Original Research Article

Assessing the Relationship Between Fluctuating Asymmetry and Cause ofDeath in Skeletal Remains: A Test of the Developmental Origins of Health andDisease Hypothesis

KATHERINE E. WEISENSEE*Department of Sociology and Anthropology, Clemson University, Clemson, South Carolina 29634

Objectives: This study examines the relationship between craniofacial fluctuating asymmetry and cause of death inan identified skeletal collection. This study tests the Developmental Origins of Health and Disease hypothesis usingfluctuating asymmetry as the measure of developmental instability.

Methods: The skeletal sample used in this study comes from Lisbon, Portugal, and individuals in the sample wereborn between 1806 and 1935. This represents a period during which Lisbon was beginning to undergo the modernhealth transition, in which mortality from infectious disease began to decline while mortality from degenerative dis-eases began to increase. Approximately equal numbers of individuals in the sample died from infectious diseases, suchas tuberculosis, and from degenerative diseases. Fluctuating asymmetry is examined using three-dimensional land-mark data collected from 392 individuals with documented causes of death. Landmark data may provide a more robustmeasure of fluctuating asymmetry, although it has not often been used in studies of fluctuating asymmetry in humanskeletal samples.

Results: The results of the study show that individuals who died from degenerative diseases have higher rates offluctuating asymmetry compared to individuals who died from infectious diseases. Males also exhibit higher rates offluctuating asymmetry compared to females.

Conclusions: The results of this study confirm earlier findings that early development has a significant impacton adult health outcomes. Furthermore, the results suggest that fluctuating asymmetry in skeletal samples may offera means of testing the Developmental Origins of Health and Disease hypothesis. Am. J. Hum. Biol. 25:411–417, 2013.VC 2013 Wiley Periodicals, Inc.

INTRODUCTION

Several studies have demonstrated that early develop-mental instability caused by environmental stress haslong-term effects on adult health outcomes (Barker et al.,2009; Syddall et al., 2005). This relationship has beentermed the Developmental Origins of Health and Disease(DOHaD) hypothesis. While many studies of the DOHaDhypothesis have focused on birth weight or weight at oneyear as a measure of developmental instability, othershave shown how fluctuating asymmetry (FA) can also pro-vide an accurate measure of developmental instability(Kieser et al., 1997; M�ller, 2006). However, the relation-ship between FA and the DOHaD hypothesis has not beenthoroughly studied. The current study examines FA incraniofacial morphology as a measure of developmentalinstability and its relationship to adult health outcomes.This study uses an identified skeletal sample with knowncauses of death to evaluate the usefulness of variation inthe level of FA as a predictor of adult health outcomes.The skeletal sample used in this study comes from Lisbon,Portugal, and individuals in the sample have birth datesranging from the mid 19th to early 20th centuries. Thesample comes from a population in the early stages of themodern health transition. In general, the early modernhealth transition is characterized as a period duringwhich mortality rates from infectious diseases began todecline while chronic diseases became more common(Gage, 2005). There is also a corresponding shift fromhigh to low childhood and infant mortality rates and over-all increased adult longevity during this period. The Lis-bon sample is unique in that many of the causes of deathrepresented, such as cholera and tuberculosis, do not

occur commonly today, and therefore the sample capturesthe transitional period. In this study, differential levels ofFA by cause of death are examined in order to test theDOHaD hypothesis.

Fluctuating asymmetry

Several studies have demonstrated that FA can be usedas an indicator of developmental instability (DeLeon,2007; Kieser et al., 1997; Kohn and Bennett, 1986). Bilat-eral structures have a tendency to develop symmetricallyunder ideal conditions. However, under conditions inwhich an individual is stressed during growth, subtledeviations from bilateral symmetry emerge. FA provides ameans for measuring developmental instability and hasbeen correlated with environmental and nutritionalstresses in several human populations and using animalmodels (Livshits et al., 1988; Livshits and Kobyliansky,1991). Higher levels of FA have been associated withincreased nutritional stress, genetic homozygosity, andexposure to extreme temperatures (Schaefer et al., 2006;Wells et al., 2006). FA provides a means of evaluating de-velopmental instability because it offers the opportunityto measure two independent replicates (i.e., right andleft sides) under conditions in which the genotype and

Correspondence to: Katherine E. Weisensee, Department of Sociologyand Anthropology, 132 Brackett Hall, Clemson University, Clemson, SC29634. E-mail: kweisen@clemson.edu

Received 3 July 2012; Revision received 6 January 2013; Accepted 22February 2013

DOI: 10.1002/ajhb.22390Published online 4 April 2013 in Wiley Online Library

(wileyonlinelibrary.com).

VC 2013 Wiley Periodicals, Inc.

AMERICAN JOURNAL OF HUMAN BIOLOGY 25:411–417 (2013)

environmental conditions are held constant. Random fluc-tuations in the development of the right or left sides pro-vide a measure of an individual’s inability to maintaindevelopmental stability.

FA has been used to assess health, developmental sta-bility, attractiveness, and reproductive fitness in severalhuman studies. Milne et al. (2003) found that individualswith higher levels of FA had higher body mass indices(BMIs) and were more likely to suffer from two or moremedical conditions, compared to individuals with lowerrates of FA. Van Dongen et al. (2009) found that fetuseswith more severe abnormalities had higher levels of FA.The link between perceived attractiveness and FA hasbeen established in a number of studies; specifically,results suggest that males and females are more likely toscore individuals with lower FA rates as being moreattractive (Hume and Montgomerie, 2001). Finally, FAhas been linked with reproductive fitness, whereby indi-viduals with lower levels of FA tend to have more sexpartners and to be perceived by the opposite sex as beingmore reproductively fit (Jones et al., 2001).

While several studies of both humans and animals havefound a significant relationship between FA and develop-mental instability, others have reported inconsistentresults (Bjorksten et al., 2000; Black, 1980). Inconsistentresults have been reported in both laboratory and fieldstudies of FA, even though laboratory studies are able tocontrol for many more variables compared to field studies(Talloen et al., 2010; Vangestel and Lens, 2011). Meta-analysis of the relationship between FA and stress acrossa number of studies has found only a weak positive rela-tionship with a high degree of variability (Leung and For-bes, 1996). The weak positive relationship may be relatedto high levels of measurement error or an inconsistentdefinition of stress across different studies (Hoffmann andWoods, 2002). There is also a debate in the literature as towhether one trait or a few traits should be used to assessthe relationship between FA and stress (Van Dongen,1998; Whitlock, 1996) or if a combined index provides abetter measure (Leary and Aldendorf, 1989; Leung et al.,2000). Methods have also been proposed by Klingenbergand Monterio (2005) that use Procrustes distance calcu-lated using landmark data to compute a single FA scoreas a measure of asymmetry.

In spite of a lack of uniformity in the relationshipbetween FA and stress, examinations of FA in pasthuman populations have found that FA can provide anaccurate picture of health. Studies of archaeologicalskeletal material have demonstrated an increase in FArates in populations experiencing stressful environmen-tal conditions, including increased population sizes,famines, and the origins of agricultural production(DeLeon, 2007; Gawlikowska et al., 2007). FA has alsobeen associated with other stress markers in skeletalsamples, such as linear enamel hypoplasias and cribraorbitalia (Corruccini et al., 2005; DeLeon, 2007; Hooveret al. 2005). Studies of FA both in living people and inpast populations demonstrate that differences in thelevel of FA among individuals may allow for the assess-ment of developmental instability.

Developmental origins of health and disease (DOHaD)

While FA may provide a measure of developmentalinstability, other studies have associated developmental

instability with adult health outcomes and adult mortal-ity. In a large longitudinal study from England, Barkerand Osmond (1986) demonstrated that individuals withlow birth weights were more likely to develop cardiovas-cular diseases as adults. Several other studies haveshown that there is a relationship between early develop-ment and adult health outcomes. This relationship hasbeen termed the Developmental Origins of Health andDisease (DOHaD) hypothesis. It was initially termed theFetal Origins hypothesis; however, more recent researchhas demonstrated that the period of interest extendsbeyond the fetal period, throughout the growth phase ofdevelopment (Kuzawa et al., 2010). For example, individ-uals with low birth weights followed by increased weightgain in childhood are at an especially high risk for devel-oping cardiovascular disease and diabetes as adults. TheDOHaD hypothesis suggests that developmental instabil-ity during a critical growth period continues to impactindividuals throughout life and predicts adult healthoutcomes.

Nutritional cues set fetal and infant growth trajecto-ries and may predict future environmental conditions,such as food availability. The long-term detrimentalconsequences of undernutrition during early growthand development have been well-established using ani-mal models. Widdowson and McCance (1963) demon-strated that undernutrition in the early postnatalperiod (1–3 weeks) in rats led to long-term stunting ofgrowth regardless of starting a normal diet after theearly period, whereas rats experiencing undernutritionslightly later (3–6 weeks after birth) experiencedcatch-up growth after resuming a normal diet. Langleyand Jackson (1994) showed that protein deficiency inpregnant rats led to functional changes in their off-spring, regardless of the mothers’ resuming a normaldiet after birth. These functional changes includedincreased rates of hypertension, increased rates of car-diovascular disease, and altered life spans.

The thrifty phenotype hypothesis is an example of theDOHaD hypothesis (Gluckman and Hanson, 2006). Thethrifty phenotype hypothesis states that if a fetus experi-ences nutritional stress in the intrauterine environment,its biology will be altered to deal with stress, and this willpotentially contribute to disease risk in later life. Thethrifty phenotype hypothesis was proposed as a develop-mental model for understanding the relationship betweenlow birth weight and adult disease risk. This model pro-poses that if the intrauterine environment is poor, the fe-tus will grow more slowly, and its developmentaltrajectory will be set toward conditions under which insu-lin resistance and other physiological changes associatedwith deprivation will be favored. The role played by themismatch between the fetal/neonatal environment andthe postnatal environment in the risk of developing dis-ease later in life provides a means for understanding theorigins of adult health outcomes. Data from the DutchHunger Winter provide support for this hypothesis. Theoffspring of Dutch women exposed to famine conditionsearly in pregnancy experienced long-term consequenceson their adult disease risk, and the effects of famine condi-tions could even be seen in the subsequent generation(Roseboom et al., 2001).

Because the nutritional environment is such an impor-tant factor affecting survival, it is therefore not surprisingthat systems related to metabolism, growth, and stress

412 K.E. WEISENSEE

American Journal of Human Biology

responses should be differentially affected. The genotype-environment interaction allows the developing organismto modify its growth to predict the future environment;this has been termed predictive adaptive responses(PARs). The presence of PARs allows an individual torespond to transient environmental change across one ormore generations, as opposed to plastic changes thatoccur in one lifetime or long-term adaptations related toDarwinian selection (Gluckman and Hanson, 2006).Gluckman and Hanson (2006) propose that when thehuman life span was relatively short, the environmentalmismatch of inappropriate PARs would not have been sig-nificant, but as adult life span has increased during thepostindustrial era, PARs have come to play an increas-ingly important role.

The link between early development and adult diseaseoutcomes suggests that growth during the fetal and post-natal period can be used as a predictor of future environ-mental conditions, and that if there is a mismatchbetween early growth and later environmental conditionsthere may be an increased risk of developing certain dis-eases in adulthood (Kuzawa, 2007). A mismatch occurswhen the early growth trajectory is set at a pace indica-tive of an environment in which undernutrition is likely,yet food resources prove to be plentiful in the adult envi-ronment. Mismatches between the early growth periodand environmental conditions during adulthood may havebecome particularly pronounced in human populationsover the past 200 years. During this period urbanizationand industrialization have caused significant changes inenvironmental conditions in a very short span of time.The sharp rise in cardiovascular disease in industrializedcountries beginning in the early 1900s may provide evi-dence of the environmental mismatch. Following severaldecades of high rates of coronary heart disease these num-bers have begun to fall in recent years; for example, in theUnited States coronary heart disease, currently the lead-ing cause of death, will soon be replaced by neoplastic dis-eases as the most common cause of death (Fall andSachdev, 2006). Given the rise in obesity rates, the declinein coronary heart disease in industrialized nations hasapparently not been the result of changes in adult life-styles and may be an indication of a better match betweenearly growth and the adult environment. Similarly, a risein coronary heart disease has more recently been docu-mented in developing countries, such as China and India(Godfrey, 2006). This pattern suggests that during periodsof rapid transition there is a mismatch in the rate of earlygrowth and the subsequent adult environment which mayresult in increases in chronic or degenerative diseases inadults.

This study explores the relationship between FA in cra-niofacial morphology and cause of death in the New Lis-bon skeletal collection using three-dimensional landmarkcoordinate data collected from individuals with knowncauses of death. A common method of evaluating develop-mental instability from skeletal remains uses FA. Thisstudy explores the effect of the modern health transitionon the levels of FA in craniofacial morphology, specificallyexamining whether factors that affect craniofacial asym-metry and differential growth and development in earlylife are related to causes of death in adulthood. This studyis unique in that the causes of death are well documented;such information is not commonly known in skeletalsamples. Moreover, this study uses three-dimensional

landmark-based methods to evaluate FA in the craniofa-cial skeleton; these methods have not been commonlyused in studies of FA in human skeletal samples. In addi-tion, due to the unique nature of the sample, the linkbetween FA and cause of death may be explored; the tran-sitional period under study presents an opportunity toobserve diseases that are not generally seen in modernpopulations and which cannot normally be diagnosedfrom skeletal material alone.

MATERIALS

The sample used in this study is from the New Lisbonskeletal collection. This collection contains the skeletalremains of identified individuals who were born in Portu-gal during the 19th and 20th centuries and were interredin Lisbon’s three main cemeteries. For most individualswho died before 1959, demographic information is avail-able and it includes cause of death, date of death, age, sex,place of birth, and occupation. For individuals who diedafter 1959, the demographic information is more limited;for instance, their causes of death are generally unavail-able. Therefore, the birth years of the individuals used inthe current sample range from 1806 to 1935. A total of392 individuals were included in the sample. The cause ofdeath and other demographic information is taken fromthe death certificates and cemetery records of the individ-uals included in the sample. Three-dimensional craniallandmarks were collected using a Microscribe 3DX digit-izer. Thirty-one landmarks were chosen for the currentanalysis because they represent mainly type I landmarks(see Figs. 1 and 2). Type I landmarks are those defined bythe intersection of sutures or other distinct junctions thatcan be found with a high degree of repeatability. By usingmainly type I landmarks, this study sought to minimizethe types of measurement error to which asymmetry anal-yses are often susceptible. The 392 individuals sampledhad known causes of death that were classified as degen-erative, infectious, neoplastic, or other diseases (afterGage, 2005; Preston et al., 1972); specific causes of deathgrouped within each category are presented in Table 1.The other category includes all causes of death that couldnot be classified as degenerative, infectious, or neoplastic.According to the disease classification outlined by Gage(2005) neoplastic diseases can be classified generally asdegenerative diseases, however several of the cancerspresent in the sample may be associated with infectiousagents, therefore neoplastic diseases were kept as a sepa-rate category for the purposes of this study. The associa-tion between level of FA and sex was evaluated. The totalnumber of individuals included in each cause of death(COD) group along with other demographic characteris-tics is shown in Table 1.

METHODS

The three-dimensional landmark coordinate data of allthe specimens were aligned into a common coordinate sys-tem using a generalized Procrustes least-squares super-imposition. A generalized Procrustes superimpositionremoves non-shape related variation due to the relativelocation, orientation, and size of the specimens, thusallowing for an examination of shape variation betweenspecimens. For objects with bilateral symmetry, suchas the crania, shape variation can be portioned into

FA AND CAUSE OF DEATH IN SKELETONS 413

American Journal of Human Biology

symmetric and asymmetric components. This is accom-plished using by mirroring the original landmark configu-ration and the relabeled reflections of the landmarkconfigurations. By relabeling and reflecting the originallandmark configuration, an object with perfect symmetryis constructed. The Procrustes distance between the origi-nal landmark configuration and the relabeled reflectionprovides a means of measuring overall FA in individuals.In this study, this method was used to calculate a single,multivariate estimate of the deviation from symmetry—an FA score—for each specimen. Landmark-based meth-ods of analyzing FA have two advantages. First, theeffects of FA can be explored at the levels of size, shape,and individual landmark. These differing levels of analy-sis allow for the more precise characterization of FA. Sec-ond, landmark-based methods are inherentlymultivariate; therefore, the FA scores calculated usingthis method have the potential to give a more robust indexof individual levels of development instability, comparedto single trait comparisons (Klingenberg and Monteiro,2005; Klingenberg et al., 2002; Palmer and Strobeck,2003).

Measurement error can have serious consequences forevaluating FA (Palmer and Strobeck, 2003). To evaluatemeasurement error in the sample, five crania were digi-tized five times with at least one day between each datacollection period. The significance of asymmetry relativeto measurement error was tested using the individual xside ANOVA procedure within the different cause of deathgroups and among all groups combined (Palmer and Stro-beck, 1986). Levene’s test for homogeneity of variance wasused to test for heterogeneity of variance among the causeof death groups and by sex. Kurtosis and skewness wasalso examined in order to determine whether antisymme-try was present in the data.

ANOVA was used to examine the level of FA within thesample and between cause of death (COD) groups and bysex. The individual Mahalanobis FA scores were calcu-lated in MorphoJ (Klingenberg, 2011) using methodsproposed by Klingenberg and Monteiro (2005). The Maha-lanobis FA score is scaled relative to the variation of

asymmetry in the sample and is therefore more robust foraccounting for non-normality in the sample. A Tukey’smultiple comparison test was used to determine any sig-nificant differences between the groups.

Fig. 1. Superior view of landmarks used in the analysis.

Fig. 2. Lateral view of landmarks used in the analysis. Abbrevia-tions: alar, most lateral point on nasal aperture; ast, point wherelambdoid; parietomastoid, and occipitomastoid sutures meet; bas,point where anterior margin of foramen magnum intersects midsagit-tal plane; brg, point where coronal and sagittal sutures intersect; dac,point of intersection of frontolacrimal and lacrimomaxillary sutures;ect, most lateral point on orbital margin; flac, most medial point on fo-ramen lacerum; fob, most lateral point on margin of foramen mag-num; hor, point of attachment of vomer and sphenoid bones; kro,most posterior extent of sphenoparietal suture; lam, point where sag-ittal and lambdoid sutures meet; nas, point of intersection of naso-frontal suture and midsagittal plane; nlhi, most inferior point onnasal aperture; ops, point where posterior margin of foramen mag-num is intersected by midsagittal plane; por, point superior to exter-nal auditory meatus; pro, most anterior point on alveolar border ofmaxilla between central incisors in midsagittal plane; ste, point oncranial base at intersection of sphenoid and temporal bones nearspine of sphenoid bone; zygoo, intersection of zygomaticomaxillarysuture at orbital margin; zyti, most inferior point on zygomaticotem-poral suture.

414 K.E. WEISENSEE

American Journal of Human Biology

RESULTS

The results from the measurement error test are seenin Table 2. The ANOVA analysis shows that the variancein FA was slightly more than seven times the variancedue to measurement error. The same analysis wasrepeated for each cause of group separately and theresults were nearly identical. The variance of the datawas also examined using Levene’s test for homogeneity.The test was not significant for either the cause of deathgroups or by sex, this indicates that the variance betweengroups does not differ significantly. The kurtosis value forthe Mahalanobis FA score is 2.47. These results suggestthat antisymmetry, or a bimodal distribution of the data,is not a significant component.

The descriptive statistics for the Mahalanobis FA scoresfor the COD groups and sex are shown in Table 3. Theresults of the ANOVA with FA score as the dependentvariable and COD and sex as between-subject factorsshowed significant main effects of COD group as seen in

Table 4. Tukey’s multiple comparison test demonstratedthat individuals who died from degenerative diseases hadsignificantly higher rates of FA compared to individualswho died from infectious diseases. The results of theTukey’s test are shown in Table 5 and Figure 3. There wasno significant difference by sex; however males yieldedsomewhat higher FA scores than did females.

DISCUSSION

The results of the study indicate that individuals whodied from degenerative diseases had significantly higherlevels of FA than individuals who died from infectious dis-eases. This suggests that FA provides a means of meas-uring developmental instability that may be linked to theDOHaD hypothesis. The DOHaD hypothesis proposesthat developmental instability during the fetal and post-natal period has long-term consequences on adult healthoutcomes. While evidence of developmental instability inthe DOHaD hypothesis has previously been measuredmainly using birth weight, this study demonstrates thepotential for craniofacial FA from skeletal remains to beused as an additional measure of developmental instabil-ity. The results also show that while males have slightly,but not significantly, higher rates of FA in this sample.

TABLE 1. Characteristics of cause of death groups in study sample

CODa groups NMean

birth yearMean

death yearMean

age (years) Male/female Specific causes of death

Degenerative diseases 168 1881 6 19 1945 6 14 64 6 16 81/87 Vascular lesions, heart disease, arterioscle-rosis, other diseases of the heart, diseasesof the arteries and circulatory system, ne-phritis, uremia, renal sclerosis, cirrhosisof liver, ulcers of stomach and duodenum,diabetes

Infectious diseases 126 1896 6 24 1938 6 12 42 6 21 73/53 Tuberculosis, syphilis, typhoid, leprosy,meningitis, influenza, pneumonia, bron-chitis, septicemia.

Neoplastic diseases 59 1882 6 18 1944 6 12 63 6 13 27/32 Stomach cancer, uterine cancer, intestinalcancer, lung cancer, brain tumors, pros-tate cancer, rectal cancer, oral cancer,breast cancer, kidney tumors, malignantcancer.

Other diseases 39 1874 6 22 1944 6 9 70 6 18 11/28 Old age, senility, sudden death, ruptureduterus, gunshot, hanging, carbonic acidintoxication.

aCause of death.

TABLE 3. Summary statistics by COD groups and sex

FA score Mean S.D.

COD groupa

Degenerative 6.39 1.37Infectious 6.00 1.18Neoplastic 6.02 0.99Other 6.43 1.51

SexMales 6.27 1.32Females 6.17 1.25

aCause of death.

TABLE 2. Analysis of variance for measurement error calculation

Source df SS MS F

Individual 18,216 1.32265936 0.0000726098 4.72a

Side 40 0.02492935 0.0006232338 40.51a

Individual 3 side 15,840 0.24371027 0.0000153857 5.60a

Measurement error 1548 0.00424989 0.0000027454

aP<0.001

TABLE 4. Tests of main effects

FA Score Sum of squares F P-value

COD groupa 16.130519 3.2964 0.0206Sex 1.894465 1.1615 ns

aCause of death.

TABLE 5. Tukey’s multiple comparison test

Degenerative Infectious Neoplastic Other

DegenerativeInfectious 2.677 (P 5 0.0386)Neoplastic 1.917 (P 5 0.222) 0.167 (P 5 0.998)Other 0.296 (P 5 0.990) 1.991 (P 5 0.193) 1.657 (P 5 0.348)

FA AND CAUSE OF DEATH IN SKELETONS 415

American Journal of Human Biology

The results also show high levels of FA in the other dis-ease category, with a correspondingly high standard devi-ation. This is not surprising given the wide variety ofcauses of death included in this category, from old age tomaternal deaths related to childbirth. Some of thesecauses of death may be related to increased developmen-tal instability, while others are not.

The sample used in this study examines a population inthe transitional period just prior to the postindustrial era.The unique nature of the sample allows a comparison of apre-industrial model of mortality, in which mortality wasstrongly affected by high rates of infectious disease, to amodern mortality model in which degenerative diseasesplay a much more significant role. This is reflected in themean year of death of individuals that died from infec-tious diseases as 1938, while the mean year of death ofindividuals that died from degenerative diseases was1945. Before the modern health transition, the infantmortality rate was extremely high. Individuals whosurvived this large selection pressure presumably experi-enced low levels of developmental stress and werewell-suited to their environment. The arrival of the mod-ern health transition brought with it major declines inchildhood and infant mortality (Gage, 2005). Beginning inthe mid to late 1800s, childhood and infant mortalitydeclines resulted in increased survivorship of individualswho would have died in previous generations. Presum-ably, these infants and children were not living underideal conditions during the period of transition, but theywere surviving. Therefore, individuals who might other-wise have died were now surviving yet carrying themarkers of a stressful early environment into adulthoodalong with an increased risk of developing degenerativediseases later in life. Alter and Riley (1989) point out thatwithin any cohort that experiences a decline in mortality,there may be a subsequent increase in morbidity becausethose individuals who would have died in previous cohortswould now survive with more ill-health consequences. Anincrease in morbidity is not necessarily indicative of a

decrease in health, only that early death is replaced bylater sickness. Moreover, the impact of the intergenera-tional effect could continue to impact the population,whereby following the transitional period individuals con-tinue to be affected by their mother’s and grandmother’sprenatal environment. Researchers have pointed out thatfollowing the modern health transition in countriesaround the world, there is an increase in cardiovasculardiseases which later levels off and begins to decline (Gage,2005). A possible explanation may be that populationsthat experienced urbanization earlier were better adaptedto environments with high rates of infectious diseases. Inthis study, individuals who died from infectious diseaseshave lower rates of FA, which may suggest that theseindividuals were not as stressed during development.Individuals who died from degenerative diseases, how-ever, have higher rates of FA, suggesting greater stressduring development and therefore a greater impact of theenvironmental mismatch.

The relationship between FA and adult health outcomesis difficult to measure throughout human history becausemost degenerative diseases do not leave diagnosablemarkers on the bone; therefore, it is difficult to determinethe rates of these types of diseases in earlier human popu-lations. Armelagos et al. (2009) discuss the support thatlinear enamel hypoplasias provide for the DOHaD hy-pothesis. They cite several studies that demonstrate thatindividuals with enamel defects have much higher ratesof early mortality compared to individuals without enameldefects. Bioarchaeological samples of past populationssuggest that individuals who survive infancy and child-hood show fewer signs of stress. It may be that if mortalityrates shift and those individuals who might otherwisehave died in the past were now surviving, stressful earlyevents would mark these survivors and subsequently leadto increased risk of developing degenerative diseases inadulthood.

Most studies of the DOHaD hypothesis have focused onrecent, postindustrial populations with low rates of mor-tality from conditions that were at one time very common,effectively excluding from their analyses diseases such astuberculosis. The results of this study suggest that byincluding those diseases we may be better able to assessthe impact of early environmental insults and the adapt-ive nature of developmental stability and adult healthoutcomes. Specifically, Gluckman and Hanson (2006) com-pared the mortality models of low population density for-agers with postindustrial populations; however theresults of the current study suggest that the impact of themortality model characterized by high rates of infectiousdisease may have been influential in human history.Many human populations lived under high rates of infec-tious disease for many generations and it likely impactedtheir adaptive strategies.

In summary, this study has demonstrated that FA ofthe craniofacial skeleton may be used to examine theDOHaD hypothesis in skeletal samples. The results ofthis analysis show a difference in the levels of FA by causeof death. As predicted by the DOHaD hypothesis, individ-uals who died from degenerative diseases had higherrates of FA compared to individuals who died from infec-tious diseases. While it is difficult to diagnosis degenera-tive diseases in skeletal samples without death records,the unique nature of the New Lisbon skeletal sample ena-bles an examination of the relationship between FA and

Fig. 3. Means and standard errors of Mahalanobis FA scores bycause of death groups; *P< 0.05 for Tukey’s multiple pairwise com-parison test.

416 K.E. WEISENSEE

American Journal of Human Biology

adult disease outcomes. The time period represented inthe Lisbon sample also allows for the examination of apopulation during the modern health transition whenpopulations moved from a mortality model characterizedby higher rates of infectious disease to the modern mortal-ity model in which degenerative diseases play a more sig-nificant role. The general reduction in childhood andinfant mortality during the modern health transition sug-gests that early developmental stress ultimately declines;however, the relationship between this transitional peri-od’s environmental mismatch and adult health outcomesmay also prove to be significant.

LITERATURE CITED

Alter G, Riley JC. 1989. Frailty, sickness, and death: models of morbidityand mortality in historical populations. Popul Stud (Camb) 43:25–45.

Armelagos GJ, Goodman AH, Harper KN, Blakey ML. 2009. Enamel hypo-plasia and early mortality: bioarchaeological support for the Barker hy-pothesis. Evol Anthropol 18:261–271.

Barker DJP, Osmond C. 1986. Infant mortality, childhood nutrition, andischaemic heart disease in England and Wales. Lancet 8489:1077–1081.

Barker DJP, Osmond C, Kajantie E, Eriksson JG. 2009. Growth andchronic disease: findings in the Helsinki Birth Cohort. Ann Hum Biol36:445–458.

Bjorksten TA, Fowler K, Pomiankowski A. 2000. What does sexual traitFA tell us about stress? Trends Ecol Evol 15:163–166.

Black TK. 1980. An exception to the apparent relationship between stressand fluctuating dental asymmetry. J Dent Res 59:1168–1169.

Corruccini RS, Townsend GC, Schwerdt W. 2005. Correspondence betweenenamel hypoplasia and odontometric bilateral asymmetry in Australiantwins. Am J Phys Anthropol 126:177–182.

DeLeon VB. 2007. Fluctuating asymmetry and stress in a medieval Nu-bian population. Am J Phys Anthropol 132:520–534.

Fall CHD, Sachdev HS. 2006. Developmental origins of health and dis-ease: implications for developing countries. In: Gluckman P, Hanson M,editors. Developmental origins of health and disease. Cambridge: Cam-bridge University Press. p 456–471.

Gage TB. 2005. Are modern environments really bad for us: revisiting the de-mographic and epidemiologic transitions. Am J Phys Anthropol 48:96–117.

Gawlikowska A, Szczurowski J, Czerwi�nski F, Miklaszewska D, AdamiecE, DzieR ciołowska E. 2007. The fluctuating asymmetry of mediaeval andmodern human skulls. Homo 58:159–172.

Gluckman PD, Hanson MA. 2006. The conceptual basis for the develop-mental origins of health and disease. In: Gluckman P, Hanson M, edi-tors. Developmental origins of health and disease. Cambridge:Cambridge University Press. p 33–50.

Godfrey K. 2006. The “developmental origins” hypothesis: epidemiology.In: Gluckman P, Hanson M, editors. Developmental origins of healthand disease. Cambridge: Cambridge University Press. p 6–32.

Hoover KC, Corruccini RS, Bonioli L, Macchiarelli R. 2005. Exploring therelationship between hypoplasia and odontometric asymmetry in IsolaSacra, an Imperial Roman necropolis. Am J Hum Biol 17:752–764.

Hume DK, Montgomerie R. 2001. Facial attractiveness signals differentaspects of “quality” in women and men. Evol Hum Behav 22:93–122.

Jones BC, Little AC, Penton-Voak IS, Tiddeman BP, Burt DM, Perrett DI.2001. Facial symmetry and judgements of apparent health: support fora “good genes” explanation of the attractiveness–symmetry relationship.Evol Hum Behav 22:417–429.

Kieser JA, Groeneveld HT, da Silva PC. 1997. Dental asymmetry, mater-nal obesity, and smoking. Am J Phys Anthropol 102:133–139.

Klingenberg CP. 2011. MorphoJ: an integrated software package for geo-metric morphometrics. Mol Ecol Resour 11:353–357.

Klingenberg CP, Barluegna M, Meyer A. 2002. Shape analysis of symmet-ric structures: quantifying variation among individuals and asymmetry.Evolution 56:1909–1920.

Klingenberg CP, Monteiro LR. 2005. Distances and directions in multidi-mensional shape spaces: implications for morphometric applications.Syst Biol 54:678–688.

Kohn LAP, Bennett KA. 1986. Fluctuating asymmetry in fetuses of dia-betic rhesus macaques. Am J Phys Anthropol 71:477–483.

Kuzawa CW. 2007. Developmental origins of life history: growth, produc-tivity, and reproduction. Am J Hum Biol 19:654–661.

Kuzawa CW, McDade TW, Adair LS, Lee N. 2010. Rapid weight gain afterbirth predicts life history and reproductive strategy in Filipino males.Proc Natl Acad Sci USA 107:16800–16805.

Langley SC, Jackson AA. 1994. Increased systolic blood pressure in adultrats induced by fetal exposure to maternal low protein diets. Clin Sci86:217–222.

Leung B, Forbes MR, Houle D. 2000. Fluctuating asymmetry as a bioindi-cator of stress: comparing efficacy of analyses involving multiple traits.Am Nat 155:101–115.

Livshits G, Davidi L, Kobyliansky E, Ben-Amitai D, Levi Y, Merlob P.1988. Decreased developmental stability as assessed by fluctuatingasymmetry of morphometric traits in preterm infants. Am J Med Genet29:793–805.

Livshits G, Kobyliansky E. 1991. Fluctuating asymmetry as a possiblemeasure of developmental homeostasis in humans: a review. Hum Biol63:441–466.

Milne BJ, Belsky J, Poulton R, Thomson WM, Caspi A, Kieser J. 2003.Fluctuating asymmetry and physical health among young adults. EvolHum Behav 24:53–63.

M�ller AP. 2006. A review of developmental instability, parasitism and dis-ease: infection, genetics and evolution. Infect Genet Evol 6:133–140.

Palmer AR, Strobeck C. 1986. Fluctuating asymmetry: measurement,analysis, patterns. Ann Rev Ecol Syst 17:391–421.

Palmer AR, Strobeck C. 2003. Fluctuating asymmetry analyses revisited.In: Polak M, editor. Developmental instability: causes and consequences.New York: Oxford University Press. p 279–319.

Preston SH, Keyfitz N, Schoen R. 1972. Causes of death: life tables fornational populations. New York: Seminar Press. 800 p.

Roseboom TJ, van der Meulen JH, Ravelli AC, Osmond C, Barker DJ,Bleker OP. 2001. Effects of prenatal exposure to the Dutch famine onadult disease in later life: an overview. Mol Cell Endocroinol 185:93–98.

Schaefer K, Lauc T, Mitteroecker P, Gunz P, Bookstein F. 2006. Dentalarch asymmetry in an isolated Adriatic community. Am J Phys Anthro-pol 129:132–142.

Syddall HE, Aihie-Sayer A, Simmonds SJ, Osmond C, Cox V, DennisonEM, Barker DJP, Cooper C. 2005. Birth weight, infant weight gain, andcause-specific mortality: the Hertfordshire Cohort Study. Am J Epide-miol 161:1074–1080.

Talloen W, Lens L, Van Dongen S, Adriaensen F, Matthysen E. 2010. Mildstress during development affects the phenotype of great tit Parus majornestlings: a challenge experiment. Biol J Linnean S 100:103–110.

Van Dongen S, Wijnaendts LCD, Ten Broek CMA, Galis F. 2009. Fluctuat-ing asymmetry does not consistently reflect severe developmental disor-ders in human fetuses. Evolution 63:1832–1844.

Vangestel C, Lens L. 2011. Does fluctuating asymmetry constitute a sensi-tive biomarker of nutritional stress in house sparrows (Passer domesti-cus)? Ecol Indicators 11:389–394.

Wells JC, Hallal PC, Manning JT, Victora CG. 2006. A trade-off betweenearly growth rate and fluctuating asymmetry in Brazilian boys. AnnHum Biol 33:112–124.

Widdowson EM, McCance RA. 1963. The effect of finite periods of under-nutrition at different ages on the composition and subsequent develop-ment of the rat. Proc R Soc Lond B Biol Sci 158:329–342.

FA AND CAUSE OF DEATH IN SKELETONS 417

American Journal of Human Biology