The status of Rotalia Lamarck (Foraminifera) and of the Rotaliidae Ehrenberg

One damaged specimen remains in the Lamarck Collection, Geneva, from the original type series of Rotalia trochidiformis (Lamarck) the remainder of which was included in the Defrance Collection in Caen, all of which was destroyed in World War II. Restudy of this sole remaining paralectotype, together with topotypes and other material held in the British Museum (Natural History) has revealed the morphological changes that occur with ontogeny in this species and has made an emended generic diagnosis necessary. In the adult, R. trochidiformis develops open septal canals with a tributary system of secondary and tertiary fissures by resorption and a dense mass of vertical pillars which effectively destroy the umbilical cover plate seen in the juvenile. The septal canals connect with the umbilical canal via supplementary apertures and with vertical canals between the umbilical pillars. They also connect with the foraminal passage via subsutural canals. The idea the Rotalia should be reassigned to the Discorbidae is rejected and the integrity of the Rotaliidae is affirmed.


INTRODUCTION
The early history of the taxonomic treatment of Rotalia trochidiformis (Lamarck, 1804) is thoroughly reviewed by Davies in his emendation and redescription of 1932. His conclusion was as follows (op,cit.: 41 1): 'It seems clear that Rotulites trochidijkwnis is the species which pre-eminently represents Lamarck's genus Rotulia. It is the first to be described and figured under that heading; and the only species mentioned at the time which fully matches the original description of the genus. It is true that the term Rotulia was first regarded as a mere equivalent of Rotalites, and that another species had previously been named as representing Rotulites [Rotulites tuherculosa Lamarck,180 11; but the latter species had never been recognisably indicated, and its own author had not only abandoned it but had also completely revised the generic description which he had drawn up to suit it. The revised generic description, with its alternative name of Rotalia, clearly applies to the species trochidiformis and not to tuherculosu '. The clear distinction that was made by Lamarck ( 1 804) in erecting his genera Discorhis ( D . vesicularis) and Rotulia became blurred under the influence of the extreme 'lumping' of the 'English School' in the mid-nineteenth century. Carpenter, Parker & Jones (1862) included both Discorhis and Rotulia in their new genus Discorhinu (a junior synonym) and thus offended against the Articles of the International Code of Zoological Nomenclature. Brady (1884) similarly took a wide view of ' D i s c~~h i n u ' .
He noted that its chief diagnostic character was the presence of umbilical lobes, ... 'In fullest development they are separated by marked constrictions from the body of the segments and form supplementary chambers ... sometimes they form solid masses of shell substance, filling the umbilicus, and markedexternally with exogenous tubercles'. However, he gave separate recognition to Rotalia, noting that the majority of species had double septa (following Carpenter, Parker & Jones, 1862); these two forms again had umbilical chamberlets and some had a "more or less complicated canal system". He also considered the fine wall structure and minute pores characteristic. Unfortunately, he chose . 'Rotalia' heccarii (Linne) as the "central species" which again offends the International Code of Zoological Nomenclature. It also happens to be the type of Ammonia Briinnich, 1772. The first worker to provide reasonable accurate drawings of specimens of Rotalia trochidiformis showing a growth series, from the type locality of Grignon in the Paris Basin, was Cushman (1927). His drawings show a juvenile with pointed umbilical lobes, radial, e\-cavated suturesand incipient side fissures, a second, larger juvenile with coalesced lobes and development of pillars and a third quite well-grown specimen with the ventral side completely incised with full development of pillars. The implications of these marked differences have since been overlooked.
Cushman's observations were confirmed and extended by Davies (1 932) who examined the original type material of Lamarck from Grignon (Defrance Collection in the Museum d'Histoire naturelle, Caen) together with other material from Parnes and Chaussy in the Paris Basin as well as from the Contentin, housed in the British Museum (Natural History), London and in the Ecole des Mines, Paris. The material examined by Davies also included the one damaged specimen from the original Lamarck Collection, housed in the MusCum d'Histoire naturelle, Geneva.
The results of this work which included the sectioning of a number of specimens led to the conclusion that the following features were generically diagnostic: the presence of double septa expressed as radial furrows on the ventral side: a transverse septum separating off the inner portion of the chamber (astral lobe) marked by a notch (astral furrow). The progressive development of pillars which tend to be confined to each whorl was also mentioned but not stressed. Davies denied that an umbilical or septal canal system existed, neither did he mention the presence of a dendritic network of fissures. Davies selected a lectotype from the original type series, a juvenile that best showed the umbilical features. Unfortunately, the specimen was destroyed with the rest of the type series in a bomb attack on Caen during World War I1 (Maync, 1952), leaving the Geneva specimen the sole remaining primary type, the only paralectotype.
It might have been thought that this detailed study would have settled most of the outstanding questions concerning the nature of the genus Rotalia. In fact, confusion has continued down to the present day. This arises, in part because although the type species has been recorded as occurring abundantly in high level, shallow water sediments deposited at the height of the successive Palaeogene transgressions, typical forms (i.e. Rotalia trochidiformis sensu stricto) are confined to the Middle Eocene, possibly even to the Paris Basin. Not only are specimens from elsewhere smaller but they are often much altered, recrystallised and even dolomitised, usually with the last chamber broken. For this reason Smout (1954) was unable to determine the apertural characters in his specimens from Qatar. He also denied the existence of the umbilical chamberlets or 'astral lobes' claimed by Davies, although the excellent figures of his lectotype clearly do show discorbidlike umbilical lobes.
LCvy et al. ( I 986) also deny that double septa are unique to the rotaliids and claim that they occur in Discorbis and a number of related genera. This leads them to subsume the Rotaliidae within the Discorbidae, with drastic repercussions for taxonomy.
In hope of solving some of these problems we have reexamined Davies' material in the British Museum (Nat. Hist.) and re-illustrated key specimens by scanning electron microscopy, as well as the sole remaining specimen from the Lamarck Collection kindly lent to us by the MusCum d' Histoire naturelle de Genitve, Switzerland. As this remaining type specimen (the sole paralectotype, following the destruction of Davies' lectotype) is badly damaged we have redescribed the genus and species on the basis of three additional undamaged topotypes from Grignon and other specimens from the Davies Collection from Pames and Chaussy, nearby in the Paris Basin, which together represent a growth series (see Plates I , 2, Text fig. 2, and also pl. 3, Text-figs 1, 3 for internal features). Efforts were made to obtain adult specimens from Grignon but, as pointed out by Professors J.W. Murray (letter 25.2.87) and D. Curry (letter 1.4.87), the species is actually very rare there, with a known maximum diameter of 1 .5mm, and the classic pit is now closed. We have therefore illustrated two specimens from the Murray & Wright Collection in the British Museum (Nat.Hist.) and another, larger specimen from Professor Murray's personal collection, subsequently donated, near the maximum size for the specimens found during recent years at the type locality.

SYSTEMATIC DESCRIPTIONS
Order Rotaliida Lankester, 1885Superfamily Rotaliacea Ehrenberg, 1839 Family Rotaliidae Ehrenberg, 1839 Subfamily Rotaliinae Ehrenberg, 1839Genus Rotalia Lamarck, 1804 Type Species: Rotalites trochidiformis Lamarck, 1804. Emended generic diagnosis. Test trochospiral with angular or keeled periphery; dorsal side evolute, ventral side involute with simple, arched basal aperture; umbilicus on ventral side secondarily closed by imperforate, sub-triangular lobes ('astral lobes'), well developed in juveniles and seen extending from the last few chambers in adults; lobe separated from primary chamber cavity by a deep notch ('astral furrow') and partial partition ('toothplate') which continues as a septal flap over the previous apertural face producing a 'double septum'; distal end of lobe pointed with peripheral lip and tucked into umbilical end of the aperture, effectively making the distal notch a supplementary aperture which communicates with the primary chamber via an umbilical opening; ventral septal sutures deeply entrenched and prolonged across umbilical lobes by resorption to form open canals; with growth there is a progressive modification of these open canals between earlier chambers to produce a dendritic network of secondary and tertiary fissures which divide the chamber walls and umbilical lobes and feed into the' septal canals; granules develop between the fissures becoming strong vertical pillars, these remain more or less confined to the chambers and lobes but partially fuse to become a complex umbilical plug; the zone of supplementary apertures and notches tends to remain clear in the last few chambers as an umbilical canal ('space') and the septal canals communicate with this spiral canal via the supplementary apertures and irregular vertical canals in the umbilical mass. Communication between the main chamber lumenand the umbilical areais viavertical, subsutural canals ('interlocular space') linking the foramina with the septal canals. Initially, there is free communication into the umbilical area from beneath the astral lobe, with a definite arched opening ('labial aperture') beneath the distal lip. Remarks. The chief differences between our emended diagnosis compared with that given by Loeblich & Tappan ( 1 988) arise because we have been able to show that there are important developments during ontogeny, as follows: (a) The pointed end of the astral lobe is tucked into the aperture, making the distal notch ('astral furrow') into a supplementary aperture which communicates with the primary chamber via the umbilical opening and foraminal passage.
(b) The groove beside the distal lip of the astral lobe becomes the site of the extension of the septal canal when the next chamber is added. The septal canal connects with the foraminal passage via a subsutural canal ('interlocular space') between the septal flap and the previous apertural face, as well as with the previous supplementary aperture.
(c) With growth the development of the open canal system and pillars destroys the umbilical cover plate present in the juvenile, taking over almost the entire umbilical area in the adult.
Septa1 canals are not mentioned in their diagnosis, only fissures which are defined as deeply cleft or incised sutures. The diagnoses agree, however, in emphasising the angular or keeled periphery, the presence of a spiral umbilical canal and the septal flap.
Di.sc~)r.hitiu tr.oc.hidiformis (Lamarck), Carpenter, Parker & Jones, 1862: 204,205. D e sc r i p t io n s. Ro ru liu t r w h idIf01.m is shows marked morphological changes with ontogeny and different populations are often dominated by different growth stages. Because this has caused confusion in the past we have taken the opportunity provided by the British Museum (Nat. Hist.) collections to describe and illustrate a full growth series before giving a summary description. The sequence is shown in Plate 1,2; Text fig. 2: Topotype (British Museum (Nat. Hist.) no. P 49283, PI 1 , figs. 1-1): Test trochoid, sinistral, biconvex with evolute dorsal side slightly higher than the involute ventral side; periphery acute, keeled, entire; about 2% whorls of chambers visible on the dorsal side with marked, slightly thickened whorl suture; rate of chamber size increase as added, low, shape trapezoid, sutures backward curving, flush; 7 chambers visible on ventral side with elongate, triangular umbilical (astral) lobes fused together to form partial coverplate; astral lobes separated from primary chamber cavity by a deep distal notch (astral furrow) and partial partition ( Murray, 1987. Hypotype (British Museum (Nat. Hist.) no. P 52261 , PI. 1, figs. 4-4,5,6); A well preservedadult specimen with unbroken final chamber in which the details closely correspond to those of the topotypes (Pl. I). It has about 4 whorls and 10 chambers visibleon theventral side. The keel is more strongly developed and the growth offissures and pillars has largely destroyed the umbilical cover-plate seen in the topotype (P49283. PI. I , Fig Hypotype (British Museum (Nat. Hist.)

Remarks:
The essential features of the species revealed by analysis of the paralectotype, topotypes and related material in the British Museum (Nat. Hist.) and the detailed redescription of a growth series are: (a) Rotuliu tt-oc,hidiformissensu str-ictocan be relatively large (specimens reaching up to4.5mm have been recorded by Davies, 1932) with five to six whorls and up to 16 or 17 chambers in the last visible whorl.
(b) Triangular umbilical (astral) lobes are a prominent feature of the Juvenile stage becoming fused into a partial coverplate with growth. They are partially separated from the primary chamber cavity by a deep notch (astral furrow) and partial partition. The distal end of the final lobe is tucked into the umbilical side of the simple, basal aperture isolating the distal notch as a supplementary aperture.
(c) With growth the deep, ventral sutures (developed between the double septa) become secondarily excavated and prolonged by resorption across the astral lobes to form a system of open canals. The side walls of these fissures become incised with a dendritic system of secondary and tertiary fissures which in the adult are developed over the entire ventral surface apart from the peripheral margin and the last one or two chambers.
(d) There is progressive development of granulations along the edges of the septal canals leading to strong development of vertical pillars eventually covering the entire ventral surface, confined to a single whorl towards the periphery but fusing into a composite mass in the umbilicus.
(e) The umbilical space beneath the astral lobes remains relatively clear of pillars in the last few chambers and can be traced between the umbilical mass and the umbilical partitions as a spiral canal. The septal canals link with this space via the supplementary apertures and also with the vertical canals in the umbilical mass. The septal canals also link with the foramina1 passage via sub-sutural canals ('interlocular spaces').

THE ROTALIA TROCHIDIFORMIS GROUP
Rotuliu trochidiformis in the wide sense shows interesting variations both geographically and with time, though the precise relationships require further study. Typical forms, as described by Davies in hisemendation of 1932and redescribed here, may be confined to the Lutetian (Middle Eocene) of France. Even largerforms with six or seven whorls occur near Hauteville in the Cotentin with clearly marked spire on the dorsal side which is less embracingdescribed by Davies as R. trochidiformis var. huutevillensis. This subspecies occurs at the Lutetian/Auversian boundary but unfortunately it is not clear if R. troc.hid$ormi.s huutevillensis represents a higher horizon than R. trochidfor-mis rrochidiformis in the same local succession.
Specimens as large as those from the Paris Basin have not been discovered outside France. Although described as

Explanation of Plate 1
Figs 1-6 Rotuliu trochidforrnis (Lamarck). Fig. 1 -1 'typical' by Smout ( 1954), specimens from the Lower Eocene of Qatar reach a maximum of only 2.3mm and average 1.6mm maximum diameter with three whorls and 12-14 chambers in the final whorl. The specimens described by Gill (1953) similarly reach a maximum of 2.3mm. In neither case were apertural features discernible and thus they can only be assigned to R. trochidiformis trochidiformis with a question mark. from the Palaeocene (Smout, 1954;Haynes, 1962) are smaller still, reaching only 1 .Omm in diameter with three to four whorls and 7-9 chambers visible at the periphery. These almost certainly represent a distinct species as they differ in apertural details. Rotalia cf. trmhidiformis of Pfender (1935) from the Palaeocene ('Montian') of Turkey similarly reaches only about 1 .Omm in maximum diameter but with up to 13 chambers in the last visible whorl. It is also high domed, without a keel and more coarsely perforate. It is illustrated in thin section only, as is Rotalia cf. tr.ochidiformis of Gaetani et al. ( I 983) from the Palaeocene, Spanboth Formation of the NW Himalayas which reaches 1.6mm diameter.
Rotalia tr.ochidifjrmis has also been reported from the Upper Cretaceous but without useful figures or description. Many of the specimens appear to be small and may belong to other species. In the British Museum (Nat. Hist.) are specimens, identified at R. tiwhidiforrnis, from the Maestrichtian of Qatar (Henson Collection) which approach 2mm in diameter; these require careful investigation beyond the scope of the present paper.
Specimens assigned to R. trochidiformis

DISCUSSION
Our analysis shows that a major cause of confusion has been a misconception about the size of R. trochidifhrrnis. Rotalia trm~hidiformis S.S. exceeds 4mm in diameter in the adult and R . ti-ochidiformis hautevillensis Davies, from the Contentin in western France, exceeds 5mm diameter. This misconception arose in part because Davies ( 1932) deliberately chose a juvenile as lectotype (op. cit., pl. 3, figs 4,5,7), reaching only 1.9mm maximum diameter, in order to illustrate astral lobes. Similarly LCvy et al. ( 1986) have illustrated juveniles of R. trmhidiformis in order to show the discorbidlike nature of the lobes ('folia'). Strikingly, the specimen chosen by them tocompare with Discorhis vrsicularis (op.cit., pl. I , fig. 1, 1.4mm maximum diameter) is actually smaller (pl. I , fig. 4, 1 .Omm maximum diameter). Although we have also described topotypes that conform with Davies' lectotype care must be taken not to confuse the issue by comparing such juveniles with the adults of D. vesicularis.
The inclusion of material from outside the type area, possibly belonging to different subspecies, has also led to the idea that R. trochidiformis is relatively small. Haynes (1 98 1 : 285) himself fell into this trapanddescribedRotalia as 'small, up to 3 or 4 whorls' basing his ideas on specimens from the Palaeocene of Libya! Because of these mistakes, workers have not paid proper attention to the ontogenetic development of Rotalia trochidiformis S.S. and to those features only well expressed in the adult. This applies particularly to the open septal canal system and the dendritic fissures characteristic of the adult and which together with the massive development of pillars well distinguish Rotalia from Discorhis. A major problem for authors has been the difficulty of obtaining well grown species in recent years (see also above, p.96). Thus the maximum size of specimens (from Grignon) studied by Parvati (1971) was 1.9mm and naturally she could not appreciate this development and described variation in the ventral aspect as ... 'generally associated with the development of secondary calcareous material, which may, or may not obliterate the coarse perforation of the primary wall, the spiral fissure, 'the astral fissures, the sutural fissures, etc. to the extent of presenting only a near chaotic pustulose and irregularly fissured surface'.
In Discorhis the septaappear to be doubled only where the toothplate is attached whereas in Rotalia this doubling continues across the septal face (see Miiller-Merz, 1980) which allows the deep entrenchment of the sutures on the ventral side. In their assumption that the doubling of the septa is essentially the same in Rotaliu and in Discorhis, Levy et al. appear to have followed Davies ( 1932) who described the entrenched sutures on the ventral side in Rotalia as furrows 'produced by the puckering up of the chamber floor to form double septa between successive chambers' which suggests restriction to the umbilical area on the ventral side.
The first worker to observe that the septal sutures become entrenched on the ventral side was Cushman ( 1927). Although denied by a number of other researchers our material clearly shows how this occurs by resorption during growth with prolongation of the canals across the umbilical lobes. The further development of dendritic fissures into the side walls, feeding into the entrenched sutures produces an open canal system which may have facilitated protoplasmic inflow concomitant with outflow from the aperture. Through concentration upon the characters of the juvenile, Levy et al.

Explanation of Plate 2
Figs 1-3 Roralia rrochid$ormis (Lamarck). Fig. 1 -1 Davies (1932: pl. 3: fig. 13: pl. 4, fig.  9). From Parnes (Lutetian), Paris Basin All SEM photographs, normal secondary electron images of coated material. missed the significance of the fissures, dismissing them as merely, 'deep sutures ' (op. cit., 1986: 65). They are also in some semantic confusion concerning canals. In their attempt to remove Rotalia from the Rotaliacea, as it were, they follow Reiss & Merling (1958) who restrict the term 'canal system' to intra-lamellar structures and favour Smout's ( 1 954) definition: 'Canal systems are complexes of essentially tubular cavities of relatively fine bore within the shell material'. However, we believe this to be unnecessarily restrictive and our usage follows the original zoological definition of a canal as a duct or groove in the shell which falls within that of Loeblich & Tappan (1988) who define it as an interlocular space, albeit 'usually tubular', and also that of Haynes (1 98 I), who described canals as 'grooves or passages'.
The intralamellar spiral canals of the higher rotaliids begin as grooves on the primary lamella. There is, therefore, an ontogenetic and phylogenetic progression from open to closed canals. The terms would also seem to apply to the ring-like, peri-umbilical space finally produced between the composite plug and the toothplates. This was called an umbilical canal by Rijsinge (1930) and Smout (1954: 183), and a spiral canal byParvati (1971( ), butdismissedbyDavies (1932 ..: What is more, there seems to be no function for a canal system lying along the umbilical line of open chamber mouths. Anything in the nature of a tube across this would not facilitate communication but block it...'. This shows he thought of a canal system as, necessarily, a kind of pipe-work. His figure of acut specimen (op.cit., pl. 2, fig. 15), reproduced here as PI. 3, fig. 2, clearly shows the umbilical space.
Further, although Davies considered that the astral lobes resembled the star-like secondary chamberlets of Asterigerina, communication with the umbilical area and the exterior remains open ('labial apertures' of Reiss & Merling, 1958) and closed umbilical chamberlets such as those in the Asterigerinacea are not formed.
One important feature of our results is the discovery of irregular vertical canals in the umbilical area (Fig. 3). These are less numerous and well developed than the 'funnels' described in Medocia by Parvati (1971). That genus also lacks secondary and tertiary fissures. CONCLUSIONS Rotalia trochidiformis S.S. approached 5mm in diameter as a well-grown adult and is distinguished by triangular umbilical (astral) lobes which make a -prominent cover-plate in the

Explanation of Plate 3
Figs 1-5 Rotalia trochidiformis (Lamarck). Fig. 1 -1, Topotype. Stereo-pair of specimen broken almost centrally to give natural vertical section revealing foramina and foramina1 passage, x52. Murray and Wright Collection, British Museum (Nat.Hist.) no. P 52262. Fig. 2, Hypotype. Reproduction of Davies' (1932) p1.2, fig. 15, showing large adult specimen rubbed down on ventral side to reveal peri-umbilical canal, x20. British Museum (Nat. Hist.) no P 28647. Fig. 3-3, Topotype. Stereo-pair of opposite edge of specimen in fig. 1-1 broken to show interior of last two chambers and reveal umbilical partition, septal flap and double septum, x52. Fig. 4, Topotype. Oblique view of juvenile specimen shown in fig. 1-1 broken across to reveal internal details, x52. juvenile. With growth the progressive development of an open canal system and massive development of vertical pillars breaks up the cover-plate and the umbilical lobes are seen only in the last one or two chambers (if preserved). The similarity of the toothplates in Rotalia and Discorhis and the discorbid-like features of the juvenile inRotaliu tl-ochidqbrmis are consistent with the idea of the origin of Rotalia from within the Discorbidae, possibly via intermediate, umbonate forms related to Rotorhinella of Bandy (1944). It need not lead US to doubt the integrity of the Rotaliidae which includes a number of large foraminifera1 genera such as Lockhurtia and Dictyoconoides with clear evolutionary links withRotalia and distinctive distribution in space and (Palaeogene) time. We should remember that on general grounds derived from Neo-Darwinism and ecological population genetics that the separation between families at point of origin is no greater than that between related species/subspecies. However, our results show conclusively that Rotalia, as represented by the type, R. tr-ochidiformis, is closer to Loc khnrtia and its allies than to the discorbids. Although we disagree with those authors who would incorporate Lockhartia and Dictyoconoides with Rotalia, the latest being Hofker (197 I), their views underline our point about the relationships of these forms. It is important here not to be unduly influenced by the currently fashionable Neo-Linnean ('punctuationist') search for morphological gaps between species and by extension for bigger breaks between genera and between families, this approach inevitably leads to lumping of intergrading taxa.
A formal proposal by Levy et ul., (1986) that Rotalia be included in the Discorbidae and the superfamily name Rotaliacea be abandoned (op.cit.: 68)is therefore rejected.
distal lip of a.l.and La. rear view of aperture 5 5