Introduction

Increasing dust outbreaks triggered by climate-induced land desertification1,2 have the potential of counteracting the negative effects from ongoing ocean stratification through providing an alternative source of nutrients (such as phosphorus and iron) fertilising primary production in an ever-warmer ocean3,4,5,6. At the same time, dust is also a source of mineral ballast which is critical for enhancing the export and sequestration of particulate organic carbon (POC) by increasing the sinking velocities of carbon-enriched material produced by phytoplankton and zooplankton in the upper ocean7.

Recent studies report on the influence of Saharan dust deposition as sporadically stimulating the productivity and export fluxes of coccolithophores (Haptophyta) (consult the Glossary in the Supplementary Material) in the heavily stratified tropical North Atlantic8,9,10. Coccolithophores are a biogeochemically important group of phytoplankton that can intracellularly biomineralize CaCO3 scales (coccoliths) which are secreted onto the cell surface to collectively form a calcitic exoskeleton (coccosphere)11. This ability contributes to some of the highest export fluxes of particulate inorganic carbon (PIC) from the photic zone to the deep sea12,13,14,15. Being primary producers, coccolithophores act as a CO2 sink, since they fix CO2 during photosynthesis and provide calcitic ballasting minerals which facilitate the export and deep-sea sequestration of POC13,16,17, similar to dust particles. In parallel, as calcification incorporates carbon into calcite, it reduces the availability of the carbonate ion in the surface ocean while releasing CO2, thereby causing coccolithophores to also act as CO2 source13.

In a series of sediment-trap studies across the tropical North Atlantic downwind of NW Africa, Guerreiro et al.8,9 reported on the occurrence of pulsed export production events by the opportunistic coccolithophores Emiliania huxleyi and Gephyrocapsa oceanica in the western part of the ocean basin (mooring site M4 at 12°N/49°W), which were linked to Saharan dust deposition. Both events were marked by a striking increase of POC and of bSiO2/CaCO3 ratios (i.e., proxy for diatom versus biogenic CaCO3 export production), and by a decrease of the coccolith-PIC/POC molar ratios (i.e., proxy for coccolithophore contribution to the PIC/POC ratio), suggesting that dust-born nutrient input had contributed to stimulate the biological carbon pump17,18. More recently, Guerreiro et al.10 reported on enhanced abundance (cells/L) of E. huxleyi and G. oceanica, and of N2-fixing filamentous cyanobacteria Trichodesmium sp. in the photic zone of the tropical NE Atlantic (~ 28°W), in response to Saharan-driven input of iron and phosphorous. While this recent finding supports the dust-related coccolith export productivity events above, it remains an open question whether they mostly reflected a nutrient-stimulated ecological (growth) response or if they were the result of increased export efficiency related to dust- and coccolith-ballasting (see17,19).

To answer this question, we have quantified the coccolith-Sr/Ca ratio from the same samples collected at trap site M4 (Fig. 1). Our approach is based on previous studies reporting higher Sr partitioning (i.e., higher amount of Sr incorporated into the calcite) as directly proportional to the coccolith calcification rate which, in turn, is a function of the coccolithophores’ growth rate (e.g.20,21,22,23). Because Sr/Ca ratios of seawater vary by less than 2% globally, and since the link between coccoliths Sr partitioning and their calcification rates is consistent with the control of Sr partitioning by calcification rate in abiogenic calcites22,23, substantial variations in Sr/Ca (> 20%) are assumed to reflect growth-related Sr partitioning in coccolith calcite. Basically, the faster coccolithophores grow, the faster they calcify, resulting in more Sr being incorporated into the calcite lattice of their coccoliths20,24. In the same way, nutrient limitation in coccolithophores has been reported to reduce the uptake of Sr compared to Ca into the calcifying vesicle, thereby decreasing the coccolith Sr/Ca ratios25,26,27,28. Therefore, variations in coccolith-Sr/Ca ratios can be used as proxies for nutrient-stimulated growth rates of coccolithophores21,29,30,31. Previous sediment-trap records in the Sargasso Sea and Arabian Sea30, and in the Eastern Mediterranean32 have already provided relevant indications about nutrient limitation of coccolithophore in situ productivity versus coccolith export efficiency out of the photic zone based on coccolith Sr/Ca ratios. Auliaherliaty et al.33 further demonstrate that Sr/Ca are not impacted by bacterial influences on algal physiology, thereby reinforcing the robustness of this ratio as a productivity proxy.

Figure 1
figure 1

Location of trap mooring site M4 (A), and a schematic representation of the main surface currents in the equatorial Atlantic Ocean, with the inset showing the seasonal eastward retroflection of the North Brazilian Current (NBC) during boreal summer (B) (adapted from8,36).

Full size image

Our study is based on new data of coccolith Sr/Ca signatures measured in sinking coccolith assemblages collected at sediment trap site M4 from October 2012 to October/November 2013, in relation to published data from the same trap/period, i.e., species-specific fluxes of coccoliths and coccolith-CaCO38,17; fluxes of carbonate, organic matter, biogenic silica—bSiO2, and mineral dust34,35. The main goal is to gain a broader perspective on the coccolith-Sr/Ca in relation to the drivers of coccolith export production, by linking the surface production with the export particle fluxes towards distinguishing the biogeochemical effects of Saharan dust acting as both a fertilizer and/or as ballast.

Material and methodology

Sediment trap sampling

Our study is based on time-series particle flux material collected by sediment-trap mooring M4 (12°N/49°W) at intervals of 16 days during one year (from 4 October 2012 to 7 November 2013) at 1200 m water depth in the western tropical North Atlantic (Fig. 1). Details of the mooring equipment, the deployment/recovery of the sediment trap, and the treatment of the recovered sample bottles are described in the cruise report37. Sediment-trap samples were initially wet-sieved over a 1 mm mesh, wet-split into five aliquot subsamples using a rotary splitter (WSD-10; McLane Laboratories), washed to remove the HgCl2 and salts, and centrifuged. Average weight differences between replicate aliquots were within 2.4% (SD = 2.2), with 87% of all samples differing < 5% between splits (detailed procedure in34).

Previous studies have already extensively described the oceanographic and meteorological settings, as well as the seasonal patterns of particle fluxes at mooring trap site M4 during the sampling period studied here8,34,35. A summary of environmental background at the location of mooring M4 is provided in the Supplementary Material.

Coccolith- and coccolith-CaCO3 flux analysis

Data of species-specific coccolith- and coccolith-CaCO3 fluxes presented in this study are from8,17. These studies are based on a minimum of 500 coccoliths counted from an arbitrarily chosen transect and each coccolith was identified to the lowest taxonomic level possible at 3000 × magnification using a Zeiss DSM 940A SEM at 10 kV of accelerating voltage. This number of counts per sample is well above the recommended minimum of 300 specimens for studies focused on quantifying species with > 3% of the assemblage, corresponding to a 95% confidence level38. The UPZtaxa/LPZtaxa ratio presented in Fig. 4c, used as a proxy for inferring variations in the depth of the nutricline, is calculated by dividing the fluxes produced by typically upper photic zone (UPZ) species Emiliania huxleyi and Gephyrocapsa spp., by the fluxes produced by lower photic zone (LPZ) species Florisphaera profunda and Gladiolithius flabellatus8,9.

The coccolith-derived CaCO3 export fluxes were determined using the mass equation of39, according to which the coccolith mass of distinct species is expressed as: Coccolith calcite (pg) = 2.7 × Ks × l3, in which 2.7 = density of calcite (CaCO3); ks = shape constant; l = coccolith size (mostly distal shield length). The size of around 3500 coccoliths were measured from most species present in the samples from trap M4 using the same SEM. A detailed description of the quantification of the coccolith-calcite masses and coccolith-CaCO3 fluxes is presented in17.

Quantification of the coccolith-Sr/Ca ratios

The coccolith Sr/Ca sample processing and analyses were adapted from29 and31, focusing on measuring Sr/Ca ratios from different enriched coccolith size fractions which had been previously separated via repeated decanting prior to the chemical analysis. This approach aims to obtain different size fractions dominated by distinct coccolith species with distinct calcite mass and sinking rate. Different species can indeed have different coccolith calcite partitioning of Sr and amplitudes of Sr/Ca variation (e.g.,40,41,42). The laboratory procedure is further described in the Supplementary Material. We present the Sr/Ca ratios determined from the bulk fraction (< 20 µm) and from the three coccolith size-fractions (> 6 µm, 3–6 µm and < 3 µm), which we compared to published data concerning the total and species-specific coccolith- and coccolith-CaCO3 export production from the same samples. In addition to the species-specific CaCO3 contribution to the < 20 µm bulk fraction, we also show their CaCO3 contribution for a series of coccolith size fractions obtained from trap samples U2, U7, U12, U14, U18, U21 and U24, which were selected for microscopic inspection. The total number of counts performed in the coccolith size fractions (150 < n < 390 liths) was high enough to ensure the quantification of rare species within the assemblage within a 95% confidence level38 which often contribute disproportionally high amounts of carbonate to the assemblage despite of its low abundance. Such was particularly the case for the very large Scyphopshaera apsteinii for which we determined the following 95% confidence intervals of coccolith counts for the fractions where this species was found: [0–3] for the intermediate fraction of U2, and for the large fractions of U12, U18 and U21; and [0–6] for the large fraction of U14. Despite reported evidence on ocean temperature affecting the partitioning of Sr into the coccolith calcite (~ 0.03 mmol/mol increase per °C increase24,43, the temperature changes during our study (up to 3 °C) can be neglected compared to the observed Sr/Ca changes (i.e., a change of 3 °C entails a maximum change of 0.9 mmol/mol in the range (0.7–12.6 mmol/mol), average (4.1 mmol/mol) and standard deviation (0.7) of our Sr/Ca results. Table 1 shows the Sr/Ca results obtained from this study. For the calculation of the normalized Sr/Ca ratios presented in Fig. 4a, we first determined a “Sr/Ca anomaly” by subtracting the Sr/Ca annual mean to the Sr/Ca ratio of each sample; and then we divided each Sr/Ca anomaly by the Sr/Ca annual mean deviation to obtain the normalized values.

Table 1 Coccolith Sr/Ca (mmol/mol) results from the chemical elemental analysis.
Full size table

Satellite remote sensing

Time series of meteorological and hydrological data obtained from satellite remote sensing were used to provide a complementary perspective on the environmental conditions during the in-situ sampling period (datasets details provided in Table I—Supplementary Material). The mixed-layer depth (MLD, defined as being the deepest ocean layer affected by wind-forced turbulent mixing) was used as an indicator of seasonal environmental variations related to the Intertropical Convergence Zone (ITCZ) in terms of ocean thermal stratification vs. wind-forced water cooling and mixing (i.e., < MLD indicates weaker trade winds due to a greater influence of the ITCZ on trap site M49. Sea surface salinity (SSS) was used as an indicator of salinity variations linked to the seasonal entrainment of Amazon River water during the sampling period8. Satellite data were retrieved using a 2° × 2° latitude–longitude area around the location of trap M4 and averaged for each sediment trapping interval. The 2° × 2° box, corresponding to ~ 108 × 108 N mi (1° =  ~ 59 N mi), was taken as representative of the catchment area of a sediment trap deployed at 1200 m depth, given the sinking speed for marine phytoplankton and algal aggregates (e.g.,8,44).

Statistical analysis

In order to investigate the relationship between the Sr/Ca ratios and the environmental conditions during the sediment trap period, a Principal Component Analysis (PCA, correlation mode, by PAST-4.11) was performed upon a data matrix with species-specific CaCO3 fluxes, Sr/Ca ratios of all studied coccolith size fractions, fluxes of organic matter, biogenic silica (bSiO2) and mineral dust, and with remotely sensed Chlorophyll-a (Chla), SSS and MLD as columns (n = 18 variables). Only the samples with Sr/Ca data available for all size fractions were considered (n = 20 cases). A Spearman correlation coefficient matrix was also built upon the same data matrix for assessing the statistical significance of the correlations obtained from the PCA, using a default p-level of 0.05. All the results from the statistical analysis are presented in the Supplementary Material.

Results

Coccolith CaCO3 size fraction separation for Sr/Ca analyses

The Sr/Ca was obtained from three coccolith size fractions, each characterized by CaCO3 content driven by a few species of similar coccolith size (Fig. 2). The bulk fraction (< 20 µm) overall mimicked the seasonal variation of the original coccolith sinking assemblages and related coccolith-CaCO3 reported by8,17. The small fraction (< 3 µm) was dominated (38–87%) by CaCO3 from deep-dwelling species Gladiolithus flabellatus and F. profunda. The intermediate (~ 3–6 µm) and large fractions (> 6 µm) were dominated by carbonate produced by Helicosphaera spp. (up to 83%) followed by Scyphosphaera apsteinii (up to 44%), Calcidiscus leptoporus (up to 19%) and Pontosphaera spp. (up to 7%). This high carbonate contribution by S. apsteinii to the large size fraction results from its unusual large size found in M4 samples, thereby contributing a disproportionally high percentage compared to its low relative abundance (Fig. I in the Supplementary Material). Note that samples U12, U14, U21 and U24 show that E. huxleyi, Gephyrocapsa spp. and the “other taxa” significantly increased their carbonate contributions in the small fraction (up to 26%, 22% and 32%, respectively), while Gephyrocapsa spp. also contributed to the intermediate/large fractions of samples U7 (8% and 7%, respectively) and U24 (17% and 7%, respectively) (Fig. 2).

Figure 2
figure 2

Species-specific coccolith-CaCO3 contribution (%) in the bulk fraction (< 20 μm) and coccolith size-fractions (small < 3 µm; intermediate 3–6 µm; and large > 6 µm), from selected sediment trap M4 samples U2, U7, U12, U14, U18, U21 and U24.

Full size image

Seasonal distribution of the Sr/Ca ratios

Here we report on the variation of the Sr/Ca for the different size fractions along the entire sediment trap period, i.e., from boreal autumn 2012 (October) until boreal autumn 2013 (October–November). We found generally higher ranges of Sr/Ca in the large fraction (> 6 µm; 1.6–12.6), followed by the bulk (< 20 µm; 1.2–5.4) and intermediate-size fractions (3–6 µm; 0.7–5.7), and finally the small size fraction (< 3 µm) with the lowest range (1–2.5). The bulk fraction revealed enough sediment to measure Sr/Ca ratios for all the analysed samples, while the small, the intermediate and the large size fractions occasionally did not (i.e., 2/23 of the samples for the small fraction; 4/23 of the samples for the intermediate-size fraction, and 1/23 of the samples for the large-sized fractions (Table 1 and Fig. 3).

Figure 3
figure 3

(a) Seasonal variation of the coccolith-Sr/Ca ratios in the bulk fraction (< 20 μm) and in the coccolith size fractions (small < 3 µm; intermediate 3–6 µm; large > 6 µm); (b) total coccolith- and coccolith-CaCO3 fluxes8,17 from sediment trap M4. Numbers refer to samples U2, U7, U12, U14, U18, U21 and U24, for which we performed a taxonomic analysis of all the studied coccolith size fractions (shown in Fig. 2). The light grey vertical bars indicate the periods during which co-increase in biogenic particle fluxes and Sr/Ca ratios was observed.

Full size image

Normalized Sr/Ca ratios revealed an overall similar seasonal pattern, generally higher and above the annual mean during spring and summer for all size fractions, despite some differences in their individual seasonal variation (Fig. 4a). All fractions started with Sr/Ca below the annual average in October 2012 (U1) from which it dropped significantly to a minimum in late November 2012 (bulk, intermediate and large fractions of sample U3) and in early December (small fraction of U4), further increasing to values within the annual average in January 2013 (U7). While there is no Sr/Ca data available for February and early March, the Sr/Ca of the bulk and large fractions was much lower in late March (U10) compared to January, dropping again to values below the annual mean. By contrast, the small fraction recorded a high Sr/Ca ratio during this period. Between April and early October 2013, all fractions recorded higher albeit variable Sr/Ca ratios, usually above the annual mean. Most notable increases occurred in mid-April (small fraction of U12), followed by late-May (bulk and intermediate fractions of U14), July (bulk and large fractions of U17-U18) and late August (bulk and intermediate fractions of U21). Towards the end of the sediment trap time-series, all fractions decreased to Sr/Ca ratios below the annual mean, except the small fraction which slightly increased in October–November (U24) (Fig. 4a).

Figure 4
figure 4

(a) Normalized coccolith-Sr/Ca ratios from the bulk fraction < 20 µm (light orange line) and coccolith size fractions (small, intermediate and large size fractions—red, blue and black lines, respectively); (b) Coccolith-CaCO3 fluxes produced by the main coccolithophore taxa found in the coccolith size fractions (flux data from17; (c) coccolith fluxes of the previous taxa and UPZtaxa/LPZtaxa ratio (purple line) used as a proxy for nutricline depth dynamics8,9; (d) detail of coccolith fluxes by large sized but less productive taxa (Helicosphaera spp. C. leptoporus, S. apsteinii and Pontosphaera spp.); (e) sea surface Chl-a concentrations, fluxes of biogenic particles (CaCO3, bSiO2 and organic matter—data from34; and (f) fluxes of mineral dust (orange—data from van der Does et al.35), mixed layer depth (blue) and sea surface salinity obtained from satellite remote sensing (data from8). Numbers refer to samples U2, U7, U12, U14, U18, U21 and U24, in which we performed a taxonomic analysis of the bulk fraction (< 20 μm) and of the coccolith small, intermediate, and large size fraction (shown in Fig. 2). The light grey vertical bars indicate the periods during which co-increase in particle fluxes and Sr/Ca ratios was observed.

Full size image

Of the three studied coccolith size-fractions, the small-size fraction showed a slightly more variable seasonal pattern, particularly during the winter-spring transition. This was also the only fraction registering an increase during the dust-related productivity event in October–November 2013 (sample U24, Fig. 3).

Discussion

Sr/Ca record: export production driven by dust fertilization vs dust-ballasting

Our observations suggest that the Sr/Ca ratio was more efficient at tracking coccolithophore enhanced productivity occurring in the upper photic zone at site M4, as shown by its co-increase during three events of pulsed flux maxima by more opportunistic placolith-bearing species in the heavily stratified western tropical North Atlantic17. Such events, reported as superimposing to a yearlong predominantly tropical assemblage dominated by deep-dwelling species F. profunda and G. flabellatus (Fig. 4c), were evidenced by the striking increase of the UPZtaxa/LPZtaxa ratio in winter (January–U7), spring (April–U12) and autumn of 2013 (October–November–U24; Fig. 4c,e). Links between Sr/Ca ratios and these high productivity events are also shown in results from the PCA and Spearman correlation analysis in Tables II and III, and Fig. II of the Supplementary Material.

Increased normalized Sr/Ca ratios in all coccolith size fractions during the winter event confirm what has been hypothesized as a biogeochemical response to some degree of nutrient-enrichment driven by seasonal cooling and deepening of the MLD, typical of winter (Fig. 4a,f). While Gephyrocapsa spp. were the main players in this event, as also evidenced by their slightly higher carbonate contributions in all coccolith size fractions from sample U7 (Fig. 2)34 also report a modest but noticeable increase in the fluxes of all biogenic particles, particularly of carbonate (Fig. 4b–e).

As the cool, dry, and windy conditions of January continued to persist, and even intensify, towards spring, the striking Sr/Ca increase in the small fraction of sample U12 clearly reflects a biogeochemical response of the UPZ coccolithophore community to short-term nutrient-enrichment, as illustrated by the increased total coccolith- and coccolith-CaCO3 fluxes (Figs. 3 and 4b,c). Deeper MLDs during this period suggest an even greater wind-forced mixing of the upper ocean compared to January, to which nutrient supply by Saharan dust also contributed, as suggested by the increase of mineral dust fluxes (Fig. 4e,f;35). While E. huxleyi was the most productive species, other taxa increased as well, including large-sized species S. apsteinii and Helicosphaera spp., and more intermediate-sized species within the “other taxa” group, in line with the observed co-increase of the normalized Sr/Ca ratios of the bulk and intermediate-size fractions (Fig. 4a,c). The striking flux increase of all biogenic particles, including organic matter (Fig. 4e), as well as planktic foraminifera (data not shown) which are unlikely to be affected by export efficiency and have a smaller statistical funnel (see31), provide additional evidence that the Sr/Ca peaks in spring reflect a real dust-related production event at the location of trap M4.

These observations agree with previous studies reporting maximum Sr/Ca under similar or even cooler ocean conditions; e.g., Stoll and Schrag20 report on high Sr/Ca ratios coinciding with highest growth rates at temperature minima in the equatorial Pacific; Stoll and Ziveri30 report on Sr/Ca variations generally matching seasonal trends in coccolith fluxes of several species from sediment traps in Bermuda and the Arabian Sea, which were also consistent with inferred seasonal variations in coccolithophore surface productivity during upwelling- and monsoon-related nutrient entrainment.

By contrast, and despite the multi-proxy evidence of enhanced export productivity in autumn, the Sr/Ca of most of the studied fractions obtained from sample U24 significantly dropped (Figs. 3a and 4a), with the small fraction being the only exception as evidenced by its ca. 42% increase considering the small-fraction Sr/Ca range of 1–2.5 for the studied year; Table 1). Guerreiro et al.8 report this event as reflecting a response of opportunistic taxa including E. huxleyi, G. oceanica and diatoms to surface nutrient enrichment derived from seasonal entrainment of Amazon water combined with wet deposition of Saharan dust (consult the Glossary in the Supplementary Material). This is clearly evidenced by the pulsed flux maxima of coccoliths and coccolith-CaCO3 of E. huxleyi and G. oceanica, and of organic material and bSiO2, coinciding with a striking SSS drop to values < 34.5 and with the maximum dust flux at site M4 (Fig. 4b,c,e,f).

To explain why dust-related pulsed productivity was linked to enhanced Sr/Ca ratios in spring but less so in autumn, we propose a few mechanisms. First, it could be reflecting a change in species composition between the two periods, with much lower Sr being incorporated by small-sized E. huxleyi and G. oceanica, which were the main coccolith-CaCO3 contributors to all Sr/Ca size fractions during the autumn event (Figs. 2 and 4b) compared to much larger species (like Helicosphaera spp., S. apsteinii) which also increased during spring (Figs. 2, 3 and 4b). Another reason could be related to differences in coccolith residence time before reaching the trap at 1200 m, between the two events. Stoll et al.32 report on unchangeable and/or attenuated Sr/Ca signals during high coccolith export periods in the eastern Mediterranean as suggestive of increased export efficiency/scavenging. Indeed, Eliason and Seget34 have reported that both the upper trap at 1200 (studied here) and the lower trap at 3500 m at mooring site M4 intercepted the export event U24 during the same 16-day interval (data not shown). This indicates a high settling velocity (of at least 140 m/day) resulting in rapid settling and little degradation of organic material (up to 51 mg/m2/day of organic matter; Fig. 4f). Dust particles deposited at the surface usually remain in suspension until the occurrence of phytoplankton blooms producing enough organic matter to form larger-sized aggregates (“marine snow”). Both “marine snow” aggregates and faecal pellets produced by zooplankton are likely to incorporate coccoliths and dust particles, which then act as mineral ballast to accelerate their settling velocities45,46,47,48. Since the efficiency of mineral ballasting is a function of the magnitude of the surface bloom17,49,50, our data suggest that surface productivity was particularly high during U24, followed by unusually fast/efficient export, to which rainfall likely contributed. We propose that the observed attenuated Sr/Ca signal in the autumn event partially resulted from a greater mixture of “freshly produced” coccoliths with high Sr/Ca (more abundant in the small fraction), “aged” coccoliths with lower Sr/Ca related to “background” (lower) productivity conditions prior to this event. Our data suggest that the Sr/Ca ratio may be less suitable as a productivity proxy beyond a certain threshold of dust input and/or when dust is rapidly deposited with rain, during which the resulting accelerated ballasting effect is likely to “mask” the freshly produced coccolith “growth” biogeochemical signature. This might be especially true when the coccolith-CaCO3 produced during the bloom is dominated by small size species with lower capacity to fractionate Sr compared to larger species.

Sr/Ca as proxy of productivity in stratified tropical ocean conditions

Some of the highest absolute and normalized Sr/Ca ratios observed during this study clearly coincided with coccolith- and coccolith-CaCO3 flux maximum by deep-dwelling taxa, which were by far the most abundant coccolith sinking species during the entire sampling period (Fig. 4c) and contributing to some of the highest coccolith fluxes ever recorded in the open Atlantic Ocean9. This was the case under the heavily stratified conditions typical of the ITCZ-influenced summer season when the fluxes of F. profunda and G. flabellatus co-increased with the Sr/Ca of the bulk, intermediate and large fractions in July 2013 (U18; Figs. 3a and 4a,b,c). The occurrence of a notable increase of dust fluxes during this period (Fig. 4e) suggest that atmospheric nutrients could have contributed to fuel export production in the LPZ. However, recent observations from the photic zone in the tropical NE Atlantic suggest that only opportunistic surface-dwelling species are likely to benefit from pulsed dust-born nutrient-enrichment10. For example, while the abundance of E. huxleyi and G. oceanica was seen increasing in response to the input of iron and phosphorous supplied by Saharan dust deposition off NW Africa (10–15°N), neither the K-selected UPZ species (more typical of the subtropical gyre (consult the Glossary in the Supplementary Material) nor the deep-dwelling communities had any ecological response. In the same way, despite the slight flux increase of E. huxleyi and G. oceanica in July 2013 (Fig. 4c), one would expect these species to have a more significant response to dust deposition compared to the tropical assemblage, as observed during the spring and autumn dusty events. This was not the case. Instead, high Sr/Ca ratios in summer seemed related to stratified conditions favouring productivity of LPZ species thriving near the deep nutricline. Our data are in good agreement with a previous study reporting high Sr/Ca ratios on the F. profunda size fraction being concomitant with the onset of increased fluxes of F. profunda in the Northern Bay of Bengal, where the coccolith sinking assemblage was also dominated by an unusually high degree of F. profunda29. Larger-sized K-selected taxa more typical of the UPZ, which also increased during this period (e.g., Helicosphaera spp. and most of the species within the group “other taxa”, several of which of larger size and related greater capacity for Sr incorporation compared to E. huxleyi and G. oceanica, as discussed below) may have also contributed to increase the Sr/Ca during summer (Fig. 4b,c,d).

Coccolith size fractions and species-specific Sr/Ca signal

While both field and culture studies show that the Sr/Ca measured in coccolith calcite can be used as a proxy for variations in the rates of growth/calcification rates of coccolithophores (e.g.,20,24,25,30,31,42,51), there is a crucial species-specific size control of Sr coccolith content25. Our data clearly support this, based on the much higher Sr/Ca ratios measured in the large (> 6 μm) coccolith size fractions. Sr/Ca ranges as high as 1.6–12.6 in this fraction (Table 1) have never been reported in natural samples before and are most likely related to the higher calcite content in larger-sized coccolith species, which are more typical of stratified ocean conditions like those of site M4. In our data, this is mainly driven by the CaCO3 contribution of S. apstenii and Pontosphaera spp. which have extremely high bioaccumulation of Sr compared to other species41, but also of Helicosphaera spp., and C. leptoporus which had their higher percentages in the intermediate and large fractions (Fig. 2).

This is in line with the existing notion that larger-sized and more robust coccoliths (i.e., higher amount of coccolith-CaCO3) incorporate more Sr during the cells’ growth, regardless of the degree to which they are ecologically responsive to nutrient-enrichment. Previous studies also describe coccolith Sr partitioning to be variable among genera and species, with larger and more heavily calcified coccoliths usually having higher Sr content compared to smaller and lighter coccoliths20,25,32,52. An especially illustrative example of this is the rare but very large species S. apsteinii; despite its very low export productivity17 and contributing less than 2% of the coccolith assemblages in all size fractions (Fig. 4d and Fig. I—Supplementary Material), this species produced disproportionally high percentages of carbonate both in the original trap samples and in the coccolith size fractions (e.g., samples U18 and U21) (Figs. 2 and 4b). According to coccolith biometric data presented in17, coccoliths of S. apsteinii were by far the largest coccoliths measured in samples from trap M4 (mean length of 15.24 μm), resulting in a coccolith calcite mass of 1665.06 pg. (Table IV in the Supplementary Material). Previous studies have also reported S. apsteinii to have unusually high, and still poorly understood, coccolith Sr/Ca ratios. Hermoso et al.40 highlights the unique ability of S. apsteinii to strongly fractionate Sr, resulting in much higher Sr/Ca compared to any other coccolithophore species (extant or extinct) for which this ratio has been determined. This results in high Sr/Ca ratios not always coinciding with maxima in export fluxes, in line with observations from previous sediment trap studies in the Sargasso Sea21,30 and in the eastern Mediterranean32. An example of this is the high Sr/Ca ratio in the small-size fraction co-occurring to low export production in late March 2013 (U10), when some of the lowest total coccolith- and coccolith-CaCO3 fluxes of the entire trap time-series were recorded (Fig. 3), including those produced by the dominant F. profunda and G. flabellatus (Fig. 4b,c). A similar example occurred in late May 2013 (U14), when high Sr/Ca ratios obtained from all size fractions coincided with a period of moderate coccolith export production and low coccolith-CaCO3 fluxes (Fig. 3). It could be that the Sr/Ca peaks observed during these periods represent the “tail” of the previous high productivity period (winter event—U5–U9, and spring event—U12, prior to March and May, respectively), as hypothesized in previous studies30,32.

The observed correlations between Sr/Ca and several of the most important export production events recorded in late January, mid-April, and late July, suggest an overall efficient coupling between enhanced nutrient-stimulated production in the upper photic zone and coccolith export along the uppermost 1200 m of the ocean. However, given the higher abundance of K-selected large-size coccolith species in typical open-ocean tropical settings such as the studied region, we recommend framing the Sr/Ca data within a multi-proxy approach for more accurately extracting its ecological significance. This should particularly be the case in the presence of S. apsteinii and Pontosphaera spp. Previous authors have also alerted to the yet several aspects that remain poorly understood about the drivers of Sr incorporation by ecological- and morphologically distinct coccolithophore species (e.g.41). For example, we find that the Sr/Ca of all the studied size fractions from trap M4 were generally higher during spring and summer (Figs. 3 and 4a), despite the Sr partitioning having been previously reported as not being affected by changes in growth and calcification rates related to light conditions25,28,51. Whether or not seasonal variations in light conditions during the monitored period also played a role in stimulating the biogeochemical incorporation of Sr remains an open question.

Conclusions

Our synoptic observations of seasonally resolved coccolith Sr/Ca ratios from a sediment trap mooring located in the western tropical North Atlantic during 2012–2013 contribute to understanding the ratio’s potential as a proxy for the biogeochemical effects of Saharan dust deposition in a heavily stratified ocean region. The main conclusions are as follows:

  • Strong correlation between Sr/Ca ratios and pulsed increase of coccolith- and coccolith-CaCO3 fluxes during the windier, cooler, and high MLD conditions of winter and spring support previous observations of nutrient-stimulated export productivity in the photic zone during these periods.

  • The Sr/Ca peak during the pulsed and CaCO3-dominated productivity event of mid-April 2013 coincided with the occurrence of dry dust deposition, in line with a scenario of Saharan dust contributing to stimulate the biological carbon pump through providing nutrients to fuel the surface coccolithophore community in the tropical North Atlantic.

  • Attenuated Sr/Ca during the pulsed and bSiO2-dominated productivity event in October–November 2013 coincided with the occurrence of seasonal inflow of nutrient-enriched Amazon water and wet dust deposition, suggesting that: (a) “freshly produced” coccoliths were, to some degree, diluted with “aged” coccoliths and rapidly dragged/scavenged by dust-induced accelerated sinking velocities; and (b) Sr incorporation was lower due to the small size species E. huxleyi and G. oceanica being the main coccolith-CaCO3 producers during this event.

  • High Sr/Ca ratios during maxima of typical tropical taxa, both surface- and deep-dwelling species, during the ITCZ-influenced summer season, validate the ratio as a biogeochemical proxy for enhanced productivity under the heavily stratified ocean conditions typical of tropical open-ocean settings.

  • Small-size coccolith fractions, where coccolith-CaCO3 was dominated by highly productive but small-sized species F. profunda, G. flabellatus and E. huxleyi, usually presented lower Sr/Ca compared to the other fractions, particularly the large-size fractions which presented unusually high Sr/Ca, to which the abnormally Sr-rich species Scyphosphaera apsteinii likely contributed. This clearly supports the existing notion of larger-sized coccoliths being able to incorporate more Sr during their cells’ growth, regardless of the degree to which they are responsive to transient nutrient enrichment.

While the Sr/Ca data confirm the occurrence of previously reported pulsed export productivity in the region, we recommend that the data should be interpreted taking into consideration the carbonate produced by size-distinct species within the coccolith sinking assemblage, and in the context of a multi-proxy framework. Our study suggests that the Sr/Ca ratio is less suitable to be used as a dust-related productivity proxy beyond a certain threshold of dust flux and/or when dust is deposited with rain, during which the dust ballasting effect may contribute to attenuate the coccolith biogeochemical “growth” original signal. Results from this study highlight the importance of doing more research addressing the physiological, biogeochemical, and abiotic drivers of Sr incorporation by ecologically- and morphologically distinct coccolithophore species.