The major part of the information available about turtle communication indicates that visual and olfactory stimuli are the signal modalities most commonly used for these animals (Kiester 1977; Alho and Pádua 1982; Galeotti et al. 2005a, 2005b). Although turtles have been historically considered the least-vocal reptiles (Gans and Maderson 1973), recent studies have shown that acoustic signals play an important role in turtle social behavior and reproduction (Galeotti et al. 2005a; Giles et al. 2009; Ferrara et al. 2013b, 2014a, 2014b, 2017, 2018). We know that at least 56 species, 14 genera, and 10 families of turtles (Campbell and Evans 1967; Campbell 1972; Mrosovsky 1972; Auffenberg 1977; Sacchi et al. 2003; Galeotti et al. 2005a, 2005b; Giles et al. 2009; Ferrara et al. 2013a, 2014a, 2014b, 2017, 2018; Lenz 2017; Monteiro et al. 2019) use sounds in different contexts such as in agonistic interactions, courtship and copulation, and migration behavior when males and females are also attracted by their movements and odors (Auffenberg 1978; Galeotti et al. 2005b; Ferrara et al. 2009, 2013a; Ibáñez et al. 2012).

We observed that Podocnemis expansa hatchling sounds were emitted at a higher frequency than those of adults, embryos began vocalizing 8–36 hrs before hatching and, during an experiment, adult females approached the hatchlings when the hatchlings began to vocalize in the river. We hypothesize that these calls have multiple functions including eliciting communal digging, synchronizing emergence, soliciting parental care, and maintenance of migratory group cohesion (Ferrara et al. 2013b). Sound communication has been documented in 6 species of sea turtles: Dermochelys coriacea, Chelonia mydas, Lepidochelys olivacea, Eretmochelys imbricata, and Caretta caretta (Mrosovsky 1972; Cook and Forrest 2005; Ferrara et al. 2014a, 2014b; McKenna 2016; Lenz 2017; McKenna et al. 2019; Monteiro et al. 2019) as well as Natator depressus (R.C.V., unpubl. data, 2014). For those species, the sound was recorded in embryos (inside the egg) and hatchlings in and out of water (Ferrara et al. 2014a, 2014b; McKenna 2016; Lenz 2017; McKenna et al. 2019; Monteiro et al. 2019), but the vocal repertoire has only been described for D. coriacea, C. mydas, C. caretta, and E. imbricata (Ferrara et al. 2014a, 2014b; Lenz 2017; Monteiro et al. 2019). Those sounds were classified as pulses, sounds with harmonic and nonharmonic frequency bands modulated in amplitude and frequency (Ferrara et al. 2014a, 2014b; Lenz-Rivera 2017; Monteiro et al. 2019), and as hybrid sounds with characteristics of both pulses and harmonic sounds (Ferrara et al. 2014a, 2014b; Lenz 2017). The objective of this study was to verify if Lepidochelys kempii embryos and hatchlings emit sounds during and after incubation in air and underwater and if their sounds are similar to those of other sea turtle species.

Methods. — We recorded sound in L. kempii during June 2016 in Playa Santander (19°54′07.8 ″N, 96°30′19.1″W), Veracruz, Mexico. All sound recordings were made using a Fostex FR-2 digital recorder adjusted to 48 kHz and a sample size of 24 bits. The underwater recordings were made with a Reson (TC 4032-1) omnidirectional hydrophone with a sensitivity of 5 Hz–120 kHz ± 2 dB. Airborne sounds were recorded using a Sennheiser K6 unidirectional microphone with a Sennheiser ME-66 windscreen. The system had a sensitivity of 40 Hz–20 kHz and 2.5 dB. The recordings were monitored using Sony MDR-7506 headphones; the recording level was established manually to maximize the signal-to-noise ratio and prevent distortions caused by excessive gain (“clipping”).

We recorded 5 nests 6 d before they were estimated to hatch (estimations made from nesting date and incubation times at local temperatures) and 2 nests that already contained hatchlings. We took the hatchlings out of the nest and placed them one by one on the surface of the water in an 800-l plastic water tank filled with saltwater to record the hatchlings underwater; the hydrophone was suspended in the center of the water column.

We used Raven Pro 1.4 (Cornell Lab of Ornithology 2010) to analyze the recordings using the following spectrographic parameters: hamming window type and fast Fourier transform (FFT) window sizes, varying from 256 to 512 samples, so we could identify the best spectrographic representation of each sound detection. Detections with sufficiently high signal-to-noise ratio were then selected to characterize the sound repertoire by grouping them into categories for subsequent classification into sound types. Only those sounds that were not overlapping with other signals were selected for extraction of the acoustic parameters used to describe each type of sound in the repertoire. We measured 6 acoustic parameters: minimum and maximum frequencies (Hz), peak frequency (Hz), fundamental frequency (Hz), delta time (s), and number of harmonics.

Results.—We detected 189 sounds from the embryos and hatchlings of L. kempii during 7 hrs (4 hrs inside the egg, 2 hrs inside the nest, 1 hr underwater). Inside the nest during the embryos phase (egg), we detected 70 samples, after egg hatching inside the nest we detected 5 samples, and underwater we detected 114 samples. The sounds were classified by 3 independent experts into 6 types according to their aural and spectral characteristics (Fig. 1; Table 1). Detected sounds include pulses, sounds with harmonic and nonharmonic frequency bands, modulated in amplitude and frequency, and hybrid sounds with characteristics of harmonic and pulsed sounds.

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Descriptive statistics of the acoustic parameters of each type of sound produced by Lepidochelys kempii.

View Fullscreen

Type I (n = 64) sounds had harmonic and nonharmonic frequency bands with frequency modulated in either ascending (up sweep) or descending (down sweep) fashion. Duration of the sound varied from 0.01 to 0.14 sec and the peak frequency varied from 93.8 to 1687.5 Hz.

Type II (n = 6) sounds were the most complex sound type in the repertoire, with part of the signal having pulse characteristics and the other part with nonharmonic or harmonic frequency bands (Watkins 1967). In this hybrid sound, the harmonic frequency bands were emitted before and after the pulses. The peak frequency varied from 187.5 to 3843.8 Hz.

Type III (n = 72) sounds were the most common sound type in the repertoire of L. kempii. There were harmonic and nonharmonic series with some frequency modulation (ascending or descending), including inflection points. Underwater, this sound was composed of 1 – 18 notes and once appeared as a bout of 6 notes in a short interval. Duration of each note in this sound varied from 0.008 to 0.149 sec and from the entire bout series lasted from 0.161 to 0.607 sec.

Type IV (n = 9) sounds were composed of nonharmonic frequency bands and did not include frequency modulation or inflexion points. Duration of this sound varied from 0.02 to 0.156 sec and there was no variation in peak frequency, which was 187.5 Hz.

Type V (n = 34) sounds were characterized by a series of short pulses lasting from 0.02 to 0.26 sec (with mean peak frequency of 513.88 Hz and standard deviation of 745.1).

Type VI (n = 4) sounds presented harmonic frequency bands and included 1–2 inflexion points. The frequency modulation was ascending (up sweep) and there were 3–9 harmonics. Duration of the sound varied from 0.05 to 0.23 sec and the peak frequency from 187.5 to 937.5 Hz.

The Type I sounds appeared in all behavioral categories (underwater, nest, and egg), Types II, III, IV, and V sounds appeared in the nest and underwater, and sound VI was detected only underwater. The Type III (n = 114) sound was the most common sound in our sample. The mean peak frequency of the sounds emitted by L. kempii was 440.61 Hz, ranging from 93.8 to 3843.8 Hz.

Discussion. — Most sound energy found in the turtle vocalizations studied so far is concentrated in frequencies less than 4 kHz (Campbell and Evans 1967; Mrosovsky 1972; Auffenberg 1978; Jackson and Awbrey, 1978; Galeotti et al. 2005a; Giles et al. 2009; Ferrara et al. 2013a, 2014a, 2014b; but see Monteiro et al. 2019). The peak frequencies of the sounds emitted by L. kempii embryos and hatchlings varied from 93.8 to 3843.8 Hz; apparently this species emits sounds in the same frequency range as other species studied.

The frequency of sounds of adults of Chelodina colliei varied from 100 to 3500 Hz, but “clicks” extended beyond the upper 20-kHz limit of the recording equipment (Giles et al. 2009). In adults, hatchlings, and embryos of P. expansa, frequency varied from 36.8 to 4500 Hz (Ferrara et al. 2013b), in embryos and hatchlings of D. coriacea varied from 119.2 to 24,000 Hz (Ferrara et al. 2014b), and in embryos and hatchlings of E. imbricata peak frequency reaches 15,000 Hz (Monteiro et al. 2019). This would make sense because the hearing range of sea turtles is similar among these species (Ridgway et al. 1969; Mrosovsky 1972; Samuel et al. 2005; Cook and Forrest 2005), so their vocalizations would be represented in similar frequencies. However, the best hearing sensitivity in subadults of L. kempii is between 200 and 400 Hz and in juveniles is between 100 and 500 Hz (Bartol and Ketten 2006), which coincides with the lower range in the peak frequencies measured in our sample but does not cover the entire frequency range of the repertoire. In P. expansa, we observed a change in their vocal pattern as the individuals developed. This change occurred with the frequency (Hz) and the vocal context, and perhaps the same is true with L. kempii. Unfortunately, we do not have enough data to test this change because we recorded only small size classes.

We found 6 types of sound in embryos and hatchlings of L. kempii, prehatching in the egg, posthatching in the nest, and underwater. These sounds varied from pulses to harmonic and nonharmonic frequency bands with modulated frequency and some inflexion points as well as hybrid sounds. These types of sounds have been observed in other turtle species (Campbell and Evans 1967; Sacchi et al. 2003; Giles et al. 2009; Ferrara et al. 2013a, 2014a, 2014b; McKenna 2016; Lenz 2017; McKenna et al. 2019). Specifically, the sound types found in L. kempii embryos and hatchlings were also found in embryos and hatchlings of P. expansa, C. mydas, D. coriacea, C. caretta, and L. olivacea (Ferrara et al. 2013b, 2014a, 2014b; McKenna 2016; Lenz 2017; McKenna et al. 2019), but no hybrid sounds were found in E. imbricata (Monteiro et al. 2019). However, in L. kempii we found multiple series of harmonic and nonharmonic frequency bands modulated with up to 18 inflexion points. This type of sound was found with a maximum of 4 notes in P. expansa and C. carretta and with a maximum of 2 notes in C. mydas and D. coriacea, and these bouts of sounds were also found inside the egg and nest in these species (Ferrara et al. 2013a, 2014a, 2014b; Lenz 2017).

The hybrid sounds in L. kempii appear with 2 or 3 notes, and in P. expansa, C. mydas, and D. coriacea we found just 2 notes (Ferrara et al. 2013b, 2014a, 2014b). Freeberg et al. (2012) suggested that species that have a great complexity of social interactions develop a morecomplex quality of sound signals than do those species with less complex social interactions. Even though the embryos and hatchlings of L. kempii do not have as extensive a vocal repertoire as in other species (Giles et al. 2009; Ferrara et al. 2013a, 2014a, 2014b), they present a variety and a high complexity in their sounds for a group that was assumed to engage in little social interaction. Therefore, we suggest that social interactions in L. kempii and other turtle species are more complex than previously thought.

Chelonian sounds show a repertoire that can range from pulses and tonal signals to more-complex signals, with harmonic frequency bands, frequency modulation, and hybrid sounds (Galeotti et al. 2005; Giles et al. 2009; Ferrara et al. 2013a, 2014a, 2014b, 2018). In addition to these characteristics being similar to the L. kempii repertoire, they are also similar to the sounds produced by cetaceans (Murray et al. 1998), suggesting that chelonians might also have a complex mechanism of sound production.

We observed that embryos and hatchlings of L. kempii underwater had a higher diversity and variation in the structure of the sound than did hatchlings inside the nest. The higher diversity and variation in the structure of the sound in P. expansa was in the hatchling stage inside the nest. In P. expansa, the greater diversity of sound types and the variation in the acoustic structure could be related to a greater necessity for interchange of information and a greater complexity of information exchanged within the nest among the hatchlings (Ferrara et al. 2013b). Probably, L. kempii has the same necessity as does P. expansa for interchange of information with a greater complexity of information but, instead, this information exchange looks more complex between embryos and hatchlings underwater. Embryos may exchange information to synchronize hatching and, underwater, hatchlings may maintain proximity in the water by means of acoustic communication in a new environmental condition.

The use of acoustic communication appears to be highly appropriate for aquatic turtles because the visibility in water is much lower than in air, and the acoustic modality of signaling is much more efficient than are visual or olfactory cues. More-specific and detailed studies need to be undertaken to better understand what role acoustic communication plays in the social behavior of the species, and how individual turtles react to the acoustic signals released by conspecifics, and if there is interspecific recognition of acoustic signals within a turtle assemblage in nature.