Species recognition between geographically isolated populations (1 Viewer)

Benjamin G. Freeman and Graham A. Montgomery (2017) Using song playback experiments to measure species recognition between geographically isolated populations: A comparison with acoustic trait analyses. The Auk: October 2017, Vol. 134, No. 4, pp. 857-870.

Abstract:

Geographically isolated populations of birds often differ in song. Because birds often choose mates on the basis of their song, song differentiation between isolated populations constitutes a behavioral barrier to reproduction. If this barrier is judged to be sufficiently strong, then isolated populations with divergent songs may merit classification as distinct species under the biological species concept. We used a dataset of 72 pairs of related but allopatric Neotropical passerines (“taxon pairs”) to compare 2 methods for measuring song divergence between isolated populations: statistical analysis of 7 acoustic traits measured from spectrograms, and field playback experiments that “ask the birds themselves” if they perceive foreign song as conspecific or not. We report 4 main findings: (1) Behavioral discrimination (defined as failure to approach the speaker in response to allopatric song) is nonlinearly related to divergence in acoustic traits; discrimination is variable at low to moderate levels of acoustic divergence, but nearly uniformly high at high levels. (2) The same nonlinear relationship held for both song learners (oscines) and nonlearners (suboscines). (3) Song discrimination is not greater in taxon pairs ranked as species compared to taxon pairs ranked as subspecies. (4) Behavioral responses to allopatric song are symmetric within a taxon pair. We conclude (1) that playback experiments provide a stronger measure of species recognition relevant to premating reproductive isolation than do acoustic trait analyses, at least when divergence in acoustic traits is low to moderate; and (2) that playback experiments are useful for defining species limits and can help address the latitudinal gradient in taxonomy, which arises because species are defined more broadly in the tropics than in the temperate zone. To this end, we suggest that 21 Neotropical taxon pairs that are currently ranked as subspecies, but that show strong behavioral discrimination in response to allopatric song, merit classification as distinct biological species.

[pdf]

suggested splits

It is interesting to note that the suggested splits within this paper supports a number of the BLI spits, eg within the Plain-tailed Wren, Citrine Warbler (in part) and Tricoloured Brushfinch groups. In addition there is further support for other long suggested splits, ie Buffy Tuftedcheek and Striped Woodhaunter

cheers, alan

Good work which confirms what many birders have suspected I think for a long time. Certainly if you're seeking a recording, you'll try and make sure it's of the relevant population.

I remember being told a long time ago in Asia that Himalayan and Vienamese Cutia will not respond to each others call.


A

This large set of play-back experiments indeed brings some interesting new information. After all, the Myiarchus taxonomy still holds to a large extent because of Lanyon's playback experiments in the 70s.

Unfortunately, discussion and conclusions are somewhat biased and over-simplified in my opinion: their methodology is compared with acoustic trait analysis but not with the integrative approach to reach taxonomic conclusions; they compare their findings with only one taxonomic list, comparing their findings with e.g. HBW/Birdlife taxonomy would have led to a much better match (not surprisingly as the latter is probably the taxonomy which has given highest importance to voice as a taxonomic parameter); they state the 10% species increase of HBW/BLI to be controversial but apparently don't think their 21 new species out of 72 taxonomic pairs (almost 30%) is; they suggest multiple new biological species based on 2 different taxa without any effort to define what to do with the other (in some cases as much as 10+) taxa of the species complex, avoiding real-life complexity such as ring-species; etc....

Play-back experiments are indeed very valuable (provided they are executed in very standardized conditions), and should be added to any other available information in order to reach robust taxonomic conclusions (=integrative approach). Relying purely on play-back experiments is however a step back in taxonomic decision making if you ask me.

With less far-reaching conclusions based on this single study this paper would have been better balanced I think, or are newspaper-style headers really needed to get it published?

Peter Boesman said:

they state the 10% species increase of HBW/BLI to be controversial but apparently don't think their 21 new species out of 72 taxonomic pairs (almost 30%) is

Click to expand...

I haven't read the paper yet but if they preselected known vocally distinct pairs then their 30% shouldn't be compared to BLI's splits. I assume they are actually confirming many BLI splits.

Look forward to reading it but I agree even based on the abstract their conclusions seem a bit rosy.

Peter Boesman said:

Play-back experiments are indeed very valuable (provided they are executed in very standardized conditions), and should be added to any other available information in order to reach robust taxonomic conclusions (=integrative approach). Relying purely on play-back experiments is however a step back in taxonomic decision making if you ask me.

Click to expand...

Totally agree Peter but I assume that any decisions which lead to the elevation of a race to a species, would be taken based on a wider evaluation of information and not just on vocalisation, you in fact make the point that it should and any other approach may be unsafe / unscientific?

One question that occurs to me is, how do we understand if the variation in vocalisation isn't just a regional accent?


A

The paper is open-access and I thought it was very stimulating, and the main takeaway for me was not so much the taxonomic recommendations (which was not the main point of the paper anyway) but the fact that the extent to which two taxa's vocalizations differ doesn't necessarily predict the extent to which they respond to each other's vocalizations. It really stresses how much playback experiments can contribute to our understanding of species limits.

And another big takeaway is the contributions that birders can make by engaging in playback experiments. I've done a little bit of that myself with some antbirds and look forward to doing a lot more if I get the chance. This is a field in which birders can really gather a lot of relevant data.

pbjosh said:

I haven't read the paper yet but if they preselected known vocally distinct pairs then their 30% shouldn't be compared to BLI's splits. I assume they are actually confirming many BLI splits.

Click to expand...

In a quick read I didn't see clear criteria on which the pairs were selected, but it was NOT based on known vocal distinctiveness (otherwise they wouldn't be able to draw a non-linear relationship between vocal discrimination and vocal distinctiveness)

cajanuma said:

the main takeaway for me was the fact that the extent to which two taxa's vocalizations differ doesn't necessarily predict the extent to which they respond to each other's vocalizations. It really stresses how much playback experiments can contribute to our understanding of species limits.

Click to expand...

They proved a non-linear relationship, so there is a clear correlation. Only for small to moderate vocal differences, response was more scattered. The authors didn't however investigate further what could be the causes.

I obviously agree that playback experiments might well be a better indicator than vocal difference, BUT:
* as it is not verifiable by others, one has to rely on the fact it was executed properly (not everybody will be as accurate as e.g. Lanyon)
* there is only a handful of cases where this info is available
* even when it is available, it is often a 'single point' test
etc..
In other words, just like DNA data, it may take a few decades before a reasonably extensive picture is available.
In the meantime, measurement of vocal difference is a decent alternative, especially when vocal differences are large (as proven by this paper). (As a matter of fact, this then leads to a high score according to the Tobias system etc.)

andyadcock said:

Totally agree Peter but I assume that any decisions which lead to the elevation of a race to a species, would be taken based on a wider evaluation of information and not just on vocalisation, you in fact make the point that it should and any other approach may be unsafe / unscientific?
A

Click to expand...

I wouldn't say unscientific, but the narrower the sources of information, the riskier it gets to rely just on these to take a decision. I think there are only a handful of cases in the Neotropics where play-back experiments were (almost) the only argument to elevate a taxon to species rank, one or two Myiarchus species come to my mind. In these cases the thoroughness of the play-back experiments as illustrated in Lanyon's papers were second to none.

cajanuma said:

And another big takeaway is the contributions that birders can make by engaging in playback experiments. I've done a little bit of that myself with some antbirds and look forward to doing a lot more if I get the chance. This is a field in which birders can really gather a lot of relevant data.

Click to expand...

As mentioned already, it will be tricky to rely on such data. A voice recording made by a birder on a specific location is 'hard data', and it can be used by any scientist by measuring vocal parameters etc.
A single playback experiment by a birder is far more subjective (according to which protocol was it executed? did the birder have sufficient experience with such set up? what were the conditions? was the proper territorial vocalization used? was it in a breeding period for that species? etc. etc.)

It can of course become useful, but one will have to take into account a considerable margin of error, at least until such practice becomes common and widely known.

I have also not read the paper yet. The first thing that sprung to my mind on reading the abstract is that I have read about birds reacting to song of other sympatric species in order to defend territory against food competition. Therefore, play-back experiments also need to be interpreted with care.

Niels

njlarsen said:

I have also not read the paper yet. The first thing that sprung to my mind on reading the abstract is that I have read about birds reacting to song of other sympatric species in order to defend territory against food competition. Therefore, play-back experiments also need to be interpreted with care.

Niels

Click to expand...

Very true, I've had Pygmy Wren Babbler coming to Limestone Wren Babbler recordings



A

They asked birds themselves. But if the bird gave wrong answers? It may just not recognise the song. Possibly it is slightly curious and changes its position to see if there is something there in the direction of the strange song, and sees nothing, at most it sees a human being there doing something uninteresting. But if there really was a similar bird than itself, moving, being visible and acting, then it will react.

As a practical example, it really seems that Common and Siberian Chiffchaffs do not respond to each others' song in the allopatric areas, far from the other taxon. They don't need to and individuals may never have heard the quite different song of the other taxa. But inside the contact zone, they react. There they hold inter-taxon territories and many are mixed singers. They really are quite similar birds, morphologically and ecologically.

I am not saying that playback experiments are not interesting, but it is still not clear how the result should be interpreted.

jalid said:

I am not saying that playback experiments are not interesting, but it is still not clear how the result should be interpreted.

Click to expand...

That is exactly why it should only be used in conjuction with other criteria when assessing a species for elevation to full species status.



A

cajanuma said:

the main takeaway for me was not so much the taxonomic recommendations (which was not the main point of the paper anyway)

Click to expand...

This may be correct, but if you read the header "Song Experiments Reveal 21 Possible New Tropical Bird Species" you may think otherwise :eek!: (see here)...

A statement like "Song experiments reveal SACC taxonomy to be seriously outdated" would have been equally correct...(it is in fact almost hilarous that exactly JV Remsen is cited in this note to support the findings...sorry, I couldn't resist |:$| )

njlarsen said:

I have read about birds reacting to song of other sympatric species in order to defend territory against food competition.

Click to expand...

Had several, e.g. Firecrest coming to Goldcrest recording.

Playback is generally very dependent from the time of breeding season, territory density and 'personality' of individual males.

I am especially missing the effect of distance per se. Do birds react much worse to a song of 'different' allopatric species from 400 km away than 'same' species from continous population from 400 km away in another direction?

Peter Boesman said:

This may be correct, but if you read the header "Song Experiments Reveal 21 Possible New Tropical Bird Species" you may think otherwise :eek!: (see here)...

Click to expand...

That's not the title of the paper, though

I actually agree with pretty much all your comments above, but I think playing up the taxonomic recommendations as was done on the AOU blog does a disservice to the real value of the paper. To me it's very much a 'heads up' to the ornithological/birding community that highlights a well-known but often overlooked way to add very significant data to our understanding of species limits, and puts it into perspective vs. only analyzing vocal parameters.

I wonder if a site such as xeno-canto could provide a forum/database for birders to begin performing such experiments and recording the results. Of course there are many obstacles, standardizing data would be difficult, there are major differences in the extent that birds respond to playback (Henicorhina wood-wrens seem to come in to just about anything!) but I think with some guidance a lot of interesting data could be uncovered
,

Just read the paper. Broadly agree with your comments Peter but this appears to be a paper gleaned from already collected data. So their conclusions might be a bit simplefied but wrt, for example, the 10% of new species being controversial vs their suggestion of 21 new species, I still don't see the problem. I took their controversial comment as a not too subtle way to highlight the amount of taxonomic disagreement in the neotropics. As well, they pre selected their taxonomic pairs from allopathic taxa with vocal divergence so it is not surprising they found what they suggest to be species level divergence in 21/27 subspecies pairs.

For me the biggest weakness is not sorting their taxonomic suggestions more carefully. Some results are more easily interpreted (Woodhaunter, some of the Marañón splits) whereas too many leave out other important taxa. Shame, for instance, that they didn't test S of Marañón Blackish Tapaculo vocalizations, Amazonian Buff-throated FG, etc. And their suggestion to divide two YO Flycatcher subsp on those results is almost comedic in comparison with the amount of work necessary in that group. Surprised they didn't study other Tolmomyias, there are lower hanging fruit on that tree!!

pbjosh said:

As well, they pre selected their taxonomic pairs from allopathic taxa with vocal divergence so it is not surprising they found what they suggest to be species level divergence in 21/27 subspecies pairs.

Click to expand...

Don't want to split hairs here, but I don't think that's correct. The authors state "These taxon pairs were chosen because they are closely related (often sister taxa) but geographically isolated populations; in some cases, whether the taxa should be classified as separate species or subspecies of the same species is controversial".

Pairs for which vocal divergence was calculated as 'very low' in their spreadsheet are e.g. two taxa of Grey-breasted Wood-wren, Collared/Golden-fronted Whitestart, Clay-colored/Ecuadorian Thrush,....

Anyway, I don't think we should dissect the paper here, in absence of the authors in this discussion.

There is still a lot to be learnt about play-back experiments and what exactly the results tell us. And a lot of caution is needed when drawing conclusions...