EARLY BRONZE AGE CRETE: REGIONAL DIVERSITY AND CYCLADICA

It is widely recognised that, during the third millennium bc, Crete was characterised by distinct cultural diversity. This has cast serious doubts upon the traditional assumption ‘that the island was inhabited by a single population with a more or less homogeneous culture’ (see discussion and relevant bibliography in Legarra Herrero Reference Legarra Herrero2009, 29, 49). In contrast, significant variations in ceramic production (e.g. Wilson and Day Reference Wilson and Day1994; Day, Wilson and Kiriatzi Reference Day, Wilson, Kiriatzi, Laffineur and Betancourt1997; Whitelaw et al. Reference Whitelaw, Day, Kiriatzi, Kilikoglou, Wilson, Laffineur and Betancourt1997), stylistic traits, values of consumption, circulation of materials and finished artefacts (e.g. Carter Reference Carter and Branigan1998; Bevan Reference Bevan, Barrett and Halstead2004; Doonan, Day and Dimopoulou-Rethemiotaki Reference Doonan, Day, Dimopoulou-Rethemiotaki, Day and Doonan2007; Sbonias Reference Sbonias, Schoep, Tomkins and Driessen2012), and the funerary record (especially tomb architecture, quantity and quality of the associated grave goods and the manipulation of the deceased within the island: e.g. Soles Reference Soles1992; Vavouranakis Reference Vavouranakis2007; Papadatos and Sofianou Reference Papadatos, Sofianou, Andrianakis, Varthalitou and Tzachili2012; Legarra Herrero Reference Legarra Herrero2009; Reference Legarra Herrero2014; Papadatos and Sofianou in press) reinforce the idea that communities in Prepalatial Crete, located sometimes in close proximity, developed different socio-economic and ideological trajectories.

Special emphasis has been placed, to date, upon relationships with other parts of the southern Aegean and in particular with the Cyclades. During the early and mid third millennium bc, places in a large area of the southern Aegean, including Crete (e.g. Hagia Photia Siteias: Davaras and Betancourt Reference Davaras and Betancourt2004; Reference Davaras and Betancourt2012; Gournes Pediados: Galanaki Reference Galanaki2006); Attica (e.g. Agios Kosmas: Mylonas Reference Mylonas1959; Tsepi: Pantelidou-Gofa Reference Pantelidou-Gofa2005), Euboea (e.g. Manika: Sampson Reference Sampson1988) and to a lesser degree the Anatolian coast (e.g. Iasos cemetery: Levi Reference Levi1961–2; Reference Levi1965–6; Pecorella Reference Pecorella1984), exhibit close cultural affinity: this is demonstrated in terms both of material culture (e.g. portable artefacts and grave forms) and of ideas and technological knowledge (Renfrew Reference Renfrew1972; Broodbank Reference Broodbank2000; Papadatos Reference Papadatos2005; Brodie et al. Reference Brodie, Doole, Gavalas and Renfrew2008; Wilson, Day and Dimopoulou-Rethemiotaki Reference Wilson, Day, Dimopoulou-Rethemiotaki, Brodie, Doole, Gavalas and Renfrew2008; Cherry Reference Cherry, Parkinson and Galaty2009; Papadatos and Tomkins Reference Papadatos and Tomkins2013). The impact of the Early Bronze Age Cycladic tradition in this area is materially expressed through marble figurines with folded arms (Branigan Reference Branigan1971; Getz-Preziosi Reference Getz-Preziosi1987; Papadatos Reference Papadatos1999), metal objects (especially daggers with raised mid-rib made of Cycladic raw materials: Branigan Reference Branigan1974; Stos-Gale Reference Stos-Gale, Scarre and Healey1993; Doonan and Day Reference Doonan, Day, Day and Doonan2007), a specific type of pottery known as the ‘Kampos Group’ style (Day, Wilson and Kiriatzi Reference Day, Wilson, Kiriatzi and Branigan1998; Wilson, Day and Dimopoulou Reference Wilson, Day, Dimopoulou, Cadogan, Hatzaki and Vasilakis2004; Day et al. Reference Day, Hein, Joyner, Kilikoglou, Kiriatzi, Tsolakidou, Wilson, Davaras and Betancourt2012; Davaras and Betancourt Reference Davaras and Betancourt2012; Nodarou 2012; Tsipopoulou Reference Tsipopoulou, Mantzourani and Betancourt2012b), the so-called ‘frying pans’ with symbolic cosmological and marine representations (Betancourt Reference Betancourt, Mantzourani and Betancourt2012), long blades of Melian obsidian (Carter Reference Carter and Branigan1998) and other artefacts.

The manner in which Cycladic-related items (hereafter Cycladica), ideas and practices were transferred from the core islands of the Archipelago has provoked considerable debate. The main topics of interest include the mode of transmission of objects, people and ideas around the southern Aegean during the third millennium bc, and the character and quantity of Cycladica, particularly since these objects frequently accompany funerary assemblages. There are two opposing interpretations of this phenomenon: Cycladic colonists settled outside their core area and transferred pottery, technical equipment, ideas, practices and technological knowledge from their original home area to their new settlements (Sakellarakis Reference Sakellarakis and Thimme1977a; Reference Sakellarakis1977b; Doumas Reference Doumas1979; Sapouna-Sakellaraki Reference Sapouna-Sakellaraki1987); alternatively, individuals and local populations adopted Cycladic imports or artefacts of Cycladic character in order to shape and probably project different identities for themselves and their local communities (see recent discussion in Papadatos Reference Papadatos, Antoniadou and Pace2007). Current approaches support the idea that a combination of the two processes took place: locals adopted Cycladic objects and traditions while Cycladic groups moved to Crete, in the context of trading networks and intermarriage, particularly during the mid-third millennium bc (Broodbank Reference Broodbank2000).

In order for the above issues to be addressed effectively, there is a need for research into cultural interchange between Prepalatial Crete and the Cyclades in the light of human mobility patterns and interbreeding practices. It is possible that analysis of the nature and scale of such patterns and practices may offer new perspectives on the role that the local populations played in cultural, material and technological interaction among the different regions of Prepalatial Crete.

THE CASE STUDY FUNERARY ASSEMBLAGES

The evidence of funerary assemblages demonstrates the integration of Crete with the wider southern Aegean. In contrast to the majority of studies to date, which have focused upon items of material culture, the current paper will focus on the people who lived in eastern Crete during the third and early second millennia bc. The wider region of eastern Crete was selected as a case study area on the basis of recently excavated and partly, or completely, studied mortuary assemblages with human skeletal remains. This material demonstrates: (a) temporal proximity, (b) cultural similarities that may suggest participation in common networks of interaction, and (c) cultural differences that may imply biological differences between the corresponding population groups.

Crete, particularly the north coast, exhibits a relatively large series of mortuary contexts manifesting Cycladic-related materials: among the most recently excavated assemblages (Fig. 1) are the rockshelter at Kephala Petras, near Siteia (Tsipopoulou Reference Tsipopoulou, Andrianakis, Varthalitou and Tzachili2010; Reference Tsipopoulou, Mantzourani and Betancourt2012b; Reference Tsipopoulou and Tsipopoulou2012c) and the tholos tomb and rockshelter at Livari-Skiadi in south-east Crete (Papadatos and Sofianou Reference Papadatos, Sofianou, Andrianakis, Varthalitou and Tzachili2012; in press). The Kephala Petras cemetery lies on the flat plateau area of a high hill, less than 1 km east of the modern town of Siteia, overlooking Siteia bay. On the west slope of the cemetery there is a burial rockshelter which faces the Protopalatial palace and the surrounding Minoan town; its use extends from Early Minoan IB to Middle Minoan IB, c.2900–1875/1850 bc (Tsipopoulou Reference Tsipopoulou, Andrianakis, Varthalitou and Tzachili2010; Reference Tsipopoulou and Tsipopoulou2012c; for absolute chronology see Manning Reference Manning and Cline2010, table 2.2). Moreover, systematic excavation since 2004 has recovered 11 house tombs with multiple rooms, special open areas and platforms for ritual activities in memory of the deceased. The use of the house tombs extends from the mid-third millennium bc to the early second millennium bc, c.2400–1850 bc (Early Minoan IIB – Middle Minoan IIA) (Manning Reference Manning1995, fig. 2; 2010, table 2.2), with peak use in the early Protopalatial period (Middle Minoan IB–IIA) (Tsipopoulou Reference Tsipopoulou, Andrianakis, Varthalitou and Tzachili2012a; Reference Tsipopoulou and Tsipopoulou2012c). The current analysis focuses on House Tomb 2, the only fully excavated mortuary complex of the cemetery, which spans the period from Early Minoan III to Middle Minoan IB/IIA, c.2200–1875/1850. There is, therefore, an overlapping period of approximately 350 years when the burial rockshelter and House Tomb 2 were in contemporary use. The Livari-Skiadi tholos tomb and rockshelter are situated on a small coastal plain opposite the islet of Kouphonisi, about 50 m from the seashore. The plain is cut by deep gorges which run from the inland Ziros plateau, while a rocky promontory protrudes into the sea, forming a small bay in front of the cemetery (Papadatos and Sofianou Reference Papadatos, Sofianou, Andrianakis, Varthalitou and Tzachili2012; in press). The tholos tomb was in contemporary use with the rockshelter from Early Minoan IB to Early Minoan IIB (Papadatos and Sofianou in press), c.2900–2200 bc (Manning Reference Manning1995, fig. 2; 2010, table 2.2), while the rockshelter seems to have been in continual use until Early Minoan III, c.2200–2100/2050 bc, that is, almost one or two centuries after the last use of the tholos tomb.

Fig. 1.

Map indicating the location of the study areas (drawn by N. Valasiadis). Key: 1 = Kephala Petras, 2 = Livari-Skiadi.

Funerary assemblages from both the Kephala Petras rockshelter and the Livari-Skiadi tholos tomb and rockshelter include various types of artefacts which follow well-known local and non-local traditions from other regions across the island: the pottery may be paralleled with material from the Mesara-Asterousia area (Papadatos and Sofianou in press), and there are obsidian blades (Carter in press) and metal objects which ‘seem to be Cretan in origin’ (Ferrence, Muhly and Betancourt. Reference Ferrence, Muhly, Betancourt and Tsipopoulou2012, 140); significantly, however, they include also a number of Cycladica. The latter consist of various pottery artefacts which are classified as part of the so-called ‘Kampos Group’ (Nodarou 2012; in press, 85; Papadatos and Sofianou Reference Papadatos, Sofianou, Andrianakis, Varthalitou and Tzachili2012, 51; Tsipopoulou Reference Tsipopoulou, Andrianakis, Varthalitou and Tzachili2012a; Reference Tsipopoulou, Mantzourani and Betancourt2012b), mid-rib copper daggers (Papadatos and Sofianou in press) and four fragments of marble figurines with folded arms (Tsipopoulou Reference Tsipopoulou, Andrianakis, Varthalitou and Tzachili2012a; Reference Tsipopoulou, Mantzourani and Betancourt2012b). It is interesting to note that the recent petrographic analysis of 45 samples belonging to the ‘Kampos Group’ pottery from the Kephala Petras rockshelter identified a highly calcite-tempered fabric (Nodarou 2012, 85), which is a common Cycladic manufacturing tradition found also at sites on the north coast of Crete which offer similarly strong evidence for off-island contacts with the Aegean; these sites include Hagia Photia, Gournes and Poros-Katsambas (Hagia Photia: Day, Wilson and Kiriatzi Reference Day, Wilson, Kiriatzi and Branigan1998; Day et al. Reference Day, Wilson, Kyriatzi and Joyner2000; Gournes: Galanaki Reference Galanaki2006; Poros-Katsambas: Dimopoulou-Rethemiotaki, Wilson and Day Reference Dimopoulou-Rethemiotaki, Wilson, Day, Day and Doonan2007).

To sum up, in three (Kephala Petras rockshelter, Livari-Skiadi tholos tomb and rockshelter) out of the four (Kephala Petras rockshelter and House Tomb 2, Livari-Skiadi tholos tomb and rockshelter) mortuary assemblages investigated in this study, there is significant evidence for off-island relations with the Cyclades. These coincide with the critical time span of Early Minoan IB–IIA (c.2900–2450/2400 based on Manning Reference Manning and Cline2010, table 2.2), that is, the peak time of interregional cultural contacts in the southern Aegean, described in the literature by Renfrew (Reference Renfrew1972) as the ‘international spirit’ phenomenon. Meanwhile, all case study sites seem to have been engaged in diverse cultural networks within Crete. Moreover, regarding temporal proximity, in the Kephala Petras and Livari rockshelters as well as in the Livari tholos tomb, there is an overlapping period of approximately 700 years of use which extends from Early Minoan IB to Early Minoan IIB, c.2900–2200 bc. In contrast, Kephala Petras House Tomb 2 is of a slightly later date, c.2200–1850 bc, partially overlapping with the later period of use of the Kephala Petras rockshelter, and has provided no evidence of Cycladica. It will be of interest therefore to explore the biological connections among these assemblages and to assess the degree to which the observed cultural similarities or diversities can be explained by means of population relocations and exogamy.

THE PEOPLE REPRESENTED: RESULTS OF THE MACROSCOPIC STUDY

The macroscopic analysis of the skeletal populations disposed of in the Kephala Petras rockshelter and House Tomb 2 (HT 2), as well as in the Livari-Skiadi tholos tomb and rockshelter (Table 1), has offered valuable evidence with regard to the manipulation of the deceased and the demographic composition of each assemblage. Disarticulated skeletal remains and major anatomical units are consistent with the deposition of secondary burials of either skeletonised or fresh bone in both rockshelters, which appear to represent one-off episodes of systematic redeposition of burials placed initially in a different disposal area (e.g. Triantaphyllou Reference Triantaphyllou2009; Reference Triantaphyllou and Tsipopoulou2012; in press; Triantaphyllou, Tsipopoulou and Betancourt in press). The manipulation of the deceased in the Kephala Petras House Tomb 2 differs slightly since, along with the secondary skeletal remains, there are a number of burials which appear to lie in their initial disposal area but manifest clear evidence of the removal of body parts shortly after death (for a synthetic discussion on secondary removal of body parts during the Pre- and Protopalatial periods, see Triantaphyllou in press; Triantaphyllou, Tsipopoulou and Betancourt in press). Finally, in the Livari-Skiadi tholos tomb bone alterations due to burning (e.g. severe cracking and warping) suggest a lengthy and systematic exposure of the body parts to high temperature as part of the secondary treatment of the deceased (Triantaphyllou in press).

Table 1.

Sample size of the populations under study.

Assessment of the population groups deposited inside the communal tombs has occupied archaeologists working on Early Bronze Age Crete since the earliest stage of research (e.g. Branigan Reference Branigan, Kastrinaki, Orfanou and Giannadakis1987b; Reference Branigan1993, 84–95; Papadatos Reference Papadatos1999, 59–63; Vasilakis and Branigan Reference Vasilakis and Branigan2010, 262 and more recently Triantaphyllou in press). Analysis has been based upon (a) survey work and the association of the tombs with nearby settlements (Bintliff Reference Bintliff1977; Blackman and Branigan Reference Blackman and Branigan1977; Branigan Reference Branigan1993; Vasilakis and Branigan Reference Vasilakis and Branigan2010; Whitelaw Reference Whitelaw, Schoep, Tomkins and Driessen2012), (b) the internal social organisation of the fully excavated Early Minoan settlement at Myrtos Fournou Korifi based on the spatial distribution of artefacts across the site and the functional interpretation of rooms, built facilities and finds (Whitelaw Reference Whitelaw, Krzyszkowska and Nixon1983 vs Warren Reference Warren1972 and more recently Whitelaw Reference Whitelaw, Westgate, Fisher and Whitley2007), (c) the number of the skulls recognised during excavation (particularly with regard to the house tombs: Soles Reference Soles1992, 252), and (d) the study of artefacts and their distribution within the tombs (Whitelaw Reference Whitelaw, Krzyszkowska and Nixon1983, fig.71). Recent analysis of skeletal remains has shed some new light on issues concerning the group size and composition of the people deposited in Prepalatial and early Protopalatial funerary assemblages (Table 2). Table 2 demonstrates that the low estimates of skeletal remains deposited per 100 years do not approach the hundreds of individuals commonly assumed (cf. e.g. Branigan Reference Branigan, Kastrinaki, Orfanou and Giannadakis1987b). In contrast, assuming that a household of about five to seven individuals contributes approximately 20 corpses per century – a figure based on ethnographic observations of agricultural communities but also on historical studies of census data from rural communities (for specific references see Chamberlain Reference Chamberlain2006, 52; for an application of population estimates in prehistoric Aegean contexts, see Bintliff Reference Bintliff1977, 83) – Table 2 suggests that the nuclear family was perhaps the basic population unit in Early Bronze Age communities on Crete, as has been convincingly demonstrated by Whitelaw (Reference Whitelaw, Krzyszkowska and Nixon1983; Reference Whitelaw, Westgate, Fisher and Whitley2007). Analysis of household models from small rural communities in Early Bronze Age Crete would sustain the idea of independent households represented by nuclear families which appear to have been bound together with close ties (Whitelaw Reference Whitelaw, Westgate, Fisher and Whitley2007, 73). In a recent study Whitelaw considered new data derived from archaeological surveys in the Mirabello area; combining this with his early work at Myrtos Fournou Korifi, he proposed a notional model of Early Minoan II site distributions and local demographic networks (Whitelaw Reference Whitelaw, Macdonald, Hatzaki and Andreou2015, fig. 4). The results of this study suggested that small hamlets located particularly in a very dry landscape of thin soils would ‘have to have maintained a wide network of connections with comparable communities to be demographically viable’ (Whitelaw Reference Whitelaw, Macdonald, Hatzaki and Andreou2015, 2).

Table 2.

Estimation of skeletons deposited per 100 years from recently examined skeletal assemblages on Crete (key: HT = House Tomb, MNI = minimum number of individuals).

Both sexes and all age groups were equally represented in the skeletal assemblages (Table 3). The Livari-Skiadi tholos tomb, however, deviates from this pattern; the demographic composition of this skeletal assemblage suggests a significant under-representation of the sub-adult age categories since only six sub-adult, as against seventy-five adult individuals, were identified (Triantaphyllou in press; forthcoming). It is evident, therefore, that the manipulation of the deceased in the third and early second millennia bc in eastern Crete is dominated by the communal character of deposition, but with significant deviations with regard to the disposal area, the treatment of the deceased and perhaps segregation according to age groups. It is equally important to point out here that, given population estimates from recently studied skeletal assemblages (Table 2) and systematic work conducted by Whitelaw (Reference Whitelaw, Westgate, Fisher and Whitley2007; Reference Whitelaw, Macdonald, Hatzaki and Andreou2015), further investigation of kinship connections and of the scale of exogamous networks between the small and fragmented Prepalatial communities of eastern Crete will be of great value to the recognition of social and developing political entities in the third millennium bc.

Table 3.

Distribution of age groups in the case study skeletal assemblages (key: HT = House Tomb).

STRONTIUM ISOTOPE ANALYSIS: GEOLOGICAL SETTING AND METHODS

Isotope work in the Aegean, with a particular focus on human mobility, was introduced recently by Nafplioti (Nafplioti Reference Nafplioti2007; Reference Nafplioti2008; Reference Nafplioti2009a; Reference Nafplioti2009b; Reference Nafplioti, Sampson, Kaczanowska and Kozlowski2010; Reference Nafplioti2011) and Vika (Vika Reference Vika2009) and was combined in one case with the study of cranial non-metric traits (Nafplioti Reference Nafplioti2007). Moreover, Nafplioti has created a preliminary map of the ⁸⁷Sr/⁸⁶Sr values in different geological and geographic settings in the Aegean as measured not only from archaeological human and animal dental enamel and bone material, but also from modern snail shells (Nafplioti Reference Nafplioti2011); however, the sample sizes upon which this map was created were very limited, so further work is required in this direction.

Geologically a major part of the island of Crete belongs to the Gavrovo isopic zone, while certain areas in the central and central-western part of the island fall within the Ionian zone (Higgins and Higgins Reference Higgins and Higgins1996, fig. 2.2). The Gavrovo zone, which includes a narrow strip of Albania and central Greece, widens out in the Peloponnese. This isopic zone was a continental fragment for the early part of its history (Higgins and Higgins Reference Higgins and Higgins1996, 19). The two study areas, Petras and Livari, are situated on the north and south coast, respectively, of the eastern part of Crete which belongs to the Gavrovo zone (Higgins and Higgins Reference Higgins and Higgins1996, fig. 2.2). However, these two areas are located in different geological formations. Petras is situated on a combination of Miocene marine sediments and exposures of older (Permian–Triassic) phyllites (Papastamatiou et al. Reference Papastamatiou, Vetoulis, Bornovas, Christodoulou and Tataris1959b). Livari-Skiadi is at the junction of two geological units which consist, to the north-north-east, of Jurassic to Eocene thick-bedded to massive grey limestones and dolomites (Gavrovo-Tripolitza series) and the sandstone conglomerate of the Tripolis flysch, whereas the area to the south-south-west consists of Miocene continental sediments (Papastamatiou et al. Reference Papastamatiou, Vetoulis, Bornovas, Christodoulou and Katsikatsos1959a; Creutzburg, Drooger and Meulenkamp Reference Creutzburg, Drooger and Meulenkamp1977).

Owing both to the disarticulated state of the remains from the two rockshelters and to budgetary limitation, sampling for strontium isotope analysis could not include postcranial bone and teeth from the same individuals; this would have allowed more accurate assessment of diagenetic contamination and of the local range of isotopic values. The sampling strategy for strontium isotope analysis is based on the assumption that migrant individuals who moved between different geologic regions will demonstrate different ⁸⁷Sr/⁸⁶Sr signatures between the teeth and bones; this is a reasonable assumption given that adult tooth enamel will capture the ⁸⁷Sr/⁸⁶Sr values of their youth, whereas bones, which remodel throughout life, will represent the ⁸⁷Sr/⁸⁶Sr values of the environment where the person spent his or her last years (Sealy, Armstrong and Schrire Reference Sealy, Armstrong and Schrire1995; Bentley Reference Bentley2006; see, however, the comments on diagenetic alteration below). In addition to the 18 tooth enamel samples extracted from the study skeletal assemblages, seven samples of dentine and one of maxillary bone were also analysed (Table 4). Analysis on the same individual was avoided by selecting teeth of the same type and side. Analysis of the strontium isotope ratios was conducted at the Bristol Isotope Group (BIG) Laboratory, School of Earth Sciences, University of Bristol.Footnote ¹ The strontium isotope ratio (⁸⁷Sr/⁸⁶Sr) in human tooth enamel, after the development of permanent dentition, reflects dietary strontium derived from soils and ground water via the food chain and undergoes relatively little change after it has mineralised in childhood and early adolescence (Hillson Reference Hillson1996; Bentley Reference Bentley2006, 169–74; Haak et al. Reference Haak, Brandt, Jong, Meyer, Ganslmeier, Heyd, Hawkesworth, Pike, Meller and Alt2008, 18229). Moreover, it is considered highly resistant to post-mortem diagenesis (Hillson Reference Hillson1996; Budd et al. Reference Budd, Montgomery, Barreiro and Thomas2000, 688). In contrast, the strontium ratios in other skeletal tissues, such as bone and dentine, are known to be subject to diagenetic change after burial. Consequently, it is likely that the ⁸⁷Sr/⁸⁶Sr ratio in these tissues will be heavily but variably modified by post-depositional strontium uptake and therefore give a closer indication of the local burial environment rather than the place of origin of the individual (Budd et al. Reference Budd, Montgomery, Barreiro and Thomas2000, 688; Haak et al. Reference Haak, Brandt, Jong, Meyer, Ganslmeier, Heyd, Hawkesworth, Pike, Meller and Alt2008, 18229). However, this is only an indication, depending on the extent to which the strontium originally in these tissues has been equilibriated with the local environment. Therefore, the measurements of strontium in dentine and bone cannot be taken as direct evidence for the values of the burial environment and may be different from the bioavailable strontium in the wider area (Budd et al. Reference Budd, Montgomery, Barreiro and Thomas2000). Nonetheless, in the absence of other biological samples (see below) it was felt that the measurement of some dentine and jaw bone samples would at least provide an indication/approximation of the ⁸⁷Sr/⁸⁶Sr ratio of the burial environment. However, ideally, analysis of human bone and teeth should be applied in parallel to analysis of biosphere samples (waters, soils, modern plants), animal teeth, bones and archaeobotanical material in order to map more accurately the local range of isotopic values in a particular area (Bentley Reference Bentley2006, 136; Evans et al. Reference Evans, Montgomery, Wildman and Boulton2010; Frei and Price Reference Frei and Price2012).

DENTAL NON-METRIC TRAITS

The biological affinity among the study population samples was assessed by means of dental non-metric traits. Non-metric traits are minor variants of skeletal and dental anatomy which are not metrically measurable and are not a sign of disease (Tyrrell Reference Tyrell, Cox and Mays2000). Many twin and family studies (e.g. Scott Reference Scott1973; Berry Reference Berry, Butler and Joysey1978; Nichol Reference Nichol1990; Townsend et al. Reference Townsend, Richards, Brown, Burgess, Travan, Rogers, Smith and Tchernov1992; Scott and Turner Reference Scott and Turner1997; Schnutenhaus and Rösing Reference Schnutenhaus, Rösing, Alt, Rösing and Teschler-Nicola1998) suggest that non-metric dental variants are inherited polygenically, while the natural environment in which growth takes place also affects trait expression. It should, however, be noted that while environmental factors influence trait expression to some extent, they have not been shown to significantly affect population trait frequencies (Scott and Turner Reference Scott and Turner1997). Accordingly, non-metric traits have been used extensively in reconstructing patterns of intra- and inter-population affiliations spanning wide geographical areas and long periods of time (e.g. Corruccini and Shimada Reference Corruccini and Shimada2002; Tomczak and Powell Reference Tomczak and Powell2003; Irish Reference Irish2006; Coppa et al. Reference Coppa, Cucina, Lucci, Mancinelli and Vargiu2007; Hanihara Reference Hanihara2010; Nikita, Mattingly and Lahr Reference Nikita, Mattingly and Lahr2012).

Each non-metric trait was scored on one tooth within a class (Nichol Reference Nichol1990). The general practice is to score both the right and left side and use the highest degree of expression of a variant for population comparisons (Turner, Nichol and Scott Reference Turner, Nichol, Scott, Kelley and Larsen1991). Given that the sample under study consisted of loose and commingled teeth, in order to avoid biases we recorded non-metric traits on the maxillary and mandibular teeth that belonged only to the right side, even though this may not have been the side with the greatest degree of trait expression.

The fact that most of the material was preserved as loose teeth, disassociated from any cranial or postcranial elements, should not have any impact on the obtained results given that the mean measure of divergence (MMD), the distance measure employed in the present study (see below for details), is calculated on the overall frequency of each trait in the sample. In respect to levels of sexual dimorphism in trait expression and the potential bias of pooling sexes in the current analyses, sex differences in dental non-metric traits have been found to be low, and pooling data on males and females constitutes standard procedure (Turner, Nichol and Scott Reference Turner, Nichol, Scott, Kelley and Larsen1991; Scott and Turner Reference Scott and Turner1997).

The ordinal scores for each trait were transformed into dichotomised presence/absence values based on each trait's morphological threshold as determined by Turner (Reference Turner1987) and Irish (Reference Irish2006). Traits that exhibited very low frequency (at least one sample of <2 cases) were removed from the dataset, as were traits that did not exhibit at least one statistically significant difference between sample pairs (Harris and Sjøvold Reference Harris and Sjøvold2004). Moreover, significantly intercorrelated traits were also dropped from the dataset (Sjøvold Reference Sjøvold1977). The non-discriminatory and intercorrelated traits were identified by means of Chi-square and Fisher's exact tests. Table 5 shows the original dataset and the traits that remained after editing.

Smith's MMD was calculated using the remaining traits and the Freeman–Tukey angular transformation was employed in order to correct for small sample sizes (Freeman and Tukey Reference Freeman and Tukey1950; Sjøvold Reference Sjøvold1973; Reference Sjøvold1977). Although the MMD has occasionally been criticised, it has been found to perform very well and to offer results that are in accordance with the generalised Mahalanobis D² distance for non-metric traits (Irish Reference Irish2010). The p-values for each MMD were estimated using a method that is based on the fact that the MMD values follow the normal distribution (Sjøvold Reference Sjøvold1973; Green, Suchey and Gokhale Reference Green, Suchey and Gokhale1979).