Palaeogene triserial planktonic foraminifera

Triserial planktonic foraminiferal species from the Palaeogene are classified into the Early Danian Guembelitria, with pore mounds on the test surface, and the Early Eocene-Late Oligocene Jenkinsina, which are also microperforate but lack pore mounds. The stratigraphic and palaeogeographic ranges are discussed, the holotype of G. cretacea Cushman (the type species of Guembelitria) is re-illustrated, and neotypes designated for G. danica (Hofker) and Jenkinsina triseriata (Terquem). Chiloguembelitria is shown to be a junior synonym of Guembelitria.


INTRODUCTION
Planktonic foraminifera with tests that are triserial throughout their ontogeny have been classified under the Suborder Globigerinina and Superfamily Heterohelicacea by Loeblich & Tappan (1987). In the Palaeogene there are only a few species: two or three in the Early Danian with strong connections with Late Cretaceous taxa, and three in the Eocene-Oligocene. The paucity of species may be due in part to their small size (100-200 microns) and they may have been missed because the fine sediment fraction is rarely examined in detail by micropalaeontologists.
All the taxa have microperforate test walls and species of Guembelitria (Cretaceous-Early Danian) and its junior synonym Chiloguembelitria (Early Danian) can be clearly distinguished from those of Jenkinsina (Early Eocene-Late Oligocene) by the possession of pore-mounds on their test wall surfaces; this is contrary to the opinion of Loeblich & Tappan (1987), who synonymized Jenkinsina with Guembelitria.
Guembrlitria appears to be cosmopolitan but some of the species of Jenkinsina are more restricted palaeogeographically. For example, J. samwelli is restricted to the latest Eocene-Late Oligocene of the Southern Hemisphere.
In order to facilitate better identification of the species, the holotype of G. cretacea Cushman (the type species of Guembelitria) is re-illustrated, and neotypes are designated for C. danica Hofker (the type species of Chiloguembelitria) and Textilaria triseriata Terquem (the senior synonym of Guembelitria stavensis Bandy, the type species of Jenkinsina); there are no type specimens of C. danica and the remaining syntype of T. triseriata is badly damaged. A case has been submitted to the ICZN for this specimen of T. triseriata to be set aside in favour of our neotype. A neotype is also selected for C. danica.
Gallitellia vivans (Cushman, 1934) is the only living triserial microperforate planktonic species and, according to Loeblich & Tappan (1987) it ranges back into the Pleistocene. They state that the main distinguishing features of Gallitellia are the very thin, hyaline, smooth and sparsely microperforate wall, simple aperture with no bordering lip, and occasional chamber proliferation in the adult test; these features distinguish it from both Guemhelitria and Jenkinsina. According to Kroon & Nederbragt (1990), G. vivans also lacks a tooth-plate. So far, this is the only record of a Neogene wholly triserial microperforate species. As a working hypothesis, iterative evolution has probably resulted in a number of triserial taxa in the Cenozoic.

GENERA AND WALL STRUCTURE
It has been the policy of phylogenetically minded micropalaeontologists to separate genetically unrelated groups of taxa with similar tests into distinct genera (Jenkins, 1965;McGowran, 1968;Steineck & Fleisher, 1978). A case in point is the Palaeocene-Eocene genus Morozovella, erected by McGowran (1 968) for keeled species unrelated to the Neogene Globorotalia. The advent of the Scanning Electron Microscope has also made it possible to base genera on fine wall structure differences. Thus,  were able to separate the Palaeocene cancellate spinose genera Eoglobigerina and Parasubbotina from the cancellate non-spinose Praemurica. Li Qianyu et a/. (1995) dispute this because. . . 'the pore-mound structure has been observed independently on several forms which are phylogenetically unrelated', but following our study we are firmly of the opinion that the presence of pore mounds in the microperforate triserial Danian Guembelitria can be used to distinguish it from the Early Eocene-Oligocene microperforate triserial Jenkinsina.

PALAEOGENE TRISERIAL GENERA Guembelitria
It has been established for some time that Guemhelitria cretacea Cushman had a microperforate wall with pores situated centrally within pore mounds. Thus Olsson (1970) illustrated G. cretacea from the Palaeocene Redbank Formation of New Jersey with 'typical wall structure and pore location within blunt spines'. Smith & Pessagno (1973) reported that G. cretacea from the Late Cretaceous Corsicana Formation of Texas had a 'wall finely perforate, each pore surrounded by a blunt cone of microgranular calcite'. More recently, Loeblich & Tappan (1987) stated that the test walls of Guembelitria are.. . 'finely and irregularly perforate, with each pore elevated on a pore mound' and they illustrate these features in topotype specimens of G. cretacea Cushman, 1933 from the Maastrichtian of Texas.
Loeblich & Tappan give the range of Guemhelitria as 'L.
-tGraham Jenkins died suddenly on 6th August 1995, shortly after completing this manuscript. His co-authors wish to dedicate the paper to his memory and in honour of his outstanding contributions in the field of planktonic foraminifera1 research.  (1985). J . samwelli ranges from the late Eocene to late Oligocene in South Australia, but has a very short range within the late Oligocene in the SW Pacific. The range of G. cretacea is based on  Cretaceous (Albian) to U. Cretaceous (Maastrichtian)', but this is no longer true. Keller (1988) showed that G. cretacea ranged into the Early Danian P l a Zone in the El Kef Section of Tunisia while Stott & Kennett (1990)

sergipensis.
We have obtained, on loan from the U.S. National Museum of Natural History, Washington, D.C., the holotype and paratype of Guembelitria cretacea Cushman (USNM nos. 19022 and 21217, respectively). For the first time, these specimens have been photographed by scanning electron microscopy and this has been achieved, without coating, using the Environmental Chamber technique described by Taylor (1986). The pore mounds can clearly be seen (pl. I , figs 8, 9).
In order to verify the morphology of C . danica a search was undertaken in the Netherlands for the type specimens. The writers are grateful to Drs J. C. den Hartog (National Museum of Natural History, Leiden) and R. J. W. van Leeuwen (Geological Survey of the Netherlands, Haarlem) who looked for them. In a letter dated 23/9/1990, Dr van Leeuwen wrote.. . 'To my regret, I have to tell you that Hofker has not deposited a type specimen of Chiloguembelitria danica in either Haarlem (Rijks Geologische Dienst) or Leiden (Rijksmuseum). I cannot think of another place in the Netherlands where he could have deposited a type, and I feel pretty sure that there is no such type'. We have also ascertained that Hofker deposited his specimens neither in the DSDP Depository at the Natural History Museum in Base1 nor in the Smithsonian Institution, Washington, D.C.
Dr van Leeuwen was able to send us Hofker's original type sample, however, and from it we have picked out several specimens which resemble his illustrated specimens of C. danica. The test surfaces of some of the specimens exhibit diagenetic overgrowth and no doubt, because of this, Loeblich & Tappan were misled. Better preserved specimens from the same sample, however, show clearly that the pustules on the test wall recorded by Hofker are indeed pore mounds (pl. 1, fig. 7). There is now no other consistent character by which Chiloguernbelitria can be distinguished from Guembelitria. All other criteria, particularly coiling and apertural characteristics, mentioned by Hofker in his original diagnosis, are also found in Guembelitria. Chiloguembelitria therefore becomes a junior synonym of Guernbelitria; the three species, G. cretacea, G. danica and G. irregularis Morozova, form a related group in the Early Danian. Their stratigraphic ranges were recorded by Hofker (1978), and by Keller (1988) from the El Kef Section. Because the type specimens cannot be found and because of the misunderstanding about the nature of the test surface, a neotype is here formally designated for G. danica.
Guernbefitriu dunicu (Hofker, 1978) (PI. other specimens ranging in size between 130 and 170 pm have also been illustrated (pl. 1, figs 2-6). The distinctive feature of the species is the elongate, high trochospire (of the triserial test), the pronounced increase in the size of the chambers in the last whorl, and the characteristic torsion of the chambers as viewed from the basal end. Previously rather neglected, these small triserial plankton have recently received somewhat more attention. Kroon & Nederbragt (1990) suggested that both G. danica and G. irregularis were junior synonyms of G. cretacea but they provided no photographic evidence to support this view. They stated that.. . 'both have the same shape as G. cretucea. We assume that both forms possess pore mounds and we regard them as synonyms of G. cretacea'. Confronted with the same problem, D'Hondt (1991) concluded: 'However, type illustrations of C. danica differ from the type specimens of G. cretacea in possessing an asymmetrically developed chiloguembelinid aperture marked by an indistinct apertural flange. In part, these taxonomic problems can be resolved by reducing Chiloguembelitria to a junior synonym of Guernbelitria while provisionally retaining G. irregularis Morozova and G. danica (Hofker) as valid species'.
Most authors have continued to refer C. danica to Hofker's genus but Keller (1993) placed it in Guembelitria. Further, MacLeod (1993) reports that comparative morphological analysis of G. danica shows that, at both Nye Klsv, Denmark and ODP Site 738C (Kerguelen Plateau), the 'C. danica morphotype does exhibit well-developed pore mounds (see pl. 3, figs 1, 5) and an apertural margin that is not qualitatively different from species belonging to either Guembelitria, Woodringia, or Chiloguembelitria'. As there are therefore no consistent criteria by which the two can be differentiated, we support D'Hondt, Keller and MacLeod in synonymizing Chiloguernbelitria with Guembelitria.
The most detailed study of the comparative morphology of Guembelitria species is that of MacLeod (1993). Following cladistic analysis of 52 morphological characters representing all ontogenetic stages, he concluded that cretacea and danica were indeed distinct species. The alternative view, as discussed above, is that G. danica is a junior synonym of G. cretacea. A definitive answer is beyond the scope of this paper, whose aim is primarily to focus on generic criteria. We have, however, chosen and figured a neotype of G. danica (pl. 1, fig. 1 Keller (1993) gives a range from POa to P l c in the Early Danian of the El Kef Section of Tunisia. See also Figure 1 for ranges of G. danica and G. cretacea.
Jenkinsinu tviseviuta (Terquem, 1882) (PI.  (Terquem). In the Terquem Collection there were three specimens labelled Textilaria triseriata: (1) a small triserial calcareous specimen, possibly a Reussella with an incipient keel; (2) a similarly sized, poorly preserved specimen, also triserial but with rounded chambers embedded in and obscured by Canada Balsam, and (3) a specimen in a tube labelled 'FT34 3 15.10', which appears to be a Jenkinsina but with the last few chambers broken off. After a comparison with Terquem's (1882) figure (pl. 15, fig. lo), it can reasonably be concluded that the third is probably the original figured specimen; Le Calvez (1970) had already come to this conclusion, although to refer to this last specimen as the holotype, as she did, is not strictly correct. 10.11, figs 5, 6. 1.
type locality at Vaudancourt (Oise) is no longer exposed, but she was able to supply us with two small topotype samples from the Le Calvez Collection. Both samples yielded shallow water benthonic foraminifera and fragments of bivalves, gastropods and bryozoa, but unfortunately, no specimens of J. triseriata were found. Because of the damaged nature of the only syntype that can definitely be ascribed to Terquem's taxon (the last few chambers are missing and the aperture cannot be seen), a case has been submitted to the ICZN to set aside this specimen in favour of a proposed neotype (pl. 2, fig. 1) from a locality near to Vaudancourt and from the same stratigraphical level. Liancourt St. Pierre (Oise) is c. 11 km west of Vaudancourt and the samples have yielded abundant specimens of J. triseriata (Terquem).
Of the three other specimens illustrated from Liancourt St Pierre (pl. 2, figs 2 4 ) , two are sinistrally coiled and one is dextral; they range in size from 115 to 150pm. The wall structure of the neotype and two specimens figured in pl. 2, figs 2, 3 is masked by an overgrowth of calcite crystals (e.g. pl. 2, fig.  9) but in the smallest specimen (pl. 2, fig. 4) it is smooth, similar to a specimen from the Middle Eocene of Selsey, southern England (pl. 2, fig. 5), which also has pustules on the final whorl. A close-up of the wall, especially of the smooth specimens, shows the minute perforations (pl. 2, fig. 10). The apertural position in the two other Liancourt specimens (pl. 2, figs 2 and 3) is similar to the neotype, but in the smallest specimen it is much more umbilical in position (pl. 2, fig. 4); this could be an ontogenetic difference. All the specimens, in side view, show the whorls to have a slightly twisted appearance.
Bignot & Le Calvez (1969)  Comparison with other species. Guembelitria stavensis Bandy (1949), described from the Middle Eocene of Alabama, is here regarded as a junior synonym of J. triseriata; Le Calvez (1970) was also of the same opinion. The similarity of the tests can be confirmed by examining the illustrations of topotypic specimens of J. stavensis in Jenkins (1978~).
Guembelitria samwelli Jenkins (19786) was originally described from the Late Oligocene G. euapertura Zone of New Zealand and is widely distributed in the Oligocene of the Southern Hemisphere (Jenkins, 1993). This species is also a Jenkinsinu and differs from J. triscriata in normally having: (1) a smaller and more rounded aperture, positioned opposite the suture separating the penultimate chamber and the first chamber of the final whorl [Jenkins, 1978b andSrinivasan (1985) illustrated a specimen with an enlarged aperture]; ( 2 ) a less thickened apertural lip; (3) in not having the chambers slightly twisted about the coiling axis, and (4) in possessing the occasional supplementary aperture (see Jenkins, 1978b, pl. I, figs 1, 4 and 5).

16.
(Jenkins). Masters (1977), on the other hand, after examining the holotype of columbiana, concluded that it was identical to G. cretacea Cushman. Stratigraphic/palaeogeographic distribution. According to Prof. W. A. Berggren (pers. comm.), the type locality of J. columbiana (Cook Mountain Formation) is roughly equivalent to Zone NP16 (Middle Eocene). In Trinidad, Beckmann (1957) figs 1 and 2), which is equivalent to Zone NP15 (Curry et al., 1978); similar specimens occur at about the same stratigraphic level at Peyreblanque, Biarritz, S.W. France (pl. 3, fig. 3). The European stratigraphic range is equivalent only to about the upper part of its Trinidad range (Fig. 1) and in England J . triseriata  does not overlap the range of J . columbiana (NP15).

J. samwelli lived in the Austral Gulf between Australia and
Antarctica from the Late Eocene to early Late Oligocene. When the two continents finally became detached at the Tasman Rise, the water flowed out into the Southwest Pacific and also carried the plankton around Antarctica, and through the Drake Passage into the South Atlantic and South Indian Ocean. J. samwelli lived for only a brief period of time and its migration marks the beginning of the circum-Antarctic current at c. 30-31 Ma (Jenkins, 1974(Jenkins, , 1978a(Jenkins, , 19783, 1993Jenkins & Srinivasan, 1985). The species had a northern limit of latitude 35"s in the South Atlantic and latitude 40"s in the South Pacific. Olsson et al. (1992) derive a number of microperforate genera from Guembelitria in the Danian PO Zone, but only the trochospirally coiled Parvularugoglobigerina which ranged into the P2 Zone retained some pore mounds. The other genera Chiloguembelina, Woodringina and Globoconusa do not have pore mounds. Thus Guembelitria cretacea is the last triserial taxon which had the distinctive pore mounds, and from the published data and from our own research it would seem that pore mounds disappeared with the extinction of G. cretacea and P. eugubina.

PHYLOGENY
There is no direct phylogenetic connection between the triserial Danian Guembelitria and the Early Eocene Jenkinsina taxa. Consequently, we can postulate that the Jenkinsina species evolved directly from microperforate taxa which did not possess pore mounds. An ancestral candidate for Jenkinsina is the biserial genus Chiloguembelina and as can be seen from Fig. 1, the Jenkinsina triserial test may have evolved twice, once in the Early Eocene and then again in the Late Eocene. To check this hypothesis, however, more work is required into the late Middle to Late Eocene planktonic assemblages.
Our contention that pore mounds were finally lost in the early Palaeocene is reinforced by the work of Li & Radford (1994) who redescribed Cassigerinelloita amekiensis Stolk, an Early-Middle Eocene microperforate taxon with triserial-planispiral coiling and which they derive from J. triseriata; it does not possess pore mounds.

CONCLUSIONS
Palaeogene triserial planktonic foraminiferal species can be classified into Guembelitria (Early Danian) and Jenkinsina (Eocene-Oligocene); Guembelitria is distinguished from Jenkinsina in possessing pore mounds. On present evidence, there does not seem to be any phylogenetic connection between the two genera ( Fig. 1). Future research should try to establish if Guembelitria and Jenkinsina are related and whether Jenkinsina evolved from Chiloguembelina or from some other genus in the Early Eocene.