Henry M. McHenry
Department of Anthropology
University of California
Davis, CA 95616
We applaud the effort by Strait, Grine and Moniz (1997) to resolve problems related to hominid phylogeny. Their exploration of methods for handling variable traits is especially welcome. Their results confirm most of the relationships between the early hominids proposed by Skelton and McHenry (1992). The most significant difference between the results of Strait, Grine and Moniz (1997) and those of Skelton and McHenry (1992) is in the relationship of A. aethiopicus to the other early hominids. In the analysis of Strait, Grine and Moniz (1997) A. aethiopicus was grouped with A. robustus and A. boisei to yield a monophyletic robust clade. In the analysis of Skelton and McHenry (1992) A. aethiopicus was placed as an early side branch, the sister clade of a group of hominids including A. africanus, A. robustus, A. boisei, and early Homo.
We believe that this relatively minor difference between the two analyses is primarily due to different assumptions and procedures related to the effect of trait list bias caused by 1) intercorrelations among traits and 2) unequal representation of functional evolutionary trends. Skelton and McHenry (1992) and Skelton et al. (1986) discussed these problems and attempted to compensate for them by exploring various methods of combining traits into trait complexes. Strait, Grine and Moniz (1997) acknowledge the problem, but do not take any steps to compensate for it because they believe that it is risky to base a phylogenetic analysis on untested functional and structural relationships.
We agree that some of the functional and structural hypotheses proposed by Skelton and McHenry (1992) are untested, and some are likely to be false. We readily acknowledge that combining traits into trait complexes may not be the best procedure for compensating for trait list bias. We believe, however, that trait list bias poses at least as great a risk to the reliability of a phylogenetic analysis as uncertainty of functional relationships, and that an analysis that fails to address this source of error can not be considered completely reliable.
Strait, Grine and Moniz (1997) did perform one analysis as a check on whether the preponderance of traits related to mastication had an effect on their result. In this analysis they omitted 30 traits that they believed were related to mastication, yet obtained a result similar to that obtained using the entire trait list. Although this was a commendable effort, we believe that their criteria for selecting traits related to mastication was too simple. Although most of the traits they omitted do relate to heavy chewing, they also omitted several traits that we believe relate to the evolution of the anterior dentition, basicranial flexion, and prognathism/orthognathism.
To illustrate the fact that compensating for trait list bias can yield a different result we herein re-analyze the data of Strait, Grine and Moniz (1997) using a method of grouping traits into complexes similar to that of Skelton and McHenry (1992). We also repeat the analysis of Strait, Grine and Moniz (1997) that omits traits related to heavy chewing, but omit a different set of traits which we believe more narrowly corresponds to those traits more directly related to the function of heavy chewing.
We used the taxa, traits, and trait scores as presented by Strait, Grine and Moniz (1997). Although we do not agree completely with all of the traits and scores, this trait list is, overall, an excellent revision and expansion of the trait list used by Skelton and McHenry (1992).
We assigned as many traits as possible to the five main functional complexes described by Skelton and McHenry (1992): heavy chewing, anterior dentition, basicranial flexion, prognathism/orthognathism, and encephalization. Traits that we could not unambiguously assign to one of these five complexes were omitted from the analysis. We were able to assign 48 of the 60 traits to complexes based on the descriptions of the relationships of the traits to those of Skelton and McHenry (1992) provided by Strait, Grine and Moniz (1997) or logical functional relationships similar to those employed by Skelton and McHenry (1992). This process is partly subjective, yet we believe that at least most of the 48 traits were assigned to appropriate complexes. Our assignment of traits to complexes is presented in Table 1.
Table 1: Traits Grouped by Functional Complex
Functional Complex Strait, Grine, and Moniz Number of
(1997) Trait Number Traits in
Complex
Heavy Chewing 2, 3, 4, 6, 7, 8, 9, 10, 11, 15, 19, 20, 27
23, 24, 25, 26, 28, 44, 47, 48, 53, 54, 55,
56, 57, 59, 60
Anterior Dentition 49, 50, 51, 52, 58 5
Basicranial Flexion 21, 22, 30, 31, 32, 33, 34, 38, 40, 41, 42 11
Prognathism/ 13, 14, 15 3
Orthognathism
Encephalization 16, 17 2
Phylogenetic analysis was performed using PAUP 3.0q. We set all PAUP options as described by Strait, Grine and Moniz (1997) for their "ORDERED" analysis. Taxon scores for each complex were generated using the procedure described by Skelton and McHenry (1992), except that in cases where multiple equally parsimonious trees were produced a 50% majority consensus was obtained rather than a strict consensus. The scores assigned by Strait, Grine and Moniz (1997) for Gorilla on some of the traits related to heavy chewing were problematic, because absolute sizes were scored for such traits and premolar and molar crown area without scaling for body size. Therefore we omitted Gorilla from the analysis of the heavy chewing traits.
The number of equally parsimonious trees obtained for each complex is shown in Table 2, along with the consistency index, homoplasy index, retention index, and rescaled consistency index values for the trees. In general, the equally parsimonious cladograms for a complex differed in their placement of the Homo species. Complex scores for each taxon are shown in Table 3.
Table 2: Number of Equally Parsimonious Trees and Indices
Complex Number of Consistency Homoplasy Retention Rescaled
Equally Index (CI) Index (HI) Index (RI) Consistency
Parsimonious Index (RC)
Trees
Obtained
Heavy Chewing 9 0.739 0.261 0.812 0.601
Anterior 2 1.000 0.000 1.000 1.000
Dentition
Basicranial 9 0.727 0.273 0.820 0.596
Flexion
Prognathism/ 6 0.714 0.216 0.833 0.695
Orthognathism
Encephalization 1 1.000 0.000 1.000 1.000
Table 3: Complex Scores by Taxon
Taxon Heavy Anterior Basicranial Prognathism/ Encephalization
Chewing Dentition Flexion Orthognathism
Pan/Gorilla 0 0 0 0 0
A. afarensis 0 1 1 1 0
A. 6 3 2 1 0
aethiopicus
A. africanus 4 2 3 2 1
A. robustus 5 5 6 3 2
A. boisei 6 5 6 3 2
H. habilis 1 3 5 3 3
H. 3 3 3 3 4
rudolfensis
H. ergaster 2 3 4 4 4
H. sapiens 2 4 4 4 4
The complex scores were entered as data and analyzed using PAUP. The cladograms for the heavy chewing and basicranial flexion complexes suggest that the robusts (at least A. robustus and A. boisei) and the Homo species had independently derived character states. Therefore, we analyzed the scores for these two complexes as unordered traits and the scores for the remaining three complexes as ordered traits.
We repeated the NON-MASTICATORY analysis of Strait, Grine and Moniz (1997), but omitted the 27 traits (listed in Table 1) that we believe are related to heavy chewing instead of the 30 traits chosen by Strait, Grine and Moniz (1997). We agree with Strait, Grine and Moniz (1997) that the choice of characters to omit is subjective.
The analysis of complex scores produced the cladogram shown in Figure 1. This cladogram places A. aethiopicus as a sister to a clade including A. africanus, A. robustus, A. boisei, and the Homo species. This is the same relationship presented by Skelton and McHenry (1992), and suggests that the robust species (aethiopicus, robustus, and boisei) do not constitute a monophyletic group. The cladogram produced by the ORDERED analysis of Strait, Grine and Moniz is reproduced in Figure 2 for comparison.
Our revision of the NON-MASTICATORY analysis of Strait, Grine and Moniz (1997) produced four equally parsimonious trees with the following descriptive statistics: CI = 0.632, HI = 0.368, RI = 0.740, RC = 0.468. A 50% majority consensus of these trees is shown in Figure 3. This cladogram differs from that in Figure 1 in the placement of H. habilis and H. rudolfensis.
The results obtained support the hypothesis that trait list bias can affect the results of a phylogenetic analysis and that different results may be obtained depending upon whether or not an attempt is made to compensate for trait list bias. In conjunction with the analysis of Strait, Grine and Moniz, these results suggest that the relationships of A. aethiopicus, H. habilis and H. rudolfensis among the early hominids are less certain that those of the other species considered.
Although the present analysis produced cladograms more consistent with the phylogenetic scheme of Skelton and McHenry (1992) than that of Strait, Grine and Moniz (1997), expecially with respect to the placement of A. aethiopicus, we nonetheless believe that the result consistently obtained by Strait, Grine and Moniz (1997) is worthy of at least equal consideration. These two schemes should probably be considered equally valid competing hypotheses until the methodological problems relating to trait list bias and testing of functional relationships among traits are resolved.
Given the uncertainty of the relationship of A. aethiopicus to the other early hominids, widespread adoption of the taxonomic scheme proposed by Strait, Grine and Moniz (1997), which is based on the premise of a monophyletic robust clade, would probably be premature.
Figure 1: Cladogram Resulting from Functional Complexes
Pan/Gorilla
A. afarensis
A. aethiopicus
A. africanus
A. robustus
A. boisei
H. habilis
H. rudolfensis
H. ergaster
H. sapiens
Figure 2: ORDERED Analysis of Strait, Grine and Moniz (1997)
Pan/Gorilla
A. afarensis
A. africanus
A. aethiopicus
A. robustus
A. boisei
H. habilis
H. rudolfensis
H. ergaster
H. sapiens
Figure 3: Cladogram produced when 27 traits related to heavy
chewing are omitted.
Pan/Gorilla
A. afarensis
A. aethiopicus
A. africanus
A. robustus
A. boisei
H. habilis
H. rudolfensis
H. ergaster
H. sapiens
REFERENCES CITES
Skelton, R. R. & McHenry, H. M. (1992). Evolutionary relationships among early hominids. J. Human Evol. 23, 309-349.
Skelton, R. R., McHenry, H. M. & Drawhorn, G. M. (1986). Phylogenetic analysis of early hominids. Curr. Anthropol. 27, 21-43.
Strait, D. S., Grine, F. E. & Moniz, M. A. (1997). A reappraisal of early hominid phylogeny. J. Human Evol. 32, 17-82.