The Zebra Finch is commonly found over much of central Australia, as well as on the Lesser Sunda Islands to the northwest (Plate 33). This colourful and energetic species is not a true finch, but a member of the family Estrildidae, a clade that originated 16.5 million years ago during the Himalayan uplift.
The northward-moving Indian plate first made contact with the Asian plate around 50 million years ago, although its maximal effect did not occur until the Miocene. The ongoing result was the formation of the Himalayas and the Tibetan plateau, which in turn led to the establishment of the southern Asian monsoon system, the region’s large rivers, and the Chinese deserts. It is likely that these dramatic environmental upheavals encouraged the ancestors of the Zebra Finch to radiate from their origins to reach Africa, south Asia, Australia and the South Pacific islands.1 At first, the early estrildid finches were unable to colonise the southern continent as the area remained covered in rainforest and lacked suitable habitat. But Australia was drifting northwards, and as it did so, its climate changed. By 10 million years ago, the continent’s Gondwanan rainforests had begun to recede, to be replaced by eucalyptus and acacia trees and large tracts of open grassland. These environmental upheavals provided the trigger for several waves of invasion by songbirds from southern China, among the first of which were the ancestors of the seed-eating Zebra Finch.2
Despite these fascinating evolutionary insights, it is the Zebra Finch’s role as the avian equivalent of the laboratory rat that concerns us here. For several aspects of the species’ biology – long-term pair-bonding, early breeding with 3–4 clutches per year, and its relative longevity – have made it the preferred subject for many biological studies. In particular, the Zebra Finch is the only tractable laboratory model suitable for the investigation of the genetics and evolution of birdsong. It is an ongoing story, but one that suggests that the brain circuits controlling complex traits like vocal learning have only a limited number of ways to evolve. At first sight, this may seem an arcane field of research, but it has turned out to be highly relevant to the understanding of human speech and its disorders. For this reason, the Zebra Finch’s story is an important one, well beyond its central role in highlighting the evolution of birdsong.
Vocal skills and the evolution of the syrinx
The oscines’ defining and most endearing feature is their ability to sing, an apomorphic characteristic prosaically encapsulated in their everyday collective name ‘songbirds’. The beauty and sheer exuberance of their melodies, especially the evocative cadences of the Atlantic Canary, Eurasian Skylark and Common Nightingale, have inspired poets and musicians throughout history. The complexity of such performances can be stunning. As Mark Cocker highlights, a male Nightingale possesses ‘as many as 250 different phrases compiled from a repertoire of 600 basic sound units’. Furthermore, ‘in the passages of song, the phrases are drawn together and sequenced in a variety of ways so that each performance is a unique composition, never to be repeated.’3
In contrast to most other birds, oscines are absolute masters of vocal learning, or mimicry, an innate ability that can result in a marked embellishment of their fundamental rhythmic baseline. On hearing a vocal duel between Song Thrushes in spring, Cocker wrote:
It is a fabulous noise that gains momentum as the season draws on, with a vocalist adding new motifs to his repertoire. A bird borrows elements from the others that it can hear, and you can imagine these scraps of melody being passed all round the country as one song thrush tosses the sound-torch to its song thrush neighbour.’4
Vocal learning reaches its apogee with the insatiable mimicry of the Marsh Warbler, a species that breeds in temperate Europe and winters in sub-Saharan Africa. The male’s song is an ornithological collage, a musical jukebox, composed of various snippets from up to 200 avian sound-bites, learned en route from their breeding territories to their wintering grounds. For Marsh Warblers routinely incorporate phrases from common European species, including tits, raptors and warblers, coupled with flourishes borrowed from African species such as drongos, doves and bee-eaters.5
Vocalisation is likely to have evolved early after the oscine–suboscine split, as mimicry is a well-developed trait in several basal oscines. Indeed, the world’s most celebrated vocal mimic is one such passerine, the Superb Lyrebird. This shy, ground-stalking species can imitate not just forest sounds, but also mechanical noises such as camera shutters, chainsaws and car alarms. Mimicry also plays a role in the courtship of another basal oscine, the Satin Bowerbird. Until recently, it has always been assumed that female bowerbirds selected mates based on their building ability – the symmetry and decoration of the bower (see The Bowerbird’s Story). It now seems that architecture is not everything. Gerald Borgia’s team at the University of Maryland has revealed that male Satin Bowerbirds ultimately make a more intimate appeal to females by approaching closely and ‘whispering’ an enticing mimicry of other Australian birds.6
Such vocal skills are now known to have evolved over millions of years as the result of sexual selection. But why should the female’s choice of mate be based on a male’s adeptness at singing and mimicry? Put another way, how can vocalisation be a characteristic that favours a species survival? The answer, it seems, is that vocal prowess mirrors genetic quality. A male oscine’s vocal learning skills are a reflection of its juvenile health, freedom from disease and parasitic infection, at a time when learning abilities develop. Also, males with more complex songs have superior cognitive skills and a greater learning capacity.7 The female, therefore, is selecting a mate with ‘good’ genes and, as a consequence, is enabling the male’s singing proficiency to be passed on to the next generation. Rival males try and outdo each other, engaging in a form of vocal ‘arms race’ where the most varied performance, containing the best mimicry, gains the optimal territory and, with it, the chance to mate. As we saw in The Bird-of-Paradise’s Story in relation to other aspects of male performance, female choice eventually leads to a runaway selection for vocalisation, as the genes underpinning the trait are the ones most likely to survive.
The song and mimicry skills of oscines result from the evolution of two linked structures: a complex syrinx and a dedicated neural control network, known as the song system. The syrinx in oscines is located at the base of the trachea, or windpipe, where it bifurcates into the two main bronchi that enter the lungs. It is a complex structure composed of extrinsic and intrinsic muscles that control the airflow, and membranes, or labia, that act as sound-generating structures. The mechanism of sound production is very similar to that used by ourselves, except that the oscine’s syrinx is a bilateral structure. It has two sound sources instead of one, with each half possessing a pair of membranes. Both sides can vibrate independently and are, to some extent, separately controlled.8 When a bird sings, the air from the lungs is forced through the syrinx, vibrating the membranes to cause a sound. Songbirds have up to four pairs of intrinsic muscles that can alter the organ’s configuration, an anatomical arrangement that correlates with song complexity and vocal learning. Also, evolutionary modifications to the membranes’ extracellular tissue (in effect, the amount of collagen and elastic fibres they contain) have facilitated the acoustic distinctiveness of oscines. It appears that the degree of asymmetry of the protein content in opposite membranes and the variation in their layering structure equates to the range of possible frequencies.9
Nobody knows how or why the early passerines split to give rise to the oscines, a suborder with a more complex syrinx. A likely scenario is that at some time during the early Palaeogene, a primitive passerine evolved the ability to produce a greater range of frequencies or combinations of sounds. Such an event would have resulted from recombinations or mutations in the founder’s genetic code that ultimately affected the syrinx’s anatomical structure and function. For example, the novel genotype could have resulted in a slight change in the position of the syrinx, or altered the arrangement or number of the syrinx’s muscles, or modified the biochemical composition of the membranes. These phenotypic changes were then subjected to positive selection, probably by female choice, so that subsequent generations inherited the modified syrinx. Evolution is an ongoing process, and genetic changes are occurring continuously. Any further mutations that enhanced a male’s singing ability would have become increasingly standard within the population. Eventually, after thousands or even millions of generations, a more sophisticated syrinx evolved, one with a capacity for diverse song and accurate mimicry. While something like this almost certainly occurred, it should be stressed that it remains an imaginary account, since the precise evolutionary pathways are unknown. What is certain, however, is that the oscine’s syrinx evolved sometime during the Palaeogene.
The song system
The ability of oscines to vocalise is also dependent on a set of specialised brain cells, or nuclei, located in the forebrain, that are collectively known as the song system. The evolution of these discrete, interconnected nerve cells was a definitive event in the history of songbirds and probably occurred at the same time as the development of their complex voice box. In addition, the song system must have evolved soon after the suboscine–oscine split since, although it is present in all oscines, it is missing from suboscines.10 As we will see, however, at least one suboscine, the Eastern Phoebe, possesses a vestigial homologue of the song system, a finding that could have evolutionary implications.11
Although the song system’s structure is exceedingly complex, a simple outline will help readers appreciate its evolutionary significance. The controlling network consists of seven discrete structures that interconnect to form two major neural pathways. The first pathway, the cortical motor pathway, originates in the high vocal centre or HVC nucleus and controls the vocal and respiratory muscles indirectly, via a specialised cluster of nerves termed the robust nucleus of the arcopallium. This pathway ultimately connects to the syrinx and is essential for the production of learned vocalisations. The second pathway, known as the anterior forebrain pathway, acts in conjunction with area X and facilitates the acquisition and imitation of songs. Crucially, the anterior forebrain pathway also underpins the syntax and social context of learned vocalisations. As young male Zebra Finches acquire their vocal skills, the volume and number of neurones in the various song nuclei increase compared to the rest of the brain and the brains of non-learning females. As one might predict, damage to the pathways when the bird is young disrupts its ability to vocalise, such that it may never be able to acquire its distinctive song.
Vocal learning is an involved process and takes many weeks to master. It commences with a phase similar to the babbling of human infants, called ‘subsong’, that consists of variable, low-amplitude sounds. From this raw material, imitations of its parents emerge, copied mostly from the father. Indeed, if young oscines are removed from the nest and reared in isolation, their song development is curtailed, leaving the bird with nothing but its innate subsong. Through trial and error, the fledgling’s imitations become recognisable and, once perfected, they become less and less variable. Nevertheless, readily identifiable dialects may occur in some species, acting as local cultural traditions. Vocal learning by sensory feedback ensures that by the time the bird is sexually mature, its song is robust enough to defend a territory and woo a mate. It is now known that the accuracy of a songbird’s mimicry depends on the release of a cocktail of growth factors from the song system nuclei, the most important of which is brain-derived neurotrophic factor (BDNF). This protein is essential for nerve regeneration, and its administration to juvenile birds markedly improves their song-learning abilities.12 Whether sexual selection by females is in any way governed by male levels of BDNF, however, remains to be seen. Like most birds, oscines possess an extensive repertoire of unlearned calls that are important in close-range communication. Recent experiments, incorporating wireless brain monitoring of Zebra Finches, have highlighted that it is the nerve cells controlling these innate calls that have evolved to give rise to the complex neural networks.13
The songbirds belong to one of only three related avian groups known to have acquired vocalisation through imitation rather than instinct, the others being parrots and hummingbirds. While parrots are synonymous with vocal mimicry, hummingbird vocalisation has only recently been demonstrated. The restriction of this behavioural trait to these three clades is surprising, since hummingbirds are not closely related to either parrots or songbirds. What is even more remarkable is that they have all evolved the same neural network: a song system consisting of seven interconnected forebrain nuclei. Psittacine brains, however, appear more complicated and contain a song system within a song system. While their ‘core’ song system is similar to that of songbirds and hummingbirds, their ‘shell’ song system is unique – and this may account for the advanced mimicry skills of parrots.14
So what do these findings tell us about the evolution of birdsong? The most likely explanation is one of convergent evolution, whereby avian vocalisation evolved on three separate occasions over the last 65 million years. According to the neurobiologist Erich Jarvis, this could have occurred if an adjacent, pre-existing system in the avian brain that controls motor learning skills, which also has seven active areas, duplicated to take control of vocalisation.15 Indeed, he also believes that brain duplication could be a general mechanism to explain the evolution of other complex behavioural traits. If convergent evolution is indeed the explanation, it implies that songbirds, parrots and hummingbirds all evolved a core system independently, and that parrots went on to develop an extra shell system. An alternative hypothesis is that the song system evolved only twice, in hummingbirds and the common ancestor of parrots and passerines, and that it was subsequently lost in suboscines. Supporters of the latter scenario emphasise that the Eastern Phoebe has a rudimentary song system, while several bellbird species have been documented to develop distinct dialects.16 Future experiments, however, will be necessary to resolve these two hypotheses.
Human parallels
One rationale for discussing the anatomy of the song system in some detail is that many of the structures are homologous to areas in the human brain known to be essential for the production and comprehension of the spoken word. Area X, for example, is homologous to the human basal ganglia, with similar connections, cell types and response to neurotransmitters, while the HVC nucleus is a homologue of Broca’s area. This remarkable neural parallelism between birdsong and human expression has allowed scientists to use the Zebra Finch as a model system to understand the molecular control of speech and communication disorders. Already a number of genes, whose expression levels change when finches sing, have been linked to medical conditions. The best understood is the Forkhead box protein 2 gene (FoxP2), a ‘master gene’ that encodes a transcription factor that controls the expression of hundreds of other song-related genes. Mutations in its DNA binding site result in a severe speech disorder in humans, known as developmental verbal dyspraxia and characterised by incomprehensible talk with linguistic and grammatical deficits.17 Not surprisingly, FoxP2 has been popularly dubbed the ‘language or grammar gene’. Other avian vocalisation genes, ones that encourage neuronal connections between the song centre and the nerves that control the syringeal muscles, have been linked to autism and dyslexia.18 It may soon be possible to develop novel therapeutic strategies for these speech disorders, because scientists now have the skills to insert human disease-causing genes into avian genomes and explore their function in more detail.
Recently, Andreas Pfenning, at the Duke University Medical Center, together with other members of the International Avian Genome Consortium, investigated how many genes might be involved in song learning.19 Using laser micro-dissected song nuclei, they compared gene expression levels or transcriptomes in samples taken from all three groups of vocalising birds, as well as from two non-vocalising species, a dove and a quail. Also, laser-captured samples were obtained from donated human brains as well as from our non-vocalising cousins, the Rhesus Macaque. What Pfenning’s team discovered was that an identical set of more than 50 genes showed the same expression levels in the song system of vocal learning birds and the homologous speech areas in humans. Such changes did not occur in the brain tissue of birds that do not vocalise, or in the non-human primates that do not speak. In other words, if a gene’s activity was increased in humans, it was also increased in songbirds, hummingbirds and parrots but not in any other avian order. What is surprising is that the same genes should be involved in bird and human vocalisation, since our common ancestor lived more than 300 million years ago. Such a finding supports Jarvis’s belief that brain circuits for complex traits may have only a limited number of ways in which to evolve.
To conclude the Zebra Finch’s story, let me quote Fernando Nottebohm, the Argentinian neurobiologist who discovered the avian song system:
It may well be that our best understanding of how complex skills are acquired and how broken circuits can be fixed will come not from humans, or other primates, but from the way birds learn their song.20