Seasonally sympatric but allochronic: differential expression of hypothalamic genes in a songbird during gonadal development

Abstract

Allochrony, the mismatch of reproductive schedules, is one mechanism that can mediate sympatric speciation and diversification. In songbirds, the transition into breeding condition and gonadal growth is regulated by the hypothalamic–pituitary–gonadal (HPG) axis at multiple levels. We investigated whether the difference in reproductive timing between two seasonally sympatric subspecies of dark-eyed juncos (Junco hyemalis) was related to gene expression along the HPG axis. During the sympatric pre-breeding stage, we measured hypothalamic and testicular mRNA expression of candidate genes via qPCR in captive male juncos. For hypothalamic mRNA, we found our earlier breeding subspecies had increased expression of gonadotropin-releasing hormone (GnRH) and decreased expression of androgen receptor, oestrogen receptor alpha and mineralocorticoid receptor (MR). Subspecies did not differ in expression of hypothalamic gonadotropin-inhibitory hormone (GnIH) and glucocorticoid receptor (GR). While our earlier breeding subspecies had higher mRNA expression of testicular GR, subspecies did not differ in testicular luteinizing hormone receptor, follicle-stimulating hormone receptor or MR mRNA expression levels. Our findings indicate increased GnRH production and decreased hypothalamic sensitivity to sex steroid negative feedback as factors promoting differences in the timing of gonadal recrudescence between recently diverged populations. Differential gene expression along the HPG axis may facilitate species diversification under seasonal sympatry.

1. Introduction

In sympatric populations, variation in reproductive timing can result in reduced between-population breeding opportunities. If differences in timing are fixed and intermediate phenotypes perform poorly, then between-population variation in timing can lead to population divergence [1–7]. While there is strong evidence from across taxa that selection can act on differences in reproductive timing at the population level, relatively few studies have examined variation in the organismal processes (i.e. genetic, neuroendocrine and behavioural) that underlie shifts or divergences in seasonal timing. Understanding the mechanisms underlying changes in seasonal biology is important not only for understanding the processes that shape diversification and speciation, but also for predicting how populations will respond to environmental changes such as climatic shifts and habitat alterations.

In birds, divergence resulting in speciation between closely related migrant and resident populations is a leading factor in species diversification [8]. An observation among migrant and resident populations is that loss of migration often results in an advancement in the seasonal timing of reproduction [9–13]. The study of closely related and seasonally co-occurring migrant and resident populations that vary in seasonal timing of reproduction provides a unique opportunity to identify early diverging neuroendocrine and genetic mechanisms that may promote population differences in reproductive timing.

Timing of reproduction in birds is regulated by the activity of the hypothalamic–pituitary–gonadal (HPG) axis; the activity of this axis is regulated by both stimulation (e.g. day length) and suppression (e.g. negative feedback). After the winter solstice, as the duration of daylight increases, a photoperiodic threshold (photosensitivity) is reached and the hypothalamus increases production of gonadotropin-releasing hormone (GnRH), which is then released at substantial rates once a critical photoperiod threshold (photostimulation) is reached [14,15]. GnRH induces the pituitary to release luteinizing hormone (LH) and follicle-stimulating hormone (FSH), which enter the circulation where they bind to receptors on the gonads to induce gonadal growth, gonadal maturation and production of sex steroids such as testosterone and oestradiol. Once secreted, testosterone and oestradiol levels can, in turn, downregulate HPG axis activity by binding with sex steroid hormone receptors to induce negative feedback, thus suppressing further hypothalamic and pituitary secretion [16]. Additionally, both the hypothalamus and gonads can release gonadotropin-inhibiting hormone (GnIH) [17,18], which can inhibit HPG axis activity. While hypothalamic GnIH probably plays a major role at the termination of the breeding season [17], changes in GnIH secretion at the beginning of the breeding season may also be important for seasonal reproductive development. While many of the key steps described above have long been established, much remains to be known about the control of gonadal recrudescence and whether this event occurs mainly via increases of stimulatory processes, release of suppressive processes or a balance between the two.

To explore this reproductive transition, several recent studies have used the dark-eyed junco (Junco hyemalis) as a model to probe which levels of the HPG axis are responsible for seasonal gonadal recrudescence and reproductive development. These studies have used migrant and resident subspecies (J. h. hyemalis and J. h. carolinensis, respectively) that are found in sympatry from early autumn (when the migrants arrive from their breeding grounds in the north) to early spring (when the migrants depart just prior to resident reproduction). By examining these two subspecies in the early spring immediately prior to migrant departure, comparisons can be made between the HPG axes of birds exposed to similar environmental stimuli yet at different stages of reproductive development; migrants are at an early stage of development while residents are at a further advanced stage [19–22].

Prior work in juncos has focused on population differences at the level of the gonad and the pituitary in regulating timing of reproduction, although we note that photodetection (i.e. opsin sensitivity) could be the main driver of reproductive timing differences [23]. At the level of the gonad, Fudickar et al. [21] found that compared with migrant males in early spring, resident males had larger testes, higher baseline testosterone and a greater testosterone response to a GnRH challenge. This may suggest that gonadal recrudescence is partially regulated at the level of the testes via seasonal variation in sensitivity to gonadotropins. At the level of the pituitary, Greives et al. [22] found that early-breeding resident males had slightly higher baseline levels of LH than migrants in early spring, but did not differ in LH response to repeated GnRH challenges. This suggests populations differ in pituitary sensitivity to natural stimulation (i.e. lengthening photoperiod) but not in pituitary sensitivity to activation by repeated GnRH pulses. At the level of the hypothalamus, however, more knowledge is needed to assess the collective role of the HPG axis in balancing stimulation and suppression of the reproductive axis across the season. Under the hypothesis that early breeders have a more stimulated HPG axis, we would predict that early breeders have greater hypothalamic production of GnRH than late breeders (figure 1a). Similarly, we might also predict that early breeders display reduced production of gonadotropin-inhibitory hormone (GnIH), a neuropeptide responsive to environmental stimuli such as stress and food availability [24,25] (figure 1a). Under another hypothesis, early breeders may exhibit lower hypothalamic sensitivity to sex steroids than late breeders, thus ‘relaxing’ sex steroid-induced negative feedback of the HPG axis. These hypotheses are not mutually exclusive (figure 1a).

Figure 1.

Although residents and migrants experience the same cues on overwintering grounds, residents show accelerated reproductive development, as evidenced by larger testes and higher circulating concentrations of testosterone (T). At the hypothalamic level, (a) differences in timing of reproductive maturation could be influenced via three different, non-mutually exclusive processes: (i) top-down stimulation or inhibition via gonadotropin-releasing hormone (GnRH) and gonadotropin-inhibiting hormone (GnIH), (ii) relative sensitivity to sex steroids, and/or (iii) relative sensitivity to glucocorticoids (CORT). Predictions for each hypothesis are depicted as the relative expression of different hypothalamic genes including GnRH, GnIH, androgen receptor (AR), oestrogen receptor (ERα), mineralocorticoid receptor (MR) and glucocorticoid receptor (GR). Arrow thickness and number of receptors depict the relative amount of different hormones and hormone receptors, respectively. At the gonadal level, (b) differences in reproductive timing could be caused via two different, non-mutually exclusive processes: (i) relative sensitivity to gonadotropins and/or (ii) relative sensitivity to glucocorticoids. Predictions for each hypothesis are depicted as the relative expression of different testicular genes including luteinizing hormones receptor (LHR), follicle-stimulating hormone receptor (FSHR), MR and GR. (Online version in colour.)

Delayed gonadal recrudescence and reproductive development could also be caused by increased activation of the endocrine stress response, as glucocorticoids are known to have suppressive effects on the HPG axis [26–28]. However, Bauer et al. [19] found that during the pre-breeding stage, resident juncos had significantly higher levels of both baseline and stress-induced corticosterone compared with migrants, thus suggesting that circulating levels of corticosterone are not suppressing reproductive development in pre-breeding juncos. However, it remains to be seen whether glucocorticoids might delay full activation of the HPG axis via increased hypothalamic and testicular sensitivity to these stress-responsive hormones (e.g. glucocorticoid and mineralocorticoid receptor abundance).

To further examine the pre-breeding activation of the HPG axis, in the current study we examined male resident and migrant juncos under natural photoperiodic conditions during the time in which they are still in sympatry but diverging in reproductive physiology. Resident and migrant male juncos were euthanized immediately prior to the time of year when migrants leave the overwintering grounds, and both testes and hypothalami were harvested. We then used qPCR analysis to examine the relative mRNA expression of candidate genes associated with initiation or delay of gonadal recrudescence. At the level of the hypothalamus, we tested the hypotheses that gonadal recrudescence and reproductive development are controlled by (i) top-down stimulation and inhibition of the HPG axis via GnRH and GnIH, (ii) hypothalamic sensitivity to sex steroid negative feedback, and (iii) hypothalamic sensitivity to glucocorticoids (figure 1a). We predicted that compared with migrants, residents would have higher GnRH and lower GnIH, androgen receptor (AR), oestrogen receptor alpha (ERα), mineralocorticoid receptor (MR) and glucocorticoid receptor (GR) relative mRNA expression in the hypothalamus. At the level of the testes, we tested the hypotheses that gonadal growth and reproductive maturation are controlled by (i) sensitivity to gonadotropins and (ii) sensitivity to glucocorticoids (figure 1b). We predicted that compared with migrants, residents would have higher luteinizing hormone receptor (LHR) and follicle-stimulating hormone receptor (FSHR) and lower MR and GR relative mRNA expression in the testes.

2. Methods

(a). Capture and housing

On 4 December 2013, we captured eleven migratory and nine resident male dark-eyed juncos in Giles County Virginia at Mountain Lake Biological Station (37.37° N, 80.52° W). Migrants and residents were classified using bill coloration and wing chord [29]. After capture, birds were housed briefly (1–10 days) in an outdoor aviary at Mountain Lake Biological Station. On 14 December, birds were transported to Indiana University (Bloomington, IN, USA) where they were housed in an indoor aviary. Aviary temperature was maintained at 16 ± 2°C and lights were adjusted every 3 ± 1 days to track the natural changing photoperiod at the site of capture.

On 27 February 2014, birds were individually housed in 61 × 46 × 46 cm cages and separated into seven replicate rooms (2.5 × 2.1 × 2.4 m). Cages were arranged so that birds were visually isolated from each other. While visual isolation can cause social stress [30], we find it unlikely that it would have affected our experiment unless social stress differentially affects gene expression during early versus late gonadal recrudescence. Food and water were provided ad libitum, and water was supplemented with Nekton-S Multi-Vitamin (Arcata Pet, Arcata, CA). Birds were fed a mix containing white millet and sunflower chips (2 : 1 ratio), meal worms, orange slices, and a soft diet containing ground puppy chow, hard boiled eggs and carrots. Total weekly handling time was approximately 35 min.

To collect hormone data for a related experiment [21], birds were injected with a standardized dose of GnRH (1.25 µg chicken GnRH-I, American peptide 54-8-23) in 50 µl of phosphate buffer solution (PBS) every 6–7 days from 26 March 2014. While previous work found no evidence for physiological or behavioural effects of ‘GnRH challenges’ beyond an initial and transient stimulation (30 min) [31], it is unknown if repeated challenges affect long-term mRNA transcription rates. However, as all birds were treated the same and as our study compared mRNA expression levels between migrants and residents, possible effects of repeated GnRH injections on gene expression likely do not affect the interpretation of our results.

(b). Euthanasia and dissections

Juncos were euthanized on 31 March 2014 (five migrants and four residents) and 1 April 2014 (six migrants and five residents). Each holding room was entered only once on the days that birds were euthanized. Upon entering a holding room, three birds were immediately removed from their cages and taken to a nearby processing room. Euthanization via inhaled 100% isoflurane was completed within 2 min (mean = 1.3 min, s.d. = 0.3 min) of entering the animal holding rooms. Birds were verified as dead when they went limp, ceased breathing and a heartbeat could not be manually detected. Secondary confirmation was ensured via rapid decapitation. The brain and left testis were immediately dissected out, flash frozen on powdered dry ice and stored at –80°C. The mean time between euthanasia and freezing was 6.9 min (s.d. = 1.8 min).

Hypothalami were dissected using a cryostat using a protocol adapted from Ashley et al. [32]. Briefly, we made frontal slices until we saw the tractus septomesencephalicus (TrSM) split. We then used a 3 mm-diameter biopsy punch and punched (approx. 2.0 mm deep) directly above the optic chiasm along the rostral-caudal axis. After collecting the hypothalamus, we then continued to slice the brain to verify that we obtained the entire hypothalamus (i.e. we could see the cerebellum where our punch ended).

(c). Brain mRNA extraction and qPCR

Hypothalami were successfully dissected from nine resident and ten migrant juncos (one migrant brain was lost during transit between institutions). RNA was extracted via the RNAzol RT (Sigma Aldrich, St Louis, MO) method, after which spectrophotometry was used to determine RNA concentration and optical density (all samples had a 260/280 ratio between 1.8 and 2.0). We then treated 6 µg of RNA with DNase (Life Technologies, Carlsbad, CA), after which we again determined RNA concentration via spectrophotometry. Finally, we performed reverse transcription PCR using oligo(DT) primers, qScript reverse transcriptase (Quanta Biosciences, Beverly, MA) and 250 ng of DNase-treated RNA in a total reaction volume of 5 µl. A portion of samples were run without reverse transcriptase to verify that our samples were free of DNA contamination.

We then used cDNA product as a template for quantitative real-time PCR (qPCR) to measure the relative abundance of mRNA expression. We examined the following genes: GnRH, GnIH, ERα, AR, MR and GR, with PPIA and GAPDH as reference genes. Gene-specific primer sequences and the species on which they were based, amplicon sizes, accession numbers, standard curve r² values, and thermocycling conditions are listed in the electronic supplementary material.

qPCR reactions were run in triplicate on a Stratagene MX3000p (Agilent Technologies, Santa Clara, CA) thermocycler. Each reaction (total volume of 10 µl) contained 3 µl cDNA (diluted 1 : 40), 5 µl perfeCTa SYBR green supermix Low ROX (Stratagene) and primers at 0.3 µM. One plate was run for each gene, as all samples fit on one plate. Intra-assay variation ranged from 0.42% to 0.72%. A final melting phase was run at the end of each plate to verify amplification of single product. We ran a serial dilution (1 : 4) of a pooled cDNA sample for each gene; efficiencies ranged from 90.4% to 99.5%.

Relative quantification was calculated using the using the 2^ΔΔCt method [33], where ΔΔC_t = (C_t^{gene of interest} − C_t^{reference average}) calibrator − (C_t^{gene of interest} − C_t^{reference average}) focal. We picked one migrant as our ‘calibrator,’ therefore making their relative quantification value equal to 1 for each gene of interest. For reference genes, we averaged the C_t values of PPIA and GAPDH as recommended [34], and these values did not significantly differ between resident and migrant juncos (t₁₇ = 1.792, p = 0.091).

(d). Testes mRNA extraction and qPCR

RNA was extracted from one testis for each bird (nine residents and eleven migrants) using the Trizol method (Invitrogen, Carlsbad, CA), after which spectrophotometry was used to determine RNA concentration and optical density (all samples had a 260/280 ratio between 1.8 and 2.0). We then treated 1 µg of RNA with DNase (Promega, Madison, WI, USA). Finally, we performed reverse transcription PCR using oligo (DT) primers, Superscript reverse transcriptase (Invitrogen, Carlsbad, CA) and 1 µg of DNase-treated RNA in a total reaction volume of 20 µl.

We then used cDNA product as a template for quantitative real-time PCR (qPCR) to measure the relative abundance of mRNA expression. We examined the following genes: LHR, FSHR, MR and GR, with RPL4 as a housekeeping gene. RPL4 is one of the most stable housekeeping genes in testes of passerines [35], and its expression did not differ between migrant and resident juncos in our study (t₁₇ = 0.308, p = 0.762). We did attempt to measure testicular GnIH mRNA expression, but levels were below the limit of detection. Gene-specific primer sequences and the species on which they were based, amplicon sizes, accession numbers, standard curve r² values and thermocycling conditions are listed in the electronic supplementary material.

qPCR reactions were run in duplicate on a Stratagene MX3000P Real-Time PCR System (Agilent Technologies, Santa Clara, CA) thermocycler. Each reaction (total volume of 10 µl) contained 2 µl cDNA (diluted 1 : 40), 5 µl SYBR green Low ROX (Stratagene) and primers at concentration of 0.3 µM. All genes for an individual were run on the same plate. As we ran multiple individuals on a single plate, we made sure to include equal numbers of migrants and residents on each plate. Intra-assay variation ranged from 0.74% to 1.24%. A final melting phase was run at the end of each plate to verify amplification of single product. We ran a serial dilution (1 : 4) of a pooled cDNA sample for each gene; efficiencies ranged from 93.8% to 106.8%. One resident was excluded due to a missing value for RPL4 that stemmed from a poor replicate in qPCR amplification. One migrant was excluded from the LHR comparison due to a poor LHR replicate in qPCR amplification.

Relative quantification was calculated using the using the 2^ΔΔCt method [33], where ΔΔC_t = (C_t^{gene of interest} − C_t^{reference average}) calibrator − (C_t^{gene of interest} − C_t^{reference average}) focal. The calibrator sample was a pool derived from two migrants and two residents included in the current study.

(e). Statistical analyses

All statistical analyses were carried out in SPSS (v. 20). Testes GR and FSHR, and log-transformed hypothalamic ERα and MR mRNA expression levels, met normality and homogeneity of variance assumptions (tested with Shapiro–Wilk and Levene's tests, respectively). To determine whether testes GR and FSHR and hypothalamic MR and ERα mRNA expression levels differed between residents and migrants, we then carried out two-tailed independent-samples t-tests. Effect sizes were calculated using Cohen's d-test, where values of 0.2, 0.5 and 0.8 correspond to small, medium and large magnitudes of effect, respectively [36]. Data transformation failed to meet homogeneity of variance assumptions for testes MR and LHR, and hypothalamic GnRH, GnIH, AR and GR mRNA expression levels. For these genes, we used the non-parametric equivalent of an independent-samples t-test (a Mann–Whitney U-test) to determine whether migrants and residents differed in mRNA expression levels. The high number of non-parametric tests compared to parametric tests indicates that migrants and residents tended to differ in the variance of their gene expression. Effect sizes were calculated using the following equation, where and values of 0.1, 0.3 and 0.5 correspond to small, medium and large magnitudes of effect, respectively [36]. To control for multiple tests within each tissue, we used the Benjamini–Hochberg procedure with a failed discovery rate of 5% [37].

3. Results

(a). Hypothalamic mRNA expression

Residents had significantly higher expression of GnRH mRNA compared with migrants (figure 2a: U = 0, p = 0.001, adjusted α < 0.009, r = 0.866). Residents did not have significantly lower expression of GnIH mRNA compared with migrants, however (figure 2b: U = 34, p = 0.37, adjusted α < 0.05, r = 0.212). As predicted, residents had significantly lower relative expression of both AR (figure 2c: U = 6, p = 0.001, adjusted α < 0.017, r = 0.750) and ERα mRNA (figure 2d: t₁₇ = –3.560, p = 0.003, adjusted α < 0.025, Cohen's d = 1.119) compared with migrants. Residents also had significantly lower expression of MR mRNA (figure 2e: t₁₇ = –2.417, p = 0.027, adjusted α < 0.033, Cohen's d = 1.572) and a trend towards lower expression of GR mRNA (figure 2f: U = 22, p = 0.060, adjusted α < 0.042, r = 0.443, respectively) compared with migrants.

Figure 2.

Hypothalamic differences in relative mRNA expression of (a) gonadotropin-releasing hormone (GnRH), (b) gonadotropin-inhibitory hormone (GnIH), (c) androgen receptor (AR), (d) oestrogen receptor alpha (ERα), (e) mineralocorticoid receptor (MR) and (f) glucocorticoid receptor (GR) in male resident (n = 9) and migrant (n = 10) dark-eyed juncos (Junco hyemalis). Whiskers represent maximum and minimum values, boxes indicate the 25–75th percentile and the middle line represents the median value. (Online version in colour.)

(b). Testes mRNA expression

Residents did not have significantly higher expression of LHR (figure 3a: U = 37, p = 0.83, adjusted α < 0.05, r = 0.063) or FSHR mRNA (figure 3b: t₁₇ = 1.118, p = 0.28, adjusted α < 0.025, Cohen's d = 0.494) compared with migrants. Also contrary to our predictions, residents did not display significantly higher expression of MR (figure 3c: U = 32, p = 0.35, adjusted α < 0.038, r = 0.234) or GR mRNA (figure 3d: t₁₇ = 4.396, p = 0.001, adjusted α < 0.013, Cohen's d = 2.108) compared with migrants.

Figure 3.

Testicular differences in relative mRNA expression of (a) luteinizing hormone receptor (LHR), (b) follicle-stimulating hormone receptor (FSHR), (c) mineralocorticoid receptor (MR) and (d) glucocorticoid receptor (GR) in male dark-eyed juncos (Junco hyemalis). Resident and migrant sample sizes are 8 and 11 for all genes, respectively, except for LHR, where sample sizes are 8 and 10 for residents and migrants, respectively. Whiskers represent maximum and minimum values, boxes indicate the 25–75th percentile and the middle line represents the median value. (Online version in colour.)

4. Discussion

We examined whether variation in gene expression along the HPG axis is related to timing of reproductive development, as this may be a mechanism underlying the process of sympatric population divergence. By using two subspecies of male dark-eyed juncos that overwinter together, we were able to compare birds that were early (migrants) and advanced (residents) in their gonadal recrudescence during late winter/early spring when populations co-occur. In the hypothalamus, we found residents showed higher expression of GnRH, lower expression of AR, ERα, and MR, and a trend towards lower expression of GR mRNA compared with migrants. These findings support our hypotheses that seasonal differences in the timing of gonadal recrudescence and reproductive development are regulated by (1) top-down stimulation of the HPG axis via GnRH and (2) hypothalamic sensitivity to sex steroid negative feedback. While we found partial support that increased hypothalamic sensitivity to glucocorticoids may slow gonadal recrudescence in the migrants, we did not find adequate support that hypothalamic GnIH affected this process. In the testes, we did not find that residents showed higher expression of LHR and FSHR and lower expression of MR and GR compared to migrants, therefore these findings do not support our hypotheses that gonadal growth and reproductive maturation are controlled by testicular sensitivity to gonadotropins or glucocorticoids. In summary, subspecies differences in timing of gonadal recrudescence in males are probably influenced by hypothalamic level sensitivity to both stimulatory (e.g. day length) and inhibitory (e.g. sex steroid negative feedback) cues. While mRNA levels are good predictors of protein abundance globally [38], we recognize that in some cases mRNA expression may not show a strong correlation with protein levels [39]. However, we consider mRNA quantification still to be a useful measure as it represents changes in future protein levels.

(a). Hypothalamic gene expression

Our data support the hypothesis that GnRH stimulates gonadal recrudescence and reproductive development in male dark-eyed juncos, as we found that residents had significantly higher hypothalamic GnRH mRNA expression levels compared to migrants during the pre-breeding stage. These findings fit well with other avian studies that show an increase in GnRH mRNA expression levels during the transition from photosensitivity to photostimulation [40,41]. Additionally, increased expression of hypothalamic GnRH could be reflective of reduced sex steroid negative feedback, as we also found that residents had significantly lower levels of hypothalamic AR and ERα compared to migrants. Differential sensitivity to photoperiod cues could also explain subspecies differences in GnRH expression levels [42].

Male dark-eyed junco reproductive development did not seem to be regulated via GnIH during the pre-breeding time point we examined, as residents and migrants did not differ in hypothalamic GnIH mRNA expression levels. Avian studies have noted a decrease in hypothalamic GnIH mRNA expression levels prior to and during testicular recrudescence [40,43]. However, it is possible that we sampled our birds too late to detect a subspecies difference in hypothalamic GnIH mRNA expression, as residents may have dropped GnIH expression levels during the early pre-breeding stage but by the time we sampled our birds, migrants had already lowered GnIH expression as well. Alternatively, decreases in hypothalamic GnIH may play a minor or insignificant role in gonadal recrudescence, as the rapid spike in hypothalamic GnIH mRNA expression during the end of breeding suggests that GnIH's major role is to promote gonadal regression as birds become photorefractory [40,41,43].

Delayed gonadal recrudescence in migrants could also be mediated by enhanced sex steroid feedback at the level of the hypothalamus. Our results support this hypothesis, as we saw increased expression of hypothalamic AR and ERα mRNA in migrants compared to residents. Suppression of the HPG axis via enhanced sex steroid negative feedback has also been supported by studies that found increased production of GnRH after gonadectomizing male European starlings [44] and cichlid fish [45]. Similar to these studies, we found higher hypothalamic GnRH mRNA expression in birds with lower AR and ERα mRNA expression, which we interpret as loosened sex steroid negative feedback control. Wacker et al. [46], however, found higher levels of hypothalamic AR mRNA in male song sparrows (Melospiza melodia) during breeding compared to other life-history stages.

Alternatively, subspecies differences in AR and ERα mRNA expression may instead represent inherent subspecies differences, rather than a temporal difference in reproductive development, as Mishra et al. [47] found no differences in either AR or ERα mRNA expression among photosensitive, photostimulated and photorefractory red-headed buntings. Subspecies differences in sex steroid receptors may instead indicate differences in reproductive behaviours such as aggression, courtship behaviour, or parental care [46,48]. To further determine whether AR and ERα play a key role in seasonal timing, future experiments could examine whether variation in AR or ERα mRNA expression correlate with gonad size at different stages of gonadal recrudescence.

Our results partially support the hypothesis that reproductive development is regulated via hypothalamic sensitivity to glucocorticoids in male dark-eyed juncos, as we did find increased hypothalamic MR and a trend towards increased GR mRNA expression in migrants compared with residents. However, while the migrant's hypothalamus may be more sensitive to glucocorticoids during this life-history stage, we do know that during this time point residents have significantly higher baseline and stress-induced corticosterone plasma levels compared with migrants [19,21]. Therefore, it is difficult to determine whether increased corticosteroid receptor abundance in the hypothalamus actually inhibits the HPG axis. Future studies may aim to test whether steroidogenic enzymes differ between residents and migrants during this time point, as that may crucially determine the level of glucocorticoids able to bind to hypothalamic receptors [49].

However, if our finding of higher hypothalamic MR and GR levels in migrants versus residents represents a temporal change where corticosteroid receptor abundance continually decreases throughout seasonal reproductive development, then our findings do fit with a study examining actual protein levels of GR in whole songbird brains, as Lattin & Romero [50] found that GR levels decreased from pre-laying to breeding in wild house sparrows. By contrast, Krause et al. [51] did not see any change in GR mRNA expression levels in the preoptic area or paraventricular nucleus between the pre-parental and parental stages in white-crowned sparrows (Zonotrichia leucophyrs). Increased expression of MR and GR in migrants may instead be important for migratory preparation and development, as glucocorticoids play important roles in hyperphagia, fattening and migratory restlessness [52]. Alternatively, lack of a subspecies difference in MR and GR mRNA expression could be due to chronic stress, as birds may have never fully habituated to captivity. We doubt this is the case, however, as Lattin & Romero [53] did not find differences in whole-brain GR and MR protein levels between chronically stressed and unstressed house sparrows.

Finally, as corticosteroid receptors are found in high numbers in several different areas of the avian brain, we also recognize that MR and GR levels in different brain regions may play important roles in mediating activity of the HPG axis. Specifically, hippocampal MR and GR play major roles in negative feedback regulation of the mammalian hypothalamic–pituitary–adrenal (HPA) axis [54]. However, more evidence is needed to determine whether hippocampal MR and GR are important for regulation of the HPG axis in birds. In situ hybridization would allow examination of how expression levels of glucocorticoid and sex steroid receptor mRNA change across the whole brain during seasonal reproductive development. This technique would also be useful to examine changes within different nuclei of the hypothalamus, as we also note that the hypothalamus is a heterogeneous structure [55] and that our broad-scale comparison may have missed important changes in specific nuclei.

(b). Testicular gene expression

Given that resident males had larger testes, higher baseline testosterone and a greater testosterone response to a GnRH challenge than migrant males [21], we tested the prediction that gonadal recrudescence is regulated at the level of the testes via earlier sensitivity to gonadotropins. Gonadal LHR and FSHR bind to gonadotropins which leads to recrudescence and testosterone production, however, we did not find any differences in LHR and FSHR mRNA expression in our comparison of migrants and residents. Therefore, our results do not support the hypothesis that differences in the seasonal timing of gonadal growth and reproductive maturation are controlled by sensitivity to gonadotropins. Our findings are in line with a recent comparison of testis LHR and FSHR mRNA expression in seasonally sympatric migrant and resident dark-eyed juncos (J. h. thurberi) in California, USA, wherein Fudickar et al. [13] found that early breeding urban juncos had similar gonadal LHR and FSHR mRNA expression in early spring compared to late breeding migrant juncos that are found in sympatry outside of breeding. Increased hypothalamic GnRH mRNA expression in residents during early spring, as we report in the current study, could lead to earlier testis growth due to increased LH and FSH production without the need for increased LHR and FSHR. In previous comparisons of LH in free-living migrant and resident juncos in Virginia and California, residents tended to have higher circulating LH than migrants in early spring [13,22].

Previous work has identified an association between elevated corticosterone and reduced activity of the reproductive system in birds [27]. However, our results do not support the hypothesis that gonadal growth and reproductive maturation are delayed by increased gonadal sensitivity to glucocorticoids. In fact, male residents in our study had higher expression of testis GR mRNA compared with migrants. In combination with the recent finding of increased baseline and stress-induced corticosterone plasma levels in residents compared to migrants at this time point [19,21], our findings suggest that glucocorticoid signalling could play an important role in early season breeding in juncos. Rivier & Rivest [56] suggested that CORT may bind to GR and MR to temporarily delay reproduction during stressful events. Late-winter storms and other unpredictable perturbations of the environment can inhibit reproduction by elevated glucocorticoids in birds [57,58]. Dark-eyed juncos breed in highly seasonal environments with unpredictable early spring conditions, therefore increased gonadal sensitivity to glucocorticoids just prior to reproduction could suppress reproduction under unfavourable conditions when juncos are in an advanced reproductive state.

Other hypotheses, however, may explain how testicular gene regulation mediates seasonal reproductive development in songbirds. As songbird testes can produce and respond to GnIH [18,59], future studies should also assess changes in both GnIH and GnIHR mRNA expression levels in avian testes during the pre-breeding stage. If GnIH plays a major role in reproductive development at the level of the testes, then we would predict that migrants would show increased expression of GnIH and/or GnIHR compared to residents. Future studies could also examine potential differences in gonadal expression of steroidogenic genes such as p450scc or CYP17, as these are associated with junco subspecies differences in testosterone synthesis and testosterone-mediated traits during the breeding season [60]. If these genes play a major role in gonadal maturation, then we would predict that migrants would show decreased expression of p450scc and/or CYP17 compared with residents, as both genes transcribe enzymes important for production of steroid hormones.

5. Conclusion

This study supports the hypothesis that pre-breeding activation of the avian HPG axis is influenced by key regulatory genes expressed in the hypothalamus. We found evidence that increased GnRH production, decreased hypothalamic sensitivity to sex steroid negative feedback and decreased hypothalamic sensitivity to glucocorticoids may be factors promoting gonadal recrudescence. Future studies should examine whether variation in hypothalamic sex steroid negative feedback sensitivity (AR and ERα mRNA expression) are directly correlated with subsequent GnRH mRNA expression levels. We did not, however, find support that reproductive development is accelerated via decreased hypothalamic GnIH production. Our findings also did not demonstrate that gene expression at the level of the testes played a major role in observed variation in seasonal reproductive development, at least in terms of sensitivity to gonadotropins and glucocorticoids. These results demonstrate that variation in hypothalamic gene expression is related to differences in timing of breeding, and thus may play an important role in diverging annual schedules of seasonally sympatric populations. Here we have demonstrated that a differential response of the HPG axis to similar environmental cues by captive individuals originating from different populations is in accord with field observations of reproductive allochrony. Future work is needed to identify the genetic and epigenetic basis for population differences in reproductive timing, as these mechanisms may be ubiquitous factors in avian diversification. Furthermore, future studies should also examine whether similar changes in hypothalamic gene expression occur in female songbirds as well.

Ethics

All sampling procedures were approved by the Indiana University Institutional Animal Care and Use Committee (protocol no. 12-050) and conducted under scientific collecting permits issued by the Virginia Department of Game and Inland Fisheries (permit no. 47553) and the US Fish and Wildlife Service (permit no. MB093279).

Data accessibility

Data are available in the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.5d9s571 [61].

Authors' contributions

T.J.G., E.D.K., A.M.F. and C.M.B. conceptualized the study; A.M.F., M.A.-A. and J.W.A. carried out the live animal portion of the study; C.M.B. and S.A.-B. extracted the hypothalami and ran qPCR; A.M.F. and M.A.-A. oversaw extraction and qPCR of the testes; C.M.B. performed the statistics; C.M.B. and A.M.F. wrote the manuscript. C.M.B., A.M.F., T.J.G. and E.D.K. contributed to the revising of the manuscript.

Competing interests

We declare that we have no competing interests.

Funding

This work was supported by the National Science Foundation (IOS-1257474 and IOS-1257527). This project was also supported by the Environmental Resilience Institute, funded by Indiana University's Prepared for Environmental Change Grand Challenge initiative.