Of 11 candidate steroids, corticosterone concentration standardized for mass is the most reliable steroid biomarker of nutritional stress across different feather types

Abstract Measuring corticosterone in feathers has become an informative tool in avian ecology, enabling researchers to investigate carry‐over effects and responses to environmental variability. Few studies have, however, explored whether corticosterone is the only hormone expressed in feathers and is the most indicative of environmental stress. Essential questions remain as to how to compare hormone concentrations across different types of feathers and whether preening adds steroids, applied after feather growth. We used liquid chromatography coupled to tandem mass spectrometry to quantify a suite of 11 steroid hormones in back, breast, tail, and primary feathers naturally grown at overlapping time intervals by rhinoceros auklet Cerorhinca monocerata captive‐reared fledglings and wild‐caught juveniles. The captive‐reared birds were raised on either a restricted or control diet. Measured steroids included intermediates in the adrenal steroidogenesis pathway to glucocorticoids and the sex steroids pathway to androgens and estrogens. Corticosterone was detected in the majority of feathers of each type. We also detected cortisone in back feathers, androstenedione in breast feathers, and testosterone in primary feathers. Captive fledglings raised on a restricted diet had higher concentrations of corticosterone in all four feather types than captive fledglings raised on a control diet. Corticosterone concentrations were reliably repeatable across feather types when standardized for feather mass, but not for feather length. Of the seven hormones looked for in uropygial gland secretions, only corticosterone was detected in one out of 23 samples. We conclude that corticosterone is the best feather‐steroid biomarker for detection of developmental nutritional stress, as it was the only hormone to manifest a signal of nutritional stress, and that exposure to stress can be compared among different feather types when corticosterone concentrations are standardized by feather mass.


| INTRODUC TI ON
The desire to assess the endocrine state of an animal with minimally invasive techniques has led to methods to quantify hormone concentrations in easily sampled tissues such as whiskers (Karpovich, Skinner, Kapronczai, Smith, & Janz, 2018), hair (Koren et al., 2019), and feathers (Bortolotti, Marchant, Blas, & German, 2008). In avian biology the technique of measuring stress hormones, glucocorticoids, in feathers is of particular interest as many species grow different groups of feathers at different times of the year. This feature of feathers makes them potentially valuable as indicators of physiological status during life stages when birds are inaccessible to researchers (e.g., Orben et al., 2018;Ramos, Llabrés, Monclús, López-Béjar, & González-Solís, 2018), over long periods of time (e.g., Fairhurst, Bond, Hobson, & Ronconi, 2015;Will, Kitaiskaia, & Kitaysky, 2018), or to assess the response of an individual to conditions over time (e.g., Ganz, Jenny, Kraemer, Jenni, & Jenni-Eiermann, 2018;. Bortolotti et al. (2008) posed the passive deposition hypothesis to explain how circulating glucocorticoids are incorporated into feather tissues via diffusion during keratinization and pigmentation of the feather in the blood quill (Jenni-Eiermann, Helfenstein, Vallat, Glauser, & Jenni, 2015). Hormones may also be incorporated into feathers during preening when uropygial gland secretions, which may contain dissolved steroids, are exogenously applied to feathers and thus may accumulate as feathers age (Jenni-Eiermann et al., 2015). Here, we test whether, as predicted by the passive deposition hypothesis, other circulating hormones and/or their metabolites are incorporated into feathers evenly throughout a bird's body and whether any other hormone is a potential candidate for measuring exposure to stress. We also address the issue of how to compare hormone concentrations across feather types, an issue regardless of how hormones arrived in the feather, and test the possibility that hormones may be incorporated into the feather via dissolved secretions in preen oils.
Corticosterone is not the only hormone transported by the circulatory system, and it is possible that other steroid hormones, including androgens and estrogens, may also be present in feather tissues.
The specific steroid, and the amount, that is incorporated into feather tissues should depend on which hormones, and how much, are in circulation at the time of feather growth and may vary depending on the feather type as a result of molt schedules occurring during different life stages, take place in geographically distinct places due to migration or because a species inhabits seasonally fluctuating environments (e.g., Romero, Soma, & Wingfield, 1998). Androgen precursor steroid hormones can also be present in the circulatory system during periods of the year when synthesis in the gonads is inactive (Soma, Scotti, Newman, Charlier, & Demas, 2008). Identifying patterns in hormone deposition across feather types may, therefore, be of use to researchers hoping to target feather sampling to capture a specific seasonal representation of endocrine activity.
To date, the analysis of hormone concentrations in feathers has been largely confined to measuring the major avian glucocorticoid, and corticosterone. One previous study linked concentrations of cortisol and testosterone, in addition to corticosterone, in house sparrow flank feathers to subsequent survival (Koren et al., 2012) and another explored the detectability of testosterone (in addition to corticosterone) in very small quantities of feathers (Bílková, Adámková, Albrecht, & Šimek, 2019). However, it is largely unknown whether other steroid hormones might also be present in feather tissues and, if so, also affected by exposure to stress. "Stress" refers to a wide range of environmental conditions such as problematic temperature changes (e.g., Xie, Romero, Htut, & McWhorter, 2017), changes in food availability (e.g., Kitaysky, Wingfield, & Piatt, 1999), predation pressure (mixed evidence: Clinchy, Zanette, Boonstra, Wingfield, & Smith, 2004;Fontaine, Arriero, Schwabl, & Martin, 2011), or pollution (contaminants: Bourgeon et al., 2012noise: Kleist, Guralnick, Cruz, Lowry, & Francis, 2018) that act directly, or indirectly via psychological stress (e.g., Cyr, Earle, Tam, & Romero, 2007), on an individual's physiology. Here, we narrowly define "stress" as developmental food limitation and build on previous work, which linked experimental exposure to nutritional stress in captive-reared rhinoceros auklet (Cerhorhinca monocerata) chicks to elevated concentrations of corticosterone in plasma and feathers (Sears & Hatch, 2008;Will et al., 2014).
Whether corticosterone is the most effective biomarker to analyze in order to quantify stress exposure has received little attention (Koren et al., 2012). For example, plasma testosterone decreases in breeding male rufous-sided sparrows Peucaea carpalis in response to physical restraint (Deviche et al., 2014) and plasma estrone decreases in reproductively active female black birds Turdus merula in response to artificial lights at night (Russ et al., 2015). In both cases, decreased concentrations of testosterone and estrone corresponded to an increase in plasma corticosterone. However, corticosterone production may be dampened during feather growth (Romero, Strochlic, & Wingfield, 2005), so in some species it may be preferable to use an alternative steroid hormone to detect exposure to environmental stress. Furthermore, the pools of precursor steroids and the metabolites produced during steroid clearance may change in response to stressful conditions. Using liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS), we measured the concentrations of 11 major steroids to examine which metabolic pathways in four types of feathers (primary, tail, back, and breast), from captive rhinoceros auklet fledglings (Figure 1), grown on either food restricted or control diets. Our previous work on these same individuals showed that rhinoceros auklets experience no ontogenetic patterns in the circulation of corticosterone in plasma, or temporal trends in the concentration of corticosterone in feathers ( Figure 2, Will et al., 2014). In free-living individuals raised by their parents, temporal patterns of corticosterone deposition varied both by colony and by year and were associated with diet composition rather than age of the bird . Due to limitations in detectability of our LC-MS/MS method, optimized for a suite of steroids, we needed to use whole feathers, and sometimes combined several feathers in the same sample, to have sufficient concentrations.
This profiles more closely the situation researchers may face in comparing whole feathers grown at different time periods in freeliving adult birds, but it may arguably sacrifice the precision that a comparison of feather segments grown at the exact same time period would yield. Since we know that there was no significant temporal ontogenetic variability in corticosterone secretion during the development of the captive-reared chicks, we have based our whole-feather comparison on the premise that all feathers were exposed to a relatively constant amount of circulating corticosterone in birds raised on a control diet.
Feather growth in rhinoceros auklets spans their entire nestling phase; primary feathers emerge at day 11  and tail feathers emerge at day 20 (A. Kitaysky, unpublished data) posthatch. In closely related tufted puffins Fratercula cirrhata, back and breast feathers emerge at day 8-11 posthatch (Piatt & Kitaysky, 2002). We estimate that back and breast feathers finish growing at between day 24 and 28 posthatch based on a growth rate of 2.22 mm/day, the rate of primary feather growth in captive-reared rhinoceros auklets . Primary feathers complete growth between day 42 and 49 and tail feathers finish F I G U R E 1 Rhinoceros auklets grow their feathers over the course of about 40 days while in the nest (photo credit: A. Will) F I G U R E 2 (a and b) are reproduced from Will et al. (2014), illustrating the lack of temporal patterns in the presence of corticosterone in both plasma (a) and primary feathers (b) in birds raised on a control diet (light gray) throughout the nestling period. (c) Schematic of feather growth patterns during rhinoceros auklet nestling development. Back and breast feathers emerge between day 8 and 11 and finish growing at day 24 to 28 posthatch. Primary feathers emerge at day 11 and complete growth between days 42 and 49. Tail feathers emerge at day 20 and finish growing sometime after fledging (after day 56), but for captive raised birds tail feather growth was terminated between days 42 and 48 (as illustrated above). Feathers are grown from tip (earliest) to base (latest) Captive-reared birds were sacrificed between ages 42 and 48 when they began to regularly leave the nest (Sears & Hatch, 2008).
Since all feathers overlapped in their timing of growth (Figure 2), we were able to examine whether an endocrine signal of nutritional stress was detectable across all feather types. We then used these results, with the addition of feathers collected from wildcaught juveniles, to resolve the issue of comparing hormone concentrations across different feather types. Bortolotti et al. (2008) originally recommended to report feather corticosterone concentrations standardized by sample length to reflect exposure to stress over time, assuming a constant growth rate, and cautioned against comparing different feather types, which may have more material, and thus more hormone, per unit of feather length. Feathers grown at different times of the year vary in size and structure, which may affect hormone deposition in the feather tissues (Harris, Madliger, & Love, 2016) and may preclude direct comparison of a bird's exposure to stressful conditions during various stages of its annual cycle. It may, however, be possible to account for differences in feather structure by standardizing concentrations by sample mass or to correct length concentrations by mass (e.g., Orben et al., 2018;Will et al., 2014;Will et al., 2015). Here, we test three methods of reporting hormone concentrations (standardized for sample length, mass, and length corrected for mass) in feathers to determine which is the least affected by the differences in feather structure among breast, back, primary, and tail feathers. Since we expect corticosterone concentrations to be similar among the four feather types, we expect that the method(s) of standardization that removes the effect of feather type on steroid concentration will show no discernable trend in steroid concentration with increasing feather size and weight.
Finally, we address the possibility that preening might deposit steroids dissolved in preening oils, including glucocorticoids, into, as well as adherent to, feathers (Jenni-Eiermann et al., 2015). We compared steroid signatures of feathers and secretion of uropygial glands between captive fledglings which had not started preening their feathers (A. Kitaysky, personal observation, birds were sacrificed prior to prefledging preening) compared to wild juveniles (caught at sea) which have engaged in preening behavior to waterproof their plumage additional time while at sea.

| Samples
We collected six back, six breast, one tail, and one primary feather by plucking, as well as uropygial gland oil by absorbent filter paper, from carcasses of captive-reared birds of fledging age and wildcaught juvenile rhinoceros auklets (likely a month post fledging).

Captive chicks were hatched and raised at the University of Alaska
Fairbanks under IACUC #06-21 on either a control (n = 8) or restricted (n = 7) diet. Diet restriction started at day 10 posthatch and ended when chicks fledged (for experiment details see Sears & Hatch, 2008). Previous studies on these same birds reported elevated concentrations of corticosterone (based on corticosterone radioimmunoassay analyses) in both plasma (Sears & Hatch, 2008) and feathers  of individuals raised on a restricted diet. Individuals included in this study were euthanized after fledging. Wild-caught juveniles (n = 8) were collected (a by-catch of commercial gill-net fisheries) within a month of leaving their nests by the National Marine Fisheries Service in Southeast Alaska. All birds were stored frozen until sampled.

| Candidate steroids quantified
The 11 steroids quantified encompass the pathway for adrenal glucocorticoid synthesis, as well as the sex steroid pathway to androgens and estrogens ( Figure 3). Specifically, the earliest common substrate steroid for the steroidogenic pathways investigated was 17-hydroxyprogesterone. The intermediate steroid on the adrenal steroidogenesis pathway from 17-hydroxyprogesterone to cortisol is 11-deoxycortisol. Finally, the first breakdown product of cortisol is cortisone, which can also serve as a substrate pool for back-synthesis of cortisol, was quantified. In the parallel corticosteroid pathway, we were not able to quantify the direct precursors of corticosterone (progesterone and 11-deoxycorticosterone), but were able to quantify 11-dehydrocorticosterone as the breakdown product/substrate pool. In these parallel pathways for biosynthesis of cortisol and corticosterone, the same enzymes act on both substrates to synthesize both products (pathways described in Taves 17-hydroxyprogesterone is also the immediate precursor for the gonadal sex steroid pathway. 17-hydroxyprogesterone is converted to the androgen, androstenedione, that is, in turn, the substrate for testosterone synthesis. Estrogens are synthesized from these two androgens by the aromatase enzyme, with androstenedione as the substrate for estrone and testosterone as the substrate for estradiol. Both estrone and estradiol can then be converted into the third estrogen quantified, estriol (pathways described in Hanukoglu, 1992).
In each case where the calibration was against a nonbioidentical steroid, results were compared to direct reading from the nondeuterated, linear, calibration curve for the bioidentical standard and results were similar. Quantitation interference in this matrix and method precluded quantitation of progesterone. HPLC grade methanol (MeOH), Optima grade acetonitrile (ACN), ammonium fluoride (NH 4 F), and Optima grade water were purchased from Fisher Scientific.

| Sample preparation
The calamus of each feather was removed to avoid blood and skin tissue contamination; feathers and uropygial gland secretions were stored dry at room temperature until analysis. Each feather (primary mean mass 0.036 g ± 0.004 standard deviation, tail 0.0166 g ± 0.002), group of 6 feathers (back total mean mass 0.017 g ± 0.004, breast 0.017 g ± 0.002), or uropygial gland secretion was placed in a 13 × 100 mm borosilicate test tube. To the test tube, 100 μl of combined internal standard containing all deuterated standards, plus 9 ml cold (4°C and held in an ice bath) methanol was added. Tubes containing the standards and the feathers were capped and stored in a 4°C fridge for 20 hr, whereas the filter paper samples that were explicitly oils, esters, and waxes were in-
The flow rate was 0.22 ml/min, and the injection volume was 12 µl.

| Quantitation method
Each run contained a 10-point calibration series for each steroid.
The range was from 100 ng/ml down to 0.1 ng/ml for cortisol, corticosterone, cortisone, and androstenedione and 50 ng/ml down to 0.05 ng/ml for the remaining steroids. The lowest calibration standard that gave <20% CV and <30% error was the lower limit of quantitation (LLOQ). LLOQ was 0.1 ng/ml for all steroids except cortisol and corticosterone, with an LLOQ of 0.2 ng/ml. When the sample fell below LLOQ, no concentration was reported. Three quality control solutions at 5.0/0.5/0.1 ng/ml or 2.5/0.25/0.05 ng/ml were included in each run. All CVs for all QC above LLOQ were <20%.
Each steroid was quantified relative to an internal standard that carried deuterium substitutions for hydrogen that increased the molecular weight without altering the biochemical properties of the molecule. Calibration curves used analyte/internal standard peak area ratios (y axis) versus analyte concentration (x axis) yielding a linear fit with 1/x weighting to improve precision at low concentrations. R 2 for all calibration lines were always higher than 0.995, and typically higher than 0.998.

| Standardization of concentrations
We standardized hormone concentrations (Table 1)

| Statistical analysis
All analyses were performed in R v. 3.4.2 (R Core Team, 2017).
We used linear mixed effects models to evaluate which method of standardization, by length, mass, or detrending length concentrations for mass, results in patterns of corticosterone concentrations of back, breast, primary, and tail feathers most similar to previously observed patterns in plasma corticosterone concentrations, where nestlings during primary feather growth do not have higher concentrations of circulating corticosterone than during body feather growth. These three standardized concentrations were modeled with linear mixed effects models using the lme4 (Bates, Maechler, Bolker, & Walker, 2015) and lme4Test (Kuznetsova, Brockhoff, & Christensen, 2017) packages in R. We included feather type ("FeatherType" = back, breast, primary) as a fixed effect and individual as a random intercept. Note: 17-OH-Progesterone is the same as 17-hydroxyprogesterone. 11-DHC is the abbreviation for 11-dehydrocorticosterone. Estriol and Estrone were not detected in any sample.

| Steroid detection by feather type
The common cortisol and sex steroid precursor 17-hydroxyprogesterone were detected in fewer than half of the feather samples, irrespective of sex (n = 23 individuals for all analyses except for primary feathers for estradiol and estrone n = 15, and tail feathers were not analyzed for estradiol and estrone, see Table 2 for detection summary). The product of 17-hydroxyprogesterone, 11-deoxycortisol, was not detected in any of the feathers analyzed. Cortisol, the product of 11-deoxycortisol, was detected in only a few back (4.35%) and breast (13.04%) feathers. Cortisone, a product of cortisol, was detected in a higher proportion of back feathers (73.91%) than in breast (17.39%), tail (4.35%), and primary (4.35%) feathers (Fisher's exact test: p < .001). In the parallel corticosteroid pathway, corticosterone was detected in half or more of the feathers of each type. Its product 11-dehydrocorticosterone, which can also serve as a substrate pool, was detected in less than half of the feathers of each type. Androstenedione, the sex steroid precursor, was detected in breast (73.91%), back (34.78%), and primary (52.17%) feathers but only in a limited number of tail feathers (17.39%, Fisher's exact test: p < .001) while testosterone, one of its products, was detected in nearly all primary feather samples (91.30%), but in fewer than half of all other feather types analyzed (Fisher's exact test: p < .001).
Estradiol, a product of testosterone, was detected in fewer than half of the primary feathers and was not detected in body feathers. The other estrogens (estrone and estriol) were not detected.

| Effect of sex
The proportion of samples in which steroids were detected did not significantly differ between male and female birds (captive-reared and wild-caught combined). Although not statistically significant, proportions of detection did vary. For example, 17-hydroxyprogesterone was detected in more male (50% n = 8) than female birds (20%, n = 12, Fisher's exact test: p = .16) and estradiol was detected in more female (50%, n = 12) than male birds (12.5%, n = 8, Fisher's exact test: p = .15). Testosterone was detected in all but one bird, a female, and androstenedione was detected in half of both female and male birds (58%, n = 12 and 50%, n = 8 respectively, Fisher's exact test: p = 1 for both steroids). To determine whether hormones may be acting in tandem or serving as a potential substrate pool, we examined the correlation matrix

| Measuring exposure to nutritional stress
Back, breast, primary, and tail feathers from birds raised on a restricted diet (n = 7) had significantly higher concentrations of TA B L E 2 The percentage of samples in which each hormone was detected above the lower limit of quantitation Note: Estradiol was detected in a higher proportion of female feathers than male feathers. Estriol, not depicted, was not detected in back, breast, or primary feathers. Tail feathers were not analyzed for estrone, estradiol, or estriol. 17-OH-Progesterone is the same as 17-hydroxyprogesterone. 11-DHC is the abbreviation for 11-dehydrocorticosterone. F 2,20 = 16.03, p < .0001; Figure 5). There was a greater difference in feather corticosterone concentrations in primary and tail feathers in food-restricted birds than in control and wild birds across all three methods of standardizing concentrations ("Treatment*FeatherType" by mass: F 6,60 = 3.105, p = .01, length:  Table 4 for effect sizes and t tests).

| Detection of steroids in preen oil
Corticosterone was detected in only one uropygial oil sample from a food restricted captive bird, although this individual did not have the highest feather corticosterone concentrations. There was no detection of 17-hydroxyprogesterone, 11-deoxycortisol, cortisol, cortisone, 11-dehydrocorticosterone, androstenedione, or testosterone in any of the preen oil samples (n = 23).

| D ISCUSS I ON
We used LC-MS/MS to quantify the concentrations of 11 steroid hormones in different feather groups grown by captive-reared fledglings and wild-caught juvenile rhinoceros auklets to test four predictions of the passive deposition hypothesis regarding the presence, signal of nutritional stress, and comparability of steroid hormones in and across feather types, as well as the possibility that steroid hormones may also be incorporated into feathers exogenously via preen oils.
We sampled feather types grown at overlapping intervals as a method of assessing the size of the steroid pool in a feather. We found that androstenedione and testosterone were detected in each of the four feather types: back, breast, primary, and tail that we tested. In the corticosteroid pathways, corticosterone and cortisone were also detected in each of the four feather types. With the exception of corticosterone, however, the proportion of detection of a particular steroid varied by feather type. Androstenedione was seldom detectable in tail feathers, testosterone was mostly detected in primary feathers, and cortisone was detected in a higher proportion of back feathers compared to the other feather types.
Assuming production of steroids is systemic, variability in the detection of hormones across feather types may be due to ontogenetic changes in the steroids in circulation. Although the four feathers we investigated overlap during their growth, primary and tail feathers begin growing slightly later and continue growing for longer than back or breast feathers (Figure 2). While this difference in timing does not explain differences in detection of cortisone in back and breast feathers, it may account for differences in the other steroids. In Greylag geese Anser anser the presence of androgen, glucocorticoid, and estrogen metabolites changed throughout the nestling period (Frigerio, Moestl, & Kotrschal, 2001). Glucocorticoids were present throughout development, while androgens and estrogens appeared at days 20 and 40, respectively (Frigerio et al., 2001). These patterns are reflected in the differential presence of these steroids in the rhinoceros auklet feathers (Table 2) and may be associated with developmental stress (glucocorticoids) and the development of the gonads and sexual behaviors (androstenedione, testosterone, and estradiol;Frigerio et al., 2001). Thus, the variability in the presence and the levels of hormones across feather types may be due in part to whole-organism differences in steroid profiles during the nestling period.
It is possible that differences in detection across feather types are due to local production of steroids. In altricial zebra finches (Taeniopygia guttata) lymphoid organs produce cortisol (and its precursors) during early development prior to the adrenals becoming fully responsive (Taves et al., 2016) and was found to be an important hormone in the development of the immune system (Schmidt and Soma, 2008). Rhinoceros auklets are, however, semiprecocial, and similar lymphoid production of glucocorticoids was not found in precocial Japanese quail or chickens (Taves et al., 2016). It is not known whether steroids may be locally produced elsewhere, for example, in the skin or the feather follicle itself. Another possible explanation is that local metabolism of hormones, perhaps at the feather follicle (Rettenbacher, Henriksen, Groothuids, & Lepschy, 2013), changes the type and availability of hormones present in the blood that circulates in the feather pulp, where hormone incorporation into the keratin structure of the feather was shown to occur (Jenni-Eiermann et al., 2015). Local metabolism may occur at different rates or express different steroidogenic enzymes, which may result in different end products.
Although we did not find another candidate hormone for reflecting continuous exposure to nutritional stress, we did find that high concentrations of testosterone were associated with high levels of corticosterone in male fledglings. In pied flycatcher (Ficedula hypoleuca), nestlings dosing with oral testosterone increased begging behavior (Goodship & Buchanan, 2007). It is not known whether testosterone is associated with begging behavior in rhinoceros auklet nestlings, however, it is possible that the positive association of testosterone and corticosterone concentrations in the feathers of male chicks (Figure 4) may be an adaptation of male nestlings to counteract preferential maternal investment in female offspring observed during poor environmental conditions (Addison, Kitaysky, & Hipfner, 2008). We used the concentrations of corticosterone in feathers naturally grown at overlapping time periods in captive-reared fledglings and wild-juvenile rhinoceros auklets to address the lingering question of whether it is meaningful to compare hormone concentrations across feather types. Lattin, Reed, Desrochers, and Romero (2011) found that hormone concentrations were very high for very small feather masses, an artifact that has not been explained by extraction efficiency (Berk, McGettrick, Hansen, & Breuner, 2016). Harris et al. (2016) compared simultaneously grown feathers in adult wild starlings and found that feather corticosterone concentrations were not comparable when standardized for feather length, however, when Lendvai, Giraudeau, Németh, Bakó, and McGraw (2013) used the per mass standardization to compare tail and breast feather corticosterone concentrations in house finches (Haemorhous mexicanus) the results were consistent within a given individual. We found that, when mass is included in the standardization, concentrations of corticosterone were comparable across back, breast, primary, and tail feathers and that patterns associated with the amount of feather material were eliminated (Figure 3), but that the signal of nutritional stress in restricted birds was still evident. When standardized for mass, an increase in concentrations of corticosterone with feather mass was still apparent in the food restricted group but not in controls. Increased nutritional stress later in the nestling period in the restricted treatment may have occurred as the chicks' energy requirements grew, but the amount of food available remained the same an experimental decision made to mimic parental provisioning in the wild (Figure 2, Sears & Hatch, 2008). This signal was not detected in measures of plasma corticosterone concentrations in food restricted birds (Figure 2). Measurements of corticosterone in plasma of the captive-reared rhinoceros auklets used in this study mirror the results obtained when concentrations are standardized for mass alone. It is critical to note that the feathers analyzed, although they overlapped in growth, did not strictly grow concurrently. Hormone concentrations standardized for length and then detrended for mass yielded similar results to comparisons made when concentrations were standardized for mass alone, but did not correct for the largest differences between light, flimsy body feathers, and dense, robust tail feathers.
We evaluated whether dissolved steroid hormones may be incorporated into the feathers from exogenously applied preening oils.
We detected only corticosterone concentrations in the preen oil of a single individual out of 23 individuals tested. We did not detect any of the other focal hormones targeted in our analysis in any of the other uropygial gland secretions. Previously, corticosterone was also not detected in preen oils using an immunoassay (Lattin et al., 2011).
Hormones were not detected in preen oils sampled from wild juvenile, individuals who had begun to apply preen oil to their feathers. methods used here were specifically chosen to reduce extraction of interfering compounds, yielding improvements in background (Di Francesco et al., 2017).

| SUMMARY AND CON CLUS I ON
We used a highly specific technique to identify the suite of hormones present in four types of feathers in a developing seabird.
We found that, even in newly fledged chicks, corticosterone was detectable in over half of the feathers analyzed and that androstenedione, testosterone, and cortisone were reliably detectable in some feather types but not others. This variability in detection of hormones across feather types followed previously reported patterns of steroid presence during nestling development. The signal of food restriction was most clearly detected in corticosterone concentrations across all feather types, and birds experiencing food restriction had higher concentrations of corticosterone. Finally, we used feathers grown during overlapping periods of development to test which method of standardizing hormone concentrations was most appropriate for comparing concentrations across different feather types. Standardizing by mass minimized the effect of feather size and type on the concentration of corticosterone.

ACK N OWLED G M ENTS
The authors would like to thank Liz Labunski and Kathy Kuletz at the United States Fish and Wildlife Service who provided the wild-caught rhinoceros auklets and would like to acknowledge the comments from M. Benowitz-Fredericks and two anonymous reviewers, which greatly improved the final manuscript. This project was funded by an Angus Gavin Migratory Bird Grant awarded to AW and the publication cost was covered by the North Pacific Board's project #1612, awarded to ASK and AW. This is publication number 1612-1. The hormone quantitation did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. However, KEWE would like to acknowledge the important role that CFI-funded infrastructure and University of Calgary/ Alberta Advanced Education and Technology start-up funds played in the conduct and analysis of this research. That support had no role in the study design, in the collection, analysis, and interpretation of data or in the writing of the report or in the decision to submit the article for publication.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

AUTH O R CO NTR I B UTI O N S
AK and KW-E conceived this study, KW-E and RZ conducted the analysis, and AW and AK analyzed the results. All authors contributed to the writing of the final manuscript.