The problems of genetic support of dividing the black kite ( Milvus migrans ) into subspecies

The black kite Milvus migrans is a common bird of prey demonstrating remarkable ecological plasticity. It inhabits a variety of habitats and is an increasingly synanthropic species. The black kite is widespread in Eurasia, Africa, Australia and adjacent islands. Palearctic kites migrate to Africa, India and China in winter, but kites of Africa and Australia are partly sedentary and partly seasonal migrants. The wide range and high mobility are the reasons of a complex population structure of the black kite. Commonly five to seven M. migrans subspecies are distinguished, each of which is widespread over extensive areas and has more or less an apparent phenotype. Recently, studies of genetic differences between black kite populations started to emerge. On the grounds of earlier studies of mitochondrial and nuclear genes of this species, we check whether there is a genetic support for separation of the black kite subspecies. Recent studies of some mitochondrial loci substantiate the recognition of at least the European (M. m. migrans), Asian (M. m. lineatus and M. m. govinda), African (M. m. aegyptius and M. m. parasitus), and Australian (M. m. affinis) black kite subspecies. Furthermore, the mitochondrial haplotype difference suggests that the African yellow-billed kite, including M. m. aegyptius and M. m. parasitus, should be a separate species as already proposed, or even two separate species.


Introduction
The black kite Milvus migrans is a diurnal raptor of the fa mi ly Accipitridae, inhabiting a wide breeding range including continental Eurasia, Africa, Australia and several islands.This bird of prey shows remarkable ecological plasticity and inhabits a variety of habitats from the forest and foreststeppe zones to dry steppes and semideserts, as well as various mountain habitats.The black kite is one of the few raptors that successfully adapt to living next to humans.It intensely populates the neighborhoods of landfills, agricultu ral and parasitus migrans livestock complexes and increasingly nests even in large cities.In the 20th cen tury, the black kite along with other raptors suffered from persecution, pes ticides, and landscape changes due to human activity.Now these dangers still exist, although at a reduced scale, since poisonous chemicals and hunt ing are tightly controlled in European countries.Populations of many birds of prey were dangerously reduced in the second half of the 20th century, and they needed protection and restoration measures (examples are the red kite Milvus milvus, the bengal griffon Gyps bengalensis, and the peregrine falcon Falco peregrinus).In contrast, the black kite is quickly and independently restor ing its population and rapidly spreading to new habitats, becoming a common synanthropic species.
Commonly from five to seven black kite subspecies are distinguished, each of which is widespread over an exten sive range.Recently, I.V. Karyakin analy zed about 800 black kite photo graphs in several ornithological data bases (mostly Raptors of the World in GBIF) and produced a map of subspe cies distribution (Karyakin, 2017).The Palearctic is populated by two subspe cies: the western part is occupied by the European black kite M. m. migrans, and the eastern part, by the blackeared kite M. m. lineatus.The kites belonging to these subspecies differ in phenotype, but their ranges overlap in Western Siberia forming a wide intergradation zone, where birds with various intermedi ate phenotypes occur (see the Figure ).M. m. migrans and M. m. lineatus are migratory birds and spend the winter mainly in Africa and the Indian subcon tinent (FergusonLees, Christie, 2001).Southern and Equatorial Africa are po pulated by the yellowbilled kite, which is usually clearly different from the Eura sian subspecies, sometimes con sidered a separate species (see below) and, in turn, divided into two subspecies M. m. parasitus and M. m. aegyptius.India and the IndoChine countries are inhabited by the sedentary pariah kite M. m. govinda, and Australia, together with the adjacent islands are inhabited by the forktailed kite M. m. aff inis, which migrates seasonally within the Australian continent (Marchant, Hig gins, 1993).The Indian and Australian subspecies are very similar in their phe notypes, but they are considered to be strictly separated by distance.Some research ers recognize a special Taiwan kite M. m. formosanus, inhabiting southeastern China (see the Figure ).
Other previously proposed races currently are considered as synonyms and are included in the subspecies mentioned above.In 1908, S.A. Buturlin described the subspecies M. m. ruf iventris from Turkmenistan, later considered as a red morph of M. m. migrans and M. m. ferghanensis from Kyrgyzstan, included in M. m. lineatus (Dement'ev, 1936).Lectotypes and paralectotypes of both races are conserved in the Zoological Museum of Moscow University (Tomkovich, 2001).In 1923, the same author described M. m. tianshanicus from Kyrgyzstan mounts, which was shortly included in M. m. lineatus, too.Probably, it resulted from interbreeding of this taxon with nominative subspecies (Dement'ev, 1936).The next two races are currently included in M. m. aegyptius.M. m. tenebrosus was described in 1933 by Grant and MackworthPraed.Wintering specimens of this kite collected in 1922 (holotype) at the Ivory Coast and in 1954 in Nigeria are stored at the Natural His tory Museum (London) (GBIF Secretariat, 2017).M. m. arabicus was described by Swann in 1922, one of specimens of this proposed subspecies was collected in Lahej, Saudi Arabia, in 1899, and now it is stored at the American Museum of Natural History as a holotype (Trombone, 2013), other specimens could be found in several museums.During the examination of yellowbilled kites from Saudi Arabia and Yemen in the Natural History Museum at Tring, England, P.J. Mundy (2011) did not distinguish skins labeled M. m. arabicus from M. m. aegyptius; supposedly, the identification is not possible by their morphology.However, the nesting range of this race is situated in a semiisolated area and the existence of this taxon can be questioned, though we should notify that the information on M. m. arabicus is insufficient.Data on genetic features for any of these races are not available; however, they may be of certain interest.

Population genetics of the black kite
Until now, researchers have been focused on the European black kite population.Usually the black kite is studied together with a close species, the red kite (Milvus milvus), which is a nearly threatened species breeding mainly in Central and South ern Europe.Roques and Negro (2005) hypothesized that the red and black kites diverged as separate species during the last glaciation in the Pleistocene, and now they are isolated reproductively, although the isolation is incomplete (Heneberg et al., 2016).These kites show sympatric speciation, the earlier archaeozoological Breeding ranges of the black kite subspecies (Karyakin, 2017).
record of Milvus representative are black kite bone remains from Germany dated by 10 000 ± 2 000 years BP, and the remains of both species come from the same European areas since the Neolithic.In that time, coniferous boreal forests, avoided by kites, gave way to temperatezone forests and agriculture started vigorously, resulting in wood fragmentation (Schreiber et al., 2000).In that early study, allozyme analysis of both species populations from SachsenAnhalt in Germany was carried out and these kites were considered very close genetically according to Nei's interspecies distance.However, the black kite is distributed much wider than the red kite and rare hybridization between these species could not apparently result in any significant changes in the black kite gene pool.
Phylogenetic studies on black kites are limited, and only a few researches analyzed relatively representative samples that allowed conclusions to be drawn about the kite population structure.Johnson et al. (2005) analyzed the CytB and ND2 mitochondrial gene sequences (2146 bp in total) in 43 black and red kite samples from 27 locations, including Western Europe, Africa, South Asia, Japan, Australia, and New Guinea.Mitochondrial control region (547 bp) was also sequenced for 26 samples of the set.The authors managed to construct an informative phylogenetic tree despite a small sample size.That tree demonstrated that CytB and ND2 haplotypes of the African M. m. aegyptius and M. m. parasitus differ from the haplotypes of the other M. migrans subspecies no less than they differ between the species M. migrans and M. milvus.Ac cording to the tree, the European M. m. migrans and Australian M. m. affinis haplotypes were found to be separate branches of the tree, but the Asian subspecies haplotypes (M.m. lineatus and M. m. govinda) were merged (Johnson et al., 2005).
The purpose of the study (Johnson et al., 2005) was to determine whether the black kites previously inhabiting the Cape Verde Islands were representatives of a particular subspecies.Therefore, almost all the samples used in this work were obtained from old museum collections and were collected mainly in the first half of the 20th century or even earlier.Since then, the black and red kite populations have experienced significant changes due to human activity.Kite populations had been reduced strongly or even disappeared completely from many areas, which was followed by repopu lation (FergusonLees, Christie, 2001).Thus, at least some black kite populations passed through a bottle neck and the founder effect, which certainly affected haplotype frequen cies.Hence the current haplotype distribution may differ from that obtained by Johnson et al. (2005).Scheider et al. (2004) used the mitochondrial CytB gene (1143 bp) to construct a phylogeny of black kites.According to this preliminary work, European M. m. migrans haplotypes and haplotypes of M. m. lineatus from Mongolia formed two separate branches, and the African M. m. parasitus was closer to the red kite than to the black kite.In a later research (Scheider et al., 2009) more individuals of M. m. parasitus were added in the analysis and most of them formed a particu lar clade between M. milvus and others M. migrans.Heneberg et al. (2016) examined 184 red kite samples and 124 black kite samples from Germany and the Czech Republic.The authors studied phylogeny on the base of mitochondrial CytB (493 bp) and CO1 (564 bp) gene fragments.The nuclear Myc (352 bp) gene was also examined, but phylogenetic trees were built for each gene separately.The authors also used sequences available in the GenBank database and concluded that the CytB gene is best suited for black kite subspecies differentiation, although a relatively small gene fragment was used.Ten closely related CytB haplotypes were found in M. m. migrans samples from Germany and the Czech Republic, including the most common core haplotype.CytB sequences of another M. migrans subspecies from GenBank were included to the phylogenetic tree.M. m. lineatus and M. m. govinda haplotypes were very close to each other and to M. m. migrans haplotypes, but Australian M. m. affinis haplotype formed a separate branch of the tree.Even this small CytB gene fragment shows that the African M. m. parasitus haplotype is different from the other M. migrans haplotypes, their genetic distances exceed those between M. migrans and M. milvus (Heneberg et al., 2016).
The same authors (Heneberg et al., 2016) found out that the nuclear Myc gene was not appropriate for discrimination of M. migrans and M. milvus species.All Myc polymorphic sites in the red kite were also polymorphic in the European black kite population.It would be impossible to determine exactly which species a particular Myc sequence belongs to, although the frequencies of Myc alleles in these two species are different.Thus, it cannot be expected that the Myc gene can be used to distinguish black kite subspecies.As nuclear genes have a slow evolution rate, such a result could be expected for young species.
Recently, our group published a small study of CytB haplo types in the M. m. lineatus population from the Tyva Republic (Andreyenkova et al., 2018).We examined 20 independent black kite families and concluded that the most common haplotypes in this population coincided with the minor hap lo types found in the European M. m. migrans population by Heneberg et al. (2016).In addition, a CytB haplotype was dis covered in Tuva M. m. lineatus, which had been found pre viously in India and Pakistan (CytB-19 in terms of a study by Heneberg et al., 2016) and been assigned to the Indian M. m. govinda.However, since the same haplotype was found in Japan (Johnson et al., 2005), where M. m. govinda does not live, then the CytB-19 haplotype unlikely belongs to M. m. govinda.

Genetic data on individuals of the M. migrans subspecies
The studies mentioned above (Scheider et al., 2004(Scheider et al., , 2009;;Johnson et al., 2005;Heneberg et al., 2016) demonstrate that some mitochondrial loci allow to distinguish, at least the Eu ropean (M.m. migrans), Asian (M. m. lineatus and M. m. govinda), African (M.m. aegyptius and M. m. parasitus), and Australian (M. m. affinis) black kite subspecies.Below we discuss genetic data for each subspecies in more detail.
Yellow-billed kite, M. m. aegyptius and M. m. parasitus.It is generally accepted that the yellowbilled kite includes two subspecies (M.m. aegyptius and M. m. parasitus), which breed in Northeastern and SubSaharan Africa, respectively (see the Figure ).These kites differ from the European M. m. migrans, which are wintering and nesting in Africa.According to all existing data on CytB gene, the yellowbilled kite differs from all other black kite subspecies no less than the black kite from the red kite, and in most cases the yellowbilled and black kites form even separate clades (Wink, SauerGürth, 2000, 2004; эволюция генома / genome evolution Scheider et al., 2004Scheider et al., , 2009;;Johnson et al., 2005).In a study of West African bird phylogeny based on the COI mitochondrial gene, one M. migrans from Nigeria, probably belonging to M. m. parasitus, also formed a clade with M. milvus, and other M. migrans of unclear origin were more distinct (Echi et al., 2015).More comprehensive phylogenetic analysis carried by Minimal Evolution and Neighbor Joining methods with bootstrap support reveals that samples of M. m. parasitus from the Ivory Coast and M. m. aegyptius found in Kenya constitute a clade close to all other M. migrans (and also Cape Verde M. milvus fasciicauda); whereas one South African M. m. parasitus gets into the clade with most of M. milvus (Scheider et al., 2009).Nevertheless, position of yellowbilled kites with respect to other taxa is not well supported in that research (values are less than 50 %), so their exact place in kite phylogeny remains open.
Additionally, CytB and ND2 genes haplotypes of M. m. parasitus/aegyptius analyzed together were clearly divided into two branches: the first one included seven samples collected in Equatorial Africa and in the south of the Arabian Peninsula, the second one included three samples from South Africa and Madagascar.These two branches were divergent from each other to the same extent as the black kite from the red kite (Johnson et al., 2005).Therefore, the mitochondrial haplotype difference suggests that the yellowbilled kite should be a separate species, as already proposed (Wink, SauerGürth, 2000), or even two separate species.
Taxonomy of M. m. parasitus/aegyptius is uncertain be cause the yellowbilled kite is poorly investigated.Young and even adult yellowbilled kites might be sometimes confused with young black kites.In addition, the yellowbilled kite is a seasonal migrant in some areas, and it can occasionally occur outside the breeding range (Clark, Davies, 2018).All these facts lead to misidentification of kites, especially if the birds are caught during migrations.Some sources consider the yellowbilled kite as a separate species Milvus aegyptus and divide it into subspecies M. a. aegyptus and M. a. parasitus (Morris, Hawkins, 1998;Sinclair, Langrand, 1998;Ferguson Lees, Christie, 2005;Gill, Donsker, 2018), but other checklists still interpret M. m. parasitus and M. m. aegyptius as black kite subspecies (Dickinson, Remsen, 2013;Clements et al., 2018;Orta et al., 2018).In any case, further research is required, including both the study of yellowbilled kite biology and molecular studies involving a large number of samples and additional mitochondrial and nuclear markers.
European black kite, M. m. migrans.This is the best characterized black kite subspecies in terms of both biological features and genetics.Relatively complete information about M. m. migrans seasonal migrations is evaluable (Panuccio et al., 2013;Literák et al., 2017).Various sequences of the mito chondrial CytB, ND2, and CO1 genes; mitochondrial control region; nuclear Myc gene; and some other European black kite genes are published in the GenBank database.However, two of minor M. m. migrans CytB haplotypes were found as major haplotypes in the nesting M. m. lineatus population in Tuva Republic (Andreyenkova et al., 2018), which probably reflect a gene stream from the eastern black kite population to the western one.In any case, analysis of a sufficient sample of other subspecies is required to clarify phylogenetic relation ship between M. m. migrans and other black kite subspecies.
Black-eared kite, M. m. lineatus.It has been earlier sug gested to consider the blackeared kite as a separate species, M. lineatus (Sibley, Monroe, 1990;Zhukov, 2012;Avibase taxonomic concepts, 2017), but this proposal is inconsistent with population research.As reported by Karyakin (2017), M. m. migrans and M. m. lineatus interbreed freely and, appar ently, penetrate deeper into the ranges of each other, expanding an intergradation zone.According to Scheider et al. (2004) and Andreyenkova et al. (2018), M. m. migrans and M. m. lineatus CytB haplotypes form two separate branches of a phylogenetic tree.A CytB haplotype from Japan (where M. m. lineatus breeds) is in the same haplogroup as samples from India and Pakistan (Johnson et al., 2005), where M. m. lineatus, M. m. govinda and, possibly, M. m. migrans breed (Karyakin, 2017).It is therefore unknown which subspecies this hap logroup belongs to.Recently complete mitochondrial genome of a black kite from South Korea was published (Jeon et al., 2018).This territory is included into the M. m. lineatus range, although the black kite is a rare species there.
We compared CytB and ND2 haplotypes from this individu al with the haplotypes published by Johnson et al. (2005) and found out that the Korean kite haplotype belongs to the same haplogroup as the samples from Japan, India, and Pakistan.Thus we conclude that this haplotype cannot be assigned to M. m. govinda, as it was done before (Heneberg et al., 2016).It is more likely that all the birds having this haplotype were representatives of M. m. lineatus, including individuals from India and Pakistan, especially since they were caught in winter, when M. m. lineatus reside in this area.On the other hand, it is still conceivable that M. m. lineatus and M. m. govinda do not differ in CytB sequence.There are only two studies using reliable M. m. lineatus samples from the Tuva Republic (An dreyenkova et al., 2018) and Mongolia (Scheider et al., 2004).Unfortunately, the latter group published only preliminary results and did not submit the CytB sequences to GenBank.In general, it is only known for now that M. m. migrans and M. m. lineatus differ in CytB sequence significantly but to a lesser degree than the species M. milvus and M. migrans.It turns out that the most widespread black kite subspecies is very poorly characterized in terms of genetics.Population biology and seasonal migrations of M. m. lineatus are also insufficiently understood.
Pariah kite, M. m. govinda.This subspecies is widespread throughout the Indian subcontinent from Pakistan in the west to IndoChina in the east (Karyakin, 2017) and is very poorly investigated, too.M. m. govinda is possibly intergradated with other black kite subspecies in the northwest and northeast of its range; moreover, M. m. lineatus spends winters in the range of M. m. govinda while M. m. govinda is considered to be sedentary.Therefore, these subspecies are difficult to distinguish despite the phenotypic differences, and in some cases the validity of subspecies identification is questionable.The only study sequencing mitochondrial genes of kites from India was performed by Johnson et al. (2005).However, all three samples from India used in that study were collected during M. m. lineatus wintering period.In addition, as men tioned above, their CytB and ND2 haplotypes were very close to the haplotype of the sample from Japan (Johnson et al., 2005) and the sample from South Korea (Jeon et al., 2018), so we suggest that all the four samples likely belonged to The problems of genetic support of dividing the black kite (Milvus migrans) into subspecies M. m. lineatus.Thereby it can be concluded that no DNA sequences of reliable M. m. govinda have been published so far.Phylogenetic relationships of M. m. govinda with other black kite subspecies remain to be established.
Fork-tailed kite, M. m. aff inis.This subspecies is wide spread in Australia, New Guinea, and Indonesia.According to current data (Karyakin, 2017) M. m. aff inis have no contacts with other black kite subspecies; therefore, significant genetic differences between M. m. aff inis and other subspecies could be expected, as is the case of the African yellowbilled kite.Unfortunately, very little data on the Australian black kite genetics is available.One Australian sample was used in the study of Scheider et al. (2004); based on CytB gene the sample was very far from the other black kite subspecies in a phylogenetic tree, both the red and black kites were ingroups.However, the authors themselves considered their result unre liable due to possible DNA degradation.This did not prevent Australian authors from the suggestion that the forktailed kite was the ancestor for other subspecies, and Australia was the center of black kite origin (Debus, 2005).These conclusions seem unfounded, since there is not enough information on the Australian black kite genetics.In addition, M. m. aff inis looks very similar to Indian M. m. govinda.These subspecies are almost indistinguishable in phenotype, reflecting possible close relation.Beside this study, only Johnson et al. (2005) se quenced CytB and ND2 genes of four M. m. aff inis birds: (two from Australia, one from Indonesia, and one from Papua New Guinea).These sequences form a special group but belong to the same branch as the M. m. lineatus/govinda haplotypes, indicating a fairly close relationship between the Australian and Asian black kite populations (Johnson et al., 2005).
There are two short (424 and 429 bp) CytB sequences in Genbank database belonging to black kites (possibly the same bird, no substitutions between the sequences) from India, Andhra Pradesh (HM804897, KC439281, submitted by S. Shivaji, A.P. Reddy, and B.G. Sasikala).It is important that the samples were collected in summer when wintering M. m. lineatus had left India.We compared these sequences with the known black kite CytB sequences and found that they clearly belonged to M. m. aff inis haplogroup from the study of Johnson et al. (2005), despite the sequences were rather short.This unexpected finding made us think that Australian black kites were not as isolated as previously thought.Another explanation is that the sequences from India belonged to M. m. govinda and therefore M. m. aff inis and M. m. govinda are close in CytB.
Taiwan kite, M. m. formosanus.This subspecies belongs to a polytypic group of subspecies together with M. m. lineatus and inhabits the eastern part of China, Taiwan and Hainan.The range is located at the junction of M. m. govinda and M. m. lineatus ranges (see the Figure).Some taxonomists do not recognize M. m. formosanus as a separate subspecies because its phenotype is intermediate (Karyakin, 2017).It is likely that the range of this subspecies is actually an intergrada tion zone of M. m. govinda and M. m. lineatus.We could not find any information on genetics of this subspecies.

Conclusions
We conclude that the current knowledge of black kite phy logenetics is very far from complete, and genetic research is limited mainly to the European subspecies M. m. migrans.The largest amount of information in the GenBank database is related to the mitochondrial CytB gene.Comparison of CytB sequences in different black kite populations indicates that this gene allows the distinction of at least part of the subspecies (Scheider et al., 2004;Johnson et al., 2005;Heneberg et al., 2016;Andreyenkova et al., 2018).As for nuclear markers, data is insufficient: only Myc gene was investigated on a rep resentative sample of red and black kites, but this sequence did not even allow to differentiate these two species (Heneberg et al., 2016).
A collection of black kite samples from the range of all subspecies and intergradation zones between subspecies is required for further study of the black kite.Mitochondrial and nuclear markers (such as microsatellites and SNPs) will clarify black kite population structure features, as well as dispersion history after the end of the Pleistocene glaciation.