- Top of page
The influence of early conditions may be particularly pronounced in early stages of growth (e.g. survival before fledging in birds; Nisbet et al. 1998), but may also span post-independence years. It has been shown that early conditions have fitness consequences lasting up to reproductive life in long-lived species (e.g. Gaillard et al. 1998). In long-lived species with deferred breeding, early conditions may influence fitness components before recruitment (i.e. survival in prebreeders), recruitment probability and fitness components after recruitment (i.e. survival and reproductive rates). In the vast majority of cases estimation of these demographic parameters using data from wild animal populations requires approaches to statistical inference incorporating explicitly the probability of capturing or resighting marked individuals that are alive and present in the study area (Clobert, Lebreton & Marzolin 1990; Lebreton et al. 1990; Nichols et al. 1990; Clobert et al. 1994; Pradel et al. 1997; Pradel & Lebreton 1999; Schwarz & Arnasson 2000; Spendelow et al. 2002; Williams, Nichols & Conroy 2002). Approaches dealing with prereproductive stages and recruitment probability were developed in the 1990s, which explains partly why the long-term consequences of early conditions on these fitness components have been rarely addressed in long-lived species. Most studies have been restricted to prefledging or first-year survival only (e.g. Nisbet et al. 1997, 1999; Hall, McConnell & Barker 2001), cumulative survival to recruitment (e.g. Harris et al. 1994) or reproductive parameters in the segment of the population that recruited (e.g. Lifetime Reproductive Success (LRS) studies; Clutton-Brock 1988; Newton 1989; Rose et al. 1998). In this paper we used multistate capture–recapture models (e.g. Nichols et al. 1994; Nichols & Kendall 1995) to address the consequences of early conditions on yearly survival before recruitment and recruitment probability in a long-lived seabird: the kittiwake (Rissa tridactyla).
A conceptual framework often used to address early conditions is the ‘cohort effect’, as part of the relationship between environmental stochasticity and population dynamics (e.g. Sæther 1997). For practical reasons, in this paper we did not focus on such collective effects. Our main interest was in survival before recruitment and recruitment probability, which were age-specific (Cam et al. 2002a). In many seabird species, young birds do not return to the breeding ground in the first years of life (e.g. Spendelow et al. 2002). Previous analyses have shown that there were very few resightings in the first age classes in the study population (Cam et al. 2002a). Assessing effects such as cohort effects in addition to age effects and the influence of early conditions required increased stratification of the data: too many parameters of the multistate model were not estimated (even after fixing the relevant parameters to zero). Consequently, here we focused on a subset of biological hypotheses relevant to early conditions characterized at the individual level.
Our first objective was to address the influence of the length of the rearing period on both local survival after independence during the prereproductive and reproductive stages of life, and on recruitment probability in the kittiwake. The length of the rearing period is viewed as a component of reproductive effort in the literature on the evolution of parental care and parent-offspring conflict (e.g. Trivers 1994; Godfray 1995). It may influence the condition of the young at independence and their fitness (e.g. Nisbet et al. 1998), and partly characterizes conditions during growth. In the kittiwake, after the first successful flight (where the chick is able to return to the nest, hereafter referred to as ‘fledging’) the young are still dependent on parents and regularly return to the nest. We used the total length of the rearing period as the best measure of total parental effort available to perform analyses using existing longitudinal data. We do not assume this is a good surrogate for the frequency of provisioning or the total amount of food delivered, but we assume there is a positive relationship between the total length of the rearing period and total parental effort. Parents involved in offspring rearing for longer periods of time have less time and energy to devote to their own maintenance, take more risks (e.g. Ydenberg 1989), and may incur greater costs.
Models of parent–offspring conflict assume a positive relationship between food provisioning and offspring fitness. However, the rearing period may be shorter than ‘optimal’ from the offspring perspective. Most models of parent–offspring conflict assume that the marginal gain in fitness of the parent declines with increased resources (e.g. Godfray 1995). The fitness interest of the parent is expected to decrease as the offspring grows and learns how to forage alone (but is still fed by the parent in the nest), and as the survival probability of the young increases. There may be a disagreement between parents and offspring over the time of termination of care (Clutton-Brock & Godfray 1991; Clutton-Brock 1991). Parents may stop responding to their chicks’ signals (begging) and terminate parental care even though they stay in colonies for several days or weeks after stopping offspring feeding. This relies on the paradigm of a trade-off between current and future fitness in parents, or the fitness of parents and the quality of offspring (Stearns 1992), and is relevant to theories about the evolution of parental investment (sensu Trivers 1994). Another non-exclusive hypothesis to explain rearing periods of shorter length than ‘optimal’ from the offspring viewpoint relies on the length of the breeding season and timing of breeding (Daan et al. 1990; Meijer, Daan & Hall 1990): the length of the rearing period may be limited in late breeders. Parents of migrating species may leave the breeding area before the young is fully independent and able to forage. However, here there is no clear relationship between timing and the length of the rearing period; nearly all the late breeders have short rearing periods, but such short periods are common during the entire breeding season (unpublished data). Note that the parent's perspective and the factors influencing the length of the rearing period are beyond the scope of this paper. Lastly, short rearing periods may also result from a decision of the young: they may decide to leave before full independence.
According to the assumption of a positive relationship between parental investment into food provisioning and offspring fitness, we might expect a positive relationship between the length of the rearing period and offspring survival. However, the relationship may not be linear. In this study, the length of the rearing period corresponds to the total period of dependency and includes the period of time the chicks are learning how to forage, a skill likely to be important for survival over the first winter at sea. It is possible that offspring relying on their parents for a longer period after being able to fly do not reach full independence, which may translate into reduced survival. This may reflect lower individual quality (incomplete behavioural development). Similarly, some chicks in poorer condition may fledge and continue to beg for food for longer periods because of poorer parental care and slower development, possibly because of lower-quality parents. Consequently, we considered models where survival varies monotically with the length of the rearing period, as well as quadratic models (i.e. where survival increases and then levels off, or decreases for longer periods). Lastly, our most general model included an interaction between age and the length of the rearing period; this should provide insight into whether the influence of initial conditions fades as individuals age. We also investigated the long-term influence of the length of the rearing period on reproductive performance in individuals that recruited (e.g. Gaillard et al. 1998).
Our second objective was to address the influence of the ‘rank’ of the chick (which depends on hatching order) on survival and recruitment probability. Elder siblings are dominant in competition with juniors for food, and sometimes eliminate them through aggression (e.g. Braun & Hunt. 1983). This source of mortality was not relevant here because we focused on survival after fledging. However, because of its possible influence on the condition of younger siblings, rank is a component of early conditions. Parents do not seem to attempt to balance the amount of food delivered to chicks. In addition, there is evidence that growth is slower in juniors in some seabird species (e.g. Nisbet et al. 1998; Pasquet & Monnat, unpublished data). This may lead to differences in condition among chicks at independence and differences in fitness (Sydeman & Emslie 1992). Offspring may be able to manipulate parental investment (Clutton-Brock 1991; Godfray 1995) via the intensity and frequency of begging, and deciding when to leave colonies. Except in extreme cases at the end of the breeding season, the length of the rearing period is determined by individual decisions and tactics of the parents and offspring. In contrast, the young cannot influence their own rank. Rank can be viewed as a random factor potentially influencing fitness. Lastly, as above, we addressed the long-term influence of rank on survival and reproductive performance in individuals that recruited.
- Top of page
Our results provided evidence of a long-term influence of early conditions on fitness. Both the length of the rearing period and rank influenced survival in prebreeders (after fledging). The influence of rank did not depend on age during that period: rank 1 chicks had higher survival than younger siblings. They also had a higher probability of recruiting at younger age. In contrast, there was an interaction between age and the length of the rearing period. In rank 1 individuals, longer rearing periods were associated with higher survival in 0-year-old individuals, but with a decrease in survival in older prebreeders. In rank 2 birds, the pattern was the same in 0- and 1-year-old individuals as in rank 1 birds, but the length of the rearing period had a slightly positive influence on local survival in older prebreeders. Due to differences in departure dates among individuals, we may have overestimated survival probability in individuals with shorter rearing periods. In spite of this, we detected an increase in survival with E in the youngest individuals; we may have underestimated the magnitude of the increase. In addition, the influence of the length of the rearing period was still perceptible after recruitment. This variable had a positive influence on reproductive performance, but the increase in success probability decelerated slightly at higher values of E. This suggests that early conditions (assessed using the length of the rearing period) contribute to shape individual quality. In contrast, we did not find evidence of an influence of rank on demographic parameters corresponding to the reproductive stage of life.
The sign of the influence of the length of the rearing period on survival before recruitment strongly depended on age. Such a phenomenon would remain undetected using cumulative survival up to recruitment. The influence of E was unambiguously positive in the first year, and negative in 1-year-old individuals. In addition, rank 1 individuals had higher survival than juniors. The interpretation of differences in survival among groups raises a classic difficulty: assessment of the contribution of natal dispersal to local survival (e.g. Nichols et al. 1992; Brownie et al. 1993; Spendelow et al. 1995; Joe & Pollock 2002; Spendelow et al. 2002). The range of the kittiwake covers the north of the Atlantic Ocean and we cannot assess long-distance dispersal in this species (e.g. Koenig, Van Duren & Hooge 1996; Boulinier et al. 1997). However, it is logical to hypothesize that individuals dominant in competition for food within broods, or individuals without siblings, or individuals receiving a larger amount of resources from their parents (e.g. with longer rearing periods), are in a better condition at independence and have lower mortality during the first winter at sea.
The presence of the interaction between ‘age’ and E in the survival part of models was consistent with our a priori hypothesis. We expected the effect of E to reduce with age. The influence of a variable can fade with age through two processes: a genuine decrease in the strength of the effect as individuals become older, or earlier death of the individuals in which the variable in question had the strongest negative effect (Vaupel & Yashin 1985a,b). Based on the hypothesis that lower survival reflects higher mortality to some extent, individuals with lower survival are expected to disappear from the sample at younger age (i.e. die). We expected within-cohort phenotypic mortality selection to operate (Curio 1983; Endler 1986; Forslund & Pärt 1995; Cam et al. 2002b). Several studies based on data from the same population have converged to the conclusion that such a selection process is operating in this population (Cam & Monnat 2000a,b; Cam et al. 2002b). The strong positive influence of E on survival during the first winter implies that, on average, individuals with shorter rearing periods disappeared earlier and did not reach older age. Our a priori hypothesis was that there would not be any long-term positive influence of longer rearing periods on survival. The results for adult survival are consistent with this. Furthermore, the positive influence of the number of years elapsed since first reproduction (T) on reproductive perpformance is consistent with previous results (Cam & Monnat 2000a,b). This may reflect a genuine combined influence of age and experience on breeding success, or result from within-generation phenotypic selection (Cam & Monnat 2000a,b; Cam et al. 2002b).
The dispersal component of local survival may partly explain the apparent inversion of the influence of E on rank 1 prebreeder survival in older age classes: long rearing periods are associated with a decrease in local survival in rank 1 prebreeders aged 1 or more. Individuals with long rearing periods may be in good condition and may have a higher probability of emigrating permanently to distant locations, even after visiting natal colonies as prospectors during the prebreeding stage. There may be a relationship between condition at independence and dispersal distance (Ims & Hjermann 2001; O’Riain & Braude 2001). The complex influence of E on local survival in this population may reflect a combination of processes whose influence may be stronger at different stages of life. Genuine mortality associated with poor condition at independence may be the predominant component of local survival during the first winter at sea, whereas permanent emigration may dominate after the initial selection process has operated (i.e. in older prebreeders).
We did not find evidence of an influence of E2 on survival in prebreeders or adults. A methodological issue may explain this. Very long rearing periods probably correspond to two situations: (1) individuals in poor condition because of lack of efficiency of parents or failure to reach behavioural independence or (2) fully independent chicks returning to the natal colony after a week or two at sea. The former are expected to survive poorly and the latter to have higher survival. In this situation, the quadratic term would be needed in the former category only, and further stratification of the data would be needed to account for such a phenomenon (Cohen 1986; Cooch, Cam & Link 2002). The relationship between the length of the dependency period and the length of the presence of the young in colonies needs to be addressed thoroughly. This will require definition of criteria allowing distinction between chicks in different situations. Daily observations in all the colonies were required to assess the length of the rearing period, but the daily probability of resighting an individual alive and present in the study area is unknown. This probability may decrease as chicks become older, become progressively independent and attend colonies less often or for shorter periods of time. It may not decrease in chicks failing to become independent and strongly relying on their parents for food.
We found evidence of an influence of rank on survival in prebreeders even though we classified chicks when we marked them and did not account for brood size during the entire rearing period. That is, some chicks classified as ‘of rank 1’ were younger siblings in broods with several siblings, one or several of which died before being marked. They were of rank ≥ 2 at birth and during a period we could not assess. We tried to limit disturbance in colonies and climbed down cliffs only when a reasonable number of chicks were old enough to be marked. In terms of access to resources, these chicks were probably dominated in competition with the other young of the brood. We probably underestimated the influence of rank on survival by assigning rank ‘1’ to some chicks that had been of rank ‘2’ at the beginning of the rearing period. However, it is also possible that the influence of rank (competition and access to resources) on survival is shaped relatively late in the rearing period (when older siblings are strong enough to monopolize resources). The initial disadvantage of rank 2 chicks may fade rapidly after the death of older siblings, and these chicks may have the same characteristics as genuine rank 1 chicks after fledging.
We classified correctly some chicks as rank 1 individuals but did not account for the size of the brood, and thus the number of siblings competing for food. This may influence the amount of resources acquired by dominant chicks. It is theoretically possible to refine the approach to classifying individuals and to define criteria that account more effectively for the degree of competition in broods and the length of the period of time chicks were competing. Accurate assessment of these conditions would permit evaluation of the hypothesis of a trade-off between number and quality of offspring (e.g. Smith et al. 1989; Stearns 1992). Some other species permitting access to the young at hatching and the use of a system to mark them before using bands should lend themselves to such a study. Despite a possible mismatch (to some extent) between the rank assigned to some individuals and their status in terms of dominance in competition for resources, we found that chicks of rank ≥ 2 had lower survival before recruitment. We also found that rank negatively influenced age of first breeding: the proportion of individuals of rank ≥ 2 breeding at that age 3, given that they survived to that age, was lower. It is possible that rank 1 individuals keep a dominant status in competition with others, which may increase the probability of acquiring a nesting site and recruit into the breeding segment of the population at younger age. Reduced age at maturity or the age at which life history transitions occur with increased growth conditions is a common pattern in many taxa (e.g. Day & Rowe 2002).
Recent studies in the areas of population dynamics (e.g. Johnson et al. 1986; Bjornstad & Hansen 1994; Conner & White 1999; White 2000), life history theory (e.g. Van-Noordwijk & de Jong 1986; Houston & McNamara 1992; McNamara & Houston 1992, 1996; Stearns 1992; Clark 1993; Marrow et al. 1996; Morris 1996; Clark & Mangel 2000) and demography (e.g. Vaupel & Yashin 1985a,b; Burnham & Rexstad 1993; Cam & Monnat 2000a,b; Cam et al. 2002a; Pledger & Schwarz 2002) have emphasized the need to account for ‘individual heterogeneity’. Quality is sometimes viewed as a permanent individual characteristic or a dynamic characteristic (e.g. Morris 1998), and is sometimes assumed to be determined genetically, or to be of epigenetic nature. The concept of quality as a permanent or a dynamic characteristic has implications in terms of modelling. Some state-dependent optimization models used to address life history evolution rely on the assumption that all the individuals have the same a priori probability of being in a given state after accounting for age, body size or condition (e.g. Clutton-Brock et al. 1996; Marrow et al. 1996). Some more realistic models consider groups of individuals (e.g. quality groups; McNamara & Houston 1992), and the dynamic component of models is group-specific (individuals of different quality have distinct a priori probability of being in a given state after accounting for age, size or body condition; McNamara & Houston 1992). Our results support the hypothesis that early conditions have long-term consequences on fitness: conditions during growth may contribute to shape a permanent component of individual quality. In addition, chicks may be able to manipulate parental investment (e.g. higher-quality chicks acquiring more resources and inciting parents to prolong care). That is, the length of the rearing period may depend on some intrinsic (perhaps determined genetically) characteristic existing at birth. However, rank is probably a factor that cannot be manipulated and influences individual characteristic at random, regardless of intrinsic quality at birth. It is likely to be an epigenetic determinant of individual quality.