letters to nature 62 NATURE | VOL 414 | 1 NOVEMBER 2001 | www.nature.com 10. Krabill, W. et al. Greenland ice sheet: high-elevation balance and peripheral thinning. Science 289, 428±430 (2000). 11. Paterson, W. S. B. The Physics of Glaciers 3rd edn 45 (Elsevier Science, Oxford, 1994). Acknowledgements We thank J. Bamber, S. Ekholm, W. Krabill and E. Mosley-Thompson for providing information and unpublished data. Correspondence and requests for materials should be addressed to N.R. (e-mail: nr@oersted.dtu.dk). ................................................................. Late Cretaceous relatives of rabbits, rodents, and other extant eutherian mammals J. David Archibald*, Alexander O. Averianov² & Eric G. Ekdale* * Department of Biology, San Diego State University, San Diego, California 92182-4614, USA ² Zoological Institute, Russian Academy of Sciences, Universitetskaya nab. 1, Saint Petersburg 199034, Russia .............................................................................................................................................. Extant eutherian mammals and their most recent common ances- tor constitute the crown group Placentalia. This taxon, plus all extinct taxa that share a more recent common ancestor with placentals than they do with Metatheria (including marsupials), constitute Eutheria 1 . The oldest well documented eutherian- dominated fauna in the world is Dzharakuduk, Uzbekistan 2 . Among eutherians that it yields is Kulbeckia, an 85±90-Myr-old member of Zalambdalestidae (a family of Late Cretaceous Asian eutherians) 3 . This extends Zalambdalestidae back by some 10 million years from sites in the Gobi Desert, Mongolia 4 . A phylo- genetic analysis of well described Late Cretaceous eutherians strongly supports Zalambdalestidae, less strongly supports `Zhelestidae' (a Late Cretaceous clade related to Tertiary ungu- lates), but does not support Asioryctitheria (a group of Late Cretaceous Asian eutherians). A second analysis incorporating placentals from clades that include rodents (Tribosphenomys), lagomorphs (Mimotona) and archaic ungulates (Protungulatum and Oxyprimus) strongly supports Zalambdalestidae in a clade with Glires (rabbits, rodents and extinct relatives) and less strongly `Zhelestidae' within a clade that includes archaic ungu- lates (`condylarths'). This argues that some Late Cretaceous eutherians belong within the crown group Placentalia. The ages of these taxa are in line with molecularly based estimates of 64± 104 Myr ago (median 84 Myr ago) for the superordinal diversi®ca- tion of some placentals 5 , but provide no support for a Late Cretaceous diversi®cation of extant placental orders. Timing of the origin of major clades of extant placental mammals remains controversial. Some molecular studies place the origin of such clades (orders) far back into the Cretaceous 6 . Most palaeo- biologists argue that such clades originated in the Palaeocene or later 7,8 . Others argue that more inclusive extant placental clades (superordinal groupings) may not be discernible even in the Late Cretaceous 9 . Some proposed exceptions are Late Cretaceous representatives of Ungulatomorpha (`Zhelestidae' 10 ), lipotyphlans (Paranyctoides and Batodon 11 ) and Glires (Zalambdalestidae 11 ). The timing of the origin and relationships of Glires are among the most controversial subjects in studies of mammal evolution. Glires includes extinct early Tertiary taxa, as well as extant Rodentia (rodents) and Lagomorpha (rabbits and pikas). The monophyly of Glires has been repeatedly upheld 5 but the timing of the origin of Glires is contested; on the basis of molecular studies one of its members, Rodentia, has been argued to appear over 100 Myr ago 6,12 . This is almost as old as the oldest eutherian fossils 13±15 . The relationships of Glires to other eutherian taxa are also disputed; it has been argued that Glires may share a common ancestry with the Late Cretaceous Zalambdalestidae 11 . We present new data for the oldest known zalambdalestid, Kulbeckia kulbecke, from the 85±90-Myr-old Dzharakuduk fauna, Bissekty Formation, Uzbekistan 2 . Until recently, zalambdalestids were known almost exclusively from 75-Myr-old fossils from the Gobi Desert (J. R. Wible, M. J. Novacek & G. W. Rougier, manu- script in preparation; and ref. 16). Kulbeckia was named in 1993, on the basis of a few teeth 17 . Possible af®nities with zalambdalestids were not recognized until 1997 (ref. 3). During ®eld seasons at Dzharakuduk (1997±2000), over 40 additional specimens were recovered including teeth, dentaries, petrosals, a partial skull, and postcrania (Figs 1, 2). Similarities of zalambdalestids and Glires warrant further consideration. The hallmark of Glires is its enlarged, medially placed pair of incisors, which are procumbent, have enamel restricted to the more ventro- or dorsolabial margin of the tooth, are open-rooted, ever- growing, and have Hunter±Schreger bands 18 . Other mammals possess incisors that have some of these traits, but only Glires has all six character states. In the lower incisors, the zalambdalestid Kulbeckia possesses the ®rst four of these states, although the enamel encompasses a greater circumference of the crown, which is a slightly more ancestral condition (Fig. 2c). We cannot demonstrate that the open-rooted lower incisor was ever-growing and scanning electron micrographs (SEMs) of the surface of the lower incisor did not reveal Hunter±Schreger bands. Although suggestive, these character states alone are not strong evidence for a phylogenetic relationship, because other eutherians possess some of these traits. When a suite of other comparisons is added, the zalambdalestid± Glires link becomes more plausible. The two larger, anterior upper incisors and much smaller third upper incisor known for Kulbeckia are preserved only as roots; thus, the disposition of enamel is unknown. One (or two) smaller, more medial incisor(s) may have been present (Fig. 2a, b). Rodents possess one upper incisor on each side while lagomorphs possess two, and thus the condition in Kulbeckia could have been ancestral to both of these orders. Kulbeckia and other zalambdalestids (Zalambdalestes and Barunlestes) warrant further comparison with Glires, especially with more basal members. All have narrow, elongate snouts (Figs 1a and 2b). Zalambdalestes and Barunlestes fall between the ancestral condition in which most of the elongation is within the maxilla and the derived condition in Glires in which half or more of the elongation is in the premaxilla. Although it shows some premaxillary elongation, Kulbeckia is close to the ancestral condition. In the morphocline Kulbeckia to Zalambdalestes to Barunlestes to Glires, we see a reduction of incisors from 3, 4, or 5/4 (5 in the upper jaw, 4 in the lower jaw) to 2/3 or 3/3 to 2/3 to 1/1 or 2/1. Homologies of the four lower incisors are open to inter- pretation. Kulbeckia may retain the ancestral therian condition of four incisor sites seen in some early eutherians. In these taxa, however, the incisors are of nearly equal size, while in Kulbeckia the most medial incisor is six times longer and perhaps as much wider than the three following teeth. In Glires embryological evidence suggests that the medial, enlarged lower tooth is the second, lower, deciduous incisor 19 . Unfortunately, embryological evidence does not help in determining homologies in available fossils, and thus without ontogenetic information we use the overall anatomy of the dentitions to support the view that the medial pair of lower incisors in zalambdalestids and Glires are homologous. In the same taxa, canines are never large, ranging from two roots above (in the upper jaw) and below (in the lower jaw) to one above and below to no canines. In Kulbeckia there are 4/4 premolars while later zalambdalestids have 3/4 or 4/4, and among some Glires, 3/3 are retained. Upper molars of both Kulbeckia and early Glires such as © 2001 Macmillan Magazines Ltd