4.1. General Trends in Fe Acquisition
 The slowest volumetric rates of Fe uptake were observed in the presence of the fungal siderophore DFB, and the fastest in the presence of phytic acid. Neither of these Fe complexes is photolabile. As previously observed [Tortell et al., 1996; Maldonado and Price, 1999], the bacterial size fraction was largely responsible for the majority of the total community Fe uptake, accounting for ∼57% of the total volumetric Fe uptake rates in all of our experiments. The 2–20 μm and >20 μm fractions accounted for the remaining ∼30% and 13%, respectively. The fast volumetric rates of Fe uptake by the bacteria were partly explained by their high C biomass, accounting for 45% of the total C biomass in two of the experiments and 15% of the total in the remaining experiment.
 When the volumetric rates were normalized to C biomass, the rates of Fe uptake were still fastest for the 0.2–2 μm size fraction and slowest for the >20 μm fraction. Within a specific ligand and treatment, the difference in C-specific Fe uptake rates between the largest and the smallest size fractions ranged from 3.3- to 57-fold, implying that the bacteria had a more efficient mechanism for acquisition of organically bound Fe, per unit C biomass. This may also reflect the higher Fe requirement of marine heterotrophic [Tortell et al., 1996] and autotrophic bacteria [Sañudo-Wilhelmy et al., 2001; Brand, 1991] relative to eukaryotic phytoplankton [Brand, 1991; Sunda and Hunstman, 1995]. Uptake rates for the eukaryotic phytoplankton ranged between 2.5 and 140 nmol Fe mol C−1 h−1, and were in agreement with those measured in previous investigations of acquisition of Fe bound to strong organic ligands by Fe-limited phytoplankton in HNLC waters, including the subarctic NE Pacific (∼88 nmol Fe mol C−1 h−1 [Maldonado and Price, 1999]) and the polar Southern Ocean (50 nmol Fe mol C−1 h−1 [Maldonado et al., 2001]).
4.2. Fe Acquisition by the Bacterial Plankton Size Fraction (0.2–2 μm)
 In general, the total C biomass in the bacterial size fraction (0.2–2 μm) was partitioned equally between the heterotrophic and the autotrophic (Synechococcus) bacteria. Therefore the uptake of Fe in this size fraction reflects the ability of both functional groups to acquire organically bound Fe. This size fraction achieved the fastest rates of Fe uptake among all size fractions, ranging from 8 to 1400 nmol Fe mol C−1 h−1. The fastest rates of Fe uptake by the bacteria (in the presence of phytic acid and gallocatechin) were 10 times faster than those measured previously with FeDFB in the subarctic NE Pacific [Maldonado and Price, 1999], and may thus reflect the difference in bioavailability of specific Fe siderophores/chelators. These fast Fe uptake rates from organically bound Fe do not indicate whether the bacterioplankton are accessing organic Fe directly or simply utilizing the inorganic Fe that is liberated from these complexes.
 Previous laboratory studies suggest that inorganic Fe acquisition by marine heterotrophic bacteria is not upregulated under Fe stress, and it is extremely inefficient compared to organic Fe acquisition [Granger and Price, 1999]. Iron limited heterotrophic bacteria, like those in the Southern Ocean [Pakulski et al., 1996; Arrieta et al., 2004], acquire Fe-siderophore complexes at very fast rates by either internalizing the Fe-siderophore complex or by an exchange reaction of the Fe between the Fe-siderophore and a transport ligand at the cell surface (see below) [Granger and Price, 1999]. When the Fe-siderophore is internalized as a complex, marine bacteria, like terrestrial and pathogenic bacteria [Byers and Arceneaux, 1998], are believed to mediate the dissociation of Fe from the siderophore intracellularly or in the periplasmic space. Physiological data indicates that similar siderophore-mediated high-affinity Fe transport systems occur in Synechococcus [Wilhelm and Trick, 1994; Trick and Wilhelm, 1995; Webb et al., 2001]. It thus seems most likely that the rates of Fe uptake by heterotrophic and autotrophic bacteria (0.2–2 μm size fraction) in the Southern Ocean reflect their ability to acquire Fe bound to strong organic ligands directly.
4.3. Fe Acquisition by the Eukaryotic Plankton Size Fractions (2–20 and >20 μm)
 Thus far, eukaryotic phytoplankton have not been shown to internalize intact organic Fe complexes, and it is believed that these organisms possess Fe transporters at the cell surface that react specifically with inorganic Fe species [Hudson and Morel, 1990; Sunda and Huntsman, 1995]. However, when the concentrations of inorganic Fe are insufficient for growth, Fe-limited eukaryotic phytoplankton are able to access Fe bound within organic Fe complexes [Soria-Dengg and Horstmann, 1995; Maldonado and Price, 1999, 2001]. For our experiments, we calculated the rate of inorganic Fe supply to the cell surface in order to determine whether eukaryotic phytoplankton during FeCycle could meet their Fe uptake demand without accessing the Fe bound within the strong organic Fe complexes. Using DFB and EDTA as model ligands, we calculated the supply rate of inorganic Fe (Fe′) in seawater at the concentration of FeL added in the uptake experiments (Fe′ supply rate = k′d × [FeL]; where k′d for FeEDTA = 1 × 10−6 s−1, and k′d for FeDFB = 63.25 × 10−12 s−1 [Hudson et al., 1992; Rue and Bruland, 1995]). Our calculation indicates that the rate of inorganic Fe supplied by thermodynamic dissociation of Fe from FeDFB and FeEDTA is 0.00045 pmol Fe L−1 h−1 and 7.2 pmol Fe L−1 h−1, respectively. In the case of FeEDTA, this rate of inorganic Fe dissociation is sufficient to supply inorganic Fe for the rates of Fe uptake by the eukaryotic phytoplankton (0.06–1.47 pmol Fe L−1 h−1, Table 3). However, in the case of DFB, the inorganic Fe supply, as a result of thermodynamic dissociation, is too slow to account for the measured rates of Fe uptake (0.012 and 0.03 pmol Fe L−1 h−1, Table 3). Thus, in our Fe uptake experiments with the strongest organic Fe complexes, which have very slow dissociation rate constants, eukaryotic phytoplankton appear to access organically bound Fe.
 Two mechanisms have been proposed to allow phytoplankton to access the Fe within organic Fe complexes. Photoreduction and biologically mediated Fe(III) reduction enhance the rate of dissociation of Fe from strong organic complexes. These reduction processes may result in higher concentrations of inorganic Fe in the vicinity of the cell, and thus faster uptake rates. The results of the present study suggest that photochemistry is not the main mechanism mediating Fe acquisition from model Fe organic ligands by plankton in the Southern Ocean. The C-specific rates of Fe uptake for all size fractions were faster for the non-photolabile (phytic acid and gallocatechin) than for the photolabile Fe complexes (DTPA, EDTA, and rhodotorulic acid). The exception was FeDFB; the rates of Fe uptake in the presence of this non-photolabile siderophore were the slowest among all ligands. The order of preference of the organically bound Fe complexes was similar for all size fractions. Iron bound to the non-photolabile Fe complex phytic acid was the preferred FeL for Fe acquisition, followed by gallocatechin, EDTA or DTPA, rhodotorulic acid, and DFB. These results strongly suggest that photolability of the model Fe-complexes is not the most important mechanism liberating Fe from these organic Fe complexes.
 While the rates of Fe uptake from all ligands were faster in the presence of light than in darkness, we believe that this artifactual effect was due to the physiological state of the plankton in the dark treatment. The cells in this treatment were placed in the dark continuously for 24–30 hours. Under this condition, most likely, the growth rates of the phytoplankton declined slightly, their Fe demand was lower, and thus their Fe uptake rates slowed down. The observation that the rates of Fe uptake for all ligands, except EDTA, were ∼2 times faster in the presence of light than in its absence, regardless of their photolability, further supports the effect of darkness on cell physiology. We thus believe that the light enhancement observed in the presence of the non-photolabile ligands (DFB, phytic acid, and gallocatechin), on average a twofold increase, was exclusively due to their physiological state in the dark treatment. Any additional light enhancement over and above this twofold increase may be due to photoreduction of the Fe complex. For example, in the EDTA treatment, the rates of Fe uptake were on average 6.5 times faster in the light than in the dark, respectively. Thus, in the presence of FeEDTA (a highly photolabile Fe complex), light-mediated reduction of organically bound Fe(III) may be an important pathway to increase the inorganic Fe pool available for Fe uptake [Anderson and Morel, 1982].
 Given that photolability of the different model organic Fe complexes did not control the bioavailability of Fe for eukaryotic plankton in the Southern Ocean, it seems likely that Fe acquisition of organically bound Fe was mediated by biological reduction at the cell surface. Unfortunately, rates of biological reduction of Fe were not determined during FeCycle. However, indirect evidence can be used to support the role of biological Fe reduction for acquisition of Fe from strong organic Fe complexes. Using previous field (NE subarctic Pacific) Fe(III) bioreduction rate data (13.6 μmol Fe mol C−1 h−1 at 0.1 μmol L−1 FeDF [Maldonado and Price, 1999]), we are able to compare the rates of supply of Fe by biological reduction and the rates of Fe uptake measured here. Note that the reduction rates of Fe(III) bound to weak (NTA, EDTA, and DTPA) and strong organic complexes (DFB and DFE) only vary by a factor of 2, like the reduction potentials of these Fe complexes [Maldonado and Price, 2001]. The reduction potentials of all the ligands used in the present study are between that of DFB (−468 mV [Boukhalfa and Crumbliss, 2002]) and that of EDTA (+94 mV [Soria-Dengg and Horstmann, 1995]), and thus the reduction rates of FeDF measured previously are probably applicable to all ligands used in our experiments. For this calculation we assumed that the Fe(III)DF reduction rates by plankton in the subarctic Pacific Ocean (summer waters T ∼ 8°–12°C) are equivalent to those by plankton in the subantarctic Southern Ocean (summer waters T ∼ 12°C), that the rates scale linearly with substrate concentration (from 2 to 100 nmol L−1 FeL [Maldonado and Price, 2001]), and that the average plankton C biomass in the 2–20 μm and >20 μm fractions is 3.5 and 2 μmol C L−1, respectively (Table 2). Under these assumptions, the rates of Fe supply resulting from biologically mediated Fe reduction of organically bound Fe complexes would be 0.95 and 0.54 pmol Fe L−1 h−1 for the 2–20 μm and >20 μm size fraction, respectively. These supply rates are in close agreement with the rates of Fe uptake we measured in the presence of various Fe complexes. The rates of Fe uptake for the 2–20 μm fraction ranged from 0.028 to 1.65 pmol Fe L−1 h−1, while those for the >20 μm fraction ranged from 0.016 to 0.62 pmol Fe L−1 h−1. These calculations and comparisons suggest that extracellular enzymatic reduction of Fe(III) may be an important mechanism for Fe acquisition of organically bound Fe by eukaryotic phytoplankton in the Southern Ocean. Once the Fe(II) is reduced, the affinity of the ligand for the reduced Fe(II) will determine the rate of Fe dissociation, and thus Fe uptake. According to the affinity constants of the ligands for Fe(II), Fe dissociation will be faster for the gallocatechin (log K Fe(II)L = 8.79 M−1, calculated using the equation: E = Eo − 0.05916 * log (KFe(III)/KFe(II)), and E and KFe(III)L values from Jovanovic et al. ) than for the rhodotorulic acid and DFB complexes (log K Fe(II)L = 10.6 M−1 and 10.3 M−1, respectively [Boukhalfa and Crumbliss, 2002; Cooper et al., 1978]). This trend is consistent with that observed for the rates of Fe uptake in the presence of these ligands during our experiments.
 Alternatively, if the uptake of organically bound Fe by phytoplankton and bacteria is mediated by the exchange of Fe(III) between ligands in seawater and specific Fe transporters at the cell surface, the denticity of the ligands (the number of donor groups from a given ligand attached to the same central Fe atom) will partially determine the bioavailability of the organically complexed Fe [Boukhalfa and Crumbliss, 2002]. According to the denticity of the ligand, Fe bound to DFB, a hexadentate siderophore, and rhodotorulic acid (tetradentate) will be less available for exchange than other ligands of lower denticity, such as phytic acid (tridentate) and gallocatechin (bidentate). Indeed, in support of the role of denticity, the rates of Fe uptake from Fe bound to strong organic complexes were the fastest for phytic acid and gallocatechin, followed by rhodotorulic acid and DFB.
4.4. Bioavailability of Fe Bound to In Situ Organic Ligands
 This study shows that plankton in the subantarctic HNLC waters are able to access Fe bound to the in situ ligands in the dark, albeit at significantly slower rates than that bound to the model ligands (EDTA, DTPA, gallocatechin, rhodotorulic acid, and phytic acid). Interestingly, in the dark, the rates of algal Fe uptake by the >20 μm size fraction from the in situ Fe-ligand treatment were identical to those in the FeDFB treatment, suggesting that these in situ organic Fe-complexes may be similar to FeDFB. Indeed, the conditional stability constants of FeDFB (Table 1) and of the surface in situ ligands (Croot et al., submitted manuscript, 2005), during the FeCycle were indistinguishable (log K′FeL, Fe′ = 11.8 ± 0.2 versus 11.8 ± 0.3, respectively). In contrast to FeDFB, however, these in situ Fe-ligands seem to be extremely photolabile, as light enhanced the rates of Fe uptake by 15-fold, on average, for all size fractions. This was the greatest light enhancement effect observed in all the Fe uptake experiments. The importance of light in enhancing Fe uptake from the in situ Fe-ligands suggests that some naturally occurring organic ligands are photolabile, and that light in the photic zone partially increases the rates of Fe uptake from the dissolved organic Fe pool. These findings are in agreement with those of recent studies on the photochemical degradation of organic iron complexes in coastal waters, which results in an increase of dissolved inorganic Fe concentrations [Powell and Wilson-Finelli, 2003].
 Two main mechanisms have been proposed to mediate photochemical reactions of organically bound Fe in the sea. The first mechanism is indirect, and involves chromophoric dissolved organic matter (CDOM) as a source of photochemical radicals (i.e. H2O2) for Fe(III) reduction. However, low CDOM concentrations and sunlight in Antarctic waters limit surface H2O2 levels [Sarthou et al., 1997], and thus this indirect photoinduced Fe(III) reduction mechanism is not believed to be significant in HNLC regions of the Southern Ocean [Rijkenberg et al., 2005]. The second photochemical reaction mechanism is direct, mediating a light-induced ligand-to-metal charge transfer reaction, and resulting in the reduction of Fe(III) and the partial oxidation of the ligand [Barbeau et al., 2001, 2003]. This direct photochemical mechanism is most likely important in oligotrophic and/or Fe-limited regions, where CDOM concentrations are low and the vast majority of the dissolved Fe is bound to strong organic complexes.
 Even though little is know about the origin and nature of in situ Fe(III) ligands in seawater, these ligands have functional groups [Macrellis et al., 2001], and Fe(III) conditional stability constants typical of siderophores [Rue and Bruland, 1995]. A recent study on the photolability of Fe(III) siderophore complexes produced by marine heterotrophic bacteria (Fe-aquachelin) indicates that the photolysis of Fe(III) aquachelin complexes leads to the formation of a lower affinity Fe(III) ligand and the reduction of Fe(III) [Barbeau et al., 2001]. This photochemical Fe(III) reduction readily occurs in polycarbonate bottles exposed to natural sunlight, and increases the bioavailability of organically bound Fe to indigenous plankton in the Atlantic Ocean [Barbeau et al., 2001]. As in the work of Barbeau et al. , we used polycarbonate bottles, which filter out most UV wavelengths (<350 nm). Our results thus suggest that the photochemical reactions of the in situ Fe-ligands are mainly mediated by the visible portion of the light spectrum (400–700 nm) and to a lesser degree by UVA (320–400 nm). Given that during austral summer in these subantarctic waters, the visible portion of the light spectrum (FeCycle 1% PAR at 50 m) and UVA penetrate deeper (FeCycle 1% UVA at ∼34 m) than UVB (280–320 nm, FeCycle 1% UVB at ∼19 m) (Croot et al., submitted manuscript, 2005), wavelengths >350 nm would be more effective in mediating photochemical reactions of ferrated in situ ligands in the euphotic zone.
 The photochemical reaction in the Fe-aquachelin complex is mediated by the α-hydroxy acid moiety [Barbeau et al., 2001, 2003], a functional group which is common in the marine siderophores characterized to date [Reid et al., 1993; Haygood et al., 1993; Martinez et al., 2000]. Indeed, most of the siderophores produced by open ocean marine bacteria are mixed-moeity ligands, containing hydroxymate and catechol functional groups, as well as the α-hydroxy carboxylate moiety [Reid et al., 1993; Haygood et al., 1993; Martinez et al., 2000; Barbeau et al., 2001, 2003]. However, the α-hydroxy carboxylate moiety is the one that imparts photoreactivity to the ferrated, mixed, marine bacterial siderophores studied to date [Barbeau et al., 2003]. The in situ Fe(III) ligands in the Southern Ocean may well be mixed-moeity siderophores, with hydroxymate or catecholate functional groups and the photoreactive α-hydroxy carboxylate moeity. The photolability of these in situ Fe-ligands may thus account for the considerably faster rates of Fe uptake in the light than in the dark. Interestingly, the concentrations of strong natural ligands during the FeCycle were slightly lower at shallow depths (∼1 versus 2 nmol L−1 (Croot et al., submitted manuscript, 2005)), perhaps indicating photodegradation of these ligands in the upper water column.