Early View (EV): 1-EV

Bird introductions thus provide opportunities to study the cultural evolution of behavioural traits. Despite the rela-tive ease with which song recordings can be obtained from free-living individuals (in contrast to, e.g., DNA samples or morphometric data) and the fact that detailed information about introduction events is often available, so far only a few studies have examined changes in bird vocalisation after introduction (Jenkins and Baker 1984, Lang and Barlow 1997, Hamao 2015).

Birdsong serves multiple purposes, including defending a territory, attracting sexual partners, and signalling a species ’ identity. Although song thus acts as a species ‘ fi ngerprint ’ , it can vary greatly within species in time and space. Th is geographic variation is apparent both on large scale, when the song diff ers among populations that do not meet and thus cannot potentially breed, but also on smaller scales, when there is an intense song sharing among neighbours (Catchpole and Slater 2008). A particularly interesting type of geographic variability in vocalisation is the occurrence of dialects with sharp borders, when song characteristics change abruptly rather than gradually (Podos and Warren 2007, Catchpole and Slater 2008).

Ecography 40: 001–009, 2017 doi: 10.1111/ecog.02779

© 2016 Th e Authors. Ecography © 2016 Nordic Society Oikos Subject Editor: Mikko Monkkonen. Editor-in-Chief: Miguel Ara ú jo. Accepted 8 December 2016

Biological invasions can be a potential threat to the environ-ment (Py š ek et al. 2012, Kumschick et al. 2015, Martin-Albarracin et al. 2015) and economy (Pimentel et al. 2005, Kettunen et al. 2009), but they can also provide a golden opportunity to study evolutionary, behavioural or ecologi-cal phenomena on spatial and temporal scales unachievable by traditional experimental approaches (Hale and Briskie 2007, Yoshida et al. 2007, Schrey et al. 2011). Animal populations often diff er between native and introduced ranges, in morphological (Blackburn et al. 2013), life-history (Congdon and Briskie 2014) or behavioural traits (Pintor and Sih 2009), and biological invasions as ‘ natural experiments ’ can encompass vast areas, including whole continents (Norbury and Jones 2015), and several human generations (Lever 2005). Invasions may therefore provide insight into a range of natural phenomena.

Birds, particularly passerines (Passeriformes), are one of the most frequently studied animal groups in invasion ecol-ogy (Blackburn et al. 2009). Passerines are characterised by their complex vocalisation, or birdsong, which plays impor-tant roles in their breeding ecology, and which is transmit-ted culturally in this taxon (Catchpole and Slater 2008).

Dialects of an invasive songbird are preserved in its invaded but not native source range

Pavel Pipek , Tereza Petruskov á , Adam Petrusek , Lucie Dibl í kov á , Mark A. Eaton and Petr Py š ek

P. Pipek (http://orcid.org/0000-0003-1116-1013) (pipek@natur.cuni.cz), T. Petruskov á , A. Petrusek, L. Dibl í kov á and P. Py š ek, Dept of Ecology, Faculty of Science, Charles Univ. Prague, Czech Republic. PP and PP also at: Inst. of Botany, Dept of Invasion Ecology, Th e Czech Academy of Sciences, Pr ů honice, Czech Republic. – M. A. Eaton, RSPB Centre for Conservation Science, Newcastle-upon-Tyne, UK.

Biological invasions are not only events with substantial environmental and socioeconomic impacts but are also interesting natural experiments, allowing the study of phenomena such as the cultural evolution of bird song following introduction. We took an excellent opportunity to compare the distribution of dialects of the yellowhammer Emberiza citrinella , a small Eurasian passerine, in its native source region (Great Britain) and invaded range (New Zealand) more than hundred years after relocation. Recent fi eld recordings (including those provided by volunteers within a citizen science project) were complemented by those from archives, each assigned to appropriate dialect by visual inspection of a sonogram, and the resulting spatial patterns of dialect distribution were interpreted using historical data on the yellowhammer invasion. Th e two countries diff er markedly in the composition and distribution of dialects. New Zealand populations sing a greater number of diff erent dialects, seven in total, fi ve of which were not detected in the current British population, but have been reported by previous studies from the continental Europe. Two identifi ed localities of capture (Brighton, Sussex, UK) and release (Dunedin, Otago, NZ) diff er even more strikingly, having no dialects in common. Th e largely sedentary nature of yellowhammers allows for two mutually exclusive explanations for European dialects being detected in New Zealand but not in Great Britain: 1) the corresponding song types have emerged de novo in New Zealand, through convergent cultural evolution; 2) the dialects have disappeared from Great Britain, while being preserved in New Zealand. Indirect evidence from the widespread occurrence of these dialects in continental Europe and the reported stability of yellowhammer song, supports the latter explanation. We suggest that the yellowhammer dialect system is an avian equivalent of a phenomenon already noted in human languages, in which ancient words or structures are retained in expatriate communities.

2-EV

Dialects are often regarded as a potential fi rst step in spe-ciation (Xing et al. 2013, Potvin and Clegg 2015), although this is not the only reason to warrant research on geographic variation in song. Studies of dialects and other types of song characteristics can reveal important information about the ecology of the study species. For example, the intensity of song variant or syllable sharing among neighbours can indi-cate the degree of population fragmentation (Laiolo and Tella 2005, Petruskov á et al. 2010), a loss of song variability and disruption of the geographical pattern of dialects may refl ect population decline (Holland et al. 1996), and dialects may also help to identify the source of a recolonizing popula-tion (Brunton et al. 2008).

Several hypotheses have been proposed for the origin of bird dialects (Podos and Warren 2007, Catchpole and Slater 2008) according to function. Dialect could be adaptive, and as such indicate that the singer is well adapted to local con-ditions (local adaptation hypothesis) or belongs to a certain social group (social adaptation hypothesis). Th ey can also be non-adaptive and arise as a by-product of selection pressures on other (e.g. morphological) traits. None of these hypoth-eses appears to be universal in application, and each can work in only some dialect systems (Podos and Warren 2007, Catchpole and Slater 2008).

Once established, dialect borders can apparently persist over decades (Harbison et al. 1999, Nelson et al. 2004, Wonke and Wallschl ä ger 2009) and can be maintained even when the songs on both sides change (McGregor and Th ompson 1988, Kopuchian et al. 2004). Dialect borders can be sta-bilised either by strong assortative mating and low dispersal (Williams and Slater 1990, Ellers and Slabbekoorn 2003, Planqu é et al. 2014), or, when dispersal is frequent, by post-dispersal learning (Ellers and Slabbekoorn 2003). For several species, playback experiments have also shown that reaction of birds towards familiar and unfamiliar dialects can diff er, both in males (Petrinovich and Patterson 1981, Ratcliff e and Grant 1985, Searcy et al. 1997) and females (Baker and Jenkins 1987, Searcy et al. 1997, Danner et al. 2011).

Th e understanding of the processes behind the evolution and maintenance of dialects can be enhanced by studying populations with the same background but which have been separated from each other, either naturally (Hill et al. 2013) or with human assistance (Parker et al. 2012).

Biological invasions give us the opportunity to study the eff ect of translocation over large geographic scales. Studies on chaffi nches Fringilla coelebs introduced to New Zealand from England (Lynch et al. 1989), Japanese bush-warblers Cettia diphone introduced to Hawai ’ i (Hamao 2015) and Eurasian tree sparrows Passer montanus (Lang and Barlow 1997) introduced to the USA have demonstrated that the structure of the song and the shape of individual syllables can change signifi cantly after introduction. Th e last-mentioned study also showed that song diversity can be maintained despite an initial population bottleneck, due to subsequent rapid cultural evolution (Lang and Barlow 1997). In studies examining song evolution following natural colonization of new areas, i.e. a process similar to invasion (Hoff mann and Courchamp 2016; but see Wilson et al. 2016), the results are equivocal. In some cases, new populations show lower song variability, in terms of number of diff erent song types and syllables (Xing et al. 2013). In other situations, the song

variation in exotic population can exceed variation in any of the original ones (Kroodsma et al. 1999, Baker et al. 2003), which might arise when individuals from multiple source populations combine (Pytte 1997).

However, all the studies mentioned above explored only post-introduction changes in song characteristics, not changes in composition and distribution of dialects. For that purpose, the invasion of yellowhammers Emberiza citrinella in New Zealand provides a rare opportunity. Th e yellowham-mer is a widespread Palearctic songbird (del Hoyo et al. 2011) with a very simple song and well defi ned dialects with sharp borders, the distribution of which is known from large areas (Petruskov á et al. 2015). As with chaffi nches, yellowham-mers were introduced to New Zealand from Great Britain in the 19th century and the history of their capture and release is well documented (Pipek et al. 2015). Moreover, for Otago and partly for Canterbury regions in New Zealand the source locality in England is known: Brighton, Sussex (Pipek et al. 2015). Unlike natural colonizers, New Zealand popu-lations of yellowhammers were isolated from Great Britain and, thus, if any change has occurred in New Zealand since the last introduction event, it has to be attributed clearly to evolution in situ. Moreover, British yellowhammers seem to be largely isolated from their conspecifi cs on the continent; there have been only fi ve recaptures of birds crossing the sea between Great Britain and mainland Europe recorded since 1906 (Robinson et al. 2015), which constitutes only 0.2% of all yellowhammer recaptures in the UK.

Here, we compare the patterns of distribution of yellowhammer dialects in their European source area, Great Britain (Britain hereafter), and in the invaded range of New Zealand. Apart from our own data collection, we benefi ted from recordings available online and in sound archives, and from contributions of the general public to a citizen science project dedicated to this purpose. We combined the pat-terns obtained with historical data on introductions (Pipek et al. 2015), to test the following hypotheses: 1) the dialect diversity is greater in the native range, as only a small num-ber of birds (ca 600 – 800 in total; Pipek et al. 2015) were taken from the large British population to New Zealand, and this already small pool of dialects could have been fur-ther reduced by cultural drift, 2) new dialects evolved in the invaded range, as before the establishment and spread of yellowhammers in New Zealand there was a lack of song tutors for introduced birds, and errors in learning could have been more frequent and 3) dialect composition is similar in the localities of capture and release in the native and invaded range, respectively, due to birds ’ common origin.

Materials and methods

Study species and its song

Th e yellowhammer is a small passerine, nowadays found typ-ically in farmlands. Its native range includes a large portion of the Palearctic, from Spain in the west to central Asia in the east (del Hoyo et al. 2011). Between 1864 and 1875, more than 600 yellowhammers were brought from London as a biological control against crop pests and liberated in vari-ous localities across New Zealand, principally in Auckland

3-EV

(North Island), Canterbury (South Island) and Otago (South Island) regions (Pipek et al. 2015). All the birds that were liberated in Dunedin (n � 39) were collected around Brighton (Sussex); some (34 out of 222) of those released in Canterbury were also caught close to Brighton, the source of the remaining 188 birds is not known. In New Zealand, yellowhammers established quickly and spread to such an extent that they later became a target of organized culling (Pipek et al. 2015).

Yellowhammers are persistent singers, singing from early spring to late summer, morning until evening (Noble 1958). Th eir characteristic song typically consists of two parts (Fig. 1). Th e initial quick repetition, usually one or two syllable types, is believed to be individually specifi c (Caro et al. 2009). Th e second part, a combination of two (rarely three) terminal syllables (labelled with letters B, C, D, E, X, of which B, C and X are shown in Fig. 1), is typically shared by males of a particular dialect area (reviewed by Petruskov á et al. 2015). Such a distinction of individual dialects, sug-gested by Hansen (1985), has been supported by quanti-tative analyses of syllable spectrotemporal characteristics (Wonke and Wallschl ä ger 2009). Males with territories at the dialect border can sometimes alternate end phrases of both dialects, and are referred to as mixed singers (Glaubrecht 1989). Another apparent outcome in dialect contact zones is the emergence of intermediate three-syllable fi nal phrases (Fig. 1, 2); these can sometimes dominate whole areas and thus be regarded as dialects themselves, such as XBC in northern Germany (Glaubrecht 1989) and Denmark (Hansen 1985). Song structure as well as dialect of a particular male are set in the fi rst two years of its life and do not change later (Sch ö n 1989).

Study areas and sampling strategy

Great Britain and New Zealand are isolated landmasses of comparable sizes, with New Zealand only about 10% larger and separated into two main islands. In New Zealand, yellowhammers reach population densities three times

higher in comparable habitats (farmlands) than in Britain (MacLeod et al. 2005). We attempted to obtain recordings from throughout both countries representing the native and invaded range, but aimed for a higher density of recordings in New Zealand ’ s South Island and in southern England, refl ecting our previous fi nding that most of the birds for release in the South Island were collected in southern England (Pipek et al. 2015).

Recording collection and analysis

We set up a citizen-science project ‘ Yellowhammer Dialects ’ , inspired by a similar Czech project (Proch á zka et al. 2013). Volunteers from Britain and New Zealand were encouraged to record songs of yellowhammers in the fi eld and upload them to a dedicated website ( < http://yellowhammers.net > ), or to send them by email to the authors. Some record-ings were already stored in digital audio or video formats online. Additional recordings were obtained from sound archives (British Library Sounds, BBC Sound Eff ects library and McPherson Natural History Unit Sound Archive). Furthermore, New Zealand Dept of Conservation shared with us recordings obtained through automatic recorders distributed across New Zealand. Finally, a signifi cant por-tion of New Zealand recordings and some of the British ones were collected by authors (PP, AP, TP) in the fi eld. Most of the recordings were recorded in 2013 – 2015, since the launch of the citizen science project ‘ Yellowhammer Dialects ’ , but we also obtained some recordings from the pre-digital era. Th e oldest British recording was from 1951 (Aberdeenshire, Scotland), the oldest recording from New Zealand was from 1955 (Fiordland National Park, South Island). Th e list of all recordings used in our analysis is given in Supplementary material Appendix 1, dataset A1.

All recordings were converted into WAV format and visua-lised in Avisoft SASLab Pro ver. 5 (Specht 2007). Recordings of low quality and those not containing the complete ter-minal part of the songs (thus not allowing assignment of dialect) were excluded from further analyses. Each recording with a complete song was assigned a corresponding dialect (or two dialects, in case of recordings of mixed singers). We used the nomenclature introduced by Hansen (1985), based on the shape and frequencies of terminal syllables. However, as the diff erentiation between dialects XsB and XlB, based just on the duration of the X syllable (Hansen 1985) is ques-tionable (Wonke and Wallschl ä ger 2009, Petruskov á et al. 2015), we pooled these two into a single dialect category XB. Similarly, we simplifi ed the nomenclature of interme-diate three-syllable combinations containing the X syllable (XBC, XBB).

Rarefaction analysis

As was already mentioned, the distribution of recordings was not even; furthermore, the sampling eff ort diff ered also between the two studied countries (see Results). In order to account for this, we performed a rarefaction analysis, in which the sampled elements were not recordings but squares 10 � 10 km, which are typically used as a basic unit in atlases of breeding birds (Robertson et al. 2007, Balmer et al.

Figure 1. Spectrograms of songs representing two most common yellowhammer dialects (XB, BC) and the intermediate dialect XBC.

4-EV

compare the saturation curves (median number of detected dialect types within 95% central range; i.e. 2.5 to 97.5% quantiles) for both countries.

Results

In total, we obtained 510 recordings of yellowhammer songs with spatial information (305 from Britain and 205 from New Zealand). After removing songs lacking the fi nal, dia-lect-specifi c phrase, 365 recordings were suitable for analysis, 224 from Britain and 141 from New Zealand. We achieved

2013). Recordings occupying the same 10 � 10 km square were pooled together and all distinct dialects among them recorded (Supplementary material Appendix 1, dataset A1). On the squares that contained at least one recording with identifi ed dialect we then performed a rarefaction analysis, developed in R (R Core Team) (Supplementary material Appendix 1, Script A1).

We created random subsamples of various size (from 1 square to the total number of squares with recordings within the given country) and counted the number of captured dia-lects within these subsamples. For each number of squares, resampling was performed 1000 times. Th is allowed us to

Figure 2. Distribution of yellowhammer dialects in Great Britain and New Zealand. Coloured symbols represent diff erent dialects. Th ese are depicted with schematic spectrograms. Th e fi gure is separated into two parts: A) comparison of both countries, B) zoomed regions of interest, source region in Great Britain (Sussex) and two target localities in New Zealand (Otago and Canterbury). Map sources: Esri, DigitalGlobe, Earthstar Geographics, CNES/Airbus DS, GeoEye, USDA FSA, USGS, Getmapping, Aerogrid, IGN, IGP, and the GIS User Community.

5-EV

South Island (n � 121), the sample was large enough to cover all but one dialect (BDB) recorded in New Zealand, including the dialect BC which we did not fi nd on South Island.

Discussion

History preserved in a song: a greater diversity of dialects in the invaded range

Our study is the fi rst to explore post-invasion changes in the song dialects of an invasive bird species. We found a striking diff erence in the total number and spatial distribution of yel-lowhammer dialects recorded in the original range in Britain and introduced range in New Zealand. Although we expected a reduction of the number of dialects in the introduced range due to the bottleneck eff ect, the opposite was true (Fig. 2A). While Britain was dominated by only three dialects, seven diff erent terminal syllable combinations were recorded in New Zealand. Th ese were often scattered over relatively small areas in a mosaic-like fashion, but in some areas, their spatial distribution was more uniform. For instance, a large part of Canterbury is dominated by the XB dialect and the borders between diff erent dialect zones might be indicated by mixed singers, as is the case in Europe (Glaubrecht 1989). Finally, the overall richness of dialects was high in both North and South Island, with six dialects each.

Two corresponding areas of origin and introduction identifi ed from historical records (Pipek et al. 2015), southern England and the Dunedin area on the South

similar coverage in both countries in terms of number of squares (10 � 10 km) in national grids (OSGB36, NZMG), from which we had at least one fi nished song (136 for Britain and 105 for New Zealand).

In total, nine distinct dialect types (i.e. distinct combina-tions of terminal syllables) were detected. Britain and New Zealand diff ered markedly in dialect composition (Fig. 2A, Table 1). While two dialects were shared (BC, XB), seven were found in only one country: two in Britain, and fi ve in New Zealand. Th is diff erence was not caused by sampling eff ort as illustrated by the rarefaction analysis (Fig. 3); the 95% central range band for the two countries ceased over-lapping even in subsamples four times smaller (28) than our actual sample size (105 for New Zealand).

We recorded four dialect types in Britain, although one of these was sung by a single male only. British recordings were dominated by the XB dialect, followed by the BC dialect and an intermediate, three-syllable dialect XBC (Table 1). Th e southwest of England was dominated by a single dialect (XB) while in other parts of the country dialects alternated across relatively short distances creating a mosaic-like distri-bution (Fig. 2A). Th ree recordings from the adjacent island of Ireland were all of XB dialect.

We found substantially more yellowhammer dialects in New Zealand (seven) than in Britain (Table 1). Th e most frequent in New Zealand was XB, while three others, BhBl, BlBh and BD, were also common and were recorded with similar frequency. Th e remaining three dialects (BBe, BDB and BC) were less frequent. Th ere were also mixed singers, capable of singing XB songs in combination with character-istics typical of another dialect: BhBl (4 recordings from 2 squares), BD (n � 1) and BlBh (n � 1).

As in Britain, the XB dialect prevailed in New Zealand, being found across large areas, particularly coastal regions of the South Island. In other areas, birds singing diff erent dialects were more intermixed. For example, three dialects (BhBl, BlBh and BBe) were found within the 10 km radius around Dunedin (Otago, South Island). Five dialects found in New Zealand (BhBl, BlBh, BD, BBe and BDB) were not found in Britain. Interestingly, dialect BC, the second most common dialect in Britain, was recorded only three times in New Zealand, in a limited area of North Island (Bay of Plenty).

Direct comparison of the Brighton area (Sussex) and two regions in South Island (Canterbury and Otago), cor-responding to the localities of capture and release, respec-tively, showed even more striking diff erences between the native and invaded areas (Fig. 2B). While the coastal area of Canterbury region was, similarly to southern England, dominated by the XB dialect, birds in Otago sang mostly dialects not found in the UK.

Even though we had substantially fewer recordings with complete songs from North Island (n � 19) than from

Figure 3. Rarefaction curves constructed in order to remove the bias potentially caused by uneven sampling eff ort and diff erent sample sizes in the two countries. We used national grids with squares 10 � 10 km (OSGB36, NZGM). We created random sub-samples from all squares containing at least one song with identi-fi ed dialect. For each size of the subsample (ranging from 1 to the total number of squares with recordings per region) we made 1000 permutations and calculated the median value of number of dia-lects (represented by solid line) and 2.5 to 97.5% quantile range (shaded areas). NZ – New Zealand, GB – Great Britain.

Table 1. Proportion of dialect types recorded in the two countries.

Country Total XB BC XBC XBB BhBl BlBh BD BBe BDB Mixed singers

Great Britain birds 224 150 42 28 1 – – – – – 3squares 136 97 38 9 1 – – – – – 3

New Zealand birds 141 66 3 – – 20 18 16 8 4 6squares 105 50 3 – – 18 13 15 6 4 4

6-EV

between Britain and New Zealand, although we do not fi nd them as plausible as the above scenario. First, since we have sampled only a small portion of birds (300) out of the whole population of yellowhammers in Britain (approx. 700 000 pairs) (Musgrove et al. 2013), some of the dialects could have ‘ missed the net ’ . Sampling bias, the simplest expla-nation for observed diff erences, seems unlikely however (Fig. 3). In fact, in only 19 complete songs from North Island of New Zealand, there were more distinct syllable patterns than in the whole of Britain from which more than ten times as many recordings were available.

An alternative but also unlikely explanation of the mis-match between the British and New Zealand dialect com-position (particularly in the well-studied southern regions of both areas) would be a diff erent origin of the birds intro-duced to New Zealand than assumed. Most of the birds for Otago departed from the UK in winter (Pipek et al. 2015), and thus some of the birds captured in southern England for this purpose might have been winter immigrants from elsewhere, possibly including continental Europe (Knox and Parkin 2009). However, yellowhammers are generally sedentary, both in Britain (Wernham et al. 2002) and else-where in Europe (Cep á k et al. 2008). Th ey rarely cross the English channel: between 1906 and 2015, only two yel-lowhammers out of 2500 recaptured in Britain were ringed outside of the British Isles (Robinson et al. 2015), and only three yellowhammers ringed in Britain were recovered else-where in that period (Robinson et al. 2015). Furthermore, the mainland European localities connected with Britain through known ringing recoveries seem to be dominated by the XB dialect (Petruskov á et al. 2015) (see Supplementary material Appendix 1, Fig. A2 and dataset A2). Th erefore, oversea migration is also a highly unlikely explanation of the observed discrepancy between British and New Zealand dialect composition.

In the 19th century it was not uncommon for birds caught in continental Europe, e.g. game birds or rare spe-cies such as ortolan buntings Emberiza hortulana , to be sold on London markets (Shrubb 2013); yellowhammers obtained that way might have sung with diff erent dia-lects. However, it was probably not economically viable to import this species to Britain, where it was amongst the most common bird species (Holloway 2002). Furthermore, this scenario cannot explain the patterns around Dunedin, as it is well documented that all yellowhammers released in that area were caught in the vicinity of Brighton (Pipek et al. 2015).

One could also argue that our sampling is biased towards southern England and that some birds from less sampled British areas, such as Scotland, could migrate southwards in winter (Balmer et al. 2013, BirdLife International 2016). However, even within Britain the birds do not move far; 95% of recaptured yellowhammers were recovered within 25 km from the locality they were ringed in, with the median distance being less than 1 km (Wernham 2002). Th e longest known distance between ringing and capture within Britain is only 153 km (Robinson et al. 2015). Th ere is no indica-tion that yellowhammers migrate over long distances in New Zealand either (Angus 2013), although it must be admitted that in this country they are banded much less frequently than in Britain (Cossee 1998, Jamieson et al. 2016).

Island of New Zealand, do not match in dialect composition (Fig. 2B). In fact, the broad areas around Brighton and Dunedin have no dialects in common (the closest bird sing-ing the XB syllable combination was recorded 65 km from the centre of Dunedin). Unfortunately, the source areas of birds liberated in other New Zealand regions are unknown (Pipek et al. 2015) and thus do not allow such direct com-parison. Interestingly, the dialects found in New Zealand but not Britain are known from continental Europe (Petruskov á et al. 2015). Th is holds even for birds singing a rare combi-nation of terminal syllables BDB (Supplementary material Appendix 1, Fig. A1), which had been previously recorded in Denmark (Hansen 1985) and the Pyrenees (Matheu 1997).

Th is is not the fi rst time that the written history of an invasion has come into apparent confl ict with the interpreta-tion derived from an alternative methodological approach. For example, molecular data indicate that the spread of European hedgehogs Erinaceus europaeus through New Zealand was in the opposite direction to that implied from historical documents (Bolf í kov á et al. 2013). However, the absence from Britain of some of the yellowhammer dialects found in New Zealand can be explained without discarding the historical record. It is often assumed that when popu-lations in native and invaded ranges diff er, it is the exotic population which must have changed (Lockwood et al. 2005, Congdon and Briskie 2014). However, it can be the other way around because factors such as changes in land use in the native range can have a similar or even stronger infl uence than the novel environment. For example, mor-phological changes during the last ca 150 yr are in fact more apparent in British birds than in those introduced to New Zealand, with contemporary British yellowhammers being smaller than those in the mid-19th century (Blackburn et al. 2013). We argue that, in a similar vein, the most likely expla-nation for the observed diff erences in song dialects is that New Zealand yellowhammers have retained song structures that were present in both countries originally, but were lost from the source population in the UK subsequently. Such a phenomenon has been demonstrated in the much more complex system of human languages, for example in colonial English dialects (Hickey 2005) or in isolated Czech villages in Romanian Banat (Costachie et al. 2011).

Yellowhammers have experienced a signifi cant population decline in Britain in recent decades, in particular between the 1980s and 1990s (Robinson et al. 2016) due to the infl uence of modern agricultural practices (Chamberlain et al. 2000). Th is apparently resulted in a loss of genetic variation (Lee et al. 2001) and could also have been associated with a loss of song variability. Corn buntings Emberiza calandra , which have declined even more sharply in response to the same pressures, lost some dialect variation as a consequence of land-use change (Holland et al. 1996). In Poland (Osiejuk and Raty ń ska 2003) and in the UK (McGregor 1981), they showed clear dialect borders only in areas with high population densities.

Alternative scenarios: no evidence for sampling bias and winter immigration

Th ere are, however, several alternative scenarios, which might have resulted in the current contrasting patterns

7-EV

recordings of BBC Sound Eff ects Library, British Library, McPherson Natural History Unit Sound Archive, and NZ Dept of Conservation, respectively. British Ornithological Union, British Guides, Ornithological Society of New Zealand, Royal Forest and Bird Protection Society of New Zealand, Wildlife Recording Society, 10 000 Birds and Veronika Meduna from Radio New Zea-land helped us spreading the information about our citizen-science project. We also thank T. Blackburn for editing the manuscript and P. Scofi eld for enriching discussions. Finally, we thank the late V. Jaro š í k for being always open to new ideas. Th e project was fi nancially supported by long-term research development project RVO 67985939, Praemium Academiae award from the Czech Academy of Sciences (to P. Py š ek) and by the Grant Agency of Charles Univ. in Prague (project no. 312213).

References

Angus, D. J. 2013. Yellowhammer. – In: Miskelly, C. M. (ed), New Zealand birds online. – < http://nzbirdsonline.org.nz > .

Baker, A. J. and Jenkins, P. F. 1987. Founder eff ect and cultural evolution of songs in an isolated population of chaffi nches, Fringilla coelebs , in the Chatham Islands. – Anim. Behav. 35: 1793 – 1803.

Baker, M. C. et al. 2003. Rapid evolution of a novel song and an increase in repertoire size in an island population of an Australian songbird. – Ibis 145: 465 – 471.

Balmer, D. E. et al. 2013. Bird atlas 2007 – 11: the breeding and wintering birds of Britain and Ireland. – British Trust for Ornithology.

BirdLife International 2016. Species factsheet: Emberiza citrinella . – < http://datazone.birdlife.org/species/factsheet/22720878 >

Blackburn, T. M. et al. 2009. Avian invasions: the ecology and evolution of exotic birds. – Oxford Univ. Press.

Blackburn, T. M. et al. 2013. Body size changes in passerine birds introduced to New Zealand from the UK. – NeoBiota 17: 1 – 18.

Bolf í kov á , B. et al. 2013. Population biology of establishment in New Zealand hedgehogs inferred from genetic and historical data: confl ict or compromise? – Mol. Ecol. 22: 3709 – 3720.

Brunton, D. H. et al. 2008. Assessing natural dispersal of New Zealand bellbirds using song type and song playbacks. – N. Z. J. Ecol. 32: 147 – 154.

Caro, S. P. et al. 2009. Song repertoires in a western European population of yellowhammers Emberiza citrinella . – Acta Ornithol. 44: 9 – 16.

Catchpole, C. K. and Slater, P. J. B. 2008. Bird song: biological themes and variations. – Cambridge Univ. Press, 2nd ed.

Cep á k, J. et al. 2008. Atlas migrace pt á k ů Č R a SR. – Aventinum. Chamberlain, D. E. et al. 2000. Changes in the abundance of

farmland birds in relation to the timing of agricultural intensifi cation in England and Wales. – J. Appl. Ecol. 37: 771 – 788.

Congdon, N. M. and Briskie, J. V. 2014. Changes in the life history traits of song thrushes Turdus philomelos introduced to New Zealand. – Bird Study 61: 143 – 151.

Cossee, R. O. 1998. Report on bird banding in New Zealand 1995/1996. – New Zealand Dept of Conservation.

Costachie, S. et al. 2011. Czech minority in Banat – Romania. A social geography survey. – Geogr. Pannonica 15: 7 – 15.

Danner, J. E. et al. 2011. Female, but not male, tropical sparrows respond more strongly to the local song dialect: implications for population divergence. – Am. Nat. 178: 53 – 63.

del Hoyo, J. et al. 2011. Handbook of the birds of the world, volume 6: mousebirds to hornbills. – Lynx Edicions.

Ellers, J. and Slabbekoorn, H. 2003. Song divergence and male dispersal among bird populations: a spatially explicit

Alternative scenarios: convergent evolution is possible but less likely than preservation of dialects in the invaded range

Finally, the patterns observed could result from convergent evolution of the yellowhammer songs in New Zealand lead-ing to similar patterns as in continental Europe. As this spe-cies ’ song has a relatively simple structure, and the dialects diff er in frequency and modulation of only two or three terminal syllables, there is not infi nite space for innovation. During the initial phase of invasion, there was probably a lack of tutors (birds from which the males learn their song in the sensitive period of their life) as well as lack of positive or negative feedback from conspecifi cs. Under such condi-tions, there is a room for errors in learning and development of song elements. For example, New Zealand native saddle-backs Philesturnus rufusater evolved completely new song types in just a few decades after translocation (Parker et al. 2012). Similarly, alien Eurasian tree sparrows have replen-ished lost song variability with brand new syllables (Lang and Barlow 1997). It seems likely that initial phrases of yel-lowhammer song also undergo fast evolution, as they can show extreme variability even on a very small scale (Caro et al. 2009). However, in our study we focused not on over-all song-element variability but only on the fi nal part of the song, which is, unlike initial phrases, clearly connected to geography. Furthermore, in contrast to saddlebacks, yel-lowhammers learn and fi x their song in the fi rst two years (Sch ö n 1989), and only extremely rarely acquire some song characteristics from other bird species (Swart and Zavadil 2015). Th erefore, it is reasonable to assume that they would more often lose syllables than develop new ones.

Unfortunately, there are only very few old recordings available, which prevents the evaluation of possible tempo-ral trends in dialect composition. While there is no record of ‘ missing ’ dialects since the 1950s in Britain, one of the ‘ new dialects ’ , BDB, was recorded in New Zealand almost 60 years ago, confi rming its long-term presence in the country.

Most dialects as found in New Zealand have been recorded in several well separated places around Europe, such as Romania, Czech Republic, Spain or Finland (Petruskov á et al. 2015). Whether this pattern has arisen through inde-pendent evolution or through rare long-distance migration remains unanswered. If new syllable combinations developed easily in low-density yellowhammer populations, however, we would expect greater dialect variability in Britain, especially following the species ’ recent decline (Robinson et al. 2016).

To conclude, we can dismiss with confi dence the sam-pling bias and winter immigration scenarios as a cause of the patterns observed in yellowhammer dialects. Although we cannot do so for the convergent evolution of song struc-tures, we believe that the more parsimonious explanation is the loss of dialects in the native range due to population declines, rather than the independent emergence of numer-ous syllable combinations in New Zealand and Europe. Th us, we propose that yellowhammer songs in New Zealand are in a sense a living archive of those sung in 19th century Britain.

Acknowledgements – We thank all authors of recordings; G. Helbling, C. Tipp, L. McPherson and J. Mortimer for providing access to

8-EV

MacLeod, C. J. et al. 2005. Can increased niche opportunities and release from enemies explain the success of introduced Yellowhammer populations in New Zealand? – Ibis 147: 598 – 607.

Martin-Albarracin, V. L. et al. 2015. Impact of non-native birds on native ecosystems: a global analysis. – PLoS One 10: e0143070.

Matheu, E. 1997. Spanish birds 98/28. – Sound recording available in British Library (SHELFMARK W1CDR0000246).

McGregor, P. K. 1981. Bird song repertoires and local dialects. – PhD thesis, Univ. of Oxford.

McGregor, P. K. and Th ompson, D. B. A. 1988. Constancy and change in local dialects of the corn bunting. – Ornis Scand. 19: 153 – 159.

Musgrove, A. et al. 2013. Population estimates of birds in Great Britain and the United Kingdom. – Br. Birds 106: 64 – 100.

Nelson, D. A. et al. 2004. Cultural evolution of puget sound white-crowned sparrow song dialects. – Ethology 110: 879 – 908.

Noble, R. 1958. Late season singing of the yellowhammer. – Br. Birds 51: 290 – 303.

Norbury, G. and Jones, C. 2015. Pests controlling pests: does predator control lead to greater European rabbit abundance in Australasia? – Mamm. Rev. 45: 79 – 87.

Osiejuk, T. S. and Raty ń ska, K. 2003. Song repertoire and microgeographic variation in song types distribution in the corn bunting Miliaria calandra from Poland. – Folia Zool. 52: 275 – 286.

Parker, K. A. et al. 2012. Th e eff ects of translocation-induced isolation and fragmentation on the cultural evolution of bird song. – Ecol. Lett. 15: 778 – 785.

Petrinovich, L. and Patterson, T. L. 1981. Th e responses of white-crowned sparrows to songs of diff erent dialects and subspecies. – Z. Tierpsychol. 57: 1 – 14.

Petruskov á , T. et al. 2010. Geographic variation in songs of the tree pipit ( Anthus trivialis ) at two spatial scales. – Auk 127: 274 – 282.

Petruskov á , T. et al. 2015. A review of the distribution of yellowhammer ( Emberiza citrinella ) dialects in Europe reveals the lack of a clear macrogeographic pattern. – J. Ornithol. 156: 263 – 273.

Pimentel, D. et al. 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. – Ecol. Econ. 52: 273 – 288.

Pintor, L. M. and Sih, A. 2009. Diff erences in growth and foraging behavior of native and introduced populations of an invasive crayfi sh. – Biol. Invasions 11: 1895 – 1902.

Pipek, P. et al. 2015. How the yellowhammer became a Kiwi: the history of an alien bird invasion revealed. – NeoBiota 24: 1 – 31.

Planqu é , R. et al. 2014. On the maintenance of bird song dialects. – J. Math. Biol. 68: 505 – 531.

Podos, J. and Warren, P. S. 2007. Th e evolution of geographic variation in birdsong. – Adv. Study Behav. 37: 403 – 458.

Potvin, D. A. and Clegg, S. M. 2015. Th e relative roles of cultural drift and acoustic adaptation in shaping syllable repertoires of island bird populations change with time since colonization. – Evolution 69: 368 – 380.

Proch á zka, P. et al. 2013. Citizen-Science-Projekte enth ü llen die Geheimnisse der Goldammerdialekte. – Vogelwarte 51: 307 – 308.

Py š ek, P. et al. 2012. A global assessment of invasive plant impacts on resident species, communities and ecosystems: the interaction of impact measures, invading species ’ traits and environment. – Global Change Biol. 18: 1725 – 1737.

Pytte, C. L. 1997. Song organization of house fi nches at the edge of an expanding range. – Condor 99: 942 – 954.

Ratcliff e, L. M. and Grant, P. R. 1985. Species recognition in Darwin ’ s fi nches (Geospiza, Gould). III. Male responses to

model testing the role of vocal learning. – Anim. Behav. 65: 671 – 681.

Glaubrecht, M. 1989. Geographische Variabilit ä t des Gesangs der Goldammer, Emberiza citrinella , im norddeutschen Dialekt-Grenzgebiet. – J. Ornithol. 130: 277 – 292.

Hale, K. and Briskie, J. V. 2007. Response of introduced European birds in New Zealand to experimental brood parasitism. – J. Avian Biol. 38: 198 – 204.

Hamao, S. 2015. Rapid change in song structure in introduced japanese bush-warblers ( Cettia diphone ) in Hawai ’ i. – Pac. Sci. 69: 59 – 66.

Hansen, P. 1985. Geographic song variation in the Yellowhammer ( Emberiza citrinella ). – Nat. Jutlandica 21: 209 – 219.

Harbison, H. et al. 1999. Long-term persistence of song dialects in the mountain white-crowned sparrow. – Condor 101: 133 – 148.

Hickey, R. 2005. Dialects of English and their transportation. – In: Hickey, R. (ed), Legacies of colonial English. Cambridge Univ. Press, pp. 33 – 58.

Hill, S. D. et al. 2013. A comparison of vocalisations between mainland tui ( Prosthemadera novaeseelandiae novaeseelandiae ) and Chatham Island tui ( P. n. chathamensis ). – N. Z. J. Ecol. 37: 214 – 223.

Hoff mann, B. D. and Courchamp, F. 2016. Biological invasions and natural colonisations: are they that diff erent? – NeoBiota 29: 1 – 14.

Holland, J. et al. 1996. Changes in microgeographic song variation of the corn bunting Miliaria calandra . – J. Avian Biol. 27: 47 – 55.

Holloway, S. 2002. Th e historical atlas of breeding birds in Britain and Ireland 1875 – 1900. – T. and A. D. Poyser.

Jamieson, S. et al. 2016. Bird banding in New Zealand 1996 – 2013. – New Zealand Dept of Conservation.

Jenkins, P. F. and Baker, A. J. 1984. Mechanisms of song diff erentiation in introduced populations of Chaffi nches Fringilla coelebs in New Zealand. – Ibis 126: 510 – 524.

Kettunen, M. et al. 2009. Technical support to EU strategy on invasive species (IAS): assessment of the impacts of IAS in Europe and the EU (fi nal module report for the European Commission). – Inst. for European Environmental Policy, Brussels.

Knox, A. and Parkin, D. 2009. Th e status of birds in Britain and Ireland. – Christopher Helm Publishers.

Kopuchian, C. et al. 2004. Temporal stability and change in a microgeographical pattern of song variation in the rufous-collared sparrow. – Anim. Behav. 68: 551 – 559.

Kroodsma, D. E. et al. 1999. Geographic variation in black-capped chickadee songs and singing behavior. – Auk 116: 387 – 402.

Kumschick, S. et al. 2015. Comparing impacts of alien plants and animals in Europe using a standard scoring system. – J. Appl. Ecol. 52: 552 – 561.

Laiolo, P. and Tella, J. L. 2005. Habitat fragmentation aff ects culture transmission: patterns of song matching in Dupont ’ s lark. – J. Appl. Ecol. 42: 1183 – 1193.

Lang, A. L. and Barlow, J. C. 1997. Cultural evolution in the Eurasian tree sparrow: divergence between Introduced and ancestral populations. – Condor 99: 413 – 423.

Lee, P. L. M. et al. 2001. Microsatellite variation in the yellowhammer Emberiza citrinella : population structure of a declining farmland bird. – Mol. Ecol. 10: 1633 – 1644.

Lever, C. 2005. Naturalised birds of the world. – T. and A. D. Poyser.

Lockwood, J. L. et al. 2005. Th e role of propagule pressure in explaining species invasions. – Trends Ecol. Evol. 20: 223 – 228.

Lynch, A. et al. 1989. A model of cultural evolution of chaffi nch song derived with the meme concept. – Am. Nat. 133: 634 – 653.

9-EV

Specht, R. 2007. Avisoft-SASLab Pro ver. 4.5, 5. Sound analysis and synthesis software. – Berlin, Germany: Avisoft Bioacoustics.

Swart, J. M. and Zavadil, V. 2015. A corn bunting (Emberiza calandra ) imitating yellowhammer ( Emberiza citrinella ) song. – Sylvia 51: 75 – 88.

Wernham, C. et al. 2002. Th e migration atlas: movements of the birds of Britian and Ireland. – T. and A. D. Poyser.

Williams, J. M. and Slater, P. J. B. 1990. Modelling bird song dialects: the infl uence of repertoire size and numbers of neighbours. – J. Th eor. Biol. 145: 487 – 496.

Wilson, J. R. H. et al. 2016. Biological invasions and natural colonisations are diff erent: the need for invasion science. – NeoBiota 31: 87 – 98.

Wonke, G. and Wallschl ä ger, D. 2009. Song dialects in the yellowhammer Emberiza citrinella : bioacoustic variation between and within dialects. – J. Ornithol. 150: 117 – 126.

Xing, X. Y. et al. 2013. Recent northward range expansion promotes song evolution in a passerine bird, the light-vented bulbul. – J. Evol. Biol. 26: 867 – 877.

Yoshida, T. et al. 2007. Biological invasion as a natural experiment of the evolutionary processes: Introduction of the special feature. – Ecol. Res. 22: 849 – 854.

playback of diff erent song types, dialects and heterospecifi c songs. – Anim. Behav. 33: 290 – 307.

Robertson, C. J. R. et al. 2007. Atlas of bird distribution in New Zealand 1999 – 2004. – Ornithological Society of New Zealand.

Robinson, R. A. et al. 2015. Th e online demography report: bird ringing and nest recording in Britain and Ireland in 2015. – BTO, Th etford.

Robinson, R. A. et al. 2016. BirdTrends 2015 : trends in numbers, breeding success and survival for UK breeding birds. – BTO Res. Rep. 678.

Sch ö n, R. 1989. Dialekte, Individualit ä t und Gesangslernen bei der Goldammer ( Emberiza citrinella L.). – PhD thesis, Univ. of Wien.

Schrey, A. W. et al. 2011. Broad-scale latitudinal patterns of genetic diversity among native European and introduced house sparrow ( Passer domesticus ) populations. – Mol. Ecol. 20: 1133 – 1143.

Searcy, W. A. et al. 1997. Th e response of male and female song sparrows to geographic variation in song. – Condor 99: 651 – 657.

Shrubb, M. 2013. Feasting, fowling and feathers: a history of the exploitation of wild birds. – Poyser.

Supplementary material (Appendix ECOG-02779 at < www.ecography.org/appendix/ecog-02779 > ). Appendix 1.