OUP user menu

Differential responses of Prochlorococcus and SAR11-dominated bacterioplankton groups to atmospheric dust inputs in the tropical Northeast Atlantic Ocean

Polly G. Hill, Mikhail V. Zubkov, Duncan A. Purdie
DOI: http://dx.doi.org/10.1111/j.1574-6968.2010.01940.x 82-89 First published online: 1 May 2010


The metabolic responses of indigenous dominant bacterioplankton populations to additions of dust were examined in the tropical northeast Atlantic. Subsurface seawater samples were treated with dust, added directly or indirectly as a ‘leachate’ after its rapid dissolution in deionized water. Samples were incubated at ambient temperature and light for up to 24 h and microbial metabolic responses were assessed by 35S-methionine (35S-Met) uptake. Prochlorococcus and low nucleic acid (LNA) cells were sorted by flow cytometry to determine their group-specific responses. Sorted cells were also phylogenetically affiliated using FISH. The high-light-adapted ecotype II dominated the Prochlorococcus group and 73±14% of LNA prokaryotes belonged to the SAR11 clade of Alphaproteobacteria. Both Prochlorococcus and LNA cells were metabolically impaired by the addition of dust (40±28% and 37±22% decrease in 35S-Met uptake compared with controls, respectively). However, LNA bacterioplankton showed minor positive responses to dust leachate additions (7±4% increase in 35S-Met uptake), while the metabolic activity of Prochlorococcus cells decreased in the presence of dust leachate by 16±11%. Thus, dust dissolution in situ appears to be more deleterious to Prochlorococcus than SAR11-dominated LNA bacterioplankton and hence could initiate a compositional shift in the indigenous bacterioplankton.

  • Bacteria
  • flow cytometric sorting
  • Aeolian
  • Saharan sand


Desert dust consists of soil particles that are lifted into the atmosphere when high winds occur over dry and sparsely vegetated land (Mahowald et al., 2005). With dust production estimated at about 1700 Tg year−1 (Jickells et al., 2005) and potentially increasing desertification (Rosenfeld et al., 2001), the effect of dust deposition on the indigenous microbial communities of the surface ocean can be significant. Desert dust, and its associated nutrients, can play a key role in regulating primary production (Guieu et al., 2002; Bonnet et al., 2005; Herut et al., 2005; Moore et al., 2006) and bacterial production (Herut et al., 2005; Pulido-Villena et al., 2008b) in the open ocean, as well as bacterioplankton and phytoplankton dynamics in lakes and reservoirs (Pulido-Villena et al., 2008a; Reche et al., 2009).

Generally, studies have shown atmospheric dust deposition to be beneficial to bacterioplankton communities. Saharan dust addition incubations have indicated the stimulation of bacterial production in a Spanish reservoir (Reche et al., 2009) and the eastern Mediterranean basin (Herut et al., 2005), nitrogen fixation in the tropical north Atlantic (Mills et al., 2004) and bacterial abundance in a high mountain lake (Pulido-Villena et al., 2008a) and the western Mediterranean Sea (Pulido-Villena et al., 2008b). However, the bacterial communities of the northwestern Mediterranean Sea (Bonnet et al., 2005) and subtropical northeast Atlantic (Duarte et al., 2006) showed little or no response to dust addition. Observations of dust deposition in situ have also indicated a positive response of bacterial abundance in a Mediterranean lake (Pulido-Villena et al., 2008a) and in the western Mediterranean Sea (Pulido-Villena et al., 2008b), and bacterial activity in the eastern Mediterranean basin (Herut et al., 2005).

More specifically, Synechococcus abundance increased and Prochlorococcus abundance decreased in response to dust addition in the eastern Mediterranean basin (Herut et al., 2005), whereas the opposite was observed in the Gulf of Aqaba in the northern Red Sea (Paytan et al., 2009).

There is a need to assess the response of individual populations of the bacterioplankton community to dust deposition. The aim of this study, therefore, was to assess the metabolic responses of key groups of oceanic bacterioplankton to dust deposition. The study focused on two bacterioplankton groups: the Prochlorococcus cyanobacteria and the SAR11 clade of Alphaproteobacteria, because in the (sub)tropical open ocean, the bacterioplankton community is often dominated by Prochlorococcus (Chisholm et al., 1988) and the globally ubiquitous and abundant SAR11 (Morris et al., 2002).

The metabolic response of these bacteria was studied because microbial metabolism, or production, is more sensitive to environmental change than abundance (Gasol & Duarte, 2000). The (sub)tropical northeastern Atlantic region was chosen because this region is regularly exposed to high Saharan dust inputs, ∼5 g m−2 of dust per year (Jickells et al., 2005), and yet few studies on the subject have been conducted there (Mills et al., 2004; Duarte et al., 2006). Dust addition incubations were used to exclude the factors associated with dust events, such as high wind speeds and surface cooling, which may lead to favourable conditions for cell growth (McGillicuddy & Robinson, 1997; Singh et al., 2008). Additions of freshly collected dust or dust ‘leachate’ (Buck et al., 2006) were made in parallel to natural seawater samples.

Materials and methods

Sampled region and sample collection

The experimental work was conducted during an oceanographic cruise on board the Royal Research Ship Discovery (cruise no. D326) in the eastern (sub)tropical North Atlantic Ocean (Fig. 1) during January–February 2008. A homogeneous surface mixed layer was characteristic for the studied region with a mixed layer depth of ∼50 m in the southwest of the region and a deeper mixed layer of ∼131 m to the north. Easterly winds prevailed in the northern region (78±13°) and north-easterly winds in the south (46.1±12°). During 25–28 January, a major dust deposition event occurred, while the ship was in the southwest of the region, making the sky brown and covering the ship in a layer of red-brown dust. The dust cloud was clearly visible in satellite images and back trajectories for these dates show that the air mass came from the Sahara region.

Figure 1

Sample collection locations (all triangles) for cellular 35S-Met uptake by Prochlorococcus and LNA bacterioplankton cells, and FISH analysis of bacterioplankton community structure. Black triangles also indicate where seawater for dust and leachate addition incubations was collected, with an indication of incubation number.

Seawater samples were collected using a trace metal clean technique from 20 m depth, to minimize iron contamination from the ship's hull, using a rosette of 20-L Niskin bottles mounted on a titanium frame with a CTD profiler (Sea-Bird Electronics) in polyoxymethylene plastic and titanium casing. Samples were decanted into 1-L HCl-cleaned polycarbonate bottles. The experiments commenced within an hour of sampling.

Dust collection and leachate production

Dust samples were collected daily, at sea, onto polypropylene filters (47 mm, 0.45 μm, Sterlitech). Rotary vein vacuum pumps filtered aerosol at 25–30 L min−1 for periods of typically 24 h, although this was reduced to 4–6 h during the major dust event on 25–28 January. The instantaneous dissolution of metals and nutrients was simulated by quickly passing 100 mL of deionized water (milli-Q) through the dust-loaded filter (Buck et al., 2006) and the leachate was subsampled into sterile 2-mL polypropylene vials.

Experimental procedure

The bacterioplankton response to dust and leachate additions was determined by time-course sampling during incubations lasting 24 h. Four incubations were performed, two in the southwest of the region and two in the north (Fig. 1). Seawater samples (34 mL) were incubated in HCl-cleaned 35-mL PTFE bottles with dust, leachate or no (control) additions. Dust was added with the polypropylene filter onto which it was collected; additions were calculated postcruise to be 0.3 mg L−1 (incubation 1), 1.5 mg L−1 (incubation 2) or 4.7 mg L−1 (incubations 3 and 4). A further control of a blank polypropylene filter was used to ensure that the bacterioplankton response was due to the dust and not the filter. Leachate additions of 700 μL supplied 100 nM inorganic N and 10 nM P to all incubations. Bottles were placed in on-deck incubators screened to allow 20% surface irradiance and cooled to in situ temperature.

The uptake rate of 35S-methionine (35S-Met) was measured at t=0, 2, 4, 6 and 24 h to determine the bacterioplankton community metabolic response to treatments (+Leachate or +Dust) as compared with controls. Two incubations were also sampled at t=8 h. At t=0 and 6 h, samples were taken to measure cellular uptake by sorted bacterioplankton groups. A further eight t=0 h samples were collected throughout the cruise to measure the cellular uptake in response to natural dust deposition in the ocean. Parallel to this, untreated samples were fixed with paraformaldehyde (1% final concentration) and stored at −80 °C before analysis of bacterioplankton community structure.

35S-Met uptake rates

Throughout the 24-h dust and leachate addition incubations, uptake rates of 50 pM 35S-Met (1175 Ci mmol−1, Perkin Elmer, Beaconsfield, UK) by total bacterioplankton were measured using time series (10, 20 and 30 min) incubations with 500-μL subsamples. Subsamples were fixed with 1% paraformaldehyde and filtered onto 0.2-μm polycarbonate membrane filters. The radioactivity retained on filters was measured using a liquid scintillation counter (Tri-Carb 3100, Perkin Elmer, UK) on board the ship and is presented in becquerels (Bq).

At t=0 and 6 h, three 1.6-mL replicate seawater samples were incubated with 0.2 nM 35S-Met for 2 h to compare the bacterioplankton metabolic response to ambient dust deposition (t=0 h), and dust and leachate addition, as compared with controls, in incubation bottles (t=6 h). Samples were fixed with 1% paraformaldehyde and stored at −80 °C until sorted by flow cytometry to determine the group-specific 35S-Met cellular uptake.

35S-Met dilution bioassays (Zubkov et al., 2003) were performed in parallel to all experiments to estimate the ambient methionine concentration, uptake rates and turnover times. These data will be published elsewhere.

Analysis of bacterioplankton groups

Bacterioplankton samples were analysed using flow cytometry (FACSCalibur, BD Biosciences, Oxford, UK). Prochlorococcus cyanobacteria were identified and flow sorted from unstained samples using their characteristic red autofluorescence (Olson et al., 1993). Bacterioplankton cells were stained with the nucleic acid stain SYBR Green I (Marie et al., 1997), and the cells with low nucleic acid (LNA) and high nucleic acid (HNA) content (Li et al., 1995; Gasol et al., 1999) were separated using a plot of side scatter (90° right angle light scatter) against green (FL1) fluorescence. Although the SAR11 clade of Alphaproteobacteria cannot be discriminated specifically by flow cytometry, they dominate the LNA bacterioplankton group (Mary et al., 2006; Schattenhofer, 2009), which can be sorted.

The isotopically labelled LNA bacterioplankton and Prochlorococcus cells were flow sorted as described previously (Zubkov et al., 2004; Mary et al., 2006). Radioactivity retained by known numbers of sorted cells from the two groups examined was measured using an ultra-low-level liquid scintillation counter (1220 Quantulus, Wallac, Finland) ashore and is presented as mBq per cell. In order to assess 35S-Met adsorption to dust, 5000 dust particles were sorted in parallel to microbial cells. The radioactivity of the dust particles was indistinguishable from the background measurements, indicating insignificant adsorption of 35S-Met to dust.

Bacterioplankton cells in samples collected for community structure analysis were sorted into the HNA and LNA groups. Cells were collected directly onto 0.2-μm pore size polycarbonate membrane filters (Millipore, Isopore) and analysed by FISH using the method described by Pernthaler (2002), with the adaptations of Zubkov (2007), and the probes detailed in Table 1.

View this table:
Table 1

Results of FISH analysis of LNA- and HNA-containing bacterioplankton groups, expressed as percentages of prokaryotes (mean±SD of at least six fields of view)

DateSample originLNAHNA
20/01/0812°33′5632°40′3494 ± 486 ± 490 ± 375 ± 836 ± 60 ± 0
21/01/0812°31′5035°46′5194 ± 476 ± 688 ± 459 ± 942 ± 60 ± 0
22/01/0812°37′1333°14′0689 ± 460 ± 594 ± 669 ± 968 ± 50.3 ± 0.6
23/01/0812°30′1930°37′3194 ± 283 ± 987 ± 562 ± 748 ± 30 ± 0
25/01/0816°11′1330°38′3789 ± 539 ± 796 ± 274 ± 349 ± 90 ± 0
30/01/0821°56′5027°04′52No data64 ± 593 ± 570 ± 444 ± 50 ± 0
30/01/0822°49′0227°11′5087 ± 480 ± 795 ± 372 ± 431 ± 30.3 ± 0.8
01/02/0826°08′5226°24′48No data79 ± 892 ± 366 ± 454 ± 70.3 ± 0.7
02/02/0826°36′2623°42′2992 ± 679 ± 494 ± 364 ± 432 ± 30.3 ± 0.8
02/02/0826°42′3223°00′46No data83 ± 393 ± 475 ± 754 ± 50 ± 0
Mean91 ± 373 ± 1492 ± 368 ± 646 ± 120.1 ± 0.2
  • Probes applied were EUB338I-III (Amann et al., 1990; Daims et al., 1999), SAR11-441 (Morris et al., 2002), 405Pro (West et al., 2001) and 645HLII (West et al., 2001).

Statistical analysis

All data are presented as means±SD for the stated number of independent observations. Statistical significance at P <0.05 was determined using Student's t-test for paired or unpaired samples depending on the compared datasets.


The SAR11 clade of Alphaproteobacteria dominated the LNA group at 72±14% of prokaryotes (Table 1). The unidentified fraction of the LNA group could not be phylogenetically affiliated using other probes including Gam42a (identifying Gammaproteobacteria), 405Pro (Prochlorococcus) or 645LL (low-light-adapted Prochlorococcus).

Prochlorococcus dominated the HNA bacterioplankton at 68±6% of prokaryotes (Table 1). The majority of Prochlorococcus cells belonged to the high-light-adapted ecotype II (HLII) (Table 1). A maximum of 2% of prokaryotes were identified by 645HLI as the HLI, with the majority of samples containing none. No more than one or two HNA cells were identified as SAR11 in each sample, with the majority containing none (Table 1).

In experimental incubations, 35S-Met uptake by LNA bacterioplankton cells increased by 4–13% in the presence of leachate (as compared with controls) in each of the four incubations, and the increase was statistically significant in two (Fig. 2). Conversely, Prochlorococcus cells, sorted unstained, took up significantly less 35S-Met in the presence of dust leachate (3–28% less than in controls) in each of the experiments (Fig. 2). Yet, in unsorted samples, the bacterioplankton community was mostly unaffected by the addition of dust leachate at each time point (four or five per incubation) sampled throughout the four incubations (paired t-test, P >0.1, n=18; Fig. 3a).

Figure 2

Uptake of 35S-Met (mBq) per Prochlorococcus (left) and LNA (right) bacterioplankton cell after 6-h incubations with no addition (control) or in the presence of dust leachate or dust (mean±SD; N=3). Dust additions were 0.3 mg L−1 (incubation 1, a and b), 1.5 mg L−1 (incubation 2, c and d) or 4.7 mg L−1 (incubations 3 and 4, e–h). Treatments were compared with controls using Student's t-test; statistical significance is indicated as follows: *P <0.05, **P <0.01, ***P <0.0001. d.f.=4 for all comparisons.

Figure 3

Impact of dust leachate (a) and dust (b) addition on the rate of bacterioplankton community 35S-Met uptake (Bq mL−1 min−1) at 2 hourly intervals throughout initial 6–8 h, and after 24 h of the incubation experiments. Symbols indicate data from individual incubations. All data are presented as mean±SE.

The effect of direct dust addition (not as a leachate) was more dramatic; 35S-Met uptake by both Prochlorococcus and LNA bacterioplankton decreased during all incubations by 21–82% and 20–68% of the control, respectively (Fig. 2). Dust addition also negatively impacted the bacterioplankton community as a whole (Fig. 3b).

During the dust deposition event, LNA bacterioplankton took up significantly more 35S-Met per cell than Prochlorococcus, paired t-test, P <0.005, n=3, suggesting reduced metabolic activity of Prochlorococcus and/or enhanced metabolic activity of the LNA bacterioplankton (Fig. 4). Outside of the dust event, Prochlorococcus cells took up more 35S-Met than the LNA cells.

Figure 4

Cellular uptake of 35S-Met by untreated Prochlorococcus (Prochl.) and LNA bacterioplankton (LNA Bpl.) cells throughout the cruise period. Uptake was measured as mBq per cell and displayed as mean±SD (N=3 for each sample). Samples were incubated with 0.2 nM 35S-Met for 2 h. The period of dust deposition is indicated by a black line and triangular symbols.


The bacterioplankton metabolic response to dust additions was measured by comparing the cellular uptake rates of radiolabelled methionine, as a proxy for bacterioplankton production. Methionine was used because it is available with a 35S label, which gives it a higher specific activity than the more traditional 14C and tritium-labelled leucine tracers used previously (e.g. Herut et al., 2005), which increases the sensitivity of the flow-sorting technique. Prochlorococcus and SAR11 have been shown to take up 35S-Met actively (Zubkov et al., 2003; Mary et al., 2006); indeed, Prochlorococcus has exhibited a preference for methionine over leucine (Mary et al., 2008) despite the leucine requirement for all proteins.

The data presented suggest that the LNA bacterioplankton, but not Prochlorococcus, benefited metabolically from dust leachate additions. This differential result was hidden when observing the community response as a whole, which suggested no stimulation or suppression of bacterial metabolism. The varying degree of stimulation of LNA bacterioplankton by leachate within the four incubations was presumably due to the variation in the ambient methionine uptake rates, as indicated by 35S-Met bioassays that were conducted in parallel (4.2–17.7 pmol L−1 h−1, P. G. Hill unpublished data).

In agreement with previous observations, the SAR11 clade of Alphaproteobacteria dominated the LNA bacterioplankton, and yet was not identified within the HNA bacterioplankton. This coverage of 72±15% LNA prokaryotes is similar to that achieved in one previous study (Schattenhofer, 2009), but higher than others (Mary et al., 2006; Zubkov et al., 2007), probably because the cells were more metabolically active, allowing more hybridizations to occur. The remaining fraction of LNA bacterioplankton cells could be identified as Bacteria, while they could not be affiliated to other groups, including Gammaproteobacteria and Prochlorococcus. The difficulty in identifying the LNA group in open ocean samples (Mary et al., 2006; Schattenhofer, 2009) suggests that they could belong to the SAR11 clade, but differ in their cellular ribosomal content.

Dust may introduce organic carbon (Duarte et al., 2006; Pulido-Villena et al., 2008b), which could benefit heterotrophic SAR11 cells more than phototrophic Prochlorococcus cells. It may also alleviate the limitation of microbial growth by inorganic N or P (Rivkin & Anderson, 1997; Caron et al., 2000); Prochlorococcus cells can assimilate these inorganic nutrients (Casey et al., 2007). Indeed, a strain of Prochlorococcus found in the Red Sea, which is relatively insensitive to metal toxicity compared with strains from the Atlantic, has been shown to increase in abundance following inorganic nutrient and Saharan dust additions (Paytan et al., 2009). However, the majority of Prochlorococcus cells in samples from the present study belonged to the HLII (Table 1), which are well adapted to oligotrophic environments (West et al., 2001; Johnson et al., 2006; Zubkov et al., 2007; Zwirglmaier et al., 2007). No more than 2% of HNA prokaryotes were identified as HLI, which has a relatively high nutrient requirement compared with HLII (Johnson et al., 2006). Given that the study region was dominated by HLII, it seems unlikely that the Prochlorococcus population would have benefited from dust-derived nutrients.

Ecotypes of both Prochlorococcus and SAR11 have maximized their ability to take up nutrients efficiently at very low nutrient concentrations. Their resultant streamlined genomes lack many of the regulatory proteins found in most marine bacteria for regulating the uptake of N, P and Fe (Rocap et al., 2003; García-Fernández et al., 2004; Giovannoni et al., 2005; Martiny et al., 2006, 2009), which could make it difficult for these groups to regulate nutrient uptake at substantially elevated concentrations. Thus, deposition of high quantities of nutrients and metals in dust may be toxic to these groups. Prochlorococcus, for example, have been shown to be particularly sensitive to copper (Mann et al., 2002). Herut (2005) also report a decline in the Prochlorococcus community in response to Saharan dust in Mediterranean waters. Furthermore, studies have shown that SAR11 is not very abundant in mesotrophic regions (Fuchs et al., 2005; Alonso-Sáez et al., 2007), which implies a disadvantage of this clade in regions of high nutrient availability.

Direct dust addition to seawater suppressed the metabolism of both Prochlorococcus and LNA cells, and this negative impact was also clear at the bacterioplankton community level. Conversely, dust additions to reservoir water showed an increase in bacterial production after a 48-h incubation, although there was evidence that this was due to the introduction of air-borne Gammaproteobacteria associated with the dust particles (Reche et al., 2009). A comparison of cellular methionine uptake by the two flow-sorted bacterioplankton groups in control samples suggests that LNA bacterioplankton benefited from and/or Prochlorococcus were inhibited by dust deposition in the field (Fig. 4). These observations support our experimental findings that small increases in dust-derived nutrients have a detrimental impact on Prochlorococcus in the region. It seems plausible, therefore, that ambient bacterioplankton communities suffer from large dust events, whereas opportunistic bacteria multiply rapidly, leading to increased bacterial production.

In summary, this study suggests differential responses of major bacterioplankton groups to dust-derived nutrients, which are hidden when studying the bacterioplankton community as one entity. However, the cause of these differential responses of the Prochlorococcus and LNA bacterioplankton groups requires further investigation.


We thank all the scientists involved in the Natural Environment Research Council (NERC) UK Surface Ocean Lower Atmosphere Study (SOLAS) project NE/C001931/1 and cruise D326, particularly Eric Achterberg, Claire Powell, Ludwig Jardillier, Micha Rijkenberg and Matthew Patey. We would also like to thank the captain and crew onboard RRS Discovery. We thank Bernhard Fuchs and Jörg Wulf at the Max Planck Institute for Marine Microbiology, Bremen, and Jane Heywood currently at the University of Bremen, for their help with FISH identification of bacterioplankton. Manuel Dall'Osto at the National University of Ireland, Galway, provided back trajectories for the dust storm. This research was funded by NERC UK SOLAS. P.G.H. is funded by a SOLAS NERC-tied studentship.


  • Editor: J. Colin Murrell


View Abstract