Allozymic differentiation in response to laboratory demographic selection of Drosophila melanogaster.
Just as there are alleles likely to cause aging through their deleterious effects late in life, so there may be alleles that enhance longevity through beneficial effects late in life. Such alleles should be kept at a low frequency by being neutral or slightly deleterious early in life. Of course, such alleles should be much rarer than alleles causing aging, since most mutations are deleterious and even a slight disadvantage early in life could eliminate an allele from the population.
In Drosophila melanogaster, postponed aging can be accomplished by allowing only the flies that can reproduce late in life to contribute to the next generation. Selecting flies for late-life reproduction has resulted in postponed aging in sets of Drosophila spp. stocks with independent origins (Wattiaux 1968a,b; Rose and Charlesworth 1981; Luckinbill et al. 1984; Rose 1984). For D. melanogaster there is no statistical evidence for the involvement of fewer than 100 loci in postponed aging (Hutchinson and Rose 1990; Fleming et al. 1993).
Because flies selected for postponed aging have enhanced resistance to stresses such as starvation, desiccation, and forced flight (Service et al. 1985; Graves and Rose 1990), gene loci that encode metabolic enzymes are expected to be among the loci at which allele frequencies change in response to selection, Electrophoretic surveys of a dozen metabolic enzymes have been performed on the long-lived and short-lived D. melanogaster populations of Luckinbill et al. (1989). They found allele frequencies to differ only at the glucose-6-phosphate dehydrogenase locus, though their low effective population sizes may have limited the diversity of genotypes available to selection.
Enzyme electrophoresis has been used previously to correlate laboratory selection with allele frequency changes at allozyme loci. Foundational work in this area included the demonstration that allele frequencies at the amylase locus could be changed by maintaining D. melanogaster populations on food varying in starch content (Hickey 1977). Cavenet and Clegg (1981) reported responses to ethanol selection at the alcohol dehydrogenase, alpha-glycerophosphate dehydrogenase, malate dehydrogenase, and 6-phosphogluconate dehydrogenase loci. The alcohol dehydrogenase locus has also been shown to respond to temperature selection (Oakeshott 1979) and accelerated development (Knibb 1987). Carfagna et al. (1980) used the phosphoglucomutase locus to demonstrate frequency-dependent and beterotic selection. Bijlsma and Kerver (1983) showed differential survival among genotypes in the presence of DDT for the penrose phosphate shunt enzymes glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase.
While all of the above studies deal with laboratory evolution, a much greater number of studies have utilized electrophoretic techniques to support hypotheses of past evolution in natural environments. Watt et al. have provided the best example of the evolutionary importance of an allozyme locus to natural insect populations. They showed that phosphoglucose isomerase (PGI) allozymes in Colias butterflies differ consequentially in kinetics and thermal stability. These biochemical differences affect daily flight capacity such that they can be used to predict differences among PGI genotypes in daily flight capacity. The flight differences result in genotypic differences in survival, male mating success, and female fecundity (Watt 1992, 1994). Using models assuming fitness is proportional to flux through a biochemical pathway, Hastings (1992) has suggested that selection pressures acting on enzyme variants may be much higher than previously thought.
This study evaluates responses to selection at several loci that encode enzymes of possible relevance to life history. Allele frequencies at these loci are calculated for populations of D. melanogaster kept on discrete generations of characteristic lengths. All the populations have in common the genetic background of a single founding population. Their current selection regimes represent an entire spectrum of different generation times. Each generation time is represented by five replicate populations.
MATERIALS AND METHODS
This study utilized six fivefold-replicated D. melanogaster stocks: B, O, CB, CO, ACB, and ACO. The B and O populations were formed in February 1980 (Rose 1984) from a single ancestral generation of the Ives stock obtained in 1975 from Massachusetts (Ives 1970) [ILLUSTRATION FOR FIGURE 1 OMITTED]. The long-lived Os have been maintained on 10-week generations since 1981, while their controls, the Bs, have been maintained on two-week generations like the Ives stock. The mean female longevity is 32.8 days for the Bs and 55.6 days for the Os (Chippindale et al., pers. comm.).
CB and CO are stocks of intermediate generation length derived from the Bs and Os, respectively [ILLUSTRATION FOR FIGURE 1 OMITTED], in 1988 (Rose et al. 1992). They serve as controls for stocks derived from them, including both the accelerated stocks mentioned below and stress selected stocks which determine the CB/CO generation time but otherwise are not relevant to this study. The CB/CO generation time is on the order of five weeks, which favors relatively late fecundity, compared to the two-week generation time of the Bs. The mean female longevity is 38.8 days for the CBs and 46.5 days for the COs.
The ACB and ACO stocks were derived from the CB and CO lines, respectively, such that the replicate number of each line corresponds to the CB or CO line from which it was derived. The ACB and ACO lines had completed approximately 100 generations of selection for accelerated development at the time of this study, each generation lasting less than two weeks. Mean female longevity equals 33.0 days for the ACBs and 31.1 days for the ACOs.
Selection for accelerated development was accomplished by allowing only approximately the first 13% of flies to emerge from pupa to contribute eggs to the next generation. Development time had decreased significantly after ten generations of selection. Viability is lower for these stocks than for stocks with longer development times (Chippindale 1994; Chippindale et al. 1994).
Individual female flies were homogenized and subjected to starch gel electrophoresis. The sample size was approximately one hundred flies for every population except O1 and O5; for these two populations five hundred flies were analyzed. For the five B populations and the five O populations, gels were stained for alcohol dehydrogenase, [Alpha]-glycerol-3-phosphate dehydrogenase, phosphoglucomutase, and CuZn-superoxide dismutase, among other enzymes (Pasteur et al. 1988; Manchenko 1994). Flies from the CO, CB, ACO, and ACB populations were stained for only the first four enzymes. Allele frequencies were calculated from the allozymes resolved on the gels, and tested for deviation from Hardy-Weinberg equilibrium using the [[Chi].sup.2] statistic.
The two main hypotheses to be tested are: (1) do the O populations, selected for postponed reproduction, have different allele frequencies from those of the baseline B populations; and (2) do the populations selected for accelerated development have different allele frequencies from those of the control populations? By using a factorial ANOVA of arcsine-transformed allele frequencies, we were able to condense our analyses to a single statistical test on the data for each locus, thus reducing the error associated with multiple comparisons. The three population types (parental, control, and accelerated development) were regarded as one factor, and the two original lineages (B and O) as another. The differentiation associated with population type was subjected to a Tukey-HSD post hoc test in order to identify differentiation associated with selection for accelerated development. The statistical tests were performed using the multivariate general linear hypotheses section of Systat software (Wilkinson 1990).
For the O1 and O5 lines, maximum likelihood estimation of gamete frequencies permitted evaluation of linkage between the phosphoglucomutase and CuZn-superoxide dismutase loci, which are both on the third chromosome. The overall hypothesis that none of the disequilibrium coefficients differ from zero was tested using the method of Weir (1990). In this equation, n is the gamete count, D is the linkage disequilibrium constant of loci u and v, and k and l are the number of alleles at locus u and locus v. respectively.
[Mathematical Expression Omitted]
[TABULAR DATA FOR TABLE 1 OMITTED]
Throughout the stocks, there are two allozymes each for alcohol dehydrogenase (ADH), [Alpha]-glycerol-3-phosphate dehydrogenase (GPDH) and CuZn-superoxide dismutase (SOD). They are designated F and S for faster and slower migration in the gel. Three allozymes for phosphoglucomutase (PGM) are present in the B and O stocks; they are designated F, M, and S.
Genotype frequencies in all the populations are at Hardy-Weinberg equilibrium. No deviations from Hardy-Weinberg equilibrium were supported, after Bonferroni correction to the [[Chi].sup.2] tests. Linkage disequilibrium between the Pgm and Sod loci is small. The linkage disequilibrium coefficients are 0.017 for the O1 population and 0.012 for the O5 population. The [[Chi].sup.2] values are 0.62 for the O1 population and 0.36 for the O5 population; at the P = 0.05 level, the critical [[Chi].sup.2] value is 3.84. Hence, the linkage disequilibrium coefficients do not differ significantly from zero.
Selection for Postponed Reproduction
Comparison of the O lines to the B lines shows allele frequencies diverging at two of the four polymorphic enzyme loci. While change was insignificant at the Adh and Gpdh loci, selection for postponed reproduction changed allele frequencies significantly at the Sod and Pgm loci (Table 1). At the Adh locus, allele frequencies are on average the same in the O lines as in the B lines. While the mean Gpdh allele frequencies of the Bs and Os appear divergent, allele frequencies at this locus are highly variable, and there is no statistically significant differentiation associated with postponed reproduction (Table 1). In contrast, at the Sod locus, an increase in the frequencies of the previously rare S allele is highly statistically significant (P [less than] 0.001). Significant change also occurred in the frequencies of the Pgm S allele (P [less than] 0.001). In all five B populations, M is the most abundant allele, F is relatively common, and S is generally rare. In all five O populations, the S allele is much more common, often exceeding the M allele in abundance.
Flies with Intermediate Generation Lengths
At the Adh, Gpdh, and Sod loci, the CB and CO lines do not differ significantly from the B and O lines (Appendix). However, the Pgm locus is exceptional (Table 1). In the CB lines, relative to the B lines, the abundance of the F allele has been halved. In the CO lines, relative to the O lines, the S allele has become substantially less common, while the M allele has increased in frequency. In both the CB and the CO lines, the frequency of the S allele has changed in the same direction longevity has changed [ILLUSTRATION FOR FIGURE 2 OMITTED].
Flies Selected for Accelerated Development
In the accelerated development lines, the more abundant allele generally increased in frequency at all four loci, and fixations were common (Table 1). At the Adh and Pgm loci, the increases represent significant differentiation between accelerated development and control stocks (Appendix). At the Gpdh locus, there is a trend toward fixation from the parental to the most derived stocks, but the accelerated development stocks are not significantly different from their controls, the CBs and COs (Appendix). At the Sod locus, the S allele is undetectable in every ACB line as well as three ACO lines (Table 1). These apparent fixations constitute significant divergence of the accelerated development lines, at least compared to the parental lines (Appendix), but for the ACB lines they represent small increases in frequency of the F allele, since the S allele was already rare in the B parental stock. Consistent fixation across accelerated development lines is more apparent at the Pgm locus (Table 1). At this locus, not only do the fixations represent highly significant changes in allele frequency, but some constitute large increases in the proportion of the M allele. An overview of the demographically selected stocks shows a progressive increase in the frequency of the Pgm S allele with mean longevity [ILLUSTRATION FOR FIGURE 2 OMITTED].
Both selection for postponed reproduction and selection for accelerated development can be correlated with significant differentiation in allele frequencies. Responses to selectively postponed aging were seen at the Sod and Pgm loci, while responses to selectively accelerated development were seen at the Adh and Pgm loci, and possibly the Sod locus. The allele frequency changes associated with the postponed aging lines represent robust responses to selection, as they are both substantial and consistent in direction across fivefold replicated populations. Since allele frequency changes in the accelerated development lines are relatively small in some cases, it may be desirable to consider genetic drift as a cause for allozymic differentiation of the accelerated development lines from their controls. Subsequently the results will be discussed by locus.
The lines selected for accelerated development were derived from populations in which the more common allozyme was already encoded by a strong majority of alleles. However, the initial allele frequencies of the accelerated development lines cannot be assumed to equal the allele frequencies of the CB and CO lines, even if the ACBs and ACOs were representative samples, because the ACB and ACO lines are separated from the CB and CO lines studied here by over 25 generations of selection in the CB and CO lines. Nevertheless, a combination of founder effect and finite population size could make genetic drift a viable explanation of the fixation of common alleles.
Effective population size is a key issue for the Adh and Sod loci, since the S allele at each of these loci is relatively rare throughout all the stocks. If we assume that the initial allele frequencies in the accelerated development lines are equal to the allele frequencies of their controls, changes resulting in fixations range from less than 1% to greater than 20%. These changes should require up to 2N generations for a neutral allele (Hartl and Clark 1989, p. 72). While the effective sizes of the populations are not precisely known, the population size is approximately 1000 flies. If we conservatively assume that the effective population size is 500, up to 1000 generations are still required for the fixations observed, given neutral alleles. Moreover, for both Adh and Sod, the direction of allele frequency change is consistent for nine of 10 accelerated development populations. Thus the significant increase in frequency of the Adh F allele is likely to be a real effect of selection.
Alcohol Dehydrogenase and [Alpha]-Glycerophosphate Dehydrogenase
Heinstra et al. (1987) suggested that natural selection may act on the Adh polymorphism in larvae, and indeed the selection regime imposed on the ACB and ACO populations acted upon the larval stage. Our association of Adh F allele frequency increases with accelerated development concurs with the findings of Knibb et al. (1987) who selected D. melanogaster for fast, intermediate, and slow development times. Their fast development selection regime, like our own, reduced the time to the first emergence of adults by using only the earliest flies to emerge from the pupa as parents for the next generation. Over the course of selection the frequency of the Adh F allele rose in three of four fast development lines. Cavener's (1983) study of developmental rate selection provides a contrasting result. He selected for fast and slow development in fivefold replicated populations from each of two different geographical localities. The Adh F allele frequency did not change significantly in the populations Cavenet selected for fast development. McKechnie and Geer (1988) have demonstrated epistasis between Adh and Gpdh in the determination of larval ethanol tolerance. Larvae of the [Adh.sup.FF] [Gpdh.sup.FF] and [Adh.sup.SS] [Gpdh.sup.SS] allelic combinations are more ethanol tolerant than larvae with other combinations. So the tendency for the Gpdh F allele to increase in the ACB and ACO populations may have been affected by the increase in the Adh F allele. Izquierdo and Rubio (1989) found the allele frequencies at [Alpha]Gpdh to be determined to a considerable extent by genotypic combinations with the Adh F allele. On the other hand, they detected a difference in development rate between [Alpha]Gpdh alleles, with the [Alpha]Gpdh S allele associated with a faster rate of development. Barnes et al. (1989) also found that flies with the [Alpha]Gpdh S/S genotype developed slightly faster. Hence, selection for accelerated development may place conflicting pressures on the [Alpha]Gpdh locus, which could explain its lack of response to selection in our study. Regardless, the Adh results found here support earlier findings indicating that the F allele at the locus fosters rapid development.
One of the most consistent and significant results of the present study is the association between increased S allele frequencies of superoxide dismutase and increased lifespan. Given this finding, what interpretation can be placed on it? Superoxide dismutase, as a free radical scavenger, has long been regarded as a probable factor in aging according to the free radical theory of aging (Harman 1956). Superoxide dismutase (SOD) catalyzes the conversion of superoxide anion radicals to hydrogen peroxide, which is converted to water by catalase. Overexpression of SOD and catalase retards accumulation of oxidative damage of age. Flies with an extra dose of the Sod and Cat genes exhibit a more gradual increase in mitochondrial hydrogen peroxide generation and slower accumulation of 8-hydroxydeoxyguanosine, a product of DNA oxidation (Sohal et al. 1995). They also had less protein oxidative damage according to two different measures: protein carbonyl levels (Orr et al. 1994) and rate of decrease in glucose-6-phosphate dehydrogenase activity oxidation (Sohal et al. 1995). Bovine as well as Drosophila SOD can make D. melanogaster more resistant to oxidative stress (Fleming et el. 1992). The flies treated with a bovine Sod gene averaged SOD expression 32.5% greater than normal. These transgenic flies had a longer lifespan than the controls; the difference was slight but significant (Reveillaud et el. 1991). For an increase in longevity it may be necessary to enhance the activity of both SOD and catalase; it was hypothesized that the toxicity of hydrogen peroxide limited SOD activity. Seto et al. (1990) reported that there was no change in oxygen metabolism or longevity when they added an additional copy of Sod, despite increases in Sod transcription and SOD activity. On the other hand, Orr et al. (1994) reported that the lifespan of transgenic flies with three copies of Drosophila Sod genes was extended up to one third.
Either Sod or a locus linked to it indeed responds to selection for postponed reproduction. If the favored locus is not Sod itself, it must be closely linked; Tyler et al. (1993) eliminated the possibility of Sod falling within an inversion polymorphism by showing a general absence of inversion polymorphisms in the Ives D. melanogaster. The S allozyme, which corresponds to the allele found at elevated frequencies in the fly lines with postponed aging, is known to be more active than the average F allozyme (Lee et al. 1981). In any case, the transgenic studies strongly indicate the causal involvement of more active Sod alleles in increasing lifespan. The results from selection for this locus, then, corroborate the transgenic findings.
The Sod F allele actually consists of a series of haplotypes indistinguishable by starch gel electrophoresis, while the S allele consists of an invariant sequence that differs from one of the Fast haplotypes at only the nucleotide site that encodes the amino acid difference between the two electrophoretically distinguishable allozymes (Hudson et al. 1994). While the invariance of the S allele supports selection, both the Slow haplotype and the Fast haplotype most like it are abundant. Therefore, the variant favored by selection is probably not the amino acid difference that changes electrophoretic mobility, such that the differentiation between B and O lines presented here is an underestimate. The lack of identity between the electrophoretic polymorphism and the selected variant may explain the failure of Tyler et al. (1993) to find a significant effect of Sod genotype on longevity.
Tests for linkage disequilibrium could also be confounded by the existence of a selectively favored Sod allele that includes the S allele but is not delimited by it. While no linkage disequilibrium between the Sod and Pgm loci was detected, it could have been obscured by our inability to distinguish between subtypes of the Sod F allele. The number of gametes carrying Pgm S and Sod S, the two alleles hypothesized to be selectively favored, is an underestimate of the number of gametes carrying the Pgm S allele and the selectively favored Sod allele. Smit-McBride et al. (1988) showed that linkage disequilibrium between Pgm and Sod is possible. They detected it in two of six experimental populations studied but in neither of two natural populations studied.
While the possibility of linkage to an unidentified selected locus exists, phosphoglucomutase is likely to be important for life-history parameters because of its key role in metabolism. Gu (1991) associated the polymorphism at Pgm with variation in flight capacity of the moth Epiphyas postvittana. Goulson (1993) found that the genotype at Pgm affected the length of time for which individual Maniola jurtina butterflies could fly continuously. Watt et al. (1985) discovered a Pgm heterozygote advantage in male mating success. Carfagna et el. (1980) provided evidence that the Pgm polymorphism has selective significance in D. melanogaster. They started two populations at each of three different frequencies of the "[Pgm.sup.B]" allele, 0.02, 0.50, and 0.98. All six populations converged to 10% to 15% "[Pgm.sup.B]."
In our study, the phosphoglucomutase locus responded to demographic selection with electrophoretically detectable divergence of substantial magnitude. The S allozyme appears strongly favored by selection for postponed reproduction, and the longevity of a fly stock can be predicted from the frequency of the Pgm S allele [ILLUSTRATION FOR FIGURE 2 OMITTED]. Furthermore, the frequency of the S allele averages less than 0.01 in four of five of the B lines, making a substantial increase without direct selection dependent on strong linkage disequilibrium. Selection intensity must exceed recombination fraction in order for linkage disequilibrium to increase, and an increase in linkage disequilibrium would be expected primarily when the favored allele is rare (Asmussen and Clegg 1981). In order for the Pgm S allele to increase from an average frequency of 0.01 to an average frequency of 0.51, a high proportion of the favored alleles would have to be linked to the Pgm S allele. If every copy of the favored allele can be traced back to a single copy, 100% of the copies of the Pgm S allele should be linked to the favored allele. This could happen only if the mutant later to be favored by selection for postponed aging were linked to the very rare Pgm S allele in every newly formed replicate O line. This is particularly unlikely given the large sizes and long establishment of these lines, and thus opportunities for recombination.
The Pgm M allele has become fixed in every line selected for accelerated development. Not only can it be argued that selection for accelerated development favors the M allele, relative to the S allele, but two pieces of evidence support recessive expression of the M allele. Firstly, its presence in the O lines may signify that M/S heterozygotes have fitness under selection for postponed reproduction comparable to that of S/S homozygotes. Secondly, its consistency of fixation over all the accelerated development lines might require more generations for a dominant allele. To corroborate this hypothesis, four sets of fifty simulations were performed using the genetic drift and selection option of the Populus software developed by Don Alstad of the University of Minnesota. All four sets of simulations assumed a selection coefficient of 0.05 and a starting frequency of the M allele of 0.75. These values were chosen based on the assumption that the real values, which are unknown, should be less than these values, making the estimates conservative. I set the population size at either 500 or 300 and assumed the M allele to be either completely dominant or completely recessive. For a population size of 500, fixation required 311 [+ or -] 158 generations when the allele was assumed to be dominant. When the allele was assumed to be recessive, fixation required 85 [+ or -] 30 generations. For a population size of 300, a dominant allele required 286 [+ or -] 200 generations for fixation, while a recessive allele required 77 [+ or -] 26 generations. Since the accelerated development lines were formed about one hundred generations ago, dominance of the M allele is highly unlikely, in the absence of a selection coefficient substantially greater than 5%. Codominance is possible, as the average fixation time with codominance is on the order of 135 [+ or -] 50 generations. However, the consistency of fixation across five populations argues against the likelihood of codominance explaining fixation within one-hundred generations.
Since genetic drift appears to be unable to account for the fixation of the Pgm M allele in all accelerated development populations, this allele must be favorable rather than neutral in a short generation-time environment. This Pgm allele or an allele at a linked locus appears to be subject to antagonistic pleiotropy, since it is favorable for reproduction at early ages but relatively deleterious for reproduction at late ages. The Pgm S allele, or a linked allele, is favorable at late ages but kept at a low frequency in the absence of selection for postponed reproduction. It appears to be relatively deleterious in flies reproduced before late ages, as shown by the CO populations. The COs were derived from the Os, and are maintained on a shorter generation time. The frequency of the Pgm S allele has fallen substantially in these populations (Table 1).
The dominance of the Pgm S allele relative to the Pgm M allele may slow its progress toward fixation in populations selected for postponed aging. Moreover, heterozygotes may have a greater probability than homozygotes of reproducing late in life, such that the Pgm S allele is prevented from reaching fixation despite its correlation with longevity. With regard to the Sod locus, Peng et al. (1991) found that F/S heterozygotes were more productive than homozygotes. While these flies were reproduced early in life only, the result is consistent with our findings of relatively modest increases in the Sod S allele.
Dominance of the alleles that responded to selection (Sod S and Pgm S) over the alleles that would affect lifespan in most flies (Sod F and Pgm M) is consistent with the findings of Buck et al. (1993) who studied the localization and regulation of longevity determinant genes in a selected strain of D. melanogaster. They showed that mean longevity is determined in large part by recessive genes on the third chromosome. Luckinbill et al. (1988) had used chromosome substitution lines to reveal that the third chromosome accounts for 66% to 72% of the observed variation in female longevity. Sod and Pgm are indeed located on the third chromosome. Buck et al. (1993) further showed that the second chromosome can inhibit the third chromosome in its longevity determinant capacity, and that the first chromosome can negate this action of the second chromosome. Graf and Ayala (1986) have identified a second chromosome modifier of Sod, which further emphasizes the consistency of the Sod locus with the expression pattern described by Buck et al.
We have identified two loci which contribute to longevity in D. melanogaster: the locus encoding phosphoglucomutase and the locus encoding CuZn-superoxide dismutase. Both of the loci may be subject to antagonistic pleiotropy, as selection for the longevity-opposing life history of accelerated development appears to render the longevity-associated alleles deleterious. Currently we are expanding our study to other species of Drosophila to determine whether the same or different loci are associated with postponed aging in other species.
Many hours of technical assistance were provided by M. Arias, D. T Frank, S. J. Mee, G. M. Vazquez, and C. P. You. This research was supported by the estate of Robert H. Tyler and by US-PHS grant AG09970 to M.R.R. Helpful comments on previous versions of the manuscript were provided by R. R. Hudson, L. D. Mueller, D. N. Reznick, and three anonymous reviewers.
ASMUSSEN, M. A. AND M. T. CLEGG. 1981. Dynamics of the linkage disequilibrium function under model of gene frequency hitchhiking. Genetics 99:337-356.
BARNES, P. T., B. HOLLAND, AND V. COURREGES. 1989. Genotype-by-environment and epistatic interactions in Drosophila melanogaster: the effects of glycerol phosphate dehydrogenase allozymes, genetic background, and rearing temperature on larval development time and viability. Genetics 122:859-868.
BIJLSMA, R., AND J. W. KERVER. 1983. The effect of DDT on the polymorphism at the Glpd and Pgd loci in Drosophila melanogaster. Genetics 103:447-464.
BUCK, S., R. A. WELLS, S. P. DUDAS, G. T. BAKER III, AND R. ARKING. 1993. Chromosomal localization and regulation of the longevity determinant genes in a selected strain of Drosophila melanogaster, Heredity 71:11-22.
CARFAGNA, M., L. FUCCI, L. GAUDIO, G. PONTECORVO, AND R. RUBINO. 1980. Adaptive value of PGM polymorphism in laboratory populations of Drosophila melanogaster. Genet. Res. 36:256-276.
CAVENER, D. R. 1983. The response of enzyme polymorphisms to developmental rate selection in Drosophila melanogaster. Genetics 105:105-113.
CAVENER, D. R., AND M. T. CLEGG. 1981. Multigenic response to ehanol in Drosophila melanogaster. Evolution 35:1-10
CHARLESWORTH, B. 1980. Evolution in age-structured populations. Cambridge Univ. Press, Cambridge.
CHIPPINDALE, A. K. 1994. The evolution of trade-offs in vomplex life cycles: the Drosophila model. Ph.D. diss. Univ. of California, Irvine.
CHIPPINDALE, A. K., D. T. HOANG, P.M. SERVICE, AND M. R. ROsE. 1994. The evolution of development in Drosophila melanogaster selected for postponed senescence. Evolution 48:1880-1899.
FLEMING, J. E., I. REVElLLAUD, AND A. NIEDZWIECKI. 1992. Role of oxidative stress in Drosophila aging. Mut. Res. 275:267-279.
FLEMING, J. E., G. S. SPICER, R. C. GARRISON, AND M. R. ROSE 1993. Two-dimensional protein electrophoretic analysis of postponed aging in Drosophila melanogaster. Genetica 91:183-198.
GOULSON, D. 1993. Allozyme variation in the butterfly, Maniola jurtina (Lepidoptera: Satyrinae) (L.): evidence for selection. Heredity 71:386-393.
GRAF, J., AND F. AYALA. 1986. Genetic variation for superoxide dismutase level in Drosophila melanogaster. Biochem. Genet. 24:153-168.
GRAVES, J. L. AND M. R. ROSE. 1990. Flight duration in Drosophila melanogaster selected for postponed senescence. Pp. 59-65 in D. E. Harrison, ed. Genetic effects on aging II. Telford Press, Inc., Caldwell, NJ.
Gu, H. 1991. Electrophoretic variation at flight related enzyme loci and its possible association with flight capacity in Epiphyas postvittana. Biochem. Genet. 29:345-354.
HALDANE, J. B. S. 1941. New paths in genetics. Allen and Unwin, London.
HAMILTON, W. D. 1966. The moulding of senescence by natural selection. J. Theor. Biol. 12:12-45.
HARMAN, D. 1956. Aging: a theory based on free radical and radiation chemistry. J. Gerontoi. 11:298-300.
HARTL, D. L., AND A. F. CLARK. 1989. Principles of population genetics. Sinauer, Sunderland, MA.
HASTINGS, I. M. 1992. The population genetics of alleles affecting enzyme activity. J. Theor. Biol. 157:305-316.
HEINSTRA, P. W. H., W. SCHARLOO, AND G. E. W. THORIG. 1987. Physiological significance of the alcohol dehydrogenase polymorphism in larvae of Drosophila. Genetics 117:75-84.
HICKEY, D. A. 1977. Selection for amylase allozymes in Drosophila melanogaster. Evolution 31:800-804.
HUDSON R. R., K. BAILEY, D. SKARECKY, J. KWIATOWSKI, AND F. J. AYALA. 1994. Evidence for positive selection in the superoxide dismutase (Sod) region of Drosophila melanogaster. Genetics 136:1329-1340.
HUTCHINSON, E. W., AND M. R. ROSE. 1990. Quantitative genetics of postponed senescence in Drosophila melanogaster. I. Analysis of outbred populations. Genetics 127:719-727.
IVES, P. T. 1970. Further genetic studies of the South Amherst population of Drosophila melanogaster. Evolution 24:507-518.
IZQUIERDO, J. I., AND J. RUBIO. 1989. Allozyme polymorphism at the [Alpha]Gpdh and Adh loci and fitness in Drosophila melanogaster. Heredity 63:343-352.
KNIBB, W. R., J. G. OAKESHOTT, AND S. R. WILSON. 1987. Chromosome inversion polymorphisms in Drosophila melanogaster. IV. Inversion and Adh allele frequency changes under selection for different development times. Heredity 59:95-104.
LEE, Y. M., H. P. MISRA, AND F. J. AYALA. 1981. Superoxide dismutase in Drosophila melanogaster: biochemical and structural characteristics of allozyme variants. Proc. Nat. Acad. Sci. USA 78:7052-7055.
LUCKINBILL, L. S., R. ARKING, M. J. CLARE, W. C. CIROCCO, AND S. A. BUCK. 1984. Selection for delayed senescence in Drosophila melanogaster. Evolution 38:996-1003.
LUCKINBILL, L. S., J. L. GRAVES, A. H. REED, AND S. KOETSAWANG. 1988. Localizing genes that defer senescence in Drosophila melanogaster. Heredity 60:367-374.
LUCKINBILL, L. S., T. A. GRUDZIEN, S. RHINE, AND G. WEISMAN 1989. The genetic basis of adaptation to selection for longevity in Drosophila melanogaster. Evol. Ecol. 2:85-94.
MANCHENKO, G. P. 1994. Handbook of detection of enzymes on electrophoretic gels. CRC Press, Inc., Boca Raton, FL.
MCKECHNIE, S. W., AND B. W. GEER. 1988. The epistasis of Adh and Gpdh allozymes and variation in the ethanol tolerance of Drosophila melanogaster larvae Genet. Res. 52:179-184.
MEDAWAR, P. B. 1946. Old age and natural death Mod. Quart. 1: 30-56.
-----. 1952. An unsolved problem of biology. H. K. Lewis, London.
OAKESHOTT, J. G. 1979. Selection affecting enzyme pomorphisms in laboratory populations of Drosophila melanogaster. Oecologia 143:341-354.
ORR, W. C., AND R. S. SOHAL. 1994. Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263:1128-1130.
PASTEUR, N., G. PASTEUR, F. BONHOMME, J. CATALAN, AND J. BRITTON-DAVIDIAN. 1988. Practical isozyme genetics. Ellis Horwood Ltd., Chichester, UK.
PENG, T. X., A. MOYA, AND F. J. AYALA. 1991. Two modes of balancing selection in Drosophila melanogaster: overcompensation and overdominance. Genetics 128:381-391.
REVEILLAUD, I., A. NIEDZWIECKI, AND J. E. FLEMING. 1991. Expression of bovine superoxide dismutase in Drosophila melanogaster augments resistance to oxidative stress. Mol. Cell. Biol. 11:632-640.
REVEILLAUD, I., J. PHILLIPS, B. DUYF, A. HILLIKER, A. KONGPACHITH, AND J. E. FLEMING. 1994. Phenotypic rescue by a bovine transgene in CuZn-superoxide dismutase-null mutant of Drosophila melanogaster. Mol. Cell. Biol. 14:1302-1307.
ROSE, M. R. 1984. Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38:1004-1010.
-----. 1985. Life-history evolution with antagonistic pleiotropy and overlapping generations. Theor. Popul. Biol. 28:342-358.
-----. 1991. Evolutionary biology of aging. Oxford Univ. Press, Oxford.
ROSE, M. R., AND B. CHARLESWORTH. 1981. Genetics of life history in Drosophila melanogaster. II. Exploratory selection experiments. Genetics 97:187-196.
ROSE, M. R., L. N. VU, S. U. PARK, AND J. L. GRAVES JR. 1992. Selection on stress resistance increases longevity in Drosophila melanogaster. Exp. Gerontol. 27:241-250.
SERVICE, P. M., E. W. HUTCHINSON, M. D. MACKINLEY AND M. R. ROSE. 1985. Resistance to environmental stress in Drosophila melanogaster selected for postponed senescence. Physiol. Zool. 58:380-389.
SETO, N. O. L., S. HAYASHI, AND G. M. TENER. 1990. Overexpression of Cu, Zn SOD in Drosophila does not affect life-span. Proc. Nat. Acad. Sci. USA 87:4270-4274.
SMIT-MCBRIDE, Z., A. MOYA, AND F. J. AYALA. 1988. Linkage disequilibrium in natural and experimental populations of Drosophila melanogaster. Genetics 120:1043-1051.
SOHAL, R. S., A. AGARWAL, S. AGARWAL, AND W. C. ORR. 1995. Simultaneous overexpression of Cu and Zn containing superoxide dismutase and catalase retards age-related oxidative damage and increases metabolic potential in Drosophila melanogaster. J. Biol. Chem. 270:15671-15674.
TYLER, R. H., H. BRAR, M. SINGH, A. LATORRE, J. L. GRAVES, L. D. MULLER, M. R. ROSE, AND F. AYALA. 1993. The effect of superoxide dismutase alleles on aging in Drosophila. Genetica 91:141-149.
WATT, W. B. 1992. Eggs, enzymes, and evolution-natural genetic variants change insect fecundity. Proc. Nat. Acad. Sci. USA 89: 10608-10612.
-----. 1994. Allozymes in evolutionary genetics: self-imposed burden or extraordinary tool? Genetics 136:11-16.
WATT, W. B., P. A. CARTER, AND S. M. BLOWER. 1985. Adaptation at specific loci. IV. Differential mating success among glycolytic allozyme genotypes of Colias butterflies. Genetics 109:157-175.
WATTIAUX, J. M. 1968a. Parental age effects in Drosophila pseudoobscura. Am. Nat. 117:1035-1039.
-----. 1968. Cumulative parental age effects in Drosophila subobscura. Evolution 22:406-421.
WEIR, B. S. 1990. Genetic data analysis. Sinauer, Sunderland, MA.
WILKINSON, L. 1990. SYSTAT: the system for statistics. SYSTAT, Evanston, IL.
APPENDIX Tukey-HSD tests on population type. Alcohol dehydrogenase F allele frequencies Accelerated Control Parental Accelerated p = 1.000 Control p = 0.013 p = 1.000 Parental p = 0.001 p = 0.532 p = 1.000 [Alpha]-glycerol-3-phosphate dehydrogenase F allele frequencies Accelerated Control Parental Accelerated p = 1.000 Control p = 0.100 p = 1.000 Parental p = 0.035 p = 0.866 p = 1.000 Cu, Zn-Superoxide dismutase S allele frequencies Accelerated Control Parental Accelerated p = 1.000 Control p = 0.066 p = 1.000 Parental p = 0.031 p = 0.931 p = 1.000 Phosphoglucomutase S allele frequencies Accelerated Control Parental Accelerated p = 1.000 Control p = 0.001 p = 1.000 Parental p = 0.000 p = 0.051 p = 1.000