Professional Documents
Culture Documents
Roksandic y Armstrong (2011)
Roksandic y Armstrong (2011)
Perspectives
ABSTRACT Paleodemography, the study of demo- In addition to a standard five-stage developmental model
graphic parameters of past human populations, relies on based on the human life history pattern, current among
assumptions including biological uniformitarianism, sta- human biologists, we suggest divisions within the adult
tionary populations, and the ability to determine point age stage that recognize the specific nature of skeletal samples.
estimates from skeletal material. These assumptions have We therefore propose the following eight stages recogniz-
been widely criticized in the literature and various solu- able in human skeletal development and senescence:
tions have been proposed. The majority of these solutions infancy, early childhood, late childhood, adolescence, young
rely on statistical modeling, and have not seen widespread adulthood, full adulthood, mature adulthood, and senile
application. Most bioarchaeologists recognize that our abil- adulthood. Striving toward a better prediction of chronolog-
ity to assess chronological age is inherently limited, and ical ages will remain important and could eventually help
have instead resorted to large, qualitative, age categories. us understand to what extent past societies differed in the
However, there has been little attempt in the literature to timing of these life stages. Furthermore, paleodemogra-
systematize and define the stages of development and age- phers should try to develop methods that rely on the type
ing used in bioarchaeology. We propose that stages should of age information accessible from the skeletal material,
be based in the human life history pattern, and their skele- which uses life stages, rather than point age estimates. Am
tal markers should have easily defined and clear endpoints. J Phys Anthropol 145:337–347, 2011. V 2011 Wiley-Liss, Inc.
C
As a study of demographic parameters of past human derived from modern human populations. This is prob-
populations, paleodemography relies on three major lematic not only because we are unable to determine
assumptions: biological uniformitarianism, stationary point age estimates, but because these categories have no
populations, and the ability to determine point age esti- underlying biological meaning insofar as these categories
mates from skeletal material. Despite numerous group individuals of different life stages together based
criticisms raised on each of these assumptions (Petersen, on chronological age, with little correlation to biological
1975; Bocquet-Appel and Masset, 1982, 1985, 1996; Sat- age. In this article we propose to work within the biologi-
tenspiel and Harpending, 1983; Horowitz et al., 1988), cal reality of the skeletal material and redefine age
and a number of unsatisfactory attempts to address these categories to suit this reality. This would likely mean
criticisms (Van Gerven and Armelagos, 1982; Buikstra avoiding cohort statistics and the assumption of biological
and Konigsberg, 1985; Greene et al., 1986; Milner et al., uniformitarianism by using only biological age groups,
1989; Konigsberg and Frankenberg, 1992, 1994), there independent of chronological age. Furthermore, the size
seems to be a reluctance in the field of paleodemography and the structure of the population at risk are not
to let go of these assumptions and examine different known; it is commonly assumed that the excavated
approaches to overcoming these difficulties. As currently (analyzed) population is an unbiased sample of the
conceptualized, the construction of mortality profiles is
based on methods developed for modern human popula-
Grant sponsor: SSHRC; Grant number: 861-2009-0071; Grant
tions, founded on cohort statistics. A cohort is defined as sponsor: NSERC-USRA.
a group that shares a common experiential trait, such as
birth year. Relationships such as this can rarely be deter- *Correspondence to: Mirjana Roksandic, Department of Anthro-
mined in an archaeological context. This has been recog- pology, University of Winnipeg, 515 Portage Avenue, Winnipeg,
nized as problematic since the very onset of paleodemo- Manitoba, Canada, R3B 2E9. E-mail: m.roksandic@uwinnipeg.ca
graphic research by Angel (1969), who recognized that it
cannot be assumed that the individuals have cohort rela- Received 29 June 2010; accepted 10 January 2011
tionships, nor can it be assumed that the sample popula-
tion is unbiased. Furthermore, paleodemography com- DOI 10.1002/ajpa.21508
monly uses 5- or 10-year age categories (Lancaster, 1990) Published online 5 April 2011 in Wiley Online Library
which stem from the application of model life tables (wileyonlinelibrary.com).
C 2011
V WILEY-LISS, INC.
338 M. ROKSANDIC AND S.D. ARMSTRONG
population at risk. This assumption is not always war- use,’’ (Angel, 1969: p 428). Despite Angel’s warnings, life
ranted, and even though a fairly robust method for the tables—a tool used to calculate life expectancy and the
examination of bias in buried populations has been pro- probability of survival at a particular age—went on to be
posed (Jackes, 1993; Jackes and Meiklejohn, 2008), it is widely used throughout the field starting with Acsádi
rarely examined or reported. and Nemeskéri (1970) and Angel himself (1971). Acsádi
Bioarchaeologists are increasingly recognizing the and Nemeskéri (1970) provided the first standard text
problems within the field and—as we show below—they on determining demographic parameters from ancient
are avoiding both point age estimates and 5-year catego- populations by using life tables, followed by a number of
ries. Increasingly, they are using biologically meaningful case studies (e.g., Blakely, 1971; McKinley, 1971; Ben-
categories that are not dependent on precise determina- nett, 1973; Owsley and Bass, 1979; Piontek and Henne-
tion of chronological age. In this article, we draw atten- berg, 1981, etc.). These case studies used abridged life
tion to the changes in the field of bioarchaeology that tables that grouped populations into 5-year age intervals
are the result of the earlier criticisms of paleodemogra- as advocated by Ubelaker (1978), who deemed these
phy, and recognize the increasing use of age categories intervals large enough to eliminate most of the probable
rather than point age estimates and 5- and 10-year age errors in age determination.
groups derived from point age estimates. These age cate- Following the initial enthusiasm for paleodemography,
gories are, however, never properly defined in relation to the field underwent its first substantial criticism by Boc-
a sound underlying principle or a theoretical concept: quet-Appel and Masset (1982). In their ‘‘Farewell to Pale-
their meaning is assumed rather than discussed in any odemography’’ the authors argued that age determination
of the papers. We propose that life history theory pro- was inaccurate and that age-at-death profiles tended to
vides the best underlying model for constructing age cat- mirror the reference samples from whence they were
egories. In that context, we examine the biology of derived (Bocquet-Appel and Masset, 1982). This latter
human growth and development using a life-history per- criticism was addressed by many in favor of using more
spective. The age categories derived from life history are appropriate reference samples (Van Gerven and Armela-
matched to stages of skeletal and dental development in gos, 1982; Buikstra and Konigsberg, 1985; Greene et al.,
non-adults. These developmental markers are examined 1986), though more recently it has been recognized that
against primate models, to stress their universality and reference population mimicking remains a valid concern
applicability across populations. We further discuss skel- (Buikstra, 1997).
etally recognizable and meaningful stages within adult- The problem of inaccuracy in age determination is one
hood that are critically important for paleodemographic that continues to plague paleodemography. This issue
and paleopathological studies. While it is evident that has been thoroughly discussed in the anthropological lit-
these groups will not in and of themselves solve the erature (Murray and Murray, 1991; Kvaal et al., 1994;
problems inherent in paleodemographic research, we Ley et al., 1994; Aykroyd et al., 1999; Nagar and Hersh-
hope that the article will prompt paleodemographers to kovitz, 2004). There is a definite tendency of archaeologi-
build their models on thus redefined, readily available cal populations to show a mortality peak between 30 and
information that, limited as it is, could potentially reveal 45 years of age and an absence of older individuals—a
important aspects of past human groups based on sub- pattern not seen in historical hunter-gatherer and sub-
stantiated biological processes. In itself—even if mortal- sistence populations. Rather than reflect a reality of
ity/survivorship profiles could not be reconstructed due archaeological populations, this pattern shows the bias
to cemetery biases and problems inherent to paleode- in skeletal age determination methods, and reflects ref-
mography (Petersen, 1975)—analyzing the population in erence populations that rarely contain individuals older
terms of relative representation of different stages could than 60 years of age (see Fig. 1 of Aykroyd et al., 1999;
advance our understanding of past human experiences Chamberlain, 2006).
of life. Furthermore, while we deliberately avoided set- In addition to the practical issues of age determination,
ting chronological boundaries to these life stages, they there are further problems with the assumption of biolog-
carry intrinsic age information that could be used to ical uniformitarianism. This assumption implies that
build mortality profiles, provided development of proper age-related changes in humans are the same throughout
paleodemographic techniques. It is also important to time and space. It has been noted that this is only par-
stress that finding better methods of point age estima- tially correct—though the pattern of age-related changes
tion is still a very important goal. Better methods in may be the same, the rate at which these changes occur
age-at-death estimation would allow us to examine at varies between males and females, and across popula-
which chronological ages past populations experienced tions (Hoppa, 2002). Cultural practices can also affect
the transition from one life-history stage to another. individual ageing patterns (Gowland, 2006).
The most problematic assumption of paleodemography,
rarely addressed by paleodemographers, is the existence
Issues and criticisms of cohort relationships in a sample population. Cohorts
are studied for the very reason that each cohort is differ-
The problems associated with using life tables in pale- ent and that ‘‘. . . successive cohorts do something other
odemography were recognized at the very onset of paleo- than merely repeat the patterns of behavior of their pred-
demography as a discipline. In his theoretical article, ecessors’’ (Weeks, 2005: p 321). It is also noted that
Angel (1969) outlined three unsubstantiated assump- though successive cohorts are expected to be similar in
tions that are made in constructing life tables from some ways, they can differ in others (namely size, compo-
cemeteries: ‘‘. . . that the cemetery represents a single sition, and age-specific patterns of survival or longevity),
generation cohort, that death rates are even at all ages which is precisely why they are studied (Weeks, 2005).
after infancy and hence directly reflected in cemetery Understanding these changes is the goal of paleodemog-
age frequencies, and that the population is virtually sta- raphy, but by their reliance on artificial cohorts and
ble biologically and socially over the period of cemetery assumption of ‘stationary populations’ paleodemographers
Current approaches
The response to the criticisms regarding inaccuracies in
age determination and in the underlying age distribution
in sample populations has focused on statistical methods
that could address these inaccuracies such as: hazard
models (Gage, 1988); maximum likelihood models (Paine,
1989); the iterated-age-length key method (Konigsberg
and Frankenberg, 1992); the use of Bayes’ theorem (Ayk-
royd et al., 1999); and weight functions (Müller et al.,
2002). Most recently, the ‘‘Rostock manifesto’’ proposed
that the Bayesian approach should be the method of
choice, but that the age distribution (f (a)) of the target
population must be established before osteologists are able
Fig. 1. Graph of age categories used in American Journal of to determine the probability of a person from that target
Physical Anthropology, in two periods: the pre-criticism period population dying at a particular age, which again depends
between 1978 and 1982 at the top and the recent 2005 and 2009 on statistical modeling to determine how the probability of
period at the bottom. It should be noted that the category death varies with age (Hoppa and Vaupel, 2002).
‘‘Other numerical’’ includes many studies using only non-adult
specimens, which generally used small age categories due to the
This somewhat circular logic and heavy reliance on
relative ease of determining age in subadults. The majority of statistics makes practical application difficult, as
articles in 2005–2009 employing 5-year age categories were self- reflected in the relative absence of case studies that
defined paleodemographic studies. have emerged from the manifesto. Furthermore, the
probabilities are calculated on the basis of the same poor
age indicators (for a recent attempt to apply this model
deny that any change occurs. Furthermore, although a see DeWitte, 2009, although the author uses additional
cohort can be any delineated population that experiences age information provided by documentary evidence).
a particular life event in a given period of time, paleode- Because of the practical and theoretical complexities of
mography uses a population as if it were a birth cohort, using the statistical methods suggested over the past
which is by definition one that is born in the same few years, a number of more recent case studies (Steyn,
defined period of time (usually a year or a decade), in 2003; Castro et al., 2004; Mosothwane and Steyn, 2004;
order to examine death by age (Glenn, 1977). Ultimately, Nagaoka and Hirata, 2007), although recognizing the
it is inappropriate to treat an archaeological population problems with the life table method, continue to use the
as a birth cohort when it is unknown when the individu- traditional paleodemographic approach.
als were born, or when the period of time in which the Consequently, a substantial rift separates theoretical
cohort as a whole was born is large, as is often the case statistical paleodemography and bioarchaeological
with archaeological populations. The result of this method attempts to understand past human populations. This is
is that it attempts to take into account only the effects of no less true with regards to age categories. As previously
age on life expectancy, while neglecting other important mentioned, 5-year intervals have been commonly used in
and inseparable effects, such as cohort effects (the effects paleodemographic reconstructions. Used for simplicity of
due to being part of a particular birth cohort) and period analysis for examining birth cohorts in modern demogra-
effects (effects due to the period of time in which that phy (Weinstein and Pillai, 2001), 5-year intervals have
individual or group lived) (Glenn, 1977). been used in paleodemography due to the difficulties in
Archaeological populations do not allow us to account establishing point age estimates from the individual skel-
for cohort and period effects, and as a result, we can etons. Outside of paleodemography, however, very few
only estimate a very crude age-specific death rate (i.e., bioarchaeologists use 5-year categories, as evidenced by a
number of individuals who died within a particular age survey of articles using age categories in the past 5 years
range over the total population) to measure mortality of the American Journal of Physical Anthropology (see
among archaeological populations, with the knowledge Fig. 1). This journal being the most relevant biological
that no cohort relationship exists between the individu- anthropology journal, the past 5 years should reflect cur-
als who perished at any particular age (Lancaster, 1990). rent trends in the use of age categories. It becomes clear
Crude death rates may be as precise as paleodemogra- that many researchers use qualitative categories (e.g., ju-
phy is capable of achieving, but even they might be a venile, adolescent, young adult, mature adult, etc.) with
lofty goal, given that the structure of the population at or without a defined year range, as opposed to paleode-
risk is hard to ascertain. While the dead are never an mography articles which tend to use strict 5- or 10-year
unbiased sample of a living population, as Jackes (2011) age categories. This may reflect the urge to use biologi-
points out, ‘‘an unbiased cemetery sample. . . closely cally meaningful age categories, as opposed to simply
outlining chronological age. When this is compared to the et al., 2004), have been heavily criticized (Renz and
five years of the journal before the major criticisms of Radlanski, 2006; Roksandic et al., 2009), we still do not
paleodemography emerged in 1982, it becomes evident have a method that will allow us to calibrate these stages
that views on the quality of methods of age determina- in relation to their chronological age expression in differ-
tion have changed—it is no longer assumed that precise ent populations. This, however, should be our ultimate
ages can be determined for adults (Table 1). goal. In the meantime, age categories currently used by
Despite some positive reviews of our ability to deter- bioarchaeologists need to be defined and linked to mean-
mine age in skeletal remains (Wright and Yoder, 2003), ingful biological processes.
determining point age estimates remains an imprecise
endeavor, subject to substantial scrutiny, particularly for Life history model
the age assessment of adult specimens. It has already
been noted that methods of age determination provide a There are many complicating factors to determining bi-
better idea of physiological or biological age rather than ological age. As Beall (1984) notes, most maturational
chronological age (Kemkes-Grottenthaler, 2002; Jackes changes finish by the end of the second decade of life and
et al., 2008). In fact, biological age is the only age that is subsequent degenerative changes are much more vari-
apparent on skeletal remains; it is usually the bioarch- able. As already noted, there has been no discussion on
aeologist who attempts to convert this into years based how to appropriately divide the human lifespan into life
on a correlation between biological and chronological age, stages, as evidenced by different descriptive categories
and the assumption of biological uniformitarianism. Yet used by researchers, not only when dealing with adults,
it is understood that, in addition to the difficulties of but also in non-adult skeletons.
establishing chronological age from skeletal remains and The model proposed here uses biologically meaningful
the issues of mimicking a reference population, chrono- stages, under the overarching framework provided by life
logical age is often an irrelevant measurement in other history theory. Life history theory is concerned with the
cultural settings (Gowland, 2006; Bocquet-Appel and trade-offs between growth, reproduction and maintenance
Bacro, 2008). The physical process of growth and senes- (McDade, 2003), and how these interacting factors affect
cence, however, is a human universal, though certainly the evolution of a particular life history pattern. This
the importance assigned to any particular maturational ontogenic process and consequent biological compromises
stage, or even the timing of these events, may be affected result in a distinct pattern of life stages where each stage
by cultural practices, genetics, and environment, among represents the sum of the coordination between behavior
others. It is well understood that the chronological timing and physiology (Del Guidice et al., 2009); in humans, this
of biological events can vary wildly between individuals process has led to the pattern characterized by helpless
and populations. Thus, for biological anthropologists it infants, slow maturation and long childhood dependency,
seems only logical to determine maturational or degener- menopause and long life span (Bogin, 1999). The use of a
ative stage, as opposed to chronological age. The increas- model of progressive life stages can circumvent many of
ing use of descriptive age categories by bioarchaeologists the difficulties that arise when determining the chrono-
indicates that this might be the next logical step for pale- logical age of individuals. It requires identifying clear de-
odemographic research as well. Ultimately, it is likely velopmental and degenerative endpoints on the skeleton,
more informative, even from a paleodemographic per- which serve as a proxy for those biological events that
spective, to discover mortality patterns by meaningful, cannot be observed on the skeleton, but through which all
biological life-stage as opposed to chronological age. It is humans must pass (Leidy, 1996). The chronological age at
nevertheless obvious that projecting a population by life which these endpoints are achieved may vary, between
stage will be further hampered by the fact that stages individuals and across populations, depending on a num-
will be of different lengths. Therefore, establishing better ber of factors. Though understanding the variation of age
methods of assigning chronological age remains an im- at these endpoints across populations remains an impor-
portant goal that will ultimately allow us to discuss the tant area of study, the chronological ages at which these
chronological ages at which different life history transi- endpoints occur is not the focus of the proposed approach.
tions were experienced in different populations. Since The model outlined here is an adaptation of the pat-
even methods such as the analysis of tooth cementum tern proposed by Bogin (1990, 1994, 1999, 2001, 2003).
annulations, that claimed high levels of accuracy in a Bogin’s five-stage model includes infancy, childhood,
known age context (Stott et al., 1982; Wittwer-Backofen juvenility or juvenescence, adolescence, and adulthood.
Since the term ‘‘juvenile’’ is used by both paleodemogra- lated to complete eruption of the deciduous dentition,
phers and bioarchaeologists (even in very important vol- occurring at about 2.5 years (Smith et al., 1994). After
umes such as Bocquet-Appel and Masset, 1977; Scheuer this time, substantially more non-maternal sources of
and Black, 2000) to indicate non-adults in general, we food are introduced into the diet, though the exact tim-
prefer to substitute the terms childhood and juvenility ing is, again, culturally dependent.
with early and late childhood, respectively. Each of these In primates, infancy is defined as the period from birth
stages is defined by distinct biological events. In addition to weaning which coincides with the eruption of the first
to Bogin’s model, which recognizes only one adult stage, permanent molar (Smith, 1991). Even in primates, wean-
we outline several stages within adulthood that are ing is a gradual process, and its timing may be affected by
based on observable skeletal changes associated with the seasonality and the availability of weaning foods (Lee,
final developments of the skeleton and the degenerative 1996; Strier, 2007). In humans, the end of infancy also
changes that follow the end of skeletal growth (Table 2). tends to coincide with weaning, but this occurs years
Given the nature of bioarchaeology and the skeletal before the first permanent tooth erupts (Bogin, 1999).
material it uses, we will focus mainly on the skeletal and Consequently, humans not only have a relatively short
dental developments, though it is understood that these infancy compared to non-human primates, but they also
developmental stages are also marked by cognitive and experience the unique stage of early childhood between
physiological changes. Dentition is of particular impor- weaning and the eruption of the first permanent molar.
tance in defining stages of growth, as it is less susceptible Early childhood, a time of rapid brain growth and
to environmental and nutritional changes, and is ‘‘... inte- extensive learning, is recognized as a stage primarily for
grated into the plan of growth and development, timed to its intellectual and social, rather than physical develop-
the growth of the skull, maturation of muscles of mastica- ment (Mace, 2000). In terms of skeletal biology this
tion, and somatic growth in general’’ (Smith, 1989: p stage corresponds to the time between the full set of de-
683). Once full dentition is reached, changes associated ciduous teeth and the eruption of the first permanent
with the post-cranial skeleton become more relevant. tooth, which tends to be the first molar, but can also be
one of the central incisors (Smith et al., 1994; Bogin,
Nonadults (juveniles) 2001; Del Guidice et al., 2009). Although all deciduous
teeth have emerged, the individual is still dependent on
The juvenile (the term is here used as is common in adults for care and feeding, and they have difficulty
bioarchaeological practice to denote a non-adult) period processing an adult diet (Bogin, 2003). At least theoreti-
can be divided into infancy, early childhood, late child- cally, the eruption of the full deciduous dentition indi-
hood, and adolescence. cates the ability of the child to process food, though
Infancy can be defined, among other biological and there is variation between different societies and individ-
cognitive developments, as the time between birth and uals as to the extent of solid food introduced in the diet
the eruption of a full set of deciduous teeth at 2–3 and the continued reliance on breastfeeding.
years of age (Bogin 2001). This also tends to coincide After weaning, nonhuman primates are relatively in-
with weaning, which is a highly culturally dependent dependent in comparison to human children, who are
event in humans, but can nonetheless be roughly corre- still helpless for a number of years during which the
child must continue to be fed by specially prepared food ductive maturity, the time which is occupied by adoles-
by care-giving adults (Del Guidice et al., 2009). The fact cence in humans, is not unique to humans as it is also
that this adult need not be the mother is important as seen in great apes. Unlike humans, in great apes this
this reduced dependency for breast milk allows for rela- gap is relatively short in terms of total lifespan (Bogin,
tively shorter birth intervals in humans (Bogin, 2003). 1999). Similarly, while sexual maturity is also the time of
Furthermore, it is possible to speculate that the food permanent canine eruption in humans, it is instead
processing behavior, particularly thermal processing marked by eruption of the third permanent molar in pri-
(cooking), seen in humans has aided the usefulness of mates such as Pan and Gorilla, an event which does not
this stage, given that children cannot process a fully occur in humans until reproductive maturity or later, if
adult diet and have a high food-energy requirement for it occurs at all (Dainton and Macho, 1999).
brain growth (Carmody et al., 2010). Childhood may pro- Though perhaps not well-recognized, the timing of hor-
vide humans with the necessary time for learning and monal changes indicating the onset of puberty in girls
understanding culture, in accompaniment with the rapid and boys coincides with the eruption of the permanent
brain growth characteristic of this stage (Bogin, 1990). canines, and consequently the canines serve as a better
Further proof for the existence of early childhood as a predictor for the onset of adolescence than the eruption
uniquely human life stage is the difference in growth of the second molars. Though he does not discuss the
profiles. Humans are distinct in that three mathematical association between the two events, Bogin (2001) does
functions are required to plot their growth, due to the separately discuss the timing of canine eruption (p 83)
mid-childhood growth spurt which is not seen in other and onset of puberty (p 87). Similarly, the eruption of
primates (Bogin, 1990). The mid-childhood growth spurt the canines is a less variable predictor of the onset of pu-
is characterized by another event not observable in the berty than menarche, an event which is under strong
skeleton, the onset of adrenal androgen production, or environmental pressures and whose variation in timing
adrenarche (Bogin, 1997). The effects of adrenarche are is well-documented (Moffitt et al., 1992; Worthman,
wide-ranging, from social and cognitive behavior, to so- 1999; Freedman et al., 2002; Chumlea et al., 2003). The
matic growth and development, to effects on the neuro- advantages to using the eruption of canines as a proxy
logical and immune systems (Campbell, 2006). for the onset of sexual maturity and adolescence is that
Following early childhood is late childhood, coinciding it is observed in both males and females, and can be
with the time between the eruption of the first perma- observed in a bioarchaeological context.
nent tooth and the eruption of the permanent canine. The onset of reproductive maturity indicates adulthood
Late childhood is the time of pre-puberty and relative in- and cessation of growth, as growth and reproduction are
dependence; it is also when the growth rate is particu- mutually exclusive in terms of life history theory
larly slow. In addition, there are a number of social (Harvey et al., 1985). The evolutionary purpose of ado-
developments—late childhood is a time of increased play lescence is far more speculative than the distinct evolu-
and increased sexual differentiation of activities (Del tionary advantages of childhood. Nonetheless, it has
Guidice et al., 2009). The eruption of the first permanent been proposed that since the timing and type of changes
tooth at the end of early childhood tends to coincide with that occur in girls and boys differ, the evolutionary rea-
the end of the mid-growth spurt that occurs in human sons behind them may differ as well. In girls, menarche
children, as well as with adrenarche (Bogin, 2001). Adre- is followed by a few years of sterility—indeed, the female
narche seems to be related to gonadarche (puberty), and pelvis is not affected by the adolescent growth spurt and
it has been proposed that it may ‘‘serve to activate amyg- according to Bogin (1994) does not reach its adult size
dala and hippocampus activity prior to the onset of sex until several years after adult stature is attained. Conse-
steroids, reducing fearfulness and anxiety and increasing quently, they appear to be fully grown adults before they
social interaction with unfamiliar individuals’’ (Camp- are in fact reproductively mature, which Bogin (1994)
bell, 2006: p 585). suggests leads adults to encourage these adolescent girls
Adolescence is a time of hormonal changes, secondary to participate in adult social interactions, and conse-
changes (e.g., pubic hair, breast buds, seminal emission, quently learn how to be reproductively successful. In
menarche, etc.), and rapid growth, namely the adolescent contrast, boys appear more child-like despite the hormo-
growth spurt (Bogin, 2001). This phase corresponds to nal changes which encourage them to participate in
the time between sexual and reproductive maturity, adult behavior. Consequently, due to their appearance,
marked by the eruption of the permanent canines and they are not perceived to be adults and do not constitute
skeletal maturity, respectively (Bogin, 2001). Skeletal ma- competition for adult males (Bogin, 1994). Ultimately, it
turity in this case refers to the end of longitudinal can be said that, ‘‘Girls best learn their adult social roles
growth, with complete epiphyseal fusion of the long while they are infertile but perceived by adults as
bones. To a lesser extent, the end of adolescence may also mature, whereas boys best learn their adult social roles
be marked by the eruption of the third molar, although while they are sexually mature but not yet perceived as
this has become a variable, and consequently, unreliable such by adults. Without the adolescent growth spurt this
event (Smith, 1989). Once adolescence has ended, adult- unique style of social and cultural learning could not
hood is achieved and the majority of growth-related occur’’ (Bogin, 1994: p 33).
changes are complete. It should be noted that reproductive maturity cannot
The characteristic dramatic increase in the rate of skel- be equated with fertility; while sexual maturity is fol-
etal growth in adolescent humans is not observed in lowed by a period of sterility, human females are fertile
nonhuman primates, although chimpanzees may demon- before they are reproductively mature (Bogin, 1994). Sim-
strate an increase in body mass in this period (Bogin and ilarly to the onset of menarche, fertility is under strong
Smith, 1996; Kappeler and Pereira, 2003). In females, environmental pressures. While in well-fed populations,
sexual maturity is marked by menarche, while reproduc- individuals can be fertile shortly after menarche; this has
tive maturity is marked by the ability to successfully not traditionally been the case, and generally females are
carry a child to term. The gap between sexual and repro- not fertile until a few years after the onset of menarche