A mini-review of the evolutionary theories of aging

Demographic Research a free, expedited, online journal
of peer-reviewed research and commentary
in the population sciences published by the
Max Planck Institute for Demographic Research
Doberaner Strasse 114 · D-18057 Rostock · GERMANY
www.demographic-research.org
DEMOGRAPHIC RESEARCH
VOLUME 4, ARTICLE 1, PAGES 1-28
PUBLISHED 8 FEBRUARY 2001
www.demographic-research.org/Volumes/Vol4/1/
DOI: 10.4054/DemRes.2001.4.1
A mini-review of the evolutionary
theories of aging.
Is it the time to accept them?
Éric Le Bourg
Table of Contents
Evolutionary theories of aging: a brief overview Studies of trade-offs in human beings and primates Direct and indirect selections for longevity in Drosophila Are the evolutionary theories of aging valid? The theory of the accumulation of mutations at old age The theory of the antagonistic pleiotropy Can we accept the evolutionary theories of aging? Demographic Research - Volume 4, Article 1 A mini-review of the evolutionary theories of aging.
Is it the time to accept them?
Éric Le Bourg 1
Abstract
This article reviews some studies testing evolutionary theories of aging and shows thatthey are not always confirmed. Nevertheless, many gerontologists consider now thatthese theories provide a general explanation of the aging process. In such conditions,we may wonder whether time has come to provisionally accept these theories in orderto redirect the research efforts of gerontologists towards other directions, such as thesearch for new means to modulate the aging process.
Laboratoire d'Éthologie et de Psychologie Animale, E.R.S. C.N.R.S. n° 2041, Université Paul-Sabatier, 118 route de Narbonne, F-31062 Toulouse cedex 4, France.
(fax: 33 5 61 55 61 54, e-mail: [email protected]) Demographic Research - Volume 4, Article 1 1. Introduction
After twenty years or so of experiments testing the evolutionary theories of aging, theyare now considered by many gerontologists as the basis of the explanation of the agingprocess. These theories explain the ultimate causes of aging (why aging occurs?), whileother theories explain the proximate causes of aging (what mechanisms do explainaging?). These two kinds of theories operate at different levels of explanation and thustheories describing proximate causes of aging do not contradict evolutionary theories.
For some scientists, proximate causes of aging deal with the deleterious actions of freeradicals (see for instance all the articles of the Sohal’s team). Others favor theimbalance of long-term low intensity stressors and of protective and repair processes(Masoro 1996), or other mechanisms as those described by Medvedev (1990) in hisfamous article compiling more than 300 theories of aging.
The matter of the present article is to pay some attention to the evolutionary theories of aging. This article describes in a few words the theory of the accumulationof mutations at old age (Medawar 1952), the theory of antagonistic pleiotropy(Williams 1957), the disposable soma theory (Kirkwood 1999), and reviews someresults purporting to test them.
Even if these theories are very appealing, we may think that the final word about their validation process has not still be said. Yet, in a recent past, it has become clearthat there is a risk for some gerontologists to consider these theories as definitivelyvalidated, which seems premature. For instance, Keller and Genoud (1997) haveexplained the long life span of queens of ants by relying on evolutionary theories ofaging. Le Bourg (1998) considered that Keller and Genoud (1997) showed “thatevolutionary theories of aging are consistent with the high longevities of queens, but…only apply a theoretical explanation to what is observed in the wild”. Following thisarticle, a debate occurred between these authors (Keller and Genoud 1999, Le Bourgand Beugnon 1999).
The present article proposes to provisionally accept evolutionary theories of aging - even if the review will show that one may have some doubts about their full validity -to redirect a part of the research efforts towards other questions. In other words, wecould accept the evolutionary theories of aging, because they currently offer the mostplausible explanation of aging and, at the same time, not accept all claims thatexperiments support them. This behavior allows to consider that testing these theories isno longer a top priority.
Demographic Research - Volume 4, Article 1 2. Evolutionary theories of aging: a brief overview
According to these theories, aging is a by-product of natural selection. Any individualhas a probability to reproduce. It is zero at birth and reaches a peak in young adults.
Then, it decreases due to the increased probability of death linked to various external(predators, illnesses, accidents) and internal causes (aging). In such conditions,deleterious mutations expressed at young age are severely selected against, due to theirhigh negative impact on fitness. Conversely, the same mutations, if they are expressedat old age only, are rather neutral to selection, because their bearers have alreadytransmitted their genes to the next generation. Note that these mutations can affectfitness directly or not. For instance, a mutation increasing the risk for leg fracture, dueto a low fixation of calcium, may be as deleterious to fitness as one impairing thenesting of the egg in the uterus. In both cases, the animal is at risk not to reproduce,either because many precocious abortions occur or because it becomes an easy prey fora predator.
This theory of the accumulation at old age of mutations (Medawar 1952) seems to be in accordance with common sense. It may be easily understood that persons loadedwith a deleterious mutation at young age have less or no chance to reproduce; forinstance, progeria patients live for about 12 years (Turker 1996). By contrast, peopleexpressing a mutation only at older ages can reproduce before the illness occurs, as it isthe case with the Huntington’s disease. In such conditions, the autosomal dominantgenetic disease progeria stems from de novo mutations and not from the genes ofparents. As an outcome, progeria is less frequent than late diseases such as theHuntington’s disease, because the deleterious alleles are not removed from the genepool and can accumulate in successive generations. Obviously, the most importantconceptual problems with the theory of the accumulation of mutations at old age arethat it predicts that diseases are more common at old age than at young age, which is amere tautology, and that the risk of dying increases with age. However, the fact thatmortality rates can decelerate at old ages is at variance with the theory (Pletcher, Houleand Curtsinger 1998).
In 1957, Williams added that pleiotropic genes with favorable effects on fitness at young age and deleterious ones at old age could exist and explain the aging process.
Such genes could be selected due to their positive effect on fitness at young age, despitetheir negative effects at old age: these negative effects are the aging process. Forinstance, let us suppose that a gene favoring the fixation of calcium in bones does exist.
This gene could have positive effects at young age, because the risk of fracture and thusof death is decreased, and negative effects at old age, because the risk of osteoarthritisis increased. In the wild, such a gene has no actual negative effect because most ofanimals die before its negative effects can be observed. There is then a trade-off Demographic Research - Volume 4, Article 1 between an actual positive effect at young age and a potential negative one at old age:this negative effect may become effective if animals live in a zoo, which is free ofpredators and supplied with a veterinarian and enough food.
The main difference between these two theories is that, for the former, genes with negative effects at old age passively accumulate from one generation to the next, whilefor the latter these genes are actively kept in the gene pool by selection. Moreexperiments have tried to experimentally test the antagonistic pleiotropy theory than thetheory of accumulation of mutations at old age, because the former theory allows moreeasily such tests.
More recently, Kirkwood (1993, 1999) accepted these two theories and considered that it is useless to invest too much energy in the soma maintenance if the chances tolive long are low. In such conditions, it is more appropriate to favor fast reproduction.
Thus, there is a balance between maintenance and reproduction for all species. Whenliving conditions improve, and thus the chance for a longer life, it is useful to switch thebalance more towards maintenance, because reproductive life increases, and the agingrate will decrease. When living conditions worsen, it is time to invest more in fastreproduction to increase fitness, which increases the aging rate since maintenance isdeserted. This theory thus implies that it is possible to allocate energy either tomaintenance or reproduction. Furthermore, this allocation does not only occur betweensuccessive generations, i.e., it is not only an evolutionary process. Rather, this processis also considered to explain individual variability of life histories, as it will be shownin the following.
These three theories, which are more complementary than antagonist (see e.g.
Kirkwood and Rose 1991), shape modern thinking in gerontology. It is correct to stressthat they offer a convincing explanation of the aging process. The hot question is toknow whether they may be considered as validated or not. Indeed, in the past, othertheories were considered to readily explain the aging process, before to be eventuallygiven up, as for instance the Pearl’s rate of living theory (1928). As emphasized byKirkwood (1999, p. 58), “the theory of natural selection is one of the best tools we haveto understand the living world”. We may share that view (I share) and consider that it isyet needed to test the predictions of evolutionary theories of aging. Such tests are ofhelp to refine these theories and it is well known that a deeper knowledge can be gainedfrom studies trying to falsify or to confirm theories than from impassioned speechesclaiming that the right theory has been discovered. To speak clearly, the review ofexisting data will show that evolutionary theories have not been fully validated, whichdoes not mean that these theories have to be put to the trash. Indeed, for the time being,these theories are probably the best ones we have.
In the following, the article focuses on the tests of evolutionary explanations of aging. Firstly, it reviews the results on trade-offs between longevity and fecundity in Demographic Research - Volume 4, Article 1 humans and other primates, secondly, studies selecting directly or indirectly for anincreased longevity in flies and, thirdly, the very few experiments modulating extrinsicmortality rates.
3. Studies of trade-offs in human beings and primates
Evolutionary theories of aging are genetic theories: they try to explain aging by relyingon the selection of genes with positive or negative effects on aging (Williams 1957) oron the accumulation of mutations at old age (Medawar 1952). At a first sight, it couldbe considered that it is nearly impossible to test these theories in human beings andother primates, since they are not easily amenable to genetic studies. This point iscorrect.
However, since both Williams’ and Kirkwood’s theories predict that trade-offs between longevity and fitness, particularly early reproduction, exist, it is valuable tolook for their existence. The existence of trade-offs in humans or other primates wouldnot prove that the theories are valid, since it would remain impossible to know whethertrade-offs are due to genetic or environmental causes (or both). Nevertheless,discovering such trade-offs would obviously stimulate the search of their causes.
3.1 Human beings
A difficulty with the study of life histories in human beings is that modern peoplestrongly limit their progeny number, which is not the best condition to study trade-offsbetween reproduction and longevity. Obviously, not limiting the progeny number inconditions where infant mortality is very low would give very large families, which isnot the wish of most people. Consequently, studying ancient populations is of interest.
It seems that only four studies have been done: Canadian women living in Quebecduring the XVII-XVIIIth centuries (Le Bourg et al. 1993), British aristocrats living inthe 740-1875 period (Westendorp and Kirkwood 1998), Germans living in the 1720-1870 period (Lycett, Dunbar and Voland 1999), European aristocratic and Finnish ruralfamilies living in the XVIII-XIXth centuries (Korpelainen 2000). Table 1 summarizesthe results.
Le Bourg et al. (1993), using the parish registers of old Quebec have correlated various life history parameters in two populations: the French immigrant womenarriving in Quebec before 1680 (range of birth dates: 1603-1666) and the first French-Canadian women born before 1700 in Canada (range: 1620-1699). At that time,contraception was unheard of and healthy living conditions spared people from many Demographic Research - Volume 4, Article 1 Table 1:
Summary of the studies of trade-offs between fecundity and longevityin human beings.
Trade-off?
Comments
used in computations, ratherthan the actual number (case offathers), low progeny number diseases. Most women were married at least once and around 8% of immigrantsremained sterile, this proportion being lower in the first Canadians (Charbonneau et al.
1987). Only women having at least one child during their life were considered in theanalysis. The main difference between the two populations was that immigrants livedfor 7 years longer than Canadians did, probably because they were a highly selectedpopulation. However, the two groups had the same mean number of children.
No clear relationship between early fecundity and longevity was observed in immigrants, while the most longevous Canadians had a higher early fecundity.
However, this relationship was due to death at young age of some women, whichobviously ends reproduction. Considering only women reaching the age of menopausemade the relationship to disappear.
To sum up, it could be concluded that in the two groups of women, those who have their first child at a young age have also a high fecundity peak and a high number ofchildren during a long reproductive life. There is thus no trade-off between fecundity,early or late, and longevity.
Westendorp and Kirkwood (1998) have also tried to discover trade-offs using the database of British aristocracy. They reported that, among women living at least to 60years, the age at first child was positively correlated with longevity and the number ofprogeny was negatively correlated with longevity. That relationship held either forwomen living before 1700 and for those living in the 1700-1875 period, when the Demographic Research - Volume 4, Article 1 progeny number was lower than before 1700. Furthermore, the same negativecorrelation was observed between progeny number and longevity of fathers.
This study seems thus to show that there is a trade-off between fecundity, early or not, and longevity, both in men and women. However, the problems of that study arehuge.
Firstly, it is clear enough that British aristocrats strongly limited their progeny number since more than one third of women remained childless and, if we take intoaccount only mothers, the progeny number was around 3.5. This is in sharp contrastwith the study of Canadians for whom that number was 8. Therefore, British aristocratsare probably not the best sample to study the relationships between fecundity andlongevity, since their fecundity is low. It is of interest that Ligtenberg and Brand (1999)showed that, when only mothers are considered in the analysis, there is no negativecorrelation between fecundity and longevity.
Secondly, Westendorp and Kirkwood (1998) when correlating progeny number with father’s longevity used the legitimate number of children, and not the actual one.
Unless to hypothesize that British aristocrats did not use their dominant social positionto obtain intercourses with maidservants, a very strong hypothesis, their legitimatenumber of children is probably poorly connected to their actual number. Thus, thenegative correlation between progeny number and father’s longevity is surely spurious.
It probably reflects an environmental component, as emphasized by Promislow (1998).
Thirdly, contrarily to Rose (1989), Le Bourg et al. (1993) and Promislow (1998), Westendorp and Kirkwood (1998) considered that the existence of trade-offs, asdeduced from phenotypic correlations, would support “the interpretation that thedecrease in progeny number in long-lived women has its basis in evolutionarygenetics”. This is maybe a too liberal attitude, since phenotypic correlations mix geneticand environmental influences, as already noticed. It cannot be argued that Westendorpand Kirkwood (1998) made a clumsy turn of phrase. Westendorp and Kirkwood (1999)answered to a criticism by Ligtenberg and Brand (1999) that the weakness of acorrelation between spouse’s life span “strongly argues against environmental factorsplaying a major role in the trade-off (between longevity and reproductive success), andsupports the hypothesis that genetic factors are important”. Finally, a correspondencebetween the author and Dr Westendorp confirms he thinks that “the data support theidea that there is a genetic variation within the human population for genes that affectlife span, and genes for fertility and that there is a trade-off between the two”.
Fourthly, the quality of the database of British aristocrats has been severely criticized (Gavrilova and Gavrilov 1999), particularly because the base is stronglymale-biased (19,380 men and 13,667 women) and women’s birth dates are unknown inmuch cases. Gavrilova and Gavrilov (1999) concluded, “this British databaseunfortunately can not be used in the scientific analysis in its present form”.
Demographic Research - Volume 4, Article 1 Therefore, the problems with that study are so numerous that it cannot provide Lycett, Dunbar and Voland (1999) studied the relationships between fecundity and longevity in the Krummhörn population (northwest Germany) during the 1720-1870period. Contrarily to British aristocrats, only 10% of women remained childless, and themean number of children in the whole population was around 5. There was no negativecorrelation between the number of children and mother’s longevity (only womenreaching 50 years of age were used in analysis).
The authors then controlled for the duration of marriage, which was positively correlated with women’s longevity, and claimed that, for the poorest group of thepopulation only (the “landless”), there was a negative correlation between the numberof children and women’s longevity (r = – 0.139, p = 0.040, n = 223). However, nocorrelations were observed in two richer groups (“farmers” and “smallholders”).
Considering the whole population resulted in a significant negative correlation (r = –0.072, p = 0.041, n = 820). The negative correlation in the poorest group was notconnected to a higher number of children in this group, since that number was similar inthe three groups.
Furthermore, when the amount of time spent in a fecund marriage was controlled for, i.e. the time between marriage and menopause, the authors observed a similarnegative correlation in the poorest group (r = – 0.074, p = 0.005, n = 276), whilepositive correlations were observed in the two richest groups. However, since thesegroups were less numerous, the significance levels of these correlations were lower (r =0.099, p = 0.043, n = 73 and r = 0.068, p = 0.082, n = 119). Considering the wholepopulation (n = 1073) resulted in a non-significant positive correlation.
Lycett, Dunbar and Voland (1999) concluded, “at least for the poorest social group, there is a trade-off between reproduction and longevity”. It could be opposed tothis rationale that the evidence for a trade-off between longevity and fecundity is notclear, since it is observed only in a given group, while positive or non-significantcorrelations are seen in other groups. More fundamentally, all correlations are weak andthus only explain a tiny part of the variance. It seems then that this study is not atvariance with that of Le Bourg et al. (1993): in both studies no clear trade-off betweenreproduction and longevity does exist.
Finally, Korpelainen (2000) studied Finnish rural families and European aristocrats living in the XVIII and XIXth centuries. Women living longer than 80 years had a lowerprogeny number than those living for 50-79 years (respectively, 4.34 and 5.40 children).
The number of offspring surviving to the age of 18 years was however not different(3.40 vs 3.88). No effect of longevity on progeny number was observed among fathers.
The author mixed in the analysis people experiencing very different living conditions, which can bias the results. As aristocrats had a lower number of offspring Demographic Research - Volume 4, Article 1 and lived longer than rural families (table 1 in Korpelainen 2000), more Finns were inthe short-lived group and more aristocrats in the long-lived one. A correspondencebetween the author and Dr Korpelainen indicated that the progeny number of aristocratwomen (±SEM, n) was 4.35 (± 0.32, n = 100) for those dying in the 50-79 age rangeand 3.97 (±0.56, n = 29) for those living at least for 80 years, a non-significantdifference.
Concerning Finn women, these numbers were respectively 5.92 (±0.21, n = 203) and 4.62 (±0.53, n = 39), a significant difference, but the number of offspring reachingadulthood did not depend on longevity (respectively for Finns living 50-79 years andmore than 80 years: 3.99±0.17 and 3.51±0.38). Since mothers living more than 80 yearshad a higher proportion of surviving offspring (78.3%, table 2 in Korpelainen 2000)than those living for 50-79 years (71.9%), this could explain why there is a contrastbetween the effect of longevity on progeny number and on surviving progeny. Long-lived mothers have less children than shorter-lived ones, which is as trade-off, but theyare more able to rear successfully their children, which is the contrary of a trade-off. Inother words, there is no trade-off between fecundity and longevity because, with alower total number of children, long-lived mothers had the same number of childrenreaching adulthood.
Considering the previous studies it seems that there is no clear trade-off between earlyand late fecundity in women, and no trade-off between fecundity, early or late, andlongevity. These results do not refute the antagonistic pleiotropy theory, because theobserved correlations are phenotypic, since they mix genetic and environmentalinfluences (see, for a discussion of genetic and phenotypic correlations, Cheverud 1988,Stearns 1992, Roff 1995, 1996, Koots and Gibson 1996). However, they show thatthere is no ground to aver that trade-offs between fecundity and longevity do exist inhuman populations, as it could be expected from the antagonistic pleiotropy theory. Itcould be that antagonist pleiotropic genes dealing with fecundity and longevity do exist,but they do not seem to have a deep impact on life histories, since their influence doesnot outmatch the effect of other genes with no such pleiotropic effects or ofenvironmental influences.
These four studies on human beings do not provide any clear evidence for trade- offs. It remains to know whether results gathered from other primates point in the samedirection.
Demographic Research - Volume 4, Article 1 3.2 Primates
Studies focusing on reproduction and longevity in primates are scarce, to say the veryleast, and Table 2 summarizes them.
Table 2:
Summary of the studies of trade-offs between fecundity and longevity inprimates. Trade-off?
Comments
between fecundity andlongevity, in optimal andbad living conditions(population decline) In 1993, Bercovitch and Berard showed, in female rhesus macaques (Macaca mulatta) observed during 30 years, that macaques reproducing when 3 years-old (“rapidreproducers”) did not live shorter than those reproducing when 5 years-old (“delayedreproducers”, respectively: 9.0 ± 4.5 years, n = 14, vs 11 ± 4.5 years, n = 11). Thisresult speaks against the existence of a trade-off between early fecundity and longevity.
Considering only these two contrasted groups, longevity was independent of age at firstparturition (r = 0.077, n = 25), but was strongly correlated with the number of offspringsurviving to the age of sexual maturity (r = 0.812, p < 0.001, n = 21), or with the totalnumber of offspring (r = 0.891, p < 0.001, n = 25). When age of death was keptconstant, rapid reproducers gave birth to more offspring reaching age at maturity thandelayed reproducers.
In summary, there is no trend for any trade-off between fecundity and longevity in female rhesus macaques, in accordance with results in humans. However, the samplesize is rather low and considering the whole population, and not only rapid and delayedreproducers, would allow refining the picture. Furthermore, conclusions drawn from asingle study need to be confirmed.
In 2000, Rhine, Norton and Wasser reported the results of a study of reproductive longevity and lifetime reproductive success in baboons (Papio cynocephalus). Lifetimereproductive success was defined as the number of offspring living at least to the age ofsexual maturity. Reproductive longevity (highly correlated to longevity: r > 0.98) wasthe time interval between sexual maturity and death. The troop of baboons experienced Demographic Research - Volume 4, Article 1 a severe population decline during the 24 years study, and it was possible to comparefemales born before the start of the study (n = 27) and those reaching adulthood in theyears before the population decline (n = 45). The first sub-sample was less affected bythe population decline and its reproductive longevity and lifetime reproductive successwere higher than those of the second sub-sample (respectively, 13.58 vs 7.38 years and3.13 vs 1.18 offspring). However, the correlation between lifetime reproductive successand reproductive longevity was positive in the two sub-samples (r = 0.70, n = 27, p <0.0001; r = 0.73, n = 45, p < 0.0001). Even if this study did not differentiate betweenearly and late fecundities, it clearly confirms a part of the conclusions of Bercovitch andBerard (1993) on macaques: there is no trade-off between fecundity and longevity.
3.3 Conclusions
This review of data correlating fecundity, particularly early fecundity, and longevitydoes not provide firm evidence in favor of the existence of trade-offs at the inter-individual level in human and non-human primates. However such trade-offs betweenlife-history parameters do exist at the inter-specific (see, e.g., Stearns 1983) and, insome cases, at the intra-specific level (Stearns 1992).
On the one hand, this absence of trade-offs does not disprove the Williams’ antagonistic pleiotropy theory of aging, but only shows that one of its predictions is notfulfilled. In such conditions, it is premature to wonder what part of the variation oflongevity is due to pleiotropic genes. It has to be said that Williams (1957) stated that“most of the genes or gene combinations that favor vigor early in life probably alsofavor longevity” and that “only a small proportion of the genes need be of the sort thatproduce opposite effects on fitness at different ages”. In such conditions, it could beargued that it would be always impossible to discover any trade-off between earlyfecundity and longevity at the inter-individual level. The only means to confirm theantagonistic pleiotropy theory of aging in humans could be to discover a human geneticdisease due to an allele conferring the illness at old age and a selective advantage atyoung age (Albin 1993).
On the other hand, the absence of trade-offs disproves the Kirkwood’s disposable soma theory of aging, because the theory does not only apply to the evolutionary level,but also to the inter-individual one.
In conclusion, the current evidence does not confirm evolutionary theories of aging in primates, as far as inter-individual trade-offs are concerned. Other studies of trade-offs between early fecundity and longevity in Drosophila melanogaster (e.g. Le Bourget al., 1988) or in medflies (Carey et al. 1998) gave similar results.
Demographic Research - Volume 4, Article 1 However, selection for longevity in a species well amenable to genetic studies, such as D. melanogaster, could provide different conclusions about the validity ofevolutionary theories of aging.
4. Direct and indirect selections for longevity in Drosophila
melanogaster
Williams (1957) predicted that “successful selection for increased longevity shouldresult in decreased vigor in youth”, such as for instance a decreased early fecundity.
This is a testable prediction, which explains that many gerontologists have tried todiscover such a trade-off in D. melanogaster. Obviously, the goal of these studies wasnot only to study the possible genetic trade-off between longevity and early fecundity,but also, if not mainly, to create lines with contrasted longevities. Two selectionprocedures have been used: direct selection for life span and indirect selection viareproduction at old age. Studies using the first procedure are less numerous than thoseusing the second one are. Tables 3 and 4, respectively, summarize the results.
4.1 Direct selection for longevity
Lints et al. (1979) tried to select directly for increased longevity. When 20% of the pairsof the line selected for increased longevity were dead, the authors kept the virginprogeny of surviving pairs. When only 25% of these pairs were still alive, their progenywas crossed to give the next generation and the progeny of the other pairs wasdiscarded. Therefore, only the progeny of long-lived flies produced the next generationand this process was done for 8 generations.
There was no longevity increase in the selected line when compared to two control lines and the realized heritability was only 0.034 (Baret, Beckers and Lints 1995).
However, longevity increased in the three lines during selection.
Zwaan, Bijlsma and Hoekstra (1995) also selected directly for longevity. They started selection for increased longevity in two replicate lines and for decreasedlongevity in two other lines, two control lines being used. In all lines, a part of thevirgin progeny was kept at 29°C to measure longevity and the other part was stored at15°C to produce the next generation. When all 29°C flies were dead, brothers andsisters of the most longevous ones were mated to produce the next generation of thelong-lived lines. A similar procedure was used to give the next short-lived generation,and flies of the control lines were randomly paired. The whole process was done for 6generations.
Demographic Research - Volume 4, Article 1 Table 3:
Summary of the studies selecting directly for longevity in Drosophilamelanogaster. Longevity
Trade-off?
Comments
decreased at all not really increase (0 day Table 4:
Summary of the studies selecting indirectly for longevity in Drosophilamelanogaster, by reproducing flies at old age. Longevity
Trade-off?
Comments
Demographic Research - Volume 4, Article 1 Zwaan, Bijlsma and Hoekstra (1995) concluded that longevity increased in the long-lived lines, and that the selection was less successful in the short-lived ones.
Consequently, they reported high realized heritabilities (up to 0.517).
These conclusions were based on the comparison of selected and control lines.
However, if the actual mean longevity values are considered, longevity decreasedduring selection in both short-lived and control lines (by 40% in one of the controllines), while a slight longevity increase was observed in long-lived males (ca 3 dayswhen compared to the parental generation) and no increase at all in females. Theauthors also measured longevity at 25°C in virgin flies. Long-lived lines derived fromthe fourth generation of selection lived longer than control ones, but the pattern was lessconsistent in flies coming from crosses between the two replicate lines. However, theauthors did not indicate whether the longevity has decreased or not when compared tothe parental generation. Finally, the progeny was lower in long-lived lines at all ages,and not only at young age, while viability of eggs, development time, body weight andstarvation resistance were similar in control and long-lived lines.
On the one hand, Lints et al. (1979) failed to increase longevity in their selected line, but longevity increased in both selected and control lines during selection. Zwaan,Bijlsma and Hoekstra (1995) increased longevity in the long-lived lines, whencompared to the control lines, but this pattern was only due to the decreased longevityof these latter lines during selection. It is difficult to accept that selection was successfulwhen mean longevity does not increase.
The whole evidence of the two experiments selecting directly for longevity is inconclusive and we may hope that studies using indirect selection for longevity willclarify the issue.
4.2 Indirect selection for longevity
The rationale of the indirect selection for longevity procedure is that reproduction at oldage in successive generations could increase longevity, provided this trait is partlyheritable, since only long-lived flies may reproduce.
If the antagonistic pleiotropy theory is correct, we may expect to observe a decreased early fecundity in lines reproduced at old age and, maybe, an increased latefecundity. Conversely, if flies are repeatedly reproduced at young age, their longevity isexpected to decrease, as well as their late fecundity, while their early fecundityincreases.
If the theory of the accumulation of mutations at old age is correct, a trade-off between early and either late fecundity or longevity is not mandatory.
Demographic Research - Volume 4, Article 1 Numerous studies have tried to confirm these expectations. The most famous experiment is that of Luckinbill and Clare (1985) after the early work of Luckinbill etal. (1984). They measured longevity in successive generations and, after 21 generationsof reproduction, longevity had regularly increased in lines reproduced at old age(hereafter called the OLD lines) and, to a lesser extent, in lines reproduced at young age(YOUNG lines). Furthermore, early fecundity decreased in OLD lines (Clare andLuckinbill 1985).
This work confirmed the results of Rose and Charlesworth (1981) and Rose (1984). However, these last authors measured longevity and fecundity only once, in thelast generation of selection, and erratic differences between OLD and YOUNG linescould explain their differences. As a matter of fact, Lints and Hoste (1974, 1977), whohave reproduced YOUNG and OLD lines for ten generations, observed large longevityand fecundity variations between generations. However, there was no increasedlongevity in OLD lines when compared to YOUNG ones.
After Luckinbill and Clare (1985), Partridge and Fowler (1992) and Engström, Liljedahl and Björklund (1992) also reported that OLD lines live longer than YOUNGor control ones, but longevity was measured only once during the selection process.
However, they failed to show a decreased early fecundity in OLD lines.
By contrast, a decreased early fecundity was observed in the OLD lines of Partridge, Prowse and Pignatelli (1999), that lived longer than YOUNG lines (longevitymeasured only once during the selection process). However, since these authorsmeasured longevity in mated females only, a cost of reproduction could explain the lowlongevity of YOUNG lines.
When all these results are considered, it seems clear that reproduction at old age does increase longevity. However, Baret and Lints (1993) reanalysed the Clare andLuckinbill’s results (1985). They noticed that, since YOUNG and OLD lines arereproduced at different ages, the same generation number occurs at a different calendartime in the two lines. If longevity differs along months, as in Lints et al. (1989), it couldbias the results. As a matter of fact, Baret and Lints (1993) showed that, when the Clareand Luckinbill’s results (1985) are expressed as a function of calendar time, and not ofgeneration number, the difference between YOUNG and OLD lines is erased. In otherwords, when YOUNG and OLD flies living at the same moment are compared, there isno longevity difference between them. This article provoked considerable attention (seethe debate in Gerontology: Fukui, Pletcher and Curtsinger 1995; Arking and Buck1995; Baret, Le Bourg and Lints 1996). Arking and Buck (1995) published new resultsusing the lines of Luckinbill et al. (1984) showing clear differences between OLD andcontrol lines reproduced at random age. However, these new results also showed thatthese differences appeared very late, not before the 30th month of selection, whichconfirms that no increased longevity in OLD lines was observed in the article of Demographic Research - Volume 4, Article 1 Luckinbill and Clare (1985). Reproduction of OLD lines has finally increased theirlongevity, but it is hard to reconcile the late and sudden longevity increase with theeffects due to a selection of quantitative genes. However, Buck et al. (2000) reported anew selection experiment with a different issue. They selected for increased longevityin lines also selected for the speed of development (flies emerged during days 1-4 ofeclosion or during days 6-10 of eclosion) and measured longevity regularly duringselection. The OLD lines showed a regularly increasing longevity in the two cases, witha clear difference from control lines 20 months after the start of selection in the fast-developing line and after 12 months in the slow-developing line. The problem with thatstudy is that the authors did not specify the sex and mating status of flies used inlongevity measurements.
While it can be now safely accepted (but see below) that reproduction at old age increases longevity, it remains that a decrease in early fecundity was not alwaysobserved, which casts some doubt on the antagonistic pleiotropy theory. Furthermore, ithas been shown that the trade-off between early fecundity and longevity could be lost(Leroi, Chippindale and Rose 1994). An absence of trade-off between longevity andother traits indirectly connected to fitness has also been reported. For instance, Serviceet al. (1985) reported that OLD lines have throughout life a higher resistance tostarvation and desiccation than YOUNG lines. Arking et al. (2000) reported that OLDlines have at young and old ages a higher superoxide dismutase activity level thancontrol lines, no difference being observed at intermediate ages. Indeed, the resultsseem more compatible with the accumulation of mutations at old age, since it could beargued that reproduction at old age has purged the gene pool from deleterious alleles,which could explain the increased longevity as well as the increased resistance to somestresses.
However, the debate about the increased longevity of OLD lines could be soon revitalized. Buck et al. (2000) have produced new OLD lines and showed that they haveup to twice the developmental lethality as have control lines. Using 8 OLD and controllines, they reported that longevity increased when developmental viability decreased (r= – 0.79). This could thus explain a part of the increased longevity of OLD lines sinceonly the fittest, and possibly the most longevous, flies reach adulthood in OLD lines,while most of flies do so in the control lines. In such conditions, a part of the increasedlongevity could be a mere statistical artifact, due to a demographic selection process,since the longevity of (short-lived?) flies dying during development cannot be observed.
It is of interest to note that Partridge and Fowler (1992) showed that larvae of OLDlines were less able to compete with larvae of a mutant than were larvae of YOUNGlines. For instance, if one third of larvae in a vial were YOUNG and two thirds weremutant, one third of emerging imagoes were YOUNG. This proportion was only 20% inOLD lines. When larval crowding increased, the differences were more important. For Demographic Research - Volume 4, Article 1 the same proportion of YOUNG larvae, the proportion of YOUNG imagoes was 45%,while it was only 15% in OLD lines. Roper, Pignatelli and Partridge (1993) confirmedthese results, but Partridge, Prowse and Pignatelli (1999) did not. Curtsinger (pers.
comm.) has indicated that the viability is currently 85% in the OLD Luckinbill’s linesand 88% in the YOUNG ones, with respective longevities of 70-80 and 35-40 days. Thesmall viability difference does not appear to explain the huge longevity difference.
4.3 Conclusions
This review of studies using reproduction at old age to increase longevity in D.
melanogaster and of those attempting to directly increase longevity was not intended tocover all results. Particularly, other studies using different species have been done (e.g.
Tucic et al. 1996 in the bean weevil Acanthoscelides obtectus). The goal of this reviewwas simply to show whether longevity has been increased, and how this result is inaccordance with evolutionary theories of aging.
Direct selection for longevity has failed to increase longevity. Zwaan, Bijlsma and Hoekstra (1995) increased longevity in their long-lived lines, when compared to thecontrol ones, but this pattern was only due to the decreased longevity of these latterlines. Indeed, the mean longevity of long-lived lines did not increase.
Reproduction at old age was more successful and it can be accepted that it increased longevity. However, at least in what concerns the Luckinbill’s lines for whichwe have regular longevity measurements during the experiment, it is difficult to explainwhy no increase occurred before 30 months of selection. The Buck’s et al. (2000) linessuffer less from this problem, but a high developmental lethality could explain a part ofthe longevity increase in OLD lines.
The study of early fecundity has shown that trade-offs with longevity have been observed only in some cases. Furthermore, other traits, such as stress resistances, werenot involved in trade-offs. The whole evidence does not totally confirm the antagonisticpleiotropy theory but shows that, at least in some cases, pleiotropic mechanisms couldbe at play.
To sum up, even if one could consider that these experiments have failed to unequivocally confirm the Williams’ theory (1957), it remains that creating long-livedlines of flies has provided the community of gerontologists with a useful research tool.
However, as emphasized by Harshman and Hoffmann (2000), inconsistent correlatedresponses to selection can be observed.
Another way of testing evolutionary theories of aging is to experimentally impose low or high extrinsic mortality rates to modify longevity.
Demographic Research - Volume 4, Article 1 5. Extrinsic mortality rates and longevity
When extrinsic mortality rates increase, there is less time to reproduce and anevolutionary response is expected to preserve the species from eventual extinction. As amatter of fact, a higher extrinsic mortality rate means a higher selection for a fastreproduction and, if evolutionary theories of aging are correct, we expect an earlyfecundity increase at the expense of late fecundity and lifespan. This is the rationale ofthe very few studies imposing higher extrinsic mortality rates as a means to testevolutionary theories of aging (see for a discussion Reznick 1997). These studies do notselect directly for longevity, nor they impose an age at reproduction. Simply, they usecontrasted extrinsic mortality rates.
Stearns, Ackermann and Doebeli (1998) defined two extrinsic mortality rates in D. melanogaster. The authors rear flies in population cages since 1993 and kill 90% ofthem twice a week (probability to survive a week: 0.01, HAM condition), or 20%(probability to survive a week: 0.64, LAM condition). In 1996, the probability tosurvive under the LAM condition was risen to 0.81 (10% of flies killed twice a week).
In order not to mix the effects of mortality rates and those of density, killed flies arereplaced by flies of the same condition (HAM or LAM) and age. Flies are thus keptunder contrasted mortality rates and their life history parameters are observed regularly.
After three years of experiment (75 generations in the high mortality conditions and 40in the low ones) the mean longevity was about 30 days in HAM flies, and 34 days in theLAM condition (mean longevity computed from fig.4 in (Stearns, Ackermann andDoebeli 1998). Furthermore, HAM flies eclosed earlier and had a higher early fecunditythan LAM ones. However, no difference was observed regarding late fecundity (Stearnset al. 2000, Gasser et al. 2000). After five years of experiment (Stearns et al. 2000), itwas observed that the early fecundity differences had reached a plateau 30 months afterthe beginning of experiment, and the same was true for development time. Oddlyenough, early fecundity decreased regularly under both HAM and LAM conditions(from 70 eggs at days 13-15 to 20), from the very beginning of experiment, whiledevelopment time variations were more erratic. The longevity had also stronglyincreased under both conditions between 1996 and 1998, since the median longevitywas above 50 days in 1998. However, the HAM flies lived shorter than LAM flies(mean longevities unknown, but median longevity estimated from fig.3 in Stearns et al.
2000), as it was observed earlier. Stearns et al. (2000) did not provide explanations forthe increased longevity between 1996 and 1998, nor they explained why early fecunditydecreased throughout experiment. In a correspondence with the author, Dr Stearns saidthat he prefers not to draw “any quantitative conclusions from the trends over time” Demographic Research - Volume 4, Article 1 because he is “much more certain of the differences between the two treatments than…of the changes within a treatment from, say, 1995 to 1998”.
Other experiments modifying extrinsic mortality rates have been done, for instance in guppies. Early fitness varies according to mortality rates, in accordance withevolutionary theories of aging, but experiments recording longevity are still in progress(Reznick 1997).
The study on flies seems to indicate that, even if some questions remain unanswered, contrasting extrinsic mortality rates makes longevity and early fecundity todiffer, in accordance with evolutionary theories of aging. Obviously, more results areneeded before to conclude unequivocally that these theories are confirmed. However,there is a good chance that they will be eventually confirmed, since their expectationsconcerning the effect of extrinsic mortality rates are very similar to those of life historystudies (Stearns 1992).
6. Are the evolutionary theories of aging valid?
This review shows that, on the one hand, there is no clear trade-off at the individuallevel between fecundity and longevity in human beings, but it could be rightly arguedthat the absence of phenotypic correlations does not disprove the antagonisticpleiotropy theory (Williams 1957). This is correct, but it has been shown above thatWestendorp and Kirkwood (1998) consider inter-individual trade-offs as arguments infavor of the Kirkwood’s disposable soma theory.
On the other hand, indirect selection for longevity in D. melanogaster was eventually successful, but it remains difficult to explain the kinetics of the selectionresponse. A progressive longevity increase was expected, but no increase was observedfor a long time in the Luckinbill’s lines. It remains that lines differing by their longevityand originating from the same genetic background are available. However, directselection for longevity appears to fail.
Finally, experiments that modify extrinsic mortality rates seem a promising tool to test evolutionary theories of aging.
Some conclusions about the validity of the evolutionary theories of aging may now 6.1 The theory of the accumulation of mutations at old age
The theory of the accumulation of mutations at old age is in accordance with commonsense, since it states that mutations occurring at old age, i.e. after the reproductive age, Demographic Research - Volume 4, Article 1 are likely to be kept in the gene pool. However, Pletcher and Curtsinger (1998) are rightwhen stressing that “a theory involving deleterious mutation pressure alone as the causeof senescence is not consistent with mortality plateaus far below 100%”.
If this theory is considered as the main or unique cause of aging, it may be ruled out. By contrast, if one considers that it describes one cause of aging among others, thetheory may be valuable. Indeed, some features of aging could be due to such alleleswith a late expression while other ones could be best explained by differentmechanisms. If the mutations at old age explain only a part of the aging process,“plateaus far below 100%” could be explained.
6.2 The theory of the antagonistic pleiotropy
The theory of the antagonistic pleiotropy has been submitted to many tests. Maybe themost important result is that lines with contrasted longevities have been created whenreproducing flies at old age.
However, the crucial idea of the theory is that “successful selection for increased longevity should result in decreased vigor in youth” (Williams 1957). This correlatedresponse to selection has been observed in some cases, but not in each selectionexperiment. It is however not difficult to imagine that some alleles with antagonisticpleiotropic effects on early fitness and longevity could be at play in some strains, butnot in others. If so, a decreased early fecundity, as a correlated response to selection onlongevity, can be observed or not. In the same way, positive genetic correlationsbetween early fecundity and longevity are not unexpected, too. For instance Clark andGuadalupe (1995) reported such a genetic positive correlation (correlations betweenline means: r = 0.286, p = 0.0398) in 52 lines of flies containing a P-element inserted atrandom into a common genetic background. Therefore, it may be that genetic variationfor longevity is associated with a decreased early fecundity in some strains, anincreased early fecundity in others, or not associated. It remains that observing adecreased early fecundity shows that the antagonistic pleiotropy mechanism is at play insome cases.
Thus, there is no debate about the validity of the antagonistic pleiotropy hypothesis: the hypothesis is valid. The debate concerns the importance of thatmechanism for the aging process, because it can be shown only in some cases that agenetic trade-off exists between early fecundity and longevity (see Table 4). Thus, itmay be concluded, as it was done for the theory of the accumulation of mutations at oldage, that the antagonistic pleiotropy theory is correct but, probably, does not describethe main cause of aging.
Demographic Research - Volume 4, Article 1 6.3 The disposable soma theory
Concerning the disposable soma theory, we have to take into account the level ofexplanation of the theory, because it can be applied either at the evolutionary level or atthe individual level.
When applied at the evolutionary level, the theory is not really different from the antagonistic pleiotropy theory. As emphasized by Kirkwood and Rose (1991), thedisposable soma theory is a causal subset of the antagonistic pleiotropy theory, “if thegenes responsible for somatic maintenance functions are regarded as … (to) prolongsurvival and … consume resources which might be used for reproduction”. In fact, “thedifference between the disposable soma theory and the antagonistic pleiotropy is partlya difference between an optimality approach… and a quantitative genetics approach”,but the two theories do not differ in their predictions regarding trade-offs betweenreproduction and longevity.
The disposable soma theory is also explicitly applied at the individual level, because trade-offs can also operate at the individual level. For instance, Kirkwood andRose (1991) consider the case when “reproductive activity reduces survival byincreased risk exposure”; in that case, “rescheduling fecundity to later ages willincrease survivorship”. This kind of non-metabolic trade-off operates at the individuallevel. However, we may be skeptical about some “direct metabolic trade-offs … thatcan happen if reproduction and maintenance draw directly from the same supply ofresources within the organisms”. If such trade-offs were so important at the individuallevel, we would observe a clear trade-off between the number of children and longevityin women, particularly at periods when only breastfeeding, a costly metabolic activity,was possible. It has been shown (Table 1) that it is not the case.
Finally, it has to be recalled that it is not possible to use phenotypic correlations to directly conclude in favor of a genetic hypothesis, as Westendorp and Kirkwood (1998)did (see 3.1). Therefore, it may be concluded that the disposable soma theory isprobably not correct when applied at the individual level.
7. Can we accept the evolutionary theories of aging?
In conclusion, the whole evidence shows that the evolutionary theories of aging explaina part of the results but not all of them. In such conditions, we may wonder whethertime has come to accept them provisionally, even if one may have doubts about some oftheir expectations, because they explain a part of the aging process. In 1989, Rose andGraves wrote, “if other biologists could accept the sufficiency of the evolutionary Demographic Research - Volume 4, Article 1 theory as the general theory of aging, then there might be a relaxation of efforts to findgeneral physiological theories of aging”. Nowadays, evolutionary theories have not anyobvious challenger. In some cases, these theories well fit to the results and, in othercases, they fail; but many gerontologists consider that they offer, more or less, aplausible explanation of the aging process by integrating natural selection as a maincause of aging. In such conditions, accepting the general frame of these theories wouldrelax efforts to test all their expectations.
Evolutionary theories of aging have faced the opposition of Sacher (1978). He considered that “the implication that… organisms are mortal only because of theaccumulation of adventitious senescence genes, is more easily reconciled with acosmology of special creation than with current scientific conceptions”. Lints (1983,1985) strongly criticized, not the evolutionary theories of aging, but rather the firstexperiments claiming to confirm them, and particularly that of Rose and Charlesworth(1980) expanded in Rose and Charlesworth (1981). The Le Bourg’s et al. (1988) articleprovoked a hard debate with Rose (1989). Later on, Le Bourg et al. (1993) criticizedRose’s (1991) logical flaws about theories testing. Since that time, new experimentshave been done to test the theories and many gerontologists seem to consider thatevolutionary theories are the current best candidates to explain the occurrence of theaging process. Particularly, connecting the evolutionary theories with the life historytheory was probably a step toward this beginning of a consensus since “trade-offs havea central role in life history theory” (Stearns 1992). In fact, longevity and earlyfecundity are life history parameters among other ones, such as number of offspring,time to maturity, and so on. Thus, they have to be considered just like these otherparameters. In this way, it could be said that evolutionary theories of aging are just asubset of the life history theory.
During the last twenty years, some gerontologists have tried to confirm or invalidate evolutionary theories of aging. We may wonder whether it is useful to spendthe next twenty years to test them again, since it is clear that they explain, at leastpartly, the aging process. These theories probably still need to be refined. New tests ofthem are surely useful. However, testing them is probably no longer a top priority. Thispriority could be to find new means to modulate the aging process in animal models(worms, flies, rodents or primates), using genetic or environmental manipulations, tofinally improve everyday life of elderly. This is of importance because the number ofelderly is growing rapidly.
That does not mean that we must definitively accept the evolutionary theories of aging, while we may think they are not so good, but simply that testing them could beonly a secondary goal or our research. Debates about the validity of evolutionarytheories will continue for the best of gerontology, but should not constitute the maininterest of the community. Current evolutionary theories have probably not told the Demographic Research - Volume 4, Article 1 final word about the aging process and articles favoring them, even when alternativehypotheses are available, may be criticized (Le Bourg 1998). Therefore, it is possible toprovisionally accept these theories while not being a dogmatic supporter of them.
It is indeed possible that new studies will challenge current evolutionary theories up to the point that they will be given up. It is also possible that, eventually, these newstudies will confirm these theories. As scientists, we have to be ready to accept anyoutcome. However, even if testing evolutionary theories is no longer a top priority,previous attempts were not wasted time and effort, because gerontologists have reachedthe beginning of a consensus. To be perfectly clear, these theories give a plausibleexplanation of the aging process, but in many occasions they do not fit to the results:since better theories are not available, we could accept them provisionally andconcentrate our efforts in other directions.
8. Acknowledgements
Many thanks are due to Philippe Baret, Nadège Minois, Vassily Novoseltsev andAnatoli Yashin for their helpful comments on a previous draft. This work wassupported by a grant (Santé-Société 98N72/0048) from the French Centre National de laRecherche Scientifique (CNRS).
Demographic Research - Volume 4, Article 1 References
Albin TL (1993). Antagonistic pleiotropy, mutation accumulation, and human genetic disease. Genetica 91: 279-286.
Arking R, Buck S (1995). Selection for increased longevity in Drosophila melanogaster: a reply to Lints. Gerontology 41: 69-76.
Arking R, Burde V, Graves K, Hari R, Feldman E, Zeevi A, Soliman S, Saraiya A, Buck S, Vettraino J, Sathrasala K (2000). Identical longevity phenotypes arecharacterized by different patterns of gene expression and oxidative damage.
Exp. Geront. 35: 353-373.
Baret PV, Beckers F, Lints FA (1995). Realized heritability of longevity in Drosophila melanogaster. Gerontology 41: 82-85.
Baret PV, Le Bourg E, Lints FA (1996). Selection for increased longevity in Drosophila melanogaster: reflections on new data. Gerontology, 42: 14-17.
Baret P, Lints FA (1993). Selection for increased longevity in Drosophila melanogaster: a new interpretation. Gerontology 39: 252-259.
Bercovitch FB, Berard J.D. (1993). Life history costs and consequences of rapid reproductive maturation in female rhesus macaques. Behav. Evol. Sociobiol. 32:103-109.
Buck S, Vettraino J, Force AG, Arking R (2000). Extended longevity in Drosophila is consistently associated with a decrease in developmental viability. J. Geront.
Biol. Sci. 55A: B292-B301.
Carey JR, Liedo P, Müller HG, Wang JL, Chiou JM (1998). Relationship of age patterns of fecundity to mortality, longevity, and lifetime reproduction in a largecohort of mediterranean fruit fly females. J. Geront. Biol. Sci. 53A: B245-B251.
Charbonneau H, Desjardins B, Guillemette A, Landry Y, Légaré J, Nault F (1987).
Naissance d’une population. Les Français établis au Canada au XVIIe siècle.
Paris: Institut National d’Études Démographiques, Presses Universitaires deFrance; Montréal: Presses de l’Université de Montréal.
Cheverud JM (1988). A comparison of genetic and phenotypic correlations. Evolution Clare MJ, Luckinbill LS (1985). The effects of gene-environment interaction on the expression of longevity. Heredity 55: 19-29.
Clark AG, Guadalupe RN (1995). Probing the evolution of senescence in Drosophila melanogaster with P-element tagging. Genetica 96: 225-234.
Demographic Research - Volume 4, Article 1 Engström G, Liljedahl LE, Björklund T (1992). Expression of genetic and environmental variation during ageing. 2. Selection for increased lifespan inDrosophila melanogaster. Theor. Appl. Genet. 85: 26-32.
Fukui HH, Pletcher SD, Curtsinger JW (1995). Selection for increased longevity in Drosophila melanogaster: a response to Baret and Lints. Gerontology 41: 65-68.
Gasser M, Kaiser M, Berrigan D, Stearns SC (2000). Life-history correlates of evolution under high and low adult mortality. Evolution 54: 1260-1272.
Gavrilova NS, Gavrilov LA (1999). Data resources for biodemographic studies on familial clustering of human longevity. Demogr. Res. (Online) 1. Available:http://www.demographic-research.org/volumes/vol1/4.
Harshman LG, Hoffmann AA (2000). Laboratory selection experiments using Drosophila: what do they really tell us? TREE 15: 32-36.
Keller L, Genoud M (1997). Extraordinary lifespans in ants: A test of evolutionary theories of aging. Nature 389: 958-960.
Keller L, Genoud M (1999). Evolutionary theories of aging. 1. The need to understand the process of natural selection. Gerontology 45: 336-338.
Kirkwood TBL (1993). The disposable soma theory of aging: evidence and implications. Neth. J. Zool. 43: 359-363.
Kirkwood TBL (1999). Time of our lives. London: Weidenfeld and Nicolson.
Kirkwood TBL, Rose MR (1991). Evolution of senescence: late survival sacrificed for reproduction. Phil. Trans. R. Soc. Lond. B 332: 15-24.
Koots KR, Gibson JP (1996). Realized sampling variances of estimates of genetic parameters and the difference between genetic and phenotypic correlations.
Genetics 143: 1409-1416.
Korpelainen H (2000). Fitness, reproduction and longevity among European aristocratic and rural Finnish families in the 1700s and 1800s. Proc. R. Soc. Lond. B 267:1765-1770.
Le Bourg E (1998). Evolutionary theories of aging: Handle with care. Gerontology 44: Le Bourg E, Beugnon G (1999). Evolutionary theories of aging : 2. The need not to close the debate. Gerontology 45: 339-342.
Le Bourg E, Lints FA, Delincé J, Lints CV (1988). Reproductive fitness and longevity in Drosophila melanogaster. Exp. Geront. 23: 491-500.
Le Bourg E, Thon B, Légaré J, Desjardins B, Charbonneau H (1993). Reproductive life of French-Canadians in the 17-18th centuries: a search for a trade-off betweenearly fecundity and longevity. Exp. Geront. 28: 217-232.
Demographic Research - Volume 4, Article 1 Leroi AM, Chippindale AK, Rose MR (1994). Long-term laboratory evolution of a genetic life-history trade-off in Drosophila melanogaster. 1. The role ofgenotype-by-environment interaction. Evolution 48: 1244-1257.
Ligtenberg T, Brand H (1999). Longevity – Does family size matter? Nature 399: 522.
Lints FA (1983). Genetic influences on lifespan in Drosophila and related species. Rev. Lints FA (1985). Drosophila. Interdiscipl. Topics Geront. 21: 91-119.
Lints FA, Hoste C (1974). The Lansing effect revisited – I. Life-span. Exp. Geront. 9: Lints FA, Hoste C (1977). The Lansing effect revisited – II. Cumulative and spontaneously reversible parental age effects on fecundity in Drosophilamelanogaster. Evolution 31: 387-404.
Lints FA, Lints CV, Bullens P, Bourgois M, Delincé (1989). Unexplained variations in life span of the Oregon-R strain of Drosophila melanogaster over a four-yearperiod. Exp. Geront. 24: 265-271.
Lints FA, Stoll J, Gruwez G, Lints CV (1979). An attempt to select for increased longevity in Drosophila melanogaster. Gerontology 25: 192-204.
Luckinbill LS, Arking R., Clare MJ, Cirocco WC, Buck SA (1984). Selection for delayed senescence in Drosophila melanogaster. Evolution 38: 996-1003.
Luckinbill LS, Clare MJ (1985). Selection for life span in Drosophila melanogaster.
Lycett JE, Dunbar RIM, Voland E (1999). Longevity and the costs of reproduction in a historical human population. Proc. R. Soc. Lond. 267: 31-35.
Masoro EJ (1996). The biological mechanism of aging: Is it still an enigma? Age 19: Medawar PB (1952). An unsolved problem in biology. London: HK Lewis.
Medvedev ZA (1990). An attempt at a rationale classification of theories of aging.
Partridge L, Fowler K (1992) Direct and correlated responses to selection on age at reproduction in Drosophila melanogaster. Evolution 46: 76-91.
Partridge L, Prowse N, Pignatelli P (1999). Another set of responses and correlated responses to selection on age at reproduction in Drosophila melanogaster. Proc.
R. Soc. Lond. B 266: 255-261.
Pearl R (1928). The rate of living. London: Knopf.
Demographic Research - Volume 4, Article 1 Pletcher SD, Curtsinger JW (1998). Mortality plateaus and the evolution of senescence: why are old-age mortality rates so low? Evolution 52: 454-464.
Pletcher SD, Houle D, Curtsinger JW (1998). Age-specific properties of spontaneous mutations affecting mortality in Drosophila melanogaster. Genetics 148: 287-303.
Promislow DEL (1998). Longevity and the barren aristocrat. Nature 396: 719-720.
Reznick DN (1997). Life history evolution in guppies (Poecilia reticulata): guppies as a model for studying the evolutionary biology of aging. Exp. Geront. 32: 245-258.
Rhine RJ, Norton GW, Wasser SK (2000). Lifetime reproductive success, longevity, and reproductive life history of female yellow baboons (Papio cynocephalus) ofMikumi National Park, Tanzania. Amer. J. Primatol. 51: 229-241.
Roff DA (1995). The estimation of genetic correlations from phenotypic correlations: a test of Cheverud’s conjecture. Heredity 74: 481-490.
Roff DA (1996). The evolution of genetic correlations: an analysis of patterns.
Roper C, Pignatelli P, Partridge L (1993). Evolutionary effects of selection on age at reproduction in larval and adult Drosophila melanogaster. Evolution 47: 445-455.
Rose MR (1984). Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38: 1004-1010.
Rose MR (1989). To the editor. Exp. Geront. 24: I.
Rose MR (1991). Evolutionary biology of aging. Oxford, New-York, Tokyo: Oxford Rose MR, Charlesworth B (1980). A test of evolutionary theories of senescence. Nature Rose MR, Charlesworth B (1981). Genetics of life history in Drosophila melanogaster.
II. Exploratory selection experiments. Genetics 97: 187-196.
Rose MR, Graves JL (1989). What evolutionary biology can do for gerontology. J. Geront. Biol. Sci. 44: B27-B29.
Sacher G (1978). Evolution of longevity. In: Schneider EL, editor. The genetics of aging. Plenum publ. Corp. pp.151-168.
Service PM, Hutchinson EW, Mackinley MD, Rose MR (1985). Resistance to environmental stress in Drosophila melanogaster selected for postponedsenescence. Physiol. Zool. 58: 380-389.
Demographic Research - Volume 4, Article 1 Stearns SC (1983). The influence of size and phylogeny on patterns of covariation among life-history traits in the mammals. OIKOS 41: 173-187.
Stearns SC (1992). The evolution of life histories. Oxford, New-York, Tokyo: Oxford Stearns SC, Ackermann M, Doebeli M (1998). The experimental evolution of aging in fruitflies. Exp. Geront. 33: 785-792.
Stearns SC, Ackermann M, Doebeli M, Kaiser M (2000). Experimental evolution of aging, growth, and reproduction in fruitflies. Proc. Natl. Acad. Sci. USA 97:3309-3311.
Tucic N, Gliksman I, Seslija D, Milanovic D, Mikuljanac S, Stojkovic O (1996).
Laboratory evolution of longevity in the bean weevil (Acanthoscelides obtectus).
J. Evol. Biol. 9: 485-503.
Turker M. (1996). Premature aging. In: Birren JE, editor. Encyclopedia of gerontology.
Age, Ageing, and the aged. Vol. 2: 341-354. Academic Press.
Westendorp RGJ, Kirkwood TBL (1998). Human longevity at the cost of reproductive success. Nature 396: 743-746.
Westendorp RGJ, Kirkwood TBL (1999). Longevity – Does family size matter? Nature Williams GC (1957). Pleiotropy, natural selection and the evolution of senescence.
Zwaan B, Bijlsma R, Hoekstra RF (1995). Direct selection for life span in Drosophila melanogaster. Evolution 49: 649-659.

Source: http://www.demographic-research.org/volumes/vol4/1/4-1.pdf