Available online at www.sciencedirect.com

Journal of Anthropological Archaeology 26 (2007) 541–575 www.elsevier.com/locate/jaa

Why is there a lack of Mode 3 Levallois technologies in East Asia? A phylogenetic test of the Movius–Schick hypothesis Stephen J. Lycett Leverhulme Centre for Human Evolutionary Studies, University of Cambridge, Fitzwilliam Street, Cambridge CB2 1QH, UK Received 3 January 2007; revision received 25 June 2007 Available online 17 August 2007

Abstract The ‘Movius–Schick hypothesis’ claims that an absence of Levallois (Mode 3) technologies in East Asia is due to the lack of a strong ancestral Acheulean (Mode 2) tradition in that region. Hence, this hypothesis is based on the assumption that similarities between Acheulean handaxes and Levallois cores can be interpreted as being phylogenetically homologous (i.e. due to common technological ancestry) as opposed to being homoplasic (i.e. due to convergent technological evolution). Here, the phylogenetic basis of this hypothesis is tested using a formal cladistic procedure. Under the framework of an ‘iterative’ approach to phylogenetic analysis, a series of post-hoc tests and re-evaluations of results follow the initial cladistic analysis. Results of these combined analyses indicate that morphological similarities between Mode 2 Acheulean handaxes and Mode 3 Levallois cores can, most parsimoniously, be seen as phylogenetically homologous. Hence, these results support the tenets of the ‘Movius–Schick’ hypothesis in suggesting that a lack of Levallois industries in East Asia may be due to the paucity of an ancestral Acheulean tradition in that region. The implications of these phylogenetic analyses for the concept of Palaeolithic ‘Modes’ are discussed. It is further suggested that low demographic levels (i.e. small effective population sizes) in East Asia may have constrained the technological phylogenetic trajectory of East Asia compared with that seen in other regions of the Old World during the Lower Palaeolithic. In addition, it is hoped that several methodological issues discussed here will contribute to the growing field of cultural phylogenetic analysis.  2007 Elsevier Inc. All rights reserved. Keywords: Lithics; Cladistics; Cultural evolution; Phylogenetics; Mode 1; Mode 2; Mode 3; Acheulean; Levallois; Movius Line

Introduction Famously, Movius (1948) noted that strong evidence for an Acheulean tradition is absent in southeast Asia, and that Mode 1 technologies (i.e. polyhedrons, ‘choppers’ and flakes) continue to be made long after distinct Acheulean assemblages appear over much of the Palaeolithic Old World.

E-mail address: [email protected]

Subsequently, the geographic demarcation line between the Mode 1 industries of southeast Asia and the Acheulean (Mode 2) industries of the Indian subcontinent became known as the ‘Movius Line’ (Swartz, 1980). Movius (1969, p. 71) also observed that Levallois (Mode 3) industries were absent in East Asia, and intimated at a possible link regarding the absence of both traditions within the region. This notion has subsequently been extended by Schick (1994, 1998) who has argued that since Levallois technologies are potentially the descen-

0278-4165/$ - see front matter  2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jaa.2007.07.003

542

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

dant of Acheulean traditions, the absence of Mode 3 Levallois traditions might ultimately be explained by the lack of a strong ancestral Mode 2 Acheulean tradition within eastern Asia. Here, the phylogenetic basis of this hypothesis is tested using a formal cladistic procedure of analysis. The Movius–Schick hypothesis Movius’s (1948, 1969) observation regarding the lack of a clear ‘Acheulean’ technological tradition in East Asia is a frequently discussed issue (e.g., Swartz, 1980; Schick, 1994; Clark, 1998; Keates, 2002; Corvinus, 2004; Norton et al., 2006). However, in addition to the paucity of Acheulean (Mode 2) assemblages, Movius (1969, p. 71) also went on to observe that ‘‘the well-known Levallois technique. . . is also completely lacking throughout the Far East as far as southeastern Asia and Northern China are concerned’’ (see also, Schick and Zhuan, 1993; Gao and Norton, 2002). Hinting at a possible link between the absence of these two technologies, Movius (1969, p. 71) commented that normally the Levallois technique ‘‘is found in association with the more specialized hand-axe cultures in other regions of the Old World, and in fact the known distribution of the two is very nearly coincidental’’. More recently, Schick (1994, p. 592) has suggested that the absence of an Acheulean Mode 2 tradition and Mode 3 technologies in southeast Asia ‘‘may serve as a corroboration of important technological differentiation between east and west and also as a possible key to potential reasons behind these differences’’. Schick (1994, p. 593), emphasis in original) also avers that the ‘‘absence of both technologies in eastern Asia is not easily explained on grounds of lack of suitable raw material, preeminent use of non-lithic raw materials for tools, or different functional requirements’’, an assertion supported by more recent work (Brantingham et al., 2000). Elsewhere, Schick (1998, p. 456) is more explicit about the link between these two traditions asserting that: ‘‘procedures of Levallois technology can in many ways be seen as an outgrowth of Acheulean procedures, with an emphasis on strategic flaking to shape the mass or form of the flakes removed, on platform preparation to achieve this end, and overall shaping of the mass of the core, it may be significant that both of these technological patterns are lacking in much of eastern Asia’’.

The view that Mode 2 Acheulean technologies are the direct ancestor of Mode 3 Levallois industries is, of course, in line with the ideas of many others (e.g., Leroi-Gourhan, 1966; Copeland, 1995; Tuffreau, 1995; Tuffreau and Antoine, 1995; Kooyman, 2000, p. 73; deBono and Goren-Inbar, 2001; Petraglia et al., 2003; Tryon et al., 2006). Hence, in drawing upon the earlier observations of Movius (1948, 1969), Schick (1994, 1998) would see the lack of Mode 3 Levallois industries in southeast Asia as the result of an absence of ‘ancestral’ Mode 2 Acheulean traditions. However, this hypothesis— which might more formally be termed the ‘Movius–Schick hypothesis’—rests entirely on the untested assumption that Acheulean bifaces should be regarded as the most appropriate ancestors of Levallois technologies. Moreover, this hypothesis also relies on the implicit assumption that morphological and technological similarities between Acheulean bifaces and Levallois cores are evidence of technological genealogical proximity, a premise that should always be tested rather than assumed (McLennan and Brooks, 2001; O’Brien and Lyman, 2003; Richter, 2005). Some workers have argued that occasional occurrences of ‘handaxe-like’ technologies in East Asia render the Movius Line obsolete (e.g., Yi and Clark, 1983; Hou et al., 2000; Gamble and Marshall, 2001). However, the resemblance of these ‘handaxes’ to those from Acheulean assemblages in terms of morphological features (i.e. degree of bifacial knapping, profile form, thickness) has led many to question the technological comparability of these artefacts with conventional Acheulean bifaces (Schick and Zhuan, 1993; Schick, 1994; Corvinus, 2004; Norton et al., 2006). Moreover, the density of accepted Acheulean sites on the Indian subcontinent (i.e. a region just west of the Movius Line) contrasts starkly with the paucity of purported ‘Acheulean’ sites to the east of the Movius Line (Chauhan, 2004; Petraglia, 2007). As Norton et al. (2006, p. 534) concluded recently, ‘‘if a map of East Asian Paleolithic sites were drawn, the conspicuous lack of biface-bearing sites in East Asia is still prominent’’. Likewise, we are gaining increased knowledge of the age and distribution of Mode 3 Levallois technologies in Africa, the Near East, the Indian subcontinent and Europe (e.g., Rolland, 1995; Misra, 2001; Shea, 2003; Tryon, 2006; Lycett, 2007b; White et al., 2006). In contrast, eastern Asia continues to exhibit a dearth of Levallois technologies (Schick and Zhuan, 1993; Schick, 1994, 1998;

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Gao and Norton, 2002). Hence, the differentiation first noted by Movius (1948) between western regions of the Palaeolithic Old World and portions of East Asia in terms of a lack of Levallois technologies, also appears to have remained valid. As one commentator noted recently, the apparently robust pattern of archaeological differentiation between regions east and west of the Movius Line ensures that ‘‘interesting questions concerning the geographic variations amongst these different stone tool industries remain to be answered’’ (Petraglia, 2007, p. 404). Phylogeny and cultural transmission in archaeology In recent years it has become increasingly recognised that technological traditions (and indeed other cultural behaviours) in humans and some nonhuman primates are learned, inherited and maintained via a process of social transmission (Durham, 1992; O’Brien and Lyman, 2000; McGrew, 2004; Shennan, 2002; Van Schaik et al., 2003; Mesoudi et al., 2004, 2006; Whiten, 2005; Whiten et al., 2005; Perry, 2006). It is also becoming increasingly recognised that wherever there is a process involving transmission, inheritance, variation and the differential representation of transmitted elements in subsequent generations, there will be evolution—or, as Darwin (1859, p. 459) put it, ‘‘descent with modification’’. Where there is evolution, an understanding of the relationship between units (i.e. phylogeny) becomes essential (O’Brien and Lyman, 2003; Mesoudi et al., 2004, 2006; Dunnell, 2006; Lipo et al., 2006). Such theoretical advances have facilitated the use of formal phylogenetic procedures to test hypotheses of technological relationships and processes of cultural evolution (e.g., Collard and Shennan, 2000; O’Brien et al., 2001; Tehrani and Collard, 2002; Jordan and Shennan, 2003; Croes et al., 2005; Collard et al., 2006; Darwent and O’Brien, 2006; Buchanan and Collard, 2007). Phylogenetic studies of material culture operate on the premise that the features (i.e. characteristics or attributes) of artefacts such as stone tools (O’Brien et al., 2001; Darwent and O’Brien, 2006), pottery designs (Collard and Shennan, 2000), basketry styles (Jordan and Shennan, 2003), carpets (Tehrani and Collard, 2002), etc., are useful for reconstructing phylogenetic relationships, in a manner similar to that of using the morphological characteristics of fossil specimens in constructing phylogenies of extinct biological taxa, or of using

543

morphological features to establish phylogenetic relationships in living taxa. This, of course, does not imply that the underlying transmission mechanisms (i.e. genetic transmission versus social transmission) are identical. One immediately obvious difference between the two is that, unlike genes, cultural attributes can be inherited from individuals other than biological parents (for further discussion of cultural versus genetic transmission and inheritance systems see e.g., Shennan, 2002; Mace, 2005; Richerson and Boyd, 2005). However, there is growing evidence to suggest that in both cases the goal of reconstructing the phylogenetic relationships of evolving entities is sufficiently comparable, both ontologically and epistemologically, to justify the use of phylogenetic methods drawn from biology in addressing questions of cultural and technological phylogeny (Mesoudi et al., 2004, 2006; Spencer et al., 2004; Collard et al., 2006; Lipo et al., 2006). The use of knapped stone artefacts by hominins from at least 2.6 Mya offered immense novel opportunities, opening up new subsistence resources and a further means of shaping practical material objects, to say nothing of the resultant potential for colonising new landscapes and occupying new adaptive niches (Ambrose, 2001). Indeed, as Dennell (2004, p. 434) has argued recently, it is for these reasons that hominins were probably technologically dependent upon stone tools from at least the Plio–Pleistocene boundary (c. 1.8 million years BP), indicating that a lineage (i.e. a line of ancestry and descent— sensu O’Brien and Lyman, 2003, p. 121) of lithic technology and associated knapping skills was evolving, via a process of descent with modification, from this time period through to the latter stages of prehistory (see also Shennan and Steele, 1999). The form of stone tools will of course be influenced by several factors, including raw material and function. Many of the current approaches in the field emphasise functional explanations of Palaeolithic technology, yet largely ignore the evolutionary, transmission and descent element of material culture, such that technological phylogenies might be analysed, as Foley and Lahr (2003) and Kuhn (2004) have also recently noted. The logic of testing ancestor–descendant relationships and phylogenetic homology using parsimony Fundamentally, phylogenetic analysis is a means of organising groups of things (be they species, populations, or artefacts) into a hierarchical pattern

544

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

that reflects closeness of relationship based on the attributes (e.g., genes or morphology) exhibited by the individuals within those groups (Quicke, 1993; Page and Holmes, 1998; McLennan and Brooks, 2001; O’Brien and Lyman, 2003; Felsenstein, 2004). In a phylogenetic sense, ‘relationship’ explicitly refers to genealogical affinities, rather than mere closeness of overall characteristics per se. Put simply, phylogenetics is an historical approach to a given data set (Smith, 1994; O’Brien and Lyman, 2000; Lipo et al., 2006). The most commonly applied means of phylogenetic analysis in biology is cladistics (Eldredge and Cracraft, 1980; Wiley et al., 1991; Minelli, 1993; Quicke, 1993; Smith, 1994; Kitching et al., 1998; Page and Holmes, 1998; Gee, 2000; Schuh, 2000; McLennan and Brooks, 2001). It is has also been adopted as the main phylogenetic method by those investigating questions surrounding cultural phylogenetics within anthropology (e.g., Foley, 1987; Collard and Shennan, 2000; O’Brien et al., 2001, 2002; Tehrani and Collard, 2002; Jordan and Shennan, 2003; O’Brien and Lyman, 2003; Shennan and Collard, 2005; Darwent and O’Brien, 2006; Jordan and Mace, 2006; Buchanan and Collard, 2007). Cladistic analysis is based on the parsimonious null model of taxonomic bifurcation and character evolution. A succinct definition of cladistic analysis has been provided by Brower (2000, p. 13), who states that it is a ‘‘method of grouping by parsimonious patterns of shared character state change’’. Parsimony is the principle that ad hoc character state changes should not be invoked in order to explain their evolution within a set of taxonomic units. Since bifurcation is the simplest model of accounting for the origination of new taxa from existing taxa, this also conforms to the principle of parsimony (i.e. economy of reasoning). It should be emphasised that cladistics does not assume that evolution always occurs in a parsimonious manner, nor that parsimony analysis will produce the correct phylogeny on all occasions (Sober, 1983; Stewart, 1993). Rather, cladistic trees explain the data in the simplest possible manner, and while more complex explanations of the data are possible, these explanations move away from the most parsimonious resolution of the data and the more ad hoc they become, requiring ever-greater complexity of reasoning in order to explain the given data (Sober, 1983). As a result, cladistic analysis provides a justifiable means of estimating phylogeny and testing hypotheses of homology (i.e. that similarity of form

is due to descent from a common ancestor rather than via alternative means) (Sober, 1983; Kitching et al., 1998; McLennan and Brooks, 2001; Richter, 2005). Several accessible introductions to cladistic methodology are now available (e.g., Kitching et al., 1998; Lipscomb, 1998), including some written specifically with archaeologists in mind (e.g., O’Brien et al., 2001; O’Brien and Lyman, 2003). However, in essence, cladistic analysis may be broken down into a small series of fundamental steps (McLennan and Brooks, 2001; Buchanan and Collard, 2007). First, the taxonomic groups to be included in the analysis are delineated. These analytical units are referred to as ‘Operational Taxonomic Units’ (OTUs). Secondly, a character state data matrix is generated describing the character states of each character for each taxon. Thereafter, the direction of evolutionary change (i.e. ‘character polarity’) for the states of each character is estimated, most commonly via comparison with an outgroup. Next, a branching diagram (i.e. cladogram) is constructed that describes relationships for each character. Finally, in accordance with the principle of parsimony, an ensemble cladogram of those branching diagrams is constructed that is consistent with the largest number of characters, and therefore requires the smallest number of ad hoc hypotheses of character change to account for the evolution of characters in the taxa analysed. It is sometimes suggested that cladistic trees reveal little about ancestors since a cladogram may be rearranged into several compatible phylogenetic hypotheses (e.g., Kitching et al., 1998, p. 15– 17). However, when given a specific hypothesis of ancestor–descendant relationships, predictions regarding the topology of a cladogram can be made if it is to be consistent with the given hypothesis. Thus, a purported ‘descendant’ taxon cannot be placed in a more plesiomorphic position relative to its hypothesised ancestor(s) (i.e. be placed lower in the cladogram than its ancestor), and cladistically an ancestor will be placed as the sister taxon to its descendant(s) when treated as a terminal taxonomic unit (Szalay, 1977; Page and Holmes, 1998, 22). By definition, sister taxa share a more recent common ancestor relative to the other taxa in the cladogram and, hence, are more phylogenetically proximate than other taxa. The use of this logic is established in palaeontological practice (Smith, 1994, pp. 130– 133), and has recently been employed in palaeoanthropology to test ancestor–descendant hypotheses

Discoids

Discoids

Discoids Mode 2 Cleavers Mode 2 Cleavers

Mode 2 Handaxes Mode 2 Handaxes

Mode 2 Handaxes Mode 2 Handaxes Mode 3 Levallois

Mode 3 Levallois

Mode 3 Levallois Mode 3 Levallois

Discoids

Discoids

Discoids Mode 2 Cleavers Mode 2 Cleavers

Mode 2 Handaxes Mode 2 Handaxes

Mode 2 Handaxes Mode 2 Handaxes Mode 3 Levallois

Mode 3 Levallois

Mode 3 Levallois Mode 3 Levallois

Mode 1 Discoids Mode 1 Discoids

Mode 1

Mode 1 Mode 1 Mode 1 Mode 1

545

Mode 1

a

Out group

in fossil hominin taxa (Strait and Grine, 2004; Kimbel et al., 2006). Moreover, it is the hypothesis of morphological homology being equivalent to phylogenetic homology that is ultimately tested via a cladistic procedure (McLennan and Brooks, 2001; Richter, 2005). Here, a formal cladistic procedure is used to test the hypothesis of technological phylogenetic homology at the crux of the Movius–Schick hypothesis. The implications of these analyses for the Movius Line are also discussed. From the outset it should be emphasised that, following the recent recommendation of Franz (2005), an ‘iterative’ cladistic procedure is undertaken. That is, results obtained in initial analyses are retested and re-evaluated in the light of new information as the analysis proceeds. It is also hoped that several methodological issues discussed here will contribute to the growing field of cultural phylogenetic analysis.

Out group

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Primary analyses: predictions, materials and methods

Mode 3 Levallois Mode 3 Levallois

Mode 3 Levallois

Mode 2 Handaxes Mode 2 Handaxes Mode 3 Levallois

Mode 2 Handaxes Mode 2 Handaxes

Discoids

Discoids Mode 2 Cleavers Mode 2 Cleavers

Discoids

Mode 1 Discoids

Mode 1 Mode 1

Mode 1

Cladistic methodology enables predictions of tree topologies both consistent and inconsistent with certain ancestor–descendant relationships to be stated explicitly. Fig. 1a–c shows three hypothetical cladograms that are consistent with the Movius– Schick hypothesis, which regards Mode 2 industries as the ancestor of later Mode 3 industries. It should be noted that in all three cladogram topologies, Mode 2 Acheulean Operational Taxonomic Units (OTUs) are placed as sister-taxa to Mode 3 Levallois OTUs, indicating that morphological similarities between these industries can—most parsimoniously—be interpreted as phylogenetically homologous, as required by an ancestor–descendant relationship. Conversely, Fig. 2a–f shows six example cladograms that are inconsistent with the Movius–Schick hypothesis. In Fig. 2a–e, Mode 2 Acheulean OTUs (i.e. handaxes and cleavers) are not the sister taxon of all Mode 3 Levallois OTUs, indicating that alternative taxa share a more recent common ancestor with at least some of the Mode 3 OTUs. Moreover, this may imply that morphological similarities between Mode 2 OTUs and Mode 3 OTUs are the product of convergence rather than via descent from a common ancestor. This latter possibility may be tested via character analysis in order to demonstrate that homoplasic character

b

Out group

Phylogenetic predictions for Movius–Schick hypothesis

c

Fig. 1. Three examples of hypothetical cladograms consistent with the Movius–Schick hypothesis. Note that Mode 2 OTUs (i.e. Handaxes and Cleavers) are positioned as sister taxa to Mode 3 Levallois OTUs.

changes occur upon the cladogram. Fig. 2f shows a topology in which multiple instances of branching (i.e. polytomies) occur. Such polytomies may suggest a lack of phylogenetic signal in the available data, or alternatively, they may indicate that there

e

Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois

Mode 1 Mode 1 Mode 1

Out group Mode 1 Mode 3 Levallois Mode 1 Cleavers Discoids Handaxes Mode 1 Handaxes Cleavers Mode 3 Levallois Discoids Handaxes Mode 1 Mode 3 Levallois Handaxes Discoids Mode 3 Levallois

d

Mode 1 Mode 1 Mode 1 Discoids Discoids Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois

Out group Mode 1 Mode 1 Discoids Discoid Cleavers Handaxes Handaxes Handaxes

c

Out group Mode 1 Mode 1 Mode 1 Mode 1 Discoids Discoids Discoids Cleavers Cleavers Handaxes Handaxes Handaxes Handaxes Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois

b

Out group Mode 1 Mode 1 Mode 1 Mode 1 Discoids Cleavers Handaxes Handaxes Cleavers Handaxes Discoid Discoid Mode 3 Levallois Discoid Mode 3 Levallois Mode 3 Levallois Discoid Mode 3 Levallois

a

Out group Discoids Discoids Discoids Cleavers Handaxes Handaxes Handaxes Handaxes Mode 1 Mode 1

Mode 1 Mode 1 Mode 1 Cleavers Cleavers Handaxes Handaxes Handaxes Discoids Discoids Discoids Discoids Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois Mode 3 Levallois

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Out group Mode 1

546

f

Fig. 2. Six examples of hypothetical cladograms inconsistent with the Movius–Schick hypothesis. Note that Mode 2 OTUs (i.e. Handaxes and Cleavers) are not indicated to be sister taxa to Mode 3 Levallois OTUs. Horizontal bars indicate convergent (i.e. homoplasic) instances of character state evolution.

are no hierarchical relationships between the OTUs such that all are equally related, in a manner inconsistent with the Movius–Schick hypothesis. It should be noted that in these multi-taxon analyses, it requires just a single discoidal or Mode 1 taxon to form a sister taxon with any of the Mode 3 taxa (with accompanying homoplastic character changes occurring independently along the branches leading

to the relevant terminal taxa) for the predictions of the Movius–Schick hypothesis not to be fulfilled. Materials and OTUs Criteria used to construct OTUs As discussed at length by O’Brien and Lyman (2000, 2003), Operational Taxonomic Units (OTUs)

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

are delineated by the analyst for the purpose of addressing particular sets of questions. In other words, the key element in the phrase ‘OTU’ is ‘operational’. As anyone who has even cursorily followed the debate concerning species definitions and ‘species concepts’ will be aware, controversies surrounding the delineation of appropriate units is no less of a problem in biology than it is in archaeology (e.g., Wheeler and Meier, 2000). Notions of convenient or self-evident ‘natural units’ appear to be unfounded, and units must be operationalized according to explicit criteria in order to approach specific research problems (O’Brien and Lyman, 2003, p. 116). It is ultimately the goals and theoretical considerations (e.g., scale) of the particular research question to be addressed that will dictate how such units should be delineated. As O’Brien and Lyman (2003, p. 121) have also noted, in the case of much of prehistory, it is at the scale of traditions (i.e. lineages) that testable hypotheses of artefact phylogeny may be constructed, rather than at the level of ‘cultures’ (i.e. discrete linguistically, socially, economically or otherwise taxonomically diagnosed groups of people). The OTUs described here are probably of this technological scale of artefact ‘tradition’, rather than ‘culture’. However, it should be emphasised that the OTUs described below were constructed specifically to test the Movius–Schick hypothesis and associated predictions, as outlined above. In this regard, it should be noted that at a phylogenetic level, the analyses undertaken here are macro rather than micro. The criteria used to delineate OTUs, the sampling strategy employed and the material ultimately analysed, reflect this broader, more macrolevel goal. Testing the Movius–Schick hypothesis requires the use of multiple OTUs representing the basic different categories of Lower-to-Middle Palaeolithic nuclei (i.e. Mode 1 nuclei, Mode 2 handaxes, Mode 3 Levallois nuclei) (Table 1). Several criteria were used to delineate the OTUs shown in Table 1. These were (1) typological considerations (i.e. based on specific attributes, nuclei can be considered to represent one of the basic lithic nuclei categories required to test the hypothesis rather than another), (2) geographic locality, and (3) stratigraphic considerations (i.e. Mode 1 material from Lower Bed II Olduvai Gorge is placed in a separate OTU to Mode 1 material from Middle Bed II Olduvai Gorge). It should be emphasised that in the case of the typological criteria, it is not merely that certain nuclei possess spe-

547

cific characteristics, but equally that they do not possess all of the characteristics that define alternative nuclei types (although they may possess some of them). The use of several different criteria to delineate OTUs ensures this is a conservative approach to the data, since it is the analysis that will link these sub-divided elements, rather than the analyst. Assemblages and OTUs were chosen, as far as possible, in order to maximise geographical and chronological coverage of the relevant modes (see Table 1). The quantity of relevant literature available was also a consideration, but always subservient to the parameters of geographic locality, stratigraphic considerations and typology. In many cases the assemblages used had previously been cited in the literature as containing Acheulean, Levallois, discoidal or Mode 1 type nuclei. Hence, the criteria applied in such circumstances were largely aimed at unambiguously identifying and delineating these elements in each case. Nuclei assigned to Mode 3 Levallois OTUs were identified on the basis of adherence to Boe¨da’s (1994, 1995) volumetric definition of Levallois core morphology. That is, such nuclei are essentially bifacial in form and possess a line of intersection produced between two discrete, but hierarchically related, regions: one area providing a surface for the removal of predetermined flakes, the other providing an area of strike platforms for the removal of such flakes. Moreover, such nuclei must exhibit some remnant of the distal and lateral convexity possessed by the core prior to the removal of the predetermined flakes. The predetermined blanks will have been removed at an angle parallel to the line of intersection. Nuclei assigned to Mode 3 Levallois OTUs also possessed a disproportionately large negative flake scar, or scars, on their superior surface when compared on a relative basis to other negative flake scars on the same surface. Thus, negative flake scars that are invasive of around 50% of a surface would require a width of around 75% of the total surface width in order to conform to such a criterion. Conversely, a flake that is clearly invasive to P75% of a core’s surface may be comparatively narrow yet still conform to the general criterion of being disproportionately large on a comparative basis. This criterion is in line with van Peer’s (1992, p. 10) observations of Levallois cores. As van Peer (1992, p. 10) notes, Levallois nuclei will bear traces of the negative scars of flakes removed prior to these disproportionately large flake scars and, as such, the larger flake scar(s) truncate the

548

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Table 1 Thirty-four operational taxonomic units employed in analyses (total n = 689 nuclei) Taxonomic unit number

Locality

n

Raw material

Nuclei type/mode

1 2 3 4 5 6 7 8

Barnfield Pit, Kent, UK Barnham St. Gregory, Suffolk, UK Lion Point, Clacton, Essex, UK Olduvai Gorge (Lower Bed II), Tanzania Olduvai Gorge (Middle/Upper Bed II), Tanzania Soan Valley, Pakistan Zhoukoudian, Locality 1, China Zhoukoudian, Locality 15, China

22 30 18 11 26 25 14 11

Chert Chert Chert Lava, chert, quartz Lava, chert, quartz Quartzite Sandstone, quartz, limestone Sandstone, quartz

Ml Ml Ml Ml Ml Ml Ml Ml

9 10 11 12 13 14 15 16 17 18 19 20

Attirampakkam, India Bezez Cave (Level C), Adlun, Lebanon Elveden, Suffolk, UK Kariandusi, Kenya Kharga Oasis (KOl0c), Egypt Lewa, Kenya Olduvai Gorge (Middle/Upper Bed II), Tanzania Morgah, Pakistan St. Acheul, France Tabun Cave (Ed) Attirampakkam, India Kariandusi, Kenya

30 30 24 30 17 30 13 21 30 30 28 21

Quartzite Chert Chert Lava Chert Lava Quartz, lava Quartzite Chert Chert Quartzite Lava

M2 M2 M2 M2 M2 M2 M2 M2 M2 M2 M2 M2

21 22 23 24 25 26 27 28 29

Baker’s Hole, Kent, UK Bezez Cave (Level B), Adlun, Lebanon El Arabah, Abydos, Egypt El Wad (Level F), Israel Fitz James, Oise, France Kamagambo, Kenya Kharga Oasis (KO6e), Egypt Muguruk, Kenya Soan Valley, Pakistan

23 28 16 27 11 13 11 12 11

Chert Chert Chert Chert Chert Quartzite, chert Chert Lava Quartzite

M3 M3 M3 M3 M3 M3 M3 M3 M3

30 31 32 33 34

Attirampakkam, India Hargesia, Somalia Morgah, Pakistan Sheik, Somalia Tabun (Ed), Israel

12 11 12 13 28

Quartzite Quartzite Quartzite Chert, quartzite Chert

Discoid Discoid Discoid Discoid Discoid

preparatory, predetermining flake scar pattern. These attributes tend to avoid confusion with discoidal type reduction strategies (even those potentially exhibiting some hierarchical organisation of faces), which merely exhibit centripetal removal of several flakes invasive to a point central to the core’s surface, yet which are all approximately equal in size. Given the disproportionately large size and invasive nature of these diagnostic ‘Levallois’ flake removals, such scars will also tend to possess a relatively deep (again, on a comparative basis with other scars of the same surface) clear negative bulb of percussion in line with Boe¨da’s (1995) observation that Levallois flake removal is achieved via hard hammer percussion. These criteria are sufficiently broad that both ‘point’ and ‘flake’/’tortoise’ varieties of Mode 3 Levallois nuclei may be incorporated.

(Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Handaxe) (Cleaver) (Cleaver)

Nuclei assigned to Mode 2 Acheulean handaxe OTUs were identified as being ‘tear-drop’, ‘triangular’ or ‘ovate’ outline shape in plan form, and lenticular or triangular shape in cross section, formed by a process of invasive bifacial knapping, resulting in a cutting edge that extends around at least a large proportion of the piece, with an overall form that exhibits an element of bilateral symmetry. As such, nuclei assigned to these Mode 2 OTUs conform to the broad and generally recognised, ‘classic’ handaxe definitions (see e.g., Oakley, 1958, pp. 41–43; Roe, 1976, 66; Isaac, 1977, pp. 117–120; Schick and Toth, 1993, p. 231; Clark, 1994; Debe´nath and Dibble, 1994, 130; Wynn, 1995, pp. 10–11; Mithen, 1996, p. 24; Clark and Kleindienst, 2001, p. 49). Hence, while such a definition may not recognise all of the variability within such a class of artefacts, it is an adequate

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

one to apply in order to test the Movius–Schick hypothesis considered here, since this is based around this ‘classic’ definition of Mode 2 handaxe morphology. This definition is also sufficiently broad to subsume handaxe nuclei manufactured on both cobbles/nodules and flake/slab blanks. Nuclei comprising cleaver OTUs were identified on the basis of being both bifacial and possessing a transverse cutting edge. The latter feature is conventionally seen as the key diagnostic feature of such artefacts (Schick and Toth, 1993; Roe, 1994; Ranov, 2001). Cleavers are generally manufactured on flake blanks, a proportion of which is frequently left unflaked so as to retain the characteristic straight transverse edge. Although cleavers may sometimes be unifacial (Ranov, 2001), only those exhibiting bifacial knapping were included here. Nuclei assigned to Mode 1 OTUs were represented by a broad range of cores, manufactured on both nodule/cobble blanks and flakes. Their diagnostic ‘feature’ is simply that they do not readily display the characteristics that may allow them to be assigned to one of the alternative OTU categories discussed above (i.e. they do not conform to a Mode 3 Levallois, Mode 2 handaxe or cleaver definition), yet all exhibit negative flake scars indicating that such nuclei potentially formed a source of flake tools. As such, the nuclei assigned to these OTUs display a range of shape characteristics including ‘chopper’, polyhedral and discoidal forms. Nuclei specifically segregated for inclusion within Discoid OTUs were delineated on a combination of features. Since experimental work has demonstrated that some discoidal forms may simply be the end product of a reduction strategy designed to provide flake tools rather than a deliberate and specific endproduct in their own right (Toth, 1985), a decision was made only to assign nuclei to a separate ‘Discoid’ OTU (rather than incorporating them within a more general ‘Mode 1’ OTU) in instances where Mode 2 Acheulean and/or Mode 3 Levallois were found within the same context/locality. The logic underlying this protocol decision was that if hominins were capable of knapping a handaxe or Mode 3 Levallois style core, they could—in principle—also deliberately set out to produce a nucleus conforming to the diagnostic ‘discoid’ characteristics, as defined below. Discoidal nuclei were recognised on the basis of a recurrent, bifacial and centripetal pattern of negative flake scars, resulting in a nucleus exhibiting a circular or slightly ovoid shape in plan, and a biconvex shape in section

549

(Terradas, 2003). In order that the ‘circular or slightly ovoid plan-shape’ criterion be applied in a consistent and unambiguous manner, an additional metric requirement was added to this basic discoid definition: the difference between the width and two 45 axis measurements (see below) should not exceed 15% of the length axis measurement. In order to ensure that sample sizes were reasonable for each OTU, it was decided that no OTU should contain less than 10 artefacts. This minimum number was decided on the basis of further methodological requirements. In particular, it should be noted that character states were ultimately coded on the basis of the statistical procedure of divergence coding (see below). The minimum number of individuals to which this method of coding has previously been applied is Lycett and Collard’s (2005) sex specific analyses of Old World monkey genera, which also contained ten individuals in each male/female OTU. For the purposes of comparison, it is notable that O’Brien and Lyman (2003, p. 157) cladistic analysis of Paleoindian projectile points contained a mean of 4.9 specimens per OTU. In order that time spent collecting basic data might be maximised toward collecting the maximum number of OTUs representing as many different localities and raw material types as possible, a further decision was also made not to measure more than 30 individual nuclei for each specific OTU. This protocol was also deemed desirable in an attempt to ensure that discrepancies between the number of individual nuclei within each OTU were not unduly excessive. Where a museum collection for a particular assemblage contained more than 30 total artefacts, each specimen was assigned a number, and 30 specimens were randomly selected from the total assemblage using the program Research Randomizer (http://www.randomizer.org). The impact of this sampling strategy on the analyses undertaken may be analysed empirically (see below). Using the criteria described, a total of 689 lithic nuclei specimens were divided into 34 OTUs for the present study (11–30 nuclei in each). The mean number of nuclei in each OTU was 20.3, with a standard deviation of 7.63 specimens. Table 1 lists the locality, raw material and sample size of individual OTUs. It should be noted that the protocol for assigning OTUs to Mode 1, Mode 2 or Mode 3 designations is consistent with the results of a previously undertaken multivariate morphometric comparative analysis of those nuclei (Lycett, 2007b [Fig. 3]).

550

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Table 2 Characters employed (see Lycett et al., 2006; Lycett 2007b; and online supplementary details for further information) 1. Core left width at 10% of length 2. Core left width at 20% of length 3. Core left width at 25% of length 4. Core left width at 30% of length 5. Core left width at 35% of length 6. Core left width at 40% of length 7. Core left width at 50% of length 8. Core left width at 60% of length 9. Core left width at 65% of length 10. Core left width at 70% of length 11. Core left width at 75% of length 12. Core left width at 80% of length 13. Core left width at 90% of length 14. Core right width at 10% of length 15. Core right width at 20% of length 16. Core right width at 25% of length 17. Core right width at 30% of length 18. Core right width at 35% of length 19. Core right width at 40% of length 20. Core right width at 50% of length 21. Core right width at 60% of length 22. Core right width at 65% of length 23. Core right width at 70% of length 24. Core right width at 75% of length 25. Core right width at 80% of length 26. Core right width at 90% of length 27. Core length distal at 10% of width 28. Core length distal at 20% of width 29. Core length distal at 25% of width 30. Core length distal at 30% of width 31. Core length distal at 40% of width 32. Core length distal at 50% of width 33. Core length distal at 60% of width 34. Core length distal at 70% of width 35. Core length distal at 75% of width 36. Core length distal at 80% of width 37. Core length distal at 90% of width 38. Core length proximal at 10% of width 39. Core length proximal at 20% of Width 40. Core length proximal at 25% of Width 41. Core length proximal at 30% of Width 42. Core length proximal at 40% of Width 43. Core length proximal at 50% of Width 44. Core length proximal at 60% of Width 45. Core length proximal at 70% of Width 46. Core length proximal at 75% of Width 47. Core length proximal at 80% of Width 48. Core length proximal at 90% of Width 49. Coefficient of surface curvature 0–180 50. Coefficient of surface curvature 90–270 51. Coefficient of surface curvature 45–225 52. Coefficient of surface curvature 135–315 53. Coefficient of edge-point undulation 54. Index of symmetry 55. Maximum width/width at orientation 56. Maximum length/length at orientation 57. Nuclei outline length 58. Area of largest flake scar 59. CV of complete flake scar lengths

Table 2 (continued) 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73.

CV complete flake scar widths Total number of complete (i.e. untruncated) flake scars Total number of negative flake scars Number of flakes removed superior and in contact with outline of nucleus Number of non-feather terminations % Cortex (entire nucleus) % Cortex 1st superior quadrant % Cortex 2nd superior quadrant % Cortex 3rd superior quadrant % Cortex 4th superior quadrant % Cortex 1st inferior quadrant % Cortex 2nd inferior quadrant % Cortex 3rd inferior quadrant % Cortex 4th inferior quadrant

Character acquisition Data for a total of 73 characters were collected for all 689 nuclei employed in the analyses (Table 2). It should be noted that this number of characters is in line with phylogenetic analyses of morphometric data in biological studies (e.g., Collard and Wood, 2000 [hominoids = 129 characters; papionins = 62 characters]; Lycett and Collard, 2005 [60 characters]), linguistic data (e.g., Holden, 2006 [92 characters]), and other examples of cladistic analyses of material culture (e.g., Tehrani and Collard, 2002 [90 characters]; Croes et al., 2005 [93 characters]). Morphometric data were collected via use of a Crossbeam Co-ordinate Caliper (hereafter CCC) (Lycett et al., 2006). Characters 1–60 (Table 2) and the protocols used to acquire them have previously been described in detail elsewhere (Lycett et al., 2006; Lycett, 2007b). Variables 1–48 listed in Table 2 (i.e. Euclidean distance variables) were size-adjusted by the geometric mean method (Jungers et al., 1995; Lycett et al., 2006) in order to remove the confounding effects of isometric differences in scale between various finished artefacts and initial blank form sizes. The geometric mean method of size-adjustment equalizes the volumes of the specimens while maintaining overall shape information (Falsetti et al., 1993; Jungers et al., 1995). In addition to these previously described characters, four characters relating to the number and morphology of negative flake scars were included (i.e. characters 61–64, Table 2), and a further nine characters were added relating to cortex (i.e. characters 65–73, Table 2). The protocols used to obtain these characters are included in the online supplementary information. Intra-observer error tests

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

undertaken elsewhere (Lycett, 2007a) have established that these variables were repeatable to an accuracy of greater than 95%. Character coding It is sometimes suggested that quantitative (i.e. continuous) characters are fundamentally different from ‘discrete’ qualitative characters and should be excluded from phylogenetic analyses (e.g., Crisp and Weston, 1987; Pimentel and Riggins, 1987; Cranston and Humphries, 1988; Crowe, 1994). However, such a position has been rebutted by numerous authors (e.g., Baum, 1988; Chappill, 1989; Thiele, 1993; Rae, 1998; Swiderski et al., 1998; MacLeod, 2002) who have shown that qualitative characters are invariably points along a continuous scale of variation yet, confusingly, are described in a manner that merely implies ‘discreteness’ (e.g., ‘moderately curved’, ‘curved’, ‘highly curved’, etc.). As MacLeod (2002, p. 103) has wryly noted, even a seemingly unambiguous and so-called ‘discrete’ character such as colour, which is sometimes used by those who criticise the use of metric characters (e.g., Pimentel and Riggins, 1987), is actually a ratio-scale variable based on the frequency spectrum of reflected light. Hence, there is no justification for rejecting the use of metric characters on the basis of an assumed fundamental difference between the nature of qualitative and quantitative statements about morphology. In addition to this lack of theoretical difference, quantitative characters offer several advantages over qualitative characters in terms of facilitating sizeadjustment so that the confounding effects of size may be diminished (Rae, 2002), and the potential to screen characters for integration via statistical procedures (Nadel-Roberts and Collard, 2005). However, it is in regard to the issue of increasing the repeatability of character coding and reducing ambiguity that quantitative characters offer particular advantages. For instance, while a character such as ‘length’ could arbitrarily be divided into ‘long’, ‘medium’ and ‘short’ character states via qualitative assessment, variations of such a character will be evident both within and between OTUs. If such states overlap, or even come close to overlapping, character state assignation will inevitably become increasingly subjective (Rae, 1998). Conversely, the use of morphometric data allows character state assignations to be made on a non-arbitrary basis via statistical analysis of character variation and difference, even in the event of some degree of overlap

551

(e.g., Thorpe, 1984). It is for these reasons that continuous characters have become increasingly used in wide range of biological phylogenetic analyses (e.g., Simonovic, 1999; Collard and Wood, 2000; Davis et al., 2001; Strait and Grine, 2004; Hibbitts and Fitzgerald, 2005; Lycett and Collard, 2005). In order for morphometric data to be suitable for use as characters in a cladistic analysis, they must be converted into a series of discrete character states. Several procedures have been proposed for converting morphometric data into discrete character states (e.g., Simon, 1983; Thorpe, 1984; Baum, 1988; Mickevich and Johnson, 1976; Archie, 1985; Thiele, 1993). As noted by Rae (1998), the most effective coding methods are those that assign character codes in a non-arbitrary manner on the basis of statistical inference. Indeed, as Richter (2005) has recently argued, the analysis of homology is a two-stage process, whereby putative morphological homology (i.e. correspondence of form) must be assayed prior to a test of phylogenetic homology using parsimony. Hence, a procedure termed divergence coding (Thorpe, 1984) was chosen as the means of character coding here because, in contrast to many alternative coding methods, it assigns character states on the basis of statistical analyses rather than arbitrary decisions, or untested assumptions of homology in the case of overlapping data. Thorpe (1984, p. 252) notes that divergence coding has the advantage of being able to reflect relatively small differences between OTUs, and the number of character states is related directly to the extent of evolution within the character. Divergence coding has been employed in a variety of phylogenetic contexts (e.g., fish (Simonovic, 1999), dung beetles (Davis et al., 2001) and snakes (Hibbitts and Fitzgerald, 2005)), but particularly in the cladistic analysis of primate morphological data (e.g., Collard and Wood, 2000, 2001; Young, 2003; Lycett and Collard, 2005; Nadel-Roberts and Collard, 2005). Divergence coding is very similar to Simon’s (1983) method of homogenous subset coding, which, on occasion (e.g., Strait and Grine, 2004), has been used as a direct substitute for divergence coding due to the virtually identical manner in which the two methods assign character states to taxonomic units. The divergence coding method proceeds as follows. The mean values of each character are placed in ascending order, and a taxon-by-taxon matrix compiled for each of the characters to be coded. Cells in the top row of each matrix are arranged

552

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

such that taxon means decrease from left to right, while in the first column they decrease from the top to the bottom. Thereafter, each cell of the matrix is assigned a score of 1, +1 or 0, depending upon the outcome of the statistical comparisons. Where the mean of the taxon in the column is significantly lower than that of the taxon in the corresponding row, a score of 1 is assigned. Where the mean of the taxon in the column is significantly higher than that of in the corresponding row, a score of +1 is assigned. When differences between taxon means are not statistically significant, a score of 0 is assigned. Once this procedure is completed, the total score for each column (i.e. the sum of every 0, 1, and +1) is computed. Lastly, the relevant integer is added to each taxon total to ensure that each score is positive. The first ten character states (from lowest to highest) may be coded as 0–9. Character states from 11–n may be coded as letters (e.g., A–Z). In coding the data set, a one-way analysis of variance (ANOVA) (a 6 0.05) with post-hoc least significant difference (LSD) pairwise comparisons was employed to test for statistical significance. In the case of the post-hoc LSD tests, there is no requirement to lower the critical alpha level (p-value) below 0.05, if the initial ANOVA is significant (Dytham, 2003, p. 116) (if the initial ANOVA is not significant, the character can be regarded as uninformative and may be discarded). Bonferroni correction procedures were not applied in the pairwise comparisons since such a procedure leads to elevated type II errors (Perneger, 1998; Nakagawa, 2004). In a phylogenetic analysis, elevated type II errors are especially problematic since this will lead to more false similarities (i.e. homoplasies) being incorporated into the data set (Lycett and Collard, 2005, p. 625), and divergence coding is regarded as a conservative approach to character coding for this very reason (Young, 2003, p. 446). Since ANOVA assumes a normal distribution (Sokal and Rohlf, 1995), a Kolmogorov–Smirnov (K–S) test was used to establish that data were normally distributed (p 6 0.05). Where data for a given character were found to be significantly divergent from normal, the data were logarithmically transformed (loge) and re-tested. During the ANOVA, Levene’s test for homogeneity of variances (p 6 0.05) was employed to ensure that data met this assumption. Any characters that were found not to be normally distributed following logarithmic transformation were subjected to a non-parametric Kruskal–Wallis

(KW) test (p 6 0.05) with pairwise Mann–Whitney U-test comparisons. Again, following Dytham (2003, p. 121), the logic behind this procedure is that if the initial Kruskal–Wallis test is significant, there is no requirement to lower the critical p-value below 0.05 for the pairwise comparisons. This procedure was used to code the cortex characters (i.e. Characters 65–73). The ANOVA, LSD, Levene’s test, Kruskal–Wallis test, pairwise Mann–Whitney U-test and K–S tests were undertaken in SPSS v.12.0.1, with subsequent aspects of the coding procedure undertaken using the spreadsheet facilities of Microsoft Excel. Character matrices were subsequently written as text files in NEXUS format for analysis (Swofford, 1998). Screening of data for non-phylogenetic correlations (integration) A fundamental requirement of cladistic analysis is that characters are independent of one another (Farris, 1983; Kluge, 1989). However, as NadelRoberts and Collard (2005, p. 216) have highlighted, it is important to be precise about what is meant by non-independence or ‘integration’ of characters in a phylogenetic setting. As these authors discuss, this is because robust phylogenetic inference relies on the incorporation of characters that are phylogenetically correlated or, in other words, integrated due to a shared phylogenetic history. Strong support for particular clades or branching relationships will depend upon such events being supported by multiple characters. With such considerations in mind, Nadel-Roberts and Collard (2005, pp. 216– 217) noted that characters should only be considered non-phylogenetically correlated (i.e. integrated) if they are found to consistently co-vary with one (or more) other trait(s) across all of the OTUs included in a particular study. Conversely, characters that do not correlate in one or more taxa can reasonably be assumed to have the potential to evolve separately, and may justifiably be employed in a cladistic analysis. Following Nadel-Roberts and Collard (2005), characters were screened for significant correlation (p 6 0.05) via Pearson (product-moment) correlation analyses. However, since Pearson correlation analysis assumes a normal distribution (Sokal and Rohlf, 1995; Dytham, 2003), a Kolmogorov–Smirnov (K–S) test (p 6 0.05) was additionally used to establish that data were normally distributed prior to analysis. Any data found not to be normally distributed were logarithmically transformed (loge)

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

and re-tested. Thereafter, where the data for any two characters were found to be significantly correlated (p 6 0.05) one of these characters was removed from the character matrix for that analysis. It should be noted that when any two variables are correlated in this fashion, it is arbitrary which of these is removed from the dataset in order to remove this redundancy of information. Again, following the protocol used by Nadel-Roberts and Collard (2005), Bonferroni correction procedures for pairwise comparisons were not employed in order to increase the likelihood of finding correlations between characters and thus adopt a more conservative approach to the search for integration within the data set. Pearson correlation analyses were undertaken in PAST v.1.37 (Hammer et al., 2001), freely available online at http://folk.uio.no/ ohammer/past. Kolmogorov–Smirnov (K–S) tests were undertaken in SPSS v.12.0.1. Due to the coded procedure for the recording of cortex, data for cortex-related characters are comprised of discrete variables rather than continuous data. Since Pearson correlation analyses require continuous data, it was not possible to screen cortex characters for non-phylogenetic correlations. The impact of this factor is further considered below. Protocol for phylogenetic analyses All characters with the exception of Character 65 (total percentage cortex remaining on entire nucleus) and any found to be integrated were employed as characters in the initial analyses. Character 65 was not included in the matrix since this would result in the double inclusion of this variable (i.e. information would not be independent of other variables), since this is also counted for the individual quadrants in Characters 66–73. Character 65 is not included until the post-hoc tests, whereupon this character is substituted for the individual quadrant characters in order to assess the impact of recording procedures for this variable (see below). Given the size of the data sets employed here (i.e. 34 OTUs · 73 characters), a heuristic-search algorithm in the phylogenetic software program PAUP*4.0 (Swofford, 1998) was used to search for the most parsimonious tree(s) in order to keep computation times to within reasonable limits. Following the same procedure employed by Holden (2006) to analyse a large linguistic data set (i.e. 75 OTUs · 92 characters), five hundred replications using the tree bisection–reconnection (TBR) algo-

553

rithm with random addition were undertaken, with 2000 trees retained in the memory for each search. Although the TBR heuristic method of branch swapping is the most costly in terms of computation time, it is also the most effective in recovering the most parsimonious cladograms (Kitching et al., 1998, p. 46). Since the characters are quantitative and may therefore be expected to have evolved serially (Rae, 1997, p. 65; Collard and Shennan, 2000, 93), all characters were treated as linearly ordered and freely reversing (Slowinski, 1993). All characters were given equal weight in the analysis. Outgroup selection Choosing an appropriate outgroup in order to root a cladogram and determine character polarity, is a fundamental step in cladistic analysis (Nixon and Carpenter, 1993; Weston, 1994; Lyons-Weiler et al., 1998; McLennan and Brooks, 2001, 12; O’Brien et al., 2002). However, choosing an appropriate outgroup is often considered controversial (O’Brien et al., 2002), since this requires some assumption to be made regarding the relationships of the OTUs even prior to a phylogenetic analysis being undertaken (i.e. it must be decided a priori that at least one of the OTUs is more plesiomorphic than the ingroup taxa). Given that Mode 1 technologies are the technological predecessor of all other forms of knapped lithic technology, were manufactured for at least one million years prior to the introduction of Mode 2 type industries (Asfaw et al., 1992; Clark, 1994), and involve the most simple of knapping techniques (Leakey, 1971; Wynn, 1981; Toth, 1985), choosing one of the Mode 1 OTUs as an outgroup would appear uncontroversial. However, this still requires some criterion to be used in order to determine which of the Mode 1 OTUs be employed for outgroup comparison. In this instance a decision was taken to employ the stratigraphic criterion, whereby the oldest OTU in the database is selected (Smith, 1994, pp. 58–59; Bryant, 2001, 325). Use of stratigraphy for determining outgroups has often been considered controversial since, as many authors have pointed out (e.g., Eldredge and Cracraft, 1980; Aiello and Dean, 1990; O’Brien and Lyman, 2003, 84), the oldest OTU in a dataset may not necessarily be the most plesiomorphic taxon due to imperfections in the stratigraphic record as a result of biased sampling. Hence, a more plesiomorphic taxon may only be found in later stratigraphic horizons due to gaps in the archaeological record.

554

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

However, several workers (e.g., Harper, 1976; Gingerich and Schoeninger, 1977; Szalay, 1977; Gingerich, 1979; Fortey and Chatterton, 1988) have provided arguments that the age of individual OTUs should not be ignored when considering phylogenetic hypotheses, which may justify usage of the oldest taxon as an appropriate outgroup in the absence of alternatives. Moreover, as Smith (1994, pp. 58–59) has argued, use of a taxon close to the basal node of a cladogram for outgroup comparison decreases the risk that the outgroup will have become so derived that it becomes a problematic taxon when determining character polarity. Smith (1994, p. 59) notes that in this sense, the age of the taxon per se is a ‘‘red herring’’, since it is due to evolutionary factors rather than first appearance date alone that an OTU occurring early in the stratigraphic record may justifiably be used as a means of determining character polarity. Following this logic, the Mode 1 OTU from Lower Bed II of Olduvai Gorge (Tanzania) was employed as an outgroup. At c. 1.75–1.6 Mya this is the oldest OTU in the data set (Leakey, 1971; Hay, 1990; Tamrat, 1995). It should be stressed that the impact of this methodological decision may be assessed empirically via post-hoc tests (see below).

Results of primary analyses Character integration Pearson correlation analyses revealed that a total of five characters were significantly (p 6 0.05) correlated with another character in all taxa employed. These were Character 7 (Core left width at 50% of length), Character 16 (Core right width at 25% of length), Character 35 (Core length distal at 75% of width), Character 40 (Core length proximal at 25% of width) and Character 43 (Core length proximal at 50% of width). The relevant characters were subsequently excluded from the character matrix for analysis. Primary cladistic results Cladistic analysis produced a single parsimony tree (Fig. 3) with a tree length of 5075. All 67 characters in the analysis were parsimony informative. With the Olduvai (Lower Bed II) Mode 1 taxon employed as the outgroup, the remaining Mode 1 OTUs were all indicated to be plesiomorphic relative to other OTUs within the analysis. Discoid taxa were determined to be the next most plesiomorphic

Fig. 3. Maximum Parsimony topology produced during initial cladistic analysis. Tree length = 5075.

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

taxa in the tree topology. The analysis supports the most basic prediction of the Movius–Schick hypothesis in providing a tree topology that depicts Mode 2 Acheulean and Mode 3 Levallois OTUs as monophyletic and, hence, sharing a more recent common ancestor with each other relative to all other taxa within the analysis. However, it should be noted that the Mode 2 assemblages are indicated to be more derived than Mode 3 assemblages, a factor not consistent with the predictions of the Movius– Schick hypothesis. Specific aspects of the cladogram topology produced in this primary parsimony analysis are inconsistent with a simple ancestor–descendant scenario between Mode 2 and Mode 3 assemblages. An issue of particular concern generated by these results is the finding that Mode 2 OTUs are more derived relative to Mode 3 industries. This result ensures that despite Mode 2 and Mode 3 taxa sharing a more recent common ancestor as the prediction states, Mode 2 assemblages cannot be seen as ancestral to the Mode 3 taxa which, according to the results produced here, are more plesiomorphic. Moreover, such a result is entirely inconsistent with the archaeological stratigraphic record, whereby Acheulean Mode 2 assemblages precede the appearance of Mode 3 technologies by around 1 million years (Leakey, 1971; Asfaw et al., 1992; Clark, 1994; Kuman and Clarke, 2000). Hence, this finding is not merely inconsistent with the predictions of the Movius–Schick hypothesis, but is contrary to our general understanding of the chronological sequence and development of Palaeolithic technologies during the Pleistocene across Africa, Asia and Europe (Schick and Toth, 1993; Bar-Yosef and Dibble, 1995; Roebroeks and van Kolfschoten, 1995; Foley and Lahr, 1997; Petraglia, 1998; McBrearty, 1999, 2001; Tryon, 2006). Such inconsistencies suggest that some unknown confounding factor may be unduly affecting the analyses. Post-hoc analyses How does sample size, raw material, cortex recording and outgroup choice affect results? In order to more accurately assess factors that may be having an effect upon the tree topology produced in the primary analysis, a series of post-hoc tests were undertaken. These were specifically designed to investigate four factors that may confound the analysis undertaken in the previous sec-

555

tion: (1) the effects of differences between the number of individual specimens included within the OTUs employed, (2) the effect of raw material similarities upon tree topology, (3) the effect of cortex presence and cortex measurement techniques, and (4) the influence of outgroup choice. Sample size As noted above, a tactical decision was taken not to measure more than 30 specimens per taxon in order to maximise data collection time and ensure that a greater number of nuclei from different sites and regions could be included in the analyses. In addition, given that the procedure for the coding of metric characters was that of divergence coding, it was also decided that no less than 10 nuclei should be included within each OTU. As it happened, no less than 11 nuclei were eventually included in all taxa examined. Such a sampling strategy did, however, lead to discrepancies between the number of individuals included in the OTUs analysed. In order to determine whether such differences in OTU sample size were a confounding factor when testing the Movius–Schick hypothesis, a further analysis was undertaken. This involved reducing each OTU down to the minimum number of 11 individual nuclei per taxon. Specimens were randomly selected from OTUs containing more than 11 individuals using the program Research Randomizer (http://www.randomizer.org). Thereafter, the raw data from each taxon were re-subjected to the divergence coding procedure, and a new character matrix compiled. The Movius–Schick hypothesis was then re-tested employing this new character matrix. Parsimony analyses were undertaken as before. The prediction was that if differences in sample size between the individual OTUs employed was a major confounding factor in the analysis undertaken in the previous section, then the new tree topologies should differ markedly from those produced in the primary analysis. Fig. 4 shows the results of parsimony analysis of the reduced data produced a single cladogram with a length of 4899. The topology of this tree (Fig. 4) is consistent with that produced during analysis of the full data set (i.e. the major OTU groupings of Mode 1, Discoidal, Mode 2 and Mode 3 assemblages are congruent with the topology shown in Fig. 3). Again, Mode 2 and Mode 3 assemblages form a clade, and thus fulfil the basic prediction of the Movius–Schick hypothesis that such technologies

556

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Fig. 4. Maximum parsimony topology produced using reduced data set (i.e. 11 nuclei per OTU). Tree length = 4899.

should share a more recent common ancestor with each other than any of these OTUs do with either Mode 1 or discoid OTUs. It is also notable that the two cleaver OTUs (Attirampakkam and Kariandusi) are depicted as sister taxa, as in the analysis of the full data set. Hence, the topology produced from the reduced data set is generally analogous to that resulting from cladistic analysis of the full data set. This is not to say that the two tree topologies are identical. For instance, there are differences between the precise arrangements of taxa within the major techno-typological groupings. However, the overall similarity between the two topologies suggests that sample size differences were not a major confounding factor in the cladistic test of the Movius–Schick hypothesis, nor were a major factor in driving the general positioning of the major techno-typological categories within the cladogram. Raw material Raw material parameters potentially place constraint upon stone tool shape and be a source of homoplasy in morphometric attributes due to the ‘mediatory’ role they play in the manufacturing pro-

cess (Goodman, 1944; Clegg, 1977). Hence, it is possible that raw material may have an influence upon cladistic topologies of stone tools. There is no standard methodology for the categorisation of lithic material types in archaeology, making raw material classifications less than straight forward (Clarkson and O’Conner, 2006; Andrefsky, 1998, p. 40). Despite these difficulties, it is possible to classify the individual artefacts examined here as either quartz, quartzite, igneous, limestone, sandstone or chert. However, given the particulars of the assemblages examined (particularly the fact that no OTU contained solely quartz, limestone or sandstone specimens) assemblages were designated to four basic raw material categories: (i) Quartzite; (ii) Chert; (iii) Igneous; (iv) Mixed. It is important to note that by only having four broad categories, rather than subdividing into more specific categories (e.g., igneous material into basalt, dolerite, etc.), subsequent tests constitute a more conservative approach when determining the influence of raw material on cladogram topology. Here, the Kishino and Hasegawa (1989) test was employed to statistically determine if the cladogram topology produced from the data set is significantly different from ‘model’ cladograms in which raw

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

material is known to be a constraint upon the topology. The test uses the standard deviation of changes in each character for a given topology and the t-statistic in order to determine if the MP topology is significantly different (p = 0.05) from that of a comparative model tree. If the trees are statistically different, the null hypothesis of ‘no difference’ may be rejected at the 95% confidence level (i.e. a = 0.05) (e.g., O’Donnell et al., 1998; Kroken et al., 2003). In essence, the K–H test allows the topologies produced during parsimony analysis to be compared with model ‘explanatory’ trees, and facilitates the possibility of statistically rejecting the model (e.g., that the topologies are entirely the result of raw material parameters) upon which the comparative trees are built. The K–H test has recently been employed in several studies of material culture to determine the goodness-of-fit between cladograms produced from the data and model ‘explanatory’ trees (e.g., Jordan and Shennan, 2003; Jordan and Mace, 2006; Buchanan and Collard, 2007). In order to implement the K–H test, a constraint tree reflecting raw material types was constructed manually in MacClade 4.02 (Maddison and Madd-

557

ison, 2001). This tree was then exported into PAUP*4.0 (Swofford, 1998) and the character matrix was re-subjected to parsimony analysis such that the cladogram reflected the structure defined by the constraint tree. This topology was then employed as the model ‘explanatory’ tree against which the maximum parsimony topology was compared. The comparison tree produced via this process is shown in Fig. 5. K–H analyses were undertaken in PAUP*4.0 (Swofford, 1998). Output of the K–H analyses is shown in Table 3. The maximum parsimony tree produced in the primary analyses is significantly (p 6 0.05) different from the model ‘explanatory’ tree. Hence, the null model that the tree produced during the primary analysis is entirely the product of raw material factors may be rejected. Cortex issues ‘Cortex’ is a general term employed by lithic archaeologists to refer to a mechanically weathered or chemically altered surface that was formed prior to the rock being knapped, such that negative flake scars truncate this surface (Kooyman, 2000, p. 15).

Fig. 5. ‘Explanatory’ raw material model tree used in Kishino–Hasegawa comparisons. This topology was produced via cladistic analysis of the data matrix, but constrained by raw material type. This topology is significantly different (p 6 0.001) from that produced in the primary cladistic analysis demonstrating that the results are not entirely driven by raw material factors (see text for discussion).

558

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Table 3 Results of K–H test Tree

Length

Length difference SD difference p-value

1 2

5075 6906

1831

107.62

<0.0001

Tree 1, maximum parsimony topology; Tree 2, model raw material ‘explanatory’ tree.

However, some nuclei may not be of a material that originally had cortex present due to the chemical and geological circumstances in which it was formed and exposed prior to exploitation by hominins. Moreover, the use of freshly struck flakes or quarried blocks of stone as blanks for cores or core tools may exacerbate this. Such issues potentially confound the use of cortex as a source of characters in cladistic analyses. It is notable that the quadrant system used here for the recording of cortex produces a total of eight characters (i.e. 11.9% of the total characters employed) relating to this single parameter. It should be also noted that the quadrant system creates a series of characters not just related to the presence of cortex, but also directly implicates the position of cortex. Any confounding factors associated with this variable, may thus be exaggerated due to the number of characters being employed in its description. Two analyses were undertaken to investigate the effect of cortex characters on the analysis. In the first analysis, all eight cortex characters (i.e. characters 66–73) were removed from the character matrix, and the parsimony analysis was re-run. In the second analysis, the eight cortex characters were again removed, but replaced with a single character (Character 65) designed to provide a measure of percentage cortex present over the entire core surface. Hence, this second analysis gives some consideration to the presence and quantity of cortex upon nuclei, but only does so via the use of a single character. Moreover, this character gives no emphasis to the position of any cortex that may be present. The prediction in both of these analyses was that if the cortex characters and/or the cortex recording techniques employed were a confounding factor, then the tree topologies produced in these new analyses should differ markedly from that produced in the primary analysis. Parsimony analysis of the character matrix with all eight cortex characters excluded produced a single tree topology with a length of 4454. This cladogram displayed exactly the same arrangement of taxa as that produced when cortex characters were

included. Likewise, with cortex represented by a single character (i.e. overall % cortex), cladistic analysis produced a single tree topology with a length of 4565, which also displayed an identical set of relationships to that suggested in the primary analysis. Hence, cortex recording techniques and any factors biasing the position of cortex do not appear to have confounded the cladistic test of the Movius–Schick hypothesis. Outgroup choice As discussed previously, use of the Olduvai Gorge (Lower Bed II) Mode 1 OTU as an outgroup may be problematic since use of the oldest taxon does not guarantee its status as the most plesiomorphic taxon. Given this fact, a further round of tests was undertaken in order to determine the effect of outgroup choice upon the results of the analysis. The data were re-subjected to parsimony analysis a further seven times. During these re-analyses a different Mode 1 OTU was employed as the outgroup on each occasion. The prediction here was that if outgroup choice was a major confounding factor, marked differences in these new tree topologies should be visible when compared with that produced in the primary analysis. The seven post-hoc analyses each produced a single parsimony tree with a length of 5075; the exact same tree length obtained in the primary analysis. Moreover, no differences of tree topology and suggested relationships between taxa are exhibited in the Discoid, Mode 2 and Mode 3 OTUs. These findings provide robust evidence that outgroup choice was not a confounding factor in the attempt to test the Movius–Schick hypothesis. Assessing the strength of phylogenetic signal and topology determinants Permutation Tail Probability (PTP) tests In order to assess the strength of phylogenetic signal within the character matrix, a permutation tail probability test (PTP test) was undertaken. The PTP test reshuffles (i.e. permutates) the original character matrix without replacement, creating a pre-defined number of pseudoreplicate character matrices. Thereafter, a maximum parsimony tree is computed for each pseudoreplicate character matrix, the lengths of which are compared to the original maximum parsimony tree computed for the unpermutated character matrix. If 95% or more

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

of the cladograms produced from the pseudoreplicate data sets are longer than the original maximum parsimony tree, the original unpermutated character matrix is considered to contain a strong phylogenetic signal (Kitching et al., 1998, p. 123). The PTP test was originally designed as a statistical test (Archie, 1989; Faith, 1991; Faith and Cranston, 1991), although subsequent criticism (e.g., Carpenter, 1992; Kitching et al., 1998), has led to suggestions that it is more usefully considered a heuristic device (Kitching et al., 1998). It is in this heuristic capacity that the PTP test is applied here. PTP tests were undertaken using PAUP*4.0 (Swofford, 1998). Following recent permutation-based analyses undertaken with biological data sets (e.g., Collard and Wood, 2000; Gibbs et al., 2000) and with cultural data (e.g., Tehrani and Collard, 2002; Buchanan and Collard, 2007), 10,000 pseudoreplicate character matrices were examined. The shortest tree returned by the PTP test had a tree length (TL) of 8977, while the longest had a TL of 9380. Hence, 100% of the trees produced from the 10,000 pseudoreplicate data sets were longer than the maximum parsimony tree (TL = 5075). Moreover, the shortest tree produced from the permutated data was 77% longer than the MP tree computed for the unpermutated data. Accordingly, the original character matrix employed to test the Movius–Schick hypothesis can confidently be considered to contain a strong phylogenetic signal. Phylogenetic bootstrapping Bootstrapping in a phylogenetic sense is a method for assessing the level of support for individual clades within a given cladogram (Felsenstein, 1985; Kitching et al., 1998, pp. 129–131). The bootstrap procedure randomly samples characters with replacement, and forms a pseudoreplicate data matrix with the same number of characters and character states as the original. Thereafter, each pseudoreplicate character matrix is subjected to parsimony analysis, with results typically presented in the form of a majority-rule consensus tree as a means of displaying the proportion of resampled data sets that support individual nodes. Like the permutation tail probability test, phylogenetic bootstrapping was initially introduced as a statistical test. However, recent criticisms have left its status as a statistical test in doubt (Kitching et al., 1998, p. 131; Page and Holmes, 1998, pp. 222–223; although see Felsenstein, 2004 for defence of its statistical validity). Given these considerations, the

559

procedure is now generally considered to be a heuristic method of investigating clade robusticity (e.g., Kitching et al., 1998, p. 131; Page and Holmes, 1998, 222–223), and it is in this capacity that the bootstrap is applied here. The analyses were undertaken using PAUP*4.0 (Swofford, 1998). Following recent bootstrap analyses of biological data sets (e.g., Collard and Wood, 2000; Gibbs et al., 2000; Strait and Grine, 2004) and cultural data sets (e.g., Tehrani and Collard, 2002), the character matrix was bootstrapped a total of 10,000 times. Thereafter, minimum length cladograms were computed for each of these pseudoreplicate matrices. A majority-rule consensus tree was subsequently computed. Fig. 6 shows the 50% majority rule topology produced from 10,000 bootstrap replicates. The percentage of instances (P50%) in which a particular node was supported in this bootstrap analysis is indicated by the figures adjacent to the relevant node. Many of the clades are strongly supported and several nodes that distinguish the relative position of the major techno-typological groupings are particularly noteworthy. One such node is that separating the Mode 1 OTUs from all other taxa, which is supported in 71% of the bootstrap replicates. The node that indicates the Discoid OTUs to be a clade is also strongly supported, being returned by 95% of the bootstrap replicates. One of the most important nodes in the parsimony test of the Movius–Schick hypothesis is that indicating Mode 2 and Mode 3 OTUs form a clade. Fig. 6 shows that in the bootstrap analysis, this node was supported in 78% of the pseudoreplicate data sets. Of particular note, is the node that unites all the Mode 2 OTUs as a clade, which is supported in 100% of the bootstrap replicates. RI comparisons When characters perfectly fit a cladistic topology (i.e. in the case of binary characters, the character changes from one state to the other just once at a single node), the characters are considered to be fully consistent with that particular topology (Kitching et al., 1998, p. 95). Conversely, homoplasic character changes (i.e. the character state changes more than once in at different nodes) are the reason why characters are generally less than 100% consistent with a given topology (Goloboff, 1991; Kitching et al., 1998, 95). The ensemble retention index (RI) is a tree statistic commonly used to determine the amount of homoplasy in a given topology (Farris, 1989a; Farris, 1989b). The RI

560

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

57

100

95 92 79 60

71

78 81

55

53 55 98 100 98 72 53 83

99 98 85

Olduvai L Bed II M1 Barnfield Pit M1 Soan Valley M1 Olduvai m/uBed II M1 Loc 1 Zhoukoudian M1 Barnham M1 Loc 15 Zhoukoudian M1 Lion Point M1 Morgah Discoids Hargesia Discoids ATPKM Discoids Sheik Discoids Tabun Discoids Soan Valley M3 Bakers Hole M3 Kharga Oasis M3 Bezez M3 El Wad M3 Kamagambo M3 Cave Valley M3 Fitz James M3 Muguruk M3 Bezez Handaxes M2 Tabun Handaxes M2 St Acheul M2 ATPKM Handaxes M2 Elveden Handaxes M2 Morgah Handaxes M2 Kariandusi Handaxes M2 ATPKM cleavers M2 Kariandusi Cleavers M2 Kharga Oasis Handaxes M2 Lewa Handaxes M2 Olduvai Bed II Handaxes M2

Fig. 6. Fifty percent majority-rule consensus bootstrap tree (10,000 bootstrap replications).

expresses homoplasy relative to the amount of character change that is retained during the hierarchical process of branching as revealed by the most parsimonious cladogram, thus making a distinction between character changes that are synapomorphic as opposed to autapomorphic (Kitching et al., 1998, p. 97; Lipscomb, 1998, 23; Skelton et al., 2002, 59). The retention index is not sensitive to differences between the dimensions of different character matrices, allowing RI values generated from different data sets to be compared directly (e.g., Collard et al., 2006). Collard et al. (2006) recently employed the ensemble RI statistic to compare 20 cladograms produced from anthropological data sets of cultural behaviour to equivalent cladograms generated for 21 biological data sets drawn from a range of behavioural, morphological and genetic studies of various

non-human taxa. Their analyses indicated that, in the case of the cultural data sets, RIs ranged from 0.42 to 0.78 with a mean of 0.59. In the case of the biological data sets, RIs ranged from 0.35 to 0.94 with a mean of 0.61. The RI of the cladogram produced in the present study may usefully be compared against Collard et al.’s results in order to evaluate the relative degree of homoplasy in the data set employed here. An RI for the cladograms produced in the primary parsimony tests of the Movius– Schick hypothesis was computed in McClade 4.02 (Maddison and Maddison, 2001), following importation of the data sets from PAUP*4.0 (Swofford, 1998). If the cladogram generated in the parsimony test the Movius–Schick hypothesis is markedly more homoplasic than those generated for the cultural data sets examined by Collard et al. (2006), we would expect the RI to fall close toward the

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

lower end of, or even fall outside, the RI range for those data sets. The ensemble RI value for the cladogram produced in the parsimony test of the Movius–Schick hypothesis was computed as 0.66. This RI falls toward the higher end of the range of cultural RIs (i.e. range = 0.42–0.78) examined by Collard et al. (2006), and also falls comfortably within the range of biological RIs they examined (i.e. range = 0.35–0.94). Moreover, the RI of the cladogram generated in the parsimony test of the Movius–Schick hypothesis is higher than the mean RI of both the cultural and the biological data sets examined in Collard et al.’s study, where the mean RIs were found to be 0.59 and 0.61, respectively. Hence, this analysis provides no evidence that the cladogram produced in the primary test of the Movius–Schick hypothesis displays a comparatively high level of homoplasy. Character mapping Character mapping was undertaken here to determine how characters, or groups of characters, are influencing the primary cladogram topology, especially any clades where the phylogenetic bootstrap returned high values. This procedure essentially involves an assessment of the fit of particular character states to the tree topologies. The logic here is that such analyses may reveal further insights into the source of any potentially confounding factor not yet revealed by the post-hoc analyses, but which is having an important influence upon the cladogram topologies used to test the phylogenetic predictions of the Movius–Schick hypothesis. Character analyses were undertaken using MacClade 4.02 (Maddison and Maddison, 2001). This program enables the importation of character-state data matrices and tree topologies generated in PAUP*4.0 (Swofford, 1998) and facilitates a visual diagrammatic assessment of character-state placement within a given topology. As indicated above, clades that returned high bootstrap values in the previous section deserve particular scrutiny here, since a specific set of characters may be having a dominant effect in supporting particular clades and, in turn, producing the hypothesised phylogenetic patterns seen in any given topology. The characters conveniently fall into three approximately equal sets of characters describing different aspects of nuclei morphology (see Table 2). These three groups of characters are the left and right lateral outline characters (n = 24), the dis-

561

tal and proximal outline characters (n = 19), and a third set of characters describing parameters such as cortex, symmetry, edge point undulation, surface curvature, flake scar counts and flake scar dimensions (n = 24). In order to perform character mapping, the character matrix was imported into MacClade. Thereafter, the parsimony topology produced in the cladistic test of the Movius–Schick hypothesis was imposed upon the data set. Thereafter, character state placement for each OTU was displayed for the relevant character group using MacClade’s ‘display data boxes’ command. This command provides a colour-coded output indicating patterns of shared character-states within a given topology for each OTU and character group assessed. If particular sets of characters are having a dominant influence upon the topology (i.e. shared character states are influential in supporting a particular clade), this will be indicated by the OTUs within particular parts of the tree sharing the same colour-coded output. Conversely, characters having less influence upon the tree’s topology will exhibit less clear patterning amongst the colour-coded output. Figs. 7a–c show the colour-coded output for the 66 characters used to generate the maximum parsimony topology which is also shown in each figure. Fig. 7a shows the pattern of shared character states within the tree topology for characters 1–26, which all measure left and right lateral outline shape of lithic nuclei. As Fig. 7a demonstrates there is a pattern of shared character states in the Mode 2 clade, indicated by the shared hues of green prevalent within this part of the figure. This indicates that these characters are influential in supporting the Mode 2 clade shown in the parsimony topology. Likewise, Fig. 7b also shows colour-coded patterning of shared character states within the tree topology for characters 27–48, which relate to distal and proximal nuclei outline shape. A pattern of shared character states in the Mode 2 clade is indicated by the shared hues of blue and purple shown in Fig. 7b. Again, this indicates that these characters are influential in supporting the Mode 2 clade. Conversely, less clear patterning is evident in the colourcoded output of Fig. 7c, which relates to characters 49–73. This indicates that characters relating to parameters such as cortex, symmetry, edge point undulation, surface curvature, flake scar counts and flake scar dimensions are less influential in supporting the Mode 2 clade shown in the parsimony topology.

562

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Fig. 7. (a) Character mapping output for characters 1–26 (i.e. left/right lateral outline characters). Characters are shown from 1 (top) to 26 (bottom). Shared character states are indicated by colour-coded output. Note that shared character states are particularly prevalent in the clade containing Mode 2 handaxe OTUs, suggesting that lateral outline characters are particularly influential in the formation of this clade. (b) Character mapping output for characters 27–48. Note that shared character states are particularly prevalent in the clade containing Mode 2 handaxe OTUs, suggesting that these (distal/proximal) outline characters are particularly influential in the formation of this clade. (c) Character mapping output for characters 49–73. Note that in contrast to the previous character groups, less clear patterning of shared character states is evident in the clade containing the Mode 2 handaxe OTUs. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this paper.)

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

563

Fig. 7 (continued)

Re-evaluation: taking account of potentially homoplasic outline variables Rationale, methods and predictions Homoplasy is defined by Lieberman et al. (1996, p. 98) as ‘‘the presence of a similar character state in more than one taxon through mechanisms other

than immediate shared ancestry’’. Homoplasic similarities between taxa will, if prevalent, be misleadingly seen as synapomorphies (i.e. shared derived characters), and lead to false relationships being proposed (Cain, 1982; Stewart, 1993; McHenry, 1996). In the previous section, character mapping indicated that outline variables play a prominent role in the cladogram topology produced during

564

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Fig. 7 (continued)

the primary test of the Movius–Schick hypothesis, particularly in supporting the anachronistic derived Mode 2 clade. However, the outline characters of Mode 2 OTUs may be due to functional parameters relating to their potential use as cutting and chopping implements (Jones, 1980; Isaac, 1986; Ohel, 1987; Schick and Toth, 1993; Ashton and McNabb, 1994; Edwards, 2001). Conversely, Mode 1 nuclei, discoids and Mode 3 nuclei can most parsimoniously be interpreted as the product of knapping in

order to obtain flakes via different techniques, rather than as ‘tools’ in the strictest sense (Toth, 1985; Toth and Schick, 1993). Functional instances of character convergence are known to be a common source of homoplasy in biology, where similar adaptive characteristics evolve in response to analogous problems of survival and fecundity (Cain, 1982; Sanderson and Hufford, 1996). The notion that technological convergences may obscure population histories has been prevalent in archaeology

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

565

for some time now (e.g., Binford and Binford, 1966; Binford, 1973). Indeed, Otte (2003, p. 183) has made specific reference to this phenomenon with regard to bifacial technologies, stating that ‘‘pointed bifacial forms were produced by multiple independently developed technological processes across the time span of human history in which they were used’’. One of the strongest statements regarding the potentially homoplastic nature of Mode 2 handaxe form, however, was made recently by McBrearty (2003, pp. 2–3) who suggested:

clades. Conversely, if the Mode 2 taxa become intermixed with either the Mode 1, Discoid or Mode 3 taxa, this would provide support for the hypothesis that the Mode 2 OTUs have multiple independent origins, thus forming a group that is polyphyletic. That is, the Mode 2 OTUs would not form a group that contains the most recent common ancestor of all its members, as in a classic instance of convergent evolution.

‘‘handaxes are made not born. They were manufactured because they fulfilled some purpose in an ancient subsistence system. . . .[T]he teardrop shape occurs repeatedly in the archaeological record in cases where there is no evidence of ‘phylogenetic’ relation. . . .[Bifaces] from Africa, Europe, the Near East, and even the New World have similar plan forms, but there is clearly no ‘phylogenetic’ relationship among them’’.

When outline (i.e. ‘plan-form’) characters were excluded from character matrix a single parsimony topology with length of 1948 was returned (Fig. 8). While Mode 2 OTUs do not form a clade in this analysis (i.e. they are not monophyletic), they do not form a polyphyletic group, as the hypothesis of multiple independent origins would predict. Rather, the Mode 2 OTUs form a paraphyletic group. Conversely, the Mode 3 OTUs form a derived clade in this analysis. Hence, the hypothesis of homoplasic outline variables cannot be rejected, but the hypothesis of multiple independent origins for Mode 2 OTUs is not supported by this analysis. It is notable that in contrast to the topology produced in the primary test of the Movius–Schick hypothesis, the removal of outline variables returns a cladogram that is consistent with the stratigraphic archaeological record, which indicates that Mode 2 technologies precede Mode 3 technologies in Africa, Asia and Europe. Indeed, the cladogram produced in the revised analysis shows Mode 3 OTUs to be the most derived technology, yet forming a clade with the Mode 2 OTUs, as the Movius–Schick hypothesis would predict. Moreover, if the greater consistency of this tree with stratigraphic information is taken as support for its accuracy, this implies that outline variables are indeed homoplasic, but as a source of homoplasy between Mode 1/Discoid OTUs and the Mode 3 OTUs, rather than among the Mode 2 OTUs per se.

Given McBrearty’s (2003, p. 3) comments regarding the potentially homoplasic nature of biface ‘‘plan forms’’ (see also McBrearty, 2001, p. 84), the employment of characters designed to quantify outline form during the cladistic test of the Movius–Schick hypothesis deserves particular scrutiny. This factor is of even greater significance if we consider that as many as 44 outline-shape characters (66% of the total character matrix) were used in the cladistic test of this hypothesis. Hence, the phylogenetic proximity of the Mode 2 assemblages indicated during the primary analysis of the Movius– Schick hypothesis, may simply be the product of homoplastic outline variables related to the probable ‘cutting edge’ function of these nuclei. In order to test this ‘functional homoplasy’ hypothesis, all outline ‘plan-form’ characters (i.e. left/right lateral and distal/proximal outline characters) were excluded from the character matrix. Thereafter, the remaining 24 characters were subjected to parsimony analysis employing the same protocol as that used in the original analysis. In order to reject the hypothesis that the monophyly of the Mode 2 taxa is entirely the result of homoplastic functionally-related outline variables, at least one of two different cladograms would need to be generated: either the Mode 2 clade remains as a derived monophyletic clade with the Mode 3 OTUs remaining as a paraphyletic group, or the Mode 2 OTUs and Mode 3 OTUs form monophyletic sister

Results

Discussion and conclusions The origins of Mode 3 Levallois technologies are traditionally taken as an important marker in dividing Lower Palaeolithic and Middle Palaeolithic industries (Schick and Toth, 1993; Mellars, 1996; Gao and Norton, 2002; Porat et al., 2002). While it is now recognised that the Lower–Middle Palaeolithic transition was an extended chronological

566

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Fig. 8. Maximum Parsimony topology produced via cladistic analysis with outline (‘plan-form’) characters excluded. Tree length = 1948.

phase between Marine Isotope Stages 9–6 (Monnier, 2006; Tryon et al., 2006), the origins of Levallois industries are widely taken to herald important changes in the technological, social, behavioural, and possibly cognitive elements of the hominin adaptive repertoire (Bordes, 1968; Schlanger, 1996; Dibble and Bar-Yosef, 1995; Foley and Lahr, 1997; Wynn and Coolidge, 2004; Hopkinson, 2007). The process and pattern of this technological shift, however, remain under-explored. For several decades, posited links between Acheulean Mode 2 technologies and Mode 3 Levallois industries have been used to support a potential ancestor–descendant relationship between these nuclei reduction schemes (e.g., van Riet Lowe, 1945; Copeland, 1995; Rolland, 1995; Tuffreau, 1995; Tuffreau and Antoine, 1995; deBono and Goren-Inbar, 2001; Petraglia et al., 2003; Tryon et al., 2006). Others (e.g., Movius, 1969; Schick, 1994, 1998) have even gone as far as to implicate this purported ancestor–descendant scenario in explaining the absence of Mode 3 Levallois industries in East Asia on the basis of a paucity of Acheulean ‘ancestral’ traditions. Here, the phylogenetic basis of this ‘Movius–Schick’ hypothesis has been tested using formal

cladistic procedures of phylogenetic analysis for the first time. The initial test of the ‘Movius–Schick’ hypothesis produced a cladogram partly consistent with its phylogenetic predictions. That is, Mode 2 Acheulean and Mode 3 Levallois OTUs were shown to be monophyletic indicating that morphological similarities between these nuclei could be seen, most parsimoniously, as phylogenetically homologous (i.e. due to descent via a common ancestor), rather than as resulting from technological convergence. However, this initial cladogram also depicted Mode 2 Acheulean OTUs as a derived clade relative to Levallois OTUs, in a manner inconsistent with a simple ancestor–descendant scenario between these technologies. Moreover, the derived Acheulean clade is entirely inconsistent with the archaeological stratigraphic record, which demonstrates that Mode 2 assemblages precede the appearance of Mode 3 assemblages by around 1 million years (Leakey, 1971; Asfaw et al., 1992; Clark, 1994; Foley and Lahr, 1997; Kuman and Clarke, 2000; Tryon, 2006). Such a result suggested that some unknown confounding factor was unduly affecting the analyses, casting the integrity of these results in doubt.

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Subsequent post-hoc analyses further determined, however, that this initial result was not confounded by sample size differences between various OTUs, cortex recording procedures or raw material factors. The result was also shown to be well supported on the basis of the Retention Index (RI) statistic, bootstrapping and permutation tail probability tests. Character mapping, however, revealed that this anachronistic Mode 2 clade was particularly influenced by outline (i.e. plan-form) characters. The finding that characters relating to outline shape were having a dominant effect in producing the anachronistic ‘derived’ Mode 2 clade was particularly important. This is because the outline planform shape of Mode 2 Acheulean biface forms is commonly linked to their functional use as cutting and chopping implements (e.g., Schick and Toth, 1993; Jones, 1994; McBrearty, 2001). In accordance with such logic, it was hypothesised that the phylogenetic proximity of the Mode 2 OTUs returned during the primary analyses may simply be a product of homoplasic outline variables related to the probable ‘cutting edge’ function of these nuclei. In order to test this hypothesis, all outline-shape characters were removed from the data set and the remaining characters re-subjected to parsimony analysis. The prediction in this analysis was that if the phylogenetic proximity of the Mode 2 OTUs is entirely due to the influence of potentially homoplasic outline characters, then this new analysis should return a topology depicting the Mode 2 taxa as a polyphyletic group. That is, the Mode 2 OTUs would not form a group that contains the most recent common ancestor of all its members, and the topology would indicate that Mode 2 OTUs were the product of multiple, independent and convergent evolutionary processes. While this analysis produced a cladogram in which the Mode 2 OTUs were not shown to be monophyletic, the phylogenetic proximity of the Mode 2 taxa was not totally destroyed by removal of the outline characters. Rather, the Mode 2 taxa formed a paraphyletic group, and thus display phylogenetic proximity via a shared common ancestor and shared primitive characters (i.e. symplesiomorphies) (Page and Holmes, 1998, p. 29). What was most striking regarding the results of this new analysis, however, was that the removal of outline-shape variables returned a cladogram consistent with the stratigraphic record, which indicates that Mode 2 technologies preceded Mode 3 technologies in Africa, Asia and Europe. This new

567

cladogram depicted Mode 3 taxa as the most derived, yet they still formed a monophyletic group with the Mode 2 taxa, as predicted by the Movius– Schick hypothesis. Hence, this analysis supports not only the Movius–Schick hypothesis in suggesting that the absence of Mode 3 Levallois technologies in East Asia is due to the lack of a well-established Acheulean tradition, but also the statements of numerous additional workers who have suggested that Mode 2 technologies are the phylogenetic ancestor of Mode 3 technologies (e.g., van Riet Lowe, 1945; Copeland, 1995; Rolland, 1995; Tuffreau, 1995; Tuffreau and Antoine, 1995; deBono and Goren-Inbar, 2001; Petraglia et al., 2003; Tryon et al., 2006). Indeed, if the stratigraphic criterion is taken into account, this analysis implies that outline variables in the primary analysis were a confounding source of homoplasy, but as a source of homoplasy between Mode 1/Discoid OTUs and Mode 3 OTUs rather than among the Mode 2 OTUs per se. The plesiomorphic position of Discoid OTUs returned throughout the analyses is also worthy of comment. Discoid core forms are sometimes implicated in the origins of Mode 3 Levallois industries and discussed alongside them (e.g., Davidson and Noble, 1993, pp. 376–378; Lenoir and Turq, 1995; Ohnuma, 1995; Mourre, 2003; Terradas, 2003; Vaquero and Carbonell, 2003). However, the plesiomorphic position of the discoid OTUs employed here, and particularly in the more stratigraphically consistent topology produced when outline characters were removed, is more consistent with suggestions (e.g., Gowlett, 1986, p. 252) that discoid cores are the technological precursor of Mode 2 Acheulean bifaces, rather than of Mode 3 Levallois cores. Modes under the phylogenetic model Following many recent discussions (e.g., Foley, 1987; Toth, 1987; Toth and Schick, 1993; Rolland, 1995; Carbonell et al., 1999; Petraglia, 1998; Tryon and McBrearty, 2002; Foley and Lahr, 2003; Schick and Clark, 2003; Carbonell et al., 2007; Chauhan, 2007), the overarching, basic taxonomic framework for Palaeolithic technologies employed here has been that of Clark’s (1969) ‘Modes’. However, it should be noted that continued use of this taxonomic scheme has not been without recent criticism (e.g., Bar-Yosef and Belfer-Cohen, 2001; Gamble, 2001; Villa, 2001). One of the disadvantages of calls to think of Modes within a phylogenetic framework

568

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

especially when depicted as a cladogram (e.g., Foley and Lahr, 1997, 2003) has perhaps been a tendency to assume this implies that ‘Modes’ are akin to ‘species’ in biology. That is, that each ‘Mode’ consists of closely-related individuals sharing a common genealogical history at a rather fine level, and ultimately exhibiting a relatively limited set of intra-taxonomic variation (e.g., Carbonell et al., 1999). Such views may also be confounded if it is assumed that a phylogenetic tree is correlated with (or implies) a chronological sequence; an assumption that is sometimes incorrect (O’Brien and Lyman, 2003, p. 84). In biology, three basic groupings of taxa may be suggested under phylogenetic criteria: monophyletic groups, paraphyletic groups and polyphyletic groups (McLennan and Brooks, 2001, pp. 13–15). Monophyletic groups are comprised of taxa sharing a single common ancestor, and are united by shared-derived characteristics (i.e. synapomorphies). Some (e.g., Mishler and Theriot, 2000) would argue that species, and perhaps genera (e.g., Strait et al., 1997), are monophyletic entities. Paraphyletic groups share a common ancestor, but do not contain all of its descendants, and are united by shared primitive characters (i.e. symplesiomorphies). Polyphyletic groups are based on the possession of similar characteristics, but similarity brought about by homoplasic factors rather than those that reflect a common evolutionary history. The analyses undertaken here suggest that the Lower–Middle Palaeolithic Modes generally fall under the category of paraphyletic groups. In other words, Modes may generally represent groups based on shared primitive character states (i.e. symplesiomorphies), even though that group does not contain all of the descendants of the most recent common ancestor. As Alexander (2006, p. 4) noted recently, ‘‘[s]ymplesiomorphies as well as synapomorphies represent character states that are shared because of a common evolutionary origin, and both are useful for recognising groups united by such character states’’. The unintended consequence of thinking of Modes as akin to species, perhaps brought about by the biological origins of phylogenetic methods, may also have encouraged the notion that Modes represent ‘monolithic’ entities (in the literal sense of the term), allowing little margin for the generation or recognition of diversity and variation within each Mode. Such thinking may also mistakenly lead to progressionist scenarios, which do not allow for the loss of some elements (e.g. Mode 2 or Mode 3 technologies) and thus technological ‘reversals’.

However, perhaps a more correct reading of Modes (i.e. that Modes generally represent broad paraphyletic groups), is that they should be thought of as akin to groups at the ‘tribe’, ‘family’ or even the ‘class’ levels of hierarchical classification in biology (see e.g., Simpson, 1945; McKenna and Bell, 1997). As O’Brien and Lyman (2003, p. 45) pointed out recently, recognition of some extremely broad biological groupings as paraphyletic entities (e.g. birds, reptiles and fish) has not caused their widespread abandonment by practising biologists (Brummitt, 2002, 2003, 2006; Nordal and Stedje, 2005; Alexander, 2006). Use of lithic Modes may operate at rather coarse levels, but no more so than the concepts of ‘birds’, ‘fish’ or ‘reptiles’ would be for biologists working at a level of global and continental comparison. Establishing the relative hierarchy of even the basic (supra-specific) taxonomic units is no less an important task in Palaeolithic archaeology than it is in biology (e.g., Simpson, 1945; McKenna and Bell, 1997; see also Bretsky, 1979, pp. 150–156). Employing the term ‘Mode’ for such units does not prevent the additional use of more specific localised terms (e.g. Oldowan, Developed Oldowan, etc.), perhaps in combination, just as Linnaean terminology allows biological units to be defined in an step-wise manner of precision through the use of hierarchically structured levels of taxonomic terms. Moreover, like many higher taxonomic levels in biology, Modes may also share properties of biological ‘grades’. That is, while they are not monophyletic, they most probably possess distinct adaptive characteristics (Huxley, 1958; Collard and Wood, 1999). The recognition of trends brought about by selective processes is as much a part of evolutionary research as the recognition of monophyletic groups (Alexander, 2006), regardless of whether they are the product of natural selection or cultural selection. Modes and the Movius Line Establishing that Acheulean handaxes make the most plausible candidates for the phylogenetic ancestors of Mode 3 Levallois technologies, does not, however, automatically explain why the pattern of technological evolution in East Asia was different to that seen in other regions. Over the years, several suggestions have been made as to why Mode 2 (and ultimately Mode 3) technologies did not become established in East Asia. Although raw material

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

factors, functional differences in each region, or a shift to the use of non-lithic tools have all featured as possible causes, none of these appear to provide compelling explanations (see Schick, 1994 for review). The potential of cultural transmission ‘‘bottlenecks’’ (Schick, 1994, pp. 592–593) may perhaps profitably provide the focus for future discussions. Recent theoretical developments have indicated that, as in the field of population genetics, demographic parameters (i.e. effective population sizes) must be considered if cultural evolutionary patterns are to be interpreted (Neiman, 1995; Shennan, 2000, 2001; Henrich, 2004). Such considerations suggest that technological evolution may be constrained or even undergo reversals in small populations with low density and social interconnectedness (Henrich, 2004; Lycett and von Cramon-Taubadel, 2008). Van Schaik et al. (2003) have shown that in the case of orangutan cultural behaviours, increased opportunity for social associations beyond those of close kin (within a group) correlates directly with the size of cultural repertoire exhibited by different groups. Henrich (2004) has argued that technological reversals due to low demographic levels (i.e. small effective population sizes) will be particularly prevalent where the level of skill required to replicate a given task is relatively more complex. Such factors may be particularly important when considering the relative skill levels required to manufacture Mode 2 versus Mode 1 artefacts (Schick and Toth, 1993; Schick, 1994). Given that pertinent barriers to dispersal from the Indian subcontinent into East Asia exist in the form of the Himalayas and the Ganges–Brahmaputra Delta (Schick and Toth, 1993; Field and Lahr, 2006), it is possible that hominin dispersals into East Asia involved relatively low numbers into what is a large geographic region. As argued recently by Lycett and von Cramon-Taubadel (2008), ‘source’ populations for any potential East Asian ‘Acheulean’ may also have had small effective population sizes due to dispersal factors (i.e. repeated instances cultural bottlenecking along a dispersal route), thus leading to loss of Acheulean technologies. As noted above, some workers (e.g., Yi and Clark, 1983; Hou et al., 2000; Gamble and Marshall, 2001) have argued that occasional occurrences of handaxe-like technologies in East Asia mitigate against a strict Mode 1 versus Mode 2 distribution straddling the Movius Line. However, the morphology of such ‘handaxes’ in terms of an affinity with genuine Acheulean examples has been questioned

569

(e.g., Schick and Zhuan, 1993; Schick, 1994; Corvinus, 2004), as have issues regarding the chronological integrity of such specimens (e.g., Schick, 1994; Dennell, 2003). Indeed, Norton et al. (2006) found statistically significant differences in the thickness of handaxe-like artefacts from the Imjin/Hantan River basins (Korea) compared with bifaces from the Acheulean sites of Hungsi–Baichbal Valleys (India) and Olorgesailie (Kenya). Issues of dating and identification aside, however, it should be noted that a demographic scenario is consistent with the possibility that Acheulean-like technological elements may have occurred sporadically in East Asia, yet never reached the proliferation or technological elaboration of the Acheulean phenomenon seen in many other regions west of the Movius Line due to small effective population sizes (Lycett and von Cramon-Taubadel, 2008). In turn, the evolution of descendant Mode 3 Levallois industries would also have been constrained in East and southeast Asia. Acknowledgements I am especially indebted to Mark Collard, Noreen von Cramon-Taubadel, Parth Chauhan, Paul Coombes, Leslie Aiello, Robin Dennell Bill McGrew and Robert Foley for their invaluable assistance and advice over the years. I also thank Paul Mellars, Michael O’Brien Kathy Schick, Stephen Shennan and Nick Toth for constructive conversations in relation to this work. Three anonymous reviewers offered comments that helped clarify the text, for which I am grateful. Of course, none of the above should be held responsible for the views expressed here, or any remaining errors. Anne Taylor and Nick Ashton kindly facilitated access to lithic material. This work was supported by a research scholarship and post-doctoral funding award from Trinity College (University of Cambridge). Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jaa.2007.07.003. References Aiello, L.C., Dean, C., 1990. An Introduction to Human Evolutionary Anatomy. Academic Press, London. Alexander, P.J., 2006. Descent with modification in evolutionary systematics. Taxon 55, 4.

570

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Ambrose, S.H., 2001. Paleolithic technology and human evolution. Science 29, 1746–1753. Andrefsky, W., 1998. Lithics: Macroscopic Approaches to Analysis. Cambridge University Press, Cambridge. Archie, J., 1985. Methods for coding variable morphological features for numerical taxonomic analysis. Systematic Zoology 34, 326–345. Archie, J., 1989. A randomization test for phylogenetic information in systematic data. Systematic Zoology 38, 219–252. Asfaw, B., Beyene, Y., Suwa, G., Walter, R.C., White, T.D., WoldeGabriel, G., Yemane, T., 1992. The earliest Acheulean from Konso-Gardula. Nature 360, 732–735. Ashton, N., McNabb, J., 1994. Bifaces in perspective. In: Ashton, N., David, A. (Eds.), Stories in Stone: Lithic Studies Society Occasional Paper No. 4. Lithic Studies Society, London, pp. 182–191. Bar-Yosef, O., Belfer-Cohen, A., 2001. From Africa to Eurasia— early dispersals. Quaternary International 75, 19–28. Bar-Yosef, O., Dibble, H.L., 1995. Preface. In: The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. ix–xiii. Baum, B., 1988. A simple procedure for establishing discrete characters from measurement data, applicable to cladistics. Taxon 37, 63–70. Binford, L.R., 1973. Interassemblage variability: the Mousterian and the functional argument. In: Renfrew, C. (Ed.), The Exploration of Culture Change. University of Pittsburgh Press, Pittsburgh, pp. 227–254. Binford, L.R., Binford, S.R., 1966. A preliminary analysis of functional variability in the Mousterian of Levallois facies. American Anthropologist 68, 238–295. Boe¨da, E., 1994. Le Concept Levallois: Variabilite´ des Me´thodes. Centre de la Recherche Scientifique (CNRS), Paris. Boe¨da, E., 1995. Levallois: a volumetric construction, methods, a technique. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 41–68. Bordes, F., 1968. The Old Stone Age. Weidenfield and Nicolson, London. Brantingham, P.J., Olsen, J.W., Rech, J.A., Krivoshapkin, A.I., 2000. Raw material quality and prepared core technologies in Northeast Asia. Journal of Archaeological Science 27, 255– 271. Bretsky, S.S., 1979. Recognition of ancestor–descendant relationships in invertebrate paleontology. In: Cracraft, J., Eldredge, N. (Eds.), Phylogenetic Analysis and Paleontology. Columbia University Press, New York, pp. 113–163. Brower, A.V.Z., 2000. Homology and the inference of systematic relationships: some historical and philosophical perspectives. In: Scotland, R., Pennington, R.T. (Eds.), Homology and Systematics: Coding Characters for phylogenetic Analysis. Taylor and Francis, London, pp. 10–21. Brummitt, R.K., 2002. How to chop up a tree. Taxon 51, 31–34. Brummitt, R.K., 2003. Further dogged defense of paraphyletic taxa. Taxon 52, 803–804. Brummitt, R.K., 2006. Am I a bony fish?. Taxon 55 268–269. Bryant, H.N., 2001. Character polarity and the rooting of cladograms. In: Wagner, G.P. (Ed.), The Character concept in Evolutionary Biology. Academic Press, London, pp. 319– 338. Buchanan, B., Collard, M., 2007. Investigating the peopling of North America through cladistic analyses of Early Paleoin-

dian projectile points. Journal of Anthropological Archaeology. doi:10.1016/j.jaa.2007.02.005. Cain, A.J., 1982. On homology and convergence. In: Joysey, K.A., Friday, A.E. (Eds.), Problems of Phylogenetic Reconstruction. Academic Press, London, pp. 1–19. Carbonell, E., Mosquera, M., Xose, P.R., Sala, R., 1999. Out of Africa: the dispersal of the earliest technical systems reconsidered. Journal of Anthropological Archaeology 18, 119–136. Carbonell, E., Mosquera, M., Rodrı´guez, P., 2007. The emergence of technology: a cultural step or long-term evolution? Comptes Rendus Palevol 6, 231–233. Carpenter, J.M., 1992. Random cladistics. Cladistics 8, 147–153. Chappill, J., 1989. Quantitative characters in phylogenetic analysis. Cladistics 5, 217–234. Chauhan, P.R., 2004. A Review of the Early Acheulian evidence from South Asia, Assemblage 8. Avaialble from: http:// www.shef.ac.uk/assem/issue8/chauhan.html. Chauhan, P.R., 2007. Soanian cores and core-tools from Toka, Northern India: towards a new typo-technological organization. Journal of Anthropological Archaeology. doi:10.1016/ j.jaa.2007.01.00. Clark, G., 1969. World Prehistory: A New Outline, second ed. Cambridge University Press, Cambridge. Clark, J.D., 1994. The Acheulian industrial complex in Africa and elsewhere. In: Corruccini, R.S., Ciochon, R.L. (Eds.), Integrative Paths to the Past. Prentice Hall, Englewood Cliffs, NJ, pp. 451–469. Clark, J.D., 1998. The Early Palaeolithic of the eastern region of the Old World in comparison to the west. In: Korisettar, R., Petraglia, M.D. (Eds.), Early Human Behaviour in Global Context: The Rise and Diversity of the Lower Palaeolithic Record. Routledge, London, pp. 437–450. Clark, J.D., Kleindienst, M.R., 2001. The stone age cultural sequence: terminology, typology and raw material. In: Clark, J.D. (Ed.), Kalambo Falls Prehistoric Site:, vol. III. Cambridge University Press, Cambridge, pp. 34–65. Clarkson, C., O’Conner, S., 2006. An introduction to stone artefact analysis. In: Balme, J., Paterson, A. (Eds.), Archaeology in Practice: A Student Guide to Archaeological Analyses. Blackwell, Oxford, pp. 159–199. Clegg, J.K., 1977. The four dimensions of artefactual variation. In: Wright, R.V.S. (Ed.), Stone Tools as Cultural Markers: Change, Evolution and Complexity. Australian Institute of Aboriginal Studies, Canberra, pp. 60–66. Collard, M., Wood, B.A., 1999. Grades among the African early hominids. In: Bromage, T.G., Schrenk, F. (Eds.), African Biogeography, Climate Change and Human Evolution. Oxford University Press, Oxford, pp. 316–327. Collard, M., Shennan, S., 2000. Processes of cultural change in prehistory: a case study from the European Neolithic. In: Renfrew, C., Boyle, K. (Eds.), Archaeogenetics: DNA and the Population Prehistory of Europe. McDonald Institute for Archaeological Research, Cambridge, pp. 89–97. Collard, M., Wood, B., 2000. How reliable are human phylogenetic hypotheses?. Proceedings of the National Academy of Sciences of the United States of America 97 5003–5006. Collard, M., Wood, B., 2001. Homoplasy and the early hominid masticatory system: inferences from analyses of extant hominoids and papionins. Journal of Human Evolution 41, 167– 194. Collard, M., Shennan, S.J., Tehrani, J.J., 2006. Branching, blending, and the evolution of cultural similarities and

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575 differences among human populations. Evolution and Human Behavior 27, 169–184. Copeland, L., 1995. Are Levallois flakes in the Levantine Acheulian the result of biface preparation? In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 171–183. Corvinus, G., 2004. Homo erectus in East and Southeast Asia, and the questions of the age of the species and its association with stone artifacts, with special reference to handaxe-like tools. Quaternary International 117, 141–151. Cranston, P., Humphries, C., 1988. Cladistics and computers: a chironomid conundrum? Cladistics 4, 72–92. Crisp, M., Weston, P., 1987. Cladistics and legume systematics, with an analysis of the Bossiaeeae, Brongniartieae and Mirbelieae. In: Stirton, C. (Ed.), Advances in Legume Systematics, Part 3. Royal Botanical Gardens, Kew, London, pp. 65–130. Croes, D., Kelly, K.M., Collard, M., 2005. Cultural historical context of Qwu?gwes (Puget Sound, USA): a preliminary investigation. Journal of Wetland Archaeology 5, 141–154. Crowe, T.M., 1994. Morphometrics, phylogenetic models and cladistics: means to an end or much to do about nothing? Cladistics 10, 77–84. Darwent, J., O’Brien, M.J., 2006. Using cladistics to reconstruct lineages of projectile points from northeastern Missouri. In: Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S. (Eds.), Mapping Our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory. Aldine Transaction, New Brunswick, NJ, pp. 185–208. Darwin, C., 1859. On the Origin of Species. John Murray, London. Davidson, I., Noble, W., 1993. Tools and language in human evolution. In: Gibson, K.R., Ingold, T. (Eds.), Tools, Language and Cognition in Human Evolution. Cambridge University Press, Cambridge, pp. 363–388. Davis, A.L.V., Scholtz, C.H., Harrison, J.D.G., 2001. Cladistic, phenetic and biogeographical analysis of the flightless dung beetle genus, Gyronotus van Lansberge (Scarabaeidae: Scarabaeinae), in threatened eastern Afrotropical forests. Journal of Natural History 35, 1607–1625. Debe´nath, A., Dibble, H.L., 1994. Handbook of Paleolithic Typology. University Museum, University of Pennsylvania, Philadelphia. deBono, H., Goren-Inbar, N., 2001. Note of a link between Acheulian handaxes and the Levallois method. Journal of the Israel Prehistoric Society 31, 9–23. Dennell, R., 2003. Dispersal and colonisation, long and short chronologies: how continuous is the Early Pleistocene record for hominids outside East Africa. Journal of Human Evolution 45, 421–440. Dennell, R.W., 2004. Early Hominin Landscapes in Northern Pakistan: Investigations in the Pabbi Hills. British Archaeological Reports, Oxford. Dibble, H.L., Bar-Yosef, O. (Eds.), 1995. The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin. Dunnell, R.C., 2006. Measuring relatedness. In: Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S. (Eds.), Mapping Our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory. Aldine Transaction, New Brunswick, NJ, pp. 109– 116.

571

Durham, W.H., 1992. Applications of evolutionary culture theory. Annual Review of Anthropology 21, 331–355. Dytham, C., 2003. Choosing and Using Statistics: A Biologist’s Guide, second ed. Blackwell Science, Oxford. Edwards, S.W., 2001. A modern knapper’s assessment of the technical skills of the Late Acheulean biface workers at Kalambo Falls. In: Clark, J.D. (Ed.), Kalambo Falls Prehistoric Site, vol. III. Cambridge University Press, Cambridge. Eldredge, N., Cracraft, J., 1980. Phylogenetic Patterns and the Evolutionary Process. Columbia University Press, New York. Faith, D.P., 1991. Cladistic permutation tests for monophyly and non-monophyly. Systematic Zoology 40, 366–375. Faith, D.P., Cranston, P.S., 1991. Could a cladogram this short have arisen by chance alone? On permutation tests for cladistic structure. Cladistics 7, 1–28. Falsetti, A.B., Jungers, W.L., Cole III, T.M., 1993. Morphometrics of the Callitrichid forelimb: a case study in size and shape. International Journal of Primatology 14, 551–572. Farris, J.S., 1983. The logical basis of phylogenetic analysis. In: Platnik, N., Funk, V. (Eds.), Advances in Cladistics, vol. 2. Columbia University Press, New York, pp. 7–36. Farris, J.S., 1989a. The retention index and homoplasy excess. Systematic Zoology 38, 406–407. Farris, J.S., 1989b. The retention index and the rescaled consistency index. Cladistics 5, 417–419. Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–791. Felsenstein, J., 2004. Inferring Phylogenies. Sinauer, Sunderland, MA. Field, J.S., Lahr, M.M., 2006. Assessment of the southern dispersal: GIS-based analyses of potential routes at Oxygen Isotopic Stage 4. Journal of World Prehistory 19, 1–45. Foley, R.A., 1987. Hominid species and stone tool assemblages: how are they related?. Antiquity 61 380–392. Foley, R.A., Lahr, M.M., 1997. Mode 3 technologies and the evolution of modern humans. Cambridge Archaeological Journal 7, 3–36. Foley, R.A., Lahr, M.M., 2003. On stony ground: lithic technology, human evolution, and the emergence of culture. Evolutionary Anthropology 12, 109–122. Fortey, R.A., Chatterton, B.D.E., 1988. Classification of the Trilobite suborder Asaphina. Paleontology 31, 165–222. Franz, N.M., 2005. Outline of an explanatory account of cladistic practice. Biology and Philosophy 20, 489–515. Gamble, C., 2001. Modes, movement and moderns. Quaternary International 75, 5–10. Gamble, C., Marshall, G., 2001. The shape of handaxes, the structure of the Acheulian world. In: Milliken, S., Cook, J. (Eds.), A Very Remote Period Indeed: Papers on the Palaeolithic Presented to Derek Roe. Oxbow Books, Oxford, pp. 19–27. Gao, X., Norton, C.J., 2002. A critique of the Chinese ‘Middle Palaeolithic’. Antiquity 76, 397–412. Gee, H., 2000. Deep Time: Cladistics, The Revolution in Evolution. Fourth Estate, London. Gibbs, S., Collard, M., Wood, B., 2000. Soft-tissue characters in higher primate phylogenetics. Proceedings of the National Academy of Sciences of the United States of America 97, 11130–11132. Gingerich, P.D., 1979. The stratophenetic approach to phylogeny reconstruction in vertebrate paleontology. In: Cracraft, J.,

572

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Eldredge, N. (Eds.), Phylogenetic Analysis and Paleontology. Columbia University Press, New York, pp. 41–77. Goloboff, P.A., 1991. Homoplasy and the choice among cladograms. Cladistics 7, 215–232. Goodman, M.E., 1944. The physical properties of stone tool materials. American Antiquity 9, 415–433. Gowlett, J.A.J., 1986. Culture and conceptualisation: the Oldowan–Acheulian gradient. In: Bailey, G.N., Callow, P. (Eds.), Stone Age Prehistory. Cambridge University Press, Cambridge, pp. 243–260. Hammer, Ø., Harpner, D.A.T., Ryan, P.D., 2001. PAST: Palaeontological statistics software package for education and data analysis. Palaeontologica–Electronica 4, 1–9. Harper, C.W., 1976. Phylogenetic inference in paleontology. Journal of Paleontology 50, 180–193. Hay, R.L., 1990. Olduvai Gorge: a case history in the interpretation of hominid palaeoenvironments in East Africa. Geological Society of America Special Paper 242, 23–37. Henrich, J., 2004. Demography and cultural evolution: how adaptive cultural processes can produce maladaptive losses— the Tasmanian case. American Antiquity 69, 197–214. Hibbitts, T.J., Fitzgerald, L.A., 2005. Morphological and ecological convergence in two nactricine snakes. Biological Journal of the Linnean Society 85, 363–371. Holden, C.J., 2006. The spread of Bantu Languages, farming, and pastoralism in sub-equatorial Africa. In: Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S. (Eds.), Mapping Our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory. Aldine Transaction, New Brunswick, NJ, pp. 249–267. Hopkinson, T., 2007. The transition from the Lower to the Middle Palaeolithic in Europe and the incorporation of difference. Antiquity 81, 294–307. Hou, Y., Potts, R., Baoyin, Y., Zhengtang, G., Deino, A., Wei, W., Clark, J., Guangmao, X., Weiwen, H., 2000. MidPleistocene Acheulean-like stone technology of the Bose Basin, South China. Science 287, 1622–1626. Huxley, J., 1958. Evolutionary processes and taxonomy with special reference to grades. Uppsala Universitet Arssks 6, 21– 38. Isaac, G.L., 1977. Olorgesailie: Archaeological Studies of a Middle Pleistocene Lake Basin in Kenya. University of Chicago Press, Chicago. Isaac, G.L., 1986. Foundation stones: early artefacts as indicators of activities and abilities. In: Bailey, G.N., Callow, P. (Eds.), Stone Age Prehistory. Cambridge University Press, Cambridge, pp. 221–241. Jones, P.R., 1980. Experimental butchery with modern stone tools and its relevance for Palaeolithic archaeology. World Archaeology 12, 153–165. Jones, P.R., 1994. Results of experimental work in relation to the stone industries of Olduvai Gorge. In: Leakey, M.D., Roe, D.A. (Eds.), Olduvai Gorge: Volume 5. Cambridge University Press, Cambridge, pp. 254–298. Jordan, P., Shennan, S., 2003. Cultural transmission, language, and basketry traditions amongst the California Indians. Journal of Anthropological Archaeology 22, 42–74. Jordan, P., Mace, T., 2006. Tracking culture-historical lineages: can ‘descent with modification’ be linked to ‘association by descent’? In: Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S. (Eds.), Mapping Our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory. Aldine Transaction, New Brunswick, pp. 149–167.

Jungers, W.L., Falsetti, A.B., Wall, C.E., 1995. Shape, relative size, and size adjustments in morphometrics. Yearbook of Physical Anthropology 38, 137–161. Keates, S.G., 2002. The Movius Line: fact or fiction? Bulletin of the Indo-Pacific Prehistory Association 22, 17–24. Kimbel, W.H., Lockwood, C.A., Ward, C.V., Leakey, M.G., Rak, Y., Johanson, D.C., 2006. Was Australopithecus anamensis ancestral to A. afarensis? A case of anagenesis in the hominin fossil record. Journal of Human Evolution 51, 134–152. Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29, 170–179. Kitching, I.J., Forey, P.L., Humphries, C.J., Williams, D.M., 1998. Cladistics: The Theory and Practice of Parsimony Analysis. Oxford University Press, Oxford. Kluge, A.G., 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Systematic Zoology 38, 7–25. Kooyman, B.P., 2000. Understanding Stone Tools and Archaeological Sites. University of New Mexico Press, Albuquerque. Kroken, S., Glass, N.L., Taylor, J.W., Yoder, O.C., Turgeon, B.G., 2003. Phylogenomic analysis of type I polyketide synthase genes in pathogenic and saprobic ascomycetes. Proceedings of the National Academy of Sciences of the United States of America 100, 15670–15675. Kuhn, S.L., 2004. Evolutionary perspectives on technology and technological change. World Archaeology 36, 561–570. Kuman, K., Clarke, R.J., 2000. Stratigraphy, artefact industries and hominid associations for Sterkfontein, Member 5. Journal of Human Evolution 38, 827–847. Leakey, M.D., 1971Olduvai Gorge: Excavations in Beds I and II, 1960–1963, vol. 3. Cambridge University Press, Cambridge. Lenoir, M., Turq, A., 1995. Recurrent centripetal debitage (Levallois and discoidal): continuity or discontinuity? In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 249–256. Leroi-Gourhan, A., 1966. La Pre´histoire. Universitaires de France, Paris. Lieberman, D.E., Wood, B.A., Pilbeam, D.R., 1996. Homoplasy and early Homo: an analysis of the evolutionary relationships of H. habilis sensu stricto and H. rudolfensis. Journal of Human Evolution 30, 97–120. Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S., 2006. Cultural phylogenies and explanation: why historical methods matter. In: Lipo, C.P., O’Brien, M.J., Collard, M., Shennan, S. (Eds.), Mapping Our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory. Aldine Transaction, New Brunswick, NJ, pp. 3–16. Lipscomb, D., 1998. Basics of Cladistic Analysis. Available from: http://www.gwu.edu/~clade/faculty/lipscomb/Cladistics.pdf. Lycett, S.J., 2007a. ‘Stone Trees’: Integrative Cladistic and Morphometric Approaches to the Phylogenetics of Palaeolithic Technologies, Ph.D. Thesis, University of Cambridge. Lycett, S.J., 2007b. Is the Soanian techno-complex a Mode 1 or Mode 3 phenomenon? A morphometric assessment. Journal of Archaeological Science 34, 1434–1440. Lycett, S.J., Collard, M., 2005. Do homoiologies impede phylogenetic analyses of the fossil hominids? An assessment based on extant papionin craniodental morphology. Journal of Human Evolution 49, 618–642.

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575 Lycett, S.J., von Cramon-Taubadel, N., 2008. Acheulean variability and hominin dispersals: a model-bound approach. Journal of Archaeological Science. doi:10.1016/j.jas. 2007.05.003. Lycett, S.J., von Cramon-Taubadel, N., Foley, R.A., 2006. A crossbeam co-ordinate caliper for the morphometric analysis of lithic nuclei: a description, test and empirical examples of application. Journal of Archaeological Science 33, 847–861. Lyons-Weiler, J., Hoelzer, G.A., Tausch, R.J., 1998. Optimal outgroup analysis. Biological Journal of the Linnaean Society 64, 493–511. Mace, R., 2005. Introduction: a phylogenetic approach to the evolution of cultural diversity. In: Mace, R., Holden, C.J., Shennan, S. (Eds.), The Evolution of Cultural diversity. UCL Press, London, pp. 1–10. MacLeod, N., 2002. Phylogenetic signals in morphometric data. In: MacLeod, N., Forey, P.L. (Eds.), Morphology, Shape and Phylogeny. Taylor & Francis, London, pp. 100–138. Maddison, D.R., Maddison, W.P., 2001. MacClade 4: Analysis of Phylogeny and Character Evolution, Version 4.02. Sinauer Associates, Sunderland, MA. McBrearty, S., 1999. The Archaeology of the Kapthurin formation. In: Andrews, P., Banham, P. (Eds.), Late Cenozoic Environments and Hominid Evolution: A Tribute to Bill Bishop. Geological Society, London, pp. 143–156. McBrearty, S., 2001. The Middle Pleistocene of East Africa. In: Barham, L., Robson-Brown, K. (Eds.), Human Roots: Africa and Asia in the Middle Pleistocene. Western Academic and Specialist Press Ltd, Bristol, pp. 81–98. McBrearty, S., 2003. Patterns of technological change at the origin of Homo sapiens. Before Farming 3, 1–5. McGrew, W.C., 2004. The Cultured Chimpanzee: Reflections on Cultural Primatology. Cambridge University Press, Cambridge. McHenry, H.M., 1996. Homoplasy, clades, and hominid phylogeny. In: Meikle, W.E., Howell, F.C., Jablonski, N.G. (Eds.), Contemporary Issues in Human Evolution. California Academy of Sciences, San Francisco, pp. 77–92. McKenna, M.C., Bell, S.K., 1997. Classification of Mammals Above the Species Level. Columbia University Press, New York. McLennan, D.A., Brooks, D.R., 2001. Phylogenetic systematics: five steps to enlightenment. In: Adrain, J.M., Edgcombe, G.D., Lieberman, B.S. (Eds.), Fossils, Phylogeny and Form: An Analytical Approach. Kluwer Academic/Plenum, New York, pp. 7–28. Mellars, P., 1996. The Neanderthal Legacy: An Archaeological Perspective from Western Europe. Princeton University Press, Princeton, NJ. Mesoudi, A., Whiten, A., Laland, K.N., 2004. Is human cultural evolution Darwinian? Evidence reviewed from the perspective of The Origin of Species. Evolution 58, 1–11. Mesoudi, A., Whiten, A., Laland, K.N., 2006. Towards a unified science of cultural evolution. Behavioral and Brain Sciences 29, 329–383. Mickevich, M., Johnson, M., 1976. Congruence between morphological and allozyme data in evolutionary inference and character evolution. Systematic Zoology 25, 260–270. Minelli, A., 1993. Biological Systematics: The State of the Art. Chapman and Hall, London. Mishler, B.D., Theriot, E.C., 2000. The phylogenetic species concept (sensu Mishler & Theriot): monophyly, apomorphy

573

and phylogenetic species concepts. In: Wheeler, Q.D., Meier, R. (Eds.), Species Concepts and Phylogenetic Theory: A Debate. Columbia University Press, New York, pp. 44–54. Misra, V.N., 2001. Prehistoric human colonization of India. Journal of Bioscience 26, 491–531. Mithen, S., 1996. The Prehistory of the Mind: A Search for the Origins of Art, Religion and Science. Thames and Hudson, London. Monnier, G.F., 2006. The Lower/Middle Paleolithic periodization in western Europe. Current Anthropology 47, 709–744. Mourre, V., 2003. Discoı¨d ou pas discoı¨d? Re´flexions sur la pertinence de crite`res techniques de´finissant le de´bitage discoı¨d. In: Peresani, M. (Ed.), Discoid Lithic Technology: Advances and Implications. British Archaeological Reports, Oxford, pp. 1–18. Movius, H.L., 1948. The Lower Palaeolithic Cultures of Southern and Eastern Asia. Transactions of the American Philosophical Society 38, 329–426. Movius, H., 1969. Lower Paleolithic archaeology in southern Asia and the Far East. In: Howells, W.W. (Ed.), Early Man in the Far East, Studies in Physical Anthropology, No.1. Humanities Press, New York, pp. 17–82. Nadel-Roberts, M., Collard, M., 2005. Impact of methodological choices on assessments of the reliability of fossil primate phylogenetic hypotheses. Folia Primatologica 76, 207–221. Nakagawa, S., 2004. A farewell to Bonferroni: the problems of low statistical power and publication bias. Behavioral Ecology 15, 1044–1045. Neiman, F.D., 1995. Stylistic variation in evolutionary perspective: inferences from decorative diversity and interassemblage distance in Illinois Woodland ceramic assemblages. American Antiquity 60, 7–36. Nixon, K.C., Carpenter, J.M., 1993. On outgroups. Cladistics 9, 413–426. Nordal, I., Stedje, B., 2005. Paraphyletic taxa should be accepted. Taxon 54, 5–7. Norton, C.J., Bae, K., Harris, J.W.K., Lee, H., 2006. Middle Pleistocene handaxes from the Korean Peninsula. Journal of Human Evolution 51, 527–536. Oakley, K.P., 1958. Man the Tool-Maker, fourth ed. Trustees of the British Museum, London. O’Brien, M.J., Lyman, R.L., 2000. Applying Evolutionary Archaeology: A Systematic Approach. Kluwer Academic/ Plenum, New York. O’Brien, M.J., Lyman, R.L., 2003. Cladistics and Archaeology. University of Utah Press, Salt Lake City, Utah. O’Brien, M.J., Darwent, J., Lyman, R.L., 2001. Cladistics is useful for reconstructing archaeological phylogenies: Palaeoindian points from the southeastern United States. Journal of Archaeological Science 28, 1115–1136. O’Brien, M.J., Lyman, R.L., Saab, Y., Saab, E., Darwent, J., Glover, D.S., 2002. Two issues in archaeological phylogenetics: taxon construction and outgroup selection. Journal of Theoretical Biology 215, 133–150. O’Donnell, K., Kistler, H.C., Cigelnik, E., Ploetz, R.C., 1998. Multiple evolutionary origins of the fungus causing Panama disease of banana: concordant evidence from nuclear and mitochondrial gene genealogies. Proceedings of the National Academy of Sciences of the United States of America 95, 2044–2049. Ohel, M.Y., 1987. The Acheulean handaxe: a maintainable multifunctional tool. Lithic Technology 16, 54–55.

574

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575

Ohnuma, K., 1995. Analysis of debitage pieces from experimentally reduced ‘classical Levallois’ and ‘discoidal’ cores. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 257–266. Otte, M., 2003. The pitfalls of using bifaces as cultural markers. In: Soressi, M., Dibble, H.L. (Eds.), Multiple Approaches to the Study of Bifacial Technologies. University of Pennsylvania, Philadelphia, pp. 183–192. Page, R.D.M., Holmes, E.C., 1998. Molecular Evolution: A Phylogenetic Approach. Blackwell Science, Oxford. Perneger, T.V., 1998. What’s wrong with Bonferroni adjustments. British Medical Journal 316, 1236–1238. Perry, S.E., 2006. What cultural primatology can tell anthropologists about the evolution of culture. Annual Review of Anthropology 35, 171–190. Petraglia, M.D., 1998. The Lower Palaeolithic of India and its bearing on the Asian record. In: Petraglia, M.D., Korisettar, R. (Eds.), Early Human Behaviour in Global Context: The Rise and Diversity of the Lower Palaeolithic Record. Routledge, London, pp. 343–390. Petraglia, M.D., 2007. The Indian Acheulian in global perspective. In: Goren-Inbar, N., Sharon, G. (Eds.), Axe Age: Acheulan Tool-making from Quarry to Discard. Equinox, London, pp. 389–414. Petraglia, M.D., Schuldenrein, J., Korisettar, R., 2003. Landscapes, activity, and the Acheulean to Middle Paleolithic transition in the Kaladgi Basin, India. Eurasian Prehistory 1, 3– 24. Pimentel, R., Riggins, R., 1987. The nature of cladistic data. Cladistics 3, 201–209. Porat, N., Chazan, M., Schwarcz, H., Horwitz, L.K., 2002. Timing of the Lower to Middle Paleolithic boundary: new dates from the Levant. Journal of Human Evolution 43, 107–122. Quicke, D.L.J., 1993. Principles and Techniques of Contemporary Taxonomy. Blackie Academic and Professional, London. Rae, T.C., 1997. The early evolution of the hominoid face. In: Begun, D.R., Ward, C.V., Rose, M.D. (Eds.), Function, Phylogeny, and Fossils: Miocene Hominoid Evolution and Adaptations. Plenum Press, New York, pp. 59–77. Rae, T.C., 1998. The logical basis for the use of continuous characters in phylogenetic systematics. Cladistics 14, 221–228. Rae, T.C., 2002. Scaling, polymorphism and cladistic analysis. In: (MacLeod, N., Forey, P.L. (Eds.), Morphology, Shape and Phylogeny. Taylor & Francis, London, pp. 45–52. Ranov, V.A., 2001. Cleavers: their distribution, chronology and typology. In: Milliken, S., Cook, J. (Eds.), A Very Remote Period Indeed: Papers on the Palaeolithic Presented to Derek Roe. Oxbow Books, Oxford, pp. 105–113. Richerson, P.J., Boyd, R., 2005. Not by Genes Alone: How Culture Transformed Human Evolution. University of Chicago Press, Chicago. Richter, S., 2005. Homologies in phylogenetic analyses—concept and tests. Theory in Biosciences 124, 105–120. Roe, D.A., 1976. Typology and the trouble with handaxes. In: de Sieveking, G., Longworth, I.H., Wilson, K.E. (Eds.), Problems in Economic and Social Archaeology. Duckworth, London, pp. 61–70. Roe, D.A., 1994. A metrical analysis of selected sets of handaxes and cleavers from Olduvai Gorge. In: Leakey, M.D., Roe, D.A. (Eds.), Olduvai Gorge: Volume 5. Cambridge University Press, Cambridge, pp. 146–234.

Roebroeks, W., van Kolfschoten, T. (Eds.), 1995. The Earliest Occupation of Europe. University of Leiden and European Science Foundation, Leiden. Rolland, N., 1995. Levallois technique emergence: single or multiple? A review of the Euro-African record. In: Dibble, H.L., Bar-Yosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 333–359. Sanderson, M.J., Hufford, L. (Eds.), 1996. Homoplasy: The Recurrence of Similarity in Evolution. Academic Press, New York. Schick, K.D., 1994. The Movius line reconsidered. In: Corruccini, R.S., Ciochon, R.L. (Eds.), Integrative Paths to the Past. Prentice Hall, Englewood Cliffs, NJ, pp. 569–596. Schick, K.D., 1998. A comparative perspective on Paleolithic cultural patterns. In: Akazawa, T., Aoki, K., Bar-Yosef, O. (Eds.), Neandertals and Modern Humans in Western Asia. Plenum Press, New York, pp. 449–460. Schick, K., Clark, J.D., 2003. Biface technological development and variability in the Acheulian industrial complex in the Middle Awash region of the Afar Rift, Ethiopia. In: Soressi, M., Dibble, H.L. (Eds.), Multiple Approaches to the Study of Bifacial Technologies. University of Pennsylvania, Philadelphia, pp. 1–30. Schick, K.D., Toth, N., 1993. Making Silent Stones Speak: Human Evolution and the Dawn of Human Technology. Weidenfeld and Nicolson, London. Schick, K.D., Zhuan, D., 1993. Early Paleolithic of China and Eastern Asia. Evolutionary Anthropology 2, 22–35. Schlanger, N., 1996. Understanding Levallois: lithic technology and cognitive archaeology. Cambridge Archaeological Journal 6, 231–254. Schuh, R.T., 2000. Biological Systematics: Principles and Applications. Cornell University Press, London. Shea, J.J., 2003. The Middle Paleolithic of the east Mediterranean Levant. Journal of World Prehistory 17, 313–394. Shennan, S., 2000. Population, culture history, and the dynamics of culture change. Current Anthropology 41, 811–835. Shennan, S., 2001. Demography and cultural innovation: a model and its implications for the emergence of modern human culture. Cambridge Archaeological Journal 11, 5–16. Shennan, S., 2002. Genes, Memes and Human History: Darwinian Archaeology and Cultural Evolution. Thames and Hudson, London. Shennan, S., Collard, M., 2005. Investigating processes of cultural evolution on the north coast of New Guinea with multivariate and cladistic analyses. In: Mace, R., Holden, C.J., Shennan, S. (Eds.), The Evolution of Cultural Diversity: A Phylogenetic Approach. UCL Press, London, pp. 133–164. Shennan, S.J., Steele, J., 1999. Cultural learning in hominids: a behavioural ecological approach. In: Box, H.O., Gibson, K.R. (Eds.), Mammalian Social Learning: Comparative and Ecological Perspectives. Cambridge University Press, Cambridge, pp. 367–388. Simon, C., 1983. A new coding procedure for morphometric data with an example from periodical cicada wing viens. In: Felsenstein, J. (Ed.), Numerical Taxonomy. Springer Verlag, Berlin, pp. 378–383. Simonovic, P.D., 1999. Phylogenetic relationships of Ponto– Caspian gobies and their relationship to the Atlantic–Mediterranean Gobiinae. Journal of Fish Biology 54, 533–555.

S.J. Lycett / Journal of Anthropological Archaeology 26 (2007) 541–575 Simpson, G.C., 1945. The principles of classification and a classification of mammals. Bulletin of the American Museum of Natural History 85, 1–350. Skelton, P., Smith, A., Monks, N., 2002. Cladistics: A Practical Primer on CD-Rom. Cambridge University Press, Cambridge. Slowinski, J.B., 1993. ‘Unordered’ versus ‘ordered’ characters. Systematic Biology 42, 155–165. Smith, A.B., 1994. Systematics and the Fossil Record. Blackwell Science, Oxford. Sober, E., 1983. Parsimony in systematics: philosophical issues. Annual Review of Ecology and Systematics 14, 335–357. Sokal, R.R., Rohlf, F.J., 1995. Biometry, third ed. W.H. Freeman and Co., New York. Spencer, M., Davidson, E.A., Barbrook, A.C., Howe, C.J., 2004. Phylogenetics of artificial manuscripts. Journal of Theoretical Biology 227, 503–511. Stewart, C.B., 1993. The powers and pitfalls of parsimony. Nature 361, 603–607. Strait, D.S., Grine, F.E., Moniz, M.A., 1997. A reappraisal of early hominid phylogeny. Journal of Human Evolution 32, 17–82. Strait, D.S., Grine, F.E., 2004. Inferring hominoid and early hominid phylogeny using craniodental characters: the role of fossil taxa. Journal of Human Evolution 47, 399–452. Swartz, B.K., 1980. Continental line-making: a reexamination of basic Palaeolithic classification. In: Leakey, R.E.F., Ogot, B.A. (Eds.), Proceedings of the 8th Congress of Prehistory and Quaternary Studies, Nairobi, 5–10th September 1977, The International Louis Leakey Memorial Institute for African Prehistory, Nairobi, pp. 33–35. Swiderski, D.L., Zelditch, M.L., Fink, W.L., 1998. Why morphometrics is not special: coding quantitative data for phylogenetic analysis. Systematic Biology 47, 508–519. Swofford, D.L., 1998. PAUP*: Phylogenetic Analysis Using Parsimony (* and other methods), version 4.0. Sinauer Associates, Sunderland, MA. Szalay, F.S., 1977. Ancestors, descendants, sister groups and testing phylogenetic hypotheses. Systematic Zoology 26, 12– 18. Tamrat, E., 1995. Revised magnetostratigraphy of the Plio– Pleistocene sedimentary sequence of the Olduvai Formation (Tanzania). Palaeogeography, Palaeoclimatology, Palaeoecology 114, 273–283. Tehrani, J., Collard, M., 2002. Investigating cultural evolution through biological phylogenetic analyses of Turkmen textiles. Journal of Anthropological Archaeology 21, 443–463. Terradas, X., 2003. Discoid flaking method: conception and technological variability. In: Peresani, M. (Ed.), Discoid Lithic Technology: Advances and Implications. British Archaeological Reports, Oxford, pp. 19–32. Thiele, K., 1993. The holy grail of the perfect character: the cladistic treatment of morphometric data. Cladistics 9, 275–304. Thorpe, R.S., 1984. Coding morphometric characters for constructing distance Wagner networks. Evolution 38, 244–255. Toth, N., 1985. The Oldowan reassessed: a close look at early stone artefacts. Journal of Archaeological Science 12, 101–120. Toth, N., 1987. Behavioral inferences from early stone artifact assemblages: an experimental model. Journal of Human Evolution 16, 763–787. Toth, N., Schick, K., 1993. Early stone industries and inferences regarding language and cognition. In: Gibson, K.R., Ingold, T. (Eds.), Tools, Language and Cognition in Human Evolution. Cambridge University Press, Cambridge, pp. 346–362.

575

Tryon, C.A., 2006. ‘Early’ Middle Stone Age lithic technology of the Kapthurin Formation (Kenya). Current Anthropology 47, 367–375. Tryon, C.A., McBrearty, S., 2002. Tephrostratigraphy and the Acheulian to Middle Stone Age transition in the Kapthurin formation, Kenya. Journal of Human Evolution 42, 211–235. Tryon, C.A., McBrearty, S., Texier, J.-P., 2006. Levallois lithic technology from the Kapthurin Formation, Kenya: Acheulian origin and Middle Stone Age diversity. African Archaeological Review 22, 199–229. Tuffreau, A., 1995. The variability of Levallois technology in northern France and neighboring areas. In: Dibble, H.L., BarYosef, O. (Eds.), The Definition and Interpretation of Levallois Technology. Prehistory Press, Madison, Wisconsin, pp. 413–427. Tuffreau, A., Antoine, P., 1995. The earliest occupation of Europe: continental northwestern Europe. In: Roebroeks, W., van Kolfschoten, T. (Eds.), The Earliest Occuptation of Europe. University of Leiden and European Science Foundation, Leiden, pp. 147–163. van Peer, P., 1992. The Levallois Reduction Strategy. Prehistory Press, Madison, Wisconsin. Vaquero, M., Carbonell, E., 2003. A temporal perspective on the variability of the discoid method in the Iberian Peninsula. In: Peresani, M. (Ed.), Discoid Lithic Technology: Advances and Implications. British Archaeological Reports, Oxford, pp. 67– 82. van Riet Lowe, C., 1945. The evolution of the Levallois technique in South Africa. Man 45, 49–59. van Schaik, C.P., Ancrenaz, M., Borgen, G., Galdikas, B., Knott, C.D., Singleton, I., Suzuki, A., Utami, S.S., Merrill, M., 2003. Orangutan cultures and the evolution of material culture. Science 299, 102–105. Villa, P., 2001. Early Italy and the colonization of western Europe. Quaternary International 75, 113–130. Weston, P.H., 1994. Methods for rooting cladistic trees. In: Scotland, R.W., Siebert, D.J., Williams, D.M. (Eds.), Models in Phylogeny Reconstruction. Clarendon Press, Oxford, pp. 125–155. Wheeler, Q.D., Meier, R. (Eds.), 2000. Species Concepts and Phylogenetic Theory: A Debate. Columbia University Press, New York. White, M., Scott, B., Ashton, N., 2006. The Early Middle Palaeolithic in Britain: archaeology, settlement history and human behaviour. Journal of Quaternary Science 21, 525–541. Whiten, A., 2005. The second inheritance system of chimpanzees and humans. Nature 437, 52–55. Whiten, A., Horner, V., de Waal, F.B.M., 2005. Conformity to cultural norms of tool use in chimpanzees. Nature 437, 737–740. Wiley, E.O., Siegel-Causey, D., Brooks, D.R., Funk, V.A., 1991. The Compleat Cladist: A Primer of Phylogenetic Procedures. The University of Kansas, Museum of Natural History, Lawrence, Kansas. Wynn, T., 1981. The intelligence of Oldowan hominids. Journal of Human Evolution 10, 529–541. Wynn, T., 1995. Handaxe enigmas. World Archaeology 27, 10–24. Wynn, T., Coolidge, F.L., 2004. The expert Neandertal mind. Journal of Human Evolution 46, 467–487. Yi, S., Clark, G.A., 1983. Observations on the Lower Palaeolithic of Northeast Asia. Current Anthropology 24, 181–202. Young, N.M., 2003. A reassessment of living hominoid postcranial variability: implications for ape evolution. Journal of Human Evolution 45, 441–464.