Wednesday 8 January 2020

How to Measure Your Caspian Gulls

If you are interested (or think you might be) in the dubious joy to be had from a discussion about Caspian Gull Identification, with particular reference to phenotypic variability and the field characteristics of hybrids, then... Welcome!

The bold print above is taken from the title of a British Birds paper by Chris Gibbins, Grzegorz Neubauer and Brian J. Small, published in December 2011 (BB 104: pp 702-742). Having already published a Caspian Gull ID paper the previous year which dealt with typical individuals, this one focussed on less typical birds and suspected hybrids. The paper provides an objective framework for the treatment of such beasts. Note the word 'objective'...

I have not forgotten that my first response to X78C was not very objective at all. I didn't like its heavily marked underparts, or its chequered greater coverts. I did like its white head, its bill, its long legs, its facial expression and various other features. In the end I decided to count it as a Casp, despite knowing it originated from a German colony where hybrids are present [I nearly typed 'where hybrids are rife' but as I have no idea of the demographic composition of these colonies I shall refrain, and try to]. So, I decided to count the bird as a Casp because I liked it sufficiently well.

Is that good enough?

Well, in the end, when pushed (thank you Steve!) I decided it wasn't. What's interesting is that I know why I liked it. Down the pub I could have argued my case with like-minded gull people. Possibly I would have got a bit exasperated when they didn't agree with my argumentation or see how incredibly solid it was. Maybe I'd have been intimidated by their greater experience and knowledge, but stuck to my guns anyway because I really, really liked it for Casp. I don't know... But what I do know is that my opinion would have been almost wholly subjective.

I find it slightly ironic that just a few days ago I wrote a post called The Expert View, all about measuring things, being objective, and yet here was I being the opposite. So this post is all about me rectifying that...

In their paper, Gibbins et al say this:

'There is frequent debate on internet forums over 'problem birds'. In many cases these debates go unresolved because the views put forward lack an objective or quantitative basis - opinion and counter opinion fuel interesting but ultimately inconclusive debate.' (bold type mine)

The paper goes on to outline a process where certain plumage and structure traits are assessed for each bird (in effect measured), scored individually, and then added together to give an overall total score for that bird. The traits chosen for first-winter birds (October to March inclusive) were as follows, with the available scores in brackets:
  • Extent of scapular moult (0-2)
  • Greater-covert pattern (0-3)
  • Ventral bulge (0-1)
  • Primary projection (0-3)
  • Moult: greater coverts (0-5)
  • Moult: median coverts (0-5)
  • Moult: tertials (0-3)
  • Darkness of head and body (0-4)
  • First-generation tertial pattern (0-3)
  • Second-generation scapular pattern (0-4)
In their study, the authors found total scores for pure Caspian Gull ranged from 12-25, for pure Herring Gull 29-37 (ie, no overlap) but for hybrids 22-32. Oh dear, there's an overlap, so any bird scoring 22-25 might be either a genuine Casp or a hybrid. Oops. How do you handle that? Gibbins et al explain...

'...we suggest that 21 should mark the upper limit of safe first-winter Caspian (as birds with scores of 22-25 could be either pure Caspian Gulls or hybrids).' (bold type and underlining mine)

And I would just like to highlight another point made in this fascinating paper...

'Particularly for first-winter birds, the data show that some traits, traditionally thought to be useful for separating the species, are not wholly reliable on their own because of the degree of overlap. Rather than necessarily seeing this overlap as a problem for field identification, we prefer to use it to suggest that certain criteria used in record assessment should be relaxed. Perhaps most obviously, it is clear that a virtually unmarked white head and body should not be a prerequisite for a record of first-winter Caspian Gull - many Caspian Gulls (16% of our sample) were actually rather well streaked (trait score 3). Similarly, some first-winters had heavily chequered greater coverts, so this feature alone should not automatically rule out a candidate bird.'

One aspect of using the total of all the trait scores means that undue weight is not given to any particular trait, and the above quote indicates why this is a reasonable approach.

So, how did X78C fare? That bird returned a trait score of 19. Which is objective enough for me.

What about yesterday's bird? Well, it came out as 18.

The mean score for Caspian Gull in the study was 18, so both birds were more-or-less average. I am guessing this will be a bit surprising to some readers. It was to me. Also surprising was how close their scores were.

Above: 7th Jan bird   Below: X78C (18th Dec)

Here's a fact. X78C comes from a German colony where there are hybrids, as well as pure Caspian Gulls. Provenance of the 7th Jan bird is unknown; it may be from the hybrid zone, it may not. But because we know about this possibility it is very tempting when we see traits in a bird that are a bit more Herring than Casp to effectively downgrade the bird as having had its genetic purity diluted, rather than just accept that such traits are within the variability of Caspian Gull and treat the bird as a whole. In reality the bird may be genetically spot on, and its less-than-classic appearance nothing whatsoever to do with any Herring Gulls in its dark, embarrassing ancestry.

It is thought-provoking that one or two fellow birders felt that yesterday's Axe Casp was a really nice example, or words to that effect. I am not so sure. In the field I thought it looked much darker on top than I would like, and not that great really. Definitely a Casp, yes, but far from a classic one in my opinion. For example, in the comparison shots above, it's pretty clear that X78C has a more Caspy bill. Using a selection of the best photos I could find, the length/depth bill ratio of X78C was almost 3.0, compared with 2.5 for yesterday's bird. Again, using several photos, it looked like X78C just pipped yesterday's bird for primary projection too. A tiny bit more Caspy. Mantle and scaps much more Caspy.

My gut feeling is that those NQS readers who are a bit into their gulls much prefer yesterday's bird to poor old X78C. If, after reading this, that's still the case, well, fair enough, but I really like that lovely white-headed weirdo. In an objective way of course...

No comments:

Post a Comment