Sequence gaps join mice and men: phylogenetic evidence from deletions in two proteins.

Recent nuclear sequence analyses have provided evidence that primates and rodents are more closely related than previously believed (Madsen et al. 2001; Murphy et al. 2001a, 2001b). This proposal is difficult to reconcile with morphological insights (Liu et al. 2001; Novacek 2001) and is not generally supported by current mitochondrial sequence data (Reyes, Pesole, and Saccone 2000; Nikaido et al. 2001; Arnason et al. 2002; Janke et al. 2002). Moreover, the supporting data and analyses have been criticized on methodological grounds (Rosenberg and Kumar 2001). Here we report deletions in two nuclear protein-coding genes that lend independent support to this contested grouping. Some 18 orders of placental mammals are currently recognized, but their phylogenetic relationships remain highly controversial. Extensive sequence comparisons of mainly nuclear genes support a basal division into four major clades (Xenarthra, Afrotheria, Laurasiatheria, and Euarchontoglires), which has far-reaching implications for early mammalian biogeography and morphological diversification (Murphy et al. 2001b). Euarchontoglires is composed of the orders Primates, Rodentia, Lagomorpha (rabbits, hares, and pikas), Scandentia (tree shrews), and Dermoptera (flying lemurs). In contrast, morphology groups Primates, Scandentia, and Dermoptera with Chiroptera (bats) in the clade Archonta, whereas Rodentia and Lagomorpha (jointly called Glires) are in a distant clade with Macroscelidea (elephant shrews) (Liu et al. 2001; Novacek 2001). Also, sequence data from 12 proteins encoded by the mitochondrial genome generally do not support Euarchontoglires (e.g., Nikaido et al. 2001) or even maintain rodent polyphyly in many cases (Reyes, Pesole, and Saccone 2000; Arnason et al. 2002; Janke et al. 2002). Only by excluding some taxa with high or atypical substitution rates (or both) can sound mitochondrial support be obtained (Waddell, Kishino, and Ota 2001). Establishing the monophyly of the most speciose eutherian order, Rodentia, and finding its sister group has indeed been most difficult to solve on the basis of sequence evidence (e.g., Graur, Hide, and Li 1991; Adkins et al. 2001; Huchon et al. 2002). As for the molecular data sets giving support to Euarchontoglires, it has been questioned whether these are actu-