Data for Al, Fe, Na, Sb, Se V, and Zn in aerosol particles from Bermuda are used in combination with parameterized dry deposition velocities to investigate trace element recycling on atmospheric sea salt. During July and August, which are in the high dust season, >10% of the monthly mean aerosol Na can be ascribed to crustal sources rather than sea salt, but in general the contribution of mineral dust to Na dry deposition is small (∼1%) relative to that from sea salt. In fall and winter, ∼10 to 20% of the aerosol Al, a commonly used indicator of mineral dust, is attributable to recycled sea salt, but the dust fluxes then are small, and longer-term averages show a recycling effect of only 1 to 5% on dry deposition. The percentage of recycled Fe appears to be even less than for Al. Noncrustal/non-sea-salt (NC/NSS) sources account for >70% of the aerosol Sb, Se, V, and Zn, but differences in the dry deposition velocities for three particle types strongly affect their air/sea exchange. Dust can account for more than half of the new dry deposition inputs of V and Zn while the amounts of Sb and Se recycled on sea spray approach or exceed their new inputs from dust and the NSS/NC sources.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The elements investigated here (Al, Fe, Na, Sb, Se, V and Zn) exist in the marine troposphere as components of aerosol particles that originate from a variety of natural and anthropogenic sources [Nriagu and Pacyna, 1988; Arimoto et al., 1995]. Sea salt and mineral dust are major components of marine aerosols in terms of mass, and enrichments of trace elements relative to these two natural sources often are demonstrated by comparing elemental ratios in samples against those in selected reference materials. Commonly used indicator elements are Na for sea salt and Al or various other elements, including Fe, Sc, Si, and non-sea-salt Ca, for crustal material [e.g., Duce et al., 1976; Rahn, 1976; Prospero and Nees, 1987]. Of the elements we consider, marine biogenic sources are significant only for Se, and unlike the other elements investigated, the biogeochemical cycling of Se also involves significant reactions in the gas phase [Mosher and Duce, 1987; Ellis et al., 1993; Amouroux and Donard, 1996].
 Previous air/sea exchange studies in the western North Atlantic Ocean have considered the relationships between continental sources and trace element fluxes [Church et al., 1990, 1991; Church and Véron, 1993; Cutter, 1993; Véron et al., 1992; Jickells et al., 1998], but most of those studies dealt primarily with wet deposition. To better characterize seasonal and longer-term variations of transport and deposition, we examine data for trace elements in aerosols collected over a 5-year period as part of the Atmosphere-Ocean Chemistry Experiments (AEROCE) at Bermuda. Specifically, the data were used to (1) evaluate trace element recycling on sea-salt particles and estimate net dry deposition rates, (2) estimate the relative contributions of natural versus anthropogenic sources on trace element concentrations and dry deposition, and (3) characterize seasonal variations in trace element deposition. The focus of the dry deposition calculations is on various geochemical implications of recycling, and this study is not intended to be an assessment of deposition models or an evaluation of the trace element deposition with respect to ocean chemistry. A companion paper (T. M. Church, Univ. of Delaware, personal communication, 2001) will focus on the wet deposition of these elements, and an integrated paper involving the dry and wet deposition estimates is also planned.
 Bulk aerosol samples were collected at Tudor Hill, a coastal site on Bermuda (32.27°N, 64.87°W), from August 1988 to July 1993. High-volume (∼0.7 m3 min−1) aerosol samples were collected on Whatman 41® filters (Whatman International Limited, Maidstone, England) from atop a 23-m tall anodized aluminum tower. The daily aerosol collections were controlled by computer with respect to wind direction and wind speed to avoid local island contamination.
 After sampling, the aerosol filters were placed in polyethylene bags and shipped to the University of Rhode Island where they were analyzed by instrumental neutron activation (INAA) [Arimoto et al., 1992]. For these analyses, the aerosol-laden filters were cut into quarters, pressed into pellets, and irradiated in the 2 MW research reactor operated by the Rhode Island Nuclear Science Center. Two INAA procedures were used, the first a short (1 to 2 min) irradiation procedure for determinations of Al, Na, and V and the second a long (∼24 hour) irradiation method for Sb, Se, and Zn.
3. Results and Discussion
3.1. Sea Salt and Mineral Dust
 When studying the air-sea exchange of trace elements, it is often desirable to determine net exchange rates, that is, the total deposition rates corrected for material recycled between the ocean and atmosphere on sea salt particles. In the marine atmosphere, the sea salt mass, often represented by Na, is mainly associated with coarse (>1 to ∼100 μm) particles, while crustal elements such as Al and Fe mostly are concentrated in particles with mass-median radii of ∼1 to 3 μm [Duce et al., 1983; Arimoto and Duce, 1986; Arimoto et al., 1997]. Trace elements enriched over the levels expected from these natural sources, often as a result of pollution emissions, tend to be concentrated in submicrometer particles.
 Various types of substances, including trace elements [Weisel et al., 1984] and radionuclides [Walker et al., 1986] are enriched on fresh sea-salt particles relative to the composition of bulk surface seawater, and one must quantify the extent of these enrichments if the recycled component and net deposition rates are to be determined. For trace elements, these enrichments can range into the tens of thousands over seawater values [Weisel et al., 1984], and they result from the scavenging of surface active material by rising bubbles [Wallace and Duce, 1978] and the rupture of the sea-surface microlayer [MacIntyre, 1974].
 As a first step in evaluating trace element fluxes and the extent of recycling, we calculated the amount of non-sea-salt Na in the aerosols by assuming that in addition to sea salt, the other significant source for Na was mineral dust. The second assumption made was that all of the Al was crustal in origin (this second assumption is evaluated below). The percentage of non-sea-salt Na in the aerosol samples was calculated as follows:
where [Na]Observed is the concentration of Na determined in samples. Crustal Na (NaDust) was calculated based on the composition of Harmattan dust [Adepetu et al., 1988, Table 1], which we considered the best available compilation of elemental data for dust from the Sahara:
where [Al]Observed is the Al concentration (μg m−3) for the sample and [Na/Al]Harmattan is the ratio of Na to Al in Harmattan dust (the concentrations of Na and Al in this source material are 6400 and 61100 μg g−1, respectively [Adepetu et al., 1988]).
 The concentrations of sea salt and mineral dust over the North Atlantic exhibit strong seasonal cycles [Arimoto et al., 1995], and these affect both air/sea exchange and recycling. In July and August, which are in Bermuda's high-dust season, an appreciable amount of the aerosol Na is not from sea salt but rather is a component of the mineral dust. For some monthly means during the high dust season, over 10% of the Na could be ascribed to the crustal source, and for some daily samples, the percentage of non-sea-salt Na was much higher, exceeding 30% on occasion (Figure 1).
 The next step in the assessment of Na and Al as indicator elements was to evaluate the assumption that all of the Al in the aerosol samples was crustal in origin. Determining the amount of sea-salt Al in an aerosol sample is not so straightforward as calculating non-sea-salt Na, however, because unlike the relatively well-defined composition of Harmattan dust, there is little information for the Al/Na mass ratio in freshly produced sea-salt aerosols ([Al/Na]Sea salt). There is, however, a way to evaluate this ratio. One can calculate [Al/Na]Sea salt using data for sea-salt particles experimentally produced with the bubble-interfacial microlayer sampling device (BIMS) [Weisel et al., 1984]. Following this approach, we calculated a sea salt Al/Na mass ratio of ∼2.3 × 10−4 using fresh sea-salt aerosols for the BIMS enrichments and the seawater composition in Weisel et al. . Sea-salt Al is then calculated as:
 The BIMS [Al/Na]Sea salt ratio is comparable to the lowest observed Al/Na ratios for the Bermuda samples (Table 1), which we assume to be indicative of the lowest relative mineral aerosol to sea-salt concentrations, and therefore arguably representative of sea-salt composition. Of the 844 samples with matched data for Al and Na, only six had an Al/Na ratio less than the BIMS value, and only one sample was less than 1.0 × 10−4, suggesting that the [Al/Na]Sea salt value we derived was a reasonable value to use for calculating sea-salt Al.
Table 1. Observed and Reference Elemental Mass Ratios
 In this simple assessment, the percentage of noncrustal Al was calculated as the concentration of sea-salt Al divided by the total observed Al ([Al]Observed). No provisions were made for Al associated with coal fly ash or other types of aerosols that contain additional Al,
 Applying the reference Al/Na ratio for sea salt aerosols to the observations shows that the amount of recycled Al on sea salt particles is generally small relative to the dust-associated Al. The arithmetic mean value for the recycled component of Al calculated for all of the daily aerosol samples is ∼6% (Table 2). Over the 5-year study, the noncrustal Al calculated on a monthly basis can amount to as much as ∼30% of the total aerosol Al, but such high percentages occur for only a few months of the year (Figure 2). The percentage of noncrustal Al in fall and winter exceeds 50% in some individual samples, but the amount of Al in the atmosphere is small at that time of year, and hence the deposition rate for Al also is small. Monthly arithmetic mean values of the percentages of noncrustal Al for November to February, averaged over the 5-year study, indicate that in late fall and winter ∼10 to 20% of the aerosol Al can be attributed to sea salt.
Table 2. Percentage Contributions of Sea Salt, Mineral Dust and Other Sources to Elemental Concentrations in Aerosols
 While the biogeochemical cycling of mineral dust is of considerable scientific interest, much of the attention given to this material lately has focused on Fe owing to its importance as a micronutrient for oceanic phytoplankton [e.g., Martin et al., 1994; Falkowski, 1995; Coale et al., 1996]. Therefore, we repeated the assessment of recycling, using the same approach as described above for Al, but this time focusing on aerosol Fe. These calculations again make use of the Weisel et al.  data, which indicate that the Fe/Na ratio for fresh sea salt aerosols is 5.6 × 10−5 (Table 1). This figure is reasonably close to (∼40% lower than) the lowest Fe/Na ratio observed in 738 aerosol samples from Bermuda for which data are available, again indicating reasonable agreement between the BIMS-derived aerosol composition and the field data.
 The seasonal trend in noncrustal Fe in aerosols (not shown) closely tracks that for noncrustal Al, but the calculated percentage of recycled Fe basis is smaller, never exceeding 5% for any month. The monthly mean percentages of noncrustal aerosol Fe between March and September generally were <2%, and they averaged ∼4% from November to February. Although these attempts to model the composition of sea salt aerosols are admittedly crude, the general conclusion we draw is that the amount of atmospheric Al and Fe contributed by sea-salt aerosols is on average small relative to the crustal source.
 The calculation of non-sea-salt Na and noncrustal Al and Fe are based on two assumptions that are in a sense mutually exclusive, that is, to calculate non-sea-salt Na, one assumes that all of the Al is crustal while in calculating noncrustal Al, one assumes that all of the Na is associated with sea salt. As shown above neither of these assumptions is strictly valid for the entire year, and for those situations in which the assumptions do not hold, the non-sea-salt Na and noncrustal Al fractions obviously would be overestimated.
 While this exercise has shown that neither Na nor Al are perfect indicators, the non-sea-salt Na calculated for Bermuda from October to March, amounts to only a few tenths of a percent of the total Na, and thus non-sea-salt Na can be considered negligible for that part of the year. During spring and summer, when sea salt concentrations fall and dust concentrations rise, non-sea-salt Na is of the order of 10% of the total Na. From March to September, the 5-year average noncrustal Al and Fe was ≤5%, and as shown below this is when most of the mineral aerosol deposition occurs. When the dust concentrations are low, ∼5 and 15% of the monthly average atmospheric Fe and Al, respectively, can be ascribed to sea salt.
 As a major focus of this paper is on deposition, we are particularly interested in evaluating the importance of the recycled component with respect to dry deposition, and this is addressed in the following section. It is worth mentioning at this point, however, that in terms of deposition, NaDust would tend to be associated with aerosol particles smaller than sea-salt Na, and hence crustal Na would have a lower dry deposition velocity than sea-salt Na. The same consideration would hold for Al, that is, AlSea salt would be on larger particles than crustal Al (AlDust), and therefore the two Al-containing aerosol populations would have different deposition velocities, with the larger sea-salt particles being deposited more rapidly.
3.2. Trace Element Recycling on Sea-Salt Aerosols
 The sea-salt and crustal components of Sb, Se, V, and Zn in aerosols were calculated based on (1) the observed concentrations of these elements, (2) the concentrations of NaSea salt and AlDust and, (3) the elemental composition of crustal material and sea salt taken from the following references as appropriate: Adepetu et al. , Taylor and McLennan , Weisel et al.  (Table 1). Estimates of the percentages of each element attributable to sea salt, mineral dust, and the source or sources responsible for the enrichments were calculated separately for all samples with complete data for all elements of interest. We note that the uncertainties associated with these calculations are difficult to quantify but more than likely large. To evaluate the representativeness of the various reference ratios, we include in Table 1 a summary of the observed elemental ratios, showing the minimum observed ratio and the average ratio for the lowest 1% and lowest 2% of the observations. As before, the assumption we make is that lowest observed values for the elemental ratios can be viewed as the best empirical indicators of the composition of the sea salt and mineral dust.
 In several cases, the elemental ratios taken from the literature were not consistent with the observations, and in those cases the composition of other reference materials were evaluated for use in the model calculations. The Zn/Al ratio for Harmattan dust, for instance, calculated from Adepetu et al.  is 3.6 × 10−2 whereas our lowest observed ratios for aerosols from Bermuda were of the order of 5 × 10−4. In comparison, a representative Zn/Al ratio for the upper continental crust is 8.8 × 10−4 according to data in Taylor and McLennan . In this case then, based on the greater similarity between the observed and crustal ratios, we used the average crustal value rather than the Harmattan ratio to calculate the percentage of Zn from the non-sea-salt/noncrustal (NSS/NC) sources. A similar situation occurred for Sb, that is, the Sb/Al ratio reported for Harmattan dust was over 200 times that calculated for average crustal material and 50 to 100 times higher than the lowest observed values. The crustal Sb/Al ratio from Taylor and McLennan, on the other hand, was only ∼50% less than the lowest observed value.
 No data are available from the BIMS study for Sb and Se, and therefore we had to rely on sample data, that is, the lowest observed values, to best approximate the Sb/Na and Se/Na ratios in sea salt. Relative to the composition of seawater [Millero, 1996], the lowest observed ratios (Table 1) would represent enrichments of 60 and 600 for Sb and Se, respectively, which are well within the range of those found for other elements in BIMS studies [Weisel et al., 1984]. Note, however, that the composition of the sea-salt aerosols could be influenced by the sorption of gas-phase Se, and therefore the Se/Na ratio used for assessing source contributions could be affected by material recycled from the gas phase as well as through the sea-salt production process.
 For the four enriched elements considered, the NSS/NC sources could account for up to ∼90% of total observed aerosol mass (Table 2). Of these elements, V was the least affected by NSS/NC sources, but even this element had a 70% contribution from them. The apportionments of Sb and Se among the three sources were similar; sea salt accounted for roughly twice the amount from mineral aerosol, but ∼85% of the mass of these two elements was from NSS/NC sources. Zinc showed approximately the same contribution from NSS/NC sources as Sb and Se, but more of the Zn was attributable to crustal material than sea salt (Table 2).
3.3. Air/Sea Exchange
3.3.1. Aluminum and Iron
 There are few direct measurements of dry deposition (Fd) in remote marine locations, and this is as much a result of concerns over the representativeness of collection substrates as it is to difficulties in sampling [Arimoto et al., 1985; Dulac et al., 1989]. Studies of dry deposition have made use of wind tunnels [Hall and Upton, 1988] and various surrogate surfaces, such as petri dishes partially filled with water [Zobrist et al., 1993] and dry surfaces [Lin et al., 1994; Shahin et al., 2000]. While each of these surrogate surfaces has certain practical advantages, none is fully representative of a natural water surface. As an alternative to actual measurements, dry fluxes (Fd) often are estimated from measured trace element concentrations in air (Ca) and model-derived or estimated dry deposition velocities (Vd) [Jickells et al., 1987; Duce et al., 1991] as follows:
 Major uncertainties arise in dry deposition rates calculated in this way, especially for submicrometer particles [Sievering, 1984], because in addition to gravitational settling, Vd typically includes terms for particle slip, impaction and diffusion, all of which vary in complex functions of particle size, particle shape, and meteorological conditions [Slinn, 1983]. Measurements of dry deposition of small particles have likewise produced conflicting results [Nicholson, 1988], but the emerging consensus is that the mass fluxes are dominated by giant particles, typically those greater than 7 to 10 μm [McDonald et al., 1982; Davidson et al., 1985; Dulac et al., 1989; Lin et al., 1994]. It is worth noting, however, that even for particles >10 μm in diameter, the disagreement among models can be considerable [Kim et al., 2000].
While this simple model has substantial uncertainties, more refined methods for calculating deposition also would be problematic. For example, while there are some data for size-separated aerosols from cascade impactor samples [Arimoto et al., 1997], it would be difficult to evaluate the fractions of recycled material for these samples because the trace element enrichments presumably vary as a function of particle size, and little or no information is available for assessing this. No less important, the inability of cascade impactors to capture giant particles is well documented [Dulac et al., 1989; Lin et al., 1994], and because these particles are critically important for deposition, the errors from undersampling them could easily lead to inaccurate flux estimates.
 In calculating net dry deposition rates, the total observed Al and Fe aerosol concentrations were separated into their crustal and noncrustal components as described above, and deposition rates were calculated based on the Vds for sea salt and mineral dust shown above. Dry deposition rates were calculated for both the daily sample data and the monthly means. The two methods produce different results because the percentages of recycled Al and Fe are comparatively small on the days when the dust concentrations and deposition rates are high. The sporadic nature of the dust fluxes makes the recycled fractions calculated from monthly or longer-term averages lower than those calculated from the daily samples.
 The deposition calculations show negligible effects of dust on the deposition of Na over the course of a year, and the contributions from sea salt amount to 5 and 10% of the Fe and Al dry deposition, respectively, when calculated on a daily basis (Table 3). Uncertainties in the percent contributions from sea salt and dust to the dry deposition of Na, Al, and Fe were assessed by manipulating the Vds in the model to calculate mean daily maxima and minima for these two components. For this assessment, the maximum percent contribution for the sea-salt component relative to dust was calculated by using the model's maximum sea salt dry Vd (6 cm s−1, see above) and the minimum Vd for dust (0.33 cm s−1). Conversely, the maximum percent crustal contributions for these elements were calculated using the minimum sea salt Vd (1.5 cm s−1) and the maximum Vd for dust (3 cm s−1).
Table 3. Percentage Contributions of Sea Salt, Mineral Dust and Non-Sea-Salt/Noncrustal Sources to Trace Element Dry Deposition: Daily Averages
 The recycled component of the Al dry deposition is important only when dust concentrations and deposition rates are low. In terms of daily averages, the recycled component of the Al dry deposition was 12%, with a large standard deviation (18%) reflecting the strong day-to-day variability, and an estimated daily minimum percent contribution of 4% and a maximum of 32% (Table 3). Maximum recycling of Al occurs in December and January when sea-salt concentrations are high and dust concentrations are low. When averaged over the 5-year study, ∼15 to 20% of the Al dry deposition during these two months can be attributed to recycled sea salt. During the high-dust season, the recycled component is negligible, amounting to only ∼1% of the total Al dry deposition for the months from May to September, with an arithmetic mean of 6% for all months combined.
 The annually averaged net dry deposition flux for Al, based on the 5-year data set is 1.1 × 105 μg m−2 yr−1, a value comparable to the Al dry deposition fluxes of 1.5 × 105 μg m−2 yr−1 for 1988–89 and 1.1 × 105 μg m−2 yr−1 for 1989–90 reported by Jickells et al.  (which did not account for recycling). These authors calculated the annual Al dry deposition rates using a subset of the 5-year data set used for the present study. The similarity in the fluxes from the Jickells et al. studies and the present work simply reflects the fact that recycling does not greatly affect the dry deposition of this element and that the dust concentrations have remained relatively stable over the study period.
 The calculated annual dry deposition of Fe is 6.2 × 104 μg m−2 yr−1, and the recycled component for Fe appears lower than that for Al reaching a maximum of ∼5% per month from December to February and averaging only a few percent per day when calculated over the duration of the study (Table 3; see also Figure 3). As was the case with Al, the impact of recycling for Fe is significant only when the dust concentrations and deposition rates are low. One implication of this result is that the new inputs of Fe during low dust periods are about 5% less than the inputs not corrected for recycling.
3.3.2. Enriched Elements: Antimony, Selenium, Vanadium and Zinc
 Recycled sea salt can account for 30 to 60% of the dry deposition of Sb, Se, V and Zn to the sea surface when calculated as arithmetic means for the daily samples (Table 3). Maximum and minimum percent contributions for the sea-salt, crustal, and NSS/NC components were estimated by using various combinations of maximum and minimum deposition velocities for the model described above. For example, the maximum sea-salt contribution was estimated by using the maximum Vd for sea salt and the minimum Vds for the crustal and NSS/NC components in the model.
 Another, perhaps more geochemically meaningful, way of assessing the importance of the various sources is by calculating their percent contributions to the estimated annual deposition rates rather than on a daily basis. For V and Zn, this calculation produces higher contributions from the crustal source compared with the calculations done on a sample-by-sample basis (67% versus 35% and 51% versus 27%, respectively; Tables 3 and 4), and this is simply another consequence of the sporadic nature of the dust fluxes noted above. When calculated from annual deposition rates, the recycled dry deposition of Sb, Se, and V was greater than the new inputs from NSS/NC sources, but for Zn the recycled fraction was smaller than inputs from the other sources.
Table 4. Contributions of Sea Salt, Mineral Dust and Other Sources to Trace Element Dry Deposition: Quarterly and Annual Averages
Dry Deposition Rate (μg m−2 d−1) and Percentage of Total Dry Deposition
 The relative importance of the recycled component for the enriched elements varies strongly with season, due in part to the seasonal variations in salt concentrations, but more as a result of the strong seasonality in the new inputs of these elements associated with atmospheric dust (Table 4 and Figure 4). The percent contributions associated with the dust flux range from ∼10% of the total inputs of Sb and Se to 67% for V and 51% for Zn (Table 4); these are equivalent to 14%, 27%, 87% and 61% of the new inputs of Sb, Se, V and Zn, respectively. Moreover, the impact of the dust on the dry deposition of these elements is highly seasonal: for V and Zn, approximately 50 to 60% of the total dry deposition from July to September is attributable to dust while for Sb and Se, the crustal contribution in that calendar quarter is about 25% (Table 4, see also Figure 4). In comparison, under low dust conditions, the contribution from dust sources to V and Zn dry deposition drops to ∼15%, and for Sb and Se the impact of the dust flux is negligible, amounting to ≤1% of the total dry deposition.
 Whether calculated as daily averages or from the annual deposition rates, less than half of our best estimates of the dry deposition for any of the enriched elements can be attributed to material originating from NSS/NC sources (Tables 3 and 4). This is a direct consequence of the low Vd in the model for the small particles on which these elements are concentrated, and this serves to emphasize the need for better estimates of the Vds for small particles. Even the average daily maxima for the NSS/NC contributions (Table 3), which are almost certainly overestimates, show that recycling has a greater effect on the dry deposition of trace elements than on their concentrations in the atmosphere.
 The second pattern that emerges from the calculations for the enriched elements is that the seasonal fluctuations in the NSS/NC component are much smaller than those associated with mineral dust or sea salt. For example, the dry deposition fluxes for Sb and Se, which showed the smallest contributions from crustal sources, when averaged over calendar quarters only varied from 1.3 to 1.6 μg m−2 d−1 and 9.5 to 11.5 μg m−2 d−1, respectively (Table 4). This is in contrast to V and Zn, which are more strongly influenced by dust and whose respective quarterly deposition rates varied by factors of 6 and 3. These temporal patterns in dry deposition are likely to be quite different from those in wet deposition because the latter are affected by the seasonal trends in precipitation as well as by the characteristics of the aerosol populations. As noted above, further studies comparing the dry and wet deposition rates, and their effects on ocean chemistry are in progress.
 The simple model calculations presented here are in several ways an oversimplification of the dry deposition process. The deposition patterns produced by the model were simply driven by the changes in aerosol concentrations observed during the course of the experiment. Thus, one major shortcoming of the model was the reliance on constant dry deposition velocities that did not take into account the effects of wind strength, relative humidity, or atmospheric stability, all of which are important determinants of particle deposition. Despite this limitation, the results demonstrate that the seasonal cycle in the dry deposition of mineral dust is much more pronounced than the dry deposition inputs of the enriched trace elements.
 Another limitation of the calculated dry deposition rates is more geochemical in nature, that is, materials originating from sea-salt, crust material and NSS/NC sources are not likely to be equally soluble in seawater, and hence their impacts on the composition of seawater and marine biota will not be in direct proportion to their mass fluxes. In general, one would expect the trace elements recycled on sea salt to be much more soluble than those associated with mineral dust, and that the solubilities of the particles would vary inversely with size [Salomons, 1984]. Aerosols from the various NSS/NC sources are a heterogeneous admixture of particle types, however [Anderson et al., 1996], and their physico-chemical properties have not been well characterized.
 Recycling of Al and Fe and presumably other crustal elements on sea salt is important only when dust concentrations and deposition rates are low, but for some enriched elements, such as Sb and Se, the recycled component can approach or exceed the new inputs from dust and NSS/NC sources. We note that the strategy of quantifying the recycled components of trace element deposition is equally relevant for wet deposition studies and may also pertain to certain organic substances. The effects of recycling on precipitation inputs are likely to differ quantitatively from dry deposition, however, owing to the weak particle-size dependence of in-cloud and below-cloud scavenging [Scott, 1981]. Nevertheless, the information presented here on the fractions of the trace elements associated with sea-salt, mineral dust, and the NSS/NC sources provides an indication of the levels of recycling that might be expected in wet deposition.
 This research was conducted as part of the AEROCE program and funded by the National Science Foundation under grants ATM 8702563, 91-14072, 9414262, 9728983 and 0002599. We appreciate the contributions of H. Maring, T. Snowdon, A. Glasspool, and M. McKay, who developed the AEROCE site at Bermuda and conducted the sampling operations there. We also are pleased to acknowledge the staff of the Rhode Island Nuclear Science Center for their support in the instrumental neutron activation analysis of the aerosol samples.