Genetic and Environmental Components of Thyroxine Variation in Mennonites from Kansas and Nebraska Genetic and Environmental Components of Thyroxine Variation in Mennonites from Kansas and Nebraska L.J. MARTIN1 AND M.H. CRAWFORD* Abstract Thyroxine is an endocrine hormone that regulates cellular and organismic metabolism. Current research on thyroxine has primarily examined its adaptive potential and genetic inheritance patterns. To date, no studies have attempted to investigate the interaction between the ge- netic and environmental components of thyroxine variation. This ap- proach is useful because hormones are on feedback regulation; thus in- teraction occurs between the environment and gene expression. The purposes of this research are to characterize the genetic and environmen- tal components of thyroxine variation using univariate statistics and to estimate the genetic and cultural heritabilities through path analysis. For univariate analyses, analyses of variance are used to determine whether or not age, sex, or community affiliation are covariates of thyroxine level. Significant differences existed in thyroxine level based on sex and com- munity affiliation ( p < 0.05). The genetic and environmental components of thyroxine variation were partitioned through path analysis. Heritability was estimated at 0.317 ± 0.109 for the genetic component and at 0.060 ± 0.029 for the environmental component. The environmental variables that contributed to the variation in thyroxine level were caffeine con- sumption, blood calcium level, and biceps skinfold thickness. Thyroxine is a hormone in the endocrine system that regulates cellular and organismic metabolism and governs calorigenesis, lipid, carbohydrate, and protein metabolism, respiratory and cardiovascular function, and growth and development (Butt 1975; Hardy 1981). There are two forms of this hormone: thyroxine (T4) and triiodothyronine (T3). T3 is recognized as the metabolically active form of thyroxine (Butt 1975; Hardy 1981). Compared to T3, T4 is secreted at a ratio of 20 to 1 and is less responsive to environmental influences (Harland and Orr 1975; Hardy 1981; Danforth 1986). Current research on thyroxine focuses primarily on its metabolic functions, although a few studies have examined its adaptive potential (Schwartz 1983; Riis and Madsen 1985; 1 Department of Anthropology, University of Kansas, Lawrence, KS 66045. Human Biology, August 1998, v. 70, no. 4, pp. 745-760. Copyright © 1998 Wayne State University Press, Detroit, Michigan 48201-1309 KEY WORDS: HORMONES, THYROXINE, ENDOCRINE SYSTEM, PHYSIOLOGY, MENNONITES http://www.jstor.org/page/info/about/policies/terms.jsp 746 / MARTIN AND CRAWFORD Ballard 1986; Legrand 1986; Pimentel 1987; Cabello and Wrutniak 1990; Maruo et al. 1992a,b; Sower et al. 1992; Facchini et al. 1997). For example, thyroid hormone fluctuations have been reported in cold adaptation studies and a relationship between thyroxine and physique has been suggested (Reed et al. 1986, 1990, 1992; Silverin et al. 1989; Gaikwad et al. 1990; Kwiecinski et al. 1991; Rone et al. 1992; Kirchengast 1994). To date, no studies have attempted to explain thyroxine variation by focusing on the interaction between genetic, environmental, and endocrino- logical factors. However, integrating information from several disciplines is particularly informative when studying hormones such as thyroxine because of their complex interrelationships. It is important to identify both the genetic and the environmental components of normal thyroxine variation because thyroxine variation is due to both genetic and environmental factors. The goals of this research are to examine the genetic and environmental compo- nents involved in thyroxine variation using univariate statistics and to estimate the genetic and cultural heritabilities using path analysis. Materials and Methods Population. The data for this study were collected from three Mennonite communities, two in Kansas (Goessel and Meridian) and one in Nebraska (Henderson) (Figure 1), as part of a multidisciplinary study of biological aging. The collected data included demographics, reproductive histories, blood specimens, nutritional data, anthropometrics, genealogies, psycholog- ical questionnaires, and neuromuscular tests (Crawford and Rogers 1982). Crawford and Rogers (1982) demonstrated that Goessel and Henderson are most similar genetically because they both originated from a single commu- nity, Alexanderwohl, during the 1870s. In contrast, Meridian is genetically heterogeneous because it was recently founded by a charismatic leader and consisted of several familial groupings of various origins. Data Collection and Analysis. From November 1980 to January 1981 data were collected from 1062 Mennonites, ranging in age from 18 to 94 years, from the 3 communities (Goessel, Henderson, and Meridian). Individ- uals who had a history of thyroid disease were excluded from the analysis, thus reducing the sample used in this study to 1042 individuals (Goessel, 460 individuals; Meridian, 85 individuals; Henderson, 497 individuals). The total sample of 1042 individuals was included in the computation of univariate statistics. However, only family data were used for the path analysis, thus reducing the sample to 378 individuals representing 112 families (111 fathers, 112 mothers, and 155 children). To assess thyroxine levels, we drew blood specimens from nonfasting volunteers. The serum T4 levels were assayed using the standard radioim- http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 747 /NEBRASKA v LINCOLN * J HENDERSON * 1 KANSAS TOPEKA INMAN, #GOESSEL •meridian VVICHITA Figure 1. Locations of the Mennonite communities of Goessel, Meridian, and Henderson in Kansas and Nebraska. munoassay technique of Consolidated Biomedical Laboratories (Roche Labs, Wichita, Kansas) immediately after the blood was drawn. Univariate Methods. Univariate methods were used to examine popula- tion and individual differences. For population level analyses means, standard deviations, and standard errors of thyroxine levels for the three Mennonite communities were compared to each other and to the normal physiological variation. For individual level analyses age effects were examined by regress- ing thyroxine variation on age and on age-squared. Sex differences in thy- roxine level were analyzed using a one-way analysis of variance (anova). Furthermore, thyroxine levels for females were partitioned into pre- and post- menopausal groups and were compared using anova. Last, the postmeno- pausal women were separated into those who had complete hysterectomies (surgical menopause) versus those who had experienced menopause without hysterectomies (natural menopause), and the thyroxine levels were compared using ANOVA. Analytical Methods. Path analysis was used to estimate the genetic and environmental components of thyroxine variation by partitioning the two http://www.jstor.org/page/info/about/policies/terms.jsp 748 / MARTIN AND CRAWFORD Figure 2. Path diagram of pathmix hi demonstrating the relationship between parents and off- spring with respect to genotype and environment. The subscripts, F, M, CI, and C2, represent father, mother, and two children. P is phenotype, G is genotype, C is trans- missible environment with index /, and B is nontransmitted common sibship environ- ment (Rao, Williams et al. 1983). components on the basis of nuclear family data by using a predetermined causal scheme, the path diagram (see Figure 2) (Wright 1918; Rao et al. 1982). This analysis uses pathmix hi, a computer program designed by D.C. Rao and colleagues to directly use nuclear family data (Rao et al. 1982; Rice et al. 1991). A cultural or environmental index was created using a phenotype de- fined by thyroxine level to characterize the familial environment (Morton 1974; Rao, Williams et al. 1983). To construct this index, we first used linear regression to correct the thyroxine level for age and sex effects. The age- and sex-corrected data were then normalized by the inverse normal transformation and were used as the phenotype values. Once the phenotype was character- ized, an environmental index was constructed by multiple regression analysis of the phenotype (forward and backward and best subsets). The following variables were used to determine the significant contributors tó T4 variation: blood glucose, low-density lipoprotein, blood calcium, smoking, alcohol con- sumption, caffeine consumption, church affiliation, weight, height, and sub- scapular, triceps, biceps, and suprailiospinal skinfold, thicknesses. The re- gression equation (Y = aX + b) produces a line that defines the relationship between the independent variables (possible contributors) and the dependent variable (thyroxine level). The Ts from the regression equation are called the predicted Ts. The normalized predicted Ts constitute the cultural index. http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 749 Table 1. Definition of the Ten Parameters in the Path Model (Rao, Williams et al. 1983) Parameter Definition h Effect of genotype on child's phenotype (square root of genetic heritability) hz Effect of genotype on adult's phenotype c Effect of environment on child's phenotype (square root of cultural heritability) cy Effect of environment on adult's phenotype u Correlation between parental environments b Effect of sib-transmitted common sibship environment on child's environment /F Effect of father's environment on that of a child he rears /M Effect of mother's environment on that of a child she rears i Effect of environment on child's index iv Effect of environment on adult's index PATHMix m uses the following linear equation to estimate the various parameters: P = G + C + R, (1) where P is the phenotype, G is the genetic contribution, C is the cultural and familial environmental contribution, and R is the residual resulting from ran- dom environmental factors (Rao, Williams et al. 1983; Friedlander et al. 1986). It is assumed that G, C, and R are uncorrected (Rao, Williams et al. 1983). The "structural equation" is produced when the variables in Eq. (1) are standardized and are all divided by the variance of the phenotype (Rao, Morton et al. 1983; Rao, Williams et al. 1983): p = hG* + cC* + rR*, (2) where G*, C*, and R * are the standardized variables, h = (Vg/Vp), c = (Yc/Vp), and r = (Vr/Vp), and V is the variance. These are also known as the standardized partial regression coefficients or path coefficients. Dividing both sides of the structural equation by the phenotypic variance yields (Rao, Mor- ton et al. 1983; Rao, Williams et al. 1983) 1 = h2 + c2 + r2, (3) where h 2 is the genetic heritability, c2 is the cultural heritability, and r2 is the residual variation. This mathematical expression is known as the equation for complete determination (Rao, Morton et al. 1983; Rao, Williams et al. 1983). The PATHMix m method contains 6 known variables (PF, phenotype of father; PM, phenotype of mother; Pc, phenotype of child; /F, index of father; /M, index of mother; /c, index of child) and 10 unknown parameters, which are displayed in the path diagram (Figure 2). Table 1 defines each of the 10 unknown parameters that were estimated by maximum likelihood. The 6 known variables yielded 15 correlations, several of which are repetitive (PF, /F and PM, /M and PF, /M and /F, PM), thus reducing the number of correlations http://www.jstor.org/page/info/about/policies/terms.jsp 750 / MARTIN AND CRAWFORD Table 2. Definitions of the Null Hypotheses Tested Null Hypothesis Mathematical Expression No intergenerational differences with respect to heritabilities y - z = 1 No marital resemblance (assortative mating) u = 0 No common sibship effect b - 0 Equal parental transmission of familial environment /F = /M No genetic heritability h = z = 0 No cultural heritability c = y = 0, i = v= 1 to 13. Moreover, the 3 sib-sib correlations raise the total number of correla- tions to 16, which serve as a basis for the interpretation of the data and a test of the fit of the general and most parsimonious models (Rao et al. 1984). PATHMix m estimates the model directly from the family data instead of by computing the model from the correlations. However, the 16 correla- tions, 6 means, and 6 variances of the 6 variables are used to test the overall log-likelihood function (Rao et al. 1984; Duggirala and Crawford 1995). To test the functions, we calculated the log-likelihood for single nuclear families using the deviation of the phenotype from the mean; then the overall log- likelihood was calculated using the log-likelihood from all nuclear families. The overall likelihood ratios were used to test the goodness of fit of given models and to test specific hypotheses. Twice the difference in likeli- hood ratios follows a chi-square distribution. To test the fit of the model using all the path coefficients and their variances, we tested the log-likelihood of the correlation model against the log-likelihood of the unconstrained path model. To test the null hypotheses listed in Table 2, we tested the log- likelihood for each constrained model against the log-likelihood for the un- constrained path model. The degrees of freedom for all tests was the differ- ence in the number of estimated parameters. Results Univariate Analysis. Means and standard deviations of thyroxine levels in the three Mennonite communities are presented in Table 3. The mean values of thyroxine level in these populations fall within normal physiological variation (Hardy 1981). The distribution of thyroxine level appears to be relatively normal but with a slight skew to the left (Figure 3). No statistically significant age effects were observed in the distribution of the T4 levels. How- ever, one-way anova revealed significant differences between the commu- nities and the sexes ( p < 0.05) (Figure 4). Moreover, thyroxine levels of pre- and postmenopausal women were significantly different ( p < 0.05) with a mean thyroxine level for the postmenopausal group of 8.38 versus 8.95 for http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 751 Table 3. Means and Standard Deviations of Thyroxine Levels in the Three Mennonite Communities Community N Mean SD Goessel Males 220 8.020 1.583 Females 240 8.524 1.445 Meridian Males 38 7.826 2.079 Females 47 8.085 1.926 Henderson Males 242 8.289 1.589 Females 255 8.707 1.711 Figure 3. Frequency distributions for thyroxine levels in the three Mennonite communities. http://www.jstor.org/page/info/about/policies/terms.jsp 752 / MARTIN AND CRAWFORD T4 Level 9.000 -r T T T □ 8.000 - 7.000 -| 1 J 1 Meridian Goessel Henderson Figure 4. Mean thyroxine (T4) levels and standard error bars in men (diamonds) and women (squares) for the three Mennonite communities. the premenopausal group. Yet no statistical difference was detected between the women who had hysterectomies and those who had experienced natural menopause. Path Analysis. Using the criteria of the F ratios, adjusted R2, mean square error, tolerance, and Mallow's Cp , we constructed the environmental index from the best subsets of the regression of all possible variables. This model contained three variables: caffeine intake, blood calcium level, and biceps skinfold thickness. The familial correlations estimated by maximum likeli- hood are found in Table 4. Only the father-child phenotype correlation was significant, suggesting a paternal effect. For the environmental index corre- lations the mother-father and child-child correlations were highly significant, whereas the other index correlations were low, suggesting intergenerational differences. Parameter estimates for the general model, a summary of the null hy- potheses tested, and parameter estimates for the most parsimonious model are found in Tables 5 through 7. The general (10-parameter) model fits the data (xl = 6.627, p = 0.36) (Table 5). The most parsimonious model was generated from the accepted null hypotheses, which include the following hypotheses: equal parental trans- http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 753 Table 4. Familial Correlations Estimated by Maximum Likelihood Variable Correlation n p Phenotype PF, PM -0.0377 94 ns PF,PC 0.3052 144 pCO.Ol PM, Pc -0.0177 138 ns Peu Pe 2 0.1258 40 ns Index /F, /M 0.4240 111 p <0.01 /F,/c 0.2281 121 p = 0.01 /M,/c 0.3144 119 pCO.Ol IciJci 0.1957 25 pCO.Ol Cross traits PF, /F = PM , /M 0.2357 207 PF, /M = /F, 0.0788 188 PF, Ic 0.0986 129 /p, Pc 0.0423 125 PM, Ic 0.0595 146 IM,PC 0.0843 117 PciJci -0.1491 55 PC2JC i 0.2243 118 mission of environment, no generational differences, and no common sibship. This model was compared for goodness of fit to the observed (Table 4) and the expected (general 10-parameter; Table 5) models using a chi-square sta- tistic. The most parsimonious model estimated the genetic heritability as 0.317 (0.109) and the cultural heritability as 0.060 (0.029). Table 5. Parameter Estimates and Their Standard Errors for the General Model Parameter Estimate Standard Error h 0.6201 0.2452 c 0.2567 0.0817 y 0.9206 0.3934 z 0.8175 0.6481 u 0.4238 0.2128 /f 0.1357 0.0980 fM 0.2402 0.0982 b 0.3571 0.2617 i 1.0000 0.0000 V 1.0000 0.2377 X2 12.625 d.f. 6 p 0.0500 http://www.jstor.org/page/info/about/policies/terms.jsp 754 / MARTIN AND CRAWFORD Table 6. Chi-Square Values and Probabilities Associated with the Tested Null Hypotheses Null Hypothesis d.f y2 Probability y = z = 1 2 0.092 0.955 u = 0 1 18.949 <0.0001 ¿7 = 0 1 0.452 0.5014 fF=fM 1 0.460 0.4976 h = z = 0 2 8.373 0.0152 c = y = 0, i = v = 1 4 19.221 0.0007 Parsimony /m = /f. y = z = 1, b = 0 4 0.980 0.928 Discussion Univariate Analysis. The significant differences in thyroxine level that exist between the three communities may be attributed to either environment or genetics. Considering that the three Mennonite communities share similar diets and lifestyles, it is unlikely that these differences are due to environment. Furthermore, Henderson and Goessel are more similar to each other with respect to mean thyroxine level than to Meridian (Figures 3 and 4). Indeed, Goessel and Henderson were once a single community that later divided. Because these communities are historically related, they should be more simi- lar genetically. Even though biological differences exist between the corn- Table 7. Parameter Estimates for the Most Parsimonious Model Parameter Tf Standard Error h 0.5630 0.0928 c 0.24434 0.0597 y (1.0000) (0) z (1.0000) (0) « 0.4217 0.1865 b (0.0000) (0) /f (0.2000) (0) /m (0.2000) (0) i 1.0000 0.0000 v 1.0000 0.2029 X2 0.980 d.f. 4 p 0.928 a. Values in parentheses represent variables that were fixed. http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 755 munities, these differences do not affect the results of path analysis because these methods are concerned with the segregation of traits within families, not within populations. Path Analysis. Path analysis was used to partition the genetic and envi- ronmental contributions to variation in thyroxine level. Because the environ- ment cannot be directly measured (Rao, Williams et al. 1983), an index was developed using the significant contributors to thyroxine variation. This index identifies which environmental factors were significant. In this analysis blood calcium level, caffeine consumption, and biceps skinfold thickness contrib- uted significantly to the expression of thyroxine level. The genetic component of path analysis yielded heritability estimates and information on intergen- erational differences, homogamy, and other factors. Correlation analysis revealed that a significant positive relationship ex- ists between caffeine intake and thyroxine level. To date, there has been no specific study of the relationship between thyroxine and caffeine; however, there is evidence for a possible interaction between these variables. Previous studies have demonstrated that thyroid hormones increase the level of aden- ylate cyclase activity and the cellular cyclic AMP (cAMP) level (Segal et al. 1985, 1989). cAMP is a secondary messenger for the action of many hor- mones. When hormones attach to their receptors on the plasma membrane, cAMP provides the necessary energy for the hormonal action (Stryer 1988). Evidence for cAMP as a secondary messenger includes stimulation of aden- ylate cyclase and positive correlations between cellular cAMP concentrations and hormonal concentrations (Stryer 1988). Segal et al. (1985, 1989) dem- onstrated that increases in thyroxine cause concomitant increases in the aden- ylate cyclase activity and cAMP levels, suggesting that thyroxine uses cAMP as a secondary messenger. Caffeine acts synergistically with hormones that use cAMP as a secondary messenger (Stryer 1988). Because of this nonad- ditive effect, positive correlations between thyroxine and caffeine level would be expected. Thus our data support a synergistic action of caffeine and thy- roxine through cAMP. Previous studies identified a relationship between thyroxine and calcium level (Davis et al. 1983; Segal et al. 1985, 1989). Segal et al. (1989) suggested that this relationship is due to calcium's regulation of thyroxine activity. This hypothesis is based on two previous findings: (1) T3 and T4 stimulation of Ca2+ ATPase activity depends on baseline levels of cytoplasmic calcium because the calcium is pumped out of the cell by Ca2+ ATPase (Davis et al. 1983; Segal et al. 1985, 1989); and (2) calcium is required for the stimulation of cAMP and adenylate cyclase by thyroxine (Segal et al. 1989). Thus calcium regulates thyroxine's activity because intracellular calcium is necessary for thyroxine's effects on cAMP and Ca2+ ATPase. When thyroxine levels are elevated, Ca2+ ATPase is stimulated and calcium moves into the bloodstream, which in turn lowers the intracellular calcium levels. When the intracellular http://www.jstor.org/page/info/about/policies/terms.jsp 756 / MARTIN AND CRAWFORD levels of calcium become too low, thyroxine fails to stimulate cAMP and adenylate cyclase, thereby diminishing thyroxine's hormonal effects. There- fore calcium concentration is a regulator of thyroxine activity through Ca2+ ATPase. However, thyroxine can also regulate blood calcium levels by re- leasing calcium from the cells. Calcium cannot exit the cell without the pump- ing action of Ca2+ ATPase. The relationship between calcium and thyroxine may explain why post- menopausal women are particularly prone to osteoporosis. There is a signifi- cant difference in pre- and postmenopausal women's thyroxine levels, with postmenopausal women having lower levels. After menopause there is less thyroxine in the bloodstream; thus less thyroxine enters the cells and less calcium is released by means of the Ca2+ ATPase pump. The body continues to require the same level of calcium, and therefore calcium is removed from bone. Indeed, the relationship between calcium and thyroxine can be seen as one maintaining an equilibrium. At menopause the equilibrium is disrupted by a drop in thyroxine level. Gradually, the body adjusts to a lower thyroxine level and the rate of bone loss is reduced. This apparent relationship is sup- ported by previous research that demonstrated that the rate of bone loss is greatest during the first five to ten years after menopause, after which the rate declines (Krolner and Nielsen 1982; Geusens et al. 1986). Biceps skinfold thickness is positively correlated with thyroxine level and may be explained through energy balance. During periods of positive energy balance and surplus energy, fat is deposited. Indeed, Kirchengast (1994) found positive correlations between the thyroid hormones and various measures of body fatness. Individuals with prolonged positive energy balance have higher thyroxine levels because additional thyroid hormone is required to metabolize the surplus food intake. Extremely low thyroxine levels also promote fat storage but only in regions of primary fat storage such as the abdomen, hips, and buttocks (Van Hardeveld 1986; Després et al. 1988). Biceps skinfold thickness is the best predictor of thyroxine level because the biceps maintain a significant portion of muscle in most individuals and thus are less likely to be a storage area for fat (Després et al. 1988; Lohman 1992). Therefore the relationship between thyroxine level and biceps skinfold thick- ness may be due to energy balance. Heritability Estimates. Unlike the univariate methods, path analysis es- timates the genetic and cultural heritabilities (which were 0.317 and 0.060, respectively) for Mennonite thyroxine levels. These estimates explain only 38% of the observed variation, and it is unclear what causes the other 62% of the thyroxine variation. The genetic heritability measured by path analysis is slightly higher than the univariate estimate of heritability for the Menno- nites (Walawender and Crawford 1992), slightly lower than the estimate for Salt Lake City male twins (Meikle et al. 1988), and similar to heritabilities measured by variance decomposition for Mexican Americans (Comuzzie et http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 757 al. 1996). The variation in genetic heritability estimates is due to several factors. First, the data types are different; the Salt Lake City probands are twins, which can inflate estimates. Second, the methods of analysis can alter the heritability estimates. Path analysis splits variation into genetic and en- vironmental components, variance decomposition separates the genetic vari- ation into the additive and dominance components, and univariate methods examine only additive genetic variation. Therefore slight differences in esti- mates are to be expected. Third, the heritability estimates are population spe- cific; thus intrapopulation variation is normal. Although the genetic heritability estimates are well within the range of previous studies, a large residual remained. One possible reason for this re- sidual is the confounding effect of menopause. Although no significant age effects were observed, there was a significant menopausal effect. Postmeno- pausal women had significantly lower thyroxine levels than did premenopau- sal women. This is problematic when estimating heritabilities because there are two distinct categories of women. A simple correction for sex may over- simplify the physiological situation by forcing females into a single category so that neither pre- nor postmenopausal females will be comparable to males or to each other. Attempts to correct for the menopausal effect were made within females; however, these corrections resulted in insignificant general models of path analysis. This could be due to the dramatic changes that occur during perimenopause and immediately after menopause. Conclusions In this study we estimated the genetic and environmental components of thyroxine variation. There are significant sex effects on thyroxine levels that must be corrected in any studies that include both sexes. Although no age effect was observed, a menopausal effect was detected. Future studies must address the question of how to statistically correct for menopausal status. Dividing the female sample into pre- and postmenopausal groups is insuffi- cient because physiological changes commence several years before meno- pause during the period of perimenopause. Thus corrections must account for perimenopausal changes in addition to menopause. Thyroxine level was influenced by caffeine intake, blood calcium level, and biceps skinfold thickness. Through an examination of the relationship of each of these variables to thyroxine, a better understanding of the variation in thyroxine levels is gained. Caffeine appears to act synergistically with thyroxine by means of cellular cAMP. Calcium regulates thyroxine activity by controlling thyroxine's entrance into the cell. Furthermore, the thyroxine- calcium relationship may explain why osteoporosis occurs after menopause. Last, biceps skinfold thickness may demonstrate how thyroxine is influenced by the long-term energy balance. Thus, by examining the variables that affect http://www.jstor.org/page/info/about/policies/terms.jsp 758 / MARTIN AND CRAWFORD thyroxine variation, we can gain a better understanding of the inheritance of thyroxine. To date, no previous research has estimated the cultural component of thyroxine variation. However, by examining the residual value for this model, we can assume that other factors (not taken into consideration) affect thyrox- ine level variation. For example, estrogen levels have been demonstrated in- directly to play a significant role in thyroxine variation (Chopra 1981; Sower et al. 1992). Unfortunately, estrogen levels were not available for this data set; thus the relationship between estrogen and thyroxine could not be ex- amined directly. Future research could address how permanent changes in estrogen levels, such as at menopause, affect thyroxine variation and other biochemical levels. Indeed, the key to understanding thyroxine variation may be through an examination of the complex interactions of the endocrine system. Received 5 March 1997 ; revision received 2 September 1997. Literature Cited Ballard, P.L. 1986. Hormones and Lung Maturation. Berlin, Germany: Springer-Verlag. Butt, W.R. 1975. Hormone Chemistry, v. 2. New York: Wiley. Cabello, G., and C. Wrutniak. 1990. Thyroid function in the newborn lamb: Physiological approach of the mechanisms inducing changes in plasma thyroxine, free thyroxine, and triiodothyronine concentration. J. Dev. Physiol. 13:25-32. Chopra, I. J. 1981. Triiodothyronines in Health and Disease. New York: Springer- Verlag. Comuzzie, A.G., J. Blangero, M.C. Mahaney et al. 1996. Genetic and environmental correlations among hormone levels and measures of body fat accumulation and topography. J. Clin. Endocrinol. Metab. 81:597-600. Crawford, M.H., and L. Rogers. 1982. Population genetic models in the study of aging and longevity in a Mennonite community. Soc. Sei. Med. 16:149-153. Danforth, E., Jr. 1986. Effects of fasting and altered nutrition on thyroid hormone metabolism in man. In Thyroid Hormone Metabolism , G. Hennemann, ed. New York: Marcel Decker, 335-358. Davis, F.B., V. Cody, P.J. Davis et al. 1983. Stimulation by thyroid hormone analogues of red blood cell Ca2+ ATPase activity in vitro. J. Biol. Chem. 258:12,373-12,377. Després, J.P., A. Tremblay, and C. Bouchard. 1988. Regionally adipose tissue distribution and plasma lipoproteins. In Fat Distribution During Growth and Later Health Outcomes , C. Bouchard and F.E. Jonhston, eds. New York: Liss, 221-242. Duggirala, R., and M.H. Crawford. 1995. Biological and environmental determinants of lipids and lipoproteins in the Mennonites: A path analytical approach. Unpublished. Facchini, F., D. Pettener, A. Rimoondi et al. 1997. Taste sensitivity to PTC and thyroid function (FT4 and TSH) in high- and low-altitude Kirghiz populations in the Pamir. Hum. Biol. 69:97-106. Friedlander, Y., J.D. Kark, and Y. Stein. 1986. Biological and environmental sources of vari- ation in plasma lipids and lipoproteins: The Jerusalem Lipid Research Clinic. Hum. Hered. 36:143-153. http://www.jstor.org/page/info/about/policies/terms.jsp Thyroxine Variation in Mennonites / 759 Gaikwad, A.S., T. Ramassarma, and C.K.R. Kurup. 1990. Changes in thryoid hormone and insulin status of the brown adipose tissue of cold acclimated rats on short-term exposure to heat. Ind. J. Biochem. 27:167-171. Geusens, P., J. Dequeker, A. Verstraeten et al. 1986. Age-, sex-, and menopause-related changes of vertebral and peripheral bone: Population study using dual and single photon absorp- timetry and radiogrammetry. J. Nuclear Med. 27:1540-1549. Hardy, R.N. 1981. Endocrine Physiology. Baltimore, MD: University Park Press. Harland, W.A., and J.S. Orr. 1975. Thyroid Hormone Metabolism. New York: Academic Press. Kirchengast, S. 1994. Body dimensions and thyroid hormone levels in premenopausal and postmenopausal women from Austria. Am. J. Phys. Anthropol. 94:487-497. Krolner, B., and S.P. Nielsen. 1982. Bone mineral content of the lumbar spine in normal and osteoporotic women: Cross-sectional and longitudinal studies. Clin. Sei. 62:329-336. Kwiecinski, G.G., D.A. Damassa, and A.W. Gustafson. 1991. Patterns of plasma sex hormone- binding globulin, thyroxine, and thyroxine binding globulin in relation to reproductive state and hibernation in female little brown bats. J. Endocrinol. 128:63-70. Legrand, J. 1986. Thyroid hormone effects on growth and development. In Thyroid Hormone Metabolism, G. Hennemann, ed. New York: Marcel Decker, 503-534. Lohman, T.G. 1992. Advances in Body Composition Assessment. Champaign, IL: Human Ki- netics Publications. Maruo, T., K. Katayama, E.R. Barnea et al. 1992a. A role for thyroid hormone in the induction of ovulation and corpus luteum function. Horm. Res. 37:12-18. Maruo, T., K. Katayama, H. Matsuo et al. 1992b. The role of maternal thyroid hormones in maintaining early pregnancy in threatened abortion. Acta Endocrinol. 127:118-133. Meikle, A.W., J.D. Stringham, M.G. Woodward et al. 1988. Hereditary and environmental influences on the variation of thyroid hormones in normal male twins. J. Clin. Endocri- nol. Metab. 66:588-592. Morton, N.E. 1974. Analysis of family resemblance. I. Introduction. Am. J. Hum. Genet. 26:318-330. Pimentel, E. 1987. Hormones , Growth Factors, and Oncogenes. Boca Raton, FL: CRC Press. Rao, D.C., P.M. Laskarzewski, J. A. Morrison et al. 1982. The Cincinnati Lipid Research Clinic Family Study: Cultural and biological determinants of lipids and lipoprotein concentra- tions. Am. J. Hum. Genet. 34:888-903. Rao, D.C., M. McGue, R. Wette et al. 1984. Path analysis in genetic epidemiology. In Human Population Genetics: The Pittsburgh Symposium, A. Chakravarti, ed. New York: Van Nostrand Reinhold, 35-81. Rao, D.C., N.E. Morton, C.J. Glueck et al. 1983. Heterogeneity between populations for mul- tifactorial inheritance on plasma lipids. Am. J. Hum. Genet. 35:468-483. Rao, D.C., W.R. Williams, M. McGue et al. 1983. Cultural and biological inheritance of plasma lipids. Am. J. Phys. Anthropol. 62:33-49. Reed, H.L., D. Brice, K.M.M. Shakir et al. 1990. Decreased free fraction of thyroid hormones after prolonged Antarctic residence. J. Appl. Physiol. 69:1467-1472. Reed, H.L., K.D. Burman, K.M.M. Shakir et al. 1986. Alterations in the hypothalmic-pituitary- thyroid axis after prolonged residence in Antarctica. Clin. Endocrinol. 25:55-65. Reed, H.L., M.M. D'Alesandro, K. Kowalski et al. 1992. Multiple cold air exposures change oral triiodothyronine kinetics in normal men. Am. J. Physiol. 263:E85-E93. Rice, T., G.P. Vogler, T.S. Perry et al. 1991. Familial aggregation of lipids and lipoproteins in families ascertained through random and nonrandom probands in the Iowa Lipid Re- search Clinics Family Study. Hum. Hered. 41:107-121. Riis, P.M., and A. Madsen. 1985. Thyroxine concentrations and secretion rates in relation to pregnancy, lactation, and energy balance in goats. J. Endocrinol 107:421-427. Rone, J.K., R.F. Dins, and H.L. Reed. 1992. The effect of endurance training on serum triio- dothyronine kinetics in man: Physical conditioning marked by enhanced thyroid hor- mone metabolism. Clin. Endocrinol. 37:325-330. http://www.jstor.org/page/info/about/policies/terms.jsp 760 / MARTIN AND CRAWFORD Schwartz, H.L. 1983. Effect of thyroid hormones on growth and development. In Molecular Basis of Thyroid Hormone Action , J.H. Oppenheimer and H.H. Samuels, eds. New York: Academic Press, 413-444. Segal, J., C. Buckley, and S.H. Ingbar. 1985. Stimulation of adenylate cyclase activity in rat thymocytes in vitro by 3,5,3 '-Triiodothyronine. Endocrinology 116:2036-2043. Segal, J., J. Hardiman, and S.H. Ingbar. 1989. Stimulation of calcium-ATPase activity by 3,5,3'- Triiodothyronine in rat thymocyte plasma membranes. J. Biochem. 261:749-754. Silverin, B., P.A. Viebke, J. Westin et al. 1989. Seasonal changes in body weight, fat deposits, and plasma levels of thyroxine and growth hormone in free-living great tits and willow tits. Gen. Comp. Endocrinol. 73:404-416. Sower, S.A., K.H. Karlson, and R.S. Fawcett. 1992. Changes in plasma thyroxine, estradiol- 17B and 17/20 dihydroxy4-preghen-3one during smoltification of coho salmon. Gen. Comp. Endocrinol. 85:278-285. Stryer, L. 1988. Biochemistry. New York: W.H. Freeman. Van Hardeveld, C. 1986. Effects of thyroid hormone on oxygen consumption, heat production, and energy economy. In Thyroid Hormone Metabolism , G. Hennemann, ed. New York: Marcel Decker, 579-608. Walawender, L.J., and M.H. Crawford. 1992. A quantitative analysis of plasma thyroxine levels from three Mennonite populations. Am. J. Phys. Anthropol. 7:73. Wright, S. 1918. On the nature of size factors. Genetics 3:367-374. http://www.jstor.org/page/info/about/policies/terms.jsp Article Contents p. [745] p. 746 p. 747 p. 748 p. 749 p. 750 p. 751 p. 752 p. 753 p. 754 p. 755 p. 756 p. 757 p. 758 p. 759 p. 760 Issue Table of Contents Human Biology, Vol. 70, No. 4 (August 1998), pp. i-ii, 643-811 Front Matter Clinal Variation in the Nuclear DNA of Europeans [pp. 643-657] Two Ethnic-Specific Polymorphisms in the Human Beta Pseudogene of Hemoglobin [pp. 659-666] Immunoglobulin Allotypes (GM and KM) in Basques from Spain: Approach to the Origin of the Basque Population [pp. 667-698] Genetic History of the Population of Sicily [pp. 699-714] Study of 15 Protein Polymorphisms in a Sample of the Turkish Population [pp. 715-728] Genetic Variability of Transferrin Subtypes in the Populations of India [pp. 729-744] Genetic and Environmental Components of Thyroxine Variation in Mennonites from Kansas and Nebraska [pp. 745-760] Impact of Maternal Body Build Characteristics on Newborn Size in Two Different European Populations [pp. 761-774] Birth Order and Male Homosexuality: Extension of Slater's Index [pp. 775-787] Brief Communications HLA-DQA1 and HLA-DQB1 Alleles and Haplotypes in Two Brazilian Indian Tribes: Evidence of Conservative Evolution of HLA-DQ [pp. 789-797] Evolution, Mutations, and Human Longevity: European Royal and Noble Families [pp. 799-804] Book Review Review: untitled [pp. 805-808] Instructions for Contributors [pp. 809-811] Back Matter