. Characterization of the coccolithophore community off Cabo Verde archipelago, including the Senghor Seamount (Eastern North Atlantic)

A systematic investigation of the extant coccolithophore community around Cabo Verde archipelago was performed during the cruise MSM49 of RV Maria S. Merian, which took place in the late fall of 2015. The description of the spatial and vertical distributions of coccolithophores was based on a survey performed to the north, east and south of Cabo Verde archipelago, between the surface and 150 m water depth. The total cell densities obtained for the studied region were relatively low, reaching to a maximum of 30 × 10 3 cell L (cid:0) 1 in the upper 50 m over the southeastern slope of the Senghor seamount. Emiliania huxleyi and Gephyrocapsa oceanica were the dominant species, followed by Florisphaera profunda . The coccolithophore distribution off Cabo Verde was essentially explained by relatively warm and nutrient-depleted waters in the region during the surveyed interval, in result of the weaker NE trade winds and the northward migration of the Intertropical Convergence Zone. In these conditions, a notable zonation of coccolithophores along depth was depicted, in consequence of the inferred general well-stratified water column. Four typical depth-related groups were identified: (i) a Shallow oligotrophic (10 – 30 m), represented by Discosphaera tubifera and Umbellosphaera spp.; (ii) an Intermediate (40 – 50 m), formed by the three placolith-bearing species E. huxleyi , G. ericsonii and G. oceanica , and by Algir-osphaera robusta , Helicosphaera spp., Michaelsarsia spp., Syracosphaera spp. and Umbilicosphaera spp.; (iii) a Deep (60 – 75 m) with F. profunda , Ophiaster spp., Oolithotus spp. and Reticulofenestra sessilis as typical members; (iv) and The Deepest ( > 80 m), composed by Gladiolithus flabellatus and Syracosphaera lamina . In addition, high abundances of G. oceanica related with the Eddy station were attributed to the transport and thriving of the coastal coccolithophore community, dominated by this species, from the African coast towards Cabo Verde.


Introduction
The eastern subtropical North Atlantic is an oligotrophic to mesotrophic region, bordered to the west and north by the oligotrophic subtropical gyre, to the east by the NW African coast, with frequent formation of eddies and filaments transporting upwelled nutrient-rich water from the coast offshore, and to the south by the equatorial convergence also often transporting upwelled waters towards Cabo Verde. In addition, an increase in nutrients in the proximity of an island, known as Island Mass Effect (IME), may rise the nearshore standing stock of phytoplankton biomass by more than 80% over background oceanic productivity (Gove et al., 2016). Acting as natural barriers in the main current pathways, seamounts lead to similar geophysical and biological effects when compared to the surrounding open ocean (Heywood et al., 1990;Hasegawa et al., 2004). The movement of water masses around these prominent features of seabed, usually associated with the upwelling of nutrient-rich deep sea water, lead to a higher primary production and the subsequent increase of fauna and flora around the peaks (Abecasis et al., 2009), thus considered hotspots of biodiversity (Morato et al., 2010). However, at Senghor seamount in particular, studies focused on zooplankton distribution patterns showed no significant differences between seamount and open sites (Denda and Christiansen, 2014). Similarly, the seamount effect on the trophic structure of zooplankton and micronekton communities was weak (Denda et al., 2017a). Since short residence times are known for phytoplankton stocks over seamounts (Mendonça et al., 2012), a detailed study focusing on spatial and temporal composition and distribution of phytoplankton throughout this topographic feature is still vital to clearly understand its dynamics.
With respect to regional investigations, they exist on integral ecological aspects related to the Cape Verde Frontal Zone (Zenk et al., 1991), upwelling filaments formed in the NW African coast (Meunier et al., 2012) and impact of Saharan dust reaching the area (Baker et al., 2007;Marañón et al., 2010). Other studies have already focused on taxonomic composition and distribution of zooplankton (Denda and Christiansen, 2014;Denda et al., 2017) but, concerning the phytoplankton community and taxonomy, only few works based on sediment traps have evaluated the coccolithophore dynamics in this regional context (Köbrich and Baumann, 2009;Köbrich et al., 2015;Guerreiro et al., 2019;Romero et al., 2019).
Coccolithophores, which represent about 10% of the global phytoplankton biomass (Tyrrell and Young, 2009), are pelagic unicellular algae, members of the haptophyte class Prymnesiophyceae Hibberd, distinguished by the ability to produce calcite platelets. These structures, referred as coccoliths, surround the living cell and form an exoskeleton called coccosphere (e.g. Winter and Siesser, 1994;Young, 1994). These phytoplankton organisms, with its first appearance in Upper Triassic sediments (Bown et al., 2004), are widely and abundantly distributed in the ocean, being one of the major contributors of pelagic carbonate on Earth (Honjo, 1996;Schiebel, 2002;Beare et al., 2013). Through biomineralization and photoautotrophy, coccolithophores influence the global carbon and sulphur cycles, actively participating in the climate system (Westbroek et al., 1993;Rost and Riebesell, 2004;Taylor et al., 2007). This fact has aroused much concern about the response of coccolithophores to climate change and subsequent ocean acidification (e.g. Oviedo et al., 2015;D'Amario et al., 2017;Smith et al., 2017).
The present study, focusing on the spatial and temporal composition and distribution of coccolithophores off Cabo Verde, aims to: i) determine the main coccolithophore contributors to the marine phytoplankton community and their overall response to the regional hydrography and physico-chemical parameters; ii) identify the preferential development along the water column relative to nutricline, that compose the upper and lower photic zone assemblages; iii) detail specific species preferences in response to the variety of ecological niches sampled; iv) understand the role of the Senghor seamount in the coccolithophore distribution.

Oceanographic setting
The study area is located around Cabo Verde islands, between 12 • -18 • N and 20 • -24 • W, including the Senghor seamount, locally also known as Nova Holanda seamount. It consists in a nearly conical seamount, which rises from 3300 m to a small summit plateau with a minimum depth of approximately 100 m. This area is influenced hydrographically by large-scale interactions between the Canary Current (CC), the North Equatorial Current (NEC) and the North Equatorial Countercurrent (NECC), forming the Cape Verde Frontal Zone (CVFZ) that is associated with potentially productive zones (Fernandes et al., 2005) (Fig. 1). This feature extends, with some seasonal variability, from Cape Blanc to the SW close to the north-western islands of the archipelago (Pérez-Rodríguez et al., 2001;Vangriesheim et al., 2003). The NEC is originated from the CC when its flux turns southwestwards leaving the African continent. This current, formed in the north of the Cabo Verde islands, is a westerly flow directly responding to the NE trade winds (Stramma et al., 2005). Between 5 • N and 10 • N the dominant surface feature is the NECC. This weaker and easterly current curves northwards close to the African coast to form the Mauritanian Current (MC) (Mittelstaedt, 1991).
The boundary between the NE and SE trade winds corresponds to the Intertropical Convergence Zone (ITCZ), located around 6 • N in February and 15 • N in August, in response to insolation received by the African continent (Molinari et al., 1986). During boreal summer and autumn, when the northernmost position of the ITCZ is observed, the stronger SE trade winds produce an increase in the NECC velocity and a weakening of the NEC and the CC (Mittelstaedt, 1991;Peterson and Stramma, 1991;Fig. 1. Sampling location on four major sites: north of Cabo Verde archipelago, Senghor Seamount, cyclonic eddy to the east of Cabo Verde islands and to the south. The main ocean surface currents that affect the study area are represented: CC-Canary Current; MC-Mauritanian Current; NEC-North Equatorial Current; NECC-North Equatorial Countercurrent; NEUC-North Equatorial Undercurrent. The northernmost and southernmost Intertropical Convergence Zone (ITCZ) locations, the Guinea Dome (GD) and the Cape Verde Frontal Zone (CVFZ) are also included. Stramma et al., 2005). When the strengthening of the NECC occurs, the warm oligotrophic equatorial waters of the MC reach further north at about 20 • N, suppressing the coastal upwelling south of this latitude (Mittelstaedt, 1991).
South of the Cabo Verde archipelago and associated with a cyclonic circulation is the Guinea Dome (GD) (Mazeika, 1967). This quasi stationary feature is part of the large-scale near-surface flow fields associated with the NEC, NECC, as well as with the North Equatorial Undercurrent (NEUC) (Siedler et al., 1992). The GD develops seasonally due to the wind-induced Ekman upwelling associated with the northwards migration of the ITCZ (Doi et al., 2009), similar to the seasonal changes in the NECC (Fernandes et al., 2005).

Sample collection
Sampling was performed during the oceanographic campaign MSM49 onboard of the RV Maria S. Merian, between 28 of November and December 21st, 2015, in the scope of the project SEAMOX-the Influence of Seamounts and Oxygen Minimum on Pelagic Fauna in the Eastern Tropical Atlantic (Christiansen et al., 2016).
Seawater samples were collected between 6 and 9 depths per station, from the first 150 m, through a CTD/Rosette system equipped with 24 Niskin bottles of 10 L volume. The water column was sampled for physical, chemical and biological data using a Seabird CTD (SBE911 plus) incorporated with conductivity, temperature and dissolved oxygen sensors (all SEA-BIRD ELECTRONICS), as well as a combined fluorescence/turbidity sensor (WET LABS, ECO-FL-NTU).

Physico-chemical parameters
In situ temperature and conductivity were measured with a Seabird CTD (SBE911 plus), with accuracies of ±0.001 • C for temperature and ±0.0003 S m − 1 for conductivity. Practical salinity and density were converted from temperature, conductivity and depth measurements.
Fluorescence was measured with a WET LABS, ECO FLNTU sensor with a sensitivity of 0.025 mg m − 3 .
Dissolved oxygen (DO) sensor data were calibrated using water samples drawn at 12 stations and analyzed by Winkler titration for dissolved oxygen.
All raw CTD data were processed following the GO-SHIP guidelines for the SBE 911 plus (e.g Hood et al., 2010). Visualization of the physical data was done using Ocean Data View software (Schlitzer, 2020). The hydrographic data are available at the PANGAEA database .
Samples for dissolved inorganic nutrients (PO 4 , NO 2 , NO 3 and SiO 4 (μmol L − 1 )) were taken together with the phytoplankton samples. Water samples (volume depending on plankton abundance) were filtered through pre-washed (10% HCl) cellulose acetate filters (Sartorius, 0.2 μm pore size) and frozen immediately until analysis. Samples were measured following the protocols of Hansen and Koroleff (1999) with an auto-analyzer (Skalar, SANPLUS; Breda/Netherlands). The detection limit of the auto-analyzer was at a concentration of 0.1 μmol L − 1 .

Coccolithophores
A total of 79 samples were collected for coccolithophore analysis. Onboard, between 4 and 5 L of water were immediately filtered through Whatman nuclepore track-etched polycarbonate membranes (0.4 μm pore size, 47 mm diameter), using low vacuum suction (200-400 mbar), for identification and quantification of this phytoplankton group. The filters were rinsed with about 1-2 ml of deionized water, to remove sea salt, and left to dry in petri dishes at room temperature. On land, at the lab, a randomly chosen 20 to 30 • angular sector of each filter was cut and permanently mounted on a slide using Entellan mounting media and a cover slip. Species diversity and abundances were determined under cross-polarized light microscopy (Leitz Ortholux II Pol-BK) at 1250x magnification. At least 400 coccospheres were counted through a randomly selected sequence of fields of view (FOV) aligned parallel to the radius of the filter. For samples with the lowest abundances (depths below 100 m), in which the density of coccospheres was much lower, the total screened area per filter reached around 10 mm 2 (equivalent to 430 FOV). The absolute abundances (Coccosphere L − 1 ) were estimated as follows: Coccosphere L − 1 = [N x (TA/EA)]/V. Where N refers to the number of counted coccospheres in the examined area, TA is the total filter area, EA the examined area and V the volume of filtered water.

Multivariate analysis
To evaluate the ecological similarities among coccolithophore taxa and their relationship with the environmental conditions, a Canonical Correspondence Analysis (CCA) was conducted using the software package PAleontological STatistics (PAST) vers. 4.0 (Hammer et al., 2001). In order to minimize variance dispersion a square root transform was applied to the data matrix composed of 27 variables (columns) and 79 samples (rows), prior to the multivariate analysis. Ten environmental parameters were determined: water density, depth, fluorescence, dissolved oxygen, salinity, temperature, nitrate, nitrite, phosphate and dissolved silicate. From the microbiological water column analysis, 16 of the more relevant coccolithophore taxa (see criteria in section 4.2.1) were selected as dependent variables in addition to the total coccolithophore absolute abundance for the CCA analysis.

Physico-chemical parameters
The maximum temperature, registered around the upper 30 m, varied between 24.5 • C at the SE slope of the Senghor seamount (Sg3) and 27.4 • C at the southernmost station S1. The minimum temperature of the studied surface layer was observed at 150 m depth, varying between 12.9 • C at the southernmost station S2 and 14.9 • C at station N2. Stations S1, S2 and SE showed a homogeneous surface layer in the upper 25-30 m, whereas in the other stations the homogeneous surface layer was observed down to 35-50 m. Stations Sg3 and N2 recorded the deepest thermocline (Fig. 2, Table 2).
In the upper 30 m of the homogeneous surface layer the salinity varied between 35.4 (S1) and 36.3 (CVO). Maximum values, between 35.7 (S1) and 36.4 (CVO), were recorded at the thermocline for the several sites. Below this depth a general salinity decrease was observed until the lowest depth recorded.
The vertical profile of density is similar to that of the temperature, with the pycnocline around 25-45 m at S1, S2 and SE and around 45-65 m for the other stations. The deepest and the greatest density gradient were recorded at the SE slope of the Senghor seamount (Sg3).

Fig. 2.
Physico-chemical parameters distribution (Temperature, salinity, density, fluorescence and dissolved oxygen) along the uppermost 150 m water depth for the 10 stations.

Table 2
Absolute (Cell L − 1 ) and relative (%) abundances of the coccolihophore taxa observed during the sampling period. Minimum and maximum values of the total coccolithophores and the physico-chemical parameters are also indicated. Fluorescence, as an indicator for phytoplankton biomass, showed the highest values at about 50 m for most of the stations, with exception of SE, at which the maximum occurred at 30 m. Moreover, the station Ed recorded the highest maximum values of fluorescence whereas the station in the NW slope of the Senghor seamount (Sg1) recorded the lowest.
Concerning the dissolved oxygen, a well-mixed saturated surface layer followed by a steep decline coincident with the fluorescence maximum was inferred. At stations Sg3 and N2 this decline started deeper, below the 50 m depth. From there and down to 150 m, steady low oxygen levels were recorded for all stations.
Nitrate, phosphate and dissolved silicate show a general similar pattern along the upper 150 m of the water column, for the 7 surveyed locations (Fig. 3). Thus, nutrient levels were depleted at surface with significant increases from 40-55 m -80 m depth and slight increases    5. Vertical distribution of the most relevant coccolithophore taxa in the first 150 m of the water column, along a S-N transect. Station S2 is not represented since it would be overlapped to S1 and its absolute abundances are too low when compared with the other sites. below 100 m. Nitrate and phosphate reach maximum values in the Ed station whereas silicate in the Ed and in the southern stations. Nitrite shows a different pattern with concentration peaks around 40-60 m depth, more pronounced in the southern station S1 and at the top of the Senghor seamount (Sg2).

Diversity and abundance of total coccolithophores
Coccolithophores were well preserved and represented in the analyzed samples. A total of 39 distinct taxa were recognized using polarizing light microscopy, 2 identified to genus level and 37 to species level. Additional SEM analysis revealed more 7 species belonging to the genus Syracosphaera and 2 species to the genus Ophiaster. The complete list of taxa, including the rare ones and the species found at least once (*), can be consulted in the Appendix.
Total absolute abundances ranged between 0.49 × 10 3 and 30.06 × 10 3 cell L − 1 , being Emiliania huxleyi (morphotype type A) and Gephyrocapsa oceanica the dominant species, together making up to 56% of the total (Table 2). Although Bendif et al. (2019) have recently proposed that E. huxleyi and G. oceanica should be included in the genus Gephyrocapsa based on genome analysis, in the present study we preferred using the traditional taxonomic designation. The other relevant taxa for the present study, in order of decreasing abundance were: Florisphaera profunda, Gephyrocapsa ericsonii, Algirosphaera robusta, Syracosphaera spp., Helicosphaera spp. (mostly H. carteri), Umbellosphaera spp., Oolithotus spp. (mostly O. antillarum), Ophiaster spp., Discosphaera tubifera, Umbilicosphaera spp. (mostly U. sibogae), Michaelsarsia spp., Syracosphaera lamina, Reticulofenestra sessilis and Gladiolithus flabellatus. The relative abundances of the considered taxa were >2% in at least five samples and >0.3% of the total coccolithophores. However, since they are representative of certain depths, they were included in the study. The absolute abundances of these coccolithophores per sample are available at the PANGAEA database (Narciso et al., 2020).
The highest total abundances were observed at the southeastern slope of the Senghor seamount (Sg3), at a water depth of 40 m, whereas the lowest coccolithophore abundances were observed at one of the southernmost stations, S2 (Fig. 4).

Vertical distribution of the most relevant taxa
A remarkable zonation can be inferred by coccolithophore depth distribution (Fig. 5). D. tubifera and Umbellosphaera spp. show the highest occurrences between depths of 15-20 m. The group formed by E. huxleyi, G. oceanica, G. ericsonii, A. robusta, Helicosphaera spp. and Michaelsarsia spp. present the major concentrations between 40 and 50 m, although G. ericsonii also exhibits a peak at 60 m. Syracosphaera spp. and Umbilicosphaera spp. show the major increases around 50 m. The group composed by R. sessilis, F. profunda, Oolithotus spp. and Ophiaster spp. display the highest occurrences around 60 m. S. lamina shows the major increase around 80 m whereas G. flabellatus shows it between 80 and 100 m. Tracking this vertical distribution we may consider four main coccolithophore groups: a Shallow (SG) one, composed by D. tubifera and Umbellosphaera spp., linked to shallower waters above the thermocline; an Intermediate Group (IG), composed by E. huxleyi, G. oceanica, G. ericsonii, A. robusta, Helicosphaera spp., Michaelsarsia spp., Syracosphaera spp. and Umbilicosphaera spp., and more associated with the thermocline; a Deep Group (DG), formed by R. sessilis, F. profunda, Oolithotus spp. and Ophiaster spp., showing clear affinities with the nutricline; and The Deepest (TDG), formed by S. lamina and G. flabellatus (Table 3, Plate 1).

Distribution of the most relevant taxa per station
The analysis of Fig. 5 also indicates that the majority of these taxa occur mainly associated with the Senghor seamount stations. However, maxima increases of D. tubifera and Umbellosphaera spp., belonging to the shallow group, of G. flabellatus and S. lamina, from the deepest one, as well as of Michaelsarsia spp. are observed at the southernmost station S1. Other exceptions are G. oceanica with the major abundances recorded in the station Ed, Helicosphaera spp. in the SE, R. sessilis in S2 (not shown in Fig. 5) and Umbilicosphaera spp. in the northern station CVO.

Multivariate analysis
The first three components of Canonical Correspondence Analysis (CCA) explained 94% of total data variability, with the first component (CCA 1) containing 72.4% and the second (CCA 2) 16.2% of total information (Fig. 6). The eigenvalues and its percentage variance are depicted in Table 4 whereas the loadings of each variable can be consulted in Table 5. Environmental variables and the direction of its maximum change are indicated by vectors, which length is proportional to the rate of change among samples. CCA 1 is mostly represented by the opposition between temperature and dissolved oxygen, in the negative semi-axis, and density, nutrients and depth, in the positive semi-axis. Concerning the coccolithophores, there is a strong opposition between the group formed by G. flabellatus, S. lamina and R. sessilis, with positive loadings, and the SG (D. tubifera and Umbellosphaera spp.), with negative ones. F. profunda, Ophiaster spp. and Oolithotus spp., which integrate the DG are also in opposition with the SG, although presenting general lower loadings. The affinity among taxa belonging to the IG is also inferred by this figure analysis. Positive correlations between the SG and temperature, as well as dissolved oxygen are disclosed. The same is established between the groups DG and TDG and the variables density, nutrients and depth. With much lower variance CCA 2 reflects the influence of salinity and fluorescence, in its negative semi-axis. Along it the groups SG and TDG plus R. sessilis (positive loadings) are kept relatively separated from the IG and DG groups, confirming its usefulness as a discriminating tool.

Environmental controls on coccolithophore species distribution
A total of 39 taxa were observed in the samples off Cabo Verde, from which only 16 were considered relevant for the present study. In the upper water column of this oligo-to -mesotrophic region the standing crop of coccolithophores was relatively low, with the maximum total cell densities reaching 30 × 10 3 cell L − 1 in the southeastern slope of the Senghor seamount (Sg3). In addition, cell densities were also significant at the Senghor top (Sg2) and at the Eddy station (Ed) (Fig. 4). Comparatively, maximum coccolithophore abundance off the Madeira archipelago, also in the northeast subtropical Atlantic, was found to be four times higher (Narciso et al., 2019). Cabo Verde's relatively low Table 3 Depth-related groups from Cabo Verde and the relative abundance of each taxon in the associated water depth. coccolithophore cell densities, with the lowest observed at the southernmost station S2 (Fig. 4), are in accordance with the regional seasonal variability in coccolithophore fluxes, with minima during fall (Köbrich and Baumann, 2009;Köbrich et al., 2015). Although the southernmost stations could have been influenced by the Guinea Dome during the surveyed interval, our data did not reveal any significant contribution of this submarine geomorphological structure to the coccolithophore stock in the region. In closer detail, Cabo Verde coccolithophore distribution showed hydrographic controls on the distribution of some of the major taxa; temperature, dissolved oxygen, density, nitrate and nitrite had the greatest influence on the assemblage variance (Fig. 6). The assemblage was dominated by E. huxleyi and G. oceanica, with a contribution of 56%, followed by F. profunda. These results are similar with previous trap studies in the region (Köbrich and Baumann, 2009;Guerreiro et al., 2019), however, these always refer the preponderance of F. profunda over G. oceanica. In fact, the high abundances of G. oceanica found in the present work are mostly due to station Ed contribution (Fig. 5). Initially detected from geostrophic currents, oceanographic data confirmed the occurrence at that time of a cyclonic eddy, which was the reason behind  the recovery of water samples at station Ed (Christiansen et al., 2016). However, with well-marked oceanographic features in higher depths, the presence of the eddy is not obvious from the uppermost water column hydrographic data. A typical feature of this region is the presence of large eddies that seed near the coast and propagate offshorewards, keeping inside water mass that retains its physico-chemical properties. Known for dominating the coastal coccolithopore communities (e.g. Andruleit et al., 2000;Guerreiro et al., 2014;Narciso et al., 2016;, G. oceanica may have been trapped inside the eddy, being transported through the currents and continuing to thrive until abnormal high values that are not matched by any other station. At same time, the development of this cold-core eddy surely led to significant nutrient enrichment in the euphotic zone around its perimeter (see Fig. 3), contributing to the maximum fluorescence peak at the Ed station (Fig. 2). The synchronous nutrient decrease during the chl-a peak may reflect the rapid growth and uptake capacity of some phytoplankton organisms within relatively nutrient-rich and turbulent environment. E. huxleyi was the most abundant species present off Cabo Verde, although it was found in relatively low absolute cell numbers compared to much higher cell concentrations reported from the high nutrient systems. This species is usually linked to productive water mass conditions, as well as upwelling regimes (e.g. Silva et al., 2009;Guerreiro et al., 2013;Ausín et al., 2018). A cosmopolitan and opportunistic behavior is also associated to E. huxleyi, since it may dominate the coccolithophore flora during most of the year, through high and low productive periods (Winter et al., 1994;Baumann et al., 2000;Andruleit, 2007). Thus, the low E. huxleyi abundances found in our samples may indicate a region slightly reduced in nutrient concentration, due to the almost non-existent migration of nutrient-enriched filaments from the coast, during the surveyed interval. Although eddies are features typically associated with upwelling systems, which dynamics may act as a trigger to costal filaments (Barton et al., 2004;Meunier et al., 2010Meunier et al., , 2012, there are no evidences in our data showing that. In the Mauritania-Senegalese upwelling zone (12-19 • N), upwelling occurs during the winter months and fades during the summer months, related to the seasonal migration of the trade winds (Cropper et al., 2014). In addition, and based on sea surface temperature upwelling indices, the downwelling regime is also present in November, when the NE trade winds are still weak and the ITCZ in a northern position. So, during the surveyed interval (late fall) we may facing a transition to upwelling conditions, which reach their peak later in February. The dominance of E. huxleyi type A over type B in our study corroborates that there is no influence of upwelling extensions from the northwest African coast, since morphotype type A shows a preference for warm waters with low nutrient conditions (Findlay and Giraudeau, 2000;Mohan et al., 2008;Patil et al., 2017). This fact is also consistent with previous observations of Lathuilière et al. (2008). Also with high affinity for warm and highly-stratified water column, but living in the surface layer, is Umbellosphaera spp. (Kinkel et al., 2000;Haidar and Thierstein, 2001;Poulton et al., 2017). In our study, the high abundances of Umbellosphaera spp. and D. tubifera characterize the shallower and mixed layer of the well-stratified water column from the surveyed area. The notable negative correlation between the cell densities of these taxa and the nutrient concentration (Fig. 6), clearly demonstrates that Umbellosphaera spp. and D. tubifera are adapted to low nutrients, consistent with the observations of Baumann et al. (2008).
Oligotrophic regions tend to be favorable to species living close to the nutricline. F. profunda is known for its high affinity for warm and highly-stratified conditions characteristic of tropical and subtropical open-ocean regions (Kinkel et al., 2000;Haidar and Thierstein, 2001;Poulton et al., 2017), indicating the vicinity of the nutricline (Winter and Siesser, 1994;Kinkel et al., 2000). At the present study, F. profunda pattern is rather consistent with these remarks. In addition, during the surveyed time interval, F. profunda may have benefited from low irradiance, since this species is apparently adapted to low light conditions (Quinn et al., 2005). Furthermore, its habitat close to the nutricline would have provided the required nutrients during this time of lower nutrient input. The position of G. flabellatus, S. lamina and R. sessilis in the ordination diagram, opposite to the SG, shows they are typically deep-living species. The high affinity displayed among F. profunda, Ophiaster spp. and Oolithotus spp. shows their preference for cooler and deeper waters, as well as nutrients availability. Concerning the taxa positioned around the central zone of the diagram, namely Syracosphaera spp. and Helicosphaera spp., the environmental control for their distribution is complex. Syracosphaera is the most diverse genus, including species reported from different zones and various ecological habitats (Jordan and Chamberlain, 1997;Dimiza et al., 2008). In this study, coccospheres of 12 Syracosphaera species were recorded, justifying the ambiguous interpretation related to these taxa. Helicosphaera spp. are known to have a clear affinity for mesotrophic conditions in a well-mixed upper water column (Haidar and Thierstein, 2001;Ziveri et al., 2004;Boeckel et al., 2006). However, and despite the low significance of the loadings, the present data reveal a link to warm and low nutrient concentration. The same observation was reported in Guerreiro et al. (2019), based on a site positioned close to our study's region, during a stable and thermally stratified water column. Umbilicosphaera is also a genus that causes some controversy. On one hand it is known for its affinity for warm and oligotrophic habitats (Winter and Siesser, 1994;Baumann et al., 2016;Liu et al., 2018), which is evidenced in the CCA biplot. On the other hand it is influenced by water-mass stratification and nutrient level , being also known for its preference for mesotrophic conditions (Hagino and Okada, 2006;Narciso et al., 2016Narciso et al., , 2019. G. ericsonii shows a positive correlation with salinity, fluorescence and nitrate. If on one hand this taxon has ecological demands similar to E. huxleyi (Haidar and Thierstein, 2001), on the other hand prefers subsurface waters with enhanced nutrient concentrations , which is closer to our findings. Positioned next to G. ericsonii in the CCA diagram, A. robusta reveals similar ecological affinities with this species, being associated to moderate productive zones (Narciso et al., 2019), from deeper (Takahashi and Okada, 2000;Andruleit, 2007) to shallower depths (Jordan and Winter 2000;Malinverno et al., 2003). Michaelsarsia spp. is mostly correlated with salinity, oxygen and temperature, and is known for its preference for moderate productive zones (Narciso et al., 2019), as well as shallower near-shore zones with dissolved oxygen increase trends, generally warm and moderately high-salinity environments (Agbali, 2014).

Coccolithophore depth distribution
The sampling resolution performed off Cabo Verde, along with the inferred well-stratified water column, allowed the identification of typical depth-related assemblages from the depth-distribution of the species. In subtropical waters, this common vertical separation of coccolithophore assemblages allows the recognition of two main photic zones: Upper Photic Zone (UPZ) and Lower Photic Zone (LPZ). Several studies refer the upper 80 m as the UPZ depth range (Winter et al., 1994;Jordan and Chamberlain, 1997) and the depth below 80-100 m Cortés et al., 2001) or even 120 m for the LPZ (Winter et al., 1994;Jordan and Chamberlain, 1997). However, these typical depth ranges are in fact coarse approaches, since local conditions cause significant variations in the actual zonal boundaries. Since the LPZ is commonly characterized by the dominance of F. profunda, we may consider that in the surveyed region off Cabo Verde, and during the late fall, the LPZ was in a higher position and, consequently, narrowing the UPZ (~the upper 60 m). Thus, in the surveyed UPZ two distinct assemblages or groups could be recognized: a shallow one and an intermediate. Typical members of the SG, thriving mainly between 10 and 30 m of the mixed layer above the thermocline, were D. tubifera and Umbellosphaera spp. The assemblage part of the IG, characteristic from the 40-50 m, was formed by the three placolith-bearing species E. huxleyi, G. ericsonii and G. oceanica, and by A. robusta, Helicosphaera spp., Michaelsarsia spp., Syracosphaera spp. and Umbilicosphaera spp. Some of these taxa are already known to occupy lower water depths, being associated with the middle photic zone (MPZ, between 70/80 m and 100/125 m; Jordan and Chamberlain, 1997;Jordan and Winter 2000;Cortés et al., 2001) or even with the LPZ, such as A. robusta (Jordan and Chamberlain, 1997;Hagino et al., 2000;Malinverno et al., 2003). Surprisingly, A. robusta seemed to be quite well adapted to the oceanographic conditions at 40-50 m depth, showing high related relative abundances (65% , Table 3). However, it is important to refer that the optimal depth-range of the species may vary seasonally, especially when surface water stratification is more pronounced (e.g. Reid, 1980). Within the LPZ, two more distinct assemblages could be recognized: a deep one and the deepest. Typical members of the DG, which was representative of the 60-75 m water depth and of the nutricline level, were F. profunda, Ophiaster spp., Oolithotus spp. and R. sessilis. These taxa are well known as low photic dwellers, but from lower depths. However, and according to Cheng and Wang (1998), these species may thrive better when the water column above is clear and the mixed layer is shallow, conditions observed in the current study. Concerning F. profunda, probably the best known deep-living species, there was not a clear dependence showing that it could only live within a rather specific depth range. Although this species dominated in this 60-75 m depth interval (42%, see Table 3), the second highest F. profunda absolute abundances were observed around 50 m depth, associated with the top and the southeastern slope of the Senghor seamount. The single 50 m peak of F. profunda observed in the station Sg3 may be explained by the associated shallower nutricline (Fig. 3). At the same time, the shoaling of the nutricline may have been responsible for the increase in E. huxleyi and G. ericsonii, which in turn increased the total absolute abundances. Under these circumstances, a thicker mixing layer (down to 50 m; see Fig. 2) is acceptable but also tricky, since it is the limit for the more important cell concentrations of F. profunda. The highest coccolithophore densities recorded in the SE slope of the Senghor seamount, as well as over the top, demonstrate the strong influence of this topographic feature on the development of this phytoplankton group. The shallower depth range found for F. profunda is consistent to what is described in Andruleit et al. (2003). Regarding to O. antillarum, similar depths were also found in previous works (Hagino et al., 2000;Andruleit et al., 2003) with the first putting this species below the thermocline, and the latter at the thermocline. The deepest assemblage, characteristic from depths below 80 m, was formed by G. flabellatus and S. lamina, consistent with the descriptions of Knappertsbusch (1993), Jordan and Chamberlain (1997), Malinverno et al. (2003), among others.
Following a different approach, Poulton et al. (2017) proposed the existence of three floral groups to describe the vertical distribution of coccolithophores in the Atlantic subtropical gyres, equatorial and temperate waters. Thus, and based on surface irradiance, these authors considered a different terminology: Upper Euphotic Zone (UEZ, >10% surface irradiance); Lower Euphotic Zone (LEZ, 10-1% surface irradiance); and Sub-Euphotic Zone (SEZ, <1% surface irradiance). Fig. 7 shows the position of our typical groups/assemblages against optical and percentage irradiance depths. For its construction we assumed that the deep chlorophyll/fluorescence maximum (DCM) approximates the depth of 1% surface irradiance (optical depth = 4.6; see description in Poulton et al., 2017). Considering our data (Fig. 2), a DCM at 50 m depth was applied. Actually, this depth is closer to the ones observed around the tropical Equator (50-60 m; Monger et al. (1997); Poulton et al. (2006)) than the ones generally accepted for the subtropical gyres (~105, Pérez et al., 2006;~125 m, Poulton et al., 2006). However, it should be noted that: (i) Cabo Verde archipelago is very close to the province border of equatorial waters; (ii) the nutricline becomes increasingly shallower towards the east, when analyzing an E-W Atlantic transect (Guerreiro et al., 2019); (iii) and the nutricline during fall should present a relatively shallower position, since the deepening of the nutricline reach its maximum during the summer in consequence of maximum thermal stratification (Cermeño et al., 2008). The depth distribution of Cabo Verde floral assemblages is rather consistent with the floral groups described in Poulton et al. (2017), with just few discrepancies like the position of the shallow group. However, in the Equatorial context these authors also place the lower limit of the UEZ floral assemblage below 10% of the surface irradiance, similar to the position occupied by the SG (see Fig. 7). The maximum abundance of G. flabellatus well below F. profunda, as well as below the 0.1% of the surface irradiance, was also found by the previous authors and brings questions about the nutritional strategies of such kind of species. The existence of mixothrophy that combine photoautotrophy and phagotrophy is a theory that has been defended by several authors (e.g. Billard and Inouye, 2004;De Vargas et al., 2007;Poulton et al., 2017;Millán and Winter 2019) and recently proven by Avrahami and Frada (2020).

Conclusions
The present study on biodiversity and on spatial and vertical distributions of coccolithophores off Cabo Verde archipelago led to the following main conclusions: 1. A total of 39 taxa were observed, from which only 16 were considered relevant for the present study. E. huxleyi and G. oceanica were the dominant species, followed by F. profunda; 2. Total cell densities were relatively low, reaching to a maximum of 30 × 10 3 cell L − 1 . The highest coccolithophore standing stock was found in the upper 50 m over the Senghor seamount, especially over its southeastern slope; 3. The high abundances of G. oceanica were mainly due to the Eddy station contribution. The formation of this eddy near the African coast may have trapped the coccolithophore community dominated by G. oceanica, promoting its transport and thriving offshorewards; 4. Four typical depth-related groups were observed: (i) a Shallow oligotrophic (10-30 m), represented by D. tubifera and Umbellosphaera spp.; (ii) an Intermediate (40-50 m), formed by the three placolith-bearing species E. huxleyi, G. ericsonii and G. oceanica, and by A. robusta, Helicosphaera spp., Michaelsarsia spp., Syracosphaera spp. and Umbilicosphaera spp.; (iii) a Deep (60-75 m) with F. profunda, Ophiaster spp., Oolithotus spp. and R. sessilis as typical members; (iv) and The Deepest (>80 m), composed by G. flabellatus and S. lamina; 5. An overall shallowing of the entire coccolithophore community, with the UPZ compressed within the first 60 m. Both the UPZ and LPZ communities disclosed a clear subdivision into sub-communities each, with the UPZ subdivided into the Shallow and Intermediate groups and the LPZ subdivided into the Deep and The Deepest groups; 6. The coccolithophore distribution off Cabo Verde was essentially explained by the general relatively warm and nutrient-depleted waters in the region during the surveyed interval, in result of the weaker NE trade winds and associated northward migration of the ITCZ.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.