<%BANNER%>

Sexual Dimorphism and the Transition to Agriculture: A Meta-Analysis


PAGE 1

SEXUAL DIMORPHISM AND THE TRANSITION TO AGRICULTURE: A META-ANALYSIS By ANNA ELIZABETH VICK A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Anna Elizabeth Vick

PAGE 3

iii ACKNOWLEDGMENTS I would like to sincerely thank Dr. Da vid Daegling and Dr. Ken Sassaman for serving on my committee. They both o ffered advice and guidance throughout this project; I am grateful for th eir mentorship. Furthermore, I owe a great deal of appreciation to Dr. Russell Bernard and Dr Larry Winner; they fielded numerous questions regarding meta-analysis with gr eat patience and thoroughness. I also would like to thank Chad Maxwell for twenty-four hou r technical support. Next, I would like to express my gratitude to The University Women’s Club. This generous organization financially assisted me during my time at UF. Most importantly, however, I would like to thank my family, Laura Greer Vick, Gilb ert Vick, and Marguerite Hardee Greer, for their infinite love, suppo rt and encouragement.

PAGE 4

iv TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi ABSTRACT......................................................................................................................v ii INTRODUCTION...............................................................................................................1 METHODS........................................................................................................................ 15 Meta-Analysis.............................................................................................................15 Choosing Data Sets.....................................................................................................15 Data Analysis: Meta-analysis....................................................................................18 Data Analysis: Non-statistical Analysis of Data Not Appropriate for MetaAnalysis..................................................................................................................22 RESULTS........................................................................................................................ ..24 The Meta-analysis.......................................................................................................24 Non-Statistical Analysis.............................................................................................28 DISCUSSION....................................................................................................................3 0 Appropriate Indices of Dimorphism...........................................................................32 The Transition to Agriculture.....................................................................................34 Limitations of Meta-Analysis.....................................................................................40 CONCLUSIONS................................................................................................................43 APPENDIX A META-ANALYSES OF FEMORAL MEASUREMENTS.......................................44 B META-ANALYSES OF HUMERAL MEASUREMENTS......................................48 C META-ANALYSES OF OTHER LONG BONE MEASUREMENTS.....................51

PAGE 5

v LIST OF REFERENCES...................................................................................................52 BIOGRAPHICAL SKETCH.............................................................................................58

PAGE 6

vi LIST OF TABLES Table page 1 Sample Composition for Meta-analysis...................................................................18 2 Sample Composition for Non-statistical Analysis...................................................23 3 Summary of Meta-Analyses.....................................................................................25 4 Non-Statistical Analysis of Cha nges in Sexual Dimorphism Ratios.......................29 5 Sexual Dimorphism Ratios in Transitional Groups.................................................39

PAGE 7

vii Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts SEXUAL DIMORPHISM AND THE TRANSITION TO AGRICULTURE: A META-ANALYSIS By Anna Elizabeth Vick August 2005 Chair: David Daegling Major Department: Anthropology The degree of sexual dimorphism found in human populations has declined throughout the history of anatomically modern humans. Researchers have specifically suggested that the transition to agriculture led to a decline in sexual dimorphism due to the reduction in gendered biomechanical loads and a nutritional decline. To test this theory, data were compiled from studies which compared populations spanning the transition to agriculture. Meta-analyses were then c onducted using the available postcranial measurements of preagricultura l and agricultural gr oups to observe any changes in sexual dimorphism. Of the five North American populations compared, only 10.5% showed a significant dec line in sexual dimorphism. A non-statistical analysis of the literature did not dem onstrate any noteworthy differe nce in the degree of sexual dimorphism in populations before and after the adoption of agri culture. In contrast to the theory that sexual dimorphism declines with the transition to agriculture, in most cases, no significant change occurs.

PAGE 8

1 INTRODUCTION In modern human groups, the average male is almost 1.1 times as tall as the average female, and, according to Krantz (1982: 86), they are “correspondingly more massive.” Compared to extant ape populat ions, the degree of sexual dimorphism in humans represents a moderate degree of di morphism. Gorillas and orangutans have an enormous degree of sexual dimorphism wher e females may only weigh half as much as their male counterparts. Gibbons and siamangs exhibit very little sexual dimorphism in body size; females are 94% of the size of male s. Humans and chimpanzees are similar in degree of dimorphism, with females at 78% and 81% of male body size, respectively (Campbell and Loy, 2000). Within modern human groups, it has b een found that the degree of sexual dimorphism varies by population. Eveleth (1975) measured individuals of African, Amerindian, and European origin and found th at in the measurement of adult stature, Africans were the least sexually dimorphi c population. Amerindians were the most sexually dimorphic population, with Eur opeans ranking in between Africans and Amerindians. However, it is important to re member that stature is just one measurement of dimorphism. In my own investigations (Vick, unpublishe d), I found that Amerindians were highly dimorphic in the measurement of maximum femur length (commonly used to calculate stature), but did not express greater dimorphi sm in other osteometric measurements such as femoral circumference or humeral length.

PAGE 9

2 In contrast to Eveleth, Gaulin and Boster (1985) have suggested that the degree of sexual dimorphism found in human populations is actually quite consistent and that evidence to the contrary may simply be the pr oduct of small sample size. Nevertheless, these researchers, along with numerous others have continued to conduct studies which attempt to explain sexual dimorphism based on cultural differences in human populations. Marriage patterns, work load, pa rental investment, di vision of labor and subsistence type are just a few of the cultura l variables tested for their association with sexual dimorphism (Gaulin and Boster, 1992; Holden and Mace, 1999; Ruff, 1987; Wolfe and Gray, 1982a; Wolfe and Gray, 1982b) Sexual dimorphism in humans has varied over time. Over the course of human evolution, from the australopithecines to anatomically modern humans, the degree of sexual dimorphism has declined (Frayer a nd Wolpoff, 1985; Krantz, 1982). Likewise, within Homo sapiens sapiens there has been a subsequent decline in sexual dimorphism from the Upper Paleolithic to the present (Borgognini Tarli and Repetto, 1997; Brace, 1973; Brace and Ryan, 1980; Frayer, 1980; Fr ayer, 1981; Frayer and Wolpoff, 1985; Meiklejohn et al., 1984). The decreases in dimorphism seen in Homo sapiens sapiens have generally been associated with changes in subsistence or technology, most notably from the Upper Paleolithic to the Mesolithic. During th e Upper Paleolithic, humans were big game hunters. Then, as the Upper Paleolithic ende d with glacial retreat and the large scale extinction of many big game species, the Mesolithic hunter had to adapt by hunting smaller species, like pigs and deer, rather than, for example, mammoths. The prevailing theory to explain this transition is that as the game became smaller, so too did the hunter

PAGE 10

3 (Brace and Ryan, 1980; Frayer, 1980; Frayer, 1981). Brace and Ryan (1980) argue that an increase in male body size occurred during th e Pleistocene as an adaptive strategy for hunting large game prior to technological sophistication. But, with technological advances, the extinction of la rge game species, and the metabolic demands of a large body, the selective pressure for large male body si ze decreased. It is believed that female activity patterns did no t change as greatly as male activity patterns during the Mesolithic transition. As a result, mean female size did not change drastically, resulting in an overall decline in sexual size dimorphism (Frayer, 1981). Brace and Ryan (1980) take this theory to the next le vel by suggesting that the degr ee of sexual dimorphism in modern populations is directly related to th e amount of time that has passed since that population’s dependence on hunting large game. This theory is based on Brace’s (1963) theory of probable mutation effect which stat es that in the absen ce of selection, random mutations will lead to the reduction of associated features. For the effects of the probable mutation theory to be observable in mode rn populations, Brace and Ryan make several assumptions. First, selection among big game hunters would need to have been constant across populations. In addition, populations would have to remain reproductively isolated. Otherwise, the differences in th e progressive reduction in dimorphism due to the probable mutation effect w ould not be discernible. Researchers have also hypothesized that a further reduction in sexual size dimorphism occurred as humans shifted from a hunting and gathering economy to agriculture (Armelagos and Van Gerv en, 1980; Boyd and Boyd, 1989; Frayer, 1980; Frayer and Wolpoff, 1985; Hinton and Carlson, 1979; Holden and Mace, 1999; Kennedy et al., 1987; Lazenby, 2002; Ruff, 1987; Wolfe and Gray, 1982b). In contrast to data

PAGE 11

4 available for the Upper Paleolithic – Meso lithic transition, which despite the small sample size, strongly support a decrease in sexual dimorphism, data for a possible decrease in sexual dimorphism after the sh ift to agriculture are more equivocal. Archaeological, morphometric, and patholog ical data have been collected which describes the transition to ag riculture. While sp ecific biological responses to the adoption of agriculture vary re gionally, there appears to have been an overall decline in human health (Cohen and Armelagos, 1984). As human groups became more sedentary, populations increased, as did rate s of infectious diseases. People who raised livestock were even more prone to diseases due to exposure to animal vectors (Ortner, 2003). Chronic malnutrition also increased with agri culture. While groups were better able to store excess food for times of shortage, the nut ritional quality of th e food declined. In contrast, hunter-gathering populations had th e advantage of mobility ; if an area was no longer productive due to drought or other ci rcumstances, they could easily move. Sedentism resulted in reduced evidence of musc uloskeletal stress such as the incidence of degenerative joint disease and overall declin e in robusticity. However, a reduction in physical stress may not be an indicator of pos itive change for several reasons. Cohen and Armelagos (1984) point out that the higher frequency of dege nerative joint disease in hunter-gatherer populations may be complicated by the higher age at death among these groups as contrasted with more sedentar y populations. Therefore, reduction in the frequency of degenerative joint disease ma y not be an indication of a decreased workload. In addition, physical activity is associated with bone remodeling. Evidence suggests that a low level of physical activity during a pers on’s life may be associated with a higher propensity for bone fractures in later life (Ruff, 1991). Overall, the

PAGE 12

5 transition to agriculture was marked by a decl ine in both the quality and duration of life (Cohen and Armelagos, 1984), characteristics which need to be considered when evaluating why changes in sexual dimorphi sm would occur during this time period. To understand how a change in the subs istence economy could affect the degree of sexual dimorphism, it is first necessary to understand the causative factors affecting sexual dimorphism. There is general consen sus that genetics affect sexual dimorphism; however, the nature or strength of this relationship is not understood. Eveleth (1975) is frequently cited as a primary source s upporting the genetic contribution to sexual dimorphism. Eveleth’s comparison of the sexua l dimorphism of adult height in different “ethnic groups” concluded that there is a di fference in the level of sexual dimorphism between these groups. While the sources of Eveleth’s data are not given, the groups compared were broadly defined as Europeans, Negroes, Amerindians, Asiatics, and New Guineans. These “ethnic groups” are reminis cent of the major continental races and the comparison is based on the assumption that thes e groups are genetically distinct from one another. However, genetic variation does not observe socially constructed racial designations. The vast majority of genetic variation occurs within racial groups rather than between them. Eveleth’s (1975) study found that the de gree sexual dimorphism found in each ethnic or racial group di d not meet the expectations of th e nutritional hypothe sis that the most dimorphic populations should have the most nutritious diet. In consequence, Eveleth concluded that there must be a gene tic factor controlling the level of sexual dimorphism in each population. To arrive at this conclusion, Eveleth only considered two causal agents affecting dimorphism. Ther e is no consideration of climate, gendered

PAGE 13

6 access to resources, or any ot her factors associated with human morphology or sexual dimorphism. While there are many criticism s of Eveleth’s studies, other studies comparing the amount of dimorphism in diffe rent populations have found similar results (Holden and Mace, 1999). We are all aware that the size of an individual is highly related to the size of hi s or her parents, but th is does little to expl ain the ultimate cause of sexual dimorphism. In the animal kingdom there is a relati onship between overall body size and the degree of sexual dimorphism found in a species (Frayer and Wolpoff, 1985). According to Rensch’s Rule (Rensch, 1959), sexual dimo rphism increases with body size in taxa where males are the larger se x (Gustafson and Lindenfors, 2004). While Rensch’s rule is widely accepted, studies suggest that this relatio nship is less apparent in primates than in other taxa (Frayer and Wolpoff, 1985). Wh en size variation is compared by a method which includes phylogenetic data, the resu lts are even less clear (Gustafson and Lindenfors, 2004). Gustafson and Lindenf ors (2004) conducted a study of sexual dimorphism in human populations where data on mean male and female height were compared to genetic phylogenies. The results indicate that both male and female stature is associated with phylogeny and that there is no evidence for an allometric relationship between male and female stature in hu man populations (Gustafson and Lindenfors, 2004). This study serves as a dditional support that the geneti c component to dimorphism needs to be considered in cross-cultural studi es and that sexual dimo rphism is not simply a byproduct of overall size. The theory of sexual selection, more specifically, intrasexual selection, is commonly used to explain how sexual dimorphism develops in non-human animals

PAGE 14

7 including primates. Intrasexua l selection is based on the idea that competition exists within the members of one sex for reproductive access to the other sex. The sex which is competed for is generally the one that has a higher energy investment in the success of the offspring. In the case of humans, as we ll as in most other mammals, females make the greater investment in young; therefore, there is male-male competition for access to females (Trivers, 1972). While the relationship of size to reproductive success is difficult to measure, larger size may be linked to dominance by offering an advantage in aggressive encounters, and some studies demonstrate a positive relationship between dominance and reproductive success. As a re sult, larger males are able to pass their genes to subsequent generations more effectiv ely, thereby increasing the degree of sexual dimorphism in size (Trivers 1972). Based on theories of sexual selection, sexual dimorphism should be greatest in populations where there is the most competition for access to females. Because paternity is often difficult to dete rmine, mating or marriage practices have been used to test the sexual selection hypot hesis. Alexander et al. (1979) conducted a study to determine whether or not breedi ng systems were correlated with sexual dimorphism in a variety of species, incl uding humans. In all nonhuman groups it was found that the degree of polygyny (measured as a deviation from monogamy by harem size) was positively correlated with sexual dimorphism in body size. Primate species, such as hamadryas baboons, which are characterized by single-male, multi-female groups, have a high degree of sexual dimo rphism in body size. Conversely, the Callitrichidae characterized by monogamy and polya ndry, have a very small degree of sexual dimorphism, with females larger than males in some cases (Rowe, 1996).

PAGE 15

8 Alexander et al. (1979) also found a co rrelation between breeding system and sexual dimorphism in human populations, but their methods and results have been challenged by Gray and Wolfe (1980) who found no such correl ation in their own research. In the studies of Alexander et al. (1979) and Gray and Wolfe (1980), all monogamous societies were divided into tw o groups: those with socially imposed monogamy, and those with ecologically impos ed monogamy. Alexander et al. (1979) argue that socially imposed monogamy should be included with polygyny and that only ecologically imposed monogamy can be expected to follow the hypothesized pattern of sexual dimorphism based on theories of se xual selection. Unfortunately the methodology utilized by Alexander et al. (1979) of estima ting heights by sight, leaves their results subject to speculation. When Gray and Wolf e (1980) reanalyzed the data, they found no significant correlation between sexual dimorphi sm and mating pattern, but they did find less variability in male and female height in polygynous societie s, perhaps suggesting they are under greater selective pressure fo r height or some correlate of height. According to the theory of sexual se lection, there is gr eater variance in reproductive success for individual males in polygynous societies as opposed to monogamous ones. In any study of mating or ma rriage it is understood that true rates of paternity are not always easy to determine. However, marriage systems provide a readily available variable to study how sexual selection affects se xual dimorphism in humans under the assumption that a husba nd is likely to be the fath er of a woman’s offspring. Sexual dimorphism can be influenced by circumstances of the environment. A number of studies have shown that sexual dimorphism in stature can decrease when people are under nutritional stress and incr ease under conditions of optimal nutrition

PAGE 16

9 (Brauer, 1982; Gray and Wolfe, 1980; Li eberman, 1982; Stini, 1969; Stini, 1982; Wolanski and Kasprzak, 1976). The theoretical basis for this is found in the fact that males and females experience differential success in dealing with stressors like starvation and disease due to hormonal and metabolic di fferences (Ortner, 2003; Stini, 1969). The greater fat and nutrient reserv es characteristic of human fe males are thought to be an adaptation for the increased metabolic demands of lactation and ge station in producing offspring. As a result of these physiologi cal differences, males experience a greater reduction of lean body mass during periods of nutritional inadequacy than their female counterparts. When periods of starvati on occur during growth, the reduction in body mass is accompanied by reduced skeletal growth (Stini, 1975). In Stini’s (1975) analysis of the transiti on to hunting and gathering, he states that “severe nutritional imbalances…are much more common in agricultural areas than among hunters and gatherers alt hough starvation is no stranger to hunting populations in most parts of the world” (p. 64). The greater prevalence of these nutritional shortages in agricultural societies forms the basis for the th eory that sexual dimor phism declined with the transition to agriculture. If nutritional de ficiencies lead to a decline in dimorphism, this shift should primarily be due to a re duction in male size. However, while it is apparent that nutrition affect s sexual dimorphism to some degree (Brauer, 1982; Hall, 1982), many researchers think that it is not the leadi ng factor affecting sexual dimorphism on an evolutionary scale (E veleth, 1975; Larse n, 1984; Stini, 1969). Related to the issue of sexual selection ar e theories of parental investment. The theory of sexual selection sugge sts that the pressure for se xual selection is decreased when both male and female invest in the re aring of their offspring. In most primate

PAGE 17

10 species, the female invests much more energy in raising the next generation than does the male. In contrast, tamarins and marmosets ar e most noted for male parental investment. It is in these species that the least degree of sexual dimorphism occurs (Rowe, 1996). Studies suggest that parental investment favors the sex where the most variance in reproductive success occurs (Morbeck, 1997). Because the variance in reproductive success is generally greater for males, inve sting in young males can greatly increase a parent’s fitness. Cultural studies of human groups also show that parental investment varies based on the gender of the child. Holden and M ace (1999) found that sexual dimorphism in stature is negatively correlated with the am ount of work women perform. Likewise, female juvenile mortality rates are higher than those for juvenile males in areas where females contribute less to subsistence. Th ese patterns follow geographical patterns of sexual dimorphism (Holden and Mace, 1999). Rivers (1982) studi ed survival rates under conditions of famine and disaster to determine whether proof could be found for theories of differential survival based on se x. It was found that sex discrimination and preferential treatment of male children confounded the re sults of the study. While females may have a natural advantage under ti mes of stress, males often receive cultural advantages that may more than make up for any differential survival rates. Similar conditions may affect the degree of se xual dimorphism found in a population. Biologists may also view sexual dimor phism as a product of optimal biomass distribution for the species. When conditions select for large males, it is advantageous for the female of the species to be as much smaller as possible while still being able to achieve reproductive success (Bramblett, 1994) The optimum female size is large

PAGE 18

11 enough to bear the physical demands of labor, but small enough to reduce the metabolic demands associated with large size. By c onsidering sexual dimorphism as an optimal biomass distribution, DeVore and Washburn (19 63) are proposing that males and females may be better able to utilize their resources if they fill different ecological niches. If niche divergence is amplified with an increa se in sexual dimorphism, then the selective pressures affecting males and female s are progressively more different. Ethnographic research (Murdock and Provost, 1973) suggests that there is greater overlap between male and female subsistence acti vities in agricultural populations than is found in hunting and gathering populations. In addition, biomechanical data suggest that the forces affecting males and females are more similar in an agricu ltural population than among hunter-gatherers. For example, Ruff (1987) found that when comparing crosssectional properties of bone, th ere was a decline in sexual d imorphism with the transition to agriculture indicating a reduc tion in the division of labor associated with agricultural tasks. Ruff (1987) did not code for nutriti onal changes in this analysis because he believes that cross-sectional data is a reflect ion of mechanical environment as opposed to dimensional variables which may be better in dicators of nutrition. Holden and Mace (1999) compared populations in the Ethnographic Atlas with regard to marriage practices, subsistence and the division of labor. They “c oncluded that in contemporary humans, neither hunting nor agriculture has any effect on sexual dimorphism. [Instead] It is the amount of subsistence work done by men and women, rather than the type of subsistence practiced, which has an effect on sexual dimorphism in different societies.” (p. 42). As women contribute more to th e subsistence economy, it appears that the degree of sexual dimorphism is reduced. Ho lden and Mace (1999) used stature as their

PAGE 19

12 only measurement of dimorphism while Ru ff (1987) used only the cross-sectional properties of bone. Ruff compared individuals from what is believed to be a genetically related population in which the transition to agriculture took place. Holden and Mace compiled data from various populations ar ound the globe, and, while hunter-gatherers and agriculturalists are both represented, there is no gene tic control which connects one group to the other. Sexual selection, parental investme nt, nutrition, environmental conditions, marriage or mating patterns, body size, ecological niche, and the division of labor by sex have all been used to explain sexual dimor phism in humans and other animal species. The etiology of sexual dimorphism has prove n itself complex and not easily understood. All the same, sexual dimorphism is still one of the indices commonly included by anthropologists to describe bi ological change in response to agriculture and in other studies of human health. From the theories presented above, we can see that there are two major forces behind the theory that sexual dimorphism dec lines with the transition to agriculture. First, it has been shown that rates of chronic malnutriti on are higher in agricultural populations than among hunter-gatherers (C ohen and Armelagos, 1984; Stini, 1975). Based on studies of secular declines in sexual dimorphism among malnourished populations (Brauer, 1982; Wolanski and Kasp rzak, 1976), it is reasonable to assume the same trend would apply to the transition to agriculture. Second, it is believed that the biomechanical forces affecting males and fema les are more similar among agriculturalists than hunter-gatherers due to a reduced division of labor among agriculturalists.

PAGE 20

13 Not all of the data collected support this theory. Wolfe and Gray (1982a) compared extant populations and found the opposite of what was expected; the agricultural populations were more sexually dimorphic in stature than were th e hunting and gathering populations they sampled. Data from archaeo logical populations that have undergone the transition to agriculture ha ve found conflicting results (c f. Boyd and Boyd, 1989; Larsen, 1984; Martin et al., 1984; Ruff, 1987). Most studies which compare the degree of sexual dimorphism in hunter-gatherers versus agriculturalists have looked at exta nt unrelated populations (Holden and Mace, 1999; Wolfe and Gray, 1982a). These studies ignore possible genetic contributions to sexual dimorphism or regional differences in the ways in which agriculture was adopted or in the nutritional quality of the food sources cultivated. Archaeological investigations into the tr ansition to agriculture have unearthed a number of sites containing human remains which temporally span the transition in subsistence economy. These remains are of ten of presumably genetically related individuals which eliminates having to account for the influence of genes on sexual dimorphism. By using meta-analysis I have aggregated da ta from individual archaeological sites which span the transition to agriculture in order to test the hypothesis that there was a decrease in sexual dimor phism with the advent of agriculture. By investigating the relati onship between sexual dimorphism and the transition to agriculture I am examining what dimorphism means in terms of human biology and the role of cultural factors in this relations hip. In archaeological contexts, inferences regarding human behavior are often made based on the degree of dimorphism in a population (Bridges, 1989; Bri dges, 2000; Ruff, 1987). In studies of human health,

PAGE 21

14 overall size and the degree of dimorphism are used as indices for the health of human populations (Brauer, 1982; Stin i, 1969; Stinson, 1985; Wolanski and Kasprzak, 1976). In order to make these inferences or utilize thes e indices, it is first necessary to understand the role of sexual dimorphism on an evolutionary scale.

PAGE 22

15 METHODS Meta-Analysis Meta-analysis is a statistical technique th at allows researchers to aggregate data from multiple individual studies. When a num ber of studies address a single issue, the results of said studies do not always agree and the call for more research is often repeated in the literature. The recurring call for a dditional research has left many questioning the validity of continuing to perf orm these individual studies a nd has created the need for cumulative techniques for approaching data (Hunter et al., 1982; Rosenthal, 1984). To explore the impact of the adoption of agriculture on human sexual dimorphism, I first conducted a survey of the results of extant studies investigating the skeletal consequences of this transition. Preliminary i nvestigation of the results of these studies shows that sexual dimorphism decreases in so me areas, increases in others, and in some, there appears to be no change. Any litera ture review can provide the information necessary for a qualitative summar y of the data, but a quantita tive review is advantageous in that it can be methodically scrutinized (Wood and Christ ensen, 2004). By aggregating all the available data collected by vari ous researchers into a single and more comprehensive result, meta-analysis serves as a tool to evaluate existing theories (Wood and Christensen, 2004). Choosing Data Sets The first step in any meta-analysis is to conduct a thorough litera ture review to find all sources of data on the subject. The articl es discussed in this project were found using

PAGE 23

16 a variety of techniques, most importantly the ancestry approach. Rosenthal (1984) describes the ancestry approach as a means of gathering new resources by using the bibliographic information found in an initial article. For this study, Paleopathology at the Origins of Agriculture (Cohen and Armelagos, 1984) served as the primary bibliographic source. I also conducted a numbe r of searches using the internet resources Web of Science and First Search. Keywords for these searches included combinations of the following terms: sexual dimorphism, di morphism, subsistence, agriculture, change, temporal, femur, femoral, osteometric, mo rphology. While these searches did provide a few relevant articles which could then be mined for bibliographic sources, they were largely unsuccessful. Likewise, on-line sear ches of individual anth ropological journals were ineffective. The keywords mentioned a bove were used in on-line searches of the following journals: American Journal of Physical Anthropology, Journal of Human Evolution, Southeastern Archaeology, Amer ican Journal of Archaeology, Journal of Anthropological Archaeology, International Journal of Osteoarchaeology, Current Anthropology, and Yearbook of Physical Anthropology The majority of relevant data were not found in studies of sexual dimorphism, but rather, in studies which investigated within sex variation over time. The purpose of the original publication did not affect the outcome of my study except in cases where P values or test statistics were provided in li eu of complete data. The statistical values provided by the original authors could not be used because different questions were being asked of the data. The primary stipulations for data inclusion in this project were twofold: 1) studies had to present data from both before and afte r the adoption of agricu lture and 2) studies

PAGE 24

17 had to demonstrate geographic continuity. Pr ior studies of tempor al changes in sexual dimorphism have not always limited their samples geographically, sometimes taking a preagricultural sample from one area and a postagricultural sample from another (Brace and Ryan, 1980; Frayer, 1980). By limiting samp les to those found in approximately the same region, I am hoping to control for gene tic factors i.e., that the preand postagricultural data are not known to be from genetically unre lated groups. While testing populations within the same geographic area cannot always assure genetic continuity, archaeological evidence can indicate whether the people in a region are from related cultural groups or whether there is eviden ce of population migrati ons. The introduction of new subsistence strategies or technologies may be suggestive of exchange with an outside population. However, I only in cluded populations wh ere archaeological information was consistent with genetic contin uity. In addition to the issue of genetic continuity, the plants cultivated in different areas of the world may have also had an effect on the biological response to a subsistence shift since st udies show that the effects of agriculture were not uniform worldwid e (Cohen and Armelagos, 1984). Geographic continuity therefore helps cont rol for this moderating factor. Although review of the lite rature provided information on changes in sexual dimorphism spanning the adoption of agricultu re, the majority of these only present a score for percent dimorphism or a difference in means. For the purposes of metaanalysis, only those studies which provided sample sizes, means, and a measure of dispersion (standard deviation or standard error) for any measurement variable were included. A variety of measurement variab les, both linear and bi omechanical, were included in the meta-analyses.

PAGE 25

18 In any meta-analytical process it is as sumed that all studies are independent. Therefore, when multiple studies are conducted using the same data, it is important that the meta-analyst does not consider any group of data more than once. This limited the number of studies included in the meta-analy sis to five, all of which were from areas within the continen tal United States. In Bridge s (1989) and Boyd and Boyd (1989), Archaic and Mississippian populations were c onsidered. Ruff and La rsen (1990) divided the populations they studied into precontact pr eagricultural and precontact agricultural. Contact populations were excluded from this analysis to control for disease and other changes associated with European contact. In the studies of Brock and Ruff (1988) and Table 1: Sample Comp osition for Meta-analysis Source Location Preagricultural sample Postagricultural sample Boyd and Boyd (1989) Tennessee Middl e and Late Archaic Late Mississippian Bridges PS (1989) Alabama Archaic Mississippian Bridges PS (2000) West Central Illinois Middle Woodland Mississippian Brock and Ruff (1988) New Mexico Early Villages Aggregated Villages Ruff and Larsen (1990) Georgia Coast Prec ontact Preagricultural Precontact Agricultural Bridges (2000), the earliest groups studied did have some horticultural supplementation to their diet, but intensive agri culture did not occur until later. Table 1 lists the original authors’ descriptions of the pr e-and post-agricultural samples used for this meta-analysis. Populations that were consider ed “transitional” by the auth or(s) were not considered in this study. Data Analysis: Meta-analysis Deriving the t-statistic. For each measurement variable in each study a t-statistic for change in dimorphism was calcu lated using the following formula: t = (X PM-X PF) (X AM-X AF) / (SE1+SE2+SE3+SE4) (1)

PAGE 26

19 where P and A represent pre-agriculturalist and agriculturalist males and females, respectively. Because probability is base d on sample sizes which are not consistent across the selected studies, a simple comparison of test statistics across studies could not be performed. Determining effect sizes. Meta-analyses rely on effect sizes to test the homogeneity of populations before the indivi dual test statistics can be compared or combined. Effect sizes are calculated from the test statistic, but are measures of magnitude rather than probability. There ar e several different types of effect size indicators, the most common being Pearson’s r Cohen’s d and Glass’ Cohen’s d and Glass’ are both used to investigate the sta ndardized differences between means of independent groups (Source: http://web.uccs.edu/lbecker/Psy590/es/htm last accessed March 23rd, 2005). For these meta-analyses, Pearson’s r was chosen as the effect size indicator because the va riables under investigati on are not considered to be independent. The effect size r was computed for each measurement variable using the following formula: r = ( t2 / t2 + (N1+N2+N3+N4-4)) (2) where N ’s represent the sample size of each sex, before and after the adoption of agriculture. The sign of the t statistic should be preserved when converting to r That is, if t is negative, r should be too. A negative t statistic occurs when the direction of the result is in the opposite direction of th e hypothesis. In the case of this study, a negative result indicates that sexual dimorphism in creases rather than decreases. A t statistic of zero

PAGE 27

20 indicates that there is no change in th e degree of dimorphism over the time period considered. Rather than performing a single analysis of all measurement variables within each study, a meta-analysis was performed on each set of variables, since using multiple data from a single site in a meta-analysis w ould violate the assumption of independent samples. For example, all the data available on the mid shaft circumference of the femur were compiled, but these data were not combined with the data on femoral length. For each measurement variable, the effect sizes were compared and a sample weighted mean correlation was com puted using the following equation: R = [Niri] / Ni (3) The dispersion was describe d by the variance formula: Sr 2 = [Ni (ri r )2] / Ni (4) Tests of homogeneity Effect size estimates were compared to determine whether the studies were from a homogeneous populat ion. Tests of homogeneity do not suggest that every variable within a gr oup is the same, but rather that there is no indication of an outside factor, or moderator variable, infl uencing the relationship being studied. Homogeneity is necessary when comparing or combining the probabi lities of individual studies. The following chi-s quare formula was used to test for homogeneity: Xk-1 2 = ( Ni / (1 r )2) Sr 2 (5) where k is equal to the number of studies used in the meta-analysis. The null hypothesis was that the studies are homogeneous and a P value of 0.05 was require d to reject the null hypothesis.

PAGE 28

21 An alternative method proposed by Hunter et al. (1982) was also performed to test for the presence of a moderator variable. Th is method compares th e error variance and uncorrected variance. The e rror variance was calculated: Ser 2 = k (1 r 2)2 / Ni (6) The error variance was then divi ded by the uncorrected variance: Ser 2/ Sr 2. (7) If Ser 2/ Sr 2 < 0.75, then a moderator variable is pr esent. A heterogeneous population or the presence of a moderator va riable suggests that there is a factor outsid e those being tested that may be influencing the results of the study. These tests do nothing to suggest what that outside influence may be. When a moderator variable is present, the probabilities of individual studi es cannot be combined with reliable results (Source: http://www.fammed.ouhsc.edu/tutor/metanal.htm last accessed March 24th, 2005). Combining probabilities There are several methods of combining the probabilities of individual studies, but the num ber of applicable methods is limited by the small number of studies included in th is analysis. The method of adding Z ’s is the most appropriate under the circumstances of this st udy since it can be us ed regardless of the size of the study. However, it is limited in that it assumes unit variance (Rosenthal, 1984). The Z statistic is based on the area under a standard normal probability curve where the mean is zero and the standard deviation is one. A Z score is associated with the probability that a given value is Z standard deviations away from the mean (Ott and Longnecker, 2001). Adding the Z ’s first requires that we convert our t statistic to Z using the following equation: Z = t (1 – (t2 / 4df)) (8)

PAGE 29

22 where df equals the degrees of freedom. In this case, the degrees of freedom were equal to the sample size of each study divided by four. The Z ’s are then added using the following formula: Z = Z / K (9) where K is the number of studies included in the analysis. Probability was then determined using the statistical tables pr ovided by Ott and Longnecker (2001). Results were determined significant at the 0.05 level. Data Analysis: Non-statistical Analysis of Data Not Appropriate for Meta-Analysis Not all of the information gathered regarding changes in morphology spanning the transition to agriculture could be include d in the meta-analysis due to insufficient data. Several publications presented the mean measurements for ma les and females with no measurement of dispersion. In the non-st atistical analysis, these studies were compared to those used in the meta-analysi s. For each of these populations, dimorphism ratios were produced usi ng the following formula: Ln X Mln X F (10) where X M is the male mean and X F is the female mean. According to the laws of logarithms, Ln X Mln X F is equal to Ln (X M/ X F), a measure of proportion. The decision to use this method of calculating a score of dimorphism was based on Smith, who demonstrated that compared to other methods for finding dimorphism ratios, this method is believed to have fewer problematic mathematical features (1999). Ratios were compared to determine whether there was an increase or decrease in sexual dimorphism over time. In this analysis there was no atte mpt to determine the degree or significance of the change. The non-statistical analysis was completed to determine whether trends found in the meta-analysis were consiste nt over a larger population sample and

PAGE 30

23 geographical area. All formulas in the meta -analysis and the comparison of ratios were calculated using Microsoft Excel. Table 2: Sample Composition for Non-statistical Analysis Source Location Preagricultural sample Agricultural sample Angel (1984) Eastern MediterraneanMesolithic Late Neolithic Boyd and Boyd (1989) Tennessee Middl e and Late Archaic Late Mississippian Bridges PS (1989) Alabama Archaic Mississippian Bridges PS (2000) West Central Illinois Middle Woodland Mississippian Brock and Ruff (1988) New Mexico Early Villages Aggregated Villages Clark (1988) Dickson Mounds, IL Pre-Mississippian Mississippian Martin (1984) Lower Nubia Preagricultural Intensive Agricultural Meiklejohn et al. (1984) Western Europe Mesolithic Neolithic Perzigan et al. (1984) Ohio Rive r Valley Late Archaic Ft. Ancient Rathbun (1984) Iran & Iraq Preagricultural (Hotu) Neolithic Rose (1984) Caddo Culture Area Fourche Maline Caddo II Ruff and Larsen (1990) Georgia Coast Prec ontact PreagriculturalPrecontact Agricultural Smith et al. (1984) Levant Natufian Neolithic and Chalcolithic Ubelaker (1984) Ecuador Sta. Elena Ayalan and Guangala

PAGE 31

24 RESULTS The Meta-analysis From the five studies selected for incl usion in the study, twenty-two individual meta-analyses were conducted. An individu al meta-analysis was performed for each variable, the majority of which were ta ken on the femur and humerus. The linear measurement variables included in meta -analyses were length and mid shaft circumference. At midshaft for both the femur and the humerus, the following biomechanical variables were included: cortical area, mini mum bending strength, maximum bending strength, shape index (Imax / Imin), and polar second moment of area (J). Two studies included subtrochanteric biom echanical data which were included in the meta-analyses. Meta-analyses of the tibia ulna, and radius are limited to midshaft circumference data as biomechanical variab les were considered for these bones in only one of the studies includ ed in this analysis. For an individual meta-analysis to be included in the final analysis, it was necessary that the effect sizes be homogeneous. A comparis on of the effect sizes by chisquare showed that a moderator variable wa s present in three meta-analyses: femoral cortical area, humeral mid shaft circumferen ce and radial mid shaft circumference. In these three cases, the presence of a modera tor variable was conf irmed using Hunter’s (1982) alternate approach. While tests of homogeneity indicate the presence of a moderator variable, they do not provide information on the nature of this outside influence. Since the source of the moderator variable could not be determined from the

PAGE 32

25 Table 3: Summary of Meta-Analyses Measurement Variables r Z P value Femur Length -0.0317 -1.7002 0.9554 Midshaft Circumference -0.0824 -1.4782 0.0929 Min Bending Strength (size standa rdized at midshaft) -0.0731 -0.9215 0.8212 Max Bending Strength (size sta ndardized at midshaft) 0.0625 0.7756 0.2206 Femoral Shape Index (Imax/Imin) 0.0452 0.4558 0.3300 Femoral J (size standardized at midshaft) 0.0401 0.4034 0.3446 Subtrochanteric Cortical Ar ea (size standardized) 0.1991 1.3859 0.0838 Subtrochanteric Min. Bending Strength (size stand.) 0.1408 1.2116 0.1131 Subtrochanteric Max. Bending Strength (size stand.) 0.1867 1.6103 0.0537 Subtrochanteric Femoral Shape Index (Imax/Imin) 0.1546 0.4167 0.3409 Subtrochanteric J (si ze standardized) 0.2010 1.7342 0.0418** Humerus Humerus Length 0.0054 -0.0941 0.5359 Humerus Cortical Area (size sta ndardized at midshaft) 0.0507 0.4692 0.3228 Humerus Min. Bending Strength (size standardized) 0.1361 1.3011 0.0968 Humerus Max. Bending Strength (size standardized) 0.1635 1.8789 0.0307** Humerus J (size standardized at midshaft) 0.1232 0.3216 0.3745 Humeral Shape Index (Imax/Imin) -0.1420 -0.8878 0.8106 Other Tibia Midshaft Circumference 0.0178 0.1956 0.4247 Ulna Midshaft Circumference 0.0131 0.1747 0.4325 Table does not include meta-analyses in wh ich a moderator variable was found to be present through the chi-square test. For additional data, see appendix. r = sample weighted mean effect size Z = test statistic ** Significant (p< 0.05) Near Significant (0.1 > p > 0.05)

PAGE 33

26 data, these three meta-analyses were removed from further study. In the meta-analysis of femur length, a chi-square test suggested th e population was homogeneous but Hunter’s (1982) alternative approach de tected the presence of a moderator variable. Although Hunter’s approach detected the presence of a moderator variable, the result of 0.68 was not far below the 0.75 threshold for significance. The chi-square test strongly suggested the population to be homogeneous and as such this meta-analysis was not excluded from the final study, leaving nineteen meta-analy ses for comparison in the final analysis. The femur. In the eleven independent meta-analyses of the femur, three (27%) of the sample weighted mean correlations ( r ) were negative. A negative effect size indicates that the degree of sexual dimorphi sm in the femur increased, rather than decreased, with the shift to agriculture. Two of the three negative effect sizes were for linear measurement variables: length and midshaft circumference. The third measurement variable was a de rived biomechanical variable. Only one femoral measurement variable experienced a significant decline with the transition to agriculture: the s ubtrochanteric polar second mome nt of area (J). J is found by summing I values from two perpendicular me asurements. When a section is perfectly cylindrical, J is a measure of strength unde r torsional twisting. Although these sections are likely elliptical, the cited au thors have used this measure as an estimate of torsional or twisting strength (Bridges, 2000; Brock a nd Ruff, 1988; Ruff a nd Larsen, 1990). A second femoral measurement variable, subtro chanteric maximum bending strength, had a near significant result of p = 0.0537. Both of these measurement variables are biomechanical and provide information on skeletal strength and robusticity.

PAGE 34

27 The humerus. Of the six meta-analyses performed on the humerus, only the humeral shape index (Imax/Imin) had a negative sample weighted mean correlation. All other results were positive suggesting that a slight decline occurs more often than an increase. In only one meta-analysis was th e change in sexual dimorphism significant. The degree of sexual dimorphism found in the maximum bending strength of the humerus is significantly less in agricultu ral than in preagricultural populations. Combined meta-analyses. The majority of meta-analyses show no significant change in the degree of sexual dimorphism between preagricultural and agricultural populations. Two measurement variables (10.5% of the total) had a significant decline in the amount of sexual dimorphism and a third measurement variable had a near significant decline. These three measurement vari ables, humeral maxi mum bending strength, subtrochanteric maximum bending strength, and subtrochanteric J, are all biomechanical rather than linear variables, indicating that sexual differen ces in strength experienced a decline. A decline in dimorphism indi cates that either the male mean is decreasing or the female mean is increasing. The original da ta from the significant and near-significant meta-analyses were examined to determine wh ether or not the within sex changes which occurred with the adoption of agriculture were largely male or female. The preagricultural and agricultural mean measurements for males and females were subtracted from one another to determine whet her the greatest tempor al differences could be attributed to the males or females. The re sults of this investiga tion were inconclusive. In some cases the males experienced the gr eater change and in other cases females

PAGE 35

28 exhibited a greater change. Therefore, it does not appear that si gnificant changes in dimorphism can be attributed to one sex across studies. Non-Statistical Analysis A comparison of the sexual dimorphism ratios (Ln X Mln X F ) for all studies, including those used in the meta-analysis, showed that sexual dimorphism decreased in 54% of cases, only slightly more than half. Measurements of the femur were less likely to decrease in dimorphism than were meas urements of the humerus. While this comparison has no statistical si gnificance, it mirrors what is found through the metaanalysis. Overall, there does not appear to be a noteworthy difference in the degree of sexual dimorphism found in populations befo re and after the adop tion of agricultural subsistence. However, there is a slight tr end toward a decline in dimorphism for all bones with the exception of the femur. Similar trends appear when linear and biomechanical data are compared. To determine whether the data availabl e provided information on geographical differences in patterns of sexual dimorphi sm, studies were divided on the basis of geographical location into New World and Old World samples. However, of the total 207 data points, only five were from Old World populations. Of these, three showed an increase in sexual dimorphism ratios and two exhibited a decrease. Four of the measurement variables were measures of st ature (without indicati on as to which long bones were used for the calculation) and one measurement variable was for maximum femur length. Due to insufficient data fr om Old World sites, it was impossible to compare Old and New World populations. However, the limited Old World data appear to reflect the trend found th roughout the total data set.

PAGE 36

29 Table 4: Non-Statistical Analysis of Changes in Sexual Dimorphism Ratios Increase Decrease Total (207) 9344%113 54% Femur (106) 6258%44 42% Humerus (56) 2137.5%21 62.5% Radius (12) 18%11 92% Tibia (14) 17%13 93% Ulna (10) 220%8 80% Linear Variables (74) 2838%46 62% Femur (27) 1970%8 30% Humerus (10) 220%8 80% Biomechanical Variables (115) 5850%57 50% Femur (72) 4157%31 43% Humerus (43) 1740%26 60%

PAGE 37

30 DISCUSSION Theoretically there are two reasons why a decline in sexual dimorphism would accompany a transition to agriculture. First, cu ltural research has shown that the type of work performed by males and females in agricu ltural cultures is more similar than the gendered duties of the hunte r-gatherer. Murdock and Pr ovost (1973) conducted a crosscultural study of work in which they code d all tasks performed by each sex and whether the activities were assigned to males or fema les, partially or comp letely. There was a greater delineation between male and female duties in hunting and gathering populations than among agriculturalists. Theoretically sexual dimorphi sm decreased as male and female mechanical loads became more simila r with the transition to agriculture. In addition, nutrition generally dec lined with the transition to agriculture. Nutritional decline occurred due to several f actors. First, the amount of pr otein in the di et declined. Agricultural sedentism meant that an area around a village could easily be over hunted. Second, for the mobile hunter-gatherer there is greater variation in the types of plant food collected. The few agricultural crops did not always provide th e nutritional value associated with variation, pa rticularly in the case of maize (Cohen and Armelagos, 1984; Larsen, 1995). Under conditions of nutritional stress, males are less likely than females to reach their full potential size, thereby re ducing the degree of sexual dimorphism (Stini, 1969). If the degree of sexual dimorphism in a population does not decline with the transition to agriculture, it may be explaine d in several ways. Fi rst, the changes in

PAGE 38

31 nutritional value may not have been great enoug h to effect a change in the morphology of a population, or, perhaps, changes in morphology were short-lived and difficult to discern in the archaeological record. Other cultural va riables, such as preferential treatment of male children, may obscure any nutritional changes which may have affected dimorphism (Ortner, 2003; Rivers, 1982). In reference to the biomechanical argument, changes in loading may not have been of the type or in tensity to change sexual dimorphism. Finally, males and females may have both changed their behaviors in such a wa y that the level of dimorphism remained the same with the adoption of agricultural subsistence. This meta-analytical survey found that th e decline in sexual di morphism with the shift to agriculture was significant in two of nineteen measurement variables (10.5% of the total). While four measurement variable s showed an increase in dimorphism over time (21% of the total), none of these resu lts were significant. The meta-analyses therefore suggest that there is a slight trend toward a reduction in dimorphism, but, overall, the temporal changes in sexual dimo rphism with the change in subsistence are not great. It is important to remember that the meta-analyses conducted in this study addressed changes in sexual dimorphism rath er than within sex changes in morphology. Many of the individual studies compared did find significa nt changes in morphology with the transition to agriculture that were not addressed by the meta-analyses conducted here. At the Upper Paleolithic – Mesolithic transition, the changes in morphology which accounted for a decline in sexual dimorphism we re mostly due to the gracilization of the male which most researchers associate with a reduction in selection for large size due to the advancement of hunting technologies and smaller game species. Female size (and theoretically, activ ity pattern) did not undergo any majo r changes at that time. Frayer

PAGE 39

32 (1980) found that the reduction in dimorphi sm which occurred at the dawn of the Neolithic was more closely associated with changes in the female form rather than the male. While he does not offer an explanation for these results it is reasonable to assume that changes in the female form may be particularly pronounced in populations where females were the primary agriculturalists. Ru ff et al. (1984) found th at of the biological changes associated with the transition to ag riculture (increased periosteal infections, increased frequency of dental caries, and decr eased stature and robusticity) the majority were more prevalent in females than males. In the case of the Georgia coastal populations these changes are t hought to result from heavier female involvement in agricultural activities and greater female consum ption of corn (Ruff, 1987). Analysis of the combined data used in this study could not attribute changes in dimorphism to either sex. Changes in individual studies may be largely male or female, but there is no consistent pattern when studies are compared. Appropriate Indices of Dimorphism The data collected for this review consis ted of twenty-two different osteological measurement variables from which sexual dimo rphism could be calculated. A review of the literature provides many other measurem ent variables of dimorphism which have been used to compare populations. This pr esents the researcher with two important questions: 1) is there an a ppropriate index of dimorphism and 2) does it depend on the questions which are being asked of the data? A number of studies have used stature, or the sexual dimorphism of stature as an indica tor of overall health in a population (Brauer, 1982; Holden and Mace, 1999; Wolanski and Kasprzak, 1976; Wolfe and Gray, 1982a). Unlike many osteometric measurements, stat ure can be easily measured in extant populations and compared to skeletal populatio ns through a number of available formulae

PAGE 40

33 for the estimation of stature. However, fo r skeletal populations, st ature may not be the most reliable indicator of dimorphism. First, all formulas used to calculate dimorphism assume some amount of error. Second, when available, the femur or tibia is most commonly used to calculate stature due to the fact that the stature calculations for those bones have a smaller error than other long bone s. In the analyses conducted for this study, the trend toward a decline in dimorphism was less evident in the femur than it was in measurements of the humerus. Therefore, perhaps stature is not the most sensitive measurement for explaining changes in sexual dimorphism over time. Of the twelve meta-analyses conducted on the femur, three ( 25%) had a negative mean effect size indicating that sexual dimor phism increased rather than declined. For the humeral meta-analyses, only one of seven (14%) mean effect sizes was negative. The twelve meta-analyses used to describe the femur were drawn from a total of 35 effect sizes, one for each measurement variable consid ered in each study. Forty two percent of the femoral effect sizes were negative comp ared to 17% for the humerus. Ruff et al. (1993) evaluated temporal changes in postcrani al robusticity and found that a decline in femoral diaphyseal robusticity was consistent in humans from the early Pleistocene through recent populations whereas trends in upper limb robusticity were more difficult to decipher. While these results may encour age researchers to evaluate the femur when studying temporal changes in robusticity, the same confidence should not be extended to studies of temporal chan ges in sexual dimorphism. Measurements of sexual dimorphism on archaeological samples are limited to adults due to the difficulty of sexing juven ile remains. However, studies of extant populations have proven that problems exist in interpreting the remains of adult sex

PAGE 41

34 differences. While juvenile populations may exhi bit differential effects of stress, many of these effects may be corrected through “cat ch-up” growth before reaching adulthood (Stini, 1975). Therefore, adult size may not the most sensitive indicator of the differential effects of stress. Ruff (1987; 1984) suggests that the differe nt measurements of sexual dimorphism may be evidence for particular causal factors. He suggests that cross-sectional data reveal more information about the forces ac ting on bone and, therefore, activity patterns, whereas changes in size or stature, linear da ta, are more likely to be due to nutritional factors (Ruff et al., 1984). If this is indeed the case, it is important to note that the only significant results in this study were for cross-sectional data. The Transition to Agriculture Theories regarding sexual dimorphism a nd the transition to agriculture are not based solely on the dietary effects of cultigen s, but rather on the changes that accompany this subsistence change. P opulation growth accompanied the transition to agriculture although theories differ as to how these two fact ors relate to one anot her. Some believe that the growth in population forced people to adopt an agricultural economy; others think that the surplus food a nd sedentary nature of agricu ltural economies allowed for population expansion (Boserup, 196 5; Bronson, 1977; Cohen, 1977). With increased sedentism and population expa nsion, the rates of in fectious diseases also increased after the transi tion to agriculture. Sedentism is particularly problematic under marginal environmental conditions since mobility can be beneficial in allowing a person to flee the worst circumstances. In the studies compiled by Cohen and Armelagos (1984) there appears to be an overall decline in the quality and length of life with the transition to agriculture. It could be argued that morphological changes associated with

PAGE 42

35 the transition to agriculture are due to popul ation expansion rather than subsistence change. However, these two changes are largely inseparable. In some areas of the world there were ch anges that accompanied the transition to agriculture that were not universal and need to be considered as possible moderating factors to any morphological ch ange. For example, Bridges (2000) points out that the atlatl was being replaced by the bow as a hunting tool around the same time that the transition to agriculture was occurring in some areas of North America. These two changes did not occur simultaneously worl dwide, but may account for some of the morphological changes seen at the transition to agriculture. Brues’ (1959) spearmanarcher hypothesis suggested that the mechan ical needs of the two weapon types would afford a selective advantage to different body ty pes. While attempts to test this theory have proven it highly suspect, it must be cons idered that any change to the mechanical loading of bone which accompanied the tran sition to agriculture may prove to be a moderating variable in studies of morphological change. The adoption of agriculture has been asso ciated with an increase in infectious diseases due to increased sedentism and a population expansion. Population expansion allowed viral diseases such as measles, mumps and small pox to be more easily communicable than ever before. Malaria, ch olera, blastomycosis, and scrub typhus are examples of diseases associated with agri culture due to an increased exposure to zoonoses when turning soil and exposure to contaminated water (Armelagos, 1990; Ortner, 2003). Similarly, new ideas and technologies, such as those associated with subsistence changes, may be introduced through trade or migration. In many ar eas of the world, the

PAGE 43

36 transition to agriculture was a ssociated with a colonizing force. None of the populations in this study were subject to the diseases introduced with European contact, but the introduction of disease should be considered as a moderating va riable in some areas. Ruff and Larsen (1990) were able to compare precontact agricultural groups with postcontact agricultural groups on the Georgia coast. On the Georgia coast it appears that sexual dimorphism in femur length increased with the adoption of agriculture but the trend then decreased after contact. The increa se in dimorphism that marked the adoption of agriculture along the Georgia coast has been associated with a culture in which the negative effects of corn agriculture appear to have affected females much more so than males. Throughout much of southeastern North America after the adoption of agriculture, females were the farmers while the primary subsistence activity of men remained hunting (Swanton, 1946). Contact wi th the Spanish brought diseases and a mission system which regimented the lives of both males and females. This example provides valuable information on the eff ect of culture on morphological changes spanning the transition to agriculture. In this study, the size of males and fema les were compared in agricultural and preagricultural populations. However, the dividing line separating these two groups is not always clear. Bronson (1977) argues that cu ltivation of plant foods began as early as the Paleolithic. By selectively discarding waste from food plants in areas where they wanted plants to grow, people began propa gating plant species. In North America, Native Americans cultivated local seed crops before maize ( Zea mays ) was introduced from Mesoamerica. Exposure to agriculture may thus have occurred long before the full adoption of an agricultural economy, and food cr ops may have been limited to a certain

PAGE 44

37 segment of the population before they b ecame dietary staples among the population at large. For the purposes of this study, a populati on had to have a dependency on at least one domesticated carbohydrate crop to be consid ered agricultural. In Bridges (2000) the Middle Woodland population described had sm all scale cultivation of food crops. However, this population is considered “pr eagricultural” in this study because it is believed that agricultural products and activit ies had not become a significant portion of this population’s diet and lifestyle. Determining the presence of agriculture in an archaeological site can be based on the identification of pl ant remains, associated material culture, or through skeletal indicators of agricultural subs istence. It is important to remember that the lack of botanical samples may be the result of poor preservation, or, as Rose et al. (1984) point out, many sites were excavated prior to the a doption of flotation techniques for retrieving paleobotanical remains. Conversely, the pr esence of plant domesticants does not necessarily indicate that agri cultural products were a dietary staple of the population at large. Several skeletal pathologies are a ssociated with agricultural populations, but an increase in the frequency of car ious lesions in the dentition is universal and as such they are used as one indicator of the presence of agriculture (Lar sen, 1984; Rose et al., 1984). Turner (1979) compared global samples and found that the average frequency of teeth affected by carious lesions in hunting a nd gathering groups is 1.72%; mixed hunting, gathering and farming groups average 4.37%; and agriculturalists average 8.56%. These rates vary regionally based on the food sources available. Agricultural diets are high in carbohydrates which are the primary cause of de ntal caries. Chemical analysis of bone

PAGE 45

38 may also be used as a method of deducing pa leodiet, particularly in the Americas where corn leaves such a clear chemical signature. Many studies contain skeletal data fro m time periods that are considered transitional with regard to agriculture. Martin et al. (1984) found that agricultural intensification, rather than ag ricultural origins, were account able for differential patterns of biological response, and, as such, only t hose groups in which agriculture was fully entrenched are used in this study. N onetheless, transitional groups can provide information about how changes occurred. Nickens’ (1976) study of stature reduction with the adoption of agriculture suggests th at body size declined with the adoption of agriculture, but then increased as humans adapted, perhaps learning to compensate for some of the negative effects of agriculture. This suggests that some of the recorded morphological changes observed with the tran sition to agriculture may represent shortterm rather than evolutionary changes. Ta ble 5 provides sexual dimorphism ratios from a study which includes transitional populations. Of particular interest are the cases where Middle Woodland and Mississippi an groups have very simila r sexual dimorphism ratios, while the intermediate populations differ gr eatly (for example, see femoral maximum bending strength, humeral cortical area, and hum eral minimum bending strength). If the transitional groups were not included, it w ould appear there wa s little morphological change with the transition to agriculture. If some morphological changes are only shortterm, they may be difficult to detect in th e archaeological recor d, particularly when comparing groups which may be at different stages in the transition to agriculture.

PAGE 46

39 Table 5: Sexual Dimorphism Ratios in Transitional Groups (Bridges 2000) Middle Early Late Late Late Woodland* Woodland Woodland Mississippian Femoral Cross-Sectional Properties Cortical Area 0.1700 0.0541 0.0228 0.0675 Min Bending Strength 0.0055 -0.0334 0.0165 0.0847 Max Bending Strength 0.2877 0.0271 0.0990 0.2281 Torsional Strength 0.2665 0.0010 0.0615 0.1674 Shape Index (Imax / Imin) 0.0455 0.1008 0.0729 0.1542 Femoral linear data and indices Bicondylar Length 0.0733 0.0864 0.0760 0.0760 AP diameter (midshaft) 0.1247 0.1520 0.1514 0.1456 ML diameter (midshaft) 0.0794 0.0551 0.0738 0.0666 Circumference (midshaft) 0.1158 0.1098 0.1088 0.1118 AP diameter (subtroch) 0.1238 0.1163 0.1223 0.1207 ML diameter (subtroch) 0.1007 0.0859 0.1076 0.0924 Circumference (subtroch) 0.1007 0.0949 0.1341 0.1037 Vert Head Diam Subtroch 0.1117 0.1051 0.1265 0.1394 Pilastric Shape Index 0.0461 0.0987 0.0741 0.0793 Platymeric Shape Index 0.0270 0.0270 0.0136 0.0270 Humeral Cross-sectional Properties Cortical Area 0.4143 0.1665 0.1032 0.4149 Min Bending Strength 0.3222 -0.1452 0.0440 0.2678 Max Bending Strength 0.2865 -0.2187 -0.1178 0.3005 Torsional Strength 0.2990 -0.1927 -0.0563 0.2900 Shape Index (Imax / Imin) -0.0357 -0.0818 -0.1542 0.0255 Humeral linear data and indices Length 0.0854 0.0881 0.0795 0.0734 Max diameter (midshaft) 0.1110 0.0745 0.0612 0.1089 Min diameter (midshaft) 0.1353 0.0831 0.0953 0.1133 Circumference (midshaft) 0.1172 0.0918 0.0712 0.1016 Max diameter (min. shaft) 0.1205 0.1112 0.0905 0.0943 Min diameter (min. shaft) 0.1550 0.0827 0.0927 0.1313 Circumference (min. shaft) 0.1161 0.0927 0.0794 0.1025 Midshaft shape index 0.0267 0.0000 0.0270 0.0132 Min. shaft shape index 0.0290 -0.0296 0.0000 0.0396 *In the meta-analysis, Middle Woodland and Mi ssissippian groups were compared as the preagricultural and agricultura l populations. Early late and late late Woodland groups are considered transitional.

PAGE 47

40 Limitations of Meta-Analysis In using meta-analysis it is important to understand the inherent assumptions and limitations of the technique. First, meta-analysis assumes that all available data is included in the analysis. For the research er compiling these data, this presents the problem of publication bias or the “file drawer” problem; studies with significant results are more likely to be published than those w ith insignificant results (Rosenthal, 1984). In this review, the nature of the publication bias was different. Most of the studies used in the meta-analysis were not primarily inve stigating sexual dimorphism. Therefore, significant results in the original publicati ons were not an issue. Rather, finding sufficient data was a problem due to the standards used for presenting data. The vast majority of publications on th e transition to agriculture provided mean values for the populations studied, with no information on the dispersion of the sample. Hence, statistical data could no t be gathered from these studies. Situations such as these present a problem for the meta-analyst as well as a ny reader hoping to criti que the results of a study. In collecting data, the hope was to find archaeological sites worldwide from which information could be drawn. However, the studies which could be included in my sample are affected by publication bias. As an English-speaker, all the journals and resources I collected were from sources written in English; all the studies I collected with data sufficient for a meta-analysis were from sites within the continental United States. Requests for data from international sources went unanswered. Furthermore, differential preservation of skeletal remains in the ar chaeological record is likely to skew any worldwide study of temporal changes in morphology.

PAGE 48

41 The number of studies which contain da ta on skeletal morphology spanning the transition to agriculture is far greater than th e number that could be included in a metaanalysis. A meta-analysis assumes independent samples. Skeletal remains are culturally sensitive materials and laws lim it the access to these remains. As such, the same skeletal materials are often used in a number of differe nt studies. No skeleton could be used more than once without violating the assumptions of meta-analysis. Due to these problems in data collect ion, the results of this study are only applicable to the transition to agriculture as it occurred in North America. While a variety of locations and culture s are considered in these North American samples, they are all of similar ancestry and adopted mai ze as the primary agri cultural carbohydrate. To assess the effect of the transition to ag riculture on sexual dimo rphism worldwide, it would be necessary to include studi es from other areas of the globe. Meta-analysis has been criticized fo r glossing over the de tails of individual studies. However, Rosenthal (1984) argues that the same can be said of any traditional review. In fact, a meta-analysis is more likely to reflect the actual results of studies rather than being overly influenced by information included in their abst racts or discussions (Rosenthal, 1984). Another common criticism of meta-analysis is that comparing different studies, by different researchers, is like comparing apples and oranges. Not only are methods heterogeneous, but the quality of the studies may also vary. To control for heterogeneity, only like measurement variables were compared in twenty-two separate meta-analyses. However, in this study several measuremen ts were combined although the methods of data collection were not identical. For example, measurements of maximum femur

PAGE 49

42 length and bicondylar length were combined for the meta-analysis of femur length. Measurements of bending strength in differe nt studies were collected using similar methods, but then each author used a differe nt formula for standardization. A metaanalysis would not have been possible had th ese measurements not been compared. To do so did not violate any assumption of independence and provided important information which would have otherwise been unavailable. Glass made an excellent point in defense of such generalizations. “O ne compares apples and oranges in the study of fruit” (Glass, 1978). The criticisms of meta-analysis have been addressed by its proponents and the method has proved reliable when used for its intended purpose (Glass et al., 1981; Hunter et al., 1982; Rosenthal, 1984). In this study the technique ha s been employed to test the theory that sexual dimorphism declined with the transition to agriculture. While this meta-analysis does not directly inform the question of causality, it provides information about patterns of dimorphism which may then serve future studies of sexual dimorphism. The availability of data has restricted the a pplicability of the results to North American populations but has provided a needed cumu lative analysis of how sexual dimorphism may be affected by the transition to ag riculture in a larg e geographic area.

PAGE 50

43 CONCLUSIONS Meta-analyses were conducted on the postc ranial measurements taken in five separate studies. Of the nineteen measur ements from which populations were deemed homogeneous, two experienced a significant de cline with the transition to agriculture (10.5 %) and a third experienced a near sign ificant decline (p = 0.0537). A comparison of the effect sizes found that 79 % of a ll measurements experienced a decline in dimorphism as opposed to the remaining 21 % in which a negative e ffect size indicated that the degree of dimorphism increased. N one of the increases in dimorphism were found to be significant. The tr end towards a decline in dimor phism is more apparent in humeral than in femoral measurements. The non-statistical analysis of agricultural versus preagricultural populations shows similar results. In contrast to the theory that sexual dimorphism declines with the transition to agriculture, in most cases, no significant change occurs.

PAGE 51

44 APPENDIX A META-ANALYSES OF FEMORAL MEASUREMENTS Femur Length (based on maximum or bic ondylar length measurements) source study # t statistic r N df Z Boyd and Boyd (1989) 1 0.23430.0103524520 0.2343 Bridges (2000) 3 -0.1495-0.0135127123 -0.1495 Brock and Ruff (1988) 4 -1.3684-0.1173138134 -1.3636 Ruff and Larsen (1990) 5 -2.195-0.34364036 -2.1216 rbar = -0.0317 Xk-1 2 = 5.5141 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0071 Ser 2/Sr 2 = 0.6802 Moderator Variable Present Ser 2 = 0.0048 Z = -1.7002 p = 0.9554, N.S. Femur Midshaft Circumference source study # t statistic r N df Z Bridges (1989) 2-2.156-0.1786174170 -2.1413 Bridges (2000) 300126122 0 Brock and Ruff (1988) 4-0.4192-0.0362138134 -0.4190 rbar = -0.0824 Xk-1 2 = 2.3564(0.5 > p > 0.1) Homogeneous Sr 2 = 0.0063 Ser 2/Sr 2 = 1.0721No moderator variable present Ser 2 = 0.0068 Z = -1.4782p =0.0929, N.S. Femoral Cortical Area (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 -1.4781-0.227644 40 -1.4579 Bridges (2000) 3 1.26170.205840 36 1.2478 Brock and Ruff (1988) 4 2.30470.342444 40 2.2282 Ruff and Larsen (1990) 5 -0.2271-0.037840 36 -0.2270 rbar = 0.0701 Xk-1 2 = 9.6710 (0.01 > p > 0.005) Heterogeneous Sr 2 = 0.0498 Ser 2/Sr 2 = 0.4736 Moderator Variable Present Ser 2 = 0.0236 Z = 0.8955 p = 0.1867, N.S.

PAGE 52

45 Femur Minimum Bending Strength (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 -2.0041-0.302144 40 -1.9538 Bridges (2000) 3 -0.6359-0.105440 36 -0.6342 Brock and Ruff (1988) 4 1.10420.172044 40 1.0958 Ruff and Larsen (1990) 5 -0.3512-0.058440 36 -0.3509 rbar = -0.0731 Xk-1 2 = 4.3424 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0298 Ser 2/Sr 2 = 0.7914 No Moderator Variable Ser 2 = 0.0236 Z = -0.9215 p = 0.8212, N.S. Femur Maximum Bending Strength (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 -0.4867-0.076744 40 -0.4860 Bridges (2000) 3 0.31080.051740 36 0.3106 Brock and Ruff (1988) 4 1.56270.239944 40 1.5389 Ruff and Larsen (1990) 5 0.18770.031340 36 0.1876 rbar = 0.0625 Xk-1 2 = 2.5951 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0136 Ser 2/Sr 2 = 1.7400 No Moderator Variable Ser 2 = 0.0236 Z = 0.7756 p = 0.2206, N.S. Femoral Shape Index (Imax/Imin) source study # t statistic r N df Z Bridges (2000) 3 -1.1831-0.193540 36 -1.1716 Brock and Ruff (1988) 4 1.27280.197344 40 1.2599 Ruff and Larsen (1990) 5 0.70350.116540 36 0.7011 rbar = 0.0452 Xk-1 2 = 3.8381 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0282 Ser 2/Sr 2 = 0.8538 No Moderator Variable Ser 2 = 0.0241 Z = 0.4558 p = 0.3300, N.S.

PAGE 53

46 Femoral J (J=Iap + Iml, polar second moment of area, size standardized at midshaft) source study # t statistic r N df Z Bridges (2000) 3 -0.2930-0.048840 36 -0.2928 Brock and Ruff (1988) 4 1.76510.268844 40 1.7307 Ruff and Larsen (1990) 5 -0.7420-0.122740 36 -0.7391 rbar = 0.0401 Xk-1 2 = 3.9915 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0297 Ser 2/Sr 2 = 0.8130 No Moderator Variable Ser 2 = 0.0241 Z = 0.4034 p = 0.3446, N.S. Subtrochanteric Cortical Area (size standardized) source study # t statistic r N df Z Brock and Ruff (1988) 4 2.38220.364741 37 1.2589 Ruff and Larsen (1990) 5 0.17670.029440 36 0.7011 rbar = 0.1991 Xk-1 2 = 3.5471 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0281 Ser 2/Sr 2 = 0.8107 No Moderator Variable Ser 2 = 0.0228 Z = 1.3859 p = 0.0838, N.S. Subtrochanteric Minimum Bending Strength (size standardized) source study # t statistic r N df Z Brock and Ruff (1988) 4 1.44230.230741 37 1.4221 Ruff and Larsen (1990) 5 0.29160.048540 36 0.2914 rbar = 0.1408 Xk-1 2 = 0.9102 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0083 Ser 2/Sr 2 = 2.8595 No Moderator Variable Ser 2 = 0.0237 Z = 1.2116 p = 0.1131, N.S. Subtrochanteric Maximum Bending Strength (size standardized) source study # t statistic r N df Z Brock and Ruff (1988) 4 1.04680.169641 37 1.0391 Ruff and Larsen (1990) 5 1.25190.204240 36 1.2382 rbar = 0.1867 Xk-1 2 = 0.0367 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0003 Ser 2/Sr 2 = 76.6613 No Moderator Variable Ser 2 = 0.0230 Z = 1.6103 p = 0.0537, Near Significant

PAGE 54

47 Subtrochanteric Femoral Shape Index (Imax/Imin) source study # t statistic r N df Z Brock and Ruff (1988) 4 -0.65000.106341 37 -0.6481 Ruff and Larsen (1990) 5 1.25100.204140 36 1.2374 rbar = 0.1546 Xk-1 2 = 0.2713 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0024 Ser 2/Sr 2 = 9.8274 No Moderator Variable Ser 2 = 0.0235 Z = 0.4167 p = 0.3409, N.S. Subtrochanteric J (J=Iap + Iml, polar second moment of ar ea, size standardized at midshaft) source study # t statistic r N df Z Brock and Ruff (1988) 4 1.44230.230741 37 1.4221 Ruff and Larsen (1990) 5 0.29160.048540 36 0.2914 rbar = 0.2010 Xk-1 2 = 0.1569 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0012 Ser 2/Sr 2 = 18.3894 No Moderator Variable Ser 2 = 0.0227 Z = 1.7342 p = 0.0418, SIGNIFICANT

PAGE 55

48 APPENDIX B META-ANALYSES OF HUMERAL MEASUREMENTS Humerus Length source study # t statistic r N df Z Ruff and Larsen (1990) 5 -0.9413-0.129456 52 -0.9373 Bridges (2000) 3 0.75010.0710115 111 0.7491 rbar = 0.0054 Xk-1 2 = 1.4492 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0084 Ser 2/Sr 2 = 1.3949 No Moderator Variable Ser 2 = 0.0117 Z = -0.0941 p = 0.5359, N.S. Humerus Midshaft Circumference source study # t statistic r N df Z Bridges (1989) 2 2.20740.1453220 216 2.1949 Bridges (2000) 3 0.49430.0467116 112 0.4941 rbar = 0.1112 Xk-1 2 = 6.4901 (0.025 > p > 0.01) Heterogeneous Sr 2 = 0.0153 Ser 2/Sr 2 = 0.3805 Moderator Variable Present Ser 2 = 0.0058 Z = 1.3445 p = 0.090, N.S. Humerus Cortical Area (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 0.23260.034749 45 0.2325 Bridges (2000) 3 0.29950.062340 36 0.2993 Ruff and Larsen(1990) 5 0.40680.056356 52 0.4065 rbar = 0.0507 Xk-1 2 = 0.0220 (0975 > p > 0.9) Homogeneous Sr 2 = 0.0001 Ser 2/Sr 2 = 150.7371 No Moderator Variable Ser 2 = 0.0206 Z = 0.4692 p = 0.3228, N.S.

PAGE 56

49 Humerus Minimum Bending Strength (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 1.87290.268949 45 1.8364 Bridges (2000) 3 -0.6359-0.105440 36 -0.6342 Ruff and Larsen(1990) 5 1.41350.192456 52 1.3999 rbar = 0.1361 Xk-1 2 = 4.5214 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0233 Ser 2/Sr 2 = 0.8564 No Moderator Variable Ser 2 = 0.0199 Z = 1.3011 p = 0.0968, N.S. Humerus Maximum Bending Strength (size standardized at midshaft) source study # t statistic r N df Z Bridges (1989) 2 1.99150.284649 45 1.9476 Bridges (2000) 3 0.31080.051740 36 0.3106 Ruff and Larsen(1990) 5 1.00100.137556 52 0.9962 rbar = 0.1635 Xk-1 2 = 1.7955 (0.5 > p > 0.1) Homogeneous Sr 2 = 0.0087 Ser 2/Sr 2 = 2.2621 No Moderator Variable Ser 2 = 0.0196 Z = 1.8789 p = 0.0307, SIGNIFICANT Humeral Shape Index (Imax/Imin) source study # t statistic r N df Z Bridges (2000) 3 -0.9142-0.187327 23 -0.9059 Ruff and Larsen(1990) 5 -0.8729-0.120256 52 -0.8697 rbar = -0.1420 Xk-1 2 = 0.0629 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0010 Ser 2/Sr 2 = 23.4217 No Moderator Variable Ser 2 = 0.0231 Z = -0.8878 p = 0.8106, N.S.

PAGE 57

50 Humeral J (J=Iap + Iml, polar second moment of area, size standardized at midshaft) source study # t statistic r N df Z Bridges (2000) 3 -0.39420.081927 23 -0.3935 Ruff and Larsen(1990) 5 1.22800.167956 52 1.0368 rbar = 0.1399 Xk-1 2 = 0.1819 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0016 Ser 2/Sr 2 = 14.2832 No Moderator Variable Ser 2 = 0.0232 Z = 0.5838 p = 0.2810, N.S.

PAGE 58

51 APPENDIX C META-ANALYSES OF OTHER LONG BONE MEASUREMENTS Tibia Midshaft Circumference source study # t statistic r N df Z Bridges (1989) 2 00121 117 0 Bridges (2000) 3 0.39140.046176 72 0.3912 rbar = 0.0178 Xk-1 2 = 0.0631 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0003 Ser 2/Sr 2 = 32.8342 No Moderator Variable Ser 2 = 0.0101 Z = 0.1956 p = 0.4247, N.S. Ulna Midshaft Circumference source study # t statistic r N df Z Bridges (1989) 2 -0.0910-0.0082140 136 -0.0910 Bridges (2000) 3 0.44060.050580 76 0.4403 rbar = 0.0131 Xk-1 2 = 0.1244 (0.9 > p > 0.5) Homogeneous Sr 2 = 0.0006 Ser 2/Sr 2 = 16.4999 No Moderator Variable Ser 2 = 0.0091 Z = 0.1747 p = 0.4325, N.S. Radius Midshaft Circumference source study # t statistic r N df Z Bridges (1989) 2 2.97190.2513142 138 2.9243 Bridges (2000) 3 0.90750.094795 91 0.9054 rbar = 0.1885 Xk-1 2 = 14.8913 ( p < 0.01) Heterogeneous Sr 2 = 0.0414 Ser 2/Sr 2 = 0.1897 Moderator Variable Present Ser 2 = 0.0078 Z = 1.9149 p = 0.0281, Significant, but not relevant due to the presence of moderator variable

PAGE 59

52 LIST OF REFERENCES Alexander RD, Hoogland JL, Howard RD, N oonan KM, and Sherman PW (1979) Sexual dimorphisms and breeding systems in pi nnipeds, ungulates, prim ates, and humans. In NA Chagnon and W Irons, (eds.): Evol utionary biology and human social behavior. North Scituate, MA.: Duxbury Press. p 402-435. Angel JL (1984) Health as cr ucial factor in the changes from hunting to developed farming in eastern Mediterranean. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agricu lture. London: Academic Press. p 50-73. Armelagos GJ (1990) Health and disease in prehistoric populations in transition. In GJ Armelagos (ed.): Diseases in populations in transition. New York: Bergin and Garvey. Armelagos GJ, and Van Gerven DP (1980) Sexu al dimorphism and hu man evolution: an overview. Journal of Human Evolution 9 :437-446. Borgognini Tarli SM, and Re petto E (1997) Sex differen ces in human populations: change through time. In ME Morbeck, A Galloway and AL Zihlman, (eds.): Evolving female. Princeton, NJ: Prin ceton University Press. p 198-208. Boserup E (1965) The conditions of agricu ltural growth: the economies of agrarian change under population pre ssure. Chicago: Aldine. Boyd DC, and Boyd CC (1989) A comparison on Tennessee archaic and mississippian maximum femoral lengths and midshaft diameters: subsistence change and postcranial variability. Southeastern Archaeology 8 :107-116. Brace CL (1963) Structural reduction in evolution. The American Naturalist 97 :39-49. Brace CL (1973) Sexual dimorphism in human evolution. In CL Brace and J Metress, (eds.): Man in evolutionary perspective. New York: John Wiley & Sons, Inc. p 238-254. Brace CL, and Ryan AS (1980) Sexual dimo rphism and human tooth size differences. Journal of Human Evolution 9 :417-435. Bramblett CA (1994) Patterns of primate behavior, 2nd ed. Prospect Heights, IL: Waveland Press, Inc.

PAGE 60

53 Brauer GW (1982) Size sexual dimorphism and secular trend: indicators of subclinal malnutrition? In RL Hall, (ed.): Sexual dimorphism in Homo Sapiens : a question of size. New York: Praeger Publishers. p 245-262. Bridges PS (1989) Changes in activ ities with the shif t to agriculture in the southeastern United States. Current Anthropology 30 :385-394. Bridges PS (2000) Changes in long bone di aphyseal strength with horticultural intensification in west-central Illinois. American Journal of Physical Anthropology 112 :217-238. Brock SL, and Ruff CB (1988) Di achronic patterns of change in structural properties of the femur in the prehistoric American s outhwest. American Journal of Physical Anthropology 75 :113-127. Bronson B (1977) The earliest farming: dem ography as cause and consequence. In CA Reed, (ed.): Origins of agricultu re. Paris: Mouton Publishers. Brues A (1959) The spearman and the archer an essay on selection in body build. American Anthropologist 61 :457-469. Buikstra JE (1984) The lower Illinois river region: a prehis toric context for the study of ancient diet and health. In MN Cohen a nd GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. London: Academic Press. p 215-234. Campbell BG, and Loy JD (2000) Humankind emerging. Needham Heights, MA: Allyn & Bacon. Clark GA (1988) New method for assessing changes in growth and sexual dimorphism in paleoepidemiology. American Jour nal of Physical Anthropology 77 :105-116. Cohen MN (1977) Population pressure and the origins of agriculture : an archaeological example from the coast of Peru. In CA R eed, (ed.): Origins of agriculture. Paris: Mouton Publishers. p 135-177. Cohen MN, and Armelagos GJ (1984) Paleopa thology at the origin s of agriculture: editor's summation. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. New York: Academic Press, Inc. p 585-601. DeVore I, and Washburn SL (1963) Baboon eco logy and human evolution. In FC Howell and F Bouliere, (eds.): African ecology and human evolution. New York: Viking Fund Publication No. 36. p 335-367. Eveleth PB (1975) Differences between ethnic groups in sex dimorphism of adult height. Annals of Human Biology 2 :35-39.

PAGE 61

54 Finkel DJ (1982) Sexual dimorphism and settle ment pattern in Middl e Eastern skeletal populations. In RL Hall, (ed.) : Sexual Dimorphism in Homo sapiens New York: Praeger Publishers. p 165-185. Frayer DW (1980) Sexual dimorphism and cultu ral evolution in the late Pleistocene and Holocene of Europe. Journal of Human Evolution 9 :399-415. Frayer DW (1981) Body size, weapon use, a nd natural selection in the European Upper Paleolithic and Mesolithic. American Anthropologist 83 :57-73. Frayer DW, and Wolpoff MH (1985) Se xual dimorphism. Annual Review of Anthropology 14 :429-473. Gaulin SJC, and Boster JS (1985) Cross-cult ural differences in sexual dimorphism: is there any difference to be expl ained? Ethology and Sociobiology 6 :219-225. Gaulin SJC, and Boster JS (1992) Human marriage systems and sexual dimorphism in stature. American Journal of Physical Anthropology 89 :467-475. Glass GV (1978) In Defense of Generaliza tion. The Behavioral and Brain Sciences 3 :394-395. Glass GV, McGaw B, and Smith ML (1981) Me ta-analysis in social research. London: Sage Publications. Gray JP, and Wolfe LD (1980) Height and sexual dimorphism of stature among human societies. American Journal of Physical Anthropology 53 :441-456. Gustafson A, and Lindenfors P (2004) Human si ze evolution: no evolutionary allometric relationship between male and female stature. Journal of Human Evolution 47 :253266. Hall RL (1982) Unit of analysis. In RL Hall (ed.): Sexual dimorphism in Homo sapiens New York: Praeger Publishers. p 189-196. Hinton RJ, and Carlson DS ( 1979) Temporal change in human temporomandibular joint size and shape. American Jour nal of Physical Anthropology 50 :325-334. Holden C, and Mace R (1999) Sexual dimor phism in stature and women's work: a phylogenetic cross-cultural analysis. American Journa l of Physical Anthropology 110 :1: 27-45. Hunter JE, Schmidt FL, and Jackson GB ( 1982) Meta-Analysis: Cumulating Research Findings across Studies. L ondon: Sage Publications. Kennedy KAR, Deraniyagala SU, Roertgen JC, Chiment J, and Disotell T (1987) Upper Pleistocene fossil hominids form Sri La nka. American Journal of Physical Anthropology 72 :441-461.

PAGE 62

55 Krantz GS (1982) The fossil record of sex. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens New York: Praeger Publishers. p 85-105. Larsen CS (1984) Health and disease in prehis toric Georgia: the transition to agriculture. In MN Cohen and GJ Armelagos, (eds .): Paleopathology at the origins of agriculture. London: Acad emic Press. p 367-392. Larsen CS (1995) Biological changes in human populations with agriculture. Annual Review of Anthropology 24 :185-213. Lazenby RA (2002) Population variation in second metacarpal sexual size dimorphism. American Journal of Physical Anthropology 118 :378-384. Lieberman LS (1982) Normal and abnormal se xual dimorphic patterns of growth and development. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens : a question of size. New York: Praeger Publishers. p 263-316. Martin DL, Armelagos GJ, Goodman AH, and Van Gerven DP (1984) The effects of socioeconomic change in prehistoric Af rica: Sudanese Nubia as a case study. In MN Cohen and GJ Armelagos, (eds.): Pale opathology at the orig ins of agriculture. London: Academic Press. Meiklejohn C, Schentag C, Venema A, a nd Key P (1984) Socioeconomic change and patterns of pathology and variation in th e Mesolithic and Neolithic of Western Europe: some suggestions. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the Origins of Agriculture. London: Academic Press. p 75-100. Morbeck ME (1997) Evolving Female. Princet on, NJ: Princeton University Press. Murdock GP, and Provost C (1973) Factors in the division of labor by sex: a cross cultural analys is. Ethnology 12 :203. Nickens PR (1976) Stature reduction as an adaptive response to food production in Mesoamerica. Journal of Archaeological Science 3 :31-41. Ortner DJ (2003) Identification of Pathologi cal Conditions in Human Skeletal Remains. New York: Academic Press. Ott LH, and Longnecker M (2001) An introdu ction to statistical methods and data analysis, 5th ed. Pacific Grove, CA: Duxbury. Perzigan AJ, Tench PA, and Braun DJ (1984) Pr ehistoric health in the Ohio River valley. In MN Cohen and GJ Armelagos, (eds .): Paleopathology at the origins of agriculture. New York: Academ ic Press, Inc. p 347-366. Rathbun TA (1984) Skeletal pathology from the Paleolithic through the Metal Ages in Iran and Iraq. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. London: Academic Press.

PAGE 63

56 Rensch B (1959) Evolution above the species level. London: Methuen & Co. Rivers J (1982) Women and children last: an essay on sex discrimination in disasters. Disasters 6 :256-267. Rose JC, Burnett BA, and Blaeuer MW (1984) Paleopathology and the origins of maize agriculture in the lower mississippi va lley and Caddoan culture areas. In MN Cohen and GJ Armelagos, (eds.): Paleopathol ogy at the origins of agriculture. New York: Academic Press, Inc. p 393-424. Rosenthal R (1984) Meta-analytical procedures for social research. Beverly Hills, CA: Sage Publications. Rowe N (1996) The pictorial guide to livi ng primates. Charlestown, RI: Pogonias Press. Ruff CB (1987) Sexual dimorphism in human lower limb bone structure: relationship to subsistence strategy and sexual division of labor. Journal of Human Evolution 16 :391-416. Ruff CB (1991) Aging and osteoporosis in Native Americans from Pecos Pueblo, New Mexico. New York: Garland Publishing, Inc. Ruff CB, and Larsen CS (1990) Postcranial biomechanical adaptations to subsistence strategy changes on the Georgia coast. Anth ropological Papers of the Museum of Natural History 68 :94-120. Ruff CB, Larsen CS, and Hayes WC (1984) St ructural changes in the femur with the transition to agriculture on the Georgia coast. American Journal of Physical Anthropology 64 :125-136. Ruff CB, Trinkaus E, Walker A, Larsen CS (1993) Postcranial robusticity in Homo. I: temporal trends and mechanical interpretation. American Journal of Physical Anthropology 91 :21-53. Smith P, Bar-Yosef O, and Sillen A (1984) Archaeological and skeletal evidence for dietary change during the Late Pleistocen e/Early Holocene in the Levant. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. London: Academic Press. p 101-136. Smith RJ (1999) Statisics of sexual size dimorphism. Journal of Human Evolution 36 :423-459. Stini WA (1969) Nutritional stress and grow th: sex differences in adaptive response. American Journal of Physical Anthropology 31 :417-426. Stini WA (1975) Ecology and human adapta tion. Dubuque, IA: Wm C. Brown Company Publishers.

PAGE 64

57 Stini WA (1982) Sexual dimorphism and nutri ent reserves. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens : a question of size. New Yo rk: Praeger Publishers. p 391-420. Stinson S (1985) Sex differences in envi ronmental sensitivity during growth and development. Yearbook of Physical Anthropology 28 :123-147. Swanton J (1946) The Indians of the South eastern United States: Smithsonian Press. Trivers RL (1972) Parental i nvestment and sexual selection. In B Campbell, (ed.): Sexual selection and the descen t of man: 1871-1971. Chicago: Aldine Publishing Company p 136-179. Turner CG (1979) Dental anth ropological indications of ag riculture among Jomon people of central Japan. American Jour nal of Physical Anthropology 51 :619-636. Ubelaker DH (1984) Prehistoric human biology of Ecuador: possible te mporal trends and cultural correlations. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. New Yo rk: Academic Press, Inc. p 491-513. Wolanski N, and Kasprzak E (1976) Statur e as a measure of environmental change. Current Anthropology 17 :548-552. Wolfe LD, and Gray JP (1982a) A Cross-cu ltural investigatio n into the sexual dimorphism of stature. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens New York: Praeger Publishers. p 197-230. Wolfe LD, and Gray JP (1982b) Subsistence practices and human sexual dimorphism of stature. Journal of Human Evolution 11 :575-580. Wood W, and Christensen PN (2004) Quantita tive research synthesis: examining study outcomes over samples, settings, and tim e. In AT Panter, (ed.): Handbook of methods in social psychology. Thousand Oaks, CA: Sage Publications.

PAGE 65

58 BIOGRAPHICAL SKETCH Anna Elizabeth Vick was born in 1975, in Chapel Hill, North Carolina. She graduated from the University of North Caro lina at Chapel Hill in 1998 with a Bachelor of Arts degree in anthropology.


Permanent Link: http://ufdc.ufl.edu/UFE0009587/00001

Material Information

Title: Sexual Dimorphism and the Transition to Agriculture: A Meta-Analysis
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0009587:00001

Permanent Link: http://ufdc.ufl.edu/UFE0009587/00001

Material Information

Title: Sexual Dimorphism and the Transition to Agriculture: A Meta-Analysis
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0009587:00001


This item has the following downloads:


Full Text












SEXUAL DIMORPHISM AND THE TRANSITION TO AGRICULTURE:
A META-ANALYSIS














By

ANNA ELIZABETH VICK


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Anna Elizabeth Vick















ACKNOWLEDGMENTS

I would like to sincerely thank Dr. David Daegling and Dr. Ken Sassaman for

serving on my committee. They both offered advice and guidance throughout this

project; I am grateful for their mentorship. Furthermore, I owe a great deal of

appreciation to Dr. Russell Bernard and Dr. Larry Winner; they fielded numerous

questions regarding meta-analysis with great patience and thoroughness. I also would

like to thank Chad Maxwell for twenty-four hour technical support. Next, I would like to

express my gratitude to The University Women's Club. This generous organization

financially assisted me during my time at UF. Most importantly, however, I would like

to thank my family, Laura Greer Vick, Gilbert Vick, and Marguerite Hardee Greer, for

their infinite love, support and encouragement.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ................................................................................................. iii

LIST OF TABLES ..................................... .. .......... .................................... vi

ABSTRACT ................................................... ................. vii

INTRODUCTION ....................................... ........... ...............................

M E T H O D S ..................................................................................................... ....... .. 15

M eta-A n aly sis ............................................................................................................. 15
Choosing Data Sets ................... ............ ............................... 15
D ata A naly sis: M eta-analy sis ........................................................... ...................... 18
Data Analysis: Non-statistical Analysis of Data Not Appropriate for Meta-
A naly sis .................................................................................. ....................... 22

R E S U L T S ................................................................................................. ..................... 2 4

T he M eta-analy si s. ............... ................... ................ .............. ......... ... ............ 24
N on-Statistical A analysis .................................................................... ................ 28

D IS C U S S IO N .................................................................................................................. .. 3 0

A appropriate Indices of D im orphism ...................................................... ................ 32
T he T transition to A agriculture ....................................... ...................... ................ 34
L im stations of M eta-A analysis ....................................... ...................... ................ 40

CONCLUSIONS....................... ......... .. ................43

APPENDIX

A META-ANALYSES OF FEMORAL MEASUREMENTS ..................................... 44

B META-ANALYSES OF HUMERAL MEASUREMENTS ..................48

C META-ANALYSES OF OTHER LONG BONE MEASUREMENTS............... 51





iv










L IST O F R EFER EN CE S ...... .................................................................... ................ 52

BIO GR APH ICAL SK ETCH ...................... .............................................................. 58


























































v
















LIST OF TABLES

Table page

1 Sam ple Com position for M eta-analysis .............................................. ................ 18

2 Sample Composition for Non-statistical Analysis .............................................23

3 Sum m ary of M eta-A nalyses....................................... ...................... ................ 25

4 Non-Statistical Analysis of Changes in Sexual Dimorphism Ratios ....................29

5 Sexual Dimorphism Ratios in Transitional Groups ............................................39















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts

SEXUAL DIMORPHISM AND THE TRANSITION TO AGRICULTURE:
A META-ANALYSIS

By

Anna Elizabeth Vick

August 2005

Chair: David Daegling
Major Department: Anthropology

The degree of sexual dimorphism found in human populations has declined

throughout the history of anatomically modern humans. Researchers have specifically

suggested that the transition to agriculture led to a decline in sexual dimorphism due to

the reduction in gendered biomechanical loads and a nutritional decline. To test this

theory, data were compiled from studies which compared populations spanning the

transition to agriculture. Meta-analyses were then conducted using the available

postcranial measurements of preagricultural and agricultural groups to observe any

changes in sexual dimorphism. Of the five North American populations compared, only

10.5% showed a significant decline in sexual dimorphism. A non-statistical analysis of

the literature did not demonstrate any noteworthy difference in the degree of sexual

dimorphism in populations before and after the adoption of agriculture. In contrast to the

theory that sexual dimorphism declines with the transition to agriculture, in most cases,

no significant change occurs.















INTRODUCTION

In modern human groups, the average male is almost 1.1 times as tall as the

average female, and, according to Krantz (1982: 86), they are "correspondingly more

massive." Compared to extant ape populations, the degree of sexual dimorphism in

humans represents a moderate degree of dimorphism. Gorillas and orangutans have an

enormous degree of sexual dimorphism where females may only weigh half as much as

their male counterparts. Gibbons and siamangs exhibit very little sexual dimorphism in

body size; females are 94% of the size of males. Humans and chimpanzees are similar in

degree of dimorphism, with females at 78% and 81% of male body size, respectively

(Campbell and Loy, 2000).

Within modern human groups, it has been found that the degree of sexual

dimorphism varies by population. Eveleth (1975) measured individuals of African,

Amerindian, and European origin and found that in the measurement of adult stature,

Africans were the least sexually dimorphic population. Amerindians were the most

sexually dimorphic population, with Europeans ranking in between Africans and

Amerindians. However, it is important to remember that stature is just one measurement

of dimorphism. In my own investigations (Vick, unpublished), I found that Amerindians

were highly dimorphic in the measurement of maximum femur length (commonly used to

calculate stature), but did not express greater dimorphism in other osteometric

measurements such as femoral circumference or humeral length.









In contrast to Eveleth, Gaulin and Boster (1985) have suggested that the degree of

sexual dimorphism found in human populations is actually quite consistent and that

evidence to the contrary may simply be the product of small sample size. Nevertheless,

these researchers, along with numerous others, have continued to conduct studies which

attempt to explain sexual dimorphism based on cultural differences in human

populations. Marriage patterns, work load, parental investment, division of labor and

subsistence type are just a few of the cultural variables tested for their association with

sexual dimorphism (Gaulin and Boster, 1992; Holden and Mace, 1999; Ruff, 1987;

Wolfe and Gray, 1982a; Wolfe and Gray, 1982b)

Sexual dimorphism in humans has varied over time. Over the course of human

evolution, from the australopithecines to anatomically modern humans, the degree of

sexual dimorphism has declined (Frayer and Wolpoff, 1985; Krantz, 1982). Likewise,

within Homo sapiens sapiens, there has been a subsequent decline in sexual dimorphism

from the Upper Paleolithic to the present (Borgognini Tarli and Repetto, 1997; Brace,

1973; Brace and Ryan, 1980; Frayer, 1980; Frayer, 1981; Frayer and Wolpoff, 1985;

Meiklejohn et al., 1984).

The decreases in dimorphism seen in Homo sapiens sapiens have generally been

associated with changes in subsistence or technology, most notably from the Upper

Paleolithic to the Mesolithic. During the Upper Paleolithic, humans were big game

hunters. Then, as the Upper Paleolithic ended with glacial retreat and the large scale

extinction of many big game species, the Mesolithic hunter had to adapt by hunting

smaller species, like pigs and deer, rather than, for example, mammoths. The prevailing

theory to explain this transition is that as the game became smaller, so too did the hunter









(Brace and Ryan, 1980; Frayer, 1980; Frayer, 1981). Brace and Ryan (1980) argue that

an increase in male body size occurred during the Pleistocene as an adaptive strategy for

hunting large game prior to technological sophistication. But, with technological

advances, the extinction of large game species, and the metabolic demands of a large

body, the selective pressure for large male body size decreased. It is believed that female

activity patterns did not change as greatly as male activity patterns during the Mesolithic

transition. As a result, mean female size did not change drastically, resulting in an

overall decline in sexual size dimorphism (Frayer, 1981). Brace and Ryan (1980) take

this theory to the next level by suggesting that the degree of sexual dimorphism in

modern populations is directly related to the amount of time that has passed since that

population's dependence on hunting large game. This theory is based on Brace's (1963)

theory of probable mutation effect which states that in the absence of selection, random

mutations will lead to the reduction of associated features. For the effects of the probable

mutation theory to be observable in modern populations, Brace and Ryan make several

assumptions. First, selection among big game hunters would need to have been constant

across populations. In addition, populations would have to remain reproductively

isolated. Otherwise, the differences in the progressive reduction in dimorphism due to

the probable mutation effect would not be discernible.

Researchers have also hypothesized that a further reduction in sexual size

dimorphism occurred as humans shifted from a hunting and gathering economy to

agriculture (Armelagos and Van Gerven, 1980; Boyd and Boyd, 1989; Frayer, 1980;

Frayer and Wolpoff, 1985; Hinton and Carlson, 1979; Holden and Mace, 1999; Kennedy

et al., 1987; Lazenby, 2002; Ruff, 1987; Wolfe and Gray, 1982b). In contrast to data









available for the Upper Paleolithic Mesolithic transition, which despite the small

sample size, strongly support a decrease in sexual dimorphism, data for a possible

decrease in sexual dimorphism after the shift to agriculture are more equivocal.

Archaeological, morphometric, and pathological data have been collected which

describes the transition to agriculture. While specific biological responses to the

adoption of agriculture vary regionally, there appears to have been an overall decline in

human health (Cohen and Armelagos, 1984). As human groups became more sedentary,

populations increased, as did rates of infectious diseases. People who raised livestock

were even more prone to diseases due to exposure to animal vectors (Ortner, 2003).

Chronic malnutrition also increased with agriculture. While groups were better able to

store excess food for times of shortage, the nutritional quality of the food declined. In

contrast, hunter-gathering populations had the advantage of mobility; if an area was no

longer productive due to drought or other circumstances, they could easily move.

Sedentism resulted in reduced evidence of musculoskeletal stress such as the incidence of

degenerative joint disease and overall decline in robusticity. However, a reduction in

physical stress may not be an indicator of positive change for several reasons. Cohen and

Armelagos (1984) point out that the higher frequency of degenerative joint disease in

hunter-gatherer populations may be complicated by the higher age at death among these

groups as contrasted with more sedentary populations. Therefore, reduction in the

frequency of degenerative joint disease may not be an indication of a decreased

workload. In addition, physical activity is associated with bone remodeling. Evidence

suggests that a low level of physical activity during a person's life may be associated

with a higher propensity for bone fractures in later life (Ruff, 1991). Overall, the









transition to agriculture was marked by a decline in both the quality and duration of life

(Cohen and Armelagos, 1984), characteristics which need to be considered when

evaluating why changes in sexual dimorphism would occur during this time period.

To understand how a change in the subsistence economy could affect the degree

of sexual dimorphism, it is first necessary to understand the causative factors affecting

sexual dimorphism. There is general consensus that genetics affect sexual dimorphism;

however, the nature or strength of this relationship is not understood. Eveleth (1975) is

frequently cited as a primary source supporting the genetic contribution to sexual

dimorphism. Eveleth's comparison of the sexual dimorphism of adult height in different

"ethnic groups" concluded that there is a difference in the level of sexual dimorphism

between these groups. While the sources of Eveleth's data are not given, the groups

compared were broadly defined as Europeans, Negroes, Amerindians, Asiatics, and New

Guineans. These "ethnic groups" are reminiscent of the major continental races and the

comparison is based on the assumption that these groups are genetically distinct from one

another. However, genetic variation does not observe socially constructed racial

designations. The vast majority of genetic variation occurs within racial groups rather

than between them.

Eveleth's (1975) study found that the degree sexual dimorphism found in each

ethnic or racial group did not meet the expectations of the nutritional hypothesis that the

most dimorphic populations should have the most nutritious diet. In consequence,

Eveleth concluded that there must be a genetic factor controlling the level of sexual

dimorphism in each population. To arrive at this conclusion, Eveleth only considered

two causal agents affecting dimorphism. There is no consideration of climate, gendered









access to resources, or any other factors associated with human morphology or sexual

dimorphism. While there are many criticisms of Eveleth's studies, other studies

comparing the amount of dimorphism in different populations have found similar results

(Holden and Mace, 1999). We are all aware that the size of an individual is highly

related to the size of his or her parents, but this does little to explain the ultimate cause of

sexual dimorphism.

In the animal kingdom there is a relationship between overall body size and the

degree of sexual dimorphism found in a species (Frayer and Wolpoff, 1985). According

to Rensch's Rule (Rensch, 1959), sexual dimorphism increases with body size in taxa

where males are the larger sex (Gustafson and Lindenfors, 2004). While Rensch's rule is

widely accepted, studies suggest that this relationship is less apparent in primates than in

other taxa (Frayer and Wolpoff, 1985). When size variation is compared by a method

which includes phylogenetic data, the results are even less clear (Gustafson and

Lindenfors, 2004). Gustafson and Lindenfors (2004) conducted a study of sexual

dimorphism in human populations where data on mean male and female height were

compared to genetic phylogenies. The results indicate that both male and female stature

is associated with phylogeny and that there is no evidence for an allometric relationship

between male and female stature in human populations (Gustafson and Lindenfors,

2004). This study serves as additional support that the genetic component to dimorphism

needs to be considered in cross-cultural studies and that sexual dimorphism is not simply

a byproduct of overall size.

The theory of sexual selection, more specifically, intrasexual selection, is

commonly used to explain how sexual dimorphism develops in non-human animals









including primates. Intrasexual selection is based on the idea that competition exists

within the members of one sex for reproductive access to the other sex. The sex which is

competed for is generally the one that has a higher energy investment in the success of

the offspring. In the case of humans, as well as in most other mammals, females make

the greater investment in young; therefore, there is male-male competition for access to

females (Trivers, 1972). While the relationship of size to reproductive success is difficult

to measure, larger size may be linked to dominance by offering an advantage in

aggressive encounters, and some studies demonstrate a positive relationship between

dominance and reproductive success. As a result, larger males are able to pass their

genes to subsequent generations more effectively, thereby increasing the degree of sexual

dimorphism in size (Trivers, 1972). Based on theories of sexual selection, sexual

dimorphism should be greatest in populations where there is the most competition for

access to females.

Because paternity is often difficult to determine, mating or marriage practices have

been used to test the sexual selection hypothesis. Alexander et al. (1979) conducted a

study to determine whether or not breeding systems were correlated with sexual

dimorphism in a variety of species, including humans. In all nonhuman groups it was

found that the degree of polygyny (measured as a deviation from monogamy by harem

size) was positively correlated with sexual dimorphism in body size. Primate species,

such as hamadryas baboons, which are characterized by single-male, multi-female

groups, have a high degree of sexual dimorphism in body size. Conversely, the

Callitrichidae, characterized by monogamy and polyandry, have a very small degree of

sexual dimorphism, with females larger than males in some cases (Rowe, 1996).









Alexander et al. (1979) also found a correlation between breeding system and

sexual dimorphism in human populations, but their methods and results have been

challenged by Gray and Wolfe (1980) who found no such correlation in their own

research. In the studies of Alexander et al. (1979) and Gray and Wolfe (1980), all

monogamous societies were divided into two groups: those with socially imposed

monogamy, and those with ecologically imposed monogamy. Alexander et al. (1979)

argue that socially imposed monogamy should be included with polygyny and that only

ecologically imposed monogamy can be expected to follow the hypothesized pattern of

sexual dimorphism based on theories of sexual selection. Unfortunately the methodology

utilized by Alexander et al. (1979) of estimating heights by sight, leaves their results

subject to speculation. When Gray and Wolfe (1980) reanalyzed the data, they found no

significant correlation between sexual dimorphism and mating pattern, but they did find

less variability in male and female height in polygynous societies, perhaps suggesting

they are under greater selective pressure for height or some correlate of height.

According to the theory of sexual selection, there is greater variance in

reproductive success for individual males in polygynous societies as opposed to

monogamous ones. In any study of mating or marriage it is understood that true rates of

paternity are not always easy to determine. However, marriage systems provide a readily

available variable to study how sexual selection affects sexual dimorphism in humans

under the assumption that a husband is likely to be the father of a woman's offspring.

Sexual dimorphism can be influenced by circumstances of the environment. A

number of studies have shown that sexual dimorphism in stature can decrease when

people are under nutritional stress and increase under conditions of optimal nutrition









(Brauer, 1982; Gray and Wolfe, 1980; Lieberman, 1982; Stini, 1969; Stini, 1982;

Wolanski and Kasprzak, 1976). The theoretical basis for this is found in the fact that

males and females experience differential success in dealing with stressors like starvation

and disease due to hormonal and metabolic differences (Ortner, 2003; Stini, 1969). The

greater fat and nutrient reserves characteristic of human females are thought to be an

adaptation for the increased metabolic demands of lactation and gestation in producing

offspring. As a result of these physiological differences, males experience a greater

reduction of lean body mass during periods of nutritional inadequacy than their female

counterparts. When periods of starvation occur during growth, the reduction in body

mass is accompanied by reduced skeletal growth (Stini, 1975).

In Stini's (1975) analysis of the transition to hunting and gathering, he states that

"severe nutritional imbalances...are much more common in agricultural areas than

among hunters and gatherers although starvation is no stranger to hunting populations in

most parts of the world" (p. 64). The greater prevalence of these nutritional shortages in

agricultural societies forms the basis for the theory that sexual dimorphism declined with

the transition to agriculture. If nutritional deficiencies lead to a decline in dimorphism,

this shift should primarily be due to a reduction in male size. However, while it is

apparent that nutrition affects sexual dimorphism to some degree (Brauer, 1982; Hall,

1982), many researchers think that it is not the leading factor affecting sexual

dimorphism on an evolutionary scale (Eveleth, 1975; Larsen, 1984; Stini, 1969).

Related to the issue of sexual selection are theories of parental investment. The

theory of sexual selection suggests that the pressure for sexual selection is decreased

when both male and female invest in the rearing of their offspring. In most primate









species, the female invests much more energy in raising the next generation than does the

male. In contrast, tamarins and marmosets are most noted for male parental investment.

It is in these species that the least degree of sexual dimorphism occurs (Rowe, 1996).

Studies suggest that parental investment favors the sex where the most variance in

reproductive success occurs (Morbeck, 1997). Because the variance in reproductive

success is generally greater for males, investing in young males can greatly increase a

parent's fitness.

Cultural studies of human groups also show that parental investment varies based

on the gender of the child. Holden and Mace (1999) found that sexual dimorphism in

stature is negatively correlated with the amount of work women perform. Likewise,

female juvenile mortality rates are higher than those for juvenile males in areas where

females contribute less to subsistence. These patterns follow geographical patterns of

sexual dimorphism (Holden and Mace, 1999). Rivers (1982) studied survival rates

under conditions of famine and disaster to determine whether proof could be found for

theories of differential survival based on sex. It was found that sex discrimination and

preferential treatment of male children confounded the results of the study. While

females may have a natural advantage under times of stress, males often receive cultural

advantages that may more than make up for any differential survival rates. Similar

conditions may affect the degree of sexual dimorphism found in a population.

Biologists may also view sexual dimorphism as a product of optimal biomass

distribution for the species. When conditions select for large males, it is advantageous

for the female of the species to be as much smaller as possible while still being able to

achieve reproductive success (Bramblett, 1994). The optimum female size is large









enough to bear the physical demands of labor, but small enough to reduce the metabolic

demands associated with large size. By considering sexual dimorphism as an optimal

biomass distribution, DeVore and Washburn (1963) are proposing that males and females

may be better able to utilize their resources if they fill different ecological niches. If

niche divergence is amplified with an increase in sexual dimorphism, then the selective

pressures affecting males and females are progressively more different.

Ethnographic research (Murdock and Provost, 1973) suggests that there is greater

overlap between male and female subsistence activities in agricultural populations than is

found in hunting and gathering populations. In addition, biomechanical data suggest that

the forces affecting males and females are more similar in an agricultural population than

among hunter-gatherers. For example, Ruff (1987) found that when comparing cross-

sectional properties of bone, there was a decline in sexual dimorphism with the transition

to agriculture indicating a reduction in the division of labor associated with agricultural

tasks. Ruff (1987) did not code for nutritional changes in this analysis because he

believes that cross-sectional data is a reflection of mechanical environment as opposed to

dimensional variables which may be better indicators of nutrition. Holden and Mace

(1999) compared populations in the Ethnographic Atlas with regard to marriage

practices, subsistence and the division of labor. They "concluded that in contemporary

humans, neither hunting nor agriculture has any effect on sexual dimorphism. [Instead] It

is the amount of subsistence work done by men and women, rather than the type of

subsistence practiced, which has an effect on sexual dimorphism in different societies."

(p. 42). As women contribute more to the subsistence economy, it appears that the

degree of sexual dimorphism is reduced. Holden and Mace (1999) used stature as their









only measurement of dimorphism while Ruff (1987) used only the cross-sectional

properties of bone. Ruff compared individuals from what is believed to be a genetically

related population in which the transition to agriculture took place. Holden and Mace

compiled data from various populations around the globe, and, while hunter-gatherers

and agriculturalists are both represented, there is no genetic control which connects one

group to the other.

Sexual selection, parental investment, nutrition, environmental conditions,

marriage or mating patterns, body size, ecological niche, and the division of labor by sex

have all been used to explain sexual dimorphism in humans and other animal species.

The etiology of sexual dimorphism has proven itself complex and not easily understood.

All the same, sexual dimorphism is still one of the indices commonly included by

anthropologists to describe biological change in response to agriculture and in other

studies of human health.

From the theories presented above, we can see that there are two major forces

behind the theory that sexual dimorphism declines with the transition to agriculture.

First, it has been shown that rates of chronic malnutrition are higher in agricultural

populations than among hunter-gatherers (Cohen and Armelagos, 1984; Stini, 1975).

Based on studies of secular declines in sexual dimorphism among malnourished

populations (Brauer, 1982; Wolanski and Kasprzak, 1976), it is reasonable to assume the

same trend would apply to the transition to agriculture. Second, it is believed that the

biomechanical forces affecting males and females are more similar among agriculturalists

than hunter-gatherers due to a reduced division of labor among agriculturalists.









Not all of the data collected support this theory. Wolfe and Gray (1982a) compared

extant populations and found the opposite of what was expected; the agricultural

populations were more sexually dimorphic in stature than were the hunting and gathering

populations they sampled. Data from archaeological populations that have undergone the

transition to agriculture have found conflicting results (cf. Boyd and Boyd, 1989; Larsen,

1984; Martin et al., 1984; Ruff, 1987).

Most studies which compare the degree of sexual dimorphism in hunter-gatherers

versus agriculturalists have looked at extant unrelated populations (Holden and Mace,

1999; Wolfe and Gray, 1982a). These studies ignore possible genetic contributions to

sexual dimorphism or regional differences in the ways in which agriculture was adopted

or in the nutritional quality of the food sources cultivated.

Archaeological investigations into the transition to agriculture have unearthed a

number of sites containing human remains which temporally span the transition in

subsistence economy. These remains are often of presumably genetically related

individuals which eliminates having to account for the influence of genes on sexual

dimorphism. By using meta-analysis I have aggregated data from individual

archaeological sites which span the transition to agriculture in order to test the hypothesis

that there was a decrease in sexual dimorphism with the advent of agriculture.

By investigating the relationship between sexual dimorphism and the transition to

agriculture I am examining what dimorphism means in terms of human biology and the

role of cultural factors in this relationship. In archaeological contexts, inferences

regarding human behavior are often made based on the degree of dimorphism in a

population (Bridges, 1989; Bridges, 2000; Ruff, 1987). In studies of human health,






14


overall size and the degree of dimorphism are used as indices for the health of human

populations (Brauer, 1982; Stini, 1969; Stinson, 1985; Wolanski and Kasprzak, 1976). In

order to make these inferences or utilize these indices, it is first necessary to understand

the role of sexual dimorphism on an evolutionary scale.















METHODS

Meta-Analysis

Meta-analysis is a statistical technique that allows researchers to aggregate data

from multiple individual studies. When a number of studies address a single issue, the

results of said studies do not always agree and the call for more research is often repeated

in the literature. The recurring call for additional research has left many questioning the

validity of continuing to perform these individual studies and has created the need for

cumulative techniques for approaching data (Hunter et al., 1982; Rosenthal, 1984).

To explore the impact of the adoption of agriculture on human sexual dimorphism,

I first conducted a survey of the results of extant studies investigating the skeletal

consequences of this transition. Preliminary investigation of the results of these studies

shows that sexual dimorphism decreases in some areas, increases in others, and in some,

there appears to be no change. Any literature review can provide the information

necessary for a qualitative summary of the data, but a quantitative review is advantageous

in that it can be methodically scrutinized (Wood and Christensen, 2004). By aggregating

all the available data collected by various researchers into a single and more

comprehensive result, meta-analysis serves as a tool to evaluate existing theories (Wood

and Christensen, 2004).

Choosing Data Sets

The first step in any meta-analysis is to conduct a thorough literature review to find

all sources of data on the subject. The articles discussed in this project were found using









a variety of techniques, most importantly the ancestry approach. Rosenthal (1984)

describes the ancestry approach as a means of gathering new resources by using the

bibliographic information found in an initial article. For this study, Paleopathology at

the Origins ofAgriculture (Cohen and Armelagos, 1984) served as the primary

bibliographic source. I also conducted a number of searches using the internet resources

Web of Science and First Search. Keywords for these searches included combinations of

the following terms: sexual dimorphism, dimorphism, subsistence, agriculture, change,

temporal, femur, femoral, osteometric, morphology. While these searches did provide a

few relevant articles which could then be mined for bibliographic sources, they were

largely unsuccessful. Likewise, on-line searches of individual anthropological journals

were ineffective. The keywords mentioned above were used in on-line searches of the

following journals: American Journal of Physical Anthropology, Journal of Human

Evolution, N.v',m/be'iel ,n Archaeology, American Journal of Archaeology, Journal of

Anthropological Archaeology, International Journal of Osteoarchaeology, Current

Anthropology, and Yearbook of Physical Anthropology.

The majority of relevant data were not found in studies of sexual dimorphism, but

rather, in studies which investigated within sex variation over time. The purpose of the

original publication did not affect the outcome of my study except in cases where P

values or test statistics were provided in lieu of complete data. The statistical values

provided by the original authors could not be used because different questions were being

asked of the data.

The primary stipulations for data inclusion in this project were twofold: 1) studies

had to present data from both before and after the adoption of agriculture and 2) studies









had to demonstrate geographic continuity. Prior studies of temporal changes in sexual

dimorphism have not always limited their samples geographically, sometimes taking a

preagricultural sample from one area and a postagricultural sample from another (Brace

and Ryan, 1980; Frayer, 1980). By limiting samples to those found in approximately the

same region, I am hoping to control for genetic factors i.e., that the pre- and post-

agricultural data are not known to be from genetically unrelated groups. While testing

populations within the same geographic area cannot always assure genetic continuity,

archaeological evidence can indicate whether the people in a region are from related

cultural groups or whether there is evidence of population migrations. The introduction

of new subsistence strategies or technologies may be suggestive of exchange with an

outside population. However, I only included populations where archaeological

information was consistent with genetic continuity. In addition to the issue of genetic

continuity, the plants cultivated in different areas of the world may have also had an

effect on the biological response to a subsistence shift since studies show that the effects

of agriculture were not uniform worldwide (Cohen and Armelagos, 1984). Geographic

continuity therefore helps control for this moderating factor.

Although review of the literature provided information on changes in sexual

dimorphism spanning the adoption of agriculture, the majority of these only present a

score for percent dimorphism or a difference in means. For the purposes of meta-

analysis, only those studies which provided sample sizes, means, and a measure of

dispersion (standard deviation or standard error) for any measurement variable were

included. A variety of measurement variables, both linear and biomechanical, were

included in the meta-analyses.









In any meta-analytical process it is assumed that all studies are independent.

Therefore, when multiple studies are conducted using the same data, it is important that

the meta-analyst does not consider any group of data more than once. This limited the

number of studies included in the meta-analysis to five, all of which were from areas

within the continental United States. In Bridges (1989) and Boyd and Boyd (1989),

Archaic and Mississippian populations were considered. Ruff and Larsen (1990) divided

the populations they studied into precontact preagricultural and precontact agricultural.

Contact populations were excluded from this analysis to control for disease and other

changes associated with European contact. In the studies of Brock and Ruff (1988) and

Table 1: Sample Comn position for Meta-analysis
Postagricultural
Source Location Preagricultural sample sample
Boyd and Boyd (1989) Tennessee Middle and Late Archaic Late Mississippian
Bridges PS (1989) Alabama Archaic Mississippian
Bridges PS (2000) West Central Illinois Middle Woodland Mississippian
Brock and Ruff (1988) New Mexico Early Villages Aggregated Villages
Ruff and Larsen (1990) Georgia Coast Precontact Preagricultural Precontact Agricultural


Bridges (2000), the earliest groups studied did have some horticultural supplementation

to their diet, but intensive agriculture did not occur until later. Table 1 lists the original

authors' descriptions of the pre-and post-agricultural samples used for this meta-analysis.

Populations that were considered "transitional" by the authors) were not considered in

this study.

Data Analysis: Meta-analysis

Deriving the t-statistic. For each measurement variable in each study a t-statistic

for change in dimorphism was calculated using the following formula:

t = (XPM-XPF) (XAM-XAF) / I (SEi+SE2+SE3+SE4) (1)









where P and A represent pre-agriculturalist and agriculturalist males and females,

respectively. Because probability is based on sample sizes which are not consistent

across the selected studies, a simple comparison of test statistics across studies could not

be performed.

Determining effect sizes. Meta-analyses rely on effect sizes to test the

homogeneity of populations before the individual test statistics can be compared or

combined. Effect sizes are calculated from the test statistic, but are measures of

magnitude rather than probability. There are several different types of effect size

indicators, the most common being Pearson's r, Cohen's d, and Glass' A. Cohen's d, and

Glass' A are both used to investigate the standardized differences between means of

independent groups (Source: http://web.uccs.edu/lbecker/Psy590/es/htm, last accessed

March 23rd, 2005). For these meta-analyses, Pearson's r was chosen as the effect size

indicator because the variables under investigation are not considered to be independent.

The effect size r was computed for each measurement variable using the following

formula:

r = / (t2 / t2 + (Ni+N2+N3+N4-4)) (2)

where N's represent the sample size of each sex, before and after the adoption of

agriculture.

The sign of the t statistic should be preserved when converting to r. That is, if t is

negative, r should be too. A negative t statistic occurs when the direction of the result is

in the opposite direction of the hypothesis. In the case of this study, a negative result

indicates that sexual dimorphism increases rather than decreases. A t statistic of zero









indicates that there is no change in the degree of dimorphism over the time period

considered.

Rather than performing a single analysis of all measurement variables within each

study, a meta-analysis was performed on each set of variables, since using multiple data

from a single site in a meta-analysis would violate the assumption of independent

samples. For example, all the data available on the mid shaft circumference of the femur

were compiled, but these data were not combined with the data on femoral length.

For each measurement variable, the effect sizes were compared and a sample

weighted mean correlation was computed using the following equation:

R = I [Niri] / INi. (3)

The dispersion was described by the variance formula:

Sr2 = X [Ni (ri- r)2] / Ni. (4)

Tests of homogeneity. Effect size estimates were compared to determine whether

the studies were from a homogeneous population. Tests of homogeneity do not suggest

that every variable within a group is the same, but rather that there is no indication of an

outside factor, or moderator variable, influencing the relationship being studied.

Homogeneity is necessary when comparing or combining the probabilities of individual

studies. The following chi-square formula was used to test for homogeneity:

Xk-12 = (iN, / (1- r)2) Sr2 (5)

where k is equal to the number of studies used in the meta-analysis. The null hypothesis

was that the studies are homogeneous and a P value of 0.05 was required to reject the null

hypothesis.









An alternative method proposed by Hunter et al. (1982) was also performed to test

for the presence of a moderator variable. This method compares the error variance and

uncorrected variance. The error variance was calculated:

Ser2 = k (1- r 2)2 / jNi (6)

The error variance was then divided by the uncorrected variance:

Ser2/ Sr2. (7)

If Ser2/ Sr2 < 0.75, then a moderator variable is present. A heterogeneous population or

the presence of a moderator variable suggests that there is a factor outside those being

tested that may be influencing the results of the study. These tests do nothing to suggest

what that outside influence may be. When a moderator variable is present, the

probabilities of individual studies cannot be combined with reliable results (Source:

http://www.fammed.ouhsc.edu/tutor/metanal.htm, last accessed March 24th, 2005).

Combining probabilities. There are several methods of combining the

probabilities of individual studies, but the number of applicable methods is limited by the

small number of studies included in this analysis. The method of adding Z's is the most

appropriate under the circumstances of this study since it can be used regardless of the

size of the study. However, it is limited in that it assumes unit variance (Rosenthal,

1984). The Z statistic is based on the area under a standard normal probability curve

where the mean is zero and the standard deviation is one. A Z score is associated with

the probability that a given value is Z standard deviations away from the mean (Ott and

Longnecker, 2001). Adding the Z's first requires that we convert our t statistic to Zusing

the following equation:


Z = t (1 (t2 / 4df))









where dfequals the degrees of freedom. In this case, the degrees of freedom were equal

to the sample size of each study divided by four. The Z's are then added using the

following formula:

Z = 1Z /
where K is the number of studies included in the analysis. Probability was then

determined using the statistical tables provided by Ott and Longnecker (2001). Results

were determined significant at the 0.05 level.

Data Analysis: Non-statistical Analysis of Data Not Appropriate for Meta-Analysis

Not all of the information gathered regarding changes in morphology spanning

the transition to agriculture could be included in the meta-analysis due to insufficient

data. Several publications presented the mean measurements for males and females with

no measurement of dispersion. In the non-statistical analysis, these studies were

compared to those used in the meta-analysis. For each of these populations, dimorphism

ratios were produced using the following formula:

Ln XM- InXF (10)

where XM is the male mean and XF is the female mean. According to the laws of

logarithms, Ln XM- In XF is equal to Ln (XM/ XF), a measure of proportion. The

decision to use this method of calculating a score of dimorphism was based on Smith,

who demonstrated that compared to other methods for finding dimorphism ratios, this

method is believed to have fewer problematic mathematical features (1999). Ratios were

compared to determine whether there was an increase or decrease in sexual dimorphism

over time. In this analysis there was no attempt to determine the degree or significance

of the change. The non-statistical analysis was completed to determine whether trends

found in the meta-analysis were consistent over a larger population sample and










geographical area. All formulas in the meta-analysis and the comparison of ratios were

calculated using Microsoft Excel.

Table 2: Sample Composition for Non-statistical Analysis
Source Location Preagricultural sample Agricultural sample

Angel (1984) Eastern Mediterranean Mesolithic Late Neolithic

Boyd and Boyd (1989) Tennessee Middle and Late Archaic Late Mississippian

Bridges PS (1989) Alabama Archaic Mississippian

Bridges PS (2000) West Central Illinois Middle Woodland Mississippian

Brock and Ruff (1988) New Mexico Early Villages Aggregated Villages

Clark (1988) Dickson Mounds, IL Pre-Mississippian Mississippian

Martin (1984) Lower Nubia Preagricultural Intensive Agricultural

Meiklejohn et al. (1984) Western Europe Mesolithic Neolithic

Perzigan et al. (1984) Ohio River Valley Late Archaic Ft. Ancient

Rathbun (1984) Iran & Iraq Preagricultural (Hotu) Neolithic

Rose (1984) Caddo Culture Area Fourche Maline Caddo II

Ruff and Larsen (1990) Georgia Coast Precontact Preagricultural Precontact Agricultural

Smith et al. (1984) Levant Natufian Neolithic and Chalcolithic

Ubelaker (1984) Ecuador Sta. Elena Ayalan and Guangala















RESULTS

The Meta-analysis

From the five studies selected for inclusion in the study, twenty-two individual

meta-analyses were conducted. An individual meta-analysis was performed for each

variable, the majority of which were taken on the femur and humerus. The linear

measurement variables included in meta-analyses were length and mid shaft

circumference. At midshaft for both the femur and the humerus, the following

biomechanical variables were included: cortical area, minimum bending strength,

maximum bending strength, shape index (Imax / Imin), and polar second moment of area

(J). Two studies included subtrochanteric biomechanical data which were included in the

meta-analyses. Meta-analyses of the tibia, ulna, and radius are limited to midshaft

circumference data as biomechanical variables were considered for these bones in only

one of the studies included in this analysis.

For an individual meta-analysis to be included in the final analysis, it was

necessary that the effect sizes be homogeneous. A comparison of the effect sizes by chi-

square showed that a moderator variable was present in three meta-analyses: femoral

cortical area, humeral mid shaft circumference and radial mid shaft circumference. In

these three cases, the presence of a moderator variable was confirmed using Hunter's

(1982) alternate approach. While tests of homogeneity indicate the presence of a

moderator variable, they do not provide information on the nature of this outside

influence. Since the source of the moderator variable could not be determined from the









Table 3: Summary of Meta-Analyses
Measurement Variables r Z P value
Femur
Length -0.0317 -1.7002 0.9554
Midshaft Circumference -0.0824 -1.4782 0.0929
Min Bending Strength (size standardized at midshaft) -0.0731 -0.9215 0.8212
Max Bending Strength (size standardized at midshaft) 0.0625 0.7756 0.2206
Femoral Shape Index (Imax/Imin) 0.0452 0.4558 0.3300
Femoral J (size standardized at midshaft) 0.0401 0.4034 0.3446
Subtrochanteric Cortical Area (size standardized) 0.1991 1.3859 0.0838
Subtrochanteric Min. Bending Strength (size stand.) 0.1408 1.2116 0.1131
Subtrochanteric Max. Bending Strength (size stand.) 0.1867 1.6103 0.0537 *
Subtrochanteric Femoral Shape Index (Imax/Imin) 0.1546 0.4167 0.3409
Subtrochanteric J (size standardized) 0.2010 1.7342 0.0418**
Humerus
Humerus Length 0.0054 -0.0941 0.5359
Humerus Cortical Area (size standardized at midshaft) 0.0507 0.4692 0.3228
Humerus Min. Bending Strength (size standardized) 0.1361 1.3011 0.0968
Humerus Max. Bending Strength (size standardized) 0.1635 1.8789 0.0307**
Humerus J (size standardized at midshaft) 0.1232 0.3216 0.3745
Humeral Shape Index (Imax/Imin) -0.1420 -0.8878 0.8106
Other
Tibia Midshaft Circumference 0.0178 0.1956 0.4247
Ulna Midshaft Circumference 0.0131 0.1747 0.4325
Table does not include meta-analyses in which a moderator variable was found to be
present through the chi-square test. For additional data, see appendix.
r = sample weighted mean effect size
Z = test statistic
** Significant (p< 0.05)
* Near Significant (0.1 > p > 0.05)









data, these three meta-analyses were removed from further study. In the meta-analysis of

femur length, a chi-square test suggested the population was homogeneous but Hunter's

(1982) alternative approach detected the presence of a moderator variable. Although

Hunter's approach detected the presence of a moderator variable, the result of 0.68 was

not far below the 0.75 threshold for significance. The chi-square test strongly suggested

the population to be homogeneous and as such this meta-analysis was not excluded from

the final study, leaving nineteen meta-analyses for comparison in the final analysis.

The femur. In the eleven independent meta-analyses of the femur, three (27%) of

the sample weighted mean correlations ( r ) were negative. A negative effect size

indicates that the degree of sexual dimorphism in the femur increased, rather than

decreased, with the shift to agriculture. Two of the three negative effect sizes were for

linear measurement variables: length and midshaft circumference. The third

measurement variable was a derived biomechanical variable.

Only one femoral measurement variable experienced a significant decline with the

transition to agriculture: the subtrochanteric polar second moment of area (J). J is found

by summing I values from two perpendicular measurements. When a section is perfectly

cylindrical, J is a measure of strength under torsional twisting. Although these sections

are likely elliptical, the cited authors have used this measure as an estimate of torsional or

twisting strength (Bridges, 2000; Brock and Ruff, 1988; Ruff and Larsen, 1990). A

second femoral measurement variable, subtrochanteric maximum bending strength, had a

near significant result ofp = 0.0537. Both of these measurement variables are

biomechanical and provide information on skeletal strength and robusticity.









The humerus. Of the six meta-analyses performed on the humerus, only the

humeral shape index (Imax/Imin) had a negative sample weighted mean correlation. All

other results were positive suggesting that a slight decline occurs more often than an

increase. In only one meta-analysis was the change in sexual dimorphism significant.

The degree of sexual dimorphism found in the maximum bending strength of the

humerus is significantly less in agricultural than in preagricultural populations.

Combined meta-analyses. The majority of meta-analyses show no significant

change in the degree of sexual dimorphism between preagricultural and agricultural

populations. Two measurement variables (10.5% of the total) had a significant decline in

the amount of sexual dimorphism and a third measurement variable had a near significant

decline. These three measurement variables, humeral maximum bending strength,

subtrochanteric maximum bending strength, and subtrochanteric J, are all biomechanical

rather than linear variables, indicating that sexual differences in strength experienced a

decline.

A decline in dimorphism indicates that either the male mean is decreasing or the

female mean is increasing. The original data from the significant and near-significant

meta-analyses were examined to determine whether or not the within sex changes which

occurred with the adoption of agriculture were largely male or female. The

preagricultural and agricultural mean measurements for males and females were

subtracted from one another to determine whether the greatest temporal differences could

be attributed to the males or females. The results of this investigation were inconclusive.

In some cases the males experienced the greater change and in other cases females









exhibited a greater change. Therefore, it does not appear that significant changes in

dimorphism can be attributed to one sex across studies.

Non-Statistical Analysis

A comparison of the sexual dimorphism ratios (Ln XM- In XF) for all studies,

including those used in the meta-analysis, showed that sexual dimorphism decreased in

54% of cases, only slightly more than half. Measurements of the femur were less likely

to decrease in dimorphism than were measurements of the humerus. While this

comparison has no statistical significance, it mirrors what is found through the meta-

analysis. Overall, there does not appear to be a noteworthy difference in the degree of

sexual dimorphism found in populations before and after the adoption of agricultural

subsistence. However, there is a slight trend toward a decline in dimorphism for all

bones with the exception of the femur. Similar trends appear when linear and

biomechanical data are compared.

To determine whether the data available provided information on geographical

differences in patterns of sexual dimorphism, studies were divided on the basis of

geographical location into New World and Old World samples. However, of the total

207 data points, only five were from Old World populations. Of these, three showed an

increase in sexual dimorphism ratios and two exhibited a decrease. Four of the

measurement variables were measures of stature (without indication as to which long

bones were used for the calculation) and one measurement variable was for maximum

femur length. Due to insufficient data from Old World sites, it was impossible to

compare Old and New World populations. However, the limited Old World data appear

to reflect the trend found throughout the total data set.









Table 4: Non-Statistical Analysis of Changes in Sexual Dimorphism Ratios
Increase Decrease

Total (207) 93 44% 113 54%

Femur (106) 62 58% 44 42%

Humerus (56) 21 37.5% 21 62.5%

Radius (12) 1 8% 11 92%

Tibia (14) 1 7% 13 93%

Ulna (10) 2 20% 8 80%

Linear Variables (74) 28 38% 46 62%

Femur (27) 19 70% 8 30%

Humerus (10) 2 20% 8 80%

Biomechanical Variables (115) 58 50% 57 50%

Femur (72) 41 57% 31 43%

Humerus (43) 17 40% 26 60%















DISCUSSION

Theoretically there are two reasons why a decline in sexual dimorphism would

accompany a transition to agriculture. First, cultural research has shown that the type of

work performed by males and females in agricultural cultures is more similar than the

gendered duties of the hunter-gatherer. Murdock and Provost (1973) conducted a cross-

cultural study of work in which they coded all tasks performed by each sex and whether

the activities were assigned to males or females, partially or completely. There was a

greater delineation between male and female duties in hunting and gathering populations

than among agriculturalists. Theoretically sexual dimorphism decreased as male and

female mechanical loads became more similar with the transition to agriculture. In

addition, nutrition generally declined with the transition to agriculture. Nutritional

decline occurred due to several factors. First, the amount of protein in the diet declined.

Agricultural sedentism meant that an area around a village could easily be over hunted.

Second, for the mobile hunter-gatherer there is greater variation in the types of plant food

collected. The few agricultural crops did not always provide the nutritional value

associated with variation, particularly in the case of maize (Cohen and Armelagos, 1984;

Larsen, 1995). Under conditions of nutritional stress, males are less likely than females

to reach their full potential size, thereby reducing the degree of sexual dimorphism (Stini,

1969).

If the degree of sexual dimorphism in a population does not decline with the

transition to agriculture, it may be explained in several ways. First, the changes in









nutritional value may not have been great enough to effect a change in the morphology of

a population, or, perhaps, changes in morphology were short-lived and difficult to discern

in the archaeological record. Other cultural variables, such as preferential treatment of

male children, may obscure any nutritional changes which may have affected dimorphism

(Ortner, 2003; Rivers, 1982). In reference to the biomechanical argument, changes in

loading may not have been of the type or intensity to change sexual dimorphism. Finally,

males and females may have both changed their behaviors in such a way that the level of

dimorphism remained the same with the adoption of agricultural subsistence.

This meta-analytical survey found that the decline in sexual dimorphism with the

shift to agriculture was significant in two of nineteen measurement variables (10.5% of

the total). While four measurement variables showed an increase in dimorphism over

time (21% of the total), none of these results were significant. The meta-analyses

therefore suggest that there is a slight trend toward a reduction in dimorphism, but,

overall, the temporal changes in sexual dimorphism with the change in subsistence are

not great. It is important to remember that the meta-analyses conducted in this study

addressed changes in sexual dimorphism rather than within sex changes in morphology.

Many of the individual studies compared did find significant changes in morphology with

the transition to agriculture that were not addressed by the meta-analyses conducted here.

At the Upper Paleolithic Mesolithic transition, the changes in morphology which

accounted for a decline in sexual dimorphism were mostly due to the gracilization of the

male which most researchers associate with a reduction in selection for large size due to

the advancement of hunting technologies and smaller game species. Female size (and

theoretically, activity pattern) did not undergo any major changes at that time. Frayer









(1980) found that the reduction in dimorphism which occurred at the dawn of the

Neolithic was more closely associated with changes in the female form rather than the

male. While he does not offer an explanation for these results it is reasonable to assume

that changes in the female form may be particularly pronounced in populations where

females were the primary agriculturalists. Ruff et al. (1984) found that of the biological

changes associated with the transition to agriculture (increased periosteal infections,

increased frequency of dental caries, and decreased stature and robusticity) the majority

were more prevalent in females than males. In the case of the Georgia coastal

populations these changes are thought to result from heavier female involvement in

agricultural activities and greater female consumption of corn (Ruff, 1987). Analysis of

the combined data used in this study could not attribute changes in dimorphism to either

sex. Changes in individual studies may be largely male or female, but there is no

consistent pattern when studies are compared.

Appropriate Indices of Dimorphism

The data collected for this review consisted of twenty-two different osteological

measurement variables from which sexual dimorphism could be calculated. A review of

the literature provides many other measurement variables of dimorphism which have

been used to compare populations. This presents the researcher with two important

questions: 1) is there an appropriate index of dimorphism, and 2) does it depend on the

questions which are being asked of the data? A number of studies have used stature, or

the sexual dimorphism of stature as an indicator of overall health in a population (Brauer,

1982; Holden and Mace, 1999; Wolanski and Kasprzak, 1976; Wolfe and Gray, 1982a).

Unlike many osteometric measurements, stature can be easily measured in extant

populations and compared to skeletal populations through a number of available formulae









for the estimation of stature. However, for skeletal populations, stature may not be the

most reliable indicator of dimorphism. First, all formulas used to calculate dimorphism

assume some amount of error. Second, when available, the femur or tibia is most

commonly used to calculate stature due to the fact that the stature calculations for those

bones have a smaller error than other long bones. In the analyses conducted for this

study, the trend toward a decline in dimorphism was less evident in the femur than it was

in measurements of the humerus. Therefore, perhaps stature is not the most sensitive

measurement for explaining changes in sexual dimorphism over time.

Of the twelve meta-analyses conducted on the femur, three (25%) had a negative

mean effect size indicating that sexual dimorphism increased rather than declined. For

the humeral meta-analyses, only one of seven (14%) mean effect sizes was negative. The

twelve meta-analyses used to describe the femur were drawn from a total of 35 effect

sizes, one for each measurement variable considered in each study. Forty two percent of

the femoral effect sizes were negative compared to 17% for the humerus. Ruff et al.

(1993) evaluated temporal changes in postcranial robusticity and found that a decline in

femoral diaphyseal robusticity was consistent in humans from the early Pleistocene

through recent populations whereas trends in upper limb robusticity were more difficult

to decipher. While these results may encourage researchers to evaluate the femur when

studying temporal changes in robusticity, the same confidence should not be extended to

studies of temporal changes in sexual dimorphism.

Measurements of sexual dimorphism on archaeological samples are limited to

adults due to the difficulty of sexing juvenile remains. However, studies of extant

populations have proven that problems exist in interpreting the remains of adult sex









differences. While juvenile populations may exhibit differential effects of stress, many of

these effects may be corrected through "catch-up" growth before reaching adulthood

(Stini, 1975). Therefore, adult size may not the most sensitive indicator of the

differential effects of stress.

Ruff (1987; 1984) suggests that the different measurements of sexual dimorphism

may be evidence for particular causal factors. He suggests that cross-sectional data

reveal more information about the forces acting on bone and, therefore, activity patterns,

whereas changes in size or stature, linear data, are more likely to be due to nutritional

factors (Ruff et al., 1984). If this is indeed the case, it is important to note that the only

significant results in this study were for cross-sectional data.

The Transition to Agriculture

Theories regarding sexual dimorphism and the transition to agriculture are not

based solely on the dietary effects of cultigens, but rather on the changes that accompany

this subsistence change. Population growth accompanied the transition to agriculture

although theories differ as to how these two factors relate to one another. Some believe

that the growth in population forced people to adopt an agricultural economy; others

think that the surplus food and sedentary nature of agricultural economies allowed for

population expansion (Boserup, 1965; Bronson, 1977; Cohen, 1977).

With increased sedentism and population expansion, the rates of infectious diseases

also increased after the transition to agriculture. Sedentism is particularly problematic

under marginal environmental conditions since mobility can be beneficial in allowing a

person to flee the worst circumstances. In the studies compiled by Cohen and Armelagos

(1984) there appears to be an overall decline in the quality and length of life with the

transition to agriculture. It could be argued that morphological changes associated with









the transition to agriculture are due to population expansion rather than subsistence

change. However, these two changes are largely inseparable.

In some areas of the world there were changes that accompanied the transition to

agriculture that were not universal and need to be considered as possible moderating

factors to any morphological change. For example, Bridges (2000) points out that the

atlatl was being replaced by the bow as a hunting tool around the same time that the

transition to agriculture was occurring in some areas of North America. These two

changes did not occur simultaneously worldwide, but may account for some of the

morphological changes seen at the transition to agriculture. Brues' (1959) spearman-

archer hypothesis suggested that the mechanical needs of the two weapon types would

afford a selective advantage to different body types. While attempts to test this theory

have proven it highly suspect, it must be considered that any change to the mechanical

loading of bone which accompanied the transition to agriculture may prove to be a

moderating variable in studies of morphological change.

The adoption of agriculture has been associated with an increase in infectious

diseases due to increased sedentism and a population expansion. Population expansion

allowed viral diseases such as measles, mumps and small pox to be more easily

communicable than ever before. Malaria, cholera, blastomycosis, and scrub typhus are

examples of diseases associated with agriculture due to an increased exposure to

zoonoses when turning soil and exposure to contaminated water (Armelagos, 1990;

Ortner, 2003).

Similarly, new ideas and technologies, such as those associated with subsistence

changes, may be introduced through trade or migration. In many areas of the world, the









transition to agriculture was associated with a colonizing force. None of the populations

in this study were subject to the diseases introduced with European contact, but the

introduction of disease should be considered as a moderating variable in some areas.

Ruff and Larsen (1990) were able to compare precontact agricultural groups with

postcontact agricultural groups on the Georgia coast. On the Georgia coast it appears that

sexual dimorphism in femur length increased with the adoption of agriculture but the

trend then decreased after contact. The increase in dimorphism that marked the adoption

of agriculture along the Georgia coast has been associated with a culture in which the

negative effects of corn agriculture appear to have affected females much more so than

males. Throughout much of southeastern North America after the adoption of

agriculture, females were the farmers while the primary subsistence activity of men

remained hunting (Swanton, 1946). Contact with the Spanish brought diseases and a

mission system which regimented the lives of both males and females. This example

provides valuable information on the effect of culture on morphological changes

spanning the transition to agriculture.

In this study, the size of males and females were compared in agricultural and

preagricultural populations. However, the dividing line separating these two groups is

not always clear. Bronson (1977) argues that cultivation of plant foods began as early as

the Paleolithic. By selectively discarding waste from food plants in areas where they

wanted plants to grow, people began propagating plant species. In North America,

Native Americans cultivated local seed crops before maize (Zea mays) was introduced

from Mesoamerica. Exposure to agriculture may thus have occurred long before the full

adoption of an agricultural economy, and food crops may have been limited to a certain









segment of the population before they became dietary staples among the population at

large.

For the purposes of this study, a population had to have a dependency on at least

one domesticated carbohydrate crop to be considered agricultural. In Bridges (2000) the

Middle Woodland population described had small scale cultivation of food crops.

However, this population is considered "preagricultural" in this study because it is

believed that agricultural products and activities had not become a significant portion of

this population's diet and lifestyle.

Determining the presence of agriculture in an archaeological site can be based on

the identification of plant remains, associated material culture, or through skeletal

indicators of agricultural subsistence. It is important to remember that the lack of

botanical samples may be the result of poor preservation, or, as Rose et al. (1984) point

out, many sites were excavated prior to the adoption of flotation techniques for retrieving

paleobotanical remains. Conversely, the presence of plant domesticants does not

necessarily indicate that agricultural products were a dietary staple of the population at

large. Several skeletal pathologies are associated with agricultural populations, but an

increase in the frequency of carious lesions in the dentition is universal and as such they

are used as one indicator of the presence of agriculture (Larsen, 1984; Rose et al., 1984).

Turner (1979) compared global samples and found that the average frequency of teeth

affected by carious lesions in hunting and gathering groups is 1.72%; mixed hunting,

gathering and farming groups average 4.37%; and agriculturalists average 8.56%. These

rates vary regionally based on the food sources available. Agricultural diets are high in

carbohydrates which are the primary cause of dental caries. Chemical analysis of bone









may also be used as a method of deducing paleodiet, particularly in the Americas where

corn leaves such a clear chemical signature.

Many studies contain skeletal data from time periods that are considered

transitional with regard to agriculture. Martin et al. (1984) found that agricultural

intensification, rather than agricultural origins, were accountable for differential patterns

of biological response, and, as such, only those groups in which agriculture was fully

entrenched are used in this study. Nonetheless, transitional groups can provide

information about how changes occurred. Nickens' (1976) study of stature reduction

with the adoption of agriculture suggests that body size declined with the adoption of

agriculture, but then increased as humans adapted, perhaps learning to compensate for

some of the negative effects of agriculture. This suggests that some of the recorded

morphological changes observed with the transition to agriculture may represent short-

term rather than evolutionary changes. Table 5 provides sexual dimorphism ratios from a

study which includes transitional populations. Of particular interest are the cases where

Middle Woodland and Mississippian groups have very similar sexual dimorphism ratios,

while the intermediate populations differ greatly (for example, see femoral maximum

bending strength, humeral cortical area, and humeral minimum bending strength). If the

transitional groups were not included, it would appear there was little morphological

change with the transition to agriculture. If some morphological changes are only short-

term, they may be difficult to detect in the archaeological record, particularly when

comparing groups which may be at different stages in the transition to agriculture.










Table 5: Sexual Dimorphism Ratios in Transitional Groups (Bridges 2000)
Middle Early Late Late Late Mississippian
Woodland* Woodland Woodland
Femoral Cross-Sectional Properties
Cortical Area 0.1700 0.0541 0.0228 0.0675
Min Bending Strength 0.0055 -0.0334 0.0165 0.0847
Max Bending Strength 0.2877 0.0271 0.0990 0.2281
Torsional Strength 0.2665 0.0010 0.0615 0.1674
Shape Index (Imax / Imin) 0.0455 0.1008 0.0729 0.1542
Femoral linear data and indices
Bicondylar Length 0.0733 0.0864 0.0760 0.0760
AP diameter (midshaft) 0.1247 0.1520 0.1514 0.1456
ML diameter (midshaft) 0.0794 0.0551 0.0738 0.0666
Circumference (midshaft) 0.1158 0.1098 0.1088 0.1118
AP diameter (subtroch) 0.1238 0.1163 0.1223 0.1207
ML diameter (subtroch) 0.1007 0.0859 0.1076 0.0924
Circumference (subtroch) 0.1007 0.0949 0.1341 0.1037
Vert Head Diam Subtroch 0.1117 0.1051 0.1265 0.1394
Pilastric Shape Index 0.0461 0.0987 0.0741 0.0793
Platymeric Shape Index 0.0270 0.0270 0.0136 0.0270
Humeral Cross-sectional Properties
Cortical Area 0.4143 0.1665 0.1032 0.4149
Min Bending Strength 0.3222 -0.1452 0.0440 0.2678
Max Bending Strength 0.2865 -0.2187 -0.1178 0.3005
Torsional Strength 0.2990 -0.1927 -0.0563 0.2900
Shape Index (Imax / Imin) -0.0357 -0.0818 -0.1542 0.0255
Humeral linear data and indices
Length 0.0854 0.0881 0.0795 0.0734
Max diameter (midshaft) 0.1110 0.0745 0.0612 0.1089
Min diameter (midshaft) 0.1353 0.0831 0.0953 0.1133
Circumference (midshaft) 0.1172 0.0918 0.0712 0.1016
Max diameter (min. shaft) 0.1205 0.1112 0.0905 0.0943
Min diameter (min. shaft) 0.1550 0.0827 0.0927 0.1313
Circumference (min. shaft) 0.1161 0.0927 0.0794 0.1025
Midshaft shape index 0.0267 0.0000 0.0270 0.0132
Min. shaft shape index 0.0290 -0.0296 0.0000 0.0396
*In the meta-analysis, Middle Woodland and Mississippian groups were compared as the
preagricultural and agricultural populations. Early late and late late Woodland groups are
considered transitional.









Limitations of Meta-Analysis

In using meta-analysis it is important to understand the inherent assumptions and

limitations of the technique. First, meta-analysis assumes that all available data is

included in the analysis. For the researcher compiling these data, this presents the

problem of publication bias or the "file drawer" problem; studies with significant results

are more likely to be published than those with insignificant results (Rosenthal, 1984). In

this review, the nature of the publication bias was different. Most of the studies used in

the meta-analysis were not primarily investigating sexual dimorphism. Therefore,

significant results in the original publications were not an issue. Rather, finding

sufficient data was a problem due to the standards used for presenting data. The vast

majority of publications on the transition to agriculture provided mean values for the

populations studied, with no information on the dispersion of the sample. Hence,

statistical data could not be gathered from these studies. Situations such as these present

a problem for the meta-analyst as well as any reader hoping to critique the results of a

study.

In collecting data, the hope was to find archaeological sites worldwide from

which information could be drawn. However, the studies which could be included in my

sample are affected by publication bias. As an English-speaker, all the journals and

resources I collected were from sources written in English; all the studies I collected with

data sufficient for a meta-analysis were from sites within the continental United States.

Requests for data from international sources went unanswered. Furthermore, differential

preservation of skeletal remains in the archaeological record is likely to skew any

worldwide study of temporal changes in morphology.









The number of studies which contain data on skeletal morphology spanning the

transition to agriculture is far greater than the number that could be included in a meta-

analysis. A meta-analysis assumes independent samples. Skeletal remains are culturally

sensitive materials and laws limit the access to these remains. As such, the same skeletal

materials are often used in a number of different studies. No skeleton could be used more

than once without violating the assumptions of meta-analysis.

Due to these problems in data collection, the results of this study are only

applicable to the transition to agriculture as it occurred in North America. While a

variety of locations and cultures are considered in these North American samples, they

are all of similar ancestry and adopted maize as the primary agricultural carbohydrate.

To assess the effect of the transition to agriculture on sexual dimorphism worldwide, it

would be necessary to include studies from other areas of the globe.

Meta-analysis has been criticized for glossing over the details of individual

studies. However, Rosenthal (1984) argues that the same can be said of any traditional

review. In fact, a meta-analysis is more likely to reflect the actual results of studies rather

than being overly influenced by information included in their abstracts or discussions

(Rosenthal, 1984).

Another common criticism of meta-analysis is that comparing different studies,

by different researchers, is like comparing apples and oranges. Not only are methods

heterogeneous, but the quality of the studies may also vary. To control for heterogeneity,

only like measurement variables were compared in twenty-two separate meta-analyses.

However, in this study several measurements were combined although the methods of

data collection were not identical. For example, measurements of maximum femur









length and bicondylar length were combined for the meta-analysis of femur length.

Measurements of bending strength in different studies were collected using similar

methods, but then each author used a different formula for standardization. A meta-

analysis would not have been possible had these measurements not been compared. To

do so did not violate any assumption of independence and provided important

information which would have otherwise been unavailable. Glass made an excellent

point in defense of such generalizations. "One compares apples and oranges in the study

of fruit" (Glass, 1978).

The criticisms of meta-analysis have been addressed by its proponents and the

method has proved reliable when used for its intended purpose (Glass et al., 1981; Hunter

et al., 1982; Rosenthal, 1984). In this study the technique has been employed to test the

theory that sexual dimorphism declined with the transition to agriculture. While this

meta-analysis does not directly inform the question of causality, it provides information

about patterns of dimorphism which may then serve future studies of sexual dimorphism.

The availability of data has restricted the applicability of the results to North American

populations but has provided a needed cumulative analysis of how sexual dimorphism

may be affected by the transition to agriculture in a large geographic area.















CONCLUSIONS

Meta-analyses were conducted on the postcranial measurements taken in five

separate studies. Of the nineteen measurements from which populations were deemed

homogeneous, two experienced a significant decline with the transition to agriculture

(10.5 %) and a third experienced a near significant decline (p = 0.0537). A comparison

of the effect sizes found that 79 % of all measurements experienced a decline in

dimorphism as opposed to the remaining 21 % in which a negative effect size indicated

that the degree of dimorphism increased. None of the increases in dimorphism were

found to be significant. The trend towards a decline in dimorphism is more apparent in

humeral than in femoral measurements. The non-statistical analysis of agricultural versus

preagricultural populations shows similar results. In contrast to the theory that sexual

dimorphism declines with the transition to agriculture, in most cases, no significant

change occurs.














APPENDIX A
META-ANALYSES OF FEMORAL MEASUREMENTS


Femur Length (based on maximum or bicondylar length measurements)
source study # t statistic r N df Z
Boyd and Boyd (1989) 1 0.2343 0.0103 524 520 0.2343
Bridges (2000) 3 -0.1495 -0.0135 127 123 -0.1495
Brock and Ruff (1988) 4 -1.3684 -0.1173 138 134 -1.3636
Ruff and Larsen (1990) 5 -2.195 -0.3436 40 36 -2.1216

rbar= -0.0317 Xk-2 = 5.5141 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0071 Ser2/Sr2 = 0.6802 Moderator Variable Present
Ser2 = 0.0048 Z = -1.7002 p = 0.9554, N.S.



Femur Midshaft Circumference
source study # t statistic r N df Z
Bridges (1989) 2 -2.156 -0.1786 174 170 -2.1413
Bridges (2000) 3 0 0 126 122 0
Brock and Ruff (1988) 4 -0.4192 -0.0362 138 134 -0.4190

rbar= -0.0824 Xk-2 = 2.3564 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0063 Ser2/Sr2 = 1.0721 No moderator variable present
Ser2 = 0.0068 Z = -1.4782 p =0.0929, N.S.



Femoral Cortical Area (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 -1.4781 -0.2276 44 40 -1.4579
Bridges (2000) 3 1.2617 0.2058 40 36 1.2478
Brock and Ruff (1988) 4 2.3047 0.3424 44 40 2.2282
Ruff and Larsen (1990) 5 -0.2271 -0.0378 40 36 -0.2270


rbar= 0.0701
Sr2= 0.0498
Ser2 = 0.0236


Xk2 = 9.6710 (0.01 > p > 0.005) Heterogeneous
Ser2/Sr2 = 0.4736 Moderator Variable Present


p = 0.1867, N.S.


Z = 0.8955











Femur Minimum Bending Strength (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 -2.0041 -0.3021 44 40 -1.9538
Bridges (2000) 3 -0.6359 -0.1054 40 36 -0.6342
Brock and Ruff (1988) 4 1.1042 0.1720 44 40 1.0958
Ruff and Larsen (1990) 5 -0.3512 -0.0584 40 36 -0.3509

rbar = -0.0731 Xk-2 = 4.3424 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0298 Ser2/Sr2 = 0.7914 No Moderator Variable
Ser2 = 0.0236 Z = -0.9215 p = 0.8212, N.S.



Femur Maximum Bending Strength (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 -0.4867 -0.0767 44 40 -0.4860
Bridges (2000) 3 0.3108 0.0517 40 36 0.3106
Brock and Ruff (1988) 4 1.5627 0.2399 44 40 1.5389
Ruff and Larsen (1990) 5 0.1877 0.0313 40 36 0.1876

rbar= 0.0625 Xk-2 = 2.5951 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0136 Ser2/Sr2 = 1.7400 No Moderator Variable
Ser2 0.0236 Z = 0.7756 p = 0.2206, N.S.



Femoral Shape Index (Imax/Imin)
source study # t statistic r N df Z
Bridges (2000) 3 -1.1831 -0.1935 40 36 -1.1716
Brock and Ruff (1988) 4 1.2728 0.1973 44 40 1.2599
Ruff and Larsen (1990) 5 0.7035 0.1165 40 36 0.7011


rbar =
Sr2
Ser =
^er


0.0452
0.0282
0.0241


Xk-2 =
Ser2/Sr2 =
Z =


3.8381
0.8538
0.4558


(0.5 > p > 0.1) Homogeneous
No Moderator Variable
p = 0.3300, N.S.









Femoral J (J=Iap + Iml, polar second moment of area, size standardized at midshaft)
source study # t statistic r N df Z
Bridges (2000) 3 -0.2930 -0.0488 40 36 -0.2928
Brock and Ruff (1988) 4 1.7651 0.2688 44 40 1.7307
Ruff and Larsen (1990) 5 -0.7420 -0.1227 40 36 -0.7391

rbar= 0.0401 Xkl2 = 3.9915 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0297 Ser2/Sr2 = 0.8130 No Moderator Variable
Ser2 0.0241 Z = 0.4034 p = 0.3446, N.S.



Subtrochanteric Cortical Area (size standardized)
source study # t statistic r N df Z
Brock and Ruff (1988) 4 2.3822 0.3647 41 37 1.2589
Ruff and Larsen (1990) 5 0.1767 0.0294 40 36 0.7011

rbar= 0.1991 Xk-2 = 3.5471 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0281 Ser2/Sr2 = 0.8107 No Moderator Variable
Ser2 0.0228 Z = 1.3859 p = 0.0838, N.S.



Subtrochanteric Minimum Bending Strength (size standardized)
source study # t statistic r N df Z
Brock and Ruff (1988) 4 1.4423 0.2307 41 37 1.4221
Ruff and Larsen (1990) 5 0.2916 0.0485 40 36 0.2914

rbar= 0.1408 Xk-2 = 0.9102 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0083 Ser2/Sr2 = 2.8595 No Moderator Variable
Ser2 = 0.0237 Z = 1.2116 p = 0.1131, N.S.



Subtrochanteric Maximum Bending Strength (size standardized)
source study # t statistic r N df Z
Brock and Ruff (1988) 4 1.0468 0.1696 41 37 1.0391
Ruff and Larsen (1990) 5 1.2519 0.2042 40 36 1.2382


rbar =
Sr2
Ser =


0.1867
0.0003
0.0230


Xk-2 =
Ser2/Sr2 =
Z =


0.0367
76.6613
1.6103


(0.9 > p > 0.5) Homogeneous
No Moderator Variable
p = 0.0537, Near Significant










Subtrochanteric Femoral Shape Index (Imax/Imin)
source study # t statistic r N df Z
Brock and Ruff (1988) 4 -0.6500 0.1063 41 37 -0.6481
Ruff and Larsen (1990) 5 1.2510 0.2041 40 36 1.2374

rbar= 0.1546 Xk-2 = 0.2713 (0.9 > p > 0.5) Homogeneous
Sr2= 0.0024 Ser2/Sr = 9.8274 No Moderator Variable
Ser2 0.0235 Z = 0.4167 p = 0.3409, N.S.



Subtrochanteric J (J=Iap + Iml, polar second moment of area, size standardized at
midshaft)
source study # t statistic r N df Z
Brock and Ruff (1988) 4 1.4423 0.2307 41 37 1.4221
Ruff and Larsen (1990) 5 0.2916 0.0485 40 36 0.2914


rbar =
Sr2
Ser =


0.2010
0.0012
0.0227


X 2-
Xk-1 =
Ser2/Sr2 =
Z =


0.1569
18.3894
1.7342


(0.9 > p > 0.5) Homogeneous
No Moderator Variable
p = 0.0418, SIGNIFICANT















APPENDIX B
META-ANALYSES OF HUMERAL MEASUREMENTS


Humerus Length
source study # t statistic r N df Z
Ruff and Larsen (1990) 5 -0.9413 -0.1294 56 52 -0.9373
Bridges (2000) 3 0.7501 0.0710 115 111 0.7491

rbar= 0.0054 Xk-2 = 1.4492 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0084 Ser2/Sr2 = 1.3949 No Moderator Variable
Ser2= 0.0117 Z = -0.0941 p = 0.5359, N.S.



Humerus Midshaft Circumference
source study # t statistic r N df Z
Bridges (1989) 2 2.2074 0.1453 220 216 2.1949
Bridges (2000) 3 0.4943 0.0467 116 112 0.4941

rbar= 0.1112 Xk-2 = 6.4901 (0.025 > p > 0.01) Heterogeneous
Sr2= 0.0153 Ser2/Sr2 = 0.3805 Moderator Variable Present
Ser2 = 0.0058 Z = 1.3445 p = 0.090, N.S.



Humerus Cortical Area (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 0.2326 0.0347 49 45 0.2325
Bridges (2000) 3 0.2995 0.0623 40 36 0.2993
Ruff and Larsen(1990) 5 0.4068 0.0563 56 52 0.4065


rbar = 0.0507
Sr2= 0.0001
Ser2 = 0.0206


Xk-2 =
Ser 2/Sr2 =


0.0220
150.7371


(0975 > p > 0.9) Homogeneous
No Moderator Variable


Z = 0.4692 p = 0.3228, N.S.









Humerus Minimum Bending Strength (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 1.8729 0.2689 49 45 1.8364
Bridges (2000) 3 -0.6359 -0.1054 40 36 -0.6342
Ruff and Larsen(1990) 5 1.4135 0.1924 56 52 1.3999

rbar= 0.1361 Xk-2 = 4.5214 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0233 Ser2/Sr2 = 0.8564 No Moderator Variable
Ser2 = 0.0199 Z = 1.3011 p = 0.0968, N.S.



Humerus Maximum Bending Strength (size standardized at midshaft)
source study # t statistic r N df Z
Bridges (1989) 2 1.9915 0.2846 49 45 1.9476
Bridges (2000) 3 0.3108 0.0517 40 36 0.3106
Ruff and Larsen(1990) 5 1.0010 0.1375 56 52 0.9962

rbar= 0.1635 Xk-2 = 1.7955 (0.5 > p > 0.1) Homogeneous
Sr2= 0.0087 Ser2/Sr2 = 2.2621 No Moderator Variable
Ser2 0.0196 Z = 1.8789 p = 0.0307, SIGNIFICANT



Humeral Shape Index (Imax/Imin)
source study # t statistic r N df Z
Bridges (2000) 3 -0.9142 -0.1873 27 23 -0.9059
Ruff and Larsen(1990) 5 -0.8729 -0.1202 56 52 -0.8697


rbar= -0.1420
Sr2= 0.0010
Ser2 = 0.0231


Xk-2 = 0.0629
Ser2/Sr2= 23.4217
Z = -0.8878


(0.9 > p > 0.5) Homogeneous
No Moderator Variable
p = 0.8106, N.S.






50



Humeral J (J=Iap + 1ml, polar second moment of area, size standardized at midshaft)
source study # t statistic r N df Z
Bridges (2000) 3 -0.3942 0.0819 27 23 -0.3935
Ruff and Larsen(1990) 5 1.2280 0.1679 56 52 1.0368

rbar= 0.1399 Xk-12 = 0.1819 (0.9 > p > 0.5) Homogeneous
Sr2= 0.0016 Ser2/Sr2 = 14.2832 No Moderator Variable
Ser2 0.0232 Z = 0.5838 p = 0.2810, N.S.














APPENDIX C
META-ANALYSES OF OTHER LONG BONE MEASUREMENTS


Tibia Midshaft Circumference
source study # t statistic r N df Z
Bridges (1989) 2 0 0 121 117 0
Bridges (2000) 3 0.3914 0.0461 76 72 0.3912

rbar= 0.0178 Xk-2 = 0.0631 (0.9 > p > 0.5) Homogeneous
Sr2= 0.0003 Ser2/Sr2 = 32.8342 No Moderator Variable
Ser = 0.0101 Z = 0.1956 p = 0.4247, N.S.



Ulna Midshaft Circumference
source study # t statistic r N df Z
Bridges (1989) 2 -0.0910 -0.0082 140 136 -0.0910
Bridges (2000) 3 0.4406 0.0505 80 76 0.4403

rbar= 0.0131 Xk-2 = 0.1244 (0.9 > p > 0.5) Homogeneous
Sr2= 0.0006 Ser2/Sr2 = 16.4999 No Moderator Variable
Ser = 0.0091 Z = 0.1747 p = 0.4325, N.S.



Radius Midshaft Circumference
source study # t statistic r N df Z
Bridges (1989) 2 2.9719 0.2513 142 138 2.9243
Bridges (2000) 3 0.9075 0.0947 95 91 0.9054


rbar= 0.1885
Sr2= 0.0414
Ser2 = 0.0078


Xkl2 = 14.8913


Serl/S


( p < 0.01) Heterogeneous


r = 0.1897 Moderator Variable Present
Z = 1.9149 p = 0.0281, Significant, but not
relevant due to the presence of
moderator variable















LIST OF REFERENCES


Alexander RD, Hoogland JL, Howard RD, Noonan KM, and Sherman PW (1979) Sexual
dimorphisms and breeding systems in pinnipeds, ungulates, primates, and humans.
In NA Chagnon and W Irons, (eds.): Evolutionary biology and human social
behavior. North Scituate, MA.: Duxbury Press. p 402-435.

Angel JL (1984) Health as crucial factor in the changes from hunting to developed
farming in eastern Mediterranean. In MN Cohen and GJ Armelagos, (eds.):
Paleopathology at the origins of agriculture. London: Academic Press. p 50-73.

Armelagos GJ (1990) Health and disease in prehistoric populations in transition. In GJ
Armelagos (ed.): Diseases in populations in transition. New York: Bergin and
Garvey.

Armelagos GJ, and Van Gerven DP (1980) Sexual dimorphism and human evolution: an
overview. Journal of Human Evolution 9:437-446.

Borgognini Tarli SM, and Repetto E (1997) Sex differences in human populations:
change through time. In ME Morbeck, A Galloway and AL Zihlman, (eds.):
Evolving female. Princeton, NJ: Princeton University Press. p 198-208.

Boserup E (1965) The conditions of agricultural growth: the economies of agrarian
change under population pressure. Chicago: Aldine.

Boyd DC, and Boyd CC (1989) A comparison on Tennessee archaic and mississippian
maximum femoral lengths and midshaft diameters: subsistence change and
postcranial variability. Southeastern Archaeology 8:107-116.

Brace CL (1963) Structural reduction in evolution. The American Naturalist 97:39-49.

Brace CL (1973) Sexual dimorphism in human evolution. In CL Brace and J Metress,
(eds.): Man in evolutionary perspective. New York: John Wiley & Sons, Inc. p
238-254.

Brace CL, and Ryan AS (1980) Sexual dimorphism and human tooth size differences.
Journal of Human Evolution 9:417-435.

Bramblett CA (1994) Patterns of primate behavior, 2nd ed. Prospect Heights, IL:
Waveland Press, Inc.









Brauer GW (1982) Size sexual dimorphism and secular trend: indicators of subclinal
malnutrition? In RL Hall, (ed.): Sexual dimorphism in Homo Sapiens: a question of
size. New York: Praeger Publishers. p 245-262.

Bridges PS (1989) Changes in activities with the shift to agriculture in the southeastern
United States. Current Anthropology 30:385-394.

Bridges PS (2000) Changes in long bone diaphyseal strength with horticultural
intensification in west-central Illinois. American Journal of Physical Anthropology
112:217-238.

Brock SL, and Ruff CB (1988) Diachronic patterns of change in structural properties of
the femur in the prehistoric American southwest. American Journal of Physical
Anthropology 75:113-127.

Bronson B (1977) The earliest farming: demography as cause and consequence. In CA
Reed, (ed.): Origins of agriculture. Paris: Mouton Publishers.

Brues A (1959) The spearman and the archer an essay on selection in body build.
American Anthropologist 61:457-469.

Buikstra JE (1984) The lower Illinois river region: a prehistoric context for the study of
ancient diet and health. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at
the origins of agriculture. London: Academic Press. p 215-234.

Campbell BG, and Loy JD (2000) Humankind emerging. Needham Heights, MA: Allyn
& Bacon.

Clark GA (1988) New method for assessing changes in growth and sexual dimorphism in
paleoepidemiology. American Journal of Physical Anthropology 77:105-116.

Cohen MN (1977) Population pressure and the origins of agriculture: an archaeological
example from the coast of Peru. In CA Reed, (ed.): Origins of agriculture. Paris:
Mouton Publishers. p 135-177.

Cohen MN, and Armelagos GJ (1984) Paleopathology at the origins of agriculture:
editor's summation. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the
origins of agriculture. New York: Academic Press, Inc. p 585-601.

DeVore I, and Washburn SL (1963) Baboon ecology and human evolution. In FC Howell
and F Bouliere, (eds.): African ecology and human evolution. New York: Viking
Fund Publication No. 36. p 335-367.

Eveleth PB (1975) Differences between ethnic groups in sex dimorphism of adult height.
Annals of Human Biology 2:35-39.






54


Finkel DJ (1982) Sexual dimorphism and settlement pattern in Middle Eastern skeletal
populations. In RL Hall, (ed.): Sexual Dimorphism in Homo sapiens. New York:
Praeger Publishers. p 165-185.

Frayer DW (1980) Sexual dimorphism and cultural evolution in the late Pleistocene and
Holocene of Europe. Journal of Human Evolution 9:399-415.

Frayer DW (1981) Body size, weapon use, and natural selection in the European Upper
Paleolithic and Mesolithic. American Anthropologist 83:57-73.

Frayer DW, and WolpoffMH (1985) Sexual dimorphism. Annual Review of
Anthropology 14:429-473.

Gaulin SJC, and Boster JS (1985) Cross-cultural differences in sexual dimorphism: is
there any difference to be explained? Ethology and Sociobiology 6:219-225.

Gaulin SJC, and Boster JS (1992) Human marriage systems and sexual dimorphism in
stature. American Journal of Physical Anthropology 89:467-475.

Glass GV (1978) In Defense of Generalization. The Behavioral and Brain Sciences
3:394-395.

Glass GV, McGaw B, and Smith ML (1981) Meta-analysis in social research. London:
Sage Publications.

Gray JP, and Wolfe LD (1980) Height and sexual dimorphism of stature among human
societies. American Journal of Physical Anthropology 53:441-456.

Gustafson A, and Lindenfors P (2004) Human size evolution: no evolutionary allometric
relationship between male and female stature. Journal of Human Evolution 47:253-
266.

Hall RL (1982) Unit of analysis. In RL Hall (ed.): Sexual dimorphism in Homo sapiens.
New York: Praeger Publishers. p 189-196.

Hinton RJ, and Carlson DS (1979) Temporal change in human temporomandibular joint
size and shape. American Journal of Physical Anthropology 50:325-334.

Holden C, and Mace R (1999) Sexual dimorphism in stature and women's work: a
phylogenetic cross-cultural analysis. American Journal of Physical Anthropology
110:1: 27-45.

Hunter JE, Schmidt FL, and Jackson GB (1982) Meta-Analysis: Cumulating Research
Findings across Studies. London: Sage Publications.

Kennedy KAR, Deraniyagala SU, Roertgen JC, Chiment J, and Disotell T (1987) Upper
Pleistocene fossil hominids form Sri Lanka. American Journal of Physical
Anthropology 72:441-461.









Krantz GS (1982) The fossil record of sex. In RL Hall, (ed.): Sexual dimorphism in
Homo sapiens. New York: Praeger Publishers. p 85-105.

Larsen CS (1984) Health and disease in prehistoric Georgia: the transition to agriculture.
In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of
agriculture. London: Academic Press. p 367-392.

Larsen CS (1995) Biological changes in human populations with agriculture. Annual
Review of Anthropology 24:185-213.

Lazenby RA (2002) Population variation in second metacarpal sexual size dimorphism.
American Journal of Physical Anthropology 118:378-384.

Lieberman LS (1982) Normal and abnormal sexual dimorphic patterns of growth and
development. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens: a question of
size. New York: Praeger Publishers. p 263-316.

Martin DL, Armelagos GJ, Goodman AH, and Van Gerven DP (1984) The effects of
socioeconomic change in prehistoric Africa: Sudanese Nubia as a case study. In
MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture.
London: Academic Press.

Meiklejohn C, Schentag C, Venema A, and Key P (1984) Socioeconomic change and
patterns of pathology and variation in the Mesolithic and Neolithic of Western
Europe: some suggestions. In MN Cohen and GJ Armelagos, (eds.):
Paleopathology at the Origins of Agriculture. London: Academic Press. p 75-100.

Morbeck ME (1997) Evolving Female. Princeton, NJ: Princeton University Press.

Murdock GP, and Provost C (1973) Factors in the division of labor by sex: a cross
cultural analysis. Ethnology 12:203.

Nickens PR (1976) Stature reduction as an adaptive response to food production in
Mesoamerica. Journal of Archaeological Science 3:31-41.

Ortner DJ (2003) Identification of Pathological Conditions in Human Skeletal Remains.
New York: Academic Press.

Ott LH, and Longnecker M (2001) An introduction to statistical methods and data
analysis, 5th ed. Pacific Grove, CA: Duxbury.

Perzigan AJ, Tench PA, and Braun DJ (1984) Prehistoric health in the Ohio River valley.
In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of
agriculture. New York: Academic Press, Inc. p 347-366.

Rathbun TA (1984) Skeletal pathology from the Paleolithic through the Metal Ages in
Iran and Iraq. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at the
origins of agriculture. London: Academic Press.









Rensch B (1959) Evolution above the species level. London: Methuen & Co.

Rivers J (1982) Women and children last: an essay on sex discrimination in disasters.
Disasters 6:256-267.

Rose JC, Burnett BA, and Blaeuer MW (1984) Paleopathology and the origins of maize
agriculture in the lower mississippi valley and Caddoan culture areas. In MN
Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture. New
York: Academic Press, Inc. p 393-424.

Rosenthal R (1984) Meta-analytical procedures for social research. Beverly Hills, CA:
Sage Publications.

Rowe N (1996) The pictorial guide to living primates. Charlestown, RI: Pogonias Press.

Ruff CB (1987) Sexual dimorphism in human lower limb bone structure: relationship to
subsistence strategy and sexual division of labor. Journal of Human Evolution
16:391-416.

Ruff CB (1991) Aging and osteoporosis in Native Americans from Pecos Pueblo, New
Mexico. New York: Garland Publishing, Inc.

Ruff CB, and Larsen CS (1990) Postcranial biomechanical adaptations to subsistence
strategy changes on the Georgia coast. Anthropological Papers of the Museum of
Natural History 68:94-120.

Ruff CB, Larsen CS, and Hayes WC (1984) Structural changes in the femur with the
transition to agriculture on the Georgia coast. American Journal of Physical
Anthropology 64:125-136.

Ruff CB, Trinkaus E, Walker A, Larsen CS (1993) Postcranial robusticity in Homo. I:
temporal trends and mechanical interpretation. American Journal of Physical
Anthropology 91:21-53.

Smith P, Bar-Yosef O, and Sillen A (1984) Archaeological and skeletal evidence for
dietary change during the Late Pleistocene/Early Holocene in the Levant. In MN
Cohen and GJ Armelagos, (eds.): Paleopathology at the origins of agriculture.
London: Academic Press. p 101-136.

Smith RJ (1999) Statisics of sexual size dimorphism. Journal of Human Evolution
36:423-459.

Stini WA (1969) Nutritional stress and growth: sex differences in adaptive response.
American Journal of Physical Anthropology 31:417-426.

Stini WA (1975) Ecology and human adaptation. Dubuque, IA: Wm C. Brown Company
Publishers.









Stini WA (1982) Sexual dimorphism and nutrient reserves. In RL Hall, (ed.): Sexual
dimorphism in Homo sapiens: a question of size. New York: Praeger Publishers. p
391-420.

Stinson S (1985) Sex differences in environmental sensitivity during growth and
development. Yearbook of Physical Anthropology 28:123-147.

Swanton J (1946) The Indians of the Southeastern United States: Smithsonian Press.

Trivers RL (1972) Parental investment and sexual selection. In B Campbell, (ed.): Sexual
selection and the descent of man: 1871-1971. Chicago: Aldine Publishing
Company p 136-179.

Turner CG (1979) Dental anthropological indications of agriculture among Jomon people
of central Japan. American Journal of Physical Anthropology 51:619-636.

Ubelaker DH (1984) Prehistoric human biology of Ecuador: possible temporal trends and
cultural correlations. In MN Cohen and GJ Armelagos, (eds.): Paleopathology at
the origins of agriculture. New York: Academic Press, Inc. p 491-513.

Wolanski N, and Kasprzak E (1976) Stature as a measure of environmental change.
Current Anthropology 17:548-552.

Wolfe LD, and Gray JP (1982a) A Cross-cultural investigation into the sexual
dimorphism of stature. In RL Hall, (ed.): Sexual dimorphism in Homo sapiens.
New York: Praeger Publishers. p 197-230.

Wolfe LD, and Gray JP (1982b) Subsistence practices and human sexual dimorphism of
stature. Journal of Human Evolution 11:575-580.

Wood W, and Christensen PN (2004) Quantitative research synthesis: examining study
outcomes over samples, settings, and time. In AT Panter, (ed.): Handbook of
methods in social psychology. Thousand Oaks, CA: Sage Publications.















BIOGRAPHICAL SKETCH

Anna Elizabeth Vick was born in 1975, in Chapel Hill, North Carolina. She

graduated from the University of North Carolina at Chapel Hill in 1998 with a Bachelor

of Arts degree in anthropology.