Trace element sampling and shipboard flow injection analysis during the June–August 2003 Climate Variability and Predictability (CLIVAR)-CO2 Repeat Hydrography A16N transect has produced a high-resolution section of dissolved Fe and Al in the upper 1000 m of the Atlantic Ocean between 62°N and 5°S. Using the surface water dissolved Al and the Model of Aluminum for Dust Calculation in Oceanic Waters (MADCOW) model we have calculated the deposition of mineral dust to the surface ocean along this transect and compare that to dissolved Fe concentrations. The lowest mean mineral dust depositions of ≤0.2 g m−2 a−1 are found to the north of 51°N; a region which also exhibits characteristics of biological Fe limitation through its low dissolved surface water Fe (∼0.1 nM) and residual macronutrients, e.g., nitrate >2 μM. To the south of this region, mean dust deposition increases by an order of magnitude reaching ∼3 g m−2 a−1 at 10°N, underneath the Saharan dust outflow. Surface water Fe values also increase along this section to >1 nM. Distinct minima in Fe concentrations at the depth of the chlorophyll maximum in the vertical profiles between 18 and 4°N illuminate the role that active biological uptake plays in Fe cycling. An extensive subsurface zone of enhanced dissolved Fe concentrations (>1.5 nM) underlying this region is a result of the biological vertical transport and remineralization of the surface water Fe and is coincident with the intermediate nutrient maximum and oxygen minimum of this region. Elevated concentrations of dissolved Al in subsurface waters seen between 30 and 20°N coincide with the domain of the subtropical mode waters (STMW) which result from the sinking of surface waters in late winter in regions imprinted by dust deposition. The magnitude of the Al enrichment observed in this water mass implies that the predominant source to the STMW is from the more dust-impacted western Atlantic, with only limited contributions from the STMW formation region near Madeira. A deeper subsurface Al enrichment (30–45°N) is associated with the outflow from the Mediterranean, another heavily dust-impacted basin. These two regions of Al enrichment show the widespread geochemical connection between atmospheric transport processes and the North Atlantic and underscore its susceptibility to imprinting by atmospherically borne materials, natural as well as anthropogenic.
 Understanding the linkages between atmospheric and oceanic processes is crucial to developing realistic global models of climate feedback [Falkowski et al., 1998, 2000; Fung et al., 2000]. Nowhere is this need more apparent than in the necessity to incorporate the factors that control the availability of Fe in surface ocean waters, the lack of which has been shown to limit biological processes and carbon sequestration in large areas of the North, South and equatorial Pacific Ocean [Martin and Fitzwater, 1988; Martin et al., 1990a]. It is well recognized that the atmospheric delivery of continental mineral dust to, and its partial dissolution within, remote surface ocean waters is an important vector of Fe delivery [Jickells et al., 2005]. However, there are no actual dust deposition data over the open ocean, only estimates extrapolated from adjacent land-based sampling sites [Duce et al., 1991]. Dissolved Al concentrations in surface waters can be used to estimate dust deposition to the surface of the open ocean [Measures and Brown, 1996; Measures and Vink, 2000], and these can then be compared to dissolved Fe distributions to infer fluxes and to characterize processes.
 However, our ability to investigate and model these processes has been severely limited by our ability to obtain data for these, and other trace elements, in sufficient quantity to resolve features and map out their geographical extent. Part of the reason for the dearth of information is due to the labor and time-intensive methods that have been used historically to collect trace element samples at sea and which has made this work incompatible with the large-scale hydrography programs that provide global sampling opportunities. We report here the first set of data from the North Atlantic, using a rosette-based sampling system for trace elements that was designed to enable the collection of a high-resolution section (approximately 1° spacing) of the dissolved trace elements Fe and Al in the upper 1000 m, as part of the international Climate Variability and Predictability (CLIVAR)-CO2, Repeat Hydrography program during the A16N cruise.
 The principal motivation for developing sections specifically for Fe and Al is related to the apparent role that the availability of the micronutrient Fe in surface waters plays in moderating oceanic biological processes, and the role that atmospheric deposition of mineral dust in surface waters, traced by dissolved Al, plays in delivering Fe to the surface waters of the remote oceans. Since the recognition that limited Fe availability in some oceanic surface waters might be an important component of the glacial-interglacial carbon dioxide feedback loop [Martin et al., 1990b], there has been an impetus to understand the systematics of oceanic Fe geochemistry, and to account for these processes in global models. In particular, this requires a better understanding of how Fe, and other biogeochemically important trace elements enter the ocean through atmospheric deposition processes. To accomplish this requires detailed trace element sections across a wide range of oceanic biogeochemical regimes and atmospheric deposition gradients. The work reported here is the first CLIVAR trace element contribution to that global goal.
 The CLIVAR-CO2 Repeat Hydrography program's A16N cruise was conducted aboard the NOAA research vessel Ron Brown which left Reykjavik, Iceland on 19 June 2003 and terminated in Natal, Brazil, on 12 August 2003. Details of the cruise track and ancillary data associated with the program can be found at http://cchdo.ucsd.edu/data_access?ExpoCode=33RO200306_01.
 Seawater samples were collected using 12L GO-FLO bottles (General Oceanic) mounted on a conventional rosette frame, containing commercially available CTD and oxygen sensors (SeaBird SBE 911and SBE 42), and a fluorometer (WetLabs FL1). The aluminum rosette frame was completely painted to eliminate most bare metal surfaces. The package was suspended on a 4 conductor Kevlar cable (Cortland Cable), which passed through a Nylatron block (General Oceanic) and was deployed using a SeaMac winch with nylon rollers and level wind. To prevent trace element contamination of the sampled water from the few uncoated metal surfaces that remained, sample bottles were closed by electrical signal from the ship only while the rosette frame was moving upward through the water column at 5–10 m min−1 into water that had not been in contact with the frame. A 12-depth profile in the upper 1000 m was routinely collected in approximately 1 h. Problems with the weight handling ability of the winch during the first leg of the cruise (Stations 5–78; 62 to 27°N) resulted in deployment of only 10 or 11 bottles on the rosette, and sampling was restricted to the upper 750 m. At Station 120, 6°N, a second pattern of sampling depths was introduced. The two sampling patterns were then alternated to more closely match the sampling depths used by the main hydrography CTD and also to improve the contouring of the TM data sets. Immediately after package recovery the 12 L bottles were removed from the rosette frame and carried into a 20 ft laboratory container van equipped with a HEPA filtered air system (Mac 10, ENVIRCO) where subsampling was completed under trace element clean conditions. A detailed description of the sampling system and its construction will be provided elsewhere (C. I. Measures et al., A rosette system for trace metal clean sampling, submitted to Limnology Oceanography: Methods, 2007).
 Seawater subsamples were filtered through 0.4 μm 47 mm polycarbonate track-etched filters (GE Poretics part number K04CP04700) held in a MFS polypropylene filter holder. Subsequent to this cruise, the combined rosette and subsampling scheme was compared with other Fe sampling methods, during the NSF-sponsored SaFe intercalibration cruise (October 2004) and was found to produce samples that were essentially identical to those from other sampling systems [Johnson et al., 2007].
 Subsamples were run on board ship within 24 h of collection using the flow injection analysis (FIA) methods for dissolved Al and Fe [Resing and Measures, 1994; Measures et al., 1995]. Detection limits and precisions during this cruise were approximately 0.5 and 0.1 nM, and 3.0 and 2.5%, respectively. Shipboard dissolved Fe data sets have been validated and corrected by shore-based dissolved Fe determinations on replicate dissolved samples returned to FSU determined by WML using the inductively coupled plasma-mass spectrometry (ICP-MS) method of Wu and Boyle . The correction consisted of quantifying and then subtracting the daily variation of a previously unidentified blank in the shipboard FIA method. This blank was estimated by correlating the shore based ICP-MS Fe determinations of a subset of each day's shipboard run of samples against their shipboard values and subtracting the derived offset from each sample run on that particular day at sea.
3. Results and Discussion
 The data, which consist of 659 samples from 62 profiles spaced at approximately 1 degree intervals between 62°N and 5°S, are presented as three contour plots produced using Ocean Data View (http://odv.awi-bremerhaven.de), Figure 1. Each of the panels is color-coded to the parameter concentration, and each is overlaid with contour lines depicting either potential density (Figure 1, top and bottom) or oxygen concentration (Figure 1, middle).
3.1. Al Distributions
 We will first describe and interpret the dissolved Al data set and use the insights that this provides to interpret the dissolved Fe section. We will concentrate our attention on the three major features that the Al section shows. These are the surface water values that can be used to infer the magnitude of dust deposition and the two subsurface regions of enriched Al between 20° and 40°N.
 In order to calculate dust deposition from dissolved Al, the Model of Aluminum for Dust Calculation in Oceanic Waters (MADCOW) model assumes that between 1.5 and 5% of mineral dust aerosols dissolve in the surface ocean. This range is derived from several published laboratory studies of partial solubilities [Maring and Duce, 1987; Prospero et al., 1987; Chester et al., 1993; Lim and Jickells, 1990] which are also close to the average solubility of Al in Asian aerosol (4.6%) sampled in Hawaii and reported by Sato . Thus the absolute values of our calculated deposition vary by a factor of ∼3 depending on the fractional solubility of the mineral aerosol we assume. In the data below, we generally will use the term “mean dust deposition,” which uses an assumed solubility of 3.3%, the mid point of our expected range. C. S. Buck et al. (The solubility and deposition of aerosol Fe and other trace elements in the North Atlantic Ocean: Data from the A16N CLIVAR-CO2 Repeat Hydrography Section, manuscript in preparation, 2008) have made DI water and seawater aerosol solubility measurements along the A16N track that are consistent with the values used for the MADCOW calculations.
3.2. Surface Waters
 The dissolved Al in the mixed layer samples along with the estimated dust depositions for 1.5% and 5% solubility (high and low deposition, respectively) are shown in Figure 2. Additionally, in Figure 2 we superimpose deposition estimates interpolated from the Group of Experts on Scientific Aspects of Marine Environmental Protection (GESAMP) model output [Duce et al., 1991] at selected latitudes along our cruise track. It should be noted that the large range of the GESAMP estimates at these latitudes in Figure 2 reflects the spacing between adjacent contour intervals which are 1 order of magnitude apart. This spacing does not represent any inherent inaccuracy in that data set but merely reflects the factor of 2–3 uncertainty in the GESAMP estimates, which is similar to our factor of ∼3 uncertainty due to atmospheric mineral solubility and other assumptions. In addition, because of the multiyear residence time of Al in surface waters, our dust deposition estimates are not instantaneous estimates, but instead represent a 5-year running average; that is, any seasonal effects at the time of our cruise should be muted significantly.
 Along the A16N cruise track the dissolved Al ranges widely, from ∼2–37 nM, implying mean dust depositions (assuming 3.3% solubility) that vary by more than 1 order of magnitude, from ≤0.2–∼3.0 g mineral dust m−2 a−1 (hereinafter referred to simply as g m−2 a−1). The lowest mean dust deposition, ∼0.2 g m−2 a−1, is found in the northern part of the section between 48° and 60°N. Estimates rise rapidly to the south reaching a maximum in mean dust deposition (∼2 g m−2 a−1) at 30°N. Values then decline to ∼1 g m−2 a−1 by 20°N. Estimates then increase once again reaching a maximum mean deposition of ∼3 g m−2 a−1 at 8°N. To the south of 8°N values decline to ca. 1.5 g m−2 a−1 at 5°S. This N-S trend, with the exception of the maximum at 30°N (discussed below), is consistent with the estimates from the Duce et al.  GESAMP model, which were based on extrapolation of data gathered at land-based sites in the North Atlantic. Comparison of the absolute values of our estimates with Duce et al.'s  estimates, indicates that our values agree with those estimates best when lower dust solubilities, yielding higher deposition, are assumed. Nevertheless, the agreement between the sets of estimates, both in trend and absolute value, are within the factor of three uncertainty of each of these approaches. While satellite imagery of suspended dust load will not necessarily mirror dust deposition, it is interesting to note that our maximum deposition region at 8°N is consistent with the average location of the large dust plume emanating from the Sahara. This plume oscillates between the equator and 20°N as it follows the migration of the Intertropical Convergence Zone during the period of maximum dust transport, which is between December and August [Husar et al., 1997].
 The maximum in deposition that we see at 30°N, 23°W though, is not apparent in either the GESAMP or the 3-month averaged satellite imagery of Husar et al. . Our sampling resolution across this feature (14 stations) and the consistency of the trends we see within it, discounts the possibility that this is a sampling or other kind of artifact. We note that our observed maximum lies approximately 370 nmiles to the west and north of the Izana, Canary Islands sampling site at 28°N and approximately 780 nmiles to the north of the Sal Island sampling site in the Cape Verde Islands at ∼17°N, the two eastern Atlantic sampling sites that were used in the GESAMP model at these latitudes. Thus it is possible to argue that the large physical separation of these two GESAMP sites might preclude observation of this feature. We also note that of all the data comparisons between GESAMP and our data set, the estimates at the latitude of Izana show the poorest agreement.
 Another, and more interesting, explanation is that the source of the mineral dust that produces this enhanced deposition we see at 30°N is propagated from the west, rather than the east, in which case the material we see entering the surface waters at this location (30°N, 23.4°W) from the marine boundary layer would not have traversed, nor be recorded at, the Izana sampling site (28°N, 16.5°W, 2360 m above sea level [Prospero, 1996]). We will attempt to provide evidence for this argument when we discuss the sources of Al to the North Atlantic Sub tropical mode water formation regions, below.
3.3. Enhanced Al in the Subsurface Waters
 The elevated Al (>20 nM) seen in our deepest samples (∼750 m) at 37/36°N, is coincident with the upper part of the Mediterranean outflow water, the presence of which is also reflected in the salinity maximum seen in Figure 1, bottom. The Al labeling of this water results from the fact that the Mediterranean basin receives upward of 10 g mineral dust m−2 a−1 resulting in highly elevated dissolved Al in the subducted surface waters that exit the Strait of Gibraltar [Hydes, 1983; Measures and Edmond, 1988]. While the distribution of the enhanced salinity signal from the Mediterranean outflow is a well documented feature of the tropical North Atlantic, our section shows that the geochemical imprint that accompanies this signal is also unmistakable, influencing Al concentrations at this depth from ∼5° to 55°N [Hall and Measures, 1998].
 The third region of elevated Al concentrations, is centered at ∼300 m, between 35 and 20°N with Al values generally >18 nM. Again, as with the Mediterranean signal, the ultimate origin of the enrichment is from the subduction of surface waters that have been enriched in Al by atmospheric dust deposition. In this case the water mass is the Subtropical “mode” water (STMW) of the North Atlantic which is found at a potential density of approximately 26.5 kg m−3. Two forms of STMW have been reported for the Atlantic [Hanawa and Talley, 2001], the classical 18° water which forms in the western basin on the boundary of the Gulf Stream [Worthington, 1959] and the Madeira mode water which forms in the eastern basin to the North and West of Madeira [Siedler et al., 1987]. Both forms of STMW are believed to form in late winter/early spring when winter cooling increases surface water densities and the mixed layers are at their deepest. The CLIVAR cruise track passed through the Madeira sub tropical mode water formation region in early July and we observed surface water Al concentrations of 10–17.5 nM between 32° and 36°N, which are considerably below the values seen in the core of the STMW where maximum values of ∼24 nM were seen at depth in our section. However, although our transit was not during the STMW formation period, the ∼5-year residence time of dissolved Al in surface waters implies seasonal variations should be ∼20%. In contrast, in the western basin of the North Atlantic, between 31° and 36°N, surface water Al values ranging from 28 to 43 nM have been reported [Measures et al., 1984, 1986; Jickells et al., 1994], more than sufficient to supply the observed Al concentrations in the STMW. The concentrations that we see in the eastern Atlantic are consistent with previous reports for dissolved Al in the surface waters of this region [Hydes, 1983; Kramer et al., 2004]. A contour plot showing these data sets and the east-west gradient across the Atlantic at this latitude is presented in Figure 3. Thus the dissolved Al concentrations that we observed in surface waters of the eastern basin, when combined with reported values for the western basin surface waters, suggests that the main volumetric contribution to the STMW of the Atlantic is from the western basin, i.e., the classic 18° variety. This is also consistent with the conclusion of Siedler et al. .
 The apparent paradox of greater Al concentrations in the surface waters of the western Atlantic at these latitudes (30–36°N) when the dust source providing the Al is on the eastern side of the Atlantic at more southerly latitudes (0–20°N), brings us back to the question of the origin and systematics of the Al concentration maximum that we observe in surface waters at 30°N shown in Figure 2.
 Our surface water data above shows clearly that there is significant deposition of mineral dust from the Saharan plume as it exits the African coast between 0 and 20°N into the adjacent surface ocean. However, much larger amounts of material are transported rapidly at elevations that are well above the marine boundary layer across the Atlantic [Prospero, 1996]. In summer months the strong circulation around the Bermuda-Azores high coupled with the subsiding air masses can bring this high-altitude material into the marine boundary layer from which it can more easily sediment out or be removed by wet deposition, and send it back across the Atlantic with an eastward trajectory. Thus the predominant westerlies at this latitude are seeded in their marine boundary layer with dust on the western side of the basin and appear as a point source that is progressively depleted through deposition along the advective track of the air mass, imprinting the surface ocean beneath them with a signal that is also progressively depleted from west to east.
 While this is a speculative argument, it should be noted that the Bermuda dust record shows elevated dust deposition during summer when the high-pressure system is stabilized, and facilitating downward transport of high-altitude Saharan material [Prospero, 1996]. Additionally, Sedwick et al.  have reported increases in surface water dissolved Fe in Sargasso Sea surface waters associated with enhanced summertime atmospheric deposition. This argument is also consistent with the plot of the distribution of dissolved Al in surface waters over this latitude range (Figure 3) which shows much higher values in the western basin than the east. Finally, the speculation is also consistent with the GESAMP model which also shows higher deposition estimates in the western basin of the Atlantic, than in the eastern basin between ∼30–45°N.
 It may also be coincidental, but enhanced deposition of atmospheric mineral dust has also been seen in the North Pacific at a similar latitude, 30°N, 150°W; this depositional feature is also believed to be the result of the transfer of high-altitude dust from the Gobi Desert into the MBL by the high-pressure zones of this region [Hiscock et al., 2006].
 If, as we suggest in the case of the Atlantic, the STMW are labeled with Al derived from atmospheric deposition, then there exists the potential to develop a paleotracer based on Al (or another atmospherically delivered trace element) that might record the intensity of mode water formation, and/or dust transport to the STMW formation region of the North Atlantic.
 Overall, it is evident from the Al distributions that a large part of the upper 1000 m of this part of the Atlantic Ocean is clearly imprinted by the process of atmospheric deposition, either directly under the Saharan plume or indirectly through the deeper water masses that are also labeled by atmospheric deposition processes in their formation regions. It is also important to note that below the Al-enriched surface waters underlying the Saharan dust plume there is no evidence of subsurface remineralization of surface water scavenged Al in contrast to the distribution for Fe (discussed below).
 While our data show the atmospheric imprinting through the distribution of dissolved Al, a natural component of continental materials, it is also likely that anthropogenic materials with continental sources and atmospheric transport vectors, are similarly imprinting the upper waters of this ocean basin as has been shown in the Sargasso for Pb [Boyle et al., 1986; Shen and Boyle, 1988; Véron et al., 1993; Wu and Boyle, 1997]. This suggests that the upper water of the North Atlantic is one of the most susceptible regions of the global oceans to atmospheric input and that it should be a sensitive recorder of both historical and future changes in atmospheric inputs.
3.4. Fe Distributions
 Surface water Fe distributions to a large degree follow the surface water Al concentrations, reflecting the importance of the role of atmospheric deposition in supplying Fe to the surface waters of much of the North Atlantic (Figure 4). The correlation is strongest (R = 0.82) between 51.5 and 9°N, roughly coinciding with the boundaries of the N Atlantic gyre. Within the gyre between 25 and 32°N, the dissolved Fe concentrations we observe are slightly higher than the October surface water values reported by Sarthou et al.  further to the east. The range and pattern of our dissolved Fe values are similar to the winter/spring concentrations reported by Bergquist and Boyle  between 30°N and 5°S to the west. To the north of the gyre, deep winter mixing (mixed layers >300 m) supplies the majority of the Fe to the surface waters from the subsurface layers that are relatively enriched in Fe through biological remineralization. In contrast, Al concentrations in both the surface waters and subsurface waters are low because of low dust deposition and also because of lack of remineralization, thus decoupling the Al:Fe correlation.
 To the south of 9°N, the degradation of the correlation is most likely a result of the North Equatorial Countercurrent (NECC), which is visible in the shipboard acoustic Doppler current profiler (E. Firing and J. Hummon, unpublished data, 2003). Between July and December, the NECC which has its origins in the seasonal retroflection of the North Brazil Current, advects surface waters across the Atlantic from the west [Richardson and Walsh, 1986; Wilson et al., 1994; Tsuchiya et al., 1992]. We speculate that the entrainment of nutrients from the shelf and the Amazon outflow stimulates productivity in this water resulting in the preferential removal of Fe.
3.5. Fe Systematics in the Subarctic Gyre
 The observation that in general the surface water Fe concentrations north of ∼51°N were extremely low (0.02–0.16 nM; average 0.09 nM) and that there were significant quantities of unutilized nutrients in the surface waters (up to 5 μM nitrate), naturally poses the question of whether this region might be Fe-limited. This combination of nutrient properties could of course simply be a result of the timing of our cruise; that is, the spring bloom had not yet reached its climax, removing all available macronutrients. However, inspection of the World Ocean Circulation Experiment database, even when restricting data sets to two summer months (June and July), also indicates a very clear demarcation in surface waters with remnant macronutrients (NO3 ∼ 2–6 μM) extant in the surface waters of north of ∼50°N. Additionally, the Joint Global Ocean Flux Study (JGOFS), May/June 1989 North Atlantic Bloom Experiment (NABE), also reported surface water nitrate concentrations of 2–9 μM between 50 and 60°N [Garside and Garside, 1993]. Thus there appears to be abundant evidence that small but significant levels of macronutrients persist in the surface waters north of ∼50°N during the summer months, and that this is coincident with our observations of this being a region of low dust input.
Martin et al.  concluded from their JGOFS NABE Fe and chlorophyll data that the growth capacity in this northern part of their sampling region was much greater than was observed in the subarctic and equatorial Pacific and thus did not appear to be similarly Fe-limited as those regions were. We agree, this is not a classic HNLC region since chlorophyll is not particularly low, and remnant nutrients are not particularly high. However, this does not preclude that Fe-limitation exists, but merely the degree to which the surface waters may be Fe-limited. To examine this concept in more detail, we wish to revisit the Fe sufficiency argument at this latitude from the standpoint of macronutrient availability and Fe supply from deep winter mixing and mineral deposition. We estimate that late winter deep mixing to ∼500 m would raise surface water Fe values in this region to a maximum of ∼0.5 nM. Additionally, an annual mineral dust deposition of 0.2 g m−2 a−1 containing 800 μmole Fe g−1 of which 3% dissolves, would provide an additional 4.8 μmoles Fe m−2 a−1 to the surface waters. Over a 50 m mixed layer (the value in summer during plankton growth) this would be equivalent to an additional 0.1 nM Fe/L, yielding a maximum of 0.6 nM. Using a N:Fe ratio of 15,000:1 for biological uptake under Fe limiting conditions (derived from a C:Fe of 1 × 105 and C:N = 6.67 [Measures and Vink, 1999]), would imply that ∼9 μM nitrate could be removed, leaving a residual of ∼6 μM nitrate from the ∼15 μM surface water winter nitrate values. While these estimates could be varied significantly by choosing different aerosol solubilities and N:Fe uptake ratios etc., the published values we have used are reasonable, and yield results that are remarkably close to the observed residual surface water nitrate values. Thus this calculation serves as test of concept, rather than a definite proof of occurrence Alternatively, using these same values, we can calculate that full utilization of the winter nitrate levels would require ∼0.4 nM additional Fe, which could be supplied by the deposition and partial dissolution of a further ∼1.0 g m−2 a−1 of mineral dust to the surface waters of the region, approximately 5 times that of our mean deposition estimate.
 Thus it would appear from these calculations, and the persistence of macronutrients in surface waters, that the region north of 50°N is on the borderline of Fe limitation. If so, then this would be a particularly interesting region for repeat studies over longer periods of time, since its Fe sufficiency may vary significantly from year to year as a result of natural, or climatically induced changes in dust flux. Thus we suggest that this would be a valuable, and logistically feasible region in which to study the systematic changes in oceanic biology and chemistry that accompany an oceanic region as it transitions between Fe sufficiency and Fe limitation.
3.6. Subsurface Fe Maximum
 The much shorter surface residence time and remineralization of Fe leads to significantly different subsurface distributions of this element from those of Al. Thus the Mediterranean and STMW do not show visible Fe maxima, above the background concentrations that develop from biological vertical transport and remineralization.
 Instead, the most dramatic feature of the subsurface Fe distribution is the large Fe maximum between ∼18 and 4°N at depths between 200 and 800 m, where Fe values reach up to 2 nM. The latitude range of the subsurface Fe maximum corresponds closely with the surface water Al maximum indicating atmospheric deposition of mineral dust is possibly fueling this process. Dissolved Fe in surface waters in this region also reach values of up to 1.5 nM. These enriched surface waters are separated from the subsurface maximum by a minimum at ∼50 m, which becomes more pronounced toward 12°N. The dissolved Fe minimum at 50 m corresponds very closely to the chlorophyll maximum, recorded by the uncalibrated fluorometer that was mounted on the TM rosette (Figure 5). Thus we can see over a significant latitude range the direct partitioning of dissolved Fe into a living biological particulate phase. The depletion between the surface mixed layer and the dissolved Fe minimum is typically 0.4 to 0.6 nM, significantly higher than the ∼0.2 nM depletion observed by Bruland et al.  in the lower-euphotic zone of the stratified waters of the North Pacific gyre or the 0.2 to 0.3 nM depletion reported by Bergquist et al.  at 10°N in the western Atlantic. While these differences may reflect greater production in the eastern versus the western North Atlantic region or the stratified North Pacific gyre, the biological implications of this are beyond the scope of this paper, or the data sets available for interpretation of the biological signal. Clearly the ability to close budgets on calculations like this would benefit greatly from the availability of a suite of basic biological parameters.
 The latitude range of the subsurface Fe maximum (18 to 4°N) corresponds very closely to the oxygen minimum and is most pronounced where O2 values are less than 80 μM, as shown in Figure 1, middle. This implies that biological remineralization of surface derived Fe supplied by the Saharan mineral dust plume plays an important role in sustaining this feature. Alternatively, the oxygen minimum could be an advected feature originating where overlying seawater contacts coastal reducing sediments. However, we discount this latter possibility as there is little dissolved Mn, which also emanates from reducing sediments, associated with this low-oxygen feature [Landing et al., 2006] In addition, south of the equator, where dust inputs and surface Fe concentrations are much lower, an almost equally intense low-oxygen zone shows little or no subsurface Fe enrichment, reinforcing the notion that the subsurface Fe enrichment north of the equator is predominantly supported by the vertical transport and remineralization of the local surface Fe enrichment. The match between the enhanced surface and subsurface Fe concentrations starts to break down around ∼7°N which is most likely a result of the NECC bringing relatively low Fe waters from the east at this latitude during the time of our cruise. Presumably elevated Fe concentrations in the poorly ventilated subsurface waters integrate over a much longer period of time than those in the relatively rapidly moving, and seasonally flow reversing, surface waters.
 If the Fe maximum is sustained by in situ remineralization, then it is instructive to look at the ratio of N:Fe within the oxygen minimum to develop some insights into the water column Fe remineralization process. Under Fe limiting conditions, the C:Fe ratio in phytoplankton is ∼105:1 [Sunda and Huntsman, 1995], which, using a typical C:N ratio of 6.7:1, translates to a N:Fe ratio of 15 × 103. Thus one would expect that the subsurface remineralization of biological material delivered from Fe-rich surface waters to have a ratio of N:Fe ≪ 15 × 103. However, within the boundaries of the oxygen minimum, the N:Fe ratio varies from ∼15–30 × 103, at or well above the “limitation” ratio. This surprisingly high value suggests that during the biological remineralization process there is a large relative loss of Fe compared to N, presumably through active scavenging of this particle reactive element. Bergquist et al.  also observed elevated total dissolved Fe values of >1.0 nM between 200 and 1000 m, associated with the oxygen minimum, at their 10°N, 45°W station. Through careful speciation determinations they were also able to attribute the bulk of that Fe at these depths as being present in the form of colloids, a form of Fe that is probably more prone to scavenging than organically complexed forms of Fe. The elevated subsurface concentrations that we see are clearly much higher than the steady state maximum value of 0.6 nM Fe postulated by Johnson et al.  for deep waters and discussed by Boyle . While our single survey of this region prevents us from addressing the seasonal variation directly, it would seem likely that in this poorly ventilated region the elevated Fe plume is a steady state feature where biological remineralization processes are balanced by scavenging of released Fe. If a large fraction of the remineralized Fe is present in a colloidal form as suggested by Bergquist et al.  then the relatively low concentration of Fe observed relative to N may be a result of continual scavenging by vertically transported of eolian dust.
 Previous observers have noted that in regions with low surface water Fe, the subsurface waters were deficient in Fe relative to N. For example, Martin and Fitzwater  reported N:Fe ratios in upwelling water of the HNLC region of the North Pacific of ∼57,000:1, and reported an even more limiting ratio of >100,000:1 in the Drake Passage region of the Southern Ocean [Martin et al., 1990b]. To observe this same phenomenon in the water column that underlies a region of relatively high surface water Fe, indicates that there must be strong geochemical fractionation of these two elements during remineralization, and that this ubiquitous loss of Fe develops the relative deficiency of Fe to N in subsurface waters. It should be noted that Measures and Vink  also found a relatively high N:Fe ratio of 23,000:1in the waters upwelling in the Arabian Sea, a region that also receives significant dust deposition 2.2–7.4 g m−2 a−1.
 Thus all subsurface waters, regardless of the source of biological material that fuels nutrient remineralization within them, will be deficient in Fe to some degree. The upwelling or deep mixing processes that returns the macro nutrients to surface waters always requires a complementary process that delivers an additional source of Fe to the surface waters. In the open ocean, far from continental shelves, the atmospheric transport of continental materials, its deposition and partial dissolution in surface waters, is usually the only available vector.
 The high-resolution sections of dissolved Al and Fe in the Atlantic Ocean along the CLIVAR A16N section clearly show the role that atmospheric deposition of continental mineral dust plays in the geochemical cycle of these two elements. The effect is manifested in surface waters by high concentrations of both elements in regions of high deposition. Subsurface waters are also enriched with Al by subduction of surface labeled waters in the Mediterranean and in the western basin where the classical 18° subtropical mode water forms. Together these water masses transmit the atmospheric signal into a large part of the upper 1000 m between 60°N and 5°S. This widespread distribution indicates the close chemical connection between atmospheric and oceanic processes in the North Atlantic. That this relationship is visible in a tracer with a relatively short residence time indicates the potential rate at which anthropogenic materials with eolian transport vectors can directly affect a large proportion of the upper waters of this basin. This clear labeling of the mode water by Al might indicate that this tracer, or another with an atmospheric flux, might be developed into a tracer of the history of atmospheric deposition and mode water formation.
 In addition, a large region of enhanced Fe concentrations between 18° and 4°N is maintained by biological uptake in enriched surface waters, vertical transport and remineralization in the oxygen minimum zone. Despite the copious quantities of Fe supplied to the surface waters of this region, the remineralized ratio of Fe:N appears to be below the Fe-limiting uptake ratio for photosynthetic organisms.
 Also significant is that to the north of 51°N, where atmospheric deposition is dramatically lower, the surface waters appear to be on the borderline of Fe-limitation, possibly explaining why this region has unused macronutrients in surface waters and suggesting the presence of a previously unrecognized “HNLC” region, in the sense of unutilized surface layer macronutrients after the phytoplankton growth season.
 This work was supported by NSF grants OCE 0223397 to CIM and OCE 223378 to WML. We thank John Bullister and Nicky Gruber for the many kindnesses they showed in helping us to integrate our new program into their larger shipboard commitments. We also thank the Captain and the crew of the RV Brown for their help with the deployment and retrieval of our package, and particularly Mike Gowan, Chief Engineer, for his help with our troublesome winch. We also thank all the members of the CLIVAR-CO2 Repeat Hydrography oversight committee, cochaired by Dick Feely and Lynne Talley, for their enthusiastic support in adding the TM component to the CLIVAR Repeat Hydrography and providing us with the ship time required for our work. This is contribution 7251 of the School of Ocean Earth Science and Technology, University of Hawaii.