Baleen hormones: a novel tool for retrospective assessment of stress and reproduction in bowhead whales (Balaena mysticetus)

Baleen samples from sixteen bowhead whales contained measurable cortisol and progesterone, and both hormones demonstrated significant correlations with sex, age class and reproductive state. Analysis of hormones in baleen shows promise as a novel tool for retrospective analysis of stress and reproductive physiology of large mysticete whales.


Introduction
Arctic marine mammals are faced with increasing exposure to a variety of ecological and anthropogenic stressors, par ticularly climate change and associated increases in seismic exploration, oil and gas development, ship traffic and fishing activity (Doney et al., 2011;Pompa et al., 2011;Reeves et al., 2011;Moore et al., 2012). The potential impacts of these rapid environmental changes on stress physiology, repro ductive physiology and related populationlevel effects are largely unknown. Moreover, such physiological responses are particularly challenging to assess in the case of mysticete whales (baleen whales), which are difficult to study with standard physiological techniques (Hunt et al., 2013).
Assessment of steroid hormones in novel tissue types offers a potential solution. Conservation physiologists are increas ingly exploring different tissue types, as diverse as urine, fae ces, hair, blubber, respiratory vapour and earwax, to measure steroid hormones relevant for addressing questions of stress physiology (e.g. adrenal steroids) and reproductive physiol ogy (e.g. oestrogens, progestins, androgens; Amaral, 2010;Hunt et al., 2013;Kellar et al., 2013;Trumble et al., 2013). In whales, the progestin content of faeces and blubber has been used to diagnose pregnancy (Rolland et al., 2005;Kellar et al., 2006Kellar et al., , 2013, and elevated faecal glucocorticoids have been shown to reflect exposure to various acute and chronic stress ors (Rolland et al., 2012;Hunt et al., 2013).
Mysticete whales have a unique tissue type that has not pre viously been explored for steroid hormone analysis, i.e. baleen. Baleen is a stratified, cornified epithelial tissue consisting of long, overlapping, fringed plates that grow downward from the upper jaw (Haldiman and Tarpley, 1993;Bragulla and Homerger, 2009;Young, 2012), which collectively form the whale's filterfeeding apparatus (Haldiman and Tarpley, 1993;Werth, 2013). It has recently become clear that most cornified epidermal tissues in vertebrates contain detectable steroid hor mones; this has been well demonstrated in human hair and fin gernails, mammalian fur, bird feathers and shed snakeskin (Bortolotti et al., 2008;Warnock et al., 2010;Ashley et al., 2011;D'AnnaHernandez et al., 2011;Lattin et al., 2011;Bechshøft et al., 2012;Berkvens, 2013). The steroid content of these cornified tissues appears to reflect circulating hormone levels accumulated during tissue growth, e.g. human hair corti sol content reflects adrenal activity of the previous few months, whereas feather corticosterone content reflects adrenal activity that occurred during feather molt (Warnock et al., 2010;Lattin et al., 2011;Sharpley et al., 2011). The glucocorticoid content of these tissues correlates significantly with exposure to known stressors (Yamada et al., 2007;Van Uum et al., 2008;Fairbanks et al., 2011;Skoluda et al., 2012) and also with consequent physiological impacts, such as reduced growth and body condi tion (Lattin et al., 2011;Macbeth et al., 2012;Kennedy et al., 2013), reduced investment in offspring (Fairhurst et al., 2012), increased disease susceptibility (Pereg et al., 2011) and mortal ity risk (Koren et al., 2012). Most research to date has focused on the adrenal glucocorticoids, but preliminary results indicate that reproductive steroids are also measurable in cornified tis sues (Koren et al., 2012;Snoj et al., 2012;Caslini, 2013).
Baleen steroid analysis may offer two unusual benefits as a conservation physiology tool. First, baleen grows continuously from the whale's upper palate; newer baleen is added gradually at the base, and the oldest baleen is steadily worn off at the distal end (Best and Schell, 1996;Lubetkin et al., 2008;Young, 2012). In mature bowhead whales (Balaena mysticetus), for example, ~15-20 cm of baleen is added per year, and a single plate may be 4 m long; thus, one baleen plate from a mature bowhead may represent a continuous physiological record of approximately the past 20-25 years (Lubetkin et al., 2012;N. Lysiak, personal communication). (Immature animals have shorter baleen and faster baleen growth rate, and would be expected to have maximal baleen growth records of ~10-15 years depending on age; Lubetkin et al., 2012;N. Lysiak, personal communication). Baleen might therefore contain information on longitudinal profiles of past reproductive cycles and physiological events that occurred during the last decade or two of the whale's life. This is of particular interest for bowhead whales, in which reproductive cycles are still poorly understood; for example, current estimates of calving intervals are based on only three observed cases (Rugh et al., 1992). A second potential advantage of baleen is that there exist historical archives of baleen plates that date back to the era of commercial whaling (Bockstoce, 1986), as well as mod ern archives of baleen collected from whales that were stranded or taken in aboriginal subsistence whaling. Steroids in corni fied tissues often remain detectable in specimens stored long term; museum specimens of hair and feathers that are decades to centuries old have been found to have detectable steroid hormones that demonstrate physiologically relevant patterns (Webb et al., 2010;Bechshøft et al., 2012). Therefore, the pos sibility exists that baleen hormones could be used as a valuable tool for retrospective comparisons of steroid profiles in past vs. presentday populations, for example, comparison of inter calving intervals (Webb et al., 2010;Bechshøft et al., 2012;Kennedy et al., 2013).
The aim of this study was to evaluate the feasibility of baleen as a novel sample type for assessing patterns in stress related hormones (cortisol and its metabolites) and reproduc tive hormones indicative of calving cycles (progesterone and its metabolites). Our specific goals were as follows: (i) to develop methodology for pulverizing baleen and extracting steroids; (ii) to determine whether immunoreactive cortisol and progesterone are detectable in baleen samples; (iii) to validate cortisol and progesterone immunoassays for baleen using standard parallelism and accuracy tests; (iv) to assess patterns of baleen cortisol and progesterone at the base of the baleen plate in relationship to current physiological status (e.g. pregnant vs. nonpregnant females); and (v) to assess hormone variation at different locations along the baleen plate, as a preliminary investigation to determine whether baleen might contain a record of past physiological events.

Baleen samples
Baleen plates were collected between 2003 and 2012 from 16 subsistenceharvested bowhead whales in the towns of Barrow (n = 10), Kaktovik (n = 3), Savoonga (n = 2), and Gambell (n = 1), AK, USA (Fig. 1). Sampling was conducted under US National Marine Fisheries Service permit #17350 00. Whales were harvested during spring (n = 4) and autumn months (n = 12). Baleen plates were stored at ambient tem peratures in Barrow (annual temperature range −48 to 26°C, with ~320 days/year below 0°C) until analysis in 2011. One baleen plate from each whale was sampled, either by cutting a longitudinal strip of ~40 cm length from the base of the plate (n = 3 whales) or by using a DeWalt 18 V cordless power drill equipped with a circular hole saw to excise a disc of baleen ~2 cm in diameter (n = 13 whales). Note that 2 cm represents ~1.2-1.6 months of baleen growth (Lubetkin et al., 2008). For all whales, baleen was sampled as close to the base of the plate as possible (i.e. most recently grown baleen), but samples could not always be taken precisely from the base of the plate due to fraying of baleen at the base. In some samples, a thin dried layer of fibrous, nonbaleen tissue was adhered to the base of the baleen plate; this tissue is lighter in colour than baleen and was discarded whenever noted, but it is possible that some sam ples contain a small contribution from nonbaleen tissue.
For a subset of whales (n = 11), additional discs were excised with the drill centred at 10, 20 and 30 cm distal to the base (Fig. 2). This incremental sampling along the baleen plate was carried out longitudinally (i.e. along the growth axis) as a preliminary investigation into temporal variation in hormone content. These distances (base to 30 cm) represent ~1.6-2.0 years of baleen growth for most animals in our study (ani mals >10 m body length; see Lubetkin et al., 2012) and were selected to capture a potential endocrine record of current pregnancies as well as pregnancies of the year before. Given the known variation in baleen growth rate in bowheads, and given the 2 cm diameter of our samples, the four baleen sam pling locations are most likely to correlate to time of growth as follows: the 'base' sample represents baleen grown between 0.0 and 1.2 months before death; the '10 cm' loca tion repre sents baleen grown between 6.6 and 7.2 months before death; the '20 cm' location, between 12.6 and 15.2 months before death; and the '30 cm' location, between 18.6 and 23.2 months before death (ranges are based on 2 cm disc diameter and on 3 Figure 1: Locations in Alaska at which baleen samples were obtained from landed bowhead whales. Ten whales were sampled at Barrow, three whales at Kaktovik, two at Savoonga and one at Gambell.

Toolbox
Conservation Physiology • Volume 2 2014 known baleen growth rate data for bowheads of >10 m body length; Lubetkin et al., 2008;N. Lysiak, personal communica tion). One animal in our study was <10 m body length, a male of 7.8 m body length; this individual may have had a faster baleen growth rate of up to 25 cm/year (Lubetkin et al., 2008).
The final sample size was 49 subsamples from 16 baleen plates (Table 1). All subsamples (i.e. strips and discs) were shipped to our laboratory at the New England Aquarium (Boston, MA, USA) for analysis.

Sex and maturity
Baleen plates were from six males and 10 females ( Table 1). Four of the females were pregnant at the time of harvest. Sexual maturity and reproductive state were defined on the basis of body length and relevant anatomical findings at har vest (i.e. testicle size, ovarian morphology), as follows: imma ture males <13.5 m body length; mature males >13.5 m; immature females <13.5 m; mature females >13.5 m or large ovarian follicles present; and pregnant females were identified by presence of a fetus and/or placenta at harvest, or (if uterus not opened) large corpus luteum and enlarged uterus (Koski et al., 1993;George et al., 1999;O'Hara et al., 2002). Final sample sizes were five immature males (the largest three of which were likely pubertal; O'Hara et al., 2002), one mature male, three immature females, three mature nonpregnant females and four pregnant females (Table 1).

Baleen sample preparation and steroid extraction
Sample preparation and extraction methods were initially tested on a single baleen plate from a North Atlantic right whale (NARW, Eubalaena glacialis). Isopropanol rinses (e.g. Davenport et al., 2006;Ashley et al., 2011;Bechshøft et al., 2011) were not used, because pilot trials showed that such rinses had no effect on apparent cortisol content of NARW baleen (see Supplementary Information).
We tested the following nine different methods of pulveriza tion on NARW baleen: (i) handmincing with scissors (follow ing Davenport et al., 2006); (ii) mortar and pestle; (iii) mortar and pestle with liquid nitrogen; (iv) handmincing followed by mortar and pestle; (v) commercial blade grinder; (vi) commer cial tobacco grinder; (vii) commercial spice grinder; (viii) labo ratory tissue homogenizer (MP Biomedicals FastPrep 24); and (ix) rotary electric grinder (Dremel Model 395 Type 5) with a flexible handheld extension and tungstencarbide ball tip. The Dremel produced the finest baleen powder with the highest apparent cortisol concentration (see Supplementary  Information), indicating superior extraction of hormones. Subsequently, all bowhead whale baleen samples were pulver ized with the Dremel grinder, with the resulting baleen powder collected on a weigh paper underneath. For the 2cmdiameter baleen discs, several areas around the disc were pulverized to obtain a representative sample; for the 40cmlong strips, a 2cmdiameter area at the base was pulverized.
Pilot trials on 12 bowhead samples indicated that 50 mg samples of baleen powder sometimes had undetectably low cortisol, whereas a 100 mg sample mass always had detect able cortisol (see Supplementary Information). Therefore, 100 mg of wellmixed baleen powder from each sample was weighed to the nearest 0.0001 g, poured into a 16 mm × 100 mm borosilicate glass tube, capped, and stored at room temperature until extraction within 2 weeks.

Steroid extraction from baleen
Extraction of steroids from baleen powder followed a consen sus protocol derived from the hair and feather glucocorticoid literature (e.g. Davenport et al., 2006;Bortolotti et al., 2008). Briefly, 4.0 ml of 100% methanol was added to 100 mg of baleen powder, vortexed for 20 h at room temperature and centrifuged for 15 min at 4000g. The supernatant was pipetted into a separate borosilicate glass tube for drying, and the pellet was rinsed twice more to recover additional hormone. For each rinse, 1.0 ml of 100% methanol was added to the pellet, vortexed for 30 s, centrifuged for 15 min at 4000g, and the supernatant was transferred to the drying tube. The combined supernatant was dried under air blow in a ThermoScientific ReactiTherm III set at 45°C. When the volume reached <0.5 ml, another 1.0 ml methanol was added to rinse down the walls of the tube, and the sample was again dried. Once fully dry, samples were reconstituted in 0.5 ml of assay buffer from the progesterone assay kit, transferred to a cryovial and stored at −20°C until assay within 1 week of extraction.
Progesterone assay parallelism was tested using three sep arate baleen extract pools from pregnant females, non pregnant females and males. These three pools were serially diluted in assay buffer and assayed alongside knowndose progesterone standards. Due to low cortisol content in most baleen samples, cortisol parallelism was tested using a single pool from samples containing the highest cortisol content in 5 Abbreviations: F, female; M, male. The lengths 0, 10, 20 and 0 cm are distances from the base of the baleen plate; 0 cm is the most recently grown baleen and 30 cm the oldest baleen tested. a 11B3 categorized as pregnant based on presence of large corpus luteum and enlarged uterus; however, abdomen was not opened at time of harvest, abdominal organs were frozen the next day, and presence of a fetus could not be confirmed.
a pilot assay. This 'highcortisol' pool was serially diluted in buffer and assayed alongside the cortisol standard curve.
Results were plotted as the percentage bound vs. the loga rithm of relative dose, and the slopes were compared for par allelism of the linear portion of the curve. Based on parallelism results, all subsequent samples were assayed at 1:4 for pro gesterone and at 1:1 (fullstrength extract) for cortisol; these dilutions were chosen to fall near the 50% bound on the standard curve, the area of greatest assay precision.
Assay accuracy was assessed by spiking standard curves with an equal volume of a 'lowprogesterone' pool (samples that had low progesterone in a pilot assay) diluted to 1:4, or a 'lowcortisol' pool (samples that had low cortisol in a pilot assay) at 1:1, and assaying alongside standards spiked only with assay buffer. Results were plotted as the observed dose vs. known standard dose and assessed for linearity, slope and yintercept.
All assays were performed in duplicate, with a full stan dard curve in each assay, nonspecific binding wells in qua druplicate and 'zero' (blank) wells in quadruplicate. Any samples with >10% coefficient of variation between dupli cates, or with the percentage bound <10 or >90%, were redi luted accordingly and reassayed. Multiple samples from the same whale (e.g. base, 10, 20 and 30 cm samples) were assayed simultaneously in the same EIA plate, except for an initial set of 12 base samples that were assayed on a separate EIA plate.

Statistical analysis
Data were analysed using InStat 3.0b for Macintosh OSX (GraphPad Software Inc., San Diego, CA, USA). Assay paral lelism was assessed with Ftests on data for the linear portion of the curves. Assay accuracy was assessed as follows: (i) good nessoffit of linear regression line on expected vs. observed dose (e.g. r 2 should ideally be >0.90); (ii) slope within 0.7-1.3; and (iii) good match of yintercept to the apparent dose when the pool was assayed alone. Cortisol data were normally distributed and therefore analysed with parametric tests (Zar, 2009). Progesterone data were highly skewed and could not be normalized with common transformations, and were anal ysed with nonparametric tests (Conover, 1999). Differences between males vs. females and nonpregnant vs. pregnant females were analysed with Student's ttests or Mann-Whitney Utests. Patterns in hormones at different locations along a baleen plate (base, 0, 20 and 30 cm) were analysed with repeatedmeasures ANOVA (with Tukey-Kramer post hoc tests) or Kruskal-Wallis tests. Data are presented as means ± SEM for normally distributed data or medians for nonnormal data. The significance level was set at α = 0.05.
Cortisol content was significantly higher in the base sam ples than in three distal samples (older baleen at 10, 20 and 30 cm), but the distal samples from the same animals were not significantly different from each other (withinindividual; F 3,10 = 12.166, P < 0.0001; P < 0.05 for post hoc comparisons of base vs. any other location; Table 1). Furthermore, in con trast to the sex difference seen in the base samples, males and females did not exhibit significant differences in cortisol at the 10, 20 or 30 cm locations.

Baleen progesterone
Progesterone (or immunoreactive progesterone metabolites) was detectable in all baleen samples (Table 1). Similar to cortisol, females had significantly higher base progesterone than males, and this difference disappeared if pregnant females were excluded (females, median = 14.65 ng/g; males, median = 6.33 ng/g; U = 9.000, P = 0.023; pregnant females excluded and mature animals only, t 5 = 0.6908, P = 0.5204; Fig. 5). All pregnant females had higher progesterone in the base sample than all nonpregnant females, and this difference was significant (U = 24.000, P = 0.0095).
In the baleen plates that were subsampled at multiple loca tions, all four males and both immature females exhibited consistently low progesterone (<20 ng/g) at all locations (Table 1 and Fig. 5). These six animals had slightly, but sig nificantly, higher progesterone at the base location compared with the other three locations (F 3,5 = 11.007, P = 0.0004; Fig. 5). In contrast, four of five mature females showed extreme variation in progesterone, spanning one to three orders of magnitude within a single individual, at different 6 locations along the baleen plate (Table 1 and Fig. 5). Mature females exhibited no relationship of progesterone with baleen location (Fr 5,4 = 2.040, P = 0.6522), and progesterone was occasionally very high even in the oldest baleen tested (Table 1). In the two pregnant females that were tested at multiple baleen locations (whales 10S1 and 11B3), both had extremely high progesterone at the base and 10 cm locations, and the smaller whale (10S1) also had very high progesterone at the 20 cm location.

Discussion
Our results indicate that immunoreactive cortisol and pro gesterone are measurable in bowhead whale baleen using commercially available immunoassay kits. Both hormones appear to reflect physiological parameters of interest, such as sex and reproductive state, with baleen progesterone exhibit ing variation consistent with occurrence of past pregnancies. To our knowledge, this is the first demonstration of successful

Baleen cortisol
The higher baleen cortisol levels found in females compared with males, at the base of the plate, is reminiscent of sex differ ences in circulating glucocorticoids seen in many mammals (Tilbrook et al., 2000;Panagiotakopoulos and Neigh, 2014). Sex differences in cortisol have been documented in other cor nified tissues as well (e.g. polar bear hair, Bechshøft et al., 2011;Macbeth et al., 2012;human hair, Dettenborn et al., 2012). In the present study, the highest baleen cortisol levels detected were in pregnant females. Female mammals commonly have elevated glucocorticoids during pregnancy (Foley et al., 2001;D'AnnaHernandez et al., 2011), and a similar pattern has been documented in faecal glucocorticoids of North Atlantic right whales (Hunt et al., 2006). Although preliminary, our results indicate that baleen, at least at the base, may preserve a hor monal signature of adrenal activity in bowhead whales.
Cortisol concentrations were markedly lower in samples taken from older baleen (10, 20 and 30 cm) compared with base samples. This pattern was also noted in progesterone data, but to a lesser degree. Cortisol is one of the most polar steroids, whereas progesterone is relatively nonpolar; it may be that cortisol is more likely to leach from baleen into sea water. Human hair loses some cortisol upon immersion in hot water (Li, 2012), and it is unknown whether a similar process might occur in baleen immersed in cold seawater over months. However, any such leaching does not seem to continue beyond 10 cm, because the 10, 20 and 30 cm sam ples had similar cortisol concentrations. If loss of hormone to seawater does occur, baleen from the core of the plate may be protected from any such loss. Our samples consisted of a mix of surface and interior baleen; future research could compare surface vs. core baleen separately. Additionally, assay of other adrenal steroids that are less polar than cortisol, such as cor ticosterone and aldosterone, may prove fruitful for distin guishing periods of higher adrenal activity in older baleen.

Baleen progesterone
The dramatic variation in baleen progesterone profiles of the mature females suggests that baleen may contain an endo crine record of pregnancy. The progesterone profiles seen in individual females are consistent with estimates of bowhead gestation length and baleen growth rate (Koski et al., 1993;Lubetkin et al., 2008). Baleen growth rate in bowheads slows with age; for the animals sampled in this study, baleen growth rate was estimated at ~19-20 cm/year for the smaller, younger females and ~17-18 cm/year for the largest and oldest female (Lubetkin et al., 2008(Lubetkin et al., , 2012C. George, personal observa tion;N. Lysiak, personal communication). For example, in the smaller pregnant female, 10S1, the 20 cm sample was probably grown ~12-13 months prior to death, while in the larger pregnant female, 11B3, the 20 cm sample was probably grown ~13.5-15 months prior to death. Given an estimated gestation length of 13-14 months (Koski et al., 1993), it therefore makes sense that the smaller female had high progesterone that spanned three locations (base, 10 and 20 cm), i.e. her 20 cm sample was probably grown during pregnancy. In contrast, the larger female had high progester one that spanned only two locations (base and 10 cm, but not 20 cm); thus, her 20 cm sample may represent prepreg nancy. Although these patterns are preliminary, they suggest that baleen may retain an endocrine signature of prior preg nancies and that baleen should be investigated further as a potential retrospective record of past reproductive cycles.
We note, however, that there was one pregnant female (07B12; Table 1) that had much lower base progesterone than the other pregnant females (though still higher than all non pregnant females; Table 1). This female was in early preg nancy, as determined by small fetal size (C. George, personal observation). Lower progesterone levels might be characteris tic of early pregnancy. Alternatively, it is possible that the base sample of this female may not have been taken precisely from the growth zone, due to fraying at the base of the plate (see Materials and methods) and may represent prepregnancy.
Overall, these data indicate that baleen progesterone should be investigated further as a retrospective record of prior reproductive cycles in bowhead whales. Baleen hor mone data could be coupled with ovarian analysis (e.g. cor poraalbicans counts; George et al., 2011) and may provide an alternative, and muchneeded, method of estimating inter calving intervals and reproductive cycles in baleen whales. Sampling at more locations along the baleen plate, ideally coupled with stable isotope analysis to discern annual cycles (Lubetkin et al., 2008), could help match adrenal and repro ductive hormone profiles to season and year.

Future methodological research
The processing and extraction methods presented here pro vide an introduction to the technique, but should not be regarded as a final protocol. Future experiments should include tests of other extraction methods (e.g. methanol vs. ethanol, vortex duration, extraction temperature), with the goal of maximizing extraction of steroids from baleen pow der. Highperformance liquid chromatography or related techniques could be used to confirm the chemical identity of steroids present in baleen, as it is possible that immunoreac tive steroids detected by EIA antibodies are not necessarily pure parent hormones; unusual tissue types sometimes con tain uncommon steroid metabolites and/or conjugated steroids (Wasser et al., 1988). Furthermore, additional anti bodies and other quantification methods should be tested. Other steroids could be tested as well, particularly testoster one, estradiol, aldosterone and corticosterone. Variation in hormone content in different sections of the baleen plate should also be investigated, e.g. lingual vs. labial edge, dorsal surface vs. ventral surface and surface vs. core of plate. Potential contributions from any nonbaleen dried tissue that adheres to the sample should be evaluated, as should variation between plates from different locations in the mouth (left vs. right side, anterior vs. posterior plates).

Conclusions
In conclusion, baleen steroid analysis represents a new tool that shows great potential for retrospective assessment of patterns of physiological stress and reproductive cycles of baleen whales. Of particular interest is the availability of historical baleen samples in museum archives (samples collected during the era of com mercial whaling) that could be used for comparisons with pres entday population data. Additionally, if this method proves reliable, continued collection of baleen from modern popula tions may allow population changes to be tracked through time (e.g. potential shifts in average intercalving intervals), as well as assessment of potential impacts of ongoing climate change and increasing human activity in the Arctic. In sum, hormone analy sis of baleen could provide an innovative means to evaluate longterm trends of stress and reproduction in whale popula tions exposed to a changing marine environment.

Supplementary material
Supplementary material is available at Conservation Physiology online.