Acoustic divergence in advertisement calls among three sympatric Microhyla species from East China

Introduction

Vocalizations, whose functions are mainly involved in reproduction, aggressive behavior and defense, is one of the important mode of communication in anuran amphibians (Gerhardt & Huber, 2002; Kelley, 2004; Wells, 2007; Toledo et al., 2015b). Advertisement calls are a form of male anuran vocalization that is most commonly heard and of the highest value in taxonomy and they play a crucial role in attracting potential mates and conveying territorial information to conspecifics. Several researchers have focused on advertisement calls of anurans (Wells, 2007; Gambale, Signorelli & Bastos, 2014; Köhler et al., 2017; Poyarkov et al., 2019; Wei et al., 2019). The variation in advertisement calls in anurans is generally studied at the intraspecific and interspecific levels (Köhler et al., 2017). Call variation within and between individuals of many anurans is influenced by individual motivation of the vocalizing male owing to numerous internal and/or external factors (Wells, 2007; Köhler et al., 2017). For example, social context could change the frequency of calls (Bee, Perrill & Owen, 2000; Reichert & Gerhardt, 2013) and temperature effects are linked to pulse rate and call duration (CD) (Lingnau & Bastos, 2007; Pröhl et al., 2007; Gasser, Amézquita & Hödl, 2009; Bee, Suyesh & Biju, 2013; Ziegler, Arim & Bozinovic, 2015) at the individual level. Furthermore, body size could constrain the frequency of calls (Gerhardt & Huber, 2002; Gingras et al., 2013) and physical or physiological damages could generate differences in calls (Hoffmann & Kloas, 2012; Pröhl et al., 2013) among individuals. Interspecific variation in calls is generally considered for species recognition in anurans (Gerhardt & Huber, 2002) and it is higher than intraspecific variation (Köhler et al., 2017). Therefore, Köhler et al. (2017) suggested that the assessment of variation in advertisement calls in anurans is important for understanding speciation and signal evolution.

Several studies on advertisement calls in anurans have focused on the variation pattern in spectral and temporal parameters (Gerhardt, 1991; Bee, Reichert & Tumulty, 2016). The static and dynamic properties of call parameters are proposed to be relevant to species recognition and mate choice, respectively (Gerhardt, 1991). In most anuran species, spectral and fine temporal call parameters are relatively stable (more static) at the within-individual level, whereas gross temporal parameters represented higher variability (more dynamic) (Gerhardt, 1991; Reinhold, 2009; Köhler et al., 2017). When a parameter has a relatively low variation, it is hypothesized to be caused by stabilizing selection of female preferences (Gerhardt, 1991; Castellano & Rosso, 2006). However, the static and dynamic traits of call parameters are not strictly inalterable, that is, these traits of a specific call parameter may vary among anuran species (e.g., dominant frequency (DF), Gambale, Signorelli & Bastos, 2014; Wei et al., 2019).

Microhyla (Anura: Microhylidae) is a genus with 50 species that are distributed from the Ryukyu Island and South China through India to Sri Lanka and through Southeast Asia to Sumatra, Borneo, Java and Bali (Frost, 2020). Nine species of Microhyla are currently known from Southeast China (AmphibiaChina, 2020). To date, there are reports on the advertisement calls of 28 Microhyla species (Table S1). Several studies have reported the taxonomy and identification of Microhyla species in some distribution areas (Sri Lanka, Myanmar and South Asia) through bio-acoustic comparison (Wijayathilaka, Meegaskumbura & Gianni, 2016; Garg et al., 2018; Poyarkov et al., 2019). In East China, M. beilunensis Zhang et al. (2018), M. fissipes Boulenger (1884) and M. heymonsi Vogt (1911) are three common Microhyla species and their distribution ranges are overlapping (Fig. 1). They breed from April to June in shallow, ephemeral pools and paddy flied (AmphibiaChina, 2020; Fei, Ye & Jiang, 2012; Z.-Q. Chen et al., 2019, unpublished data). In this study, we quantified call parameters in these three species. The results will be of importance for systematics, rapid identification of species, behavioral experiments and species protection. Particularly, they will be useful to identify species quickly and accurately in field surveys. We quantified the spectral and temporal parameters of advertisement calls in the three sympatric Microhyla species in the Jiulongshan National Nature Reserve, Lishui City, Zhejiang Province, East China. We also explored the relationships among snout-vent length (SVL), CD, number of pulses (NP) and DF using previously published data, while accounting for phylogenetic relationships.

Materials and Methods

Results

The advertisement calls of all three species had approximately the same spectrogram shape: first a pulse of low amplitude, which then increases, then the main body of the call, then a final pulse, during which the amplitude decreases again. It is a spindle-shaped trend in which the amplitude gradually increased, and then gradually decreased (Fig. 2). However, there were species-specific differences in call composition (Fig. 2). All species exhibited significant differences in the frequency of the NP. The call of M. beilunensis was composed of 2–7 separate pulses (Figs. 2B and 2C); a six-pulse call occurred most frequently, comprising 37.70% of the total calls. The call of M. fissipes had 8–21 separate pulses (Figs. 2F and 2G); the 16-pulse call occurred most frequently, comprising 33.75% of thetotal calls. The call of M. heymonsi was composed of 3–11 separate pulses (Figs. 2K and 2L); the six-pulse call occurred most frequently, comprising 43.40% of the total calls. Males of all three species have an external single subgular vocal sac and their throat color differs among the three species (Figs. 2E, 2J and 2O). M. beilunensis has small black markings around the throat (Fig. 2E). The throat of M. fissipes has large black markings on all sides and small black markings in the center (Fig. 2J) and that of M. heymonsi has small brown markings (Fig. 2O).

The mean SVL differed among the three species (F_{2, 42} = 11.51, P < 0.001) and was the largest in M. fissipes (23.2 ± 1.2 mm; ranged, 21.6–25.8 mm; N = 13) and the smallest in M. heymonsi (21.5 ± 1.2 mm; ranged,19.6–23.6 mm; N = 15), with M. beilunensis (22.5 ± 0.7 mm; ranged, 21.2–23.5 mm; N = 17) in between. The correlation analysis between SVL and the 10 call parameters showed that SVL negatively correlated with the DF in M. fissipes (ρ = −0.698) and M. beilunensis (ρ = −0.652), with the F1 in M. heymonsi (ρ = −0.558) and M. beilunensis (ρ = −0.494) and with the F3 in all three species (M. beilunensis: ρ = −0.553, M. fissipes: ρ = −0.733, M. heymonsi: ρ = −0.526).

All the 10 call parameters differed significantly among the three species (Table 1). The CD, DF, F2 and F3 were higher in M. fissipes and M. heymonsi than in M. beilunensis; the CI and CIT were the highest in M. heymonsi and the lowest in M. beilunensis; the NP was the highest in M. fissipes and the lowest in M. beilunensis; the PR was higher in M. fissipes than in M. heymonsi and M. beilunensis; and the F1 and F4 were lower in M. fissipes than in the other species (Table 1).

In all the species, four temporal parameters (CD, CI, NP and PR) showed larger CV_W and CV_A, whereas six spectral parameters presented smaller coefficients (Table 2). The mean CV_A was larger than the mean CV_W for all the parameters (the ratio of CV_A/CV_W > 1.0), and these differences were statistically significant (Table 2).

The PCA analysis suggested that two components (eigenvalue ≥ 1) from the original ten acoustic parameters accounted for 78.39% of the variation (Table 3). The first component (explained 50.60% of variance) had high positive loading scores for NP, PR, F2 and F3 and a high negative loading score for F1; the second axis (explained 27.79% of variance) mainly represented CI and CIT (Table 3). The resulting scores differed significantly among the three species on both axes (first axis: F_{2, 42} = 323.24, P < 0.0001; MH > MF > MB; second axis: F_{2, 42} = 131.73, P < 0.0001; MH > MB > MF) (Fig. 3A).

The DFA analysis presented results similar to those of the PCA and generated two discriminant functions (Table 3). The first canonical variable (explained 71.15% of variance) mostly represented NP and F3; the second canonical variable (explained 28.85% of variance) was largely represented by the CI, NP and PR (Table 3). The two functions differed significantly among the three species (Wilks’ λ = 0.001, P < 0.0001); even after excluding the first function, the second function was still different for each species (Wilks’ λ = 0.057, P < 0.0001). The centroids were −7.55, −1.52 for M. beilunensis; 7.11, −4.18 for M. fissipes; and 2.40, 5.34 for M. heymonsi (Fig. 3B).

By incorporating data from 29 Microhyla species, we found that the mean male SVL in the genus ranged from 14.9 mm to 33.6 mm; mean CD ranged from 0.062 s to 1.806 s; mean NP ranged from 4 to 97; and mean DF ranged from 1,650 Hz to 5,029 Hz (Table S1). The PGLS analysis showed that phylogenetic relationships affect CD vs. SVL, NP vs. SVL, DF vs. CD, NP vs. CD and NP vs. DF (all λ > 0.77), but not DF vs. SVL (λ = 0) (Table 4). DF was negatively related to SVL (Fig. 4), but the relationship between CD and SVL, NP and SVL, DF and CD, and NP and DF were not significant in the OLS and PGLS models (Table 4). NP was positively related to CD in the OLS model, but not in the PGLS model (Table 4). Based on likelihood ratio tests, the PGLS model was a better fit to the data than the OLS model for NP vs. SVL, NP vs. CD and NP vs. DF (Table 4).

Discussion

To the best of our knowledge, the present study is the first to report the advertisement call of M. beilunensis consisting of 2–7 pulses per call, which has relatively few NPs among the 29 studied Microhyla species. Similar to that in other Microhyla species reported previously (Wijayathilaka, Meegaskumbura & Gianni, 2016; Sun, 2017; Nguyen et al., 2019; Poyarkov et al., 2019; Garg et al., 2018), oscillograms and spectrogram of advertisement calls in M. beilunensis were conformed to the pulse-repetition sound based on Beeman’s (1998) sound categories.

The patterns of variation in call parameters of anurans are correlated with female preferences and different parameters contain different kinds of biologically significant information (Gerhardt, 1991). Our results showed that the examined spectral parameters (e.g., DF, F1–F4 and CIT) were not or weakly variable as their CV_W is smaller than 5% and a temporal parameter (CI) was variable as its CV_W was larger than 10% in the three sympatric Microhyla species, according to the classification of Gerhardt (1991). These results were similar to those of most of reported 48 anurans, of which the DF of 69% species was classified as a static property and CI of 40% species was classified as a dynamic property (Köhler et al., 2017). Static properties with lower CV_W are more important in species recognition, caused by stabilizing selection, because females usually prefer the values of individual calls at or near the mean of the population. Whereas, dynamic properties with higher CV_W are generally supposed to be beneficial for mate choice and are driven by directional selection, because females tend to prefer extreme values of male calls (Gerhardt, 1991, 1994; Castellano & Giacoma, 1998; Wollerman, 1998; Friedl, 2006; Reinhold, 2011; Köhler et al., 2017). Therefore, our results indicated that all spectral parameters contributed to species recognition and the CI might encode information on mate quality in the three sympatric Microhyla species.

Generally, call parameters with stable within-individual variation have a low CV_A and within-individual dynamic call parameters are also more variable on the CV_A (Köhler et al., 2017). However, this phenomenon does not necessarily mean that among-individual variations are lower in static than in dynamic properties (Köhler et al., 2017). In our study, the CVs of spectral parameters were higher among individuals than within individual in the three Microhyla species. Especially, the values of CV_A were more than 1.8 times of those of CV_W for DF of the three Microhyla species. These variations among individuals were supposed to serve sexual selection and play a role in male–male competitions (Gerhardt, 1991; Howard & Young, 1998; Friedl & Klump, 2002; Köhler et al., 2017). Furthermore, body size is an important determinant of variations in call parameters among individuals (McClelland, Wilczynski & Ryan, 1996; Wang et al., 2012; Rodríguez et al., 2015). Our results showed that spectral parameters (e.g., DF, F1 and F3) are under the physical body size constraint, with smaller individuals producing calls at higher frequencies; in contrast, temporal parameters were not influenced by body size in the three Microhyla species. Similar results were also found in other Microhyla species (Wijayathilaka, Meegaskumbura & Gianni, 2016) indicating that spectral parameters of calls might encode the information of male’s body size in each Microhyla species. Therefore, researchers suggested that the spectral parameters of anurans’ advertisement calls, as an honest signal, transfer information about the body size of vocalizing individuals and thus, possibly information about its strength and quality to females and competitors (Davies & Halliday, 1978; Wells, 2007). In addition, although spectral parameters are relatively stable, they also have certain plasticity under the influence of social context (Lopez et al., 1988; Wagner, 1992; Bee & Perrill, 1996; Bee, Perrill & Owen, 2000). For example, Lithobates clamitans could lower the DF of their calls in response to broadcasts of conspecific calls (Bee & Perrill, 1996; Bee, Perrill & Owen, 2000). However, this phenomenon was not examined in our three Microhyla species. Therefore, further experiments are required to ascertain whether Microhyla species alter spectral parameters of advertisement calls in accordance with different social contexts.

Interspecific variation of advertisement calls in anurans serve in species recognition (Gerhardt & Huber, 2002; Köhler et al., 2017). Sympatric anuran sister species share the same acoustic space and tend to exhibit similar acoustic signals and behavioral traits (Toledo et al., 2015a). It is not clear how to minimize interference in the acoustic niche including spatial, temporal and structure dimensions (Rabin et al., 2003; Sinsch et al., 2012). The three Microhyla species studied used similar microhabitat and vocalizing time during the breeding season (Fei, Ye & Jiang, 2012; Z.-Q. Chen et al., 2019, unpublished data), but differences in acoustic structure were significant among the three sympatric Microhyla species. Besides, it cannot be ignored that phylogenetic history is perhaps the main restriction on the evolution of acoustic characteristics in anurans (Mclean, Bishop & Nakagawa, 2013). Many studies have shown that call differences between anuran species can largely be accounted by phylogeny (Bosch & De la Riva, 2004; Cocroft & Ryan, 1995; Mclean, Bishop & Nakagawa, 2013). Our PGLS results showed that the CD vs. SVL, NP vs. SVL, DF vs. CD, NP vs. CD and NP vs. DF relationships were significantly influenced by Microhyla phylogeny. However, the relationship between DF and SVL was not influenced by phylogeny, possibly owing to inter-sexual selection, which could weaken or even modify phylogenetic effects on the evolution of acoustic characteristics (Mclean, Bishop & Nakagawa, 2013). The level of call modification may also be affected by the properties of a specific signal. For example, behavioral traits are often more highly modified than physiological features (Blomberg, Garland & Ives, 2003; Mclean, Bishop & Nakagawa, 2013). Several Microhyla species are sympatric (Lee et al., 2016; Poyarkov et al., 2019; Frost, 2020; Z.-Q. Chen et al., 2019, unpublished data). To prevent inter-specific hybridization events, sympatric species must be able to distinguish conspecifics, via behavioral and physical features. Gerhardt & Huber (2002) found that inter-sexual selection is the main driving force behind the evolution of acoustic communication in anurans. Our results provide considerable evidence that the three sympatric Microhyla species have acoustic divergence in call structure, facilitating inter-sexual selection, which may therefore help prevent hybridization.

Conclusion

Based on the results, we conclude the following: (1) CVs of all spectral parameters were smaller than 5%, whereas the CV of CI was larger than 10% within individual of the three sympatric Microhyla species. (2) Body size was a key factor that leads to among-individual variation in advertisement calls of Microhyla species. (3) Acoustic divergence in call structure existed in the three sympatric Microhyla species. (4) The PGLS analysis showed that phylogeny affected the NP, DF and SVL vs. CD and SVL and DF vs. NP but not DF vs. SVL.

Supplemental Information