. M., Gamo, T., Obata H., & Norisuye, K. (2014). Recent
distribution of lead in the Indian Ocean reflects the impact of regional emissions.
Proceedings of the National Academy of Sciences of the United States of America,
111(43), 15328-15331. https://doi.org/10.1073/pnas.1417370111

Eldridge, M. L., Trick, C. G., Alm, M. B., DiTullio, G. R., Rue, E. L., Bruland, K. W., …
Wilhelm, S. W. (2004). Phytoplankton community response to a manipulation of
bioavailable iron in HNLC waters of the subtropical Pacific Ocean. Aquatic Microbial
Ecology, 35(1), 79–91. https://doi.org/10.3354/ame035079

Ellis, E. L., & Delbruck, M. (1939). The Growth of Bacteriophage. Journal of General
Physiology, 22(3), 365–384. https://doi.org/10.1085/jgp.22.3.365

Elrod, V. A., Berelson, W. M., Coale, K. H., & Johnson, K. S. (2004). The flux of iron from
continental shelf sediments: A missing source for global budgets. Geophysical
Research Letters, 31(12), 2–5. https://doi.org/10.1029/2004GL020216

Emery, K. O. (1960). The Sea Off Southern California, A Modern Habitat of Petroleum.
John Wiley & Sons, Inc. https://doi.org/10.5408/0022-1368-8.2.69

EPA. (2012). Revised air quality standards for particle pollution and updates to the air
quality index (AQI).

162
Field, C. B., Behrenfeld, M. J., Randerson, J. T., & Falkowski, P. (1998). Primary
production of the biosphere: Integrating terrestrial and oceanic components.
Science, 281(5374), 237–240. https://doi.org/10.1126/science.281.5374.237

Finke, J. F., Hunt, B. P. V., Winter, C., Carmack, E. C., & Suttle, C. A. (2017). Nutrients
and other environmental factors influence virus abundances across oxic and
hypoxic marine environments. Viruses, 9(6), 1–15.
https://doi.org/10.3390/v9060152

Flores, J. M., Bourdin, G., Bourdin, G., Altaratz, O., Trainic, M., Lang-Yona, N., … Koren, I.
(2020). Tara Pacific Expedition’s atmospheric measurements of marine aerosols
across the Atlantic and Pacific Oceans: Overview and preliminary results. Bulletin of
the American Meteorological Society, 101(5), E536–E554.
https://doi.org/10.1175/BAMS-D-18-0224.1

Forchhammer, G. (1865). On the Composition of Sea-water in the different parts of the
Ocean. Philosophical Transactions of the Royal Society of London, 155, 203–262.
https://www.jstor.org/stable/108888

Fovell, R., & Gallagher, A. (2018). Winds and Gusts during the Thomas Fire. Fire, 1(3), 47.
https://doi.org/10.3390/fire1030047

Freer, J. J., Partridge, J. C., Tarling, G. A., Collins, M. A., & Genner, M. J. (2018).
Predicting ecological responses in a changing ocean: the effects of future climate
uncertainty. Marine Biology, 165(1), 1–18.
https://doi.org/10.1007/s00227-0173239-

Fu P., Kawamura K., & Miura K. (2011). Molecular characterization of marine organic
aerosols collected during a round-the-world cruise. Journal of Geophysical
Research: Atmospheres, 116(13), 1–14. https://doi.org/10.1029/2011JD015604

Fuhrman, J. A. (1999). Marine viruses and their biogeochemical and ecological effects.
Nature, 399(6736), 541–548. https://doi.org/10.1038/21119

Fuss, J. O., Tsai, C. L., Ishida, J. P., & Tainer, J. A. (2015). Emerging critical roles of Fe-S
clusters in DNA replication and repair. Biochimica et Biophysica Acta – Molecular
Cell Research, 1853(6), 1253–1271. https://doi.org/10.1016/j.bbamcr.2015.01.018

Gallaher, B. M., & Koch, R. J. (2004). Cerro Grande Fire Impacts to Water Quality and
Stream Flow near Los Alamos National Laboratory: Results of Four Years of
Monitoring. Retrieved from
http://www.lanl.gov/environment/air/docs/cgf/LA14177.pdf

163
Gallon, C., Ranville, M. A., Conaway, C. H., Landing, W. M., Buck, C. S., Morton, P. L., &
Flegal, A. R. (2011). Asian industrial lead inputs to the north pacific evidenced by
lead concentrations and isotopic compositions in surface waters and aerosols.
Environmental Science and Technology, 45(23), 9874–9882.
https://doi.org/10.1021/es2020428

Gartman, A., & Findlay, A. J. (2020). Impacts of hydrothermal plume processes on oceanic
metal cycles and transport. Nature Geoscience, 13(6), 396–402.
https://doi.org/10.1038/s41561-020-0579-0

GEOTRACES Intermediate Data Product Group. (2021). The GEOTRACES Intermediate
Data Product 2021 (IDP2021). NERC EDS British Oceanographic Data Centre
NOC. https://doi.org/10.5285/cf2d9ba9-d51d-3b7c-e053-8486abc0f5fd

German, C. R., & Angel, M. V. (1995). Hydrothermal fluxes of metals to the oceans: A
comparison with anthropogenic discharge. Geological Society Special Publication,
87(87), 365–372. https://doi.org/10.1144/GSL.SP.1995.087.01.28

Gerringa, L. J. A., Alderkamp, A. C., Laan, P., Thuróczy, C. E., De Baar, H. J. W., Mills, M.
M., … Arrigo, K. R. (2012). Iron from melting glaciers fuels the phytoplankton
blooms in Amundsen Sea (Southern Ocean): Iron biogeochemistry. Deep-Sea
Research Part II: Topical Studies in Oceanography, 71–76, 16–31.
https://doi.org/10.1016/j.dsr2.2012.03.007

Gledhill, M., Devez, A., Highfield, A., Singleton, C., Achterberg, E. P., & Schroeder, D.
(2012). Effect of metals on the lytic cycle of the Coccolithovirus, EhV86. Frontiers in
Microbiology, 3, 1–10. https://doi.org/10.3389/fmicb.2012.00155

Gledhill, M., & van den Berg, C. M. G. (1994). Determination of complexation of iron(III) with
natural organic complexing ligands in seawater using cathodic stripping
voltammetry. Marine Chemistry, 47(1), 41–54.
https://doi.org/10.1016/0304-4203(94)90012-4

Gobler, C. J., Hutchins, D. A., Fisher, N. S., Cosper, E. M., Sanudo-Wilhelmy, S. A., &
Sañudo-Wilhelmy, S. A. (1997). Release and bioavailability of C, N, P, Se, and Fe
following viral lysis of a marine chrysophyte. Limnology and Oceanography, 42(7),
1492–1504. https://doi.org/10.4319/lo.1997.42.7.1492

Godsey, S. E., Kirchner, J. W., & Clow, D. W. (2009). Concentration-discharge
relationships reflect chemostatic characteristics of US catchments. Hydrolological
Processes, 23, 1844–1864. https://doi.org/10.1002/hyp

Goldberg, E. D. (1958). The processes regulating the composition of sea water. Journal of
Chemical Education, 35(3), 113–115. https://doi.org/10.1021/ed035p116
164

Gorsky, G., Bourdin, G., Lombard, F., Pedrotti, M. L., Audrain, S., Bin, N., … Karsenti, E.
(2019). Expanding Tara Oceans Protocols for Underway, Ecosystemic Sampling of
the Ocean-Atmosphere Interface During Tara Pacific Expedition (2016–2018).
Frontiers in Marine Science, 6. https://doi.org/10.3389/fmars.2019.00750

Goss, M., Swain, D. L., Abatzoglou, J. T., Sarhadi, A., Kolden, C. A., Williams, A. P., &
Diffenbaugh, N. S. (2020). Climate change is increasing the likelihood of extreme
autumn wildfire conditions across California. Environmental Research Letters, 15(9),
094016. https://doi.org/10.1088/1748-9326/ab83a7

Guieu, C., Bonnet, S., Petrenko, A., Menkes, C., Chavagnac, V., Desboeufs, K., …
Moutin, T. (2018). Iron from a submarine source impacts the productive layer of the
Western Tropical South Pacific (WTSP). Scientific Reports, 8(1), 1–9.
https://doi.org/10.1038/s41598-018-27407-z

Guieu, C., Bonnet, S., Wagener, T., & Loÿe-Pilot, M. D. (2005). Biomass burning as a
source of dissolved iron to the open ocean? Geophysical Research Letters, 32(19),
1–5. https://doi.org/10.1029/2005GL022962

Hadas, H., Einav, M., Fishov, I., & Zaritsky, A. (1997). Bacteriophage T4 development
depends on the physiology of its host Escherichia coli. Microbiology, 143(1), 179
185. https://doi.org/10.1099/00221287-143-1-179

Hadas, H., Einav, M., & Zaritsky, A. (1995). Division-inhibition capacity of penicillin in
Escherichia coli is growth-rate dependent. Microbiology (Reading), 141, 1081
1083. https://doi.org/10.1099/13500872-141-5-1081

Häder, D. P., Villafañe, V. E., & Helbling, E. W. (2014). Productivity of aquatic primary
producers under global climate change. Photochemical and Photobiological
Sciences,13(10), 1370–1392. https://doi.org/10.1039/c3pp50418b

Hansen, A. D. A., Rosen, H., & Novakov, T. (1983). The Aethalometer - An Instrument for
the Real-Time Measurement of optical Absorption by Aerosol Particles.

Hatta, M., Measures, C. I., Wu, J., Roshan, S., Fitzsimmons, J. N., Sedwick, P., & Morton,
P. (2015). An overview of dissolved Fe and Mn distributions during the 2010-2011
U.S. GEOTRACES north Atlantic cruises: GEOTRACES GA03. Deep-Sea Research
Part II: Topical Studies in Oceanography, 116, 117–129.
https://doi.org/10.1016/j.dsr2.2014.07.005

165
Hawco, N. J., Ohnemus, D. C., Resing, J. A., Twining, B. S., & Saito, M. A. (2016). A
dissolved cobalt plume in the oxygen minimum zone of the eastern tropical South
Pacific. Biogeosciences, 13(20), 5697–5717.
https://doi.org/10.5194/bg-13-5697-2016

Hawco, N. J., Yang, S. C., Foreman, R. K., Funkey, C. P., Dugenne, M., White, A. E., …
John, S. G. (2020). Metal isotope signatures from lava-seawater interaction during
the 2018 eruption of Kīlauea. Geochimica et Cosmochimica Acta, 282, 340–356.
https://doi.org/10.1016/j.gca.2020.05.005

Hayward, T. L., & Venrick, E. L. (1998). Nearsurface pattern in the California Current:
coupling between physical and biological structure. Deep Sea Research Part II:
Topical Studies in Oceanography, 45, 1617–1638.
https://doi.org/10.5565/rev/papers/v40n0.1650

Ho, T., Quigg, A., Zoe, V., Milligan, A. J., Falkowski, P. G., & Morel, M. M. (2003). The
elemental composition of some phytoplankton. Journal of Phycology, 39(6),1145
1159. https://doi.org/10.1111/j.0022-3646.2003.03-090.x

Hoffmann, L. J., Breitbarth, E., Boyd, P. W., & Hunter, K. A. (2012). Influence of ocean
warming and acidification on trace metal biogeochemistry. Marine Ecology
Progress Series, 470, 191–205. https://doi.org/10.3354/meps10082

Hogle, S. L., Barbeau, K. A., & Gledhill, M. (2014). Heme in the marine environment: From
cells to the iron cycle. Metallomics, 6(6), 1107–1120.
https://doi.org/10.1039/c4mt00031e

Hogle, S. L., Cameron Thrash, J., Dupont, C. L., & Barbeau, K. A. (2016). Trace metal
acquisition by marine heterotrophic bacterioplankton with contrasting trophic
strategies. Applied and Environmental Microbiology, 82(5), 1613–1624.
https://doi.org/10.1128/AEM.03128-15

Hopkinson, B. M., & Morel, F. M. M. (2009). The role of siderophores in iron acquisition by
photosynthetic marine microorganisms. BioMetals, 22(4), 659–669.
https://doi.org/10.1007/s10534-009-9235-2

Hornberger, G. M., Raffensperger, J. P., Wiberg, P. L., & Eshleman, K. N. (1998).
Elements of physical hydrology. Baltimore, Maryland: John Hopkins University Press.

Hosseini, S., Urbanski, S. P., Dixit, P., Qi, L., Burling, I. R., Yokelson, R. J., … Cocker, D.
R. (2013). Laboratory characterization of PM emissions from combustion of
wildland biomass fuels. Journal of Geophysical Research: Atmospheres, 118(17),
9914–9929. https://doi.org/10.1002/jgrd.50481

166
Hsu, S. C., Wong, G. T. F., Gong, G. C., Shiah, F. K., Huang, Y. T., Kao, S. J., … Tseng,
C. M. (2010). Sources, solubility, and dry deposition of aerosol trace elements over
the East China Sea. Marine Chemistry, 120(1–4), 116–127.
https://doi.org/10.1016/j.marchem.2008.10.003

Hunsinger, G. B., Mitra, S., Warrick, J. A., & Alexander, C. R. (2008). Oceanic loading of
wildfire-derived organic compounds from a small mountainous river. Journal of
Geophysical Research: Biogeosciences, 113(2), 1–14.
https://doi.org/10.1029/2007JG000476

Hurst, M. P., & Bruland, K. W. (2007). An investigation into the exchange of iron and zinc
between soluble, colloidal, and particulate size-fractions in shelf waters using low
abundance isotopes as tracers in shipboard incubation experiments. Marine
Chemistry, 103(3–4), 211–226. https://doi.org/10.1016/j.marchem.2006.07.001

Hutchins, D. A., & Bruland, K. W. (1994). Grazer-mediated regeneration and assimilation of
Fe, Zn and Mn from planktonic prey. Marine Ecology Progress Series, 110(2–3),
259–270. https://doi.org/10.3354/meps110259

Hutchins, D. A., Hare, C. E., Weaver, R. S., Zhang, Y., Firme, G. F., DiTullio, G. R., …
183 Bruland, K. W. (2002). Phytoplankton iron limitation in the Humboldt Current
and Peru Upwelling. Limnology and Oceanography, 47(4), 997–1011.
https://doi.org/10.4319/lo.2002.47.4.0997

Hutchins, D. A., & Boyd, P. W. (2016). Marine phytoplankton and the changing ocean iron
cycle. Nature Climate Change, 6, 1072-1079.
https://doi.org/10.1038/nclimate3147

Hutchins, D. A., DiTullio, G. R., & Bruland, K. W. (1993). Iron and regenerated production:
evidence for biological Fe recycling in two marine environments. Limnology and
Oceanography, 38(6), 1242–1255. https://doi.org/10.4319/lo.1993.38.6.1242

Hutchins, D. A., Witter, A. E., Butler, A., & Luther, G. W. (1999). Competition among
marine phytoplankton for different chelated iron species. Nature, 400(6747), 858
861. https://doi.org/10.1038/23680

Jacquot, J. E., & Moffett, J. W. (2015). Copper distribution and speciation across the
International GEOTRACES Section GA03. Deep-Sea Research II: Topical Studies in
Oceanography, 116, 187-207. http://doi.org/10.1016/j.dsr2.2014.11.013

Jickells, T. (1995). Atmospheric inputs of metals and nutrients to the oceans: their
magnitude and effects. Marine Chemistry, 48(3–4), 199–214.
https://doi.org/10.1016/0304-4203(95)92784-P

167
Jickells, T. D., An, Z. S., Andersen, K. K., Baker, A. R., Bergametti, C., Brooks, N., …
Torres, R. (2005). Global iron connections between desert dust, ocean
biogeochemistry, and climate. Science, 308(5718), 67–71.
https://doi.org/10.1126/science.1105959

John, S. G., & Conway, T. M. (2014). A role for scavenging in the marine biogeochemical
cycling of zinc and zinc isotopes. Earth and Planetary Science Letters, 394, 159
167. https://doi.org/10.1016/j.epsl.2014.02.053

John, S. G., Geis, R. W., Saito, M. A., & Boyle, E. A. (2007). Zinc isotope fractionation
during high affinity and low-affinity zinc transport by the marine diatom Thalassiosire
oceanica. Limnology and Oceanography, 52(6), 2710–2714.

John, S. G., Helgoe, J., & Townsend, E. (2018). Biogeochemical cycling of Zn and Cd and
their stable isotopes in the Eastern Tropical South Pacific. Marine Chemistry, 201,
256–262. https://doi.org/10.1016/j.marchem.2017.06.001

Kang, C. M., Gold, D., & Koutrakis, P. (2014). Downwind O3 and PM2.5 speciation during
the wildfires in 2002 and 2010. Atmospheric Environment, 95, 511–519.
https://doi.org/10.1016/j.atmosenv.2014.07.008

Karl, D. M. (1999). A sea of change: Biogeochemical variability in the North Pacific
Subtropical Gyre. Ecosystems, 2(3), 181–214.
https://doi.org/10.1007/s100219900068

Kauffman, K. M., Brown, J. M., Sharma, R. S., VanInsberghe, D., Elsherbini, J., Polz, M., &
Kelly, L. (2018). Data Descriptor: Viruses of the Nahant Collection, characterization
of 251 marine Vibrionaceae viruses. Scientific Data, 5, 1–11.
https://doi.org/10.1038/sdata.2018.114

Kean, J. W., Staley, D. M., Lancaster, J. T., Rengers, F. K., Swanson, B. J., Coe, J. A., …
Lindsay, D. N. (2019). Inundation, flow dynamics, and damage in the 9 January
2018 Montecito debris-flow event, California, USA: Opportunities and challenges for
post-wildfire risk assessment. Geosphere, 15(4), 1140–1163.
https://doi.org/10.1130/GES02048.1

Keller, M. D., Bellows, W. K., & Guillard, R. R. L. (1988). Microwave treatment for
sterilization of phytoplankton culture media. Journal of Experimental Marine Biology
and Ecology, 117(3), 279–283. https://doi.org/10.1016/0022-0981(88)90063-9

168
Kelly, R. L., Bian, X., Feakins, S. J., Fornace, K. L., Gunderson, T., Hawco, N. J., … John,
S. G. (2021). Delivery of Metals and Dissolved Black Carbon to the Southern
California Coastal Ocean via Aerosols and Floodwaters Following the 2017 Thomas
Fire. Journal of Geophysical Research: Biogeosciences, 126(3), 1–25.
https://doi.org/10.1029/2020JG006117

King, A. L., & Barbeau, K. A. (2011). Dissolved iron and macronutrient distributions in the
southern California Current System. Journal of Geophysical Research: Oceans,
116(3), 1–18. https://doi.org/10.1029/2010JC006324

King, A. L., & Barbeau, K. A. (2007). Iron distribution and phytoplankton iron limitation in
the southern California Current System. Marine Ecology Progress Series, 342, 91
103.

Kirchman, D. L., Meon, B., Cottrell, M. T., Hutchins, D. A., Weeks, D., & Bruland, K. W.
(2000). Carbon versus iron limitation of bacterial growth in the California upwelling
regime. Limnology and Oceanography, 45(8), 1681–1688.
https://doi.org/10.4319/lo.2000.45.8.1681

Knapp, A., Casciotti, K., Berelson, W., Prokopenko, M., & Capone, D. (2016). Low rates of
nitrogen fixation in eastern tropical South Pacific surface waters. Proceedings of the
National Academy of Sciences of the United States of America, 113(16), 4398
4403. https://doi.org/10.1073/pnas.1515641113

Knauer, G. A., Martin, J. H., & Gordon, R. M. (1982). Cobalt in north-east Pacific waters.
Nature, 297(5861), 49–51. https://doi.org/10.1038/297049a0

Koukouzas, N., Hämäläinen, J., Papanikolaou, D., Tourunen, A., & Jäntti, T. (2007).
Mineralogical and elemental composition of fly ash from pilot scale fluidised bed
combustion of lignite, bituminous coal, wood chips and their blends. Fuel, 86,
2186–2193. https://doi.org/10.1016/j.fuel.2007.03.036

Kranzler, C. F., Brzezinski, M. A., Cohen, N. R., Lampe, R. H., Maniscalco, M., Till, C. P.,
… Thamatrakoln, K. (2021). Impaired viral infection and reduced mortality of
diatoms in iron-limited oceanic regions. Nature Geoscience, 14(4), 231–237.
https://doi.org/10.1038/s41561-021-00711-6

Kranzler, C. F., Krause, J. W., Brzezinski, M. A., Edwards, B. R., Biggs, W. P., Maniscalco,
M., … Thamatrakoln, K. (2019). Silicon limitation facilitates virus infection and
mortality of marine diatoms. Nature Microbiology, 4(11), 1790–1797.
https://doi.org/10.1038/s41564-019-0502-x

169
Krishnaswami, S., Lal, D., Amin, B. S., & Soutar, A. (1973). Geochronological studies in
Santa Barbara Basin: 55Fe as a unique tracer for particulate settling. Limnology and
Oceanography, 18(5), 763–770. https://doi.org/10.4319/lo.1973.18.5.0763

Kuang, X. M., Scott, J. A., da Rocha, G. O., Betha, R., Price, D. J., Russell, L. M., …
Paulson, S. E. (2017). Hydroxyl radical formation and soluble trace metal content in
particulate matter from renewable diesel and ultra low sulfur diesel in at-sea
operations of a research vessel. Aerosol Science and Technology, 51(2), 147–158.
https://doi.org/10.1080/02786826.2016.1271938

Laès, A., Blain, S., Laan, P., Ussher, S. J., Achterberg, E. P., Tréguer, P., & De Baar, H. J.
W. (2007). Sources and transport of dissolved iron and manganese along the
continental margin of the Bay of Biscay. Biogeosciences, 4(2), 181–194.
https://doi.org/10.5194/bg-4-181-2007

Lagerström, M. E., Field, M. P., Séguret, M., Fischer, L., Hann, S., & Sherrell, R. M. (2013).
Automated on-line flow-injection ICP-MS determination of trace metals (Mn, Fe, Co,
Ni, Cu and Zn) in open ocean seawater: Application to the GEOTRACES program.
Marine Chemistry, 155, 71–80. https://doi.org/10.1016/j.marchem.2013.06.001

Landry, M. R., Barber, R. T., Bidigare, R. R., Chai, F., Coale, K. H., Dam, H. G., … White,
J. R. (1997). Iron and grazing constraints on primary production in the central
equatorial Pacific: An EqPac synthesis. Limnology and Oceanography, 42(3), 405-
418. https://doi.org/10.4319/lo.1997.42.3.0405

Lannuzel, D., Schoemann, V., de Jong, J., Tison, J. L., & Chou, L. (2007). Distribution and
biogeochemical behaviour of iron in the East Antarctic sea ice. Marine Chemistry,
106, 18–32. https://doi.org/10.1016/j.marchem.2006.06.010

Lee, J. M., Boyle, E. A., Echegoyen-sanz, Y., Fitzsimmons, J. N., Zhang, R., & Kayser,
R. A. (2011). Analysis of trace metals (Cu, Cd, Pb, and Fe) in seawater using single
batch nitrilotriacetate resin extraction and isotope dilution inductively coupled
plasma mass spectrometry. Analytica Chimica Acta, 686(1–2), 93–101.
https://doi.org/10.1016/j.aca.2010.11.052

Letelier, R. M., Bidigare, R. R., Hebel, D. V, Ondrusek, M., Winn, C. D., & Karl, D. M.
(1993). Temporal variability of phytoplankton community structure based on
pigment analysis. Limnology and Oceanography, 38(7), 1420–1437.
https://doi.org/10.4319/lo.1993.38.7.1420

Lis, H., Kranzler, C., Keren, N., & Shaked, Y. (2015). A comparative study of Iron uptake
rates and mechanisms amongst marine and fresh water Cyanobacteria: prevalence
of reductive Iron uptake. Life, 5(1), 841–860. https://doi.org/10.3390/life5010841

170
Lohse, K. A., Brooks, P. D., McIntosh, J. C., Meixner, T., & Huxman, T. E. (2009).
Interactions Between Biogeochemistry and Hydrologic Systems. Annual Review of
Environment and Resources, 34(1), 65–96.
https://doi.org/10.1146/annurev.environ.33.031207.111141

Longhurst, A. (1998). Ecological Geography of the Sea. San Diego: Academic Press.

Luo, C., Mahowald, N., Bond, T., Chuang, P. Y., Artaxo, P., Siefert, R., … Schauer, J.
(2008). Combustion iron distribution and deposition. Global Biogeochemical Cycles,
22(1), 1–17. https://doi.org/10.1029/2007GB002964

Lynn, R. J., & Simpson, J. J. (2008). The California Current system: The seasonal variability
of its physical characteristics. Journal of Geophysical Research, 92(C12), 12947.
https://doi.org/10.1029/jc092ic12p12947

Maat, D. S., Crawfurd, K. J., Timmermans, K. R., & Brussaard, C. P. D. (2014). Elevated
CO2 and phosphate limitation favor micromonas pusilla through stimulated growth
and reduced viral impact. Applied and Environmental Microbiology, 80(10), 3119–
3127. https://doi.org/10.1128/AEM.03639-13

Mackas, D. L., Stub, P., Thomas, A., & Montecino, V. (2006). Eastern ocean boundaries
pan-regional overview. In The Sea, (pp. 21-59).

Mackey, D. J., O’Sullivan, J. E. Os., & Watson, R. J. (2002). Iron in the western Pacific: A
riverine or hydrothermal source for iron in the Equatorial Undercurrent? Deep-Sea
Research Part I: Oceanographic Research Papers, 49(5), 877–893.
https://doi.org/10.1016/S0967-0637(01)00075-9

Mahowald, N. M., Engelstaedter, S., Luo, C., Sealy, A., Artaxo, P., Benitez-Nelson, C., …
Siefert, R. L. (2009). Atmospheric Iron Deposition: Global Distribution, Variability,
and Human Perturbations. Annual Review of Marine Science, 1, 245–278.
https://doi.org/10.1146/annurev.marine.010908.163727

Mahowald, N. M., Hamilton, D. S., Mackey, K. R. M., Moore, J. K., Baker, A. R., Scanza,
R. A., & Zhang, Y. (2018). Aerosol trace metal leaching and impacts on marine
microorganisms. Nature Communications, 9(1). 2614.
https://doi.org/10.1038/s41467-018-04970-7

Maldonado, M. T., & Price, N. M. (2001). Reduction and transport of organically bound iron
by Thalassiosira oceanica (Bacillariophyceae). Journal of Phycology, 37(2), 298–
310. https://doi.org/10.1046/j.1529-8817.2001.037002298.x

171
Maldonado, Maria T., & Price, N. M. (1999). Utilization of iron bound to strong organic
ligands by plankton communities in the subarctic Pacific Ocean. Deep-Sea
Research Part II: Topical Studies in Oceanography, 46(11–12), 2447–2473.
https://doi.org/10.1016/S0967-0645(99)00071-5

Manck, L. E., Espinoza, J. L., Dupont, C. L., & Barbeau, K. A. (2020). Transcriptomic
Study of Substrate-Specific Transport Mechanisms for Iron and Carbon in the
Marine Copiotroph Alteromonas macleodii. MSystems, 5(2).
https://doi.org/10.1128/msystems.00070-20

Manck, L. E., Park, J., Tully, B. J., Poire, A. M., Bundy, R. M., Dupont, C. L., & Barbeau,
K. A. (2022). Petrobactin, a siderophore produced by Alteromonas, mediates
community iron acquisition in the global ocean. ISME Journal, 16(2), 358–369.
https://doi.org/10.1038/s41396-021-01065-y

Maranger, R., & Bird, D. F. (1995). Viral abundance in aquatic systems: A comparison
between marine and fresh waters. Marine Ecology Progress Series, 121(1–3), 217–
226. https://doi.org/10.3354/meps121217

Marchetti, A., Maldonado, M. T., Lane, E. S., & Harrison, P. J. (2006). Iron requirements of
the pennate diatom Pseudo-nitzschia: Comparison of oceanic (high-nitrate, low
chlorophyll waters) and coastal species. Limnology and Oceanography, 51(5),
2092–2101. https://doi.org/10.4319/lo.2006.51.5.2092

Mari X., Van T. C., Guinot B., Brune J., Lefebvre J. P., Raimbault P., … Niggemann J.
(2017). Seasonal dynamics of atmospheric and river inputs of black carbon, and
impacts on biogeochemical cycles in Halong Bay, Vietnam. Elementa: Science of
the Anthropocene, 5(75). https://doi.org/10.1525/elementa.255

Maring, H. B., & Duce, R. A. (1987). The impact of atmospheric aerosols on trace metal
chemistry in open ocean surface seawater, 1. Aluminum. Earth and Planetary
Science Letters, 84(4), 381-392. https://doi.org/10.1016/0012-821X(87)90003-3

Martin, J. H., & Fitzwater, S. E. (1988). Iron deficiency limits phytoplankton growth in the
north-east pacific subarctic. Nature, 331(6154), 341–343.
https://doi.org/10.1038/331341a0

Martin, J. H., & Michael Gordon, R. (1988). Northeast Pacific iron distributions in relation to
phytoplankton productivity. Deep Sea Research Part A, Oceanographic Research
Papers, 35(2), 177–196. https://doi.org/10.1016/0198-0149(88)90035-0

Martin, J. M., & Whitfield, M. (1983). The Significance of the River Input of Chemical
Elements to the Ocean. In Trace metals in sea water (pp. 265–296). Springer,
Boston, MA.
172

Marques, J. S. J., Dittmar, T., Niggemann, J., Almeida, M. G., Gomez-Saez, G. V., &
Rezende, C. E. (2017). Dissolved black carbon in the headwaters-to-ocean
continuum of Paraíba do Sul River, Brazil. Frontiers in Earth Science, 5, 1–12.
https://doi.org/10.3389/feart.2017.00011

Matsui, H., Mahowald, N. M., Moteki, N., Hamilton, D. S., Ohata, S., Yoshida, A., …
Flanner, M. G. (2018). Anthropogenic combustion iron as a complex climate forcer.
Nature Communications, 9(1). https://doi.org/10.1038/s41467-018-03997-0

Mawji, E., Schlitzer, R., Dodas, E. M., Abadie, C., Abouchami, W., Anderson, R. F., …
Zimmer, L. A. (2014). The GEOTRACES Intermediate Data Product 2014. Marine
Chemistry, 177, 1–8. https://doi.org/10.1016/j.marchem.2015.04.005

Mcdermid, J. M., Jaye, A., Loeff, M. F. S. Van Der, Todd, J., Bates, C., Austin, S., …
Prentice, A. M. (2007). Elevated iron status strongly predicts mortality in West
African adults with HIV infection. Mortality, 46(4), 498–507.
https://doi.org/10.1097/qai.0b013e31815b2d4b

McDonald, B. C., Goldstein, A. H., & Harley, R. A. (2015). Long-term trends in California
mobile source emissions and ambient concentrations of black carbon and organic
aerosol. Environmental Science & Technology, 49(8), 5178–5188.
https://doi.org/10.1021/es505912b

Measures, C. I., Landing, W. M., Brown, M. T., & Buck, C. S. (2008). A commercially
available rosette system for trace metal-clean sampling. Limnology and
Oceanography: Methods, 6(9), 384–394. https://doi.org/10.4319/lom.2008.6.384

Meixner, T., Fenn, M. E., Wohlgemuth, P., Oxford, M., & Riggan, P. (2006). N saturation
symptoms in chaparral catchments are not reversed by prescribed fire.
Environmental Science and Technology, 40(9), 2887–2894.
https://doi.org/10.1021/es051268z

Merchant, C. J., Embury, O., Bulgin, C. E., Block, T., Corlett, G. K., Fiedler, E., … Donlon,
C. (2019). Satellite-based time-series of sea-surface temperature since 1981 for
climate applications. Scientific Data, 6(1), 1–18.
https://doi.org/10.1038/s41597-019-0236-x

Michael, S., Resing, J., Lacan, F., Buck, N., Pradoux, C., & Jeandel, C. (2021).
Constraining the Solomon Sea as a source of Al and Mn to the Equatorial
Undercurrent. Deep-Sea Research Part I: Oceanographic Research Papers,
174, 103559. https://doi.org/10.1016/j.dsr.2021.103559

173
Middag, R., de Baar, H. J. W., Bruland, K. W., & van Heuven, S. M. A. C. (2020). The
Distribution of Nickel in the West-Atlantic Ocean, Its Relationship With Phosphate
and a Comparison to Cadmium and Zinc. Frontiers in Marine Science, 7, 1–17.
https://doi.org/10.3389/fmars.2020.00105

Minshall, G. W., Brock, J. T., & Varley, J. D. (1989). Wildfires and Yellowstone’s Stream
Ecosystems. BioScience, 39(10), 707–715. https://doi.org/10.2307/1311002

Millero, F.J., Woosey, R., Ditrolio, B., & Water, J. (2009). effect of Ocean acidification on
the Speciation of Metals in Seawater. Oceanography, 22(4), 72–85. Retrieved from
http://tos.org/oceanography/assets/docs/22-4_millero.pdf

Mioni, C. E., Poorvin, L., & Wilhelm, S. W. (2005). Virus and siderophore-mediated transfer
of available Fe between heterotrophic bacteria: Characterization using an Fe-
specific bioreporter. Aquatic Microbial Ecology, 41(3), 233–245.
https://doi.org/10.3354/ame041233

Moffett, J. W., & Brand, L. E. (1996). Production of strong, extracellular Cu chelators by
marine cyanobacteria in response to Cu stress. Limnology and Oceanography,
41(3), 388–395. https://doi.org/10.4319/lo.1996.41.3.0388

Moffett, J. W., & German, C. R. (2018). The U.S.GEOTRACES Eastern Tropical Pacific
Transect (GP16). Marine Chemistry, 201, 1–5.
https://doi.org/10.1016/j.marchem.2017.12.001

Moffett, J. W., & Zika, R. G. (1987). Reaction kinetics of hydrogen peroxide with copper
and iron in seawater. Environmental Science & Technology, 21(8), 804–810.
https://doi.org/10.1021/es00162a012

Montecito Fire Protection District. (2018). A Defensible Community? A Retrospective Study
of Montecito Fire Protection District’s Wildland Fire Program during the 2017
Thomas Fire. Montecito, CA, USA.

Moody, J. A., Shakesby, R. A., Robichaud, P. R., Cannon, S. H., & Martin, D. A. (2013).
Current research issues related to post-wildfire runoff and erosion processes.
Earth-Science Reviews, 122, 10–37.
https://doi.org/10.1016/j.earscirev.2013.03.004

Moore, C. M. M., Mills, M. M. M., Arrigo, K. R. R., Berman-Frank, I., Bopp, L., Boyd, P. W.
W., … Ulloa, O. (2013). Processes and patterns of oceanic nutrient limitation.
Nature Geoscience, 6(9), 701–710. https://doi.org/10.1038/ngeo1765

174
Moore, L. R., Coe, A., Zinser, E. R., Saito, M. A., Sullivan, M. B., Lindell, D., … Chisholm,
S. W. (2007). Culturing the marine cyanobacterium Prochlorococcus. Limnology
and Oceanography: Methods, 5(10), 353–362.
https://doi.org/10.4319/lom.2007.5.353

Moore, W. S. (2010). The effect of submarine groundwater discharge on the ocean.
Annual Review of Marine Science, 2(1), 59–88.
https://doi.org/10.1146/annurevmarine-120308-081019

Morel, F. M. M., Milligan, A. J., & Saito, M. A. A. (2003). Marine Bioinorganic Chemistry:
The Role of Trace Metals in the Oceanic Cycles of Major Nutrients. In Treatise on
Geochemistry (pp. 113–143). Elsevier Pergamon. https://doi.org/10.1016/B0-08-
043751-6/06108-9

Morel, F. M. M., Reinfelder, J. R., Roberts, S. B., Chamberlain, C. P., Lee, J. G., & Yee, D.
(1994). Zinc and carbon co-limitation of marine phytoplankton. Nature, 369(6483),
740–742. https://doi.org/10.1038/369740a0

Morrison, R. J., Denton, G. R. W., Bale Tamata, U., & Grignon, J. (2013). Anthropogenic
biogeochemical impacts on coral reefs in the Pacific Islands-An overview. Deep-
Sea Research Part II: Topical Studies in Oceanography, 96, 5–12.
https://doi.org/10.1016/j.dsr2.2013.02.014

Morrissey, J., & Bowler, C. (2012). Iron utilization in marine cyanobacteria and eukaryotic
algae. Frontiers in Microbiology, 3, 1–13.
https://doi.org/10.3389/fmicb.2012.00043

Moulin, C., Lambert, C. E., Dulac, F., & Dayan, U. (1997). Control of atmospheric export of
dust from North Africa by the North Atlantic Oscillation. Nature, 387(6634), 691–
694. https://doi.org/10.1038/42679

Myers-Pigg, A. N., Louchouarn, P., & Teisserenc, R. (2017). Flux of dissolved and
particulate low-temperature pyrogenic carbon from two high-latitude rivers across
the spring freshet hydrograph. Frontiers in Marine Science, 4, 1–11.
https://doi.org/10.3389/fmars.2017.00038

Nadler, D. W. (2019). Decision support: using machine learning through MATLAB to
analyze environmental data. Journal of Environmental Studies and Sciences, 9(4),
419–428. https://doi.org/10.1007/s13412-019-00558-9

NASA. (2022). EOSDIS Worldview. Retrieved from
https://worldview.earthdata.nasa.gov/?v=-148.32622082965645,
24.692629403117117,-25.626600950351538,46.52671342092567

175
NASA. (2017a). EOSDIS Worldview. Retrieved from
https://worldview.earthdata.nasa.gov/?v=-148.32622082965645,
24.692629403117117,-25.626600950351538,46.52671342092567

NASA. (2017b). Smoke and Fire in Southern California. Retrieved from
https://visibleearth.nasa.gov/images/91379/smoke-and-fire-in-southern
california/91379f

Neilands, J. B. (1995). Siderophores : Structure and Function of Microbial Iron Transport
Compounds. The Journal of Biological Chemistry, 270(45), 26723–26726.
https://doi.org/10.1074/jbc.270.45.26723

Ni M., Huang J., Lu S., Li X., Yan J., & Cen K. (2014). A review on black carbon emissions,
worldwide and in China. Chemosphere, 107, 83-93.
https://doi.org/10.1016/j.chemosphere.2014.02.052

Nicewicz, D., & Szczepkowski, A. (2008). The Content of Heavy Metals in the Wood of \
Healthy and Dying Beech Trees (Fagus Sylvatica L.) *. Journal Acta Scientiarum
Polonorum Silvarum Colendrum Ratio et Industria Lignaria, 7(4), 35–44.

NOAA. (n.d.). NOAA HYSPLIT Trajectory Model. Retrieved from
https://ready.arl.noaa.gov/hypub-bin/trajtype.pl

NOAA. (2022). Sea Surface Temperature Countour Charts. Retrieved from
https://www.ospo.noaa.gov/Products/ocean/sst/contour/

NOAA. (2020). National Temperature and Precipitation Maps. Retrieved from
https://www.ncdc.noaa.gov/temp-and-precip/us
maps/1/201712?products[]=divisionaltavgrank#maps

Nyholm, S., Mann, G. J., Johansson, A. G., Bergeron, R. J., Graslund, A., & Thelander, L.
(1993). Role of ribonucleotide reductase in inhibition of mammalian cell growth by
potent iron chelators. Journal of Biological Chemistry, 268(35), 26200–26205.
https://doi.org/10.1016/s0021-9258(19)74300-2

Obata, H., Shitashima, K., Isshiki, K., & Nakayama, E. (2008). Iron, manganese and
aluminum in upper waters of the western South Pacific Ocean and its adjacent
seas. Journal of Oceanography, 64(2), 233–245.
https://doi.org/10.1007/s10872-008-0018-0

Odigie, K. O., & Flegal, A. R. (2011). Pyrogenic remobilization of historic industrial lead
depositions. Environmental Science & Technology, 45(15), 6290–6295.
https://doi.org/10.1021/es200944w

176
Odigie, K. O., & Flegal, A. R. (2014). Trace metal inventories and lead isotopic composition
chronicle a forest fire’s remobilization of industrial contaminants deposited in the
Angeles National Forest. PLoS ONE, 9(9).
https://doi.org/10.1371/journal.pone.0107835

Ohnemus, D. C., Auro, M. E., Sherrell, R. M., Lagerström, M., Morton, P. L., Twining, B.
S., …Lam, P. J. (2014). Laboratory intercomparison of marine particulate digestions
including Piranha: A novel chemical method for dissolution of polyethersulfone
filters. Limnology and Oceanography, 12, 530–547.
https://doi.org/10.4319/lom.2014.12.530

Oliveira-Filho, E. C., Brito, D. Q., Dias, Z. M. B., Guarieiro, M. S., Carvalho, E. L., Fascineli,
M. L., … Grisolia, C. K. (2018). Effects of ashes from a Brazilian savanna wildfire on
water, soil and biota: An ecotoxicological approach. Science of the Total
Environment, 618, 101–111. https://doi.org/10.1016/j.scitotenv.2017.11.051

Olivella, M. A., Ribalta, T. G., De Febrer, A. R., Mollet, J. M., & De Las Heras, F. X. C.
(2006). Distribution of polycyclic aromatic hydrocarbons in riverine waters after
Mediterranean forest fires. Science of the Total Environment, 355(1–3), 156–166.
https://doi.org/10.1016/j.scitotenv.2005.02.033

Overland, J. E., & Wang, M. (2007). Future climate of the North Pacific Ocean. Eos, 88(16),
178-182. https://doi.org/10.1029/2007EO160003

Paulson, S. E., Gallimore, P. J., Kuang, X. M., Chen, J. R., Kalberer, M., & González, D. H.
(2019). A light-driven burst of hydroxyl radicals dominates oxidation chemistry in
newly activated cloud droplets. Science Advances, 5(5), 1–8.
https://doi.org/10.1126/sciadv.aav7689

Paytan, A., Mackey, K. R. M., Chen, Y., Lima, I. D., Doney, S. C., Mahowald, N., … Post,
A. F. (2009). Toxicity of atmospheric aerosols on marine phytoplankton, 106(12),
4601–4605. https://doi.org/10.1073/pnas.0811486106

Peers, G., & Price, N. M. (2006). Copper-containing plastocyanin used for electron
transport by an oceanic diatom. Nature, 441(7091), 341–344.
https://doi.org/10.1038/nature04630

Perdrial, J. N., McIntosh, J., Harpold, A., Brooks, P. D., Zapata-Rios, X., Ray, J., …
Chorover, J. (2014). Stream water carbon controls in seasonally snow-covered
mountain catchments: Impact of inter-annual variability of water fluxes,
catchment aspect and seasonal processes. Biogeochemistry, 118(1–3), 273–290.
https://doi.org/10.1007/s10533-013-9929-y

177
Pinedo-González, P., Hawco, N. J., Bundy, R. M., Virginia Armbrust, E., Follows, M. J.,
Cael, B. B., … John, S. G. (2020). Anthropogenic Asian aerosols provide Fe to the
North Pacific Ocean. Proceedings of the National Academy of Sciences of the
United States of America, 117(45), 27862–27868.
https://doi.org/10.1073/pnas.2010315117

Pinedo-González, P., Hellige, B., West, A. J., & Sañudo-Wilhelmy, S. A. (2017). Changes
in the size partitioning of metals in storm runoff following wildfires: Implications for
the transport of bioactive trace metals. Applied Geochemistry, 83, 62–71.
https://doi.org/10.1016/j.apgeochem.2016.07.016

Pinedo-González, P., West, A. J., Rivera-Duarte, I., & Sañudo-Wilhelmy, S. A. (2014). Diel
changes in trace metal concentration and distribution in coastal waters: Catalina
Island as a study case. Environmental Science and Technology, 48(14), 7730–
7737. https://doi.org/10.1021/es5019515

Pio, C. A., Legrand, M., Alves, C. A., Oliveira, T., Afonso, J., Caseiro, A., … Gelencsér, A.
(2008). Chemical composition of atmospheric aerosols during the 2003 summer
intense forest fire period. Atmospheric Environment, 42(32), 7530–7543.
https://doi.org/10.1016/j.atmosenv.2008.05.032

Pitman, R. M. (2006). Wood ash use in forestry - A review of the environmental impacts.
Forestry. 79(5), 563–588. https://doi.org/10.1093/forestry/cpl041

Pomeroy, L. R., le Williams, P. J. B., Azam, F., & Hobbie, J. E. (2007). The microbial loop.
Oceanography, 20, 28–33. https://doi.org/10.5670/oceanog.2007.45

Poorvin, L, Rinta-Kanto, J. M., Hutchins, D. A., & Wilhelm, S. W. (2004). Viral lysis of iron
and its bioavailability to marine plankton. Limnology and Oceanography, 49(5),
1734–1741. https://doi.org/10.4319/lo.2004.49.5.1734

Poorvin, L., Rinta-Kanto, J. M., Hutchins, D. A., & Wilhelm, S. W. (2004). Viral release of
iron and its bioavailability to marine plankton. Limnology and Oceanography, 49(5),
1734–1741. https://doi.org/10.4319/lo.2004.49.5.1734

Powell, C. F., Baker, A. R., Jickells, T. D., W. Bange, H., Chance, R. J., & Yodle, C. (2015).
Estimation of the atmospheric flux of nutrients and trace metals to the eastern
tropical North Atlantic Ocean. Journal of the Atmospheric Sciences, 72(10), 4029–
4045. https://doi.org/10.1175/JAS-D-15-0011.1

Price, N. M., & Morel, F. M. (1998). Biological cycling of iron in the ocean. In Metal ions in
biological systems (pp. 1–36).

178
Queirolo, F., Valenta, P., Stegen, S., & Breckle, S. W. (1990). Heavy metal concentrations
in oak wood growth rings from the Taunus (Federal Republic of Germany) and the
Valdivia (Chile) regions. Trees, 4, 81–87.

Rafter, P. A., Sigman, D. M., & Mackey, K. R. M. (2017). Recycled iron fuels new
production in the eastern equatorial Pacific Ocean. Nature Communications, 8(1).
https://doi.org/10.1038/s41467-017-01219-7

Raymond, P. A., Saiers, J. E., & Sobczak, W. V. (2016). Hydrological and biogeochemical
controls on watershed dissolved organic matter transport: pulse-shunt concept.
Ecology, 97(1), 5–16.

Redfield, C. A. (1934). On the proportions of organic derivatives in sea water and their
relation to the composition of plankton. James Johnstone Memorial Volume, 176–
192. Retrieved from http://ci.nii.ac.jp/naid/10006798977/en/

Resing, J. A., Sedwick, P. N., German, C. R., Jenkins, W. J., Moffett, J. W., Sohst, B. M.,
& Tagliabue, A. (2015). Basin-scale transport of hydrothermal dissolved metals
across the South Pacific Ocean. Nature, 523(7559), 200–203.
https://doi.org/10.1038/nature14577

Roebuck, J. A., Medeiros, P. M., Letourneau, M. L., & Jaffé, R. (2018). Hydrological
Controls on the Seasonal Variability of Dissolved and Particulate Black Carbon in
the Altamaha River, GA. Journal of Geophysical Research: Biogeosciences, 123(9),
3055–3071. https://doi.org/10.1029/2018JG004406

Roshan, S., & DeVries, T. (2021). Global Contrasts Between Oceanic Cycling of Cadmium
and Phosphate. Global Biogeochemical Cycles, 35(6), 1–18.
https://doi.org/10.1029/2021GB006952

Rue, E. L., & Bruland, K. W. (1995). Complexation of iron(III) by natural organic ligands in
the Central North Pacific as determined by a new competitive ligand
equilibration/adsorptive cathodic stripping voltammetric method. Marine Chemistry,
50(1–4), 117–138. https://doi.org/10.1016/0304-4203(95)00031-L

Rusiecka, D., Gledhill, M., Milne, A., Achterberg, E. P., Annett, A. L., Atkinson, S., ... &
Connelly, D. (2018). Anthropogenic signatures of lead in the Northeast Atlantic.
Geophysical Research Letters, 45(6), 2734-2743.
https://doi.org/10.1002/2017GL076825

Saager, P. M., De Baar, H. J. W., & Burkill, P. H. (1989). Manganese and iron in Indian
Ocean waters. Geochimica et Cosmochimica Acta, 53(9), 2259–2267.
https://doi.org/10.1016/0016-7037(89)90348-7

179
Saffari, A., Daher, N., Shafer, M. M., Schauer, J. J., & Sioutas, C. (2013). Seasonal and
spatial variation in reactive oxygen species activity of quasi-ultrafine particles
(PM0.25) in the Los Angeles metropolitan area and its association with chemical
composition. Atmospheric Environment, 79, 566–575.
https://doi.org/10.1016/j.atmosenv.2013.07.058

Saito, M. A., Bertrand, E. M., Dutkiewicz, S., Bulygin, V. V., Moran, D. M., Monteiro, F. M.,
… Waterbury, J. B. (2011). Iron conservation by reduction of metalloenzyme
inventories in the marine diazotroph Crocosphaera watsonii. Proceedings of the
National Academy of Sciences of the United States of America, 108(6), 2184–2189.
https://doi.org/10.1073/pnas.1006943108

Saito, M. A., & Moffett, J. W. (2001). Complexation of cobalt by natural organic ligands in
the Sargasso sea as determined by a new high-sensitivity electrochemical cobalt
speciation method suitable for open ocean work. Marine Chemistry, 75(1–2), 49–
68. https://doi.org/10.1016/S0304-4203(01)00025-1

Samanta, S., & Dalai, T. K. (2018). Massive production of heavy metals in the Ganga
(Hooghly) River estuary, India: Global importance of solute-particle interaction and
enhanced metal fluxes to the oceans. Geochimica et Cosmochimica Acta, 228,
243–258. https://doi.org/10.1016/j.gca.2018.03.002

Sander, S. G., & Koschinsky, A. (2011). Metal flux from hydrothermal vents increased by
organic complexation. Nature Geoscience, 4(3), 145–150.
https://doi.org/10.1038/ngeo1088

Santa Barbara County Public Health Department. (2017). Air Quality Watch Cancelled for
Santa Barbara County. Retrieved from https://www.countyofsb.org/asset.c/3601

Santoso, M., Dwiana Lestiani, D., & Hopke, P. K. (2013). Atmospheric black carbon in
PM2.5 in Indonesian cities. Journal of the Air & Waste Management Association,
63(9), 1022–1025. https://doi.org/10.1080/10962247.2013.804465

Santschi, P. H., Adler, D. M., & Amdurer, M. (1983). The Fate of Particles and Particle-
Reactive Trace Metals in Coastal Waters: Radioisotope Studies in Microcosms. In
Trace metals in sea water (pp. 331–349). Springer, Boston, MA.

Schauer, J. J., & Sioutas, C. (2012). Source Apportionment of Carbonaceous Aerosols
Using Integrated Multi-Variant and Source Tracer Techniques and a Unique
Molecular Marker Data Set Prepared for the California Air Resources Board and the
California Environmental Protection Agency, (07).

180
Schaule, B. K., & Patterson, C. C. (1981). Lead concentrations in the northeast Pacific:
evidence for global anthropogenic perturbations. Earth and Planetary Science
Letters, 54(1), 97–116.
https://doi.org/https://doi.org/10.1016/0012-821X(81)90072-8

Schoffman, H., Lis, H., Shaked, Y., & Keren, N. (2016). Iron-nutrient interactions within
phytoplankton. Frontiers in Plant Science, 7, 1–12.
https://doi.org/10.3389/fpls.2016.01223

Schwyn, B., & Neilands, J. B. (1987). Universal chemical assay for the detection and
determination of siderophore. Analytical Biochemistry, 160, 47–56.
https://doi.org/10.1016/0003-2697(87)90612-9

Sedwick, P. N., Church, T. M., Bowie, A. R., Marsay, C. M., Ussher, S. J., Achilles, K. M.,
Lethaby, P. J., Johnson, R. J., Sarin, M. M., & McGillicuddy, D. J. (2005). Iron in the
Sargasso Sea (Bermuda Atlantic Time-series study region) during summer: Eolian
imprint, spatiotemporal variability, and ecological implications. Global
Biogeochemical Cycles, 19, GB4006. https://doi.org/10.1029/2004GB002445

Sedwick, P. N., Marsay, C. M., Sohst, B. M., Aguilar-Islas, A. M., Lohan, M. C., Long, M.
C., … Ditullio, G. R. (2011). Early season depletion of dissolved iron in the Ross Sea
polynya: Implications for iron dynamics on the Antarctic continental shelf. Journal of
Geophysical Research: Oceans, 116(12), 1–19.
https://doi.org/10.1029/2010JC006553

Sedwick, P. N., Sholkovitz, E. R., & Church, T. M. (2007). Impact of anthropogenic
combustion emissions on the fractional solubility of aerosol iron: Evidence from the
Sargasso Sea. Geochemistry, Geophyics, Geosystems, 8(10).
https://doi.org/10.1029/2007GC001586

Segovia-Zavala, J. A., Delgadillo-Hinojosa, F., & Alvarez-Borrego, S. (1998). Cadmium in
the coastal upwelling area adjacent to the California-Mexico border. Estuarine,
Coastal and Shelf Science, 46(4), 475–481.
https://doi.org/10.1006/ecss.1997.0296

Shelley, R. U., Morton, P. L., & Landing, W. M. (2015). Elemental ratios and enrichment
factors in aerosols from the US-GEOTRACES North Atlantic transects. Deep-Sea
Research Part II: Topical Studies in Oceanography, 116, 262–272.
https://doi.org/10.1016/j.dsr2.2014.12.005

Shirmohammadi, F., Hasheminassab, S., Saffari, A., Schauer, J. J., Delfino, R. J., &
Sioutas, C. (2016). Fine and Ultrafine Particulate Organic Carbon in the Los Angeles
Basin: Trends in Sources and Composition. Science of the Total Environment, 541,
1083–1096. https://doi.org/10.1016/j.scitotenv.2015.09.133.
181

Sholkovitz, E. R., & Gieskes, J. M. (1971). A physical-chemical study of the flushing of the
Santa Barbara Basin. Limnology and Oceanography, 16(3), 479–489.

Sigel, S. P., & Payne, S. M. (1982). Effect of iron limitation on growth, siderophore
production, and expression of outer membrane proteins of Vibrio cholerae. Journal
of Bacteriology, 150(1), 148–155. https://doi.org/10.1128/jb.150.1.148-155.1982

Simon, N., Cras, A. L., Foulon, E., & Lemée, R. (2009). Diversity and evolution of marine
phytoplankton. Comptes Rendus - Biologies, 332(2–3), 159–170.
https://doi.org/10.1016/j.crvi.2008.09.009

Simoneit, B.R.T., Schauer, J. J., Nolte, C. G., Oros, D. R., Elias, V. O., Fraser, M. P., …
Cass, G. R. (1999). Levoglucosan, a tracer for cellulose in biomass burning and
atmospheric particles. Atmospheric Environment, 33, 173–182.
https://doi.org/10.1016/s1352-2310(98)00145-9

Simoneit, B.R.T., & Elias, V. O. (2000). Organic tracers from biomass burning in
atmospheric particulate matter over the ocean. Marine Chemistry, 69(3–4), 301
312. https://doi.org/10.1016/S0304-4203(00)00008-6

Skrabal, S. A., Donat, J. R., & Burdige, D. J. (1997). Fluxes of copper-complexing ligands
from estuarine sediments. Limnology and Oceanography, 42(5 I), 992–996.
https://doi.org/10.4319/lo.1997.42.5.0992

Sohrin, Y., & Bruland, K. W. (2011). Global status of trace elements in the ocean. TrAC -
Trends in Analytical Chemistry, 30(8), 1291–1307.
https://doi.org/10.1016/j.trac.2011.03.006

Sohrin, Y., Urushihara, S., Nakatsuka, S., Kono, T., Higo, E., Minami, T. T., … Umetani, S.
(2008). Multielemental determination of GEOTRACES key trace metals in seawater
by ICPMS after preconcentration using an ethylenediaminetriacetic acid chelating
resin. Analytical Chemistry, 80(16), 6267–6273. https://doi.org/10.1021/ac800500f

Song, T., Liu, H., Lv, T., Zhao, X., Shao, Y., Han, Q., … Zhang, W. (2018). Characteristics
of the iron uptake-related process of a pathogenic Vibrio splendidus strain
associated with massive mortalities of the sea cucumber Apostichopus japonicus.
Journal of Invertebrate Pathology, 155, 25–31.
https://doi.org/10.1016/j.jip.2018.05.001

Spencer, M. J., Betzer, P. R., & Piotrowicz, S. R. (1982). Concentrations of cadmium,
copper, lead and zinc in surface waters of the Northwest Atlantic Ocean - A
comparison of Go-flo and teflon water samplers. Marine Chemistry, 11(4), 403–410.
https://doi.org/10.1016/0304-4203(82)90034-2
182

Steenari, B. M., Schelander, S., & Lindqvist, O. (1999). Chemical and leaching
characteristics of ash from combustion of coal, peat and wood in a 12 MW CFB – a
comparative study. Fuel, 78(2), 249–258.
https://doi.org/10.1016/S00162361(98)00137-9

Stein, E. D., Brown, J. S., Hogue, T. S., Burke, M. P., & Kinoshita, A. (2012). Stormwater
contaminant loading following southern California wildfires. Environmental
Toxicology and Chemistry, 31(11), 2625–2638. https://doi.org/10.1002/etc.1994

Strzepek, R. F., Maldonado, M. T., Higgins, J. L., Hall, J., Safi, K., Wilhelm, S. W., & Boyd,
P. W. (2005). Spinning the “ferrous wheel”: The importance of the microbial
community in an iron budget during the FeCycle experiment. Global
Biogeochemical Cycles, 19(4). https://doi.org/10.1029/2005GB002490

Stubbins, A., Niggemann, J., & Dittmar, T. (2012). Photo-lability of deep ocean dissolved
black carbon. Biogeosciences, 9(5), 1661–1670.
https://doi.org/10.5194/bg-91661-2012

Stubbins, A., Spencer, R. G. M., Mann, P. J., Holmes, R. M., McClelland, J. W.,
Niggemann, J., & Dittmar T. (2015). Utilizing colored dissolved organic matter to
derive dissolved black carbon export by arctic rivers. Frontiers in Earth Science,
3, 1–11. https://doi.org/10.3389/feart.2015.00063

Sunda, W. G. (1989). Trace metal interactions with marine phytoplankton. Biological
Oceanography, 6(5–6), 411–442.
https://doi.org/10.1080/01965581.1988.10749543

Sunda, W. G., Price, N. M., & Morel, F. M. M. (2005). Trace metal ion buffers
and their use in culture studies. In Algal culturing techniques (pp. 35–63).
https://doi.org/10.1007/s13398-014-0173-7.2

Sunda, W. G. (2012). Feedback interactions between trace metal nutrients and
phytoplankton in the ocean. Frontiers in Microbiology, 3, 1–22.
https://doi.org/10.3389/fmicb.2012.00204

Sunda, W. G., & Huntsman, S. A. (1995). Iron uptake and growth limitation in oceanic
and coastal phytoplankton. Marine Chemistry, 50(1–4), 189–206.
https://doi.org/10.1016/0304-4203(95)00035-P

Suttle, C. A. (2005). Viruses in the sea. Nature, 437, 356–361.
https://doi.org/10.1038/nature04160

183
Suttle, C. A. (2007). Marine viruses - Major players in the global ecosystem. Nature
Reviews Microbiology, 5(10), 801–812. https://doi.org/10.1038/nrmicro1750

Swain, D. (2017). 2017 Hottest Summer in California History. Retrieved from
https://www.climatesignals.org/headlines/2017-hottest-summer-california-history

Świetlik, R., Trojanowska M., & Rabek P. (2013). Distribution patterns of Cd, Cu, Mn, Pb
and Zn in wood fly ash emitted from domestic boilers. Chemical Speciation &
Bioavailability, 25(1), 63–70. https://doi.org/10.3184/095422912X13497968675047

Tagliabue, A., Bowie, A. R., Boyd, P. W., Buck, K. N., Johnson, K. S., & Saito, M. A.
(2017). The integral role of iron in ocean biogeochemistry. Nature, 543(7643), 51–
59. https://doi.org/10.1038/nature21058

Thunell, R. C. (1998). Particle fluxes in a coastal upwelling zone: sediment trap results from
Santa Barbara Basin, California. Deep-Sea Research Part II: Topical Studies in
Oceanography, 45(8–9), 1863–1884.
https://doi.org/10.1016/s09670645(98)80020-9

Tian D., Hu Y., Wang Y., Boylan J. W., Zheng M., & Russell A. G. (2009). Assessment of
biomass burning emissions and their impacts on urban and regional PM2. 5: A
Georgia case study. Environmental Science & Technology, 43(2), 299-305.
https://doi.org/10.1021/es801827s

Tortell, P. D., Maldonado, M. T., & Price, N. M. (1996). The role of heterotrophic bacteria in
iron-limited ocean ecosystems. Nature, 383, 330-332.
https://doi.org/10.1038/383330a0191

Tortell, P. D, Maldonado, M. T., Granger, J., & Priece, N. M. (1999). Marine bacteria
and geochemical cycling of iron in the oceans. FEMS Microbiology Ecology, 29, 1–
11. https://doi.org/10.1111/j.1574-6941.1999.tb00593.x

Tovar-Sanchez, A., Sañudo-Wilhelmy, S. A., Kustka, A. B., Agustí, S., Dachs, J., Hutchins,
D. A., … Duarte, C. M. (2006). Effects of dust deposition and river discharges on
trace metal composition of Trichodesmium spp. in the tropical and subtropical
North Atlantic Ocean. Limnology and Oceanography, 51(4), 1755–1761.
https://doi.org/10.4319/lo.2006.51.4.1755

Trostle, K. D., Runyon, J. R., Pohlmann, M. A., Redfield, S. E., Pelletier, J., McIntosh, J., &
Chorover, J. (2016). Colloids and organic matter complexation control trace metal
concentration-discharge relationships in Marshall Gulch stream waters. Water
Resources Research, 52, 7931–7944.
https://doi.org/10.1111/j.17521688.1969.tb04897.x

184
Twining, B. S., & Baines, S. B. (2013). The trace metal composition of marine
phytoplankton. Annual Review of Marine Science, 5, 191–215.
https://doi.org/10.1146/annurev-marine-121211-172322

Umbreit, J. (2005). Iron deficiency: A concise review. American Journal of Hematology,
78(3), 225-231. https://doi.org/10.1002/ajh.20249

USGS. (2020a). California Drought. Retrieved from
https://ca.water.usgs.gov/california-drought/california-drought-runoff.html

USGS. (2020b). NWIS Site Information for USA: Site Inventory - Site 11118500. Retrieved
from https://waterdata.usgs.gov/nwis/inventory/?site_no=11118500

Van Geen, A., & Husby, D. M. (1996). Cadmium in the California Current system: Tracer of
past and present upwelling. Journal of Geophysical Research C: Oceans, 101(C2),
3489–3507. https://doi.org/10.1029/95JC03302

Van Hulten, M., Middag, R., Dutay, J. C., De Baar, H., Roy-Barman, M., Gehlen, M., …
Sterl, A. (2017). Manganese in the west Atlantic Ocean in the context of the first
global ocean circulation model of manganese. Biogeosciences, 14(5), 1123–1152.
https://doi.org/10.5194/bg-14-1123-2017

Van Leeuwen, T. T., Van Der Werf, G. R., Hoffmann, A. A., Detmers, R. G., Rücker, G.,
French, N. H. F., … Trollope, W. S. W. (2014). Biomass burning fuel consumption
rates: A field measurement database. Biogeosciences, 11(24), 7305–7329.
https://doi.org/10.5194/bg-11-7305-2014

Vink, S., Boyle, E. . A., Measures, C. . I., & Yuan, J. (2000). Automated high resolution
determination of the trace elements iron and aluminium in the surface ocean using a
towed Fish coupled to flow injection analysis. Deep Sea Research Part I:
Oceanographic Research Papers, 47(6), 1141–1156.
https://doi.org/10.1016/S0967-0637(99)00074-6

Wagner, S., Cawley, K. M., Rosario-Ortiz, F. L., & Jaffé, R. (2015). In-stream sources and
links between particulate and dissolved black carbon following a wildfire.
Biogeochemistry, 124(1–3), 145–161. https://doi.org/10.1007/s10533-015-0088-1

Wagner, S., Ding, Y., & Jaffé, R. (2017). A new perspective on the apparent solubility of
dissolved black carbon. Frontiers in Earth Science, 5, 1–16.
https://doi.org/10.3389/feart.2017.00075

Wagner, S., Jaffé, R., & Stubbins, A. (2018). Dissolved black carbon in aquatic
ecosystems. Limnology and Oceanography Letters, 3(3), 168–185.
https://doi.org/10.1002/lol2.10076
185

Wang, F. J., Chen, Y., Guo, Z. G., Gao, H. W., Mackey, K. R., Yao, X. H., … Paytan, A.
(2017). Combined effects of iron and copper from atmospheric dry deposition on
ocean productivity. Geophysical Research Letters, 44(5), 2546–2555.
https://doi.org/10.1002/2016GL072349

Wang, X., Xu C., Druffel E. M., Xue Y., & Qi Y. (2016). Two black carbon pools transported
by the Changjiang and Huanghe Rivers in China. Global Biogeochemical Cycles,
30(12), 1778–1790. https://doi.org/10.1002/2016GB005509

Wang, Z. L., & Liu, C. Q. (2003). Distribution and partition behavior of heavy metals
between dissolved and acid-soluble fractions along a salinity gradient in the
Changjiang Estuary, eastern China. Chemical Geology, 202, 383-396.
https://doi.org/1010.1016/j.chemgeo.2002.05.001

Ward, T. J., Hamilton, R. F., Dixon, R. W., Paulsen, M., & Simpson, C. D. (2006).
Characterization and evaluation of smoke tracers in PM: Results from the 2003
Montana wildfire season. Atmospheric Environment, 40(36), 7005–7017.
https://doi.org/10.1016/j.atmosenv.2006.06.034

Warrick, J. A., DiGiacomo, P. M., Weisberg, S. B., Nezlin, N. P., Mengel, M., Jones, B. H.,
…Farnsworth, K. L. (2007). River plume patterns and dynamics within the Southern
California Bight. Continental Shelf Research, 27(19), 2427–2448.
https://doi.org/10.1016/j.csr.2007.06.015

Warrick, J. A., Hatten, J. A., Pasternack, G. B., Gray, A. B., Goni, M. A., & Wheatcroft, R.
A. (2012). The effects of wildfire on the sediment yield of a coastal California
watershed. Bulletin of the Geological Society of America, 124(7–8), 1130–1146.
https://doi.org/10.1130/B30451.1

Wear, E. K., Carlson, C. A., James, A. K., Brzezinski, M. A., Windecker, L. A., & Nelson, C.
E. 2015). Synchronous shifts in dissolved organic carbon bioavailability and
bacterial community responses over the course of an upwelling-driven
phytoplankton bloom. Limonolgy and Oceanography, 60(2), 657–677.
https://doi.org/10.1002/lno.10042

Webb, E. A., Moffett, J. W., & Waterbury, J. B. (2001). Iron Stress in Open-Ocean
Cyanobacteria (Synechococcus, Trichodesmium, and Crocosphaera spp.):
Identification of the IdiA Protein. Applied and Environmental Microbiology, 67(12),
5444–5452. https://doi.org/10.1128/AEM.67.12.5444-5452.2001

186
Wei, W., Zhang, R., Peng, L., Liang, Y., & Jiao, N. (2018). Effects of temperature and
photosynthetically active radiation on virioplankton decay in the western Pacific
Ocean. Scientific Reports, 8(1), 2–10.
https://doi.org/10.1038/s41598-018-19678-3

Weitz, J. S. (2016). Quantitative Viral Ecology: Dynamics of Viruses and Their Microbial
Hosts (Vol. 73). Princeton: Princeton University Press.
https://doi.org/https://doi.org/10.1515/9781400873968

Westrich, J. R., Ebling, A. M., Landing, W. M., Joyner, J. L., Kemp, K. M., Griffin, D. W., &
Lipp, E. K. (2016). Saharan dust nutrients promote Vibrio bloom formation in marine
surface waters. Proceedings of the National Academy of Sciences of the United
States of America, 113(21), 5964–5969. https://doi.org/10.1073/pnas.1518080113

Whitfield, M., & Turner, D. R. (1979). Water-rock partition coefficients and the composition
of seawater and river water. Nature, 278, 132–137.
https://doi.org/10.1038/278132a0

Wilhelm, S. W. (1995). Ecology of iron-limited cyanobacteria: A review of physiological
responses and implications for aquatic systems. Aquatic Microbial Ecology, 9(3),
295–303. https://doi.org/10.3354/ame009295

Wilhelm, S. W., MacAuley, K., & Trick, C. G. (1998). Evidence for the importance of
catechol-type siderophores in the iron-limited growth of a cyanobacterium.
Limnology and Oceanography, 43(5), 992–997.
https://doi.org/10.4319/lo.1998.43.5.0992

Wilhelm, S. W., & Trick, C. G. (1994). Iron‐limited growth of cyanobacteria: Multiple
siderophore production is a common response. Limnology and Oceanography,
39(8), 1979–1984. https://doi.org/10.4319/lo.1994.39.8.1979

Wilhelm, S. W, & Suttle, C. A. (1999). Viruses and nutrient cycles in the sea: viruses play
critical roles in the structure and function of aquatic food webbs. BioScience,
49(10), 781–788. https://doi.org/10.2307/1313569

Wilson, S. T., Hawco, N. J., Armbrust, E. V., Barone, B., Björkman, K. M., Boysen, A. K.,
… Karl, D. M. (2019). Kīlauea lava fuels phytoplankton bloom in the North Pacific
Ocean. Science, 365(6457), 1040–1044. https://doi.org/10.1126/science.aax4767

Wilson, W. H., Carr, N. G., & Mann, N. H. (1996). the Effect of Phosphate Status on the
Kinetics of Cyanophage Infection in the Oceanic Cyanobacterium Synechococcus
Sp. Wh78031. Journal of Phycology, 32(4), 506–516.
https://doi.org/10.1111/j.0022-3646.1996.00506.x

187
Winton, V. H. L., Edwards, R., Bowie, A. R., Keywood, M., Williams, A. G., Chambers, S.
D., … Paton-Walsh, C. (2016). Dry season aerosol iron solubility in tropical northern
Australia. Atmospheric Chemistry and Physics, 16(19), 12829–12848.
https://doi.org/10.5194/acp-16-12829-2016

Witter, A. E., Hutchins, D. A., Butler, A., & Luther, G. W. (2000). Determination of
conditional stability constants and kinetic constants for strong model Fe-binding
ligands in seawater. Marine Chemistry, 69(1–2), 1–17.
https://doi.org/10.1016/S0304-4203(99)00087-0

Wong, G. T., Pai Su-Cheng, & Chung Shi-Wei. (1995). Cobalt in the Western Philippine
Sea. Oceanologica Acta, 18(6), 631–638.

Worsfold, P. J., Clough, R., Lohan, M. C., Monbet P., Ellis, P. S., Quétel, C. R., ...
McKelvie, I. D. (2013). Flow injection analysis as a tool for enhancing oceanographic
nutrient measurements—A review. Analytica Chimica Acta, 803, 15-40.
https://doi.org/10.1016/j.aca.2013.06.015

Wu, J., Roshan, S., & Chen, G. (2014). The distribution of dissolved manganese in the
tropical-subtropical North Atlantic during US GEOTRACES 2010 and 2011 cruises.
Marine Chemistry, 166, 9–24. https://doi.org/10.1016/j.marchem.2014.08.007

Xie, S. P., Lu, B., & Xiang, B. (2013). Similar spatial patterns of climate responses to
aerosol and greenhouse gas changes. Nature Geoscience, 6(10), 828–832.
https://doi.org/10.1038/ngeo1931

Yoon, V. K., & Stein, E. D. (2008). Natural Catchments as Sources of Background Levels
of Storm-Water Metals, Nutrients, and Solids. Journal of Environmental Engineering,
134(12), 961–973. https://doi.org/10.1061/(asce)0733-9372(2008)134:12(961)

Young, D. R., & Jan, T. K. (1977). Fire fallout of metals off California. Marine Pollution
Bulletin, 8(5), 109–112. https://doi.org/10.1016/0025-326X(77)90133-3

Young, R., Wang, I. N., & Roof, W. D. (2000). Phages will out: Strategies of host cell lysis.
Trends in Microbiology, 8(3), 120–128.
https://doi.org/10.1016/S0966- 842X(00)01705-4

Zhang, R., Kelly, R. L., Kauffman, K. M., Reid, A. K., Lauderdale, J. M., Follows, M. J., &
John, S. G. (2019). Growth of marine Vibrio in oligotrophic environments is not
stimulated by the addition of inorganic iron. Earth and Planetary Science Letters,
516, 148–155. https://doi.org/10.1016/j.epsl.2019.04.002

188
Zhuang, G., Yi, Z., Wallace, G. T., Wolf, A., Zheng, T., Witzel, K., … Fitzwater, S. E. (1988).
Iron deficiency limits phytoplankton growth in the north-east Pacific subarctic.
Nature, 331(6154), 341–343. https://doi.org/10.1038/331341a0

Zimmerman, A. E., Howard-Varona, C., Needham, D. M., John, S. G., Worden, A. Z.,
Sullivan, M. B., … Coleman, M. L. (2020). Metabolic and biogeochemical
consequences of viral infection in aquatic ecosystems. Nature Reviews
Microbiology, 18(1), 21–34. https://doi.org/10.1038/s41579-019-0270-x

Ziolkowski, L. A., & Druffel, E. R. M. (2010). Aged black carbon identified in marine
dissolved organic carbon. Geophysical Research Letters, 37(16), 4–7.
https://doi.org/10.1029/2010GL043963

Zurbrick, C. M., Gallon, C., & Flegal, A. R. (2017). Historic and industrial lead within the
Northwest Pacific Ocean evidenced by lead isotopes in seawater. Environmental
Science & Technology, 51(3), 1203-1212. https://doi.org/10.1021/acs.est.6b04666

189
Appendix
Iron recycling through the microbial loop in the North Pacific
Subtropical Gyre

Author Contributions
Rachel Kelly designed and performed the experiments and organized the manuscript.
Sallie Chisholm provided the Prochlorococcus and paired phage stocks. David Caron
provided the Uronema marinum and Paraphysomonas bandaiensis cultures.

Abstract
The persistence of marine microbial life in iron limiting conditions has conjured up
interest in studies focusing on the recycling of Fe amongst marine microbial communities.
In this study, we attempted to quantify the relative importance of three Fe cycling
mechanisms simultaneously in the natural environment, recycling through viral lysis and
byproducts of grazing, as well as cycling through siderophores. Using stable Fe isotopes
as tracers (
54
Fe,
57
Fe,
58
Fe) we were able to successfully label three different Fe ligands,
viral lysates, grazing byproducts, and siderophores. However, it is unclear if we observed
the biological uptake these labeled Fe ligands. There is some evidence that we may
observed the biological uptake of
57
Fe bound to siderophores (deferoxamine, DFB) in the
natural North Pacific Subtropical Gyre though the presence of other unexplainable
fluctuations in particulate Fe during the incubations makes us hesitant to draw any
conclusions from the dataset. We believe the lack of useful data from our study is the
result of four flaws in our experimental set up. By assessing the shortcomings of our
experiments we provide recommendations for future similar studies.

190
1. Introduction
The marine microbial community is responsible for producing 50 to 80% of the
world’s oxygen (Field et al., 1998; Pomeroy et al., 2007; Simon et al., 2009) and for
removing about one third of the anthropogenic carbon dioxide from the atmosphere (Basu
& Mackey, 2018; Häder, et al., 2014). In order for the marine microbial community to
efficiently take up carbon and produce oxygen phytoplankton need the nutrients essential
for primary productivity to be bioavailable. Most of the world’s oceans are far from sources
of essential nutrients therefore a substantial amount of productivity is nutrient limited
(Moore et al., 2013), yet microbial life continues to subsist. In a paper by Azam et al. (1983)
the term “microbial loop” was first introduced which refers to the idea that microbes are
recycling nutrients amongst themselves and sometimes these nutrients can be transferred
to higher trophic levels through grazing (Wilhelm & Suttle, 1999). The microbial loop is
important for the recycling of bioavailable iron (Fe) which is an essential cofactor for
proteins needed in photosynthesis (e.g. ferrodoxin, cytochromes) and respiration (e.g. Fe-
S proteins)(Twining & Baines, 2013). Relieving Fe limitation is crucial for marine microbes
as one third of the world’s ocean is thought to be Fe limited (Boyd et al., 2007; Moore et
al., 2013).
A major source of recycled Fe in the microbial loop is cell debris, which can be a
byproduct of the viral lysis or grazing of microbes. Previous studies have shown that
important nutrients such as C, N, P, and Fe are recycled through the “viral shunt” (Wilhelm
& Suttle, 1999), the process in which dissolved, bioavailable nutrients associated with cell
debris are released after viral lysis (lysates; Gobler et al., 1997; Poorvin et al., 2004). Other
191
studies have also observed that nutrients, including Fe, can be recycled through grazing
(Hutchins & Bruland, 1994; Hutchins et al., 1993). Grazers release nutrients in the
dissolved phase when cells are not fully consumed, or cellular material is excreted post
digestion. Laboratory incubations have found that grazers can also ingest refractory
particulate or colloidal Fe, which is then converted into dissolved bioavailable Fe in the
acidic food vacuoles of the grazer (Barbeau et al., 1996).
Several Fe tracer experiments have been designed to try and quantify the recycling
rates and bioavailability of Fe in viral lysates, grazing byproducts, and other ligands. A
study using radiolabeled
55
Fe as a tracer concluded that 19 to 75 pM Fe d
-1
is regenerated
by viral lysis, Fe from bacterial lysates are more bioavailable than inorganic Fe and Fe
bound to synthetic chelators (EDTA), and that lysate-bound Fe supported ~90% of the
primary production in high nutrient low chlorophyll (HNLC) regions (Poorvin et al., 2004).
Additionally, using Fe-dependent bioluminescent reports, Mioni et al. (2005) found that Fe
in viral lysates was the most bioavailable source of Fe in comparison Fe bound to naturally
produced siderophores > EDTA > a synthetic siderophore (trihydroxamate desferrioxamine
B). In contrast, a study using radiolabeled
59
Fe determined that there was no net release of
dissolved Fe (dFe) after lysis, but that subsequent photochemical degradation of the cell
debris enhanced dFe levels by 50% (Gobler et al., 1997). Grazer-focused incubations in
the equatorial Pacific, an HNLC region, showed that over 93% of the
59
Fe in diatom cells
(10 µm in size) was recycled and taken up by Synechococcus cells (≤1 µm in size) when a
grazer was present, suggesting that grazing promoted this transfer of Fe (Hutchins et al.,
1993). Hutchins & Bruland (1994) also found that dissolved concentrations of
55
Fe and
59
Fe
192
increased by 3 to 7 times when a grazer was present compared to the control incubations
with no grazer. Lastly, a study done by Strzepek et al. (2005) in which an unperturbed
mesoscale patch of HNLC water was labeled with SF6 concluded that 0.4 to 28 pM Fe d
-1

was regenerated by viral lysis and 15 to 25.5 pM Fe d
-1
was regenerated by grazing.
Most prior work has concentrated on one or two Fe recycling pathways in isolation,
but in the natural environment there are multiple components of the microbial loop’s Fe
cycle. These studies (Gobler et al., 1997; Hutchins & Bruland, 1994; Hutchins et al., 1993;
Mioni et al., 2005; Poorvin et al., 2004; Strzepek et al., 2005) have no doubt given us a lot
of insight as to how these Fe recycling pathways operate, however we must start to
investigate how these Fe ligands work side-by-side. No study to date has tried to quantify
the relative importance of three Fe cycling mechanisms simultaneously in the natural
environment, recycling through viral lysis and grazing along with cycling through
siderophores, as we attempted to do here.
To study the relative bioavailability of Fe associated with viral lysates, grazing
byproducts, and siderophores we conducted incubations with stable Fe isotope tracers
bound to each of these Fe sources. First, we successfully labeled each of the Fe ligands,
viral lysates, byproducts of grazing, and siderophores, using three different stable Fe
isotope spikes,
54
Fe,
57
Fe, and
58
Fe. We then attempted to add each of the prelabeled Fe
ligands to incubations with natural microbial communities. Our incubation experiments
were conducted during the 2021 SCOPE PARticles And Growth in the Oceanic Nutricline
(SCOPE-PARAGON) cruise in the center of the North Pacific Subtropical Gyre (NPSG), the
largest circulation feature throughout the world’s ocean (Karl, 1999). We believe that using
193
stable Fe isotope tracers will be a more effective approach to determining the relative
bioavailability of different Fe complexes because we will be able to trace multiple Fe
ligands in a single incubation bottle with better sensitivity in comparison to other tracer
techniques, such as radiolabeled Fe and bioluminescent reports.
Prior work has concentrated on Fe recycling mechanisms in lab incubations and in
Fe-limited HNLC regions, whereas this study focuses on an oligotrophic region with low Fe
concentrations that are not always growth limiting (~0.2 nM dFe). Much of the microbial life
in oligotrophic NPSG waters have become acclimated to the low Fe environment, for
example the ubiquitous cyanobacteria Prochlorococcus (Campbell & Vaulot, 1993; Karl,
1999; Letelier et al., 1993), by taking up siderophore-bound Fe (Bundy et al., 2018;
Neilands, 1995; Schwyn & Neilands, 1987; Wilhelm, 1995; Wilhelm & Trick, 1994). For
instance, Bundy et al. (2018) found that the concentration of Fe bound to siderophores at
Station ALOHA in the NPSG were two orders of magnitude higher than the inorganic Fe
present, and that siderophores were present at nearly every depth. A study by Hutchins et
al. (1999) also found that Fe bound to a variety of siderophores was preferentially taken up
by cyanobacteria, in comparison to eukaryotic phytoplankton. Thus we believe that even
though lab cultures have shown that the Fe bound to viral lysates and grazing byproducts
is generally more bioavailable, we suspect the natural microbial community in the NPSG
will already be producing Fe siderophores and will preferentially take up the Fe bound to
similar ligands.

194
2. Methods
Two major steps had to be done in order for these stable Fe isotope tracer
experiments to be successful. First, we prelabeled each of the Fe ligands (viral lysates,
byproducts of grazing, and siderophores) using three different stable Fe isotope spikes
(
54
Fe,
57
Fe, and
58
Fe). Second, we incubated whole seawater which included the natural
microbial assemblage from the NPSG and added prelabeled Fe complexes.
2.1 Labeling the iron complexes
2.1.1 ALOHA media
PRO99 natural seawater medium made with low-metal, surface water from Station
ALOHA was used to grow in lab Prochlorococcus cultures (Moore et al., 2007). The
surface Station ALOHA water was first filtered through a 0.2 µm Supor
®
filter and then
microwave sterilized (Keller et al., 1988). Aliquots of filter-sterilized (0.2 µm) nutrient stocks
were added to the sterilized ALOHA seawater to a final concentration of 50 µM
NaH2PO4·H2O, 800 µM NH4Cl, 1.17 µM Na2EDTA·2H20 (unless noted otherwise), 1.18 µM
FeCl3·6H2O (unless noted otherwise), 0.008 µM ZnSO4·7H2O, 0.090 µM MnCl2·4H2O,
0.005 µM CoCl2·6H2O, 0.003µM Na2MoO4·2H2O, 0.010 µM Na2SeO3, and 0.010 µM
NiCl2·6H2O. Both NaH2PO4·H2O and NH4Cl stocks were chelexed prior to being filter-
sterilized and added to the seawater base.
2.1.2 Prochlorococcus and grazer cultures
The most prevalent organism in the NPSG is the cyanobacteria Prochlorococcus,
therefore much of the prelabeled organic Fe-binding complexes in our experiments were
products of Prochlorococcus. Three Prochlorococcus strains were grown: MED4ax,
195
NATL2A, and MIT9313 (paired phage: P-SSP7, P-RSM2, and P-SS2, respectively).
Culturing procedures were adapted from Moore et al. (2007). Cool white fluorescent lamps
were used at low, continuous light levels of about 20 µmol Q m
-2
s
-1
and exposed to the
incubations through window screening. The incubations were placed in an incubator set at
23°C. Cultures were grown in 500 mL acid washed and sterilized polycarbonate (PC)
Nalgene
®
bottles. Cell growth was monitored by fluorometry, and cultures were diluted
about every 4 days with new media.
Two grazers that feed on Prochlorococcus were also cultured: Uronema marinum,
a ciliate, and Paraphysomonas bandaiensis, a chrysophyte. Both grazers were cultured at
the same light level and temperature as Prochlorococcus. The consumption of
Prochlorococcus (and the resulting growth of the grazer) was monitored by fluorometry,
and grazer cultures were diluted about every 4 days using Prochlorococcus cultures.
2.1.3 Labeling viral lysates
To make stable Fe isotope labeled viral lysates an amended version of the protocol
found in Poorvin et al. (2004) was used. Prochlorococcus cultures were first grown in
ALOHA media without any added Fe or EDTA for multiple generations. After 4 to 5
transfers, the Prochlorococcus cultures were diluted with ALOHA media with 0.2 µM or 2
µM of either
54
Fe,
57
Fe, or
58
Fe stable isotope spike. Once the Fe-labeled Prochlorococcus
cells had reached exponential growth (2 to 3 days) the cells were then collected onto a 0.8
µm Supor
®
filter using gravity filtration. The cells on the Supor
®
filter were rinsed with
filtered and sterilized Station ALOHA seawater to remove any excess extracellular Fe. The
filter with the rinsed cells was placed in a 500 mL PC bottle with about 25 mL of no added
196
Fe or EDTA ALOHA media just so the filter was barely submerged. A 500 µL stock of
paired phage was added to the bottle, and the bottle was left in the incubator for at least
48 hours. The mixture was then 0.2 µm filtered to exclude cells and the filtrate which
contained the viral progeny was collected and labeled as the new viral stock.
An axenic Prochlorococcus culture was then split into two PC bottles, the new viral
stock was added to one bottle, the experimental bottle, and the health of the cultures in
both the experimental and control bottles were monitored using fluorometry to determine
whether the host cultures in the experimental bottle had been successfully infected. If
successfully infected, which was noted by steadily decreasing absorption values, the
experimental bottle would be 0.02 µm filtered and the filtrate was kept at 7°C as the
labeled-Fe viral lysate stock.
2.1.4 Labeling grazing byproducts
The method used to label grazing byproducts was very similar to the one used to
label viral lysates. Instead of phage being introduced to the rinsed host cells submerged in
25 mL of no added Fe or EDTA ALOHA media, a grazer was introduced. If grazing
occurred, which was noted by steadily decreasing absorption values, the experimental
bottle would be 0.2 µm filtered and the filtrate was kept at 7°C as the labeled-Fe grazing
byproduct stock.
2.1.5 Labeling siderophores
A siderophore, deferoxamine mesylate (DFB) salt (≥92.5%, TLC), was used in the
incubations. DFB is a hexadentate organic ligand with hydroxomate functionalities
(Albrecht-Gary & Crumbliss, 1998). The chelator was introduced to dissolved stable Fe
197
isotope spike at a 2:1 ratio (DFB to Fe spike). The mixture was then filtered through a 0.2
µm Supor
®
filter and the filtrate was kept at 7°C as the labeled-Fe siderophore stock.
2.1.6 Measuring
54
Fe,
57
Fe, and
58
Fe in stocks
The Fe isotope analyses for each of the Fe complex stocks were performed in a
class-100 clean room. A 15 mL subsample of each of the Fe complex stocks was acidified
to pH 2 using distilled HCl. After 24 hours of acidification, the 15 mL subsamples of each
stock were extracted onto a column with Nobias chelating resin PA-1 (Sohrin et al., 2008)
at pH 6 with an ammonium acetate/acetic acid buffer, and the preconcentrated sample
was eluted into 0.5 mL 1 M HNO3 which included 1 ppb indium (In) as an internal
standard. Final concentrations in all samples were measured on an Element 2 inductively
coupled plasma mass spectrometer (ICP-MS, Thermo Fisher Scientific). The concentration
and isotopic composition for each of the Fe spikes was quantified relative to a 10 ppb Fe
standard.
2.2 Fe addition incubations
2.2.1 Incubations in the field
Shipboard incubations were done onboard the R/V Kilo Moana during the 2021
SCOPE-PARAGON cruise in an anticyclonic eddy north of the Hawaiian islands at about
22°N, 156°W. Seawater for incubations was collected using a trace metal rosette and
trace metal clean (TMC) sampling protocols (Measures et al., 2008). Two experiments
were conducted during the span of the cruise. For Experiment 1, each incubation bottle
had a total volume of 200 mL which consisted of 25 m whole seawater and 1 nM of each
labeled Fe ligands (total of 3 nM dFe added to incubations). In Experiment 2, each
198
incubation bottle had a total volume of 200 mL which consisted of 200 m whole seawater,
30 mL of 200 m net trap material, and 1 nM of each labeled Fe ligand (total of 3 nM dFe
added to incubations).
For each experiment there were three treatments: (1) control – untreated, unfiltered
whole seawater, (2) heat killed control – whole seawater was boiled and cooled before
start of incubation and isotopically labeled Fe ligands were then added, (3) experimental –
untreated whole seawater with isotopically labeled Fe ligands. Triplicates were done for
each treatment, therefore a total of nine 500 mL sterilized TMC PC bottles were used in
each experiment. Experiment 1 bottles, with 25m whole seawater, were incubated in on
deck incubators with a flow-through seawater system with blue Rohm and Haas Plexiglas
filters. Experiment 2 bottles, with 200 m whole seawater, were incubated in a Caron
®
incubator at 19.5°C in the dark.
Samples were taken twice in each experiment, at 0 hours (T0) and 24 hours (T24).
At each timepoint two 50 mL subsamples were taken, one for total particle associated Fe
concentrations and another for intracellular Fe concentrations. The total particle associated
Fe subsample was filtered onto a TMC 47 mm 0.2 µm Supor
®
filter and stored. For the
intracellular Fe subsample, 5 mL of oxalate-EDTA solution was added to the subsample,
mixed for 30 minutes, and then filtered onto a TMC 47 mm 0.2 µm Supor
®
filter. Then 10
mL of corresponding (25 or 200 m) dissolved seawater was passed through the
intracellular Fe subsample filter before the filter was collected and stored. The filtrate and
filters were collected for each subsample in TMC 50 mL and 15 mL centrifuge tubes,
respectively. All samples were stored at room temperature in the dark until being
199
processed at the University of Southern California where the rest of the sample digestions
and ICP-MS analyses were done.
2.2.2 Sample digestions
Sample filters were placed in clean Savillex perfluoroalkoxy alkanes (PFA) vials. For
each of the PFA vials, 5 mL of 50% nitric acid was added, vials were then capped tightly,
and placed on a hot plate at 95°C for 2 to 3 days. Over the 2 to 3 day period each of the
vials were inverted at least once a day. Sample filters were then removed from the vials,
the vials were placed back on the hot plate, uncapped, and left to dry overnight at 150°C.
100 µL of pure HNO3 and 100 µL of pure HCl were then added to each dried vial, vials
were recapped, and placed back on the hot plate for approximately 4 hours at 150°C. The
vials were then uncapped and left to dry down. 1.5 mL of 0.1M HNO3 with 10 ppb In was
added to each of the vials, left to sit overnight, and transferred to TMC 15 mL centrifuge
tubes for ICP-MS analysis. The concentration and isotopic composition for each of the Fe
spikes was quantified relative to a 10 ppb Fe standard.

3. Results and Discussion
The experiments presented here had two methodological challenges, labeling viral
lysates, grazing byproducts, and siderophores with
54
Fe,
57
Fe, and
58
Fe (Figure A-1), and
setting up incubations with a sufficient amount of Fe ligands and biomass to induce the
biological uptake of isotopically labeled Fe (Figure A-2). Ultimately, we successfully labeled
each of the Fe ligands, though it was unclear if we observed the biological uptake of the
pre-labeled Fe ligands. We hope by discussing the flaws in our experimental set up and
200
the outcomes of our incubations we can provide valuable information for similar future
studies.
In the first step of the experiment we successfully labeled viral lysates, grazing
byproducts, and siderophores with
54
Fe,
57
Fe, and
58
Fe, but unfortunately the stock
solutions were very diluted. The dissolved Fe concentrations (dFe) of the isotopically
labeled stocks ranged between 1.37 to 13.6 nM dFe however the volumes of the stocks
ranged from 50 to 275 mL (Table A-1). The small dFe to volume ratios meant that the
stocks were quite diluted and so the stock solutions made up a large portion of the total
200 mL per incubation bottle (approximately 34 to 53%). The final Fe stock to natural
seawater volume ratios meant that the phytoplankton assemblage was diluted in the
incubations. Prior to the incubation experiments, the Fe stocks could have been
concentrated using Amberlite XAD-16 resin followed by rotary evaporation (Poorvin et al.,
2004; Mioni et al., 2005). The Amberlite XAD-16 resin would have isolated and extracted
the organic fraction of the stock solution. The rotary evaporation would then concentrate
the Fe bound ligands even further.
Most of the particulate Fe data from the incubation experiments shows no clear
sign of biological uptake for any of the isotopically labeled Fe ligands. In Experiment 1,
there is a significant increase in total particulate
58
Fe concentration (p
58
Fe) in the heat killed
treatment after 24 hours (paired t-tests, n=3, p<0.05; Figure A-3, Table A-2). This increase
of total p
58
Fe could be the adsorption of DFB-bound
58
Fe to particulate matter since we
did not observe an increase in the intracellular p
58
Fe. In Experiment 2, total and intracellular
p
57
Fe significantly increased in the experimental treatment after 24 hours (Figure A-4, Table
201
A-3). All other fluctuations in pFe after the incubation period for both Experiments 1 and 2
are insignificant. Higher p
54
Fe values in both experiments is unsurprising as
54
Fe is the
most naturally abundant of the three stable Fe isotopes used in these experiments (%
natural abundances:
54
Fe – 5.85%,
57
Fe – 2.12%,
58
Fe – 0.28%).
There is a chance that the significant increase in experimental total and intracellular
p
57
Fe in Experiment 2 could be a real signal and there was biological uptake of DFB-bound
57
Fe (Figure A-4). Past studies have shown that DFB-bound Fe is bioavailable to
phytoplankton (Maldonado & Price, 2001; Maldonado & Price, 1999). Others have
hypothesized that cyanobacteria can effectively take up Fe from DFB due to ligand-ligand
Fe exchange between the DFB and the cyanobacteria-produced siderophores (Eldridge et
al., 2004). According to preliminary calculations using the average primary productivity in
surface Station ALOHA water (6 mg C m
-3
d
-1
) and the C:Fe ratio (121000:12.2) by
Cunningham & John (2017) we estimated that ~5 pM of dFe is taken up a day at Station
ALOHA. If we assume this Fe uptake rate applies to our study region, then the 24 hour
increases of total (33.72 pM) and intracellular (13.54 pM) p
57
Fe in Experiment 2 could be
real signals of DFB-Fe uptake. Even when the amount of final intracellular p
57
Fe in the
control is subtracted from the final experimental intracellular p
58
Fe, we still see an increase
of 9.78 pM p
58
Fe. If this increase in p
58
Fe is a true signal than this result supports our
original hypothesis, that the natural community in the NPSG near Station ALOHA would
preferentially take up Fe bound to siderophores.
There are three plausible reasons why the isotopically labeled Fe uptake
incubations were not as successful: (1) a 24 hour incubation period was too short to
202
observe Fe uptake, (2) dFe in the incubations was adsorbed onto excess DFB, and (3)
there was not enough living biomass to take up the dFe. Past studies have shown the
uptake of labelled Fe in under 24 hours (Lis et al., 2015; Mioni et al., 2005; Poorvin et al.,
2004; Strzepek et al., 2005) therefore we assumed 24 hours was a suitable incubation
period. A 24 hour incubation period was also chosen because we were concerned about
the exchange of labelled Fe amongst the different ligands after an extended period of time.
Measuring and characterizing the Fe-binding conditional stability constants (logKFe’L or
Kcond) for viral lysates and grazing byproducts (Kcond = 9.14 to 9.34; Witter et al., 2000), and
DFB (Kcond = 8.56 to 16.5, Buck et al., 2010; Rue & Bruland, 1995; Witter et al., 2000) has
been a highly contended area of research (Hutchins et al., 1999). These metal ligands are
routinely categorized into a stronger ligand class (L1; Kcond > 11) and a weaker ligand class
(L2; Kcond < 11) (Buck et al., 2010; Rue & Bruland, 1995). Based off the similarities in Kcond
for each of the mostly L2-type ligands in our experiments, we chose to use a shorter
incubation period.
The presence of any excess ligands in our incubations could be problematic, due to
the similarities between Fe-binding capacity of the viral lysates, grazing byproducts, and
DFB. The DFB stock used in our study contained a 2:1 DFB to Fe spike ratio to ensure
that all of the dFe added to the stock solution was bound to DFB. The higher ratio of
DFB:Fe meant that some uncomplexed DFB was added to the incubations (~1 nM DFB).
Even though some of the estimates for DFB’s stability constant are similar to viral lysates
and grazing byproducts (Kcond =8.56, Witter et al., 2000), there are some higher estimates
of DFB’s stability constant which indicate there is a possibility the DFB could “pull” Fe from
203
the other ligands (Kcond =12.9, Buck et al., 2010; Kcond = 16.5, Rue & Bruland, 1995). Even
if the lower estimates of DFB stability constants are accurate previous studies have shown
naturally occurring Fe-ligands are more susceptible to dissolution via photochemical
activity in comparison to Fe bound to DFB, which is considered relatively photo-stable
(Barbeau et al., 2003; Eldridge et al., 2004). Future experiments may want to consider
using siderophores with lower Fe-binding conditional stability constants and adding
stability constant measurements to their sampling. If DFB is used, future studies should
consider using a 1:1 DFB to labeled Fe ratio.
We acknowledge that Fe uptake rates, incubation time, and the amount of biomass
in the incubations are not independent of one another. Fe uptake rates should increase
with higher biomass due to the physical interactions between bioavailable Fe and living
cells (Sunda & Huntsman, 1995). Using chlorophyll concentrations as a proxy for biomass,
we find that our study region and depths sampled had low levels of productivity (average
chlorophyll a: 0.13 µg/L at 25 m & 0.09 µg/L at 250 m). The low levels of biomass and the
high volume Fe stock additions in each incubation most likely reduced Fe uptake rates. In
an effort to add more biomass we added net trap material to our Experiment 2
incubations. Coincidentally, the only plausible sign of biological Fe uptake was observed in
the experimental p
57
Fe samples in Experiment 2 (Figure A-4). The added net trap material
therefore may have increased the living biomass and subsequently the biological uptake of
Fe in these incubations.

204
4. Conclusion
The experiments presented here are novel in that no published study to date has
attempted to compare the bioavailability of Fe bound to viral lysates, grazing products, and
siderophores in single bottle incubations using stable Fe isotopes as tracers. Although
these incubations were unsuccessful in answering our original experimental questions, we
hope that what we have learned from our shortcomings will be useful to the design of
similar future studies. The use of naturally produced L2-type siderophores is
recommended to better ensure that the isotopically labeled Fe stays with its original ligand.
After being isotopically labeled each of the Fe stocks should be concentrated to avoid
diluting the biomass in incubations. Then, in an effort to increase living biomass, we
suggest that preliminary incubations either be conducted with laboratory cultures or more
productive marine communities (e.g. coastal waters). We also recommend that the
incubation period be extended with more frequent sampling in case 24 hours is not
enough time to observe sufficient Fe uptake. Finally, volume permitting, we advise
measuring the Fe-binding conditional stability constants of the Fe-labeled ligands and
monitoring primary productivity and biomass in the incubations.

205
Table A-1
Detailed information on Fe-labeled (
54
Fe,
57
Fe,
58
Fe) viral lysate and grazing byproduct
stocks. A total of six stocks were made, but only B, C, and D filtrate were used in
incubations.

Stock ID Pro Host
Host Media
dFe (μM)
Top-Down Control
Labelled
Fe
Stock
d
54
Fe (nM)
Stock
d
57
Fe (nM)
Stock
d
58
Fe (nM)
Volume needed
for 1 nM Fe per
200 mL
A MIT9313 0.2 P-SS2 phage
57
Fe 0 3.89 0.4 51.41
B NATL2A 0.2 P-RSM2 phage
58
Fe 0 0.5 3.73 53.62
C MIT9313 2
Paraphysomonas
bandaiensis grazer
54
Fe 13.6 0.1 0.32 14.71
D NATL2A 2 P-RSM2 phage
57
Fe 0.1 8.6 0.72 23.26
E NATL2A 0.2
Paraphysomonas
bandaiensis grazer
58
Fe 0 0.1 1.37 145.99
F NATL2A 2
Paraphysomonas
bandaiensis grazer
57
Fe 0 6.5 0.75 30.77

206
Table A-2
Results from Experiment 1 where 1 nM of each Fe-labeled ligand was added to 25 m
whole seawater, incubated in on deck incubators, and sampled at 0 hours and 24 hours.
Three treatments were done, control, heat killed control, and experimental, and each was
sampled for total particulate and intracellular iron. A paired t-test was performed between
the replicate Fe concentrations from 0 hours and 24 hours for each treatment. Asterisks
next to some 24 hour Fe concentrations indicates a p-value <0.05.

Treatment
54
Fe (nM)
57
Fe (nM)
58
Fe (nM)
Grazing Byproduct Viral Lysate Siderophore
0 hours 24 hours 0 hours 24 hours 0 hours 24 hours
Total Control 19.36 ± 11.99 32.01 ± 14.11 0 0 0 0
Intracellular Control 16.02 ± 24.46 31.32 ± 20.09 0 1.97 ± 3.41 2.40 ± 4.16 5.59 ± 6.43
Total Heat Killed 81.15 ± 55.65 113.00 ± 63.37 7.09 ± 3.17 10.27 ± 9.05 8.37 ± 4.12 25.85* ± 5.34
Intracellular Heat Killed 11.81 ± 12.65 18.16 ± 10.92 0 0 1.91 ± 3.30 0
Total Experimental 48.22 ± 6.60 2.87* ± 4.98 9.07 ± 4.91 11.95 ± 8.78 3.69 ± 3.65 8.00 ± 4.50
Intracellular Experimental 13.32 ± 2.79 20.90 ± 16.44 0 0 0 0.70 ± 1.22

207
Table A-3
Results from Experiment 2 where 1 nM of each Fe-labeled ligand was added to 200 m
whole seawater with 200 m net trap material, incubated in dark Caron® incubators, and
sampled at 0 hours and 24 hours. Three treatments were done, control, heat killed control,
and experimental, and each was sampled for total particulate and intracellular iron. A
paired t-test was performed between the replicate Fe concentrations from 0 hours and 24
hours for each treatment. Asterisks next to some 24 hour Fe concentrations indicates a
pvalue <0.05.

Treatment
54
Fe (nM)
57
Fe (nM)
58
Fe (nM)
Grazing Byproduct Siderophore Viral Lysate
0 hours 24 hours 0 hours 24 hours 0 hours 24 hours
Total Control 41.89±11.27 47.62±41.26 0 0.47±0.81 0 0
Intracellular Control 117.32±81.29 32.04±18.78 0 3.76±6.51 2.62±4.53 0
Total Heat Killed 218.39±30.21 260.68±40.05 0 0 23.66±3.53 41.81±45.43
Intracellular Heat Killed 79.80±6.09 61.55±23.94 3.29±5.70 1.13±1.96 2.52±4.37 5.97±6.08
Total Experimental 89.84±82.36 78.38±30.42 12.53±21.70 46.25*±19.99 28.02±29.98 43.71±11.49
Intracellular Experimental 42.66±14.09 7.62±11.36 0 13.54*±4.60 0 12.53±11.53

208

Figure A-1
A schematic of how the grazer byproducts and viral lysates were labeled with stable Fe
isotopes (
54
Fe,
57
Fe,
58
Fe). The steps are oversimplified for this figure and the organisms,
Fe, viruses, and bottles/tubes are not to scale.

Fe
Fe
Prochlorococcus cell
Extracellular Fe
Intracellular Fe
Paraphysomonas bandaiensis grazer
Prophage virus
2. Incubate
3. Collect &
Wash Cells
1. Add Labeled
Fe to Cultures
4. Add Grazers or Viruses
5. Collect
Cell Debris
6. Measure Stock [Fe]
Fe
Fe
Fe
Fe Fe
Fe
Fe
Fe
Fe
Fe
Fe Fe Fe Fe Fe Fe
Fe
Fe
Fe
209

Figure A-2
A schematic of how the stable Fe isotopes (
54
Fe,
57
Fe,
58
Fe) tracer incubations were
executed. The steps are oversimplified for this figure and the organisms, Fe, bottles/tubes,
and instruments are not to scale.

Prochlorococcus cell
54
Fe labeled ligand
57
Fe labeled ligand
58
Fe labeled ligand
2. Sample at T0 & T24 4. Collect Filter
5. Process & Measure
on MC-ICP-MS
3. Subsample for total
particulate and intracellular Fe
1. Add Labeled Fe Stocks
to Natural Seawater
Unwashed Oxalate Washed
210

Figure A-3
Data from Experiment 1 of the labeled-Fe addition incubations in which 1 nM of each Fe-
labeled ligand was added to 25 m whole seawater, incubated in on deck incubators, and
sampled at 0 hours and 24 hours. Three treatments were prepared, control, heat killed
control, and experimental, and each was sampled for total particulate (navy) and
intracellular (green) iron. A paired t-test was performed between the replicate Fe
concentrations from 0 hours and 24 hours for each treatment. Asterisks next to some 24
hour Fe concentrations indicates a p-value <0.05. Fe concentrations above the y-axis
scale are labeled appropriately.
Amount of Labeled
Spike Added
Type of Ligand Stock ID
1 nM Fe54 grazing byproduct C Filtrate
1 nM Fe57 phage lysate D Filtrate
1 nM Fe58 siderophore DFB
Experimental
*
0
25
50
75
100
T0 T24 T0 T24 T0 T24
0
25
50
75
100
T0 T24 T0 T24 T0 T24
0
25
50
75
100
T0 T24 T0 T24 T0 T24
Total Intracellular
Heat Killed
Control
54
Fe
grazing byproduct
57
Fe
phage lysate
58
Fe
siderophore
113
Fe concentration(nM) Feconcentration(nM) Fe concentration(nM)
211

Figure A-4
Data from Experiment 2 of the labeled-Fe addition incubations in which 1 nM of each Fe-
labeled ligand was added to 200 m whole seawater with 200 m net trap material,
incubated in dark Caron
®
incubators, and sampled at 0 hours and 24 hours. Three
treatments were prepared, control, heat killed control, and experimental, and each was
sampled for total particulate (navy) and intracellular (green) iron. A paired t-test was
performed between the replicate Fe concentrations from 0 hours and 24 hours for each
treatment. Asterisks next to some 24 hour Fe concentrations indicates a p-value <0.05. Fe
concentrations above the y-axis scale are labeled appropriately.
Amount of Labeled
Spike Added
Type of Ligand Stock ID
1 nM Fe54 grazing byproduct C Filtrate
1 nM Fe57 siderophore DFB
1 nM Fe58 phage lysate B Filtrate
0
25
50
75
100
T0 T24 T0 T24 T0 T24
0
25
50
75
100
T0 T24 T0 T24 T0 T24
0
25
50
75
100
T0 T24 T0 T24 T0 T24
Total Intracellular
Control
Heat Killed
Experimental
54
Fe
grazing byproduct
57
Fe
siderophore
58
Fe
phage lysate
117
218 261
*
*
Fe concentration(nM) Feconcentration(nM) Fe concentration(nM)

Asset Metadata

Creator Kelly, Rachel Lauren (author)

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

Contributor Electronically uploaded by the author (provenance)

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Geological Sciences

Degree Conferral Date 2022-05

Publication Date 04/12/2022

Defense Date 03/07/2022

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

Tag aerosols,floodwaters,marine,Metals,OAI-PMH Harvest,seawater,virus,wildfire

Format application/pdf (imt)

Language English

Advisor John, Seth G. (committee chair), Hutchins, David A. (committee member), Moffett, James W. (committee member), West, A. Joshua (committee member)

Creator Email kellyrl@usc.edu,rachelaurenk@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC110937259

Unique identifier UC110937259

Document Type Dissertation

Format application/pdf (imt)

Rights Kelly, Rachel Lauren

Type texts

Source 20220412-usctheses-batch-922 (batch), 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 uscdl@usc.edu

Abstract (if available)

Abstract Since the first accurate marine metal measurements in the 1970s, trace metals in the ocean have been identified as critical components in major biogeochemical cycles, because most metals serve as essential nutrients for phytoplankton. Even though the field of trace metal marine chemistry has advanced considerably over the last 50 years, simple sampling methods are needed to make metal analyses more routine so that the effects of metals on the marine ecosystem can continue to be explored. With frequent metal sampling and a better understanding of the various roles metals play in marine biology we can better predict how life in the global ocean will look like under the future climate. The work presented in this thesis first investigates if simple pole sampling can produce reliable metal data that captures the nuances in surface metal concentrations throughout the world’s oceans. Subsequent chapters then look at the transport of metals by wildfires, which are expected to intensify worldwide due to climate change, and observe how changes in metal concentrations, particularly Fe, can affect both bottom-up and top-down controls on the marine ecosystem.
Using a simple pole sampler, 242 surface metal samples collected from the North Atlantic, North Pacific, and South Pacific Oceans produced pristine trace metal measurements, that were comparable to previous measurements made by the GEOTRACES program. Machine learning algorithms and environmental data were then used to identify the likely sources of metals to the sampling regions, such as upwelling in the South Pacific and anthropogenic aerosol deposition in the North Pacific. In another study, a wildfire was directly surveyed through aerosol, floodwater, and seawater samples. Aerosol and floodwater samples collected during the wildfire and subsequent flash flood event indicated that metals, along with black carbon, were mobilized by the fire, but seawater samples showed that fire-associated metals did not greatly affect the overall coastal metal inventory. The fire also did not appear to immediately impact organic matter cycling in the coastal ocean. However it is possible that the marine microbial communities could be impacted as fires transport metals to the open ocean. Finally, I show how changes in Fe concentration in laboratory cultures have varying effects on the viral infection of marine heterotrophs, specifically Vibrio bacteria. While higher Fe was found to increase the growth rates of both strains of Vibrio bacteria, higher Fe only increased the viral progeny of one infected strain of Vibrio and not the other. If Fe also has the same impacts on viral dynamics in natural ocean communities, an increase in viral progeny due to increases in Fe could impact diversity and nutrient cycling within microbial communities.