"Given the considerable amount of parallelism and convergence that is believed to have occurred in virtually all components of the rodent masticatory system, as well as the uncertainties that surround the recognition of homology or homoplasy, we believe that those phylogenetic hypotheses which can be corroborated by data from several different (and preferably unrelated) organ systems are more likely to reflect the true phylogeny of a group, than are those hypotheses corroborated by only a single character complex." (Luckett and Hartenberger 1985)

Currently I am in a collaboration with Scott Steppan at Florida State University working on resolving the phylogenetic relationships among members of the family Muridae, which comprise almost 25% of all mammals. The following is an excerpt of a related proposal submitted prior to our collaboration.

Significance of rodent systematics

Rodents are extremely important to evolutionary biology, medicine, genomics, ecology, and agriculture. Of the 4,206 species of mammals 1,752 are rodents (42%) and 1,122 are in the family Muridae. In other words, slightly less than half of all mammals are rodents, and almost 27% belong to a single family. Mus and Rattus account for almost 1.6 million of the 1,603,895 rodent nucleotide sequences in GenBank, and over 35,000 articles using rodents were cataloged in PubMed in 1999 alone. The draft of the Mus musculus genome expected within a year will become a standard for comparative genomics and genome structure. In industrialized countries, rodents consume 1-20% of crops and as much as 50% of crops in less industrialized countries. In Pakistan rodent control produced a 21.4% increase in rice production. In natural environments, rodents account for 50% of the plant material consumed by herbivores and routinely out-consume large herbivores.

Rodents serve as direct or indirect vectors of many human diseases. Plague epidemics have been initiated from fleas carried by Rattus, Meriones, Tatera, and Rhabdomys, and at least 200 other rodents carry plague-bearing fleas. Hantaviruses are an extremely interesting case. At least 13 species in 7 different genera of rodents have directly infected humans or are carriers of various hantavirus strains. Importantly, the phylogeny of hantaviruses parallels the phylogeny of their rodent hosts, implying a long („ 10 My) period of coevolution and cospeciation of rodents and viruses. Therefore, a complete understanding of the evolution, ecology and biogeography of hantaviruses requires an accurate understanding of the phylogeny and biogeography of their rodent hosts.

The phylogeny of rodents sheds light on many fields of evolutionary biology. Biogeography: Rodents naturally occupy all the continents except Antarctica, and their phylogeny and speciation dates reveal much about biogeography and biotic exchange. For example, the date of entry of murid rodents into South America is highly controversial, and the resolution of this debate is dependent on accurate phylogenies and speciation dates. Morphology and developmental biology: Rodents exhibit levels of morphological homoplasy nearly unparalleled among mammals. This confounds attempts to resolve their relationships but provides "replicate experiments" within which to study the evolution of skeletal, myological, and developmental traits. At least twice rodents have adapted to arboreal, gliding, fossorial, saltatorial, terrestrial and semiaquatic lifestyles. Having a well-corroborated phylogeny will simplify the task of distinguishing synapomorphies from convergent adapatations. Cospeciation: Rodents commonly exhibit patterns of cospeciation with their pathogens (i.e., hantavirus) or parasites (i.e., lice). Once again, robust phylogenies are necessary to correlate patterns of host-parasite speciation. Molecular evolution and dating: Rodents accumulate nucleotide substitutions and molecular homoplasy faster than most other mammals, complicating phylogenetic analysis. Also, estimation of divergence dates within rodents is difficult. Usually, a molecular clock is calibrated based on the Mus-Rattus split at ~14 Mya. While efficient, this approach is reliant upon the accuracy of the Mus-Rattus date. Alternatively, rates are based on external calibrations (i.e., horse-rhino at ~50 Mya or human-Old World monkey at ~22 Mya), perhaps applied after linearization of the phylogeny. Uniformly, these studies produce divergence dates highly incongruent with the fossil record, suggesting that the rate acceleration in rodents is not being fully accommodated. Power of molecular systematics: Rodents have undergone periods of rapid speciation, and these provide tests of the limits of molecular characters to resolve relationships. For example during the Miocene the first undisputed fossils of several murid subfamilies appear, suggesting a burst of speciation. To date, molecular data have given only very weakly supported and conflicting resolutions of this radiation.

Historically, rodents are divided into paraphyletic grades of evolution. Sciurognathy (plesiomorphic) defines a paraphyletic group of taxa. Hystricognathy is distinguished by the angular process of the jaw originating laterally to the alveolus of the incisor. Based on the pattern of origination of the masseter muscles, rodents are further divided into four groups. In protrogomorphs (plesiomorphic) the masseter muscles originate entirely on the zygomatic arch (retained in Aplodontia). In sciuromorphs (squirrel-like rodents, gophers, and beavers) part of the lateral masseter has shifted onto the rostrum. Hystricomorphs (hystricognaths, jumping mice, springhare, and gundis) have the medial masseter originating on the rostrum and passing through an enlarged infraorbital foramen. Finally, in myomorphs (most rats and mice, dormice) the anterior part of the lateral masseter originates on a special projection of the zygomatic arch and the anterior part of the medial masseter passes through an enlarged infraorbital foramen.

Resolving these grades of evolution into monophyletic clades has been extremely difficult due to complicated patterns of morphological homoplasy in many taxa and the absence of some key fossils. For this reason, molecular data is an invaluable complement to morphological study. My laboratory has made significant progress towards resolving some of the most difficult phylogenetic problems among rodents using nucleotide sequence data from the growth hormone receptor gene (~1000 bp) and the BRCA1 gene (~3400 bp).

Muridae - "Still it presents the irony that if we devalue or dismiss all characters thought to exhibit homoplasy within the realm of Muroidea, then it is not apparent what evidence remains upon which to ground our phylogenetic hypotheses." (Carleton and Musser 1984)

The family Muridae (taxonomy of Carleton and Musser is followed) contains about 1/4 of all mammals, and the inter-relationships of the major clades are among the most perplexing problems in mammalian systematics. Relationships among the taxa are based primarily on dental morphology, and the problem of inter-relationships is so acute that the subfamilies are currently ranked equally with few assertions of sister group relationship. Murids explosively radiated near the beginning of the Miocene, when the first unambiguous fossils of several subfamilies appear (affinities of most Oligocene fossils are controversial). The rapidity of this radiation is reaffirmed by the lack of well-supported subfamilial associations in molecular studies.

Given the diversity of Muridae, a dissection of their relationships is a vast undertaking. Here I propose the beginning of a long-term project to decipher the relationships among the major clades of murid rodents. This work focuses on relationships at the subfamilial level. I realize that several subfamilies are probably paraphyletic, and only a subset of issues are mentioned here. This work is unique in that it does not rely strictly on DNA sequencing, but also incorporates an exciting new source of data, unique SINE insertions. The combination of these two sources of data promises to give strong support to several clades of murid subfamilies.

Short interspersed elements (SINEs) - SINEs are tRNA derived sequences bearing an RNA polymerase III promoter and can be present >10,000 in a genome. To retrotranspose SINEs must rely on reverse transcriptase encoded by a gene elsewhere in the genome, presumably a LINE. The use of unique SINE insertions to define monophyletic clades has been performed with great success for cetartiodactyls, salmon, and cichlids. The key requirements for the use of SINE elements in phylogenetics is that the elements can be uniquely amplified from orthologous regions and that the elements were actively being retrotransposed during speciation. B2 SINEs are derived from tRNA^lys, are absent from dipodids, but are present in all murid subfamilies surveyed (Murinae, Gerbillinae, Arvicolinae, Spalacinae). Therefore, B2 elements were actively being retrotransposed during the radiation of murid subfamilies. There are ~100,000 copies of B2 per murid genome, and the average sequence divergence among elements is 10%. B1 SINEs are members of the second major class of rodent retroposons, are derived from 7SL RNA, as are primate Alu elements, and are ubiquitous among rodents. Both B2 and B1 elements can be divided into subfamilies that are defined by unique substitutions, implying successive waves of amplification of elements over time. The six subfamilies of B1 have intra-subfamily levels of divergence as high as 30.9% and as low as 9.8%. Most new SINE elements are inactive and should accumulate mutations at the neutral rate, much like introns. Li determined an average divergence of 14.4% for Mus and Rattus introns. The levels of divergence between pairs of B2 or B1 elements in murids range above and below the level of intron divergence, and this indicates that both elements have been undergoing active retrotransposition throughout the radiation of Muridae. ID elements are produced by BC1 RNA and are present in all rodents. They have been very active in rodents, as demonstrated by copy number differences of 2-3 orders of magnitude among rodents. Rattus has the largest number of ID elements (~130,000) of any rodent. This is the third major SINE element whose phylogenetic pattern of active retrotransposition indicates that it has been active during the murid radiation, although most ID insertions seem to be species-specific. A major goal of this project is to use orthologous SINE insertions to define monophyletic groups within Muridae. This approach will be an invaluable complement to the collection of nucleotide sequence data and is the first application of unique SINE insertions to rodent systematics. Previous studies have used the presence/absence of unique elements in various rodents but have not explicitly determined the orthology of insertion events.

A long-term interest of mine has been the interordinal relationships among placental mammals. I am particularly interested in resolving relationships among the members of a clade supported by molecular data that includes rodents, lagomorphs (rabbits & pikas), primates, tree shrews (Scandentia), and colugos (Dermoptera). The following is an excerpt from a proposal submitted a few years ago that should give you a flavor for my interests. I realize some of the text is outdated in the face of papers published since 1999.

Despite over 100 years of scrutiny the relationships among the orders of placental mammals (Eutheria) have yet to reach a well-supported consensus among morphological or molecular systematists. Indeed, significant conflicts have arisen in the last several years concerning the affinities of Xenarthra (armadillos, sloths, and anteaters) and Pholidota (pangolin), the primitive status of Insectivora (shrews, moles, hedgehogs, and kin), and the monophyly of Insectivora, Rodentia, and Artiodactyla, to mention only a few. These few examples demonstrate how much progress remains to be made in resolving eutherian interordinal relationships. For a molecular systematist two main obstacles exist to the resolution of these questions. First is the inadequate sampling of the full diversity of eutherian mammals between and within orders. Although Stanhope et al. have produced data from each order of extant eutherians, most nodes remain unresolved and the most speciose orders (i.e., Rodentia and Chiroptera) are represented by few taxa. All other molecular studies of Eutheria have sampled comparatively few orders or few species within each order. Second is ignorance of the best genes or evolutionary parameters (i.e., sequence length, taxon sampling, substitution rate, combination of genes) that have the highest probability of resolving relationships among taxa that rapidly speciated more than 60 million years (My) ago.

The work proposed here is part of my long-term objectives to resolve eutherian interordinal relationships as well as to determine the genes and evolutionary parameters most appropriate for resolving those relationships. As outlined later, a number of benefits will result from this study, but there are two most important benefits. First, the combination of data generated in this study with that being produced by other laboratories will result in a much more resolved and better supported phylogeny of mammals based on molecular data. Second, if previous molecular studies are any indication, the molecular phylogeny will differ from morphological phylogenies in several respects, and this will inspire careful examination of molecular and morphological evidence. The simulations proposed in this study will permit one to determine if these conflicts are due to misleading biases in the molecular data (i.e., long branch attraction). Hypotheses concerning the paraphyly of rodents are an excellent example in which inadequate taxonomic sampling in molecular studies has led to striking, but poorly supported, conflicts with morphological data. If no potential source of bias in the molecular data can be found, this should inspire morphologists to carefully reevaluate the characters underlying that part of their phylogeny. The recently proposed affiliation between cetaceans and hippopotamuses is an example of a molecular hypothesis that has inspired fruitful reevaluation of the morphological traits supporting a cetacean-mesonychid relationship.

Archonta. Determining the sister-group relationships of Primates, Dermoptera, Scandentia (Euarchonta), Rodentia and Lagomorpha (Glires) are the two initial priorities of this project. Archonta was proposed by Gregory to contain a Primates/Scandentia (tree shrews) clade sister to a Chiroptera (bats)/Dermoptera (colugos) clade. Archonta is maintained in most morphological and paleontological studies of eutherian relationships, although the sister group relationship of Primates and Scandentia is disputed. However, molecular analyses strongly separate the bats from the other archontan orders, although they are ambiguous regarding the exact placement of Primates, Dermoptera and Scandentia (Euarchonta). A combination of 12s rRNA, vWF, IRBP and some morphological traits weakly supports a Dermoptera-Scandentia clade, but Mhc-DRB intron sequences support a Primates-Scandentia clade. The Mhc-DRB results are suspect because only five orders were represented, and the sequences were very divergent (interordinal sequence divergence = 40-86%). I am surveying the diversity within all four orders and expect to resolve confidently any affinities among the Archonta. Results based on only a third of the total proposed data set (GHR) support a Dermoptera-Scandentia clade, although with weak support (preliminary results).

Volitantia. Volitantia (bats + colugos) is usually singled out by morphological studies as having particularly strong support. This hypothesis is attractive because a close relationship between volant (bats) and gliding (colugos) mammals opens intuitively appealing explanations of the origin of flight. Nevertheless, molecular data refute a sister group relationship between bats and colugos. This necessitates a reevaluation of the origin of mammalian flight and the evolution of myriad morphological traits involving the inner ear, skeletal system, and the integument and musculature of the forelimbs. Associated with Volitantia is a recent hypothesis based on neural traits that Microchiroptera and Megachiroptera are not sister taxa and that Megachiroptera are closely related to Primates. This hypothesis receives no support from other morphological or molecular studies. An examination of this hypothesis is a natural outgrowth of this study because both Microchiroptera and Megachiroptera are being thoroughly sampled.

Rodent monophyly. Potential paraphyly of rodents has attracted very much attention lately, as a partial literature review illustrates. Many nuclear and mitochondrial sequences support a closer relationship of Mus and Rattus to primates and artiodactyls than to the guinea pig. More recently, D'Erchia et al. used complete sequences of mitochondrial DNA to separate the guinea pig from mouse and rat. However, the data was inadequate to statistically address the monophyly of rodents, artiodactyls or primates. Moreover, numerous morphological traits and other molecular data support rodent monophyly. A more recent challenge to rodent monophyly is the placement of a dormouse, Glis glis, sister to the guinea pig and well separated from mouse and rat based on complete mitochondrial sequences. This result is unexpected because dormice, Mus, and Rattus are all members of the infraorder Myomorpha. This hypothesis receives some support from the shared absence of B2 and B2-like retroposons, although this latter data must be viewed with caution given the instability of mobile elements. Reanalysis of the mtDNA data with more recently published genomes tends to support the monophyly of rodents. The most parsimonious tree for vWF supports the monophyly of rodents and the placement of dormice as the basal lineage of rodents with weak bootstrap support. Preliminary data from this study supports the monophyly of all rodents, including guinea pig and dormice.

Glires and Anagalida. Morphological systematists vary in their support for Glires (Rodentia + Lagomorpha), but recent morphological work on extant and extinct taxa favors this grouping based on the acquisition of at least six unique traits. Graur et al. used numerous proteins but only four taxa at a time to support a close relationship of lagomorphs with Euarchonta. However, Halanych suggested that long branch attraction problems existed and four-taxon trees are an inferior method of analysis compared to use of all available taxa. Mitochondrial and nuclear gene sequences are ambiguous regarding lagomorph relationships, in that combined 12s-16s/alpha-2B adrenergic receptor/ aquaporin/IRBP/vWF supports a close relationship between a rabbit and human (Scandentia and Dermoptera were not represented) but combined 12s rRNA/vWF/IRBP and morphological data support Glires. Clearly, Glires requires closer examination at the molecular level. Morphological data provide some support for a sister group relationship between Glires and the order Macroscelidea (elephant shrews) in the superordinal taxon Anagalida, although this taxon is characterized by only two unique synapomorphies. Molecular data consistently fails to support this clade, instead suggesting a close relationship between Macroscelidea and paenungulate taxa. The preliminary results from this study strongly support Glires and a macroscelidid-paenungulate clade. The strong support for Glires likely can be ascribed to improvement in phylogenetic reconstruction brought about by thorough taxonomic sampling.

Home