The validity of incorporating temporal data in phylogenetic analysis has been an area of recent debate among researchers interested in the reconstruction of the evolutionary history. While arguments have been made that the incomplete nature of the fossil record could bias the results of analyses incorporating stratigraphic data in hypothesis testing, several researchers have proposed methods for doing so. One such method is stratocladistics, which incorporates temporal data in the form of their sequence of appearance events in the stratigraphic record. Stratocladistics evaluates the morphological and temporal data simultaneously in hypothesis testing under a parsimony criterion, giving priority to neither class of data.

Morphologic and stratigraphic data were compiled from the primary literature for five extant and thirteen fossil taxa at the genus-level for the primate Superfamily Hominoidea. The morphological data matrix consisted of 200 characters. Stratigraphic data consisted of first and last appearance events for each taxon correlated to the paleomagnetic timescale. The stratigraphic data divides the stratigraphic range of the Hominoidea into fifteen distinct stratigraphic levels. These data were used to perform cladistic and stratocladistic analyses in an effort to reconstruct the evolutionary history of the Hominoidea.

The morphological analyses provide insight into possible evolutionary relationships among the hominoids. Cladistic analysis of the morphological data matrix recovered four most parsimonious cladograms, indicating two distinct evolutionary radiations of primates by the earliest Miocene. Subsequent analyses for the strength of support for topologies proposed by the most parsimonious cladograms indicate support in the morphological data for a postcranially primitive "parahominoid" clade, comprised of early to middle Miocene hominoids from eastern Africa, and a postcranially derived "euhominoid" clade delineated by the derived postcranial morphology witnessed in the early Miocene hominoid Morotopithecus. Included below is the preferred morphological cladogram for this study. The "euhominoid" clade includes all extant and middle to late Eurasian hominoids. However, the strength of support for this division of the hominoids suffers from the fragmentary (proportion of preserved anatomy) fossil record for the majority of the early to middle Miocene hominoids. Strong morphological support also exists for the separation of African great apes (Homininae) from Asian great apes (Ponginae).

Incorporating the stratigraphic data to perform a stratocladistic analysis reveals a single most parsimonious phylogenetic tree. Relative to the morphological data, this tree is equally as parsimonious as the conservative morphologic hypothesis (above). However, it is topologically different in several significant areas. These topological differences highlight areas of significant incongruence between the morphologic and stratigraphic information preserved in the fossil record. Below, the stratigraphic ranges of each taxon are plotted relative to (A) the morphologically preferred cladogram, and (B) the stratocladistic cladogram. The erasure of the "parahominoid"/"euhominoid" distinction (C) occurs in the stratocladistic hypothesis, where East African taxa are interpreted to form a continuous hominoid lineage through the early Miocene. This difference points to an incomplete fossil record in eastern Africa in the early Miocene. However, this incompleteness may be either in the form of missing ranges of fossil taxa as implied by (A), or in the form of missing anatomy of already known fossil taxa as implied by (B). Additionally, a similar discrepancy exists in the late Miocene with the variable position of Lufengpithecus and Oreopithecus (D).

A close examination of the morphologically preferred hypothesis (above) indicates that there is a larger degree of homoplasy recorded in the dental characters than in the postcranial skeletal characters. However, the postcrania are both more poorly preserved in the fossil record and represent a significantly larger proportion of the characters coded in the data matrix. Both of these factors could have important effects on observed homoplasy. Low preservation in the fossil record could mask actual homoplasy, therefore the lower amount of homoplasy observed in the postcrania could be a function of their incompleteness. Additionally, the proportion of characters in the data matrix strongly influences the optimized topology. Thus, with a greater proportion of characters in the data set, the morphologically preferred cladogram could be fundamentally biased in favor of the postcrania and against dental characters, serving to raise the level of homoplasy observed in dental characters.

One hundred rarefactions were performed to eliminate potential biases from these variables to determine if the pattern of higher homoplasy in dental characters was a real phenomenon. The completeness of the dental characters was randomly degraded to that of the postcranial characters, and character weighting was used to equalize the input of both sets of characters in determining the optimal cladogram. These analyses show that the lower homoplasy in postcrania is not a function of biases in preservation. Below, is a histogram of the results for the rarefaction analyses . These results indicate that postcranial characters have higher RI scores (higher RI scores imply lower homoplasy) than dental characters. The results are highly significant (P << 0.0001) and imply postcranial characters could potentially be the more reliable indicators of phylogeny.

Download various programs for generating randomly re-sampled distributions around discrete data sets. Program routines include bootstrapping and jackknifing, and a program designed to created rarefactions of cladistic character-by-taxon matricies. Executable and Readme file (in .pdf format) are included in the following WinZip files. Feel free to use these programs as you please.

• Boot Version 3.0         (137 Kb)
• Jacknife Version 1.0   (137 Kb)
• Rarefact Version 2.5   (138 Kb)