The Consequences of Linear Growth Stunting: Influenceon Body Composition Among Youth in the BolivianAmazon

Susan Tanner,1* William R. Leonard,2 Victoria Reyes-Garc�ıa,3 and TAPS Bolivia Study Team4,5

1Department of Anthropology, University of Georgia, Athens, GA 306022Department of Anthropology, Northwestern University, Evanston, IL 602083ICREA and Institut de Ciencia I Technologia Ambientals, Universitat Aut�onoma de Barcelona, 08193 Cerdanyoladel Valles, Spain4Brandeis University, Waltham, MA 024545Tsimane’ Amazonian Panel Study, Correo Central, San Borja, Beni, Bolivia

KEY WORDS growth and development; developmental plasticity; panel study; Bolivia

ABSTRACT Stunting, or linear growth retardation,has been documented in up to half of all children in ruralindigenous populations of South America. Stunting is wellunderstood as a signal of adverse conditions duringgrowth, and has been associated with developmentallyinduced modifications to body composition, including bodyfat and muscularity, that stem from early growth restric-tion. This article examines the relation between shortstature and three anthropometric indicators of body com-position during childhood and adolescence among a rural,indigenous population of forager-horticulturalists.Anthropometric data were collected annually from 483Tsimane’ youth, ages 2–10 years, in 13 communities inthe Beni region of Bolivia for 6 consecutive years (2002–2007). Baseline height-for-age was used to indicate stunt-ing (HAZ < 22.0) and compared with z-scores of body

mass index (BMI), sum of two skinfolds, and arm musclearea. Multilevel regression models indicate baselinestunting is associated with lower BMI z-scores (B 520.386; P < 0.001), body fatness (ZSkinfold, B 5 20.164;P < 0.001), and arm muscularity (AMAZ, B 5 20.580; P <0.001) in youth across a period of 6 years. When split bysex, there was a stronger relation between baseline stunt-ing and lower skinfold body fat scores among girls (B 520.244; P < 0.001) than boys (B 5 20.080; P 5 0.087). Incontrast, baseline stunting was associated with lower armmuscularity in both girls (B 5 20.498; P < 0.001) andboys (B 5 20.646; P < 0.001). The relation between lineargrowth restriction and indicators of body composition per-sist into adolescence, providing additional insight into theinfluence of adverse conditions during growth. Am J PhysAnthropol 153:92–102, 2014. VC 2013 Wiley Periodicals, Inc.

Stunting, or linear growth retardation (< 22.0 SD inheight-for-age Z-score), is common in many rural indige-nous populations of South America, and most recentestimates suggest that about 165 million children underthe age of five worldwide were growth stunted (de Oniset al., 2012; Black et al., 2013). The bulk of research ongrowth patterns in biological anthropology (Kuzawa,1998; Bogin, 1999; Walker et al., 2006; Bogin et al.,2007) draws on evolutionary life history, which explicitlygrapples with how trade-offs between overlappingdomains of growth, reproduction, and maintenance serveto enhance survival and reproduction in heterogeneousenvironments. In terms of understanding human pheno-typic variation, these developmental trade-offs connectearly growth deficits to long-term consequences includ-ing shorter stature (Stein et al., 2010; Sterling et al.,2012) and increased risk of developing metabolic dis-eases including cardiovascular disease and type 2 diabe-tes (Barker et al., 2002). While there is considerableevidence linking growth restriction in utero to adult out-comes, research on the associations between postnatalstunting and later-life body composition has producedmore variable results. Body fat reserves play importantroles in survival and reproduction and developmentaloscillations in body composition throughout childhoodand adolescence may have evolutionary consequences(Zafon, 2007; Adair, 2008; Wells, 2010). Therefore,

research examining how growth restriction or stuntingin childhood is related to body composition may deserveadditional attention.

Over the past several decades, research has shownlinks between early growth stunting and later body com-position but the direction of the association likely varieswith local context. Among some urbanizing populations,childhood stunting exists alongside rapidly rising ratesof obesity. This dual nutritional burden is hypothesizedto stem from developmental adaptations in which meta-bolic adjustments early in life result in increased risk ofmetabolic disease, including obesity, later in life (Frisan-cho, 2003; Leonard et al., 2009; Wilson et al., 2012). Sev-eral large studies have identified the co-occurrence ofboth stunting and overweight or obesity in children(Popkin et al., 1996; Fernald and Neufeld, 2007) and

Grant sponsor: National Science Foundation, Cultural and Physi-cal Anthropology Programs.

*Correspondence to: Susan Tanner, University of Georgia Athens,250A Baldwin Hall, Athens, GA 30602. E-mail: stanner1@uga.edu

Received 26 April 2013; accepted 10 October 2013

DOI: 10.1002/ajpa.22413Published online 7 November 2013 in Wiley Online Library

(wileyonlinelibrary.com).

� 2013 WILEY PERIODICALS, INC.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 153:92–102 (2014)

evidence for metabolic trade-offs associated with lineargrowth stunting is mounting (Hoffmann et al., 2000a,b).For example, in large surveys from Russia, Brazil,China, and the Republic of South Africa, Popkin et al.(1996) found that stunted children had a 1.7–7.8 timesincreased risk of being overweight than their non-stunted peers, depending on country and householdincome. Research among Mayans in urban Mexico hasalso demonstrated associations between child stature,lean body mass, and total energy expenditure (Wilsonet al., 2012) and that the effects of shortness and stunt-ing may persist across generations (Varela-Silva et al.,2009; Azcorra et al., 2013).

Several longitudinal or cohort studies have failed to findclear associations between stunting and weight or body fatduring childhood or adolescence. For example, in their lon-gitudinal research among South African urban children,Cameron et al. (2005) found children who were stunted at2 years of age were shorter and lighter than their non-stunted peers at 9 years of age, but they did not identifyassociations between stunting and later BMI, body compo-sition, or fat patterning. Research from Guatemala (Liet al., 2003) and Jamaica (Walker et al., 2002, 2007) hasalso suggested early growth restriction is associated withless body fat and muscularity over time. The results in thisliterature indicate that the consequences of linear growthrestriction are complex and likely depend on local environ-ment, diet, and developmental timing. Therefore, as sev-eral researchers have noted (Baker et al., 2009; Varela-Silva et al., 2012), longitudinal research in rural areas oflow-income countries is needed to better understand thefull range of consequences of growth trade-offs.

The goal of this article is to examine the short-run bio-logical consequences of linear growth stunting through-out childhood and adolescence in a foraging-horticulturalgroup with high rates of stunting and infection but lim-ited evidence for overnutrition. We explore the conse-quences of growth stunting on both muscularity andbody fatness using anthropometric data collected fromTsimane’ youth from 2002 to 2007. Prior work amongBolivia’s Tsimane’ has documented both frequent stunt-ing (Foster et al., 2005) and evidence for catch-upgrowth during childhood associated with household com-position and income (Godoy et al., 2010a). Additionally,previous research suggests immune activation is relatedto linear growth retardation and reduced body fatreserves (McDade et al., 2008). Given these associations,we first predict that tracking childhood growth will iden-tify associations between growth stunting and body com-position across a 6-year period. Second, we predict thatsex will modify the relations between growth stunting,muscularity, and body fat based on literature suggestingthat boys may be more sensitive to the nutritional envi-ronment than girls (Stinson, 1985; Kuzawa, 2007).Third, we predict that the association between stuntingand body composition will vary with age and, more spe-cifically, become stronger at later ages as growthbecomes canalized (Bogin, 1999). Finally, we predict thatthe association between stunting and body fatness willremain after controlling for maternal, household, andcommunity effects.

METHODS

Background

Research was conducted among Tsimane’, an indige-nous group of �15,000 people living in the Beni Depart-

ment of Bolivia. This study draws on data from theTsimane’ Amazonian Panel Study (TAPS) collected dur-ing 6 consecutive years of annual observations (2002–2007) in 13 communities (Leonard and Godoy, 2008).The communities ranged in size from 10 to 26 house-holds and the average household size was �6.0 people(range 1–16). Large households are often multigenera-tional and cross-cousin marriage is commonly practiced(Ellis, 1996). This results in a dense web of kinship andsocial networks in which households may be closelyrelated to several other households in each community.

Communities were initially selected to capture a spec-trum of variation in access to local markets and distanceto the Bolivian town of San Borja (mean 5 26.0 km; SD5 16.7). Across the panel study, the area experiencedimportant economic and political changes (Godoy et al.,2009) including expansion of cattle ranching, farmers,and logging firms in the region. The region saw the con-tinued expansion of public health services (Gurvenet al., 2007) including child vaccination. In 2006, theBolivian government distributed solar panels and visualequipment to encourage adult education night classes,which also brought limited electricity to some commun-ities (Godoy et al., 2009). Finally, in 2006, a regionalflood disrupted many households. Against the backdropof these changes, many aspects of daily life remainedrelatively continuous. From 2002 to 2007, the majorityof households relied primarily on swidden agriculturesupplemented with hunting, gathering, and a smallamount of purchased foods for subsistence. The primarysource of income for adult men was unskilled wage laborwith logging or cattle ranching or selling crop and forestgoods. Although roads and motorized canoes improvedaccessibility of many communities, roads often becomeimpassable during rainy periods and the more remotecommunities remain accessible only through canoe orwalking path.

Several recent books contain discussions of the ethno-graphic and historical background of the area and people(Huanca, 2008; Ringhofer, 2010), and here we provide abrief description of child growth and body composition.Previous research among Tsimane’ youth has docu-mented that growth stunting is common, with �45% ofchildren ages 2–10 years stunted (Foster et al., 2005)but little evidence of sex-based disparities in anthropo-metric indicators of short- or long-run nutritional statusor perceived health between girls and boys 2–13 years ofage (Godoy et al., 2006). Linear growth stunting appearsto begin around 6 months of age for both Tsimane’ boysand girls and continues through childhood (Foster et al.,2005). Frequent immune system activation, assessedthrough C-reactive protein, has been associated withlower skinfold measurements and arm circumference inchildren (McDade et al., 2005) and reduced lineargrowth over a three-month window (McDade et al.,2008). Gastrointestinal disease and parasitic infectionsare endemic, but short-term studies have not foundstrong associations between the infections and measuresof nutritional status including child height-for-age scores(Tanner et al., 2009). While malaria is not present in thearea, respiratory and gastrointestinal illnesses are com-mon. Infectious diseases, especially respiratory and gas-trointestinal infections, have been the most commoncause of death over the past 50 years (Gurven et al.,2007). Gurven and colleagues (2007) found that infantmortality rates are high in the first year of life anddecrease across early childhood. From 1950 to 1989,

LINEAR GROWTH STUNTING AND BODY COMPOSITION 93

American Journal of Physical Anthropology

they estimate that the average survivorship to age 5was 79.3%. In more recent years, Gurven (2012) foundthat infant mortality rates in the first year of life werehigh (�13%) but had declined from 1990 to 2002.

In order to examine catch-up growth, a recent paperexplored the growth trajectories of 2- to 7-year-old chil-dren in the 2002–2006 TAPS panel dataset (Godoy et al.,2010a). Using year-to-year change in age and sex-standardized height, this research found children whowere stunted at baseline exhibited faster growth thantheir non-stunted peers. Growth rates were not statisti-cally different between boys and girls: 25% of girls andonly 13% of boys who were stunted at baseline were nolonger stunted in 2006. Variables that were associatedwith slower linear growth were 1) living in a communityclose to a Bolivian town, 2) having additional youngersiblings, and 3) having higher household income levels.Here, we build on these findings by adding an additionalyear of data and by considering the association betweenlinear growth stunting and anthropometric indicators ofbody composition.

Recent research has suggested people throughout theAmazon are experiencing increasing rates of overweightand obesity (Benefice et al., 2007; Lourenco et al., 2008),although others have documented modest changes innutritional status (Piperata et al., 2011). In one of asmall number of studies to consider child body composi-tion in rural lowland Bolivia, Benefice et al. (2007) docu-mented the rates of overweight among Amerindianchildren living along the Beni River to be about 12%,with less than 1% of children obese according to BMI.Research among Tsimane’ has found evidence for gain inweight over the past decade among adults; however, thesignificance of this trend is elusive. Weight increaseshave not clearly been linked to any of the commonlyinvoked explanations including wealth, degree of cul-tural consonance, acculturation, or household measuresof food availability (Reyes-Garc�ıa et al. 2010; Zeng et al.,2012).

Methods

In 2002, a baseline survey was conducted duringMay–August to capture demographic, socio-economic,cultural, and health data along with anthropometricmeasurements of all participating individuals. The ini-tial survey attempted to recruit all individuals over 2years of age living in the 13 selected communities andfollow-up surveys were conducted annually. In this arti-cle, we limit our analyses to 483 children (225 girls and258 boys) who were between the ages of 2 and 10, inclu-sive, in 2002. Children were tracked for 6 consecutiveyears and, at the end of data presented here, the chil-dren were 7–15 years of age. The age bracket in thisstudy follows children through a period in which theyare sensitive to growth faltering due to infection andundernutrition (Bogin, 1999).

Age was reported by the child’s principal caretaker ineach year of the survey. When possible, parents reportedthe child’s exact birth date or showed a birth certificate.When the parent was not able to provide an exact age orbirthdate, surveyors estimated age in years based onsurveys with parents that focus on season of birth, birthorder within a household, and age in relation to locallyimportant events. The study protocol was reviewed bythe Institutional Review Boards of Northwestern andBrandeis Universities.

In addition to annual surveys with primary caretakerson a range of individual-level and household-level char-acteristics including demographics, income, and wealth(Leonard and Godoy, 2008), anthropometric measure-ments were taken annually on all participating individu-als following Lohman (1988). Stature was measuredwith a portable stadiometer. Body weight was measuredwith a Tanita scale and recorded to the nearest 0.2 kg.Mid-arm circumference was measured to the nearestmillimeter, using plastic tape measures. Triceps and sub-scapular skinfold thicknesses of were measured to thenearest 0.5 mm using Lange calipers. The technicalerror of measurement for anthropometry is not avail-able. Across the study period, measurements were takenby multiple surveyors who received training in anthropo-metric methods before data collection.

Anthropometric measures of body composition wereexamined through three derived anthropometric indices.Body Mass Index was calculated as weight (kg)/height(m)2. A skinfold measure of body fatness was calculatedusing the sum of tricep and subscapular skinfolds (Sumtwo skinfolds). Upper arm muscle area was calculatedfollowing Frisancho (2008). For analysis, height-for-ageand weight-for-height measures were standardized rela-tive to the National Center for Health Statistics (NCHS)percentiles as z-scores (i.e., HAZ, WHZ; Hamill et al.,1979). BMI z-scores were calculated for all children olderthan 5 years of age also, using the 2007 WHO growthreference based on NCHS data (Hamill et al., 1979; deOnis et al., 2007). Finally, age- and sex-standardized z-scores for the sum of two skinfolds (ZSkinfold) and armmuscle area z-score (ZAMA) were calculated according toFrisancho (2008), which provide growth references val-ues based on the NHANES III databases.

Data analysis

Analysis was conducted in Stata 10.0 in two stages.First, descriptive statistics were used to examine theassociation between stunting and body compositionusing a sub-sample of 321 youth who participated inboth 2002 and 2007. Here we consider if the youth wasstunted in 2002 (baseline stunting) and patterning inthe outcomes of interest: a) BMI, b) skinfold measure-ments, and c) arm muscle area. Because boys and girlsdiffer in body composition and rates of growth, descrip-tive data are presented for each sex separately. Second,panel linear multiple regressions with individual ran-dom effects, clustering by child, and robust standarderrors are used to estimate the association betweenbaseline stunting and measures of body composition.Random effects modeling allows the inclusion of timeinvariant characteristics (sex, baseline stunting) andexamines if differences across individuals are related tothe outcome variable of interest (Rabe-Hesketh andSkrondal, 2008). Of the 483 children measured at base-line, �49% were present during all 6 surveys and �21%were present in 5 survey years. About 6% were presentonly in the first year and 5% only in the first 2 years.The remaining �19% were present in 3 or 4 surveyyears. Children who left the sample during the studywere not more likely to be stunted than children whoremained in sample (Chi-square test of base stunting vs.eventual attrition 5 0.0695; P 5 0.792). There was alsono difference in stunting rates among children who par-ticipated in <4 years of anthropometric measurements

94 S. TANNER ET AL.

American Journal of Physical Anthropology

than those who participated in 4 or more years (Chi-square 5 0.617; P 5 0.432).

Other publications have discussed rounding error ordigit heaping (Godoy et al., 2010a) and there is evidencefor digit heaping in height and skinfold measurementsin this study. For example, digit heaping in skinfoldmeasurements occurred with values ending in 5, 6, or 7mm (they account for nearly 50% of values instead ofthe 30% that would be expected with a normal distribu-tion. The consequences of this measurement error arediscussed in the conclusions.

Byron (2003) found that Tsimane’ girls reach pubertyby 12–13 years of age and, although puberty is a con-founding factor in the associations examined in this arti-cle, we do not have accurate information on puberty.Therefore, in order to facilitate data presentation andminimize the confounding effects of puberty, sex and agecategories are used for descriptive data presentation: a)2–4 years at baseline (7–9 at follow-up), b) 5–7 at base-line (10–12 at follow-up), and c) 8–10 at baseline (13–15at follow-up). We eliminated six measurements fromyoung women who were either pregnant or lactating.

RESULTS

Measures of body compositionin 2002 and 2007

Anthropometric assessments of body composition showthat Tsimane’ children and adolescents are relativelyshort, but not experiencing acute nutritional stress(Tables 1 and 2). BMI scores among the sub-set of chil-dren and adolescents (n 5 321) that participated in both2002 (baseline) and the last year of data used here,2007, indicate that Tsimane’ youth are not experiencingacute malnutrition (mean BMI range from 16.4–19.7 forboys, 15.8–21.4 for girls). Similarly, the age- and sex-standardized BMI z-scores also indicate Tsimane’ youthare comparable to international reference populations(mean BMIz for all boys 5 0.26; girls 5 0.25 in 2007). In

2002, 11% of girls (11 of 86) and 25% of boys (26 of 104)over the age of 5 years had a BMI z-score above 1.0,indicating at risk of overweight (de Onis et al., 2010). By2007, the frequency of overweight remained constant ingirls (11%, 116 of 145) and declined for boys (11%, 19 of176). Thus, Tsimane’ youth were lean, but skinfoldmeasurements of body fatness were not so low as to indi-cate severe energy stress (mean ZSkinfold 5 20.40 forall girls and 21.0 for all boys in 2007) and muscularitylevels do not indicate acute protein malnutrition (ZAMA5 20.52 for girls and 20.93 for boys in 2007).

Overall, by 2007, youth who were stunted (S) at base-line had lower measures of adiposity (Fig. 1) and muscu-larity (Fig. 2) than their peers who were not stunted(NS) at baseline. Children who were stunted in 2002had lower average skinfold body fat measurements thantheir non-stunted peers in 2007, although the differenceis more apparent during the ages of 10 and 14 for bothboys and girls.

Similar to body fatness, Tsimane’ boys (Fig. 2a) andgirls (Fig. 2b) who were stunted at baseline had loweraverage values of upper arm muscularity in 2007 thantheir non-stunted peers. In contrast to skinfold measuresof body fatness, the difference in average arm muscular-ity between stunted and non-stunted children appearssimilar for all youth and consistent from ages 2 through10 years.

Youth who were stunted in both 2002 and 2007 hadsignificantly poorer indicators of body composition thantheir peers who were not stunted in either year or thosewho changed stunting status from 2002 to 2007 (Table3). Those who were never stunted had a higher averageBMI z-score than youth who were stunted in both years(ANOVA F 5 9.46, df 5 3; P < 0.001). When consideringarm muscularity, youth who were not stunted in either2002 or 2007 also had significantly higher AMA z-scorethan their peers who were stunted in both years (meanvalues of 20.43 vs. 21.23; F 5 22.06; df 5 3; P < 0.001).A similar pattern emerged with respect to skinfold thick-ness (ANOVA F 5 5.31, df 5 3; P 5 0.001).

TABLE 1. Descriptive statistics for selected anthropometric dimensions among Tsimane’ boys measured in 2002 and 2007

MeasuresAge at baseline

(years) Mean (SD) in 2002 Mean (SD) in 2007 D mean valuea

Stature (cm) 2–4 92.5 (8.68) 120.3 (6.86) 27.5 (3.21)5–7 109.8 (6.48) 133.7 (8.27) 24.1 (3.30)8–10 122.2 (7.39) 151.0 (9.42) 28.8 (4.71)Total 107.4 (14.68) 133.2 (15.13) 26.9 (4.23)

Weight (kg) 2–4 14.4 (2.40) 24.2 (3.40) 9.7 (1.71)5–7 19.1 (2.71) 31.4 (5.59) 11.8 (3.13)8–10 25.2 (4.13) 45.0 (8.83) 19.8 (5.89)Total 19.6 (5.55) 33.1 (10.66) 13.6 (5.88)

BMI (kg/m2) 2–4 16.7 (1.09) 16.8 (1.10) -0.1 (1.43)5–7 16.4 (1.28) 17.6 (1.46) 1.2 (1.08)8–10 16.7 (1.20) 19.7 (1.82) 3.0 (1.48)Total 16.6 (1.19) 17.9 (1.91) 1.3 (1.84)

Sum of two skinfolds (mm) 2–4 12.4 (2.66) 12.2 (2.85) -0.2 (3.61)5–7 11.6 (2.77) 13.7 (4.17) 2.2 (4.31)8–10 11.5 (2.35) 14.4 (3.85) 2.9 (3.53)Total 11.9 (2.62) 13.4 (3.74) 1.5 (4.04)

Arm muscle area (cm2) 2–4 14.6 (2.57) 19.6 (3.39) 4.9 (3.33)5–7 18.1 (3.10) 24.0 (5.50) 6.1 (4.77)8–10 20.1 (3.47) 33.0 (8.21) 12.9 (7.21)Total 17.4 (3.83) 25.2 (8.16) 7.8 (6.31)

Sample size in age groups: 2–4 (n 5 65); 5–7 (n 5 56); 8–10 (n 5 56).a Values not adjusted for exact time duration between measurements.

LINEAR GROWTH STUNTING AND BODY COMPOSITION 95

American Journal of Physical Anthropology

Longitudinal patterns during childhood and earlyadolescence

Although the comparisons above suggest that stuntingis associated with lower BMI, skinfold measurements,and arm muscularity over time, panel data allow an

examination of year-to-year patterns. In the entire sam-ple of youth, descriptive analyses of the panel data from2002 to 2007 suggest two points: a) stunting appears tobe associated with smaller skinfold measurements ingirls; and b) while all Tsimane’ children diverge from

Fig. 1. Average value of sum of triceps and subscapularskinfolds (mm) of Tsimane’ (a) boys and (b) girls in 2007 thatwere stunted (S) or non-stunted (NS) at baseline (2002). Fig. 2. Average value of arm muscle area (cm2) of Tsimane’

(a) boys and (b) girls in 2007 who were stunted (S) or non-stunted (NS) at baseline (2002).

TABLE 2. Descriptive statistics for selected anthropometric dimensions among Tsimane’ girls measured in 2002 and 2007

Measures Age at baseline (years) Mean (SD) in 2002 Mean (SD) in 2007 D mean valuea

Stature (cm) 2–4 90.7 (7.62) 117.7 (6.51) 28.0 (3.61)5–7 109.8 (6.48) 138.7 (8.05) 29.0 (3.26)8–10 124.0 (8.72) 148.0 (5.30) 25.6 (4.80)Total 106.1 (15.27) 133.9 (13.97) 27.8 (3.98)

Weight (kg) 2–4 13.6 (2.27) 22.8 (3.26) 9.6 (1.50)5–7 19.1 (2.71) 35.2 (7.62) 16.1 (5.38)8–10 26.1 (5.5) 46.6 (7.52) 21.5 (4.42)Total 18.8 (5.99) 33.3 (11.04) 14.8 (6.17)

BMI (kg/m2) 2–4 16.5 (1.19) 16.4 (1.06) 20.1 (1.17)5–7 15.8 (1.00) 18.1 (2.21) 2.4 (1.85)8–10 16.8 (1.60) 21.4 (2.57) 4.7 (1.95)Total 16.3 (1.31) 18.3 (2.70) 1.98 (2.48)

Sum of two skinfolds (mm) 2–4 14.5 (2.78) 13.7 (3.49) 20.3 (3.64)5–7 12.9 (2.44) 17.3 (5.17) 4.4 (5.04)8–10 14.6 (4.12) 26.9 (7.96) 12.6 (6.84)Total 13.9 (3.15) 18.3 (7.40) 4.5 (7.06)

Arm muscle area (cm2) 2–4 13.4 (2.40) 18.8 (2.76) 5.5 (2.73)5–7 16.9 (2.51) 25.5 (5.65) 8.6 (5.32)8–10 20.8 (3.87) 31.9 (5.90) 11.6 (5.23)Total 16.6 (4.08) 24.4 (6.93) 8.1 (5.02)

Sample size in age groups: 2–4 (n 5 56); 5–7 (n 5 54); 8–10 (n 5 34).a Values not adjusted for exact time duration between measurements,.

96 S. TANNER ET AL.

American Journal of Physical Anthropology

their US peers with respect to arm muscularity, stuntedchildren have lower scores of arm muscularity than theirnon-stunted peers.

Figure 3 shows the average yearly values of age- andsex- adjusted z-scores of skinfold measurements and armmuscularity for youth who were stunted and non-stunted at baseline in 2002. For boys, skinfold fatnessdoes not appear to differ between boys who were stuntedand non-stunted in 2002 but, by 2007, boys who werestunted at baseline had significantly lower age- and sex-adjusted skinfold measurements (NS 5 20.94, S 521.13; t 5 2.184 (181); P 5 0.030 (Fig. 3a). Similarly,

among girls, there is little difference in z-scores of bodyfatness between those who were stunted and non-stunted at baseline (NS 5 20.39, S 5 20.51; t 5 1.468(200); P 5 0.144). However, by 2007, girls who were notstunted at baseline have significantly higher averagevalues of skinfold body fat than girls who were stuntedat baseline (ZSkinfold: 20.28 vs. 20.58; t 5 3.542 (153);P 5 0.001). Finally, descriptive analysis of each yearshows that both boys (Fig. 3a) and girls (Fig. 3b) whowere stunted in 2002 had lower average z-scores for armmuscularity than their peers who were not stunted in2002 across all years in this sample.

Regression results

Tables 4 and 5 contain the main regression results.First, consistent with the graphical and descriptive anal-ysis presented above, stunting early in life is related tolater body composition for Tsimane’ youth. Across thestudy period, youth who were stunted in 2002 had lowerBMI z-scores (B 5 20.386; P < 0.001; Table 4, model 1)than their non-stunted peers. The association remainedconsistent when boys and girls were analyzed independ-ently and there was no evidence for interactions betweensex and baseline stunting. Because BMI z-scores aremost commonly used for children over 5 years of age,additional models using weight-for-height as the out-come variable for children who were 2–4 years old atbaseline showed that, among the younger cohort of thepanel, there was little relation between baseline stunt-ing and weight-for-height z-score during 2002–2007 (B 520.084; P 5 0.271, Table 4, Model 1).

The association between baseline stunting and bodyfatness and arm muscularity from 2002 to 2007 (Table5) was similar to the pattern observed with BMI. Youthwho were stunted at baseline had significantly lowerskinfold z-score (B 5 20.164; P < 0.001; Model 1) andarm muscle area z-score (B 5 20.580; P < 0.001; Model1) than their non-stunted peers after controlling forcovariates in the model. Additionally, results show evi-dence for a stronger relation between baseline stuntingand lower skinfold z-score among girls (B 5 20.244; P <0.001; Table 5, model 2) than among boys (B 5 20.080;P 5 0.087; model 3) when the models were run sepa-rately for boys and girls. An additional regression modelfor the full sample including an interaction term (male3 baseline stunting) indicated that the difference infavor of boys was statistically significance at conven-tional levels (full results not shown, P 5 0.026). For armmuscularity, the results observed with the full sampleremained consistent when boys and girls were analyzedseparately. Baseline stunting was associated with loweraverage scores of arm muscularity in girls (B 5 20.498;P < 0.001; Table 5, model 2) and boys (B 5 20.646; P <

TABLE 3. Comparison of body composition for Tsimane’ youth in 2007 with respect to stunting status in 2002 and 2007

N BMIZ ZSkinfold AMAZ

Not stunted (2002 or 2007) 160 0.43 (0.60) 20.61 (0.66) 20.43 (0.79)Stunted only 2007 16 0.29 (0.80) 20.79 (0.72) 20.56 (0.90)Stunted only 2002 50 0.05 (0.62) 20.96 (0.71) 20.91 (0.75)Stunted (both 2002 and 2007) 95 0.04 (0.69) 20.88 (0.64) 21.23 (0.78)

F 5 9.46; P < 0.001 F 5 5.31; P 5 0.001 F 5 22.06; P < 0.001

Values are mean (SD).

Fig. 3. Average yearly Z-score of Sum of Skinfolds (ZSkin-fold) and arm muscle area (AMAZ) for Tsimane’ youth whowere stunted (S) and non-stunted (NS) at baseline (in 2002).

LINEAR GROWTH STUNTING AND BODY COMPOSITION 97

American Journal of Physical Anthropology

0.001; model 3) than their non-stunted peers after con-trolling for confounding variables.

The association between stunting and anthropometricmeasurements of body composition differed across earlylife. Regression models that examined skinfold measure-ments and muscularity independently for both theyoungest ages (2–4 at baseline) and oldest ages (8–10 atbaseline) show the effects discussed above remained onlyfor the older children (Table 6, rows 1 and 2). Among theyoungest children, baseline stunting was associated withlower values of arm muscularity from 2002 to 2007, butthe associations with skinfold z-scores were not statisti-cally significant at conventional levels. In contrast,among the older age category, which is most likely toinclude youth in multiple stages of puberty, baselinestunting remained inversely associated with lower scoresof both arm muscularity and body fatness among girls.

Finally, the results presented above remain robust toadditional analysis that included maternal characteris-tics (mother’s age, height, weight, and years of com-pleted education) and household conditions (number ofchildren in the household, household income, and

wealth). Including these covariates slightly strengthenedthe relation between stunting and lower skinfold z-scores (Table 6, row 3). Adding additional covariates toaccount for community level effects also had only a smalleffect on the associations (Table 6, row 4). For example,children stunted at baseline had significantly lowerAMA z-scores than their non-stunted peers (B 5 20.580;P < 0.001) in the original model, but controlling formaternal, household, and community effects slightlyreduces the effect size (B 5 20.518; P < 0.001). House-hold wealth, number of children in the household, andmother’s height and weight were statistically associatedwith measures of body composition among youth (resultsnot shown), but the associations between stunting andbody composition remained robust after controlling forthese confounding effects.

DISCUSSION AND CONCLUSION

Indigenous groups throughout South America sufferdisproportionately high rates of infection and undernu-trition (Hurtado et al., 2005), but several studies also

TABLE 4. Random-effect panel linear regression models examining the relation between baseline stunting (2002) and the child’sBMI or weight-for-height z-score from 2002 to 2007

BMI z-score (ages 5–10) Weight for height z-score (ages 2–4)

All children Girls Boys All children Girls BoysExplanatory variable (1) (2) (3) (1) (2) (3)

Baseline stunting (yes) 20.386*** 20.467*** 20.310*** 20.084 20.192* 0.023(0.066) (0.104) (0.086) (0.076) (0.110) (0.104)

Age (years) 20.313*** 20.396*** 20.261*** 0.149*** 0.189*** 0.105(0.048) (0.062) (0.064) (0.054) (0.069) (0.084)

Age2 0.014*** 0.020*** 0.010*** 20.005 20.008 20.001(0.002) (0.003) (0.003) (0.005) (0.006) (0.007)

Male 0.106 0.013(0.067) (0.076)

Constant 2.015*** 2.203*** 2.037*** 20.128 20.184 20.046(0.238) (0.305) (0.330) (0.165) (0.205) (0.253)

Number of observations 1,418 646 772 952 465 487Number of individuals 287 130 157 200 98 102

Robust standard errors in parentheses.*P < 0.1, **P < 0.05, ***P < 0.01.

TABLE 5. Random-effect panel linear regression models examining the relation between baseline stunting (2002) and the measuresof body fatness (ZSkinfold) and arm muscularity (AMAZ) from 2002 to 2007 among Tsimane’ youth

Skinfolds z-score Arm muscle area z-score

All children Girls Boys All children Girls BoysExplanatory variable (1) (2) (3) (1) (2) (3)

Baseline stunting (yes) 20.164*** 20.244*** 20.080* 20.580*** 20.498*** 20.646***(0.038) (0.058) (0.047) (0.057) (0.080) (0.077)

Age (years) 20.011 20.148*** 20.097*** 20.206*** 20.202*** 20.220***(0.028) (0.040) (0.033) (0.035) (0.052) (0.046)

Age2 0.002* 0.007*** 0.011*** 0.011*** 0.013** 0.010***(0.002) (0.002) (0.002) (0.002) (0.003) (0.002)

Male 0.426*** 20.293***(0.039) (0.056)

Constant 20.063 0.452** 20.677*** 0.549*** 0.327 0.497**(0.127) (0.181) (0.149) (0.164) (0.235) (0.210)

Number of observations 2,339 1,092 1,247 2,341 1,097 1,244Number of individuals 483 225 258 483 225 258

Robust standard errors in parentheses.*P < 0.1, **P < 0.05, ***P < 0.01.

98 S. TANNER ET AL.

American Journal of Physical Anthropology

have found evidence of overweight and obesity amongadults (Benefice et al., 2007; Zeng et al., 2012). Thisstudy draws on panel data to examine the relationbetween childhood stunting and body composition amonga sample of Tsimane’ youth. Overall, we found that lin-ear growth retardation early in life has implications forbody composition throughout adolescence. Among Tsi-mane’ youth, stunting was associated with lower ageand sex adjusted scores of BMI, skinfold body fatreserves, and arm muscularity after a period of 6 years.The effects of linear growth retardation are particularlyevident among girls and older youth.

Findings in this article are consistent with literaturesuggesting early growth restriction may permanentlyconstrain lean mass development and modify body fatthroughout adolescence and into adulthood (Cameron,2007; Wells, 2007). For example, in Guatemala, Li et al.(2003) found that growth retardation in early childhood(0–2 years of age) was associated with leanness and lessbody fat at 21–27 years of age. In a prospective cohortstudy in Kingston, Jamaica, Walker et al. (2002) foundthat, after controlling for the effects of low birth weight,children who were “chronically stunted” (stunted both at9–24 months and 7 years) had less fat and lower BMIthan non-stunted children at 11 years of age. Althoughstunted children had less body fat than non-stunted chil-dren, stunting was also associated with a more centralfat distribution (as assessed by the ratio of subscapular:triceps skinfolds), indicating that stunting may modifyhow fat is distributed across the body.

Our findings and other research among Latin Ameri-can children (Varela-Silva et al., 2012, Wilson et al.,2012) are relevant to understanding the co-occurrence ofchildhood stunting and overweight or obese status (Pop-kin et al., 1996). Although there is variation in thesestudies with respect to location, study design, and meth-odology, they point to important role of environment inhuman growth and downstream health (Cameron, 2007).

A potential explanation for differences in the literatureis that possible links between stunting and overweightor abdominal fat patterning may be visible only amongyouth with access to high-fat or calorically dense foods(Popkin et al., 1996; Frisancho, 2003; Kain et al., 2005)and moderate levels of daily activity. A high-fat diet maybe particularly important in these associations, asSawaya et al. (1998) found that a high-fat diet resultedin greater body fat gains in stunted girls than non-stunted girls after controlling for the effects of energyintake and activity levels in Brazil. Among the Tsimane’communities participating in this study, householdswere relatively independent of market foods throughoutthe panel study. Although the pattern is beginning tochange, food is primarily produced at the householdlevel and consists of crops such as rice, corn, plantains,and manioc supplemented with wild game, fish, orhousehold domestic animals such as chickens and pigs.Additionally, children and adolescents are activethroughout the day, which would influence both muscu-larity and fatness. Some attend school in the morningsand youths spend the remainder of the day fishing orhunting, playing in groups throughout the community,or assisting parents or older siblings in household dutieswhich may include child-care or working in fields (Aiello,2013). If the consequences of linear growth retardationdiffer with context, we might expect that the increasesin body fat or BMI for age may occur only when at leastone of the double burdens of infection and moderateundernutrition is removed.

A second possible explanation may be population vari-ation in stature. Tsimane’ adults are also relatively shortand about 40% of Tsimane’ adults are classified asstunted when compared with sex and age peers in theUS (Godoy et. al., 2010b). Recently several scholars havesuggested adaptive explanations for small body size(Perry and Dominy, 2008). Although previous researchamong Tsimane’ suggests catch-up growth is influenced

TABLE 6. Prediction testing and results of random-effects panel linear regression models examining the relation between baselinestunting and measures of body composition in youth.

Skinfolds z-score Arm muscle area z-score

All children Girls Boys All children Girls BoysExplanatoryvariables (1) (2) (3) (1) (2) (3)

Youngestchildren (2–4years in2002; n 5 195)

20.029 (0.059) 20.011 (0.095) 20.041 (0.072) 20.393*** (0.088) 20.317** (0.116) 20.484*** (0.130)

Older children(8–10 yearsin 2002;n 5 142)

20.266*** (0.087) 20.474*** (0.118) 20.088 (0.078) 20.653*** (0.099) 20.529*** (0.160) 20.712*** (0.111)

Includes controlsfor motherand householdconditiona

(n 5 477)

20.189*** (0.038) 20.197*** (0.050) 20.077* (0.040) 20.558*** (0.057) 20.445*** (0.082) 20.657*** (0.079)

31 communityeffectsb

20.174*** (0.036) 20.261*** (0.053) 20.072 (0.039) 20.518*** (0.055) 20.436*** (0.080) 20.572*** (0.079)

Values are Beta (Robust standard error).a Control variables include mom’s age, education, height and weight, number of children in household, and household income andwealth.b Adds dummy variables for community of residence.*P < 0.1, **P < 0.05, ***P < 0.01.

LINEAR GROWTH STUNTING AND BODY COMPOSITION 99

American Journal of Physical Anthropology

by local ecology (Godoy et al. 2010a), research also sug-gest that adult stunting does not carry negative conse-quences for socioeconomic indicators of well-being suchas education or income (Godoy et al., 2010b; Undurragaet al., 2012). Additional research is needed to under-stand the longer-term consequences of shortness orstunting.

A second important finding of this study is that sexappeared to modify the association between stuntingand body fatness as assessed through skinfold measure-ments. When boys and girls were analyzed independ-ently, baseline stunting was associated with both lowerage- and sex-adjusted measures of body fat and armmuscularity among girls across a period of 6 years.Among boys, baseline stunting was associated with armmuscularity but the association with skinfold measuresof body fatness was not statistically significant at con-ventional levels of P < 0.05. Although this result shouldbe interpreted with caution in this study, it is consistentwith research in Guatemala where women who werestunted early in life had lower amounts of both lean andfat mass than women who had not been stunted, whilemen who experienced early childhood stunting had adultreductions only in lean mass compared to their non-stunted peers (Li et al., 2003).

Although research among Tsimane’ children has consis-tently found little evidence for disparities between girlsand boys with respect to nutritional status, infection,immune system activation, or rates of catch-up growth(Foster et al., 2005; McDade et al., 2005; Godoy et al.,2006; McDade et al., 2008), a few studies have also indi-cated that there may be sex differences in responses toenvironmental stressors. Prior research has indicatedthat environmental conditions during growth, and specifi-cally greater variation in yearly rainfall levels, are associ-ated with reduced adult height of Tsimane’ women butnot of Tsimane’ men (Godoy et al., 2008b). In contrast, theassociation between climatic conditions and child and ado-lescent height was less clear as the height of boys 2–12years of age was more susceptible to climate events thanthe height of girls of the same age (Godoy et al., 2008a).Life history theory and the developmental origins ofhealth research suggest that boys may be more sensitiveto environmental insults than girls, especially in utero(Stinson, 1985; reviewed in Kuzawa and Pike, 2005). Evi-dence for differential susceptibility to environmentalinfluences after birth is somewhat more conflicting, possi-bly because of the buffering effects of parental behaviorand/or cultural traditions that give preference to boys orgirls (reviewed by Stinson, 1985).

Overall, our findings also suggest that additionalresearch examining diet, activity, and the metabolic con-sequences of stunting in diverse contexts is needed.Work by Hoffman and his colleagues in the shantytownsof Brazil (Schroeder et al., 1999; Hoffman et al., 2000;Martins et al., 2004; Hoffman et al., 2007) and others(Schroeder et al., 1999; Walker et al., 2001; Fernald andNeufeld, 2007;Adair, 2008) have found that child stunt-ing correlates with changes to substrate metabolism,metabolic rate, and blood pressure in children and ado-lescents. This research suggests that stunted childrenhave a reduced capacity to burn (oxidize) fat and thisenhanced fat storage capacity might predispose them toincreased body fatness later in life (Grillol et al., 2005;Walker et al., 2007).

A strength of this study is the fact that this is one ofonly a handful of studies to track body composition in an

indigenous group through puberty. Our finding that lin-ear growth stunting at 2–10 years of age is associatedwith body composition across a 6-year period indicatesthe importance of considering multiple stages of thegrowth process. An important shortcoming of this studyis that we do not have accurate birth weight data or fre-quent measures of linear growth or weight gain duringthe first 2 years of life; therefore, we are unable to eval-uate or control for the role of prenatal growth, birthweight, and early postnatal growth. We also do not havedata on puberty, which could be an important confound-ing factor among the oldest age category. Additionally,information on metabolic rates and activity level amongchildren and adolescents is needed to shed light on theunderlying mechanism linking growth perturbations tolife-long body composition. Finally, rounding error andrandom measurement error in anthropometric measure-ments would inflate standard errors and weaken theresults. These measurement errors suggest that theinverse association between stunting and later anthropo-metric measurements of muscularity and skinfold bodyfat likely underestimates the magnitude of the true rela-tion in this sample.

In sum, these results present a short window ofinsight into the connection between linear growth andbody composition in a rural, indigenous population. Wefound evidence that linear growth stunting was associ-ated with short-run anthropometric measures of bodycomposition among Tsimane’ youth, including lower BMIz-scores and measures of arm muscularity and body fat-ness. Given the importance of developmental plasticityand the potential for intergenerational effects of lineargrowth retardation (Kuzawa, 2007), additional researchthat examines the full consequences of growth stuntingis needed to examine the links between nutrition,health, and disease.

ACKNOWLEDGMENTS

The authors thank Tsimane’ participants and the Tsi-mane’ Grand Council for participation and assistancewith research. Additionally they thank the editor, associ-ate editors, and two anonymous reviewers for helpfulcomments on the manuscript.

LITERATURE CITED

Adair LS. 2008. Child and adolescent obesity: epidemiology anddevelopmental perspectives. Physiol Behav 94:8–16.

Aiello M. 2013. The children of Tsimane’: exploring how chil-dren create and negotiate health outcomes. Doctoral Disserta-tion, Northwestern University.

Azcorra H, Varela-Silva MI, Rodriguez L, Bogin B, Dickinson F.2013. Nutritional status of Maya children, their mothers, andtheir grandmothers residing in the city of Merida, Mexico:revisiting the leg-length hypothesis. Am J Hum Biol 25:659–665.

Baker J, Hurtado AM, Pearson OM, Hill KR, Jones T, Frey MA.2009. Developmental plasticity in fat patterning of Ache chil-dren in response to variation in interbirth intervals: a prelim-inary test of the roles of external environment and maternalreproductive strategies. Am J Hum Biol 21:77–83.

Barker D, Eriksson J, Forsen T, Osmond C. 2002. Fetal originsof adult disease: strength of effects and biological basis. Int JEpidemiol 31:1235–1239.

Benefice E, Lopez R, Monroy SL, Rodr�ıguez S. 2007. Fatnessand overweight in women and children from riverine Amerin-dian communities of the Beni River (Bolivian Amazon). Am JHum Biol 19:61–73.

100 S. TANNER ET AL.

American Journal of Physical Anthropology

Black RE, Victora CG, Walker SP, Bhutta ZA, Christian P, deOnis M, Ezzati M, Grantham-McGregor S, Katz J, MartorellR, Uauy R, Maternal and Child Nutrition Study Group. 2013.Maternal and child undernutrition and overweight in low-income and middle-income countries. Lancet 382:427–451.

Bogin B. 1999. Patterns of human growth. New York: Cam-bridge University Press.

Bogin B, Silva MIV, Rios L. 2007. Life history trade-offs inhuman growth: adaptation or pathology? Am J Hum Biol 19:631–642.

Byron E. 2003. Market integration and health: the impact ofmarket and acculturation on the self-perceived morbidity,diet, and nutritional status of the Tsimane’ Amerindians oflowland Bolvia. Doctoral Dissertation, University of Florida.

Cameron, N. 2007. Growth patterns in adverse environments.Am J Hum Biol 19:615–621.

Cameron N, Wright MM, Griffiths PL, Norris SA, Pettifor JM.2005. Stunting at 2 years in relation to body composition at 9years in African urban children. Obes Res 13:131–136.

de Onis M, Bl€ossner M, Borghi E. 2010. Global prevalence andtrends of overweight and obesity among preschool children.Am J Clin Nutr 92:1257–1264.

de Onis M, Bl€ossner M, Borghi E. 2012. Prevalence and trendsof stunting among pre-school children 1990–2020. PublicHealth Nutr 15:142–148.

de Onis M, Onyango A, Brghi E, Siyam A, Nishida C,Siekmann J. 2007. Development of a WHO growth referencefor school-aged children and adolescents. Bull World HealthOrgan 85:661–668.

Ellis R. 1996. A taste for movement: an exploration of the socialethics of the Tsimane’ of lowland Bolivia. Ph.D. dissertation.St. Andrews University, Scotland.

Fernald LC, Neufeld LM. 2007. Overweight with concurrentstunting in very young children from rural Mexico: preva-lence and associated factors. Eur J Clin Nutr 61:623–632.

Foster Z, Byron E, Reyes-Garc�ıa V, Huanca T, Vadez V, ApazaL, P�erez E, Tanner S, Gutierrez Y, Sandstrom B, Yakhedts A,Osburn C, Godoy R, Leonard WR. 2005. Physical growth andnutritional status of Tsimane’ Amerindian children of lowlandBolivia. Am J Phys Anthropol 126:343–351.

Frisancho AR. 2003. Reduced rate of fat oxidation: a metabolicpathway to obesity in the developing nations. Am J Hum Biol15:522–532.

Frisancho AR. 2008. Anthropometric standards: an interactivenutritional reference of body size and body composition forchildren and adults. Ann Arbor: University of MichiganPress.

Godoy R, Goodman E, Reyes-Garc�ıa V, Eisenberg DT, LeonardWR, Huanca T, McDade TW, Tanner S, Jha N, TAPS BoliviaResearch Team. 2008a. Rain, temperature, and child-adolescent height among Native Amazonians in Bolivia. AnnHum Biol 35:276–293.

Godoy R, Tanner S, Reyes-Garc�ıa V, Leonard WR, McDade TW,Vento M, Broesch J, Fitzpatrick IC, Giovannini P, Huanca T.2008b. The effect of rainfall during gestation and early child-hood on adult height in a foraging and horticultural society ofthe Bolivian Amazon. Am J Hum Biol 20:23–34.

Godoy R, Magvanjav O, Nyberg C, Eisenberg DT, McDade TW,Leonard WR, Reyes-Garc�ıa V, Huanca T, Tanner S, GravleeC. 2010b. Why no adult stunting penalty or height premium?Estimates from native Amazonians in Bolivia. Econ Hum Biol8:88–99.

Godoy R, Nyberg C, Eisenberg DTA, Magvanjav O, Shinnar E,Leonard WR, Gravlee C, Reyes-Garc�ıa V, McDade TW,Huanca T, Tanner S. 2010a. Short but catching up: staturalgrowth among native Amazonian Bolivian children. Am JHum Biol 22:336–347.

Godoy R, Reyes-Garc�ıa V, Gravlee Clarence C, Huanca T,Leonard W, McDade T, Tanner S. 2009. Moving beyond asnapshot to understand changes in the well-being of nativeAmazonians: panel evidence (2002–2006) from Bolivia. CurrAnthropol 50:563–573.

Godoy R, Reyes-Garc�ıa V, McDade T, Tanner S, Leonard WR,Huanca T, Vadez V, Patel K. 2006. Why do mothers favor girls

and fathers, boys? A hypothesis and a test of investment dis-parity. Hum Nat 17:169–189.

Grillol LP, Siqueira AA, Silva AC, Martins PA, Verreschi IN,Sawaya AL. 2005. Lower resting metabolic rate and highervelocity of weight gain in a prospective study of stunted vsnonstunted girls living in the shantytowns of S~ao Paulo, Bra-zil. Eur J Clin Nutr 59:835–842.

Gurven M, Kaplan H, Supa AZ. 2007. Mortality experience ofTsimane Amerindians of Bolivia: regional variation and tem-poral trends. Am J Hum Biol 19:376–398.

Gurven M. 2012. Infant and fetal mortality among a high fertil-ity and mortality poulation in the Bolivian Amazon. Soc SciMed 75:2493–2502.

Hamill PV, Drizd TA, Johnson CL, Reed RB, Roche AF, MooreWM. 1979. Physical growth: National Center for Health Sta-tistics percentiles. Am J Clin Nutr 32:607–629.

Hoffman DJ, Martins PA, Roberts SB, Sawaya AL. 2007. Bodyfat distribution in stunted compared with normal-height chil-dren from the shantytowns of S~ao Paulo, Brazil. Nutrition 23:640–646.

Hoffman DJ, Sawaya AL, Coward WA, Wright A, Martins PA,de Nascimento C, Tucker KL, Roberts SB. 2000a. Energyexpenditure of stunted and nonstunted boys and girls livingin the shantytowns of Sao Paulo, Brazil. Am J Clin Nutr 72:1025–1031.

Hoffman DJ, Sawaya AL, Verreschi I, Tucker KL, Roberts SB.2000b. Why are nutritionally stunted children at increasedrisk of obesity? Studies of metabolic rate and fat oxidation inshantytown children from Sao Paulo, Brazil. Am J Clin Nutr72:702–707.

Huanca T. 2008. Tsimane’ oral tradition, landscape, and inden-tity in a tropical forest. La Paz, Bolivia: Wa-Gui.

Hurtado AM, Lambourne CA, James P, Hill K, Cheman K, BacaK. 2005. Human rights, biomedical science, and infectiousdisease among South American indigenous groups. Annu RevAnthropol 34:639–665.

Kain J, Uauy R, Lera L, Taibo M, Albala C. 2005. Trends inheight and BMI of 6-year-old children during the NutritionTransition in Chile. Obes Res 13:2178–2186.

Kuzawa CW. 1998. Adipose tissue in human infancy and child-hood: an evolutionary perspective. Am J Phys Anthropol 107:177–209.

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

Kuzawa CW, Pike IL. 2005. Introduction: fetal origins of devel-opmental plasticity. Am J Hum Biol 17:1–4.

Leonard WR, Godoy R. 2008. Tsimane’ Amazonian Panel Study(TAPS): the first 5 years (2002–2006) of socioeconomic, demo-graphic, and anthropometric data available to the public.Econ Hum Biol 6:299–301.

Leonard WR, Sorensen MV, Mosher MJ, Spitsyn V, ComuzzieAG. 2009. Reduced fat oxidation and obesity risks among theBuryat of southern Siberia. Am J Hum Biol 21:664–670.

Li H, Stein AD, Barnhart HX, Ramakrishnan U, Martorell R.2003. Associations between prenatal and postnatal growthand adult body size and composition. Am J Clin Nutr 77:1498–1505.

Loman TG, Roche AF, Martorell R. 1988. Anthropometric stand-ardization reference annual. Windsor, Ontario: HumanKinetics Publishers.

Lourenco A, Santos R, Orellana J, Coimbra Jr. C. 2008. Nutri-tion transition in Amazonia: obesity and socioeconomicchange in the Surui Indians from Brazil. Am J Hum Biol 20:564–571.

Martins PA, Hoffman DJ, Fernandes MTB, Nascimento CR,Roberts SB, Sesso R, Sawaya AL. 2004. Stunted childen gainless lean body mass and more fat mass than their non-stunted counterparts: a prospective study. Br J Nutr 92:819–825.

McDade TW, Leonard WR, Burhop J, Reyes-Garc�ıa V, Vadez V,Huanca T, Godoy RA. 2005. Predictors of C-reactive proteinin Tsimane’ 2–15-year-olds in lowland Bolivia. Am J PhysAnthropol 128:906–913.

LINEAR GROWTH STUNTING AND BODY COMPOSITION 101

American Journal of Physical Anthropology

McDade TW, Reyes-Garc�ıa V, Tanner S, Huanca T, LeonardWR. 2008. Maintenance versus growth: investigating thecosts of immune activation among children in lowland Boli-via. Am J Phys Anthropol 136:478–484.

Nyberg CH, Leonard WR, Tanner S, McDade T, Huanca T,Godoy RA, TAPS Bolivia Study Team. 2012. Diurnal cortisolrhythms and child growth: exploring the life history conse-quences of HPA activation among the Tsimane’. Am J HumBiol 24:730–738.

Perry GH, Dominy NJ. 2008. Evolution of the human pygmyphenotype. Trends Ecol Evol 24:218–225.

Piperata BA, Spence JE, Da-Gloria P, Hubbe M. 2011. Thenutrition transition in Amazonia: rapid economic change andits impact on growth and development in Ribeirinhos. Am JPhys Anthropol 146:1–13.

Popkin BM, Richards MK, Montiero CA. 1996. Stunting is asso-ciated with overweight in children of four nations that areundergoing the nutrition transition. J Nutr 126:3009–3016.

Reyes-Garc�ıa V, Gravlee CC, McDade TW, Huanca T, LeonardWR, Tanner S, TAPS Bolivia Study Team. 2010. Cultural con-sonance and body morphology: estimates with longitudinaldata from an Amazonian society. Am J Phys Anthropol 143:167–174.

Rabe-Hesketh S, Skrondal A. 2008. Multilevel and longitudinalmodeling using STATA. College Station, TX: Stata Press.

Ringhofer L. 2010. Fishing, foraging, and farming in the Boliv-ian Amazon: on a local society in transition. London:Springer.

Sawaya AL, Grillo LP, Verreschi I, de Silva AC, Roberts SB.1998. Mild stunting is associated with higher susceptibility tothe effects of high fat diets: studies in a shantytown popula-tion in Sao Paulo, Brazil. J Nutr 128:415S–420S.

Schroeder DG, Martorell R, Flores R. 1999. Infant and childgrowth and fatness and fat distribution in Guatemalanadults. Am J Epidemiol 149:177–185.

Stein AD, Wang M, Martorell R, Norris SA, Adair LS, Bas I,Sachdev HS, Bhargava SK, Fall CHD, Gigante DP, CohortsGroup. 2010. Growth patterns in early childhood and finalattained stature: data from five birth cohorts from low- andmiddle-income countries. Am J Hum Biol 22:353–359.

Sterling R, Miranda JJ, Gilman RH, Cabrera L, Sterling CR,Bern C, Checkley W. 2012. Early anthropometric indices pre-dict short stature and overweight status in a cohort of Peru-vians in early adolescence. Am J Phys Anthropol 148:451–461.

Stinson S. 1985. Sex differences in environmental sensitivityduring growth and development. Am J Phys Anthropol 28 S6:123–147.

Tanner S, Leonard WR, McDade TW, Reyes-Garc�ıa V, Godoy R,Huanca T. 2009. Influence of helminth infections on childhoodnutritional status in lowland Bolivia. Am J Hum Biol 21:651–656.

Undurraga EA, Zebrowitz L, Eisenberg DTA, Reyes-Garc�ıa V,TAPS Bolivia Study Team, Godoy R. 2012. The perceived ben-efits of height: strength, dominance, social concern, andknowledge among Bolivian Native Amazonians. PLOS One 7:e35391.

Varela-Silva MI, Azcorra H, Dickinson F, Bogin B, FrisanchoARF. 2009. Influence of maternal stature, pregnancy age, andinfant birth weight on growth during childhood in Yucatan,Mexico: a test of the Intergenerational Effects Hypothesis.Am J Hum Biol 21:657–663.

Varela-Silva MI, Dickinson F, Wilson HJ, Azcorra H, GriffithsPL, Bogin B. 2012. The nutritional dual-burden in developingcountries: how is it assessed and what are the health implica-tions. Coll Anthropol 36:39–45.

Walker SP, Gaskin P, Powell CA, Bennett FI, Forrester TE,Grantham-McGregor S. 2001. The effects of birth weight andpostnatal linear growth retardation on blood pressure at age11–12 years. J Epidemiol Community Health 55:394–398.

Walker SP, Gaskin P, Powell CA, Bennett FI. 2002. The effectsof birth weight and postnatal linear growth retardation onbody mass index, fatness and fat distribution in mid and latechildhood. Public Health Nutr 5:391–396.

Walker SP, Chang SM, Powell CA. 2007. The associationbetween early childhood stunting and weight status in lateadolescence. Int J Obes 31:347–352.

Walker R, Gurven M, Hill K, Migliano A, Chagnon N, De SouzaR, Djurovic G, Hames R, Hurtado AM, Kaplan H, Kramer K,Oliver WJ, Valeggia C, Yamauchi T. 2006. Growth rates andlife histories in 22 small-scale societies. Am J Hum Biol 18:295–311.

Wells JCK. 2010. The evolutionary biology of human body fat-ness: thrift and control. Cambridge: Cambridge UniversityPress.

Wilson HJ, Dickinson F, Hoffman DJ, Griffiths PL, Bogin B,Varela-Silva MI. 2012. Fat free mass explains the relationshipbetween stunting and energy expenditure in urban MexicanMaya children. Ann Hum Biol 39:432–439.

Zafon C. 2007. Oscillations in total body fat content throughlife: an evolutionary perspective Obes Rev 8:525–530.

Zeng W, Eisenberg DTA, Jovel KR, Undurraga EA, Nyberg C,Tanner S, Reyes-Garc�ıa, Leonard WR, Casta~no J, Huanca T,McDade TW, TAPS Bolivia Study Team. 2013. Adult obesity:panel study from native Amazonians. Econ Hum Biol 11:227–235.

102 S. TANNER ET AL.

American Journal of Physical Anthropology