General and Comparative Endocrinology 204 (2014) 95–103
Contents lists available at ScienceDirect
General and Comparative Endocrinology journal homepage: www.elsevier.com/locate/ygcen
Response of testosterone and corticosterone plasma levels to the challenge of sibling competition: A study in common terns Alexander Braasch a,⇑, Peter H. Becker a, Ton G.G. Groothuis b a b
Institute of Avian Research ‘‘Vogelwarte Helgoland’’, An der Vogelwarte 21, D-26386 Wilhelmshaven, Germany Behavioural Biology, Centre for Life Sciences, University of Groningen, Nijenborgh 7, 9747 AG Groningen, The Netherlands
a r t i c l e
i n f o
Article history: Received 6 July 2013 Revised 3 May 2014 Accepted 5 May 2014 Available online 21 May 2014 Keywords: Begging behavior Aggression Hatching order Sibling rivalry Dominance hierarchy
a b s t r a c t The hormonal response to social challenges has been widely studied, however, most work focused on adult behavior in a reproductive context although developing animals also encounter important social challenges early in life. We studied the relationship between acute sibling competition and plasma corticosterone (CORT) and testosterone (T) in common tern (Sterna hirundo) chicks, a species whose young compete for access to food by scramble interactions. Blood samples were taken in nests with two and only one single chick both immediately after a feeding bout and in non-challenged controls. We found that T levels were lower in siblings challenged by a feeding bout as compared to controls, which may be explained by the fact that T suppresses begging behavior and is only elevated in response to territorial intrusion but not sibling competition in a related species. Singletons had, corrected for body condition, generally lower CORT levels than siblings suggesting that growing up with siblings creates a competitive environment in which high CORT levels are sustained irrespective of a social challenge. CORT levels were also negatively correlated with body condition and were higher in males than in females. The latter may be related to sex-specific food requirements and susceptibility to stress. Our results suggest a possible suppressive effect of acute sibling competition on T secretion, and a positive effect on CORT levels by longer term sibling competition. The degree to which these dynamics are related to begging or aggression, or both, needs further experimental work. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction In many vertebrate species steroid hormones are involved in the organization and regulation of social behavior (Adkins-Regan, 2005). Studies on the relationship between steroid hormones and social behavior mostly concern the role of androgens in a competitive context, such as forming dominance hierarchies, territorial conflict and reproductive interactions (Soma, 2006). Since such competitive episodes are usually short-term, and long-term elevation of steroid hormones may have costs (for a review on androgens see Wingfield et al. 2001; for glucocorticoids see Korte et al. 2005), plastic secretion of these hormones may function as a powerful tool that animals can use to adapt to varying environmental conditions (Dufty et al., 2002). In this context, most studies have focused on agonistic adult behavior in a reproductive context (e.g. ‘‘Challenge Hypothesis’’, Wingfield et al., 1990), while our knowledge about the hormonal control of competitive behavior
⇑ Corresponding author. Fax: +49 4421 968955. E-mail address: [email protected] (A. Braasch). http://dx.doi.org/10.1016/j.ygcen.2014.05.007 0016-6480/Ó 2014 Elsevier Inc. All rights reserved.
early in ontogeny is limited, perhaps due to the assumption that androgen production is insignificant in sexually immature animals. Birds serve as good models for studying behavioral endocrinology before independence because they are easy to observe and show clear sibling competition (Mock and Parker, 1997). Moreover, the endogenous secretion of steroid hormones can already be substantial very early during ontogeny (e.g. Adkins-Regan et al., 1990; Ottinger et al., 2001; Ros et al., 2002). In many avian taxa, the chick period is the most competitive period in the life cycle (Wright and Leonard, 2002), having major implications on survival prospects (Mock and Parker, 1997). Nestlings mostly compete by begging to parents and scrambling for resources, and this begging, as well as early aggression within and between broods, is at least in some species affected by either androgens or corticosterone (CORT), the principal stress hormone in birds (see below for details). As far as androgens are concerned, testosterone (T) was suggested to be a major candidate in regulating the expression of competitive behaviors in nestling birds (reviewed in Ros 2008; Smiseth et al. 2011). However, the relationships among T and begging, competition for food, and aggressive interactions have been shown to be inconsistent and differ between species, which
96
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
may either depend on the developmental mode (altricial vs. semiprecocial), the method used or on the social context (see also Boncoraglio and Groothuis 2013). For example, in the altricial pied flycatcher (Ficedula hypoleuca), T facilitates begging (Goodship and Buchanan, 2007, 2006), while in the spotless starling (Sturnus unicolor) broods with increased sibling competition displayed lower T levels (Gil et al., 2008). In the semi-precocial black-headed gull (Croicocephalus ridibundus) chicks, T inhibits begging and growth (Groothuis and Ros, 2005), whereas T levels were temporarily elevated during conspecific territory defense under seminatural conditions (Ros et al., 2002), suggesting a trade-off between benefits and costs of endogenous T in this species. In the obligate siblicidal and altricial Nazca Booby (Sula granti), both the larger and smaller nest mates show elevated T levels during fights, but not before and after fights, and not in chicks lacking a nest mate (Ferree et al., 2004). In the closely related blue-footed booby (Sula nebouxii), a facultative siblicidal species, dominant chicks did however not show higher T levels than subdominant chicks (Nuñez-de la Mora et al., 1996; Ramos-Fernandez et al., 2000). Although often neglected, many types of sibling competition may be simultaneously competitive and stressful and thus possible co-variation with CORT should be considered (Ros, 2008). Moreover, there is also experimental evidence that CORT has a direct effect on begging and aggression in black-legged kittiwake (Rissa tridactyla) chicks, as CORT-implanted individuals begged more frequently and were more aggressive than controls (Kitaysky et al., 2003). Further studies in developing birds point in the same direction as they have demonstrated that variations in baseline and acute stress-induced levels of CORT are associated with increased begging rate and activity (Kitaysky et al., 2001b; Loiseau et al., 2008; Wada and Breuner, 2008), although the results for altricial birds are mixed (reviewed in Smiseth et al. 2011). Studies on siblicidal species, suggest that CORT may regulate sibling dominancesubordination relationships (Nuñez-de la Mora et al., 1996; Ramos-Fernandez et al., 2000; Tarlow et al., 2001) or intra-brood hierarchies (Love et al., 2003; Müller et al., 2010). Unfortunately, few studies have addressed the relationship between chick competition and both T and CORT in the same species during the same interaction (but see Gil et al. 2008; Nuñez-de la Mora et al. 1996; Ramos-Fernandez et al. 2000). In the present study, we test the effect of short-term competition in which siblings compete for food delivered by the parents, and long term effects (dominance hierarchy in the brood) on the dynamics of both T and CORT in chicks of the common tern (Sterna hirundo). This species serves as a suitable study species: the semi-precocial young are mobile within a few days after hatching and grow up together (one to three siblings) in a relatively short time span in which competition for food is intense, especially during the last days pre-fledging (Becker and Ludwigs, 2004). Moreover, terns are related to black-headed gulls in which T suppressed begging and was unrelated to within-brood competition (Groothuis and Ros, 2005), and neither species show overt within-brood aggression. However, common tern chicks show scramble competition over a monopolizable food source resulting in a clear ‘‘winner’’ and ‘‘loser’’ outcome of the interaction. This situation might be comparable with some altricial young, in which parents may provide a non-partitioned food item to one chick, based on competitive T-dependent begging behavior (Goodship and Buchanan, 2007, 2006; Smiseth et al., 2011). Both are in contrast to the situation in black-headed gull chicks, whose parents regurgitate food on the ground that cannot be monopolized and, as a result, the majority or all of the chicks within the brood obtain a share of the food; as such, begging may be viewed as a ‘cooperative’ endeavor that is T-independent (Groothuis and Ros, 2005).
Focal sibling broods were sampled at the time all nest mates were observed to be engaged intensely in scramble interactions during feeding, including vigorous begging, and pecking and pulling the food item. As a control, we included both non-challenged sibling broods that were sampled outside the agonistic context (no feeding bout, named ‘‘non-challenged’’ condition hereafter) and single-chick broods (singletons). Given the inconsistency in findings on the relationship between T and agonistic behavior in chicks, we did not formulate a clear prediction for T. Assuming that scramble interactions are simultaneously competitive and stressful, we hypothesized that CORT should be higher in challenged sibling broods as compared to non-challenged siblings. In the absence of an opponent, singletons neither have to share resources (less hunger-induced stress) nor are they engaged in within-brood scramble interactions. Thus, we predict singletons should have lower CORT levels than siblings. Additionally, we were interested whether (a) the outcome of scramble competition is related to T and CORT, (b) competitive asymmetries between siblings established by an asynchronous hatching interval are reflected in their hormonal response, and (c) the two sexes respond differentially to social challenges. 2. Material and methods 2.1. Study species Common terns have broods of two or three chicks that hatch asynchronously over a 2–3 day period (Bollinger, 1994). Because the thermoregulatory ability of newly hatched chicks is underdeveloped, they depend on parental brooding during the first few days post-hatch, and thus generally remain close to the nest and one another. However, as chicks’ mobility increases, they move further from the nest, seeking cover within a small territory which they defend against intruders (conspecifics, both adults and chicks from neighboring broods) until fledging (Becker and Ludwigs, 2004; Langham, 1972). Chicks are able to discriminate between siblings and non-familiar young within a few days after hatching (Palestis and Burger, 1999). A size-based within-brood hierarchy is caused by hatching asynchrony (Bollinger, 1994; Bollinger et al., 1990), and competitive abilities are reflected by position in the hatching sequence. As in many other larids, chick mortality is highest during the first week of life, often to the disadvantage of the last-hatched chick (Bollinger et al., 1990; Langham, 1972). Typically, adults are single-load feeders carrying only one prey item, mostly a small fish or crustacean (Becker and Ludwigs, 2004), back to the colony. Consequently, each feeding attempt represents a non-shareable food resource that could be monopolized by only one chick. Once they hear the parent approaching with food, common tern chicks exhibit a conspicuous begging display involving both calls and postural changes. Outrunning their siblings and trying to reach the parent first as it lands seems to be advantageous, because in most cases feedings are directed to the first chick to reach the parent (Smith et al., 2005). As aggressive intrabrood interactions, involving non-lethal pecking and biting of siblings, are extremely rare and almost exclusively restricted to feeding events (Palestis and Burger, 1999; Braasch et al., unpublished data), intrabrood aggression is considered to be mainly a result of competition for food (Palestis and Burger, 1999). 2.2. Study site and general field work procedures Field work was conducted between May and July 2008 in a monospecific common tern colony of about 380 breeding pairs
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
situated in the harbor area of Wilhelmshaven, Lower Saxony, Germany (53°300 4000 N, 8°060 2000 E). The colony site consists of six equally sized rectangular islands (10.7 4.6 m) arranged in a line 0.9 m apart and surrounded by concrete walls (60 cm high) avoiding chick losses by drowning. Two permanently fixed hides provided suitable sampling facility in close proximity to focal broods. Being the subject of a long-term population study, this colony has been under investigation since the early 1980s (for details see Becker and Wendeln 1997; Becker et al. 2008, 2001). We individually marked new nests with numbered stakes and checked their contents daily during egg-laying and hatching periods. Thereafter, the fate of each brood was followed during regular control checks every two or three days. On the day a chick was found for the first time, it was banded and its age was designated as 0 if the chick was freshly hatched, otherwise as day 1. First-, second-, and third-hatched chicks were denoted as a-, b- and c-chicks, respectively. Usually, hatching order corresponds with laying order (Bollinger, 1994), however, hatching spread is shorter than laying interval (Nisbet and Cohen, 1975), and thus occasionally two chicks can be found in a nest on the same day. In such cases information on hatching order can be derived by the size of the remains of the yolk sac on the chicks’ belly, which diminishes with age (Wagener, 1998). During regular demographic monitoring a subsample of the cohort (on two of six islands) was weighed to the nearest gram by means of a digital balance. The sex of all focal chicks was determined by using body feathers and standard PCR methods (for details see Becker and Wink 2003). 2.3. Experimental design and behavioral observations To test the effects of a social challenge (competition during a feeding bout) on T and CORT levels of siblings we temporarily enclosed (on average 3 h) sibling broods and singletons with wire netting 30 cm high and approximately 1 m in diameter. Using enclosures allowed us to control for territorial intrusion by neighboring broods and kleptoparasitic attacks from older chicks or adults (Oswald et al., 2012). Before setting up the enclosures the focal broods were observed to gather information about their preferred territory. To discriminate between nest mates, chicks were individually marked with color-tape on head and back. The enclosures were set up at least one hour before the experiment started so that the chicks got used to the new environment. The experimental broods were observed from either one of the two hides situated within the colony to ensure that parental provisioning was not inhibited by the enclosures. As we have assured that our setup had no obvious negative effects on behavior of both parents and offspring we proceeded with observations. Once a parent approached the enclosure with food, feedings as well as behavioral parameters (begging, aggression, the outcome of scramble competition) were recorded. In order to categorize the data in two clearly different contexts with respect to sibling competition, we analyzed the data of the following two groups: If all siblings were actively engaged in scramble interactions displaying the most intense form of vigorous begging and locomotive activities, such as wing flapping and jumping with body full erect (cf. Smith et al., 2005), those broods were chosen as challenged broods. Enclosed sibling broods not engaged in scramble interactions during our observation period were sampled outside a feeding bout and considered as ‘‘non-challenged’’ controls. Correspondingly, singletons were sampled directly after (n = 6) and outside a feeding event (n = 7). All observations were conducted by only one person so that observer bias could be ruled out. In general, all measurements from focal broods were taken during the end of the pre-fledging period, a phase of high food demands (Klaassen et al., 1992). Possibly due to limited food supply (Dänhardt and Becker, 2011), brood reduction reduced
97
three-chick broods (broods sampled: n = 1) by that time, consequently the focus was on two-chick broods (challenged: n = 9; non-challenged: n = 9). Of the one three-chick brood sampled we considered only a- and b-chick. All single-chick broods (8 ‘‘achicks’’, 5 ‘‘b-chicks’’) were initially multi-nestling broods that have been reduced by egg and/or chick losses prior to the experiment. Sibling broods were sampled at an age of 18.4 d ± 0.7 (n = 19, range: 15–26 d), which was determined according to the age of the a-chick (b-chick was on average 0.8 d ± 0.2, n = 19, range: 0–2 d, younger than a-chick). Single-chick broods were on average 19.3 d ± 0.5 (n = 13, range: 16–22 d) old when sampled. 2.4. Sampling procedure and chick measurements We focused on appropriately timed blood samples taken shortly after a feeding bout. After the food item was monopolized we waited another 5 min until chick removal, which also applies to singletons being fed. As feeding events and chick handling took a few minutes, timing of our blood collection was approximately 10 min after the start of the feeding bout. This was considered because earlier studies have shown that plasma T levels increase substantially after about 10 min after the start of the challenge (e.g. Ros et al., 2002). Control broods were not influenced by a parental feeding attempt prior to nest removal for on average 81 min (singletons: 49.7 ± 10.1 min, n = 7, range: 24–96 min; two-chick broods: 93.7 ± 20.6 min, n = 9, range: 18–290 min), thus we assumed to measure baseline levels of both plasma T and CORT. Upon capture, siblings were kept in separate boxes to avoid getting in physical contact with each other and were handled in random order thereafter, on average in less than 2 min (handling time: 1.9 ± 0.1 min, n = 37). However, as siblings were collected simultaneously we also registered the time of initial disturbance at the nest until finishing the blood withdrawal (capture time: 4.5 ± 0.4 min, n = 37) and controlled for capture time in the statistical analysis (see Section 2.7.). After taking a small blood sample (200 ll), obtained by puncturing the superficial brachial wing vein with a sterile 23- to 27-gauge needle, we measured wing length and body mass (digital balance, accuracy 1 g). Blood was collected in self-sealing, heparinized hematocrit microcapillary tubes (capacity 75 ll) and kept in a cool bag until (within 1 h) transferred to the laboratory and centrifuged at 10,000 rpm for 8 min at 4 °C. Plasma was separated and then stored at 20 °C until proceeding with laboratory analyses at the University of Groningen, Netherlands. 2.5. Body condition index As an approximation for condition, we calculated the condition index according to Stienen and Brenninkmeijer (2002) using the formula
CI ¼
M Mexp 100% M exp
The condition index (CI %) obtained is the age-specific ratio of the deviation of the measured body mass (M) at sampling from the expected body mass (Mexp), which is the arithmetic mean of all chicks of a specific age weighed regularly (on two of six islands) until fledging successfully (Becker and Wink, 2003). Body condition values as a function of hatching order and sex are given in Table 1. We did not correct by structural size (wing length) as common tern chicks reach peak mass followed by a mass loss prior to fledging; thus body condition indices relative to age are more appropriate than mass controlled for measures of structural size. 2.6. Hormone assays Plasma concentrations of T and CORT were quantified by radioimmunoassay (RIA). To extract the hormones, 52–458 mg of
98
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
Table 1 Body condition of sibling common tern chicks in relation to hatching order (achick = first-hatched sibling; b-chick = second-hatched sibling) and sex. Given are means ± SE, (N). Sex
# $ Total
Hatching order a-chick
b-chick
Singleton
0.3 ± 2.5 (7) 5.6 ± 4.8 (12) 3.6 ± 3.1 (19)
10.0 ± 6.4 (9) 12.7 ± 6.8 (10) 11.4 ± 4.6 (19)
8.3 ± 8.8 (5) 7.6 ± 3.4 (8) 1.5 ± 4.3 (13)
see Text 2.5 for definition and calculation of body condition index.
plasma was weighed (accuracy 0.001 g). Twenty ll of 3H-labelled testosterone was added to trace the recovery of extracted hormones during the extraction procedure (total count 1992 CPM). This solution was incubated for 15 min at 37 °C before being extracted in 2 ml of diëthylether/petroleum ether (DEE/PE, 70/30 v/v) by vortexing for 60 s. Extracted samples were centrifuged at 2000 rpm for 3 min (4 °C) to separate the ether phase, the samples were snap-frozen and the ether/hormone phase decanted into a fresh tube. The extraction procedure was repeated with an additional 2 ml of DEE/PE and vortexed for 30 s. Next, the extracts were dried under nitrogen. Hormone extracts were rinsed in 1.5 ml of 70% methanol to precipitate any lipids and stored at least overnight at 20 °C. Subsequently, the tubes were centrifuged, decanted into a fresh tube, re-dried under nitrogen and stored at 20 °C. Prior to assay, extracts were thawed and dissolved in 100 ll of phosphate-buffered-saline with gelatin (PBSG). Subsequently, 50 ll of sample was used for T determination using a kit purchased from DSL Benelux Office, (‘DSL-4000’, Sinsheim, Germany). For CORT determination 100 ll of sample (11.3 dilution) using a kit purchased from DSL Benelux Office, (‘RS49011’, MP Biomedicals, Ohio, USA). Standards were prepared using dilution series from pre-prepared stock and ranged from 0.04 to 20 ng ml1 for T and 25 to 1000 ng ml1 for CORT. Calculation of the recoveries failed because although we obtained the data for the individual samples, those for the 100% recoveries, needed for final calculation, were lost. We therefore calculated the average CPM recovery over all individual samples and set this to the value of the averaged recoveries of the same hormone in 4 same extractions using the same label that we recently performed. This amounted to 4980 CPM (stdev. 6.6%). Given the small standard deviation this is a reliable method. Moreover, the correlation between the amounts of T corrected and not corrected for recoveries was very high (Pearson r = 0.97), indicating that calculation of the recoveries (average: 79%, stdev. 4.8%) would not influence our further statistical analyses. Indeed, both data sets yielded similar results. The intra-assay coefficient of variation was 1.9% for T and 1.0% for CORT.
2.7. Statistics Sibling broods were analyzed using Linear Mixed Effects (LME) models with a Restricted Maximum Likelihood Estimator (REML) for parameter estimation (Pinheiro and Bates, 2000). This procedure can produce unbiased estimates of variance and covariance parameters, can handle unbalanced data sets, and it offers the opportunity to incorporate random effects. We considered that chicks from the same brood were not independent and, hence, controlled for a potential effect of sharing the same parents by including brood identity (brood-id) as a random effect (random intercept term) in the analysis. First, we analyzed the subset of challenged sibling broods to assess whether variations in T and CORT levels were related to the outcome of the interaction (‘‘winning or losing’’). Additionally,
the model included hatching order and sex as fixed factors, whilst body condition and capture time were controlled for by including both as covariates. We also tested whether hatching order, sex and/or body condition influenced ‘‘winning or losing’’ by fitting the respective interaction terms. Secondly, we used simple Pearson’s correlations to analyze whether the time elapsed after the last feeding (‘‘feeding time’’) was associated with T and CORT levels using the subset of broods sampled in the non-challenged condition. To analyze the effect of challenge, we used the same model as described above in which we replaced the factor ‘‘winning or losing’’ with challenge (Yes/No), now using the data of sibling broods in both the challenged and non-challenged condition (for details see Table 2). To avoid a confounding influence of singletons, the effect of brood size (two-level factor: multi-nestling brood vs. singleton) was tested separately in a model that included brood size, challenge, body condition and their interactions, and capture time. If not otherwise stated, we followed a stepwise backward model simplification procedure by removing non-significant terms starting with the interactions to reach the most parsimonious model. We tested the fit of our models by checking the residuals for normality and homoscedasticity and by plotting the residuals against the predicted values. As neither plasma T nor CORT values were normally distributed, they were (1+)log-transformed to meet parametric test assumptions. Sample sizes are reported throughout the text because they varied slightly between different subsets as data were unbalanced due to missing values in some categories. All analyses were conducted in the R environment version 2.12.2 (R Development Core Team, 2011) using the function ‘lme’ to fit LME models (‘‘nlme’’ package version 3.1-102; Pinheiro et al., 2011). The level of significance was set at a 6 0.05, and data are presented as mean ± standard error (SE).
Table 2 Results of a Linear Mixed Effects (LME) model analyzing the effects of hatching order, sex, and social challenge on plasma testosterone and corticosterone levels of common tern chicks. Term
Testosterone Fixed effects Intercept Hatching order (factor) Sex (factor) Challenge (factor) Hatching order challenge (factor) Sex challenge (factor) Body condition (covariate) Capture time (covariate) Corticosterone Fixed effects Intercept Hatching order (factor) Sex (factor) Challenge (factor) Hatching order challenge (factor) Sex challenge (factor) Body condition (covariate) Capture time (covariate)
Statistics F
df
P
b(SE)
2499.2 0.678 0.015 5.540 1.069
1,18 1,15 1,14 1,17 1,13
Contents lists available at ScienceDirect
General and Comparative Endocrinology journal homepage: www.elsevier.com/locate/ygcen
Response of testosterone and corticosterone plasma levels to the challenge of sibling competition: A study in common terns Alexander Braasch a,⇑, Peter H. Becker a, Ton G.G. Groothuis b a b
Institute of Avian Research ‘‘Vogelwarte Helgoland’’, An der Vogelwarte 21, D-26386 Wilhelmshaven, Germany Behavioural Biology, Centre for Life Sciences, University of Groningen, Nijenborgh 7, 9747 AG Groningen, The Netherlands
a r t i c l e
i n f o
Article history: Received 6 July 2013 Revised 3 May 2014 Accepted 5 May 2014 Available online 21 May 2014 Keywords: Begging behavior Aggression Hatching order Sibling rivalry Dominance hierarchy
a b s t r a c t The hormonal response to social challenges has been widely studied, however, most work focused on adult behavior in a reproductive context although developing animals also encounter important social challenges early in life. We studied the relationship between acute sibling competition and plasma corticosterone (CORT) and testosterone (T) in common tern (Sterna hirundo) chicks, a species whose young compete for access to food by scramble interactions. Blood samples were taken in nests with two and only one single chick both immediately after a feeding bout and in non-challenged controls. We found that T levels were lower in siblings challenged by a feeding bout as compared to controls, which may be explained by the fact that T suppresses begging behavior and is only elevated in response to territorial intrusion but not sibling competition in a related species. Singletons had, corrected for body condition, generally lower CORT levels than siblings suggesting that growing up with siblings creates a competitive environment in which high CORT levels are sustained irrespective of a social challenge. CORT levels were also negatively correlated with body condition and were higher in males than in females. The latter may be related to sex-specific food requirements and susceptibility to stress. Our results suggest a possible suppressive effect of acute sibling competition on T secretion, and a positive effect on CORT levels by longer term sibling competition. The degree to which these dynamics are related to begging or aggression, or both, needs further experimental work. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction In many vertebrate species steroid hormones are involved in the organization and regulation of social behavior (Adkins-Regan, 2005). Studies on the relationship between steroid hormones and social behavior mostly concern the role of androgens in a competitive context, such as forming dominance hierarchies, territorial conflict and reproductive interactions (Soma, 2006). Since such competitive episodes are usually short-term, and long-term elevation of steroid hormones may have costs (for a review on androgens see Wingfield et al. 2001; for glucocorticoids see Korte et al. 2005), plastic secretion of these hormones may function as a powerful tool that animals can use to adapt to varying environmental conditions (Dufty et al., 2002). In this context, most studies have focused on agonistic adult behavior in a reproductive context (e.g. ‘‘Challenge Hypothesis’’, Wingfield et al., 1990), while our knowledge about the hormonal control of competitive behavior
⇑ Corresponding author. Fax: +49 4421 968955. E-mail address: [email protected] (A. Braasch). http://dx.doi.org/10.1016/j.ygcen.2014.05.007 0016-6480/Ó 2014 Elsevier Inc. All rights reserved.
early in ontogeny is limited, perhaps due to the assumption that androgen production is insignificant in sexually immature animals. Birds serve as good models for studying behavioral endocrinology before independence because they are easy to observe and show clear sibling competition (Mock and Parker, 1997). Moreover, the endogenous secretion of steroid hormones can already be substantial very early during ontogeny (e.g. Adkins-Regan et al., 1990; Ottinger et al., 2001; Ros et al., 2002). In many avian taxa, the chick period is the most competitive period in the life cycle (Wright and Leonard, 2002), having major implications on survival prospects (Mock and Parker, 1997). Nestlings mostly compete by begging to parents and scrambling for resources, and this begging, as well as early aggression within and between broods, is at least in some species affected by either androgens or corticosterone (CORT), the principal stress hormone in birds (see below for details). As far as androgens are concerned, testosterone (T) was suggested to be a major candidate in regulating the expression of competitive behaviors in nestling birds (reviewed in Ros 2008; Smiseth et al. 2011). However, the relationships among T and begging, competition for food, and aggressive interactions have been shown to be inconsistent and differ between species, which
96
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
may either depend on the developmental mode (altricial vs. semiprecocial), the method used or on the social context (see also Boncoraglio and Groothuis 2013). For example, in the altricial pied flycatcher (Ficedula hypoleuca), T facilitates begging (Goodship and Buchanan, 2007, 2006), while in the spotless starling (Sturnus unicolor) broods with increased sibling competition displayed lower T levels (Gil et al., 2008). In the semi-precocial black-headed gull (Croicocephalus ridibundus) chicks, T inhibits begging and growth (Groothuis and Ros, 2005), whereas T levels were temporarily elevated during conspecific territory defense under seminatural conditions (Ros et al., 2002), suggesting a trade-off between benefits and costs of endogenous T in this species. In the obligate siblicidal and altricial Nazca Booby (Sula granti), both the larger and smaller nest mates show elevated T levels during fights, but not before and after fights, and not in chicks lacking a nest mate (Ferree et al., 2004). In the closely related blue-footed booby (Sula nebouxii), a facultative siblicidal species, dominant chicks did however not show higher T levels than subdominant chicks (Nuñez-de la Mora et al., 1996; Ramos-Fernandez et al., 2000). Although often neglected, many types of sibling competition may be simultaneously competitive and stressful and thus possible co-variation with CORT should be considered (Ros, 2008). Moreover, there is also experimental evidence that CORT has a direct effect on begging and aggression in black-legged kittiwake (Rissa tridactyla) chicks, as CORT-implanted individuals begged more frequently and were more aggressive than controls (Kitaysky et al., 2003). Further studies in developing birds point in the same direction as they have demonstrated that variations in baseline and acute stress-induced levels of CORT are associated with increased begging rate and activity (Kitaysky et al., 2001b; Loiseau et al., 2008; Wada and Breuner, 2008), although the results for altricial birds are mixed (reviewed in Smiseth et al. 2011). Studies on siblicidal species, suggest that CORT may regulate sibling dominancesubordination relationships (Nuñez-de la Mora et al., 1996; Ramos-Fernandez et al., 2000; Tarlow et al., 2001) or intra-brood hierarchies (Love et al., 2003; Müller et al., 2010). Unfortunately, few studies have addressed the relationship between chick competition and both T and CORT in the same species during the same interaction (but see Gil et al. 2008; Nuñez-de la Mora et al. 1996; Ramos-Fernandez et al. 2000). In the present study, we test the effect of short-term competition in which siblings compete for food delivered by the parents, and long term effects (dominance hierarchy in the brood) on the dynamics of both T and CORT in chicks of the common tern (Sterna hirundo). This species serves as a suitable study species: the semi-precocial young are mobile within a few days after hatching and grow up together (one to three siblings) in a relatively short time span in which competition for food is intense, especially during the last days pre-fledging (Becker and Ludwigs, 2004). Moreover, terns are related to black-headed gulls in which T suppressed begging and was unrelated to within-brood competition (Groothuis and Ros, 2005), and neither species show overt within-brood aggression. However, common tern chicks show scramble competition over a monopolizable food source resulting in a clear ‘‘winner’’ and ‘‘loser’’ outcome of the interaction. This situation might be comparable with some altricial young, in which parents may provide a non-partitioned food item to one chick, based on competitive T-dependent begging behavior (Goodship and Buchanan, 2007, 2006; Smiseth et al., 2011). Both are in contrast to the situation in black-headed gull chicks, whose parents regurgitate food on the ground that cannot be monopolized and, as a result, the majority or all of the chicks within the brood obtain a share of the food; as such, begging may be viewed as a ‘cooperative’ endeavor that is T-independent (Groothuis and Ros, 2005).
Focal sibling broods were sampled at the time all nest mates were observed to be engaged intensely in scramble interactions during feeding, including vigorous begging, and pecking and pulling the food item. As a control, we included both non-challenged sibling broods that were sampled outside the agonistic context (no feeding bout, named ‘‘non-challenged’’ condition hereafter) and single-chick broods (singletons). Given the inconsistency in findings on the relationship between T and agonistic behavior in chicks, we did not formulate a clear prediction for T. Assuming that scramble interactions are simultaneously competitive and stressful, we hypothesized that CORT should be higher in challenged sibling broods as compared to non-challenged siblings. In the absence of an opponent, singletons neither have to share resources (less hunger-induced stress) nor are they engaged in within-brood scramble interactions. Thus, we predict singletons should have lower CORT levels than siblings. Additionally, we were interested whether (a) the outcome of scramble competition is related to T and CORT, (b) competitive asymmetries between siblings established by an asynchronous hatching interval are reflected in their hormonal response, and (c) the two sexes respond differentially to social challenges. 2. Material and methods 2.1. Study species Common terns have broods of two or three chicks that hatch asynchronously over a 2–3 day period (Bollinger, 1994). Because the thermoregulatory ability of newly hatched chicks is underdeveloped, they depend on parental brooding during the first few days post-hatch, and thus generally remain close to the nest and one another. However, as chicks’ mobility increases, they move further from the nest, seeking cover within a small territory which they defend against intruders (conspecifics, both adults and chicks from neighboring broods) until fledging (Becker and Ludwigs, 2004; Langham, 1972). Chicks are able to discriminate between siblings and non-familiar young within a few days after hatching (Palestis and Burger, 1999). A size-based within-brood hierarchy is caused by hatching asynchrony (Bollinger, 1994; Bollinger et al., 1990), and competitive abilities are reflected by position in the hatching sequence. As in many other larids, chick mortality is highest during the first week of life, often to the disadvantage of the last-hatched chick (Bollinger et al., 1990; Langham, 1972). Typically, adults are single-load feeders carrying only one prey item, mostly a small fish or crustacean (Becker and Ludwigs, 2004), back to the colony. Consequently, each feeding attempt represents a non-shareable food resource that could be monopolized by only one chick. Once they hear the parent approaching with food, common tern chicks exhibit a conspicuous begging display involving both calls and postural changes. Outrunning their siblings and trying to reach the parent first as it lands seems to be advantageous, because in most cases feedings are directed to the first chick to reach the parent (Smith et al., 2005). As aggressive intrabrood interactions, involving non-lethal pecking and biting of siblings, are extremely rare and almost exclusively restricted to feeding events (Palestis and Burger, 1999; Braasch et al., unpublished data), intrabrood aggression is considered to be mainly a result of competition for food (Palestis and Burger, 1999). 2.2. Study site and general field work procedures Field work was conducted between May and July 2008 in a monospecific common tern colony of about 380 breeding pairs
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
situated in the harbor area of Wilhelmshaven, Lower Saxony, Germany (53°300 4000 N, 8°060 2000 E). The colony site consists of six equally sized rectangular islands (10.7 4.6 m) arranged in a line 0.9 m apart and surrounded by concrete walls (60 cm high) avoiding chick losses by drowning. Two permanently fixed hides provided suitable sampling facility in close proximity to focal broods. Being the subject of a long-term population study, this colony has been under investigation since the early 1980s (for details see Becker and Wendeln 1997; Becker et al. 2008, 2001). We individually marked new nests with numbered stakes and checked their contents daily during egg-laying and hatching periods. Thereafter, the fate of each brood was followed during regular control checks every two or three days. On the day a chick was found for the first time, it was banded and its age was designated as 0 if the chick was freshly hatched, otherwise as day 1. First-, second-, and third-hatched chicks were denoted as a-, b- and c-chicks, respectively. Usually, hatching order corresponds with laying order (Bollinger, 1994), however, hatching spread is shorter than laying interval (Nisbet and Cohen, 1975), and thus occasionally two chicks can be found in a nest on the same day. In such cases information on hatching order can be derived by the size of the remains of the yolk sac on the chicks’ belly, which diminishes with age (Wagener, 1998). During regular demographic monitoring a subsample of the cohort (on two of six islands) was weighed to the nearest gram by means of a digital balance. The sex of all focal chicks was determined by using body feathers and standard PCR methods (for details see Becker and Wink 2003). 2.3. Experimental design and behavioral observations To test the effects of a social challenge (competition during a feeding bout) on T and CORT levels of siblings we temporarily enclosed (on average 3 h) sibling broods and singletons with wire netting 30 cm high and approximately 1 m in diameter. Using enclosures allowed us to control for territorial intrusion by neighboring broods and kleptoparasitic attacks from older chicks or adults (Oswald et al., 2012). Before setting up the enclosures the focal broods were observed to gather information about their preferred territory. To discriminate between nest mates, chicks were individually marked with color-tape on head and back. The enclosures were set up at least one hour before the experiment started so that the chicks got used to the new environment. The experimental broods were observed from either one of the two hides situated within the colony to ensure that parental provisioning was not inhibited by the enclosures. As we have assured that our setup had no obvious negative effects on behavior of both parents and offspring we proceeded with observations. Once a parent approached the enclosure with food, feedings as well as behavioral parameters (begging, aggression, the outcome of scramble competition) were recorded. In order to categorize the data in two clearly different contexts with respect to sibling competition, we analyzed the data of the following two groups: If all siblings were actively engaged in scramble interactions displaying the most intense form of vigorous begging and locomotive activities, such as wing flapping and jumping with body full erect (cf. Smith et al., 2005), those broods were chosen as challenged broods. Enclosed sibling broods not engaged in scramble interactions during our observation period were sampled outside a feeding bout and considered as ‘‘non-challenged’’ controls. Correspondingly, singletons were sampled directly after (n = 6) and outside a feeding event (n = 7). All observations were conducted by only one person so that observer bias could be ruled out. In general, all measurements from focal broods were taken during the end of the pre-fledging period, a phase of high food demands (Klaassen et al., 1992). Possibly due to limited food supply (Dänhardt and Becker, 2011), brood reduction reduced
97
three-chick broods (broods sampled: n = 1) by that time, consequently the focus was on two-chick broods (challenged: n = 9; non-challenged: n = 9). Of the one three-chick brood sampled we considered only a- and b-chick. All single-chick broods (8 ‘‘achicks’’, 5 ‘‘b-chicks’’) were initially multi-nestling broods that have been reduced by egg and/or chick losses prior to the experiment. Sibling broods were sampled at an age of 18.4 d ± 0.7 (n = 19, range: 15–26 d), which was determined according to the age of the a-chick (b-chick was on average 0.8 d ± 0.2, n = 19, range: 0–2 d, younger than a-chick). Single-chick broods were on average 19.3 d ± 0.5 (n = 13, range: 16–22 d) old when sampled. 2.4. Sampling procedure and chick measurements We focused on appropriately timed blood samples taken shortly after a feeding bout. After the food item was monopolized we waited another 5 min until chick removal, which also applies to singletons being fed. As feeding events and chick handling took a few minutes, timing of our blood collection was approximately 10 min after the start of the feeding bout. This was considered because earlier studies have shown that plasma T levels increase substantially after about 10 min after the start of the challenge (e.g. Ros et al., 2002). Control broods were not influenced by a parental feeding attempt prior to nest removal for on average 81 min (singletons: 49.7 ± 10.1 min, n = 7, range: 24–96 min; two-chick broods: 93.7 ± 20.6 min, n = 9, range: 18–290 min), thus we assumed to measure baseline levels of both plasma T and CORT. Upon capture, siblings were kept in separate boxes to avoid getting in physical contact with each other and were handled in random order thereafter, on average in less than 2 min (handling time: 1.9 ± 0.1 min, n = 37). However, as siblings were collected simultaneously we also registered the time of initial disturbance at the nest until finishing the blood withdrawal (capture time: 4.5 ± 0.4 min, n = 37) and controlled for capture time in the statistical analysis (see Section 2.7.). After taking a small blood sample (200 ll), obtained by puncturing the superficial brachial wing vein with a sterile 23- to 27-gauge needle, we measured wing length and body mass (digital balance, accuracy 1 g). Blood was collected in self-sealing, heparinized hematocrit microcapillary tubes (capacity 75 ll) and kept in a cool bag until (within 1 h) transferred to the laboratory and centrifuged at 10,000 rpm for 8 min at 4 °C. Plasma was separated and then stored at 20 °C until proceeding with laboratory analyses at the University of Groningen, Netherlands. 2.5. Body condition index As an approximation for condition, we calculated the condition index according to Stienen and Brenninkmeijer (2002) using the formula
CI ¼
M Mexp 100% M exp
The condition index (CI %) obtained is the age-specific ratio of the deviation of the measured body mass (M) at sampling from the expected body mass (Mexp), which is the arithmetic mean of all chicks of a specific age weighed regularly (on two of six islands) until fledging successfully (Becker and Wink, 2003). Body condition values as a function of hatching order and sex are given in Table 1. We did not correct by structural size (wing length) as common tern chicks reach peak mass followed by a mass loss prior to fledging; thus body condition indices relative to age are more appropriate than mass controlled for measures of structural size. 2.6. Hormone assays Plasma concentrations of T and CORT were quantified by radioimmunoassay (RIA). To extract the hormones, 52–458 mg of
98
A. Braasch et al. / General and Comparative Endocrinology 204 (2014) 95–103
Table 1 Body condition of sibling common tern chicks in relation to hatching order (achick = first-hatched sibling; b-chick = second-hatched sibling) and sex. Given are means ± SE, (N). Sex
# $ Total
Hatching order a-chick
b-chick
Singleton
0.3 ± 2.5 (7) 5.6 ± 4.8 (12) 3.6 ± 3.1 (19)
10.0 ± 6.4 (9) 12.7 ± 6.8 (10) 11.4 ± 4.6 (19)
8.3 ± 8.8 (5) 7.6 ± 3.4 (8) 1.5 ± 4.3 (13)
see Text 2.5 for definition and calculation of body condition index.
plasma was weighed (accuracy 0.001 g). Twenty ll of 3H-labelled testosterone was added to trace the recovery of extracted hormones during the extraction procedure (total count 1992 CPM). This solution was incubated for 15 min at 37 °C before being extracted in 2 ml of diëthylether/petroleum ether (DEE/PE, 70/30 v/v) by vortexing for 60 s. Extracted samples were centrifuged at 2000 rpm for 3 min (4 °C) to separate the ether phase, the samples were snap-frozen and the ether/hormone phase decanted into a fresh tube. The extraction procedure was repeated with an additional 2 ml of DEE/PE and vortexed for 30 s. Next, the extracts were dried under nitrogen. Hormone extracts were rinsed in 1.5 ml of 70% methanol to precipitate any lipids and stored at least overnight at 20 °C. Subsequently, the tubes were centrifuged, decanted into a fresh tube, re-dried under nitrogen and stored at 20 °C. Prior to assay, extracts were thawed and dissolved in 100 ll of phosphate-buffered-saline with gelatin (PBSG). Subsequently, 50 ll of sample was used for T determination using a kit purchased from DSL Benelux Office, (‘DSL-4000’, Sinsheim, Germany). For CORT determination 100 ll of sample (11.3 dilution) using a kit purchased from DSL Benelux Office, (‘RS49011’, MP Biomedicals, Ohio, USA). Standards were prepared using dilution series from pre-prepared stock and ranged from 0.04 to 20 ng ml1 for T and 25 to 1000 ng ml1 for CORT. Calculation of the recoveries failed because although we obtained the data for the individual samples, those for the 100% recoveries, needed for final calculation, were lost. We therefore calculated the average CPM recovery over all individual samples and set this to the value of the averaged recoveries of the same hormone in 4 same extractions using the same label that we recently performed. This amounted to 4980 CPM (stdev. 6.6%). Given the small standard deviation this is a reliable method. Moreover, the correlation between the amounts of T corrected and not corrected for recoveries was very high (Pearson r = 0.97), indicating that calculation of the recoveries (average: 79%, stdev. 4.8%) would not influence our further statistical analyses. Indeed, both data sets yielded similar results. The intra-assay coefficient of variation was 1.9% for T and 1.0% for CORT.
2.7. Statistics Sibling broods were analyzed using Linear Mixed Effects (LME) models with a Restricted Maximum Likelihood Estimator (REML) for parameter estimation (Pinheiro and Bates, 2000). This procedure can produce unbiased estimates of variance and covariance parameters, can handle unbalanced data sets, and it offers the opportunity to incorporate random effects. We considered that chicks from the same brood were not independent and, hence, controlled for a potential effect of sharing the same parents by including brood identity (brood-id) as a random effect (random intercept term) in the analysis. First, we analyzed the subset of challenged sibling broods to assess whether variations in T and CORT levels were related to the outcome of the interaction (‘‘winning or losing’’). Additionally,
the model included hatching order and sex as fixed factors, whilst body condition and capture time were controlled for by including both as covariates. We also tested whether hatching order, sex and/or body condition influenced ‘‘winning or losing’’ by fitting the respective interaction terms. Secondly, we used simple Pearson’s correlations to analyze whether the time elapsed after the last feeding (‘‘feeding time’’) was associated with T and CORT levels using the subset of broods sampled in the non-challenged condition. To analyze the effect of challenge, we used the same model as described above in which we replaced the factor ‘‘winning or losing’’ with challenge (Yes/No), now using the data of sibling broods in both the challenged and non-challenged condition (for details see Table 2). To avoid a confounding influence of singletons, the effect of brood size (two-level factor: multi-nestling brood vs. singleton) was tested separately in a model that included brood size, challenge, body condition and their interactions, and capture time. If not otherwise stated, we followed a stepwise backward model simplification procedure by removing non-significant terms starting with the interactions to reach the most parsimonious model. We tested the fit of our models by checking the residuals for normality and homoscedasticity and by plotting the residuals against the predicted values. As neither plasma T nor CORT values were normally distributed, they were (1+)log-transformed to meet parametric test assumptions. Sample sizes are reported throughout the text because they varied slightly between different subsets as data were unbalanced due to missing values in some categories. All analyses were conducted in the R environment version 2.12.2 (R Development Core Team, 2011) using the function ‘lme’ to fit LME models (‘‘nlme’’ package version 3.1-102; Pinheiro et al., 2011). The level of significance was set at a 6 0.05, and data are presented as mean ± standard error (SE).
Table 2 Results of a Linear Mixed Effects (LME) model analyzing the effects of hatching order, sex, and social challenge on plasma testosterone and corticosterone levels of common tern chicks. Term
Testosterone Fixed effects Intercept Hatching order (factor) Sex (factor) Challenge (factor) Hatching order challenge (factor) Sex challenge (factor) Body condition (covariate) Capture time (covariate) Corticosterone Fixed effects Intercept Hatching order (factor) Sex (factor) Challenge (factor) Hatching order challenge (factor) Sex challenge (factor) Body condition (covariate) Capture time (covariate)
Statistics F
df
P
b(SE)
2499.2 0.678 0.015 5.540 1.069
1,18 1,15 1,14 1,17 1,13
