Checklist, assemblage composition, and biogeographic assessment of Recent benthic foraminifera (Protista, Rhizaria) from São Vincente, Cape Verdes

We describe for the first time subtropical intertidal foraminiferal assemblages from beach sands on São Vincente, Cape Verdes. Sixty-five benthic foraminiferal species were recognised, representing 47 genera, 31 families, and 8 superfamilies. Endemic species were not recognised. The new checklist largely extends an earlier record of nine benthic foraminiferal species from fossil carbonate sands on the island. Bolivina striatula , Rosalina vilardeboana and Millettiana milletti dominated the living (rose Bengal stained) fauna, while Elphidium crispum , Amphistegina gibbosa , Quinqueloculina seminulum , Ammonia tepida , Triloculina rotunda and Glabratella patelliformis dominated the dead assemblages. The living fauna lacks species typical for coarse-grained substrates. Instead, there were species that had a planktonic stage in their life cycle. The living fauna therefore received a substantial contribution of floating species and propagules that may have endured a long transport by surface ocean currents. The dead assemblages largely differed from the living fauna and contained redeposited tests deriving from a rhodolith-mollusc carbonate facies at <20 m water depth. A comparison of the Recent foraminiferal inventory with other areas identified the Caribbean and Mediterranean as the most likely source regions. They have also been constrained as origin points for littoral to subtidal macroorganisms on other Cape Verdean islands. Micro-and macrofaunal evidences assigned the Cape Verde Current and North Equatorial Current as the main trajectories for faunal immigrations. The contribution from the NW African coast was rather low, a pattern that cannot be explained by the currently available information. The suprageneric foraminiferal classification as proposed by Pawlowski et al . (2013) and Holzmann & Pawlowski (2017) was applied in the present study. If a certain genus or a family was not rooted in the tree of gene sequences by the authors, it was affiliated to the respective family or superfamily after the classification scheme of Loeblich and Tappan (1988). Species are listed in alphabetical order under the respective genera. The type reference, as retrieved from the Ellis and Messina (1940) catalogue, and at least one reference to a high-quality image in a recent publication is given for each species. References to the specimens images in this paper are given in square brackets after the type reference.


Introduction
The species inventory of oceanic islands, their evolution and their connectivity has been the subject of scientific research for decades (McArthur & Wilson 1967, and references therein). Earlier studies focused on immigration, evolution, and extinction controlling the species richness. Later studies investigated the role of isolation, size and age of the islands in balancing the ratio of endemic and immigrated taxa (e.g. Whittaker & Fernándes-Palacios 2007;Whittaker et al. 2008). Shallow-water organisms living on and around the islands came into focus of recent investigations (e.g. Hachich et al. 2015;Cunha et al. 2017;Pinheiro et al. 2017). They suggested that gene flow, i.e. connectivity, and island age were important for the biogeographical distribution of species. Furthermore, Pleistocene sea-level changes were considered a critical variable in the evolution and dispersal of marine organisms on and between oceanic islands (Ávila et al. 2018).
Benthic foraminifera are millimetre to sub-millimetre sized, shell-bearing unicellular organisms living in all marine environments. Foraminifera are well adapted to their habitat, have short generation times, and their tests are readily preserved in the sediment after reproduction. Benthic foraminifera are therefore considered as sensitive indicators for the prevailing environmental conditions for today and the geological past (Murray 2006 and references therein). Global warming and sea-level rise is expected to profoundly change the assemblage structure of intertidal and near-shore foraminiferal communities in the near future (Schmidt et al. 2011;Weinmann & Goldstein 2016;Müller-Navarra et al. 2017). In order to constrain their resilience, to assess and monitor the impact of Global Change on near-coastal ecosystems in the future, baseline studies

Material and methods
Surface sediment samples were taken by opportunity on an excursion to São Vicente during the HOSST Cape Verde Summer School at Mindelo in May and June 2018. The sample locations were chosen according to accessibility and wetness of the sediment surface. In particular : a shallow lagoon behind the breakwater wall at Baia das Gatas, the landward slope of the beach bar in a runnel at Sao Pedro, and the top of the swash zone on the beach of Calhau were sampled. Sampling was done close to the actual water level. The uppermost 1-2 cm of the surface sediment was scraped off with a spoon, transferred into 100 ml PVC (Kautex ® ) bottles, and immediately preserved and stained with a solution of 2 g rose Bengal in 1 l alcohol (40 %, vodka) (Lutze & Altenbach 1991;Schönfeld 2012: there p. 58). The preservative was exchanged two weeks later in order to rise the ethanol concentration to >90 % for long-term storage . Global Positioning System (GPS) coordinates and sampling time were retrieved from a mobile phone and noted on the vials (Appendix Table 1). The hights of the samples were obtained from the tide curve for Porto Grande at Mindelo (16.8667°N, 24.9833°W), as calculated by the web application of Xtide for the respective sampling day (http://tides.mobilegeographics.com/locations/5064.html) following Rashid (2015). They were consequently referred to the mean tidal level (MTL) at Mindelo.
The foraminiferal samples were prepared and analysed following the procedures described by Schönfeld et al. (2013), and Lübbers & Schönfeld (2018). The volume of the samples ranged from 54 to 77 cm 3 . The residues 63-2000 μm were split using a HAVER RT 6,5 sample splitter to facilitate microscopic work. One sixteenth was wet picked for rose Bengal stained foraminifera that were alive at the time of sampling. This split was dried and further subdivided with an Otto microsplitter to an aliquot that contained approximately 100-300 unstained (dead) foraminiferal tests. These subsamples were dry picked. The specimens from the living fauna and dead assemblages were sorted by species in separate Plummer cell slides, fixed with glue, and counted. Once rose Bengal stained specimens were not found in the 1/16 split, the remaining 15/16 were treated with a sodium polytungstate solution of 2.303 g cm -3 density to recover all foraminifera from the sample (Parent et al. 2018). The float ate and deposited parts of the sample were washed with tap water to remove remaining polytungstate and dried at 50 °C. The floatate was carefully examined for living foraminifera and added to the deposited part of the sample again after picking.
Benthic foraminiferal morphotypes may include different genotypes (OTUs) that are depicted by subtle morphological characters (e.g. Darling et al., 2016). Once these characters are disclosed, they may be used to discriminate morphotypes representing a single OTU Kucera et al. 2017;Richirt et al. 2019). We therefore attempted to discern the species up to a comprehensible level, in particular miliolids that were previously lumped as "Quinqueloculina spp.". Species were determined by using both taxonomic literature and online data bases (Ellis and Messina 1940;WoRMS Editorial Board 2018). Literature from all ocean basins was considered, since organisms in a submillimeter size range tend to have a more cosmopolitan rather than endemic distribution (e.g. Fenchel & Finlay, 2004). Preference was given to literature providing high-quality images and accurate taxonomic references for the respective taxon. Once the holotype figure was ambiguous, we gathered all accessible information. For instance, we approached the Smithonian collections at Washington DC, USA, for supplementary type images, and examined reference slides with living foraminifera (rose Bengal stained, plankton catchments) from well investigated areas (southern Portugal, Gulf of Cadiz, Puerto Rico) (Schönfeld 1997(Schönfeld , 2002Kucera et al. 2017). In the literature, foraminiferal species were often reported with several different genus names. We have chosen that genus which has not been amended later, which was in best agreement with the morphological characters of the species, or which has been validated by recent genetic investigations. A genus is regarded as monophyletic group comprising at least two different species. The use of monospecific genera (e.g. Lobatula) was therefore avoided in the present study. Images were taken with a Keyence VHX-700 FD digital microscope at the Institute of Geosciences, Chris-tian-Albrechts University Kiel, and with a Zeiss LEO 1455VP Scanning Electron Microscope (SEM) at the Institute for Geology, University of Hamburg. Due to the explorative character of the present study, we documented as many species as possible.
Regional species occurrences were compiled by also considering literature without images but with taxonomic references, provided the species names' lists were given with the author and publication year. Google ® , Bing ® and Mac OS X Spotlight search functions were used to retrieve notes on the occurrence of each species.
Grainsize distribution and coarse fraction composition were analysed by using remaining aliquots from the sample residues. They were dry sieved with ISO 3310-1 certified sieves with a mesh sizes of 125,150,250,355,500, and 1000 μm. The size fraction >2000 μm was considered as well. Coarse fraction analysis was performed by applying the protocol of Sarnthein (1971). The subfractions were weighed and further subdivided with an Otto microsplitter. Between 103 and 265 grains were identified and counted under a stereo-microscope from each subfraction. Only three major groups of components were considered, i.e. volcanic grains, biodetritic grains and foraminifera (e.g. Wolf & Thiede 1991). Therefore, the accuracy of the census is deemed acceptable (van der Plas & Tobi 1965Fatela & Taborda 2002. The composition of the sand-sized fraction >63 μm was calculated by combining the proportion of individual components with the weight of the subfractions examined. The weighed proportions were added for each component, and percentages were calculated in reference to the sum of all subfraction weights (Sarnthein 1971).

Composition of the coarse-grained fraction.
The grain-size distributions revealed that the deposits were pure medium to coarse sands. The grains were well rounded, the surfaces were smooth, and the carbonate grains appeared as having been polished. Some of these carbonate grains showed a layered and punctuated internal structure of translucent and opaque material resembling the reticulate structure of coralline red algae. Spines of regular echinoids, coral fragments, gastropod shells and fragments, bivalve shells and debris, fragments of balanid plates, bryozoans and benthic foraminifera could be identified in the size fractions >500 μm. Judging only from the aspect of the grains' surface, rhodoliths could not be distinguished with certainty from strongly bioeroded or encrusted shell debris of molluscs. The biogenic components were therefore summarized as undifferentiated bioclasts. Their proportion ranged from 78-92 %. Benthic foraminifera were rather rare with 1-3 % (Appendix Table 1). The foraminiferal tests were mostly well or moderately preserved. Only a few species showed a moderately to poor preservation. SEM images revealed that in particular tests of Amphistegina gibbosa have been bioeroded by filamentous microborings (e.g. Young & Nelson, 1988;Günther, 1990).
The terrigenous fraction comprised basalts, single mafic minerals, altered or silicified pumice of light green and brown colour, dark volcanic glass particles, and ash charts. Light coloured charts were common at Calhau only. They may derive from the young strombolian volcanic cones at Calhau and Bahia das Gatas (Ancochea et al. 2010;Ricardo Ramalho, pers. comm. 6 th December 2018). The proportion of the volcanic particles ranged from 7-19 % of the coarse fraction (Appendix Table 1). These figures are in good agreement with point counting data on thin sections of Pleistocene dune sands on other Cape Verdian islands (Johnson et al. 2013).
Foraminiferal check lists. Sixty-five benthic foraminiferal species were recorded in the present study, of which 42 species were recorded in the living fauna and 52 species in the dead assemblages (Appendix Table 2). They belonged to 47 genera, 31 families and 8 superfamilies.
The genus Quinqueloculina and Bolivina, with 10 and 5 different species, was rather diverse. Elphidium showed 3 species, while Lepidodeuterammina showed 2 and Peneroplis showed 3 species. All other genera were represented by a single species. Endemic species were not recognised. These figures rejected our initial hypothesis; a distance of >500 km between the Cape Verdes and Western Africa is not an effective zoogeographic barrier for near-shore benthic foraminifera.
Fifty-seven fossil benthic foraminiferal species were recognised on the Cape Verdian islands (Torres & Soares 1946). Ten of these species were extinct (Appendix Table 3). From the nine species recorded on São Vincente by the authors, only Elphidium cripsum was found in the present study.

Family Cassidulinidae d'Orbigny 1839b
Genus  (Cushman 1918), which is very similar in shape and chamber arrangement. The last 4 to 5 chambers of the latter species extend into triangular, axe-shaped flabs on the umbilical side, each off set from the previous, and thereby forming a rosette collar around the umbilicus. This feature is poorly developed in the present species. The flabs are only recognisable at very high microscope magnifications. Consequently, umbilical flabs neither have been mentioned nor drawn by d' Orbigny (1839). The specimens figured by Brady (1884) also show no flabs, thus mirroring the 19 th century view of this species. Furthermore, d'Orbigny's Rosalina bertheloti is much more compressed than the holotype of Hanzawaia concentrica (Cushman 1918 fig. 9. Loeblich & Tappan (1994), p. 153, pl. 329, figs. 1-12. Note: only microspheric specimens without a floating chamber were found in the present study. Adult specimens show a beehive-shaped outline in lateral view, and the umbilicus is surrounded by four inflated chambers. The species differs from Cymbaloporetta squammosa d'Orbigny 1839b by the distinct sutures that are hardly visible in the latter, and by the umbilicus, which is not covered by an umbilical disc as in Cymbaloporetta squammosa (e.g., Rückert-Hilbig 1983). Such an umbilical disc has not been seen in our specimens from São Vincente.

Foraminiferal assemblages
Living (rose Bengal-stained) benthic foraminifera were common in the sample from Baia das Gatas. The samples from Calhau and São Pedro were barren of living foraminifera. The population density at Baia das Gatas, with 478 individuals per 10 cm 3 , was well in the range of standing stocks on sand substrates in the lower intertidal zone under warm climatic conditions (ca. 100-1000 individuals per 10 cm 3 ; e.g. Culver & Horton 2005). The abundance of empty tests of the dead assemblage varied from 29 to 298 specimens per gram of sediment (Appendix Table 2). At Baia das Gatas, their concentration per volume, i.e. 2160 specimens per 10 cm 3 , outnumbered the concentration of living specimens by a factor of 4.5, which is in reasonable agreement with literature data (1:4 to 1:6, Scott & Medioli 1980). Bolivina striatula, Rosalina vilardeboana and Millettiana milletti dominated the living fauna with 24, 15 and 13 % respectively. Ammonia tepida, Quinqueloculina bosciana and Quinqueloculina laevigata were common with 4-3 %. The remaining 36 species of the living fauna were rather rare with <3 %. Only 22 of 42 living species were also recorded in the dead assemblage.
Quinqueloculina seminulum, Ammonia tepida, Triloculina rotunda and Glabratella patelliformis dominated the dead assemblage at Baia das Gatas with 13-9 %. Quinqueloculina lamarckiana, Millettiana milletti, Peneroplis carinatus and Quinqueloculina lata were common with 5-3 %, and the remaining 42 species were rare with <3 %. It has to be noted that only two of the eight frequent or common species from the dead assemblage were also frequent in the living fauna.
The Fisher alpha diversity index of the living fauna and dead assemblage at Baia das Gatas of 15.06 and 16.96 respectively was very similar, depicting rich and highly diverse foraminiferal assemblages at this site. The Fisher alpha index was substantially lower at São Pedro and Calhau with 1.75 and 1.95 (Appendix Table 2).

Composition of the living foraminiferal fauna.
Beach sands in subtropical environments have been considered as barren of living benthic foraminifera (Pamela Hallock, St. Petersburg, USA, pers. comm.), which might explain why we did not record living foraminifera at Sao Pedro and Calhau. However, early studies on the microhabitat depth of near-shore foraminifera in intertidal sands demonstrated that they live deeper in the sediment than usually sampled (Richter 1964a,b;Giese 1991;Langer et al. 1989). Own observations from Esteiro Ancão backbarrier sands, Ria Formosa, Portugal, at warm climatic conditions and high salinities, revealed that living benthic foraminifera were with 1-4 individuals per 10 cm 3 indeed very rare in the uppermost 1-2 cm of beach sands above MTL. It is therefore conceivable that the abundance maximum of living foraminifera was encountered at Baia das Gatas but missed at the other sampling locations.
The living fauna at Baia das Gatas did not resemble littoral faunas at temperate to warm climatic conditions in the Atlantic. Typical species for coarse-grained substrates, for instance Asterigerinata mamilla (Williamson 1858), Haynesina depressula or Cibicides lobatulus, were missing. Arenaceous species were underrepresented when compared to West African sites (e.g. ). On the contrary, such a high proportion as 28 % Bolivina spp. was not observed in intertidal settings elsewhere in the tropics or subtropics (e.g. Laut et al. 2017). Only the proportion of Ammonia tepida and Quinqueloculina spp. was in the expected range. Therefore, the fauna may be considered as being incomplete.
The life cycle and ecology of frequent and some rare species deserves attention in this respect. Millettiana milletti matures to a planktonic stage before reproduction (Rückert-Hilbig, 1983). However, specimens with a globular floating chamber were not observed in the living fauna from Baia das Gatas. The tychopelagic life mode of Bolivina variabilis is well constrained (Darling et al. 2009). However, other Bolivina spp. tend to float under environmental disturbance as well (Kucera et al. 2017;Glock et al. 2019). They were widely dispersed off NW Africa (Lutze 1980). Trifarina bella was recorded in plankton catchments south of Puerto Rico during certain days at high num-CHECKLIST OF BENTHIC FORAMINIFERA SPEICES Zootaxa 4731 (2) © 2020 Magnolia Press · 165 bers (Schmucker 2000;Anna Jentzen, Kiel, and own unpubl. data). Even though recent studies revealed that foraminiferal propagules are of local or regional rather than of remote origin (Weinmann & Goldstein 2017;Weinmann et al. 2019), it is reasonable to assume that either floating propagules or the capability of a transient planktonic lifestyle were important for sustaining the benthic foraminiferal population at São Vincente.
Generation of the dead assemblages. The living fauna forms the dead assemblage over the course of many generations (Murray 1982). Species proportions and abundances may differ from the living fauna due to bias inferred by dissolution, bioerosion and lateral transport (e.g. Murray et al. 1982;Schröder 1988;Berkely et al. 2009). The living fauna may be subjected to a strong seasonal and interannual variability in species proportions and population densities (e.g. Murray & Alve 2000;Bouchet et al. 2007). None-the-less, common and frequent taxa of both, living fauna and dead assemblage should be the same.
At Baia das Gatas, only Millettiana milletti and Ammonia tepida are common or frequent in the living fauna and dead assemblage. However, about half of the living species were found in the dead assemblage as well. The missing species do not have delicate, fragile tests that easily could be destroyed (Kotler et al. 1982). Population densities in high-energy, near-shore sands are usually low because most species can not stand the permanent redeposition of their substrate (e.g. Langer et al. 1989;Humphreys et al. 2018). Only a few attached species may persevere. Such specialists, e.g. Trochamminidae, were rare in both assemblages at Baia das Gatas. A possible seasonal variability of the living fauna at Baia das Gatas is difficult to constrain since. The seasonality on São Vincente is characterised by changes in trade wind intensity. During sampling in late May and early June, the trades were on the decline but still strong. On the other hand, the difference in dead foraminiferal assemblage composition between samples from the windward locations Baia das Gatas and Calhau was higher than between the latter and the leeward sample from São Pedro.
The most plausible explanation for the difference between living and dead assemblages is that the dead assemblage received a substantial amount of redeposited tests. For instance, the frequent species Elphidium crispum is a symbiont-bearing, epifaunal species living in shallow, subtidal waters in temperate to tropical environments (Leutenegger 1984;Lee & Lee 1989;Parker & Gischler 2015). Peneroplis proteus is frequent in tropical shallow waters where moderate redeposition prevails (Wilson & Wilson 2011). Amphistegina gibbosa prefers coral reef rubble as substrate (Eichler et al. 2019, and references therein). Eponides repandus is adapted to high energy environments as well and prefers coarse sands and gravels (Edwards 1982). Empty tests of these species were probably taken up by wave turbulences in subtidal, shallow waters and were transported on the shore by surf action. The similarity of the dead assemblages from all three sampling sites, and the good agreement of the coarse fraction composition with literature data from fossil sands in other areas, suggest that this redepositional process is extensively prevailing around the Cape Verdian islands, in that beach sands are augmented by shallow water rhodolith-mollusc carbonate production sites.
Comparison of checklists. In fossil carbonate sands from São Vincente, nine benthic foraminiferal species were recognised (Appendix Table 3; Torres & Soares 1946). Two of them date back to the late Pliocene (Faujasina carinata) and Bajocian (Epistomina regularis), whereas the maximum depositional age of these sands is late Pleistocene (0.33 million years; Ramaldo et al. 2010). The identification of foraminiferal species by thin section analyses has been proven suitable for investigations of the internal structures of large, shallow-water benthic foraminifera, as well as for assessments of the planktonic foraminiferal inventory of limestones from the late Paleozoic to Neogene (e.g. Wernli et al. 1997;BouDagher-Fadel 2008;Asis et al. 2018). The determination of modern foraminifera by thin section examination is difficult however, in particular once test shape, apertural characteristics, and surface structures are diagnostic features. Misidentifications are likely. Therefore, the comparison of the 1946 checklist with data of the present study is not a straightforward way to accomplish the Recent benthic foraminiferal inventory of Cape Verdian islands.
Source regions and trajectories. The Cape Verdian archipelago is separated from the African continent by more than 500 km of a 3000 m deep ocean. As endemic species were not recognised, any Recent littoral or near-shore foraminifera must have been transported to the islands. A long-range transport of propagules by ocean currents or adult individuals by transocean rafting, a medium-range displacement by ichtyochory, and the introduction of alien species by marine traffic or migratory birds have been invoked as dispersal mechanisms for benthic foraminifera (McGann et al. 2000;Alve & Goldstein 2003;Riedel et al. 2011;Polovodova Asteman & Schönfeld 2016;Guy-Haim et al. 2017;Finger 2018). Indeed, source regions of the displaced species have to be identified first before migration routes are delineated and transport mechanisms are constrained (Lübbers & Schönfeld 2018).
We explored data from Bahia Reefs, Brazil, the Caribbean region, Bermuda, the Algarve coast of the Gulf of Cadiz, the Mediterranean, Gran Canaria and, more importantly, West Africa for benthic foraminiferal species co-occurring with São Vincente (Appendix Table 2, and references therein). The numbers of foraminiferal species in common with the potential source regions ranged from 28-59 %. Between 2-8 % of the species encountered on São Vincente were assigned to a single source region. None-the-less, 11 % of the species were not recorded in the tropical to temperate northern Atlantic to date and may derive from remote areas, e.g. eastern Pacific or Indian Ocean. The highest proportion of >50 % co-occurrences as well as the highest single-source region matches were with the Mediterranean and Caribbean regions (Figure 2). The adjacent NW African margin showed a markedly lower similarity with 44 % co-occurrences.
These far-reaching relationships are corroborated by data on shallow-water macrofauna from Cape Verdean islands. The barnacle Chthamalus stellatus (Poli 1791) dominated the epizoan fauna on intertidal rocky shores. This species is common in the Mediterranean, western Europe, and Atlantic Islands, but absent from West Africa. The Cape Verdean corals and algae came from the Caribbean and eastern America (Morri et al. 2000). Three of 13 new records of marine invertebrates on the Cape Verde an islands derive from the Caribbean and Mediterranean, and only one species came from West Africa (Wirtz 2009). From nine keyhole limpet species of the genus Diodora, five derive from the western Atlantic and one from the Mediterranean, whereas two came from the Pacific and one is endemic (Cunha et al. 2017). All authors emphasized the role of ocean currents as dispersal vectors, and the latter also considered the durability and resilience of larval stages as being crucial for their survival during long-distance transport. Evidences for invasions of alien species transported by marine traffic or fishing gear were not reported to date. São Vincente is bathed by surface currents coming from the Canary Islands or Brazil, depending on the seasonal position of the CVFZ (Fig. 1). It is conceivable that these currents have brought propagules or floating specimens of meroplanktonic foraminifera to the Cape Verdes. The composition of the living fauna at Baia das Gatas suggests that the latter are of particular importance as contributors. It remains enigmatic, however, why the relationship to West Africa is so sparsely developed, even though the CVC follows the African shelf break over a long distance (Fig. 1). One may speculate that coastal upwelling creates an effective border that inhibits the off-shore proliferation of propagules of near-shore foraminifers in this area.  Table 2 for references.

Conclusions
Sandgrain composition and dead benthic foraminiferal assemblages corroborated earlier evidences from petrographic studies that the beach sands from São Vincente had derived from a subtidal rhodolith-mollusc carbonate facies off shore. The maximum depth can be constrained to approximately 20 m because coralline algae were absent below (Morri et al. 2000). The preferred depth range of Amphistegina gibbosa is 14-40 m due to the specific light demands of their symbionts, even though the species may be found as shallow as 10 m on patch reefs, and as deep as 100 m on carbonate shelf sediments (e.g. Hallock et al. 1986;Martin & Liddell 1988;Hallock 1999). The sandgrain composition indicated that the sublittoral carbonate environments were highly detritus productive in comparison to the extensive coastal erosion of volcanic rocks at the cliffs around the island. The good agreement in the composition of bioclasts and benthic foraminiferal constituents in Recent beach deposits and fossil eolian sands revealed that the near-shore environmental conditions at São Vincente have not substantially changed since at least 330 thousand years.
The comparison of benthic foraminiferal checklists from Cape Verdian fossil sands and Recent dead foraminiferal assemblages from São Vincente revealed several misidentifications of foraminiferal species in thin sections. The island had not yet emerged from the sea at times when the respective species had lived. Even though thin section analyses provide reliable data on larger shallow-water foraminifera from Carboniferous to Miocene limestones, in particular once the internal structures are diagnostic, the technique should not be applied to Pleistocene and Recent sediments.
The living fauna at Baia das Gatas was missing typical species for coarse-grained substrates under warm climatic conditions. The proportion of arenaceous species was also rather low. Instead, species that either had a planktonic stage in their life cycle (Millettiana milletti), were tychopelagic (Bolivina variabilis), or had the ability to float (Trifarina bella, Bolivina spp.) were common. The living fauna was therefore augmented by a substantial contribution of floating species and propagules that may stand a long, transocean transport by surface ocean currents.
On the remote Marshall and Mariana Islands in the Pacific Ocean, 84 % of Recent foraminiferal species were common on other islands as well, 6 % were endemic, and 10 % were of unknown origin, i.e. derived from remote areas (Cushman et al. 1954;Todd 1966). These figures are in good agreement with our data from São Vincente where 11 % of the species came from remote areas. Todd (1957) speculated that reef-dwelling foraminifera probably have a planktonic stage facilitating their dissemination by ocean currents. Even though propagules of reef-dwelling fora-minifera have not been caught neither individuals were raised from open ocean surface water samples, our results from Cape Verde corroborate the conclusions of these early, complementary studies from the Pacific.
A comparison with checklists from other regions in the tropical to temperate North Atlantic depicted the Caribbean and Mediterranean as probable source regions. The same processes and source regions have been described for littoral to subtidal macroorganisms on other Cape Verdian islands. The combined micro-and macrofaunal evidences identified both CVC and NEC as main trajectories for faunal immigrations. However, t he contribution from the NW African coast was considered to be rather low. The available information does not offer a plausible explanation for this. Perhaps coastal upwelling or other oceanographic processes on the West African shelf create effective barriers that inhibit an off-shore proliferation of foraminiferal propagules.