Phylogenetic Systematics

The phylogenetic history of any species will impact its ability to adapt to changing environmental-ecological conditions. A major problem in identifying phylogenetically significant anatomical features is the impact that function has on the development of morphological form. In order to differentiate between functionally and phylogenetically significant features, it is important to identify homologies and homoplasies within and between assumed closely related genera.

The definition of homology used in most paleontological studies is the phylogenetic concept. Homology implies shared common ancestral and shared derived characters; both are, of course, relative to a given point. The brain, for example, is a homologous character because it occurs in the same position in the body, protected by the skull and linked to the major sense organs according to a common pattern (R.D. Martin, 1990). While expansion and development of the frontal and temporal lobes of the brain may be considered synapomorphic characters uniting the fossil and extant hominins, a phylogenetic interpretation of homology is therefore to equate homology with shared ancestry. Homoplasy is the result of either morphological convergence or parallelism and not the result of immediate shared common ancestry.

How does one identify possible homologies and homoplasies? The best approach to tackling this problem is to use phylogenetic systematics (cladistics). Phylogenetic systematics is concerned with identifying the evolutionary branching sequence and establishing sister-group relationships between taxa. The allocation of fossil taxa to specific groups cannot be based on overall morphological similarity, because of the primitive retention of features as well as parallel evolution. In order to work out the phylogeny of a specific group, it is important that we reconstruct phyloge-nies independent of previously held assumptions.

Phylogenetic systematics enables us to identify likely homologies and homoplasies without reference to a previously constructed phylogeny, for it works in the following way: (1) Always assume that characters at the first instance are homologies; (2) use an outgroup comparison to distinguish more general (primitive) from more specialized (derived) features; (3) group species on the basis of shared derived features (synapomorphies); (4) in the event of conflicting information, tend to choose the phylogenetic relationships that are supported by the largest number of features; and (5) interpret inconsistent results, post hoc, as homoplasies. Thus homologies, which determine phylogenetic relationships, are identified a priori, without reference to a phylogeny, and later confirmed as homologies or reinterpreted as homoplasies (see Wiley, 1981; Brooks & McLennan, 1991).

Phylogenetics is based on the allocation of taxa to groups based on the recognition that they share unique characters (synapomorphies) that distinguish them from their ancestors as well as more distantly related groups. The identification of these derived characters is based on the study of an outgroup. An outgroup is a group of closely related taxa that are not directly part of the study group. If we were studying hominids, for example, the obvious outgroup would be the early Miocene hominoids and the Old World monkeys.

Synapomorphies establish sister-group relationships between taxa, while primitive features that are shared by at least two outgroups are assumed to be symplesiomorphies and cannot be used to allocate species to a clade. For example, when examining the New World and Old World monkeys, an obvious symplesiomorphic feature is the presence of a tail; but when the great apes are included in such a study, the absence of a tail is a synapomorphy linking all primates without tails as hominoids. When looking at a finer resolution of detail, let's say groups within the hominoids, the absence of a tail is now of no phylogenetic significance because all hominoids are defined by not having a tail. The absence of a tail in hominoids is meaningless in trying to determine which species are closely related. In this case the absence of a tail becomes a primitive, or plesiomorphic, feature of the hominoids. A synapomorphy of the hominins, however, might be an increase in brain size (relative to body weight).

Recently Collard and Wood (2000, 2001a, 2001b) have published a number of influential papers using parsimony analysis to reconstruct the phylogeny of extant great apes and papionins based on craniofacial morphology. These studies are used to help determine the consistency of phy-logenetic trees generated from morphological information against trees generated from molecular data. In all cases the morphological analyses fail to reproduce the molecular tree. They conclude that the four regions of the craniofacial complex that they studied were equally affected by homoplasy and were, therefore, "equally unreliable for phylogenetic reconstruction" (2001a: 167). We would suggest, however, that the inclusion of closely related fossil species is essential in helping to determine the correct polarity of characters. This is clearly the case in the significant impact that cranial expansion has had on hominin craniofacial morphology. Thus the recognition of "intermediate" stages in evolutionary development, which may be missing from extant studies alone, are likely to be crucial in helping to define the polarity of characters and their phylogenetic significance.

Fossils are the only direct morphological evidence that we have pertaining to the evolution of form, which enables us to examine likely selective process associated with adaptations. Hennig (1966) originally argued that the inclusion of fossil material within any phylogenetic analysis is crucial because it allows us to determine the likely "true" polarity of characters, which enables us to identify homoplasies and homologies (also see Wiley, 1981). The use of extant taxa alone will provide only a small sample of all the character combinations that have existed, and the most stringent test of character homology comes from including fossil and extant taxa in an analysis (Harvey & Pagel, 1998). Indeed, fossils that lie near the base of a cladogram are likely to be crucial, because these will tend to have retained more ancestral characters that have been modified in the later fossils and extant descendants (see Donaghue et al., 1989; M.V.H. Wilson, 1992; Kemp, 1999).

Gauthier et al. (1988), in their study of tetrapods, demonstrated that the inclusion of fossil taxa made an appreciable difference to the "correct interpretation" of the phylogeny, as suggested by extant molecular studies. For example, the use of 109 morphological characters from extant taxa alone suggested that mammals were the sister group to crocodiles plus birds and that turtles were the sister group of that clade. But when fossil taxa were included, mammals were placed as a sister group to all other tetrapods, and it was lizards that were placed as a sister group to crocodiles and birds (see also Donaghue et al., 1989; Novacek, 1992; M.V.H. Wilson, 1992; A.B. Smith, 1994; Kemp, 1999).

It is crucial that, prior to any study, the characters considered for analysis are themselves examined to determine their likely phylogenetic value. This includes determining whether characters are redundant. Character redundancy is the presence of characters whose overall appearance are intricately related and thus should be considered as representing just one character, such as robust canines and developed canine juga. In order to help identify redundancy, we must understand the functional and developmental significance of the phenotypic features that are being considered for analysis. Strait (2001) demonstrates such an approach, proposing a method for the integration of phenotypic characters into functional and developmental complexes in an examination of the hominin cranial base. His method is based on the construction of a data matrix, a functional/structural analysis of metric values, a character analysis to help identify phylogenetic independence of characters within the matrix (with functional subsets identified), followed by a parsimony analysis of the proposed integrated complexes. In using the cranial base as an example, Strait was able to show, using factor analysis and the resulting generated correlation matrices, how some characters are correlated. From this he was able to identify three hierarchical inferences of integration: uniform, compatible, and functionally weighted complexes. From the character analysis, a number of procedures followed, including character deletion, replacing character states with a binary system generated from parsimony analysis of functional "subsets," and the differential weighting of some characters. Strait was able to apply this method successfully to the cranial base, although he found very few characters that could be considered truly integrated.

While it was tempting to try to adopt aspects of Strait's method here, it was finally deemed unwise because it would require the construction of a number of hypothesized and untestable integrative complexes that could not be substantiated by factor analysis (due to missing variables). Indeed, as Strait warns, "uncritical hypotheses of integration could also have a deleterious effect on phylogenetic analyses. . . . [T]he concepts of integration should be used in phylogenetic analysis only if hypotheses of integration can be adequately tested" (2001:294). We explain in the appendix our attempt to identify likely redundant/duplication of features.

Skelton and McHenry (1992, 1998), believing that the masticatory apparatus is dominated by homoplasies, attempted to remove the "bias" of this complex by reducing masticatory features in their parsimony analyses. In doing so, they concluded that P. walkeri should be removed from the robust australopithecine clade, because the synapomorphies uniting these taxa were demonstrated to be largely masticatory characters. They suggest that, when this "bias" is removed, KNM-WT 17000 becomes a more primitive hominin rather than part of a Paranthropus clade. Characters helping to define the masticatory apparatus, especially at the species or genus level, we would argue, however, must be considered as part of the phylogenetic history of the taxa concerned, for there is no evidence to suggest that the masticatory apparatus is more prone to homoplasy than any other anatomical region (see Strait et al., 1997; Collard & Wood, 2000, 2001a, 2001b). The studies of Collard and Wood have concluded that the phylogenetic information supplied by the masticatory complex should not be underestimated. They show that characters not directly associated with the masticatory apparatus are no more reliable for phylogenetic reconstruction than are characters often used to help define this complex (see also Strait et al., 1997). This can be tied to the concept of phylogenetic niche conservatism.

The principle of phylogenetic niche conservatism is based on the proposition that past and present members of a lineage are likely to have occupied similar environments, because only those species that are best suited to particular environments are likely to survive (Harvey & Pagel, 1998). For example, the masticatory correlates between Paranthropus species of eastern and southern Africa suggest that not only did these species at some time in the past, share a similar environmental niche, but they also shared an immediate common ancestor. Their features must have at least partially evolved as a common adaptive response to similar adaptive pressures that were also operating on the last common ancestor. In other words, it is likely that the speciation event resulting in P robustus and P boisei was not related to a dramatic shift from a forest to a desert environment (thus resulting in a functionally distinct morphology), because it is unlikely that any ancestral population could survive such a dramatic environmental shift. It is more likely that an ancestral population moved into a slightly more or less wooded environment, because this is the adaptive strategy of its immediate common ancestor. This pattern of speciation is also closely aligned with vicariance, in which the ancestral species extends its range under favorable and specific environmental circumstances and then becomes geographically restricted to paleohabitats by geographic and/or environmental events (see Andrews & Bernor, 1999; Strait & Wood, 1999). There is a real danger in trying to "disentangle" processes and patterns that are intimately related. This is because numerous functional "homoplasies" may in fact be homologies. As such, studies that attempt to disentangle function from phylogeny may be "throwing out the baby with the bathwater."


The taxa and the specimens used to define the species examined are listed in Table 5.1. We have been intentionally conservative in the allocation of specimens to taxa. Given the current taxonomic confusion concerning both OH 9 and OH 62, these are not considered. The same applies to the early Homo specimens from Georgia (H. georgicus) and the recently reported material from Danakil (Afar) and Bouri (Middle Awash), both from Ethiopia (see Abbate et al., 1998; Gabunia et al., 2000a; Asfaw et al., 2002). Currently only one specimen of P. walkeri is recognized here (KNNM-WT 17000), and none of the lower Omo "robust" mandibles are considered. Given the poor preservation of the specimens from Chad allocated to A. bahrelghazali, they have not been included in this study.

Because Sahelanthropus is currently dated to the Late Miocene of Africa, the outgroup consists of Middle and Late Miocene African genera

TABLE 5.1 ► Specimens Included in the Hypodigm of the Hominidae


Middle Miocene specimens from Fort Ternan, Kenya.


Miocene specimens from Hungary and Spain.


Miocene specimens from northern Greece.


Late Miocene specimens from the Djurab Desert, northern Chad.

Ardipithecus ramidus:

Specimens from Pliocene strata at Aramis, Middle Awash, Ethiopia.


Specimens from Allia Bay, Kenya.


0 0

Post a comment