Variability of phyto-and zooplankton communities in the Mauritanian coastal upwelling between 2003 and 2008 2

16 Continuous multiyear records of sediment trap-gained microorganism fluxes are scarce. Such studies are important to identify and to understand the main forcings behind seasonal and 18 multiannual evolution of microorganism flux dynamics. Here, we assess the long-term flux variations and population dynamics of diatoms, coccolithophores, calcareous and organic 20 dinoflagellates, foraminifera and pteropods in the Eastern Boundary Upwelling Ecosystem (EBUE) of the Canary Current. A multiannual, continuous sediment trap experiment was conducted at the 22 mooring site CBeu (Cape Blanc eutrophic, ~20°N, 18°W; trap depth = ca. 1,300 m) off Cape Blanc, Mauritania (northwest Africa), between June 2003 and March 2008. Throughout the study, the 24 reasonably consistent good match of fluxes of microorganisms and bulk mass reflects the seasonal occurrence of the main upwelling season and the contribution of microorganisms to mass 26 flux off Mauritania. A clear successional pattern of microorganisms, i.e. primary producers followed by secondary producers, is not observed. High fluxes of diatoms, coccolithophores, organic 28 dinoflagellates cysts, and planktonic foraminifera occur simultaneously. Peaks of calcareous https://doi.org/10.5194/bg-2019-314 Preprint. Discussion started: 27 August 2019 c © Author(s) 2019. CC BY 4.0 License.


Introduction 52
A way to obtain insights into the impact of climate variability on marine ecosystems is monitoring multiannual evolution and changes covering key species or groups of species information about organism groups of different trophic levels are practically unknown or cover only a few species (e.g., Schlüter et al., 2012;Rembauville et al., 2016). 60 Eastern Boundary Upwelling Ecosystems (EBUEs) are among the most important marine ecosystems, both ecologically and economically (Cropper et al., 2014). Despite the fact that they 62 cover only 10% of the global surface ocean area, they provide about 25% of the global fish catch (Pauly and Christensen, 1995) and build extraordinary marine hotspots of high primary production 64 and biodiversity (Arístegui et al., 2009). In doing so, they play a key role in the marine biological pump, as together with other continental margins may be responsible for more than 40% of the 66 CO 2 ocean sequestration (Muller-Karger et al., 2005). As EBUEs are highly dynamic with large seasonal and interannual variability (e.g., Chavez and Messié, 2009;Fischer et al., 2016), gaining 68 information on their long-term variability is essential to understand their potential response to current global climate change. 70 One of the EBUEs that have been thoroughly studied over the past three decades is the coastal ecosystem off Mauritania (northwest Africa), which is part of the Canary Current (CC) EBUE 72 (Cropper et al., 2014). The Mauritanian system is characterized by intense offshore Ekman transport and strong mesoscale heterogeneity, which facilitate the exchange of neritic and pelagic 74 water masses (Mittelstaedt, 1983;Zenk et al., 1991;Van Camp et al., 1991;Arístegui et al., 2009;Chavez and Messié, 2009;Meunier et al., 2012;Cropper et al., 2014). In addition, regional factors 76 such as nutrient trapping efficiency (Arístegui et al., 2009), the giant chlorophyll filament (Gabric et al., 1993;Barton, 1998;Lange et al., 1998;Helmke et al., 2005); dust deposition (Fischer et al., 78 2016(Fischer et al., 78 , 2019 and/or the shelf width (Hagen, 2001;Cropper et al., 2014) strongly affect the temporal dynamics of primary and secondary producers communities in surface waters along the 80 Mauritanian coast. In this ecosystem, several long-term, continuous, sediment trap-based monitoring records are available since the late 1980s. Until now, studies monitoring variability of 82 this seasonally dynamic ecosystem mostly focused on the variability of bulk fluxes (Fischer et al., 1996(Fischer et al., , 2009(Fischer et al., , 2016(Fischer et al., , 2019Bory et al., 2001;Marcello et al., 2011;Skonieczy et al., 2013), particular 84 groups of microorganisms (Lange et al. 1998;Romero et al., 1999Romero et al., , 2002Romero et al., , 2003Köbrich and Baumann, 2008;Romero and Armand, 2010;Zonneveld et al., 2010;Köbrich et al., 2016;Romero 86 6 Plankton variability off Mauritania northeasterly winds cause the coastal upwelling to move further offshore and the upper slope is filled with upwelling waters. South of Cape Blanc (Fig. 1), northerly winds dominate year through 144 but surface waters remain stratified and the PUC occurs as a subsurface current. South of Cape Timiris (ca. 19°30'N), the PUC intensifies during summer-fall and remains at the subsurface during 146 winter-spring (Pelegrí et al., 2017). The encountering of the northward flowing MC-PUC system with the southward flowing currents in the Canary Basin leads to flow confluence at the CVFZ 148 (Zenk et al., 1991) and the offshore water export visible as the giant Mauritanian chlorophyll filament (Gabric, 1993;Pelegrí et al., 2006;Pelegrí et al., 2017). This filament extends over 300 150 km offshore (e.g., Van Camp et al., 1991;Arístegui et al., 2009;Cropper et al., 2014) and carries South Atlantic Central Water (SACW) offshore through an intense jet-like flow (Meunier et al.,152 2012; Fig. 1). Intense offshore transport forms an important mechanism for the export of cool, nutrient-rich shelf and upper slope waters offshore Mauritania. Based on satellite imagery and in 154 situ data, it has been estimated that the giant Mauritanian filament could export about 50% of the particulate coastal new production offshore toward the open ocean during intervals of most intense 156 upwelling, while coastal phytoplankton at the surface might be transported as far as 400 km offshore (Gabric et al., 1993;Barton, 1998;Lange et al., 1998;Helmke et al., 2005). The transport 158 effect could extend to even more distant regions in the deep ocean, since sinking particles are strongly advected by lateral transport (Fischer and Karakaș, 2009;Karakaș et al., 2006, Fischer et 160 al., 2009. The nutrient concentration of the upwelled waters off Mauritania varies depending on their origin 162 (Fütterer, 1983;Mittelstaedt, 1991;Zenk et al., 1991). The source of upwelling waters off Mauritania are either North Atlantic Central Water (NACW), north of about 23°N, or SACW (south 164 of 21°N, Fig. 1). Both water masses are mixed in the filament area off Cape Blanc. The SACW occurs in layers between 100 and 400 m depth off Cape Blanc and the Banc d'Arguin. The 166 hydrographic properties of the upwelling waters on the shelf suggest that they ascend from depths between 100 and 200 m south off the Banc d'Arguin (Mittelstaedt, 1983). North of it, the SACW 168 merges gradually into deeper layers (200-400 m) below the CC (Mittelstaedt, 1983). During intense upwelling, the stratification of the shelf waters weakens, and so is the stratification further offshore, 170 usually within the upper 100 m (Mittelstaedt, 1991 Table 1) and 0.5 m² openings, equipped with a honeycomb baffle (Kremling et al., 1996), were used. Traps 182 were moored in intermediate waters (between 1,256 m and 1,296 m) and sampling intervals varied between 6.5 and 23 days (Table 1). Uncertainties with the trapping efficiency due to strong 184 currents (e.g. undersampling, Buesseler et al., 2007) and/or due to the migration and activity of zooplankton migrators ('swimmer problem') are assumed to be minimal in this depth range. 186 Prior to each deployment, sampling cups were poisoned with 1 ml of concentrated HgCL 2 per 100 ml of filtered seawater. Pure NaCl was used to increase the density in the sampling cups up to 188 40 ‰. Upon recovery, samples were stored at 4°C and wet-split in the MARUM sediment trap laboratory (University Bremen, Bremen) using a rotating McLANE wet splitter system. Larger 190 swimmers, such as crustaceans, were handpicked with forceps and removed by carefully filtering through a 1-mm sieve. All flux data hereafter refer to the size fraction of <1 mm. Detailed 192 information about sampling and laboratory analysis is given in Mollenhauer et al. (2015) where the bulk fluxes are given for the deployments CBeu 1-4. Additionally to the fluxes, alkenone derived 194 sea surface temperature (SST) for the CBeu deployments 1-4 were provided by these authors.
Using ¼ or 1 / 5 wet splits, analysis of the <1 mm fraction was carried out (Fischer and Wefer, 196 1991;Fischer et al., 2016). Samples were freeze-dried and homogenized before being analyzed for bulk (total mass), organic carbon (OC), calcium carbonate (CaCO 3 ) and biogenic silica (BSi,198 opal carbonate with 2 N HCl. Overall analytical precision based on internal lab standards was better than 0.1% (±1σ). Carbonate was determined by subtracting OC from total carbon, the latter being 202 measured by combustion without pre-treatment with 2N HCl. Organic matter was estimated by multiplying the content of total organic carbon by a factor of two as about 50-60% of marine 204 organic matter is constituted by OC (Hedges et al., 2002).
BSi was determined with a sequential leaching technique with 1M NaOH at 85°C (Müller and 206 Schneider, 1993). The precision of the overall method based on replicate analyses is mostly between ±0.2 and ±0.4%, depending on the material analyzed. The lithogenic fluxes were 208 estimated by subtracting the masses of CaCO 3 , BSi, and 2 x OC from the total mass flux.

Diatoms
For this study 1/25 and 1/125 splits of the original samples were used. Samples were prepared 212 for diatom studies following the method proposed by Schrader and Gersonde (1978). A total of 185 sediment trap samples were processed. Each split was treated with potassium permanganate, 214 hydrogen peroxide, and concentrated hydrochloric acid following previously used methodology (Romero et al., 2002(Romero et al., , 2009a(Romero et al., , b, 2016(Romero et al., , 2017. Identification and count of the species assemblage 216 were done on permanent slides (Mountex® mounting medium) at x1000 magnification using a Zeiss ® Axioscop with phase-contrast illumination (MARUM, Bremen). The counting procedure and 218 definition of counting units follows Schrader and Gersonde (1978). Depending on valve abundances in each sample, several traverses across each slide were examined. Total amount of 220 counted valves per slide ranged between 300 and 800. At least two cover slips per sample were scanned in this way. Valve counts of replicate slides indicate that the analytical error of 222 concentration estimates is ca. 10 %. The resulting counts yielded abundance of individual diatom taxa (absolute and relative) as well as daily fluxes of valves per m-2 d-1 , calculated according to 224 Sancetta & Calvert (1988).

Coccolithophores 226
Aliquots of each sample were 1/125 of the <1 mm fraction. Depending on the total flux, samples were further split down to 1/625 to 1/2500 of the original sample volume and were filtered 228 afterward onto polycarbonate membrane filters (Schleicher and Schuell TM 47mm diameter, 0.45µm

Organic-walled and calcareous dinoflagellate cysts 236
1/125 splits of the original trap material was ultrasonically treated and sieved with tap water through a high precision metal sieve (Storck-Veco®) with a 20µm pore size. The residue was 238 transferred to Eppendorff cups and concentrated to 1 ml of suspension. After homogenization of the material, a known aliquot was transferred to a microscope slide where it was embedded in 240 glycerin-gelatine, covered with a cover slip and sealed with wax to prevent oxidation of the organic material. After counting, cyst fluxes were calculated by multiplying the cyst counts with 242 the aliquot fraction and the split size (1/125) and dividing through the amount of days during which the trap material was sampled as well as the trap-capture surface. No chemicals were 244 used to prevent dissolution of calcite and silicate. Cyst assemblages were determined by light microscopy (Axiovert, x400 magnification). Taxonomy of organic walled dinoflagellate cysts is 246 according to Zonneveld and Pospelova (2015), taxonomy of calcareous dinoflagellate cysts is according to Vink et al. (2002) and Elbrächter et al. (2008)

Planktonic foraminifera and pteropods
Depending on the absolute magnitude of the total mass flux, a 1/5 or a 1/25 split of the wet 250 solution (fraction <1mm) was used to pick planktonic foraminifers and pteropods (pelagic mollusks). Specimens of both groups of calcareous microorganisms were rinsed three times by 252 using tap water, dried at 50°C in an oven overnight and then separated from each other.  A slightly different, miniaturized analysis procedure has been applied for the CBeu trap 5 266 samples. 1/5 wet splits of the freeze-dried <1mm fraction were weighted in 10 ml Pyrex tubes and a known amount of an internal standard (n-Nonadecan-2-one) was added. Samples were then 3x 268 ultrasonically extracted with a mixture of 3 ml dichloromethane/methanol (9:1 vol./vol.), centrifuged and the supernatant solvent combined as total lipid extract (TLE). TLEs were evaporated to 270 dryness and saponified in a 0.1M potassium hydroxide solution in methanol/water (9:1 vol./vol.) for two hours at 80°C. Neutral lipids, recovered with hexane, were afterwards separated into fractions 272 of different polarity by silica gel chromatography and elution with hexane, dichloromethane/hexane (1:1 vol./vol.) and dichloromethane/methanol (9:1 vol./vol.), respectively. The second fraction 274 containing the alkenones was dried, re-dissolved in 20µl hexane and analyzed on a 7890A gas chromatograph (GC, Agilent Technologies) equipped with a cold on-column injection system, a 276 DB-5MS fused silica capillary column (60 m, ID 250 µm, 0.25 µm film coupled to a 5 m, ID 530 µm deactivated fused silica precolumn) and a flame ionization detector (FID). Helium was used as 278 carrier gas (constant flow, 1.5 mL/min) and the GC oven was heated using the following temperature program: 60 °C for 1 min, 20 °C/min to 150 °C, 6 °C/min to 320 °C and a final hold 280 time of 35 min. Alkenones were identified by comparison of the retention times with a reference sample composed of known compounds. Peak areas were determined by integrating the 282 respective peaks.
The U 37 K' index was calculated using the following equation (Prahl and Wakeham, 1987) 284  (Table 1). In the research area, SST at the trap position is influenced by seasonal air 292 temperature changes as well as the presence of upheld water surfacing at the trap position.
To compensate for seasonal air temperature changes the SSTA is calculated by subtracting the

Multivariate analyses 302
The ordination techniques Principal Component (PCA) and Redundancy (RDA) analyses have been performed with the software Package Canoco 5 (ter Braak and Smilauer, 2012;Smilauer and 304 Leps, 2014). To obtain insights into the temporal relationship between fluxes of organism groups (diatoms, coccolithophores, organic-walled dinoflagellate cysts, calcareous dinoflagellate cysts, 306 planktonic foraminifera and pteropods) and bulk components as well as the environmental conditions in surface waters and low atmosphere a RDA has been performed. RDA compares the 308 total flux of organism groups with environmental parameters and TOC, BSi, CaCO 3 and lithogenic fluxes ( Table 2) To better understand the relationship within the individual organism groups, a PCA has been performed ( Table 2). For these analyses, the total flux of the organisms/species groups have been 320 normalized to values between 0 and 1000 according to formula 2:

Bulk fluxes and fluxes of organism groups 332
On average, the carbonate fraction (CaCO 3 ) dominates the mass flux (41% to the total mass flux) and is mainly composed of coccolithophores, foraminifera, calcareous dinoflagellates and 334 pteropods (see also Fischer et al., 2009Fischer et al., , 2016. CaCO 3 is followed by BSi (average = 14.5%, mostly diatoms, Romero and Fischer, 2017), and organic carbon (6.5%, delivered by diatoms, 336 coccolithophores and organic dinoflagellate cysts). The lithogenic fraction -mostly composed of mineral dust-makes up 31.5% of the total mass for the entire sampling period of CBeu 1-5 (2003-338 2008, Table 1). Bulk fluxes for the CBeu deployments 1 to 4 were already presented in

Species-and group-specific composition of assemblages 368 Plankton variability off Mauritania
The studied plankton community at the CBeu site is highly diverse and is composed by at least 220 identified species. Table 3 presents the species-specific composition of groups depicted in Fig.  370 4.
Out of 170 marine diatom species, the 70 most abundant diatom taxa (average relative 372 contribution >0.75% for the entire studied interval) were attributed in four groups, according to the main ecological conditions they represent: (1) benthic, (2) coastal upwelling, (3) coastal planktonic, 374 and (4) open-ocean waters (see also Romero and Fischer, 2017  the predominantly high spring-summer total diatom flux remained unaltered (Fig. 3a).
Coccolithophores are consistently dominated by Emiliania huxleyi and Gephyrocapsa 384 oceanica, whose contribution is always higher 50% of the community throughout the sampling period (Fig. 4b). Oligotrophic upper photic zone (UPZ, e.g., Umbellosphaera tenuis, U. irregularis) 386 and lower photic zone species (LPZ, e.g., Florisphaera profunda, Gladiolithus flabellatus) make up the majority of the remaining species. Whereas E. huxleyi showed a less clear seasonal pattern, 388 G. oceanica tends to be more abundant during late spring and early fall (Fig. 4b). In contrast, UPZ and LPZ taxa have higher relative contributions during winter and spring. The appearance of 390 Umbilicosphaera anulus (present in consistently low relative abundances of 5-10% until the summer of 2006) accounts for up to 65% of the community in winter 2005/06. Other common taxa 392 with an average relative contribution >0.75% for the entire studied interval are listed in Table 3.
Organic dinoflagellates can be attributed to five groups based on the relationship between 404 their geographic distribution in surface sediments from the Cape Blanc area and the environmental conditions in surface and subsurface waters as well as long-term surveys of their 406 seasonal cyst production (Susek et al., 2005;Holzwarth et al., 2010;Smayda and Trainer, 2010;Trainer et al., 2010;Zonneveld et al., 2012Zonneveld et al., , 2013: (1) Table 2). 436 All microorganism groups are ordinated at the positive part of the first axis showing a positive relationship with all bulk parameters (Fig. 5). This implies that the fluxes of all studied 438 microorganisms groups increase with increasing fluxes of total mass, TOC, lithogenic, BSi and CaCO 3 (Fig. 5). Fluxes of planktonic foraminifera, diatoms and -to a lesser extent-440 coccolithophores and organic dinoflagellates are ordinated at the negative site of SST and, with exception of organic dinoflagellates, positive side of MLD (Fig. 5). This implies that their fluxes are 442 enhanced whenever SST is low and MLD is deep, i.e. under a well-mixed uppermost water column. Diatoms, coccolithophores, organic dinoflagellates and planktonic foraminifera also show 444 a positive correlation with SSTA, implying that enhanced fluxes of these microorganisms occur when temperature anomalies between waters overlying site CBeu and the offshore pelagial is 446 large. The fluxes of pteropods and calcareous dinoflagellate cysts are positively related to the average wind direction, and negatively to MLD and average wind speed (Fig. 5). 448 To better understand the correlation of the fluxes of the species groups within the organism groups, PCA has been performed (Fig. 6, Table 2). The first two axes correspond to 26. 3 % and 450 16.2% of the variance within the dataset respectively. Based on their ordination on the first and second axis, three groups are recognized (Fig. 6):  The atmospheric, hydrographic and biochemical conditions deliver the physical and nutrient 486 frames that determine the temporal pattern of population dynamics as recorded by the CBeu trap.
Wind and upper water conditions off Mauritania show a clear seasonal pattern of variability (Fig.  488 7a-e). The highly stratified uppermost water column (above 40 m water depth) overlying site CBeu is an effect of winds blowing mainly from the N-NE between late winter and early summer (Fig. 7a,  490 b, e). The stratification breaks down mostly in early to middle winter, when the predominant winds turn from N-NE into S-SE (Fig. 7a). Following this setting, upwelling off Mauritania reaches its 492 highest intensity between late winter/early spring and early summer (Mittelstaedt, 1991;Meunier et al., 2012;Cropper et al., 2014). The SST record (Fig. 7d) (Fig. 2b, d) support the scenario of calcareous primary and secondary producers 504 (coccolithophores, foraminifera and pteropods) dominating the plankton community in the Mauritanian upwelling system (Fischer et al., 2019). Diatoms are the main contributors to the BSi 506 flux (Fig. 3a, 2d; Romero et al., 2002;Romero and Fischer, 2017).
A strong match among fluxes of diatoms, coccolithophores, organic-walled dinoflagellate cysts 508 and planktonic foraminifera with lithogenic fluxes at times of enhanced upwelling is observed (Figs. 2e, 3 a-c, e). The RDA supports this correlation (Fig. 5). The good correlation between lithogenic 510 and microorganisms fluxes demonstrates that winds -responsible for the water column mixing off Mauritania (Mittelstaedt, 1983;Meunier et  Mauritanian ocean waters in the form of dust that it is transported from the Sahara and the Sahel (Romero et al., 2003;Friese et al., 2017). Numerous studies have thoroughly documented that the 516 particle flux off Mauritania predominantly occurs in the form of aggregates, often rich in lithogenic particles (e.g., Karakaş et al., 2009;Nowald et al., 518 2015;Fischer et al., 2016;van der Jagt et al., 2018). Recent experiments have also shown that aggregates' abundance and sinking velocities increase toward deeper waters when aggregates 520 are ballasted with lithogenic particles, whereas aggregates are not able to scavenge lithogenic material from deeper waters (van der Jagt et al., 2018). 522 A remarkable finding of our multiannual trap experiment is that flux maxima of diatoms, coccolithophores, organic-walled dinoflagellate (all primary producers) and planktonic foraminifera 524 (secondary producers) seem to occur fairly simultaneously (Figs. 3,5). We propose three possible interpretations: (i) no clear short-term succession of the microorganism groups occurred (no 526 temporal turnover in phytoplankton composition within a few days, Roelke and Spatharis, 2015), (ii) the succession is not properly captured due to low temporal resolution of some sediment trap 528 intervals (Table 1), and/or (iii) the microorganisms -originally produced in surface and subsurface waters by different communities-sink with different velocities through the water column toward the 530 ocean bottom and get 'mixed' during their sinking, mainly due to dissimilar weights and sizes of their remains. However, the high-resolution intervals of CBeu deployments 4 and 5 (up to 7.5 days 532 per sample, Oct 2006-March 2008, Table 1) should have captured a possible short-term succession of major groups (e.g., diatoms quickly reacting to increasing nutrient availability, 534 whereas photosynthetic dinoflagellates becoming more abundant during upwelling relaxation, Margalef, 1963;Jiménez-Quiroz et al., 2019). Although we do not observe a clear pattern of 536 succession within studied populations, at this stage we do not disregard either its occurrence. It should be kept in mind that the deployed traps capture those microorganism remains that reach 538 the trap cups at around 1,300 m water depth, while they do not capture green algae or cyanobacteria thriving in surface waters. CBeu traps at ca. 1,300 m water depth capture a mixed 540 signal of sinking particles from a surface catchment area of at least ca. 100 km 2 (Siegel andDeuser, 1997, Fischer et al., 2016,) due to (i) differential settling velocities of particles (Fischer and 542 https://doi.org/10.5194/bg-2019-314 Preprint.  Karakaş, 2009;, van der Jagt et al., 2018, and (ii) highly heterogeneous and dynamic surface water conditions due to filament and eddy activity off Mauritania (Mittelstaedt, 544 1991;Gabric et al., 1993;Meunier et al., 2012;Cropper et al., 2014). Additionally, the trapped signal is always affected by dissolution of particular species and/groups of organisms sinking 546 through the water column intro deeper waters (e.g., Romero et al., 1999).

Temporal variations of the species-specific composition of the plankton community 548
We are aware that 1,900 days of continuous sampling cannot deliver a definite picture of all temporal changes affecting the composition of the plankton community in the very dynamic 550 Mauritanian upwelling. However, the overall temporal pattern observed let us to propose a general sequence of seasonal variability. Most of the major microorganisms groups occur simultaneously 552 and clear successional trends are not quite distinguishable (Fig. 3). A consistent seasonal pattern in the occurrence of species or groups of species is yet recognized. Figure 4 shows the seasonal 554 evolution of populations responding to the temporal dynamics of nutrient availability, e.g. following short-period dust events (Fig. 2e) and/or vertical mixing events associated with stronger winds 556 (Fig. 7a, e). Based on the visual data examination and the statistical analysis, four groups of species are recognized (Figs. 3,6). Populations of group 1 (Dia-bent, Dia-upw, Co-Gocean, OD-558 upw, For-cold = in blue in Fig. 6) have higher relative contribution during the most intense phase of the upwelling season (mainly between late winter/early spring and early summer; Mittelstaedt, 560 1983, Cropper et al., 2014. Group 1 quickly responds to intense mixing and lowered SST at the CBeu site (Fig. 7d, e) and represents the typical upwelling-related association off Mauritania. This 562 observation confirms the ecological characterization of the species groups that has been separately presented in previous biogeographical/ecological studies (Romero et al., 2002;Kucera, 564 2007;Köbrich et al., 2008Köbrich et al., , 2016Zonneveld et al., 2013;Romero and Fischer, 2017).

Diatoms of coastal regions (Dia-coast, non-upwelling related) and those thriving in open ocean 566
waters (Dia-ocean) together with other calcareous dinoflagellates (CD-other), cosmopolitan organic dinoflagellates (OD-cosm) and 'other coccolithophores' (Co-other) are assigned to group 2 568 (in brown in Fig. 6). Except for the cosmopolitan organic dinoflagellates cysts, all components of group 2 are primary producers and occur more abundantly between early fall and late winter (Fig.  570   4), at times of deepening of the ML and upwelling relaxation (Fig. 7e). Group  under weakened upwelling, when winds predominantly blow from the N-NE, SST start decreasing after their summer peak, and the uppermost water column stratifies (Fig. 7a, d, e). 574 Except for warm waters (CD-warm) and dust input-sensitive (CD-min) calcareous dinoflagellates cysts, group 3 is mainly composed by secondary producers: warm-water planktonic 576 foraminifera and all pteropods (Fig. 4e, f). As such, this group represents the calcareous fraction of zooplankton feeding on other (primary) phytoplankton, occurring mainly during phases of 578 predominantly warmer SSTs (Fig. 7d), N-NE-originated winds (Fig. 7a) and stratified uppermost water column (Fig. 7e). The distribution and abundance of planktonic foraminifera species is 580 strongly linked to surface-water properties. SST appears to be the most important factor controlling assemblage composition of planktonic foraminifera (Kucera, 2007). Large, symbiont-bearing 582 specialists like Globigerinoides ruber and G. sacculifer are adapted to more oligotrophic and warmer waters. They show their maximum abundance in warm waters with a deeper mixed-layer 584 depth (Fig. 7e,f).
The seasonal dynamics of group 4 is similar to that of group 3 (intervals of weakened upwelling 586 conditions), but they differ in their composition: group 4 is mainly made of calcareous primary producers. These populations dominate the plankton community during intervals of weakened 588 upwelling, shallow MLD and predominantly oligotrophic water conditions. Similar to group 3, group 4 consists mainly of coccolithophores (the dominant E. huxleyi, accompanied by UPZ and LPZ, U. 590 anulus, Figs. 4b, 6), as well as organic dinoflagellate cysts characteristic for upwelling relaxation phases (CD-upw relax). The contribution of E. huxleyi and accompanying coccolithophore taxa, 592 and upwelling-relaxation organic dinoflagellate cysts shows highest relative values from early fall through early spring and decreases into the most intense upwelling season (when G. oceanica 594 increases, Fig. 4b). As such, this group also bears some resemblance to group 2, though coastal and open-ocean water diatoms are component of the latter, while diatoms are absent in group 4. 596 This difference possibly reflects the distinct nutrient and water depth conditions in which E. huxleyi and other coccolithophores (group 4) and diatoms (group 2) typically thrive. The persistent seasonal pattern of the groups' and species occurrence experiences occasional 600 shifts. Several events, which altered the 'regular' pattern of temporal occurrence of species or group species at site CBeu, were observed between late 2004 and late 2006 (Fig. 7f-j). We identify 602 three main shift stages in the species-specific composition of assemblages: 1.
A certain degree of interannual variability of the physical setting (Mittelstaedt, 1983(Mittelstaedt, , 1991614 Cropper et al., 2014) might explain the shifts in the species-specific composition of the assemblages. The almost disappearance of warm-water planktonic foraminifera in 2004 (Fig. 7f) 616 was most probable the response to lower-than-usual water temperatures (Fig. 7d). However, the SST decrease is not recorded by satellite imagery. The overall climate evolution indicates a longer 618 warm and dry period from 2001-2004 in the Sahel and Sahara (east of site CBeu) and anomalously warm temperatures in the Eastern Atlantic (Zeeberg et al., 2008;Alheit et al., 2014). 620 2004 is the only year of our study with the largest lag between satellite and U !" !" -based temperature (Fig. 7d). This temperature gap suggests a certain decoupling between the temperature signal of 622 the uppermost centimeters of the water column (satellite) and subsurface waters where the alkenone-forming coccolithophores dwell (E. huxleyi and G. oceanica; Conte et al., 1995). As 624 planktonic foraminifera mainly react to SST variability (Kucera, 2007), cooler than usual subsurface waters between middle winter and early fall 2004 (Fig. 7d) might have been responsible for the 626 strong decrease of the warm-water planktonic foraminifera contribution (Fig. 7f) (Fig. 2e, 7e).
Exceptionally, the winter season 2004/2005 is characterized by a high total flux (Fig. 2a); this 630 extraordinarily high seasonal value matches well highest fluxes of TOC and CaCO 3 for the studied interval. 632 The extraordinary high relative abundance of U. anulus in fall 2005 has not yet been observed in similar or other settings, although it is often listed in studies of large-scale distribution patterns of 634 coccolithophores (e.g., Böckel and Baumann, 2008;Estrada et al., 2016;Poulton et al., 2017). So far only Steinmetz (1991) has found U. anulus (described as U. calvata and U. scituloma) in 636 'frequent' abundances in sediment traps deployed in the equatorial Atlantic, central Pacific, and in the Panama Basin, but without adding appropriate information such as fluxes, the timing of its 638 occurrence or its ecological significance. In most of earlier trap studies, U. anulus has been grouped together with other umbilicosphaerids coccolithophores, since it did not reached high 640 abundances (e.g., Köbrich et al., 2016;Guerreiro et al., 2017). Nevertheless, umbilicosphaerids seem to favor warm and more oligotrophic conditions , so that the increased 642 input of tropical surface waters transported northward via the MC (Mittelstaedt, 1991)  7h)-can be also explained by the increased influence of warmer surface waters of southern origin.
Heliconoides inflatus is known as a rather cosmopolitan species, occurring across a wide range of 648 oceanic provinces (Bé and Gilmer, 1977;Burridge et al., 2017), whereas L. bulimoides seems to prefer waters of subtropical gyres (although it was also present in low numbers in the equatorial 650 region, Burridge et al., 2017). A stronger transport of the MC from the south may have led to the deterioration of the adequate environmental conditions for H. inflatus, as can be seen by the 652 extremely low total pteropods flux during winter 2005 to spring 2006 (Fig. 3e), and, thus, to the relative enrichment of L. bulimoides. The fact that the latter species is absent again in winter 2008 654  (Romero and Fischer, 2017). Observational and model 660 experiments show that the transport of particles from the Mauritanian shelf and the uppermost slope via nepheloid layers significantly contributes to the deposition upon the lowermost slope and 662 beyond than the direct vertical settling of particles from the surface layer (Nowald et al., 2014;Karakaş et al., 2006;Fischer et al., 2009;Zonneveld et al., 2018). The relevance of advective 664 processes within nepheloid layers has been already proposed for similar settings (Puig and Palanques, 1998;Inthorn et al., 2006). We speculate that the longer predominance of N-NE winds 666 between 2005 and 2007 (Fig. 7a) might have possibly intensified the transport of benthic diatoms from the shallow coastal area into the hemipelagic CBeu trap via the MC (Fig. 1). Enhanced lateral 668 transport has important environmental implications for the final burial of organic matter in EBUEs.
As the organic matter can be effectively displaced from the area of production (Inthorn et al., 670 2006), carbon depocenters generally occur at the continental slopes between 500 and 2,000 m. In the CC-EBUE around Cape Blanc, the depocenter with up to 3% of organic carbon has a depth 672 range between 1,000 and 2,000m (Fischer et al., 2019).

Most of the populations affected by and responding to shifting environmental conditions off 674
Mauritania between 2004 and 2006 returned to their 'regular' seasonal pattern of occurrence after 2006 (Fig. 4). However, some shifts persisted still after summer 2006. Limacina bulimoides still 676 dominated the pteropod assemblage (Fig. 7h), the total pteropod flux showed the highest maxima for the entire studied interval (might be due to the large food supply and organic matter as 678 represented by high total fluxes of diatoms, Fig. 3a, e), and warm-water calcareous dinoflagellate cyst increased during late fall 2006 (Fig. 7j). An exception to this pattern is the high relative 680 contribution of benthic diatoms (Figs. 4a, 7i;Romero and Fischer, 2017). At this stage, we cannot fully disregarded that the shift in the species-specific composition of the diatom community (also 682 present after 2008; Romero and Fischer, 2017;Romero, unpublished observations) might be due to the natural long-term variability due to external forcings (e.g., North Atlantic Oscillation) or due to 684 on-going climate change.
Our multiannual trap experiment provides a unique opportunity to study the long-term evolution EBUE (Fischer et al., 2016(Fischer et al., , 2019Romero et al., 2002Romero et al., , 2016Romero et al., , 2017  MARUM-deliver a broad observational basis on the occurrence of persistent seasonal pattern as well long-lasting variations of microorganisms changes in response to key forcings, such as 724 nutrient input, water masses variability, lateral transport and/or climate change. The applicability of the flux dynamics of primary and secondary producers here presented is not limited to the 726 Mauritanian upwelling system, and it might comparable to other EBUEs.

Author Contributions 732
All authors collected the data. Oscar E. Romero wrote the manuscript. All authors contributed to results interpretation and discussion. 734

Competing Interests 736
The authors declare that they have no conflict of interest.  Res., 54, 73-98, 1996.  CaCO 3 =calcium carbonate; mixed layer depth; SST=sea surface temperature; SSTA=sea surface temperature anomalies. For interpretation of the references to color in this figure  1158 legend, the reader is referred to the web version of this article. 1160  (calcareous dinoflagellate cysts): -cosm = cosmopolitan group, -min = terrestrial mineral group, -other = species that do not fit in one of the other ecological groups, -upw = upwelling, warm: 1168 warm waters; OD (organic-walled dinoflagellate cysts): -cosm = cosmopolitan group, -other = species that do not fit in one of the other ecological groups, -tox = potential toxic group, -upw = 1170 upwelling, -upw relax = upwelling relaxation; For (foraminifera): -cold: cold water group, -upw = upwelling group; -warm = warm water group; and Pt (pteropods): -Hinf = Heliconoides inflatus, -1172 Lbul = Limacina bulimoides, -uncoi: uncoiled. Groups of microorganisms are identified by colors (light blue, group 1; brown, group 2; black, group 3; and red, group 4). The species-specific 1174 composition of groups is presented in Table 3. For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.