Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Discerning local and long-range causes of deoxygenation and their impact on the accumulation of trace, reduced compounds
(USC Thesis Other)
Discerning local and long-range causes of deoxygenation and their impact on the accumulation of trace, reduced compounds
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Discerning local and long-range causes of deoxygenation and their impact on the accumulation of trace, reduced compounds by Natalya Evans 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 (OCEAN SCIENCES) December 2022 Copyright 2022 Natalya Evans Acknowledgements The work described in this thesis would not have been possible without the financial support of the Moffett lab. During my doctoral work, I analyzed data from more than 13 separate cruises, which also required significant financial support. I would like to acknowledge NSF OCE-1636332, DEB-1542240, OCE-1046017, OCE-1029951, OCE-1657958, OCE-2023708, OCE-2124188; Oregon Sea Grant Project number: R/HBT-23-Reimers2022; support from the University of Southern California Wrigley Institute for Environmental Studies; the Schmidt Ocean Institute; and the MIT Ally of Nature Award. Other oceanographic programs and findings were fundamental to my research in this project. I relied on the water mass analysis packages omp2 written by Mattias Tomczak and Johannes Karstensen as well as pyompa written by Avanti Shrikumar with support from Karen Casciotti. I rely on data from the California Cooperative Fisheries Investigations, the 2013 World Ocean Atlas, the 2018 World Ocean Database, a 1,200 year-long sediment core record measured by Caitlin Tems and surely many others, and repeated transects on the 110 °W line spanning 50 years. During my dissertation, I went to sea five times, and my work there could not have been possible without the crew, technicians, and my collaborators on the R/V Falkor, R/V Kilo Moana, and the R/V Oceanus. I would first like to thank Jim Moffett, my advisor, for his guidance, unorthodox research sug- gestions, perspective, and support during the past four years. Without his interest in taking on a student with no oceanography background, I would not be here today. I’d also like to thank my committee for their support. Naomi Levine has consistently provided insight from new perspec- tives on my research, as well as extremely useful information in her modeling class. Seth John’s excitement for ocean sciences has always inspired me, and while Matthew Pennell is new to USC, ii he has grounded my committee. Though he is not on my committee, I’d also like to acknowledge Will Berelson for occasional feedback and challenging yet important questions. My lab members were instrumental in my work. Kenny Bolster taught me how to measure Fe(II), but more importantly, illustrated how to approach oceanographic questions and life as a seagoing researcher. Rin Moriyasu taught me how to measure iodide. The Oregon chapters could not have happened without Alexis Floback and Justin Gaffney, who helped prepare, mobilize, deploy, recover, filter, prepare, and measure during those cruises. During my five years at USC, I was assisted by excellent collaborators. I’d like to thank Al Devol for helping me with my second and third chapters, as well as trusting me to publish his 50 year time series of the ETNP ODZ. His guidance on the nitrogen cycle made these manuscripts possible. AlongwithAl, I’dliketothankWendiRuefforherworkcollatingandquality-controlling the data from these eight cruises. I’d like to thank my collaborators at Oregon State University, mostnotablyClareReimers,PeterChace,NicoleCoffey,andReneBoiteau,fortheirassistancewith our work on that continental margin and inviting me back for the last cruise on the R/V Oceanus. During my water mass analysis, sulfide modeling, and iron measurements, Daniele Bianchi and his lab have provided stimulating discussion and pushed my work to new heights. In the Bianchi lab, I’d specifically like to thank Daniel McCoy, who also helped us at sea on OC2102A. I also worked with excellent undergraduate students, who I was proud to advice. Amanda Taing, Juliana Tichota, and Emma Johnson all performed tasks that no undergraduate had done before in the Moffett lab and did them so well that we offered to hire them down the road. I’d also like to acknowledge Dalton Hardisty, who has been a pleasure to collaborate with the first two weeks of grad school to the last semester. In the Hardisty lab, I’d also like to thank Alexi Schur and Kirsten Fentzke for welcoming me as well as Todd Lydic for his upkeep on the ICP-MS. I’d also like to acknowledge SungHyun Nam for advising me on my first paper as began to dabble iii in water masses. This supervision arose from a single meeting in a Starbucks in La Jolla when I was a first-year grad student, but likely was set the foundation for me to began my path through water mass analysis. In a way, I also have to thank Peter Brewer for desk rejecting this paper as well. His question of advection versus remineralization motivated me to answer that question in the southern California Current System using water masses. I could not have done this research withoutthegraciouscollaborationofStevenBograd,MercedesPozoBuil,MichaelJacox,andIsaac Schroeder, who shared their data, discussed their results, and worked with me to put together this paper during quarantine. I’d also like to thank all of the friends I made on land and at sea while working in this program for their support during long slogs of research and strong swells. Elisa, Diana, Kyla, Jennifer, Naomi, Ali, Hank, Rachel, Xiaopeng, Phil, Emily, Anna, JL, Emmett, Khadijah, Matt, Anna, Pete, Alexi, Kirsten, and many others helped me through hard times. As Matt said, ”a cruise is just asking other people for a hand and vice versa for three weeks”. I’d also like to thank Lucia, Chris, and Karina for camaraderie while trying to do a PhD in a pandemic. The most important acknowledgement here belongs to my fiance Kareesa Kron for countless hoursofcompanionship,emotionalsupport,adventuresacrosscontinents,anddeliciousfood. Their intellectual contributions wee also vital to my dissertation work, as they advised my statistics, reviewed my writing, checked my code, and fixed my math numerous times. My parents also provided unwavering support throughout this journey and my work in academia could not have occurred without their help. I’d also like to acknowledge Kaylee, Bear, and Olivia, the small dogs that always brightened our moods. I also need to acknowledge the community that I’ve joined the past few years. Though there are few of us in academia, this group has become a pillar for support and guidance. I work forward iv to working alongside and celebrating with Avery, Sean, Zara, Anne-Marie, Dani, JL, Jule, Taylor, Emma, Lana, Morgan, Mae, Sim´ on(e), and Krisha for many years to come. v Contents Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxiii Chapter 1: Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Eastern Boundary Upwelling Systems and Oxygen Deficient Zones . . . . . . . . . . 1 1.2 Anaerobic metabolisms and ODZ strength . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Anaerobic processes in hypoxic waters . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Chapter 2: The Role of Water Masses in Shaping the Distribution of Redox Active Com- pounds in the Eastern Tropical North Pacific Oxygen Deficient Zone and Influencing Low Oxygen Concentrations in the Eastern Pacific Ocean . . . . . . . . . . . . . . . . . . . . . 6 2.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.3 Data and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3.1 Data collection and processing . . . . . . . . . . . . . . . . . . . . . . . . . . 10 vi 2.3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.4 Water mass analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4.1 OMPA theory and application . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4.2 Previous water mass definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4.3 Basis set of water types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.4.4 Error analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.5.1 13CW shapes the Oxygen Deficient Zone . . . . . . . . . . . . . . . . . . . . 17 2.5.2 Offset between hydrographic results and secondary nitrite maximum . . . . . 20 2.5.3 Role of mesoscale eddies on 13CW distribution and nitrite accumulation . . . 22 2.5.4 Coupling between the nitrogen and iodine cycles in low oxygen waters . . . . 23 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.7 Other information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.8 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Chapter 3: Prolific nitrite re-oxidation across the Eastern Tropical North Pacific Ocean . . . 33 3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.3 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.3.1 Natural variability in Redfield 106:16:1 C:N:P stoichiometry . . . . . . . . . . 35 3.3.2 The water masses of the ETNP ODZ . . . . . . . . . . . . . . . . . . . . . . . 37 3.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 3.4.1 Parameterizing nitrite re-oxidation using regression . . . . . . . . . . . . . . . 38 3.4.2 Water mass analysis for high-resolution nitrite re-oxidation . . . . . . . . . . 40 3.5 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 vii 3.5.1 Basin-scale quantification of nitrite re-oxidation . . . . . . . . . . . . . . . . . 42 3.5.2 Spatiotemporal variability of nitrite re-oxidation . . . . . . . . . . . . . . . . 47 3.5.3 Future and global implication . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Chapter 4: Rapid expansion of fixed nitrogen deficit in the eastern Pacific Ocean revealed by 50-year time series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 4.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 4.3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.3.1 Sample acquisition and measurement . . . . . . . . . . . . . . . . . . . . . . . 55 4.3.2 Fixed nitrogen loss via integration method . . . . . . . . . . . . . . . . . . . 57 4.3.3 Fixed nitrogen loss via water mass analysis . . . . . . . . . . . . . . . . . . . 58 4.3.4 Data processing for time comparison . . . . . . . . . . . . . . . . . . . . . . . 59 4.3.5 Natural variability estimation via sediment core conversion . . . . . . . . . . 61 4.4 Fifty years of fixed nitrogen loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.5 Observed fixed nitrogen loss likely occurred locally . . . . . . . . . . . . . . . . . . . 66 4.6 Natural variability of the ETNP ODZ . . . . . . . . . . . . . . . . . . . . . . . . . . 68 4.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 4.8 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 4.9 Open Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Chapter 5: The role of seasonal hypoxia and benthic boundary layer exchange on margin- derived iron cycling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 viii 5.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 5.2.1 Sampling description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 5.2.2 Chemical measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.2.3 Computational methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 5.3 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 5.3.1 Cross-shelf variability in iron . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 5.3.2 Benthic release of iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 5.4.1 Controls on iron(II) concentration . . . . . . . . . . . . . . . . . . . . . . . . 88 5.4.2 Off-shelf iron transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 5.6 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 5.7 Open Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Chapter 6: More than deoxygenation: linking iodate reduction to nitrogen, iron, and sulfur chemistry in reducing regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 6.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 6.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 6.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6.3.1 Sample collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 6.3.2 Iodine measurement methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 6.3.3 Water mass analysis and iodate deconvolution. . . . . . . . . . . . . . . . . . 106 6.3.4 Computational methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 6.4 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 6.4.1 Eastern Tropical North Pacific Oxygen Deficient Zone . . . . . . . . . . . . . 108 ix 6.4.1.1 Distribution of redox-active compounds . . . . . . . . . . . . . . . . 108 6.4.1.2 Deconvolution of iodate into source waters and respiration pathways 110 6.4.1.3 Extrapolating an iodate reduction rate from water mass deconvolu- tion results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 6.4.2 Oregon coastal waters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 6.4.2.1 Distribution of redox-active compounds . . . . . . . . . . . . . . . . 114 6.4.2.2 Benthic boundary layer profiles . . . . . . . . . . . . . . . . . . . . . 116 6.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 6.5.1 What factors control iodate depletion? . . . . . . . . . . . . . . . . . . . . . . 118 6.5.2 Robustness of the I/Ca proxy . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drivers of Subsurface Deoxygenation in the Southern California Current System . . . . . 146 1.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 1.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 1.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 1.3.1 Data from the southern California Current System . . . . . . . . . . . . . . . 147 1.3.2 Optimum multiparameter analysis in CCS . . . . . . . . . . . . . . . . . . . . 147 1.3.3 Linking water masses to deoxygenation . . . . . . . . . . . . . . . . . . . . . 148 1.4 Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 1.4.1 Impact of advection versus remineralization on deoxygenation . . . . . . . . . 149 1.4.2 Deconvoluting accumulated remineralization within water masses . . . . . . . 151 1.4.3 Deoxygenation from water mass advection-derived remineralization . . . . . . 151 x 1.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 1.5.1 Water mass analysis for high-resolution nitrite re-oxidation . . . . . . . . . . 152 1.6 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 1.7 Data availability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 xi List of Tables 2.1 Basis set of water types used in these water mass analyses. . . . . . . . . . . . . . . . . . 16 3.1 Ranges for filtering WOA13 and WOD18 data into coherent water masses. Starred variables were only used to filter the data for linear regression of nutrient ratios in Fig. 3, not for geographic distributions in Fig. 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 3.2 RatiosofnutrientsdepictedinFig. 2usinglinearfitsandreactionstoichiometryforcomparison. 45 3.3 Percent of nitrite re-oxidized based on slopes fit in Table 2. . . . . . . . . . . . . . . . . . 46 6.1 Deconvolution results for simulating the iodate distribution in the ETNP ODZ. . . . . . . . 110 6.2 Summary of iodate depletion mechanisms and their alignment with evidence presented in this study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 xii List of Figures 2.1 Predictedoxidation/reductionrangesforrelevantchemicalspeciesinseawateratpH 7.5 and practical salinity of 35. Data from Stumm & Morgan (1995) but presented in the style of Rue et al. (1997). Each redox couple has a range based on variations in activities within seawater. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2 Map detailing the cruise transects on the R/V Falkor (FK180624, orange circles) and the R/V Roger Revelle (RR1804, black diamonds). The FK180624 transect began at (14º N, 102 ºW) and ended at (18º N, 119 ºW), whereas the RR1804 transect began at (18.5º N, 110 ºW) and ended at (16º N, 124 ºW). . . . . . . . . . 11 2.3 Percentages of (a) 13CW, (b) NEPIW, (c) AAIW, and (d) error (χ 2 parameter) for the FK180624. This fit has an error of χ 2 ν =0.078. The percent 13CW water mass figure includes white contours of nitrite concentration at 0.5 and 1.5 µ M, allowing a comparison between the 13CW water mass and the secondary nitrite maximum. Black dots represent sampling locations. . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.4 Same as Figure 4 but for the RR1804. This fit had an error of χ 2 ν =0.131. . . . . . . 21 2.5 Sea surface height anomaly above geode at (a) 30/6/2018, (b) 8/7/2018, and (c) 6/4/2018. The maximum value shown in red is 0.939 m, whereas the minimum value shown in blue color is 0.551 m for all maps. The black dot represents the approximate location of the R/V Falkor during the transect, traveling westward, while the mesoscale features including the eddies propagate westward. . . . . . . . . 22 xiii 2.6 Same as Figure 6 but for the RR1804, where is sampled at (a) 2018/4/6 and (b) 2018/4/10. NotethatRR1804hadseverallegsanditsfullcruisepathisdocumented in Moriyasu et al. (in review). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.7 Percent of 13CW with (a) white contours of 460 and 540 nM iodide and (b) with white contours of 0.5 and 1.5 µ M nitrite and black contours of 100 and 250 nM iodate. (c) The concentration of nitrite in µ M with white contours of 10, 30, and 300 nM iodate, appropriately. Black dots represent sampling locations on FK180624. 25 2.8 The above schematic depicts the extent of 13CW and NEPIW/EqPIW across the Eastern Pacific at isopycnals of 26.2 σ θ -26.5σ θ as well as the currents that transport 13CW based on this work and cited literature. When the 13CW overlaps with low ventilation and high organic matter fluxes, an ODZ forms, until NEPIW/EqPIW upwells and replaces the 13CW. Pockets of NEPIW appear within the ETNP as the mesoscale eddies carrying 13CW wean. . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.1 Depth profiles from the ETNP ODZ at 14 ºN, -110 ºE on FK180624 [1]. The left- most plot depicts raw fluorescence values from the CTD as well as nitrite concentra- tions from bottle samples. The center plot illustrates the distribution of the 13CW, NEPIW,andAAIWbetween26-27kgm − 3 . Therightmostplotprovidestheoxygen concentration measured via SBE43 electrodes, which are not accurate in the oxygen deficient layer but do present where the upper and lower oxyclines are. . . . . . . . . 37 3.2 Maps of (a) 13 ºC Water, (b) Northern/Southern Equatorial Pacific Intermediate Water, and (c) Antarctic Intermediate Water, depicting PO 3– 4 /µ mol kg − 1 . The ma- genta line depicts the concentration past which anaerobic respiration appears to begin, as given in Fig. 3. Water mass formation sources are noted. . . . . . . . . . . 43 xiv 3.3 Evolution of nutrients within the 13CW, NEPIW, and AAIW in the northern hemi- sphere using (a) NO – 3 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOA13, (b) NO – 3 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOD18 for comparison with WOA13, (c) total CO 2 /µ mol kg − 1 vs NO – 3 /µ mol kg − 1 from WOD18, and (d) total CO 2 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOD18. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 3.4 Oxygen concentrations and pertinent water mass analysis results for cruises on the 110 ºW line. Figures in the left column are from 1994, the middle is 2007, and the right is 2016. Each row corresponds to the property specified on the color bar to the right. Oxygen concentrations were plotted using raw data, whereas nitrite reoxidation, aerobic remineralization, and anaerobic remineralization were plotted using the mean of 10 calculations. The uncertainties in nitrite re-oxidation were plotted using the standard error of these 10 calculations. . . . . . . . . . . . . . . . . 49 4.1 Map presenting the transects from the eight cruises analyzed in this study and the location of the Pescadero Basin coring site. These data are presented over a contour map of the fraction of ODZ conditions observed on the 26.5 kg m − 3 isopycnal. . . . 56 xv 4.2 a-c) Integrated –N* values for each cruise, subset between the specified potential density ranges such that (a) is the total ODZ, (b) is the upper oxycline, and (c) depicts both the core ODZ and the deep ODZ, which contains the deeper oxycline. Both2016cruisesarepresented,andSKQ201617Sisthesystematicallylowpointnot connected by a line on each plot, also labeled as “SKQ”. (d) depicts the maximum fixed N lost, measured on each cruise and calculated by eOMP, within the potential densityrangesintheintegrationsin(c). Errorbarscorrespondtostandarddeviation in the maximum 10% of data. The depth of the 13CW is plotted on the right y-axis. (e)depictsthemeanoxygenconcentrationmeasuredbetween100-400monCalCOFI cruises for southern California, and error bars correspond to the seasonal standard deviation. (f)depictsnormalizedparticulateorganicnitrogenisotopesmeasuredand published in Deutsch et al. (2014) for comparison. . . . . . . . . . . . . . . . . . . . 63 4.3 Cross-sectionsoffixednitrogenlossascalculatedusingeOMPintheODZonthe110 ºW line for each year except 1972. We apply a cutoff at 8 µ mol kg − 1 to visualize the spatial coverage of fixed nitrogen loss, rather than this value being a true definition. 64 4.4 NO − 3 -PO 3– 4 plothighlightingNO – 2 concentrationsinthesecondarynitritemaximum. The arrow depicts the stoichiometry of anaerobic remineralization using in water massanalysis,andthepointsarethereferencenutrientconcentrationsforthe13CW and NEPIW. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 xvi 4.5 a) Map depicting the potential density of the ODZ core, as defined as the midpoint between the top and bottom of the ODZ using the atlas developed by Kwiecinski & Babbin (2021). This map matches the blue inset in (b), and (b) also illustrates the boundariesfortheROMSsimulationperformedbyMargolskeeetal. (2019). (c)and (d) reproduce the results of Margolskee et al. (2019), where the sections represent the locations where water enters the ODZ on the northwest boundary (c) and the southern boundary (d). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 4.6 Figure 6. a) δ 15 N-PON Pescadero core data replotted from Tems et al. (2016) with themaximumfixedNlossfromthecoreODZfromtherelevantyearsoverlaidasblue diamonds. b) Points depicting the maximum fixed N loss from the core ODZ and a histogram of δ 15 N-PON Pescadero core data from Tems et al. (2016) converted to maximumfixednitrogenlossbasedonthecomparisonin(a). Thisfigureincludesthe 99%confidenceintervalsofthisdistributionindashedmagentalines. Errorbarshere correspond to error bars in Fig. 2. c) NO – 3 - PO 3– 4 plot of data from the eight cruises, where the magenta box corresponds to the 99% confidence interval determined in (a). 69 5.1 a) Sampling map for OC2107A stations with bathymetry shaded in blue. A solid black line indicates the coastline. b) Same coastline and bathymetry, but the mea- sured sediment mud content (% silt fraction) from the usSEABED database is pre- sented. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 xvii 5.2 Depth profiles of Fe(II), sampled Fe(II) which is corrected for oxidation between sampling and measurement, dFe, and O 2 for selected shelf stations. The upper row illustrates the Fe data, whereas the bottom row depicts O 2 . Station numbers are specified on the top of each Fe profile, and this station number applies to the O 2 profile below it. The gray region on the bottom of each profile indicates the depth of the seafloor, as estimated by the Echosounder while on station. Measured Fe(II) concentrationsweresolowatstation29thatsampledFe(II)couldnotbecalculated. pH measurements were not collected at stations 19.5 and 20 so sampled Fe(II) at these stations used estimated pH values. . . . . . . . . . . . . . . . . . . . . . . . . . 85 5.3 Fe(II), sampled Fe(II), dFe, and O 2 profiles from two benthic lander deployments on the Oregon continental margin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 5.4 a) Scatter plot of O 2 and Fe(II) measured across the continental margin, with the colorbar highlighting the depth from the seafloor. Round points indicate the sample was collected from the water column with a CTD, whereas diamonds indicate that the sample came from a benthic lander (replotted from Fig. 3). b) Scatter plot of the sediment mud content at each station versus the deepest Fe(II) sample collected at that station, with the colorbar highlighting the deepest O 2 concentration. . . . . 89 5.5 Scatter plot of Fe(II) concentrations compared to their half-lives. Sample color re- flects the depth where that sample was collected, whereas error bars consist of the uncertainty in Fe(II) measurements and half-lives, appropriately. . . . . . . . . . . . 91 xviii 5.6 Sections of a) Fe(II) and b) O 2 on the Heceta Bank. These sections consist of stations 31-33, where stations 32 and 33 repeat Umpqua River influenced stations from Severmann et al. (2010). Stations 32-33 have Fe(II) concentrations far higher than the colorbar maximum, as seen in Fig. 2, but the colorbar range was chosen to visualize the lower off-shelf Fe(II) concentrations. The 40 and 47 nM Fe(II) concentrationsannotatedontheplotindicatethemaximumconcentrationsmeasured at stations 32 and 33, appropriately. Labeled contour lines represent the potential density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 6.1 Predicted reduction potentials for terminal electron acceptors in seawater at pH 7.5, salinity of 35 psu, and temperature of 10 ºC. This figure originates from Cutter et al. (2018) but uses data from Stumm and Morgan (1995), and each redox couple is presented as a horizontal range due to variability in the activities of each compound. 101 6.2 a) Map of stations in the ETNP ODZ during FK180624 overlaid on the fraction of ODZ conditions observed on the 26.5 kg m − 3 isopycnal. b) Map of stations on the Oregon coast during OC2107A and OC2111A cruises overlaid on seafloor depth. . . 102 6.3 Distributions of a) oxygen, b) total iodine determined as the sum of iodate and iodide, c) iodate, and d) iodide measured on FK180624. . . . . . . . . . . . . . . . . 109 6.4 a) Map of selected stations during RB1603 where Fe(II) was measured and (b) a section of Fe(II). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 6.5 Distributions of a) iodate, b) residuals of fit for simulated iodate, c) iodate depletion duetothepresenceofthe13CWwatermass,andd)iodatedepletionduetoanaerobic respiration. (a) repeats a subset of the data seen in Fig. 3c but is plotted with potential density instead of depth, because water mass features are coherent across potential density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 xix 6.6 Comparison between measured iodate and simulated iodate calculated using Eq. 5 and the coefficients specified in Table 1. . . . . . . . . . . . . . . . . . . . . . . . . . 112 6.7 Distributionofa)oxygen,b)Fe(II),c)totaldissolvediodine,andd)iodidemeasured on the Heceta Bank. Total dissolved iodine here is measured as the sum of iodate, iodide,anddissolvedorganiciodide. Thegraybackgroundrepresentsthebathymetry of Heceta Bank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 6.8 ProfilescollectedfromabenthicboundarygradientlanderatstationsMT2(top)and MT0(bottom). Iodidesamplesmeasuredwithahangingmercurydropelectrodeare labeled as (HMDE) whereas samples measured with an ICPMS are labeled (ICPMS).117 1.1 Map of the annual climatology of O 2 at 300 m in the North Pacific for the period 1955-2017, with the regions where we defined upper and deeper PSUW (blue box), ENPCW (green box), and PEW (red box). The map illustrates the main surface (white) and subsurface (yellow dotted) currents. O 2 data obtained from WOA18 (Garcia et al., 2019). The inset map highlights the nominal CalCOFI survey grid and bathymetry of the region, obtained from ETOPO2 (National Geophysical Data Center, 2006). Attached on the bottom right is the TS diagram for PSUW (blue), ENPCW (green), and PEW or 13CW/NEPIW (red), modified from Bograd et al. (2019). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 xx 1.2 a) Mean O 2 measured in the CalCOFI study region between 100 and 400 m. Oxy- genation occurs from 1985-1993 at 2.3±0.6 µ M year − 1 and deoxygenation occurs from 1993-2018 at -0.97±0.09 µ M year − 1 . The deoxygenation trend in 2a is recon- structed by both b) mean advective (de)oxygenation from NEPIW with a long-term trend of -0.487±0.001 µ M year − 1 and c) deoxygenation from remineralization calcu- lated in eOMP with a long-term trend of -0.44±0.04 µ M year − 1 . Since eOMP does not attribute remineralization to specific water masses, 2c is further deconvoluted in 2d-f. d) O 2 lost through accumulated internal remineralization in dPSUW with a long-term trend of -0.12±0.03 µ M year − 1 and e) O 2 lost through internal remineral- ization in NEPIW with a long-term trend of -0.20±0.03 µ M year − 1 . f) Deoxygena- tion caused by water mass advection-derived remineralization from NEPIW with a long-term trend of -0.11±0.04 µ M year − 1 . Error bars for 2a consist of the uncer- tainty in the water mass contribution combined with the uncertainty in the NEPIW O 2 watermassdefinition,buttheseerrorbarsaretoosmalltobeseen. Uncertainties were not determined for 2b and values less than 40 µ M were not included in the fit. Error bars are not presented for this 2c-e. Error bars in 2c present the uncertainty in the SVD fit propogated through the water mass contribution. . . . . . . . . . . . 150 xxi 1.3 Summary schematic for the processes and contributions of dPSUW and NEPIW on deoxygenation in the CCS, with the amount of deoxygenation written next to each relevant process. On the left, all three processes are linked to dPSUW but it only impacts deoxygenation through accumulated internal remineralization, hence thesolidarrow. Ontheright,NEPIWactivelycontributestodeoxygenationthrough all three processes. Advection-derived remineralization is highlighted in light blue because the OMP deconvolutes it as remineralization, but it depends on increased NEPIW advection from 1984-2018. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 xxii Abstract Climate change is driving deoxygenation across the world’s oceans. This deoxygenation restruc- tures ecosystem energy transfer, foraging habitats, and the abundance of both critical and toxic compoundsinthewatercolumn. Intheabsenceofoxygen,microbesrespireusingterminalelectron acceptors such as iodate, nitrate, ferric (oxy)hydroxides, and sulfate. These metabolisms produce reduced compounds such as iodide, nitrite, Fe(II), and sulfide, which have different properties and distributions in the water column. The research in this dissertation analyzes the causes and interconnections in the distribution of these reduced compounds. Nitrite, an intermediate in dissimilatory nitrate reduction, indicates that this metabolism is occurring in marine Oxygen Deficient Zones (ODZs). Spatial heterogeneity in nitrite distributions observed during multiple sampling campaigns in the Eastern Tropical North Pacific (ETNP) ODZ could not be explained using oxygen or nutrient concentrations. By deconvoluting the source waters sampled on these cruises, I determined that mesoscale features transporting the 13 ºC Water mass westward led to nitrite accumulation in the ETNP ODZ. With this source water mass framework, I identified the basin-wide stoichiometry of anaerobic respiration in the ETNP ODZ. These results reveal that 50%-70% of the nitrite produced is re-oxidized to nitrate in subsurface waters, and I developed a method that can estimate nitrite re-oxidation using only nutrient and carbon measurements. I also applied this source water mass framework to seven cruises spanning a 50-year time series on the 110 ºW line. With these data, I reveal that the ETNP ODZ became xxiii 30% stronger in 2019 than 1994. More importantly, I also calculated the first confidence interval for the strength of the ETNP ODZ and concluded that anthropogenic climate change has not yet influenced the strength of the ETNP ODZ, but likely will soon. PreviousresearchintheMoffettlabhasinvestigatedthefluxofFe, specificallyFe(II),aswellas iodide from ODZs. In these regions, mesoscale features transport Fe(II) from waters that intersect the continental margin. This phenomenon transports extremely high iodide concentrations and iodide has a significantly higher residence time in the subsurface ocean than Fe(II). These findings suggest that iodide could be used to trace sources of Fe(II) in low oxygen waters. To examine the processes that control the distributions and couplings of these compounds, I studied seasonally hypoxic waters on the Oregon continental margin. The presence of low oxygen water in this region prevents water column denitrification like in the ETNP ODZ, but sediment denitrification still occurs. I explored the factors that control Fe(II) accumulation and transport on the Oregon continental shelf, where Fe(II) concentrations exceed 50 nM. Surprisingly, we observed minimal iodide accumulation in this region, unlike in ODZs. With this comparison and a re-analysis of iodate in the ETNP ODZ, I contrast the nitrogen, Fe, and sulfur cycles in these two reducing margins. These findings suggest that sulfide accumulation likely drives iodate depletion in these low oxygen waters. The findings within this dissertation highlight the physical processes that influence reduced compounds and the interconnections of these elemental cycles. xxiv Chapter 1 Introduction 1.1 Eastern Boundary Upwelling Systems and Oxygen Deficient Zones In subsurface waters near continental margins, complex mixing processes convolute the distribu- tion of different compounds. Water mass mixing, poleward undercurrents, mesoscale eddies, and turbulent jets all create heterogenous structure in the water column. These physical features must be untangled to identify the sources and fates of compounds in the water column. My dissertation focusesspecificallyonEasternBoundaryUpwellingSystems(EBUS),becausetheiruniquehydrog- raphy makes them highly productive ecosystems that provide many important ecosystem services [2]. EBUS are productive due to the mixing of cold, oxygen-rich waters and warm, nutrient-rich water near the continental margin. The California Current System (CCS), which covers near-shore waters from Baja California to Washington, is a particularly well-studied and economically im- portant EBUS. In the CCS, the California Current advects cold, oxygen-rich, and nutrient-poor waters from the Gulf of Alaska, while the California Undercurrent advects warm, oxygen-poor, and nutrient-rich waters from the west coast of Mexico [3, 4]. This undercurrent runs along the continental shelf-break, and topographical features can detach eddies from this undercurrent that transport coastal material into the interior of the Pacific Ocean [5]. The equatorial source waters for the CCS originate from the Eastern Tropical North Pacific Oxygen Deficient Zone (ETNP ODZ) off the west coast of Mexico. The ETNP ODZ is the largest ODZ and the least studied [6]. The subsurface oxygen deficient layer is created by the introduction 1 of low oxygen waters that become anoxic due to remineralization and low mixing with exterior, oxic waters [7]. Ch. 2 identifies the source waters of the ETNP ODZ and uses these source waters to explain the unexpectedly heterogenous distribution of compounds observed on my first cruise aboard the R/V Falkor. These water masses also provide the foundation for the research in Ch. 3-4, because they allow me to quantify the precise amount of remineralization that has occurred in the ETNP ODZ. This water mass framework analysis represents a significant advance over the use of N* because N* quantifies the amount of fixed nitrogen lost versus a global average, whereas my approach references the exact nutrient concentrations where the waters feeding the ETNP ODZ become anoxic. 1.2 Anaerobic metabolisms and ODZ strength Since ODZs are anoxic, their primary metabolisms reduce nitrogen compounds instead of oxygen. Theseregionsareglobalhotspotsforfixednitrogenloss,estimatedat22-26Tgyear − 1 [8]. Thefirst intermediate in denitrification, nitrite, accumulates within ODZs. High rates of nitrite re-oxidation have been measured in ODZs, sometimes higher than the rate of fixed nitrogen loss [9, 10, 11]. Therefore, nitrite acts as a key intermediate that controls if fixed nitrogen is recycled or lost. Ch. 3 develops a method using only commonly measured nutrients and dissolved inorganic carbon to calculate the percent nitrite re-oxidized in that sample. With this method, we provide estimates of nitrite re-oxidation with both high spatial resolution as well as temporal resolution. We observe intermediate depth minima in nitrite re-oxidation, likely due to increase nitrite reduction rates at these depths, and this feature is consistent between 1994, 2007, and 2016. Wecalculatenitritere-oxidationfromnutrientconcentrationsbycomparingtheremineralization stoichiometries against referencevalues. Froelichetal. (1979)derivesthe stoichiometryfororganic matterremineralizationvianitratereductiontodinitrogengaswithRedfieldianorganicmatter[12]. Thepercentofnitritere-oxidizediscalculatedbycomparingtheobservedstoichiometryagainstthis 2 reference stoichiometry. We developed this framework using a new water mass analysis package, pyompa, [13]. This methodology enables us to calculate the nitrite re-oxidation for each measured nutrient sample, providing extremely high spatial resolution. OnekeyquestionforoceanographersishowODZwillchangeduetoclimatechange. Significant deoxygenation has been observed in the Pacific [14, 15, 16, 17], and this deoxygenation will likely causeODZstostrengthenandexpand. Manyscientistspositthatclimatechangehasalreadycaused ODZ intensification; however, we cannot confirm that climate change caused ODZ intensification without knowing the natural variability in ODZ strength. Deutsch et al. (2014) demonstrated that the strength of the ETNP ODZ varies with the depth of the 13 ºC isotherm, which is controlled by tropical trade wind strength. They calculated the ETNP ODZ strength using a sediment core recordofnitrogenisotopes,andfurtherworkonthiscoregenerateda1,200year-longreconstruction [18]. In Ch. 4, we built on their results using a 50-year time series of cruises on the 110 ºW line through the ETNP ODZ. Ouranalysisof50yearsoffixednitrogenlossintheETNPODZilluminatesseveralkeyfindings. WedeterminedthattheETNPODZwas30%strongerin2019than1972. Thestrengthofthecore ODZ is correlated with the depth of the 13CW as well as oxygen concentrations in the southern CCS. In the deep ODZ, we see a dramatic increase in strength from 2016-2019, which likely occurs north of 18 ºN and may be unprecedented. We correlated the fixed nitrogen loss in the water column to the 1,200 year long sediment core record in Tems et al. (2016) and calculated the first confidence interval for the strength of the core ETNP ODZ. 1.3 Anaerobic processes in hypoxic waters The rest of my dissertation contrasts seasonally hypoxic Oregon continental margin waters against the permanently anoxic ETNP ODZ. Comparisons between these regions help to identify what controls various phenomena that occur only under low oxygen. One such phenomenon is the 3 way that ODZs act as sources of Fe to the ocean. ODZs contain deep plumes of Fe, though the mechanism causing these deep plumes is currently uncertain [19]. The ETNP ODZ has a deep plume of Fe, as well as a far shallower plume of Fe(II) in waters that intersect the continental margin [20]. Recent observations of deep Fe plumes in every single ODZ (ETNP: [20]; ETSP: [21]; Arabian Sea: [22]) have renewed interest in the transport of Fe from continental margins. We investigated how off-shelf Fe transport differed between a seasonally hypoxic shelf and these permanently anoxic margins. The Oregon continental shelf is known to have high benthic Fe fluxes [23] due to river inputs [24] and high organic matter content [25]. We observe over 50 nM Fe(II) in bottom waters on the Oregon continental margin, and Ch. 5 provides a thorough analysis of the causes, variability, and fate of this Fe(II). We contrast the iodine distribution in the ETNP ODZ, some of which I measured [26], against the iodine distribution that we measured on the Oregon continental margin in Ch. 6. We observe minimaliodideaccumulation,whichsurprisedusconsideringthe50nMFe(II)thathadaccumulated in the bottom waters, especially when comparing iodide and Fe(II) concentrations in the ETNP ODZ.Manyprocesseshavebeensuggestedtocauseiodatereduction,thoughtheactualmechanisms andimportanceisunknown. UsingourcomparisonbetweentheETNPODZandOregoncontinental waters, we suggest that sulfide accumulation in sediments near the sediment-water interface likely causes iodate depletion. This finding is particularly significant because the incorporation of iodine intocarbonateshasbeenusedtoreconstructbottomwateroxygenconcentrations, andourfindings refine the use of this proxy. This last chapter integrates many of the processes discussed in my previous chapters and con- nects well to this dissertation’s overall theme of deconvoluting the causes of a reducing process and theirinteractionsacrossasuiteofelementalcycles. Inthischapter, Iapplythewatermassanalysis framework developed in Ch. 2-3 deconvolute the controls on iodate depletion, and the background 4 on ETNP ODZ hydrography in Ch. 2 assists in describing the missing factor(s) controlling iodate depletion. Comparing iodine speciation between the ETNP ODZ and Oregon continental margin waters relies on contrasting the cycles of nitrogen, iron, and sulfur between these regions. With the watermassdeconvolutiontechnique, Iusethedifferentpropertiesofeachofthesecyclestoidentify which likely causes iodate depletion, an open and important question in that field. 5 Chapter 2 The Role of Water Masses in Shaping the Distribution of Redox Active Compounds in the Eastern Tropical North Pacific Oxygen Deficient Zone and Influencing Low Oxygen Concentrations in the Eastern Pacific Ocean 2.1 Abstract Oceanic oxygen deficient zones (ODZs) influence global biogeochemical cycles by enabling the accumulation of reactive intermediates not present in the rest of the ocean such as iodide and nitrite. By applying Optimum Multiparameter Analysis to two cruises in the Eastern Tropical North Pacific (ETNP), this study provides a water mass analysis to investigate the extent that water masses govern the ODZ and influence its chemistry. This analysis reveals that within the deoxygenated waters of the ODZ, oxygen deficient conditions are linked to the 13 ºC (13CW) water mass, which is distributed via eddies branching from the California Undercurrent. Nitrite accumulates within these eddies and slightly below the core of the 13CW, and between these maxima in nitrite, complete reduction of iodate is observed. This water mass analysis also reveals that the 13CW and deeper Northern Equatorial Pacific Intermediate Water (NEPIW) act as the two Pacific Equatorial source waters to the California Current System. The 13CW is identified as the same water mass as the Equatorial Subsurface Water (ESSW) and Subtropical Subsurface Water (StSsW), and this study labels this water mass the 13CW based on its source and history. 6 Since the 13CW has been found to dominate under low oxygen concentrations within the Eastern Tropical South Pacific ODZ and the Peru-Chile Undercurrent, the 13CW and the NEPIW shape diminished dissolved oxygen concentrations within the entire eastern Pacific Ocean. 2.2 Introduction ThePacificOceanhousestwoofthelargestOxygenDeficientZones(ODZs), horizontallyextensive regions where oxygen is scarce enough to be anoxic [27, 28]. This deoxygenation originates from reduced exposure to more oxygenated water masses and elevated respiration of sinking organic matter [29], which forms through natural processes [30, 28] as compared to eutrophication driving coastal deoxygenation [31]. These regions have been determined to expand in tropical and sub- tropical regions over the recent decades, and they are predicted to expand further in the future [32, 33]. These minima in dissolved oxygen lead to radically different ecosystems on microbial [34, 35, 36, 37, 38] and macroorganism scales [39, 40], cause approximately 22-26 Tg of yearly fixed nitrogen loss from the ocean [41, 42, 43], release approximately 6 Tg of the potent greenhouse gas nitrous oxide [44], and could act as a source of iron to the iron-limited Southern Ocean [45, 46]. These ODZs occur due to a combination of nutrient-rich upwelling, sluggish mixing, and high biological surface productivity that positions them in the subtropics north and south of the equator[32],calledtheEasternTropicalNorthandSouthPacific(ETNPandETSP),appropriately. Bothregionshaveapolewardundercurrent,knownastheCaliforniaUndercurrent(CUC)andPeru- ChileUndercurrent(PCUC),nearthecontinentalshelf,whichareknowntospawnmesoscaleeddies thatpropagatewestthroughtheirODZs[47,48,49]. Mesoscaleeddiesincreaseproductivityinboth the ETNP and ETSP by transporting nutrients [50], and coherent poleward undercurrent eddies possess elevated respiration rates compared to surrounding waters [51] that lead to increased N 2 O production [52] and fixed nitrogen loss [53]. 7 A water mass analysis deconvolutes an oceanographic transect into source water masses, allow- ingforspecificfeaturesinthetransecttoberelatedtophysicalproperties. Awatermassanalysisof theETSPincludingtheODZwasrecentlypublishedbasedontheUSGEOTRACESGP16Eastern Pacific Zonal Transect [54], and this study provides a similar analysis of the ETNP ODZ. This wa- ter mass analysis focuses on the intermediate water masses that interact with the ODZ within this region. Thelarge-scalecirculationresponsibleforthelocationoftheETNPODZwithinthePacific Ocean, such as low ventilation [55, 56], and drivers for deoxygenation such as elevated respiration compounded on low oxygen solubility [57, 33] have been characterized. Nevertheless, thorough in- vestigations into the water masses responsible for the depth range and fine structure of the oxygen minimum layer have not been carried out. Previous works on the ODZs in the ETNP and ETSP have noted that their dissolved oxygen minima accompany the salinity maximum at isopycnals between 26.0σ θ and 27.0σ θ . That salinity maximum was identified as Equatorial Subsurface Water (ESSW), and this report aims to investigate the fidelity of the ODZ to this water mass, though we argue that the ESSW should be called the 13 degree Celsius Water (13CW) in section 4.1. Itiswell-establishedthatODZsarehydrographicallycharacterizedbyhighspiciness(π )aswell as low oxygen [49, 48]. Spiciness is a conservative variable orthogonal to potential density anomaly that correlates with warmer, saltier, and lower dissolved oxygen concentrations [58, 49]. ODZs containanitritemaximumduetodissimilatorynitratereductioncompetingwithoxygenrespiration [59]. Within an ODZ, dissimilatory nitrate reduction requires extremely low concentrations of dissolved oxygen, often quoted as 10 µ mol kg − 1 . Since the true dissolved oxygen concentrations are below the detection limit for the conventional CTD-integrated SBE 43 oxygen sensors [60], the subsequent nitrite maximum from dissimilatory nitrate reduction, known as the secondary nitrite maximum, is often used to indicate the presence of the ODZ. Nevertheless, the ODZ is observed to extend deeper than the secondary nitrite maximum [61]. Nitrite has been frequently studied in 8 ODZs and further discussion of the secondary nitrite maximum can be seen in additional literature ulloa m icrobial 2 012,buchwald n itrogen 2 015,babbin m ultiple 2 017. In addition to nitrite, iodate and iodide were measured as proxies for anoxia as well as hypoxia. Iodate and iodide are the primary chemical species of inorganic iodine in the ocean, and the concentration of iodate dominates that of iodide within the oxygenated ocean [62]. Within the open ocean, iodide can accumulate via iodate reduction by phytoplankton [63], though shelf inputs also act as an iodide source [64] . Reducing conditions in shelf sediments release iodide into the watercolumn,aswellasotherspeciessuchasferrousiron[65,66]. Thisphenomenonisthefirststep in the shelf-to-basin iron shuttle [67, 68], after which subsurface mesoscale features sweep the iron off the shelf. A subsequent series of reductions and precipitations transport iron from continental shelves into deep waters in the ocean’s interior. Iodide does not precipitate or scavenge easily, enabling it to accumulate in the water column. Once iodide accumulates, its abiotic oxidation to iodate is kinetically hindered without light [62], such that this oxidation is attributed to biological processes [69]. The reduction potential of iodate is less thermodynamically favorable than that of oxygenbutmorethermodynamicallyfavorablethanthatofnitrate(Cutteretal. 2018),asdepicted in Figure 1, which enables these proxies to be effective under both anoxia and hypoxia. This intermediate reduction potential enables this reduced species to act as an indicator of deoxygenated water [46, 63, 70] as well as shelf inputs [64]. Due to the lower rates of iodine redox interconversion [71, 72] compared to rapid nitrite interconversion [73], iodine species could serve as better indicators of the history of dissolved oxygen concentrations within a parcel of water. It has also been theorized that microbes can oxidize nitrite via iodate when oxygen is scarce [73]. Nitrite has been shown to reduce iodate to iodine and iodide abiotically in aqueous regions between ice crystals, demonstrating that this reaction is possible with pre-concentration [74] which microbes could capitalize upon. This paper applies water masses as a framework to analyze their role in 9 Figure 2.1: Predicted oxidation/reduction ranges for relevant chemical species in seawater at pH 7.5 and practical salinity of 35. Data from Stumm & Morgan (1995) but presented in the style of Rue et al. (1997). Each redox couple has a range based on variations in activities within seawater. definingtheoxygenminimumlayer,thecouplingofthesecondarynitritemaximumtohydrographic features, and the relative control of nitrite versus oxygen on iodate in low oxygen conditions. 2.3 Data and methods 2.3.1 Data collection and processing This manuscript focuses on data acquired by the R/V Falkor (FK180624) from June to July 2018, though includes data from the R/V Roger Revelle (RR1804) from March to April 2018 to test if results observed from the R/V Falkor are consistent between cruises and times (Figure 2). These cruise transects will be referred to hereafter by their cruise names. Data from FK180624 and RR1804 are available at Rolling Deck 2 Repository (http://www.rvdata.us/catalog/FK180624) and (https://www.rvdata.us/search/cruise/RR1804). On both cruises, pressure, temperature, salinity, andoxygen datawere acquiredwith aSea-BirdElectronics (SBE)9/11 plusConductivity- Temperature-Depth (CTD), and the data collected during only upcasts were used to improve the quality of the data by limiting hysteresis in oxygen measurements via SBE 43 sensors. Nitrite and phosphate were measured onboard with the standard photometric methods (Babbin et al., 2014; 10 Figure 2.2: Map detailing the cruise transects on the R/V Falkor (FK180624, orange circles) and the R/V Roger Revelle (RR1804, black diamonds). The FK180624 transect began at (14º N, 102 ºW) and ended at (18º N, 119 ºW), whereas the RR1804 transect began at (18.5º N, 110 ºW) and ended at (16º N, 124 ºW). Stricklandetal.,1972),usingahigh-sensitivity10cmquartzcell(Starna)onFK180624,andiodate and iodide were measured via spectrometry and voltammetry, respectively [26]. 2.3.2 Methods The primary result of this paper is a water mass analysis produced via Optimum Multiparameter Analysis(OMPA).ThepackagefortheOMPAwasdevelopedbyJohannesKarstensenandMatthias Tomczak. This package operated in the MATLAB statistical computing language and the authors usedMATLABR2018b[75]. Apreviousversionofthispackagecanbefoundat https://omp.geom ar.de/,thoughthisversionwasnotcompatiblewithMATLABR2018bandanupdatedversionwas distributed by Johannes Karstensen via personal communication. While this package relied on the EOS-80seawaterlibraryratherthanthecurrentstandardofTEOS-10(http://www.teos-10.org/ software.htm) [76], we used it to maintain continuity with previous water mass analyses. Instead, the authors adapted the OMPA package to input hydrographic data from TEOS-10 with IBM ILOG CPLEX Optimization Studio V12.8.0 as optimizer. For all the cruises, in situ temperature 11 andsalinitywereconvertedtoconservativetemperature(θ )andabsolutesalinity(S A )withTEOS- 10, and this information was used to calculate potential density referenced to surface (σ θ ) and spiciness (π ). This modification to TEOS-10 maintains the continuity of the OMPA package while including updated oceanographic standards. Interpolated figures were generated using the contour and contourf functions with linear interpolation in MATLAB and displayed contour lines were selected using scatter plots to ensure that the contours were representative of the distribution. Due to the importance of mesoscale features on lateral transport, sea surface height anomaly (SSHA) data was retrieved from satellite altimetry. These data enabled us to analyze the influence of mesoscale eddies in shaping the water mass and nitrite distributions. The publicly accessible Copernicus Marine Environmental Monitoring Service (http://marine.copernicus.eu/) data product “Global Ocean Gridded L4 Sea Surface Height and Derived Variables NRT” with product id“SEALEVEL GLO PHY L4 nrt OBSERVATIONS 008 046”wereusedtovisualizeSSHAatthe start date (30/6/2018), mid-date (8/7/2018), and end date (14/7/2018) of the FK180624 transect as well as the start date (6/4/2018) and end date (10/4/2018) of the RR1804 transect. 2.4 Water mass analysis 2.4.1 OMPA theory and application OMPA assumes that measured and calculated transect properties such as temperature, salinity, oxygen, phosphate, and/or potential vorticity can be decomposed into a linear combination of source water mass properties via inverse modeling [77, 78, 54]. Each water mass is assigned a water type, which is a theoretical point in n-parameter space defined as the source or centered within the cluster of properties of the observed water mass [79]. During an OMPA, the fraction of each water mass must be non-negative [80]. Though the extent that water mass properties are conservative across a transect varies [81], classical OMPA has been used and described in a number of other oceanographic studies [80, 77, 82, 83, 84, 85]. Nevertheless, chemical parameters 12 such as oxygen, phosphate, and nitrate are often only semi-conservative across large distances due to biological metabolisms. An extended version of OMPA was developed to account for nutrient non-conservative deviations from source water mass properties using the Redfield ratio combined with an internal parameter, the extent of remineralization [83, 80, 54]. Conservativetemperature,absolutesalinity,oxygen,phosphate,andspicinesswereusedforthis OMPAstudyduetotheavailabilityandwell-characterizednatureoftheseparameters. Nitratewas not used because the denitrification in the ODZ renders this parameter non-conservative. While spiciness (π ) has not been implemented in water mass analyses before, it is a conservative state variableorthogonaltopotentialdensityanomaly[58]commonlyusedtodifferentiatewarmer,saltier watersatsimilarisopycnalswithintheCaliforniaCurrentSystem[86,48,49,4]. Highspicinesshas been correlated with lower dissolved oxygen and pH within the higher portion of Pacific Equatorial Water in the California Current System [49], and we include π in this study to aid in resolving low oxygen waters. While this is the first implementation of π in OMPA, other physical parameters, suchasbetweenpotentialdensity(σ θ )[87]andpotentialvorticity[88],havebeenusedorconsidered in differentiating water masses. This OMPA was weighted using the default coefficients in the package, θ , S A , O 2 , PO 3− 4 , π , mass=24, 24, 7, 2, 7, 24, and spiciness was weighted equally to oxygen since it is implemented as a proxy for oxygen in waters including and below the thermocline. In this study, extended OMPA was applied to analyze data from FK180624 and RR1804 cruises independently. This water mass analysis was split between different potential density horizons for separate iterations of OMPA: surface (σ θ < 26.0 kg m − 3 ), intermediate (26.0 kg m − 3 <σ θ < 27.0 kg m − 3 ), and deep (σ θ > 27.0 kg m − 3 ). These cruises lacked sufficient deep casts to provide a rigorous OMPA in this potential density horizon. Water mass distributions are plotted against σ θ rather than depth or pressure due 13 to the inherent coherence observed along the isopycnal surface. The OMPA results from FK180624 and RR1804 are presented in section 4.2. 2.4.2 Previous water mass definitions The basis set for the water masses in this analysis was identified via literature on general hydrog- raphy and circulation in the tropical Pacific [89, 90], off Mexico [91, 92, 93], the ETNP [94], and the ETSP [95, 96, 54]. This literature suggests that the relevant water masses are Tropical Surface Water (TSW), Equatorial Surface Water (ESW), Pacific Subarctic Upper Water (PSUW), 13 ºC Water (13CW), North Equatorial Pacific Intermediate Water (NEPIW), Antarctic Intermediate Water (AAIW), and Upper Circumpolar Deep Water (UCDW). Water masses such as Subtropical Subsurface Water, Pacific Equatorial Water, and Pacific Intermediate Water were deemed to have too general of properties to be relevant for this analysis. A brief discussion of naming conventions for 13CW versus Subtropical Subsurface Water versus Equatorial Subsurface Water is included in section 4.1. Surface waters include TSW, ESW, and PSUW [89, 91, 96, 92]. Since FK180624 and RR1804 focused on the oxygen deficient zone, surface sampling was not emphasized and most of the surface data came from FK180624. The OMPA results also indicate that some intermediate water could be seen bleeding into surface waters. Intermediate waters in these transects include 13CW and NEPIW [90, 96, 54], in addition to the ODZ layer. AAIW contributes slightly to this potential density range and is included to improve the quality of the OMPA fit, though the center of AAIW is within deeper layers. For the purposes of this OMPA, NEPIW could be expressed as a linear combination of 13CW and AAIW, rendering its presence debatable by OMPA [79]. Nevertheless, previous studies have identified a Pacific Intermediate Water [92] and even NEPIW itself (Peters et al. 2018), justifying its inclusion. While ESW also appears to be a linear combination based on 14 the θ − S A , relationship in Figure 3a, it is an endmember in O 2 − PO 3– 4 space, as Figure 3b depicts, and therefore must be a separate water mass. The basis set of water types for this analysis was developed using previously established defi- nitions for each water mass then refined using histograms of water properties from each transect, where relative maxima in a θ -S A diagram indicated the θ and S A water type all water masses ex- ceptPSUWandUCDW.Subsequenthistogramsof θ versusotherparametersidentifiedtheirwater types. Using histograms to identify water masses has been established previously [97]. Since the California Current injects PSUW into the ETNP, it influences the region but does not have a large volume of water associated with it. Therefore, PSUW was defined as the coldest, freshest point on the θ -S A diagram. This assumption implies that the “true” PSUW value cannot be identified from this data, but waters that deviate from the TSW-13CW mixing line with lower temperatures and salinities have mixed with PSUW. Since this PSUW proxy is noticeably cooler and fresher (lower π ) than native ETNP waters [86, 92, 49], this assumption is reasonable. The sources of the water masses in this transect are relatively well-known. TSW is locally produced, whereas ESW originates from a mixture of upwelling and the Equatorial Undercurrent (EUC) [98, 99]. PSUW is transported equatorward to the ETNP from the colder North Pacific by theCaliforniaCurrent[100,49]. 13CWisformedintheTasmanSea[101,102], butitisextensively modified in the South Pacific Subtropical Gyre before the EUC transports it to the Eastern Pacific [103]. WhilethesimulationsinQuetal. (2009)donotinclude13CWinthenorthernhemisphere,it has been identified in this region due to transport by the Northern Subsurface Countercurrent [90]. Pacific Intermediate Water has also been identified in this area [92], and based on its geographical location and physical properties, this water mass was specified as NEPIW. Diapycnal mixing with PacificDeepWater(PDW)andAAIW,withsomeinfluencefromNPIW,formsNEPIW[104]. The AAIW forms in the Antarctic Circumpolar Current and has been well-characterized [105]. 15 Table 2.1: Basis set of water types used in these water mass analyses. Water mass θ /°C S A /g kg − 1 O 2 /µ mol kg − 1 PO 3− 4 /µ mol kg − 1 π TSW 26.95 (FK); 34.06 (FK); 192.7 0.223 4.97 (FK); 26.56 (RR) 33.83 (RR) 4.36 (RR) ESW 23.28 (FK); 34.56 (FK); 212 0.17 3.8 (FK); 21.26 (RR) 34.44 (RR) 3.3 (RR) PSUW proxy 15.93 (FK); 34.13 188.4 0.437 1.79 (FK); 16.71 (RR) 1.91 (RR) 13CW 12.54 (FK); 34.98 (FK); 0.59 (FK); 2.73 1.58 (FK); 13.08 (RR) 34.95 (RR) 0.69 (RR) 1.68 (RR) NEPIW 9.56 34.8 0.64 (FK); 3.1 0.894 0.90 (RR) AAIW 6.03 34.7 4.8 3.48 0.418 UCDW proxy 1.42 34.87 118 2.67 -0.11 2.4.3 Basis set of water types The combination of FK180624 and RR1804 data helped to create a robust basis set of water types [79]forthisOMPA,becausetheRR1804dataallowedtheFK180624basissettobeverified. Values for the basis set are presented in Table 1 and scatter plots of these water types are presented in Figure 3. Conservative temperature and absolute salinity for surface water types varied between cruises,likelyduetothechangesinseasonalthermalstratificationasFK180624andRR1804cruises were conducted in different seasons (spring and summer, respectively), and the FK180624 water types are preferred over the RR1804 types due to better vertical distribution of sampling. The presented values for dissolved oxygen in the 13CW and NEPIW water types are not necessarily the “true” dissolved oxygen for these water masses because the SBE 43 dissolved oxygen sensors cannot report quantitative concentrations within the range of dissolved oxygen concentrations in the ODZ. Nevertheless, these dissolved oxygen concentrations are consistent within each cruise, as depicted in Figure S1. 16 2.4.4 Error analysis The quality of OMPA fits was tested with the chi-squared goodness of fit test. A chi-squared parameter was calculated for each parameter of every data point, then summed to test the contri- bution of each parameter and the goodness of the OMPA fit itself. To account for autocorrelation, the degrees of freedom were reduced using the lag-1 correlation coefficient [106, 107]. Removing autocorrelation caused the large data sets (n∼ 200 to 1000) to be reduced (n∼ 40 to 100), but these degrees of freedom still led the chi-squared test to become overpowered [108, 109, 110], such that the p-values returned either converge beyond p=0.01 or diverge to p=1. In this manner, the fit of an OMPA was regarded as divergent or convergent, where a convergent fit possesses statistical significance. Since the purpose of this error analysis was to determine if the fit was statistically significant, this binary response proved acceptable. Instead of reporting a p-value, a reduced chi- squared statistic (χ 2 ν ) is provided for each fit. Traditionally, χ 2 ν is applied to fitting models to data where χ 2 ν >> 1 indicates that the fit is poor, χ 2 ν = 1 indicates that the uncertainty in the fit is weighted equally to the uncertainty in the observations, and χ 2 ν ¡ 1 indicates overfitting [111]. In this study, χ 2 ν ≥ 1 results in a divergent p-value, and overfitting in this context does not warrant penalization. All fits presented have χ 2 ν < 1 and oxygen had the most error of any parameters, particularly when modeling the mixing of TSW and 13CW, likely due to the large variance in oxygen concentrations between those two water masses. 2.5 Results and discussion 2.5.1 13CW shapes the Oxygen Deficient Zone We label the salinity maximum (S A :34.8− 35.0gkg − 1 ) at 26.2σ θ as 13CW following [101], though this water mass has been assigned other names such as the Equatorial Subsurface Water (ESSW) [89] and the Subtropical Subsurface Water (StSsW) [94]. The water mass possessing this salinity maximumhasbeenhistoricallyreferredtoasESSWintheETSPandsouthernhemisphere[112,95, 17 54], though13CWhasalsobeenused[47]. OffMexicoitisoftenreferredtoasStSsW[91,113,114, 115, 93], though ESSW has also been used [116]. This water mass has the lowest concentrations of dissolved oxygen and maximum spiciness of intermediate waters identified in this study. A previous water mass analysis on the GEOTRACES GP16 transect included the ETSP ODZ and found that the ESSW, with a potential temperature around 13º C and a practical salinity around 34.9, dominated the low oxygen waters of the ODZ [54], similar to the ETNP ODZ. The ESSW, seen in the ETSP, and the low oxygen StSsW, seen in the ETNP, both branch off from the Equatorial Countercurrent [90, 55, 32]. The Northern Subsurface Countercurrent (NSSCC) transports the ESSW to the ETNP from the equator, and the Southern Subsurface Countercurrent (SSSCC) transports it to the ETSP [32]. The 13CW may originate from the Tasman Sea [101], where the corresponding water properties are found during the summer [102], as formed via winter mixing with temperature-salinity properties and carried into the Eastern Pacific via the EUC. Models suggest that the South Pacific gyre extensively modifies the chemistry of the 13CW, and it can take up to 70 years for it to be transported into the Eastern Equatorial Pacific [103] (Qu et al. 2009). Sincethe13CWandESSWarebothtransportedtotheEasternPacificviatheEquatorialCoun- tercurrents/Undercurrentsandsharenearlyequivalenttemperature,salinity,andoxygenproperties, these names are synonymous for the same water mass. Therefore, a single water mass, the 13CW, shapes both the ETNP and ETSP ODZs. We chose to refer to this water mass as the 13CW, since the convention for naming water masses is based on their source and this water mass is not likely formed in the Equatorial Pacific. This paper refers to it as 13CW in an attempt to unify these three redundant location-specific names, since ESSW and StSsW refer to the same water mass [89, 117, 90], though StSsW is often a broader definition that includes the 13CW mixed with other 18 water masses. Unifying these water masses as the 13CW is important because this water mass is associated with diminished dissolved oxygen concentrations in the eastern Pacific Ocean. The 13CW dominates both Pacific ODZs, and deficient oxygen conditions from the 13CW extend into NEPIW in the ETNP. In the ETSP, the 13CW is bounded by the Equatorial Pacific IntermediateWater(EqPIW)below[54], whichisthesamewatermassasNEPIWbutfoundcloser to the equator and in the Southern hemisphere with a different nutrient profile [104]. In the ETSP, the EqPIW retains 20 µ M O 2 [54], whereas this study identified dissolved oxygen concentrations below the limit of quantification for the SBE 43 oxygen sensors [60]. This northerly decrease in oxygen is attributed to increased productivity north of the equator that depletes the oxygen of EqPIW [104]. EqPIW forms from subsurface mixing of AAIW, PDW, and some North Pacific Intermediate Water [104], and this lack of ventilation leads to oxygen depletion, as does the lack of ventilation of 13CW. The characteristic spiciness maximum of 13CW within the thermocline may factor into its role as the most deoxygenated water mass, since increases in spiciness lead to lower oxygen solubility and higher respiration rates as oxygen solubility decreases with increases in both temperature and salinity independently (Lynn and Simpson 1990; Brewer and Peltzer 2016). The salinity maximum of the 13CW reinforces the pycnocline that separates the intermediate layers from surface mixing, andthisprocessmaintainstheredoxgradientnecessarytoreachtheODZconditionsbelow(Figure 8b). The long residence time and lack of ventilation of this water mass within the subtropical gyre or its transport under and parallel to the highly productive EUC could lead to biologically-driven oxygen depletion as well. It is likely a combination of physical and biological factors that leads to the distinctive oxygen minima of 13CW in the ETNP (and ETSP) ODZ . 19 Figure 2.3: Percentages of (a) 13CW, (b) NEPIW, (c) AAIW, and (d) error (χ 2 parameter) for the FK180624. This fit has an error of χ 2 ν =0.078. The percent 13CW water mass figure includes whitecontoursofnitriteconcentrationat0.5and1.5µ M,allowingacomparisonbetweenthe13CW water mass and the secondary nitrite maximum. Black dots represent sampling locations. 2.5.2 Offset between hydrographic results and secondary nitrite maximum The intermediate waters contain the lowest oxygen concentrations in this region, and their distri- butions can be seen in Figures 4 and 5, for FK180624 and RR1804, respectively. The ODZ itself is defined as the space where oxygen concentrations are so low that nitrate competes and replaces it as the major oxidant. This shift in metabolism is associated with the secondary nitrite maximum, where nitrite accumulates, despite being an intermediate in denitrification. The secondary nitrite maximum is often used as a proxy for the oxygen minimum layer, because the accumulation of nitrite indicates denitrification is occurring. This water mass analysis enables the transect to be deconvoluted into specific water masses, and we were interested in determining the relationship be- tween the oxygen minimum layer and the 13CW. This relationship was analyzed by superimposing nitrite concentrations over the distribution of the 13CW. 20 Figure 2.4: Same as Figure 4 but for the RR1804. This fit had an error of χ 2 ν =0.131. Elevated nitrite concentrations align well with greater than 70% 13CW, 110 ºW (Figures 4 and 5). East of 110 ºW, this nitrite feature extends below the isopycnal of 26.5σ θ and into NEPIW (Figure 4). This expansion of nitrite into deeper waters is attributed to increased productivity and particle fluxes closer to shore. The NEPIW water mass appears to interweave with the 13CW and nitritedoesnotaccumulatewithintheNEPIW,thoughithasextremelylowoxygenconcentrations. The secondary nitrite maximum only fades where NEPIW uplifts into isopycnals of 26.2σ θ -26.4σ θ , except at 106 ºW (Figure 4). When FK180624 was sampling, 106 ºW is near the core of an anticyclonic eddy (Figure 6), and strong downwelling of oxygenated waters (Figure S5) interferes with the accumulation of nitrite. Nitrite accumulation also ends near the core of an anticyclonic eddy in Figures S4 and S5a. Figure5cdepictsAAIWincreasingwestof125ºWaround26.05σ θ ,buttheco-locatedmaximum oftheχ 2 parameterinFigure5dindicatesthattheOMPAhadsignificanterrorinfittingthisregion. AAIWisnotexpectedatthisshallowisopycnal, soitislikelythewatermassobservedinthispoint 21 Figure 2.5: Sea surface height anomaly above geode at (a) 30/6/2018, (b) 8/7/2018, and (c) 6/4/2018. The maximum value shown in red is 0.939 m, whereas the minimum value shown in blue color is 0.551 m for all maps. The black dot represents the approximate location of the R/V Falkor during the transect, traveling westward, while the mesoscale features including the eddies propagate westward. is PSUW-proxy. PSUW is far fresher than 13CW and NEPIW, but PSUW-proxy was not included asawatertypewithintheintermediateOMPA.ThefreshestwatermassintheintermediateOMPA is AAIW, hence why it was selected, but Figure 5d indicates that this attribution is incorrect. 2.5.3 Role of mesoscale eddies on 13CW distribution and nitrite accumulation Thesecondarynitritemaximumcenteredatisopycnalsof26.3σ θ -26.4σ θ doesnotalignwithawater mass; rather, theCaliforniaUndercurrent(CUC)hasbeenidentifiedatthispotentialdensityrange [48, 49]. Mesoscale features created by the CUC maintain this isopycnal while propagating west. Sea surface height anomaly is a common tool for identifying mesoscale dynamics [118, 119, 120], and can even identify subsurface eddy dynamics [121, 122]. Analysis of satellite altimetry reveals that minima in sea surface height anomaly accompany a shift from 13CW to NEPIW at isopycnals of 26.2σ θ -26.4σ θ , observed specifically at 117 ºW and 114 ºW in FK180624 and 120 ºW and 114 ºW in RR1804. 22 Figure 2.6: Same as Figure 6 but for the RR1804, where is sampled at (a) 2018/4/6 and (b) 2018/4/10. Note that RR1804 had several legs and its full cruise path is documented in Moriyasu et al. (in review). Subsurface mesoscale eddies are responsible for the maxima in sea surface height anomaly observed in Figures 6 and 7. These mesoscale eddies transport 13CW westward, extending the range of the ODZ and the area where nitrite accumulates. Subsurface mesoscale eddies caused by polewardundercurrentspossesselevatedrespirationratescomparedtosurroundingwaters(Frenger et al. 2018) that lead to increased N 2 O production [52] and fixed nitrogen loss [53]. The shift in the core of the 13CW to deeper isopycnals in the western region of these transects, particularly noticeableinFigure5,isduetothepropagationofmesoscaleeddiesatthe26.3σ θ -26.4σ θ isopycnals of the CUC, which are deeper than the 26.1σ θ -26.3σ θ isopycnals of the core of the 13CW. Figure 8b reveals the 13CW dominates up to isopycnal 25.8σ θ as well. The mesoscale eddies observed on FK180624 appear to be transient (Figure S5), but coherent subsurface eddies can be formed via poleward undercurrents (Frenger et al. 2018). 2.5.4 Coupling between the nitrogen and iodine cycles in low oxygen waters In addition to nitrite, iodine redox chemistry can also be an indicator for deoxygenated water (Figure 1). While the appearance of iodide would be expected as an indicator for when oxygen is absent, lateral advection of iodide from reducing shelf sediments complicates this analysis [26, 46, 81]. The distribution of iodide within the oxygen minimum layer is depicted in Figure 8a. A station from a previous cruise demonstrates that the 13CW shoals onto the continental shelf (Section S6), so this water mass contains the highest iodide concentrations from this additional 23 sourceanditsinherentlowoxygenconcentration. Sinceiodideoxidationisnotarapidprocess[62], its distribution is more muddled by mixing processes than nitrite is (Figure 8a versus Figure 4a). The lack of iodate is a better indicator for low oxygen conditions than an increase of iodide because iodide has an additional source. The reduction potential of iodate is slightly below the oxygen-water reduction potential [70] and the abiotic oxidation of iodide is slow [62], so the lack of iodate indicates the history of oxygen concentrations within a parcel of water. Figure 8b displays a general alignment of these proxies for low oxygen concentrations within the 13CW as sampled in FK180624, which exists in shallower waters with intermediate iodate concentrations and zero nitrite. These results demonstrate that the 13CW houses the redox gradient leading to the oxygen minimum layer, as seen in the decrease of iodate concentration with depth. This trend in iodate persists until NEPIW appears, and the differing concentrations of iodate in these water masses exemplifies the increased oxygen in NEPIW compared to 13CW. Broadly, the secondary nitrite maximum overlaps with minima in iodate, but regions with iodate minima, seen in Figure 8c, appear to avoid maxima in nitrite. The accumulation of nitrite occurs below the iodate gradient within the 13CW, which could suggest that nitrite accumulation may be hindered by the presence of iodate. Within the ETNP ODZ, there are water parcels with zero iodate [26], unlike in the ETSP ODZ [46]. Figure 8c demonstrates that these parcels exist around and between the secondary nitrite maxima in the ETNP ODZ, since these nitrite maxima are shaped by subsurface mesoscale eddies. In the ETSP ODZ, the 13CW and corresponding secondary nitrite maximum is a coherent region, as seen in Section S5. Though there appears to be a discontinuity at 84 ºW, this discontinuity is an artifact caused by interpolating across the 26.0σ θ -26.8σ θ isopycnals without any samples. The ETSPdoesnotrequiresubsurfacemesoscalefeaturesfornitriteaccumulation, whereastheperiodic accumulation of nitrite and the 13CW in the ETNP reveals that subsurface mesoscale features are 24 Figure 2.7: Percent of 13CW with (a) white contours of 460 and 540 nM iodide and (b) with white contours of 0.5 and 1.5 µ M nitrite and black contours of 100 and 250 nM iodate. (c) The concentration of nitrite in µ M with white contours of 10, 30, and 300 nM iodate, appropriately. Black dots represent sampling locations on FK180624. 25 necessary for the secondary nitrite maximum (Figure S5). Complete iodate reduction also appears to be a property of the ETNP, possibly accompanying eddy-driven nitrite accumulation. We have two hypotheses for the complete iodate reduction in the ETNP compared to the ETSP. The first hypothesis is that stronger upwelling in the ETSP than the ETNP [123] supplies additional iodate from EqPIW to the ETSP ODZ than the iodate supply of NEPIW to the ETNP ODZ. The second hypothesis invokes an accelerated iodine cycling within the microbial activity responsible for nitrite accumulation that generates essentially a steady-state background iodate concentration. Without the increased reaction rates within the eddies [51] where both iodate and nitriteare present, microbialprocesses lessensuchthatthealreadylowiodateconcentrations reach zero. These non-eddy ODZ regions only appear in the ETNP since the ETSP secondary nitrite maximum is a coherent region (Figure S6b). Nevertheless, the mechanisms that lead to complete iodate reduction outside of regions of nitrite accumulation are difficult to determine without a detailed understanding of microbial iodine cycling. 2.6 Conclusion The overlap between a single water mass, the 13 ºC Water (13CW), and physiochemical forcings that drive deoxygenation creates the environment required for denitrification within the secondary nitritemaximumintheEasternTropicalNorthPacific(ETNP)OxygenDeficientZone(ODZ).The 13CW is centered at approximately 150 dbars and isopycnals of 26.2σ θ - 26.3σ θ and it is thickest on the eastern side of the transects. SBE 43 oxygen sensors report the lowest dissolved oxygen concentrations in this water mass for both FK180624 and RR1804, ranging from 0.6 to 0.7 µ M. Whiletheseconcentrationsarebelowthelinearresponserangeforthesesensors[60], themaximum spiciness of 13CW suggest that it has the minimum oxygen of any water mass. NEPIW also has an extremely low oxygen concentration, but not as low as the 13CW. This conclusion can be inferred by the distribution of nitrite versus the distribution of 13CW (Figures 4a and 5a), because nitrite 26 does not accumulate within the NEPIW. Despite the 13CW having the lowest oxygen, the region where denitrification occurs does not align with its core. Thepercentof13CWdoesnotdecreasemonotonicallyacrossthesetransectsbecausesubsurface mesoscale eddies intersperse Northern Equatorial Pacific Intermediate Water (NEPIW) into the isopycnals of 13CW. At the western edge of the transect, it appears that NEPIW completely eclipses 13CW, but there are insufficient data to support this claim definitively. Within latitudinal and longitudinal ranges of the ODZ as controlled by the physiochemical drivers of deoxygenation, the fine structure of the ETNP ODZ is shaped by subsurface mesoscale eddies and their influence on the ratio of 13CW to NEPIW. When the ETSP water mass analysis [54] is repeated for 26.0σ θ -27.5σ θ , as discussed in the supplemental information, the 13CW and the secondary nitrite maximum are coupled as seen in the ETNP. EqPIW appears to replace the 13CW past 100 ºW, and the secondary nitrite maxi- mum disappears along with the 13CW. In the ETSP data presented, the signature of mesoscale eddies cannot be seen from satellite altimetry (Figure S6e), though previous studies have identified infrequent subsurface eddies from the Peru-Chile Undercurrent [124]. In the ETSP, subsurface mesoscale eddies are infrequent and nitrite accumulates independently of them, but in the ETNP, nitrite only accumulates within subsurface mesoscale eddies, suggesting that the elevated respira- tion rate of these features may be required for nitrite accumulation in the ETNP. Nevertheless, the presence of the 13CW versus NEPIW/EqPIW shapes the secondary nitrite maximum in both PacificODZsandinfluencesseasonalhypoxiaoffSouthAmerica[112,95]. AseparateOMPAinthe California Current System using locally defined water types has extracted upper and lower Pacific EquatorialWaters(PEW)atisopycnalsof26.2σ θ -26.8σ θ [125]responsiblefordeoxygenationwithin theSouthernCaliforniaBight[126,86,49]. Thesedefinitionsfit13CWandNEPIWalmostexactly and the entrainment of 13CW within mesoscale eddies in the ETNP suggests that this water mass 27 is present in the California Undercurrent. These results suggest that the 13CW water mass shapes diminished oxygen concentrations across the entire Eastern Pacific Ocean from subtropical gyres poleward to Alaska and Chile. Figure 9 depicts the known distribution of 13CW, the currents that transport it, and role of NEPIW/EqPIW in defining the boundaries of the 13CW and ODZs. Climatechangedrivespresentandfutureoceandeoxygenationandthesubsequentexpansionof ODZs [30, 32, 127, 128, 15, 28, 129]. Current global biogeochemical models struggle with modeling oxygen due to the complex set of factors that control its distribution (Fu et al. 2018). Since the 13CW and NEPIW are the low oxygen waters in the eastern Pacific, other tracers for these water masses could be implemented to simulate oxygen concentrations and the advection of low oxygen waters. Iodine speciation clearly differentiates the 13CW and the NEPIW in the ETNP ODZ. Within the13CW,iodaterangesfrom0-200nM,comparedtoits200-400nMconcentrationrangewithinthe NEPIW. Iodide reaches 400-850 nM concentrations within the 13CW since this water mass shoals onto the shelf. The shoaling of the 13CW onto the continental shelf (Section S6) suggests that the 13CW may act as a conduit in the shelf-to-basin shuttle for reduced species from shelf sediments to the deep interior of the ETNP. Within this 13CW conduit, mesoscale features resulting from the CUC drive parcels of water enriched by sediment porewaters. Compared to the iodide maxima around 13CW, iodide is completely oxidized within NEPIW. This difference in iodine speciation supports the higher dissolved oxygen of NEPIW compared to 13CW . The reduction of iodate within the 13CW indicates that this water mass houses the redox gradient between isopycnals of 25.8σ θ -26.2σ θ required for ODZ conditions between isopycnals of 26.2σ θ -26.6σ θ . The presence of iodate within this redox gradient may set a shallow depth horizon for nitrite accumulation. This distribution of iodate, iodide, denitrification, and oxygen between these two waters in- fluences the conditions of the California Current System, since the CUC transports both water 28 Figure 2.8: The above schematic depicts the extent of 13CW and NEPIW/EqPIW across the Eastern Pacific at isopycnals of 26.2 σ θ -26.5σ θ as well as the currents that transport 13CW based on this work and cited literature. When the 13CW overlaps with low ventilation and high organic matter fluxes, an ODZ forms, until NEPIW/EqPIW upwells and replaces the 13CW. Pockets of NEPIW appear within the ETNP as the mesoscale eddies carrying 13CW wean. 29 masses with a summer shift towards 13CW [49]. A water mass-based distribution may be seen in other redox active compounds besides iodine. For example, 13CW appears to possess fourfold higher concentrations of riboflavin than NEPIW [116]. Identifying these water masses as sources of PEW enables a better understanding of the nutrient fluxes within the California Current System. It is interesting to note that within the ETNP ODZ, iodate is depleted, whereas this phenomenon does not occur in the ETSP or Arabian Sea ODZ [46, 63], though the Arabian Sea does not share water masses with the other ODZs [130, 80]. Iodate depletion occurs outside of eddy-induced sec- ondary nitrite maxima (Figure 8c), and future research on iodate reaction mechanisms is required to understand this observation fully. Within the California Current System, the 13CW is seasonally transported and associated with lower dissolved oxygen and pH than the NEPIW, which are both locally referred to as PEW [49]. While iodine was not measured in Nam et al. (2015), the seasonal advection of 13CW is likely correlated with a shift from iodate to iodide-rich waters which could be used as a tracer for the relativecontributionsofthesetwowatermasses. Silvaetal. (2009)suggestedthatintegratednitrate and nitrite deficit, calculated via N* [41], could be used as a tracer for the 13CW, but the spatial offset of nitrite accumulation with the 13CW core in both the ETNP and ETSP indicates that dissolved inorganic iodine species may be more accurate. Focusing on the 13CW signature would improve attempts at modeling deoxygenation in the Pacific Ocean due to the strong correlation of this water mass with low dissolved oxygen and high spiciness. Rather than using fundamental models to predict ODZ expansion in the future, simulating the accumulation of 13CW will likely result in more precise estimates. Estimating the fluxes and shoaling of the 13CW could also be implemented to anticipate advection-driven coastal deoxygenation. Identifying the role of the 13CW, NEPIW, and subsurface mesoscale eddies on the ETNP ODZ allows some climate change predictions to be made. The increased temperatures of climate 30 change have been associated with increased surface water stratification, deepening isopycnals, and decreasing dissolved oxygen concentrations [131]. These conditions would lead to a stretching of the redoxcline into the 13CW from the surface as the 13CW lowers in depth but upper waters lower in dissolved oxygen, shaping the iodine/iodate distribution. A more interesting consequence is the elevation of the CUC isopycnal with temperature [48]. The extent of this undercurrent’s variability with temperature has not been characterized, but likely leads to the seasonal advection of 13CW versus NEPIW. Eddies from the CUC lead to nitrite accumulation in the ETNP ODZ, and increased temperatures could better align the 13CW and the secondary nitrite maximum. The amount of 13CW in the ETNP and ETSP may also increase, as the EUC accelerates [132], though the net accumulation can only occur if the Tsuchiya jets and poleward undercurrents remain at a constant or lower velocity. 2.7 Other information A Schmidt Ocean Institute grant to Karen Casciotti and A.R.B. supported the FK180624 cruise. Additional support provided by the MIT Ally of Nature and Heflinger Funds to A.R.B. NSF OCE- 1636332supportedR.M.ontheRR1804cruiseandtheUniversityofSouthernCaliforniasupported N.E. and R. M. on the FK180624 cruise. The authors declare no conflicts of interest. 2.8 Acknowledgements We would like to express our appreciation for the various assistance received when working on this project. The crew of the FK180624 enabled the collection of the high resolution data required for the precise water mass analysis. Onboard FK180624, Dalton Hardisty assisted in measuring iodide, while onboard the RR1804, Al Devol measured nitrite. The publicly available OMPA script written by Johannes Karstensen and Matthias Tomczak at https://omp.geomar.de/ enabled N.E. to implement this water mass analysis, in addition to feedback from Brian Peters, the author 31 of the 2018 Eastern Pacific Zonal Transect water mass analysis. Peters also provided his basis set of water types from the ETSP for comparison. Daniele Bianchi provided useful clarification on the classification of mesoscale features observed in these transects. Seth John and Julien Emile-Geay provided MATLAB scripts for interpolation and lag-1 correction for autocorrelation, respectively. Kareesa Kron provided useful criticism and suggestions on statistical analysis for this manuscript. This study has been conducted using E.U. Copernicus Marine Service Information. Data from FK180624 and RR1804 are available at Rolling Deck 2 Repository (http://www.rvdata.us/) 32 Chapter 3 Prolific nitrite re-oxidation across the Eastern Tropical North Pacific Ocean 3.1 Abstract Fixed nitrogen limits primary productivity in large expanses of the ocean, and marine Oxygen Deficient Zones (ODZs) are hotspots of fixed nitrogen loss. This fixed nitrogen loss occurs pri- marily through denitrification, where the stepwise reduction of nitrate to nitrite and ultimately to dinitrogen gas is coupled to organic matter oxidation. Nitrite, the first intermediate in den- itrification, can also be re-oxidized back to nitrate in a reaction by chemoautotrophic microbes. Nitrite’spartitioningbetweenreductionandoxidationdeterminesifmarinefixednitrogenislostor recycled. Nitrite oxidation in anoxic waters has been previously studied through stable and tracer isotope experiments, but the scarcity of these measurements has limited their geographical distri- bution and therefore requires extrapolation to understand the impact on nitrogen cycling. Using basin-scale data, we analyzed the stoichiometry of nutrient regeneration within the three water masses that feed the Eastern Tropical North Pacific Oxygen Deficient Zone. Significant deviations from the expected stoichiometry for denitrification demonstrate that 79% of the nitrite produced in the Oxygen Deficient Zone between the 26.2-26.4 kg m − 3 isopycnal is re-oxidized, whereas only 54% of the nitrite produced between the 26.7-26.9 kg m − 3 isopycnal is re-oxidized. We applied a water mass analysis framework to repeat cruises on the 110 ºW line, revealing high spatiotemporal variability in nitrite re-oxidation. These results reveal significant fixed nitrogen recycling across the Eastern Tropical North Pacific. 33 3.2 Introduction Denitrification consists of a series of reduction reactions that convert fixed nitrogen, which is gen- erally available for biological reactions, to dinitrogen or other biologically unavailable gases. These reactions occur within the oxygen deficient layers of the water column within marine Oxygen De- ficient Zones (ODZs) as well as within marine sediments. Denitrification is the primary process removing fixed nitrogen in the marine fixed nitrogen budget [133] and therefore an important regulatory control on marine productivity. During denitrification, organic matter is oxidized to carbon dioxide and inorganic phosphate is released from organic phosphorus compounds, causing a stoichiometric decrease in nitrate and increases in total carbon as well as phosphate. While the inorganic C:N:P stoichiometry for aerobic respiration of organic matter is 106:16:1 and commonly referred to as Redfield stoichiometry, similar Redfield relationships exist for other metabolic path- ways. The inorganic C:N:P stoichiometry during denitrification from nitrate to dinitrogen gas is 106:-94.4:1 [134, 135], and estimates of fixed nitrogen loss commonly rely on comparing nitrate and phosphate concentrations, such as N* [41, 136]. Recent isotopic tracer and stable isotope studies have reported measured rates of nitrite ox- idation and, in some studies, estimated the relative partitioning of nitrite to reduction versus re-oxidation. In the Costa Rica Dome, the southern side of the Eastern Tropical North Pacific (ETNP) ODZ, 50% of the nitrite was re-oxidized [10]. Data from near the center of not only the ETNP ODZ [137] but also the Eastern Tropical South Pacific (ETSP) ODZ [138] reveal con- sistently high rates of nitrite oxidation throughout the ODZ. These high rates include both the oxygen deficient layer as well as the upper oxycline [11, 139]. Nevertheless, due to the scarcity of these measurements, estimates of the impact of nitrite oxidation across the ETNP ODZ requires significant extrapolation. Here we estimate the magnitude of nitrite re-oxidation in the Eastern Tropical North Pacific by analyzing deviations from Redfield nutrient stoichiometry, specifically, 34 the-94.4:1nitrate:phosphatestoichiometry. Watermassesserveasausefulframeworkforanalyzing these deviations in nutrient stoichiometries because they reveal how nutrients have changed versus their source values. 3.3 Background 3.3.1 Natural variability in Redfield 106:16:1 C:N:P stoichiometry The Redfield ratio, 106(CH 2 O):16(NH 3 ):PO 4 , originated 86 years ago and it is now a fundamental ratio in marine geochemistry. In 1934, Redfield first published the N:P ratio of average plankton as about 20:1 [140]; later, carbon was added, and the ratio was refined to 106:16:1 [141]. After moreresearch,theseminalpaper“TheinfluenceoforganismsonthecompositionofSea-water”was published and the regeneration ratio of 106:16:1 C:N:P was established for marine aerobic systems [12]. Richards added equations for the oxidation of Redfield organic matter by denitrification and sulfatereductioninhispaperinChemicalOceanographyVol. 1anddemonstratedthattheseequa- tions also held for sulfate reduction [134]. Investigations into nutrient profiles in marine sediments confirmed that Richards’ equations were consistent for Redfield organic matter oxidation through nitrate, sulfate, iron oxide, and manganese oxide oxidation [135]. In our analysis, we interpret deviations from the 106:-94.4:1 C:N:P stoichiometry of nitrate reduction in Froelich et al. (1979) to estimate the magnitude of nitrite re-oxidation occurring within the water masses of the Eastern Tropical North Pacific. Natural variations in this 106:-94.4:1 C:N:P stoichiometry, specifically the -94.4:1 ratio, will artificially influence our calculations. Fortunately, Peng and Broecker (1987) looked at C:P ratios in marine detritus and found them consistent with previous estimates. Anderson and Sarmiento (1994)didanextensiveanalysisofC:N:PratiosofGEOSECSnutrientandoxygenprofilesfromthe Atlantic, Pacific and Indian Oceans and found that the regenerated nutrients closely followed the Redfieldorganicmatteroxidationequations. Withregardstothe-94.4:1ratio, thisspecificnumber 35 derives from the charge- and mass-balanced equation for Redfield, Ketchum, and Richards-type organic matter [12]. The only way to change this ratio is to alter the organic matter composition. Previous studies have used a slightly more reduced organic matter and invoked a -104:1 ratio [142]. Should we apply their ratio, our data would be shifted by approximately 11%. The Redfield ratio does not imply that the C:N:P in marine particulate matter fractions and plankton is constant [143], only that the mean ratio of nutrients regenerated throughout the oceans is 106:16:1. Should nitrite re-oxidation occur, we would expect to observe positive deviations in the -94.4:1 N:P value in 106:-94.4:1 C:N:P oxidation stoichiometry [135]. Nitrite re-oxidation impacts both the C:N and N:P stoichiometries. Nitrite oxidation occurs through dissimilatory metabolisms and therefore at far higher rates than biological phosphate assimilation, such that nitrite oxidation has a minimal impact on changes in C:P ratio. However, there are other potential explanations for positive deviations in the -94.4:1 ratio. If the proportion of carbon relative to nitrogen in organicmatterbeingoxidizedislower[144], itwouldslightlydecreasetheC:Nstoichiometry. Also, carbonfixationthroughbothphoto-andchemo-autotrophyaswellascarbonatedissolutionoccurs, which lowers or elevates the C:N and C:P stoichiometries. Only nitrite re-oxidation impacts the N:P stoichiometry of nutrient regeneration without significantly changing the C:P stoichiometry. This behavior suggests that the C:P stoichiometry identifies the magnitude of the second and third processes on these water masses. Therefore, positive deviations in the N:P stoichiometries relative to the C:P stoichiometries indicate the fraction of nitrite that is re-oxidized rather than further reduced. These positive deviations do not specify what mechanism is responsible for the re-oxidation, though. With regards to fixed nitrogen loss, the causal mechanisms diverge at the appearance of nitrite due to the possibility of anammox. Our analysis focuses on the net processes influencing fixed nitrogen loss, and we cannot differentiate what process removes nitrite. 36 3.3.2 The water masses of the ETNP ODZ ThreewatermassescomposeboththePacificODZs, withslightdifferencesbetweentheETNPand ETSP [1]. In the Eastern Tropical North Pacific, these water masses (Fig. 1) are the 13 ºC Water (13CW), Northern Equatorial Pacific Intermediate Water (NEPIW), and Antarctic Intermediate Water (AAIW). While these source water masses are coherent between Pacific ODZs, there are slight differences in the water masses between the ETNP and the ETSP ODZ. For example, the EquatorialPacificIntermediateWaterhasbothnorthernandsouthernarchetypes,asitisformedby the subsurface mixing of primarily Pacific Deep Water (PDW) and AAIW with some NPIW [104]. Since Antarctic Intermediate Water has far higher dissolved oxygen concentrations than Pacific Deep Water, the Southern Equatorial Pacific Intermediate Water remains oxygenated. Therefore, the ETSP ODZ is shallower than the ETNP ODZ [145]. Figure 3.1: Depth profiles from the ETNP ODZ at 14 ºN, -110 ºE on FK180624 [1]. The leftmost plot depicts raw fluorescence values from the CTD as well as nitrite concentrations from bottle samples. The center plot illustrates the distribution of the 13CW, NEPIW, and AAIW between 26-27 kg m − 3 . The rightmost plot provides the oxygen concentration measured via SBE43 electrodes, whicharenotaccurateintheoxygendeficientlayerbutdopresentwheretheupperand lower oxyclines are. 37 Previous analysis of the water masses of the ETNP ODZ in Evans et al. (2020) focused on the northernpartoftheETNPODZfrom14-18ºN,andarepresentativedepthprofile(Fig. 1)from14 ºN,110ºWdepictstheoverlapbetweenthesewatermassesandnotablefeaturessuchaschlorophyll maxima, the secondary nitrite maximum, and the upper and lower oxyclines. In this region, the 13CW is truly anoxic and its core sets the upper oxycline depth. The secondary chlorophyll maximum co-occurs near the 13CW core, causing the formation of hypoxic microenvironments. Nitrite accumulates slightly deeper than the 13CW core due to these sources of oxygen. The presence of the NEPIW interferes with nitrite accumulation in the secondary nitrite max- imum, likely because it is barely oxic at this location. A trace oxygen profile at 19 ºN, 108 ºW reveals that oxygen content is about 10 nM oxygen at similar depths (Larsen et al. 2016). The next deepest water mass, the Antarctic Intermediate Water, acts as the lower boundary for the oxygen deficient layer. We note that these depth distributions are true for this northern region of the ETNP ODZ, but closer to the Costa Rica Dome, previous studies have found nitrite accumu- lation occurs notably deeper [146]. These differences in the nitrite accumulation depth likely occur due to the oxygen concentration in the 13CW and the NEPIW, which varies geographically due to their different entry locations into the shadow region of the ODZ [7, 147]. Since we estimate nitrite re-oxidationusingnutrientconcentrations, wecannotidentifyifnitritere-oxidationoccurredatthe sampling location or was advected from elsewhere, nevertheless, our results present general trends in nitrite re-oxidation across the 110 ºW line. 3.4 Methods 3.4.1 Parameterizing nitrite re-oxidation using regression We analyzed Pacific-wide data products spanning large spatial and temporal ranges. Gridded data from World Ocean Atlas 2013 (Garcia et al. 2013) were acquired via the WOA13 1.00deg 1955- 2012 Annual file available from the Ocean Data View web portal (https://odv.awi.de/dat 38 a/ocean/world-ocean-atlas-2013/). Available temperature, salinity, oxygen, phosphate, and nitrate, and dissolved inorganic carbon were extracted from the World Ocean Database 2018 [148] using the WODselect tool (https://www.ncei.noaa.gov/access/world-ocean-datab ase-select/dbsearch.html). Both the WOA13 and the WOD18 data were processed and visualized with Ocean Data View version 5.1.7 [149] and MATLAB R2018b [75]. TEOS-10 (http: //www.teos-10.org/software.htm) [76] was used to convert in situ temperature and salinity to conservative temperature and absolute salinity, as well as calculate potential density anomaly, using IBM ILOG CPLEX Optimization Studio V12.8.0 as an optimizer. Nutrient ratios for each water mass were calculated using Type II linear regressions with pre-existing code [150]. All data were plotted, but outliers were excluded from these linear regressions. With regards to visualizing data, the viridis colormap was selected for scientific use of color and accessibility [151], rather than the “odv” option which is used as default. We subdivided this data into the three water masses that compose the ETNP ODZ using con- servative parameters centered around water mass endmembers specified in a water mass analysis of this region [1]. Table 1 provides these specifications, which are generously wide in an effort to in- cludesufficientdatathatoutliersdonotinterferewithcalculations. Withthesewatermasssubsets, we analyzed the C:N:P regeneration ratios resulting from anaerobic processes. The percentages of nitrite re-oxidized were calculated using Eqs. 1-2, a common formulation for analyzing the percent yield. A sample calculation is included in Eq. 3, whereas Eq. 4 depicts how we calculated uncer- tainties for nitrite re-oxidation using these regressions. WOA13 data lacked carbon measurements, so we used C:N ratios in the WOD18 data set to verify that nitrite re-oxidation was occurring, rather than to quantify the process. NO − 2 oxidized (%)=100 (NO − 3 :PO 3− 4 ) expected − (NO − 3 :PO 3− 4 ) measured (NO − 3 :PO 3− 4 ) expected ! (3.1) 39 Table 3.1: Ranges for filtering WOA13 and WOD18 data into coherent water masses. Starred variables were only used to filter the data for linear regression of nutrient ratios in Fig. 3, not for geographic distri- butions in Fig. 2. 13CW NEPIW AAIW Min Max Min Max Min Max θ /ºC 12.5 13.5 9 10 5 6 S A /g kg − 1 34.8 35.2 34.7 34.85 34.67 34.72 σ θ /kg m − 3 26.2 26.4 26.7 26.9 27.2 27.3 *Latitude/ºN 0 n/a 0 n/a 0 n/a *NO – 3 /µ mol kg − 1 15 n/a n/a n/a n/a n/a NO − 2 oxidized (%)=100 − 94.4− (NO − 3 :PO 3− 4 ) i − 94.4 (3.2) 54%=100 − 94.4− (− 43) i − 94.4 (3.3) Uncertainty in NO − 2 oxidized (%)=100 (NO − 3 :PO 3− 4 ) measured uncertainty (NO − 3 :PO 3− 4 ) expected ! (3.4) 3.4.2 Water mass analysis for high-resolution nitrite re-oxidation For our regression analysis, we binned data into water masses to enabled calculation nitrite re- oxidation percentages from a large number of measurements, thus, increasing the precision of our calculations. Unfortunately, thisapproachlimitstheresolutionofthedataproduct,astherequired large number of measurements prevents comparisons between geographic coordinates. Binning by water masses also removes vertical structure. To create a higher resolution estimate of nitrite re-oxidation in the ETNP ODZ, we implemented a novel approach to water mass analysis. The water mass analysis performed in this paper builds on previous work in this region[1, 152] using the omp2 water mass analysis package [77]. These studies focused first on only mixing [1] (Evans et al., 2020), then combined mixing with anaerobic remineralization [147]. Unfortunately, the omp2 package used in these analyses only included one remineralization term and is thus not appropriate for analysis within an ODZ. Fortunately, the pyompa water mass analysis software packagecancalculatemultipledifferentremineralizationterms,eachwithaflexibleremineralization 40 stoichiometry [13]. This flexible stoichiometry feature can calculate the anaerobic remineralization stoichiometry for every measured sample independently, whereas our regression analysis needed to bin this data. In both cases, we compare the calculated stoichiometry the -94.4:1 reference stoichiometry, yielding estimates of nitrite re-oxidization at far higher spatial resolution. This water mass analysis simultaneously calculates the water mass contributions to a sample, the amount of aerobic remineralization that has occurred in that sample, the amount of anaerobic remineralization that has occurred in that sample, the stoichiometry of aerobic remineralization that has occurred, and the stoichiometry of anaerobic remineralization that has occurred, for each sample independently. To perform these calculations, pyompa needs endmembers for each water mass as well as upper and lower boundaries for the stoichiometries of aerobic and anaerobic rem- ineralization. The temperature and salinity endmembers for each water mass were taken from the endmembers from Evans et al. (2022). Nutrient endmembers were determined using the results of the WOD18 data analysis (Fig. 3b-d). Since we are interested in anaerobic processes, we de- fined the nutrient endmembers as the approximate concentrations where aerobic remineralization switches to anaerobic remineralization. These concentrations are specified in Table S2. The upper and lower bounds for flexible remineralization stoichiometries were set using the range of stoi- chiometries observed in the WOD18 regression analysis (Table 2, Table S1 for more information), and were applied uniformly to each measured sample. We present the WOD18 data that we used to derive our water mass endmembers, these endmembers, and the 110 ºW data we analyzed in Fig. S4 to help visualize our data analysis framework. During our analysis, we found that the nitrite re-oxidation values determined through water mass analysis were sensitive to the nutrient concentrations of our endmembers. To account for this sensitivity,weperformedourwatermassanalysistentimeswithslightvariationsintheendmember nutrient concentrations. The results we present are the average of those ten simulations and the 41 uncertainty is the standard error of these results. It is important to remember that in water mass analysis, remineralization is calculated as the difference in the concentration of nutrients in a sample and the concentration calculated using a given combination of endmembers. Therefore, negativeremineralizationresultsindicatealoss,ratherthanagain,ofnutrientconcentrations. The supplemental information contains far more detail about the setting and mathematics used in this water mass analysis. We performed water mass analysis on only the WOD18 data because the inclusion of carbon measurements constrains the remineralization stoichiometry, preventing it from acting as a free parameter that auto-fits the data. Carbon measurements are limited in the ETNP ODZ (Fig. S2), and we focused our analysis on three cruises on the 110 ºW line from the tip of Baja to 10 ºN. This transect is near the core of the ODZ, which limits other waters from the Pacific Gyre or the California Current from mixing into the region of interest. South of 10 ºN, we observed a water mass transition that prevented this region from being included. In our analysis, we only processed data between 26.4-27.2 kg m − 3 , unlike previous studies in this region. 27.2 kg m − 3 represents the core of the AAIW, which acts as a lower bound for the oxygen deficient layer [1]. We set 26.4 kg m − 3 asanupperboundtoomitthermoclinesamplesfrompyompa. Thethreecruisesanalyzedhere sampled in 1996, 2007, and 2016; information and results from these cruises have been published in other locations as well [153, 147]. 3.5 Results 3.5.1 Basin-scale quantification of nitrite re-oxidation Figure 2 depicts the geographical extent of each water mass using subsets of World Ocean Atlas 2013 gridded data (WOA13) based on the temperature and salinity of each water mass, listed in Table1. The13CWsubsetwasselectedtobeslightlydeeperthanthatfoundinEvansetal. (2020). This decision is based on the results of particle backtracking into the ETNP ODZ, which found the 42 entryofthiswatermassthroughthesouthernboundaryoftheODZslightlydeeper[7]. Inaddition, while other references [104] define the center of the NEPIW and its southern counterpart, SEPIW, closertoσ θ =27kgm − 3 ,theNorthernEquatorialUndercurrentJetsthatinjecttheNEPIWintothe ETNPODZarecenteredatσ θ =26.8kgm − 3 (Qiuetal. 2013;Margolskeeetal. 2019). Torepresent the waters of the NEPIW that feed the ETNP ODZ, we adjusted our subset to be shallower, even though this process selects specifically the upper region of these water masses. Figure 3.2: Maps of (a) 13 ºC Water, (b) Northern/Southern Equatorial Pacific Intermediate Water,and(c)AntarcticIntermediateWater,depictingPO 3– 4 /µ molkg − 1 . Themagentalinedepicts the concentration past which anaerobic respiration appears to begin, as given in Fig. 3. Water mass formation sources are noted. The phosphate concentrations of each water mass are used to depict how the water masses age as they are transported from their source regions to the Pacific ODZs (Fig. 2). The magenta contour represents the approximate location where the signal of anaerobic metabolisms can be detected. This concentration is identified in Fig. 3 and described in surrounding text, however, these phosphate concentrations are based on the ETNP ODZ and do not reflect the ETSP as well. Nevertheless, theseplotsconfirmthatthePacificODZssharecoherentsourcewatermasses, aswell as visualize the flow paths for these water masses entering the ETNP ODZ. 43 The AAIW has several formation regions across the Southern Hemisphere, two of which are displayed in Fig. 2c; however, the geographical extent of this water mass endmember does not intersect directly with either of these locations. Instead, the AAIW observed in both the ETNP and ETSP ODZ is likely a form of AAIW that has been mixed with other water masses, likely PDW, as Tasman Sea AAIW has a northwestern mixing location [154, 155]. While the ocean’s fixed nitrogen budget depends on the relative partitioning of nitrite lost via further denitrification versus recycled via oxidation, measurements of nitrite reactions are partic- ularly challenging. Investigations into nitrite oxidation typically extrapolate results based on a small number of cruises with limited spatiotemporal sampling. To overcome these limitations, we analyzed WOA13 data for nitrate:phosphate ratios within the source water masses for the ETNP ODZ.Ifeachstepofdenitrificationoccursirreversibly,thenitrate:phosphateratioshouldbe-94.4:1 for each water mass. However, the 13CW, NEPIW, and AAIW had slopes of -20±1, -43±1, and -34±3, appropriately(Fig. 3aandTable2). SinceWOA13didnotcontaincarbondata, weverified that these deviations in the nitrate:phosphate ratio were due to nitrate by comparing regressions of both nutrients against total inorganic carbon data from WOD18 (Fig. 3c and 3d). We found thatcarbon:phosphatedeviatedlessthan12%fromtheexpected106:1ratioandthisdeviationwas positive for the NEPIW and AAIW but not the 13CW. These positive deviations are likely due to carbonate dissolution, since this is one of the few processes that increases the carbon:phosphate ratio and it can occur within these deeper waters (Hernandez-Ayon et al. 2019). In contrast, the negative deviation in the 13CW occurs due to photosynthesis near the base of the euphotic zone, which the 13CW can mix into. Despite the fact that carbon:phosphate ratios deviated by less than 12%, the carbon:nitrate ratios were -3±0.4 and -3±0.2 for 13CW and NEPIW, respectively. These 44 ratios of approximately -3 are significantly lower than the -1.12 predicted from reaction stoichiom- etry (Froelich et al. 1979), indicating that denitrifying a single nitrate requires almost three times the carbon expected. Figure 3.3: Evolution of nutrients within the 13CW, NEPIW, and AAIW in the northern hemisphere using (a) NO – 3 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOA13, (b) NO – 3 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOD18 for comparison with WOA13, (c) total CO 2 /µ mol kg − 1 vs NO – 3 /µ mol kg − 1 from WOD18, and (d) total CO 2 /µ mol kg − 1 vs PO 3– 4 /µ mol kg − 1 from WOD18. Table3.2: RatiosofnutrientsdepictedinFig. 2usinglinearfitsandreactionstoichiometryforcomparison. Data Source Water mass Nutrients Respiration Slope WOA13 13CW NO – 3 :PO 3− 4 Anaerobic -20±1 NEPIW NO – 3 :PO 3− 4 Anaerobic -43±1 AAIW NO – 3 :PO 3− 4 Anaerobic -34±3 WOD18 13CW NO – 3 :PO 3− 4 Anaerobic -43±4 NEPIW NO – 3 :PO 3− 4 Anaerobic -59±6 13CW Total CO 2 :NO – 3 Anaerobic -3±0.4 NEPIW Total CO 2 :NO – 3 Anaerobic -3±0.2 13CW Total CO 2 :PO 3− 4 Aerobic 99±2 NEPIW Total CO 2 :PO 3− 4 Aerobic 118±3 AAIW Total CO 2 :PO 3− 4 Aerobic 118±4 Froelich et al. (1979) NO – 3 :PO 3− 4 Anaerobic -94.4 Total CO 2 :NO – 3 Anaerobic -1.12 Total CO 2 :PO 3− 4 Anaerobic 106 45 The approximate phosphate concentrations where anaerobic processes begin is defined as the pointatwhichthenitrate:phosphatestoichiometrydeviatesfrom16:1inFig. 3a. Theseconcentra- tionsareplottedasinFig. 2tovisualizetheapproximaterangewhereeachwatermasscontainsthe signal of denitrification. In this northern region of the ETNP ODZ, Evans et al. (2020) found that the Antarctic Intermediate Water is hypoxic based on measured oxygen concentrations. Despite the hypoxia of these water masses, our basin scale analysis revealed that denitrification has oc- curred in the Northern Equatorial Pacific Intermediate Water and even the Antarctic Intermediate Water. Previous studies have identified dissimilatory nitrate reduction occurring in seawater with dissolved oxygen concentrations reaching up to 20 µ M [34], and this low but nonzero threshold would facilitate this metabolism in this environment. Since the Northern Equatorial Pacific Inter- mediate Water is introduced into the north side of the ETNP ODZ [7], this water mass is likely driven to anoxia as it migrates southeast through the ETNP ODZ. Nitrite data from the Costa Rica Dome [10, 156] supports this geographical differentiation in water mass oxygen availability, as nitrite accumulates far deeper in the south of the ETNP ODZ than the north [1]. Elevated phos- phate concentrations in Fig. 2c indicate that in the AAIW, this denitrification signal intensifies within the southern portion of the ETNP ODZ, where the ETNP ODZ is deeper. Previous studies have found a deeper ODZ in this region [146] but had not identified the water masses linked to this process. We propose that the vertical shifts in where nitrite accumulates in the ETNP ODZ can be explained by the general circulation of water masses, as they slowly reach anoxia while moving away from their introduction points. Table 3.3: Percent of nitrite re-oxidized based on slopes fit in Table 2. Data source Water mass Nitrite oxidized (%) WOD18 13CW 66±12 NEPIW 64±8 WOA13 13CW 79±7 NEPIW 54±2 AAIW 64±9 46 Using the stoichiometries for anaerobic remineralization calculated for each water mass using linear regression (Table 2), we can calculate the percent of nitrite re- oxidized (Table 3). We observe the largest nitrite re-oxidation for the 13CW using the WOA13 dataset. The 13CW is the shallowest water mass and it can shoal into the photic zone, enable oxygen production that creates hypoxic microenvironments [157]. Nitrite reoxidation is reported to be responsible for 40%-80% of the oxygen consumption in the edge of the ETNP ODZ [158], and this fact supports the significant fraction of nitrite oxidation we observed in the 13CW using the WOA13 dataset. A similar trend canbeseeninthenitritere-oxidationvaluesforthe13CWandNEPIWintheWOD18data,though thesevaluesareclosertothestoichiometricestimateof-94.4. TheanaerobicportionoftheWOD18 data originates almost solely from cruises on the CLIVAR P18 line. This repeat transect occupies 110 ºW and therefore bisects through the middle of the ETNP ODZ (Fig. S2), emphasizing the processes within the oxygen deficient layer. The WOA13 gridded data encompasses the ETNP ODZ causing its boundaries to be well-represented in this dataset, leading to the higher impact of hypoxic microenvironments on nitrite re-oxidation. 3.5.2 Spatiotemporal variability of nitrite re-oxidation Weperformedawatermassanalysisondatafromthe110 ºWlinetoanalyzethespatialvariability ofnitritere-oxidation. Therehavebeenthreecruisesonthe110ºWlinewithcarbonmeasurements, which enabled analysis of temporal variability as well. Oxygen across this transect within the 26.4- 27.2 kg m − 3 potential density range was consistently below the detection limit of SBE43 sensors [145], except for shallow intrusions to the south and north (Fig. 4). The most intense of aerobic remineralization occurred primarily between 600-800 m and it strengthens with depth. Aerobic remineralization was also slightly stronger to the south of this transect. The amount of aerobic remineralizationbetween200-400mhasalsoincreasedoverthesethreecruises. Ontheotherhand, 47 anaerobic remineralization occurred primarily north of 14 ºN shallower than 400 m, and it has strengthened over these years, as previously observed (Evans et al. 2022). The percent of nitrite re-oxidized is far more spatiotemporally variable than oxygen or either remineralization. A core of low nitrite re-oxidation persisted between 300-400 m north of 10 ºN, though it varies with regards to the magnitude and latitudes where it primarily manifests. This depth ranges matches with the core of the Northern Equatorial Pacific Intermediate Water (Fig. 1, Fig. S5), and these results match the lower nitrite re-oxidation within the NEPIW determined via regression. The intensity of anaerobic remineralization appears to influence the upper boundary of this low nitrite re-oxidation core, as the hotspots of anaerobic remineralization have slightly higher nitrite re-oxidation than their neighbors. This interference is most obvious in 2016. Interestingly, this area is co-located with the Antarctic Intermediate Water, but the AAIW itself covers a far larger region where nitrite re-oxidation is far higher. This higher re-oxidation might swamp out the hotpots here when analyzed through regression. However, the most notable feature of nitrite re-oxidation is the significantly lower values observed in 2007. We are uncertain what caused this significant different in nitrite re-oxidation difference. The uncertainty in nitrite re-oxidation, calculated as the standard error, increased when the nitrite re-oxidation decreased. This behavior likely occurred because at high nitrite re-oxidation, this high value is not sensitive to the endmember nutrient concentrations, while lower values of nitrite re-oxidation become more sensitive to endmember nutrient concentrations. This water mass analysis method suffers from lower precision than the regressions, but the relative values and their depths match well with the regression results. Interestingly, while the uncertainty in nitrite re- oxidation was high, the uncertainties in aerobic and anaerobic remineralization were quite low, with relative standard errors less than 10% and 1%, appropriately (Fig S7). Nevertheless, this water mass analysis provides a robust characterization of the spatiotemporal variability of nitrite 48 Figure 3.4: Oxygen concentrations and pertinent water mass analysis results for cruises on the 110 ºW line. Figures in the left column are from 1994, the middle is 2007, and the right is 2016. Eachrowcorrespondstothepropertyspecifiedonthecolorbartotheright. Oxygenconcentrations were plotted using raw data, whereas nitrite reoxidation, aerobic remineralization, and anaerobic remineralization were plotted using the mean of 10 calculations. The uncertainties in nitrite re- oxidation were plotted using the standard error of these 10 calculations. re-oxidation in the ETNP ODZ and suggests that nutrient data, on its own, can be mined to calculate nitrite re-oxidation. We encourage further analysis into anaerobic processes in Pacific ODZs using this water mass analysis framework. 3.5.3 Future and global implication We can estimate the percent of nitrite that is re-oxidized in the ETNP by comparing data against the-94.4nitrate:phosphateratio..ThesecomparisonsrevealthatwithinthecoreoftheETNPODZ, assampledbytheWOD18data,38±10%-54±9%ofthenitriteproducedisre-oxidized. Thehigher end of this estimate is similar to a previous study that found 50% nitrite re-oxidation [10], as well as another study that found nitrite oxidation rates exceeded nitrate reduction rates [9]. Extending this data to the entire ETNP, as presented with WOA13 data in Table 2, reveals that the 13CW in theupperwatercolumncontains79±7%nitriteoxidation, whichisnoticeablyhigherthanprevious 49 estimates. Nitrite oxidation is likely highest in the 13 ºC Water in the WOA13 because previous studies have also noted that the upper oxycline contains higher measured nitrite oxidation rates than the oxygen deficient layer [11], and the WOA13 dataset contains the entire upper oxycline in the ETNP ODZ. Nitrite re-oxidation estimates from the WOA13 data are consistently higher than that observed in WOD18 data. Beyond the tip of Baja, the WOD18 data only contains open ocean samples (Fig. S2), whereas the WOA13 data contains several coastal transects (Fig. S1). Since nitrite re-oxidation is higher in coastal waters, their inclusion in WOA13 likely resulted in the observed 20% increase in nitrite re-oxidation. To analyze the spatiotemporal variability of nitrite re-oxidation in the ETNP ODZ, we also performed a water mass analysis with flexible remineralization stoichiometries. This is the first quantitativewatermassanalysisperformedwithmultiplemetabolismsthatweareawareof,aswell as the most sophisticated water mass analysis ever performed in an ODZ. The software and code for this project is open source, free, and documented to facilitate its use by future researchers. The flexiblestoichiometryfeatureinpyompaenabledustocalculatethepercentofnitritere-oxidizedfor every measured sample. These results reveal a consistent core of low nitrite re-oxidation between 300-400 m, which matches the depth of the Northern Equatorial Pacific Intermediate Water (Fig. S4). These results also indicate significant temporal variability in nitrite re-oxidation, with 2007 having notably low amounts. Unfortunately, this method could not be applied to data in the core of the 13 ºC Water, which overlaps with the thermocline, but the trend of increasing nitrite re- oxidation shallower than 300 m matches expected results for the 13CW. In addition, the Antarctic Intermediate Water also has far more nitrite re-oxidation than the Northern Equatorial Pacific Intermediate Water as observed in the WOA13 data, except a small region north of 16 ºN. Though watermassanalysiswithpyompacalculatesnitritere-oxidationwithfarmoreuncertaintythanour water mass linear regressions, it provides a far more detailed picture into this process. Notably, 50 there is no apparent correlation between nitrite re-oxidation and measured oxygen concentrations. This lack of correlation occurs because the oxygen sensors used on these cruises, SBE43 electrodes, lack sufficient detection limits to differentiate oxygen concentrations in the anoxic core, where oxygen is often under 50 nM [159]. Thesesignificantproportionsofnitritere-oxidationindicatethatsteady-statenitrateandphos- phate concentrations cannot indicate the rate or extent of nitrate reduction without a 38%-79% correction, depending on location within the ETNP. Inverse modeling of ODZs requires nitrite ox- idation closely paired with nitrate reduction to maintain realistic concentration and isotope distri- butions (Martin et al. 2019) or artificially lower remineralization rates to prevent nitrate depletion (Su et al. 2015). Models that simulate fixed nitrogen loss using N* without including prolific and depth-dependentnitritere-oxidation[136,42,160]likelyoverestimatethefixednitrogenlosswithin marine ODZs. The equation for N* is displayed in Eq. 5. Fu et al. (2018) predicts that ODZs will expanduntilapproximately2100thencontract, butthesecalculationsrelyonafixedstoichiometry and increasing the oxycline area of the 13 ºC Water may alter the amount of nitrite denitrified ver- sus recycled. Prolific nitrite re-oxidation causes N* to overestimate the fixed nitrogen loss because it recycles the fixed N but does not significantly alter the phosphate released by nitrate reduction. Therefore,higherproportionsofnitritere-oxidationtonitratereductionleadstodisproportionately high phosphate concentrations relative to nitrate, which lowers the N* and overestimates the fixed N loss. N ∗ =(NO − 3 +NO − 2 )− 16 PO 3− 4 (3.5) This study is the first effort to use basin-wide data to determine the partitioning of nitrite be- tween reduction and oxidation rather than extrapolating from a subset of points or matching data with models. This approach suffers from significant scatter in water mass data, causing large un- certainties, though these uncertainties may be derived from the fact that the nitrite re-oxidation 51 varies widely across the ETNP. In addition, this analyze focuses on the sources and sinks of fixed nitrogen, and therefore cannot differentiate between mechanisms of nitrite reduction such as fur- ther denitrification or anammox. Ultimately, sparse measurements of both inorganic nitrogen and carbon parameters within Pacific ODZs limit the ability to derive robust conclusions. This large- scale data manipulation also cannot determine the oxidant for nitrite oxidation, though oxygen, iodate, and even nitrite dismutation have been proposed [161, 9, 162]. We calculate that between 54%-79% of nitrite produced in the ETNP is re-oxidized, with 38%-54% in the core of the ETNP ODZ. We find the highest nitrite oxidation in the shallow 13CW because this water mass con- tains hypoxic microenvironments because the secondary chlorophyll maximum manifests within it, which matches previous studies reporting high nitrite re-oxidation in shallow waters. This analy- sis demonstrates the significance of nitrite recycling, which exceeds nitrite loss across the ETNP ODZ, and the impact of this recycling on elevating phosphate comparisons. Most importantly, the disconnect between nitrate reduction rates from concentrations of nitrate and phosphate influences how oceanographers and ocean modelers represent fixed nitrogen loss in the ocean. 3.6 Acknowledgements N.E.andJ.W.M.weresupportedbyNSFOCE2023708andJ.T.wassupportedbytheUniversityof Southern California Wrigley REU program. We would like to acknowledge the extensive programs that went into compiling the World Ocean Atlas and World Ocean Database such that these resources could be publicly available. Subsets of these datasets used for this analysis, as well as the code for processing them, is available at https://github.com/NatalyaEvans/ETNP stoich. Amanda Taing provided copy edits for this manuscript. Avanti Shrikumar assisted immensely in training the authors in using the pyompa package. JL Weissman helped proofread this manuscript as well. 52 Chapter 4 Rapid expansion of fixed nitrogen deficit in the eastern Pacific Ocean revealed by 50-year time series 4.1 Abstract Climate change is expected to increase the strength of ocean Oxygen Deficient Zones (ODZs), but we lack detailed understanding of the temporal or spatial variability of these ODZs. A fifty-year time series in the Eastern Tropical North Pacific (ETNP) ODZ revealed that it strengthened by 30% from 1994 to 2019. We subdivided the ODZ into a core and a deep layer based on potential density and revealed that different processes control the magnitude of fixed nitrogen loss in these two regions. We postulate that the depth of the upper ETNP ODZ water mass, the 13 ºC water, influences the organic carbon supply to the core ODZ and therefore its strength. We correlated the fixed nitrogen loss in the core ODZ with a nearby sedimentary nitrogen isotope record and found that this recent, rapid increase has only occurred a few times over the last 1,200 years. Using this correlation,wederivedthefirstconfidenceintervalforthestrengthofthecoreETNPODZ,9.2-12.5 µ mol kg − 1 of fixed nitrogen loss. While the current increase is comparable to only two previous events, it is still within this confidence interval. Nevertheless, climate driven intensification could lead to unprecedented changes within the next decade. The deep ODZ also strengthened from 2016-2019 by approximately 30%, even more rapidly than the core ODZ. This dramatic increase was not observed over the rest of the 50-year time series. 53 4.2 Introduction Open ocean Oxygen Deficient Zones (ODZs) play a pivotal role in the global nitrogen cycle [73, 42, 163,164],carboncycle[165,166],andironcycle[167,19],aswellasshapingecosystemproductivity (Gallo & Levin, 2016) and predator foraging habitat [168, 169]. These ODZs are expected to expand due to climate change as ocean warming elevates respiration rates, increases stratification, and lowers oxygen solubility. Subsurface deoxygenation in the tropical Pacific Ocean has already been observed [170, 16], but the lack of time series in ODZs hinders our ability to differentiate expansion signals from natural variability [171]. The Eastern Tropical North Pacific (ETNP) ODZ is the largest of the three ocean ODZs [172, 6] and sediment core proxies indicate that its strength has fluctuated dramatically over the past thousand years [14, 173, 18]. Nevertheless, attributing ODZ strengthening to anthropogenic climate change requires estimates of its natural variability. We analyzed water column properties measured during eight cruises to the ETNP between 1972-2019, seven of which transited through the center of the ETNP ODZ along the 110 ºW line. The sampling coverage provided by these cruises is highlighted in Fig. 1, which also depicts the frequency of oxygen deficient conditions sampled on the 26.5 kg m − 3 isopycnal, based on an atlas of Pacific ODZs [145]. This atlas reveals that the 110 ºW line crosses through some of the most permanently oxygen deficient waters. By comparing cross-sections of the ODZ on the 110 ºW line over time, we assessed the spatiotemporal variability of the ETNP ODZ. Many previous studies analyzing ODZ variability over time focus on ODZ size [174, 175, 56]. Most notably, Deutsch et al. (2014) interpreted a record of sediment particulate organic nitrogen isotopes from 1850- 2010 to demonstrate that ETNP ODZ only began strengthening around 1993, before which it was weakening. Additional work with this sediment core extended this record from 160 years to 1,200 years (Tems et al., 2016). In our study, we report the concentration of fixed nitrogen lost for each cruise transect and use it as a measure of the ODZ’s strength. We focus on strength because 54 our cross-sections cannot accurately capture ODZ volume. In addition, mesoscale features such as eddies spread and distribute ODZ water outside of its traditional bounds [1, 176] as well as inject non-ODZ water into its domain [7], hindering the ability of discreet sampling to quantify the size of an ODZ. The source waters to the oxygen deficient layer of the ETNP ODZ have been identified as the 13 ºC water (13CW) and the Northern Equatorial Pacific Intermediate Water (NEPIW), with Equatorial Surface Water and modified Antarctic Intermediate Water acting as upper and lower boundaries [1, 177]. Evans et al. (2020) found that nitrite accumulation occurred below maxima in 13CW as well as sea surface height anomaly, indicating that eddies and meanders from the California Undercurrent introduce 13CW hotspots into the ETNP ODZ that influence nitrogen cycling in this region. Interestingly, the depth of the 13 ºC isotherm also correlates with the strength of the ETNP ODZ [174]. The 13CW and NEPIW also act as the Pacific Equatorial endmembers to the California Current System (CCS) [1]. A 40-year record of the CCS indicates that its deoxygenation correlates with the strength of the ETNP ODZ, and the NEPIW caused 81% of the deoxygenation observed [178]. The time series presented in this manuscript on the 110 ºW line also serves as a time series through the Pacific Equatorial endmember of the CCS, which provides relevant information for analyzing deoxygenation in the CCS. 4.3 Materials and Methods 4.3.1 Sample acquisition and measurement The primary data for this publication were acquired on eight cruises, seven of which were on the 110 ºW longitude line. This time series spans a total of 47 years. The stations for these cruises in the ETNP ODZ are plotted in Fig. 1 and descriptive metadata for these cruises and contained in Table S1. Temperature, salinity, oxygen, nitrate, nitrite, phosphate, and silicate were all measured with methods standardized in the U.S. Joint Global Ocean Flux Study (http: 55 //usjgofs.whoi.edu/protocols rpt 19.html). Additional information about the first four cruises can be found in previous studies [146, 153]. Following Horak et al. (2016), we corrected for systematic errors between expeditions by applying quality control to the nitrate and phosphate data, using sigma-4 surfaces in the Global Ocean Data Analysis Project (GLODAP) Pacific data as the reference. These correction factors were applied if the offset was greater than 2% for nitrate or phosphate, and the correction factors are included in Table S2. More information about this quality control can be found in Horak et al. (2016). Figure 4.1: Map presenting the transects from the eight cruises analyzed in this study and the location of the Pescadero Basin coring site. These data are presented over a contour map of the fraction of ODZ conditions observed on the 26.5 kg m − 3 isopycnal. 56 4.3.2 Fixed nitrogen loss via integration method Fixed nitrogen loss was calculated using –∆N*, as provided in Eq. 1. In this formulation, larger –∆N* values indicate more fixed nitrogen has been lost and therefore the ODZ is stronger. We also applied a post-hoc correction for nitrite re-oxidation to this integrated –∆N*, described in the Supplemental Information, to the data presented in Fig. 2a-c. − ∆ N ∗ =− ((NO − 3 +NO − 2 )− 16PO 3− 4 +2.9) (4.1) These –∆N* measurements were gridded every 0.5 degrees between 14 ºN and 23ºN and 1 m, then integrated with a depth cumulative cubic integration routine between the integrands specified for eachpotentialdensityrangeusingMATLABR2021A.ThisintegrationapproachwasusedinHorak et al. (2016), however, we applied different potential density ranges. Potential density ranges were selectedbasedonthecoresofwatermassesintheETNPODZ[1], exceptfortheshallowestdensity surface, 24.75kgm − 3 , whichwasusedinHoraketal. (2016). Thepotentialdensityof26.2kgm − 3 correspondstothe13ºCWater(13CW)[90,103],26.8kgm − 3 correspondstothepotentialdensity where the Northern Equatorial Undercurrent jets inject Northern Equatorial Pacific Intermediate Water (NEPIW) into the ETNP ODZ [179], and 27.2 kg m − 3 corresponds to modified Antarctic IntermediateWater(AAIW)[155]. Wereferredtotheregionbetweenthecoresofthe13CWandthe NEPIWasthecoreODZbecauseattheselatitudes,oxygenislowestandnitriteaccumulateswithin this potential density range [1, 159]. We separate the deep ODZ into a different potential density range because typically, it is barely aerobic, though oxygen concentrations are below the detection limits for conventional sensors [180, 60]. We set the bottom of the ODZ at 27.2 kg m − 3 because, in this region of the ODZ, oxygen concentrations become measurable [1]. For readers interested in the depths for each of these potential density horizons, Fig. S1 illustrates the relationship between 57 depth and potential density. We note that N* indicates the history of fixed nitrogen loss in a water parcel,ratherthanthatfixednitrogenlossoccurringduringsampling. Therefore,thefixednitrogen loss we report may have occurred elsewhere and been transported to the sampling location. To address where fixed nitrogen loss occurred in this ODZ, we compare our results with an atlas of Pacific ODZs [145] as well as a ROMS model of the ETNP ODZ [7]. 4.3.3 Fixed nitrogen loss via water mass analysis Wealsocalculatedfixednitrogenlossusingawatermassanalysiswithextendedoptimummultipa- rameter analysis (eOMP). This method allowed us to specify the stoichiometry of remineralization to match the stoichiometry measured in the ETNP ODZ [177], increasing the accuracy of our cal- culations. This computational approach also implicitly accounts for the nitrite re-oxidation that occurs in this region, because it uses the same stoichiometry that is observed, unlike N*. Extended optimum multiparameter analysis (eOMP) calculations were performed using the GUI option in a modified version of the omp2 MATLAB package written by Johannes Karstensen and Matthias Tomczak, which has been uploaded to https://doi.org/10.5281/zenodo.6519316. The original version of this software can be found at https://www.mathworks.com/matlabcentral/fileex change/1334-omp-analysis. All computations besides the integrations were performed using MATLAB R2018B. The basis of water types for eOMP were selected primarily based on their defi- nitions in Evans et al. (2020), except upper Pacific Subarctic Water (uPSUW). Instead, this water type was taken from the California Current System (Bograd et al., 2019). In a previous water mass analysis of the ETNP ODZ, nutrient concentrations were adjusted to the most representative values for each cruise. While this method provides robust analysis of water mass distributions, it does not allow us to compare the accumulated remineralization between each cruise, which is stoichiometrically equivalent to the fixed nitrogen loss. Conservative temperature (Θ), absolute salinity (S A ), phosphate (PO 3– 4 ), nitrate (NO – 3 ), silicate (SiO 2– 4 ), and potential density anomaly 58 (σ θ ) were used as parameters for this eOMP for each cruise. Thermodynamics Equation of State 10 (TEOS-10) was used to convert in situ temperature and salinity to conservative temperature and absolute salinity, as well as calculate potential density anomaly, using IBM ILOG CPLEX Optimization Studio V12.8.0 as an optimizer [76]. PO 3– 4 , NO – 3 , and SiO 2– 4 concentrations for the 13CW, NEPIW, and AAIW definitions were selected by identifying the approximate PO 3– 4 concentration where remineralization switches from aerobic to anaerobic, as identified from WOD18 data [177], then scaling that concentration to the appropriateNO – 3 andSiO 2– 4 concentrationsusingthestoichiometricrelationshipsspecifiedinEvans et al. (2022). The stoichiometry for anaerobic remineralization was selected from the same data, then this stoichiometry, water mass definitions, and weightings were refined by slight adjustments to minimize the sum of squared residuals output from the eOMP. The water mass definitions superimposedontheinputdataaredepictedinFig. S2,andtheSupplementalInformationcontains more details about this eOMP. The strength of the ODZ is calculated by taking the mean of the maximum 10% quantile of the fixed nitrogen loss for each cruise. This quantile range was selected toensurethatasingleoutlierdidnotbiasthedata, butvaluesstillrepresentthemostintensefixed nitrogen loss sampled on each cruise. 4.3.4 Data processing for time comparison In this paper, we compare several time series derived from the eight cruises in this region, as well as relevant ancillary information (Fig. 2). Calculations for integrated fixed nitrogen loss (Fig. 2a-c) are described entirely in section 2.2. For the maximum fixed nitrogen loss (Fig. 2d), we converted the accumulated anaerobic remineralization to NO – 3 equivalents by scaling it with 62, the stoichiometry we used for relating NO – 3 to PO 3– 4 . We then subset this fixed nitrogen loss into the appropriate potential density ranges. We calculated the maximum of the fixed nitrogen loss by extracting values greater than or equal to the 90% quantile for each cruise and taking their mean. 59 Error bars correspond to the standard deviation of this extracted data. Table S5 reports the mean, standard deviation, and number of samples for each potential density range in the 90% quantile. Previous work has identified that the source waters to the California Current System originate fromalocationonthe110ºWline. Togeneralizeourtimeseriesforbroaderaudiences,wecompared our time series against a record of data from the CCS collected by the California Cooperative Fisheries Initiative. For the CalCOFI O 2 data presented in Fig. 2e, all samples between 100 m and 400 m for every station were averaged for every year and quarter, as performed in Evans et al. (2020). This intermediate dataset was uploaded to the same Zenodo repository for repeatability and attached as Table S6. A Grubbs test for outliers was performed for the four quarterly cruises each year, and then the mean and standard deviation was calculated for the remaining values. PreviousworkhascorrelatedthestrengthoftheETNPODZ,asmeasuredvianitrogenisotopes, with the depth of the 13 ºC isotherm taken from the World Ocean Database [174]. We analyzed the depth of the core 13CW, similar to the depth of the 13 ºC isotherm, on the eight cruises in our time series to compare against ODZ strength. The depth of the 13CW water was calculated by converting the 1 m binned data for each cruise into conservative parameters via TEOS-10, as previously described, identifying the depth of all samples where 13.31 ºC ≤ Θ ≤ 13.51 ºC, then taking the mean and standard deviation of these samples between 14 – 19 ºN. Stations north of 19 ºN were removed because the Gulf of California leads to a confluence of water masses that creates a transition zone at these locations [181], as well as introducing Gulf of California Water [182], which could influence this calculation. The mean, standard deviation, and number of samples for each expedition is specified in Table S7 and the depth for each station and expedition is presented in Fig. S5. To examine whether the fixed nitrogen loss we observed was advected from elsewhere in the ODZ, we analyzed published data from an atlas of Pacific oxygen deficient conditions as well as a 60 ROMS model of the ETNP ODZ. We calculated the core oxygen deficient layer potential density by averaging the top and bottom potential densities of the ODZ specified in Kwiecinski & Babbin (2021), and we recreated a figure of water entry into the ODZ from Margolskee et al. (2019) with data shared by the author and contour lines added by eye. The data and code for these analyses are included in the Github repository for this paper. 4.3.5 Natural variability estimation via sediment core conversion A primary goal in this study was to assess if modern changes in ETNP ODZ strength are unprece- dented and therefore could be attributed to anthropogenic climate change. For this analysis, we required a time series of the ETNP ODZ with enough temporal measurements to assess its natural variability. Tems et al. (2016) measured sedimentary nitrogen isotopes from the Pescadero basin that spanned nearly 1,200 years, which Deutsch et al. (2014) correlated with ODZ strength. The supplemental information in Tems et al. (2016) contains the entire Pescadero basin sediment core record. The 20-point smoothing algorithm used for the data they published removed the last 10 points,whichoverlapswithourwatercolumndata,sowesmoothedtheoriginaldatawithamoving boxcar approach to retain the tail of these data. We applied a length 7 boxcar and the “rloess” smoothing method in MATLAB R2018B because this smoothing algorithm best fit their data from 1970-2010. The initial data, their smoothed data, and our smoothed data are displayed in Fig. S5. We applied a linear interpolation to this smoothed data to estimate the values at the same time- points as the water column data and extrapolated with the same linear method from 2010 to 2012. Our converted data are presented in Fig. S5. This extrapolation allowed us to use four points, ratherthanthree, forcomparisonwiththePescaderobasindata. Wedeterminedaconversionfrom these interpolated Pescadero basin points to the maximum fixed nitrogen loss samples before 2016 using singular value decomposition [150] with the formulation in Eq. 2. These confidence intervals 61 can be directly added to the water mass nutrient definitions of the 13CW and NEPIW. Additional statistical information is provided in the Supplemental Information. MaxfixedN =m(δ 15 N interp )+b (4.2) 4.4 Fifty years of fixed nitrogen loss ODZsaredefinedasoceanicregionsthatfacilitatefixednitrogenlossthroughdenitrification,anam- mox, and other processes, due to their absence of sufficient oxygen as a terminal electron acceptor. We first examined ODZ variability by calculating the nitrogen deficit relative to phosphate concen- trations using –∆N*, then integrating that parameter across the ODZ using a previously defined method [153]. We found that the integrated fixed nitrogen loss in this cross-section of the ETNP ODZ has increased almost monotonically from 1972-2019, with slight decreases in 1994 and 2018 (Fig. 2a). The upper oxycline of the ODZ contributes negligibly to the total fixed nitrogen loss (Fig. 2b), whereas the core ODZ and the deep ODZ share similar contributions to the total fixed nitrogen loss (Fig. 2c). The integrated fixed nitrogen loss in 2019 is 30% larger than what was measured in 1994, and this intensification was caused by an already intense core ODZ combined with a dramatic increase in deep ODZ strength after 2016. The magnitude of this post-2016 deep ODZ intensification is not observed in the previous 40 years and may even be unprecedented. 62 Figure 4.2: a-c) Integrated –N* values for each cruise, subset between the specified potential density ranges such that (a) is the total ODZ, (b) is the upper oxycline, and (c) depicts both the core ODZ and the deep ODZ, which contains the deeper oxycline. Both 2016 cruises are presented, and SKQ201617S is the systematically low point not connected by a line on each plot, also labeled as “SKQ”. (d) depicts the maximum fixed N lost, measured on each cruise and calculated by eOMP, within the potential density ranges in the integrations in (c). Error bars correspond to standarddeviationinthemaximum10%ofdata. Thedepthofthe13CWisplottedontherighty- axis. (e) depicts the mean oxygen concentration measured between 100-400 m on CalCOFI cruises for southern California, and error bars correspond to the seasonal standard deviation. (f) depicts normalized particulate organic nitrogen isotopes measured and published in Deutsch et al. (2014) for comparison. For robustness, we compared this integrated fixed nitrogen loss, calculated via –N* (Eq. 1), againstthemaximumfixednitrogenloss,calculatedviaextendedoptimummultiparameteranalysis (eOMP, Eq. S15-17). The mean of the maximum 10% fixed nitrogen loss within samples measured on each cruise (Fig. 2d) represents the ODZ strength, and it ranged from 9.43±0.65 to 11.83±0.35 µ mol kg − 1 in the core and 7.23±0.16 to 9.74±0.48 µ mol kg − 1 in the deep ODZ. We demonstrated that this ODZ strength correlates with the integrated fixed nitrogen loss (Fig. 2c,d), and we visualized cross-sections of the ODZ over time (Fig. 3). The strengthening of the core ODZ (σ θ 26.2-26.8 kg m − 3 ) presented in Fig. 2(a, d) is linked to shallowing of the ODZ, most obvious in the trend from 1994 to 2012. Deepening of the ODZ past σ θ 26.8 kg m − 3 is responsible for the rapid intensification in the deep ODZ observed in 2018 and 2019. This deepening does not follow the same temporal trend as the shallowing of the core ODZ, suggesting that these processes have different causes. 63 Figure 4.3: Cross-sections of fixed nitrogen loss as calculated using eOMP in the ODZ on the 110 ºW line for each year except 1972. We apply a cutoff at 8 µ mol kg − 1 to visualize the spatial coverage of fixed nitrogen loss, rather than this value being a true definition. Previous research suggests that water masses from the ETNP ODZ are responsible for 81% of the deoxygenation observed in the California Current System (CCS) [178]. The 110 ºW line studied in this 50-year time series contains the location for the Pacific Equatorial endmember of theCCS[3,4], andusingourdata, weconfirmthattheanaerobicstrengthoftheETNPODZ(Fig. 2d) correlates with the mean oxygen concentration of the southern, subsurface CCS (Fig. 2e). We also confirm that a previous record of nitrogen isotopes from sediment cores, replotted in Fig. 2f, correlated with ODZ strength in the water column (Fig. 2d). Deutsch et al. (2014) initially used this isotopic record to argue for changes in ODZ size (Deutsch et al., 2014), but since it reflects the extent of denitrifying processes, it also represents the maximum fixed nitrogen loss. Therefore, we interpreted this record of sediment nitrogen isotopes as a record of ODZ strength. Deutsch et al. (2014) also found that this sedimentary nitrogen isotope record correlated with the depth of the 13 ºC isotherm, which is a defining property of the 13 ºC water mass (13CW) in this region [178]. We discovered a strong correlation between the depth of this water mass and the strength of the ODZ core (Fig. 2d) and an even stronger correlation between the 13CW depth and the mean oxygen concentration in the southern, subsurface CCS. Since the 13CW depth sets the thermocline depth in the ETNP, we suggest that shoaling of the 13CW leads to higher 64 organiccarbonfluxintothesubsurfacewatersoftheETNP,whichfuelsbothanaerobicandaerobic metabolisms, causing the results observed in Fig. 2d-e. Therefore, the depth of the 13CW plays an important role on remineralization strength across the ETNP. Figure 4.4: NO − 3 - PO 3– 4 plot highlighting NO – 2 concentrations in the secondary nitrite maximum. The arrow depicts the stoichiometry of anaerobic remineralization using in water mass analysis, and the points are the reference nutrient concentrations for the 13CW and NEPIW. Previousresearchonthe13CWalsofoundnitriteaccumulationincreasedbelow13CWhotspots, which were introduced by westward-propagating eddies [1]. We examined the controls on nitrite accumulation in this dataset by comparing nitrite concentrations versus the depth of the sample as well as the amount of accumulated fixed nitrogen loss. In Fig. 4, sample points become shallower as the move down and to the left, whereas the concentration of fixed nitrogen lost accumulates as they move down and to the right along the given arrow. ForthedeepestcoreODZsamples,seenaround2.75µ molkg − 1 PO 3– 4 ,nitriteaccumulationonly occurs for samples with significant fixed nitrogen loss. As sampling shallows but the magnitude of fixed nitrogen loss remains the same, as seen by moving down and to the left perpendicular to the given arrow, the nitrite concentration increases. Also, it requires less fixed nitrogen loss to 65 facilitate nitrite accumulation. The highest nitrite accumulation always occurs at the maximum fixednitrogenlossforeachpotentialdensitysurface, andthisismostnoticeableatthe26.3kgm − 3 isopycnal, where the 13CW is located. Shallower samples are farther from the NEPIW core, which is suboxic at these depths, and therefore these samples are exposed to less oxygen. In addition, these samples are closer to the surface and likely have more availability for particulate organic carbon. Based on the impact of the 13CW on basin-wide ODZ strength and mesoscale nitrite accumulation, we propose that the depth and relative contribution of the 13CW act as controlling variables on nitrate reduction within the ETNP ODZ across broad spatial scales. 4.5 Observed fixed nitrogen loss likely occurred locally While the deep ETNP ODZ clearly has strengthened in 2018 and 2019 relative to the past 40 years, the location where this strengthening occurred is not clear because this signal could have been advected from elsewhere in the ODZ. The depth of the oxygen deficient layer varies across the ETNP, where it tends to be shallower in the north and deeper near the Costa Rica Dome [146]. Evans et al. (2020) suggested that the ETNP ODZ manifests due to the presence of two old water masses, the 13CW and NEPIW. Both of these water masses are oxic when they enter the region, however, respiration within the shadow zone of the ETNP forces them into anoxia. Literature on the pathways for the 13CW and the NEPIW into the ETNP ODZ closely match particle backtracking via ROMS simulation [7]. This literature reveals that the 13CW is advected into this region by the Northern Subsurface Countercurrent in the southeast, whereas the upper partoftheNEPIWisinjectedintothisregiononitswesternboundarybyTsuchiyajets(references in Evans, Boles et al., (2020)). As these water masses move away from the locations where they are introduced into the ETNP, they become anoxic and the spatial distribution of oxygen deficient waters depends on the relative distance from these entry points. 66 Figure 4.5: a) Map depicting the potential density of the ODZ core, as defined as the midpoint between the top and bottom of the ODZ using the atlas developed by Kwiecinski & Babbin (2021). This map matches the blue inset in (b), and (b) also illustrates the boundaries for the ROMS simulation performed by Margolskee et al. (2019). (c) and (d) reproduce the results of Margolskee et al. (2019), where the sections represent the locations where water enters the ODZ on the northwest boundary (c) and the southern boundary (d). The ROMS particle backtracking results from Margolskee et al. (2019) are reproduced in Fig. 5c,d and they demonstrate that (c) the NEPIW enters from the west around 12-15 ºN at 26.8 kg m − 3 , whereas the 13CW enters from the south primarily at -90 ºE at 26.2-26.3 kg m − 3 , though its entry deepens and continues west to -140 ºE. A high-resolution atlas of oxygen deficient waters, obtainedusing1mbinneddata[145],illustrateshowtheentrypointsofthesewatermassesinfluence thepotentialdensitywheretheoxygendeficientlayeriscentered(Fig. 5a). Inthesoutheastcorner of Fig. 5a where the 13CW enters at 26.2 kg m − 3 , the core oxygen deficient layer is quite deep, around 26.8 kg m − 3 . On the west side of the ODZ where the NEPIW enters at 26.8 kg m − 3 , the 67 oxygen deficient layer becomes shallower and centered closer to the 13CW. These results indicate that the circulation pathways of these water masses dictate the depths where the oxygen deficient layer forms. The locations of ODZ intensification during this time series differ for the core versus the deep ODZ (Fig. 3). In 2007, 2012, and 2019, the core ODZ is stronger at and south of 18 ºN, whereas in 2018 and 2019, the deep ODZ is stronger at and north of 18 ºN. Comparing these locations again the 110 ºW line on Fig. 5a, the oxygen deficient layer is centered around 26.7 kg m − 3 at 18 ºN and deepens closer to Baja. On the other hand, the oxygen deficient layer gets far shallower south on the 110 ºW line. This distribution of the oxygen deficient layer matches the strengthening of the ODZ (Fig. 3). Therefore, we propose that the approximately 2.4 µ mol kg − 1 intensification of the core ODZ occurred between 14-18 ºN whereas the approximately 2.5 µ mol kg − 1 intensification of the deep ODZ occurred between 18-22 ºN. These locations match the results of Fig. 3. This conclusion is notable because the results deviate from the conventional trend that deeper oxygen deficient layer occur farther south in the ETNP, and thus the fixed nitrogen loss we observed was likely advected to the north. While this trend is true and can be seen in Fig. 5a, the intensification that we observed actually occurred closer to Baja. 4.6 Natural variability of the ETNP ODZ SedimentnitrogenisotoperecordscanilluminatehistoricaltrendsinthestrengthofthecoreETNP ODZ. We converted the 1,200-year long Pescadero sediment core nitrogen isotope record (Tems et al., 2016) to equivalent water column data by fitting overlapping nitrogen isotope data to the maximum fixed nitrogen loss in the core ODZ, as plotted in Fig. 4a. We then extrapolated this fit totheentireisotopicrecord,whichprovidedarecordofthemaximumfixednitrogenlossinthecore ODZoverthepast1,200years. Wedeterminedthefrequencydistributionofthisproxynitrogenloss (Fig. 6b) and applied a 99% confidence interval to identify thresholds for virtually certain climate 68 changeforcings[183]ontheETNPODZ.Thisdatatreatmentindicatesthatthenaturalvariability over 1,200 years of the core ETNP ODZ on the 110 ºW line is 9.2-12.5 µ mol kg − 1 of fixed nitrogen lossfromourreferenceeOMPwatermassdefinitions. Tofacilitatetheuseofthisconfidenceinterval by future scientists, we present it as a region on a nitrate-phosphate plot of ETNP ODZ data from the 110 ºW line, which removes the eOMP water mass definitions. The upper left and lower right sides of the parallelogram represent the 99% confidence interval for the maximum fixed nitrogen loss, as plotted in Fig. 4b. Should the maximum 10% of nutrients sampled on the 110 ºW line between σ θ 26.2-26.8 kg m − 3 exceed the lower right side of that parallelogram, we can be virtually certain that climate change is responsible for ETNP ODZ intensification. Overall, this analysis demonstrates that the ETNP ODZ’s current conditions are near the limit of its natural variability and it could exceed its historical variability within the next decade. Figure 4.6: Figure 6. a) δ 15 N-PON Pescadero core data replotted from Tems et al. (2016) with the maximum fixed N loss from the core ODZ from the relevant years overlaid as blue diamonds. b) Points depicting the maximum fixed N loss from the core ODZ and a histogram of δ 15 N-PON PescaderocoredatafromTemsetal. (2016)convertedtomaximumfixednitrogenlossbasedonthe comparison in (a). This figure includes the 99% confidence intervals of this distribution in dashed magenta lines. Error bars here correspond to error bars in Fig. 2. c) NO – 3 - PO 3– 4 plot of data from the eight cruises, where the magenta box corresponds to the 99% confidence interval determined in (a). 69 Our findings reveal that the strength of the ETNP ODZ, as characterized along the 110 ºW line, decreased slightly from 1972 to 1994 then nearly monotonically increased by 30% in 2019. We subdivided the ODZ into the upper oxycline, the core ODZ, and the deep ODZ and found that the core and deep ODZ contribute similarly to the integrated fixed nitrogen loss. The deep ODZ intensified significantly after 2016, which is unprecedented throughout this 50-year time series and contributes to the strongest ODZ in 2019. Using an extended optimum multiparameter analysis, we confirmed that the maximum fixed nitrogen loss in the core ODZ correlates with sediment core nitrogen isotopes and the mean oxygen in the southern, subsurface California Current System. Deutsch et al. (2014) used this nitrogen isotope record to argue that the ETNP ODZ contracted before its recent expansion in 1994-2010. With the sediment record from the Pescadero basin, we estimatedthatthenaturalvariabilityofthecoreODZfixednitrogenlossis9.2-12.5 µ molkg − 1 (99% confidence interval). Only two events around 1230 and 1400 CE possessed the rapid strengthening observed in the past 30 years, and both events were followed by ODZ weakening. It remains to be seen if the current event will follow this pattern, and should the core ODZ strength continue increasing, it will exceed historical precedent within the next decade. While this analysis relies on only four measurements, it provides a first estimate at the natural variability of fixed nitrogen loss in the ETNP ODZ. More recent sediment data could extend this comparison from four to seven points, and the supplemental information in this paper includes all the data and code to improve our calculations should this data become available. Fig. 4c depicts the region of natural variability in nitrate-phosphate space for reference, and these values are also included in the supplemental information. In addition, this analysis of natural variability focused only on the core ODZ and did not include the deep ODZ, which also became 30% stronger from 2016 to 2019. A sediment core from a location that records processes in the deep ODZ would be useful to address this topic through a similar analysis. 70 4.7 Conclusions Expansion of Oxygen Deficient Zones is a globally relevant issue due to their significant role in biogeochemical cycling and influence on ecosystem biogeography. Prokaryotic metabolisms drive most of biogeochemical cycling in ODZ regions. These metabolisms employ electron donors and acceptors other than carbon and oxygen such as trace metals, sulfur, and particularly nitrogen via N 2 gas production as well as N 2 O cycling. N 2 O is a potent greenhouse gas, while N 2 production influences the marine fixed nitrogen inventory. In vast areas of the ocean, the availability of fixed- nitrogen limits primary production, and the balance between N 2 fixation and production regulates this inventory. A strengthening ODZ could result in increased denitrification, lowered primary production and therefore a diminished biological carbon pump. This nearly 50-year time series is, to the best of our knowledge, the longest time series from an ODZinscientificliterature, and itprovidesvaluable insightintochangesthathave occurredwithin the ETNP ODZ. Most importantly, we generated a first estimate for the natural variability of this ODZ. This result indicates that the core ODZ has only been this strong twice in the past 1,200 years, but we cannot yet attribute this recent ODZ intensification to climate change. Nevertheless, correlations between the 13CW water mass, the strength of the ETNP ODZ, nitrite accumulation in the ETNP ODZ, and the amount of deoxygenation in the southern CCS reveal that the 13CW significantly impacts ETNP processes. Due to the slight lag between core ODZ strength and CCS oxygen concentrations, we suggest that real time monitoring of the 13CW depth in the ETNP ODZ, potentially with Argo floats, could forecast deoxygenation in the CCS. This monitoring, as well as confidence intervals for natural variability, are crucial for future oceanographers and geoscientists as we monitor, forecast, and respond to climate change and its consequences on ocean biogeochemistry, ecosystem health, and fishery production. 71 4.8 Acknowledgments Sampling cruises on the 110 ºW line were funded by NSF OCE-1636332, DEB-1542240, OCE- 1046017,OCE-1029951,andOCE-1657958. NEwasfundedbyOCE-2023708duringtheprocessing and writing of this manuscript. JT was funded to work on this project by a fellowship through the UniversityofSouthernCaliforniaWrigleyInstituteforEnvironmentalStudies. Wegreatlyappreci- ate the efforts required in funding, coordinating, and measuring data in the WOCE, CLIVAR, and CalCOFI programs, as well as the work required to create consensus nutrient concentrations in the GLODAP program and the effort that went into acquiring these extensive sediment core records. We thank Mattias Tomczak and Joseph Karstensen for writing the MATLAB omp2 package used forconductingthiswatermassanalysis,aswellasDavidGlover,WilliamJenkins,andScottDoney for documenting several statistical techniques used in this analysis in their book Modeling Meth- ods for Marine Science. We would like to acknowledge Isaac Schroeder for sharing CalCOFI data updated to 2019, as compared to data circa 2018 as used in Evans et al. (2020). We would like to acknowledge Caitlin Tems for sharing the Pescadero and Santa Monica basin particulate organic nitrogen isotope core record data. 4.9 Open Research Almost all the data products used in this analysis are stored in Zenodo at https://doi.org/ 10.5281/zenodo.6519188, and almost all the code used in this analysis is stored at in Zenodo at https://doi.org/10.5281/zenodo.6519316. Data and code used to calculate the depth of the 13CW using 1 m binned data, as well as to plot the ODZ fraction in Fig. 1, is stored at https://github.com/NatalyaEvans/ETNP ODZ time series expanded. All code is well- commented to facilitate future use and the data products that are specifically plotted in this paper are saved as separate, labeled files for convenience. In addition to these repositories, the CalCOFI 72 dataset can be found at www.calcofi.org. While the MATLAB code for the water mass analysis and the TEOS-10 conversions were acquired at https://www.mathworks.com/matlabcentral/fi leexchange/1334-omp-analysis and http://www.teos-10.org/, appropriately, they have also been uploaded to the same Zenodo repository to enable other researchers. The authors declare no conflicts of interest. 73 Chapter 5 The role of seasonal hypoxia and benthic boundary layer exchange on margin-derived iron cycling 5.1 Introduction Iron is a crucial element for biological processes and can limit primary productivity, especially in nutrient-richecosystems[184]. Thislimitationisparticularlysignificantincoastalregionswithhigh productivity. The continental margin acts as one of the primary sources of Fe to these near-shore waters,andtheformofironinfluencesitsfateintheocean. MostFesourcesintroduceFeasFe(III), which is not stable in oxic seawater. Aqueous Fe(III) quickly precipitates to Fe (oxy)hydroxides that sink out of the water column. Fe(III) can also be bound to organic ligands, keeping the Fe dissolved and increasing its lifetime in the water column [185]. Underanoxicconditions, Fe(III)canbereducedtoFe(II)bymicroorganismsthroughdissimila- tory iron reduction [186]. Through reductive dissolution of Fe(III) (oxy)hydroxides, Fe(II) remains aqueous and can enter oxic waters, where it persists until it is re-oxidized to Fe(III). The reduc- tive dissolution of iron within shelf sediments and its release into the overlying waters acts as the first two steps of the shelf-to-basin iron shuttle [187, 188]. The shelf-to-basin iron shuttle has been proposedasoneoftheprimarymechanismsforFetransportoffshorefromcontinentalmargins[67]. In the shelf-to-basin shuttle, eddy diffusion mixes Fe(II) produced by reductive dissolution into the water column, and mesoscale features can transport this Fe(II) off the continental margin as it oxidizes [189, 190]. In Oxygen Deficient Zones, where the dominant metabolisms in the water column reduce nitrogen compounds[20, 191]. This off-shelf Fe(II) precipitates to Fe(III) and settles 74 on the deep continental slope, and from there it is mobilized and forms a deep plume of dissolved Fe(III) [19]. These deep plumes have been observed in all three oceanic Oxygen Deficient Zones: The ETNP, ETSP, and the Arabian Sea [20, 21, 22] as well as highly productive Oxygen Minimum Zones [192, 193], though their appearance is not predicted by classical geochemical models [19]. The observations of deep Fe plumes have prompted renewed interest in off-shelf Fe transport. Mechanistic and modeling studies have suggested the importance of nonreductive dissolution of iron [194] and its transport via colloids [195]. In coastal waters, both resuspension of iron particles [196] as well as reductive dissolution have been identified as sources of iron [197, 23]. Studies into Fe flux from reductive dissolution have parameterized it as a function of both bottom water O 2 concentrations as well as carbon oxidation rate [198, 199, 23]. More recent parameterization of Fe in a regional model for the United States west coast has shifted to using only the bottom water O 2 concentration [200]. Studies into the reductive dissolution of Fe have characterized the United States west coast thoroughly, but have used only a single method, benthic chambers, for these studies [201]. In addition, these studies provide excellent data on benthic fluxes but rarely link these benthic flux measurements to water column Fe concentrations. Other sources of Fe to the ocean include dust deposition and river inputs. Dust deposition is a major source of Fe to the open ocean; however, dust reaching the Pacific Ocean originates primarily from Asia [202]. Therefore, this source of Fe is less significant in Oregon coastal waters, the region of focus for this study. Rivers are also a source of Fe to the ocean, and off the coast of Oregon,theColumbiaRiverprovidessignificantfreshwaterinputeventhoughstrongFescavenging occurs across the salinity gradient to the ocean during downwelling conditions. During upwelling conditions, waters with low O 2 and elevated Fe (20-50 nM) become entrained in this plume, which facilitates their transport [203]. The Umpqua River is a smaller river than the Columbia River, but it also acts as a source of particulate Fe to these coastal water [204]. Studies on the west coast 75 have found that sediments on the continental margin act as the primary source of Fe to coastal waters [196]. CoastalOregonbottomwaterscontainhighconcentrationsofbothFe(III)andFe(II).Aprevious study found 10-23 nM Fe(III) and 7-27 nM Fe(II) in bottom waters across the Oregon continental shelfduringsummerhypoxia[205]. Oregoncoastalwatersarepronetosummerhypoxiadueinpart toupwelling[206],withhypoxiceventsbecominglongerandmorepronouncedoverthe21stcentury [207]. A combination of low O 2 waters upwelling onto the continental shelf, retention of these waters, and respiration on these organic matter-laden continental shelves leads to these hypoxic conditions[208]. Shelfwidthsignificantlyinfluencesorganiccarbonandparticulateironenrichment in coastal sediments due to the amount of river flux off Oregon. Wider shelves catch more material duringpeakwinterriverfluxes,andthismaterialisthenremineralizedduringsummerhypoxia[24]. Analysis of sediment on the Heceta Bank, a particularly wide section of shelf that retains material fromtheUmpquaRiver, revealedanextremelyhighcorrelationbetweenreactiveironmineralsand the percent of sediment organic matter [25]. With the recent and broad interest in continental margin Fe sources, and this knowledge of Fe supply to the continental margin in Oregon, we were particularly interested in analyzing Fe accumulation, speciation, and transport across the Oregon shelf. OregonprovidesanexcellentstudysitefortheFecyclenotonlybecauseofthepreviousresearch there but also its hydrographic features. Oregon is near the end of the California Undercurrent, which together with upwelling supplies nutrient-rich but O 2 -poor water [4]. As the undercurrent moves along the continental margin, elevated remineralization rates continue to lower the oxygen within these waters [1, 48], such that these waters have especially low O 2 once they reach Oregon. AnoxiawasobservedofftheOregoncoastin2006, andthetypicalbenthiccommunitywasreplaced with microbial mats of sulfide oxidizing bacteria [209]. In any year as summer hypoxia intensifies, 76 the strength of reductive dissolution will increase, likely increasing the magnitude of the regional continental margin Fe source. The switch from hypoxic to anoxic respiration has major ramifications for a highly productive ecosystem such as the northern California Current System. In the summer of 2002, shoaling of hypoxic waters onto the shallow shelf damaged Oregon’s Dungeness crab population, a fishery worth $50 million annually [210]. Anoxic respiration leads to significant community restructuring as trophic energy transfer switches from mobile predators to microbial communities [31]. This metabolic change will also significantly magnify the flux of Fe from this continental margin into the Northeast Pacific Ocean, potentially causing positive feedback to ocean deoxygenation due to increasedsurfaceproductivityandparticleexport. Understandingthecurrentprocessescontrolling the continental margin Fe source, such as its speciation, accumulation, and export, may improve our ability to predict future ecosystem conditions in this region. 5.2 Materials and Methods 5.2.1 Sampling description For this study, water samples from the Oregon continental margin were from the R/V Oceanus during cruise OC2107A which was conducted from July to August 2021 (Fig. 1). Stations 32-33 sampled the water column in the same locations as the Umpqua River stations in Severmann et al. (2010), and stations 25-29 were on the well-characterized Newport Hydrographic Line (Adams et al., 2019). Station 20 was located at the Columbia River depocenter (Nittrouer et al., 1979), whereas station 34 was located at the Umpqua River depocenter. 77 Figure 5.1: a) Sampling map for OC2107A stations with bathymetry shaded in blue. A solid black line indicates the coastline. b) Same coastline and bathymetry, but the measured sediment mud content (% silt fraction) from the usSEABED database is presented. Oxygen and nutrient samples were collected using a standard Seabird CTD-rosette equipped with10LNiskinbottlesandaSBE43sensor. Ironsampleswerecollectedprimarilyusing5LTeflon coated external spring Niskin-type bottles (Ocean Test Equipment) with a stand-alone trace metal clean rosette (Sea-Bird Electronics) tripped by an autofire module continuously during upcasts using a separate winch system. These bottles were hand-carried to a clean lab space constructed from polyethylene plastic sheeting with HEPA filters to prevent contamination. The trace metal clean rosette lacked an altimeter and therefore could not sample close to the seafloor, so a few near-bottom Fe(II) samples were collected using the traditional CTD-rosette NiskinbottlesinhighFe(II)bottomwaters. WhencollectingFe(II)samplesfromtheCTDNiskins, we overlapped the sampled depths with the trace metal clean depth profile to enable comparison betweenthesetwomethods. Atstations19.5and33,abenthicboundarygradientsamplermounted on an aluminum tripod frame was deployed using the ship’s 9/16” trawl wire to collect samples 78 within 1.5 m of the seafloor. This ”lander” held a syringe sampler designed to draw up to 18 samples simultaneously from known heights above the seafloor [211] and was triggered using a timed burn-wire (Chace et al., in prep). Sampling syringes were flushed with N 2 -purged water and kept in a continuously purging bath for at least 6 hours prior to sampling to reduce oxygen contamination. While the syringe sampler is designed to flush sample lines using in situ fluid prior to drawing samples, all sampling lines and syringes were further flushed with N 2 -purged deionized water to limit diffusion from walls of sample lines and syringes. 5.2.2 Chemical measurements All Fe samples were filtered using Pall Gelman Supor 0.45 µ m polyethersulfone filters with ei- ther Millipore Swinnex polypropylene 25 mm filter holders or Advantec-MFS type PP47 47 mm polypropylene inline filter holders, as recommended by the GEOTRACES Cookbook version 3.0, 2017. N 2 gas was used to pressurize the Go-Flo bottles during filtration. Filtered fractions for Fe(II) measurement were transferred to a syringe using a 3-way Luer-lock adaptor to prevent the introduction of air, then measured within 2 hours of sample collecting. This offset between sam- pling and measurement time is used to correct for the Fe(II) concentration at sampling. Fe(II) was measured via chemiluminescence with luminol, consisting of a continuous 2 mL min − 1 flow of sample and luminol solution at a 1:1 ratio into a standard quartz flow cell with a Hamamatsu HC135 photon counter via a peristaltic pump. The signal value for each sample was determined as the mean of the signal for 30 seconds (n=50), once the signal reached a plateau. Since luminol chemiluminescence is not selective for Fe(II), each sample was also treated with diethylenetriamine pentaacetate(DTPA),aselectiveFe(II)chelatorusedasamaskingligand. TheDTPA-treatedsam- ples, with a final concentration of 0.5 mM DTPA, serve to correct for chemiluminescence caused by compounds besides Fe(II). DTPA-treated samples were processed with the exact same analytical 79 train as untreated Fe(II) samples. The order of sample measurement was randomized to prevent systematic error due to sample oxidation. Fe(II) calibration was performed each time the system was powered on and calibration curves consisted of six or more concentrations within the appropriate working range of the measured Fe(II). These concentrations were created by spiking a seawater solution that aged for at least 24 hoursinanamberbottlewithanappropriatevolumeof1 µ MFe(II)workingstockinmilliQwater. This working stock was created fresh before calibration from a 0.01 M Fe(II) standard solution at pH 2, prepared monthly by dissolving Optima grade hydrochloric acid (Fisher) and ammonium ferroussulfate(Fluka)inmilliQwater. Luminolsolutionswerepreparedfirstusingastocksolution with 0.796 g of sodium luminol (Sigma), 250 mL of Optima grade ammonium hydroxide (Fisher), approximately 45 mL of Optima grade hydrochloric acid (Fisher), and milliQ water. The working luminol reagent was produced by diluting this stock to one-fourth its starting concentration, then heating it at 50 ºC for 9-12 hours. The DTPA solution was created by diluting 2.4584 g of DTPA (Millipore-Sigma) with sodium hydroxide (Aldrich, trace metal basis) in milliQ water for a final concentrationof50mMDTPAand200mMNaOH.Additionaldetailsaboutthismethodincluding reagent preparation, instrumental settings, and procedures have been previously reported [212]. The notable difference between this sample processing and Bolster et al. (2018) was that Fe(II) samples were not preserved using 3-(N-morpholino)propanesulfonic acid in this study, which lowers sample pH. Lowering the pH could have shifted the equilibrium speciation to favor Fe(II), and sample pH was not adjusted, unlike previous Fe(II) measurements in this region [205]. Many of the Fe(II) samples were also measured for pH using an Oakton pH 150 portable meter kit. TotaldissolvedFe(dFe)sampleswerecollectedinto500mLLDPEbottles. Withinonemonthof collection,thesesampleswereacidifiedtoapHofapproximately1.7byadding1mLofhydrochloric acid (Optima, Fisher Scientific) per 500 mL of seawater. Measurement samples were prepared in 80 triplicate using the seaFAST-pico (Elemental Scientific) offline method, where 10 mL of acidified sample was added to a Nobias resin chelation column then eluted with ultra-pure, distilled 5% nitric acid (Optima, Fisher Scientific) for a final volume of 0.5 mL. This method is similar to the method described in [213] and [214]. Fe concentrations were quantified with a Finnegan Element 2 (Thermo Scientific) Inductively Coupled Plasma-Mass Spectrometer and Apex desolvation system in medium resolution. A 1 ppb indium eluent is used during sample preparation as an internal standard to correct for instrument drift. Instrument optimization and tuning are performed daily with a 1 ppb indium and uranium tuning solution. Procedural blanks were prepared in triplicate from pH 1.7 hydrochloric acid to correct for contamination during acidification of samples. Sample concentration of Fe was determined via isotopic dilution method [215], in which acidified samples were spiked with a multi-element standard, including 57Fe, enriching it over the natural isotopic abundances. Oxygenconcentrationsfromthebenthicboundarygradientsamplerweremeasuredlessthanan hour from recovery using a Presens TX-3 microoptode in a flow cell fitted to the sampling syringe outlets. Between each O 2 measurement, brief flushes of nitrogen gas were also used to check for calibration shifts and remove contamination between samples. The order of sample measurement was randomized to reduce the influence of any contamination post-recovery. 5.2.3 Computational methods MATLAB R2018B [75] was used to read in and average signal values for Fe(II) measurement. The calibration curve for Fe(II) using this method follows a second-order polynomial, so the “nllsqr” function [150] was used to fit this data. Sample concentrations were calculated by first subtracting the DTPA signal then applying the “vpasolve” command with the given calibration curve. The uncertainty in measured Fe(II) concentrations was calculated using the standard error of the signal 81 plateau(n=50). ForFe(II)samplesmeasuredinduplicate,theuncertaintyconsistsofthisstandard error and the standard deviation of the two measurements added in quadrature. No samples collected on this cruise were anoxic, so all Fe(II) samples were oxidizing during retrieval, filtration, and measurement. For samples with greater than 0.2 nM measured Fe(II), we also calculated the concentrations of Fe(II) at the time of sampling. Measured pH values were used to calculate the Fe(II) oxidation rates. Stations 19.5 and 20 lacked pH measurements, as did benthic lander samples, so the pH was approximated using a correlation between pH and O 2 . This pH approximation is described further in the supplemental text 3. Details about data processing for Fe(II) measurement as well as the sampled Fe(II) are in the “computational methods” section. Sampled Fe(II) concentrations were calculated by integrating the Fe(II) oxidation rate between sampling time and measurement time. The rate law for Fe(II) oxidation is provided in Eq. 1. However, the pseudo-first order assumption specified in Eq. 2 was used to derive the rate law in Eq. 3andanempiricalrelationshipfortherateconstantinEq. 4[216]. Thisempiricalrelationship for the rate constant was calculated for the O 2 concentration at saturation with the atmosphere, whereas our samples were hypoxic. Therefore, we calculated the O 2 saturation using TEOS-10 and scaled the rate constant by the difference in O 2 concentrations (Eq. 5). Uncertainty in the sampled Fe(II) concentrations was calculated by combining the uncertainty in the measured Fe(II) with the error in the measured pH. These calculations are specified explicitly in the supplemental information, and the equation to calculate the half-life is included in Eq. 6. dFe(II) dt =− k[O 2 ][Fe(II)] (5.1) k ′ =k[O 2 ] sat (5.2) dFe(II) dt =− k ′ [Fe(II)] (5.3) 82 log(k ′ )=35.627− 6.7109(pH)+0.5342(pH) 2 − 5362.6 T − 0.04406(S) 0.5 − 0.002847(S) (5.4) dFe(II) dt =− k ′ [O 2 ] [O 2 ] sat [Fe(II)] (5.5) τ 1/2 = ln(2) k ′ [O 2 ] [O 2 ]sat (5.6) Sediment properties were downloaded from the usSEABED database (https://www.usgs.gov/p rograms/cmhrp/science/usseabed), a unified, comprehensive database of seafloor properties in the United States Exclusive Economic Zone. The US9 EXT data product, consisting of discrete samples, wasusedforthisstudy. Thisdataproductconsistsofdiscretesamples, soweinterpolated the sediment mud content to our station locations to analyze its relationship with water column Fe concentrations. These sediment data were cleaned by removing any samples deeper than 500 m, west of -123.5 ºE, outside of 42-47 ºN, and missing mud percentage. Some samples within regions of interest were missing depth, so we first interpolated longitude and latitude to depth using the scatteredInterpolant command and the “natural” interpolation option in MATLAB R2018B. We interpolated the sediment mud content using the same computational set up, but we included longitude, latitude, anddepthtocreatetheinterpolationsurfaceforsedimentmudcontent. Weset a constant uncertainty for the interpolated mud content using the mean of the residuals between the interpolated and the original mud values, which is 3.3%. Scatter and section plots were generated using MATLAB R2018B, and TEOS-10 calculations [76]wereperformedinMATLABaswellwiththeIBMILOGCPLEXOptimizationStudioV12.8.0 asanoptimizer. MapswereplottedusingPython3.7.13[217]inSpyder5.1.5[218]withtheBasemap package [219], and perceptually uniform colormaps were used to improve accessibility as well as scientific accuracy [220, 221]. Bathymetry data were downloaded from the National Centers for 83 Environmental Information 3 arc-second resolution coastal relief model (https://www.ngdc.noa a.gov/mgg/coastal/crm.html). 5.3 Results 5.3.1 Cross-shelf variability in iron Dissolvedironconcentrationsincontinentalmarginbottomwatershavebeenlinkedtomanyfactors, such as oxygen concentration, carbon oxidation rate, shelf width, and sediment mud content. We analyzed Fe data at eight stations along the mid-shelf (depth<150 m) to probe the roles of these different factors. These stations are organized south-to-north in Fig. 2. Stations 33, 34, and 34.5 are the farthest south, on the wide Heceta Bank, north of where the Umpqua River outflows, and station 34 is located at the Umpqua depocenter [25]. Stations 27, 29, and 38 are near the middle of the sampling region and on a narrower section of the shelf. Stations 29 and 27 were located at and nearby the Newport Hydrographic line repeat stations NH15 and NH20, appropriately. Stations 19.5and20arethefarthestnorth,atawidersectionoftheshelfwheretheColumbiaRiveroutflows, andstation20islocatedattheColumbiaRiverdepocenter[222]. TheFeprofileswithinthesethree groups – the Heceta Bank, Newport Hydrographic Line, and Columbia River –are more similar to each other than they are between groups. The Heceta Bank stations exhibited only 25-30 µ M O 2 in their bottom waters and they had the highest dFe and Fe(II) concentrations with 40-60 nM dFe and 22-58 nM Fe(II). The Columbia River stations had similarly low O 2 , with 23 and 37 µ M, but drastically different Fe profiles. At station 19.5, dFe reached 37 nM, but Fe(II) was only 6.2 nM, whereas station 20 was lower in bothcompounds. TheNewportHydrographiclinestationshadthehighestO 2 duringthesampling campaign, ranging from 68-76 µ M, and the lowest Fe, with 5.5-25 nM dFe and 1.2-4.4 nM Fe(II). For all of the samples with more than 0.25 nM Fe(II), sampled Fe(II) reveals that oxidation was minimal between recovery and measurement. 84 Figure 5.2: Depth profiles of Fe(II), sampled Fe(II) which is corrected for oxidation between sampling and measurement, dFe, and O 2 for selected shelf stations. The upper row illustrates the Fe data, whereas the bottom row depicts O 2 . Station numbers are specified on the top of each Fe profile, and this station number applies to the O 2 profile below it. The gray region on the bottom of each profile indicates the depth of the seafloor, as estimated by the Echosounder while on station. Measured Fe(II) concentrations were so low at station 29 that sampled Fe(II) could not be calculated. pH measurements were not collected at stations 19.5 and 20 so sampled Fe(II) at these stations used estimated pH values. The comparison in Fig. 2 highlights that shelf width correlates with bottom water Fe, as the Heceta Bank is the widest and has the most Fe, whereas the Newport Hydrographic line has the narrowest shelf extent and the least Fe. This comparison also raises a few notable discrepancies. First, station 19.5 has lower O 2 than the Heceta Bank stations, but the dFe at station 19.5 is slightly lower and its Fe(II) is significantly lower than the Heceta Bank stations. The low Fe(II) here may suggest that the supply of elevated dFe at station 19.5 may originate further from this specific station or that the dFe comes from resuspended sediments, rather than a reductive source. Second,theFeprofilesforstation27neartheshelfbreakaremarkedlydifferentfromotherNewport HydrographicLinestations,withelevateddFeacrossthewatercolumnandespeciallyinthebottom waters, along with elevated Fe(II). 85 5.3.2 Benthic release of iron Reducing shelf sediments serve as the dominant source of Fe to this region, but our bottom water samples can only reach approximately 3 m from the seafloor during good sampling conditions. Nevertheless, we are interested in the release and transport of Fe from the shelf across the benthic boundary layer and into the water column. Previous studies have analyzed trace metal profiles within the benthic boundary layer [211, 223], but have not yet published a Fe(II) profile from this layer. Comparing Fe(II) with dFe in the benthic boundary layer can illuminate the contribution of reductive versus non-reductive dissolution on Fe supply as well as the extent of Fe trapping due to Fe(II) oxidation. Wecollectedhighresolutionbenthicboundarylayersamplesatstations20and33, representing both the Columbia River and Heceta Bank regions. The near-bed O 2 concentration at station 20 was twice that of station 33, and the resulting Fe(II) and dFe concentrations in the samples nearest the seafloor at station 20 were approximately half that of station 33. At both stations, the concentrations of Fe(II) and dFe were nearly equal near the seafloor, suggesting that Fe(II) was the dominant form of Fe. This evidence points to Fe being released from the sediments through reductive dissolution, which is expected based on the low oxygen concentrations measured at these locations. For several depths, we observe that measured the Fe(II) concentration actually exceeds the independently measured dFe concentration. While unexpected, these discrepancies are likely a product of measurement error or minor losses of dFe during sample storage. When deploying this lander, we expected the Fe concentration profiles to follow a smooth ex- ponential decay curve moving away from the seafloor. This type of profile is expected for diffusion- dominated system where the source is the sediment-water interface. Instead, these profiles do not 86 Figure 5.3: Fe(II), sampled Fe(II), dFe, and O 2 profiles from two benthic lander deployments on the Oregon continental margin. depict a smooth exponential decay for Fe(II) or dFe at either station. We attribute the disconti- nuities in these profiles to wave motion as well as cross-shelf advection, which would introduce Fe released elsewhere on the shelf to our sampling site. The Fe(II) profiles presented in Fig. 3 are the first profiles of Fe(II) within 1 m of the seafloor, to our best knowledge. The first study with this syringe sampler reported total Fe, as well as other parameters such as Mn, PO 3− 4 , NH + 4 , carbon, As, Hg species, and volatilizable sulfur compounds from a 7 m deep station in a bay and a 6 m deep station in a lagoon [211]. Profiles from these locations had micromolar concentrations of trace metals and smooth profiles, likely because the vertical flux was far higher than any horizontal advective fluxes in these locations. A benthic landerwithadifferentdesign, holdingfivebottlesspacedevery60cm, wasdeployedata30mdeep 87 site in the Baltic Sea [223]. Plass et al. (2022) reports commonly measured first row transition metals in nanomolar concentrations, closer to those observed in our deployments. These Fe profiles weresmootherthantheoneswereport,withonepointfallingoutsideofwhatappearstobealinear trend. The depth range covered by this sampler differs significantly from the other, and the data reported in Plass et al. (2022) likely does not reflect the logarithmic layer of the benthic boundary layer. Comparing these results indicates that the magnitude of benthic flux, cross-shelf advection, and sampling resolution all influences the results of these near-bed profiles. 5.4 Discussion 5.4.1 Controls on iron(II) concentration Recent observations of deep Fe plumes in all three ocean Oxygen Deficient Zones [20, 21, 22] as well as recent investigations into the mechanisms of off-shelf Fe transport [195, 194] have prompted interest in the conditions that enhance off-shelf Fe transport. Previous research has determined the importance of the benthic carbon oxidation rate [199] as well as bottom water oxygen concen- trations [23] on influencing benthic Fe fluxes [198]. A primary finding is that bottom water oxygen concentrations of 60-80 µ M O 2 are required for benthic Fe fluxes to be observable with benthic chamber methods [201]. Our Fe measurements enable us to connect these benthic flux estimates to their impact on the water column. Carbon oxidation rate measurements were not performed during this cruise. Previous research ontheHecetaBankcorrelatedreactiveFemineralswithhigherorganiccarboninsurfacesediments [25]. Unfortunately, the dataset presented in Roy et al. (2013) was limited to a specific region of the shelf near the mouth of the Umpqua River, including parts of the Heceta Bank. Instead, we used the usSEABED dataset and its percent mud parameter to provide a sediment proxy able to cover our sampling area. With these results, we can compare the relative impact of bottom water 88 oxygenconcentrationsandpercentmud,likelylinkedtocarbonoxidationrate,onthebottomwater Fe(II) samples (Fig. 4). Figure 5.4: a) Scatter plot of O 2 and Fe(II) measured across the continental margin, with the colorbar highlighting the depth from the seafloor. Round points indicate the sample was collected from the water column with a CTD, whereas diamonds indicate that the sample came from a benthic lander (replotted from Fig. 3). b) Scatter plot of the sediment mud content at each station versusthedeepestFe(II)samplecollectedatthatstation,withthecolorbarhighlightingthedeepest O 2 concentration. The results presented in Fig. 2 demonstrate that elevated Fe(II) concentrations were only observed in samples with low O 2 concentrations, which is expected because oxidation by O 2 serves as the primary sink for Fe(II). Almost all of the high Fe(II) samples were collected within 30 m of the seafloor, with two exceptions slightly below 40 nM Fe(II). These samples were collected at station22inasubmarinecanyonandlikelyreceivedthishighFe(II)fromcanyonwallsourcesrather than the seafloor. These results refine the O 2 threshold for Fe flux presented in Severmann et al. (2010) to be closer to 75 µ M O 2 , a conclusion that can be integrated into models that simulated Fe concentrations. We can invoke this 75 µ M O 2 threshold to explain the differences in Fe distribution between stations 20 and stations 33. At station 33, the water column within the benthic boundary layer had approximately 31 µ M O 2 and O 2 concentrations were below 60 µ M until 30 m above the seafloor. These low concentrations permitted the deepest bottle samples at station 33 to retain approximately 40 nM of Fe(II) and dFe, whereas at station 20, Fe(II) concentrations were less than 89 5 nM in the deep bottles samples while dFe was 20 nM dFe. Notably, station 20 exceeded the 75 µ M O 2 threshold for Fe(II) accumulation within 2 m from the seafloor. These results suggest that while Fe(II) is certainly being released from the seafloor across the Oregon continental shelf, when the O 2 is above 75 µ M, Fe trapping occurs, which converts Fe(II) to Fe(III) and limits dFe concentrations. The trend between O 2 and Fe(II) is far stronger than sediment mud content and Fe(II). The results depicted in Fig. 4b reveal that Fe(II) only accumulates to concentrations greater than 5 nM at locations where the sediment mud content exceeds 40%. Nevertheless, high sediment mud content does not guarantee Fe(II) accumulation due to bottom water O 2 and potentially other factors. TheseobservationsindicatethatsedimentmudcontentcanbeusedtoinformwhereFe(II) accumulationmaybeexpected, butbottomwaterO 2 isrequiredtodetermineifFe(II)willactually accumulate and persist. In addition, we cannot be confident that the higher mud content itself is fueling more Fe(II) accumulation. Higher sediment mud content may merely be correlated with a wider continental shelf and therefore longer residence times for waters upwelled onto the shelf, which also promotes Fe(II) accumulation. 5.4.2 Off-shelf iron transport The lack of O 2 in Oxygen Deficient Zones enables Fe(II) to persist far longer than it does in a hypoxic region. In the Eastern Tropical South Pacific Oxygen Deficient Zone, the half-life of Fe(II) has been estimated to be 200-2900 hours [189]. Since the Oregon continental shelf has waters that are usually oxic or hypoxic, residence times of Fe(II) will be far shorter, limiting the transport of Fe(II) as well as dFe in general. To evaluate the persistence of Fe(II) across the Oregon continental margin, we calculated the predicted half-lives of Fe(II) oxidation by O 2 for each Fe(II) sample. These half-lives reflect only the rate of Fe(II) removal, rather than its lifetime, because near-shelf samples may have significant Fe(II) fluxes. 90 Calculatingtheoxidationhalf-livesrevealsthatmosthighFe(II)sampleshavehalf-livesgreater than 20 hours. Interestingly, samples collected deeper than 600 m tend to have significantly longer half-lives, despite their low Fe(II) concentrations. These extended lifetimes likely occur due to the presence of the oxygen minimum zone, where the low O 2 concentrations, pH, and temperature all stabilize Fe(II). Our Fe(II) samples collected in the Oregon oxygen minimum zone had half-lives approaching that range (100-200 hours) with O 2 concentrations of 8-15 µ M. The long half-lives of Fe(II) in these deep waters suggests that trace amounts of Fe(II) may persist for a significant amount of time adjacent to the deep continental margin off Oregon, which is unexpected based on the low concentrations observed there. Figure5.5: ScatterplotofFe(II)concentrationscomparedtotheirhalf-lives. Samplecolorreflects the depth where that sample was collected, whereas error bars consist of the uncertainty in Fe(II) measurements and half-lives, appropriately. Deep Fe plumes have received a significant amount of attention due to their prevalence seaward of several reducing continental margins with Oxygen Deficient Zones. For deep Fe plumes to form, 91 though,Fehastobetransportedoff-shelffirst. InPacificOxygenDeficientZones,Feistransported off-shelf via Fe(II) export in thin fingers of water that intersects the continental margin. This phenomenon is centered at 26.4 kg m − 3 in the Eastern Tropical South Pacific [46] and 26.5 kg m − 3 in the Eastern Tropical North Pacific [20]. After leaving the shelf, Fe(II) in appears to support deep plume formation as it is transported to depth via particles [19]. In our study, we analyzed the Fe(II) distribution across the continental shelf to compare Fe plume formation under hypoxic shelf conditions against these previously studied anoxic conditions. Figure5.6: Sectionsofa)Fe(II)andb)O 2 ontheHecetaBank. Thesesectionsconsistofstations 31-33, where stations 32 and 33 repeat Umpqua River influenced stations from Severmann et al. (2010). Stations32-33haveFe(II)concentrationsfarhigherthanthecolorbarmaximum, asseenin Fig. 2, but the colorbar range was chosen to visualize the lower off-shelf Fe(II) concentrations. The 40 and 47 nM Fe(II) concentrations annotated on the plot indicate the maximum concentrations measuredatstations32and33,appropriately. Labeledcontourlinesrepresentthepotentialdensity. Unlike distributions off Oxygen Deficient Zones, we did not observe evidence of Fe(II) being transported off-shelf in waters that intersect with the continental shelf, between 26.5-26.6 kg m − 3 (Fig. 6). We have two possible explanations for this lack of an Fe(II) finger occurring off Heceta Bank. First, the approximately 75 µ M O 2 concentrations within these off-shelf waters were too high to allow Fe(II) to persist that far off-shelf. Second, our sampling location was close to the southern tip of the Heceta Bank, and south of this bank the seafloor deepens. While sampling off-shelf, we observed an extremely strong northward California Undercurrent (approximately 0.2 m s − 1 , Fig. S3), and this strong undercurrent would likely move material released from the 200 92 m deep Heceta Bank north, rather than directly off-shelf. We discuss the relative control of these two mechanisms on the lack of a 26.5-26.6 kg m − 3 Fe(II) finger using our analysis of the 26.8 kg m − 3 feature below. As for the 600 m Fe(II) feature, we attribute this to Fe(II) released from the continental slope and stabilized by the oxygen minimum zone. Wedidobserveoff-shelfFe(II)transportnearthe26.8kgm − 3 isopycnal,however,thisisopycnal does not shoal onto Heceta Bank. There are several potential explanations for this Fe(II) plume. A high-resolution model of Fe on the Washington coast depicts an off-shelf plume of Fe centered at 26.8 kg m − 3 [208], similar to Fig. 6. This model specifies that the plume consists of material detached from the benthic boundary layer during a downwelling event. When we sampled at this location, the upwelling strength is best described as being during a relaxation from upwelling, rather than intense downwelling (Fig. S2), which renders this mechanism a less likely cause of the Fe(II) plume. Nevertheless, previous analysis of hypoxia on the Oregon shelf found that the period of relaxation after upwelling leads to the strongest deoxygenation [224], and this phenomenon may explain why the Fe concentrations here are elevated. The second potential mechanism is that this strong undercurrent swept Fe(II) accumulating on the continental slope to the southwest and transported it to our sampling location (Barth et al., 2005). This mechanism is likely because south of Heceta Bank, seafloor deepens with a small, approximately400mshoulder. Thisdepthissimilartothedepthofthe26.8kgm − 3 Fe(II)feature, and the lack of a shallow shelf would prevent shallower Fe(II) features. We verified that this 26.8 kg m − 3 Fe(II) feature originated from the southwest by identifying northwesterly currents at that depth using the shipboard Acoustic Doppler Current Profiler (ADCP). To cross-check that this shelf shoulder could be the only source of Fe(II) to our sampling site, we calculated the Fe(II) concentration that would have accumulated in the shelf shoulder bottom water. We estimated this value using the velocities measured via ADCP and compared it against the 12 hour half-life. This 93 calculationrevealsthatbottomwatersontheshelfshoulderwouldneedtocontainapproximately3 nM Fe(II) for the observed Fe(II) concentration off Heceta Bank at 26.8 kg m − 3 . Since our off-shelf Heceta station was influenced by this region to the southeast, which lacks a 100-200 m shelf, we conclude that we did not observe a 26.5-26.6 kg m − 3 Fe(II) finger due to the lack of a nearby Fe(II) source. It is important to note that the iron distribution we observed across Heceta Bank may be a more exaggerated version of those that exist more broadly across hypoxic Eastern Boundary Upwelling Systems. Heceta Bank is a particularly large topographical feature, so it retains river- deposited organic matter and experiences elevated amounts of remineralization [225] as well as being a semi-retentive region [226]. These factors facilitate oxygen depletion, Fe accumulation, and induce off-shelf transport events linked to the California Undercurrent. Nevertheless, these results indicate that off-shelf iron transport in hypoxic regions differs significantly from Oxygen Deficient Zones due to the relatively higher oxygen concentrations in shelf waters. 5.5 Conclusions Thedataproducedbythisstudyprovideatimelyandhigh-resolutionanalysisofironaccumulation and speciation on the Oregon continental margin during summer. The seasonal hypoxia observed inthisregioncausesittobeaparticularlysignificantsourceofirontonearbyhighlyproductiveand often iron-limited ecosystems. The potential emergence of seasonal anoxia within these waters also suggests that this region could become a far larger source of iron to the North Pacific in the future. We analyzed our data within the context of many previously established relationships with iron, such as bottom water O 2 , shelf width, and sediment organic carbon to improve our mechanistic understanding of iron released from continental reducing margins. For some of our samples, we measured extremely high Fe(II) and dFe concentrations, up to 58 nM. These concentrations are far higher than a previous study in this area [205], likely due to 94 the areas where we sampled. We observed the highest Fe concentrations over Heceta Bank due to the especially low O 2 present there as well as the higher sediment mud content. Since Heceta Bank is a region where the shelf widens [24], is fed by the Umpqua river [25], and acts as a semi- retentivearea[226],thislocationwasexpectedtobeahotspotforFereleaseandoff-shelftransport. Near the seafloor of Heceta Bank, the dFe concentration was almost entirely Fe(II), suggesting an intense supply of this material to the overlying waters. On the shallow and narrower shelf at the Newport Hydrographic line, Fe concentrations were far lower and dFe concentrations dwarfed Fe(II)concentrations,suggestingthatthisdFeoriginatedfromelsewhereonthecontinentalmargin. Near the Columbia River depocenter, the dFe was higher than that observed on the Newport Hydrographic Line but less than that of Heceta Bank. This observation fits with the intermediate width shelf at the Columbia River outflow. By measuring the first high resolution profiles of Fe(II) within a meter of the seafloor, we observed 20 nM Fe(II) near the seafloor at the Columbia River depocenter; however, this Fe(II) quickly attenuated within 1 m of the seafloor. This loss of Fe(II) likely occurred due to oxidation, as O 2 concentrations in the bottom waters of the Columbia River depocenter were approximately twice that of Heceta Bank, where this rapid Fe(II) attenuation was not predicted. Comparing the measured Fe(II) versus O 2 concentrations for our samples reveals that Fe(II) does not significantly accumulateinsamplesabove75µ MO 2 andthedistancefromtheseaflooractsasaprimarycontrol on Fe(II), since the seafloor is the primary source of Fe(II) to this region. This 75 µ M O 2 threshold updatesaprevious60-80µ MO 2 rangedeterminedbySevermannetal. (2010)forbenthicFefluxes. Conversely, we did not find a robust empirical trend between bottom water Fe(II) concentrations and sediment mud content. Previous research found a strong correlation between sediment organic matter content and sediment highly reactive iron minerals [25], which we approximated using the sedimentmudcontent,butwecouldnotconfirmthatthistrendextendstodictateconcentrationsin 95 the water column. We did find that elevated Fe(II) concentrations occur only at sites with greater than 40% mud content, however, this does not guarantee Fe(II) accumulation in their bottom waters. The bottom water O 2 concentration may serve as a more important variable for Fe fluxes in this region, near locations where this relationship was first quantified. A primary goal of this study was to analyze Fe off-shelf transport in a hypoxic region. We only observed off-shelf transport of Fe(II) seaward of Heceta Bank. Unlike Pacific Oxygen Deficient Zones, we did not observe a finger of Fe(II) between 26.5-26.6 kg m − 3 , likely due to the strong California Undercurrent and our sampling location. Instead, we observed a plume of Fe(II) at the 26.8 kg m − 3 isopycnal. While there are a number of potential sources for this Fe(II) plume, we attribute it to transport by the California Undercurrent from a nearby shoulder of the Heceta Bank. The importance of topographical features and horizontal transport suggests that off-shelf Fe transport is highly spatial variable off the Oregon continental margin, though continental slope- derived Fe(II) may persist for a significant amount of time. Hypoxia is increasing in strength and duration on the Oregon continental shelf [207], and intermediate waters of the North Pacific are also experiencing deoxygenation [17]. Lower O 2 will lead to elevated Fe(II) release and persistence in the water column, increasing the role of the Oregon continental margin as an Fe source. Coastal Oregon and North Pacific Ocean waters are often iron-limited [227, 228], and this increased Fe source could act as a positive feedback to Pacific deoxygenation by increasing primary production and therefore carbon supply to hypoxic intermediate waters. Understanding and simulating continental margin Fe supply will provide helpful information for predicting the condition of the future coastal ocean. Intensive modeling and monitoring already occurs for these waters due to their economic significance [229], and an improved understanding of Fe, a critical nutrient in this region, will advance these efforts. 96 5.6 Acknowledgments We would like to acknowledge NSF OCE 2023708 for supporting this research. The development of the benthic sampler lander was supported by Oregon Sea Grant Project number: R/HBT-23- Reimers2022 with the syringe sampler (Susane) graciously provided from Ifremer. The authors declare no conflicts of interest. We appreciate the support provided by R/V Oceanus crew and marine technicians during our cruise, specifically Emily Shimada, Michael Tepper-Rasmussen, and Sabrina Taraboletti. We would also like to thank our collaborators in the Bianchi lab, specifically Daniel McCoy, Pierre Damien, and Anh Pham. 5.7 Open Research Alldataandcodeusedinthispapercanbefoundat https://github.com/NatalyaEvans/Oregon Fe2. We analyze several other data products in this paper, and we include them in this repository as well. 97 Chapter 6 More than deoxygenation: linking iodate reduction to nitrogen, iron, and sulfur chemistry in reducing regimes 6.1 Abstract The apparent simplicity of oceanic iodine speciation has motivated its use in several proxies. In subsurface, oxic waters, iodine speciation is dominated by iodate. In anoxic and hypoxic waters, iodate is reduced to iodide, and this shift in speciation implies that iodate can be used as a tracer forlowoxygenwaters. Sinceiodatecanbeincorporatedintocalciumcarbonatesbutiodidecannot, theiodine/calciumproxyhasbeenusedtoreconstructbottomwateroxygenconcentrations. Iodide concentrationshavealsobeenusedasaproxywithdissolvedironfromcontinentalmargins,because bothiodideandiron(II)arereleasedfromreducingshelfsediments. Despitetheuseofiodineinthese proxies, the processes that control iodate reduction and subsequent iodide accumulation are not well-known. We re-analyzed iodine data from the Eastern Tropical North Pacific Oxygen Deficient Zone, where we observed complete iodate depletion. We applied a novel analytical framework and determined that anaerobic respiration was not the primary cause of iodate depletion in this region. We also present new iodine data from the seasonally hypoxic Oregon continental margin, where we did not observe iodide accumulation. Contrasting these results for iodine speciation, we propose that the primary cause of iodate depletion is its reduction with sulfide slightly deeper than the sediment-water interface. The mechanism that causes iodate depletion dictates the robustness of the I/Ca ratio for bottom water oxygen reconstructions as well as iodide accumulation for iron 98 fluxes from continental margins. Understanding the controls on iodine speciation will improve the accuracy of these proxies. 6.2 Introduction The use of dissolved iodine as a proxy for reducing processes in marine waters has been motivated by the apparent simplicity of its chemistry. The mean oceanic concentration of dissolved iodine is approximately 470 nM [62, 26], and below the thermocline, iodine is almost entirely in the form of iodate [230, 231]. The other stable form of inorganic iodine in seawater is iodide, and it accumulates in surface waters as well as oxygen deficient waters [232, 70, 233, 26]. Slow rates of iodide oxidation [62, 234, 235] as well as sometimes slow rates of iodate reduction [69, 236] cause these oxidation state endmembers to be stable in oxic, euphotic seawater. Interestingly, dissolved iodine has been measured far in excess of 470 nM, and in these situations, iodide has been the primary form of dissolved iodine [63, 46, 26]. This excess iodine originates from reducing shelf sediments, where accumulated organic matter is respired and iodate is reduced, producing elevated iodideconcentrations. Thesebottomwaterswithexcessiodideconcentrationsarethentransported into the Oxygen Deficient Zones (ODZs) [1, 26]. Reducing shelf sediments were identified as the sourceofthisiodidebecausetheratioofiodinetocarboninmarineorganicmatteris10 − 4 [237,238], which is far lower than the amount required to produce this excess iodide signal. These controls on iodine have enabled its use in several proxies for reducing conditions. Both Fe(II) and excess iodide originate from reducing shelf sediments [67, 19], but iodide persists in the watercolumnwhileFe(II)oxidizesandprecipitates. ThesepropertiesmotivatedCutteretal. (2018) to use excess iodide to analyze an Fe(II) plume. A more commonly used proxy, iodine/calcium (I/Ca)ratios,relyonthefactthatiodatebutnotiodidecanreplacecarbonateinmarinecarbonates [239, 240, 241]. The I/Ca ratio has been identified as a proxy for reconstructing bottom water oxygen concentrations [242], and it has gained popularity in recent years [243, 244, 245, 246, 247]. 99 The I/Ca ratio requires iodate depletion to occur under anoxic [248] or even hypoxic conditions [249]. However,theprocessesthatreduceiodatetocreatethedepletionrequiredfortheI/Caproxy to simulate bottom water oxygen concentrations are not well-constrained. Iodate reduction occurs in low oxygen waters because the reduction potential of iodate is less favorable than oxygen [70], as seen in Fig. 1. A facultative anaerobe that reduces iodate chemo- heterotrophically has been identified [250], though iodate reduction can be a byproduct of nitrate- reducing enzymes [69, 251]. Reduced compounds produced in other anaerobic metabolisms such as nitrite and sulfide can also reduce iodide biotically [73] and abiotically [252]. Fig. 1 indicates the relative potential of different electron acceptors in seawater to provide context for these reactions. In this manuscript, we compare dissolved iodine speciation between an Oxygen Deficient Zone and a hypoxic continental shelf to examine the processes that cause iodate depletion. We focus on the Eastern Tropical North Pacific (ETNP) ODZ due to the extensive analysis of its iodine distri- bution [26], iodate reduction rates [236], and hydrographic features [1]. Water masses were found to have a significant role on the iodate distribution [236], and we apply a new water mass analysis package [13] to identify the mechanisms causing iodate depletion in this region. Our continental shelfdatacomesfromOregon,whichexperiencesseasonalhypoxia[206]. Summerhypoxiaisknown to produce extremely high Fe(II) concentrations in bottom waters, suggesting that this continental shelf can become extremely reducing [225]. Despite this high reducing power, Oregon continental shelf waters typically remain hypoxic, though occasional sulfidic events have occurred [209]. This hypoxic region provides an excellent comparison with the consistently denitrifying region of the ETNP ODZ, and this comparison elucidates the factors that cause causes iodate depletion. 100 Figure 6.1: Predicted reduction potentials for terminal electron acceptors in seawater at pH 7.5, salinity of 35 psu, and temperature of 10 ºC. This figure originates from Cutter et al. (2018) but usesdatafromStummandMorgan(1995),andeachredoxcoupleispresentedasahorizontalrange due to variability in the activities of each compound. 6.3 Methods 6.3.1 Sample collection In our manuscript, we compare a re-analysis of iodine speciation data from the Eastern Tropical North Pacific Oxygen Deficient Zone (ETNP ODZ) with new iodine speciation data from the hypoxic Oregon continental margin. The ETNP ODZ data were first presented in Moriyasu et al. (2020),andwespecificallyanalyzedatacollectedonboardtheR/V FalkorduringcruiseFK180624, which sailed from June to July 2018. The transect for FK180624 is presented in Fig. 2a. Samples were collected using a Seabird CTD equipped with twenty-four 12 L GO-FLO bottles and a SBE43 sensor for measuring oxygen concentrations. After recovery, samples were filtered using a 0.2 µ m 101 AcroPak 200 filter within two hours of sample collection. Most samples were measured within 6 hours of filtration, however, some were refrigerated and measured within 24 hours. Moriyasu et al. (2020) presents the dissolved iodide and iodate concentrations as well as the excess iodine measured on this cruise, whereas Evans et al. (2020) uses a water mass analysis to explain the spatial distribution of these iodine compounds. Figure 6.2: a) Map of stations in the ETNP ODZ during FK180624 overlaid on the fraction of ODZ conditions observed on the 26.5 kg m − 3 isopycnal. b) Map of stations on the Oregon coast during OC2107A and OC2111A cruises overlaid on seafloor depth. Oregon continental margin samples were collected on the R/V Oceanus during the OC2107A cruisefromJulytoAugust2021aswellastheOC2111AcruiseduringNovember2021. Wesampled during summer hypoxia in OC2107A, whereas OC2111A had bottom water oxygen concentrations closer to 100 µ M. The sampling locations for these two cruises are presented in Fig. 2b. Samples were collected from a Seabird CTD equipped with 10 or 12 L GO-FLO bottles and a SBE43 sensor for measuring oxygen concentrations. Samples were filtered using 0.45 µ m Corning disposable bottletopfilterswithinthreehoursofcollectionandthenfrozenforanalysisonland. ForOC2107A, this manuscript focuses primarily on stations 31-33, which were collected on the Heceta Bank. This region serves as a hotspot for deoxygenation [225] and benthic iron fluxes because it is a semi-retentive shelf [226] fed organic matter from the Umpqua River [23]. 102 DuringOC2111AatstationsMT0andMT2,wecollectediodinesamplesusingabenthicbound- ary gradient sampler mounted on an aluminum tripod frame was deployed using the ship’s 9/16” trawl wire to collect samples within 1.5 m of the seafloor. This ”lander” held a syringe sampler designed to draw up to 18 samples simultaneously from known heights above the seafloor [211] and was triggered using a timed burn-wire (Chace et al., in prep). Sampling syringes were flushed with N 2 -purged water and kept in a continuously purging bath for at least 6 hours prior to sam- pling to reduce oxygen contamination. While the syringe sampler is designed to flush sample lines using in situ fluid prior to drawing samples, all sampling lines and syringes were further flushed with N2-purged deionized water to limit diffusion from walls of sample lines and syringes. Oxygen concentrations from the benthic boundary gradient sampler were measured less than an hour from recovery using a Presens TX-3 microoptode in a flow cell fitted to the sampling syringe outlets. Between each O 2 measurement, brief flushes of nitrogen gas were also used to check for calibra- tion shifts and remove contamination between samples. The order of sample measurement was randomized to reduce the influence of any contamination post-recovery. Fe(II) was also measured following the methods described in Evans et al. (in review) for these lander samples. These near-bottom samples had high Fe(II), which lowers the potential risk of contamination. We measured Fe(II) to compare their benthic profiles, since both of these compounds have the same reducing shelf source. 6.3.2 Iodine measurement methods Iodide was quantified with cathodic square wave stripping voltammetry using a hanging mercury drop electrode and either a calomel or an Ag/AgCl reference electrode (Rue et al., 1997), adapted from Luther et al. (1988). Each 10 mL seawater sample was measured in duplicate by treating them with 150 µ L of 0.2% Triton X 100 (Sigma Aldrich – BioX grade) and purging them with argon for five minutes to avoid oxygen interference [253]. Previous studies have found that argon 103 is required for removing dissolved oxygen, as nitrogen was insufficient (Moriyasu et al. in review). Scans used a drop size of 7, deposition time of 30 sec, and 5 sec of quiet time. Scan increments were set to 2 mV with a scan range between -140 to -700 mV, and the square wave amplitude and frequency were 25 mV and 125 Hz. Each sample was measured in duplicate and their signals averaged. Calibration was performed using standard additions of potassium iodide to a seawater sample. ForOxygenDeficientZonesamples,thisstandardadditionwasperformedusingsequential addition to the same sample volume, starting with a sample that had low (100-200 nM) iodide and iterating until 500 nM iodide had been added. For Oregon samples, this standard addition was performedusingaGEOTRACESGP15samplewithlessthan20nMiodidewasusedasthesample matrix and increasing 20 nM increments were added to fresh splits from that GP15 sample until 120 nM had been added. These adaptations to the calibration protocol were required to measure the extremely low iodide concentrations in the Oregon samples. Dissolved iodate was also measured, but only for the Oxygen Deficient Zone samples. The protocol for measuring dissolved iodate was adapted from Rue et al. (1997) who adapted their method from Wong and Brewer (1977). Measurements were made on a spectrophotometer (Perkin Elmer Lambda 35) using a 10 cm quartz cuvette. One milliliter of 0.12 M sulfanilamide (Sigma Aldrich – ACS grade) in 1% sulfuric acid (Macron Fine Chemicals) was added to each 25 mL sample to prevent nitrite interference. The sample sits for 5 min, after which 1 mL of 0.12 M potassium iodide in deionized water is added to the sample to form triiodide. Absorbance of triiodide is measured at 353 nm within one minute of iodide addition. Iodate concentrations were determined by the method of standard additions using potassium iodate (Baker Analyzed – ACS Reagent). Precision, based on five replicates, was ±12 nM for a seawater sample with 286 nM iodate. For Oxygen Deficient Zone samples, total iodine was calculated by adding iodate and iodide concentrations. 104 IodatewasnotmeasuredfortheOregonsamples. Aswewereprimarilyinterestedintheamount of dissolved iodine released from the continental margin, we measured the total dissolved iodine using a similar method to Hardisty et al. (2020). This method measures total iodine as the sum of dissolved iodide, dissolved iodate, and dissolved organic iodine. Total iodine was measured by reducing these iodine species to iodide by acidifying samples to pH < 2 with hydrochloric acid (Fisher) and adding sodium bisulfite (Fisher) for a final concentration of 0.6 mM then letting the samples sit overnight. Iodide was extracted from the seawater samples by running the samples through AG1-X8 resin then eluting with 15 mL of 2.0 M nitric acid (Fisher) and 18% tetramethy- lammonium hydroxide (TMAH, Fisher). While TMAH is commonly used to stabilize dissolved iodine, it is a category 1 central nervous system toxin. Recent research has found that ammonium hydroxide performs similarly to TMAH [247], and we encourage researchers to replace TMAH with ammonium hydroxide to increase workplace safety. Once iodide was eluted in 2.0 M nitric acid and 18% TMAH, these solutions were diluted to 1/40th of this concentration and quantified on a Thermo-FisheriCAPTQInductivelyCoupledPlasma-MassSpectrometer(ICP-MS)usinginternal standards of cesium, indium, and rhodium (Inorganic Ventures). This ICP-MS was tuned before each set of measurements using Thermo-Fisher iCAP TQ tuning solution (Fisher) and calibrated for iodide using eight standards diluted from a stock iodide solution (Inorganic Ventures). 40 sam- ples were measured for total dissolved iodine, and of these, nine samples had anomalously low total dissolved iodine with less than 400 nM. These samples were removed from the analysis. We also extracted iodide to measure on the ICP-MS, which allowed us to intercompare this method with the hanging mercury drop electrode method. For this protocol, we did not reduce the samples before adding them to our columns. Instead, we added untreated seawater to the AG1-X8 resin columns to retain the iodide then eluted the iodate and organic iodine fractions using milliQ water and 0.2 M potassium nitrate rinses, then eluted the iodide with the same 15 mL of 2.0 M 105 nitric acid and 18% tetramethylammonium hydroxide solution. Due to the low concentrations of iodide, we measured these solutions with a 1/20th dilution rather than a 1/40th dilution on the ICP-MS, otherwise, this analytical chain was identical to the total iodine measurements. 6.3.3 Water mass analysis and iodate deconvolution Watermassesprovideaframeworkforanalyzingthecausesofspatialvariabilityaswellasidentify- ingprocessesrelatedtorespiration. Evansetal. (2020)foundthatthe13ºCWater(13CW),North- ern Equatorial Pacific Intermediate Water (NEPIW), and Antarctic Intermediate Water (AAIW) were the primary water masses that compose the Eastern Tropical North Pacific Oxygen Deficient Zone(ETNPODZ).Evansetal. (2020)deconvolutedETNPODZsamplesintotheirsourcewaters, but this deconvolution relied on local nutrient concentrations for their water mass definitions and therefore could not calculate respiration. With this source water deconvolution, Hardisty et al. (2020) indicated that processes besides water mass mixing control the iodate distribution of the ETNP ODZ. Unfortunately, these previous analyses could not fully identify the factors influencing the iodate distribution of the ETNP ODZ without including respiration. Using ETNP ODZ source water mass definitions from Evans et al. (2020), Evans et al. (2022) analyzed global datasets to determine the progression of these water masses into the ETNP ODZ. Thesedatasetsprovidetheconcentrationswhereeachwatermassswitchesfromaerobictoanaerobic respiration, inadditiontoprovidingthestoichiometryforanaerobicrespirationintheETNPODZ. With this information, we can improve on previous water mass analyses by deconvoluting water masses as well as the amount and type of respiration that has occurred. We applied extended optimum multiparameter analysis (eOMP) to FK180624 data using the pyompa package [13] and similarmethodsasEvansetal. (inreview). Thismethodenablesustoquantifytheamountofboth aerobicandanaerobicrespirationthathasoccurredineachsample, andthenetreactionsweuseto define these respirations are specified in Eq. 1-2. The stoichiometry for these net reactions varies 106 in the ETNP ODZ (Evans et al., in review), so we allowed the stoichiometry of these reactions to be flexible. We specify the stoichiometric ranges used for each reaction in Table S1. (CH 2 O) 106 (NH 3 ) 16 (PO 4 )+O 2 →CO 2 +NO 3 +PO 4 +H 2 O (6.1) (CH 2 O) 106 (NH 3 ) 16 (PO 4 )+NO 3 →CO 2 +N 2 +PO 4 +H 2 O (6.2) Whenwerefertotheamountofarespirationthathasoccurred,wearereferringtothecontribution of that reaction (ξ ) to the nutrient concentration as specified in Eq. 3, whereas the initial concen- trationiscalculatedfromtheendmembervalues. Evansetal. (inreview)hasadetaileddescription for performing eOMP using pyompa, but slight adjustments had to be made to accommodate the FK180624 data. We elaborate on these calculations in Supplemental Information section 3. [PO 3− 4 ] f =ξ aerobic +ξ anaerobic +[PO 3− 4 ] 0 (6.3) In Hardisty et al. (2020), the contributions of water masses to the iodate distribution were calcu- latedusingtheequationspecifiedinEq. 4andtheleast-squaresnonnegative“lsqnonneg”command in MATLAB R2018B. In our re-analysis of this data, we instead use a mixed-effect linear fit model withthe“fitlme”commandtocalculatethecontributionsofthesewatermassestotheiodatedistri- bution. This command is far more robust than the lsqnonneg command and provides uncertainties as well as statistical information. We deconvoluted the iodate distribution using the source water masses as well as the respiration pathways as specified in Eq. 5. Iodate values calculated from this deconvolution indicate the amount of iodate corresponding to that water mass or respiration. To 107 prevent the flexibility in respiration stoichiometry from interfering with the amount of respiration calculated, we standardized the amount of respiration as described in Evans et al. (in review). [IO − 3 ]=a 1 x 13CW +a 2 x NEPIW +a 3 x AAIW (6.4) [IO − 3 ]=a 1 x 13CW +a 2 x NEPIW +a 3 x AAIW +a 4 ξ aerobic +a 5 ξ anaerobic (6.5) 6.3.4 Computational methods Pyompa was performed in Google Colaboratory using the scripts provided in https://github .com/NatalyaEvans/Iodine Frontiers. After performing water mass analysis, the results wereplottedandprocessedusingMATLABR2018B[75]andvisualizedusingperceptuallyuniform colormaps [221]. TEOS-10 calculations [76] were performed in MATLAB as well with the IBM ILOG CPLEX Optimization Studio V12.8.0 as an optimizer. Linear regression comparing the measured and simulated iodate was performed using the “lsqfitma” command in MATLAB [150]. Maps were plotted using Python 3.7.13 [217] in Spyder 5.1.5 [218] with the Basemap package [219]. BathymetrydataweredownloadedfromtheNationalCentersforEnvironmentalInformation 3 arc-second resolution coastal relief model (https://www.ngdc.noaa.gov/mgg/coastal/crm.ht ml). The fraction of ODZ conditions presented in Fig. 2a and 4a depicts how consistently certain locations are found to be oxygen deficient using an atlas of oxygen deficient conditions [145]. 6.4 Results 6.4.1 Eastern Tropical North Pacific Oxygen Deficient Zone 6.4.1.1 Distribution of redox-active compounds The distributions of oxygen, dissolved inorganic iodine, iodate, and iodide measured on FK180624 [26] are presented in Fig. 3. These results demonstrate that even though oxygen concentrations 108 are extremely low throughout the ODZ between approximately 100 m to 1000 m, iodate depletion onlyoccurswithinawedgeofthiswater. Thisregionofiodatedepletioniswiderclosertothecoast and is closely mirrored by an increase in iodide concentrations. In certain locations, particularly more coastal ones, iodide concentrations exceed 500 nM, the maximum iodate observed. The total iodine distribution (Fig. 3b) reveals that additional iodide inputs occurred across this transect. We are interested in using the excess iodide released from reducing shelf sediments as a proxy for other material released from this source, specifically Fe(II). Figure6.3: Distributionsofa)oxygen,b)totaliodinedeterminedasthesumofiodateandiodide, c) iodate, and d) iodide measured on FK180624. Unfortunately, Fe(II) and iodide have not been measured on the same transect in the ETNP ODZ. Instead, we compare Fe(II) measured in the ETNP ODZ on the R/V Ronald Brown cruise RB1603,whichsailedin2016[20]. ThistransectcoveredaslightlydifferentareaoftheETNPODZ closer to the coast (Fig. 4a), however, the Fe(II) distribution (Fig. 4b) is also more pronounced closer to the coast, just like the iodide distribution (Fig. 3d). This comparison also reveals that near-complete depletion of iodate in off-shore waters, as indicated by greater than 400 nM iodide, is comparable to approximately 0.8 nM Fe(II). 109 Figure 6.4: a) Map of selected stations during RB1603 where Fe(II) was measured and (b) a section of Fe(II). 6.4.1.2 Deconvolution of iodate into source waters and respiration pathways We define our water mass nutrient endmembers at the concentrations where respiration switches from aerobic to anaerobic (Evans et al., in review). This formulation is founded on the fact that Eastern North Pacific water masses contain an approximately uniform iodate concentration of 476 ± 9 nM, and in subsurface waters, anaerobic processes are the primary cause of iodate depletion. We deconvoluted the iodate distribution measured in the ETNP ODZ on FK180624 using the formulation specified in Eq. 5, and the best fit coefficients from this deconvolution are presented in Table 1. These coefficients reveal that the starting iodate concentrations in the NEPIW and AAIW are close to the uniform 476 ± 9 nM, iodate concentration, though AAIW is slightly lower. Iodate depletion is correlated with anaerobic respiration, and we estimate that this process reduces -1.8 ± 0.2 IO − 3 /PO 3− 4 . These three results match our understanding of iodine chemistry within an ODZ, since we presume that these water masses have starting concentrations of approximately 476 ± 9 nM iodate before entering the ODZ, after which anaerobic processes deplete this iodate. Table 6.1: Deconvolution results for simulating the iodate distribution in the ETNP ODZ. Name Coefficient Standard error 13CW 103 nM 28 nM NEPIW 479 nM 25 nM AAIW 413 nM 26 nM Aerobic respiration 116 nM IO − 3 /µ M PO 3− 4 54 nM IO − 3 /µ M PO 3− 4 Anaerobic respiration -1750 nM IO − 3 /µ M PO 3− 4 200 nM IO − 3 /µ M PO 3− 4 110 Our iodate deconvolution reveals that the 13CW has a starting concentration of only 100 ± 30 nM iodate. Therefore, this water mass reaches our transect in the northern ETNP ODZ with a deficit of approximately 350 nM iodate. Since this water mass has a mean transit time of approximately 50 years from formation to the southern edge of the ODZ [7], it likely had sufficient time to develop an iodate concentration closer to the mean subsurface oceanic value. The inclusion of anaerobic respiration in our iodate deconvolution means that the 13CW iodate deficit was not caused by this respiration or any process correlated with its stoichiometry. Although our simulated iodate results approximate the measured iodate (Fig. 5b), the presence of the 13CW is responsible formostlowiodateconcentrationsontheFK180624cruise(Fig. 5c). Ontheotherhand,anaerobic respiration causes small decreases in iodate concentrations across the entire transect as well as significant decreases in a few specific hotspots (Fig. 5d). Figure 6.5: Distributions of a) iodate, b) residuals of fit for simulated iodate, c) iodate depletion due to the presence of the 13CW water mass, and d) iodate depletion due to anaerobic respiration. (a) repeats a subset of the data seen in Fig. 3c but is plotted with potential density instead of depth, because water mass features are coherent across potential density. Wecan identifythe robustness ofthis deconvolutionbycomparing thesimulated iodateagainst the measured iodate (Fig. 6). Our simulated iodate concentrations are similar to the measured 111 concentrations and this fit is superior to the one performed in Hardisty et al. (2020). Our regres- sion underestimates iodate concentrations, with a slope of 0.73±0.04 and an intercept of 51±6. Variability between the simulated and measured iodate suggests that the parameters used in our deconvolution cannot fully explain the iodate distribution. Nevertheless, the results of this decon- volution can hint at the missing process or processes that shape iodate concentrations. Figure 6.6: Comparison between measured iodate and simulated iodate calculated using Eq. 5 and the coefficients specified in Table 1. 6.4.1.3 Extrapolatinganiodatereductionratefromwatermassdeconvolutionresults We can compare our estimate for an anaerobic iodate reduction rate against previously measured iodatereductionratesintheETNPODZusingarateofanaerobicrespirationthatcanbeconverted to iodate reduction using the -1.8 ± 0.2 IO − 3 /PO 3− 4 stoichiometry. Unfortunately, anaerobic respi- ration rates and iodate reduction rates have been measured extremely infrequently in the ETNP ODZ. The only measured anaerobic respiration rates originate from a study of carbon export near 112 the mouth of the Gulf of California [166]. The calculations for anaerobic respiration rate from Devol and Hartnett (2001) are included in supplemental information section 4. Hardistyetal. (2020a)reportstheonlyiodatereductionratesintheETNPODZ,andthesedata were measured during FK180624. Since Devol and Hartnett (2001) sampled near-shore whereas Hardisty et al. (2020a) measured iodate reduction in more open ocean conditions, we would expect the respiration rates to differ. Nevertheless, the anaerobic respiration data included in Devol and Hartnett (2001) enable us to compare our -1.8 ± 0.2 IO − 3 /PO 3− 4 stoichiometry against iodate reduction rates in the ETNP ODZ. The iodate reduction rate for a sample collected in the upper oxycline of the ETNP ODZ at 95 m with 11 µ M oxygen was 218.8 nM day − 1 . Applying our stoichiometry to the respiration rate calculatedbyDevolandHartnett(2001)at95m,weestimateaniodatereductionrateof99±12nM day − 1 . ThisvalueisthesameorderofmagnitudeoftheHardistyetal. (2020a)measurement, even though our stoichiometry is specifically for anaerobic respiration and both Hardisty et al. (2020a) and Devol and Hartnett (2001) reported oxic conditions at 95 m. For an anoxic depth (310 m), we estimate iodate reduction to be 60±7 nM day − 1 . These values differ significantly from anoxic iodatereductionratesmeasuredintheODZ,whichwereconsistently¡23nMday − 1 (Hardistyetal., 2020a). However,theseincubationswereperformedshipboard,anditisalwayspossiblethatoxygen contamination occurred during incubation set-up. While oxygen contamination could slightly alter the microbial community in a hypoxic sample, it would significantly alter the microbial community in an anoxic sample, and oxygen contamination might explain the limited rates of iodate reduction observed in anoxic samples but not hypoxic samples. Farrenkopf et al. (1997) also measured iodate reduction rates, but in the Arabian Sea ODZ. They report that in shipboard incubations, the reduction rate was not noticeable before increasing the iodate concentration to 1500 nM. This concentration is significantly higher than the native concentration as well as the amount added 113 in Hardisty et al. (2020). Therefore, their high iodate reduction rates may have been artificially inflated by this spike. 6.4.2 Oregon coastal waters 6.4.2.1 Distribution of redox-active compounds During our summer cruise in Oregon continental margin waters, we sampled consistently hypoxic bottomwatersontheHecetaBank. Thisregionisknownformoreintensehypoxia[225]andhigher benthicironfluxes[23]thanitssurroundings. Whenwesampledthere,theoxygenconcentrationson shelfbottomwaters(stations33-32)were28-50µ Mandtheoxygenconcentrationonthecontinental slope bottom water (station 31) was 8 µ M (Fig. 7a). Despite the fact that these bottom waters were hypoxic rather than anoxic, we observed extremely high concentrations of Fe(II), between 40-47 nM (Fig. 7b). These results indicate that even though the water column is hypoxic, high amounts of organic carbon in the sediment cause porewaters to become strongly anoxic near the sediment-water interface. 114 Figure 6.7: Distribution of a) oxygen, b) Fe(II), c) total dissolved iodine, and d) iodide measured on the Heceta Bank. Total dissolved iodine here is measured as the sum of iodate, iodide, and dissolved organic iodide. The gray background represents the bathymetry of Heceta Bank. In the ETNP ODZ, elevated iodide concentrations indicate that iodate reduction has occurred either in situ or some time before and was then transported to the sampling location. Therefore, iodide accumulation serves as a diagnostic for nearby iodate reduction. We observe minimal iodide accumulation on the Heceta Bank (Fig. 7d). The highest iodide concentration is a surface sample measured off-shelf, though there is one bottom water sample with 72 nM iodide. We also observe a mid-depth iodide feature on our most coastal Heceta Bank station, and we attribute this mid- depth feature to iodide flux from the Umpqua River. Total dissolved iodine (Fig. 7c) shares this mid-depth feature at the most coastal station, suggesting that this iodide flux corresponds with a flux of total iodine. The mid-Heceta Bank station where iodide accumulates to 72 nM has 478 nM total dissolved iodine, whereas other samples at this station have 420 and 485 nM total dissolved iodine. Therefore, the iodide accumulation at this bottom water sample is due to water column 115 iodate reduction to iodide, rather than benthic iodide release. The most obvious total dissolved iodine feature is a single measurement with 540 nM off-shelf, and this feature corresponds with Fe(II) transport off-shelf rather than significant iodide accumulation. Besides this single point, these total dissolved iodine concentrations do not indicate iodide flux from the reducing sediments, a process that would be anticipated given the 40 nM Fe(II). These results may be expected based on the water column oxygen concentrations, but they do not follow when comparing the Oregon Fe(II) concentrations against the ETNP ODZ Fe(II) concentrations. For Fe(II) to reach the high concentrations measured in Oregon shelf bottom waters, the sed- iment porewaters must be anoxic, and these sediments must be strongly reducing to create the gradient from anoxic porewaters to hypoxic bottom waters. Thermodynamically, iodate reduction becomes favorable before iron (oxy)hydroxide reduction [186]. In the ETNP ODZ, we observe 400-600 nM iodide but only 0.8 nM Fe(II). Despite the prolific Fe(II) accumulation on the hypoxic Oregoncoast,iodideonlyreachesconcentrationsofnM.Theseresultsindicatethatiodatereduction is not controlled purely by the reducing strength of the environment. If iodate depletion does not correspondwithreducingstrength, thentheiodateintercalatedwithincalciumcarbonatealsodoes not correspond with reducing strength. These results suggest that while the iodine/calcium ratio may measure a process often co-occurs with deoxygenation, this process does not always co-occur. 6.4.2.2 Benthic boundary layer profiles Reducing shelf sediments are expected to be the primary source of iodide in continental margins [26]. Therefore, high resolution profiles of iodine across the benthic boundary layer can provide useful information about benthic fluxes. We deployed a benthic boundary gradient sampler at two sitesonthenorthernHecetaBankinNovember2021. ThesestationswereMT2at198mandMT0 at 259 m (Fig. 2b). Unfortunately, this expedition occurred in the winter and therefore we did not 116 sample hypoxic waters, even in the benthic boundary layer. Nevertheless, we observed both Fe(II) and iodide in these samples, despite the 80-90 µ M of oxygen, as illustrated in Fig. 8. While oxygen depletion is clearly observed closer to the seafloor, trends in iodide and Fe(II) are far less coherent. At MT2, none of the benthic boundary layer profiles have the same trend. At MT0, iodide and Fe(II) do have similar profiles, though neither match the oxygen profile. In addition, the profiles for iodide and Fe(II) appear to decrease closer to the seafloor, though the seafloor is supposed to be the source for this material. While counterintuitive at first, these data profiles can be explained from the near-bottom velocities measured by the lander during samplecollection. WeproposethatsinceMT0wascollectedonthecontinentalslope, thesebenthic profilessampledmaterialduringcoherentoff-shelftransport,whereasonthecontinentalshelf,MT2 experienced less coherent advection and therefore this profile reflects material from a wider range of source locations on the shelf. The water column profiles of iodide and oxygen corresponding to these benthic boundary layer deployments are presented in Fig. S9 for reference. Figure 6.8: Profiles collected from a benthic boundary gradient lander at stations MT2 (top) and MT0 (bottom). Iodide samples measured with a hanging mercury drop electrode are labeled as (HMDE) whereas samples measured with an ICPMS are labeled (ICPMS). 117 These benthic lander data reveal high correlations between iodide and total dissolved iodine concentrations. These results suggest that the continental margin is acting as a source of excess iodine to the water column. This conclusion was difficult to acquire from water column profiles, likely due to the influence of river iodine and increased mixing in the water column, and these two factors are less significant for the benthic boundary gradient profiles. It is also interesting to note that we observe 20-40 nM increases in iodide and total dissolved iodine despite approximately 90 µ M oxygen concentrations. Nevertheless, the Fe(II) measured in these benthic lander data accumulates to higher concentrations than in the ETNP ODZ, so reductive dissolution must still be occurring in the continental shelf sediments. In addition, this Oregon Fe(II) accumulation leads to nearly a tenth less iodide accumulation than in the ETNP ODZ. These lander data also represent a coastal intercomparison of the hanging mercury drop elec- trode (HMDE) method for quantifying iodide as well as the AG1-X8 resin/inductively coupled plasma mass spectrometry (ICPMS) method for quantifying iodide. These methods demonstrated good agreement for deep water samples with low iodide. Our results reveal that coastal low iodide samplesalsohavegoodagreement, eventhoughtheiodideconcentrationsinthesesamplesarenear the detection limit of the HMDE method (10-15 nM). We recommend a methods intercomparison with ODZ samples, as these have high iodide concentrations, to confirm agreement between these methods. 6.5 Discussion 6.5.1 What factors control iodate depletion? The iodine/calcium proxy is commonly used to reconstruct oxygen concentrations because iodate depletion occurs in anoxic waters. However, iodate concentrations range from 0-500 nM in waters with oxygen below the detection limits of SBE43 sensors in the ETNP ODZ [26]. These concentra- tions span the entire range of oceanic iodate while Fe(II) only accumulates to 0.8 nM in the ETNP 118 ODZ [20]. On the other hand, iodide concentrations only reach 72 nM whereas Fe(II) accumulates as high as 47 nM on-shelf and 2 nM off-shelf on the hypoxic Oregon continental margin. These results reveal that iodate depletion, as measured through iodide accumulation, is not generally correlated with bottom water oxygen or the amount of dissimilatory iron reduction, measured with Fe(II) accumulation. In select regions like ODZs, iodate depletion and Fe(II) accumulation are linked [46], which allows iodide to be implemented as a non-oxidizing tracer for shelf-derived Fe. Nevertheless, the results from our comparison suggest that linkages between the iodine cycle and oxygen concentrations, as well as the iodine cycle and Fe(II) concentrations, may be regionally specific rather than broadly applicable. Our water mass deconvolution reveals that anaerobic respiration, or a correlated process, is one of the factors responsible for iodate depletion in the ETNP ODZ (Table 1, Fig. 5). Nevertheless, mostoftheiodatedepletionobservedintheETNPODZoccurredthroughaprocessthatprimarily affects the 13 ºC water mass (13CW) but does not influence nutrients like anaerobic respiration does. The 13CW overlaps with continental shelf sediments [1] and the West Mexican Coastal Current advects this water mass along the continental margin as it moves to the northern ODZ [254]. This transport means that the 13CW has been significantly exposed to processes that occur nearandinsedimentsbeforereachingthenorthernODZ.Processesknowntoreduceiodateinclude lowsubstrateaffinitydissimilatorynitratereduction[69,251],chemoautotrophiccouplingtonitrite oxidation [73], chemoheterotrophic reduction [250], and reduction via abiotic oxidation of sulfide [252]. Of these processes, we expect iodate reduction via low substrate affinity dissimilatory nitrate reduction to be the best correlated with anaerobic respiration. Therefore, this process does not explain the 13CW iodate deficit and is not the cause of iodate depletion due to anaerobic respi- ration. Iodate reduction via chemoautotrophic nitrite oxidation may be correlated with anaerobic 119 respiration in the water column of ETNP ODZ but it is certainly not correlated with anaerobic respiration in Oregon continental shelf sediments. Nitrite re-oxidation is common in water column denitrification [255, 11, 9], but rare in sediment denitrification [133, 256]. This discrepancy implies that chemoautotrophic nitrite oxidation may be responsible for the iodate depletion in the ETNP ODZ but not in Oregon coastal waters, matching our observations. We would expect that increased amounts of anaerobic respiration would increase the amount of nitrite re-oxidation, suggesting that this mechanism of iodate reduction may be correlated with anaerobicrespiration. However,anaerobicrespirationandnitritere-oxidationhavedistinctlydiffer- ent stoichiometries, and this stoichiometric difference would prevent our water mass deconvolution fromcorrelatingnitritere-oxidationwithanaerobicrespiration. Despitethisuncertainty, nitritere- oxidation cannot generate the magnitude of iodate reduction that we observe in the ETNP ODZ. We observe up to 500 nM excess iodide in the ETNP ODZ, and this signal originates from the continental margin. Since nitrite re-oxidation is rare in shelf sediments, some iodate depletion may occur due to chemoautotrophic nitrite oxidation, but other processes must be more significant. A facultative anaerobe that can chemoheterotrophically reduce iodate has been identified re- cently. Metagenomics have identified its presence in the ETNP ODZ as well as in the California Current System [250]. Chemoheterotrophic iodate reduction could occur alongside dissimilatory nitrate reduction, but it could also occur independently to it as a separate reaction. Our lack of knowledge about this metabolism limits our ability to determine its role in iodate depletion. The redox reaction between sulfide and iodate is rapid [252], far faster than sulfide and oxy- gen [234, 257]. Plumes of sulfide have periodically emerged in the ETSP ODZ [258], but this phenomenon has not yet been observed in the ETNP ODZ. Sulfide production does occur within reducing microenvironments in large sinking particles in the ETNP ODZ [259], but this is a rela- tivelysmallsourceofsulfidetothewatercolumn. Oregoncontinentalshelfwatershaveexperienced 120 several sulfidic pulses (Chan et al., 2008), but summer hypoxia is far more common [206]. While thewatersoftheETNPODZremainnitrogenous, themaximumsulfatereductionrateappearsless than 2 cm below the sediment-water interface. On the other hand, sediments on the Washington shelf do not reach their maximum sulfate reduction rates until 5 cm from the sediment-water inter- face [260]. This difference in the depth of the maximum measured sulfate reduction rate relative to the sediment-water interface suggests that porewater sulfide is more likely to reach bottom waters and reduce iodate in the ETNP ODZ. The elevated influence of sediments on the 13CW supports this process as the primary cause of iodate depletion in the ETNP ODZ. In addition, sulfide accumulation may explain why iodate depletion was observed in hypoxic waters in the Benguela Upwelling System [249] but not Oregon waters. The Benguela Upwelling System is known for sulfidic bottom waters [261], and the strong hypoxia observed by Truesdale and Bailey (2000) could have caused sulfidic sediment conditions, which would lower iodate con- centrations. These factors all increase the confidence that sulfide accumulation, especially near the sediment-water interface in reducing shelf sediments, serves as one of the primary drivers of iodate depletion in the water column. Table2summarizestheprocesseswehypothesizecauseiodatedepletionandthelinesofevidence supporting or eliminating them. The primary driver of iodate depletion cannot correlate with anaerobic respiration in the ETNP, explain the iodate depletion in the 13CW, and does not occur in Oregon shelf waters. While chemoautotrophic coupling to nitrite re-oxidation seems plausible, this reaction rarely occurs in reducing sediments and therefore cannot supply the excess iodine signal that we observed. This fact suggests that abiotic reduction with sulfide is most likely, though chemoheterotrophic reduction remains a possibility. 121 Table 6.2: Summary of iodate depletion mechanisms and their alignment with evidence presented in this study. Mechanism Correlates with Explains 13CW Occurs in or near anerobic resp. in iodate deficit? Oregon shelf waters? the ETNP? Low substrate affinity Yes, high No, high Yes, high dissimilatory nitrate reduction confidence confidence confidence Chemoautotrophic coupling Maybe, low No, medium No, high to nitrite oxidation confidence confidence confidence Chemoheterotrophic reduction Maybe, high Maybe, high Maybe, high uncertainty uncertainty uncertainty Abiotic reduction No, high Yes, high No, medium with sulfide confidence confidence confidence 6.5.2 Robustness of the I/Ca proxy Theselinesofevidencesuggestthatabioticreductionwithsulfidecausesmostiodatedepletion. This reaction pre-dominantly occurs near the sediment-water interface and the benthic boundary layer where Oxygen Minimum Zones overlap the continental shelf. Previous work has correlated bottom wateroxygenconcentrationswiththeI/Caratio[242,246],however,thesamplesforthiscorrelation originate primarily from regions nearby sulfidic sediments, such as the ETSP and Arabian Sea ODZs. The I/Ca proxy works well in these regions because bottom water properties do limit the degree of sulfide accumulation, however, using this proxy to regions without sulfide accumulation near the sediment-water interface will likely result in artificially elevated oxygen concentrations. We suggest that the I/Ca proxy only be applied in sediments where the dominant redox state of the sediment and water column is known. Iodate depletion clearly occurs in regions with sulfidic water columns [237, 232], and this proxy would also be effective in these conditions. The lack of iodate depletion on the Fe-rich Oregon continental shelf indicates that the I/Ca proxy decouples fromoxygeninFe-rich,reducingregions. Theabundanceofironlikelypreventssulfideaccumulation and therefore iodate reduction. The reliability of the I/Ca proxy in denitrifying waters has more complications. The I/Ca proxy is highly reliable in modern ODZs because of sulfidic porewaters 122 near the sediment-water interface, but what about water columns that are denitrifying but lack nearbysulfidicsediments? Thesesystemscouldstillhaveiodatedepletionduetochemoautotrophic nitrite oxidation, however, the relative magnitude of this process versus sulfide oxidation on iodate depletion is unknown. Regions with denitrifying water columns but without sulfidic porewaters near sediment-water interfaces are rare. Fortunately, the incomplete denitrification in the Bay of Bengal implies that high amounts of nitrite re-oxidation occur there [262], however, porewater sulfate concentrations reveal that sulfate reduction occurs coupled to methane oxidation [263], rather than heterotrophic sulfate reduction. We encourage research into iodine speciation in the Bay of Bengal as a useful test case to determine how robust the I/Ca ratio is in nitrogenous waters. We also suggest more research into iodatedepletionviachemoheterotrophicreductionbecausethispathwayisnotwell-knownandmay be significant in these systems. Nevertheless, we encourage researchers using iodine speciation to remember that the stability of iodine species means that its signal of low oxygen conditions can be transported far from the location where this process occurs. Water mass analysis and other mass transfer deconvolutions can aid in identifying the specific mechanisms causing iodate reduction. 123 References [1] Natalya Evans, Elisabeth Boles, Jarek V. Kwiecinski, Susan Mullen, Martin Wolf, Allan H. Devol, Rintaro Moriyasu, SungHyun Nam, Andrew R. Babbin, and James W. Moffett. The role of water masses in shaping the distribution of redox active compounds in the Eastern Tropical North Pacific oxygen deficient zone and influencing low oxygen concentrations in the eastern Pacific Ocean. Limnology and Oceanography, 65(8):1688–1705, 2020. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/lno.11412. [2] Edward B. Barbier. Marine ecosystem services. Current Biology, 27(11):R507–R510, June 2017. [3] Steven J. Bograd, Isaac D. Schroeder, and Michael G. Jacox. A water mass history of the Southern California current system. Geophysical Research Letters, 46(12):6690–6698, 2019. [4] Richard E. Thomson and Maxim V. Krassovski. Poleward reach of the California Undercur- rent extension. Journal of Geophysical Research: Oceans, 115(C9), September 2010. [5] DanielMcCoy,DanieleBianchi,andAndrewL.Stewart. Globalobservationsofsubmesoscale coherent vortices in the ocean. Progress in Oceanography, 189:102452, November 2020. [6] A. Paulmier and D. Ruiz-Pino. Oxygen minimum zones (OMZs) in the modern ocean. Progress in Oceanography, 80(3):113–128, March 2009. [7] A. Margolskee, H. Frenzel, S. Emerson, and C. Deutsch. Ventilation Pathways for the North Pacific Oxygen Deficient Zone. Global Biogeochemical Cycles, 33(7):875–890, 2019. [8] NicolasGruber. TheDynamicsoftheMarineNitrogenCycleanditsInfluenceonAtmospheric CO2 Variations. In Mick Follows and Temel Oguz, editors, The Ocean Carbon Cycle and Climate, pages 97–148. Springer Netherlands, Dordrecht, 2004. [9] Andrew R. Babbin, Carolyn Buchwald, Fran¸ cois M. M. Morel, Scott D. Wankel, and Bess B. Ward. Nitrite oxidation exceeds reduction and fixed nitrogen loss in anoxic Pacific waters. Marine Chemistry, 224:103814, August 2020. [10] CarolynBuchwald,AlysonE.Santoro,RachelH.R.Stanley,andKarenL.Casciotti. Nitrogen cyclinginthesecondarynitritemaximumoftheeasterntropicalNorthPacificoffCostaRica: NITROGEN CYCLING IN THE SNM. Global Biogeochemical Cycles, 29(12):2061–2081, December 2015. [11] Xuefeng Peng, Clara A. Fuchsman, Amal Jayakumar, Sergey Oleynik, Willm Martens- Habbena, Allan H. Devol, and Bess B. Ward. Ammonia and nitrite oxidation in the Eastern Tropical North Pacific. Global Biogeochemical Cycles, 29(12):2034–2049, 2015. 124 [12] A. C. Redfield, B. H. Ketchum, and F. A. Richards. The Influence of Organisms on the Composition of the Sea Water. In The Sea, volume 2, pages 26–77. Interscience Publishers, New York, 1963. [13] Avanti Shrikumar, Rian Lawrence, and Karen L. Casciotti. PYOMPA Technical Note, April 2022. Archive Location: world Publisher: Earth and Space Science Open Archive Section: Oceanography. [14] Konstantin Choumiline, Ligia P´ erez-Cruz, Andrew B. Gray, Steven M. Bates, and Timo- thy W. Lyons. Scenarios of Deoxygenation of the Eastern Tropical North Pacific During the PastMillenniumasaWindowIntotheFutureofOxygenMinimumZones. Frontiers in Earth Science, 7, 2019. [15] LisaA.Levin. Manifestation,Drivers,andEmergenceofOpenOceanDeoxygenation. Annual Review of Marine Science, 10(1):229–260, January 2018. [16] Lothar Stramma and Sunke Schmidtko. Tropical deoxygenation sites revisited to investigate oxygenandnutrienttrends. Ocean Science, 17(3):833–847, July2021. Publisher: Copernicus GmbH. [17] FrankA.Whitney, HowardJ.Freeland, andMarieRobert. Persistentlydecliningoxygenlev- elsintheinteriorwatersoftheeasternsubarcticPacific. ProgressinOceanography,75(2):179– 199, October 2007. [18] Caitlin E. Tems, William M. Berelson, Robert Thunell, Eric Tappa, Xiaomei Xu, Debo- rah Khider, Steve Lund, Oscar Gonz´ alez-Yajimovich, and Yvonne Hamann. Decadal to centennial fluctuations in the intensity of the eastern tropical North Pacific oxygen mini- mum zone during the last 1200 years. Paleoceanography, 31(8):1138–1151, 2016. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/2015PA002904. [19] Phoebe J. Lam, Maija I. Heller, Paul E. Lerner, James W. Moffett, and Kristen N. Buck. UnexpectedSourceandTransportofIronfromtheDeepPeruMargin. ACS Earth and Space Chemistry, 4(7):977–992, July 2020. Publisher: American Chemical Society. [20] Kenneth M. Bolster, Maija I. Heller, Margaret R. Mulholland, and James W. Moffett. Iron andmanganeseaccumulationwithintheEasternTropicalNorthPacificoxygendeficientzone. Geochimica et Cosmochimica Acta, 334:259–272, October 2022. [21] Joseph A. Resing, Peter N. Sedwick, Christopher R. German, William J. Jenkins, James W. Moffett, Bettina M. Sohst, and Alessandro Tagliabue. Basin-scale transport of hydrothermal dissolved metals across the South Pacific Ocean. Nature, 523(7559):200–203, July 2015. [22] Huong Thi Dieu Vu and Yoshiki Sohrin. Diverse stoichiometry of dissolved trace metals in the Indian Ocean. Scientific Reports , 3(1):1745, April 2013. Number: 1 Publisher: Nature Publishing Group. [23] Silke Severmann, James McManus, William M. Berelson, and Douglas E. Hammond. The continental shelf benthic iron flux and its isotope composition. Geochimica et Cosmochimica Acta, 74(14):3984–4004, July 2010. [24] Zanna Chase, Peter G. Strutton, and Burke Hales. Iron links river runoff and shelf width to phytoplankton biomass along the U.S. West Coast. Geophysical Research Letters, 34(4), 2007. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2006GL028069. 125 [25] Moutusi Roy, James McManus, Miguel A. Go˜ ni, Zanna Chase, Jeffry C. Borgeld, Robert A. Wheatcroft, Jesse M. Muratli, Meghan R. Megowan, and Alan Mix. Reactive iron and manganese distributions in seabed sediments near small mountainous rivers off Oregon and California (USA). Continental Shelf Research, 54:67–79, February 2013. [26] Rintaro Moriyasu, Natalya Evans, Kenneth M. Bolster, Dalton S. Hardisty, and James W. Moffett. The Distribution and Redox Speciation of Iodine in the Eastern Tropical North Pacific Ocean. Global Biogeochemical Cycles, 34(2):e2019GB006302, 2020. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2019GB006302. [27] Bo Thamdrup, Tage Dalsgaard, and Niels Peter Revsbech. Widespread functional anoxia in the oxygen minimum zone of the Eastern South Pacific. Deep Sea Research Part I: Oceano- graphic Research Papers, 65:36–45, July 2012. [28] Denise Breitburg, Lisa A. Levin, Andreas Oschlies, Marilaure Gr´ egoire, Francisco P. Chavez, DanielJ.Conley, V´ eroniqueGar¸ con, DenisGilbert, DimitriGuti´ errez, KirstenIsensee, GilS. Jacinto, Karin E. Limburg, Ivonne Montes, S. W. A. Naqvi, Grant C. Pitcher, Nancy N. Rabalais, Michael R. Roman, Kenneth A. Rose, Brad A. Seibel, Maciej Telszewski, Moriaki Yasuhara, andJingZhang. Decliningoxygenintheglobaloceanandcoastalwaters. Science, 359(6371), January 2018. [29] XiaoliZhou,HughC.Jenkyns,WanyiLu,DaltonS.Hardisty,JeremyD.Owens,TimothyW. Lyons, and Zunli Lu. Organically bound iodine as a bottom-water redox proxy: Preliminary validation and application. Chemical Geology, 457:95–106, May 2017. [30] John J Helly and Lisa A Levin. Global distribution of naturally occurring marine hypoxia on continental margins. Deep Sea Research Part I: Oceanographic Research Papers, 51(9):1159– 1168, September 2004. [31] Robert J. Diaz and Rutger Rosenberg. Spreading Dead Zones and Consequences for Marine Ecosystems. Science, 321(5891):926–929, August 2008. [32] Lothar Stramma, Gregory C. Johnson, Eric Firing, and Sunke Schmidtko. Eastern Pacific oxygen minimum zones: Supply paths and multidecadal changes. Journal of Geophysical Research: Oceans, 115(C9), September 2010. [33] Sunke Schmidtko, Lothar Stramma, and Martin Visbeck. Decline in global oceanic oxygen content during the past five decades. Nature, 542(7641):335–339, February 2017. [34] Tim Kalvelage, Marlene M. Jensen, Sergio Contreras, Niels Peter Revsbech, Phyllis Lam, Marcel G¨ unter, Julie LaRoche, Gaute Lavik, and Marcel M. M. Kuypers. Oxygen Sensitivity of Anammox and Coupled N-Cycle Processes in Oxygen Minimum Zones. PLOS ONE, 6(12):e29299, December 2011. [35] Tim Kalvelage, Gaute Lavik, Marlene M. Jensen, Niels Peter Revsbech, Carolin L¨ oscher, Harald Schunck, Dhwani K. Desai, Helena Hauss, Rainer Kiko, Moritz Holtappels, Julie LaRoche,RuthA.Schmitz,MichelleI.Graco,andMarcelM.M.Kuypers. AerobicMicrobial Respiration In Oceanic Oxygen Minimum Zones. PLOS ONE, 10(7):e0133526, July 2015. [36] Osvaldo Ulloa, Donald E. Canfield, Edward F. DeLong, Ricardo M. Letelier, and Frank J. Stewart. Microbial oceanography of anoxic oxygen minimum zones. Proceedings of the Na- tional Academy of Sciences, 109(40):15996–16003, October 2012. 126 [37] Molly T. Carolan, Jason M. Smith, and J. M. Beman. Transcriptomic evidence for micro- bial sulfur cycling in the eastern tropical North Pacific oxygen minimum zone. Frontiers in Microbiology, 6, 2015. [38] Daniele Bianchi, Thomas S. Weber, Rainer Kiko, and Curtis Deutsch. Global niche of ma- rine anaerobic metabolisms expanded by particle microenvironments. Nature Geoscience, 11(4):263–268, April 2018. [39] C. Deutsch, A. Ferrel, B. Seibel, H.-O. Portner, and R. B. Huey. Climate change tightens a metabolic constraint on marine habitats. Science, 348(6239):1132–1135, June 2015. [40] Jackson W. F. Chu and Katie S. P. Gale. Ecophysiological limits to aerobic metabolism in hypoxia determine epibenthic distributions and energy sequestration in the northeast Pacific ocean. Limnology and Oceanography, 62(1):59–74, January 2017. [41] Nicolas Gruber and Jorge L. Sarmiento. Global patterns of marine nitrogen fixa- tion and denitrification. Global Biogeochemical Cycles, 11(2):235–266, 1997. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/97GB00077. [42] L. A. Codispoti, Jay A. Brandes, J. P. Christensen, A. H. Devol, S. W. A. Naqvi, Hans W. Paerl, and T. Yoshinari. The oceanic fixed nitrogen and nitrous oxide budgets: Moving targets as we enter the anthropocene? Scientia Marina, 65(S2):85–105, December 2001. [43] Jay A. Brandes and Allan H. Devol. A global marine-fixed nitrogen isotopic budget: Im- plications for Holocene nitrogen cycling. Global Biogeochemical Cycles, 16(4):67–1–67–14, 2002. [44] Daniele Bianchi, John P. Dunne, Jorge L. Sarmiento, and Eric D. Galbraith. Data-based estimates of suboxia, denitrification, and N2O production in the ocean and their sensitivities to dissolved O2. Global Biogeochemical Cycles, 26(2), June 2012. [45] Florian Scholz, James McManus, Alan C. Mix, Christian Hensen, and Ralph R. Schneider. Theimpactofoceandeoxygenationonironreleasefromcontinentalmarginsediments. Nature Geoscience, 7(6):433–437, June 2014. [46] Gregory A. Cutter, James W. Moffett, Maria C. Nielsd´ ottir, and Virginie Sanial. Multi- ple oxidation state trace elements in suboxic waters off Peru: In situ redox processes and advective/diffusive horizontal transport. Marine Chemistry, 201:77–89, April 2018. [47] GregoryC.JohnsonandKristeneE.McTaggart. EquatorialPacific13 °CWaterEddiesinthe Eastern Subtropical South Pacific Ocean. Journal of Physical Oceanography, 40(1):226–236, January 2010. [48] M. Meinvielle and Gregory C. Johnson. Decadal water-property trends in the California Undercurrent, with implications for ocean acidification. Journal of Geophysical Research: Oceans, 118(12):6687–6703, December 2013. [49] SungHyun Nam, Yuichiro Takeshita, Christina A. Frieder, Todd Martz, and John Ballard. Seasonal advection of Pacific Equatorial Water alters oxygen and pH in the Southern Cali- fornia Bight. Journal of Geophysical Research: Oceans, 120(8):5387–5399, August 2015. 127 [50] NicolasGruber, ZouhairLachkar, HartmutFrenzel, PatrickMarchesiello, MatthiasM¨ unnich, James C. McWilliams, Takeyoshi Nagai, and Gian-Kasper Plattner. Eddy-induced reduc- tion of biological production in eastern boundary upwelling systems. Nature Geoscience, 4(11):787–792,November2011. Bandiera abtest: aCg type: NatureResearchJournalsNum- ber: 11Primary atype: ResearchPublisher: NaturePublishingGroupSubject term: Biogeo- chemistry;Physical oceanography Subject term id: biogeochemistry;physical-oceanography. [51] Ivy Frenger, Daniele Bianchi, Carolin St¨ uhrenberg, Andreas Oschlies, John Dunne, Curtis Deutsch, Eric Galbraith, and Florian Sch¨ utte. Biogeochemical Role of Sub- surface Coherent Eddies in the Ocean: Tracer Cannonballs, Hypoxic Storms, and Microbial Stewpots? Global Biogeochemical Cycles, 32(2):226–249, 2018. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1002/2017GB005743. [52] Damian L. Ar´ evalo-Mart´ ınez, Annette Kock, Carolin R. L¨ oscher, Ruth A. Schmitz, Lothar Stramma, and Hermann W. Bange. Influence of mesoscale eddies on the distribution of nitrousoxideintheeasterntropicalSouthPacific. Biogeosciences,13(4):1105–1118,February 2016. [53] M. A. Altabet, E. Ryabenko, L. Stramma, D. W. R. Wallace, M. Frank, P. Grasse, and G.Lavik. Aneddy-stimulatedhotspotforfixednitrogen-lossfromthePeruoxygenminimum zone. Biogeosciences, 9(12):4897–4908, December 2012. [54] Brian D. Peters, William J. Jenkins, James H. Swift, Christopher R. German, James W. Moffett,GregoryA.Cutter,MarkA.Brzezinski,andKarenL.Casciotti. Watermassanalysis of the 2013 US GEOTRACES eastern Pacific zonal transect (GP16). Marine Chemistry, 201:6–19, April 2018. [55] William S. Kessler. The circulation of the eastern tropical Pacific: A review. Progress in Oceanography, 69(2):181–217, May 2006. [56] Lothar Stramma, Sunke Schmidtko, Lisa A. Levin, and Gregory C. Johnson. Ocean oxygen minima expansions and their biological impacts. Deep Sea Research Part I: Oceanographic Research Papers, 57(4):587–595, April 2010. [57] Peter G. Brewer and Edward T. Peltzer. Ocean chemistry, ocean warming, and emerging hypoxia: Commentary. Journal of Geophysical Research: Oceans, 121(5):3659–3667, May 2016. [58] P Flament. A state variable for characterizing water masses and their diffusive stability: spiciness. Progress in Oceanography, 54(1):493–501, July 2002. [59] Peter G. Brewer, Andreas F. Hofmann, Edward T. Peltzer, and III Ussler, William. Evalu- ating microbial chemical choices: The ocean chemistry basis for the competition between use of O2 or NO3 as an electron acceptor. Deep Sea Research Part I: Oceanographic Research Papers, 87:35–42, May 2014. [60] Niels Peter Revsbech, Lars Hauer Larsen, Jens Gundersen, Tage Dalsgaard, Osvaldo Ulloa, and Bo Thamdrup. Determination of ultra-low oxygen concentrations in oxygen minimum zones by the STOX sensor. Limnology and Oceanography: Methods, 7(5):371–381, May 2009. [61] A.R.Babbin,R.G.Keil,A.H.Devol,andB.B.Ward. OrganicMatterStoichiometry,Flux, and Oxygen Control Nitrogen Loss in the Ocean. Science, 344(6182):406–408, April 2014. 128 [62] George W. Luther, Jingfeng Wu, and John B. Cullen. Redox Chemistry of Iodine in Seawa- ter. In Aquatic Chemistry, volume 244 of Advances in Chemistry, pages 135–155. American Chemical Society, May 1995. [63] A.M.FarrenkopfandG.W.Luther. Iodinechemistryreflectsproductivityanddenitrification intheArabianSea: evidenceforfluxofdissolvedspeciesfromsedimentsofwesternIndiainto the OMZ. Deep Sea Research Part II: Topical Studies in Oceanography, 49(12):2303–2318, January 2002. [64] Peng Zhou, Xiuxian Song, Yongquan Yuan, Xihua Cao, Wentao Wang, Lianbao Chi, and Zhiming Yu. Water Mass Analysis of the East China Sea and Interannual Variation of Kuroshio Subsurface Water Intrusion Through an Optimum Multiparameter Method. Jour- nal of Geophysical Research: Oceans, 123(5):3723–3738, May 2018. [65] D. A. Hutchins, G. R. DiTullio, Y. Zhang, and K. W. Bruland. An iron limitation mosaic in the California upwelling regime. Limnology and Oceanography, 43(6):1037–1054, September 1998. [66] Brian M. Hopkinson and Katherine A. Barbeau. Interactive influences of iron and light limitation on phytoplankton at subsurface chlorophyll maxima in the eastern North Pacific. Limnology and Oceanography, 53(4):1303–1318, July 2008. [67] Silke Severmann, Timothy W. Lyons, Ariel Anbar, James McManus, and Gwyneth Gordon. Modernironisotopeperspectiveonthebenthicironshuttleandtheredoxevolutionofancient oceans. Geology, 36(6):487–490, June 2008. [68] Florian Scholz, Silke Severmann, James McManus, Anna Noffke, Ulrike Lomnitz, and Chris- tian Hensen. On the isotope composition of reactive iron in marine sediments: Redox shuttle versus early diagenesis. Chemical Geology, 389:48–59, December 2014. [69] AnnaM.Farrenkopf, MichaelE.Dollhopf, Sin´ eadN´ ıChadhain, GeorgeW.Luther, andKen- neth H. Nealson. Reduction of iodate in seawater during Arabian Sea shipboard incubations and in laboratory cultures of the marine bacterium Shewanella putrefaciens strain MR-4. Marine Chemistry, 57(3):347–354, July 1997. [70] Eden L. Rue, Geoffrey J. Smith, Gregory A. Cutter, and Kenneth W. Bruland. The response of trace element redox couples to suboxic conditions in the water column. Deep Sea Research Part I: Oceanographic Research Papers, 44(1):113–134, January 1997. [71] Shizuo Tsunogai and Shinichiro Noriki. Organic matter fluxes and the sites of oxygen consumption in deep water. Deep Sea Research Part A. Oceanographic Research Papers, 34(5):755–767, May 1987. [72] K Bluhm, P Croot, K Wuttig, and K Lochte. Transformation of iodate to iodide in marine phytoplankton driven by cell senescence. Aquatic Biology, 11(1):1–15, October 2010. [73] A. R. Babbin, Brian D. Peters, Calvin W. Mordy, Brittany Widner, Karen L. Casciotti, and Bess B. Ward. Multiple metabolisms constrain the anaerobic nitrite budget in the Eastern Tropical South Pacific: Nitrogen Dynamics in the Eastern Tropical South Pacific. Global Biogeochemical Cycles, 2017. 129 [74] Kitae Kim, Jinjung Ju, Bomi Kim, Hyun Young Chung, Lubica Vetrakova, Dominik Heger, Alfonso Saiz-Lopez, Wonyong Choi, and Jungwon Kim. Nitrite-Induced Activation of Iodate into Molecular Iodine in Frozen Solution. Environmental Science & Technology, March 2019. [75] Inc. The MathWorks. MATLAB Release 2018b. Natick, MA, 2018. [76] T. J. McDougall and P.M. Barker. Getting started with TEOS-10 and the Gibbs Seawater (GSW) Oceanographic Toolbox. SCOR/IAPSO WG127, 2011. [77] Johannes Karstensen and Matthias Tomczak. Ventilation processes and water mass ages in the thermocline of the southeast Indian Ocean. Geophysical Research Letters, 24(22):2777– 2780, November 1997. [78] Heidi Leffanue and Matthias Tomczak. Using OMP analysis to observe temporal variability in water mass distribution. Journal of Marine Systems, 48(1):3–14, July 2004. [79] Matthias Tomczak. Some historical, theoretical and applied aspects of quantitative water mass analysis. Journal of Marine Research, 57(2):275–303, March 1999. [80] Eastward shift and maintenance of Arabian Sea oxygen minimum zone: Understanding the paradox. Deep Sea Research Part I: Oceanographic Research Papers,115:240–252,September 2016. [81] Peng Zhou, Xiuxian Song, Yongquan Yuan, Wentao Wang, Lianbao Chi, Xihua Cao, and Zhiming Yu. Intrusion of the Kuroshio Subsurface Water in the southern East China Sea and its variation in 2014 and 2015 traced by dissolved inorganic iodine species. Progress in Oceanography, 165:287–298, July 2018. [82] Il-Nam Kim, Dong-Ha Min, Dae Hyun Kim, and Tongsup Lee. Investigation of the physico- chemical features and mixing of East/Japan Sea Intermediate Water: An isopycnic analysis approach. Journal of Marine Research, 68(6):799–818, November 2010. [83] Robert Poole and Matthias Tomczak. Optimum multiparameter analysis of the water mass structure in the Atlantic Ocean thermocline. Deep Sea Research Part I: Oceanographic Re- search Papers, 46(11):1895–1921, November 1999. [84] Matthias Tomczak and Daniel G. B. Large. Optimum multiparameter analysis of mixing in the thermocline of the eastern Indian Ocean. Journal of Geophysical Research: Oceans, 94(C11):16141–16149, November 1989. [85] L. Zhang, J. Zhang, P. W. Swarzenski, and Z. Liu. Radium Isotope Tracers to Evaluate Coastal Ocean Mixing and Residence Times. In M. Baskaran, editor, Handbook of Environ- mental Isotope Geochemistry, pages 331–343. 2012. [86] Steven J. Bograd, Mercedes Pozo Buil, Emanuele Di Lorenzo, Carmen G. Castro, Isaac D. Schroeder,RalfGoericke,ClarissaR.Anderson,ClaudiaBenitez-Nelson,andFrankA.Whit- ney. Changes in source waters to the Southern California Bight. Deep Sea Research Part II: Topical Studies in Oceanography, 112:42–52, February 2015. [87] W. J. Jenkins, W. M. Smethie, E. A. Boyle, and G. A. Cutter. Water mass analysis for the U.S. GEOTRACES (GA03) North Atlantic sections. Deep Sea Research Part II: Topical Studies in Oceanography, 116:6–20, June 2015. 130 [88] YYou, NobuoSuginohara, MasaoFukasawa, IchiroYasuda, IkuoKaneko, HiroyukiYoritaka, and Michio Kawamiya. Roles of the Okhotsk Sea and Gulf of Alaska in forming the North Pacific Intermediate Water. Journal of Geophysical Research: Oceans, 105(C2):3253–3280, February 2000. [89] Klaus Wyrtki and Robert Kendall. Transports of the Pacific Equatorial Countercurrent. Journal of Geophysical Research, 72(8):2073–2076, April 1967. [90] Paul C. Fiedler and Lynne D. Talley. Hydrography of the eastern tropical Pacific: A review. Progress in Oceanography, 69(2):143–180, May 2006. [91] C. E. Warsh, K. L. Warsh, and R. C. Staley. Nutrients and water masses at the mouth of the Gulf of California. Deep Sea Research and Oceanographic Abstracts, 20(6):561–570, June 1973. [92] JushiroCepeda-Morales,GilbertoGaxiola-Castro,EmilioBeier,andV´ ıctorM.God´ ınez. The mechanismsinvolvedindefiningthenorthernboundaryoftheshallowoxygenminimumzone in the eastern tropical Pacific Ocean off Mexico. Deep Sea Research Part I: Oceanographic Research Papers, 76:1–12, June 2013. [93] EstherPortela,EmilioBeier,EricD.Barton,Rub´ enCastro,VictorGod´ ınez,EmilioPalacios- Hern´ andez, Paul C. Fiedler, Laura S´ anchez-Velasco, and Armando Trasvi˜ na. Water Masses and Circulation in the Tropical Pacific off Central Mexico and Surrounding Areas. Journal of Physical Oceanography, 46(10):3069–3081, August 2016. [94] P. C. Garfield, T. T. Packard, G. E. Friederich, and L. A. Codispoti. A subsurface particle maximum layer and enhanced microbial activity in the secondary nitrite maximum of the northeastern tropical Pacific Ocean. Journal of Marine Research, 41(4):747–768, November 1983. [95] Nelson Silva, Nora Rojas, and Aldo Fedele. Water masses in the Humboldt Current System: Properties,distribution,andthenitratedeficitasachemicalwatermasstracerforEquatorial Subsurface Water off Chile. Deep Sea Research Part II: Topical Studies in Oceanography, 56(16):1004–1020, July 2009. [96] Evgenia Ryabenko. Nitrogen Isotopes in the Atlantic and Pacific Oxygen Minimum Zones . Doctoral Thesis, Christian-Albrechts Universit¨ at Kiel, 2011. [97] Robert H. Bourke and June G. Pattullo. Seasonal variation of the water mass along the Oregon-northern California coast. Limnology and Oceanography, 19(2):190–198, March 1974. [98] Klaus Wyrtki. Oceanography of the Eastern Equatorial Pacific Ocean. Oceanography and Marine Biology Annual Review, 4:33–68, 1966. [99] MizukiTsuchiya,RogerLukas,RanaA.Fine,EricFiring,andEricLindstrom. Sourcewaters of the Pacific Equatorial Undercurrent. Progress in Oceanography, 23(2):101–147, January 1989. [100] Ronald J. Lynn and James J. Simpson. The flow of the undercurrent over the continental borderland off southern California. Journal of Geophysical Research: Oceans, 95(C8):12995– 13008, August 1990. 131 [101] Mizuki Tsuchiya. The Origin of Equatorial 13 C Water. Journal of Physical Oceanography, 11:794–812, 1981. [102] E. C. V. Butler, J. A. Butt, E.J. Lindstrom, P.C. Teldesley, S. Pickmere, and W.F. Vincent. Oceanography of the Subtropical Convergence Zone around southern New Zealand. New Zealand Journal of Marine and Freshwater Research, 26(2):131–154, June 1992. [103] Tangdong Qu, Shan Gao, Ichiro Fukumori, Rana A. Fine, and Eric J. Lindstrom. Origin and Pathway of Equatorial 13°C Water in the Pacific Identified by a Simulated Passive Tracer and Its Adjoint. Journal of Physical Oceanography, 39(8):1836–1853, August 2009. [104] Helen C. Bostock, Bradley N. Opdyke, and Michael J. M. Williams. Characterising the intermediate depth waters of the Pacific Ocean using 13C and other geochemical tracers. Deep Sea Research Part I: Oceanographic Research Papers, 57(7):847–859, July 2010. [105] L. D. Talley. Antarctic Intermediate Water in the South Atlantic. In The South Atlantic, pages 219–238. Springer Berlin Heidelberg, Berlin, Heidelberg, 1996. [106] Myles R. Allen and Leonard A. Smith. Investigating the origins and significance of low- frequency modes of climate variability. Geophysical Research Letters, 21(10):883–886, May 1994. [107] Jun Hu, Julien Emile-Geay, and Judson Partin. Correlation-based interpretations of pale- oclimate data – where statistics meet past climates. Earth and Planetary Science Letters, 459:362–371, February 2017. [108] Gail M. Sullivan and Richard Feinn. Using Effect Size—or Why the P Value Is Not Enough. Journal of Graduate Medical Education, 4(3):279–282, September 2012. [109] Bj¨ orn Lantz. The large sample size fallacy. Scandinavian Journal of Caring Sciences, 27(2):487–492, June 2013. [110] Daniel Bergh. Sample Size and Chi-Squared Test of Fit—A Comparison Between a Random Sample Approach and a Chi-Square Value Adjustment Method Using Swedish Adolescent Data. Pacific Rim Objective Measurement Symposium (PROMS) 2014 Conference Proceed- ings, pages 197–211, 2015. [111] Rene Andrae, Tim Schulze-Hartung, and Peter Melchior. Dos and don’ts of reduced chi- squared. arXiv:1012.3754[astro-ph, physics:physics, stat],December2010. arXiv: 1012.3754. [112] Marcus Sobarzo, Luis Bravo, David Donoso, Jos´ e Garc´ es-Vargas, and Wolfgang Schneider. Coastal upwelling and seasonal cycles that influence the water column over the continental shelf off central Chile. Progress in Oceanography, 75(3):363–382, November 2007. [113] S. Alvarez-Borrego and R. A. Schwartzlose. Water Masses of the Gulf of California. Ciencias Marinas, 6(1):43–63, March 1979. [114] C. G. Castro, F. P. Chavez, and C. A. Collins. Role of the California Undercurrent in the export of denitrified waters from the eastern tropical North Pacific. Global Biogeochemical Cycles, 15(4):819–830, December 2001. 132 [115] J. A. Segovia-Zavala, M. L. Lares, F. Delgadillo-Hinojosa, A. Tovar-S´ anchez, and S. A. Sa˜ nudo-Wilhelmy. Dissolved iron distributions in the central region of the Gulf of California, M´ exico. Deep Sea Research Part I: Oceanographic Research Papers, 57(1):53–64, January 2010. [116] Sergio A. Sa˜ nudo-Wilhelmy, Lynda S. Cutter, Reginaldo Durazo, Emily A. Smail, Laura G´ omez-Consarnau,EricA.Webb,MariaG.Prokopenko,WilliamM.Berelson,andDavidM. Karl. Multiple B-vitamin depletion in large areas of the coastal ocean. Proceedings of the National Academy of Sciences, 109(35):14041–14045, August 2012. [117] Susan E. Wijffels, John M. Toole, Harry L. Bryden, Rana A. Fine, William J. Jenkins, and JohnL.Bullister. Thewatermassesandcirculationat10°NinthePacific. Deep Sea Research Part I: Oceanographic Research Papers, 43(4):501–544, April 1996. [118] Feng Nan, Zhigang He, Hui Zhou, and Dongxiao Wang. Three long-lived anticyclonic eddies in the northern South China Sea. Journal of Geophysical Research: Oceans, 116(C5), May 2011. [119] Dudley B. Chelton, Michael G. Schlax, and Roger M. Samelson. Global observations of nonlinear mesoscale eddies. Progress in Oceanography, 91(2):167–216, October 2011. [120] SungHyun Nam and Uwe Send. Resonant Diurnal Oscillations and Mean Alongshore Flows Driven by Sea/Land Breeze Forcing in the Coastal Southern California Bight. Journal of Physical Oceanography, 43(3):616–630, January 2013. [121] Alexis Chaigneau, Marie Le Texier, G´ erard Eldin, Carmen Grados, and Oscar Pizarro. Ver- tical structure of mesoscale eddies in the eastern South Pacific Ocean: A composite analysis fromaltimetryandArgoprofilingfloats. Journal of Geophysical Research: Oceans,116(C11), November 2011. [122] Cori Pegliasco, Alexis Chaigneau, and Rosemary Morrow. Main eddy vertical structures observed in the four major Eastern Boundary Upwelling Systems. Journal of Geophysical Research: Oceans, 120(9):6008–6033, September 2015. [123] J. Timothy Pennington, Kevin L. Mahoney, Victor S. Kuwahara, Dorota D. Kolber, Ruth Calienes, and Francisco P. Chavez. Primary production in the eastern tropical Pacific: A review. Progress in Oceanography, 69(2):285–317, May 2006. [124] J. E. Johnson, S. M. Webb, K. Thomas, S. Ono, J. L. Kirschvink, and W. W. Fischer. Manganese-oxidizing photosynthesis before the rise of cyanobacteria. Proceedings of the Na- tional Academy of Sciences, 110(28):11238–11243, July 2013. [125] Steven Bograd, Isaac Schroeder, and Michael Jacox. A water mass history of the Southern California Current System, April 2018. [126] StevenJ.Bograd,CarmenG.Castro,EmanueleDiLorenzo,DanielM.Palacios,HelenBailey, William Gilly, and Francisco P. Chavez. Oxygen declines and the shoaling of the hypoxic boundary in the California Current. Geophysical Research Letters, 35(12), June 2008. [127] WilliamF.Gilly, J.MichaelBeman, StevenY.Litvin, andBruceH.Robison. Oceanographic and Biological Effects of Shoaling of the Oxygen Minimum Zone. Annual Review of Marine Science, 5(1):393–420, January 2013. 133 [128] N.D. Gallo and L.A. Levin. Fish Ecology and Evolution in the World’s Oxygen Minimum Zones and Implications of Ocean Deoxygenation. In Advances in Marine Biology, volume 74, pages 117–198. Elsevier, 2016. [129] Weiwei Fu, Francois Primeau, J. Keith Moore, Keith Lindsay, and James T. Randerson. Reversal of Increasing Tropical Ocean Hypoxia Trends With Sustained Climate Warming. Global Biogeochemical Cycles, 32(4):551–564, April 2018. [130] Y. You and M. Tomczak. Thermocline circulation and ventilation in the Indian Ocean de- rived from water mass analysis. Deep Sea Research Part I: Oceanographic Research Papers, 40(1):13–56, January 1993. [131] Peng Xiu, Fei Chai, Enrique N. Curchitser, and Frederic S. Castruccio. Future changes in coastalupwellingecosystemswithglobalwarming: ThecaseoftheCaliforniaCurrentSystem. Scientific Reports , 8(1):2866, February 2018. [132] S. Coats and K. B. Karnauskas. A Role for the Equatorial Undercurrent in the Ocean Dynamical Thermostat. Journal of Climate, 31(16):6245–6261, May 2018. [133] Jay A. Brandes and Allan H. Devol. Isotopic fractionation of oxygen and nitrogen in coastal marine sediments. Geochimica et Cosmochimica Acta, 61(9):1793–1801, May 1997. [134] Francis A. Richards. Anoxic basins and fjords. pages 611–645. Academic Press, New York, USA, 1965. [135] P. N. Froelich, G. P. Klinkhammer, M. L. Bender, N. A. Luedtke, G. R. Heath, Doug Cullen, Paul Dauphin, Doug Hammond, Blayne Hartman, and Val Maynard. Early oxidation of organic matter in pelagic sediments of the eastern equatorial Atlantic: suboxic diagenesis. Geochimica et Cosmochimica Acta, 43(7):1075–1090, July 1979. [136] Curtis Deutsch, Nicolas Gruber, Robert M. Key, Jorge L. Sarmiento, and Alexandre Ganachaud. Denitrification and N2 fixation in the Pa- cific Ocean. Global Biogeochemical Cycles, 15(2):483–506, 2001. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2000GB001291. [137] AndrewR.Babbin,ElisabethL.Boles,JensM¨ uhle,andRayF.Weiss. Onthenaturalspatio- temporal heterogeneity of South Pacific nitrous oxide. Nature Communications, 11(1):3672, July 2020. Number: 1 Publisher: Nature Publishing Group. [138] Karen L. Casciotti, Carolyn Buchwald, and Matthew McIlvin. Implications of nitrate and nitrite isotopic measurements for the mechanisms of nitrogen cycling in the Peru oxygen deficient zone. Deep Sea Research Part I: Oceanographic Research Papers, 80:78–93, October 2013. [139] J. M. Beman, S. M. Vargas, J. M. Wilson, E. Perez-Coronel, J. S. Karolewski, S. Vazquez, A. Yu, A. E. Cairo, M. E. White, I. Koester, L. I. Aluwihare, and S. D. Wankel. Substantial oxygen consumption by aerobic nitrite oxidation in oceanic oxygen minimum zones. Nature Communications, 12(1):7043, December 2021. Bandiera abtest: a Cc license type: cc by Cg type: Nature Research Journals Number: 1 Primary atype: Research Publisher: Nature Publishing Group Subject term: Element cycles;Marine chemistry;Metagenomics;Microbial ecology Subject term id: element-cycles;marine-chemistry;metagenomics;microbial-ecology. 134 [140] A. C. Redfield. On the proportions of organic derivatives in sea water and their relation to the composition of plankton. James Johnstone Memorial Volume, pages 176–192, 1934. Publisher: The University Press. [141] R.H.Fleming. Compositionofplanktonandunitsforreportingpopulationsandproduction. Proceedings 6th Pacific Science Congress , 3:535–540, October 1940. [142] Laurence A. Anderson and Jorge L. Sarmiento. Redfield ratios of remineralization deter- mined by nutrient data analysis. Global Biogeochemical Cycles, 8(1):65–80, 1994. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/93GB03318. [143] Adam C. Martiny, Chau T. A. Pham, Francois W. Primeau, Jasper A. Vrugt, J. Keith Moore, Simon A. Levin, and Michael W. Lomas. Strong latitudinal patterns in the elemental ratios of marine plankton and organic matter. Nature Geoscience, 6(4):279–283, April 2013. Bandiera abtest: a Cg type: Nature Research Journals Number: 4 Primary atype: Research Publisher: Nature Publishing Group Subject term: Biogeochemistry;Biooceanography Sub- ject term id: biogeochemistry;biooceanography. [144] Benjamin A. S Van Mooy, Richard G Keil, and Allan H Devol. Impact of suboxia on sinking particulateorganiccarbon: Enhancedcarbonfluxandpreferentialdegradationofaminoacids via denitrification. Geochimica et Cosmochimica Acta, 66(3):457–465, February 2002. [145] JarekV.KwiecinskiandAndrewR.Babbin. AHigh-ResolutionAtlasoftheEasternTropical Pacific Oxygen Deficient Zones. Global Biogeochemical Cycles, 35(12):e2021GB007001, 2021. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2021GB007001. [146] L. A. Codispoti and F. A. Richards. An analysis of the horizontal regime of denitrification in the eastern tropical North Pacific1: Horizontal denitrification regime. Limnology and Oceanography, 21(3):379–388, May 1976. [147] Natalya Evans, Juliana Tichota, Wendi Ruef, James Moffett, and Allan Devol. Rapid ex- pansion of fixed nitrogen deficit in the eastern Pacific Ocean revealed by 50 year time series, January 2022. Section: Oceanography. [148] Hernan E. Garcia, Timothy P. Boyer, Ricardo A. Locarnini, John I. Antonov, Alexey V. Mishonov, Olga K. Baranova, Melissa M. Zweng, James R. Reagan, and Daphne R. Johnson. Worldoceanatlas2013.Volume3,Dissolvedoxygen,apparentoxygenutilization,andoxygen saturation. September 2013. Publisher: U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Environmental Satellite, Data and Information Service. [149] Reiner Schlitzer. Ocean Data View. 2020. [150] Donald M. Glover, William J. Jenkins, and Scott C. Doney. Modeling Methods for Marine Science. Cambridge University Press, 2011. [151] Fabio Crameri, Grace E. Shephard, and Philip J. Heron. The misuse of colour in science communication. Nature Communications, 11(1):5444, October 2020. Number: 1 Publisher: Nature Publishing Group. [152] Natalya Evans, Hal Van Ryswyk, Marc Los Huertos, and Tanja Srebotnjak. Robust spa- tial analysis of sequestered metals in a Southern California Bioswale. Science of The Total Environment, 650:155–162, February 2019. 135 [153] Rachel E. A. Horak, Wendi Ruef, Bess B. Ward, and Allan H. Devol. Ex- pansion of denitrification and anoxia in the eastern tropical North Pacific from 1972 to 2012. Geophysical Research Letters, 43(10):5252–5260, 2016. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1002/2016GL068871. [154] Tangdong Qu and Eric J. Lindstrom. Northward Intrusion of Antarctic Intermediate Water in the Western Pacific. Journal of Physical Oceanography, 34(9):2104–2118, September 2004. Publisher: American Meteorological Society Section: Journal of Physical Oceanography. [155] Helen C. Bostock, Phil J. Sutton, Michael J. M. Williams, and Bradley N. Opdyke. Review- ing the circulation and mixing of Antarctic Intermediate Water in the South Pacific using evidence from geochemical tracers and Argo float trajectories. Deep Sea Research Part I: Oceanographic Research Papers, 73:84–98, March 2013. [156] Panagiota-Myrsini Chronopoulou, Felicity Shelley, William J. Pritchard, Susanna T. Maanoja, and Mark Trimmer. Origin and fate of methane in the Eastern Tropical North Pacific oxygen minimum zone. The ISME Journal, 11(6):1386–1399, June 2017. Number: 6 Publisher: Nature Publishing Group. [157] L. Tiano, E. Garcia-Robledo, T. Dalsgaard, A.H. Devol, B.B. Ward, O. Ulloa, D.E. Canfield, and N. Peter Revsbech. Oxygen distribution and aerobic respiration in the north and south eastern tropical Pacific oxygen minimum zones. Deep-Sea Research Part I: Oceanographic Research Papers, 94:173–183, 2014. [158] Aur´ elienRibes,Sa¨ ıdQasmi,andNathanP.Gillett. Makingclimateprojectionsconditionalon historicalobservations. ScienceAdvances,7(4):eabc0671,January2021. Publisher: American Association for the Advancement of Science Section: Research Article. [159] Morten Larsen, Philipp Lehner, Sergey M. Borisov, Ingo Klimant, Jan P. Fischer, Frank J. Stewart, Donald E. Canfield, and Ronnie N. Glud. In situ quantifica- tion of ultra-low O2 concentrations in oxygen minimum zones: Application of novel optodes. Limnology and Oceanography: Methods, 14(12):784–800, 2016. eprint: https://aslopubs.onlinelibrary.wiley.com/doi/pdf/10.1002/lom3.10126. [160] T.DeVries,C.Deutsch,P.A.Rafter,andF.Primeau. Marinedenitrificationratesdetermined from a global 3-D inverse model. Biogeosciences, 10(4):2481–2496, April 2013. Publisher: Copernicus GmbH. [161] Ingrid A. van de Leemput, Annelies J. Veraart, Vasilis Dakos, Jeroen J. M. de Klein, Marc Strous, and Marten Scheffer. Predicting microbial nitrogen pathways from basic principles. Environmental Microbiology, 13(6):1477–1487, 2011. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1462-2920.2011.02450.x. [162] Xin Sun, Claudia Frey, Emilio Garcia-Robledo, Amal Jayakumar, and Bess B. Ward. Micro- bial niche differentiation explains nitrite oxidation in marine oxygen minimum zones. The ISME Journal, pages 1–13, January 2021. Publisher: Nature Publishing Group. [163] Nicolas Gruber and James N. Galloway. An Earth-system perspective of the global nitrogen cycle. Nature, 451(7176):293–296, January 2008. Number: 7176 Publisher: Nature Publish- ing Group. 136 [164] Phyllis Lam and Marcel M.M. Kuypers. Microbial Nitrogen Cycling Processes in Oxygen Minimum Zones. Annual Review of Marine Science, 3(1):317–345, January 2011. [165] E.L.Cavan,M.Trimmer,F.Shelley,andR.Sanders. Remineralizationofparticulateorganic carbon in an ocean oxygen minimum zone. Nature Communications, 8(1):1–9, March 2017. Number: 1 Publisher: Nature Publishing Group. [166] Allan H. Devol and Hilairy E. Hartnett. Role of the oxygen-deficient zone in transfer of organic carbon to the deep ocean. Limnology and Oceanography, 46(7):1684–1690, 2001. eprint: https://aslopubs.onlinelibrary.wiley.com/doi/pdf/10.4319/lo.2001.46.7.1684. [167] Maija I. Heller, Phoebe J. Lam, James W. Moffett, Claire P. Till, Jong-Mi Lee, Brandy M. Toner, and Matthew A. Marcus. Accumulation of Fe oxyhydroxides in the Peruvian oxygen deficientzoneimpliesnon-oxygendependentFeoxidation. GeochimicaetCosmochimicaActa, 211:174–193, August 2017. [168] Ga¨ el Le Croizier, Jeroen E. Sonke, Anne Lorrain, Sandrine Serre, Lucien Besnard, Gau- thier Schaal, Felipe Amezcua-Martinez, and David Point. Mercury stable isotopes suggest reducedforagingdepthinoxygenminimumzonesforbluesharks. Marine Pollution Bulletin, 181:113892, August 2022. [169] Lothar Stramma, Eric D. Prince, Sunke Schmidtko, Jiangang Luo, John P. Hoolihan, Martin Visbeck, Douglas W. R. Wallace, Peter Brandt, and Arne K¨ ortzinger. Expansion of oxygen minimum zones may reduce available habitat for tropical pelagic fishes. Nature Climate Change, 2(1):33–37, January 2012. Number: 1 Publisher: Nature Publishing Group. [170] D.Espinoza-Morriber´ on,V.Echevin,D.Guti´ errez,J.Tam,M.Graco,J.Ledesma,andF.Co- las. Evidences and drivers of ocean deoxygenation off Peru over recent past decades. Scien- tific Reports , 11(1):20292, October 2021. Bandiera abtest: a Cc license type: cc by Cg type: Nature Research Journals Number: 1 Primary atype: Research Publisher: Nature Pub- lishing Group Subject term: Biogeochemistry;Environmental sciences;Ocean sciences Sub- ject term id: biogeochemistry;environmental-sciences;ocean-sciences. [171] E. Hawkins and R. Sutton. Time of emergence of climate signals. Geophysical Research Let- ters, 39(1), 2012. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2011GL050087. [172] Johannes Karstensen, Lothar Stramma, and Martin Visbeck. Oxygen minimum zones in the eastern tropical Atlantic and Pacific oceans. Progress in Oceanography, 77(4):331–350, June 2008. [173] Jorge Feliciano Ontiveros-Cuadras, Robert Thunell, Ana Carolina Ruiz-Fern´ andez, Claudia Benitez-Nelson,Mar´ ıaLuisaMachain-Castillo,EricTappa,andJoan-AlbertSanchez-Cabeza. Centennial OMZ changes in the NW Mexican Margin from geochemical and foraminiferal sedimentary records. Continental Shelf Research, 176:64–75, March 2019. [174] Curtis Deutsch, William Berelson, Robert Thunell, Thomas Weber, Caitlin Tems, James McManus, John Crusius, Taka Ito, Timothy Baumgartner, Vicente Ferreira, Jacob Mey, and Alexander van Geen. Centennial changes in North Pacific anoxia linked to tropical trade winds. Science, 345(6197):665–668, August 2014. [175] Lothar Stramma, Gregory C. Johnson, Janet Sprintall, and Volker Mohrholz. Expanding Oxygen-Minimum Zones in the Tropical Oceans. Science, 320(5876):655–658, May 2008. 137 [176] L. Resplandy, M. L´ evy, G. Madec, S. Pous, O. Aumont, and D. Ku- mar. Contribution of mesoscale processes to nutrient budgets in the Ara- bian Sea. Journal of Geophysical Research: Oceans, 116(C11), 2011. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2011JC007006. [177] NatalyaEvans, JulianaTichota, JamesMoffett, andAllanDevol. Prolificnitritere-oxidation across the Eastern Tropical North Pacific Ocean, March 2022. Archive Location: world Publisher: Earth and Space Science Open Archive Section: Oceanography. [178] Natalya Evans, Isaac D. Schroeder, Mercedes Pozo Buil, Michael G. Jacox, and Steven J. Bograd. Drivers of Subsurface Deoxygenation in the Southern California Current System. Geophysical Research Letters, 47(21):e2020GL089274, 2020. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2020GL089274. [179] Bo Qiu, Shuiming Chen, and Hideharu Sasaki. Generation of the North Equatorial Under- current Jets by Triad Baroclinic Rossby Wave Interactions. Journal of Physical Oceanogra- phy, 43(12):2682–2698, December2013. Publisher: AmericanMeteorologicalSocietySection: Journal of Physical Oceanography. [180] Emilio Garcia-Robledo, Aurelien Paulmier, Sergey M. Borisov, and Niels Peter Revsbech. Sampling in low oxygen aquatic environments: The deviation from anoxic conditions. Limnology and Oceanography: Methods, n/a(n/a). eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/lom3.10457. [181] Fernando Contreras-Catala, Emilio J. Beier, Laura S´ anchez-Velasco, Victor M. God´ ınez, Elvia D. S´ anchez-P´ erez, and Eric D. Barton. Water masses and larval fish habitats in the Pacific tropical-subtropical convergence off Mexico. Continental Shelf Research, 230:104575, November 2021. [182] R. Castro, A. S. Mascarenhas, R. Durazo, and C. A. Collins. Seasonal variation of the temperatureandsalinityattheentrancetotheGulfofCalifornia,Mexico. Ciencias Marinas, 26(4):561–583, March 2000. [183] Michael D. Mastrandrea, Katharine J. Mach, Gian-Kasper Plattner, Ottmar Edenhofer, Thomas F. Stocker, Christopher B. Field, Kristie L. Ebi, and Patrick R. Matschoss. The IPCCAR5guidancenoteonconsistenttreatmentofuncertainties: acommonapproachacross the working groups. Climatic Change, 108(4):675–691, October 2011. Company: Springer Distributor: Springer Institution: Springer Label: Springer Number: 4 Publisher: Springer Netherlands. [184] John H. Martin, Steve E. Fitzwater, and R. Michael Gordon. Iron deficiency limits phyto- plankton growth in Antarctic waters. Global Biogeochemical Cycles, 4(1):5–12, March 1990. [185] Julia M. Vraspir and Alison Butler. Chemistry of Marine Ligands and Siderophores. Annual review of marine science, 1:43–63, 2009. [186] Werner Stumm. Equilibrium Concepts in Natural Water Systems. page 20. [187] Donald E. Canfield, Timothy W. Lyons, and Robert Raiswell. A model for iron deposition to euxinic Black Sea sediments. American Journal of Science, 296(7):818–834, September 1996. Publisher: American Journal of Science Section: Articles. 138 [188] Daniel C. Reed, Bo G. Gustafsson, and Caroline P. Slomp. Shelf-to-basin iron shuttling enhances vivianite formation in deep Baltic Sea sediments. Earth and Planetary Science Letters, 434:241–251, January 2016. [189] Peter L. Croot, Maija I. Heller, and Kathrin Wuttig. Redox Processes Impacting the Flux of Iron(II) from Shelf Sediments to the OMZ along the Peruvian Shelf. ACS Earth and Space Chemistry, March 2019. [190] Florian Scholz, Carolin R. L¨ oscher, Annika Fiskal, Stefan Sommer, Christian Hensen, Ulrike Lomnitz, Kathrin Wuttig, J¨ org G¨ ottlicher, Elke Kossel, Ralph Steininger, and Donald E. Canfield. Nitrate-dependent iron oxidation limits iron transport in anoxic ocean regions. Earth and Planetary Science Letters, 454:272–281, November 2016. [191] JagrutiVedamati,TylerGoepfert,andJamesW.Moffett. Ironspeciationintheeasterntrop- ical South Pacific oxygen minimum zone off Peru. Limnology and Oceanography, 59(6):1945– 1957, November 2014. [192] Mariko Hatta, Chris I. Measures, Jingfeng Wu, Saeed Roshan, Jessica N. Fitzsimmons, Pe- ter Sedwick, and Pete Morton. 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, June 2015. [193] Abigail E. Noble, Carl H. Lamborg, Dan C. Ohnemus, Phoebe J. Lam, Tyler J. Goepfert, Chris I. Measures, Caitlin H. Frame, Karen L. Casciotti, Giacomo R. DiTullio, Joe Jennings, and Mak A. Saito. Basin-scale inputs of cobalt, iron, and manganese from the Benguela- Angola front to the South Atlantic Ocean. Limnology and Oceanography, 57(4):989–1010, 2012. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.4319/lo.2012.57.4.0989. [194] D. K¨ onig, T. M. Conway, M. J. Ellwood, W. B. Homoky, and A. Tagli- abue. Constraints on the Cycling of Iron Isotopes From a Global Ocean Model. Global Biogeochemical Cycles, 35(9):e2021GB006968, 2021. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2021GB006968. [195] William B. Homoky, Tim M. Conway, Seth G. John, Daniela K¨ onig, FeiFei Deng, Alessandro Tagliabue, and Rachel A. Mills. Iron colloids dominate sedimentary supply to the ocean interior. Proceedings of the National Academy of Sciences, 118(13), March 2021. Publisher: National Academy of Sciences Section: Physical Sciences. [196] Kenneth S. Johnson, Francisco P. Chavez, and Gernot E. Friederich. Continental-shelf sedi- mentasaprimarysourceofironforcoastalphytoplankton. Nature,398(6729):697–700,April 1999. Number: 6729 Publisher: Nature Publishing Group. [197] C. A. Fuchsman, A. H. Devol, Z. Chase, C. E. Reimers, and B. Hales. Benthic fluxes on the Oregon shelf. Estuarine Coastal and Shelf Science, 163:156–166, September 2015. Patent Number: B Place: London Publisher: Academic Press Ltd- Elsevier Science Ltd WOS:000362604800018. [198] A. W. Dale, L. Nickelsen, F. Scholz, C. Hensen, A. Oschlies, and K. Wall- mann. A revised global estimate of dissolved iron fluxes from marine sediments. Global Biogeochemical Cycles, 29(5):691–707, 2015. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/2014GB005017. 139 [199] Virginia A. Elrod, William M. Berelson, Kenneth H. Coale, and Kenneth S. Johnson. The flux of iron from continental shelf sediments: A missing source for global budgets. Geophysical Research Letters, 31(12), 2004. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2004GL020216. [200] Curtis Deutsch, Hartmut Frenzel, James C. McWilliams, Lionel Renault, Faycal Kessouri, EvanHoward,Jun-HongLiang,DanieleBianchi,andSimonYang. Biogeochemicalvariability intheCaliforniaCurrentSystem. bioRxiv,page2020.02.10.942565,February2020. Publisher: Cold Spring Harbor Laboratory Section: New Results. [201] WilliamB.Homoky, ThomasWeber, WilliamM.Berelson, TimM.Conway, GideonM.Hen- derson, Marco van Hulten, Catherine Jeandel, Silke Severmann, and Alessandro Tagliabue. Quantifying trace element and isotope fluxes at the ocean–sediment boundary: a review. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374(2081):20160246, November 2016. Publisher: Royal Society. [202] T. D. Jickells, Z. S. An, K. K. Andersen, A. R. Baker, G. Bergametti, N. Brooks, J. J. Cao, P. W. Boyd, R. A. Duce, K. A. Hunter, H. Kawahata, N. Kubilay, J. laRoche, P. S. Liss, N. Mahowald, J. M. Prospero, A. J. Ridgwell, I. Tegen, and R. Torres. Global Iron Connec- tions Between Desert Dust, Ocean Biogeochemistry, and Climate. Science, 308(5718):67–71, April 2005. Publisher: American Association for the Advancement of Science. [203] Kenneth W. Bruland, Maeve C. Lohan, Ana M. Aguilar-Islas, Geoffrey J. Smith, Bettina Sohst, and Antonio Baptista. Factors influencing the chemistry of the near-field Columbia River plume: Nitrate, silicic acid, dissolved Fe, and dissolved Mn. Journal of Geophysical Research: Oceans, 113(C2), February 2008. [204] Michael S. Wetz, Burke Hales, Zanna Chase, Patricia A. Wheeler, and Michael M. Whitney. Riverine input of macronutrients, iron, and organic matter to the coastal ocean off Oregon, U.S.A., during the winter. Limnology and Oceanography, 51(5):2221–2231, 2006. eprint: https://aslopubs.onlinelibrary.wiley.com/doi/pdf/10.4319/lo.2006.51.5.2221. [205] Maeve C. Lohan and Kenneth W. Bruland. Elevated Fe(II) and Dissolved Fe in Hypoxic Shelf Waters off Oregon and Washington: An Enhanced Source of Iron to Coastal Upwelling Regimes. Environmental Science & Technology, 42(17):6462–6468, September 2008. [206] Katherine A. Adams, John A. Barth, and Francis Chan. Temporal vari- ability of near-bottom dissolved oxygen during upwelling off central Oregon. Journal of Geophysical Research: Oceans, 118(10):4839–4854, 2013. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/jgrc.20361. [207] Stephen D. Pierce, John A. Barth, R. Kipp Shearman, and Anatoli Y. Erofeev. Declining Oxygen in the Northeast Pacific*. Journal of Physical Oceanography, 42(3):495–501, March 2012. [208] S. A. Siedlecki, A. Mahadevan, and D. E. Archer. Mechanism for export of sediment-derived iron in an upwelling regime. Geophysical Research Letters, 39(3), 2012. [209] F. Chan, J. A. Barth, J. Lubchenco, A. Kirincich, H. Weeks, W. T. Peterson, and B. A. Menge. Emergence of Anoxia in the California Current Large Marine Ecosystem. Science, 319(5865):920–920, February 2008. 140 [210] Francis Chan, John Barth, Kristy Kroeker, Jane Lubchenco, and Bruce Menge. The Dynam- ics and Impact of Ocean Acidification and Hypoxia: Insights from Sustained Investigations in the Northern California Current Large Marine Ecosystem. Oceanography, 32(3):62–71, September 2019. [211] Jo¨ el Knoery, Daniel Cossa, Bastien Thomas, Germain Gregory, and Sylvain Rigaud. Susane, a device for sampling chemical gradients in the benthic wa- ter column. Limnology and Oceanography: Methods, 17(6):331–342, 2019. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/lom3.10317. [212] K. M. Bolster, M. I. Heller, and J. W. Moffett. Determination of iron(II) by chemilumi- nescence using masking ligands to distinguish interferences. Limnology and Oceanography: Methods, 16(11):750–759, November 2018. [213] S. L. Jackson, J. Spence, D. J. Janssen, A. R. S. Ross, and J. T. Cullen. Determination of Mn, Fe, Ni, Cu, Zn, Cd and Pb in seawater using offline extraction and triple quadrupole ICP-MS/MS. Journal of Analytical Atomic Spectrometry, 33(2):304–313, February 2018. Publisher: The Royal Society of Chemistry. [214] Insa Rapp, Christian Schlosser, Dagmara Rusiecka, Martha Gledhill, and EricP. Achterberg. AutomatedpreconcentrationofFe,Zn,Cu,Ni,Cd,Pb,Co,andMninseawaterwithanalysis using high-resolution sector field inductively-coupled plasma mass spectrometry. Analytica Chimica Acta, 976:1–13, July 2017. [215] Angela Milne, William Landing, Michael Bizimis, and Peter Morton. Determination of Mn, Fe, Co, Ni, Cu, Zn, Cd and Pb in seawater using high resolution magnetic sector inductively coupled mass spectrometry (HR-ICP-MS). Analytica Chimica Acta, 665(2):200–207, April 2010. [216] J. Magdalena Santana-Casiano, Melchor Gonz´ alez-D´ avila, and Frank J. Millero. Oxidation of nanomolar levels of Fe(II) with oxygen in natural waters. Environmental Science & Tech- nology, 39(7):2073–2079, April 2005. [217] Python Software Foundation. Python Language Reference, 3.7.13 Documentation. 2022. [218] P Raybaut. Spyder 5.1.5 documentation. 2009. [219] John D. Hunter. Matplotlib: A 2D Graphics Environment. Computing in Science & Engi- neering, 9(3):90–95, May 2007. Conference Name: Computing in Science & Engineering. [220] Ander Biguri. Perceptually uniform colormaps. MATLAB Central File Exchange, 2021. [221] Kristen Thyng, Chad Greene, Robert Hetland, Heather Zimmerle, and Steven DiMarco. True Colors of Oceanography: Guidelines for Effective and Accurate Colormap Selection. Oceanography, 29(3):9–13, September 2016. [222] C.A. Nittrouer, R.W. Sternberg, R. Carpenter, and J.T. Bennett. The use of Pb-210 geochronology as a sedimentological tool: Application to the Washington continental shelf. Marine Geology, 31(3-4):297–316, May 1979. [223] Anna Plass, Anna-Kathrin Retschko, Matthias T¨ urk, Tim Fischer, and Florian Scholz. A novel device for trace metal-clean sampling of bottom water and suspended particles at the ocean’slowerboundary: TheBenthicTraceProfiler. Limnology and Oceanography: Methods, 20(2):102–114, 2022. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/lom3.10473. 141 [224] Alexander Gal´ an, Marnie Jo Zirbel, Gonzalo S. Sald´ ıas, Francis Chan, and Ricardo Letelier. The role of upwelling intermittence in the development of hypoxia and nitrogen loss over the Oregon shelf. Journal of Marine Systems, 207:103342, July 2020. [225] S. A. Siedlecki, N. S. Banas, K. A. Davis, S. Giddings, B. M. Hickey, P. MacCready, T. Con- nolly, and S. Geier. Seasonal and interannual oxygen variability on the Washington and Oregon continental shelves. Journal of Geophysical Research: Oceans, 120(2):608–633, 2015. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1002/2014JC010254. [226] Jianping Gan and J. S. Allen. Modeling upwelling circulation off the Ore- gon coast. Journal of Geophysical Research: Oceans, 110(C10), 2005. eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1029/2004JC002692. [227] GiselleF.Firme,EdenL.Rue,DebraA.Weeks,KennethW.Bruland,andDavidA.Hutchins. Spatial and temporal variability in phytoplankton iron limitation along the California coast and consequences for Si, N, and C biogeochemistry. Global Biogeochemical Cycles, 17(1), March 2003. [228] C. M. Moore, M. M. Mills, K. R. Arrigo, I. Berman-Frank, L. Bopp, P. W. Boyd, E. D. Galbraith, R. J. Geider, C. Guieu, S. L. Jaccard, T. D. Jickells, J. La Roche, T. M. Lenton, N. M. Mahowald, E. Mara˜ n´ on, I. Marinov, J. K. Moore, T. Nakatsuka, A. Oschlies, M. A. Saito, T. F. Thingstad, A. Tsuda, and O. Ulloa. Processes and patterns of oceanic nutrient limitation. Nature Geoscience, 6(9):701–710, September 2013. [229] Jane Lubchenco, Bruce Menge, John Barth, Mark Carr, Jennifer Caselle, Francis Chan, Heather Fulton-Bennett, Steven Gaines, Kristy Kroeker, Kristen Milligan, Steven Palumbi, andJ.WilsonWhite. ConnectingSciencetoPolicymakers,Managers,andCitizens. Oceanog- raphy, 32(3):106–115, September 2019. [230] M. L. A. M. Campos, A. M. Farrenkopf, T. D. Jickells, and G. W. Luther. A comparison of dissolvediodinecyclingattheBermudaAtlanticTime-seriesStationandHawaiiOceanTime- series Station. Deep Sea Research Part II: Topical Studies in Oceanography, 43(2):455–466, January 1996. [231] V. W Truesdale and R Upstill-Goddard. Dissolved iodate and total iodine along the British east coast. Estuarine, Coastal and Shelf Science, 56(2):261–270, February 2003. [232] George W. Luther, Timothy Ferdelman, Charles H. Culberson, Joel Kostka, and Jingfeng Wu. Iodine chemistry in the water column of the Chesapeake Bay: Evidence for organic iodine forms. Estuarine, Coastal and Shelf Science, 32(3):267–279, March 1991. [233] RosieJ.Chance, LiselotteTinel, Tom´ asSherwen, AlexR.Baker, ThomasBell, JohnBrindle, Maria Lucia A. M. Campos, Peter Croot, Hugh Ducklow, He Peng, Frances Hopkins, Ba- bette Hoogakker, Claire Hughes, Timothy D. Jickells, David Loades, Dharma Andrea Reyes Macaya, Anoop S. Mahajan, Gill Malin, Daniel Phillips, Ieuan Roberts, Rajdeep Roy, Amit Sarkar,AlokKumarSinha,XiuxianSong,HelgeWinkelbauer,KathrinWuttig,MingxiYang, Zhou Peng, and Lucy J. Carpenter. Global sea-surface iodide observations, 1967–2018. Sci- entific Data , 6(1):286, November 2019. Number: 1 Publisher: Nature Publishing Group. [234] George W. Luther. The Role of One- and Two-Electron Transfer Reactions in Forming Thermodynamically Unstable Intermediates as Barriers in Multi-Electron Redox Reactions. Aquatic Geochemistry, 16(3):395–420, June 2010. 142 [235] Martin R. Wadley, David P. Stevens, Tim D. Jickells, Claire Hughes, Rosie Chance, Helmke Hepach, Liselotte Tinel, and Lucy J. Carpenter. A Global Model for Iodine Speciation in the Upper Ocean. Global Biogeochemical Cycles, 34(9):e2019GB006467, 2020. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2019GB006467. [236] D. S. Hardisty, T. J. Horner, N. Evans, R. Moriyasu, A. R. Babbin, S. D. Wankel, J. W. Moffett, and S. G. Nielsen. Limited iodate reduction in shipboard seawater incubations from the Eastern Tropical North Pacific oxygen deficient zone. Earth and Planetary Science Letters, page 116676, November 2020. [237] George T. F Wong and Peter G Brewer. The marine chemistry of iodine in anoxic basins. Geochimica et Cosmochimica Acta, 41(1):151–159, January 1977. [238] Henry Elderfield and Victor W. Truesdale. On the biophilic nature of iodine in seawater. Earth and Planetary Science Letters, 50(1):105–114, October 1980. [239] Zunli Lu, Hugh C. Jenkyns, and Rosalind E.M. Rickaby. Iodine to calcium ratios in marine carbonate as a paleo-redox proxy during oceanic anoxic events. Geology, 38(12):1107–1110, December 2010. [240] J.Podder,J.Lin,W.Sun,S.M.Botis,J.Tse,N.Chen,Y.Hu,D.Li,J.Seaman,andY.Pan. Iodate in calcite and vaterite: Insights from synchrotron X-ray absorption spectroscopy and first-principles calculations. Geochimica et Cosmochimica Acta, 198:218–228, February 2017. [241] Xiaolei Feng and Simon A. T. Redfern. Iodate in calcite, aragonite and vaterite CaCO3: Insights from first-principles calculations and implications for the I/Ca geochemical proxy. Geochimica et Cosmochimica Acta, 236:351–360, September 2018. [242] N. Glock, V. Liebetrau, and A. Eisenhauer. I/Ca ratios in benthic foraminifera from the Peruvian oxygen minimum zone: analytical methodology and evaluation as a proxy for re- dox conditions. Biogeosciences, 11(23):7077–7095, December 2014. Publisher: Copernicus GmbH. [243] Dalton S. Hardisty, Zunli Lu, Andrey Bekker, Charles W. Diamond, Benjamin C. Gill, Gan- qing Jiang, Linda C. Kah, Andrew H. Knoll, Sean J. Loyd, Magdalena R. Osburn, Noah J. Planavsky, Chunjiang Wang, Xiaoli Zhou, and Timothy W. Lyons. Perspectives on Protero- zoic surface ocean redox from iodine contents in ancient and recent carbonate. Earth and Planetary Science Letters, 463:159–170, April 2017. [244] Wanyi Lu, Andy Ridgwell, Ellen Thomas, Dalton S. Hardisty, Genming Luo, Thomas J. Al- geo, Matthew R. Saltzman, Benjamin C. Gill, Yanan Shen, Hong-Fei Ling, Cole T. Edwards, Michael T. Whalen, Xiaoli Zhou, Kristina M. Gutchess, Li Jin, Rosalind E. M. Rickaby, Hugh C. Jenkyns, Timothy W. Lyons, Timothy M. Lenton, Lee R. Kump, and Zunli Lu. Late inception of a resiliently oxygenated upper ocean. Science, page eaar5372, May 2018. [245] Wanyi Lu, Catia F. Barbosa, Anthony E. Rathburn, Priscila da Matta Xavier, Anna P. S. Cruz, Ellen Thomas, Rosalind E. M. Rickaby, Yi Ge Zhang, and Zunli Lu. Proxies for paleo-oxygenation: A downcore comparison between benthic foraminiferal surface porosity and I/Ca. Palaeogeography, Palaeoclimatology, Palaeoecology, 579:110588, October 2021. [246] Wanyi Lu, Yi Wang, Delia W. Oppo, Sune G. Nielsen, and Kassandra M. Costa. Comparing paleo-oxygenation proxies (benthic foraminiferal surface porosity, I/Ca, authigenic uranium) 143 onmodernsedimentsandtheglacialArabianSea. GeochimicaetCosmochimicaActa,331:69– 85, August 2022. [247] Helge Winkelbauer, Kathy Cordova-Rodriguez, Dharma Reyes-Macaya, Jennifer Scott, Nico- laas Glock, Zunli Lu, Elliott Hamilton, Simon Chenery, Phil Holdship, Charlotte Dor- mon, and Babette Hoogakker. Foraminifera Iodine to Calcium Ratios: Approach and Cleaning. Geochemistry, Geophysics, Geosystems, 22(11):e2021GC009811, 2021. eprint: https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1029/2021GC009811. [248] A. M. Farrenkopf, G. W. Luther, V. W. Truesdale, and C. H. Van Der Weijden. Sub- surfaceiodidemaxima: evidenceforbiologicallycatalyzedredoxcyclinginArabianSeaOMZ during the SW intermonsoon. Deep Sea Research Part II: Topical Studies in Oceanography, 44(6):1391–1409, January 1997. [249] V. W. Truesdale and G. W. Bailey. Dissolved Iodate and Total Iodine During an Extreme Hypoxic Event in the Southern Benguela System. Estuarine, Coastal and Shelf Science, 50(6):751–760, June 2000. [250] Victor Reyes-Umana, Zachary Henning, Kristina Lee, Tyler P. Barnum, and John D. Coates. Geneticandphylogeneticanalysisofdissimilatoryiodate-reducingbacteriaidentifiespotential nichesacrosstheworld’soceans. The ISME Journal,pages1–12,July2021. Bandiera abtest: a Cc license type: cc by Cg type: Nature Research Journals Primary atype: Research Pub- lisher: Nature Publishing Group Subject term: Biogeochemistry;Microbial ecology Sub- ject term id: biogeochemistry;microbial-ecology. [251] Seigo Amachi, Nahito Kawaguchi, Yasuyuki Muramatsu, Satoshi Tsuchiya, Yuko Watan- abe, Hirofumi Shinoyama, and Takaaki Fujii. Dissimilatory Iodate Reduction by Marine Pseudomonas sp. Strain SCT. Applied and Environmental Microbiology, 73(18):5725–5730, September 2007. [252] Zhang Jia-Zhong and Michael Whitfield. Kinetics of inorganic redox reactions in seawater: I. The reduction of iodate by bisulphide. Marine Chemistry, 19(2):121–137, May 1986. [253] R.C.TianandE.Nicolas. IodinespeciationinthenorthwesternMediterraneanSea, method and vertical profile. Marine Chemistry, 48(2):151–156, January 1995. [254] F.G´ omez-Valdivia,A.Par´ es-Sierra,andA.L.Flores-Morales. TheMexicanCoastalCurrent: A subsurface seasonal bridge that connects the tropical and subtropical Northeastern Pacific. Continental Shelf Research, 110:100–107, November 2015. [255] J. Michael Beman, Joy Leilei Shih, and Brian N. Popp. Nitrite oxidation in the upper water column and oxygen minimum zone of the eastern tropical North Pacific Ocean. The ISME Journal, 7(11):2192–2205, November 2013. Bandiera abtest: a Cg type: Nature Research Journals Number: 11 Primary atype: Research Publisher: Nature Publishing Group Sub- ject term: Biogeochemistry;Marine microbiology Subject term id: biogeochemistry;marine- microbiology. [256] Moritz F. Lehmann, Daniel M. Sigman, Daniel C. McCorkle, Julie Granger, Sharon Hoff- mann, Greg Cane, and Brigitte G. Brunelle. The distribution of nitrate 15N/14N in marine sediments and the impact of benthic nitrogen loss on the isotopic composition of oceanic nitrate. Geochimica et Cosmochimica Acta, 71(22):5384–5404, November 2007. 144 [257] GeorgeW.Luther, AlyssaJ.Findlay, DanielJ.MacDonald, ShannonM.Owings, ThomasE. Hanson, Roxanne A. Beinart, and Peter R. Girguis. Thermodynamics and Kinetics of Sul- fide Oxidation by Oxygen: A Look at Inorganically Controlled Reactions and Biologically Mediated Processes in the Environment. Frontiers in Microbiology, 2, 2011. [258] Harald Schunck, Gaute Lavik, Dhwani K. Desai, Tobias Großkopf, Tim Kalvelage, Car- olin R. L¨ oscher, Aur´ elien Paulmier, Sergio Contreras, Herbert Siegel, Moritz Holtappels, Philip Rosenstiel, Markus B. Schilhabel, Michelle Graco, Ruth A. Schmitz, Marcel M. M. Kuypers, and Julie LaRoche. Giant Hydrogen Sulfide Plume in the Oxygen Minimum Zone off Peru Supports Chemolithoautotrophy. PLOS ONE, 8(8):e68661, August 2013. [259] M. R. Raven, R. G. Keil, and S. M. Webb. Microbial sulfate reduction and organic sul- fur formation in sinking marine particles. Science, December 2020. Publisher: American Association for the Advancement of Science Section: Report. [260] Erik Kristensen, Allan H. Devol, and Hilairy E. Hartnett. Organic matter diagenesis in sediments on the continental shelf and slope of the Eastern Tropical and temperate North Pacific. Continental Shelf Research, 19(10):1331–1351, August 1999. [261] Thomas Ohde and Isabelle Dadou. Seasonal and annual variability of coastal sulphur plumes in the northern Benguela upwelling system. PLOS ONE, 13(2):e0192140, February 2018. Publisher: Public Library of Science. [262] Eric D’Asaro, Mark Altabet, N. Suresh Kumar, and M. Ravichandran. Structure of the Bay ofBengaloxygendeficientzone. Deep Sea Research Part II: Topical Studies in Oceanography, 179:104650, September 2020. [263] A. Mazumdar, A. L. Paropkari, D. V. Borole, B. R. Rao, N. H. Khadge, S. M. Karisiddaiah, M. Kocherla, and Hilda M. Jo¨ ao. Pore-water sulfate concentration profiles of sediment cores fromKrishna-GodavariandGoabasins, India. GEOCHEMICAL JOURNAL,41(4):259–269, 2007. 145 Appendix 1 Drivers of Subsurface Deoxygenation in the Southern California Current System 1.1 Abstract A confluence of subarctic, tropical, and subtropical water masses feed the California Current Sys- tem(CCS),supportingahighlyproductiveecosystemandwidearrayofmarineecosystemservices. Long-term declines in oxygen have been observed in this region, causing habitat compression and other ecosystem consequences. Here, we quantify the water masses and processes causing deoxy- genation in the subsurface CCS from 1993-2018, and we find that deoxygenation was caused both by changes in the advection of source waters and increased remineralization in the source waters. The historical deoxygenation trend can be attributed primarily (81%) to the Northern Equatorial Pacific Intermediate Water, the deep Pacific Equatorial Water mass transported in the California Undercurrent. We also find that advection and remineralization share nearly equal contributions to deoxygenation. This improved understanding of the mechanisms affecting the aerobic habitat of the CCS will inform projections of ecological impacts and mitigation of future deoxygenation. 1.2 Introduction The California Current System (CCS), a coastal upwelling biome off the west coast of the United States, supports a highly productive ecosystem and a wide array of marine ecosystem services. These ecosystem services include storm protection, pollution control, food, recreation, and carbon sequestration (Barbier, 2017). Climate change is expected to impact the CCS by raising tem- peratures, lowering oxygen concentrations, and causing more acidic waters, all of which change ecosystem productivity, structure, and function (Deutsch et al., 2015; Howard et al., 2020; Koslow et al., 2013). Off southern California, long oceanographic time series have documented significant physical and biogeochemical variability; trends in subsurface warming and deoxygenation have corresponded with enriched nutrients and enhanced acidification (Bograd et al., 2008, 2015; Mc- Clatchie et al., 2010; Meinvielle & Johnson, 2013; Nam et al., 2015; others). This biogeochemical variabilityisconsistentwiththeeffectsexpectedfromclimatechange: warmerwatersreducingoxy- gen solubility, increasing water column stratification, limiting ocean ventilation, and accelerating oxygen consumption via respiration (Breitburg et al., 2018; Brewer & Peltzer, 2016; Doney et al., 2012; Gruber, 2011; Keeling et al., 2010; Levin, 2018). While several studies have suggested that the observed biogeochemical trends are associated with changes in the properties and/or advection of source waters to the CCS (Bograd et al., 2015, 2019), the explicit processes and water masses causing deoxygenation have not been quantified. A recent study performed an extended optimum multiparameter analysis (eOMP) (Tomczak & Large, 1989) on the long-term hydrographic data from the California Cooperative Oceanic Fish- eries Investigations (CalCOFI) program to quantify the spatiotemporal variability of source water contributions to the southern CCS (Bograd et al., 2019). That study described the distribution of six source water masses that feed the southern CCS and found significant long-term variability in the relative contribution and distribution of upper and deeper Pacific Equatorial Waters. These water masses are transported to the CCS within the California Undercurrent (CUC), and they are relatively low in oxygen relative to the other water masses (Borgrad et al., 2019, Meinvielle and Johnson, 2013). In this study, we extended this eOMP by framing deoxygenation as a function 146 of water mass contributions, which allowed us to quantify the relative contributions of different processes, specifically changes in the advection of source water masses versus remineralization ac- companying these source water masses. Improving our understanding of deoxygenation in the CCS will highlight priorities for simulation developments that can better project biogeochemistry in the region, informing potential mitigation strategies. 1.3 Methods 1.3.1 Data from the southern California Current System We used historical hydrographic data from the CalCOFI program, which has consistently sampled the southern CCS from San Diego to Pt. Conception quarterly since 1984, with target months of January, April, July, and October (Fig 1), amounting to more than 6,700 station occupations from 1984-2018. Variables obtained include temperature (T), salinity (S), oxygen (O 2 ), nitrate (NO − 3 ), phosphate (PO 3− 4 ), and silicate (SiO 2− 4 ). CalCOFI sampling extends to 500 m or the continental shelf, if shallower. Further details can be found in any of the CalCOFI data reports (e.g., Scripps Institution of Oceanography, 2012) or online at www.calcofi.org. 1.3.2 Optimum multiparameter analysis in CCS OMP analysis (Tomczak & Large, 1989) estimates the relative contributions of source water types that mix together to form an observed water mass, assuming that water mass properties are quasi- conservative. In extended OMP (eOMP) analysis, nutrient concentrations for each water mass are set at a location upstream of the location of interest (Anderson & Sarmiento, 1994; Brzezinski, 1985; Karstensen & Tomczak, 1997), and the change in phosphate, ∆P, accounts for nutrient concentration changes along the advective pathway (Garc´ ıa-Ib´ a˜ nez et al., 2015; Poole & Tomczak, 1999). ∆P is related to changes in oxygen, nitrate, and silicate through Redfield ratios (Anderson & Sarmiento, 1994; Brzezinski, 1985) that occur due to organic matter remineralization. Bograd et al. (2019) performed eOMP on the 1984-2018 CalCOFI data. They first defined six source water masses using three 10°×10° boxes centered at 45°N, 135°W with depths of 101 m and 178 m; 27°N, 139°W with depths of 85 m and 219 m; and 5°N, 108°W with depths of 131 m and 399 m, corresponding to Pacific Subarctic Upper Waters (PSUW), Eastern North Pacific Central Waters (ENPCW), and Pacific Equatorial Waters (PEW), respectively. Each of these three locations has an upper and a deeper water mass, denoted uPSUW and dPSUW, respectively. They obtained available data within each source water box from the World Ocean Database 2018. While these three boxes act as the source regions for water masses entering the CCS, these boxes are not the ultimate formation regions for these six water masses. Withsixsourcewatermasses,definedastime-invariantusingWOD18datafrom1984-2018,and six input variables (T, S, O 2 , NO − 3 , PO 3− 4 , and SiO 2− 4 ), Bograd et al. (2019) solved for the relative contributions of each water mass at each CalCOFI station, depth, and sampling time. During these calculations, ∆P is estimated using the equations in Bograd et al. (2019) for each sample, such that each station and depth has its own ∆P value. This ∆P term allows us to calculate the extent that remineralization has accumulated in source water masses to the CalCOFI region after water mass formation. To deconvolute the drivers of this deoxygenation signal, we averaged water mass contribution for each sample from 100 to 400 m depth across every station for a given season and a given year. Beyond these depths, the eOMP error was greater than 10%, and therefore these samples were omitted from this analysis. Since there is significant seasonal variability in the subsurface CCS (Nam et al., 2015), we analyzed each season individually for long-term trends. Here we note that the upper and deeper PEW source waters used in Bograd el al. (2019, Sup- plementary Table 1) are more broadly referred to as 13ºC Water (13CW) and Northern Equatorial PacificIntermediateWater(NEPIW),followingEvansetal. (2020). Wewillusethisnomenclature 147 as it better represents these water masses within the Pacific Ocean and links them to other papers that describe their sources and transport (Bostock et al., 2010; Fiedler & Talley, 2006; Qu et al., 2009; Tsuchiya, 1981). Evans et al. (2020) contains additional references discussing these water masses. -220 -200 -180 -160 -140 -120 -100 -80 Longitude (°) 0 10 20 30 40 50 60 Latitude (°) 0 50 100 150 200 250 O 2 ( M) North Equatorial Countercurrent Equatorial Undercurrent North Pacific Current California Current CalCOFI Survey Grid California Undercurrent California Current ENPCW PSUW PEW North EUC jet A B Longitude/ °W Latitude/ °N Longitude/ °W Latitude/ °N depth/ m Salinity Temperature / °C 32.5 33.0 33.5 34.0 34.5 35.0 35.5 5 10 15 20 25 30 13CW NEPIW Figure 1.1: Map of the annual climatology of O 2 at 300 m in the North Pacific for the period 1955-2017, withtheregionswherewedefinedupperanddeeperPSUW(bluebox), ENPCW(green box), and PEW (red box). The map illustrates the main surface (white) and subsurface (yellow dotted) currents. O 2 data obtained from WOA18 (Garcia et al., 2019). The inset map highlights thenominalCalCOFIsurveygridandbathymetryoftheregion,obtainedfromETOPO2(National GeophysicalDataCenter,2006). AttachedonthebottomrightistheTSdiagramforPSUW(blue), ENPCW (green), and PEW or 13CW/NEPIW (red), modified from Bograd et al. (2019). 1.3.3 Linking water masses to deoxygenation eOMP deconvolutes observed hydrographic properties into a ratio of source water masses and an amount of remineralization. The ratio of source water masses sets the maximum possible O 2 while remineralization lowers it to the observed value. In this paper, advective (de)oxygenation refers to changes in O 2 caused only by changing the ratio of source water contributions. While eOMP does not attribute remineralization to specific water masses, this was a primary goal for this analysis. For this purpose, we separated the OMP-derived remineralization into two categories: internal and advection-derived. Internal remineralization refers to ∆P that has accumulated within each water mass, independent of each other, between their formation regions and the CCS. Advection-derived remineralization refers to a change in the amount of ∆P accumulated within the CCS because the ratio of water mass contributions has changed. Our results indicate that the source water masses experience different amounts of ∆P between their defined source regions and the CCS, and as advection alters the ratio of water mass contributions, it also alters the ∆P observed in the CCS as a function of these contributions. Eq 1 depicts the deconvolution we performed to identify the causes of deoxygenation in the southern CCS, where the first term refers to advective 148 (de)oxygenation, whereas the second and third terms are internal and advection-derived reminer- alization, appropriately. Bold variables represent matrices, whereas plain text variables represent scalars or vectors. d[O 2 ] dt = ∂x ∂t O 2,x + − 170µMO 2 µM ∆ P ∂(∆ P) ∂t x + ∂(∆ P) ∂x i ∂x ∂t (1.1) To determine advective (de)oxygenation, we calculated the theoretical O 2 concentration that would be present in the CCS given the contributions of each water mass without remineralization. We compared this to the O 2 concentration if one of the water masses did not contribute. While completely omitting one of these water masses is unrealistic for the CCS, these calculations are intended to generate a long-term trend for how changes in source water mass advection change the concentration of O 2 , rather than an absolute value. RemineralizationlowerstheO 2 concentrationduringthetransportofeachwatermassaswellas locally in the CCS. Within a water mass, increased remineralization due to elevated temperatures is the most prevalent cause of deoxygenation (Brewer & Peltzer, 2016), but changes in production, export, and surface warming can also alter properties within a water mass (Rykaczewski & Dunne, 2010). Mathematically, eOMP groups these processes into the ∆P term, and stoichiometrically scaling this parameter by -170 O 2 /PO 4 provides the deoxygenation from remineralization, as this scaling factor was implemented in the OMP in Bograd et al. (2019). For each season and year, we averaged a given property, such as O 2 (Fig. 1) or the water mass contribution, across all stations and depths from 100-400 m, and we presented these values as time series for each season. Thedeoxygenationfromaccumulatedinternalremineralizationwithineachwatermassisquan- tified by tracking how much remineralization varies over time in the core of each water mass. Since we deconvoluted these water masses from CCS data and the maximum contribution of each water mass differs significantly, we analyzed the amount of O 2 lost from remineralization at the same quantile for each water mass. Quantiles are useful because the contributions of NEPIW and dP- SUW changed monotonically across this time period, so we must implement a threshold below the absolute maxima of each water mass. We analyzed remineralization between the 84% and 85% quantiles to ensure all years of the analysis were represented, and justifications for this quantile selection are discussed in detail within the SI. For these calculations, samples with water mass contributions corresponding to 84%-85% of each water mass were selected, and the ∆P calculated for these samples was averaged and scaled stoichiometrically to O 2 . Deoxygenation from advection-derived remineralization was calculated by deconvoluting the OMP-derived ∆P term into components for each water mass. Each water mass component was then regressed linearly over time to determine how changes in the advection of that water mass intotheCCSinfluencedtheaccumulatedremineralization. Yearswithgreaterthan20%errorfrom the initial OMP-derived remineralization were omitted, as presented in Fig S5c. These data were deconvoluted using singular value decomposition (SVD), as described in Glover et al. (2011), with the source water mass contributions. All linear regressions performed in this paper were Type 1 using code provided in Glover et al. (2011), and linear regressions were performed on combined data from all seasons. 1.4 Results 1.4.1 Impact of advection versus remineralization on deoxygenation In the southern, subsurface CCS between 100 m and 400 m, we found that 0.97±0.09µ M year − 1 of deoxygenation occurred from 1993-2018, and this value matches with depth-dependent estimates in Bograd et al. (2008). 149 1980 1990 2000 2010 2020 Year -70 -65 -60 -55 -50 -45 -40 -35 Average O 2 change from remineralization/ M OMP with all water masses 1980 1990 2000 2010 2020 Year -50 -45 -40 -35 -30 -25 -20 -15 -10 Average O 2 change from advection/ M NEPIW 1980 1990 2000 2010 2020 Year 55 60 65 70 75 80 85 90 95 100 105 Average O 2 / M CalCOFI data b. a. c. Deconvolution of the remineralization in OMP Advection vs. remineralization in the CCS deoxygenation 1980 1990 2000 2010 2020 Year -25 -20 -15 -10 -5 0 Average O 2 at 84%-85% quantile/ M dPSUW 1980 1990 2000 2010 2020 Year -35 -30 -25 -20 -15 -10 -5 0 Average O 2 change from advection-derived remineralization/ M NEPIW d. e. f. 1980 1990 2000 2010 2020 Year -35 -30 -25 -20 -15 -10 -5 0 Avera NEPIW 1980 1990 2000 2010 2020 Year -25 -20 -15 -10 -5 0 Average O 2 at 84%-85% quantile/ M NEPIW Winter Spring Summer Fall Best annual fit Figure 1.2: a) Mean O 2 measured in the CalCOFI study region between 100 and 400 m. Oxy- genation occurs from 1985-1993 at 2.3±0.6 µ M year − 1 and deoxygenation occurs from 1993-2018 at -0.97±0.09 µ M year − 1 . The deoxygenation trend in 2a is reconstructed by both b) mean ad- vective (de)oxygenation from NEPIW with a long-term trend of -0.487±0.001 µ M year − 1 and c) deoxygenation from remineralization calculated in eOMP with a long-term trend of -0.44±0.04 µ M year − 1 . Since eOMP does not attribute remineralization to specific water masses, 2c is further deconvoluted in 2d-f. d) O 2 lost through accumulated internal remineralization in dPSUW with a long-term trend of -0.12±0.03 µ M year − 1 and e) O 2 lost through internal remineralization in NEPIW with a long-term trend of -0.20±0.03 µ M year − 1 . f) Deoxygenation caused by water mass advection-derived remineralization from NEPIW with a long-term trend of -0.11±0.04 µ M year − 1 . Error bars for 2a consist of the uncertainty in the water mass contribution combined with the uncertainty in the NEPIW O 2 water mass definition, but these error bars are too small to be seen. Uncertainties were not determined for 2b and values less than 40 µ M were not included in the fit. Error bars are not presented for this 2c-e. Error bars in 2c present the uncertainty in the SVD fit propogated through the water mass contribution. NEPIW is the only water mass driving advective (de)oxygenation, as it has the lowest O 2 con- centration of the six water masses and the contribution of this water mass monotonically increases. IncreasingNEPIWadvectioncaused-0.487±0.001µ Myear − 1 ofdeoxygenation(Fig2b). Thetotal deoxygenation in the CCS can be reconstructed by summing the -0.44±0.04 µ M year − 1 of deoxy- genation from accumulated remineralization calculated in the eOMP (Fig 2c) and the advective (de)oxygenation from NEPIW (Fig 2a). This reconstructed total deoxygenation is -0.92±0.04 µ M year − 1 , which is statistically similar to the -0.97±0.09 µ M year − 1 observed from 1993-2018 in the O 2 data, presented in Fig 2a. This treatment matches the observed deoxygenation and increased advection of NEPIW is responsible for 50% of the deoxygenation from 1993-2018, such that it shares nearly equal contribution with accumulated remineralization. Oxygenation from 1984-1993 can be attributed to the advection of both PSUW water masses, which have the highest O 2 concentrations of the six water masses. This oxygenation trend is most clear in uPSUW from 1984-1993, though dPSUW reached a relative minimum in deoxygenation in the late 1980s. Fitting these individual time periods yields oxygenation trends of 0.62±0.07 µ M year − 1 for uPSUW and 1.1±0.5 µ M year − 1 for dPSUW, causing a net oxygenation of 1.7±0.6 µ M year − 1 , compared to the 2.3±0.6 µ M year − 1 calculated from the CalCOFI data set in Fig 2a. The magnitude of dPSUW suggests that it is primarily responsible for this oxygenation trend, however, 150 it leads to deoxygenation after 1989, unlike the 1984-1993 trend observed in the CalCOFI data. While the magnitude of oxygenation can be explained via these two water masses, the scatter in dPSUW deoxygenation undercuts this conclusion, such that we cannot fully explain the early oxygenation trend with these results. Fig S2 and table S2 expand on this information. This trend of CCS oxygenation sharply switching to deoxygenation at 1993 is remarkably similar to that of ∆ δ 15 N measurements of particulate organic nitrogen in sediments from Baja California and the Santa Monica Basin (Deutsch et al. 2014). These measurements were used as a proxy for the size of the Eastern Tropical North Pacific Oxygen Deficient Zone (ETNP ODZ), and Deutsch et al. attributed these ∆ δ 15 N-PON trends to equatorial wind stress weakening, then intensifying. Based on the similarity of this data to O 2 concentrations in the CCS and the connection of these regions, this equatorial wind stress likely influences the CCS as well and is worth further investigation. 1.4.2 Deconvoluting accumulated remineralization within water masses The eOMP results suggest that remineralization has accumulated most in the deepest waters, par- ticularlybelow300m(FigS2a). ThisdepthrangecorrespondstodPSUWandNEPIW(FigS2b,c), so internal remineralization within the dPSUW and NEPIW are most significant to accumulated remineralization in the CCS. Internal remineralization within other water masses has less influence on the average O 2 concentration of the region but could important in specific locations. At the 84%-85% quantile, the dPSUW water mass contributes 81%, while the NEPIW water mass only contributes 37%, as seen in Fig S3. The accumulation of internal remineralization in the dPSUW has increased monotonically dur- ing the study time (Fig 2d). This process appears more seasonally variable in the NEPIW (Fig 2e) than the dPSUW, and other non-monotonic factors appear to influence it in NEPIW, as seen by the lower scatter from 1994-2004. The increased accumulation of internal remineralization in the NEPIWisapproximately1.5xthatofthedPSUW,whichcouldbeafunctionoftheorganiccarbon flux into these water masses or the temperature difference between the two water masses. The NEPIW is transported into the southern CCS from the shadow zone around the ETNP ODZ. This internal remineralization likely occurs in that region off Mexico, since the circulation is sluggish and the residence time of this water mass is long. 1.4.3 Deoxygenation from water mass advection-derived remineralization EachofthesourcewatermassesenteringthesouthernCCSaccumulatesadifferentamountofrem- ineralization as it is advected from the source location to the CCS. Therefore, changing the ratio of water mass contributions to the CCS also changes the accumulated remineralization calculated in eOMP. NEPIW was found to be the only significant contributor to advection-derived reminer- alization with -0.11±0.04 µ M year − 1 (Fig 2f). The advection-derived remineralization and the accumulated internal remineralization calculated in section 3.2 sum to -0.42±0.05 µ M year − 1 while theeOMPcalculated-0.44±0.04µ Myear − 1 ofdeoxygenation,suggestingthatthesedeconvolutions reproduce the deoxygenation from eOMP remineralization with statistical significance. 1.5 Discussion Climate change-driven deoxygenation is a significant stressor on the global ocean, with eastern boundary upwelling systems being particularly vulnerable regions. The consequences of deoxy- genationhavealreadybeenobservedwithintheCCS(Koslowetal., 2011; Koslowetal., 2013). We analyzed the drivers of deoxygenation in the CCS through water mass deconvolution of the Cal- COFIdataset(Bogradetal., 2019). WefoundthattheaverageO 2 concentrationinthesubsurface CCSincreasedby20±5µ Mfrom1985-1993anddecreasedby24±2µ Mfrom1993-2018. Byquanti- fyingdeoxygenationasafunctionofwatermasses,weobservedthatadvectionandremineralization share nearly equal impact in causing deoxygenation in the CCS. 151 Latitude/ °N Longitude/ °W Depth/ m Internal remineralization -0.12 ± 0.03 Advection Advection-derived remineralization O2/ !M Internal remineralization -0.20 ± 0.03 Advection -0.487 ± 0.001 Advection-derived remineralization -0.11 ± 0.04 NEPIW -0.97 ± 0.09 μM O2 year -1 -120 -125 -130 25 30 35 40 45 -400 -300 -200 -100 0 50 100 150 200 250 300 dPSUW Figure 1.3: Summary schematic for the processes and contributions of dPSUW and NEPIW on deoxygenation in the CCS, with the amount of deoxygenation written next to each relevant process. On the left, all three processes are linked to dPSUW but it only impacts deoxygenation through accumulated internal remineralization, hence the solid arrow. On the right, NEPIW ac- tivelycontributestodeoxygenationthroughallthreeprocesses. Advection-derivedremineralization is highlighted in light blue because the OMP deconvolutes it as remineralization, but it depends on increased NEPIW advection from 1984-2018. Fig 3 illustrates the processes and water masses that have caused deoxygenation in the CCS from 1993-2018. NEPIW is responsible for 81% it caused -0.78±0.05 µ M year − 1 of deoxygenation from 1984-2018 with elevated summer impacts. The average contribution of NEPIW to the CCS has increased monotonically since 1984 and this increased advection has caused -0.487±0.001 µ M year − 1 of deoxygenation and contributed to -0.11±0.04 µ M year − 1 . The residual -0.19±0.03 µ M year − 1 of NEPIW-driven deoxygenation has been caused by accumulated internal remineralization within this water mass. 1.5.1 Water mass analysis for high-resolution nitrite re-oxidation For our regression analysis, we binned data into water masses to enabled calculation nitrite re- oxidation percentages from a large number of measurements, thus, increasing the precision of our calculations. Unfortunately, thisapproachlimitstheresolutionofthedataproduct,astherequired large number of measurements prevents comparisons between geographic coordinates. Binning by water masses also removes vertical structure. To create a higher resolution estimate of nitrite re-oxidation in the ETNP ODZ, we implemented a novel approach to water mass analysis. While increasing surface ocean temperatures from climate change could cause elevated rem- ineralization rates, this factor is likely not responsible for the increased accumulation of internal remineralization within the NEPIW, as transit times for NEPIW to reach the ETNP are too long (¿130 years; Karstensen et al. 2008) for recent warming trends to have manifested. Instead, the internal remineralization observed in the CCS likely occurs nearby, off Mexico. The accumulation rates of internal remineralization in the uPSUW and dPSUW are similar, suggesting processes in the Northeastern Pacific drive the trends for these two water masses (Durski et al. 2017). We suggest that increased subsurface stratification (Rykaczewski & Dunne, 2010) within these waters may be responsible for this deoxygenation trend. From 1984-2018, the time series of NEPIW contribution on the southern CCS closely fits a linear trend (Fig 2b), even though oscillations such as ENSO impacts the NEPIW contribution by approximately 10% (Bograd et al., 2019). It is likely that our data treatment averaged out 152 the impact of this oscillation. Therefore, repeating this type of deoxygenation deconvolution for specific lines and stations of interest in the CCS will likely provide more nuanced information. In the northern CCS, subsurface deoxygenation appears to be more severe (Chan et al., 2008; Connolly et al., 2010). Internal remineralization accumulated within the NEPIW causes 19% of the deoxygenation in the southern CCS, and this deoxygenation may be greater in the northern CCSduetoincreasedremineralizationthatoccursduringthetransitofthiswatermassintheCUC. Nevertheless, the deoxygenation in the northern CCS may be driven more by upwelling of low O 2 waters than remineralization inside a specific water mass. We recommend replicating the water mass analysis performed in Bograd et al. (2019) and the deoxygenation deconvolution described in this paper for other time series in the CCS, such as the Newport Hydrographic line off Oregon, as wellastheHumboldt,Benguela,andCanaryEasternBoundaryUpwellingSystems. TheHumboldt system shares the same uPEW and similar dPEW as the CCS (Evans et al. 2019), facilitating this investigation. This study aimed to quantify the relative contributions of advection, remineralization, and the six source water masses in driving deoxygenation in the southern, subsurface CCS. Laffoley & Baxter (2019) estimated that the concentration threshold for coastal low oxygen zones is 60 µ M. If equatorial wind stress continues to increase (Deutsch et al. 2014) and this causes deoxygenation to continuelinearlyattheratescalculatedinthisstudy,the150-325mdepthrangeoftheNEPIWwill reachthisthresholdin2034. WatersdeeperthanthisthresholdhaveminimalNEPIWcontribution and therefore will deoxygenate slower than this rate. We have revealed that in the southern, subsurfaceCCS,81%oftheobserveddeoxygenationfrom1993-2018occurredthroughthesouthern boundary and at the depths of the NEPIW. This improved understanding of the mechanisms affectingtheaerobichabitatoftheCaliforniaCurrentSystemwillinformbiogeochemicalmodeling, allowing improved projections of ecological impacts and mitigation of future deoxygenation. 1.6 Acknowledgements We acknowledge the quality and longevity of the CalCOFI program and the many scientists and seagoing staff who have contributed to the collection, processing, and analysis of this excellent dataset. We also acknowledge the California Current Ecosystem Long-Term Ecosystem Research (CCE-LTER) project, supported by a grant from NSF (OCE-0417616). We would like to acknowl- edge Donald M. Glover, William J. Jenkins, and Scott C. Doney for writing Modeling Methods for Marine Science and Naomi Levine for teaching her course on MATLAB, which included this textbook. We appreciate clarifying feedback by Ryan Rykaczewski during internal NOAA review and Jack Barth for review of this submission. N.E conceived of this project and performed the analysis,afterbeinggivendatafromS.J.B.,I.D.S,M.G.J.,andM.P.B.N.Ewrotemostofthepaper withcontributionsfromS.J.B.S.J.B., I.D.S,M.G.J., andM.P.Bprovidedfeedbackandsuggestions throughout this process. The authors declare no conflicts of interest. 1.7 Data availability This study did not use any new data, rather, it re-processed the CalCOFI data set. The original CalCOFI data are available at http://www.calco.org. The MATLAB OMP toolbox used to generate water mass contributions in Bograd et al. (2019) is available at https://www.mathwork s.com/matlabcentral/leexchange/1334ompanalysis. The MATLAB code for linear regression can be found as attached electronic content for Doney et al. (2012) at https://www.cambridge. org/us/academic/subjects/earth-and-environmental-science/oceanography-and-marine -science/modeling-methods-marine-science?format=HB. 153
Abstract (if available)
Abstract
Climate change is driving deoxygenation across the world's oceans. This deoxygenation restructures ecosystem energy transfer, foraging habitats, and the abundance of both critical and toxic compounds in the water column. In the absence of oxygen, microbes respire using terminal electron acceptors such as iodate, nitrate, ferric (oxy)hydroxides, and sulfate. These metabolisms produce reduced compounds such as iodide, nitrite, Fe(II), and sulfide, which have different properties and distributions in the water column. The research in this dissertation analyzes the causes and interconnections in the distribution of these reduced compounds.
Nitrite, an intermediate in dissimilatory nitrate reduction, indicates that this metabolism is occurring in marine Oxygen Deficient Zones (ODZs). Spatial heterogeneity in nitrite distributions observed during multiple sampling campaigns in the Eastern Tropical North Pacific (ETNP) ODZ could not be explained using oxygen or nutrient concentrations. By deconvoluting the source waters sampled on these cruises, I determined that mesoscale features transporting the 13 ºC Water mass westward led to nitrite accumulation in the ETNP ODZ. With this source water mass framework, I identified the basin-wide stoichiometry of anaerobic respiration in the ETNP ODZ. These results reveal that 50\%-70\% of the nitrite produced is re-oxidized to nitrate in subsurface waters, and I developed a method that can estimate nitrite re-oxidation using only nutrient and carbon measurements. I also applied this source water mass framework to seven cruises spanning a 50-year time series on the 110 ºW line. With these data, I reveal that the ETNP ODZ became 30% stronger in 2019 than 1994. More importantly, I also calculated the first confidence interval for the strength of the ETNP ODZ and concluded that anthropogenic climate change has not yet influenced the strength of the ETNP ODZ, but likely will soon.
Previous research in the Moffett lab has investigated the flux of Fe, specifically Fe(II), as well as iodide from ODZs. In these regions, mesoscale features transport Fe(II) from waters that intersect the continental margin. This phenomenon transports extremely high iodide concentrations and iodide has a significantly higher residence time in the subsurface ocean than Fe(II). These findings suggest that iodide could be used to trace sources of Fe(II) in low oxygen waters. To examine the processes that control the distributions and couplings of these compounds, I studied seasonally hypoxic waters on the Oregon continental margin. The presence of low oxygen water in this region prevents water column denitrification like in the ETNP ODZ, but sediment denitrification still occurs. I explored the factors that control Fe(II) accumulation and transport on the Oregon continental shelf, where Fe(II) concentrations exceed 50 nM. Surprisingly, we observed minimal iodide accumulation in this region, unlike in ODZs. With this comparison and a re-analysis of iodate in the ETNP ODZ, I contrast the nitrogen, Fe, and sulfur cycles in these two reducing margins. These findings suggest that sulfide accumulation likely drives iodate depletion in these low oxygen waters. The findings within this dissertation highlight the physical processes that influence reduced compounds and the interconnections of these elemental cycles.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Anaerobic iron cycling in an oxygen deficient zone
PDF
Comparative behavior and distribution of biologically relevant trace metals - iron, manganese, and copper in four representative oxygen deficient regimes of the world's oceans
PDF
Tracking fluctuations in the eastern tropical north Pacific oxygen minimum zone: a high-resolution geochemical evaluation of laminated sediments along western North America
PDF
Going with the flow: constraining the lateral advection of redox-active metals from continental margins under differing oxygen regimes
PDF
The distributions and geochemistry of iodine and copper in the Pacific Ocean
Asset Metadata
Creator
Evans, Natalya
(author)
Core Title
Discerning local and long-range causes of deoxygenation and their impact on the accumulation of trace, reduced compounds
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Ocean Sciences
Degree Conferral Date
2022-12
Publication Date
11/22/2022
Defense Date
11/21/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
13CW,anoxia,California current system,climate change,denitrification,eastern boundary upwelling systems,Fe(II),ferrous,fixed nitrogen loss,hypoxia,iodate,iodide,Iodine,Iron,nitrite,northern equatorial Pacific intermediate water,OAI-PMH Harvest,Oregon,oxygen deficient zones,oxygen minimum zones: eastern tropical north Pacific,respiration,water mass analysis
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Moffett, James W. (
committee chair
), John, Seth G. (
committee member
), Levine, Naomi M. (
committee member
), Pennell, Matthew W. (
committee member
)
Creator Email
natalyacevans@gmail.com,ncevans@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC112542559
Unique identifier
UC112542559
Identifier
etd-EvansNatal-11331.pdf (filename)
Legacy Identifier
etd-EvansNatal-11331
Document Type
Dissertation
Format
theses (aat)
Rights
Evans, Natalya
Internet Media Type
application/pdf
Type
texts
Source
20221201-usctheses-batch-993
(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
cisadmin@lib.usc.edu
Tags
13CW
anoxia
California current system
climate change
denitrification
eastern boundary upwelling systems
Fe(II)
ferrous
fixed nitrogen loss
hypoxia
iodate
iodide
nitrite
northern equatorial Pacific intermediate water
oxygen deficient zones
oxygen minimum zones: eastern tropical north Pacific
respiration
water mass analysis