Do females plastically adjust hatching asynchrony as a reproductive tactic in response to climatic cues?

Hatching asynchrony is understood as a female tactic that shapes family dynamics, but its adjustment mechanisms and adaptiveness remain unclear. Using 32 years of individual data on a Neotropical seabird, the blue-footed booby, Sula nebouxii , we examined variation in hatching asynchrony in relation to sea surface temperature, a proxy for food abundance, and hatching asynchrony's adaptiveness. Under warm sea surface temperature, signalling low food abundance, females extended laying interval, resulting in increased hatching asynchrony. Longer hatching asynchrony was associated with more probable and earlier brood reduction. When brood reduction occurred, longer hatching asynchrony improved females' prospect of breeding the next year but not their timing of laying or production of ﬂ edglings and recruits. By extending laying interval, female boobies increase hatching asynchrony to cope with poor food conditions by prompting early brood reduction, thereby reducing reproductive costs and enhancing the probability of breeding the next year. Understanding the temperature sensitivity of plastic reproductive traits is crucial for predicting organisms' responses and resilience to global warming. © 2024 The Author

Hatching asynchrony is widespread in avian broods, with intervals between first and last nestlings ranging from hours to several days (Stoleson & Beissinger, 1995).Hatching asynchrony derives from laying interval, delay in the onset of incubation and differential investment in successive eggs (Tom as, 2014).Lack (1954) interpreted hatching asynchrony as part of a parental strategy to manage the impact of food scarcity by creating a competitive hierarchy among nestlings, enabling the earliest hatched to outcompete their younger siblings for parentally provided food and precipitate their death through selective starvation (facultative brood reduction hypothesis).Other proposals hold that hatching asynchrony reduces parents' workload by decreasing sibling competition (rivalry reduction hypothesis) or peak parental workload by spacing out nestlings' maximum food requirements over time (peak load reduction hypothesis).Alternatively, hatching asynchrony could be a by-product of hurried incubation to accelerate hatching and fledging and minimize both predation and exposure to food scarcity at the end of the breeding season (hurry-up hypothesis) or to maintain egg viability in hot environments (egg viability hypothesis).Similarly, nutritional stress might prompt females to increase laying gaps, thus increasing hatching asynchrony, or conversely, decrease hatching asynchrony by delaying onset of incubation as they spend more time foraging to accumulate energy reserves (hypotheses reviewed in Stoleson & Beissinger, 1995).
We address phenotypic plasticity of individual females (Barrientos et al., 2016;Kontiainen et al., 2010;Stenning, 2008), one of the least explored aspects of hatching asynchrony, to test whether it is correlated with environmental indices of food availability and to assess its adaptiveness.As global warming advances, understanding climate-induced variations in traits associated with reproductive success is crucial, as warming could influence the microevolution of reproductive traits and the persistence of populations (Parmesan, 2006).For example, adaptively plastic hatch intervals may help individuals cope with increasingly stressful conditions by finding the right balance between maximizing the number of offspring and preserving their own condition, thereby ensuring future reproduction and survival (Velando & Alonso-Alvarez, 2003).Such plasticity should be especially important for long-lived species, whose microevolution is slow (Charmantier & Gienapp, 2014), and for dietary specialists, which are vulnerable to changes in food availability (Barrientos et al., 2016;S ¸ekercio glu et al., 2012).
Our focal species, the blue-footed booby, Sula nebouxii, is a piscivorous Neotropical seabird whose broods of two chicks hatch at intervals of 0e13 days (mean ± SD ¼ 4.3 ± 2.4; N ¼ 146 dailymonitored broods in 1985, 1986and 1990;Bizberg-Barraza & Drummond, 2023).In the study population, asymmetries in development and size among siblings facilitate emergence of relatively stable, age-related, pairwise dominance relationships in broods of two (Drummond et al., 1986) and three chicks (Valderr abano-Ibarra et al., 2007), in which dominance assures feeding priority.When food is scarce, the youngest/most subordinate chicks' share of food declines, sometimes resulting in their death (facultative brood reduction) from starvation or expulsion.During their long lives (mean ± SD ¼ 9.2 ± 4 years), breeders experience interannual variation in sea surface temperature, food abundance and reproductive success associated with El Niño Southern Oscillation (ENSO; Ancona et al., 2011;Ballance et al., 2006), setting the stage for adaptive plasticity in hatching asynchrony (Clutton-Brock & Sheldon, 2010).
The reproduction of our highly philopatric (Kim et al., 2007) population of boobies banded at fledging was monitored over 32 years and six ENSO events.For our study of two-chick broods, we used two modelling approaches ('sliding window' and 'withinsubject centering') to identify the temporal windows when local sea surface temperature best explained hatching asynchrony variation and to partition the population level relationship between sea surface temperature and hatching asynchrony into within-and between-individual components.We also applied both analyses to laying interval, a variable previously shown to influence hatching asynchrony in this species (Anderson, 1989).This approach enabled us to determine whether and when females plastically adjust hatching asynchrony to sea surface temperature, a proxy of fish abundance (Lanz et al., 2009).We evaluated the adaptiveness of hatching asynchrony under varying food abundance by testing for its effects on the probability and latency of brood reduction and on females' probability of breeding, timing of laying and production of fledglings and recruits the following year.Studies of hatching asynchrony functionality have generally been conducted over narrower ranges of environmental conditions (Parejo et al., 2015) and did not test for effects of hatching asynchrony on parents' future reproductive success (Callery et al., 2022), leaving adaptiveness unclear.
Boobies of the study population feed exclusively on cold-water pelagic fish (Taylor et al., 2011) and their reproductive success declines with increased sea surface temperature (Ancona et al., 2011).We hypothesized that when water is warm around the time of laying, females increase their laying and hatching intervals to facilitate and accelerate brood reduction in the current year and that the resulting reduction in parental investment pays off in the following year by increasing females' probability of breeding, enabling earlier laying or increasing production of fledglings and recruits.

Study Population
Subjects were from the socially monogamous blue-footed booby population on Isla Isabel, Mexico (21 50 0 57 00 N, 105 52 0 53 00 W), where recruits normally breed each season (DecembereJuly) close to their natal neighbourhoods (Kim et al., 2007).Almost 60% of females lay two eggs while 30% and 11% lay one and three eggs, respectively.Hatching failure and egg predation contribute to the prevalence of single-chick broods (approximately 47%) in the colony, followed by broods of two (approximately 44%; Bizberg-Barraza et al., 2024).The two caretaking parents alternate incubation shifts approximately equally (Guerra & Drummond, 1995) during roughly 6e7 weeks (mean ± SD ¼ 40.12 ± 3.96 days/egg), then both feed the brood during 3e4 months, with female parents (larger and 25% heavier) providing three times as much food as male parents (Guerra & Drummond, 1995).Elder chicks attack and threaten their younger siblings daily throughout the nestling period, and parents neither interfere in sibling agonism nor provide compensatory feeding to last-hatched chicks (Drummond et al., 1986(Drummond et al., , 1991)).In broods of two, 40% of junior chicks and 29% of senior chicks die during the nestling period (Drummond et al., 2011).

Data Collection
Between 1989 and 2021, all nests in our two study areas (26 889 m 2 ) were monitored annually, from roughly mid-February through mid-July.Nest contents were recorded every 3 days until late in the breeding season, then every 6 days until the last nestling fledged (Drummond et al., 2003).Nestlings were marked within 3 days of hatching with coloured plastic bands according to hatch order, replaced by alphanumeric steel bands at fledging (70 days old; Drummond et al., 2003).We used daily average sea surface temperature (AVHRR Pathfinder Version 5.3) and rainfall (CHIRPS Version 2.0) from 1989 to 2021, extracted from the National Oceanic and Atmospheric Administration website (http://www.nodc.noaa.gov)over an area of approximately 60 Â 60 km centred on Isla Isabel, roughly equivalent to blue-footed booby average foraging ranges (Zavalaga et al., 2008).

Statistical Analyses
All analyses were performed using R software 4.2.0 (R Core Team, 2022).Covariates were standardized prior to their inclusion in models.We used the variance inflation factor (VIF) analysis from the 'performance' package (Lüdecke et al., 2021) to assess collinearity between covariates (acceptable VIF < 1.5).

Identifying climatic drivers
To analyse sea surface temperature influence on laying and hatching intervals, a sliding window analysis ('climwin' package for R; van de Pol et al., 2016) identified the weekly intervals (i.e.number of weeks prior to clutch and brood completion, respectively) in which sea surface temperature best explained the variation in laying interval and hatching asynchrony.To account for potential confounding effects of rainfall on seabird foraging effort (De Pascalis et al., 2022) and its potential adverse impact on incubation temperatures affecting embryonic development and hatching date (Hilton et al., 2004), we also identified critical time windows for rainfall.Initial search windows were determined through preliminary analysis to find the shortest time window including all climate signals, to limit the probability of finding false positives (van de Pol et al., 2016).We tested the linear and quadratic terms over multiple descriptive metrics for sea surface temperature (maximum, minimum, mean) and rainfall (maximum, cumulative sum) in all analyses.
The sliding window approach for laying interval was carried out using all 1813 ringed females that established a clutch of two during 3-day monitoring, and initial search windows for sea surface temperature and rainfall were set at 10 weeks prior to clutch completion.The sliding window analysis for hatching asynchrony was performed using all 972 ringed females with clutches of two that hatched both eggs during 3-day monitoring.Initial search windows for sea surface temperature and rainfall encompassed, respectively, 12 and 14 weeks preceding brood completion.
For both sliding window analyses, the baseline model (model without climatic covariates) used for comparison with a model including climatic covariables was a linear model ('lm' from the package 'lme4'; Bates et al., 2015) with Gaussian error structure.Given the 3-day nest monitoring interval, laying and hatching intervals (the response variables, in days) were estimated by subtracting the estimated date (midpoint between two checks of the nest where the egg or chick was first observed) of the first egg/chick from the estimated date of the second egg/chick.Both models included as covariates laying date (relative rank among breeders of the same year; values close to zero designate early breeders while values close to one indicate late breeders) and female age (linear and quadratic terms) as both can affect hatching asynchrony of birds (Callery et al., 2022;Wiebe et al., 1998).A randomization procedure assessed the likelihood that a found climate signal was not a false positive (more details in van de Pol et al., 2016).When both sea surface temperature and rainfall were identified as relevant climatic variables, we reran the sliding window analysis, including in the baseline model the best-supported window of the remaining variable.Finally, the newly best-supported environmental signals were extracted and retested in a linear mixed model ('lmm').The model was implemented using the 'lmer' function from the R package 'lme4' (Bates et al., 2015) and included female identity (ID) and year as random effects to account for sampling year and avoid pseudoreplication.

Within-versus between-individual variation
To attribute variation in laying interval and hatching asynchrony in the population to within-versus between-individual variation in responsiveness to sea surface temperature, we carried out a withinsubject centering analysis on a subset of females that reproduced at least twice, selected from the larger sample used in the sliding window analysis (laying interval: N ¼ 914 females; hatching interval: N ¼ 228 females) (van de Pol & Wright, 2009).Withinsubject centering was implemented by replacing the sea surface temperature terms in the lmm models with the mean sea surface temperature experienced by each parent across all of its breeding attempts (between-subject: b B ) and the mean-centred sea surface temperature experienced per parent at each reproductive event (within-subject: b W ). We tested the statistical difference between the between-subject and within-subject slopes to explicitly test the importance of the within-subject component on the population level response to sea surface temperature (van de Pol & Wright, 2009).This was done by replacing the within-subject term in the previous model with the best-supported window for sea surface temperature experienced by the female each year.

Testing adaptiveness of hatching asynchrony
We first evaluated the effect of hatching asynchrony on brood reduction probability under multiple sea surface temperatures (interaction: hatching asynchrony)sea surface temperature) in broods of two (monitored at 3-day intervals; N ¼ 1261 observations).Brood reduction was defined as the death of the junior chick, leaving the senior one alive, irrespective of whether the senior chick eventually fledged.Our model was a logistic regression including hatching interval as a continuous predictor and five variables proven to influence the probability of brood reduction (Bizberg-Barraza et al., 2024): laying date, linear and quadratic terms of female age and mean sea surface temperature experienced by the clutch/brood during incubation and parental care (the 110 days following clutch initiation).Female ID and cohort were included as random effects.
To evaluate the consequence of hatching asynchrony on brood reduction latency, we fitted a linear regression (lme4; Bates et al., 2015) using all 499 broods of two, monitored at 3-day intervals, in which brood reduction occurred.The model included the victim's age of death as the response variable, hatching interval as a continuous predictor, laying date, female age (linear and quadratic terms) and sea surface temperature experienced by the second-hatched egg/chick during incubation and parental care (linear and quadratic terms) and female ID and cohort as random effects.
The consequences of hatching asynchrony on female breeding probability (being sighted with a nest; model 1), earliness of laying (model 2) and number of fledglings (model 3) and recruits produced (model 4) in the next year were evaluated by fitting a generalized linear mixed model (GLMM from the R package 'glmmTMB'; Brooks et al., 2017) with a binomial distribution (N ¼ 1440 observations, model 1), a linear mixed model (N ¼ 844 observations, model 2) and a cumulative link mixed models from the R package 'ordinal' (Therneau, 2020) (N ¼ 484 observations, model 3; N ¼ 385 observations, model 4).Since most boobies recruit within their first 6 years (Oro et al., 2010), we restricted model 4 to reproductive events occurring between 1989 and 2015, allowing a 6-year recruitment period for the 2015 cohort.All models included four covariates from the year prior to the focal year: hatching interval (continuous predictor of interest), laying date, brood reduction and female age (linear and quadratic terms) as control predictors and female ID and cohort as random effects.In addition, the interaction between hatching interval and brood reduction was added to test whether brood reduction influenced the effect of hatching asynchrony on females' reproductive parameters.In models 3 and 4, we included two additional variables from the focal year: laying date and mean sea surface temperature experienced by females during the 110 days following clutch initiation.

Ethical Note
Colony monitoring followed ASAB/ABS (2012) Guidelines for animal treatment and was approved by Mexican authorities: SEMAR, CONANP, INE, SEMARNAT and Parque Nacional Isla Isabel.Two observers carried out nest inspections with minimum handling of the birds.To approach the nests, observers maintained a slow pace and walked in single file to minimize disturbance to the birds.Blue-footed boobies are highly tolerant of human presence, with most parents remaining on their nests during our inspections.In the rare event that a parent flew off, it typically returned within a minute.If necessary, observers used a forked stick to gently nudge caretaking adults away from the nest to determine the contents of nests and parents' IDs.Upon completion of the inspection, observers remained nearby until the parent covered the eggs and/or nestlings.If parents flew away and seagulls were nearby, the observers covered the eggs with leaves to prevent predation.

Population Level Response to Climatic Variables
Variation in laying interval and hatching interval was best explained, respectively, by the maximum weekly sea surface temperature 1 week prior to clutch completion (Appendix, Fig. A1) and 5e9 weeks before brood completion (Appendix, Fig. A2).As predicted, laying and hatching intervals increased as sea surface temperature rose in those windows (laying interval:  S1eS2).

Within-versus Between-individual Variation
The population level relationship between sea surface temperature and laying interval was driven by individual females increasing their intervals in response to warm sea surface temperature (phenotypic plasticity), as well as by females with consistently longer intervals being more likely to breed during high sea surface temperature (Table 1,  In contrast, population level adjustment of hatching asynchrony to sea surface temperature was driven exclusively by betweenfemale differences, as evidenced by the significance of only b B (Table 1).However, as the difference between b B and b W slopes was barely significant, apparent lack of within-female plasticity in hatching asynchrony should be interpreted with caution (Roast et al., 2022)
The interaction between hatching asynchrony and brood reduction influenced female reproduction in the subsequent year.Females that experienced brood reduction were more likely to breed in the following year if their hatch interval was long (brood reduction)hatching asynchrony: b ± SE ¼ 0.74 ± 0.34, 95% CI: [0.08, 1.4]; Fig. 5, Appendix, Table A3).These individuals often bred successfully: among the females that experienced brood reduction and bred the next year, 42% fledged at least one chick in the next year and 19% produced at least one recruit.However, among females that bred the next year, hatching asynchrony had no effect on the timing of laying or the number of fledglings or recruits produced, whether the females experienced brood reduction or not (Appendix, Table A3).

DISCUSSION
This study assessed hatching asynchrony as a female reproductive tactic to cope with scarcity of prey fish.Females increased hatching asynchrony by extending the laying interval in response to warm sea surface temperature during the week before completing laying (roughly their laying period).As hypothesized, greater hatching asynchrony precipitated more frequent and earlier brood reduction and increased brood-reducing females' probability of breeding the subsequent year, presumably by enabling them to conserve resources.
Long hatch intervals were associated with more prompt brood reduction (Fig. 4), and among females that experienced brood reduction, those with longer hatch intervals were more likely to breed the following year (Fig. 5).Under poor food conditions, longer hatch intervals may therefore increase brood-reducing females' probability of breeding the next year by curtailing investment in the current brood (Nur, 1988;Vedder et al., 2017Vedder et al., , 2019)).This finding aligns with a previous experimental study showing that early brood reduction improved female boobies' body condition (Velando & Alonso-Alvarez, 2003).However, among females that bred in two consecutive years, neither laying date nor number of fledglings or recruits the second year improved with longer hatch intervals in the first year.Therefore, a function of longer intervals may be enabling brood-reducing females to gain an extra breeding season in the subsequent year, rather than enhancing the reproductive outputs of females that breed in the next season.
Females increased their hatching intervals by delaying laying of the second egg in response to warm sea surface temperature in the week before clutch completion (Table 1).Long intervals could be forced by current nutritional stress (Aparicio, 1994;Nilsson & Svensson, 1993;Tov & Wright, 1993), or they could be a reproductive tactic in anticipation of poor-food conditions.Adjustment of laying interval in response to environmental cues could also be a tactic enabling female birds to fine-tune synchronization of hatching and peak food demands of their clutches with peak food abundance (review in Tom as, 2014).Although our study cannot discount a possible role of current nutritional stress, we note four arguments supporting the idea of an anticipatory strategy in these boobies.First, costs of egg production by blue-footed boobies are expected to be low because (1) clutches are small and each egg constitutes only 3.5% of female body mass (Kim et al., 2011; see also Townsend & Anderson, 2007)  .Interaction effect of hatching asynchrony and brood reduction (0 ¼ no, 1 ¼ yes) on the probability of breeding the next year for females that experienced brood reduction (blue) and females that did not (red).Intervals of zero days are an artefact of scoring hatching dates based on nest inspections every 3 days.Shaded areas depict the 95% confidence intervals.
weight between prelaying and the conclusion of incubation (Lerma et al., 2022) and (3) calcium deficiencies, which can delay egg production (Astheimer, 1985), should be uncommon in seabirds, which obtain calcium from fish and mollusc shells (Zavalaga et al., 2007).Second, at laying, food shortage may sometimes be only incipient, because in the Gulf of California, which borders our colony's foraging area, sea surface temperature correlates better with fish abundance 2 weeks later than current abundance (Lanz et al., 2009).Third, younger females, which are more prone to nutritional stress due to less efficient foraging (Riotte-Lambert & Weimerskirch, 2013), did not lay at longer intervals.Finally, Nazca boobies, Sula granti, experimentally provided with additional fish prey did not shorten their laying intervals (Clifford & Anderson, 2001).
In the blue-footed booby, we suspect that laying interval is the primary influence on hatching asynchrony.The boobies of Isla Isabel cover their clutches as soon as the first egg is laid (Drummond et al., 1986), and variation in this species' laying interval explains 79% of variation in its hatching asynchrony (Anderson, 1989).Unlike most species in the hatching asynchrony literature, boobies nest on the ground in a hot environment where daytime air (Nolasco-Luna et al., 2022) and substrate (Osorno et al., 2005) temperatures can exceed incubation temperature (z 28 C in this species; Morgan et al., 2003).Because delayed incubation in environments above physiological zero reduces egg hatchability, boobies may be constrained by their nesting microclimate to initiate incubation upon egg laying to maintain egg viability (Stoleson & Beissinger. 1999).Furthermore, female boobies may have limited control over incubation onset as both sexes contribute similarly to egg incubation (Castillo-Alvarez & Chavez-Pe on, 1983).In principle, female birds might influence embryonic development rates and, hence, hatching asynchrony through unequal provision of androgens to successive eggs (Gil et al., 2007;Martin & Schwabl, 2007), but successive blue-footed booby eggs have similar androgen levels (Drummond et al., 2008; but see Dentressangle et al., 2008).
In summary, our results suggest regulation of hatching asynchrony is a tactic of female blue-footed boobies to cope with unfavourable environmental conditions.Through plasticity in their laying intervals, they adjust hatching intervals to short-term cues forecasting future food availability.This flexibility allows them to mitigate the cost of brood care by facilitating siblicidal brood reduction when faced with food scarcity and to minimize the risk of brood reduction associated with longer hatching asynchrony during favourable conditions.Unlike dietary generalist birds that can rely on alternative foods, dietary specialist birds may rely heavily on flexible control of hatching asynchrony to cope with substantial fluctuations in prey availability (Barrientos et al., 2016).The plastic control of this critical life-history trait holds particular significance in light of climate change and its associated stressors.Hatching asynchrony can enable species to buffer against the effects of adversely high sea surface temperature in their foraging ranges (Sanchez-Cabeza et al., 2022).
Given the intricate interplay of multiple intrinsic and extrinsic factors influencing hatching asynchrony, further investigation is warranted in tropical species, which have received limited attention.Supplementation experiments conducted across different stages of reproduction could corroborate our conclusions and help distinguish between longer laying intervals forced by nutritional stress and those driven by adaptive female reproductive strategies.Similarly, exploring the impact of food supplementation in biparental incubators, which may experience fewer incubation time constraints due to foraging, could yield insights into the onset of incubation as a reproductive tactic (cf.Wiebe & Bortolotti, 1994; but see Parejo et al., 2012).Finally, experimental manipulations of laying and hatching intervals (under varying food abundance conditions) on the condition and survival of chicks and parents could significantly deepen our understanding of the trade-offs associated with hatching asynchrony and female birds' reproductive strategies.

Figure 1 .
Figure 1.Effect of sea surface temperatures (SST) one week prior to clutch completion and 5e9 weeks prior to brood completion on (a) laying interval and (b) hatching interval, respectively.Intervals of zero days are an artefact of scoring laying and hatching dates based on nest inspections every 3 days.

Figure 2 .
Figure 2. Within-and between-individual relationships between laying interval and sea surface temperature (SST) in the week prior to clutch completion.Twelve randomly selected females with two to five reproductive events are presented for illustration.Each coloured line represents the within-individual response to SST obtained from a linear regression between observed laying intervals and SSTs of a single female.Intervals of zero days are an artefact of scoring laying dates based on nest inspections every 3 days.Dots represent the mean laying interval and mean SST of each female (coloured dots) or all females (black dot).The bold black line represents the between-individual response to SST, estimated by fitting a linear regression model between average laying interval and average SST for each female across all of its breeding attempts.

Figure 4 .
Figure 4. Effect of hatching asynchrony on latency of brood reduction.Intervals of zero days are an artefact of scoring hatching dates based on nest inspections every 3 days.Shaded area depicts the 95% confidence interval.

Table 1
Within-individual centering analyses showing the effect of sea surface temperature (SST) on laying interval (N ¼ 2707 observations) and hatching interval (N ¼ 519 observations), partitioned into within-subject (b W ) and between-subject (b B ) effects , (2) females do not appear to lose Interaction effect of sea surface temperature (SST) and hatching asynchrony on the probability of brood reduction.Intervals of zero days are an artefact of scoring hatching dates based on nest inspections every 3 days.Shaded areas depict the 95% confidence intervals.