Assymmetrical use of vocalisations? (1 Viewer)

Vocal differences are increasingly relied upon in avian taxonomy, to the point that Birdlife recently split at least two species (an African cuckoo and barbet) pretty much entirely on the basis of vocalisations. As far as I can tell, the morphological / plumage differences involved are trivial, and no genetic analysis was undertaken. I don't take issue with this - Birdlife has a "quick-and-dirty" approach to species-level taxonomy, and it's likely that genetics will support these splits.

However, one thing I have noticed is that a lack of vocal difference is often ignored when species limits are revised on the basis of morphological differences, and I wonder whether there is an inconsistency here?

I can think of instances where I have had very strong response to playback from the "wrong" closely-related species (e.g. Mentawai Scops-owl reacting to Sunda Scops-owl; Rufous-vented Prinia reacting to Swamp Prinia).

Is it correct to treat vocal differences / lack of differences assymmetrically in this way, or does this reflect a bias in favour of splitting?

It's bias I would say. Now, there might be very good reasons why you acknowledge two forms as separate species even if they respond to each other's vocalizations...there might be other factors preventing widespread hybridization. I don't know enough about either of your examples to say if those might play a role or not.

DMW said:

Vocal differences are increasingly relied upon in avian taxonomy, to the point that Birdlife recently split at least two species (an African cuckoo and barbet) pretty much entirely on the basis of vocalisations. As far as I can tell, the morphological / plumage differences involved are trivial, and no genetic analysis was undertaken. I don't take issue with this - Birdlife has a "quick-and-dirty" approach to species-level taxonomy, and it's likely that genetics will support these splits.

However, one thing I have noticed is that a lack of vocal difference is often ignored when species limits are revised on the basis of morphological differences, and I wonder whether there is an inconsistency here?

I can think of instances where I have had very strong response to playback from the "wrong" closely-related species (e.g. Mentawai Scops-owl reacting to Sunda Scops-owl; Rufous-vented Prinia reacting to Swamp Prinia).

Is it correct to treat vocal differences / lack of differences assymmetrically in this way, or does this reflect a bias in favour of splitting?

Click to expand...

I would guess that the prevalence of mimicry would have a part to play in some cases...

However, examination of sonograms by experts in that field has revealed quite often that vocal differences detectable by birds are often undetectable by humans. We are poor at discrimination of sounds at high repetition rates, and if that's where the differences lie, it's a key to beginning to evaluate what a bird 'understands' in the vocal message.

If the sonograms of different individual birds from different populations are , within limits, identical, then there may be a bias towards splitting. However, given that various forms of integrative taxonomy currently used and in development rely on a majority of features, then the splitting on a single character surely is less likely?

To my ear, Dunnock song early in the breeding season changes when Whitethroats arrive and sing in the vicinity; it becomes harsher - I would regard this as a precautionary, 'let's not take any chances' approach by Dunnocks. Perhaps your experience with those more exotic species contained that element?
MJB

MJB said:

I would guess that the prevalence of mimicry would have a part to play in some cases...

However, examination of sonograms by experts in that field has revealed quite often that vocal differences detectable by birds are often undetectable by humans. We are poor at discrimination of sounds at high repetition rates, and if that's where the differences lie, it's a key to beginning to evaluate what a bird 'understands' in the vocal message.

If the sonograms of different individual birds from different populations are , within limits, identical, then there may be a bias towards splitting. However, given that various forms of integrative taxonomy currently used and in development rely on a majority of features, then the splitting on a single character surely is less likely?

To my ear, Dunnock song early in the breeding season changes when Whitethroats arrive and sing in the vicinity; it becomes harsher - I would regard this as a precautionary, 'let's not take any chances' approach by Dunnocks. Perhaps your experience with those more exotic species contained that element?
MJB

Click to expand...

All fair points, but I think the use of playback perhaps negates the question of whether we aren't able to discern differences, since it is the birds themselves deciding that a song is sufficiently similar to merit a strong and immediate response. I've read papers where non-reaction to playback of sister-taxa is used as positive evidence to split, but I don't recall seeing the converse.

Regarding use of integrative taxonomy, I agree, but Birdlife / HBW seems to eschew this approach. In the examples I referenced, I believe the observable differences between the taxa were rather slight, and that no genetics work was undertaken. My take is that the split is therefore based entirely on vocals. Notably, IOC has these down as proposed splits, presumably in response to the Birdlife split.

Mysticete said:

It's bias I would say. Now, there might be very good reasons why you acknowledge two forms as separate species even if they respond to each other's vocalizations...there might be other factors preventing widespread hybridization. I don't know enough about either of your examples to say if those might play a role or not.

Click to expand...

With the specific examples I quoted (African cuckoo and barbet), the splits probably are justified. Cuckoos in particular seem to have very stable songs, and a significantly different song is likely to be highly significant in evolutionary terms. To this extent, they were probably bad examples for me to use!

I have not re-read it just now, but I think I remember that this thread has some relevant info: https://www.birdforum.net/showthread.php?t=350519

Niels

DMW said:

With the specific examples I quoted (African cuckoo and barbet), the splits probably are justified. Cuckoos in particular seem to have very stable songs, and a significantly different song is likely to be highly significant in evolutionary terms. To this extent, they were probably bad examples for me to use!

Click to expand...

This is the first thing that occurs to me as well, that in your example these are both not only non-oscines but non-passerines. Thus for these species song is genetically coded, non learned, so a difference of song should, theoretically, precipitate a genetic difference.

Assessing playback trials and song differences is difficult and I don't think any one set of rules apply. In some groups of birds, I feel like congeneric or even confamiliar (is that a valid term?) responses are common. For instance Antbirds and Tapaculos regularly respond to other species and even other genera at times. Owls are notorious for territorial responses to other owls.

An interesting example is Western Flycatcher vs Pacific-slope and Cordilleran if considered separately. It appears to now be increasingly apparent that voice intergrades over a large area and that they are perhaps best re-lumped.

I also find the vocally and morphologically messy species groups fascinating, and in many cases there is still a lot of work to be done to fully understand them and try to shoehorn them into human-defined species. I'm thinking here of things like Yellow-Olive Flycatcher, House Wren, Sedge/Grass Wren, Spangled/Hair-crested/Wallacean Drongos, etc.

A very interesting thread. I’m a non-biologist but I’ve had many of the same thoughts as others here.

pbjosh said:

This is the first thing that occurs to me as well, that in your example these are both not only non-oscines but non-passerines. Thus for these species song is genetically coded, non learned, so a difference of song should, theoretically, precipitate a genetic difference.

Assessing playback trials and song differences is difficult and I don't think any one set of rules apply. In some groups of birds, I feel like congeneric or even confamiliar (is that a valid term?) responses are common. For instance Antbirds and Tapaculos regularly respond to other species and even other genera at times. Owls are notorious for territorial responses to other owls.

An interesting example is Western Flycatcher vs Pacific-slope and Cordilleran if considered separately. It appears to now be increasingly apparent that voice intergrades over a large area and that they are perhaps best re-lumped.

I also find the vocally and morphologically messy species groups fascinating, and in many cases there is still a lot of work to be done to fully understand them and try to shoehorn them into human-defined species. I'm thinking here of things like Yellow-Olive Flycatcher, House Wren, Sedge/Grass Wren, Spangled/Hair-crested/Wallacean Drongos, etc.

Click to expand...

Excellent points, Josh. Although it's clearly highly relevant, my question isn't specifically about the utility of playback, but about (what I perceive as) the rather lopsided use of vocal differences in avian taxonomy.

I should say that I don't have any axe to grind here - and I actually think there may be a case to be made for being lopsided (i.e. using vocal differences as support for splits, but not using a lack of difference to argue against it).

It just seems an interesting philosophical issue that I haven't really seen discussed as such.

Absolutely the situation is asymmetrical. I think that is mostly reasonable though.

I am assuming that most cases where vocal differences are used in taxonomic decisions involve two related but isolated populations that do not actually interact with each other in life at present. Think related populations that live on different islands, or in different isolated montane regions. In these cases we are trying to measure what would happen if they came back together - would they happily interbreed and fuse together (= same species), or would they remain distinct (= distinct species)? We should have the humility to recognize this question is difficult-to-impossible (what would happen if...), but we can use careful analysis to provide our best answer.

So we analyze acoustic features of songs, and come up with a statistical metric of how different the populations sing. Or, better yet, we do playback experiments and see if the birds themselves respond to their relative's song.

I've done lots of playbacks in the field, and have many cases where territorial birds consistently totally ignore playback of their relative's song. And every time (at least so far), when I go do the reciprocal experiments, the other population ignore's the relative's song as well. To me this situation represents strong evidence for the "two species" treatment - territorial individuals of the two populations do not recognize each other as territorial competitors, and hence are unlikely to recognize each other as prospective mates. Things get trickier when the variation is complex (as is often the case, e.g. the yellow-olive flycatcher complex mentioned above). And note that the South American Classification Committee is not enthusiastic on defining species based on the observation that they ignore each other in playback experiments.

But imagine they happily respond to each others' songs. Here the interpretation is a bit murkier. Mainly, it is impossible to rule out the idea that they respond to each others' territorial signals but would still fail to interbreed. The playback experiments measure a territorial response, not which mates the birds would prefer. In general birds will be less picky on territorial response and more picky when choosing mates. So if they ignore a song in a territorial context, they likely also ignore in a mate choice context. But if they respond in a territorial context, they may still ignore in a mate choice context, making it difficult to argue for same species status only on the basis of playback experiments. Key here would be to do experiments on female birds in mating condition to see how they perceive the different songs. Unfortunately such experiments are far harder to do than playback experiments, which themselves are harder to do than statistical analyses of songs.

Still, you may be interested that the AOS accepted two lumps of Central American taxa (Cherrie's/Passerini's Tanager & Olive-crowned/Chiriqui Yellowthroat) based in part on playback data. In both cases the birds are similar in plumage, genetics, and also happily respond to each other's songs.

All that said, there is absolutely a general bias in favor of splitting (and against lumping).

DMW said:

I should say that I don't have any axe to grind here - and I actually think there may be a case to be made for being lopsided (i.e. using vocal differences as support for splits, but not using a lack of difference to argue against it).

It just seems an interesting philosophical issue that I haven't really seen discussed as such.

Click to expand...

I agree it's an interesting discussion, well worth pondering. And thanks to Ben for chiming in! Reading this and thinking about it more I quite agree that it makes sense that it should be asymmetrical. If a rigorous trial shows a lack of response in sympatric or parapatric taxa, you've got pretty good data. Proving a response is interesting as well, but never as strong as proving a lack of response... I'm guessing you could pick some Screech or Scops Owl species and get a very high response rate for wildly unrelated taxa at the right time of year if you are in the territories of breeding birds.

An interesting thread indeed! Good to raise this question.

We should indeed not mix up analysis of vocal differences with playback experiments. These are two different things, also in terms of results (as has been shown by Ben Freeman, there is a non-linear relationship between both, and especially in the case of 'small vocal differences as defined by measuring some human-defined basic sound-parameters' there is no clear relationship with play-back response).

I think most people agree that (whenever possible) an integrative approach is the most robust process to decide on the taxonomic status of non-sympatric taxa/populations. And typically that includes analysis of phenotypic characters, voice and genetics.
The larger the difference in these three dimensions, the more reason to provide species status.

Where there is far less consensus is what to do when not all of these three dimensions point in the same direction…

A conservative approach could be to lump anything that does not show differences in all three dimensions. But then this would lead to lumping quite a few all-time good species (which lack either clear morphological, vocal or genetic differentiation).
I don't think anyone would be in favour of this, so then -to be consequent- we are left with the possibility to give species status to taxa which differ significantly in only two of the three dimensions.

This then leads to acceptance of species which are clearly vocally different but hardly morphologically (think Tapaculos), but equally species which show clearly morphological differences but hardly vocally.

So I don't think this can be called a bias towards splitting. In my opinion it is rather being consequent (worldwide) as to which level of distinctiveness is required to assign species status.

This being said, as a birdsound-man myself, I am always a bit suspicious when there are no vocal differences among newly-split species (e.g. the recently split Rote Warbler was said not to differ vocally, a possible first among Phylloscopus warblers). In such cases:
- a possibility is that one didn't select the right parameters to define vocal distinctiveness.
- another possibility is that (unlike morphology) vocal characters did not evolve significantly as there was apparently no need or benefit for it
- or else it is simply not a 'good species' :eek!:

Its not only about vocalizations. It is general cherry-picking arguments favoring taxonomic change and ignoring arguments against status quo.

One could point further flaws. Most playback experiments do not measure female mating preference at all, only territoriality. That one equals another is an assumption usually not tested and generally not transferable between different species. Another flaw is that hybridization does occur between birds singing dissimilar songs (for example, gene flow between distant populations of one species with different local songs).

The reason is primary a drive to increase 'weight' of publications. This is biologists exploiting a loophole in science: re-assigning species or higher taxa gets published as a scientific publication, although it is not science. Discovering underlying biological reality is science, re-assigning species is a type of bureaucracy.

The result is a drive to make publications with added maximum overhead of proposing taxonomic changes to a minimum of new actual biological knowledge, or no new knowledge at all. The ornithology goes to maximum taxonomic changes independently of biological facts.

fun stuff to think about!

A couple more things occurred to me:

- playback experiments & allopatry vs. sympatry. Playback experiments between populations that are isolated and do not encounter each other in real life serve as "simulated meet-ups" that give us some idea about whether birds might recognize each other in the event that they expanded their ranges. I think a lack of response in these cases suggests the two-species treatment, while a response is more difficult to interpret for taxonomic decisions, for the reasons discussed above.

Playback experiments between birds that actually encounter each other in real life tell us something different. For example, sometimes very different species compete with one another and because they compete, they respond aggressively to each other's song ( e.g. Least Flycatcher and American Redstart). So positive responses may be about competition rather than anything mating related.

When populations actually interact in real life, we can use genetics to see if they interbreed. When populations are isolated not so much, so we do things like evaluate song or plumage differences to do our best to assess reproductive isolation.

- playback experiments vs. acoustic differences. As Peter mentions, these two approaches are aiming at the same target but are not the same thing. I like playback experiments because they "ask the birds themselves" if they care about a difference in song.

I found the correlation between acoustic differentiation and response to playback to be weak at low to moderate levels of acoustic differences. So if you told me that two populations sang a bit different, I couldn't tell you anything about whether they would respond to one another (could be yes, could be no). When songs are really quite different, the two methods agree -- really big statistical differences in song pretty much always mean the birds also fail to respond to playback. Meaning that there is little need to do playback experiments when songs are really quite different (measuring what constitutes "really quite different" is a bit involved but not too tricky).

- integrative approach to taxonomy. This is a worthy aim. But these analyses are hard to do, and there aren't many incentives for academics to work on these types of projects. So not many of them get done.

To echo Peter and Josh, we do think different traits are more important for some birds than others. For example, I'd be happy to split two isolated populations of tapaculos solely on the basis of good documentation that they sing really differently even if they look and measure the same and without knowing any genetics. And some taxa seem to hardly ever sing, so song differences are close to meaningless in these taxa (or we are missing the boat and song is important, but hardly ever given in contexts where humans hear and record the songs).

- song is a good marker of good biological species in sympatric taxa. Historically people categorized birds on the basis of specimens in museums, and here differences in plumage and morphology are easy to see. Then field people started to notice song differences between similar-looking species, and there is a long track record of successful discoveries of new taxa based on songs, e.g. Ted Parker mentioning "hey, the Fasciated Antshrikes in SE Peru sing two different songs" - one would prove to be Bamboo Antshrike. There are many such examples, and I think subsequent research has usually (or always??) shown that yes, the two populations in question are quite different genetically. Song gurus have been right a lot.

if song (at least in many groups of birds) is so important in helping us differentiate sympatric taxa that are good biological species, then that is the basis for thinking it should be useful for making taxonomic decisions for isolated allopatric taxa as well.

- asymmetries in splitting vs. lumping. Here I was meaning the social/personal incentives for splitting taxa It's more fun and interesting to say oh look here are a bunch of taxa that should be classified as species, and people get more excited about it. Much less so for lumping. The bar for evidence is also waaaay higher for a lump than a split, e.g. the redpoll discussions. Also there is no shortage of cases that deserve to be split.

all good stuff!

MJB said:

To my ear, Dunnock song early in the breeding season changes when Whitethroats arrive and sing in the vicinity; it becomes harsher - I would regard this as a precautionary, 'let's not take any chances' approach by Dunnocks.

Click to expand...

I've noticed the same with Robins when Blackcaps arrive, their song starts to imitate Blackcap a bit. And same with Chiffchaffs, when Willow Warblers arrive a month after they do, they often imitate WW (not very well!) for a day or two presumably to maintain their established territories against the new incomers :t:

Peter Boesman said:

An interesting thread indeed! Good to raise this question.

I think most people agree that (whenever possible) an integrative approach is the most robust process to decide on the taxonomic status of non-sympatric taxa/populations. And typically that includes analysis of phenotypic characters, voice and genetics.
The larger the difference in these three dimensions, the more reason to provide species status.

Click to expand...

FWIW as a complete layman in such matters, I totally agree with this.

If vocalisations are sufficiently distinct, its icing on the cake but surely it cannot be assumed to be diagnostic in itself?

andyadcock said:

If vocalisations are sufficiently distinct, its icing on the cake but surely it cannot be assumed to be diagnostic in itself?

Click to expand...

I don't pretend to be an authority on this, but in my experience reading papers, seeing/hearing birds in the wild, and watching taxonomic authorities discuss cases, it really depends on which birds you're talking about. As Ben mentioned above, voice in Tapaculos is generally regarded as gold standard. Physical differences are minimal and many taxa are sympatric with several and allopatric with several others at the same time. So, for instance, if you have sympatric taxa and see their genetic distances, then other allopatric taxa with similar genetic differences AND similarly different voices, you can make an inference that voice correlates to (presumed) reproductive isolation. As well, with non-oscine passerines and with non-passerines, voice is more telling than in oscines.

A couple of cases mentioned above - Golden-bellied Warbler and Tricolored Brushfinch, are illustrative of the harder cases with oscines. These birds learn their song, they don't have it genetically coded. I think in both cases most everyone expects / believes they are different species, but as they are oscines taxonomists may be more reluctant to make a decision based solely on voice. Though I don't recall details I believe there is genetic evidence available that for the Atlapetes, the two are not even each others closest relatives.

andyadcock said:

FWIW as a complete layman in such matters, I totally agree with this.

If vocalisations are sufficiently distinct, its icing on the cake but surely it cannot be assumed to be diagnostic in itself?

Click to expand...

For species that are finding their mates in the night (or otherwise are truly dependent on voice for mate choice) and for which voice is genetically determined, I would say yes. Tropical Owls (those that are not active during the day) would be one example.

Niels

njlarsen said:

For species that are finding their mates in the night (or otherwise are truly dependent on voice for mate choice) and for which voice is genetically determined, I would say yes. Tropical Owls (those that are not active during the day) would be one example.

Niels

Click to expand...

Indeed! Here's an illustrative case of a recent lump of a Screech-Owl (Colombian into Rufescent) where voice and genetics correlated well despite morphological differences:

http://www.museum.lsu.edu/~Remsen/SACCprop770.htm

And here's the other half of the coin, this revision of the Vermiculated Screech-Owl complex split some forms out but not all, based again on corresponding vocal and genetic data:

http://www.museum.lsu.edu/~Remsen/SACCprop771.htm

The mode of mate recognition in birds is a homoplasic trait, not an evolutionary conserved trait. It is well known that females of different species of birds use very different characteristics (color, song, different parameters of song) for evaluating a partner.

Therefore the assumption that different songs probably mean reproductive isolation is false. This is even if one gives several positive examples, this trait is not conserved evolutionary.