This study is one of the first comparisons of acoustic characteristics in a sizeable number of closely related species of fishes [see also ]. The most important insight was found at the level of the relationship between fish size and both dominant frequency and pulse duration. These kinds of relationships were already well known in fishes and have been found in numerous species from different taxa [26, 27, 32–36]. In the present study and for the first time, it appears that these relationships are observed at a higher taxonomic level in the case of clownfishes (i.e. spread over the entire tribe Amphiprionini; see ) since dominant frequency and pulse duration are strongly predicted by body size among the 14 different species. It clearly shows all clownfish species use the same mechanism of vocalization , which has remained largely conserved throughout evolution. Moreover, different species having different body shape and different types of teeth (A. akallopisos, A. frenatus and A. ocellaris) but the same size produce sounds that display the same power spectrum and the same oscillogram, demonstrating these morphological variations do not make significant differences in the sounds produced by the jaw teeth snapping. The size influence highlights that all the fourteen species might have a major overlap at the level of pulse length and dominant frequency. Surprisingly, using the pulse duration/body size and dominant frequency/body size ratios removes the effect of body size at the level of pulse duration (Figure 2A), but not at the level of dominant frequency (Figure 2B). Regarding the dominant frequency, the ratio continues to be smaller in larger species, but highly reduces the overlap between species. For example, some Amphiprion clarkii, measuring between 55 and 110 mm in SL, produced a frequency range that was between 700 and 400 Hz, overlapping also the frequency of smaller A. ocellaris (625-900 Hz; see also ). However, the ratio of dominant frequency/body size was 1.40 in A. clarkii and 1.62 in A. ocellaris, clearly distinguishing the species. Therefore, a factor other than body length must be important in determining the absolute frequency values. The volume of the swimbladder could be the determining factor. Moreover, the data of this last factor are also correlated with fish size. The question arises whether the fish is able to determine the relationship (or the discrepancy) between the emitter size and the sound frequency it detects. If so, it could enable the fish to distinguish whether the emitter is conspecific or not. In the future, it would be interesting to compare these size-related relationships between Amphiprion and other pomacentrids. Comparison with Pomacentrus partitus data  seems to indicate these fish might also be placed on the same slope. In this case, we could expect there is an ancestral relationship between dominant frequency and body size, and this relationship is not important in the taxon diversification because it does not change between species. However, more precise data from other pomacentrids are needed.
Variations in sounds are usually considered as pre-zygotic isolating mechanisms leading to speciation [7, 37, 38]. In Amphiprion species, acoustic properties can contribute to the differentiation of species because some of them showed differences in at least one of the acoustic characteristics recorded. However, the significant overlap in acoustic data due in part to the conserved mechanism of sound production supports the acoustic communication cannot be considered as the unique isolating barrier and does not seem to be the main driving force in the evolution of clownfishes. The fact that all species have the same biomechanics implies these fishes do not have many possibilities to develop variations in their calls: they can differ in dominant frequency and pulse duration through their body size, in pulse period and in number of pulses in a call. The alternative hypothesis would be some differences in the calls are a by-product of the size variation. Body size as a trait of natural selection has already been demonstrated in the speciation event of some stickleback species in which this difference is thought to be an adaptation to alternative foraging habitats . In damselfishes, evolutionary change in body size (i.e. gigantism or nanism) could be assumed as a relatively common phenomenon [40, 41] and could therefore be involved in the isolation of some Amphiprion species. From the proximal cause point of view, the size appears really important in maintaining the hierarchy existing within groups. In Amphiprion akallopisos, all aggressive interactions (biting, chasing, frontal and lateral display, body jerking) appear to be preferentially directed towards individuals adjacent in ranks . In Amphiprion percula, rank was the only factor associated with the probability of mortality; low-rank individuals suffer from a higher mortality rate than high-rank individuals. The most likely explanation for this pattern is competition for rank , preventing smaller fish from having access to reproduction. It means that acoustic communication can be an important factor for mating access.
From the ultimate cause point of view, competition for the limited anemone resource may have resulted in niche partitioning through specialization for different anemone species , most clownfish species remaining in close contact with their hosts and rarely interact with other species on the reef . However, some clownfish species appear to partition the anemone resource with other species by having a refuge in size [42, 43]: small A. sandaracinos or A. leucokranos cohabit with A. chrysopterus in the region of Madang (Papua New Guinea), while small A. perideraion use the same host sea anemone as A. clarkii in the region of Okinawa (Japan). In both cases, the different sizes of cohabiting species imply they possess clear differences in their acoustic repertoire (Table 1), size-related call characteristics such as main frequency and pulse length being a by-product of the evolutionary trait. In the Japanese heterospecific groups, small A. perideraion are not considered as competitors and should receive less aggressive attention from larger congener A. clarkii. Although A. clarkii suppresses the growth and reproduction of A. perideraion , subadult A. perideraion are able to mature in heterospecific groups, and change to female when they are the largest among conspecific members. This suggests that A. perideraion in heterospecific groups prepare for reproduction before the disappearance or emigration of larger A. clarkii. Thus, they adopt a mating strategy that involves waiting for vacated breeding posts because of their low mobility and a low host density .
Due to the relative simplicity of many central and peripheral vocal mechanisms, fish typically lack the ability to produce complex and dynamic, frequency-modulated calls . Vocal differences among fish species are usually due to variations in temporal patterning [27, 31]. Pulse period has been shown to be the most important acoustic feature involved in species recognition in pomacentrids [46, 47]. Divergence in this character seems to be sufficient to drive pre-zygotic isolation , because differences in the calling characteristics are able to prevent the signaller to be considered as a competitor. Myrberg et al.  conducted playback experiments to test the responsiveness of different Stegastes species. Although sounds of each species were able to elicit responses of all the other species, males significantly more responded to sounds of their own species than to sounds from congeners. Interestingly, species that cohabit individual sea anemones (i.e. A. sandaracinos with A. chrysopterus in the region of Madang, or A. perideraion with A. clarkii in the region of Okinawa) present a completely different pulse period. As previously stated, non-overlap in this character may have been important in the taxon diversification. However, pulse period is not systematically significantly different among sympatric species: A. clarkii, A. frenatus and A. ocellaris have the same pulse period range (Figure 2D) while living in sympatry on the fringing reef around Sesoko island . These three species inhabit different host species, being Heteractis crispa for A. clarkii, Entacmaea quadricolor for A. frenatus and Stichodactyla gigantea for A. ocellaris [44, 49], which suggests overlap in pulse period among these species is of minor importance.