Abstract

Sexual displays are typically given in the absence of predators. One possible exception to this is seen in male splendid fairy-wrens (Malurus splendens), which sing display-like vocalizations (Type II song) in response to predator calls. In this study, we explored the function of this vocal “hitchhiking” using playback experiments. We documented that: 1) males responded to Type II songs of other males as if they were displays intended for conspecifics, 2) females became more attentive after hearing a predator vocalization than a control (nonpredator) vocalization, and 3) females tended to respond more strongly to Type II songs when the songs were preceded by a predator vocalization than when they were given solo. Males of all age and status classes produced Type II songs and production was not dependent on any of 3 morphological measures of male quality, so we do not find evidence to support the hypothesis that Type II singing behavior is condition dependent. Because predator calls increased female attentiveness and tended to increase subsequent response to Type II songs, it appears that males may be using predator vocalizations as alerting signals for their own conspecific-directed displays.

Understanding the evolution of elaborate sexual displays requires study of the signals themselves and the context in which the signals are given. One context that is counterintuitive is displaying in the presence of predators, and with few exceptions (e.g., Evans et al. 2002; Ellis 2009; Kim et al. 2009) most research to date has demonstrated that both signalers and receivers should reduce sexual behavior when predators are near (Magurran and Seghers 1990; Magnhagen 1991; Zuk and Kolluru 1998; Godin and McDonough 2003; Bernal et al. 2007; Møller et al. 2008). Displaying to conspecifics in the presence of a predator, however, could be an adaptive strategy under honest signaling theory (Zahavi 1975) if high-quality individuals perform such potentially costly displays more than low-quality individuals (Møller et al. 2006). Alternatively, displaying in the presence of predators could be advantageous based on theories of efficient signal detection (Johnstone 1998; Rowe and Skelhorn 2004; Wiley 2006) because receivers may be especially attentive when danger is near (Elgar 1989). Thus, the presence of a predator may stimulate male sexual displays if alerted or “scared” females are more receptive to conspecific signals.

Although displaying to conspecifics in the presence of predators is considered to be the exception rather than the rule, 3 species of fairy-wrens (Aves: Maluridae) exhibit a vocal behavior that involves “hitchhiking” their song to the end of heterospecific vocalizations, specifically those of avian predators (Langmore and Mulder 1992; Zelano et al. 2001; Greig et al. 2010). Our study focused on the splendid fairy-wren (Malurus splendens) in which males give trills, “Type II songs,” in response to the vocalizations of gray butcherbirds (Cracticus torquatus) (Zelano et al. 2001; Greig and Pruett-Jones 2008). Butcherbirds are potential predators of both adult fairy-wrens and their nests (Lea and Gray 1935–1936; Tarburton 1991; Gardner 1998). Nevertheless, butcherbird vocalizations cause male fairy-wrens to sing so promptly and reliably that the joint vocalizations resemble a duet (Greig and Pruett-Jones 2009). Working with superb fairy-wrens, which also give Type II songs in response to avian predator calls, Langmore and Mulder (1992) hypothesized that Type II songs are sexual displays directed to females and that predator calls function as alerting signals that increase signal efficacy. They also suggested that giving a display in response to an unpredictable external stimulus adds a dimension of difficulty that enforces the honesty of the signal.

Two alternative reasons to sing in response to a predator are to give an alarm directed to conspecifics or as a pursuit deterrent directed to the predator. We can reject these possibilities in splendid fairy-wrens, however, based on previous results (Greig and Pruett-Jones 2009) that strongly suggest that neither is the case. In particular, males do not give Type II songs in the presence of silent butcherbirds but only in response to butcherbird vocalizations. Additionally, in response to butcherbird playbacks, males are no more likely to give Type II songs in the presence of a butcherbird mount than in the presence of a nonpredator mount (Greig and Pruett-Jones 2009). This leaves the possibility that Type II songs are a form of sexual or territorial display to conspecifics.

Here, we experimentally tested the hypotheses that Type II songs are displays directed to conspecifics and that predator calls function as alerting signals for intraspecific communication. We predicted: 1) If Type II songs are displays to conspecifics, then males should respond to Type II songs from an intruding male on their territory in an aggressive or territorial manner, 2) if the purpose of giving Type II songs after a predator is to enhance signal detection, then females should become more alert and attentive after hearing the calls of that predator, 3) if predator calls effectively enhance Type II song detection, then females should be more responsive to Type II songs when they are preceded by a predator vocalization. We investigate the possibility that Type II singing is a condition-dependent signal of male quality, and we document additional conspecific-directed display behaviors given by male fairy-wrens in response to predator vocalizations.

MATERIALS AND METHODS

Study species

We worked on a color-banded study group of splendid fairy-wrens that has been monitored from 1992–1999 and 2004–2009 (Pruett-Jones et al. 2010). Splendid fairy-wrens are a cooperatively breeding, socially monogamous, and reproductively promiscuous passerine widely distributed throughout southern and southwestern Australia (Rowley and Russell 1997; Webster et al. 2004). Both male and female splendid fairy-wrens produce a song that is used in territorial encounters and during the dawn chorus, referred to as “Type I song” (Rowley and Russell 1997). Both primary and helper males, but rarely females, also give Type II songs (Greig and Pruett-Jones 2008). In addition to the predator context, Type II songs are also given during the dawn chorus and occasionally during the day in the absence of predators (Greig and Pruett-Jones 2008). Both sexes also give “seet” alarm calls, short trilled contact calls, and brief “chatter” calls that function either as contact calls or mobbing calls depending on the speed with which concurrent chatters are delivered (Greig and Pruett-Jones 2008).

Playback experiments

Our playback experiments were conducted primarily during the October–December breeding season of 2008 at the Brookfield Conservation Park (BCP) in South Australia. We included playback experiments conducted in 2006 and 2007 for an analysis of the condition dependence of male Type II singing behavior. We recorded all vocalizations used in the playback experiments at BCP using a Marantz PMD 660 solid-state digital recorder (D&M Professional, Itasca, IL), 44.1 kHz sampling rate, and a Sennheiser ME66 shotgun microphone capsule and K6 power module (Sennheiser Electronic Corporation, Old Lyme, CT). We checked the quality of all recordings and filtered noise below 1.0 kHz using Raven 1.2 (Cornell Laboratory of Ornithology, Ithaca, NY). Playbacks were uncompressed Audio Interchange File Format audio files. In the field, we used a Saul Mineroff Electronics amplified field speaker (Saul Mineroff Electronics Inc., Elmont, NY) and an iPod mini (Apple Inc., Cupertino, CA). We standardized the relative amplitude of playbacks in Raven 1.2 and then tested actual amplitude with the field playback equipment using a Radio Shack sound–pressure level (SPL) meter (model number 33–2050, Radio Shack Corporation, Fort Worth, TX), set at C-weighting, fast response. Amplitude was approximately 87.0 dB SPL (re 20 μPa) at 1.0 m for Type I songs and Type II songs, 83.0 dB SPL for nondisplay controls, and 90.0 dB SPL for butcherbird and nonpredator vocalizations.

We tested focal birds with vocalizations recorded from nonneighboring individuals (potential territory intruders), and we used a different sample individual for every focal bird and test vocalization within each experimental treatment (playback type). We tested only birds that did not have young, so all birds were at a similar nesting stage and generally responsive to playbacks. We conducted playbacks in the morning (0600–1100 h) while focal birds were on their territory with their social mate and clearly visible. For each playback, we positioned the speaker ∼20 m from the focal bird at a height ∼1 m above ground.

We randomized the order of experimental treatments for those individuals that received multiple treatments over the course of a season. Focal birds received a maximum of one playback in a day, and when multiple playbacks were conducted on an individual they were done within a period of no longer than 2 weeks. We observed focal birds for a minimum of 5 min before beginning the playback to be sure that there were no disturbances, such as an intrusion from an extragroup fairy-wren or the appearance of a predator. In the event of a disturbance, we discontinued the trial and attempted it at another time. We recorded the entire playback session using a Marantz PMD 660 digital recorder so that we could accurately record both fairy-wren vocal responses and dictated behavioral responses. Blind scoring of behavioral responses was not feasible because we could not capture playbacks on video for subsequent analysis, so to compensate for this, we chose behavioral responses that were unmistakable when the bird was in view (described below) and that happened immediately after or during the playback. Because the behavioral responses we chose were typically immediate and brief, we were not able to dictate information on latency and duration that was accurate enough for analysis, so we categorized these responses simply as present/absent. Although this results in a more simplified data set, the data are less susceptible to interpreter bias and error. If a focal bird moved out of view and prevented unambiguous scoring of a response, then we excluded the playback from our analysis of that response. We still used that playback for analysis of other responses that we were able to score unambiguously. Because vocal responses were captured on a digital recorder and their presence or absence confirmed subsequent to the playbacks, these response variables were not subject to interpreter bias.

To determine male response to Type II songs, we used 3 playback types: 1) a display (Type I) song (n = 15), 2) a Type II song (n = 16), or 3) a nondisplay control that was a series of 3–4 seet alarms (n = 9), contact calls (n = 11), or chatter calls (n = 5). Spectrograms of all vocalization types are shown in Greig and Pruett-Jones (2008). We recorded 3 nonexclusive responses: “look,” “sing” (Type I song), and “chase female.” We chose the response variable “look” to capture changes in vigilance to different playback types. A look response consisted of a male turning his body or changing perches to face the speaker. We chose the response variable “sing” because Type I songs are regularly given by males during territorial encounters (Payne et al. 1988; Rowley and Russell 1997; Greig and Pruett-Jones 2008) and therefore likely represent an aggressive response to a conspecific in this context. Using Raven 1.2, we created spectrograms of the playback trial and measured the latency of song responses as the time from the end of the playback to the beginning of the song response. We considered songs that overlapped the playback as having a latency of zero. We chose the response variable “chase female” because it was an unambiguous response that occurred immediately after the playback and consisted of the male flying directly to his social mate and displacing her multiple times in quick succession. Although we were unsure if the chase response functioned in mate guarding or display, it undoubtedly indicated a heightened state of agitation during a territorial intruder. Focal birds would often move into vegetation and out of view during chase responses, so we were not able to determine precise duration information and so code chases simply as present or absent.

Because our playbacks consisted of only a single song or brief series of nondisplay calls, males did not engage in prolonged territorial responses of the sort typically observed in playback studies, such as flying toward the speaker or singing multiple response songs. These prolonged territorial responses can be readily elicited from splendid fairy-wrens using playbacks of multiple songs (Payne et al. 1988; Greig EI and Pruett-Jones S, personal observation), but we wanted to keep our playback regime as simple as possible and therefore chose to only use playbacks of single songs and measures of immediate or near-immediate responses.

To determine female response to predator and nonpredator calls, we used 2 playback types: 1) a butcherbird (n = 18), or 2) a nonpredator (n = 18). For the nonpredator category, we used vocalizations of 3 species that were common in the park, do not prey on adult fairy-wrens (Slater et al. 1989), and that frequently produced loud conspicuous vocalizations: white-winged choughs (Corcorax melanorhamphos), galahs (Cacatua roseicapilla), and gray shrike-thrushes (Colluricincla harmonica). We recorded 3 nonexclusive response variables: “look,” “fly,” and “seet” (alarm call). We conducted all of these playbacks when females were on the ground foraging, so the “look” response always consisted of the female ceasing her foraging behavior and standing erect while looking toward the speaker. The “fly” response consisted of the female ceasing her foraging behavior and flying from the ground to nearby vegetation. These behavioral responses either did not occur at all or occurred immediately after the playback such that variation in latency was too small to accurately dictate into the microphone. We attempted to record details on the destination of female flight (e.g., to covered or to open vegetation), but this proved problematic because categorization of vegetation was often ambiguous, and the initial location of females at the time of playback determined what types of vegetation were available and at what relative distances. Thus, we do not consider destination of flight variables further. “Seet” calls were captured on the audio recordings so we were able to extract accurate information on latency and number of calls, but because seet calls were never given in response to nonpredator playbacks, coding seet response simply as present or absent sufficiently captured variation between playback types, so we do not analyze seet responses in more detail.

To determine female response to solo and butcherbird-preceded Type II songs, we used 3 playback types: 1) a solo Type II song (n = 25), 2) a solo butcherbird call (n = 18), or 3) a Type II song preceded by a butcherbird call (n = 20) (Figure 1). We used the results from butcherbird playbacks from the previously described playback series in this analysis because those playbacks were conducted in the same manner and during the same time frame. To create the butcherbird-preceded Type II song playbacks, we used Raven 1.2 to join a butcherbird recording and a Type II song, with a delay between vocalizations of 0.2–0.5 s. We used the same pool of butcherbird calls and Type II songs as for the solo butcherbird and Type II song playbacks; however, we were careful never to test a female with the same butcherbird call or Type II song she had heard before. We recorded the response variables “look”, “fly”, and “sing” (Type I song). We chose the response variable “sing” because females will occasionally give Type I songs in response to territorial intruders (Payne et al. 1988; Rowley and Russell 1997; Greig and Pruett-Jones 2008). The function of female song in response to intruding males may be either territorial (e.g., Cooney and Cockburn 1995) or may be for the purposes of encouraging extrapair displays (e.g., Langmore 2000), but in either case it represents a response to an intruding individual and therefore allows us to measure the degree of female interest in a male playback. We chose the response variables “look” and “fly” to capture changes in female vigilance among playback types.

Figure 1

Spectrogram of a butcherbird vocalization followed by a Type II song that was constructed and used for playback to females. BB indicates the beginning of the butcherbird call and TII indicates the beginning of the Type II song.

Figure 1

Spectrogram of a butcherbird vocalization followed by a Type II song that was constructed and used for playback to females. BB indicates the beginning of the butcherbird call and TII indicates the beginning of the Type II song.

Condition dependence of Type II singing behavior

We collected 5 morphological and social indicators of male condition: 1) age, either known or minimum age, 2) social status, breeder or helper, 3) body condition, residuals of a regression of mass on tarsus length, 4) cloacal protuberance volume (Tuttle et al. 1996), and 5) percent blue on breast, range from 0% in completely brown males to 95% in very blue individuals (Tarvin et al. 2005). Percent blue on breast was estimated visually in the field and was done by the same observer (S.P-J.) for all males so that any error or subjectivity in the estimation was consistent across males. To determine male tendency to sing Type II songs, we played a single butcherbird recording to each of 90 males and noted whether or not the focal male gave a Type II song in response, which always occurred immediately after the playback or not at all. We used 31 different butcherbird recordings, and each male was tested only once. For descriptive purposes, we also recorded the occurrence of 3 display behaviors associated with butcherbird playbacks; “face fan” (erecting and spreading the cheek feathers; Rowley and Russell 1990) “sea horse flight” (slow, upright, undulating flight; Rowley and Russell 1990), and “chase female.” We played 1–4 additional consecutive butcherbird recordings when documenting these display behaviors, which often occurred after the second or third butcherbird playback.

Statistical analysis

For playback experiments, we used 2-tailed Fisher's Exact tests to investigate pairwise comparisons of interest in response variables. Because our comparisons were planned (a priori) based on the experimental design and our questions of interest, we do not make Bonferroni adjustments to P values and consider results significant at P ≤ 0.05.

For our analysis of the condition dependence of male Type II singing behavior, we scored male Type II response as 1 (yes) or 0 (no) and used logistic regression for continuous variables (body condition, cloacal protuberance volume, and % blue on breast) because this test does not require normality in the data. We used a chi-square distribution to test for an effect of age on Type II singing behavior and categorized age as 1, 2, 3, or 4+. We used a Fisher's Exact test to compare Type II song responses of breeder and helper males. We did not conduct additional multivariate analyses because testing each condition variable independently biases us toward detecting an effect, but we found no effect for any variable and had a sufficiently large sample size (n = 90) to minimize Type II error.

RESULTS

Males were more likely to sing and tended to chase their mate more in response to Type I (display) songs and Type II songs than to nondisplay calls (Table 1; Figure 2), although the difference in chase response between Type II songs and nondisplay calls was nonsignificant (P = 0.052). There were no significant differences in song latency or occurrence of male sing or chase responses between Type 1 and Type II songs, and there were no significant differences in male look responses between any of the 3 playback types. Thus, only Type I and Type II songs elicited behaviors consistent with territorial aggression or agitation, although males became more attentive in response to all playback types.

Table 1

Male response to Type I song (TI), Type II song (TII), and nondisplay (ND) playbacks

Response TI (15)
 
TII (16)
 
ND (25)
 
P 
Yes No Yes No Yes No 
Look 17 NS 
Chase 12 13 25 0.046 TI 
0.052 TII 
Sing 25 0.001 TI 
0.003 TII 
Song Latency 1.20 ± 1.13 s 1.37 ± 1.54 s — NS 
Response TI (15)
 
TII (16)
 
ND (25)
 
P 
Yes No Yes No Yes No 
Look 17 NS 
Chase 12 13 25 0.046 TI 
0.052 TII 
Sing 25 0.001 TI 
0.003 TII 
Song Latency 1.20 ± 1.13 s 1.37 ± 1.54 s — NS 

Sample sizes are given in parentheses after each playback type. Song latency is given as mean ± SD. P values are shown for comparisons of either Type I (TI) or Type II (TII) verses nondisplay playbacks. No significant differences exist between comparisons not shown. All tests are 2-tailed Fisher's Exact tests.

Figure 2

Number of males with positive (yes) responses for 3 response variables look, chase, and sing recorded for 3 playback types. TI represents display songs, TII represents Type II songs, and ND represents nondisplay vocalizations (seet, contact, and chatter calls). Error bars indicate binomial standard error.

Figure 2

Number of males with positive (yes) responses for 3 response variables look, chase, and sing recorded for 3 playback types. TI represents display songs, TII represents Type II songs, and ND represents nondisplay vocalizations (seet, contact, and chatter calls). Error bars indicate binomial standard error.

Females were significantly more likely to look, fly, and seet in response to butcherbird calls than in response to nonpredator controls (Table 2; Figure 3). Females were also more likely to look and fly in response to butcherbird calls than in response to solo Type II songs, more likely to fly in response to butcherbird-preceded Type II songs than solo Type II songs, and tended to look more in response to butcherbird-preceded Type II songs than solo Type II songs (Table 3; Figure 4), suggesting that butcherbird calls increased female attentiveness more effectively than Type II songs. This is supported by the observation that, for the butcherbird-preceded Type II song playbacks, the look and fly responses always occurred immediately after the butcherbird vocalization and before the Type II song had begun to play.

Table 2

Female responses to nonpredator (NP) and butcherbird (BB) playbacks

Response NP (18)
 
BB (18)
 
P 
Yes No Yes No 
Look 10 15 0.003 
Fly 14 13 0.002 
Seet 18 11 <0.001 
Response NP (18)
 
BB (18)
 
P 
Yes No Yes No 
Look 10 15 0.003 
Fly 14 13 0.002 
Seet 18 11 <0.001 

Sample sizes are given in parentheses after each playback type. All tests are 2-tailed Fisher's Exact tests.

Table 3

Female responses to butcherbird (BB), Type II song (TII), and butcherbird-preceded Type II song (BB + TII) playbacks

Response BB (18)
 
TII (25)
 
BB + TII (20)
 
P 
Yes No Yes No Yes No 
Look 15 11 13 0.001 TII versus BB 
0.093 TII versus BB + TII 
Fly 13 15 12 0.003 TII versus BB 
0.020 TII versus BB + TII 
Sing 17 23 13 0.045 BB + TII versus BB 
0.057 BB + TII versus TII 
Response BB (18)
 
TII (25)
 
BB + TII (20)
 
P 
Yes No Yes No Yes No 
Look 15 11 13 0.001 TII versus BB 
0.093 TII versus BB + TII 
Fly 13 15 12 0.003 TII versus BB 
0.020 TII versus BB + TII 
Sing 17 23 13 0.045 BB + TII versus BB 
0.057 BB + TII versus TII 

Sample sizes are given in parentheses after each playback type. P values are shown for comparisons of interest. No significant differences exist between comparisons not shown. All tests are 2-tailed Fisher's Exact tests.

Figure 3

Number of females with positive (yes) responses for 3 response variables look, fly, and seet recorded for 2 playback types. BB represents butcherbird playbacks and NP represents nonpredator controls. Error bars indicate binomial standard error.

Figure 3

Number of females with positive (yes) responses for 3 response variables look, fly, and seet recorded for 2 playback types. BB represents butcherbird playbacks and NP represents nonpredator controls. Error bars indicate binomial standard error.

Figure 4

Number of females with positive (yes) responses for 3 variables look, fly, and sing recorded for 3 playback types. TII represents a solo Type II song, BB represents a solo butcherbird call, and BB + TII represents a butcherbird vocalization immediately followed by a Type II song. Error bars indicate binomial standard error.

Figure 4

Number of females with positive (yes) responses for 3 variables look, fly, and sing recorded for 3 playback types. TII represents a solo Type II song, BB represents a solo butcherbird call, and BB + TII represents a butcherbird vocalization immediately followed by a Type II song. Error bars indicate binomial standard error.

Females sang significantly more frequently in response to butcherbird-preceded Type II songs than solo butcherbird calls and tended to sing more often in response to butcherbird-preceded Type II songs than solo Type II songs (Table 3; Figure 4). The female song response always occurred after the Type II song had played, so it was not in response to the butcherbird call but rather in response to the Type II song that followed the butcherbird call. Butcherbird-preceded Type II song playbacks were longer (mean ± standard deviation [SD] = 4.6 ± 0.84 s) than either solo Type II (1.85 ± 0.39 s) or solo butcherbird (2.32 ± 0.86 s) playbacks, but within the butcherbird-preceded Type II song playbacks (range: 3.19–7.06 s), variation in playback length had no effect on female look, fly, or sing responses (logistic regression; look, χ2 = 0.69, n = 19, P = 0.404; fly, χ2 < 0.01, n = 20, P = 0.980; sing, χ2 = 0.26, n = 20, P = 0.606).

Males giving Type II songs in response to butcherbirds were not significantly different than those not singing for all 5 social and morphological traits (P > 0.40 for all comparisons). When females (social mates) were present during the playbacks, males occasionally (in 15 of 78 cases, 19.2%) exhibited additional display behavior, primarily face fans, in response to butcherbird playbacks. Face fans were usually conspicuously directed toward the resident female, which was evident because the focal male would move close to the female and face her during the display.

DISCUSSION

The response of male fairy-wrens to both Type I and Type II songs strongly suggests that Type II songs are displays directed to conspecifics, despite often being associated with the presence of a predator. The degree to which Type II songs are territorial or sexual in nature is uncertain, but they are clearly functionally distinct from nondisplay vocalizations such as seet alarms and contact calls, which elicited no sing or chase responses in males. It is noteworthy that nondisplay vocalizations elicited the same proportion of look responses as Type I and Type II songs because it confirms that the lack of territorial response to these nondisplay vocalizations was not because males were unaware of the playback but instead because males were not motivated to respond in a territorial manner. The number of sing and chase responses to Type I and Type II song playbacks may appear low for a territorial response (<50% of focal males responded), but it is important to note that these are responses to only a single intruder song. The majority of males will sing and approach a speaker in response to playbacks of multiple intruder songs (Payne et al. 1988; Greig EI and Pruett-Jones S, personal observation), so the comparatively low level of response is likely because our intruder playbacks were brief. Thus, the sing responses from focal males are likely to be territorial in function but simply of a lower intensity than territorial responses to multiple intruder songs.

In support of the interpretation that Type II songs are displays directed toward conspecifics is the observation that butcherbird playbacks sometimes caused focal males to engage in visual display behaviors such as face fans and sea horse flights, which were directed toward females and were often paired with Type II songs. Not only were Type II songs sometimes used in conjunction with visual conspecific-directed displays but the fact that butcherbird calls stimulated such visual displays supports the idea that male fairy-wrens engage in courtship behavior in response to predator vocalizations.

Our playbacks of butcherbirds and nonpredators to females indicate that females can distinguish between predators and nonpredators based solely on their vocalizations and that butcherbird calls increase female vigilance more effectively than nonpredator calls. This may explain why males preferentially sing in response to butcherbirds rather than other avian species (Zelano et al. 2001), if singing Type II songs after heterospecific calls is a strategy for enhancing conspecific signal detection. Additionally, females tended to sing more in response to butcherbird-preceded Type II songs than to solo Type II songs, which is consistent with the interpretation that hitchhiking a Type II song to a butcherbird call is an effective communication strategy. One caveat of this result is that butcherbird-preceded Type II song playbacks were longer in duration than solo Type II song or solo butcherbird playbacks. It is therefore possible that the increased level of female response to butcherbird-preceded Type II song playbacks was due to the longer duration of that playback type rather than the alerting properties of butcherbird vocalizations. Although we cannot rule out that possibility using our current data set, variation in playback duration within the butcherbird-preceded Type II song playback type did not influence female response, which suggests that playback duration may be less important than playback content.

The function of female song in response to male Type II songs remains uncertain. Females may be responding to intruding males in a territorial manner or may be singing in response to Type II songs from intruding males because those males represent potential extrapair partners. We did not consistently measure the responses of mates during playbacks to focal birds and we therefore do not statistically analyze mate response, but it is worth noting that after approximately 70% of both solo butcherbird and butcherbird-preceded Type II song playbacks, the female's mate responded to the butcherbird vocalization with a Type II song. Because females did not sing in response to solo butcherbird playbacks even when their mate gave a Type II song, female song response appears to be preferentially stimulated by butcherbird-preceded Type II songs from intruding males rather than Type II songs from their mate. Females were occasionally observed singing in response to Type I song and Type II song playbacks directed to focal males, and in all cases, these were playbacks in which the focal male sang a Type I song in response to the playback. In the few instances where focal females sang in response to Type II song playbacks directed to those females, these were also accompanied by Type I song responses from their mate. Thus, female song response seemed to be associated with mate Type I song response, with the exception of female song response to butcherbird-preceded Type II songs; in these playbacks, females sang even when their mate did not sing a Type I song. Future work specifically investigating the influence of mate response on focal female response would be valuable in determining the relative importance of territoriality and extrapair mate attraction in the function of female song, particularly in response to butcherbird-preceded Type II songs.

Predator vocalizations may provide a form of costly signal elaboration, either because it is dangerous to sing in the presence of a predator (Møller et al. 2006) or because it is difficult to respond to such an unpredictable stimulus (Langmore and Mulder 1992). These hypotheses predict that giving a Type II song is condition dependent, but in all social and morphological traits measured for this study, males that gave Type II songs in response to butcherbirds were not significantly different than nonresponding males. This may be because variation in a male's motivation to sing Type II songs is more dependent on immediate circumstances than on individual condition. It is also possible that a male's tendency to respond to butcherbird calls is not adequately captured with a single playback trial and measuring male Type II song responses to multiple butcherbird playbacks is necessary to reliably estimate Type II singing tendency. Even if singing a Type II song in response to a predator is not a costly, condition-dependent behavior, it is still possible that aspects of Type II song acoustic structure, such as trill length (e.g., Dalziell and Cockburn 2008) or trill rate, contain condition-dependent information about male quality.

We do not rule out condition dependence of Type II singing behavior with our current data set, but the lack of evidence for condition dependence is consistent with the idea that singing after a butcherbird is an alerting strategy that any male, regardless of quality, can employ. In this scenario, the predator call functions as an “attractive modifier” (Gualla et al. 2008) that does not necessarily make more obvious any information contained in the male's signal but simply makes the message more likely to be received. Under this interpretation, the reason that females tend to sing more in response to butcherbird-preceded Type II songs than to solo Type II songs is because females are particularly alert or “excited” after hearing a butcherbird call and are thus more likely to detect and respond to subsequent Type II songs. The observation that males sing only in response to conspicuous predator vocalizations, but not in response to silent predators (Greig and Pruett-Jones 2009), strengthens this interpretation; it is not merely the presence of a potentially unnoticed silent predator that is important but rather the presence of an obvious predator that has attracted the attention of receivers. Using attentiveness caused by the presence of a predator to enhance signal detection is akin to an individual adding alerting components to its own signals. Alerting components are simple, conspicuous, easily recognizable additions that get a receiver's attention and precede a more complex message (Richards 1981). Organisms as taxonomically widespread as mammals, birds, amphibians, and reptiles are thought to incorporate alerting components into their signals (e.g., Gerhardt 1976; Richards 1981; Mitchell et al. 2006; Ord and Stamps 2008) but the effectiveness and extent of this phenomena have been relatively little studied (Wiley 2006).

If the only requirement for an effective alerting signal is that it attracts the attention of receivers, then vocalizations from many species of avian predators could potentially be used for this purpose. Splendid fairy-wrens at our study site, however, have a strong preference for singing Type II songs in response to butcherbirds and only sporadically sing in response to vocalizations of other avian predators (Zelano et al. 2001). Butcherbirds are sit and wait predators that hunt by swooping to the ground to capture prey (Greig EI and Pruett-Jones S, personal observation), so they may present a predation threat to fairy-wrens that are foraging in the open and unaware of their presence but not to fairy-wrens that are alerted and in vegetation. This would explain why fairy-wrens become attentive and fly to vegetation in response to butcherbird vocalizations but then appear to be willing to engage in behaviors such as songs and face fans. Preferentially singing in response to butcherbirds may therefore be because butcherbirds are an ideal species to use as an alerting signal for conspecific display; they give conspicuous and recognizable vocalizations, they require immediate attention by potential prey, and they may not be extremely dangerous once potential prey are aware of their presence.

We attempted additional playbacks in 2009 to determine if nonpredator vocalizations also increased female response to Type II songs or if butcherbird calls were unique in this respect. These playbacks, however, were uninformative because during that season female song response to all playbacks was particularly low (and the comparison thus uninformative). We suspect that the overall lack of female response was due to their advanced nesting stage at the time of the playback experiments compared to the previous year. Although a comparison of female response to butcherbird-preceded Type II songs and nonpredator-preceded Type II songs would be a valuable experiment, the lack of this experiment does not undermine our current interpretation for 2 reasons. First, even if nonpredator calls resulted in increased female response to Type II songs relative to solo Type II songs, that would simply imply that heterospecific vocalizations other than butcherbird calls may function as alerting signals, which is not unreasonable. Second, the observations that females are more attentive in response to butcherbirds and that males preferentially sing Type II songs in response to butcherbirds (Zelano et al. 2001) indirectly support the interpretation that butcherbird calls are used by males because they make particularly effective alerting signals.

Sexual or territorial displays are typically thought to be negatively associated with the presence of predators, but our results indicate that Type II songs are displays directed to conspecifics despite being closely associated with the presence of a predator. Our results also suggest that that the presence of a predator may actually enhance signal detection in receivers. Predator-associated behaviors are often thought to have some defensive antipredator function, but we have shown that may not always be the case. We suggest that this signaling strategy may occur in other species (e.g., fairy gerygones, Gerygone palpebrosa; Murphy 2002) and need not apply only to auditory traits but could be used for any signal that is more readily detected by alerted receivers.

FUNDING

National Science Foundation to S.P-J.; University of Chicago Hinds Fund; American Ornithologists' Union; Animal Behaviour Society; Frank M. Chapman Memorial Fund; Graduate Assistance in Areas of National Need training fellowship to E.G.

We thank John Bates, Jill Mateo, Trevor Price, and Tim Wootton for comments on the manuscript. Carolyn Johnson, Elizabeth Scordato, and Jill Mateo loaned equipment for fieldwork. Nick Brandley, Katherine Spendel, Bryce Masuda, Alex Tuchman, Karan Odom, Ben Parker, Leah Fisher, Emily Kay, Alex Smith, and Caroline Novak provided assistance in the field. Approval for all behavioral manipulations was obtained from the University of Chicago Animal Care and Use Committee (ACUP permit number 71708) and the Government of South Australia Department of Environment and Heritage (Wildlife Ethics Committee approval number 21/2006, Scientific Research Permit number C25249 and animal use license number 187).

References

Bernal
XE
Rand
AS
Ryan
MJ
Sexual differences in the behavioral response of tungara frogs, Physalaemus pustulosus, to cues associated with increased predation risk
Ethology
 , 
2007
, vol. 
113
 (pg. 
755
-
763
)
Cooney
R
Cockburn
A
Territorial defense is the major function of female song in the superb fairy-wren, Malurus cyaneus
Anim Behav
 , 
1995
, vol. 
49
 (pg. 
1635
-
1647
)
Dalziell
AH
Cockburn
A
Dawn song in superb fairy-wrens: a bird that seeks extra-pair copulations during the dawn chorus
Anim Behav
 , 
2008
, vol. 
75
 (pg. 
489
-
500
)
Elgar
MA
Predator vigilance and group-size in mammals and birds: a critical review of the empirical evidence
Biol Rev Camb Philos Soc
 , 
1989
, vol. 
64
 (pg. 
13
-
33
)
Ellis
JMS
Anti-predator signals as advertisements: evidence in white-throated magpie-jays
Ethology
 , 
2009
, vol. 
115
 (pg. 
522
-
532
)
Evans
JP
Kelly
JL
Ramnarine
IW
Pilastro
A
Female behavior mediates male courtship under predation risk in the guppy (Poecilia reticulata)
Behav Ecol Sociobiol
 , 
2002
, vol. 
52
 (pg. 
496
-
502
)
Gardner
JL
Experimental evidence for edge-related predation in a fragmented agricultural landscape
Aust J Ecol
 , 
1998
, vol. 
23
 (pg. 
311
-
321
)
Gerhardt
HC
Significance of two frequency bands in long distance vocal communication in the green treefrog
Nature
 , 
1976
, vol. 
5562
 (pg. 
692
-
694
)
Godin
JGJ
McDonough
HE
Predator preference for brightly colored males in the guppy: a viability cost for a sexually selected trait
Behav Ecol
 , 
2003
, vol. 
14
 (pg. 
194
-
200
)
Greig
EI
Pruett-Jones
S
Splendid songs: the vocal behavior of splendid fairy-wrens (Malurus splendens melanotus)
Emu
 , 
2008
, vol. 
108
 (pg. 
103
-
114
)
Greig
EI
Pruett-Jones
S
A predator-elicited song in the splendid fairy-wren: warning signal or intraspecific display?
Anim Behav
 , 
2009
, vol. 
78
 (pg. 
45
-
52
)
Greig
EI
Spendel
K
Brandley
NC
A predator elicited vocalisation in the Variegated Fairy-wren (Malurus lamberti)
Emu
 , 
2010
, vol. 
110
 (pg. 
165
-
169
)
Gualla
F
Cermelli
P
Castellano
S
Is there a role for amplifiers in sexual selection?
J Theor Biol
 , 
2008
, vol. 
252
 (pg. 
255
-
271
)
Johnstone
RA
Conspiratorial whispers and conspicuous displays: games of signal detection
Evolution
 , 
1998
, vol. 
52
 (pg. 
1554
-
1563
)
Kim
TW
Christy
JH
Dennenmoser
S
Choe
JC
The strength of a female mate preference increases with predation risk
Proc R Soc B Biol Sci
 , 
2009
, vol. 
276
 (pg. 
775
-
780
)
Langmore
NE
Espmark
Y
Amundsen
T
Rosenqvist
G
Why female birds sing
Animal signals: signalling and signal design in animal communication
 , 
2000
Trondheim (Norway)
Tapir Academic Press
(pg. 
317
-
327
)
Langmore
NE
Mulder
RA
A novel context for bird song: predator calls prompt male singing in the kleptogamous Superb Fairy-Wren, Malurus cyaneus
Ethology
 , 
1992
, vol. 
90
 (pg. 
143
-
153
)
Lea
AM
Gray
JT
The food of Australian birds: an analysis of the stomach contents. Parts II through IV
Emu
 , 
1935
, vol. 
35
 (pg. 
63
-
98
145–178, 251–280, 335–347
Magnhagen
C
Predation risk as a cost of reproduction
Trends Ecol Evol
 , 
1991
, vol. 
6
 (pg. 
183
-
185
)
Magurran
AE
Seghers
BH
Risk sensitive courtship in the guppy (Poecilia reticulata)
Behaviour
 , 
1990
, vol. 
112
 (pg. 
194
-
201
)
Mitchell
BR
Makagon
MM
Jaeger
MM
Barrett
RH
Information content of coyote barks and howls
Bioacoustics
 , 
2006
, vol. 
15
 (pg. 
289
-
314
)
Møller
AP
Nielsen
J
Garamszegi
LZ
Song post exposure, song features, and predation risk
Behav Ecol
 , 
2006
, vol. 
17
 (pg. 
155
-
163
)
Møller
AP
Nielsen
JT
Garamzegi
LZ
Risk taking by singing males
Behav Ecol
 , 
2008
, vol. 
19
 (pg. 
41
-
53
)
Murphy
S
Why do male fairy gerygones Gerygone palpebrosa burst into song on hearing predators or loud noises? Sunbird
2002
, vol. 
32
 (pg. 
62
-
66
)
Ord
TJ
Stamps
JA
Alert signals enhance animal communication in “noisy” environments
Proc Natl Acad Sci U S A
 , 
2008
, vol. 
105
 (pg. 
18830
-
18835
)
Payne
RB
Payne
LL
Rowley
I
Kin and social relationships in splendid fairy-wrens: recognition by song in a cooperative bird
Anim Behav
 , 
1988
, vol. 
36
 (pg. 
1341
-
1351
)
Pruett-Jones
S
Greig
EI
Rowe
M
Roche
E
The effects of sex, age and social status on annual survival in the splendid fairy-wren
Condor
 , 
2010
, vol. 
112
 (pg. 
369
-
377
)
Richards
DG
Alerting and message components in songs of rufous-sided towhees
Behaviour
 , 
1981
, vol. 
76
 (pg. 
223
-
249
)
Rowe
C
Skelhorn
J
Avian psychology and communication
Proc R Soc B Biol Sci
 , 
2004
, vol. 
271
 (pg. 
1435
-
1442
)
Rowley
I
Russell
EM
Fairy-wrens and grasswrens: Maluridae
 , 
1997
New York
Oxford University Press
Rowley
I
Russell
EM
“Philandering”—a mixed mating strategy in the splendid fairy-wren Malurus splendens
Behav Ecol Sociobiol
 , 
1990
, vol. 
27
 (pg. 
431
-
437
)
Slater
P
Slater
P
Slater
R
The Slater Field Guide to Australian Birds
 , 
1989
Willoughby (Australia)
Weldon Publishing
Tarburton
MK
Predation of willie wagtail and fledglings by pied butcherbird
Aust Bird Watch
 , 
1991
, vol. 
14
 (pg. 
68
-
69
)
Tarvin
KA
Webster
MS
Tuttle
EM
Pruett-Jones
S
Genetic similarity of social mates predicts the level of extrapair paternity in splendid fairy-wrens
Anim Behav
 , 
2005
, vol. 
70
 (pg. 
945
-
955
)
Tuttle
EM
Pruett-Jones
S
Webster
MS
Cloacal protuberances and extreme sperm production in Australian fairy-wrens
Proc R Soc B Biol Sci
 , 
1996
, vol. 
263
 (pg. 
1359
-
1364
)
Webster
MS
Tarvin
KA
Tuttle
EM
Pruett-Jones
S
Reproductive promiscuity in the splendid fairy-wren: effects of group size and auxiliary reproduction
Behav Ecol
 , 
2004
, vol. 
15
 (pg. 
907
-
915
)
Wiley
RH
Brockmann
HJ
Slater
PJB
Snowdon
CT
Roper
TJ
Naguib
M
Wynne-Edwards
KE
Signal detection and animal communication
Advances in the study of behavior
 , 
2006
, vol. 
Vol. 36
 
San Diego (CA)
Elsevier Academic Press Inc
(pg. 
217
-
247
)
Zahavi
A
Mate selection—a selection for a handicap
J Theor Biol
 , 
1975
, vol. 
53
 (pg. 
205
-
214
)
Zelano
B
Tarvin
KA
Pruett-Jones
S
Singing in the face of danger: the anomalous type II vocalization of the splendid fairy-wren
Ethology
 , 
2001
, vol. 
107
 (pg. 
201
-
216
)
Zuk
M
Kolluru
GR
Exploitation of sexual signals by predators and parasitoids
Q Rev Biol
 , 
1998
, vol. 
73
 (pg. 
415
-
438
)