Phylogeny of the Eocene Antarctic Tapetinae Gray, 1851 (Bivalvia, Veneridae) from the La Meseta and Submeseta formations

Abstract. Systematic analysis shows that the Southern Hemisphere bivalve genus Retrotapes includes the Antarctic species R. antarcticus, R. newtoni, and R. robustus and recognizes for the first time the presence of Katelysia represented by K. florentinoi. Two new genera were erected in this study: Marciachlys new genus to include M. inflata new combination, and Adelfia new genus, which includes A. austrolissa new combination and A. omega new species from the Eocene of Antarctica, and the late Eocene Chilean A. arenosa new combination. Eurhomalea carlosi was synonymized with K. florentinoi; Cyclorismina marwicki with R. antarcticus; Gomphina iheringi was considered an indeterminate species; and Cockburnia lunulifera was excluded from the Tapetinae. These systematic assignments are supported by a phylogenetic analysis, which recognizes an Austral clade of Tapetinae, comprising all the genera mentioned above, along with Marcia, Paleomarcia, Atamarcia, and Protapes.

Because E. florentinoi, E. carlosi, E. inflata, and Cockburnia lunulifera were defined on articulated specimens, Cyclorismina marwicki has a hinge plate with very similar characters to those of Retrotapes antarcticus, and Eumarcia austrolissa is known through one partially eroded valve with broken cardinal teeth, it is necessary to carry out a full revision of these species in the light of the new material collected by MJA in the 2014 field season performed by the Instituto Antártico Argentino (IAA; Argentinean Antarctic Institute).
The Submeseta Formation was informally named by Montes et al. (2013) (Fig. 2), being the uppermost part of the former La Meseta Formation and is dated 43.4-33.9 Ma according to Montes et al. (2013). However, a new chronologic scheme for During the 2014 field season several specimens were collected from concretionary facies of Cerro Jonas (Acantilados II Allomember, La Meseta Formation). This new material has a hinge plate with long and thin anterior and posterior cardinal teeth, a very short middle one, and an anterior lateral tooth in the left valve. These features are not coincident with those of Tapetinae, and added to the presence of a sub-central umbo and a very large and wide lunule, suggest it belongs to the Family Corbiculidae; but more studies are being carried out to confirm this assignment.
As previously mentioned, Gomphina iheringi Zinsmeister, 1984 (Fig. 3.10, 3.11) has no internal diagnostic characters because the only known specimen is filled with sedimentary matrix, and Zinsmeister placed it in Gomphina because of its shape. However, this feature is similar to that of some Meretricinae or Corbiculidae species; for that reason, we follow the proposal of Beu (2009) and consider this specimen nomen dubium.
Zinsmeister (1984, p. 1522, figs. 10A-10C) named the species 'Cyclorismina' marwicki ( Fig. 3.3-3.6) (Eocene; La Meseta Formation) based on its subcircular shells, adducing that although the generic position is uncertain, the combination of the shell-shape and internal features are similar to those of the genus Cyclorismina Marwick, 1927(type species C. woodsi Marwick, 1927Late Cretaceous, New Zealand;Fig. 3.7-3.9). Cyclorismina has circular-shaped valves, a small umbo, sunken lunule, unbounded from the rest of the shell, absent escutcheon, almost vertical nymphs, divergent cardinal teeth, entire 1, and long pallial sinus, which is triangular, ascendant, and with a sharp apex. The type material illustrated by Zinsmeister (1984), as well as those specimens donated by this author to the MACN-Pi collection, do not have any of the features described for Cyclorismina, even the shape of both species is not similar; therefore the Eocene species do not belong to this genus. In the present contribution, 'C.' marwicki is synonymized with Retrotapes antarcticus (Fig. 3.1-3.3) because of its similarity to the globoid morphotypes of this species analyzed by Alvarez and Pérez (2016). The only significant difference is that the contacts between the anterior and posterior margins with the dorsal and ventral ones are slightly more rounded, giving the subcircular shape described by Zinsmeister (1984); but it is within the range of shapes observed in the globoid morphotypes of R. antarcticus.
Another species from the Eocene of Antarctica included in the genus Eurhomalea is 'Eurhomalea' claudiae Stilwell, 2000, from McMurdo Sound (Ross Sea). Stilwell (2000) tentatively included this species in Eurhomalea due to its similarity to the shape of Retrotapes newtoni (Wilckens, 1911) and Katelysia florentinoi (Zinsmeister, 1984), both previously considered as Eurhomalea by Zinsmeister (1984) and Stilwell and Zinsmeister (1992), but mentioned that it might belong to Retrotapes del Río, 1997 instead. However, 'E.' claudiae has divergent cardinal teeth, which means that it does not belong to Retrotapes. The teeth are very small, the middle one is vertical and the posterior is slightly curved, resembling those of K. florentinoi. However, it is not possible to determine if the middle teeth of both valves are bifid, a diagnostic character of Katelysia, either in the material illustrated or in the description provided by Stilwell (2000). In addition, neither muscular characters nor the pallial sinus can be observed. A search of the type material Alvarez and del Rio-Eocene Antarctic Tapetinae 801 supposedly housed at Smithsonian Museum of Natural History was made, but the specimens were not in the collection, and are presumably lost. Therefore it is possible that 'Eurhomalea' claudiae belongs to a new genus, but the lack of key features and any type material makes it impossible to determine.
The assignment of Eurhomalea florentinoi Zinsmeister, 1984 into the genus Katelysia expands the stratigraphic and geographic ranges of the latter genus to the Eocene of the Antarctic Peninsula. Katelysia differs from Eurhomalea Cossmann, 1920 (type species E. rufa [Lamarck, 1818], Pacific Ocean between the central region of Chile and Panama) because of its smaller size, suboval shape, shorter cardinal teeth that do not exceed  Emended diagnosis.-Pallial sinus short (but longer than that observed in extant species). Sculptured with low commarginal ribs, which are closer towards the ventral margin of the disk.
Description.-Shell thin, medium to small sized, suboval to elliptical. Umbo small, placed in anterior third of length. Posterodorsal and anterior margins tend to be dorsally straight and ventrally convex, ventral margin convex. Lunule lanceolate, short, bounded by a line. Escutcheon very narrow and short, wider in the left valve, with commarginal sculpture similar to the rest of the shell. Nymph narrow and smooth. Hinge plate narrow, curved behind the cardinal teeth, which are divergent, short, and do not exceed the ventral margin of the hinge plate. Right hinge with 3a tooth lamellar, sloped forwards; 1 triangular, narrow, bifid, tilted backwards; 3b wide, bifid, sub-horizontal; posterodorsal region of right valve with a groove for the insertion of left valve. Left hinge with 2a triangular, wide, bifid, higher than the other teeth; 2b narrow, bifid; 4b lamellar, sub-horizontal, and separated from the nymph by a groove. Adductor muscle scars isomyarian, the anterior one is deeper; anterior pedal retractor scar placed below the anterior margin of the hinge plate and separated from the adductor muscle scar; posterior pedal retractor scar joined to the posterior adductor muscle scar; and small pedal elevator muscle scars under the hinge plate. Pallial sinus short, triangular, dorsal and ventral margins straight, and apex rounded. Shell sculptured with low and narrow commarginal ribs, which are closer to each other towards ventral margin of the disk, and wider than the interspaces. Some well-preserved specimens have very fine radial ribs. Measurements.-Holotype USNM 365516: length 41 mm, height 30 mm (Appendix 2).
Remarks.-Zinsmeister (1984) erected Eurhomalea florentinoi describing hinge plate characters, but without providing any internal images, and erected Eurhomalea carlosi based only on articulated specimens, considering the presence of a lunule in E. carlosi as the only difference from E. florentinoi. The revision of specimens of both species herein allows us to synonymize them because they have the same shape, sculpture, position of the umbones, and hinge plate. Regarding the presence or absence of a lunule depends on the grade of erosion of the external surface; with significant erosion, the line that bounded the lunule is not recognizable. The only difference between both species is that E. carlosi is smaller than E. florentinoi. The specimens of E. carlosi ( Fig. 4.9) have four to six annual growth lines, whereas specimens of E. florentinoi have more than ten. This fact, added to the morphological evidence discussed before, indicates that the specimens of E. carlosi are probably young specimens of E. florentinoi.
Most of the new specimens of K. florentinoi ( Fig. 4.6-4.8) have an initial shell shape that is similar in form and number of annual growth lines to those of the specimens of E. carlosi. However, in the same specimens, the convexity of the shell and the number of annual grow lines increases, developing a globoid shape, similar to that observed in R. antarcticus, which is associated with the great longevity of this Antarctic taxon (Alvarez and Pérez, 2016).
Eurhomalea florentinoi does not belong to the genus Eurhomalea (type species E. rufa) because this genus has a large and subquadrate shell, with sub-central umbo, narrow hinge plate with thin and high cardinal teeth that exceed the hinge plate margin, and a pallial sinus, which is large, triangular, and with an acute apex.
This species is included in the genus Katelysia because of its medium-sized shell, suboval, laterally compressed shape, with very narrow lunule, which is lanceolate and bounded by a line. The escutcheon is very narrow, but wider in the left valve. The hinge plate is narrow, with short cardinal teeth that do not exceed the hinge plate margin and bifid 1 and 2a teeth. The pallial sinus is very short, with rounded apex. The shell is sculptured with low commarginal ribs, wider than the interspaces. Some well-preserved specimens have very fine radial ribs.
Beu (2009) assigned K. florentinoi to the genus Retrotapes del Río, 1997 and commented that the lack of lunule would support its inclusion in the genus Frigichione Fletcher, 1938. The differences between Katelysia and Retrotapes are discussed Journal of Paleontology 94(5):799-818 above. Still, Katelysia florentinoi cannot be assigned to Frigichione because of its smaller and thinner shells, suboval shape, with narrow escutcheon and thinner teeth.
Katelysia florentinoi is distinguishable from the Miocene Australian and New Zealand species (K. corioensis and K. lunulata) by its suboval shape, larger shell, and sculpture of low commarginal ribs. Katelysia florentinoi differs from the extant species of the genus by its larger shell, longer pallial sinus, and sculpture of low commarginal ribs. However, its low sculpture is closer to that of K. florentinoi polita Nielsen, 1963 (p. 223, pl. 1, figs. 4-6). The flat morphotype of K. florentinoi is less inflated than the extant species, which are similar in convexity to the globoid morphotype ( Fig. 4.8, 4.14). Diagnosis.-Shell very inflated, umbo prominent and rounded. Hinge plate narrow with divergent cardinal teeth, short and straight, the anterior ones are angled forward, 1 tooth almost vertical, 2b entire. Pallial sinus short (shorter than K. florentinoi), ascendant, with apex rounded.
Etymology.-Marcia of the mist. The name refers to Marcia because the external similarity to this genus and Achlys refers to the Greek word for mist, because the new specimens of this new genus were found at the northern end of the Marambio Base airstrip in a day of thick mist.
Remarks.-Marciachlys n. gen. differs from Retrotapes in having a lunule bounded by a line, divergent cardinal teeth, with the anterior ones very much tilted forward, entire 2b, and ascendant pallial sinus. Tooth 3b is horizontal in Retrotapes, but is ventrally oriented in Marciachlys n. gen. This genus is distinguishable from Katelysia in having a more convex shell, prominent and rounded umbo, wider lunule, larger hinge plate and teeth, entire 2b, and longer pallial sinus.
Marciachlys n. gen. differs from the New Zealand Eumarcia and Atamarcia by its smaller shell, prominent and rounded umbo, entire 2b, and shorter and ascendant pallial sinus. The specimens MACN-Pi 2531-2533 ( Fig. 5.8) from Tolhuin in the vicinity of Kaiken Hostel (Tierra del Fuego Province, Argentina), of Paleocene age, could be assigned to Marciachlys n. gen. since they have similar shape, convexity, umbo, and sculpture. However, these specimens are larger and no inner characters are known.
Marciachlys inflata (Zinsmesiter, 1984), new combination Figure 5. 1-5.7 1984  Marwick, 1927;Miocene, New Zealand). Both species have similar shell shapes, but Adelfia austrolissa n. comb. does not belong in Atamarcia because of its flat lunule bounded by a line in its entire extension, shorter 4b, bifid 2b with a shallower groove, narrower 3b, smooth 1, and triangular, shorter, and subhorizontal pallial sinus. In addition, the interspaces of the sculpture have the same depth throughout the valve and are closer to each other towards the ventral margin in Adelfia n. gen., while in Atamarcia the interspaces are deeper in the anterior margin of the valve and are closer to each other towards the posterior one.
Ortmann (1899) erected Venus arenosa (PRI 72690; Fig. 6.10, 6.11) (Loreto Formation, late Eocene, Punta Arenas, Chile) based on three right valves embedded in sedimentary matrix. This taxon is included in Adelfia n. gen. because it shows a similar shell shape, with a hinge plate with cardinal teeth similar to those of Adelfia austrolissa n. comb., but with a margin more curved behind the teeth, and the sculpture when preserved is similar to that of Adelfia n. gen. The lunule is lanceolate and slightly concave, and this is the only difference from species of Adelfia n. gen. The scarcity of observable characters in the eroded and matrix-embedded right valves, as well as the complete absence of left valves, plus the fact that the differences with Adelfia austrolissa n. comb. are very little, make it difficult to differentiate both species. Based on these characters and due to the geographic and stratigraphic distance between both taxa, they are recognized here as valid species. If new evidence on Adelfia arenosa n. comb. would appear, allowing us to synonymize them, the specific name arenosa would have priority over austrolissa n. comb.
Adelfia n. gen. differs from Retrotapes by its flat lunule bounded by a line, divergent cardinal teeth, entire 1, and narrower 3b. It is distinguishable from Katelysia by its larger and more convex shell, wider escutcheon, higher hinge plate with longer teeth, entire 1, and longer and subhorizontal pallial sinus. Adelfia n. gen. differs from Marciachlys n. gen. by its lessconvex shell, less-prominent umbo, narrower lunule, less-tilted forwards anterior cardinal teeth, backward sloping 1, subhorizontal 3b, and longer and subhorizontal pallial sinus. The sculpture of Adelfia n. gen. differs from that observed in the three compared taxa, and consists of broad, low, and flat ribs.
Emended Diagnosis.-Shell medium to large sized. Up to 10 pedal elevator muscle scars. Pallial sinus with dorsal margin subhorizontal, ventral margin curved, and apex small and truncated.
Description.-Shell thin, large size, suboval. Umbo small, placed at anterior quarter of length. Dorsal margin slightly convex; ventral, anterior, and posterior margins convex. Lunule narrow, flat, and bounded by a line. Escutcheon narrow, wider in the left valve. Nymph narrow and smooth. Hinge plate wide, slightly curved behind the cardinal teeth, which are divergent, narrow, and do not exceed the ventral margin of the hinge plate. Right hinge with 3a tooth lamellar, straight, sloped forward, and shorter and lower than the 1, which is thin, entire, and tilted backwards; 3b triangular, bifid, and sub-horizontal; posterodorsal region of right valve with a groove for the insertion of left valve. Left hinge with 2a triangular, straight; 2b triangular, bifid, and higher than 2a; 4b lamellar, curved, sub-horizontal, and separated from the nymph by a groove. Adductor muscle scars deep; anterior pedal retractor scar placed below the anterior margin of the hinge plate and separated from the adductor muscle scar; posterior pedal retractor scar joined to the posterior adductor muscle scar; up to 10 pedal elevator muscle scars small, deep. Pallial sinus short, triangular, with dorsal margin straight and subhorizontal, and ventral margin curved, and apex small and truncated. Sculpture of broad, low, and flat ribs, closer to each other towards the ventral margin, separated by interspaces of equal depth throughout the valve. Measurements.-Holotype USNM 441638: length 60.5 mm; height 46.5 mm (Appendix 2).
Remarks.-Adelfia austrolissa (Stilwell and Zinsmeister, 1992) n. comb. was erected based on only one specimen found in the Cucullaea I Allomember (middle Eocene). During the 2014 field season, new articulated material was collected at the cliff near Campamento Point (64°13 ′ 45.6 ′′ S; 56°39 ′ 55.9 ′′ W), from the Campamento Allomember, extending the stratigraphic range of this species. Later, based on this new material, other specimens of this species were recognized in the collection of MACN-Pi from Acantilados II and Cucullaea II allomembers. The external surface of the holotype is almost smooth, and its name is derived from this fact, but the surfaces of the new material are not smooth, revealing that the holotype is partially eroded and the outer layer of the shell is almost absent.
Adelfia austrolissa n. comb. differs from the other Tapetinae from Acantilados II to Cucullaea II allomembers (Retrotapes antarcticus and R. robustus), by its suboval shape and characteristic sculpture. The articulated specimens of A. austrolissa n. comb. and R. newtoni have the same shape and they can be easily confused with each other. The difference between them is that R. newtoni has a concave lunule, bounded by a deep groove, whereas A. austrolissa n. comb. has a flat lunule, bounded by a line, which is not visible in eroded specimens.
Adelfia omega new species Figure 6.12-6.15 Holotype. Etymology.-Omega is the last letter of the Greek alphabet, and the name refers to some of the specimens of this taxon collected in the last bed with veneroids of the Submeseta Formation, found just below the Weddell Sea Formation (Pliocene) outcrops.
Material Remarks.-During the 2014 expedition several specimens of this new species were found in the beds with veneroids at the top of the Submeseta Formation. They have the same shape, umbo, lunule, and similar sculpture and hinge plate to those of Adelfia austrolissa n. comb., which is why this species is included in Adelfia n. gen. All specimens are articulated, so it is not possible to observe internal features. However, some cracked shells were broken to have access to the inner cast and these showed some internal characters. This demonstrated that there are some differences between the pallial sinuses of both species, which are used as diagnostic characters. In addition, the elements of the sculpture of the new species are lower than those of Adelfia austrolissa n. comb.
The most conspicuous difference is the size, with A. austrolissa n. comb. being larger than A. omega n. gen. n. sp. (Appendix 2). This could indicate that the new species is a juvenile of the type species, but counting of the annual growth ribs observed in eroded specimens of A. omega n. gen. n. sp. revealed that some specimens are more than 30 years old.
Specimens of Marciachlys inflata n. comb. and Retrotapes newtoni were also collected from the beds where A. omega n. gen. n. sp. was found. These three species have a suboval shape and most of their specimens are eroded. Although it is very hard to identify them, there are some distinguishing features. Marciachlys inflata n. comb. has a larger lunule, prominent umbo, and is wider than the other two species. Retrotapes newtoni has a concave lunule bounded by a deep groove, and A. omega n. gen. n. sp. has a smaller and flatter lunule and the disk is sculptured with low, broad, and flat ribs.
In order to minimize the loss of information, most of the reviewed material was studied first hand. Only 3.57% of the entries are missing in the data matrix. Characters from 18 to 23 are lineal measurements (not ratios), therefore, in order to compare the measurements, these were rescaled to the average size of Tapes literatus (Linnaeus, 1758) (average height value of 45.9 mm). The ratios of the other continuous characters were logarithmized (log10) following Mongiardino Koch et al. (2015).
Ingroup.-Previously, Alvarez (2019) tested the monophyly of the genus Retrotapes. The results allowed him to argue about the possibility of the existence of an Austral Tapetinae clade that includes the genera Retrotapes, Atamarcia, Paleomarcia, and Katelysia. With the objective of evaluating the relationships of the new genera described herein within this Austral clade, the selected terminals of the ingroup settled in Alvarez (2019) were resampled. Only four of the 13 known species of the genus Retrotapes (Alvarez et al., 2014) were included in the matrix: the three Eocene Antarctic species (R. antarcticus, R. robustus, R. newtoni) and the type species, R. ninfasiensis del Río, 1997, the type species of the genera Adelfia (A. austrolissa n. comb.) and A. omega n. gen. n. sp., as well as the type species of Marciachlys n. gen. (M. inflata n. comb.) and the type of Katelysia (K. scalarina) and K. florentinoi. Other austral taxa that share some features with Retrotapes were also included: Atamarcia sulcifera (Marwick, 1927) (type species of Atamarcia), Eumarcia fumigata (Sowerby, 1853) (type species of Eumarcia), and Paleomarcia tatei (Fletcher, 1938).
Search.-A phylogenetic analysis was performed following the maximum parsimony criterion using the TNT 1.5 software (Goloboff et al., 2008) through a heuristic search of 100 replicates of Wagner trees (with addition of random sequences) followed by TBR branch swapping algorithm holding 10 trees per replicate. Characters 1-25 were considered as continuous. The methodology of character weighting was implied weighting (Goloboff, 1993), performing 100 searches for k values between 1 and 100, because bivalves and mollusks in general are a homoplastic group. However, an exploratory search without implied weighting was also performed. Support values were estimated by resampling using frequency differences under Bootstrap (BS) (Felsenstein, 1985) and Jackknife (JK) (Farris et al., 1996), with a p = 8 (equivalent to removing 10% of the characters) (Goloboff et al., 2003), and performing 1,000 pseudo-replicates.

Results
Each search performed with a different k value (k between 1 and 100) resulted in a single topology, obtaining ranges of k where the recovered topologies are similar to each other. The ranges of trees that have similar topologies are 6-28 ( Fig. 7.2) and 29-100 (Fig. 7.3). The BS and JK values were calculated and informed on each topology (Fig. 7). The tree obtained for the k value 29 is the most abundant (71 of 100 trees), and is similar to the one obtained in an exploratory search performed without implied weighting; it also has the best BS and JK values, and the discussion is based on it.
In all the performed searches, Adelfia n. gen., Katelysia, and Retrotapes (closely related to Paleomarcia and Atamarcia) are monophyletic groups. Marciachlys n. gen. is the sister taxon of the clade Marcia + Protapes, and in the search performed with equal weighting (Fig. 7.1), which was a k value of 29, it is possible to recognize a large clade of Austral or sub-Antarctic taxa that was previously unknown.
Adelfia n. gen. is supported by high JK values in all searches, and has four synapomorphies on the topology with k values ranging from 29 to 100: (c7) width of the pallial sinus 17% of the total height; (c30) high hinge plate; (c39) smooth tooth 1; and (c70) presence of growth commarginal ribs with thin ribs interspaced. Marciachlys n. gen. is recovered as the sister taxon of the clade Marcia + Protapes with low values of support in all searches and has only three synapomorphies: (c5) inclination of tooth 1 of 107-110°; (c10) tooth 4b wide (52% of 2b width); and (c63) smooth angle between the posterior and dorsal margins.

Discussion
In all the performed searches, subtropical taxa are recovered as basal for the clade Tapetinae (e.g., Eumarcia, Pahia, and Neotapes for the searches with equal weighting and with k values of 29-100, and the clade Polititapes + Eurhomalea + Venerupis + Ruditapes in searches performed with k values of 6-28), which suggests a possible subtropical origin for the subfamily Tapetinae. On the other hand, for the searches with equal weighting and with k values of 29-100, a clade of Austral taxa with clear affinities with Eocene Antarctic genera is recovered. Within this group, Adelfia n. gen. and Katelysia (including its type and extant species from the South Australia, K. scalarina) are recovered basal to a clade that includes Paleomarcia (Miocene of Kerguelen Island) and Atamarcia (Miocene of New Zealand) as a sister taxa of Retrotapes. Concerning the clade Retrotapes, the Eocene Antarctic species are recovered basal to the type species of the genus, R. ninfasiensis (late Miocene, Patagonia). In all searches, Marciachlys n. gen. is basal to a clade that includes the genus Marcia and Protapes. These last two genera have a subtropical distribution nowadays and this result raises a possible evolutionary history connected with taxa from the Eocene of Antarctica.
In view of the results previously discussed, a possible scenario for the evolution of Tapetinae is that this subfamily, as a whole, had a subtropical origin and later migrated to higher latitudes, in this case to the south. Once in Antarctica, the group would have diversified during the Eocene, to migrate once again, but this time towards lower latitudes, which led in part to the conformation of the Neogene Austral faunas.
As Beu (2009) argued, it may be too risky to claim an Antarctic origin for the aforementioned fauna due to the scarcity of early Cenozoic record in the southern hemisphere. But the phylogenetic results obtained here can shed light on the origin and distribution of part of the modern fauna from the Austral seas, which without a doubt has a close link with the Eocene fauna of Antarctica.
Although it is not the objective of the present contribution, it is important to discuss the phylogenetic position of the genus Atamarcia. This fossil genus was erected by Marwick (1927) as a subgenus of the extant Eumarcia. Later, Stilwell and Zinsmeister (1992) and Beu (2009) included some species from the Eocene of Antarctica within this genus, including Atamarcia austrolissa, which we considered as the type species of the new genus Adelfia. Eumarcia and Atamarcia were not grouped together in any of our phylogenetic results. Eumarcia is part of the tropical clade of the Tapetinae, whereas Atamarcia is recovered as an Austral Tapetinae. The genera Adelfia n. sp. and Atamarcia were never recovered as sister taxa. Atamarcia is more closely related to Retrotapes than to Adelfia n. sp., in agreement with the morphological differences recognized between the type species of Adelfia n. sp. and Atamarcia discussed in the systematic section, which in consequence led us to name the new genus for the Eocene Antarctic species. We use only the type species of Atamarcia for both the systematic comparison and the phylogenetic study because a preliminary search of the taxa included in Atamarcia allowed us to conclude that a major systematic revision of this genus is needed. There is a big morphological disparity among the species originally assigned by Marwick (1927) as Atamarcia.
Considering only some hinge characters as an example, A. sulcifera has slightly curved cardinal teeth, whereas A. benhami (Hutton, 1874) and A. crassatelliformis Marwick, 1927 have straight cardinal teeth, and A. crassa Marwick, 1927 has a straight right anterior cardinal tooth, joined to the lunule and strongly curved middle and posterior ones.
The presence of K. florentinoi (Zinsmeister, 1984) in the Eocene of Antarctica considerably extends the stratigraphic and geographic occurrences of Katelysia Römer, 1857, which nowadays inhabit the marine coast of the Southern Australia.
Retrotapes, Katelysia, Adelfia n. gen., and Marciachlys n. gen. are grouped together in a clade with other Austral taxa, such as Paleomarcia and Atamarcia, and with two subtropical taxa, Marcia and Protapes, a relationship previously ignored. This reinforces the importance of Antarctica as a center of origin and distribution of fauna during the Cenozoic. The basal position of the clade is occupied by Adelfia n. gen. and Katelysia, represented by the Eocene Antarctic K. florentinoi and the extant K. scalarina (Southern Australia). Marciachlys n. gen. is basal to the clade, formed by the extant genera Marcia and Protapes. The fossil genera Paleomarcia (Miocene, Kerguelen Island) and Atamarcia (Miocene, New Zealand) are the sister groups of Retrotapes. The latter is represented here by its Eocene Antarctic taxa and the Miocene Patagonian R. ninfasiensis, being the most ancient species of the genus, and R. newtoni, which is basal to the rest of the species. Stilwell, J.D., and Zinsmeister, W.J., 1992