Early mammalian radiations

INTRODUCTION

THE SEVENTY-FIFTH anniversary of the Journal of Paleontology presents a felicitous opportunity to review major changes in interpretation of mammalian phylogeny. Founding of the journal coincides with the nascence of the career of the most influential paleomammalogist of the past century, George Gaylord Simpson (1902-1984). It occurred at a time when now-archaic models for mammalian systematics and evolution, such as the aristogenesis of H. F. Osborn (1857-1935) and the typological concept of taxa, were prevalent (e.g., Simpson, 1945). These models were soon to give way to "new ways of going at things" (Laporte, 2000, p. 87); most significantly, the incorporation of quantitative methods and the evolutionary synthesis (Simpson, 1944). Subsequent decades witnessed the rise and/or sophistication of other applications and perspectives in fossil-based interpretation of mammalian systematics, including form-function analysis (e.g., Szalay, 1994) and, particularly, cladistic approaches (e.g., McKenna, 1975). Within these broad ideological frameworks, major paradigm shifts have resulted from new discoveries, conceptual changes, or (most commonly) a combination of both. Finally, mammalian systematics currently lie at the verge of a monumental paradigm shift, providing important direction for the future.

Given the brevity of this review, I focus mainly on the base and early branches of the mammalian tree: separate chapters would be required to undertake more than superficial coverage of Cenozoic radiations of placentals and marsupials. Likewise, despite the increasingly important role that molecular data have come to play in interpreting mammalian relationships (e.g., summary by Novacek, 1993), I deal here mainly with the primary subject matter of the journal itself: the fossil record. For organizational purposes, it is convenient to follow some broad systematic framework for mammals, notwithstanding the fact that placement of several major groups is unsettled and the subject of intense debate. Herein, I generally follow the phylogeny proposed by Luo et al. (2001). Taxonomic names for higher categories of mammals have proliferated in recent years, as a result of node-- based naming. Herein, I restrict usage to such terms as are most appropriate in the present context. Where traditionally-recognized groups appear to be paraphyletic but the terms remain nonetheless useful in a descriptive sense, I have adopted the common practice (e.g., Nessov et al., 1998) of citing names in quotes. Quotes are not used, however, when the same terms are used in reference to morphology (e.g., a "triconodont" has molars with a triconodont cusp pattern).

EARLIEST AND EARLIEST-DIVERGING MAMMALS

What is a mammal? In the 1920s, as for a number of decades previously, certain mammal-like reptiles (tritylodontids) were included among Mammalia, partly because their strikingly advanced cheektooth pattern is similar to what is seen in multituberculates (Simpson, 1928). Tritylodontids later (e.g., Watson, 1942) became universally regarded as non-mammalian synapsids, though their relative proximity to mammals remains contentious (e.g., Rowe, 1993; Crompton and Luo, 1993; Hopson, 1994). The union card for membership in the Mammalia long was based exclusively on diagnostic characters or character systems, especially those thought to have significant adaptive value. Pre-eminent among features observable in fossils are those related to the masticatory system, particularly the jaw joint(s). In primitive synapsids, the lower jaw is made of multiple elements, and articulation occurs between one of the so-called postdentary elements (the articular) and the quadrate of the skull; in advanced mammals, these elements are incorporated into the auditory system, and the jaw joint is instead formed by the tooth-bearing bone of the lower jaw (dentary) and an element on the side of the skull (squamosal). Complexity arose from new discoveries, together with reinterpretation of previously-known fossils. These showed that both jaw joints were actually present in some Triassic and Jurassic taxa generally considered to be mammals (e.g., K. A. Kermack and Mussett, 1958; K. A. Kermack et al., 1973). Concomitantly, an early suggestion that mammals may have had a polyphyletic origin (Simpson, 1928, 1929a) became widely adopted by the midtwentieth century, as a result of detailed studies of non-mammalian synapsids, as well as newly-discovered mammals from the Triassic-Early Jurassic (e.g., Olson, 1944, 1959; Simpson, 1959, 1960; see review by Hopson and Crompton, 1969). More recent studies have universally upheld a monophyletic origin for Mammalia (e.g., Rowe, 1988); the presence of a well-developed or dominant dentary-squamosal joint has remained either the principal defining character (Hopson, 1994), or one of a system of features relating to masticatory function (Crompton and Jenkins, 1973, 1979; Crompton, 1995). This definition has been challenged in recent years, however, by those advocating phylogenetic taxonomy, wherein taxa are defined on the basis of contents or subordinate taxa, rather than on characters or character complexes. Rowe (1988, 1993), for example, has proposed a node-based, crown group definition of Mammalia: that is, the most recent common ancestor of the three living groups (monotremes, marsupials, and placentals), plus all of its descendants. In this case, fossil relatives presumed to fall outside of this major clade are relegated to a hierarchy of other, successively more inclusive taxa, some of which have been named. This scheme has received mixed support, being followed by some authors (e.g., Wible et al., 1995; Rougier et al., 1996a) and ignored by others (e.g., Hopson, 1994). One of the problems inherent to phylogenetic nomenclature (see Lucas, 1992; Benton, 2000), instability of contents, is particularly acute with early mammals-the position of several key groups with respect to living taxa is highly uncertain. For the purposes of this general review, it is practical to view the contents of Mammalia in a traditional, inclusive sense. This is consistent with node-based definitions that include fossil taxa lying outside of the crown group (e.g., Lillegraven and Krusat, 1991; Cifelli and Muizon, 1997). Thus conceived, mammals are represented by rather complete, informative fossils dating back to the Early Jurassic, with other possible records as old as Carnian, or early Late Triassic (Lucas and Luo, 1993; Datta and Das, 1996).