Family legacies: short- and long-term fitness consequences of early-life conditions in female European rabbits


*Correspondence author. E-mail:


  • 1Environmental conditions during an animal's early life can have profound long-term consequences and affect its fitness. In particular, maternal and sibling effects, which can strongly influence the early growth of altricial mammals may be important. Few studies have investigated the influence of such early-life parameters in small mammals, because in these species the early post-natal stage is difficult to monitor under natural conditions.
  • 2We quantified the effects of litter size (i.e. number of litter siblings), maternal social rank and age and reproductive history of the mother (i.e. whether or not the mother had given birth to a previous litter during this season), and the individual date of birth and social rank on two fitness components of female European rabbits (Oryctolagus cuniculus L.) from a field enclosure population. Analyses were based on data on survival to maturity of 1836 female pups from 10 annual cohorts, and on lifetime reproductive success (LRS; here: the summed up number of offspring surviving to maturity) of 81 adult females from eight annual cohorts.
  • 3Both fitness components were correlated with the size of the females’ original litter and with the age of their mother. Litter size was related to survival to maturity and to LRS in a nonlinear (quadratic) way being highest in females from medium-sized litters. Maternal age also exerted quadratic effects on LRS, which peaked in females born to 2- to 3-year-old mothers. In contrast, survival to maturity increased with increasing age of the mother.
  • 4Survival to maturity and LRS were decreased in females born later in the breeding season, likely because of the longer time for growth that early born young enjoy before the winter season. In addition, LRS was lower in females which occupied a higher social rank at the onset of their first breeding season.
  • 5Our results emphasize that factors during early development, in particular parameters of the early social environment, do not only affect juvenile survival but have the potential to exert long-term fitness consequences throughout life.


The environmental conditions experienced by an animal during early-life stages can have an important influence on its fitness (Lindström 1999; Metcalfe & Monaghan 2001; Lummaa & Clutton-Brock 2002). Such early-life conditions often show short-term effects, for example by affecting early growth or survival (mammals: Loison & Langvatn 1998; Rödel et al. 2004b, 2009; Millar 2007; birds: Reed et al. 2003; Dawson, Lawrie & O’Brien 2005). Moreover, conditions during early development may show delayed or long-term consequences (‘silver spoon effect’: Grafen 1988) on various fitness components such as the age at maturity, life span, reproductive performance, senescence and breeding success (mammals: Albon, Clutton-Brock & Guinness 1987; Forchhammer et al. 2001; Nussey et al. 2007; Descamps et al. 2008a,b; birds: Cam, Monnat & Hines 2003; Van de Pol et al. 2006), or health during adulthood (humans: reviewed in Lummaa & Clutton-Brock 2002). Most of these studies, particularly the ones on mammals, consider population-level factors such as food availability (Descamps et al. 2008), density and weather conditions (Albon et al. 1987; Forchhammer et al. 2001). Such factors usually influence all young born during a particular span of time in a similar way. This may lead to cohort effects, such as similarities in the performance in all recruits of a particular year (Rose, Clutton-Brock & Guinness 1999). However, studies considering potentially fitness-relevant parameters on the individual level may add valuable information for understanding the strong between-individual variation in fitness, which can be observed in many vertebrate populations (Jones et al. 2005).

For example, parental effects, i.e. nongenetic effects of the mother/parents on the offspring's phenotype have the potential to induce such individual fitness variation. Particularly in animals with altricial young, parental effects might be expected to be pronounced, because the early development of the young strongly depends on the provisioning with food by the mother (in most mammals) or by both parents (Mousseau & Fox 1998). Furthermore, the presence of siblings represents an important component of the early developmental environment in most mammal and bird species (Mock & Parker 1997; Hudson & Trillmich 2008). However, studies on fitness consequences of such individual-specific factors are rare, in particular in altricial mammals which are difficult to monitor during early post-natal life (cf. Millar 2007).

Here we investigated the fitness consequences of early-life variables in female European rabbits (Oryctolagus cuniculus L.) from a population living in a field enclosure. The European rabbit provides a useful model for such a study: maternal characteristics are well known to affect offspring growth and survival. For example, it has been shown that the offspring of mothers with a higher social rank show a higher probability of nest survival (von Holst et al. 2002; Rödel et al. 2009). Furthermore, pre-weaning growth is maximal in offspring of middle-aged mothers of 2 to 3 years, and is lower in pups born to mothers, which gave birth to a previous litter during the respective breeding season, most probably due to short-term effects of reproductive costs (Rödel, Hudson & von Holst 2008a). In comparison to many other mammals, maternal care in the European rabbit is very low and almost exclusively restricted to the first 3 to 4 weeks of life. The mother leaves the pups immediately after giving birth in a subterranean breeding burrow, and only returns for a few minutes once a day to nurse them (Hudson, Bilkó & Altbäcker 1996). The comparatively low maternal investment excluding any brooding of the young emphasizes the importance of litter siblings. Although rabbit litter mates scramble for the maternal milk supply leading to lower growth in pups from larger litters (Drummond et al. 2000; Bautista et al. 2005; Rödel et al. 2008b), the presence of siblings also provides mutual benefits in reducing the costs of thermoregulation, in particular during the first post-natal days (Bautista et al. 2003; Gilbert et al. 2007). The balance between both effects has been shown to lead to a medium litter size being optimal with respect to post-natal growth and nest survival (Rödel et al. 2008a, 2009). Furthermore, animals from litters born later in the season usually have a lower survival probability during their first winter season (Rödel et al. 2004b).

The objective of our study was to identify short-term and long-term fitness consequences of different early-life conditions in female European rabbits. We considered litter size (i.e. the number of litter mates), maternal age and social rank, the reproductive history of the mother (i.e. whether or not she gave birth to a previous litter during this breeding season), and the date of birth. Based on our previous studies on juvenile growth and survival (von Holst et al. 2002; Rödel et al. 2004b; 2008b,d), we predicted these factors to exert clear short-term effects on survival from birth to maturity. In particular, we expected nonlinear (quadratic) effects of litter size and maternal age, linear or nonlinear effects of the females’ date of birth and linear effects of their mother's social rank. Linear effects might consist of higher survival probabilities in animals born earlier in the season and in animals born to higher-ranking mothers. Nonlinear effects might become apparent as survival optima in offspring from middle-aged mothers due to the comparatively low performance of young mothers and senescent effects in older females (Rödel et al. 2004a), and in offspring from medium-sized litters due to the balance between the costs of sibling competition and the thermal benefits of sibling presence in the nest (Gilbert et al. 2007; Rödel et al. 2008a). Furthermore, the date of birth might also be related to survival in a nonlinear way, since offspring which are born very early in the season (i.e. in late winter/early spring) might still suffer from adverse weather conditions and low food quality during early youth. We also expected higher survival probabilities of young born to females which had not given birth to a previous litter during the season, i.e. to mothers who did not have to carry the costs of a previous reproductive event (Koivula et al. 2003; Rödel et al. 2008b). Our main goal, however, was to investigate whether the influence of these early-life parameters is still evident during later life, i.e. in animals which managed to survive until maturity. Therefore, we tested and compared the effects of the different factors on female lifetime reproductive success, measured as their summed-up number of offspring which managed to survive until maturity (recruits).

Material and methods

study population

The study was conducted on animals from a fenced population of European rabbits living in a 20 000-m2 enclosure situated in Upper Franconia, Germany (49·55N, 11·36E, elevation 359 m a.s.l.). Vegetation consisted of grassland interspersed with groups of trees and bushes. In addition to the burrows and breeding stops dug by the rabbits (around 40 to 50), the area contained 16 artificial concrete warrens with interconnected chambers and removable tops. The whole study site could be observed from two towers and all animals of the population could be identified by individual ear tags.

The population consisted of descendants of animals that had been caught in the wild (Upper Palatinate, Germany) in 1983. During the study period (1990 to 2002), the number of animals at the start of the breeding season ranged from 27 to 94 (on average 61), and the sex ratio (males/females) varied between 0·42 and 1·05 (on average 0·70). For further details on this population, see von Holst et al. (2002) and Rödel et al. (2008a).

All animals were marked individually by fitting the young on post-natal day 12 with a numbered plastic tag (Dalton Rototag, 20 × 5 × 1 mm, 0·25 g; Dalton Continental GmbH, Bocholt, Germany) in one ear. When the animals reached a body mass of about 1000 g (usually in autumn), we removed the small plastic tag and replaced it with a coloured aluminium tag (450 × 200 mm, 1·1 g) fixed with a plastic tag (Dalton Rototag, 35 × 10 × 2 mm, 1·5 g).

population counts and collection of survival data

We followed the changes in population size by records during regular behavioural observations (three to four times per week during the breeding season; e.g. von Holst et al. 2002; Rödel et al. 2008c) and by records of animals found dead during the daily check walks. Based on these data, we determined the population size and presence/absence of each individual in mid-March, i.e. shortly before or around the onset of the annual breeding season (Rödel et al. 2005).

identification of mothers and collection of reproductive data

During the breeding season, we trapped the rabbits once a month using peanut-baited wooden traps. Trapping success was around 70% of the adult population. The traps were set overnight and checked at dawn the next morning. We dyed the abdominal fur of the adult females with different colours (silk colour, Marabu, Germany), and shortly afterwards we returned the animals to the enclosure. We did not find any indication that this procedure affected nest mortality or female breeding success.

Every morning we checked for newborn litters. To this end, we prepared all natural warrens and breeding stops dug by the animals with artificial vertical openings to the nest chambers, which we covered with concrete flagstones. By checking the nests daily, we could record the birth of all litters within 24 h. We recorded litter size, and 12 days later (about 1 week before the young rabbits leave their breeding burrow) we determined the sex and marked the pups individually (see above for details). For analysis, we used litter size on day 1. Litter size on day 1 and at weaning was usually identical, because mortality rarely occurred in single pups but typically affected all pups of a litter (Rödel et al. 2009). Individual body mass on post-natal day 1 was not routinely registered, because of the difficulties of marking newborn pups individually.

As female rabbits pluck out abdominal hair to build their nests (Denenberg et al. 1963), we were able to determine the mother of each litter by the location of the nest in combination with the colour of the hair found in it. Identity of the mothers was additionally confirmed by the analysis of female reproductive status during the regular trapping sessions (detection of pregnancies by abdominal palpation) and by behavioural observations (females entering particular breeding burrows; copulation as a sign of post-partum oestrus; nest defence against other females: Rödel et al. 2008d). Based on these data, we were able to document the reproductive history of each female during the breeding season, i.e. whether or not the mother had given birth to a previous litter during this season.

determination of female social rank

Each year, we determined the social rank of all females by regular direct observations throughout the breeding season. Each female was observed 8 to 10 times per month for 30 min a day during the last 3 to 4 h before twilight, when rabbits usually show the peak of their daily activity (Wallage-Drees 1989). The observation units were distributed equally over the whole breeding season, resulting in an average observation time of about 4 to 5 h per female and month. We determined the social rank of each adult female by the occurrence and direction of intra-group aggression between females (details in Holst et al. 2002). European rabbits are organized in social groups with linear intrasexual rank hierarchies and have a polygynous mating system. The number of adult individuals per group in our study varied from one to three adult males and from one to six adult females (see Cowan 1987a; von Holst et al. 1999). Social rank = 1 represents the dominant position in the group.

study design and data analysis

In a first step, we determined for each female pup (n = 1836) whether or not it survived to maturity, i.e. until mid-March of the following year shortly before or around the young females started first reproduction.

From 1991 to 1998, n = 81 females reached maturity. In autumn 2002, the population crashed due to myxomatosis. Therefore, the females of our data set born in 1998 could potentially reproduce during three breeding seasons, which exceeds the average reproductive life span of a female rabbit of about two breeding seasons (von Holst et al. 2002). For each of these 81 females, we calculated the lifetime reproductive success (LRS), defined as the sum of offspring (males and females) reaching maturity, i.e. which survived until mid-March of the year following their year of birth (see distribution in Fig. 1).

Figure 1.

Model graphs (solid lines for covariates, grey bars for fixed factors) for the partial effects of the significant predictor variables (Table 1) on survival to maturity, given as probabilities between 0 and 1. Data points represent the averages per group. Sample sizes for each group are given. For calculating the predicted values for each explanatory variable, the respective variable was varied within the observed range of the data while the others were set constant at their means.

The early-life parameters considered to potentially affect survival to maturity and LRS were the female's original litter size (i.e. the number of litter siblings of both sexes), age, social rank and reproductive history of the mother (i.e. whether or not the mother gave birth to a previous litter during this breeding season) and date of birth. We used the social rank that the mother had when she gave birth to the respective litter. All of these explanatory variables were used as covariates, except the mother's reproductive history which was used as a fixed factor with two levels. In order to account for the differences in group sizes among years which results in an imbalance in the occurrence of females with very low social ranks (i.e. mothers with rank 6), we categorized the maternal social rank as 1, 2, 3, 4 and > 4.

We did not correct the social rank for the number of female rabbits per group, because it is not the relative social rank per se that causes stress in a low-ranking female, but the number of female group member dominating it (von Holst 1998). We tested for linear effects of all covariates. Based on our predictions, we also tested for quadratic effects of litter size, maternal age and the date of birth in order to detect possible optima in survival or LRS. Maternal age was log-transformed when tested for quadratic effects, because we explicitly expected a steep, left-skewed increase (either in juvenile survival or LRS) due to the pronounced difference in body condition and reproductive performance between 1-year-old and older females (Rödel et al. 2004a). The date of birth (covariate) was measured in 1-week intervals with respect to the onset of the annual breeding season, i.e. the occurrence of the first litter per year. This was carried out in order to correct for potential inaccuracies in the determination of birth date during our check walks.

For the analysis of LRS, we also included the individual social rank which the female obtained at the beginning of its first breeding season, because previous studies have shown effects of this parameter on female reproductive performance and longevity in European rabbits (Mykytowycz & Fullagar 1973; von Holst et al. 2002). We entered the individual social rank as a covariate, since we expected a linear effect on female performance (see von Holst et al. 2002). However, we pooled the data of females reaching rank 1 (= dominant) and 2 and of females with rank 5 or 6 due to the low number of cases in these rank positions. Furthermore, we considered the year of birth as a fixed factor in order to test for cohort effects. We also included the identity of the female's mother as a random factor (for modelling LRS and survival to maturity) because several of the tested females descended from the same mothers. To avoid overfitting and the associated problem of detecting spurious effects, we did not consider interaction between the predictors due to the moderate sample size (Burnham & Anderson 2002).

The effects of these predictor variables on the different fitness components were analysed using generalized linear mixed models (GLMM, Pinheiro & Bates 2000), carried out with the program r, version 2·7·2 (R Development Core Team 2008). Mixed-effects models were fitted with the package lme4 with the Laplace approximation of the likelihood function (Bates 2005). We used GLMMs for binomial error structure with logit-link function for modelling the data on survival to maturity. Note that the survival of all individual animals until maturity was known with certainty (re-sighting rates = 1) and was not based on probability estimates. LRS, which represents count data with a right-skewed distribution was analysed using Quasi-Poisson models with a log-link function. We used quasi-likelihood estimators, because these are useful for data with overdispersion (Faraway 2006). A direct test for overdispersion cannot be conducted for GLMMs in r because the exact calculation of the degrees of freedom is not implemented yet. However, the calculation of a Poisson GLM (i.e. without the random factor) indicated overdispersion due to the high ratio of deviance and degrees of freedom (inline image = 3·01). Colinearities between different covariates occurred, but the correlations were rather weak (Spearman rank: litter size vs. date of birth: rs = 0·381, P < 0·001; maternal age vs. date of birth: rs = –0·224, P = 0·045; maternal age vs. individual social rank: rs = –0·294, P = 0·008). All other correlations between covariates were not significant (maternal age vs. litter size: rs = –0·107, P = 0·341; maternal social rank vs. litter size: rs = –0·019, P = 0·864; litter size vs. individual social rank: rs = 0·135, P = 0·228; maternal social rank vs. date of birth: rs = 0·119, P = 0·291; date of birth vs. individual social rank: rs = 0·199, P = 0·074; maternal age vs. individual social rank: rs = –0·112, P = 0·318; maternal social rank vs. individual social rank: rs = –0·102, P = 0·365).

P values were calculated using likelihood-ratio tests based on changes in deviance when each term was dropped from the full model. The respective full model (for survival to maturity and LRS) included the additive combination of all explanatory variables. We tested for quadratic effects by calculating P values for the changes in deviance when the full quadratic term (xi + inline image) was removed (given in Table 1). In addition, we directly compared the support for linear and quadratic effects of the respective predictor by dropping the quadratic term from the full model (given in the text).

Table 1.  Effects of different predictor variables on fitness components of female European rabbits. Survival to maturity of females pups (n = 1836, from nine different years) was modelled by a binomial GLMM with logit-link function; lifetime reproductive success (LRS) of adult females (n = 81, females born during 8 years) was modelled by a Quasi-Poisson GLMM with log-link. Mother ID was always included as a random factor. We tested for quadratic effects of date of birth, litter size and maternal age; maternal age was log-transformed when included as quadratic effect. P values were calculated by likelihood-ratio tests based on changes in deviance when each term was dropped from the full model. Significant P values are given in bold
TraitSource of variationχ2d.f.Estimates (SE)P
  • *

    fixed factor.

(a) Survival to maturityDate of birth19·391–0·274 (0·059)< 0·001
Date of birth + date of birth220·932–0·011 (0·296)< 0·001
–0·033 (0·030)
Litter size 7·551–0·202 (0·084)0·006
Litter size + litter size210·6520·507 (0·440)0·005
0·066 (0·042)
Maternal age 4·6010·198 (0·095)0·032
Maternal age + maternal age2 5·2121·202 (0·961)0·074
–0·360 (0·561)
Maternal social rank 0·1510·089 (0·159)0·698
Maternal reproductive history* 0·0310·246 (0·382)0·596
Year*78·618 < 0·001
(b) LRSDate of birth 7·641–0·118 (0·047)0·006
Date of birth + date of birth2 7·642–0·206 (0·280)0·022
0·001 (0·017)
Litter size 2·2410·128 (0·102)0·135
Litter size + litter size214·3322·131 (0·559)< 0·001
–0·191 (0·054)
Individual social rank 4·751–0·314 (0·166)0·023
Maternal age 4·201–0·223 (0·153)0·040
Maternal age + maternal age2 8·0422·829 (0·989)0·018
–1·565 (0·523)
Maternal social rank 1·061–0·084 (0·193)0·304
Maternal reproductive history* 1·3410·728 (0·540)0·247
Year*12·907 0·075

Effect sizes of the different significant predictor variables were calculated as the range in the predicted values given by the GLMM. For this, the effects of the respective other significant predictor variables were set constant at their means. We calculated a likelihood-ratio Pseudo-R2 (Kramer 2005) in order to assess the variance explained of the model including all statistically significant predictor variables.


survival to maturity

The survival of rabbit females to maturity was significantly related to litter size, maternal age and date of birth, and showed significant variation among years. Maternal social rank or reproductive history, i.e. whether or not the mother gave birth to a previous litter during this season, did not show any significant effects (Table 1a). Survival to maturity increased with the mother's age (Fig. 1b), and was negatively correlated with the date of birth (Fig. 1a). Pups born earlier in the breeding season had a higher chance to survive, whereas none of the pups born later than 16 weeks after the onset of the annual breeding season ever reached maturity during the study period (Fig. 1c).

We found significant linear as well as nonlinear (quadratic) effects of the date of birth; however, a direct comparison of both models by a likelihood-ratio test revealed that the inclusion of the quadratic slopes did not add any significant information (χ2 = 1·54, d.f. = 1, P = 0·214). Survival was also related to litter size, where the inclusion of the quadratic term significantly increased the fit compared to the model only including the linear slope (χ2 = 8·90, d.f. = 1, P = 0·022). This nonlinear effect predicted a survival optimum in animals from litters of three to four pups (Fig. 1a).

The additive combination of all significant predictors (excluding the quadratic effects of litter size) explained R2 = 23·5% of the variance in survival probabilities. A comparison of the effect sizes of the different significant explanatory variables, calculated as the range in the predicted survival probabilities (inline image), revealed that the effect of the factor ‘year’ was largest (inline image = 0·185), although the model seemed to overestimate the maximum survival in 1996 (Fig. 2a4). The effects of date of birth (inline image = 0·087), maternal age (inline image = 0·068), and of litter size (inline image = 0·049) did not differ markedly.

Figure 2.

Distribution of LRS in female European rabbits which survived until maturity (n = 81).

lifetime reproductive success

Overall, LRS, measured as the number of offspring (both sexes) surviving to maturity, showed a strongly right-skewed distribution; about 52% of the adult females of our study population did not even produce one recruit (Fig. 2). Nevertheless, most (96·3%) of the females were reproductively active. On average, they gave birth to 6·7 litters over lifetime (range: 0 to 23) making up a total of 31·1 pups (range: 0 to 110). The average number of litters per breeding season born to a female was 2·5 (range: 0 to 6) and the average number of pups per season was 11·8 (range: 0 to 25). Adult female lived on average for 2·6 years (oldest female: 7·8 years); 40·1% died during or short after their first breeding season and 16·0% managed to survive longer than 5 years.

LRS in adult females was related to litter size, the date of birth, maternal age, and to the individual social rank which the females occupied at the onset of their first breeding season. The effects of maternal social rank or reproductive history and variation among years were not significant (Table 1b).

LRS was negatively correlated with the date of birth and the individual social rank, i.e. high-ranking females and females born earlier in the season produced a larger number of recruits (Fig. 3c,d). As in survival to maturity, the date of birth showed a significant linear as well as a quadratic effect; however, a direct comparison between both models revealed that the inclusion of the quadratic term did not significantly increase the fit (χ2 < 0·01, d.f. = 1, P = 0·994). In contrast, the inclusion of the quadratic slope significantly increased the effects of maternal age on LRS (χ2 = 6·17, d.f. = 1, P = 0·013). Litter size showed quadratic but no significant linear effects on LRS (Table 1b). The shape of the nonlinear model graphs predicted a comparatively low LRS in females born in litters of two pups with a maximum in females from medium-sized litters (Fig. 3a). Furthermore, LRS was low in females born to 1-year-old mothers, was maximal in females born to middle-aged mothers and decreased in females born to older mothers (Fig. 3b).

Figure 3.

Model graphs for the partial effects of the significant predictor variables (Table 1) on lifetime reproductive success. Data points represent the averages per group; the size of the circles codes for the (log-transformed) frequency of cases per group. Sample sizes for each group are given. For calculating the predicted values for each explanatory variable, the respective variable was varied within the observed range of the data while the others were set constant at their means.

The explained variance of the model including all significant predictor variables was R2 = 41·1%. The date of birth (linear effects) showed the largest effect size (inline image = 3·6 offspring), followed by litter size (inline image = 2·4 offspring), the individual social rank (inline image = 2·1 offspring) and the mother's age (inline image = 1·9 offspring).


Our study detected short-term as well as striking long-term consequences of early-life conditions on fitness components in female rabbits. ‘Family effects’ such as the effects of litter size and maternal age influenced female survival to maturity as well as LRS in adult females. In addition, both fitness components were lower in females born late in the season. We also found a negative correlation between the social rank a female reached at the onset of the first breeding season and LRS, confirming the results of an earlier study (von Holst et al. 2002). However, we did not find any evidence for long-term effects of the mother's social rank or reproductive history on female fitness.

survival to maturity

As expected, several early-life conditions, which have been previously identified to affect growth and survival of juvenile European rabbits during different life stages, contributed to shaping survival to maturity. Variation among years showed the strongest effect compared to the other predictor variables, indicating cohort effects on juvenile survival. We suggest that particularly the variation in winter temperature conditions, which has been shown to strongly limit the survival of subadults (Rödel et al. 2004b), might account for this strong year-to-year variation. The effect sizes of all early-life parameters considered which were found to be significant were of similar magnitude. Litter size-dependent survival was best described by an optimum function predicting survival to peak in females born into litters of three to four pups. This is consistent with previous findings on nest survival, where survival probability was highest in medium-sized litters (Rödel et al. 2009), most likely because the costs of thermoregulation decrease (Bautista et al. 2003; Gilbert et al. 2007) but scramble competition for milk increases with litter size (Bautista et al. 2005). The lower survival in offspring from 1-year-old mothers might be explained by maternal effects on nest survival (though nest survival was rather related to the mother's social rank than to her age; von Holst et al. 2002; Rödel et al. 2009), but might also be due to delayed (post-weaning) survival effects of the comparatively lower pre-weaning growth in females born to 1-year-old mothers (Rödel et al. 2008b). Such delayed effects on post-weaning survival might also apply to females born in larger litters, being usually smaller around weaning than pups with a lower number of litter mates (domestic rabbits: Drummond et al. 2000; wild rabbits: Rödel et al. 2008b). Furthermore, an early date of birth has been reported to positively influence juvenile rabbits’ probability of winter survival, likely because of the longer time for growth that early born young enjoy before the winter season (European rabbits, Cowan 1987b; Rödel et al. 2004b; see similar findings in Kraus et al. 2005a for dark-backed cavies Cavia magna Ximinez and in Feder et al. 2008 for bighorn sheep, Ovis canadensis Shaw). Surprisingly, we did not find any evidence for decreased survival probabilities in females born to older mothers, as one might expect due to the notably lower pre-weaning growth in offspring from mothers older than 3 years (see Rödel et al. 2008b). This suggests that other factors than pre-weaning growth were involved in mediating maternal effects on offspring survival. For example, differences in the transfer of components of the immune system from mother to young, either prenatally via the placenta or post-natally via the milk, can influence the development of the offspring immune system (Zhou et al. 2000), and thus may have the potential to affect juvenile survival probabilities.

lifetime reproductive success

Interestingly, the same combination of early-life parameters affecting juvenile survival also influenced LRS in the 5% of the females which managed to reach maturity. Similar results were reported from a long-term study on oystercatchers (Haematopus ostralegus L.), where individuals born and raised in high-quality habitats had a higher juvenile survival probability and also a higher lifetime reproductive success (Van de Pol et al. 2006). It is commonly thought that an important pathway for such long-term consequences of an animal's early developmental environment acts through the effects on its early growth: a low body mass at birth or reductions in post-natal growth may lead to a lower age-specific size or body condition, which might in turn affect survival and reproductive performance (Festa-Bianchet, Jorgenson & Réale 2000; Lummaa & Clutton-Brock 2002; Jones et al. 2005). Such a relationship between maternal body mass and reproductive performance has also been described for European rabbits (Rödel et al. 2005).

Unfortunately, complete long-term data on individual birth mass or weaning mass were not available for our study population. However, maternal age and litter size have been shown to strongly affect the post-natal growth in the European rabbit (Drummond et al. 2000; Rödel et al. 2008b). One-year-old females are regularly not fully grown when starting to reproduce (Rödel et al. 2004a), whereas old females often show signs of senescence such as a decrease in reproductive function (Kirkwood & Austad 2000; Monaghan et al. 2008), leading to a maximum in body mass or growth of dependent offspring of middle-aged mothers (reindeer Rangifer tarandus L., Weladji et al. 2002; European rabbit, Rödel et al. 2008b). Although we only found a decreased juvenile survival probability in offspring from young mothers, the offspring of both age classes showed a comparatively lower LRS than offspring from middle-aged mothers. Thus, our study suggests that 1-year-old females as well as older mothers produce offspring of lower quality, and provides evidence for intergenerational fitness effects of the mother's age in the European rabbit (see Descamps et al. 2008a for a review on similar studies in other mammal species).

Nonlinear (quadratic) effects of litter size, as they occurred with respect to LRS, have also been shown to shape the post-natal growth of European rabbit pups in a similar way (Rödel et al. 2008a). This may point to a causal relation between LRS and early growth or some related traits of the juveniles’ early physiological development. However, effects of litter size might already occur during prenatal life; pups from larger litters are frequently born with a lower body mass (deer mouse Peromyscus maniculatus Wagner, Myers & Master 1983; red squirrel Sciurus vulgaris L., Wauters, Bijnens & Dhondt 1993; dark-backed cavy, Kraus, Trillmich & Künkele 2005b; domestic and wild European rabbit, Breuer & Claussen 1977; von Holst et al. 2002). In this context, it has been proposed that disturbances during the prenatal development might exert direct influence on an animal's later life, independently of its post-natal environment (‘foetal programming’, Lucas 1998; Lummaa & Clutton-Brock 2002; Foxcroft et al. 2006).

Studies in European rabbits have shown that young females, which integrated together with their sister into the same social group started to breed earlier than same-age females without sisters in their group, most likely because social bonds among sisters may have helped to buffer negative consequences of stress (Rödel et al. 2008c). Such sibling effects could provide a further explanation for the comparatively low LRS in females from litters of two pups, because (i) a larger number of litter mates might increase the chance that a female settles in a social group together with a sister, and (ii) an early onset of breeding or age at maturity, as it is reported for such females, can essentially increase an animal's breeding success (Cole 1954; Oli, Hepp & Kennamer 2002).

Relative to the other variables considered, the date of birth appeared to be the most important determinant of LRS. The main mechanism underlying long-term effects of the date of birth on female fitness may be the inability of late-born females to compensate for their lower body mass or body condition before their first winter season (Rödel et al. 2004b), leading to persistent downstream effects on female performance or longevity. An additional, non-exclusive explanation rests on the differential social environment which early- and late-born females might experience during their youth. Juveniles born late in the season usually grow up at much higher population densities than animals born in spring. Such effects of crowding during early life might negatively affect an animal's development by increased levels of social stress (Christian 1980; von Holst 1998), and may also increase the risk of infection with diseases. In turn, juveniles born early in the season, which usually dominate later born young in aggressive encounters (unpublished data), might have a higher chance of attaining a high social position after reaching maturity by already establishing dominance structures during juvenile life. Finally, differences in the performance of females born early or late in the season may be due to seasonal variation in the intrauterine environment. Rabbit mothers show high glucocorticoid levels during the early breeding season, when intrasexual competition for breeding resources is high (von Holst et al. 1999). Consequently, early born young might be exposed to comparatively higher levels of maternal stress hormones during their foetal development than later-born offspring, and studies in laboratory rats using psychological stressors reported that offspring of stressed mothers were less anxious and more explorative (Lordi et al. 2000; Götz & Stefanski 2007). Such a behavioural style may improve the ability to successfully compete for a dominant social position during adulthood, which in turn might positively affect several components of their reproductive fitness.

In summary, our study on European rabbits provides a striking example, where lifetime reproductive success is profoundly influenced (with an explained variance of about 41%) by factors of the individual's early developmental environment. Most of the described effects here might be due to chance, such as the mother's age, timing of parturitions (which determines the pups’ date of birth) and litter sizes. All of these traits show considerable intra-individual variation among different years of the mother's life (i.e. the pattern of parturitions over the season) and also within the season (i.e. litter size; see Rödel et al. 2004a, 2005). Although many studies have described short-term effects of such early-life conditions on growth and survival, their long-term fitness consequences are scarcely studied and might be still largely underestimated (cf. Van de Pol et al. 2006). Such long-term or intergenerational effects on the individual level might even contribute to the dynamics of a population (Sutherland 1996; Beckerman et al. 2002). For example, delayed effects on population growth might emerge when the average age of the female population is high and consequently most recruits descend from old, senescent mothers (Boonstra 1994).


We thank all our colleagues of the Department of Animal Physiology who helped with the field work. We are grateful to Raquel Monclús, Robyn Hudson and to Carsten Schradin for their helpful comments on our manuscript. Permission for population biology studies on European rabbits was granted by the government of Middle Franconia (211-3894a).