Journal of Geophysical Research: Atmospheres

Seasonal and spatial distributions and dry deposition fluxes of atmospheric total and labile iron over the tropical and subtropical North Atlantic Ocean

Authors


Abstract

[1] The tropical and subtropical region of the North Atlantic Ocean is under the influence of mineral dust plumes transported from North Africa, resulting in high atmospheric iron (Fe) deposition. Atmospheric Fe, as a major Fe source to this open oceanic region, may influence the rate of nitrogen fixation since Fe is a critical nutrient cofactor for the nitrogenase enzyme in diazotrophic microorganisms. Field measurements of Fe concentrations and Fe speciation in aerosols provide crucial information to understanding the biological role of this atmospheric Fe flux to the ocean. In this study, 24-hour aerosol samples were collected during winter (6 January 2001 to 18 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over this North Atlantic region. Three labile Fe species, that included ferrous iron (Fe(II)) and reducible ferric iron (Fe(III)) species, were measured in aerosol samples using a sequential aqueous extraction method. Microwave digestion of the aerosol samples followed by inductively coupled plasma-mass spectrometry was used to quantify total elemental concentrations. A spatial gradient of over nearly 3 orders of magnitude in the total Fe concentrations (from 1.6 ng m−3 at 28.6°N to 1688 ng m−3 at 10.9°N) was observed during the winter, while this gradient was not as strong in the summer. The mean total Fe concentrations were approximately a factor of 2 higher during the winter (mean value of 670 ng m−3 between 5°N and 26°N) than in the summer (mean value of 324 ng m−3 between 6°N and 26°N). The highest percentage of labile Fe to total Fe was observed between 26°N and 30°N in the winter with a mean value of 32%, which corresponded to low concentrations of total Fe. At latitude 0°N to 10°N, where the lowest Fe concentrations were observed in the summer, the labile Fe fraction with a mean of 5.0% was similar to the region from 10°N to 20°N where there were much higher mineral aerosol concentrations. Air mass back trajectories showed that mineral dust transport from North Africa is a significant atmospheric Fe source in this Atlantic region. However, the highest labile Fe to total Fe ratios were observed in air masses that had been over the ocean for greater than 7 days and also corresponded to high ratios of non-seasalt-sulfate (NSS-sulfate) to total Fe and oxalate to total Fe. The correlation with NSS-sulfate and oxalate suggests that labile Fe concentrations may have been influenced by anthropogenic activities from North America or Europe.

1. Introduction

[2] Fe is a critical micronutrient that limits primary productivity in the high-nutrient low-chlorophyll (HNLC) regions of the ocean [Behrenfeld et al., 1996; Boyd et al., 2000]. Fe is also a crucial micronutrient for diazotrophic microorganisms and therefore may influence nitrogen fixation in oligotrophic oceans [Falkowski, 1997; Gruber and Sarmiento, 1997]. Atmospheric deposition is the primary source of Fe to the oligotrophic oceans, and previous studies have investigated the concentrations, deposition and solubility of atmospheric Fe. The reported solubility of atmospheric Fe in seawater spans over 2 orders of magnitude (1% to 50%) although most are at the low end of this range [Zhuang et al., 1990; Chester et al., 1993b; Zhu et al., 1992, 1993; Jickells, 1999]. Measuring this soluble Fe fraction along with understanding the process that controls the Fe solubility is important since this soluble Fe may be associated with the amount of bioavailable Fe derived from the atmospheric deposition.

[3] There are relatively few field studies investigating aerosol Fe speciation in the marine boundary layer (MBL) of oligotrophic oceans, which results in large uncertainties associated with model calculations of Fe and bioavailable Fe fluxes to world oceans. Zhu et al. [1993] measured the total soluble Fe(II), total soluble Fe and total Fe concentrations in marine aerosol samples collected at Barbados, and showed that only 1% of the total Fe and 7.5% of the soluble Fe were in the Fe(II) oxidation state. A clear diel variability in the concentration of soluble Fe(II), with day values (mean 3.7 ng m−3) about twice night values (1.5 ng m−3), was also observed in marine mineral aerosols at Barbados [Zhu et al., 1997]. Zhu et al. [1993] measured the soluble Fe by leaching a filter with the aerosol sample in a pH 1.0 NaCl solution for 5 min at ambient temperature, and the leached solution was filtered through a 0.2 μm Nuclepore filter for Fe(II) analysis. Siefert et al. [1999] reported that the aqueous labile Fe(II) concentrations were between 4.75 and <0.4 ng m−3 during the intermonsoon but below the detection limit (<0.34 ng m−3) during the southwest monsoon over the Indian Ocean. Siefert et al. [1999] measured the aqueous labile Fe(II) by extracting a filter with the aerosol sample in a pH 4.2 formate buffer solution for 30 min and then filtering the extract through a 0.2 μm cellulose acetate syringe filter. Most of the labile Fe(II) was in the fine aerosol size fraction (diameter <2.5 μm) during the intermonsoon although most of the total Fe was in the coarse fraction. This fine-fraction “enrichment” was also observed over the tropical Atlantic ocean where the labile Fe(II) was about 0.5% of total aerosol Fe and correlated with NSS-SO42− and oxalate concentrations [Johansen et al., 2000]. The correlations with NSS-SO42− and oxalate may be due to either the chemistry occurring in the aerosol or possibly a common origin for these species (e.g., anthropogenic activities [Johansen et al., 2000]).

[4] The North Atlantic and North Pacific oceans are the two oceanic basins with the highest atmospheric Fe fluxes, accounting for 48% and 22% of the total Fe flux to the world oceans, respectively [Gao et al., 2001]. This deposition occurs as both wet and dry deposition and models predict that the fluxes due to both processes have a similar magnitude over the coastal and open oceans [Gao et al., 2003]. In wet deposition the aerosol Fe is incorporated into the raindroplets by either serving as the cloud condensation nuclei or through within-cloud or below-cloud scavenging processes. The dry deposition flux can be calculated by multiplying the atmospheric concentration by a deposition velocity that is primarily a function of particle size, wind speed and relative humidity. Several comprehensive models have been developed for particle deposition to natural water surfaces [Slinn and Slinn, 1980, 1981; Williams, 1982]. Slinn and Slinn's [1981] model showed that the particles with 1 to 10 μm radius have a dry deposition velocity around 1.0 cm s−1 under the condition of 5 m s−1 wind speed and 100% relative humidity. Jickells and Spokes [2001] suggested that a mean value of 1.0 cm s−1 properly represented dry deposition velocity of atmospheric Fe by comparing dust deposition estimates and sediment trap records in the Sargasso Sea. The North Atlantic Ocean is under the influence of mineral dust plumes transported from North Africa (Sahel and Saharan regions) as well as anthropogenic sources from North America and Europe. Asian dust storms can sometimes travel all the way across the Pacific Ocean and North America and reach this region, as observed by satellite based sunphotometry measurements [Thulasiraman et al., 2002] and aerosol data from the Interagency Monitoring Program for improved Visual Environments (IMPROVE) network [Jaffe et al., 2003]. Great variability in Fe concentrations and liable Fe(II) fractions has been observed in this tropical and subtropical North Atlantic region [Johansen et al., 2000].

[5] This paper presents further measurements of labile Fe concentrations using a new extraction method that analyzes for reducible labile fractions of Fe along with aqueous labile Fe(II) species [Chen and Siefert, 2003]. This more detailed investigation of labile forms of Fe in aerosols is important to understanding the processes controlling labile Fe fractions along with quantifying the flux of these species to oceanic regions where labile Fe may be an important rate limiting nutrient. Labile Fe concentrations were measured in aerosol samples collected during two research cruises (winter and summer) over the tropical and subtropical North Atlantic Ocean. Mineral dust plumes transported by trade winds from the Sahel region of North Africa are a dominant source of Fe to this North Atlantic region. The zone of maximum dust transport off North Africa coast moves north from about 5°N in winter to 20°N in summer as evident in satellite images [Husar et al., 1997; Moulin et al., 1997]. This shift is driven by the seasonal migration of the intertropical convergence zone (ITCZ). The data from the two cruises in this study were taken between 0°N and 30°N and spanned this zone of maximum dust transport. The aerosol samples were collected using a dichotomous aerosol sampler and the analysis was performed on both the fine-fraction (diameter <2.5 μm) and coarse-fraction (diameter >2.5 μm) aerosol samples. Three labile Fe species, including Fe(II) species and reducible Fe(III) species, were analyzed using an aqueous extraction procedure and long path length absorbance spectroscopy (LPAS) [Chen and Siefert, 2003]. The labile Fe fractions measured using this method are hypothesized to have different degrees of bioavailability to marine microorganisms. The analysis of labile Fe species was performed immediately after sample collection to minimize the effects due to sample storage (e.g., redox reactions). Total Fe concentrations were determined back in the laboratory using microwave assisted strong acid digestion of the filter samples followed by analysis of the digestion solution using inductively coupled plasma mass spectrometer (ICP-MS). The atmospheric Fe data will be presented along with the air mass back trajectories (AMBTs), giving a detailed map of distributions of Fe and labile Fe fractions in aerosols over the tropical and subtropical North Atlantic Ocean. The potential atmospheric Fe sources are discussed and mean atmospheric total and labile Fe dry depositional fluxes are calculated.

2. Methods

2.1. Aerosol Collection

[6] Aerosol samples (approximately 24 hour sampling times) were collected during two cruises (6 January 2001 to 19 February 2001 and 27 June 2001 to 15 August 2001) over the tropical and subtropical North Atlantic Ocean. A high-volume dichotomous virtual impactor (HVDVI) [Solomon et al., 1983] was used to collect two size fractions of ambient aerosols with an aerosol aerodynamic diameter cutoff of 2.5 μm. The fine- and coarse-fraction aerosols were collected on two 90 mm diameter Zefluor Teflon membrane filters (Gelman Zefluor, 1 μm pore size). The HVDVI aerosol collector was constructed out of polycarbonate with nylon screws in order to minimize trace metal contamination and had a total flow rate of 330 L min−1. The filters, HVDVI and laboratory equipment were acid-cleaned using ultrapure acids (Seastar Chemicals, Inc.) and 18.2 MΩ-cm Nanopure water (Barnstead). An orifice plate meter, gas meter and critical orifice were used to control and measure the flow rates through the collector. A sector sampling system was used to control the aerosol collector. The system was configured to allow collection of ambient aerosol samples only when the relative wind direction was plus or minus 75° relative to the ship's bow during both the winter and summer Atlantic cruises. New filters were loaded every 24 hours, and the sampling duration depended the time that the system was in sector.

2.2. Labile Fe Analysis

[7] Labile Fe(II), labile Fe(III) and reducible particulate Fe, were the three labile Fe species investigated using an aqueous extraction procedure and measured using LPAS immediately after sample collection. This method is briefly outlined below, and a more detailed discussion of the method is given by Chen and Siefert [2003]. Labile Fe species were operationally defined by the extraction time and reagents. The extraction time was determined by following the release of Fe over time and noting the characteristic time where Fe had reached a maximum for each step in the aqueous extraction procedure. A 47 mm diameter subsample from the 90 mm aerosol filter sample was cut for labile Fe measurements. The remaining portion of the aerosol filter sample was stored in a freezer and later subsampled and analyzed for total metals and soluble ions. The 47 mm subsample was placed in a Teflon jar and “wetted” by adding approximately ten 0.01 mL drops of spectrophotometric grade ethanol (total ≈0.1 mL) to increase the affinity between the aqueous extraction solution and the hydrophobic Teflon membrane filter. The extraction solution (50 mL of pH 4.5, 0.5 mM formate-acetate buffer solution) was then added to the jar, and the jar was covered and gently swirled. After 2 min, a 2 mL aliquot and a 1 mL aliquot were removed and transferred to 5 mL sample vials; 5 μL of 10 mM ferrozine solution was added to the sample vial with 2 mL aliquot, while the 1 mL aliquot was used for a background absorbance spectrum. The ferrozine reagent forms a colored complex with Fe(II) in solution that can be used to quantify Fe(II) using absorbance spectrometry [Stookey, 1970]. Aliquots were removed from the extraction solution every 30 min without filtration and the ferrozine-Fe(II) absorbance was measured using LPAS with a 1 m path length [Waterbury et al., 1997]. The measured Fe(II) concentration at the 90 min extraction time was defined as the labile Fe(II) since Fe(II) concentrations typically reached a maximum by 90 min in the extraction solution [Chen and Siefert, 2003]. At 90 min another two aliquots of the extraction solution were removed and hydroxylamine (HA, 50 mM) added to the aliquots with the ratio of 3.3 μL of HA per mL of extraction solution. The Fe(II) measured in the aliquot with HA included both the labile Fe(II) and labile Fe(III) concentrations, and was used to determine the labile (Fe(II) + Fe(III)) species. The labile Fe(III) concentration was then calculated by subtracting the labile Fe(II) from the labile (Fe(II) + Fe(III)) concentration at 90 min. HA was also added to the remaining extraction solution in the Teflon jar at 90 min, that was in contact with the filter subsample to dissolve labile Fe(III) particles that can undergo reductive dissolution. The unfiltered aliquots were removed for Fe(II) measurements every 30 min, and the measured Fe(II) concentration at 180 min of the extraction procedure was defined as total labile Fe. The reducible particulate Fe species was then defined by subtracting the labile (Fe(II) + Fe(III)) from the total labile Fe (“reducible particulate Fe” was named “reducible Fe” by Chen and Siefert [2003]. Chen and Siefert [2003] observed that photochemical reduction of Fe(III) species using ambient sunlight had similar Fe(II) production rates as extractions using hydroxylamine as a reducing agent.

2.3. Total Metal Analysis

[8] A strong-acid microwave digestion procedure followed by ICP-MS (HP 4500) was used to measure total elemental concentrations. The digestion procedure was run in a batch of six Teflon bombs with four filter samples, one blank and one sample of Standard Reference Material (SRM) 2709 San Joaquin Soil (U.S. Department of Commerce, National Institute of Standards and Technology) for quality control. Two grams of 10 N nitric acid (Seastar Chemical, Inc.) were added to each Teflon bomb. The microwave heating cycle was 200 W for 5 min and then 700 W for 2 min. After being cooled to the room temperature, the bombs went through a second heating cycle: 200 W for 10 min, addition of 0.1g of 28N hydrofluoric acid (Seastar Chemical, Inc.), and then 700 W for 2 min. Finally, 28 g of 18.2 MΩ-cm water was added to each bomb. The mass of each bomb was monitored to check for venting of acids in the digestion process, and the total loss was controlled within 0.03 g to minimize the effects of matrix variability. A multielement internal standard of Sc, In and Bi (SPEX CertiPrep, Inc.) was added to the sample before analysis of total elemental concentrations in the digestion solution using ICP-MS. The atmospheric concentrations were then calculated using the mass of the extraction solution, the concentration of the element, the volume of the air sampled and the fraction of the filter sample used in analysis.

2.4. Ion Analysis

[9] A subsample of the 90 mm Teflon filter sample was analyzed for anions and cations using an aqueous extraction technique [Derrick and Moyers, 1981] and ion chromatography using a Dionex DX-600 system. Anions were separated and eluted using an AS15 anion column (Dionex) using a KOH eluent in gradient mode, and cations were separated and eluted using a CS12A cation column (Dionex) using a methanesulfonic acid (MSA) eluent in gradient mode. The atmospheric concentrations were then calculated using the mass of the extraction solution, the concentration of the ion, the volume of the air sampled and the fraction of the filter sample used for the analysis. NSS-sulfate concentrations were calculated by subtracting the sulfate contribution to the aerosol due to sea salt by using the average ratio between sulfate and sodium in the seawater along with the sodium concentrations in the aerosol [Duce et al., 1983].

2.5. Air Mass Back Trajectories (AMBTs)

[10] AMBTs were calculated from the National Oceanic and Atmospheric Administration (NOAA) FNL database using the Hybrid Single-Particle Langrangian Integrated Trajectories (HY-SPLIT) program [Draxler, 2002]. AMBTs were performed at 100 m, 500 m and 1500 m height levels over the sampling position at 2000 UTC (corresponding to 7:00 PM local time that was about midway during the sample collection period) for each day of the cruises. Atmospheric aerosols do not always follow these trajectories due to scavenging processes and gravitational settling of the aerosol [Ellis and Merrill, 1995], and there are errors associated with these AMBTs due to the data sets and models. However, the AMBTs still provide useful information about the synoptic situation and general source of the air mass sampled.

3. Results and Discussion

3.1. Spatial and Seasonal Distributions of Atmospheric Fe

[11] Table 1 lists the concentrations of labile Fe(II), labile Fe(III), reducible particulate Fe, and total Fe in both the fine- and coarse-fraction aerosols along with the sampling dates and locations. The data have been placed into five categories: (1) 26°N to 30°N Atlantic region in winter (WIN26), (2) 5°N to 26°N Atlantic region in winter (WIN15), (3) 26°N to 30°N Atlantic region in summer (SUM26), (4) 6°N to 26°N Atlantic region in summer (SUM15), and (5) 0°N to 6°N Atlantic region in summer (SUM5). This categorization was done according to total Fe concentrations, locations, and dates of aerosol sample collection (see Figure 1). Total non-sea-salt-sulfate (NSS-sulfate) and oxalate concentrations are also listed in Table 1.

Figure 1.

Spatial, size, and seasonal distributions of total Fe in aerosols collected during winter (6 January 2001 to 19 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over the Atlantic Ocean. The bubble areas are proportional to the concentrations of total Fe. Note that there is a bias toward showing the “coarse total Fe” bubbles when data are located close together on the map since all of the “coarse total Fe” bubbles are placed over all of the “coarse and fine total Fe” bubbles.

Table 1. Concentrations of Labile and Total Fe, NSS-Sulfate and Oxalate in Both Fine and Coarse Fraction Aerosols Collected During the Winter (6 January 2001 to 19 February 2001) and Summer (27 June 2001 to 15 August 2001) Research Cruises Over the Atlantic Ocean
LocationDate, Julian DayLabile Fe(II), ng m−3Labile Fe(III), ng m−3Reducible Particulate Fe, ng m−3Total Fe, ng m−3NSS-Sulfate Coarse Plus Fine, nmoles m−3Oxalate Coarse Plus Fine, nmoles m−3
Latitude, NLongitude, WCoarseFineCoarseFineCoarseFineCoarseFine
  1. a

    The detection limit varied for each sample and species measured because of the detection limit of the analytical method and the air volumes collected for each aerosol filter sample.

26°N to 30°N Atlantic region in winter (WIN26)
--6------17.813.520.41.49
27.875.57------25.915.119.41.05
28.170.980.371.980.130.120.140.256.054.5820.31.21
28.466.990.320.320.420.340.080.102.84<2.55.270.4
28.663.3100.100.340.090.080.180.090.810.838.940.2
29.255.2120.050.360.040.140.070.26<0.32.442.690.13
29.551.313------0.722.111.990.15
29.948.2140.040.13<0.050.020.040.203.732.081.610.1
29.646.5150.010.05<0.040.020.030.050.1<0.41.080.1
27.845160.020.10.030.0400.090.291.671.680.15
Average  0.130.470.140.110.080.156.475.29  
 
5°N to 26°N Atlantic region in winter (WIN15)
25.345170.070.440.130.170.110.2413.921.32.190.23
21.545180.040.730.250.310.200.3618.9183.870.32
16.845190.301.130.942.300.65<0.071081867.540.27
13.245201.432.890.510.163.89<0.0624939717.81.24
10.245.2210.893.870.4215.21.790.8616172915.80.59
10.246.5220.493.310.479.381.754.991812309.920.59
10.547.8290.161.780.781.830.251.3632.987.72.840.24
9.3647.5341.254.150.280.990.582.04671165.970.59
7.4148.2353.7110.80.677.813.298.9332064714.81.62
6.3147.1360.986.881.796.274.0613.51424548.720.85
7.2245.1372.1010.71.7614.73.231.241395299.090.6
7.1743383.7910.10.785.332.035.331374186.860.57
8.6141.3392.485.44<0.0512.42.497.821283015.280.36
9.3441.5401.755.521.950.900.769.961703455.190.43
10.942.4417.9019.95.975.066.9217.236940416.91.09
10.144.7425.0212.12.6310.02.7314.33527295.020.92
9.8144.5432.6722.45.655.323.4314.4383836100.89
10.646.6446.7115.80.776.072.3012.12134845.420.82
9.4449.2455.9714.42.685.111.6913.514941514.30.79
9.0851.8463.487.800.754.961.403.39154351110.61
9.4155.3475.7712.50.8412.13.267.75500108416.60.87
10.956.1483.689.542.3512.34.977.2252111677.310.61
11.354.8498.8912.12.9716.74.347.523526074.730.5
Average  3.028.461.616.752.447.34211459  
 
26°N to 30°N Atlantic region in summer (SUM26)
29.227.4178<0.050.070.060.070.030.132.673.153.990.24
29.329.61790.020.080.040.090.060.071.792.743.720.31
29.433.5180<0.030.060.020.100.050.084.882.073.30.28
29.537.41810.090.110.070.080.170.1066.42.43.850.38
29.539.31820.040.250.200.390.070.0779.123.26.910.29
29.643.21830.010.240.060.400.020.037.178.536.810.25
29.6451840.170.740.842.281.711.5415480.510.20.48
Average  0.070.220.180.490.300.2945.128.8  
 
6°N to 26°N Atlantic region in summer (SUM15)
25.548.61850.040.780.531.730.621.0913770.35.10.26
22.651.3186<0.030.560.171.530.350.8262.959.14.570.46
16.356.81870.050.940.461.960.481.211451021.820.24
--188------1141155.330.31
11.854.4190<0.050.530.431.570.560.9861.779.71.410.17
11.854.41900.080.270.430.990.570.8177.491.44.350.23
10.448.11910.411.541.486.881.222.873563208.510.52
10.448.11920.081.020.713.150.992.401171994.050.34
10.448.1193-0.33-0.85-1.1481.110811.80.59
9.845.31940.17-0.85-0.65-1952236.410.29
10.145.41950.090.630.893.280.721.632662567.90.32
10.245.51960.090.660.572.400.961.852832056.420.4
1149.31970.221.361.455.650.321.585583046.960.49
--198------2422715.10.32
11.658.22000.010.701.086.010.801.972632294.670.6
10.356.32010.260.490.802.970.981.2614792.83.940.68
10.256.32020.091.151.123.690.972.201381432.330.55
10.256.32030.20-0.68-0.93-61.21063.560.71
11.954.92070.020.500.342.340.661.7495.71042.370.25
10.4532080.100.810.583.241.313.352492506.620.31
8.74512090.020.600.411.600.553.1667.31374.020.33
7.2348.5210<0.030.200.170.480.170.272022.44.380.93
8.2152.82190.100.820.501.990.641.411471203.660.4
10.555221<0.031.340.603.680.892.221572234.750.13
10.655.8222<0.041.440.461.540.842.321261253.240.3
12.555.12230.081.420.317.181.552.452903847.010.3
12.554.12240.260.941.194.110.972.54240194<0.10.01
11.453.8225<0.051.610.282.511.021.4848.290.31.540.28
11.854.62260.050.180.110.330.170.052315.74.420.95
Average  0.120.830.642.870.761.71164160  
 
0°N to 6°N Atlantic region in summer (SUM5)
5.6546.42110.020.160.140.320.160.4723.719.82.560.45
4.7643.9212<0.030.0300.050.030.060.752.181.120.08
3.8342.82130.010.260.050.060.060.456.3712.2<0.1<0.1
3.2744.22140.050.070.060.030.140.198.643.1620.24
3.9346.1215<0.070.180.010.080.160.2411.30.391.750.24
5.79482160.110.220.040.100.040.359.29.61<0.20.27
6.1750.22170.030.130.070.130.130.1610.15.862.10.46
Average  0.040.150.050.110.100.28107.6  

[12] Total Fe concentrations in the fine- and coarse-fraction aerosols were extremely low near latitude 30°N during the winter (Figure 1). A spatial gradient of over nearly 3 orders of magnitude in the total Fe concentrations (from 1.6 ng m−3 at 28.6°N to 1688 ng m−3 at 10.9°N) was observed during the winter, while this gradient was not as strong in the summer. A similar spatial gradient, although not as large in magnitude as aerosol Fe, has been observed for the dissolved Fe concentrations in Atlantic surface seawater [Wu and Boyle, 2002; Sanudo-Wilhelmy et al., 2001]. The observed Fe concentrations were 0.2 nM at latitude 31°N, 0.8 nM at 26°N, and 0.77 nM at 10°N to 16°N, respectively. It was suggested that the gradient distribution of surface seawater Fe is mostly attributed to the latitudinal variation of atmospheric Fe deposition that is evident in the winter data (Figure 1). However, the latitudinal gradient of atmospheric total Fe concentrations over this oceanic region was not as significant during the summer. One reason for this decreased latitudinal gradient during the summer is that the maximum zone of African dust plume migrated to latitude 20°N with the ITCZ [Husar et al., 1997; Moulin et al., 1997], which extended the dust impacts to higher latitudes (up to 30°N). The total Fe concentration (fine and coarse) reached 235 ng m−3 at the position 29.6°N 45°W, while the averaged Fe concentrations in SUM15 were 324 ng m−3 during the summer (Figure 1). Low concentrations of atmospheric total Fe (mean 17.6 ng m−3) were observed at lower latitudes where the influence of mineral dust plumes from North Africa was weaker due to the northward migration of the ITCZ in summer. The low Fe concentrations and weaker latitudinal gradient may also be caused by the episodic nature of dust plumes. Overall, the atmospheric total Fe concentrations in the MBL were approximately a factor of 2 higher during the winter than in the summer when comparing the mean total Fe concentrations in the maximum dust plume zone (mean 670 ng m−3 in WIN15 and 324 ng m−3 in SUM15).

[13] Long-term measurements of mineral dust concentrations have been done at Sal Island (16°45′N, 22°57′W) located in the zone of maximum dust transport. These measurements showed a pronounced seasonal pattern with the maximum dust concentrations during winter [Chiapello et al., 1995]. However, Gao et al. [2001] indicated that the highest Fe flux including wet deposition to the tropical and subtropical Atlantic Ocean occurred in the summer using their model. At Barbados, Gao et al. [2001] predicted the average Fe flux to be about 45 mg m−2 month−1 in the summer and significantly higher than the average Fe flux in the winter (∼20 mg m−2 month−1). This different seasonal pattern at Barbados is also probably a consequence of the long distance between North Africa and Barbados (it takes about 1 week for a dust event to travel from North Africa to the Caribbean). This travel time weakens the impact of the African mineral dust, especially in winter when a low-level dust transport replaces a high-level dust layer in summer [Guelle et al., 2000]. Higher concentrations of atmospheric total Fe (mean 900 ng m−3) were also observed for a couple days during the summer cruise when the ship was close to Barbados, where other atmospheric Fe sources (e.g., crustal source from South America) cannot be excluded. The winter versus summer trends observed for the aerosol Fe during these two cruises highlights the strong gradients and episodic nature of the dust events and does not apply in general for seasonal variations since this data are limited to certain regions during 1 year.

3.2. Labile Fe Features

[14] Three labile Fe species in the aerosol samples were measured using an aqueous extraction procedure and LPAS. Most of the total labile Fe concentrations (mean 82%) were found in the fine-fraction aerosol (see Figure 2 and Table 1). However, the fine fraction only contributed half (mean 50%) of the total Fe concentrations, showing the “enrichment” of the labile Fe species in the fine fraction compared to the coarse aerosols. This has previously been observed in aerosol samples collected during the intermonsoon season over the Arabian Sea by Siefert et al. [1999] where more than 80% of total atmospheric aqueous-labile Fe(II) concentrations were present in the fine fraction, and over the tropical Atlantic Ocean [Johansen et al., 2000]. Siefert et al. [1999] indicated that fine and coarse fractions of aerosol particles might have different origins, and over the Arabian Sea, pollution sources (e.g., combustion of fossil fuels) may have contributed to the fine aerosols with high labile Fe(II) concentrations. However, differences in the atmospheric processing of the fine and coarse aerosol may also explain these observations. Most atmospheric Fe in this Atlantic region is carried by the mineral dust plumes originating from North Africa, and this long-range transport will favor the suspension of fine particles over larger particles due to the greater settling velocity of larger particles. The fine particles will therefore have more opportunities to undergo condensation-evaporation cycles and be exposed to reductive processes (e.g., photochemical redox reactions) and acidic pH conditions that can transform refractory Fe species into more labile Fe species [Chen and Siefert, 2003]. Fine particles also have a higher surface area to volume ratio than coarse particles, which may enhance their aqueous dissolution.

Figure 2.

Spatial, size, and seasonal distributions of labile Fe in aerosols collected during winter (6 January 2001 to 19 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over the Atlantic Ocean. The bubble areas are proportional to the concentrations of total labile Fe. Note that there is a bias toward showing the “coarse total labile Fe” bubbles when data are located close together on the map since all of the “coarse total labile Fe” bubbles are placed over all of the “coarse and fine total labile Fe” bubbles.

[15] Higher percentages of total labile Fe to total Fe ratios were observed in the fine-fraction aerosol than in the coarse fraction for most of the samples except during in WIN26 where the percentages of total labile Fe to total Fe in both the coarse and fine fractions became extremely large (mean 33% and 37%, respectively) and not significantly different (see Figure 3). For labile Fe(II) species (see Table 1), fine-fraction aerosols had even higher percentages than the coarse fraction, including the WIN26 where the labile Fe(II) percentage to total Fe in the fine fraction (mean 19%) was about twice as much as that in the coarse fraction (mean 8.6%), accounting for most of the total labile Fe (∼58%) observed in fine-fraction aerosols. This large contribution of labile Fe(II) to labile Fe species (58% of total labile Fe) in the fine fraction in WIN26 suggest a longer atmospheric transportation time for the fine aerosols, or possibly a distinctive air mass source in this Atlantic region during this time period. These fine aerosols may be associated with high concentrations of organic matter that could stabilize the labile Fe(II) produced by photochemical reduction process.

Figure 3.

Spatial, size, and seasonal distributions of the ratio of labile Fe to total Fe concentrations for aerosols collected during winter (6 January 2001 to 19 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over the Atlantic Ocean. The ratio in all cases is to the sum of the coarse and fine total Fe concentrations. Note that there is a bias toward showing the “coarse total labile Fe to total Fe” bubbles when data are located close together on the map since all of the “coarse total labile Fe to total Fe” bubbles are placed over all of the “coarse and fine total labile Fe to total Fe” bubbles.

[16] Total Fe concentrations were strongly correlated with the concentrations of total Mn and total Al in aerosols collected during both cruises with the exception of the fine-fraction aerosols in WIN26 (Table 2). Aluminum is typically used as the tracer to quantify the mineral aerosol abundance in the atmosphere [Taylor and McLennan, 1985]. Crustal enrichment factors (EFs) based on X/Al ratios have been widely used to identify contribution of crustal and noncrustal sources on observed concentrations of trace elements [Kaya and Tuncel, 1997; Al-Momani et al., 1998; Chester et al., 1993a; Yatin et al., 2000; Huang et al., 2001]. High correlation coefficients between the total Fe and total Al (Table 2) suggested that the mineral dust transported from the Sahel region was the dominant Fe source over the tropical and subtropical North Atlantic Ocean. However, total Fe measured in fine aerosols in WIN26 showed a weak correlation with Al (0.386) but large correlation coefficients with Cu (0.755), Ni (0.809), and V (0.817), which indicated that the aerosol Fe was probably influenced strongly by the noncrustal anthropogenic sources from the North America or Europe. The second Aerosol Characterization Experiment (ACE-2) showed that both North American and European urban/industrial sources contributed to the aerosols over the North Atlantic region, with North American sources dominating under conditions of a strong Azores high [Benkovitz et al., 2003]. The northeastern United States burns residual oil in winter, and some of the aerosol associated with this oil combustion may be transported to the subtropical North Atlantic region [Huang et al., 2001]. Noncrustal V in the atmosphere is most often associated with the combustion of heavy fuel oil [Rahn and Lowenthal, 1984; Yatin et al., 2000]. High loadings of Cu at Mumbai, India, were generally from nonferrous industrial emissions, and Ni may be from pollution sources such as oil and refuse burning [Venkataraman et al., 2002]. The Fe observed in the fine aerosols during this specific time and region may be caused by wearing of metals used in motor vehicles [Yatin et al., 2000] or ferrous industries (e.g., metallurgic plants, steel mills, castings) in North America.

Table 2. Correlation Matrix Between the Concentrations of Trace Elements and Total Fe in Both Coarse- and Fine-Fraction Aerosols Collected During Winter (6 January 2001 to 19 February 2001) and Summer (27 June to 15 August 2001) Over the Tropical and Subtropical North Atlantic Oceana
ElementsCoarse-Fraction AerosolsFine-Fraction Aerosols
WIN26WIN15SUM15SUM5WIN26WIN15SUM15SUM5
  • a

    Definitions are as follows: WIN26, 26°N to 30°N Atlantic region in winter; WIN15, 5°N to 26°N Atlantic region in winter; SUM26, 26°N to 30°N Atlantic region in summer; SUM5, 0°N to 6°N Atlantic region in summer.

  • b

    The three largest values (i.e., close to 1) are indicated.

Al0.726b0.989b0.956b0.882b0.3860.994b0.954b0.972b
Ca0.1790.8920.9420.2260.5030.8840.908b−0.096
K0.1550.915b0.8340.4680.7150.931b0.728−0.042
Na0.372−0.0440.5510.0990.5150.2520.429−0.358
Mg0.0720.5280.5750.0950.4150.8810.460−0.344
Cr0.2940.6240.8150.624−0.0800.9170.776−0.031
Co_0.3020.1620.378_0.7060.416_
Cu0.755b−0.1740.670−0.1760.773−0.0550.6370.044
Pb0.0370.4930.6870.5140.7120.7110.335−0.077
Mn0.763b0.961b0.998b0.892b0.844b0.986b0.994b0.950b
Ni0.243−0.1790.8400.4630.809b0.0550.748−0.084
V0.5970.5300.971b0.766b0.817b0.9270.863−0.124
Zn0.0380.5860.7820.0990.7760.6490.670−0.261
Fe11111111

[17] An increase in labile Fe to total Fe ratio generally corresponds to a decrease in total Fe concentration (Figure 4). In WIN26 the percentage of labile Fe to total Fe (∼35%) was approximately 7 factors higher than that in WIN15 (∼5.0%) when an opposite spatial gradient of atmospheric total Fe concentrations was observed (Figure 3). The percentage of labile Fe to total Fe in SUM26 was around 5.7%, only slightly higher than that in SUM15 ∼2.6%, which corresponded to the summer pattern of total Fe concentrations in the atmosphere with a weaker spatial gradient (Figure 1). In addition, SUM5 had the lowest concentrations of atmospheric total Fe observed; however, the labile Fe to total Fe ratio (∼5.0%) was still close to the northern regions. Therefore a low concentration of atmospheric total Fe would not always imply a high percentage of labile Fe fractions. Moreover, the correlation coefficients between the Fe and Al concentrations in SUM5 were high for both fine- (0.972) and coarse- (0.882) fraction aerosols (Table 2), indicating a crustal source origin for Fe (mineral dust).

Figure 4.

Ratio of labile Fe to total Fe versus total Fe concentrations in aerosols collected during the winter (6 January 2001 to 19 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over the tropical and subtropical North Atlantic Oceans.

[18] Ion concentrations in the aerosol samples showed that WIN26 had much higher ratios of oxalate and NSS-sulfate anions to total Fe (oxalate/Fe and NSS-sulfate/Fe mean ratios are 4.8 and 88, respectively) than in SUM5 (oxalate/Fe and NSS-sulfate/Fe mean ratios are 1.1 and 10, respectively) (Figure 5). These species along with condensation-evaporation cycles in clouds and photochemical redox processes may produce the high labile Fe fractions observed in these aerosol samples. The source of the Fe may also be noncrustal as indicated by the weak correlation with Al (Table 2). Oxalate is the final product of photochemically induced reactions involving many organic precursors [Kawamura and Ikushima, 1993], and it is known to be an efficient electron donor for the photochemical reduction of Fe(III) in atmospheric waters [Zuo and Hoigne, 1992]. Sulfate may indicate a lower aerosol pH which could increase the stability of Fe(II) with respect to oxidation [Johansen et al., 2000] and also increase the solubility of Fe minerals. Thus the air mass characterized by the highest percentage of labile Fe to total Fe in WIN26 may have a different source contribution (polluted air masses from North America or Europe) from the others characterized by much lower labile Fe fractions.

Figure 5.

Ratios of NSS-sulfate to total Fe and oxalate to total Fe versus the ratio of labile Fe to total Fe in aerosol samples collected during the winter (6 January 2001 to 19 February 2001) and summer (27 June 2001 to 15 August 2001) research cruises over the tropical and subtropical North Atlantic Oceans.

[19] Previous studies have investigated the percentage of labile Fe(II) species in total Fe concentrations in aerosols using various extractions techniques. The labile Fe(II) results in this study are comparable to these previous observations. Siefert et al. [1999] reported that never more than 4% of the total Fe was released as Fe(II) after 22 hours of extraction for aerosol samples collected over Arabian Sea, and Fe(II)labile/Fetotal fractions in Barbados aerosol samples were observed between 0.47 and 0.92% [Zhu et al., 1993], which are consistent with our measurements that labile Fe(II) species that were approximately 1.8%, 1.0%, 0.33%, and 1.1% of total Fe concentrations in WIN15, SUM26, SUM15, and SUM5, respectively. However, much higher percentages of labile Fe(II) to total Fe (10–100% and 2.2–49%) have been observed in marine aerosols over the central North Pacific Ocean and Barbados [Zhuang et al., 1992], which was explained as a result of increased cloud processing of the aerosols. In this study, a high percentage of labile Fe(II) to total Fe (∼16%) was also observed in WIN26, which may be attributed to different sources (e.g., anthropogenic sources from North America or Europe) of the aerosol particles.

3.3. Air Mass Back Trajectories (AMBTs)

[20] AMBTs were calculated using the HY-SPLIT (isentropic program) for a 7-day period. Figure 6 shows seven representative AMBTs for this data set. The AMBTs showed the dominant northeasterly trade winds transporting mineral dust from North Africa over the 0°N to 30°N Atlantic region. However, at both ends of this region the air masses were seasonally affected by other circulations as a result of latitudinal shifting of the ITCZ, which effected the variations of total Fe concentrations and labile Fe fractions.

Figure 6.

Representative 7 day air mass back trajectories for starting altitudes of 100 m, 500 m, and 1500 m above ground level (AGL) calculated from the National Oceanic and Atmospheric Administration's FNL database using the Hybrid Single-Particle Langrangian Integrated Trajectory (HY-SPLIT) model (markers are at 6 hour increments): (a) 16 January 2001 (28°N, 45°W), 2000 UTC; (b) 18 January 2001 (21°30′N, 45°W), 2000 UTC; (c) 22 January 2001 (10°N, 46°30′W), 2000 UTC; (d) 4 July 2001 (25°30′N, 48°30′W), 2000 UTC; (e) 15 July 2001 (10°N, 45°30′W), 2000 UTC; and (f) 31 July 2001 (5°N, 44°W), 2000 UTC.

[21] The spatial gradient of increased total Fe concentrations from 30°N to 10°N in the winter indicated a stronger impact of African dust in this region. The AMBTs for 30°N to 10°N in the winter (Figures 6b and 6c) were consistent with these higher concentrations and showed the air masses to have either passed over North Africa or to have come close to the coast. Back trajectories at 28°N 45°W (where the ship was located on 16 January 2001) showed that the air mass had circulated over the ocean for more than 7 days without contact with the land (Figure 6a), which was consistent with an extremely low concentration of atmospheric total Fe (see discussion in section 3.2).

[22] Chiapello et al. [1997] labeled samples as “non-dusty samples” of oceanic origin when the AMBTs had circulated over the ocean for more than 5 days. Many of the AMBTs in this study had circulated over the ocean for more than 7 days; therefore the oceanic aerosols in this study were aerosols that not only had circulated over the ocean for more than 7 days but also were not near a land mass after 7 days (e.g., the 100 m and 500 m and 1500 m AMBTs in Figure 6a). The oceanic origin aerosols in this study (e.g., at 28°N 45°W) contained the highest labile Fe fractions. This high labile Fe fraction may be because the particles had more time to undergo photochemical processing than the fresh terrestrial dust or that the aerosol Fe had a different source. Almost equal percentages of labile Fe to total Fe were observed in both the fine- and coarse-fraction aerosols (labile Fe fraction is dominant in fine fraction of mineral dust, see discussion 3.2), which may indicate an impact of anthropogenic Fe source (e.g., ferrous-industries from North America or Europe) on these oceanic origin aerosols at 28°N (Figure 6a). The two lower back trajectories at latitude 21°30′N extended along the North African coast and corresponded to a high concentration of total Fe (Figure 6b). The back trajectories at 10°N passed over North Africa (Figure 6c) and had even higher Fe concentrations. These two locations had similar African dust origins and similar labile Fe fractions. The maximum concentrations of atmospheric total Fe observed in WIN15 were probably due to an additional lower layer transport (below 1.5 to 3 km in altitude) of African dust [Chiapello et al., 1995].

[23] Dust transport off North Africa during summer occurs at high altitudes, from about 1.5 km to 5–7 km above sea level, which allows for long-range transport and influences a larger North Atlantic region [Prospero and Carlson, 1972]. The 1500 m back trajectories at latitude 25°30′N and 10°N (Figure 6d) passed over North Africa, while the two lower back trajectories circulated over the ocean at 25°30′N or extended along the North African coast at 10°N (Figure 6e). This was consistent with the weaker spatial gradient of atmospheric total Fe in the summer and the lower Fe concentrations in the MBL during the winter. Back trajectories at 5°N 44°W during the summer showed that the air mass was from the South Atlantic Ocean (Figure 6f), which indicated that this region was not significantly impacted by African dust, corresponding to the low concentrations of atmospheric total Fe.

3.4. Atmospheric Dry Deposition of Fe

[24] The dry deposition velocity is primarily a function of dust particle size, wind speed, and relative humidity. Previous studies have found that the mass median diameters (MMD) of aerosols collected over the tropical and subtropical North Atlantic Ocean were between one to several micrometers [Prospero, 1995]. This study shows that labile Fe is a function of particle size and is typically enriched in the fine aerosol compared to total Fe, and therefore it is expected that these two size fractions would have different deposition velocities and other microphysical properties. However, it is somewhat arbitrary to use different deposition velocities for the fine- and coarse-fraction aerosols investigated in this study since more information is needed about the aerosol microphysical properties (e.g., MMD). Slinn and Slinn [1981] showed that particles with 1 to 10 μm radius have a dry deposition velocity around 1.0 cm s−1 under the condition of 5 m s−1 wind speed and 100% relative humidity. The mean deposition velocity of 1.0 cm s−1 was also suggested by comparison between dust deposition estimates and sediment trap records [Jickells, 1999]. Therefore Fe dry deposition rates over the tropical and subtropical North Atlantic Oceans were calculated by multiplying Fe concentrations in the aerosols with the dry deposition velocity of 1.0 cm s−1. Fe dry deposition rates were calculated for four subregions WIN26, WIN15, SUM15, and SUM5 according to the spatial and seasonal distributions of atmospheric Fe concentrations (see discussion in section 3.2). The dry deposition rates of the various Fe species were calculated. The three labile Fe fractions measured in this study represent Fe species with varying abilities to dissolve in aqueous solutions. The order from most labile to least labile for these fractions would be (1) labile Fe(II), (2) labile Fe(III), (3) reducible particulate Fe, and (4) the refractory Fe pool determined by strong acid digestion. The bioavailability of these Fe fractions would be expected to follow this same order. The reducible particulate Fe enlarges the bioavailable Fe pool by considering Fe reduction processes in the atmosphere or seawater [Chen and Siefert, 2003]. The maximum dry deposition rates of total Fe (631 μg m−2 d−1) and labile Fe occurred in WIN15 (Table 3), which is comparable to the Fe dry and wet deposition rates (∼20 mg m−2 month−1 or 667 μg m−2 d−1) at Barbados calculated by Gao et al. [2001]. The dry deposition of total Fe in SUM15 was approximately 278 μg m−2 d−1, about half of the maximum dry deposition in the winter. Correspondingly, labile Fe dry deposition decreased even further (Table 3). Low dry deposition rates of total Fe were calculated in WIN26 and SUM5 with the former region even lower, approximately 2.5 and 15 μg m−2 d−1, respectively. However, the dry deposition rates of labile Fe species were similar between these two regions (Table 3), indicating that the atmospheric dry depositions may provide almost equal amount of bioavailable Fe to these two oceanic regions.

Table 3. Atmospheric Dry Depositions of Total Fe and Labile Fe Species Over the Tropical and Subtropical North Atlantic Oceans During the Winter (6 January 2001 to 19 February 2001) and Summer (27 June 2001 to 15 August 2001)a
 WIN26WIN15SUM15SUM5
  • a

    Distributions are given as mean plus/minus SD in μg m−2 d−1.

Total Fe2.5 ± 1.44631.1 ± 355.48278.4 ± 170.8515.2 ± 10.93
Labile Fe(II)0.5 ± 0.699.9 ± 7.150.7 ± 0.430.2 ± 0.08
Labile Fe (III)0.2 ± 0.227.2 ± 4.862.8 ± 1.930.1 ± 0.12
Reducible particulate Fe0.2 ± 0.107.9 ± 5.752.0 ± 1.030.3 ± 0.15

4. Conclusions

[25] Field measurements of Fe concentrations and speciation in aerosols were conducted during the winter and summer cruises over the tropical and subtropical North Atlantic Ocean. A spatial gradient was observed of nearly 3 orders of magnitude in the total Fe concentrations, from a low concentration of 1.6 ng m−3 in WIN26 to a high concentration of greater than 1688 ng m−3 in WIN15, while this gradient was not as significant in the summer due to the migration of the ITCZ. The atmospheric total Fe concentrations in the MBL were approximately a factor of 2 higher during the winter (mean value 670 ng m−3 in WIN15) than in the summer (mean value 324 ng m−3 in SUM15). The highest percentage of labile Fe to total Fe (35%) was observed in WIN26, corresponding to low concentrations of total Fe and relatively high concentrations of oxalate and nss-sulfate. However, in SUM5, where the lowest Fe concentrations were measured, the labile Fe fraction (5.0%) was similar to SUM15. AMBTs showed that mineral dust transport off North Africa is a dominant Fe source in this region. However, labile Fe appears to be influenced by anthropogenic activity in some of the winter samples (around 30°N latitude) where anthropogenic metals (V, Cu, Ni) were relatively high along with other species (i.e., oxalate, nss-sulfate) consistent with anthropogenic sources. The highest calculated dry deposition fluxes of total Fe and labile Fe occurred in WIN15, whereas the lowest fluxes were shown in both WIN26 and SUM5. The highest dry deposition fluxes of labile Fe occurred in the region of high N2-fixing diazotrophic activity in the tropical North Atlantic region (5°N to 25°N, oligotrophic ocean where diazotrophs are favored). A strong spatial gradient of Trichodesmium biomass (7 times higher at 10°N to 16°N Atlantic than at 0°N to 6°N Atlantic region) observed in April 1996 [Sanudo-Wilhelmy et al., 2001] may be related to the spatial pattern of atmospheric deposition of labile Fe, instead of the dissolved Fe concentrations in surface seawater. Trichodesmium blooms in offshore waters of the west Florida shelf [Lenes et al., 2001] in summer may also relate to the high atmospheric deposition of labile Fe that extended northward in summer.

Acknowledgments

[26] This material is based upon work supported by the National Science Foundation: Ocean Sciences Division - Biological Oceanography under grants OCE-998 1218 and OCE-998-1252 (Biocomplexity: Phase I). We also appreciate the thoughtful comments provided by the reviewers.

Ancillary