Sequences of a possible distress call of a juvenile gray whale found in a shallow lagoon within the Gulf of California

Gray whales (Eschrichtius robustus) use sound to communicate (Dahlheim, 1987), and some of these sounds have been associated with activities, such as socialization, feeding, and maternal care (López-Urbán et al., 2016). Gray whale sounds are produced in frequencies ranging from 30 to 3,000 Hz (Crane & Lashkari, 1996; Dahlheim, 1987; Dahlheim et al., 1984; Swartz, 2018), and have been reported throughout the species' distribution (Burnham et al., 2018; Guazzo et al., 2017; Haver et al., 2020; López-Urbán et al., 2016; Ollervides, 2001; Rice et al., 2021; Stafford et al., 2007). A broad classification of sound types that have been reported in literature can be found in Table 1. In general, a sound is named with an "S" plus a number if it is recorded at a breeding area, and a "M" plus a number if recorded at a migration route. However, some studies do not use a letter, using just numbers and subclasses (e.g., 1a, 1b), or others use a letter code very specific ("BMC" for those recorded in Bahia Magdalena Complex). A consensus on nomenclature for the species will be helpful to standardize naming of new calls. Most vocal sounds consist of pulses or sequences of pulses, and a few tonal calls, such as S3 and S9. Type 1 calls are considered the most commonly recorded call type in breeding lagoons (Charles, 2011; Dahlheim & Castellote, 2016; Ollervides, 2001), while type 3 calls have been reported more commonly during migration (Burnham et al., 2018; Crane & Lashkari, 1996; Cummings et al., 1968). A mix of more than one type of call, is also possible and has been documented in San Ignacio Lagoon (Frouin-Mouy et al., 2020; Laguna San Ignacio Ecosystem Science Program, n.d.; Ponce, et al. 2012; Seger et al., 2015). When a known signal is found at the same time in recurrent association with an unknown signal, it is assumed to be produced by the same individual. This is helpful to later identify these new signals when uttered as single calls. For instance, the northeastern Pacific blue whale (Balaenoptera musculus) C call has occasionally been found between A and B calls during songs (Clark & Fristrup, 1997; Stafford et al., 1999), and fin whale (Balaenoptera physalus) triplet calls were found associated with the typical doublet call during passive acoustic monitoring in the Gulf of California (Bonilla-Garzón, 2015; Širović et al., 2017). The present research reports sequences possibly produced by a juvenile gray whale that include a new tonal call type which in some instances is uttered together with a known S1 call. The Ensenada de La Paz is a shallow coastal lagoon connected through a natural channel to La Paz Bay in the Gulf of California, Baja California Sur, Mexico. Gray whales have been reported in La Paz Bay, with a few strandings inside Ensenada de La Paz (Flores-Ramírez et al., 1996; Urbán et al., 2022), with a report of a pair of gray whales stranding together in an area outside the entrance of the lagoon (Urbán et al., 2022). In March 2019, the BCS Stranding Network reported two juvenile gray whales (~9 m in length, <2 years old) which were observed in an advanced emaciation stage. One stranded inside the Ensenada de La Paz on March 3, and the other was sighted alive in the channel area on March 13. It is known that during migration individuals are segregated by age, sex, and reproductive condition (Swartz, 2018), where immature individuals of both sexes are the last cohort in the migration. Thus, it seems reasonable to find potential grouping of immature/juvenile individuals. As part of an ongoing bottlenose dolphin passive acoustic monitoring project, continuous recordings were collected with a moored buoy 350 m off the coast within the natural channel of Ensenada de La Paz from the afternoon of March 3 to the afternoon of March 5, 2019 (Figure 1). Visual observations were not obtained during the recording period. The buoy recorded signals with a bandwidth 0.02–22 kHz, gain 30 dB, using a Teledyne Benthos AQ-2000 piezoelectric hydrophone (0.001–10 kHz frequency response, sensitivity −201 dB 1 V/μPa). Recordings were resampled to 6,250 Hz sample rate, since gray whale calls are generally produced below 3 kHz (Crane & Lashkari, 1996; Dahlheim, 1987; Dahlheim et al., 1984; Swartz, 2018). Then, files were analyzed in the time-frequency domain using Audacity 3.0.1 (1,024 FFT points, Hann window 1,024 size, 50% overlap) in the frequency range resolution 0–3,000 Hz. The recordings showed tonal signals that did not match sounds from a database (maintained since 2017) of biotic, abiotic, and anthropogenic sounds recorded at different points of the lagoon. No dolphin whistles or echolocation clicks were detected at the same time of the tonal signal. Besides, through further analysis, pulsed S1 calls—the most commonly reported gray whale signal in literature—were found along the tonal signals. Spectrograms of calls recorded at the breeding lagoons, migratory route, and feeding grounds (Burnham et al., 2018; Charles, 2011; Dahlheim, 1987; Dahlheim & Castellote, 2016; Frouin-Mouy et al., 2020; Guazzo et al. 2017, 2019; López-Urbán et al., 2016; Ponce et al., 2012; Seger et al., 2015) were used to confirm putative gray whale calls detected in Ensenada de La Paz. The fact that the tonal signals were within the reported frequency range for the species, in some cases together with a known call (S1), and the sightings of an individual stranded the morning before the recordings and other whale sighted days after in the area, allowed us to assume that the signals were produced by a gray whale. Peak energy, peak frequency, maximum frequency, minimum frequency, and duration were measured from spectrograms of putative calls following a threshold energy limit standardization as previously proposed for Bryde's whale (Balaenoptera edeni) calls (Viloria-Gómora et al., 2021). The signal to noise cutting threshold was selected by calculating the difference between the peak energy of each of 100 calls selected randomly, and the immediate value of ambient noise, giving a 27-dB threshold. The frequency and time measurements were therefore made at 27 dB below the peak energy. Additionally, qualitative characteristics, such as modulation and sequences of calls were also observed. The S1 calls identified in the recordings consisted of one or several broadband pulses with a peak around 300–600 Hz (Dahlheim & Castellote, 2016; Frouin-Mouy et al., 2020; López-Urbán et al., 2016; Ponce et al., 2012). A comparison of the S1 calls in San Ignacio and the ones detected in Ensenada de La Paz can be observed in Figure 2. Contrary to the S1 signals of San Ignacio to those form Ensenada de La Paz, lower frequency components are not so strong. However, it could be expected considering that La Paz inlet is at most 10 m in depth, and lower frequency components travel less efficiently in very shallow areas (Urick, 1979) because the wavelength is greater than the water column. In some cases (84 times), the new tonal signal appeared on average 0.27 ± 0.04 s (mean ± SD) after a S1 call with similar intensity, and so it was considered as produced by the same individual, and therefore as a putative gray whale call. This mixed S1-tonal signal also appeared in sequences (Figure 3a,b). The tonal call mostly appeared without the S1 (Figure 3c), thus, the name S10 was assigned to keep a numerical continuity of the last new calls (S8 and S9) described in the San Ignacio Lagoon (López-Urbán et al., 2016). It was decided to maintain the nomenclature "S" plus number for calls in the breeding southernmost distribution of the species for now, until a future consensus in the rules of naming is reached by the research community. Sometimes the S10 call seemed connected with other S10 calls (Figure 3d), and some S10 calls were produced as a "hissing" sound within a sequence but following the sequence pattern (Figure 3e). A total of 1,046 (874 disconnected and 172 connected) S10 calls were detected during approximately 4 hr (13:00–17:00) on March 3, 2019, with the highest energy calls (662 signals) detected around 2 hr (13:20–15:20). Qualitatively, the S10 calls showed mostly a "convex" modulation or slightly "down sweep," but they could also be found as a nonmodulated "flat" tone. The calls showed a mean peak frequency of 437.3 Hz, in a range from 405.3 Hz (mean minimum) to 457.6 Hz (mean maximum), and a mean duration of 1.15 s (Table 2). S10 belongs to type 3 calls "Tonal with or without harmonics," resembling the "convex" modulation of the fundamental component of the M3 call shown in (Širović, 2011). Compared to other calls within this type S10 has similar duration, but the highest fundamental frequency and no harmonic content. Calls were classified as: S10 (tonal), S1 (pulsed), and S10 connected (tonal). The calls that were "blurred" in the spectrogram ("hissing") were not considered. A sequence was defined as a continuous production of calls separated by less than 10 s; if more than 10 s were separating two calls, then a new sequence was considered. A total of 246 sequences were obtained containing 1 to 76 calls. The S10 call was mostly detected in sequences (Figure 3c), ranging from 2 to 11 S10 calls repeated in a sequence, although single S10 calls also occurred (48 times). Most sequences consisted of clearly separated S10 calls (128 sequences), but some of them included S10 connected calls (38 sequences). S10 calls were also found to be associated in sequences with S1 calls (30 sequences) in alternate or continuous production. For sequence analysis the encoding information paradigms (Kershenbaum, 2013) identified were (1) Repetition (for example sequences of S10 from 2 to 11 calls); (2) Diversity (sequences including "S10, S1 and S10 connected"); and (3) Ordering (sequences S10, S10, S1, S1 and S1, S1, S10, S10 mirrored for example). Sequences presented Repetition, Diversity, and Order encoding but not a clear structure that would lead to grammar, therefore, a Markov chain analysis was selected to establish whether the signals were randomly produced (Kershenbaum et al., 2016). The first Markov chain model order obtained from the sequences showed that there was a higher probability of a S10 connected signal of being repeated (0.8108) than followed by an S10 (0.1892), and no chance of being followed by an S1 (unconnected nodes). Once a S10 appeared there was a high probability of being repeated (0.8509) and a low one of being followed by an S1 pulse (0.0978) but slightly greater than switching to a S10 connected (0.0512). S1 had a higher probability of switching to S10 (0.9126) rather than repeating itself (0.0874), and no chance (unconnected nodes) of switching to S10 connected (Figure 4). The first-order Markov model of the sequences revealed a sequential structure with forbidden and preferred state transitions. This nonrandom utterance implies information encoding in the structure of the sequences. S10 calls in itself convey a message (as any other call), but the linear ordering of calls can have different meaning to receivers (Kershenbaum, 2013). To our knowledge this tonal call without harmonics has not been reported in the literature or on research websites. Tonal calls from gray whales, such as S3 and S9, have been reported in the breeding lagoons at lower frequencies (<100 Hz), shorter in duration, and with or without harmonic content (Dalheim, 1987; López-Urbán et al., 2016). However, a solitary gray whale calf (<2 years old) kept under human care was reported to produce tonal moans at higher frequencies (270 Hz) that adults (Wisdom et al., 2001). The new S10 signal is a tonal signal with higher frequency than those previously reported (433.3 Hz median). The Ensenada de La Paz channel is a shallow area (<10 m depth) in which calls >400 Hz would propagate better compared with very low frequency calls. In terrestrial animals, high frequency tonal signals with a simple modulation have been associated with fear or appeasing (August & Anderson,1987; Briefer, 2012; Morton, 1977), or distress (Briefer, 2018; Lingle et al., 2012). Humpback whales have been reported to use high frequency calls during splitting and joining group members (Cusano et al., 2021; Dunlop et al., 2008). Moreover, Norris et al. (1977) recorded broadband pulses, grunts, and low frequency pulses from stranded alive calves; when calves were released, spaced pulsed sounds (often alone) were recorded as they reunite with their mothers. The gray whale sighted alive was a juvenile in an advanced emaciation stage that was likely associated with at least another juvenile in the same condition (the stranded gray whale found inside the lagoon). Thus, the authors speculate that both whales might have been migrating together, then entered the lagoon (Ensenada de La Paz), and one individual stranded in shallow waters. The other whale remained in the channel area, possibly vocalizing while going back and forth towards the bay, hence recorded at the time of the acoustic monitoring. It is unclear if S10 calls or mixed S1–S10 sequences have a distinctive function related to the context of the whales found in Ensenada de La Paz, but from their acoustic characteristics, repetition, and the context, they might function as a contact call, or even as a distress call. However further research is needed to confirm the meaning of these calls and the information encoded in the sequences. Assuming that this is a distress call could help lead to action when detecting this whale species visiting shallow areas in which they could potentially strand, especially if those areas lie outside their typical range. The authors thank Marco Gauger for his help deploying the recording system, Roberto Robles-Hernandez (PROFEPA) and Red de Varamientos BCS for the gray whale data in the lagoon, and Diana Fischer who improved the English. Moreover, to editor Daryl Boness, associated editor Denise Risch, and two anonymous reviewers for their valuable input to improve this note. Braulio Leon-Lopez: Data curation; formal analysis; investigation; methodology; visualization; writing – original draft. Aurora Paniagua-Mendoza: Formal analysis; investigation; methodology; validation; writing – review and editing. Eduardo Romero-Vivas: Formal analysis; investigation; methodology; software; supervision; validation; writing – review and editing.