Report

A Delayed Wave of Death from Reproduction in Drosophila

See allHide authors and affiliations

Science  24 Dec 1999:
Vol. 286, Issue 5449, pp. 2521-2524
DOI: 10.1126/science.286.5449.2521

Abstract

Mortality rates typically increase rapidly at the onset of aging but can decelerate at later ages. Reproduction increases the death rate in many organisms. To test the idea that a delayed impact of earlier reproduction contributes to both an increase in death rates and a later deceleration in mortality, the timing of the surplus mortality produced by an increased level of egg production was measured in femaleDrosophila. Reproduction produced a delayed wave of mortality, coincident with the sharp increase in death rates at the onset of aging and the subsequent deceleration of mortality. These results suggest that aging has evolved primarily because of the damaging effects of reproduction earlier in life, rather than because of mutations that have detrimental effects only at late ages.

A cost of reproduction is prevalent in animals and plants: Greater reproductive activity causes a decrease in life-span and in subsequent fecundity (1). Much less well understood is the time course of the impact of reproductive costs. Mortality rates in a population typically increase rapidly as aging occurs but have been seen to drop below this initial rate of increase at late, mainly postreproductive, ages in experimental animals and in humans (2). A delayed impact of earlier reproduction could produce both the high initial rate of increase in death rates and the deceleration of mortality rates. If the increase in death rates from reproduction is delayed and temporary, earlier reproduction would cause a subsequent wave of mortality, with an initial increase in mortality rates, followed by deceleration after the peak of the wave has passed.

One theory about the evolution of aging suggests that genes with beneficial effects early in life also have deleterious effects later on (3, 4). A delayed impact of early reproduction on later mortality rates would provide a general mechanism for this type of time-delayed pleiotropic gene action. Experimental manipulations of reproductive rate can have both immediate (5) and delayed (6) effects on mortality, but the timing of the effects of natural genetic variation for early fecundity has not been examined. We used artificially selected lines of the fruit flyDrosophila melanogaster and sterility induced by x-ray irradiation or by a female-sterile mutant to test the hypotheses that (i) early reproduction causes a delayed wave of elevated mortality, and (ii) delayed effects of natural genetic variation for early fertility underlie the evolution of aging.

We used sets of five replicate artificially selected lines derived from a wild-type base stock of D. melanogaster. “Old” lines were selected by breeding adults at an old age, whereas “young” lines were propagated from young breeders. To contribute to future generations in the old lines, flies had to survive to the time of egg collection, and there was hence selection for increased adult survival. Adult life-span evolved to be greater in the old lines than in the young lines, and their fecundity in early life evolved to be lower, demonstrating a cost of reproduction. There were no significant differences in preadult development time, larval competitive ability, body size, or late adult fecundity between the lines from the two selection regimes (4). The difference in early fecundity between the young and old lines was apparent by day 7 and ceased by day 28 (4, 7). Egg production had ended by day 40. In both experiments reported here, we first abolished the difference in reproductive rate between the young and the old lines to determine whether it was responsible for the difference in their death rates. We could then measure the timing of the impact of the differing levels of early reproduction in the young and old lines by comparing their death rates directly with each other and also with those of sterile flies. Egg production, exposure to males, and mating have all been shown to reduce survival in female D. melanogaster (8). We examined the impact of egg production on mortality trajectories in females, in either once-mated females or in virgins, in the absence of males.

In a first experiment, using a single cohort of flies, we compared the death rates of once-mated egg-laying females of the old and young lines with each other and with the death rates of females that were x-ray–irradiated as late pupae, which halts oogenesis (7,9). In the x-irradiated females, the significant difference in mortality rates between the young and old lines was abolished (Fig. 1A) (10). The measured difference in late-age mortality rates between reproductive young- and the old-line females was therefore explained by the difference between them in early fecundity. The egg-laying young-line females had significantly greater mortality rates after day 30 than did the old-line females, and the difference in death rates reached a peak at day 40 and thereafter tended to decline (Figs. 1B and2A) (10); whereas in the irradiated females, differences in mortality rates were nonsignificant (Fig. 2B). The higher early egg production rate of the young-line females, which was over by day 28 (4,7), therefore acted with a time delay to produce a wave of excess mortality later in their lives. X-rays would be expected to reduce mortality rates if their only effect was to induce sterility. However, x-irradiation significantly increased mortality rates in the old lines and had no consistent effect in the young lines (11). This finding suggested that x-rays also caused somatic damage, and we confirmed this by x-irradiation of females whose oogenesis was already blocked by the dominant mutantovoD1 (12). The full reduction in mortality caused by sterility was therefore not apparent in the x-irradiated females.

Figure 1

Mean of the natural logarithms of the age-specific mortality rates for the five replicate old and young selection line females when (A) irradiated as late pupae and (B) mated once and producing eggs. Solid circles and solid line, old selection lines; open circles and dashed line, young selection lines.

Figure 2

Mean difference [and 95% confidence limits (CLs), shown by error bars] in the natural log of the mortality rates of egg-laying young- and old-line females (A) when mated once and producing eggs and (B) when irradiated as late pupae. Positive values indicate that young-line females had higher mortality rates. CLs were calculated from the five replicate differences between the matched pairs of old and young lines. The comparison was ended when the first young line went extinct.

In a second experiment, we abolished egg production by a different method, which would be expected to have no direct effect on the adult soma, using the dominantovoD1 mutant. ovo encodes a zinc finger protein that is active only in the female germ line, and the mutant halts oogenesis at stage 4 (13). We hybridized males carrying the mutant, in the genetic background of the base stock from which the selection lines were derived, with each of the young and old selection lines and measured the mortality rates of the resulting F1 hybrid sterile females in single-sex groups. To produce genetically comparable egg-laying females, we hybridized each selection line to the base stock itself and examined the mortality rates of the resulting fertile hybrid virgin females in the same experiment as the sterile hybrid females (14). These two sets of females were therefore both genetically hybrids between the Dahomey base stock and each selection line, except for the presence of heterozygousovoD1 in the sterile females.

In the sterile females with ovoD1 present, the significant difference in mortality rates between old and young lines was again abolished (15) (Figs. 3A and 4A), demonstrating that it was caused entirely by the difference in early egg production rate. ovoD1 extended life span, and this extension was greater for the young- than for the old-line females, as would be predicted by the higher early fecundity of the former (16). In the egg-laying females, a wave of significantly increased mortality in the young-line females peaked at day 52, with nonsignificant differences in age-specific death rates until day 32 and after day 58 (Figs. 3B and 4B). Direct comparison of fertile and sterile females showed that the low early egg production rates of the fertile old-line hybrid females were not sufficient to elevate their subsequent death rates above those of the comparable sterile old-line hybrid females (Fig. 4C). In the young-line fertile hybrid females there was again evidence for a wave of increased mortality rates after the difference in reproductive rate had ceased (Fig. 4D). The low egg production of the old-line females at early ages may therefore have been insufficient to elevate their mortality to an extent that could be detected with these sample sizes.

Figure 3

Mean natural logarithm of the age-specific mortality rates for (A) sterile females that were hybrids between each selection line and the Dahomey base stock males carrying ovoD1 and (B) fertile females that were hybrids between each selection line and the Dahomey base. Solid circles and solid line, old selection lines; open circles and dashed line, young selection lines.

Figure 4

Mean difference (and 95% CLs) in the natural logarithm of the age-specific mortality rates for (A) old and young sterile hybrid females, (B) old and young fertile hybrid females, (C) old-line fertile and sterile hybrid females, and (D) young-line fertile and sterile hybrid females. CLs were calculated from the five replicate differences between the matched pairs of old and young lines or between fertile and sterile hybrids.

The mortality rate deceleration, seen at late ages in other studies, was clearest in the present study in the egg-laying females illustrated in Fig. 3B, which lived the longest of the egg-laying groups. Deceleration was greatest in the young-line females and commenced at day 54, when death rates started to drop. This coincided with the declining impact of earlier reproduction, which had peaked at day 52 (Fig. 4, B and D). Females of these selection lines become postreproductive by day 40. The wave of mortality consequent upon earlier egg production thus contributed to the mortality rate deceleration at postreproductive ages in these females.

The mechanisms by which reproduction acts with a time delay to increase mortality rates require further investigation. Reproduction may cause damage directly, and the effects may accumulate with time. Reproduction may also divert nutrients from repair and defense, resulting in more rapid accumulation of damage (17). The mortality rate deceleration at later ages could be explained as a consequence of the subsequent repair of damage from earlier ages, and of a diversion of resources to repair as reproduction wanes. Heterogeneity in frailty between the individuals in a population may also contribute to a declining impact of the cost of reproduction with time, if the survivors of the initial impact of reproduction tend to be a more robust subset of their cohort (18).

The abolition of the difference in late-age mortality rates by sterility, whether induced by x-irradiation orovoD1 , has important implications for the evolution of aging. The young and old lines were the products of different naturally occurring genetic variants for the rate of aging because they were produced by artificial selection from a wild-type base stock. First, our data show that the lower death rate of the old lines is not a consequence of constitutive up-regulation of repair and defense processes (17), because this would be expected to persist when the difference in early egg production was abolished. Second, the results support the idea that aging evolves as a result of genes that have both beneficial effects at early ages and detrimental effects at late ages (3, 4). Furthermore, the results imply that the two effects are not the result of the effects of expression of these genes at different ages, but rather a direct impact of the early beneficial process that they produce (high egg production) on mortality at later ages. Finally, the results indicate that mutation accumulation causing late-age mortality (19) was not detectable in the young lines with these sample sizes, because any effects of mutation accumulation would be expected to persist after the removal of the difference in early reproductive rate. Mutations that specifically elevate death rates at late ages may therefore be rare (20).

  • * To whom correspondence should be addressed E-mail: l.partridge{at}ucl.ac.uk

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article