Wednesday, 29 September 2021

A Typical Caspian Gull?

As any gull enthusiast based in the South-west will tell you, Caspian Gull is still a rare bird down here, or at best very scarce. For example, the Dorset totals for the last few years look like this: 2015 (2), 2016 (1), 2017 (5), 2018 (8) and 2019 (5). The count of 8 in 2018 is the highest one-year total so far. In a brief Twitter exchange with Mark Golley I learned that 2018 was the best-ever year in Norfolk too, with a staggering total of almost 140! Even so, and despite the big numbers, until a couple of years ago Caspian Gull was still a description species in Norfolk. As far as Dorset is concerned (plus Devon, Cornwall and Somerset of course), Casp is likely to remain a description species for a long time yet.

However, Caspian Gull is spreading inexorably westwards in Europe, and occurences on this side of the North Sea are bound to increase. Great. But unfortunately its tendency to interbreed on occasion with Herring Gull means that we will doubtless see a corresponding increase in hybrids and backcrosses, and the identification challenges they pose. What does this mean in practice? Basically, encountering a candidate Caspian Gull in the field presents us with just two choices:

1. Look away immediately. If that is our favoured approach, we need read no further!

2. Get stuck in. Try to nail it one way or the other.

Caspian Gull ID is a topic with the capacity to bore like few others, and this post should already be ringing all kinds of alarm bells. If gulls are not your thing, er...clang, clang, clang!

I have written this post for a very specific reason. At West Bexington, on Saturday 25th September, I found (very badly) a Caspian Gull. I would be the first to say that it was hardly a textbook bird, a 'classic' example of the taxon, but in this post my intention is to prove that it was a 'typical' example. My intention is to demonstrate that a Casp candidate can be analysed in sufficient detail to establish its identity to a degree which will satisfy not only you the observer, but also any records committee tasked with assessing your bird for the purposes of posterity.

A Caspian Gull is identified by a combination of features, which all examples of the taxon will show to some degree or other. But that phrase 'to some degree or other' implies a level of variability, and therein lies the challenge. Hybrids and backcrosses (and Herring Gulls...and Yellow-legged Gulls!) also will show these features to some degree or other. In January 2020 I wrote a post entitled How to Measure Your Caspian Gulls. It referenced a 2011 Caspian Gull identification paper written by Chris Gibbins et al, which aimed to objectively quantify some critical features, in order that you could numerically 'score' an individual bird. The reason I wrote that post was to make a case for the identification of a striking (but not classic!) Casp I had discovered on the Axe Estuary. A Devon or Dorset Caspian Gull is a major prize, and no matter how much they might wriggle I am not going to let one go without a fight! So...

In the Gibbins et al ID paper, 95% of their sample of 63 1st-winter Caspian Gulls scored between 12 and 24, with a mean of 18. So I reckon it is fair to say that a typical Casp will score around 18. The authors concluded that the upper limit for 'safe' identification as Caspian Gull should be 21; anything higher might not be pure, even though genuine Casps did score up to 25. What I intend to do now is analyse Saturday's bird, feature by feature, scoring each trait as we go. So this could easily wind up the most tedious NQS post ever... 


1. Extent of scapular moult

0     no first-generation feathers remaining
1     a small number (< 1/3) of first-generation feathers remaining
2     a significant number (>1/3) of first-generation feathers remaining

Score = 1

So this bird would score 1 for 'extent of scapular moult'.

At this point I should mention that the Gibbins et al trait scoring for 1st-winters was aimed at birds in the period October to March, so at least one week older than this individual. I reckon it is highly likely that those few first-gen lower scaps will be moulted out very soon, so a score of 1 is a bit harsh. A nice, round zero would be the generous option, but we'll stick with 1 for the moment.


2. Greater-covert pattern

0     simple pattern with brown centres and sharp white edges, with no white vermiculation or notching
1     white edges with delicate notches or vermiculation; or dark brown centre with white tip to 1/3 of length (i.e. white restricted to tip or distal third)
2     clear white notches/barringcreating a delicate 'piano key' pattern along the whole edge/feather; but much of feather dark
3     lots of white (more than 1/2 of coverts looking white) distributed along the whole feather, or a bold notching ('piano key' pattern)

Score = 2

A score of 2 for greater-covert pattern is probably fair. If the inner coverts were still first-generation they would definitely be quite chequered, so despite the outer coverts being largely plain I couldn't give a score of 1. Even 1.5 would be a bit too generous I reckon. 


3. Ventral bulge

0     present
1     absent

Score = 1


4. Primary projection

As measured on a photo. This was about the best photo I had for assessing primary projection.  More or less a side-on view, and just about scrapes a zero score.

Score = 0


5. Moult: greater coverts

0     all or almost all new (>75%)
1     51-75% new
2     34-50% new
3     10-33% new
4     one or two feathers moulted
5     no moult

Score = 3

6. Moult: median coverts

Same scoring criteria as greater coverts above.

Score =3

7. Moult: tertials

0     3 or more new
1     2 new
2     1 new
3     all old

Score = 3

8. Darkness of head and body

0     totally white
1     reduced grey wash or streaking (confined to flanks and/or single streaks around nape)
2     light streaking/wash to head (incl. some dark around eye); isolated streaks/blotches on body. Overall, body looks more white than brown
3     well streaked: dark mask around eye and/or streaking covering the whole head/face; body with extensive but moderately dense streaks/mottles
4     strong and dense streaking/mottling on body and head making it appear almost wholly dark

This is another trait where my bird suffers due to youth. A few weeks older and it is likely to look a lot whiter. Even so it's not bad, and I am happy to give it a score of 2.

9. First-generation tertial pattern

0     Diffuse white tip (like Common Gull Larus canus)
1     fine pale fringe around distal portion (like classic michahellis), possibly also with some vermiculations
2     edges moderately notched
3     edges strongly notched and/or some dark barring or pale patches across the feather on some or all tertials

[See photo in 7]

The white part of the feather is restricted to the tip and does not extend very far down the edges. Also there is no notching. So although it doesn't quite match the description in score = 0, I still feel comfortable giving it nil points.

10. Second-generation scapular pattern

0     uniformly silvery-grey, darker patterning absent or very faint
1     silvery-grey background, pattern stronger than on 0, but lacks strong barring or central dark diamonds (only dark shafts and subtle anchors), with only a minority of such feathers (one or two) admixed
2     strong, contrasting shaft-streaks, anchors and/or dark central diamonds, but these more patterned feathers are less than half of all; ground colour creamy or silvery grey, possibly with some grey feathers mixed in
3     strong pattern described in 2 on most (more than half) of feathers, but possibly also one or two plain grey feathers or feathers with grey ground tone.
4     all feathers contrastingly patterned (with dark cross bars or diamonds), lacking plain grey feathers; feather centres buffy brown

Score = 3

Final score

Totalling up the scores for each trait gives the following:

1+2+1+0+3+3+3+2+0+3 = 18

As mentioned above, the date of 25th September means this bird hasn't really completed its post-juvenile moult. Which means it would likely score zero for 'extent of scapular moult' rather than 1, were we to revisit in December, say. Even so, a score of 18 puts it comfortably below the 'safe' identification threshold of 21. In fact, as 18 was the mean score of all genuine Caspian Gulls in the ID paper's sample, it would not be stretching things to call this bird an average, or typical Caspian Gull.

Additional evidence

Before we leave this bird, take a look at this pic...

Caspian Gull at rear, Herring Gull foreground - both in 1st-winter plumage

Notwithstanding the bird's slightly deformed bill (looks like a swollen lower jaw in fact) this gull has a very Caspish structure. High-chested (as if holding its breath), with drooping, attenuated rear end; quite long legs. All quite normal - and good pointers - for Caspian Gull.

Finally, that head and bill...

I took one of the photos, printed it out and literally performed surgery - with scalpel and sellotape - to produce this...

As the inset shows, not a particularly long bill, but slim, and with minimal gonydeal angle.

The 'bill shape trait score' was one of the characteristics which Gibbins et al used to help identify adult birds (not 1st-winters) and again it is a case of the lower the score, the better. I think I could have been a bit more generous with the scalpel, reduced the gap between the mandibles a little further and given the bird an even slimmer bill. This would have given a ratio greater than 2.47. In the ID paper, 87% of genuine adult Caspian Gulls (sample size = 100) scored 1, and the remaining 13% scored 2. Of the 12 confirmed hybrids examined, 43% scored 1, 25% scored 2 and another 25% scored 3.

Personally, I see nothing about this bird's head and bill to make me doubt its identity as a Caspian Gull, and the bill measurements support this view. 

Late Edit...

How could I forget the underwing?!

The superbly white underwing is a massively pro-Casp feature. Not all 1st-winters display such a clean underwing as this. The tail pattern is also spot on. Neither characteristic was used in the Gibbins et al trait scoring system, so all I shall say is that in the case of this individual, both unequivocally support the identification as Caspian Gull. As does the 'Venetian-blind' effect on the inner primaries, with sharp contrast between the dark outer webs and pale inner webs.


The conclusion actually is very simple. An objective analysis of this bird's characters, by means of the trait-scoring system devised by Gibbins et al, gives an unambiguous result: it is a typical Caspian Gull.

Question: do you think I applied the trait scores in a fair manner? After all, I did have a vested interest in a pro-Casp result. I would argue that even the most stringent application of this system would still leave you with a final score of 21 or less, and the nagging feeling that you'd been unduly harsh.

Lastly, a small soap-box...

All this talk of 'German muck' is doing the lovely Caspian Gull a massive disservice I think. The expression is unhelpfully pejorative, insinuating that any Casp which fails to display textbook or 'classic' (rather than 'typical') features, is genetically tainted. And if it is known to originate from a mixed colony...well...kiss of death! This is grossly unfair. In the preparation of the Gibbins et al paper, pure Caspian Gulls from the core range, with known provenance, scored as high as 25 on the trait score test. In other words, birds which are much muckier-looking than the one above can still be pure Caspian Gulls. And all of them deserve a warm hug.


  1. A classic post about Caspian Gulls if ever there was one? Gav. Though I should add that my view of the matter is similar to the one I have of the videos of Sabine Hossenfelder 🙄. One must try.

    1. Just looked her up Ric. Quantum physics even harder than gulls! 😄