Sensory biases in response to novel complex acoustic signals in male and female grey treefrogs, Hyla chrysoscelis

Abstract

The sensory bias hypothesis proposes that female preferences for male sexual signalling traits evolved in contexts other than mating. Individuals of both sexes may experience similar selection pressures in these contexts; thus males may have similar biases to females for variation in signal traits. We tested this prediction in the grey treefrog, Hyla chrysoscelis, in which males produce simple advertisement calls, but females are more attracted to certain novel complex stimuli. We recorded males' responses to playbacks of both simple advertisement calls and complex calls consisting of the advertisement call with an acoustic appendage (filtered noise, or heterospecific call pulses) either leading or following the call. We tested females’ preferences for the same stimuli in phonotaxis tests. We found evidence for a sensory bias in both sexes: males gave more aggressive calls in response to complex stimuli and females sometimes preferred complex over simple calls. These biases were not universal and depended on both temporal order and appendage characteristics, but how these effects manifested differed between the sexes. Ultimately, our approach of studying biases of both sexes in response to novel mating signals will shed light on the origin of mating preferences, and the mechanisms by which sensory biases operate.

1. Introduction

The sensory bias hypothesis is a leading explanation for the evolution of sexually selected traits by female mate choice [1–3]. According to this hypothesis, sensory or neural processing mechanisms that evolved in a context other than mate evaluation nevertheless affect how females respond to male mating signals, for instance if enhanced sensitivity to certain wavelengths of light due to foraging preferences affects how females respond to male visual signals [4]. Furthermore, ‘hidden’ preference models emphasize that female preferences may arise as entirely incidental consequences of neural processing mechanisms [5]. Sensory biases are typically described as a characteristic of females [6–8]. However, all of these processes that lead to sensory biases generally apply to both sexes. Thus, we would expect that males would usually have similar sensory biases to females. For instance, female guppies, Poecilia reticulata, have a preference for orange mates, which is thought to be due to a sensory bias caused by foraging preferences for orange food [9]. Male guppies also have a preference for orange food [9], which supports the sensory bias hypothesis because it suggests a common underlying preference for orange that did not evolve specifically in the context of female mate choice.

In many species, both male and female receivers respond to male sexual advertisement signals [10–12], providing an opportunity to test directly whether sensory biases are found in both sexes. In these cases, the sensory bias hypothesis would predict that if females exhibit differential responses to certain mating signals, then males should do the same because of the shared sensory and neural machinery that generated the bias. For instance, in some fiddler crabs of the genus Leptuca, males construct conspicuous vertical mounds of sand next to burrows where courtship takes place, and both males and females are attracted to these structures [13,14]. Interestingly, this preferential attraction to sand mounds occurs even in species that do not actually build such structures, and it is likely that this is a sensory bias that arose in the context of shelter-seeking behaviour to avoid predation [15].

Sensory biases may be responsible for the evolution of complex signals, defined as a signal consisting of multiple distinct components [16], from an ancestral uni-component signal [6,17]. Indeed, in some species where males produce only simple mate attraction signals, females have hidden preferences for complex signals [18,19]. However, preferences for complexity are not universal: only some novel signal elements add to the attractiveness of the base signal [20]. Temporal order, in terms of whether novel elements are placed before or after the existing signal, appears to be an important factor in determining the attractiveness of novel complex signals, and in many species that already produce complex signals there are strong preferences for the ordering of signal elements [21–23]. Even in species that do not naturally produce complex signals, receivers are often biased towards a specific temporal ordering of signal elements: in some cases, they are more attracted to complex signals containing novel acoustic elements placed before (i.e. leading) the species-specific signal, while in other cases, they are more attracted to signals with novel elements placed at the end of (i.e. following) the species-specific signal [24–28].

In the grey treefrog, Hyla chrysoscelis, males produce a simple trilled advertisement call that serves to both attract females and interact with competing males [29]. Although males do not naturally produce complex calls, female H. chrysoscelis show sensory biases, preferring some novel complex calls in which an acoustic appendage is combined with the advertisement call over the advertisement call itself [24,27]. This pattern of response has no obvious ecological correlate and instead is likely a hidden preference arising as an incidental consequence of the mechanism of auditory processing. Females' responses to complex calls depend on the temporal order of signal elements, although one study reported that females preferred complex calls with following appendages and were biased against complex calls with leading appendages [24], while another reported that complex calls with leading appendages were the most attractive [27]. These contrasting results may be due to population differences or to differences in the specific appendage used, but in either case reveal that females of this species have sensory biases for complex stimuli. Males also respond to advertisement calls by adjusting characteristics of their own calls, and their response depends on specific temporal characteristics of the stimulus [30]. Therefore, we hypothesized that male H. chrysoscelis would show a sensory bias similar to that of females, with their response affected by call complexity, and responding differently to novel complex calls with a leading appendage compared with those with a following appendage. We used acoustic playbacks to measure the response of males and females to the same set of simple and complex advertisement calls.

2. Methods

All male playbacks were performed in the field at a pond on the McPherson Preserve, Payne County, Oklahoma, USA from 23 May to 2 July 2020, and 5–7 July 2021. Each night (21.00–24,00), we identified calling male H. chrysoscelis for recording, removing any other calling male within 2 m to reduce immediate competition. We measured male body temperature using an infrared thermometer (Fluke 62 Max+), recorded the calls using the procedure described below and then captured the male for marking with a unique visible implant elastomer code for individual identification. Female phonotaxis tests were performed from May to June 2022. Gravid females were captured at the study site and returned to the laboratory, where they were placed in containers on melting ice until testing [31]. All procedures were carried out with the approval of the Institutional Animal Care and Use Committee at Oklahoma State University (protocol number AS-19-4) and with permits from the Oklahoma Department of Wildlife Conservation (letter numbers 2117, 2211).

(a) . Playback stimuli

All stimuli were generated as .wav files (44.1 kHz sampling rate) using a custom script in R v. 3.6.3 software [32] using the Seewave v. 2.1.3 [33] and tuneR v. 1.3.3 [34] packages. Waveform displays and a complete description of the characteristics of all stimuli are given in electronic supplementary material, figure S1. Call pulse rates vary with temperature [35], so we generated stimuli with characteristics that corresponded to one of three different temperatures (15, 20 or 25°C), and chose the stimulus that was the closest match to male body temperature for playback. All stimuli were broadcast at a rate of one call every 6.11 s for the duration of the 3 min stimulus.

We built novel complex calls by combining the standard call, a synthetic stimulus with characteristics of a typical H. chrysoscelis call based on average values from our study population, with an acoustic appendage (electronic supplementary material, figure S1). We used two different types of appendage to test whether sensory biases for complex stimuli generalize across different novel signal elements [36]. The first appendage type was a noise appendage; this consisted of a 200 ms segment of white noise that was filtered to approximate the frequency characteristics of the standard call (electronic supplementary material, figure S1 for details). This appendage type was chosen as a sound that is novel in the sense that no male produces it, but nevertheless is likely to be stimulating to the sensory system because it contains a conspecific frequency spectrum. The second appendage type was a heterospecific appendage; this consisted of an abbreviated synthetic H. versicolor call (four pulses, total duration 165 ms; electronic supplementary material, figure S1). This appendage type was chosen because H. versicolor is a direct descendent of an ancestral lineage of H. chrysoscelis, and thus its calls, and auditory processing, evolved from ancestral H. chrysoscelis [37]. While H. chrysoscelis individuals are overall less responsive to H. versicolor calls [38], given the species’ close evolutionary relationship to H. versicolor, we considered it plausible that there may nevertheless be hidden biases for complex calls including some H. versicolor pulses. Each of these two appendage types was combined with the standard call in one of two temporal orders to make the four complex call stimuli. Appendages in the leading temporal position came 50 ms before the start of the standard call, while appendages in the following temporal position came 60 ms after the end of the standard call. The root mean square (RMS) amplitude of the appendages was modified to match that of the standard call.

We also created two control stimuli: (i) the standard call with no appendage and (ii) a longer version of the standard call (long control), which was a simple call with approximately the same duration as the complex calls, to control for the effects of stimulus duration on response. As additional controls, we used each appendage type on its own to examine how frogs would respond to a novel sound in the absence of the standard call.

(b) . Male playback procedure

Immediately prior to the playback we recorded 20 spontaneous calls from each male to serve as a baseline. We then randomly chose one of the playback stimuli (stimuli were done in sets such that recordings with the heterospecific appendage stimuli took place from 15 June to 2 July 2020 and from 5 to 7 July 2021, and all others took place between 23 May and 20 June 2020) and broadcast it to the frog, recording its responses for the 3 min duration of the stimulus. Each male was tested with only a single playback stimulus. Playbacks were broadcast from a Pignose 7–100 portable speaker and FiiO M3K MP3 player at a sound pressure level (SPL) of 90 dB (RMS, reference pressure: 20 µPa) at 1 m, calibrated at the beginning of the evening with a sound pressure level meter (Extech 407703A).

(c) . Acoustic analyses

All advertisement calls given by males during each recording period were analysed in Raven Pro 1.6 software (Cornell Lab of Ornithology). We noted the onset and offset of each call, measured call duration and counted the number of pulses per call (hereafter, pulse number) from a waveform display. We calculated the mean values for pulse number for each recording session, along with two derived variables: call rate and duty cycle. Call rate was quantified as the number of calls (excepting the final call of the sequence) divided by the time between the start of the first call and the start of the last call on the recording. Duty cycle was calculated as the total number of pulses produced (again, excepting the final call) divided by the time between the start of the first call and the start of the final call on the recording. The above characteristics were calculated for all advertisement calls given by males. We chose these three characteristics because they are important in determining the attractiveness of the call, and because males generally respond to conspecific calls by altering the pulse number and call rate of their own calls [30,39]. The other main response of males to conspecific calls is to switch from advertisement calling to aggressive calling [40]. Indeed, in this study, some frogs also gave aggressive calls in response to the playback. We did not analyse specific characteristics of these calls because not all males produced them, but we did note the number of aggressive calls and used this to calculate the proportion of a male's calls during the playback that were aggressive calls.

(d) . Female preference testing

We used two-choice phonotaxis tests to measure female preferences for novel complex signals. The calls used to create the phonotaxis stimuli were identical to those used to test male responses. Females were tested at 25 ± 1°C, so we used the call stimuli corresponding to 25°. Testing took place in a darkened semi-anechoic chamber containing a circular arena 2 m in diameter, surrounded by dark cloth. The female was placed under an acoustically transparent cage in the centre of the arena, the chosen stimulus broadcast for 30 s, and then the cage was removed by pulling a rope attached to it. The female's movements were monitored remotely via an infrared camera. Females were given 5 min to make a choice, defined as approaching within 10 cm of one of the playback speakers, which were separated from one another by 90°. Playbacks were performed at 90 dB SPL (fast, C-weighted) at 1 m, calibrated using a Larson Davis 831c sound-level meter. The stereo stimulus was played from a PC and broadcast through Orb Mod1 speakers via an external sound card (Motu 16A), an attenuator (Tucker-Davis PA5) and an amplifier (Crown XLS 2002).

Each female was tested with up to 14 stimuli, in a random order. For each of the two appendage types (heterospecific, noise), we tested the following combinations: (i) the standard call versus either the leading or following complex call, (ii) the long call versus either the leading or following complex call, (iii) the leading complex call versus the following complex call, and (iv) the appendage on its own versus either the leading or following complex call. The speaker that each track of these stereo stimuli was broadcast from was randomly selected. We tested for significant preferences for each stimulus pair using a binomial test against the null hypothesis of equal choices to each speaker if the female were to choose at random.

(e) . Statistical analyses

No male was recorded more than once per appendage type (noise or heterospecific). Note that the same set of recordings of male responses to the standard call and long control were used for comparisons to responses to both the noise and heterospecific appendage stimuli. For the analyses of responses to the complex call stimuli with a heterospecific appendage, three males were recorded on separate nights responding to a complex call playback and to the long control playback; in these cases, we excluded the response to the long control playback so that our samples remained independent.

For each advertisement-call characteristic (pulse number, call rate and duty cycle) and each appendage type (noise, heterospecific), we calculated a separate linear model to examine whether playback stimulus characteristics affected male calling. Each model included the call characteristic as the dependent variable, with the stimulus (short control, long control, leading complex, following complex and appendage only), male body temperature and the value of that call characteristic from the male's spontaneous calling during the baseline recording (to control for differences in initial motivation or calling tendencies of males in the sample) as independent variables. Only recordings with at least five advertisement calls were included in these analyses. We created custom contrasts using the emmeans function [41] and a Sidak correction to test the following specific hypotheses for the effects of the playback stimulus on call characteristics: (i) Calling is affected by the complexity of the playback stimulus. We therefore compared call characteristics given in response to each appendage stimulus versus the standard call. (ii) Complexity affects calling over and above its effects on the duration of the stimulus. We therefore compared call characteristics given in response to each appendage stimulus versus the long control stimulus. (iii) The temporal order of the appendage affects calling. We therefore compared call characteristics given in response to leading versus following appendage stimuli. (iv) Finally, we compared the response to the complex calls and the controls versus the appendage itself.

For the analyses of the effects of playback stimulus on aggressive calling, we used generalized linear models with the proportion of aggressive calls as the (binomial) dependent variable, temperature and playback stimulus as independent variables, and contrasts as above. All statistical analyses were performed in R v.4.1.0 [32]. Raw data associated with this manuscript are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.tx95x6b1t.

3. Results

(a) . Male response to calls with a noise appendage

For complex calls with a noise appendage, the appendage's temporal position did not affect advertisement-call characteristics. Males had significantly lower call rates, but no difference in pulse number or duty cycle, in response to either complex call compared with the standard call (table 1, figure 1). However, there was no difference between advertisement calls given to the complex stimuli and either the long control or the noise appendage itself (table 1, figure 1). The proportion of aggressive calls was also not affected by the temporal position of the appendage in complex calls (figure 1d). However, males gave a significantly greater proportion of aggressive calls in response to either complex call than to any of the control stimuli (table 1). The appendage on its own also elicited more aggressive calls than did either the standard call or long control (table 1).

Table 1.

Test statistics for contrasts of response to the different playback stimuli. Each column shows for a given appendage type and call characteristic the test statistics from a contrast analysis on the estimated marginal means from linear or general linear models with the call characteristic as the dependent variable and playback stimulus, temperature, and, for the three advertisement-call characteristics, the baseline call characteristic, as independent variables. Actual values of the estimated marginal means are shown in figures 1 and 2. Full model results are in the electronic supplementary material, tables S1 and S2. Statistics correspond to t-values for pulse number, call rate and duty cycle (linear model), and z-values for proportion aggressive (binomial proportion variable, general linear model). Asterisks indicate statistical significance after a Sidak correction (corrected for the nine different comparisons within each appendage/call characteristic combination). *p < 0.05, **p < 0.01, ***p < 0.001.

	noise appendage			proportion aggressive	heterospecific appendage			proportion aggressive
	pulse number	call rate	duty cycle		pulse number	call rate	duty cycle
leading−following	−0.92	−0.13	−0.10	−1.94	0.79	−2.42	−1.7	6.88***
leading−long	−0.20	−1.77	−1.71	14.23***	−0.01	−2.61	−2.85*	12.64***
leading−standard	0.94	−3.17*	−2.63	11.56***	0.91	−3.60**	−3.73**	10.69***
leading−appendage	2.05	−1.98	−0.65	4.2***	3.92**	−4.51***	−3.25*	12.94***
following−long	0.74	−1.57	−1.56	15.8***	−0.73	−0.15	−1.06	6.89***
following−standard	1.86	−3.02*	−2.53	13.34***	0.09	−1.09	−1.87	4.04***
following−appendage	2.75	−1.72	−0.51	6.19***	2.87*	−1.85	−1.37	5.94***
standard−appendage	1.19	1.13	1.97	−7.71***	3.15*	−0.78	0.60	1.81
long−appendage	2.34	−0.27	1.03	−10.90***	4.04*	−1.86	−0.30	−2.10

Figure 1.

Call characteristics for males in response to the different playback stimuli with the noise appendage. The bottom row shows the waveform display of each stimulus; lead: complex call with leading appendage, N = 19; follow: complex call with following appendage, N = 19; standard: the standard simple advertisement call, N = 21; long: the long control advertisement call, N = 21; append.: the appendage on its own, N = 22. Each point represents the raw value of the average across the recording for each male's advertisement-call characteristics: (a) pulse number, (b) call rate, (c) duty cycle, and (d) proportion of calls during the recording that were aggressive. Points have been slightly jittered along the x-axis to reduce overlap. Black square and error bars represent the estimated marginal mean value ± 95% confidence interval.

(b) . Male response to calls with a heterospecific appendage

As in the noise appendage, for complex calls with a heterospecific call segment as an appendage, the appendage's temporal position did not affect advertisement-call characteristics (table 1, figure 2). Males had a lower call rate and lower duty cycle in response to the leading appendage stimulus compared with the standard call, but there were no such differences for males with the following stimulus, nor was there a difference between these stimuli for pulse number (table 1). Duty cycle was lower in response to leading appendages compared with the long control, but otherwise there were no differences between the advertisement-call responses to the complex calls and this control (table 1). Males had higher pulse numbers in response to either complex stimulus, and males had lower call rates and duty cycles in response to the leading, but not the following, appendage stimulus, than to the heterospecific appendage on its own (table 1, figure 2). For complex calls with a heterospecific appendage, the proportion of aggressive calls was greater in response to the leading appendage stimulus than to the following appendage stimulus (figure 2d). Either complex call stimulus elicited more aggressive calls than any of the control stimuli (table 1).

Figure 2.

Call characteristics for males in response to the different playback stimuli with the heterospecific appendage. Labels and points as in figure 1. lead, N = 18; follow, N = 16; standard, N = 21; long, N = 18; append., N = 20.

(c) . Female preferences

Females significantly preferred the call with a following noise appendage to the standard call and had a similar but non-significant preference for the leading noise appendage stimulus (figure 3a). However, there was no preference for complex calls with the heterospecific appendage in either position compared with the standard call (figure 3b). For both appendage types, females preferred the long control stimulus to the complex calls (figure 3). When given the choice between complex calls with the appendage in either the leading or following position, females showed no preference in the case of a heterospecific appendage, but had a significant preference for the leading noise appendage stimulus over the following noise appendage stimulus (figure 3). Females always significantly preferred the complex call stimulus to the appendage on its own (electronic supplementary material, table S3).

Figure 3.

Female preferences. (a) Responses in two-choice tests involving complex calls with a noise appendage. Each row represents a single test, with waveforms of the two stimulus choices. The number of females choosing each stimulus is indicated by the ratio in the centre (with the number on the left corresponding to choices for the stimulus illustrated on the left; likewise for the number on the right), below which is the p-value from a two-tailed binomial test against the null hypothesis of equal choices to each stimulus. (b) Responses in two-choice tests involving complex calls with a heterospecific appendage. All waveforms depict the actual stimuli used and illustrate one second of audio. Full statistical results are given in electronic supplementary material, table S3, and an alternative graphical depiction of the results including confidence intervals is given in the electronic supplementary material, figure S2.

4. Discussion

The sensory bias hypothesis for the evolution of female mate choice proposes that the path of signal evolution is determined by female preferences that already existed in their sensory systems prior to the evolution of the signalling trait, and that these biases towards certain stimuli arose in contexts other than selection on mating preferences [1]. This is an alternative to models of female preference evolution in which preferences coevolve with male signal traits because of the benefits of mating with certain males [42]. The role of sensory biases in sexual selection is often examined at the interspecific level, by using phylogenetic analyses to test the prediction that the female preference evolved prior to the preferred male trait [2,43,44]. However, the sensory bias hypothesis can also be tested at the intraspecific level, because it predicts that males and females should show similar response patterns to novel signals, although this possibility has received little attention [15]. In at least one case, in darters of the genus Etheostoma, male preferences seem to have evolved before female preferences [45], which further emphasizes the importance of examining sensory biases in males. We found mixed evidence in favour of our prediction that males and females would show similar biases in response to novel signals depending on signal complexity and the temporal order of signal components. Although both sexes showed evidence for a sensory bias, as described below there were sex differences in the details of how this bias was expressed.

Males gave more aggressive calls in response to any of the novel complex stimuli than they did to conspecific advertisement calls, consistent with the sensory bias hypothesis. Stimulus complexity elicited responses above and beyond the effects of the necessarily increased duration of complex signals, as shown by comparisons of the response to complex calls and the response to the long control, a simple call with the same duration as the complex calls. Males also gave more aggressive calls to complex stimuli than they did to the novel signal element on its own. Thus, it was the combination of signal elements—the appendage and the species-specific advertisement call—that led to the increase in aggressive calling. However, there was variation among the complex stimuli in their effects on male aggressive calling. For the complex calls with a heterospecific appendage, there was an effect of temporal order, with the leading appendage stimulus eliciting more aggressive calls than the following appendage stimulus. This temporal order effect was not seen for the noise appendage stimulus, indicating that characteristics of the appendage itself also play a role in response to complex stimuli. Indeed, the noise appendage alone elicited a large amount of aggressive calling, although there was an even stronger aggressive response to complex calls with a noise appendage. There was limited evidence for effects of call complexity on male advertisement calling, and most effects were likely byproducts of the increase in aggressive calling (e.g. lower advertisement call rate; across all stimuli, advertisement call rate was negatively correlated with the proportion of aggressive calls, r = −0.43).

Previous studies of H. chrysoscelis described sensory biases in female response to complex calls [24,26,27], and in our study, we also found some evidence for a sensory bias in female preferences towards the same stimuli we presented to males. There were both similarities and differences between the sexes in their response to novel complex calls. First, the overall influence of novel appendages was weaker for females than it was for males. Although females preferred complex calls with the noise appendage to the standard call, this may have been because the complex call had a longer duration. They strongly preferred a long control with no appendages over a complex call of the same duration, as was also the case in a previous study [27]. The heterospecific appendage was apparently neutral because females did not prefer a complex call with this appendage type over the standard call. The neutrality of this stimulus is somewhat surprising because of the inclusion of a heterospecific call element, to which females are normally much less attracted compared with a conspecific call [29]. By contrast, at least in terms of aggressive calling, males differed strongly in their response to all of the complex calls compared with non-complex calls. Second, there was some evidence for a temporal order effect in females, as there was in males. In the direct comparison between the leading and following complex stimuli, females showed a strong preference for a call with a leading noise appendage versus a call with a following noise appendage. However, temporal order did not seem to have much of an influence on female preferences for complex stimuli over non-complex alternatives. In addition, males only showed a temporal order effect towards complex calls with a heterospecific appendage, while females only showed a temporal order effect towards complex calls with a noise appendage. Interestingly, when there was an effect, it was the leading stimulus that elicited the stronger response in both males and females. The female preference for leading appendages is consistent with one study of a population of H. chrysoscelis from the same genetic lineage as our study population [27], although a study performed on a different genetic lineage of this species found that calls with following appendages were more attractive than calls with leading appendages [24].

One challenge for the general application of our approach is that in many species, males may simply not be responsive to conspecific mating signals and, even when they are, their response behaviour and/or the function of the response differs from that of females. Male H. chrysoscelis, like conspecific females and males of many anuran species [46], are responsive to advertisement calls, but whereas females respond by phonotaxis, males respond by adjusting call characteristics or switching to aggressive calls. In another frog, Engystomops pustulosus, when males and females were tested with the same behavioural assay, phonotaxis, they showed similar responses to call variation. However, results from a different behavioural assay of males, calling in response to playbacks, did not correspond to patterns of female phonotactic response [47,48]. By contrast, in H. chrysoscelis, we found some common effects of signal complexity and the temporal order of signal components on male and female responses, which suggests some commonalities in the underlying stimulus processing mechanisms, perhaps because of shared biases.

It is important to keep in mind that H. chrysoscelis males do not naturally produce such complex calls, and we are not necessarily proposing that sensory biases are responsible for female preferences for extant characteristics such as longer duration calls [39]. Nevertheless, our finding of hidden biases gives insights into what would happen should complex calls evolve, in particular suggesting that a complex call with certain kinds of leading appendage would be favoured in terms of direct attractiveness to females, but perhaps also costly because it would elicit more aggressive calls from other competitors, which can indirectly reduce a given male's attractiveness [49]. Other types of complex calls are apparently more neutral in their effects on attractiveness to females, but all of the complex stimuli we tested elicited more aggressive calls from competing males. Our findings are thus an example of the general phenomenon of trade-offs in the production of complex signals that arise because of multiple types of receivers, some of whose responses are beneficial, and some detrimental, for the signaller [50]. Such studies have focused on trade-offs between attracting conspecifics versus heterospecific predators [51,52], but the trade-off may also operate at the intraspecific level if both males and females have sensory biases in response to signals.

Temporal order is increasingly recognized as a critical component of signal design, and complex signal evolution is likely shaped by the effects of novel signal elements on temporal processing in the sensory system [53–55]. On the one hand, in many cases, novel elements placed at the end of an existing signal are more effective at eliciting a response than those at the beginning of the signal [24,25]. Such findings may be explained by neural processing similar to a drift–diffusion model, in which elements at the beginning of a signal are weighted more heavily than those at the end, and negative weighting on novel components is stronger than the positive weighting of typical conspecific signal elements [56–58]. On the other hand, in some cases, leading novel signal elements generate neural accommodation that improves response to subsequent species-typical signal components [59] or function as alerting elements [60].

Importantly, most previous demonstrations of common biases across the sexes come from examples in which the bias was generated by ecological selection on specific sensory capabilities [9,13–15]. By contrast, the differential responses to complex acoustic signals that we observed in male and female treefrogs were more likely hidden biases that arose as incidental consequences of the neural mechanisms of acoustic signal processing. Cross-sex comparisons of neural processing of complex signals will be an important next step for determining whether there is a common mechanistic basis for the responses of males and females to novel complex signals, further supporting the sensory bias hypothesis of signal evolution. The landscape of female preferences for complex calls appears to be highly variable in H. chrysoscelis, as different studies using somewhat different characteristics for call and appendage have come to different conclusions in terms of the nature of sensory biases, and particularly the effects of temporal order of call components [24,27]. An important approach for future studies of sensory bias in this and other species is therefore to use a function-valued approach to measure the response of both sexes to systematic variation in appendage characteristics and temporal position [25].

Data accessibility

Raw data associated with this manuscript are available in the Dryad Digital Repository (https://doi.org/10.5061/dryad.tx95x6b1t) [61] and in the electronic supplementary material [62].

Authors' Contributions

M.S.R.: conceptualization, formal analysis, investigation, methodology, project administration, and writing—original draft; I.d.l.H.: data curation, investigation, methodology, and writing—review and editing.

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

Conflict of interest declaration

The authors declare no competing interests.

Funding

We received no funding for this study.