Among one of the most remarkable of animal behaviours is the ability of parrots to imitate human speech with great accuracy. This ability, referred to as vocal production learning, is also seen in a number of other animal groups – though rarely with the same level of exact mimicry. Recent research has suggested that this form of vocal learning is made up of many different modules, and that some species, like parrots, have all of the modules, while others may only have a limited number, and limited ability. As a result, the ability to replicate sounds likely exists along a spectrum – from very basic and limited to elaborate mimicry of new sounds. Examining the abilities of species with more rudimentary abilities may provide novel insights into the array of vocal learning abilities and new understanding of when and how vocal learning evolved.

In our paper, we use a combination of innovative and interdisciplinary approaches to investigate predispositions for vocal production learning in a unique group of birds: the New Zealand Wrens. 

The New Zealand wrens occupy a key phylogenetic position to elucidate the evolution of vocal production learning in Passeriformes. They have always been assumed to be vocal non-learners, but are they really?  

First and foremost, New Zealand wrens are part of the Passeriformes, which share a common ancestor with parrots. They are the earliest diverging passerine group pre-dating the oscine/suboscine divergence and situated between the oscines and the parrots. An understanding of New Zealand wrens’ vocal learning abilities would shed light on when vocal learning likely evolved. 

There are a number of indicators that suggest that the New Zealand wrens may have rudimentary vocal learning abilities. First, although New Zealand wrens lack intrinsic syrinx muscles (often associated with advanced vocal learning), their lack of intrinsic muscles is perhaps the result of a loss. Indeed, a fine and detailed observation of their syrinx morphology shows an unusually large broad base on which intrinsic muscles would have otherwise been attached. Selective forces on New Zealand wrens’ extremely small size (~5-7g) may have resulted in the loss of their intrinsic muscles. But, if vocal production learning was present in their ancestors, New Zealand wrens may have lost the intrinsic muscles and preserved the neurobiological basis for it. 

Another important aspect of New Zealand wrens is that, similar to many Austral birds, they live in complex social groups and are likely to have experienced strong social selection - two driving forces that we think contribute to the evolution of vocal production learning. New Zealand wrens are sedentary cooperative breeders that share foraging grounds with neighbors (often related), and they have helpers that participate in the rearing of nestlings at the nest. Most helpers are kin but some are not. Such a complex social environment may provide the right selective pressures for vocal production learning to evolve.

This leads to the next aspect of New Zealand wrens’ behaviour that aligns with behavioural features found in vocal production learning. Social interactions in New Zealand wrens involve the production of a complex call repertoire with many types of short high pitched calls. Calls often do not receive the same level of interest as the broastcast songs – as they can be extremely short in duration, more conspicuous and less diverse in their vocal features, making them perhaps harder to study. As a result, calls have been assumed to be non-learned vocalizations, but there is increasing evidence that many call types are learned.

Finally, New Zealand wrens have a relatively long vocal developmental period for birds of their size. Vocal learners need a similarly long exposure to their vocal tutors to develop accurate copies of their tutors, which they use to survive and navigate complex social environments.  

All of this means that there are good reasons for testing for vocal learning in this group. But one reason it hasn’t been done before, is that traditional methods for testing for vocal learning are not possible in this species. They are threatened and declining, and do not survive in captivity. So traditional methods, such as rearing birds in controlled environments in the lab, would not suit. Instead, we had to develop unconventional and indirect techniques that would allow us to address the question in the field. 

In our paper, we focus on detecting any forms of evidence for call imitation in adult riflemen, one of the only two extant species of New Zealand wrens. We first show that the calls of the rifleman have unique individual vocal signatures. In other words, each rifleman has a distinct voice. We also find that they maintain a nest vocal signature. This is particularly relevant to our search for predispositions for vocal learning, as kin and non-kin from the same nest sound more similar to one another than any other groups, including relatives that do not live close. This behaviour is known as vocal convergence, and it does not result in an exact sound matching, but it likely requires some level of auditory feedback and vocal modulation. This behaviour seems most similar to what is commonly known as vocal accommodation in human linguistics, an ability that enables speakers to adjust their ways of speaking in different social settings, such as exposure to other dialects and hierarchy. 

Demonstrating whether this form of vocal convergence is learned, especially in a wild population of animals, is a difficult task. We attempted to solve this problem by breaking the phenotypic variances of rifleman calls down to their social proximity and genetics and compared the ratios to another vocal learner’s phenotypic variances. We find that social proximity plays an essential role in explaining the phenotypic variance of rifleman calls. Second, those phenotypic ratios were similar to those of zebra finches, a well-studied vocal learner. This clearly indicated to us that New Zealand wrens’ calls were strongly shaped by their social environment, similar to what has been reported in vocal learners. 

Our finding opens new avenues to explore vocal predispositions for vocal learning abilities in species with subtle or rudimentary vocal learning abilities. The presence of vocal convergence in rifleman also reveals that this behaviour was likely present in the common ancestor of songbirds and parrots (Psittacopasserans). Future research is needed to determine the underpinning neurobiological mechanisms for call convergence in New Zealand wrens. Further exploring the evolution of vocal convergence, a behaviour that allows for considerable vocal flexibility in social contexts, may reveal new insights into the evolution of vocal production learning. 

Research article: Moran et al. Social proximity and vocal convergence shape the call feature of one of the most basal Passeriformes, New Zealand Wrens. Communications Biology (2024) 7:575. https://doi.org/10.1038/s42003-024-06253-y.