• No results found

Age-Specific Patterns of Genetic Variance in Drosophila melanogaster. II. Fecundity and Its Genetic Covariance With Age-Specific Mortality

N/A
N/A
Protected

Academic year: 2020

Share "Age-Specific Patterns of Genetic Variance in Drosophila melanogaster. II. Fecundity and Its Genetic Covariance With Age-Specific Mortality"

Copied!
10
0
0

Loading.... (view fulltext now)

Full text

(1)

Copyright 0 1996 by the Genetics Society of America

Age-Specific Patterns of Genetic Variance

in Drosophila melanogaster.

11.

Fecundity and Its Genetic Covariance With Age-Specific Mortality

Marc

Tatar, Daniel

E.

L. Promislow,' Aziz

A.

Khazaeli

and

James

W.

Curtsinger

Department of Ecology, Evolution and Behavior, University of Minnesota, St. Paul, Minnesota 55108

Manuscript received July 12, 1995

Accepted for publication February 29, 1996

ABSTRACT

Under the mutation accumulation model of senescence, it was predicted that the additive genetic variance (V,) for fitness traits will increase with age. We measured age-specific mortality and fecundity from 65,134 Drosophila melanogaster and estimated genetic variance components, based on reciprocal crosses of extracted second chromosome lines. Elsewhere we report the results for mortality. Here, for fecundity, we report a bimodal pattern for VA with peaks at 3 days and at 17-31 days. Under the antagonistic pleiotropy model of senescence, it was predicted that negative correlations will exist between early and late life history traits. For fecundity itself we find positive genetic correlations among age classes > 3 days but negative nonsignificant correlations between fecundity at 3 days and at older age classes. For fecundity us. age-specific mortality, we find positive fitness correlations (negative genetic correlations) among the traits at all ages > 3 days but a negative fitness correlation between fecundity at 3 days and mortality at the oldest ages (positive genetic correlations) . For age-specific mortality itselfwe find overwhelmingly positive genetic correlations among all age classes. The data suggest that mutation accumulation may be a major source of standing genetic variance for senescence.

P

ATTERNS of genetic variance and covariance for fitness traits play a crucial role in theories of life history evolution (ROW 1992; STEARNS 1992; CHARLES

WORTH 1994), and knowledge of the genetic variance- covariance matrix can shed light on three key issues. First, how and why is genetic variance for fitness traits maintained ( LANDE 1976; BULMER 1989; HOULE 1989; BARTON 1990)? Second, can we use the genetic vari- ance-covariance matrix to predict the direction of short- term evolutionary change (e.g., LANDE 1982; HOULE 1991 ) ? Third, and the focus of this paper, what is the evolutionary basis for senescence? In particular, can we use age-dependent changes in additive genetic variance

(V,) and covariance to test two specific models for the evolution of aging, mutation accumulation ( MEDAWAR 1952) and antagonistic pleiotropy (WILLIAMS 1957) ?

Both mutation accumulation and antagonistic pleiot- ropy theories grew out of the recognition that in an age-structured population, the force of natural selection declines with age (FISHER 1930; HALDANE 1941; MEDA-

WAR 1952). In the first case, MEDAWAR pointed out that deleterious mutant alleles with late-acting effects can in- crease in frequency since they are not strongly opposed by selection. Senescence then results from the age-re- lated increase in the frequency of mutations with late expressed deleterious effects (PARTRIDGE and BARTON

Currapmding author: Marc Tatar, Department of Ecology, Evolution

and Behavior, University of Minnesota, 1987 Upper Buford Circle,

St. Paul, MN 55108-6097. E-mail: mtatar@ecology.umn.edu

'

Present address: Department of Genetics, Life Sciences Bldg., Uni- versity of Georgia, Athens, GA 30602-7223.

Genetics 1 4 3 849-858 (June, 1996)

1993; CHARLESWORTH 1994). Models of this process sug- gest that it can lead to an age-related increase in additive variance for fitness traits ( CHARLESWORTH 1990).

In the second case, antagonistic pleiotropy, the fit- ness costs of late-acting deleterious mutations could be more than compensated for if those alleles have bene- ficial pleiotropic effects on fitness early in life. For ex- ample, increases in late-life mortality may be balanced by increases in early fecundity or rate of development. Most models of this process have predicted negative genetic covariance between at least some pairs of early- and late-age fitness traits ( LANDE 1980; ROSE 1982, 1985; CHARLESWORTH 1994; but see PARTRIDGE and BARTON 1993). However, while negative genetic corre- lations can be taken as support for antagonistic pleiot- ropy, failure to find negative correlations does not ex- clude the possibility that antagonistic pleiotropic alleles cause senescence (PEASE and BULL 1988; CHARLES-

WORTH 1990; HOULE 1991 ) .

Several quantitative genetic studies have estimated standing genetic variance to test the predictions of an age-related increase in V, for demographic fitness com- ponents, including fecundity and male mating ability (ROSE and CHARLESWORTH 1981a; ENGSTROM et al. 1992; HUGHES 1995). Taken together, the results are equivocal at best. ROSE and CHARLESWORTH (1981a) used a sib-mating design to estimate age-specific addi- tive variance for fecundity in Drosophila melanogaster.

(2)

850 M. Tatar et at.

related increase in additive variance in their study may have been biased downward by the differential loss, due to mortality at later ages, of genotypes with relatively high fecundity (CLARK 1987; ENGSTROM et al. 1992; PAR- TRIDGE and BARTON 1993). To avoid this difficulty, EN-

GSTROM et al. (1992) restricted their analysis of age- specific variance components of fecundity to females that survived for the duration of the assay period. Al- though ENGSTROM et al. ( 1992) found a dramatic in- crease in V,, their result may be confounded by the logarithmic transformation that they used on fecundity data ( G. ENGSTROM, personal communication)

.

This transformation can lead to an artifactual increase in variance with age for a trait that is actually Poisson dis- tributed and declines with age, as is the case for fecun- dity in Drosophila. KOSUDA (1985) also reports an in- crease in total genetic variance for male mating ability in D. melanogaster, but this analysis, as with HUGHES

( 1995), does not account for the increase in binomial sampling variance that accompanies the decrease in trait mean with age.

To test predictions arising from antagonistic pleiot- ropy, a second set of studies has used analyses of pedi- gree data to estimate genetic covariance between re- production and mean age of death. Again, when considered as a whole, no clear pattern emerges from the results. In Drosophila, genetic covariance between reproduction and mean age of death may be positive

( GIESEL 1986; HUGHES and CLARK 1988; ENGSTROM et al. 1992), negative (ROSE and CHARLESWORTH 1981a; TUCIC et al. 1988; SERVICE 1993), or not significantly different from zero ( SCHEINER et al. 1989; HUGHES 1995). In other invertebrates, the genetic covariances are predominantly positive, e.g., milkweed bug (HEG MANN and DINGLE 1982), bean beetle ( NOMURA and YONEZAWA 1990; TANAKA 1993), and Daphnia ( SPITZE

et al. 1991).

Unfortunately, a host of factors may detract from the interpretation of these studies. First, with one notable exception (HUGHES and CLARK 1988), adult life expec- tancy was used to measure mortality. But if we find a negative correlation between early fecundity and life ex- pectancy (i.e., mean age at death measured from sexual maturity), we cannot tell whether life expectancy de- clined due to an increase in mortality late in life

(as

predicted by antagonistic pleiotropy) or due to changes in mortality at early ages (perhaps due to immediate physiological costs of reproduction, e.g., PARTRIDGE and ANDREWS 1985; CHAPMAN et al. 1995). Furthermore, im- portant differences in mortality late in life may only prc- duce negligible differences in life expectancy from sex- ual maturity. For each of these reasons, the use of life expectancy weakens our ability to ascribe genetic correla- tions between early fecundity and late-age mortality.

Second, various factors can lead to spurious positive genetic correlations between traits and age classes. For

example, we might observe positive genetic correlations if genetic lines become inbred and previously masked deleterious recessive alleles are expressed, which could depress values for fitness related traits at all ages (ROSE 1984a; BELL and KOUFOPANOU 1986; CLARK 1987). Sim- ilarly, if a population is placed in a novel environment (e.g., from a natural to a laboratory setting) in those genotypes that are poorly adapted to the novel environ- ment, some or all fitness-related traits could decline (SERVICE and ROSE 1985; FRY 1993). Again, this could lead to positive genetic correlations that do not reflect the correlations existing in the natural population.

Finally, in a companion paper (PROMISLOW et al. 1996) , we pointed out several statistical pitfalls for those who wish to study the genetics of demographic charac- ters. At all but the largest sample sizes, we will likely underestimate the true additive variance for mortality rates at early and late ages and so may also underesti- mate the additive genetic covariance between fecundity and mortality. Further, for both mortality ( PROMISLOW

et al. 1996) and fecundity (as noted above)

,

if the data are incorrectly transformed one may draw erroneous conclusions.

Bearing these methodological and statistical prob- lems in mind, we set out to estimate genetic variance components for fecundity as a function of age, as well as patterns of genetic covariance both within and between age-specific fecundity and mortality. Our aim was to test predictions arising from models for the evolution of senescence, but at the same time to minimize the im- pact of potentially confounding factors discussed above. To that end, we estimated age-specific patterns of fecun- dity and mortality throughout the life span of the co- hort. We used flies from a large laboratory-adapted pop- ulation and a breeding design that eliminated the risk of inbreeding effects. Finally, we used a large number of flies ( -35,000 females and 30,000 males) to minimize sampling error in our mortality estimates.

Taken together, these data on age-specific mortality and fecundity provide us with a powerful tool for a fine- scale examination of the structure of genetic variances and covariances, their causes, and their consequences.

MATERIALS AND METHODS

We estimated genetic variance components for fecundity and mortality in D. melanogaster using a reciprocal partial dial- lel cross (North Carolina I1 cross, COMSTOCK and ROBINSON 1952; also see HUGHES 1995) of second chromosome balancer lines ( DOBZHANSKY and SPASSKY 1953; details for these stocks in FUKUI et at. 1996) . We have described the results for mortal- ity in PROMISLOW et at. ( 1996), where details of the genetic crosses and maintenance procedures are elaborated. Briefly, 53 balancer second chromosome lines were extracted from a large, outbred, laboratory population from which we selected

(3)

Genetic Variance for Life Histories 85 1

6-10; these crosses yielded 25 lines, plus reciprocals, of sec- ond chromosome heterozygote flies. The adult offspring of these final heterozygous lines were used to measure age-spe- cific mortality and fecundity. Two cohort cages, each of -650 flies (mixed sex), were set up for each of the heterozygous lines and their reciprocals for a total of 100 cages. Cohorts were collected over a 48-hr period.

Experimental stocks were derived from W 3 5 0 (WEBER and DIGGENS 1990), a laboratory population founded in 1981 and maintained in population cages at 24", constant light and 55- 65% relative humidity on agar-yeast-molasses-cornmeal me- dia. In the experiments reported here, the environmental conditions and media are the same of those of LF350. Bal- ancer lines were derived from the LF350 base populations and in the process were maintained in either 8-dram vials or half-pint milk bottles for a total of 12 generations. At the 12th generation, these lines were crossed and the adults of the next generation were placed in 3.8-1 cages made of inverted plastic jars. These cages were designed in such a way that we could remove dead flies daily to estimate age-specific mortal- ity; a screen covered the jar mouth, a screen window on the side provided air flow, and a gasket aperture provided a way to aspirate dead flies out of the cage. To feed flies, dishes of food ( 68 cm2, fashioned from the j a r lids) were placed against the screen at the bottom of the cage. Females laid eggs through the screen onto the food medium.

Fresh food was provided four times each week through the end of the study; on Sunday and Wednesday afternoons (between 1600 and 1700 h r ) , and on Monday and Thursday mornings (between 900 and 1000 h r )

.

With this schedule, between Sunday night and Monday morning, and between Wednesday night and Thursday morning, females were ex- posed to fresh food for a set period of 17 hr. In the first five weeks, starting at age 3 days, we counted eggs on these "17- hr" dishes to estimate age-specific fecundity. We stopped cen- susing eggs in the middle of the fifth week when no cage produced more than five eggs, even though all cages had at least 18 females alive at this time.

We estimated the number of eggs laid on each dish by count- ing all the eggs within each of six pie-shaped radial sections. Each section covered 1 / 48 of the total dish area and the six sections were distributed uniformly around the dish. The radial sections were superimposed on the medium by placing over the food dish a plastic petri plate with etched polar coordinates.

Eggs were counted within 48 hr of collection. Dishes were

stored at 4" until counted to prevent hatching.

We calculated average per capita fecundity for each cage at each age by dividing the total number of counted eggs by the number of females present at the census. Fecundity, consisting of counts, is Poisson distributed and will naturally show a correlation between mean and variance. Accordingly, all estimates of fecundity were square-root transformed ( S o -

KAL and ROHLF 1995 ) .

Variance component analysis: The genetic analyses were

conducted using Quercus, a package of maximum likelihood analysis of quantitative genetic data ( SHAW 1987; SHAW and SHAW 1992). With Qwcus we can analyze quantitative traits from individuals of known pedigree and evaluate the signifi- cance of variance components with log-likelihood ratio tests (see PROMISLOW et al. 1996). We used the restricted maxi- mum likelihood (REML) option to find unbiased estimates of components of variance (V,, additive; V,, dominance; V,, environmental) and maternal effects (V,) for age-specific fecundity. As in the companion paper on mortality ( PROMIS.

LOW et al. 1996), here we treat each cage as an individual in the model, since per capita fecundities are known for each cage but not for each fly within a cage. In the genetic analysis

1 2 3 4 5

Age, weeks

FIGURE 1 .-Per-capita fecundity for females, square-root- transformed, measured biweekly.

then, cages are treated as individuals with known pedigree based on their maternal and paternal stocks, and our esti- mates of variance components reflect the variation among cohorts and not among individuals within cages. This ap- proach is appropriate since all the individuals within a given reciprocal cross are genetically identical at the second chro- mosome. However, our interpretation of environmental varia- tion is consequently unconventional since it refers only to cage effects and overlooks the among-individual variation within cages; our measures of VE are underestimates and should not be used to calculate heritability.

We also used Quercus to estimate the additive genetic corre- lations from the data on fecundity presented in this paper and from the mortality data presented in PROMISLOW et al.

( 1996). Mortality data include estimates for weeks 1-8 for males and weeks 1-7 for females. We found the genetic corre- lations between fecundities across each pair of age classes, between mortality rates across each pair of age classes and between fecundity and mortality rate across each pair of age classes. We analyzed mortality rates as in PROMISLOW et al.

( 1996) with the use of logarithm-transformed values of instan- taneous mortality, that is, In ( p , ) , where pr x -In (P,) and

P, is the probability of surviving from age x to age x

+

1

( ELANDT-JOHNSON and JOHNSON 1980). For some pairs of

traits, the REML algorithm would not converge within our iteration limits. In these cases, we present point estimates of additive genetic correlation (marked by parentheses in the tables and figures). The point estimates were calculated from the mean square and cross products estimated from a general linear model ( COCKERHAM and WEIR 1977). Since our design lacks block replicates, we cannot conduct hypothesis tests with these point estimates.

In all analyses with Quercus, we estimated variance compo- nents assuming an inbreeding coefficient of 1.0 in parents, which accounts for the identity by descent of the second chro- mosome among individuals within each of the parental lines.

RESULTS

Fecundity: Daily per capita fecundity reached a maxi- mum of eight to 16 eggs (Figure 1 ) . The levels of egg production we observe are smaller than those we find for females housed singly in vials in our laboratory

(4)

852 M. Tatar d nl.

due partly to the behavior of females in cohort cages relative to vials and to the fact that some eggs stick to the screen. However, this bias in the mean should have little effect on the relative differences in the number of eggs laid among genotypes. The square-root-trans- formed fecundity data are approximately normal [ SHA- PIRO-WILKS test ( SHAPIRO and WILKS 1965) on the re- sidual values determined from the genetic variance component analyses: only Days 21 and 31 show a sig- nificant difference from normality, respectively, P = 0.037 and P

<

0.0001

3.

After transformation there is no mean-variance correlation ( = 0.05, F'>, = 0.426, P = 0.53) ; the phenotypic variance of the square-root- transformed values appears roughly constant across all ages.

We calculated genetic variance component5 for gross fecundity (the within-cage sum of daily per capita fe- cundity) for comparison with previous studies. Additive genetic variance for gross fecundity is marginally greater than zero ( V I = 0.190, log-likelihood ratio test

X:

= 3.39, P = 0.07), and the dominance variance is small but significant (VI, = 0.025, log-likelihood ratio test

x:

= 3.98, P

<

0.05). We can compare the additive variance estimate to those of other studies, scaled for trait means, with the coefficient of additive genetic vari- ance,

CY,,

= &/mean. For gross fecundity, CV,, =

6.46, which is similar to the value for Drosophila male mating ability reported in HUGHES ( 1995), but some- what lower than the values for female fecundity calcu- lated by CHARLESWORTH ( 1987).

We find a bimodal pattern for age-specific V,, for fecundity (Figure 2 ) . Additive genetic variance for fe- cundity is significantly greater than zero at age 3 days, is not measurably different from zero at ages 7, 10 and

14 days, and is greater than zero from ages 17-31 days, excluding age 21 days (each age, log-likelihood ratio test, d.f. = 1, P

<

0.01 )

.

The bimodal pattern remains after we correct for multiple comparisons; the V,, at ages 3, 28 and 31 days retain significance ( a = 0.05,

Dunn-SidBk correction, SOKAL and ROHLF 1995). We also tested for differences in additive variance between ages using ficrfl, a two sample version of Quercus that compares the variance components among indepen- dent populations (see PROMISLOW et al. 1996). These tests support the suggestion that V,, for fecundity is bimodal with respect to age: V,, for fecundity at age 14

days is significantly less than at age 3 days and at ages

17-31 days exclusive of age 21 days (each pair of days: log-likelihood ratio test, d.f. = 1 , P

<

0.03), and the variance for fecundity at age 7 days is less than the variance at ages 28 and 31 days (each pair of ages: log- likelihood ratio test, d.f. = 1 , P

<

0.05 )

.

However none of these tests retain significance after correction for multiple comparisons ( a = 0.05, Dunn-SidBk correc- tion, SOKAL and ROHLF 1995). We also note that these

*t * t

C

*E

0.025-

s

0.02

-

0.015

-

0.01

-

0.005

-

0 -r

.

.

.

1 . . . . .

3 7 10 14 17 21 24 28 31

Age, days

FIGLIRE 2. Age-specific variance components for fecundity.

Additive variance (V,,) is shown in the top of the figure.

The bottom, plotted on the same vertical scale, illustrates

the estimates for dominance variance ( V I , ) , environmental

variance (V,) and maternal effects (V,)

.

Note that VE here

does not include environmental variation among individuals (see text). Additive variance is significantly greater than zero

at the 95% confidence level ( * ) when not corrected for multi-

ple comparisons, and

( t )

when corrected for multiple com-

parisons (see text).

data are consistent with a linear age-specific increase in

V,, with age (across ages,

r"

= 0.46, P

<

0.05). In this case the VA at age 3 days may simply reflect a positive deviation from this regression. To resolve whether V, at age 3 days is biologically real, we would need to conduct genetic analyses of fecundity on a finer scale. Finally, dominance variance is small in magnitude rela- tive to the additive variance (Figure 2 ) and different from zero at ages 14, 28 and 31 days, but only when we do not correct for multiple comparisons (log-likelihood ratio test, d.f. = 1 , P

<

0.05). Maternal effects (Figure

2 ) were negligible and not significantly different from zero in any age class. Measures of V, at advanced ages may be depressed by the early extinction of high fecun- dity genotypes, but this bias is unlikely here since fewer than 5% of females died by the age at which we ob- served a significant increase in V,, (age 17 days). Fur- thermore, in all the cages at least 18 females were still alive by the age when reproduction was essentially com- pleted (day 35)

.

Finally, we note that these age-depen- dent patterns for genetic variance may be confounded with temporal changes in the laboratory environment. For instance, the genetic variance components for age- specific mortality show differences in the amount of genetic variation among blocks run at different times

(5)

Genetic Variance for Life Histories 853

3 day8

"v

I I I 1 I

7 10 14 17 21 24 28 31

Age of fecundity, day

FIGURE 3.-The additive genetic correlations between mea- sures of fecundity at each age us. all subsequent ages. Each line represents the correlation between fecundity at one age

(indicated by the label for the line) and fecundity at all subse- quent ages (indicated on the abscissa). Symbols in parenthe-

ses represent point estimates based on linear models and are made without estimates of errors (see text). All the other

points were estimated by REML, and the filled symbols repre- sent correlations significantly different from zero without cor-

rection for multiple comparisons (see text).

such period effects, it will be necessary that future stud- ies on the genetics of demographic traits examine repli- cate cohorts that are initiated at different times.

Genetic Correlation: We estimate additive genetic co- variance between three sets of age-specific traits: ( 1 ) fecundity, ( 2 ) fecundity and mortality in females, and

( 3 ) mortality in females and in males. We estimate covariances even when the additive genetic variance of one trait is not significantly greater than zero, because this failure to reject the null hypothesis does not imply that the variance actually is zero.

( 1 ) The genetic correlation between fecundity at dif- ferent ages shows two distinct patterns (Figure 3 )

.

First, there is no statistically significant relationship between fecundity at age 3 days and any other age. However six of eight correlations are negative in value, which may suggest a trade-off between early and late fecundity, although this proportion of negative values is not sig- nificantly different from random (sign test, exact prob- ability = 0.109). Second, we find positive genetic corre- lations among all fecundities at ages > 3 days. Each of the pair-wise correlations involving fecundities after age 14 days are significantly greater than zero (each pair, log-likelihood ratio test, d.f. = 1, P

<

0.05). Since we are most interested in the overall sign of the correla- tions among age-specific fecundities, we note that the table-wide sign is significantly positive (sign test, normal approximation, SOKAL. and ROHLF 1995, t, = 3.77, n =

34, P

<

0.001 ) .

( 2 ) As with fecundity alone, we find that the patterns of correlation between fecundity and age-specific mor- tality are qualitatively different for early us. later ages

-1.54

fl

(-2.17)

I I I I

3 4 5 6 7

Age of mortality, week

FIGURE 4,"Genetic correlation between age-specific fe- male mortality and fecundity. Each line represents fecundity at a given age (indicated by the label for the line) vs. mortality

at ages 1-7 weeks (indicated on the abscissa). Symbols in parentheses represent point estimates based on linear models and are made without estimates of errors (see text). All the other points were estimated by REML, and the filled symbols represent correlations significantly different from zero with- out correction for multiple comparisons (see text). Values

>

?2 are given rather than plotted.

(Figure 4 )

.

Fecundity at age 3 days has a positive ge- netic correlation with mortality in 6 of the

7

weeks, and the correlation was greatest and statistically significant with mortality estimated at week

7

(log-likelihood ratio test,

x:

= 4.06, P

<

0.05)

.

Since mortality is the comple- ment of age-specific survival, this positive genetic corre- lation between fecundity and mortality represents a negative correlation between fitness components. In contrast, for ages > 3 days the genetic correlations be- tween fecundity and mortality are negative in 49 of 63 cases, i.e., there are positive correlations between the fitness components of reproduction and survival. As the age of the cohort advances, there is an increase in both the magnitude of the correlations and the occurrence of statistical significance (log-likelihood ratio test, d.f.

= 1, P

<

0.01, for all pairs of mortality at weeks 5 and

6 us. fecundity at ages 21-31 days; log-likelihood ratio test, d.f. = 1, P

<

0.05, for mortality week 4 us. fecundity age 28 days and for mortality week

7

us. fecundity age 31 days). The table-wide pattern of correlations taken over all ages is significantly negative in sign (sign test, normal approximation, t, = 4.41, n = 63, P

<

0.001 ) ;

overall, the fitness components of age-specific fecundity and survival are positively correlated.

( 3 ) Last we present the genetic correlations among age-specific mortalities (Table 1 ) . For males, we find a nearly uniform pattern of positive correlations (26 of

(6)

854 M. Tatar et al.

TABLE I

Genetic correlations between age-speciiic mortality estimates

Week

Week 2 3 4 5 6 7 8

Males

1 (2.71) 0.6454 0.937 0.777 0.778 0.723 0.898

2 0.861 1.249 0.838 0.808 0.838 0.3642

3 0.5815 0.4244 0.1914 0.0943 -0.27

4 0.941 0.825 0.791 0.1897

5 0.969 0.904 0.0485

6 1.008 0.3967

7 0.8398

Females 1 (0.68) 0.228 -0.396 -0.248 0.5769 1.3195

2 0.7611 0.1143 0.009 0.0322 -0.554

3 0.85 7 0.656 0.582 0.2635

4 0.862 0.749 0.4008

5 1.004 0.866

6 0.973

Entires represent genetic correlations. Log-likelihood ratio test,

x;,

of unconstrained covariance relative to covariance con- strained to zero. Bold entries, P < 0.01; italic entries, significance could not be determined due to lack of convergence of the maximum likelihood algorithm when the covariance was constrained to zero; parentheses entries, point estimates based on linear model, no error gstimates available (see text).

also find a tendency toward positive genetic correlations among age-specific mortalities, but the case is not as strong as for males. We find positive correlations in

17

of 20 cases, and P

<

0.10 for three pairs (Table 1 )

.

As

with males, the table-wide pattern of correlation is significantly positive (sign test, normal approximation, t, = 3.13, n = 20, P

<

0.01 )

.

Overall, these results indicate that there is a strong genetic association among mortality rates across ages in both males and females; again, age-specific fitness components are positively cor- related.

DISCUSSION

The results described here bear on two models for the evolution of senescence: mutation accumulation

( MEDAWAR 1952) and antagonistic pleiotropy ( WIL LIAMS 1957). In the introduction to this paper, we dis- cussed two expectations that have been derived from MEDAWAR’S and WILLIAMS’ proposals, an age-related in- crease in VA for fitness components under mutation accumulation and a negative genetic correlation be- tween early-age and late-age fitness components for an- tagonistic pleiotropy.

We designed the present study to test the mutation accumulation model by estimating age-specific genetic variance components for fecundity, and to test the an- tagonistic pleiotropy model by testing for negative ge- netic correlations between early-age and late-age fitness components. The experiment was designed to over-

come many of the problems encountered by previous studies, including inbreeding, small genetic and demo- graphic sample size, and the effects of novel environ-

ments (ROSE 1984a; BELL and KOUFOPANOU 1986;

CHARLESWORTH 1987; CLARK 1987; PARTRIDGE and BAR-

T O N 1993). Our results indicate that both mechanisms may contribute to the senescence observed in our labo- ratory population of

D.

melanogaster, but they also sug- gest that mutation accumulation may be the dominant source of standing genetic variance for late-age fitness traits.

Fecundity: In a companion paper, we showed that

V, for age-specific mortality increased early in life and then declined at later ages ( PROMISLOW et al. 1996)

.

In marked contrast, for fecundity we find significant VA

very early in life, at age 3 days. At ages 7-14 days, there was no detectable additive variance for fecundity, but after age

17

days V, became significant once again. The late rise in V, is consistent with the prediction for mutation accumulation as a cause of senescence in

D.

melanogaster. However, since egg laying stops after week

5 in these stocks and in many wild-type stocks, we can make no inferences about patterns of variance at ages as late as those observed for mortality rates.

(7)

Genetic Variance for Life Histories 855

fecundity (ROSE 1985)

.

However, this seems unlikely since the negative trade-offs we observe between these traits primarily involve mortality rates at the latest ages where the fitness effects of death will be negligible. On the other hand, antagonistic pleiotropy across age classes of fecundity may provide a basis for maintaining V, for fecundity at age 3 days, and we observe a ten- dency toward such negative genetic correlations, as have other pedigree-based analyses of genetic covari-

ance in

D.

melanogaster (ROSE and CHARLESWORTH

1981a; TUCIC et al. 1988; SCHEINER et al. 1989). How- ever, substantial dominance variance is expected to arise if selection maintains VA for life history traits through antagonistic pleiotropy ( CURTSINGER et al. 1994; HUGHES 1995), and we find little evidence for dominance variance for either age-specific fecundity or mortality ( PROMISLOW et al. 1996). Second, VA for early fecundity may be a consequence of selection that main- tains variation in larval developmental rate. Variation in larval development rate could, in turn, lead to varia- tion in the age at first reproduction or peak reproduc- tion (e.g., HIRAIZUMI 1961; HUDAK and GROMKO 1989; ROPER et al. 1993). Thus, when same-aged females are assayed, variation in the rate of ovarian maturation in adults will appear as genetic variation for the rate of early egg production. Further empirical and theoretical work is required to evaluate this hypothesis.

Covariance among fitness components: Models of se- nescence based on antagonistic pleiotropy predict neg- ative genetic correlations between early-age fecundity and late-age mortality (ROSE 1991; but see PARTRIDGE and BARTON 1993). Studies of standing genetic covari- ance in

D.

melanogaster, both through sib-analysis and selection, claim to support this prediction (ROSE and CHARLESWORTH 1981a,b; ROSE 1984b; LUCKINBILL et al. 1984; TUCIC et al. 1988; SERVICE 1993; ZWAAN et al. 1995; but see ENGSTROM et al. 1992; PARTRIDGE and FOWLER 1992; STEARNS and KAISER 1993). However, these re- ports are each limited in some crucial way ( CURTSINGER et al. 1995). For instance, selection studies typically in- crease the longevity of a strain by restricting reproduc- tion to specific ages, but this practice produces direct selection both on life span and on the schedule of fe- cundity. Thus, we cannot isolate the correlated from the direct selection responses ( CURTSINGER et al. 1995; ZWAAN et al. 1995)

.

Only recently has ZWAAN et al. (1995) selected on life span without direct pressure on fecundity.

Furthermore, for both selection and sib-analysis, stud- ies that compare fecundity to longevity cannot make the critical distinction between effects on mortality that are associated with senescence from those that are inde- pendent of aging. Reproduction may affect longevity because it is “risky” (PARTRIDGE 1987); age-specific mortality may be increased only at or near the time of reproduction, as appears to be the case for male

D.

melanogaster (PARTRIDGE and A N D R E W ~ 1985)

.

Repro- duction may also increase mortality by an equal amount at all ages in which case longevity is decreased but the rate of senescence is unaffected as occurs in the beetle Callosobruchus maculatus ( TATAR et al. 1993; TATAR and CAREY 1995). Finally, reproduction may affect late-age mortality alone or it may increase the rate of change of mortality with age. In these cases, reproduction would affect senescence since the changes reflect a progressive decline in fitness as a function of age (ROSE 1991 )

.

We report a positive genetic correlation between early fecundity and late-age mortality or, in the usual frame- work, a negative correlation with late-age survival. To our knowledge, this is the first demonstration of a trade- off between these demographic parameters, and it sup- ports the notion that early fecundity has a genetic trade- off explicitly with senescence.

Although we find at least one set of negative fitness correlations, the fitness correlations in this study are dominated by positive values, not only between age- specific fecundity and mortality, but also between age classes within fecundity and within mortality. Positive fitness correlations have often been observed in genetic studies (REZNICK 1985; BELL and KOUFOPANOU 1986; GIESEL 1986; HUGHES and CLARK 1988; ENGSTROM et al. 1992; HUGHES 1995) and are sometimes explained as the result of inbreeding or of novel environments (ROSE 1984a; SERVICE and ROSE 1985 ) . Neither of these factors are likely to account for our results since our demographic analysis involved second chromosome heterozygote flies that originated from a large labora- tory-adapted population, and we conducted our experi- ments under the same environmental conditions of this base population.

Positive fitness correlations could arise as a conse- quence of the functional structure of physiological trade-offs that underlie the relationship between traits

(PEASE and BULL 1988; CHARLESWORTH 1990; HOULE

1991; DE JONG and VAN NOORDWIJK 1992). HOULE

( 1991 ) and CHARLESWORTH ( 1390) developed muta- tion-selection balance models that incorporated such functional structure in terms of independent loci for resource allocation and acquisition. Both authors found that at least some positive genetic correlations result when mutations affect a large number of loci for acqui- sition relative to allocation. The positive genetic correla- tions that we observed, therefore, may be due to a pre- dominance of genetic variance for acquisition despite the occurrence of genetic trade-offs between traits. However, none of these models specify across which ages we should expect to find positive rather than nega- tive genetic correlations. The nature of the age struc- ture of genetic correlations observed in our data sug- gests a need for greater specificity of such models.

(8)

856 M. Tatar et al.

with pleiotropic deleterious late-age effects. The genetic lines analyzed in our study originated from population cages and were then passed through two-week discrete culture for 12 generations. In discrete culture there is great potential for the accumulation of de novo muta- tions because, in practice, flies older than 1 week con- tribute very little to the next generation. Since life his- tory mutations are likely to have substantial positively correlated effects ( MUKAI andYAMAzAKll971; SIMMONS

et al. 1980; HOULE et al. 1994), deleterious effects on age-specific fecundity and mortality will tend to cose- gregate at older ages. In this way, mutations with age- specific effects could lead to positive fitness correlations among traits at ages as early as

7

days post-eclosion and lead to senescence marked by the decline in demo- graphic fitness traits.

The feasibility of this explanation is tempered by two issues. First, 12 generations in discrete culture are not likely to produce substantial mutational variance

( CHARLESWORTH 1987). Potentially, mutation accumu-

lation at ages beyond

7

days also occurred in the LF350 stock, but the actual age-dependent intensity of selec- tion is unknown for the demography of population cages. Second, when the deleterious effects of muta- tions act across early and late ages, selection at the early ages may forestall their accumulation (PARTRIDGE and

BARTON 1993; HOULE et al. 1994). The extent to which mutational effects correlate across ages is a critical but largely unresolved question ( HOULE et al. 1994; CLARK and GUADALUPE 1995 )

.

We tentatively conclude that the patterns of genetic variance for age-specific fecundity and of covariance for age-specific mortality and fecundity are consistent with the simultaneous action of both antagonistic pleiotropy and mutation accumulation. This agrees with SERVICE

et al.’s (1988) interpretation of the evolution of post- poned senescence in

D.

melanogaster. SERVICE et al.

( 1988) applied reverse selection for early reproduction in ROSE’S (1984b) late reproducing lines and found that starvation resistance decreased as early reproduc- tion increased. They interpret this correlated response as evidence that antagonistic pleiotropy played a role in the original increase of starvation resistance in the late reproducing lines. At the same time, they found no change in two other traits, ethanol resistance and desiccation resistance, that also increased in ROSE’S

(1984b) late-reproducing lines. SERVICE et al. (1988) interpret this stasis as evidence that mutation accumula- tion was the source of the original variation of ethanol and desiccation resistance in the base population.

Our study advances in two important ways the notion that both mutation accumulation and antagonistic plei- otropy may play an important role in the evolution of senescence. First, SERVICE et al. (1988) did not report how life span responded to reverse selection on early fecundity, even though an increase in life span was the

central demographic trait used by ROSE (1984b) to in- fer postponed senescence in these lines. Here, we ex- plicitly analyze the genetic correlates of age-specific mortality. We suggest that mutation accumulation is a major source of genetic variance for life span and senescence.

However, we also find a genetic basis for trade-offs across age-specific traits, which is consistent with antag- onistic pleiotropy as a source of variation for senes- cence. This observation relates to our second point.

PARTRIDGE and BARTON ( 1993) emphasized that to

fully explain the evolution of senescence, we need to understand the relative contribution of antagonistic pleiotropy and mutation accumulation to observed pat- terns of physiological and demographic deterioration. In our population of

D.

melanogaster, we suggest that mutation accumulation plays the dominant role since the fitness correlations among age-specific traits begin- ning as early as age

7

days were overwhelmingly positive. While the analysis of standing genetic variation will not tell the whole story about the evolution of senescence

(PARTRIDGE and BARTON 1993), our analysis provides

a clear implication for how we interpret experiments that select for postponed senescence. If mutations with late-age effects accumulate in laboratory-adapted popu- lations before they are subjected to selection, then se- lection for late-age fitness components will produce a rapid response for improvement across many late-age life history traits. Even if antagonistic pleiotropy makes a contribution to the selection response, much of the phenotypic difference between selected and control lines will reflect the culling of late-acting deleterious mutations and not the effects of changes in gene fre- quency for loci with alternative fitness effects across age classes.

For their help with conceptual, experimental and statistical issues, we thank P. ABRAMS, J. CAREY, B. CHARLESWORTH, G. ENGSTROM, C. GEYER, H. FUKUI, P. HARVEY, K. HUGHES, L. PARTRIDGE, S. PLETCHER, F. SHAW, R. SHAW, M. TURELLI, J. VAUPEL, and once again, Dr. DOGMA. For their steady technical assistance, we thank J. ASHI.EY, K. BRAAS, H. FUKUI, K. KELLY, A. KIRSCHER, C. MISIAK, S. PLETCHER, A. RESLER,

N. STAUS, and D. WOODWARD. This work was supported by National Institutes of Health grants AGO971 and AG11722 to J.W.C.

LITERATURE CITED

BARTON, N. H., 1990 Pleiotropic models of quantitative variation. Genetics 1 2 4 773-782.

BELL, G., and V. KOUFOPANOU, 1986 The cost of reproduction, pp. 83-131 in Oxjmd Suwtys in Evolutionary Biology, edited by R. DAWNS and M. RIDLEY. Oxford University Press, Oxford. BULMER, M. G., 1989 Maintenance of genetic variability by muta-

tion-selection balance: a child’s guide through the jungle. Ge- nome 31: 761-767.

CHAPMAN, T., L. LIDDLE, J. KALB, M. WOLFNER and L. PARTRIDGE, 1995 Male seminal fluid components cause the cost of mating

in Drosophila melanogaster females. Nature 373: 241-244.

CHARLESWORTH, B., 1987 The heritability of fitness, pp. 21-40 in

Sexual Selection: Testing the Alternatives, edited by J. W. BRADBURY

(9)

Genetic Variance for Life Histones 857

CHARLESWORTH, B., 1990 Optimization models, quantitative genet- ics, and mutation. Evolution 44: 520-538.

CHARLESWORTH, B., 1994 Evolution i n Age-Structured Populations.

Cambridge University Press, Cambridge.

CIM, A. G., 1987 Senescence and the geneticcorrelation hang- up.

Am.

Nat. 1 2 9 932-940.

CLARK, A. G., and R. N. GUADALUPE, 1995 Probing the evolution of senescence in Drosophila melanogaster with Pelement tagging. Genetica 96: 225-234.

COCKERHAM, C. C., and B. S. WEIR, 1977 Quadratic analyses of recip rocal crosses. Biometrics 33: 187-203.

COMSTOCK, R. E., and H. F. ROBINSON, 1952 Estimation of average dominance of genes, pp. 494-516 in Heterosis, edited by J. W. GOWEN. Iowa State College Press, Ames, IA.

CURTSINGER, J. W., P. M. SERVICE and T. PROUT, 1994 Antagonistic pleiotropy, reversal of dominance, and genetic variation. Am. Nat. 144 210-228.

CURTSINGER, J. W., H. H. FUKUI, A. W L I , A. KIRSCHER, S. PLETCHER et al., 1995 Genetic variation and aging. Annu. Rev. Genet. 29: 553-575.

DE JONG, G., and A. J. VAN NOORDWJK, 1992 Acquisition and alloca- tion of resources: genetic (co)variances, selection, and life his- tories. Am. Nat. 139: 749-770.

DOBZHANSKY, T., and B. SPASSKY, 1953 Genetics of natural popula- tions. XXI. Concealed variability in two symmetric species of

Drosophila. Genetics 3 8 471 -484.

ELANDT-JOHNSON, R., and N. L. JOHNSON, 1980 Survival Models and

Data Analysis. Wiley, New York.

ENGSTROM, G., L.-E. LIIJEDAHL and T. BJORKLUND, 1992 Expression of genetic and environmental variation during ageing. 2. Selec- tion for increased life span in Drosophila melanogaster. Theoret. Appl. Genet. 8 5 26-32.

FISHER, R. A., 1930 The Genetical Theoly of Selection. Clarendon Press, Oxford.

FRY, J. D., 1993 The “general vigor” problem: can antagonistic plei- otropy be detected when genetic covariances are positive. Evolu- tion 47: 327-333.

FUKUI, H. H., L. ACKART and J. W. CURTSINGER, 1996 Deceleration of age-specific mortality rates in inbred and non-inbred Drosoph-

ila melanogaster. Exp. Geront. (in press).

GIESEL, J. T., 1986 Genetic correlation structure of life history vari- ables in outbred, wild Drosophila melanogaster: effects of photope- riod regimen. Am. Nat. 128: 593-603.

HALDANE, J. B. S., 1941 New Paths i n Genetics. Allen and Unwin, London.

HEGMANN, J. P., and H. DINGLE, 1982 Phenotypic and genetic covari- ance structure in Milkweed Bug life history traits, pp. 177-185

in Evolution and Genetics of Lye Histories, edited by H. DINGLE and

J. P. HEGMANN. Springer-Verlag, New York.

HIRAIZUMI, Y., 1961 Negative correlation between rate of develop- ment and female fertility in Drosophila melanogaster. Genetics 46: 615-624.

HOULE, D., 1989 The maintenance of polygenic variation in finite populations. Evolution 43: 1767-1780.

HOULE, D., 1991 Genetic covariance of fitness correlates: what ge- netic correlations are made of and why it matters. Evolution 43: 630-648.

HOULE, D., K. A. HUGHES, D. K. HOFFMASTER, J. I m , S. A~SIMACO. ~ 0 ~ 1 . 0 s et al., 1994 The effect of spontaneous mutation on quantitative traits. I. Variance and covariance of life history traits. Genetics 138: 773-785.

HUDAK, M. J., and M. H. GROMKO, 1989 Response to selection for early and late development of sexual maturity in Drosophila mela-

nogaster. Anim. Behav. 38: 344351.

HUGHES, K. A,, 1995 The evolutionary genetics of male life-history characters in Drosophila melanogaster. Evolution 49: 521-537. HUGHES, K. A., and B. CHARLESWORTH, 1994 A genetic analysis of

senescence in Drosophila. Nature 367: 64-66.

HUGHES, D. M., and A. G. C I1988 ~ Analysis of the genetic struc- ture of life history of Drosophila melanogaster using recombinant extracted lines. Evolution 42: 1309-1320.

KIRKWOOD, T. B. L., and M. R. ROSE, 1991 Evolution of senescence: late survival sacrificed for reproduction. Phil. Trans. R. SOC. Lond. B 3 3 2 15-24.

KOSUDA, K., 1985 The aging effect o n male mating activity in D r e

sophila melanogaster. Behav. Genet. 15: 297-303.

LANDE, R., 1976 The maintenance of genetic variability by mutation in a polygenic character with linked loci. Genet. Res. Camb. 26: 221-35.

LANDE, R., 1980 The genetic covariance between characters main- tained by pleiotropic mutations. Genetics 94: 203-215. LANDE, R., 1982 A quantitative genetic theory of life history evolu-

tion. Ecology 3: 607-615.

LUCKINBILI., L. S., R. ARKING, M. J. CLARE, W. C. CIROCCO and S. A. BUCK, 1984 Selection for delayed senescence in Drosophila mela-

nogaster. Evolution 38: 996-1003.

MEDAWAR, P. B., 1952 An unsolved problem of biology, pp. 28-55

in The Uniqueness of the Individual. Dover, New York.

MUM, T., and T. YAMAZAKI, 1971 The genetic structure of natural populations of Drosophila melanogaster. X. Developmental time and viability. Genetics 69: 385-398.

NOMURA, T., and K. YONEZAWA, 1990 Genetic correlations among life history characters of adult females in the Azuki Bean Weevil,

Callosobruchw chinensis (L.) (Coleoptera: Bruchidae) . Appl. En-

tomol. 2001. 25: 423-430.

PARTRIDGE, L., 1987 Is accelerated senescence a cost of reproduc- tion? Func. Ecol. 1: 317-320.

PARTRIDGE, L., and R. ANDREWS, 1985 The effect of reproductive activity on the longevity of male Drosophila melanogaster is not caused by an acceleration of aging. J. Insect Physiol. 31: 393-395. PARTRIDGE, L., and N. H. BARTON, 1993 Optimality, mutation and

the evolution of aging. Nature 362: 305-311.

PARTRIDGE, L., and K. FOWLER, 1992 Direct and correlated re- sponses to selection of age at reproduction in Drosophila melane

gaster. Evolution 46: 76-91.

PEASE, C. M., and J. J. BULL, 1988 A critique of methods for measur- ing life history trade-offs. J. Evol. Biol. 1: 293-303.

PROMISLOW, D. E. L., M. TATAR, A. A. WLI and J. W. CURTSINGER, 1996 Age-specific patterns of genetic variance in Drosophila mel-

anogaster. I. Mortality. Genetics 143: 839-848.

REZNICK, D., 1985 Costs of reproduction: an evaluation of the empir- ical evidence. Oikos 4 4 257-267.

ROFF, D. A., 1992 The Evolution ofL+ Histories. Chapman and Hall, New York.

ROPER, C., P. PIGNATELLI and L. PARTRIDGE, 1993 Evolutionary ef- fects of selection on age at reproduction in larval and adult

Drosophila melanogaster. Evolution 47: 445-455.

ROSE, M. R., 1982 Antagonistic pleiotropy, dominance, and genetic variation. Heredity 48: 63-78.

ROSE, M. R., 1984a Genetic covariation in Drosophila life history: untangling the data. Am. Nat. 123: 565-569.

ROSE, M. R., 1984b Laboratory evolution of postponed senescence

in Drosophila melanogaster. Evolution 38: 1004-1010.

ROSE, M., 1985 Life history evolution with antagonistic pleiotropy and overlapping generations. Theor. Pop. Biol 2 8 342-358. ROSE, M. R., 1991 Evolutionaly Biology of Aging. Oxford University

Press, New York.

ROSE, M. R., and B. CHARLESWORTH, 1981a Genetics of life history

in Drosophila melanogaster. I. Sib analysis of adult females. Genetics

97: 173-186.

ROSE, M. R., and B. CHARLESWORTH, 1981b Genetics of life history

in Drosophila melanogaster. 11. Exploratory selection experiments.

Genetics 97: 187-196.

SCHEINER, S. M., R. L. CApwand R. F. LYMAN, 1989 A search for trade-offs among life history traits in Drosophila melanogaster. Evol. Ecol. 3: 51-63.

SERVICE, P., 1993 Laboratory evolution of longevity and reproduc- tive fitness components in male fruit flies: mating ability. Evolu- tion 47: 387-399.

SERVICE, P. M., and M. R. ROSE, 1985 Genetic covariation among life-history components: the effect of novel environments. Evolu- tion 39: 943-945.

SERVICE, P. M., E. W. HUTCHINSON and M. R. ROSE, 1988 Multiple genetic mechanisms for the evolution of senescence in Drosophila

melanogaster, Evolution 42: 708-716.

(10)

858 M. Tatar et al.

SHAW, R. G., 1987 Maximum-likelihood approaches applied to quan- titative genetics of natural populations. Evolution 4 0 812-826. SHAW, R. G., and F. H. SHAW, 1992 QUERCUS: programs for quan-

titative genetic analysis using maximum likelihood. Published electronically o n the Internet, available via anonymous ftp from ecology.umn.edu: directory path pub/quercus and from

ftp.bio.indiana.edu; directory path biology/quantgen/quercus.20.

SIMMONS, M. J., C. R. PRESTON and W. R. ENGEIS, 1980 Pleiotropic effects on fitness of mutations affecting viability in Drosophila melanogaster. Genetics 94: 467-475.

SOW, R. R., and F. J. ROHLF, 1995 Biometly. W. H. Freeman, New York.

SPITZE, R, J. BURNSON and M. LYNCH, 1991 The covariance structure of life-history characteristics in Daphnia Pulex. Evolution 4 5 1081-1090.

STEARNS, S. C., 1992 TheEuolution OfLijeHisturies. Oxford University Press, Oxford.

STFARNS, S. C., and M. KAISER, 1993 The effects of enhanced expres- sion of elongation factor EF-la on lifespan in Drosophila melanogas- tu. lV. A summary of three experiments. Genetica 91: 167-182. TANAKA, Y., 1993 A genetic mechanism for the evolution of senes-

cence in Callosobruchw chinensis ( t h e azuki bean weevil). Hered- ity 70: 318-321.

TATAR, M., and J. R. CAREY, 1995 Nutrition mediates reproductive trade-offs with age-specific mortality in the beetle Callosobruchw maculatus. Ecology 76: 2066-2073.

TATAR, M., J. R. CAREY and J. W. VAUPEL, 1993 Long term cost of

reproduction with and without accelerated senescence in Culloso- bruchus maculatus: analysis of age-specific mortality. Evolution 47:

TUCIC, N., D. CVETKOVI~: and D. MILANOVIC, 1988 The genetic varia-

tion and covariation among fitness components in Drosophila melanogaster females and males. Heredity 60: 55-60.

WEBER, K. E., and L. T. DIGGINS, 1990 Increased selection response in larger populations. 11. Selection for ethanol resistance in D m

sophila melanogaster. Genetics 125: 585-597.

WILLIAMS, G . C., 1957 Pleiotropy, natural selection, and the evolu- tion of senescence. Evolution 11: 398-41 1 .

ZWAAN, B., R. BIJLSMA and R. F. HOEKSTRA, 1995 Direct selection on life span in Drosophila melanogaster. Evolution 49: 649-659. 1302-1312.

Figure

FIGURE  1 .-Per-capita fecundity for females, square-root- transformed,  measured biweekly
FIGURE 3.-The  additive  genetic line  represents  the  correlation between  fecundity made  without  estimates  of sent  correlations  significantly  different  from  zero quent ages (indicated on  the abscissa)
TABLE I Genetic  correlations  between  age-speciiic mortality estimates

References

Related documents

Expected Results: improved public relations due to ADA entrance to mansion; 10% increase in mansion bookings; offer art classes at the Cutrer Cultural Arts Center; rental of

In this section, we turn from tensor networks to genuine field theories defined on p -adic spacetimes: either in the bulk of the tree T p (or possibly a quotient by a Schottky group)

The edge coe ffi cients are shown su ffi cient for closely reconstructing the images, while contour representations by means of chains of edges reduce the information redundancy

Specically, we consider the absorption rate of an ideal reverberant random sound eld as dened in..

The importance of charge optimization has also been demonstrated by a study looking at the effect of overdosing a system with an excess of micropaticles accompanying cPAM