Phylogeny of Glires

Conventionally, there is little doubt about the monophyly of either Rodentia or Lagomorpha. The monophyly of Glires, however, has proven to be one of the most controversial issues in higher-level mammalian systematics (Wilson, 1949; Wood, 1957; McKenna, 1975; Luckett and Hartenberger, 1985, 1993; Li and Ting, 1985; Novacek, 1985; 1992; Li et al., 1987; Meng and Wyss, 2001). Rodentia and Lagomorpha have been at one time or another related to a variety of mammalian groups, either together or individually, such as rodent + primate (McKenna, 1961), rodent + leptictid (Szalay, 1977, 1985), and lagomorph + zalambdalestid to the exclusion of rodents (Szalay and McKenna, 1971; McKenna, 1975). All of these hypotheses would dismiss the similarities between rodents and lagomorphs as convergence. As pointed out elsewhere (Meng and Wyss, 2001), these hypotheses of convergence require identification of some third taxon that shares a unique common ancestry with one of the two groups but lacks the derived similarities common to both. Such a “third taxon” has never seriously threatened the sister-group relationship between lagomorphs and rodents. Several recent morphological studies favored monophyly of Glires (Li et al., 1987; Novacek, 1992; Luckett and Hartenberger, 1993; Shoshani and McKenna, 1998; Meng and Wyss, 2001). The present study supports this relationship.

During the last decade the challenges to the monophyly of Rodentia or Glires by some molecular studies have been severe. Based on analysis of amino acid sequence data, Graur et al. (1991) first questioned whether the guinea pig (Cavia porcellus) is a rodent. Their result suggested a closer grouping of the guinea pig with Primates, not with other rodents, which would make the conventional Rodentia, and therefore Glires, nonmonophyletic. Some molecular analyses also rejected the monophylies of Glires and Rodentia (Li et al., 1992; Graur et al., 1996; D'Erchia et al., 1996), while others were less certain with respect to the relationships of rodents and lagomorphs (Ma et al., 1993; Martignetti and Brosius, 1993; Honeycutt and Adkins, 1993; Porter et al., 1996; Stanhope et al., 1996; Huchon et al., 1999). As pointed out by several studies, earlier molecular studies that claimed the nonmonophyly of Glires may have been biased by inappropriate methods and deficient quality of data used in analyses (Allard et al., 1991; Hasegawa et al., 1992; Graur, 1993; Novacek, 1993; Catzeflis, 1993; Sullivan and Swofford, 1997). Recent molecular studies that include more taxa and sequence data (Madsen et al., 2001; Murphy et al., 2001a, 2001b), however, concur with recent morphological studies in supporting the Glires monophyly.

It has been questioned that some derived characters shared by rodents and lagomorphs, such as loss of canines, have been attributed to the evolution of gnawing function; therefore, they were regarded as parallelism (Graur et al., 1996). Gnawing, or incisor biting, is a common function in eutherian mammals, but the gnawing in gliroid mammals is unique in two aspects. First, the upper and lower cheek teeth must be distantly separated when the upper and lower incisors are engaged, or vice versa. Second, the lower incisors can be protruded anterior to the upper incisors so that the tips of the lower incisors can be honed against the tips of the upper incisors. Some other placental mammals, such as plesiadapids, have enlarged incisors, but they cannot gnaw the same way as do the gliroid mammals. Still, some pleciadapids lack canines. It seems to me there is no evidence for the relationship between the loss of canines and the function of gnawing; nor is there reason to think, based on our current knowledge of relationships of gliroid mammals, that the enlarged incisors and other derived similarities shared by rodents and lagomorphs evolved independently.

Graur et al. (1996: 335) also suggested that

Based on the phylogeny and distributions of gliroid and related eutherian mammals (Meng and Wyss, 2001; Meng et al., 2003), a rodentlike ancestral eutherian morphotype is not supported by paleontological data. Features, such as losses of the incisors, canines, and premolars, modification of the glenoid fossa, and enlarged upper and lower incisors, are apparently derived conditions evolved in the lineage toward rodents and lagomorphs and are not retention of a rodentlike eutherian morphotype that was present in the Cretaceous.

Divergence Times of Glires

On the molecular time scale, sciurognath rodents made their first appearance approximately 112 ± 3.5 Ma, followed by the divergence of hystricognaths (109 ± 3.2 Ma), Gerbillidae/Muridae (66.2 ± 7.6 Ma), Muridae/Cricetidae (65.8 ± 2.2 Ma), and mouse/ rat (40.7 ± 0.9 Ma) (Kumar and Hedges, 1998; see also Janke et al., 1994). Lagomorpha would have appeared at 90.8 ± 2.0 Ma (Kumar and Hedges, 1998). Other molecular analyses also suggest that the common ancestor of rodents split from other eutherian orders in the Cretaceous (Janke et al., 1994; Huchon et al., 2000). However, there is no evidence in the fossil record supporting such an early date. The earliest rodent, Acritoparamys (which may be considered a sciurognath) occurred in the late Paleocene Clarkforkian of North America, about 57 Ma (Dawson and Beard, 1996). Its age is only half the divergence time for the Sciurognathi estimated by the molecular clock hypothesis (Kumar and Hedges, 1998). The earliest known fossil record of hystricognaths is in the early Eocene (McKenna and Bell, 1997), approximately 55 Ma. Clearly, the molecular clock hypothesis tends to push divergence times of the Glires at various levels deeper into history than what the fossils have documented. Five hypotheses for the divergence time of placental mammals and the discrepancy between fossil and molecular dates have been postulated: (1) the conventional view that placental radiation occurred after the K-T boundary, (2) poor preservation of Cretaceous mammals, (3) a lack of morphological diagnostic features in Cretaceous taxa, (4) hidden Cretaceous fossil records in southern continents, and (5) deep branching of superordinal clades of placental mammals (fig. 7.3).

The conventional view of the eutherian radiation (H1 in fig. 7.3; “Explosive Model” of Archibald and Deutschman, 2001) emphasizes a post K-T boundary radiation of modern eutherian orders, presumably having taken place in the empty niches left by the extinction of dinosaurs (Gingerich, 1977; Novacek, 1992). The geological distributions of the monophyletic Glires, including basal gliroid taxa such as eurymylids, mimotonids, and alagomyids, are found in the Paleogene (figs. 7.1, 7.2) and therefore support the conventional view. Moreover, the earliest occurrences of immediate outgroups to Glires, such as anagalids, pseudictopids, macroscelideans, and primates, are present only in the early Paleogene. These would indicate that the cladogenesis of rodents and lagomorphs took place no earlier than the K-T boundary and contradict the much earlier divergence times for rodents and lagomorphs estimated from some molecular studies. If the fossil record roughly reflects cladogenesis of gliroid mammals, then one has to question the accuracy of the molecular clock. Problems concerning the vagaries of the molecular clock have been reviewed elsewhere (e.g., Ayala, 1997; Bromham et al., 1999; Rodríguez-Trelles et al., 2001) and are beyond the scope of this study.

Hypothesis 2 (H2 in fig. 7.3., which is similar to the “Short Fuse Model” of Archibald and Deutschman, 2001) suggests that the split of modern placental orders occurred early in the Cretaceous and may be concurrent with the breakup of continents (Hedges et al., 1996), but that fossils of presumably Cretaceous placental mammals did not show up in the record because these early forms were small and fragile, so that their preservation as fossils was less likely (Cooper and Fortey, 1998; Foote et al., 1999a). As Foote et al. (1999a, 1999b) demonstrated, however, assuming even very low preservation rates for Cretaceous mammals, the gap between the earliest fossil record and the divergence time based on molecular data is unacceptably large. Several major groups of mammals, including “triconodonts”, symmetrodonts, multituberculates, therians, and eutherians (Lillegraven et al., 1979; McKenna and Bell, 1997; Novacek et al., 1998) flourished in the Cretaceous. It seems unreasonable that these taxa, most of which were small, would be preserved as fossils while placental taxa were not. Although Cenozoic mammals are usually larger than Cretaceous mammals, and preservation rates of Cenozoic fossils are often greater than those of Cretaceous fossils (Cooper and Fortey, 1998; Foote et al., 1999a, 1999b), these generalizations do not seem applicable to the Glires because early gliroid mammals are generally small and many are smaller than most Cretaceous mammals.

It should also be noted that the molecular clock hypothesis extends not only the divergence times of the sciurognath and hystricognath rodents into the Cretaceous, but also those of lower-level rodent groups into much earlier periods of the Tertiary, such as the mouse-rat split at about 41 Ma. Fossil records show that Mus and Rattus diverged at about 8 to 14 Ma (Jacobs and Pilbeam, 1980; Jaeger et al., 1986), with Mus not appearing until 5.7 Ma (Jacobs and Downs, 1994). The argument of poor preservation for Cretaceous mammals does not seem applicable for these Tertiary rodents.

Hypothesis 3 (H3 in fig. 7.3; phylogenetic fuse of Cooper and Fortey, 1998) implies that cladogeneses of modern placental orders took place about 110 Ma, but the lineages remained morphologically cryptic until they evolved diagnostic characters after the K-T boundary (Cooper and Fortey, 1998; Foote et al., 1999a). Hypotheses 1 and 3 both imply rapid morphological modifications during a short period of time. The difference between them is that in hypothesis 1 the morphological changes are coeval with cladogeneses, whereas in hypothesis 3 cladogeneses occur throughout the Cretaceous and are decoupled from morphological changes. The relationship of molecular and morphological evolution is not fully understood, but some studies suggest that they are generally correlated, not decoupled (Omland, 1997). Moreover, lack of diagnostic characters is a poor argument for the absence of a subgroup of rodents, such as Mus, in early Tertiary records. This is particularly so when the cladogenesis of Mus and its sister taxon was recognized in fossil records (Jacobs and Downs, 1994).

Hypothesis 4 (H4 in fig. 7.3), or the “Garden of Eden” hypothesis (Foote et al., 1999a), suggests that deep splitting of modern placental mammals took place in areas where fossil records of Cretaceous eutherians are unknown or sediments of relevant ages are not preserved (Foote et al., 1999a). Potential areas of this kind include Africa, Australia, and Antarctica. This hypothesis further assumes that in the early Tertiary, about 65 to 60 Ma, early members of placental orders dispersed to northern continents and were recognized as the post–K-T radiation (Bromham et al., 1999). This hypothesis implies that known Cretaceous eutherians from northern continents are unrelated to modern orders, which contrasts with studies that suggest that higher-level taxa such as ungulates may have evolved from these northern eutherians (Archibald, 1996). Nonetheless, the “Garden of Eden” hypothesis can be tested with fossil finds from southern continents (Foote et al., 1999a). Ausktribosphenos nyktos from the Early Cretaceous in Australia (Rich et al., 1997, 1999) was considered to be a possible erinaceid and thus evidence corroborating the “Garden of Eden” hypothesis (Rich et al., 2001). However, the status of A. nyktos as a genuine erinaceid remains controversial (Kielan-Jaworowska et al., 1998; Musser and Archer, 1998; Archer et al., 1999; Rich et al., 2001). Additional material from the cranial and postcranial skeleton of A. nyktos would help clarify its taxonomic identification.

The phylogeny and distribution of basal gliroid mammals show that their early divergence area, if not their center of origin, was in Asia. All basal gliroid taxa, such as Mimotona, Mimolagus, Gomphos, Rhombomylus, Matutinia, Eurymylus, Heomys, Tribosphenomys, and Alagomys are known only from Asia, except for one species of Alagomys that was found in North America (Dawson and Beard, 1996). Assuming gliroid mammals originated in the southern continents during the Cretaceous, the large bodies of water present between Asia and the southern continents (Smith et al., 1994) were obvious geographic barriers for the migration of gliroid mammals in the Late Cretaceous. Finds of basal gliroid mammals in areas that are possible dispersal routes may also test the “Garden of Eden” hypothesis. However, the absence of basal Glires, such as mimotonids and eurymylids, in Tertiary sediments outside Central Asia challenges the theory of the origin of the Glires in southern continents. To reiterate, the “Garden of Eden” hypothesis does not explain the absence of subgroups of rodents, such as Mus and Rattus, in the early Tertiary as predicted by the molecular clock hypothesis. Given that rodents and lagomorphs have a relatively dense fossil record since the early Eocene, hidden origin centers for various subgroups of rodents are difficult to defend.

Hypothesis 5 (H5 in fig. 7.3; the “Long Fuse Model” of Archibald and Deutschman [2001]) predicts a deep branching of superordinal clades of eutherian mammals. It argues that although members of modern placental orders are not present in the Cretaceous, stem taxa to the modern placental orders can be found in the Cretaceous. This hypothesis may be supported by the relationship of zhelestids and ungulates (Archibald, 1996; Archibald et al., 2001), but that relationship has been put in doubt by other studies (Novacek et al., 1998, 2000). More recently, Archibald et al. (2001: 64) presented another case study of the Long Fuse Model that included zhelestids, ungulates, zalambdalestids, and Glires. These authors concluded that the results of their analysis “support a superordinal clade including zalambdalestids and Glires” and that the presence of zalambdalestids in the 85–90 Ma faunas at Dzharakuduk “argues that the superordinal clade including Glires had separated from other superordinal placental clades by this time” (Archibald et al., 2001: 64). Archibald et al. (2001) considered that the result of their study was concordant with molecular-based estimates for the superordinal diversification of placentals, and thus called zalambdalestids Late Cretaceous relatives of rabbits and rodents, a scenario once favored by McKenna (1982, 1994).

However, although Archibald et al. (2001) considered zalambdalestids to be more closely related to Glires than other Cretaceous eutherians selected for their analysis, they did not sufficiently test whether Glires, represented by Tribosphenomys and Mimotona in their analysis, is the Tertiary group that is most closely related to zalambdalestids. As mentioned above, rodents and lagomorphs have been frequently related to primates, leptictids, tree shrews, pseudictopids, anagalids, macroscelideans, zalambdalestids, eurymylids, and mimotonids. All these groups, except zalambdalestids, are Tertiary mammals and an appropriate phylogenetic analysis to test Glires relationships should include them. Only two Tertiary ungulates, Protungulatum and Oxyprimus, in addition to Tribosphenomys and Mimotona, were included in Archibald et al.'s (2001) study. Because these ungulates are distantly related to the Glires, the zalambdalestid-Glires link (Archibald et al., 2001) is insufficiently supported. When relevant taxa are included in the analyses (Meng and Wyss, 2001; Meng et al., 2003; this study), zalambdalestids are not the closest group to Glires.

Nonetheless, because zalambdalestids were nested within crown placentals in this and previous studies (Meng and Wyss, 2001), it is not impossible to think that at a certain level of mammalian phylogeny, such as the clade of “Euarchontoglires” of Murphy et al. (2001b), placentals may have occurred within the Cretaceous. Murphy et al. (2001b) actually suggested that zalambdalestids may be early representatives of “Euarchontoglires”, which is congruent with McKenna's (1982: 215) earlier suggestion that “If one accepts a broad association with lagomorphs of rodents (including eurymylids), living elephant shrews, anagalids, pseudictopids, and zalambdalestids, then the clade that includes them all dates from some much earlier time in the Cretaceous.” The scenario of a Cretaceous split of higher-level placental clades, however, needs further testing in the broader scope of mammalian phylogeny.