The Chiffchaff superspecies complex occupies almost the entire Palearctic and includes many taxa of different ranks. It is traditionally considered to be one of the most complex problems in the taxonomy of Palearctic birds. We present new data on the genetics, morphology and bioacoustics of the Chiffchaff taxa found in Turkey, the Caucasus, Transcaucasia and Kopet Dagh: ‘greenish' brevirostris, caucasicus, menzbieri and ‘brownish' lorenzii. In southern Turkey, two Chiffchaffs were identified which carried a novel haplotype, recently discovered in Northern Israel. Both individuals from Turkey looked somewhat brighter and more yellowish than typical brevirostris, but their song clearly corresponded to the brevirostris dialect. The form brevirostris, inhabiting the western and central regions of Northern Turkey, was found to be the most distinct among all ‘greenish’ taxa, due to its small body size, wing formula and distinct vocal dialect. Paradoxically, the mitochondrial DNA of brevirostris is almost identical with caucasicus, whereas, in this respect, menzbieri is different from both. We believe that this paradox can be explained if we suppose that caucasicus originated during the ancient hybridization of brevirostris and menzbieri. The relationships between brevirostris and lorenzii, in the mountainous regions of eastern Turkey, show the mismatch between phenotypic and genotypic traits in some individuals that could also be a result of hybridization.
The study of geographical variation in birds, being closely related to the research on microevolution, taxonomy and zoogeography, is a highly developed area of ornithology. Researchers working in this field are especially interested in polymorphic species with extensive and complexly fragmented distribution ranges. The Chiffchaff Phylloscopus collybita, the range of which extends from North Africa and southwestern Europe to Lake Baikal and Chukotka, provides an excellent example (Figure 1). A number of taxa of various ranks have been identified within the Chiffchaff's range. At present, they are divided into four to five species, according to genetic studies (Helbig et al. 1996, del Hoyo & Collar 2016, Gill et al. 2021).
Differences between Chiffchaff taxa are expressed in their plumage coloration, body size, wing shape and vocalization (both songs and calls). In particular, all the taxa can be divided into ‘greenish’ and ‘brownish’, groups according to their plumage coloration (Ticehurst 1938, Martens 1982). In secondary contact zones, some taxa hybridize and there are individuals with intermediate coloration, size and a mixed song (Salomon 1987, Marova & Leonovich 1993, Hansson et al. 2000, Helbig et al. 2001, Bensch et al. 2002, Marova et al. 2013, 2017, Shipilina et al. 2017).
Figure 1.
Distribution ranges of the Chiffchaff taxa and the sampling locations. The oval shape contains the area considered in this article. The white arrows show hybrid zones.
Numerous studies of the phylogeny and taxonomy of the Chiffchaff have been published (Ticehurst 1938, Watson 1962, Martens 1980, 1982, Martens & Meincke 1989, Loskot 1991, Marova & Leonovich 1993, Helbig et al. 1996, Clement & Helbig 1998, Dean & Svensson 2005, Marova et al. 2013, 2017). Recently, a new Chiffchaff haplotype was discovered in the Hermon mountains in northern Israel. Although no specific morphological attributes of this Israeli Chiffchaff were reported, its genetic characteristics were found to be clearly distinct (Raković et al. 2019). There is no further information on this form.
Nevertheless, many questions concerning the well-known Chiffchaff taxa which have long been known still remain unanswered. In particular, the situations in Turkey, the Caucasus, Elburs, Khorasan and the west of Turkmenistan, are only vaguely understood. Three morphologically similar ‘greenish' taxa, brevirostris, caucasicus, menzbieri, as well as the ‘brownish’ lorenzii, are found in this area. Since Watson's (1962) description of the Chiffchaff of Asia Minor, little information has appeared on the Chiffchaffs found in the region (Kirwan et al. 2008). It was only recently that Raković et al. (2019) published their comprehensive genetic study of these taxa (but not including brevirostris).
Meanwhile, the form brevirostris, found in Turkey, remains an unresolved problem. In the first description of brevirostris, Strickland (1836) noted its significant morphological similarity with the European collybita and abietinus. Stresemann (1928) considered a breeding bird from central Turkey (Eregli) to be inseparable from collybita, and Roselaar (1995) claimed that “recognition of brevirostris is provisionally maintained, but examination of a larger number of breeding specimens may show that brevirostris is not separable from collybita”. However, a year later, an article by Helbig et al. (1996) was published, which showed brevirostris to be genetically distinct from collybita and abietinus, but almost indistinguishable from caucasicus. Later, in accordance with this finding, Kirwan et al. (2008) proposed brevirostris to be considered as a senior synonym of caucasicus. In eastern Turkey, ‘greenish' brevirostris co-occurs with ‘brownish' lorenzii (Kumerloeve 1967a,b, Roselaar 1995). In addition, lorenzii is widespread in the Caucasus and Transcaucasian areas. Almost everywhere in the latter two regions the distribution range of lorenzii overlaps with that of ‘greenish’ caucasicus. Normally these taxa inhabit different altitudes (Chunihin 1962) but sometimes they breed in close proximity (Jordanya 1962, Molamusov 1967, Martens 1982, Marova 2002).
The form menzbieri represents another poorly known taxon of ‘greenish’ Chiffchaffs. Since it was first described in 1937, very little knowledge has been gained about its morphology, ecology and song (Marova & Leonovich 1997, Loskot 2002, Dean 2017). The distribution range of menzbieri is centred in the mountains of Khorosan and extends eastward up to the western Kopet Dagh. The northern and western limits of the range are poorly known but the genetic markers of menzbieri were recently found throughout southern areas in the Transcaucasia, and introgression was supposed to exist between caucasicus and menzbieri (Raković et al. 2019).
The aim of the present paper is to analyse our new data on the genetics, bioacoustics and morphology of all Chiffchaff taxa breeding in the vast region between the Bosporus and Kopet Dagh, including Turkey (where in collecting our data a significant part of the territory of the latter was surveyed), Ciscaucasia, Caucasus and Transcaucasia (Figure 1 and 2). In particular, we place a special emphasis on brevirostris and menzbieri. Thus, the novelty of our study lies in the fact that we analyse quantitative bioacoustics parameters along with morphological and genetic characters in those Chiffchaff taxa which remain currently the least studied in terms of all the aspects mentioned above. Finally, a new hypothesis will be proposed to explain the intriguing similarity between mitochondrial genomes in brevirostris and caucasicus despite a significant difference in body size and song structure between them.
METHODS
Study areas and capture methods
The song recording, genetic sampling and ecological observations were made during our field research in Turkey, in the western and central Caucasus (Black Sea coast, Tuapse, Abkhazia, Caucasian state reserve, Teberda state reserve, Northern Ossetia), in Transcaucasia (southern Armenia, Meghri region), Ciscaucasia (Krasnodar oblast) and western Kopet Dagh (Turkmenistan, Syunt-Hasardagh nature reserve; Figure 2). Our song recordings from southern Germany, European Russia, the southern Urals and eastern Siberia were used for comparison. In all these areas, we conducted the observations, recordings and trapping at a time when the presence of migrating individuals in the population became unlikely. All males recorded and captured were observed actively singing and defending their territories. Most of them were in pairs, although the presence of females could not be identified in all cases. Some females were observed building nests.
Figure 2.
Locations of the data collection. The circles denote the locations where genetic samples were collected and song recordings were made. The asterisks indicate the locations where only song recordings were made. 1: Ciscaucasia, 2: Caucasian State Reserve, 3: Teberda State Reserve, 4: Tuapse, 5: Abkhazia, 6–10: northern Turkey, 11–12: southern Turkey, 13–14: eastern Turkey, 15: southern Armenia (Meghri), 16: western Kopet Dagh, 17: northern Ossetia.
Singing males were tape-recorded and then captured in mist nets after being lured by the song playback. Almost all males of menzbieri, caucasicus and even brevirostris (whose song is the most distinctive) were successfully attracted to mist nets by song of abietinus. We did not make any skin collections. Once pictured, measured and blood sampled (from the vena brachialis), the bird was immediately released back into its territory.
Morphology
All the males studied (n = 63) were caught at the beginning and in the middle of the nesting season and their plumage was characterized by a medium extent of wear. All measurements (wing and tail length) were taken by the same persons (IM and PK). We also analysed the wing formula, i.e. the length of the second outermost primary compared with the 6th, 7th and 8th ones. The measurement and comparison between primary lengths was done according to Svensson (1992).
Genetics
We examined mtDNA gene cytochrome-b (1043 bp). A total of 97 individuals from 4 species and 11 subspecies were used for genetic analysis, including 34 sequences from the GenBank (see Table S1 for the numbers). In addition, sequences of two Willow Warblers Phylloscopus trochilus were taken from the GenBank as an external group. The cladograms were constructed using maximum likelihood and Bayesian analysis algorithms.
DNA was extracted from blood samples using a standard phenol-chloroform procedure (Sambrook et al. 1989). PCR was performed for 35 cycles (30 s at 94°C, 30 s at 54°C and 30–120 s at 72°C) with the DNA-100 PCR kit (Isogen, Moscow, Russia). The cytb gene was amplified with the primers L-14995 and H-16065 (Helbig et al. 1995). PCR products were purified with alcohol precipitation and a PCR purification kit (Isogen, Moscow, Russia).
Sequences were initially aligned and then optimized by eye using BioEdit Sequence Alignment Editor v. 5.0.9 (Hall 1999). Mean uncorrected genetic distances (p-distances) between and within sequences were determined by MEGA v. 6.0 (Tamura et al. 2013). Phylogenetic analyses were conducted with Treefinder (Jobb et al. 2004) and MrBayes 3.1.2 (Huelsenbeck & Ronquist 2001, Ronquist & Huelsenbeck 2003) software. Confidence in tree topology was tested by nonparametric bootstrap analysis (Felsenstein 1985) with 1000 replicates and by posterior probability for Bayesian inference in MrBayes v. 3.1.2 (Huelsenbeck & Ronquist 2001).
Bioacoustics
We used digital sound recorders (Marantz PMD 660, PMD 222) with external condenser microphones (Sennheiser ME 66 with K6 module) for audio recording. The sonograms were created and analysed in Avisoft SASLab Pro with settings FFT = 512 and Hamming window. The peak frequency was determined as the frequency at maximum amplitude with 0.336 Hz resolution and the bandwidth threshold was set to 20 dB. The following eleven samples were created for bioacoustics analysis: (1) collybita from southern Germany (10 males), (2) abietinus from the Moscow region (10 males), (3) caucasicus from the vicinity of Tuapse (10 males), (4) caucasicus from Abkhazia (7 males), (5) lorenzii from the Caucasian Reserve (10 males), (6) lorenzii from the Teberda Reserve (10 males), (7) lorenzii from eastern Turkey (3 males), (8) menzbieri from the western Kopet Dagh (8 males), (9) menzbieri from southern Armenia (10 males), (10) brevirostris from western and central Turkey (10 males), (11) three Chiffchaff males recorded in the south of Turkey. The Chiffchaffs in sample (11) differed markedly from typical brevirostris in the plumage colouration and were analysed as a separate group.
An individual catalogue of song elements (distinct notes) was created for each male included in these samples. The overall (population) catalogue of the elements was created for each of the 11 samples. We used the most complete individual catalogues to ensure their contributions to the overall catalogue were as equal as possible. Seven measurements were made for each element: (1) maximum frequency, (2) minimum frequency, (3) peak frequency, (4) frequency range, (5) duration, (6) number of frequency peaks (modulation peaks) and (7) position of the first point in the change from ascending to descending modulation (Figure 3). In addition, the ratio of the elements with ascending and descending modulation (at the very beginning of the element) was calculated for each overall catalogue.
Figure 3.
Some measured parameters of the most typical song elements. Fmax: maximum frequency, Fmin: minimum frequency, black arrows: frequency of the first change of modulation from descending to ascending, grey arrows: frequency peaks. (A) typical element of brevirostris, not included in the repertoire of other taxa, (B) typical element of caucasicus, menzbieri and lorenzii, not included in the repertoire of other taxa, (C) typical element of menzbieri and caucasicus, only rarely performed by other taxa, (D) typical element with an ascending initial modulation from the repertoire of lorenzii, never executed by ‘greenish’ Chiffchaffs, but common in tristis and sindianus (Helbig et al. 1996, own data).
Statistics
Statistical analysis was conducted with Statistica v. 8 (Stat Soft Inc., USA) and PAST v. 2.13 (Hammer et al. 2001). We used the nonparametric Kruskal-Wallace test to identify the subspecies differences in wing and tail length. The measurement data of the song elements (notes) were subjected to discriminant analysis using the populations as a grouping variable. Among the results, a matrix of inter-group distances of Mahalanobis was obtained. The cluster analysis was then carried out on this matrix, using Euclidian distance and the clustering was done according to the algorithm Unweighted Pair Group Average.
RESULTS
Mitochondrial DNA variability
According to our analysis, all the Chiffchaff subspecies studied are divided into several clades, within which separate mitochondrial lineages can be distinguished (Figure 4). All the main clades correspond to the four well-separated species which are currently recognized. The first clade is formed by the Iberian Chiffchaff P. brehmii, the second one by the Canary Island Chiffchaff P. canariensis. The third main clade corresponds to the Mountain Chiffchaff P. sindianus. In our study, this species is represented by the subspecies lorenzii found in the Caucasus. This clade combines sequences of this subspecies from GenBank, North Ossetia (Caucasus), and one specimen from Eastern Turkey in the foothills of the Armenian highlands. According to its plumage coloration this bird obviously belonged to the group of ‘brownish’ Chiffchaffs.
The fourth clade combines the ‘greenish' collybita, abietinus, caucasicus, brevirostris, menzbieri and the ‘brownish’ tristis. This clade is divided into six lineages:
(1) The caucasicus/brevirostris lineage includes two GenBank sequences belonging to caucasicus and brevirostris, and unites all the ‘greenish' Chiffchaffs we caught in Western and Northern Turkey from the Bosporus to the Armenian highlands, as well as two ‘brownish’ birds from the mountainous regions of Eastern Turkey. Both of the latter upon capture were identified by their plumage coloration as typical lorenzii, but later it turned out that they carry the haplotype caucasicus/brevirostris. This lineage also includes four Chiffchaffs from Israel: three of them were caught on mount Hermon during nesting time and one near lake Tiberias in winter.
(2) The menzbieri lineage unites all the ‘greenish’ Chiffchaffs that we caught in southern Armenia (Meghri province) with a single individual menzbieri from the western Kopet Dagh, where only this form of Chiffchaff is known to occur. Thus, our data confirm the conclusion of Raković et al. (2019) on the distribution of menzbieri in southern Transcaucasia. The difference between our menzbieri sample and others is small (p = 0.37–0.69%) and intra-group variability is also low (p = 0.08%; Table 1).
(3) The P. c. unclassified lineage appears to be the most interesting. It combines two birds caught by us in southern Turkey (points 11 and 12; Figure 2) together with three birds caught earlier in the Hermon mountain in northern Israel. Both our birds had rather specific features in their plumage coloration (see below), but the birds from Israel were reported as not differing in this respect from the typical ‘brevirostris/caucasicus’ morphotype (Raković et al. 2019). According to the assumption of Raković et al. (2019), all three of these Israeli Chiffchaffs belong to a new form, not described previously.
(4) The tristis lineage corresponds to the Siberian Chiffchaffs found from Arkhangelsk and the Ural Mountains to Baikal Lake and the Chukotka peninsula and belong to the ‘brownish’ taxa. It is interesting that the tristis lineage is closely associated with the previous branch, i.e. P. c. unclassified.
(5) The collybita lineage combines ‘greenish’ Chiffchaffs belonging to the nominative subspecies that live in Western Europe. This group also included one bird from Israel (Mount Hermon, August 15).
(6) The abietinus lineage unites all representatives of the abietinus form (sequences from the GenBank). This form also belongs to the ‘greenish’ Chiffchaff and lives in Eastern Europe and in the European part of Russia.
Figure 4.
Phylogeny of Chiffchaff taxa studied according to cytochrome b sequencing data. The tree is based on 1041 nucleotides. The figures to the right of the geographical names indicate the number of individuals studied. The figures at the bottom of the cluster show its support (Bayesian/maximal likelihood).
Bioacoustics
The frequency parameters of the song and the shape of the modulation of the elements are subject to considerable variability; however, stable differences between taxa can be identified (Figure 5). The main results are as follows: (1) Among all ‘greenish' Chiffchaffs, the song brevirostris was significantly higher in all three frequency parameters (Figure 6). The remaining ‘greenish' taxa did not differ from each other in the frequency parameters of the song. (2) Brownish lorenzii performed a higher-frequency song than all the greenish Chiffchaffs, except brevirostris. The frequency range of the lorenzii song was considerably narrower compared with all other taxa. (3) The songs of all Caucasian Chiffchaffs – caucasicus, menzbieri and lorenzii – contained highly distinct elements with an acute V-shaped or V-shaped first modulation, whereas a smooth L-shaped modulation predominated in the songs of the collybita and abietinus. (4) The song of brevirostris was clearly distinguished by the number of frequency peaks in a single element. Many elements contain three and even four peaks. (5) The song of brevirostris differed by the position of the first frequency minimum (measured from the beginning of the element). (6) The elements with ascending modulation (at the initial segment) were found only in the repertoire of brownish lorenzii and were completely absent in all ‘greenish' taxa. It should be noted, however, that the proportion of elements with ascending modulation in lorenzii was rather small (on average 10.2%), and some of our recordings were completely devoid of such elements. For comparison, in the tristis song, the ascending elements were about 50%. (7) Unlike all ‘brownish' Chiffchaffs (including lorenzii), all ‘greenish’ Chiffchaffs tended to repeat the same notes during singing. This habit was particularly pronounced in brevirostris, caucasicus and menzbieri, in the songs of which up to 10–12 repetitions of the same note were found. Such singing sounds much more monotonous than the usual Chiffchaff singing with the ordered alternation of notes: not as “chiff-chaff-chiff-chaff-chiff-chaff”, but as “chiff-chiffchiff-chiff-chaff-chaff-chaff-chaff”.
Table 1.
Pairwise genetic distances between Chiffchaff taxa based on 1041 nucleotides of the mitochondrial cytochrome b gene. Numbers below the diagonal: proportion of nucleotides differing between taxa (in %). In the diagonal: intragroup variations.
On average, 48.5% (from 14.3% in abietinus to 96.7% in brevirostris) of the song elements were classified correctly by discriminant analysis (Wilks lambda = 0.14, F = 10.4, P < 0.001). All variables except the minimum frequency significantly contribute to the separation of samples. To visualize the results, we conducted a cluster analysis of the Mahalanobis inter-group distance matrix obtained in the course of discrimination (Figure 7). The Chiffchaffs from western, northern and southern Turkey are clustered into one node, clearly separated from all the others. Both Turkish samples are reliably grouped together, although the differences between them (Mahalanobis distance 0.49) are greater than, for example, between two Caucasian samples of lorenzii (0.21), between two samples of caucasicus (0.18) and between European collybita and abietinus (0.13). Two samples of caucasicus and two samples of menzbieri are clustered into one node; however, all three samples from the Caucasus, including the Armenian menzbieri, form a well-evidenced distinct node and appear to be separate from the Kopet Dagh's menzbieri. The sample from eastern Turkey is, with good evidence, associated with two Caucasian samples of lorenzii.
Figure 5.
Example sonograms of the song elements (notes) of all taxa studied: (A) collybita, (B) abietinus, (C) caucasicus, (D) menzbieri (Kopet Dagh), (E) menzbieri (Armenia), (F) brevirostris (western and central Turkey), (G) two males with unusual plumage coloration from southern Turkey, (H) lorenzii (Northern Caucasus), (I) lorenzii (eastern Turkey).
Figure 6.
Variations in the characteristics of the song elements (notes): (A) maximum frequency, (B) minimum frequency, (C) frequency range, (D) peak frequency, (E) position of the first point in the change of modulation from descending to ascending, (F) duration, (G) number of frequency peaks (modulation peaks), (H) ratio of the song elements (notes) with ascending and descending frequency modulation. The numbers along the horizontal axis designate the taxa and locations: 1: menzbieri, Kopet Dagh, 2: menzbieri, Armenia, 3: caucasicus, Abkhazia, 4: caucasicus, Tuapse, 5: brevirostris, western and central Turkey, 6: unclassified males, southern Turkey, 7: lorenzii, eastern Turkey, 8: lorenzii, Caucasian Reserve, 9: lorenzii, Teberda Reserve, 10: collybita, Germany, 11: abietinus, Moscow. In diagrams A–F, the mean and the standard deviation, as well as the limits are shown for each sample. The colour of the boxes identifies the ‘greenish' and ‘brownish’ Chiffchaffs. In diagrams G and H, the ratio of different vocal patterns is shown as a percentage of the total number of song elements. In diagram G, black columns: 1, white columns: 2, dark grey columns: 3, light grey columns: 4 frequency peaks per element. In panel H, grey columns: elements with descending modulation, black columns: elements with ascending modulation.
Figure 7.
The relationship between the geographical samples of the Chiffchaffs studied, based on the analysis of their population song repertoires. The numbers in square parenthesis correspond to the location point numbers in Figure 2. The numbers at the nodes show bootstrap support (pair-group average, Euclidean distance).
The song elements on the phonograms of the two Chiffchaffs recorded by us in March 2017, in the north of Israel (the Golan Heights region), contained three to four frequency modulation peaks, and most likely belong to the brevirostris dialect. Unfortunately, these recordings are very short, and therefore have not been included in the analysis.
Morphology
The nonparametric Kruskal-Wallace test shows that the subspecies differences in wing and tail length (Table 2) are statistically significant (P < 0.001). According to pairwise comparisons, brevirostris significantly differs in wing length from each of the other three subspecies (P < 0.05), while no significant differences were found among the latter. In terms of tail length, significant differences were shown to occur in pairs: menzbieri/abietinus (P < 0.05), brevirostris/caucasicus (P < 0.01) and brevirostris/abietinus (P < 0.001). Thus, brevirostris is the smallest of all four subspecies of the ‘greenish’ Chiffchaffs.
The wing shapes of caucasicus and menzbieri were found to be identical (the projection of the 2nd primaries between the 7th and 8th). The top of the wing in abietinus is most pointed (projection of 2nd primaries between 6 and 7), while brevirostris has the most rounded wing (projection of 2nd primaries between 8–9 and 10–11); menzbieri and caucasicus take an intermediate position and do not differ from one another (projection of 2nd primaries between 7 and 8 in both
Table 2.
Wing and tail length in the four ‘greenish’ Chiffchaff taxa (only males included).
DISCUSSION
The geographical area considered in this article contains the highest taxonomic diversity of the entire Chiffchaff range. Among all the forms of Chiffchaff studied, caucasicus and menzbieri have been found to be very similar, according to their morphological features (plumage colour, wing and tail length and wing shape). The songs of these two are also very similar and differ significantly from the brevirostris song on the ‘morphology’ of individual elements and frequency-time characteristics. In particular, the song of brevirostris with its smallest body size was found to have the highest frequency, which corresponds to the well-known inverse relationship between body size and song frequency characteristics (Hall et al. 2013, Liu et al. 2017). Curiously, the distribution of genetic traits seems to form the opposite pattern. Our data strongly corroborate the findings of Helbig et al. (1996) and Raković et al. (2019), that caucasicus and brevirostris are genetically very close, whereas menzbieri, in contrast, is much more distinct.
The western and northern regions of Turkey are traditionally included in the distribution range of the brevirostris subspecies (Kumerloeve 1967, Roselaar 1995, Kirwan et al. 2008). The morphological, genetic and bioacoustics features of all our specimens from this area (including those caught on the west and east coast of the Bosporus) were similar to each other and no doubt belong to this subspecies. Three Chiffchaffs from the mountainous areas of eastern Turkey were typical representatives of the ‘brownish’ group and their song did not differ significantly from the Caucasian lorenzii (Figure 5). Thus, based on the totality of the phenotypic characteristics (colour, size and song), all these Chiffchaffs should be assigned to the form lorenzii, which fully corresponds to the conclusion reached by Roselaar (1995). Surprisingly, only one specimen of the three was genetically identical with the Caucasian specimen of lorenzii from GenBank, while the other specimens carried mitochondrial DNA identical to the greenish brevirostris/caucasicus.
Figure 8.
The external appearance of Chiffchaffs (A) carrying the new haplotype from southern Turkey and (B) the typical brevirostris.
Two Chiffchaffs caught in the south of Turkey (points 11 and 12; Figure 2) had an unusual plumage colouration. Both individuals looked somewhat brighter than typical brevirostris and were covered with clear yellowish markings on the throat, chest, abdomen and even on the head (Figure 8). The song of these two Chiffchaffs clearly corresponded to the brevirostris dialect, although it had some peculiarity (Figure 5). The mitochondrial DNA tree shows good evidence that both these individuals clustered in a single node with three individuals from Israel (Raković et al. 2019). Unexpectedly, this node then closely associates with the Siberian Chiffchaff (tristis) This too is supported by good evidence. We assume that two of our specimens from southern Turkey belong to a relict population described by Raković et al. (2019) in Israel. It could be hypothesized also that this population is closely related to tristis, which some researchers consider to be the most ancient form of Chiffchaffs, close to the common hypothetical ancestor of the ‘greenish' and ‘brownish’ forms (Martens 1982). It is noteworthy that both Chiffchaffs caught in southern Turkey lived in small scattered remnants of relict forest (Lebanese Cedars Cedrus libani) that have survived in southern Turkey and adjacent territories. We assume that this relict population inhabits not only Israel, but the Lebanon, Syria and Turkey as well.
Our study has revealed brevirostris to be the most different among all three greenish taxa, due to its small size, wing formula and distinct vocal dialect. Brevirostris clearly differs from caucasicus and menzbieri in all these characteristics, and, in turn, caucasicus and menzbieri are more similar to each other. Paradoxically, the mitochondrial DNA of brevirostris was found to be almost identical to caucasicus, whereas menzbieri is very distinguishable, in this respect, from both of them (Helbig et al. 1996).
This paradox can be explained if we hypothesize that caucasicus originated as a result of ancient hybridization between brevirostris and menzbieri. The results of such hybridization could be a complete replacement of mitochondrial DNA in the population of one of the two species that were affected by the hybridization process (Tegelström 1987, Fedorov et al. 2009, Irwin et al. 2009). The history could be hypothesized as follows. Brevirostris originated in the mountains of northern Turkey, which were once covered with broadleaved forests. From that area, it moved eastward to the Transcaucasia, along the Black Sea coast and along the river valleys of the Armenian uplands. When brevirostris reached Transcaucasia, it encountered here the menzbieri, which had moved westward from its homeland in the broad-leaved forests of Elburz and the western Kopet Dagh. As a result of subsequent hybridization, the caucasicus emerged as a hybrid form, which quickly settled the Transcaucasia, the Caucasus, and is now rapidly settling northward in the Ciscaucasia steppes (our data). Phenotypically and genetically pure populations of menzbieri are retained in southern Transcaucasia, Elburz and Kopet Dagh. At the northern limit of the distribution of menzbieri, in the Transcaucasia, its hybridization with caucasicus seems very likely (Raković et al. 2019), considering the similarity of their habitats, appearance and, in particular, their song.
ACKNOWLEDGEMENTS
We are grateful to Thomas Kleinbub for improving our writing and Ilya Ukolov for excellent photo of the Caucasian Chiffchaff. We would like to thank Yaroslav Redkin and two anonymous referees for helpful comments on the manuscript. Our study received financial support from Russian Science Foundation (grant 20-14-00058). The dataset generated during this study is included both in this published article and in GenBank (see Table S1). The access to most of our study areas was not restricted in any way. No specific permissions were required for conducting our study in these areas. When working in nature reserves, we received permission to record and catch birds in accordance with the laws of the Russian Federation. The study did not involve endangered or protected species.
REFERENCES
Appendices
SAMENVATTING
Het over bijna het gehele Palearctische gebied voorkomende Tjiftjaf Phylloscopus collybita superspecies complex omvat vele taxa (soorten en/of ondersoorten). Dit superspecies wordt traditioneel als een van de meest complexe problemen binnen de taxonomie van Palearctische vogels beschouwd. Wij presenteren hier nieuwe gegevens over de genetische structuur, morfologie en geluiden van taxa van Tjiftjaffen uit Turkije, de Kaukasus, Transkaukasië en het grotendeels in Iran gelegen Kopet-Daggebergte (brevirostris, caucasicus, menzbieri en lorenzii). De eerste drie zijn groenig van kleur, de laatste is bruinig getint. In het zuiden van Turkije werden twee exemplaren aangetroffen met het nieuwe haplotype dat onlangs in het noorden van Israël is ontdekt. Beide individuen zagen er wat helderder en geler uit dan de typische brevirostris, maar hun zang kwam duidelijk overeen met het brevirostris-dialect. De vogels van het taxon brevirostris, dat in het westen en midden van Noord-Turkije voorkomt, waren door de kleinere grootte, vleugelformule en hun uitgesproken geluidenrepertoire het duidelijkst van de groenig getinte individuen van de andere taxa te onderscheiden. Paradoxaal genoeg is het mitochondriale DNA van brevirostris bijna identiek aan dat van caucasicus, terwijl menzbieri in dit opzicht van beide verschilt We menen dat deze paradox kan worden verklaard door aan te nemen dat caucasicus lang geleden is ontstaan uit hybridisatie van brevirostris en menzbieri. De relaties tussen brevirostris en lorenzii, voorkomend in de bergachtige streken van Oost-Turkije, laten een discrepantie zien tussen fenotypische en genotypische eigenschappen in sommige individuen, die ook het gevolg van hybridisatie zouden kunnen zijn.
