Maternal age alters offspring lifespan, fitness, and lifespan extension under caloric restriction

Maternal age has a negative effect on offspring lifespan in a range of taxa and is hypothesized to influence the evolution of aging. However, the mechanisms of maternal age effects are unknown, and it remains unclear if maternal age alters offspring response to therapeutic interventions to aging. Here, we evaluate maternal age effects on offspring lifespan, reproduction, and the response to caloric restriction, and investigate maternal investment as a source of maternal age effects using the rotifer, Brachionus manjavacas, an aquatic invertebrate. We found that offspring lifespan and fecundity decline with increasing maternal age. Caloric restriction increases lifespan in all offspring, but the magnitude of lifespan extension is greater in the offspring from older mothers. The trade-off between reproduction and lifespan extension under low food conditions expected by life history theory is observed in young-mother offspring, but not in old-mother offspring. Age-related changes in maternal resource allocation to reproduction do not drive changes in offspring fitness or plasticity under caloric restriction in B. manjavacas. Our results suggest that the declines in reproduction in old-mother offspring negate the evolutionary fitness benefits of lifespan extension under caloric restriction.

Offspring lifespan, fecundity, and response to caloric restriction. We conducted life table experiments as previously described 78 . To avoid residual undefined parental effects on our experimental populations, we synchronized the maternal ages of the great-grand and grand-maternal generations for the experimental maternal (F0) cohort by collecting eggs from 3-5 d old females for two generations. Briefly, we harvested eggs from a batch culture by vortexing and micropipette isolation, let these hatch and then grow in ad libitum T. suecica (AL; 6 × 10 5 cells ml −1 ) for 5 days, and then collected the eggs from that culture. This was repeated twice, so that the maternal and grand-maternal ages for our experimental F0 cohort were 3-5 days old.
To obtain the F0 generation, eggs from this age-synchronized culture were harvested as above, allowed to hatch over 16 hours, and neonates were randomly deposited individually into 1 ml of 15 ppt seawater and AL T. suecica in wells of unshaken 24-well tissue culture plates (n = 187). Every 24 h, we recorded survival, reproductive status (whether carrying eggs), and the number of live offspring and unhatched, dropped eggs for each individual; the female was then transferred to a new well with fresh algae of the appropriate concentration. To obtain the F1 cohorts, at the specified maternal ages we isolated one female neonate hatched within the previous 24 h from each F0 female and placed these in wells of unshaken 24-well plates with 1 ml of the appropriate food concentration (n = 69-72 for each F0 age X F1 diet cohort). All F1 cohorts were collected from the same set of 187 mothers. Offspring were randomly distributed among food treatments. We tested for effects of non-independence of offspring lifespan using linear regression and found no correlation in lifespan between individual F0s and their offspring for any maternal age or diet cohort. Sample size was determined by power analysis to detect a 0.75 d (approx. 7%) difference in lifespan using the program G*Power 79 .
As a measure of offspring ability to mount a beneficial adaptive response, we subjected F1 individuals from 3, 5, 7, and 9 d old mothers (F1 3 , F1 5 , F1 7 , and F1 9 , respectively) to either chronic caloric restriction (CCR; 10% of AL food levels; 6 × 10 4 cells ml −1 ) or intermittent fasting (IF; feeding AL or starving every other day), two treatments known to increase lifespan in rotifers (experimental design in Fig. 1). Survival, reproductive status, and numbers of offspring and unhatched eggs were recorded every 24 hours for the caloric restriction experiments. No blinding was used.

Maternal investment.
To determine if maternal investment in reproduction changes with maternal age, we conducted a separate experiment to measure size and shape of female and male eggs from 3, 6, 9, and 11 d old mothers. Age-synchronized females were placed 2 per well in 1 ml of 6 × 10 5 cells ml −1 T. suecica in 15 ppt Instant Ocean in 24-well plates, and transferred daily to new wells with fresh T. suecica. At the specified ages, 48-72 rotifers were collected and vortexed or sheared through a 23 gauge needle to separate eggs (normally carried externally by females until hatching) from females. For egg size and shape, we fixed samples in 5% formalin (final concentration). Before imaging, formalin was removed by centrifugation and aspiration, and eggs were washed twice with Instant Ocean. At least 25 each of male and female eggs were imaged with a Zeiss AxioCam at 400X magnification on an Axioskop (Carl Zeiss, Inc., Thornwood, NY). We measured egg diameter, area, and roundness (inverse of aspect ratio between longest and shortest axes) using the image analysis software, Fiji 80 .
As a quantitative assessment of changes in nutrient allocation to offspring, in a separate experiment we measured neutral lipids in newly hatched F1 neonates from 3, 6, 9, and 11 d old mothers. These lipids are maternally distributed to offspring and used as a source of nutrition by neonates post-hatching. We anesthetized 6-h old neonates in 1.0 µM bupivicain for 10 minutes before fixation in 2.5% formalin. Neonates were stained with 0.5 μg μl −1 Nile Red in acetone for 5 minutes and washed twice with 15 ppt Instant Ocean. For each maternal age, we imaged 20 stained and 5 unstained neonates at 200X with a Zeiss LSM 710 Confocal Microscope (Carl Zeiss, Inc., Thornwood, NY) using a 514 nm laser excitation with 559-621 nm emission and a 458/514 nm main beam Figure 1. Experimental design to test the combined effects of maternal age and offspring diet on offspring lifespan and fecundity. Newly-hatched offspring (F1) were collected from ad libitum fed, age-synchronized amictic (asexual) maternal females (F0, n = 180) at maternal ages of 3, 5, 7, and 9 days (F1 3 , F1 5 , F1 7 , and F1 9 , respectively). Offspring were subjected to an ad libitum diet (AL; 6 × 10 5 cells ml −1 Tetraselmis suecica), chronic caloric restriction (CCR; 6 × 10 4 cells ml −1 T. suecica, a 90% reduction in food relative to AL); or intermittent fasting (IF; alternate day AL and starvation). All rotifers were housed individually in 1 ml 15 ppt Instant Ocean and algae in 24-well plates. Survival and reproduction of the F0 and F1 were scored daily until all rotifers had died. For each F1 maternal age X diet cohort, n = 69-72.
www.nature.com/scientificreports www.nature.com/scientificreports/ splitter, imaging the entire animal volume with 1 µM slices. Lipid volume per animal volume was quantified using Fiji 80 .
For an additional estimate of changes in maternal investment in reproduction with increasing maternal age, we measured resistance to starvation in unfed F1s from 3, 6, 9 and 11 d old mothers in a separate experiment. We isolated eggs from mothers as described above. Eggs hatched overnight in 15 ppt Instant Ocean, so that neonates were never fed, after which we placed 2 neonates per well in 1 ml of 15 ppt Instant Ocean in 24-well plates (n = 48 for each maternal age cohort). As above, sample size was determined by power analysis to detect a difference in lifespan of 0.75 d between groups 79 . We scored survival twice per 24 hours, at 8 and 16-hour intervals, until all individuals had died.
Statistical analyses. We used Prism 7.0a for graphing and statistical analyses. From lifespan data, we calculated median and maximum (age of 5% survivorship) lifespan. Kaplan Meier survivorship curves were constructed from lifespan data; data were right-censored in the event an individual was lost prior to death or due to accidental death caused by mishandling. Significance of differences between median lifespans was calculated using a Mantel-Cox log-rank test. We used ANCOVA to determine significance of differences between mortality rate (the slope, β) and onset of senescence (the intercept, α) from a Gompertz function fitted to age-specific hazard rate. We used one-way ANOVA with Tukey's test for multiple comparisons to determine significant differences between egg size, shape, and lipid content across maternal ages. We used two-way ANOVA with Tukey's test for multiple comparisons to determine significant differences in lifetime reproduction, non-viable embryos, or reproductive period between F1s due to maternal age or F1 diet. To determine the effect of interaction between maternal age and F1 diet on F1 lifespan and F1 reproduction, we used two-way ANOVA. Correlations between lifespan and reproduction were fit with a second-order polynomial (quadratic) equation; we also tested linear and third-order polymomial regressions, and found quadratic equations to be the best fit. Differences between reproduction-lifespan correlations were determined using an extra-sum-of-squares F-test.

Results
Offspring Lifespan. Both maternal age and offspring diet affected offspring lifespan (Figs 2,3). Median offspring lifespan declined significantly with increasing maternal age, but maximum lifespan did not change (Supplementary Table 1). Both CCR and IF significantly increased median and maximum lifespan in all F1s, regardless of maternal age (Fig. 3, Supplementary Table 1). Under CCR, the percent increase in median lifespan was greater for F1 5 -F1 9 than for F1 3 , and under IF it was greater for F1 7 and F1 9 than for F1 3 and F1 5 (Supplementary Table 1). There was no significant correlation between lifespans of individual mothers and their offspring for any maternal age or under any F1 diet (data not shown, R 2 < 0.07 for all linear regressions). There was a significant interaction between maternal age and offspring diet to determine F1 lifespan (4.26% of total variance, F 6,791 = 7.33, p < 0.0001). F1 diet had a greater influence on F1 lifespan (10.61% of total variance, F 2,791 = 54.78, p < 0.0001) than did F0 age (7.74% of total variance, F 3,791 = 26.66, p < 0.0001).
We estimated the onset and rate of aging by fitting a Gompertz function to the age-specific hazard rate ( Fig. 2D-F). Linear regression showed a change in both onset and rate of mortality under caloric restriction (Fig. 3), though this varied depending on maternal age ( Fig. 2; Supplementary Table 1). The onset of mortality (α) was delayed under CCR and IF for offspring of all maternal ages. For younger maternal ages (F1 3 and F1 5 ), the onset of mortality was later under IF than CCR, while the reverse was true for offspring from later maternal ages (F1 7 and F1 9 ). While the rate of aging (β) was significantly lower under both CCR and IF for F1 3 , at later maternal ages there was no significant difference between the Gompertz regression slopes under AL and CCR or IF, and differences in lifespan under caloric restriction were primarily due to decreased mortality at early ages, rather than a decline in the rate of aging.
Offspring fecundity and reproductive schedule. Increasing maternal age suppressed daily and total reproduction in the F1 (Figs 2-4, Supplementary Table 2). While F1 CCR slightly depressed daily reproduction relative to AL, the reproductive period was extended, resulting in the same lifetime reproduction (Supplementary  Table 3, Figs 3,4). Although the reproductive period was also extended under IF (Supplementary Table 3, Fig. 4), daily reproduction was approximately half that of AL, and discontinuous reproduction (days of no reproduction interrupting the reproductive period) was significantly higher (Supplementary Table 4), leading to significantly lower lifetime reproduction under IF (Fig. 4).
Total lifetime fecundity declined significantly with increasing maternal age under all F1 diets (Fig. 4). While lifetime fecundities were similar under AL and CCR for young and middle maternal ages (F1 3 -F1 7 ), for F1 9 the decline in fecundity under AL was partially rescued by CCR. Maternal age and F1 diet interacted significantly to determine lifetime F1 fecundity (2.0% of the total variance, F 6,787 = 3.39, p = 0.0026). Maternal age had a greater impact on fecundity (13.7% of the total variance, F 3,787 = 46.62, p < 0.0001) than did F1 diet (7.1% of the variance, F 2,787 = 36.04, p < 0.0001).
Non-viable embryo production was strongly influenced by both maternal age (2.24% of total variance, F 3,785 = 6.33, p = 0.0003) and F1 diet (4.55% of total variance, F 2,785 = 19.32, p < 0.0001) (Fig. 4). The total number of non-viable offspring (unhatched eggs) produced by F1s under AL increased significantly with maternal age, doubling for F1 5 and F1 9 (Supplementary Table 2). The timing of non-viable embryo production was strikingly different between the offspring of young and old mothers under AL. For F1 3 , non-viable embryos were low early in life and reached a maximum of 0.2 ind −1 d −1 late in life at age 11 d. In contrast, non-viable embryos peaked near 0.4 ind −1 d −1 for F1 5 and F1 7 at ages 9 d and 8 d, respectively. For F1 9 , non-viable embryos were produced at a relatively high rate throughout life, peaking at 0.4 ind −1 d −1 at age 6 d. Non-viable embryo production declined significantly under CCR and IF relative to under AL for F1s from older mothers (p < 0.001, except for www.nature.com/scientificreports www.nature.com/scientificreports/ F1 7 under CCR, which was not significantly lower) but was still low in early life for offspring of young mothers and high in early life for F1 9 .
Under all food conditions, increasing maternal age significantly decreased the reproductive period and increased the post-reproductive period as a percentage of total lifespan (p < 0.0007 for F1 7 and F1 9 relative to F1 3 ; Figs 5,6; Supplementary Table 3). This effect was greatest for AL and CCR diets, under which the reproductive period was significantly shorter and the post-reproductive period was significantly longer for F1 5 , F1 7 , and F1 9 than for F1 3 , both in actual days and as a percent of lifespan (Supplementary Table 3, Figs 5,6). The pre-reproductive period was not significantly changed by either maternal age or diet except for a slight increase as a percent of total lifespan (though not in actual days) for F1 9 under AL conditions. Both diet and maternal age changed reproductive continuity (Table 4). Only 2.8% of the F1 3 AL cohort had discontinuous reproduction; this increased with maternal age to 7.7% for F1 9 AL, although the difference was not significant. Diet had the greatest effect, with reproductive discontinuity ranging from 12.3-17.8% under CCR and 31.5-57.7% under IF. conditions. Significant differences from F1 3 are given by *(p < 0.05), **(p < 0.01), ***(p < 0.001), or ****(p < 0.0001), or ns = not significant. Additional significance of differences in survivorship and hazard rate is given in Supplementary www.nature.com/scientificreports www.nature.com/scientificreports/ The age of maximum reproduction was generally younger under IF than AL, and for any given diet, the age of maximum reproduction was lower with increasing maternal age (Fig. 4). The interaction between maternal age and diet was not significant (p = 0.08), though both variables significantly impacted age of maximum reproduction independently, with maternal age accounting for 9.7% of the variation (F 3,777 = 28.71, p < 0.0001) and diet for 1.4% of the variation (F 2,777 = 6.25, p = 0.002). Under AL conditions, the age of maximum reproduction was a full 1.15 days earlier for F1 9 than for F1 3 . This demonstrates earlier reproductive senescence rather than earlier development, given that there was no difference in the length of time to first reproduction among any maternal age cohorts (Fig. 5). At the peak of reproduction for F1 9 the number of offspring per individual was already higher for F1 3 ; F1 3 reproduction peaked a day later, when F1 9 reproduction was already declining.
Maternal age and F1 diet interacted to determine the age of reproductive senescence (Fig. 4), accounting for 3.8% of the total variance (F 6,778 = 6.95, p < 0.0001). Diet alone accounted for 11.1% of variance (F 2,778 = 60.69, p < 0.0001), and maternal age for 12.3% of the variance (F 3,778 = 44.71, p < 0.0001). For any given diet, the age of reproductive senescence in F1s was significantly younger with increasing maternal age, except for F1 7 under IF, in which reproductive senescence was later than for F1 5 (Fig. 4, Supplementary Table 4). Both CCR and IF significantly delayed reproductive senescence, relative to AL-fed F1s.
Maternal Investment. We measured female and male egg size and shape, female neonate lipid content, female offspring time to reproductive maturity, and female offspring starvation resistance as estimates of maternal investment in reproduction ( Supplementary Fig. 2). As maternal age increased from 3 d to 9 d, female egg area and roundness increased significantly (One-way ANOVA with Tukey's multiple comparison test, p < 0.0001 for all comparisons). At a maternal age of 11 d, female egg area and roundness significantly decreased. Male eggs showed a similar pattern of increase in size with increasing maternal age and decreased roundness at the oldest maternal age (Supplementary Fig. 2). Neonate lipid content decreased slightly with increasing maternal age, and was only significantly different between F1 3 and F1 11 (One-way ANOVA, p = 0.044). We found no significant difference in the time to first reproduction between F1 maternal age or diet cohorts (Figs 5,6). Mean lifespan of offspring under starvation conditions decreased significantly with maternal age between F1 3 and F1 9 , then remained constant for F1 11 (Supplementary Fig. 2).  www.nature.com/scientificreports www.nature.com/scientificreports/ Relative offspring fitness. CCR and IF had different impacts on the correlation between total lifetime fecundity and lifespan. For all F1 maternal age cohorts, the lifespan-reproduction relationship was significantly different between AL and IF (p < 0.0001, extra-sum-of-squares F-test for difference in best-fit values between quadratic equations; Fig. 7). The lifespan-reproduction correlation under CCR was significantly different from under AL for F1 3 and F1 7 (p < 0.01) but not for F1 5 (p = 0.06) or F1 9 (p = 0.22). The slope for lifespan versus reproduction was much lower under IF than under either AL or CCR, suggesting a greater decrease in lifetime reproduction with increasing lifespan under IF. Under food limitation, the slope of the lifespan-reproduction correlation decreased significantly in the F1 5 , F1 7 , and F1 9 cohorts relative to the F1 3 (p < 0.05; extra-sum-of-squares  www.nature.com/scientificreports www.nature.com/scientificreports/ F-test for difference in best-fit values between quadratic equations) suggesting a reduction in fecundity with increasing lifespan under CCR and IF in offspring from older mothers (Fig. 8).
We measured age-specific fitness as l x m x , in which reproduction, l, is multiplied by survivorship, m, for a given day, x. Relative age-specific fitness, defined here as the difference in l x m x between calorically restricted and AL-fed rotifers within a maternal age cohort (Fig. 9) or between older maternal age cohorts and F1 3 for a given diet (Fig. 10), declined with maternal age. Under both CCR and IF, fitness of all F1 maternal age cohorts was much lower in early life relative to AL (Fig. 9). Relative fitness was greater under CCR or IF only late in life, at ages beyond which most AL rotifers were post-reproductive and survivorship was low; this late-life fitness benefit decreased with increasing maternal age. The cumulative relative fitness, measured as net area under the curve, was negative for all maternal age comparisons for a given diet, and for all diet comparisons for a given age, except for F1 9 CCR relative to F1 9 AL, which was slightly positive (Fig. 9). The relative fitness under IF was generally lower than that under CCR throughout life. For a given diet, the fitness of older maternal age cohorts relative to F1 3 was lower throughout life and decreased with increasing maternal age (Fig. 10).

Discussion
To our knowledge, this study provides the first evidence that maternal age affects offspring response to caloric restriction, a lifespan extending intervention conserved across a range of taxa. Because maternal environment, physiology, and age are all known to influence offspring phenotype and lifespan, we hypothesized that maternal age might affect offspring adaptive response to caloric restriction. Previous studies have investigated the effect of maternal age on offspring phenotype, but prior work has not examined the combinatorial effect of maternal age, maternal investment, and offspring environment on offspring lifespan, daily and total reproduction, and reproductive schedule. Increasing maternal age changed not only offspring lifespan and the degree of lifespan extension www.nature.com/scientificreports www.nature.com/scientificreports/ under caloric restriction, but also the length of the reproductive period, fecundity, and the trade-off between lifespan extension and reproduction. The finding that maternal age impacts the magnitude of lifespan extension and the level of resource allocation trade-off has implications for the widespread use of CR and CR mimetics as anti-aging therapies in humans, and should be verified in mammalian models of aging.
Increasing maternal age decreases offspring lifespan and fecundity. Consistent with earlier studies in rotifers and other species, offspring of the oldest mothers had a significantly shorter median lifespan than the www.nature.com/scientificreports www.nature.com/scientificreports/ offspring from the youngest mothers 23,28,29,31,33,40,81,82 . Among AL-fed F1s, earlier onset of aging, rather than an increased rate of aging, appeared to be responsible for the observed decrease in lifespan in old-mother offspring.
The decline in offspring fecundity with increasing maternal age found in this and previous rotifer studies 23,76 differs from some reports for C. elegans, Daphnia, and Drosophila, in which offspring from the youngest or smallest mothers have been shown to have lower lifetime reproduction than those from older mothers 34,40,83 . One possible explanation is that differences in reproductive strategy may drive the differences in offspring outcomes with changing maternal age among varied small, short-lived invertebrate species. Brachionus manjavacas makes a relatively large investment in each offspring, producing a maximum of 25-30 eggs over its lifespan, with each embryo approximately one-third the size of its mother. In comparison, hermaphroditic C. elegans produces up to 300 offspring of only 30-50 µm in size over a shorter reproductive period, laying up to 140 eggs per day 72,74 . Drosophila lay up to 100 eggs per day with approximately 600 total offspring, and Daphnia produce nearly 100 offspring in multiple synchronized batches that are coordinated with adult molting 73,84,85 . Additionally, these Relationships are fitted with second-order polynomial (quadratic) equations, and differences between the best-fit values for AL and CCR or AL and IF were determined with an extra-sum-of-squares F-test. CCR was significantly different from AL for F1 3 and F1 7 (p < 0.01) but not for F1 5 (p = 0.06) or F1 9 (p = 0.22). The regression for IF was significantly different from that for AL for all cohorts (p ≤ 0.0001). For each F1 maternal age X diet cohort, n = 69-72.  (CCR; B), or intermittent fasting (IF; C). Relationships were fitted with second order polynomial (quadratic) equations, and differences between F1 3 and older maternal age cohorts were tested using an extra-sum-of-squares F-test. Under CCR and IF, the lifespan-reproduction correlation was significantly different for the F1 5 , F1 7 , and F1 9 cohorts relative to the F1 3 (p < 0.05). For each F1 maternal age X diet cohort, n = 69-72.
www.nature.com/scientificreports www.nature.com/scientificreports/ other invertebrates are indirect developers, producing offspring with larval stages or that undergo multiple metamorphoses before becoming reproductively mature; B. manjavacas, in contrast, has direct development, with neonates emerging from the egg as a small version of the adult form.
The early, high-investment, and direct reproductive strategy may be adaptive for B. manjavacas, although it is different from the r-selection strategy expected for a microscopic invertebrate that evolved in ephemeral habitats where it was subject to high extrinsic mortality due to predation and rapidly changing environmental conditions 86 . A larger investment in each embryo increases chances of neonate survival, but high external mortality likely decreases the selection pressure to produce high-quality offspring at late maternal ages 63 . Differences in life history strategy, even among short-lived invertebrate models evolving under similar environmental and predation selective pressures, must be considered when determining the applicability of results among species and from model organisms to humans. www.nature.com/scientificreports www.nature.com/scientificreports/ Offspring fitness is not determined by simple changes in gross maternal resource allocation. This study suggests that offspring size is not a sufficient measure to determine the quality or quantity of maternal investment in reproduction. Despite larger egg size and neonate body size 76 , we did not observe the accelerated development time or greater early-life reproductive output that has been associated with earlier onset of senescence in Daphnia old-mother offspring 40 . Lipid reserves and starvation resistance were slightly lower in old-mother offspring, suggesting decreased offspring provisioning, though likely not enough to account for the 21% reduction in lifespan and 46% decline in reproduction between the youngest mother and oldest mother offspring. The relatively synchronous time to death in maternal age cohorts under starvation conditions suggests that maternal provisioning to offspring is relatively consistent among offspring for a given maternal age cohort, but may decrease slightly with increasing maternal age. Lifespan extension under CCR and IF demonstrates that lifespan is plastic for all maternal age F1 cohorts, and is not solely determined by maternal provisioning. Taken together, these findings suggest that epigenetic or cellular mechanisms beyond simple changes in gross maternal investment play a role in decreased offspring fitness with increasing maternal age.
Maternal age alters offspring response to caloric restriction. The maternal age of the experimental cohort changes the magnitude of lifespan extension and degree of reproductive trade-off in response to caloric restriction, and may thus change interpretation of the mechanism. Lifespan extension under caloric restriction was greater for the offspring of the oldest mothers; under IF relative to AL, less trade-off between lifespan and reproduction was observed for F1 9 (no reduction in mean lifetime reproduction for a 28% increase in lifespan) than for F1 3 (35% reduced net reproduction and 21% lifespan increase). A previous study in B. manjavacas similarly showed that old-mother offspring had greater lifespan extension when their mothers were calorically restricted 76 . While the rate of aging decreased under caloric restriction in young mother offspring, only the onset of aging and not the aging rate were altered in old mother offspring. These results suggest that the offspring www.nature.com/scientificreports www.nature.com/scientificreports/ of the youngest mothers may already be closer to potential maximum lifespan, or alternatively, are less able to up-regulate caloric restriction-induced protective pathways. Given that F0 survivorship was 67% at 9 d old, we cannot rule out that a change in phenotypic composition of the population due to mortality led to the observed changes in offspring lifespan, fecundity, and caloric restriction response. However, as the tested rotifer population was isogenic and no other external environmental variables changed over the course of the experiment, the observed differences in both magnitude and mechanism of the response to caloric restriction are likely due to maternal age. Increasing maternal age leads to changes in offspring gene expression in C. elegans, which likely cause differential offspring responses to environmental conditions 83 . The maternal age of experimental cohorts is not always controlled or consistent among separate aging studies and thus maternal age effects may be a source of the observed variability and inconsistencies seen among caloric restriction experiments.

Caloric restriction increases relative fitness only in late life.
To assess the effects of maternal age and to evaluate the Disposable Soma theory and the evolutionary theory of the response to caloric restriction, many studies focus on end-point assessments such as median and maximum lifespan and on trade-offs between longevity and lifetime reproduction 29,[87][88][89][90] . In reality, evolutionary fitness is an age-specific combination of innate lifespan, resistance to external mortality, and reproduction. Relative l x m x provides an age-specific measure of relative fitness that incorporates age-specific survivorship, fecundity, latency to reproduction, and timing of reproductive senescence.
Averaged over lifetime, the shift to lower daily reproduction and extended lifespan under CCR and IF appears maladaptive relative to the reproductive strategy under AL; the integrated relative lifetime l x m x is negative for CCR and IF. CCR and IF both provide an age-specific late life benefit, however, supporting the Disposable Soma theory for the evolution of lifespan extension in response to caloric restriction. It is hypothesized that those individuals that are able to reallocate resources from reproduction to maintenance of the soma during times of famine have a selective advantage; this strategy allows the organism to make it through the period of starvation, and produce offspring later when resources become available 55,59 . In the current study, the relationship between lifespan and lifetime reproduction was positive within each population, showing that longer-lived individuals tended to reproduce more. When food was limited by CCR or IF, however, the slope of the lifespan-reproduction correlation decreased, suggesting a trade-off between extending lifespan and producing offspring when resources are limited, as is the expectation under the Disposable Soma theory 55 .
The caloric restriction-mediated late-life relative l x m x benefit greatly declines for F1s with increasing maternal age. Given that offspring of older mothers had a proportionally greater increase in lifespan, this does not indicate a decreased selective pressure for an adaptive response to caloric restriction. Rather, it suggests that the overall decreased lifespan and fecundity in old-mother offspring negates any fitness benefit of the caloric restriction response. While the early life cost of caloric restriction appears to decline with increasing maternal age, this is due primarily to the decrease in l x m x under AL conditions throughout life with increasing maternal age. The decline in relative offspring fitness with increasing maternal age supports the hypothesis that the force of natural selection decreases with increasing age 63,91,92 .
The increase in lifespan with concomitant decrease in daily reproduction is not in itself an adaptive response that increases fitness. Indeed, fitness will only be increased if reproduction is upregulated once food is restored. The ability to re-establish reproduction in late life after early life caloric restriction should be tested in the context of maternal age. Given the low rates of reproduction in old-mother offspring even under full food conditions, it is unclear that there is an evolutionary benefit to increasing lifespan in the face of limiting food resources for old-mother offspring, rather than maximizing early reproduction.

Conclusions
Because the selection gradient on both mortality and fecundity are decreasing with increasing age, changes in these parameters that affect older age classes are theorized to have less impact than the same changes at earlier ages 63,[92][93][94] . This study provides empirical support for this hypothesis, and demonstrates that maternal age affects not only offspring fitness, but also offspring response to interventions. We observed changing levels of caloric restriction-mediated lifespan extension and reproductive trade-off in different maternal age offspring. Additional work is needed to determine if maternal age has a similar impact on other lifestyle, diet, or pharmaceutical interventions, or if there are differences in maternal age effects among varied genotypes. Controlling for maternal age in experimental populations will be important for replication of experimental results, appropriate interpretation of findings, and assignment of mechanism.

Data Availablility
Upon publication, data will be included in online supplementary material.