Early onset of postnatal individual vocal recognition in a highly colonial mammal species

Abstract

Mother–young vocal recognition is widespread in mammals. The features of vocal recognition are known to be shaped by the ecological constraints faced by each species. In some species, a rapid establishment of mother–young vocal recognition is crucial for offspring's survival. However, knowledge of the precise features of this recognition system, especially the timing of the onset in the first hours after birth, is often lacking. Here we show that Cape fur seal females can recognize their pup's voice 2–4 h after parturition and that pups develop this aptitude 4–6 h after birth. This study is the first to investigate this mechanism in a wild and free-ranging mammal from only 2 h after birth. We report the fastest establishment of mother–young vocal recognition for any mammalian species, including humans, described to date. Such early vocal identification in pups suggests an in utero vocal imprinting. These findings highlight the synergistic role of environmental constraints and biological traits in optimizing the timing of individual vocal recognition onset in vertebrates.

1. Introduction

In most mammals, only the mother provides care to the offspring. Mother–young individual recognition is a significant way to optimize maternal investment. It reduces costs by avoiding misdirected care and thus enhances mothers' reproductive success and increases offspring's chances of survival [1]. In many mammal species, the acoustic channel plays a major role in this recognition process as mother and young develop the ability to identify each other through the vocal signals (i.e. vocalizations) they produce [2–8]. Although the occurrence of individual vocal recognition has been well documented in mammals, some features such as the timing of the onset of recognition, its temporal pattern throughout lifetime and the individual signature are not well understood and rarely investigated in wild mammals [9,10].

Rapid establishment of recognition between the mother and her young is crucial to ensure the survival of the young, particularly in species experiencing stringent ecological constraints such as a high risk of confusion between individuals, early pup mobility and early separations between mother and young due to foraging needs. Experimentally testing the development of the vocal recognition in mother–young pairs of wild mammals within the first hours after birth is challenging for reasons of access to the animals, tracking of births and the ability to conduct experiments. Only a few studies have been carried out so far, involving five domestic [11–15] and five wild species [3,16–19] (pinniped species only). In most cases, individuals were tested from a few days after birth (i.e. later than 24–48 h), except in sheep (Ovis aries) for which ewes and lambs were tested within 6–12 h after parturition [12] and for female Australian sea lions (Neophoca cinerea), which were tested within 12 h after parturition [16]. The timing in development of vocal recognition, especially in the first 24 h after birth, is thus poorly understood in mammals, including humans [2,20].

The Cape fur seal (Arctocephalus pusillus pusillus) is a particularly interesting species in which to investigate the characteristics of vocal recognition in mammals because mother–pup pairs are under strong ecological constraints to develop individual recognition. As one of the most colonial mammals in the world, Cape fur seals reproduce over a short breeding season (lasting from mid-October to early January each year [21]) and within massive colonies of up to several hundred thousand typically tightly packed individuals. As with most otariid species, allo-nursing is rare in this species [22,23] (i.e. pups are fed only by their own mother) and mother and pup are frequently separated throughout the lactation period (maternal attendance periods on shore are interspersed with multi-day foraging trips at sea by the mother). The first foraging trip is reported to occur 6 days after the pup's birth [24]. Females and pups produce harmonic-structured vocalizations (figure 1a) to communicate with each other, respectively named pup-attraction calls (PAC) and female-attraction calls (FAC) [25]. PAC and FAC are used for both long-distance communication i.e. when the female returns after a foraging trip at sea or after a short swim for thermoregulation, and short-range communication when both are gathered in the colony [25]. Previous studies showed that PAC and FAC display a high degree of individual stereotypy [26] and are involved in a mutual vocal recognition mechanism between the mother and her young [27].

Figure 1.

Behavioural responses of mothers and pups to playback experiments conducted to assess the timing of the onset of the mother–pup vocal recognition. (a) Playback experiments set-up. Each tested individual was exposed to the broadcast of two series of calls from a related individual (mother or filial pup) or a stranger individual (non-mother or non-filial pup). (b) General behavioural response of mothers (n_ind = 38) to playback trials (n_trials = 59) conducted from 2–4 h to 12–24 h following parturition. Mothers' response is indicated by the composite score (PC1 scores) obtained from six behavioural variables. (c) General behavioural response of pup (n_ind = 36) to playback trials (n_trials = 57) conducted from 2–4 h to 24–48 h following birth. Pups’ response is indicated by the composite score (PC1 scores) obtained from two behavioural variables. LME followed by a post hoc analysis of estimated marginal means. Significance code is p-value [0, 0.001]: ***, (0.001, 0.01]: **, (0.01, 0.05]: *, (0.05, 1]: NS. Boxplots present median values with first and third quartiles (lower and upper hinges) and whiskers represent the min and max values.

The mother–offspring vocal recognition processes in mammals are known to be shaped by the ecology of the species (social organization and breeding strategy) [10,28,29]. Therefore, it is likely that the extreme ecological constraints that the Cape fur seal currently is facing act as a selective pressure for the establishment of a rapid mother–pup individual recognition in this species. We hypothesized that this recognition would be established within the first days after parturition (i.e. before the first mother–pup separation) for both females and pups. Using playback experiments, we tested Cape fur seal mothers and pups at different hours after birth to assess the timing of the onset of recognition of the pup's voice by the mother and the mother's voice by the pup. Furthermore, vocal recognition is established through an imprinting period during which the animals are repeatedly exposed to vocal signals that they learn to recognize [22]. The vocal activity of both mother and pup immediately after birth is therefore likely to play a role in the onset of vocal recognition. We investigated the link between the behavioural responses of mothers and pups to early playback trials (conducted 2–4 h after birth) and the vocal activity (call production rate) of respective pups and females within the first 2 h after birth.

2. Materials and methods

(a) . Study site and animals

Recordings and playback experiments were conducted at Pelican Point Cape fur seal breeding colony, Namibia (25°52.2′S, 14°26.6′E) during the breeding season—from 6 November to 29 December 2021 (over 33 days). Of the 51 mother–pup pairs included in this study, we witnessed the birth of 30 pups. For 21 other cases, the fresh placenta was observed through seagulls gathering around the placenta to eat it and the mother–pup pair carefully approached. In Cape fur seals, the placenta is usually expulsed within 30 min after birth [23]. Our observations of 30 live births with subsequent placenta expulsion supported this, with an average time of 18.2 ± 22.3 min for placenta expulsion after parturition (electronic supplementary material, table S1). For the 21 births that were identified through gulls eating the freshly expulsed placenta, the time of birth was considered to be 30 min earlier.

At Pelican Point, Cape fur seals are not habituated to the close presence of humans in the colony. To avoid disturbing the animals and inducing fearful reactions that may extend to stampedes, seals were approached carefully by crawling along the sand. Recordings and experiments were carried out from a distance, using a 5 m pole to bring the equipment close to the animals without disturbing them. To identify mother–pup pairs over time, pups were bleach-marked using hair-dye (Blonde highlight kit, ©Kair). The marks consisted of numbers applied on their flank using a 10 cm-wide wooden pad attached to the end of a 5 m pole. The pups were marked at least 1 h after birth. Females were not marked but identified through their consistent association with and nursing of their marked pup. Females nurse their own pups only, allo-nursing being rare in otariid species [22].

(b) . Recording procedure

Mothers’ and pups’ calls (respectively, PAC and FAC) were recorded during the first 2 h after the birth of the pup. Vocalizations were recorded using a Sennheiser ME67 directional shotgun microphone (frequency range: 40–20 000 Hz ± 2.5 dB) at 44.1 kHz sampling frequency connected to a two-channel NAGRA LB digital audio recorder. The distance between the microphone and the focal animals ranged from 3 to 6 m during recording sessions.

The duration of the recordings varied among mother–pup pairs (ranging from 5 to 120 min in total for a given pair) depending on their vocal activity and the recording possibilities i.e. accessibility of the mother–pup pair in the colony, movements of the animals around them, wind conditions, etc. The number of calls produced by each individual was counted during the total recording duration to quantify their vocal activity and expressed as the average number of calls produced per 10 min.

(c) . Playback stimuli

Stimuli (i.e. playback track) were created based on a previous study on Cape fur seal mother–pup vocal recognition [27]. Playback stimuli were prepared on-site from a laptop, using Avisoft SAS Lab Pro software (R. Specht, version 5.2.14, Avisoft Bioacoustics, Berlin, Germany). Series consisted of six calls from the same individual, each separated by 2 s of silence. Whenever possible, the tracks were composed of six different calls, but sometimes fewer than six good-quality calls (good signal-to-noise ratio and no overlap) could be recorded from the same individual within 2 h of birth and some calls had to be replicated in the playback track. The tracks had an average duration of 22.1 ± 4.9 s for females' calls and 17.1 ± 3.1 s for pups’ calls. The amplitude level of the playback tracks was adjusted to the natural amplitude of PAC and FAC (respectively, 88 ± 2 dB SPL at 1 m and 80 ± 2 dB SPL at 1 m). To ensure that the amplitude level was correct and consistent between trials, both the volumes of the sound player and the loudspeaker were calibrated on-site prior to conducting the experiments. During calibration, received levels were measured with a ‘Testo 815’ sound level meter. In addition, the amplitude level of all sound files corresponding to playback tracks was set at a standardized value during their building on the software (amplitude normalized to 95% of the dynamic range).

(d) . Playback procedure

A playback trial consisted of exposing the tested individual to two stimuli series: a series of calls recorded from a related individual i.e. the focal pup's own mother or the focal female's own pup—and a series of calls recorded from an unrelated individual (of the same population and pups of the same age). The two playback series were separated by at least 2 min and the order of the series was randomized [27]. Females were tested up to 24 h after their pup's birth and pups were tested up to 48 h. Playbacks on females were classified into 4 categories of time elapsed since the pup's birth: [2–4 h), [4 –6 h), [6–12 h) and [12–24 h] (i.e. [2–4 h) refers to experiments ≥ 2 h and < 4 h). Playbacks on pups were divided into five categories: [2–4 h), [4–6 h), [6–12 h), [12–24 h] and (24–48 h]. Each individual was tested at different time categories between 2 and 24–48 h, but with only one trial per category and a minimum duration of an hour and a half between two successive playback trials. In cases where the tested individual showed no behavioural response to both series of a trial (due to a lack of motivation or a sleeping animal that was not woken by the played-back calls), the trial was discarded from the analyses. After such ‘no-response’ trials, the break duration before the next trial was reduced to 30 min. Two trials could thus occur in the same time category but only if the first trial induced no response to either series.

Females were tested in the presence of their pup, while pups could be tested during their mother's absence (i.e. females thermoregulating in the water during hot periods). Calls were broadcast using a waterproof and wireless high-powered loudspeaker (JBL Charge 3, 2 × 10W, frequency response: 65 Hz–20 kHz) connected to a Bluetooth sound player (Sony NW-A35). The loudspeaker was placed from 1 to 3 m from the focal individual at a 45–90° orientation to induce searching behaviour and thus facilitate the evaluation of a behavioural reaction. Playback experiments were filmed using an Olympus Tough TG-6 camera to allow further analysis.

(e) . Behavioural response

Responses to the playback series were all assessed through video analysis with the software BORIS [30]. Behavioural responses were observed for 30 s from the beginning of the playback. For mothers, response variables were latency to look towards the loudspeaker (s), look duration towards the loudspeaker (s), latency to check her pup (s), duration of pup check (s), latency to call (s) and number of calls. The variable ‘pup check’ is defined as a female looking at and/or sniffing her own pup. From previous observations and the results reported by Martin and colleagues [27], very young pups (i.e. a few hours or days old) are not able to locate the sound source accurately. Indeed, during the broadcast of their mother's calls, we noticed that they look around without directing their eyes towards the loudspeaker. Furthermore, they are not very mobile and never move towards the loudspeaker as pups of several weeks of age can do [27]. In this study, the response variables for pups were thus latency to call(s) and number of calls. For both mothers and pups, the absence of a given behavioural variable (e.g. no call) was assigned a default value of 30 s for latency.

(f) . Statistical analysis

(i) . Onset of vocal recognition in mother–pup pairs

The overall behavioural response of each tested individual was obtained by combining the raw data of the response variables (six for females and two for pups) in a principal component analysis (PCA) and obtaining a composite score of the behavioural response [31]. Principal components (PCs) with eigenvalues greater than 1 were retained (Kaiser's criterion) and corresponding PC scores were used as a composite score to quantify the level of response of tested mothers and pups. PC scores were compared between filial and non-filial (or mother and non-mother) series and by time category after the pup's birth using a linear mixed-effects model (LME): PC scores were set as a ‘response variable’, while the series (filial or non-filial, mother or non-mother) and the time category were set as fixed effects. In addition, the playback trial (combination of both ID and hour, e.g. m32 tested at [2–4 h) or p28 tested at (24–48 h]) was defined as a ‘random effect’ to account for the fact that data are non-independent (each trial consists of observing the same focal individual over two series but during the same time period). The model was run with the lme4 R package [32]. A post hoc analysis of estimated marginal means was conducted to investigate differences between the two series in each time category using the emmeans package in R [33].

(ii) . Link between vocal recognition and vocal activity within the first hours after birth

The vocal activity or call rate (i.e. number of calls produced per 10 min) during the first 2 h of the pup's life was calculated for each individual of the mother–pup pairs in which at least one of the two protagonists was tested 2–4 h postpartum (n = 35 pairs). The relationship between the vocal activity of a mother and those of her pup during the first 2 h postpartum was tested by calculating a Pearson correlation coefficient (stats package, R Core Team, 2021). We investigated whether the vocal activity of a pup within its first 2 h of life could impact the behavioural responses of its mother when tested early after parturition, i.e. in a playback trial conducted 2–4 h postpartum. For this, we performed a linear regression between pups' call rate and the difference in their mother's behavioural response between the filial and non-filial series (PC1 scores filial – PC1 scores non-filial) for females tested at [2–4 h). The same approach was used to evaluate the impact of the mother's vocal activity on the pup's behavioural response (indicated by the difference PC1 scores mother – PC1 scores non-mother) to early playbacks conducted at [2–4 h,) indicated by the difference between PC1 scores mother and PC1 scores non-mother.

Second, we assessed the effect of the vocal activity of an individual (either mother or pup, during the first 2 h after birth) on its behavioural response to the two types of series conducted at [2–4 h). For mothers, two linear regressions were conducted between the mothers' call rate and their response to filial or non-filial series. Similarly, two linear regressions were carried out between pups’ call rate and their behavioural response to either the mother series or the non-mother series.

3. Results

(a) . Onset of vocal recognition in mother–pup pairs

We investigated the ontogeny of mother–pup vocal recognition by testing females and pups early after birth and within the first 24 h (females) or 48 h (pups) of the pup's life using an acoustic playback approach. Each trial consisted of exposing the tested individual to two vocal stimuli: one from the pup (filial) or the mother of the tested individual and another one from a stranger individual (non-filial pup or non-mother female, figure 1a).

We studied 51 mother–pup pairs for this study and performed a total of 162 playback trials. Among them, trials that did not elicit any behavioural responses for both stimuli were excluded (i.e. animal remained sleeping; electronic supplementary material, table S2), leaving 116 trials for analysis (n = 59 for mothers and n = 57 for pups; figure 1b,c).

For females, due to small sample size for [6–12 h) and [12–24 h] age categories (n = 8 and n = 6 respectively), playbacks trials conducted from 6 to 24 h were grouped into a single [6–24 h) category to ensure effectiveness of statistical analysis. The behavioural response of females to the filial stimulus (i.e. calls recorded from her own pup) was significantly stronger than the response to the non-filial stimulus for all pup age categories between 2 and 24 h ([2–4 h): estimate = 1.57, s.e. = 0.282, p < 0.0001; [4–6 h): estimate = 1.86, s.e. = 0.420, p < 0.0001; [6–24 h): estimate = 2.20, s.e. = 0.420, p < 0.0001; figure 1b). When exposed to a series of vocalizations from their own pup, females looked at the sound source (loudspeaker) more quickly after the start of the broadcast and for a longer duration. They also responded vocally more quickly after the start of the playback and produced more calls (look and vocal responses represented by PC1 scores; table 1; electronic supplementary material, Video S1). No significant difference was found for variables related to the pup check (PC2 scores, p-values from estimated marginal means analysis ranged from 0.054 to 0.23; table 1).

Table 1.

Summary of the first two PCs (PC1 and PC2) resulting from the two PCA performed on mothers' and pups’ playback trials with, respectively, six and two behavioural variables. For mothers, the two PC were retained (eigenvalues > 1): PC1 includes both look and vocal components of the behavioural response, whereas PC2 relates only to the check of the pup by the mother. For pups, only PC1 was retained.

	playbacks on mothers		playbacks on pups
	PC1	PC2	PC1	PC2
eigenvalues	2.87	1.73	1.66	0.34
% cumulative variance	47.85	76.70	83.01	100
correlation coefficients between PC and variables
latency to look	−0.70	−0.07
look duration	0.87	0.18
latency to check her pup	−0.13	0.91
duration of pup check	0.11	−0.92
latency to call	−0.90	−0.002	−0.91	0.41
number of calls	0.89	0.005	0.91	0.41

Similar playback trials were performed on pups. Playback trials conducted on pups from 6 to 24 h old were also grouped in a single [6–24 h) category because of small sample sizes (n = 5 for [6–12 h) and n = 8 for [12–24 h]). As the behavioural response of newborn pups to their mother's calls is exclusively vocal (i.e. they are not yet able to localize the sound source accurately and thus they did not look towards the loudspeaker or approach), their behavioural response to the playback series was only assessed by the latency to call and the number of calls produced. No difference in the pups' behavioural response was observed between the mother and non-mother series in playback trials conducted at 2–4 h after birth (estimate = 0.139, s.e. = 0.390, p = 0.7221; figure 1c). However, the pups' behavioural response was stronger to their mothers' calls compared to the non-mother series in all of the three following age categories spanning from 4 to 48 h after birth ([4–6 h): estimate = 0.983, s.e. = 0.471, p = 0.0418; [6–24 h]: estimate = 0.946, s.e. = 0.471, p = 0.0498; (24–48 h]: estimate = 1.457, s.e. = 0.490, p = 0.0044; figure 1c, electronic supplementary material, Video S2). Therefore, pups' vocal recognition of their mothers' calls appeared to develop between 4 and 6 h after birth. However, 7 out of 19 pups responded more strongly to their mothers' calls than to non-mother calls, suggesting that an earlier recognition onset could be possible for some individuals.

(b) . Link between vocal recognition and vocal activity within the first hours after birth

We next examined how the vocal activity of an individual during the imprinting period could influence the timing of the onset of mother–pup vocal recognition. The vocal activity (i.e. call rate, number of calls produced per 10 min) during the first 2 h of the pup's life was measured for each target mother–pup pair, for which at least one of the two protagonists was tested 2–4 h after the pup's birth (n = 35 pairs, electronic supplementary material, table S3).

We found no correlation between the vocal activity of a mother and the vocal activity of her pup (Pearson's product–moment correlation, t = 0.305, d.f. = 33, p = 0.762). This means that vocalizations produced by mothers and pups soon after birth are not always in response to the other's calls. The mothers' vocal rate was highly individually variable and ranged between 0 and 55.9 calls/10 min (mean value: 8.4 ± 10 calls/10 min) (figure 2c; electronic supplementary material, table S3). Where labour and birth were observed, females were not observed to vocalize during labour, only after the pup's birth. Pups were generally more vocal compared to mothers and were already calling a few minutes after birth. Their call rate, also highly variable between pups, ranged between 21.6 and 220.7 calls/10 min (mean value: 68.4 ± 48.7 calls/10 min) (figure 2f; electronic supplementary material, table S3).

Figure 2.

Link between mothers' and pups’ behavioural responses to playback trials conducted at 2–4 h after birth and their vocal activity during the first 2 h postpartum. (a) Relationship between pups' vocal activity during the first 2 h after birth (in number of calls produced per 10 min) and the difference in their mother's behavioural response to the filial and non-filial playback series broadcasted 2–4 h after parturition (PC1 scores filial – PC1 scores non-filial). (b) Relationship between mothers' vocal activity during the first 2 h postpartum and their behavioural response to the filial series (PC1 scores filial, in purple) and the non-filial series (PC1 scores non-filial, in grey). (c) Spectrograms of calls produced by a female (m44) showing a high vocal activity (average call rate = 55.9 calls/10 min) during recordings made within 2 h after parturition. (d) Relationship between mothers' vocal activity during the first 2 h after parturition (in number of calls produced per 10 min) and the difference in their pup's behavioural response to the mother and non-mother playback series broadcasted 2–4 h after birth (PC1 scores mother – PC1 scores non-mother). (e) Relationship between pups' vocal activity during the first 2 h after birth and their behavioural response to the mother (PC1 scores mother, in orange) and the non-mother series (PC1 scores non-mother, in grey). (f) Spectrograms of calls produced by a pup (p66) showing a high vocal activity (average call rate = 220.7 calls/10 min) during recordings made within 2 h after birth.

We then assessed if the pups' vocal activity within the first 2 h could influence the onset of maternal vocal recognition. We used the difference in mothers' response to filial and non-filial series (PC1 scores filial – PC1 scores non-filial) in playback trials conducted at 2–4 h after parturition, as a proxy of their ability to recognize their pup's voice early. Positive values indicated a stronger maternal response to the vocalizations of their own pup, while negative values showed a stronger response to the calls of a non-filial pup (figure 2a). We found no significant relationship between the pups' call rate during the first 2 h after birth and the ability of mothers to distinguish between filial and non-filial calls at 2-4 h (F_1,29 = 0.009, p = 0.927, adjusted R² = −0.034, figure 2a). In other words, females showing stronger responses to the broadcast of their pup's calls at 2–4 h after parturition did not necessarily have a highly vocal pup. Similarly, mothers with the most vocal pups (figure 2f) did not show a stronger difference in their behavioural response to the two playback series. In addition, the mothers’ vocal activity early after parturition (within 2 h) had no influence on their own behavioural response to the filial and non-filial playback series (F_1,28 = 0.945, p = 0.339, adjusted R² = −0.002 for filial series and F_1,28 = 0.057, p = 0.814, adjusted R² = −0.034 for non-filial series). In other words, highly vocal females do not necessarily respond more to the playback series.

Lastly, as some pups identified their mother's call between 2 and 4 h after birth, we investigated if such an ability could be influenced by their mother's vocal activity within 2 h after birth. As for mothers, the difference between PC1 scores mother and PC1 scores non-mother was used as a proxy to assess the occurrence of an early vocal recognition. Positive values indicated a stronger response to the mother's vocalizations, while negative values showed a stronger response to the calls of a non-mother female (figure 2d). Similar to females, no relationship was found between mothers' call production and pups' recognition (F_1,17 = 1.982, p = 0.177, adjusted R² = 0.052; figure 2d). Although some pups were exposed to their mother's calls many times (i.e. very vocal females) in the early stages after birth, they were not able to recognize their mother's voice within the first 2 h: they either responded equally to both series or more strongly to the non-mother series (figure 2d). By contrast, among the pups who successfully identified their mother's vocalizations, some had relatively quiet mothers (figure 2d), such as m29 (mother of p29), which had only produced one single call during the first 2 h after birth (average call rate of 0.1 calls/10 min; electronic supplementary material, table S3). Again, there was no relationship between the vocal activity of a pup and its' behavioural response to the mother and non-mother series (F_1,17 = 0.043, p = 0.837, adjusted R² = −0.056 for mother series and F_1,17 = 1.701, p = 0.210, adjusted R² = 0.037 for non-mother series).

4. Discussion

This study was the first to investigate the ontogeny of mother–young vocal recognition in a wild and free-ranging mammal species from only 2 h after birth. We showed that female Cape fur seals were able to recognize their pup's voice within 2 h after parturition. In other mammal species, the vocal recognition of an offspring by its mother is established later: 24 h in sheep [12], 48 h in mink [11] and goat [13], 24–48 h in the northern elephant seal [18] and 48 h Australian sea lion [16]. Even in humans, only 40% of women succeed in recognizing their own baby's cries after 24 h [2]. Such early recognition is likely to involve highly evolved cognitive mechanisms to enable Cape fur seal females to learn and memorize the vocal signature of their pup's vocalizations within 2 h and to respond specifically and appropriately when they hear them.

Regarding Cape fur seal pups' vocal recognition of their mother's calls, we showed that it is established between 4 and 6 h after birth. Again, this timing is much shorter than in other mammal species such as lambs [12] or goat kids [13,14] (48 h), but also compared to other pinniped species: the Subantarctic fur seal (2–5 days) [3], the Galapagos fur seal and the Galapagos sea lion (10 days) [19] or the Australian sea lion [17] (still not established before the first mother–pup separation occurring at 10–14 days after birth). In humans, babies seem to be able to recognize their mother's voice by one month of age [34,35]. However, to our knowledge, there is no study investigating such recognition earlier or assessing the timing of its onset. In sheep [12] or in Australian sea lion [17], the mother–young vocal recognition is mutual but occurs slightly later in offspring than in mothers. Here in Cape fur seals, the timing of the onset of vocal recognition appeared to be relatively synchronous.

Considering the extreme ecological constraints that Cape fur seal mother–pup pairs are facing, we expected the vocal recognition to be mutual and to develop early, i.e. before the first mother–young separation for foraging occurring at 6 days post-birth. Surprisingly, our findings revealed a particularly early recognition in this species: 2 h for females to recognize their pups and 4–6 h for pups to recognize their mothers. Our observations at the Pelican Point breeding colony (Walvis Bay, Namibia) showed that pups experienced earlier separations from their mothers, well before the female's first extended foraging trip to sea. During hot days (sunny days with temperature above 20°C), we observed females leaving their pup alone in the colony for a short period of time (15–30 min to an hour—although not systematically recorded) to swim and thermoregulate. During our fieldwork, four marked pups were seen alone in the colony less than 3 days after their birth—specifically, at 19, 32, 47 and 50 h postnatally—and all were successfully reunited and seen in the presence of their mother afterwards. While these are short separations and the mobility of newborn pups is limited, they still require a reunion process between mother and pup, taking place among other conspecifics of the colony, which would involve a process of individual vocal recognition. In addition to the high temperatures, predation events are also responsible for early mother–pup separations. Indeed, Cape fur seals are exposed to different terrestrial predators such as black-backed jackals, brown hyenas and even desert lions [36], and these attacks may induce panics and thus separations in mother–pup pairs. A reliable vocal communication will thus facilitate reunion between the two. In addition to these factors, the high density of individuals in the breeding colony might represent an additional challenge for successful reunion that could catalyse an early development of mother–young vocal recognition. Indeed, PAC and FAC of Cape fur seals are highly individually distinctive [26]. Identity information appears to be encoded in acoustic features such as the duration of the call, the fundamental frequency and the energy spectrum [26]. The perception and discrimination of these features are likely involved in the vocal recognition process, supporting the identification abilities of females and pups.

Considering the strong inter-individual variations in mothers' and pups' vocal activity within the first 2 h after parturition and the fact that, during this time, pups are a lot more vocal than females, we thus assessed whether vocal production can affect the early vocal recognition between mothers and pups. The results showed that, in both cases, the behavioural response of one individual (female or pup) to a playback trial performed between 2 and 4 h after birth is not linked to the vocal activity of the other individual of the pair, and thus to the amount of calls to which the tested individual was exposed before the playback trial. According to our experiments, the call rate of the least vocal pups (at 20–40 calls/10 min, figure 2a) was sufficient to allow their mothers to learn and memorize their vocal signature. Maternal vocal recognition in Cape fur seals is thus based on a learning process that is rapid and highly efficient. With an average of about 70 calls produced by a pup every 10 min, mothers are exposed early and intensely to the voice of their offspring after birth. The rich imprinting period is therefore likely to facilitate rapid maternal vocal learning.

By contrast, the pup's exposure to its mother's voice is much more variable among mother–pup pairs and sometimes very low during the first hours—from 0 to about 56 calls produced by a female every 10 min. These differences in vocal activity of females can be explained by a variety of factors: the duration of the labour, the presentation of the fetus (cephalic or breech), their physiological state (fatigue, stress), their personality and their experience (e.g. primiparous or multiparous females). Some females were not very vocal during these first 2 h postpartum and started to call at a later stage, after a few hours of rest. However, we found no correlation between the pup's vocal activity during the first 2 h after birth and the mother's behavioural response to a playback trial conducted between 2 and 4 h after parturition.

In spite of a much lower vocal exposure in pups, our behavioural experiments revealed that pups can recognize their mother's voice as early as 4–6 h after birth. In particular, we found that a 3 hour-old pup (p29) was able to recognize its mother's calls during the playback (higher number of calls produced with a lower latency in response to the mother series expressed as a difference in PC1 scores of 2.06; figure 2d), even if only exposed to a single maternal call. Similarly, three other pups (p55, p81 and p82) recognized their mother's voice in spite of a low maternal vocal rate (up to 3.6 calls/10 min; electronic supplementary material, table S3). The lack of a significant relationship between maternal call production after birth and the vocal recognition by the pup indicates that the imprinting period may not be limited to the first few hours after birth. As one call is likely not enough for the pup to imprint the voice of its mother, a longer exposure would be probably required for establishment of vocal recognition, especially for a colonial species where pups are exposed simultaneously to the calls of many females. Our results indicate that the learning of the mother's voice by pups might start to take place in utero. In mammals, the uterine auditory environment is rich: it is composed of sounds generated inside the mother (associated with digestive, respiratory or cardiovascular activities) and exogenous vocal signals like conspecific voices. It has been shown in humans and cetaceans that hearing abilities develop progressively during the development of the fetus and begin to function before birth [37,38] (the sound, especially low frequencies, being mainly transmitted to the inner ear by bone conduction [39–41]). Among exogenous signals such as human voices, the mother's voice is dominant because the sound waves are also conveyed directly to the womb through the mother's body (soft tissue and liquid conduction) [42]. This double internal and external transmission causes the sound to be less attenuated or distorted [43], as well as being slightly louder compared to other voices [44]. In humans, fetuses were reported to respond differently to their mother's voice (i.e. increase in fetal heart rate) than to the voice of a stranger woman (decrease), suggesting discrimination [45]. However, the ability of human babies to recognize the maternal voice has never been tested earlier than one month after birth [34,35]. During the gestation period of the Cape fur seal (which lasts for 8 months, excluding the delayed implantation period [46]), females also nurse their pup from the previous year (lactation period lasts from 9 to 11 months [24]) and thus produce calls to communicate with them. Throughout its development, the fetus is thus frequently exposed to its mother's voice. There may be a pre-natal learning process during which the fetuses are imprinting on their mother's calls features [47] (vocal signature) allowing for recognition of these calls to develop. This could allow the newborn pups to develop a rapid vocal recognition soon after birth. It is likely that this learning process would rather be supported by acoustic features linked to the source and not the filter, as the fetus can assess its mother's voice from inside. So, source features such as temporal features, including duration, amplitude and/or frequency modulations of the calls, as well as the value of the fundamental frequency could be learned by the fetus in utero. Some of these source-related features have been shown to be individual-specific in females' calls in a previous study [26]. By contrast, features related to the filter, such as the spectral characteristics, cannot be assessed reliably as they are likely filtered by the mother's body [44]. Such pre-natal vocal learning mechanisms have been shown in humans [40,48] and birds [49,50], and might also occur in pinnipeds in response to the strong ecological constraints for an early, reliable and efficient mother–pup recognition.

In summary, the establishment of the mother–pup vocal recognition in Cape fur seal occurs 2–4 h after birth in females, and 4–6 h after birth in pups. This appears to be the fastest establishment of mother–young vocal recognition described to date for any mammalian species (including humans). We suggest that pre-natal learning is likely to occur and thus facilitates the establishment of vocal recognition soon after birth for pups. Such precocial learning of the voice is likely driven by the complex environment and the strong ecological constraints to which this highly colonial species is exposed. The high density of individuals generates both an acoustic and visual jamming, and the frequent and early separations of the mother from her pup are selective pressures that have shaped their recognition system, and led to an early onset of mother–pup recognition.

Ethics

All procedures were approved by the Research Ethics Committee (Animal Care and Use) of Stellenbosch University (ACU-2021-15015) and authorized by the Namibian National Commission on Research Science and Technology (NCRST; Authorization no.: AN202101095).

Data accessibility

Raw data are available on Zenodo repository at https://doi.org/10.5281/zenodo.6768678 [51].

The data are provided in the electronic supplementary material [52].

Authors' contributions

M.M.: conceptualization, data curation, formal analysis, writing—original draft and writing—review and editing; T.G.: project administration, resources and writing—review and editing; S.E.: project administration, resources and writing—review and editing; I.C.: conceptualization, data curation, funding acquisition, project administration, supervision and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This project has received financial support from the Centre National de la Recherche Scientifique (CNRS, France) through the MITI interdisciplinary programs.