Limnol. Oceanogr. 66, 2021, 3190–3208 © 2021 The Authors. Limnology and Oceanography published by Wiley Periodicals LLC on behalf of Association for the Sciences of Limnology and Oceanography. doi: 10.1002/lno.11872 Carbonate fluxes by coccolithophore species between NW Africa and the Caribbean: Implications for the biological carbon pump Catarina V. Guerreiro ,1,2* Karl-Heinz Baumann,3,4 Geert-Jan A. Brummer,5 André Valente,1 Gerhard Fischer,3,4 Patrizia Ziveri,6,7 Vanda Brotas,1,8 Jan-Berend W. Stuut5,9 1 MARE, Marine and Environmental Sciences Centre, Faculty of Sciences of the University of Lisbon, Lisbon, Portugal 2 IDL, Instituto Dom Luiz, Faculty of Sciences of the University of Lisbon, Lisbon, Portugal 3 Department of Geosciences, University of Bremen, Bremen, Germany 4 MARUM, Center for Marine and Environmental Sciences, University of Bremen, Bremen, Germany 5 Department of Ocean Systems, NIOZ Royal Netherlands Institute for Sea Research, Den Burg, The Netherlands 6 ICREA, Catalan Institution for Research and Advanced Studies, Barcelona, Spain 7 ICTA-UAB, Institut de Ciència i Tecnologia Ambientals—Universitat Autònoma de Barcelona, Barcelona, Spain 8 Department of Plant Biology, Faculty of Sciences of the University of Lisbon, Lisbon, Portugal 9 Faculty of Earth and Life Sciences, VU Vrije Universiteit, Amsterdam, The Netherlands Abstract Coccolithophores are among the most important calcifying pelagic organisms. To assess how coccolithophore species with different coccolith-carbonate mass and distinct ecological resilience to ocean warming will influence the “rain ratio” and the “biological carbon pump”, 1 yr of species-specific coccolith-carbonate export fluxes were quantified using sediment traps moored at four sites between NW Africa and the Caribbean (i.e., CB-20 N/21 W, at 1214 m; M1-12 N/23 W, at 1150 m; M2-14 N/37 W, at 1235 m; M4-12 N/49 W, at 1130 m). Highest coccolith-CaCO3 fluxes at the westernmost site M4, where the nutricline is deepest along the tropical North Atlan- tic, were dominated by deep-dwelling small-sized coccolith species Florisphaera profunda and Gladiolithus flabellatus. Total coccolith-CaCO3 fluxes of 371 mg m2 yr1 at M4 were followed by 165 mg m2 yr1 at the north-easternmost CB, 130 mg m2 yr1 at M1, and 114 mg m2 yr1 at M2 in between. Coccoliths accounted for nearly half of the total carbonate flux at M4 (45%), much higher compared to 23% at M2 and 15% at M1 and CB. At site M4, highest ratios of coccolith-CaCO3 to particulate organic carbon fluxes and weak correlations between the carbonate of deep-dwelling species and particulate organic carbon suggest that increasing productiv- ity in the lower photic zone in response to ocean warming might enhance the rain ratio and reduce the coccolith- ballasting efficiency. The resulting weakened biological carbon pump could, however, be counterbalanced by increasing frequency of Saharan dust outbreaks across the tropical Atlantic, providing mineral ballast as well as nutrients to fuel fast-blooming and ballast-efficient coccolithophore species. Introduction phytoplankton groups, crucial for the biologically mediated Among mineralizing phytoplankton, coccolithophores fixation and conversion of atmospheric CO2 to organic carbon (Haptophyta) are the most important group of unicellular (via photosynthesis), and its subsequent export to the deep marine algae, covering their cells with a calcified exoskeleton ocean, also termed “biological carbon pump” (BCP) (Cermeño (the coccosphere) composed of numerous tiny, calcified scales et al. 2008). Coccolithophores contribute less to global marine ( 1 to 20 μm across), the so-called coccoliths (Pienaar 1994). primary production ( 1–10%) compared to diatoms, which Coccolithophores are among the main functional are the main exporters of particulate organic carbon (POC) in the modern ocean (e.g., Poulton et al. 2007; Malviya et al. 2016 and refs. therein). Yet, they are among the main *Correspondence: cataguerreiro@fc.ul.pt exporters of particulate inorganic carbon (PIC) from the pho- This is an open access article under the terms of the Creative Commons tic zone to the deep sea, thereby performing a unique dual Attribution-NonCommercial-NoDerivs License, which permits use and dis- tribution in any medium, provided the original work is properly cited, the role in the ocean’s carbon cycle since 220 million years ago, use is non-commercial and no modifications or adaptations are made. when they first appeared in the geological record (e.g., Rost and Riebesell 2004; Broecker and Clark 2009; Hutchins 2011). Additional Supporting Information may be found in the online version of this article. Following the death of the cells, some of the coccolith-CaCO3 3190 Guerreiro et al. Coccolithophore species-specific carbonate fluxes may dissolve within the photic zone (Poulton et al. 2006) while Romero et al. 2020). As tropical oceans provide a benchmark most of it sinks out of the surface waters inside zooplankton for future primary productivity in increasingly warmer and fecal pellets and marine snow aggregates or incorporated nutrient-depleted waters, it is crucial to constrain the main into the agglutinating shells of tintinnid microzooplankton drivers of coccolith-CaCO3 export in these regions. (Knappertsbusch and Brummer 1995 and refs. therein). Here, we provide a synoptic, synchronized, and biweekly The ratio of biogenic PIC to POC export from the surface resolved record of coccolith-CaCO3 fluxes of the more abun- down to the deep ocean, termed the rain ratio, is a key parame- dant species across the entire tropical North Atlantic, in rela- ter for biogeochemical models exploring the long-term effi- tion to atmospheric and oceanographic processes. The study ciency of atmospheric CO2 drawdown (Archer and Maier- builds on an earlier transatlantic transect extending down- Reimer 1994; Zondervan 2007). It represents the opposing wind from NW Africa (off Cape Blanc) to the Caribbean relationship between the production of organic matter acting (Fig. 1; Supplementary Table S1), right underneath the core of as a CO2 sink (via photosynthesis) and the production of the main dust plume originating from the Saharan desert CaCO3 acting as a CO2 source (via calcification) (e.g., Rost and (Guerreiro et al. 2019). Here, we relate coccolith-CaCO3 fluxes Riebesell 2004; Hutchins 2011 and refs. therein). As the rain with other biogenic particle fluxes collected at four trap sites, ratio is what largely determines the biologically mediated CO2 including total biogenic carbonate, biogenic silica and organic flux between the ocean and the overlying atmosphere matter, as well as POC (e.g., Fischer et al. 2012, 2013; Korte (e.g., Rost and Riebesell 2004; Cermeño et al. 2008), any et al. 2017). The molar ratio between the fluxes of coccolith- changes in the PIC and POC fluxes triggered by shifts in the CaCO3 and POC (Coccolith-CaCO3/POC) and the species- coccolithophore communities are likely to alter the earth’s cli- specific coccolith-ballasting effect are analyzed to investigate mate. While nutrient depletion triggered by increasing ther- how ecologically different species modulate the export of mal stratification is more likely to favor coccolithophores over organic carbon. The spatiotemporal flux distribution of atmo- diatoms (e.g., Rivero-Calle et al. 2015), the subsequent spheric dust along the transect (van der Does et al. 2020) is increase of PIC production is likely to weaken the efficiency of also included to explore links between Saharan dust deposi- the BCP through increasing the rain ratio. This may however tion and coccolith-CaCO3 export production. For understand- be counterbalanced by coccolithophores providing ballast ing the processes involved, we combine our in situ material acting to accelerate the sinking velocities of organic- transatlantic observations with time series of surface-water rich aggregates formed at the surface (Armstrong et al. 2002; concentrations of chlorophyll a (Chl a) and PIC determined Ziveri et al. 2007; Fischer and Karakas 2009). This is supported by satellite remote sensing during the same year. by up to 83% of global POC fluxes being reportedly linked to CaCO3 (Klaas and Archer 2002), reflecting the high density of calcite, and its higher abundance in the open ocean compared Environmental settings to terrigenous material (e.g., Klaas and Archer 2002; Ziveri Our study is based on time-series material from four sedi- et al. 2007). ment trap moorings collecting sinking particles during  1 yr Given the unequivocal importance of coccolithophores for along a transatlantic transect extending from the Caribbean the ocean’s carbon cycle (Rost and Riebesell 2004; into the mesotrophic region offshore Cape Blanc (NW Africa) Zondervan 2007; Hutchins 2011), significant attention has (Fig. 1; Supplementary Table S1). Traps M4 (12 N 49 W), M2 been given to exploring the dynamics of their export produc- (14 N 37 W), and M1 (12 N 23 W) collected particles in tion over the last few decades, particularly in the North Atlan- synchronous intervals of 16 d, from 19 October 2012 to tic (e.g., Broerse et al. 2000; Sprengel et al. 2002; Guerreiro 07 November 2013, while trap CB (21 N 20 W) collected par- et al. 2019). Time-series sediment traps collecting settling par- ticles at variable intervals, from 04 October 2012 to ticles (organic and inorganic) over periods of time ranging 21 February 2014. Details of the mooring equipment, the from weeks to years are the standard approach for studying deployment/recovery of the sediment traps, and the treatment the impact of seasonal variations in coccolithophore export of the recovered sample bottles are described in the cruise productivity on the functioning of BCP (e.g., Milliman 1993; reports by Stuut et al. (2013) and Fischer et al. (2012, Baumann et al. 2005; Salter et al. 2015). However, most of the 2013, 2014). existing studies express the contribution of coccolithophores Wind-forced oceanographic processes, both geostrophic to total carbonate export in terms of total coccolith-CaCO3 and seasonal, and transatlantic Saharan dust deposition have flux without discussing its dependence on the spatiotemporal been recently reported to influence the distribution of dynamics of ecologically distinct species. In addition, most coccolith species fluxes along the studied transect during the observations report from mid- to high-latitude regions in the monitored sampling interval (Guerreiro et al. 2017, 2019). NE Atlantic Ocean, where coccolith production is more sea- The region is characterized by westward deepening of the sonal and dominated by fast-blooming placolith-bearing spe- nutricline related to large-scale geostrophic deepening of cies (e.g., Emiliania huxleyi) compared to low-latitude open- the thermocline from the eastern to the western tropical ocean regions (e.g., Broerse et al. 2000; Sprengel et al. 2002; North Atlantic (Merle 1980; Katz 1981). In contrast, the 3191 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fig 1. Location of the trap mooring sites M4, M2, M1, and CB and a schematic representation of the main surface currents in the equatorial Atlantic Ocean, including the Mauritanian Current (MC) flowing northwards off NW Africa (see Guerreiro et al. 2019 and refs. therein). dynamics of the seasonal mixed layer depth is function of April and October–November 2013, triggered by combination wind forcing over the uppermost  75–80 m water depth of enhanced dust deposition, wind-forced mixing, and fresh- along the transect (Guerreiro et al. 2019). Seasonality was water dispersal linked to the seasonal inflow of the Amazon weak and atmospheric conditions were similar at all four loca- River Plume into the western part of the transect (e.g., Molleri tions, modulated by the seasonal migrations of the Intertropi- et al. 2010; Guerreiro et al. 2017; Korte et al. 2020). The cal Convergence Zone (Basha et al. 2015; Guerreiro sediment trap samples bulk composition at all four sites is et al. 2019; see Fig. 1). Higher precipitation rates, weaker trade presented in Supplementary Table S4. winds, and enhanced thermal stratification occurred in sum- mer and autumn, while little to no precipitation, intensified winds and a deeper mixed layer depth in winter and spring. Methods Toward the northeast, conditions were less influenced by the Laboratory and microscope analysis Intertropical Convergence Zone but mostly determined by For the detailed description of the procedures regarding the variable magnitude of yearlong upwelling off Cape Blanc treatment of the sediment trap samples, the readers are (Pastor et al. 2013 and refs. therein). referred to Fischer et al. (2016) and Korte et al. (2017). Proce- Carbonate fluxes were on average higher at M4 (41 g m2 dures for the coccolith analysis detailed in Guerreiro d ) and M1 (38 g m2 d1) compared to the oligotrophic site 1 et al. (2017, 2019), including the coccolith species counting M2 (22 g m2 d1), while the fluxes of organic matter and bio- method and the calculation of coccolith fluxes. The coccolith- genic silica where higher at M1, followed by M4 and M2 derived CaCO3 export fluxes were determined using the mass (Korte et al. 2017). New flux data from trap CB indicate the equation of Young and Ziveri (2000), according to which the highest fluxes of carbonate (92 g m2 d1) and organic matter coccolith mass of distinct coccolithophore species is along the transect, reflecting the high productivity of calcify- expressed as: ing plankton induced by upwelling (Fischer et al. 2016; Guerreiro et al. 2019; Romero et al. 2020). While the organic Coccolith calcite ðpgÞ ¼ 2:7  ks x l3 matter fluxes at CB were only slightly higher compared to the eastern site M1, the fluxes of biogenic silica were lower. Saha- in which 2.7 = density of calcite (CaCO3); ks = shape constant; ran dust fluxes available for sites M1 (14 g m2), M2 (3 g m2), l = coccolith size (mostly distal shield length). The size of around and M4 (3 g m2) were highest at M1 forced by strong trade- 3500 coccoliths was measured from most species present in samples wind intensities at lower altitudes during spring, as well as fol- from traps M1, M2, and M4 using a Zeiss DSM 940A scanning lowing the occurrence of sporadic dust storms during sum- electron microscope. Most of the used coccolith shapes come from mer/autumn (van der Does et al. 2020). Site M4 was marked Young and Ziveri (2000). For some of the species, the shape con- by the occurrence of two high surface productivity events in stants have been slightly adjusted within their shape range, as 3192 Guerreiro et al. Coccolithophore species-specific carbonate fluxes indicated in Young and Ziveri (2000), such that a meaningful rela- Supplementary Table S3). The PIC product has been validated tionship would be achieved between the coccolith masses of the in various locations with in situ measurements (e.g., Balch measured species. The measured coccolith lengths and the subse- et al. 2005), providing considerable insights into the pheno- quently determined species-specific coccolith masses are provided logical characteristics of coccolithophore surface blooms at a in Supplementary Table S2. For trap CB, the coccolith-CaCO3 global scale (e.g., Hopkins et al. 2015). Both Chl a and PIC masses are based on coccolith measurements performed by data were retrieved for each trap location using a 2  2 Köbrich and Baumann (2009) for the same location. The only latitude-longitude area around the trap location and averaged exceptions to this include Gladiolithus flabellatus, Hayaster per- for each sediment trapping interval. The box, corresponding plexa, Oolitothus antillarum, and Umbilicosphaera hulburtiana for to  108  108 Nm (1 = 59 Nm), was taken as representa- which we used the actual measurements of coccoliths from the tive of the catchment area of a trap deployed at 1200 m depth, neighboring trap site M1. Carbonate content of rare coccolith spe- given the mean sinking speed of marine phytoplankton and cies for which we did not perform size measurements was based on algal aggregates (e.g., Waniek et al. 2000). Data were processed coccolith size information available from Nannotax (http://www. for the study period of October 2012–October/November mikrotax.org/Nannotax3/) and estimates from Young and 2013 at sites M1, M2, and M4, and up to January 2014 at Ziveri (2000) (Supplementary Table S2). A multivariate statistical site CB. analysis using Principal Components Analysis (PCA, correlation mode; PAST-3 software) and a Pearson correlation coefficient matrix were undertaken to explore and statistically support the Results and discussion observed relationships between species-specific coccolith-CaCO3 Coccolith-carbonate export production fluxes and the other particle fluxes (Supplementary Information). The coccolith-CaCO3 fluxes were highest and most season- ally variable at the opposite ends of the transatlantic array, Satellite remote sensing particularly at the westernmost site M4, while lower and less Time series of sea surface concentrations of Chl a and PIC seasonally variable fluxes occurred at the more central sites in were obtained from satellite remote sensing, used as proxies between, particularly at mid ocean site M2 (Fig. 2a). From for phytoplankton biomass (see Guerreiro et al. 2017, 2019) October 2012 to October–November 2013, site M4 received a and coccolithophore carbonate surface production across the total of 371 mg m2, followed by M1 (130 mg m2) and M2 transect (e.g., Balch et al. 2005) (data sets details provided in (114 mg m2), while from October 2012 to January 2014, site Fig 2. Fluxes of coccolith-CaCO3 (light green bars) and unspecified-CaCO3 (black line) (a), deviation from the annual mean coccolith-CaCO3 flux at each trap site (dark gray bars) and percent contribution of coccoliths carbonate to total CaCO3 flux (b), total carbonate flux (blue line) and particulate inorganic carbon (PIC) concentration at the ocean surface (c) at sites M4, M2, M1, and CB. Light and dark gray vertical bars indicate summer and autumn, respectively. 3193 Guerreiro et al. Coccolithophore species-specific carbonate fluxes CB received a total of 165 mg m2. Highest fluxes at M4 and westwards of CB ( 0.01 mg L1 at site M1) suggests a much CB ultimately reflect the higher coccolith export production lower coccolith-CaCO3 export at the other sites, which is not closer to the Caribbean and to the upwelling-induced high- what we find in the traps, especially in M4 (Fig. 2c). productivity region off NW Africa, compared to the more Such seemingly contradicting observations seem related to open and oligotrophic sites M1 and M2 (Fig. 2a). The 2–3 technical limitations related to the maximum satellite- times higher fluxes at M4 mirror the much higher coccolith observation depth range, reflecting the transition from a export productivity reported by Guerreiro et al. (2019) com- more markedly dynamic region with a shallower nutricline at pared to the other trap sites, including the mesotrophic and CB toward a nearly permanently stratified ocean setting (see carbonate-dominated region off Cape Blanc. Guerreiro et al. 2019). That coccolith-CaCO3 fluxes at site CB Albeit occasional mismatches, the temporal distribution of were mostly produced by fast-blooming surface-dwelling spe- carbonate produced by coccolithophores was similar to that cies including the dominant E. huxleyi (Fig. 3), probably con- of the total carbonate flux (Fig. 2a,c) and of the unspecified- tributes to produce larger and denser surface blooms that are CaCO3 flux. The latter is assumed to include carbonate from optically reflective enough to be detected by remote sensing coccolith fragments and coccoliths within aggregates (which (e.g., Holligan et al. 1993; Balch et al. 2018). On the other were not included in the counts) and/or produced by other hand, highest coccolith contribution to carbonate in October calcifying planktonic organisms (Fig. 2a). Significant differ- 2012 at site CB not reflected in terms of PIC (Fig. 2b,c) sug- ences were noticed, however, between the distribution of sur- gests that either coccolithophores were temporarily thriving face PIC and carbonate fluxes from our in situ trap records below the satellite depth range or that coccolith-CaCO3 pro- (Fig. 2c). Much higher PIC with markedly distinct seasonality duction and export were not coupled. This illustrates the at site CB compared to the other sites indicates that the satel- extent to which the remotely sensed PIC is strongly limited lite PIC data only represent a portion of the coccolith-CaCO3 by the optical depth of the water column (Gordon and production between NW Africa and the Caribbean, particu- McCluney 1975) in spite of its successful application to larly at the more open ocean sites M1–M4. Site CB was the explore the occurrence and frequency of large-scale region where surface PIC was both highest (up to coccolithophore surface blooms in the North Atlantic 0.032 mg L1 in early spring) and more seasonally similar to (e.g., Balch et al. 1991; Holligan et al. 1993; Shutler coccolith- and total-carbonate fluxes, all three strikingly et al. 2013). This presents a major constraint for detecting peaking during spring (Figs. 2c). We also observed a PIC pulse coccoliths in tropical regions where coccolithophores typi- in early October 2013 at the westernmost site M4 (up to cally thrive at greater depths (Poulton et al. 2017). While 0.02 mg L1 in M4-23), which did coincide with surface ocean massive surface blooms shown by satellites may have a cru- fertilization (recorded by the trap at 1200 m depth in M4-24; cial biogeochemical impact on regional scales (Iglesias- Guerreiro et al. 2017; Korte et al. 2020), and reaching concen- Rodríguez et al. 2002), they probably represent a minor frac- trations almost as high as the spring peak recorded offshore tion compared to the background levels of CaCO3 produced Cape Blanc (Fig. 2c). Still, the drastic drop of surface PIC south annually on a global scale. Fig 3. Percent coccolith-CaCO3 production by the most abundant sinking coccolithophore taxa at trap sites M4, M2, M1, and CB. Taxa similar in spatio- temporal distribution were grouped together. 3194 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Species contributions Species constituting less than 5% of the total coccolith sinking As expected, large-sized taxa such as Helicosphaera spp. and assemblage (Guerreiro et al. 2019) contributed a total of Calcidiscus leptoporus showed the highest coccolith-CaCO3 1 mg m2d1. mass, followed by Rhabdosphaera spp., Umbellosphaera spp, In terms of seasonality, higher coccolith-CaCO3 fluxes in Umbilicosphaera spp., Gladiolithus flabellatus, Gephyrocapsa spring and summer along the transect (Fig. 2b) reflect the sea- oceanica, G. muellerae, Florisphaera profunda, Emiliania huxleyi sonal latitudinal migration of the Intertropical Convergence and G. ericsonii (Fig. 4; Supplementary Table S3). While differ- Zone modulating the temporal dynamics of taxa with distinct ences between taxa were a persistent feature at all four trap ecological preferences (Basha et al. 2015; Guerreiro locations, certain species varied substantially in carbonate et al. 2017). Enhanced thermal stratification resulting from mass along the transect. For example, C. leptoporus showed a weaker winds across the tropical North Atlantic in summer clear coccolith mass increase from west to east, while and autumn led to increased carbonate produced by species G. flabellatus revealed the opposite tendency. Taxa such as more typical of the lower photic zone (LPZspp.) (Figs. 3 and G. oceanica, Rhabdosphaera spp. and U. foliosa had their highest 5c). By contrast, enhanced wind-forced mixing leading to carbonate mass at site CB and their lowest at site M1, while cooler and more nutritious surface conditions in winter and Helicosphaera spp. was more heavily calcified at sites CB and spring has favored carbonate export by placolith-bearing taxa M2 (Fig. 4). more typical of the upper photic zone (UPZspp). The subsequently determined coccolith-CaCO3 fluxes rev- Placolith-bearing taxa produced higher carbonate fluxes at ealed, however, that both large-sized Helicosphaera spp. both ends of the transect, with C. leptoporus dominating at the (1.6 mg m2d1) and smaller-sized species such as northeastern site CB and gephyrocapsid species at the western G. flabellatus (2.3 mg m2d1) and F. profunda (1 mg m2d1) site M4. Higher carbonate production by C. leptoporus, contributed the most to the mean transatlantic coccolith- E. huxleyi, and Umbilicosphaera spp. at site CB reflects their CaCO3 production, with 26%, 17%, and 13%, respectively. higher affinity for more seasonally variable and upwelling- The latter were followed by C. leptoporus (0.8 mg m2d1; influenced settings off Cape Blanc compared to the other loca- 11%), and by G. oceanica, Umbellosphaera spp., Umbilicosphaera tions (Figs. 3 and 5a, see Guerreiro et al. 2019). Similar sea- spp., and E. huxleyi (0.2–0.5 mg m2d1; 4–5%), while the sonal distribution between the fluxes of coccolith-CaCO3 by remaining taxa individually contributed less than 4%. Of these taxa and of total CaCO3 at CB highlights the link these, G. ericsonii, G. muellerae, and Reticulofenestra sessilis pro- between the productivity of placolith-bearing taxa and car- duced less than 0.1 mg m2d1 (Supplementary Table S4). bonate production in this mesotrophic region (Fischer and Fig 4. Mean species-specific carbonate mass (in pg) of coccoliths species from the most abundant sinking species at trap sites M4, M2, M1, and CB. 3195 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fig 5. Fluxes of coccolith-CaCO3 (left y-axes) and percentage of total carbonate (right y-axes) produced by the most abundant coccolithophore taxa at sites M4, M2, M1, and CB. Taxa with a similar spatiotemporal distribution (see Guerreiro et al. 2019) were grouped into: (a) placolith-bearing E. huxleyi, G. oceanica, G. muellerae and G. ericsonii, Umbilicosphaera spp. and C. leptoporus, (b) Helicosphaera spp., Rhabdosphaera spp., Umbellosphaera spp., R. sessilis, and (c) lower photic zone species (LPZspp) F. profunda and G. flabellatus. Karakas 2009; Romero et al. 2020). Of these species, the low- components analysis (PC1 and PC2, Supplementary Fig. S1 flux and large-sized C. leptoporus contributed the most (up to and Table S5), providing evidence on the potential of short- 41% of coccolith-CaCO3 production in the winter–spring term upper photic zone productivity for enhancing transition). coccolithophore contribution to total carbonate flux in highly Pulsed coccolith-CaCO3 export production by G. muellerae, stratified tropical settings. G. ericsonii, G. oceanica, and E. huxleyi in January, April, and Helicosphaera spp. were by far the dominant coccolith- October–November 2013 at site M4 (M4-6-7, M4-12, CaCO3 producers at the more central sites, particularly at M1, and M4-24, respectively) contributed to slightly increase the but also reaching high fluxes at site CB. That these species pro- flux temporal variability in this area compared to M2 and M1 duced more carbonate during both spring and summer suggest (Fig. 5a). Carbonate fluxes produced by these species during a strong affinity for well-illuminated conditions in the photic such high surface productivity events were similar or even zone but a lesser dependency on stratification and nutrient higher compared to their maxima at the upwelling-influenced availability/supply (Figs. 3 and 5b). Rhabdosphaera spp. and site CB (Fig. 5a). During the spring and autumn events, min- Umbellosphaera spp. produced higher carbonate fluxes at sites eral dust, organic matter, and biogenic silica, as well as the M4 and M2, showing a rather weak seasonality along the UPZspp/LPZspp ratio (i.e., ratio between coccolith fluxes of entire transect. Carbonate by Rhabdosphaera spp. was slightly the upper photic zone species E. huxleyi and G. oceanica, and higher at M2 and M1 compared to Umbellosphaera spp., which the lower photic zone species F. profunda and G. flabellatus, were more productive at M4 throughout the year, and during used as a proxy for the nutricline depth dynamics; Guerreiro spring at CB. R. sessilis revealed low coccolith-CaCO3 along et al. 2017, 2019) (Figs. 6a,f), also increased. Such combina- the entire transect but slightly higher at site M2 (Figs. 3 tion of proxies marks the occurrence of surface ocean fertiliza- and 5b). tion resulting from dust deposition combined with wind- Higher species-specific coccolith-carbonate production forced vertical mixing and Amazon water dispersal, all three along the transect by Helicosphaera spp. and C. leptoporus processes seasonally influencing site M4 (Guerreiro et al. 2017; (Fig. 4) supports previous notions that total coccolith- Korte et al. 2017, 2020). These ecologically r-selected events carbonate flux is largely determined by larger species in spite were clearly illustrated in the results from the principal of their lower productivity (e.g., Broerse et al. 2000; Young 3196 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fig 6. Spatiotemporal variation of (a) UPZspp/LPZspp ratio (light green) in coccolith fluxes of E. huxleyi and G. oceanica typical of the upper photic zone (UPZ), and F. profunda and G. flabellatus typical of the lower photic zone (LPZ) (used as a proxy for nutricline depth dynamics; Guerreiro et al. 2017, 2019), and surface chlorophyll a concentrations (Chl a, black line); (b) total fluxes of particulate organic carbon (POC) (green) and molar ratio between fluxes of biogenic silica (bSiO2) and total CaCO3 (bSiO2/CaCO3, orange line) as a proxy for the ratio of silicifying to calcifying plankton (see Cermeño et al. 2008), (c) rain ratio (CaCO3/POC, in blue) and Coccolith-CaCO3/POC molar ratios (black line); (d) relative contribution of coccolith-CaCO3 to total CaCO3 fluxes by the most important carbonate producing coccolith taxa; (e) total coccolith-CaCO3 fluxes (blue) and surface ocean concentrations of particulate inorganic carbon (PIC) (black line); and (f) fluxes of (from bottom to the top) CaCO3, organic matter, biogenic silica, and lithogenic dust (M1, M2, and M4 data from Korte et al. 2017 and van der Does et al. 2020). The mass flux of unspecified particles (i.e., residual fraction) results from subtracting the weights of biogenic constituents (i.e., CaCO3, organic matter and biogenic silica) from the total mass and has been used as a proxy for dust deposition in the Atlantic Ocean (Jickells et al., 1998), including our study area (e.g., Fischer et al. 2016; Guerreiro et al. 2017; Korte et al. 2017). and Ziveri 2000; Baumann et al. 2004). This is also the case for 15 mg m2d1 at site M4 (Supplementary Table S3). Still, and the rare but very large-sized species Scyphosphaera apsteinii, albeit their much lighter and smaller-sized coccoliths, included in the category “others,” which produced up to G. flabellatus and F. profunda were the most important 3197 Guerreiro et al. Coccolithophore species-specific carbonate fluxes carbonate producers along the transect, contributing to 2–3 values show a clear westward increasing contribution of times higher coccolith-CaCO3 flux at M4 (16 mg m2 d1) coccolithophores to total carbonate production along the compared to the other sites (8 mg m2d1 at CB and tropical North Atlantic. While the percent contribution at 5 mg m2d1 at M1 and M2) (Figs. 3 and 5c). Higher mean the most oligotrophic site M2 is in line with some of the loca- fluxes by LPZspp at M4 (10 mg m2d1 representing 66%), tions above, site M4 was substantially higher, also compared even compared to large-sized Helicosphaera spp. at site M1 to, for example, the 39% and 24% in the subtropical and tem- (52%) and C. leptoporus at site CB (27%) (Supplementary perate eastern North Atlantic, respectively (Broerse Table S4), highlight the importance of coccolith-CaCO3 pro- et al. 2000), and to 28–33% off the Canary Islands (Sprengel duction in the lower photic zone in highly stratified settings et al. 2002). Köbrich and Baumann (2009) report a contribu- typical of tropical oceans (Guerreiro et al. 2019). tion of 18% in 1988–1989 at site CB that is similar to M1 and The observed spatiotemporal patterns in species-specific CB during our studied period, but lower than at M4 and M2. carbonate fluxes were also reflected in terms of their relative The only regions in the North Atlantic reporting even lower contributions to total carbonate flux. Maximum coccolith per- percentages include the Bay of Biscay (12%; Beaufort and cent contribution of 82% in the autumn-winter 2012–2013 at Heussner 1999) and the western tropical North Atlantic (9% at site CB was dominated by C. leptoporus (up to 32%) (Fig. 5a), 13 N/54 W; Steinmetz 1991). Although total mass fluxes to which Helicosphaera spp. (up to 13%) (Fig. 5b) and along the tropical North Atlantic are all dominated by bio- G. oceanica (12%) (Fig. 5a) also contributed. At site M4, highest genic carbonate (Fig. 6f; Supplementary Table S4), M4 was the carbonate contributions were during the autumn 2012 and only location approaching the conventional notion of summer 2013, dominated by F. profunda and G. flabellatus coccolithophores dominating carbonate production and (up to 37%) (Fig. 5c), whereas Helicosphaera spp. contributed export in the global ocean (e.g., Baumann et al. 2004; the most throughout the year at sites M2 and M1 (up to 22%) Baumann and Freitag 2004; Broecker and Clark 2009). (Fig. 5b). Similar fluxes of coccolith- and total-carbonate at sites M2, M1 and CB compared to previous trap studies suggest that Coccolithophores and carbonate export coccolith dissolution did not cause the observed low percent- Both the total- and coccolith-carbonate export fluxes across ages, consistent with the notion that coccolith preservation in the transect were generally in line with previous trap studies North Atlantic sediments is excellent (lysocline at  4000– in the North Atlantic. While carbonate fluxes at sites M4, M2, 4500 m; Milliman 1993; Milliman et al. 1999 and refs. and M1 (15, 8, and 14 g m2 yr1, respectively) were much therein). A certain proportion of broken coccoliths that were higher compared to 2–5 g m2 yr1 reported as typical of cen- not included in the coccolith-CaCO3 budget could, however, tral ocean gyres, site CB (34 g m2 yr1) falls within the 30– contribute to underestimate the coccolithophorid percent 40 g m2 yr1 described as characteristic of eastern boundary contributions. This is supported by previous traps studies upwelling systems (Milliman 1993) (Supplementary Table S4). reporting that such coccolith fragments add to the coccolith- Likewise, coccolith-CaCO3 fluxes of 1.8 g m2 yr1 at M2 and dominated sediment size fraction (Knappertsbusch and M1 were slightly higher compared to previous studies Brummer 1995; Ziveri et al. 2007). Such biogenic remnants, from the western tropical North Atlantic (13 N/54 W) which also include other carbonate particles of both biogenic (Steinmetz 1991) and near the Canary Islands (ESTOC- (including juvenile foraminifera) and non-biogenic origin, 29 N/17 W and LP-29 N/17 W; Sprengel et al. 2002). At site appear to significantly add to the so-called biogenic carbonate M4, however, mean fluxes of 5.8 g m2 yr1 were higher not powder (Ziveri et al. 2007). only compared to M1, M2, and CB and to the sites mentioned Biological breakage and partial dissolution of coccoliths by above, but also to other locations at higher latitudes in the zooplankton grazing, as well as driven by microbial oxidation North Atlantic. These include the Sargasso Sea (Honjo 1978), of organic matter in flocculates and aggregates, could also the Bay of Biscay (MS1-44.5 N/2.5 W; Beaufort and explain the much lower coccolith percent contributions at the Heussner 1999), and the temperate and subtropical North eastern sites CB and M1 (Knappertsbusch and Brummer 1995; Atlantic (NABE48-48 N/21 W and NABE34-34 N/21 W; Milliman et al. 1999 and refs. therein). Higher surface Chl Broerse et al. 2000; Ziveri et al. 2000). In contrast, mean fluxes a concentrations, as well as higher UPZspp./LPZspp. ratios and of 2.9 g m2 yr1 offshore Cape Blanc were nearly half of the POC fluxes off Cape Blanc (Fig. 6a,b), rooted in strong and 4.9 g m2 yr1 earlier reported for the same site (Köbrich and yearlong coastal upwelling (Pastor et al. 2013), could result in Baumann 2009), pointing to substantial interannual variabil- higher carbonate removal rates compared to the highly strati- ity in this mesotrophic region. fied open ocean regions at M4 and M2 (Milliman et al. 1999). Albeit the similar or even higher export of coccolith-CaCO3 This is in line with reportedly observed differences in species observed across the tropical North Atlantic compared to more diversity between plankton and sediment traps samples linked seasonal regimes at higher latitudes, the average coccolith to the selective destruction of smaller and/or more delicate contribution to the total carbonate was only 45% at M4, 23% forms during sinking (e.g., holococcolithophores) (Baumann at M2, and 15% at M1 and CB (see Figs. 2b and 6d). These et al. 2005 and refs. therein). Some authors go even further to 3198 Guerreiro et al. Coccolithophore species-specific carbonate fluxes suggest that only the coccoliths and coccospheres that sink Hernandez et al. 2020) (see Table 2). Therefore, the volumet- incorporated intact inside of fecal pellets and marine snow ric/biometric method was assumed to be the most advanta- aggregates, or into the shells of tintinnid microzooplankton, geous for our transatlantic data set, since both the taxonomic are likely to be preserved in the traps (Knappertsbusch and analysis and the morphometric measurements of the Brummer 1995; Beaufort and Heussner 1999; Baumann coccoliths had been performed using a scanning electron et al. 2005 and refs. therein). Thereby, export productivity is microscope (Guerreiro et al. 2017, 2019). This is undoubtedly also a function of daily fluctuations in the production of larger the most accurate approach for identifying coccoliths to the biogenic particles (Knappertsbusch and Brummer 1995; species level, as well as for measuring the coccoliths length, Fischer et al. 2009) and of the factors modulating their own allowing to produce a consistent coccolith biogeochemical seasonal ecological dynamics. Whereas these processes may data set without the errors linked to the optical/birefringence have led to underestimate the sinking coccoliths from which analysis. It also provided for accurately measuring the the coccolith-CaCO3 fluxes were calculated, enhanced propor- coccoliths’ length of all species, independently of their thick- tions of the “biogenic carbonate powder” would still contrib- ness and birefringence. ute to the observed total carbonate fluxes. This is supported Overall, our study suggests that existing notions on the rela- by the general agreement between the coccolith- and tive contribution by distinct calcifying plankton groups to total unspecified-CaCO3 fluxes, which may include a significant carbonate export may not be accurately constrained, albeit the coccolith contribution (Fig. 2a). potentially underestimated coccolith-CaCO3 contributions Other calcifying planktonic organisms (e.g., foraminifera, presented here. Quantifying the coccolith-CaCO3 inside fecal pteropods, heteropods, calcareous dinophytes) could have also pellets and marine snow aggregates (see Knappertsbusch and contributed more carbonate than it is usually assumed. As Brummer 1995; Beaufort and Heussner 1999) and within most studies dealing with these issues do not include forami- the “biogenic carbonate powder” (Ziveri et al. 2007) is likely to nifera < 150 μm, which make up a large but poorly con- increase the accuracy of estimated coccolithophore contribution strained fraction of the total foraminifera, their total relative to total carbonate flux. contribution to the carbonate flux is likely to be under- estimated (Brummer et al. 1986). This is supported by occa- Coccolithophore carbonate and the rain ratio sional pulsed coccolith-CaCO3 fluxes coinciding with In spite of the much higher POC fluxes at sites M1 and CB decreased coccolith percent contributions to carbonate flux (Fig. 6b), as expected from its higher proximity to a productive along the transect, suggesting that other calcifying organisms eastern boundary upwelling system (e.g., Chavez et al. 2011; were playing a greater role for carbonate production Signorini et al. 2015), the rain ratio remained fairly stable (e.g., April and October–November 2013 at M4, early March along the entire transect, only slightly lower at site M1 2013 at M1, and February–March 2013 at CB) (Figs. 2b (Fig. 6c). This highlights the importance of biogenic carbonate and 5a). production toward the eastern tropical North Atlantic. The Finally, the volumetric/biometric method used in our study Coccolith-CaCO3/POC ratios, however, clearly decreased from (Young and Ziveri 2000) could have also contributed to under- west to east, reaching 2–4 times higher mean ratios at sites M4 estimating the species-specific coccolith-CaCO3 fluxes, since it and M2 (Fig. 6c), in line with the westward increase of the is only applicable for intact coccoliths and assuming an identi- coccolith percent contribution to total carbonate flux dis- cal shape and width proportions for all individual coccoliths cussed above. This clearly suggests an increasing role of of the same species (Ziveri et al. 2007). While some of these coccolithophores in modulating the rain ratio toward more issues have been reportedly reduced with methods based on stratified conditions, compared to other calcifying organisms. the coccoliths’ optical properties (i.e., calcite birefringence Higher Coccolith-CaCO3/POC ratios at M4 were mostly used for estimating the coccolith-carbonate mass; Beau- produced by lower photic zone species F. profunda and fort 2005; Beaufort et al. 2014; Bollmann 2014, Fuertes G. flabellatus (see Figs. 6d and 7b). The negative correlation et al. 2014), the latter are not exempt of uncertainties. In fact, between the Coccolith-CaCO3/POC and the UPZspp/LPZspp- important errors have been reported, driven by difficulties in CaCO3 ratios (Fig. 8b) shows just that, while also illustrating calibrating the microscope light intensity for measuring the increasing role of coccolithophores in rising the rain ratio coccolith calcite, with subsequent uncertainties on the with the deepening of the nutricline toward the west. This is coccolith mass estimates. Secondly, the optical method is not consistent with the alkalinity model of Sarmiento et al. (2002) applicable to weakly birefringent coccoliths nor to coccoliths yielding higher rain ratios in the tropics possibly linked to a thicker than a certain valid measurable thickness. This results dominant contribution by low-latitude non-bloom forming in the systematic weight underestimation in coccolith calcite coccolithophores to global calcification. In addition, discrep- from larger taxa compared to the morphometric method, and ancies in the size and carbonate mass of G. flabellatus along the exclusion of the larger coccoliths that are often the most the transect, with twice heavier coccoliths at M4 and M2, important coccolith-carbonate contributors (Beaufort seem to have substantially contributed to increase the total et al. 2014; Bollmann 2014; Fuertes et al. 2014; Rigual coccolith-CaCO3 fluxes and Coccolith-CaCO3/POC ratios 3199 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fig 7. Spatiotemporal variation in the molar ratios of Coccolith-CaCO3 to particulate organic carbon (POC) fluxes produced by (a) Helicosphaera spp. (yellow line) and C. leptoporus (brakish green) and (b) LPZ species F. profunda and G. flabellatus (black) and the UPZ species E. huxleyi and G. oceanica (green line) (where UPZ and LPZ stand for upper- and lower-photic zones). Fig 8. Relationships between (a) percentages of UPZspp-CaCO3 and LPZspp-CaCO3, and between the UPZspp/LPZspp-CaCO3 ratio (where UPZspp and LPZspp stand for species typical of the upper- and lower-photic zones) and (b) the rain ratio, that is, ratio of CaCO3 to particulate organic carbon (POC) fluxes, and (c) the Coccolith-CaCO3 to POC molar ratios, along the transatlantic transect. toward the west. Whether this is linked to differently sized mixing, light and nutrient availability), to which some taxa morphotypes of G. flabellatus and/or to a greater susceptibility are more resilient than others (e.g., Poulton et al. 2007; of its calcification to increasingly dynamic water conditions Zondervan 2007; Sett et al. 2014). But while a decline in toward NW Africa remains as an open question. coccolithophore calcification would be expected to reduce the Assuming that deep-dwelling species are ecologically better rain ratio (e.g., Beaufort et al. 2011; Hutchins 2011), our obser- adapted to thrive in the face of ocean warming (Guerreiro vations indicate just the opposite. Indeed, yearlong-dominant et al. 2019), our 1-yr transatlantic snapshot suggests that lower photic zone species produced 3–8 times higher coccolithophore communities are likely to become less calci- Coccolith-CaCO3/POC ratios at M4 even compared to ratios fied in response to ocean stratification. This is in line with by much larger-sized Helicosphaera spp. and C. leptoporus at M1 Beaufort et al. (2011) who projected a decline in coccolith and CB (see Fig. 7a,b). mass through ecological selection of less-calcified species Lower Coccolith-CaCO3/POC ratios at M1 compared to M2 rather than from reduced calcification within individual spe- reflect the yearlong higher POC fluxes toward east (Fig. 6b,c). cies. In other words, the biogeochemical impact of Although site M1 is marked by low coccolithophore export coccolithophores may be mostly due to variations in abiotic production, high fluxes of POC and biogenic silica in this factors affecting their ecology (i.e., temperature, stratification/ region (Fig. 6f) suggest an increase in productivity linked to a 3200 Guerreiro et al. Coccolithophore species-specific carbonate fluxes dominant surface-dwelling phytoplankton assemblage CaCO3 flux was more consistently linked to POC at all sites (including diatoms) in response to the eastward shallowing of (Fig. 9a,b). At M4, the correlation strongly depends on two the nutricline (Romero and Fischer 2017; Guerreiro outliers representing distinct events of Amazon and possibly et al. 2019). This is supported by the rising bSiO2/CaCO3 dust fertilization of the upper photic zone in the western trop- ratios from M4 to M1 (Fig. 6b), in parallel with surface Chl ical North Atlantic (i.e., Guerreiro et al. 2017; Korte a concentrations and the UPZspp/LPZspp ratio (Fig. 6a). Lower et al. 2017, 2020). The importance of these outliers at M4 is and more seasonally variable bSiO2/CaCO3 ratios just at the also reflected in the stronger correlation between upper photic mesotrophic site CB (Fig. 6c), however, suggest a much higher zone species and POC export toward the western part of the production of calcifying plankton compared to silicifying transect (Fig. 9d), consistent with the pulsed carbonate pro- plankton. While eastern boundary upwelling systems are more duction by E. huxleyi and G. oceanica in spring and autumn. traditionally linked to diatom production (Abrantes Overall, the positive correlation between the UPZspp/LPZspp- et al. 2016), the observed high POC fluxes offshore Cape Blanc CaCO3 ratio and POC fluxes (Fig. 9d—all sites; Supplementary are associated to carbonate-dominated production instead. Table S7) reflects the enhanced coccolith-CaCO3 production Pteropod-dominated carbonate production reported by Fischer in the upper photic zone linked to higher productivity in et al. (2016) may explain the observed high rain ratios, low shallower nutricline conditions, whether due to geostrophic Coccolith-CaCO3/POC ratios, and low coccolith contribution wind forcing (i.e., sites CB and M1) or to the occurrence of to total CaCO3 flux observed at the mesotrophic site CB. transient surface blooms (site M4; see Guerreiro et al. 2017, Assuming that higher rain ratios result in a slower BCP, the 2019; Korte et al. 2020). similar rain ratios observed at both the upwelling-site CB and In spite of the much higher unspecified-CaCO3 fluxes com- the thermally stratified sites M4 and M2 suggest that the BCP pared to coccolith-CaCO3 (Fig. 2a), the ratio between was not more efficient off NW Africa, contrary to expectations coccolith- and unspecified-CaCO3 shows little correlation to from a region close to such a productive eastern boundary the POC export at individual trap sites (Fig. 9c), and only a upwelling system (see Fischer et al. 2009, 2016, 2019). Our weak negative correlation over the entire transect (Fig. 9c—all transatlantic observations further suggest that increased sites, Supplementary Table S7). This suggests a minor coccolithophore productivity in the lower photic zone in coccolith-ballasting effect relative to other calcifying groups, response to ocean warming is unlikely to lower the rain ratio. and that compositional changes in the calcifying plankton On the contrary, if combined with a decreased biological community might, to some degree, alter the efficiency of POC drawdown of atmospheric CO2 due to nutrient depletion, the export along the transect. relative increase of these species may actually contribute to More efficient coccolith-ballasting has been reported from weaken the BCP (see Cermeño et al. 2008). This argues for a the NE Atlantic, where coccolith-CaCO3 fluxes are dominated potentially greater role of export productivity from the lower by C. leptoporus (Ziveri et al. 2007). This species was also the photic zone in marine carbon cycling in the face of ocean main coccolith-carbonate producer in the northeast of our warming. transect (Fig. 5a) and was positively correlated to POC fluxes at site CB (Supplementary Table S6 and Fig. S2). However, Coccolith-ballasting higher carbonate contributions by other taxa toward the The incorporation of mineral ballast—such as biogenic southwest, including both large- and small-sized coccolith spe- carbonate—in marine snow aggregates and fecal pellets adds cies, resulted in weaker correlations with POC, for example, to their density, thereby increasing their sinking velocities. large-sized Helicosphaera spp. compared to smaller-sized This is critical for reducing the exposure of POC to E. huxleyi and G. oceanica (Supplementary Table S6 and remineralization during settling, hence ensuring its successful Fig. S2). The efficiency of coccolith-ballasting seems, hence, to downward export and sequestration in the deep ocean be less related to the coccolith-carbonate mass but more with (Armstrong et al. 2002; Fischer et al. 2016, 2019; van der Jagt the species’ affinity to bloom in high productivity conditions, et al. 2018). Positive correlation between POC and carbonate which, in turn, are linked to higher POC fluxes. Supporting produced by coccoliths (Fig. 9a), as well as by other calcifying this hypothesis is the lack of correlation between POC and the plankton groups (i.e., unspecified-CaCO3 flux, Fig. 9b) along lower photic zone species F. profunda and G. flabellatus, not the transect, does suggest such a link between carbonate bal- only at sites M4 and M2 where they dominate coccolith- lasting and POC sequestration (Supplementary Table S7). This CaCO3 production, but also elsewhere along the transect (see is consistent with previous observations based on regional Supplementary Table S6 and Fig. S2). Lower POC production/ (Fischer et al. 2009; Fischer and Karakas 2009) and global-scale export from the lower photic zone in highly stratified condi- sediment trap records (Ziveri et al. 2000, 2007; Klaas and tions could explain the reduced ballasting efficiency of these Archer 2002), as well as from model simulations (Armstrong deep-dwelling species, in spite of their high carbonate et al. 2002). contribution. Strongest correlations between coccolith-CaCO3 and POC Our data suggest that coccoliths will be more effective bal- export were found at sites M2 and CB, while the unspecified- lasts when produced by fast-blooming surface-dwelling species 3201 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fig 9. Correlation between particulate organic carbon (POC) fluxes and (a) total coccolith-CaCO3 fluxes, (b) unspecified-CaCO3 fluxes, (c) the ratio between coccolith- and unspecified-CaCO3 fluxes, and (d) ratio between the coccolith-CaCO3 fluxes produced by taxa typical of the upper- and lower- photic zone (UPZspp/LPZspp-CaCO3). POC flux data for M1, M2, and M4 from Korte et al. (2017). See Corresponding Pearson correlation matrix in Sup- plementary Table S7. such as E. huxleyi and G. oceanica. Not only these species have not only as a source of mineral ballast but also as a nutrient higher affinity to thrive in more productive conditions, they fertilizer to fuel coccolithophore productivity. Indeed, as are also more likely to be grazed upon by microzooplankton referred above, our data do show a sharp increase in carbonate compared to larger taxa like Helicosphaera spp. and of G. oceanica and E. huxleyi (Fig. 5a) during the Amazon and C. leptoporus (Rost and Riebesell 2004), thus increasing their possibly dust fertilization of the upper photic zone in the chances to sink inside fecal pellets. It seems that the produc- western tropical North Atlantic, reflecting their fast response tion/export of organic matter is what modulates the efficiency to short-term nutrient input (i.e., Guerreiro et al. 2017; Korte of coccolith-ballasting, in line with Passow and De La et al. 2017, 2020). Interestingly, though, such coccolith- Rocha (2006) and Fischer et al. (2016) who argued that the carbonate increase resulted in lower Coccolith-CaCO3/POC POC flux determines the flux of minerals to the deep ocean ratios due to the drastic co-increase of POC during these dusty and not vice-versa. events (Figs. 6a,b). Albeit their slightly heavier coccolith mass, Coccolith-CaCO3/POC ratios by E. huxleyi and G. oceanica Coccolithophore carbonate and Saharan dust were lower even compared to those produced by deep- Being adjacent to the world’s largest source of soil dust, our dwelling species at site M4 (Fig. 7b). Bory et al. (2001) also transatlantic region is a natural laboratory to explore the reported a drastic increase in POC fluxes following an effects of Saharan dust deposition in the marine environment, event of high export of coccoliths by E. huxleyi (> 93%) and 3202 Guerreiro et al. Coccolithophore species-specific carbonate fluxes of lithogenic particles off NW Africa, to which enhanced input effect on either POC production (via fertilization) at site M1, of dust-born nutrients could have contributed. consistent with its negative correlation to UPZspp/LPZspp- Concerning the dust ballasting effect, our data do show a CaCO3 (Fig. 10b), nor in promoting its export (via ballasting) in stronger positive correlation of POC to Saharan dust fluxes this region. (Fig. 10a) than to coccolith-CaCO3 (Fig. 9a), suggesting more For the mesotrophic CB, the residual (unspecified) mass efficient ballasting. This is consistent with the weak negative cor- flux—often used as a proxy for Saharan dust fluxes (Fischer and relation between the Coccolith-CaCO3/dust ratio and POC along Karakas 2009; Fischer et al.2016; van der Does et al. 2020)— the transect (Fig. 10c—M1, M2, and M4; Supplementary Tables S7 suggests much lower dust deposition northwards of the Saha- and S8). Since there were no changes in coccolith species that ran dust plume core where M1 is located ( 12 N) (Fig. 7f; would indicate dust fertilization east of M4, one might expect Supplementary Table S4 and Fig. S3). Previous studies have that increased dust ballasting with increasing proximity to the reported the interaction of very high fluxes of this residual dust sources in Africa would have greatly contributed to the fraction with abundant formation of organic-rich marine snow observed high POC export at M1 (see Supplementary Table S5 particles at site CB to increase carbon sequestration offshore and PC2 in Fig. S1). However, the weaker correlation between NW Africa relative to other eastern boundary upwelling systems POC and dust (Fig. 10a) compared to POC and coccolith-CaCO3 (e.g., Fischer and Karakas 2009; van der Jagt et al. 2018). fluxes at sites M2 and M1 (Fig. 9a) suggests a higher efficiency Although we do not have dust data for site CB to confirm in of coccolith-ballasting. It seems that dust had no clear added which of these eastern locations Saharan dust acted as a more Fig 10. Correlation between (a) fluxes of dust and particulate organic carbon (POC), (b) UPZspp/LPZspp-CaCO3 ratio (where UPZspp and LPZspp stand for species typical of the upper- and lower-photic zones) and dust, and (c) ratio coccolith-CaCO3/dust and POC fluxes. POC and dust data for M1, M2, and M4 from Korte et al. (2017) and van der Does et al. (2020). See Corresponding Pearson correlation matrix in Supplementary Table S8. 3203 Guerreiro et al. Coccolithophore species-specific carbonate fluxes effective ballast, the observed similarly weak correlation • Remotely sensed surface PIC revealed a good potential for between the residual fraction and POC at both M1 and CB do exploring the occurrence of coccolithophore blooms in con- not point toward any significant dust-ballasting effect in any of ditions and/or regions where the nutricline is shallower, these locations (Supplementary Fig. S3). The proposed link that is, toward the eastern boundary upwelling system off between Saharan dust deposition and coccolith-CaCO3 produc- NW Africa and during the occurrence of transient fertiliza- tion in the upper photic zone is, however, reinforced by the tion events in open ocean stratified conditions. It is, how- stronger correlation between the UPZspp/LPZspp-CaCO3 ratio ever, unsuitable for regions and/or conditions where and dust (Fig. 10b) compared to the residual mass flux along coccolith production occurs deeper in the photic zone, M1-M4 (Supplementary Fig. S3b). hence below the maximum satellite-observation depth Our study suggests that an increasing dominance of lower range; photic zone species in response to ocean warming may poten- • Positive correlations between large-sized Helicosphaera spp. tially lead to reducing the drawdown of atmospheric CO2 by and POC fluxes were weaker compared to smaller-sized simultaneously (1) increasing the rain ratio and (2) reducing G. oceanica and E. huxleyi, suggesting that the efficiency of the coccolith-ballasting efficiency, both processes likely to coccolith-ballasting is ecologically dependent. That is, it has weaken the BCP. Climate-driven increasing levels of atmo- less to do with the carbonate mass per species but with its spheric dust across the tropical North Atlantic could, however, affinity for thriving in more productive conditions. It is the contribute to counterbalance this tendency. Either through POC flux that seems to determine the flux of coccoliths to fertilizing the ocean and/or by acting as ballast material, dust the deep ocean and not vice versa; has the potential to reduce the Coccolith-CaCO3/POC molar • Higher Coccolith-CaCO3/POC molar ratios and weaker cor- ratios and to promote a more efficient coccolith-ballasting, relations between the carbonate of lower photic zone spe- both contributing to the accelerate the BCP. cies and POC fluxes at site M4 suggest that increasing abundances of F. profunda and G. flabellatus in response to ocean warming might contribute to weaken the BCP Conclusions through (1) rising the rain ratio and (2) reducing the Our synoptic observations of seasonally resolved coccolith- coccolith-ballasting efficiency. CaCO3 export fluxes along a transatlantic transect of four sedi- • A striking flux increase of coccolith-CaCO3 by fast- ment trap sites contribute to understand the role of calcifying blooming G. oceanica and E. huxleyi, POC, and biogenic sil- phytoplankton in the total carbonate export production. The ica during two dust-related fertilization events at site M4 potential of coccolithophore species for influencing the BCP resulted in lower Coccolith-CaCO3/POC molar ratios and through modulating the rain ratio and ballasting POC along higher bSiO2/CaCO3 ratios. This suggests that increasing the tropical North Atlantic was investigated. The main conclu- Saharan dust outbreaks across the tropical North Atlantic sions are as follows: will contribute to compensate the projected weakening of the BCP, either by providing nutrients to fuel fast-blooming • Fluxes of coccolith-CaCO3 at sites M2, M1, and CB were surface-dwelling coccolithophore production and by pro- similar or higher compared to previous trap studies at moting the export of POC via both dust- and coccolith- higher latitudes in the North Atlantic. At site M4, however, ballasting. lower photic zone species F. profunda and G. flabellatus pro- duced dramatically higher coccolith-CaCO3 fluxes, contrib- Our transatlantic observations suggest that changes in uting to 2–3 times higher coccolith-CaCO3 production in coccolithophore species composition and distribution in response the western tropical North Atlantic (16 mg m2 d1) com- to ocean warming are likely to result in community-level shifts in pared to the other trap sites (5 mg m2 d1 at M2 and M1, total organic and inorganic carbon production/export/ballasting and 8 mg m2 d1 at CB). This highlights the importance by this group. This is expected to impact the cycling and seques- of carbonate produced by coccolithophores in stratified- tration of carbon at both regional and global scales. We highlight oligotrophic conditions, typical of tropical oceans; the need of addressing the role of coccolithophore ecological • Coccolith percent contributions of 45% to total carbonate dynamics and species-specific carbonate production in modulat- in the western tropical North Atlantic were high compared ing the rain ratio for more accurately project the functioning of to only 23% at M2, and 15% at M1 and CB, suggesting that the BCP in the face of ongoing climate change. other calcifying plankton were more important carbonate contributors eastwards of site M4. This points to the need for more accurately estimating the flux of distinct carbonate-producing pelagic organisms, as well as of quanti- References fying the coccoliths within aggregates/fecal pellets and Abrantes, F., P. Cermeno, C. Lopes, O. Romero, L. Matos, J. investigating the nature of particles within the “biogenic Van Iperen, M. Rufino, and V. Magalh~aes. 2016. Diatoms Si carbonate powder”; uptake capacity drives carbon export in coastal upwelling 3204 Guerreiro et al. Coccolithophore species-specific carbonate fluxes systems. Biogeosciences 13: 4099–4109. doi:10.5194/bg-13- mass of thin carbonate particles such as coccoliths. Nat. 4099-2016 Protoc. 9: 633–642. Archer, D., and E. Maier-Reimer. 1994. Effect of deep-sea sedi- Bollmann, J. 2014. Technical note: Weight approximation of mentary calcite preservation on atmospheric CO2 concen- coccoliths using a circular polarizer and interference colour tration. Nature 367: 260–263. derived retardation estimates – (the CPR method). Bio- Armstrong, R. A., C. Lee, J. I. Hedges, S. Honjo, and S. G. geosciences 11: 1899–1910. Wakeham. 2002. A new, mechanistic model of organic car- Bory, A., and others. 2001. Particle flux within different pro- bon fluxes in the ocean based on the quantitative associa- ductivity regimes off the Mauritanian upwelling zone (EU- tion of POC with ballast minerals. Deep-Sea Res. Pt. II 49: MELI program). Deep-Sea Res. Pt I 48: 2251–2282. 219–236. Broecker, W., and E. Clark. 2009. Ratio of coccolith CaCO3 to Balch, W. M., P. M. Holligan, S. G. Ackleson, and K. J. Voss. foraminifera CaCO3 in late Holocene deep-sea sediments. 1991. Biological and optical properties of mesoscale Paleoceanography 24: PA3205. doi:10.1029/2009PA001731 coccolithophore blooms in the Gulf of Maine. Limnol. Broerse, A. T. C., P. Ziveri, J. E. van Hinte, and S. Honjo. 2000. Oceanogr. 36: 629–643. Coccolithophore export pro- duction, species composition Balch, W. M., H. R. Gordon, B. C. Bowler, D. T. Drapeau, and and coccolith- CaCO3 fluxes in the NE Atlantic (34 N E. S. Booth. 2005. Calcium carbonate budgets in the surface 21 W and 48 N 21 W). Deep-Sea Res. Pt. II 47: 1877–1906. global ocean based on MODIS data. J. Geophys Res Oceans Brummer, G.-J., C. Hemleben, and M. Spindler. 1986. Plank- 110 (C7): C07001 07010.01029/02004JC002560. tonic foraminiferal ontogeny and new perspectives for Balch, W. M., B. C. Bowler, D. T. Drapeau, L. C. Lubelczyk, micropaleontology. Nature 319: 50–52. and E. Lyczkowski. 2018. Vertical distributions of Cermeño, P., S. Dutkiewicz, R. P. Harris, M. Follows, O. coccolithophores, PIC, POC, biogenic silica, and chloro- Schofield, and P. G. Falkowski. 2008. The role of nutricline phyll a throughout the global ocean. Global Biogeochem. depth in regulating the ocean carbon cycle. Proc. Natl. Cycles 32: 2–17 https://doi.org/10.1002/2016GB005614. Acad. Sci. USA 105: 20344–20349. Basha, G., P. Kishore, M. Venkat Ratnamc, T. B. M. J. Ouarda, Chavez, F. P., M. Messié, and J. T. Pennington. 2011. Marine I. Velicogna, and T. Sutterley. 2015. Vertical and latitudinal primary production in relation to climate variability and variation of the intertropical convergence zone derived change. Ann. Rev. Mar. Sci. 3: 227–260. using GPS radio occultation measurements. Remote Sens. Fischer, G., and G. Karakas. 2009. Sinking rates and ballast Environ. 163: 262–269. composition of particles in the Atlantic Ocean: Implica- Baumann, K.-H., and T. Freitag. 2004. Pleistocene fluctuations tions for the organic carbon fluxes to the deep ocean. Bio- in the Benguela current system as revealed by coccolith geosciences 6: 85–102. doi:https://doi.org/10.5194/bg-6-85- assemblages. Mar. Micropaleontol 52: 195–215. 2009 Baumann, K.-H., B. Bockel, and M. Frenz. 2004. Coccolith Fischer, G., C. Reuter, G. Karakas, N. Nowald, and G. Wefer. contribution to South Atlantic carbonate sedimentation, 2009. Offshore advection of particles within the Cape Blanc p. 367–402. In H. R. Thierstein and J. R. Young [eds.], filament, Mauritania: Results from observational and Coccolithophores – From molecular processes to global modelling studies. Prog. Oceanogr. 83: 322–330. impact. Springer-Verlag. Fischer, G., Basse, A., Baumann, K.-H., Klann, M., Klawonn, I., Baumann, K.-H., H. Andruleit, B. Bockel, M. Geisen, and H. Küchler, R., Nowald, N., and Ruhland, G. 2012. Report and Kinkel. 2005. The significance of extant coccolithophores preliminary results of RV POSEIDON Cruise P425. Las Pal- as indicators of ocean water masses, surface water tempera- mas – Las Palmas, 16.01.2012–30.01.2012. Berichte, ture, and paleoproductivity: A review. Paläontol. Z. 79: Fachbereich Geowissenschaften, Universität Bremen, 93–112. No. 287: 32 pp. Beaufort, L., and S. Heussner. 1999. Coccolithophorids on the Fischer, G., and others. 2013. Report and preliminary results continental slope of the Bay of Biscay – Production, trans- of R/V POSEIDON cruise POS445. Las Palmas – Las Palmas, port and contribution to mass fluxes. Deep-Sea Res. Pt. II 19.01.2013–01.02.2013. Berichte, MARUM – Zentrum für 46: 2147–2174. Marine Umweltwissenschaften, Fachbereich Beaufort, L. 2005. Weight estimates of coccoliths using the Geowissenschaften, Universität Bremen, No. 298: 30 pp., optical properties (birefringence) of calcite. Micropaleontol- ISSN 2195-9633. ogy 51: 1–9. Fischer, G., and others. 2014. Report and preliminary results Beaufort, L., and others. 2011. Sensitivity of coccolithophores of R/V POSEIDON cruise POS464, Las Palmas (Canary to carbonate chemistry and ocean acidification. Nature Islands) – Las Palmas (Canary Islands), 03.02. 2014– 476: 80–83. doi:10.1038/nature10295 18.02.2014. Berichte, MARUM – Zentrum für Marine Beaufort, L., N. Barbarin, and Y. Gally. 2014. Optical measure- Umweltwissenschaften, Fachbereich Geowissenschaften, ments to determine the thickness of calcite crystals and the Universität Bremen, No. 304: 29 pp., ISSN 2195-7894. 3205 Guerreiro et al. Coccolithophore species-specific carbonate fluxes Fischer, G., and others. 2016. Deep ocean mass fluxes in the Katz, E. J. 1981. Dynamic topography of the sea surface in the coastal upwelling off Mauritania from 1988 to 2012: Vari- equatorial Atlantic. J. Mar. Res. 39: 53–63. ability on seasonal to decadal time- scales. Biogeosciences Klaas, C., and D. Archer. 2002. Association of sinking organic 13: 3071–3090. doi:https://doi.org/10.5194/bg-13-3071- matter with various types of mineral ballast in the deep sea: 2016 Implications for the rain ratio. Global Biogeochem. Cycles Fischer, G., and others. 2019. Changes in the dust- 16: 1116. doi:10.1029/2001GB001765 influenced biological carbon pump in the canary current Knappertsbusch, M., and G.-J. A. Brummer. 1995. A sediment system: Implications from a coastal and an offshore sedi- trap investigation of sinking coccolithophores in the North ment trap record off Cape Blanc, Mauritania. Global Bio- Atlantic. Deep-Sea Res. Pt. I 42: 1083–1109. geochem. Cycles 33: 29 https://doi.org/10.1029/ Köbrich, M. I., and K.-H. Baumann. 2009. Coccolithophore 2019GB006194. flux in a sediment trap off Cape Blanc (NW Africa).  J.-A. Flores, and F. J. Sierro. 2014. The use of cir- Fuertes, M.-A., J. Nannoplankton Res. 30: 83–96. cularly polarized light for biometry, identification and esti- Korte, L. F., and others. 2017. Downward particle fluxes of bio- mation of mass of coccoliths. Mar. Micropaleontol. 113: genic matter and Saharan dust across the equatorial North 44–55. doi:https://doi.org/10.1016/j.marmicro.2014.08.007 Atlantic. Atmos. Chem. Phys. 17: 6023–6040 doi:10.5194/ Gordon, H., and W. McCluney. 1975. Estimation of the depth acp-17-6023-%202017. of sunlight penetration in the sea for remote sensing. Appl. Korte, L. F., G.-J. Brummer, M. van der Does, C. V. Guerreiro, Optics 14: 413–416. F. Mienis, C. I. Munday, L. Ponsoni, and S. Schouten. 2020. Gordon, H. R., G. C. Boynton, W. M. Balch, S. B. Groom, D. S. Multiple drivers of production and particle export in the Harbour, and T. J. Smyth. 2001. Retrieval of western tropical North Atlantic. Limnol. Oceanogr. 9999: coccolithophore calcite concentration from SeaWiFS imag- 1–17. doi:10.1002/lno.11442 ery. Geophys. Res. Lett. 28: 1587–1590. https://doi.org/10. Malviya, S., and others. 2016. Insights into global diatom dis- 1029/2000gl012025 tribution and diversity in the world’s ocean. Proc. Natl. Guerreiro, C. V., K.-H. Baumann, G.-J. A. Brummer, G. Fischer, Acad. Sci. USA 113: E1516–E1525, doi/www.pnas.org/cgi/ L. F. Korte, U. S. C. Merkel, H. de Stigter, and J.-B. W. Stuut. 10.1073/pnas.1509523113. 2017. Coccolithophore fluxes in the open tropical North Merle, J. 1980. Seasonal variation of heat storage in the tropi- Atlantic: Influence of thermocline depth, Amazon water, cal Atlantic Ocean. Oceanol. Acta 3: 455463. and Saharan dust. Biogeosciences 14: 4577–4599. doi: Milliman, J. D. 1993. Production and accumulation of calcium https://doi.org/10.5194/bg-14-4577-2017 carbonate in the ocean: Budget of a non-steady state. Guerreiro, C. V., K.-H. Baumann, G.-J. A. Brummer, G. Fischer, Global Biogeochem. Cycles 7: 927–957. L. F. Korte, C. Sa, and J.-B. W. Stuut. 2019. Wind-forced Milliman, J. D., P. J. Troy, W. M. Balch, A. K. Adams, Y. H. Li, transatlantic gradients in coccolithophore species fluxes. and F. T. Mackenzie. 1999. Biologically mediated dissolu- Prog. Oceanogr. 176: 102140. doi:https://doi.org/10.1016/ tion of calcium carbonate above the chemical lysocline? j.pocean.2019.102140 Deep-Sea Res. Pt I Oceanogr. Res 46: 1653–1669. Holligan, P. M., and others. 1993. A biogeochemical study of Molleri, G. S. F., E. M. L. Novo, and M. Kampel. 2010. Space– the coccolithophore, Emiliania huxleyi, in the north Atlan- time variability of the Amazon River plume based on satel- tic. Global Biogeochem. Cycles 7: 879–900. lite ocean color. Cont. Shelf Res. 30: 342–352. Honjo, S. 1978. Sedimentation of materials in the Sargasso Sea Passow, U., and C. De La Rocha. 2006. Accumulation of min- at a 5367 m deep station. J. Mar. Res. 36: 469–492. eral ballast on organic aggregates. Global Biogeochem. Hopkins, J., S. A. Henson, S. C. Painter, T. Tyrrell, and A. J. Cycles 20: GB1013. doi:10.1029/2005GB002579 Poulton. 2015. Phenological characteristics of global Pastor, M. V., J. B. Palter, J. L. Pelegrí, and J. P. Dunne. 2013. coccolithophore blooms. Global Biogeochem. Cycles 29: Physical drivers of interannual chlorophyll variability in 239–253. the eastern subtropical North Atlantic. J. Geophys. Res. Hutchins, D. 2011. Forecasting the rain ratio. Nature 476: 41–42. Oceans 118: 3871–3886. Iglesias-Rodríguez, M. D., C. W. Brown, S. C. Doney, J. Pienaar, R. N. 1994. Ultrastructure and calcification of Kleypas, D. Kolber, Z. Kolber, P. K. Hayes, and P. G. coccolithophores, p. 63–82. In A. Winter and W. G. Siesser Falkowski. 2002. Representing key phytoplankton func- [eds.], Coccolithophores. Cambridge University Press. tional groups in ocean carbon cycle models: Poulton, A. J., R. Sanders, P. M. Holligan, T. Adey, M. Coccolithophorids. Global Biogeochem. Cycles 16: 1–20. Stinchcombe, L. Brown, and K. Chamberlain. 2006. Phyto- doi:10.1029/2001GB001454 plankton mineralisation in the tropical and subtropical Jickells, T. D., S. Dorling, W. G. Deuser, T. M. Church, Atlantic Ocean. Glob. Biogeochem. Cycles 20: GB4002 R. Arimoto, and M. Prospero. 1998. Air-borne dust fluxes to 4010.1029/2006GB002712. a deep water sediment trap in the Sargasso Sea. Global Bio- Poulton, A. J., T. R. Adey, W. M. Balch, and P. M. Holligan. geochemical Cycles 12: 311–320. 2007. Relating coccolithophore calcification rates to 3206 Guerreiro et al. Coccolithophore species-specific carbonate fluxes phytoplankton community dynamics: Regional differ- Signorini, S. R., B. A. Franz, and C. R. McClain. 2015. Chloro- ences and implications for carbon export. Deep-Sea Res. phyll variability in the oligo- trophic gyres: Mechanisms, Pt. II Chapman Calcification Conference - Special 54: seasonality and trends. Front. Mar. Sci. 2: 11. 538–557. Sprengel, C., K.-H. Baumann, J. Henderiks, R. Henrich, and S. Poulton, A. J., P. M. Holligan, A. Charalampopoulou, and T. R. Neuer. 2002. Modern coccolithophore and carbonate sedi- Adey. 2017. Coccolithophore ecology in the tropical and mentation along a productivity gradient in the Canary subtropical Atlantic Ocean: New perspectives from the Islands region: Seasonal export production and surface Atlantic Meridional Transect (AMT) programme. Prog. accumulation rates. Deep-Sea Res. Pt. II 49: 3577–3598. Oceanogr 158: 150–170. https://doi.org/10.1016/j.pocean. Steinmetz, J. 1991, p. 41. In S. Honjo [ed.], Calcareous nanno- 2017.01.003. plankton biocoenosis: Sediment trap studies in the Equato- Rigual Hernandez, A. S., and others. 2020. Coccolithophore rial Atlantic, Central Pacific, and Panama Basin. Woods biodiversity controls carbonate export in the Southern Hole Oceanographic Institution Woods Hole Ocean. Biogeosciences 17: 245–263. Massachusetts. Stuut, J.-B. W., and others. 2013. Cruise report and prelimi- Rivero-Calle, S., A. Gnanadesikan, C. E. Del Castillo, W. M. nary results (64PE378), TRAFFIC II: Transatlantic fluxes of Balch, and S. D. Guikema. 2015. Multidecadal increase in Saharan dust (Las Palmas de Gran Canaria, Spain – North Atlantic coccolithophores and the potential role of St. Maarten). Royal NIOZ, p. 54. rising CO2. Science 350: 1533–1537. van der Does, M., G.-J. A. Brummer, F. C. J. van Crimpen, L. F. Romero, O., and G. Fischer. 2017. Shift in the species Korte, N. M. Mahowald, U. Merkel, and others. 2020. Tropi- composition of the diatom community in the eutrophic cal rains controlling deposition of Saharan dust across the Mauritanian coastal upwelling: Results from a multi-year North Atlantic Ocean. Geophys. Res. Lett. 47: sediment trap experiment from 2003 to 2010. Prog. Ocean- e2019GL086867. doi:https://doi.org/10.1029/ ogr. 159: 31–44. doi:https://doi.org/10.1016/j.pocean.2017. 2019GL086867 09.010 van der Jagt, H., C. Friese, J.-B. W. Stuut, G. Fischer, and M. H. Romero, O., K.-H. Baumann, K. Zonneveld, B. Donner, J. Iversen. 2018. The ballasting effect of Saharan dust deposi- Hefter, B. Hamady, V. Pospelova, and G. Fischer. 2020. Flux tion on aggregate dynamics and carbon export: Aggrega- variability of phyto- and zooplankton communities in the tion, settling, and scavenging potential of marine snow. Mauritanian coastal upwelling between 2003 and 2008. Limnol. Oceanogr. 63: 1386–1394. doi:https://doi.org/10. Biogeosciences 17: 187–214. 1002/lno.10779 Rost, B., and U. Riebesell. 2004. Coccolithophores and the bio- Waniek, J., W. Koeve, and R. D. Prien. 2000. Trajectories of logical pump: Responses to environmental changes, p. 99– sinking particles and the catchment areas above sediment 125. In H. R. Thierstein and J. R. Young [eds.], traps in the Northeast Atlantic. J. Mar. Res. 58: 983–1006. Coccolithophores: From molecular processes to global Young, J. R., and P. Ziveri. 2000. Calculation of coccolith vol- impact. Springer. ume and its use in calibration of carbonate flux estimates. Salter, I., Schiebel, R., Ziveri, P., Movellan, A., Lampitt, R. S., Deep-Sea Res. Pt. II 47: 1679–1700. and Wolff, G. A. 2015. Carbonate counter pump stimulated Ziveri, P., A. Rutten, G. J. de Lange, J. Thomson, and C. by natural iron fertilization in the Southern Ocean, 2015 Corselli. 2000. Present-day coccolith fluxes recorded in cen- Aquatic Sciences Meeting, Granada, Spain, 22–27 February tral eastern Mediterranean sediment traps and surface sedi- 2015. ments. Palaeogeogr. Palaeoclimatol. Palaeoecol. 158: Sarmiento, J. L., J. Dunne, A. Gnanadesikan, R. M. Key, K. 175–195. Matsumoto, and R. Slater. 2002. A new estimate of the Ziveri, P., B. de Bernardi, K.-H. Baumann, H. M. Stoll, and CaCO3 to organic carbon export ratio. Glob. Biogeochem. P. G. Mortyn. 2007. Sinking of coccolith carbonate Cycles 16: 1107. and potential contribution to organic carbon ballasting in Sett, S., L. T. Bach, K. G. Schulz, S. Koch-Klavsen, M. Lebrato, the deep ocean. Deep-Sea Res. Pt. II 54: 659–675. and U. Riebesell. 2014. Temperature modulates Zondervan, I. 2007. The effects of light, macronutrients, trace coccolithophorid sensitivity of growth, photosynthesis and metals and CO2 on the production of calcium carbonate calcification to increasing seawater pCO2. PLoS ONE 9: and organic carbon in coccolithophores—A review. Deep- e88308 doi:10.1371/journal.pone.0088308. Sea Res. Pt. II 54: 521–537. Shutler, J., P. Land, C. Brown, H. Findlay, C. Donlon, M. Medland, R. Snooke, and J. Blackford. 2013. Acknowledgments Coccolithophore surface distributions in the North Atlantic The authors thank the crews of Meteor cruise M89, Pelagia cruise 64PE378, RV Poseidon cruises POS425, POS445 and POS464, as well as and their modulation of the air-sea flux of CO2 from the NIOZ and MARUM technicians for their contributions. Moorings M4, 10 years of satellite Earth observation data. Biogeosciences M2, and M1 were managed by the NIOZ in the framework of the projects 10: 2699–2709. TRAFFIC funded by NWO (no. 822.01.008), and DUSTTRAFFIC funded by 3207 Guerreiro et al. Coccolithophore species-specific carbonate fluxes ERC (no. 311152), directed by Jan-Berend W. Stuut. Moorings CB-23 and Atlantic Area Programme (contract EAPA-165/2016), CALMED CB-24 were managed by Gerhard Fischer as part of the MARUM project (#CTM2016-79547-R) and the Generalitat de Catalunya MERS (#2017 GB1 (Particle flux, carbon turnover and nutrient regeneration). Lab prepa- SGR-1588). The authors acknowledge the NASA’s Ocean Biology ration of the 1/5 split of the original sediment trap sample for M4, M2 Processing Group (https://oceancolor.gsfc.nasa.gov), as well as the con- and M1 was conducted at the NIOZ; the splitting, filtering, and SEM tax- structive suggestions and encouraging remarks by two anonymous onomical analysis were performed at Uni-Bremen, Germany. The quantifi- reviewers and by the Editor-in-Chief Dr. K. David Hambright. cation and discussion of the coccolith-CaCO3 fluxes were performed at MARUM, Germany, and at MARE/Uni-Lisbon, Portugal. The first author benefited from a Marie Sklodowska-Curie Fellowship supported by Uni- Conflict of Interest Bremen and the EU FP7 COFUND (grant no. 600411) and from a Marie None declared. Sklodowska-Curie European Fellowship supported by the EU H2020-MSCA-IF-2017 (grant no. 796802) within DUSTCO. Currently, the first author benefits from a research contract funded by FCT (contract Submitted 05 August 2020 CEECIND/00752/2018/CP1534/CT0011) linked to project CHASE (www. Revised 25 January 2021 chase-dust.com). This study was also supported by PORTWIMS (https:// Accepted 29 May 2021 www.portwims.org) funded by the EU’s H2020 Research and Innovation Programme (grant no. 810139), iFADO from ERDF funds of the INTERREG Associate editor: Thomas R. Anderson 3208