Reassessing Hominoid Phylogeny: Evaluating Congruence in the Morphological and Temporal Data
Total Page:16
File Type:pdf, Size:1020Kb
Paleobiology, 30(4), 2004, pp. 614±651 Reassessing hominoid phylogeny: evaluating congruence in the morphological and temporal data John A. Finarelli and William C. Clyde Abstract.ÐThe phylogenetic relationships of fossil and extant members of the primate superfamily Hominoidea are reassessed by using both conventional (morphological) cladistic and stratoclad- istic (incorporating morphological and temporal data) techniques. The cladistic analysis recovers four most parsimonious cladograms that distinguish postcranially primitive (``archaic'') and de- rived (``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 recov- ered 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 signi®cantly higher level of homoplasy than post- cranial 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 char- acters 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 ®rst 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 strato- cladistic 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 elimi- nation 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 hy- potheses highlight two intervals of signi®cant discord between the morphological and temporal dataÐthe early Miocene of eastern Africa and the late Miocene of Eurasia. The ®rst 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 homi- noid Oreopithecus. John A. Finarelli. Committee on Evolutionary Biology, University of Chicago, 1025 East Fifty-seventh Street, Culver Hall 402, Chicago, Illinois 60637, and Department of Geology, The Field Museum of Natural History, 1400 South Lake Shore Drive, Chicago, Illinois 60605. E-mail: [email protected] William C. Clyde. Department of Earth Sciences, 56 College Road, University of New Hampshire, Durham, New Hampshire 03824. E-mail: [email protected] Accepted: 18 February 2004 Introduction date no phylogenetic analysis of the Homi- Despite a considerable research effort to re- noidea has explicitly incorporated temporal ®ne our understanding of the evolutionary re- data in hypothesis testing, and no quantitative lationships among known fossil and extant evaluation of the congruence of the morpho- hominoid taxa, little consensus has emerged logical and temporal data for the hominoid (e.g., Begun 1992a,b, 1994, 1995; Begun et al. fossil record has occurred. In an attempt to 1997a; de Bonis and Koufos 1993; Dean and clarify these evolutionary relationships and to Delson 1992; Andrews et al. 1996; MacLatchy evaluate congruence across data types, a phy- et al. 2000). Although their use in phylogenetic logenetic reassessment of the Hominoidea was analysis remains controversial, temporal data performed incorporating both morphological represent a potentially relevant class of infor- and stratigraphic data. mation that is often overlooked (Fisher 1991, Although cladistic hypotheses are evaluat- 1994; Huelsenbeck 1994; Wagner 1995). To ed with no explicit reference to time, all clad- q 2004 The Paleontological Society. All rights reserved. 0094-8373/04/3004-0007/$1.00 DATA CONGRUENCE IN HOMINOID PHYLOGENY 615 ograms make implicit statements about the ered under stratocladistic analysis, without si- relative time of divergence among taxa by the multaneously observing signi®cant decreases order of branching events (Fisher 1991; Wag- in the ®t of morphological data. In computer ner 1995). The sequence of branching events in simulations in which fossil records were cre- a morphological cladistic hypothesis is often ated for hypothetical taxa with known evolu- harmonized with the fossil record of the in- tionary histories, stratocladistics signi®cantly group through the creation of ``ghost lineag- outperformed conventional cladistic analysis es,'' arti®cial extensions of a taxon's range be- in recovering the true phylogenies (Fox et al. yond its observed ®rst appearance in the fossil 1999). Additionally, in cases where neither record (Norell 1993). This approach essential- method was able to recover the correct phy- ly erases any discrepancy between the ob- logeny, the stratocladistic hypotheses more served order of appearance events and the or- closely matched the true phylogenies (Fox et der implied by the hypothesis. Insofar as al. 1999). ghost lineages explain away discrepancies be- tween (stratigraphic) observation and (cladis- Methods tic) hypothesis, they may be considered ap- Morphological Data peals to ad hoc support, analogous to the way homoplasy is invoked to explain away mor- Thirteen fossil and ®ve extant genera of the phological data that are incongruent with a primate superfamily Hominoidea constituted cladistic hypothesis (Fisher 1991, 1994). the ingroup of this analysis. The morpholog- Stratocladistics is a parsimony-based crite- ical data matrix was modi®ed from the char- rion that evaluates competing phylogenetic acter-by-taxon matrix of Begun et al. (1997a). hypotheses relative to both morphological That study examined 240 morphological char- characters and a stratigraphic character de- acters for eight fossil and ®ve extant genera. rived from the stratigraphic record of the in- Modi®cations to the Begun et al. (1997a) data group taxa (Fisher 1991, 1992, 1994). The mor- set are discussed below. The character state phological component is evaluated as in con- descriptions and the character by taxon matrix ventional cladistic analyses, where each in- used in this phylogenetic analysis have been stance of homoplasy imparts a unit of included as Appendices 1 and 2, respectively. ``morphological parsimony debt'' upon the Added Taxa. The Begun et al. (1997a) ma- hypothesis. In addition, each instance of in- trix was ®rst expanded to include ®ve addi- congruence between the stratigraphic data tional taxa (Turkanapithecus, Equatorius [sensu and the hypothesis, where a taxon is predicted Ward et al. 1999], Griphopithecus, Morotopithe- to exist by the branching order yet is not ob- cus [sensu Gebo et al. 1997], and Ankarapithe- served, imparts a unit of ``stratigraphic par- cus). Morphological data for these taxa were simony debt'' (Fisher 1992). Stratocladistics obtained from the literature. Character coding then sums the morphological and stratigraph- for the added taxa preferentially followed cla- ic parsimony debt values of each hypothesis, distic treatments of these taxa in other studies creating a single ``total parsimony debt'' val- by the authors of the Begun et al. (1997a) anal- ue. The minimum total parsimony debt value ysis to ensure consistency in character coding. over the set of possible phylogenetic hypoth- Additional character state information was eses determines the overall most parsimoni- also incorporated from descriptions of holo- ous phylogenetic hypothesis (Fisher 1992, types and additional fossil material. 1994). Morotopithecus was coded following Begun Two recent studies have shown that strato- and GuÈlecË 1998 and Ward 1997a, and with de- cladistics may outperform conventional cla- scriptions of new postcranial fossils from the distic analyses in recovering evolutionary his- Moroto II locality by Gebo et al. (1997) and tories for fossil taxa. Working with published MacLatchy et al. (2000). Turkanapithecus was cladistic analyses, Clyde and Fisher (1997) coded following Rae 1997, Rose 1997, and noted signi®cant increases in the ®t of strati- Ward 1997a. Additional character information graphic data to phylogenic hypotheses recov- was obtained from Leakey et al. 1988. Follow- 616 JOHN A. FINARELLI AND WILLIAM C. CLYDE ing the separation of Equatorius from Kenyapi- logical trait, characters were consolidated to thecus by Ward et al. (1999), coding of Equa- avoid unduly weighting the impact of these torius followed the character coding for Keny- characters on the cladistic analysis. Combined apithecus africanus by Rose (1997) and Ward pairs (again following the numbering scheme (1997a). Additional character information was of Begun et al. 1997) include: characters