Introduction

The Armenian Gull Larus armenicus is poorly known on its breeding grounds, although these are easily accessible. Its field identification is surrounded by conflicting information (esp. Frede & Langbehn 1997), its phylogenetic affinities are unknown and its taxonomic status as a distinct species is poorly supported by published information. The taxon was first described as subspecies of Larus taymirensis (Buterlin 1934), then classified as a subspecies of Larus argentatus (e.g. Cramp & Simmons 1983) and finally ranked as a full species (Haffer 1982, Snow & Perrins 1998, Sangster et al. 1998).
However, except for one detailed study on the Armenian breeding grounds (Filchagov 1993, see also Buzun 1993a), little new information has come to light that would justify such changes in taxonomic opinion. In particular, a genetic study that might shed new light on the potential reproductive isolation and phylogenetic affinities of this taxon has been lacking.

Material and methods

From 22 May to 06 June 1999 we visited breeding colonies of Armenian Gulls at three lakes in interior Turkey (Beysehir, Tuz Golu, Van Golu). Our goals were to collect information about breeding numbers, diet and conservation status, to quantitatively characterize the phenotype of breeding adults (plumage, size, vocalizations) and appropriately document these characters photographically and sonographically and finally to determine the genetic distinctiveness and the phylogenetic affinities of armenicus by sequencing mitochondrial DNA of a sufficiently large number of individuals. For that purpose we noted plumage and bare parts features in several hundred adults, photographed many individuals, caught 10 adults at Tuz Golu and collected blood samples from these plus 87 chicks. A highly variable portion of the mitochondrial genome (control region domain I, 420 nucleotides) was sequenced from all samples.
Several hours of sound recording were used to detect possible differences in the stereotyped long call between armenicus and michahellis. Data gathered for armenicus were compared to material (phenotypic and genetic) we had previously obtained for all West Palaearctic taxa that it might be related to. For adults caught in the field, sex was determined by PCR (polymerase chain reaction) using sex-specific primers (Kahn et al. 1998). Results presented here are part of a larger study involving the entire Larus argentatus-fuscus-cachinnans complex (Liebers & Helbig in prep.).

Colonies, breeding phenology, behaviour

In the northern part of Tuz Golu we visited two colonies on small adjacent islands about 7 km of the western shore. A total of 350-370 pairs of gulls bred here, 95% of which had chicks ranging in age from one day to about 3 weeks, the remaining pairs were incubating clutches, almost invariably containing 3 eggs. Gulls foraged only in terrestrial habitats around the lake and fed on insects, rodents and other small mammals (e.g. Spermophilus xanthoprymnus).
At Lake Beysehir there was a mixed colony of 40-50 pairs of Yellow-legged L. michahellis and Armenian Gulls plus many intermediate phenotypes breeding on a small rocky islet. Nests contained chicks about 2 weeks old and three fresh replacement clutches. Gulls brought only aquatic prey (fish, dragonfly larvae) to the nests. This colony has been known for 35 years (Vauk 1973), but compared to 1964 was dramatically impoverished; Dalmatian Pelicans (83 pairs), Cormorants, Night Herons and Little Egrets all have vanished due to disturbance by local people. Only three clutches of Mallard, one pair of Ruddy Shelduck and many Jackdaws and Rock Pigeons remained.
At Van Golu we visited two colonies, one on Ahtamar Island plus a small off-lying rock and one on Carpanak Island. At Ahtamar, where previously up to 1.500 breeding pairs of Armenian Gulls had been recorded (Beaman 1975, 1978), we found only 15 pairs with nests on the small rock, plus c. 50 empty nests at the steep western coast of the main island, where all broods seemed to have been lost through predation (apparently by large snakes, of which we saw one). Currently, the largest Armenian Gull colony in Turkey is located on Carpanak Island, where we estimated 2.000 pairs to be present. Hatching success appeared to be very high here; most clutches had just hatched or were in the process of hatching on 2 June. Although only c. 2 km from the shore, the island seems to be visited rarely by local people or tourists and is free of terrestrial predators. Gulls nested everywhere from the pebble beaches over grassy or boulder-strewn slopes, on the ruins of an old Armenian church up to the rocky hill top. Adults were very aggressive and easy to observe from close distance. Other noteworthy breeding birds included Night Heron (c. 15 pairs), Caspian Tern (4 pairs) and Ruddy Shelduck (one pair). Judging from remains at the nests, the diet at both colonies at Van Golu appeared to consist of fish, only a single endemic species (Pearl Mullet Chalcalburnus tarichi) being available in the lake.

Turkish and world population

The current Turkish breeding population of Armenian Gull is about 2.400 pairs. It seems to have been constant at Tuz Golu during the past three decades, but apparently declining at Van Golu (due to predation at Ahtamar Island). Together with 11.000-13.000 pairs at two lakes in Armenia and 4.000-5.000 pairs at one lake in Iran, the world population may be around 20.000 breeding pairs. All large colonies are located on predator-free islands in large highland lakes. With 98% of the total being concentrated on only four lakes, the species must be regarded as potentially threatened. The most likely threat would be introductions of terrestrial predators to the major breeding islands.

Plumage and bare parts of adults

At the Beysehir colony adults of typical michahellis, typical armenicus and intermediate phenotypes bred together, indicating that hybridization is going on. The following description refers only to adults at Tuz and Van Golu, which belonged to the same, fairy uniform phenotype (fig. 5, 6). These birds were characterized by a rather dark grey mantle (8 on Kodak Grey Scale), presence of obvious black bill marking, yellow legs and large amount of black in the outer primaries. However, all characters showed some variation, the extremes of which were indistinguishable from character states observed in michahellis or other taxa.

Upper part grey-tone
The upper part grey-tone was checked when 10 birds were caught at Tuz Golu. Seven birds scored 8 on the Kodak Grey Scale, two birds were darker (Kodak 8.5) and one was paler (Kodak 7.5). This is slightly darker than michahellis (Kodak 6-7), but identical to barabensis (pers. obs. by authors).

Size and shape
Armenian Gull proved to be smaller than michahellis and cachinnans in head-bill length and wing length (tarsus and toe not measured), but there is some overlap (broader in females) and size measures are only useful if sex is known (fig. 21).

Head shape, often stated in the literature to be a useful character, is highly variable (e.g. depending on stress or other external factors) and differs between sexes: females have a rounded head with steep front, together with their shorter bill creating a more friendly facial expression, while males have a much flatter front, more angular nape, longer bill and more aggressive facial expression. It is possible that the misconception about head shape in the literature is due to the fact that mostly females winter in Israel (most identification papers were based on winter observations in Israel).

Bill
Over 99% of all individuals had at least some black markings on the bill (Table 1), clearly rejecting Frede & Langbehn’s (1997) contention that armenicus lacks this character in the breeding season. However, bill markings were quite variable and were poorly developed (hard to see, small and diffuse) in about 5% of adults (n = 463). Hence, 95% of the breeding birds had a “half-moon shaped” black marking on the upper mandible (not a circle), which continued in many birds (more or less clearly) on the lower mandible. Only one single bird at Van Golu and two birds at Tuz Golu (2 out of 182) really lacked any black on the bill. The very tip of the bill, in front of the black marking, was whitish, together with the red gonys and yolk-yellow proximal part of the bill creating a four-coloured impression (Fig. 7). The red gonys spot was well developed on the lower mandible and only marginally continued on the upper mandible. The base of the bill is more intense yellow than it is in cachinnans from the Ukraine (Fig. 9), but similar to michahellis.

Leg colour
Leg colour was always yellow, but with two stages of saturation, one pale yellow (with a grey-green cast like in cachinnans) in 70% (n = 125, Table 1), the other intensely yellow (like in breeding adult michahellis) in 30% of armenicus.

Iris colour
Iris colour varied from completely pale (14%), to predominantly dark (44%) with all kinds of intermediate stages (n = 163 at Van Golu). Birds classified with a predominantly dark iris appeared “complete dark eyed” in the field at 20 m. distance (Fig. 8). At close range in the hand it appeared that a complete pale iris still showed small dark spots, and that a complete dark iris in fact was a dark spotted iris on a pale background. The orbital ring was orange-red in all birds.

Primaries
Three types of wing-tip patterns were found (Fig.18), all had black in 7 primaries and in many birds there was also black on the greater primary coverts:
* Type I with a white mirror only in P10 was most common (79%, n = 232),
* Type II with a second small white mirror on the inner web of P9 was less common (20%).
* Type III, typical of michahellis, was rare (only 2 birds, is 0.9%) in pure armenicus colonies (Fig. 19). The number of mirror spots was unrelated to sex in both armenicus and barabensis (contra Jonsson 1998). Armenian Gulls lacked a clearly delimited pale “tongue” on the inner web of P10 (the inner web was predominantly black and only at the base was a diffuse transition from this black to grey, and on P9 this tongue was much shorter and usually less clearly demarcated than in barabensis (Fig. 16, 17) or cachinnans.

Long call display
Armenian Gulls performed long call displays similar to michahellis and fuscus, never spreading the wings to an “albatross posture” typical of cachinnans. In large white-headed gulls, the long call has a stereotype pattern: it starts with one, only rarely two long elements, then miau-calls, then an intermediate call, followed by 8 to 24 short elements (Fig. 22). Long call vocalizations differed significantly from those of michahellis in 5 parameters measured sonographically (duration of first 8 short elements, frequency of miau-call and first short element, intensity distribution of harmonics and frequency modulation of miau-call), but differences were slight and gradual in nature. Several of these parameters may be related to body size, i.e. differences from michahellis may reflect smaller body size of armenicus.

Mitochondrial DNA
Phylogenetically, Armenian Gull was found to be most closely related to michahellis to the exclusion of taymirensis, barabensis and cachinnans (Fig. 24). Michahellis and armenicus are represented by two distinct clusters in the mitochondrial haplotype phylogeny, but introgression (unidirectional geneflow) was evident with two michahellis haplotypes (shaded in cluster I, Fig. 24) having invaded the armenicus population at Tuz Golu. These and a third michahellis haplotype were found in 62% of the birds in Beysehir, 14% at Tuz and 0% at Van Golu (Fig. 25). This indicates that hybridization does occur, hybrids are fertile and backcross at least with armenicus in central Anatolia. Whether geneflow in the other direction (into the michahellis population of the E Mediterranean) also occurs requires further study. Notwithstanding the ongoing introgression, the phenotypically “pure” armenicus population at Tuz and Van Golu was genetically significantly differentiated from michahellis (p < 0.0001). The genetic divergence of the two haplotype clusters (average 2% at 420 nucleotides of control domain I, Fig. 26) indicates a relatively ancient separation and long independent evolution between michahellis and armenicus, but this has so far not resulted in full reproductive isolation.

Conclusions

All evidence presented here agrees in showing that Armenian Gull does not differ qualitatively from other gulls of the argentatus-fuscus-cachinnans complex (contra Sangster et al. 1998), i.e. there is no single character allowing an unambiguous diagnosis. A combination of several characters (bill marking, wing-tip pattern, mtDNA) achieves almost 100% diagnosability in the core breeding areas of Armenian Gull (Tuz, Van), but not in the zone of secondary contact with michahellis (Beysehir), where intermediate birds occur. With respect of other taxa of the complex, in particular cachinnans, barabensis and taymirensis, diagnosability is 100% (using mtDNA and wing-tip pattern). At present it is unclear whether intrinsic barriers (mating preferences, impaired viability or fertility of hybrids) restrict the geneflow between michahellis and armenicus or whether the observed introgression is the result of recent secondary contact. The two taxa only partially fulfill the requirements of biological species status (diagnosability, reproductive isolation). However, giving the high degree of diagnosability in the core breeding areas, the limited amount of geneflow and the substantial genetic divergence between their respective mitochondrial haplotype lineages, the two taxa should be regarded as semi- or allospecies under the BSC. The situation may be comparable with the early stages of contact between argentatus and fuscus at the beginning of this century (early 1900’s), when hybridization was much more frequent than it is today. Phylogenetically, armenicus is clearly the sister taxon of michahellis (including atlantis), while it is more distantly related to barabensis, taymirensis and heuglini with which relationships have been hypothesized in the past.

Literature

Beaman, M., & S. Madge (1998): The handbook for bird identification for Europe and Western Palaearctic. London.

Buterlin, S. A. (1934): Larus taymirensis armenicus, subsp. Nov. Ibis ser. 13 (4): 171-172.

Buzun, V.A. (1993a): Armenian Gull Larus armenicus Buterlin, 1934: morpho-biometrical and behavioural distinctions with indication of taxonomical status. Russian J. Ornithol. 2: 471-490 (Russian, with English summary).

Cramp, S., & K.E.L. Simmons (1983): Handbook of the birds of Europe, the Middle East and North Africa. Bd. 3. Oxford, London, New York.

Doherty, P. (1992): Armenian Gull and Californian Gull. Birding World 5: 117-118.

Filchagov, A.V. (1993): The Armenian Gull in Armenia. British Birds 86: 550-560.

Frede, M., & H. Langbehn (1997): A contribution to the identification and distribution of the Armenian Gull. Alula 3: 102-108.

Grant, P.J. (1987): Notes on the Armenian Gull. In: International Bird Identification. Proc. 4th Intern Identification Meeting Eilat: 43.

Haffer, J. (1982): Systematik und Taxonomie der Larus argentatus-Artengruppe. In: U.N. Glutz von Blotzheim & K.M. Bauer: Handbuch der Vogel Mitteleuropas, Bd. 8. 502-515. Wiesbaden.

Hirschfeld, E. (1992): More gulls with bill bands. Birding World 5: 116.

Hume, R.A. (1983): Herring Gulls in Israel. British Birds 76: 189-191.

Jonsson, L. (1998): Yellow-legged Gulls and yellow-legged Herring Gulls in the Baltic. Alula 3: 74-100.

Kahn, N.W., J.S. John & T.W. Quinn (1998): Chromosome specific intron size differences in the avian CHD gene provide efficient method for sex identification in birds. Auk 115: 1074-1078.

Madge, S. (1990): Eine Armenienmowe Larus (cachinnans) armenicus ohne schwarze Schnabelzeichnung. Limicola 4: 216-217.

Sangster, G., C.J. Hazevoet, A.B. van den Berg & C.S. Roselaar (1998): Dutch avifaunal list: species concepts, taxonomic instability, and taxonomic changes in 1998. Dutch Birding 20: 22-32.

Satat, N., & B. Laird (1992): The Armenian Gull. Birding World 5: 32-36.

Shirihai, H., & D. Christie (1996): The Macmillan birder’s guide to European and Middle Eastern birds. London.

Snow, D.W., & C.M. Perrins (1998): The birds of the Western Palaearctic. Concise Edition. Bd. 1. Oxford, New York.

Suter, W. (1990): Comments on the breeding range of the Armenian Gull. OSME 25: 12-15.

Yesou, P., & E. Hirschfeld (1997): Which large gulls from the Larus fuscus-cachinnans-argentatus complex of (sub)species occur in Bahrain? Sandgrouse 19:111-121.

.