Senescence in natural populations of animals: Widespread evidence and its implications for bio-gerontology

1.1. Medawar and the origins of the fallacy

The notion that animals in nature do not senesce – that environmental challenges whether they be predators, floods, famine or something else kill all wild animals before aging can take a measurable toll – can be traced at least as far back as Peter Medawar’s first full theoretical treatment of the evolution of senescence (Medawar, 1952). This idea is clearly fallacious as we will show, undercut by both subsequent theoretical work and copious empirical data from a wide range of animals, yet it was crucial to the development of Medawar’s central hypothesis about the genetic mechanism by which senescence could evolve.

Medawar’s paradigm-shifting contribution to the evolutionary understanding of aging was his insight that due to the inevitability of death from environmentally driven causes, the ability of natural selection to favour or disfavour genetically-based traits depended on the age at which those traits appeared. As he phrased it, “the force of natural selection weakens with increasing age”. This specific insight forms the basis of all subsequent analyses of the evolution of senescence. Based on this idea, Medawar proposed a particular genetic mechanism – that senescence evolves by the accumulation in the genome of harmful alleles, such as those predisposing to cancer, dementia, or heart disease, whose effects appear sufficiently late in life that “the force of natural selection will be too attenuated to oppose their establishment and spread.” In other words, such deleterious alleles could spread only when the probability that they could have a measurable effect on reproductive success was very low. Consequently observable senescence should only occur at ages “which the great majority of the population do not reach” (all quotes from Medawar, 1952). Only under conditions in which animals are protected from natural hazards, he theorized, such that they commonly survive to ages they would very seldom or never achieve in the wild, would senescence be manifest. He was quite specific about this latter point, repeating it four times in the same monograph. To cite one of these,

“Whether animals can, or cannot, reveal an innate deterioration is almost literally a domestic problem; the fact is that under the exactions of natural life they do not do so. They simply do not live that long.”

– Medawar, 1952 (italics in original)

As evidence for this claim, Medawar cites a personal communication from field mammalogist Dennis Chitty, in which Chitty states that “wild mammals of any perceptible degree of senility” seldom turned up in his traps, and if they did he assumed that some cause other than senescence (e.g. infection, injury) was responsible for their condition (Medawar, 1952). For wild birds, Medawar’s source was David Lack, the pioneering field ornithologist. Lack had noted in several papers in the 1940’s, that the probability of death seemed to be independent of age in birds in nature, implying that senescence, at least in its actuarial sense, did not occur (Lack, 1943a, b). Medawar’s claim was repeated in the first comprehensive consideration of senescence as a general biological phenomenon - Alex Comfort’s classic tome, The Biology of Senescence (Comfort, 1956). Since then, the idea that senescence fails to occur in nature has been often repeated within bio-gerontology (e.g. Hayflick, 2000; Kirkwood and Austad, 2000; Rose, 1991).

In Medawar and Comfort’s time there was in fact scant evidence for senescence in the wild because few detailed long-term studies of natural populations existed. Those that were available tended to be short-term, cross-sectional and often did not know the exact age of adults in the population. It has since become apparent that long-term studies that monitor marked individuals from birth to death are required to reliably detect and investigate senescence in the wild. In an earlier monograph on aging and death published in 1946, Medawar seems to acknowledge this, highlighting the importance of studies of exactly this kind:

“No-one has yet made a systematic study of whether even mammals in their natural habitat do indeed live long enough to reach a moderate though certifiable degree of senility… The difficulties of constructing life tables in the wild are technically formidable, but they must be solved.”

– Medawar, 1946 (italics in original)

As we will demonstrate, despite Medawar and Comfort’s subsequent assertions to the contrary, there was never any theoretical reason to adduce the absence of senescence in the wild (Williams, 1957), and there has been an avalanche of recent data from wild populations clearly demonstrating the senescence does indeed occur commonly in natural populations (Bennett and Owens, 2002; Brunet-Rossinni and Austad, 2006; Nussey et al., 2008a).

1.2. Unravelling the fallacy

Quite probably, Medawar’s belief that animals in nature failed to senesce allowed him to dismiss one clear implication of his insight. While he focused nearly exclusively on the accumulation of late-acting deleterious alleles, he mentioned in passing – even giving an example – that because of the gradual weakening strength of natural selection with age, “a relative small advantage conferred early [in life]…may outweigh a catastrophic disadvantage withheld until later” (Medawar, 1952). This, of course, is a succinct description of an idea George Williams later developed: antagonistic pleiotropy, in which alleles with beneficial effects on survival or reproduction early in life can be actively favoured by natural selection despite negative effects on health and fitness later in life (Williams, 1957). So although Medawar described antagonistic pleiotropy, he failed to appreciate its significance. In summarizing the implications of his theory at the end of his 1952 paper, he never mentioned it, focusing instead on the accumulation of late-acting alleles and on hypothetical modifier alleles that might postpone the effects of deleterious alleles to ages at which they would be effectively neutral (Medawar, 1952). However, if antagonistic pleiotropy is a common mechanism of senescence, then there is no necessary expectation that wild animals will fail to display progressive deterioration of health in later life or that such deterioration need be subtle. Williams’ clearly appreciated this, referring to Comfort’s argument that senescence was ‘outside the developmental program that concerns natural selection’, as follows:

“I believe that this theory is incorrect. Its fallacy lies in the confusion of the process of senescence with the state of senility, and in an inaccurate conception of the relationship of age to selection processes.”

– Williams, 1957

By way of example, Williams noted that an examination of athletic records or life tables reveals ‘rampant’ senescence in humans as early as their 30’s, a period which no-one could disagree humans commonly reached even in a state of nature (Williams, 1957). Williams highlighted two crucial points that alter the way we think about senescence in natural populations. First, in an evolutionary sense, senescence is the progressive physiological process of deterioration leading to a decline in fitness with age, which is not synonymous with infirmity and frailty associated with extreme old age in humans and captive animals. Classical evolutionary theory does not refer to or consider a state of senility in very late adulthood, rather it predicts that senescence should begin at the age of sexual maturity and progress from that point as the force of natural selection weakens (Hamilton, 1966; Williams, 1957). Second, under antagonistic pleiotropy, natural selection is expected to favour the evolution of life histories in which senescence has a detectable fitness cost within the natural life expectancies of organisms, as long as the genes associated with this deterioration confer a sufficient fitness benefit in earlier life.

Kirkwood’s (1977) disposable soma theory of senescence is also compatible with the manifestation of senescence in the wild. Developed from a different theoretical framework – optimization theory as opposed to population genetics theory – the disposable soma theory makes largely similar predictions to antagonistic pleiotropy. Briefly, it hypothesizes that because critical resources such as energy are limited, natural selection will adjust the allocation of cellular and physiological resources between the fundamental processes of somatic maintenance and reproduction appropriately for an organism’s ecological context (Kirkwood, 1977; Kirkwood and Rose, 1991). Since natural selection is expected to weaken with age, selection will tend to favour investment in early reproduction over long-term somatic maintenance, and senescence will result (Kirkwood and Rose, 1991). Antagonistic pleiotropy and disposable soma theories, often referred to together as ‘life history’ theories of aging (Partridge and Barton, 1996), now form the basis of the majority of theoretical work on the evolution of aging. This emerging body of evolutionary theory now clearly demonstrates that selection can favour senescence occurring within a species’ natural life expectancy, and that the pattern of senescence can vary depending on the specifics of organism’s life history, ecology and the interplay between “intrinsic” and “extrinsic” factors influencing mortality (e.g. Abrams and Ludwig, 1995; Baudisch, 2008; Cichon, 2001; Mangel, 2008; McNamara et al., 2009; Williams and Day, 2003).

Current empirical support from model laboratory organisms for disposable soma theory and antagonistic pleiotropy as mechanisms of senescence is dramatically stronger than for mutation accumulation. For instance, many single gene mutations known to extend life in model laboratory organisms have detrimental effects on early components of Darwinian fitness (Table 1). To pick one illuminating example, a partial loss of function mutation in daf-2, the C. elegans ortholog of the vertebrate insulin/IGF receptor, doubles the longevity of worms in the laboratory, yet its depressive effects on early reproduction (Chen et al., 2007) cause it to be quickly eliminated when directly competed again worms carrying a wild-type allele (Jenkins et al., 2004). If antagonistic pleiotropy and disposable soma are the main mechanisms responsible for the evolution of senescence and the maintenance of genetic variation in aging and lifespan, we should expect to observe senescence in the wild.

3.1. Sex differences in senescence

In humans, sex differences in lifespan are well documented, with evidence suggesting women have lower mortality rates throughout life but a similar increase in mortality risk with age than males (Austad, 2006, 2011). However, reproductive senescence – in form of menopause – happens much earlier in females than in males, and the degree to which other patterns of physiological deterioration with age are sex-biased remains a source of debate (Bonduriansky et al., 2008; Graves et al., 2006). Mammalian and invertebrate laboratory model systems also show sex differences in lifespan and sexes often differ in their response to genetic or dietary manipulations associated with lifespan extension (e.g. Bokov et al., 2011; Clancy et al., 2001; Selman et al., 2009; Strong et al., 2008). Evolutionary theory offers explanations for how and why differences in lifespan and aging might have arisen under natural selection (Bonduriansky et al., 2008; Williams, 1957). Based on the idea that the strength of natural selection against senescence hinges on the rate of ‘extrinsic mortality’ experienced in nature, Williams (1957) predicted “where there is a sex difference [in “extrinsic” mortality], the sex with the higher mortality rate and lesser rate of increase with fecundity should undergo the more rapid senescence.” Male-biased natural mortality rates are very widely observed in polygynous species, including many mammals and insects, where males tend to be larger and show mating behaviours and secondary sexual characteristics associated with substantial energy expenditure and mortality risk, but evidence for sex differences in senescence rates from the wild is much more limited (Bonduriansky et al., 2008; Clutton-Brock and Isvaran, 2007). Some support for the prediction has been obtained from a comparative study using data from long-term field studies of birds and mammals which found that male-bias in survival senescence rates increased with the degree of polygyny, and was largely absent in monogamous birds (Clutton-Brock and Isvaran, 2007). However, most studies of senescence in the wild have tended to focus on a single sex – typically females – and direct and detailed comparison of aging patterns between the sexes remain rare.

There is evidence that males show more rapid actuarial senescence than females from several wild ungulate populations (Catchpole et al., 2004; Gaillard et al., 1993; Jorgenson et al., 1997; Loison et al., 1999; Toigo et al., 2007), and from a single study of common lizards (Massot et al., 2011). Perhaps unsurprising, given their typically monogamous mating system and generally lack of sex bias in mortality; we have not been able to find evidence in the literature for significantly sex-biased survival senescence in wild birds. The few relevant studies on insects are puzzling: black field crickets studied in the wild showed no sex differences in the Gompertz function describing the rate of age-related increase in mortality risk, but when kept and monitored in “semi-natural” enclosures females showed a more rapid mortality rate increase but with a later onset than males (Zajitschek et al., 2009a; Zajitschek et al., 2009b). In stalk-legged flies, males were found to show rapid actuarial senescence in the wild but females were not (Kawasaki et al., 2008).

Studies directly comparing rates of reproductive senescence between the sexes in the wild remain very rare, not least because of the considerable challenges in determining reproductive success of males in polygynous species and in separating the contributions of the sexes to reproductive success in socially monogamous species. Field studies have provided evidence for age-related declines in a variety of male reproductive traits including competitive behaviour (Bonduriansky and Brassil, 2005; Jennings et al., 2010; Nussey et al., 2009), secondary sexual characteristics (Moller and De Lope, 1999; Vanpé et al., 2007; Velando et al., 2010), sperm performance and damage levels (Moller et al., 2009; Velando et al., 2011), and ultimately the number of offspring sired (Coltman et al., 1999; Nussey et al., 2009; Vanpé et al., 2009). Likewise, there is evidence across a large suite of females traits associated with reproductive performance – from timing of egg laying or parturition to number of offspring and egg or offspring size, right through to maternal investment after hatching or birth and offspring survival (e.g. Beamonte-Barrientos et al., 2010; Bouwhuis et al., 2009; Hewison and Gaillard, 2001; Martin and Festa-Bianchet, 2011; Nussey et al., 2009) – and several recent studies suggest important longer-term effects of parental age on offspring fitness (Bouwhuis et al., 2010b; Torres et al., 2011).

The challenge remains to quantitatively integrate and compare these changes in reproductive performance with age, and ultimately the differences in fitness costs of senescence, between the sexes (Bonduriansky et al., 2008; Promislow, 2003). In birds, social monogamy makes separating the contributions of female and male age from reproductive performance traits challenging, although a few studies have found evidence of sex-differences in aging of reproductive performance traits (Lecomte et al., 2010; Reed et al., 2008; Reid et al., 2003), or interactions between male and female age influencing such traits (Auld and Charmantier, 2011). In mammals, age-dependent fecundity trajectories have been compared between sexes in wild red deer and Soay sheep; both studies suggested that age-related declines begin slightly later but progresses more rapidly in males than in females (Nussey et al., 2009; Robinson et al., 2006), although the generality of this pattern remains to be confirmed.

From an evolutionary perspective, a central question is whether differences in natural selection on life history between sexes (sexually antagonistic selection) has resulted in the evolution of genetically programmed differences in allocation to somatic maintenance, and therefore in lifespan and aging patterns. Some support for this hypothesis comes from a cross-sectional study of culled red deer, in which – like many other ungulates – adult teeth are not replaced and tooth wear has been suggested as an important determinant of mortality in older animals (Carranza et al., 2004). The study took starting tooth size as an indicator of resource allocation to somatic maintenance and wear rate as a proxy of the need for repair and showed that, despite males being considerably larger and heavier than females and showing faster molar wear, sexual dimorphism in molar height in early life was minimal (Carranza et al., 2004). This supports life history theories, suggesting males allocate less in maintenance in accordance with their shorter life expectancy (Carranza et al., 2004). An alternative possibility is that earlier or more rapid senescence in males may be the direct physiological result of their heavier energetic allocation to growth and reproduction in earlier life, compared to females. Supporting evidence from wild populations remains scarce, although there is evidence of increased costs of reproduction in older individuals in single sex studies (e.g. Descamps et al., 2009; Torres and Velando, 2007). Studies comparing age-dependent costs of reproduction across sexes are lacking, although a very recent analysis of four European populations of barn swallows suggests males, but not females, pay a longevity cost of allocating resources to a secondary sexual characteristic, namely having long tail feathers (Balbontin et al., 2011).

3.2. Variation in senescence among traits

Williams (1957) predicted that senescence in different physiological systems associated with fitness should progress in synchrony, and reiterated the “expected evolution of synchrony” in a much later monograph on aging (Williams, 1999). This was odd as in both papers he discussed at length the discrepancies between the onset of athletic decline, menopause and life expectancy in humans. He appeared to think this obvious heterochrony was a quirk of our own species, stating: “modern human populations… have grossly abnormal environments that may greatly disrupt the coupling of senescence” (Williams, 1999). Evidence from the laboratory has revealed similarly complex and divergent patterns of senescence across physiological systems and functions (Burger and Promislow, 2006; Grotewiel et al., 2005; Herndon et al., 2002), leading to calls for more careful consideration of the system of heterochronous interactions associated with organismal aging (Promislow et al., 2006; Walker and Herndon, 2010). Of course, these examples might also simply reflect vagaries of the ‘grossly abnormal environments’ experienced in the laboratory compared to ecological reality. However, emerging evidence from natural populations refutes Williams’ prediction and suggests heterochrony is also a reality in evolutionarily relevant natural contexts.

Long-term individual-based field studies often collect longitudinal data on a host of phenotypic traits, including behavioural, reproductive and physiological parameters, as well as information on survival, as already discussed (see Section 3). There has been a very recent surge in studies directly comparing senescence rates across such traits, following calls for more integrated studies of aging in the wild (Massot et al., 2011; Nussey et al., 2008a). Evidence for uncoupling of patterns of survival and fecundity senescence remain quite scarce. In females, post-reproductive survival has been documented in wild mammals, but with the possible exception of some toothed whales, this seems more likely the product of chance than some adaptive reproductive cessation (Cohen, 2004; Packer et al., 1998; Thompson et al., 2007; Foote, 2008; Ward et al., 2009). There is also some evidence from wild female ungulates that fecundity senescence may begin later and progress more rapidly than age-related declines in survival probability (Bérubé et al., 1999; Catchpole et al., 2004; Jorgenson et al., 1997; Nussey et al., 2009). There is much clearer evidence for heterochrony of senescence among traits associated with reproduction in both males and females (e.g. Beamonte-Barrientos et al., 2010; Berman et al., 2009; Descamps et al., 2007; Ericsson et al., 2001; Evans et al., 2011; Hewison and Gaillard, 2001; Massot et al., 2011; Moller and De Lope, 1999; Nussey et al., 2009). For instance, emerging data suggest that male secondary sexual traits – despite theoretical expectations that they should be physiologically costly (Andersson, 1994) – do not actually senesce, although reproductive performance clearly does (Evans et al., 2011; Nussey et al., 2009). Evidence remains mixed on this front, with several studies documenting age-related declines in such traits (Moller and De Lope, 1999; Mysterud et al., 2005; Vanpé et al., 2007; Velando et al., 2010). Further careful integrated cross-trait comparisons are now required to understand why and how such heterochrony might evolve under natural conditions (Massot et al., 2011; Nussey et al., 2008a; Promislow et al., 2006).

3.3. Individual, genetic and environmental variation in senescence rates

Just as patterns of senescence appear to vary across systems and traits in humans and in model systems, so they are known to vary dramatically among individuals. Field studies testify that age-related declines in fitness-associated traits vary among individuals in natural populations. A number of studies have directly quantified individual variation in aging rates (Brommer et al., 2010; Brommer et al., 2007; Wilson et al., 2007), whilst others have provided indirect evidence by documenting associations between early-life environment or life history and aging rates later on (Bouwhuis et al., 2010a; Hayward et al., 2009; Massot et al., 2011; Nussey et al., 2007; Reed et al., 2008). Several studies have also compared whether declines in such traits are better predicted by chronological age or years until death. The latter would provide support for the idea that declines were to some degree age-independent. Evidence is mixed, although there is growing support for some degree of age-independent deterioration, presumably reflecting differences in past experience or genetic make-up among individuals (Hayward et al., 2009; Martin and Festa-Bianchet, 2011; Nussey et al., 2011).

Classical evolutionary theory predicts that variation in the aging process should have a genetic basis (Rose, 1991), and work in model laboratory systems supports this prediction both through artificial selection experiments and through genetic manipulations that influence aging and lifespan (Kirkwood and Austad, 2000; Partridge, 2010). This research suggests conserved or ‘public’ genetic and physiological pathways, which are modulated by diet, across evolutionarily distant related taxa to modulate aging and lifespan (Fontana et al., 2010; Partridge, 2010). However, the generality of these findings to longer-lived organisms or even to the same species experiencing the more complex and challenging environments of the natural world has yet to be established, as has the evolutionary significance of laboratory mutations, knock-outs and knock-ins associated with lifespan extension. Recent work on nematodes, flies and mice shows that changing the environmental context profoundly alters how long these mutants live and how well they perform, and suggests that the responses of wild-caught animals to typical treatments can differ markedly from those subjected to years of selection for laboratory life (e.g. Chen et al., 2007; Giorgio et al., 2012; Harper et al., 2006; Linnen et al., 2001; Sgro and Partridge, 2000; van Voorhies et al., 2005). In many respects, understanding these complex genotype-by-environment interactions and the lifelong fitness and health consequences of dietary, pharmaceutical and genetic manipulations that extend lifespan for lab models in long-lived organisms under more complex environmental conditions is pivotal to hopes of meaningful intervention to alleviate senescence in humans and domestic animals.

Several studies of wild birds and mammals have used reconstructed population pedigrees and quantitative genetic methods to test whether aging rates in fitness-related traits are heritable (reviewed by Wilson et al., 2008). Specifically, in red deer, Soay sheep, and mute swans increased genetic variation with age and, by implication, heritable variation in aging patterns, has been documented (Charmantier et al., 2006a; Wilson et al., 2007). Importantly in the swan study, the association between early first reproduction and early last reproduction mentioned earlier was shown to have a genetic basis, providing perhaps the most compelling evidence yet for antagonistic pleiotropy from a wild animal (Charmantier et al., 2006b). Also, in red deer, data suggest that the association between increased maternal resource allocation towards reproduction in early life and rapid reproductive senescence in red deer had a genetic basis, the effect was marginally statistically non-significant (P = 0.06; Nussey et al., 2008b). Additionally, studies in common gulls and collared flycatchers documented variation in aging rates but did not find evidence that this variation had a genetic basis (Brommer et al., 2010; Brommer et al., 2007). It remains unclear whether these latter studies failed to find a genetic basis for variation in aging rate due to a lack of statistical power or to real biological differences among species. Researchers studying aging in fish and reptiles have tended to adopt a different approach, bringing individuals from populations experiencing different selective pressures into the laboratory and comparing their survival and physiology under controlled laboratory conditions. This approach clearly identifies genetic differences among populations that influence aging rates or maintenance and repair mechanisms providing some support as well as important challenges to classical evolutionary theories of aging (Reznick et al., 2004; Robert and Bronikowski, 2010; Robert et al., 2007).

Studies in wild animals provide spectacular examples of the effect that the environment can have on the aging process. For instance, a comparison of mortality curves in wild and laboratory stalked-legged flies revealed that males senesce at least twice as rapidly under natural conditions as in the laboratory (Kawasaki et al., 2008). Also, wild medflies brought into the laboratory as adults often survived considerably longer than laboratory-reared flies did from hatching, suggesting a profound effect of early environment on aging rates (Carey et al., 2008). In longitudinal field studies of birds and mammals poor conditions in early life are seen to exacerbate the rate and onset of decline in survival and reproductive performance (Beamonte-Barrientos et al., 2010; Bouwhuis et al., 2010a; Hayward et al., 2009; Nussey et al., 2007; Reed et al., 2008), and commonly show that the survival of older adults is more sensitive to environmental conditions (e.g. Gaillard et al., 2000; Reichert et al., 2010; Toigo et al., 2007). An interesting pattern emerged from a recent study of tawny owls, which experience profound variation in food availability associated with population cycles of their main prey. Food availability early in life affected survival and reproductive performance in old age only. However, it did not appear to influence aging rates per se (Millon et al., 2011). The relationship between environmental conditions and senescence is clearly very complex in natural populations, and identifying the precise environmental or physiological causes of relationships is often likely to be impossible. However, field researchers, particularly those working on birds, have a long history of integrating longitudinal field data collection with experimental manipulation. The integration of field observation and field experimentation could allow us to determine the effects of factors such as nutrition, infection, thermal stress and competition for lifespan and aging rates in nature.

3.4. Tests of life history theories of aging

Both antagonistic pleiotropy and disposable soma theories of aging rest on the central idea of trade-offs between early life reproduction and later life maintenance of physiological function and reproductive output (Kirkwood and Rose, 1991). As already discussed, work on laboratory model systems generally supports these ‘life history’ theories of aging (Table 1, Kirkwood and Austad, 2000; Partridge and Barton, 1996). However, environmental conditions are likely to alter both the nature and magnitude of the costs involved and the way that natural selection acts on life history allocation, and laboratory conditions are profoundly different to those experienced by wild animals. To understand how the environment and natural selection interact to shape variation in senescence among species, populations and individuals it is crucial to test these hypotheses in natural settings. The emergence of long-term, individual-based field studies and statistical tools to dissect within- and between-individual variation has now made it possible to robustly test the consequences of an individual’s early life history decisions for life expectancy and aging patterns in later life (Cam et al., 2002; Rebke et al., 2010; van de Pol and Verhulst, 2006). Importantly, evidence is now emerging from natural populations supporting the central prediction that an increased allocation of resources to early life reproduction is associated with earlier or more rapid aging in later life.

For instance, a long-term study of collared flycatchers in Sweden, where researchers experimentally increased the brood size of females in early adulthood, has shown that these females produced consistently smaller subsequent broods, with a suggestion that brood size also declined more rapidly with age, compared to control females (Gustafsson and Part, 1990). Subsequently, studies of female great tits, guillemots and red deer have all demonstrated that increases in fecundity or reproductive performance in early adulthood are associated with more rapid declines in reproductive performance in later life (Bouwhuis et al., 2010a; Nussey et al., 2006; Reed et al., 2008). In meerkats, a social mammal in which a dominant pair almost completely monopolize breeding attempts within any group, females experiencing more stressful ascensions to breeding dominance – as a result of competition with same-age siblings in the group – show more rapid reproductive senescence in later life (Sharp and Clutton-Brock, 2011). Also, mute swans that start their breeding careers earlier in life end their reproductive lifespan earlier as well (Charmantier et al., 2006b), and red squirrels that start breeding early have shorter subsequent life expectancies (Descamps et al., 2006). An intriguing variation on this pattern is seen in female common lizards. Those that allocate relatively more in offspring at two years of age actually show increased subsequent reproductive success but at the cost of more rapid declines in survival probability, in essence sacrificing lifespan for high reproductive output (Massot et al., 2011). Similar evidence of a trade-off between parental lifespan and offspring survival has been recorded in a bird species, the red-billed chough (Reid et al., 2010). While these studies support the expectation of maintenance or survival costs of early reproduction, some also show that natural selection favours individuals allocating more resources towards early-life reproduction, as theory would predict (Bouwhuis et al., 2010a; Descamps et al., 2006; Gustafsson and Part, 1990). Where field studies have tested the predictions of life history theories of aging with detailed longitudinal data and robust statistics, they have tended to find support (Peron et al., 2010).