Concentrations of soluble (<0.025 μm), dissolved (<0.4 μm) and total (unfiltered) iron (Fe) were measured over the continental shelf and slope of the northern Gulf of Alaska (GoA) during spring-summer. Large cross-shelf gradients of surface water Fe concentrations were observed in these productive shelf waters during both seasons. Most of the particulate (>0.4 μm) and colloidal (0.025–0.4 μm) Fe size fractions were removed from surface waters within the inner and mid shelf. As a result the contribution of soluble Fe to the total Fe concentration increased from the inner shelf to the shelf break/slope waters. Surface water dissolved Fe concentrations on the northern GoA continental slope were higher than those previously observed in the central GoA gyre. Variations in surface water Fe concentrations from spring through summer appear to result from the changes in freshwater discharge and physical processes on the shelf.
 The continental shelf and slope of the northern Gulf of Alaska (GoA) is biologically productive [Sambrotto and Lorenzen, 1986] and supports diverse marine populations that include several important commercial fisheries [Ware and McFarlane, 1989]. Sambrotto and Lorenzen  and Stabeno et al.  hypothesize that the high productivity results from the cross–shore convergence of iron (Fe)-poor, but nitrate-rich, offshore waters with iron-rich, nitrate-poor, coastal waters. The large annual freshwater discharge to the coastal GoA from surrounding glaciers and mountains carries an enormous suspended load [Milliman and Syvitski, 1992], which is likely a major source of Fe (∼4% Fe by weight for mineral particles). The cross-shelf transport of freshwater and its iron content are mediated by the winds and the shelf and slope current systems and likely involve a number of poorly resolved processes. Re-suspended Fe-rich bottom sediments are an unlikely Fe source for the shelf euphotic zone during summer, except during episodic storms and upwelling events [Stabeno et al., 2004], because the northern GoA shelf, in addition to being subject to downwelling year-round, is deep (>150 m) and capped by a strong pycnocline from late spring through early fall. However, deep winter mixing can occur [Royer, 1975] and potentially supply sub-surface Fe to the surface prior to the spring bloom [Wu and Luther, 1996].
 We present Fe data from seawater samples collected opportunistically during two Northeast Pacific Global Ocean Ecosystem Dynamics (NEP-GLOBEC) cruises in the northern coastal GoA. Cross-shelf gradients of Fe concentrations in different size fractions and their relationships to biological, macronutrient, and physical parameters are described to provide insights on the shelf Fe supply and transport pathways.
2. Sampling and Analytical Methods
 Seawater samples for Fe analyses were collected in May and July 2004 in the northern GoA at stations shown in Figure 1. Sampling occurred in Prince William Sound (HB4 and MS2), upstream (HE stations) and downstream (GAK stations along the Seward Line) of the Sound, and within the Alaska Coastal Current (ACC), slope current, and mid-shelf domain. Surface samples were collected at 0.5 m depth using an underway sampling system [Vink et al., 2002] and sub-surface samples were collected using the UAF ATE/Vane Fe sampler [Wu, 2007]. Sample handling and analysis were conducted using rigorous trace metal clean procedures [Wu et al., 2001]. Sample replicates were filtered through 0.4 μm Nuclepore and 0.025 μm Millipore filter membranes. All samples were acidified to pH 2 with the equivalent of 1 ml of 12 M hydrochloric acid (Optima grade, Fisher Scientific Inc. with Fe blank ∼2 nM) per liter of sample. Acidified samples were stored at room temperature for 6 months prior to Fe analysis using Mg(OH)2 coprecipitation isotope dilution ICPMS [Wu, 2007] with a detection limit of ∼0.03 nM and a precision of ∼3% at 0.3 nM Fe level. The operationally defined size fractions of Fe referred herein are as follows according to Wu et al. . Iron that has passed through 0.025 μm Millipore filter membranes is defined as soluble Fe (SFe), whereas dissolved Fe (DFe) is the Fe that passed through 0.4 μm Nuclepore filter membranes. Colloidal Fe (CFe) is the difference between DFe and SFe, total Fe (TFe) is the Fe that can be dissolved at pH 2 in unfiltered samples, and particulate Fe (PFe) is the difference between TFe and DFe.
3. Results and Discussion
Figures 2a–2f show the cross-shelf distributions of salinity, chlorophyll, Si(OH)3 and NO3 + NO2 (N + N) averaged between 0–30 m (the approximate depth of the euphotic zone), and different size fractions of surface water (0.5 m depth) Fe for May and July on the Seward Line (Additional cruise data: http://www.ims.uaf.edu/GLOBEC/results/index.html). In May, the ACC was within 50 km of the coast, with euphotic zone salinities of 30.8–31.8. The spring bloom within the ACC was in progress as evident by the relatively high chlorophyll (2–4.5 μg l−1) and low N + N (∼2–10 μM) and Si(OH)3 (∼10–18 μM). Offshore of the ACC, chlorophyll concentrations were low (∼1.5μg l-1) although N + N (∼7–10 μM) and Si(OH)3 (15–20 μM) were abundant. There were two offshore bands of elevated chlorophyll (3–4.5 μg l-1) and reduced concentrations of N + N and Si(OH)3 centered on the shelfbreak (175 km) and slope (225–250 km). These bands may be filaments shed from a Yakutat Eddy that appears to have been impinging upon the slope ∼100 km east of the Seward Line in May based upon inspection of satellite altimetry [Janout et al., 2009]. In July, ACC salinities had freshened (30.4 at the coastal-most station) due to increasing runoff, and low-salinity ACC surface waters extended further ∼100 km offshore due to the seasonal relaxation in downwelling wind. Chlorophyll and N + N concentrations were low (1–2 μgl−1; ∼2 μM), while Si(OH)3 concentrations ranged from 8–15 μM. Stratification was strong (<0.05 kg m−4 at ∼25 m; data not shown) over the entire section. The concentration of Fe in all size fractions was elevated within the ACC, and decreased offshore by 1–3 orders of magnitude with a stronger gradient in July than in May (Figures 2d–2f). Iron concentrations in the inner shelf stations were higher in July than in May reflecting the larger coastal fresh water discharge in July, as evident in the lower surface salinities (Figures 2d–2f). The concentrations of TFe and DFe in offshore stations were higher in May than in July, likely reflecting deep winter mixing (the mid- and outer shelf was unstratified in May) or perhaps the effects of upwelling within the shelfbreak front.
 The TFe and DFe cross-shelf gradients observed along the Seward Line mainly resulted from the variations of PFe and CFe. For example, the gradient in TFe (from ∼730 nM to ∼0.5 nM) from GAK1 to GAK7 in July was mainly due to a large decrease in PFe (∼720 nM). Similarly, the gradient in DFe (from ∼8.5 nM to 0.2 nM) from GAK1 to GAK7 in July was caused by a decrease in CFe of ∼6.4 nM. As shown in Figure 3a, the bulk of the PFe and the CFe was rapidly removed from surface waters. This indicates that a large fraction of the coastal Fe input to the northern GoA is effectively trapped in the inner shelf, and we propose this occurs mainly through the removal of PFe and CFe. While PFe removal can result from particle settling, the removal of CFe may be due to colloid coagulation, particle scavenging, and/or the lack of Fe-binding organic ligands in the colloidal size fraction that results in lower residence times for CFe. Runoff and the mean downwelling winds generate the coastally-trapped, low-salinity ACC [Stabeno et al., 1995; Weingartner et al., 2005] on the shelf. The large annual coastal runoff is at least 24000 m3 s−1 [Royer, 1982; Weingartner et al., 2005], and carries an enormous suspended load of Fe-rich particles. The discharge is a minimum in winter, increases rapidly in summer due to snowmelt, and is a maximum in fall when coastal precipitation rates are also a maximum [Royer, 1982]. Thus, we view the ACC as an Fe reservoir (and alongshore vehicle) because it receives a continual, albeit seasonally-varying, Fe supply via the Gulf-wide coastal freshwater discharge, and seems to prevent the bulk of this Fe input from being transported offshore. In addition, cross-shelf dispersal of ACC waters is constrained by the onshelf Ekman transport of surface waters from the GoA basin due to the year-round downwelling winds [Weingartner et al., 2005]. In contrast to PFe and CFe, SFe exhibits a nearly linear relationship with salinity across the shelf, suggesting a longer residence time for this size fraction (Figure 3a). Cross-shelf transport of Fe, however, may be possible through episodic (and unpredictable) events such as interactions of the ACC with the bottom topography or the coastline [Weingartner et al., 2005] and/or mesoscale eddies and meanders along the shelfbreak [Janout et al., 2009]. The frequency of mesoscale eddies in the GoA is at a maximum in spring/summer [Henson and Thomas, 2008] when freshwater discharge with its enormous suspended particulate load is increasing due to snowmelt. Cross-shelf transport of Fe via mesoscale eddies has been shown in the southeastern side of the GoA basin [Johnson et al., 2005]. In addition, coastal downwelling could induce a sub-surface offshore export of Fe, some of which may become available to the surface waters upon winter mixing. Lateral delivery of iron from the northern GoA continental margin, in conjunction with deep winter mixing, has been suggested as a wintertime iron source to surface waters as far offshore as Ocean Station Papa [Lam et al., 2006]. Figure 3b shows vertical profiles of DFe, SFe, and CFe in July in the offshore station GAK 11. In both May and July, DFe was lower in the near surface waters and increased with depth. The main form of DFe throughout the water column in July was SFe. As in surface waters, the removal of CFe from inshore to offshore likely takes place throughout the water column. This suggestion is supported by data from Nishioka et al.  that show a large cross-shelf gradient in “small colloidal Fe” throughout the water column in the southeastern GoA (line P), with soluble Fe being more abundant than “small colloidal Fe” at offshore stations. The concentration of SFe throughout the water column at GAK11 and in the surface from GAK1-5, is higher than inorganic Fe solubility (∼0.08 nM, as determined in ligand free, UV irradiated seawater [Wu et al., 2001]), suggesting that a portion of SFe may be bound to soluble Fe-complexing natural organic ligands that helped stabilize and keep Fe in solution. The DFe concentration at GAK 11 (Figure 3b) was higher than those reported for station Papa in the GoA central basin [Martin and Fitzwater, 1988; Nishioka et al., 2001]. If these waters could be mixed or transported south by eddies or other mechanisms, they could provide Fe to the central gyre, as has been suggested by Johnson et al. . Strom et al.  suggested that the large cross-shelf gradients in phytoplankton biomass, cell size, species composition, growth rates, and macronutrient utilization in the northern GoA were consistent with cross-shore gradients in iron availability. Iron data was not presented in support their argument. However, the strong cross-shelf gradients in Fe we observed during both sampling periods along the Seward Line are supportive of Strom et al.  contention.
 Along the Seward Line strong cross-shelf gradients in the different size fractions studied were more pronounced in July than in May. The difference between the two seasons was mainly due to increases in freshwater runoff during July resulting in higher Fe concentrations at the inshore-most station. Iron concentrations as a function of salinity showed rapid removal of PFe and CFe within the ACC (Figure 3a). The coastally-trapped, low-salinity ACC constrains the input of freshwater to the inner shelf, and prevents the bulk of Fe delivered via freshwater runoff from being transported offshore. In contrast to PFe and CFe, an almost linear relationship existed between SFe and salinity (Figure 3a), suggesting a longer residence time for this fraction of DFe in surface waters. As a result the contribution of soluble Fe to the total Fe concentration increased from the inner shelf to the shelf break/slope waters. Offshore, the vertical distribution of DFe was also dominated by the SFe fraction (Figure 3b).
 This work is supported by funding from NSF (OCE-0220978, OCE-0321402, OCE-0728930 and ARC-0612538 to J.W.) and CNSF (No. 40776042 to J. Wu and No. 40528007 and 90411016 to J. Wu and M. Chen) and SCSIO. We are grateful to Yihua Cai and Russell Hopcroft who help collecting seawater samples and the captain and crew of R/V Alpha-Helix.