Differences in Siderophore Production During the Growth of Phytoplankton in Southern Ocean Incubation Experiments Under High‐ and Low‐Iron Conditions Authors: Randelle M. Bundy, Kristen N. Buck, Bethany D. Jenkins, P. Dreux Chappell, Daniel J. Repeta, Alexa Sterling, Shannon M. Yang, Laura Holland, Léo Mahieu, Laura E. Moore Published: 2026 ABSTRACT ABSTRACTIron is an important micronutrient required by marine microorganisms. The Southern Ocean is an ecosystem where primary production is limited by the availability of iron, but how microorganisms in this region adapt to low‐iron conditions or respond to episodic iron inputs is understudied. We tested how external iron additions would impact bacterial production of siderophores during phytoplankton growth using shipboard incubation experiments under a range of iron conditions. Similar heterotrophic bacteria communities and phytoplankton biomass were observed whether iron was added from the continental margin or as inorganic iron. Under both high‐ and low‐iron conditions, the heterotrophic bacteria community remained the same but produced distinct iron‐binding siderophores. Siderophores were detected in the highest concentrations when phytoplankton and heterotrophic bacteria were iron‐limited, and unique compounds were produced in low‐iron conditions compared to when iron was added. Hydrophobic siderophores dominated when iron concentrations were low, and hydrophilic compounds when iron concentrations were higher, and this was likely due to environmental factors rather than changes in the microbial siderophore producers. Our results also highlight that phytoplankton growth likely played an important role in stimulating siderophore production across both low‐ and high‐iron conditions, with important implications for phytoplankton and bacteria interactions. Introduction Marine productivity in the ocean is in part controlled by the availability of the major nutrients, such as nitrate and phosphate, as well as minor nutrients, such as vitamins and trace elements (Moore et al. 2002, 2013; Browning and Moore 2023). Iron (Fe) is a trace nutrient that is required by microorganisms in intracellular concentrations that are greater than 100 times more than its concentration in seawater, requiring organisms to adapt to the low Fe conditions present in the majority of the ocean (Moore et al. 2013). Fe is ultimately introduced into the ocean through an external supply mechanism from the Earth's crust, such as rivers, sediment, dust or hydrothermal vents (Moore and Braucher 2008; Tagliabue and Resing 2016). Fe that is introduced into seawater from these supply pathways is poorly soluble in oxygenated seawater and will precipitate without the presence of organic compounds that can complex Fe and keep it in the dissolved phase (< 0.2 μm) (Liu and Millero 2002). Although marine organic matter is extremely heterogeneous, many compounds have functional groups that can bind Fe and are part of the organic Fe‐binding ligand pool that controls the solubility and bioavailability of Fe in the ocean (Gledhill and Buck 2012; Buck and Bundy 2024; Whitby et al. 2024). The majority of dissolved Fe in seawater is associated with organic matter; thus, organisms must either release Fe from the organic compound before taking it up or internalise the entire Fe‐organic ligand complex to meet their Fe demands (Shaked et al. 2005; Sutak et al. 2012). The chemical identity of organic Fe in seawater therefore plays an important role in its availability (Lis et al. 2015), and may select for organisms with specific Fe uptake pathways depending on the organic compound that is bound to Fe (Hutchins et al. 1999). An important class of organic ligands that complex Fe in seawater is siderophores. Siderophores are organic compounds that have an extremely high affinity for Fe and are specifically produced by bacteria or fungi for the purposes of acquiring Fe (Butler 2005; Vraspir and Butler 2009). Siderophores are well characterised in terrestrial ecosystems and enteric systems and are used primarily in these systems to solubilise otherwise inaccessible Fe phases from particles, or to outcompete other organisms for this limited resource (Vraspir and Butler 2009). Siderophores have only recently been studied in the marine environment, as it was originally assumed they would not be an advantageous Fe acquisition strategy in the dilute oceanic environment, and because other organisms can ‘steal’ siderophores once produced (i.e., the ‘public goods dilemma’), siderophore production might not be favoured in the ocean (Griffin et al. 2004; Leventhal et al. 2019). However, measurements of both bulk organic ligand binding strength from voltammetry studies (Witter et al. 2000; Gledhill and Buck 2012) and mass spectrometry measurements that are able to identify specific organic compounds provide evidence that siderophores are present in a wide variety of ocean environments (Macrellis et al. 2001; Whitby et al. 2024). Although originally characterised in laboratory cultures of marine organisms (Vraspir and Butler 2009), siderophores have now been detected in environments ranging from coastal (Boiteau et al. 2019; Moore et al. 2021) and hydrothermal systems (Hoffman et al. 2024) to the open ocean (Boiteau et al. 2016; Bundy et al. 2018; Park et al. 2023; Li et al. 2024). Their presence in such diverse marine ecosystems was initially surprising, given the energetic cost of producing siderophores and the relatively high Fe concentrations of some of these ecosystems. Regardless, the ability to produce and consume siderophores appears widespread in the ocean and, thus far, is not necessarily influenced only by dissolved Fe distributions (Hopkinson and Barbeau 2012; Garber et al. 2020; Park et al. 2023). These observations are consistent with the ocean as an overall Fe‐poor environment and/or that the chemical form of Fe in seawater is not readily accessible to the microbial community. This leads to several remaining questions about when and why siderophores are produced, and how they might be influencing the bioavailability of Fe for phytoplankton and bacteria in marine ecosystems. Siderophores have now generally been found in a wide range of ocean environments, yet the variety of factors that may stimulate their production remains unknown. For example, no phytoplankton have been definitively shown to possess the biosynthesis genes to produce siderophores, but some are able to internalise siderophores through various uptake mechanisms (Maldonado et al. 2005; Sutak et al. 2012; Kazamia et al. 2018; Coale et al. 2019), though this might be a specialised adaptation for some taxa. Many bacteria, on the other hand, can both produce and consume siderophore‐bound Fe (Hopkinson and Barbeau 2012), with some ‘cheaters’ who only take up siderophores and do not produce them (Cordero et al. 2012). Thus, siderophore production that is observed in the marine environment is assumed to be due to production from heterotrophic bacteria. However, the conditions under which siderophore production in the ocean is favoured are poorly understood. It is also unclear whether siderophore production makes Fe more or less bioavailable to certain members of the phytoplankton community, as it likely depends on whether or not they have Fe‐siderophore uptake pathways. This study explores the dynamics influencing siderophore production in the Southern Ocean, the most severely Fe‐limited oceanic ecosystem (Martin et al. 1991). The Southern Ocean is the largest high‐nutrient, low‐chlorophyll region in the world, where the availability of Fe is the primary limiting nutrient to phytoplankton growth and primary production. In controlled incubation experiments conducted near the Antarctic Peninsula, we explored how the in situ phytoplankton and bacteria communities responded to a range of external Fe additions, both in the form of added inorganic Fe and naturally added Fe from the continental margin. The bulk phytoplankton response and nutrient and trace metal uptake rates from these experiments have been previously published (Burns et al. 2023). This work focuses on the experimental conditions that stimulated siderophore production, giving insights into where and when we observe siderophore production in seawater, with implications for the competition and bioavailability of Fe in Fe‐limited ecosystems. Methods Incubation Experimental Design Here, we present data from two incubation experiments, the details of which have been described elsewhere (Burns et al. 2023). The experiments presented in this work are Incubation Experiment #1 and Incubation Experiment #3; we have retained these names to be consistent with previous work (Burns et al. 2023). In brief, the two incubation experiments were initiated near the Antarctic Peninsula in the Southern Ocean (Figure 1A) with the overarching goal of testing whether externally added Fe as inorganic Fe (FeCl3) had similar bioavailability to the in situ phytoplankton community as naturally added Fe from shelf sediment sources, related in part to the presence of siderophores. The experiments were conducted on board the R/V/I/B Nathaniel B. Palmer during austral spring from 7 September to 14 October 2016. Seawater was collected using a trace metal‐clean CTD rosette equipped with an SBE32 epoxy‐coated pylon and Kevlar conducting cable and 12 Teflon‐coated external spring Niskin (X‐Niskin, Ocean Test Equipment) bottles (Cutter and Bruland 2012). Incubation #1 was initiated using trace metal clean seawater collected offshore (−62.33°N, −64.65°E) in low Fe waters (Figure 1A). The unfiltered seawater from the Niskin bottles was passed through a 150 μm acid‐cleaned nylon mesh to remove large grazers and dispensed into 50 L acid‐cleaned polypropylene carboys to homogenise the collected water, before finally being dispensed into acid‐cleaned 4 L polycarbonate bottles (Nalgene) for each treatment (Figure 1B). The experiment consisted of a series of treatments that were incubated in the light, including a control (no amendments) and three added Fe treatments (+ 1, 4 and 10 nM Fe) to which the Fe was added as 57FeCl3 (Burns et al. 2023). The same treatments were also included as replicates that were incubated in the dark, and these dark treatments were prepared in the same way as the light bottles. Dark treatments were incubated alongside light treatments to measure the impact of Fe conditions on only the bacteria community, without significant phytoplankton growth. Data from the dark treatments are shown in Supporting Information. Incubation #3 was similar, in that it was initiated in offshore waters from the same station as Incubation #1. However, in this incubation, the Fe was added as a natural Fe source from shelf waters (Figure 1A). The shelf waters were collected near the Bransfield Strait (−62.46°N, −59.57°E) and were filtered through a 0.2 μm filter (Acropak 200, Pall Corporation) and stored at 4°C in a 50 L carboy until use. Incubation #3 consisted of a low Fe (offshore waters only) and a high Fe (50% offshore water and 50% shelf water) treatment (Figure 1B). Similar to Incubation #1, dark treatments were also prepared for this experiment in the same way as the treatments that were incubated in the light (Figure 1B). All incubation bottle caps were sealed with parafilm, and bottles were gently inverted to mix the contents. Light treatments were incubated in a temperature‐ and light‐controlled incubation van (temperature = 3.5°C ± 1.5°C, n = 3405; light = 108.1 ± 37.5 μmol photons m−2 s−1, n = 42), and dark treatments were double‐bagged in heavy‐duty black polypropylene bags and incubated in the same temperature‐controlled incubation van. Each bottle was assigned a unique number along with its treatment designation to track it throughout the experiment, and replicates were sampled randomly at each time point (Burns et al. 2023). Study region and experimental design. (A) Map of the study region denoting the location of Incubation Experiment #1 (green star) and the location of filtered seawater that was collected for Incubation Experiment #3 (red star). (B) Experimental design of Incubation #1, where low‐iron water was collected offshore (green star; control) and was amended with 1, 4 and 10 nM iron added as inorganic iron (FeCl3). Incubation #3 contained a low‐iron treatment of water that was also collected offshore (green star, low iron) and then was amended with filtered high‐iron waters from the shelf (red star, high iron). Details on the experimental design can be found in Burns et al. (2023). Nutrient, Chlorophyll a and Trace Element Analyses Dissolved nutrient samples were filtered through sequential 3.0 and 0.4 μm polycarbonate track‐etched filters (PCTE, Whatman Nucleopore) under vacuum on Teflon filter holders (Savillex Corporation). Macronutrient samples for dissolved nitrate, soluble reactive phosphorus (phosphate) and silicic acid (silicate) were collected in 50 mL polypropylene tubes (Falcon) and stored at 4°C until analysed shipboard. Details of the shipboard nutrient analyses can be found in Burns et al. (2023). Briefly, nutrient samples were analysed within 24 h of collection on a Lachat 8500 QuickChem analyser and also in the laboratory on a Seal Analytical Technicon AAII (Parsons et al. 1984). Chlorophyll a samples were collected by filtering through 0.8 μm glass fibre filters (GF/F, Whatman) under gentle vacuum filtration. Chlorophyll a was extracted from the filters with 95% ethanol after collection and stored in the dark at room temperature for 12–24 h. Chlorophyll a concentrations were measured using a 10AU fluorometer (Turner Designs, Incorporated) before and after acidification with 10% hydrochloric acid (Jespersen 1987; Morison and Menden‐Deuer 2015). Dissolved Fe samples were filtered as described above for the macronutrient samples. Samples were collected in 125 mL low‐density polyethylene (LDPE, Nalgene) bottles, acidified to pH ~ 1.8 with 0.024 mol L−1 HCl (Optima, Fisher Scientific) and were stored at room temperature for several months prior to analysis (Burns et al. 2023). For analysis, samples were preconcentrated using a seaFAST‐pico system (Elemental Scientific) and eluted with 5% triple‐distilled nitric acid containing a 10 ppb indium internal standard into 2.0 mL polyvinylidene fluoride vials (Elemental Scientific). Dissolved Fe was quantified by standard addition using a calibration curve both from a mixed metal standard (Ultra Scientific) and from a standard curve of the 57FeCl3 stock that was used to spike the incubation experiments. Samples, standards and blanks were analysed on a magnetic sector Element XR high‐resolution inductively coupled plasma mass spectrometer (HR‐ICP‐MS; Thermo Scientific). For samples that had been spiked with 57FeCl3, the concentration of 57Fe was calculated by first subtracting the natural abundance 57Fe determined from the mixed metal standard addition curve, and then quantifying the 57Fe using the 57FeCl3 standard curve. Additional details on quality control and the measurements of additional trace metals are described elsewhere (Burns et al. 2023). Prokaryotic Community Composition The prokaryotic community composition was determined by sequential filtration of approximately 3.2 L of water onto 3.0 μm (PETE, Sterllitch) and 0.2 μm 25 mm filters (Supor, Pall). Prokaryotic sequences found on the 3.0 μm filters were considered particle‐associated bacteria, while those in the 3.0–0.2 μm size fraction were considered free‐living bacteria. Universal primers were used to recover DNA barcodes of the V4 region of the 16S rRNA gene for bacteria from these samples, including the initial time point for both experiments and day 12 for Incubation #1 and day 9 for Incubation #3 from all treatments in the light from both experiments. The primers included the 515FB forward and 806RB reverse primers from the Earth Microbiome Project. Sequencing was performed in duplicate (2 × 300 bp) on Illumina MiSeq, and amplicon sequence variants (ASVs) were created using DADA2 and taxonomy was assigned using the SILVA database (v138) with DADA2's naïve Bayes classifier. For an ASV to be considered as belonging to a species, it must have been a 100% match. After assigning taxonomy to the bacterial 16S V4 reads, only reads belonging to Kingdom Bacteria were retained and any reads identified as chloroplast or mitochondria were removed (Sterling et al. 2023). Organic Iron‐Binding Ligand Measurements Total dissolved organic Fe‐binding ligand measurements are presented here from Incubation #1 in the control treatment (light and dark) and from Incubation #3 in the low Fe and high Fe treatments from the initial and final time points. Total dissolved organic Fe‐binding ligand samples were filtered as described above into 500 mL fluorinated polyethylene bottles (FPE, Nalgene) that had been acid‐cleaned and conditioned according to GEOTRACES protocols (Cutter et al. 2017). Samples were immediately frozen upright at −20°C until analysis in the laboratory. Before analysis, samples were thawed in the dark at room temperature and swirled to homogenise. The details of the organic Fe‐binding ligand analyses are described elsewhere (Mahieu et al. 2024). In brief, 10 mL aliquots of each sample were placed in 50 mL polypropylene tubes (MetalFree, Labcon) that had been conditioned following Mahieu et al. (2024). Next, 50 μL of a 1.5 M boric acid buffer in 0.4 M ammonia was added to each aliquot for a final pH of 8.2, followed by Fe additions in the form of FeCl3. The Fe additions were 0, 0.1, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 4.0, 5.0, 7.5, 10 and 15 nM, followed by a final 20 nM spike that was performed directly in the voltammetric cell at the end of each titration to verify the saturation of the organic Fe‐binding ligands. The buffer and Fe additions equilibrated with the natural organic ligands for at least 2 h at room temperature, and then 25 μM of salicylaldoxime was added to each aliquot at least 10 min before analysis. Aliquots were analysed on a controlled‐growth mercury electrode equipped with an Epsilon2 analyser (BASi Research Products). The procedure for peak height determinations and interpretation of the data is described in detail elsewhere (Omanović et al. 2015; Mahieu et al. 2024). Siderophore Analyses and Identification Dissolved siderophore samples were collected at a subset of the sampling time points for each incubation experiment. For Incubation #1, dissolved siderophores were sampled on days 0, 1, 9, 12 and 14, and for Incubation #3 they were collected on days 0, 5, 7, 9 and 12. Due to volume constraints, siderophore samples were only taken from one replicate at each time point except the final time point, where three to four replicates were sampled. Siderophore samples from each time point were taken as filtrate from the DNA filtration (described above) and were collected in 1 L acid‐cleaned polycarbonate bottles (Nalgene). Two to three of the replicates at the final time points were also filtered in this manner, but the fourth replicate was from a separate 4 L polycarbonate bottle that had been incubated in the same way as all other bottles from the experiments, but had not been subsampled at the earlier time points and was completely reserved for siderophore samples on the final day of the incubation. This enabled > 1 L of filtrate to be collected in these replicates (replicate C or D for the final time point). The C or D replicates that were filtered separately from the DNA samples on the final time point were filtered through a 0.4 μm PCTE filter as described above for dissolved trace metals. The details of the siderophore extraction method from seawater and subsequent analysis have been described previously (Boiteau et al. 2013; Bundy et al. 2018, 2024), but a brief summary is provided here. Dissolved siderophore filtrate was slowly pumped onto methanol‐activated and Milli‐Q‐rinsed 6 mL Bond Elut ENV columns (Agilent Technologies) to extract organic matter and siderophores from seawater. The solid phase extraction columns were then rinsed thoroughly with Milli‐Q and stored at −20°C until analysis. Prior to analysis, columns were thawed and rinsed three times again with Milli‐Q to remove any remaining salt. Each column was eluted with 12 mL of distilled methanol into 15 mL acid‐cleaned and distilled‐methanol‐rinsed polypropylene centrifuge tubes (Fisher Scientific). The eluted samples were concentrated to < 500 μL in a SpeedVac (Thermo Scientific), and samples were spiked with an internal standard of 1 μM cyanocobalamin before being analysed. Samples were analysed by first being injected onto a PEEK‐lined C18 column, and the compounds were separated using a Dionex 3000 liquid chromatography (LC) system (Thermo Scientific). The chromatography and instrument settings are described in detail elsewhere (Park et al. 2023; Bundy et al. 2024). Briefly, the LC was first coupled to an ICP‐MS (iCap‐Q, Thermo Scientific) and the counts per minute of 56Fe, 57Fe, 54Fe and 59Co were monitored in kinetic energy discrimination mode (KED). Distinct peaks in Fe that were chromatographically resolved from the background were determined as putative siderophores. The area of each peak was converted to a concentration of Fe in the putative siderophore using a standard curve of 56Fe in the siderophore ferrioxamine E. Since all known siderophores in nature contain one Fe atom, the concentration of Fe in the siderophore was equivalent to the concentration of the putative siderophore. Some putative siderophore peaks contained a combination of natural Fe (predominantly 56Fe) as well as added 57Fe, the Fe isotope that was added in our +Fe treatments in the incubation experiments. In these cases, the 56Fe and 57Fe peaks were both integrated and the areas determined, and the 57Fe peak area was corrected for the natural abundance of 57Fe present in the ferrioxamine E standard. The total siderophore concentrations presented at each time point are a sum of the siderophores that contained both 56Fe and 57Fe after the natural abundance of Fe had been corrected. To determine the identity of the putative siderophores, the samples were analysed a second time with the same chromatography and LC settings but coupled to an Orbitrap Fusion (Thermo Scientific) using electrospray ionisation mass spectrometry (ESI‐MS). The detailed instrument settings are described in Bundy et al. (2018). The ICP‐MS and ESI‐MS data were aligned using the cyanocobalamin internal standard, and each sample was searched for the presence of a targeted list of > 500 siderophores using an in‐house R script. Identified siderophores with peak shapes and retention times matching a peak in 56Fe or 57Fe from the ICP‐MS were considered for further analysis (Monreal et al. 2025). A particular siderophore identity was considered further if there was a match in m/z to the list of siderophores, if the Fe‐ and metal‐free (apo) version of the siderophore were present, and if a 13C isotopologue was also present of either the metal or apo‐version of the compound. Fragmentation data (MS2) of some of the compounds were also collected along with the MS1 data, and this was compared to metabolite libraries in cases where publicly available fragmentation data was available. Any siderophores that were only found in one or two samples within a given incubation experiment were discarded. An example output from the siderophore identification analysis is presented in Supporting Information. All putative siderophores presented here are also denoted with a confidence flag in Supporting Information, as very few siderophores are commercially available and thus can be definitively identified with high confidence. Results Microbial Community Response in the Incubation Experiments A response to external Fe addition by the phytoplankton community was observed in both Incubation #1 and #3 (Figures 2 and 3). A detailed description of the macronutrient (nitrate, phosphate and silicate), micronutrient (trace metals) and phytoplankton biomass response (chlorophyll a) has been published elsewhere (Burns et al. 2023). Some of the results from the previous work, however, have been reproduced here as essential context for the organic ligand, siderophore and prokaryotic community composition data presented. In Incubation #1, where external Fe was added to low Fe (< 0.4 nM) offshore waters in the form of inorganic Fe (see Section 3.1), the largest phytoplankton biomass response was observed in the two highest Fe treatments (+4 and 10 nM Fe; Figure 2B). The dissolved Fe, nitrate and other macronutrients (Burns et al. 2023) were drawn down in the added Fe treatments, with the greatest macro‐ and micronutrient uptake also observed in the highest Fe treatments (Figure 2A,C). All nitrate was also consumed in the highest Fe treatments by the final time point. The nitrate: dissolved Fe (μmol:nmol) was also examined to assess nitrate availability relative to Fe. The nitrate:dissolved Fe over time in the experiment evolved from > 75 in the initial conditions to < 12 in the added Fe treatments (Figure 2D), while remaining elevated (> 45) in the controls. Together, these results indicated that the initial unamended waters for Incubation #1 were Fe‐limiting for the in situ phytoplankton community, and that inorganic Fe addition caused a growth response (Burns et al. 2023). Nutrient and trace metal data from Incubation Experiment #1. The average dissolved iron (A), chlorophyll a (B), nitrate (C) and calculated nitrate:dissolved iron (D) concentrations measured over the course of the 14‐day experiment in Incubation Experiment #1 from the initial conditions (dark blue), control (light blue) and added iron treatments (+1, 4 and 10 nM Fe, light orange to dark red). The horizontal lines in (D) represent published nitrate‐to‐dissolved iron ratios that appear to be limiting to diatoms (solid line; 12 μmol:nmol) and to heterotrophic bacteria (dashed line; 20 μmol:nmol). Error bars represent the standard deviation from triplicate sampling bottles. These data are replotted from Burns et al. (2023). Nutrient and trace metal data from Incubation Experiment #3. The dissolved iron (A), chlorophyll a (B), nitrate (C) and calculated nitrate:dissolved iron (D) concentrations measured over the course of the 12‐day experiment in Incubation #3 in the low‐iron (dark blue) and high‐iron (orange) initial conditions, and low‐iron (light blue) and high‐iron (red) treatments. The horizontal lines in (D) represent published nitrate‐to‐dissolved iron ratios that appear to be limiting to diatoms (solid line; 12 μmol:nmol) and to heterotrophic bacteria (dashed line; 20 μmol:nmol). These data are replotted from Burns et al. (2023). The bacterial community response was similar for Incubation #3 when a natural sedimentary external Fe source was added to low Fe offshore waters (Figure 3). The filtered shelf waters that were added to the low Fe offshore water initially had ~2 nM dissolved Fe, between the +1 and +4 nM inorganic Fe treatments in Incubation #1. The resulting phytoplankton biomass (via chlorophyll a proxy) was remarkably similar in magnitude to Incubation #1 of the +1 and +4 nM Fe treatments on day 12 (comparing Figure 3B to Figure 2B). Dissolved Fe and nitrate were drawn down over the first 7 days of the experiment (Figure 3A,C), and then nitrate continued to be taken up for the remaining 5 days of the experiment to near zero (Figure 3C). The dissolved Fe had some replicates which slightly increased for the last two sampling points (Figure 3A). The nitrate:dissolved Fe ratios (Figure 3D) were initially low in the high Fe treatments and became slightly elevated until the termination of the experiment. The nitrate:dissolved Fe ratios in the low Fe treatments were elevated initially (> 75), similar to Incubation #1, and then remained high until approximately day 7, when the dissolved Fe concentrations caused a decrease in the ratios until the final time point on day 12 (Figure 3D). The heterotrophic bacteria community composition shifted significantly from initial conditions to the last sampling time point in both incubation experiments. The initial bacteria community in both Incubations #1 and #3 was diverse, and then it altered to one dominated by Alteromonadales, Flavobacteriales and Oceanospirillales by the later sampling time point in both experiments (Figure 4A,B, Figure S1). Although the initial bacteria community composition had significantly changed later in the experiments (Figure S2A), the community composition was not statistically distinct at the last sampling time point between treatments (Figure S2B, Incubation #3 not shown). The community composition between the free‐living (0.2–3.0 μm) and particle‐associated bacteria (> 3.0 μm) was distinct at all time points (Figure 4, Figure S2). Overall, in both the free‐living and particle‐associated fractions, the bacteria community was very similar to each other within size classes between treatments at the last sampling time point in both experiments (Figure 4). Bacteria community composition from Incubations #1 and #3. Bacteria community composition at the Order level from the particle‐associated bacteria fraction (3.0 μm filter) in Incubation #1 (A) and in Incubation #3 (B) from the initial time point and Day 12 for Incubation #1 and Day 9 for Incubation #3 and final time points. Each bar represents one replicate of each treatment. Organic Ligand Production in High‐ and Low‐Iron Conditions Dissolved siderophores were detected in all treatments and sampling time points, and were produced in both high and low Fe treatments (Figure 5). Siderophore concentrations were relatively consistent across treatments for the first 9 days in Incubation #1 (Figure 5A). Some separation between treatments was observed on days 12–14. The highest concentration of siderophores from Incubation #1 was observed in the control treatments on the final day of the experiment, with greater than 150 pM observed in some treatment bottles (Figure 5A). Siderophore concentrations on the final day in the remaining added Fe treatments were ~10–15 pM. A slight increase in siderophore concentrations was also observed in the dark control treatments, but no difference in siderophores was seen in the dark added Fe treatments (Figure S3A). The siderophore production in the dark was much lower (12 pM) in the control compared to the average increase observed at the final time point in the light control treatments (> 100 pM). Siderophore concentrations in each experiment. Total dissolved siderophore concentrations in Incubation #1 (A) and Incubation #3 (B) over the course of each experiment. Note the different y‐axis scale on panel B. Siderophore concentrations plotted against dissolved iron (C) and nitrate:dissolved iron (μmol:nmol) (D) concentrations from Incubation #1 (blue) and Incubation #3 (red). The vertical solid and dashed lines in (D) are the same as plotted from panel D in Figures 2 and 3. Error bars represent the standard deviation of replicate sample bottles. If no error bar is present, then only a single replicate was measured for that particular time point. Siderophores were also observed at all time points of Incubation #3 and were produced both in the high and low Fe treatments, though at concentrations an order of magnitude lower than in Incubation #1 (Figure 5B). The initial siderophore concentrations in Incubation #3 in the offshore water (low Fe treatment) were similar to those from Incubation #1 (Figure 5A,B). Higher siderophore concentrations were observed in the low Fe treatment compared to the high Fe treatment that contained Fe added from the shelf, but siderophore concentrations in both treatments were higher than the initial concentrations by the final days of the incubation. At the final time point (day 12), siderophore concentrations were elevated in the low Fe relative to the high Fe treatment, but not significantly so, due to high variability between replicates. There was a modest increase in siderophore concentrations at the final time point in the low Fe treatments in the dark (~9 pM), and no significant increase in the high Fe treatment in the dark (Figure S3B). Although siderophore production was observed in all treatments from both experiments, in general, higher concentrations of siderophores were observed in treatments where dissolved Fe concentrations were low (Figure 5C). The highest siderophore concentrations were measured when dissolved Fe concentrations were < 1 nM, but low Fe concentrations did not guarantee that high siderophore concentrations were observed, and there was significant variability between samples. Higher siderophore concentrations were generally measured in samples with elevated nitrate:dissolved Fe (Figure 5D). High nitrate:dissolved Fe conditions were observed throughout both incubation experiments unless significant Fe was added, but it was not until there was phytoplankton biomass accumulation (Figures 2B and 3B) that high siderophore concentrations were measured (Figure 5A,B). For example, the dissolved Fe concentrations were low, and the availability of nitrate relative to dissolved Fe was high in the dark controls in Incubation #1, but only modest siderophore production occurred (Figure S3A). Overall, there was not a linear relationship between siderophore concentrations and either dissolved Fe or nitrate:dissolved Fe. Rather, the highest concentrations of siderophores were measured in samples where there was a combination of low Fe, high nitrate:dissolved Fe and growth of phytoplankton. Subtle differences were observed in the total organic Fe‐binding ligand pool (as measured by voltammetry) between the initial and final conditions of both experiments (Figure 6A–F). The total and excess (excess:total ligand–Fe concentration) ligand concentrations modestly increased in Incubation #1 in the control treatment from the initial to final conditions, and the average binding strength of the ligands remained consistent over time (Figure 6E). Total organic ligand concentrations and binding strengths in the dark treatments also remained relatively constant over time (Figure S3C,E). In Incubation #3, the initial and final ligand pool was similar in both the low and high Fe treatments (Figure 6B), with lower excess ligands observed in the high Fe treatments (Figure 6D). At the final time point of Incubation #3, the total ligand binding strengths in the high Fe treatments were stronger than in the low Fe treatments, but the average binding strength was statistically indistinguishable (Students t‐test, p > 0.05). Total organic ligand concentrations in the dark treatments decreased slightly by the final time point in the low Fe treatment, but remained constant in the high Fe treatment (Figure S3D). The ligands measured on day 12 in the high Fe treatment had conditional binding strengths indicating the majority of the ligands were strong (log KFe′,FeLcond>12.0), while the average ligand pool in both Incubation #1 and the initial conditions of Incubation #3 were more moderate in terms of binding strength logKFe′,FeLcond~11.3–11.7. The binding strength of the ligands in the dark treatments in Incubation #3 did not change significantly over time in either treatment (Figure S3F). Total dissolved organic iron‐binding ligand characteristics in each experiment. Average total dissolved organic iron‐binding ligand characteristics in Incubation #1 (left panels, blue) and Incubation #3 (right panels, red). Average total organic iron‐binding ligand concentrations for a single ligand class in control treatments in Incubation #1 (A) and in the low‐iron and high‐iron treatments for Incubation #3 (B). Average total ligand concentrations that are in excess of the dissolved iron concentrations (excess ligands) in Incubation #1 (C) and Incubation #3 (D). The conditional binding strength of the determined ligand class in Incubation #1 (E) and Incubation #3 (F). Error bars represent the standard deviation of replicate sample measurements. Identification of Siderophores Produced in Low‐ and High‐Iron Conditions Distinct siderophores were found in the high versus low Fe treatments across both experiments (Figure 7). Qualitatively, the differences in the types of siderophores that were detected can be seen in example 56Fe chromatograms from each incubation at the final time points of each experiment, which contained the most phytoplankton biomass and the largest differences in dissolved Fe concentrations (Figure 7A). For example, in the control treatments from Incubation #1, the high concentrations of siderophores observed in the final time point were due to the production of a suite of hydrophobic and amphiphilic siderophores that eluted at later retention times in the chromatogram (Figure 7A, Figures S4 and S5). In the example chromatogram from the final time point in the high Fe treatment in Incubation #3, smaller peaks that appeared at earlier retention times in the chromatogram are present (Figure 7A). These peaks represent siderophores that are hydrophilic. Hydrophilic siderophores were observed commonly across treatments, but tended to dominate the siderophore pool when dissolved Fe concentrations were elevated (> 1 nM). A higher median proportion of the siderophore pool was hydrophobic in all samples with low dissolved Fe concentrations (< 1 nM, n = 25) compared to high dissolved Fe concentrations (> 1 nM, n = 13) (Figure 7B), though high variability was also observed. For example, two samples with high Fe concentrations had > 90% of the total number of putative siderophores as hydrophobic, while all 13 other high Fe concentration samples had < 8% of the siderophores as hydrophobic. In contrast, in the low Fe samples, more than half of the samples (n = 14) contained greater than 8% of the siderophore pool as hydrophobic siderophores. Siderophore identification in high‐ versus low‐iron conditions. (A) Example chromatograms of 56Fe counts per minute from LC‐ICPMS analyses of siderophores, where peaks in the chromatograms represent a putative siderophore. Peaks that emerge early in the chromatogram represent siderophores that are hydrophilic, while later peaks represent hydrophobic and/or amphiphilic compounds. The chromatogram from Incubation #1 is from one replicate of the control treatment on Day 14, while the chromatogram from Incubation #3 is one replicate from the high‐iron treatment on Day 12. (B) Violin plot of the percentage of the total siderophore pool that was comprised of hydrophobic siderophores in low‐ (< 1 nM; blue) versus high‐iron (> 1 nM; red) conditions across all experiments and treatments. The horizontal bar represents the median percentage of hydrophobic siderophores that were detected, and the shape of the violin plot represents the distribution and variation of data points in each category. Seventy unique siderophores were putatively identified in the combination of all experiments and treatments (Table S1). The confidence in these putative siderophores is noted in the Supporting Information, and all identifications should be considered as putative since very high‐confidence siderophore identification is currently difficult given the availability of standards (Park et al. 2023). Four of the identified siderophores were only found in Incubation #3, while all others were found in both experiments. The types of siderophores detected fell into 11 general classes, with only a few additional compounds detected that were outside of these classes. The major classes of compounds included amphibactins, ferrioxamines, aminochelins, asperchromes, pyoverdins, mycobactins, nocardimicins, monoachelins, synechobactins, tenacibactins and terpenibactins (Figure 8). The additional compounds detected that fell outside of these classes were 2‐deoxymugineic acid, hydroxycoprogen, anachelin 1, IC202 and hydroxl‐ferribactin D. Fewer types of siderophores were observed in Incubation #3 compared to Incubation #1 (Figure 8), but there was significant overlap in the types of compounds found across both experiments (Table S1). While some of the siderophores found here were originally isolated from fungi, we assume the majority of the compounds we identified were produced by heterotrophic bacteria, due to their numerical abundance and genetic potential to produce the majority of the compounds we identified. However, some fungal production of siderophores could be possible, though it has yet to be explored in the marine environment. Siderophore class clusters from each experiment. Hierarchical clusters of the presence (red) or absence (blue) of siderophores (columns) or classes of siderophores in Incubation #1 (A) and Incubation #3 (B). The rows in each (treatments and time points) are ordered by dissolved iron concentrations measured in the sample and are represented with a colour bar. Major clusters are labelled with numbers and are referred to in the text. Letters next to a treatment name indicate a sample replicate for that time point. Although many of the siderophores observed fell into broad classes, distinct compounds were often found in different bottle replicates. For example, ferrioxamines were the most common siderophores found in both experiments (Figure 8). Fourteen of the 70 compounds were distinct ferrioxamines (Table S1), but different ferrioxamines often appeared in different treatment bottles. Ferrioxamine X3 was only observed in the +4 and +10 nM treatment bottles in Incubation #1, while ferrioxamine A1 was observed in 9 out of 26 bottles from Incubation #1 (Table S1). In contrast, in Incubation #1, a few siderophores were found commonly across all treatments and bottle replicates (Table S1). These included mycobactin T, ferrioxamine X4 and acyl‐ferrioxamine 3. In Incubation #3, the siderophores most common across all treatments were hydroxycoprogen and pyoverdine 51W (Table S1). Although not the most common siderophores in Incubation #1, these two compounds were also found often in that experiment (Table S1). Amphiphilic siderophores were also commonly detected in both experiments, and these included amphibactins, mycobactins and moanachelins (Figure 8). Other amphiphilic siderophores were also detected, such as synechobactins, terpenibactins and tenacibactins, though not as often (Figure 8). When the treatments and time points from both experiments were ordered by measured dissolved Fe concentration (Figure 8; y‐axis order) and the presence or absence of different classes of siderophores were clustered, patterns emerged in the types of siderophores present in different treatments and time points. Cluster 1 from both experiments represented siderophores that were present across multiple treatments and time points. These included a variety of ferrioxamines (Figure 8). Cluster 2 in both experiments represented amphiphilic siderophores that were present both in the lowest Fe samples and in some other samples with mid‐range Fe concentrations, though they were not universally found in consistent treatments or time points. Cluster 3 from Incubation #1 represented two classes of siderophores that were present in many of the treatments and across the range of Fe concentrations, with slightly higher rates of occurrence in the lower Fe samples (Figure 8A). This was similar for Clusters 3 and 3a in Incubation #3, except for one replicate from the final time point of the high Fe treatment (Figure 8B). Cluster 4 from Incubation #1 consisted of the amphibactins and mycobactins, and this cluster showed the clearest pattern, where the presence of these compounds was almost exclusively in the lowest Fe samples (Figure 8A). Cluster 5 from Incubations #1 and #4 from Incubation #3 represented a suite of siderophores that were also present across the range of observed Fe concentrations, but they were distinct from Cluster 1 in both experiments since they were less commonly detected than compounds in Cluster 1. Although some clear patterns emerged, there was also a high degree of variability between replicates (Figure S4). Discussion Environmental Nutrient Conditions Stimulate Siderophore Production Siderophore Production Is Not Only Controlled by Dissolved Fe This study aimed to test the hypothesis that distinct external Fe sources would differentially stimulate organic ligand production by the bacterial community, with implications for the bioavailability of Fe to the phytoplankton community. Previous work showed that when an in situ Southern Ocean diatom community was limited by Fe, an increase in strong organic Fe‐binding ligand concentrations was observed, presumably to facilitate Fe uptake and alter the bioavailability of the Fe pool (Buck et al. 2010). This previous study was completed in the same region as the work here but during austral winter, and the observed phytoplankton bloom in bottle incubations started with relatively high initial dissolved Fe concentrations (~5 nM), which decreased rapidly over time in the incubation. It was not until the dissolved Fe concentrations had been sufficiently drawn down and phytoplankton biomass had accumulated that the production of strong organic ligands was observed (Buck et al. 2010). The ligands observed in that experiment were strong Fe‐binding organic ligands consistent with siderophores, but the experiment predated the development of methods capable of measuring these compounds in seawater. The authors hypothesised that siderophores were produced by the in situ heterotrophic bacteria community and not the Fe‐limited diatoms, implying there was a connection between the metabolism of the diatom community and the associated heterotrophic bacteria community. In this study, we saw that siderophore production was most pronounced under Fe‐limiting conditions for the phytoplankton community, consistent with previous work (Buck et al. 2010). Although we observed a similar pattern in strong ligand production as Buck et al. (2010), there were also some important differences. The experiments performed here were initiated in austral spring, later in the growing season and further offshore than those done previously (Buck et al. 2010). Initial dissolved Fe concentrations in the water column of our study were very low (Figure 2A), and a small amount of phytoplankton biomass had already accumulated at this point in the season (Figure 2B). The initial conditions of Incubation #1 had nitrate:dissolved Fe ratios that were within the range that has been previously observed to be limiting to marine diatom communities (~12; King and Barbeau 2007, 2011; Buck et al. 2010) and possibly the heterotrophic bacteria community (~20; Li et al. 2024), and yet the initial siderophore concentrations were not elevated (< 20 pM; Figure 4A). In control treatments from incubation #1, where the nitrate:dissolved Fe remained above the estimated Fe‐limitation threshold for diatoms throughout the experiment (> 12; Figure 2D), siderophore concentrations did not increase until the final time point when significant phytoplankton biomass had accumulated (Figure 2B). Siderophore concentrations began to increase earlier in the +1 and +4 nM Fe treatments, coinciding with a slightly earlier increase in chlorophyll a (Figure 2B). A different pattern was observed in incubation #3. Overall, the siderophore concentrations were 10 times lower throughout the experiment in all treatments (< 15 pM compared to ~150 pM), and although the low Fe treatment had Fe‐limiting conditions for diatoms for all 12 days (nitrate:dissolved Fe > 12 μmol:nmol), there was no significant siderophore accumulation in the final time points when phytoplankton biomass increased. However, the decreases in dissolved Fe and the increases in chlorophyll a over time were not as dramatic in Incubation #3, which could account for some of the differences in siderophore production between the experiments. Although there were low Fe conditions in the dark treatments from both experiments, there was not a similar level of siderophore production as there was in the light treatments (Figure S3A,B). The microbial biomass in the dark treatments was assumed to be dominated by heterotrophic bacteria, in contrast to the light treatments, but we do not have flow cytometry data or another measurement of bacteria biomass in the dark treatments to confirm this assumption. There is no genetic evidence in the literature that diatoms themselves are capable of producing siderophores, and thus it is assumed that the siderophore production in our experiments primarily occurred from either the free‐living (0.2–3.0 μm) or particle (phytoplankton)‐associated (> 3.0 μm) heterotrophic bacteria community (Sandy and Butler 2009). Siderophore production was consistently higher in the lower Fe treatments from both experiments, but there was also high variability between replicates (Figure S4). We suspect this is biological variability rather than analytical, since similar variability was observed in the dissolved metal and chlorophyll a measurements (Burns et al. 2023). Siderophore Production Related to Iron Availability and Phytoplankton Growth Together, the results from these experiments indicated that certain environmental conditions likely induced siderophore production from the marine heterotrophic bacteria community, connected to the growth of phytoplankton and not only the Fe conditions. The limiting nutrient for growth in heterotrophic bacteria in many regions of the ocean is organic carbon (Kirchman 2016). Most of the dissolved organic carbon in the surface waters in the coastal Southern Ocean derives from productivity on the shelves (Carlson et al. 2000), and the labile organic carbon assimilated by heterotrophic bacteria likely derives from organic carbon that is either passively released from cells due to lysis or death, or actively excreted by the phytoplankton during growth. In addition to organic carbon and nutrients, both heterotrophic bacteria and phytoplankton need Fe (Morel et al. 1991) and are thus competing for this scarce resource (Fourquez et al. 2015, 2020; Ratnarajah et al. 2021). Whereas phytoplankton use Fe primarily for nutrient assimilation and photosynthesis (Twining and Baines 2013), heterotrophic bacteria use Fe predominantly for electron transport in the citric acid cycle (Mazzotta et al. 2020). Although it is well documented that Fe limits the growth of phytoplankton in the Southern Ocean, Fe and carbon co‐limitation of bacteria have also been demonstrated in this region (Church et al. 2000; Fourquez et al. 2015, 2020). More recently, the interaction of carbon and Fe limitation to heterotrophic bacteria has also been shown in the California Current (Manck et al. 2024), in the upper mesopelagic open ocean (Li et al. 2024; Mellett et al. 2026) and in the deep chlorophyll maximum (Manck et al. 2026). Together, these studies indicate that if labile organic carbon is available, Fe is likely the next limiting nutrient for heterotrophic bacteria. The production of siderophores is one potential indicator of low Fe availability to the bacterial community and can be used as a biomarker of Fe stress (Li et al. 2024; Mellett et al. 2026). Evidence from our experiments indicates that siderophore production was primarily stimulated when phytoplankton growth had occurred, and Fe was still low. This could indicate that the heterotrophic bacteria community had sufficient organic carbon, provided by phytoplankton growth, but not enough Fe due to an increase in competition for Fe with phytoplankton and/or an increase in the bacteria community demand for Fe from increases in bacteria abundances. In Incubation #1, the peak in siderophore production occurred only in the later stages of the incubation when phytoplankton biomass had accumulated, likely indicating a sufficient source of organic carbon to the heterotrophic bacteria community from phytoplankton exudates. Previous experiments and water column siderophore data have shown that the greatest siderophore production on a per‐cell basis occurs when sufficient labile organic carbon is present, but Fe concentrations are low (Sijerčić and Price 2015; Mellett et al. 2026). The estimated threshold for Fe limitation of heterotrophic bacteria relative to carbon occurs at values of approximately 34 ± 6.6 μmol:mol Fe:C in the surrounding environment (Li et al. 2024; Manck et al. 2024). Unfortunately, we do not have measurements of dissolved organic carbon or the heterotrophic bacteria fraction of particulate organic carbon in these experiments to calculate when the heterotrophic bacteria might have reached this Fe:C threshold, but we can use nitrate relative to Fe as a proxy for Fe stress (Li et al. 2024). When the average Fe and carbon cellular quota that have been estimated in heterotrophic bacteria is converted to nitrogen:Fe, the average cellular requirements of heterotrophic bacteria are ~20 μmol nitrogen:nmol Fe (Bundy et al. 2018; Li et al. 2024). This estimated cellular requirement is derived from one published study on Fe‐limited heterotrophic bacteria in surface waters of the North Pacific (Tortell et al. 1996), but appears to be an appropriate threshold (Li et al. 2024). Thus, we can instead use this nitrate:Fe threshold as a proxy for when the heterotrophic bacteria may have become Fe‐limited in our experiments relative to the phytoplankton community. In the majority of samples, when nitrate:dissolved Fe exceeded 20 μmol:nmol, siderophore concentrations greatly exceeded the initial concentrations (~2 pM). Siderophore concentrations exceeded initial concentrations in 67% of samples when nitrate:dissolved Fe was > 20 μmol:nmol, compared to 48% of samples when nitrate:dissolved Fe was less than this threshold. The greatest siderophore production occurred when nitrate:dissolved Fe ratios indicated that both the phytoplankton and heterotrophic bacterial communities were Fe‐limited (Bruland et al. 2001; Burns et al. 2023; Li et al. 2024), suggesting there are interactions between the availability of carbon, nitrogen and Fe to the phytoplankton and bacterial communities that contribute to siderophore production. For example, mutualistic interactions may be common in this region, whereby phytoplankton provide heterotrophic bacteria with organic carbon during growth, and bacteria produce siderophores to facilitate Fe acquisition for the broader microbial consortia, including phytoplankton (Amin et al. 2009; Sterling et al. 2023). The heterotrophic bacteria may thus alter the bioavailability of Fe for the microbial consortia, ‘fixing’ otherwise inaccessible forms of Fe (e.g., particulate Fe), enabling the uptake of Fe by both phytoplankton and other heterotrophic bacteria (Manck et al. 2022). This hypothesis requires that the phytoplankton cell have a siderophore uptake pathway, which has been observed in the Southern Ocean and in diatoms in general (Timmermans et al. 2001; Hassler and Schoemann 2009; Kazamia et al. 2018; Coale et al. 2019). In control treatments in Incubation #1, greater than 50% of the dissolved Fe pool was bound to siderophores and the diatom community was still able to grow (Figure 2), indicating that internalisation of siderophore‐bound Fe was possible. Network analyses have also shown that bacteria and diatom associations are common in our study region, supporting a mutualistic interaction (Sterling et al. 2023). Intriguingly, very little siderophore production was observed in Incubation #3. However, the nitrate:dissolved Fe ratio only exceeded the 20 μmol:nmol threshold before day 9 in the low Fe treatment, which was where the highest siderophore concentrations were observed (Figure 5). Significant siderophore production was not observed in the dark treatments from both experiments, where no phytoplankton growth occurred. This likely indicates there was lower organic carbon availability and less competition for Fe, providing additional evidence that there is a connection between phytoplankton growth and siderophore production (Figure S3). Distinct Siderophores Produced in High‐ vs. Low‐Iron Conditions The total organic Fe‐binding ligand data indicated there were distinct ligand dynamics in the high Fe versus low Fe conditions, and that the natural and inorganic Fe amendments had comparable bioavailabilities. Although the total ligand concentrations did not change dramatically in the treatments where they were measured, the excess organic ligand concentrations were lower in the high Fe treatments, reflecting the occupation of Fe‐binding sites of the natural ligands with the added Fe. This suggests that the Fe added as inorganic Fe or as natural Fe from shelf waters was able to efficiently bind to the natural ligand pool present in the experimental conditions. This indicates that the natural Fe was either inorganic (e.g., as Fe(II)) or weakly bound in shelf waters, allowing exchange with the strong excess ligands present in offshore waters used for incubation #3. It also demonstrates that the inorganic Fe additions and natural shelf Fe additions behaved similarly in terms of binding to the natural ligand pool and thus similarly in terms of their speciation and bioavailability. This is important because large differences in Fe bioavailability despite similar Fe concentrations could induce differences in siderophore production in high Fe conditions. However, this is likely not the case in these experiments. The greatest siderophore production occurred when the heterotrophic bacteria community was Fe‐limited, but modest siderophore production also occurred when Fe concentrations were elevated. Interestingly, the siderophores that were produced in low Fe conditions were distinct from those that were produced in higher Fe conditions (Figures 7 and 8). The production of different siderophores in high versus low Fe conditions suggests that those compounds could have different microbial sources and/or different ecological functions in the marine environment. While more hydrophilic compounds were often present across experimental time points and treatments, they were the dominant compounds in high Fe conditions (Figure 7B). These compounds were composed of a mix of ferrioxamines (Figure 8). Ferrioxamines have extremely high Fe‐binding affinities, in some cases several orders of magnitude higher than other siderophores whose conditional stability constants in seawater have been measured (Bundy et al. 2018). Ferrioxamines are also unique in that they have been shown to solubilise Fe from particles and can shift the Fe pool from the particulate to the dissolved phase (Borer et al. 2005; Kessler et al. 2020). Other strong siderophores like petrobactin have also been shown to be utilised by the heterotrophic bacteria in order to access particularly refractory or inaccessible Fe phases, such as from minerals or colloids (Manck et al. 2022). The production of these particularly strong siderophores across many of the treatments, particularly in high Fe conditions, indicates that specialists in the microbial community may specifically produce these siderophores to solubilise precipitated Fe or natural Fe that was added from continental margin Fe additions. Ferrioxamines have been observed in other high dissolved and particulate Fe marine environments, such as hydrothermal vents (Hoffman et al. 2024), oxygen‐deficient zones (Moore et al. 2021) and coastal environments (Boiteau et al. 2016, 2019). Importantly, they have been found in higher relative abundance in these high Fe environments than other siderophores found in lower Fe regions (Boiteau et al. 2016). In contrast to ferrioxamines, the hydrophobic and amphiphilic siderophores were produced in higher concentrations overall and were the dominant class of siderophores in lower Fe conditions (Figures 7B and 8). Amphiphilic siderophores, such as the amphibactins and mycobactins identified here, are siderophores that have the potential to be tethered to the cell membrane, limiting their diffusive loss to the surrounding environment. In this study, we cannot distinguish between compounds that might have been present in the phycosphere or tethered to the membrane and those that were released into the dissolved phase. Regardless, siderophore measurements in seawater to date indicate that amphibactins and other amphiphilic siderophores are more commonly found in the ocean than in terrestrial systems and may be more widely conserved in marine bacteria (Vraspir and Butler 2009; Vraspir et al. 2011). Amphiphilic compounds appear to be more common in low Fe environments, and also where other inorganic and/or organic nutrients are elevated, and Fe remains low, such as in surface HNLC waters (Boiteau et al. 2016), the mesopelagic ocean (Bundy et al. 2018, 2024; Park et al. 2023; Li et al. 2024) and low metal hydrothermal vent systems (Hoffman et al. 2024). Other low Fe and low macronutrient waters, such as the subtropical gyre surface waters, do not have an accumulation of these compounds (Boiteau et al. 2016). The results from our experiments are consistent with these findings from the field, as the amphiphilic compounds predominantly appeared when the availability of nitrate relative to dissolved Fe was high in the incubations. Heterotrophic Bacteria Community Siderophore Production and Implications for Iron Bioavailability Although distinct siderophores were observed depending on Fe conditions, a similar heterotrophic bacterial community was present across treatments. This implies that the different compounds produced could represent a specific response to the environmental conditions rather than distinct microbial sources of the unique siderophores. In both incubation experiments, the heterotrophic bacterial community near the end of the experiments was very similar between treatments and was less diverse than the initial conditions (Figure 4, Figures S1 and S2). The similarity in community composition between treatments was surprising, since the total phytoplankton biomass between treatments was very different, likely impacting the amount and type of organic carbon that was available for heterotrophic growth. The dominant bacterial groups in the later sampling time point of each of the incubations were largely copiotrophic bacterial groups that tend to respond when nutrients and organic carbon are plentiful, are fast growing and are generally larger cells with large genomes and greater genetic potential. This was particularly true in the particle‐associated size fraction of the bacterial community in both experiments, where Alteromonadales, Flavobacteriales and Oceanospirillales were the most dominant Orders (Figure 4). It is also possible that as phytoplankton biomass increased in some treatments, the dominance of the particle‐associated bacteria also increased, which may have influenced the differences in the siderophores produced. However, we are unable to tell with only the microbial community composition data if a small subset of the bacterial community was responsible for producing the siderophores we observed, or if specialists had an outsized impact on siderophore production in our experiments. Many of the particle‐associated bacteria we observed were likely responsible for producing the siderophores detected in our experiments. During the same expedition, 153 bacteria isolates were cultured from the particle‐associated size fraction, and 134 of them showed positive siderophore production responses via a colorimetric CAS‐assay (Sterling 2021), indicating that siderophore production potential is high in Southern Ocean particle‐associated bacteria. The dominant groups represented by these isolates were Gammaproteobacteria Pseudoalteromonas spp., Pseudomonas sp., Colwellia spp., Moritella sp., Psychrobacter spp., Psychromonas sp., Halomonas sp., Bacteroidia Polaribacter sp. and Arcticiflavibacteria sp., many of which were observed in our incubation experiments (Figure 4). Four of these isolates, including a Colwellia spp., Moritella sp., Psychrobacter spp. and Pseudoalteromonas spp., were selected for whole‐genome sequencing, and all except for the Pseudoalteromonas strain contained known siderophore biosynthesis genes (Sterling 2021). Interestingly, some of the isolates contained at least partial siderophore biosynthesis pathways for multiple siderophores, including pyoverdine, a common siderophore observed in our experiments (Figure 8). The other siderophore biosynthesis pathways found in these genomes were not detected in our study (Sterling 2021). Beyond the insights gained from these isolates, it remains difficult to connect specific microbes to the siderophores they produce, as we did not collect prokaryotic transcriptomes in this study, the biosynthesis genes for many siderophores are unknown, and the specific siderophore that should be produced based on an identified biosynthetic gene cluster is difficult to predict (Minowa et al. 2007; Hopkinson and Barbeau 2012). Many marine studies have used assays to detect siderophore production in culture, but do not often chemically characterise the siderophores produced. For example, P. haloplanktis produces hydroxamate siderophores in culture under both high Fe and low Fe conditions and had the largest siderophore production rates when organic carbon availability was high, and Fe availability was low (Sijerčić and Price 2015). The exact identity of the siderophore(s) produced in these conditions was unknown, but the authors hypothesised the siderophores might be ferrioxamines, compounds that were common in our experiments (Figure 8). Several marine bacteria have been shown to produce the amphibactins and mycobactins found in the low Fe treatments. A suite of amphibactins was isolated from Halomonas meridiana str. HC4321C1 of the Order Oceanospirillales (Figure 8). Other members of H. meridiana have been isolated from the Antarctic, indicating this group could potentially be present in the Southern Ocean and be a source of amphibactins (James et al. 1990). Mycobactins have only been characterised in pathogenic bacteria thus far, but are far more likely to be associated with the cell membrane than even other amphiphilic siderophores detected in previous studies (Gauglitz and Butler 2013; Kem and Butler 2015). It is possible that other marine microorganisms possess the biosynthesis genes for both amphibactin and mycobactin synthesis beyond those that have been characterised in culture, since they were common across our experiments. The results from this study provide compelling new evidence that heterotrophic bacteria in the Southern Ocean produce distinct siderophores depending on the in situ environmental conditions. While this has never been shown in the field to our knowledge, in culture Shewanella putrefaciens produced either ferrioxamine B or putrebactin depending on substrate availability in the media (Soe et al. 2012). Representative isolates from the Southern Ocean also appear to have the genetic potential to produce multiple, distinct siderophores (Sterling 2021); thus, it is possible that bacteria in this region produce different compounds depending on the environmental conditions and the ecological function of the siderophore. Siderophores have been shown in non‐marine systems to serve a wide range of functions beyond simply Fe acquisition, including cell‐signalling and antibiotic resistance and can be important in a variety of ecological interactions among organisms (Kramer et al. 2020). Certain common siderophores like ferrioxamines might function, for example, as a cooperative shared resource or a ‘public good’, while other siderophores might be ‘privatised’ and are more likely to be upregulated under competitive scenarios (Niehus et al. 2017; Leventhal et al. 2019). Further exploration of this hypothesis is needed in the marine environment, but evidence from this work, along with emerging environmental patterns in siderophore distributions, indicates that different classes of marine siderophores may serve distinct ecological functions in the marine environment, with implications for the overall bioavailability of Fe to the microbial community. Conclusions The experiments presented here demonstrate that siderophore production was stimulated by environmental conditions impacting both phytoplankton and bacteria. The greatest siderophore production occurred when both the phytoplankton and heterotrophic bacteria communities were likely experiencing Fe‐limitation, and in these low Fe conditions more than 50% of the dissolved Fe pool was associated with siderophores. Low Fe conditions alone were not enough to stimulate siderophore production, and phytoplankton growth also played an important role either by providing organic carbon to stimulate bacteria growth or increasing the competition for Fe. Siderophores were also produced when Fe concentrations were elevated, but the types and amount of siderophores produced were distinct from low Fe conditions despite similar heterotrophic bacteria communities. This indicates that the same groups of heterotrophic bacteria or a small subset of the population may produce different siderophores depending on Fe availability and accompanying phytoplankton growth, and that distinct siderophores might have unique ecological functions in terms of facilitating Fe acquisition or altering Fe bioavailability. How siderophore production affects Fe bioavailability, and the interactions between phytoplankton and heterotrophic bacteria in Fe‐limited environments will be important to constrain in future studies. Author Contributions Shannon M. Yang: investigation, writing – review and editing, data curation, methodology. Kristen N. Buck: conceptualization, funding acquisition, writing – review and editing, project administration, supervision, resources. P. Dreux Chappell: conceptualization, funding acquisition, writing – review and editing, visualization, formal analysis, project administration, data curation, supervision, resources. Randelle M. Bundy: investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, data curation, supervision. Alexa Sterling: investigation, writing – review and editing, formal analysis, data curation. Léo Mahieu: investigation, methodology, formal analysis, data curation. Laura E. Moore: investigation, methodology, formal analysis, data curation. Daniel J. Repeta: investigation, methodology, writing – review and editing, supervision. Laura Holland: investigation, methodology, data curation, writing – review and editing. Bethany D. Jenkins: conceptualization, funding acquisition, writing – review and editing, project administration, resources, supervision. Funding This work was supported by the National Science Foundation (PLR‐1443483, PLR‐1443474, PLR‐1443646). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Community composition of free‐living bacteria in Incubations #1 and #3. The bacteria community composition from the free‐living fraction (0.2–3.0 μm filter) from Incubation #1 (A) and Incubation #3 (B) at the level of Order. Each bar represents a single sampling replicate. Figure S2: Statistical differences in bacteria community composition in Incubation #1. A non‐metric multidimensional scaling (NMDS) analysis of the 16S bacteria community composition in Incubation #1 from the initial and final time points (A) and from just the final time points (B). Colours represent the treatments while shapes represent the filter size fraction. Figure S3: Total iron‐binding ligand composition and siderophores in the dark treatments. Ligand characteristics in the dark treatments from Incubation #1 (panels A, C and E; left panels) and Incubation #3 (B, D and F; right panels). Total siderophores (A, B), total organic ligand concentrations (C, D) and conditional binding strengths (E, F) are shown for the control treatment (blue) and the low‐ and high‐iron treatments (light and dark red). Time points shown for Incubation #1 are from the initial and final (Day 14) time points. Time points shown for Incubation #3 are from the initial and final (Day 12) time points. Figure S4: Replicates of putative siderophore chromatograms from the final time point of Incubation #1. The ICP‐MS chromatogram of 56Fe counts per minute from each of the replicates of the control treatment on Day 14 (final time point) of Incubation #1, highlighting the variability in treatment replicates. Figure S5: Putative siderophore identifications of a control replicate from Incubation #1. The ICP‐MS chromatogram of 56Fe counts per minute from the ‘RA2’ replicate (see Table S1) of the control treatment on Day 14 (final time point) of Incubation #1. Each peak that was identified as a putative siderophore is labelled with a letter, and the corresponding putative siderophore identification is shown in the table inset. Some siderophores were found to co‐elute at a similar retention time (e.g., peak d). Other peaks that were identified via ICP‐MS did not have a corresponding siderophore that was identified either via a targeted or untargeted search (e.g., peaks k, m and o). Two peaks were found to correspond to unknown compounds that appear to bind iron (peaks n and p). Table S1: Putative siderophore identifications across all sample replicates. The dissolved putative siderophores identified in all treatments and time points. The ‘x’ indicates the siderophore was present, ‘—’ indicates it was not present, and ‘nd’ indicates that siderophore was not detected at all in any replicates in that experiment. The confidence flag for each siderophore is indicated at the bottom of the table, where a Flag 1 indicates high confidence (siderophore presence is confirmed with a standard), Flag 2 indicates medium confidence (siderophore has been found in previously published studies and the retention time and fragmentation data are consistent with these studies), Flag 3 indicates low confidence (siderophore m/z ratio was found, the compound binds Fe, but the compound has not been confirmed with a standard).