Taxonomic research update Greenish Warbler, 'Two-barred Greenish Warbler', and the speciation process Martin Collinson

ABSTRACT This Taxonomic Research Update draws birdwatchers' attention to the work of Irwin et at. (2001) on Greenish Warblers Phylloscopus trochiloides..Their research shows that two , P. t. viridanus and P. t. plumbeitarsus, both of which occur as vagrants in western , do not interbreed in the region where their ranges overlap because they do not recognise each other's songs.The subspecies of Greenish Warbler are perhaps the best example of an avian 'ring ', and in this case sexual selection for increased complexity of song is driving the speciation process.

Introduction The Greenish Warbler Phyllo­ scopus trochiloides is a poly­ typic species that breeds across much of temperate , and as far west as eastern Europe. Six subspecies are recognised (Ticehurst 1938), their breeding distributions being shown in fig. 1. Greenish Warblers of the subspecies viridanus are regular vagrants to , although P.t.plumbeitarsus ('Two-barred Greenish Warbler) and P. t. nitidus ('') have Fig. 1. Distribution of the races of Greenish Warbler Phylloscopus also occurred there (Dean trochiloides. Racial variation is clinai through the intergradation of 1985; see also pages 284-288). the subspecies viridanus (eastern Europe to western Asia and: The Greenish Warbler Afghanistan), ludlowi (southeast Afghanistan to Kumaon),:. trochiloides (central and eastern ), obscuratus (central complex is one of the few iChina) and plumbeitarsus (eastern Russia); the dotted lines indicate putative avian examples of a a gap in the distribution due to deforestation.The 'Green Warbler' ''. Ring species, P.t. nitidus is fully allopatric, and is not considered further.All other most influentially described by subspecies interbreed with adjacent subspecies with the exception Mayr (1963), are deemed to be of viridanus and plumbeitarsus, which show no apparent intergradation in the zone where their distributions overlap a possible result of one species (indicated by *). 'Two-barred Greenish Warbler' P. t. plumbeitarsus expanding its range in a ring­ was regarded as a separate species, P. plumbeitarsus, by Cramp like manner around an area of (1992). unsuitable (see fig. 2).

278 British 94:278-283, June 2001 Collinson: Greenish Warbler, 'Two-barred Greenish Warbler, and the speciation process

of a ring species are not fully understood, and the validity of some of the classic exam­ ples, such as the Herring Gull Larus argen- tatus/Lesser Black-backed Gull L. fuscus circumpolar ring, and the Eurasian ring of Great Tits Parus major/Grey Tits P. afer, has been seriously challenged by new taxonomic data. Similarly, although Greenish Warblers of the subspecies viridanus and plumbeitarsus have been reported not to interbreed (Cramp 1992), there has been considerable ambiguity concerning the rela­ tionship between the two, no thorough analysis of which had been undertaken until recently, with the results published in the journal Nature (Irwin et al. 2001).

Recent research Many of the fundamental gaps in our knowl­ edge about the subspecies of Greenish Warbler have been addressed in the Nature paper mentioned above. Irwin et al. used behavioural, morphological and genetic data to show that viridanus and plumbeitarsus apparently do not recognise each other as conspecifics, and that there is no detectable between the two.They are linked, however, around the southern edge of the Fig. 2. Hypothetical formation of a ring species. by a ring of intergrading, Panel A represents the ancestral distribution of a intermediate subspecies. Greenish Warblers species, shown in green. In B, the species' range hence fulfil many of the critical require­ starts to expand, perhaps as a result of climatic 1 amelioration, around the sides of an area of ments of a 'typical' ring species. unsuitable habitat (in this case, a mountain In the study, Greenish Warblers of all sub­ range): as expansion proceeds, populations on species were observed at 29 sites through­ each side of the mountain range evolve genetic out the entire range. Various biometric and and morphological differences (represented by: blue and yellow), either by chance or because plumage characters were recorded, sono­ ; the selection pressures on each one are ; grams were constructed and analysed, the different; nevertheless, there is limited gene flow responses to songs of the same and of dif­ and intergradation between adjacent populations ferent subspecies were measured, and blood (represented by double-headed arrows). In C, the opposite ends of the ring meet on the far side of samples were taken for genetic analysis. the area of unsuitable habitat: the races of the Northern races of Greenish Warbler, viri­ ring species at these terminal points of the chain danus and plumbeitarsus, have longer and have diverged to such an extent that they are reproductively isolated, i.e. they do not : more complex songs than the nominate interbreed, and thus behave as separate species;: southern subspecies trochiloides. The gene flow is maintained between all other races authors demonstrated parallel gradients of in the ring. increasing song length and complexity on both the eastern and the western sides of Under certain conditions, a chain of racial the Tibetan plateau. By breaking down the variants may arise, each race able to breed songs into their principal components, they and intergrade with its neighbours, except were able to show that there are smooth for those co-existing at the two overlapping clines in song structure from trochiloides to 'ends' of the ring, which have diverged so viridanus on the west side of the ring, and much that they behave as different species. from trochiloides to plumbeitarsus on the The conditions that lead to the formation east side. It is presumed that, during the

British Birds 94:278-283, June 2001 279 Collinson: Greenish Warbler, 'Two-barred Greenish Warbler', and the speciation process Hugh Harrop Hugh Harrop

162. Greenish Warbler Phylloscopus 163. Greenish Warbler Phylloscopus trochiloides of the subspecies viriddnus, trochiloides of the subspecies viridanus, Quendale, Shetland, September 1994. Sumburgh, Shetland, August 1996.

species' northward range expansion fol­ blers of each subspecies to playback of the lowing glaciation, sexual selection encour­ songs of all other subspecies, and measured aged this increasing complexity, favouring the strength of their response. They found males with more elaborate songs. Although that, in general, Greenish Warblers reacted there have been parallel pressures to evolve strongly to the songs of their own or neigh­ longer, more complex songs, however, the bouring populations or subspecies, but less songs have developed differently on each strongly to those of more distant popula­ side of the Tibetan plateau: viridanus have tions. For example, trochiloides recognised songs of high frequency range, composed of ludlowi and obscuratus as being from the three or four long song units, while same species, but showed a weaker response plumbeitarsus have songs with a lower fre­ to viridanus songs; ludlowi showed a strong quency range, composed of seven or eight response to many viridanus and shorter units. This divergence, although trochiloides, but a weaker response to smooth and gradual, has, therefore, led to the obscuratus; and so on. In general, individuals manifestation of noticeably different songs in reacted aggressively to the songs of sub­ viridanus and plumbeitarsus. species which occurred within approxi­ What is perhaps more interesting is that mately 1,500 km of their own territory, but the differences in song structure that we less strongly or not at all to songs of those hear are of biological significance to the beyond that range, presumably because the birds themselves. Typically, male latter's songs were too different from their birds exhibit a strong territorial response to own. One notable exception was the reac­ playback of songs which, although not iden­ tion of viridanus to plumbeitarsus. Individ­ tical to their own, they recognise as coming uals of these two subspecies within their from a competing male of the same species. zone of overlap showed no response to each If they do not recognise the song as being other's songs, and, indeed, sometimes shared that of a competitor, for a territory or a mate, territories. The clear, measurable differences they do not react. in their songs, as a result of their being at the opposite ends of the ring of subspecies varia- Irwin et al. subjected male Greenish War­

280 British Birds 94:278-283, June 2001 Collinson: Greenish Warbler, 'Two-barred Greenish Warbler', and the speciation process

tion, are therefore also registered by the eastern and western genetic lineages falls birds themselves. They behave as different within the range of one of the subspecies, species. ludlowi. Around this boundary, there are no To determine whether this incipient spe­ morphological or song differences between ciation of the Greenish Warblers was warblers carrying the eastern-clade DNA and reflected in their genetic history, a phyloge- those with the western-clade DNA structure, netic tree was drawn up.This is based on an and the playback experiments confirm that analysis of 1,200 base pairs of mitochondrial the birds do not distinguish between individ­ DNA (mtDNA) from 149 individuals of all uals carrying the different mtDNA lineages. races, taken from sites throughout the entire While direct gene flow between viridanus range of the species. These data are sum­ and plumbeitarsus in Siberia does not occur, marised in a simplified form in fig. 3. This it would seemingly be possible for gene gene tree demonstrates a deep genetic split exchange to take place 'via the back door' between western and eastern birds. In the (i.e. through the chain of intermediate sub­ zone of overlap of viridanus and species encircling the southern edge of the plumbeitarsus in Siberia, genetic differences Tibetan plateau). Nevertheless, gene flow were fully in accordance with differences in around this ring is apparently not great physical appearance: i.e. all viridanus had enough to disrupt the genetic differentiation the 'western-clade' DNA, and all of the Greenish Warbler populations; this is plumbeitarsus had the 'eastern-clade' DNA. presumably a classic case of 'isolation by dis­ There is therefore no evidence of gene flow tance'. between the populations of viridanus and The trends in other genetic (microsatel- plumbeitarsus in Siberia, a finding which lite) and morphological (width of the supports the behavioural evidence derived greater-covert wing-bar) characters were fol­ from the song-playback experiments. lowed for individuals in the ring of sub­ Interestingly, the boundary between the species from viridanus, through ludlowi, trochiloides and obscuratus, to plumbeitarsus. Again, an apparent lack of gene flow in the Siberian overlap zone was demonstrated, but there was significant con­ tinuing or recent gene flow between adja­ cent subspecies along the southern side of the ring.

Discussion The classic explanation of the Greenish Warbler as constituting a ring species would postulate that the ancestral populations inhabited the forested sub-Himalayan region. During the Ice Ages, much of Siberia would

Fig. 3. A gene tree, derived from analysis of mitochondrial DNA, showing the deep genetic split between eastern and western subspecies of Greenish Warbler Phylloscopus trochiloides. For clarity, the shaded boxes represent clusters of genetically similar individuals. A western-clade DNA (») contains all the individuals of viridanus and nitidus and a proportion of the ludlowi populations, whereas an eastern-clade DNA (§) is representative of other ludlowi individuals, as well as all trochiloides, obscuratus and plumbeitarsus. Individuals labelled ludlowi/trochiloides were taken from the ludlowi end of the intergrade zone between those two races. See text for discussion.

British Birds 94:278-283, June 2001 281

British Birds 94:260-267, June20 2001 261 Collinson: Greenish Warbler,'Two-barred Greenish Warbler', and the s/K-ciation process

Tibetan plateau (where the birds recognised each other as potential mates and inter­ bred), and one to the north (where they did not recognise each other). There are some controver­ sial differences between the two models outlined here, but birdwatchers need not be too bothered about them.The con­ ditions that lead to the forma­ tion of a ring species may well, in practice, involve periods of range-splitting and subsequent divergence in allopatry. If we think of the of the subspecies of Greenish Warbler in terms of the classic ring species, based on expansion of an ancestral Himalayan population, but David Tipling/Wmdmsh then superimpose the possi­ 164. Greenish Warbler Phylloscopus trochiloides of the subspecies bility that genetic divergence plumbeitarsus ('Two-barred Greenish Warbler'), Happy Island, has been maintained, or at , May 1994. least helped along, by periods have been treeless; as the glaciers retreated of range contraction and splitting during and lowland temperate Asia became refor­ glacial periods, followed by secondary con­ ested, Greenish Warblers expanded their tacts between the races during range expan­ range northwards, skirting the edges of the sion, we are probably not too far away from a still treeless Tibetan plateau. The genetic split conceptual basis for understanding what has in mtDNA lineages between eastern and happened. western birds may have occurred by chance, Males of plumbeitarsus and viridanus do as a result of random genetic drift. Sexual not recognise each other as conspecifics; selection caused significant divergence of there are no intermediate individuals in the both the songs and the morphology of popu­ Siberian overlap zone; and, while direct gene lations on the east and west sides of Tibet in flow between the two would be expected to a geographically smooth manner, such that, occur if females of one subspecies were when the eastern and western birds came mating with males of the other, none has back into contact in Siberia, they no longer been detected. Furthermore, although it has 'recognised each other and did not inter­ been possible until recently for gene flow to breed. occur indirectly, via the ring of intermediate The data cannot, however, exclude a subspecies, this ring has now been broken second possibility, namely that, during the by deforestation in China (fig. 1). There Ice Ages, there were two genetically diver­ seems little chance that the separate evolu­ gent, allopatric subspecies of Greenish tionary lines taken by viridanus and Warbler: a western subspecies near the plumbeitarsus will ever again merge into Himalayas; and an eastern subspecies, with one. Perhaps those birders who have seen double wing-bars, presumably somewhere in plumbeitarsus, or 'Two-barred Greenish southeast Asia. These two subspecies Warbler', are justified in following the line expanded and met in the postglacial period, taken by BWP, giving this separate each one evolving more complex songs as it specific status. Were such a split to be moved northwards. Two new zones of adopted, however, some or all of the contact were formed: one to the south of the southern subspecies might conceivably be

282 British Birds 94:278-283, June 2001 Collinson: Greenish Warbler, 'Two-barred Greenish Warbler', and the speciation process left in limbo, since it would be impossible to assign them to either Greenish Warbler or Two-barred Greenish Warbler. It is often assumed that a complete understanding of the stages by which specia­ tion occurs is not possible, because no single study has demonstrated the evolution of a new species through all the 'grey-zone' intermediates (as shown diagrammatically in Collinson 2001). The process by which one species may, over time, split into two increasingly divergent popula­ tions which eventually become species, fully isolated reproductively, is illustrated neatly, however, by ring species. All the intermediate stages are extant and repre­ David Tipling/Windrush sented in geographical space. 165. Greenish Warbler Phylloscopus trochiloides of the subspecies Examples of ring species such plumbeitarsus ('Two-barred Greenish Warbler'), Happy Island, as the Greenish Warbler are China, May 1993. perhaps, therefore, the most graphic argument to counter the views of those people who continue to believe that Darwinian evolutionary theory cannot explain how one species becomes two.

References Collinson, M. 2001. Shifting sands: taxonoraic changes in the world of the field ornithologist. Brit. Birds 94: 2-27. Cramp, S. (ed.) 1992. The Birds of the Western Palearctic. vol. 6. Oxford. Dean, A. R. 1985. Review of British status and identification of Greenish Warbler. Brit. Birds 78: 437-451. Irwin, D. E., Bensch, S., & Price.T. D. 2001. Speciation in a ring. Nature 409:333-337. Mayr, E. 1963- Species and Evolution. Cambridge, Mass.

Ticehurst, C. B. 1938. A Systematic Colin Bradshaw Review of the Genus Phyllos- copus. New York. 166. Greenish Warbler Phylloscopus trochiloides of the subspecies plumbeitarsus ('Two-barred Greenish Warbler'), China, spring.

Martin Collinson, 22 Tippet Knowes Park, Winchburgh EH52 6UJ; e-mail: Martin. Collinson @ed. ac. uk

British Birds 94:278-283, June 2001 283