Costs of growing up as a subordinate sibling are passed to the next generation in blue-footed boobies


Correspondence: Hugh Drummond, Instituto de Ecología, Universidad Nacional Autónoma de México, AP 70-275, 04510 D.F., Mexico. Tel.: +(5255) 562 29007; fax: +(5255) 561 61976; e-mail:


As stresses in early development may generate costs in adult life, sibling competition and conflict in infancy are expected to diminish the reproductive value of surviving low-status members of broods and litters. We analysed delayed costs to blue-footed booby fledglings, Sula nebouxii, of junior status in the brood, which involves aggressive subordination, food deprivation and elevated corticosterone, but little or no deficit in size at fledging. In ten cohorts observed for up to 16 years, juniors showed no deficit in breeding success at any age, independent of lifespan, including in a sample of sibling pairs. Among females, juniors actually outreproduced seniors across the 16-year span. However, offspring produced by juniors in the first 3 years of life were less likely to recruit into the breeding population than offspring of seniors. Since junior fledglings survive, recruit and compete as well as seniors (shown earlier), and breed as successfully as seniors across the lifespan, it appears the delayed cost of subordination is passed to offspring, and only to those few offspring produced in the first 3 years of life. These correlational results indicate that systematic competition-related differences in developmental conditions of infant siblings can alter their reproductive value by affecting the viability of their eventual offspring.


Experiments have shown that when birds, fish and mammals experience deficient or stressful environments before birth or during infancy (‘poor starts’), they show physiological, morphological or behavioral deficits at some stage of their adult life (Lindström, 1999). Even when food-deprived infants have the opportunity to catch up and attain normal size, profound and persistent deficits can appear later on, and deficits can occur on diverse timescales and persist into old age (Metcalfe & Monaghan, 2001). Most studies have been of captive members of short-lived species and impacts have mostly been measured only in early adulthood. However, descriptive field studies, often of long-lived animals, suggest that poor starts can result in diminished reproductive success of wild animals over their lifetimes (Albon et al., 1987; Reid et al., 2003; van de Pol et al., 2006; Wilkin & Sheldon, 2009). But we currently lack a general appreciation, not only of causal mechanisms (Monaghan, 2008), but of the natural contexts and types of species in which poor starts prejudice development (Wilkin & Sheldon, 2009).

In species with parental care, sibling competition during infancy often involves aggressive subordination, selective resource deprivation and other stresses (reviews in Mock & Parker, 1997; Drummond, 2006), and low-ranking members of broods and litters are therefore expected to show phenotypic deficits during adulthood, coupled with diminished reproductive success. Body size/mass of animals at independence is sometimes positively related to survival and reproduction (e.g. Albon et al., 1987; Krementz et al., 1989; Magrath, 1991), so when sibling competition results in low body mass at independence, undersized individuals are likely to survive or reproduce less well afterwards (Thomas et al., 1999). However, even in the absence of such body size effects, a low-ranking infant's breeding success over its lifespan seems likely to be prejudiced by delayed and long-term effects of early stresses. Although sibling competition and conflict are widespread in infant birds and mammals, and numerous authors have suggested that dominance hierarchies in broods and litters could affect competitive abilities, status or breeding success in later life (e.g. Geist, 1978; Bekoff, 1981; Spear & Nur, 1994; Thomas et al., 1999), possible impacts on reproduction across the lifespan have not, to our knowledge, been successfully measured in any species (for an inconclusive attempt, see Cam et al., 2003). We do not know whether the privations and stresses of low-ranking siblings fall within their zone of tolerance (sensu Monaghan, 2008) or exact a fitness penalty. Competition-related fitness penalties are of considerable theoretical importance not only because they imply systematic variation in the reproductive value of offspring but also because they affect the inclusive fitness interests of all family members and potentially influence the evolution of parent–offspring conflict, sibling conflict and parental life history allocations.

In birds, experimental manipulations of food abundance or quality, and brood size, have shown that poor diet and growth of nestlings, or subsequent compensatory growth, can induce a variety of phenotypic deficits in the juvenile or adult, prejudice breeding success and even produce deleterious effects in the next generation. Deficits have been induced in body size and mass (Richner et al., 1989; Ohlsson & Smith, 2001), plumage (Birkhead et al., 1999), sexual ornament development (Gustafsson et al., 1995; de Kogel & Prijs, 1996; Ohlsson et al., 2002) metabolic rate (Verhulst et al., 2006), capacity to assimilate dietary antioxidants (Blount et al., 2003), social dominance (Richner et al., 1989) and cognitive performance (Fisher et al., 2006). Similarly, experimental elevation of circulating corticosterone, the stress hormone, in nestling birds induced diverse deficits in adulthood, including a smaller higher vocal centre in the brain (Buchanan et al., 2004), diminished problem-solving ability (Kitaysky et al., 2003) and reduced exploratory behaviour and competitiveness (Spencer & Verhulst, 2007). Impacts of poor starts on breeding early in life and intergenerational effects were demonstrated by experiments on captive domestic zebra finches (Taeniopygia guttata). Females raised by parents on a poor quality diet laid their clutches later and at a slower rate (Blount et al., 2006), and females that developed in enlarged broods showed reduced nestling growth followed by delayed laying and reduced fledging success (Alonso-Alvarez et al., 2006). After females were given a low-protein diet during brood care, their daughters laid smaller clutches; and after females were maintained on a low-protein diet prior to laying, their sons showed reduced fledging success in the first year of life and their daughters during the first 2 years of life (Gorman & Nager, 2004). Intergenerational effects were induced in captivity by enlarging broods: daughters of females that developed in such broods were undersized as nestlings and as adults; they also showed depressed hatching success and produced fewer fledglings (Naguib & Gil, 2005; Naguib et al., 2006). This developmental vulnerability of captive birds raises the expectation that in nature stressful sibling competition in the nestling phase could, through a variety of causal pathways, result in production of fewer or less viable fledglings at some stage of adult life.

To test this, we analysed the impact of a three-pronged poor start arising from severe sibling competition on the number and viability of offspring produced by a long-lived marine bird, the blue-footed booby (Sula nebouxii). In all two-chick broods, one chick (usually the younger one) experiences three stresses: violent aggressive subordination throughout the 3-month nestling period; restricted food ingestion and poor growth during the first half of the nestling period; and high circulating corticosterone (plus, in females, diminished immune responsiveness) during the first weeks of life (see Study species, below). However, juniors largely catch up with seniors in body size and mass by the time they fledge, so this species provides an unusual opportunity to measure the effects of an array of stresses in infancy on breeding success, independent of effects of small body size at fledging. Hitherto, tests have shown that after fledging juniors are generally equivalent to seniors: although female juniors show a deficit in body condition during the first 6 years of life, juniors show similar immune responsiveness, natal dispersal, nest defence, annual recruitment and annual mortality (references in Study species) and breed as successfully during the first several years of adult life (Drummond et al., 2003). However, costs of the poor start could be deferred until late in life (Morgan & Metcalfe, 2001; Alonso-Alvarez et al., 2006) or passed on to the next generation, so we compared the annual production of offspring by junior and senior boobies up to an advanced adult age (16 years) and also compared the probability of recruitment of the offspring they produce in the first 6 years of life.

Materials and methods

Study species

The blue-footed booby lays 1–3 eggs and raises 1–3 chicks, but sometimes adjusts brood size to food availability by facultative, siblicidal brood reduction (Drummond et al., 1986). In the colony on Isla Isabel, Mexico, 46.5% of broods comprise two chicks. The chicks hatch 4 days apart and from roughly age 5 days through at least age 10 weeks juniors are pecked, bitten and threatened daily by their siblings (Drummond et al., 2003), with aggression peaking at an average 60 pecks/bites per 12-h day when aggressive dominance is finally established at 16–21 days. After 3 weeks of aggressive subordination, juniors are ‘trained losers’ that submit promptly to roughly 90% of aggressions (Valderrábano Ibarra et al., 2007), rarely attack and fight ineffectively against experimentally provided younger and smaller opponents (Drummond & Osorno, 1992; Drummond & Canales, 1998). Compared with their siblings, subordinates receive 17% less fish during the first 7 days of life and an estimated 13% less fish between 12 and 35 days (Anderson & Ricklefs, 1995; Guerra & Drummond, 1995), making juniors 11% lighter than seniors at age 20 days (Drummond et al., 1986). In the second half of the nestling period, juniors largely catch up in mass and size (Drummond et al., 2003; Drummond et al., 2003) and at age 70 days they are only 3% lighter than seniors (n = 2107 sibling pairs, unpublished data). Due to food deprivation, circulating corticosterone levels are 109% higher in subordinate chicks than in dominant chicks and singletons at ages 11–20 days and possibly throughout the period of deprivation (Núñez de la Mora et al., 1996). Similarly, in females, the cell-mediated immune response of subordinates is 32% lower than that of dominants at age 20 days (Carmona et al., unpublished data). Junior nestlings are more likely to die before fledging than seniors (40% vs. 29% mortality, unpublished data from 24 seasons) because dominant chicks respond to parental underfeeding by killing their siblings (Drummond & García Chavelas, 1989).

Females and males breed for the first time at 3.85 ± 0.08 years and 4.32 ± 0.09 years, respectively (Drummond et al., 2003), and show senescent decline in reproductive success after 8–10 years (Velando et al., 2006; Beamonte–Barrientos et al., 2010). Juniors appear generally equivalent to seniors in adult life: juniors show similar cell-mediated immune responsiveness (Carmona et al., unpublished data), disperse over a similar distance (Drummond et al., 2010), defend their nests as vigorously (Sánchez-Macouzet & Drummond, 2011), recruit and survive as well at all ages up to 20 years (Drummond et al., 2011) and during the first 5–10 years of life at least, breed as early in the year and as successfully (Drummond et al., 2003).

Comparisons of annual breeding success of senior and junior fledglings over the lifespan are unlikely to be biased by any association between status (senior vs. junior) and either sex, post-fledging mortality or dispersal. Status and sex are not related in two-chick broods (Torres & Drummond, 1999a), and junior and senior fledglings disperse similar distances to their first breeding site (Drummond et al., 2010) and nest in that vicinity during at least the first 8 years of life (Kim et al., 2007). Analysis of 7927 fledglings from 20 cohorts using multi-state mark recapture models showed that after fledging, juniors and seniors have similar probabilities of survival and recruitment at all ages up to age 20 years (Drummond et al., 2011).


Between 1989 and 2010, reproduction was monitored in two study areas on Isla Isabel every year by marking all nest sites with a clutch or brood and inspecting each one every 3 or 6 days throughout the 5-month season until all chicks fledged (survived to age 70 days) or died (details in Drummond et al., 2003). Chicks were marked with plastic rings at hatching and steel rings at fledging, and all ringed breeders were identified. Breeders were sexed by voice (females grunt, males whistle; Nelson, 2005) and fledglings were sexed when they recruited.

We analysed annual breeding success of fledglings that grew up in two-chick broods that suffered no nestling mortality and which subsequently recruited into the breeding population. We assumed that all senior fledglings were dominant and all junior fledglings were subordinate in their natal broods since a dominance-subordination relationship emerges in every booby brood. However, in 9.7% of two-chick broods, the junior chick dominates during at least some of the nestling period (Drummond et al., 1986, 2003; Valderrábano Ibarra et al., 2007) and data from these inverted broods could tend to obscure differences in breeding success of dominants and subordinates in our analyses.

More juniors than seniors die in the nestling period (37% vs. 24% in the ten cohorts studied), but there is no evidence that this facultative brood reduction falls on poor quality juniors so it should not result in poor quality juniors being underrepresented in two-chick broods that suffer no mortality (the subjects of this study). First, descriptive and experimental analyses have shown that differential loss of junior nestlings occurs when seniors escalate their aggression in response to poor personal growth (due to parental underfeeding) rather than when juniors are growing poorly (Drummond et al., 1986; Drummond & García Chavelas, 1989). Second, to the extent that the quality of siblings is correlated, removal from the sample of each junior chick that died was partially counter-balanced by automatic removal of its senior sibling too. The accumulated standardized breeding success of the 62 pairs of sibling recruits up to age 16 years was correlated (= 0.31, P = 0.015), indicating that quality of siblings is indeed correlated.

Statistical analysis

Breeding success

We analysed age-specific annual breeding success (number of chicks fledged) in two overlapping samples that permitted analysis of fledglings up to different ages: (1) up to age 10 years of recruits from ten cohorts (1989, 1991, 1993–2000; in 1990, no fledglings were ringed and in 1992, an El Niño year, no chicks fledged) comprising 683 seniors and 622 juniors, and (2) up to age 16 years of recruits from the earliest three fledgling cohorts (1989, 1991 and 1993) comprising 125 seniors and 107 juniors. In any year between recruitment and death (assumed to occur after its last observed breeding attempt) when a booby skipped breeding (was not observed with clutch or brood), its reproductive success was scored as zero. Breeding success was standardized in relation to the mean breeding success of all pairs breeding in the colony in the same year. In theory, differential post-fledging mortality of the junior fledglings most affected by sibling conflict could bias comparisons based exclusively on recruits (fledglings that survived to breed). However, no bias is expected because junior and senior chicks that fledge along with a sibling do not differ in their annual survivorship or annual probability of recruiting at any age (Drummond et al., 2011).

First, we compared all seniors and juniors in each sample of recruits. Second, to control for genetic quality and condition of parents as well as birth date and natal territory quality of recruits, we limited the comparison to broodmate pairs: the 332 and 62 pairs of siblings in the first and second samples, respectively, that both recruited (first sample: 142 female seniors, 190 male seniors, 158 female juniors, 174 male juniors; second sample: 26 female seniors, 36 male seniors, 26 female juniors, 36 male juniors, all from the 1989 and 1991 cohorts).

For both samples, we used general linear mixed models (GLMMs) with normal error distributions and an identity link function; normal distributions of our samples were confirmed in all cases by graphical analyses of residuals (normal QQ plots; Wilk & Gnanadesikan, 1968). Recruit/breeder identity was included as a random effect and in comparisons of pairs of sibling recruits sibling pair was included as an additional random effect. In addition to status (senior or junior), sex, cohort, age and age2, explanatory variables of initial models included laying date, which substantially affects booby breeding success (e.g. Drummond et al., 2003). Age at death (when last observed breeding) was also included in analyses of the second sample, because annual breeding success could covary with lifespan differently in juniors and seniors. Models included the two-way interactions of interest: status*sex, status*cohort, status*age2 and status*age at death.

Recruitment of offspring

We compared the probability of recruitment during the first 6 years of life of offspring fledged by recruits of both status categories belonging to seven cohorts (1989, 1991, 1993–1997), considering these recruits' first 5 years of life, the period during which offspring recruitment increases linearly before plateauing (Torres et al., 2011). Failure of fledglings (70-day old nestlings) to recruit implies death during fledging, the transition to independence or early adulthood. The recruits comprised 110 female seniors, 129 male seniors, 102 female juniors and 97 male juniors. More than 95% of male and female recruits breed for the first time by age 6 years (Drummond et al., 2003). We used GLMMs with binomial error distribution and logit link function, with three random factors: recruit/breeder identity, recruit/breeder cohort and, to control for interannual variation in conditions experienced, annual proportion of recruitment in each fledgling cohort (including fledglings of all status categories and brood sizes). In addition to status, sex and age of the breeder/recruit, the model included hatching date of the offspring and two-way interactions of interest (status*sex, status*age). In a second model, all recruits that died during the first 5 years of life were excluded, to discount the possibility that observed age-related changes were due to differential mortality.

We simplified models by sequentially dropping nonsignificant interactions and main terms. To compare the simplified minimal adequate model with the model including a nonsignificant term or with the model excluding a significant term, we used likelihood ratio tests for GLMMs with normal error distributions and χ2 tests for GLMMs with binomial error distributions (Crawley, 2007). Significant status*sex interactions in GLMMs with normal distribution and identity link function (the only interactions that were significant in these models) were followed up by analysing each sex separately in a model including all terms of the minimal adequate model. All GLMMs were fitted using R (v.2.10.1; library lme4 and nlme; Crawley, 2007).


Breeding success

In the ten cohorts of recruits monitored up to age 10 years, annual number of fledglings produced increased with age (age: < 0.0001; Table 1) but did not vary significantly among sexes (F1,1276 = 0.026, P = 0.872), cohorts (F9,1276 = 0.48, P = 0.825) or status categories (F2,1276 = 0.092, P = 0.761; Fig. 1). Similarly, in the sample of broodmate pairs monitored for 10 years, breeding success increased with age (age: P < 0.0001; Table 1) but did not vary significantly among sexes (F1,319 = 0.005, P = 0.943), cohorts (F8,323 = 0.79, P = 0.613) or status categories (F1,319 = 0.04, P = 0.842).

Table 1. General linear mixed models examining the effects of status (senior and junior), sex, cohort, age, age2, laying date and age at death on annual standardized number of fledglings produced by male and female recruits over the first 10 or 16 years of life. Breeder identity and sibling pair were included as random effects and initial models included two-way interactions of status with sex, cohort, age2 and age at death. For the first 10 years of life 10 cohorts of recruits were analysed, for the first 16 years 3 cohorts were analysed
 Estimate ± SEd.f. F P
First 10 years
Females + Males
Age0.231 ± 0.0301,507398.17< 0.0001
Age2−0.013 ± 0.0021,507328.64< 0.0001
Broodmate pairs
Age0.066 ± 0.0491,216453.25< 0.0001
Age2−0.002 ± 0.0031,21647.050.008
Laying date−1.582 ± 0.0661,2164571.61< 0.0001
First 16 years
Status0.234 ± 0.0721,1048.030.0055
Age0.169 ± 0.0501,7297.970.0049
Age2−0.008 ± 0.0021,7299.090.0027
Laying date−0.469 ± 0.1901,1046.070.0154
Age at death0.340 ± 0.0121,7297.350.0068
Age0.253 ± 0.0391,12748.410.0038
Age2−0.013 ± 0.0021,127436.25< 0.0001
Broodmate pairs
Age0.259 ± 0.0511,8325.090.0235
Age2−0.013 ± 0.0031,83222.98< 0.0001
Laying date−0.421 ± 0.2061,594.410.0493
Figure 1.

Standardized number of fledglings (mean ± SE) produced by former junior (grey symbols) and former senior (black symbols) blue-footed boobies from 10 cohorts over the first 10 years of life. Scores of the 683 former seniors and 622 former juniors of both sexes did not differ significantly.

In the three cohorts of recruits monitored up to age 16 years, the relationship between status and annual breeding success differed between the sexes (status*sex: F1,226 = 5.57, P = 0.019). In females, breeding success increased with age up to 10–12 years then declined (age2: P = 0.002; Table 1, Fig. 2a), and differed among status categories (status: P = 0.005; Table 1), with juniors outperforming seniors across most of the 16-year period (Fig. 2a). In males, breeding success increased with age up to 9–11 years then declined (age2: P < 0.0001; Table 1, Fig. 2b), but did not differ among status categories (status: F1,121 = 0.017, P = 0.894; Fig. 2b). In the sample of broodmate pairs monitored for 16 years, status*sex was not significant (F1,58 = 0.0006, P = 0.936), and although juniors tended to outperform seniors across most of the 16-year period the difference between them was not significant (status: F1,59 = 2.15, P = 0.15; Fig. 2c). Age at death was unrelated to breeding success of males (F1,792 = 0.298, P = 0.585) or broodmate pairs (F1,832 = 0.959, P = 0.327) and positively related to breeding success of females (females: F1,729 = 7.355, P = 0.007), but in none of these samples was there an effect of status*age at death (females: F1,728 = 0.434, P = 0.510; males: F1,792 = 0.639, P = 0.425; broodmate pairs F1,830 = 2.367, P = 0.124). Thus, both the reported similarities between juniors and seniors, and the superiority of female versus male juniors in the second sample, were independent of their lifespans.

Figure 2.

Standardized number of fledglings (mean ± SE) produced by former junior (grey curves) and former senior (black dashed curves) blue-footed boobies from three cohorts over the first 16 years of life. (a) Females and (b) males of the whole sample of 125 seniors and 107 juniors and (c) a subsample of 62 pairs of siblings including both sexes. The quadratic effect of age (shown here) generally explained more variance than the linear effect. Juniors were not inferior to seniors: female juniors outscored female seniors (P = 0.0055), otherwise juniors and seniors did not differ significantly.

Oddly, in the above results superior breeding success of juniors was revealed by analysis of females of three cohorts across the first 16 years of reproduction (i.e. no age2*status interaction: F1,727 = 0.253, P = 0.615) but not by analysis of the same females along with females of seven other cohorts, over just the first 10 years of breeding. Importantly, when we inspected mean accumulated standardized breeding success over the 10 years, the cohorts included in the 10-year analysis showed no general tendency for superiority of junior females: in four cohorts juniors outperformed seniors and in six cohorts seniors outperformed juniors (data not shown).

Recruitment of offspring

Recruitment of offspring did not differ significantly between male and female parents (χ2 = 0.178, P = 0.673) or among recruit/breeder cohorts (χ2 = 2.00, P = 0.156), but varied with offspring hatching date (χ2 = 15.74, P < 0.0001) and with annual proportion of recruitment in each fledgling cohort (χ2 = 37.043, P = 0.0001), and was differentially related to parental age in seniors and juniors (status*age: χ2 = 6.358, P = 0.011; Fig. 3, Table 2). The probability that offspring of former junior chicks would recruit was relatively low when juniors were 2–3 years old, then increased to equal that of former senior chicks' offspring by age 4–5 years (Fig. 3). A similar pattern was obtained when the analysis was repeated using only the 167 seniors and 151 juniors that survived and reproduced until at least age 5 years (status*age: χ2 = 5.49, P = 0.019).

Table 2. Generalized linear mixed models examining the effects of status (senior and junior), sex, age and age2 of 438 breeders/recruits of both sexes on probability of recruitment of offspring they fledged in their first 5 years. In a second model, all 120 breeders/recruits that died during the first 5 years of life were excluded. Models included breeder identity, breeder cohort and proportion of recruitment per cohort as random effects; initial models included hatching date of the offspring and two-way interactions of status with sex and age
 Estimate ± SEd.f. χ 2 P
All breeders/recruits
Proportional hatch date−1.671 ± 0.975115.740< 0.0001
Status2.273 ± 0.90326.3650.041
Age0.328 ± 0.20426.3730.041
Status : age−0.529 ± 0.20716.3580.011
Breeders/recruits that survived at least 5 years
Proportional hatch date−1.935 ± 0.475116.719< 0.0001
Status2.394 ± 1.02025.5130.063
Age0.271 ± 0.22525.5670.061
Status : age−0.548 ± 0.23115.4970.019
Figure 3.

Estimated probability of recruitment of offspring of former juniors (grey curves) and seniors (black dashed curves) fledged during the parents' first 5 years of life. Status : age, P = 0.011. Probabilities were estimated from the mixed logit model described in Methods, with parameters estimated from the offspring of 201 seniors and 182 juniors from seven cohorts. Hatching dates substantially affected recruitment, although similarly for offspring of juniors and seniors. For illustration, Figure shows recruiting probabilities of offspring hatching mid-way through each reproductive season (hatch rank 0.5).


Contrary to expectation, junior recruits produced at least as many fledglings as senior recruits at all ages up to 16 years, independent of their lifespan. Despite an estimated 90% of them having experienced aggressive subordination, food deprivation and slow growth (followed by compensatory growth) and elevated circulating corticosterone (plus, in females, diminished immune responsiveness) during the nestling period, both female and male juniors from two-fledgling broods produced as many offspring as their senior counterparts. Even in samples of sibling recruits, which controlled for parental quality and natal environment, within-pair comparisons revealed no tendency for juniors to produce fewer offspring. Thus, unreduced broods of two boobies provide no evidence for any deleterious effect of junior chicks' poor start on their breeding success at any age up to 16 years, or at the end of the individual's lifespan, and no evidence for any beneficial effect of senior chicks' exercise of aggressive dominance on their breeding success. The results of the first analysis of impacts of sibling competition and conflict on reproduction of a wild vertebrate across the lifespan largely contradict expectations arising from the studies of other vertebrates (summarized in Introduction) that have documented impacts of experimentally imposed poor starts on diverse adult traits.

However, the fledglings produced by former juniors of both sexes in their first 3 years of life were less likely to recruit than those produced by former seniors, implying that the junior booby's poor start prejudices its condition in early adult life, with negative consequences for its offspring, an intergenerational effect. There was progressive improvement in the viability of offspring over former juniors' first 4 years of life, even when individual juniors that died during that period were excluded from the sample, indicating that improvement was due not to poor quality recruits dying very young but to individuals fledging increasingly viable offspring as they aged and their own condition improved. An intergenerational effect was shown in domestic zebra finches when daughters of females that had grown up on limited rations in enlarged broods showed deficits in adult body size and hatching success (Naguib & Gil, 2005; Naguib et al., 2006). An important difference is that whereas the booby parents that suffered subordination in infancy were generally equivalent to seniors during most of adult life, the finch parents that suffered experimental deprivation in infancy were permanently stunted by it.

Studies of blue-footed boobies on Isla Isabel have now shown that former juniors are equivalent to former seniors not only in natal dispersal (Drummond et al., 2010), territory defence (Sánchez-Macouzet & Drummond, 2011) and immune responsiveness (in females; Carmona et al., unpublished data) but also in exhibiting no deficit at any post-fledging age in survival, recruitment (Drummond et al., 2011) or annual breeding success. As far as we know, juniors' only performance deficit is the poor quality of the offspring they fledge in early adulthood. Blue-footed boobies produce few offspring in the first 3 years of life, because most birds recruit after age three (Drummond et al., 2003) and nest success is especially low at ages one to 3 years (Fig. 1). Our correlational results document for the first time that competition between infant broodmates can affect viability of offspring of the next generation. They show for a long-lived species in which severe subordination to a sibling does not result in a substantial deficit in size or mass at independence, that the cost can be passed to the next generation and confined to offspring produced very early in life.

Experimental poor starts that have produced diverse phenotypic deficits in birds, fish and mammals have created an image of great developmental vulnerability in vertebrates. However, experimental treatments have not been designed to mimic natural challenges and they have usually been applied to short-lived species in captive contexts where scope for making facultative and ameliorative responses is constrained (Drummond et al., 2011). Effects of poor starts on breeding success over the lifespan have seldom been analysed. The boobies' poor start was, of course, natural in all respects and it is experienced by sufficient blue-footed boobies to constitute a major selective pressure: 24.1% of Isla Isabel fledglings cohabited with an elder sibling throughout the nestling period (unpublished data on 24 cohorts). Adverse conditions can certainly prejudice development of vertebrate infants and exact a reproductive cost in adulthood, but this could select for developmental plasticity. Female great tits (Parus major) have apparently evolved resilience to long-term effects of unfavourable natal environments (Wilkin & Sheldon, 2009) and boobies may have evolved to withstand the stresses associated with subordination. However, their plasticity, although great, is limited. In the terminology of Monaghan (2008), the stresses of subordination slightly exceed the levels to which the junior booby can adjust without any fitness penalty (its zone of tolerance) and oblige it to mitigate their detrimental potential by downgrading the quality of its first progeny. In species where competition among infant siblings results in some individuals suffering privation or abuse, we should not assume that as adults they will show important deficits, or reduced survival or fitness, and we should suspect that low-status individuals may pass the cost to the next generation. Sibling competition resulting in diminished body size/mass at independence, on the other hand, seems likely to prejudice both survival and reproduction, because undersized animals tend to be outcompeted by conspecifics.

How boobies minimize developmental damage to juniors during infancy is not clear. There is no evidence that stresses during the nestling stage are counteracted by maternal favouritism in the egg stage: in an El Niño year at least, second eggs were similar to first eggs in size and yolk androgens, and actually contained 10% less yolk than first eggs (Drummond et al., 2008). Animals can reduce and may be able to eliminate impacts of natural poor starts by adopting alternative reproductive strategies or making age-related changes in life history allocations (Birkhead et al., 1999; Auer, 2010), or by developmental buffering against stress (Nijhout, 2003). Similar reproductive and life history strategies in junior and senior boobies are implied by similar natal dispersal and similar age-related mortality, recruitment and reproductive success. Production of low-recruiting offspring in the first 3 years of life could reflect a minor adjustment of a life-history trade-off: a reduction in early-life reproductive investment that enables subsequent survival and reproduction of juniors to reach the same level as in seniors. Nonetheless, former juniors' general equivalence to former dominants can probably be attributed to evolved physiological buffering against subordination, food deprivation and elevated cortico-sterone.

In addition, at the behavioural level, parents and siblings may have been selected to enhance their own inclusive fitness by moderating the challenges faced by subordinates. That junior boobies are scarcely underweight at fledging could imply that parents and siblings facilitate their compensatory growth. Long-lived birds may often be able to keep challenges to their offspring below damaging levels by facultatively adjusting their clutch and brood sizes, their progeny sex ratios and their food provisioning, or by postponing breeding. High annual breeder survival of the blue-footed booby (generally ˜90%, Oro et al., 2010) should give it increased scope to moderate or postpone investment until circumstances are more favourable.

Greater production of fledglings by junior females than by senior females in a sample of three cohorts observed up to age 16 years demands an explanation. We attribute this result mainly to sampling error among cohorts rather than a general superiority of junior females in the Isla Isabel population, since nestling status did not affect reproduction of either sex in the sample of ten cohorts (including those three) observed for 10 years. Even so, it is intriguing that in a small sample of cohorts junior females significantly outperformed seniors, implying that the particular social or ecological contexts in which those cohorts developed propitiated enhanced development of junior females relative to senior females. In principle, challenging early-life experiences expected to affect development negatively could sometimes include elements that enhance aspects of it, and enhancements could even outweigh negative effects. Enhancements could be sex-biased (cf. Appleby et al., 1997; Reid et al., 2003) and they could be contingent on variation in early-life environments; they may be most likely when animals respond to challenges by learning new skills or modifying their social relationships. For example, the female-biased enhancement of breeding success that we observed in former juniors could arise from females' experience challenging dominance by males in those broods (roughly 25% of total) where the elder chick is male and the younger chick female, behaviour that is seldom seen in broods with other sex/hatch order combinations (Drummond et al., 2003). Experience challenging their elder but smaller male siblings in the second half of the nestling period might enable females to develop social skills that yield benefits in their adult interactions with suitors, mates and colony neighbours. Another possibility is that in some contexts long-term development of females (the faster growing sex; Drummond et al., 2003; Torres & Drummond, 1999b) is better served in the first half of the nestling period by the junior chick's constrained growth (due to food restriction) than the senior chick's faster growth, since fast growth in infancy can generate delayed costs (Metcalfe & Monaghan, 2003).


We thank José Luis Osorno, Roxana Torres and numerous student volunteers for dedicated work in the field and on the database, Sin-Yeon Kim and Dani Oro for comments on the manuscript, Alejandra Ramos and Oscar Sánchez for help with figures and Salvador Sánchez Colon and an anonymous reviewer for advice on statistics. Annual fieldwork on Isla Isabel depended on the generous support of many fishermen and the logistical support of the Secretaría del Medioambiente y Recursos Naturales and the Mexican navy. This work was supported by the Consejo Nacional de Ciencia y Tecnología (1986–88, grant numbers 4722-N9407, C01-47599, D112-903581, PCCNCNA-031528, 31973-H, 34500V, 104313), the Universidad Nacional Autónoma de México (grant numbers PAPIIT, 1991–93 IN211491, IN-200702-3) and the National Geographic Society (1992, 1996).