We investigated effects of natal conditions and early life reproduction on rates of reproductive senescence in female great tits, and found evidence for both: rates of late life age-specific deterioration in fledgling number were associated with female immigrant status and early life fledgling production. Quantitatively similar effects in terms of point estimates, although non-significant, were found for late life declines in recruit production with respect to both factors. A range of other measures of the natal and early life reproductive environment were unrelated to the rate of senescence. Hence, these data indicate that although some elements of prior experience contribute to variation in rates of senescence in this population, these effects are found only for a limited range of variables.
Our results show that female great tits immigrating into Wytham Woods achieve similar reproductive output as locally born females in the first two years of life, but subsequently suffer from elevated rates of reproductive senescence. As local knowledge is expected to increase with breeding experience and age, such a pattern seems unlikely to result from late expressed differences in local adaptation. Increased costs of reproduction for the same level of early life reproductive output, perhaps because of poorer initial local knowledge, seems a more likely candidate mechanism underlying this observation. Alternatively, as some of these birds are known to originate from low-quality areas surrounding Wytham Woods (Riddington & Gosler 1995; Verhulst, Perrins & Riddington 1997), immigrants might suffer from long-term effects of natal conditions that were not captured in our measures of natal conditions for locally born females. Finally, immigrants could have come from areas in which rates of senescence have evolved to be higher, for example, because of higher levels of extrinsic mortality (Medawar 1952; Williams 1957). However, genetic differentiation to this extent seems implausible (especially because of the high level of immigration reported in this study), and at present too little is known about the genetic basis of senescence in passerines (Brommer, Wilson & Gustafsson 2007), or about variation in rates of senescence between populations of the same species (Reznick et al. 2004), to evaluate the likelihood of this option.
As well as accelerated reproductive senescence, immigrants also show a reduction in lifetime reproductive success of 15% in our sample of birds (Fig. 3a; also see McCleery & Clobert 1990). In contrast to the situation in a Dutch population of great tits (Verhulst & van Eck 1996), this difference cannot be attributed to a shorter reproductive life span of immigrants, as previous analyses in this population have shown similar survival rates for immigrant and locally born females (0·41 and 0·40, respectively; Clobert et al. 1988), and we also found no difference in average ALR between them in this study. Recapture probability of immigrants in our population is, however, lower than that of locally born birds (0·77 and 0·84, respectively; Clobert et al. 1988), and indeed an analysis of the number of missing breeding attempts over all ages in the 1029 females reveals this number to be higher for immigrants than locally born birds (average: immigrants: 0·31, locally born birds: 0·27, χ²1 = 6·37, P = 0·011). Missing breeding attempts include failed attempts (6% of all attempts), successful attempts within the study area for which the female was not caught (4% of all attempts), skipped attempts and perhaps breeding attempts in natural cavities or outside the study area, although the frequency of these last two is thought to be low (Harvey, Greenwood & Perrins 1979; Perrins 1979). As the extent to which each type of breeding attempt contributes to the number of missing attempts in relation to female immigrant status, and therefore the extent to which they might bias our estimates of LRS, is presently unknown, we should be cautious in interpreting the difference. But, assuming a difference in LRS exists, our findings of accelerated reproductive senescence and lower lifetime fledgling production in immigrants suggest it to be at least partly because of reduced reproductive performance. The remaining status effect could originate from both reduced survival of fledglings of immigrant females, and a higher probability for offspring of immigrant females to leave the study area if natal dispersal is heritable (Doligez & Pärt 2008).
Besides female immigrant status (which could be regarded as an indirect natal effect), we found no evidence that direct effects of natal environmental conditions (density, year quality, beech mast crop), or natal brood characteristics (maternal age, hatching date, sibling number) were associated with rates of reproductive senescence within Wytham-born females. This result supplements recent analyses of other life-history traits in locally born females in this population, which also showed that natal environmental variation has at most very weak effects on post-recruitment fitness components in females (Wilkin & Sheldon 2009). These findings contrast with previous studies of senescence in populations of red deer and red squirrels, which reported associations between natal density (Nussey et al. 2007) and food abundance (Descamps et al. 2008a) and rates of reproductive senescence, respectively. With respect to natal density, the absence of a relationship with rates of senescence in our population might perhaps be explained by birds escaping resource competition effects by dispersal or emigration; no such behavioural response was possible for the red deer born on the Isle of Rum. Indeed, local recruitment is known to be inversely related to natal density in Wytham (Wilkin et al. 2006), although it is unknown whether emigration or mortality underlies this pattern. But if natal density drives emigration, late acting costs of natal density in terms of accelerated reproductive senescence may be apparent only in immigrants, as found here. Studies in which fledglings are followed until recruitment over larger spatial scales are required to test this hypothesis.
With respect to food abundance in the year of fledging, we found no effect on rates of reproductive senescence, but we did find a positive correlation with average late life reproductive performance. Beech mast crop has previously been shown to be important for subsequent reproductive success of female great tits in adult life (Verhulst 1998), and the long-term effect reported here raises the question of the proximate mechanism by which high quantities of beech mast in the first winter of life affect female reproductive output in the long run. As with beech mast crop, sibling number did not relate to rates of reproductive senescence, but correlated positively with average late life reproductive performance. This is in contrast to experimental evidence for a negative effect of natal brood size on offspring future reproduction in collared flycatchers Ficedula albicollis (Gustafsson & Sutherland 1988), and perhaps suggests positive effects because of the heritability of clutch size (McCleery et al. 2004) to outweigh negative effects of being reared in a larger brood; experiments are probably needed to disentangle these effects.
Previous analyses of maternal age on offspring age-specific reproductive performance in our population showed that offspring born from older mothers perform better early in life, but suffer from an earlier onset, and stronger rate, of reproductive senescence later in life (Bouwhuis et al. 2010). A negative interaction between maternal age and female age to explain late life fledgling and recruit production was therefore expected. Parameter estimates for these interactions were indeed negative, but failed to reach statistical significance (Table 1 and Fig. S1e,f). Apparently, the effect of maternal age on their daughter’s life-history trajectory works over the entire age range, and is not just driven by the senescent phase of life.
Early life reproduction
In addition to an effect of female immigrant status, we found individual rates of reproductive senescence to be associated with early life fledgling production. Birds with high early life reproductive performance maintained a high level at the start of the senescent phase, but then suffered from accelerated deterioration, whereas birds with low early life reproductive performance showed a less pronounced senescent decline from their lower starting point. Such a late life cost of reproduction, not in the level of late life performance but in its rate of change, and apparent only in a subset of individuals, may explain why a previous study of the cost of reproduction in our population failed to find one (Doligez et al. 2002). It, however, supplements an earlier finding of a delayed survival cost of reproduction in the female great tits in our study area (McCleery et al. 1996). In addition, the pattern found here matches recent findings in long-lived red deer (Nussey et al. 2006) and common guillemots (Reed et al. 2008), and supports the assumption of the disposable soma hypothesis of senescence of a trade-off between early life and late life performance (Kirkwood & Rose 1991). Furthermore, our results support the idea that a disposable soma mechanism can promote the evolution of senescence, as birds adopting a strategy of high early life reproductive performance and high rates of reproductive senescence enjoyed the highest lifetime reproductive success. In fact, a selection analysis revealed strong positive directional selection on early life fledgling production, such that an evolutionary response towards higher levels of early life reproduction and stronger rates of senescence is expected if the trade-off between early life and late life reproduction has a genetic basis and is unconstrained by other genetic correlations. Quantitative genetic analyses of age-specific reproduction (e.g. Charmantier et al. 2006a,b; Brommer, Wilson & Gustafsson 2007; Wilson et al. 2007; Nussey et al. 2008b; Wilson, Charmantier & Hadfield 2008) are required to address this issue.
Note that the previous discussion is based on a subset of females in our data set for which we had complete early life histories. This subset only included about half of the entire set of females reaching a minimum age of 3, the other half of which was subject to experiments in early life, or had missing breeding attempts. Adding these females to the data set, assuming missing attempts to have resulted in no fledglings, showed that, encouragingly, females who had undergone experimental manipulation in early life showed a qualitatively similar trade-off between early life and late life reproduction as females with complete life histories. This was not the case for females with missing breeding attempts in early life. Because, as mentioned before, missing breeding attempts can be the result of a heterogeneous set of causes, it is difficult at present to interpret the discrepancy. Earlier identification of breeding females, as well as improved opportunities to track individuals over time are required to investigate this, but given the life history differences among these groups this is generally advisable. Indeed, greater attention to the problem of missing data seems advisable for a wide range of studies that estimate fitness and selection in wild populations (also see Nakagawa & Freckleton 2008).
Finally, we found no associations between environmental conditions during early life reproduction and rates of late life performance declines. Surprisingly, such effects were found in a long-lived seabird, the common guillemot (Reed et al. 2008), even though long-lived organisms might be expected to adjust reproductive effort to environmental conditions to a greater extent than shorter-lived ones, and therefore to suffer less repercussion of early life conditions (Williams 1966). Perhaps the magnitude of environmental variation was larger for common guillemots compared with great tits, or the damage done was of a different type. Study of physiological correlates of reproductive senescence and early life environmental conditions may shed light on this matter.