Dunnock Prunella modularis is a common songbird species that is endemic to the Western Palearctic. Some north European populations migrate short distances to winter in southern Europe, whereas populations breeding in the latter region are largely sedentary (del Hoyo & Collar 2016). No fewer than 15 subspecies have been described, with several having ranges largely circumscribed by peninsular (e.g., Iberia), insular (British Isles) or montane regions (e.g., Caucasus). Subspecific descriptions have been based on plumage differences, and in a few cases qualitative assessments of mensural characters; assessments of vocalisations have been restricted to P. m. occidentalis from England (e.g., Cramp 1988). Since their formal description, many Dunnock taxa have been subsumed into other races (Table 1), and recent treatments vary in recognising as many as eight (Cramp 1988, del Hoyo & Collar 2016, Gill et al. 2020) to just three subspecies (Dickinson & Christidis 2014, Shirihai & Svensson 2018). None of these decisions was based on genetic data.

In our recent phylogeographic study of the Dunnock (Drovetski et al. 2018a), sampling included individuals from the distributions of P. m. modularis (most of central and northern Europe), P. m. obscura Hablizl, 1783 (north-east Turkey, Caucasus, north Iran), P. m. occidentalis E. Hartert, 1910 (Britain and west France), P. m. mabbotti Harper, 1919 (French Pyrenees, Iberia, and possibly the Italian Apennines and Greece; recent treatments are in conflict) and P. m. meinertzhageni Harrison & Pateff, 1937 (former Yugoslavia and Bulgaria).

The results of the Drovetski et al. (2018a) study placed Dunnocks into one of three highly supported clades, each distributed in different Pleistocene refugia: a Caucasus clade, an Iberian clade, and a clade comprising all other individuals, including from the Balkans. Gene-flow analysis indicated that the Caucasus and Iberian clades were isolated from each other, as well as from the remaining clade, in which extensive gene flow was evident between Italy, the Balkans (Montenegro, Serbia and Greece) and broader Europe. Based on this genetic isolation, we suggested that Dunnock comprises three species (Drovetski et al. 2018a), which was supported by molecular species delimitation. Here we summarise in more detail the morphological and vocalisation data supporting the elevation of two Dunnock subspecies, corresponding to the Iberian and Caucasus clades described by Drovetski et al. (2018), to species status. In addition, we suggest that P. m. meinertzhageni of the former Yugoslavia and Bulgaria be synonymised with the nominate subspecies.

TABLE 1

The eight subspecies of Prunella modularis currently recognised by Gill et al. (2020) and their type localities, and their treatment under other modern classifications.

Material and Methods

Based on our genetic results (Drovetski et al. 2018), we compared specimens collected during the breeding season from the distributions of the three Dunnock lineages to their original descriptions (as subspecies) in order to verify the accuracy and validity of the proposed distinguishing characters. We analysed specimens held in the American Museum of Natural History, New York (AMNH), National Museum of Natural History, Smithsonian Institution, Washington DC (USNM), Biodiversity Research and Teaching Collection Texas A&M University, College Station, TX (TCWC), and Museo Civico di Storia Naturale di Carmagnola, Italy (MCCI). Specifically, we sought to assess if the plumage characters used in the original descriptions were in fact diagnostic. We analysed 50 specimens of P. m. modularis, 17 of P. m. mabbotti and 14 of P. m. obscura (Appendix 1). To visualise plumage variation among these taxa, we present photographs of the ventral, lateral and dorsal aspects of specimens of these lineages, including the holotype of P. m. mabbotti at USNM (Fig. 1). All photographed specimens were collected between late April and early June, and are therefore in breeding plumage. We also provide pictures of live individuals of the same taxa, photographed in the breeding season (Fig. 2).

Figure 1.

Specimens of the three Dunnock taxa considered in this work in dorsal, ventral and lateral views. A. Prunella [modularis] modularis, USNM 640862 (female; left) and USNM 640847 (male; right), Greece. B. P. [m.] mabbotti, first pair USNM 256648 (female; left) and USNM 256755 (type specimen, male; right), French Pyrenees; second pair USNM 317499 (female; left), and USNM 317500 (male; right), Spain. C. P. [m.] obscura, USNM 640349 (female; left) and USNM 640358 (male; right), Greater Caucasus (S. V. Drovetski)

Figure 2.

Individuals of the three Dunnock taxa considered herein. A. Prunella [modularis] modularis, UK (© L. Pulawski). B. P. [m.] modularis, Ireland (© B. Carruthers). C. P. [m.] modularis, UK (© K. Johnson). D. P. [m.] mabbotti, Spain (© D. Jauvin). E. P. [m.] mabbotti, Spain (© D. Petterson). F. P. [m.] mabbotti, Spain (© P. Alves). G. P. [m.] obscura, Armenia (© P. Adriaens). H. P. [m.] obscura, Georgia (© O. Nabrovenko). I. P. [m.] obscura, Iran (© C. Mroczko)

We also compared songs of the three lineages, by using the Xeno-canto database ( www.xenocanto.org) to select good-quality recordings made during the breeding season in the distributions of the three taxa. We analysed 125 phrases of P. m. modularis, 31 of P. m. mabbotti and 11 of P. m. obscura (Appendix 2). The sonograms (Figs. 3–4) and the analyses were prepared using the software Praat (Boersma & Weenink 2020). For each analysed phrase, we calculated the total length (in seconds) and the frequency range, including min. and max. (in kHz). We also evaluated whether rattles were present (Constantine & The Sound Approach 2006: 38–39) and, if so, assessed their length and frequency range. To assess if differences between taxa were significant, we calculated p-value and Cohen's d value (Table 2).

Results

Plumage characteristics.—Dunnock P. modularis sensu lato has a grey to brownish head with streaky ear-coverts, plain to mottled crown, a heavily streaked mantle, and grey to brown breast grading to dusky white on the central belly, with the body-sides streaked buff-brown. The three subspecies assessed here (central European P. m. modularis, Iberian P. m. mabbotti and Caucasian P. m. obscura) show clear plumage differences relative to each other (Figs. 1–2). In particular, P. m. mabbotti (Figs. 1B, 2D–F) can be distinguished from other taxa of P. modularis by plumage differences listed in the original description (Harper 1919). Specifically, it differs from P. m. modularis (Figs. 1A, 2A–C) by its greyer, less rufescent back, mantle and wings, with their feather edgings buffy brown and smoke grey instead of cinnamon-brown as in P. m. modularis (Harper 1919), and their centres are brown instead of black as in P. m. modularis. Ventrally, P. m. mabbotti is less grey than P. m. modularis but closer in appearance to the latter than either is to P. m. obscura (Fig. 1C). Furthermore, the head of P. m. mabbotti is not as grey as, and lacks the brown streaking on crown and cheeks of, P. m. modularis and the former's ear-coverts are not as brown as those of the latter, conveying the impression of a plain face in P. m. mabbotti.

Figure 3.

Sonograms of songs of Dunnock taxa considered in this work. A. Prunella [modularis] modularis, XC 392787 (J. Matusiak;  www.xeno-canto.org/392787), Slovakia. B. P. [m.] modularis, XC 132703 (J. Matusiak;  www.xeno-canto.org/132703), Poland. C. P. [m.] mabbotti, XC 342428 (A. Tomás;  www.xeno-canto.org/342428), Portugal. D. P. [m.] mabbotti, XC 471699 (J. Calvet;  www.xeno-canto.org/471699), Spain. E. P. [m.] obscura, XC 480650 (A. Lastukhin;  www.xeno-canto.org/480650), Azerbaijan. F. P. [m.] obscura, XC 139532 (F. Deroussen; w www.xeno-canto.org/139532), Georgia. Sonograms generated using Praat (Boersma & Weenink 2020).

P. m. obscura is distinguished from the other taxa of P. modularis primarily by the plumage characters reported in the original description (Hablizl 1783; see also del Hoyo & Collar 2016). These differences are particularly evident on the head: buff-tinged chin, brownish nape (not grey), light brown plain supercilium (vs. pure grey in P. m. modularis), and light brown and finely streaked crown. Further differences include back feathers with ill-defined chestnut streaks, and distinctly darker breast, compared to the other lineages, and throat to breast extensively marked with off-white or pale buff feather tips, forming a scaly pattern. Overall, P. m. obscura is much darker ventrally and more rufous-brown overall than the other two lineages, with less contrast between the grey and brown tones, a pattern approaching that of Japanese Accentor P. rubida.

Figure 4.

Sonograms of calls of Dunnock taxa considered in this work. A. Prunella [modularis] modularis, XC107150 (P. Åberg;  www.xeno-canto.org/107150), Sweden. B. P. [m.] mabbotti, XC335177 (C. Mroczko;  www.xeno-canto.org/335177), Spain. C. P. [m.] obscura, XC512728 (C. Mroczko;  www.xeno-canto.org/512728), Iran. Sonograms generated using the software Praat (Boersma & Weenink 2020).

Mensural data reported in the literature show great overlap between the various subspecies, with none available for breeding birds from Iberia and southern France (P. m. mabbotti) (Cramp 1988, Shirihai & Svensson 2018), thus we did not consider these parameters in our analysis.

Vocalisations.—The song of P. modularis sensu lato is a rather loud warble, usually 2.0–3.5 seconds long, and quite formless in that the notes lack a distinct structure (Fig. 3). The song of P. modularis sensu lato shows more prolonged units in the frequency range 4–6 kHz, and brief and sharply modulated units at 3–7 kHz with peaks up to 8 kHz. All units typically comprise different notes, apart from repetitions of a trill or rattle, usually once each phrase. Each male has repertoire of up to six (perhaps more) different songs, with each song usually different in all units from the others; modification to an individual's song does occur between years and during counter-singing (Cramp 1988). These factors combine to render comparisons across lineages difficult, and furthermore the availability of recordings of P. m. mabbotti and P. m. obscura song during the breeding season is reduced compared to those of P. m. modularis.

TABLE 2

Measurements of sound parameters in Prunella [modularis] modularis, P. [m.] mabbotti and P. [m.] obscura. Means and standard deviations are given. s = seconds; Cohen's d = effect size in terms of magnitude.

Despite these difficulties, our comparison of songs did reveal some variation that seems to permit discrimination between P. m. modularis, P. m. mabbotti and P. m. obscura. Analysis demonstrates that the songs of P. m. mabbotti are higher pitched (with peaks up to 9 kHz) with the rattle repeated more frequently within the phrase (Fig. 3C–D). Specifically, P. m. mabbotti had two rattles in 81% of examined phrases and occasionally a third at the start of the phrase (Table 2), whereas in P. m. modularis 64% of examined songs possessed just a single rattle. On the other hand, the song of P. m. obscura (Fig. 3E–F) is longer and lower pitched (barely exceeding 7 kHz) than both P. m. modularis and P. m. mabbotti, and lacks the typical rattles of the other two (with one exception where a faint rattle was present). Most differences we identified are statistically significant, but effect size is relatively low based on Cohen's d values (Table 2).

Dunnock also produces a wide variety of calls (Cramp 1988), which makes their comparison rather complicated. Similar types of calls across the three taxa show very slight differences, with P. m. obscura being of lower frequency (kHz) and repeated more frequently than the other two (Fig. 4). As such, calls seem to have little bearing for distinguishing taxa.

Species concept criteria.—Diagnosability is an obvious threshold for recognising species, and diagnosable characters vary across species. In a recent study that used both morphological and genetic data to assess avian diversity, Barrowclough et al. (2016) postulated that recognised biological species each harbour on average 1.97 phylogenetic species based on a criterion of morphological diagnosability. Their genetic estimates found an average of 2.4 phylogenetic species per biological species. In other words, each biological species appears to include on average two phylogenetic species.

Our previous genetic results (Drovetski et al. 2018a), which included a species delimitation method, and Bayesian and gene-flow analyses, clearly indicated that the three lineages are reproductively isolated with no gene flow. It is highly unlikely that adding more genes (we used 11) or samples (we included 13 for Iberia, 36 for the Caucasus and 66 for the remaining clade) would alter our interpretation of reproductive isolation (see, e.g., Felsenstein 2006). Based on reproductive isolation and independent evolutionary trajectories (Drovetski et al. 2018a), the obvious plumage differences identified by previous works (summarised above), and song differences documented here, we suggest that P. modularis comprises three species under the Phylogenetic Species Concept (Cracraft 1983). Considering that an independent evolutionary history is what ultimately identifies a species (see Zachos et al. 2013), we consider our genetic data sufficient to also recognise the three lineages as species in accord with both the Biological Species Concept (for which reproductive isolation is the paramount issue; Mayr 1963) and the Comprehensive Biological Species Concept (Johnson et al. 1999). Our proposal to recognise three species is also consistent with the criteria proposed by Helbig et al. (2002), as all are diagnosable by plumage, mitochondrial and nuclear DNA, with differences in song.

Offering further support is that two Dunnock lineages (P. m. modularis and P. m. mabbotti) are probably parapatric (we can find no records to indicate sympatry) in southwest France, and that P. m. obscura is isolated from P. m. modularis (and other described subspecies) in the Caucasus and adjacent regions. Finally, we suggest that the length of time since lineages are estimated to have diverged should not be a factor in recognising species; to suggest otherwise seems notional rather than evidence-based. For example, Motacilla species diverged more recently than Prunella species, but the former are often accepted as separate species (Drovetski et al. 2018b). Among the Prunella radiation, of the 14 recognised species, ten (including P. modularis) are estimated to have diverged from their closest relative within the last two million years, and the most recent divergence was just 0.13 million years ago (between Brown Accentor P. fulvescens and Kozlov's Accentor P. koslowi; Drovetski et al. 2013).

Taxonomic assessment

Considering the various aspects detailed above, we propose the following taxonomic arrangement for European Dunnock taxa.