Glucocorticoid–temperature association is shaped by foraging costs in individual zebra finches

Increased concentrations of glucocorticoid hormones (GCs) are often assumed to be an indication of ‘stress’ (reviewed in Dantzer et al., 2014; Koolhaas et al., 2011). The term ‘stress’, however, has been under scientific debate since its first use in physiological and biomedical research by Hans Selye (1950), and in the field of ecology has been used to refer to different concepts including noxious stimuli, coping responses by organisms and the overstimulation of such responses that results in disease (reviewed in Koolhaas et al., 2011). Through the synthesis and release of GCs, organisms mobilize body reserves (i.e. glucose, fatty acids and proteins; Remage-Healey and Romero, 2001; Sapolsky et al., 2000) to provide the resources needed to cope with a current or anticipated increase in energy expenditure (Herman et al., 2016; McEwen and Wingfield, 2003; Romero et al., 2009). In this context, GCs are considered as mediators of allostasis (i.e. achieving stability through change; McEwen and Wingfield, 2003), integrating physiology and associated behaviours in response or anticipation to changing internal and external conditions (McEwen and Wingfield, 2003; Romero et al., 2009). GC levels fluctuate in daily and seasonal patterns (i.e. not only in response to perturbations), and also rise rapidly when a perturbation occurs. For example, a GC increase is often observed in response to colder weather, which induces a higher metabolic rate (Jenni-Eiermann et al., 2008; Lendvai et al., 2009; Thiel et al., 2011; Jimeno et al., 2017a; reviewed in Jessop et al., 2016). Despite this fact, much research on GCs in the last decades has focused on establishing associations between variation in GC concentrations and animal welfare or fitness prospects under the assumption that GCs are indicators of stress, usually without explicitly considering the role of variation in energy metabolism. This tendency contrasts with the numerous studies published during the early stages of GC research that provided insights into their metabolic role (e.g. Snedecor et al., 1963; Edens and Siegel, 1975; Harlow et al., 1987; reviewed in Siegel, 1980; Munck et al., 1984).

Although the negative association between ambient temperature and GCs is often found, it is not ubiquitous (e.g. differing between sexes or among taxa; Lendvai et al., 2009; Jessop et al., 2016; Jimeno et al., 2017b), for reasons that are not well understood. A potential explanation for the heterogeneous findings is that there is environmental or individual variation in the extent to which ambient temperature affects metabolic rate and hence GCs. Negative results when testing for a GC–metabolism relationship may also arise from a reliance on cross-sectional data, in which variation between individuals can partially mask existing patterns within individuals (Briga and Verhulst, 2017). In the present study, we therefore concentrated on within-individual variation.

Lower ambient temperature requires higher heat production, leading individuals to increase their energy requirements (Jimeno et al., 2017a; Cohen et al., 2008). When food acquisition costs energy, which will usually be the case in the wild, but rarely so in captivity, a lower ambient temperature further increases foraging effort, because the foraging costs themselves need to be covered with more foraging. Building on the hypothesis that GCs are primarily regulated with respect to energetic demands, we predicted the GC–temperature associations to be steeper in environments with higher foraging costs. We tested this prediction in captive zebra finches [Taeniopygia guttata (Vieillot 1817)] living permanently in outdoor aviaries with either low or high foraging costs, by comparing baseline corticosterone (CORT; the main bird glucocorticoid) measurements taken on the same individuals in two consecutive years at different ambient temperatures.

Housing and rearing conditions of the birds used in this study are described in Briga et al. (2017). In brief, zebra finches were bred indoors, and when the oldest chick was maximally 5 days old, chicks were randomly cross-fostered to create small and large broods, always within the range observed in the wild. After reaching 100 days of age, individuals were assigned randomly to one of eight outdoor aviaries (310×210×150 cm), evenly distributed between easy and hard foraging environments. Each aviary contained 20–25 individuals of one sex, and an approximately equal number of birds reared in small and large broods.

The foraging manipulation is described in detail in Koetsier and Verhulst (2011). In brief, in each aviary, a food container with five holes on each side was suspended from the ceiling. In the easy foraging environment, food boxes had perches just below the holes, allowing the birds to perch while eating (low foraging costs). In the hard foraging environment, the perches were absent, forcing birds to remain in flight when obtaining food (high foraging costs). Birds could sustain themselves well in these conditions (Koetsier and Verhulst, 2011), but in the long run the manipulation shortened lifespan of birds reared in large broods, but not of birds reared in small broods (Briga et al., 2017). The foraging manipulation did not, on average, affect baseline corticosterone levels over all groups, but more complex patterns emerged in more detailed analyses (Jimeno et al., 2017b).

Ambient temperature at the aviaries was recorded each hour (HOBO, Onset Computer Corporation). Following Jimeno et al. (2017b), for temperature we used the average ambient temperature during the hour prior to sampling.

Blood samples were collected in May 2014 and May 2015, always within 2 min of entering the aviary, in the context of the study described in Jimeno et al. (2017b). Samples were taken from the brachial vein, collected in heparinized microcapillary tubes and stored on ice until centrifugation. Plasma was separated from all samples and stored at −20°C until analysed.

Plasma CORT concentrations were determined using an enzyme immunoassay kit (ADI-900-097, ENZO Life Sciences, Lausen, Switzerland), following previously established protocols (Jimeno et al., 2017b). In brief, aliquots of 10 μl along with a buffer blank and two positive controls (at 20 ng ml⁻¹) were extracted twice with diethylether and redissolved in 280 μl assay buffer after evaporation. On the next day, two 100 μl duplicates of each sample were added to an assay plate and taken through the assay. Buffer blanks were at or below the assay's lower detection limit (27 pg ml⁻¹). Samples with coefficients of variation higher than 20% were re-assayed. Final hormone concentrations were corrected for average loss of sample during extraction in our laboratory (i.e. 15%).

To test our hypothesis, we applied model selection using the Akaike information criterion corrected for small sample size (AICc; Burnham and Anderson, 2004). Difference in corticosterone (2015–2014) was the dependent variable, and the model representing our hypothesis contained temperature difference (2015–2014), foraging treatment and their interaction. The alternative models we considered are listed in Table 1. Statistical analyses were performed using R version 3.2.2 (https://www.r-project.org/) with the  function ‘lm’ in the R package nlme (http://cran.r-project.org/package=nlme), and the functions ‘dredge’ and ‘r.squaredGLMM’ in the R package MuMIn (https://cran.r-project.org/package=MuMIn). Logarithmic transformations were performed to normalize CORT; corticosterone change was calculated as the difference between lnCORT2015 and lnCORT2014.

We previously showed that being reared in a small or large brood had no effect on the association between ambient temperature and corticosterone (Jimeno et al., 2017b). Analyses on the present dataset confirmed this result, and therefore we do not report the effects of the brood size manipulation in the present analyses.

All methods and experimental procedures were carried out under the approval of the Animal Experimentation Ethical Committee of the University of Groningen, licence 5150E, and in accordance with the approved guidelines.

CORT difference in response to a temperature difference was known for 49 individuals that were sampled in both years (Fig. S1; 28 in the easy and 21 in the hard foraging environment; 27 females and 22 males). In agreement with our prediction, the model that best explained the within-individual change in CORT concentrations (i.e. lowest AICc) included temperature difference, foraging treatment and their interaction (Table 1), while the next best model (ΔAICc=+1.25) was similar except for the addition of sex as a main effect. Thus, between-year differences in ambient temperature were associated with differences in CORT, and this association was affected by foraging treatment, with birds living in the energetically more demanding environment showing a steeper slope (Fig. 1). The effect of temperature differences on CORT differences did not differ between the sexes; i.e. including the interaction between sex and foraging treatment or temperature resulted in poorer model fits (ΔAICc>5.9). The lack of an interaction with sex is in agreement with our earlier results collected under temperature-controlled conditions (Jimeno et al., 2017a). Removing the interaction between foraging treatment and temperature difference from any of the models always increased AICc values (ΔAICc>4). Thus, when experiencing natural variation in ambient temperature, individuals that had to expend more energy to forage (i.e. fly more to obtain food) showed stronger CORT responses to variation in ambient temperature compared with individuals in the less demanding foraging environment.

The increase in metabolic demands induced by lower temperatures will require an increase in fuel supply (i.e. glucose, the main fuel molecule in birds) to match those needs. Glucose can be absorbed in the intestine during digestion, or synthesized in the liver from glycogen or fat reserves (Braun and Sweazea, 2008). As GCs are required for the latter process, a correlation between energetic (i.e. glucose) needs and GCs may only be detected when individuals do not have access to food, or when access to food is energetically costly, as in our hard foraging treatment. Glucose synthesis and mobilization may also be needed when food is easily available but the short-term energetic demands are too high to be fulfilled by current absorption from food; however, further research is needed to test this idea and the involvement of the glucose mobilization processes. Nevertheless, this strong effect of the foraging costs, together with reliance on cross-sectional measurements, could partly explain the inconsistency between studies testing for correlations between metabolism and GCs (e.g. Wikelski et al., 1999; Buehler et al., 2012).

Our results can be integrated into the allostasis model (McEwen and Wingfield, 2003), which provides a framework for understanding GC secretion through modelling the energetic requirements of an individual in relation to the energy available in its environment. This model incorporates concepts that allow replacing the controversial word ‘stress’ (Blas, 2015). Allostatic load is the cumulative energetic requirement of an organism (the ‘workload’) at a particular moment, including predictable and unpredictable demands. Meanwhile, allostatic overload is defined as the state in which energy requirements exceed the capacity of the animal to replace that energy from environmental resources. According to this theoretical framework, birds in a hard foraging environment would have a higher allostatic load compared with birds in an easy foraging environment, because the energy required to obtain food will be higher in the hard foraging environment. This difference will be larger at colder temperatures, because birds need to allocate more energy to thermoregulation, and at very low temperatures the individuals in the hard foraging treatment will have a higher risk of experiencing allostatic overload unless energy-saving mechanisms such as hypothermia are triggered (Briga and Verhulst, 2017).

In conclusion, our findings underline the importance of accounting for variation in energy metabolism when interpreting variation in glucocorticoid concentrations. More generally, the present study illustrates the importance of investigating variation in (physiological) traits in multiple environments that differ in ecologically relevant variables, particularly in laboratory conditions, which usually differ strongly from the environment in which the trait evolved.

We thank Sabine Jörg for expertly running all assays, and students who helped in the sampling sessions.