Dorsally ([Fig. 3]), the premaxillae are not visible. The proportionally gigantic septomaxillae are visible anterior to the nasals. The moderate-sized nasals are separated medially and in broad contact with the ethmoid posteriorly. The palatine process of the nasal does not meet the frontal process of the maxilla. A large frontoparietal fontanelle is evident between the frontoparietals. The tegmen tympani are much reduced and maintain only cartilaginous contact with the posterior arms of the squamosals. The foramen magnum, occipital condyles, and exoccipitals show no unusual features. The pars facialis and frontal process of the maxilla are greatly reduced. The maxilla and premaxilla are articulated. The high, narrow alary processes of the premaxillae extend dorsally about two-thirds of the height of the snout. A cartilaginous internasal septum is illustrated ([Fig. 3]), but sectioning is necessary to determine the true nature and extent of this element.
Ventrally, the skull lacks palatines. The maxillae, premaxillae, and prevomers are edentate. The parasphenoid is large with relatively short, stout alary (lateral) processes. The sphenethmoid is extensive in ventral aspect and forms the major supporting structure in the anterior part of the skull. The pterygoid has a broad articulation with the maxilla, a tenuous contact with the squamosal, but is not attached to the proötic. The anterior (zygomatic) process of the squamosal is greatly reduced (only about one-third the length of the posterior process).
DISCUSSION
The skull of Allophryne is definitely non-hylid. Most of the post-cranial features do not help to clarify relationships. Allophryne shares several osteological features with the Dendrobatidae: T-shaped terminal phalanges, general cranial morphology and procoelus vertebrae. But, the dendrobatids possess firmisternal pectoral girdles and lack epicoracoidal horns. Also, no dendrobatid has intercalary elements in the digits. We are, therefore, left with a taxonomic enigma. In one or more characters generally regarded as important, Allophryne differs from all presently defined families of frogs. The Hylidae and Dendrobatidae are the only currently recognized families in which the genus might be placed.
The function and taxonomic importance of the large septomaxillae are unknown and are probably associated with the modification of the sphenethmoid-prevomer area. A more detailed study of the cranial osteology of Allophryne, especially the structural relationships of the sphenethmoid-prevomer area may elucidate the relationships of Allophryne.
The relationships of Allophryne cannot be understood without a re-analysis of some of the features used as major criteria in frog classification (the nature of an intercalated cartilage; the nature of the sternal complex; the relative value of cranial osteology; the vertebral structure; and the thigh musculature). Some of these features have been investigated by other workers, most notably Griffiths, but others have not and need re-examination. A re-analysis of some of the major criteria used in frog classification is in progress (Callison, Lynch, and Trueb) and upon completion of that study we think the relationships of Allophryne will become apparent.
A more comprehensive study of the cranial anatomy of certain hylids, leptodactylids, dendrobatids, and atelopodids along with that of Allophryne is needed to clarify the relationships of Allophryne, and might indicate that the recognition of a fifth family is necessary.
CONCLUSION
Among currently recognized families of frogs, Allophryne is least different from the Hylidae although it is our opinion that inclusion of this genus in the Hylidae probably represents an unnatural classification. However, the present evidence suggesting that Allophryne should be in another family is less convincing than evidence suggesting it should be in the Hylidae. We tentatively place Allophryne in the Hylidae.