Role of Testosterone in Stimulating Seasonal Changes in a Potential Avian Chemosignal
- First Online:
- Cite this article as:
- Whittaker, D.J., Soini, H.A., Gerlach, N.M. et al. J Chem Ecol (2011) 37: 1349. doi:10.1007/s10886-011-0050-1
- 193 Views
Songbird preen oil contains volatile and semivolatile compounds that may contain information about species, sex, individual identity, and season. We examined the relationship between testosterone (T) and the amounts of preen oil volatile and semivolatile compounds in wild and captive dark-eyed juncos (Junco hyemalis). In wild males and females, we observed an increase in volatile compound relative concentration early in the breeding season. This increase mirrored previously described seasonal elevation in T levels in wild males and females, suggesting a positive relationship between hormone levels and preen gland secretions, and a possible role for these secretions in signaling receptivity. In females, the greatest relative concentrations of most compounds were observed close to egg laying, a time when steroid hormones are high and also the only time that females respond to an injection of gonadotropin-releasing hormone with a short-term increase in T. In a study of captive juncos held on short days, we asked whether the seasonal increases observed in the wild could be induced with experimental elevation of T alone. We found that exogenous T stimulated the production of some volatile compounds in non-breeding individuals of both sexes. However, of the 15 compounds known to increase during the breeding season, only four showed an increase in relative concentration in birds that received T implants. Our results suggest that testosterone levels likely interact with other seasonally induced physiological changes to affect volatile compound amounts in preen oil.
Key WordsTestosteroneChemical communicationChemical signalsBirdsDark-eyed juncoPasserineSteroid hormonesPreen gland
Many species of vertebrates, including birds, produce chemical signals that may play an important role in social and reproductive behavior (Wyatt, 2003). The use of chemical communication by birds was doubted until recent years, when several studies began to decode information contained in these chemosignals (e.g., Mardon et al., 2010; Whittaker et al., 2010; Shaw et al., 2011), and provided evidence of the ability of birds to detect these chemicals (Bonadonna and Mardon, 2010; Whittaker et al., 2011) and a role of these chemicals in social and reproductive behavior (reviewed in Balthazart and Taziaux, 2009). In most species of birds studied to date, compounds that may play a role in chemical signaling are present in preen oil secreted from the uropygial, or preen, gland (Bonadonna et al., 2007; Soini et al., 2007). Birds spread this oil over feathers while preening, where it functions to protect the feathers from exposure to the environment, enhance insulation, and lower ectoparasite load (Jacob and Ziswiler, 1982). The volatile and semivolatile compounds in preen oil also contribute an odor to a bird, which varies qualitatively among species (Mardon et al., 2010), and quantitatively within species, between sexes, populations, individuals (Soini et al., 2007; Mardon et al., 2010; Whittaker et al., 2010), and age classes (Shaw et al., 2011).
In seasonally reproducing species, many physiological and behavioral changes occur during the transition from non-breeding to breeding condition. Many of these changes are mediated by steroid hormones induced by changes in photoperiod (Farner and Wingfield, 1980). Our study species, the dark-eyed junco (Junco hyemalis), is a North American passerine bird that breeds in the summer months. In juncos, like many seasonally breeding birds, circulating testosterone (T) levels increase sharply early in the breeding season, drop when the female lays eggs, and stay low while both parents care for the offspring (Ketterson et al., 2001, 2005). This pattern is observed in both sexes, although T levels typically are higher in males than females at most or all stages of breeding (Ketterson et al., 2001, 2005). Throughout the breeding season, males respond to an injection of gonadotropin-releasing hormone (a “GnRH challenge”) with a short-term increase in T (Jawor et al., 2006). The only time that females respond this way is during the 7 days before egg laying, when they are depositing yolk in eggs (Jawor et al., 2007).
In juncos, concentrations of preen oil volatile and semivolatile compounds are higher in captive birds held in long-day, compared to birds in short-day photoperiods, especially for the linear 1-alkanols, 1-undecanol through 1-hexadecanol, the methyl ketones, 2-undecanone through 2-pentadecanone, and the carboxylic acids, dodecanoic acid, tetradecanoic acid, and hexadecanoic acid (Soini et al., 2007). For convenience, we henceforth will refer to both volatile and semivolatile compounds as “volatile compounds,” but acknowledge that the heavier compounds with lower vapor pressures, such as hexadecanoic acid, are considered “semivolatile.” Males have significantly higher proportions of two methyl ketones, 2-tridecanone and 2-pentadecanone, in preen oil, while females have a blend with higher proportions of 1-undecanol and the three carboxylic acids above (Whittaker et al., 2010). The function of these compounds is not yet clear, but they appear capable of transmitting information about identity that could be used in mate choice (Mardon et al., 2010; Whittaker et al., 2010). Juncos in breeding condition detect these odors and exhibit preferences based on sex and individual characteristics of the bird that produced the preen oil (Whittaker et al., 2011).
This relationship among volatile compound concentrations, sex, and reproductive condition suggests that changes in preen oil concentration may be linked to hormone levels, even though the role of hormones in mediating changes in avian chemosignal production is not well understood. The uropygial gland contains androgen receptors (Daniel et al., 1977; Shanbhag and Sharp, 1996), and treatment with exogenous T increased rates of protein synthesis in the uropygial gland (Maiti and Ghosh, 1972; Amet et al., 1986). The similarity of the avian uropygial gland and the testosterone-dependent mammalian preputial gland has been noted previously (Maiti and Ghosh, 1972). The mammalian preputial gland can be the source of pheromones related to reproductive behavior, such as the farnesenes secreted by the mouse preputial gland (Novotny et al., 1990). Certain volatile compounds and proteins produced by the rat preputial gland are stimulated by testosterone (Ponmanickam et al., 2010). These observations suggest that, in birds, T might stimulate production of volatile compounds in the uropygial secretion. However, in the only study to examine this relationship so far, concentrations of volatile carboxylic acids decreased in the preen oil of gray catbirds receiving T treatment (Whelan et al., 2010).
We present two studies designed to elucidate the relationship between variation in steroid hormone levels and concentrations of preen oil volatile compounds. The first studied the increase in preen oil volatile compounds during the early spring, when juncos return to breeding grounds and undergo physiological and behavioral shifts into breeding condition. We asked whether the preen oil of wild juncos would have the same or different composition as captive junco preen oil. We also asked whether the effect of day length on preen oil concentration observed in the laboratory (Soini et al., 2007) was repeated under field conditions, and whether any changes would correlate with the rise in testosterone that is known to occur during early spring.
In the second study, we again used male and female juncos in captivity. The birds were held on short days, and had T levels elevated experimentally in order to test whether T alone, without the influence of increased photoperiod, stimulated production of volatile compounds in preen oil similar to the increases observed during long days. We hypothesized that birds with experimentally elevated T levels would show increases in preen oil volatile compound concentrations. We also asked whether birds with higher levels of T would have greater increases in concentrations of those compounds.
Methods and Materials
Our study subjects were wild and captive dark-eyed juncos (Junco hyemalis carolinensis) from Mountain Lake Biological Station (MLBS), Pembroke, VA, USA (previous studies of chemical communication in juncos have included this subspecies as well as J. h. thurberi in southern California and J. h. aikeni from South Dakota: Soini et al., 2007; Whittaker et al., 2010, 2011). This population has been the subject of a long-term behavioral and endocrinological study since 1983 (Ketterson et al., 2001, 2005; Nolan et al., 2002). From about April 15 to May 15 of every year, juncos are captured using mist nets and traps to census returning individuals and to band new birds with USFWS bands. From May 15 to July 15, we intensively search for nests and monitor them from laying to hatching to fledging.
Study 1: Early Season Profiles of Preen Oil Volatile Compound Relative Concentrations in Wild Males and Females
We collected preen oil from every bird captured opportunistically during the early season census from April to May 16, 2008. Sex was determined by observation of a brood patch (females) or cloacal protuberance (males), and also by plumage and wing length (Nolan et al., 2002). We collected preen oil by gently rubbing the uropygial gland with a 100 μl glass capillary tube (Drummond Scientific, Broomall, PA, USA), which stimulates the gland to secrete 1–3 mg of preen oil (Whittaker et al., 2010). Most samples were collected soon after capture, to minimize the impact of stress hormones. However, unlike circulating hormone levels in blood, which are known to change very quickly (less than 15 min), preen oil content is not expected to change over such a short time period. We have observed, when we collect preen oil from a bird, it can take more than 24 h before that individual has enough preen oil to sample again (Whittaker, personal observation). Preen oil was stored at −20°C within 10 min of collection and kept frozen until analyzed by gas chromatography–mass spectrometry (GC-MS), as described below. We analyzed the volatile compound content of preen oil from all birds captured on at least three different days over the course of this period, for a total of 16 females and 35 males. To compare the change in compounds over time, we binned the collection dates into 4 weeks, and used week as a categorical variable. We measured GC-MS peak areas for 15 volatile compounds that are known to differ between males and females (Whittaker et al., 2010), and between breeding and non-breeding season (Soini et al., 2007). The data were not normally distributed, so we natural-log transformed peak areas. We used a multivariate analysis of variance (MANOVA) to compare ln-transformed peak areas of each compound across the 4 weeks, and used Fisher’s LSD post-hoc test to determine when changes occurred. All statistics were two-tailed.
Study 2: Response of Preen Oil Volatile Compounds in Captive Male and Female Juncos Held on Short Days to Experimental Elevation of Testosterone
In October 2009, we captured adult juncos at MLBS using mist-nets and transported them to Indiana University, Bloomington, where they were held in an indoor, climate-controlled aviary on a natural light:dark cycle. Birds were given ad libitum access to water, millet and sunflower seeds, mealworms, orange slices, and a mixture of puppy chow, hardboiled eggs, and carrots (Whittaker et al., 2010). The temperature was maintained at 60–65°C, and photoperiod was set to match the daylight schedule in Pembroke, VA, until late December 2009, when we fixed the light:dark period at 10 L:14D. During short days, juncos are in non-breeding condition and have low levels of T and low concentrations of preen oil volatile compounds (Ketterson et al., 2005; Soini et al., 2007).
Males and females were randomly assigned to one of three treatment groups: control (C), which received empty implants made from 7 mm Silastic ® tubing (Dow Corning, 1.47 mm i.d., 1.96 mm o.d.); low T, which received implants packed with 2.5 mm (females) or 5 mm (males) of crystalline T (Sigma-Aldrich, Inc.); and high T, which received implants with 5 mm (females) or 10 mm (males) of T (Ketterson et al., 1991, 1992). The high T doses were selected to produce circulating T levels similar to natural peaks observed during the breeding season (Ketterson et al., 1991, 1992, 2005; Clotfelter et al., 2004). The implants were inserted subcutaneously along the left flank after anesthetizing birds with isofluorane.
There were a total of 42 birds in the study: 24 males (8 C, 8 low T, 8 high T) and 18 females (5 C, 6 low T, 7 high T). Sex was confirmed using a molecular sexing technique in which we amplified fragments of the CHD gene on the W and Z chromosomes (Griffiths et al., 1998).
Birds were housed in individual cages across three rooms. Birds were assigned to cages and rooms randomly with respect to treatment, but we maintained similar sex ratios across all three rooms (F:M ≈ 0.8). The birds were able to see and hear each other. A blood sample was taken 1 week before, and 1 week after, implantation to assay for T concentration. We took preen oil samples from the implanted birds 1 week before implantation, and 2 and 4 weeks after implantation (Dec. 28, 2009 and Jan. 11, 2010).
Although randomly assigned to treatment groups, differences in pre-implant preen oil volatile relative concentrations were observed between control males and males assigned to one of the testosterone treatments. Therefore, we restricted our analysis to the change in ln-transformed volatile compound relative concentrations at 2 and 4 weeks after implant, relative to pre-implant values. Because the data were not normally distributed, we compared treatment groups using nonparametric tests (Kruskal-Wallis for comparing three groups, Mann–Whitney U test for comparing two groups). All statistics are two-tailed.
We measured T levels in plasma samples using a commercial enzyme immunoassay (EIA) kit (Assay Designs #901-065). Assay methods are described in detail elsewhere (Clotfelter et al., 2004). The primary change to the manufacturer’s protocol is that we added 2000 cpm of tritiated testosterone to the samples in order to calculate extraction efficiencies. We determined testosterone concentrations with a 4-parameter logistic curve-fitting program (Microplate Manager; BioRad Laboratories, Inc.) and corrected them for incomplete recoveries.
Gas Chromatography–Mass Spectrometry (GC-MS)
Our GC-MS method has been described previously (Soini et al., 2007; Whittaker et al., 2010). We added an internal standard (8 ng of 7-tridecanone, Sigma-Aldrich Chemical Company, Milwaukee, WI, USA) dissolved in 5 μl of methanol (Baker Analyzed®, Mallinckrodt Baker, Inc., Phillipsburg, NJ, USA) to samples, and extracted volatile compounds with the Twister® stir bar (10 × 0.5 mm polydimethylsiloxane, PDMS) for 60 min (Gerstel GmbH, Mülheim an der Ruhr, Germany). After extraction, the stir bar was rinsed with high-purity OmniSolv® water, dried with a paper tissue, and placed in the thermal desorption autosampler tube. Quantitative analysis was performed using an Agilent 6890N gas chromatograph connected to a 5973i MSD mass spectrometer (Agilent Technologies, Inc., Wilmington, DE, USA) with a Thermal Desorption Autosampler and Cooled Injection System (TSDA-CIS 4 from Gerstel). Samples were thermally desorbed in a TDSA automated system, followed by injection into the column with a cooled injection assembly, CIS-4.
All major compounds were identified by comparison to standard substances obtained from Sigma-Aldrich, using mass spectra and retention times (Soini et al., 2007). Peak areas of the identified compounds were used for quantitative comparisons among the groups. Peak areas were integrated either from the TIC (total ion current) profiles or from the post-run selected ion current (SIC) profiles at m/z 55 for linear 1-alkanols, m/z 58 for 2-ketones and m/z 60 for carboxylic acids. Peak areas of the internal standard (7-tridecanone) were integrated from the post-run m/z 113 profiles. Peak areas of the compounds of interest were normalized by dividing each peak area by that of the internal standard in corresponding runs, yielding relative concentrations (i.e., relative amounts per 100 μl of preen oil). Relative standard deviation (RSD, a measure of reproducibility) of the internal standard peak area was 13% (N = 12).
All work was conducted in compliance with the Bloomington Institutional Animal Care and Use Committee guidelines (BIACUC protocol 09-037).
Study 1: Early-Season Preen Oil Volatile Compound Profiles
We confirmed that wild junco preen oil had the same volatile compound composition as captive junco preen oil. We identified the same 19 volatile compounds in wild junco preen oil as found in captive juncos (Soini et al., 2007; Whittaker et al., 2010). Four of the compounds (1-nonanol, 1-heptadecanol, nonanoic acid, and decanoic acid) were below detectable levels in many individuals and were subsequently left out of our statistical analyses; three of these (1-nonanol, nonanoic acid, and decanoic acid) were shown previously to have low intra-individual repeatability (Whittaker et al., 2010).
The pattern in female juncos was different: no change in any compound was observed between weeks 1 and 2, but all volatile compounds had increased by week 3 (P < 0.05, all comparisons; Fig. 1, right side, P > 0.1).
Study 2: Experimental Elevation of T in Captive Birds Held on Short Days
Males (N = 24) had pre-implant T levels of 0.52 ± 0.15 ng.ml−1 (mean ± SD), and post-implant T levels of 0.73 ± 0.26 in C males (N = 8), 5.11 ± 1.48 in low T males (N = 8), and 11.60 ± 3.09 ng.ml−1 in high T males (N = 8). Females (N = 18) had pre-implant T levels of 0.73 ± 0.58 ng.ml−1, and post-implant T levels of 0.60 ± 0.17 in C females (N = 5), 5.05 ± 0.84 in low T females (N = 6), and 8.11 ± 3.27 ng.ml−1 in high T females (N = 7). These post-implant T levels were different among the three treatment groups within each sex [ANOVA, F = 60.87 (males) and 18.69 (females), df = 2, P < 0.001 (both comparisons)].
We tested whether testosterone stimulates the production of volatile and semivolatile compounds present in preen oil, and whether the concentrations of these compounds varied in relation to an individual’s testosterone levels. We found that volatile compounds increased in concentration during the early breeding season, in tandem with the previously described seasonal increase in circulating levels in T (Ketterson et al., 2005) and seasonal response to GnRH (Jawor et al., 2006, 2007) in wild male and female juncos. Our data showed a concentration peak for females in the same week that most females in the population laid their first egg (week 3), as would be expected if steroid levels influence volatile compound production. In males, we observed a steady increase across the 4 weeks, with volatile compound concentration higher in the fourth than in the first week. Previous studies found that males also experience a drop in T when their mate begins incubating eggs (Ketterson et al. 2005), but the drop is not as sharp as in females. It may be that the males in our study had not yet experienced this decrease in circulating T, or that the decrease was smaller and did not affect volatile compound production. The volatile compounds we observed in wild junco preen oil were the same as those previously described in captive junco preen oil (Soini et al., 2007; Whittaker et al., 2010).
Treatment with exogenous testosterone stimulated the production of some, but not all, preen oil volatile compounds and, for some compounds, birds treated with a high dose of T did not have greater concentrations of preen oil volatile compounds compared to birds that received half the dose. These data suggest that while T in the absence of long days is sufficient to stimulate the production of some volatile compounds, it is not positively correlated with the concentrations of individual compounds. We did not detect differences between the sexes in the implant study, likely due to the fact that both males and females in non-breeding condition are known to display low levels of all preen oil volatile compounds (Soini et al., 2007), and the only manipulation in the study was the circulating level of T. The only volatile compound that is typically in greater concentrations in males than in females, and that responded significantly to T treatment, was 2-undecanone. In another study (Whittaker et al., 2011), we found a positive correlation between the relative proportions of “male-like” volatile compounds (those that make up a greater proportion of the volatile compounds present in male preen oil compared to female preen oil) and a proxy for body size. Thus, there may be another physiological factor involved in producing a male-like chemical profile, and these other factors may not be directly under T control, unlike volatile compounds related to reproductive behavior in rats (Ponmanickam et al., 2010).
Changes in photoperiod affect the production of many different hormones, besides T, which may play a role in stimulating volatile compound production. For example, exposure to light inhibits the production of melatonin from the pineal gland. In mammals, melatonin has been suggested to play a role in regulating the hypothalamic-pituitary-gonadal (HPG) axis, affecting the timing of the production of steroid hormones (Tamarkin et al., 1985). Although melatonin does not appear to be responsible directly for timing reproductive activity in birds, it does play a role in seasonal changes in immune function and song production (Bentley et al., 1999; Bentley, 2001). Thyroid hormone also has been shown recently to play a role in seasonal changes in both mammals and birds (Watanabe et al., 2007). The short photoperiod in the experimental study likely was associated with increased melatonin levels in the birds and decreased levels of the bioactive form of thyroid hormone (T3), which could have affected the changes in volatile compound concentrations.
The uropygial gland has been described as “androgen-dependent,” due to both the high concentration of androgen receptors and the previously observed responses of the gland to treatment with testosterone (Maiti and Ghosh, 1972; Abalain et al., 1983; Amet et al., 1986). In addition to androgen receptors, the uropygial gland contains receptors for other steroid hormones, including receptors for estradiol (Daniel et al., 1977). In order for T to influence target tissues, it must, sometimes, first be converted to an active form, such as dihydrotestosterone (DHT), through the action of 5α-reductase, or estradiol, through the process of aromatization (Massa, 1984). The seasonal changes observed in wild females could have been related to changing estradiol levels, and exogenous T may have been converted to estradiol. Steroid hormone metabolism can take place in the brain to affect behavior (Riters et al., 2001), as well as in peripheral tissues (Edwards et al., 2005), and the abundance of the enzymes involved in this conversion can vary seasonally (Soma et al., 2003). Thus, the inability of our T implant study to replicate volatile compound concentrations observed in breeding birds may be related to differences in concentration of other hormones like melatonin and thyroxine, or a lack of 5α-reductase in males, the quantity of which is known to be decreased in non-breeding birds (Soma et al., 2003).
Other steroid hormones may also play a role in regulating volatile compounds. In the crested auklet, a seabird that produces a citrus-like odorant that does not originate from preen oil, odorant emission was related to levels of the steroid hormone progesterone in males, but not in females (Douglas et al., 2008). Relatively few progesterone, compared to androgen, receptors have been found in the uropygial gland, but progesterone and androgens are known to exhibit cross affinity for receptors (Daniel et al., 1977). Corticosterone, a steroid hormone related to stress, also has been shown to have an effect on cellular activity in the uropygial gland (Maiti and Ghosh, 1969), and corticosterone levels are known to be correlated with T levels in juncos (Ketterson et al., 1991). Thus, a better understanding of the relationship between preen oil volatile compounds and progesterone or corticosterone levels may be necessary to understand seasonal changes in volatile profiles.
While our results shed light on the role of testosterone on preen oil composition, the role of these volatile compounds in male preen oil remains unknown. Behavioral data suggest female and male juncos are attracted to the scent of males (Whittaker et al., 2011), indicating potential behavioral roles for these compounds. However, their function in females is less clear. The positive relationship between T and “female-like” volatile compounds (those that are typically present in higher proportions in female preen oil) suggests that these compounds could function as a signal of ovulation or receptivity, when T levels are highest. In many species of birds, female T levels peak before or during nest-building (Ketterson et al., 2005), and treatment with T can induce sexual receptivity in many female vertebrates (Adkins-Regan, 1981), most likely as a result of aromatization of T into estradiol (Winkler and Wade, 1998). Data from other avian species also suggest that females produce an olfactory signal that is attractive to males. In domestic chickens, experimental removal of the female’s uropygial gland resulted in significantly fewer copulation attempts by males, suggesting that this gland is the source of an attractive signal (Hirao et al., 2009). Female mallard ducks in breeding condition produce preen oil compounds that have been described as pheromones, and these compounds can be induced in nonbreeding males and females by injecting them with exogenous estradiol (Bohnet et al., 1991).
Future research should focus on the action of additional hormones, including estradiol, progesterone, thyroxine, and melatonin, on preen oil volatile compound concentrations. In particular, the relationship between female breeding condition and chemical signals, which is well established in some mammals, should be further investigated in birds.
We thank Ryan Kiley, Christine Bergeon Burns, Kim Rosvall, and Mark Peterson for assistance in capturing juncos and maintaining them in captivity. Thanks to Jim Goodson for lending equipment and Kevin McLane for help processing samples. We thank Greg Demas for comments on an earlier draft. All work was conducted in compliance with the Bloomington Institutional Animal Care and Use Committee guidelines (BIACUC protocol 09-037) and with permission from the US Department of Fish and Wildlife, the Virginia Department of Game and Inland Fisheries, and the US Forest Service. This research was supported by the Indiana Academy of Sciences, and the Indiana University Faculty Research Support Program. Chemical analysis was sponsored jointly by the METACyt Initiative of Indiana University, a major grant from the Lilly Endowment, Inc., and the Lilly Chemistry Alumni Chair funds (to M.V.N.).