Main

The morphology of the ankle can be used to evaluate these hypotheses. Ankle specializations are universally used to characterize Artiodactyla, and would provide an excellent test for the inclusion of whales in that order. Unfortunately, the few cetacean ankle bones known are too incomplete or too reduced to allow meaningful comparison with other mammals.

We have recently recovered fragmentary Eocene astragali (ankle bones) from pakicetid and ambulocetid cetaceans5 in Pakistan. We identified them as cetaceans because the deeply grooved trochlea resembles the partial astragalus of the holotype of Ambulocetus natans5, and the large size of the astragali matches only a few mammals known from the associated freshwater and marine faunas, in particular perissodactyls, anthracobunids and sirenians. Astragali for known representatives or relatives of these mammals do not match the morphology of the new bones. The ambulocetid astragalus was found in marine sediments.

Three articular facets of artiodactyl ankles are highly specialized and are important for the relationship of whales. First, the astragalar head of artiodactyls is trochleated, meaning that it is wide, gently concave mediolaterally, and strongly convex dorso-plantarly, with the axis of this convexity perpendicular to the median plane. Second, the sustentacular facet is rectangular and covers the entire posterior aspect of the astragalus. Finally, the ectal (posterior calcaneo-astragalar) facet is reduced and placed on the lateral side of the bone. The combination of these features is found in all artiodactyls but not in any other mammal.

The cetacean astragalar head (Fig. 1) is wide and nearly flat both mediolaterally and dorsoplantarly. This is unlike the condyle of mesonychians, but is also unlike the convex trochleated head of artiodactyls. This important feature, often cited as the main defining character of artiodactyls6, is inconsistent with the hypothesis that cetaceans should be included in the artiodactyls.

Figure 1: Astragali of: a-c, ?pakicetid cetacean shown in dorsal (a), plantar (b) and medial (c) view (H-GSP 97227, Locality 300); d, mesonychian (plantar view, Disaccus europaeus, MNHN Br 21 L); e, ?ambulocetid cetacean (lateral view, H-GSP 97113, Locality 9205, distal part missing); and f, artiodactyl (lateral view, Sus scrofa).
figure 1

Shown are ectal facet (ect.), sustentacular facet (sus.) and astragalar head (hd.). Left scale bar is for a-d, right scale bar is for e-f.

Full size image

The cetacean sustentacular facet resembles that of artiodactyls in being long, but unlike that of artiodactyls it is narrow. In primitive mammals, including mesonychians, the sustentacular facet is short and rounded. The cetacean ectal facet is strongly reduced and placed laterally as in artiodactyls, not plantarly as in mesonychians and other primitive mammals. This position of the ectal facet is highly derived7 and unique, occurring only in artiodactyls and these Eocene cetaceans. These features argue against close phylogenetic ties between cetaceans and mesonychians.

Our new ankle data do not unambiguously support either of the predominant hypotheses of cetacean relationships. Inclusion of Cetacea in Artiodactyla to the exclusion of mesonychians is consistent with the position of the ectal facet and the shape of the sustentacular facet. But the absence of a trochleated astragalar head argues against the inclusion of Cetacea in Artiodactyla, unless the flat head of the cetacean is interpreted as a secondary aquatic adaptation. Inclusion of Cetacea in Artiodactyla is also inconsistent with the derived similarities of the dentition and basicranium of cetaceans and mesonychians8. Sister-group relations between mesonychians and cetaceans are inconsistent with the derived similarities in the sustentacular and ectal facets between artiodactyls and cetaceans, both characters with little or no homoplasy in mammals. But, in any case, extensive convergence or reversals must have occurred in the dentition, basicranium and/or tarsus.