Playback experiment shows no evidence for vocal learning in titipounamu nestlings (Acanthisitta chloris)

ABSTRACT A recent reshuffling in the avian phylogeny indicates that New Zealand wrens and songbirds share a close common ancestor with parrots – making New Zealand wrens an excellent group to test for vocal learning. The New Zealand wrens have previously been classified as vocal non-learners, but their vocal learning ability has never been experimentally tested. Here, we explore the potential presence of vocal learning in one species of New Zealand wren, the titipounamu (Acanthisitta chloris). We expose nestlings to synthetic playback stimuli that simulate adult feeding calls and determine whether the nestlings altered their calls after exposure to the playback stimuli. We found that experimental nestlings did not alter their calls towards the playback stimuli. While this indicates that no vocal imitation occurred during the nestling period, other developmental stages should be tested for vocal learning, particularly during the fledgling or first year adult phase.


Introduction
The extent and evolutionary origins of vocal learning in birds are the focus of considerable debates (Kroodsma and Konishi 1991;Kroodsma et al. 2013;Liu et al. 2013;de Lima et al. 2015;Jarvis 2019;Wirthlin et al. 2019;Davenport and Jarvis 2022).For example, the evolution of vocal learning in Passeriformes remains unresolved mainly because vocal learning has not been experimentally established in many suboscine species and New Zealand wrenstwo Passeriform clades that share a common ancestor with vocal learning oscines and parrots (Jarvis and Kaas 2006;Hackett et al. 2008;Suh et al. 2011;Zhang et al. 2014).Thus, while advanced vocal learning is well established in the oscines (Brainard and Doupe 2002;Rundstrom and Creanza 2021), the picture is less clear for suboscines (Kroodsma and Konishi 1991;Kroodsma et al. 2013;Liu et al. 2013;Touchton et al. 2014;de Lima et al. 2015) and New Zealand wrens.
Overall, little is known about New Zealand wrens' vocalisations and their vocal learning ability (if at all present).Their vocal repertoire has been assumed to be non-learned based on their simple vocal features, non-territorial and non-courtship vocalisations, and a syrinx that lacks complex intrinsic muscles found in vocal learners in birds (Stidolph 1950;Ames 1971;Sherley 1985;Moran 2021).Furthermore, New Zealand wrens do not produce broadcast territorial songs as most vocal learning songbirds do; instead they only produce calls, which are short functional units that have historically been regarded as innate rather than learned (Marler and Mundinger 1975;Sewall 2011;Maat et al. 2014;Sewall et al. 2016).Although call learning is not as well documented as song learning, it has been demonstrated in a variety of species, including parrots (Medina-García et al. 2015;Wright and Dahlin 2017), songbirds (Mundinger 1970;Zann 1990Zann , 1985)), and more recently in musk ducks (Ten Cate and Fullagar 2021) and black-headed gulls (Ten Cate 2021) (although more tests are needed in the latter species).Other recent works have also revealed that one species of New Zealand wrens, the titipounamu (rifleman, Acanthisitta chloris), has a protracted vocal development (Loo 2021), vocal dialects (Withers 2013), and a vocal convergence and vocal plasticity that resemble those of vocal learners (Moran et al. in review).These traits are often associated with vocal learning and their coexistence in titipounamu could indicate the underlying presence of vocal learning.
To test for the presence of call learning in New Zealand wrens, tutoring playback experiments (i.e. in which tutors are simulated using playback stimuli) are an ideal non-invasive and reliable experimental method (Mennill et al. 2018;Schroeder and Podos 2023).Traditionally, tutoring playback experiments reared young birds in captivity under acoustic isolation and exposed to a novel tutor.These methods have been extremely successful at demonstrating the presence of vocal learning, and identifying the critical time period when young birds learn and memorise their adult songs or calls (Zann 1996;Brainard and Doupe 2002).During this period, young learners initially produce vocalisations with substantial variability, but eventually learn to imitate the acoustic features of their parents, or conspecifics and heterospecifics (Brainard and Doupe 2002;Villain et al. 2015).New Zealand wrens do not thrive in captivity due to their dietary requirements (i.e.tree trunk insectivorous diet) (Moeed and Fitzgerald 1982;O'Donnell and Dilks 1994), thus alternative non-traditional tutoring playback methods are more suitable for testing vocal learning in wild populations of New Zealand wrens.One such approach uses field setups (Mennill et al. 2018;Schroeder and Podos 2023), in which young birds are exposed to artificial playback stimuli despite the presence of live tutors.
Here, to test for the presence of call learning in New Zealand wrens, we conduct a field playback tutoring experiment during the nestling period of titipounamu (one of the two species of New Zealand wrens) by playing a combination of unique synthetic calls at their natal nests.By comparing nestling calls between experimental and control nests and by investigating call shifts of experimental nestlings towards playback stimuli, we gain further insights into titipounamu vocal imitation abilities.

Field site and study species
We conducted a playback tutoring experiment in a wild population of titipounamu at Boundary Stream Mainland Island reserve, New Zealand (39 • 06 ′ 15.8 ′′ S, 176 • 48 ′ 06.1 ′′ E) during two breeding seasons from September to February 2018-20 (approved the Department of Conservation, New Zealand and Auckland Animal Ethics Committee).Titipounamu are a small (5-7 g) cavity-nesting species that build nests at all heights, including under leaf litter or high in tree cavities (Sherley 1990a(Sherley , 1985;;Hunt and Mclean 1993;Higgins 2001).Due to the inaccessibility of most natural nests (i.e.tight spherical nests in tree cavities), we could not band nestlings, which limited information about the relatedness among individuals and their natal nests.The timing of titipounamu breeding is asynchronous (Sherley 1985), which enabled us to monitor nests simultaneously and continuously throughout the breeding seasons.For monitoring and identification purposes, adults were captured and uniquely colour banded (one of the colour bands was embedded with a Passive Integrated Transponder or PIT tag; 2.3 mm EM4102; Eccel Technology Ltd, UK).Birds were sexed and aged based on their sexual size dimorphism and plumage colouration (Hunt and Mclean 1993).The developmental period in the nest is relatively long for such a small bird and can last up to about 24 days (Sherley 1985).Parental care extends for four to five weeks into the fledgling period.
Titipounamu adults have a vocal repertoire of short and simple calls, and lack long, complex, broadcast songs (Loo et al., in review).During their vocal development, nestlings are exposed to adult titipounamu feeding calls (i.e.zip and purr calls, Figure 1A) which are produced prior to feeding nestlings (Withers 2013;Loo 2021;Loo and Cain 2021).These feeding calls are visually distinguishable (from spectrograms) from other call types due to their 'S' shape and higher frequency (∼ 6-14 kHz; Figure 1A).The frequency of adult visits at the nests increases as nestlings grow, and can range from 4-20 visits an hour (Sherley 1985).Titipounamu have a facultative cooperative breeding structure; relatives and unrelated neighbours or floaters sometimes assist breeding pairs in feeding nestlings (Sherley 1990b;Preston et al. 2016Preston et al. , 2013)).As a result, nestlings are exposed to the feeding calls of parents and helpers.These feeding 'zip' calls have unique individual features whose parameters can be modified to resemble those of socially close individuals (Moran et al. in review), making them an excellent call type to test for vocal learning.
Over their vocal development, titipounamu nestlings produce acoustically variable calls which are difficult to classify into clear call type categories.Nonetheless, some calls clearly act as begging calls because they are produced before and after the visit of an adult and are often long high pitch calls.Some other calls are part of a multi-note assemblage in which some note resemble adult zip feeding calls (Loo 2021).

Playback experiment
We located and monitored 23 nests.Half were randomly assigned to experimental (playback) treatment (n = 12 nests), and half to the control treatment (n = 11 nests).The only difference between experimental and control nests was that experimental nests were exposed to the playback stimuli.Because our playback experiment was conducted in the wild, both control and experimental groups received visitations from live tutor adults.
Following similar playback methods from Mennill et al. (2018), we exposed nestlings to a diversity of unique synthetic playback stimuli (n = 16 stimuli) and stimuli combinations to increase the probability of nestlings shifting their vocalisations in response.All 16 unique stimuli had distinct and unique synthetic acoustic signatures (Supplementary Fig. S1).Playback stimuli were created by modifying natural feeding calls recorded in previous years (i.e.stretching, inverting, or shifting frequencies; digital recorder Zoom H6 and a shotgun Sennheiser ME62 K6 microphone, 20,000 Hz frequency).Audio manipulation was done using Audacity sound editing software v.2.3.0 (AudacityTeam 2018).We then normalised the amplitude of the playback stimuli to −1 dB to ensure all playback stimuli were played at the same amplitude.
The playback stimuli were played individually or combined in sequences to mimic natural adult feeding call sequences (zip followed by purr calls, Figure 1).Based on previously recorded nests with helpers, adult natural visitation rate was once roughly every 361 s.Playback stimuli at each nest were played with this minimum interval.The amplitude of speakers at experimental nests was set to ∼ 34db (half of the estimated natural amplitude of adult feeding calls) until nestlings were ∼ 11 days old (about halfway through the nestling period) to avoid any potential nest abandonment.The amplitude was then increased to the natural amplitude of titipounamu feeding zip calls at the nests (∼ 68db; determined in the field using a sound decibel metre placed one metre away from a nest entrance).For nests found during the nestling stage, we adjusted the amplitude of the playback stimuli depending on nest stages (i.e. to either ∼ 34db or ∼ 68db).
Because many nests were found during the nestling stage, onset of playback stimuli ranged from 2 days prior to hatching to 25 days after hatching (onset of playback = 11.63 ± 7.81 days after hatching).The overall playback stimuli exposure on nestlings ranged from 3 to 30 days of playback exposure (average = 15.2 ± 7.3 days of playback stimuli exposure).During these playback exposure periods, experimental nestlings received between one to five types of playback stimuli (2.83 ± 1.52 types of playback stimuli per nest) programmed to play at a pre-set interval schedule throughout the day (see below).The highly asynchronous nature of titipounamu nestling combined with the low sample sizes due to predation (Moran et al. 2019) required us to use all possible experimental nest attempts for playback.
For all experimental nests, playback stimuli were played within 1 m of titipounamu nest entrances.Experimental and control nests were at least 25 m apart, so nestlings at experimental nests could not hear playback stimuli from other nests.For low nests (i.e. less than ∼2 m high, n = 7 nests), we deployed a custom-built apparatus that paired a PIT tag reader with a built-in speaker powered by rechargeable batteries.Stimuli were programmed to play at a pre-set interval schedule throughout the day (Supplementary Table S3).When parents or helpers banded with PIT tag bands visited nests, the playback apparatus detected the PIT tag bands and temporarily stopped the playback stimuli to avoid overlapping with live birds' vocalisations.Playback stimuli using this approach were scheduled to start at 0600 and ended at 1800.For higher nests (n = 4 nests), the playback apparatus consisted of speakers (model Zleader Touch Speaker SoundAngel A8) mounted on wooden poles placed within 1 m from nest entrances.The soundtracks were 14 h long and consisted of loops of the experimental feeding call stimuli.The speakers were replaced daily around 0600-1000 with recharged speakers and removed around 1800-2100.One nest received playback stimuli from both methods, one after the other.Our playback apparatus and stimuli described above generated the expected behavioural responses from nestlings as experimental titipounamu nestlings reacted to the playback stimuli by producing begging calls.

Recording titipounamu nestlings at nests
We recorded wild titipounamu nestlings at the nest.Control nests were recorded during the 2018-20 breeding seasons and experimental nests were recorded during the 2019-20 breeding season (Figure 1C).We used Bioacoustics Automated Recorders (BARs) (Frontier Labs version 1.4, WAV format with a sampling frequency of 44,100 Hz and 32 bits sampling depth) to record nestlings.Each BAR microphone was placed within 1 m from the nest entrance and was connected to a cable linked to the BAR unit 10-15 m away from the nest to minimise nest disturbance when changing batteries and SD cards.We programmed the BAR to record from one hour before sunrise to two hours after sunset daily.We stopped recording once nestlings fledged.

Sound processing
We annotated nestling calls from experimental and control nests by selecting a two-hour recording period within one week before their fledgling day.We chose this specific time in the nestling period because some of the earliest onset of signs of vocal learning in vocal learners start from 3-4 weeks after hatching and/or during their nestling period (Brittan-Powell et al. 1997;Tchernichovski et al. 2001;Berg et al. 2012).We used Raven Pro v.1.6.1 software (Cornell Lab of Ornithology 2019) for the annotation process.We could not clearly distinguish between individual titipounamu nestling calls (i.e.enclosed cavity nests with unbanded nestlings).Additionally, no clear nestling call types could be precisely categorised due to their high acoustic variability, so to minimise biases in our call annotation process, we annotated all nestling call types from the selected recording sessions.This annotation process resulted in 3876 nestling calls (mean = 378.45nestling calls ± 127.62 per nest) from 11 control nests, and 6124 nestling calls (mean = 522.58nestling calls ± 254.08 per nest) from 12 experimental nests.

Statistical and acoustic analysis
If experimental nestlings shift their calls towards playback stimuli, we would expect higher resemblance, similarity, and acoustic clustering of the experimental nests to the stimuli.To test for effects of the playback stimuli on nestling calls, we first compared the acoustic differences between groups (control vs experiment, control vs stimuli, and experimental vs stimuli) using permutation tests under the null hypothesis that groups had equal means (Npermutations = 10,000 using an in house code).The permutation null model accounted for the fact that nestling call clips from the same nest were correlated to each other (regardless of treatment).For all sets of comparisons, we measured a variety of acoustic parameters including, but not limited to frequency, symmetry of signal (skew), tonality of a signal (spectral flatness sfm), and kurtosis (peakedness of a signal), using the function specan from warbleR v.1.1.26(Araya-Salas et al. 2017) (Table S2).We compared the mean differences of acoustic parameters of each clip for each group (group1 -group2/number of group pairs; part 1 in Supplemental Fig. S2) and relative mean difference of acoustic parameters ((group1 -group2)/mean average of the full data set; Figure 2A; part 2 in Supplemental Fig. S2).Each acoustic parameter was normalised by its absolute mean.
As an alternative comparative method, we performed cross-correlations of all pairwise combinations of each nestling calls and stimuli using xcorr from warbleR v.1.1.26(option window length = 300; overlap = 90; correlation method = pearson), which resulted in a matrix of 10,016 × 10,016 cross-correlations.To test for treatment differences in dissimilarities score, we compared pairwise sound dissimilarities (dissimilarities = 0.5*(1 -cross_correlations)) using Kruskal's non-metric multidimensional scaling plot (MDS ordination, Figure 2B) generated by MASS (isoMDS) v.7.3.51.6 (Venables and Ripley 2010).The number of dimensions was set to k = 2, the maximum number of iterations was set to maxit = 200.We calculated the Kruskal's stress of the MDS plot based on nestling call pairwise cross-correlations as a measure of the goodness of fit (i.e.represents the similarity between a set of objects; Figure 2B).We then performed a nested permanova test of pairwise sound dissimilarities by randomly permuting nestling call clips while maintaining nest association because individual nestling call clips were not independent of one another (i.e.produced by same individuals in the nest; Permanova npermutations = 1000) (Anderson and Walsh 2013;Anderson 2017Anderson , 2008)).
To further determine whether and to what extent experimental nests shifted their vocalisations in response to the playback stimuli, we examined the effects of stimulus exposure on nestling calls (Figure 3 and Table S1).We plotted the distributions of nestling call cross-correlations for each stimulus (Figure 3A) and across different experimental nests (Figure 3B and Supplementary Fig. S3) using the package ggplot2 v.3.3.2 (Wickham 2009).We then built linear models that related the mean cross-correlations ('proximity') between natural nestling calls and stimuli (response variables) to the durations of stimulus exposure (predictor variables; Table S1).Because each experimental nest was exposed to multiple stimuli, and most stimuli were played at multiple nests, a separate model was built for each stimulus in order to examine its unique effect on calls, and for each model, the cross-correlations between natural calls and the stimulus were averaged over each nest.Hence, the number of samples for each model was equal to the number of control nests plus the number of nests exposed to the specific stimulus.The linear models were built using the function lm from the R package Stats v4.0.2 (RCoreTeam 2016), with the formula (mean cross correlation) ∼ (exposure time).We reported the intercept, slope, p values, sd intercept, sd slope, R-squared and number of samples (Table S1).

Results
When examining specific acoustic parameters, titipounamu nestlings exposed to experimental stimuli showed no detectable call differences from control nestlings.We measured 28 acoustic parameters, and saw no statistically detectable difference in any of them (Figure 2A, Supplementary Fig. S2).For all acoustic parameters measured, the nestling calls from control and experimental nests differed from playback stimuli in the same acoustic parameters in the same direction and to the same extent from the playback stimuli (Supplementary Fig. S2).The multidimensional scaling analysis showed similar results.Cross-correlation pairwise dissimilarities between nestlings at experimental nests, control nests, and with stimuli showed that calls from experimental nests overlapped and did not differ from control nestling calls (isoMDS Kruskal's stress = 0.19; and A, Relative mean differences of selected nestling call features between control nest (n = 11 nests with no playback stimuli received, n = 3876 nestling calls), experimental nests (n = 12 nests with playback stimuli received, n = 6124 nestling calls) and playback stimuli (i.e.n = 16 stimuli).Acoustic parameters were normalised by their absolute mean with p-value-threshold = 0.05.Black contoured circles represent significant p-values.Blue coloured bubbles indicate a higher value for the first represented group while red coloured bubbles show a higher value for the second represented group (e.g.control nestling calls are slightly higher in their dominant frequency slope (dfslope) than experimental nestlings).For a complete acoustic comparison for all acoustic parameters refer to Supplementary Fig. S2.B, Non-metric multidimensional scaling plot of nestling calls' pairwise crosscorrelation dissimilarities from experimental and control nests, and stimuli.The blue points represent the nestlings' calls from experimental nests, the orange points represent the nestlings' calls from control nests and the pink points represent the 16 playback stimuli (isoMDS Kruskal's stress = 0.19; and Permanova F = 25.5, mean random F = 61.3,standardised effect size = −0.77,p-value = 0.77).
When investigating vocal shifts towards playback stimuli among experimental nestlings, we found no evidence that titipounamu nestlings in experimental nests alter their calls towards playback stimuli.Experimental and control nest cross-correlation distributions overlapped and did not reach 1 (Figure 3, Supplementary Fig. S3), and stimuli exposure time had no effect on vocal proximity with stimuli in experimental nests (Table S1).More precisely, the null model for no response (zero slope in the linear models) was not rejected for any of the stimuli (p > 0.1 in all cases), which further confirmed that no vocal shift occurs and longer exposure to playback stimuli did not result in more vocal shifts in experimental nestlings.

Discussion
In this study, we found no evidence that titipounamu nestlings alter their vocalisations in response to playback stimuli during their nestling vocal development.Control titipounamu nestlings did not differ from experimental nests.This might suggest that titipounamu feeding calls are innate and lack the vocal plasticity and vocal matching ability of vocal learners.However, interpretative caution should be applied because of the nature of this experiment and because vocal learning may manifest in other developmental stages or in other call types.
In some species of vocal learners, nestlings learn to imitate adults during their nestling stage, such as green-rumped parrotlets (Forpus passerinus) (Berg et al. 2012) and little ravens (Corvus mellori) (Jurisevic 1999).However, in many species, song and call learning occurs after fledgingafter a long vocal maturation and ontogeny process in which birds take time to accurately copy their learned vocalisations (e.g. up to a year in temperate regions) (Lanyon 1958;Nottebohm 1970;Funabiki 2012;van der Kant et al. 2013;Loo and Cain 2021).This may be the case in titipounamu (Loo et al., in review).In our study, nestling vocalisations were examined during the last week of the nestling period, which may still be a sensory phase in titipounamua period during which nestlings cannot yet imitate sounds (Lanyon 1958;Nottebohm 1970;Funabiki 2012;van der Kant et al. 2013;Loo and Cain 2021).Thus, vocal imitation of playback stimuli may occur once individuals become adults and feed their own nestlings.Recording experimental individuals during adulthood would have provided better resolution to this possibility; however, monitoring nestlings at a later stage was a challenge in our study due to several limitations: (1) wide dispersal rates from natal territory made recapture difficult, (2) fledgling nest memberships were uncertain due to difficulty banding titipounamu nestlings, (3) unusually high predation rates caused low survival rates of nestlings, and ( 4) lockdowns (2020-22 global pandemic) resulted in unforeseen limitation to site access.Repeating our experiment by monitoring the vocal behaviour of fledglings and firstyear-old titipounamu should be considered in future studies.
Further, the onset and timing of our playback stimuli is an important aspect to take into account for best interpretation of our results.Some studies show that birds start assimilating adult-like calls and songs earlier than previously assumed, as early as in their eggs (Colombelli-Négrel et al. 2016;Katsis et al. 2018;Kleindorfer et al. 2018;Rivera et al. 2018).In our playback experiment, titipounamu were exposed to their parents' incubation calls while still in the eggs, which may have influenced their vocal behaviour and vocal preference towards their conspecifics.Furthermore, many titipounamu nests were found during the post-hatching period, which meant that some nests received playback stimuli later in their vocal development.The late onset of some playback experiments at some nests may have influenced the ontogeny of experimental nestlings towards their parents' calls.Future studies should consider repeating the playback experiment starting from the pre-hatching period instead of post-hatching period.
Finally, titipounamu nestlings may have had an auditory preference towards conspecific calls of their own species.In the majority of vocal learners, individuals show a preference towards learning conspecific vocalisations (Marler 1970;Brainard and Doupe 2002).Titipounamu nestlings that were exposed to playback stimuli could still hear conspecifics (i.e.parents and helpers) who were feeding them immediately after vocalising.This constant visual and social adult exposure combined with food intake may have created a positive reinforcement loop for nestlings to imitate their live tutors.Several studies show that young vocal learners favour songs from live and interactive tutors rather than stimuli coming from inanimate tutors, such as from stand-alone playback speakers (Baptista and Petrinovich 1984;Beecher and Burt 2004).Thus, exposure to live social stimulations may also explain why no vocal shift or imitation occurred towards the playback stimuli.Future studies may consider playing stimuli from a vocally distinct population from the incubation period to the fledgling period, and monitor experimental individuals over several years.

Conclusion and future directions
To a large extent, our understanding of which avian clade is capable of vocal learning is based on an absence of evidence rather than evidence of absence.This is because the logistical requirements of traditional methods for testing the presence of vocal learning are substantial and not feasible for many wild species.However, using methods such as in this study, offers novel options for testing the presence of vocal learning in wild populations, particularly if wild birds are tractable at their nest and after fledgling.In our study, there was no direct evidence of vocal production learning during titipounamu nestling period, but signs of vocal production learning may still occur in other developmental stages which need to be further tested.Alternatively, another ideal approach to test for the presence of vocal learning in New Zealand wrens, would be to disentangle the genetic and social contributions of titipounamu vocalisations using cross-fostering techniques (i.e.switching nestlings from nests in populations that use nest boxes).These approaches could be implemented to further test for the presence of vocal learning in titipounamu and will likely provide important insights into the evolution of vocal learning in Passeriformes.

Figure 1 .
Figure 1.Diagram of playback methods conducted at titipounamu nests in Boundary Stream Mainland island, Aotearoa-New Zealand.A, Spectrograms of a natural adult titipounamu feeding zip (left) and a purr call (right).B, Examples of 2 of the 16 synthetic playback stimuli played to nestlings from experimental nests (other playback stimuli can be found in Supplementary Fig.S1).The spectrograms were plotted with R Seewave v.2.1.6(Sueur et al. 2008).C, Simplified diagram showing recording and playback methods used at control and experimental nests.

Figure 2 .
Figure 2. Acoustic comparison showing lack of difference between control and experimental nestling calls.A, Relative mean differences of selected nestling call features between control nest (n = 11 nests with no playback stimuli received, n = 3876 nestling calls), experimental nests (n = 12 nests with playback stimuli received, n = 6124 nestling calls) and playback stimuli (i.e.n = 16 stimuli).Acoustic parameters were normalised by their absolute mean with p-value-threshold = 0.05.Black contoured circles represent significant p-values.Blue coloured bubbles indicate a higher value for the first represented group while red coloured bubbles show a higher value for the second represented group (e.g.control nestling calls are slightly higher in their dominant frequency slope (dfslope) than experimental nestlings).For a complete acoustic comparison for all acoustic parameters refer to Supplementary Fig. S2.B, Non-metric multidimensional scaling plot of nestling calls' pairwise crosscorrelation dissimilarities from experimental and control nests, and stimuli.The blue points represent the nestlings' calls from experimental nests, the orange points represent the nestlings' calls from control nests and the pink points represent the 16 playback stimuli (isoMDS Kruskal's stress = 0.19; and Permanova F = 25.5, mean random F = 61.3,standardised effect size = −0.77,p-value = 0.77).

Figure 3 .
Figure 3.Comparison of nestlings' calls cross-correlation distribution at the stimuli level and nest level showing absence of convergence towards playback stimuli in experimental nestling calls and overlap of distribution between control and experimental nests.A, Comparison of nestlings' call cross-correlation distribution at the stimuli level.For each listed stimuli, the distributions of the nestling call cross-correlations from experimental nests (represented in blue) combine nests that received this specific stimuli.For each stimulus, the distribution of the nestling call cross-correlations for control nests combined cross-correlations from all control nests (n = 11 nests).Values closer to 1 indicate that the distributions are more similar to the playback stimuli.B, Examples of nestling call cross-correlation distributions for each nest that received playback stimulus PB0014zz.Values closer to 1 indicate that the distributions are more similar to the playback stimuli.A complete comparison of the distribution of nestling call cross-correlations at the nest level for each playback stimuli can be found in Supplementary Fig. S3.