Stress physiology of migrant birds during stopover in natural and anthropogenic woodland habitats of the Northern Prairie region

This study addresses the question of whether increased anthropogenic influence on migration stopover sites/migration corridors affects the stress physiology of migratory landbirds. Our data suggest that the reduction of natural riparian corridor habitats and subsequent increased reliance of woodland migrants on anthropogenic woodlots for stopover is not detrimental to the stress physiology of these birds in a region with limited woodland habitats.


Introduction
Migration is energetically expensive, and woodland migrant birds must refuel at stopover habitats along the migratory route (Moore and Woodrey, 1993;Yong and Moore, 1997;Moore and Aborn, 2000;Wikelski et al., 2003;Faaborg et al., 2010). Woodland habitats are scarce in the Northern Prairie region of North America and consist of natural riparian corridor woodlands (hereafter corridors) and small anthropogenic woodlots (hereafter woodlots; Castonguay, 1982;Swanson et al., 2005;Gentry et al., 2006). These natural corridors have been greatly reduced and degraded since the time of European settlement (Johnson et al., 1976;Hesse, 1996;Dixon et al., 2012). The reduction of corridor habitats, however, has occurred simultaneously with an increase in the number of small woodlots of anthropogenic origin (Bakker, 2003;Swanson et al., 2005). Given that woodland migrants occupy both corridor and woodlot habitats in this region during migratory stopover (Martin, 1980;Dean, 1999;Bakker, 2003;Swanson et al., 2003Swanson et al., , 2005, these woodlots have the potential to provide a partial substitute for lost and degraded riparian woodlands as stopover habitat for woodland migrant birds. One aspect of the biology of migratory birds that may be influenced by anthropogenic alterations to the natural landscape is their stress physiology, which can impact the timely and successful completion of migration, with potential fitness consequences (Heglund and Skagen, 2005;Carlisle et al., 2009;Eikenaar et al., 2014). Corticosterone (CORT) is the major glucocorticoid secreted in birds (Holmes and Phillips, 1976) and is a primary component of the stress response . In general, plasma CORT levels are low prior to a stressor and increase rapidly (within minutes) in response to stress, reaching a peak after 30-60 min (e.g. Holberton et al., 1996). The increase in plasma CORT in response to a stressor is known as the stress response and acts to direct the organism away from energetically costly activities, such as migration, breeding or courtship, to focus on activities that will promote immediate condition improvement and enhance the probability of survival, such as foraging (Wingfield et al., , 1995. Indeed, high plasma CORT may enhance fattening during migration (Astheimer et al., 1992;Long and Holberton, 2004;Landys et al., 2006;Lõhmus et al., 2006;Holberton et al., 2007Holberton et al., , 2008; but see Eikenaar et al., 2013Eikenaar et al., , 2014. However, plasma CORT levels and stopover duration may show a U-shaped relationship, with migrants potentially having higher baseline CORT (CORT B ) upon arrival at stopover sites, showing a decline as their body condition improves and an increase shortly before their departure, when fat stores are high (Piersma et al., 2000;Landys-Ciannelli et al., 2002;Eikenaar et al., 2013). Such a U-shaped relationship between stopover duration and plasma CORT may still be consistent with CORT promoting fattening if fat migrants decrease foraging and fattening rate prior to departure (e.g. Loria and Moore, 1990), concurrent with elevated plasma CORT concentrations to stimulate departure. The elevated plasma CORT concentrations upon departure may function in relationship to other hormones as the cue to induce migratory flight (Schwabl et al., 1991;Landys-Ciannelli et al., 2002;Eikenaar et al., 2013).
Plasma CORT levels in birds may be related to body condition, and many studies suggest that fat birds may show lower levels of CORT B and a greater stress response than lean birds (Schwabl et al., 1991;Marra and Holberton, 1998;Holberton, 1999;Jenni et al., 2000;Piersma et al., 2000;Mizrahi et al., 2001;Landys-Ciannelli et al., 2002;Long and Holberton, 2004;Raja-aho et al., 2010;Fokidis et al., 2011). Migrants in corridor and woodlot sites in the present study showed similar refuelling rates, as indicated by plasma metabolite analyses (Liu and Swanson, 2014), but whether the foraging effort required to achieve similar refuelling rates differs between corridors and woodlots is unknown. Highly fragmented landscapes, such as the woodlots in the present study, often reduce food availability and require increased foraging effort in comparison to more contiguous habitats, such as the corridor habitats in the present study (Ellis et al., 2012). If differences in foraging effort exist between corridor and woodlot habitats, such differences could influence the stress physiology of migrants (Kitaysky et al., 1999;Pravosudov et al., 2001;Mark and Rubenstein, 2013), despite similar refuelling rates.
Differences in the size and structure of corridor and woodlot habitats could also affect other ecological factors, which could, in turn, influence stress physiology of migrant birds. In southeastern South Dakota, the average size of corridor woodlands is more than three times larger than woodlots, which results in much greater amounts of woodland edge for woodlot habitats, and corridors also harbour more diverse vegetation than woodlots (Castonguay, 1982;Gentry et al., 2006;Liu and Swanson, 2014). These differences in size, structure and vegetative diversity of corridors and woodlots may produce differences in predation pressure, competition among migrants and thermal microclimates (Ellis et al., 2012), all of which could potentially influence stress physiology. Predation pressure can alter the stopover biology of migrants (Lank et al., 2003;Pomeroy, 2006;Taylor et al., 2007) and influence the stress response (Cockrem and Silverin, 2002;Clinchy et al., 2004;Newman et al., 2013). Smaller woodland parcel size and increased edge facilitate the access of predators to smaller woodlands (Donovan et al., 1997;Tewksbury et al., 2006), so predation pressure is expected to be higher in woodlots than in corridors in our study area. Such differences in predation pressure could result in differences in stress physiology between migrants in corridors and woodlots in the present study.
Competition may also influence stress physiology, and better competitors, such as birds with better body condition or of larger size, may show lower CORT B and a greater magnitude of the stress response (Lehnen and Krementz, 2005;Braasch et al., 2014). Hatch-year birds are usually socially subordinate to adults, which may restrict the former's access to limited resources due to competitive interactions with adults Faaborg et al., 2010). In addition, because juveniles in autumn are experiencing their first migration, foraging efficiency during stopover may be lower than that of experienced adults (Yong et al., 1998;Heise and Moore, 2003). Consequently, the stress response might differ between ages, with higher CORT B and a lower stress response in juveniles than in adults. The age ratios of migrants at woodlot and corridor study sites differ, with higher proportions of juveniles at woodlot sites, perhaps resulting from competition with adults (Dean et al., 2004;Liu and Swanson, 2014); therefore, differences in stress physiology between birds in corridors and woodlots as a function of differing competitive interactions or differing age ratios might be expected.
Finally, the smaller size and increased edge of woodlots relative to corridors could affect thermal microclimates of birds in the two habitats. These changes would be likely to contribute to increased penetration of wind into woodlot habitats, which would function to reduce operative temperatures for birds in woodlots relative to corridors (Olson and Grubb, 2007). Such reduced temperatures may have metabolic consequences for woodland birds (Olson et al., 2010), and thermal stress can influence the stress response in birds, particularly during periods of inclement weather (Wingfield et al., 1983;Rogers et al., 1993;Romero et al., 2000). Thus, differences in microclimate could also contribute to differences in stress physiology for migrants in corridor and woodlot study sites.
In the present study, therefore, we tested for differences between birds in corridors and woodlots for CORT B and the magnitude of the stress response for individual species, taxonomic families and foraging guilds to determine whether migrants in anthropogenic habitats demonstrate different stress physiology from those in natural riparian corridor habitats. As an additional objective, we also examined whether stress physiology in migrants was correlated with refuelling rates in the two habitats, as measured by plasma metabolites (Liu and Swanson, 2014).

Study area
This study was conducted in southeastern South Dakota, USA, where woodlands cover ~4% of total land area and the Missouri River corridor provides the most extensive riparian woodlands in the region (Castonguay, 1982). Woodlots cover ~1% of the total landscape area in southeastern South Dakota and they are smaller, more isolated and less vegetatively diverse than corridor woodlands (Castonguay, 1982;Swanson et al., 2003Swanson et al., , 2005. Corridor woodlands are mostly dominated by cottonwoods (Populus deltoides), with some later successional tree species, such as green ash (Fraxinus pennsylvanica), elm (Ulmus spp.), and hackberry (Celtis occidentalis), among other trees (Hesse, 1996;Swanson et al., 2005;Dixon et al., 2012). The woodlot study sites are comprised primarily of elm, mulberry (Morus alba), box elder (Acer negundo), hackberry and green ash (Swanson et al., 2003;Gentry et al., 2006). Three representative study sites were sampled from both corridor (all along the Missouri River) and woodlot habitat types in Clay County, South Dakota ( Fig. 1), from sites sampled previously by Swanson et al. (2005) and Gentry et al. (2006). Woodlot and corridor study sites were separated by at least 9.5 km.

Bird capture and blood sampling
Migrants were captured with 9 m, 36-mm-mesh, nylon mist nets. Each of the six study sites was sampled once per week, and we rotated through the sites in the same order, with woodlot and corridor sites sampled on alternating days during spring (mid-April to early June, 2010-2012) and autumn (mid-August to mid-October, 2010-2011) migration periods. Mist nets were operated for a few hours, beginning at sunrise, and during similar weather conditions, i.e. no rain and relatively low wind speed (<30 kph). We erected up to three nets per individual field worker at each site, and monitored nets continuously so that we could extract birds from the nets immediately upon capture. The initial blood samples for CORT B determination were collected within 3 min of birds hitting the net. We collected blood samples with 50 µl heparinized capillary tubes via venipuncture from the brachial vein on the underside of the wing (Schwabl et al., 1991;Wingfield et al., 1992;Palm et al., 2013). A second sample was taken 30 min after capture (CORT 30 ), with birds held in cloth bags during the 30 min restraint stress . Depending on bird body mass, a maximum of 100-200 µl blood was sampled from each individual bird (Voss et al., 2010). After collection, blood samples were transferred into microcentrifuge tubes and stored on ice while in the field. Upon return to the laboratory, blood samples were centrifuged for 10 min at 4°C and 3000g. The plasma was then removed to new microtubes and immediately stored at −60°C until later measurement of plasma CORT.
3 Figure 1: Study sites in Clay County, SD, USA. Open circles denote corridor sites and filled circles woodlot sites. Contiguous woodland areas for the study sites ranged from 1 to 1.6 km 2 for corridor sites and from 0.17 to 0.2 km 2 for woodlot sites.

Bird banding and morphological measurements
Prior to release following blood sampling, we banded birds with United States Geological Survey aluminum leg bands, identified each bird to species, and aged birds (autumn only; using skull ossification, Pyle, 1997) as after hatch year (AHY) or hatch year (HY). The date and time of capture were recorded for each bird. Unflattened wing chord and tarsus length were measured with a wing ruler to the nearest 0.5 mm or with callipers to the nearest 0.1 mm. We measured body mass to the nearest 0.1 g with an electronic scale (Ohaus Model LS200) and scored visible fat in the furculum and abdomen on a 0-5 scale (Helms and Drury, 1960). Following blood sampling and morphometric measurements, birds were released at the site of capture. Bird capture, blood sampling and banding were carried out under a Master Banding Permit from the United States Geological Survey Bird Banding Laboratory to D.L.S. All methods in the study were approved by the Institutional Animal Care and Use Committee at the University of South Dakota (protocol # 84-03-11-14B).

Plasma metabolite measurement
Plasma metabolite data are from Liu and Swanson (2014), who measured plasma concentrations of triglycerides (TRIG), glycerol and β-hydroxybutyrate (BUTY) from the same individual birds for which CORT was measured in the present study. We measured plasma metabolites with commercially available spectrophotometric assay kits (Sigma-Aldrich Corp., St Louis, MO, USA) using a Beckman DU-7400 spectrophotometer (Liu and Swanson, 2014). We did not obtain sufficient volumes of blood from all individuals for all plasma metabolite and CORT measurements, so we determined the relationships between plasma CORT and plasma metabolites from a subset of individuals, for which we had measurements of both CORT and plasma metabolites.

Plasma CORT measurement
We measured plasma CORT concentrations with commercially available microplate spectrophotometric end-point assay kits (Enzo Life Sciences ADI-901-097; Assay Designs Inc., Ann Arbor, MI, USA), according to the manufacturer's instructions (Forster et al., 2008). We distributed samples across assay plates evenly with respect to season and species to minimize effects of inter-assay variation on subsequent analyses. All samples, controls and corticosterone standards (0-2000 ng/ml) were run in duplicate. Plasma corticosterone levels were detected by absorbance of samples at 405 nm using an automated plate reader and KineticCalc Jr software (Bio-Tek Instruments, Wonooski, VT, USA). Absorbance values from samples were applied to the standard curve generated (r 2 = 0.99), and CORT levels were expressed as nanograms per millilitre. Absorbance values were also used to calculate the percentage of maximal binding, which ranged from 17.94 to 23.56% (n = 21), and the percentage of non-specific binding, which ranged from 2.11 to 2.98% (n = 21). The minimal detection limit was 27.0 pg/ml. Inter-assay samples (pooled from blood plasma of multiple species) were included in each of the plasma CORT assays to calibrate measured CORT levels and to monitor variation among assays. Inter-assay variance (average variation among assays) was 6.28% and intra-assay variance (average variation between duplicates) 9.15%.

Statistics
We performed all statistical analyses with SAS software (Version 9.3, SAS Institute Inc., Cary, NC, USA). Values of CORT B , CORT 30 and the magnitude of the stress response (CORT 30 − CORT B ) were compared for individual species, taxonomic groups (species grouped by families) and foraging guilds (Martin and Finch, 1995;Liu and Swanson, 2014; Table 1), with sample sizes ≥10 in each habitat. Post hoc power analyses, assuming a power of 0.80 and an α value of 0.05, revealed that these sample sizes are sufficient to detect significant differences at moderate to large effect sizes (>0.40) for all statistical comparisons (Cohen, 1988). Data were log 10 transformed prior to analyses to meet the assumptions of normality and equal variance. We present CORT data as mean values ± SEM (in nanograms per millilitre). We conducted the taxa and foraging guild comparisons to increase overall sample sizes for between-habitat comparisons and to include data from species for which we did not have sufficient sample sizes for individual species comparisons. We reasoned that, in general, species within the same taxonomic family or the same foraging guild would respond in a similar manner to potential variation in resources or other ecological factors (e.g. predation, microclimates) between corridor and woodlot habitats, so we pooled data from species in these groups, after controlling for species effects (see below), for betweenhabitat comparisons. Moreover, the conservation of woodland habitat for woodland migrants may need to focus on bird groups requiring similar resources or responding in a similar manner to other ecological factors (e.g. families or foraging guilds), in addition to individual species, so comparisons among taxa and foraging guilds may better inform conservation planning.
To adjust for individual differences in body size, we used principal component analysis to incorporate structural measurements (wing chord and tarsus length) to produce a principal component (PC1) for each individual species describing body size (Pedhazur, 1997;Guglielmo et al., 2005). We used a backward stepwise multiple regression approach to generate predictors with CORT B , CORT 30 or the magnitude of the stress response as the dependent variable. Independent variables in the model included the principal component for body size, fat score (the average of furcular and abdominal fat scores), age (autumn only), season (foraging guilds only), capture time, capture date and year, with variables retained in the model at P < 0.10. The variables retained after multiple regressions were maintained as covariates in an analysis of covariance (ANCOVA) to test for differences between birds from corridors and woodlots. We applied ANCOVA (for CORT B and CORT 30 ) or repeatedmeasures ANCOVA (for the magnitude of the stress response) to test for differences between habitat types and ages (autumn only) in both spring and autumn. If no covariates were generated by multiple regressions, we used analysis of variance (ANOVA; for CORT B and CORT 30 ) or repeated-measures ANOVA (for the magnitude of the stress response) to test for differences between birds from corridors and woodlots. Repeated-measures ANOVA or ANCOVA also allowed us to test whether CORT B and CORT 30 differed significantly for all species, taxa and foraging guilds.
To control for possible individual species effects on plasma CORT levels in analyses of foraging guilds and taxa, we included species and habitat type as categorical independent variables, along with the species × habitat type interaction term, in the general linear model ANCOVA. The rationale behind this correction for species effects is that different species may show varying raw plasma CORT concentrations, so significant biases could exist in pooled species comparisons if species are combined without controlling for species differences. For individual species with sample sizes ≥10 in both corridors and woodlots, we also applied unpaired Student's t-tests to compare fat scores.
Finally, for correlation analyses between CORT (CORT B , magnitude of the stress response) and plasma metabolites (TRIG and BUTY) and between CORT B and the magnitude of the stress response, we normalized data for individual birds to species averages (for species with n ≥ 10), so that we could pool data from individual birds in our comparisons. To normalize the CORT and plasma metabolite data for each individual bird, the individual values for CORT B , the magnitude of the stress response and plasma TRIG and BUTY levels were divided by the mean value for that species. These normalized data from individual birds were then used for correlation analyses of CORT B and the magnitude of the stress response against TRIG and BUTY concentrations and of the magnitude of the stress response against CORT B .

Comparisons of fat scores
No species showed significant between-habitat differences in fat scores, although the difference in fat scores for yellowrumped warblers (Setophaga coronata) in autumn approached significance (F 1,51 = 6.37, P = 0.08), with greater fat scores from corridors than from woodlots. We obtained sufficient sample sizes for comparisons between juveniles and adults only for warbling vireos (Vireo gilvus) and dark-eyed juncos (Junco hyemalis) in autumn, and fat scores were not statistically different between ages for these species.

Between-habitat comparisons of CORT levels
No individual species, taxon or foraging guild (see Table 1 for list of species captured) showed significant differences in CORT B between habitat types for either spring or autumn migration (Fig. 2). All individual species, taxa and foraging guilds in both habitats showed significantly elevated plasma CORT after 30 min of restraint relative to baseline levels, indicating that all birds were capable of mounting a stress response during spring and autumn migration (Figs 2 and 3, Supplementary Table 1). The CORT 30 did not vary significantly between birds from the two habitats for any group (Supplementary Table 1). Moreover, we did not detect any significant between-habitat differences in the magnitude of the stress response for any individual species, taxon or foraging guild from either spring or autumn migration (Fig. 3). During autumn migration, both adults and juveniles showed significant stress responses. However, no significant between-age differences occurred for CORT B , CORT 30 or the magnitude of stress response for any individual species, taxon or foraging guild.
We found no significant species effects for between-habitat comparisons for either foraging guilds or taxa in spring. In autumn, a significant species effect occurred for CORT B only for the sparrow group (F 5,50 = 3.32, P = 0.01) and for the magnitude of the stress response only for the foliage-gleaning guild (F 5,60 = 3.21, P = 0.02). A significant species effect indicates a strong influence of particular species on plasma CORT values for the taxon or foraging group as a whole. After between-habitat comparisons were adjusted statistically for these significant   Tables 1 and 2. species effects, between-habitat comparisons of CORT B , CORT 30 and the magnitude of the stress response remained non-significant for these bird groups. The interaction between species and habitat type for comparisons of plasma CORT was not significant for any foraging guild or taxon in either season.

Effects of covariates on CORT levels
Several covariates influenced plasma CORT levels, for CORT B , CORT 30 or for the magnitude of stress response. The CORT B was significantly higher in the autumn of 2011 than in the autumn of 2010 for all migrant groups tested, except for red-eyed vireos (Vireo olivaceus), which showed no significant effect of sampling year on CORT B (Table 2), but year was not a significant effector of CORT levels in other comparisons. Fat scores showed a significant positive correlation with CORT B and significant negative correlations with CORT 30 and the magnitude of the stress response for the ground-foraging guild, warbling vireos and sparrows during autumn migration (Table 2). Body size (described by the principal component analysis) did not exert a significant influence on plasma CORT for most bird groups, with the exception of the ground-foraging guild in spring and the foliage-gleaning guild and vireos in autumn, where smaller birds showed higher CORT B and a lower CORT 30 and magnitude of the stress response (Table 2). Autumn migrants did not show any significant difference in CORT B , CORT 30 or the magnitude of the stress response between juveniles and adults. We found no significant effects of capture date or time on CORT B , CORT 30 or the magnitude of the stress response for any individual species, taxon or foraging guild during either spring or autumn migration, except for a positive effect of capture date on the magnitude of the stress response in autumn migrant vireos, where vireos captured later in the season showed a greater stress response ( Table 2). The CORT B was significantly higher in spring than in autumn only for the ground-foraging guild. The CORT 30 and the magnitude of the stress response were significantly higher in spring than in autumn for both foliage-gleaning and groundforaging guilds.

Discussion
Anthropogenic woodlots might be expected to provide less favourable stopover habitats for landbird migrants than 7 natural riparian corridor woodlands because of their smaller size, increased isolation, increased edge and lower vegetative diversity (Castonguay, 1982;Swanson et al., 2003;Gentry et al., 2006). As a consequence of these differences, migrants stopping over in woodlots could be exposed to different levels of food availability, competition intensity, predation pressure and thermal microclimates than migrants stopping over in corridors, and these factors might lead to differences in stress physiology for birds in woodlots and corridors (Clinchy et al., 2004;Kitaysky et al., 2007;Olson and Grubb, 2007;Olson et al., 2010;Braasch et al., 2014). In the present study, however, we found no significant differences for any CORT variables (CORT B , CORT 30 or the magnitude of the stress response) between birds captured in corridors and woodlots, suggesting that the stress physiology of migrant birds in both corridors and woodlots is generally similar. In addition, all individual species, taxa and foraging guilds from both spring and autumn migration showed a significant stress response after a 30 min restraint stress, which suggests that birds were in sufficient body condition in both woodland habitats to mount a robust stress response (i.e. a significant increase in plasma CORT with restraint stress). These data suggest that the increased reliance of migrating birds on anthropogenic woodlands with the loss and degradation of natural riparian corridor habitats in the Northern Prairie region, where woodland habitat is scarce, does not negatively impact their stress physiology. This result is consistent with the finding of similar refuelling rates and body condition between birds in the two habitat types at these same study sites (Liu and Swanson, 2014). Collectively, these data suggest similar habitat quality in the two habitat types and support the idea that anthropogenic woodlot habitats can, at least partly, substitute for lost and degraded riparian corridor woodlands as  stopover habitat for migrant birds in the Northern Prairie region of North America.
Among the covariates influencing plasma CORT levels in our study, sampling year significantly affected CORT B for several bird groups in autumn, and CORT B levels were higher in 2011 than in 2010 for all of these groups. This result suggests that ecological or environmental conditions (e.g. food availability, predation pressure, intensity of competition, weather) may vary among years at our study sites in a manner that influences stress physiology of migrants. Defining which, if any, of these factors might be associated with the elevated CORT B levels in 2011 relative to 2010 will require further study, but two factors potentially related to this pattern bear mentioning. First, a large and prolonged flood impacted the Missouri River throughout the summer and early autumn of 2011 (Munes, 2014). In addition, autumn 2011 was considerably drier than autumn 2010 near Vermillion, Clay County, southeastern South Dakota, with precipitation in September and October of 2011 only 39% of that during the same months in 2010 (US Climate data, http://www.usclimatedata.com/). Liu and Swanson (2014) detected lower plasma TRIG levels in autumn 2011 than in autumn 2010 for several migrant bird groups at these same study sites, suggesting that conditions in autumn 2011 resulted in lower refuelling rates than in autumn 2010, which is consistent with relatively lower quality stopover habitat resulting in elevated CORT B levels in autumn 2011.
Body size may also be related to plasma CORT B levels in birds, with smaller birds exhibiting higher baseline levels than larger birds (Lehnen and Krementz, 2005). In the present study, the principal component describing body size had a significant effect on plasma CORT levels for only a few bird groups. The general absence of an effect of body size on plasma CORT in migrating birds is consistent with the findings of Thomas (2008), who documented that the body size of migrating shorebirds did not influence the overall CORT response. However, for those bird groups in our study where body size was significantly related to plasma CORT levels (ground-foraging guild in spring and foliage-gleaning guild and vireos in autumn), body size was negatively related to CORT B and positively related to the magnitude of the stress response. This result is consistent with the relationship documented by Lehnen and Krementz (2005) and suggests that larger birds are in better condition than and/or outcompete smaller individuals for resources in these bird groups. We have no obvious explanation for why such a relationship occurs for only some bird groups and not for others, but perhaps differential use of or access to resources is involved.
Hatch-year migrants are undertaking their first migration in autumn, so they are less experienced foragers than adults, which could result in lower refuelling rates, although agerelated differences in body condition and refuelling rates are seldom detected at stopover sites (e.g. Seewagen et al., 2013;Liu and Swanson, 2014). Moreover, juvenile migrants are usually socially subordinate to adults, which may restrict the former's access to limited resources due to competitive interactions with experienced adults (Yong et al., 1998;Heise and Moore, 2003;Moore et al., 2003;Faaborg et al., 2010). As a consequence, the stress response may differ between age classes, with higher CORT B and a lower stress response in juveniles than in adults. In the present study, we did not detect significant differences in CORT B , CORT 30 or the magnitude of the stress response between adults and juveniles. This is similar to the results of Thomas (2008), who found little difference in plasma CORT levels between adult and juvenile least (Calidris minutilla) and semipalmated sandpipers (Calidris pusilla). This suggests that ecological and environmental conditions at our stopover sites provided effective stopover habitat for both adults and juvenile birds such that age-related differences in stress physiology were not evident.
We obtained sufficient samples sizes for between-season comparisons only for foliage-gleaning and ground-foraging guilds and found that both groups had higher CORT B and a greater magnitude of the stress response in the spring compared with autumn migration. Spring migration is potentially more urgent than autumn, because early arrival on the breeding grounds generally shows a positive correlation with fitness (e.g. Smith and Moore, 2005;Cooper et al., 2011;Gargallo et al., 2011;Gienapp and Bregnballe, 2012). Spring migrants typically migrate at a faster pace than autumn migrants (Pearson and Lack, 1992;Ellegren, 1993;Fransson, 1995;Kemp et al., 2010), which may reflect differences in their physiological capacities (Swanson, 1995;Swanson and Dean, 1999). These differences might result in higher CORT B in spring than in autumn migrants and promote greater foraging intensity and better refuelling performance than for birds during the more leisurely autumn migration. Such an explanation is consistent with our previous study suggesting better refuelling performance in spring migrants than in autumn migrants at these same study sites (Liu and Swanson, 2014).
The migration modulation hypothesis proposes that migrants maintain elevated CORT B levels during migration and show a suppressed stress response when faced with additional acute stressors (Holberton et al., 1996), thereby facilitating fattening while sparing muscle protein. Consistent with this hypothesis, elevated CORT B levels, a suppressed stress response, or both, occur in a number of wild migrant or photostimulated, migratory-ready captive birds (Holberton et al., 1996(Holberton et al., , 2007Piersma et al., 2000;Landys-Ciannelli et al., 2002;Long and Holberton, 2004). However, numerous studies also document significant stress responses in wild migrant or photostimulated captive birds (Romero et al., 1997;Piersma et al., 2000;Landys-Ciannelli et al., 2002;Falsone et al., 2009;Nilsson and Sandell, 2009), suggesting that full suppression of the stress response does not occur in most migrants or that the degree of suppression changes with the stage of migratory stopover. However, the magnitude of the stress response in these birds may sometimes be influenced by body condition (Mizrahi et al., 2001;Jenni-Eiermann et al., 2009;Raja-aho et al., 2010), so energetic condition could influence the modulation of the stress response during migration. A corollary prediction of the migration modulation hypothesis is that CORT B and the magnitude of the stress response should be inversely related, even if the stress response is not fully suppressed. The suppressed stress response when facing additional acute stressors might result from negative feedback of elevated CORT B during migration (Raja-aho et al., 2010). In the present study, we found a significant negative relationship between CORT B and the magnitude of the stress response, despite robust stress responses for all bird groups studied, which is consistent with this corollary prediction of the migration modulation hypothesis.
High plasma CORT concentrations may promote elevated levels of gluconeogenesis and lipogenesis, leading to fat deposition, while also stimulating protein catabolism (Holmes and Phillips, 1976;Berdanier, 1989; but see Eikenaar et al., 2013Eikenaar et al., , 2014. Plasma levels of TRIG and BUTY are considered the best metabolites for plasma metabolite profiling in birds (Guglielmo et al., 2002Smith and McWilliams, 2010;Seewagen et al., 2011) and they are now routinely used to measure fattening rates of birds during migration because plasma TRIG increases during fat deposition and plasma BUTY increases during fat catabolism (Schaub and Jenni, 2001;Seaman et al., 2005;Guglielmo, 2006, 2010). We found a significant negative correlation of CORT B and a significant positive correlation of the magnitude of the stress response with plasma TRIG for birds in the present study. Plasma BUTY showed significant opposite patterns of correlations with CORT B and the magnitude of the stress response in comparison to plasma TRIG. These data support our hypotheses that high-quality habitat promotes effective fattening and reduced CORT B , while allowing a robust stress response.
Furthermore, CORT measures can potentially provide additional information regarding the relationships among the stress response, fattening, body condition and stopover biology in migrants. Plasma CORT shows a U-shaped relationship with fat stores in shorebirds (Piersma et al., 2000), and Eikenaar et al. (2013) found a positive correlation between plasma CORT and fat stores before departure for a passerine bird, the northern wheatear (Oenanthe oenanthe). If landbird migrants also show a U-shaped relationship between plasma CORT and fat stores during stopover, but a linear relationship with plasma metabolites (present study), analyses including CORT, plasma metabolites and fat stores might allow teasing apart of the effects of different stopover durations prior to capture on plasma metabolite comparisons. The U-shaped relationship between plasma CORT B with stopover duration in shorebirds (Piersma et al., 2000) suggests that migrants have high CORT B levels upon arrival at a stopover site and then show a decline as their body condition improves, only to increase CORT again before departure, when fat stores are high (e.g. Eikenaar et al., 2013). Such a pattern might be driven by differential sensitivity of receptors to CORT at different CORT levels or by regulation of the hypothalamo-pituitary-adrenal axis at the brain level (e.g. de Kloet et al., 1999;Matthews, 2002;Sapolsky, 2002). As a consequence, migrating birds may show considerable variation in the relationship between CORT B and fat stores. Indeed, negative relationships (Long and Holberton, 2004;Raja-aho et al., 2010), positive relationships (Romero et al., 1997;Holberton, 1999, Landys-Ciannelli et al., 2002Landys et al., 2004;Eikenaar et al., 2013) and no relationship (Holberton et al., 1996;Mizrahi et al., 2001) between CORT B and fat stores or energetic condition have been documented among migratory birds. We found no significant relationships between fat scores and CORT B for any individual species, taxon or foraging guild during spring migration. During autumn migration, fat score was a significant covariate in multiple regression models for several bird groups and showed a consistent positive relationship with CORT B and negative relationship with the magnitude of the stress response in these groups. One possible explanation for these results is that fat birds within these groups were ready to migrate, and elevated CORT B levels and a smaller stress response prior to departure served to prepare them for upcoming behavioural and metabolic transitions during migratory flights, similar to the suggestion of Piersma et al. (2000) for shorebirds and Eikenaar et al. (2013) for northern wheatears. Tracking these variables concurrently during stopover might allow more precise estimation of the stage of migratory stopover from a single blood sample than has been possible to date. However, validation of the potential for this method will require much additional research and larger sample sizes for individual species than those in the present study.

Supplementary material
Supplementary material is available at Conservation Physiology online.