Molecular phylogeny, morphology and life-history comparisons within Circus cyaneus reveal the presence of two distinct evolutionary lineages

Abstract

Background

Circus cyaneus is a medium-sized bird of prey that is widely distributed across the Northern Hemisphere. There are two currently recognized forms, the Palearctic form C. c. cyaneus (Hen Harrier), and the Nearctic form C. c. hudsonius (Northern Harrier). The forms have recently been split by the British Ornithologists’ Union but the American Ornithologists’ Union and some other taxonomic committees have not yet made any change. Here we examine the phylogenetic relationship between the two forms using sequence data from multiple nuclear and mitochondrial genes and examine breeding biology, body size, morphology, dispersal and other behaviors.

Methods

In order to fully compare cyaneus and hudsonius, we carried out a full literature review, measured museum skins and carried out phylogenetic analysis using a number of different mitochondrial genes and compare our findings to other recent work.

Results

We find that these two allopatric taxa form reciprocally monophyletic groups, show substantial mtDNA sequence divergence, and further differ significantly with respect to body size, plumage characters, breeding biology, dispersal and other behavioral traits.

Conclusions

Based on an array of consistently divergent characteristics, it is suggested that the two forms are best regarded as separate species, Hen Harrier (Circus cyaneus) and Northern Harrier (Circus hudsonius).

Background

Circus cyaneus is a medium-sized diurnal raptor found across most of the Northern Hemisphere. Known as “Annoch-kee-naepeek-quaeshew” (Snake Hunter) by the Cree Indians (Swainson and Richardson 1831), it was first described by Edwards in 1750 as “the Ring-tailed Hawk” (Edwards 1750) and subsequently classified as Falco cyaneus by Linné in his original 1766 classification of birds (Linné 1766). Edwards also described the bird as “The Blue Hawk” (Edwards 1758) and then “Marsh Hawk” (Edwards 1760). After being associated with a number of different scientific and common names, the Nearctic form was officially designated Circus hudsonius when the genus Circus was first recognized (Swainson and Richardson 1831). The British Ornithologists’ Union (BOU) later adopted the scientific name Circus cyaneus in 1883, recognizing “an allied form” in North America (BOU 1915). In 1920 the Palearctic form was recognized as a subspecies, C. c. cyaneus (Witherby 1920), again acknowledging the existence of a different, but unspecified, race in North America. The British common name for the form has always been Hen Harrier.

This taxonomy remained so until 1931 when the Nearctic form was relegated to sub-specific status and recognized as C. cyaneus hudsonius by the American Ornithologists’ Union (AOU). The AOU considered the Nearctic and Palearctic forms to be conspecific (Peters 1931), such that the Eurasian form continued to be recognized as C. c. cyaneus. The North American common name of Marsh Hawk for the Nearctic form remained so until 1983, when the currently recognized common name became Northern Harrier when it was first recognized by the AOU (1983).

Recently though, the BOU has acknowledged the status of Circus cyaneus as representing two species—Hen Harrier (Circus cyaneus) and Northern Harrier (Circus hudsonius) (Sangster et al. 2016), based on differences in plumage and morphometrics, substantial genetic divergence between cyaneus and hudsonius in mitochondrial and nuclear DNA (divergence similar to or larger than between several other recognized species of Circus) and a closer relationship between hudsonius and Cinereous Harrier C. cinereus than to cyaneus. This move has also been followed by a number of other authors (Ferguson-Lees and Christie 2001; Rasmussen and Anderton 2005; Gill and Donsker 2015). This movement of splitting the forms into two species has not been followed by the AOU (Chesser et al. 2015, 2016) and others.

North Americas’ current temperate climate allows access to suitable Northern Harrier breeding areas north of the Arctic Circle. These same areas become completely uninhabitable during the winter and the populations that breed there migrate south in autumn to overwinter in more hospitable regions. Terrestrial bird species that normally undertake an annual long-distance migration typically face constraints that inhibit them from dispersing in an east–west direction between the major landmasses (Boehning-Gaese et al. 1998). Such dispersal constraints, along with the notable tendency to avoid crossing large bodies of water, may account for the lack of dispersal that has been observed between the Nearctic and Palearctic forms of Circus cyaneus, although extremely rare occurrences of C. c. hudsonius in the Western Palearctic have been documented (Martin 2008; Mullarney and Forsman 2011).

Previous phylogenetic research on harriers has concentrated mainly on the relationship of the genus Circus to various other raptors. Both cyaneus and hudsonius were included in a phylogenetic reconstruction of Mediterranean raptors (Wink and Seibold 1996), showing they were separate, sister taxa but no specific comment regarding their historical relationship was made. An additional study of genetic relationships within Holarctic raptors also included both of the forms and commented that the two forms are considerably genetically divergent and might represent distinct species (Wink et al. 1998). Furthermore, the same author also observed that cyaneus and hudsonius show 1.7 % divergence in the Cytochrome b gene (Wink and Sauer-Gürth 2000). In “Harriers of the World” (Simmons 2000), a molecular phylogeny of the harriers (Circus spp.), considers the cyaneus and hudsonius forms to be separate species and notes that they are treated as such because “their genomes are more strongly divergent than those of species already separated on other grounds”.

Historically, there has clearly been some level of disagreement, and perhaps even confusion, regarding the appropriate taxonomic ranking of cyaneus and hudsonius. Recently however, an in-depth molecular phylogeny of all the harrier species was carried out, using one mitochondrial and three nuclear loci and concluded that cyaneus and hudsonius were two distinct species, with Cinereous Harrier (Circus cinereus) being a sister species of hudsonius (Oatley et al. 2015). However, despite the different taxonomies employed by previous workers studying Circus cyaneus, no formal systematic treatment of the two forms has been performed. Here, we build on previous phylogenetic work (Oatley et al. 2015) and carry out a comparative analysis of several key characters associated with the life-history and morphology of the two forms cyaneus and hudsonius and discuss their taxonomic ranking in light of these new perspectives.

Methods

A complete review of the scientific literature for Circus cyaneus was carried out, which are covered in the “Discussion” section under “Vocalization”, “Habitat”, “Distribution, dispersal and migration” and “Breeding behavior (Mate choice, Nest site and Male desertion)”. In addition, analyses of phylogeny and morphology (size difference and plumage) were also carried out as follows.

Phylogenetic analyses

Frozen tissue samples for nine representatives of hudsonius and five representatives of cyaneus were obtained from various museum collections for the molecular phylogenetic analyses of the mitochondrial Cytochrome b gene (Additional file 1: S1). A single additional Cytochrome b sequence of C. c. cyaneus obtained from GenBank (accession number X86745) (Benson et al. 2002) served as an independent verification of sequencing accuracy during alignment, and was included in the final phylogenetic analyses. A Cytochrome b sequence for Circus aeruginosus (Western Marsh Harrier) from GenBank (accession number AY987305) was designated as the outgroup in the phylogenetic reconstructions. Full details of tissue extraction and DNA isolation can be found in the Additional file 2: S4.

Of the 9 hudsonius samples obtained, 8 amplified well enough for sequencing, while of the 5 cyaneus samples obtained, only 3 amplified sufficiently to allow sequencing. Specifically, the toe-pad samples acquired from The University of Copenhagen did not amplify. The hudsonius samples were designated EX1.1–1.3, EX2.1 and EX2.3–2.6, while the cyaneus samples were identified as EX3.1–3.3.

In addition to our Cytochrome b sequence data, we also used publicly available data for various gene sets. To create our second dataset, we examined the Cytochrome c oxidase subunit 1 (COI) gene for which a number of barcoding sequences (7 for cyaneus and for 4 hudsonius) are available in GenBank.

Finally, for our third dataset we downloaded sequences from a previous study (Oatley et al. 2015), where the authors sequenced a 1.2 kb fragment of NADH dehydrogenase (ND1). We extracted all Circus cyaneus sp. sequences from ND1 and carried out identical analyses as used in our Cytochrome b analyses. We also refer to the authors’ findings for three nuclear loci, namely, Myoglobin intron-2 (MB), Beta Fibrinogen intron-5 (FGB) and TGFβ2 intron-5 (TGFB2) (see “Discussion” section).

All datasets were aligned using ClustalW (Thompson et al. 2002) through the graphical interface MEGA6 (Tamura et al. 2013). For each dataset the corresponding gene for C. aeruginosus (Western Marsh Harrier) was used as an outgroup. The model test method in MEGA6 was used to identify the best Maximum Likelihood (ML) substitution model for each set of aligned sequences. The best model was noted and the appropriate tree construction algorithm for the data was used to construct a ML phylogenetic tree, with 1000 bootstrap replications. Also, after assigning the sequences to groups (“hudsonius” or “cyaneus”), pairwise, inter-group and intra-group genetic distances (p-distance) were calculated, along with nucleotide content.

Morphology

New measurements of museum specimens, plus data from material referenced in the literature, were used to examine size differences between the two forms. Data collected from the literature (Scharf and Hamerstrom 1975; Watson 1977; Cramp and Simmons 1980; Palmer 1988; Johnsgard 1990; Wheeler and Clarke 1995; MacWhirter and Bildstein 1996; Simmons 2000) was not combined with our new measurements of study skins in the final statistical analyses, due to the confounding effects of individual variation in the biometric measurements taken by different workers. The following measurements were taken: body length (distal tip of central tail feather to proximal tip of bill), wing cord, tail length (distal tip of central tail feather to base of tail), bill-nostril to tip, bill-cere to tip, bill depth (dorsal surface of upper mandible to ventral surface of lower mandible at base of bill) and tarsus length. The measurements taken follow those made by other authors. The biometric data were analyzed according to a multivariate analysis of variance (MANOVA) for joint analysis of males and females between both species and two-way ANOVA for single-sex comparisons, using the Genstat software package (Payne et al. 2008).

Plumage

In order to compare variation in plumage characteristics of the harriers, notes and photographs were taken whilst examining museum specimens and additional reference was made to photographs published in the literature and found online.

Results

Phylogenetic analyses

We aligned cyaneus and hudsonius gene datasets for Cytochrome b (Cytb), Cytochrome c oxidase subunit 1 (COI) and NADH dehydrogenase (ND1), calculated the best ML substitution model for the data and used that model to construct the best phylogenetic tree for the alignment (Fig. 1). For all three datasets, HKY85 was found to be the best model (Hasegawa et al. 1985).

Fig. 1

Phylogenetic trees with bootstrap values for C. c. cyaneus and C.c. hudsonius for a Cytochrome b, b Cytochrome Oxidase subunit 1 and c NADH dehydrogenase 1. C. aeruginosus (Western Marsh Harrier) was used as the outgroup for all trees

As can be seen from the phylogenies in Fig. 1, all of the sequences from cyaneus and all the sequences from hudsonius cluster into separate clades supporting the monophyly of each clade.

As a summary, for each dataset, genetic distances (p-distance) within and between each form, as well as overall minimum, maximum and mean genetic distances are shown in Table 1.

Table 1 Summary of pairwise genetic distances matrix for Cytochrome b, ND1 and COI sequences providing minimum, maximum and mean distances (%) for both intra-form and inter-form comparisons

Cytochrome b

After alignment, the sequences were edited manually (gapped regions trimmed at start and end), resulting in a total sequence length of 719 nucleotides for all but one sample (C. cyaneus EX3.2, which was 653 nucleotides long). A close examination of the molecular data shows that the Cytochrome b sequences for the hudsonius samples are identical, except for EX1.2, which has a single transition mutation (C to T). The cyaneus sequences, on the other hand, showed a slight degree of variation with two variable sites. No indels were located in any of the sequences. From the above mutations, each of the forms had one non-synonymous change. In hudsonius the single transition leads from an Arginine to a Tryptophan and in cyaneus the non-synonymous transition leads from a Glycine to an Asparagine.

There are also 11 positions where all samples of one form consistently show a different nucleotide sequence to all samples of the other. All of these differences have occurred through transitions (A-G or T-C).

Nucleotide frequencies were also calculated for each form. The overall nucleotide frequency for both forms was A—0.28, T—0.24, C—0.35 and G—0.13 (GC content = 0.48).

A pairwise genetic distance matrix for Cytochrome b is shown in Additional file 3: S5. As can be seen from the summary in Table 1, the overall genetic distance within both forms is very low, while the genetic distance between each form is much higher with the maximum value between the forms (0.01838) being over 13 times greater than the highest value within either form (0.00139 in cyaneus). The minimum distance value between the two forms (0.0153) is also five times greater than the maximum distance value calculated within either form (0.00307 in cyaneus).

COI

The aligned COI sequences resulted in an alignment 555 nucleotides in length. Unlike the other datasets, all the cyaneus sequences were identical and there was one transversion in hudsonius, with one sample having a T to G substitution. There were seven positions in the alignment where cyaneus sequences showed constant differences to hudsonius sequences of which six were transitions and one a transversion.

Nucleotide frequencies were also calculated for each form. The overall average nucleotide frequency was A—0.26, T—0.27, C—0.31 and G—0.16 (GC content = 0.47).

A pairwise genetic distance matrix is shown in Additional file 4: S7 and summarized in Table 1. As found in the other datasets, the overall genetic distance within both forms is very low, while the genetic distance between each form is much higher with the maximum value between the forms (0.01292) being 14 times greater than the highest value within either form (0.00093 in hudsonius).

ND1

For the ND1 sequences, all the hudsonius ND1 sequences were identical, with no substitutions across the complete 1223 nucleotide stretch of sequences. The two cyaneus sequences, on the other hand, showed a slight degree of variation. At one position, a non-synonymous A to G transition can be found which results in an Isoleucine to Valine amino acid change along with another synonymous transition of a C to T.

One indel was identified between hudsonius and cyaneus, giving a gap in hudsonius and a “T” in cyaneus, although this was outside the ND1 coding region. SNPS were identified at a further 21 locations, of which 20 were transitions and one was a transversion. Two of the transitions within the coding regions resulted in non-synonymous changes.

Nucleotide frequencies were also calculated for each form. The overall nucleotide frequency was A—0.29, T—0.26, C—0.32 and G—0.13 (GC content = 0.45).

A pairwise genetic distance matrix is shown in Additional file 5: S6 and summarized in Table 1. As seen in the distance matrix for other datasets, the overall genetic distance within both forms is very low, while the genetic distance between each form is much higher with the maximum value between the forms (0.01802) being 11 times greater than the highest value within either form (0.00164 in cyaneus).

Morphology

We carried out a range of measurements from museum specimens of both taxa and provide the mean of those measurements in Table 2. A complete list of all our museum measurement data and measurements from the literature can be found in the Additional file 6: S2 for our museum measurements and Additional file 7: S3 for measurements from the literature.

Table 2 Mean measurements (with SD) in mms of seven different features for 43 museum specimens

Males versus females—As in most raptor species, female harriers are significantly larger than males. The MANOVA test indicated that there is a significant size difference between the sexes in both cyaneus and hudsonius (averaged across taxa, p ≤ 0.001). Also, all of the mean measurements of male size are smaller than those for female size (Additional file 6: S2; Additional file 7: S3).

Cyaneus versus hudsonius—Our data show that with respect to the individual measurements, in all but one case both male and female hudsonius are consistently larger than their corresponding sex in cyaneus. The only exception to this is male tail length, where cyaneus shows a 0.2 cm longer tail. Although this may represent a real difference between the two forms it should be noted that tail length is notoriously difficult to measure accurately and consistently as the tail feathers disappear into the rump of the birds, and finding the actual base of the tail can be difficult. The variation in tail length measurements reported by other authors also seems to echo this inconsistency (Additional file 7: S3). The MANOVA test suggests that the mean of the seven individual size measurements averaged across both sexes is significantly different between cyaneus and hudsonius (p = 0.017). Two-way ANOVA test also suggests a significant difference between only males of the two forms (p = 0.067, p = 0.037 when tail length not included) and only females of the two forms (p = 0.012).

Plumage

Circus cyaneus obtains full adult plumage in its third calendar year, going through various plumages before reaching maturity (Fig. 2).

Fig. 2

The range of plumages found in hudsonius and cyaneus. a hudsonius male, b, c cyaneus male, d hudsonius female, e cyaneus female, f hudsonius juvenile and g cyaneus juvenile. A full account of plumage characteristics is discussed in the text. (Photo credits: a—Simon Richards, b—Hiyashi Haka, c—Dirk-Jan Hoek, d, f—Julian Hough, e—Matti Suopajärvi and g—Peter Blanchard)

The differences between each sex and age group are as follows:

Adult males

Adult male hudsonius (Fig. 2a) differs from cyaneus (Fig. 2b, c) in being a much darker and more streaked bird overall. It has a dark “saddle” across the back, darker wing-coverts and more variably streaked under-parts, including the under-wing. Adult male hudsonius also has five black outer primaries (p6‒10), p7‒10 having roughly half of the primary base coloured white on the dorsal surface, and with only the tip of p6 being black. Adult male cyaneus has six black outer primaries, with almost the entire vane of p6‒10 being black, with a notably longer black tip to p5. Adult male cyaneus also show an even black trailing edge on all of the inner primaries and secondaries, whereas hudsonius has less well-marked blackish sub-terminal spots on the inner five primaries, but broader black tips to the secondaries. Adult male hudsonius also exhibits black barring on the secondaries and rufous barring on the under-wing coverts and axillaries, while cyaneus typically shows only white on the under-wing coverts and axillaries, with very little streaking or barring on the flight feathers or under-wing coverts. The dark appearance of the upper-wing of hudsonius is created by the combined effect of the dark tips on the greater and median wing coverts (including the alula), almost completely dark lesser coverts, dark central streaking on the primary coverts and the “background” grey of the wing being of a darker shade than in cyaneus. Furthermore, the grey secondaries and primaries are surrounded by darker feathers (primaries, trailing edge, and coverts), which forms a grey “window” in the flight feathers of hudsonius. The scapular and mantle feathers of hudsonius are all very dark, lending a dark “saddled” effect to the back. In comparison, adult male cyaneus has a consistently pale grey back and upper wing, with contrasting black primaries. The back of cyaneus sometimes appears darker, but never to the degree that it does in hudsonius, and may very well be just a characteristic which varies among younger adults. In cyaneus the dark tips to the secondaries appear as a dark grey band along the length of the trailing edge of the open wing and just a few dark-shafted coverts create the only hint of streaking.

The tail of adult male hudsonius is also very different from that of cyaneus. It is a shade of grey lighter than that of the secondaries, with a thick black sub-terminal band and between four and six thinner, but distinct bars across all but the much paler outer rectrices, where they appear to be fainter and thinner. Also, the sixth (basal) tail bar is often obscured by the upper-tail coverts. Adult male cyaneus has an evenly pale-grey tail, usually lacking any noticeable tail bars and only the faintest of any other dark markings.

The under-parts of hudsonius and cyaneus also differ considerably. Adult male cyaneus has a white vent, belly, and lower breast, with a smoky-grey upper breast, throat and chin which extends up onto the side of the head and is very much consistent with the overall greyness of the bird. Adult male hudsonius, on the other hand, has a completely white background to the under-parts, with extensive rufous streaking, barring, and spotting. Generally, the thighs, flanks, and under-tail coverts are spotted, the side of the breast is barred, and the central region of the breast, throat and chin have long, vermiculated streaking that creates a rufous-grey breast band on more heavily marked individuals. Furthermore, whereas cyaneus has an unmarked grey head, hudsonius has a finely streaked head and often shows a whitish forecrown, supercilium and lower-eye crescent.

Sub-adult males

Male hudsonius acquire their first adult-type plumage through a protracted moult from April to October during the second calendar year (Wheeler 2003), whilst male cyaneus has a similarly timed second-year moult that occurs between May and October (Forsman 1999). Sub-adult male hudsonius are much darker and more strongly patterned than fully mature adult males, often retaining signs of their juvenile plumage around the head, neck and breast. They also show extensive streaking on the breast and throat, with the thighs, vent, axillaries and under-wing-coverts being heavily spotted with rufous, although by this age they do show the adult-type distribution of black on the primaries. While rufous spotting and streaking may be present in sub-adult cyaneus, it is restricted to the breast and upper belly, and when present, it tends to be very light on the thighs and vent. Even by this age, the smoky-grey upper breast and throat characteristic of adult male cyaneus is present and the streaking does not extend far onto this area. The mantle feathers of cyaneus at this age are also browner, as opposed to the dark grey mantle of adult hudsonius.

Another notable difference in the plumage of sub-adult males is in the tail markings. In cyaneus, the only complete tail bar is also the most terminal and is usually quite subtle. The outer rectrices of cyaneus have thin, faint barring while the remainder of the tail feathers exhibit barring that does not reach the outer edges of the feather and is restricted mainly to the inner webs of the feathers. As in the adults, sub-adult hudsonius have up to seven tail bars (including sub-terminal) that are all quite thick and extend the whole width of the feather.

Juveniles

Typically-plumaged juvenile hudsonius (Fig. 2f) has deep chestnut-orange under-parts with little or no streaking, from the throat to the undertail-coverts. It has a dark hooded appearance, created by dark ear-coverts and dark sides to the neck forming a solid “boa” around the bird’s neck. This coloration makes it appear quite similar to juvenile Pallid Harrier (Circus macrourus), a Palearctic species. Juvenile hudsonius also has dark upper-parts that contrast markedly with their white rumps. Juvenile cyaneus (Fig. 2g) typically have a buff background to their under-parts and heavy streaking across the breast, belly and flanks. Although cyaneus also has dark ear-coverts, it rarely shows the dark “boa” of hudsonius and those cyaneus individuals that do exhibit a dark “boa” are usually at the more extreme-dark end of the plumage spectrum and also show heavily marked under-parts (Mullarney and Forsman 2011). Juvenile cyaneus also have paler upper-parts than hudsonius, rarely reaching the dark sepia-brown that is so characteristic of the latter.

While juvenile hudsonius can be distinguished from juvenile cyaneus by a range of characters, there are extremely rare occasions where they may appear quite similar overall (Thorpe 1988). Although the amount of streaking present in cyaneus is rarely as sparse as that of hudsonius, where it is restricted to the flanks and upper breast, the most heavily streaked hudsonius individuals may overlap with some of the least-streaked cyaneus individuals (Mullarney and Forsman 2011).

With respect to the under-wing of hudsonius, p6‒8 usually show five or six strong, yet thin, bars plus a dark tip. In cyaneus, there are usually only three or four such bars present on these same feathers, and they are thicker in comparison. Also p10 (the shortest, outer primary) in cyaneus has three bars, whereas hudsonius typically has four bars, but may show only three on occasion (Mullarney and Forsman 2011). Juveniles of both forms exhibit a series of three bars across the under-wing on the secondaries, with the terminal bar (nearest the tip), being the thickest. In hudsonius, the central bar is usually noticeably thinner than that of cyaneus, although there is substantial variation and overlap in this particular character.

The under-wing pattern of juvenile hudsonius is generally very much like that of adult females, but with a buff, not white, background to the flight feathers. The under-wing coverts of hudsonius juveniles are streaked, rather than spotted as in females, and the background coloring is fairly rufous. The under-parts of juvenile cyaneus are usually streaked and look very similar to adult females, although with comparatively warmer tones to the under-parts and under-wing. Juveniles of both forms can be sexed according to iris color with females having completely dark eyes and males having a light grey-green iris.

Adult females

Adult female hudsonius (Fig. 2d) and cyaneus (Fig. 2e) superficially resemble juvenile birds in that they have dark brown upper-parts, white rumps and similarly marked under-wings. The under-wing markings previously noted for juveniles also apply generally to females of each form. In adult female hudsonius the breast, belly, vent and flanks are light buff, lacking the orange tones characteristic of juveniles, and have strong streaking over the entire under-parts. Female hudsonius usually show diamond shaped markings on their flanks, whereas cyaneus usually shows streaking in this area, although some of the cyaneus museum specimens examined also had diamond shaped markings on their thighs. Furthermore, in the field the bars on the under-wing of hudsonius are easier to see than in cyaneus, due to the paler background color of the secondaries in hudsonius.

Discussion

We have shown how the two forms of Circus cyaneus—C. c. cyaneus and C. c. hudsonius vary in a number of morphological and genetic characters and we discuss this, along with other work further.

Mitochondrial and nuclear phylogeny

A complete phylogenetic analysis has been carried out using three mitochondrial loci which consistently show that the two forms represent divergent linages.

Previous work (Oatley et al. 2015) was carried out using a total of 2032 bp from three nuclear genes [Myoglobin intron-2 (MB), Beta Fibrinogen intron-5 (FGB) and TGFβ2 intron-5 (TGFB2)] as well as the mitochondrial ND1 gene from 39 Accipitriformes, which included 16 different Circus taxa. They also showed cyaneus and hudsonius to be separate species, with Circus cinereus (Cinereous Harrier) being a sister species to hudsonius and then cyaneus sister to those two species. This work subsequently led to the recognition of hudsonius as a full species by some authorities (Gill and Donsker 2015; Sangster et al. 2016).

To demonstrate further the substantial degree of divergence between the two forms, examples of differences in overall appearance, behavior and life history are provided. Differences in habitat, dispersal, mate choice, nesting site and other ecological characters are numerous.

The most notable morphological variation between cyaneus and hudsonius may be observed in the consistent differences in their respective plumages. These differences were also compared between each sex and age group. Morphologically, individuals from each form are diagnosable by a number of qualitative differences. Adult males can be distinguished by approximately 13 different characters, females by about 4 characters and juveniles by at least three. hudsonius also averages slightly larger than cyaneus over a range of morphological characters. The differences in morphology (both body size and plumage details), vocalization and ecology between the two forms mirrors their existence in separate ecological niches. The expansion of hudsonius into North America would have meant that it underwent fundamental ecological changes in order to adapt to the new environment, climate and habitats.

The genetics of the two forms can also be examined from the molecular work. As can be seen from Table 1, the intra-taxa genetic distances are quite low, but the inter-taxa genetic distances are relatively high. The intra-taxa genetic distances are generally higher in cyaneus than in the more genetically uniform hudsonius. The general mutation rate for Cytochrome b in birds has been estimated at around 2 % per million years (Avise et al. 1987; Weir and Schluter 2008) but other estimates suggest as little as 0.64 % (Pereira and Baker 2006). Using these two rates (0.64 and 2.0) as a minimum and maximum mutation rate, we can estimate a divergence time for our sequences. For Cytochrome b, using the maximum genetic distance between forms of 0.01838, a divergence time of between 0.765 and 2.39 mya is found. Using the minimum genetic distance between forms of 0.0153, a divergence time of between 0.915 and 2.87 mya is found. For ND1 (in which the min/max are the same), a divergence time off 0.9 and 2.815 mya is found, echoing the estimates calculated for Cytochrome b. Oatley et al. (2015) also examined divergence times of the harriers and estimated that the emergence of the Circus clade occurred between 2.7 and 6.6 mya during the expansion of the C4 grasses, followed by the diversification of the steppe harrier clade (in which cyaneus and hudsonius are placed) between 2.2 and 5.5 mya. These estimates overlap and agree well with our estimates of Circus cyaneus divergence times. These estimates have been examined further. Generation time and body size were found to be correlated with the rate of mitochondrial genome evolution and caused biases in molecular dating (Nabholz et al. 2016). The authors re-examined the study by Oatley et al. (2015) and calibrated the emergence of the Circus clade to between 11 and 13.1 mya, with the subsequent splitting of the cyaneus/hudsonius complex of 2.1–2.5 mya. The authors conclude that although their estimates are much older, the dates still largely agree with the appearance of C4 grasses during the mid-Miocene.

Much avian phylogenetic work in the Accipitriformes has been carried out using mitochondrial DNA sequences, of which the resulting genetic distances between well-defined species are comparable to those found in our study (approaching 2 %). In a study of Old and New World Vultures (Wink 1995), species within the Torgos and Gyps genera typically showed genetic divergence of only 2 % with Griffon Vulture (Gyps fulvus) and Cape Vulture (G. coprotheres) having a genetic divergence as low as 0.9 %. In a study of the Old World Buzzards (Buteo) (Kruckenhauser et al. 2004), genetic distances between undisputed species ranged between 1.0 and 1.6 %. In a mitochondrial phylogeny of Sea Eagles (Haliaeetus) (Wink et al. 1996) genetic distances between seven species of sea eagles varied between 0.3 and 9.8 %. Also, a Cytochrome b sequence divergence of only 1.75 % was found between Greater Spotted Eagle (Aquila clanga) and Lesser Spotted Eagle (A. pomarina) (Helbig et al. 2005).

There are also numerous differences between cyaneus and hudsonius when it comes to vocalization, habitat, distribution and movements, mate choice and breeding biology.

Vocalization

Comparisons were made between single-samples of previously published spectrographs of hudsonius and cyaneus (Cramp and Simmons 1980; MacWhirter and Bildstein 1996). Male harriers sometimes give a distress call when attacking a potential predator. In cyaneus eleven “keks” are emitted per second, each of which starts at 2 kHz and finishes at less than 6 kHz (Cramp and Simmons 1980). In hudsonius six “kek” calls are emitted per second, starting at 0 kHz and finishing at 6 kHz (MacWhirter and Bildstein 1996).

Female harriers often give a distress call when they are approached at the nest by a potential predator. In cyaneus the call consists of eight “keks” per second at a frequency between 2 and 5 kHz (Cramp and Simmons 1980), while hudsonius emits six “keks” per second at a frequency between 0 and 6 kHz (MacWhirter and Bildstein 1996).

Habitat

hudsonius

As its old common name “Marsh Hawk” suggests, hudsonius prefers marshes, fresh and brackish wetlands, and damp meadows with undisturbed vegetation during the breeding season, especially in northeast and Midwest regions of North America. Upland prairies, dry grasslands, agricultural areas, and riparian woodlands up to little more than 2400 m above sea level are also used, but mainly in western North America, while dense forest habitats are avoided. The winter range can be more variable, with birds tending to frequent most open habitats, especially in lowlands (Apfelbaum and Seelbach 1983; MacWhirter and Bildstein 1996).

cyaneus

In Britain and Ireland, cyaneus breeds almost exclusively on moorland and in young coniferous forests. In Holland, sand dunes are occupied as breeding habitat, whereas in Scandinavia, they nest both on high conifer plateaus and around lowland sedge-fringed lakes. In winter, as in most Circus species, cyaneus may be found frequenting lowland plains and marshes (Simmons et al. 1987; Simmons 1988; Etheridge and Summers 2006).

Distribution, dispersal and migration

The hudsonius population breeds south of a line that runs roughly from northern Alaska, down along the southern shore of the Hudson Bay and into southern Quebec and the Maritime Provinces, and north of a line running from central California, through northern Texas, up to the Great Lakes and then south-east to New Jersey. It winters throughout much of the lower 48 United States, south through Central America, various Caribbean islands and into South America, but usually no further south than Colombia and Venezuela.

The cyaneus population breeds across Eurasia, south in Europe to Portugal and north to Finland, then east across Asia to the Kamchatka Peninsula in the north and south to eastern China. In Asia, some individuals winter as far south as Iran and northwest Pakistan, across to Indo-China and possibly the northern Philippines (Ferguson-Lees and Christie 2001). In autumn, there is a usually a short southerly migration, starting in late September. Females, and maybe non-breeding birds, that are dispersing earlier from the breeding territory, mainly undertake these short migrations. These birds generally appear in Ireland and northwest Europe (Belgium, France, etc.) by late October, while males do not arrive until mid-November. The northern most regions of the breeding range are almost completely deserted by males, which often complete a south-westerly migration in severe weather, while females can endure complete snow cover due to their ability to capture larger prey items.

Breeding behaviour

Mate choice

Both males and females engage in a display flight during the breeding period, termed “sky-dancing”, the frequency of which is correlated with food availability (Simmons 1991). It has been noted that in Orkney, Scotland, female cyaneus displayed up to five times as often as males, but in New Brunswick (Canada) hudsonius males displayed 12 times as frequently as females (Simmons et al. 1987). Although the female:male sex ratio in Orkney (3:1) was more greatly skewed than the sex ratio of the population in New Brunswick (3:2), it is unknown if this imbalance accounts for the gender and frequency differences in display behavior or if it is reflection of some divergence in mating behavior since the two forms became fundamentally allopatric (Simmons et al. 1987).

Nest site

cyaneus

There is a preference in cyaneus to place the nest in heather that is taller than the surrounding vegetation and a further preference for vegetation that is 40‒50 cm high. In a study of 52 Scottish cyaneus nests, it was found that cyaneus showed a clear preference for nesting in heather (Calluna vulgaris) (Redpath et al. 1998). The average length of heather used to nest in was not that of the average length of heather. Nearly 50 % of birds nested in heather between 40 and 50 cm high, with 40 % of birds evenly occupying heather 30‒40 and 50‒60 cm high. It was also found in a separate study that out of 922 nests, 76 % were located in heather moorland. Other nest sites included upland grassland and open canopy and closed canopy woodlands (Redpath et al. 1998).

hudsonius

hudsonius occupies a wide range of habitats. In a survey of 428 nests, 17 % were located in wet sedge meadows, 18 % were in freshwater reed marshes, 26 % were located in dry grasslands and 8 % were in agricultural fields (Apfelbaum and Seelbach 1983). The nest is usually constructed in standing water, on floating platforms that are raised above the water level, or in tall vegetation (e.g. reeds, cattails, etc.).

Male desertion

Male harriers generally supply food to the female who feeds the young herself, although when the young are old enough to feed themselves, males may provision them at the nest directly. After this initial period, by the time the chicks are about three weeks old, male investment in feeding the young often declines. In some instances food provisioning was found to cease, with the male deserting the nest completely (Simmons 2000). In one study of hudsonius, females at 7 out of 11 nests that were deserted by the male managed to fledge their young alone, while the nestlings in the other 4 nests died of starvation. Contrastingly, cyaneus females can rarely raise young without help from their mate (Simmons et al. 1987).

Conclusions

Circus cyaneus and C. hudsonius fulfil all the criteria set out by the BOU for assigning species rank to allopatric taxa (Helbig et al. 2002). They are fully diagnosable in several characters (i.e. 13, 4 and 3 characters for male, female and juvenile respectively) and have different DNA sequences. Should the two taxa ever become sympatric and create contact zones in the future, morphology, vocalization, breeding biology, and habitat should clearly act as prezygotic barriers.

Our finding of a genetic distance between the forms hudsonius and cyaneus of up to 2 % falls well within the range of other well-established species and taking into account other morphological and ecological differences found between the two forms, we suggest that cyaneus and hudsonius represent distinct evolutionary lineages and should be treated as separate species. We recommend the scientific name for the (currently nominate) Eurasian species remain as Circus cyaneus retaining the common name of Hen Harrier, and that the American form is given the scientific name of Circus hudsonius with the common name of Northern Harrier.