The development of sex differences in song in a tropical duetting wren

Abstract

The study of song development has focused on temperate zone birds in which typically only males sing. In the bay wren, Cantorchilus nigricapillus, both sexes sing, performing precisely timed, female-initiated duets in which birds alternate sex-specific song phrases. We investigated the development of these sex differences by collecting bay wren eggs and nestlings, and hand-raising them in individual acoustic isolation chambers. Each bird was tutored with either monophonic or stereophonic recordings of bay wren duets or heard no song. As adults, each tutored bird sang repertoires of complete duets, singing both male and female phrases. In addition, some birds sang only the male or female part of some duets to which they were exposed. Mono-tutored birds showed no sex-specificity in these solo songs, whereas stereo-tutored birds only sang solos consistent with their sex. In addition, stereo-tutored birds acquired songs over a longer period than did mono-tutored birds, and stereo-tutored females showed more sex-specificity than did males during early song production. Finally, we observed that tutored and acoustically isolated birds of both sexes invented male-like songs, whereas only males invent songs in the wild. These results reveal the relative roles of environmental versus innate influences in the development of sex-specific song in this species.

1. Introduction

Over the last 70 years, bird song has emerged as a complex behaviour whose function, underlying neuroendocrine mechanisms and development are well understood [1–3]. However, the bird song model is based on studies of species in which only male birds sing, most of which reside in the temperate zone [4–6]. Studies in the last 30 years have expanded this model to include females in two ways. First, female song is now acknowledged as ancestral [7] and has been described in many temperate zone species, although females usually sing with less intensity and complexity than their male counterparts [8–10]. Second, it has been recognized that in the tropics, where over 85% of avian species reside, females usually sing [11], with males and females often singing together in vocal duets [12,13]. While an increasing number of studies have focused on female song (reviewed in [14]) [3,15], few have attempted to study song learning in species in which both males and females regularly sing full, complex songs [2,10,14], despite their potential for expanding models for the origin of sex differences (and similarities) in animal communication [16,17]. Fewer still have focused on song learning in species in which males and females couple their songs in vocal duets [17].

Vocal duetting occurs when males and females non-coincidentally sing together in time. Styles of coordinating song between males and females can range from completely overlapping phrases to precisely timed alternations, ‘antiphonal song’ [18]. In addition, duetting species also vary in the degree to which syllables and phrases produced by males and females are sex-specific.

Song acquisition in duetting species is necessarily more complex than in species in which only males sing, which have been the focus of song learning studies. These now classic studies have demonstrated that males learn their songs from parents or neighbours before and/or after they disperse [3]. Young birds first silently memorize song phrases to which they are exposed and later practise them by performing ‘subsong’, improving what they sing by comparing it to their stored memory (template) of songs they have heard, until finally crystallizing a song into its final form (e.g. [19]). The degree of genetic constraints on what songs are accepted as templates and the length of the memorization and subsong phases of learning differ among species (e.g. [20]). In duetting species, there is an additional layer of complexity since young birds develop in an acoustic world where both males and females sing. Learning becomes even more complicated when there are sex differences in song behaviour, which is true of most duetting species [12,18]. In these cases, a young bird must first properly separate a duet into its component male and female parts, then acquire only the set of song phrases consistent with its sex, and, finally, it must learn to perform this song in coordination with that of an opposite-sexed partner. To date, there have only been two studies of song development in more controlled aviary environments in species in which both sexes regularly sing: slate-coloured boubous [21], Laniarius funebris, and blue-capped cordon bleus [22], Uraeginthus cyanosephalus, only the latter of which is a duetting species. In both studies, pairs of nestlings housed with parents or other adults appeared to learn most of their songs from adults of the same sex, although some birds also copied opposite-sex songs. To date, there has been no controlled study of sex-specific song development in individually housed birds.

Duetting has received a resurgence of attention in the last 20 years [12,17], particularly in the robustly singing group of neotropical wrens of the genus formerly known as Thryothorus, which molecular data have reorganized into four genera [23], so that the songs of over a dozen species have now been described (e.g. [13,15,24–35]). Comparisons across these and other wren species have detailed the occurrence of sex-specific song phrases, leader–follower roles in singing, rules by which male and female song phrases are combined, and duet precision [36].

Field studies of free-living neotropical wren species have provided initial insight into how duetting wrens might learn their sex-specific songs and how to coordinate them with those of a mate. In many of these species, young birds remain on their natal territories for several months [17,26,37], suggesting that, at least initially, song and linkage rules (‘codes’ sensu [38]) might be acquired from same-sex parents and/or neighbours. This argument is supported by observations in several wren species of young birds singing along with their parents (R.N.L. 1982–1986, personal observation) [24,33,37], or, in the case of cooperatively breeding striped-backed wrens, Campylorynchus nuchalis, same-sex relatives in their family groups [39]. When young birds practise singing by themselves, individuals tend to practise only the songs of one sex [26,37], although these observations might, in part, reflect the timing of when the songs of young birds are audible to an observer. In addition, differences between the sexes in the degree of song sharing within and between populations have been interpreted as reflecting sex differences in the timing and duration of song acquisition and/or sex differences in distance of dispersal [29,40,41].

The bay wren, Cantorchilus nigricapillus, offers a unique opportunity for understanding the development of sex differences in song because song behaviour, function and its underlying neuroendocrine mechanisms have been well studied in this species [13,26,27,42–45]. In the Republic of Panama, both male and female bay wrens regularly sing; most song occurs as precisely timed bouts of duetting in which males and females repeatedly alternate song phrases. Bay wren song behaviour was described extensively by Levin [13,26], who determined that duets are always initiated by females. Levin [13] reported that males and females have similarly sized repertoires of 15–20 sex-specific song phrases which are combined to form repertoires of 17–21 duets per pair through individually specific linkage rules guiding how each male responds to any given female song phrase. The most notable difference in the syntax of male and female song phrases is that males tend to repeat the last 1–2 notes within each phrase in a trill-like fashion, whereas females do not. Occasionally, birds sing their sex-specific phrases alone as solo songs. Females have each of their song phrases in common with at least one other female in the local population, whereas roughly 20% of the song phrases performed by each male are unique [26].

The origin of sex-specificity in bay wren song behaviour is of particular interest since previous work on this species has shown that adult males and females do not differ in the volume of the forebrain song control nuclei [42], in steroid sensitivity of these areas [43,44], or in plasma hormone levels [45]. Furthermore, years of observations of young bay wrens on their natal territories revealed that only one young bird ever produced the songs of the opposite sex during subsong [26]; the sex of this bird is unknown as it evaded capture (R.N.L. 1982–1986, personal observation).

It is unclear from these observations which, if any, aspects of sex-specificity in song are under direct genetic control in bay wrens. In temperate zone species, often there are innate aspects of song acquisition, a preactive template [46] that includes a set of general syntactical guidelines that allow a young bird to accept conspecific songs as models from which to learn and to reject others (e.g. [20]). Thus, in bay wrens, it is possible that the decision to accept male or female songs as appropriate models is under genetic control. To examine the relative roles of environmental versus innate influences on song acquisition, we examined song development in hand-raised bay wrens while controlling exposure to acoustic and visual stimuli, thus providing the first direct investigation into the development of sex-specific song in a duetting species.

2. Methods

We performed three experiments in this study in three consecutive years, 1993–1995. The methods employed for each experiment were identical with the exception of the method and degree of song exposure. Birds collected as eggs and young nestlings were used in each experiment. We designed Experiment 1 to determine whether bay wrens could be successfully hand-raised in captivity and whether they would learn to sing from tutor tapes. Thus, in Experiment 1, we exposed eight birds to monaural tutor tapes broadcast through one speaker in their cage (mono-tutoring), given that our observations showed in the wild, young birds often hear their parents singing from a distance so that the male and female phrases in duets appear to be coming from a single point source. It is likely, however, that nestlings also hear their parents singing nearby from separate locations. To simulate this possibility, in Experiment 2, we presented the same tutor songs used in the first experiment to 10 new birds in stereo, with the male and female contributions to each duet broadcast from separate speakers mounted opposite to each other in each cage (stereo-tutoring). In Experiment 3, we raised eight birds in acoustic isolation, without any tape tutoring, to see what aspects of song acquisition did not require song exposure. All experiments were approved by the Pomona College Institutional Animal Care and Use Committee.

(a) . Collection and hand-rearing

We collected nests in the lowland rainforest along roads and streams in the province of Colon in the Republic of Panama under permits issued by the Instituto Nacional de Recursos Naturales Renovables, during the two peaks of the breeding season (May–August and October–December), 1993–1995. To avoid the high rate of nest predation in this species [26], eggs or nestlings were collected upon discovery. Eggs were incubated at 37.2°C until hatching. All birds were transported under quarantine with permits from the United States Department of Agriculture via commercial airlines to a USDA certified animal quarantine facility at Pomona College. Nestlings were hand-fed a diet modified from Lanyon [47] (J. Marler & D. Kroodsma 1993, personal communication) and hydrated with Pedialyte every 20–40 min throughout the day until they drank water on their own (9–11 weeks of age). The same food was presented to adults, ad libitum, and changed at 2–2.5 h intervals throughout the day to prevent spoilage. Water was changed daily.

At approximately 15 days post-hatching, each bird was housed in a 47.0 × 25.4 × 27.9 cm wire cage placed within a 57.8 × 42.5 × 35.6 cm (interior dimensions) cold-rolled steel, sound-insulated Acoustic Systems sound isolation chamber. Chambers were oriented so that birds were visually isolated from each other. Each chamber was fitted with a Realistic PZM microphone and full-range Realistic 10.2 cm flush mount speakers (in Experiments 1 and 2) through which vocalizations could be recorded or broadcast, respectively, from a remote location. White noise was broadcast into the facility (at 72 db SPL measured in the centre of the room) to further reduce song transmission between chambers. Bird rooms were maintained at 25.5–27.8°C on photoperiods matching those of their collection site.

(b) . Song tutoring

Beginning at 15 days post-hatching, birds in chambers were exposed twice daily to a taped, 1 h sequence of repetitions of five different bay wren duets that had been originally recorded from wild birds in their natural habitat (described below). Tutoring sessions were conducted within the first and last 2 h of daylight to be coincident with times that most song occurs in populations of free-living bay wrens [26]. The five duet types used on tutor tapes were changed 6, 12, 28 and 44 weeks following the onset of tutoring, so that, over time, birds were exposed to five different sets of five duet types. The order of delivery of different duet types within a set was changed weekly to prevent habituation.

Duets used for tutoring were recorded from the duet repertoire of one pair of adults living within the same song dialect region from which eggs and nestlings were collected. Recorded duets were edited so that a female–male pair of song phrases was repeated four or five times in rapid succession, comprising a duet bout. Bouts of duets occurred at 90 s intervals throughout each tutor tape. Each of the five duet types on a tutor tape was presented eight times in a sequence resembling natural bay wren song behaviour. Tutor tapes were broadcast from a Realistic SCP-31 tape player through a Realistic MPA-45 35-watt PA amplifier and into Realistic flush mount speakers (catalog no. 112-1704) in each bird's chamber. Playback intensity was standardized at 72 dB SPL measured at the centre of each acoustic chamber.

In Experiment 1, tapes were broadcast from a single speaker placed in each chamber. The same tapes were used in Experiment 2, but male and female phrases of each duet were presented from speakers on opposite sides of each chamber. The location from which male and female phrases were broadcast was exchanged halfway through each tutor session. Stereo tapes were prepared from the original mono recordings used in the first experiment by using Digidesign ProTools sound editing software.

(c) . Recording and song analysis

Vocalizations produced by each bird were recorded through their chamber by a Realistic PZM microphone (catalogue no. 33-1090B) connected to a Yamaha MLA-7 8 channel mic to line amplifier to a Nakamichi DR-3 tape recorder. Birds were recorded weekly for the first hour of the day two weeks after tutoring started (i.e. at four weeks of age) until they were at least five months of age; tutoring and recording stopped after the age at which free-living birds first breed and sing full adult song. Sex of each bird was then determined using surgical laparotomy or by taking a drop of whole blood from a brachial vein and drying it on paper for DNA analysis (Zoogen, Davis, CA).

We displayed and printed recorded songs using a Kay Model 5500-1 DSP Sonagraph and Kay P6OU video copy processor, using fixed settings and aspect ratios to permit direct comparison. Sonograms of adult birds were then classified by at least two independent observers as to whether they matched the male, female or full duet phrase of any of the duets used on the tutor tapes. Interobserver reliability was 100%. At least the last two recording sessions were used to assess the adult repertoires of each bird; a repertoire was considered to be completely sampled when a plot of the number of song types detected versus the number of duet bouts recorded showed that repertoire size had reached an asymptote [48].

We also evaluated the early song performance of each bird in Experiment 2. We traced back to when each bird first produced a vocalization that was recognizable as a song in its eventual adult repertoire. These first performances were then rated using a 1–5 rating scale as follows: 1—song would not have been recognizable as a bay wren song phrase without knowing it was produced by a subsinging bay wren; 2—song is barely recognizable (several notes produced at the wrong frequency, not quite the correct frequency versus time ‘shape’, missing, or have extra frequency sweeps at their end); 3—missing at least one note and/or at least one note is misshapen; 4—one note missing or misshapen; and 5—performance matches the quality of each bird's adult performance of that song. The performance of male and female phrases in each duet was scored separately and the scores for all of the male and female song phrases in an individual bird's repertoire were then averaged by sex. Thus, at the end of this analysis, each bird had two scores, one each for the male and female phrases from their entire repertoire. Two observers scored each bird's song performance; interobserver reliability was at least 95% for each bird.

(d) . Statistical analyses

We used Wilcoxon signed-rank tests to compare different aspects of song behaviour within a group of birds and Mann–Whitney U-tests were used to compare song behaviour between different groups of birds. We used a chi-squared test to compare the distribution of learned tutor songs between mono-tutored and stereo-tutored birds. In all cases, statistical significance was determined at a p-value less than or equal to 0.05.

3. Results

Both mono- and stereo-tutored birds demonstrated the ability to learn songs from tutor tapes. Almost all birds acquired songs to which they were exposed during weeks 1–6 and 7–12 of tutoring (from approximately 2 to 14 weeks post-hatching). However, stereo-tutored birds acquired songs for longer than did mono birds; only two of the eight mono-tutored birds acquired songs 13–28 weeks after tutoring began, whereas only two stereo-tutored birds failed to do so (χ² = 5.45, d.f. = 1, p = 0.02).

(a) . Experiment 1: mono-tutored birds

At adulthood, the majority of the songs produced by each of the eight birds in this experiment were complete duets; that is, each bird sang both the male and female phrases of the duets they had acquired from tutor tapes. (figure 1). As is true in many species in which song learning has been studied [49], each bird crystallized a slightly different individual variation of any given song (figure 1b,c). Individual repertoire size ranged from 3–11 different duets (figure 2). Two females each had one song which they only performed as a solo song, showing no evidence that they ever performed both parts of this duet. In one of these cases (female F), the female produced only the female phrase from the duet, and in the other case (female G), the female produced only the male phrase.

Figure 1.

Sonogram of a duet from a tutor tape and imitations produced by hand-raised birds. (a) Frequency versus time representation (sonogram) of one duet that was broadcast to birds in this study. Each duet consists of repeated presentations of a female–male phrase sequence, as indicated by the brackets. (b,c) sonograms of the duet in A as performed by birds B and D, respectively, as adults. Both the male and female parts of the duet were produced by each bird.

Figure 2.

Duet repertoire size of monoaurally tutored birds. The height of each bar indicates the total number of different duet types performed by each of eight birds. Open areas of each bar indicate the number of duet types which were always performed as complete duets. All but one bird also performed a subset of the duets in their repertoire as male (heavily shaded) or female (hatched) solo songs. Birds did not show a significant tendency to perform solo songs that were consistent with their genetic sex. Birds F and G each had one song that was never performed as a duet only as a male (G) or female (F) solo song.

In addition to singing complete duets, six of the eight mono-tutored birds occasionally sang only one half of a duet in their repertoire, performing only the male or female phrase as a solo song. An example of this is illustrated in figure 3, which shows a singing bout performed by a female that began as a duet and ended as a female solo song when she stopped singing the male phrase. This was not a frequent occurrence; out of a total of 639 bouts of singing, only 76 contained at least one repetition of a solo song phrase, and all birds sang significantly more bouts of duets than bouts that included solo songs (Wilcoxon signed-ranks test, W = 36.0, p = 0.014). Mono-tutored birds did not show a significant preference for singing solo songs that were consistent with their genetic sex (Wilcoxon signed-ranks test, W = 9.5, p = 0.832). Three birds also sang chimeric solo songs, performing song phrases which included parts of both male and female contributions to a duet.

Figure 3.

Sonogram of a bout in which bird G began singing a complete duet and ended the bout with a ‘female solo song’ in which only the female portion of the duet was performed. Birds can also sing an entire bout consisting only of solo song.

(b) . Experiment 2: stereo-tutored birds

As adults, the 10 stereo-tutored birds in this experiment most commonly sang complete renditions of the full duets to which they had been exposed, acquiring repertoires of 4–11 duets (figure 4). Like the mono-tutored birds in Experiment 1, most of the birds in this experiment also sang portions of their duets as solo songs but did so more often than did the birds in Experiment 1 (stereo-tutored birds: 32.08% ± 17.94% of all bouts, mean ± s.d., mono-tutored birds: 13.36% ± 12.62%, Mann–Whitney U = 121.0, p = 0.02). In contrast with the mono-tutored birds, however, stereo-tutored birds showed a significant tendency for singing the song phrases appropriate to their sex (Wilcoxon signed-ranks test, W = 0, p = 0.014). Stereo-tutoring did not eliminate mistakes, however; half of the birds that produced sex-typical solo songs also sang solo songs appropriate to the opposite sex, and three birds sang chimeric solo songs in which they incorrectly separated the male and female parts of a duet.

Figure 4.

Duet repertoire size of stereo-tutored birds. Open areas of each bar indicate the number of duet types which were always performed as complete duets. Birds could also perform a subset of the duets in their repertoire as male (heavily shaded) or female (hatched) solo songs. There was a significant tendency for birds to perform solo songs that were consistent with their genetic sex.

Like some of the birds in Experiment 1, stereo-tutored birds had some songs which they sang only as solo songs; in these songs, they apparently learned only the male or female phrase in a duet. This was true for all but one of the stereo-tutored birds, in contrast with the only two of eight birds that exhibited this behaviour in the first experiment. For the stereo-tutored birds, these songs were completely consistent with each bird's genetic sex (figure 5). However, this was evident in only a small proportion of each bird's vocal behaviour since these birds still sang full duets more often than they sang any type of solo song (Wilcoxon signed-ranks test, W = 51.5, p = 0.02).

Figure 5.

Repertoire size of songs performed only as solo songs by stereo-tutored birds. The performance of male (heavily shaded) or female (hatched) solo songs was entirely consistent each bird's genetic sex.

We explored sex-specific song behaviour further by examining vocal behaviour of stereo-tutored birds when song production first began and scoring performance of male versus female song phrases as they began to become recognizable. This analysis revealed that at the stage that any part of a song was first recognizable, the averaged performance scores of all young birds were better for sex-appropriate song phrases than the song phrases of the opposite sex (Wilcoxon signed-ranks W = 0.0, p = 0.006). Males and females performed the song phrases of their own sex equally well (Mann–Whitney U = 7.0, p = 0.25). However, at this stage, young males were more adept at singing female song phrases than young females were at singing male song phrases (Mann–Whitney U = 38.0, p = 0.036).

In both Experiments 1 and 2, sample size prevented us from determining whether males and females differed in the degree to which they sang solo songs appropriate to their own sex. However, when pooled together, the results of the two experiments indicated that females were more likely than males to sing sex-appropriate solo songs (Mann–Whitney U = 72.0, p = 0.04).

In both experiments, one or more birds of each sex also sang invented, unique songs that were not on the tutor tapes; in all, four males and five females created a total of 13 unique song types. Nine of these song types were duets in which the female song phrase was copied from a tutor tape and the male song phrase in the duet was completely invented. Only one of the remaining four song types resembled a duet; this song was performed by a female and contained both male- and female-like invented song phrases. The remaining three invented song types were unlike typical bay wren song phrases [24].

(c) . Experiment 3: acoustically isolated birds

Of the five females and three males raised in acoustic isolation, none sang male, female or duet phrases of song types that were on tutor tapes. However, isolate birds did produce repertoires of song-like phrases that were within the frequency and timing parameters of typical bay wren songs (figure 6) and seven of the eight isolate birds at least occasionally sang in the species-typical fashion of repeating a song phrase several times within a bout of singing. The structure of song phrases produced by isolate birds was variable and unique to each bird. Songs produced by males and females did not appear to differ in ways that might indicate sex-specific constraints on song construction. For example, both male and female isolate birds sang phrases that contained repeated notes within a phrase- a trait that is typical only of male songs in wild bay wrens [24], although the repeated notes did not always occur at the end of a song phrase as they do in free-living males.

Figure 6.

Sonograms of the songs of three males and three females that were raised without exposure to bay wren songs. Each of these birds created songs that included male-like syntax.

4. Discussion

The experiments reported here represent the first attempt to hand-raise and song-tutor duetting birds, and to provide initial insights into innate versus environmental influences on the ontogeny of sex-specific song behaviour. The fact that full duets were the primary vocalization of each adult bird in both mono- and stereo-tutored bay wrens suggests that there are no genetic, physiological or anatomical constraints that restrict birds to producing the songs of only one sex; as adults, both male and female tutored birds initiated singing and did so producing male and female song phrases equally well. Furthermore, our observations of occasional solo singing by mono-tutored birds suggest that bay wrens appear to be predisposed to splitting a song to which they are exposed into its component parts and performing only half of it, and most birds were able to find the break point between male and female song phrases within a duet. However, it was only when birds were provided with the opportunity to hear male and female song phrases from separate speakers that young bay wrens showed the ability to pick the part that corresponded with their biological sex. Finally, the fact that even birds with no song exposure produced invented song phrases at a species-typical repetition rate and general syntax suggests that this aspect of song development is also innate, as has been suggested for other species originally placed in the Thryothorus genus [36].

It is possible that wild bay wrens learn to sing by first memorizing entire duets. As they subsequently begin to travel with their parents, they would have the opportunity both to hear the two parts of the duet coming from different locations and to engage in social interactions with their parents and siblings, as suggested by Rivera-Cáceres et al. [37]. As a result, birds might then correctly identify the break point between male and female contributions to each duet and socially learn which part of each memorized duet to crystallize into their final adult repertoire. This scenario is reminiscent of the overproduction and selective attrition of song types first described by Marler & Peters [50] and later found in many temperate zone species (reviewed in [51]). The suggestion that retention of some songs and the deletions of others can be the result of a young bird paying attention to or receiving attention from another individual has been made for other species [1,51–55]. Regardless, bay wrens that were tutored in stereo demonstrated the degree to which birds can develop sex-specific song without social interactions with other birds.

It may be that bay wrens enter the song-learning process with a preference for the songs of their own genetic sex that is only apparent when male and female songs are presented from two different locations, as it is for stereo-tutored and free-living birds. Under these circumstances, bay wrens might be guided by a predisposition to learn either the first (female) or second half (male) of each song phrase, or by an innate preference for song phrases with male versus female syntax. Regardless, it is clear that some likely social element was missing in the environment of tutored birds that plays a critical role in the development of sex-specific song since most of the vocal behaviour of even stereo-tutored birds was the singing of full duets.

Why should bay wrens at least initially learn the songs of the other sex? In temperate zone birds, it has been suggested that songs that males acquire and rehearse, but don't crystallize (i.e. ‘overproduced’ songs) may be beneficial, rather than simply being a learning artefact. ‘Overproduced’ songs might play a role in song sharing, territorial defense, mate attraction and/or neighbour assessment in males (reviewed in [51,52]) and facilitate mate assessment and recognition by females (e.g. canaries [56,57]). In the case of bay wrens, we found sex differences in the sex-specificity of song learning and performance that might reflect a differential benefit to learning the songs of the opposite sex. In early subsong, female bay wrens did not sing opposite-sex songs as well as did males and as adults, females were more likely than males to sing same-sex solo songs. This is consistent with the fact that bay wren male and female songs serve different functions. Levin [13,27] concluded that female bay wrens sing to defend territories against other females and that males vocally respond to female song as a form of acoustic mate-guarding, resulting in a duet. Thus, wild female bay wrens may only need to recognize and respond to female song, while male bay wrens must coordinate and link each of their songs with those of their mate, a task which might require young males to attend to and/or learn both male and female songs. The idea that difference in song function might explain sex differences in song learning has been suggested in recent reviews of song learning [2,10]. This interpretation is supported by results in Pheugopedius euphrys, another tropical duetting wren species in which females sing the lead phrase in duet songs [35,36]. In this species, it was found that for both male and female wrens, female song syllables elicited a stronger neural response in HVC, a major forebrain song nucleus, than did male song syllables [58], possibly reflecting the greater need by males to respond to opposite-sex song.

Our experiments also provide insight into the development of sex differences in plasticity and creative ‘invention’ in bay wren song development. In free-living bay wrens, only males sing unique songs [13]. By contrast, in both mono- and stereo-tutored bay wrens, adults of both sexes sang invented duets, but, with only one exception, they only invented the male part of the duets that they sang, coupling it with a female phrase from a tutored song. Similarly, the songs produced by birds raised in acoustic isolation also contained repeated syllables typical of male song phrase syntax This difference between laboratory-raised and wild bay wrens suggests that the ability to create male songs is common to both sexes, but is only expressed by males in the wild. The process by which invention is suppressed in wild females is unclear, but it is reminiscent of other sex-specific behaviours that have been suggested to be the result of sex-specific cell death [59–61].

Regardless of whether the initial learning and production of songs of the opposite sex is an artefact or an adaptation, the observation that stereo- but not mono-tutored birds produced only the sex-appropriate song phrase when singing solo songs is intriguing. The change from monoaural to stereophonic tutoring also resulted in a longer song learning period. The latter result is reminiscent of the extension of song learning that results from social interactions in other species (reviewed in [54,62]). These results suggest that wild bay wrens acquire completely sex-specific song either by observing vocal interactions of their parents or by interacting with or by simply copying their same-sex parent. Interactions with parents might also suppress the expression of unique song types by females.

It can be difficult to clearly identify the role played by social influences in the development and maintenance of sex differences in a complex vertebrate behaviour where both sexes exhibit the behaviour, but do so differently. In the bay wren, previous work has not found an obvious neuroendocrine basis for sex differences in adult song [42–45]. Examination of song development in this species has provided a unique glimpse into the nature of the interplay of genetic and social influences on the expression of sex differences in a behaviour expressed by both sexes.

Ethics

This research was approved by the Institutional Animal Care and Use Committee of Pomona College. Collection, transportation and housing of bay wrens were conducted under permits issued by the Instituto Nacional de Recursos Naturales Renovables of the Republic of Panama with the assistance of the Smithsonian Tropical Research Institute, and the United States Department of Agriculture.

Data accessibility

Data from tutoring experiments described in this paper are accessible via the Dryad Digital Repository: https://doi.org/10.5061/dryad.qz612jmj0 [63].

Authors' contributions

R.N.L.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—original draft and writing—review and editing; T.I.P.: data curation, investigation, methodology, resources, supervision and writing–review and editing; J.K.B.-M.: data curation, formal analysis, methodology, validation and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This research was funded by an NSF Young Investigator Award, an NSF-RUI grant and a grant from the Irvine Foundation to R.N.L.