Portal de Periódicos da CAPES

Individual optimization of reproduction in a long-lived migratory bird: a test of the condition-dependent model of laying date and clutch size

2010; Wiley; Volume: 25; Issue: 3 Linguagem: Inglês

10.1111/j.1365-2435.2010.01824.x

1365-2435

Sébastien Descamps, Joël Bêty, Oliver P. Love, H. Grant Gilchrist,

Bird parasitology and diseases

Functional EcologyVolume 25, Issue 3 p. 671-681 Free Access Individual optimization of reproduction in a long-lived migratory bird: a test of the condition-dependent model of laying date and clutch size Sébastien Descamps, Corresponding Author Sébastien Descamps Norwegian Polar Institute, Fram Centre, 9296 Tromsø, Norway Département de Biologie, UQAR-CEN, Rimouski, Québec G5L 3A1, Canada Correspondence author. E-mail: sebastien.descamps@npolar.noSearch for more papers by this authorJoël Bêty, Joël Bêty Département de Biologie, UQAR-CEN, Rimouski, Québec G5L 3A1, CanadaSearch for more papers by this authorOliver P. Love, Oliver P. Love Department of Biological Sciences, University of Windsor, Windsor, Ontario N9B 3P4, CanadaSearch for more papers by this authorH. Grant Gilchrist, H. Grant Gilchrist National Wildlife Research Centre, Environment Canada, Ottawa, Ontario K1A 0H3, CanadaSearch for more papers by this author Sébastien Descamps, Corresponding Author Sébastien Descamps Norwegian Polar Institute, Fram Centre, 9296 Tromsø, Norway Département de Biologie, UQAR-CEN, Rimouski, Québec G5L 3A1, Canada Correspondence author. E-mail: sebastien.descamps@npolar.noSearch for more papers by this authorJoël Bêty, Joël Bêty Département de Biologie, UQAR-CEN, Rimouski, Québec G5L 3A1, CanadaSearch for more papers by this authorOliver P. Love, Oliver P. Love Department of Biological Sciences, University of Windsor, Windsor, Ontario N9B 3P4, CanadaSearch for more papers by this authorH. Grant Gilchrist, H. Grant Gilchrist National Wildlife Research Centre, Environment Canada, Ottawa, Ontario K1A 0H3, CanadaSearch for more papers by this author First published: 23 December 2010 https://doi.org/10.1111/j.1365-2435.2010.01824.xCitations: 61AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary 1. Optimality theory predicts that both timing of arrival and arrival state on the breeding area will determine reproductive timing and investment in migratory organisms. We tested this idea using a condition-dependent individual optimization model (Ardea68, 1980, 225 and The American Naturalist143, 1994, 698) in common eider ducks through descriptive data, path analyses and experimental manipulation. 2. Our results support the causal pathways drawn from the optimization model indicating that individuals adjust their reproductive decisions as a function of their arrival date and body condition at arrival. 3. Independent of body condition, early-arriving females had a longer pre-laying period, but still initiated their nests earlier, and produced larger clutches than late-arriving birds. Independent of arrival date, females in good condition laid earlier than those in poor condition. Manipulation of pre-laying female body condition confirmed that the relationship between condition and laying date was causal. 4. Female common eiders appear to optimize reproductive decisions in response to both their external (i.e. environmental conditions affecting the egg-value) and internal (i.e. body condition) states. These adjustments seem to minimize the fitness costs of reproduction, in which higher clutch size is not associated with an apparent lower survival or future breeding probability. 5. Our study emphasizes the importance of (i) simultaneously considering the timing of migration, the state of individuals and the seasonal change in egg-value to understand clearly birds’ breeding decisions and (ii) appreciating the potential proximate and ultimate factors explaining why some individuals delay breeding and/or produce small clutches. Introduction Two of the most important decisions for seasonally breeding organisms are when to initiate reproduction and how many offspring to produce. For example, in bird species living in seasonal environments, hatching success and offspring post-hatching survival are generally higher for individuals that reproduce early compared to late (e.g. Perrins 1970), typically because the environment is less favourable later in the season (Verhulst & Nilsson 2008). Such environmental changes during a given breeding season lead to changes in egg-value, with eggs laid early in the season being of greater value (i.e. having a higher recruitment probability) than eggs laid late in the season (Lepage, Gauthier & Menu 2000). Parental characteristics, such as body condition, are also expected to affect reproductive output. Indeed, as producing an offspring is energetically demanding, the number and quality of eggs or offspring produced and the parental provisioning or parental care can be affected by parental body condition (e.g. Karell et al. 2008; Öst, Smith & Kilpi 2008). Theoretically then, there should be a conflict between the advantages of breeding early to maximize the value of the eggs vs. the advantages of delaying breeding to improve parental body condition and increase energetic allocation to reproduction (i.e. to lay a larger clutch or increase the ability to raise a larger brood). In a state-dependent life history framework, individuals breeding in seasonal environments are expected to adjust their breeding decisions (timing and investment) to their external (i.e. the environment) and internal (i.e. their condition) states (sensu McNamara & Houston 1996; McNamara 1998). In the case of migratory organisms, reproductive timing and investment should thus be determined by the seasonal variation in egg-value, the timing of migration (and then the date of arrival at the breeding grounds) and the body condition at arrival. This idea was first proposed by Drent & Daan (1980) and then formulized by Rowe, Ludwig & Schluter (1994) into a dynamic model that predicts the optimal laying date and clutch size as a function of arrival date and body condition at arrival. To obtain an accurate understanding of what determines reproductive decisions in seasonal breeding birds requires concurrent consideration of arrival date, body condition at arrival, laying date and clutch size (see Bêty, Gauthier & Giroux 2003 for an example with snow geese). As all of these variables may affect one another reciprocally, and appear in the model as both a response variable and a predictor, analysing all the relationships between these variables simultaneously is not an easy task. However, structural equation modelling techniques (Shipley 2000) provide an ideal statistical framework for such analyses (see Thomas et al. 2007 for an application on blue tits); these multiple-equation regression models represent structural relationships among a number of variables, some of which may affect one another mutually. We begin by testing Drent & Daan (1980) and Rowe, Ludwig & Schluter (1994) optimization model (simply called hereafter: the optimization model) in Arctic-breeding common eiders (Somateria mollissima) using long-term individual monitoring data and structural equation modelling. The relationships between arrival date on the breeding grounds, condition at arrival, delay between arrival and laying, laying date and clutch size expected from this model are schematized in Fig. 1 and on the path diagram in Fig. 2. The model assumes that there (i) are direct negative effects of arrival date and condition at arrival on the delay before laying, (ii) is a direct positive effect of arrival date on laying date, (iii) is a direct negative effect of laying date on clutch size, (iv) is no direct effect of body condition on laying date and clutch size, (v) is no direct effect of arrival date on clutch size, and (vi) is no covariance between date of and condition at arrival. Finally, there should be an obvious direct effect of the delay before laying on laying date (Table 1). The optimization model is based on some basic assumptions. First, individuals are predicted to improve their condition after arrival on the breeding grounds; based on the model, this improvement is expected to lead to a greater clutch size and thus greater fitness benefits (Rowe, Ludwig & Schluter 1994; Fig. 1). A previous study conducted at our focal eider colony showed that females feed after their arrival on the breeding grounds and partially use nutrients acquired during the pre-laying and laying periods to cover the costs of egg formation (Sénéchal 2009; Sénéchal, Bêty & Gilchrist in press). This agrees with the idea that pre-laying female eiders need to acquire resources on the breeding areas, and thus gives some support to this first assumption. Secondly, egg value (i.e. the probability that an egg survives until recruitment) is expected to decline with increasing laying date. It has been shown that duckling post-hatching survival is typically higher in earlier than later broods in this eider colony (Love et al. 2010). Here, we also tested whether laying date was negatively associated with the probability that an egg survives from laying to hatching. Combined together, these results would support the hypothesis that egg-value decreases as the season progresses. Figure 1Open in figure viewerPowerPoint Graphical representation of the condition-dependent optimization model (modified from Bêty, Gauthier & Giroux 2003 and Rowe, Ludwig & Schluter 1994). The thick line represents optimal combinations of clutch size and laying date assuming a trade-off between the cost (decreasing offspring value) and the benefit (increasing condition and hence clutch size) of a delay in laying date. Letters A, B and C represent individuals with different initial condition (A and C have the same condition, and B has a lower condition) or arrival date on the breeding grounds (A and B arrive at the same date and C arrives later). Dashed lines illustrate the increase in condition; bottom arrows indicate the delay between arrival and laying. Individuals must first reach a minimum condition threshold (dotted line) before they can produce a clutch and incubate eggs. Our body condition experiment (see Methods for details) can be represented by individuals A (control individuals) and B (individuals with experimentally decreased body condition). The expected responses of our treatment are thus a later lay date and a smaller clutch size. Figure 2Open in figure viewerPowerPoint (a) Path diagram showing the hypothesized causal structure linking arrival date and body condition at arrival to breeding parameters (delay between arrival and laying, laying date and clutch size) in a common eider colony, East Bay, Southampton Island, Nunavut, Canada. Solid lines indicate the predicted structure based on optimal model shown in Fig. 1. Dashed lines indicate other alternative paths. Signs above each arrow indicate the sign of the predicted effect (positive, negative or no effect); bidirectional arrow represents the covariance between date of and condition at arrival. (b) Standardized path coefficients in hypothesized structural model (n = 318). Table 1. Test of conditional independence implied by the path diagram (Fig. 2a). (X; Y) | {Z} means that variables X and Y are independent conditional of variable Z (i.e. if Z is held constant, variation in X does not imply variation in Y). The associated mixed model used to test the independence claims are Y∼Z + X + 1|Year, where Year represents a random effect Basis set Partial slope ± SE (variable tested) t-value Null probability (Body condition; Arrival date) | {ø}* 1·81 ± 1·96 0·93 0·35 (Clutch size; Arrival date) | {Laying date} −0·013 ± 0·012 −1·14 0·26 (Laying date; Body condition) | {Arrival date; Delay} 0·12 × 10−3 ± 0·14 × 10−3 0·88 0·38 (Clutch size; Body condition) | {Laying date} 0·36 × 10−3 ± 0·33 × 10−3 1·09 0·28 (Clutch size; Delay) | {Arrival date; Body condition; Laying date} 0·095 ± 0·14 0·70 0·49 *Variables Body condition and Arrival date are expected to be independent if we hold constant none of the other variables. Structural equation models examine potential causal relationships using observational data, but cannot actually prove causality. As such, we combined descriptive analyses with an experimental decrease in female eider body condition prior to laying to confirm that the relationship between body condition and delay before laying, and hence lay date and clutch size, was causal. Experimental approaches testing for causal relationships between body condition and breeding parameters (clutch size and laying date) are rare (but see Nooker, Dunn & Whittingham 2005), and most studies have tried to manipulate female condition with food supplementation (see Boutin 1990; Schoech & Hahn 2008 for reviews). However, a major drawback of food supplementation is that providing food ad libitum can affect several parameters at once, including pre-laying condition and the rate of condition gain (through changes in feeding rates). Such changes in the rate of condition gain could affect the predicted relationships between condition, laying date and clutch size (see Rowe, Ludwig & Schluter 1994 for details) and even lead to counter-intuitive relationships. Indeed, in some circumstances (see Rowe, Ludwig & Schluter 1994, p. 710 and Fig. 4), for the same arrival date and body condition at arrival as un-manipulated birds, individuals food-supplemented prior to laying could actually lay later (but with a larger clutch). This can lead to an apparent negative relationship between body condition and laying date. As such, manipulating body condition prior to laying rather than feeding rate during reproduction is required to understand adequately the proximate role of body condition in reproductive decisions. Our experimental approach should thus provide strong insight into our understanding of the proximal role of body condition during reproduction; we predicted that for a given arrival date, females with experimentally reduced pre-laying condition should lay later, and hence produce smaller clutches, than controls. Figure 4Open in figure viewerPowerPoint Relationships between body mass at capture, arrival date and delay between capture and laying date in the East Bay common eider colony, Southampton Island, Nunavut, Canada (n = 318). Arrival dates are expressed in days since January 1. Finally, we investigated the long-term fitness consequences of reproductive decisions in female eiders. Indeed, based on the optimization model, individuals are expected to adjust their reproductive decisions to their state (i.e. it is expected that large clutches characterize females in good condition and/or arriving early and small clutches characterize females in poor condition and/or arriving late). Consequently, large clutches should not represent a greater investment in reproduction (sensu Evans 1990) than small clutches, and an increase in clutch size should not be associated with any apparent decrease in survival or future breeding probability. This does not mean that reproduction does not incur any fitness cost, but rather that such costs should not be apparent in a state-dependent reproductive framework (Reznick 1992). Similarly, in the context of our experiment, individuals with reduced pre-laying condition are expected to adjust their laying date and clutch size so that they do not jeopardize their future survival and/or reproductive prospects. Therefore, manipulated females should achieve the same subsequent survival and future breeding probabilities as controls. Materials and methods Study population The study was conducted on Mitivik Island (64°02′N, 81°47′W), which supports the largest known nesting colony of common eiders in the Canadian Arctic (up to 8500 pairs annually). Details of the colony and the biology and capture of eiders are given in Appendix S1 (Supporting Information). Mass at capture (mean ± SD: 2174 ± 165 g), delay between capture and laying (mean ± SD: 9 ± 7 days), laying date (date of first egg laid; mean ± SD: June 30th ± 7), clutch size (maximum number of eggs found in a nest after the start of incubation; mean ± SD: 2·8 ± 1·0 eggs) and hatching success (probability to hatch ≥1 egg; mean ± SD: 0·58 ± 0·5) were determined for 318 nasal-tagged females. Among these females, the number of hatched ducklings was known for 49 individuals (mean ± SD: 2·1 ± 1·1 ducklings). Females were caught very early in the season when they were flying above the colony as soon as the first eiders arrive at the colony; we therefore assumed that capture date was a good proxy of arrival date at the colony (see also Discussion). A previous study in the same eider population indicated that body mass (not corrected for structural size) of pre-laying females explains 60% of individual variation in pre-laying abdominal fat mass and performs as well as body mass adjusted for structural size as a measure of condition (Descamps et al. 2010). Indeed, structural size (tarsus length) explains only 1% of body mass variation in our colony (Descamps et al. 2010). We thus used body mass at capture as a proxy of female condition at arrival. Study design To test the condition-dependent optimization model in our common eider population, we proceeded in four steps. First, we tested the assumption that egg value decreases as laying date increases within a given reproductive season. Specifically, we tested for the effect of laying date on hatching success (i.e. the probability to hatch ≥1 offspring), and then for the effect of laying date on the number of hatched ducklings among successful nests (i.e. nests that hatched ≥1 duckling) while controlling for clutch size. This should tell us whether or not clutches laid late in the season were less successful and productive than clutches laid early in the season (for a given clutch size). This result, combined with the observed negative association between laying date and post-hatching survival (Love et al. 2010), should confirm that egg value generally decreases with increasing laying date. Our second step was to test Drent & Daan (1980) and Rowe, Ludwig & Schluter (1994) optimization model through descriptive data and structural equation models as described in Fig. 2a. Then, in the third step, we used data from an experimental manipulation of pre-laying body condition (see below) to test for a causal relationship between condition at arrival, laying date and clutch size. Finally, in the fourth step, we tested the fitness costs of reproductive decisions through modelling of capture–mark–recapture data. Experiment In 2002, 2003 and 2004, we kept 112 females in outdoor cages (wooden structure surrounded by wire mesh, 1·2 × 2·5 × 1·2 m) for 24 h with water but no food. Manipulated birds were randomly chosen among all captured individuals. Thirty-five manipulated females were re-sighted as breeders (34 of known laying date and 31 of known clutch size). Some of the manipulated females may have bred following the manipulation, but remained undetected in the colony because of a high bird density and re-sighting rate 0·3). Our experiment lacks a true control group, that is, a group of females kept in captivity for 24 h with both water and food to prevent any decrease in body condition. A previous experiment (G. Gilchrist, unpublished data) indicated that eiders temporarily held in captivity do not feed even if food is provided ad libitum. Therefore, it was simply impossible to have this type of control group for such an experiment. However, it is important to note that ‘unmanipulated’ females were also kept in captivity after capture for c. 1 h (time needed to remove, measure and mark a group of individuals captured in the nets). This absence of this control group will be discussed in light of our results. Statistical analyses Decrease in egg-value with increasing laying date (test of model assumption) To test for a decrease in hatching success and number of hatched ducklings with increasing laying date (step 1), we performed linear mixed models with a binomial (and logit link function) or normal error distribution, respectively. We used the lmer and lme functions of software R (R Development Core Team, 2010) and included a random Year effect in each model. Inspection of residuals from the model with the number of hatchlings as a dependent variable indicated no violation of the assumption of normality (Shapiro–Wilk test, P = 0·083). Dispersion parameter from the model with hatching success as a dependent variable was close to 1 (1·28), indicating that our model did not suffer from significant over-dispersion. Path analysis To test the optimization model and thus the structural relationships between reproductive parameters, we performed a path analysis (i.e. a special case of structural equation model with no latent variable, Shipley 2000). The principle of the method is to specify how the variables are linked together in terms of direct and indirect causal effects. Figure 2a shows the expected causal relationships based on Drent & Daan (1980) and Rowe, Ludwig & Schluter (1994) model (Fig. 1). We first tested whether the direct causal relationships defined in this model were significant or not. To do so, and obtain path coefficients, we performed linear mixed models to regress each variable on its direct causes using the lme function of R software (R Development Core Team, 2010). A random Year effect was included in each model. Then, to test the validity of our causal model as a whole, we performed simultaneous tests of all independence claims, known as a directional-separation test (d-sep tests of path models, Shipley 2009). The validity of the model is based on a statistic C that follows a chi-square distribution with 2k degrees of freedom, k being the number of independence claims (see details in Appendix S2). Manipulation of pre-laying body condition To test for an effect of our body condition manipulation on the delay before laying and then clutch size, we performed linear mixed models with a normal error distribution, using the lme function of R software (R Development Core Team, 2010) with a random Year effect. Inspection of residual distribution indicated no departure from normality when considering the model with delay before laying as dependent variable (Shapiro–Wilk test: P = 0·088), and a moderate departure from normality when considering the model with clutch size as the dependent variable (Shapiro–Wilk test: P = 0·021). Results were the same for this later model when considering a Poisson error distribution (and indicated no over-dispersion, 0·27). Therefore, the ‘iii’ assumption (i.e. independence of fates and identity of rates among individuals), required for CMR analyses was met. As non-breeders rarely come to the colony before the hatching period (mid/end of July) and because re-sighting mostly occurred early in the season, a vast majority of females re-sighted at the colony were likely breeding females or failed breeders. As a consequence, re-sighting probability at the colony should represent a good proxy of breeding probability. To test for a negative association between clutch size in year t and survival between years t and t + 1 or breeding probability in year t + 1, we tested for a clutch-size effect using clutch size as an individual covariate. Clutch size was available only in the year of first capture, so we tested for an effect of clutch size on survival only for the year following banding. We started our model selection from a general model including a year (t) effect and two ‘age-classes’ (i.e. year following capture and 1 year after capture onwards; model ). We considered the same sample of females used in previous steps, but restricted our sample to years 2002–2005. Indeed, from 2006 onwards, severe avian cholera outbreaks occurred in our colony and costs of reproduction were likely greater, leading to a negative association between survival and clutch size (see Descamps et al. 2009 for details). This variation in costs of reproduction with the presence of avian cholera in the context of the condition-dependent optimization model will be discussed. Our data set used for survival analyses thus corresponded to 189 females with known clutch size, banded from 2002 to 2005 and monitored from 2002 to 2006. To test for an effect of our body condition manipulation on future survival and breeding probability, we considered two groups of individuals (manipulated and control) and tested for a group effect. Our sample consisted of 109 manipulated and 398 control females with known capture-recapture history. We started our model selection from a general model including a year effect and two groups (i.e. groups of manipulated [m] and control [c] females; model ). We considered the time period 2002–2006 which represents the period before the severe cholera outbreaks (Descamps et al. 2009). In both cases, model selection was based on the Akaike Information Criterion corrected for small sample sizes (AICc), as recommended when several non-nested models are fitted (Burnham & Anderson 2002). We used ΔAICc (difference in AICc between a given model and the model with lowest AICc) as a criterion to choose the best models among all tested models. A ΔAICc <2 between two competing models means that they cannot be distinguished in their ability to model the data (Burnham & Anderson 2002). When ΔAICc between two nested models was <2, the simplest one was selected. Results Relationship between laying date and egg value The probability of hatching at least one egg in a clutch decreased with increasing laying date (slope of −0·07 ± 0·02 SE on a logit scale; z = −3·58, P = 0·0003; Fig. 3), after controlling for clutch size (slope of the clutch-size effect: 0·34 ± 0·13 SE on a logit scale; z = 2·66, P = 0·008). The interaction between laying date and clutch size was not significant (P = 0·84) and was therefore not included in the previous model. Among successful nests with a known number of hatchlings, the number of hatched offspring in a given clutch did not vary with laying date (slope of 0·004 ± 0·022 SE; t = 0·18, P = 0·86), after controlling for clutch size (slope of 0·63 ± 0·13 SE; t = 4·70, P < 0·001). Again, the interaction between laying date and clutch size was not significant (P = 0·93) and was not included in the previous model. Figure 3Open in figure viewerPowerPoint Hatching success (probability to hatch ≥1 egg) as a function of laying date (expressed in days since January 1) in a common eider colony, East Bay, Southampton Island, Nunavut, Canada (n = 318; for the sake of clarity, data have been pooled for laying dates ≥190 and ≤170). Test of the condition-dependent optimization model: a path analysis approach Three structural equations linking date of arrival, condition at arrival, delay before laying, laying date and clutch size can be derived from Fig. 2, but only two are of interest. Indeed, the delay before laying was calculated as (laying date-arrival date) so that the structural equation linking laying date to arrival date and delay before laying is simply Laying date = Arrival date + Delay before laying (i.e. the path coefficient is equal to 1). The two other structural equations obtained from linear mixed models are: Numbers in brackets represent the standard errors of the path coefficients. All path coefficients are significantly different from 0 at the 0·001 level. 1|Year represents the random year effect. Body condition at arrival and arrival date (and the random year effect) explained 39% of the variation in delay before laying, and laying date (and the random year effect) explained 13% of variation in clutch size (calculated as , where represents the average value for the trait considered, the predicted value for in