Reassessing hominoid phylogeny: evaluating congruence in the morphological and temporal data

Paleobiology, Fall 2004 by Finarelli, John A, Clyde, William C

Abstract.-The phylogenetic relationships of fossil and extant members of the primate superfamily Hominoidea are reassessed by using both conventional (morphological) cladistic and stratocladistic (incorporating morphological and temporal data) techniques. The cladistic analysis recovers four most parsimonious cladograms that distinguish postcranially primitive ("archaic") and derived ("modern") hominoid clades in the earliest Miocene of East Africa and supports distinct hominine and pongine clades. However, the relationships among the pongines and hominine clades and other Eurasian hominoids remain ambiguous and there is weak support (Bremer decay indices, reduced consensus, and bootstrap proportions) for several other parts of the proposed phylogeny.

An examination of the partitioning of homoplasy across the two major hominoid clades recovered in the cladistic analysis indicates that the majority of the observed homoplasy resides in the postcranially derived clade. An examination of the partitioning of homoplasy across anatomical regions indicates that dental characters display a significantly higher level of homoplasy than postcranial characters. A rarefaction analysis demonstrates that the higher homoplasy associated with the dental characters is not the result of sampling biases, indicating that postcranial skeletal characters are likely the more reliable phylogenetic indicators in the hominoids.

The branching order of the most parsimonious cladograms shows better than average congruence with the observed ordering of first appearances in the fossil record, implying that the hominoid fossil record is surprisingly good. As with morphologic parsimony debt, most of the stratigraphic parsimony debt in these cladograms is associated with the "modern" hominoid clade. A stratocladistic analysis of the data recovers a single most parsimonious phylogenetic tree with a different cladistic topology from the morphological cladogram. The most striking difference is the elimination of the postcranially primitive clade of hominoids in the early Miocene in favor of a pectinate succession of taxa. The relative position of the late-appearing taxon Oreopithecus is also altered in the stratocladistic hypothesis. Topological differences between the cladistic and stratocladistic hypotheses highlight two intervals of significant discord between the morphological and temporal data-the early Miocene of eastern Africa and the late Miocene of Eurasia. The first discrepancy is likely the result of poor preservation and morphological homoplasy in Morotopithecus, as the fossil record in the early Miocene of eastern Africa for the ingroup is rather good. The second discrepancy is likely the result of the unusual preservation conditions associated with the late Miocene hominoid Oreopithecus.

Introduction

Despite a considerable research effort to refine our understanding of the evolutionary relationships among known fossil and extant hominoid taxa, little consensus has emerged (e.g., Begun 1992a,b, 1994, 1995; Begun et al. 1997a; de Bonis and Koufos 1993; Dean and Delson 1992; Andrews et al. 1996; MacLatchy et al. 2000). Although their use in phylogenetic analysis remains controversial, temporal data represent a potentially relevant class of information that is often overlooked (Fisher 1991, 1994; Huelsenbeck 1994; Wagner 1995). To date no phylogenetic analysis of the Hominoidea has explicitly incorporated temporal data in hypothesis testing, and no quantitative evaluation of the congruence of the morphological and temporal data for the hominoid fossil record has occurred. In an attempt to clarify these evolutionary relationships and to evaluate congruence across data types, a phylogenetic reassessment of the Hominoidea was performed incorporating both morphological and stratigraphic data.

Although cladistic hypotheses are evaluated with no explicit reference to time, all cladograms make implicit statements about the relative time of divergence among taxa by the order of branching events (Fisher 1991; Wagner 1995). The sequence of branching events in a morphological cladistic hypothesis is often harmonized with the fossil record of the ingroup through the creation of "ghost lineages," artificial extensions of a taxon's range beyond its observed first appearance in the fossil record (Norell 1993). This approach essentially erases any discrepancy between the observed order of appearance events and the order implied by the hypothesis. Insofar as ghost lineages explain away discrepancies between (stratigraphie) observation and (cladistic) hypothesis, they may be considered appeals to ad hoc support, analogous to the way homoplasy is invoked to explain away morphological data that are incongruent with a cladistic hypothesis (Fisher 1991, 1994).

Stratocladistics is a parsimony-based criterion that evaluates competing phylogenetic hypotheses relative to both morphological characters and a stratigraphic character derived from the stratigraphic record of the ingroup taxa (Fisher 1991,1992, 1994). The morphological component is evaluated as in conventional cladistic analyses, where each instance of homoplasy imparts a unit of "morphological parsimony debt" upon the hypothesis. In addition, each instance of incongruence between the stratigraphic data and the hypothesis, where a taxon is predicted to exist by the branching order yet is not observed, imparts a unit of "stratigraphic parsimony debt" (Fisher 1992). Stratocladistics then sums the morphological and stratigraphic parsimony debt values of each hypothesis, creating a single "total parsimony debt" value. The minimum total parsimony debt value over the set of possible phylogenetic hypotheses determines the overall most parsimonious phylogenetic hypothesis (Fisher 1992, 1994).