A POPULATION AFFINITY STUDY USING
DENTAL MORPHOLOGICAL DATA
COLLECTED FROM THE ANCIENT CITY OF LEPTIMINUS
Cherie K. Walth
B.A., California State University at San Jose, 1990
B.S., University of Phoenix, 1981
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
This thesis for the Master of Anthropology
Cherie K. Walth
has been approved
David P. Tracer
Walth, Cherie K. (M.A., Anthropology)
A Population Affinity Study Using Dental Morphological Data Collected From the
Ancient City of Leptiminus
Thesis directed by Assistant Professor Mark Spencer
This study is part of a multi-disciplinary investigation being conducted at Leptiminus,
Tunisia to help elucidate the social, cultural, political, economic, and biological
relationships of this town to the contemporary Mediterranean region. This study
analyzes dental morphological data gathered from human remains excavated from two
Roman period Leptiminus cemeteries. The use of dental morphological data for
population affinity studies provides information on the origins, relationships,
migrations, and admixtures of populations living and extinct. The frequency of
occurrence of these traits provides a means of estimating biological affinity. The
dental morphological traits gathered from Leptiminus are statistically compared to
populations from Italy, Egypt, Nubia, Carthage, and southern Levant. The results
show that the Carthaginian samples (Carthage and Leptiminus) are phenetically similar
supporting the hypothesis that the Carthaginians are a biologically distinct population.
Phenetic similarity is also seen between one of the two Leptiminus cemeteries and a
sample from near Naples, Italy providing support for the assertion that Italian settlers
were installed at Leptiminus. The phenetic similarity between two Italian samples and
two non-Italian samples provides support for the biological influence of Roman
expansion in the Mediterranean. The isolation by distance hypothesis, based on the
premise that affinity can be predicted on the geographic proximity between groups,
was not fully supported. 'With samples that were within 1500 km of each other, there
was a moderate correlation between distance and phenetic similarities based on the
dental traits. Studies such as this aid in the understanding of the variability of human
populations by tracing historic relationships.
This abstract accurately represents the content of the candidates thesis. I recommend
Signed _ _____________
My thanks to the directors of the Leptiminus Project; Dr. John Humphrey, Dr. David
Mattingly, Dr. Lea Stirling, and Dr. David Stone, for allowing me to be a part of this
project. I also wish tothank Dr. Robert Jurmain for getting me started in Physical
Anthropology and for sharing with me his expertise in this field.
1. INTRODUCTION ............................................... 1
2. THE SCOPE OF DENTAL MORPHOLOGICAL TRAIT ANALYSIS 3
Non-metric Traits ........................................3
Evolutionary Origins of Human Tooth Form ................5
Non-genetic Influences on Dental Form in Humans...........7
Inheritance of Dental Variation in Humans.................8
Evolutionary Changes in Dental Form in Humans ............9
Genetic Drift .................................... 10
Natural Selection ................................ 11
Gene Flow ........................................ 12
Previous Affinity Studies Using Dental Traits .......... 13
Advantages and Disadvantages to Using Dental Trait Analysis ... 15
3. BACKGROUND OF LEPTIMINUS .................................. 18
The Leptiminus Project ................................. 18
The Carthaginians .......................................21
4. GOALS OF THE ANALYSIS ......................................25
Hypotheses to be Tested .................................27
Biological Distinctiveness of the Carthaginians....27
Biological Influence by Roman Expansion ..............27
Isolation by Distance ................................29
5. MATERIALS AND METHODS ..........................................30
The Leptiminus Samples ...............................31
Samples from Published Sources .......................31
Dental Morphological Traits .................................34
Traits on Maxillary Dentition ........................34
Traits on Mandibular Dentition .......................40
Scoring Procedures ..........................................45
6. RESULTS AND DISCUSSION .........................................56
Descriptive Analysis ........................................56
Mean Measure of Divergence Analysis .........................57
Comparison of MMD Results.............................62
Visual Statistical Methods ..................................64
Cluster Analysis .....................................64
Multidimensional Scaling .............................68
Discussion of Results........................................72
Carthaginian Biological Distinctiveness ..............74
Biological Influence by Roman Expansion ..............76
Isolation by Distance.................................81
7. CONCLUSIONS AND SUMMARY ........................................86
Hypothesis #1: Biological Distinctiveness of the Carthaginians .. 86
Hypothesis #2: Biological Influence by Roman Expansion ......86
Hypothesis #3: Isolation by Distance ........................87
A. RAW DATA...................................................89
Raw Data for Leptiminus Samples ........................89
Maxillary Traits .................................89
Frequency Data for All Samples..........................97
Maxillary Traits .................................97
Mandibular Traits............................... 101
REFERENCES ....................................................... 104
2.1 Project Area Location ..................................... 19
2.2 Location of Cemeteries at Leptiminus .......................20
5.1 Geographic Location of Samples .............................33
6.1 Cluster Dendrogram with the First Sample Set................65
6.2 Cluster Dendrogram with the Second Sample Set ..............66
6.3 Cluster Dendrogram with the Third Sample Set ...............66
6.4 Cluster Dendrogram with All Samples ........................67
6.5 Multidimensional Scaling Model with the First Sample Set ...69
6.6 Multidimensional Scaling Model with the Second Sample Set ... 69
6.7 Multidimensional Scaling Model with the Third Sample Set ...70
6.8 Multidimensional Scaling Model with All Samples ............70
6.9 Flow Chart Depicting Hypothesized Contact Between Various
Mediterranean Populations ..................................76
6.10 Scatter Plot of Distance and MMD Value for All Sample Pairs .. 82
6.11 Scatter Plot of Distance and MMD Value with a Subset of Sample
5.1 The Dental Samples...........................................32
5.2 Tooth, Trait, and Break Point Combinations for Statistical
5.3 Results of Trait-to-Trait Correlation Using Kendalls tau-b and
Spearmans Rho ................................................51
6.1 MMD Results with First Sample Set .........................59
6.2 MMD Results with Second Sample Set ........................59
6.3 MMD Results with Third Sample Set .........................60
6.4 MMD Results with All Samples...............................61
6.5 Dental Trait Frequency for Five World Populations ...........73
6.6 Percentage of Residual Genotype After a Per-Generation Gene Flow
of 1%, 5%, 10%, and 20% .......................................79
6.7 Subset of Sample Pairs with their Distance and MMD Value .... 83
This study is part of a multi-disciplinary investigation being conducted at Leptiminus,
Tunisia. Leptiminus was a port city established in the Punic period (post 800 B.C.),
that gained importance in the Roman period (1st century AD), and was abandoned at
the end of the Byzantine period (ca. AD 600-700). The ancient city of Leptiminus is
being explored through a cooperative archaeological investigation involving
Canadian, British, American, and Tunisian researchers. The broad goals of this
project are to elucidate the social, cultural, political, economic, and biological
relationships of this town to the contemporary Mediterranean region.
Carthaginians, the people who established the city of Leptiminus, had a strong empire
that challenged the might of Rome. After losing to Rome in 149 B.C. they fell into
obscurity. There are archaeological and historical references that tell us of the once
great empire but little is known of the people biologically. Carthaginians, like their
parent population the Phoenicians, spread their influence around the Mediterranean
through their prowess on the sea. As sea-going merchants, there were many
opportunities to interact with various Mediterranean populations. Did this exchange
result in gene flow and consequent biological affinity with those groups in the
Mediterranean? Or, is there a distinctiveness, a recognizable biological population
that is Carthaginian?
The expansion of Rome resulted in political, social, cultural, economic, and linguistic
influence throughout the invaded lands. Roman rule eventually spread across all of
the Mediterranean. What is the biological influence of the Roman expansion?
Historic references assert that Italian settlers (Roman citizens) were installed at
Leptiminus. Is there a discemable biological presence in Leptiminus by Italian
This study is directed at one of the main goals of the Leptiminus project; the
elucidation of the relationship of this port town to its local and regional areas.
Morphological data gathered from the study of the human remains from two Roman
period cemeteries are used to elucidate the biological relationship of these people to
other contemporary groups from the Mediterranean. This is accomplished by
recording dental morphological traits, determining the frequency of occurrence of the
dental traits, and statistically comparing them to other populations from the
Mediterranean. The use of dental morphological data for population affinity studies
can provide a means of estimating biological affinity. Studies such as this aid in the
understanding of the variability of human populations by tracing historic
This report is organized into chapters that explore the relationship of Leptiminus with
various Mediterranean groups. This introduction is followed by a chapter that
describes the use of dental morphological traits in determining biological affinity. The
third chapter provides background information on Leptiminus and the Carthaginians.
The specific objectives and hypotheses of the present study are detailed in chapter
four. Chapter five gives information on the samples and the methods used. The
results are discussed in chapter six and conclusions in chapter seven.
THE SCOPE OF DENTAL MORPHOLOGICAL TRAIT ANALYSIS
Dental anthropology today is an important area of study that is used to understand
human growth and development, genetics, health, and evolution. Teeth can be used
to aid in the taxonomic classification of fossils and to study the human species and its
origins. This chapter briefly describes the use of dentition in affinity studies and
includes general information for affinity assessment using non-metric traits (such as
dental variants), development of dentition, inheritance of dental traits, and non-
genetic and microevolutionary changes to dental form. The advantages and
disadvantages of dental morphological trait analysis are also assessed in this chapter.
Dental morphological traits are one type of non-metric trait that can be recorded on
human remains. Minor skeletal variants, referred to as non-metric, discrete,
discontinuous, or quasi-continuous traits, occur in all vertebrate species. Minor
skeletal variants that are not easily quantified by measuring on an interval scale are
frequently assigned to discrete states, such as present or absent.
Researchers use non-metric trait analysis to make inferences about relationships
within or between prehistoric human populations. The frequency of occurrence of
these traits in a particular population is considered a characteristic of that gene-pool
(Berry and Berry, 1972; Sjovold, 1973). The theoretical approach of (non-metric)
studies follows a biological population model, which assumes that these traits are
primarily under genetic control (Saunders, 1989:95). The use of non-metric traits to
determine biological distance is based on the assumption that trait frequencies are
directly proportional to distances based on gene frequencies with fewer shared traits
indicating greater biological distance. Conclusions about genetic distances should be
approached with caution because, the observation that two populations are
distinguishable by non-metric trait incidence does not prove genetic dissimilarity
because it is a phenotypic observation (Saunders, 1989:104). Many studies do,
however, suggest that the variation or frequency of various non-metric traits within
or between populations may indicate possible underlying genetic causes.
The use of non-metric traits for population affinity assessment is based on studies of
inheritance of skeletal non-metric traits in mice (Berry and Berry, 1972; Griineberg,
1963). These studies show that these traits are not inherited in a simple Mendelian
fashion. The presence or absence of traits is typically due to the action of multiple
genes (Scott, 1973). The fact that a non-metric trait is present is considered
diagnostic and not to what extent it is present (Sjovold, 1973). The studies with
mice showed that despite the complex genetic background of these traits, inbred
strains of mice show that populations of the same strain have approximately the same
proportions of variants (Berry and Berry, 1972; Sjovold, 1973 and 1977). Thus, the
frequency of occurrence of these variants (non-metric traits) is a characteristic of the
gene-pool in question. These non-metric traits are inherited in the same way in
humans as in mice (Berry, 1967). These early studies demonstrated the utility of
non-metric trait studies in mice and also indicated its utility in other organisms,
Evolutionary Origins of Human Tooth Form
Human teeth are complex structures and, as with other mammalian teeth, they
evolved from a three-cusp (tritubercular) form, which itself evolved from a simple
single-cusp homodont form (Scott and Turner, 1997; Weiss, 1990). There is
evidence that each mammalian cusp still passes, at least transitorily, through a
tritubercular developmental stage. One can speculate that the original entire tricuspid
metameric structure has duplicated a number of times in evolution, and we have seen
that this often means the sets of controlling genes (Hox genes) have also duplicated
or that the same set of perhaps-contiguous genes are expressed in each major
segment (Weiss, 1990:16 [parenthesis mine]).
There are two general models to explain differentiation in the development of
mammalian dentition, including human tooth form. The first, proposed by Butler
(1937, 1939), is the morphogenetic field model. Butler described mammalian teeth
as existing in three fieldscontrolled by the action of a morphogen. Depending on
its position in the dental arc, the action of the morphogen determines the type of
tooth; incisor, canine, or molar. Secondarily, the influence of the morphogen varies
through the field. The mesial tooth in each field is more stable in form and size with
more variability in the distal members of the field (Hillson, 1986; Butler, 1982). The
more stable tooth is under greater control of the morphogen, termed the polar
tooth (Lombardi, 1975). Less control is exerted on the teeth away from the polar
tooth allowing for greater variation. It is a hierarchical sequence having to do with
placement of the tooth germ which will determine, in order, tooth, class of tooth,
type of tooth within a class, and primary and secondary cusp morphology (Scott and
Turner, 1997; Hillson, 1986; Weiss, 1990).
A competing explanation was developed by Osborn (1978) and is a clone model
(Scott and Turner, 1997; Weiss, 1990). In this model the type of tooth is
determined early on in the development of teeth with existence of three primordia at
an early stage of odontogenesis associated with specific classes of teeth: an anterior
primordium (incisors), a canine primordium (canines), and a posterior primordium
(molars and premolars) (Scott and Turner, 1997:82). Both models offer important
insights into the variation in development of dental structures, though neither explain
fully the range of differentiation seen. The final solution will likely include elements
of both models (Weiss, 1990).
Considerable work remains to be done to understand the mechanisms involved in
producing the phenotypic expression of dental traits but progress is being made.
Dental form, like similar segmented, complex structures in the vertebrate skeleton
(e.g,. vertebrae, limb, stomach), are likely controlled by homeobox (Hox) genes
(Weiss, 1990,1997). These are gene segments that are homologous structures
created through duplication with variation. The morphology of complex
structures like the dentition, consists of repeats of similar anatomic units
(metameres) (Weiss, 1997:137 [parenthesis mine]).
The above understanding of the basic structure and development of teeth provides
important links between the genotype and resulting expression of a trait, the
phenotype. To be useful in estimation of biological affinity, the phenotype must be
largely genetically determined with minimal effect from the environment.
Realistically, there are no features that are wholly unaffected by non-genetic
influences. It has been recognized that the size and form of teeth are to a large extent
under genetic control and to a greater degree than other calcified tissue in the skeletal
structure (Berry and Berry, 1972; Scott and Turner, 1997; Sjovold, 1973).
Non-genetic Influences on Dental Form in Humans
To determine what, if any, environmental factors may influence tooth morphology,
research has been done with many different human groups in different environmental
settings. These studies include analysis of European colonists who settled in the
Americas, South Africa, and Australia, Chinese immigrants in America, and Norse
colonists in Iceland and Greenland (Lasker, 1945; Halffinan et al., 1992; Scott et al.,
1992; Scott and Alexandersen, 1992). These populations moved great distances into
new geographic areas with changes in environmental conditions such as temperature,
humidity, radiation, altitude, fluoride levels, and diet. The results showed that the
descendant populations had similar tooth morphology to the ancestral group and it
was distinct from the indigenous populations of those areas.
The morphological variants studied in dental trait analysis are secondary
characteristics that vary within a species. Researchers have recorded approximately
100 different morphological dental traits in the human dentition. Of these, about 40
crown and root traits are subjected to detailed analysis. The variation seen in human
dentition is largely expressed in the secondary characteristics, i.e., the size and form
of cusps, grooves, and the variation in root form and number. A possible explanation
for the occurrence of variability in these characteristics is that they may have little
effect on an individuals fitness (Butler, 1982). Thus, there is more freedom for these
secondary characteristics to vary within and between populations.
Non-genetic effects, or variability in dentition from environmental influences, can
occur due to differences in diet and/or health. Examples of these changes are enamel
hypoplasia (an interruption of the growth of the tooth resulting in lines or pits),
attrition, and caries. Enamel hypoplasia results in gross changes on the tooth but not
in the expression of the cusps and grooves or root structures which are the basis of
the non-metric dental traits (Hillson, 1990; Scott and Turner, 1997). Diet and health
can effect the quantity and severity of carious lesions (which can destroy a tooth) and
can influence the rate of attrition. Both of these factors affect the ability to score the
presence or absence of dental traits, hence affecting sample size, but not the actual
expression of the trait. Studies conducted on populations of mice have shown that
environmental influences occur in only a few traits which may result in changes in
overall size due to influences at critical developmental stages (Berry and Berry, 1972;
Sjovold, 1973). Thus, it is only those traits that are contingent on size that may be
affected and these influences were found to have no effect when a large number of
traits were considered for population characterization (Sjovold, 1973). Fluctuating
asymmetry, i.e., expression of a trait that is different on the left and right side, has
also been attributed to environmental factors (Hillson, 1986; Scott and Turner,
1997). These examples indicate that there is probably some influence in trait
occurrence that is environmental. While environmental factors influence trait
expression to some extent, they have not been shown to significantly affect
population trait frequencies (Scott and Turner, 1997:164).
Inheritance of Dental Variation in Humans
Inheritance studies have been conducted to better understand mechanisms of
inheritance for dental morphological traits (Nichol, 1989; Scott, 1973). Scott (1973)
analyzed 20 crown traits in American white families and found that the traits fell into
two general groups. There are traits that followed the expectations of a recessive
inheritance model and traits that correspond to a dominant inheritance model.
However, there were anomalies that did not fit either model and Scott suggested a
threshold model of inheritance. This model states that there are multiple loci
responsible for the expression of a trait, which seems logical given the metameric
nature of teeth as described above. For low frequency traits there is a higher
threshold to express a trait as present than there is for the high frequency traits. High
frequency traits have a lower threshold, requiring fewer loci for the present state to
be expressed. Complimentary to Scotts study is one conducted by Nichol (1989).
Nichols study suggests that 13 of the 24 traits examined are under major gene or
locus control, two are of polygenetic origin, and the remaining nine are
indeterminate. Of the 13 traits under major locus control, nine follow a model of
dominant inheritance, three follow a recessive model, and one is unclear.
No simple mode of inheritance for any crown or root trait exists, thus phenotypic
dental trait frequencies cannot be used to infer gene frequencies. Heritability is a
population specific value; it is the degree of genetic contribution to observed
variation between individuals in a population, not to the trait as it develops in an
individual (Scott and Turner, 1997:154). Heritability estimates suggest that genetic
influences are 40-80% for tooth morphology (Mizoguchi, 1978; Scott and Turner,
1997). There is some latitude for minor environmentally induced phenotypic
adjustments but these do not significantly affect population trait frequencies. It is not
essential to know the exact mode of inheritance to use dental morphological traits as
an estimate of biological affinity between populations. The mode of inheritance may
vary for different variants which is why a suite of traits should be used for dental
morphological trait analysis and not any one trait. It is the frequency of occurrence
of the suite of traits that is characteristic of the population.
Evolutionary Changes in Dental Form in Humans
The differences in trait frequencies between populations are usually the result of
microevolutionary forces such as mutation, genetic drift, gene flow, and natural
selection. Microevolution is changes in gene frequencies in a population or species.
This is within species changes and not macro changes that lead to speciation, i.e.,
creation of new species. Dental morphological studies can be used to assess the
degree of relationship between human populations. Greater biological difference
between groups suggests less affinity and a longer period of separation (Irish, 1993).
Genetic drift is a stochastic process, i.e., involving chance events in small
populations. These random events are evidenced in several ways. For example, there
is the random nature of transmitting alleles from generation to generation. The
frequency of a gene will not be exactly reproduced in the next generation because of
sampling error (Suzuki et al, 1989). In addition, there are random events that can
influence trait frequency, such as natural disasters. If a natural disaster drastically
reduces the number of individuals in a population, gene frequencies in the surviving
members may, by chance alone, be quite different from those of the larger pre-
disaster population. This situation is known as the bottleneck effect (Campbell,
Founders effect, a type of genetic drift, can result in changes in the population
frequencies of traits within a relatively short time. Small groups are separated from
the ancestral population and gene flow is restricted due to barriers such as
geographic, linguistic, cultural, or social influences, resulting in group divergence.
The effect of genetic drift depends primarily on population size. It is much more
influential in small populations.
Natural selection refers to the changes in frequencies of traits in populations due to
differential reproductive success. A phenotype (and its underlying genotype) may
provide a reproductive advantage in a certain environment. Organisms adapt through
natural selection to environmental stimuli. The dental traits that have received the
most attention by researchers examining the question of adaptation are shovel-shaped
incisors, Carabellis cusp, winging, cusp 5 (upper molar), and cusp 6 (lower
molar)(Dahlberg, 1963; Cadien, 1972; Mizoguchi, 1985; Hylander, 1977; Townsend
et al., 1986). Potential advantages and disadvantages for variation in root number
has also been examined (Hylander, 1977; Turner, 1987; Guthrie, 1996). Advantages
for tooth and root traits have been discussed in terms of increasing the strength and
durability of the dentition and masticatory function. Disadvantages of various traits
are discussed as increased risks of periodontal disease and carious lesions (Tsatsas et
al., 1973). In terms of the contribution of natural selection and the adaptability of
dental traits, (dental) features may provide some advantage in terms of enhancing
function, inhibiting wear, preventing tooth loss, etc. While these traits may serve a
function, it none the less remains difficult to move from inferred function to a
demonstration of fitness (Scott and Turner, 1997:250). Natural selection may play a
role in the patterns of dental variation but the changes are marked in hundreds of
thousands of years (Butler, 1982; Scott and Turner, 1997). Differences between
subpopulations, for the time scale of this and many other biological affinity studies
(i.e., 15-20,000 years or less), are largely a result of gene flow and genetic drift; e.g.,
migration, founders effect, and mating practices (Weiss, 1988; Scott and Turner,
To understand the pattern of human population differences, gene flow models have
been developed to explain the diversity seen between groups. Population movement,
or migration, can be into areas that are unpopulated by other groups or into areas
with an established population. In the case of the later, migration can lead to the
exchange of genetic material (admixture) between two populations. Relationships
between human groups are complex and social or cultural isolationism (e.g.,
language, status, behavior, and mating practices) can create barriers to gene flow.
One model to explain population diversity is isolation by distance. This model
describes the tendency of populations to exchange mates (genes) with their
geographically close neighbors. Geographically close neighbors will have greater
biological affinity and differences increase with geographic distance largely due to
genetic drift (Cavalli-Sforza et al., 1994; Weiss, 1988). This model assumes that
there are no barriers to gene flow except geographic distance. The results are often
shown as clinal maps, which show changes in gene frequency by geographic distance
(Piazza et al., 1981).
Expansion of human populations into inhabited lands can result in genetic exchange
between groups that had been geographically, and probably genetically, distinct.
Weiss (1988) describes this process of gene flow and population expansion as
invasion. This invasion model is a process of population expansion with
incorporation of the invaded population into the resulting gene pool (Weiss,
1988:130). The mass movements of human groups can be instigated by various
phenomena such as agricultural innovation, technological advances, transportation
advances, and military or political conquest. The resulting admixture of the two
groups can generate a new population, intermediate between the two original groups
(Cavalli-Sforza et al, 1994). The original gene pool of the inhabited group can
completely change; the speed of change depending on population size (of both the
inhabited and migrating group), amount of time (number of generations), and
percentage of gene flow. If gene replacement is nearly complete, then determining
genetic origin (ancestry) may not be possible.
In terms of biological affinity studies, similarity could be a result of common ancestry,
admixture, or convergent evolution. Determining which is more likely requires
examination of alternative lines of evidence such as historic records, archaeological
data, and other biological data (for example, DNA analysis or metrical data). Dental
morphological trait analysis can provide estimates of distance or affinity between
Previous Affinity Studies Using Dental Traits
Research in the early 1900s by Hrdlicka (1920), Heilman (1928), Tratman (1938),
Chappel (1927) and Yamada (1932) began to reveal differences between human
populations based on the occurrences of certain dental traits. It was recognized that
certain traits could be used to identify populations, such as incisor shoveling in
Mongoloid populations (Hrdicka, 1920; Hanihara, 1963, 1967) and Carabellis cusp
in Caucasoid populations (Kraus, 1951). Dental traits could help to characterize
populations and thus aid in assessing biological relationships and estimations of
genetic admixture between populations (Turner, 1967; Scott and Turner, 1997).
Since 1965, dental anthropology has become an important branch of Physical
Anthropology and the decades since have seen a large volume of literature devoted to
various aspects of dental anthropology. Descriptions of dental morphological traits
and characterizations of populations using dental traits has also benefitted from this
increased interest in dental anthropology. Hrdlicka (1920) presented some of the first
attempts at standardizing the variation in dental morphology. Dahlberg (1956) and
researchers at the Zoller Dental Lab created reference plaques for use in describing
various dental morphological traits. Other researchers, such as Hanihara (1963) and
Morris (1965) also developed methodological standards for scoring dental traits. The
most recent development has been by Turner, Morris, and students at Arizona State
University (Turner et al., 1991) with the creation of the Arizona State University
Dental Anthropology System (ASU DAS) which consists of 23 dental plaques and
descriptions for 42 dental morphological traits.
Dental morphological trait analysis has been used on various populations around the
world and at various magnitudes or levels of differentiation from global, regional, to
local. Dental trait analysis has been used to examine global research issues such as
the peopling of the Americas, the origins of the Polynesians, Micronesians, and
Australians (Hanihara, 1990, 1992; Swindler et al., 1995; Townsend et al., 1990;
Turner, 1984, 1985a, 1990,1992). On a regional level comparative studies have been
done on groups such as the prehistoric people of India who were found to be
intermediate between Europeans and Asians (Lukacs and Walimbe, 1984). Greene
(1967,1972,1982) and Greene et al. (1967) examined the relationship of three
Nubian populations to answer the question of continuity or population replacement
for Nubian Africa. On a local level, a study by Johnson and Lovell (1994) on three
predynastic Egyptian cemeteries examined the phenetic relationship to determine if
inbreeding or immigration explained the differences seen. Coppa et al. (1998)
examined populations from Iron Age Italy to determine if the Apennine Mountains
created a geographic barrier for population movement. These are but a few of the
examples of studies done around the world using dental morphological traits. Scott
and Tinner (1997:Appendix B 325-336) list many references, divided by geographic
region, of research using dental morphological traits and offer researchers an
excellent source for assessing work in a particular region.
Advantages and Disadvantages to Using Dental Trait Analysis
In summary, the advantages and disadvantages of dental trait analysis are described
as follows. First, the major disadvantages or limitations are: 1) loss of information
from attrition, pathologies, or diagenetic processes; 2) understanding of the exact
mode of inheritance; 3) lack of standardization; 4) use of published material; 5) lack
of information on regional population variation. Some of the advantages of dental
trait analysis help to counter these disadvantages. Teeth have many useful
characteristics such as: 1) teeth are durable and often the best preserved portion of
skeletal samples; 2) they possess a high genetic component; 3) except for attrition
and caries, they do not change shape once formed (as bone does); 4) they are less
affected by the environment than bone; 5) they can be studied in living and extinct
populations; 6) dental trait analysis does not require expensive equipment and can be
done by one researcher to help minimize time and expense.
Loss of information can occur through normal attrition, which obliterates the cusp
and groove pattern that is the major focus of dental morphological study. Caries and
tooth loss can also remove teeth from the sample. Teeth are subject to disintegration
from natural diagenetic processes. Teeth are hard, especially the enamel structure
and will often preserve better than other bony tissue. There is little to be done about
the loss of information from wear and pathologies. However, if one antimere is
missing or damaged, the other antimere can be used. The loss of information can
limit sample size and is an unavoidable aspect of studying human remains.
The second disadvantage is a lack of understanding of the exact mode of inheritance.
Dental geneticists think some dental traits are polygenic with quasi-continuous range
of expression (Dahlberg et al., 1982; Scott, 1973; Harris, 1977) and other traits
under a major gene model (Nichol, 1990; Turner, 1967, 1969). It is possible the
inheritance is different for different variants, individuals, and populations (Hillson,
1986). If the main mechanism is quasi-continuous, what part of the variation is most
informative; presence/absence or the range of variation? There is still work to be
done using inheritance studies to understand the underlying genetic mechanism.
However, dental morphology does posses a high genetic component in expression
(Alvesalo and Tigerstedt, 1974; Berry, 1978; Scott and Turner, 1997) and knowing
the exact mode of inheritance is not necessary for affinity estimation (Scott and
Turner, 1997; Irish, 1993).
The third limitation is the lack of standardization in observation and recording of
dental traits. Hrdlicka (1920) and Dahlberg (1956) were some of the first researchers
to try to standardize descriptions and recording of dental traits. The development of
the ASU DAS is an extension of these earlier works and offers a means of
standardizing the recording of dental traits. There is some observer error that is
possible but this is generally minimized by the use of the ASU DAS dental plaques
and experience with the recording process. In the study by Nichol and Turner (1986)
examining results from the ASU DAS, the intra-observer concordance was 90% and
inter-observer concordance 87%. To help negate this limitation, the present study
uses the ASU DAS system and the comparative samples are by researchers who also
used this system.
The next disadvantage is in using published material for comparative purposes. This
is associated with the lack of standardization in that other researchers may have
recorded traits differently. Not all researchers use the ASU DAS system and they
may not fully describe their basis for trait recording. Even when the ASU system is
used, researchers may not use the same break point for determining presence/absence
of a trait or the same suite of traits and not all traits scorable may be presented in the
published sources. All factors must be the same for the variants used to ensure the
best results possible. The best solution is to have the same researcher score all the
teeth used. This is often not possible due to availability of the samples (e.g., some
may be repatriated after initial analysis), time, and cost.
The last limitation is one that this study will, in part, address. There is a lack of
information on the dental variation in different populations from around the world.
North Africa has had few studies completed on recovered human remains. Data exist
for Egyptians and Nubians but little until recently has been done for Carthaginians.
There is one study, Irish (1993), that reported results of dental traits for 24
individuals from Carthage. This study will remedy this situation in part, by presenting
information on people from this region of the world and thus, will help in better
understanding their origins and affinities.
BACKGROUND ON LEPTIMINUS
The Leptiminus Project
Leptiminus is located near the modem town of Lamta along the east coast of Tunisia
(Figure 2.1). The Leptiminus project was initiated at the request of the Tunisian
Ministry of Archaeology (the Institut National du Patrimoine) to explore the ancient
city in areas being impacted by expansion of the modem town of Lamta. Rescue
excavations included archaeological recovery of human remains from two cemeteries.
The first cemetery, Site 10, is dated to the Roman period, 2nd to 4th century A.D., and
is located near the Bath house along the coastal shore (Figure 2.2). The second
cemetery, Site 200 (Figure 2.2) is also Roman period (2nd to 4th century A.D.) and is
located up the hill from Site 10 along the east side of the ancient city on the Dahar
Silma ridge near the ancient kilns (amphora and ceramic manufacturing center).
The city has a Punic origin (i.e., Phoenician or Carthaginian), gaining in importance
and prosperity during the Roman period. Leptiminus was promoted to the rank of
colonia in the reign of Trajan. The people of Leptiminus are described as follows by
Lazreg and Mattingly (1992:60-61); Citizens of Leptiminus are attested as
belonging to four of the Roman tribes, the Palatina, Claudia, Quirina, and Papiria.
Those belonging to the Palatina and Claudia may be descendants of Italian settlers
(Roman citizens) installed at Leptiminus in the 1st century B.C.
Figure 2.1. Project Area Location
Figure 2.2. Location of Cemeteries at Leptiminus
Map reproduced from Stirling et al., 2001 :Fig. 3.1. p216.
Carthaginians are Phoenicians who established a city that grew into an empire on the
north coast of Africa. The Phoenicians came from the land of Phoenicia, a narrow
strip of land between the mountains and the sea, on the Mediterranean coast in what
is today Lebanon (Starr, 1991). Little is known of Phoenicia through archaeology
because their towns are still occupied. As Egypt declined, local political and cultural
independence rose for small states; one of these was Phoenicia (Starr, 1991). It is
thought that Sea People (probably Mycenaean Greeks who settled in the Levant
around 1200 B.C. (Trump, 1980)) and Canaanites combined to become Philistines in
the south and Phoenicians in the north.
Phoenicians are best known for two things; their prowess on the sea and their
contributions to writing. Historic references say that Necho II, Egyptian ruler in the
7th century B.C., commissioned an expedition of Phoenician ships which sailed
southwards down the Red Sea to seek a route round Africa. Incredible though it may
seem, they succeeded, and three years later, they sailed back into the Mediterranean
from the Atlantic (Trump, 1980:232). In terms of their contribution to writing, the
Phoenician alphabet used 22 signs which stood for consonantal sounds and was
written in script from right to left. Many diverse viewpoints on the influence of the
Phoenicians exist. Their contribution to writing is uncontested. Otherwise, there are
those who feel their contributions were rather insignificant (Starr, 1991) and those
who believe their influence was greater (Trump, 1980).
Trade was clearly the Phoenician driving force (Trump, 1980:249). Sites, such as
Carthage, were chosen for their strategic control of the sea routes between primary
resources (e.g., metal ores) and markets of the Near East. Carthage was founded
about 814 B.C. by Princess Dido, otherwise known as Elissa. The colony of
Carthage acquired considerable control and influence in the Mediterranean after the
fall of the Phoenician homeland about 650 B.C. Carthage attained an empire equal to
that of Roman Italy with an estimated population of 3,000,000 (Starr, 1991).
Numerous historical references of the Carthaginians detail their antagonistic
relationship with Rome. Carthage and Rome fought three Punic Wars (Punic is a
Greek word for Phoenician). The most famous of these wars is the second Punic
War lead by Hannibal. Hannibal lead a large army onto the Italian peninsula from
Spain. He crossed the Alps with his army which included elephants perhaps the
best known fact from these wars. Under the guidance of Hannibal, the Carthaginians
seriously challenged the might of Rome. Hannibal remained undefeated on Italian
lands but the war was lost because Rome sent an army of their own to challenge
Carthage. Carthage surrendered and, one of the conditions of the surrender, was the
recall of Hannibal from Italy. The third Punic War ended the reign of Carthage.
Rome utterly defeated Carthage in 146 B.C. after laying siege to the city for over a
Small port towns, like Leptiminus, sprang up in the empire of the Carthaginians
where valuable resources occurred. For Leptiminus, one of the reasons for its
establishment and later prosperity was due to its excellent port facilities. Leptiminus
lies in a protected, though shallow bay. The harbor facilities built at Leptiminus
allowed all but the largest merchantmen vessels(Davidson, 1992:174) to be loaded
and unloaded. The harbor facilitated the export of products such as fine wares
(African Red Slip produced at Leptiminus) and agricultural products from the interior
of Tunisia (such as olive oil, fish garum (a popular fish sauce), and grains), which
were shipped to various Mediterranean destinations.
Leptiminus was established in the Punic period by the expansion of the Carthaginian
empire and came under Roman control after the fall of Carthage in the third Punic
War, post 146 B.C. The citys prosperity rose rapidly under Roman rule and
maintained its importance into the Byzantine period. Reasons for the citys decline
and abandonment are unclear, perhaps related to incursions by Arab invaders. The
followers of the Prophet Mohammed successfully conquered the North African
region by A.D. 700. The entire region remained under Islamic rule until the French
colonization of the area in more recent times.
Since the early 1800s, visitors, archaeologists, and looters have systematically
pillaged and destroyed parts of the ancient city of Leptiminus. Particularly impacted
were the cemeteries looted for antiquities to sell to the French (Lazreg and Mattingly,
1992). Leptiminus was not alone in this campaign of excavations and lootings which
have occurred at other Tunisian, and regional, sites. Archaeological sites in Tunisia
have only recently been systematically studied by archaeologists. Many of these
locations are now popular tourist destinations. Today, tourism brings important
economic benefits to Tunisia which help to preserve and protect these important
Since the mid 1970s, a UNESCO project (Save Carthage Project) has greatly
increased our knowledge of social and cultural aspects of the Carthaginians.
Archaeological excavations have been conducted in Carthage by a host of countries
who are trying to save the heritage of the city before modem development erases all
evidence. Despite this intensive effort, there is very little that we know about the
Carthaginians and their parent population, the Phoenicians, in biological terms. Early
excavations were done in Tunisia in the 1800s by the White Fathers (Catholic
Priests). Some of these individuals routinely excavated areas to better understand
Christian history in the Mediterranean. Occasionally, human crania were kept from
these excavations but most human remains were not kept or studied. The
excavations in the mid 1970s seemed to follow a similar practice. Tombs were noted
and human remains were either not kept or have since been lost. This practice
changed in the late 1980s and 1990s but few remains are systematically studied or
published. As more data are published, a better understanding of these people will
emerge. This study is part .of the effort to better understand who the Carthaginians
were and what their relationship was to the other people of the Mediterranean.
GOALS OF THE ANALYSIS
Scant record exists regarding the biology of the extinct population of Carthaginians,
about their origins, and their relationships to other Mediterranean populations. This
study will help augment the record of the Carthaginians in the following ways.
First, dental morphological traits are examined, recorded, and quantified from two
cemeteries from the ancient town of Leptiminus. Descriptive analysis examining
frequencies of these dental traits are computed identifying common local and regional
occurrences of specific traits. This data set helps to fill the void regarding the
biological characteristics unique to these people.
Secondly, phenetic similarities can be assessed, with the aid of inferential statistical
analysis, by comparing the suite of dental traits from the Leptiminus samples and
other local and regional populations. Phenetic similarity is assumed to indicate
biological affinity based on the results of the inferential statistics with less affinity
equal to greater biological distance. Frequencies of dental traits from the Leptiminus
samples can be used, in this and other studies, to better understand the origin and
affinities of ancient human populations in North Africa.
This study assesses Carthaginian population distinctiveness and affinity with other
contemporary, regional Mediterranean groups, and examines possible biological
influence from the Roman political and military expansion in the Mediterranean. By
systematically analyzing dental morphology of the Leptiminus population and
statistically quantifying the intra- and inter-population relationships, this study seeks
to reveal the biological affinity of the Leptiminus group with other local
(Carthaginians from Carthage) and regional populations (Egyptian, Nubian,
Levantine, and Italian). A hypothesis of Carthaginian distinctiveness will be tested by
examining the relationship of the Carthaginian groups to each other (Leptiminus and
Carthage). To assess the biological influence of the conquering Romans, the
relationship of the Italian populations to the other regional groups, may reveal
possible admixture through gene flow of the invading Romans into the lands of
various Mediterranean populations.
The significance of this study is twofold. First, the scarcity of published biological
information regarding the Carthaginians is very real. This study provides baseline
research information on a group of people who have previously been represented
only through their archaeological remains and historical references. These were an
important and powerful group of people relegated to biological obscurity. Therefore,
standardized recording and descriptive data are important in and of themselves.
Secondly, the use of dental morphological data for population affinity studies
provides information on the origins, relationships, migrations, and admixture of these
people with other contemporary Mediterranean populations. This aids in advancing
our understanding of the variability of human populations by tracing historic
Hypotheses To Be Tested
Biological Distinctiveness of the Carthaginians
Who are the Carthaginians? Very little is known, biologically, about the
Carthaginians and their parent population the Phoenicians. The Carthaginians gained
social, economic, and political influence throughout the Mediterranean after the 6th
century B.C. They were known as merchants suppling goods around the
Mediterranean capitalizing on their prowess on the sea. How is this regional
influence reflected biologically? Does this exchange of goods also mean an exchange
of genes? Is theirs a melting pot society with biological affinity with various groups
in the Mediterranean region? Or, is there social cohesion that is reflected in
biological cohesion, creating a biologically distinct population recognizable as
Hypothesis #1. The Carthaginians are a distinct biological poulation. They are not
an amalgam of individuals from various regional populations. If the Leptiminus and
Carthage samples are found to be phenetically similar to one another phenetically
different from other populations in the area, then the hypothesis that the
Carthaginians are a biologically distinct population will be supported. Findings to the
contrary would not support the presence of a distinct biological Carthaginian
Biological Influence bv Roman Expansion
The period of Roman influence, roughly 5th century B.C. through 5th century A.D.,
reflects the rise, expansion, and political and military control of Rome throughout the
Mediterranean region. Roman rule began in the western Mediterranean and
eventually encompassed all of the Mediterranean, the Balkans, France, and England.
The military expansion of Rome resulted in political, social, cultural, economic, and
linguistic influence throughout the invaded lands. The basic tenets of an invasion
model of gene flow (Weiss, 1988) describe population expansion into inhabited lands
via invasion. The process of mate exchange between populations is gene flow; a
combination of invasion and gene flow is referred to as demic diffusion defined as,
expansion with incorporation of the invaded population into the resulting gene pool
(Weiss, 1988:130). Is there demic diffusion associated with the invasion of Roman?
How is this regional influence reflected biologically? Is this unity under political and
military power reflected in greater gene flow and admixture resulting in phenetic
similarity across diverse regional populations? Or, are there ethnic and social barriers
maintaining isolationism reflected in social and geographical biological cohesion?
After the fall of Carthage, Rome gained control of the region, including Leptiminus.
Historical references assert that Italian settlers were installed in Leptiminus circa 1st
century B.C. Is this assertion discemable biologically? Is there biological influence
in Leptiminus by this Roman expansion as historical references suggest?
Hypothesis #2. Roman expansion in the region includes biological influence in
regional populations. The various regional Mediterranean populations are
phenetically similar to one or more of the Italian populations. If the Leptiminus,
Carthage, Egyptian, Nubian, or Leventine samples and any one or several of the
Italian samples are found to be phenetically similar to one another, then the
hypothesis that Roman biological influence is present will be supported. Further,
phenetic similarity of any one of the Italian samples and Leptiminus will also support
the historical reference to the presence of Italian settlers. Findings to the contrary
would not provide support for this hypothesis.
Isolation bv Distance
Isolation by distance, a primary model evaluating biological affinity among groups, is
based on the tendency of groups to exchange mates with groups that are near in
distance, resulting in greater affinity between geographically close groups and less
affinity with more distant groups (Cavalli-Sforza et al, 1994; Weiss, 1988). The
basic tenet of this model suggests that there are no barriers to gene flow and it is
distance alone that predicts affinity. In this model, geographic barriers are important
to assessing affinity. How well does this model predict the aflSnities between these
Mediterranean populations? Is the sea or desert an effective barrier to gene flow?
Hypothesis #3. Phenetic similarity between groups is closely correlated with
geographic proximity. If the groups nearest each other exhibit the greatest phenetic
similarity with less similarity as distance increases, then the hypothesis that phenetic
similarity is based on distance alone will be supported. Findings to the contrary
would not provide support for this hypothesis.
MATERIALS AND METHODS
The samples used in this study include dental data from populations found around the
Mediterranean region. The dental remains from the ancient town of Leptiminus were
examined by the author and the other samples are from published sources. The
samples, the date(s) they represent, and source of information are summarized in
Table 5.1. The geographical areas these samples are from are shown in Figure 5.1.
These samples are all roughly contemporaneous relating to the period of Roman
influence, approximately 5th century B.C. through 5th century A.D. These samples
were selected because they represent contemporaneous populations with similar
dental morphological data.
The dental remains from the two Leptiminus cemeteries were scored on 32 dental
traits. The dental remains varied in completeness and in preservation quality. Site 10
had a greater number of burials with dentition in fair to good condition. Site 200
remains were poorly preserved. The burials with any dental remains were examined
and scored. To increase the sample from Site 200, loci with disarticulated remains
which included scorable teeth were included.
Samples from published sources use the same dental scoring method but varied in the
number of traits recorded, number of teeth these traits were scored on, and varied in
the dichotomization points for determining presence or absence of a trait. This
variability is unavoidable when using data from published sources and is accounted
for in calculating the comparisons.
The Leptiminus Samples
The dental remains for Leptiminus are from two cemeteries dating from the 2nd to 4th
century (A.D. 100-400) designated Site 10 (LEP 10) and Site 200 (LEP 200). The
site 10 sample is from 26 burials for a total of 502 teeth. The site 200 sample
includes remains from 31 burials and 8 loci of disarticulated bone for a total of 463
teeth. Together there are 965 teeth from which the 32 dental traits could be scored.
The author scored all the teeth and re-scored a randomly chosen sample of228 teeth
from 12 burials to test for intra-observer error. The remains are housed at the Lamta
Museum in Lamta, Tunisia.
Samples from Published Sources
The comparative samples are all from published sources. These samples include one
from Carthage, one from Egypt, two from Nubia, five from Italy, and one from the
southern Levant, Israel (see Figure 1). Presently there is only one other sample of
Carthaginian remains with dental morphological information collected. This sample
is from Irish (1993) and includes dental remains from 28 individuals. The Carthage
(CAR) sample includes 24 individuals excavated in the late 1800s by P. Delattre and
are simply classified as Punic which for Carthage is 751 146 B.C. The other four
individuals are from the early Roman period 146 B.C. A.D. 435. These remains
were scored by Irish and are housed in the Musee de lHomme (Irish, 1993).
Table 5.1 The Dental Samples
Sample Name Geographic Location Date Source Sample Size
Leptiminus Site 10 (LEP10) Leptiminus, Tunisia A.D.2d-4'hC Author this report 26
Leptiminus Site 200 (LEP200) Leptiminus, Tunisia A.D. 2nd 4"' C Author this 39
Carthage (CAR) Carthage, Tunisia 751 B.C.-A.D. 435 Irish 1993 28
El Hesa (HES) Egypt A.D. 200-400 Irish 1993 72
Meriotic (MER) Semna South, Sudan 100 B.C.-A.D. 350 Irish 1993 91
X-Group (XGR) Semna South, Sudan A.D. 350-550 Irish 1993 39
EnGedi (GED) Israel 250 B.C.-A.D. 590 Lipschultz 1996 170
Etruscan (ETC) Italy 4*-2i C B.C. Coppa et al. 102
Latini (LAC) 1998 94
Campani (CAC) 54
Piceni (PCC) 211
Sulmona (SUL) 52
One Egyptian and two Nubian samples are also reported in Irish (1993). El Hesa
(HES) is an Egyptian sample from a cemetery on an island in the Nile near Aswan.
This cemetery dates from A.D. 200-400. The sample of 72 individuals was scored by
Irish and is part of the von Luschan collection in the American Museum of Natural
History. Little is known about these people except that they may be middle class
agriculturalists (Irish, 1993). There is one Meriotic sample (MER) consisting of data
collected from 91 Nubians from a Meriotic period (100 B.C. A.D. 350) cemetery at
Semna South, Sudan. These people are reported to have an economy based on
agriculture and animal husbandry, in addition to extensive trade with Greek and
Roman rulers of Egypt (Irish, 1993:114). The other sample from Irish is 39
individuals from an X-Group (XGR) period (A.D. 350-550) cemetery also from
Semna South, Sudan. These individuals may represent a genetic continuation of
Meroitic Nubians (Greene, 1982; Irish, 1993). Both of these samples are housed at
Figure 5.1 Geographic Location of Samples
North Africa: Leptiminus (LEP), Carthage (CAR), El Hesa (HES), Meriotic (MER), X-Group
(XGR), East Asia (Levant): En Gedi (GED), Italy: Etruscan (ETC), Latini (LAC), Campani (CAC),
Piceni (PCC), Sulmona (SUL).
the Arizona State University and were scored by Irish.
The southern Levant sample is from the site of En Gedi (GED) in Israel dating from
250 B.C. A.D. 590 and is published in Lipschultz (1996). This sample of 170
individuals was scored by Lipschultz and was reburied following analysis. This
Levantine sample is south of the home land of the Phoenicians, the ancestors to the
The Italian samples come from Coppa et al (1998). These samples date from the
late 4th 2nd century B.C. (400 100 B.C.). There is one Etruscan (ETC) sample
(102 individuals) from north of Rome, a Latini (LAC) sample (94 individuals) from
near Rome, a Campani (CAC) sample (54 individuals) from south of Naples, and two
samples from the high plains east of the mountains which are Piceni (PCC) from 211
individuals and Sulmona (SUL) from 52 individuals (Coppa et al., 1998:372-373).
Dental Morphological Traits
This study is concerned with determining biological affinity using dental
morphological data for phenotypic representation. The dental data for all samples in
this study were collected using the ASU dental system, thus promoting replicability
of results between researchers. The traits are scored using a combination of
descriptive information and plaster casts. These traits are described in detail in
Turner et al. (1991), Liu and Turner (1993), and Scott and Turner (1997) and are
briefly described below. Numbering of traits below follows Turner et al. (1991).
Traits on Maxillary Dentition
Winging Incisor 1. The central incisors can be rotated mesiolingually resulting in a
V-shaped appearance. No reference plaque is used and is scored as one of four
1. Bilateral winging resulting in V-shape.
2. Unilateral winging where only one of the cental incisors is rotated.
3. Straight or no expression of this trait.
4. Counter winging with incisors rotated distolingually.
Labial Curve Incisor 1. The labial surface of the upper incisors can be convex in
appearance when viewed from the occlusal aspect. Reference plaque ASU UI1 labial
curvature is used with five possible scores for this trait.
0. Labial surface flat or no expression of trait.
1. Trace convexity.
2. Weak convexity.
3. Moderate convexity.
4. Pronounced convexity.
Shoveling Incisors. Lingual marginal ridges can be seen on upper incisors giving the
tooth a shovel-like appearance. Reference plaques used are ASU UI1 shovel and
ASU UI2 shovel. There are six possible grades for UI1 and seven for UI2.
0. No shoveling or no expression of trait.
1. Faint ridges.
2. Trace ridges.
3. Semishovel shaped.
4. Semishovel shaped but stronger expression than in grade 3.
5. Shovel shaped.
6. Marked shovel shaped.
7. UI2 only: Barrel shaped.
Double-Shovel Incisors. Labial marginal ridges can be present on upper incisors.
Reference plaque ASU UI1 double-shovel is used. Possible scores are 0 6 for this
0. Smooth surface or no expression of trait.
1. Faint ridge possibly on only one margin.
2. Trace ridges.
5. Pronounced double-shovel.
6. Extreme double-shovel.
Interruption Groove Incisors. Groove on lingual surface of incisors crossing the
cingulum. No reference plaque is used. Trait is scored as absent or present in mesio-
and/or disto-lingual border or in medial area of cingulum.
Tuberculum Dentale Incisors and Canine. This trait can take on the characteristic of
ridges or cusps and is located on the lingual surface of the tooth in the cingulum area.
There are eight possible scores for this trait using ASU UI1 t.d. and ASU UC distal
accessory ridge reference plaques.
0. No expression of trait.
1. Faint ridge.
2. Trace ridge.
3. Strong ridge.
4. Pronounced ridge.
5-. Weak cuspule.
5. Weak cuspule with free apex.
6. Strong cusp.
Mesial Ridge Canine. Also known as Bushman canine because of the frequency of
occurrence in that population, this trait is characterized by a mesial ridge that is larger
than the distal ridge on the lingual surface of the canine. In pronounced cases the
tuberculum dentale will be incorporated. Reference plaque ASU UC mesial ridge is
used and four possible grades of occurrence are scored.
0. No expression with mesial and distal ridges equal size.
1. Mesiolingual ridge larger and weakly attached to tubercle dentale.
2. Mesiolingual ridge larger and moderately attached to tubercle dentale.
3. Mesiolingual ridge is much larger and fully attached to tubercle dentale.
Distal Accessory Ridge Canine. An additional ridge on the distolingual surface can
occur and occasionally be quite pronounced. Reference plaque ASU UC DAR is
used. Scores can range from 0 to 5.
0. No expression of trait.
1. Faint ridge.
2. Weak ridge.
3. Moderate ridge.
4. Strong ridge.
5. Pronounced ridge.
Tricusned Premolars. A premolar with three cusps. No reference plaque is used.
Trait is scored as present or absent.
Hvpocone Molars. Cusp 4, the distolingual cusp, can vary greatly from absent to
quite large. Reference plaque ASU UM hypocone is used. Trait can be scored for
seven possible grades.
0. No hypocone.
1. Faint ridging.
2. Faint cuspule.
3. Small cusp.
3.5. Moderate cusp.
4. Large cusp.
5. Very large cusp.
Cusp 5 (Metaconulel Molars. Cusp 5 can occur between the metacone (cusp 3) and
hypocone (cusp 4) with six possible grades that can be scored. Reference plaque
ASU UM cusp 5 is used.
0. No cusp 5.
1. Faint cuspule.
2. Trace cuspule.
3. Small cuspule.
4. Small cusp.
5. Medium cusp.
Carabellis Trait Molars. Carabellis trait occurs on the upper molars on the lingual
surface of the mesiolingual cusp (cusp 1). This follows the system developed by
Dahlberg (1956) and uses the Zoller Laboratory UM Carabelli cusp reference plaque.
There are eight possible grades for this trait.
0. No expression of trait.
1. Groove present.
2. Pit present.
3. Small Y-shaped depression.
4. Large Y-shaped depression.
5. Small cusp with attached apex.
6. Medium cusp with attached apex.
7. Large free cusp.
Parastvle Molars. This trait occurs on the buccal surface of the mesiobuccal cusp
(cusp 2). Its presence is rare and any presence of pit to cusp is scored as present.
Reference plaque ASU UM parastyle is used. There are seven possible scores.
0. No expression.
1. Pit in or near buccal groove.
2. Small attached cusp.
3. Medium free cusp.
4. Large free cusp.
5. Very large free cusp.
6. Free peg-shaped crown attached to root.
F.namel Extension Molar 1. Extension of the enamel border towards the root apex.
No reference plaque is used. Scores vary from 0-3.
0. No expression of trait.
1. Faint extension (~ 1.0 mm).
2. Medium extension (~ 2.0 mm).
3. Lengthy extension (> 4.0 mm).
Root Number Premolars. Number of separate roots, roots must be separated for
more than 1/4 to 1/3 total root length to be counted as separate. No reference plaque
is used. There are three possible scores.
1. One root (tip may be bifurcated).
2. Two roots.
3. Three roots.
Root Number Molars. As with above premolar root number, this trait counts the
number of separate roots. No reference plaque is used. There are four possible
1. One root.
2. Two roots.
3. Three roots.
4. Four roots.
Peg-shaped Lateral Incisor. An incisor that is reduced in size and crown is not
normal but peg-shaped. No reference plaque. Three possible grades for this trait.
0. No expression, tooth is normal.
1. Reduced in size with normal crown.
2. Reduced in size with peg-shaped crown.
Peg-shaped Molar 3. As with the incisor, the third molar can be reduced in size with
abnormal crown morphology. No reference plaque. Three possible scores for this
0. No expression, tooth is normal.
1. Reduced in size (7- to 10-mm buccolingual diameter).
2. Reduced (<7 mm buccolingual diameter) and crown is peg-shaped.
Odontome Premolars. Any pin-size, spike-shaped enamel and dentine projection on
the occlusal surface of the premolars. No reference plaque is used. Trait is scored as
present or absent.
Congenital Absence Molar 3. The third molar may not form in adult individuals.
Trait is scored as present or absent with no reference plaque used.
Traits on Mandibular Dentition
Shoveling Incisor 2. Same as scoring for upper incisor with six possible grades and
use of reference plaque ASU LI (see above for possible scores).
Distal Accessory Ridge Canine. As with the upper canines, the lower canines are
also scored for this trait. Reference plaque ASU LC DAR is used. Scores are from 0
to 5 and same as listed above with upper canine DAR.
Lingual Cusp Number Premolars. The number and relative size of lingual cusps on
the premolars are scored. Reference plaque ASU LP2 cusp is used. Trait can vary in
expression from no lingual cusp to three.
A. No lingual cusp.
0. One lingual cusp.
1. One or two lingual cusps.
2. Two lingual cusps, mesial cusp much larger than distal.
3. Two lingual cusps, mesial cusp is larger than distal.
4. Two lingual cusps, mesial and distal cusps equal.
5. Two lingual cusps, distal is larger than mesial.
6. Two lingual cusps, distal is much larger than mesial.
7. Two lingual cusps, distal is very much larger than mesial.
8. Three lingual cusps, all about equal size.
9. Three lingual cusp, mesial cusp is much larger than medial and/or distal
Anterior Fovea Molar 1. This trait is located on the anterior occlusal surface. It can
be expressed as a depression with a ridge that connects the mesial margin of cusps 1
and 2. It should be scored only in individuals of age 12 or less since it is easily
obliterated from attrition. Reference plaque ASU LM1 anterior fovea is used. There
are five possible grades that can be scored.
0. No expression.
1. Weak ridge with faint groove.
2. Larger ridge and deeper groove.
3. Longer groove.
4. Groove is very long and ridge is robust.
Groove Pattern Molars. Groove pattern on the occlusal surface formed by cusp
contact. No reference plaque is used. There are three possible scores.
Y. Cusps 2 and 3 contact.
+. Cusps 1-4 contact.
X. Cusps 1 and 4 contact.
Cusp Number Molars. Number of cusps present on the lower molars (note: cusp 7 is
not counted in this trait). No reference plaque. Score of 4-6 is possible.
4. Four cusps present.
5. Five cusps present.
6. Six cusps present.
Deflecting Wrinkle Molar 1. This trait records the presence and variety of form of
the medial ridge on cusp 2. It should be scored on individuals 12 years or younger
because it is easily obliterated by wear. Reference plaque ASU LM deflecting
wrinkle is used. There are four possible scores.
0. No expression, wrinkle is absent.
1. Medial ridge straight with mid-point constriction.
2. Medial ridge deflected distally.
3. Medial ridge deflected distally forming an L-shaped ridge.
Middle Trigonid Crest Molar 1. A ridge that connects the middle aspect of cusps 1
and 2. Reference plaque ASU LM mid trigonid crest is used. Trait is scored as
present or absent with two variations on present state.
0. No expression.
IA. Sharp crest connecting cusps.
IB. Wide ridge connects cusps.
Distal Trigonid Crest Molar 1. Similar to the mid trigonid crests, this is a ridge that
connects the distal aspect of cusps 1 and 2. This trait is also scored as present or
absent. Reference plaque developed by Hanihara (1961) for deciduous teeth used as
Protostvlid Molars. This trait can vary in expression from a pit to a cusp occurring
on the buccal surface of cusp 1. Reference plaque Zoller Laboratory LM protostylid
is used. There are eight possible scores for this trait.
0. No expression.
1. Pit in the buccal groove.
2. Buccal groove curved distally.
3. Faint second groove mesial to buccal groove.
4. Second groove more pronounced.
5. Second groove easily seen.
6. Small cusp.
7. Cusp with free apex.
Cusp 5 Molars. The presence of the hypoconulid on the distal occlusal surface of the
molar. It is graded according to size with scores of 0-5 possible. Reference plaque
ASU LM cusp 5 is used.
0. No cusp 5.
1. Very small.
5. Very large.
Cusp 6 Molar 1. The presence of the entoconulid, cusp 6, is lingual to cusp 5. It is
scored according to size in relationship to cusp 5. Reference plaque ASU LM cusp 6
is used. There are six possible grades for this trait.
1. Much smaller than cusp 5.
2. Smaller than cusp 5.
3. Equal in size.
4. Larger than cusp 5.
5. Much larger than cusp 5.
Cusp 7 Molars. Cusp 7, the metaconulid, occurs in the lingual groove between cusps
2 and 4. Reference plaque ASU LM cusp 7 is used. Trait can be scored as one of six
grades from 0-4.
0. No expression.
1. Faint cusp.
1 A. Faint tipless cusp.
2. Small cusp.
3. Medium cusp.
4. Large cusp.
Root Number Canine. The number of roots are counted. Separation is defined in the
same manner as explained above with maxillary premolars. Canine root number is
scored as one or two roots.
Root Number Molars. As with the maxillary teeth, the number of separate roots are
scored. Molars scored as one, two, or three roots.
The data are recorded on a Dental Recording Form. The frequency of occurrence of
a trait is calculated by determining the number of teeth which display presence of the
trait divided by the total number of teeth scorable for that trait. The point (termed
break point or dichotomization point) at which a trait is considered present usually
follows established standards (see Turner, 1985a, 1987) based on each traits
morphological threshold (Haeussler et al., 1988). As many traits as possible should
be used to characterize a population. There is no single biological trait that divides
the world populations and with dental morphological traits, most occur in all the
major groups of people to some degree. These traits are variable in the frequency of
their occurrence; human population variation in dental morphology is mostly
quantitative rather than qualitative, and it is essential to consider as many variables as
possible (Scott and Turner, 1997:11).
Sexual dimorphism in tooth size is quite common but is rarely seen in tooth
morphology (Berry and Berry, 1972; Sjovold, 1973, 1977). It is standard procedure
to pool the sexes (e.g., Coppa et al., 1998; Greene, 1982; Hanihara, 1992a, 1992b)
and all the comparative samples for this study have pooled their data. The
Leptiminus samples were, for most burials, poorly preserved and sex was
indeterminate. Therefore, in the Leptiminus samples both sexes were pooled.
Teeth are normally identical on the two sides, exhibiting a high degree of symmetry.
Differences in size and shape of antimeric teeth that is random between sides is
known as fluctuating asymmetry. In recording dental morphological traits, both
antimeres are scored and the antimere exhibiting the greatest expression of the trait is
used to represent the individual (individual count method). The reasoning behind this
method is that the tooth exhibiting the greatest expression best represents the
underlying genetic potential for that trait (Turner, 1985b). There are other methods
available such as counting either the right or the left antimere for each individual.
The individual count method is used for this study because of its potential to
maximize the genetic information, maximize sample size, and helps promote
comparability between samples with standardizing the method used.
The tooth one chooses to represent the presence of a trait is dependent on the goals
of the study and constraining factors such as the comparable data published from
other researchers. It is recognized that a trait, like Carabellis trait, can occur on
more than one member of a class of teeth, e.g., upper molars. This is predicted by
both the morphogenetic fields and clone models (Butler, 1937,1939; Weiss, 1990).
For biological distance studies, only one tooth in the class is used to represent the
occurrence of the trait in an individual. For most traits, the first tooth in a field (e.g.,
incisor, premolar, molar) usually expresses the trait more fully. Thus, if it is
important to record the greatest expression for any trait, the first tooth should be
used. If reduction or simplification is of particular importance, then the last tooth in
the field should be used. When there is little or no variation in the sample for the key
tooth for a particular trait, it may be more informative to select a different tooth in
that field. For some traits the second tooth in the field is more variable, thus
providing better discrimination of dental relationships between populations. For the
most part, traits are independent of one another with associations occurring within
fields rather than between traits (Berry and Berry, 1972; Sjovold, 1973).
Inter- and intra-observer error has been calculated for dental morphological traits by
Sofaer et al. (1972), Scott (1973), and Nichol and Turner (1986). Results among
these three studies showed concordance. In looking at traits scored as present or
absent (ignoring the differences in rank), the intra-observer concordance was 90%
and inter-observer concordance 87%. In the study by Nichol and Turner (1986),
misclassifications occurred at the present/absent break point one-third of the time,
suggesting that it is at the threshold level of a trait that misjudgements can occur.
To test for intra-observer error, a random sample of228 teeth from 12 burials, was
re-analyzed by the author in the summer of2000, eight months after the initial
recording. The data varied in three ways; 1) changes in the rank score that did not
change the number of occurrences counted as present, 2) changes that did result in
the number of occurrences counted as present, and 3) traits that were observed in one
scoring and not in the other. In the case of the latter, teeth exhibited varying amounts
of attrition and postmortem damage and were considered questionable but were
scored in one episode but not in the other. The number of scores that changed in
rank but did not result in a change of presence/absence were not quantified because
they did not change the overall frequency of the tooth/trait under consideration. The
overall concordance between the results of the original and the retested samples is
98%. There are 17 scores that changed due to a difference in the number of teeth
observed for that tooth/trait and eight that changed due to a difference in counting
the number present/absent. Because of the small sample sizes (i.e., number of teeth
observed for any one tooth/trait combination), a difference of one in either n (sample
size) or in number present, results in a large difference for the frequency of that trait
between the two episodes of scoring. There are only two traits that differ by more
than one in either sample size or number observed as present; Labial Curve for upper
incisor 1 and Distal Accessory Ridge for upper canine. The overall difference is well
within the accepted range as specified in the above studies for intra-observer error.
All teeth/traits possible were scored from the Leptiminus samples. Limiting factors
were poor preservation of many of the burials and attrition on the occlusal surface.
These are problems that are common to all research but do reduce sample size.
Using published data for comparison purposes introduced other limiting factors.
Published data may list only a limited number of traits, may use a different tooth to
represent a trait, or may select other than the standard break point for
dichotomization of the data. The frequency data and dichotomization points for all
samples used in this study are presented in Appendix A. This Appendix also presents
the raw data for all traits for the Leptiminus samples to provide future researchers a
greater latitude in selecting teeth/trait combinations as necessary. The tooth/trait and
break point that will be used for statistical comparisons for this study are summarized
in Table 5.2.
Descriptive and multivariate statistical techniques are used in this study which offer a
means of identifying unique dental morphological traits in the Leptiminus samples and
a method for assessing biological affinity between populations. A simple descriptive
assessment is provided by calculating frequency of occurrence for each tooth/trait in
the sample. The frequency of occurrence of a trait is calculated by determining the
number of teeth displaying the presence of the trait divided by the total number of
teeth scorable for that trait. This allows a basic characterization of each group and a
Table 5.2 Tooth, Trait, and Break Point Combinations for Statistical
LEP samples compared with the following groups with tooth/trait at the
listed break point for present/absent.
TRAIT BREAKPOINT CAR HES MER XGR CAR HES MER XGR GED LAC ETC PCC CAC SUL All Samples
Winging UI1 1/1-4 X X
Shoveling UI1 3-6/0-6 X
Dbl Shovel UI1 2-6/0-6 X X X X
Interup. Groove UI2 +/0-+ X X X X
Tubercle Dent. UI2 2-6/0-6 X X X X
Distal Ridge UC 2-5/0-5 X X X X
Hypocone UM2 2-5/0-5 X
Cusp 5 UM1 1-5/0-5 X
Carbellis UM1 2-7/0-7 X X X X
ParastyleUM3 1-5/0-5 X X X X
Enamel Ext. UM1 1-3/0-3 X X
Root No. UPM1 1/1-2 X X X X
Root No. UM2 3/1-3 X X X X
Peg-shaped UI2 2/0-2 X
Cusp No. LP2 2-9/0-9 X X X X
Anterior Fovea LM1 2-4/0-4 X
Groove Pattern LM2 Y/Y,X,+ X X X X
Cusp No. LM1 6/4-6 X X X X
Cusp No. LM2 4/4-6 X
Cusp No. LM2 5+/4-6 X
Cusp No. LM3 4/4-6 X
Distal Crest LM1 1/0-1 X X X X
Protostylid LM1 1-7/0-7 X X X X
Cusp 7 LM1 1-4/0-4 X
Root No. LC 2/1-2 X X
Root No. LM1 3/1-3 X X X X
Root No. LM2 1/1-3 X X X X
# OF TRAITS 24 18 22 15
North Africa: Leptiminus (LEP), Carthage (CAR), El Hesa (HES), Meriotic (MER), X-Group
(XGR), East Asia (Levant): En Gedi (GED), Italy: Etruscan (ETC), Latini (LAC), Campani (CAC),
Piceni (PCC), Sulmona (SUL).
means of comparisons with other dental samples.
As many traits as possible should be used to characterize a population for affinity
studies (Sjovold, 1973,1977). However, the traits selected should not be correlated.
To test for correlation between traits, Kendalls tau-b and Spearmans rho can be
used. If correlation is found one of the traits should be removed, otherwise
differential weighting of the underlying dimension can lead to erroneous results. In
this study, correlation analysis was run and a few traits eliminated due to correlation
with other traits. Choosing between the traits to keep or remove was a factor of
removing those that were correlated with more than one other trait and to maximize
the number of traits that could be used in the analysis.
There were 10 traits found to have a moderate to strong correlation with significance
at the .01 level using the data from the samples described in Appendix A. This
resulted in the removal of six traits from the subsequent multivariate statistical test,
mean measure of divergence. The traits removed are labial curvature of upper central
incisor, mesial ridge of upper canine, cusp number on upper second premolar,
odontome on upper first and/or second premolar, congenital absence of upper third
molar, and deflection wrinkle on lower first molar. Of the four remaining traits, three
have a weak to moderate correlation and one has a moderate to strong correlation
(Table 5.3). The trait-to-trait combination for these four traits and results of the
correlation tests (Kendalls tau-b and Spearmans rho) are shown in Table 5.3, along
with the significance (p value), and r2. The value r2 indicates how much of the
variance of one trait is related to the second trait (Bernard, 1995). For example, the
amount of variance (r2) for parastyle UM3 and Carabellis UM1 is .21-.39, indicating
that 21-39% of the variance in Carabellis UM1 is related to that of parastyle UM3.
The strongest correlation is between tuberculum on upper lateral incisor with cusp
number on lower first molar. There is no developmental reason for these two traits
to be correlated (Scott and Turner, 1997) and in Irish (1993) comparison of the same
traits with a different sample set resulted in no correlation between these two traits.
It is possible this correlation is due to chance alone. There is a 5% probability that
traits will result in a significant correlation when none actually exists.
Table 5.3. Results of Trait-to-Trait Correlation using Kendalls tau-b and
Correlated Traits Kendalls tau b Spearmans rho p value r2
Tuberculum UI2/Cusp # LM1 .667 .830 .002-.000 ,45-.69
Tuberculum UI2/Cusp # LM2 -.523 -.625 .019-.030 .21-39
Parastyle UM3/Carabellis UM1 .459 .626 .037-.022 .21-.39
Parastyle UM3/Distal Crest LM1 .477 .575 .039-.040 ,23-.33
Questionable data are removed from the study, i.e., one of any pair of correlated
traits or traits where intra-observer error is too great. Traits that are present in
proportions of 0 percent and 100 percent in all samples should also be removed from
the comparisons (Sjovold, 1973). Traits based on too few observations may create a
single variance that is too large and these traits should also be removed from the
study (Sjovold, 1973). The remaining traits can then be used for comparison with
other dental samples to assess phenetic relationships. All questionable traits have
been removed and the list of traits to be used in the multivariate statistic are those
shown in Table 5.2.
Different multivariate methods can be utilized to determine phenetic relationships
between populations. For example, Hanihara (1992a and 1992b) uses the B-squared
distance coefficient. This method has the advantage of providing the contribution
rates of each trait to the distance obtained. Factorial correspondence is used by
Coppa et al. (1998) to compare the samples. An advantage to this method is that
populations and dental variables are represented simultaneously with respect to the
same axes in multidimensional space, thereby permitting visual interpretation of
dental differences among populations and the relative participation of each dental
variable in the dispersion (Coppa et al, 1998:374).
A commonly used method for assessing biological affinity (especially common with
dental morphological data) is C.A.B. Smiths Mean Measure of Divergence (MMD)
with Freeman and Tukey angular transformation to correct for small sample sizes
(Green and Suchey, 1976; Irish, 1993,1998; Lipschultz, 1996; Scott and Turner,
1997; Sjovold, 1973,1977; Turner, 1985a, 1987; Turner and Markowitz, 1990).
This statistic provides a quantitative estimate of biological divergence among samples
based on the degree of phenetic similarity for the entire suite of dental traits. This is
a dissimilarity measure, so a lower number indicates greater affinity than a higher
number. To determine if samples differ significantly, each MMD is compared to its
standard deviation (Sjovold, 1973). If the MMD is greater than two times the
standard deviation, then the null hypothesis (PI = P2, where P is sample population)
is rejected at the 0.025 significance level.
The MMD formula usedis: (5.1)
MMD = E (On-Go)2 [(l/(nn+.5)) + (l/(n a+.5))]
The variance is:
varMMD= 2 l [(l/(n(1+.5)) + (l/(n c+.5))]2
The standard deviation is equal to the square root of the variance.
N, = the number of traits used
O, j = the transformation of the zth trait for the first population
0,2 = the transformation of the zth trait for the second population
na = the sample size for the zth trait for the first population
n2, = the sample size for the zth trait for the second population
Green and Suchey (1976) compared several transformation formulae to assess what
method is best to stabilize the variance of the measures used to calculate biological
distance. Their recommendation is the Freeman and Tukey inverse sine
transformation (formula 5.3). The reasoning for the transformation is as explained by
Sjovold (1973:207); the observed (sample) proportion p gives an increasingly
better approximation of the true (population) proportion P when n (number of
observed occurrences of a trait) increases. When n is small there may be great
discrepancies between p and P. To correct this, and have all proportions be of equal
importance, then p is transformed to a new variable that is independent of P
(parentheses mine). The transformation results in variances in the measures of
divergence dependent on the number of observed occurrences (zz) for the population
sample and not on the proportions of the total population from which the sample is
drawn. This also explains why those traits with a small n would result in a great
amount of variance and should not be used in the divergence calculations.
Transformation formula: (5.3)
0 = Vzsin'1 [l-2k/(n+l)] + ^sin'1 [l-2(k+l)/(n+l)]
k = occurrence of a trait (i.e., number present for that trait)
n = number of observations for that trait
For this study the MMD statistic is applied multiple times using different suites of
traits with different samples. As stated above, as many traits as possible should be
used; however, traits used in each comparison must be the same for all samples.
When using published data, the tooth/trait combination may vary between sources as
well as the present/absent break point. All these factors must correspond to be used
in the MMD statistic (Sjovold, 1973). Table 5.2 detailed the tooth/trait and break
point that is used for comparison with the various samples. For example, the
Leptiminus samples (LEP 10 and LEP 200) can be compared with the Carthage,
Egyptian, and Nubian (CAR, HES, MER, XGR) samples on a suite of 24 traits. The
LEP samples and those from Italy can be compared using 22 traits. The advantage of
this technique is maximization of the dental information used in calculating the MMD
value between each sample pair in the matrix. More traits are used and therefore, are
more likely to reflect the phenetic relationship. The disadvantage is that the MMD
values can not be directly compared with the MMD values obtained in a matrix where
different number of traits and different traits are used.
To compensate for the above disadvantage, the MMD value was also calculated with
all samples but there are only 15 traits common to all samples. The MMD values will
be comparable and reflect the phenetic relationship between all sample pairs. One
disadvantage here is that fewer traits are used and results are not as robust as with a
greater suite of traits. These methods are complimentary and, when used together,
should more accurately reflect the overall relationship between the samples.
The MMD statistic has provided reliable results in studies using non-metric data and
with different vertebrate species (Sjovold, 1977). Results from dental morphological
studies have been compared to results from independent evidence such as blood
group frequency data, linguistic data, archaeological data, metrical, and other non-
metrical data with excellent concordance between resulting measures of biological
affinity (Berry and Berry, 1972; Coppa et al., 1998; Greene, 1982; Irish, 1993;
Prowse and Lovell, 1996; Scott and Turner, 1997; Turner, 1985a; among others).
The final statistical procedures applied will provide a visual reference of the results
using the MMD values. The two procedures are cluster analysis and
multidimensional scaling (MDS). Cluster analysis results in cluster dendograms to
show degrees of relationship according to branching points in the display. Wards
method has provided excellent results with analyses of known cluster structure
(Aldenderfer and Blashfield, 1984). Wards method groups units according to the
least increase in error of the sum of squared deviations. This process continues until
all samples are compared and the number of groups is reduced to one. The second
method, MDS, provides a two-dimensional plot of the relationship among N variables
(Bernard, 1995). The plot is like a map of the relative relationship between the
samples. The results of the two visual methods should compliment each other.
RESULTS AND DISCUSSION
The first analysis of the dental morphological data presented is the frequency of each
of the dental traits. Frequency is calculated by counting the number of teeth
displaying the presence of a trait divided by the total number of teeth scorable for that
trait. The frequency of the dental traits offers a means to characterize each sample in
terms of the entire suite of traits and is the basis of the phenetic comparison between
The results for the Leptiminus samples, with the number of occurrences by grade
scored for each trait, are presented in the Raw Data for the Leptiminus Samples in
Appendix A. The frequency, with the dichotomization (break point) for all the
samples, is presented in the Frequency Data for All Samples, also in Appendix A.
The Leptiminus samples are characterized by the following dental morphological
traits. Leptiminus Site 10 and 200 have a high frequency of occurrence 60
percent) of Hypocone on upper molars, 3-rooted UM1 and UM2, Y-cusp pattern on
LM1, and 4-cusps on LM2 and LM3. The Leptiminus samples have a mid-range
occurrence (30-60 percent) for single-rooted UP1, >l-cusp on LP2, Anterior Fovea
on LM1, Y-cusp pattern on LM3, Protostylid on LM1 and LM2, and single-rooted:
LM1. Traits that occur in low frequency are Winging UI1, Labial Curve UI1,
Shoveling and Double Shoveling on all incisors, Interruption Groove upper incisors,
Tuberculum Dentale UI1 and UC, Canine Mesial Ridge, Canine Distal Accessory
Ridge UC and LC, 3-cusp UP1 and UP2, 5-cusp UM3, Carabellis Cusp upper
molars, Parastyle upper molars, Enamel Extension UM1, Peg-shaped UI2 and UM3,
Odontome upper premolars, Congenital Absence UM3, >l-cusp UP1, X-cusp pattern
on lower molars, 6-cusp LM1, >4-cusp LM2, Deflecting Wrinkle LM1, Middle
Trigonid Crest LM1, Distal Trigonid Crest LM1, Cusp 5, 6, or 7 on lower molars, 2-
rooted LC, 3-rooted LM1, and 1-rooted LM2.
A few of the traits vary in the range of occurrence between the Leptiminus samples.
Site 10 is in the high range for Hypocone on UM2 and UM3 and Site 200 is in the
mid-range. Site 200 has a high occurrence of single-rooted UP2 with Site 10 in the
mid-range. Site 10 is in the mid-range for Tuberculum UI2, Cusp 5 UM1 and UM2,
3-rooted UM3, and Protostylid LM3 with Site 200 in the low range for those traits.
Site 200 has a mid-range occurrence of Carabellis on UM1 and Y-cusp pattern on
LM2 with Site 10 in the low range for these two traits. Overall there seems to be
close correspondence between the frequencies of traits between Site 10 and Site 200.
Examining the correspondence between Leptiminus and the other samples, there are
fewer differences in the frequency of occurrence of traits with the Carthage,
Egyptian, Nubian, and Leventine samples than with all the Italian samples. These
relationships will be examined statistically with the MMD calculation.
Mean Measure of Divergence Analysis
The MMD statistic is a quantitative estimate of biological divergence based on the
degree of phenetic similarity for a suite of dental traits. This is a dissimilarity
measure, so a lower number indicates a greater affinity than a higher number. The
first three comparisons include samples common to a region. For example, the first
group is the North African samples, the second is North African/Near East, the third
is the Italian Peninsula. All comparisons included the Leptiminus samples. The
fourth comparison includes all samples. The MMD statistic was run three times
using different regional samples to maximize the number of traits used in those
calculations. The fourth calculation used fewer traits but included all the samples.
In the tables presenting the results of the analysis, the MMD statistic is shown in the
upper portion of the table and the standard deviation is in the lower portion of the
table in parentheses. To determine significance, the MMD is compared to its
standard deviation (SD) (Sjovold, 1973). If the MMD is > 2 SD, the MMD is
significant at the .025 level and is shown in bold text. If the MMD is > 3 SD, it is
in bold and underlined text, and indicates samples which differ from each other at the
.01 level. An insignificant MMD is, therefore, in plain text and indicates either that
the samples are phenetically similar or that the MMD is based on a suite of traits with
individual traits that have small sample sizes (less than 10 observations). Small
sample sizes can result in an excessively large SD (Sjovold, 1973,1977). Caution is
advised when the MMD value is close to the critical value (2 SD) and in these cases
dissimilarity is not assumed to be proved (Sjovold, 1973). The following four tables
present the results of the MMD calculations.
Table 6.1. MMD Results with First Sample Set
LEP 10 LEP 200 CAR HES MER XGR
LEP 10 0.038 0.027 0.070 0.112 0.106
LEP 200 (0.048) 0.011 0.087 0.105 0.149
CAR (0.053) (0.049) 0.000 0.108 0.117
HES (0.034) (0.061) (0.036) 0.052 0.051
MER (0.031) (0.028) (0.033) (0.014) 0.000
XGR (0.048) (0.045) (0.050) (0.033) (0.030)
(SD); bold = MMD>2*SD; bold = MMD>3*SD; Plain text = non-significant MMD; Number of
dental traits used is 24 (refer to Table 5.2 for specific traits). Leptiminus=LEP; Carthage=CAR; El
Hesa=HES; Meriotic=MER; X-Group=XGR.
Table 6.2. MMD Results with Second Sample Set
LEP 10 LEP 200 GED CAR HES MER XGR
LEP 10 0.013 0.185 0.045 0.091 0.131 0.178
LEP 200 (0.057) 0.207 0.003 0.061 0.064 0.132
GED (0.047) (0.044) 0.206 0.138 0.153 0.119
CAR (0.064) (0.058) (0.048) 0.000 0.146 0.136
HES (0.040) (0.036) (0.027) (0.043) 0.077 0.048
MER (0.038) (0.033) (0.025) (0.040) (0.016) 0.000
XGR (0.056) (0.053) (0.045) (0.059) (0.038) (0.034)
(SD); bold = MMD>2*SD; bold = MMD>3*SD; Plain text = non-significant MMD; Number of
dental traits used is 18 (refer to Table 5.2 for specific traits). Leptiminus=LEP; Carthage=CAR; El
Hesa=HES; Meriotic=MER; X-Group=XGR; En Gedi=GED.
Table 6.3. MMD Results with Third Sample Set
LEP 10 LEP 200 LAC ETC PCC CAC SUL
LEP10 0.032 0.228 0.095 0.202 0.067 0.088
LEP 200 (0.045) 0.223 0.154 0.235 0.108 0.262
LAC (0.059) (0.052) 0.044 0.030 0.166 0.132
ETC (0.037) (0.033) (0.045) 0.000 0.022 0.023
PCC (0.029) (0.024) (0.040) (0.017) 0.134 0.063
CAC (0.039) (0.035) (0.049) (0.028) (0.019) 0.119
SUL (0.044) (0.040) (0.051) (0.033) (0.024) (0.034)
(SD); bold = MMD>2*SD; bold = MMD>3*SD; Plain text = non-significant MMD; Number of
dental traits used is 22 (refer to Table 5.2 for specific traits). Leptiminus=LEP; Latini=LAC;
Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL.
Table 6.4. MMD Results with All Samples
LEP10 LEP 200 LAC ETC PCC CAC SUL GED CAR HES MER XGR
LEP10 0.006 0.340 0.134 0.279 0.055 0.146 0.251 0.070 0.092 0.115 0.188
LEP200 (0.058) 0.250 0.129 0.222 0.095 0.275 0.268 0.040 0.092 0.085 0.176
LAC (0.083) (0.074) 0.057 0.047 0.235 0.180 0.061 0.478 0.275 0.171 0.137
ETC (0.048) (0.043) (0.064) 0.006 0.034 0.010 0.142 0.186 0.064 0.096 0.110
PCC (0.038) (0.031) (0.057) (0.022) 0.194 0.085 0.216 0.437 0.214 0.171 0.238
CAC (0.052) (0.046) (0.069) (0.036) (0.025) 0.144 0.178 0.099 0.102 0.137 0.201
SUL (0.056) (0.050) (0.070) (0.041) (0.030) (0.043) 0.206 0.314 0.129 0.157 0.164
GED (0.046) (0.040) (0.063) (0.031) (0.020) (0.033) (0.039) 0.277 0.174 0.172 0.141
CAR (0.061) (0.056) (0.075) (0.047) (0.036) (0.049) (0.055) (0.045) 0.019 0.199 0.190
HES (0.042) (0.037) (0.060) (0.028) (0.016) (0.030) (0.035) (0.026) (0.041) 0.085 0.064
MER (0.039) (0.033) (0.058) (0.024) (0.012) (0.026) (0.031) (0.022) (0.037) (0.018) 0.000
XGR (0.061) (0.057) (0.073) (0.049) (0.038) (0.050) (0.057) (0.047) (0.062) (0.044) (0.039)
(SD); bold = MMD>2*SD; bold = MMD>3*SD; Plain text = non-significant MMD; Number of dental traits used is 15 (refer to Table 5.2
for specific traits). Leptiminus=LEP; Latini=LAC; Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL; Carthage=CAR; El
Hesa=HES; Meriotic=MER; X-Group=XGR; En Gedi=GED
As stated above, an insignificant MMD can indicate a phenetic similarity or can be
due to small sample sizes (for individual traits in the suite of traits used) resulting in a
large SD. If the latter occurs a Type I error may result where the null hypothesis (PI
= P2) is accepted when it should have been rejected. Several samples contain traits
that are based on less than 10 observations. The number varies depending on which
suite of traits are being used in the four MMD calculations. The samples that have
traits with less than 10 observations are Leptiminus 10 (6 to 8 depending on traits
used), Leptiminus 200 (2 to 4), Carthage (1 to 3), X-Group (5), Latini (4), and
Campani(l). Ameanof 17 out of 151 traits (11.3 percent of all traits) are based on
less than 10 observations for a particular dental trait. When analyzing the resulting
MMD values the possibility of a Type I error will be taken under consideration.
Comparison of MMD Results
Tables 6.1-6.4 show a consistency of results for the relationships between various
groups and some disparity between others. All four calculations consistently show
the two Leptiminus samples to be phenetically similar. Both of these samples do
have several traits with less than 10 observations and thus a large SD. However, the
results are consistent with the different number of traits in each of the calculations.
The individuals all lived in the same small town and it is quite likely that this reflects a
close biological relationship. The Leptiminus samples also show a consistent
phenetic similarity to the Carthage sample. Using a similar line of reasoning, it is
quite likely that this indicates a close biological relationship with Carthage which may
represent a parent population.
The Meriotic and X-Group Nubians consistently show a close relationship with each
other which would confirm the possible genetic continuation of the X-Group from
the Meriotic group (Berry and Berry, 1972; Greene, 1967,1972, and 1982; Greene et
al., 1967; Irish, 1993). The Egyptian sample (El Hesa) shows a significant
difference, phenetically dissimilar, relative to the Meriotic Nubians but indicate a
phenetic similarity with the X-Group Nubians. Phenetic similarity is indicated with
no significant difference between El Hesa and the Carthage sample. The southern
Levant sample (En Gedi) appears to be significantly different from most the other
samples, even those from Egypt and Nubia despite being fairly close geographically.
The exception is phenetic similarity between En Gedi and one Italian sample (Latini).
In the Italian samples, the Etruscan group appears to share a close phenetic
relationship with all the other Italian samples. This is consistent with modem blood
group analysis (Cavalli-Sforza et al., 1994). If the local population of southern
Tuscany had a strong demographic development at some early time (in the case of
Etruscans, the beginnings of the Iron Age around 3,000 years ago), and if later
migration from the outside were limited, then the local gene pool would be
reasonably resistant to later modification Cavalli-Sforza et al. (1994:279). Before
the Roman civilization, the Etruscan civilization developed and flourished. The city
of Rome was originally a small village. Romans had little genetic influence because
their original population size was small and grew with the in-migration of people who
lived around the Italian peninsula.
One Italian sample (Campani from south of Naples) indicates a phenetic similarity
with Leptiminus 10. Leptiminus 10 does have several traits based on fewer than 10
observations and this may have resulted in a Type I error. However, no other close
relationship is seen with any other Italian sample. If sample size was a problem, other
relationships may be expected, especially given the phenetic similarity of the Etruscan
sample to all other Italian samples. It is plausible that the Italian settlers were from
that region of Italy, hence, the close phenetic relationship seen.
The Egyptian sample, El Hesa, shows a significant difference to the Leptiminus 10
sample. The relationship between El Hesa and Leptiminus 200 shows a disparity in
the results. The results in the first two calculations indicate phenetic similarity and
the fourth calculation is significantly different, thus phenetically dissimilar. The first
calculation, Table 6.1, includes the greatest number of traits and is a more robust,
thus more accurate, measurement of the relationship. In addition, since two of the
three calculations indicate phenetic similarity, it is likely that they represent the actual
phenetic relationship. The disparity may be due to the relationship that both have
with Carthage. There may be a shared common ancestor, possibly the Phoenicians.
The traits that differ between the first and fourth calculations may be ones that are
important for gauging the phenetic dissimilarity. The traits included in that grouping
but not in the last may be the traits shared with the common ancestor.
There is a disparity in the results between Leptiminus 200 and Meriotic Nubians.
Tables 6.1 and 6.4 show a significant difference and Table 6.2 shows no significant
difference. Given that two of three calculations indicate a significant difference, it is
likely that those reflect the actual relationship and these samples are not phenetically
Visual Statistical Methods
For data reduction and interpretation, two visual statistical methods were used. The
first is Cluster Analysis using Wards method. This results in a dendritic diagram
showing the relationship of the samples based on the MMD values. The resulting
cluster dendrograms are shown in Figures 6.1- 6.4. In the dendrograms the bars
connecting samples represent their degree of similarity where shorter bars joining
samples are more similar than those joined by longer bars (Kachigan, 1986).
Figure 6.1 depicts the calculations in the first MMD run with the North African
samples. In the dendrogram the clustering of samples follows well with the results of
the MMD statistic. The samples that have phenetic similarity as measured by the
MMD statistic are clustered together. Figure 6.2 also corresponds well with the
relationships as indicated by the MMD statistic in the second run. The En Gedi
sample (Levant sample) was not phenetically similar to any of the other samples in
this set. The cluster dendrogram placed it alone and with a relatively long bar
connecting it with the Meriotic and X-Group samples. This suggests that it is
different but with a degree of similarity closer to the Meriotic and X-Group samples
than with the other North African samples.
Figure 6.1. Cluster Dendrogram with the First Sample Set
Leptiminus=LEP; Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR
Figure 6.2. Cluster Dendrogram with the Second Sample Set
0 5 10 15 20 25
Leptiminus=LEP; Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR En Gedi=GED
Figure 6.3. Cluster Dendrogram with the Third Sample Set
0 5 10 15 20 25
Leptiminus=LEP; Latini=LAC; Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL
Figure 6.4. Cluster Dendrogram with All Samples
O 5 10 15 20 25
Leptiminus=LEP; Latini=LAC; Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL;
Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR; En Gedi=GED
The dendrogram depicting the MMD run with the Italian samples and Leptiminus is
shown in Figure 6.3. This dendrogram does not show all the relationships as
anticipated given the MMD results. The problem is in showing the phenetic similarity
of the Campani sample with both the Leptiminus 10 and the Etruscan sample. The
clustering program placed the Campani sample in close proximity with Leptiminus 10
with no indication of the phenetic relationship with the Etruscan sample. The
dendrogram also places Leptiminus 10 as closer to Campani than Leptiminus 200.
Otherwise the dendrogram does show two main groupings, one with the Italian
samples and the other with the North African.
Figure 6.4 shows the results of the cluster dendrogram with the results from the
fourth MMD run with all the samples. This dendrogram also does not depict the
phenetic relationships as well as anticipated given in the MMD results. For example,
the Campani sample is shown to be close to El Hesa in the dendrogram and these two
samples are not phenetically similar according to the results of the MMD (refer to
Table 6.4). Leptiminus 10 and 200 are clustered together but are shown further from
Carthage than Campani and El Hesa. A two-dimensional dendrogram places data
into a hierarchical relationship and does not always result in a satisfactory
classification when portraying multidimensional data (Wilcock, 1995). The
multidimensional scaling visually represent the phenetic relationships better than the
cluster analysis with the last two MMD runs.
MDS is a complementary method to cluster analysis for data reduction and
interpretation. MDS offers a visual representation of the MMD results that is easy to
understand and that represents the affinities between the samples better than the
cluster dendrograms. In MDS, the MMD values are considered as Euclidean
distances. Those that share lower values are nearer each other and those with higher
values are further apart. Figures 6.5 6.8 are two-dimensional models of the
relationships between the various samples. It is more difficult to discuss and portray
results when more than two or three dimensions are used and defeats the goal of data
reduction for interpretation (Kachigan, 1986). A MDS stress value, r2 value, and r
value are shown at the bottom of each of the diagrams. Stress is the statistical
measure of difficulty or loss of information for a particular dimensional
representation. A stress value of .15 or lower is usually considered satisfactory
(Kachigan, 1986:418). The stress values for the four calculations vary from .08 to
.13. The r2 value is a measure of variance between the MMD and MDS values
(Bernard, 1995). For example, in Figure 6.5, the r2 value is .951 which indicates that
95 percent of the variance is explained by the MDS model. The square-root of the
value, or r, is the correlation coefficient. In the same example, r = .975, the data
matrices are highly correlated suggesting that the model is an accurate representation
of the MMD dental relationship.
Figure 6.5. Multidimensional Scaling Model with the First Sample Set
-2.0 -1.5 -1.0 -.5 0.0 .5 1.0 1.5
Stress = .086, r2 = .951, r = .975
Leptiminus=LEP; Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR
# * * / CAR'. i
* # . i
fXGR \ LEP 10;
i MER ; i LEP 200 w .
* ' J 1 i 1
Figure 6.6. Multidimensional Scaling Model with the Second Sample Set
GED .LEP 200 LEP 10
; XGR ;
-2.0 -1.5 -1.0 -.5 0.0 .5 1.0 1.5
Stress = .102, r2 = .935, r = .966
Leptiminus=LEP; Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR; En Gedi=GED
Figure 6.7. Multidimensional Scaling Model with the Third Sample Set
LAC m .
% \ % % LEP 200 4
P.CC ETC'. t t 4 4 4 *
* 1 1 t \ CAC i
* t * SUL / t LEP 10 / /
-1.5 -1.0 -.5 0.0 .5 1.0 1.5 2.0
Stress = .077, r2 = .957, r = .978
Leptiminus=LEP; Latini=LAC; Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL
Figure 6.8. Multidimensional Scaling Model with All Samples
Stress = .133, r2 = .891, r = .943
Leptiminus=LEP; Latini=LAC; Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmona=SUL;
Carthage=CAR; El Hesa=HES; Meriotic=MER; X-Group=XGR; En Gedi=GED
In the MDS models, the circled areas indicate clusters according to the
neighborhood method of MDS interpretation which groups samples according to
their positions relative to each other and the half or quadrant of the graph they appear
(Kruskal and Wish, 1978). In Figure 6.5, the close phenetic relationship between the
Carthage and Leptiminus samples, El Hesa and X-Group Nubians, and the X-Group
and Meriotic Nubians is clear. In general, the Carthage and Leptiminus samples are
shown to be more similar to each other than to the Egptian and Nubian samples.
Figure 6.6 portrays the relationships from the second MMD run. In this model El
Hesa is grouped with the Carthage and Leptiminus samples because of the close
phenetic relationship between El Hesa and Carthage. The X-Group and Meriotic
Nubians are grouped together and En Gedi is separate from the other two groupings.
The MDS model for the third MMD run with the Italian samples and Leptiminus
(Figure 6.7), appears to better represent the relationship between the samples than
the cluster dendrogram although the same two main groups are clear in both. The
Campani sample is grouped close to Leptiminus 10 but the phenetic relationship to
the Etruscan sample is suggested because it is the nearest Italian sample. The
phenetic relationship of the Italian samples indicates a division between the affinity of
northern Italian populations and southern Italian populations. This was also seen in
the modem blood group analysis (Cavalli-Sforza et ol., 1994). Northern Italians are
more similar to central Europeans, whereas southern Italians are closer to other
Mediterranean people (Cavalli-Sforza et al., 1994:278).
Figure 6.8 is the MDS model based on all the samples. This model appears to
visually represent the expected relationship between samples better than the cluster
dendrogram. In Figure 6.8, the diagonal line indicates the Italian samples (below the
line) and suggests the phenetic relationship of the Italian samples relative to the
Etruscan sample. The similarity of the Campani sample with other Mediterranean
groups is shown with the grouping of the Campani, Egyptian (El Hesa), Carthage,
and Leptiminus samples together. This indicates that these samples are more similar
to each other than to all the others. The other samples create a second main cluster
with suggestions as to sub-groupings within that cluster. Sub-groupings would
include the X-Group and Meriotic Nubians together, En Gedi with Latini, and
Etruscan with Piceni and Sulmona. However, the Latini sample is not as close to the
Etruscan sample as anticipated given the phenetic similarity. Additionally, X-Group
Nubia is shown as close to En Gedi despite the phenetic dissimilarity indicated by the
MDS results. Neither the cluster analysis nor the MDS models give a complete
concordance of the phenetic relationships as indicated by the MMD results. In
general, the MDS models are easy to interpret and visually represent the expected
relationships better than the cluster analysis dendrograms.
Discussion of Results
North Africa borders the Mediterranean Sea and as such the populations in North
Africa share many characteristics (including dental morphology) with other
Mediterranean groups, in both Europe and the Near East. In a global perspective
North Africa is grouped with Western Eurasia (Scott and Turner, 1997). Table 6.5
shows the general occurrence of nine dental traits for five major world populations.
These traits have been found to be characteristic of various populations around the
world in their frequency of occurrence (Hrdicka, 1920; Scott, 1973; Haeussler et al.,
1989; Kraus, 1951; Turner, 1985a) and illustrates the differences between them.
Table 6.5. Dental Trait Frequency for Five World Populations
Trait LEP W. Europe S. Africa N.&S. America SE. Asia Australia
Shoveling UI1 Low Low Low High Middle Low
Canine Mesial Ridge Low Low Middle Low Low Low
Carabellis UM1 Middle High Middle Low Low Middle
Cusp 5 UM1 Middle Low Middle Low Middle High
Cusp 6 LM1 Low Low Middle Middle Middle High
Cusp 7 LM1 Low Low High Low Low Low
4-Cusp LM2 High High Middle Low Middle Low
>l-Root UP1 Middle Middle High Low Middle Middle
2-Root LC High High Low Low Low Low
N. & S. American is American Indians, prehistoric; LEP is Leptiminus. Table is adapted from Scott and Turner, 1997:177-236.
In general the Western Europe dental morphology suggests dental simplification.
There are fewer cusps and ridges then are found in the more complex dentition of
sub-Saharan groups or Southeast Asians. The dental morphology of the Leptiminus
samples follows the general pattern of the Western European populations but not
completely. The Leptiminus samples have a low occurrence of shoveling, canine
mesial ridge, cusp 6, and cusp 7, a mid range occurrence of >l-root UP1, and a high
occurrence of 4-cusp LM2 and 2-rooted LC. However, there is a mid range
occurrence for cusp 5 UM1 and Carabellis UM1. This variation may be explained
by some influence from sub-Saharan Africa. The Berbers are thought to be
indigenous to North Africa (Hiemaux, 1975). They are descendants of Capsian and
Mechta people. The Berbers had contact with other North African people, including
Egyptians and later with the Carthaginians. It is likely that through either the Mechta
people, whose origins are not well known, or through later contact with other
African groups, some sub-Saharan traits became part of the gene pool. It is also
possible that this is a result of local mutation rather than an ancient common heritage.
However, a more parsimonious explanation is that through trade and contact with
other African groups some sub-Saharan traits became part of the gene pool.
Carthaginian Biological Distinctiveness
In general, the dental morphology of the Leptiminus samples, like that of Western
Europe, suggest a simplification of the dentition. There are fewer cusps and ridges
than are found in the mass-additive traits of the Sub-Saharan Africans (Irish, 1993).
This simplification of dental morphology is shared by other North African groups
(Irish, 1993). The dental morphological data of Leptiminus and Carthage resulted in
no significant difference for MMD values in all calculations. In other words, the
results indicate a close biological affinity among the people of Leptiminus and
Carthage. The close phenetic relationship implies that the Carthage and Leptiminus
samples are from one biological population. Carthaginians are probably the parent
population as suggested in historic references (Lazreg and Mattingly, 1992). The
non-significant MMD values, indicating phenetic similarity, support the hypothesis
that the Carthaginians are a biologically distinct population.
Populations that are genetically different can come into contact resulting in an
exchange of genes. Such an expansion of a population into occupied areas has
occurred due to innovations such as agricultural advances, technical innovations, or
military conquests. Some of the most important expansions for the Mediterranean
area include the Phoenicians advances into the Western Mediterranean because of
their nautical improvements (Cavalli-Sforza et al., 1994). The Phoenicians may share
common ancestry with the Egyptians because both originated in the Levant (West
Asia) (Mourant, 1983). In addition, both groups had contact with the Greeks and
with each other. The Carthaginians also had contact with 26th Dynasty Egyptians
(672-525 B.C.) (Warmington, 1981). A close phenetic relationship between
Carthage and El Hesa may be due to gene flow, from common ancestry, or both.
The Phoenicians established the city of Carthage in an area inhabited by the Berbers.
The exact nature of the relationship between the Carthaginians and local Berbers is
not known (Warminton, 1981). It is quite likely that some gene flow did occur
resulting in the Carthaginian population being a Phoenician/North African hybrid.
Carthaginians continued to have contact with Egypt, in part because of their nautical
expertise. This continuing relationship may be reflected in the close phenetic
relationship between El Hesa and Carthage. However, there may be less direct
contact between El Hesa and Leptiminus and thus the phenetic relationship between
them is less clear.
Egypt was a major center of Mediterranean culture. It has been invaded and ruled by
many diverse groups and has had contact with many others. It is not surprising that
there is phenetic similarity with Carthage and a less clear phenetic relationship with
Leptiminus. The disparity is likely a result of different genetic influences both have
had. El Hesa is likely to have had exchange with many other groups including
Greeks, Nubians, and Levantines (including Phoenicians). Leptiminus is a
Carthaginian population (probably a Phoenician/North African hybrid) with
continuing genetic influences from other Carthaginians, possibly continuing exchange
with the Berbers, and influence from Italy. Thus, the relationship between El Hesa
and Leptiminus includes some shared ancestry but they are diverging because of their
own unique histories.
The phenetic relationship between Carthage and El Hesa and to a lesser extent
Leptiminus and El Hesa, does not negate the hypothesis of biological distinctiveness
of the Carthaginians. It does show the complexity of relationships in the
Mediterranean due to common ancestry and movements of people throughout history
(see Figure 6.9). The relationship between Carthage and Leptiminus was shown to
be phenetically similar in all MMD calculations. There is undoubtedly a close
relationship between these groups. Future research can help to qualify the biological
distinctiveness of the Carthaginians and help to define the range of variation in dental
morphology for this population.
Figure 6.9 Flow Chart Depicting Hypothesized Contact Between Various
Biological Influence by Roman Expansion
The military expansion of Rome resulted in political, social, cultural, economic, and
linguistic influence throughout the invaded lands. Roman rule eventually
encompassed all of the Mediterranean region. Demic diffusion occurs when military
expansion is accompanied by migration into the invaded lands and there is gene flow
between the two populations. Admixture would result in a hybrid between the
conquering and invaded populations. The dental morphological data would thus
indicate a close phenetic relationship to the migrating Roman population if sufficient
gene flow between the two populations occurred.
Biological influence with the military expansion by Rome appears to be present but
minimal as indicated by the dental morphological relationships. Table 6.4 shows the
MMD values for all the samples. Of the five Italian samples (Latini, Eruscan, Piceni,
Campani, and Sulmona), two indicate biological affinity with two non-Italian
samples. Campani (from near Naples) is shown to be pheneticaJly similar to
Leptiminus 10 and is discussed in the following section on the Italian Settlers in
Leptiminus. Latini is shown to be similar to En Gedi (Levant sample). The phenetic
relationship between Latini and En Gedi could be a result of admixture or the burial
of Italian settlers in the same cemetery. It is also possible that the phenetic
relationship indicated by the MMD values is spurious. The Latini sample includes
four of the 15 traits that are based on less than 10 observations which could have
resulted in a large SD and a Type I error. Corroborating evidence from other
sources, such as archaeological, other biological data, historical records, would help
to validate this apparent phenetic relationship.
Italian Settlers in Lentiminus. Historical references assert that Italian settlers were
installed in Leptiminus circa 1st century B.C. (Lazreg and Mattingly, 1992). The
dental morphological data (Tables 6.3 and 6.4) indicate biological affinity between
Leptiminus 10 and Campani, the Italian sample from Naples area. However, the
Campani group and Leptiminus 200 do not indicate phenetic similarity in any of the
calculations. This suggests that there is some phenetic similarity but only between
one Italian sample and one Leptiminus sample.
When one population moves into an inhabited area of a genetically separate
population, admixture and gene flow can occur. Gene flow can result in a complete
replacement of the gene pool in the long run. However, this is contingent on such
factors as the size of both populations and the amount of gene flow between them.
Restrictions to gene flow (genetic barriers) can occur for many culturally defined
reasons, such as economics, status, class, religion, or other social reasons.
Admixture between two groups will result in a hybrid population, intermediate
between the two that will likely be unique (Cavalli-Sforza et al., 1994). The change
is usually slow with small changes per generation. It may be difficult to detect
original populations depending on amount of time and gene flow rate. Estimating the
residual genotype (gene pool of original populations) can be calculated following the
formulae of Cavalli-Sforza et al. (1994). Table 6.6 shows the residual genotype after
a per-generation gene flow of 1%, 5%, 10%, and 20% over 500 years (Cavalli-Sforza
et ah, 1994:55). The point is that after 500 years with 10 or 20 percent gene flow
the original populations genotype may be nearly completely replaced. It may be
difficult to determine genetic origins of a group unless there has been little gene flow
The Italian settlers were reported to have arrived at Leptiminus in the 1st century
B.C. and the cemetery of Leptiminus 10 is dated to A.D. 2nd 4th century. There is at
least 200 years, or 8 generations (assuming 25 years per generation) of co-habitation.
At its height of occupation, Leptiminus is estimated to have had a population of
approximately 3,000 people (D. Stone and L. Stirling personal communication,
2001). This is based on the estimated number of households during the time period of
A.D. 2nd 4th century. The size of the population of Italian settlers is unknown. The
MMD results indicates biological affinity between the Leptiminus 10 sample with
both the other Carthaginian samples and the Campani sample. If there had been a
high rate of gene flow, the original Carthaginian population affinity may not have
been indicated. It is likely that a new, unique, hybrid group (Leptiminus and
Campani) had not been generated but that some percentage of admixture may have
Table 6.6. Percentage of Residual Genotype After a Per-Generation Gene Flow
of 1%, 5%, 10%, and 20%
Time Gene Flow % per Generation
Generations Years 1 5 10 20
0 0 100 100 100 100
1 25 99 95 90 80
4 100 96 81 66 41
8 200 92 66 43 17
20 500 82 36 12 1.1
Uses formulae 1.17.2, Cavalli-Sfoza et al., 1994:55.
Another possibility exists to explain the MMD results. It is possible that the Italian
settlers are buried in the same cemetery (Site 10) as were Leptiminus Carthaginians
and that little to no admixture occurred. Little similarity was found between the
Campani and the Leptiminus 200 samples, suggesting that no Italian settlers were
buried at Site 200 and little or no gene flow occurred. The cemeteries are in different
parts of the city but otherwise exhibit the same range of tomb types and grave goods.
There is no obvious difference in status between the individuals buried at the two
cemeteries (Stirling, 2001, personal communication). The data suggest that Italian
settlers phenetically similar to those from near Naples were at Leptiminus. The data
are inconclusive about whether admixture occurred between the two populations. If
gene flow occurred and given that the two Leptiminus cemeteries represent a similar
range of classes of individuals, then some similarity between Campani and Leptiminus
200 would have been expected. This suggests that the Italian settlers and Leptiminus
Carthaginians may not have exchanged genes and it is the presence of the settlers
themselves and not a Leptiminus/Campani hybrid that resulted in the phenetic
relationship. The hypothesis of Italian presence is supported but it is not clear if gene
flow occurred as well.
A few burials in Leptiminus 10 appeared different than most other individuals buried
at Leptiminus. These few individuals were taller and more robust than the other
individuals in the cemetery. These may represent the descendants of the Italian
settlers. This could be tested by removing the data associated with those individuals
and re-calculating the MMD statistic. If the phenetic similarity with Leptiminus 10
and Campani is no longer indicated than it would support a hypothesis that the
settlers themselves are buried at the same cemetery. One possible problem with this
test is that it would reduce the sample size for Leptiminus 10. The tall and robust
individuals were the best preserved skeletal remains and possessed a full set of
dentition which contributed to the overall sample size.
Sicily and southern Italy were influenced by the Greeks who settled and lived in the
area around the 8th to 9th centuries B.C. Carthaginians may share some ancestral
traits because of the Greek influence in the Phoenician population (possibly
established by Mycenaean Greeks who settled in the Levant in 1200 B.C.). The
Carthaginians also were known to have colonized Sardinia. It is possible that some
influence and gene flow could have occurred between the Sardinians and occupants
in the area south of Naples. However, if there were common ancestry or admixture,
then a close phenetic relationship should have been seen between Campani and
Carthage and Campani and Leptiminus 200. No phenetic similarity was indicated
between these groups with the dental morphological data. Therefore, the close
phenetic relationship seen between Campani and Leptiminus 10 is unlikely to be due
to shared common ancestry. It is more likely to be a recent population movement of
Italian settlers into Leptiminus that results in the biological affinity between the
Campani and Leptiminus 10 samples.
The results of the MMD calculations do support the hypothesis of biological
influence by Roman expansion. There is a phenetic similarity seen between Latini and
En Gedi samples and with the Campani and Leptiminus 10 samples. Whether gene
flow actually occurred between these groups is not clear. Migration of Italian settlers
is likely to have occurred in Leptiminus as the historical references claim. Future
research could help to verify if Italians also migrated to En Gedi. As additional
samples from around the Mediterranean become available, this hypothesis can be re-
tested to gain a better understanding of the biological influence of Roman expansion
throughout the region.
Isolation bv Distance
Isolation by distance is a model evaluating biological affinity among groups based on
the tendency of groups to exchange mates with those that are near in distance.
Distance alone predicts affinity; no other barriers to gene flow are assumed to be
Figure 6.8, the MDS model based on the MMD values for all samples, does generally
depict the geographic relationship of the samples. This figure shows the Italian
samples all in the lower left portion of the graph and the North African/Near East
samples in the upper right.
To test for the strength of the relationship between distance and affinity., correlation
analysis was used. The geographic distance between each sample pair was
determined and the MMD value was used as a measure of affinity. Figure 6.10 is a
scatter plot of all the sample pairs with their distance and affinity. A correlation of
0.120 with a P-value of 0.336 was obtained The scatter plot depicts a division of the
sample by distance. There is a random scatter of points when distance is great (>
1500 km). A second calculation was done using a subset of sample pairs that are less
than 1500 km in distance. The result with the subset is a moderate positive
correlation of 0.515 with a P-value of 0.002 (Figure 6.11). This subset includes the
Italian, Carthage, and Leptiminus samples (Table 6.7) but excludes the pair-wise
comparison of the Italian, Carthage, and Leptiminus samples to the Egyptian, Nubian,
and Levantine samples.
Figure 6.10 Scatter Plot of Distance and MMD Value for All Sample Pairs
Figure 6.11 Scatter Plot of Distance and MMD Values with a Subset of Sample
The solid circles are MMD values that indicate the samples are phenetically dissimilar and open
circles are samples that are phenetically similar.
Table 6.7 Subset of Sample Pairs with their Distance and MMD Value SamDle Pair Distance tin km) MMD Value Sample Pair Distance tin km) MMD Value
MER-XGR 0.0 0.000 LAC-CAR 570.0 0.478
LEP10-LEP200 0.5 0.006 ETC-CAR 600.0 0.186
LEP10-CAR 70.0 0.070 LEP10-CAC 600.0 0.055
LEP200-CAR 70.0 0.040 LEP200-CAC 600.0 0.095
LAC-ETC 80.0 0.057 LEP10-LAC 630.0 0.340
PCC-SUL 100.0 0.085 LEP200-LAC 630.0 0.250
LAC-SUL 150.0 0.180 LEP10-ETC 670.0 0.134
LAC-CAC 180.0 0.235 LEP200-ETC 670.0 0.129
CAC-SUL 190.0 0.144 SUL-CAR 700.0 0.314
ETC-PCC 190.0 0.006 PCC-CAR 720.0 0.437
ETC-SUL 190.0 0.010 LEP10-SUL 750.0 0.146
LAC-PCC 200.0 0.047 LEP200-SUL 750.0 0.275
ETC-CAC 270.0 0.034 GED-HES 820.0 0.174
PCC-CAC 290.0 0.194 LEP10-PCC 820.0 0.279
HES-MER 380.0 0.085 LEP200-PCC 820.0 0.222
HES-XGR 380.0 0.064 GED-MER 1180.0 0.172
CAC-CAR 510.0 0.099 GED-XGR 1180.0 0.141
Bold indicate significant MMD values (MMD>2*SD); Leptiminus=LEP; Latini=LAC;
Etruscan=ETC; Piceni=PCC; Campani=CAC; Sulmone=SUL; Carthage=CAR; El Hesa=HES;
Meriotic=MER; X-Group=XGR; En Gedi=GED
Correlation analysis may not be the best means of testing the relationship. Distance
can continue to increase but genetic distance will have an upper limit. Thus, the
relationship may show a strong initial increase but as distance increases the line will
flatten out at the limit of genetic difference. When data from many genes are
averaged, regularity increases and observed curves tend to a common shape that
corresponds reasonably well with that expected at equilibrium of drift: an initial
strong increase followed by a progressive flattening out of the curve (Cavalli-Sforza
et al, 1994:122). To understand how closely the data follow with a curve of this
shape Figures 6.10 and 6.11 are examined more closely.
To help determine the limit of phenetic similarity, as measured by the dental
morphological data, the MMD value was used as the genetic variable. Those values
that indicate phenetic similarity are shown as open circles on Figure 6.11. They
generally occur at or below 0.08 on the graph. If distance is a reasonably good
predictor then the graphs should show no phenetically similar groups at a great
distance and conversely those that are the most phenetically dissimilar should be the
farthest in distance. In Figure 6.10 there are a cluster of points on the right side of
the graph and three of those points fall below the 0.08 line. Two of those are MMD
values that indicate phenetic similarity (Latini/En Gedi and Carthage/El Hesa). The
third point is the Etruscan and El Hesa samples which had a MMD value of 0.064 but
was determined to be phenetically dissimilar because the MMD value was greater
than two times its standard deviation. Standard deviation varies considerably
between the different sample pairs, largely due to trait frequencies being based on a
minimum number of observations. It may be beneficial to find a transformation
calculation that could be preformed on the MMD value to help standardize the values
given the variability of the standard deviations. This may help to clarify the overall
relationship between this measure of genetic similarity and distance. In examining
Figure 6.11 and considering the expectations of the curve of genetic similarity to
distance, there are a few points that do not meet expectations. At just over 500 Km
in distance there is a great range of points. There is at least one that indicates
similarity which is the Campani and Leptiminus 10 samples, and there are two points
that show the greatest amount of dissimilarity which are the Latinin/Carthage samples
and the Piceni/Carthage samples. If the expectation is that the curve will initially
increase and then begin to flatten out, then 1) at a certain point in distance there
should be no phenetically similar samples; and 2) the most dissimilar should be at the
greatest distance. In general, the data do show some concordance with expectation
but there are notable exceptions.
By this period in the history of the Mediterranean, populations have been interacting
due to changes in agricultural practices, technological advances (such as nautical
expertise), and military conquests. Distance alone is not as useful for predicting
biological affinity as it is with less mobile groups (Cavalli-Sforza et al., 1994). The
data indicate that affinity is correlated with geographic distance to a certain point but
it is likely that other factors are also involved. Future research may be able to
identify some of the factors which affect affinity such as social status, economics,
class, religion, or other social reasons. This study shows that military expansion can
bring populations that are genetically distinct into contact irrespective of the
geographic distance that once separated them; for example the Latini and En Gedi
samples and the Campani and Leptiminus samples. The hypothesis of affinity being
predicted by geographic distance alone is not fully supported by this data.
CONCLUSIONS AND SUMMARY
Two main objectives of this study were to 1) examine, record, and quantify the dental
morphological information from two cemeteries from the ancient town of Leptiminus,
and 2) assess phenetic similarities with the aid of inferential statistical analysis by
comparing the suite of dental traits from Leptiminus with other local and regional
populations. The results of the statistical analysis is used to support or refute three
Hypothesis #1: Biological Distinctiveness of the Carthaginians
The Leptiminus and Carthage samples are phenetically similar supporting the
hypothesis of a distinct biological Carthaginian population. Future research can help
to qualify the biological distinctiveness of the Carthaginians and help to define the
range of variation in dental morphology for this population. If, or when, biological
samples from ancient Phoenician and North African Berber populations are available
for study, the origins of the Carthaginians may be better understood. With
Phoenician and Berber samples the question of whether the Carthaginians are a
Phoenician/North African (Berber) hybrid may be answered.
Hypothesis #2: Biological Influence bv Roman Expansion
Two of the five Italian samples are phenetically similar to non-Italian samples
supporting the hypothesis that there is biological influence in regional Mediterranean
populations by Roman expansion. Phenetic similarity was shown to exist between
one of the Leptiminus samples and a sample from south of Naples, Italy (Campani).
This also supports the assertion that Italian settlers were installed at Leptiminus as
the historical references state. There is phenetic similarity between the Latini sample
from Italy and the sample from the Levant. There is a biological presence in these
two areas by an Italian population. However, gene flow is not necessarily the reason
for the affinity. The settlers may be buried in the same cemeteries. Other biological
data may be helpful in determining if gene flow occurred particularly DNA analysis.
Excavation of other cemeteries at Leptiminus could help qualify the biological
relationship between the Leptiminus Carthaginians and Campani settlers.
Archaeological investigations at Leptiminus could also help by determining if ethnic
areas exist which may suggest some social separation between the two groups.
Future research could also help to verify that Italians migrated to En Gedi. Other
cemeteries in En Gedi may be able to further qualify the relationship between Latini
and En Gedi. As additional samples from around the Mediterranean become
available, this hypothesis can be re-tested to gain a better understanding of the
biological influence of Roman expansion throughout the region.
Hypothesis #3: Isolation bv Distance
The hypothesis that phenetic similarity between groups is closely correlated with
geographic proximity is not fully supported. When all samples are tested there is
essentially no correlation between phenetic similarity and distance. When a subset is
tested, using only samples that are within 1500 km to each other, there is a moderate
positive correlation. Affinity is correlated with geographic distance to a point but
other factors are also involved. Future research may be able to identify some of these
factors which can affect biological affinity such as social status, economics, class,
religion, or other social reasons.
This research is important in helping us to better understand the Carthaginians by
providing new information on these people who were previously biologically obscure.
The use of dental morphological data for population affinity studies informs us of
origins, relationships, migrations, and admixture of the Carthaginians with other
Mediterranean populations. This and similar studies can advance our knowledge of
the variability of human populations by tracing historic relationships.
Raw Data for Leptiminus Samples
Site 10 Site 200
Trait Grade Count Percent Grade Count Percent
Winging UI1 1 0 0.0 1 1 14.3
3 13 100.0 3 6 85.7
Labial Curve UI1 0 10 55.6 0 10 47.6
1 8 44.4 1 10 47.6
2 0 0.0 2 1 4.8
Shoveling UI1 0 9 52.9 0 15 75.0
1 5 29.4 1 3 15.0
2 2 11.8 2 2 10.0
3 1 5.9 3 0 0.0
Shoveling UI2 0 8 50.0 0 7 46.7
1 4 25.0 1 6 40.0
2 3 18.8 2 1 6.7
3 0 0.0 3 1 6.7
4 1 6.3 4 0 0.0
Shoveling UC 0 14 93.3 0 15 93.8
1 1 6.7 1 1 6.3
Dbl Shoveling UI1 0 18 100.0 0 20 100.0
Dbl Shoveling UI2 0 18 100.0 0 16 100.0
Dbl Shoveling UC 0 15 100.0 0 17 100.0
Maxillary Traits (contcTl
Site 10 Site 200
Trait Grade Count Percent Grade Count Percent
Dbl Shoveling UP1 0 14 100.0 0 16 100.0
Interuption Groove UI1 0 16 100.0 0 23 100.0
Interuption Groove UI2 0 15 88.2 0 11 84.6
Medial 2 11.8 Medial 2 15.4
Tubercle Dentale UI1 0 5 29.4 0 12 54.5
1 8 47.1 1 6 27.3
2 3 17.6 2 2 9.1
3 0 0.0 3 1 4.5
4 0 0.0 4 0 0.0
5 1 5.9 5 1 4.5
Tubercle Dentale UI2 0 8 50.0 0 10 71.4
1 2 12.5 1 1 7.1
2 1 6.3 2 0 0.0
3 1 6.3 3 1 7.1
4 3 18.8 4 0 0.0
5 1 6.3 5 2 14.3
Tubercle Dentale UC 0 10 71.4 0 15 71.4
1 1 7.1 1 2 9.5
2 0 0.0 2 1 4.8
3 2 14.3 3 1 4.8
4 0 0.0 4 1 4.8
5 1 7.1 5 1 4.8
Canine Mesial Ridge 0 12 92.3 0 13 100.0
3 1 7.7 3 0 0.0
Canine Distal Ridge 0 7 63.6 0 8 80.0
1 2 18.2 1 1 10.0
3 2 18.2 3 1 10.0
3-Cusp UP1 0 14 100.0 0 17 100.0
Maxillary Traits (confer)
Site 10 Site 200
Trait Grade Count Percent Grade Count Percent
3-Cusp UP2 0 14 100.0 0 16 100.0
Hypocone UM1 3 0 0.0 3 1 6.7
4 9 40.9 4 3 20.0
5 13 59.1 5 11 73.3
Hypocon UM2 0 1 7.1 0 6 37.5
1 1 7.1 1 1 6.3
2 0 0.0 2 1 6.3
3 3 21.4 3 2 12.5
3.5 2 14.3 3.5 2 12.5
4 6 42.9 4 1 6.3
5 1 7.1 5 3 18.8
Hypocone UM3 0 3 37.5 0 6 54.5
3 1 12.5 3 0 0.0
3.5 1 12.5 3.5 2 18.2
4 3 37.5 4 2 18.2
5 0 0.0 5 1 9.1
Cusp 5 UM1 0 10 35.7 0 12 80.0
1 5 17.9 1 2 13.3
2 1 3.6 2 1 6.7
4 1 3.6 4 0 0.0
5 1 3.6 5 0 0.0
Cusp 5 UM2 0 9 69.2 0 14 93.3
1 3 23.1 1 1 6.7
2 1 7.7 2 0 0.0
Cusp 5 UM3 0 6 85.7 0 8 88.9
1 0 0.0 1 1 11.1
3 1 14.3 3 0 0.0
Carabelli's UM1 0 17 77.3 0 6 40.0