I am very bad at remembering and recognizing melodies. It doesn’t hurt too much: I am so amusical that I rarely notice this deficiency. It was a much bigger problem, though, when I was a child – and it’s because I was a birdwatching child.
Recognizing birds by their songs was a dramatically hard thing for me. The mnemotechnic that I found the most useful was learning folk transcriptions of melodies. Take a yellowhammer as an example – song of this yellow beauty was traditionally transcribed in Polish as nie będzie suchej kobyle niiic (in a loose translation – the skinny mare will be fiiine). In English, the same song is transcribed as little bit of bread and no cheeese . The absolute absurdity of a bird comforting an owner of an underfed horse was bold enough to make me remember this song well – I can now recognize a yellowhammer’s song wherever I go.
But this song motif is not exactly the same everywhere – yellowhammers have their dialects. Czech scientists initiated even a citizen science project – you can record a yellowhammer in your neighbourhood and send them your recording. Authors use this data to make a map of dialects – you can find the interactive version here; click on a coloured point to hear the dialect.
Songs of birds can also differ in time. Yellowhammers in my hometown might sing a song that is slightly different from the one that my grandmother could hear in her childhood – this phenomenon of cultural evolution was observed in savannah sparrows living on an island. They were recorded for 30 years; it turned out that their song changed – they added some new elements and modified others.
Both phenomena – dialects and cultural evolution – are caused by the fact that yellowhammers and savannah sparrows, just like 4000 other species of songbirds, learn their song from other individuals, more or less like we learn our language. And learning, as you may know, is never perfect; some notes might not be copied properly by some birds and, in effect, a song will slowly change in time and space, as some birds migrate.
What is interesting, though, is that even though yellowhammers have different dialects and their song may change in time, it is anyway always clearly recognizable as a yellowhammer song. Birdsong, though learned, is very often surprisingly stable and universal.
Its robustness was shown in many studies. If you isolate a young songbird from its peers, making it unable to learn the song, it will usually develop a chaotic, noisy isolate song. But there is an order in this chaos – isolate song has some characteristics of a species-typical song – e.g. song duration, numbers of notes in a song and the duration of intervals between notes.
Much more spectacular proof of the stability of birdsong is a study done in 2009 by Olga Feher and others. They isolated young zebrafinches and let them develop an isolate song. After it was developed, they used those isolated singers as tutors for the next generation of birds. Afterwards, zebrafinches from the second generation became tutors for the next generation. And well, results were astonishing – birds that learned isolate songs were making it more species-typical in every generation; and after three generations of learning the song was pretty similar to the normal song of a zebrafinch. It happened even though none of those birds had ever heard a normal zebrafinch song!
In my last post I wrote that we should call such stable behaviors robust behaviors and avoid terms like instinctive or innate. This time I would like to show what are mechanisms that ensure the robustness and stability of such behaviors.
In a search for the template
As I wrote previously, whenever we face an aspect of animal behavior that seems to be independent of learning, universal and developmentally stable, people usually assume that it has to be somehow hardwired in the brain, with a scheme for this wiring being encoded in the genome. Aforementioned stable aspects of birdsong are no exception; as Douglas Nelson and Peter Marler wrote in their paper on species universals in birdsong:
“[Our model] postulates extensive pre-encoding of information about species-specific song structure, embodied in innately-specified brain circuitry.”Marler, P., & Nelson, D. (1992)
Authors mention innately-specified brain circuitry for an obvious reason: as we all know, the brain controls behavior; if genes encode some aspects of behavior, they must do it by encoding how neurons connect with each other. Genes influence behavior by their direct or indirect influence on the brain development, neuron properties and wiring – that is how the story goes. As another example of this conviction let me cite Kevin Mitchell about human behavior in his book Innate:
“Somehow, in the molecules of DNA in a fertilized egg from any of these species is a code or program of development that will produce an organism with its species-typical nature. Most importantly, that entails the specification of how the brain develops in such a way that wires in these behavioral tendencies and capacities. Human nature, thus defined, is encoded in our genomes and wired into our brains in just the same way”.Kevin Mitchell, Innate: How the Wiring of Our Brains Shapes Who We Are
And it is obviously true. Genes can have an influence on a bird’s brain to make some of its parameters universal and stable in many ways.
First of all, genes may influence the properties of neurons that take part in song generation. By changing e.g. expression or structure of different ion channels you can easily obtain neurons that might spike with a different frequency. A good example is the case of fruit flies’ courtship song. Two species of Drosophila – Drosophila simulans and Drosophila mauritiana – produce courtship song (which is produced by wing vibrations) that differs in carrier frequency by 9.7 Hz. It turns out that this difference can be explained by an intronic retroelement present in the gene coding for slowpoke calcium-activated potassium channel in D. simulans, which in some way changes properties of neurons that are involved in song generation.
In the case of birds, by influencing properties of neurons that take part in the generation of a song, genes may limit the scope of songs that are possible for a bird to sing. But they can also influence the properties of neurons that process sensory information – cells in auditory areas might be so tuned that they respond preferentially to species – specific sound frequencies. In this way a bird would have a limited scope of songs that it can learn from a tutor. Birds prefer to learn songs of their own species. Even if they learn a song of another species, they will make it sound similar to what they would normally sing. Some authors postulated the existence of innate perceptual filters, that would just filter out songs of other species. You can make an auditory region of the brain in a way that will respond much more strongly to your species song – and it seems to be the case.
In those – and many other ways – genes, by influencing properties of the nervous system, channel the development of a song to ensure it’s stability. But can we say that a yellowhammer’s song is really encoded in those genes? As I want to show you, we probably cannot, and it’s because there is much more to a song generation and learning than just the brain.
Bird has a body
If you were not born on another planet, or – just like poor birds in the aforementioned experiments – were not kept in isolation, you must have seen the picture of Darwin’s finches’ beaks. Yes, they differ, and thanks to their differences birds can occupy different niches in a harsh environment of the Galapagos islands. But what is important in our case is that the size of the beak influences also the bird’s song.
Birds actively change their beak gape rapidly during singing in a way dependent on the frequency of a song, increasing the gape when the song’s frequency is higher and vice versa. The function of those rapid adjustments is filtering – the beak acts as an acoustic filter and the change of its gape allows it to serve its filtering function over a wide range of song frequencies. If you add an extra mass to a canary’s beak, limiting bird’s ability to move its beak rapidly and precisely, the result will be a song of lower tonal purity.
As you might expect, having a naturally heavier beaks should also influence the song – and it does; as Jeffrey Podos has shown, Darwin finches with larger beaks produce songs with lower repetition rates of syllables and narrower frequency bandwidth than birds with smaller beaks. The same is true for neotropical woodcreepers; in blue cardinalids a length of a beak is negatively correlated with a note rate.
A bird’s song repertoire may thus me restrained by the size and shape of its beak. But the beak is not the only morphological trait that may constrain a song of a bird.
A bird’s vocal organ is the syrinx. It is located near the base of the trachea; the sound is generated when airflow causes vibratory tissues – membranes or labia – to vibrate. Morphology of syrinx differs between different groups of birds. It is especially elaborate in songbirds – their syrinx is a duplex voice organ, with two pairs of labia in each bronchus that can be controlled independently.
Sound can be shaped by muscles that change the tension of the tissues in the syrinx. It seems, though, that the properties of the song depend only on the commands received from the neuromuscular system. Some authors suggest that vibrating membranes or labia in bird’s syrinx may act like coupled nonlinear oscillators, producing a range of phenomena typical for such systems, such as sudden frequency jumps. Carel ten Cate and Gabriel Beckers claimed that song of doves from genus Streptopelia exhibits acoustic phenomena that are caused by such nonlinear dynamics intrinsic to their syrinx and may not be a result of direct muscle control. What is interesting, they speculate that inter-species differences in vocalizations may be caused by minor changes in e.g. the morphology of the syrinx.
One of the key features of nonlinear systems is that small and gradual changes in control parameters can cause large, sudden, and qualitative changes in dynamics. If true in bird song, the intrinsic dynamics of the sound production organ itself would provide a source of major and qualitative acoustic variation. […] Seemingly strong interspecific differences, e.g. between the tonal coos of S. risoria and the noisy ones of S. orientalis or S. tranquebarica, might thus have resulted from minor changes in underlying mechanisms of vocalization, without the need for large changes in syringeal structure or control mechanisms.Beckers, G. J. ., & Cate, C. (2006)
A similar phenomenon was observed in case of calls of Xenopus frogs. These calls frogs are produced by the larynx built of two apposed discs. Fast separation of those discs excites the larynx and the tissues around, resulting in the generation of harmonic frequencies. In a Xenopus male mating call, each sound pulse is composed of two frequency bands produced simultaneously, called dyads. One band, called DF2, has a higher dominant frequency and the other – DF1 – lower. DF2/DF1 frequency ratio is different in different Xenopus clades and conserved within a clade. What is interesting, the properties of those dyads are intrinsic to the larynx and not dependent on the neural control: you can obtain dyads identical to natural ones by stimulating isolated larynges. As authors write:
Both species-specific individual DFs and the clade-specific dyad ratio are thus intrinsic to the larynx rather than the result of laryngeal or respiratory muscle modulation by neural circuitry. Which, as yet unidentified, characteristics of laryngeal tissue geometry and properties result in species-specific DFs and their ratios remain to be determined, but are likely to reflect a common tuning mechanism in descendants of ancestral Xenopus speciesKelley, D. B., Hall, I. C., Tobias, M. L., Elias, D. O., Kwong-Brown, U., & Elemans, C. P. (2019)
It might be thus possible that the aspects of a bird’s song that we label as innate may stem not from the way its brain is hardwired, but from the way its syrinx is shaped.
There are more ways in which the bird’s body can shape its song. The most obvious one is the body size – small birds are unable to produce songs of a very low frequency, as the production of such sounds requires a bigger sound generating mechanism and is more costly energetically. Small birds also have smaller lungs and, as vocal production and respiration are coupled, they are not able to sing longer sounds without a break.
Many aspects of a bird’s song that are labelled as innate in isolation studies may not in fact stem from hardwired templates, but be a result of the way the bird’s body is built. But there are also other factors – outside of a bird’s body – that may ensure the stability of the song in a natural environment.
Bird has an environment
As I mentioned, there are perceptual filters that help birds to learn only their species-typical song. But it can be said that there are also filters that are external to the animal – its environment can also act as a filter. It is because different environments carry – or degrade – different sounds differently.
In forests, for example, sounds can be reflected off canopy or echoed off the trunks of trees. Such reverberations are especially degrading sounds with rapid frequency modulations, while pure whistles are usually much less influenced by them. Environmental factors, such as temperature and humidity, also influence sound transmission – humid environments may enhance sound transmission, and temperature gradients (e.g. when air nearby the ground is warmer) may lead to distortions of sound transmission (e.g sound shadows). Position of the animal also matters – the transmission low-frequency sounds is disrupted when a bird is positioned closer to the ground due to ground interference.
Birds adapt their songs in a way that allow them to be transmitted most efficiently through the environment. Woodland birds use a frequency window of 1600 – 2500 Hz to optimize their song transmission in a forest; they also use pure tones more frequently than birds living in open spaces. Species living in the field, on the other hand, use more trills and frequency-modulated signals. Song properties vary also with the substrate of the environment – birds living on marshes produce songs of lower frequency than those living on grasslands, as marshes absorb reflections that cause ground interference that distorts low sounds.
The environment in which a bird is living will thus to some extent filter sounds that arrive at its ears, again limiting the set of songs that it may learn and later produce. The effect of such environmental filtering might be strengthened by a bird’s ‘hardwired‘ preference to learn song less degraded by the environment, as was shown by Stephen Nowicki in swamp sparrow. I would speculate that birds might also be able to hear how song elements produced during the learning phase are transmitted. They may then choose the elements that e.g. do not produce echoes.
Such environmental filters might not be restricted only to the physical environment. It was shown that nests of Australian passerines that are build of mud can effectively attenuate noise that reaches the inside of a nest. What is more, in tree swallows that make nests in cavities, the geometry of the nest influences the frequency of nestlings’ calls and their length. As the auditory system of nestlings and embryos is responding to conspecific songs and the early auditory experience influences adult vocal behavior (examples here, here, and here), one could speculate that the nest structure may also have an impact on a bird’s song, e.g. acting as a filter that attenuates certain sounds.
Social interactions may also shape the song more directly. In many songbird species males initially produce many different songs and retain only some of them in their adult life, discarding the rest. One criterion for inclusion of a song element might be how well it matches the song of a tutor that was memorized by a bird. But it is only a part of the story – birds construct their song repertoire also during interactions with other birds, males and females.
In brown-headed cowbirds, male selects song elements that it will include in his final repertoire based on the reaction of a female. Females produce subtle signs – wing strokes and beak gapes – when a male is singing; males retain those elements of their repertoire that elicited the response of a female and discard those that did not.
In the case of male field sparrows, it is other males’ behavior that matters. When young birds settle on new territory, they sing a few different song types. They engage in vocal interactions with other males around and, after some time, retain only the song that is most similar to the male that is most active in their neighbourhood. Similarly, song sparrows initially overproduce song types, only to end up with a repertoire of songs that are also sung by many other males around.
As a result of this bias, songs sung by birds in a given area might become more uniform; songs that differ too much from the most popular type are discarded. It is yet another mechanism that may prevent bird’s song from diverging too much from a species-typical pattern.
I think that we tend to view some parts of the natural world as somehow fluid and amorphous. Every time we are confronted with a natural phenomenon that is stable across generations, universal or developmentally robust, we tend to assume that it has to be encoded in some stable structure. This way of thinking might be traced back at least to the famous Schrödinger’s What is life. As Lenny Moss writes on the Schrödinger vision:
Schrödinger begins with a naive notion (but perhaps justifiable for his time) of the cell as a disorganized bag of atoms and argues his way to the need for a solid-state “aperiodic crystal” to serve as that bedrock of order, continuity, and heroic resistance to entropy which makes life possible.What Genes can’t do, Lenny Moss
This bedrock of order is now usually thought to be DNA, but the way of thinking stays the same. In the case of animal behavior, this bedrock can be hardwired neural circuits. As I wrote above, people who try to explain the stability of bird’s song assume that somewhere in the brain there must be a template, some pre-encoded information that will ensure the development of species-typical song even in the face of developmental perturbances.
And it is to some extent true – it is a truism to say that genes influence the way brain is wired during the development, and depending on the variant of an ion channel you can have neurons that can produce patterns that constrain the range of possible songs that a bird can produce.
But – what I wanted to show you – it is not the whole story. The bird’s song repertoire is also constrained by the way a bird’s body is built. A small bird will not sing a low-frequency song – just because it’s physically impossible. But the shape of the song will be also influenced by its vocal organ morphology and the size of its beak. The songs bird will learn are, on the other hand, constrained by the environment that transmits different sounds better or worse. And even if bird memorized songs that are different from the species-specific canon, interactions with other birds – males and females – will help to prevent any attempt to be original. There may be no single template, hardwired in the brain, but the whole set of filtering mechanisms*, internal or external to the bird, that ensures the stability of the song.
Well, you can say, everything can be anyway traced back to genes, as genes determine the way body is built, the preferences of female, the tendency of a bird to match other bird’s song and it’s preference for a given type of an environment that it will inhabit. It’s just that there genes’ influence on behavior cannot be limited to the genes coding for the wiring of brain parts responsible for song learning and generation.
And if its the message that you will get from why this text, it’s good enough. What are the implications? If you are looking for genes related to different behaviors, you should not limit your search to genes that are directly related to how the brain is wired during development; you may find that there are genes beyond that set that also influence behavior. It may be especially important in studies that aim to find genes that make behavior of two closely related species differ. As in the aforementioned Drosophila study, the reason might be a difference in a gene encoding an ion channel, but in some cases one might find genes related to body shape or organ morphology that are responsible for the difference.
But, in fact, you cannot reduce those aforementioned behaviors to genes only – to start with, female preferences for different types of male songs are not really hardwired and require experience to develop. Bird’s experience and social learning also influence the selection of a nest’s site (for examples in long-tailed tits, mountain bluebirds, pinyon jays and collared flycatchers).
Instead of putting genes as a bedrock of order, we can assume that order, stability and robustness of behaviors can be ensured by interactions of many factors, internal and external to an animal, that recur usually in every generation and that are reliably present in a life of every individual. If one of them fails – for example, when a bird doesn’t have enough tutoring from a conspecific at a young age – others may help it develop a trait properly. A bird can, for example, develop different song types by improvisation and then select only those that are close enough to those sung by other males in a neighbourhood.
A good metaphor for this way of thinking might be tensegrity structures, in which the stability is achieved not by means of building stable support elements, like columns or pillars, but by balancing the counteracting forces of elements with continuous tension and discontinuous compression. There is no bedrock of order in tensegrity structures, just the dynamic interaction of elements that results in a stable structure.
That’s how I view the causes of stability of many animal behaviors. Animal, with its body and environment, can be – at least to some extent – viewed as a system, and only by looking at different parts of this system we can fully understand the genesis of behavior. There is more to it than just genes and brain. There is no single place where the yellowhammer’s song is encoded; the information necessary to ask for a little bit of bread and no cheeese is distributed.
* I really doubt that they can be really named mechanisms, but I did not find a better name for them.