A hypothesis is presented to explain the origin of a fundamental difference between the nervous systems of tardigrades and arthropods and to show how the hypostome of arthropods evolved. In arthropods the stomodeal nervous system is fused to a more posterior pair of trunk ganglia to form the tritocerebrum, whereas in tardigrades the two systems are not comparably linked. It is proposed that this difference resulted from a key transformation of the head in the stem group to Euarthropoda. The hypothesis specifies, based on a presumed plesiomorphic condition found in onychophorans, that an appendage-bearing metamere gave rise to the mouth cone of onychophorans, tardigrades and stem lineage arthropods. The metamere provided jaws and a circumoral ring of plates or sensory structures that in certain stem lineage arthropods is manifested as a "Peytoia." A simple model is developed to illustrate how in the stem group to Euarthropoda the mouth cone or "Peytoia" rotated, elongated in a posterior direction and became sclerotized on its anterior surface to form the hypostome. The hypothesis postulates further that backward rotation of the mouth cone brought the stomodeal system and a posterior pair of ganglia into juxtaposition. Thus the hypothesis offers a plausible explanation for the origin of the tritocerebrum and shows why the hypostome appears to be innervated by a neuromere of a more posterior segment. In general the model refines a construct introduced by LANKESTER (1904) that one of the distinctive features of arthropods is the formation of preoral segments by back migration of the mouth. The dramatic shift in orientation of the mouth in the stem group of arthropods appears to have been correlated with a change in feeding strategy. It is suggested that the complex evolutionary history of the hypostome, including derivation of the mouth cone or "Peytoia " from an appendage bearing segment and its subsequent rotation, elongation and scleroitization proposed here is concordant with this view. The hypothesis agrees with proposals that the hypostome is segmental but contradicts postulations that it formed from medially fused appendages. The model also reinterprets the absence of typical engrailed expression in the hypostome and predicts that remnants of a "labral" segment are confined to the areas innervated by the stomodeal system - regions previously considered to be nonsegmental. Finally, the origin of several extant taxa may be understood in terms of the transformation of the mouth cone. The presence of a biradially symmetrical cone in tardigrades suggests that they derived from the paraphyletic "Peyroia" group, even though they diverge basal to them in a recent cladistic analysis using a larger set of morphological characters (DEWEL & DEWEL 1997). Possession of a mouth cone argues also that the clade containing Tardigrada arose prior to the evolution of the hypostome, a structure that nonetheless predates the radiation of the arthropod crown group. Furthermore, it can be inferred from the location of all postantennal limbs caudal to the hypostome in several fossil arachnates (CHEN et al. 1997; RAMSKOLD et al. 1997; HOU & RERGSTROM 1997), that these taxa lacked a tritocerebrum as-defined for extant arthropods. The same seems to be true for modern Chelicerata. Molecular data suggest that the cheliceral segment is homologous to Al (TELFORD & THOMAS 1998; DAMEN et al. 1998). Thus the stomodeal system appears to be linked anatomically to homologues of the Al and not A2 ganglia, and hence the tritocerebrum, sensu stricto, may not be a synapomorphy for the crown group of Arthropoda.
