Morphology-based systematics (MBS) and problems with fossil hominoid and hominid systematics

Authors


Abstract

The generalized/primitive nature of the hominoid dentition and often fragmentary nature of fossils, coupled with enthusiastic optimism for making revolutionary finds, has wreaked havoc with recognition of early human ancestors and reconstruction of fossil hominoid phylogeny. As such, the history of paleoanthropology is one of repeated misidentification of fossil ancestors and of occasional fraud. Although this history has led many workers to lose confidence in morphology based systematics (MBS), past and present misidentifications are actually due to a disregard of systematic methodology. Systematics depends on the continuity of life and gains its objectivity largely from the order alpha taxonomy imposes on morphologic discontinuities in closely related taxa (i.e., species and genera). Transformation of characters fixed in species into character complexes, as manifested in taxa nested at different levels of relationship, form the foundation for higher-level taxonomy and for phylogeny. Because in most cases, hominoid fossils are unable to provide the data needed to resolve alpha taxonomy, classification and phylogeny of fossil taxa must be guided by analogies to living taxa. Hominid and hominoid fossil taxonomy and phylogeny, however, has been based largely on preevolutionary notions and on misinterpretations of the polarity of assumed diagnostic characters. More often than not, fossils lack resolution for the taxonomic level or rank they are assigned to and taxa are erected without appropriate analogies to living forms. As such, phylogenies based on these classifications are unlikely to be correct. More in-depth anatomical studies that are in accordance with systematic methodology are likely to hold the key to correctly classifying fossils and unraveling hominoid and hominid phylogeny. Anat Rec (New Anat) 269:50–66, 2002. © 2002 Wiley-Liss, Inc.

INTRODUCTION

One of the main tasks of paleoanthropology is to classify fossil hominid remains, and unravel their phylogeny. As a long-established science, morphology based systematics (MBS) has provided effective tools for accomplishing this task. The primitive/generalized nature of the hominoid dentition and the often-fragmentary nature of fossils (Sarmiento, 1987), both coupled with enthusiastic optimism for making revolutionary finds, have played havoc with the recognition of early human ancestors and reconstruction of fossil hominoid phylogeny (see Walker, 1969; Pilbeam, 1970, 1972; Simons, 1965, 1972; Szalay and Delson, 1979; Begun, 1992; Conroy et al., 1992; White et al., 1994).

On the basis of fossil evidence, the human lineage until recently was believed to extend back to the middle to late Miocene (Keith, 1915; Simons, 1967, 1972; Leakey, 1967, 1968; Pilbeam, 1968, 1969, 1970). Gorillas and chimpanzees were envisioned to be separate lineages in the early Miocene (Proconsul major [gorilla]; Walker and Rose, 1968; Simons and Pilbeam, 1965; Pilbeam, 1969; and Proconsul africanus [chimpanzee] Pilbeam, 1969), and great apes (Aegyptopithecus; Simons, 1965, 1972) and purported hominids (Propliopithecus; Schlosser and von Zittel, 1923; Gregory, 1934; Simons, 1963; Pilbeam, 1967, 1968) were believed to appear during the Eocene or near the Eocene-Oligocene boundary (Osborn, 1922; Kappelman et al., 1992; Van Couvering and Harris, 1993). For nearly four decades, most paleoanthropologists regarded the doctored remains of Piltdown as the earliest human ancestor (Woodward, 1917; Eliott-Smith, 1924; Keith, 1934, 1940; Oakley and Hoskins, 1950; Oakley, 1952). Yet, they were initially unwilling to recognize the possibility that the South African australopithecines (Keith, 1925, 1940; Eliott-Smith, 1925; Woodward, 1925) and the Trinil remains (Virchow, 1895) were hominids.

When more fossil material became known and more sober assessments were made, none of the early hominoid phylogenetic claims were supported, and in retrospect, some seem absurd (Gingerich, 1973; Sarmiento, 1987, 1995; Pilbeam et al., 1990). Piltdown was exposed to be a hoax constructed from the doctored remains of a modern human and an orangutan (Weiner et al., 1953; Weiner, 1955). Thereabouts, australopithecines became widely accepted as early human ancestors with little objective evaluation (Mayr, 1950; Le Gros Clark, 1950a,b, 1955; Oakley, 1954; Von Koenigswald, 1956; Simpson, 1963; Robinson, 1954, 1962, 1965).

The rapidly changing behavioral and morphologic interpretations of the hominoid fossil record that have since ensued have caused considerable confusion among paleoanthropologists (Larson, 1998; Foley, 2001) and have led many to lose confidence in MBS (Chaline, 1999; Lockwood and Fleagle, 1999; Wood and Collard, 1999; Collard and Wood, 2000). Partly, this loss of confidence centers on the very meaning of homology and the inherent problems of identifying homologies and separating these from parallelisms/convergences and reversals (collectively termed homoplasies; Hall, 1994; Lieberman, 1999). This has spurred a recent interest in turning from MBS toward biomolecular analysis (Ruvolo et al., 1991, 1993, 1994; Pilbeam, 2000) or in couching morphology based analyses within (1) development (Lieberman, 1995; Lieberman et al., 1996; McCollum, 1999, 2000), (2) phenetics (Howells, 1973, 1989), or (3) cladistic techniques based exclusively on fossils and quantitative characters (Chamberlain and Wood, 1987; Lieberman et al., 1996). Some paleoanthropologists have suggested that phylogenetic analyses should be based solely on categories of skeletal limb characters weighed according to their own interpretations of function and developmental patterns (see Lovejoy et al., 1999, 2000). As a result, some workers seem to suggest that systematics and its progress are based more on general consensus rather than the logical organization of empirical evidence (Franciscus, 1999). Past misinterpretations of the hominoid fossil evidence, however, do not reflect the shortcomings of MBS but have occurred mainly because of misunderstandings in its methodology and theoretical basis. This essay presents a review of the reasoning underlying MBS to (1) show its usefulness for interpreting hominoid fossil evidence; (2) explain past misinterpretations and misclassification of fossils; and (3) provide a framework that serves as a guideline for future studies in hominoid systematics.

Past misinterpretations of hominoid fossil evidence do not reflect the shortcomings of morphology based systematics but have occurred mainly because of misunderstandings in its methodology and theoretical basis.

Because anatomical characters form the foundation for vertebrate systematic studies and are useful for untangling relationships between taxa at all levels of relatedness, anatomists stand to make significant contributions to hominoid and hominid systematics. It is the overall goal of this essay to provide anatomists with both the tools and the incentive to do so.

THE POWER OF MORPHOLOGY-BASED SYSTEMATICS

The moment one willingly accepts that all living organisms share a common origin (Darwin, 1859; Huxley, 1895), the powers of systematic studies are obvious (Simpson, 1961; Mayr, 1963). At its lowest level, each taxon serves as a natural experiment that explores adaptive possibilities afforded to it by the environment and, thus, reveals evolutionary processes. Closely related species arrive at different solutions to environmental problems by using the inherent variation in a largely shared phenotype. Thus, studies of closely related species reveal the type of characters underlying overall differences in phenotypic expression, and the anatomical regions where these characters are found. When linked to ecological data for each member taxon, these studies have the potential to reveal the associated environmental differences and, thus, the variables bearing on speciation for the concerned taxonomic group (Vrba et al., 1995; Sarmiento et al., 1996; Sarmiento and Oates, 2000). Examination of nested taxa with varying relationships at ascending levels of classification provide insights into how (1) initial characters marking species divergence may become evolutionary trends, (2) evolutionary trends may be molded through time into the complex structures that define higher taxonomic categories, (3) complex structures once incorporated in ancestral lineages are subsequently modified to best suit the needs of descendant taxa. Founded on the continuity of all life forms, systematics assumes that existing morphologic discontinuities in living (crown) taxa at various levels of relatedness must have been bridged by intervening populations, which are presently extinct.

LOWER LEVEL (ALPHA) TAXONOMY

Alpha taxonomy is concerned with the links among geographic distribution; reproductive communality; speciation; relative evolutionary time; and behavioral morphologic, and adaptive differences in closely related taxa (usually within the same genus). It serves as the basis for systematics and as the building blocks for all higher-level taxonomy or phylogeny. Alpha taxonomy bridges the gaps between individuals, populations, and species and strives to sort phenotypic variation due to sex, developmental age, and environmental/habitat variables (e.g., altitude, temperature, rainfall, cloud cover, vegetation, etc.), arriving at the limits of intraspecific variation. It is therefore essential for identifying those characters (behavioral, morphologic, or genetic) associated with reproductive and phenotypic discontinuity (i.e., distinctiveness) and for distinguishing species-specific characters from intraspecific variation within a taxonomic group (Schuh, 2000). By exploring the minimum differences representing reproductive and phenotypic discontinuities, alpha taxonomy allows the recognition of characters and associated adaptations that define closely related species and, thus, the variables associated with species divergence.

Species, and the characters that define them, may be objectively arrived at based on whether or not distinct populations maintain their distinctiveness in areas of sympatry (Fig. 1; Simpson, 1961; Mayr, 1963; see also, Claridge et al., 1997; Lherminer and Solignac, 2000) and are diagnostically different (Cracraft, 1983). Among primates, closely related species usually exhibit frequency differences in specific characters, and/or differences in absolute size and proportions of structures, or of localized anatomy (Schultz, 1930, 1936, 1963). The largely similar phenotypes of closely related taxa forces systematists to focus on divergent morphology and magnitude of differences (discontinuities) to define lower taxonomic levels (Simpson, 1961; Mayr, 1963), for taxa that are not based on objective criteria (i.e., subspecies, allopatric species, genus, and to some degree family). Usually, the greater the magnitude of morphologic differences between closely related taxa, the higher the corresponding taxonomic category.* Because alpha taxonomy concerns itself with the fundamentals of identifying species, and species are in theory the most objective of all taxonomic categories (naturally defined by reproductive communality and phenotypic similarity), alpha taxonomy imparts some objectivity to all higher level taxonomy.

Figure 1.

Geographic distribution of individuals (dots) belonging to three populations. The red and yellow populations maintain their distinctiveness in overlapping areas and are, thus, by definition different species. Without a range of overlap, there is no objective test to decide whether blue and red populations are different species. This decision is in part subjective and depends on whether or not magnitude of differences distinguishing the two is comparable to, or more than that distinguishing objectively determined species (i.e. the red and yellow populations). Species determination for blue and yellow populations is also partly subjective. It depends on the magnitude of differences between the two and the degree to which their hybrids show grading of diagnostic differences. A gradual cline with a wide zone of hybridization between them suggests one species. A very abrupt cline with a narrow hybrid zone as shown here indicates different species.

HIGHER LEVEL TAXONOMY

With the advent of speciation and associated reproductive isolation, morphologic differences (i.e., discontinuities) greater than those initially marking the divergence of species accrue with time—either guided by selection or meandering due to drift—and with the subsequent divergence of additional species from the ancestral lineage. Unlike the localized and quantitative differences distinguishing closely related species, the distinctive (i.e., diagnostic) morphology characteristic of higher taxonomic categories may be reflected throughout the organism's anatomy. It may involve the appearance of novel anatomical elements (e.g., bones, ligaments, muscles, arteries, and nerves) and rearrangement of anatomical elements to form new structures.

Categories higher than species are defined by the divergence of descendant lineages subsequent to that point in time when structural innovations are achieved (Mayr, 1963; McKenna and Bell, 1997). In this regard, these categories often reflect the success of novel structures in solving particular physiological or environmental problems and the subsequent evolutionary radiation within a lineage that typically follows such a solution. Large discontinuities characteristic of the highest taxonomic levels are correlates of very successful solutions. In time, the strong competitive edge these solutions afford results in large discontinuities between a successful lineage and its closest related outgroup. The number of structural innovations in a lineage and the associated proliferation of descendant lineages, therefore, determine how finely a higher taxonomic group (i.e., an ancestral lineage and its descendants) may be divided into hierarchical categories (Simpson, 1961; Mayr, 1963). Guided by consistency in a classification, the taxonomist decides to what taxonomic level above species any one magnitude of morphologic difference (or morphologic complexity for categories above the family level) is assigned.* Because morphologic changes accrue over generations (a factor of time), increasingly higher taxonomic levels are distinguished by increasingly greater morphologic discontinuities (Simpson, 1961; Mayr, 1963). The widely divergent phenotypes of species comprising a higher-level taxon (above family) usually require that higher categories be based on diagnostic characters (morphologic complexes or qualitative characters). Because morphologic differences between species comprising higher level taxa may be pervasive throughout the anatomy, overall quantitative differences are rendered superfluous for classification at higher levels.

Analysis of characters diagnostic of ascending taxonomic categories within nested groups at various hierarchical levels provides insight into how the characters defining differences in closely related species are transformed through time into diagnostic structures or morphologic complexes.

PHYLOGENY

Reconstruction of phylogeny depends on taxonomy and relies on the morphologic discontinuities that exist among taxa. Without discontinuities, neither taxonomy nor phylogeny would be possible. There would be conflation of ancestor and descendant taxa and of the evolutionary sequence in which characters and structures develop (i.e., polarity). Although direction of evolutionary change and ancestor-descendant relationships are in part implied by increasing complexity of structure, structural simplification and/or loss would confound interpretations on the direction of evolutionary change. Out of necessity, therefore, direction of change and identification of ancestor-descendant taxa would be subjective and dependent on the evaluation of the taxonomist. The loss of intervening populations and the morphologic discontinuities that, over time, result from such a loss permit the reconstruction of phylogeny and in part the determination of the direction of evolutionary change. Because the loss of intervening populations does not follow any set pattern, the taxa existing at any one point in time are nested within phylogenetic groups united at different hierarchical levels, and show varying degrees of relationship and varying degrees of morphologic discontinuity. Nesting of taxa within groups and designation of outgroups is as central to phylogenetic reconstruction at higher levels as is morphologic discontinuity.

To emphasize relationships as opposed to summarizing degree of differences, phylogenetic reconstructions are based on shared characters (Tattersall and Eldredge, 1977). Unlike taxonomic categories, phylogenetic groups are solely defined by the shared characters of its member taxa (Hennig, 1966). Characters acquired and modified by descendant taxa after their divergence from the common ancestor of the group (autapomorphies) cannot affect the common phylogenetic history of the group, no matter how strongly derived these characters may be. Markedly derived characters, however, may affect the taxonomic level a group is assigned to. Regardless, the characters used in phylogeny, and their identification as shared characters, ultimately relies on taxonomic studies distinguishing species and making evident intra- and interspecific character variation.

At intermediate taxonomic levels (e.g., suborder and family), it may be uncertain whether characters present in a phylogenetic group's crown species were also present in the groups stem ancestor and in the earliest members of each of its crown lineages. At these levels, morphologic discontinuities may be relatively too great, “shared” characters too labile, and the number of taxa and/or hierarchical levels within a group too few to provide certainty. With the corresponding decrease in complexity (i.e., fewer interrelated anatomical elements) of distinctive morphology distinguishing member taxa at decreasingly lower taxonomic levels, homoplasies are increasingly more difficult to detect (Gregory, 1934; Sarmiento, 1987; McHenry, 1996; Lockwood and Fleagle, 1999). Divergence of descendant taxa may be too recent a phenomenon to enable the evolution of a complex morphology (Schultz, 1936; Sarmiento, 1987, 1988, 1998) so that there is no complexity to leave evidence of homoplasy (Szalay, 1976; Eldredge and Cracraft, 1980; Bock, 1981; Szalay and Bock, 1991). Moreover, closely related taxa are prone to acquiring homoplasies, because they are predisposed by a largely shared phenotype to arrive at similar solutions when encountering similar problems in the environment. As such, the idea that characters common to all members of a taxonomic group are homologous is an assumption, which makes phylogenetic reconstruction a theoretical endeavor. As theories, phylogenies are subject to testing and revision with accumulation of new data.

Phylogenetic reconstruction is a theoretical endeavor. As theories, phylogenies are subject to testing and revision with accumulation of new data.

CLADISTICS

To sidestep problems of homoplasies in phylogenetic reconstruction, cladistic analyses assume all apparent similarities are homologies and resolve character conflicts by using the principle of parsimony (Hennig, 1966; Forey et al., 1992; Schuh, 2000). The usefulness of cladistic logic lies in its ability to identify possible homoplasies among assumed homologies (Hennig, 1966; Kluge, 1983; McHenry, 1996). Testing of key homologies (i.e., those which are critical for defining any one clade), and their rejection as homologies, subsequently leads to alternate phylogenies and alternate homologies for further testing. Once identified, possible homologies can be tested through (1) more rigorous character analyses comprising character function and underlying anatomical details (Davis, 1964; Packer and Sarmiento, 1984; Sarmiento, 1987), (2) comparisons of character lability within closely related groups at the same taxonomic levels (Mayr, 1963), (3) reconstruction of evolutionary stages pitting structural continuity of functional complexes against ecological transitions and adaptive shifts. (Sarmiento, 1995, 1998; Nielsen, 1998), (4) fossil evidence.

Ultimately, such a process leads to the most likely phylogeny, one that assumes a minimum number of homoplasies and assumes them for characters that (1) are labile at the taxonomic levels concerned; (2) show no clear polarity in member taxa; (3) are expected to be paralleled or reversed, given the functional constraints of evolutionary stages hypothesized for descendant lineages (Sarmiento, 1987, 1995, 1998; Lockwood and Fleagle, 1999).

In cases, where hypothesized homoplasies do not conform to any of the above criteria, they are still assumed homoplasies if the phylogenetic signal they provide conflicts with that provided by the majority of characters. Where there is no overwhelming majority or minority signal and none of the conflicting characters or character sets fit “homoplasy criteria,” the phylogeny may not be resolvable.

Cladistic analyses may not be feasible at lower taxonomic levels, with reticular taxa and the resulting lateral transmission of characters. As such, there is a “line of death” or a “threshold of futility” to cladistic studies that must be determined empirically (Wheeler and Platnick, 2000).

INCONGRUITY OF TAXONOMY AND PHYLOGENY

Taxonomy usually reflects phylogeny (Simpson, 1961; Mayr, 1963; Tattersall and Eldredge, 1977, McKenna and Bell, 1997). The characters defining different taxonomic levels within a group usually define relationships between members in a group. The lower the taxonomic level shared, the closer the implied relationship between taxa (Simpson, 1961; Mayr, 1963). Among contemporaneous taxa, correspondence between phylogeny and taxonomy is especially close at higher levels (above family). At higher levels, groups have a long life and inclusion in them out of necessity is based more on specific diagnostic characters than on magnitude of differences.

However, phylogeny cannot always reflect taxonomy. On the one hand, phylogeny spans long periods, but any one taxonomic designation applies to a relatively small cross-section of time (Andersson, 1990). On the other hand, there does not seem to be a constant rate of phenotypic, genotypic, or behavioral change between evolving lineages (Simpson, 1961; Mayr, 1963, 1982). In some cases, the distinctiveness acquired by a particular taxon since the time of its divergence from its sister taxon may be best summarized by assigning it to an equal or higher taxonomic level than the one which unites its sister taxon to other more distally related taxa (Simpson, 1961; Mayr, 1963). However, a taxonomy that is out of sync with phylogeny is not the sole provenance of subjective taxonomic categories. New species may occasionally arise from one of the many subspecies comprising a species (Jolly, 1993). A higher taxon composed of taxa more distantly related to each other than those it excludes is termed paraphyletic, as opposed to monophyletic (i.e., a taxon in concordance with phylogeny). Although at higher levels (above genus) a paraphyletic taxon is unacceptable, at lower levels it is inevitable.

When dealing with ancestor descendant relationships through time, correspondence between phylogeny and taxonomy breaks down (Andersson, 1990; Szalay, 1993). This breakdown is exemplified in the inherent problems posed by the binomial system, which at some point must force an arbitrary change in the generic designation given to a long succession of species within an evolving lineage (Fig. 2). It is also illustrated with the initial divergence of two closely related species each of which goes on to give rise to a higher-level group. Taxonomically at the time of divergence, the two stem species must be placed within the same genus. Phylogenetically, however, these two species belong to different higher-level groups. This incongruity exists because phylogenetic groups are defined at their point of origin and are continuous through time, whereas taxa are defined relative to each other (at the same point in time) and increase in level (i.e., categories reflecting discontinuities) by quantum leaps.

Figure 2.

Schematic diagram of an evolving lineage composed of six species (A, A′, A′′, B, C, D), illustrating discordance between taxonomy and phylogeny. Where along a descent continuum new taxa are designated, is subjective. Along time continuum II, taxa designation is especially problematic, because it involves both a specific and a generic change (by definition taxa A′ and B are sister species and must belong to a different genus than A′′ and D) between what must ultimately be a parental population and its immediate descendants. Moreover, if both A and C are sister species representing stem ancestors of two contemporaneous family groups, time continuum III must also accommodate a change in family designation.

PROBLEMS WITH FOSSIL SYSTEMATICS

Taxonomy

Fossils may represent either ancestors of living taxa (i.e., crown taxa) or descendant lineages with no living members that diverged from the stem ancestor of the group before, during, or after the divergence of crown taxa (McKenna and Bell, 1997). Fossils are useful for systematic studies, because they may bridge the gap between the morphologic discontinuities presented by living taxa. Because many of the variables bearing on systematics are rarely known for fossils (e.g., complete phenotype, geographic distribution, population variation, behavior, ecology, development, and genotype), living taxa out of necessity must form the backbone of systematic studies. Classification of fossils always depends on analogies made to living taxa and is nearly always restricted to osseous or dental morphology. The magnitude of morphologic differences and/or types of morphologic characters that are chosen to define taxonomic categories for fossils must always be guided by those differences that define corresponding taxonomic categories for living taxa (Simpson, 1961; Mayr, 1963, 1982).

Unique morphologic complexes or structures (i.e., qualitative characters) distinctive and diagnostic of high-level taxa provide unequivocal means for initial classification of fossils.* Composed of inter-related anatomical elements, morphologic complexes (e.g., the mammalian middle ear [Romer, 1974; Lombard and Hetherington, 1993] or mammalian atlanto-occipital and atlanto-axial joint [Courant and Marchard, 2000]) are formed through a series of adaptive stages within an evolving lineage over a relatively long evolutionary time span (de Beer, 1937; Hanken and Hall, 1993). As such, it seems unlikely they can be converged upon or paralleled without leaving evidence in the anatomy. Likewise, once lost in ancestral lineages, morphologic complexes are unlikely to reappear in their descendants without leaving evidence of nonhomology.

The more completely represented a taxon is by fossil remains, the more detailed its morphology, and the more characters it exhibits, the better the resolution it provides for low-level classification. Enumeration of differences between fossil taxa without reference to those characters defining taxonomic levels among their closest living relatives, however, has no basis in systematic studies. Fossil taxa cannot be classified at lower taxonomic levels than those specified for the living taxa used to guide their classification. Without comparisons to properly classified living taxa, it is impossible to discern if the distinctiveness perceived for fossil taxa is a factor of population variation; is of the same magnitude diagnostic of subspecies, species, or generic differences; or actually corresponds to higher taxonomic levels. Whether a fossil species has the same level of intraspecific variation as do objectively defined species—i.e., those defined on the basis of sympatry (Mayr, 1963) —within the same taxonomic group may never be known. Nevertheless, assuming fossil species have a level of variation within the range observed in their most closely related living species is a conservative estimate that maintains consistency in the taxonomy.

Considering that, at the lowest levels, taxonomic resolution (i.e., subspecies and species) often depends on a complete phenotype and on population variation, only exceptional fossil assemblages with high representation of specific taxa may provide the data necessary for low-level classification. In the majority of cases, however, fossil remains fail to show the complete individual or the range of population variation and are, therefore, likely to underestimate species or subspecies numbers (Tattersall, 1992, 1993). On the other hand, the less likely scenario that fragmentary fossil remains exhibit only the distinctive and diagnostic localized morphology of two closely related species may conceivably result in the two being classified as different genera. In this case, the assumption is made that the unknown anatomy in its entirety is as distinctive as the known fragments. The taxa, therefore, are separated at a higher taxonomic level.

In the majority of cases, fossil remains fail to show the complete individual or the range of population variation and are, therefore, likely to underestimate species or subspecies numbers.

Fossils representing stem ancestors or early members of higher phylogenetic lineages may never provide enough evidence to place them within their respective higher lineage, or to exclude those closely related contemporaries outside of those lineages. This may be the case even if all of these fossils are nonreticular taxa. The parallelisms and reversals that are bound to occur in closely related taxa further confound the resolution these fossils provide. Fragmentary fossils isolated from other taxa by large time gaps may present at the time they are unearthed a unique and distinctive morphology that may very well set them apart as unique species or even genera. When remains of other more complete, closely related and contemporary fossil taxa are found, however, the number and type of characters present in the original fossil may not allow taxonomic resolution from its fossil contemporaries. If the more complete, newly found fossils clearly represent more than one species, the original fossil taxon, which was made nondiagnostic by the new finds, is in effect a nomen nudum.

Phylogeny

Use of fossils in phylogenetic reconstruction depends on fossils presenting the relevant morphology to provide some degree of taxonomic resolution. Fossils with uncertain classification may lead to circular reasoning when postulating shared derived characters for taxa. For example, misclassified fossils may be used to determine character polarity and the shared condition in early members of a lineage and, based on this determination, argue for their inclusion into the taxonomic group corresponding with this lineage. Although phylogenetic analysis may be possible when classifying fossil taxa at the highest taxonomic levels, the lower the level a fossil can be unequivocally classified at, the lower the level of the taxonomic group that needs to be analyzed. At lower taxonomic levels, comparisons can include a narrower range of living and fossil taxa, so that estimates as to where along a lineage phylogenetic events and homoplasies occur are more precise. In this regard, phylogenetic analyses benefit from unequivocal classification of fossils at low taxonomic levels (family or below). As long as the fossil is correctly classified even if it presents a single character, it can provide tests for character polarity, demonstrate parallelisms and reversals, and provide a minimum date (from the present with a known age of deposit) for phylogenetic events within the group. However, many more characters may be necessary to correctly classify a fossil and unravel its phylogenetic relationships.

As is usually the case for taxonomy, the more complete the remains of fossil taxa are, the greater the degree of phylogenetic resolution they afford. Because morphologic complexes indicative of intermediate taxonomic levels (family, superfamily, or infraorder) are often localized, fossil remains lacking these complexes may only provide initial taxonomic resolution at higher levels (e.g., order or class). In these cases, as long as fossils present sufficient diagnostic characters, finer phylogenetic resolution may be possible through cladistic analyses. Fossils with relatively few characters, all of which are labile at low taxonomic levels (i.e., genus, species, and subspecies within the taxonomic group the fossil initially provides resolution for), may never provide certain phylogenetic resolution. In these cases, the odds that shared characters are homoplasies are too high to use these characters as diagnostic of a group (Figure 5). Fossils that resist classification except at the highest taxonomic levels also confound phylogenetic resolution. Increasingly higher taxonomic levels suggest increasingly longer evolutionary time spans and higher numbers of intervening and crown taxa. Both the latter increase the odds of parallelisms and reversals between and along descendant lineages, respectively.

Fossil remains that fail to present discontinuous characters, and show only continuous ones, may also defy phylogenetic resolution. Although continuous characters may be converted into discontinuous ones by bracketing angular or metric values, it is unclear what advantages a cladistic analysis based exclusively on such characters has over phenetics (Bookstein, 1994; Sarmiento and Marcus, 2000). Throughout the evolution of a lineage, parallelisms and reversals may conceivably occur many times in continuous characters without leaving evidence in the anatomy.

When dealing with continuous characters, phenetic studies based on multivariate analyses may be more useful than cladistic studies based on bracketing metric traits (Sarmiento and Marcus 2000).* These analyses and the phenetic trees they arrive at, however, reflect degree of similarity and may or may not reflect true relationships. Although overall phenotypic similarity equates with shared taxonomic categories at lower levels, similarities in localized anatomy may or may not suggest close relationships. Simple shapes with few metric or angular variables may be easily paralleled or converged upon, but parallelisms or convergences become increasingly less likely with increasing shape complexity. The similarities present in most localized areas of the anatomy in taxa belonging to the same genus or family suggest that localized similarities in complex shape must at least show a corresponding level of relatedness.

HOMINOID SYSTEMATICS

As a group, hominoids present both benefits and drawbacks to classification and phylogenetic reconstruction. On the negative side, living hominoid species, with the exception of hylobatids are few in number. Therefore, the condition in one or two crown taxa usually makes the difference as to whether a character is to be considered a parallelism, a primitive retention or reversion, or shared derived (Sarmiento, 1988, 1995, 1998). On the plus side (Figure 5), hominoids are a modern group. Analogies made to living hominoids are, thus, more likely to be both relevant and useful. Moreover, hominoids are known from many fossil taxa (Szalay and Delson, 1979; Hill and Ward, 1988; Bonis et al., 1990; Begun and Kordos, 1993; Benefit and McCrossin, 1995; Moya-Sola and Kohler, 1996; Ishida and Pickford, 1997; Ishida et al., 1999; Senut et al., 2001). As such, character polarity and phylogenetic events reconstructed from living taxa can be amply tested against fossil evidence. Finally, the crown hominoid elbow and wrist joints exhibit a unique morphologic complex of interrelated anatomical elements that allow the hand and radius to rotate nearly 180 degrees around the ulna (Sarmiento, 1985, 1988). As summarized below, this complex permits the initial classification of hominoid remains.

Wrist and Elbow Joint Complex

Figures 3 and 4 and Table 1 summarize the elbow and wrist joint characters that are mechanical requisites of forearm rotation and shows the various character states for the complex and the most parsimonious transformations among the five hominoid genera. Excluding orangutans, variation in the complex within Hylobates, Pan, Gorilla, and Homo unites these latter genera in a cline, in order from most primitive to most derived. The orangutan complex clearly shares the hominoid condition, but it is markedly derived in a separate direction from humans and African apes, more than likely from a hylobatid-like condition (Sarmiento, 1985, Sarmiento, 1988). There is little doubt that the hominoid elbow and wrist joint complex is a shared derived structure, as indicated by the following: (1) the unique hominoid complex is formed by a large number of inter-related anatomical elements, including some novel elements. Yet, the complex in all genera corresponds in anatomical detail exhibiting intrageneric variation that unites the various hominoid wrist and joint types in a cline (Lewis, 1969; Sarmiento, 1988). (2) There is a large morphologic discontinuity between the hylobatid complex (the least derived among hominoids) and the corresponding elbow and wrist joint characters of monkeys, indicating several intervening adaptive stages are necessary to arrive at the hominoid complex (Lewis, 1969; Sarmiento, 1985, 1988). (3) The evolutionary sequences suggested by differences in the joint complex within hominoids agrees with hominoid phylogenies as independently arrived in countless studies based on other morphologic characters and on biomolecular analysis (see Stewart and Disotell, 1998). (4) The living hominoid genera practice widely divergent locomotor behaviors with contrasting different upper limb use, but the joint complex is common to all hominoids and has been modified by each to best suit practiced behaviors (Sarmiento, 1985, 1988). As a shared derived complex, the hominoid elbow and wrist joint serves as a diagnostic structure that can be used to identify hominoids (Sarmiento, 1987, 1988).

Figure 3.

Schematic diagram of character transformations during hominoid wrist joint evolution, emphasizing the ulnocarpal and pisotriquetral joints. Table 1 summarizes characters present in each hominoid wrist type and in a hypothesized ancestral catarrhine. (NOTE: From top to bottom, the first six colors in the key correspond to columns numbered 1–6 in Table 1.) The evolutionary order of development hypothesized for the modern hominoid wrist joint types is based on increasing character complexity and increasing divergence away from the generalized catarrhine condition (Lewis, 1969; Sarmiento, 1985; 1988).

Figure 4.

The hominoid humeroanterobrachial joint compared with that of a generalized catarrhine. Bony characters associated with the unique hominoid forearm pronation and supination range are emphasized (Sarmiento, 1985; 1988). In addition, hominoids also show soft tissue correlates of forearm rotation, e.g., short heads of pronator and supinator muscles and annular radioulnar ligament.

Table 1.  Systematic characters in the wrist joints of humans, apes, and a generalized catarrhine precursor
 1: Carpoantebrachial2: Pisotriquetral & Pisomeniscal*3: Distal radioulnar4: Pisostylotriquetral & Meniscoulnar5: Mid-Carpal6: Os daubentontriquetral
  • *

    For the purpose of coinciding with past works, Lewis' (1969) term ‘semi-lunar meniscus’, ‘meniscus’ for short, or ‘menisco/meniscal’ in combination are used here. Functionally however, this structure is best referred to as an anular ligament.

  • Column numbers correspond to joint colors as described in Figure 3 legend.

Generalized Cartarrhine Precursor1. Mild mediolateral curvature 2. Triangular ligament variably articular with ulnar head 3. Mediolaterally narrow radiolunate facet1. Concavo-convex pisotriquetral facet 2. ‘Meniscus’ absent1. Incipient synovial distal radioulnar joint 2. Transversely narrow joint with its curvature subtending small central angle1. Robust ulnostyloid process with facets on distal and radial aspect 2. Bulky cuboidal-like triquetrum with bifaceted proximal articulation 3. Bifaceted, robust, and elongated pisiform1. Unfused os centrale 2. Proportion of hamate to capitate forming ball of joint? 3. Large mediolateral curvature1. Os daubentonii absent
Gibbons & Siamangs5. Ulnar shelf of radius excluding ulnar head from joint 6. Mediolaterally broad radiolunate facet 7. Tight mediolateral curvature of ulnar side3. Bifaceted Pisiform articular with triquetrum and os daubentonia/‘semilunar meniscus’ 4. Separate ‘semilunar meniscus’ and triangular disc 5. Triquetrum's pisiform facet poorly demarcated3. Diarthroidal 4. Transversely broad joint (i.e. large semilunar ulnar head) with its curvature subtending large central angle4. Reduced wedge-shaped triquetrum with vestigial stylotriquetral contact 5. Distal migration and reduced pisiform with loss of ulnopisiform contact 6. Circumferential facet on ulnar styloid for ‘semilunar meniscus’4. Ball of joint formed largely by head of hamate 5. Tight mediolateral curvature of ulnar side 6. Trapezium with large elongated and curved hamulus2. Os daubentonii articular with ulnar styloid
Orangutan(5–7 as in Gibbon) 8. Hypertrophied lunate with mediolateral curvature for radial facet subtending large central angle6. Unifaceted pisiform articular only with triquetrum(3 & 4 as in Gibbon)7. Short, markedly abbreviated non-articular ulnar styloid process 8. Extensive fusion of ulnar ligaments to each other and to triquetrum separating radiocarpal and pisotriquetral joints 9. Reduced, cylyndrical-shaped triquetrum(5 as in Gibbon) 7. Ball of joint formed largely by head of capitate (rarely variable) 8. Triquetrum migrated distally to midcarpal joint 9. Lunotriquetral joint shallow ball and socket 10. Variable hamate pisiform facet3. Os daubentonii absent
Chimpanzee(6 & 7 as in Gibbon) 9. Bifaceted ulnar head for triangular disc and distal radius 10. Variable triquetral-triangular disc facet(4 & 5 as in Gibbon) (6 as in Orangutan) 7. Reduced, but palmodorsaly elongated pisiform 8. Pisotriquetral joint concavo-convex(3 & 4 as in Gibbon)(5 & 6 as in Gibbon) 10. Reduced wedge-shaped triquetrum non-articular with ulnar styloid(5 as in Gibbon) 11. Spiral hamate triquetral articulation 12. Os centrale fusion 13. Ball of joint formed mainly by capitate head4. Os daubentonii or its vestige rarely present
Gorilla(9 as in Chimpanzee) 11. Mild mediolateral curvature of ulnar side 12. Triquetrum articular disc facet(6 as in Orangutan) (7 & 8 as in Chimpanzee) 9. Triquetrum's pisiform facet well demarcated 10. Fusion of ‘semilunar meniscus’ and triangular ligament, forming single articular disc(3 & 4 as in Gibbon)(8 & 9 as in Gibbon) (10 as in Chimpanzee) 11. Short markedly abbreviated non-articular ulnar styloid process excluded by articular disc from proximal carpal joint(11–13 as in Chimpanzee) 14. Mild mediolateral curvature of ulnar side(4 as in Chimpanzee)
Human(Same as Gorilla)(9 & 10 as in Gorilla) 11. Planar pisiform-triquetral facet 12. Short pisiform(3 & 4 as in Gibbon)(10 as in Chimpanzee) (11 as in Gorilla)(Same as Gorilla)(4 as in Chimpanzee)

Considering the loss of intervening populations, the marked morphologic discontinuity between hominoid and Old World monkey wrist-joints indicates that there must have been early lineages of extinct hominoids, including the stem ancestor of the group without full development of the hominoid complex. Identification of these taxa as early hominoids is possible if they exhibit a complex approaching the common condition shared by crown hominoids. However, it is problematic if the complex is incipient, as it is expected to be in those ancestral hominoids recently diverged from nonhominoid catarrhines. At a time, when the distinction between hominoid and nonhominoid catarrhines is at the species or generic level, diagnoses as to whether a fossil belongs to the hominoid lineage must depend on rather complete remains.

HOMINOIDS AND THE CAUTIOUS CLIMBING COMPLEX

Aside from the elbow and wrist joint complex, crown hominoids share a suite of characters, associated with cautious climbing behaviors that as a set are diagnostic of the group (Sarmiento, 1987, 1988, 1995). However, not one of these characters individually presents the morphologic complexity to attest to parallelisms or reversals. Hence, when occurring in isolation, they are not diagnostic of hominoids (Sarmiento, in preparation). In fact, to one degree or another, many of these characters have been paralleled by other primates and converged on by other mammals (Sarmiento, 1995). Because many of these characters are quantitative and unlikely to document reversals, some hominoids may exhibit very few cautious climbing characters or remnants of these characters. Without knowledge as to the evolutionary transformations that led to the shared hominoid condition, presence in a fossil of one or two characters that are either part of the cautious climbing complex or part of the elbow and wrist joint complex cannot be taken as diagnostic of hominoids.

CHARACTERS DISTINGUISHING THE HOMINOID TAXA

There are a large number of morphologic characters known to distinguish the five hominoid genera and to group these at higher levels (Keith, 1915, 1934, 1940; Gregory, 1922; Wood-Jones, 1929; Schultz, 1936, 1968; Lewis, 1969; Sarmiento; 1987, Sarmiento, 1988, 1995, 1998) (Fig. 5). Siamangs and gibbons form one of these groups (hylobatids) and humans and African apes the other (hominoids). Although orangutans appear to share a common evolutionary history with humans and African apes exclusive of hylobatids (Schultz, 1936; Sarmiento, 1985, 1988, 1998), the morphologic characters orangutans share with humans and African apes lack the complexity to unequivocally resolve this. Because humans and African ape taxa are closely related, as are the various species of hylobatids, relationships within either of these two groups are not always clear. The larger number of similarities between gorillas and humans to the exclusion of chimpanzees is as likely a result of shared derived ancestry as of parallelisms owing to a terrestrial lifestyle (Sarmiento, 1983, 1985, 1988, 1994, 1998). Cladistic analyses aimed at resolving relationships between humans, gorillas, chimpanzees, orangutans, and hylobatids that consider all of the characters presently known to distinguish each hominoid genera, including biomolecular evidence have yet to be reported on.

Because humans and African ape taxa are closely related, as are the various species of hylobatids, relationships within either of these two groups are not always clear.

Figure 5.

Juvenile (a) orangutan, (b) gorilla, (c) human, (d) pygmy chimpanzee and (e) common chimpanzee skulls illustrating the presence or absence of a mastoid notch and the cranial bone relationships at pterion. Cladograms (i–vii, on right) show minimum number of homoplasies that must be posited for characters given currently accepted great ape and human phylogeny. The mastoid notch character shows three alternative cladograms (v, vi, vii), all equally parsimonious with the same minimum number of homoplasies. Cranial bone relationships at pterion produce a single most parsimonious cladogram, but each of the other alternatives (i, ii, iv) have only one additional homoplasy. Due to low sample number (i.e. number of compared taxa) and high character heterogeneity within sample, both characters fail to provide phylogenetic resolution. Because both characters are too simple to confidently detect homoplasies and alternate character states may exist in low frequency within the same hominoid taxa, they are equivocal for resolving higher-level taxonomy and usually poor for taxa diagnosis.

Hominoid alpha taxonomy also needs to be worked out. This is especially the case in great apes where distinctive populations are allopatric and species differences have been decided on magnitude of morphologic differences as opposed to objective criteria (Groves, 1993; Sarmiento and Butynski, 1996; Sarmiento et al., 1996; Sarmiento and Oates, 2000).

PROBLEMS IN HOMINOID AND HOMINID FOSSIL CLASSIFICATION

An initial dependence on very fragmentary and incomplete fossils, and the tradition in vertebrate paleontology of relying on teeth for classification, is in part to blame for past misinterpretations of the hominoid fossil evidence. With only dental remains considered, hominids and hominoids were initially diagnosed on characters that are either primitive for, or labile in catarrhines, and/or lack the complexity to test for homoplasies (i.e., thick molar enamel, a small canine and a nonsectorial p3, and the 4 cuspid upper and 5 cuspid lower molars; Gaudry, 1890; Branco, 1898; Pilgrim, 1910; Schlosser, 1911; Schwalbe, 1915; Gregory, 1916, 1922; Eliott-Smith, 1924; Abel, 1931; Simons and Pilbeam, 1965; Szalay and Delson, 1979). Although the corresponding characters appeared to be diagnostic for hominids and hominoids when considering only the crown members of sister taxa (i.e., African apes and cercopithecines, respectively), the characters were never examined in outgroups to establish polarity (primitive vs. derived), or within member taxa to gauge their lability at corresponding taxonomic levels. When alleged fossil hominoid skeletal anatomy, became better known, the number of parallelisms between the two major catarrhine groups that were required under certain fossil classifications forced progressive workers to examine outgroups and recognize the 4 cuspid upper and 5 cuspid lower molar pattern as primitive for catarrhines, if not anthropoids (Von Koenigswald, 1969; Sarmiento, 1987). Likewise, to avoid an inordinate number of parallelisms among catarrhines, hominoids, or the human/African ape clade, it was necessary to recognize thick enamel, a small canine, and nonsectorial p3 as labile in anthropoids and likely to develop in parallel in hominoids and early catarrhines. Hallowed by usage, however, provisional hominoid and hominid classifications proposed early on in the infancy of paleoanthropology on the basis of very fragmentary Eocene, Oligocene, or Miocene fossils, and on nondiagnostic characters, are still followed by some workers, even in the face of very strong conflicting evidence (Bloch et al., 1997; McKenna and Bell, 1997).

Underlying fossil misclassifications are also uninformed preconceptions, which blinded workers to alternative scenarios and gave credence to using solely dental evidence to diagnose hominoids and hominids. These preconceptions, which can trace their origin to the very beginnings of vertebrate paleontology and human evolutionary theory, can be summarized in the following practices: (1) reconstructing whole animals based on fragmentary fossil remains and attempts to correctly classify such fragmentary fossils at the lowest taxonomic levels (species or subspecies); (2) assuming characters present in humans are always the most progressive and derived, and seldom or never represent an ancestral condition (also applies to those characters humans share with great apes and/or hylobatids); (3) assuming all catarrhines with a small canine and a nonsectorial p3 are hominids.

The first of these has an origin in Cuvier's (1812) principle of correlation and is based on a creationist philosophy in which organism types are immutable. The second finds its origin in Lamarck's scala naturae (1809) in which all living organisms are striving to become human. It underlies many preconceptions as to character polarity, thus, resulting in misclassification. The third can be traced to Darwin's Descent of Man (1871) and the special significance he accorded to canine reduction as one of a constellation of interdependent human characters, including large brain, manual dexterity, tool use, and bipedalism. Although Darwin never argued that any one of these characters alone is diagnostic of humans, providing examples of their independent development in other animals, their subsequent enthronement in human evolutionary theory has given them mythic powers as diagnostic hominid characters. In this regard, many workers still fail to realize that (1) many of the early nonhominid hominoid fossils have small canines and nonsectorial premolars (e.g., Oreopithecus, Ouranopithecus, and Ramapithecus), (2) several nonhominoid anthropoids have independently arrived at canine reduction (e.g., Brachyteles, Callicebus, and Alouatta), (3) many nonhominoid catarrhine females have reduced canines (e.g., Rhinopithecus, Papio, Presbytis, and Pygathrix). All of which indicate that reduced canines may be ancestral for hominoids, relatively labile at intermediate taxonomic levels and in themselves not reliable for diagnosing hominid affinities or higher taxonomic levels.

Guided by Cuvier's principle of correlation and Lamarck's scala naturae, Darwin's musings have been taken to the extreme by paleoanthropologists working on alleged fossil hominids. In this regard, any character present in modern humans, which can be correlated with bipedalism, tool use, or some relative brain enlargement, have also been argued to be diagnostic of hominids (Hurzeler, 1960; Simons and Pilbeam, 1965; Leakey, 1968), especially Homo (Mayr, 1950; Leakey et al., 1964). This is the case regardless of whether the characters are common to nonhominid primates, ancestral for hominoids, and/or associated with other behaviors aside from bipedalism (Sarmiento, 1998, 2001; Sarmiento and Marcus, 2000) or tool use (Sarmiento, in preparation). Even in cases where fossil australopithecines are known from relatively complete remains, hominid classification has still relied on a reduced canine, a nonsectorial premolar, thick molar enamel, and some human-like skeletal characters allegedly correlated with modern human bipedality. The possibility these characters, many of which are quantitative, may be labile in, or ancestral for, hominoids is rarely given serious consideration. The presence in an early hominoid (e.g., Oreopithecus; Straus, 1963; Sarmiento, 1987; Kohler and Moya Sola, 1997) of many of the same alleged bipedal characters argued to be diagnostic of hominids suggests that these characters may be primitive for and/or labile within hominoids.

The Piltdown hoax speaks volumes for the power of these preconceptions to misguide hominid classification. At the time, the human ancestor everyone hoped to find was expected to be associated with tools, endowed with a large brain, and have relatively small canines and human-like premolar wear patterns. Thus, Piltdown was popularly embraced as a human ancestor for nearly half a century (Mayr, 1950), withstanding the scrutiny and suspicions of the most eminent workers (Miller, 1915; Gregory, 1922; Weidenreich, 1943; Weiner et al., 1953; see also Milner, 1999). Notably, the initial proposal that Oreopithecus was a human ancestor never took hold, despite the presence of reduced canines, bicuspid nonsectorial p3s, and many of the alleged bipedal characters, i.e., strong ischial spine, bowl-shaped pelvis, strong femoral bicondylar angle and relatively short pubic symphysis (Sarmiento, 1987; Kohler and Moya Sola, 1997). Other factors, therefore, must also play a role in the acceptance of fossils as ancestors.

One of us (Sarmiento, 1987) has noted that incomplete and fragmentary fossils are more likely to be popularly embraced as ancestors than are more complete ones, which divulge more specific details. Incomplete fossils accommodate everyone's preconceptions and can be made to support personal hypotheses or serve individual agendas. In this regard, Kenyapithecus and Rampithecus (Simons, 1972) known at the time only from dental remains, and exhibiting a somewhat reduced canine and a nonhoning premolar, were also posited to be bipedal tool users, and evidence was found to support these claims (Leakey, 1968). The economic and sociopolitical implications of finding human ancestors, however, are considerable, and acceptance of fossils as such rests in part with these concerns.

PRESENT AND FUTURE PROBLEMS IN SYSTEMATICS

Many of the problems confronting hominoid fossil systematics deal with resolution of late Miocene and Plio-pleistocene fossils and separation of those taxa that predate the human/African ape split from those that are members or offshoots of each of the three crown lineages. It is more than suspicious that on the African continent, currently inhabited by chimpanzees and gorillas, there are no African ape ancestors, nor offshoots of either of these lineages, nor taxa ancestral to the African ape/human clade during the Plio-pleistocene, only hominids. Aside from clinging to the same preconceptions and relying on the same nondiagnostic characters that incorrectly led Eocene, Oligocene, and Miocene fossils to be classified as hominids, paleoanthropologists have also erred in classifying late Miocene and Plio-pleistocene fossil hominoids (1) as new and/or separate taxa based solely on enumeration of differences between fossils (ignoring intraspecific variation and the types of characters and magnitude of difference defining lower taxonomic levels in living taxa), (2) without reference to diagnostic characters that distinguish higher level taxonomic groupings in living hominoids, (3) at lower levels than those specified for the living taxa used to guide classifications, (4) on too few supposedly diagnostic characters in localized areas of the anatomy without examining character lability or polarity, (5) on fragments that lack diagnostic morphology for the taxonomic level in question. It is very unlikely, therefore, that African Plio-pleistocene “hominid fossils” are correctly classified.

Phylogenetic analyses based on incorrect classifications, are unlikely to be accurate, because relevant outgroups may be excluded from analysis, and intraspecific and inter-specific variation conflated. Current fossil hominoid phylogenies, however, are also hampered by cladistic analyses that (1) are made up in large part of subjectively bracketed continuous characters; (2) sample too few member taxa and/or no outgroups, so that hypothesized homoplasies and character polarities are poorly supported; (3) present initial results as final and fail to test the likelihood of assumed homologies and hypothesized homoplasies; (4) use only skull or dental differences in analyses and ignore diagnostic characters in other areas of the anatomy, when such evidence is available; (5) attempt to bridge large gaps in intervening populations with characters that are labile at the subspecies and species level; (6) fail to consider appropriate outgroups given the taxonomic level of resolution enabled by the fossils.

By no means are these problems the sole provenance of Late Miocene or Plio-pleistocene hominoid systematics. These same errors also hamper systematics of early and middle Miocene catarrhines (Walker et al., 1986; Gebo et al., 1997; Rae, 1999; Senut et al., 2000) and hominoids (Ward et al., 1999) and also cause havoc throughout mammalian systematics. Unlike other systematists, however, most hominid paleoanthropologists have yet to recognize these problems and make the necessary adjustments.

PROMISE OF DEVELOPMENT

Lovejoy et al. (1999) suggested that, within closely related species (i.e., African apes and humans), adult mammalian limb bones can provide important data for phylogenetic or cladistic analyses if the interplay of local regulating mechanisms (SAMs) and these authors' own interpretations of developmental differences in assignment of positional information (PI) in the limb bud are considered. Combining their own interpretations on pelvic limb development and mechanical function, they arrive at the following five character categories: Type 1, trait(s) that differ in expression, due to resultant effects of PI on local pattern formation, e.g., hominid ilium superoinferior length; Type 2, trait(s) that represent field-derived pleiotropy and whose morphologic consequences are selected for (but with no interaction to natural selection processes), e.g., hominid pubis superoinferior length; Type 3, trait(s) that differs between taxa due to systemic modifications, e.g., hominid body size; Type 4, trait(s) that differs in expression between taxa due to interaction of SAMs, e.g., human femoral bicondylar angle; Type 5, same as Type 4, but of no diagnostic value, e.g., femoral anteversion. Lovejoy et al. (1999) suggest these have varying significance in fossil hominoid systematics and in functional interpretations.

Unfortunately, the categories proposed by Lovejoy et al. (1999) are unrealistic and have little heuristic value for inferring phylogenetic scenarios. It is difficult to envision how, on the basis of adult mammalian limb bones (i.e., fossils) alone, one would be able to interpret differences in PI assignment or claim that differences in length are the result of PI. Interspecific differences in adult limb bone length are also the result of postembryonic or postnatal differences in rates of cellular expression, morphogen gradients, and hormones, all of which are mediated by organism interplay with environmental forces. Limb lengths in great apes can differ without differences in PI, and this may especially be the case among closely related taxa (Sarmiento, 1985).

A category composed of limb bone lengths lacking a mechanical function as proposed by Lovejoy et al. (1999, 2000) is also unrealistic. It is well known that skeletal segment lengths affect the lever arms of muscles or of adjoining segments and also the moment of inertia of segments, so that limb bone length affects mechanical function. The same applies to joint sets (i.e., femoral bicondylar angle or femoral anteversions). The claim by Lovejoy et al. (1999) that the criteria for assigning characters to each of their five categories is not actually necessary, because allocation of extinct organisms into their categories is hypothetical,* is puzzling. As such, it is unclear how careful categorization of characters into their system will clarify either phylogeny or systematics. Nesting unrealistic hypotheses of development and function within hypotheses of phylogeny can only result, at best, in unrealistic speculations. The recommendation to construct cladograms based on characters that vary intraspecifically and/or are continuous (i.e., quantitative) to clarify lower level systematics defies logic and shows a surprising naiveté and/or irreverence for modern systematics.

Notably, the majority of current research delimiting developmental mechanisms at the cellular level has been done on chick embryos. Until similar research is done on primates and mammals, in a systematic fashion, there is no way to know (1) the various growth variable contributions to adult bone shape; (2) how this contribution varies with distance of relationship; (3) how PI simultaneously affects various dimensions of an adult bone.

Nevertheless, it is well known that in closely related taxa where distinctive body segment proportions are not developed until adulthood, length differences are more likely the result of rates of cellular expression as mediated by environmental forces, rather than of PI (Sarmiento, 1985). Such hypotheses about differences in PI among closely related mammalian taxa, at this point, are not likely to help unravel human-African ape phylogeny.

CONCLUSIONS

If there is any testament to the accuracy of morphology-based systematics, it is provided by the recent onslaught of biomolecular studies, which more often support rather than contradict morphology-based taxonomies and phylogenies. Notably, in those cases where there are clear contradictions (Graur et al., 1991; Milinkovitch, et al., 1993; Garner and Ryder, 1996), the biomolecular evidence presented has been shown to be unreliable for reconstructing phylogeny (Luckett and Hartenberger, 1997; Philippe, 1997; Naylor and Brown, 1998; Seaman, 2000). Mammalian taxonomy as based on morphology, has thus remained more or less unchanged. Progressive workers, therefore, argue for biomolecular and morphologic character integration within simultaneous analyses striving for “total evidence” (Kluge, 1989; Nixon and Carpenter, 1997).

However, the systematic interest molecular studies have generated has shown that many currently accepted phylogenies and classifications have very little molecular or anatomical data supporting them (Nielsen, 1998). The need to gather these data to construct well-supported phylogenies and classifications, especially as pertains to hominoids and hominids, is fertile ground for anatomical studies. Regardless of the anatomical subdiscipline (development, histology, gross anatomy, etc.) these data originate from, they must adhere to systematic methodology to be applicable. Hopefully, the future revolution in hominoid and hominid systematics that is sure to occur will not cause additional loss of confidence in morphology-based methods but will inspire anatomists to make significant contributions. The reason for this change will certainly not be founded in the inaccuracy of MBS, but on the long history of preconceptions and the sociopolitical and moral baggage that comes with unraveling our own history. The latter is the challenge that makes hominoids such an enjoyable group to work with.

  • *

    To avoid redundancy, the phylogenetic taxonomy of De Queiroz and Gauthier's (1990, 1992, 1994) and De Queiroz's (1997) does not recognize categories or levels (i.e., ranks). As such, it is not utilized to summarize biological differences and similarities between organisms but is strictly a system of lineages and evolutionary relationships

  • *

    As a rule of thumb, mammalian taxa with (1) distinctive, quantitative differences throughout their anatomy are placed in different genera (Gregory, 1910; Robinson, 1962); (2) a unique and distinctive morphologic complex in a localized area are given the rank of families (Gregory, 1916; Simpson, 1945; Anderson and Jones, 1984); and (3) many unique and distinctive morphologic complexes throughout their anatomy are usually ranked in their own order or class

  • *

    Although not treated here, initial classification also entails the correct identification of fossilized anatomy. Occasionally, the nonhomologous anatomy of disparate life forms may superficially resemble one another, leading to absurd phylogenetic and functional conclusions (Boaz, 1980)

  • *

    Morphometrics or shape analysis may provide some resolution to the problem of using quantitative characters in cladistics. Although no convincing methods have as of yet been proposed, algorithms based on quantitative shape variables may be used to summarize discontinuities in shapes between taxa, and these in theory could be successfully implemented in cladistic analysis

  • Whereas cladistics arrives at a shared range of metric values (bracketed values) for a phylogenetic group and hypothesizes maximum parsimony to reconstruct the ancestral condition, phenetics arrives at mean distances (D) between measurements and hypothesizes minimum distance (i.e., maximum parsimony) for the ancestral condition. In the case of phenetics, minimum mean distances are free of subjective categories introduced by taxonomists when bracketing continuous characters

  • *

    “How are traits to be allocated to one of these five categories, given that virtually nothing is known about their actual genetic basis and that such knowledge is virtually unobtainable for extinct organisms? Our proposed classification is not intended to require such knowledge, but only to encourage observers to formally state the presumed morphogenetic basis of each of the traits they choose to include in a functional or phyletic analysis” (Lovejoy et al., 1999 p. 13,251)

Biographical Information

Dr. Sarmiento is Research Associate in the Division of Vertebrate Zoology, American Museum of Natural History (AMNH). Dr. Stiner is a Scientific Assistant in the Division of Vertebrate Zoology at AMNH. Dr. Mowbray is Collections Associate in the Division of Anthropology at AMNH.

Ancillary