Caldwell 1995
provided a series of growth stages of the tarsus of Hovasaurus that chronicle the appearance of the ankle bones. Here is an animation of the same (Fig. 1).
Figure 1. From Caldwell 1995, an ontogenetic series showing the growth of the tarsus in the basal diapsid Hovasaurus. Scale bar = 1 cm. Since these specimens were not found as part of a family assemblage, there are some specimens that appear to diverge from others in terms of carpal element shapes. Plus, we see here a certain amount of individual variation, the driving force behind evolution. There is a large discontinuity between K and L due to a lack of fossils at that stage of growth. C = calcaneum. A = Astragalus. c = centralia.
Unfortunately
Caldwell was under the impression that the basal diapsid Hovasaurus was close to the ancestry of extant lepidosaurs. The large reptile tree (LRT, 1028 taxa) invalidates that hypothesis with the addition and inclusion of more taxa.
According to the LRT
Hovasaurus is a marine younginiform, basal to those diapsids that ultimately produced members of the Enaliosauria, a large clade of marine (new) archosauromorphs. Lepidosaurs had a separate origin going back to basalmost amniotes (= reptiles) like Gephyrostegus.
Figure 2 Tangasaurus, Hovasaurus and Thadeosaurus, three marine younginiformes, apparently have no scapula.
Hovasaurus is interesting
in that it developed a plesiosaur-style pectoral girdle without being directly related to plesiosaurs. Hovasaurus and Tangasaurus (Fig. 2) look like they are missing a scapula. In related Thadeosaurus the scapula has been reported only on juvenile taxa (gray box).
References
Caldwell MW 1995. Developmental constraints and limb evolution in Permian and extant lepidosauromorph diapsids.
The Pterosaur Heresies 