Acoustic communication mediates interactions within the specific social contexts to which individuals are subjected during their life cycle (Fitch et al. 2010). Acoustic communication can be defined as a sign-mediated and rule-governed acoustic interaction between at least 2 agents often susceptible to changes in the needs of species during their lifetime (Witzany 2014). The production of vocalizations can be context-mediated and susceptible to change.

Some species adapt their calls following environmental changes: for example, when exposed to anthropogenic noise, the Tokay gecko (Gekko gecko) increases the duration of its call syllables (Brumm and Zollinger 2017) and neotropical frogs of the genus Boana switch the dominant frequencies of their calls (Caorsi et al. 2017) to maintain signal transmission. Likewise, the acoustic traits of begging calls produced by the grey warbler (Gerygone igata) chicks simultaneously indicate their age and hunger level, so that their parents can respond accordingly to their needs (Anderson et al. 2010).

Similarly, species’ vocal repertoire size and complexity can change accordingly to their life stage: Asian elephants (Elephas maximus) and Nile crocodiles (Crocodylus niloticus) present simpler repertoires with fewer call types in early life stages compared to adult and juvenile individuals (de Silva 2010; Vergne et al. 2009, respectively). In some species, the basic elements of adult vocalizations are already present at birth but show a shift in organization and relative abundance with age, as in singing mice (Scotinomys teguina and Scotinomys xerampelinus; Campbell et al. 2014).

Ontogenetic changes in the vocal repertoire may occur due to different cognitive abilities and social needs experienced by individuals at different life stages (Campbell et al. 2014). Such changes are known in representatives of many different clades, but a comprehensive understanding of the distribution and complexity of these changes, and how these changes are coupled with changes in the social or environmental context, is limited by the lack of reports for certain groups, especially those historically understudied. Aside from bringing new insight into the natural history of targeted species, studies of the changes of acoustic repertoires can facilitate a clearer understanding of social behavior and cognitive functions in a broad phylogenetic perspective (Fitch et al. 2010).

Although acoustic communication is widespread among vertebrates (Ladich and Winkler 2017; Jorgewich-Cohen et al. 2022b; Rice et al. 2022), some clades have only recently received attention. Turtles, for instance, occupy a new space in bioacoustics studies (Galeotti et al. 2005; Giles et al. 2009; Ferrara et al. 2012, 2013; Geller and Casper 2019b; Jorgewich-Cohen et al. 2022b). Currently, around 100 species from all families have been reported to emit sounds in different behavioral contexts (Frank 2022; Jorgewich-Cohen et al. 2022b, 2024; Zhou et al. 2022, 2023b), including complex behaviors such as synchronous hatching and possibly parental care (Ferrara et al. 2012).

Despite the fact that the use of sounds for communication among turtles seems to be more important than previously reported—starting from embryos until adults, and associated with specific ecological and social needs in each life stage (Ferrara et al. 2012; Jorgewich-Cohen et al. 2024)—there are currently no extensive descriptions of species’ repertoires that include all life stages, precluding the understanding of their social behavior in its full extent. Exceptionally, only a few turtle species have been reported to produce sounds in different life stages (Fig. 1). Among these species, the South American river turtle (Podocnemis expansa) and the snapping turtle (Chelydra serpentina) some of only ones known to produce sounds throughout their lives, starting from embryos and hatchlings (Ferrara et al. 2012, 2014c; Geller and Casper 2019a; Lacroix et al. 2022) to adult specimens (Dodd and Brodie 1975; Ferrara et al. 2014c; Jorgewich-Cohen et al. 2022b). Although the descriptions of the acoustic repertoires are not exhaustive, in some cases with only an indication that they are able to emit sounds, changes in the acoustic behavior from embryo to hatchling already indicate the possibility that ontogenetic changes may occur in these species.

View Figure

• Download as Powerpoint
• Download Figure

Here we investigate if the acoustic repertoires of turtles undergo ontogenetic changes, and if such changes are similar throughout turtle phylogeny. To answer these questions, we collected data in the form of sound recordings from 6 species in different life stages. We describe the sounds produced by these species and compare the repertoire of specimens in different life stages to discuss the biological significance of our findings in a broad phylogenetic perspective.

METHODS

RESULTS

In total, we analyzed approximately 115 hrs of sound recordings. The number of sound types produced by each species in each life stage greatly differed (Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Acoustic repertoires were described accordingly to results from human-based analyses and were later compared to results from machine-based analyses. Audio files containing each sound type together with detailed acoustic characteristics can be found in Supplemental Material 5 and 6, respectively.

Ontogenetic Changes in Repertoire Size. —

The proxies for repertoires size based on MCP areas were calculated for each life stage for all species (Table 1). MCP plots can be visualized alongside the metrics for mean sound duration, mean peak frequency, and number of sounds per hour produced by all species in each life stage, including within-egg calls (Fig. 3). Repertoire size estimates should be interpreted with caution as the sample size and strict captive settings of the present work likely did not cover the whole range of vocalizations of the studied species.

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Proxies for repertoire size (units) in each life stage of the studied species based on the areas of the minimum convex polygons based on MDS call distribution. Areas are only comparable within species’ age groups and not among species.

View Fullscreen

DISCUSSION

Our study found that significant diversity exists in the trajectories of ontogenetic change of the vocal repertoire among turtle taxa, and that the acoustic repertoire of some species does not grow in a linear way across ontogeny. In the case of K. subrubrum, for example, the repertoire grows substantially from hatchling to juvenile but decreases at adulthood. In B. baska, hatchlings produce a greater diversity of call types and have a larger vocal repertoire occupancy than juvenile and adults. This is shown in a pronounced way by polygon areas representing acoustic occupancy, in which the repertoire size decreases consecutively through the life of this species by orders of magnitude. All the other 4 species (P. expansa, C. serpentina, C. indica, and D. reticularia) have a relatively linear growth in their repertoire based both on sound types and polygon area (although C. serpentina, C. indica, and D. reticularia lack data from juveniles).

We also observed that the rate of call emission changes across ontogeny in the turtles we studied. In crocodilians, most species vocalize progressively less as they age, with the most vocal period being the first days or weeks of life (Vergne et al. 2009). This pattern was also observed in the number of vocalizations per hour produced by hatchlings of B. baska and C. serpentina, which produce significantly more sounds than individuals in all other life cycle phases. This trait may be associated with synchronous behaviors and within-egg vocalizations exhibited by these 2 species (Lacroix et al. 2022; Jorgewich-Cohen et al. 2024), but it does not appear to hold for P. expansa, which exhibits most sound production at other life stages. It is worth noting that all crocodilian species display parental care and cannibalism toward smaller individuals, which may stimulate and inhibit vocalizations, respectively (Somaweera et al. 2013). However, neither behavior is common or even present in most turtle species, so comparisons between turtles and crocodilians need to be interpreted cautiously.

Our findings on peak frequencies fit with the hypothesis proposed by Vergne et al. (2009), which suggests that crocodilian embryos seem to emit lower frequency sounds in comparison to hatchlings due to sound filtering by the eggshell. Furthermore, the average peak frequency raises consistently with age in half of our studied turtle species, partially contradicting the observation that sound frequency decreases with growth in body size, based on vocalization produced by tortoises (Galeotti et al. 2005).

Some freshwater turtle species, like the red-eared slider (Trachemys scripta elegans) and the Chinese softshell turtle (Pelodiscus sinensis), display greater differences in average peak frequency between sexes than between age groups (Zhou et al. 2022, 2023a). The differences in the average peak frequency between age groups are not significant in most call types produced by these species (Zhou et al. 2022, 2023a). The variety in ontogenetic changes (or stability) in the vocalizations of turtles may be a reflection of specificities in the natural history of each species. For instance, sounds produced by hatchlings have nearly always the shortest mean duration in comparison to sounds produced by individuals in other life stages for nearly all studied species (except for B. baska and adult P. expansa). Sound production may represent a greater cost for hatchlings than for individuals at other stages, as they have a more limited capacity to store energetic resources (Ryan 1988).

Other acoustic characters may be more informative regarding repertoire maturation and social mediation. In this study, tonal sounds were displayed only by adults of B. baska, and C. indica (although they have been documented for embryos and hatchlings in other turtle species such as Graptemys ouachitensis [Geller and Casper 2019], C. serpentina [Lacroix et al. 2022], and Apalone spinifera [Geller and Casper 2023]). Moreover, some sound types have been frequently found in combination in the recordings of adult C. serpentina and P. expansa. While speculative, this could be an indication that calls can be used to construct new meanings in a syntax-like fashion—a behavior reported only for mammals and birds (Engesser and Townsend 2019). The development of complex vocalizations and/or a more refined use of sound combinations, compared to an apparent random use of a repertoire composed by simpler sounds, could be an indication of ontogenetic changes associated with maturation.

Moreover, sociality in other taxa seems to promote the selection of more complex and tonal sounds (May-Collado et al. 2007; Freeberg et al. 2012; see also Peckre et al. 2019), a pattern that may also hold for chelonians. This seems to be the case of P. expansa, currently the only turtle species assumed to care for their hatchlings in a sound-mediated social behavior (Ferrara et al. 2012). This atypical highly social behavior may explain why hatchlings and juveniles of this species produce various types of tonal sounds. Similarly, a complex acoustic repertoire can be observed in species that potentially use sounds to mediate various within-nest behaviors (Jorgewich-Cohen et al. 2024)—which may or may not include the snapping turtle (Lacroix et al. 2022), the South American river turtle (Ferrara et al. 2012, 2014c), and all sea turtles (McKenna et al. 2019; Field et al. 2021), a topic that needs further investigation.

All turtle species known to vocalize during synchronous behavior have similar, convergent nesting behavior (Jorgewich-Cohen et al. 2022a, 2024). Specifically, they all lay many small eggs in a clutch and synchronize hatch, which already implies the existence of at least limited social behavior (Spencer 2012), although the underlying mechanisms are currently unknown. Although research into sound production is still in its early stages, those turtle species studied to date that lay only 1 or few eggs generally do not appear to make sounds within the egg and do not exhibit synchronous hatching behavior (i.e., D. reticularia and K. subrubrum [Jorgewich-Cohen et al. 2024]). The lack of coordinated embryo/hatchling behavior in hatching and nest emergence in some turtle species could explain the accentuated ontogenetic growth in the acoustic repertoire (i.e., D. reticularia, K. subrubrum, C. indica, and potentially C. oblonga [Giles et al. 2009]) in contrast to the more social mentioned species, as complex social interactions happen only later in their life.

Likewise, hatchlings of the green sea turtle (Chelonia mydas) have been shown to produce 4 different sounds from within the egg and nest (Ferrara et al. 2014a), and later, in water, juveniles produce 10 different sound types (Charrier et al. 2022). The ontogenetic growth in the vocal repertoire of C. mydas seems to be even larger than what is observed in some other species that also produce a large diversity of calls, like B. baska and P. expansa. However, adults, juveniles and hatchlings of P. expansa occur in the same habitat and appear to have significant social interactions (Portelinha et al. 2014), which may explain the similarities in the sound types produced by individuals in all age groups. In contrast, in both C. mydas and C. serpentina there is a change in habitat use in different life stages (Arthur et al. 2008; Congdon et al. 1993, respectively), potentially coupled with changes in vocal repertoire. We hypothesize that changes in social interactions and habitat use during ontogeny pressures for changes in vocal repertoire, whereas continuity leads to more uniformed vocalizations across life stages, especially in species that socialize among individuals of all ages.

Vocalizations in the Leatherback Sea turtle (Dermochelys coriacea) have also been recorded in different life stages. Hatchlings of this species have been found to produce more sound types while still inside the egg, compared to after hatching (Ferrara et al. 2014b). The only information about the acoustic repertoire of adult leatherback turtles comes from nesting females (Cook and Forrest 2005), which limits interpretation as they might be only side effects of laying behavior and not necessarily intent for communication. Nonetheless, we predict that this species probably has a limited acoustic repertoire, as they travel long distances in open ocean (Eckert 2002) and encounter fewer individuals of the same species (Benson et al. 2007) compared to most turtle species.

The fact that different studies that described acoustic repertoires of turtles used different methodological approaches can complicate comparisons and ecological interpretation. Here we used a series of different methods that allow comparisons with different studies and also help circumventing weaknesses of a specific analysis. While it is prudent to prioritize methods that are not subjected to human bias, human perception seems to be the most sensitive method to differentiate sounds (Piczak 2015)—especially when the sample is small and there is no reference library.

The machine-based approach, although less sensitive, described similarities among some of the sounds selected with the human-based approach and can bring insights when interpreting the natural history of the species in concern. For instance, in the spectral analysis of the common snapping turtle, sound types I to VI were present only in the recordings from an isolated female, while types VII to XIII were produced only by the couple (Supplemental Material 1 and 6). This could be an indication that snapping turtles have different dialects, as occurs in many other groups of animals such as whales, primates, and birds (Ford 1984; Mitani et al. 1992; Tubaro and Segura 1994). Nevertheless, the pDFA recovered 5 different sound types, combining some of the categories produced by the lonely female and by the couple, indicating that the differences between these 2 groups are most probably context dependent, making the dialect hypothesis less parsimonious.

In fact, recordings covering all sex and age groups in different ecological conditions for each of our studied species will likely reveal larger acoustic repertoires. This can be demonstrated by comparing our description of the South American river turtle acoustic repertoire to a description produced by long-term studies in the wild (Ferrara et al. 2012, 2014c). We sampled around 70% of the sounds this species is known to produce (Fig. 2), likely because captive animals are not exposed to the whole range of social situations they experience in the wild. Our descriptions establish a baseline for future research on captive and wild populations of the species studied here. Regardless of repertoire coverage, our standardized approach enables insightful comparisons among species. However, further work is needed to enhance the precision of these comparisons and to clarify aspects of each species' life history.

Our findings shed light on the patterns of ontogenetic changes in turtle acoustic repertoire based on sounds produced by 6 different species representing distinct phylogenetic groups. We found that acoustic communication comes in great diversity within turtle taxa and likely reflects differences in species’ life history, including the degree to which sounds may play roles in synchronous hatching and emergence from nests, the extent of social interactions at different life stages, and ontogenetic changes in habitat use. Our understanding of turtle vocal communication would benefit from further comparative research on the ontogeny of acoustic repertoires of additional taxa and ecological niches.

DEFICIENCIES AND FUTURE STEPS

Despite extensive research, even well-studied animal species continue to surprise researchers with previously unrecorded vocalizations. For instance, ostriches (Struthio camelus), a paleognath bird that has been the focus of studies since the 19th century (Forbes 1881), have been found to produce vocalizations that were, until recently, undocumented (Chiappone et al. 2024). Bonobos (Pan paniscus), one of our closest living relatives, have only recently had their acoustic repertoire comprehensively documented (Wegdell et al. 2025). These examples highlight the challenges associated with carefully cataloguing a species’ acoustic repertoire.

Sampling the full range of sounds produced by a species is inherently difficult, and it is impossible to ascertain with confidence that all vocalizations have been captured. However, by analyzing vocalizations from multiple individuals of the same species and monitoring the accumulation of unique call types, researchers can observe an asymptote—a point at which no new sounds are observed even after continuous data collection. This method provides a degree of confidence in the representation of the most common and characteristic elements of the species' acoustic repertoire. Nevertheless, rare sounds that are produced only in specific ecological or social context may remain undetected. This issue is exacerbated in captive settings, where animals are not exposed to all contexts they encounter in the wild.

Considering that our study was conducted exclusively in captivity and that our sample size was limited by the availability of specimens, we caution against interpreting the available repertoires and their life-stage ontogenies as exhaustive descriptions. We highlight the need to further explore the acoustic behavior of turtle species in general, considering it is a group historically unstudied regarding they vocal abilities.

Here we show that the trajectory of ontogenetic change in the acoustic repertoire of turtles is not always linear. This may reflect cases where either differences are just the result of morphological changes that constrain vocalizations (e.g., frequency increases across ontogeny), or that such changes evolve according to different social or environmental contexts throughout their lives, likely not resulting in linear changes in their acoustic behavior. Future studies should aim to investigate the underlying mechanisms driving these nonlinear changes, including the role of morphological development, social interactions, and environmental factors. Additionally, exploring how these factors influence the functional significance of vocalizations across different life stages will provide deeper insights into the evolutionary and ecological implications of acoustic communication in turtles.

Further research on complex tonal sounds, call combination, and acoustical differences at the populational level probing for the potential presence of dialects would also be welcome additions to our understanding of turtle communication in relation with their natural history and sociality. Long-term research focused on disentangling the nuances of vocal behavior in turtles are necessary for a broader understanding of their ecology and the evolutionary patterns of sociality within amniotes.