The evolution of Early Cretaceous Dorothiinae (Foraminiferida)

Dorothia and Marssonella are ataxophragmiacean genera typically possessing canaliculate walls. They are believed to have evolved independently from noncanaliculate Praedorothia gen. nov. and Protomarssonella gen. nov. of Early Cretaceous time.


INTRODUCTION
High spired Ataxophragmiacea have long been considered to possess agglutinated, noncanaliculate walls (e.g. Loeblich & Tappan, 1984, p. 14). However, it has more recently been shown (Banner & Desai, 1985) that initially triserial genera (Clavulinoides Cushman, Valvulina d'Orbigny, Clavulinopsis Banner & Desai etc.) independently develop canaliculate walls in their evolution from noncanaliculate ancestors, during or at the end of the Late Cretaceous. Gaudryina bulletta Carsey was designated by Plummer (1931) as the type species of her new genus Dorothiu, the main point of distinction being that Dorothia contains "more than three" chambers in the initial part of the test. This multiserial group is studied, in this paper, for its canaliculation.
Topotypes of D . bulletta were figured by Loeblich & Tappan (1964) from Texas (Onion Creek, Texas). Hofker (1969) studied specimens of D. bulletta from the type level (Navarro); his figures (1969, p. 31, fig. 69) show clearly that the agglutinated walls are distinctly canaliculate. Loeblich & Tappan (1985, p. 201-202, pl. 12) again showed that Dorothia Plummer exhibits a finely canaliculate wall. Hofker (1969) also observed the canaliculate nature of the wall in D . pupoides (Reuss) from the Santonian. We are able to confirm this canaliculate structure in Dorothiri pupa (Reuss), from the Campanian chalk of Norfolk. D. Hofkcr (1976) showed that Recent D. curta (Cushm ; i n ) and D. scabra (Brady) were also canaliculate (1976, p. 184-87, figs. 51-52). Again, we are able to confirm the canaliculate nature of Recent species of Ihrothia. Plate 2 illustrates essentially similar developments of canaliculi in the walls of three different species from Recent Caribbean sediments deposited from the upper slope to the bathyal.
Gaudryina oxycona was designated as the type species of his new genus Marssonella by Cushman in 1933. Many authors (e.g. Trujillo, 1960Loeblich & Tappan, 1964) believed that Marssonella Cushman and Dorothia Plummer are synonyms and Marssonella, being the junior synonym, should be disused. However, for the reasons given by Barnard & Banner (1980, p. 391, 392) and by Bartenstein et al. (1971, p. 131). Marssonella should be distinguished from Dorothia at generic level. All of the group of species which can be referred to Marssonella have concave or flat septa, and have an evolution into Pseu~ote.~riilariella in the Early Cretaceous (to Cenomanian). In contrast. the group of species which relate to Dorothia have convex septa and persist to the Recent, not being internally complex until the Campanian (Mantanziu) . Specimens of Marssonella oxycona (PI. 3 , figs. a-i) are here shown to possess walls which are canaliculate with a very distinctive "keriothecal" structure (PI. 5 ,. This is the first time this pronounced structure has been observed in the Ataxophragmiacea, and is quite different from the wall structure seen in species of Dorothia, be they obtained from the Late Cretaceous (PI. 1) or from the Recent (PI. 2).
We are now able to show, however, that not all species which have been referred to the genera Dorothia or Marssonella possess canaliculate walls. Species described below, from the Early Cretaceous, may have solid walls, with no trace of canaliculations. These are worthy of distinction at generic level, although we would not go so far as to separate them into a distinct family. When Loeblich & Tappan (1985) reviewed the systematic position of Dorothia, they referred the genus (and the subfamily Dorothiinae) to the family Eggerellidae because of the canaliculate wall in the type species of both Dorothia and Eggerella. However, the fact that canaliculate walls develop independently in different lineages (Banner'& Desai, 1985) and the fact that it may be difficult to recognise the early evolutionary development of such walls, indicate that it is unwise in practise and improper in taxonomy to separate canaliculate and noncanaliculate genera (which are otherwise similar) into different, high-category, suprageneric groups. The noncanaliculate species are here referred to the new genera Praedorothia and Protomarssonella, which are defined, described and discussed below. We believe them to be directly ancestral to the respective canaliculate Dorothia and Marssonella of younger sediments.
The diagnosis given by Loeblich & Tappan (1984) to distinguish between the superfamilies Verneuilinacea Cushman, Ataxophragmiacea Schwager and Textulariacea Ehrenberg cannot be maintained. Loeblich & Tappan (op. cit.) defined all of the Verneuilinacea (including the family Textulariopsidae), as well as all of the Ataxophragmiacea (including the Dorothiidae), as being noncanaliculate, distinguishing them on this supposedly primary character from the canaliculate Textulariacea (including the biserial Textulariidae and the initially triserial Valvulinidae). Not only have terminally uniserial canaliculate genera (e.g. Clavulinoides and Clavulinopsis) evolved independently within the Verneuilinacea, wholly triserial forms may also have become canaliculate. An example of this last situation is a species obtained from the Early Aptian of D.S.D.P. site 398 (PI. 4, figs. la-lb) which is externally a Verneuilina but which is canaliculate. This is the earliest evolution of canaliculation in the Verneuilinidae yet known to us (compare Banner & Desai, 1985, p. 87-89) and the species predates solid walled true Verneuilina of the Late Cretaceous. This confirms that wall canaliculation has evolved independently at different times in several generic groups and cannot be taken to be a primary phyletic character.
Therefore, it is proposed that the superfamilies discussed here should be redefined as follows:  (1877), Dorothiidae Balakhmatova (1972) and Eggerellidae Cushman (1937). Superfamily Textulariacea: essentially biserial (occasionally with a third chamber in the very first whorl); wall canaliculate (as in the Textulariidae) or solid (as in the Textulariopsidae). Also here can be accommodated the families Plectorecurvoididae, Pseudobolivinidae and Nouriidae.

MATERIALS
For the purposes of this paper we have used material which has been collected from the following localities: -1. Speeton, North Yorkshire, England; the Speeton Clay outcrop is on the coast of the southern part of Filey Bay, where Cretaceous beds from Berriasian to Aptian age are exposed in the cliff and on the foreshore. The section was collected on three occasions (once under the guidance of Dr. P. F. Rawson) using the stratigraphy described by Kay (1964), Rawson (1971), Rawson and Mutterlose (1983), Fletcher (1969) and Neale (1960Neale ( , 1962. 3.
Type species: Gaudryina bulletta Carsey, 1926. Emended Diagnosis. Early stage trochospiral, with 4 or more chambers to a whorl, later stage reduced to biserial; wall agglutinated, may be of calcareous particles; aperture an interiomarginal slit; septa convex throughout; chamber walls (and sometimes the septa) canaliculate.
Dorothia moorbergensis sp. nov (Pl. 1, figs. la-le) Description. This species has a finely agglutinated wall with calcareous cement, smooth both exteriorly and interiorly but penetrated by fine canaliculi which open into the interior of the chambers but which are sealed exteriorly by a very thin imperforate granular layer (approximately equal in thickness to the, lining of the chamber interiors). The septa appear to be partly or wholly noncanaliculate. The initial trochospiral part of the test has about four chambers in the earliest whorls, rapidly reducing to three per whorl; it is circular in cross-section and relatively rapidly tapering (apical angle about 50"). The biserial part tapers much more slowly (angle of taper about 15") and contains chambers which are almost as high as they are broad; as in the holotype, three pairs of biserial chambers comprise about 213 of the total test length. The apertural and

Explanation of Plate 2
Modern species of Dorothiu from Recent sediments of the Caribbean Sea. (All specimens by courtesy of the British Museum, Natural History). All specimens have been dissected and their destruction has precluded their subsequent cura tion.
These three species all possess the convex septa of Dorothia (compare PI. 1, fig. 2a) but include the closest gross homeomorphs of Marssonelfu available for study from Recent material.  Remarks. Moullade (1984) reviewed the published references to "D". praehauterivian and excluded the citations by Neagu (1972) and Sliter (1980) (their figures were reproduced in Moullade, 1984, pl. 7 figs. 11, 12-13, respectively); in the taxonomy employed in this paper, those references would be to Protomarssonella, and not even to Praedorothia, because of their flat to concave septa. The specimens figured here, in this paper, are virtual topotypes and are included to enable reliable comparison to be made with the other species recorded. The wall is solid (Pl. 4, fig. 2b, d).
Distribution. This very cylindrical Praedorothia dominates the pre-Hauterivian dorothiid assemblage. The figured specimens (PI. 4, figs. 2a-d) were obtained from the type locality (Orosei, Sardinia) and are dated as Late Valanginian (Dieni & Massari, 1966). Moullade (1984) and Luterbacher (1972) confirm the Early Valanginian to Late Valanginian occurrence of this species from various DSDP cores. Kuznetsova & Gorbachik (1985) recorded "D". praehauteriviana from the Tithonian of the Crimea; without study of their specimens it is difficult to assess this identification but their drawing suggests that they have misidentified a species of a Protomarssonella.

Explanation of Plate 3
Figs. a-i. Marssonella oxyconu (Reuss); type species of Marssonella Cushman; all specimens from the Campanian (Belemnitella mucronutu Chalk), Burgh, Norfolk, England (Rowe locality 163).   (Moullade) S.S. but the biserial chambers are lower, being markedly broader than high, especially towards the apertural end of the test, and are strongly overlapping. The septa and the terminal faces are less convex and are sometimes somewhat flattened near the apertural face, which is lower and less steep than in P. zedlerae S.S. Remarks. The morphological difference between P. zedlerae S.S. and P. zedlerae luterbacheri parallels that between Dorothia pupa (Reuss) and D . pupa depressa Barnard & Banner (1953, p. 191, pl. (Luterbacher, 1975(Luterbacher, ). pl. 4, figs. 15-16. 1984 . This designation is in full accord with the opinion expressed by Cushman (1937, p. 57) that this may be taken as the type locality. The age of the mucronata-chalk at this locality is Campanian and is very close to that of the mucronata chalk of the area of Norwich (Norfolk, England) from which we have obtained our comparative material. The specimens of M . oxycona from the mucronata chalk of Burgh, Norfolk, which have been studied for this research, show that the calcium carbonate of the chamber lateral walls is morphologically a set of elongate grains rigorously arranged perpendicularly to the inner chamber surface (PI. 3, figs. c and e), externally coated by a crust of equidimensional, randomly arranged grains. This is a structure similar to that previously observed in Verneuilina tricarinata d'Orbigny (Banner & Desai, 1985, pl. 3) but as yet unknown in either Dorothia or Praedorothia. However, in Marssonella oxycona (but not observed in V. tricarinata) , canaliculi are regularly developed between the aligned carbonate columns of the chamber walls (Pl. 3, figs. c, f and g). The canaliculi are closed exteriorly by the outer crust, but open into the chamber lumen over the whole of the inner surface of the lateral chamber walls. Where the crust is abraded, the caniculi become exposed to the exterior (Pl. 3, fig. a). The terminal face and septa are as fibrous as the lateral chamber walls (Pl. 3, figs. d and h) but do not possess canaliculi; confinement of the canaliculi to the lateral chamber walls is common in Verneuilinids and Valvulinids (e.g. Banner & Desai, 1985, pl. 1, fig. 9; Banner & Pereira, 1981) but this structure is newly described for the genus Marssonella.
We report below on the structures of species from the Early Cretaceous which possess the test form, coiling mode and chamber shape of Marssonella but which have solid walls, lack both grain alignment and canaliculi, and which we here distinguish as Protomarssonella n. gen.
The genus Marssonella Cushman ranges from the Late Cretaceous to Paleocene or Early Eocene; e.g. M. oxycona floridana Applin & Jordan was first recorded from the Velasco Formation (Late Paleocene) and M . lodoensis Israelsky has its type horizon in the Lodo Formation of California, which may be as young as Early Eocene. Other species which are stratigraphically younger may not be referable to Marssonella, e.g., M . keijzeri van Bellen, from the Middle Eocene of the Netherlands, has a broad valvuline-like tooth and M . altisuturafis Poag, from the Alabama, appears to be triserial throughout.

Explanation of Plate 5
Figs. la-ld. Praedorothia zedlerae (Moullade)   Remarks. This genus differs from Marssonella just as Praedorothia differs from Dorothia. Each of these genera contains a distinct group of species and each characterises a distinct stratigraphical interval. For the reasons noted above (in the Introduction), the differences in wall structure are not regarded as of supregeneric significance.
Protomarssonella may occur in beds as old as Late Jurassic (e.g., "Marssonella" donesiana Dain, as figured by Luterbacher, 1972 (PI. 1, figs. 16, 17); the specimens called "Dorothia praehauteriviana Diena and Massari", obtained by Kuznetsova & Gorbachik, 1985, from the Early Tithonian, are, as noted above, also referable to a (new) species of Protomarssonella), and is shown below to be well established in Early Cretaceous time. P. hechti differs from P. kummi in its regularly tapering, conical test, very weakly depressed or flush sutures, and more depre\\ed chambers. Distribution. The primary types of this species, like the topotypes figured here, are from the Late Valanginian of Sardinia. Bartenstein & Kaever (1973) and Bartenstein et al. (1971) recorded " M . kummi" from the Valanginian to Middle Barremian; they included specimens referable to P. hechti in this identification and the specimen figured by Bartenstein et at. (1971) (correctly identifiable as P. hechti), was obtained from the Barremian of Bulgaria, confirming that rangc for this species.
Protomarssonella kummi has the solid walls characteristic of this new genus. This is observed in specimens from the Hauterivian of the Speeton Clay. P. kummi possesses a narrowly and slowly tapering, subconical test with very weakly depressed sutures. P. hechti (described above) differs in its more broadly conical, more rapidly and regularly tapering test and more depressed chambers. Distribution. The primary types of P. kummi were obtained by Zedler from Late Valanginian to Early Hauterivian of Germany; Dieni & Massari's specimens came from the Late Valanginian of Sardinia. The specimens figured here were obtained from the Hauterivian of Speeton as were the specimens obtained by Lott et al. (1986).

DISCUSSION
New evidence has been presented here to modify previously published opinions about the Early Cretaceous evolution of the Dorothiinac; c.g. Moullade (1984) suggested a phylogeny very different from that proposed here (Fig. 1). This paper distinguishes marssonellid from dorothiid tests and recognises the development of canaliculations in the walls of both of them; this leads to a finer and more reliable taxonomic discrimination than in the previously published record (e.g. Moullade, 1984). Because the canaliculate wall is phylogenetically more advanced than the solid agglutinated wall in other groups of genera (see e.g. Banner & Desai, 1985), it is reasonable to assume that the same applies to the Dorothiinae. This is confirmed by the solid walls of all marssonellid species so far studied from the Early Cretaceous, and with the appearance of canaliculi in true Marssonella in the Late Cretaceous. The stratigraphic succession suggests that Protomarssonella kummi either descended from P. hechti or that they both had a common ancestor in the Berriasian or E. Valanginian. The succession of species of Praedorothia suggests the evolutionary sequence P. praehauteriviana to P. praeoxycona to P. zedlerae, as depicted on figure 1. Dorothia moorbergensis n. sp. is now the oldest known true Dorothia, and its Early-Middle Hauterivian occurrence suggests its descent from P. praehauteriviana. D. moorbergensis n. sp. is not yet known above the Hauterivian and would therefore seem to be phylogenetically unrelated to Dorothia spp. of the Late Cretaceous and younger beds; Dorothia is probably polyphyletic and it seems as though canaliculation evolved independently at least twice in this stock. This hypothesis is in accord with that already published for the repeated evolution of the canaliculate wall in Late Cretaceous -Paleocene Verneuilinidae and Valvulinidae (Banner & Desai, 1985). It is likely that Praedorothia and Protomarssonella had a common ancestor in the Late Jurassic. Because ataxophragmiids, dorothiids, verneuilinids and valvulinids of Late Cretaceous and younger strata are well known to show repeated evolution involving reduction of the number of chambers per whorl (e.g. to develop a biseriality and even uniseriality), it may be supposed that the ancestor to the Dorothiinae possessed more chambers per whorl than did Protomarssonella or Praedorothia. It is possible that the ancestor was a Tethyan species. A suitable candidate is Riyadhella Redmond (see also Barnard & Banner, 1980, p. 391, 400), which is a high trochospire, initially with four to six chambers in the whorl reducing to three or four (Pseudomarssonella possesses a broad apertural flap and is therefore unlikely to be the ancestor). Further reduction, during terminal growth, to ultimate biseriality , would produce Protomassonella and Praedorothia. Protomarssonella arose in the Tithonian, probably giving rise both to Pseudotextulariella (see Barnard & Banner, 1980) in the Berriasian/ Valanginian and, quite separately, to Praedorothia, at least by Valanginian time. Dorothia first evolved, from Praedororhia, in the Hauterivian, and then, again in the Aptian (Barnard & Banner, 1980, op. cit., p. 392), supplanting Praedorothia by the Late Cretaceous and persisting into Recent Seas. Marssonella is not known to have evolved from Protomarssonella until the Late Cretaceous and seems to have become extinct in the Eocene. The evolutionary history of the two principal lineage-stocks seems to have been distinct since the beginning of Cretaceous time.