Abstract
Why do many animals live well beyond their reproductive period? This seems counter to the theory that the fraction of life spent reproducing should be maximized in order to maximize the number of offspring produced in each generation. To resolve this paradox, hypotheses have been developed that evoke parental or grandparental care as reasons for post-reproductive life (e.g., the Mother and Grandmother Hypotheses). However, these hypotheses fail to explain the presence of post-reproductive life in organisms that do not care for their young, such as Caenorhabditis elegans. Here we show that a candidate proxy of the stress of childbirth explains a large portion of the variance in post-reproductive lifespans across many species. A remarkably simple metric, the “offspring ratio” (ratio of the size or weight of offspring to that of the mother) explained 77% of the variance of the post-reproductive lifespan in a sample drawn from widely dispersed taxa. Our results suggest that the stress of childbirth is an important and conserved determinant of post-reproductive lifespan. Thus, long post-reproductive lifespan may simply be a byproduct of the somatic health required for reproduction of large progeny, regardless of parental care.
Introduction
Although post-reproductive life is often thought of as a result of modern medicine’s extensions of lifespan, even before these developments, human females were documented to go through menopause and spend the remainder of their lives without the ability to reproduce (“post-reproductive life”) [1,2]. Women who were able to reproduce late in life (without modern reproductive assistance) also lived longer, suggesting a positive correlation between lifespan and reproductive span [3]. Previous work has hypothesized that while connected, reproductive and total lifespan could be under differential control (perhaps even trading off against one another) [4,5], but the reasons for this differential control have not been elucidated [4,5]. In addition to maternal aging effects on progeny quality, the onset of menopause has serious biological implications due to its effects on normal regulatory processes. For example, rates of cancer, cardiovascular disease, and other degenerative processes drastically increase post-menopause [6-8]. Therefore, elucidating the mechanisms that regulate onset of post-reproductive life and subsequent effects on aging has become more critical.
Several theories have been proposed to explain the existence of post-reproductive life through direct or indirect parental care. For instance, the “Mother Hypothesis” theorizes that females stop reproducing in order to concentrate their efforts and resources in raising already-birthed offspring [9,10]. Moreover, the presence of menopause and post-reproductive life can also protect existing offspring by discouraging males from mating with older mothers [11]. The “Grandmother Hypothesis” posits that post-reproductive females assist in the reproductive success of their daughters, through care of grandprogeny [9,12-14]. Mother and Grandmother Hypotheses concentrate upon direct benefits to children and grandchildren.
However, we and others have shown that C. elegans, like women, have a proportionally long post-reproductive life span (PRLS), despite the fact that they do not care for their young [15-19]. In addition, the existence of PRLS has been suggested in studies of brine shrimp (genus Artermia), Drosophila, Mabuya buettneri (the African Skink, a reptile), and other organisms [20-26]. Mother and Grandmother theories cannot account for the long post-reproductive lives of C. elegans and other organisms with non-human social structures. The limiting factor of C. elegans reproductive span is oocyte quality decline with age, as it is in mammals [16]. Moreover, oocyte quality is governed by similar gene sets in mammals and worms [16], suggesting that factors that regulate reproductive aging may be conserved evolutionary. Therefore, we wondered whether there is a also conserved determinant of post-reproductive life span (PRLS) across species.
Here we show that such a factor does exist: the ratio of offspring size to mother size correlates well with length of post-reproductive life span across many species. We hypothesize that the offspring ratio may indicate the level of somatic integrity necessary to successfully reproduce, and our tests of this model in C. elegans suggest that altering these ratios can have deleterious effects on the mother’s survival during reproduction.
Methods
Statistical Analyses
Linear regressions were conducted using native R functionality and the tools available from the “mlbench” package [27,28]. Graphical labels were constructed using the “calibrate” package [29].
Calculations
Several values were computed from the data: Post-Reproductive Lifespan (“PRLS”), Offspring to Maternal Size Ratio (“Offspring Ratio”), Reproductive Window (the proportion of life in which the species can reproduce), Maturity Proportion (the proportion of life spent maturing to reproductive age), Gestational Proportion (the proportion of life spent in gestation in a single reproductive cycle), and Weaning Proportion (the proportion of life spent weaning offspring from one reproductive cycle; only applicable to mammals).
PRLS was computed as the ratio of life after last documented reproduction to maximum documented life expectancy minus maturity time (the final factor facilitates comparisons outside of mammals) [30]. Offspring ratio was computed by taking the ratio of the offspring to maternal size or weight (depending on the data available). When weights were unavailable, the cube of the length of offspring at birth and mother was used. Litter size-adjusted ratio was computed by multiplying the average size of a litter by the offspring ratio. Reproductive Window was computed as the proportion of maximum lifespan between the age of maturity and reproductive senescence. Maturity Proportion was computed solely for mammals and birds as the ratio of maturation age (for females) to maximum lifespan (the data for non-mammals was not reliable enough). Gestational Proportion was computed similarly as the ratio of gestational time to maximum age.
The equations used to compute the values, then, were as follows:
†Offspring Ratio was calculated using weights if possible, but if not, lengths were used. If offspring length could not be found, it was approximated using the length of the female gamete (in the case of sea urchins). While among mammals it has been found that length to the fourth power (not third) relates to body weight, the use of logarithmic regression parameters renders the exact exponent irrelevant. Moreover, as a wide range of animals (not just mammals) were studied, it appeared more appropriate to use the natural geometric relationship [31].
‡This is only computed for mammals and some birds. There was not enough non-mammalian information is available to compute a similar value.
Data
A dataset was constructed in order to probe various aspects of animal aging. Data for mammals and most birds (both in captivity and in natural habitats) were primarily obtained from anAge: The Animal Aging and Longevity Database and the references contained therein [32]. Reproductive Senescence data was supplemented using previous reviews of primate and mammalian aging [33-35]. The selection of non-mammalian models was guided by previous aging studies [19,36-39]. Data for the following species were also obtained independently: Strongylocentrotus franciscanus (Red Sea Urchin) [40-45], Strongylocentroltus purpuratus (Purple Sea Urchin) [41-43,46,47], Oncorhynchus tshawytscha (Chinook Salmon) [48-51], Oncorhynchus kisutch (Coho Salmon) [49,50,52], Oncorhynchus keta (Chum Salmon) [42,49,50,53], Drosophila Melanogaster [21-24], Mabuya buettneri (African Skink, a reptile) [25,26], Gallus gallus domesticus (Leghorn-breed chicken) [54-56], Galeorhinus galeus (School Shark) [57-60], and Alligator mississippiensis (American Alligator) [61-63]. Different measures of longevity (and their values in various species) were obtained from analyses of the ISIS database of Zoo collections [64].
The following assumptions were made (also made explicit in the data tables): In general, the maximum documented age was either obtained from records or computed by summing the longest possible life history for an animal. Age of maturity and reproductive senescence were found by computing averages of given data. Where given, it was favored to take data from the same paper or source in order to maintain consistency among measurements. However, the sources did not fundamentally disagree with one another-and the results were relatively robust to changes in values. For the three species of salmon and two species of sea urchins, a post-reproductive period of one day was assumed (although the assumption was robust to increasing the PRLS and supported by observations), and age to reproductive maturity was assumed to be negligible in the case of the sea urchins because of their relatively long lives [65]. For the shark and alligator, the age of reproductive senescence was assumed to be the average lifespan of the organism, consistent with previous theoretical and empirical work [66]. In addition, age at reproductive maturity was used from either or both genders when considering all mammals-and separated by gender when considering mammals (due to data limitations). Since not explicit, for sea urchins, litter size and birth size was estimated by the number and size of (female) gametes. Since urchins are external fertilizers, though, we assume that this release is the most stressful part of “birthing” and will suffice.
Data sets
Datatable 1 (Mammals)
Datatable 2 (All Animals)
Results
To assess whether there is a conserved determinant of post-reproductive life span (PRLS) across species, we gathered information about life history features on a variety of animal taxa (96 species) both within and outside Mammalia, including sea urchins, salmon, Drosophila melanogaster, and species of reptiles and birds (Supplemental Table 1; Supplemental Table 2), for which we could obtain information on reproductive senescence, e.g., life span, age at reproductive maturity, adult size, progeny size at birth, and average litter size (see Supplemental information for all available variables). We then compared these features to the post-reproductive life span ratio (Figure 1):
Within mammals, no one factor accounted for the majority of post-reproductive lifespan, perhaps indicating that several different parameters contribute to the determination of post-reproductive lifespan (Supplemental Table 1). In addition, humans (data from the hunter-gatherer-like Hadza and IKung) were not outliers compared to other mammals.
However, a surprisingly simple metric, the ratio of offspring size to mother size, or “Offspring ratio,” correlated well with PRLS across the larger set of species. The measure was computed in two ways depending on data available:
The unadjusted offspring ratio could explain approximately 75% of the variance of post-reproductive lifespan (R2 = 0.771; p = 0.000174) (Figure IB). That is, larger offspring with respect to the mother is associated with longer post-reproductive period of the mother. (Note that while litter size has an inverse and highly correlated (R2 = 0.869) relationship with PRLS (Figure 1C), it is also correlated with and inversely related to offspring ratio (R2 = 0.781), and thus cannot be distinguished from offspring ratio (Figure ID).
In order to attempt to separate the effects of phylogeny from taxonomic adaptations, methods outlined previously were utilized [67,68]. Species-level data among mammals were averaged to the family level, recapitulating similar results and suggesting the presence of adaptation. Graphing the residuals of separate offspring and adult weight regressions against PRLS separated the available placental mammal from the single marsupial data point. As for the non-mammals, in order to capture more data points, we measured offspring ratio through two methods (outlined in Methods). However, the non-mammals are selected from a wide variety of taxa, and even when some of the closely-related species are averaged (e.g., sea urchins and salmon), the results do not change, lessening the concern about these differences emerging from phylogeny (although methods to separate phylogeny and adaptation are unavailable in this case).
Our comparison of offspring ratio to PRLS also roughly groups organisms by reproductive strategy. Clustered in the low offspring ratio/low PRLS area of the graph are red sea urchins, purple sea urchins, and salmon species-all organisms that release unfertilized gametes into the environment (Figure IB). The African Skink, Drosophila melanogaster, School Shark, and Alligator form another group, producing fertilized embryos that are released into the environment with no parental care (Figure IB). Finally, birds and mammals-animals that care for their young-appear to form another cluster at the high offspring ratio/high PRLS region of the graph (Figure IB).
When looking at the diversity of animals, it appears that those with the largest offspring ratios have the longest post-reproductive lives, and vice versa. This result is counter to the notion that larger offspring deplete resources, resulting in a shorter post-reproductive life span. Indeed, our results suggest that there is not a direct “tradeoff” between reproductive and post-reproductive life [4]. Instead, we posit that the offspring ratio may be a proxy for stress of childbirth or progeny production. There is likely a point at which an organism cannot devote adequate energy or quality maintenance to reproduction but still has the necessary strength and physical integrity to live. Although in the absence of parental care, neither positive nor negative selection acts on PRLS, these correlations suggest a reason for the particular length of PRLS [5]. In essence, PRLS in many species is simply an unselected residual of life [15,69,70], but the reason for this residual has not been tested previously. We suggest, based on the correlations found here, that more stressful or physically demanding forms of reproduction may require greater strength and integrity in the soma to successfully produce offspring, resulting in greater somatic integrity in the post-reproductive period and correspondingly longer PRLS. The physical manifestation of this threshold can be found when females are pushed to reproduce beyond a typical time: in humans, the most common cause of death of older mothers before the introduction of modern medical interventions was hemorrhaging during childbirth [71]. In the most extreme example in the other direction, sea urchins release millions of one-cell male and female gametes into their environment and reproduce nearly to the end of their ~200 year life span, essentially exhibiting no PRLS.
To test this hypothesis, we perturbed size and reproductive span parameters in C. elegans, a model system whose long post-reproductive lifespan has been previously assumed to simply be a lab artifact. [Note: to avoid circularity, we held C. elegans out of the correlation analysis in Figure 1.] Our model predicts that if either body size ratio or RS/LS ratio are altered, there would be a suboptimal effect on reproduction or lifespan. TGF-b Sma/Mab mutants are defective in their coupling of longevity and reproductive aging: their germline and oocyte quality is maintained, extending reproductive span (Figure 2A), but their somatic tissues age at the same rate as wild-type worms (Figure 2B) [15]. Thus, their somatic integrity does not match the high quality of their germlines [16], and their PRLS is compressed without adjusting their offspring ratio proportionally (in fact, their eggs are the same size as wild-type, but their bodies are smaller, thus increasing their offspring ratio) [72]. The effect of this uncoupling is fatal for the animals in late reproduction: because their reproductive span is so long, TGF-b mutants often die from matricide (internal hatching of offspring, Figure 2C) while still reproductive, truncating their reproductive lives (Figure 2D, * indicates matricide). Wild-type worms usually avoid this fate: C. elegans seems to have tied somatic integrity to the offspring ratio, thus tuning post-reproductive lifespan to maximize reproductive span. Further extension of reproductive span without increased somatic integrity results in matricide.
When would the maintenance of somatic integrity extending through late reproduction be important? Limited nutrient conditions, a common situation in the wild, pose such a situation. Dietary restriction extends the lifespan of all animals tested thus far [73]; additionally, dietary restriction delays reproduction [74] (Figure 2E). For example, in mammals, fertilized oocyte implantation can be delayed under low nutrient conditions [75], and in humans this process is modulated through FOXO and Insulin signaling [76,77], which also regulates both lifespan and reproductive span in C. elegans (Figure 2F). In order for reproduction to resume once nutrients become available, the soma must be healthy enough to enable birth, even at advanced ages. Longevity extension, and thus extended PRLS, under dietary-restricted conditions may simply be the result of the coupling of somatic and reproductive aging that is necessary to allow a plastic reproductive response to varying nutrient conditions.
Discussion
What prevents evolutionary pressure from bringing somatic aging in synchrony with reproductive aging, if for no other reason than to prevent wasting resources that could go to younger generations? For instance, salmon die shortly after spawning; although not instantaneously after reproduction, several days after a multi-year life represents an excellent matching of the two types of aging-despite at least partially separate genetic and regulatory control [69]. Moreover, in the case of salmon specifically, reproduction follows an exhausting migration and severe lack of nutrients (salmon stop feeding), further accelerating somatic decline and death after reproduction [78]. Indeed, lifespan appears to be influenced by hormonal signals (that are in turn partially environmentally influenced) from the reproductive system. For instance, castration of salmon gonads before development prolongs lifespan significantly [79]. Similar findings have been reported in C. elegans, mediated through the DAF-12/Nuclear hormone receptor and DAF-16/FOXO signaling pathways [80,81]. Moreover, in humans, the onset of menopause is associated with increased healing time and the rise of cardiovascular disease and other pathologies, and estrogen has been shown to be protective against various health risks-although the mechanisms have not been completely elucidated [82,83]. What remains, though, is to posit why some organisms live very long after they stop reproducing and could theoretically carry another brood. At least a component of this prolonged PRLS could be explained by the nature of biological anti-aging mechanisms. It has been posited that aging is not caused by environmental damage, but rather by the failure to repair that damage [66]. Thus, by that same reasoning, aging depends partly on that damage occurring. If such damage does not occur or occurs more slowly than expected, then aging will be slowed. The case of Werner’s syndrome illustrates; the failure of one such repair mechanism takes years to kill-but eventually it does [84].
On the other hand, in addition to repairing damage to the reproductive tract and even the germ line, reproduction requires “positive control”. That is, certain processes (hormonal, regulatory, or otherwise) must be allowed for reproduction to proceed. If these prerequisites are not met, reproduction halts and the organism enters reproductive senescence. All that has to occur, then, for reproduction to stop-unlike in general somatic aging-is for some prerequisite process to stop or be damaged severely. In a sense, reproduction is much more sensitive to aging because it is a high-stress endeavor and requires so many relatively independent processes at various levels to act. Moreover, these mechanisms are also subject to much greater selection by evolution as they act earlier in life and are intimately concerned with an organism’s evolutionary fitness [85]. Nonetheless, this is not to claim that either type of aging is determined. Medical treatments have intervened in both processes. As witnessed by the dramatic increase in lifespan over the 20th century and the development of in vitro fertilization and general reproductive medicine, both mechanisms of aging can be altered. However, oocyte quality ultimately limits human reproductive span, as in C. elegans [69]).
The fact that the simple offspring ratio can correlate parameters across a great number of highly unrelated taxonomic groups, from the most primitive to complex animals, suggests deeper relationships at the genetic and regulatory levels, revealing the intricate connection between reproduction and the structure and parameters of life history. Indeed, previous work has suggested that several factors affect PRLS in less developed organisms in which parental care is a less important factor and social structures differ from those of highly developed mammals [86]. Going forward, it will be essential to develop proxies for predation and parental care [87,88], when appropriate, to account for the remaining variance in PRLS. While parental care may certainly modulate the length of post-reproductive life in some animals, considering offspring size ratio as a proxy for childbirth stress, and PRLS as a byproduct of somatic health during reproduction, offers a new perspective in predicting the post-reproductive lifespan across animal taxa. This model explains the existence of post-reproductive lifespan in animals that do not display parental care, disposing of the need to invoke a “purpose” for PRLS in most species.
Author Contributions
CTM developed the concept; GM carried out all data gathering and analysis; GM and CTM wrote the paper.
Author Information & Financial Interests Declaration
Original data and sources are available in Supplemental Information. The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to ctmurphy@princeton.edu.
Abbreviations
PRLS: Post-Reproductive Lifespan, RS: Reproductive Span, LS: Life span
Acknowledgements
We thank Zemer Gitai, Maureen Barr, Daniel Rubenstein, Andrea Bodnar, and members of the Murphy lab for critical discussion of the work, and Shijing Luo and Jasmine Ashraf for assistance with Figure 2.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.
- 14.↵
- 15.↵
- 16.↵
- 17.
- 18.
- 19.↵
- 20.↵
- 21.↵
- 22.
- 23.
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.
- 35.↵
- 36.↵
- 37.
- 38.
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.
- 56.↵
- 57.↵
- 58.
- 59.
- 60.↵
- 61.↵
- 62.
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵