Technical note: Measures of differentiation for quantitative traits.

Studies of anthropological genetics and bioarcheology often examine the degree of among-group variation in quantitative traits such as craniometrics and anthropometrics. One comparative index of among-group differentiation is the minimum value of Wright's F ST as estimated from quantitative traits. This measure has been used in certain population-genetic applications such as comparison with F ST estimated from genetic data, although some inferences are limited by how well the data and study design fit the underlying population-genetic model. In many cases, all that is needed is a simple measure of among-group variation. One such measure is R 2 , the proportion of total phenotypic variation accounted for by among-group phenotypic variation, a measure easily obtained from analysis of variance and regression methods. This paper shows that R 2 and minimum F ST are closely related as Min F ST ≈ R 2 / 2 - R 2 . R 2 is computationally easy and may be useful in cases where all we need is a simple measure of relative among-group differentiation.

Bone functional adaptation does not erase neutral evolutionary information.

OBJECTIVES: To investigate whether diaphyseal and craniofacial variation similarly reflect neutral genetic variation among modern European and South Africans. MATERIALS AND METHODS: Diaphyseal and craniofacial data were collected on English, South European, and South African samples. The Relethford-Blangero model was used to compare predicted among-population relationships generated by limb bones relative to those generated by the crania and, further, to test whether adaptive plasticity affected these predicted relationships. Evidence of adaptive plasticity was confirmed by comparing J, an indicator of limb bone robusticity, among individuals who worked different occupations in industrializing Lisbon (Portugal) and Bologna (Italy). RESULTS: Diaphyses were more variable than were crania and more robust in individuals with physically demanding occupations-both consistent with expectations of adaptive plasticity. However, diaphyseal variation still generated among-population relationships consistent with neutral genetic predictions and Mantel tests confirmed a high, significant correlation between diaphyseal and craniofacial distance matrices. This pattern was not strongly affected by adaptive plasticity. DISCUSSION: Among-population patterns of diaphyseal variation are consistent with neutral expectations and are consistent with historical data on population composition, genetics, and migration. Furthermore, plasticity induced by Industrial-era levels of physical activity does not erase these neutral signatures. Diaphyseal variation may therefore be useful to infer neutral (presumably genetic) information across populations, and controlling for existing relationships may strengthen inferences of physical activity made when comparing limb bone structure across populations.

Cranial measures and ancient DNA both show greater similarity of Neandertals to recent modern Eurasians than to recent modern sub-Saharan Africans.

OBJECTIVES: Ancient DNA analysis has shown that present-day humans of Eurasian ancestry are more similar to Neandertals than are present-day humans of sub-Saharan African ancestry, reflecting interbreeding after modern humans first left Africa. We use craniometric data to test the hypothesis that the crania of recent modern humans show the same pattern. MATERIALS AND METHODS: We computed Mahalanobis squared distances between a published Neandertal centroid based on 37 craniometric traits and each of 2,413 recent modern humans from the Howells global data set (N = 373 sub-Saharan Africans, N = 2,040 individuals of Eurasian descent). RESULTS: The average distance to the Neandertal centroid is significantly lower for Eurasian crania than for sub-Saharan African crania as expected from the findings of ancient DNA (p < 0.001). This result holds when examining distances for separate geographic regions of humans of Eurasian descent (Europeans, Asians, Australasians, Native Americans, and Pacific Islanders). Most of these results are also seen when examining distances partitioning size and shape variation. DISCUSSION: Our results show that the genetic difference in Neandertal ancestry seen in the DNA of present-day sub-Saharan Africans and Eurasians is also found in patterns of recent modern human craniometric variation.

Comparison of observed and expected levels of genetic diversity based on surname frequencies: An example from historical Massachusetts.

OBJECTIVES: The inverse of random inbreeding based on surname frequencies (1/Fr ) is an estimate of genetic diversity, and its expectation is a function of the number of migrants into a population. Observed and expected values of (1/Fr ) were compared to determine if observed diversity matches theoretical expectations under conditions of rapid demographic change using data from historical Massachusetts. MATERIALS AND METHODS: Data on migration and surnames were taken from 6,038 marriage records from the eighteenth and nineteenth centuries from five towns in north-central Massachusetts. Data for each town were broken down into a number of time cohorts defined by year of marriage, giving 33 samples based on town and year of marriage. The number of migrants (M) and the inverse of the random component of inbreeding (1/Fr ) were derived for each sample based on surname frequencies. RESULTS: There was no significant difference between observed and expected values of 1/Fr in samples where there were 100 or more migrants. However, 1/Fr was significantly higher in samples where M < 100, which is possibly due to these samples not having reached equilibrium, resulting in higher than expected values of 1/Fr . Regression of residual values of 1/Fr (observed-expected) on the number of years since settlement supports this interpretation. CONCLUSION: The number of migrants affects the level of genetic diversity inferred from surname frequencies, and the relationship between observed and expected measures depends on the number of migrants and the proximity of a given sample to an equilibrium state.

Genetic drift and the population history of the Irish travellers.

The Irish Travellers are an itinerant group in Ireland that has been socially isolated. Two hypotheses have been proposed concerning the genetic origin of the Travellers: (1) they are genetically related to Roma populations in Europe that share a nomadic lifestyle or (2) they are of Irish origin, and genetic differences from the rest of Ireland reflect genetic drift. These hypotheses were tested using data on 33 alleles from 12 red blood cell polymorphism loci. Comparison with other European, Roma, and Indian populations shows that the Travellers are genetically distinct from the Roma and Indian populations and most genetically similar to Ireland, in agreement with earlier genetic analyses of the Travellers. However, the Travellers are still genetically distinct from other Irish populations, which could reflect some external gene flow and/or the action of genetic drift in a small group that was descended from a small number of founders. In order to test the drift hypothesis, we analyzed genetic distances comparing the Travellers to four geographic regions in Ireland. These distances were then compared with adjusted distances that account for differential genetic drift using a method developed by Relethford (Hum Biol 68 (1996) 29-44). The unadjusted distances show the genetic distinctiveness of the Travellers. After adjustment for the expected effects of genetic drift, the Travellers are equidistant from the other Irish samples, showing their Irish origins and population history. The observed genetic differences are thus a reflection of genetic drift, and there is no evidence of any external gene flow.

Population-specific deviations of global human craniometric variation from a neutral model.

Past studies have revealed that much of human craniometric variation follows a neutral model of population relationships. At the same time, there is evidence for the influence of natural selection in having shaped some global diversity in craniometrics. In order to partition these effects, and to explore other potential population-specific influences, this article analyzes residuals of craniometric distances from a geographically based neutral model of population structure. W.W. Howells' global craniometric data set was used for these analyses, consisting of 57 measurements for 22 populations around the world, excluding Polynesia and Micronesia because of the relatively recent settlement of these regions. Phenotypic and geographic distances were derived between all pairs of populations. Three-dimensional multidimensional scaling configurations were obtained for both distance matrices, and compared using a Procrustes rotation method to show which populations do not fit the geographic model. This analysis revealed three major deviations: the Buriat, Greenland Inuit, and Peru. The deviations of the Buriat and Greenland Inuit appear to be related to long-term adaptation to cold environments. The Peruvian sample is more similar to other New World populations than expected based on geographic distance alone. This deviation likely reflects the evolutionarily recent movement of human populations into South America, such that these populations are further from genetic equilibrium. This same pattern is seen in South American populations in a comparative analysis of classical genetic markers, but not in a comparative analysis of STR loci, perhaps reflecting the higher mutation rate for the latter.

Race and global patterns of phenotypic variation.

Phenotypic traits have been used for centuries for the purpose of racial classification. Developments in quantitative population genetics have allowed global comparison of patterns of phenotypic variation with patterns of variation in classical genetic markers and DNA markers. Human skin color shows a high degree of variation among geographic regions, typical of traits that show extensive natural selection. Even given this high level of geographic differentiation, skin color variation is clinal and is not well described by discrete racial categories. Craniometric traits show a level of among-region differentiation comparable to genetic markers, with high levels of variation within populations as well as a correlation between phenotypic and geographic distance. Craniometric variation is geographically structured, allowing high levels of classification accuracy when comparing crania from different parts of the world. Nonetheless, the boundaries in global variation are not abrupt and do not fit a strict view of the race concept; the number of races and the cutoffs used to define them are arbitrary. The race concept is at best a crude first-order approximation to the geographically structured phenotypic variation in the human species.

Postfamine stature and socioeconomic status in Ireland.

Previous research has documented socioeconomic stratification of secular trend in height in historical populations. Using data from 4,900 males and 1,430 females born between 1840 and 1910 collected as part of the Harvard Anthropological Survey of Ireland (1934-1936), this study examined the secular changes in postfamine Ireland using several socioeconomic variables, including: occupation, migration, education, siblings, birthplace, and occupation of father and mother's father. Correlations were also calculated between height and various historical economic indices. Significant differences in the height of Irish males were found by occupation, education, and socioeconomic status of father and maternal grandfather. Males employed in agriculture, or whose fathers or grandfathers were so employed, were significantly taller than other males. For the smaller female sample, only occupation and grandfather's socioeconomic status had a significant impact on height. An inverse correlation was also found between the British Cost of Living Index (BCL) and male heights. Our results suggest that availability of resources plays an important role in the overall nutritional status reflected in terminal adult height.

Geostatistics and spatial analysis in biological anthropology.

A variety of methods have been used to make evolutionary inferences based on the spatial distribution of biological data, including reconstructing population history and detection of the geographic pattern of natural selection. This article provides an examination of geostatistical analysis, a method used widely in geology but which has not often been applied in biological anthropology. Geostatistical analysis begins with the examination of a variogram, a plot showing the relationship between a biological distance measure and the geographic distance between data points and which provides information on the extent and pattern of spatial correlation. The results of variogram analysis are used for interpolating values of unknown data points in order to construct a contour map, a process known as kriging. The methods of geostatistical analysis and discussion of potential problems are applied to a large data set of anthropometric measures for 197 populations in Ireland. The geostatistical analysis reveals two major sources of spatial variation. One pattern, seen for overall body and craniofacial size, shows an east-west cline most likely reflecting the combined effects of past population dispersal and settlement. The second pattern is seen for craniofacial height and shows an isolation by distance pattern reflecting rapid spatial changes in the midlands region of Ireland, perhaps attributable to the genetic impact of the Vikings. The correspondence of these results with other analyses of these data and the additional insights generated from variogram analysis and kriging illustrate the potential utility of geostatistical analysis in biological anthropology.

Boas and beyond: migration and craniometric variation.

Migration is expected to affect craniometric variation in three ways: 1) movement into a different environment leading to developmental plasticity; 2) movement into a different environment followed by in situ adaptation through natural selection; and 3) changes in among-group differentiation and genetic distance through the action of gene flow. The relative influence of these three factors has been argued in the literature, most recently in a series of articles debating the statistical and biological significance of Boas's immigration studies as they relate to cranial plasticity. The Boas debate is discussed within the broader context of debate over genetic and environmental influences on craniometric variation. Additional examples are provided from an ongoing study of global craniometric variation. Although developmental plasticity and climatic adaptation have had an impact on craniometric variation, these factors tend not to erase, or even obscure greatly, underlying patterns of population structure and history that fit a neutral model of quantitative variation. Thus, craniometric data can be used to explore questions of gene flow and genetic affinity.

Global patterns of isolation by distance based on genetic and morphological data.

The isolation-by-distance model predicts that genetic similarity between populations will decrease exponentially as the geographic distance between them increases, because of the limiting effect of geographic distance on rates of gene flow. Many studies of human populations have applied the isolation-by-distance model to genetic variation between local populations in a limited geographic area, but few have done so on a global level, and these few used different models and analytical methods. I assess genetic variation between human populations across the world using data on red blood cell polymorphisms, microsatellite DNA markers, and craniometric traits. The isolation-by-distance model provides an excellent fit to average levels of genetic similarity within geographic distance classes for all three data sets, and the rate of distance decay is the same in all three. These results suggest that a common pattern of global gene flow mediated by geographic distance is detectable in diverse genetic and morphological data. An alternative explanation is that the correspondence between genetic similarity and geographic distance reflects the history of dispersal of the human species out of Africa.

Local extinction and recolonization, species effective population size, and modern human origins.

A primary objection from a population genetics perspective to a multiregional model of modern human origins is that the model posits a large census size, whereas genetic data suggest a small effective population size. The relationship between census size and effective size is complex, but arguments based on an island model of migration show that if the effective population size reflects the number of breeding individuals and the effects of population subdivision, then an effective population size of 10,000 is inconsistent with the census size of 500,000 to 1,000,000 that has been suggested by archeological evidence. However, these models have ignored the effects of population extinction and recolonization, which increase the expected variance among demes and reduce the inbreeding effective population size. Using models developed for population extinction and recolonization, we show that a large census size consistent with the multiregional model can be reconciled with an effective population size of 10,000, but genetic variation among demes must be high, reflecting low interdeme migration rates and a colonization process that involves a small number of colonists or kin-structured colonization. Ethnographic and archeological evidence is insufficient to determine whether such demographic conditions existed among Pleistocene human populations, and further work needs to be done. More realistic models that incorporate isolation by distance and heterogeneity in extinction rates and effective deme sizes also need to be developed. However, if true, a process of population extinction and recolonization has interesting implications for human demographic history.

Heritability and anthropometric influences on human fertility.

This study researched the impact of anthropometrics and size-of-family of orientation on women's fertility by using path analysis. The data were collected as part of the anthropological study conducted in Ireland by Harvard University personnel before the Second World War. The women included in this analysis were all over age 49 and were either married or widowed at the time of the survey. Our results indicate that the heritability of fertility is moderate in this sample and that there is a tendency for heavy women to have a higher fertility. However, when anthropometrics and size-of-family of orientation were entered as independent variables in a path diagram, an insignificant portion of the variation of fertility was explained. In this Irish population, the main cause of differential fertility was cultural rather than biological. A large portion of women never married and no unmarried woman reported producing a child.

Apportionment of global human genetic diversity based on craniometrics and skin color.

A number of analyses of classical genetic markers and DNA polymorphisms have shown that the majority of human genetic diversity exists within local populations (approximately 85%), with much less among local populations (approximately 5%) or between major geographic regions or "races" (approximately 10%). Previous analysis of craniometric variation (Relethford [1994] Am J Phys Anthropol 95:53-62) found that between 11-14% of global diversity exists among geographic regions, with the remaining diversity existing within regions. The methods used in this earlier paper are extended to a hierarchical partitioning of genetic diversity in quantitative traits, allowing for assessment of diversity among regions, among local populations within regions, and within local populations. These methods are applied to global data on craniometric variation (57 traits) and skin color. Multivariate analysis of craniometric variation shows results similar to those obtained from genetic markers and DNA polymorphisms: roughly 13% of the total diversity is among regions, 6% among local populations within regions, and 81% within local populations. This distribution is concordant with neutral genetic markers. Skin color shows the opposite pattern, with 88% of total variation among regions, 3% among local populations within regions, and 9% within local populations, a pattern shaped by natural selection. The apportionment of genetic diversity in skin color is atypical, and cannot be used for purposes of classification. If racial groups are based on skin color, it appears unlikely that other genetic and quantitative traits will show the same patterns of variation.

Re-examination of secular change in adult Irish stature.

Data from a large cross-sectional study conducted in the 1930s were used to examine secular changes in adult stature in post-Famine Ireland. The sample consists of 4061 males and 804 females, who were 30 years of age or older at the time of measurement (mostly in 1935). The Himes-Mueller method was used to adjust stature for the effects of aging so that secular change could be assessed. The linear regression of adjusted stature on year of birth was significant for both males and females, and there was no significant difference in slopes between the sexes. These results show an increase in stature of approximately 0.35-0.40 cm per decade, which is not that different from increases found in other studies of 19th century European populations. No significant curvilinearity was detected. These results show that an earlier analysis of a much smaller portion of the sample was biased due to small numbers. © 1995 Wiley-Liss, Inc.

Unequal adaptive value of changing cigarette use during pregnancy for heavy, moderate, and light smokers.

To determine the effects of changes in maternal cigarette use during pregnancy on birthweight, a sample of Caucasian births free of major malformations (n = 9,943) was examined. Births were stratified by level of maternal smoking in the first trimester and subdivided according to whether the mother continued at the same level, reduced, or quit by the second trimester. Birthweights were adjusted statistically for extraneous variables. As expected, second and third trimester cigarette use was associated with birthweight, but so too was cigarette use during the first trimester, and the effect of quitting varied significantly with the level of first trimester smoking. Among moderate and light smokers, who comprise the majority of smokers, quitting before the second trimester is associated with heavier infants. However, infants of heavy smokers who quit by the second trimester did not weigh significantly more than infants of mothers who continued to smoke heavily throughout pregnancy, and weighed significantly less than infants of nonsmokers or other smokers who quit. Thus, quitting by the end of the first trimester may not completely negate the effect of heavy first trimester smoking, and the adaptive value of qutting is unequal among different levels of first trimester smoking. Further research on prenatal growth should take cigarette smoking in all trimesters into account. © 1994 Wiley-Liss, Inc.

Relationship between population density and rates of injury mortality in New York State (exclusive of New York City), 1978-1982.

The relationship between population density and rates of mortality from unintentional and intentional injuries is examined using mortality data from New York State (exclusive of New York City), 1978-1982. Records for 26,118 individuals with an underlying cause of death due to injury were assigned to population density quintiles based on residence of decedent at time of death. Mortality rates for each population density quintile were examined separately by sex and for 11 causes of injury death. Overall, injury mortality is highest in the most rural and most urban populations. For both sexes, there is an inverse relationship between mortality from motor vehicle incidents and population density, and a positive relationship between homicide and population density. Male mortality from unintentional poisoning deaths shows a positive relationship with population density. Male mortality from fires shows a U-shaped relatinship with population density, with the highest mortality in the most rural and most urban populations. The relationships observed here between injury mortality and population density are most likely due to concomitant variation with aspects of the physical and cultural environments, such as socioeconomic status, ethnicity, and individual risk behaviors.

Sex differentials in unintentional injury mortality in relation to age at death.

Sex differentials in unintentional injury mortality were examined using death records for New York State residents that died as the result of an unintentional injury between the years 1984 and 1988 (n = 22,547). Male/female ratios were computed for nine age groups and for the four leading causes of unintentional injury death: motor vehicle incidents, falls, fire, and drowning. Male mortality is significantly higher than female mortality for all causes (P < 0.05). The age-specific pattern of mortality varies among the four causes of death examined, but the age-specific pattern of male/female mortality ratios is consistent among all four causes. All four leading causes of unintentional injury death show a peak in relative male risk in young adulthood (usually between 15 and 24 years of age). This common peak may reflect increased risk associated behaviors of young males, including alcohol. Two causes of death, motor vehicle incidents and fire, show a second peak in relative male risk among the elderly (75+ years), perhaps reflecting sex differences in alcohol and tobacco use.