Introduction

Hares (genus Lepus Linnaeus, 1758) are represented in the modern fauna by 32 species, five of which are distributed in Europe (Smith et al. 2018). Molecular genetic studies suggest that the genus Lepus most probably originated in North America ca. 12 Myr ago and later dispersed across other continents via an ancestral lineage that crossed the Bering Strait to Asia about 7–5 Myr ago (Matthee et al. 2004). The geographic ranges of modern European species of hares partly overlap, and hybridisation is quite common in this group (Alves et al. 2003, 2008; Melo-Ferreira et al. 2005), possibly having an adaptive effect (Thulin et al. 1997, 2006; Ferreira et al. 2021; Pohjoismaki et al. 2021). Systematic treatments of lagomorphs in general are often controversial because of not only hybridisation and introgression, but also rapid radiation and local adaptations to a variety of habitats (Melo-Ferreira et al. 2012), which resulted in high intraspecific variability of widespread species. Molecular phylogenetic research on leporids has also provided conflicting results in relation to data inferred from morphology-based analyses (Robinson and Matthee 2005). Previous studies have revealed recurrent introgressions at all stages of the Lepus radiation, which resulted in genetic connections among all major clades of Eurasian hares (e.g., Thulin et al. 1997, 2006; Alves et al. 2003, 2006; Melo-Ferreira et al. 2005, 2007, 2012; Fredsted et al. 2006; Ahlgren et al. 2016; Ferreira et al. 2021).

Remains of hares are known from a number of Pleistocene and Early Holocene localities in the territory of Ukraine (Fig. 1). They were usually identified as any of three species — the brown hare Lepus europaeus Pallas, 1778, the mountain hare Lepus timidus Linnaeus, 1758, or the extinct Don hare Lepus tanaiticus Gureev, 1964 (Gromov 1961; Rekovets 1985, 1994, 1995; Rekovets and Topachevsky 1988; Averianov 1994, 1998, 2001; Krokhmal’ and Rekovets 2010). The latter was erected based on morphological characters of a right mandible from the Late Pleistocene of the Southern Urals. According to the original description, L. tanaiticus is also present in the fossil record of the Novgorod-Siverskyi locality in Ukraine and other Late Pleistocene and Early Holocene faunal assemblages of Eastern Europe and Northern Asia (Rekovets 1985, 1995; Averianov 1995, 1998, 2001; Kosintsev 2007; Prost et al. 2010; Sharko et al. 2023).

Fig. 1
figure 1

Pleistocene and Holocene record localities of hares in the territory of Ukraine (after Krokhmal’ and Rekovets 2010, modified after Krokhmal’ et al. 2021, 2022; Gorobets et al. 2023): 1 — Melekino; 2 — Kairy; 3 — Lymany; 4 — Ushkalka; 5 — Cherevychne; 6 — Luzanivka; 7 — Karai-Dubina; 8 — Bilshovyk; 9 — Protopopovka; 10 — Syniakovo; 11 — Ozerne; 12 — Morozivka; 13 — Krasnosilka; 14 — Gunky; 15 — Demydivka; 16 — Stari Kodaky; 17 — Kiik-Koba; 18 — Chokurcha; 19 — Kabazi; 20 — Gintsi (Gontsy); 21 — Molodova; 22 — Adzhi-Koba; 23 — Syuren 1; 24 — Buran-Kaya; 25 — Nyzhnie Kryvche; 26 — Novgorod-Siverskyi; 27 — Anetivka; 28 — Mezhyrich; 29 — Pushkari; 30 — Amvrosiivka; 31 — Dobranichivka; 32 — Chulativ; 33 — Mizyn (Mezin); 34 — Divochi Skeli; 35 — Osokorivka; 36 — Proshchalnaya; 37 — Alymivskyi Navis; 38 — Murzak-Koba; 39 — Syuren 2; 40 — Shan-Koba; 41 — Mala Ugolka; 42 — Raspopyntsi. Location numbers of the samples analysed in the study are indicated in bold.

Full size image

The morphological traits that distinguish L. tanaiticus as an independent species include its large body size (10% exceeding that of L. timidus) and features of dental morphology, particularly the high dental portion of the mandible, especially on p3). It has, however, been argued that body size and dental morphology are highly adaptive traits and thus poor markers for species assignment (see Prost et al. 2010), and the species status of L. tanaiticus has repeatedly been questioned. A comparative phylogenetic analysis of L. tanaiticus and extant species of the arctic hare group (Lepus arcticus Ross, 1819, Lepus othus Merriam, 1900, and L. timidus) based on a mitochondrial D-Loop fragment provided no reliable support for the species status of the Don hare, and it was suggested to be a distinct morphotype of L. timidus (Prost et al. 2010). Sharko et al. (2023) analysed the mitochondrial genome of Pleistocene Don hares from north-eastern Eurasia and revealed that specimens of L. tanaiticus younger than 39 ka BP are phylogenetically close to L. timidus, whereas older specimens form a distinct mitochondrial clade. The authors speculated that the genetic proximity between ‘young’ Don hares and the mountain hare is a result of hybridisation that occurred at the end of the Late Pleistocene.

Phylogenetic studies of L. tanaiticus have usually included specimens from localities in the territory of Russia, but specimens assigned to this species and recovered from other parts of Europe have never been analysed before using molecular genetic approaches. The aim of our study was to carry out a comparative analysis of mitochondrial DNA sequences of representatives of the arctic hare group, including, for the first time, specimens of L. tanaiticus recovered from Late Pleistocene localities of Ukraine, and contribute to the understanding of its taxonomic status and relationship to other species of arctic hares.

Materials and methods

The specimens considered (mandible fragments and parts of the postcranial skeleton) represent the remains of eight individuals of the genus Lepus recovered from four localities in the territory of Ukraine dated to 16,000–20,000 years BP (Table 1). They are housed in the Department of Palaeontology at the National Museum of Natural History, National Academy of Sciences of Ukraine (Kyiv). These specimens were assigned to Lepus tanaiticus (n = 5) and Lepus sp. (n = 3) based on morphological characters (Gureev 1964; Rekovets 1985; Rekovets and Topachevsky 1988).

Table 1 Details of the Lepus samples analysed in this study
Full size table

The molecular analysis was carried out in the Department of Genetics at Wrocław University of Life Sciences and in the Institute of Genetics and Biotechnology of the University of Warsaw (Poland). The phenol-chloroform method (Bilton and Jaarola 1996; Capelli and Tschentscher 2005) accompanied with three additional methods — with Dextran Blue (Kalmar et al. 2000), silica (Höss and Pääbo 1993; Rohland and Hofreiter 2007) and Chelex (Walsh et al. 1991) — was applied to extract the mtDNA from the specimens considered.

Ancient DNA is usually highly fragmented thus it is recommended to use primer pairs for a few relatively short fragments (Poinar 2003), product length ca. 80–200 bp (Table 2). In order to verify the results, at least three replications of the PCR and sequencing reactions were performed for each reaction. Five short overlapping fragments with a total length of about 300 bp were used for the amplification of the control-region sequence (Prost et al. 2010).

Table 2 Primer pairs used to amplify the ancient mtDNA sequences
Full size table

For the amplification of the cytochrome b sequence, five overlapping fragments with a total length of about 300 bp were used (Smith et al. 2017). In each case, the PCR was prepared in a volume of 25 µl. It was performed with 2.5 µl of PCR buffer, 0.5 µl of each primer (10 µM), 0.5 µl of dNTP (10 µM), 1 µl of MgCl (2.5 mM), 2.5 µl of BSA (1 mg/ml), 0.4 µl of Hot Start polymerase (Qiagen, Germany), distilled water, and the appropriate volume of DNA depending on the quality and concentration of the template. The mixtures were amplified in the Eppendorf Mastercycler EP thermocycler using the following program: 95 °C — 10 min, then 45 cycles (for CR sequences) and 55 cycles (for CytB) [94 °C — 30 s., 48–50 °C (in the case of CR) or 50–51 °C (in the case of CytB sequences) (depending on the primer) — 30 s., 72 °C — 30 s.], 72 °C — 7 min, 12 °C ‘for ever’. PCR products were purified with Exo/SAP (USB Corp.) and sequenced using ABI BigDye v3.1 terminators (Applied Biosystems) on an ABI 3730XL sequencer at the sequencing facility of the Institute of Biochemistry and Biophysics, Polish Academy of Sciences in Warsaw (Poland).

Three replicates of the PCR for each specimen and each primer were aligned and a consensus sequence was constructed. Then, the sequences from both primers for the same DNA fragment were aligned and a consensus sequence was created for a given primer pair in BioEdit software (Hall 1999). The sequences of all fragments were aligned into the final consensus sequence. The sequences obtained in this way for all individuals, combined with those obtained from GenBank database, were aligned using the Muscle algorithm (Edgar 2004) implemented in Seaview (Gouy et al. 2010). The obtained sequences were compared with the available control-region sequences of L. tanaiticus (Prost et al. 2010) and control-region and cytochrome b sequences of selected representatives of the genus Lepus from GenBank database (Halanych et al. 1999; Pierpaoli et al. 1999; Waltari et al. 2004; Kasapidis et al. 2005; Waltari and Cook 2005; Wu et al. 2005; Melo-Ferreira et al. 2007, 2012; Prost et al. 2010; Ramirez-Silva et al. 2010) (Tables 3 and 4).

Table 3 The list of cytochrome b sequences from GenBank used in this study
Full size table
Table 4 The list of control-region sequences from GenBank used in this study
Full size table

After alignment, the analysed sequences were truncated in order to obtain a compact block of DNA. The sets of sequences prepared in this way were used to create phylogenetic trees using Maximum Likelihood (ML) and Bayesian Analysis (BA) methods. For analyses with the ML method, the PHYML program was used with the implemented smart model selection option for automatic selection of the substitution model (Guindon et al. 2010). The tree topology was verified with 1000 bootstrap replicates. The MrBayes 3.2 software was used to create the trees using the Bayesian method (Ronquist et al. 2012). The jModelTest 2.1.10 application was used to estimate the best-fit nucleotide substitution models used in the MrBayes program (Darriba et al. 2012). For protein coding sequences (CytB), separate models were selected for each codon position — JC, HKY, and K2P (K80) for I, II, and III position of the codon, while a single general model (HKY + I + G) was selected for the control-region sequence. During the analysis in the MrBayes programme, we used the option of two independent runs, each consisting of four Markov chains. The trees were sampled every hundredth generation for 20 000 000 generations (using 25% burn-in values). The analysis was completed when the average standard deviation of split frequencies for both runs stabilised at a level below 0.01. In many cases, standard tree-based phylogenetic analysis is unable to fully resolve close phylogenetic relationships; therefore, a haplotype network was created for individual sequences. The Median Joining method (Bandelt et al. 1999) implemented in the PopART 1.7 programme (Leigh and Bryant 2015) was used to create the network. For their implementation, the same data sets were used as in the case of particular phylogenetic trees.

Results

The obtained sequences combined with data from previous molecular analyses allowed the construction of phylogenetic trees and haplotype networks to study the genetic relationships among the hares considered. We suggest that the position of Lepus individuals from the Late Pleistocene of Ukraine in the phylogenetic tree based on the cytochrome b and control-region sequence analyses is reliable, considering the high posterior probability level and support of all nodes (Figs. 2 and 3).

Fig. 2
figure 2

Phylogenetic tree of the genus Lepus based on the analysis of cytochrome b mtDNA sequences. Bayesian posterior probabilities (BPP) and NJ bootstrap values are shown at each branching point (BBP ⁄ NJ bootstraps). Branches indicated with an asterisk are not supported under the 70% majority consensus in the NJ tree. Sequences highlighted in bold belong to the specimens from Ukraine (L. tanaiticus, Lepus sp.). Sequences of L. europaeus were used as outgroup

Full size image
Fig. 3
figure 3

Phylogenetic tree of the genus Lepus based on the analysis of the control-region mtDNA sequences

Full size image

Cytochrome b sequence analysis

The cytochrome b mtDNA sequences were obtained from four individuals of the genus Lepus whose remains were recovered from Mizyn and Gintsi. They were compared with the respective sequences of extant members of the arctic hare group (L. timidus, L. arcticus, and L. othus) now distributed in Northern Eurasia, Central Europe, and North America. Two distinct clades were obtained with high support of nodes; a polytomic clade A including the majority of analysed specimens and clade B represented by a single L. timidus from western Russia (Fig. 2). Within clade A, three sub-groups can also be noted comprising specimens of L. timidus from the Alps and eastern Russia together with those from Mizyn and Gintsi (A1), specimens from Norway (A2) and the Urals (A3). Otherwise, however, representatives of L. timidus are scattered in the tree. The individuals of L. arcticus and L. othus do not form clades separate from L. timidus. Two individuals from Mizyn and Gintsi form their own branch with a high Bayesian support (0.88). The sequences of two other individuals from Gintsi are located separately, in a different part of the tree (Fig. 2).

The cytochrome b sequences ranging in length from 290 to 364 bp and those from GenBank were used to build the haplotype network of the analysed Lepus species (Fig. 4). As in the case of the phylogenetic tree, there are two groups recognisable in the network. Group 1 consists of representatives of the brown hare, whereas all other individuals considered (L. arcticus, L. othus, and L. timidus from GenBank as well as L. tanaiticus from Mizyn and Lepus sp. from Gintsi) are placed in the second group with a star-like topology. Group 2 consists of a single haplotype located in the centre, which was the most common among all the analysed individuals, and six other haplotypes (Fig. 4).

Fig. 4
figure 4

Haplotype network of species of the genus Lepus based on the analysis of cytochrome b mtDNA sequences. The size of the nodes is proportional to their haplotype frequency. Missing intermediates are indicated by black dot. The lengths of the branches are proportional to the number of substitutions separating the haplotypes

Full size image

Control-region sequence analysis

The control-region mtDNA sequences of hares ranging in length from 115 to 332 bp were analysed separately. They were obtained from seven individuals of the genus Lepus from Mezhyrich, Mizyn, Novgorod-Siverskyi, and Gintsi. The respective sequences of Lepus arcticus, L. europaeus, L. othus, and L. timidus from GenBank were used for comparison. In addition, five sequences indicated as L. t. tanaiticus by Prost et al. (2010) were also considered.

Both Bayesian and NJ tree reconstructions resulted in a consistent tree topology with ten (A–J) recognisable clades (Fig. 3). The clades A to F have a moderate Bayesian support. A single specimen of L. timidus from eastern Russia was the earliest to emerge on the tree, forming its own clade J, followed by other specimens of L. timidus and L. t. tanaiticus from various geographical localities, scattered from clade B to clade I. The clade B also includes specimens of L. othus, which is now restricted to westernmost Alaska, forming a separate sub-group (B1), while the rest of the specimens in the clade come from Russia, Norway, and Eastern Europe (B2–B3). In contrast to the tree based on CytB sequences, the North American L. arcticus forms a well-separated clade (A). The next three clades (C, D, and E) are formed by individuals of the mountain hare from Eastern Europe and Russia (Fig. 3). The specimen of L. tanaiticus from Mizyn is placed within the clade D, while another one from Novgorod-Siverskyi forms its own clade F. Such a distribution suggests some distinctness between their sequences. Clade G includes four specimens of Pleistocene hares from all of the studied Ukrainian localities. One individual from Gintsi (identified as Lepus sp.) together with L. t. tanaiticus are nested in the clade H. The clade I includes the sequences of two individuals of L. timidus from Sweden (Fig. 3).

The haplotype network of the control-region sequences (Fig. 5) resembles the distribution of specimens in the phylogenetic tree. The haplotypes of L. othus form a separate cluster (Group 1) showing greater differences in sequences compared to Group 2, which consists of sequences of L. timidus and two those of L. t. tanaiticus sensu Prost et al. (2010). Group 2 is connected with a dispersed series of haplotypes of L. timidus, L. tanaiticus from Mezhyrich, Mizyn, and Novgorod-Siverskyi, Lepus sp. from Gintsi, and three haplotypes of L. t. tanaiticus (Fig. 5). A separate group of haplotypes is formed by the sequences of L. arcticus (Group 3), although two specimens of this species fell outside this group. It is characterised by a high number of substitutions compared to other species.

Fig. 5
figure 5

Haplotype network of species of the genus Lepus based on the analysis of the control-region mtDNA sequences. Missing intermediates are indicated by black dots. Haplotypes indicated with asterisk belong to L. t. tanaiticus

Full size image

Discussion

Molecular genetic techniques, especially the analysis of ancient DNA extracted from collection specimens, have revolutionised taxonomic studies and allow clarifying the relationships between extinct and extant species. Fossil taxa have traditionally been erected based on detailed morphological descriptions, although the range of variation of those diagnostic traits often remains poorly understood, especially when the number of available specimens is limited. Palaeogenetic studies allow a deeper and more accurate analysis of phylogenetic relationships and thus have the potential to resolve contradicting morphology-based conclusions. It is especially relevant when estimating past taxonomic diversity of various geographical regions and epochs in the context of major extinction events. The Late Pleistocene glaciation had a major impact on the evolution of numerous mammalian groups, including the lagomorphs that had played an important ecological role in periglacial ecosystems, although genetic and morphological data often contradict when analysing the phylogenetic relationships among various members of this group (e.g., Prost et al. 2010; Rabiniak et al. 2023).

The Don hare (Lepus tanaiticus) — a Late Pleistocene hare of the Palaearctic — is one of those dubious taxa, the systematic position of which has been controversial and remains unresolved (Prost et al. 2010; Sharko et al. 2023). The Don hare went extinct in the Holocene and most of its remains have been reported from the territory of present-day Russia. A large sample of hares from the Late Pleistocene of Ukraine, however, was also identified by Rekovets (1985) as L. tanaiticus based on characters described by Gureev (1964). At the same time, Rekovets (1985) amended the diagnosis of L. tanaiticus by adding a few additional characters, in particular a shortened diastema of the lower jaw, a more elongated palate, and more strongly developed zygomatic processes of the upper jaw. Although Rekovets (1985) treated the Don hare as an independent species, he also highlighted the substantial morphological similarity between L. tanaiticus and L. timidus and hypothesised that they could be phylogenetically close or even conspecific.

Previous studies of mtDNA sequences of arctic hares have revealed that L. tanaiticus falls within the range of variation of the extant L. timidus and phylogenetic tree analyses do not support the monophyly of the Don hare (Prost et al. 2010; Sharko et al. 2023). In this study, the analysis of cytochrome b and control-region sequences of extant arctic hares and specimens assigned to L. tanaiticus, including those from Late Pleistocene localities of Ukraine and first analysed genetically, has provided similar results: the Don hare is scattered in the obtained phylogenetic trees and haplotype networks among specimens of other species. Based on the control-region sequences, only specimens of L. arcticus (distributed in northern Canada and Greenland) and L. othus (distributed in western Alaska) form monophyletic groups (see Fig. 3). At the same time, L. othus is closer to L. timidus, which can be a sign of more recent introgression between these two species in periods when the Bering land bridge opened.

Genes of L. timidus are present in European hare species because of repeated introgressions in the past (Alves et al. 2003; Melo-Ferreira et al. 2005, 2007, 2012; Fredsted et al. 2006; Thulin et al. 1997, 2006; Ferreira et al. 2021), but molecular genetic data do not support the idea of phylogenetic distinctness of L. tanaiticus. Instead, it seems more likely that the Don hare was a distinct morphotype of L. timidus, as Prost et al. (2010) suggested earlier.

It is now commonly accepted that one should be careful when inferring the evolutionary history of species solely based on a limited number of genetic markers. Complete mitochondrial and nuclear genome analyses are therefore considered to provide more reliable implications for mammalian population genetics and phylogeny (e.g., Moska et al. 2016; Westerman et al. 2016; Urantowka et al. 2017). Complete mtDNA sequences were obtained for a number of Palaearctic Lepus species (e.g., see Ding et al. 2016; Giannoulis et al. 2018; Sharko et al. 2023), whereas the number of studies of whole-genome sequences of hares, particularly of L. timidus, is limited (e.g., Marques et al. 2020; Michell et al. 2022). Our study provides new valuable mtDNA data that both support previous findings of genetic and taxonomic research and supply new, important genetic information for further phylogenetic studies of Lepus species.

Nonetheless, the phylogenetic position of L. tanaiticus as a separate morphotype rather than an independent species inferred from the analysis of mtDNA fragments is remarkably similar to that of some other Late Pleistocene morphology-based small-mammal taxa. Rekovets (1985) considered the Don hare a common representative of the Late Pleistocene periglacial fauna of Eastern Europe, together with Citellus severskensis Gromov, 1958, Citellus superciliosus (Cuvier, 1825), Dicrostonyx gulielmi (Sanford, 1870), and Ochotona spelaea (Owen, 1846). Molecular genetic studies, however, questioned the species status of several Late Pleistocene mammalian taxa that were described based on morphological traits. The most recent examples are the pika species O. spelaea (Rabiniak et al. 2023) and the vole species Microtus bifrons Jeannet and Fontana, 2015 (Nadachowski et al. 2023), which turned out to be ancient morphotypes of the extant species Ochotona pusilla (Pallas, 1769) and Microtus arvalis (Pallas, 1778), respectively.

The main feature of the Late Pleistocene was glaciation, which also affected biome shifts and substantial range dynamics. The periglacial tundra expanded to the south along with the geographical range of species associated with this biome, whereas the range of temperate species also shifted, and they survived the periods of glaciation in southern refugia (e.g., see Jones et al. 2020; Sommer 2020). Consequently, range expansion could have increased the geographical variation of northern species and the appearance of well-distinguished morphotypes with adaptations to the specific conditions of the vast tundra steppe (Lister et al. 1987; Stewart 1999; Meiri et al. 2013; Lagerholm et al. 2017; Rabiniak et al. 2023). As a result, ecomorphological traits have been given taxonomic significance by researchers and distinct morphotypes were recognised as independent taxa or separate subspecies. Although, Rabiniak et al. (2023) argue that defining separate subspecies based on morphological data cannot be sustained by phylogenetic analysis as they are too closely related to extant representatives. Respectively, we can expect that morphological characters that were used to describe the Don hare as a separate species are also overestimated and are within the range of geographical variation that characterised L. timidus during the Late Pleistocene.

In the light of recent findings, the taxonomic status of various representatives of the Late Pleistocene vertebrate fauna will likely be subject to revision, which, in turn, will have a crucial role in the estimates of the impact of the Quaternary extinction event on local and global biodiversity (Barnosky et al. 2004). For instance, it was already suggested that the Pleistocene/Holocene transition in the Crimea, Ukraine, was not as drastic in terms of faunal turnover as in other parts of Eastern Europe (Benecke 1999; Kovalchuk et al. 2021). It is therefore possible that the Quaternary extinction (at least of small mammals) occurred at a much smaller scale and mainly involved ancient morphotypes of extant species.