Cretaceous Therian Tarsals and the Metatherian-Eutherian Dichotomy

Abstract

Diverse metatherian and eutherian tarsal remains from the Late Cretaceous (middle-late Turonian) Bissekty Formation, Kyzylkum Desert, Uzbekistan (ca 90 MYA) are described. Their functional and taxonomic properties, along with those of other tarsal evidence, led to a reassessment of polarity hypotheses of therian, metatherian, and eutherian cruropedal attributes, and the consequences of this for phylogeny of taxa. There are calcaneal remains of several types of marsupials, and a single astragalus that probably belongs to one of these. This represents greater taxonomic diversity than the dental record suggests. Exceptionally large and distally extending peroneal processes, and small and steeply angled calcaneocuboid articulations facing mediodistally, as seen in the Early Cretaceous Sinodelphys and other Cretaceous and Paleogene taxa, attest not only to the metatherian status of these specimens, but also to the retention of many ancestral therian features, even more so than in both the Tiupampa and Itaborai marsupials of the South American Paleocene (both the calcanea and the astragalus suggest therian traits that are decidedly unlike those of symmetrodonts). Calcanea allocated to the deltatherioid species at Bissekty testify unequivocally to their metatherian affinity. The morphology of the best represented sample of eutherian calcanea from Bissekty, presumably of a number of zhelestid species, appears to be more derived than that of the Late Cretaceous/Paleocene Protungulatum in having a much more reduced peroneal process and a calcaneocuboid articulation that faces distally, oriented nearly at a 90° angle to the long axis of the calcaneus. In fact, this distally facing facet, common in later eutherians (except for lineages in the Paleogene record, and various Carnivora), may not be diagnostic of either the protoeutherian, or even of the protoplacentalian, in spite of its presence in Eomaia. Many putatively “basal” lineages have derived characters, hence such outgroups should not be considered the unequivocal repositories of only ancestral character states.