Evolution, mutations, and human longevity: European royal and noble families

NATALIA S. GAVRILOVA,1,2 LEONID A. GAVRILOV,1,3 GALINA N. EVDOKUSHKINA,2 VICTORIA G. SEMYONOVA,2 ANNA L. GAVRILOVA,1 NATALIA N. EVDOKUSHKINA,2 YULIA E. KUSHNAREVA,3 VYACHESLAV N. KROUTKO,3 AND ALEXANDER YU. ANDREYEV3

Abstract The evolutionary theory of aging predicts that the equilibrium gene frequency for deleterious mutations should increase with age at onset of mutation action because of weaker (postponed) selection against later-acting mutations. According to this mutation accumulation hypothesis, one would expect the genetic variability for survival (additive genetic variance) to increase with age. The ratio of additive genetic variance to the observed phenotypic variance (the heritability of longevity) can be estimated most reliably as the doubled slope of the regression line for offspring life span on paternal age at death. Thus, if longevity is indeed determined by late-acting deleterious mutations, one would expect this slope to become steeper at higher paternal ages. To test this prediction of evolutionary theory of aging, we computerized and analyzed the most reliable and accurate genealogical data on longevity in European royal and noble families. Offspring longevity for each sex (8409 records for males and 3741 records for females) was considered as a dependent variable in the multiple regression model and as a function of three independent predictors: paternal age at death (for estimation of heritability of life span), paternal age at reproduction (control for parental age effects), and cohort life expectancy (control for cohort and secular trends and fluctuations). We found that the regression slope for offspring longevity as a function of paternal longevity increases with paternal longevity, as predicted by the evolutionary theory of aging and by the mutation accumulation hypothesis in particular.

The evolutionary theory of aging predicts that the equilibrium gene frequency for deleterious mutations should increase with age at onset of mutation action because of weaker (postponed) selection against later-acting mutations (Medawar 1946, 1952; Finch 1990; Rose 1991; Partridge and Barton 1993; Charlesworth 1994). According to this mutation accumulation hypothesis, one would expect the genetic variability for survival (additive genetic variance) to increase with age (Partridge and Barton 1993; Charlesworth 1994). In general, both the additive genetic component of variance and the dominant component are expected to increase with age under the mutation accumulation hypothesis (because for traits affected by rare deleterious alleles, both components increase with increasing mutant allele frequency) (Charlesworth 1987; Falconer 1989; Hughes and Charlesworth 1994). The ratio of additive genetic variance to the observed phenotypic variance (the heritability of longevity) can be estimated most reliably as the doubled slope of the regression line for offspring life span on paternal age at death (the regression of offspring on mothers sometimes gives too high an estimate because of maternal effects, as it would, for example, with body size in most mammals) (Falconer 1989). That is why the slope of the regression line on paternal age at death was used in this study as an estimate for heritability of human longevity. Thus, if longevity is indeed determined by late-acting deleterious mutations, one would expect this slope to become steeper at higher paternal ages (Gavrilova et al. 1997). The purpose of this brief communication is to report the preliminary results of a test of this prediction of the evolutionary theory of aging and of the mutation accumulation hypothesis in particular.

Materials and Methods

In this study the most reliable and accurate existing data on human familial longevity were collected, computerized, and analyzed: the genealogical data on longevity in European royal and noble families published in Van Hueck's Genealogisches Handbuch des Adels (1980,1985, 1986,1987,1988, 1991, 1992, 1993, 1994) and in other professional genealogical sources listed elsewhere (Gavrilov et al. 1996). These data were chosen to minimize the social heterogeneity of the population under study and to avoid overstating the familial component of longevity when a mixture of families with different social statuses is analyzed. Thus, although the sample analyzed in this study is biased toward higher social status and does not represent the whole human population, it is the best possible sample where the effects of population heterogeneity are minimized with regard to social status.

Offspring longevity was analyzed for adults (those who survived by age 30) to study the effects of late-acting mutations. The data for offspring born in the twentieth century were excluded from the analysis so that the estimates of longevity for extinct birth cohorts were unbiased. The data for offspring born before the nineteenth century were also excluded to minimize the heterogeneity of the population.

For each birth cohort the sex-specific mean expectation of life at age 30 was calculated and used as a dependent variable in a multiple linear regression to control for cohort and secular effects on human longevity. Offspring longevity for each sex (8409 records for males and 3741 records for females) was considered a dependent variable in the multiple regression model (program 1R in the BMDP statistical package) and a function of three independent predictors: (1) paternal age at death (for estimation of heritability of life span), (2) paternal age at reproduction [a control for parental age effects, because recent studies have demonstrated that daughters born to older fathers live shorter lives (Gavrilov et al. 1996, 1997; Gavrilov and Gavrilova 1997a,b)], and (3) sex-specific cohort life expectancy (a control for cohort and secular trends and fluctuations in life span).