Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Description of the behavioural contexts of underwater sound production in juvenile green turtles Chelonia mydas

Description of the behavioural contexts of underwater sound production in juvenile green turtles... Green sea turtles Chelonia mydas have the ability to hear and produce sounds under water, with some of them potentially involved in social communication. To investigate the potential biological function of these sounds, we used a combination of acoustic, video and multi-sensor recordings of 23 free-ranging juvenile green turtles and we examined the co-occurrences of sounds with behaviours or external events. Our study revealed that most of the sounds were produced when the sea turtles were resting or swimming. However, four sound types were produced in more specific contexts. Long sequences of rumbles were recorded after sunset and mainly during resting. All these rumbles appear to have been produced by several individuals recorded simultaneously, suggesting that rumbles may be used for social interactions. The frequency modulated sound was highly associated with scratching behaviour. The grunt that was produced occasionally when green turtles were vigilant or approaching a conspecific. The long squeak was produced significantly by a small number of individuals in the presence of humans. The grunt and the long squeak may be the first evidence of an alarm or warning signal for intra-specific com- munication in green turtles. Our results mark a significant milestone in advancing the understanding of sound production in the behavioural ecology of sea turtles. Further experimental investigations (i.e., playback experiments) are now required to test the hypotheses suggested by our findings. Warning signals could be used to prevent sea turtles of a danger and may contribute to their conservation. Significance statement Underwater sound production in Chelonioidea is yet not well documented. We investigated the contexts of sound production in green sea turtles. Our results show that juvenile green sea turtles produced at least 4 identified sounds in specific contexts. This is a crucial step, as it will provide a solid basis for understanding the acoustic behaviour of green sea turtles and for improving current conservation methods. To date, the lack of knowledge on sea turtle behavioural ecology and acoustic communication hinders the implementation of mitigation measures to effectively reduce mortality and disturbance from human activities. Our findings offer the possibility of using species-specific sounds in a deterrent system to prevent them from potentially dangerous areas, including areas where seismic surveys, construction work, or areas with fishing activities (fishing Communicated by F. Jensen. * Léo Maucourt Institut des Neurosciences Paris-Saclay, CNRS, Université [email protected] Paris-Saclay, 91400 Saclay, France * Damien Chevallier Association ACWAA , Quartier l’Etang, [email protected] 97217 Les Anses d’Arlet, France Department of Mathematical Sciences, Stellenbosch Unité de Recherche BOREA, MNHN, CNRS 8067, SU, IRD University, Victoria Street, 7602 Stellenbosch, South Africa 207, UCN, UA, Station de Recherche Marine de Martinique, Quartier Degras, Petite Anse, 97217 Les Anses dArlet, African Institute for Mathematical Sciences (AIMS), 7 Martinique, France Melrose Rd, 7945 Muizenberg, Cape Town, South Africa Université des Antilles, Campus de Schoelcher, 97275 Schoelcher Cedex, Martinique, France Vol.:(0123456789) 25 Page 2 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 nets) occurs with the aim of reducing the risk of temporary 200 Hz for Kemp’s ridley Lepidochelys kempii (Bartol and or permanent hearing damage or accidental by-catch. Ketten 2006) and 50 and 1200 Hz with maximum sensitivity between 100 and 400 Hz for hatchlings leatherback Dermo- Keywords Behaviour · Chelonia mydas · Chelonians · chelys coriacea (Piniak et al. 2012). Juveniles · Underwater sound production Airborne sound production was observed and character- ised in nesting female sea turtles, such as the leatherback (Carr 1952; Mrosovsky 1972; Lutcavage and Lutz 1996; Introduction Cook and Forrest 2005). Additionally, sound production was demonstrated in embryos and hatchlings of several sea Intraspecific communication plays a crucial role in social turtles species, such as hawksbill Eretmochelys imbricata interactions such as finding a mate, interacting within (Monteiro et al. 2019), Kemp’s ridley (Ferrara et al. 2019), social partners (particularly during foraging or alerting to green (Ferrara et al. 2014a, b), olive ridley Lepidochelys the presence of a predator), and providing care to offspring olivacea (Ferrara et al. 2014b; McKenna et al. 2019), and (Bradbury and Vehrencamp 2011). There are many ways of leatherback (Ferrara et al. 2014b, c). However, the biologi- conveying information, with acoustic signals being among cal function and sound production mechanisms are not well the most widely recognised and studied signals (Bradbury understood, and further studies are required to draw conclu- and Vehrencamp 2011). Among vertebrates, sound produc- sions regarding the occurrence of acoustic communication in tion has been well studied in birds, mammals and anurans these species (McKenna et al. 2019). Studies carried out on but much less is known about non-avian reptiles which are other hatching non-avian reptiles have demonstrated sound regarded to produce sounds relatively rarely in compari- production in crocodilians, such as the Spectacled caiman son (Vergne et al. 2009; Jerem and Mathews 2021; Russell Caiman crocodilus and American alligator Alligator missis- and Bauer 2021). Several studies have been conducted on sipiensis (Garrick and Garrick 1978; Britton 2001; Higgs freshwater turtles, but little is known about sea turtle sound et al. 2002; Vergne et al. 2009). In the Nile crocodile Croco- production. dylus niloticus, it was shown that sounds produced by the Sound production in freshwater turtles (in air and under embryos can synchronise hatching, aiding their emergence water) and acoustic communication has recently been dem- from the nest (Vergne and Mathevon 2008). Nevertheless, onstrated (Ferrara et al. 2013, 2014b, 2017; Papale et al. the sounds produced by sea turtle embryos do not appear to 2020; Jorgewich-Cohen et  al. 2022), suggesting that the synchronise hatchlings in olive ridley, green and leatherback produced sounds could be involved in social behaviour (e.g. sea turtles (McKenna et al. 2019; Nishizawa et al. 2021). oblong turtle Chelodina oblonga, Giles et al. 2009; arrau Underwater sound production in sea turtles has been turtle Podocnemis expansa, Ferrara et al. 2014c, d). These neglected likely due to the difficulties to record sea turtles in findings marked the initial phase of challenging prevailing their natural environment. The development of multisensory concepts regarding chelonian sound production and social tags with hydrophone has opened new research questions behaviours (Charrier et al. 2022). Until the early 2000s, including the study of underwater acoustic production of sea turtles were considered a ‘silent group’ (Campbell and free-ranging wild sea turtles (Charrier et al. 2022). The first Evans 1972). underwater sounds described in juvenile green sea turtles Additionally, due to the absence of a visible tympanum in (Charrier et al. 2022) exhibited similarities with the general sea turtles (possessing only the middle and inner ear with- acoustic structure of the underwater sounds produced by the out an external ear, Bartol and Musick 2002), they have oblong turtle and Kemp’s ridley sea turtle hatchlings (Giles historically been considered as deaf. However, behavioural et al. 2009; Ferrara et al. 2019). However, both frequency and electrophysiological studies have shown their hearing and temporal features varied across these studies, attribut- abilities in air and under water (DeRuiter and Larbi Douk- able to size differences between hatchlings and juveniles. ara 2012; Lavender et al. 2014). Hearing seems to be well In each of the three species studied (i.e. oblong, Kemp’s adapted to their underwater environment, as the subtympanic ridley and green), sound production ranged from simple fatty tissue of sea turtles has a density similar to that of pulses to more complex sounds endowed with a harmonic water, which reduces sound attenuation and optimises sound structure and a frequency modulation pattern. The source transmission to the inner ear (Ketten 2008). The juvenile sea of sounds described in Charrier et al. (2022) was supported turtles’ underwater hearing ability ranges between 50 and by control recordings carried out in the green turtle natural 1600 Hz with maximum sensitivity between 200 and 400 habitat, but without the presence of green turtles nearby. Hz for green Chelonia mydas (Piniak et al. 2016), 50 and Indeed, recordings showed that while some detected sounds 1100 Hz with maximum sensitivity between 100 and 400 were similar in their main structure, the frequency and/or Hz for loggerhead Caretta caretta (Lavender et al. 2014), temporal characteristics were different from the sounds pro- 100 and 500 Hz with maximum sensitivity between 100 and duced by green sea turtles. These variations suggest that Behavioral Ecology and Sociobiology (2025) 79:25 Page 3 of 14 25 some of these sounds were likely produced by other marine nuchal shell and pygales plate. The freediver then positioned species, such as crustaceans or fishes, present in the green the turtle against his chest with the hind flippers against his turtle’s habitat. Although the biological function of sound breastplate and rose to the surface. A second diver held the production is not fully understood, Charrier et al. (2022) fore flippers and helped to lift the turtle on to the deck of demonstrated that turtles’ squeaks were individual-specific the boat for measurements and tagging (Nivière et al. 2018; and could potentially be used for individual recognition. Fur- Bonola et al. 2019). Once on a boat, each individual was thermore, all recorded individuals produced sounds at fre- identified by scanning its Passive Integrated Transponder quencies within the hearing range of green turtles, suggest- (PIT) or tagged with a new PIT if it was unknown, as ing potential implications for social communication. Other described in Siegwalt et al. (2020) and Lelong et al. (2024). sounds in their repertoire, such as the rumble, the toc and the Identifying individuals with PIT tags enables the population Frequency Modulated Sound (FMS) are within the best audi- demography to be monitored, particularly during Capture- tory sensitivity of juvenile green turtles (the frequency of the Mark-Recapture campaigns in these areas. The CATS Cams highest energy of these sounds ranging from 200 to 400 Hz), device was attached to the carapace using four suction cups, but they did not show any individual identity. These sounds as described in Jeantet et al. (2020). This suction cup attach- may thus be involved in intraspecific communication, but are ment method avoids the use of glue on the carapace. Air was probably not involved in individual recognition processes. manually expelled from the cups, which were held in place The aim of the present study was to investigate the behav- by the use of a galvanic timed-release system, used to limit ioural contexts during which free-ranging juvenile green the duration of the deployment. The dissolving of galvanic turtles, equipped with a biologger CATS Cam (Custom- timed-release system by seawater and the slightly positive ized Animal Tracking Solution, Germany), a multi-sensor buoyancy of the device (23.3 × 13.5 × 4 cm for 785 g) led tag associated with a video camera and one hydrophone, to the remote release of the device with the removal of the produced sounds, thereby providing insights into their asso- suction cups from the shell several hours to two days later, ciated biological functions. To automatically identify their thus avoiding the need to recapture the turtle and minimis- behaviour, we used a deep learning algorithm (Jeantet et al. ing the stress associated with a second capture to recover 2021), trained to predict the observed behaviour of green the device. A CATS Cams device included a video-recorder −1 turtles from data recorded by the CATS Cams device (from (1920 × 1080 pixels at 30 frames.s , viewing angle of 100°) the accelerometer, gyroscope and depth). combined with a tri-axial accelerometer, a tri-axial gyro- scope, a tri-axial magnetometer, time-depth recorder, hydro- phone (HTI 96 min, frequency response: 2 Hz to 30 kHz, METHODS sensitivity: −165 dB re 1 V/mPa), thermometer, luminosity and a GPS tracker. All auxiliary data were sampled at 20 Study Site and Data Collection from Free‑ranging Hz. Devices were recovered using a goniometer (RXG-134, Green turtles CLS, France) by geolocation of an Argos SPOT-363 A tag (MK10, Wildlife Computers Redmond, WA, USA), glued This study was carried out from May 2018 to May 2022 to the CATS Cams device. Due to low light conditions after in coastal waters of Grande Anse d’Arlet (14°30.158’ N, sunset, the cameras were programmed to record from 05:00 61°5.271’ W), Anse Noire (14°31.683’ N, 61°5.320’ W) to 19:00, but others sensors are still recording. CATS Cams and Anse Dufour (14°31.562’N, 61°5.425’W), Martinique devices were deployed on 23 juvenile green turtles (8 in island (French West Indies), where juvenile green turtles 2018 were included in the study by Charrier et al. 2022, p. recruit. They originate from various Caribbean and Atlantic 10 in 2021 and 5 in 2022). Due to the different configura- nesting sites (Chambault et al. 2018). There they spend sev- tions between 2018 and the following years, or the early eral years, feeding on seagrass beds located in shallow shel- release of the device, only 12 of the 23 devices were record- tered bays (Siegwalt et al. 2020; Lelong et al. 2024). Once ing after sunset. A total of 247 h of recorded tag data (sound, they reach a size close to sexual maturity (i.e. at around 80 video and accelerometer) were investigated, with an average cm curved carapace length), they embark on a major post- duration of recording of 10h44 (range: 3h24–18h37, n = 23), developmental migration and they join Caribbean and Atlan- Recording started at 7:45 am at the earliest and stopped at tic adult feeding grounds (Chambault et al. 2018). Juvenile 03:23 am next day at the latest. However, the retained part green turtles were captured along the coast by freedivers at of the deployments for the analyses (data usable without various sites with depths up to 25 m. The capture of each uncertainty, i.e. no behaviour could be identified from the turtle was performed by up to three freedivers when the tur- accelerometer data) lasted 207 h in total. tle was static (i.e. resting or feeding on the sea floor). The All data were collected in the wild from free-ranging ani- freediver silently dived towards the turtle to avoid detection mals, thus it was not possible to use blinded methods. and once close enough and above the animal, seized the 25 Page 4 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 the sixth behaviour labelled surface activity, included all Acoustic recordings and analyses activities (e.g. breathing, basking) occurring between 0 and 0.3 m depth (depth data was collected from the CATS CATS Cams device recorded acoustic data in mono at a frequency sampling rate of 24 kHz (16 bit). Listening and Cams device). With this method, surface activity is not exclusive, and occurs at the same time as the turtle is spectrogram labelling of sound files were performed using Avisoft SASlab Pro (Avisoft Bioacoustics, Version 5.3.01, swimming, as in the study by Jeantet et al. (2020). Similar to sound labelling, a label started at the beginning of the 14 May 2022). A label starts at the beginning of the sound (turtle sounds or boat noise) and stops at its end. To improve behaviour and stopped at its end. In addition to monitoring behaviour, three external visualisation of the sounds on spectrograms (Hamming, Fast Fourier Transform [FFT] size 1024 pts), all sound l fi es were events (boat noise and presence of conspecifics or humans, Table  1) were recorded to assess their potential impact down sampled at 22 kHz, as there was no energy for frequen- cies above 10 kHz. on sound production of the turtles. All video files were processed using the software VLC media player so that Sounds, behaviours and external events labelling the timing of each external event of interest was recorded, using slow motion and frame-by-frame modes if necessary. Eleven sound types were considered in the present analysis External events, including the presence of a conspecific or a human, were considered if at least one turtle other (Table 1). Ten of which (mono, doublet, triplet, multipulse, toc, croak, rumble, FMS, short squeak and long squeak) than the equipped turtle or at least one human was vis- ible in the camera field with direct interaction (e.g. touch- have been described in the juvenile green turtle sound reper- toire in Charrier et al. (2022). They were classified into four ing, smelling, biting and intimidating for turtles; follow- ing and touching the turtle for humans) or without (e.g. main sound categories: pulse, Low Amplitude Call (LAC), FMS and squeak. The grunt is a new sound type and a new resting, swimming, feeding for turtles; being in the water, swimming, snorkelling for humans). If the conspecific or sound category not previously described (see Fig. S1 in the Supplementary Information for spectrograms of the eleven the human left the camera’s field of view and then can be observed again at a later stage (i.e., beyond the next sounds type). We defined six main behavioural categories (feeding, two minutes), then the label ended at the end of the first observation, and the second observation was considered gliding, resting, scratching, swimming and surface activ- ity, Table  1; see Jeantet et al. 2020; for the behavioural as a new observation and a new label was thus created. Indeed, during the “blind period”, we cannot be sure that definitions). For data collected in 2018, behaviours were visually analysed and defined using the software VLC the same individual stayed in the vicinity of the equipped green turtle. Finally, the boat noise events (i.e. boat motor media player (version 3.0.18 Vetinari, 13 October 2022; VideoLAN, Paris, France). For the analysis of data col- noise) were detected on the audio recordings. Since the different loggers could start in a non-synchro- lected in 2021 and 2022, we used a deep learning algo- rithm coded in custom Python scripts to automatically nous way, the behaviours detected from the accelerometer data using the automatic classification algorithm (Jeantet identify the behaviours from the accelerometer, gyroscope and depth (Jeantet et al. 2021). The algorithm was trained et al. 2021) were synchronised with the sound data using external events (e.g. breathing at surface). on the behaviour dataset labelled from the visual analysis of the 2018 video files. Details on this algorithm can be Dummy Coding Procedure found in Jeantet et al. (2021). The behaviours obtained with this method included different variants and were To assess the behavioural context of underwater sound pro- pooled into five behavioural categories (feeding, gliding, resting, scratching, and swimming). In the same manner, duction in juvenile green turtles, we investigated the co- occurrence of sounds with behaviours and external events. Table 1 List of the labels for the Sounds Behaviour External Events sounds, behaviours and external events (see fig. S1 in the Pulse 1. Mono Grunt 7. Grunt I. Feeding A. Boat noise supplementary information for 2. Doublet LAC 8. Croak II. Gliding B. Conspecific spectrograms of sounds type) 3. Triplet 9. Rumble III. Resting C. Human 4. Multipulse Squeak 10. Long Squeak IV. Scratching 5. Toc 11. Short Squeak V. Surface activity FMS 6. FMS VI. Swimming Behavioral Ecology and Sociobiology (2025) 79:25 Page 5 of 14 25 Table 2 Descriptive data on individuals with sound production (for a total of 20 666 sound 1-s bins) Mono Doublet Triplet Multipulse Toc Croak Rumble FMS Grunt Short Squeak Long Squeak Number of individuals 18 21 18 12 18 17 20 11 21 17 19 Number of 1-s bins 341 435 825 94 1024 185 14 560 171 347 1096 1588 To do this, each recording day and all label durations were not possible to calculate their detection rate. To estimate the split into 1-second bin (1-s bin), using a “Dummy coding” total sound production (Fig. 1) and the behavioural budget method coded in R. Basically, the occurrence of a sound (Fig. 2) of green turtles, we calculated the percentage of 1-s or a behaviour or an external event at a given time took the bins for each sound and behaviour types. All dummy coding value of 1. If absent, it took the value of 0. This method procedures and Monte Carlo simulation were performed in quantified the number of 1-s bins of a given sound that RStudio (version 4.3.0, 2023-04-21). co-occurred with the 1-s bins of a given behaviour or an external event. Only sounds occurring during an identified behaviour or an external event were considered. RESULTS To assess if co-occurrences were significantly different from chance, we performed a Monte Carlo simulation on The behavioural contexts of underwater sound production of the percentage of each co-occurrence (i.e. the number of 23 juvenile green turtles were investigated. From the dataset 1-s bins of a given sound co-occurring with a behaviour of recorded tag data, 20 666 sound 1-s bins (representing or an external event, divided by the total of 1-s bins of this approximately 6 h of sound recordings), 746 382 behaviour sound). Each co-occurrence was simulated 10 000 times 1-s bins (~ 207 h) and 92 549 external event 1-s bins (~ 26 by randomly assigning 1-s bins of sounds with behaviours h) were identified (see Table S2 in the Supplementary Infor - and external events, while maintaining the same total num- mation for the full dataset). Individual produce sounds for ber of 1-s bins for each sound. To obtain an empirical P 0.15–14.14% of the time they were recorded. value from Monte Carlo simulation, we use the formula (r + 1)/(n + 1), where r is the number of these replicates Sound production that produce a test statistic greater than or equal to that calculated for the actual data and n is the number of repli- Among the eleven recorded sounds, the rumble was the most cate samples that have been simulated (n = 10 000) (Davi- produced. It accounted for 70.5% of total sound production son and Hinkley 1997). Monte Carlo simulations estimate (Table 2; Fig. 1a). The remaining 29.5% of the total sound significance but do not measure it (North et al. 2002). production was distributed among the other ten sound cat- Since a sound can occur during 1-s bins of two different egories. Among these, the pulse and the squeak categories external events, it was included in both external events as were the most represented, each accounting for 13.3% and they are not mutually exclusive (e.g. boat noise can occur in 13% of the total sound production, respectively (Table 2; the presence of a human or a conspecific) and it is therefore Fig.  1b). In contrast, the FMS category represented only Fig. 1 Pie charts of (a) total sound production and (b) sound production without rumble (the remaining 29.5% of total sounds production). The pulse category includes Mono, Doublet, Triplet, Multipulse and Toc. The LAC category includes Croak and Rumble. The Squeak category includes Short and Long Squeak 25 Page 6 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Table 3 Descriptive data on Feeding Gliding Resting Scratching Swimming Surface activity individuals with behaviour budget (for a total of 746 382 Number of individuals 22 23 23 21 23 23 behaviour 1-s bins) Number of 1-s bins 41 861 17 521 368 847 18 178 299 975 47 659 0.8% of the total sound production (Table 2; Fig. 1b). The LAC category (including the croak and the rumble) was the most produced sound category, with rumble contributing to the large majority of it. Additionally, 99.2% of rumbles 1-s bins were recorded after sunset. Behavioural Budget and External Event occurrences Among the six defined behaviours, resting and swimming were the most observed, each accounting for 49.4% and 37.8% of total behaviour budget, respectively (Table  3; Fig. 2). Among the three external events, the boat noise was the most observed (Table 4). The interaction of the tagged turtle with a conspecific was much more frequent than with human (Table 4). Co‑Occurrences of Sounds with Behaviours Ten of the eleven types of sound described (excluding FMS) Fig. 2 Pie chart of total behavioural budget. As turtles were always were all generally produced during resting and swimming swimming during surface activity, its budget was included in the swimming behaviour. It accounted for 6.4% of the total behavioural (Fig. 3a, b, d, e). The five sound types of the pulse category budget and always co-occurred with swimming and the rumble were mostly produced during resting (all of them are significantly different from chance with p < 0.05, Table 5a), with 74.9% of rumbles were produced during rest- Table 4 Descriptive data on individuals with external event produc- tion (for a total of 92 549 external event 1-s bins) ing (p = 0.0001, Table 5a). Rumbles were mostly recorded in three individuals whose showed an intense rumble produc- Human Conspecific Boat noise tion after sunset, mainly during resting (Fig. 4a, b): 881 1-s Number of individuals 15 22 23 bins from 20:11 to 21:27, 6 753 1-s bins from 18:44 to 21:41 Number of 1-s bins 955 14 765 76 829 and 6 783 1-s bins from 18:41 to 21:31. However, only 12 out of 23 individuals were recorded after sunset. The grunt and the two sound types of the squeak category were mostly produced during swimming (all of them are Co‑occurrence of sounds with external events significantly different from chance with p < 0.001, Table 5a), with 68.9% of grunts were produced during swimming Very few sounds were heard during external event-driven (p = 0.0001, Table 5a). contexts except during boat noise (mono and croak were The FMS was quite rare (171 1-s bins recorded in produced significantly differently from chance with total on 11 individuals, Table 2) as well as the scratching p < 0.01, Table  5b). Only FMS and long squeak were behaviour (accounting for 2.4% of the total behavioural produced in the presence of humans (p < 0.05, Table 5b), budget, Fig. 2). However, the FMS was highly associated whereas eight sound types were produced in the presence with this scratching behaviour (Fig.  3c), with 89.5% of of conspecifics in the camera’s field of view (four sound FMS were produced during scratching (11 individuals, types, including grunt, were produced significantly dif- p = 0.0001, Table 5a). ferently from chance with p < 0.05, Table 5b). However, Behavioral Ecology and Sociobiology (2025) 79:25 Page 7 of 14 25 Fig. 3 Barplots of total co-occurrences of (a) pulses, (b) LAC, (c) FMS, (d) grunts and (e) squeaks with juvenile green turtle behaviours the long squeak production in the presence of human DISCUSSION was significantly different from chance (p = 0.0001, Table 5b). This is the sound heard the most during such This study provides new knowledge on the acoustic behav- event (10 1-s bins). iour of green sea turtles based on a multi-year research program investigation focusing on immature green turtle 25 Page 8 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Table 5 Co-occurrence of sounds with (a) behaviours and (b) exter- events. A Monte Carlo simulation was performed to assess if the per- nal events. For each cell, the first line corresponds to the number of centage of a given co-occurrence was significantly different from the individuals in which the co-occurrence happened (n ind), the second one obtained randomly (* indicate p < 0.05, ** indicate p < 0.01, *** line is the number of 1-s bins of sounds co-occurring with (a) behav- indicate p < 0.001; see table S3 in the supplementary information for iours or (b) external events, and the third line is the percentages of p-values) 1-s bins of sounds, co-occurring with (a) behaviours or (b) external (a) Feeding Gliding Resting Scratching Swimming Surface activity Pulse Mono n ind 1 5 12 2 17 5 Total of 1-s bins 1 15 190 2 133 9 1-s bins rate 0.3% 4.4% ** 55.7% ** 0.6% 39% 2.6% Doublet n ind 1 3 15 1 17 4 Total of 1-s bins 1 7 256 2 169 11 1-s bins rate 0.2% 1.6% 58.9% *** 0.5% 38.9% 2.5% Triplet n ind 2 4 13 1 15 8 Total of 1-s bins 2 12 522 2 287 20 1-s bins rate 0.2% 1.5% 63.3% *** 0.2% 34.8% 2.4% Multipulse n ind No co-occur- 1 3 No co-occur- 12 4 Total of 1-s bins rence 2 57 rence 35 5 1-s bins rate 2.1% 60.6% * 37.2% 5.3% Toc n ind 5 4 15 3 15 2 Total of 1-s bins 22 32 636 73 261 2 1-s bins rate 2.1% 3.1% 62.1% *** 7.1% *** 25.5% 0.2% LAC Croak n ind 2 1 10 2 13 5 Total of 1-s bins 4 2 94 15 70 6 1-s bins rate 2.2% 1.1% 50.8% 8.1% *** 37.8% 3.2% Rumble n ind 5 7 18 2 14 3 Total of 1-s bins 564 195 10 911 21 2869 236 1-s bins rate 3.9% 1.3% 74.9% *** 0.1% 19.7% 1.6% FMS n ind 3 No co-occur- 2 11 6 No co-occurrence Total of 1-s bins 4 rence 5 153 9 1-s bins rate 2.3% 2.9% 89.5% *** 5.3% Grunt n ind 8 5 17 5 20 4 Total of 1-s bins 25 11 63 9 239 24 1-s bins rate 7.2% 3.2% 18.2% 2.6% 68.9% *** 6.9% Squeak Short Squeak n ind 2 9 10 2 15 11 Total of 1-s bins 6 83 447 2 558 29 1-s bins rate 0.5% 7.6% *** 40.8% 0.2% 50.9% *** 2.6% Long Squeak n ind 4 7 6 2 16 11 Total of 1-s bins 19 185 653 7 724 30 1-s bins rate 1.2% 11.6% *** 41.1% 0.4% 45.6% *** 1.9% (b) Human Conspecific Boat noise Pulse Mono n ind No co-occur- 3 8 Total of 1-s bins rence 17 51  1-s bins rate 5 % *** 15 % ** Doublet n ind No co-occur- 2 8 Total of 1-s bins rence 8 54 1-s bins rate 1.8 % 12.4 % Triplet n ind No co-occur- 2 9 Total of 1-s bins rence 7 88 1-s bins rate 0.8 % 10.7 % Multipulse n ind No co-occur- 2 4 Total of 1-s bins rence 4 14 1-s bins rate 4.3 % * 14.9 % Toc n ind No co-occur- 3 4 Total of 1-s bins rence 14 42 1-s bins rate 1.4 % 4.1 % Behavioral Ecology and Sociobiology (2025) 79:25 Page 9 of 14 25 Table 5 (continued) LAC Croak n ind No co-occur- 2 6 Total of 1-s bins rence 10 31 1-s bins rate 5.4 % ** 16.8 % ** Rumble n ind No co-occur- 2 6 Total of 1-s bins rence 3 45 1-s bins rate 0 % 0.3 % FMS n ind 1 No co-occur- 4 Total of 1-s bins 1 rence 17 1-s bins rate 0.6 % * 9.9 % Grunt n ind No co-occur- 4 6 Total of 1-s bins rence 11 21 1-s bins rate 3.2% * 6.1 % Squeak Short Squeak n ind No co-occur- No co-occur- 7 Total of 1-s bins rence rence 111 1-s bins rate 10.1 % Long Squeak n ind 3 No co-occur- 6 Total of 1-s bins 10 rence 73 1-s bins rate 0.6 % *** 4.6 % Significant values (marked with one, two or three *) are in bold Fig. 4 Boxplots of occurrences of (a) resting behaviour and (b) rum- twenty minutes in each time slot, and the n corresponds to the num- bles produced at each hour of the day. Boxes indicate the inter quar- ber of sampled* individuals resting (a) or producing a rumble (b) at tile range, with the central line depicting the median and the whisk- least once in each time slot. No sea turtles were recorded before 8am ers extending to min and max values and outliers. For each hour, the and after 3am. * Individuals with a tag that was recording data N corresponds to the number of individuals recorded for more than 25 Page 10 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 population in the Lesser Antilles. The investigations on the Grunts were found to be commonly produced during rest- co-occurrences of sounds and behaviours of juveniles at ing and mostly during swimming. However, when watch- their foraging grounds are key to understanding the potential ing the behaviour of the green turtles producing grunts, importance of acoustic communication in this endangered we found a stereotyped retreat head movement coinciding species. Altogether, these findings offer new and innovative with the production of grunts, which suggests a potential perspectives for improving sea turtle conservation. link between this sound and a specific behavioural response Our first finding revealed a highly variable production in green turtles. Indeed, few grunts were produced in the rate within the sound repertoire of the green turtle. Doublet, presence of conspecifics (p = 0.0443) during an intimida- grunt and rumble were the sounds most commonly produced tion interaction (i.e. two conspecifics swimming around among juvenile green turtles (n ≥ 20 individuals). However, each other, which is considered swimming behaviour in the doublets and grunts were produced far less than the rumble. results, p = 0.0001). Thus, such head movement could be a Indeed, the doublet and the grunt accounted for 2.1% and visual agonistic component analogous to aggressive threat 1.7% of the total sound production, respectively, while the signals observed in others species, such as lizard waving rumble accounted for 70.5% of the total sound production. their tails at approaching predators (Bradbury and Vehren- In contrast, the FMS and the Multipulse were the sounds camp 2011). Given that grunts were also produced when produced least commonly among individuals (n ≤ 12 indi- sea turtles exhibited a vigilance posture (head up, look- viduals) and accounted for 0.8% and 0.5% of the total sound ing around, leaning on its front legs) while swimming or production, respectively. The croak was also little produced, feeding or when approaching conspecifics, we suggest that accounting for 0.9% of the total sound production. However, grunts could function as warning signals, playing a role in since the tags recorded for short amounts of time (i.e. less intra-specific acoustic communication among juvenile green than 24 h at a time and not on several successive days), turtles. it is likely that all sounds, behaviours and external events We previously assessed that the squeak might be a good could not be sampled for each individual within a recording candidate for intra-specific communication due to its indi- session. vidual stereotypy (i.e., individual-specific) (Charrier et al. This study provided the first investigation focusing on 2022). Our recordings show that long squeaks were recorded the relationship between behaviours, external events and for three individuals during human avoidance events. Two underwater sounds produced by juvenile green turtles. The produced long squeaks just after being released from the pulse, the LAC, the grunt and the squeak categories were tagging boat and one produced a long squeak while swim- generally produced while juvenile green turtles were rest- ming away from three swimmers. Long squeaks were also ing or swimming. These prevailing co-occurrences can be observed on one other individual after the sunset. The data attributed to their predominant activity patterns during the from the hydrophone, the pressure logger and the 3D-accel- tag deployment which was mainly resting and swimming. As erometer showed that the green turtle was about to surface, seven out of the eleven described sounds types (mono, dou- stopped, then dove rapidly after producing such sound, blet, triplet, multipulse, toc, croak and short squeak), were suggesting an avoidance behaviour. The observation of not produced in a specific context, it is not possible to sug- long squeaks produced by green turtles during avoidance gest any biological function or to conclude on the absence behaviour (which is considered swimming behaviour in the of function for these seven sounds types. Thus, we cannot results, p = 0.0001) in the presence of humans (p = 0.0001) confirm whether these sound types are used for communica- provides interesting insights into the potential link between tion. Furthermore, as no sound was recorded during physical this sound in response to perceived threats and anti-predator interactions between conspecifics (affiliative or agonistic), or avoidance behaviour. The limitations of the camera’s field we suggest that sound may not be the medium used for direct of view (100°) highlight the possibility that important con- social interaction, as sight or smell remain effective at close textual cues, such as the presence of conspecifics or another range (Bartol and Musick 2002). However, we found some potential threats (animal or a human), may not have been sounds produced by a small number of turtles while in pres- captured, thus complicating the interpretation of behav- ence of conspecifics (without physical contact). ioural responses. Indeed, we often saw that the turtle was Among the context-specific sound productions, the FMS vigilant and looking around, but we could not explain why was mostly produced during scratching behaviour (Fig. 3c). it remained alert. The long squeak may thus constitute a Both FMS and scratching are individually rare compare to first evidence of an alarm acoustic signal in juvenile green others sounds and behaviours (Tables 2 and 3). However, the turtles, used to alert conspecifics from a threat. While evi- average percentage of co-occurrence is very high (83.9%, dence of alarm acoustic signals has been shown in other Table 5a). The FMS was generally produced when sea turtles non-avian reptile species (e.g. leopard lizard Gambelia wisli- seem to be alone and undisturbed. However, we are not able zenii, Wever et al. 1966; fossorial snakes, Young et al. 2013), to assess any biological function for this sound. confirmation of the alarm function of the long squeak in Behavioral Ecology and Sociobiology (2025) 79:25 Page 11 of 14 25 green turtles would require experimental validation using (2012) reported that loggerhead turtles dived immediately playback experiments. following an airgun shot, while Weir (2007) reported that We recorded long sequences of rumbles after sunset and 83% of sea turtles (including olive ridley, leatherback and during resting behaviour. The overlapping and varied ampli- loggerhead turtles) continued to bask at the surface during tude levels of these rumbles suggest that several individuals airguns exposure and as the vessel and towed equipment in the vicinity may contribute to these long-lasting acoustic moved past. The variability of behavioural responses to interactions, potentially engaging in some form of coordi- a noisy event is highly complex in many species includ- nated sound production. It may thus constitute first evidence ing non-avian reptiles. For example, freezing is a typi- of a social acoustic communication among juvenile green cal stress response in non-avian reptiles, as in the Eastern turtles during night-time resting periods. Similar production, blue tongued lizard Tiliqua scincoides when exposed to called choruses, are observed in other taxa like birds, insects the noise of mining machinery (Mancera Alarcon 2016). frogs and fishes (Farina and Ceraulo 2017). It highlights the Given impact of anthropogenic noise on non-avian reptile potential for social communication and coordination among behaviour remains relatively understudied compared to green turtles during night-time resting periods. The only other taxa (Simmons and Narins 2018; Jerem and Mathews example is the Travancore tortoise Indotestudo travancorica, 2021), the interpretation of the results regarding the effect in which several individuals called together, with individual of boat noise on the behaviour of juvenile green turtles is sound productions appearing to be regularly spaced, at night constrained. Further research is required to compare the (Campbell and Evans 1972). While chorusing behaviour is behaviour of green turtles occurring in areas with low and well documented among insects, fishes, frogs and birds and high-levels of boat traffic. has associated with diverse ecological functions (e.g. ener- The four sounds types, rumble, FMS, grunt and long getic and behavioural matters in birds, Farina and Ceraulo squeak were the only ones produced by juvenile green tur- 2017), its occurrence in non-avian reptiles, particularly Che- tles in specific behavioural contexts. Such findings could lonians, remains relatively unknown. potentially contribute to sea turtle conservation. Indeed, our Although we did not examine the sound level of the boat findings highlight intra-specific social interactions, but also noise we recorded, our findings show there is no specific alert or vigilance sounds. It provides a strong baseline to test sound produced, nor any behaviour that stopped during free-ranging green turtle with playback experiments using boat noise events. This suggests that juvenile green turtles the long squeak, the grunt or the rumble to assess if we can may not alter their acoustic behaviour in direct response elicit a behavioural reaction of the tested individuals. It has to boat activities, as it was reported in diamondback ter- been shown that some non-avian reptiles are sensitive to rapins Malaclemys terrapin (Lester et al. 2013). However, heterospecific calls and can adapt their behavioural response the absence of change in behaviour through their sound from threat signals (marine iguana Amblyrhynchus crista- production or behaviour does not mean that boat activi- tus, Vitousek et al. 2007; e.g. brown anoles Anolis sagrei, ties do not induce physiological stress or cause hearing Cantwell and Forrest 2013). Some of these sounds could be loss (e.g. red-eared slider Trachemys scripta elegans, Salas used in deterrent systems aiming to reduce by-catch of sea et al. 2023). Although green turtles have not altered their turtles by preventing them from being entangled in fishing sound production in response to boat noise, they appear nets (Chevallier et al. 2024). to surface less often to breathe (LM, unpubl. data). This Indeed, the use of acoustic pingers on harbour porpoise alteration in surfacing behaviour may indicate a poten- Phocoena phocoena for instance has shown that acoustic tial physiological response to the presence of boats and signals can reduce the occurrence of by-catches (Kraus et al. associated noise, suggesting a possible impact on their 1997; Trippel et al. 1999; Gearin et al. 2000). As part of respiratory patterns or diving behaviour. A comparable the TOPASE program (Martinique-Guadeloupe) aiming to behavioural response has previously been observed in a reduce by-catch of sea turtles in fishing nets, recent investi- juvenile green turtle, wherein it appears to remain station- gations based on the findings of this present study has been ary on or near the sea floor when ships pass nearby (Tyson carried out. Selected green turtle sounds (squeaks and rum- et al. 2017). Most studies of sea turtle responses to boat bles) were used in playback experiments on free-ranging noise have focused on exposure to high-intensity seismic green turtles during their foraging activities in Martinique airguns in a closed or semi-closed environment, which (Chevallier et al. 2024). Encouragingly, the playback tests limits the ability to assess the behaviour of free-rang- showed that a majority of green turtles responded to these ing turtles exposed to different boat noises (O’Hara and sounds by exhibiting behavioural alertness. Further investi- Wilcox 1990; Moein et al. 1994; McCauley et al. 2000). gations are now required to confirm such findings. Specifi- The rare studies on free-ranging turtles exposed to seis- cally, attention can be directed towards examining specific mic airgun surveys reported highly variable behavioural sounds such as the grunt and the long squeak to confirm responses of turtles. Indeed, DeRuiter and Larbi Doukara their role in avoiding danger from conspecifics or humans 25 Page 12 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 the stress associated with capture. The use of suction cups instead of and to explore the biological function of the rumble in social glue makes the attachment system much less invasive. The slightly interactions and social grouping pattern during night-time. positive buoyancy and hydrodynamics of the camera ensure that the Finally, to gain a deeper understanding of the biological turtle’s movements are not constrained when diving or surfacing. The function of these sounds, it would be valuable to perform automatic release of the device after a maximum of two days avoids the stress of a second capture. similar recordings using animal-borne tags on adult green turtles. By comparing the sound repertoire of juvenile to Open Access This article is licensed under a Creative Commons Attri- those of adult green turtles, we could potentially highlight bution 4.0 International License, which permits use, sharing, adapta- any variations in acoustic parameters and reveal ontogenetic tion, distribution and reproduction in any medium or format, as long information regarding the development of sound-producing as you give appropriate credit to the original author(s) and the source, organs. Moreover, this approach could reveal new sound pro- provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are ductions used only during the breeding period, providing included in the article’s Creative Commons licence, unless indicated further insights into the acoustic behaviour of adult green otherwise in a credit line to the material. If material is not included in turtles. the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will Supplementary Information The online version contains supplemen- need to obtain permission directly from the copyright holder. To view a tary material available at https://doi. or g/10. 1007/ s00265- 025- 03561-z . copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Acknowledgements This study was carried out within the framework of the Plan National d’Action Tortues Marines Antilles. We thank the two anonymous reviewers for their helpful comments. References Author contributions DC, IC and LM designed research; DC, NA, OB, LJ, NL, FL, PL, ML, JM and SR performed data collection; LM and Bartol SM, Ketten DR (2006) Turtle and tuna hearing. In: Swimmer CH designed the code of the Dummy Coding method; LM, IC, LJ and Y, Brill RW (eds) Sea turtle and pelagic fish sensory biology: DC analysed data; DC acquired funding; LM wrote the original version Developing techniques to reduce sea turtle bycatch in longline of the paper; IC, DC and CH edited the paper. fisheries. NOAA Tech. Mem. NMFS-PIFSC-7. National Ocean and Atmospheric Administration (NOAA), US Department of Funding Open access funding provided by Université des Antilles. Commerce, pp 98–105. http:// www. pifsc. noaa. gov/ tech/ NOAA_ The Program BEPHYTES of the Centre National de la Recherche Sci-Tech_ Memo_ PIFSC_7. pdf entifique (CNRS) was co-funded by the FEDER Martinique (Fonds Bartol SM, Musick JA (2002) Sensory biology of sea turtles. In: Lutz Européen de Développement Régional PO/FEDER/FSE 2014– PL, Musick JA, Wyneken J (eds) The biology of sea turtles. CRC 2020, Conventions 246239), Collectivité Territoriale de Martinique Press, Boca Raton, pp 79–102 (CTM, convention 258342), the Direction de l′Environnement, de Bonola M, Girondot M, Robin J-P et al (2019) Fine scale geographic l′Aménagement et du Logement (DEAL, Martinique, France (Conven- residence and annual primary production drive body condition tion N°2017/164894), the Office De l′Eau (ODE) Martinique, France of wild immature green turtles (Chelonia mydas) in Martinique (Convention n°180126) and the Office Français de la Biodiversité Island (Lesser antilles). Biol Open 8:bio048058. https:// doi. org/ (OFB, Parc Naturel Marin de Martinique, OFB-23-0563), the town of 10. 1242/ bio. 048058 Les Anses d’Arlet and ANSLO’S. The Program TOPASE of the CNRS Bradbury JW, Vehrencamp SL (2011) Principles of animal communica- was funded by Fonds Européen pour les affaires maritimes et la pêche tion, 2nd edn. Sinauer Associates, Sunderland, MA (FEAMP), the Minister of Agriculture and FranceAgriMer. Léo Mau- Britton ARC (2001) Review and classification of call types of juvenile court PhD scholarship was supported by the Collectivité Territoriale crocodilians and factors affecting distress calls. Crocodilian Biol of Martinique (CTM). Evol 364:364–377 Campbell HW, Evans WE (1972) Observations on the vocal behavior Data availability All datasets generated or analysed during this study of chelonians. Herpetologica 28:277–280 are included in this published article and its supplementary informa- Cantwell LR, Forrest TG (2013) Response of Anolis sagrei to acoustic tion files. calls from predatory and nonpredatory birds. J Herpetol 47:293– 298. https:// doi. org/ 10. 1670/ 11- 184 Carr AF (1952) Handbook of turtles. Comstock Publishing Associates, Declarations Ithaca, NY Chambault P, de Thoisy B, Huguin M et al (2018) Connecting paths Conflict of interest The authors declare no conflict of interest. between juvenile and adult habitats in the Atlantic green turtle using genetics and satellite tracking. Ecol Evol 8:12790–12802. Ethics approval Fieldwork was performed in accordance with the https:// doi. org/ 10. 1002/ ece3. 4708 French legal and ethical requirements. Specifically, the protocol was Charrier I, Jeantet L, Maucourt L, Régis S, Lecerf N, Benhalilou A, approved by the Conseil National de la Protection de la Nature and Chevallier D (2022) First evidence of underwater vocalizations in the French Ministry for Ecology (permit numbers: 201710-0005 and green sea turtles Chelonia mydas. Endanger Species Res 48:31– R02-2020-08-10-006) and followed the recommendations of the Police 41. https:// doi. org/ 10. 3354/ esr01 185 Prefecture of Martinique. Fieldwork was carried out under the certifica- Chevallier D, Maucourt L, Charrier I et al (2024) The response of tion of DC (prefectural authorisations’ owner) under strict compliance sea turtles to vocalizations opens new perspectives to reduce of the Police of Martinique’s recommendations to minimize animal dis- their bycatch. Sci Rep 14:16519. https:// doi. or g/ 10. 1038/ turbance. Indeed, the capture, although stressful for the animal, respects s41598- 024- 67501-z the safety conditions for the divers as well as for the sea turtle. The Cook SL, Forrest TG (2005) Sounds produced by nesting leatherback handling time on the boat did not exceed 10 min in order to minimise sea turtles (Dermochelys coriacea). Herpetol Rev 36:387–390 Behavioral Ecology and Sociobiology (2025) 79:25 Page 13 of 14 25 Davison AC, Hinkley DV (1997) Bootstrap methods and their applica- vertebrates. Nat Commun 13:6089. h t t p s : / / d o i . o r g / 1 0 . 1 0 3 8 / tion. Cambridge University Press, Cambridges41467- 022- 33741-8 DeRuiter SL, Larbi Doukara K (2012) Loggerhead turtles dive in Ketten DR (2008) Underwater ears and the physiology of impacts: response to airgun sound exposure. Endanger Species Res 16:55– comparative liability for hearing loss in sea turtles, birds, and 63. https:// doi. org/ 10. 3354/ esr00 396 mammals. Bioacoustics 17:312–315. https:// doi. or g/ 10. 1080/ Farina A, Ceraulo M (2017) The acoustic chorus and its ecological 09524 622. 2008. 97538 60 significance. In: Farina A, Gage SH (eds) Ecoacoustics: the eco- Kraus SD, Read AJ, Solow A, Baldwin K, Spradlin T, Anderson E, logical role of sounds. Wiley, Hoboken, NJ, pp 81–94 Williamson J (1997) Acoustic alarms reduce porpoise mortality. Ferrara CR, Vogt RC, Sousa-Lima RS (2013) Turtle vocalizations as Nature 388:525. https:// doi. org/ 10. 1038/ 41451 the first evidence of posthatching parental care in chelonians. J Lavender AL, Bartol SM, Bartol IK (2014) Ontogenetic investiga- Comp Psychol 127:24–32. https:// doi. org/ 10. 1037/ a0029 656 tion of underwater hearing capabilities in loggerhead sea tur- Ferrara CR, Mortimer JA, Vogt RC (2014a) First evidence that hatch- tles (Caretta caretta) using a dual testing approach. J Exp Biol lings of Chelonia mydas emit sounds. Copeia 2014:245–247. 217:2580–2589. https:// doi. org/ 10. 1242/ jeb. 096651 https:// doi. org/ 10. 1643/ CE- 13- 087 Lelong P, Besnard A, Girondot M et al (2024) Demography of endan- Ferrara CR, Vogt RC, Giles JC, Kuchling G (2014b) Chelonian vocal gered juvenile green turtles in face of environmental changes: communication. In: Witzany G (ed) Biocommunication of ani- 10 years of capture-mark-recapture efforts in Martinique. Biol mals. Springer, Dordrecht, pp 261–274 Conserv 291:110471. https:// doi. org/ 10. 1016/J. BIOCON. 2024. Ferrara CR, Vogt RC, Harfush MR, Sousa-Lima RS, Albavera E, 110471 Tavera A (2014) First evidence of leatherback turtle (Dermo- Lester LA, Avery HW, Harrison AS, Standora EA (2013) Recreational chelys coriacea) embryos and hatchlings emitting sounds. boats and turtles: behavioral mismatches result in high rates of Chelonian Conserv Biol 13:110–114. https:// doi. org/ 10. 2744/ injury. PLoS ONE 8:e82370. https://doi. or g/10. 1371/ jour nal. pone. CCB- 1045.100823 70 Ferrara CR, Vogt RC, Sousa-Lima RS, Tardio BMR, Bernardes VCD Lutcavage M, Lutz PL (1996) Diving physiology. In: Lutz PL, Musick (2014) Sound communication and social behavior in an amazo- JA (eds) The biology of sea turtles. CRC Press, Boca Raton, pp nian river turtle (Podocnemis expansa). Herpetologica 70:149– 277–296 156. https://d oi.o rg/1 0.1 655/H ERPET OLOGI CA-D-1 3-0 0050R 2 Mancera Alarcon K (2016) Effects of anthropogenic noise on the Ferrara CR, Vogt RC, Eisemberg CC, Doody JS (2017) First evidence behaviour, physiological traits and welfare of two animal mod- of the pig-nosed turtle (Carettochelys insculpta) vocalizing under- els: wild mice (Mus musculus) and Eastern blue tongued lizard water. Am Soc Ichthyol Herpetol 105:29–32. https:// doi. org/ 10. (Tiliqua scincoides). PhD thesis, The University of Queensland 1643/ CE- 16- 407 McCauley RD, Fewtrell J, Duncan AJ, Jenner C, Jenner M-N, Penrose Ferrara CR, Vogt RC, Sousa-Lima RS, Lenz A, Morales-Mávil JE JD, Prince RIT, Adhitya A, Murdoch J, McCabe K (2000) Marine (2019) Sound communication in embryos and hatchlings of Lepi- seismic surveys — a study of environmental implications. APPEA dochelys kempii. Chelonian Conserv Biol 18:279–283. https://doi. J 40:692. https:// doi. org/ 10. 1071/ aj990 48 org/ 10. 2744/ CCB- 1386.1 McKenna LN, Paladino FV, Tomillo PS, Robinson NJ (2019) Do Garrick LD, Garrick RA (1978) Temperature influences on hatch- sea turtles vocalize to synchronize hatching or nest emergence? ling Caiman crocodilus distress calls. Physiol Zool 51:105–113. Copeia 107:120–123. https:// doi. org/ 10. 1643/ CE- 18- 069 https:// doi. org/ 10. 1086/ physz ool. 51.2. 30157 859 Moein SE, Musick JA, Keinath JA, Barnard DE, Lenhardt M, George R Gearin PJ, Gosho ME, Laake JL, Cooke L, DeLong R, Hughes KM (1994) Evaluation of seismic sources for repelling sea turtles from (2000) Experimental testing of acoustic alarms (pingers) to reduce hopper dredges. Virginia Institute of Marine Science, College of bycatch of harbour porpoise, Phocoena phocoena, in the state William & Mary, Gloucester Point, VA of Washington. J Cetacean Res Manag 2:1–9. https:// doi. org/ 10. Monteiro CC, Carmo HMA, Santos AJB, Corso G, Sousa-Lima RS 47536/ jcrm. v2i1. 483 (2019) First record of bioacoustic emission in embryos and hatch- Giles JC, Davis JA, McCauley RD, Kuchling G (2009) Voice of the lings of hawksbill sea turtles (Eretmochelys imbricata). Chelonian turtle: the underwater acoustic repertoire of the long-necked fresh- Conserv Biol 18:273–278. https:// doi. org/ 10. 2744/ CCB- 1382.1 water turtle, Chelodina oblonga. J Acoust Soc Am 126:434–443. Mrosovsky N (1972) Spectrographs of the sounds of leatherback tur- https:// doi. org/ 10. 1121/1. 31482 09 tles. Herpetologica 28:256–258 Higgs DM, Brittan-Powell EF, Soares D, Souza M, Carr C, Dooling R, Nishizawa H, Hashimoto Y, Rusli MU, Ichikawa K, Joseph J (2021) Popper A (2002) Amphibious auditory responses of the American Sensing underground activity: diel digging activity pattern during alligator (Alligator mississipiensis). J Comp Physiol A 188:217– nest escape by sea turtle hatchlings. Anim Behav 177:1–8. https:// 223. https:// doi. org/ 10. 1007/ s00359- 002- 0296-8doi. org/ 10. 1016/j. anbeh av. 2021. 04. 013 Jeantet L, Planas-Bielsa V, Benhamou S et al (2020) Behavioural infer- Nivière M, Chambault P, Pérez T et al (2018) Identification of marine ence from signal processing using animal-borne multi-sensor key areas across the Caribbean to ensure the conservation of the loggers: a novel solution to extend the knowledge of sea turtle critically endangered hawksbill turtle. Biol Conserv 223:170–180. ecology. R Soc Open Sci 7:200139. https:// doi. org/ 10. 1098/ rsos. https:// doi. org/ 10. 1016/j. biocon. 2018. 05. 002 200139 North BV, Curtis D, Sham PC (2002) A note on the calculation of Jeantet L, Vigon V, Geiger S, Chevallier D (2021) Fully convolutional empirical P values from Monte Carlo procedures. Am J Hum neural network: a solution to infer animal behaviours from multi- Genet 71:439–441. https:// doi. org/ 10. 1086/ 341527 sensor data. Ecol Modell 450:109555. https:// doi. org/ 10. 1016/j. O’Hara J, Wilcox JR (1990) Avoidance responses of loggerhead turtles, ecolm odel. 2021. 109555 Caretta caretta, to low frequency sound. Copeia 1990:564–567. Jerem P, Mathews F (2021) Trends and knowledge gaps in field https:// doi. org/ 10. 2307/ 14463 62 research investigating effects of anthropogenic noise. Conserv Papale E, Prakash S, Singh S, Batibasaga A, Buscaino G, Piovano S Biol 35:115–129. https:// doi. org/ 10. 1111/ cobi. 13510 (2020) Soundscape of green turtle foraging habitats in Fiji, South Jorgewich-Cohen G, Townsend SW, Padovese LR et al (2022) Com- Pacific. PLoS ONE 15:e0236628. https:// doi. org/ 10. 1371/ journ mon evolutionary origin of acoustic communication in choanate al. pone. 02366 28 25 Page 14 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Piniak WED, Eckert SA, Harms CA, Stringer EM (2012) Underwa- Tyson RB, Piniak WED, Domit C, Mann D, Hall M, Nowacek DP, ter hearing sensitivity of the leatherback sea turtle (Dermochelys Fuentes MMPB (2017) Novel bio-logging tool for studying coriacea): assessing the potential effect of anthropogenic noise. fine-scale behaviors of marine turtles in response to sound. U.S. Dept. of the Interior, Bureau of Ocean Energy Management, Front Mar Sci 4:219. https:// doi. or g/ 10. 3389/ fmars. 2017. Headquarters, Herndon, VA 00219 Piniak WED, Mann DA, Harms CA, Jones TT, Eckert SA (2016) Vergne AL, Mathevon N (2008) Crocodile egg sounds signal hatch- Hearing in the juvenile green sea turtle (Chelonia mydas): a com- ing time. Curr Biol 18:513–514. https://doi. or g/10. 1016/j. cub. parison of underwater and aerial hearing using auditory evoked 2008. 04. 011 potentials. PLoS ONE 11:e0159711. https://doi. or g/10. 1371/ jour n Vergne AL, Pritz MB, Mathevon N (2009) Acoustic communication al. pone. 01597 11 in crocodilians: from behaviour to brain. Biol Rev 84:391–411. Russell AP, Bauer AM (2021) Vocalization by extant nonavian reptiles: https:// doi. org/ 10. 1111/j. 1469- 185X. 2009. 00079.x a synthetic overview of phonation and the vocal apparatus. Anat Vitousek MN, Adelman JS, Gregory NC, St Clair JJH (2007) Hetero- Rec 304:1478–1528. https:// doi. org/ 10. 1002/ ar. 24553 specific alarm call recognition in a non-vocal reptile. Biol Lett Salas AK, Capuano AM, Harms CA, Piniak WED, Mooney TA (2023) 3:632–634. https:// doi. org/ 10. 1098/ rsbl. 2007. 0443 Temporary noise-induced underwater hearing loss in an aquatic Weir CR (2007) Observations of marine turtles in relation to seismic turtle (Trachemys scripta elegans). J Acoust Soc Am 154:1003– airgun sound off Angola. Mar Turt Newsl 116:17–20 1017. https:// doi. org/ 10. 1121/ 10. 00205 88 Wever EG, Hepp-Reymond MC, Vernon JA (1966) Vocalization and Siegwalt F, Benhamou S, Girondot M et al (2020) High fidelity of sea hearing in the leopard lizard. P Natl Acad Sci USA 55:98–106. turtles to their foraging grounds revealed by satellite tracking and https:// doi. org/ 10. 1073/ pnas. 55.1. 98 capture-mark-recapture: new insights for the establishment of key Young BAA, Mathevon N, Tang Y (2013) Reptile auditory neuro- marine conservation areas. Biol Conserv 250:108742. https://doi. ethology: what do reptiles do with their hearing? In: Köppl C, org/ 10. 1016/j. biocon. 2020. 108742 Manley G, Popper A, Fay R (eds) Insights from Comparative Simmons AM, Narins PM (2018) Effects of anthropogenic noise Hearing Research, vol 49. Springer, New York, NY, pp 323–346. on amphibians and reptiles. In: Slabbekoorn H, Dooling R, https:// doi. org/ 10. 1007/ 2506_ 2013_ 30 Popper A, Fay R (eds) Effects of anthropogenic noise on animals, Springer Handbook of Auditory Research, vol 66. Publisher’s note Springer Nature remains neutral with regard to Springer, New York, pp 179–208. https:// doi. or g/ 10. 1007/ jurisdictional claims in published maps and institutional affiliations. 978-1- 4939- 8574-6_7 Trippel EA, Strong MB, Terhune JM, Conway JD (1999) Mitigation of harbour porpoise (Phocoena phocoena) by-catch in the gillnet fishery in the lower bay of fundy. Can J Fish Aquat Sci 56:113– 123. https:// doi. org/ 10. 1139/ cjfas- 56-1- 113 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral Ecology and Sociobiology Springer Journals

Description of the behavioural contexts of underwater sound production in juvenile green turtles Chelonia mydas

Download PDF
14 pages

Loading next page...
 
/lp/springer-journals/description-of-the-behavioural-contexts-of-underwater-sound-production-ZpF0E4vBHQ

References (54)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2025
ISSN
0340-5443
eISSN
1432-0762
DOI
10.1007/s00265-025-03561-z
Publisher site
See Article on Publisher Site

Abstract

Green sea turtles Chelonia mydas have the ability to hear and produce sounds under water, with some of them potentially involved in social communication. To investigate the potential biological function of these sounds, we used a combination of acoustic, video and multi-sensor recordings of 23 free-ranging juvenile green turtles and we examined the co-occurrences of sounds with behaviours or external events. Our study revealed that most of the sounds were produced when the sea turtles were resting or swimming. However, four sound types were produced in more specific contexts. Long sequences of rumbles were recorded after sunset and mainly during resting. All these rumbles appear to have been produced by several individuals recorded simultaneously, suggesting that rumbles may be used for social interactions. The frequency modulated sound was highly associated with scratching behaviour. The grunt that was produced occasionally when green turtles were vigilant or approaching a conspecific. The long squeak was produced significantly by a small number of individuals in the presence of humans. The grunt and the long squeak may be the first evidence of an alarm or warning signal for intra-specific com- munication in green turtles. Our results mark a significant milestone in advancing the understanding of sound production in the behavioural ecology of sea turtles. Further experimental investigations (i.e., playback experiments) are now required to test the hypotheses suggested by our findings. Warning signals could be used to prevent sea turtles of a danger and may contribute to their conservation. Significance statement Underwater sound production in Chelonioidea is yet not well documented. We investigated the contexts of sound production in green sea turtles. Our results show that juvenile green sea turtles produced at least 4 identified sounds in specific contexts. This is a crucial step, as it will provide a solid basis for understanding the acoustic behaviour of green sea turtles and for improving current conservation methods. To date, the lack of knowledge on sea turtle behavioural ecology and acoustic communication hinders the implementation of mitigation measures to effectively reduce mortality and disturbance from human activities. Our findings offer the possibility of using species-specific sounds in a deterrent system to prevent them from potentially dangerous areas, including areas where seismic surveys, construction work, or areas with fishing activities (fishing Communicated by F. Jensen. * Léo Maucourt Institut des Neurosciences Paris-Saclay, CNRS, Université [email protected] Paris-Saclay, 91400 Saclay, France * Damien Chevallier Association ACWAA , Quartier l’Etang, [email protected] 97217 Les Anses d’Arlet, France Department of Mathematical Sciences, Stellenbosch Unité de Recherche BOREA, MNHN, CNRS 8067, SU, IRD University, Victoria Street, 7602 Stellenbosch, South Africa 207, UCN, UA, Station de Recherche Marine de Martinique, Quartier Degras, Petite Anse, 97217 Les Anses dArlet, African Institute for Mathematical Sciences (AIMS), 7 Martinique, France Melrose Rd, 7945 Muizenberg, Cape Town, South Africa Université des Antilles, Campus de Schoelcher, 97275 Schoelcher Cedex, Martinique, France Vol.:(0123456789) 25 Page 2 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 nets) occurs with the aim of reducing the risk of temporary 200 Hz for Kemp’s ridley Lepidochelys kempii (Bartol and or permanent hearing damage or accidental by-catch. Ketten 2006) and 50 and 1200 Hz with maximum sensitivity between 100 and 400 Hz for hatchlings leatherback Dermo- Keywords Behaviour · Chelonia mydas · Chelonians · chelys coriacea (Piniak et al. 2012). Juveniles · Underwater sound production Airborne sound production was observed and character- ised in nesting female sea turtles, such as the leatherback (Carr 1952; Mrosovsky 1972; Lutcavage and Lutz 1996; Introduction Cook and Forrest 2005). Additionally, sound production was demonstrated in embryos and hatchlings of several sea Intraspecific communication plays a crucial role in social turtles species, such as hawksbill Eretmochelys imbricata interactions such as finding a mate, interacting within (Monteiro et al. 2019), Kemp’s ridley (Ferrara et al. 2019), social partners (particularly during foraging or alerting to green (Ferrara et al. 2014a, b), olive ridley Lepidochelys the presence of a predator), and providing care to offspring olivacea (Ferrara et al. 2014b; McKenna et al. 2019), and (Bradbury and Vehrencamp 2011). There are many ways of leatherback (Ferrara et al. 2014b, c). However, the biologi- conveying information, with acoustic signals being among cal function and sound production mechanisms are not well the most widely recognised and studied signals (Bradbury understood, and further studies are required to draw conclu- and Vehrencamp 2011). Among vertebrates, sound produc- sions regarding the occurrence of acoustic communication in tion has been well studied in birds, mammals and anurans these species (McKenna et al. 2019). Studies carried out on but much less is known about non-avian reptiles which are other hatching non-avian reptiles have demonstrated sound regarded to produce sounds relatively rarely in compari- production in crocodilians, such as the Spectacled caiman son (Vergne et al. 2009; Jerem and Mathews 2021; Russell Caiman crocodilus and American alligator Alligator missis- and Bauer 2021). Several studies have been conducted on sipiensis (Garrick and Garrick 1978; Britton 2001; Higgs freshwater turtles, but little is known about sea turtle sound et al. 2002; Vergne et al. 2009). In the Nile crocodile Croco- production. dylus niloticus, it was shown that sounds produced by the Sound production in freshwater turtles (in air and under embryos can synchronise hatching, aiding their emergence water) and acoustic communication has recently been dem- from the nest (Vergne and Mathevon 2008). Nevertheless, onstrated (Ferrara et al. 2013, 2014b, 2017; Papale et al. the sounds produced by sea turtle embryos do not appear to 2020; Jorgewich-Cohen et  al. 2022), suggesting that the synchronise hatchlings in olive ridley, green and leatherback produced sounds could be involved in social behaviour (e.g. sea turtles (McKenna et al. 2019; Nishizawa et al. 2021). oblong turtle Chelodina oblonga, Giles et al. 2009; arrau Underwater sound production in sea turtles has been turtle Podocnemis expansa, Ferrara et al. 2014c, d). These neglected likely due to the difficulties to record sea turtles in findings marked the initial phase of challenging prevailing their natural environment. The development of multisensory concepts regarding chelonian sound production and social tags with hydrophone has opened new research questions behaviours (Charrier et al. 2022). Until the early 2000s, including the study of underwater acoustic production of sea turtles were considered a ‘silent group’ (Campbell and free-ranging wild sea turtles (Charrier et al. 2022). The first Evans 1972). underwater sounds described in juvenile green sea turtles Additionally, due to the absence of a visible tympanum in (Charrier et al. 2022) exhibited similarities with the general sea turtles (possessing only the middle and inner ear with- acoustic structure of the underwater sounds produced by the out an external ear, Bartol and Musick 2002), they have oblong turtle and Kemp’s ridley sea turtle hatchlings (Giles historically been considered as deaf. However, behavioural et al. 2009; Ferrara et al. 2019). However, both frequency and electrophysiological studies have shown their hearing and temporal features varied across these studies, attribut- abilities in air and under water (DeRuiter and Larbi Douk- able to size differences between hatchlings and juveniles. ara 2012; Lavender et al. 2014). Hearing seems to be well In each of the three species studied (i.e. oblong, Kemp’s adapted to their underwater environment, as the subtympanic ridley and green), sound production ranged from simple fatty tissue of sea turtles has a density similar to that of pulses to more complex sounds endowed with a harmonic water, which reduces sound attenuation and optimises sound structure and a frequency modulation pattern. The source transmission to the inner ear (Ketten 2008). The juvenile sea of sounds described in Charrier et al. (2022) was supported turtles’ underwater hearing ability ranges between 50 and by control recordings carried out in the green turtle natural 1600 Hz with maximum sensitivity between 200 and 400 habitat, but without the presence of green turtles nearby. Hz for green Chelonia mydas (Piniak et al. 2016), 50 and Indeed, recordings showed that while some detected sounds 1100 Hz with maximum sensitivity between 100 and 400 were similar in their main structure, the frequency and/or Hz for loggerhead Caretta caretta (Lavender et al. 2014), temporal characteristics were different from the sounds pro- 100 and 500 Hz with maximum sensitivity between 100 and duced by green sea turtles. These variations suggest that Behavioral Ecology and Sociobiology (2025) 79:25 Page 3 of 14 25 some of these sounds were likely produced by other marine nuchal shell and pygales plate. The freediver then positioned species, such as crustaceans or fishes, present in the green the turtle against his chest with the hind flippers against his turtle’s habitat. Although the biological function of sound breastplate and rose to the surface. A second diver held the production is not fully understood, Charrier et al. (2022) fore flippers and helped to lift the turtle on to the deck of demonstrated that turtles’ squeaks were individual-specific the boat for measurements and tagging (Nivière et al. 2018; and could potentially be used for individual recognition. Fur- Bonola et al. 2019). Once on a boat, each individual was thermore, all recorded individuals produced sounds at fre- identified by scanning its Passive Integrated Transponder quencies within the hearing range of green turtles, suggest- (PIT) or tagged with a new PIT if it was unknown, as ing potential implications for social communication. Other described in Siegwalt et al. (2020) and Lelong et al. (2024). sounds in their repertoire, such as the rumble, the toc and the Identifying individuals with PIT tags enables the population Frequency Modulated Sound (FMS) are within the best audi- demography to be monitored, particularly during Capture- tory sensitivity of juvenile green turtles (the frequency of the Mark-Recapture campaigns in these areas. The CATS Cams highest energy of these sounds ranging from 200 to 400 Hz), device was attached to the carapace using four suction cups, but they did not show any individual identity. These sounds as described in Jeantet et al. (2020). This suction cup attach- may thus be involved in intraspecific communication, but are ment method avoids the use of glue on the carapace. Air was probably not involved in individual recognition processes. manually expelled from the cups, which were held in place The aim of the present study was to investigate the behav- by the use of a galvanic timed-release system, used to limit ioural contexts during which free-ranging juvenile green the duration of the deployment. The dissolving of galvanic turtles, equipped with a biologger CATS Cam (Custom- timed-release system by seawater and the slightly positive ized Animal Tracking Solution, Germany), a multi-sensor buoyancy of the device (23.3 × 13.5 × 4 cm for 785 g) led tag associated with a video camera and one hydrophone, to the remote release of the device with the removal of the produced sounds, thereby providing insights into their asso- suction cups from the shell several hours to two days later, ciated biological functions. To automatically identify their thus avoiding the need to recapture the turtle and minimis- behaviour, we used a deep learning algorithm (Jeantet et al. ing the stress associated with a second capture to recover 2021), trained to predict the observed behaviour of green the device. A CATS Cams device included a video-recorder −1 turtles from data recorded by the CATS Cams device (from (1920 × 1080 pixels at 30 frames.s , viewing angle of 100°) the accelerometer, gyroscope and depth). combined with a tri-axial accelerometer, a tri-axial gyro- scope, a tri-axial magnetometer, time-depth recorder, hydro- phone (HTI 96 min, frequency response: 2 Hz to 30 kHz, METHODS sensitivity: −165 dB re 1 V/mPa), thermometer, luminosity and a GPS tracker. All auxiliary data were sampled at 20 Study Site and Data Collection from Free‑ranging Hz. Devices were recovered using a goniometer (RXG-134, Green turtles CLS, France) by geolocation of an Argos SPOT-363 A tag (MK10, Wildlife Computers Redmond, WA, USA), glued This study was carried out from May 2018 to May 2022 to the CATS Cams device. Due to low light conditions after in coastal waters of Grande Anse d’Arlet (14°30.158’ N, sunset, the cameras were programmed to record from 05:00 61°5.271’ W), Anse Noire (14°31.683’ N, 61°5.320’ W) to 19:00, but others sensors are still recording. CATS Cams and Anse Dufour (14°31.562’N, 61°5.425’W), Martinique devices were deployed on 23 juvenile green turtles (8 in island (French West Indies), where juvenile green turtles 2018 were included in the study by Charrier et al. 2022, p. recruit. They originate from various Caribbean and Atlantic 10 in 2021 and 5 in 2022). Due to the different configura- nesting sites (Chambault et al. 2018). There they spend sev- tions between 2018 and the following years, or the early eral years, feeding on seagrass beds located in shallow shel- release of the device, only 12 of the 23 devices were record- tered bays (Siegwalt et al. 2020; Lelong et al. 2024). Once ing after sunset. A total of 247 h of recorded tag data (sound, they reach a size close to sexual maturity (i.e. at around 80 video and accelerometer) were investigated, with an average cm curved carapace length), they embark on a major post- duration of recording of 10h44 (range: 3h24–18h37, n = 23), developmental migration and they join Caribbean and Atlan- Recording started at 7:45 am at the earliest and stopped at tic adult feeding grounds (Chambault et al. 2018). Juvenile 03:23 am next day at the latest. However, the retained part green turtles were captured along the coast by freedivers at of the deployments for the analyses (data usable without various sites with depths up to 25 m. The capture of each uncertainty, i.e. no behaviour could be identified from the turtle was performed by up to three freedivers when the tur- accelerometer data) lasted 207 h in total. tle was static (i.e. resting or feeding on the sea floor). The All data were collected in the wild from free-ranging ani- freediver silently dived towards the turtle to avoid detection mals, thus it was not possible to use blinded methods. and once close enough and above the animal, seized the 25 Page 4 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 the sixth behaviour labelled surface activity, included all Acoustic recordings and analyses activities (e.g. breathing, basking) occurring between 0 and 0.3 m depth (depth data was collected from the CATS CATS Cams device recorded acoustic data in mono at a frequency sampling rate of 24 kHz (16 bit). Listening and Cams device). With this method, surface activity is not exclusive, and occurs at the same time as the turtle is spectrogram labelling of sound files were performed using Avisoft SASlab Pro (Avisoft Bioacoustics, Version 5.3.01, swimming, as in the study by Jeantet et al. (2020). Similar to sound labelling, a label started at the beginning of the 14 May 2022). A label starts at the beginning of the sound (turtle sounds or boat noise) and stops at its end. To improve behaviour and stopped at its end. In addition to monitoring behaviour, three external visualisation of the sounds on spectrograms (Hamming, Fast Fourier Transform [FFT] size 1024 pts), all sound l fi es were events (boat noise and presence of conspecifics or humans, Table  1) were recorded to assess their potential impact down sampled at 22 kHz, as there was no energy for frequen- cies above 10 kHz. on sound production of the turtles. All video files were processed using the software VLC media player so that Sounds, behaviours and external events labelling the timing of each external event of interest was recorded, using slow motion and frame-by-frame modes if necessary. Eleven sound types were considered in the present analysis External events, including the presence of a conspecific or a human, were considered if at least one turtle other (Table 1). Ten of which (mono, doublet, triplet, multipulse, toc, croak, rumble, FMS, short squeak and long squeak) than the equipped turtle or at least one human was vis- ible in the camera field with direct interaction (e.g. touch- have been described in the juvenile green turtle sound reper- toire in Charrier et al. (2022). They were classified into four ing, smelling, biting and intimidating for turtles; follow- ing and touching the turtle for humans) or without (e.g. main sound categories: pulse, Low Amplitude Call (LAC), FMS and squeak. The grunt is a new sound type and a new resting, swimming, feeding for turtles; being in the water, swimming, snorkelling for humans). If the conspecific or sound category not previously described (see Fig. S1 in the Supplementary Information for spectrograms of the eleven the human left the camera’s field of view and then can be observed again at a later stage (i.e., beyond the next sounds type). We defined six main behavioural categories (feeding, two minutes), then the label ended at the end of the first observation, and the second observation was considered gliding, resting, scratching, swimming and surface activ- ity, Table  1; see Jeantet et al. 2020; for the behavioural as a new observation and a new label was thus created. Indeed, during the “blind period”, we cannot be sure that definitions). For data collected in 2018, behaviours were visually analysed and defined using the software VLC the same individual stayed in the vicinity of the equipped green turtle. Finally, the boat noise events (i.e. boat motor media player (version 3.0.18 Vetinari, 13 October 2022; VideoLAN, Paris, France). For the analysis of data col- noise) were detected on the audio recordings. Since the different loggers could start in a non-synchro- lected in 2021 and 2022, we used a deep learning algo- rithm coded in custom Python scripts to automatically nous way, the behaviours detected from the accelerometer data using the automatic classification algorithm (Jeantet identify the behaviours from the accelerometer, gyroscope and depth (Jeantet et al. 2021). The algorithm was trained et al. 2021) were synchronised with the sound data using external events (e.g. breathing at surface). on the behaviour dataset labelled from the visual analysis of the 2018 video files. Details on this algorithm can be Dummy Coding Procedure found in Jeantet et al. (2021). The behaviours obtained with this method included different variants and were To assess the behavioural context of underwater sound pro- pooled into five behavioural categories (feeding, gliding, resting, scratching, and swimming). In the same manner, duction in juvenile green turtles, we investigated the co- occurrence of sounds with behaviours and external events. Table 1 List of the labels for the Sounds Behaviour External Events sounds, behaviours and external events (see fig. S1 in the Pulse 1. Mono Grunt 7. Grunt I. Feeding A. Boat noise supplementary information for 2. Doublet LAC 8. Croak II. Gliding B. Conspecific spectrograms of sounds type) 3. Triplet 9. Rumble III. Resting C. Human 4. Multipulse Squeak 10. Long Squeak IV. Scratching 5. Toc 11. Short Squeak V. Surface activity FMS 6. FMS VI. Swimming Behavioral Ecology and Sociobiology (2025) 79:25 Page 5 of 14 25 Table 2 Descriptive data on individuals with sound production (for a total of 20 666 sound 1-s bins) Mono Doublet Triplet Multipulse Toc Croak Rumble FMS Grunt Short Squeak Long Squeak Number of individuals 18 21 18 12 18 17 20 11 21 17 19 Number of 1-s bins 341 435 825 94 1024 185 14 560 171 347 1096 1588 To do this, each recording day and all label durations were not possible to calculate their detection rate. To estimate the split into 1-second bin (1-s bin), using a “Dummy coding” total sound production (Fig. 1) and the behavioural budget method coded in R. Basically, the occurrence of a sound (Fig. 2) of green turtles, we calculated the percentage of 1-s or a behaviour or an external event at a given time took the bins for each sound and behaviour types. All dummy coding value of 1. If absent, it took the value of 0. This method procedures and Monte Carlo simulation were performed in quantified the number of 1-s bins of a given sound that RStudio (version 4.3.0, 2023-04-21). co-occurred with the 1-s bins of a given behaviour or an external event. Only sounds occurring during an identified behaviour or an external event were considered. RESULTS To assess if co-occurrences were significantly different from chance, we performed a Monte Carlo simulation on The behavioural contexts of underwater sound production of the percentage of each co-occurrence (i.e. the number of 23 juvenile green turtles were investigated. From the dataset 1-s bins of a given sound co-occurring with a behaviour of recorded tag data, 20 666 sound 1-s bins (representing or an external event, divided by the total of 1-s bins of this approximately 6 h of sound recordings), 746 382 behaviour sound). Each co-occurrence was simulated 10 000 times 1-s bins (~ 207 h) and 92 549 external event 1-s bins (~ 26 by randomly assigning 1-s bins of sounds with behaviours h) were identified (see Table S2 in the Supplementary Infor - and external events, while maintaining the same total num- mation for the full dataset). Individual produce sounds for ber of 1-s bins for each sound. To obtain an empirical P 0.15–14.14% of the time they were recorded. value from Monte Carlo simulation, we use the formula (r + 1)/(n + 1), where r is the number of these replicates Sound production that produce a test statistic greater than or equal to that calculated for the actual data and n is the number of repli- Among the eleven recorded sounds, the rumble was the most cate samples that have been simulated (n = 10 000) (Davi- produced. It accounted for 70.5% of total sound production son and Hinkley 1997). Monte Carlo simulations estimate (Table 2; Fig. 1a). The remaining 29.5% of the total sound significance but do not measure it (North et al. 2002). production was distributed among the other ten sound cat- Since a sound can occur during 1-s bins of two different egories. Among these, the pulse and the squeak categories external events, it was included in both external events as were the most represented, each accounting for 13.3% and they are not mutually exclusive (e.g. boat noise can occur in 13% of the total sound production, respectively (Table 2; the presence of a human or a conspecific) and it is therefore Fig.  1b). In contrast, the FMS category represented only Fig. 1 Pie charts of (a) total sound production and (b) sound production without rumble (the remaining 29.5% of total sounds production). The pulse category includes Mono, Doublet, Triplet, Multipulse and Toc. The LAC category includes Croak and Rumble. The Squeak category includes Short and Long Squeak 25 Page 6 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Table 3 Descriptive data on Feeding Gliding Resting Scratching Swimming Surface activity individuals with behaviour budget (for a total of 746 382 Number of individuals 22 23 23 21 23 23 behaviour 1-s bins) Number of 1-s bins 41 861 17 521 368 847 18 178 299 975 47 659 0.8% of the total sound production (Table 2; Fig. 1b). The LAC category (including the croak and the rumble) was the most produced sound category, with rumble contributing to the large majority of it. Additionally, 99.2% of rumbles 1-s bins were recorded after sunset. Behavioural Budget and External Event occurrences Among the six defined behaviours, resting and swimming were the most observed, each accounting for 49.4% and 37.8% of total behaviour budget, respectively (Table  3; Fig. 2). Among the three external events, the boat noise was the most observed (Table 4). The interaction of the tagged turtle with a conspecific was much more frequent than with human (Table 4). Co‑Occurrences of Sounds with Behaviours Ten of the eleven types of sound described (excluding FMS) Fig. 2 Pie chart of total behavioural budget. As turtles were always were all generally produced during resting and swimming swimming during surface activity, its budget was included in the swimming behaviour. It accounted for 6.4% of the total behavioural (Fig. 3a, b, d, e). The five sound types of the pulse category budget and always co-occurred with swimming and the rumble were mostly produced during resting (all of them are significantly different from chance with p < 0.05, Table 5a), with 74.9% of rumbles were produced during rest- Table 4 Descriptive data on individuals with external event produc- tion (for a total of 92 549 external event 1-s bins) ing (p = 0.0001, Table 5a). Rumbles were mostly recorded in three individuals whose showed an intense rumble produc- Human Conspecific Boat noise tion after sunset, mainly during resting (Fig. 4a, b): 881 1-s Number of individuals 15 22 23 bins from 20:11 to 21:27, 6 753 1-s bins from 18:44 to 21:41 Number of 1-s bins 955 14 765 76 829 and 6 783 1-s bins from 18:41 to 21:31. However, only 12 out of 23 individuals were recorded after sunset. The grunt and the two sound types of the squeak category were mostly produced during swimming (all of them are Co‑occurrence of sounds with external events significantly different from chance with p < 0.001, Table 5a), with 68.9% of grunts were produced during swimming Very few sounds were heard during external event-driven (p = 0.0001, Table 5a). contexts except during boat noise (mono and croak were The FMS was quite rare (171 1-s bins recorded in produced significantly differently from chance with total on 11 individuals, Table 2) as well as the scratching p < 0.01, Table  5b). Only FMS and long squeak were behaviour (accounting for 2.4% of the total behavioural produced in the presence of humans (p < 0.05, Table 5b), budget, Fig. 2). However, the FMS was highly associated whereas eight sound types were produced in the presence with this scratching behaviour (Fig.  3c), with 89.5% of of conspecifics in the camera’s field of view (four sound FMS were produced during scratching (11 individuals, types, including grunt, were produced significantly dif- p = 0.0001, Table 5a). ferently from chance with p < 0.05, Table 5b). However, Behavioral Ecology and Sociobiology (2025) 79:25 Page 7 of 14 25 Fig. 3 Barplots of total co-occurrences of (a) pulses, (b) LAC, (c) FMS, (d) grunts and (e) squeaks with juvenile green turtle behaviours the long squeak production in the presence of human DISCUSSION was significantly different from chance (p = 0.0001, Table 5b). This is the sound heard the most during such This study provides new knowledge on the acoustic behav- event (10 1-s bins). iour of green sea turtles based on a multi-year research program investigation focusing on immature green turtle 25 Page 8 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Table 5 Co-occurrence of sounds with (a) behaviours and (b) exter- events. A Monte Carlo simulation was performed to assess if the per- nal events. For each cell, the first line corresponds to the number of centage of a given co-occurrence was significantly different from the individuals in which the co-occurrence happened (n ind), the second one obtained randomly (* indicate p < 0.05, ** indicate p < 0.01, *** line is the number of 1-s bins of sounds co-occurring with (a) behav- indicate p < 0.001; see table S3 in the supplementary information for iours or (b) external events, and the third line is the percentages of p-values) 1-s bins of sounds, co-occurring with (a) behaviours or (b) external (a) Feeding Gliding Resting Scratching Swimming Surface activity Pulse Mono n ind 1 5 12 2 17 5 Total of 1-s bins 1 15 190 2 133 9 1-s bins rate 0.3% 4.4% ** 55.7% ** 0.6% 39% 2.6% Doublet n ind 1 3 15 1 17 4 Total of 1-s bins 1 7 256 2 169 11 1-s bins rate 0.2% 1.6% 58.9% *** 0.5% 38.9% 2.5% Triplet n ind 2 4 13 1 15 8 Total of 1-s bins 2 12 522 2 287 20 1-s bins rate 0.2% 1.5% 63.3% *** 0.2% 34.8% 2.4% Multipulse n ind No co-occur- 1 3 No co-occur- 12 4 Total of 1-s bins rence 2 57 rence 35 5 1-s bins rate 2.1% 60.6% * 37.2% 5.3% Toc n ind 5 4 15 3 15 2 Total of 1-s bins 22 32 636 73 261 2 1-s bins rate 2.1% 3.1% 62.1% *** 7.1% *** 25.5% 0.2% LAC Croak n ind 2 1 10 2 13 5 Total of 1-s bins 4 2 94 15 70 6 1-s bins rate 2.2% 1.1% 50.8% 8.1% *** 37.8% 3.2% Rumble n ind 5 7 18 2 14 3 Total of 1-s bins 564 195 10 911 21 2869 236 1-s bins rate 3.9% 1.3% 74.9% *** 0.1% 19.7% 1.6% FMS n ind 3 No co-occur- 2 11 6 No co-occurrence Total of 1-s bins 4 rence 5 153 9 1-s bins rate 2.3% 2.9% 89.5% *** 5.3% Grunt n ind 8 5 17 5 20 4 Total of 1-s bins 25 11 63 9 239 24 1-s bins rate 7.2% 3.2% 18.2% 2.6% 68.9% *** 6.9% Squeak Short Squeak n ind 2 9 10 2 15 11 Total of 1-s bins 6 83 447 2 558 29 1-s bins rate 0.5% 7.6% *** 40.8% 0.2% 50.9% *** 2.6% Long Squeak n ind 4 7 6 2 16 11 Total of 1-s bins 19 185 653 7 724 30 1-s bins rate 1.2% 11.6% *** 41.1% 0.4% 45.6% *** 1.9% (b) Human Conspecific Boat noise Pulse Mono n ind No co-occur- 3 8 Total of 1-s bins rence 17 51  1-s bins rate 5 % *** 15 % ** Doublet n ind No co-occur- 2 8 Total of 1-s bins rence 8 54 1-s bins rate 1.8 % 12.4 % Triplet n ind No co-occur- 2 9 Total of 1-s bins rence 7 88 1-s bins rate 0.8 % 10.7 % Multipulse n ind No co-occur- 2 4 Total of 1-s bins rence 4 14 1-s bins rate 4.3 % * 14.9 % Toc n ind No co-occur- 3 4 Total of 1-s bins rence 14 42 1-s bins rate 1.4 % 4.1 % Behavioral Ecology and Sociobiology (2025) 79:25 Page 9 of 14 25 Table 5 (continued) LAC Croak n ind No co-occur- 2 6 Total of 1-s bins rence 10 31 1-s bins rate 5.4 % ** 16.8 % ** Rumble n ind No co-occur- 2 6 Total of 1-s bins rence 3 45 1-s bins rate 0 % 0.3 % FMS n ind 1 No co-occur- 4 Total of 1-s bins 1 rence 17 1-s bins rate 0.6 % * 9.9 % Grunt n ind No co-occur- 4 6 Total of 1-s bins rence 11 21 1-s bins rate 3.2% * 6.1 % Squeak Short Squeak n ind No co-occur- No co-occur- 7 Total of 1-s bins rence rence 111 1-s bins rate 10.1 % Long Squeak n ind 3 No co-occur- 6 Total of 1-s bins 10 rence 73 1-s bins rate 0.6 % *** 4.6 % Significant values (marked with one, two or three *) are in bold Fig. 4 Boxplots of occurrences of (a) resting behaviour and (b) rum- twenty minutes in each time slot, and the n corresponds to the num- bles produced at each hour of the day. Boxes indicate the inter quar- ber of sampled* individuals resting (a) or producing a rumble (b) at tile range, with the central line depicting the median and the whisk- least once in each time slot. No sea turtles were recorded before 8am ers extending to min and max values and outliers. For each hour, the and after 3am. * Individuals with a tag that was recording data N corresponds to the number of individuals recorded for more than 25 Page 10 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 population in the Lesser Antilles. The investigations on the Grunts were found to be commonly produced during rest- co-occurrences of sounds and behaviours of juveniles at ing and mostly during swimming. However, when watch- their foraging grounds are key to understanding the potential ing the behaviour of the green turtles producing grunts, importance of acoustic communication in this endangered we found a stereotyped retreat head movement coinciding species. Altogether, these findings offer new and innovative with the production of grunts, which suggests a potential perspectives for improving sea turtle conservation. link between this sound and a specific behavioural response Our first finding revealed a highly variable production in green turtles. Indeed, few grunts were produced in the rate within the sound repertoire of the green turtle. Doublet, presence of conspecifics (p = 0.0443) during an intimida- grunt and rumble were the sounds most commonly produced tion interaction (i.e. two conspecifics swimming around among juvenile green turtles (n ≥ 20 individuals). However, each other, which is considered swimming behaviour in the doublets and grunts were produced far less than the rumble. results, p = 0.0001). Thus, such head movement could be a Indeed, the doublet and the grunt accounted for 2.1% and visual agonistic component analogous to aggressive threat 1.7% of the total sound production, respectively, while the signals observed in others species, such as lizard waving rumble accounted for 70.5% of the total sound production. their tails at approaching predators (Bradbury and Vehren- In contrast, the FMS and the Multipulse were the sounds camp 2011). Given that grunts were also produced when produced least commonly among individuals (n ≤ 12 indi- sea turtles exhibited a vigilance posture (head up, look- viduals) and accounted for 0.8% and 0.5% of the total sound ing around, leaning on its front legs) while swimming or production, respectively. The croak was also little produced, feeding or when approaching conspecifics, we suggest that accounting for 0.9% of the total sound production. However, grunts could function as warning signals, playing a role in since the tags recorded for short amounts of time (i.e. less intra-specific acoustic communication among juvenile green than 24 h at a time and not on several successive days), turtles. it is likely that all sounds, behaviours and external events We previously assessed that the squeak might be a good could not be sampled for each individual within a recording candidate for intra-specific communication due to its indi- session. vidual stereotypy (i.e., individual-specific) (Charrier et al. This study provided the first investigation focusing on 2022). Our recordings show that long squeaks were recorded the relationship between behaviours, external events and for three individuals during human avoidance events. Two underwater sounds produced by juvenile green turtles. The produced long squeaks just after being released from the pulse, the LAC, the grunt and the squeak categories were tagging boat and one produced a long squeak while swim- generally produced while juvenile green turtles were rest- ming away from three swimmers. Long squeaks were also ing or swimming. These prevailing co-occurrences can be observed on one other individual after the sunset. The data attributed to their predominant activity patterns during the from the hydrophone, the pressure logger and the 3D-accel- tag deployment which was mainly resting and swimming. As erometer showed that the green turtle was about to surface, seven out of the eleven described sounds types (mono, dou- stopped, then dove rapidly after producing such sound, blet, triplet, multipulse, toc, croak and short squeak), were suggesting an avoidance behaviour. The observation of not produced in a specific context, it is not possible to sug- long squeaks produced by green turtles during avoidance gest any biological function or to conclude on the absence behaviour (which is considered swimming behaviour in the of function for these seven sounds types. Thus, we cannot results, p = 0.0001) in the presence of humans (p = 0.0001) confirm whether these sound types are used for communica- provides interesting insights into the potential link between tion. Furthermore, as no sound was recorded during physical this sound in response to perceived threats and anti-predator interactions between conspecifics (affiliative or agonistic), or avoidance behaviour. The limitations of the camera’s field we suggest that sound may not be the medium used for direct of view (100°) highlight the possibility that important con- social interaction, as sight or smell remain effective at close textual cues, such as the presence of conspecifics or another range (Bartol and Musick 2002). However, we found some potential threats (animal or a human), may not have been sounds produced by a small number of turtles while in pres- captured, thus complicating the interpretation of behav- ence of conspecifics (without physical contact). ioural responses. Indeed, we often saw that the turtle was Among the context-specific sound productions, the FMS vigilant and looking around, but we could not explain why was mostly produced during scratching behaviour (Fig. 3c). it remained alert. The long squeak may thus constitute a Both FMS and scratching are individually rare compare to first evidence of an alarm acoustic signal in juvenile green others sounds and behaviours (Tables 2 and 3). However, the turtles, used to alert conspecifics from a threat. While evi- average percentage of co-occurrence is very high (83.9%, dence of alarm acoustic signals has been shown in other Table 5a). The FMS was generally produced when sea turtles non-avian reptile species (e.g. leopard lizard Gambelia wisli- seem to be alone and undisturbed. However, we are not able zenii, Wever et al. 1966; fossorial snakes, Young et al. 2013), to assess any biological function for this sound. confirmation of the alarm function of the long squeak in Behavioral Ecology and Sociobiology (2025) 79:25 Page 11 of 14 25 green turtles would require experimental validation using (2012) reported that loggerhead turtles dived immediately playback experiments. following an airgun shot, while Weir (2007) reported that We recorded long sequences of rumbles after sunset and 83% of sea turtles (including olive ridley, leatherback and during resting behaviour. The overlapping and varied ampli- loggerhead turtles) continued to bask at the surface during tude levels of these rumbles suggest that several individuals airguns exposure and as the vessel and towed equipment in the vicinity may contribute to these long-lasting acoustic moved past. The variability of behavioural responses to interactions, potentially engaging in some form of coordi- a noisy event is highly complex in many species includ- nated sound production. It may thus constitute first evidence ing non-avian reptiles. For example, freezing is a typi- of a social acoustic communication among juvenile green cal stress response in non-avian reptiles, as in the Eastern turtles during night-time resting periods. Similar production, blue tongued lizard Tiliqua scincoides when exposed to called choruses, are observed in other taxa like birds, insects the noise of mining machinery (Mancera Alarcon 2016). frogs and fishes (Farina and Ceraulo 2017). It highlights the Given impact of anthropogenic noise on non-avian reptile potential for social communication and coordination among behaviour remains relatively understudied compared to green turtles during night-time resting periods. The only other taxa (Simmons and Narins 2018; Jerem and Mathews example is the Travancore tortoise Indotestudo travancorica, 2021), the interpretation of the results regarding the effect in which several individuals called together, with individual of boat noise on the behaviour of juvenile green turtles is sound productions appearing to be regularly spaced, at night constrained. Further research is required to compare the (Campbell and Evans 1972). While chorusing behaviour is behaviour of green turtles occurring in areas with low and well documented among insects, fishes, frogs and birds and high-levels of boat traffic. has associated with diverse ecological functions (e.g. ener- The four sounds types, rumble, FMS, grunt and long getic and behavioural matters in birds, Farina and Ceraulo squeak were the only ones produced by juvenile green tur- 2017), its occurrence in non-avian reptiles, particularly Che- tles in specific behavioural contexts. Such findings could lonians, remains relatively unknown. potentially contribute to sea turtle conservation. Indeed, our Although we did not examine the sound level of the boat findings highlight intra-specific social interactions, but also noise we recorded, our findings show there is no specific alert or vigilance sounds. It provides a strong baseline to test sound produced, nor any behaviour that stopped during free-ranging green turtle with playback experiments using boat noise events. This suggests that juvenile green turtles the long squeak, the grunt or the rumble to assess if we can may not alter their acoustic behaviour in direct response elicit a behavioural reaction of the tested individuals. It has to boat activities, as it was reported in diamondback ter- been shown that some non-avian reptiles are sensitive to rapins Malaclemys terrapin (Lester et al. 2013). However, heterospecific calls and can adapt their behavioural response the absence of change in behaviour through their sound from threat signals (marine iguana Amblyrhynchus crista- production or behaviour does not mean that boat activi- tus, Vitousek et al. 2007; e.g. brown anoles Anolis sagrei, ties do not induce physiological stress or cause hearing Cantwell and Forrest 2013). Some of these sounds could be loss (e.g. red-eared slider Trachemys scripta elegans, Salas used in deterrent systems aiming to reduce by-catch of sea et al. 2023). Although green turtles have not altered their turtles by preventing them from being entangled in fishing sound production in response to boat noise, they appear nets (Chevallier et al. 2024). to surface less often to breathe (LM, unpubl. data). This Indeed, the use of acoustic pingers on harbour porpoise alteration in surfacing behaviour may indicate a poten- Phocoena phocoena for instance has shown that acoustic tial physiological response to the presence of boats and signals can reduce the occurrence of by-catches (Kraus et al. associated noise, suggesting a possible impact on their 1997; Trippel et al. 1999; Gearin et al. 2000). As part of respiratory patterns or diving behaviour. A comparable the TOPASE program (Martinique-Guadeloupe) aiming to behavioural response has previously been observed in a reduce by-catch of sea turtles in fishing nets, recent investi- juvenile green turtle, wherein it appears to remain station- gations based on the findings of this present study has been ary on or near the sea floor when ships pass nearby (Tyson carried out. Selected green turtle sounds (squeaks and rum- et al. 2017). Most studies of sea turtle responses to boat bles) were used in playback experiments on free-ranging noise have focused on exposure to high-intensity seismic green turtles during their foraging activities in Martinique airguns in a closed or semi-closed environment, which (Chevallier et al. 2024). Encouragingly, the playback tests limits the ability to assess the behaviour of free-rang- showed that a majority of green turtles responded to these ing turtles exposed to different boat noises (O’Hara and sounds by exhibiting behavioural alertness. Further investi- Wilcox 1990; Moein et al. 1994; McCauley et al. 2000). gations are now required to confirm such findings. Specifi- The rare studies on free-ranging turtles exposed to seis- cally, attention can be directed towards examining specific mic airgun surveys reported highly variable behavioural sounds such as the grunt and the long squeak to confirm responses of turtles. Indeed, DeRuiter and Larbi Doukara their role in avoiding danger from conspecifics or humans 25 Page 12 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 the stress associated with capture. The use of suction cups instead of and to explore the biological function of the rumble in social glue makes the attachment system much less invasive. The slightly interactions and social grouping pattern during night-time. positive buoyancy and hydrodynamics of the camera ensure that the Finally, to gain a deeper understanding of the biological turtle’s movements are not constrained when diving or surfacing. The function of these sounds, it would be valuable to perform automatic release of the device after a maximum of two days avoids the stress of a second capture. similar recordings using animal-borne tags on adult green turtles. By comparing the sound repertoire of juvenile to Open Access This article is licensed under a Creative Commons Attri- those of adult green turtles, we could potentially highlight bution 4.0 International License, which permits use, sharing, adapta- any variations in acoustic parameters and reveal ontogenetic tion, distribution and reproduction in any medium or format, as long information regarding the development of sound-producing as you give appropriate credit to the original author(s) and the source, organs. Moreover, this approach could reveal new sound pro- provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are ductions used only during the breeding period, providing included in the article’s Creative Commons licence, unless indicated further insights into the acoustic behaviour of adult green otherwise in a credit line to the material. If material is not included in turtles. the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will Supplementary Information The online version contains supplemen- need to obtain permission directly from the copyright holder. To view a tary material available at https://doi. or g/10. 1007/ s00265- 025- 03561-z . copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Acknowledgements This study was carried out within the framework of the Plan National d’Action Tortues Marines Antilles. We thank the two anonymous reviewers for their helpful comments. References Author contributions DC, IC and LM designed research; DC, NA, OB, LJ, NL, FL, PL, ML, JM and SR performed data collection; LM and Bartol SM, Ketten DR (2006) Turtle and tuna hearing. In: Swimmer CH designed the code of the Dummy Coding method; LM, IC, LJ and Y, Brill RW (eds) Sea turtle and pelagic fish sensory biology: DC analysed data; DC acquired funding; LM wrote the original version Developing techniques to reduce sea turtle bycatch in longline of the paper; IC, DC and CH edited the paper. fisheries. NOAA Tech. Mem. NMFS-PIFSC-7. National Ocean and Atmospheric Administration (NOAA), US Department of Funding Open access funding provided by Université des Antilles. Commerce, pp 98–105. http:// www. pifsc. noaa. gov/ tech/ NOAA_ The Program BEPHYTES of the Centre National de la Recherche Sci-Tech_ Memo_ PIFSC_7. pdf entifique (CNRS) was co-funded by the FEDER Martinique (Fonds Bartol SM, Musick JA (2002) Sensory biology of sea turtles. In: Lutz Européen de Développement Régional PO/FEDER/FSE 2014– PL, Musick JA, Wyneken J (eds) The biology of sea turtles. CRC 2020, Conventions 246239), Collectivité Territoriale de Martinique Press, Boca Raton, pp 79–102 (CTM, convention 258342), the Direction de l′Environnement, de Bonola M, Girondot M, Robin J-P et al (2019) Fine scale geographic l′Aménagement et du Logement (DEAL, Martinique, France (Conven- residence and annual primary production drive body condition tion N°2017/164894), the Office De l′Eau (ODE) Martinique, France of wild immature green turtles (Chelonia mydas) in Martinique (Convention n°180126) and the Office Français de la Biodiversité Island (Lesser antilles). Biol Open 8:bio048058. https:// doi. org/ (OFB, Parc Naturel Marin de Martinique, OFB-23-0563), the town of 10. 1242/ bio. 048058 Les Anses d’Arlet and ANSLO’S. The Program TOPASE of the CNRS Bradbury JW, Vehrencamp SL (2011) Principles of animal communica- was funded by Fonds Européen pour les affaires maritimes et la pêche tion, 2nd edn. Sinauer Associates, Sunderland, MA (FEAMP), the Minister of Agriculture and FranceAgriMer. Léo Mau- Britton ARC (2001) Review and classification of call types of juvenile court PhD scholarship was supported by the Collectivité Territoriale crocodilians and factors affecting distress calls. Crocodilian Biol of Martinique (CTM). Evol 364:364–377 Campbell HW, Evans WE (1972) Observations on the vocal behavior Data availability All datasets generated or analysed during this study of chelonians. Herpetologica 28:277–280 are included in this published article and its supplementary informa- Cantwell LR, Forrest TG (2013) Response of Anolis sagrei to acoustic tion files. calls from predatory and nonpredatory birds. J Herpetol 47:293– 298. https:// doi. org/ 10. 1670/ 11- 184 Carr AF (1952) Handbook of turtles. Comstock Publishing Associates, Declarations Ithaca, NY Chambault P, de Thoisy B, Huguin M et al (2018) Connecting paths Conflict of interest The authors declare no conflict of interest. between juvenile and adult habitats in the Atlantic green turtle using genetics and satellite tracking. Ecol Evol 8:12790–12802. Ethics approval Fieldwork was performed in accordance with the https:// doi. org/ 10. 1002/ ece3. 4708 French legal and ethical requirements. Specifically, the protocol was Charrier I, Jeantet L, Maucourt L, Régis S, Lecerf N, Benhalilou A, approved by the Conseil National de la Protection de la Nature and Chevallier D (2022) First evidence of underwater vocalizations in the French Ministry for Ecology (permit numbers: 201710-0005 and green sea turtles Chelonia mydas. Endanger Species Res 48:31– R02-2020-08-10-006) and followed the recommendations of the Police 41. https:// doi. org/ 10. 3354/ esr01 185 Prefecture of Martinique. Fieldwork was carried out under the certifica- Chevallier D, Maucourt L, Charrier I et al (2024) The response of tion of DC (prefectural authorisations’ owner) under strict compliance sea turtles to vocalizations opens new perspectives to reduce of the Police of Martinique’s recommendations to minimize animal dis- their bycatch. Sci Rep 14:16519. https:// doi. or g/ 10. 1038/ turbance. Indeed, the capture, although stressful for the animal, respects s41598- 024- 67501-z the safety conditions for the divers as well as for the sea turtle. The Cook SL, Forrest TG (2005) Sounds produced by nesting leatherback handling time on the boat did not exceed 10 min in order to minimise sea turtles (Dermochelys coriacea). Herpetol Rev 36:387–390 Behavioral Ecology and Sociobiology (2025) 79:25 Page 13 of 14 25 Davison AC, Hinkley DV (1997) Bootstrap methods and their applica- vertebrates. Nat Commun 13:6089. h t t p s : / / d o i . o r g / 1 0 . 1 0 3 8 / tion. Cambridge University Press, Cambridges41467- 022- 33741-8 DeRuiter SL, Larbi Doukara K (2012) Loggerhead turtles dive in Ketten DR (2008) Underwater ears and the physiology of impacts: response to airgun sound exposure. Endanger Species Res 16:55– comparative liability for hearing loss in sea turtles, birds, and 63. https:// doi. org/ 10. 3354/ esr00 396 mammals. Bioacoustics 17:312–315. https:// doi. or g/ 10. 1080/ Farina A, Ceraulo M (2017) The acoustic chorus and its ecological 09524 622. 2008. 97538 60 significance. In: Farina A, Gage SH (eds) Ecoacoustics: the eco- Kraus SD, Read AJ, Solow A, Baldwin K, Spradlin T, Anderson E, logical role of sounds. Wiley, Hoboken, NJ, pp 81–94 Williamson J (1997) Acoustic alarms reduce porpoise mortality. Ferrara CR, Vogt RC, Sousa-Lima RS (2013) Turtle vocalizations as Nature 388:525. https:// doi. org/ 10. 1038/ 41451 the first evidence of posthatching parental care in chelonians. J Lavender AL, Bartol SM, Bartol IK (2014) Ontogenetic investiga- Comp Psychol 127:24–32. https:// doi. org/ 10. 1037/ a0029 656 tion of underwater hearing capabilities in loggerhead sea tur- Ferrara CR, Mortimer JA, Vogt RC (2014a) First evidence that hatch- tles (Caretta caretta) using a dual testing approach. J Exp Biol lings of Chelonia mydas emit sounds. Copeia 2014:245–247. 217:2580–2589. https:// doi. org/ 10. 1242/ jeb. 096651 https:// doi. org/ 10. 1643/ CE- 13- 087 Lelong P, Besnard A, Girondot M et al (2024) Demography of endan- Ferrara CR, Vogt RC, Giles JC, Kuchling G (2014b) Chelonian vocal gered juvenile green turtles in face of environmental changes: communication. In: Witzany G (ed) Biocommunication of ani- 10 years of capture-mark-recapture efforts in Martinique. Biol mals. Springer, Dordrecht, pp 261–274 Conserv 291:110471. https:// doi. org/ 10. 1016/J. BIOCON. 2024. Ferrara CR, Vogt RC, Harfush MR, Sousa-Lima RS, Albavera E, 110471 Tavera A (2014) First evidence of leatherback turtle (Dermo- Lester LA, Avery HW, Harrison AS, Standora EA (2013) Recreational chelys coriacea) embryos and hatchlings emitting sounds. boats and turtles: behavioral mismatches result in high rates of Chelonian Conserv Biol 13:110–114. https:// doi. org/ 10. 2744/ injury. PLoS ONE 8:e82370. https://doi. or g/10. 1371/ jour nal. pone. CCB- 1045.100823 70 Ferrara CR, Vogt RC, Sousa-Lima RS, Tardio BMR, Bernardes VCD Lutcavage M, Lutz PL (1996) Diving physiology. In: Lutz PL, Musick (2014) Sound communication and social behavior in an amazo- JA (eds) The biology of sea turtles. CRC Press, Boca Raton, pp nian river turtle (Podocnemis expansa). Herpetologica 70:149– 277–296 156. https://d oi.o rg/1 0.1 655/H ERPET OLOGI CA-D-1 3-0 0050R 2 Mancera Alarcon K (2016) Effects of anthropogenic noise on the Ferrara CR, Vogt RC, Eisemberg CC, Doody JS (2017) First evidence behaviour, physiological traits and welfare of two animal mod- of the pig-nosed turtle (Carettochelys insculpta) vocalizing under- els: wild mice (Mus musculus) and Eastern blue tongued lizard water. Am Soc Ichthyol Herpetol 105:29–32. https:// doi. org/ 10. (Tiliqua scincoides). PhD thesis, The University of Queensland 1643/ CE- 16- 407 McCauley RD, Fewtrell J, Duncan AJ, Jenner C, Jenner M-N, Penrose Ferrara CR, Vogt RC, Sousa-Lima RS, Lenz A, Morales-Mávil JE JD, Prince RIT, Adhitya A, Murdoch J, McCabe K (2000) Marine (2019) Sound communication in embryos and hatchlings of Lepi- seismic surveys — a study of environmental implications. APPEA dochelys kempii. Chelonian Conserv Biol 18:279–283. https://doi. J 40:692. https:// doi. org/ 10. 1071/ aj990 48 org/ 10. 2744/ CCB- 1386.1 McKenna LN, Paladino FV, Tomillo PS, Robinson NJ (2019) Do Garrick LD, Garrick RA (1978) Temperature influences on hatch- sea turtles vocalize to synchronize hatching or nest emergence? ling Caiman crocodilus distress calls. Physiol Zool 51:105–113. Copeia 107:120–123. https:// doi. org/ 10. 1643/ CE- 18- 069 https:// doi. org/ 10. 1086/ physz ool. 51.2. 30157 859 Moein SE, Musick JA, Keinath JA, Barnard DE, Lenhardt M, George R Gearin PJ, Gosho ME, Laake JL, Cooke L, DeLong R, Hughes KM (1994) Evaluation of seismic sources for repelling sea turtles from (2000) Experimental testing of acoustic alarms (pingers) to reduce hopper dredges. Virginia Institute of Marine Science, College of bycatch of harbour porpoise, Phocoena phocoena, in the state William & Mary, Gloucester Point, VA of Washington. J Cetacean Res Manag 2:1–9. https:// doi. org/ 10. Monteiro CC, Carmo HMA, Santos AJB, Corso G, Sousa-Lima RS 47536/ jcrm. v2i1. 483 (2019) First record of bioacoustic emission in embryos and hatch- Giles JC, Davis JA, McCauley RD, Kuchling G (2009) Voice of the lings of hawksbill sea turtles (Eretmochelys imbricata). Chelonian turtle: the underwater acoustic repertoire of the long-necked fresh- Conserv Biol 18:273–278. https:// doi. org/ 10. 2744/ CCB- 1382.1 water turtle, Chelodina oblonga. J Acoust Soc Am 126:434–443. Mrosovsky N (1972) Spectrographs of the sounds of leatherback tur- https:// doi. org/ 10. 1121/1. 31482 09 tles. Herpetologica 28:256–258 Higgs DM, Brittan-Powell EF, Soares D, Souza M, Carr C, Dooling R, Nishizawa H, Hashimoto Y, Rusli MU, Ichikawa K, Joseph J (2021) Popper A (2002) Amphibious auditory responses of the American Sensing underground activity: diel digging activity pattern during alligator (Alligator mississipiensis). J Comp Physiol A 188:217– nest escape by sea turtle hatchlings. Anim Behav 177:1–8. https:// 223. https:// doi. org/ 10. 1007/ s00359- 002- 0296-8doi. org/ 10. 1016/j. anbeh av. 2021. 04. 013 Jeantet L, Planas-Bielsa V, Benhamou S et al (2020) Behavioural infer- Nivière M, Chambault P, Pérez T et al (2018) Identification of marine ence from signal processing using animal-borne multi-sensor key areas across the Caribbean to ensure the conservation of the loggers: a novel solution to extend the knowledge of sea turtle critically endangered hawksbill turtle. Biol Conserv 223:170–180. ecology. R Soc Open Sci 7:200139. https:// doi. org/ 10. 1098/ rsos. https:// doi. org/ 10. 1016/j. biocon. 2018. 05. 002 200139 North BV, Curtis D, Sham PC (2002) A note on the calculation of Jeantet L, Vigon V, Geiger S, Chevallier D (2021) Fully convolutional empirical P values from Monte Carlo procedures. Am J Hum neural network: a solution to infer animal behaviours from multi- Genet 71:439–441. https:// doi. org/ 10. 1086/ 341527 sensor data. Ecol Modell 450:109555. https:// doi. org/ 10. 1016/j. O’Hara J, Wilcox JR (1990) Avoidance responses of loggerhead turtles, ecolm odel. 2021. 109555 Caretta caretta, to low frequency sound. Copeia 1990:564–567. Jerem P, Mathews F (2021) Trends and knowledge gaps in field https:// doi. org/ 10. 2307/ 14463 62 research investigating effects of anthropogenic noise. Conserv Papale E, Prakash S, Singh S, Batibasaga A, Buscaino G, Piovano S Biol 35:115–129. https:// doi. org/ 10. 1111/ cobi. 13510 (2020) Soundscape of green turtle foraging habitats in Fiji, South Jorgewich-Cohen G, Townsend SW, Padovese LR et al (2022) Com- Pacific. PLoS ONE 15:e0236628. https:// doi. org/ 10. 1371/ journ mon evolutionary origin of acoustic communication in choanate al. pone. 02366 28 25 Page 14 of 14 Behavioral Ecology and Sociobiology (2025) 79:25 Piniak WED, Eckert SA, Harms CA, Stringer EM (2012) Underwa- Tyson RB, Piniak WED, Domit C, Mann D, Hall M, Nowacek DP, ter hearing sensitivity of the leatherback sea turtle (Dermochelys Fuentes MMPB (2017) Novel bio-logging tool for studying coriacea): assessing the potential effect of anthropogenic noise. fine-scale behaviors of marine turtles in response to sound. U.S. Dept. of the Interior, Bureau of Ocean Energy Management, Front Mar Sci 4:219. https:// doi. or g/ 10. 3389/ fmars. 2017. Headquarters, Herndon, VA 00219 Piniak WED, Mann DA, Harms CA, Jones TT, Eckert SA (2016) Vergne AL, Mathevon N (2008) Crocodile egg sounds signal hatch- Hearing in the juvenile green sea turtle (Chelonia mydas): a com- ing time. Curr Biol 18:513–514. https://doi. or g/10. 1016/j. cub. parison of underwater and aerial hearing using auditory evoked 2008. 04. 011 potentials. PLoS ONE 11:e0159711. https://doi. or g/10. 1371/ jour n Vergne AL, Pritz MB, Mathevon N (2009) Acoustic communication al. pone. 01597 11 in crocodilians: from behaviour to brain. Biol Rev 84:391–411. Russell AP, Bauer AM (2021) Vocalization by extant nonavian reptiles: https:// doi. org/ 10. 1111/j. 1469- 185X. 2009. 00079.x a synthetic overview of phonation and the vocal apparatus. Anat Vitousek MN, Adelman JS, Gregory NC, St Clair JJH (2007) Hetero- Rec 304:1478–1528. https:// doi. org/ 10. 1002/ ar. 24553 specific alarm call recognition in a non-vocal reptile. Biol Lett Salas AK, Capuano AM, Harms CA, Piniak WED, Mooney TA (2023) 3:632–634. https:// doi. org/ 10. 1098/ rsbl. 2007. 0443 Temporary noise-induced underwater hearing loss in an aquatic Weir CR (2007) Observations of marine turtles in relation to seismic turtle (Trachemys scripta elegans). J Acoust Soc Am 154:1003– airgun sound off Angola. Mar Turt Newsl 116:17–20 1017. https:// doi. org/ 10. 1121/ 10. 00205 88 Wever EG, Hepp-Reymond MC, Vernon JA (1966) Vocalization and Siegwalt F, Benhamou S, Girondot M et al (2020) High fidelity of sea hearing in the leopard lizard. P Natl Acad Sci USA 55:98–106. turtles to their foraging grounds revealed by satellite tracking and https:// doi. org/ 10. 1073/ pnas. 55.1. 98 capture-mark-recapture: new insights for the establishment of key Young BAA, Mathevon N, Tang Y (2013) Reptile auditory neuro- marine conservation areas. Biol Conserv 250:108742. https://doi. ethology: what do reptiles do with their hearing? In: Köppl C, org/ 10. 1016/j. biocon. 2020. 108742 Manley G, Popper A, Fay R (eds) Insights from Comparative Simmons AM, Narins PM (2018) Effects of anthropogenic noise Hearing Research, vol 49. Springer, New York, NY, pp 323–346. on amphibians and reptiles. In: Slabbekoorn H, Dooling R, https:// doi. org/ 10. 1007/ 2506_ 2013_ 30 Popper A, Fay R (eds) Effects of anthropogenic noise on animals, Springer Handbook of Auditory Research, vol 66. Publisher’s note Springer Nature remains neutral with regard to Springer, New York, pp 179–208. https:// doi. or g/ 10. 1007/ jurisdictional claims in published maps and institutional affiliations. 978-1- 4939- 8574-6_7 Trippel EA, Strong MB, Terhune JM, Conway JD (1999) Mitigation of harbour porpoise (Phocoena phocoena) by-catch in the gillnet fishery in the lower bay of fundy. Can J Fish Aquat Sci 56:113– 123. https:// doi. org/ 10. 1139/ cjfas- 56-1- 113

Journal

Behavioral Ecology and Sociobiology – Springer Journals

Published: Feb 1, 2025

Keywords: Behaviour; Chelonia mydas; Chelonians; Juveniles; Underwater sound production

There are no references for this article.