Seawater was collected from the Bermuda Atlantic Time Series (BATS) and Hydrostation S sites in the Sargasso Sea in September 2018. Depth profile investigations at BATS were taken at 32.343°N 64.594°W at 21 separate depths between 1 m and 4500 m. Hydrostation S samples were taken at 32.165°N 64.501°W at 10 depths between 1 m and 500 m. Incubation water was taken from two depths (1 m and 240 m) and collected into four carboys (two euphotic (1 m) and two subphotic (240 m)) between 20:30 and 22:30 ADT. One carboy from each depth was filtered using a 0.2 µm filter to remove bacteria and other biology and particles while another was left unfiltered. ¹²⁹I (t_1/2 ≅ 15.7 My) (Eckert and Ziegler Isotope Products ^©) (Hardisty et al., 2020; Hardisty et al., 2021), was added directly to each of the carboys at a targeted concentration of ~70 nM ¹²⁹I^- for investigating iodine redox reactions in natural seawater over time. ¹²⁹I^- was added before aliquoting the carboy water for individual incubations to ensure homogenous ¹²⁹I^- concentrations at t₀ for all incubations. 200 ml from each carboy were fractionated into separated incubation containers. Samples for t₀ were immediately subsampled from spiked incubation containers, with this and subsequent (t₁, t₂, t_f) subsamples being ~50 ml. All subsamples were immediately filtered at 0.2 µm to end interaction with biology after sampling. Subsamples were refrigerated and stored at 4°C until they returned to Michigan State University and were frozen for storage. Campos (1997) showed that seawater samples stored refrigerated (4°C) or frozen (-20°C) did not show signs of degradation in total iodine measurements over a one-to-three-year period – well over the eight-month timeframe in which measurement of these samples began after collection.

Five incubation factors were used to create 20 incubation trials using a ship-deck light-filtering incubator to mimic at-depth light filtration, cooled with a continuous resupply of ambient surface seawater and stored in translucent and amber high-density polyethylene (HDPE) Nalgene bottles for dark incubations: each done in triplicate ( Table 1 ). Factors included: 1) filtering of samples through a 0.2 µm syringe filter, meant as a control to screen filtered seawater of bacteria and macro-organisms and particles, kept in either the light or the dark depending on incubation, (Campos et al., 1996; Farrenkopf et al., 1997; Hardisty et al., 2020); 2) addition of O₂ ^•− dismutase (SOD) to incubations both filtered and unfiltered, but all left in the dark, intended as a control to remove ambient O₂ ^•− in seawater (Li et al., 2012; Diaz et al., 2013; Sutherland et al., 2020); 3) addition of superoxide thermal source (SOTS) or hydrogen peroxide (H₂O₂) to filtered samples kept in the dark in separate experiments, both suspected of being able to aid in oxidation of I^- to IO₃ ^- in seawater; 4) unfiltered water in the dark to determine the role, if any, of photochemical reactions that may cause the reduction of IO₃ ^- to I^- in the presence of organic matter (Spokes and Liss, 1996; Chance et al., 2014); 5) additions of MnCl₂ to iterations of the above in order to consider the potential of preferential Mn²⁺ oxidation relative to I^-. Note that controls 2 and 5 were only relevant if I^- oxidation was detected in the other controls.

Superoxide thermal source was kept frozen (-80°C) until it was added daily by pipette to incubations 11 and 19 ( Table 1 ) as a combination of 1 ml dimethyl sulfoxide (DMSO) + 1 mg SOTS (3028 µM SOTS) (Cayman Chemicals, CAS number 223507-96-8) and diluted to 15 µM SOTS within seawater samples, which provides ≥25 nM O₂ ^•− at the surface water temperatures at BATS and Hydrostation S (Heller and Croot, 2010). This was made fresh daily immediately before adding to samples and added daily to account for natural decay. The O₂ ^•− concentration of the SOTS stock was not analyzed but O₂ ^•− concentration was analyzed in one experiment a few hours post-SOTS addition – to allow to reach steady state concentrations – to confirm O₂ ^•− accumulation. Hydrogen peroxide (30%) was added at a volume targeting 50 nM H₂O₂ in each solution. SOD was added by pipette daily – thus accounting for decay and titration via potentially newly formed O₂ ^•− within the incubations – from a stock concentration of 4 kU/ml to incubations to produce samples with SOD concentration of 0.32 kU/ml. Given potential long oxidation timescales of I^-, all incubations were performed over a 140-hour time period, with subsamples collected for iodine species measurement at t₀, ~t₄₀, ~t₈₈, and ~t₁₄₀ hours.

Depth profiles of I^- and IO₃ ^- (nM), O₂ ^•− (pM), temperature (°C), and dissolved oxygen (μM) at BATS and Hydrostation S are detailed in Figure 2 and show predictable changes throughout depth of IO₃ ^- and I^-. The dark, particle-associated superoxide steady-state concentration at BATS and Hydrostation S span 4-720 pM through the water column between 1 m and 4553 m at BATS and Hydrostation S sampling sites. Superoxide concentrations were highest between the surface and 1000 meters and fall within the range of previously reported water column values (Hansard et al., 2010; Rusak et al., 2011; Roe et al., 2016). Temperature and dissolved oxygen values are available from cruise CTD data at http://batsftp.bios.edu/. Depth profiles of I^- and IO₃ ^- mirror that of previous studies of iodine in an oxygenated seawater, with I^- accumulation in the surface elevated at the expense of IO₃ ^- and with iodine below the euphotic zone being almost completely IO₃ ^- except for the bottom water sample at 4500 m which has relatively elevated I^-. We also show I^- and IO₃ ^- reported from previous studies at BATS which were limited to shallower depths (2500 m) than that studied here (4500 m).

It is notable that there appears to be elevated superoxide concentrations at some intermediate water sample depths (e.g., 500 m). This is somewhat at odds with the expectation that light-independent superoxide production will scale with cell abundance and activity. While we did not measure additional biological parameters that might allow us to interpret these concentrations, it is apparent that elevated superoxide concentration do coincide with significant dissolved oxygen gradients. It is possible that elevated superoxide is either a direct or indirect result of remineralization processes occurring at these depths.

In incubation samples, initial [IO₃ ^-] values measured spectrophotometrically were found to be within the range of 209 nM to 452 nM, while I^- separated from incubation samples via ion-exchange chromatography was found to be in the range of 92 nM to 235 nM from ICP-MS measurements ( Table 2 , Figures 3 – 5 ). Iodate values were spot-checked via exchange chromatography and ICP-MS and were found to be consistent with spectrophotometric measurements but showed a larger standard deviation in most cases, consistent with potential variability in yields ( Table 2 ). It is interesting to note that, in some cases, filtered conditions have a lower measured [IO₃ ^-] than unfiltered conditions ( Figure 3A , subphotic depth).

Measured [I^-] shows no change over time in any of the incubations investigated ( Figures 4 – 6 ). This is also true for [IO₃ ^-] for most incubation trials; however, IO₃ ^- values in incubations that included sequential additions of O₂ ^•− (10 nM/24 hrs) and H₂O₂ (50 nM/48 hrs) showed a decrease in [IO₃ ^-] over time, from 309 nM to 249 nM and 255 nM to 204 nM, respectively. Given an interference causing a baseline shift in these incubations (discussed in Supplementary Information ), these samples were corrected for by selecting the minimum trough value between 300 nm and 350 nm for spectrophotometer concentration calculations (section 2.3.2) instead of the exact 320 nm value, which was impacted and artificially increased by the interference shift. With correction, a decrease in IO₃ ^- concentration was still observed. Since there was no corresponding increase in [I^-] or decrease in ¹²⁹I^-/¹²⁷I^-, which would be anticipated for ¹²⁷IO₃ ^- reduction to I^-, this could indicate reduction of IO₃ ^- to an iodine intermediate not identified in this study. That said, it more likely reflects interferences from SOTS degradation products which have an overlapping absorbance range with I₃ near 320 nm. Specifically, 4-Formyl Benzoic acid is a degradation product of SOTS-1 (Ingold et al., 1997; Konya et al., 2000; Heller and Croot, 2010) and ROS-induced oxidation products of CDOM have overlapping absorption peaks. Notably, the SOTS degradation products do not account for the same observations made for the same trend observed in H₂O₂ observations. Importantly, we acknowledge that the spectrophotometric IO₃ ^- measurement is highly prone to interferences (Truesdale, 1978; Luther et al., 1988). Seawater background measurements at 350 nm have the potential to correct these interferences (Hepach et al., 2020; Jones et al., 2023).

Note that the controls with addition of SOD and MnCl₂ (meant to test the potential to stop I^- oxidation via scrubbing of superoxide or added preferred electron donor) were only relevant if I^- oxidation was observed in other controls, so are not considered further. These data are available in Supplementary Table 1 .

DOI was not a focus of this study but was quantified in some instances for concentration ( Table 2 ) as well as for ¹²⁹I/¹²⁷I ratios in incubations (described in the next section). We note that DOI concentrations in measured incubations (~7.5-9.5% of total dissolved iodine) were larger than we anticipated for open ocean areas where DOI is commonly negligible or uncharacterized (Wong and Cheng, 1998). DOI has previously been found to account for up to 10% of the total iodine pool in coastal areas (Chance et al., 2014).

While all incubations were measured for [IO₃ ^-], we did not measure ¹²⁹I/¹²⁷I ratios for all incubations. This is because of the lack of variation observed in ¹²⁹I/¹²⁷I for the targeted ROS and other incubations. Measured ¹²⁹I/¹²⁷I ratios measured from chosen incubation experiments show no change over time beyond error bars in any of the incubations investigated ( Figures 3 – 5 ). This includes IO₃ ^-, I^-, and DOI. This also includes the ¹²⁹I/¹²⁷I of I^- in the ROS-based incubations where a decrease in spectrophotometrically quantified [IO₃ ^-] was observed.

Initial ¹²⁹I/¹²⁷I ratios were consistent for measured I^- samples, which was expected given that the spike was added to a larger stock volume of seawater that was then aliquoted for the incubations. Average ¹²⁹I/¹²⁷I ratios for I^- at t₀ ranged from 0.29 ± 0.004 to 0.32 ± 0.002 at the surface. Initial isotope ratios for IO₃ ^- range from 0.0016 ± 0.0006 to 0.088 ± 0.0005 and 0.0008 ± 0.0001 to 0.002 ± 0.0002 for 1 m (photic) and 240 m (subphotic) depths, respectively, with incubation 8 (photic, dark, filtered) being a slight outlier at 0.07 ± 0.00024. It is notable that ¹²⁹I was observed in both the DOI and IO₃ ^- fractions, though it was added as I^-. Initial IO₃ ^- and DOI values were still predictably quite low, as Hardisty et al. (2020) demonstrated the spike is mostly ¹²⁹I^- with only minor ¹²⁹IO₃ ^-, with the same being expected for DOI. Initial DOI values ranged between 0.005 ± 0.002 and 0.01 ± 0.003, with incubation 8 again being a slight outlier at 0.05 ± 0.004 ( Figure 4 ).

While oxidation was not observed, we quantified the maximum possible rates that would maintain our time series for ¹²⁹I/¹²⁷I of IO₃ ^- within the error (1 s.d.). Maximum daily gross rates of I^- oxidation determined by incubation conditions were calculated using isotope mass balance equations outlined in Hardisty et al. (2020). Average and standard deviation of triplicate initial and final incubation timepoint spectrophotometer measurements of ¹²⁷[IO₃ ^-], ICP-MS measurements of ¹²⁷[I^-], and MC-ICP-MS measurements of I^{- 129}I/¹²⁷I ratios and IO₃ ^{- 129}I/¹²⁷I ratios were used to first calculate the total IO₃ ^- created from I^- oxidation in incubations between t₀ to t₁₄₀ (t_final). These calculations were then used to determine the rate (nM/day) of I^- oxidation to IO₃ ^-.

The [ i o d a t e ]   127   i (initial [IO₃ ^-] pre-spike addition) and the isotope ratios of I^- and IO₃ ^- at a given time t ( R i o d i d e t and R i o d a t e t , respectively) are measured (Equations 1, 2). Since I^- oxidation to IO₃ ^- is the only quantifiable source of ¹²⁹I to R i o d a t e t and negligibly fractionated, we can assume that changes in R i o d a t e t from ¹²⁹I and ¹²⁷I are from I^- oxidation (i.e., Δ [ i o d a t e ] in Equation 3) are contributed at an isotope ratio equivalent to R i o d i d e t (Equation 1). As such, Equation 1 can be rearranged to solve for Δ   127 [ i o d i d e ] t which can be substituted into Equation 2 to solve for Δ   129 [ i o d i d e ] t . Plugging Δ   129 [ i o d i d e ] t into Equation 1 then allows to solve for Δ   127 [ i o d i d e ] t which then can be used in Equation 3 to solve for Δ [ i o d a t e ] . Based on the measured standard deviations of R i o d i d e t ,   R i o d a t e t ,   [ i o d a t e ]   127   i   values and the hours of the incubation, we can then solve for the rate of I^- oxidation to IO₃ ^- taking place in the incubations in nM/day ( Table 3 ).

This is the first report of O₂ ^•− at BATS, although iodine speciation at BATS and Hydrostation S has been previously reported through investigations of depth profiles of [I^-] and [IO₃ ^-] over time to a depth of up to 2500 m (Jickells et al., 1988; Campos et al., 1996) ( Figure 2A , Figure 6 ). The similarities between measured iodine speciation profiles provide some constraints on the reproducibility of variable iodine speciation techniques used between labs. Our current measured data for depth profiles of IO₃ ^- and I^- at BATS generally agree with that of Campos et al. (1996) and demonstrate little change in iodine speciation with time. We also extended those investigations here to the bottom waters at 4500 m. While not the focus here, we note that our new bottom water data additionally show elevated I^-, consistent with other studies demonstrating likely benthic I^- fluxes in abyssal plains (Francois, 1987a; Francois, 1987b; Kennedy and Elderfield 1987a; Kennedy and Elderfield 1987b; Moriyasu et al., 2023).

Our data from Hydrostation S also generally overlaps with the range observed in Jickells et al. (1988); however, we note that our I^- data have a range that extends below that of these previous data. One explanation is that Jickells et al. (1988) measured I^- via difference between total iodine – determined via a UV-oxidation technique – and the spectrophotometrically measured IO₃ ^-, as was done here. Note that the opposite is true for Campos et al. (1996), who reduced IO₃ ^- to I^- and calculated IO₃ ^- via difference following voltametric analysis (Total I – I^-). For both cases, given that total iodine includes organic iodine, this implies the potential that DOI is contributing to either the I^- or IO₃ ^- pool, depending on the technique. While not thoroughly evaluated in this study, we note that DOI concentrations in measured incubations were up to 10% of the total iodine pool, so this could explain the difference between ours and earlier studies.

Alternatively, the difference is Hydrostation S data ranges between the two studies could reflect real changes in hydrography or primary production. For the seasonal study of Jickells et al. (1988), these previous seasonal IO₃ ^- lows and I^- high values were generally observed in the summer and interpreted to reflect more limited water mass exchange, which allows the combination of local Bermuda inshore inputs and primary productivity to limit IO₃ ^- availability. In the absence of a seasonal iodine context, it is difficult to explain the differences from our September 2018 data, but it is likely that increased water mass exchange – increasing the supply of high IO₃ ^- and low I^- from deeper waters and the Sargasso Sea generally – plays an important role. For example, the similarities in our 2018 BATS and Hydrostation S data suggest the potential for increased hydrographic exchange between these two sites. Further, the September 2018 sampling overlapped with the passage of Hurricane Florence 750 miles southeast of Bermuda on September 9^th, 2018. While the mixed layer depth from September 2018 is like that of previous years, hurricanes have been documented to influence nutrient and other chemical parameters via increased mixing (Babin et al., 2004).

We note that while seasonal iodine variations may be driven by variations in primary productivity and mixed layer depth, other factors may contribute to longer-term evolution of iodine speciation. Specifically, Hughes et al. (2021) demonstrate that I^- oxidation to IO₃ ^- may be linked to nitrification, which is sensitive to a range of factors – e.g., light, O₂, temperature, pH, and photochemically produced ROS (Pajares and Ramos, 2019; Morris et al., 2022). Given the baseline iodine speciation conditions documented here and in previous studies, BATS may be an ideal location to track potential future changes in iodine speciation linked to predictions of evolving nitrification from global warming and ocean acidification (Bemen et al., 2012). To our knowledge, the only other locations with multiple iodine measurements are from the Black Sea (Wong and Brewer, 1977; Luther and Campbell, 1991; Truesdale et al., 2001a) and Baltic Sea (Hou et al., 2001; Truesdale et al., 2001b; Aldahan et al., 2007; Truesdale et al., 2013), but iodine speciation changes there will additionally be sensitive to the well-known presence of hypoxia and euxinia.

Prior to interpretations, it is important to discuss the limitations of our experimental and analytical approaches. First, while isotope ratio variations (or lack thereof) form the basis of our interpretations, we note that variability in triplicate measurements in isotopes and concentrations measured via ICP-MS and spectrophotometry all contribute to the uncertainty of our rate calculations ( Table 3 ). As previously mentioned, we acknowledge that the spectrophotometric method of measuring [IO₃ ^-] is prone to interferences and relatively low precision (Truesdale, 1978; Luther et al., 1988). For example, variation observed in the IO₃ ^- concentration but not in IO₃ ^- isotope ratios for the ROS-amended incubations ( Figure 5 ) is likely attributed to interferences and not iodine species variations (see Supplementary Information ). That said, beyond this example, there was general agreement between IO₃ ^- concentration and iodine isotope trends in this study and between ours and previously measured IO₃ ^- concentrations at Hydrostation S and BATS ( Figures 2 and 6 ).

Another important limitation is the duration of the experiments, which likely induced bottle effects in unfiltered conditions. For example, a subset of our experiments was unfiltered in order to retain particles and native biological communities, but changes in community composition is well known within prolonged bottle experiments, even within 24 hours (Berg et al., 1999). Our experiments lasted ~140 hours (due to the known sluggish iodine oxidation kinetics) and the biological communities of unfiltered controls were not monitored. As there were no nutrient enrichments in our unfiltered incubations, it is likely that native cells underwent physiological changes that altered the balance between phytoplankton and bacteria. While not observed here, such a turnover could have shifted the balance between iodine oxidation and reduction, or commenced cell senescence, which favors I^- release (e.g., Bluhm et al., 2010). This places a limitation on our ability to interpret biologically induced iodine oxidation and reduction in unfiltered incubations; however, biological turnover is not a factor in our filtered controls testing additional oxidation-reduction pathways.

An additional limitation on our experiments is uncertainty regarding the superoxide concentrations and decay rate in the SOTS amended incubations. Specifically, the superoxide concentration and decay rate were not quantified directly in these incubations. That said, SOTS was added at a concentration estimated to produce an excess superoxide (25-55 nM instantaneous superoxide) relative to the natural waters of BATS ( Figure 2 ). The instantaneous superoxide concentration can be estimated from temperature-dependent decay constants – 4.3-9.9x10^-5 s^-1 at 28.2-29.2°C (Heller and Croot, 2010) – and the observed superoxide decay constant from analogous oligotrophic waters (0.01 s^-1; (Roe et al., 2016)). While experiments were conducted in a shipdeck chamber continuously refilled with local surface water, the temperature was not directly monitored; however, we note that the biggest uncertainty in k_SOTS comes from uncertainty in the values at a given temperature and not from the temperature range. Regardless, the conservative estimate of 25 nM superoxide is still in excess of that measured in water column profiles from this study ( Figure 2 ).

Understanding the rates of I^- oxidation to IO₃ ^- is essential for already applied marine iodine cycling models (Lu et al., 2018; Wadley et al., 2020), but direct constraints are limited. In this study, all incubations showed no discernable change in [IO₃ ^-], [I^-], or the ¹²⁹I/¹²⁷I of IO₃ ^- beyond uncertainties over the timeframe of the incubations (~140 hours) ( Figures 3 – 5 ).

While direct observations of IO₃ ^- production from I^- remains an important goal, our use of a radiotracer still allows us to place novel constraints on the maximum rates of I^- oxidation to IO₃ ^- based on the limitations of our analytical uncertainties ( Table 3 ). Specifically, a lack of change seen over time in these incubations in both concentration and ¹²⁹I/¹²⁷I ratios are used to constrain a rate of <2.99 nM/day ± 0.53 nM/day. Previously calculated values of I^- oxidation have been reported from BATS from mass balance studies to be between 0.74 and 18 nM/day ( Table 4 ). The value calculated for our incubation studies fits well in this range ( Table 2 ), supporting previous claims that oxidation of I^- to IO₃ ^- is slow in natural seawaters (Hardisty et al., 2020; Carpenter et al., 2021).

Further, we observe non-negligible ¹²⁹I in the initial post-spike sample for both the IO₃ ^- and DOI incubation pools. Hardisty et al. (2020) made a similar observation and provided some evidence that the iodine-129 spike, while overwhelmingly I^-, contains some oxidized iodine. An alternative explanation is that the IO₃ ^- or iodine intermediates rapidly form upon addition to seawater or other solutions. This observation implies the potential for near instantaneous formation of ¹²⁹IO₃ ^- or DOI upon addition of spike to samples. Since the increase in ¹²⁹I/¹²⁷I was not ongoing, this would require rapid quenching of an existing oxidant in seawater, thus inhibiting further oxidation.

We provide here the most detailed and direct constraints to date of the potential for the ROS O₂ ^•− (added as the chemical source SOTS) and H₂O₂ to oxidize I^- to IO₃ ^- in natural seawater. To the point, in our incubations with added O₂ ^•− and H₂O₂, no oxidation of I^- to IO₃ ^- was observed in BATS seawater, as shown in ¹²⁹I/¹²⁷I ratios of IO₃ ^- ( Figure 5 ). We emphasize that this does not completely rule out the role of ROS in I^- oxidation broadly. Below we consider the implications of these results in coordination with previous studies evaluating the role, if any, of ROS on iodine cycling.

First, it is possible that ROS is preferentially reacting with dissolved organic matter or another component of seawater (e.g., Mn, NOx) (Wuttig et al., 2013; Li et al., 2014; Sutherland et al., 2021). We did not constrain alternative electron acceptors here but given the oligotrophic nature of BATS and known low dissolved Mn (Wu et al., 2014), this would imply that preferential extracellular O₂ ^•− scavenging relative to I^- is common throughout most of the ocean, and intracellular ROS have been shown to be a significant oxygen sink (Sutherland et al., 2021), hence diminishing the likely role of O₂ ^•− in marine I^- oxidation more broadly. Some evidence for preferential reaction of ROS with other redox-active species may come from the interference observed for our spectrophotometric measurements of IO₃ ^-. Second, thermodynamic calculations predict that O₂ ^•− alone forms iodine intermediates and cannot fully oxidize I^- to IO₃ ^-, so it is likely that iodine intermediates such as I₂, HOI, or DOI, instead of IO₃ ^-, in samples with excess H₂O₂ or SOTS added, are forming. This implies a potential role of ROS in iodine redox species cycling but contrasts previous findings. As discussed in detail in Luther (2023), reactions with O₂ ^•− and I^- forms I₂ and subsequently I₂ ^- (Bielski et al., 1985; Schwarz and Bielski, 1986). I₂ reacts quickly with organic matter to form DOI, which may act as a bottleneck in some environments to titrate I₂ and prevent subsequent oxidation (e.g., Hardisty et al., 2020). Since the ¹²⁹I/¹²⁷I ratio of DOI did not change in experiments ( Figure 5E ), it implies relatively little, if any, oxidation to intermediates forming DOI. Beyond this, O₂ ^•− is not favorable to subsequently oxidize I₂/I₂ ^- to form IO₃ ^-. Instead, •OH or O₃ are required for subsequent HOI formation, and OH, O₃, and H₂O₂ can then oxidize to IO₂ ^-, and then subsequently to IO₃ ^-. O₃ is not prevalent beyond the marine micro-layer and thus is unlikely responsible for I^- oxidation elsewhere in seawater.

Ultimately, these thermodynamic considerations imply that combinations of O₂ ^•- and H₂O₂, and perhaps OH radicals produced during their reduction, are necessary for complete oxidation from I^- to IO₃ ^- via ROS. Therefore, ROS most likely play a minor, if any, role in IO₃ ^- formation given the combination of: slow predicted kinetics (e.g., Wong and Zhang, 2008), limited ROS in seawater relative to iodine, the up to four step oxidation sequence each consuming ROS (I^- ➔ I₂ ➔ HOI ➔ IO₂ ^- ➔ IO₃ ^-), abundance of DOM and other preferred electron acceptors/donor – such as Br and Mn – for ROS, titration of HOI with DOM, and the likelihood of back reactions of iodine intermediates to I^-.

The lack of change in [IO₃ ^-], [I^-] or ¹²⁹I/¹²⁷I isotope ratios over the incubation period (~140 hours) ( Figure 4 ) for the incubations exposed to dark vs light conditions additionally provide constraints on the likelihood of iodine redox reaction pathways. First, interactions between light and organic matter represent an abiotic pathway for ROS production (Morris et al., 2022). If ¹²⁹IO₃ ^- formation had been observed in the light experiments, an abiotic ROS mechanism could have been inferred from a lack of ¹²⁹IO₃ ^- in both the dark and light +SOD controls. Second, natural light is also suggested to aid in photo-reduction of IO₃ ^- in seawater, although it is not known how important this reduction pathway may be for I^- accumulation in the surface ocean (Spokes and Liss, 1996; Chance et al., 2014). Conversion of IO₃ ^- to I^- observed in both isotopes and concentrations would provide evidence for such a pathway, but this was not observed in this study.

Our results support mounting evidence that iodine cycling in some ocean regions, if not more broadly, may be considered semi-conservative. From this perspective, given the relatively oligotrophic conditions at BATS (Lipschultz et al., 2002), it is perhaps not surprising that IO₃ ^- production may be slow or isolated to specific depths or even seasons. For example, laboratory cultures have identified nitrification as a possible pathway of IO₃ ^- production, which occurs specifically at the nitrite maximum at BATS and other similar localities (Newell et al., 2013; Hughes et al., 2021). In addition, Luther (2023) provides thermodynamic constraints that OH is a powerful oxidant capable of producing IO₃ ^-. Oxygenated waters supporting dissolved Fe – such as hydrothermal plumes, some low oxygen zones, and the benthic boundary layer – produce elevated •OH via Fenton chemistry, which could oxidize I^- to IO₃ ^- (e.g., Shaw et al., 2021). Luther (2023) also suggests that O₃ and N₂O can form IO₃ ^-, indicating that marine microlayer (Carpenter et al., 2021) and the oxycline of oxygen deficient zones (Babbin et al., 2015) may be important for IO₃ ^- production, respectively. Lastly, IO₃ ^- reduction in surface waters is more clearly linked to phytoplankton, which show seasonal distributions at BATS (Michaels et al., 1994; Michaels and Knap, 1996; DuRand et al., 2001).

Ex situ mixing of source waters from “hotspots” of high primary productivity, oxygen deficient zones, and pore waters may initiate iodine speciation gradients in surface waters whose rates of change slow as water masses extend to the open ocean from coastal or productive settings. As a result, water mass mixing may have a more dramatic effect on the distribution of iodine species in the open surface ocean than previously known. For example, Chance et al. (2020) and Chance et al. (2010) provided some evidence that diffusion/advection of IO₃ ^- from below the mixed layer may be important for controlling surface ocean IO₃ ^- abundance. Even in oxygen deficient zones, Hardisty et al. (2021) provide evidence that below the oxycline that IO₃ ^- has the potential to reflect water mass mixing, as opposed to clear in situ IO₃ ^- reduction.