Volume 30, Issue 24
Oceans
Free Access

Atmospheric deposition of nutrients to the Atlantic Ocean

A. R. Baker

A. R. Baker

School of Environmental Sciences, University of East Anglia, Norwich, UK

Search for more papers by this author
S. D. Kelly

S. D. Kelly

School of Environmental Sciences, University of East Anglia, Norwich, UK

Search for more papers by this author
K. F. Biswas

K. F. Biswas

School of Environmental Sciences, University of East Anglia, Norwich, UK

Search for more papers by this author
M. Witt

M. Witt

School of Environmental Sciences, University of East Anglia, Norwich, UK

Search for more papers by this author
T. D. Jickells

T. D. Jickells

School of Environmental Sciences, University of East Anglia, Norwich, UK

Search for more papers by this author
First published: 31 December 2003
Citations: 7

Abstract

[1] The role of atmospheric deposition of iron, nitrogen and phosphorus in supplying nutrients to marine systems has been described, individually, in previous works. Here we examine atmospheric dry deposition of all these nutrients simultaneously, using samples collected during two meridional transects of the Atlantic Ocean. We find that, in line with previous work, desert dust supplies excess iron to the water column. However, primary production promoted by aerosol nitrogen can be sufficient to consume all of the soluble aerosol iron input in some situations. Aerosol N:P is universally very high, so that aerosol is always deficient in P relative to phytoplankton requirements. Nitrogen fixation stimulated by any excess atmospheric iron supply and phytoplankton utilisation of atmospheric nutrient inputs will therefore tend to drive the ecosystem towards P limitation. This emphasises the need to study the biogeochemical impact of atmospheric nutrient deposition in an integrated manner.

1. Introduction

[2] Deposition of aerosol from the atmosphere can be an important source of nutrients to marine systems. Much attention has focussed on the deposition of iron, particularly from desert dust [Jickells and Spokes, 2001], because iron's low solubility in seawater makes atmospheric transport the dominant source of Fe to the remote ocean [Jickells and Spokes, 2001]. This may be of particular importance in High Nutrient Low Chlorophyll areas, where Fe is thought to limit primary production [Martin et al., 1994]. Nitrogen deposition can significantly contribute to eutrophication problems in coastal areas [Paerl, 1997; Spokes et al., 2000], but there are very few studies of atmospheric phosphorus supply to the ocean [Graham and Duce, 1982; Ridame and Guieu, 2002]. These studies have concentrated on atmospheric inputs of individual nutrients. Here however, we examine dry deposition of Fe, N and P along two meridional transects of the Atlantic Ocean because we believe that the impact of atmospheric nutrient supply is dependent on the relative rate at which nutrients are supplied with respect to plankton requirements.

2. Methods

[3] Aerosol samples were collected during cruises of RV Polarstern (ANT18-1, Bremerhaven to Cape Town, October 2000) and RRS James Clark Ross (JCR, Grimsby to Falkland Islands, September/October 2001) (Figure 1). Two high volume (1 m3 min−1) collectors were used on each cruise, one equipped with acid-washed substrates for trace metal analysis, the other with unwashed substrates for major ion analysis. Trace metal samples were collected using 3-stage cascade impactors (to separate aerosol particles at a radius of 1 μm) onto, normally, cellulose (Whatman 41) substrates, although during ANT18-1 the upper stages contained quartz fibre substrates [Sarthou et al., 2003]. Major ion collection for ANT18-1 used a single Whatman 41 paper only, while for JCR a 3-stage cascade impactor and Whatman 41 papers were used. Whatman 41 papers are known to also collect gas phase nitric acid and, since this deposits to the ocean faster than particulate nitrate, we may underestimate the atmospheric N deposition flux. However, the concentrations of nitric acid in the remote marine boundary layer are low and nitric acid deposition is suppressed in the presence of seasalt aerosol [Pryor and Sorensen, 2002], so that the effect of this artifact is unlikely to be large. Samples were generally changed once each day, average air volume sampled ∼1400 m3.

Details are in the caption following the image
Aerosol sampling start positions during ANT18-1 (grey circles) and JCR (white circles) cruises. Black markers indicate end of sampling.

[4] Methods for soluble Fe [Sarthou et al., 2003] and nitrogen species (total soluble N [Cape et al., 2001], NO3 [Spokes et al., 2000], NH4+ [Jickells et al., 2003]) analysis are described elsewhere. Soluble phosphorus was determined spectrophotometrically [Parsons et al., 1984] after extraction in ultrapure water buffered at pH 7, as preliminary experiments indicated that the fraction of phosphate extracted was pH dependent and the extraction pH varied in unbuffered samples. Aerosol Fe and P are only sparingly soluble in seawater [Jickells and Spokes, 2001; Ridame and Guieu, 2002], so that only a small fraction of their total aerosol inputs are bioavailable. Their relative solubilities estimated using the methods above will be discussed elsewhere [A. R. Baker et al., in preparation, 2003]. Preliminary results indicate that our Fe solubility estimate is of the order of 3–4%, similar to the 1–2% value proposed for the overall solubility of aerosol Fe deposited in seawater [Jickells and Spokes, 2001]. Our estimates of soluble (and excess) Fe deposition and nitrogen fixation potential (below) are therefore upper limits.

[5] Dry deposition rates were calculated from atmospheric concentrations using deposition velocities of 0.02 m s−1 for coarse (>1 μm) and 0.001 m s−1 for fine (<1 μm) particles [Duce et al., 1991]. Deposition velocities are poorly constrained and dependent on particle diameter [see Duce et al., 1991], so that our flux estimates have an inherent uncertainty of around a factor of 3 (error bars in Figure 2 indicate analytical uncertainty for each component). In order to estimate deposition fluxes for N and P during ANT18-1 where size-segregated data were not available, the distributions of NO3 and PO43− between the coarse and fine aerosol fractions were estimated to be 85% coarse (NO3) and 40% coarse (PO43−), using our results from the JCR cruise and those of other workers [Johansen et al., 2000; Spokes et al., 2000]. Again using our results from the JCR cruise, nitrate comprises only ∼40% of total soluble N concentrations in aerosol, but contributes ∼80% to the total dry N deposition because it occurs primarily on the coarse mode aerosol, while NH4+ and organic N are associated with fine mode aerosol which deposits much more slowly. We have therefore assumed that NO3 deposition for ANT18-1 (Figure 2) represents 80% of the total N deposition during that cruise and have multiplied the ANT18-1 NO3 data by 1.25 to calculate N:P and N:Fe (Figure 3). We have calculated the quantity of Fe removed by non-diazotrophic production stimulated by aerosol N input (FeR) by:
urn:x-wiley:00948276:media:grl17571:grl17571-math-0001
and nitrogen fixation potential (Nfix) by:
urn:x-wiley:00948276:media:grl17571:grl17571-math-0002
where Naero and Feaero are the nitrogen and iron deposition fluxes and (Fe/N)phyto and (Fe/N)fixers are the cellular Fe:N ratios of non-diaztrophic phytoplankton and N fixing organisms respectively. The values selected for these ratios are discussed below.
Details are in the caption following the image
Dry deposition fluxes of soluble (a) Fe, (b) N and (c) P along meridional transects of the Atlantic Ocean in 2000 (solid symbols) and 2001 (open symbols). N deposition in 2000 is shown for NO3 (the only parameter available), for 2001 total soluble N deposition is shown. Note split scale of x-axis in (a).
Details are in the caption following the image
Nutrient ratios in dry aerosol deposition; (a) N:Fe, (b) N:P and (c) P:Fe. Solid symbols 2000, open symbols 2001. Dashed lines indicate Redfieldian nutrient ratios.

[6] Air mass back trajectory analysis (NOAA Air Resources Laboratory, HYSPLIT model, FNL data set) and aerosol major ion chemistry were used to assess aerosol sources.

3. Results

[7] We first consider the atmospheric deposition of soluble Fe, N and P along our transects. Soluble Fe deposition was dominated by input from the Sahara desert in 2000 (Figure 2a). The coarse fractions of samples collected between 25°N and 8°N in both years showed the red/brown colouration indicative of desert dust, and dust was also visible in a sample collected at ∼23°S in 2001. In this latter case, air mass back trajectories indicate that the dust originated in Patagonia. The air masses that transported desert dust in 2000 also contained high concentrations of NO3 (Figure 2b), the origin of which is probably anthropogenic pollution from Europe or North Africa [Savoie et al., 1989]. In 2001 the soluble Fe inputs associated with the Saharan plume were much weaker than in 2000 reflecting the highly episodic nature of dust transportation events [Jickells and Spokes, 2001]. Aerosol N deposition was generally higher in the relatively polluted northern hemisphere, with low deposition rates observed in the remote South Atlantic. In the South Atlantic, relatively high deposition rates of Fe, N (NO3) and P were associated with air mass arrivals from southern Africa at ∼10°S. Soluble P deposition was highest in the tropics and just offshore of South America (31°–35°S).

[8] We now consider the impact of this deposition on water column net primary production. In our discussion we consider only that fraction of new production [Dugdale and Goering, 1967] that can be directly attributed to the aerosol nutrient input and exclude new production driven exclusively by nutrients from other sources (e.g., upwelling). Aerosol N inputs will be utilised by all primary producers, including N2 fixing species [Mulholland et al., 1999], so that one effect of atmospheric N inputs may be to suppress apparent N2 fixation rates. The amount of new production supported solely by aerosol nutrient input is dependent on the component having the lowest concentration with respect to Redfield ratios. In all cases this was P, as the aerosol input N:P was considerably higher than the Redfield value of 16 (Figure 3b). Once the aerosol P input is consumed, further aerosol-stimulated primary production may occur if sufficient P is present in the water column. In most of the samples this further production would result in the complete removal of the aerosol N input, so that ultimately some fraction of the original soluble Fe input remains in excess. The extent of Fe removal will be dependent on the cellular Fe:N quota of algal growth (equation 1). This quota lies in the range 13–86 μmol mol−1 depending on the Fe status of the water column [Sunda, 1997]. The upper end of this range corresponds to Fe replete waters, and we have used 86 μmol mol−1 in our calculations since dissolved Fe concentrations in the water column along our transects are unlikely to limit primary production [Bowie et al., 2002; Sarthou et al., 2003]. In most cases the fraction of aerosol Fe input removed through aerosol N-stimulated growth (FeR/Feaero) was less than 20%. However in a few instances (Figure 4a) with low Fe inputs and relatively high N inputs, this fraction exceeded 80% and the excess atmospheric Fe input was essentially zero. It is interesting to note that although Saharan input in 2000 is the dominant feature of aerosol Fe deposition along our transects, the influence of this Saharan dust input on N:Fe ratio is rather small when compared to meridional variations in N:Fe.

Details are in the caption following the image
(a) Percentage of soluble aerosol Fe deposition removed by primary production stimulated by aerosol N deposition and (b) nitrogen fixation potential calculated from dry aerosol input of excess soluble Fe. Solid symbols 2000, open symbols 2001. Note split scale of x-axis in (b).

[9] In tropical N-limited water columns the excess Fe input may stimulate N2 fixation by diazotrophs [Paerl et al., 1994; Rueter et al., 1992]. We have calculated the maximum potential N2 fixation (Nfix, equation 2) arising from the observed atmospheric deposition within the latitude range 40°N–40°S (Figure 4b) by assuming that all of the excess atmospheric input of soluble Fe is available to N2 fixers and using the cellular Fe:C and C:N ratios reported for Trichodesmium sp. [Berman-Frank et al., 2001] as an example N2 fixing organism. The highest Nfix values were obtained for the more intense Saharan dust deposition event of 2000, with values up to ∼180 μmol N m−2 d−1 - comparable to N2 fixation rates observed in tropical waters [Capone, 2001]. Nfix values were much lower in the southern hemisphere (<10 μmol N m−2 d−1) and were lower than the observed atmospheric N deposition fluxes (Figure 2b), implying that N2 fixation makes only a small contribution to external nitrogen supply outside of the areas affected by the Saharan dust plume. There is currently some debate about the Fe requirement of marine diazotrophs, and much lower Fe:C values have been proposed by other workers [Kustka et al., 2003; Sanudo-Wilhelmy et al., 2001]. Although use of these Fe:C values in our calculation leads to much higher values of Nfix, the implied N2 fixation rates in the southern hemisphere only become comparable to, but do not significantly exceed, the rate of N input directly from the atmosphere. While diazotrophs provide a means to overcome water column N-limitation, eventually available P will become exhausted [Sanudo-Wilhelmy et al., 2001]. We note that the influence of aerosol input on water column nutrient status [Falkowski, 1997; Tyrrell, 1999] will be to drive the water column towards short-term P limitation because utilisation of the aerosol N and Fe components both tend to deplete water column P. These results illustrate that evaluating the impact of atmospheric deposition of Fe on marine productivity cannot be done without consideration of the relative inputs of the other key nutrients N and P.

Acknowledgments

[10] We thank the Masters and crews of RV Polarstern and RRS James Clark Ross. This work was funded by the EU programme IRONAGES (EVK2-1999-00031). Additional support for the JCR cruise was provided by the UK Natural Environment Research Council (NER/B/2000/00116) and studentships to M. W. (University of East Anglia, School of Environmental Sciences) and K. F. B. (Association of Commonwealth Universities). The comments of two anonymous reviewers are gratefully acknowledged.