The genus category and cranial morphometrics of the Catarrhini with implications for fossil hominins
Jack Andrew Coate
A thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy
Postgraduate Board University of New South Wales
September 2007
PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet
Surname or Family name: Coate
First name: Jack Other name/s: Andrew
Abbreviation for degree as given in the University calendar: PhD
School: SOMS Faculty: Anatomy
Title: The genus category and cranial morphometrics of the Catarrhini with implications for fossil hominins
Abstract 350 words maximum: (PLEASE TYPE) Recently, the number of hominin genera has increased dramatically. Prior to the announcement of Ardipithecus, only two genera were used by paleoanthropologists: Australopithecus and Homo. Presently, up to eight hominin genera are used: Sahelanthropus, Orrorin, Ardipithecus, Australopithecus, Praeanthropus, Kenyanthropus, Paranthropus and Homo. Unlike species concepts, the genus category has not received wide critical examination. To investigate the use of the genus category in paleoanthropology, a comparative framework drawing on morphometric data from a large number of catarrhines is developed. Cranial variables include 36 standard linear measurements from representatives of catarrhine genera across the major tribes/families. This study seeks to assess whether too few or too many hominin genera have been recognized compared with extant catarrhines. Moreover, two published hypotheses about the use of Homo are examined: 1) Wood & Collard’s (1999) proposal to transfer Homo habilis/rudolfensis to Australopithecus; and 2) Goodman et al’s (1998) classification of both humans and chimpanzees in Homo. To analyze these cranial variables and a number of shape indices calculated from them, as well as to assess competing hypotheses, univariate, bivariate and multivariate statistical approaches are used. The results allow the identification of a set of variables and shape indices which distinguish genera across the catarrhines. Importantly, body size seems to be the major separator of catarrhine genera, reinforcing the idea that they occupy discrete adaptive zones. Moreover, differences between these genera mostly represent contrasts in the size of the neuro- versus the viscerocranium. When applied to hominins, a picture emerges which distinguishes them from extant catarrhines: cranial shape rather than size is the major component distinguishing them; this suggests that extinct hominins occupied similar habitats and adaptive zones; variability in size and shape within hominin genera is much lower than extant catarrhines; and the major differences seen in shape among hominins are the result of encephalization in Homo. It is concluded here that both Wood & Collard’s (1999) and Goodman et al.’s (1998) proposals appear to be premature. Moreover, while the earliest hominins may be too finely split at the genus level, the evidence for distinction of Australopithecus and Paranthropus is solid.
Declaration relating to disposition of project thesis/dissertation
I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).
…………………………………………………………… ……………………………………..……………… ……….………4/14/07...…….… Signature Witness Date
The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.
FOR OFFICE USE ONLY Date of completion of requirements for Award:
THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS Abstract:
Recently, the number of hominin genera has increased dramatically. Prior to the announcement of Ardipithecus, only two genera were used by paleoanthropologists: Australopithecus and Homo. Presently, up to eight hominin genera are used: Sahelanthropus, Orrorin, Ardipithecus, Australopithecus, Praeanthropus, Kenyanthropus, Paranthropus and Homo. Unlike species concepts, the genus category has not received wide critical examination. To investigate the use of the genus category in paleoanthropology, a comparative framework drawing on morphometric data from a large number of catarrhines is developed. Cranial variables include 36 standard linear measurements from representatives of catarrhine genera across the major tribes/families. This study seeks to assess whether too few or too many hominin genera have been recognized compared with extant catarrhines. Moreover, two published hypotheses about the use of Homo are examined: 1) Wood & Collard’s (1999) proposal to transfer Homo habilis/rudolfensis to Australopithecus; and 2) Goodman et al.’s (1998) classification of both humans and chimpanzees in Homo. To analyze these cranial variables and a number of shape indices calculated from them, as well as to assess competing hypotheses, univariate, bivariate and multivariate statistical approaches are used. The results allow the identification of a set of variables and shape indices which distinguish genera across the catarrhines. Importantly, body size seems to be the major separator of catarrhine genera, reinforcing the idea that they occupy discrete adaptive zones. Moreover, differences between these genera mostly represent contrasts in the size of the neuro- versus the viscerocranium. When applied to hominins, a picture emerges which distinguishes them from extant catarrhines: cranial shape rather than size is the major component distinguishing them; this suggests that extinct hominins occupied similar habitats and adaptive zones; variability in size and shape within hominin genera is much lower than extant catarrhines; and the major differences seen in shape among hominins are the result of encephalization in Homo. It is concluded here that both Wood & Collard’s (1999) and Goodman et al.’s (1998) proposals appear to be premature. Moreover, while the earliest hominins may be too finely split at the genus level, the evidence for distinction of Australopithecus and Paranthropus is solid.
II Acknowledgements: I would personally like to thank my supervisor and co-supervisor, Dr. D. Curnoe and Prof. K. Ashwell, as well as, many other close friends and colleagues, including; H. Cohen, E. Danielson and family, T. Furlong, H. Green, D. Harris, Dr. A. Herries, S. Kiker, J. Louys, K. McQualter, D. Neuweger, Dr. S. van Holst, and L. van der Weyde. In addition, I would like to especially thank my parents and family for emotional and financial support throughout the years and during this thesis project. Furthermore, I would also like to personally thank the following institutions and their Physical Anthropology, Zoology and/or Mammalogy staff for allowing me to gather craniometric data from primate skulls in their care – Dr. S. Ingleby and T. Ennis at the Australian Museum, Sydney, Australia; Prof. F. Thackeray, Dr. T. Kearney, S. Potze and T. Perregil at the Transvaal Museum, Pretoria, South Africa; Prof. M. Raath, M. Dayal, M. Spocter, B. Billings and J. Hemingway at the University of Witwatersrand, School of Anatomical Sciences, Jo-burg, South Africa; J. Spence and E. Sarmiento at the American Museum of Natural History, NYC, NY, USA; Dr. Y. Haile-Selassie, L. Jellema and Prof. B. Latimer at the Cleveland Museum of Natural History, Cleveland, OH, USA, which houses the Hamann-Todd Collection; M. Schulenburg at the Field Museum, Chicago, IL, USA; L. Gordon at the Smithsonian Institute, National Museum of Natural History, Washington, D.C., USA; and P. Jenkins and D. Hills at the Natural History Museum, London, UK. In addition, Dr. R. Raaum and Dr. A. Tosi provided genetic information and distance matrices for some of the primates studied herein and both are greatly appreciated by the author for their data and assistance.
III Table of contents:
Abstract: II Acknowledgements: III Table of contents: IV Abbreviations: V-VII
Chapter 1: Introduction, pp.1-18. Chapter 2: Materials and Methods, pp.15-50. Chapter 3: Results for genera of Cercopithecini, pp.51-113. Chapter 4: Results for genera of Papionini, pp.114-179. Chapter 5: Results for genera of Colobinae, pp.180-277 Chapter 6: Results for genera of Hominoidea, pp.278-337 Chapter 7: Summary and Discussion of results for the extant Catarrhini genera, pp.338- 356. Chapter 8: Results for genera of the Hominina, pp.357-412. Chapter 9: Conclusions, pp.413-420. References: pp.421-489. Appendix: pp.490-500.
IV
Anatomical, Genus and Journal Abbreviations (in alphabetical order): AJP – American Journal of Primatology AJPA – American Journal of Physical Anthropology Al. – Allenopithecus An. – Ankarapithecus Ar. – Ardipithecus Au. – Australopithecus Bas – Basion Bien – Interentoglenoid Breadth bp – basepairs Br - Bregma Cc. – Cercocebus Cp. – Cercopithecus Ch. – Chlorocebus Co. – Colobus Cum. – Cumulative CV – Coefficient of Variation or Canonical Variate(s) CVA – Canonical Variate Analysis D. – Dryopithecus Df: - Degrees of Freedom Dm3 – Distal M3 E. – Erythrocebus Ecm – Ectomolare Enm – Endomolare EQ – Encephalization Quotient Ex. – Excluding FAD – First Appearance Date Fmo – Frontomalarorbitale Fmt – Frontomalartemporale G – Glabella Go. – Gorilla H. – Homo Ho. – Homo H.he – “early Homo”, H. habilis sensu lato and H. erectus sensu lato Homo. - Homogeneity Hy. – Hylobates IJP – International Journal of Primatology Iv – Incisivion JAS – Journal of Anthropological Science JHE – Journal of Human Evolution Ka – Thousand years ago Ka. – Kenyanthropus Kp. – Kenyapithecus L. – Lophocebus V LAD – Last Appearance Date Lam – Lambda Ln – Natural Log Mn. – Mandrillus Mc. – Macaca Ma. – Mega Annum (or Megaannum; i.e. One million years) Max. – Maximum MANOVA – Multivariate Analysis of Variation Maxalvlen – Maxillo-alveolar Length (pros-dm3) Maxnawi – Maximum Nasal Aperture Width Mi. – Miopithecus Min. – Minimum MPE – Molecular Phylogenetics and Evolution MSV – Mosimann Shape Variable(s) Myr – Million years N. – Nasalis Nas – Nasion Non-metric MDS – Non-Metric Multidimensional Scaling No. – Number(s) Ns – Nasospinale O – Opisthiocranion Ol – Orale Opn - Opisthion Or. - Orrorin Ou. – Ouranopithecus p. - page Para. – Paranthropus PC – Principal Component(s) PCA – Principal Component Analysis Pi. – Piliocolobus PNAS – Proceedings of the National Academy of Sciences, USA Pms – Palatomaxillary suture Po. – Pongo Pp. – Papio pp. – pages Prae. – Praeanthropus Pre. – Presbytis Pro. – Procolobus Pros – Prosthion Py. – Pygathrix R. – Rhinopithecus Rhi – Rhinion SD – Standard Deviation Sa. – Samburupithecus SAJS – South Africa Journal of Science Se. – Semnopithecus Sh. - Sahelanthropus VI Si. – Simias – Small Sigma = Standard deviation SS – Social Structure Sta – Staphylion Sv. – Sivapithecus Th. – Theropithecus Tr. – Trachypithecus TREE – Trends in Evolution and Ecology Var. – Variance or Variation ¯x – Sample mean or average Zi – Zygomaxillare inferior Zs – Zygomaxillare superior
VII The genus category and cranial morphometrics of the Catarrhini with implications for fossil hominins
Chapter 1: Introduction - Historical and current trends in anthropology & taxonomy The genus and species categories are the foundation of Linnaean binomial nomenclature and modern taxonomy. The official commencement of Linnaean binomial nomenclature for many vertebrates and invertebrates begins with the tenth edition of Linneaus’ Systema Naturae, published in 1758. This hierarchical classification scheme is internationally accepted and has become embodied in the International Commission of Zoological Nomenclature (ICZN). In fact, the combination of these categorical names brought order and provided stability to taxonomy, which is something that had eluded Linneaus’ naturalist predecessors. While a considerable amount of time and energy has been devoted to species concepts, the genus category has received little attention and is generally thought to be ‘unproblematic’ (Wood & Collard, 1999a). In its most basic capacity, the genus category is simply either one monotypic species or a group of closely related species. Despite this simple definition, implementing the genus category in practice is much more difficult, especially when evaluating living species which may have fossil ancestors. The problem lies in the philosophical reasoning and varied evidence (morphology, genetics, ecology, behavior, etc.) that are employed to demarcate generic boundaries, which can, depending on the researcher, produce different results. The study and classification of human evolution has been and continues to be a contentious area of research. Despite nearly 250 years (2007 C.E. minus 1758 C.E. = 249 years) of taxonomic nomenclature and meticulous research, the placement of modern humans and their closest living relatives, the great apes, remains a debated topic. In fact, major interpretative (i.e. philosophical) differences and analyses of varied evidence (biomolecular and morphological) exist in paleoanthropology and primate systematics as to how and why modern humans, fossil species and extant primates, are classified. The biological classification of humans and closely related fossil species has nearly always been a highly debated topic, engendering multidisciplinary research and taxonomic refinement or revision (Marks, 2005). However, this should not come as a surprise considering that humans are the most studied of all primates with no lack of scientific theory. Modern humans belong to the genus Homo Linneaus, 1758 and are designated specifically by H.
sapiens - a fact which is not questioned. Although created in a pre-evolutionary framework, the development and use of the genus Homo includes many complementary and diverse forms of evidence and inferences. However, currently within the paleoanthropological literature there are numerous proposed hominin genera in use. In fact, there are polarized extremes of their possible relationships. Although the ‘robust australopithecines’ or ‘paranthropines’ of Paranthropus Broom, 1938 are not considered to be on the direct line to modern humans, they are thought to be closely related because of bipedal adaptations and cranial synapomorphies (Tobias, 1988). For this reason, the taxonomy of genera employed in paleoanthropology is as “chaotic” as it was in the first half of the twentieth century (Marks, 2005) when purported hominin genera were proposed for fossil remains, most of which were thought to date to the Early and Middle Pleistocene. In fact, Groves (2001a) lists nineteen generic synonyms for the genus Homo (see also Smith, 2002). However, the reasons for their creation are different to those proposed in recent years. While the genera of the first half of the twentieth century were created due to a lack of evolutionary theory and misunderstanding of the purpose of taxonomy, the recently proposed hominin genera are more a product of the predominance of cladistic theory and phylogenetic systematics. Lastly, the definition of the family Hominidae and genus Homo, and the genera, and species (extinct and extant) attributed to both are under critical debate (Wood & Collard, 1999a; Groves, 1999 & 2001b; Begun, 2001 & 2004a; Cela-Conde, 2001; Wildman et al, 2003). This thesis hopes to address and contribute to these issues. Over the last decade and a half four purported hominin genera have been described. Prior to these only two genera were commonly used or necessary, Australopithecus Dart, 1925 and Homo Linnaeus, 1758. Once not generally used in the literature (Tobias, 1967 & 1968; cf. Robinson, 1972), the genus Paranthropus is now readily employed by paleoanthropologists (e.g. Wood & Strait, 2004; cf. Suwa et al, 1997; Alemseged et al, 2002). Ardipithecus ramidus was initially announced as a species of Australopithecus (White et al, 1994), but White et al (1995) later decided a new genus was necessary to accommodate the fossil material. From what has been published, Ardipithecus is very primitive and bipedality for this species remains to be substantiated (Haile-Selassie, 2001; Haile-Selassie et al, 2004). The announcement, analysis and subsequent debate of Orrorin continues to draw controversy (Senut, 2001; Aiello & Collard, 2001; Sawada et al, 2002;
2
Pickford et al, 2002; Galik et al, 2004 & 2005;Ohman et al, 2005; White, 2006). Aside from its initial announcement, Kenyanthropus (Leakey, 2001; Leiberman, 2001) has received little attention (two exceptions, Cameron, 2003; Cela-Conde & Ayala, 2003); no doubt due to its poor preservation (White, 2003). In addition, large and statistically robust cladistic analyses of craniodental variables by Strait et al (1997) and Strait & Grine (2005) have argued for the removal of Australopithecus afarensis from Australopithecus and instead be placed in its own genus, Praeanthropus in an effort to remove paraphyly (an argument which has gained support and use by others - Wood & Collard, 1999a; Cameron, 2003; Cela-Conde & Ayala, 2003; Grine et al, 2006). Interestingly, the afarensis hypodigm has for years been subject to debate, particularly in regard to its locomotor repertoire, species composition and phyletic position (Tobias, 1980; Olson, 1981 & 1985; White et al, 1981; Susman et al, 1984 & 1985; Kimbel et al, 1984, 1985 & 1988; White, 1985; Ward, 2002). Despite this, strong evidence presented by Kimbel et al (2006) and White et al (2006) may demonstrate a ‘chrono-species’ relationship between Au. anamensis and Au. afarensis based on fossil mandibular and dental specimens from sites in Ethiopia, Kenya and Tanzania, between ~ 4.2 to 3 Mya. This casts doubts on the cladistic analyses by Strait et al (1997). Lastly, Sahelanthropus, dated to nearly ~ 7 million years, is perhaps the most thoroughly documented and analyzed taxon (Brunet et al, 2005; Zollikofer et al, 2005; Guy et al, 2005) – and may be a close relative, but not the last common ancestor of humans and chimpanzees (Wolpoff et al, 2002 & 2006). Understanding an evolutionary lineage from only fossilized skeletal remains is extremely difficult, both practically and theoretically (Tattersall, 1986 & 1992; Wood, 1988, 1991 & 1993; Cope, 1993; Collard & Wood, 2000; Plavcan & Cope, 2002). Still confounding matters is the great potential for homoplasy and differing amounts of variation-covatiation in closely, and even distantly related species (Lieberman et al, 1996; Larson, 1998; Lockwood & Fleagle, 1999; Ackermann, 2002 & 2003). As to taxonomy below the genus level, some workers argue for many species of hominins (Tattersall, 1986 & 2000; Cameron, 2003). For example, in Wood & Richmond’s (2000) review of hominin taxonomy and paleobiology, based on fossil, geological age and paleoecological evidence, four genera compromising more than 15 species are discussed (not including Sahelanthropus, Orrorin or Kenyanthropus). Justification for these species
3
mainly rests on morphological differences, although geological age, location of discovery and associated fauna play a role as well. What is more, the possible inferences from these data may elucidate behavioral and ecological differences. Different analyses based on ecological variables such as home range size and body mass (Foley, 1991; and Conroy, 2002 & 2003) indicates that the number of species used within paleoanthropology is reasonable within a mammalian or primate context. Yet, Eckhardt (2000) and Curnoe & Thorne (2003) have suggested far fewer species, <5, based on molecular data. On the other hand, most researchers would classify modern humans, chimpanzees and gorillas into separate genera. In general, the modern usage of the genus Homo is relatively agreed upon and represents an evolutionary unit or threshold bound by anatomy, locomotion, archaeology, ecology, life-history variables and inferred behaviors (Hughes & Tobias, 1977; Duchin, 1990; Stanley, 1992; Wood, 1992; Grine et al, 1996; Kimbel et al, 1996; McHenry & Berger, 1998; Aiello & Wells, 2000; McHenry & Coffing, 2000; Moggi- Cecchi, 2001; Dunsworth & Walker, 2002; Stiner, 2002; Bramble & Lieberman, 2004; Bobe & Behrensmeyer, 2004). However, recently, some anthropologists and molecular researchers have suggested placing humans and chimpanzees in the same genus and perhaps the gorilla as well, because of their small genetic distance from each other and their estimated recent divergence in the Late Miocene/Early Pliocene (Avise & Johns, 1999; Castresana, 2001; Curnoe & Thorne, 2003; Watson et al, 2001 and Wildman et al, 2003). However, even extant species are difficult to demarcate, evident from numerous observations of natural hybridizations between species in the wild and zoos (Van Gelder, 1977; Bernstein & Gordon, 1980; Struhsaker et al, 1988; Vervaecke & Van Elsacker, 1992; Jolly et al, 1997; Detwiler, 2002; Grubb et al, 2003; Brandon-Jones, 2004). Yet, some could argue the species category is the most objective category in any classification design (Simpson, 1961; Mayr, 1969; Groves, 2004). Still others have discussed the arbitrary nature of Linnaean taxonomy above the species level (see Anderson, 1975; Groves, 1986; Cela-Conde, 1998; de Quieroz & Gautheir, 1992; McKenna et al, 1997; Ereshefsky, 2001). Furthermore, there is no consensus on the definition, identification or delimitation for any of the lower categories within Linnaean hierarchal taxonomy for extant populations or species known only from fossils. Additionally, researchers use different paradigms, definitions and identification
4
procedures which they feel is the most appropriate or correct (e.g. cladistics or evolutionary systematics). There are also strong arguments for abandoning Linnaean hierarchal taxonomy and instead using nonranked monophyletic clades – e.g. the Phylocode (see de Quieroz & Gautheir, 1992; Benton, 2000; Groves, 2004). Most importantly though, there are inherent limitations as to how theoretical paradigms and hierarchical systems can be applied to the natural world and its phenomena (Mayr, 1974 & 1981; Knox, 1998; Szalay, 1999; Vrba, 1999; Mayr & Bock, 2002). Taking these comments into consideration, it is peculiar that over the past decade and half there have been numerous genera and species created or resurrected within paleoanthropology, each one (excluding Paranthropus and perhaps Kenyanthropus) vying for the position of the ‘direct ancestor’ of modern humans. As these examples and others make clear, recent analyses and discoveries have made hominin taxonomy volatile and unstable. This is in part by order of discovery, analysis and scientific announcement but another problem is the inadequate guidelines in demarcating higher taxa. Hominin genera currently used in paleoanthropology include, in nearly geological order from the oldest to the most recent:
Sahelanthropus Brunet et al, 2002, ~ 6 - 7 Ma; Orrorin Senut et al, 2001, ~ 6 Ma; Ardipithecus White et al, 1995, ~ 5.8 - 4.4 Ma; Australopithecus Dart, 1925, ~ 4.2 - 2 Ma; Praeanthropus Senuyek, 1955; ~ 3.7 - 2.9 Ma (Resurrected by Strait et al, 1997; see also Day et al, 1980; and Strait & Grine, 2004); Kenyanthropus Leakey et al, 2001, ~ 3.5 Ma; Paranthropus Broom, 1938, ~ 2.5 - ~ 1 Ma; & Homo Linneaus, 1758, ~ 2.5 - Present.
Please note, other fossil hominids which most likely contributed to the evolution of African hominid genera, such as Morotopithecus, Kenyapithecus, Samburupithecus, Dryopithecus and Ouranopithecus, are not examined within this thesis (see Leakey, 1967; de Bonis et al, 1990; McCrossin & Benefit, 1993; Ishida & Pickford, 1997; de Bonis &
5
Kuofos, 1997; Begun, 1992a & 1994; Begun et al, 1997; MacLatchy et al, 2000; Kordos & Begun, 2002; Young & MacLatchy, 2004; MacLatchy, 2004; Finarelli & Clyde, 2004). This large number of closely related genera within the span of less than seven million years is unusual when compared to the evolution and classification of other primates. In addition, other generic and species names are available for fossil hominin material (Groves, 1999). There are several extant and extinct primate genera which span nearly four to five million years or greater - Colobus Illiger, 1811; Macaca Lacepede, 1799; Victoriapithecus von Koenigswald, 1969; Sivapithecus Pilgrim, 1910; Libypithecus Stromer, 1913; Theropithecus Andrews, 1916; Gigantopithecus von Koenigswald, 1935; Aotus Illiger, 1811 and Tarsius Storr, 1780 (see Delson, 1975 & 1980; Andrews, 1981; Szalay & Delson, 1979; Tattersall et al, 1988; Kohler et al, 2000; Jablonski, 2002; Simons, 2003; Raaum et al, 2005). Groves (2001a, p.19; 2001b, p.296) also provides many nonprimate examples of genera and families with genera ~ 6 to 4 million years in time length; e.g. Ursus, Canis, Elephantidae, Bovidae. Nonetheless, the likely length of time for any genus should be kept in mind (Stanley, 1978; Simons, 2003). Moreover, there are several other hominoid genera from the Middle and Late Miocene of Africa and Eurasia and Africa (Hill & Ward, 1988; Andrews, 1992; Andrews et al, 1997; Andrews & Humphrey, 1999) which most likely contributed to, and are closely related to, the origin of the African ape clade - extant Subfamily Homininae (= Gorilla, Pan and Homo) (Groves, 1986; Cote, 2004). Some of these may include Kenyapithecus (McCrossin & Benefit, 1993 & 1997); Samburupithecus (Ishida & Pickford, 1997; Pickford & Ishida, 1998); Dryopithecus (Begun, 1992a, 1994 & 1995; Moya-Sola & Kohler, 1993 & 1996; Begun & Kordos, 1997; Kordos & Begun 2002); and Ouranopithecus (De Bonis et al, 1990; De Bonis & Kuofos, 1993 & 1997). Wood & Collard (1999a) are correct in stating that the genus category has received little attention in comparison to the consideration given to various species concepts, which is perhaps reason for its misuse and misunderstanding. Some notable exceptions to this would be discussions and work by Inger (1958), Tappen, (1960), Wolpoff (1978), Wolpoff & Lovejoy (1974), Walker (1976), Stanley (1978), Tobias (1978 & 1991), Wood (1992) and Groves (1989 & 2001a, b). The genus is the first rank above species in Linnaean binomial nomenclature - the first grouping, or hierarchical rank, of closely related species.
6
Figuratively, the genus is the fundamental ‘keystone’ of many more overarching, inclusive categories; the maximum and minimum number of which has estimated to be ~ 11 to 16 categories for a group of 100 species (Anderson, 1975). Klein (1999, p.11) lists a potential twenty-one levels in Linnaean hierarchal taxonomy and McKenna et al (1997) use twenty- five in their classification of Mammalia above the species level (Class, subclass, infraclass, superlegion, legion, sublegion, infralegion, supercohort, cohort, magnorder, superorder, grandorder, mirorder, order, suborder, infraorder, parvorder, superfamily, family, subfamily, tribe, subtribe, infratribe, genus & subgenus). Whereas the possible number of evolutionary trees is nearly beyond human comprehension (Felsenstein, 1978). As quoted by Mayr (1969, p.93), Linnaeus’ dictum (1737), “It is the genus that gives the characters, and not the characters that make the genus”, does intuitively have some truth. A formal definition was offered by Mayr, “A genus is a taxonomic category containing a single species, or a monophyletic group of species, which is separated from other taxa of the same rank [other genera] by a decided gap” (p.92). Mayr continues, “It is recommended for practical reasons that the size of the gap be in inverse ratio to the size of the taxon. In other words, the more species in the group the smaller gap needed to recognize it as a separate taxon, and the smaller the species group the larger the gap needed to recognize it. One of the function of the genus, from Linnaeus’ time on, is to relieve the memory (to facilitate information retrieval), and the “inverse ratio” recommendation prevents recognition of a burdensome number of monotypic genera” (p. 92). To delimit genera as species groups of optimal size is an operation that requires experience, good judgment, and common sense” (p.92). Later, Mayr (1982) would also comment that a genus is, “…the lowest collective category, an aggregate of species sharing certain joint properties” (p. 175). A definition not too different from more recent researchers in standard biological and anthropological texts, such as Tattersall et al (1988), Jones et al (1992) and Harrison et al (1995). In an analysis of the genus Papio (the baboons), Buettner-Janusch (1966) commented that the genus category could be discussed “ad nauseam” and yet admits it does recognize “natural phenomenon” in the biological world (p.290). Unfortunately, the genus category is difficult to define and consistently implement, which can lead to differences in interpretation, accompanied with an inflation of genera that may or may not have any biological usefulness. For example, Groves (2001a,
7
p.308; see also Smith, 2002, p.442, table 26.1; Campbell, 1965; McKenna et al, 1997) lists twenty generic names that were proposed for fossils (most of which are thought to date to the Middle or Late Pleistocene) which are now all included within Homo. These examples are evidence for a lack of a solid philosophical and empirical basis for the genus category (and therefore, hierarchal rankings as well), which has been discussed by others (e.g. Groves, 1989, 2001a & 2004; de Quieroz & Gauthier, 1992; Valentine & May, 1996; Knox, 1998; Cela-Conde, 1998; Cantino et al, 1999; Wood & Collard, 1999a & b, 2001a & b; Ereshefsky, 2001; Sarmiento et al, 2002; Cela-Conde & Ayala, 2003; Wildman et al, 2003). This research hopes to contribute to this debate and refine generic classification. Yet, consensus remains elusive while the number of fossil primate genera increases unabated (e.g. pliopithecoids - Moya-Sola et al, 2001; Rossie & MacLatchy, 2006; cercopithecoids - Frost, 2001; Hlusko, 2006; hominoids - Moya-Sola et al, 2004; Ward et al, 1999; Gebo et al, 1997) alongside robust genetic data revealing previously unrecognized groupings (e.g. Catarrhines - Page & Goodman, 2001; Papionins - Disotell et al, 1992; Disotell, 1994; Harris& Disotell, 1998; Harris, 2000; Cercopithecins - Tosi et al, 2002a & b, 2003, 2004; African Bush babies - Delpero et al, 2000). However, attempts must be made and tested. As stated earlier, the incentive for the present thesis began with a critical examination of two very different proposals defining the genus Homo (Wood & Collard, 1999a; Goodman et al, 1998), which inevitably led to broader questions; 1) what is a genus and 2) is it applied to extant and extinct primates equivalently? Equivalency, here operationally defined as containing comparable amounts of biological content and potential inferences from available evidence. More precisely though, when evaluated comparatively with extant catarrhine genera, both quantitatively and qualitatively, are the recently proposed hominin genera (e.g. Sahelanthropus, Orrorin, Ardipithecus, Kenyanthropus) in paleoanthropological taxonomy and phylogenetic reconstructions equivalent and/or comparable to extant nonhominin genera? Certainly the available evidence on extant primate genera far outweighs the inferable evidence from the fossil record but can the currently used genera in taxonomic nomenclature and systematics serve as comparative benchmarks to gauge the validity of the recently proposed hominin genera?
8
Wood & Collard’s genus argument: Wood & Collard’s (1999a) analysis of the genus category was directly concerned with the fossil species attributed to the genus Homo and their relation to modern humans, Homo sapiens, which is also the type species of the genus. They concluded that a genus could be defined as “a species, or monophylum, whose members occupy a single adaptive zone” (Wood & Collard, 2001a; p.67). This definition is influenced and compromised by ideas from evolutionary systematics and cladistics. However, an ‘adaptive zone’ or ‘plateau’ is not a new idea (see Mayr, 1950; Van Valen, 1971; Wolpoff & Lovejoy, 1975). Thus defined, their critical examination of the available evidence led them to conclude that H. habilis and H. rudolfensis should be removed if paraphyly is to be avoided while a coherent adaptive zone is upheld. Their conclusions are very similar to Wolpoff’s (1999; see Wolpoff et al, 1999). This argument has been accepted by some, such as Aiello & Andrews (2000), while others are at least “sympathetic” to the proposal (Conroy, 2005, p. xxiii). Furthermore, for any species attributed to the genus Homo (or for that matter, any other genus or species - extant, extinct, and fossil lineages with living descendants), “First, the species should belong to the same monophyletic group as the type species of that genus. Second, the adaptive strategy of the species should be closer to the adaptive strategy of the type species of the genus in which it is included, than it is the type species of any other genus” (Wood & Collard, 2001a, p. 67). Moreover, these requirements agree quite well with taxonomic rules of the ICZN which does require a genus to be defined by its type species and any member compared to it (Tattersall et al, 1988, p.222). Justification for the removal of H. habilis and H. rudolfensis relies on a number of inferred life-history and ecological variables; as well as an objective evaluation of the assumed qualities or behaviors of the species generally allocated to the genus Homo. One implicit or explicit criterion of the genus Homo was spoken language. For example, Tobias (1991 & 1994) has argued that with the appearance of H. habilis there are diagnosable features on the endocasts of this species which indicate a human-like brain configuration, perhaps indicating the possession of spoken language. In addition, the appearance of stone tools was assumed to be made by members of Homo as well (Leakey et al, 1964; Tobias, 1967; however, bear in mind, there have been numerous field observations of many monkey and ape species that fashion and use tools; Yamakoshi, 2004). Although, some
9
have questioned this inference because gross fossilized endocast features do not necessarily indicate neurological capabilities as chimpanzees have similarly developed frontal and temporal cortical areas as well (Gannon et al, 1998; Semendeferi et al, 2002). Interestingly, Elton et al (2001) demonstrate that other hominins (e.g. Paranthropus) also experienced brain encephalization. Moreover, the Plio-Pleistocene deposits that yield the earliest stone tools also contain contemporaneous and sympatric hominin species, all of which predate any fossils that have been allocated to Homo, although temporally it is very close (Semaw et al, 1997; de Heinzelin et al, 1999; Kimbel et al, 1996). Still, another example includes the analysis of postcranial elements of H. habilis and H. rudolfensis, which Wood & Collard argue are quiet different from H. sapiens, most likely reflecting different locomotor behaviors and habitat use. However, Haeusler & McHenry (2004) have more recently reanalyzed OH 62, a fragmentary skeleton presumably representing H. habilis, and their results indicate that the lower limb was more human-like then had been originally reconstructed and interpreted but that the upper arms in this species were still relatively longer than modern humans.
Goodman et al genus argument: The definition of the genus category offered by Goodman et al (1998 and 2001; see also Wildman et al, 2003) was derived by the direct comparison of the DNA of living primate species complemented by fossil evidence, plus strict adherence to phylogenetic principles (e.g. taxa are monophyletic clades and the age of the clade decides the rank of the taxa; both originally proposed by Hennig, 1966). The nuclear DNA under study was the -globin gene cluster, ~ 60- to 80-kilobase (kb), and within this cluster are -type genes ( and ). Furthermore, the nuclear regions examined include a pseudogene, noncoding introns of the functional genes and the flanking regions of these sections which all evolve at a faster rate compared to the coding sequences. Thus, these regions are selectively neutral, provide local molecular clocks, and the -globin locus (a 1.7-kb section of three exons and two introns) has been sequenced for 43 primate species. The fossil record is used to calibrate these molecular results. For example, the fossil primate genera Victoriapithecus (a cercopithecoid) and Proconsul (a hominoid) both contemporaneous during the early Miocene of East Africa provide direct evidence that by ~
10
22 Ma (at the latest) the lineages leading to the Old World Monkeys and Apes had already diverged. As a result, 25 Ma is the temporal datum point for the cercopithecoid-hominoid split. In another example, Catopithecus (a eucatarrhine) at 35 Ma (late Eocene, at the latest) again provides direct evidence that the platyrrhine and catarrhine lineages had separated by this time. Consequently, 40 Ma is the inferred timing of divergence for the platyrrhine and catarrhine lineages. In addition the results presented by Goodman et al (1998 & 2001) are congruent with many other primate morphological and molecular studies (e.g. Primates - Shoshani et al, 1996 and Poux & Douzery, 2004; Strepsirhines -Yoder et al, 1996; Platyrrhines - Schneider, 1996 and Horovitz et al, 1998; Catarrhines - Raaum et al, 2004; Hominoids - Caccone & Powell, 1989; Ruvolo, 1997a, b). The conclusions reached by Goodman et al (1998 & 2001) suggest that the age for a genus treated as ‘crown’ groups should 6-4 Ma; and 11-7 Ma if the ‘total’ group is considered (“A crown group is an extant monophyletic taxon defined by its LCA [Last Common Ancestor]; therefore it includes the LCA and all lineages that descend from this LCA to both extinct and extant species, but it does not include the stem of the LCA, whereas the total group does include the stem” (ibid. p. 592).). Autonomously, similar generic age estimates were obtained by Curnoe et al (2006). Primate genera defined as such, treated either as a crown or total group, requires modern humans and the chimpanzees both to occupy the genus Homo because they are estimated to have diverged only 6 Ma and are sister species (Groves, 1986; Martin, 1986; Page & Goodman, 2001; Gibbs et al, 2002; Wildman et al, 2003; Uddin et al, 2004); thus resulting in, H. (Homo) sapiens, H. (Pan) troglodytes & H. (Pan) paniscus. The fruition of J. Diamond’s (1988 & 1992) prediction has been realized (DNA-based phylogenies of the three chimpanzees. Nature 332, 685- 686; and The rise and fall of the third chimpanzee. London: Vintage.). Consequently, all putative hominin fossil genera are then grouped under Homo and then separated subgenerically. Independently, other molecular researchers and anthropologists such as Avise & Johns (1999), Watson et al (2001), Castresana (2001) and Curnoe & Thorne (2003) have also proposed the congeneric status of humans and chimpanzees. However, genetic techniques and hypotheses generated to demarcate taxonomic boundaries are not without criticism (e.g. Ferguson, 2002).
11
This is counter to other anthropologists who recognize up to 8 different hominin genera (Cameron, 2003; Strait & Grine, 2004; Wolpoff et al, 2006) excluding Pan, which to some is not a hominid (e.g. Martin, 1990; Fleagle, 1998; Sarmiento, 1998; Foley, 1999b; Nowak, 1999; Haile-Selassie et al, 2004). Despite theoretical developments (e.g. Hennig, 1966), modern genetics (e.g. Kumar, 2005) and an ever expanding fossil record (e.g. Hartwig, 2002), have Simpson’s (1945) critical comments regarding primate taxonomy influenced many investigators (see also Martin, 1981; Eckhardt, 2000; Tattersall, 2000; Foley, 2001; Wildman et al, 2003; Marks, 2005)? For example, Simpson wrote, “The primates are inevitably the most interesting of mammals to an egocentric species that belongs to this order. No other mammals have been studied in such detail, yet from a taxonomic point of view this cannot be considered the best-known order, and there is perhaps less agreement as to its classification than for most other orders. A major reason for this confusion is that much of the work on primates has been done by students who had no experience in taxonomy and who were completely incompetent to enter this field, however competent they may have been in other respects, and yet once their work is in print it becomes necessary to take cognizance of it. For this reason, if no other, it is not surprising that most primates have alternative names and that hardly two students use the same nomenclature for them. The importance of distinctions within the group has also been exaggerated that almost every color phase, aberrant individual, or scrap of fossil bone or tooth has been given a separate name, almost every really distinct species has been called a genus, and a large proportion of the genera have been called families. The peculiar fascination of the primates and their publicity value have almost taken the order out of the hands of sober and conservative mammalogists and have kept, and do keep, its taxonomy in a turmoil. Moreover, even mammalogists who might be entirely conservative in dealing, say, with rats are likely to lose a sense of perspective when they come to the primates, and many studies of this order are covertly or overtly emotional” (pp. 180-181); [and] “Perhaps it would be better for the
12
zoological taxonomist to set apart the family Hominidae and to exclude its nomenclature and classification from his studies” (188). Several other researchers have made similar comments – Anderson & Jones (1967); Simons (1963); Buettner-Janusch (1966); Tattersall & Eldredge (1977). Instead it would seem paleoanthropology and its taxonomic nomenclature are repeating past trends but for different reasons. While purported hominin genera prior to the evolutionary synthesis and cladistics were described because of non-population interpretations, misunderstandings of variation within species and misguided taxonomy, modern hominin genera have become splintered clades all deserving recognition. The similarities between Wood & Collard’s (1999a) arguments to those of Goodman et al (1998 & 2001) lie in their adherence to cladistic theory and recognition of monophyletic taxa only. However, the recognition of these taxa results from different forms of evidence (molecular and morphological) and analysis, which consequently result in dissimilar classifications (or each offers different resolution of macroevolutionary events), each potentially valid and useful; and both deserving attention and empirical testing. There has been a long debate and intense research on the relationship between molecular and morphological data and its potential inferences for primates and mammals in general (King & Wilson, 1975; Cronin & Meikle, 1979; Cherry et al, 1978 & 1982; Andrews, 1986 & 1987; Wayne et al, 1991; Smith & Littlewood, 1994; Purvis, 1995; Mann & Weiss, 1996; Shoshani et al, 1996; Pilbeam, 1996; Horovitz et al, 1998; Omland, 1997; Lovejoy et al, 1999 & 2000; Hillis & Wiens, 2000; Eckhardt, 2000; Weiss & Fullerton, 2000; Tosi et al, 2003; Curnoe & Thorne, 2003; Roseman, 2004). Needless to say, cladistic theory and associated methods (also having emerged alongside molecular biology in the second half of the twentieth century) have come to dominate particular aspects of biological classification and evolutionary theory, particularly within paleoanthropology. In fact, large and robust phylogenetic analyses for extinct and extant hominins and nonhuman primates have been performed, producing varied results with most disagreements the outcome of rank assignment and taxon content (Eldredge & Tattersall, 1975; Delson et al, 1977; Schwartz et al, 1978; Kluge, 1983; Boaz, 1983; Ciochon, 1983; Skelton et al, 1986; Strasser & Delson, 1987; Harrison, 1987; Stringer, 1987; Andrews & Martin, 1987; Chamberlain & Wood, 1987; Begun 1992a & 2001 and Begun et al, 1997; Groves, 2000;
13
Gibbs et al, 2003; Strait & Grine, 2004; Finarelli & Clyde, 2004). However, there have also been many other researchers who urge caution and question the results of cladistics and the demand for strict monophyly in classifications and instead argue for other forms of evidence with different approaches to taxonomic schemes (Mayr, 1974 & 1981; Hull, 1979b; Martin, 1981; Harrison, 1993; Corruccini, 1994; Begun, 1995 & 2004a; Eckhardt, 2000; Collard & Wood, 2000 & 2001b; Curnoe, 2003; Ackerman, 2003; Hawks, 2004; Hlusko, 2004; Nadal-Roberts & Collard, 2004; Andrews & Harrison, 2005). This thesis explores the definition and use of the genus category in extant Catarrhine systematics and paleoanthropology. In particular, what is the most appropriate and objective manner(s) in determining the boundaries for the genus Homo based on the available evidence. Likewise, can extant Catarrhini genera and the available evidence they provide be drawn upon to assess the recently proposed hominin genera; in addition to other arguments to include perhaps the chimpanzees within Homo. Biological classification above the species level is a strongly debated topic. Arguments and evidence used to define the genus Homo can be based on behavioral, morphological (including soft tissue, skeletal features and fossils) and genetic data. Two hypotheses defining the genus Homo tested here are those of Wood & Collard (1999a) and Goodman et al (1998). The former is grounded upon various forms of evidence which can be gleaned from the available extant species and fossil evidence, which in its totality apparently defines an ‘adaptive zone’ of one or more species. The latter relies almost exclusively on biomolecular evidence (specifically, sequenced nucleotides and protein structure) but is calibrated by the fossil record and insists time-ranked monophyletic clades. The cranium was chosen as a skeletal element to record morphological variation within and between genera and because many primate crania have been recovered from the fossil record.
Taxonomy and Nomenclature: One source of confusion in taxonomy is simply human language. Language, spoken and written, can be at times repetitive, ambiguous and contradictory. However, it is nonetheless the medium with which we physically and mentally engage with other humans; chimpanzees are anatomically incapable of articulate human-like speech. Taking this into consideration, it is paramount to know exactly the content of our subjects but also how to
14
appropriately and accurately discuss them. Any taxonomic discussion or analysis must first set out precise and clear definitions for the words (i.e. nomenclature) used (Simpson, 1961). The analysis herein has adopted the taxonomic nomenclature proposed by Groves (2001a) although species and subspecies assignments are not the main focal point. Please note however, this research in no way questions the validity of the species proposed by Groves. They most certainly are the most recent and accurate descriptions available of extant primate populations which are diagnosable from others. It should be noted that Groves uses the phylogenetic species concept (see Cracraft, 1983; and Kimbel & Rak, 1993). Therefore, other researchers can and do disagree with his species groupings and their hierarchical rank. However, his detailed research in the matter has produced a comprehensive taxonomic nomenclature with which we may accurately and appropriately (in terms of nomenclature history and priority) discuss the Order Primates. Furthermore, Groves’ taxonomic placement of taxa is not too radically different from other recent schemes (e.g. Hill, 1966, 1974 & 1970; Delson & Andrews, 1975; Delson, 1977; Schwartz et al, 1978; Wolfheim, 1983; Schwartz, 1986; Groves, 1970 &1989; Napier, 1981 & 1985; Conroy, 1990; Rowe, 1996; Shoshani et al, 1996; McKenna et al, 1997; Fleagle, 1998; Ankel-Simons, 1999; Nowak, 1999). Furthermore, Groves’ taxonomy is supplemented by the data and observations of Grubb et al (2003) and Brandon-Jones et al (2004); both of which Groves is a co-author and represent some of the most recent syntheses for the Old World Primates. Of course, though, some genera are monotypic which inevitably leads to comparing a genus of one species to a genus with many and vice versa; although, some hominin genera are monotypic as well (e.g. Kenyanthropus, Orrorin, Sahelanthropus). Yet, analyses presented here are directly concerned with the genus category and how this is used in modern catarrhine systematics compared to that of fossil hominin genera (Wood & Collard, 1999a,b & 2001a,b); as well as, considering molecularly derived time-ranked taxonomic hypotheses (Goodman et al, 1998 & 2001; Wildman et al, 2003). Accordingly, the genus is the ‘operational taxonomic unit’ for the analyses presented here and therefore is concerned with macroevolutionary events and their classification. Validation for this level of analysis is supported by the results from Purvis & Webster (1999; see also Purvis, 1995; Purvis et al, 1995) and Hardt & Wenke (2006).
15
Purvis & Webster’s (1999) macroevolutionary analysis suggest the use of generic means allows for some form of independent contrasts. While Hardt & Wenke’s (2006) study included geometric morphometrics of hominoids and cercopithecoids and despite heterogeneous craniometric samples their results still contain reliable biological information and are appropriate because of their ability to successfully discriminate superfamilies and genera. However, please note, this thesis is not directly concerned with the cladistic relationships of the genera under study, nor can the cranial morphometrics collected definitively answer specific phylogenetic questions - they are not meant to. (By and large the major phylogenetic relations of the Catarrhini have been determined.) Rather, do the species of the genera presently in use actually share particular cranial features and dimensions due to their supposed adaptive zones? Lastly, in accordance with molecular findings since the 1960’s (Goodman, 1963a & 1963b), humans (Homo) and the great apes (Pan, Gorilla & Pongo) together belong within the family Hominidae and thus all are considered ‘hominids’, a term traditionally used to describe only humans and their bipedal fossil ancestors (e.g. Campbell, 1962 & 1965; Pilbeam, 1962). Consequently, the term ‘hominin(s)’ (Tribe Hominini (= Pan-clade and Homo-clade), Subtribe Hominina (= Homo + bipedal species more closely related to modern humans) is necessary in discussion to distinguish fossil human ancestors from the great apes. Within the Hominidae are two subfamilies, the Ponginae which includes the only extant Asian ape, Pongo spp., and fossil genera such as Sivapithecus, Lufengpithecus and Khoratpithecus (Chaimanee et al, 2003 & 2004). The other subfamily, the Homininae designates the African Apes (Gorilla, Pan and Homo). Groves (2001a) goes no further then subfamilial classification for the African apes and will be dealt with more fully in Chapters 6-9. Thus, ‘hominins’ (as ‘homininans’ is a bit cumbersome and perhaps unnecessary; cf. Andrews & Harrison, 2005) in this analysis refers to fossil species thought to be more closely related to humans than to chimpanzees (Chapter 6, Figure 6.1). Additionally, there is no fossil evidence for chimpanzees older than the Middle Pleistocene (McBrearty & Jablonski, 2005). As such, there should be no ambiguity or confusion as to what is being discussed or referred to. That is, until ancient (>3-5 Ma) fossils are recovered and their relationship to modern chimpanzees is demonstrated.
16
However, there are still some who prefer the traditional, gradistic classification in which the great apes are placed in the Pongidae, while humans and their immediate ancestors are the only members of the Hominidae (e.g. Martin, 1990; Fleagle, 1999; Nowak, 1999; Haile-Selassie et al, 2004; Grehan, 2006). Moreover, ‘cercopithecines’ of the subfamily Cercopithecinae refers to the cheek-pouch monkeys; ‘cercopithecins’ refers to members of the tribe Cercopithecini (Allenopithecus, Miopithecus, Cercopithecus, Chlorocebus & Erythrocebus) (Figure 3.1); and lastly, ‘papionins’ refers to members of the tribe Papionini (Cercocebus, Mandrillus, Macaca, Lophocebus, Papio & Theropithecus) (Figure 4.1). ‘Colobines’ refers to any member of the subfamily Colobinae (multi- chambered stomach monkeys) but the subtribe Colobina will be used to discuss the African colobines (Colobus, Piliocolobus and Procolobus); Presbytina designates the South and Southeast Asian langurs and leaf-monkeys (Semnopithecus, Trachypithecus and Presbytis); and lastly, Nasalina will be employed for the East and Southeast Asian odd-nosed colobines (Pygathrix, Rhinopithecus, Nasalis and Simias) (Figure 5.1). Finally, ‘cercopithecoids’ refers to any member of the superfamily Cercopithecoidea (Old World Monkeys), while ‘hominoids’ denote the lesser and great apes (including modern humans) of the superfamily Hominoidea (Figure 6.1).
17
Classification outline for the extant Catarrhini: Order: Primates Suborder: Strepsirrhini Suborder: Haplorrhini Infraorder: Tarsiiformes Infraorder: Simiiformes (Eusimiiformes) Parvorder: Platyrrhini Parvorder: Catarrhini (Eucatarrhini) (Neocatarrhini) Superfamily: Cercopithecoidea Family: Cercopithecidae Subfamily: Cercopithecinae Tribe: Cercopithecini Allenopithecus nigroviridus Miopithecus spp. (2 species) Cercopithecus spp. (25 species) Chlorocebus spp. (6 species) Erythrocebus patas Tribe: Papionini Cercocebus spp. (6 species) Mandrillus spp. (2 species) Macaca spp. (20 species) Lophocebus spp. (3 species) Papio spp. (5 species) Theropithecus gelada Subfamily: Colobinae Subtribe1: Colobina (African colobines) Colobus spp. (5 species) Piliocolobus spp. (9 species) Procolobus verus Subtribe: Presbytina (Langurs and leaf-monkeys) Semnopithecus spp. (7 species) Trachypithecus spp. (17 species) Presbytis spp. (11 species) Subtribe: Nasalina (Odd-nosed colobines) Pygathrix spp. (3 species) Rhinopithecus spp. (4 species) Nasalis larvatus Simias concolor Superfamily: Hominoidea Family: Hylobatidae Hylobates spp. (4 subgenera and 14 species) Family: Hominidae Subfamily: Ponginae Pongo spp. (2 species) Subfamily: Homininae Gorilla spp. (2 species) Pan spp. (2 species) Homo sapiens
1 Please note, Groves (2001a) does not formally designate nor define colobine subtribes but these are employed herein to facilitate taxonomic discussion and provide comparative boundaries. Bold numbers indicate genera with >10 species. 18 Chapter 2: Material and Methods
2.1 Aims and Objectives: The impetus for this research commenced with a critical evaluation of two proposals defining the genus Homo; 1) Wood and Collard (1999a) and 2) Goodman et al (1998). Initially, this endeavor began with a thorough investigation of the anthropological and primatological literature for examples of use, and evidence employed to define the genus category in extant Catarrhini systematics because humans and their closest relatives, living or extinct, are catarrhines as well. However, this soon revealed that even many extant genera, similar in some ways to fossil hominin genera, are also marked by taxonomic differences of opinion and interpretation. To evaluate the genera used within Catarrhini systematics and paleoanthropology many lines of evidence must be assessed and attempting equal consideration. But how can biomolecular data, morphology, ecology, substrate use, diet and the fossil record be synthesized and interpreted on common ground (e.g. Pilbeam, 1996; Pilbeam & Young, 2004)? Questions one to six will be answered and discussed at the end of chapters three to six while questions seven to nine will be dealt with in the final chapter. Questions explored in this thesis include: 1. Are the extant Catarrhini genera adequately defined and do they compose an ‘adaptively coherent’ grouping of one or more species which occupy an ‘adaptive zone’, as defined by Wood & Collard (1999a)? 2. In terms of cranial morphology, how distinct are genera? In other words, how readily may a genus be determined from cranial dimensions, considered either singular or multiple (both metrical and nonmetrical traits)? 3. How much morphological variation is encompassed within a Catarrhine genus? Or, in other words, how does a ‘genus’ or ‘species’ samples behave statistically? 4. How much has one genus morphologically and genetically diverged from another? 5. Do cranial morphometric similarities and/or differences reflect adaptive zones? 6. What analogies may be drawn from extant Catarrhini genera for the interpretation of fossil hominin genera?
7. Are the recently proposed fossil hominin genera adaptively coherent occupying an adaptive zone different from that of other hominin genera based on comparisons with extant Catarrhini genera? 8. Is the congeneric placement of humans and chimpanzees (Goodman et al, 1998 & 2001 and others) more parsimonious than each occupying separate genera? 9. Would the congeneric placement of humans and chimpanzees better explain and synthesize the available evidence or oversimplify evolutionary differences? Or put more simply, what types of evidence best delineate generic boundaries?
2.2 Introduction: The guiding principles which this thesis is built upon are taxonomic uniformitarianism and equivalence. Modern osteological and dental specimens provide a baseline from which the fossil record may be gauged and interpreted (Wolpoff, 1971 & 1978; Wood, 1991; Plavcan & Cope, 2002). In addition, by knowing the behavioral and ecological context of extant skeletal material the paleobiology of fossil genera and species may also be inferred (Chiochon, 1993; Bromage, 1999; Elton, 2001; Reed, 2002). Quantifiable and morphological features, either singular or multiple, further reveal critical similarities and important functional differences, both of which may be due to either close phylogenetic relationship or evolutionary convergence (or homoplasy). Of course phenetic studies have practical and theoretical limitations (de Quieroz & Good, 1997) but they do nonetheless organize, quantify and synthesize biological data, from which many sound inferences and arguments are possible (Olson & Miller, 1958; Sneath & Sokal, 1973; Corruccini, 1975). The evolution and classification of modern Homo sapiens is perhaps best achieved and interpreted within a comparative evolutionary framework which also provides for some degree of objectivity via numerous pairwise comparisons (Gittleman & Luh, 1992; MacLarnon, 1999). Modern humans belong to the genus Homo Linnaeus, 1758, a member of the Catarrhini (Old World monkeys (OWM) and apes), which is the focus of this study. The Catarrhini were chosen because they contain relatives of Homo, and represent a natural grouping bound by anatomy, time, paleoclimatic events and biogeography (Delson & Andrews, 1975; Fleagle & Kay, 1985; Delson, 1985; Andrews, 1985; Andrews et al, 1996;
20
Jablonski & Kelley, 1997; Harrison, 2004). More importantly, there is a long and productive history of comparative studies, providing behavioral, morphological, and genetic insights (Wolpoff, 1978; Dunbar, 1983; Foley, 1987; Fleagle & Reed, 1996 & 2004; Reed, 1997; Isbell et al, 1998; Jolly, 2001; Elton, 2006). The definitions of the genus category examined here are those of Wood & Collard (1999a) and Goodman et al (1998 & 2001). Both definitions are quite different yet do share some theoretical criteria. As discussed previously, Wood & Collard’s solution to this problem was to use elements from both evolutionary systematics and cladistics. Their definition of a genus is, “a species, or monophylum, whose members occupy a single adaptive zone” (2001a, p. 67), defined by such features as body size, relative brain size, diet, locomotion and masticatory features, some of which may be inferred from cranial morphologies and other forms of complementary evidence. However, the definition provided by Wood & Collard, if useful and compelling, should apply not only to modern humans and their fossil ancestors but also to extant primates and their fossil antecedents. This requirement also applies to the time- ranked genus categories proposed by Goodman et al (1998 & 2001), which if convincing should more accurately depict evolutionary relationships then previous classifications as well as synthesizing the available evidence. To test these definitions, direct comparisons between primate genera were necessary. Comparisons between genera will remain within their respective superfamily, subfamily, subtribe and tribe. In addition, interpreting results from different (though intimately related) forms of evidence (molecular and morphological) is crucial. The skeletal complex chosen for generic comparison in this investigation are the dimensions of the neuro- and viscerocranium (ex. the mandible) of provenienced adult catarrhines, which provide many indicators of the genera’s adaptive zone and ecomorphology (Bromage, 1999). In addition, interpreting results from different (though intimately related) types of evidence (biomolecular and morphological) is crucial. However, Hillis & Wiens (2000) are quite astute in stating, “Morphological and molecular studies can each address questions that cannot be addressed by the other” (p.1).
21
Table 2.1: Catarrhini genera and sample sizes included in this study (listed in alphabetical order).
Males Females 4 2 Allenopithecus nigroviridis (Lang, 1923) Allen’s Swamp Monkey; Africa 10 9 Cercocebus spp. (E. Geoffroy, 1812) Mangabeys; Tropical Africa 12 12 Cercopithecus spp. (Linnaeus, 1758) Guenons; Tropical Africa 13 11 Chlorocebus spp. (Gray, 1870) Vervets or Grivets; Africa 15 11 Colobus spp. (Illiger, 1811) Black & White colobines; Tropical Africa 9 7 Erythrocebus patas (Schreber, 1774) Red Patas Monkey; Africa 17 17 Gorilla spp. (I. Geoffroy, 1853) Gorillas; Tropical Africa 11 9 Homo sapiens (Linnaeus, 1758) Humans; Global 16 15 Hylobates spp. (Illiger, 1811) Gibbons & Siamangs; East & Southeast Asia 11 12 Lophocebus spp. (Palmer, 1903) Pseudo-mangabeys; Tropical Africa 15 19 Macaca spp. (Lacepede, 1799) Macaques; N. Africa, Gibraltar & East Asia 8 14 Mandrillus spp. (Ritgen, 1824) Mandrill & Drill; Tropical Africa 9 10 Miopithecus spp. (I. Geoffroy, 1842) Talapoin; Tropical Africa 18 10 Nasalis larvatus (E. Geoffroy, 1812) Proboscis Monkey; Borneo 19 14 Pan spp. (Oken, 1816) Chimpanzees & Bonobos; Tropical Africa 19 12 Papio spp. (Erxleben, 1777) Baboons; Africa & Arabian Peninsula 15 13 Piliocolobus spp. (Rochebrune, 1887) Red colobines; Tropical Africa 14 12 Pongo pygmaeus (Lacepede, 1799) Orangutans; Borneo & Sumatra 8 10 Presbytis spp. (Eschscholtz, 1821) Surilis; East & Southeast Asia 3 2 Procolobus verus (Rochebrune, 1887) Olive colobine; West Africa 14 11 Pygathrix spp. (E. Geoffroy, 1812) Douc Langurs; East Asia 5 12 Rhinopithecus spp. (Milne-Edwards, 1872) Snub-nosed colobines; E. Asia 5 4 Semnopithecus spp. (Desmarest, 1822) Hanuman langurs; S. & E. Asia 2 5 Simias concolor (Miller, 1903) Simakobu; Mentawai Islands 7 7 Theropithecus gelada (I. Geoffroy, 1843) Gelada; Ethiopia, Africa 15 18 Trachypithecus spp. (Reichenbach, 1862) Lutungs; South & East Asia 294 278 Total 572
22
2.3 The Primate Cranium: The skeletal complex chosen for study was the primate skull and its various dimensions (Figures 2.1 & 2.2). The skull provides many indications of species life history, evolution and behavior, providing adaptively and phylogenetically significant data and observations for all species of primates thus far studied (Albrecht & Miller, 1993). For example, Aiello & Wood (1994) and Spocter & Manger (2007) successfully employ cranial variables to estimate body mass of hominin fossil species. Furthermore, cranial morphology may also reveal ecomorphologies (Bromage, 1999). Some of these include nuchal breadth relative to body size and the presence or absence of a sagittal crest. Additionally, just as the size and shape of the eye orbits can reflect diurnal or nocturnal behaviors (Martin, 1990), are there any traits which can differentiate genera and their adaptive zones? The cranial measurements collected for this study serve many functions. First, many variables of an adaptive zone, as defined by Wood & Collard (1999a), may be inferred from cranial dimensions; some of these include brain size and masticatory apparatus dimensions. Second, the cranial dimensions recorded serve as a surrogate measure of generic ranges of morphological variation and composition. Third, craniometric study is a historically established form of anthropological discourse generating both continued scientific debate and research (Howells, 1973 & 1989; Liebermann, 1995; Lahr, 1996) and also allows the results from this analysis to be compared and evaluated with previous work (e.g. Verheyen, 1962; Vogel, 1966 & 1968; Howells, 1973 & 1989; Creel & Preuschoft, 1976; Marroig & Cheverud, 2001a & 2001b; Pan & Groves, 2004; O’Higgins, & Pan, 2004; Miller et al, 2004). Incidentally, many primate crania and cranial fragments have been recovered from the fossil record of Africa and Eurasia (see Wood, 1991; Tobias, 1991; Schwartz & Tattersall, 2002, 2003 & 2005). Lastly, recent cranial studies have suggested that when cranial morphometrics are treated with multivariate statistics the results are similar to and behave like neutral genetic evolution (Relethford, 1994 & 2002; Roseman, 2004; Brace, 2005; Harvati & Weaver, 2006), which better allows these results to be compared to genetic data. The data presented here are based on primate cranial morphometrics, whilst the genetic, postcranial and ecological data were obtained through the published literature. Fortunately primate skulls are heavily sampled by national museums and universities.
23
Determination of sex with many primate genera was relatively easy, especially those with extreme sexual dimorphism, such as, Papio or Pongo. Still, if there were questions regarding sexual assignment, the institute’s records and notes were employed. Extant primate crania measured and included within this study were gathered from the following institutions (listed in order of visit); The Australian Museum, Sydney, Australia; University of Witwatersrand, School of Anatomical Sciences, Johannesburg, South Africa; Transvaal Museum, Pretoria, South Africa; American Museum of Natural History, New York City, New York, USA; Hamann-Todd Collection at the Cleveland Museum of Natural History, Cleveland, Ohio, USA; Chicago Field Museum, Chicago, Illinois, USA; British Museum of Natural History, London, UK; and the Smithsonian Institute, Washington, D.C., USA. Craniometric data of fossil hominins was gathered from the pertinent literature (e.g. Weindenreich, 1943; Santa Luca, 1980; Pilbeam, 1982; Kimbel et al, 1984, 1994 & 2004; De Bonis et al, 1990; Kidder et al, 1992; Alpagut et al, 1996; Arsuaga et al, 1997; Lockwood & Tobias, 1999 & 2002; Gabunia et al, 2000; Vekua et al, 2002; De Lumley et al, 2006; Curnoe & Tobias, 2006; Rightmire et al, 2006). But of course, any phenetic examination depends on the subject and medium of investigation, including the biological populations (of any taxonomic rank) involved. In other words, “In interpreting these results it would seem wise to bear in mind that any conclusions which are drawn must relate to the parameters which were selected for analysis” (Day, 1967, p. 324). Day continues to reason, “A multivariate technique, by taking account of the correlations between individual parameters in each bone, can present the functional morphology of that bone as a unified concept and assign it a position in relation to others; clustering of points will produce a grouping of bones with similar functional affinities. It is likely that these groups will represent populations of a Linnaean species, but this is not inevitable because Linnaean taxonomy is usually based on a wider selection of parameters than those which can be realized by examining a single bone” (p. 324).
2.4 Cranial Measurements:
The linear measurements chosen for this investigation fall into two categories; 1) the absolute size of a particular structure or complex, and 2) the measured distance between suture junctions or their length. In turn, these two categories document the two
24
developmental and functional regions of the skull, the neurocranium and viscerocranium. Length (or height) measurements are generally in line the median plane whereas breadth (or width) dimensions are usually transverse to the median plane. However, taking into consideration of the extent to which modern humans have experienced unprecedented encephalization (e.g. Ruff et al, 1997), a limited number of measurements of the neurocranium were selected. This was done so as not to unduly overweight measurements of the neurocranium. Instead, several facial and suture length dimensions were collected because these are more indicative of phylogenetic relationships (Ackermann, 2002; Young, 2005) and social (e.g. sexual dimorphism between males and females) and/or dietary or positional adaptations (e.g. a sagittal crest or strongly developed nuchal crest). Furthermore, cranial morphometric studies in general reveal phylogenetic signals and/or disposition while postcranial morphometric examinations retrieve functional similarities and/or convergence (Oxnard, 1975 & 1998). Additionally, most researchers agree that genera should be monophyletic (see Wood & Collard, 2001a & 2001b; Groves, 2001a & 2004; Tosi et al, 2004 for examples and discussion). Thus, it is perhaps more appropriate to record and examine crania rather than the postcranial skeleton for testing generic boundaries. Wood’s (1991) monograph on the Plio-Pleistocene hominin cranial remains from Koobi Fora, Kenya, served as a guide and reference to the craniometric data collected for the analysis herein, including some abbreviations, although other measurements and terminology have been employed. These include Buikstra & Ubelaker (1994), Bass (1995), White (2000), Ackermann (2002) and Ackermann & Cheverud (2004). Moreover, many other researchers influenced theoretical, practical and descriptive aspects of the present examination, such as, Albrecht & Miller (1993), Relethford (1994 & 2002), Aiello et al (2000), Pan & Oxnard (2001), Frost et al (2003), Guy et al (2003), and Lieberman et al (2004). All measurements are chord lengths recorded in millimeters, many divided by others to produce proportional measures. Additionally, measuring suture lengths between cranial elements provides important phylogenetic information; also allowing homologous features across diverse primate genera to be measured and compared (Ackermann 2002; Ackermann & Cheverud, 2004). For paired measurements the average of right and left was recorded, although, if a specimen was damaged, measurements were taken on the intact and
25
preserved portions. Occasionally specimens are damaged and cannot be recorded. Thus, substituting the sample’s mean is perhaps necessary for some statistical computation (Rightmire, 1969; Howells, 1973). Additionally, statistical programs (PAST ver. 1.53- Hammer et al, 2006; and SPSS 11.0.1, 2001) optionally insert means for various tests or provide casewise deletion for missing values. Substituting the sample mean for certain measurements of particular individuals was necessary for only a limited number of individuals. Furthermore, these individuals were both included and excluded from statistic analyses to gauge their impact. From a preliminary investigation it was determined that this procedure did not greatly alter the results and thus were included. For assorted portions of this analysis different spreadsheet and statistical computer programs were utilized. These include, Microsoft Office XP Excel 2003 with the add-in StatistiXL ver. 1.5; SPSS 11.0.1 (2001); and PAST ver. 1.53 (Hammer et al, 2006). In addition, StatSoft, Inc. (2007; see http://www.statsoft.com/textbook/stathome.html) was routinely consulted for statistical advice, underlying assumptions and explanations of statistical methods. Lastly, the measurement of error (see White, 2000, p.307) determined for eighteen skulls (comprising eleven catarrhine genera) measured twice for thirty-six measurements was < 3.3% (range 1.7-6.4) with a standard deviation of 0.02. In addition, the average difference in distance between standard calipers and 3-D digitizing microscribe (using Rhinoceros 3.0 to generate linear distances between craniometric points) was 1.4 mm (range 0.61 - 2.8) with a standard deviation of 0.60. There should be no discrepancy between measurements taken with calipers or microscribe. Franklin et al (2005) produced very similar results using a 3-D digitizer compared to those based on data collected with standard calipers.
1. Cranial vault length - The chord distance between glabella and opisthiocranion (g-o) was determined with spreading (or hinge) calipers with the basicranium placed on a cranial rest and the right or left lateral side facing the observer. The chord length from glabella to opisthiocranion was included because this dimension represents the greatest length of the neurocranium in the median plane (Wood, 1991 - no. 1, p. 287).
26
2. Maximum cranial length - The chord distance between prosthion and opisthiocranion (pros-o) was determined with spreading calipers with the basicranium placed on a cranial rest and the right or left lateral side facing the observer. Please note, in some primate genera opisthiocranion is inion, or immediately superior to it – e.g. Colobus, Macaca, Papio, etc. The chord length from prosthion to opisthiocranion was chosen because it represents the maximum length of the entire cranium in the median plane (Wood, 1991 - no. 2, p. 287). 3. Biasterionic breadth (biast) - The chord distance between left and right asterion was measured by dial (or sliding) calipers using the lower arms, with the neurocranium placed on a cranial rest and the occiput facing the observer. If the area of asterion is complicated with swirling ossicles, the greatest width and extent of the occipital bone (although near the junction of the temporal and parietal bones) was considered to be asterion. Biasterionic breadth was included because it represents the greatest width of the occipital bone and is the juncture of three important neurocranium bones (the parietal, occipital and temporal bones). Furthermore, the occipital bone provides surface area for many functionally important nuchal muscle attachments (Wood, 1991 - no. 14, p. 288). 4. Biporionic breadth (bipor) - The chord distance between left and right porion was measured with dial calipers with the lower arms positioned superolaterally to the most lateral extent of the bony tube of the external auditory meatus with the tips of the calipers resting on the inferior root of the zygomatic process with the neurocranium placed on a cranial rest and the occiput facing the observer. Biporionic breadth was chosen because it is an excellent indicator of basicranium breadth perpendicular to the median plane and is significantly correlated with body weight (Aiello & Wood, 1994; Wood, 1991 - no. 11, p. 287). 5. Biauriculare breadth (biaur) - The chord distance between left and right auriculare was measured using spreading calipers using the rounded end points of the calipers resting in the deepest incurvature at the root of the zygomatic processes, with the basicranium placed on a cranial rest and the occiput facing the observer. Biauriculare breadth for this study is the operational measurement for the greatest width of the cranium perpendicular to the median plane because it was a
27
homologous region across the entire sample. If bieuryonic breadth was recorded, the greatest width for humans would be located on the parietals, but for many other catarrhines, the widest point of the skull (ex. bizygomatic breadth) would be the lateral development of the temporal bones associated with nuchal muscle attachments. Thus, biauriculare breadth served as the measurement depicting the greatest cranial width across all genera by measuring from homologous features (Buikstra & Ubelaker, 1994 - no. 9, p. 75). 6. Bizygomatic breadth (bizygo) was determined using spreading calipers with the basicranium placed on a cranial rest and the viscerocranium facing the observer. Bizygomatic breadth was chosen because it represents one of the widest points of the skull perpendicular to the median plane and is a homologous and reproducible dimension across taxa (Wood, 1991 - no. 52, p. 289). 7. Bifrontomalarorbitale breadth (bifmo) was measured using dial calipers by placing the upper arms at the junction of the fronto-zygomatic suture on the lateral borders of the right and left eye orbits, with the basicranium placed on a cranial rest and the viscerocranium facing the observer. Bifrontomalarorbitale breadth was included because this dimension is correlated with body weight in hominins (Aiello & Wood, 1994) and describes the upper facial breadth excluding the lateral development of the zygomatic and frontal bones. Furthermore, bifmo is very similar or nearly approximate to biorbital breadth (bi-ektoconchion, ek-ek; Wood, 1991 - no. 50, p. 289) and is reported for many fossil specimens. 8. Superior facial breadth (bifrontomalartemporale breadth; bifmt) was determined using dial calipers by placing the lower arms at the junction of the frontal and zygomatic bones on the most superolateral extent of the fronto-zygomatic suture, with the basicranium placed on a cranial rest and the viscerocranium facing the observer. Bifrontomalartemporale breadth was chosen because this dimension represents the greatest breadth of the upper face, including the lateral development of the zygomatic and frontal bones (Wood, 1991 - no. 49, p. 289). 9. Bizygomaxillare superior breadth (bizs) was measured using the lower arms of dial calipers and placing them upon the superior junction of the maxilla and zygomatic bones on the inferior borders of the left and right orbits, with the basicranium placed
28
on a cranial rest and the viscerocranium facing the observer. Superior bizygomaxillare breadth was included because this measurement describes the greatest superior breadth of the maxilla and its contribution to the lower borders of the eye orbits (Ackermann, 2002, p.171). 10. Bizygomaxillare inferior breadth (bizi) was measured using the lower arms of dial calipers and placing them upon the most inferolateral right and left junctions of the zygomatico-maxillary suture with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. Inferior bizygomaxillare breadth was included because this dimension depicts the greatest inferior breadth of the maxilla and this region is associated with vital mastication muscles and their attachment (Ackermann, 2002, p.171). Furthermore, this measurement is similar or near approximate to bimaxillary breadth (zygomaxillare, zm-zm; Wood, 1991 - no. 53, p. 289), which is also reported for some fossil specimens. 11. Zygomatico-maxillary suture length (superior zygomaxillare to inferior zygomaxillare; zs-zi) was measured using the lower arms of dial calipers, from zygomaxillare superior to zygomaxillare inferior. Position or hold cranium as necessary. This measurement was chosen to document the chord length of the zygomatico-maxillary suture (Ackermann, 2002, p.171). 12. Maximum length of the zygomatic bone (zs-zgyi) was measured with the lower arms of dial calipers from superior zygomaxillare to the posteroinferior point of the zygo-temporal suture along the zygomatic arch. Position or hold cranium as necessary. This measurement was included because the inferior border of the zygomatic bone is an important region associated with masticatory muscles, particularly the masseter muscles (Ackermann, 2002, p.171). 13. Superior facial length (basion to prosthion; bas-pros) was measured with spreading calipers with the skull resting on its left lateral side on a cranial rest and the basicranium facing the observer. The distance between basion and prosthion was chosen because this dimension describes the size and projection of the face (Wood, 1991 - no. 44, p. 289). 14. Inferior cranial length (basion to nasion; bas-nas) was measured with spreading calipers; if nasion was a complicated area with ossicles, the most superiomedial
29
point of the discernable nasal bones and their junction with the frontal bone was nasion. Position or hold cranium as necessary. The length between basion and nasion was included because this measurement represents the area of juncture between the viscero- and neurocranium (Wood, 1991 - no. 5, p. 287). 15. Cranial height (basion to bregma; bas-br) was determined with spreading calipers; if bregma was a complicated juncture between the frontal and parietals, the most posteromedial part of the frontal was taken to be bregma, including the sagittal crest if present (Whitehead et al, 2005). Position or hold cranium as necessary. Cranial height was chosen because this dimension represents the cranium’s greatest height in the median plane (Wood, 1991 - no. 4, p. 287). 16. Superior facial height (nasion to prosthion; nas-pros) was measured using the lower and/or upper (depending on size of specimen) arms of dial calipers depending on primate genera. Position or hold cranium as necessary. The chord length between nasion and prosthion was included because this measurement depicts the greatest length of the viscerocranium in the median plane (Wood, 1991 - no. 43, p. 289). 17. Frontal sagittal chord (nasion to bregma; nas-br) was measured using the lower arms of dial calipers, including the sagittal crest if present. Position or hold cranium as necessary. The frontal, parietal and occipital sagittal chord lengths were included to depict the dimensions of the neurocranium in the median plane, representing homologous dimensions across all primates. 18. Parietal sagittal chord (bregma to lambda; br-lam) was measured using the lower arms of dial calipers, although if lambda is a complicated area with ossicles the most superiomedial point of the occipital where it meets the parietals was considered lambda. Position or hold cranium as necessary (Wood, 1991 - no. 25, p. 288). 19. Occipital sagittal chord (lambda to opisthion; lam-opn) was measured using the lower arms of dial calipers. Position or hold cranium as necessary (Wood, 1991 - no. 39, p. 288). 20. Sagittal length of the nasal bones (nasion to rhinion; nas-rhi) was measured using the lower arms of dial calipers from nasion to the rhinion. Please note, in some genera, particularly Rhinopithecus (also observed in Pygathrix and Pongo), the
30
nasal bones may or may not be present (Wang et al, 1998). For some specimens, nasion may be a complicated area with ossicles. If so, the most mediosuperior point of the discernable nasal bones and their juncture with the frontal bone was considered nasion. Position or hold cranium as necessary. The sagittal length of the nasal bones and nasal aperture (no. 22, below) was included because the nasal region and its configuration are highly diagnostic for many primate genera (Verheyen, 1962; Wood, 1991 - no. 71, p. 290). 21. Inferior breadth of the nasal bones (inbrnabo) was measured with the lower arms of dial calipers placing each arm on the most right and left lateral points of the nasal bones where they meet the nasal aperture. Place the basicranium on a cranial rest with the viscerocranium facing the observer. The inferior breadth of the nasal bones was included to examine the nasal bones contribution to the superior border of the nasal aperture (Wood, 1991 - no. 74, p. 290). 22. Sagittal length of the nasal aperture (rhinion to nasospinale; rhi-ns) was measured using the upper arms of dial calipers. Position or hold cranium as necessary (Wood, 1991 - no. 70, p. 290). 23. Maximum width of the nasal aperture (maxnawi) was determined using the upper arms of dial calipers. Place the basicranium on a cranial rest with the viscerocranium facing the observer (Wood, 1991 - no. 68, p. 290). Naso-facial measurements are important because they are highly diagnostic between species (e.g. Verheyen, 1962) and can be influenced by climate (e.g. Harvati & Weaver, 2006). 24. Nasal height (nasion to nasospinale; nas-ns) was measured using the upper arms of dial calipers. Position or hold cranium as necessary. The distance between nasion and nasospinale was included because this dimension describes the length of the nasal region only and not the further projection of the premaxilla (Wood, 1991 - no. 69, p. 290). 25. Maxillo-alveolar length (prosthion to distal M3; maxalvlen) was determined with spreading calipers from prosthion to the central midpoint of an imaginary line connecting the right and left points where the maxillo-alveolus meets distal M3, with the skull resting on its left lateral side on a cranial rest and the basicranium
31
facing the observer. Please note, Wood (1991; no. 87, p. 292) measures to the maxillary tuberosities, however, measuring to distal M3 proved to be a more definable and reliable point between species and many specimens’ maxillary tuberosities were damaged or perhaps very smooth and difficult to demarcate posteriorly. The maxillo-alveolar length was included because this dimension depicts the area necessary for the entire dental arcade in the median plane. Furthermore, the differences between these landmarks (distal M3 and maxillary tuberosity not terribly great. 26. Maxillo-alveolar breadth (or, biectomolare breadth or outer alveolar breadth; biecm) was measured using dial calipers by placing the lower arms on the maxillary alveoli just superior to the approximate center of M2, with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. Biectomolare breadth is included because this measurement represents the maximum width of the maxilla including the contribution of the postcanine dentition perpendicular to the median plane (Wood, 1991 - no. 88, p. 292). 27. Incisivion to the palatomaxillary suture (iv-pms) - Length between the most posteriomedial point of foramen incisive (or incisive canal) to the central intersection of the palatomaxillary suture was measured using the upper arms of dial calipers, with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. The chord length between incisivion and the central intersection of the palatomaxillary suture was chosen because this dimension describes the length of the maxilla only and not the contribution of the premaxilla and palatine bones in the median plane (Wood, 1991 - no. 92, p. 292). 28. I1- I2 alveolar length (prosthion to the septum between I2 & C1; i1i2) was measured using the lower arms of the dial calipers from prosthion to the septum between I2 and C1. Please note, if large diastema exists between I2 and C, measure from prosthion (pros) to the maxillo-premaxillary suture in line with the dental arcade). I1- I2 alveolar length was included because this dimension records the area needed to house the medial and lateral incisors excluding the canines. Position or hold cranium as necessary (Wood, 1991 - no. 94, p. 292).
32
29. Internal breadth between the upper canines (bicanin) was measured using the upper arms of dial calipers with the tips of the arms on the palatal alveolus surrounding the canines perpendicular to the median plane with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. Internal breadth between the upper canines was included to record the breadth of the anterior portion of the palate. 30. External breadth of the upper canines or intercanine distance (Wood, 1991 - no. 98, p. 293) (bicanex) was measured using the lower arms of dial calipers with the tips of the arms on the maxillary alveolus surrounding the canines perpendicular to the median plane with the basicranium placed on a cranial rest and the viscerocranium facing the observer. External breadth between the upper canines was included to record the maximum breadth of the anterior portion of the maxilla including the area necessary for the canines. 31. Palatal height at M2 (palhei) was determined using coordinate calipers by placing the long arms near the left and right ectomolare, allowing the measuring arm to be centered on the palatal roof with the neurocranium placed on a cranial rest and the posterior basicranium and occiput facing the observer (please note, Wood, 1991 measures palatal height at M1, no. 103, p. 293). 32. Biseptal breadth (biseptal) was measured using the lower arms of dial calipers. Please note, if large diastema exists between I2 and C, measure from prosthion to the maxillo-premaxillary suture centered in line with the dental arcade. Position or hold cranium as necessary. 33. Palatal breadth (biendomolare breadth; bienm) was measured using the lower arms of dial calipers not on the M2 with the tips of the arms on the maxillary alveolus surrounding M2 perpendicular to the median plane with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. Biendomolare breadth was included as this measurement describes the maximum breadth of the palate excluding the contributing dimensions of the postcanine dentition (Wood, 1991- no. 91, p. 292). 34. Palatal length A was measured using the upper arms of the dial calipers from orale to staphylion (ol-sta) with the skull resting on its left lateral side on a cranial rest
33
and the basicranium and palate facing the observer. Palatal length A was chosen because this dimension is the maximum length of the entire palate in the median plane (Wood, 1991 - no. 89, p. 292). 35. Palatal length B was measured using the upper arms of the dial calipers from orale to the central intersection of palato-maxillary suture (ol-pms) with the skull resting on its left lateral side on a cranial rest and the basicranium and palate facing the observer. Palatal length B was recorded because this distance provides the length of the palate minus the contribution of the palatines in the median plane (Wood, 1991 - no. 90, p. 292). 36. Interentoglenoid breadth (bien) was measured using dial calipers by placing the lower arms on the tips of the left and right entoglenoid processes with the neurocranium placed on a cranial rest and the anterior basicranium and palate facing the observer. Interentoglenoid breadth was chosen because it represents a vital dimension of the basicranium and its relationship with the mandible (Wood, 1991 - no. 15, p. 288). Other cranial measurements (orbital height and width, interorbital pillar width, superior breadth of the nasal bones and bi-mastoid breadth) were initially collected but were later excluded from the analyses because 1) there was a question of their accuracy and repeatability, 2) they were highly variable 3) specimens were damaged or 4) measured incorrectly by the author and were thus omitted.
34
br fmt
g lam nas fmo
zs o rhi aur
ns por zgyi zi pros ast
opn bas ol
ecm
iv pms
zi zi zi
sta enm
dm3 zgyi zgyi
en en bas por por aur aur
opn ast ast
Figure 2.1: Adapted from Lang (1923) with some modification. A. Norma lateralis. B. Norma inferior.
35
From Ackermann, 2002
From Singleton, 2002
Figure 2.2: Arrows indicate abbreviations used within this thesis. Other measurements and abbreviations are from Wood, 1991; Buikstra & Ubelaker, 1994; Bass, 1995; & White, 2000.
36
2.5 The Craniometric Sample: The craniometric sample collected for this study contains representatives from all of the genera of the Catarrhini according to the taxonomy of Groves (2001a). The goal of this data collection was to document the comparative cranial diversity that can exist among populations of genera and species. The craniometric sample captures, morphometrically, broad macroevolutionary trends. This is validated by the results of discriminant function analysis of raw data. Whether entire multivariate dataset or partitioned groups (superfamily, family, subfamily, tribe and subtribe) more than 90% of individuals are correctly classified to their respective genus, cross-validated using SPSS 11.0. Thus, it fair to say that there must be some biological reality to the morphometric content of this dataset and genus level classification. Nevertheless, it must be admitted, the craniometric samples cannot possibly represent the full range of variation because some generic samples do not contain all species representatives. In addition, it was sometimes necessary to include a subadult specimen because no other specimens were available. Please note, the limited subadults (i.e. individuals with their third molar not fully erupted to the occlusal plane) included less than twenty individuals and had their third molar at least in the process of erupting. Although, if a particular measurement for a subadult fell well outside the range of adults, the dimension was excluded and the mean of the sample was substituted; this was only necessary for very few individuals. However, where deficiencies exist other published accounts from the primatological and anthropological literature can supplement results and conclusions. Furthermore, when pooled sex samples are less than ten, as is the case with some genera (Allenopithecus, Procolobus, Semnopithecus and Simias), results related to these genera should be considered approximate but are included nonetheless for heuristic and comparative purposes. In addition, genera with small sample sizes (<10) or subadults were both included and excluded from the multivariate statistic procedures to measure their impact. This revealed whether or not these genera skewed the results, which they did not, and so are included and reported.
37
2.5.1 Primates, Haplorrhini, Simiiformes, Catarrhini, Cercopithecoidea, Cercopithecidae, Cercopithecinae, Cercopithecini (n=89): The tribe’s craniometric sample is composed of 89 adults with subequal numbers of male and female adults, with a few subadults. While the samples for Erythrocebus, Allenopithecus and Miopithecus include all recognized species (although the first and second are monotypic genera) the same cannot be said for Cercopithecus and Chlorocebus. The sample for Cercopithecus includes only nine species although Groves (2001a) recognizes a potential 25. Thus, the sample for Cercopithecus is not exhaustive and cannot possibly represent the full range of variation within this genus. The same must be admitted for Chlorocebus, which is comprised mainly of vervets from South Africa even though Groves (2001a) has distinguished six species. However, this allows the comparison of samples from geographical extremes of supposed closely related species. That is, the specimens for Cercopithecus include species from tropical equatorial Africa, while specimens for Chlorocebus are from South Africa. Other previous craniometric studies of the guenons and their allies include the work of Verheyen (1962), Hill (1966), Vogel (1966 & 1968), Alexander (1981), Martin & MacLarnon (1988), Shea (1992) Jungers et al (1995) Ravosa & Profant (2000) and Cardini et al (2007). Table 2.2: Craniometric sample for the Cercopithecini. Allenopithecus Craniometric sample includes four males and two females, three of which were subadults nigroviridis (n=6). Miopithecus spp. Craniometric sample includes both species, M. talapoin ( -8, -7) and M. ogouensis ( - 1, -3) (n=19, -9, -10). Cercopithecus spp. Craniometric sample includes nine species, C. ascanius ( -1, -2), C. erythrogaster ( - 1, -1), C. mitis ( -2, -2), C. neglectus ( -2, -2), C. diana ( -1, -1), C. petaurista ( -2), C. cephus ( -1), C. nictitans ( -2, -2) and C. wolfi ( -2) (n=24, -12, -12). Chlorocebus spp. Craniometric sample includes specimens mainly C. p. pygerythrus from Southern Africa although two male C. p. hilgerti were also included (n=24, -13, -11). Erythrocebus patas Craniometric sample includes nine males and seven females (n=16).
38
2.5.2 Primates, Haplorrhini, Simiiformes, Catarrhini, Cercopithecoidea, Cercopithecidae, Cercopithecinae, Papionini (n=143): The tribe’s craniometric sample is represented by 143 adults with a few subadults. While the samples for Mandrillus and Theropithecus document all species, the same cannot be said for Cercocebus, Macaca, Lophocebus or Papio. The sample for Cercocebus contains representatives from all species except C. sanjei. The sample for Macaca includes only nine species although Groves (2001a) recognizes a possible twenty. The sample for Lophocebus includes mainly L. albigena, with only three L. aterrimus and no specimens of L. opdenboschi. The sample for Papio comprises mainly specimens of P. ursinus although there are some representatives of P. anubis, P. papio and P. cynocephalus but P. hamadryas is not represented. Other craniometric studies of the papionins include the work of include Hill (1970 & 1974), Albrecht (1978), Groves (1978), Alexander (1982), Cheverud (1989), Fooden & Albrecht (1993); Martin (1993), Setchell et al (2001); Singleton (2002 & 2004a), Pan et al (2003), Frost et al (2003), Pan & Oxnard (2004), Hamada et al (2005) and Elton & Morgan (2006).
Table 2.3: Craniometric sample for the Papionini. Cercocebus spp. Craniometric sample includes four species - C. atys ( -1, -2), C. torquatus ( -4, -5), C. agilis ( -5, -1) and C. galeritus ( -1) (n=19, -10, -9). Mandrillus spp. Craniometric sample includes both species, M. sphinx ( -3, -7) and M. leucophaeus ( -5, -7) (n=22, -8, -14). Macaca spp. Craniometric sample includes ten species, M. sinica ( -1), M. nigra ( -2, -4), M. arctoides ( -4, -3), M. nemestrina ( -3, -1), M. ochreata ( -1, -1), M. assamensis ( -1, -1), M. fascicularis ( -2, -3), M. hecki ( -2), M. mulatta ( -1, -1), M. silenus ( -1), and M. thibetana ( -2) (n=34, -15, -19). Lophocebus spp. Craniometric sample includes specimens mainly of L. albigena ( -9, -11) but a few L. aterrimus ( -2, -1) are included (n=23, -11, -12). Papio spp. Craniometric sample for Papio is mainly comprised of P. ursinus ( -12, -7) from South Africa, although some P. anubis ( -4, -4), P. cynocephalus ( -2, -1) and P. papio ( -1) were collected (n=31, -19, -12). Theropithecus gelada Craniometric sample includes seven males and seven females (n=14).
39
2.5.3 Primates, Haplorrhini, Simiiformes, Catarrhini, Cercopithecoidea, Cercopithecidae, Colobinae (n=196): Other craniometric studies of the African colobines include the work of Schultz (1958), Verheyen (1962), Hull (1979) and O’Higgins & Pan (2004). Further craniometric studies of the Asian langurs include the work of Hooijer (1962), Aimi & Bakar (1992), and Pan & Groves (2004). Lastly, craniometric studies of the Asian odd-nosed colobines include the work of Groves (1970) and Pan & Oxnard (2001).
Table 2.4: Craniometric sample for the Colobinae. Colobini, Colobina Craniometric sample includes three species - C. angolensis ( -2, -2), (n=59), C. polykomos ( -1, -1), and C. guereza ( -12, -8) Colobus spp. (n=26, -15, -11). Piliocolobus spp. Craniometric sample includes four species - P. badius ( -7, -6), P. rufomitratus ( -5, -1), P. tephrosceles ( -3, -1) and P. tholloni ( -3, -2) (n=28, -15, -13). Procolobus verus Craniometric sample includes three males and two females (n=5). Presbytini, Presbytina Craniometric sample includes three species - S. entellus ( -2, -1), (n=60), Semnopithecus S. schistaceus ( -1, -1) and S. priam ( -2, -2) (n=9, -5, -4). spp. Trachypithecus spp Craniometric sample includes six species - T. obscurus ( -2, -2), Tr. pileatus ( -1, - 3), T. phayerei ( -2, -2), T. cristatus ( -6, -7), T. (Kasi) johnii ( -1, -1), and T. (Kasi) vetulus ( -3, -3) (n=33, -15, -18). Presbytis spp. Craniometric sample includes five species - P. rubicunda ( -6, -6), P. potenziani ( - 2), P. melalophos ( -1), P. natunae ( -2), and P. thomasi ( -1) (n=18, -8, -10). Presbytini, Nasalina Craniometric sample includes both species, P. nemaeus ( -6, -5) (n=77), and P. nigripes ( -8, -6) (n=25, -14, -11). Pygathrix spp. Rhinopithecus spp. Craniometric sample includes two species - R. roxellana ( -3, -10) and R. avunculus ( -2, -2) (n=17, -5, -12). Nasalis larvatus Craniometric sample includes eighteen males and ten females (n=28). Simias concolor Craniometric sample includes two males and five females, two of which were subadults (n=7).
40
2.5.4 Primates, Haplorrhini, Simiiformes, Catarrhini, Hominoidea (n=144): The superfamily is represented by 144 adults with very few subadults. The craniometric samples for the Hominoidea are not exhaustive, nor do they contain the full range of variation within these genera. The sample for Hylobates consists of seven species from four subgenera although Groves (2001a) has distinguished a possible 14. The sample for Pongo includes specimens of only P. pygmaeus (Bornean orangutan) and none for P. abelii (Sumatran orangutan). The sample for Gorilla contains specimens of only western lowland gorillas (G. gorilla) and the sample for Pan contains mainly specimens of Pan troglodytes although one female Pan paniscus was included. The sample for modern Homo contains adults from North America, South Africa and East Asia. Other craniometric studies of the hominoids include the work of Krogman (1969), Creel & Preuschoft (1971 & 1976), Cramer (1977), Shea (1985), Shea & Coolidge (1988), Groves et al (1992), Uchida (1996), Ackermann (2002), Albrecht et al (2003), Guy et al (2003), Miller et al (2004), Preuschoft (2004) and Brown et al (2005).
Table 2.5: Craniometric sample for the Hominoidea. Hylobatidae, Craniometric sample includes seven species from all four subgenera – Hylobates spp. H.(Hy.) lar ( -4, -2), H. (Hy.) agilis ( -1) and H. (Hy.) muelleri ( -1, -4); H.(S.) syndactylus ( -8, -5); H.(Ho.) ( -1, -1); H.(N.) concolor ( -1), H.(N.) gabriellae ( -1) and H.(N.) leucogenys ( -1, -1) (n=31, -16, -15). Hominidae, Craniometric sample includes only Pongo pygmaeus, fourteen males and twelve females Ponginae, (n=26). Pongo pygmaeus Hominidae, Craniometric sample includes only Gorilla gorilla, seventeen males and seventeen females Homininae, Gorilla (n=33). gorilla Pan spp. Craniometric sample includes specimens consists mostly of Pan troglodytes and one female Pan paniscus (n=33, -19, -14). Homo sapiens Craniometric sample includes eleven males and nine females from modern populations (N. America, South Africa and East Asia) (n=20).
41
2.6 Statistical Methods: 2.6.1 Descriptive, Univariate and Bivariate Statistics: The most commonly employed methods for making statistical inferences in biology are the parametric tests for mean differences, such as Student’s t-test and Analysis of Variance. These tests make several assumptions about the data under study. Important among them is the assumption that they follow a normal (Gaussian) distribution, and that variances and sample sizes are similar. Failure to meet these assumptions results in spurious results under a parametric model. They additionally assume that the data under study represent random samples and are independent. The first of these assumptions is difficult to test directly in paleontology. Fossil samples represent individuals whose remains have survived complex and unknown taphonomic histories. Thus, they might be biased in some unknown way, such as towards larger-bodied individuals such as males, or sample individuals or taxa from a narrow segment of their geographic or temporal distribution. There is simply no way of controlling for these factors, and in the absence of a complete record for hominins, fossils must and usually do get treated as random samples. The assumption of independence essentially requires that the value of one datum is unrelated to any other datum (i.e. knowing the value of one observation gives you no information about the value of any other). This is reasonable and uncontroversial here. Univariate data and descriptive statistics (mean, median, maximum (Max.), minimum (Min.), range, standard deviation (SD) and coefficient of variation (CV) of modern skeletal elements of known provenience provide many basic functions in comparative analysis and particularly in relation to biological classification of fossil material (Wood, 1991; Plavcan, 1993; Plavcan & Cope, 2002). In addition, examination of descriptive and univariate statistics identifies any errors in the craniometric datasets. Males and females were pooled together so as to capture and experience the full range of morphological variation and affects of sexual dimorphism within and between genera. Box plots are presented because they allow for quick assessment as to a sample’s spread and possible skewness, as well as identifying significant differences where no overlap of the
42
boxes occurs. Genera are arranged by mean generic body weight (grams - g) from least to greatest (left to right). Cranial variables and indices which distinguish at least two or more genera via examination of box-plots (i.e. where no overlap of boxes occur) were then subject to three different analyses.
2.6.2 Shapiro-Wilk Test: First, each generic sample was tested for normality by Shapiro-Wilk. If normality could not be assumed, that particular cranial variable was later excluded from one-way Analysis of Variation (ANOVA). The Shapiro-Wilk Test is a simple one-sample method for testing the assumption that a sample follows a normal distribution. In this study, Shapiro-Wilk tests were performed using the program PAST. The level of significance was set at p=0.05. Probability values shown indicate the probability that a sample is normally distributed (e.g. p=0.77 provides a probability of 77% that the sample is normally distributed).
2.6.3 Kruskal-Wallis Test: Second, to test statistical significance, all variables were subject to Kruskal-Wallis, which is a generally considered non-parametric ANOVA (Hammer et al, 2006). That is, it allows for comparison of multiple, randomly drawn, independent samples, and makes no assumptions about the data (i.e. normality and similar variances and sample sizes). It differs from ANOVA in providing a test of the equality of medians (instead of means). The Kruskal-Wallis procedure defines a ratio symbolized by H: numerator is the observed value of the sum-of-squares between group deviates (SSbg) based on ranks and the denominator is the mean of the sampling distribution of SSbg. As with ANOVA, the Kruskal-Wallis test does not indicate which of the samples compared show significantly (median) differences. Again, a post hoc test making pair-wise comparisons is required. Commonly, Mann-Whitney U tests are used for this purpose. Moreover, the Mann-Whitney test is usually altered with Bonferroni correction of p in order to maintain the overall probability of a Type I error at p=0.05. Here, Kruskal-Wallis tests were performed using PAST. The level of significance was set at p=0.05. The results of Bonferroni-corrected Mann-Whitney U-tests are also
43
given in PAST to allow assessment of pair-wise comparisons. Matrices of these the results are presented below.
2.6.4 One-way ANOVA Lastly, Where samples were found to follow a normal distribution using the Shapiro-Wilk test, the parametric inferential method Analysis of Variance (ANOVA) was employed to test the null hypothesis that sample means are identical. While the Student’s t- test is restricted to paired (two-sample) comparisons, ANOVA allows for multi-group comparisons to be undertaken (i.e. the means of more than two independent samples can be compared in a single test). When large numbers of paired tests of mean difference are undertaken it is a statistical inevitability that about 5% of such tests will be wrong (i.e. result in Type-I errors: rejection of a null hypothesis that is actually correct). For example, if 100 t-tests are undertaken then about 5 of them are expected to represent Type-I errors when the significance level is set at p=0.05. ANOVA allows this ambiguity to be overcome when more than one comparison is made. ANOVA essentially allows us to ask the question: Is there an overall indication of mean differences among the various samples? ANOVA is based on the comparison of variability (sums-of-squares) between groups (SSbg) and within groups (SSwg), the difference between these variance estimates provided by Fisher’s F-distributions. If the variance between two groups is large, the difference between means will be large, and the F-ratio will also be large. A major limitation of ANOVA is that is fails to indicate which of the samples compared are significantly different. To determine this, an additional (post hoc) test making pair-wise comparisons is required. Commonly, the Tukey’s HSD (Honest Significant Difference) is employed for such purposes as it is generally considered more robust than a t-test; being more stringent, and decreasing the probability of obtaining significant results by chance alone. Here, one-way ANOVA tests were performed using PAST. The level of significance was set at p=0.05. PAST also tests for equality of variance in all ANOVA’s, and only where this assumption was violated will these results be discussed in the test. Tukey’s HSD pair-wise comparisons are also automatically undertaken in PAST and the results (p-values) are presented in matrices in the results. Cranial variables which have a
44
normal distribution were subject one-way ANOVA. However, because of their small sample sizes, Allenopithecus, Procolobus, Semnopithecus and Simias were excluded from all ANOVA analyses but were examined via Shapiro-Wilk and Kruskal-Wallis. To examine the relationship between cranial size and body weight, the pooled sex mean for particular cranial dimensions were plotted against mean generic body weights collected from the published literature. Bivariate plots of body weight (x-variable) to cranial variables (y-variable) are a useful tool for examining large-scale macroevolutionary trends, including allometric trajectories and niche separation (Aiello & Day, 1982; Benefit & McCrossin, 1993; Aiello & Wood, 1994; Jungers et al, 1995).
2.7 Multivariate Statistics: 2.7.1 Principal Component Analysis (PCA): (cite Relethford, 1994 & 2002) Principal Components Analysis (PCA) is a robust ordination multivariate statistical procedure which can reduce a large number of variables to a few very important ‘principal components’ (PC) based on the variance-covariance (or correlation) matrix of the multivariate dataset. However, despite data reduction, very little information is lost. PCA has been applied successfully to numerous biological and anthropological studies for both morphometric and ecological data with significant results (Andrews & Williams, 1973; Lessertisseur et al, 1974; Albrecht, 1978; Howells, 1989; Strasser, 1992; Shea, 1992; Shea et al, 1993; Fleagle & Reed, 1996; Pan et al, 2003; Harvati, 2003; Frost et al, 2003; O’Higgins & Pan, 2004; Pan & Groves, 2004; Miller et al, 2004; Hamada et al, 2005). PCA was employed within this study because this technique has several features which are amenable to cranial morphometrics and their interpretation as well as identifying similar cases that may cluster together because of functional similarities or phylogenetic relationships. For example, linear measurements of cranial dimensions are highly interrelated which can produce spurious or false readings through redundancy. Complex interrelationships such as these are generally referred to as morphological integration or modularity (Olson & Miller, 1958; Cheverud, 1982; Ackermann & Cheverud, 2004b; Goswanji, 2006; Polanski & Franciscus, 2006). Fortunately, the computation for PCA uncorrelates each variable by creating linear combinations (or functions) of the original variables. These linear combinations maximize the inherent variation of the multivariate
45
dataset. In addition, PC’s are independent from each other by way of plotting axes perpendicular (or orthogonal) to one another. In theory, these PCs describe the underlying structure of the multivariate dataset. What’s more, these patterns may represent both phylogeny and adaptation. However, as with most statistical techniques there are assumptions and limitations to the methods. Multivariate statistics rely on conditions such as large representative samples, multivariate normality, and for some methods, the same units of measurements. Since this thesis is concerned with cranial variables which expose maximum generic differences or separation, only PCs 1 & 2 and PCs 2 & 3 will be discussed for Ln transformed data, whereas results for MSV will be limited to PCs 1 & 2 because size has been controlled for and PCs 2 & 3 rarely explain a large part of the variation between genera.
2.7.2 Canonical Variates Analysis (CVA): Canonical variates analysis (CVA) is another rigorous multivariate statistical technique which has some similarities to PCA, as well as sharing theoretical assumptions, although there are significant differences. As Albrecht (1978) has succinctly expressed, “The critical difference in the two techniques is that canonical analysis relies on the belief that it is of value to weight differences between the groups using the within-group dispersion as a standard of comparison” (p.37), which PCA does not (see also Albrecht, 1980). Within biological anthropology, CVA has been effectively applied to many primate genera and species and has produced statistically significant results (Albrecht & Miller, 1993). The computation of PCA and CVA produce similar results and graphical representation; and is used here in conjunction with PCA. In addition, both methods can be used to gauge the results of each other. Like PCA, CVA maximizes the variation between groups. Recently, Harvati (2003) has combined results of PCA and CVA demonstrating significant and interesting differences between Neandertals and Modern humans in temporal bone morphology. CVA has been successively applied to many primate species thus far studied morphometrically (cranial and postcranial), which have produced statistically significant results (Albrecht & Miller, 1993). CVA (is essentially multi-group discriminant function analysis; Albrecht, 1980; Hammer et al, 2006), like PCA, creates new, uncorrelated variables which are linear combinations of the original variables. Similar
46
to results for PCA, only CV axes 1 & 2 will be examined and discussed because these two axes explain the majority of the variation between genera while subsequent axes explain much less.
2.8 Morphological and Genetic Distances: To summarize the total morphological variation within a genus’ multivariate dataset, Euclidean distances between all individuals were recorded and the averages, maximums and minimums for each genus were presented. This summarizes the extent to which members of within a genus differ from other congenerics. Then, to measure morphological divergence between genera, again, Euclidean distances based on generic mean PCA and CVA object scores (for both Ln transformed and MSV data) from the first five PCs or CV axes were recorded generated to create inter-generic distance matrices. Generating Euclidean distances in this manner is very similar to Mahalanobis distances (Mimmack et al, 2001; Elmore & Richman, 2001). This allows for quick assessment as to how much one genus has diverged from other closely allied genera.
Morphological Distance – M: In addition, inter-generic M distances were computed for all groups studied. The “M” distance has been applied to several orders of animals with robust results (Cherry et al, 1978 & 1982; Atchley et al, 1980; Wilson et al, 1984). This computation provides evidence of shape divergence not size because each measured trait is converted into a proportion of the geometric mean, which is very similar to Mosimann shape variables. In addition, this statistic also utilizes the standard weighted deviation which corrects for small sample sizes. As Cherry et al (1978) describe, “M [morphological difference (M)] is the average number of standard deviations by which the two species [or in this analysis, genus] differ per trait measured” (p. 210).
Genetic Distances: Like morphological distances, genetic distances also summarize numerous variables and reduce data to one figure. However, unlike morphological distances, genetic distances increase with time and the same cannot be assumed for morphological features. Genetic
47
distances were compiled from the published literature (Tosi et al, 2003 & 2004; Page et al, 1999; Page & Goodman, 2001; Whittaker et al, 2006).
2.9 Data Transformation: Despite all measurements being recorded using the same unit of measurement, millimeters (mm), most statistical procedures advise or require data transformation prior to analysis. This is necessary for a number of reasons. Obviously the largest source of difference and variation between genera and species is absolute size (Jungers et al, 1995). In fact, larger species have larger ranges of variation (Albrecht, 1978). Accordingly, steps must be taken to control and account for the influence of size but it cannot be eliminated completely. Another consideration is the possibility of skewness in the data due to perhaps small sample size, sizes differences between males and females due to sexual dimorphism, or whether the dataset has multivariate normal distribution, as well as, similar variance- covariance patterns; the latter of which cannot be assumed for even closely related species (Ackermann, 2003). To adjust and comply with these requirements, prior to multivariate analysis all raw data was first examined for errors and then natural log (Ln) transformed. By doing this all measurements are distributed linearly with some correction by removing skewness as well as providing log normal distribution. In addition, the raw data was also converted to Mosimann Shape Variables (MSV); a method which in theory controls for size within the multivariate dataset (Nadal-Roberts & Collard, 2005). This is accomplished by computing the geometric mean of all measurements for one individual followed by dividing each measurement by the geometric for that individual (see Mosimann, 1970; Mosimann & James, 1979; Darroch & Mosimann, 1985; Jungers et al, 1995). Thus, two sets of multivariate analyses (PCA and CVA) were conducted on 1) A Ln transformed multivariate dataset; and 2) A size corrected multivariate dataset (Mosimann Shape Variables; MSV).
2.10 Structure of Results: Results reported in succeeding chapters are grouped at various taxonomic levels (tribe - Cercopithecini and Papionini; subfamily - Colobinae; and superfamily - Hominoidea) for comparative purposes. For example, the genera which are grouped within the tribe Cercopithecini are only compared with other genera of that tribe and so forth.
48
These groupings represent the comparative boundaries (defined by phylogeny, morphology, genetics, ecology and behavioral data) for the genera under study. By applying this criterion, group comparisons are naturally composed by a biological yardstick approach. Comparisons outside of these groupings are unnecessary because there are no generic taxonomic disagreements between these groupings. In other words, there are no taxonomic arguments for the inclusion of the Colobus within Papio or Hylobates and vice versa. Instead, differing generic taxonomic schemes are within these groupings. For example, within the Papionini tribe, Goodman et al (1998) suggests (based on genetic data) Mandrillus is a subgenus within Cercocebus. Another example is the gibbons which can be arranged into one genus with four subgenera or purely four genera. Still, another example includes the African colobines; most recognize Colobus as a distinct genus but disagreements remain concerning whether or not to recognize separate genera for Procolobus and Piliocolobus, or each a subgenus within the former due to shared derived characters (Oates et al, 1994). In fact, this last example is also true for the genera of Presbytina and Nasalina (Bennett & Davies, 1994). Due to the large amount of generic comparative data, results must follow similar and repetitive formats for clarity in the synthesis of evidence. First, each genus is briefly introduced noting some characteristic features, including body weight, habitat choice, diet, etc. Second, the taxonomic nomenclature history of the genera under consideration is provided, including differences of opinion and alternative classification schemes. Third, descriptive, box-plots and univariate statistics are presented and discussed for the cranial variables. Then, similar treatment is employed for the cranial indices. Third, a selection of cranial variables and associated sample mean, which significantly sorted genera, was then plotted against mean generic body weights (grams). Fourth, the results of multivariate statistics (PCA and CVA) of all cranial variables are offered and discussed. Fifth, important postcranial, behavioral, locomotion and ecological features are noted and discussed. These other types of evidence are necessary because genera are defined by many variables not including the cranium and many interact directly with a species’ habitat and locomotion. However, a complete review is beyond the scope of this thesis but features which differentiate genera and their adaptive zones are. Thus, data is limited mainly to skeletal elements which may be encountered in the fossil record. Sixth, results from recently
49
published genetic analyses are discussed and summarized. This is necessary to satisfy Wood & Collard’s (1999a) requirement of genus’ members to be more cladistically related to each than to any other species from another genus (i.e. taking into consideration the inter- and intra-generic genetic relationships). Lastly, each generic synopsis is outlined and evaluated along with morphological descriptions and adaptive features genus; and where necessary, genera which are in need of further refinement are mentioned. After these chapters per grouping are completed the analysis will then consider and evaluate fossil hominin genera based on the available evidence by similar statistical techniques. Through these generic comparisons, this thesis will answer the question - Are the recently proposed hominin genera, such as Sahelanthropus, Orrorin, Ardipithecus and Kenyanthropus, adequately defined and necessary when the extant Catarrhini genera are used as a comparative benchmark?
2.11 Limitations of data and analysis and theoretical assumptions The cranial morphometrics of extant catarrhines are interpreted within the principle of taxonomic uniformitarianism and evolutionary morphology or ecomorphology; both are equally important (Reed, 1997 & 2002; Bromage, 1999). The datasets provide a ‘bird’s eye’ as opposed to a ‘worm’s eye’ perspective (Andrews & Harrison, 2005). The majority of the comparative multivariate craniometric dataset is composed of monkeys from the Cercopithecoidea (n=428) while that of the apes from the Hominoidea is considerably less (n=144). However, this is more a product of macroevolutionary trends rather then deliberate choice. Since the Miocene there has been a dramatic decrease in the number of ape genera and species in comparison to monkeys (Andrews, 1981 & 1992; Fleagle, 1999).
50 Chapter 3: Results for genera of Cercopithecini
Allochrocebus Allenopithecus Miopithecus Cercopithecus Chlorocebus Erythrocebus