The Cranial Characters of the Brevicipitid Genus Cacosternum (Boulenger).

The Cranial Characters of the Brevicipitid Genus Cacosternum (Boulenger). By C. G. S. de Villiers, M.A., Ph.D., Professor of Zoology, University of Stellenbosch, South Africa. With 12 Text-figures. THE present paper is a continuation of my work on the cranial characters of the two South African Brevicipitid genera lacking the claviculo-procoracoidal arch in the pectoral girdle. The genus Phrynomerus (Noble) was discussed in a previous paper in this Journal (vol. 73), to which the reader is referred for details of technique, which was, however, slightly varied for purposes of the present research, inasmuch as double bulkstaining was resorted to. The nuclear stain, haemalum, was applied in the usual way, and after all traces of alum used for differentiation were washed out, the object was bulk-stained for twenty-four hours in a strong aqueous solution of Bismarck brown and differentiated in ordinary tap-water for forty-eight hours. I have not yet experimented with bulk-staining for connective tissue and muscle; stains recommended for this purpose are fuchsin and light green or eosin respectively. The use of van Gieson in conjunction with Bismarck brown is considered by many workers to lead to confusing results, so that pure fuchsin is recommended for connective tissue and for decalcified bone. All material was decalcified with Ebner's solution in preference to nitric acid, which is, however, used with great success for the same purpose by Professor Stadtmiiller of Gottingen, according to a personal communication from him. The material of Cacosternum bottgeri was kindly supplied by the Director of the Transvaal Museum; Cacosternum capense was collected by the author on the Stellenbosch Flats after the first rains in May 1928. The eighteen specimens collected are the only ones known besides the type

276 C. G. S. DB VILLIBRS specimens collected by Mr. Eose of Cape Town. Three specimens of Cacosternum namaquense were collected by Dr. Herre of Stellenbosch University during the phenomenal Namaqualand rains of 1929; but owing to the rarity of the material, this species was not microtomized. Mr. Hewitt has recently upheld the species and recorded the existence of two specimens referable to it in the collection of the South African Museum. REVIEW OF THE EXISTENT LITERATURE ON THE GENUS. The genus Cacosternum was first described by Boulenger (1887, page 51). The new genus was referred to the Engyostomatidae and the following osteological details were enumerated as being characteristic of it: palate toothless, without dermal ridges; tympanum hidden; no procoracoids; coracoids slender; sternum extremely small, cartilaginous; diapophyses of sacral vertebrae strongly dilated. The presence of strong subarticular tubercles of the fingers and toes, also present in Phrynomerus, was mentioned. The first South African species to be described was C. bottgeri in Boulenger's catalogue of 1882, where it is referred to as Arthroleptis bottgeri (p. 118). Werner (1910) described a new species of Cacosternum, C. namaquense. Boulenger (1906-9) first called C. bottgeri by the name now generally used. No cranial characters are mentioned in this work. Valuable osteological data are supplied by Hewitt (1911) from which I select the following: (p. 215) 'maxillary and premaxillary teeth present, no vomerine teeth, palate without dermal ridges, tympanum hidden. In the skull a fronto-parietal foramen is present and the frontoparietals are but feebly ossified'. Hewitt then regarded Cacosternum as allied to the Dyscophids, on account of the presence of teeth. Andersson (1911) merely notes the occurrence of Phrynomerus and Cacosternum in British East Africa, where the Swedish Zoological Expedition made a collecting tour in 1911. Methuen (1913, p. 123) classifies C. bottgeri under the sub-family Dyscophinae of the Engyostomatidae, probably on account of the presence of teeth on the maxilla. No cranial characters are mentioned. Hewitt (1919) discusses

CRANIUM OF CACOSTERNUM 277 in detail the affinities of the Cacosternum-Anhydrophryne group (pp. 184r-7) and concludes that the absence of procoracoids is of greater importance than the presence of maxillary teeth and stresses the affinities with the Malagasy Dyscophids. The presence of a fronto-parietal fontanelle in Breviceps and Cacosternum is not considered by the author as of great systematic importance. That part of Noble's work (1922) dealing with Cacosternum and its allies is of extreme importance, as it discards the toothed ' Dyscophidae' as an independent family and classes them with allied toothless forms under the new family Brevicipitidae. Noble's paper of 1924 contains interesting speculations on the origin of the Ethiopian Brevicipitidae, which are referred to the older African fauna whichflourishedwhen the mainland was still connected with Madagascar (pp. 277 and 278), and suggests possible affinities between the Cacosternum-Anhydrophryne group and the Malagasy Anodonthyla, the affinities of which genus are further discussed by Noble (1926, pp. 2-4), the conclusion arrived at being that Anodonthyla is derived from a form with divided, dentigerous vomer. In 1926 Hewitt contributed valuable information regarding the morphology and systematics of the genus Cacosternum: C. capense, the new Cape species described by the same author in 1925, was shown to possess a divided coracoid, which may be of great value in a comparison of Cacosternum and Phrynomerus. Werner's species (1910) of C. namaquense was upheld and found to be allied to C. capense, from which it differs mainly in being toothless. A new sub-species of C. bottgeri, C. bottgeri albiventer was described. C. namaquense, of which Werner's specimens were supposed to be the only ones extant, was rediscovered by Dr. Herreat IKoeboes, Eichtersfeld, Namaqualand. The three specimens collected in October 1929 agree in every detail with those described by Werner, who collected them at Steinkopf. The new specimens are in the Department of Zoology of the University of Stellenbosch. Noble (1926 a, p. 4) again refers to the possible relationship of Cacosternum and Anhydrophryne, and of the latter genus with the genera Arthroleptis and Arthroleptella, without, NO. 294 T

278 C. G. S. DE VILMBRS in the opinion of the author of this paper, making good his point. Nieden (1926, p. 11) groups Cacosternum under the Engyostomatidae and is therefore not aware of the discovery of maxillary teeth in the genus by Hewitt (1911). Power published a paper dealing with Cacosternum in 1927. On p. 250 he comes to the remarkable conclusion that ' Cacosternum has bufonid affinities', but bases his argument solely on the pattern of the chitinous tooth-rows and the position of the anus. The present author published a paper (1929 a) on the macroscopic development of C. capense, whose spawn and larvae are extremely common round Stellenbosch, and is of opinion that the larvae are in appearance and in habit as unlike Bufo larvae as they possibly can be, and resemble much more closely those of the genus Pyxicephalus. Mr. Rose, the discoverer of C. capense, gave a preliminary account of its habits and life-history in 1926 (p. 437). The latest paper up to date, referring to Cacosternum, is that of Noble (1927, pp. 116 and 117), in which the following conclusions are reached: (1) The pectoral girdles of Cacosternum and Phrynomerus agree closely; (2) the larvae of the former genus are ranid and unlike those of Phrynomerus ; (3) Anhydrophryne was evolved from Arthroleptella, from which it differs by the absence of a clavicle; (4) the close relationship of Cacosternum and Anhydrophryne is stressed; (5) Cacosternum is possibly also derived from Arthroleptella but is in any case ' closely related to ranids and not to other Brevicipitids'. In a paper read before the British Association at the Cape Town Session, the author of this paper has fully described the non-aquatic life-history of Arthroleptella and has proved that the genus is perfectly ranid. Contrary to what Noble presupposes, it has no clavicle, but a procoracoid only. It was suggested that Arthroleptella, Cacosternum, and Anhydrophryne might have to be removed to the Ranidae. RESULTS OF OWN INVESTIGATIONS. It is proposed in the following account to discuss chiefly those features of the cranial characters of the genus Cacosternum

CRANIUM OF CACOSTERNUM 279 in which it differs from Phrynomerus, the other South African genus lacking a procoracoid. The following description applies in the main to Cacosternum capense: the difference between it and the smaller species, C. bb'ttgeri, will duly be noted. C. namaquense Avas not sectioned as the material is too rare. THE OLFACTORY CAPSULE. Both cartilagines praenasales are present as in Phrynomerus ; the superior is short, the inferior long and flexed beneath the solum nasi, and forms the main support for the premaxilla. It is imbedded in the tubules of the large glandula intermaxillaris. The same conditions prevail in 0. bottgeri. The two vestibular 'Wiilste' are both present and the plica obliqua is blunt and short and suspended from the cartilago obliqua as in Phrynomerus, not from the tectum nasi as in Eana; the same applies to C. bottgeri. The recessus sacciformis, described by Gaupp for Kan a, was found to be absent in Phrynomerus. In Cacosternum it is, however, present, but its anatomical relations are not the same as in Eana. Gaupp described the organ on pages 625 and 633 of the third volume of the ' Anatomie des Frosches' (1904). On page 625 we read: 'Der Wulst [i.e. the one taking the place of the cartilago alaris] ist von unten und hinten her gewissermassen unterminiert durch den Becessus sacciformis, der hinten in die Vestibularnische tibergeht, medial- und ventralwarts mit dem Infundibulum und dem Cavum medium zusammenhangt.' Additional details are furnished on p. 633: 'Lateral von diesem Wandwulst stiilpt sich ein schlaffwandiger Schleimhautblindsack, der Recessus sacciformis, nach vorn und oben hin vor, der seinen Ausgang von der erwahnten Kommunikationsspalte und im Anschluss daran (nach vorn hin) auch noch von der lateralen Kante des Cavum medium nimmt.' In Phrynomerus an evagination of the infundibulum and cavum medium is absent, but a sac-like recess is present in Cacosternum, although it does not undermine the large' Wulst', is short and wide instead of high and narrow as in Ran a (see Gaupp, loc. cit., fig. 140, p. 627), and does not communicate with the vestibule. The T 2

280 C. G. S. DE VILLIEBS recessus in C. capense is figured in Text-fig. 1, from which it will be seen that it originates from both infundibulum and cavum medium; it does not represent the point of communication of the ductus nasolacrimalis with the cavum medium, as the TEXT-FIG. 1. s n Transverse section through the three narial cavities of the left side in Cacosternum eapense. bl.v., blood-vessel; cv.i., cavum inferius; cv.m., cavum medium; cv.p., cavum principale ; gl.imx., glandula nasalis intermaxillaris; g.n.l., glandula nasalis lateralis; inf., infundibulum; l.if., lamina inferior cristae intermediae; l.sp., lamina superior cristae intermediae; rc.m., recessus medialis; r.sc, recessus sacciformis; s.n., solum nasi; s.o.c, side of olfactory capsule; spmx., septomaxillary. latter receives the duct at its blind, lateral division, after the recessus sacciformis has disappeared from sections. It will be seen from the figure that the recess is surrounded by the septomaxillary bone, which forms a capsule for it. In C. bottgeri the recess has exactly the same disposition as in 0. capense. In Phrynomerus two remarkable prechoanal sacs were described (loc. cit.); they were shown to be derived from an unpaired prechoanal sac in the young form. The apparatus is easily derived from a fusion of the two ' Gaumenleisten' (Gaupp) or palatal ridges. In Cacosternum, Text-fig. 2, there is an blv

CRANIUM OF CACOSTERNUM 281 unpaired preohoanal sac as in the young Phrynomerus, but the choanae no longer open into, but just beyond it. In C. bottgeri the sac was filled with dense mucous material. The premaxilla and maxilla have the usual relations with neighbouring structures, but are not separated palatally by the ventrallyflexedcrista subnasalis as in Phrynomerus. They are moreover toothed: the teeth consist of a functional 6c c Transverse section through the head ofcacosternum capense in the anterior region of the olfactory capsule, bc.cv., buccal cavity; cr.sb., criata subnasalis; g.n.m., glandula nasalis medialis; M.c, Meckel's cartilage; pch.s., prechoanal sac; pl.t., planum terminale; sjp.n., septum nasi; ten., tectum nasi. Other abbreviations as for Text-fig. 1. row (acrodont), and a few additional rows waiting to take the place of the former. The premaxillary teeth of C. bottgeri are sketched in Text-fig. 3; they have the normal histological structure of anuran teeth as described for Ran a by Krause (1923), and were discovered in the genus by Hewitt (1911). C. namaquense differs from the other species in being toothless. The vomer in C. capense is a very small bone investing the edge of the solum nasi region of the choana. There is no considerable prolongation of the bone into the narial cavity, as Cacosternum lacks the enormous cartilaginous axis of the eminentia olfactoria, which the vomer invests in

282 C. G. S. DE VILLIERS Phrynomerus. C. bottgeri has a comparatively large vomer, its size being due to the extent of the squame investing the ventral surface of the solum nasi. The bone encapsules a few tubules of the glandula intermaxillaris. In both species the vomer is entirely edentulous, and separated from its fellow on the other side by fibrous connective tissue. In G. capense the two bones approach each other most closely. The palatine is widely separated from the vomer, so that no vomero- pwx TEXT-FIG. 3. pn pmx L sk Section through the left premaxilla of Cacosternum bottgeri. c.pn.i., cartilago praenasalis inferior; epm., epithelium of buccal cavity; pl.c., pulp cavity of tooth; pmx.l., left premaxilla; pmx.r., right premaxilla; pn.pmx.l., prenasal portion of left premaxilla; sk., socket of tooth; t., tooth. Other abbreviations as for previous figures. p a 1 a ti n e is formed as in Phrynomerus. The bone invests the ventral surface of the processua antorbitalis as in Rana and is of course quite edentulous. The septomaxillary is topographically confined to the lamina superior cristae intermediae and, as stated above, encapsules the recessus sacciformis. As in Phrynomerus, the bone terminates in front of the planum terminale of the cartilago obliqua. The relation of nasals and os en ceinture in Cacosternum was first remarked upon by Noble (1926), who pointed out that the

CRANIUM OF CACOSTERNUM 283 nasals are very small bones appearing on the sides of the large os en ceinture, which latter is prolonged into the tectum nasi. At their maximal width a nasal in Cacosternum covers about one-sixth of the tectum nasi, so that two-thirds of the latter are exposed; laterally the nasal does not articulate with the maxilla as in R a n a. This peculiarity is also met with in P h r y n o - merus, but the posterior bay in the nasal is absent. For the rest, the region of the olfactory capsule and associated structures differs very little from the well-known type met with in the frog. The absence of an enlarged eminentia olfactoria and associated cartilaginous axis encountered in Phrynomerus is probably to be interpreted as lack of specialization in this direction. OSSIFICATIONS IN THE SPHENETHMOID EEGION. In neither of the two species sectioned is the os en ceinture ossified so far forwards as one might expect. The first traces of the bone are met with in the nasal septum at the level of the choana and the posterior limits of the vomers. This is exactly the state of affairs in Ran a as well. The ossification spreads to the tectum nasi at the region of the anterior limits of the processus antorbitalis, but the ventral portion of the narial skeleton only begins to show signs of ossification towards the posterior region of the processus antorbitalis, which is not partially incorporated into the os en ceinture as in Ran a. These features are drawn in Text-fig. 4, a. In the region of the roots of the olfactory nerves (Text-fig. 4, b) the os en ceinture appears upon transverse section as a densely ossified trough with no traces of cartilage in its walls. In Phrynomerus the free dorsal ends of the trough persist as cartilage, which also constitutes the central portion of its bottom. More posteriorly these three tracts of cartilage also appear in Cacosternum ; it is interesting to note that the appearance of the mid-ventral tract indicates the presence of a posterior ventral notch in the os en ceinture. This is not represented in R a n a. The anterior division of the os en ceinture in C. bottgeri is similar in appearance to that of C. capense, but the posterior portion is like that of Phrynomerus, inasmuch as no complete

284 C. G. S. DB VILLIERS ossification occurs, the three tracts of cartilage persisting as in Phrynomerus. It will be convenient to leave the discussion of the other cartilage bones to a later stage and to consider forthwith the fronto-parietal bones and the so-called fronto-parietal TEXT-FIG. 4 6. o e c en t 4a: Transverse section of the skull ofc. capense cut posterior to the antorbital process. 4 b: Transverse section through the region of the antorbital process of C.capense. b.op.7i., branch of the optic nerve; cn.t., connective tissue; frp., fronto-parietal; mx., maxilla; o.e.c, os en ceinture; op.n., optic nerve; pal., palatine; pr.aob., processus antorbitalis. Other abbreviations as for previous figures. fontanelle. Hewitt (1911) remarked upon the weak ossification of the fronto-parietals and called attention to the presence of the fontanelle in C. bottgeri. These two features are also met with in Cacosternum capense. The tips of the fronto-parietals are figured in Text-fig. 4, a; they are imbedded in tough connective tissue which forms the roof of the brain case. The bones do not increase greatly in size in the interorbital region but are always joined by a wide expanse of connective

CRANIUM OF CACOSTBRNUM 285 tissue as in Phrynomerus; they attain to their maximal size in the region of the anterior limits of the otic capsule, but even so they do not touch in the middle line, although they are considerably more strongly developed than in Phrynomerus. Conditions in C. bottgeri correspond very closely to those in C. capense, but the fronto-parietals are even smaller and hardly project from the otic capsule into the connective-tissue cranial roof. This is almost exactly similar to the Phrynomerus pattern. The p r o - o t i c: The optic foramen is bounded anteriorly by cartilage, but its postero-dorsal margin is formed by the p r o - otic bone. Exactly the same conditions prevail in C. bottgeri. All ossification of the septum dividing the cavity of the otic capsule from the brain-cavity disappears before the foramen acusticum and the foramen endolymphaticum are sectioned, whereas in Bana and Phrynomerus the anterior boundary of the former foramen is part of the pro-ootic. Posterior to the two foramina mentioned, the otic capsule is almost entirely cartilaginous. Weak ossification of the capsule was also a feature of Phrynomerus, in which the supra- and infracristal ossifications were likewise absent. C. bottgeri has the same type of otic capsule as C. capense, except that in the latter species the posterior division contains much more persistent cartilage than in the former. The cranial roof consists of fibrous connective tissue in the orbital region; towards the anterior boundary of the otic capsule the tip of the taenia tecti medialis is sectioned; when this spreads laterally and passes into what should represent the taenia tecti transversalis, the cartilaginous cranial roof shows no more foramina, so that as in Phrynomerus and Arthroleptella (de Villiers, 1929) the parietal foramen is absent and the taenia tecti transversalis is confluent with the tectum synoticum. It is of course also possible to consider this state of affairs to be due to the absence of the transverse taenia; but comparative data are not available; Gaupp's original description referred to the European E a n a fusca. A remarkable feature of the tectum synoticum of Oacosternum is its weak development towards the foramen magnum, where it possesses a deep

286 C. G. S. DB VILLIEES notch filled with connective tissue (Text-fig. 5). The exoccipitals are comparatively small; ventrally between them a portion of the planum basale persists as cartilage, which, however, shows no traces of a notochord. The p a r a s p h e n o i d stretches from themid-a^entral division of the os en ceinture to the ventral occipital region, is a daggershaped bone as in Eana and, as in that genus, possesses two posterior notches separating the longitudinal portion from the TEXT-FIG. 5. t sn en t ot Transverse section through the posterior limits of the tectum synoticum of C. capense. br., brain; oi.c, otic capsule; t.sn., tectum synoticuni. Other abbreviations as for previous figures. lateral ones, which attain their greatest size at the level of the posterior region of the otic capsule. The cartilaginous cranial roof lacks a taenia transversalis in C. bdttgeri also. The exoccipitals are comparatively large and the tectum synoticum has no bay dorsal to the foramen magnum. The parasphenoid is dagger-shaped, but posteriorly the longitudinal portion is not marked off from the lateral ones by deep notches. The morphology of the quadratomaxillary and paraquadrate was reinvestigated by the author in Arthroleptella (1929) and Phrynomerus (Quart. Journ. Micr. Sci., vol. 73. Cacosternum is most remarkable in possessing a totally unossified quadrate cartilage, whereas in other Anuran genera, including Eana, as described by Gaupp, the quadrate ossifies perichondrally and enchondrally and fuses with the

CRANIUM OF CACOSTERNUM 287 quadratomaxillary. The relations of the above-mentioned bones in Cacosternum are sketched in Text-fig. 6, a and b; a represents a transverse section of the suspensorial region of the head and passes through the Eustachian tube. The quadratomaxillary occupies a position external to the chewing muscles, but TEXT-FIG. 6 a AND 6. an ty Me Transverse sections through the articular region of the skull of C. capense to show the relations of the quadratomaxillary to the rest of the skeleton (muscles are ruled), an.ty., annulus tympanicus; dmar., dermarticular; EX., Eustachian tube; hy., hyale; p.pt., processus pterygoideus; p.qd., pars quadrata processus pterygoidei; prq., paraquadrate; ptg., pterygoid; qmx., quadratomaxillary. Other abbreviations as for previous figures. more posteriorly it gradually shifts in between these latter, and appears as an investing bone of the quadrate cartilage. It is, however, easily distinguishable from dermal bones by the fact that it is not separated from the quadrate cartilage by a connective-tissue lamella (Text-fig. 6, b). The bone shoots a few diminutive rootlets into the invested cartilage, but the characteristic cartilaginous structure of the latter is in no way affected. Exactly the same conditions prevail in C. bottgeri. The paraquadrate is a comparatively small bone in

288 C. G. S. DE VILLIERS Cacosternum and is first encountered in sections at the level of the closure of the optic foramen; it then lies external to the temporal muscle and does not articulate anteriorly with any bone as it does in Pyxicephalus adspersus. Upon reaching the transitional region between the processus oticus and the crista parotica, the paraquadrate forms a bony sheath TEXT-FIG. 7 a, b, c, d..prq vat Consecutive transverse sections through the tympanic region of C- capense. (It is impossible to locate definitely the common boundary of the crista parotica and the processus oticus. This region is therefore labelled trn., transitional cartilage.) d.at., dorsal portion of annulus; epl., extraplectral enlargement of the pars externa plectri; m.e., middle ear; p.ot., procesaua oticus; t.m., tympanic membrane; v.at., ventral portion of annulus. Other abbreviations as for previous figures. for the latter. For a considerable stretch no connective tissue is present to separate the bone from the cartilage, which therefore simulates perichondral cristal ossification. Enchondral ossification of the cartilage is absent. The paraquadrate is typically triradiate as in all Anura known to me; the two transverse posterior rays are normal investing bones of the quadrate cartilage and crista parotica. The relations of the paraquadrate, crista parotica, processus oticus and annulus tympanicus are sketched in Text-fig. 7, a, b, c, d. C. b 611 g e r i

CRANIUM OF CACOSTERNUM 289 does not differ very much from 0. capense, but a stage corresponding to Text-fig. 7, b, is missed out, so that the paraquadrate forms a groove, and not a sheath for the transitional cartilage. The pterygoid invests the dorsal and inner surfaces of the processus pterygoideus and possesses in its dorsal portion a welldeveloped marrow cavity. On the whole it may justly be maintained that the pterygoid is much more strongly developed dorsal to the processus pterygoideus than in Ban a. The anterior portion of the processus basalis is entirely ensheathed by the pterygoid, which is not in this region separated from the cartilage by connective tissue (Text-fig. 8, a). The two anteriorly directed horns of the processus basalis are sectioned in Textfig. 8, b; the tip of the lower touches the pterygoid bone, but its upper division and the dorsal anterior horn of the processus basalis are again separated from the investing pterygoid by connective tissue. Text-fig. 8, b, shows the junction of the processus oticus and crista parotica, and Text-fig. 8, c, marks the fusion of this cartilage mass with the processus basalis. Posteriorly the pterygoid persists as a diminished bone on the inner surface of the pars quadrata. As far as comparative data are available, these conditions agree topographically quite well with those in Kana. C. bottgeri agrees with C. capense even in the minutest details, the only point of difference being the circumstance that the lower anterior hom of the processus basalis is not entirely ensheathed by the pterygoid. THE AUDITORY APPARATUS Typical transverse sections through the anterior region of the annulus tympanicus and middle ear are shown in Text-fig. 7, a and b; since the annulus has the form of a disk with a central perforation, it will appear in transverse section as two cartilages, the ventral of which is the larger on account of the eccentricity of the perforation referred to above. The pars externa plectri begins to appear in Text-fig. 7, c as a longish cartilage imbedded in the mesodermal portion of the tympanic membrane. Such an extraplectral enlargement of the pars externa was also encountered in Phrynomerus. Text-fig. 8, a, b, and c,

Consecutive transverse sections through the tympanic region of C. capense to show its relations with the processes of the pars quadrata. cr.p., crista parotica; l.a.h.pr.b., lower anterior horn of processus basalis; u.a.h.pr.b., upper anterior horn of processus basalis. Other abbreviations as for previous figures. h prb

CRANIUM OP CACOSTERNUM 291 represent further sections of the auditory apparatus, not differing essentially from Text-fig. 7, c, so far as the plectral and annular arrangements are concerned. It will, however, be noticed that the pars externa plectri shifts gradually to a more dorsal position, and in Text-fig. 9, a, its dorsal portion shows TEXT-FIG. 9 a AND b. pmp P r 1 prtj P P b Consecutive transverse sections through the annular region of C. capense. j-v., jugular vein; p.e.p., pars externa plectri; p.m.p., pars media plectri; vd.at., ventral and dorsal portions of annulus passing over into each other at the posterior limits of the annulus; VII, seventh cranial nerve. Other abbreviations as for previous figures. perichondral and enchondral ossification, although a marrow cavity is not developed. The section is in fact cut through the transition from pars externa to pars media plectri, which is the only part of the plectral apparatus to ossify in A n u r a. It will be noticed further that the dorsal portion of the annulus has now disappeared from section, although it subsequently reappears. The annulus is in fact not a complete ring as in R an a, but is discontinuous dorsally as in P h r y n o m e r u s. The open, crescentic form of the annulus seems to be quite common in Anura and particularly in Brevicipitid-Engyostomatidae. Since the region of the pars media plectri is now reached, atten-

292 C. G. S. DE VILLIERS tion should be called to the absence of a pars ascendens plectri effecting a communication between the pars externa plectri and the crista parotica. This cartilage was described by Gaupp for Bana fusca, but I have never come across it in South African frogs and toads, so that its universal occurrence in Anura should not be assumed. Text-fig. 9, b, represents a section through the body of the pars media, which is seen to consist of a perichondral osseous sheath and a central, permanently cartilaginous core. As the pars media approaches the otic capsule, the ossification ceases to be exclusively peripheral and osseous substance may also be seen traversing the cartilage as in Text-fig. 10, a. The pars media is overlain by the jugular vein and the hyomandibular branch of the seventh nerve. Textfig. 10, b, represents the point of maximal fusion of the pars media with the otic capsule; it will be noticed that the perichondral ossification is no longer sheath-like, but disappears towards the inner aspect of the pars media, where it is in cartilaginous continuity with the otic capsule. For a short distance the sheath-like pattern is represented again, Text-fig. 10, c, and all traces of ossification disappear at the level of the anterior boundary of the operculum (Text-fig. 10, d), where the plectrum is represented by the pars interna. The plectral apparatus is therefore a continuous cartilaginous structure, but in the region of the pars media it is perichondrally ossified. The operculum has the bowl-like shape typical of Anura and is fused to the dorsal margin of the fenestra ovalis upon the disappearance of the pars interna plectri. The fossa fenestrae ovalis is large as in Phrynomerus. The operculum possesses a well-developed musculus opercularis, which is not, however, attached to a special cartilaginous tuberosity of the operculum. C. bottgeri has an auditory apparatus which differs quite considerably from that of C. capense. The annulus tympanicus is very widely open dorsally, but the plectrum has much the same histological and anatomical structure as in the larger species. The operculum is, however, relatively much larger in the smaller species and in its anterior division it is straightened out, and more Avatch-glass shaped than bowl shaped. The consequence is that the fossa fenestrae ovalis and the ductus

CRANIUM OF CACOSTERNUM 293 fenestrae vestibuli both appear to be smaller than in the larger species, although the increase in height compensates for the TEXT-FIG. 10 a, b, c, d. Consecutive transverse sections through the sound conducting apparatus of C. capense. opm., operculum; pb.p.m.p., posterior boundary of the pars media; p.i.p., pars interna plectri; IX, ninth cranial nerve. Other abbreviations as for previous figures. decrease in breadth. Posteriorly the operculum acquires the usual bowl-like shape again (Text-fig. 11). The musculus opercularis is comparatively enormous and joined by means of an NO. 294 u

294 C. G. S. DE VILLIERS aponeurosis to the musculus levator scapulae. The opercular muscle is attached to the operculum, which develops a special tubercle or tuberosity for this purpose: the tubercle is absent in C* capense although the muscle is present. According to TEXT-FIG. 11. spsc m L tb- Transverse section through the opercular region of the ear of C. bottgeri. f.f.o., fossa fenestrae ovalis; f.o., f enestra ovalis; m.l.s., musoulus levator scapulae; m.op., musculus opercularis; spsc, suprascapula; tb., tubercle for opercular muscle. Other abbreviations as for previous figures. Versluys (1924) an operculum was evolved as a response to terrestrial life; C. bottgeri should therefore be more terrestrial than C. capense. Of the latter species I may definitely accentuate the aquatic habits and the same is probably true of C. bottgeri. The hyoid apparatus was imbedded and sectioned with

CRANIUM OF CACOSTBRNUM 295 the skull, but was also dissected out to make sure of the gross anatomy. It was found to be so similar to that of E a n a, that it was not considered necessary to give a drawing of it. The bay between the manubria is a little shallower than in E a n a, the processes alares taper slightly more towards their posterior tips, and the processes thyroidei are comparatively longer, so that they are still encountered in section towards the posterior limits of the coracosternum. This arrangement is, however, quite secondary and is due to the forward position of the shoulder girdle in the genus. The hyale is fused to the base of the otic capsule behind the processus basalis. The processes anteriores are absent. The hyoid apparatus in C. bottgeri is very similar to that of C. capense, but the thyroids are not prolonged posteriorly to such an extent; possibly the pectoral girdle is not so considerably shifted forwards as in the larger species. The hyale, moreover, is not fused with the otic capsule, but merely articulates with it, in a fossa below the lower lip of the fenestra ovalis (Text-fig. 12). It should be further noted that the thyroid processes in both species possess well-developed marrow cavities but end posteriorly in long cartilaginous tips. The loaver jaw lacks all traces of the peculiar modifications encountered in Phrynomerus, and agrees in all essential details with that of E a n a. The mento-mandibular is, however, ossified perichondrally only in the form of a sheath to Meckel's cartilage, Avhereas in Eana enchondral ossification is also initiated. The same conditions prevail in C. bottgeri as in C. capense. COMPARISON OF THE SKULL OF CACOSTBRNUM WITH THAT OF PHRYNOMERUS. The reasons for investigating the skulls of Phrynomerus and Cacosternum were (1) to give some new histological details of the A n u r a n skull in general, (2) to ascertain whether the two Brevicipitid (Engyostomatid) genera are allied, as the absence of a procoracoid might lead one to expect, and (3) to attempt to enumerate some cranial characters by which these two genera are both distinguishable from Eanids; in other words, to attempt to discover some specific Brevicipitid-Engyo- U2

296 C. G. S. DE VILLIEBS stomatid cranial characters. It will probably be most convenient to tabulate the evidence under the headings of the various cranial unities. A. The vestibule of the olfactory organ. The vestibular ' Wiilste' described by Gaupp for Eana are developed in both TEXT-FIG. 12. Transverse section through the otic region of the skull of C. b 611 - geri to show the method of articulation of the hyale. f-hy., fossa in otic capsule for the hyale; j?.&., processus basalis. Other abbreviations as for previous figures. species of Cacosternum, and in Phrynomerus. The plica obliqua is suspended in Eana from the tectum nasi; in the Brevicipitids the plica is short and blunt and suspended from the cartilago obliqua. B. The prenasal cartilages are present in the Brevicipitids as in Sana, C. The premaxilla is toothed in C. bottgeri and C. cap ens e but edentulous in C. namaquense (Werner, 1910)

CRANIUM OF CACOSTERNUM 297 and Phrynomerus. [Noble has repeatedly maintained that no great systematic value can be attached to the absence or presence of teeth. The South African Heleophryne would, according to Noble, have to be considered as a 'toothed Bufonid']. D. The maxilla is separated from the outer palatal squame of the premaxilla in Phrynomerus only. E. The eminentia olf actoria is very high in Phrynomerus, normal in Cacosternum and Eana. ~F. The vomer in Phrynomerus is considerably prolonged into the choana; in Eana and Cacosternum such dorsoventral enlargement of the bone does not occur. G. The palatine in Phrynomerus fuses with the vomer to form a vomero-palatine, but in Eana and Cacosternum the two bones are wide apart. H. The recessus sacciformis is large in Eana, diminished in Cacosternum, and purely vestigial in Phrynomerus. I. The nasals in Phrynomerus and Eana leave little of the anterior portion of the os en ceinture exposed, whereas in the Cacosternum-Anhydrophryne group the os en ceinture is prolonged anteriorly between the nasals, which are small and laterally placed. J. Phrynomerus has two prechoanal sacs, Cacosternum an unpaired one, whereas in Eana the structure is absent. The organ in the Brevicipitids is to be derived from a prechoanal fusion of the ' Gaumenleisten', and not improbably represents buccal vestiges of Jacobson's organ, which is supposed to be reduced to the recessus medialis in Anura. K. The os en ceinture is paired in Phrynomerus, and notched posteriorly in Cacosternum, whereas in Eana it is girdle-like as its name implies. The lateral trabecular derivates bounding the fenestra parieto-frontalis are cartilaginous in Phrynomerus and C. bottgeri but ossified in C. capense and Eana. L. The parieto-frontal bones are mere lateral tracts in Phrynomerus and Cacosternurn, whereas in Eana they are broad and closely approximated mid-dorsally. Whereas the dorsal cranial roof in the latter genus is therefore bony, it

298 C. G. S. DE VILLIERS is formed by thick, denselyfibrousconnective tissue joining the two parieto-frontals in the two Brevicipitid genera. M. The optic and pro-otic foramina are bounded anteriorly and posteriorly by bone in Phrynomerus. In Cacosternum the Eanid type prevails, in which the anterior margin of the optic foramen is not reached by the os en ceinture. N. The otic capsule in Phrynomerus lacks the supra- and sub-cristal ossifications. Weak ossification of the otic capsule is also a feature of Cacosternum, particularly of C. capense. 0. The transverse taenia of the Eanid cranial roof is absent in both Brevicipitid genera and also in Arthroleptella. P. The occipital region of the adult Anuran skull is remarkably constant in structure. The exoccipitals are joined ventrally by the persistent planum basale. All traces of an intracranial notochord disappear. In C. capense the foramen magnum possesses a deep dorsal notch. Q. The parasphenoid shows no appreciable variation. E. The paraquadrate is comparatively short in Phrynomerus and inclined to fuse with or invade the crista, which is ossified. In Cacosternum the crista is cartilaginous, encapsuled, but not invaded by the paraquadrate although the separating connective-tissue lamella disappears. S. The pterygoid invades and fuses with the ossified periphery of the processus basalis in Phrynomerus. This does not take place in Cacosternum, in which the separating connective tissue merely disappears. No histological details are available for Ran a. T. The quadratomaxillary invades the articular division of the quadrate process in Ran a (Gaupp), Arthroleptella, and Phrynomerus. In Cacosternum the quadrate cartilage is not ossified, but merely overlain directly by the quadratomaxillary. U. The two Brevicipitid genera differ from Rana in the absence of a processus ascendens plectri and in the incompleteness of the annulus tympanicus. They share these features with Arthroleptella. The middle ear and Eustachian tube are present in all the three non-eanid genera mentioned as well as in Rana. The plectrum and operculum are Eanid, except for

CRANIUM OF CACOSTERNUM 299 the following differences. The operculum is particularly large and shallow in G. bottgeri, and develops a relatively large musculus opercularis attached to a special tuberosity. I have not seen an opercular muscle in Phrynomerus. The pars externa plectri is enlarged to form an extraplectral imbedded in the mesodermal layer of the tympanic membrane in the two Brevicipitid genera, but not in Ran a. The pars media plectri lies ventral to the crista in Phrynomerus, but in Cacostern u m is closely applied to the ventral lip of the fenestra ovalis as in Ran a. V. The hyalia are fused with the otic region of the skull except in C. bottgeri. The Brevicipitid genera lack the anterior processes of the hyoid apparatus, present in Rana. The alar processes of Cacosternum are like those of Rana, but they are enlarged and blade-like in Phrynomerus; in the latter genus, as well as in Cacosternum, the thyroid processes are met with in the pectoral region, a condition which may be due to the pectoral girdle being shifted forwards. The overlapping is most pronounced in C. capense. W. In the lower jaw the mento-mandibular is exclusively perichondrally ossified in Cacosternum. In Rana and Phrynomerus enchondral ossification also takes place. The latter genus has a remarkably specialized mental region, with backwardly directed diverticula of Meckel's cartilage and a modified gular musculature. These features are absent in Cacosternum and Rana. It is always dangerous to base affinities on the study of one particular system of organs only; it is therefore necessary to state explicitly that no final conclusion regarding the mutual relationships of Phrynomerus and Cacosternum can be arrived at by a comparison of their cephalic skeletons, but that the results embodied in this paper and the previous one on Phrynomerus may aid the solution of the problem. The whole question of the validity of the Brevicipitidae as an autonomous family of the Firmisternia, and of the monophyletic origin of the South African Brevicipitid genera, will be fully discussed in a dissertation by one of my students. I shall therefore restrict myself to the cephalic skeleton.

300 C. G. S. DE VILLTERS The most important feature which Phrynomerus and Cacosternum have in common is the extreme reduction of the parieto-frontal bones, and the connective-tissue-like nature of the pretectal cranial roof. Whether the feature is met with in any other Brevicipitids, I cannot say. The otic region of both genera is characterized by the incompleteness of the annulus tympanicus and the enlargement of the pars externa plectri to form an extraplectral cartilage embedded in the tympanic membrane. Moreover, the pars ascendens, effecting a junction of the pars externa plectri and the crista parotica, is not developed. But these features, as well as the absence of a taenia tecti transversalis, are not exclusively Brevicipitid, as they have been proved (de Villiers, 1929) also to occur in the Eanid Arthroleptella. The hyoid apparatus lacks the anterior processes in Phrynomerus and Cacosternum, but again comparative data for other genera are not available, and the processes are also absent in an arciferous toad, Bufo angusticeps. The relations of membrane bones like the paraquadrate, quadrato-maxillary and pterygoid, to the cartilages they invest are of great osteogenic interest but are probably not of systematic importance. Many features of the skull of Phrynomerus are indicative of extreme specialization and cannot be used for comparison with Cacosternum; such are, e.g., the presence of a vomeropalatine, the absence of a recessus sacciformis, the enlarged eminentia olfactoria and the intrachoanal elongation of the vomer. But these features probably represent individual characteristics of Phrynomerus and probably do not represent specifically Brevicipitid modifications. As far as comparative data are available, the skull of Anura, with the exception of Aglossa, which have become secondarily aquatic, is remarkably constant in structure, and great deviation from the Eanid type should not be expected and has not been recorded in the extant literature on the subject. But it is hoped that the account of the anatomy of the microtomized skulls of Phrynomerus and Cacosternum may help to revive interest in the osteology of one of the most primitive groups of terrestrial vertebrates; their phylogeny may be obscure, but as

CRANIUM OF CACOSTBENUM 301 a group they are at any rate end-products of evolution and their genealogy is not obscured by phylogenetic neoteny, as is the case with the Urodeles. Stellenbosoh, April, 1930. LITERATURE CITED. For a more comprehensive list, the reader is referred to the Author's paper on. Phrynomerus ; the following are additional references: Andersson, L. G. (1911). "Reptiles, Batrachians, and Fishes collected by the Swedish Expedition to British East Africa. 2. Batrachians", 'Svenska Vetensk. Akad. Handl,', vol. xlvii. Boulenger, G. A. (1882). 'Catalogue of the Batrachia salientia sive ecaudata in the collection of the British Museum', 2nd ed. (1887). "Description of new Reptiles and Batrachians in the British Museum", Part III, 'An. and Mag. Nat. Hist.', vol. xx. (1906-9). "A revised list of the South African Reptiles and Batrachians", 'An. S. Afr. Mus.', vol. v. Gaupp, E. (1904). 'Die Anatomie des Frosches', Bd. 3. Hewitt, J. (1911). "A Key to the species of the South African Batrachia together with some notes on the specific characters and a synopsis of the known facts of their distribution", 'Rec. Albany Mus.', vol. ii. (1919). "Anhydrophryne rattrayi, a remarkable new frog from Cape Colony", 'Rec. Albany Mus.', vol. iii. (1925). "On some new species of Reptiles and Amphibians from South Africa", ibid. (1926). "Descriptions of some new species of Batrachians and Lizards from South Africa", 'An. Natal Mus.', vol. v. Krause, R. (1923). "Mikroskopische Anatomie der Wirbeltiere in Einzeldarstellungen", Teil III, 'Amphibien'. Berlin, de Gruyter. Methuen, P. (1913). "The Percy Sladen Memorial Expedition to Great Namaqualand, 1912-1913: Zoology", 'An. Trvl. Mus.', vol. iv. Nieden, F. (1926). "Anura II: Engyostomatidae" : 49ste Lieferung des 'Tierreichs'. Berlin, de Gruyter. Noble, G. K. (1922). "The Phylogeny of the Salientia: I. The osteology and the thigh musculature; their bearing on classification and phylogeny", 'Bui. Am. Mus. Nat. Hist.', vol. xlvi. (1924). "Contributions to the herpetology of the Belgian Congo, based on the collection of the American Museum Congo Expedition, 1909-1915", 'Bui. Am. Mus. Nat. Hist.', vol. xlix. and Parker, H. W. (1926). "A synopsis of the Brevicipitid toads of Madagascar", 'Am. Mus. Novitas', No. 232. (1926 a). "The Importance of larval characters in the classification of South African Salientia", ibid., No. 237.

802 C. G. S. DB VILLIBRS Noble, G. K. (1927). "The value of life-history data in the study of the evolution of Amphibia", 'An. N. Yk. Ac. So.', vol. xxx. Power, J. H. (1927). "Some tadpoles from Griqualand West", 'Trans. Roy. Soc. S. Afr.', vol. xiv. Versluys, J. en anderen (1924). 'Leerboek der vergelijkende Ontleedkunde van de Vertebraten', Deel II. Utrecht, Oosthoek. Villiers, C. G. S. de (1929). "The Development of a Species of Arthroleptella from Jonkershoek, Stellenbosch", 'S. Afr. Journ. Sc.', vol. xxvi. (1929 a). "Some observations on the breeding habits of the Anura of the Stellenbosch Flats, in particular of Cacosternum capense and Bufo angusticeps", 'An. Trvl. Mus.', vol. xiii. (1930). " On the Cranial Characters of the South African Brevicipitid, Phrynomerus bifasciatus", 'Quar. Journ. Micr. Sci.', vol. 73. Werner, F. (1910). 'Eeptilia et Amphibia aus den zoologischen und anthropologischen Ergebnissen einer Forschungsreise im westlichen und zentralen Siidafrika', Bd. IV.