The dual benefits of synchronized mating signals in a Japanese treefrog: attracting mates and manipulating predators

Abstract

In dense mating aggregations, such as leks and choruses, acoustic signals produced by competing male conspecifics often overlap in time. When signals overlap at a fine temporal scale the ability of females to discriminate between individual signals is reduced. Yet, despite this cost, males of some species deliberately overlap their signals with those of conspecifics, synchronizing signal production in the chorus. Here, we investigate two hypotheses of synchronized mating signals in a Japanese treefrog (Buergeria japonica): (1) increased female attraction to the chorus (the beacon effect hypothesis) and (2) reduced attraction of eavesdropping predators (the eavesdropper avoidance hypothesis). Our results from playback experiments on female frogs and eavesdropping micropredators (midges and mosquitoes) support both hypotheses. Signal transmission and female phonotaxis experiments suggest that away from the chorus, synchronized calls are more attractive to females than unsynchronized calls. At the chorus, however, eavesdroppers are less attracted to calls that closely follow an initial call, while female attraction to individual signals is not affected. Therefore, synchronized signalling likely benefits male B. japonica by both increasing attraction of females to the chorus and reducing eavesdropper attacks. These findings highlight how multiple selective pressures likely promoted the evolution and maintenance of this behaviour.

This article is part of the theme issue ‘Synchrony and rhythm interaction: from the brain to behavioural ecology’.

1. Introduction

Males of many species aggregate and produce conspicuous acoustic displays to attract mates. Examples of this phenomenon occur across taxa in the choruses and leks of mammals, birds, insects and anurans. Attracting mates in aggregations can be challenging, however, as mating signals produced at the same time among neighbouring males can overlap and acoustically interfere, reducing the ability of females to detect, localize and discriminate between signals [1–3]. In addition, when acoustic signals overlap but are offset (i.e. the signals only partially overlap in time), females often exhibit a bias for the initial ‘leading’ signal over the second ‘following’ signal. Known as the ‘precedence effect’ [4], or leader–follower preferences, this bias can override other acoustic preferences in female receivers [5–10]. Given the costs of fine-scale signal overlap, especially for producing a following signal, evolution by sexual selection is expected to result in signal timing strategies that reduce signal overlap. In anuran choruses, for example, conspecific neighbouring males commonly offset the timing of signal production in an alternating pattern, forming duets, trios, and quartets [11]. Similar patterns of signal overlap avoidance are common in insects [12] and birds [13] and in mating signals produced in other sensory modalities [14,15].

While alternation of signals is widespread, there is considerable variation in signal timing strategies across species [16]. Opposite to signal alternation, males of some species produce signals in ‘synchrony’, completely or partially overlapping their mating signals with neighbouring conspecifics in a chorus (auditory and visual signals in insects and anurans [14,17], birds [18], mammals [19], crustaceans [15,20]). In many of these synchronizing species, males actively produce following signals in response to the signals of neighbouring males [21–23]. The following signallers are thus responsible for signal synchronization, and potentially incur additional costs as females generally prefer leading over following signals [5–10]. Unlike the benefits of mating signal alternation (increased female attraction to the individual), the benefits of synchrony are less understood. In addition, any advantages males might gain from synchronizing must weigh against the assumed cost of reduced female attraction, both from signal interference and leader–follower preferences. Two potential benefits have been proposed to explain the use of synchronized mating signals: increased attraction of females from a distance (the beacon effect hypothesis [17,24]) and reduced attraction of eavesdropping predators and parasites (the eavesdropper avoidance hypothesis [25]).

Synchronized signals may have a greater active space compared to unsynchronized signals, acting as an acoustic ‘beacon’ [17,24] that attracts females from a greater distance [26–28]. Once within hearing range, males producing synchronized acoustic signals may also increase female attraction if the synchronized calls are more intense, as females across species generally prefer higher-amplitude signals [29]. The beacon effect is only advantageous, however, if the per capita increase in female attraction is higher for aggregations producing synchronized signals compared to unsynchronized ones [17]. An equivalent trade-off has received some attention in relation to group size and the evolution of aggregate breeding in general [30,31]. Larger aggregations of male signallers attract more females, but chorus size is limited by the increase in male–male competition and per capita mating advantage [17]. In this case, the selective forces that drive group size dynamics and signal timing strategies are analogous.

Synchronized mating signals may also benefit males in evading eavesdropping predators and parasites [25]. The conspicuousness of mating displays often exposes signallers to non-target receivers, including natural enemies [32,33]. Just as signal overlap may reduce female attraction through signal interference and leader–follower preferences, so too might it reduce the attraction of predatory eavesdroppers. Indeed, previous studies on a synchronizing neotropical frog found that predators prefer unsynchronized frog calls over synchronized calls [25,34], and bias leading over following calls within a synchronized chorus [35]. Thus, by producing following overlapped signals in response to calls produced by neighbouring conspecifics, males may mask their own signals and exploit the perceptual leader–follower biases of their enemies.

Here, we investigate these two non-mutually exclusive functions of mating signal synchronization, the beacon effect and eavesdropper avoidance hypotheses. Specifically, we conducted a series of field playback experiments to examine the benefits of synchronized calling in a Japanese stream-breeding treefrog, the Ryukyu Kajika frog (Buergeria japonica). To test the beacon effect hypothesis, we compare the amplitudes of synchronized and unsynchronized calls, and how each transmit through the environment. We also assess female preference for synchronized and unsynchronized calls at a distance from the chorus. If male B. japonica call synchronization acts as a beacon attracting females from a greater distance, we expect synchronized calls to reach higher amplitudes and for females to display a bias towards synchronized calls at a distance. To test the eavesdropper avoidance hypothesis, we assess the leader–follower preferences of micropredators (frog-biting midges and mosquitoes) that eavesdrop on B. japonica calls. If male synchronization helps avoid eavesdroppers, we expect reduced eavesdropper attraction to following calls. Finally, we weigh these potential benefits of signal synchrony against one potential cost: reduced female attraction through the leader–follower preferences of female B. japonica.

2. Methods

(a) . Study system and location

The Ryukyu Kajika frog (B. japonica), in the Rhacophoridae family, ranges from Taiwan through the Ryukyu Archipelago of Japan. During their breeding season, March to October, males of this species form choruses along the banks of streams and roadside ditches [36]. Chorus attendance is correlated with rainfall, and large choruses of over 300 males have been observed on nights following periods of heavy rains [37]. Although specific calling behaviour is variable between populations, calls are generally divided into two types, one of which is produced in synchrony [37]. This synchronized call is a trill of variable duration (ranging from less than 0.5 s to more than 3.0 s [38]) with a dominant frequency around 3 kHz and likely plays a role in attracting mates [39,40]. Male B. japonica respond to the calls of neighbouring males at an average latency of 0.35 ± 0.05 s, resulting in around 70% overlap in call duration between leading and following calls [37]. In addition, B. japonica calls attract multiple species of eavesdropping insects (figure 1a), frog-biting midges (Corethrella spp.) and mosquitoes (Uranotaenia spp.), which take blood meals from calling males [41,42]. The costs of attracting frog-biting insects include blood loss [43] and spread of diseases [44–46]. All field experiments were conducted in July and August 2016 on Iriomote Island in southern Okinawa, Japan (24°23'30.3″ N, 123°52'48.8″ E).

Figure 1.

The study system (a), including the sender, target receiver and non-target receivers of B. japonica calls. Oscillograms and spectrograms (b) of a natural B. japonica call and the synthetic call used in this study. A larger-scale view of the spectrograms can be found in the electronic supplementary material, figure S1. (Online version in colour.)

(b) . Acoustic playbacks

For each experiment, acoustic playbacks of synthetic B. japonica calls were broadcast using Pignose portable amplifier speakers (Model 7–100) at a peak amplitude of 80 ± 1 dB SPL re. 20 µP (C-weighting, Brüel and Kjær digital sound-level meter Type 2250) measured at 1 m from the speaker at ground level. Preliminary studies suggested this amplitude was representative of calling intensity of males from this species in the field. The amplitude of each speaker was remeasured before each trial in each experiment. Speakers were paired in sets, spaced 1 m apart, to broadcast either synchronized (0.30 s of latency) or unsynchronized (alternating) calls at a rate of one call every 3.4 s. The synthetic B. japonica call was constructed in CoolEdit2000 (Syntrillium Software) using a representative note from a pre-recorded B. japonica call. The synthetic call was designed to roughly match the average characteristics of male calls in this population (dominant frequency = 3.25 kHz, call duration = 1.62 s; figure 1b, electronic supplementary material, table S1).

(c) . Signal transmission

As acoustic signals transmit through the environment they attenuate (reduce in amplitude) and degrade (erode in their spectro-temporal characteristics). We compared these aspects of signal transmission by broadcasting synchronized and unsynchronized B. japonica calls, and then rerecording these broadcasts at a distance of 1, 2, 3, 4, 5, 10, 20, 30, 40, 60 and 80 m from the centre point between the speakers at ground level (following previously established methods [47]). This signal transmission experiment was conducted at five interior forest sites. Each site was tested once (n = 5). Fast Fourier transform (FFT) recordings of the transmitted calls were collected using a Brüel and Kjær digital sound-level meter Type 2250 (10 s recording, 50 Hz frequency resolution over a range of 500–6000 Hz, Z-weighting). Differences in synchronized and unsynchronized call attenuation were assessed by comparing amplitudes of the dominant frequency (3.25 kHz) at each distance. Differences in spectral degradation were assessed using cross-correlation coefficients of calls at each distance. Coefficients were calculated using a Pearson correlation analysis comparing the FFT recording at a given distance with the recording taken at 1 m from the speaker.

(d) . Female synchronized–unsynchronized call preference

Following standard procedures used in phonotaxis experiments with female frogs [48], pairs of male and female B. japonica were collected in amplexus from a naturally occurring chorus. Each female frog (n = 24) was tested individually. Just prior to testing, a female was separated from the male and positioned under a cup at the centre of a circular phonotaxis arena. The arena was 2 m in diameter with open sides marked at 15° increments, located in a haphazardly selected forest site at a distance of more than 100 m from any breeding B. japonica. Two sets of speakers were placed on either side of the arena (180° from each other) at 40 m from the arena's centre. One speaker set broadcast synchronized calls while the other broadcast unsynchronized calls. While the speaker sets were always placed at 180° from each other, the positions of the sets relative to the arena were varied between trials to control for side biases. Individual speakers were also rearranged among the sets between trials. The arena was illuminated using red light, and female movement in the arena was observed from a blind placed perpendicular to the speakers. Females were given 5 min to acclimate to the arena under the cup, before being remotely released. The point at which the female left the arena to the closest 15° mark was recorded. Females were tested once and then returned with their paired male to their exact collection location.

(e) . Leading–following call preferences in eavesdropping micropredators

Frog-biting midges and mosquitoes were collected using standard sound traps for insects [49]. A collection device was placed over each speaker in a pair spaced 1 m apart broadcasting synchronized calls. Within a trial, one speaker always broadcast leading calls while the other always broadcast following calls. Which speaker broadcast which stimuli, leading or following calls, was randomized between trials. Collection sites were haphazardly chosen in the forest within 100 m of a water source. Traps were run once per night for 45 min starting 30 min after sunset [34]. Over a sampling period of 20 nights, a total of 97 frog-biting midges and 37 frog-biting mosquitoes were collected. At the end of each night, frog-biting insects were counted and preserved in 75% ethanol. Species were identified using established keys (Corethrella [50], Uranotaenia [51]). The number of mosquitoes and midges attracted each night to either the leading or following calls was used as an indicator of preference.

(f) . Leading–following call preferences in female frogs

Following the previously described procedures for phonotaxis experiments with female frogs, female B. japonica leading–following call preferences were assessed in a 2 m diameter open sided circular phonotaxis arena. Briefly, females were captured (n = 14), placed under a cup at the centre of the arena and given 5 min to acclimate before being released remotely. A pair of speakers broadcasting synchronized calls was placed at a distance 2 m from the centre of the arena. The speakers were spaced 1 m apart. Which speaker broadcast leading or following calls was randomized between trials. Female choice was scored after the female approached within 10 cm of a speaker [48]. Females were tested once and then returned to their exact collection locations with their paired males.

(g) . Statistical analyses

All statistical analyses were conducted using program R v. 3.5.2 [52]. To assess differences in call transmission for synchronized and unsynchronized calls, changes in amplitude and Pearson correlation coefficients were compared using Linear Mixed Models in the glmmTMB package [53]. The main effects and interaction between distance and call overlap (synchronized or unsynchronized) were included as fixed factors, with site included as a random factor. Differences in amplitude at the dominant frequency (3.25 kHz) between synchronized and unsynchronized calls across all distances were compared using the emmeans function in the emmeans package [54]. To assess the comparative attenuation of synchronized and unsynchronized calls, differences in the rate of change in amplitude over distance (slopes of the covariates in each model) were compared using the emtrends function in the emmeans package [54]. For call degradation, Pearson correlation coefficients of calls at each distance were calculated using the standard cor function in R, and the rates at which synchronized and unsynchronized calls degraded were compared using the emtrends function in the emmeans package [54]. To assess female B. japonica preference for synchronized–unsynchronized calls, the circular directions in which females left the arena (to the nearest 15° increment) were analysed using the CircStats package [55]. Parameters for the von Mises distribution of the angles, mean (μ) and dispersion (κ), were calculated from maximum-likelihood estimates. A modified Rayleigh test (V-test) of uniformity was used to test the null hypothesis that females left the arena in a random direction [56]. Leading–following call preferences in eavesdropping predators were analysed using two-tailed exact symmetry tests in the coin R package [57], testing the null hypothesis that leading and following calls attracted equal numbers of frog-biting insects (following [35]). The numbers of captured frog-biting mosquitoes and frog-biting midges were independently compared, blocked within each night. Finally, the leading–following call preferences of female B. japonica were assessed using a two-tailed exact binomial test. Given the small sample size for the female leading–following preference experiment, concerns about type II error were assessed using a G*Power 3.1 statistical power analysis [58].

3. Results

(a) . Signal transmission

Differences in amplitude at the dominant frequency between synchronized and unsynchronized B. japonica calls were variable over distance (electronic supplementary material, table S2a). Overall, however, the dominant frequency of synchronized calls was more intense by about 2 dB (estimate = 2.10 ± 0.99; t₁₀₀ = 2.120, p = 0.0364). Calls attenuated at a rate relative to the square root of distance, with no significant difference in the rate at which amplitude decreased between synchronized and unsynchronized calls (synchronized: estimate = −7.18 ± 0.27, unsynchronized: estimate = −7.05 ± 0.27; t₁₀₀ = −0.35, p = 0.729; figure 2a). Calls spectrally degraded at a linear rate, also with no significant difference between synchronized and unsynchronized calls (both rates: estimates = −0.01 ± < 0.01; t₁₀₀ = 1.79, p = 0.076; figure 2b, electronic supplementary material, table S2b).

Figure 2.

Attenuation (a) and degradation (b) of synchronized and unsynchronized B. japonica calls broadcast from five interior forest locations. Values for attenuation are the amplitude (dB) of the dominant frequency (3.25 kHz), while values for degradation are Pearson correlation coefficients. Regression lines are illustrated. (Online version in colour.)

(b) . Female synchronized–unsynchronized call preference

Female B. japonica displayed a bias for sets of synchronized calls over unsynchronized calls broadcast at a 40 m distance. When the synchronized speakers are set at 0°, and the unsynchronized speakers in turn are set at 180°, females exited the phonotaxis arena at a mean angle of 4.96° (κ = 1.13), significantly oriented towards the synchronized speaker set (mean resultant length = 0.492, p < 0.001; figure 3).

Figure 3.

Angle (to the nearest 15° increment) at which female B. japonica left the phonotaxis arena with synchronized calls broadcast at 0° and unsynchronized calls broadcast at 180°. Both stimuli were located at a distance 40 m from the centre of the arena. Each black circle represents an individual female frog. (Online version in colour.)

(c) . Leading–following call preferences in eavesdropping micropredators

Overall, the number of eavesdropping insects collected per night was low, less than 10 individuals on average. Leading speakers attracted about 2 more frog-biting midges per night compared to speakers broadcasting following calls (3.30 ± 2.62 midges for leading calls versus 1.35 ± 1.66 midges for following calls; Z = 2.65, p = 0.004; figure 4a). No significant difference in the attraction of frog-biting mosquitoes was detected between speakers (1.15 ± 1.14 mosquitoes for leading calls versus 0.90 ± 0.64 mosquitoes for following calls; Z = 0.85, p = 0.514; figure 4b).

Figure 4.

Receiver preferences for leading or following calls. Numbers of non-target receivers, midges (a) and mosquitoes (b), are the average collected per night (mean ± s.e.). Numbers of the target receiver, female B. japonica (c), represent a single choice by an individual female. A ‘*’ indicates a statistically significant difference (p < 0.05). (Online version in colour.)

(d) . Leading–following call preferences in female frogs

Female B. japonica displayed no significant preference for leading or following calls, with only 6 of the 14 females tested choosing the leading speaker (p = 0.791; figure 4c). This result contrasts with previous experiments examining leading–following call preferences in other frog species, which have found strong leading call preferences [7,9,34,59–62]. Using the average effect size from these studies, our female phonotaxis experiment had statistical power 1 − β = 0.69 (α = 0.05, effect size = 0.360). If female B. japonica had leading–following call preferences comparable to other anuran species, we might expect 12 of the 14 females tested to choose leading calls.

4. Discussion

Our results support both functions of synchronized mating signals: the beacon effect and as a predator avoidance strategy. Although synchronized and unsynchronized frog calls had similar rates of attenuation and degradation, synchronized calls were generally higher amplitude regardless of distance. In addition, female B. japonica displayed a bias for synchronized calls at a distance beyond the chorus. These findings suggest that a group of synchronizing male B. japonica may enjoy an advantage over a group of unsynchronized males in attracting female frogs. Our results also show that individuals within a synchronized chorus that produce following calls attract fewer frog-biting midges compared to individuals producing leading calls. Thus, midges that eavesdrop on B. japonica calls display a precedence effect, suggesting that male B. japonica may reduce the attraction of at least some micropredators to their individual calls by synchronizing. Furthermore, by also assessing leading–following call preferences in female B. japonica, we weigh both advantages of synchrony against one of the potential costs: reduced female attraction to following calls. Female B. japonica, however, did not display a strong preference for leading calls. A bias for leading calls is common in anuran species (e.g. [61,62]), suggesting that female B. japonica have reduced leading–following call preferences, releasing males from this trade-off of mating signal synchrony [34].

(a) . Signal synchronization as a ‘beacon’ for females

We found no evidence that synchronized and unsynchronized signals attenuate or degrade differently as they are transmitted through the environment. However, synchronized B. japonica calls were generally higher in amplitude compared to unsynchronized calls at the same distance. As predicted, we found that females displayed a bias for pairs of synchronized calls over unsynchronized calls at a distance from the chorus. It is presently unclear, however, why female B. japonica display this bias for synchronized calls. Higher-amplitude synchronized signals may have a greater active space (acting as a more intense ‘beacon’) and thus may be perceived by females from a greater distance (the traditional mechanism of the beacon effect hypothesis [17,24]). Previously, the beacon effect hypothesis has been used to explain the evolution of chorusing in general [30]. When animal aggregations are dense enough, signals unintentionally overlap to create high-amplitude ‘chorus noise’ that can be used by females to locate the chorus [63]. Intentional overlap of signals, therefore, may be a behavioural extension of this benefit. A female bias for synchronized calls could also result from a preference for higher-amplitude signals, where females perceive differences in amplitude between groups of males producing synchronized and unsynchronized calls. A preference for higher-amplitude signals is a common phenomenon in other frog species [64,65] and in general for female receivers of mating signals across taxa [29].

The benefits of higher-amplitude acoustic signals may be emphasized in certain habitats, such as those with high levels of abiotic background noise. High levels of abiotic background noise, such as from high wind conditions or running water, reduce the ability of receivers to detect and discriminate signals [66,67]. Through the beacon effect, synchronized calls may be more detectable in background noise compared to unsynchronized calls [12]. We therefore might expect to observe synchrony in species that chorus in noisy environments. Indeed, synchronization has been observed in several frog species that call near waterfalls and streams, including B. japonica, the pug-nosed treefrog (Smilisca sila [25]) and the canyon treefrog (Hyla arenicolor; unpublished data discussed in [12]). Furthermore, if signal synchronization is an adaptation to noisy environments, synchronizing species may also be more resilient to other types of background noise, such as anthropogenic noise from aircraft and automobile traffic, which can negatively impact anuran behaviour [68].

(b) . Synchronization as a predator avoidance strategy

Synchronized signals may reduce the attraction of eavesdropping predators if overlapping signals are harder to detect, recognize and localize, and if the eavesdroppers prefer leading to following calls [34,35]. We found that frog-biting midges are biased towards leading calls. Thus, by responding to the calls of neighbouring males with following calls, male B. japonica exploit the leader–follower preferences of this eavesdropper and reduce the risk of predation by midges. Similar leading–following call preferences have been observed in eavesdropping predators in other systems, both insect and mammalian [35,69].

If B. japonica synchronize their calls to avoid eavesdroppers, we might expect high rates of eavesdropper attraction in general and thus high selective pressure from predators. However, only about three or four midges were captured per night over the 45 min collection period. It seems unlikely that avoiding two or three more midges by producing following calls compared to leading calls would confer a biologically meaningful benefit for a male frog. This low abundance of eavesdroppers may also explain why we found no leader–follower differences in frog-biting mosquitoes, as our playbacks of calls only attracted about one individual per night. In other systems, such as frogs in the Neotropics, the same eavesdropper collection methods result in capturing hundreds of frog-biting insects in the same 45 min period [34,70]. In addition, B. japonica produce two calls types, a synchronized call type (the focus of our study) and a call type that is produced in alternation between neighbouring males [37]. When both call types are broadcast without overlap, both eavesdropping midges and mosquitoes prefer the alternating call type [71]. If signal synchrony functions to avoid eavesdroppers, why would male B. japonica synchronize the call type that is less attractive to those eavesdroppers? Given the small benefit that males may gain from avoiding eavesdropping predators in this system, it is unlikely that eavesdropper attraction is the primary selective pressure driving the use of synchronization in B. japonica. While the current abundance of eavesdroppers may not represent past selective pressures, our results suggest a greater payoff from the beacon effect function of synchrony, while avoiding eavesdroppers may be a tangential benefit.

(c) . Relaxed sexual selection as a driver of signal synchronization

Synchronization of mating signals may provide multiple benefits to male B. japonica, but at the potential cost of reduced short-distance female attraction to individual males. It is possible, however, that relaxed female selection for unsynchronized signals could release males from this trade-off imposed on signal synchrony [34]. Our results indicate that female B. japonica do not display strong leading–following call preferences, suggesting that male B. japonica may enjoy the benefits of producing following synchronized calls without incurring this cost. The precedence effect is a widespread phenomenon among animals that communicate acoustically (e.g. [61,72–74]). It is intriguing that reduced or reversed leader–follower preferences have been observed in other synchronizing anuran species in addition to B. japonica (Dendropsophus ebraccatus [75], Kassina fusca [22], S. sila [35]). Thus, relaxed sexual selection may be the key that allows signalling males to take advantage of mating signal synchrony.

(d) . Limitations and future research directions

In this study, we assess two potential functions of mating signal synchronization (the beacon effect and eavesdropper avoidance hypotheses) and one potential cost (leader–follower preferences of female B. japonica). There are remaining aspects of both functions, and additional costs to synchrony, that warrant further investigation in this system. For the beacon effect hypothesis, we assessed female attraction at only one distance (40 m). Yet, female preference for groups of synchronized or unsynchronized calls likely changes over distance. In another synchronizing frog species, the Neotropical pug-nosed treefrog (S. sila), females lack a preference for synchronized or unsynchronized calls at 1.5 m distance [34]. Further studies that examine female choice across space, at multiple distances within and beyond the chorus, would improve our understanding of how distance modulates the cost and benefits of female attraction.

For the eavesdropper avoidance hypothesis, we assessed the leader–follower preferences of predators for pairs of synchronized calls, but not preferences between synchronized and unsynchronized calls. A synchronized chorus may attract more eavesdroppers than an unsynchronized chorus through the beacon effect. For call synchronization to be advantageous, the increased per capita attraction of eavesdroppers to the chorus would have to be offset by the reduced risk to individual frogs producing following calls. Thus, similar to the need to examine female choice at varying distances, further studies are needed to better understand eavesdropper attraction within and beyond the chorus. More broadly, investigating how females and eavesdroppers detect, localize and discriminate individual signals and signal groups with different timing through the environment would be valuable for understanding the trade-off between natural and sexual selection that ensues by attracting both types of receivers.

The interpretations of our results for the female leading–following preference experiment are limited by our small sample size owing to the low abundance and responsiveness of female B. japonica. While our results suggest that sexual selection has relaxed in this species, a more robust female choice experiment with varying degrees of temporal overlap between signals is needed to better understand female signal timing preferences. In addition, auditory physiology experiments comparing neural responses to synchronous and non-synchronous species, specifically differences in temporal processing (e.g. [76]), would be valuable steps to understand the mechanisms underlying how females of some anuran species overcome the precedence effect.

In the experiments in this study, the stimuli were presented from sets of two speakers broadcasting synchronized calls or unsynchronized calls. In nature, however, choruses often include numerous signallers. For synchronizing species, one leading call in the chorus can be responded to by multiple followers (B. japonica: H. Legett 2016, personal observations; S. sila [25]). It is unclear how the costs and benefits associated with the leader–follower preferences of female receivers and eavesdroppers differ for synchronizing males when there are multiple following callers. Other complex timing interactions between multiple signallers can also arise. A following call may be produced in response to another following call, resulting in a ‘domino’ pattern of calls in the chorus (observed in S. sila and Smilisca sordida [77]). Furthermore, in the context of the beacon effect, the active space of synchronized calls may be influenced by the spatial positioning of males within the chorus. Male B. japonica form elongated linear choruses along the banks of streams [36], but in other synchronizing species males are clustered together over small areas (e.g. male S. sila form choruses on rocks and trees around waterfalls [25]). Experiments using a greater number of speakers, timed and positioned to replicate natural choruses, are critical to address these questions and move forward towards understanding the functions of mating signal synchronization in nature.

Finally, there is a gap in our knowledge of the natural history of species that vary in their mating signal timing strategies. Synchronized signalling is assumed to be uncommon in anurans, having only been reported in a handful of species in addition to B. japonica (Assa darlingtoni [78], Cochranella granulosa [79], D. ebraccatus [76], Diasporus diastema [80], H. arenicolor [12], K. fusca [22], K. kuvangensis [81], K. senegalensis [82], S. sila [21]). However, for anurans, timing of signal production is often not reported with other standard call features, such as signal duration and dominant frequency (e.g. [11]). The signalling behaviour of B. japonica, for example, has been the focus of previous studies [38,40], including the population on Iriomote Island [39] where this study was performed. Yet, synchronization in B. japonica has only recently been described and discussed [37]. Even in long-studied and common anuran species, signal synchronization has received limited consideration. For example, the beacon effect function of signal synchrony in anurans was originally proposed for American toads (Anaxyrus americanus), which produce long trills that sometimes overlap [83]. Female preferences in this toad species can be influenced by call overlap [84], yet follow-up empirical studies describing the contexts in which male A. americanus synchronize their calls have not been conducted. Thus, further natural history studies are needed, especially in species where synchronization has been suggested but has not been described and measured (e.g. A. americanus [83] and H. arenicolor [12]). Such studies will provide a much-needed foundation to fuel experimental studies that provide a broader perspective of the mechanisms and function of call synchronization in anurans.

Ethics

This research was approved by Purdue University (IACUC Protocol no. 1504001224A001). This research also followed appropriate ethical and legal guidelines under the direction of the Japan Ministry of the Environment, and all field experiments on Iriomote Island were conducted outside of the state protected zone.

Data accessibility

Data supporting this study are publicly available in the Purdue University Research Repository (https://purr.purdue.edu/publications/3716/1) (doi:10.4231/J9RZ-QF79).

Authors' contributions

H.D.L., I.A. and X.E.B. conceived and designed the study. H.D.L. collected data, performed the statistical analyses and drafted the manuscript. I.A. and X.E.B. helped draft and edit the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

This research was supported by an East Asia and Pacific Summer Institute Graduate Fellowship from the National Science Foundation and Japan Society for the Promotion of Science (award no. 1515380) and a Graduate School Summer Research Grant from Purdue University, both to H.D.L.