Economics and the evolution of life histories

Alan R. Rogers* Department of , William Stewart Building, 270 South 1400 East, Room 102, , Salt Lake City, UT 84112

irth and death are subjects of newborn because the 15-year-old is less look at the problem from the other di- perennial interest. They are likely to die before reproducing. Thus, rection, how does selection weed out shaped by both physiology and he argued, selection affects genes more harmful mutations that increase mortal- behavior, which are themselves strongly if they are expressed in 15-year- ity late in life? The force of selection Bshaped (presumably) by natural selec- olds than if they are expressed in new- affecting genes expressed in 50-year-old tion. Thus, it seems natural to suppose borns. From this perspective, high new- women should depend on the contribu- that evolutionary ideas would form the born mortality results from the relatively tion that such women make to future core of demography. This, however, is weak selection acting on them. generations. But if these women have not the case. Students of birth and This argument seemed to fit the facts stopped reproducing, this contribution death are largely divided: social scien- and held sway for 36 years, until Hamil- would seem to be nil. Thus, harmful tists call the subject demography, and ton (8) found its flaw in 1966. Fisher mutations acting late in life should accu- evolutionists call it life history evolution. had been right to ask what contribution mulate and death should follow soon In this issue of PNAS, Ronald Lee (1) each age class makes to future genera- after reproduction stops (11). proposes a theory that may help heal tions. But what matters is the contribu- In most species, this is exactly what this divide. tion of the age class as a whole, not the does happen. But there are exceptions. Lee’s theory is about aging. Within contribution per individual. Correcting Among humans and some whales, fe- evolutionary biology, discussion of this males may live long after reproduction subject has been dominated for decades has ceased (12–14). Williams (11) sug- by two theories: mutation accumulation If selection gested that older females have more to and antagonistic pleiotropy (2). The first gain by helping their existing children of these holds that selection is less ef- favors production of and grandchildren than by producing fective at removing harmful mutations more of their own. Thus, selection op- that act in old age, so such mutations children, how could poses mutations that would increase ei- accumulate. The second holds that some ther the mortality or the fertility of mutations are beneficial in youth but it favor an early older females. harmful later on. Such mutations accu- Although other explanations have mulate because selection is more re- end to fertility? been proposed (15–18), there is an abid- sponsive to early effects than to late ing interest in variants of Williams’s hy- ones. Kirkwood’s (3) ‘‘disposable soma’’ pothesis. In 1993 I was able to reject the theory describes a mechanism that this mistake, Hamilton found that the view that early menopause evolved to can plausibly generate antagonistic force of selection on genes affecting protect women from childbirth mortal- pleiotropy. mortality is constant during childhood. ity. I was unable to reject the ‘‘opportu- Lee’s focus is on parental care and It is as strong for newborns as it is for nity cost’’ hypothesis, which holds that other transfers of resources between in- 15-year-olds. Thus, the revised theory each new child reduces the woman’s dividuals of different ages. When a fails completely to account for the facts ability to enhance the survival and fer- woman dies at, say, age 30, the death of juvenile mortality. tility of existing children and grandchil- deprives her children of the care they There have been other suggestions (9, dren (19). Since then, Shanley and Kirk- would have received from her. Thus, a 10). Perhaps the most influential idea is wood (20) have found a reasonable fit mutation that increases the mortality of what Hamilton (8) called the principle to a model combining both of these hy- 30-year-olds has harmful effects at sev- of sibling replacement. In species with potheses, and ethnographic research has eral ages. This has the flavor of antago- intense parental care, offspring are generated clear evidence that older nistic pleiotropy, but the theory is costly. When one dies, the parental re- women have a beneficial effect on chil- broader in that it encompasses cases in sources that are freed can be used to dren and grandchildren in traditional which the early and late effects are both produce another. If this must happen, it societies (21, 22). Comparisons across harmful or both helpful. The real value is best for it to happen early, for this primate taxa show that, after regressing of the new theory, however, is in its suc- minimizes the parental resources squan- out the lengths of childhood and life- cess in wedding an economic model of dered on reproductive failures. Thus, span, birth rates are higher in humans exchange between individuals to the selection should shove vulnerabilities than in other apes (23, 24). This is all evolutionary theory of aging. into early childhood. This idea was an- consistent with the ‘‘grandmother hy- Lee’s article addresses several impor- ticipated by Fisher (7) and has been dis- pothesis,’’ which holds that the labor of tant questions. The first of these has to cussed by many later authors. Until now, older women accelerates the rate of do with the age profile of mortality dur- however, there has been no theoretical childbearing in humans. Yet there is still ing childhood. Among humans and model capable of predicting the magni- room for skepticism. We might expect many other species, death rates decline tude of the effect. Lee’s article fills this human fertility to exceed that of apes dramatically during the initial part of hole and sheds light on a second issue: simply because the human population life (4, 5). This pattern is pronounced in the early menopause of human females. has been growing and ape populations some species but apparently absent in This is a puzzle that takes us to the have been declining in recent decades. others (6). Why should this be? The first opposite end of the female lifespan. If Other efforts to test variants of Wil- attempt at an answer was made 70 years selection favors production of children, ago by Fisher (7), who observed that how could it ever favor an early end to an average 15-year-old will contribute fertility? Why do women not continue See companion article on page 9637. more to future generations than will a producing babies into old age? Or, to *E-mail: [email protected].

9114–9115 ͉ PNAS ͉ August 5, 2003 ͉ vol. 100 ͉ no. 16 www.pnas.org͞cgi͞doi͞10.1073͞pnas.1733942100 Downloaded by guest on October 1, 2021 COMMENTARY

liams’s idea have failed to yield signifi- itly assumes clonal reproduction in lutionary equilibrium, the Hamilton ef- cant results (25). Thus, this issue is still which each individual has a single par- fect falls to zero so that selection is af- open after a decade of accelerating re- ent of identical genotype. fected by the transfer effect alone. search interest. There is a long tradition in ecology of The transfer effect increases through- Lee (1) brings a new style of mathe- building clonal models to study sexual out childhood because children consume matical model to these issues. He begins organisms (26). Such models are useful more than they produce. Thus, muta- with the standard renewal equation of because they simplify problems that are tions that increase mortality are re- life history theory (10). Onto this he otherwise complex, yet often reveal es- moved more rapidly if they act late in grafts a ‘‘balance equation,’’ which stipu- sential features. They should, of course, childhood than if they act early. This lates that mean lifetime consumption be regarded cautiously, because one may account for the decline in mortality must equal mean lifetime production. sometimes has to eat one’s words when during childhood. It formalizes the sib- This latter condition carries a lot of the sexual model fails to replicate the ling replacement hypothesis of Hamilton freight: it allows an analysis of evolu- clonal one (27). (8). Net productivity is also positive in tionary equilibria in the context of pa- The model has one other limitation. postmenopausal women, so the transfer rental care and other transfers of re- It depends critically on the shape of sources between individuals. what Lee calls the ‘‘balance curve.’’ This effect remains positive after reproduc- This model is very general. It assumes is the relationship, as implied by the bal- tion has ceased. This formalizes the hy- that fertility increases with consumption, ance equation, between growth rate and pothesis of Williams (11). Thus, both that mortality declines with consump- ␥, an index of consumption. The analy- hypotheses receive support in the con- tion, and so on, but does not make spe- sis depends on this curve having the text of a single model. cific assumptions about the forms of any shape of an inverted ‘‘U,’’ but there is Although the analysis is entirely quali- of these functions. It does, however, no clear delineation of the circum- tative, the equilibrium result is in terms make one restrictive assumption: it as- stances under which this is true. This is of a quantity, the transfer effect, that sumes that production equals consump- no fatal flaw, for it is easy to show that can be estimated. Thus, Lee is able to tion within each genotypic class. This the inverted ‘‘U’’ is there in at least show an excellent agreement between means that there are no transfers of re- some cases, but it will be important to the entire age profile of human mortal- sources between individuals of different work out the conditions under which ity rates and the predicted force of se- genotypes. But in an outbred sexual this assumption holds. lection. This is the most comprehensive population, a rare mutant is equally Lee shows that the force of selection evolutionary theory of aging that we likely to have a normal or a mutant on mutations affecting mortality has two have seen to date. mother. If transfers are from mother to components, one measuring expected daughter, half of the transfers received future births (the Hamilton effect) and I am grateful for comments from Elizabeth by a rare mutant will be across geno- the other measuring expected future net Cashdan, Kristen Hawkes, Henry Harpending, typic classes. Thus, Lee’s model implic- production (the transfer effect). At evo- Jon Seger, James Wood, and Ronald Lee.

1. Lee, R. D. (2003) Proc. Natl. Acad. Sci. USA 12. Kasuya, T. & Marsh, H. (1984) in Reproduction 18. Austad, S. N. (1997) in Between Zeus and the 100, 9637–9642. in Whales, Dolphins, and Porpoises: Proceedings Salmon: The Biodemography of Longevity, eds. 2. Rose, M. R. (1991) Evolutionary Biology of Aging of the Conference, Cetacean Reproduction, Esti- Wachter, K. W. & Finch, C. E. (Natl. Acad. Press, (Oxford Univ. Press, ). mating Parameters for Stock Assessment and Washington, DC), pp. 161–174. 3. Kirkwood, T. B. & Rose, M. R. (1991) Philos. Management, La Jolla, California, November 28– 19. Rogers, A. R. (1993) Evol. Ecol. 7, 406–420. Trans. R. Soc. London B 332, 15–24. December 7, 1981, eds. Perrin, W. F., Brownell, 20. Shanley, D. P. & Kirkwood, T. B. L. (2001) 4. Hill, K. & Hurtado, A. M. (1996) Ache Life R. L. & DeMaster, D. P. (International Whal- BioEssays 23, 282–287. History: The Ecology and Demography of a Foraging ing Commission, Cambridge, U.K.), Vol. 6, pp. 21. Hawkes, K., O’Connell, J. F. & Blurton Jones, People (Aldine de Gruyter, Hawthorne, NY). 260–309. N. G. (1997) Curr. Anthropol. 38, 551–578. Other Ways of Growing Old Proc. 5. Mace, R. (2000) Anim. Behav. 59, 1–10. 13. Hrdy, S. B. (1981) in , 22. Sear, R., Mace, R. & McGregor, I. A. (2000) eds. Amoss, P. & Harrell, S. (Stanford Univ. Press, R. Soc. London Ser. B 267, 1641–1647. 6. Martinez, D. E. (1998) Exp. Gerontol. 33, Stanford, CA), pp. 59–76. 23. Alverez, H. P. (2000) Am. J. Phys. Anthropol. 113, 217–225. 14. Finch, C. E. (1990) Longevity, Senescence, and the 435–450. 7. Fisher, R. A. (1958) The Genetical Theory of Genome (Univ. of Chicago Press, Chicago). 24. Hawkes, K., O’Connell, J. F., Blurton Jones, N. G., (Dover, New York), 2nd Ed. 15. Weiss, K. M. (1981) in Other Ways of Growing Old, Alvarez, H. & Charnov, E. L. (1998) Proc. Natl. 8. Hamilton, W. D. (1966) J. Theor. Biol. 12, 12–45. eds. Amoss, P. & Harrell, S. (Stanford Univ. Press, Acad. Sci. USA 95, 1336–1339. 9. McNeill, W. H. (1976) Plagues and Peoples Stanford, CA), pp. 25–52. 25. Hill, K. & Hurtado, M. (1991) Hum. Nat. 2, ͞ (Anchor Doubleday, Garden City, NY). 16. Wood, J. W., O’Connor, K. A., Holman, D. J., 313–350. 10. Charlesworth, B. (1994) Evolution in Age- Brindle, E., Barsom, S. H. & Grimes, M. A. 26. Rogers, A. R. (1990) Ethol. Sociobiol. 11, 479–495. Structured Populations (Cambridge Univ. Press, (2003) in Homo-unsere Herkunft und Zukunft, ed. 27. Rogers, A. R. (1994) in Human Reproductive Cambridge, U.K.), 2nd Ed. Schultz, M. (Cuvillier, Gottingen, Germany), pp. Decisions: Biological and Biosocial Perspectives, ed. 11. Williams, G. C. (1957) Evolution (Lawrence, 483–490. Dunbar, R. (Macmillan, Basingstoke, U.K.), pp. Kans.) 11, 398–411. 17. Turke, P. W. (1997) Evol. Hum. Behav. 18, 3–13. 76–95.

Rogers PNAS ͉ August 5, 2003 ͉ vol. 100 ͉ no. 16 ͉ 9115 Downloaded by guest on October 1, 2021