Dissolved Al was determined on ∼3500 surface water samples collected in the NW Pacific during the 2002 Intergovernmental Oceanographic Commission (IOC) Contaminant Baseline survey. In addition, dissolved Al was determined on samples collected at 9 vertical stations occupied along the cruise track. Surface water Al distributions, when converted to annual mineral dust deposition (Measures and Brown, 1996), imply extremely low depositions (mean values ≤0.3 g m−2 yr−1) to surface waters of the subarctic gyre despite the atmospheric transport of large amounts of mineral dust across this region from the desert regions of China. The calculated deposition values are also much lower than the 1–10 g m−2 yr−1 predicted for this region by the GESAMP model (Duce et al., 1991). The lowest mineral dust depositions correspond to the High Nutrient Low Chlorophyll (HNLC) region of the NW Pacific, which implies that the biogeochemical status of this region is the result of low atmospheric micronutrient input, similar to other HNLC regions. In the subtropical gyre the Al-derived dust deposition (mean values 0.1–0.6 g m−2 yr−1) agrees well with the (<1 g m−2 yr−1) predictions of the GESAMP model. Enriched concentrations of dissolved Al in the coastal waters near Kauai, from fluvial sources, indicate the potential role of remote high islands in adding reactive trace elements to the nearby surface waters. In the vertical profiles, elevated dissolved Al concentrations (4–8 nM) were found to correspond to the potential density of the western Pacific “Subtropical Mode” waters which form in the recirculation zone of the Kuroshio.
 A fundamental goal of contemporary Earth Science studies is to develop a sufficiently detailed understanding of environmental feedback mechanisms, so that they can be modeled mathematically and used in a predictive mode to describe likely planetary responses to changes in biogeochemical forcing parameters, e.g., rising atmospheric carbon dioxide levels. Under this broad umbrella, a recent area on which significant attention has been focused is the role that the abundance of Fe in surface ocean waters plays in moderating the biological processes. Interest in this subject arises from the speculation that insufficient dissolved Fe in oceanic surface waters may lead to limitation of biological production in high latitude and equatorial High Nutrient, Low Chlorophyll (HNLC) regions [Martin and Fitzwater, 1988; Martin, 1990; Martin et al., 1990a, 1990b]. This hypothesis has been confirmed dramatically by the observation of enhanced biological growth as a result of deliberate Fe addition to a variety of high and low latitude HNLC regions [Martin et al., 1994; Boyd et al., 2000; Coale et al., 2004]. The importance of the biological carbon pump in transferring atmospheric carbon dioxide to deep ocean water and the apparently important role of the micronutrient Fe in moderating that process highlights the need to develop a systematic understanding of the geochemical cycle of Fe and the mechanisms that determine its rate of supply to the surface waters of the ocean.
 Speculation that the deposition and partial dissolution of atmospheric dust likely plays a crucial role in Fe supply to the surface ocean has come from the fact that upwelled waters contain biologically insufficient Fe relative to macronutrients, and that in the open ocean, remote from continental shelves, the atmosphere is the only viable source of this required Fe. Additional support for this idea comes from the observation that HNLC waters appear to coincide with regions that are believed to receive low atmospheric dust fluxes. Another impetus for investigating the relationship between atmospheric dust deposition and biological processes comes from the observation that enhanced dust fluxes recorded during glacial periods [De Angelis et al., 1987; Petit et al., 1999; Kumar et al., 1995; Laj et al., 1997] correspond to periods of lowered atmospheric carbon dioxide levels. A link between enhanced dust (and thus Fe) deposition to the surface ocean and lowered atmospheric carbon dioxide levels may exist, and could provide a mechanism by which small changes in planetary insolation could lead to large glacial-interglacial atmospheric carbon dioxide transitions [Martin, 1990].
 As a result of the forgoing linkages, there is clearly a need to incorporate the geochemical effects of dust deposition into global climate models, since the magnitude of Fe delivery to the surface ocean via this route could affect considerably the rate at which atmospheric carbon dioxide levels are equilibrated with the ocean via biological processes [Falkowski et al., 1998]. Thus a key requirement for global modeling to progress is the ability to produce accurate estimates of Fe deposition to the surface ocean via atmospheric routes. To do this requires a modeling system that can accurately predict the amount of atmospheric mineral deposition to the surface ocean and a knowledge of the factors that control the release of biologically-available Fe from this material to the waters of the euphotic zone.
 Currently it is almost impossible to obtain atmospheric deposition estimates in the open ocean through direct sampling [Guerzoni et al., 1999]. The alternative, converting remotely sensed parameters from satellites into a deposition flux, is still in its infancy and suffers from various interferences such as absorbance from nondust atmospheric material. It is expected that advances in sensor design and interpretation will eliminate many of these problems; nevertheless, estimates from such sources will need significant ground-truthing before they can be used reliably in coupled models [Measures and Vink, 2000]. In addition, the biogeochemical systematics that relate dust deposition to the release of biologically available Fe to surface waters are still poorly understood and are likely to be beyond sensor technology for the foreseeable future.
 This paper reports a further application of that model to data from a region of the NW Pacific where several dust deposition models [e.g., Duce et al., 1991; Fung et al., 2000] predict high levels of dust deposition, but the presence of HNLC-like conditions implies low surface water Fe availability. If indeed this is a region of high dust deposition and high Fe availability, it would challenge our assumptions about the causes of HNLC regions. Conversely, if this were a region of high dust deposition but low Fe availability, it would challenge our understanding of the mechanism by which atmospheric dust deposition adds Fe to oceanic surface waters. This paper presents the results of the Al distributions obtained during an oceanographic cruise to that region; a companion paper by Brown et al.  presents the accompanying Fe distributions.
2. Data and Methods
 Water samples were collected at nine vertical profile stations between 1 May and 5 June from the RV Melville during the 2002 IOC western Pacific cruise using procedures described by C. I. Measures et al. (Hydrographic observations during the 2002 IOC Contaminant Baseline Survey in the western Pacific Ocean, submitted to Geochemistry, Geophysics, Geosystems, 2005) (hereinafter referred to as Measures et al., submitted manuscript, 2005). In addition, surface water samples were obtained semi-continuously while the ship was underway between stations using the towed “fish” system described by Vink et al. . This system, which allows continuous peristaltic pumping of surface water from ∼0.5 m depth, resulted in a total of 3549 surface water determinations of dissolved Al during the cruise. The surface water was filtered in-line using a Pall Gelman 0.2 μm Suporcap Capsule filter and the sample stream was acidified with 0.15 N subboiled HCl to match the pH of the standards prior to its introduction into the onboard FIA system [Resing and Measures, 1994]. Blanks and standards were run approximately every 4 hours. Precision was assessed by multiple FIA determination of the same sample and was approximately 1.6% at 7 nM. The methods employed were essentially the same as those reported by Vink and Measures .
3. Results and Discussion
3.1. Surface Waters
 The surface water results are presented as a contour plot indicating the geographical relationship of dissolved Al concentrations (Figure 1). The data were contoured using Generic Mapping Tools [Wessel and Smith, 1995]. Contours were computed after fitting a minimum curvature surface with tension to the gridded data. Setting maximum tension in the surface grid constrains the solution so that local maximum or minimum values occur only at points where there are data [Smith and Wessel, 1990].
 The data show a fairly narrow range of concentrations (excluding the nearshore data from Kauai, Hawaii; see below), with values ranging from <1 to ∼8 nM Al. The highest values (4–8 nM) were found in the subtropical gyre, south of the Kuroshio and the Kuroshio extension. The lowest values were found in the subarctic gyre, where values ranged from <0.1 nM to 2.2 nM. In addition to the variation in surface water concentrations, there was also a gradient in the variability of Al concentrations with much higher variability in the surface waters of the subtropical gyre than in the surface waters of the subarctic gyre (Figure 2).
 During the northbound transit between 142 and 153°E, the mixed water region showed a dramatic decrease in Al from ∼5 nM at ∼35°30′N, to <2 nM at 36°30′N (Figure 3), coincident with the northward propagating front of the spring bloom (Measures et al., submitted manuscript, 2005). During the southbound transit along 171°E, the increase in surface water Al was much less pronounced, with values rising to ∼2 nM at 36°N. Presumably this difference is a result of both the later timing of the southbound transit (i.e., after the passage of the spring bloom which will scavenge some of the dissolved Al), and also the fact that this transit is further offshore, i.e., further downstream of the Asian dust plume. Clearly, with one sampling it is not possible to disentangle the relative importance of these two complementary factors.
 Within the subtropical gyre, along the western part of the section, the highest Al values (6–8 nM) were found immediately south of the Kuroshio extension, between 32°35′N and 32°50′N, along 145°40′E (Figure 2). This recirculation region is associated with westward recirculation of some of the Kuroshio extension waters [Huang and Qui, 1994], thus probably leading to a relative enrichment of Al during the longer residence time of the surface waters in this region of relatively higher dust deposition.
 In the eastern part of the subtropical gyre (Stns 5–9), surface water Al values increase toward the south, reaching ∼4–5 nM as the ship approached its most southerly point (Station 7) at 24°15′N. As the ship headed eastward toward Station 8, the Al values decreased to ∼3 nM by 175°W. As the ship's track reached and ran down the axis of the NW Hawaiian Island chain, average Al values climbed slowly reaching ∼5 nM at 160°W. At this point, Al values then increase dramatically, reaching ∼50 nM as the ship transited close to the Na Pali (north) coast of Kauai (Figure 4). This coastal runoff effect rapidly diminished to ∼4 nM as the ship's track headed offshore toward Station 9. The final part of the cruise, NE of the Hawaiian Island chain, intersected a limited region of elevated Al values (6–8 nM) between 155 and 153°W, 24°N. This region was resampled on the return transect, and the same result was found. In contrast, to the NE of this region, Al values dropped back to 2–4 nM. The location of elevated Al concentrations was close to, but NW of, the region of high Al observed by Johnson et al.  in May 2001, indicating that this region might be a preferential site of eolian deposition. Recent unpublished data (Measures, Brown, Landing, and Buck, personal communication, 2004) also show a region of elevated Al values between ∼155° and 140°W along 30°N.
3.2. Modeled Dust Deposition
 The Al concentrations presented here (with the exception of the region affected by coastal runoff around Kauai, discussed below), can be transformed into a dust deposition estimate using the MADCOW model [Measures and Brown, 1996; Measures and Vink, 2000]. The model, which assumes that dissolved Al in oceanic surface waters results from the partial dissolution of atmospheric dust, requires assumptions about mixed layer depth, residence time of dissolved Al and the fraction of the aerosol that dissolves. The mixed layer depth values used for the calculation, and the effect of different assumptions about aerosol solubility on the calculated deposition are shown in Table 1. In the case of mixed layer depths, we have used the values observed at our vertical profile stations in each of the representative regions.
Table 1. Regional Dust Deposition Estimates Along the Cruise Track
Range is derived from solubility estimates of 1.5 or 5%.
approaching Stn 2
Stn 2 to Stn 3
approaching Stn 4
Stn 4 to Stn 5
Stn 5 to Stn 6
Stn 6 to Stn 7
Stn 7 to Stn 8
Stn 8 to Na Pali coast
Na Pali to Stn 9
elevated deposition region
elevated deposition region
 The residence time of Al in surface waters has been assumed to be 5 years [Orians and Bruland, 1986; Jickells et al., 1994; Measures and Brown, 1996] and invariant across the region. Clearly the residence time will vary in some manner with export production, but no attempt is made here to elaborate the nature of that relationship. At this time, there are no values for the partial solubility of the Al from the dust in this region and thus the range of 1.5 to 5.5%, with a mean value of 3.3%, has been applied to the data set. This range and mean value, which are 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] are also close to the average solubility of Al in Asian aerosol (4.6%) sampled in Hawaii and reported by Sato .
 The dust deposition equation can be reduced to
where G is the apparent dust input in g m−2 yr−1, A is the mixed layer concentration of dissolved Al in nanomoles/L, and ML is the mixed layer depth in meters.
 The results of applying this model to the data set are shown in Figure 5 as a contour plot of the deposition of mineral dust calculated using a mean dust solubility of 3.3%; the ranges and means for geographic areas are presented in Table 1. While there are considerable uncertainties in the values of mixed layer depth, solubility, etc., that are used to calculate these values, there is a clear disparity between the MADCOW regionally derived aerosol depositions and those from the GESAMP model used by Duce et al.  and plotted as an overlay in Figure 5. It should be stressed however, that even though the cruise represents an extremely short sampling period (∼1 month), the residence time of Al in surface waters (5 years) means that the calculated dust input should represent something like a 5-year running average.
 In the subtropical gyre near Japan, mean dust deposition estimates (3.3% solubility) range from 0.35 to 1.17 g m−2 yr−1 with an average of 0.69 g m−2 yr−1. The individual solubilities give ranges that are 0.16 to 0.54 g m−2 yr−1 (assuming 5% solubility) and 0.53 to 1.8 g m−2 yr−1 (assuming 1.5% solubility). All of these values are well below the >10 g mineral dust m−2 yr−1 predicted for this region. Similarly, in the subarctic gyre, the mean dust deposition ranges from 0.015 to 0.6 g mineral dust m−2 yr−1, with ranges of <0.01 to 0.3 g mineral dust m−2 yr−1 for 5% solubility and 0.02 to 0.9 g mineral dust m−2 yr−1 for 1.5% solubility. As before, these values are well below the projected deposition in this region by the Duce et al.  GESAMP model, which predicts depositions in the range of 1–10 g m−2 yr−1.
 In the eastern part of the transect through the subtropical gyre, mean dust deposition estimates (up to, but excluding, the vicinity of Kauai; see below) range from 0.10 to 0.50 g mineral dust m−2 yr−1, which are similar to those predicted by the Duce et al.  model. However, within this region there are distinct trends. For example, the mean dust deposition drops slowly from ∼0.4 g mineral dust m−2 yr−1 at 173°E to less than 0.3 g m−2 yr−1 at 175°W. Further to the east, although data are more variable, deposition values then start to rise again reaching 0.4 g mineral dust m−2 yr−1 at 161°W (Figure 6).
 Applying the model in the vicinity of the Hawaiian Islands would lead to computed dust fluxes of >1 g mineral dust m−2 yr−1 (see below). However, as discussed below, the increase in Al concentrations that underlie this effect are more likely driven by fluvial runoff from the islands than solely by dust deposition. The “island effect” disappeared rapidly as the ship's track veered offshore north of the islands toward Station 9 (ALOHA), where numbers returned to ∼0.3 g dust m−2 yr−1, close to the 0.44 g dust m−2 yr−1 measured by Uematsu et al.  on Oahu and the 0.11–0.62 g mineral dust m−2 yr−1 estimated by Prospero et al.  for the central and Eastern North Pacific. To the NE of Station 9, the ship's track encountered a region of elevated Al between 155 and 153°W, where calculated depositions doubled to ∼0.7 g mineral dust m−2 yr−1, before falling back to 0.2–0.3 g mineral dust m−2 yr−1, further to the NE corner of the cruise track.
 It is clearly possible that some of the difference between the deposition calculated from the GESAMP and the MADCOW models in the western part of our cruise track could be accounted for by invalid estimates of fractional solubility of the mineral aerosol and/or surface water residence time of Al used in MADCOW. However, these are unlikely to be the most important source of the disagreement in the western Pacific since our values agree with the GESAMP predictions quite well in the central Pacific. Since the model is linear, estimations of deposition one order of magnitude greater than those reported here would require a commensurate one order of magnitude reduction in the combination of surface water residence time and solubility used in MADCOW. We consider an error of this size to be quite unlikely since this would have to affect both the subtropical and subarctic gyres but only in the cruise track region close to the Asian continent. It also seems to us unnecessary to invoke such an error, since there is a simpler, meteorological explanation. This is that the dust that exits the Asian continent likely becomes entrained in an atmospheric process observed off the Asian coast, and known as the “Warm Conveyor Belt” [Hannan et al., 2003]. Through this process, surface air is swept into the free troposphere, above the marine boundary layer, and thus little of the mineral aerosol can be deposited locally to the surface ocean.
3.3. Al and Salinity Signal in the Surface Waters Near Kauai
 The enriched concentrations of Al close to Kauai (Figure 4) have not been modeled as dust deposition since it is believed that these elevated values result from fluvial runoff from the islands (were we to model this as dust deposition, the values would range around 3–4 g mineral dust m−2 yr−1, which is almost tenfold higher than the nearby values). While it is likely that dust generated by erosion of the nearby islands contributes somewhat to this signal, it is unlikely that its magnitude is this large; even downwind of major deserts, e.g., the Sahara, dust deposition is only in the 10–20 g mineral dust m−2 yr−1. However, this is a subject area where clearly more information would allow the relative roles of eolian and fluvial contributions of high islands to the seeding of the surrounding oceans with reactive trace elements to be disentangled.
 Nevertheless, it is likely that the major contribution to this enhanced Al feature is fluvial runoff and this view is supported by the surface water salinities in this region. Although fairly uniform through the subtropical gyre, the surface water salinities are noticeably lower in the vicinity of the NW Hawaiian Island chain, particularly Kauai. Surface water salinities which are fairly uniform at 35.5 from ∼170°E, start dropping around 168°W and 24.9°N (approximately at the Gardner Pinnacles). The salinities continue to fall along the ship's track, reaching 34.82 at 160°W, which is west of Kauai, and upstream of it in the sense of advective flow of the NW Hawaiian Ridge Current (Figure 7). Runoff from recent rainfall on Kauai was evident in the coastal waters along the Na Pali coast as the ship transited the region (distinct red coloration of inshore surface waters). A plot of Al versus salinity along the Na Pali coast (between dates 29.98 and 30.08; Figure 8) shows the negative correlation between these parameters that would result from a fluvial source of Al to the surface waters. This correlation could be used to calculate a composite fresh water end-member of ∼9 μM Al. However, the possibility of other coastal inputs (e.g., sediment resuspension), and the fact that this transect was not designed to address the complex processes of fresh water mixing, would render this value unreliable.
 To what degree this lowering of salinity is a result of the steady state effect of high island-induced rainfall or rainfall coincident with our particular cruise is impossible to distinguish from a single sampling. However, it is evident that, at least on a temporary basis, the island masses affect the surface water geochemistry of the surrounding ocean through a combination of fluvial and perhaps eolian inputs.
3.4. Vertical Distributions of Al at the Stations
 The primary input of Al to the open ocean surface waters is transmitted to the oceanic interior by subduction of surface waters masses. While Al is efficiently scavenged by particulate uptake in surface waters (biologically or nonbiologically), it does not appear to be remobilized within the water column during the remineralization of that particulate material, even in regions underlying extremely high levels of particulate Al input and high levels of export production [Moran and Moore, 1992; Measures and Edmond, 1990; Measures and Landing, unpublished data]. Thus, in the open ocean the subsurface distribution of Al results primarily from the subduction and advection of water masses containing Al, the concentration of which was determined by surface ocean processes in the outcrop regions where the water mass originated. Once below the surface layers, dissolved Al appears to have a residence time of the order of 150 years [Orians and Bruland, 1986] and thus the dissolved Al is slowly removed by scavenging processes from the water masses as they advect through the ocean interior.
 Data from the nine profile stations are presented in Figures 9 and 10, and the hydrographic interpretation of these stations is presented by Measures et al. (submitted manuscript, 2005). While the data cover a wide concentration range, there are a number of common features that occur at these stations and those features will be discussed in relation to the hydrographic setting of the particular stations below.
3.4.1. Subarctic Gyre: Stations 2 and 3
 As discussed above, surface waters of the subarctic gyre displayed the lowest dissolved Al concentrations. Not surprisingly, the subsurface waters at the two subarctic gyre stations (2 and 3) also display very low concentrations. Surface water values of ∼1 nM at Station 2 (KNOT) rise slowly reaching ∼1–2 nM in the bottom waters at 4000–5000 m.
 Station 3 also has similarly low surface water values (∼1 nM) which also persist unchanged well into the extremely low oxygen waters, where oxygen concentrations drop to <40 μM O2 by 200 m (Measures et al., submitted manuscript, 2005, Figure 10). At this station, though, an abrupt, but small increase in the Al values to ∼3 nM is seen at 300 m, and these values are maintained to 1200 m.
3.4.2. Subtropical Gyre: Stations 1 and 5–9
 At the subtropical gyre stations in the southern part the cruise track (1, 5–9) surface water Al values are significantly higher, ranging from 3–6 nM, reflecting both the greater eolian dust deposition at these latitudes and the shallower mixed layers. In general at each of these stations, below the bottom of the mixed layer, there is a subsurface minimum which is immediately followed by an intermediate maximum between 100 and 300 m where Al values reach 5–9 nM. Below this depth, values decrease to a minimum, which is seen in all stations at about 1000 m, the depth of the oxygen minimum. The eastern subtropical gyre stations (8 and 9) also show surface water enrichments of 4–6 nM, with values dropping rapidly below the mixed layer. While at Station 8 there is a small subsurface maximum at 170 m, there is none at Station 9, where Al values continue to decrease to <2 nM at the oxygen minimum found at ∼800 m.
 Most stations were not sampled below 1500 m and thus the deepest samples at those stations are not significantly different from the 1000 m values. Station 7, sampled to intercept the Antarctic Bottom Water (AABW) that flows into the North Pacific, shows a uniform layer of enriched Al (∼4 nM) below 3000 m. At Station 9, deep-water values are somewhat lower, with a uniform layer below 2000 m, where Al values are ∼2.3 nM and with a slight increase in the deepest sample.
3.4.3. Origin of the Intermediate Maximum in Al
 A plot of the intermediate maximum in Al against potential density (Figure 11) shows that this maximum occurs between potential densities of 25.2 and 25.7 kg m−3. The potential temperature across this density range at these stations is between 14.7 and 17.5°C, and salinity varies from 34.55 to 34.76. These values are consistent with this water mass being subtropical mode water, which has been identified in the North Pacific by several authors [e.g., Masuzawa, 1969; McCartney, 1982; Talley, 1985; Yuan and Talley, 1992; Bingham, 1992]. The formation and occurrence of this water mass in the context of this cruise is discussed by Measures et al. (submitted manuscript, 2005). Huang and Qui , while pointing out that the exact definitions of this water mass varied between authors, defined the mode water as having a density range from 25.2 to 25.6 and calculated an annual formation rate of ∼5 Sv, yielding a renewal time of 9–17 years across this density range, somewhat longer than the 7-year estimate for its North Atlantic equivalent.
 The outcropping of the 25.2 to 25.6 range of isopycnals occurs along the northern boundary of the subtropical gyre across the Pacific ocean. However, Huang and Qui  calculate that the major locus of formation of the mode water formed from this density surface was in the region of 30–34°N, and between 140–160°E, which encompasses the recirculation region of the Kuroshio. The Al concentration in the surface samples that we obtained as our cruise track passed through the eastern edge of this region on the way to Station 1 were among the highest observed during the expedition (with the exception of the Kauai coastal waters). The 6–8 nM Al concentrations we observed in the surface waters of this region (Figure 2), although not necessarily exactly representative of the late winter conditions when subduction takes place, are strikingly similar to the 5–9 nM Al concentration range observed in the mode waters (Figure 11). In fact, along our cruise track, these were the only surface waters that we encountered that were sufficiently enriched in Al to provide the observed Al concentrations in the mode water.
 The Al maximum in the mode water is greatest at Station 1, probably indicative of the proximity to the formation region. Surprisingly though, the maximum at Station 7 is higher than those seen at Stations 5 or 6. In addition, there is no sign of the mode water signal in the mixed water region at Station 4, i.e., this feature appears to be just in the subtropical gyre. In addition, the feature is not present further to the east at Stations 8 or 9. Whether this variation in the intensity of the Al maximum is an indication of interannual variability in the concentration of Al in the subducted surface water or an indication of the circulation and mixing of the subducted water is impossible to distinguish with such a limited data set. Recent unpublished data (Measures, Brown, Landing and Buck, personal communication, 2004) from a transect along 30°N indicates that the Al maximum in the mode water is a pervasive feature in the western subtropical gyre that persists to approximately 175°W.
3.4.4. Al Minimum in the Intermediate Waters
 Below the mode water maximum, the Al values decline toward the salinity minimum of the North Pacific Intermediate Water (NPIW), which is found at a potential density of ∼26.7 (depth range 600–750 m). Al values in the NPIW are 2–4 nM, with no obvious geographic trend. Lowest Al values are seen below this depth at around 1000 m in all profiles, with concentrations ranging from 0.9 to 2.1 nM. This depth, which varies from 775 to 975 m, corresponds closely with the oxygen minimum, which varies from its least intense at Station 1 (43 μM) to its most intense at Stations 2 and 3 (∼20 μM). Mid water Al values at Station 9 are ∼1 nM higher than those reported by Orians and Bruland  at the Vertex-4 station (155°W, 28°15′N), approximately 5°30′ further to the north. It is interesting to note that the oxygen levels in the minimum at 30°N (http://www.coaps.fsu.edu/RVSMDC/CLIVAR/html/data.shtml) (somewhat north of Vertex-4) are ∼10 μM, which is significantly lower than the 25 μM seen at Station 9. While Al differences this small are on the edge of interpretability, the lower concentrations again appear to be associated with lower oxygen levels in the water column. Since Al is removed from the subsurface waters via scavenging processes, the coincidence of the lowest Al values with the oxygen minimum is not surprising. The development and intensity of the oxygen minimum in the mid waters of the Pacific is a combination of both the length of time since this water mass was ventilated at the ocean's surface and the amount of organic matter that has been remineralized within it. Thus the initial Al concentration in the water mass, imprinted in the surface waters, is continually eroded from the subducted water masses by scavenging processes as it advects through the basin interior. In the subtropical gyres then, with relatively small gradients in export flux, the oxygen minimum is a rough proxy for the integrated magnitude of the scavenging process that the water mass has undergone.
3.4.5. Al in the Bottom Waters
 The bottom waters of the Pacific are composed of AABW that forms to a large extent in the Weddell Sea and enters the North Pacific basin through deep passages west of the Mariana Ridge. This relatively recently ventilated water mass fills the entire bottom of the western and eastern North Pacific and mixes up into the oxygen minimum. Thus the Al gradient in the deep water primarily represents a mixing between the AABW and the heavily scavenged water of the oxygen minimum, with an overprint of regional circulation.
 At Station 7, Al values rise abruptly from 2.7 nM at 2500 m to 3.8 nM at 3000 m and remain at this value to the deepest sample (5400 m). In comparison, at Station 9 Al concentrations are similar to Station 7 at 2500 m (2.4 nM), but then remain fairly constant and lower than those at Station 7. Oxygen values are also slightly lower in the bottom waters of Station 9 and thus the lowered Al signal at Station 9 may be indicative of progressive scavenging of Al within the AABW as it advects from its entry point, to the south of Station 7 within the North Pacific. By comparison, Al in AABW was determined to be approximately 2 nM in the AABW of the Weddell Sea [Moran et al., 1992] and 7 nM in the South Atlantic further from its formation region [Measures, 1995]. Orians and Bruland  reported similar Al values (∼2 nM) in AABW, north of our Station 9, and found lower concentrations (<1 nM) in deep waters further to the NE approaching the US West coast.
 Despite the advection of significant amounts of dust from the desert regions of Asia across the NW Pacific Ocean, it is clear that only a small amount of this material enters the surface waters of the region. This lack of dust deposition, and thus lack of input of eolian Fe, when coupled with the large supply of macronutrients to the surface waters, is responsible for the HNLC character of these waters. Thus the HNLC nature of the NW Pacific results from the same combination of large macronutrient supply from deep winter mixing, and low eolian micronutrient supply, as seen in the HNLC regions of the equatorial Pacific and Southern Ocean.
 Dust deposition to the surface waters of the NW Pacific calculated here from surface water Al concentrations are significantly lower than the GESAMP model predicts (by more than one order of magnitude near the Asian continent). However, in the majority of the subtropical gyre, our Al-based estimates are close to the values predicted by the GESAMP model. While it is likely that the overestimation in the NW Pacific is the result of the peculiar meteorology of that particular region (which is beyond the scope of this manuscript), it is clear that a considerable amount of work must be done to ground-truth dust deposition models and assess their regional validity if they are to be productively incorporated into global predictive models.
 The incorporation of the surface water derived Al into the mode waters of the Pacific makes it an interesting tracer of the locus of mode water formation since Al is not uniformly distributed in surface waters.
 The enrichment of Al in the coastal waters of the Hawaiian Islands indicates that high islands in remote locations might be important local sources of reactive trace elements to surface waters.
 The observation of a region of elevated Al in surface waters NW of the Hawaiian Islands, similar to that reported by Johnson et al. , suggests that there might be surface water regions that preferentially receive atmospheric deposition as a result of persistent meteorological conditions, and that these might be useful areas to study the inter-relationships between geochemical inputs and biological responses.
 We wish to thank Captain Dave Murline, resident technician Gene Pillard, and the officers and crew of the research vessel Melville for their enthusiastic and professional support of this project and their help in ensuring a successful expedition. We would also like to thank all the participating scientists for their contributions to the success of this expedition. We would like to thank the Intergovernmental Oceanographic Commission for their continued support of the baseline cruises and the United States National Science Foundation for its financial support of this project through grant OCE-0117917 to C.I.M., G.A.C., and W.M.L. We would like to thank Brad Moran and one anonymous reviewer and the associate editor Michiel Rutgers van der Loeff for their thoughtful reviews and constructive suggestions for improving the paper. This is contribution 6625 of the School of Ocean Earth Science and Technology, University of Hawaii.