A. P. Møller, Laboratoire de Parasitologie Evolutive, CNRS UMR 7103, Université Pierre et Marie Curie, Bât. A, 7ème étage, 7 quai St. Bernard, Case 237, F–75252 Paris, Cedex 05, France. Tel.: +33 1 44272594; Fax: +33 1 44273516; e-mail: email@example.com
Evolutionary theories of senescence suggest that aging evolves as a consequence of early reproduction imposing later viability costs, or as a consequence of weak selection against mutations that act late in life. In addition, highly social species that live in sites that are protected from extrinsic mortality due to predation should senesce at a slower rate than solitary species. Therefore, species that start reproducing late in life should senesce at a slower rate than species that start reproducing early. In addition, social species should senesce more slowly than solitary species. Here I investigate the rate of senescence using an extensive data set on longevity records under natural field conditions to test predictions about the evolution of senescence among 271 species of birds. Longevity records increased with sampling effort and body mass, but once these confounding variables were controlled statistically, there was a strongly positive relationship between relative longevity and relative adult survival rate. Relative longevity after controlling statistically for sampling effort, body mass and adult survival rate, increased with age at first reproduction, but not with degree of breeding sociality. These findings suggest that the evolution of senescence is related to timing of first reproduction, but that the evolution of breeding sociality has played a negligible role in the evolution of senescence.
Senescence is the deterioration of the phenotype with increasing age, causing a reduction in physiological functioning and a reduction in residual reproductive value (Fisher, 1930). Current theories of senescence can broadly be categorized as (1) those relating to evolutionary theories associated with accumulation of mutations or antagonistically pleiotropic mutations, (2) those relating to trade-offs in use of antioxidants or (3) those relating to accumulation of damage to the disposable soma caused by an inability to maintain efficient DNA, cell, tissue and organ repair (Williams, 1957; Hamilton, 1966; Rose, 1991; Partridge & Barton, 1993; Kirkwood & Austad, 2000). The evolutionary theories of senescence (1 above) differ in the mechanisms that are generating ageing, although all assume that senescence arises as an adaptation of life histories to differences among species in the effects of the rate of increase in extrinsic mortality (Rose, 1991; Finch & Kirkwood, 2000; Kirkwood & Austad, 2000). In contrast, the mechanistic theories of senescence (2 and 3 above) assume that the inevitable deterioration in repair mechanisms, for example due to damage to DNA caused by a deficiency of antioxidants, results in a deterioration of the phenotype at old age (Rose, 1991; Finch & Kirkwood, 2000; Finkel & Holbrook, 2000).
Evolutionary theories of ageing suggest that the rate of ageing should increase as the rate of extrinsic mortality increases, while no such prediction can be made from the mechanistic theories of senescence (Williams, 1957; Hamilton, 1966; Rose, 1991; Partridge & Barton, 1993,1996; Charlesworth, 1994). The reason is that as extrinsic rates of mortality increase, this should result in the elimination of more individuals late in life. If rates of fecundity change during the lifetime of individuals of a species, this should result in an increase in the rate of ageing in species in which fecundity does not increase late in life compared to those in which it does increase (Abrams, 1993).
Previous tests of theories of senescence have relied on selection experiments, phenotypic manipulations or comparative analyses. First, selection experiments that have selected for late start of reproduction in general have found a delay in onset of senescence. For example, selection for a delay in start of reproduction in Drosophila melanogaster caused a delay in senescence (Partridge & Fowler, 1992). Second, phenotypic manipulation experiments that have artificially increased early investment in reproduction have shown an earlier onset of senescence than in control treatments that maintained their clutch size. For example, brood size manipulation experiments on the collared flycatcher Ficedula hypoleuca have shown that increased early parental effort due to an experimental increase in brood size caused an increase in the rate of senescence (Gustafsson & Pärt, 1990). Likewise, barn swallows Hirundo rustica that invested disproportionately in offspring during the first year of breeding aged more rapidly than individuals with small parental investment (Saino et al., 2002). Third, comparative studies of interspecific variation in longevity have indicated that the protected and sheltered life of eusocial insects is associated with an increase in longevity as compared to noneusocial species (Keller & Genoud, 1997). Furthermore, ant species with high extrinsic rates of mortality had high rates of senescence (Keller & Genoud, 1997). However, that study did not control statistically for sampling effort, and hence for the probability of finding an extremely long-lived individual, or for body size and hence for the fact that large species (and individuals) generally survive longer than small species (e.g. Promislow, 1991).
The objective of the present study was to test the prediction arising from evolutionary theories of senescence that the rate of ageing and hence the maximum record of longevity should increase in species with delayed onset of reproduction. In particular, in species that are severely limited by availability of sites of reproduction and hence are subject to the evolution of delayed onset of reproduction, we should expect the rate of senescence to decrease. To test this prediction I exploited social reproduction or coloniality as a test case. Colonial breeding has evolved independently numerous times, and colonially reproducing individuals are restricted to a few sites that are highly protected against potential predators (Lack, 1968; Burger et al., 1980). Colonial breeding is characterized by breeding individuals feeding on unpredictable and hence nondefensible food resources (Lack, 1968; Burger et al., 1980). Therefore, colony size may increase until available space for nests is exhausted, or until competition for limiting food resources makes it more beneficial for individuals to move to another colony (Furness & Birkhead, 1984). Limited availability of nest sites should select for delayed onset of reproduction, which in turn could be predicted to result in a delay in onset of senescence. Hence, we should expect the rate of senescence to be slower in colonially than in solitarily breeding species of birds. In addition, colonially breeding species of birds should start to reproduce at an older age than solitary species, and once interspecific differences in onset of reproduction were controlled statistically, there should be no more explanatory power of coloniality for the evolution of senescence. To test these predictions I used an extensive data set on longevity records of 271 free-living bird species from Europe, obtained from the literature, but adjusted for variation in sampling effort as reflected by the number of recoveries and recaptures of the different species, and adjusted for extrinsic mortality. Extrinsic mortality was estimated from annual adult mortality rate while intrinsic mortality was estimated from maximum longevity records of banded birds.
Materials and methods
I extracted information on the maximum longevity of all European species from Cramp & Perrins (1977–1994) and Glutz & Bauer (1985–1997). The information reported in these handbooks is based on extensive literature search and data provided by major bird ringing schemes.
Longevity records only provide reliable information on maximum lifespan if records are adjusted for sampling effort. Among 271 species of common birds in Europe for which longevity records are available, the total number of recoveries and recaptures of banded birds across Europe ranged from 106 to 187 764 (http://www.euring.org). Thus, variation is sampling effort by over three orders of magnitude has to be controlled statistically to allow study of biologically relevant information. Therefore, I used the total number of recoveries reported by the European bird ringing organization EURING as a measure of variation in sampling effort for all records of longevity.
Species with high survival rates will by necessity also have extreme longevity records. However, relatively long lifespan for a given survival rate will provide information about rate of senescence, since longevities that are greater than predicted for a given survival rate will suggest a slow rate of senescence, while short records of longevity will suggest a fast rate of senescence. I extracted information on adult annual survival rate from Cramp & Perrins (1977–1994) and Glutz & Bauer (1985–1997). While some of these estimates are based on capture-mark-recapture methods that provide rigorous ways of adjusting for heterogeneity in capture probability, other are based on less rigorous methods. However, there is no reason to believe that differences in methods should bias estimates of rates of senescence in any particular direction, while such heterogeneity is likely to make any tests conservative.
Age at first reproduction
I extracted information on age at first reproduction from the sections on reproduction in Cramp & Perrins (1977–1994) and Glutz & Bauer (1985–1997). In species where there was a range of values reported for age at first reproduction, I used the minimum value reported.
Colony size was extracted as the maximum colony size reported by Cramp & Perrins (1977–1994) and Glutz & Bauer (1985–1997). Colony sizes were subsequently transformed to a log10-scale with the values 0, 1, 2, 3, 4 and 5, where 0 is solitary and 5 is a colony size exceeding 100 000. A previous study of coloniality in swallows and martins has shown that maximum and mean colony sizes are strongly positively correlated (Møller et al., 2001), suggesting that maximum colony size provides a reliable estimate of the average size of colonies of a particular species.
I extracted information on mean body mass of adult birds from Dunning (1993).
The data set is reported in electronic Appendix 1.
I controlled for similarity in phenotype among species due to common ancestry by calculating standardized independent linear contrasts (Felsenstein, 1985), using the computer program CAIC (Purvis & Rambaut, 1995). All regressions were forced through the origin (Garland et al., 1992), and standardization of contrast values was checked by examination of absolute values of standardized contrasts vs. their standard deviations (Garland, 1992; Garland et al., 1992). Plotting the resulting contrasts against the variances of the corresponding nodes revealed that these transformations made the variables suitable for regression analyses. In cases where extreme residuals were recorded, I tested for the robustness of the findings by excluding contrasts with studentized residuals greater than 3.00 (Jones & Purvis, 1997). Likewise, I ranked the independent variable to test if the conclusions remained unchanged, and in no case did this procedure give rise to conclusions different from those obtained with the contrast values.
I log10-transformed research effort, longevity, age at first reproduction and body mass, while adult survival rate was square root arcsine-transformed to obtain variables that were normally distributed.
Records of longevity by necessity must be controlled for sampling effort because it is by definition easier to record an extremely old individual in a large than a small sample. While this is obvious, no previous comparative study of senescence has to the best of my knowledge controlled statistically for this sampling effect. I used sample size as a simple procedure to control for sampling effort, by using the log-transformed number of recoveries as an independent variable in the analyses.
A second cause of bias in comparative analyses of senescence is that large species have greater longevity, survive better and have later start of first reproduction than small species (e.g. Promislow, 1991; Bennett & Owens, 2002). In addition, colonial species have larger body size than solitary species (Lack, 1968). Therefore, apparent covariation between longevity and age at first reproduction and coloniality, respectively, may arise simply because of covariation with body size. Hence, allometric effects must be controlled statistically in order to avoid such spurious correlations, and I used log-transformed body mass in the statistical analyses to control for allometry effects.
Relative maximum longevity after controlling statistically for sampling effort and body mass should by definition be positively related to relative survival rate after controlling for sampling effort and body mass, because a high survival rate should eventually produce extremely old individuals. However, species with a relative maximum longevity larger than predicted for a given relative survival rate could be considered to senesce slowly, while species with a smaller relative maximum longevity could be considered to senesce rapidly. Relative longevity was estimated from a multiple regression analysis maximum longevity as the dependent variable and sampling effort, body mass and adult survival rate as independent variables.
This procedure for estimating rate of senescence was cross-validated with estimates of actuarial rates of senescence reported by Ricklefs (1998).
I tested for a hypothetical relationship between relative longevity and degree of coloniality, using multiple regression analysis with relative longevity as defined above. This regression analysis was repeated for species-specific data and for standardized linear contrasts. I tested for an effect of relative age at first reproduction on senescence by relating relative longevity to age at first reproduction with body mass as an additional independent variable. In this multiple regression analysis absolute longevity was the dependent variable, while sampling effort, body mass, adult survival rate, and age at first reproduction were used as independent variables. Finally, I used multiple linear regression analysis to test for the independent effects of coloniality and age at first reproduction on relative longevity. If the effect of age at first reproduction fully accounted for interspecific variation in senescence, coloniality should not explain additional variation in the variable reflecting senescence. This was done in a multiple regression analysis with absolute longevity as the dependent variable and sampling effort, body mass, adult survival rate, age at first reproduction and colony size as independent variables.
Information was missing for some species for certain variables, causing sample sizes to differ slightly among analyses.
Estimates of longevity were positively related to research effort of different species of birds (Table 1). Likewise, longevity increased with body mass (Table 1).
Table 1. Correlation matrix between the six variables investigated. n = 169 species.
Age at first reproduction
*P < 0.0001
Age at first reproduction
Estimates of adult survival rate increased with body mass (Table 1). Estimates of adult survival rate may as all variables be confounded by sampling effort, with greater variance in estimates at low than at high sample sizes. However, adult survival rate was not significantly related to research effort (Table 1). Thus adult survival rate had to be adjusted statistically for body mass in order to obtain an estimate that was unbiased by effects of allometry.
Relative longevity, after adjusting for research effort and body mass, increased with adult survival rate [Fig. 1; partial regression: F = 33.07, d.f. = 1165, r2 = 0.166, P < 0.0001, slope (SE) = 0.434 (0.075)]. A multiple regression analysis of contrasts with longevity as the dependent variable and adult survival rate, research effort and body mass as independent variables revealed a significant partial regression for adult survival rate [partial regression for longevity and adult survival rate in an analysis of contrasts: F = 4.61, d.f. = 1163, r2 = 0.028, P = 0.033, slope (SE) = 0.166 (0.077)]. Thus, species with relatively high adult survival rates for their body mass also had relatively long lifespan for body mass and research effort of that species.
Relative longevity after adjusting for research effort, body mass and survival rate increased with maximum colony size [Fig. 2; partial regression: F = 11.63, d.f. = 1164, r2 = 0.066, P = 0.008, slope (SE) = 0.035 (0.010)]. A multiple regression of contrasts for longevity as the dependent variable and adult survival rate, research effort, body mass and maximum colony size as independent variables did not reveal a significant partial regression for colony size (partial regression for longevity and colony size based on analysis of contrasts: F = 0.28, d.f. = 1162, r2 = 0.002, P = 0.60). Thus, the tendency for reduced rate of senescence in colonial species was a peculiarity of specific taxa with many species.
Colonial species had relative late start of first reproduction, after adjusting for body mass [Fig. 3; F = 71.24, d.f. = 1268, r2 = 0.210, P < 0.0001, slope (SE) = 0.070 (0.008)]. However, a multiple regression of contrasts with age at first reproduction as the dependent variable and body mass and maximum colony size as independent variables did not reveal a significant partial regression for colony size (partial regression for longevity and colony size based on analysis of contrasts: F = 0.30, d.f. = 1264, r2 = 0.0001, P = 0.58). This implies that the relatively late start of first reproduction in colonial species was a feature of particular taxa with many species.
After entering research effort, body mass, survival rate, colony size and relative age at first reproduction as predictor variables, relative longevity was predicted by relative age at first reproduction and colony size (Fig. 4; Table 2). A multiple regression of contrasts with longevity as the dependent variable and research effort, body mass, survival rate, colony size and age at first reproduction as independent variables only revealed a significant partial regression for age at first reproduction, but not for colony size (Table 2). Thus, relative age at first reproduction covaried with relative longevity when taking similarity among taxa due to common descent into account. In addition, there were significant partial regressions for research effort, body mass and adult survival rate (Table 2).
Table 2. Multiple linear regression analyses of longevity of 169 bird species in relation to body mass, sampling effort, adult survival rate, age at first reproduction and colony size. See Methods for further information on transformations and calculations. The statistics for the models were for species F = 42.91, d.f. = 5, 163, r2 = 0.57, P < 0.0001 and for standardized linear contrasts F = 38.93, d.f. = 5, 161, r2 = 0.55, P < 0.0001.
Age at first reproduction
Age at first reproduction
The measure of senescence used here was cross-validated against the measure of actuarial senescence recommended by Ricklefs (1998). The parameter ω scaled significantly with body mass for all estimates reported by Ricklefs (1998) for birds [F = 42.33, d.f. = 1.16, r2 = 0.73, P < 0.0001, slope (SE) = −0.063 (0.010)]. Therefore, I related longevity to sampling effort, survival rate, body mass and the senescence parameter ω, showing a strong relationship between the two measures of senescence [partial regression for ω for the species that were common to the two data sets: F = 73.98, d.f. = 1.7, r2 = 0.91, P < 0.0001, slope (SE) = −2.18 (0.25)]. This shows that the two senescence parameters reflect the same underlying phenomenon.
The main findings of this study were that (1) longevity records for birds provided estimates of senescence, after having accounted for variation due to sampling effort, allometry and adult survival rate. (2) Colonially breeding species did not senesce at a slower rate than solitarily breeding species. (3) Species that started to reproduce relative late in their life senesced at a slower rate than species that started to reproduce early. I will discuss each of these findings briefly.
The occurrence of senescence is well established for birds based on calculations of actuarial senescence (Ricklefs, 1998,2000). In addition, the rate of senescence has been shown to be an increasing function of baseline mortality rate (Ricklefs, 1998). Here I used longevity records for 271 individually banded free-living bird species to test evolutionary hypotheses of ageing. Relative longevity after adjusting for sampling effort and allometry increased with relative adult survival rate after adjusting for allometry. Both longevity and adult survival rate are known to scale with body mass (e.g. Bennett & Owens, 2002). Although, this relationship explained 20% of the variance, most variance remained unexplained. Maximum longevity does not provide information on extrinsic or intrinsic mortality rates (Ricklefs, 1998). Some unexplained variance in relative longevity after adjusting for sampling effort and allometry could be due to senescence, with species with positive values senescing at a slower rate than predicted for a given survival rate, while species with negative values senesced faster. Variation in relative longevity, after adjusting for sampling effort, body mass and adult survival rate, was positively related to degree of coloniality in analyses of species-specific data, but not in analyses of contrasts. This implies that rate of senescence is not significantly associated with breeding sociality, and that the pattern observed for species-specific data was due to a high degree of clustering of senescence and sociality in a few taxa. This finding is at contrast with previous reports on social insects, suggesting that the evolution of sociality was associated with a reduction in rate of senescence (Keller & Genoud, 1997). A possible explanation for this discrepancy is given below.
Colonially breeding species of birds start their first attempt of reproduction at an old age (Lack, 1968; Burger et al., 1980), although the present study showed that this was not a general phenomenon across taxa (as demonstrated by an analysis of contrasts). While age at first reproduction across bird species was negatively related to rate of senescence, as shown by analyses of contrasts, breeding sociality did not confound that relationship. Although, coloniality has evolved in many different taxa, and although it is associated with a suite of behavioural, life history and ecological characteristics (e.g. Lack, 1968; Burger et al., 1980), I found no general relationship between evolution of senescence and evolution of coloniality in birds. Once the statistical effects of relative age at first reproduction on relative longevity (as an estimate of the rate of senescence) had been controlled statistically, there was no further covariation between coloniality and relative longevity. This suggests that relative age at first reproduction (or a variable closely associated with it) rather than breeding sociality is a determinant of senescence.
The analyses presented here were based on longevity records that strongly depended on research effort. In fact, research effort accounted for 10% of the variance in longevity among species and 34% of the variance in analyses of contrasts, which are proportions of variance that exceeded the effects of coloniality and age at first reproduction on the estimates of senescence. Previous comparative studies of senescence have neglected the effects of sampling effort (e.g. Promislow, 1991; Keller & Genoud, 1997; Tella et al., 2002). Given the amount of variance explained by sampling effort alone this may raise questions about the robustness of conclusions in previous comparative studies of senescence.
Colonially breeding bird species did not generally senesce at a slower rate than solitary species in the present comparative analysis. This finding is surprising given suggestions about the role of sociality in the evolution of senescence (Keller & Genoud, 1997), but perhaps unsurprising given a range of findings concerning extrinsic mortality in colonial species of birds. These species have more species of parasites than solitarily breeding species (Tella, 2002). In addition, colonial species have a stronger impact of parasite-induced mortality (Møller et al., 2001), and they have therefore evolved stronger immune responses than solitary species (Møller & Erritzøe, 1996; Møller et al., 2001). Due to the high levels of competition for limiting breeding sites, females of colonial species, but not males, have higher levels of testosterone to back up intense competition for reproduction as compared to solitary species (Møller et al., 2005). Since females of colonial species have high levels of testosterone compared to females of solitary species, this has resulted in higher concentrations of maternal testosterone in eggs (Smith et al., 2005). This is a general phenomenon across birds, with highly colonial species producing eggs with high levels of testosterone (D. Gil, A.P. Moller, N. Saino, C. Spottiswoode & P. Surai unpublished). Since testosterone and other androgens can have negative effects on immunity (Folstad & Karter, 1992), elevated levels of testosterone in female colonially breeding bird species and their eggs may have implications for host-parasite interactions. However, rates of senescence did not differ between colonially and solitarily breeding species. Although, a previous comparative study of mammals indicated that species with high richness of parasites senesce at a higher rate than species with few parasites (Morand & Harvey, 2000), the present study suggests that the disproportionate effect of parasites on colonially breeding species has had no impact on senescence. Otherwise a greater rate of senescence would be expected in colonial species.
The implications of the present study are several. First, this study has shown a relationship between life history and evolution of senescence, since a slow rate of senescence was associated with a delayed onset of reproduction. Therefore I would predict that situations with intense competition for reproductive vacancies such as in co-operatively breeding birds and mammals would result in a similar retardation of senescence due to a delay in start of first reproduction. Likewise, species in tropical environments that are likely more often to be close to carrying capacity should have slower rates of senescence than closely related species in temperate climates. Second, if the evolutionary scenario for reduced senescence in species with late onset of first reproduction presented here turns out to be corroborated this would also have consequences for intraspecific variation in rates of senescence. For example, onset of senescence should become delayed among invading species during colonization when population size eventually reached carrying capacity.
I thank C. Spottiswoode for constructive discussions and W. Blanckenhorn and L. Keller for their comments on the manuscript.