The ontogeny of the cranium of Dipnoi is restudied. The investigation especially refers to the basic components of the dipnoan cranium and several functional and developmental aspects of the structure of the larval skull ofNeoceratodus. There are fundamental differences even in the early development and composition of the chondrocranium ofNeoceratodus and Lepidosirenidae. This result, and comparison with several osteichthyans and Tetrapoda, requires a reinterpretation of the components of the dipnoan skull base. The pterygoid processes are not reduced, but incorporated into the cranial base early in ontogeny. The characteristic elongate trabecular rods, which in Gnathostomata usually bridge the ethmoidal plate and the orbito-temporal base of the chondrocranium, are much delayed in development inNeoceratodus, or even seem absent inLepidosiren andProtopterus. Accordingly, in Dipnoi no typical basitrabecular junction is formed in early ontogeny. Instead, the pars quadrata is fused to the mesodermal basis cranii posteriorly. InNeoceratodus a mesially directed basal process of the palatoquadrate is recognizable, which topographically corresponds to the basal process of Urodela and the pseudobasal process of anuran larvae. The ethmosphenoid region of the dipnoan skull also develops quite differently. In Lepidosirenidae, the palatoquadrates are interconnected anteriorly by a distinct commissura palatoquadrati, whereas inNeoceratodus a continuous planum ethmoidale (“trabecular plate”) is formed. The primary embryonic quadrato-trabecular connection persists as a commissura quadratocranialis anterior below the foramen opticum, at the root of the ectethmoid process. The formation of the skull base in living Amphibia appears to provide the best model for comparison, though it is difficult to propose any undisputable shared derived character states of the cranium of Dipnoi and Tetrapoda beyond this similarity. A similar difficulty presents the phylogenetic interpretation of the hyoid arch. In contrast to the absence of any dorsal hyoid arch elements inLepidosiren, the small hyomandibula ofNeoceratodus is surprisingly complete. In larvae it consists of a laterohyale, an epihyal part, and a processus symplecticus. A stylohyal cartilage is present, which forms rather late in ontogeny. The major chondral components of the hyoid arch are thus comparable to those of living Actinopterygii, except that a distinct symplecticum is not separated off, the components are relatively smaller, and they do not ossify. In view of the early-immobilized palatoquadrate, the hyomandibula ofNeoceratodus has no suspensorial function, but represents part of an opercular hinge and opening mechanism. The hamuloquadrate knob at the posterior face of the quadrate body is comparable to the processus hyoideus in some Urodela. It provides a pivoting joint for the ceratohyale, and therefore functions in buccal expansion. The closed spiracular canals include mechanoreceptive lateral line organs, which probably represent proprioreceptive organs for adjustment of mandibular, hyoid, and opercular movements. It is concluded that considerable differences between the skull architecture of Dipnoi and other Osteognathostomata (Teleostomi) can be assigned to the fact that palatoquadrate and trabecular anlagen fail to separate, resulting in a dramatic and highly adaptive change of palatoquadrate development in early ontogeny. Though these differences include several characters that suggest a plagiostomate condition of the jaw apparatus, this can be explained as a secondary acquisition. The multitude of retained plesiomorphies observed in the cranium of Dipnoi do not exclude a sister group-relationship to Tetrapoda. However, the ancestral osteognathostome suspensorial pattern still presents a problem of interpretation, for we lack a detailed survey of the development and significance of different quadrato-neurocranial connections.
Read full abstract