The past decades have seen a number of studies on molecular diversity, systematics and phylogeography of European Odonata. Weekers et al. (2001) used the highly variable nuclear ribosomal ITS region to explore the phylogeny and diversity of European Calopteryx species. They found an overall east-west division at the species level, but also a clear separation between C. splendens splendens (Harris, 1780) and the Italian subspecies C. splendens caprai Conci, 1956. Sadeghi et al. (2010) in an AFLP based study of C. splendens found a complex pattern that only partially corresponds to the current subspecific taxonomy. They found indication of a post-glacial recolonization of southern Europe from the southeast. Guan et al. (2013) studied the phylogeny of the genus Pyrrhosoma Charpentier, 1840 (red damselflies) based on the ITS region and the mitochondrial gene COI with special focus on the status and position of the endangered P. elisabethae Schmidt 1948. They found that while P. elisabethae was indeed monophyletic it grouped within the widespread species P. nymphula (Sulzer, 1776) and—surprisingly—closely related to P. nymphula specimens from Morocco, clearly indicating that more studies on this species complex are needed. Hinojosa et al. (2017) found a distinct pattern—albeit with low variation—in the ITS region and COI of the skimmer Sympetrum vulgatum (Linnaeus, 1758) with the three Western Palearctic subspecies S. vulgatum vulgatum (most of Europe), S. vulgatum ibericum Ocharan, 1985 (Iberian Peninsula), and S. vulgatum decoloratum Selys, 1884 (Anatolia) appearing separated and distinct. In that study, they suggested that the three subspecies represented three different glacial refugia in the Balkans, the Iberian Peninsula, and Anatolia, respectively. Schneider et al. (2015) and Simonsen et al. (2020) studied the Aeshna cyanea (Müller, 1794) and A. vercanicaSchneider et al., 2015 species complex. Their results indicate that A. cyanea colonized Western and Central Europe from the Iberian Peninsula, which had previously been colonized from the Maghreb region in Northern Africa. In contrast to the studies above, which provide detailed information on the diversity and distributional patterns of Western Palearctic Odonata, other studies have found little or no variation within the Western Palearctic region. Bernard et al. (2011) found evidence for a rapid expansion from East Asia through western Siberia into West Palearctic in Nehalennia speciosa Charpentier, 1840 with little variation across its range. Kohli et al. (2018; 2021) found very little variation and no phylogeographic pattern in the circumpolar Somatochlora sahlbergi Trybom, 1889 across its range. While Kohli et al. (2021) generally found clear distinction between Nearctic and Palearctic populations in Aeshna juncea (Linneaus, 1758), Aeshna subarctica Walker, 1908, Libellula quadrimaculata Linneaus, 1758, and Sympetrum danae (Sulzer, 1776), as well as some internal Nearctic population structure and occasional distinction between Japan and the remaining Palearctic, they found little variation and no internal structure in the Western Palearctic for either four species.

The molecular diversity and phylogeography of Odonata in the Western Palearctic thus appear highly variable with the potential to illuminate a number of phylogeographic patterns. European phylogeography is currently dominated by four paradigms associated with expansion from the three peninsular refugia, on the Iberian, Italian and Balkan peninsulas (see Hewitt 1999; 2004; Schmitt 2007 for review and details): the Hedgehog Paradigm with postglacial expansion from the Iberian, Italian and Balkan peninsulas; the Bear Paradigm with postglacial expansion from the Iberian Peninsula and an eastern refugium (not necessarily the Balkan Peninsula); the Butterfly Paradigm with postglacial expansion from the Italian and Balkan peninsulas; and the Grasshopper Paradigm with postglacial expansion from the Balkan Peninsula. Especially the Iberian and Italian peninsulas may show historic connections to the North African Maghreb regions (see Husemann et al. 2014; Simonsen et al. 2020 and references therein). In addition to the four paradigms and North African refugia, two other patterns appear to be important in European phylogeography (e.g., Schmitt 2007). First, there is a growing body of evidence for refugia in Europe north of the Pyrenees, the Alps, and the Balkans (e.g., Ursenbacher et al. 2006; Schmitt 2007; Schmitt et al. 2007; Gratton et al. 2008; Varga & Schmitt 2008; Hammouti et al. 2009; Simonsen & Huemer 2014). Second, it has been hypothesized that a considerable number of European species originate in the East Palearctic (de Lattin 1967). Even though the latter has been challenged (e.g., Schmitt 2007), some of the studies listed above (Bernard et al. 2011; Kohli et al. 2018; 2021) indicate an East Palearctic origin of some West Palearctic odonate species.

Here we present a phylogeographic study of the damselfly Platycnemis pennipes (Pallas, 1771) in Europe based on the mitochondrial gene COI. P. pennipes is a widespread species in continental Europe (Fig. 1) except for the Iberian Peninsula, and central and northern Scandinavia and Finland. It is also missing from Ireland, northern England and Scotland, and most Mediterranean islands. Outside Europe it occurs in Anatolia, the Caucasus, and extends eastwards into western Siberia and northern Kazakhstan as far east as the Yenisei River (Boudot et al. 2015). Generally, the taxonomy of P. pennipes is considered uncomplicated, but a separate subspecies, P. penniepes nitidula (Brullé, 1832), occurs in the coastal regions of Montenegro, Albania and Greece where it may hybridize with the nominal subspecies (Boudot et al. 2015). Throughout its range P. pennipes is found associated with a variety of slow flowing and standing waters with surface and edge vegetation (Nielsen 1998; Askew 2004; Boudot et al. 2015). At least in Denmark it is considered a species with low mobility.

While there are problems associated with using COI as a single genetic marker for diversity and phylogeography studies (e.g. Kondandaramaiah at al. 2013; Brunet et al. 2017; Roe et al. 2017), several of the studies mentioned above demonstrate that it can be used (albeit with some caution) in Odonata. Utilising COI also allow us to combine new data with the extensive, and rapidly growing, DNA barcode data that are already available online. Recently, two major studies of European Odonata (Galimberti et al. 2020; Geiger et al. 2021) have made a considerable number of European odonatan COI sequences available. By combining these and other publicly available sequences with several new sequences we are able to analyse an extensive geographical dataset of a European damselfly to address key question about European phylogeography.

Our results reveal a highly distinctive pattern with two clearly separated haplotype groups of P. pennipes in Europe—one restricted to Italy, and one found in the rest of the species’ distribution area in Europe from Georgia to the UK and from southern France to Scandinavia. This clearly indicates that P. pennipes remained in at least two independent glacial refugia in Europe during the Weichsel Glaciation (Gibbard & Cohen 2008) (Fig. 4), and perhaps even before that. The genetic variation within and between the two haplotype groups in P. pennipes (Table 2) are much higher than in S. vulgatum (Hinojosa et al. 2017) or N. speciosa (Bernard et al. 2011), but similar to A. cyanea (Schneider et al. 2015; Simonsen et al. 2020), and Orthetrum cancellatum (Linnaeus, 1758) and O. coerulescens (Fabricius, 1798) (Simonsen et al. 2021). If the rate of evolution in P. pennipes is comparable to that found in A. cyanea by Simonsen et al. (2020), the difference between the two haplotype groups in P. pennipes (0.015) corresponds roughly to the difference between North African and European A. cyanea (0.013—Simonsen et al. 2020, table 2). The split between the two latter groups was estimated to have occurred ca. 400 ky ago (Simonsen et al. 2020, fig. 3), meaning that the split between the two haplotype groups in P. pennipes could date back to the onset of the Saale Glaciation, rather than the onset of the Weichel Glaciation (Gibbard & Cohen 2008).

One of these refugia was obviously in Italy as haplotype group A is restricted to that area. As the greatest variation in haplotype group B is found in the east and southeast, we infer that the second refugium was in the Balkan-Black Sea region of Europe. This pattern is directly comparable to the so-called Grasshopper Paradigm, so named because the pattern is found in the grasshopper Pseudochorthippus parallelus (Zetterstedt, 1821) (Hewitt 1999; 2004; Schmitt 2007). According to these authors P. parallelus recolonized Europe north of the Alps and the Pyrenees from a refugium in the northern Balkans with isolated populations still existing in the southern Balkans/Anatolia as well as in the Iberian Peninsula and the Italian Peninsula. P. pennipes does not occur in the Iberian Peninsula (Boudot et al. 2015), but the Italian population is isolated from the rest of Europe, and it does not appear to have crossed the Alps, possibly because it rarely occurs more than 1000 m above sea level (Boudot et al. 2015). Our regional sampling from the southern Balkans and Anatolia is too limited to allow us to infer the extent of the Balkan refugium for P. pennipes with certainty, but the presence of a specimen with a haplotype very similar to Romanian specimens in Georgia (ENT-DNA-917) could indicate that it extended to the eastern shores of the Black Sea (Fig. 4). While we cannot exclude that a single, large Balkan refugium existed during the Weichsel Glaciation, the presence of two genetically highly distinct specimens from Bulgaria (ENT-DNA-1011, haplotype A7) and Montenegro (ZPLOD834-20, haplotype A8) could indicate that several subrefugia existed in the region, possibly in the south as in P. parallelus. As mentioned earlier, the distinct subspecies P. pennipes niti­dula occurs in costal Montenegro as well as Albania and Greece. It is therefore also possible that ZPLOD834-20 represents this subspecies. However, neither the original publication (Galimberti et al. 2020) nor the record on BOLD mention subspecific status for the specimen. The low genetic variation in the European group—with the exception of the two Balkan specimens mentioned above—indicate a rapid, phalanx-like post-glacial expansion (Fig. 4) as seen in several common butterfly species (see Schmitt 2007 for a review). The low genetic variation in the European group does not allow us to assess how P. pennipes colonized southern Scandinavia or the British Isles (Fig. 4). To reach southern Scandinavia it may have used the area that is today Denmark as steppingstones, or it may have crossed the Baltic Sea directly, either by long-distance dispersal or at a time when the sea levels were lower. Similarly, it may have reached the British Isles via the ancient landmass Doggerland (Lambeck 1995) as hypothesized for the ghost moth Hepialus humuli (Linnaeus, 1758) by Simonsen & Huemer (2014), or it may have crossed the English Channel either by long-distance dispersal or reached what is today southern England before the English Channel formed 10,000-8000 yrs ago (Lambeck 1995).

In addition to P. paralellus, several European species show a similar post-glacial dispersal pattern, and Hewitt (2004) stated that the Balkans comprise the main refugium for many European species. The two tree species Fagus sylvatica Linnaeus, 1753 and Alnus glutinosa (Linnaeus, 1790), and the water newt Triturus cristatus (Laurenti, 1768) all show a pattern with expansion across Europe from the Balkans, with populations remaining isolated in the Iberian Peninsula and the Italian Peninsula (Demesure et al. 1996; King & Ferris 1998; Wallis & Arntzen 1989; Hewitt 1999; 2004; Schmitt 2007). Ursenbacher et al. (2006) found that in the viper Vipera berus (Linnaeus, 1758) there is a distinct Italian refugium clearly separated from a Balkan/Northern European group, while Sztencal-Jablonka et al. (2015) found potential support for a southeastern refugium in the smooth snake Coronella austriaca Laurenti, 1768. In Lepidoptera, Louy et al. (2013) found a clear split between Italian and Balkan populations in the butterfly Coenonympha rhodopensis Elwes, 1900, while Simonsen & Huemer (2014) found deep division between Italian (and southern Austrian) and northern (including northern Austrian) populations in H. humuli. Simonsen & Huemer (2014) concluded that H. humuli likely survived the Weichsel Glaciation in peripheral alpine refugia to the south, northwest, north, and northeast of the Alps. They did, however, mention the possibility that the northeastern refugium could have extended further to the southeast.

Within Odonata several different patterns have been elucidated in the Western Palearctic region. Weekers et al. (2001) found in a study of the highly variable nuclear ITS in C. splendens that there was a clear gap between Italy and the rest of the Western Palearctic, indicating an isolated Italian refugium. In an AFLP study of the same species, Sadeghi et al. (2010) found indication of a southeastern refugium. The two studies thus point to a pattern similar to what we find here. Hinojosa et al. (2017) similarly found evidence for an isolated Iberian refugium, an Anatolian refugium, and a Balkan refugium in S. vulgatum with post-glacial expansion to the rest of Europe from the latter. In contrast to these studies, Guan et al. (2013) found in a phylogenetic study of the genus Pyrrhosoma with focus on the relationship of the widespread P. nymphula and the endangered south Balkan endemic P. elisabethae that P. elisabethae is closer related to P. nymphula from Morocco than either are to P. nymphula for the Balkans or western Europe. Schneider et al. (2015) and Simonsen et al. (2020) found a complex pattern in A. cyanea with sequential glacial refugia in the Caucasus, the Maghreb region in Northern Africa, and the Iberian Peninsula, all related to different glacial events in the Quaternary. Simonsen et al. (2021) found complex and admixed patterns in O. cancellatum and O. coerulescens in Europe. However, they did find evidence for post-glacial expansion from an Italian refugium in both species, as well as evidence for post-glacial expansion from a Balkan refugium in O. cancellatum. In O. coerulescens they found some evidence of contact between the Italian refugium and the Balkan region, as well as evidence for east-west Mediterranean contact. Simonsen et al. (2023) found evidence in COI for the current division of the small emerald spreadwing Lestes virens (Charpentier, 1825) into the subspecies L. virens virens and L. virens vestalis Rambur, 1842, probably originating from a southwestern and a combined Italian/Balkan refugium respectively. However, they also found evidence for a separate Sicilian/Mediterranean refugium. Finally, Bernard et al. (2011), and Kohli et al. (2018; 2021) found no internal pattern in N. speciosa, and A. juncea, A. subarctica, S. sahlbergi, L. quadrimaculata and S. danae, respectively, but rather evidence for colonization of the Western Palearctic region from the east.

T.J.S. and K.O. designed the study, secured funding and collected material. M.D. carried out laboratory work, data mining and the initial analyses, and drafted parts of the text. O.F.N. provided information on biology and natural history, and collected material. T.J.S. carried out the bulk of the analyses and drafted the text. All authors contributed to the Discussion and the final version of the paper.

Funding was provided by The Research Fund under the Danish Ministry for Culture (Grant: FORM.2015-0023), the 15 June Foundation (Grant: 2015-A-89), Schjøtz-Christensen’s Memorial Fund (Denmark), and the EU Synthesys Program (SE-TAF-5543).

The authors have declared that no competing interests exists.