Introduction

The Rhinolophus hipposideros group is one of the numerous groups that divide the genus Rhinolophus Lacépède, 1799, the only genus of the chiropteran family Rhinolophidae. The group currently contains a single species, the lesser horseshoe bat, Rhinolophus hipposideros (André, 1797). Originally, it was defined by Andersen (1905) as the Rhinolophus midas group, comprising two species, R. hipposideros and R. midas Andersen, 1905. This definition of the group was based on a typical structure of the sella of the nose-leaf, bearing a very low and rounded off posterior connecting process, and an extremely narrow basioccipital bone of the skull, reported to be distinct in both characters from other groups of the genus Rhinolophus. Since Andersen (1918) joined the two species into one under the prior name R. hipposideros, this name was also transferred to the group name. The R. hipposideros group was then reported as a separate and monotypic unit within the genus by numerous followers, despite the variable numbers and contents of other groups considered (Allen, 1939; Ellerman and Morrison-Scott, 1951; Koopman, 1994; Horáček et al., 2000; Csorba et al., 2003; Simmons, 2005; Burgin, 2019; etc.).

Besides Andersen's (1905) original definition made on the simple comparison of a few morphological characters, justification of the determination of the R. hipposideros group within the genus Rhinolophus was supported by the results of additional analyses of morphometric data by Bogdanowicz (1992) and genetic data by Guillén Servent et al. (2003), Stoffberg et al. (2010), Foley et al. (2015), and Dool et al. (2016). The basal and very separate position of this group within the genus Rhinolophus was stressed by Guillén Servent et al. (2003), who suggested delimiting it into the subgenus Phyllorhina Leach, 1816.

Because the group currently consists of a single species, R. hipposideros, its intraspecific variation also represents the only variation detectable in the group. This bat is a typical faunal element of the western Palaearctic (Fig. 1), where it occurs in a broad belt of the Mediterranean and temperate zones of Europe, North Africa, and western Asia (Csorba et al., 2003; Gaisler, 2013; Burgin, 2019; Bendjeddou et al., 2022); its distribution range comprises the Mediterranean Maghreb (Morocco to Tripolitania), southern, western and central Europe (from Portugal, Ireland, and Germany to western and southern Ukraine, as well as the Balkans), numerous Mediterranean islands; the Levant, including Sinai; Anatolia; Crimea; the Caucasus region; Iran; Afghanistan; Kashmir; and West Turkestan. Moreover, R. hipposideros also marginally extends to the Afrotropics; it occurs in south-western Arabia, Eritrea, Djibouti, Ethiopia, and Sudan (Fig. 1). Within this broad range, the bat is considered a polytypic species; up to seven subspecies have been defined and recognised (Andersen, 1918; Ellerman and Morrison-Scott, 1951; Koopman, 1994). Although several attempts to analyse the intraspecific structure of R. hipposideros have been made, this issue is still considered unresolved (see Burgin, 2019).

Fig. 1.

Map of the distribution range of the Rhinolophus hipposideros group (pale grey; after numerous sources) and the localities of origin and grouping of the examined samples (some symbols can denote more sites); full squares indicate the samples examined in both molecular genetic and morphological comparisons, open squares indicate the samples examined in the morphological comparison only, and open circles the samples used in molecular genetic analysis only. Grey dots denote isolated records out of the known distribution range

Based on the body size, structure of the infraorbital region of the skull, and the presence and position of the small lower premolars, Andersen (1905, 1907, 1918) defined six subspecies in R. hipposideros, and these taxa have been listed as tentatively valid by various authors until present (see Ellerman and Morrison-Scott, 1951; Koopman, 1994; Horáček et al., 2000; Roer and Schober, 2001; Csorba et al., 2003; Simmons, 2005; Burgin, 2019). Csorba et al. (2003) and Burgin (2019) defined the distribution ranges of these subspecies as follows: R. h. hipposideros (André, 1797) [type locality (t.l.) Germany] in continental Europe north of the Alps, from the Netherlands to southern Ukraine; R. h. minutus (Montagu, 1808) [t.l. Wiltshire, England] in western Ireland and south-western Great Britain; R. h. minimus von Heuglin, 1861 [t.l. Kérén in den Bogosländern (= western Eritrea)] in Mediterranean Europe from Portugal to the Levant, including Sinai and the Mediterranean islands, and in western Arabia, southern Sudan, Eritrea, Djibouti, and central Ethiopia; R. h. midas Andersen, 1905 [t.l. Jask, Persian Gulf (southern Iran)] in western Asia, from the Caucasus, Transcaucasia, and northern Iraq, to southern Kazakhstan, western Kirghizstan, and Kashmir; R. h. majori Andersen, 1918 [t.l. Patrimonia, northern Corsica] in Corsica; and R. h. escalerae Andersen, 1918 [t.l. Mogador (= Essaouira), Ha-ha, Morocco] in the Mediterranean zone of north-western Africa.

With the exception of the latter two names, all of the above-mentioned forms were originally described as separate species that were, however, soon synonymised with R. hipposideros (Blasius, 1857; Peters, 1871; Dobson, 1876; Trouessart, 1879; Andersen 1904, 1918). Moreover, several authors have demonstrated the morphological inadequacies of these numerous subspecies and showed them to be difficult to identify, because the particular characters exhibit a variable occurrence in particular populations of the species (Miller, 1912; Grulich, 1949; Panouse, 1951; Saint Girons and Caubère, 1966; Felten et al., 1977; Palmeirim, 1990). Although such variability led to the description of additional taxa, both at the species and subspecies levels (currently invalid), see e.g., R. h. alpinus Koch, 1865 [t.l. the Alps], R. phasma Cabrera, 1904 [t.l. Madrid, Spain], R. h. vespa Laurent, 1937 [t.l. Korifla, Morocco], or R. moravicus Kostroň, 1943 [t.l. Ponikev and Kadeřín, Moravia (= Czech Republic)], the mosaic-like occurrence of traditional identification characters resulted in the abandonment of taxonomic division at small geographic scales (Corbet, 1978). The variability in various morphological characters was thus interpreted as an individual variation influenced by local environmental conditions rather than a result of phylogenetic separation (Saint Girons and Caubère, 1966; Palmeirim, 1990; Salinas-Ramos et al., 2021).

Felten et al. (1977) proposed the only revision of intraspecific taxonomy in R. hipposideros. Using an evaluation of the characters suggested by Andersen (1905, 1918) — body size, shape of rostrum, and size and position of premolars — Felten et al. (1977) delimited four population groups in the species and tentatively identified with the subspecies: R. h. hipposideros in Europe (including Corsica) and the Levant, R. h. minimus in northeastern Africa and Crete, R. h. midas in the Middle East from north-eastern Turkey to Afghanistan, and an unnamed form in the islands of the central Mediterranean (Sicily, Pantelleria) and in western Turkey [Felten et al. (1977) did not evaluate some populations, e.g., those of North Africa, British Isles, or West Turkestan]. This geographic division of morphotypes in R. hipposideros was revised only to a small extent and only to certain populations; the examined specimens of the Middle Eastern populations were found to fit the morphotypes defined by Felten et al. (1977) — see Benda et al. (2006) and Benda and Gaisler (2015). However, bats from Crete did not fit the morphological criteria that Felten et al. (1977) gave for R. h. minimus when a large set of samples was examined (Benda et al., 2009). In general, subsequent authors did not follow the conclusions that Felten et al. (1977) suggested regarding intraspecific relationships in R. hipposideros and their taxonomic arrangement.

Another type of evidence of the geographic variability in R. hipposideros was found and widely documented in karyotype (Zima et al., 1992; Zima, 2004; Volleth et al., 2013; Arslan and Zima, 2014; Kacprzyk et al., 2016); three chromosome races were described in R. hipposideros, (1) the populations of 2n = 54 from Ireland, Spain, Germany, and Switzerland; (2) 2n = 56 from Italy, Greece, Bulgaria, Czech Republic, and Slovakia; and (3) 2n = 58 from Jordan, Syria, Turkey, and Iran. These chromosome races thus seem to be geographically well-defined forms, with one living in the western part of Europe, another in the eastern part of Europe, and a third in the Middle East.

Molecular genetic analyses focused on intraspecific variation in R. hipposideros (Kůs, 2008; Dool et al., 2013; Shahabi et al., 2019) have shown — in both mitochondrial and nuclear markers — a split of the species into two main lineages, the western one comprising the Maghrebian and European populations (Maghreb, British Isles, central and southern Europe, Sardinia, Malta, and Crete) and the eastern one covering the Asian populations (Turkey, Cyprus, Levant, Iran, Tajikistan).

Burgin (2019) recently summarised the main message of the review presented above, although he did not propose a taxonomic synthesis revising the old intraspecific arrangement of R. hipposideros (suggested already by Andersen, 1905). However, the available data suggest this arrangement is untenable and the bat's intraspecific relationships need a profound revision. Thus, to identify the phylogenetic pattern in R. hipposideros, we carried out a morphological examination of a set of more than 270 museum specimens with the aim of defining the positions of particular populations from its whole distribution range. Simultaneously, we subjected a geographically representative subset of these specimens to a molecular genetic comparison. In addition, we compared the echolocation data from various parts of the species range. Results of these approaches are presented here, and we propose a revised view of the systematic relationships within the R. hipposideros group, including its taxonomic interpretation.

Materials and Methods

Results

Molecular Genetic Analysis

The resulting Cyt-b dataset comprised 81 sequences which were pruned to 54 unique haplotypes. The nuclear dataset comprised 47 ACOX, 63 BGN, 44 COPS, 11 ROGDI, and 47 STAT sequences that were pruned to 46 haplotypes. For other Rhinolophus species, we added 129 sequences from GenBank in total. The Cyt-b sequences contained 403 parsimony informative positions (35.73% of total length) and this marker showed a much larger genetic differentiation within Rhinolophus species than the nuclear markers (due to the faster mutation rate). The amount of parsimony informative positions in concatenated nuclear dataset was 386, i.e. 13.33% of its total length (for substitution models of mitochondrial and nuclear trees see  Supplementary Table S3 (01-AC-24-2-p-269-298-Supplement.pdf)).

The ML and BI tree of the nuclear dataset showed slightly different topologies, nonetheless, the different nodes had a low branch support. We showed the ML tree (Fig. 2). The genus Rhinolophus was divided into four well supported clades. Rhinolophus hipposideros formed a separate clade, however, its exact position remained unclear due to the low branch support of deep nodes. Other groups were: (1) pusillus group including the species R. shameli, R. pearsonii, and R. pusillus; (2) trifoliatus group including R. trifoliatus and R. luctus; and (3) Afro-Palaearctic clade that includes the species groups euryale, fumigatus, ferrumequinum, capensis, and landeri.

Fig. 2.

Maximum likelihood tree of reconstructed phylogenetic relationships of the Rhinolophus hipposideros group with selected species of the families Rhinolophidae and Hipposideridae based on the nuclear (left) and mitochondrial (right) datasets, respectively. Branch support values are shown by pie charts on the nodes

The phylogenetic trees obtained by both ML and BI analyses of the Cyt-b dataset showed slightly different topologies. The ML tree was fully resolved and had a higher branch support than the BI tree, therefore we showed the ML tree (Fig. 2). The tree topology of the genus Rhinolophus was different than the topology of the nuclear tree. Within Rhinolophus, R. hipposideros formed a separate branch with an uncertain position. Another separate branch led to the trifoliatus group including only R. luctus. The rest of all Rhinolophus species formed a third branch within the genus tree supported by SH-aLRT and aBayes. This clade comprised African, European, and Asian species including the supported species groups ferrumequinum, fumigatus, maclaudi, capensis, and landeri. Nevertheless, the relationships among these groups and other ungrouped species were not satisfactorily resolved and neither were the relationships between three major clades.

Intraspecifically, R. hipposideros split into two lineages in both nuclear and mitochondrial trees. The first lineage ranged from Morocco and Ireland through Central Europe, the Balkans, Levant, and Iran to Tajikistan and Ethiopia; the second lineage comprised samples from north-eastern Oman. In the nuclear tree, the first lineage was not internally branched, and therefore the lineage was genetically uniform through almost the whole R. hipposideros range, except Oman which formed the second lineage. In the Cyt-b tree, the first lineage was formed by samples from the majority of the R. hipposideros distribution range and was further divided into four well-supported sublineages: (1) Afro-European, comprising samples from the Maghreb (Morocco), Central Europe (Slovakia), and the Balkans (Albania, Bulgaria, Greece); (2) Ponto-Levantine, with the samples from Crimea, Rhodes, and the Levant (Cyprus, Syria, Lebanon, Jordan); (3) Central Asian, with the samples from Iran and Tajikistan; and (4) Ethiopian, which includes a single sample from Ethiopia. The relationships among the sublineages were not resolved due to the low branch support. The uncorrected p-distances on Cyt-b between the two lineages were 8.93–10.75% and between the sublineages 2.30–7.02% (Table 1). The resolution of the nuclear gene trees was in accordance with the genetic variation of each nuclear marker ( Supplementary Figs. S2–S6 (01-AC-24-2-p-269-298-Supplement.pdf)). However, the basic split of R. hipposideros into two lineages was evident in all gene trees for which we obtained sequences from both lineages.

Table 1.

Percentage values of uncorrected genetic p-distances of Cyt-b among mitochondrial subgroups (lineage/sublineage) of the Rhinolophus hipposideros group (below the diagonal). The diagonal corresponds to the within-group genetic divergence estimated for Cyt-b in each subgroup

For the R. hipposideros only mitochondrial tree ( Supplementary Fig. S1 (01-AC-24-2-p-269-298-Supplement.pdf)), we added 155 Cyt-b sequences from GenBank to make a dataset of 213 sequences with the total length of 1,103 bp ( Supplementary Table S1B (01-AC-24-2-p-269-298-Supplement.pdf)). In this tree, five basic lineages in R. hipposideros were recovered. It corresponded to the topology with the above division based on the whole Cyt-b gene, and it covered almost a complete distribution range of the species; viz. (1) the Afro-European lineage, comprising sequences from the Maghreb (Morocco, Tunisia), Mediterranean Europe (Spain, Italy, France, Slovenia, Albania, Bulgaria, Greece, European Turkey), Mediterranean islands (Malta, Crete), British Isles (Ireland, Great Britain), and Central Europe (Austria, Slovakia, Romania); (2) Ponto-Levantine lineage, composed of the sequences from the Levant (Rhodes, western Anatolia, Cyprus, Syria, Lebanon, Israel, Jordan) and Crimea; (3) Eastern lineage comprising the sequences from the eastern part of the Middle East (eastern Anatolia, Iran) and West Turkestan (Tajikistan); (4) the Ethiopian lineage comprising one sequence from northern Ethiopia; and the last and most distant (5) Omani lineage from the sequences from north-eastern Oman. All five lineages had a high branch support (0.99–1.00 posterior probability [PP] and 97–100 bootstrap percentage [BP]). However, the relationships between the lineages did not always show high support, only the sister position of lineages 2 and 3 had marginal to moderate high support (0.82 PP and 93 BP), and the ML analysis supported the crown position of lineages 1–4 (98 BP). The uncorrected p-distances within lineages were 0–3.42%, between sublineages 2.30–7.02%. The Omani lineage differed from other lineages with the distances of 8.93–11.40%.

Our results of the Bayesian phylogenetics and phylogeography (BPP) analyses demonstrated the delimitation probabilities of the replicated runs being affected by the prior choice of parameters. It was especially apparent when a large effective population size was chosen in our pruned dataset ( Supplementary Table S4 (01-AC-24-2-p-269-298-Supplement.pdf)). Nevertheless, all the results for R. hipposideros and its populations had PP ≥ 0.95. It means that two clades, one from Oman and another from the rest of the distribution range, were strongly delimited within this lineage.

The topology of the calibrated tree (Fig. 3) showed the same four clades of the genus Rhinolophus as displayed by the topology of the nuclear ML/BI tree, however, their positions differed. The basal split occurred 37.8 Ma (95% highest posterior density [HPD]: 37.1–39.0 Ma) and divided the trifoliatus group from the rest of Rhinolophus species. A second split took place 32.0 Ma (95% HPD: 27.4–36.6 Ma) between the Indomalayan group and the Afro-Palaearctic group including R. hipposideros. Finally, the Afro-Palaearctic group diverged from R. hipposideros 29.4 Ma (95% HPD: 24.8–34.4 Ma). In the tree, all the nodes were statistically supported except three: between the Indomalayan group and the Afro-Palaearctic group including R. hipposideros; between the groups euryale and landeri (including only R. alcyone); and between R. fumigatus and R. hildebrandtii. In the R. hipposideros clade, two lineages used in our study split 7.1 Ma (95% HPD: 4.3–10.0 Ma). For the reconstruction based on a younger root calibration see  Supplementary Fig. S7 (01-AC-24-2-p-269-298-Supplement.pdf). The topology of both reconstructions remained identical, however, the splits of each group estimated in the alternative reconstruction occurred much later (16.7 Ma [16.1–17.4 Ma], 14.3 Ma [12.3–16.4 Ma], 13.2 Ma [11.0–15.3 Ma], and 3.2 Ma [2.0–4.9 Ma], respectively).

Fig. 3.

Chronogram of the family Rhinolophidae based on a Bayesian inference of the nuclear dataset (according to the model by Stoffberg et al., 2010). The numbers at nodes show mean divergence time estimates (Ma) and horizontal boxes 95% highest posterior density intervals of these estimates. The asterisk (*) indicates nodes with low branch support, the rest of the nodes were supported (PP ≥ 0.95)

Morphometric Comparison

In accordance with the geographic separation of lineages in the examined mitochondrial markers (see above) and the origin of the comparative material (Appendix I), all of examined material of R. hipposideros was sorted into six sample sets (Tables 2–4). The comparison of morphometric characters of the population sets documented a remarkable variation in the body, skull, and tooth sizes as well as in the skull and tooth shapes. In most dimensions, both in their absolute and relative values, the dimension ranges in particular sets overlapped with or exceeded the ranges of other sets. However, metric trends in the population sample sets were easily detectable from the comparison of the basic statistical values (Tables 2 and 3).

Regarding body size, two basic groups could be delimited among the examined samples, the large bats (x̄ LAt > 39 mm) from Central Europe and Central Asia, and the small bats (x̄ LAt < 38 mm) from the Mediterranean and Oman; a single sample from north-eastern Africa is medium-sized in this respect (LAt 38.7 mm).

Table 2.

External and cranial dimensions of the examined sample sets of the Rhinolophus hipposideros group (* — after DeBlase, 1980); midas, escalerae = dimensions of the respective type specimens; for the sample set delimitations and dimension abbreviations see Materials and Methods. Mean values shown in bold

Large skull size (x̄ LCc > 14.0 mm) was found in the bats from Central Europe; a small skull (x̄ LCc < 13.5 mm) was seen in the samples from the East Mediterranean, Oman, and north-eastern Africa; and a medium-sized skull (x̄ LCc 13.5–13.8 mm) was observed in the bats from the West Mediterranean and Central Asia (Fig. 4). An absolutely and relatively wide skull (x̄ LaZ > 7.4 mm; x̄ LaZ/LCc > 0.545) was observed in the bats from Central Europe, West Mediterranean, and Central Asia, whereas an absolutely and relatively narrow skull (x̄ LaZ < 7.4 mm; x̄ LaZ/LCc < 0.545) was seen in the bats from Oman and an absolutely narrow but relatively wide skull (x̄ LaZ < 7.4 mm; x̄ LaZ/LCc > 0.545) was found in the bats from the East Mediterranean and north-eastern Africa.

An absolutely and relatively wide braincase (x̄ LaNc > 6.5 mm; x̄ LaNc/LCc > 0.475) was found in the samples from the West Mediterranean and north-eastern Africa, whereas an absolutely and relatively narrow braincase (x̄ LaNc < 6.5 mm; x̄ LaNc/LCc < 0.475) was observed in the bats from Central Asia and Oman; an absolutely wide but relatively narrow braincase (x̄ LaNc > 6.5 mm; x̄ LaNc/LCc < 0.475) was observed in the bats from Central Europe; and an absolutely narrow but relatively wide braincase (x̄ LaNc < 6.5 mm; x̄ LaNc/LCc > 0.475) was seen in the bats from the East Mediterranean. Two shape types were found concerning the absolute and relative height of braincase; the bats from Central Europe, the Mediterranean, and Central Asia had a high braincase (x̄ ANc > 4.5 mm; x̄ Nc/LCc > 0.335), and the bats from north-eastern Africa and Oman had a low braincase (x̄ ANc < 4.5 mm; x̄ ANc/LCc < 0.335). An absolutely and relatively large tympanic bulla (LBT > 2.9 mm; LBT/LCc > 0.2) was observed in the bats from Oman, whereas a small bulla (LBT < 2.8 mm; LBT/LCc < 0.2) was seen in the bats from all other geographic sample sets (Fig. 5).

Table 3.

Dental dimensions of the examined sample sets of the Rhinolophus hipposideros group; midas = dimensions of the respective type specimen; for the sample set delimitations and dimension abbreviations see Materials and Methods. Mean values shown in bold

The rostral part of the skull was absolutely and relatively long (x̄ CM³ > 5.4 mm; x̄ CM³/LCc > 0.395) in the bats from Central Asia and Oman (Fig. 5); an absolutely long but relatively short rostrum (x̄ CM³ > 5.4 mm; x̄ CM³/LCc < 0.385) was observed in the bats from Central Europe, whereas an absolutely short but relatively long rostrum (x̄ CM³ < 5.4 mm; x̄ CM³/LCc > 0.385) was found in the bats from the Mediterranean and north-eastern Africa. A relatively narrow rostrum (x̄ LaInf/LCc < 0.262) was found in the bats from Central Europe and the Mediterranean; a relatively very wide rostrum (x̄ LaInf/LCc > 0.264) was observed in the bats from Oman; and a relatively medium-sized rostrum width (x̄ LaInf/LCc 0.262–0.264) was seen in the bats from Central Asia and north-eastern Africa.

Table 4.

Descriptive features of morphotypes of particular sample of the Rhinolophus hipposideros group (average states of absolute metric values are defined)

Fig. 4.

Bivariate plot of skull dimensions of the examined samples of the Rhinolophus hipposideros group: condylocanine length of skull (LCc) against length of the upper tooth-row (CM³); values in mm

Although the tooth metric characters largely followed the size trends in the skulls, certain shape variability and size trends were detectable in particular teeth. The largest upper canines (Cs; x̄ LaCs > 0.90 mm) were observed in the bats from Oman, the smallest (x̄ LaCs < 0.81 mm) were seen in the bats from the West Mediterranean and north-eastern Africa, and the upper canines (x̄ LaCs 0.82–0.85 mm) in the bats from Central Europe, the East Mediterranean, and Central Asia were medium size (Fig. 6). The small upper premolar (P²) was found to be large (LP² > 0.52 mm; x̄ LP²×LaP² > 0.25 mm²) in the bats from Central Europe, small (x̄ LP² < 0.42 mm; x̄ LP²×LaP² < 0.20 mm²) in the bats from Oman, and medium-sized (x̄ LP² 0.46–0.50 mm; x̄ LP²×LaP² 0.20–0.25 mm²) in the bats from the Mediterranean, Central Asia, and north-eastern Africa. The large upper premolar (P⁴) was found to be large (x̄ LP⁴1 > 0.98 mm) in the bats from Central Europe and Oman, small (x̄ LP⁴1 < 0.95 mm) in the bats from the East Mediterranean and northeastern Africa, and medium-sized (x̄ LP⁴1 0.95–0.98 mm) in the bats from the West Mediterranean and Central Asia; P⁴ was relatively wide (x̄ LaP⁴/LP⁴1 > 1.6) in the bats from the East Mediterranean and Central Asia, and relatively narrow (x̄ LaP⁴/LP⁴1 < 1.6) in the bats from the remaining four sample sets; it was relatively long in its medial portion (i.e., with a smallest posterior concavity in the distal margin of talon; x̄ LP⁴2/LaP⁴ > 0.36) in the bats from north-eastern Africa, short (x̄ LP⁴2/LaP⁴ < 0.34) in the bats from the East Mediterranean and Central Asia, and medium length (x̄ LP⁴2/LaP⁴ 0.34–035) in the bats from Central Europe, the West Mediterranean, and Oman.

Fig. 5.

Bivariate plot of skull dimensions of the examined samples of the Rhinolophus hipposideros group: relative length of rostrum (CM³/LCc) against relative horizontal length of tympanic bulla (LBT/LCc)

The first upper molar (M¹) was found to be large (x̄ LM¹ > 1.4 mm; x̄ LM¹×LaM¹ > 2.7 mm²) in the bats from Central Europe and Central Asia, small (x̄ LM¹ < 1.3 mm; x̄ LM¹×LaM¹ < 2.5 mm²) in the bats from north-eastern Africa, and medium-sized (x̄ LM¹ 1.35–1.38 mm; x̄ LM¹×LaM¹ 2.5–2.7 mm²) in the bats from the Mediterranean and Oman; M¹ was relatively wide (x̄ LaM¹/LM¹ > 1.4) in the bats from Central Europe, Central Asia, and northeastern Africa, relatively narrow (x̄ LaM¹/LM¹ 1.36) in the samples from Oman, and medium width (x̄ LaM¹/LM¹ 1.39–1.40) in the bats from the Mediterranean. The third upper molar (M³) was large (x̄ LM³ > 1.1 mm; x̄ LM³×LaM³ > 1.5 mm²) in the bats from Central Asia and Oman and small (x̄ LM³ < 1.1 mm; x̄ LM³×LaM³ < 1.5 mm²) in the bats from Central Europe, the Mediterranean, and north-eastern Africa; M³ was relatively wide (x̄ LaM³/LM³ > 1.4) in the bats from Central Europe, relatively narrow (x̄ LaM³/LM³ <1.3) in the samples from the West Mediterranean, Central Asia, and Oman, and medium width (x̄ LaM³/LM³ 1.3–1.4) in the bats from the East Mediterranean and north-eastern Africa. In relation to M¹, M³ was found to be large (x̄ LM³×LaM³/LM¹×LaM¹ > 0.6) in the bats from Oman, small (x̄ LM³×LaM³/LM¹ ×LaM¹ < 0.5) in the bats from Central Europe, and medium-sized (x̄ LM³×LaM³/LM¹×LaM¹ 0.5–0.6) in the bats from the remaining four sample sets (Fig. 6).

The lower canine (Ci) was observed to be large (x̄ LCi > 0.7 mm) in the bats from Central Europe and Central Asia and small (x̄ LCi < 0.6 mm) in the remaining four sample sets; in relation to the first lower molar (Mi), the Ci was relatively large (x̄ LCi/LMi > 0.5) in the bats from Central Europe, the West Mediterranean, Central Asia, and northeastern Africa, and relatively small (x̄ LCi/LMi < 0.5 mm) in the bats from the East Mediterranean and Oman. The first lower premolar (P₂) was large (x̄ LaP₂ > 0.52 mm; x̄ LP₂×LaP₂ > 0.3 mm²) in the bats from Central Europe and the Mediterranean, small (x̄ LaP₂ < 0.48 mm; x̄ LP₂×LaP₂ < 0.27 mm²) in the bats from Oman, and medium-sized (x̄ LaP₂ 0.48–0.52 mm; x̄ LP₂×LaP₂ 0.28–0.30 mm²) in the bats from Central Asia and north-eastern Africa. In relation to the last lower premolar (P₄), P₂ was very small (x̄ LP₂×LaP₂/LP₄×LaP₄ < 0.53) in the bats from Oman. In all other sample sets, this tooth was found to be large or very large (x̄ LP₂×LaP₂/LP₄ ×LaP₄ > 0.58). The last lower premolar (P₄) was large (x̄ LaP₄ > 0.75 mm; x̄ LP₄×LaP₄ > 0.48 mm²) in the bats from Central Europe, Central Asia, and Oman and small (x̄ LaP₄ < 0.75 mm; x̄ LP₄×LaP₄ > 0.48 mm²) in the bats from the Mediterranean and north-eastern Africa. The small lower premolar (P₃) was found to be large (x̄ LP₃ > 0.23 mm) in the Omani samples, but small (x̄ LP₃ < 0.23 mm) in all other sample sets. The first lower molar (Mi) and the lower molar-row were large (x̄ LMi > 1.4 mm; (x̄ M₁M₃ > 3.95 mm) in the bats from Central Asia and Oman and small (x̄ LMi <1.4 mm; x̄ M₁M₃ < 3.95 mm) in the four remaining sample sets.

Fig. 6.

Bivariate plot of tooth dimensions of the examined samples of the Rhinolophus hipposideros group: relative crown size of third upper molar (LM³×LaM³/LM¹×LaM¹) against crown size of upper canine (LCs×LaCs)

In summary, the comparison demonstrated certain characters were unique in four of the six examined sample sets (Tables 2–4 and  Supplementary Tables S5–S7 (01-AC-24-2-p-269-298-Supplement.pdf); see Appendix II for a review of the state conditions of evaluated metric characters in the particular sample sets). The Omani sample set was shown to be the most distinct among all bats (including the statistical comparison —  Supplementary Table S7 (01-AC-24-2-p-269-298-Supplement.pdf)). Within the matrix of 25 metric characters evaluated above, the Omani bats showed in ten characters a state to be unique in relation to all other sample sets: an absolutely and relatively very narrow skull with a relatively very wide rostrum and an absolutely and relatively very large tympanic bulla, very large upper canine, very small first (small) upper premolar (P²), relatively very narrow (palato-labially short) first upper molar (M¹) and a relatively very large third upper molar (M³), an absolutely as well as relatively very small first lower premolar (P₂), and a very large small-lower premolar (P₃). Six unique characters were found in the samples from Central Europe: a very large skull size with an absolutely wide but relatively narrow braincase and an absolutely long but relatively short rostrum, a very large first (small) upper premolar (P²), and a very small and relatively very wide third upper molar (M³). Only two unique characters were documented in the sample set from north-eastern Africa (large upper premolar (P⁴) being relatively narrow as a whole but relatively wide in its medial portion, and a very small first upper molar, M¹), and one was documented in the East Mediterranean set (an absolutely narrow but relatively wide braincase). No unique character among the evaluated metric traits was observed in the bats from the West Mediterranean and Central Asia.

The examined skulls of holotype specimens of two names of the R. hipposideros group (R. midas and R. h. escalerae) were compared with the above-defined morphotypes (Tables 2 and 3). The type of escalerae from western Morocco conforms in most characters to the Mediterranean populations, namely in the skull size and shape (i.e., the skull width, absolute and relative length and width of the rostrum, absolute and relative width and height of the braincase, and mandible length). The type of midas from southern Iran conforms in most respects to the bats from Oman. Similarities were found in all types of characters, in the skull size and shape and in the sizes and shapes of teeth. The type skull of R. midas is large with large tooth rows, although it is absolutely and relatively narrow; the braincase is very low; the rostrum is rather wide; and the tympanic bullae are very large. Although the Omani skulls are slightly smaller than the type skull of midas is in absolute dimensions, they well agree in the relative dimensions as well as in the absolute and relative size of tympanic bullae (Figs. 4 and 5). Even more pronounced than in the skull dimensions, the similarity of the midas type and Omani bats is apparent in the tooth dimensions. These bats are very similar in the extremely small size of the small upper premolar (P²), large absolute and mainly relative size of the last upper molar (M³), small relative and absolute size of the lower canine (Ci), small absolute and relative size of the first lower premolar (P₂), and large absolute size of the smallest lower premolar (P₃).

The separate position of the Omani samples plus the type specimen of midas in relation to all other sample sets is also illustrated by the results of the UPGMA cluster analysis (Fig. 7). Both the results calculated from the skull and tooth data showed similar positions of the Omani set together with midas positions in separate clusters, whereas the remaining sample sets from Central Europe, the Mediterranean, Central Asia, and north-eastern Africa in other clusters showed variable inner topology of particular sets. The results of a canonical analysis calculated from nine selected plain skull dimensions (LCc, LaZ, LaI, CM³, CC, M³M³, LBT, ACo, CM₃) and seven relative skull dimensions (LaZ/LCc, LaInf/LCc, LaN/LCc, LaM/LCc, ANc/LCc, LBT/LCc, CM³/LCc) conformed to the results of the empirical comparisons and cluster analysis ( Supplementary Fig. S8 (01-AC-24-2-p-269-298-Supplement.pdf) — CV1 46.37% of variance, CV2 32.62%). They clearly separated the Omani bats as the most distinct sample set (CV1 > 1.6; CV2 ≤ 6.5) in CV2 without an overlap with the four other sample sets, which, however, overlapped in both canonical variables with each other. Because only one skull was available from north-eastern Africa and the type series of escalerae and midas are composed only of holotypes, the relative positions of these three samples to other populations were not evaluated by the canonical analysis.

Fig. 7.

Results of the cluster analysis (UPGMA): differences between mean values of morphometric traits among the particular sets of samples of the Rhinolophus hipposideros group calculated from 27 plain and relative dimensions of skull (A), and from 36 plain and relative dimensions of teeth (B). Samples and sample sets: CEU — Central Europe; WMT — West Mediterranean; EMT — East Mediterranean; CAS — Central Asia; NEA — north-eastern Africa; OMA — Oman; mid — type specimen of R. midas Andersen, 1905

Comparison of Echolocation Call Parameters

The parameters of echolocation calls of R. hipposideros show similar values in the majority of characteristics throughout most of its distribution range, including central and southern Europe and south-western Asia, although the samples are rather small in some populations and the descriptive information level of the associated data could be limited (Table 5). Two exceptions among the population samples were found in the data from Malta and Oman (Mifsud and Vella, 2019; own data); whereas in the Maltese bats, the frequency values (start, end, peak) were reported to be much higher than in all other populations (≥ 115 kHz in all these parameters, without a value overlap with other populations), in the Omani bats, the peak frequency was found to be extremely low (< 101 kHz, without an overlap with the respective values from other populations; see Fig. 8).

Table 5.

Echolocation parameters in various populations of the Rhinolophus hipposideros group; based on published and original data. Abbreviations: SF = start frequency, EF = end frequency, PF = peak frequency (shown in bold), D = pulse duration, IPI = inter-pulse interval

Discussion

Our analysis uncovered the existence of an unexpected diversity within the R. hipposideros group, challenging its existing phylogenetic and taxonomic arrangement as concluded by Koopman (1994), Horáček et al. (2000), Csorba et al. (2003), Simmons (2005), or Burgin (2019). Genetic and morphological examinations of representative sets of specimens showed two main, geographically exclusive phylogenetic lineages within the group that are well delimited by molecular characteristics and possess two distinct morphotypes and two distinct echotypes. The genetic separation of the lineages is deep and detectable in both nuclear and mitochondrial genomes; in the Cyt-b gene, the uncorrected p-distance of 8.9–10.8% was found, which is roughly twice that considered sufficient for a taxonomic split (Baker and Bradley, 2006); this distance is even higher than those reported by Demos et al. (2019) for various species-pairs in Rhinolophus (or than the results for the species-pairs obtained here, e.g., on average 5.80% for capensis-swinnyi, 4.99% for euryale-mehelyi, 4.07% for ferrumequinumclivosus, or 3.65% for willardi-kahuzi). Thus, the degree of genetic separation of the two lineages within the hipposideros group is sufficient to allow us to consider them as two separate species.

Although one lineage/species was detected in the majority of the distribution range of the R. hipposideros group stretching across the whole southwestern Palaearctic (Europe, north-western and north-eastern Africa, north of the Middle East, Afghanistan, and West Turkestan), the other lineage/species was discovered in a very limited area in the north-eastern regions of Oman. The divergence of these two lineages is estimated to have occurred in the interval 4.3–10.0 Ma, when the more realistic (concerning the fossil evidence) estimation model of Stoffberg et al. (2010) is applied. This age approximately corresponds with the late Miocene period or with the Miocene-Pliocene transition (7.0–5.4 Ma — Herbert et al., 2016); that is, with the periods of dramatic environmental changes that could have led to the separation of species lineages. Alternatively, when Dool's et al. (2016) model is used, the estimated divergence occurred in the interval of 2.0–4.9 Ma, which is roughly at the Pliocene-Pleistocene transition (2.6 Ma; Gibbard et al., 2010) and linked with massive environmental changes as well. However, both time estimations mainly correspond to the main splits of species groups in the Afro-Palaearctic clade of the genus Rhinolophus and are associated with much older periods than most of the estimated divergences of crown pairs of species are within this clade (Stoffberg et al., 2010; Dool et al., 2016).

Fig. 8.

Spectrograms of echolocation call examples of the Rhinolophus hipposideros group (original data); A — an individual recorded inside the Aksamitka Cave, Slovakia; B — an individual foraging at Arjank, Iran (cf. Benda et al., 2012); C — a handled individual recorded at Zingrogh, Tajikistan; D — a handled individual recorded at Misfah, Oman. For details see Materials and Methods

The first, broadly distributed lineage/species can be easily identified with R. hipposideros (André, 1797) s.str. described from Germany, because only this genetic lineage was discovered in Europe. Based on genetic data, this species was confirmed to occur in the prevailing part of the range of the group as described by e.g., Horáček et al. (2000), Csorba et al. (2003), and Burgin (2019), with the exception of the Caucasus region, southern Iran, and southwestern Arabia.

The Omani lineage/species represents a recently discovered population of the lesser horseshoe bat (cf. Harrison and Bates, 1991; Horáček et al., 2000; Benda et al., 2013). It is known from just six localities (including those where only echolocation call recordings were made) in the Al Hajjar Mountains, situated between Sal Alah (26°02′N, 56°22′E) in the north and Tayman (22°31′N, 59°20′E) in the southeast (some 550–600 km in a line along the Al Hajjar range). Only four specimens were available for examination, which represent a morphotype very distinct from all other examined populations of the lesser horseshoe bat, typified by a very narrow skull with a relatively long rostrum, very large tympanic bullae, a very small first (small) upper premolar (P²), very large third upper molar (M³), very small first lower premolar (P₂), and very large second (small) lower premolar (P₃). Because the identical morphotype was also detected in the holotype specimen of R. midas Andersen, 1905, this name could also be applied for the Omani species/lineage, and the species has two geographic parts: Omani (known from four bats) and Iranian (known from the type specimen).

The type locality of R. midas is Jask, Hormozgan Province, Iran (25°40′N, 57°49′E), on the Iranian side of the Gulf of Oman, just opposite to the Al Hajjar Mountains of Oman. From the biogeographical perspective, the range of R. midas, lying in limited areas on both sides of the Gulf of Oman, is understandable. Similar geographical patterns of distribution range have been documented in other bats endemic to the Middle East (Harrison and Bates, 1991; Benda et al., 2012); namely, Rhinopoma muscatellum Thomas, 1903 (Rhinopomatidae) to a slightly larger geographical extent than in R. midas and Hypsugo arabicus (Harrison, 1979) (Vespertilionidae), with a distribution pattern very similar to R. midas. Besides R. midas, also R. hipposideros is distributed in Iran. However, according to the available records, it occurs in parapatry with R. midas, in uplands of the central and northern parts of the country (Benda et al., 2012). Only one record of the lesser horseshoe bat from Iran can be theoretically attributable to R. midas besides the type specimen, a bat observed in a cave on Qeshm Island in the Strait of Hormuz (Benda et al., 2012), only 106 km north-west of Sal Alah, the northernmost known site of this bat in Oman, and 250 km west–north-west of Jask, the type locality. The closest site of occurrence of R. hipposideros s.str. in Iran, confirmed by the genetic analysis (and the closest site as well), is the Tadovan Cave in the Zagros Mts. (1,190 m a.s.l.; Fars Prov., 28°51′N, 53°20′E — Shahabi et al., 2019), some 330 km NW of Qeshm Island.

Besides the genetic and morphological differences between R. midas and R. hipposideros s.str., the two species also differ in the pattern of their echolocation calls. Whereas in R. hipposideros s.str. the frequency of maximum energy (peak frequency, PF) of the call was detected around 110 kHz in most populations and only occasionally was it documented within the interval of 100–105 kHz (Benda et al., 2010; Győrössy et al., 2020), in R. midas, the PF was recorded in the interval of 94.1–100.6 kHz (x̄ = 98.2 kHz), i.e., at values much lower than in R. hipposideros s.str. Although no important differences in body size were found between R. hipposideros s.str. and R. midas, the latter species is of a similar size as the Mediterranean populations of the former species, a difference between the species could possibly be present in the auricle size as well as the size of the inner ear (i.e., in characteristics linked with the frequency value of the echolocation call — Huihua et al., 2003). The limited available data suggest such type of difference. The ear length in R. midas from Oman was 18.0–19.0 mm (x̄ = 18.5 mm), and in R. hipposideros s.str. from Iran it was 15.8–18.6 mm (x̄ = 17.1 mm — Benda et al., 2012), and in the bats from Lebanon 14.2–18.8 mm (x̄ = 17.4 mm ― Benda et al., 2016), whereas the forearm length in R. midas was 36.8–38.1 mm (x̄ = 37.4 mm), and in R. hipposideros s.str., it was 37.7–40.9 mm (x̄ = 39.0 mm) from Iran and 35.3–39.4 mm (x̄ = 37.7 mm) from Lebanon. Therefore, although the body size in R. midas is on average smaller than or similar to R. hiposideros s.str. from its geographically closest populations, the ear size seems to be on average larger in R. midas than it is in R. hipposideros s.str. (the species name midas also refers to the large ear size; it was selected by Andersen (1905) most probably after King Midas, a character from Greek mythology who had donkey ears). However, the differences in the external characteristics that would allow species identification remain to be found and tested; the currently available number of samples is too small for any conclusion. However, the size and shape of the nose-leaf of R. midas from Oman seems to be of identical parameters to those in R. hipposideros s.str. (see Fig. 9).

Hence, the R. hipposideros group (or the subgenus Phyllorhina Leach, 1816) now comprises two species, R. hipposideros and R. midas, identically as originally suggested by Andersen (1905) when he established the group. This author originally described R. midas as a separate species; later, he included it into the species rank of R. hipposideros (Andersen, 1918), and this unique morphotype, known from a single specimen until now, was for a long time overlooked. The midas morphotype seems to be rather conservative and perhaps more similar to the ancestral one because it exhibits a smaller degree of the reduction of distal molars and tiny premolars than known in the hipposideros s.str. morphotype.

The results of our analysis can also contribute to a revision of the intraspecific taxonomy of R. hipposideros s.str. Traditionally, the systematic reconstructions were based on body and skull size, nose-leaf shape, and several selected skull and tooth characters, an approach that resulted in delimitations of numerous taxa, namely in the Mediterranean area (see Introduction), a conception introduced by Andersen (1905) and accepted by many authors up to today (see Csorba et al., 2003; Simmons, 2005; Burgin, 2019). The molecular genetic analysis and broad evaluation of morphological characters brought a different view of the phylogenetic relationships within this species. The genetic analysis revealed the existence of two main genetic sublineages within the species, the western lineage, comprising most populations of Europe, including the British Isles, Sardinia, Malta, and Crete, and the Maghreb, and the eastern lineage, comprising the populations of Asia, including Eastern Mediterranean islands (Rhodes and Cyprus), and of Crimea. Both mitochondrial and nuclear markers showed the single Ethiopian sample to be a part of the eastern lineage, although without support for the mtDNA results (this could be a consequence of relatively large geographical distance between localities of the samples from Ethiopia and the Levant). However, the limited samples from West Turkestan (Tajikistan) were placed differently in the topology of both marker types, either into the eastern lineage (mtDNA) or into a separate lineage (nDNA) in a sister position to the above grouping. However, the West Turkestani samples are very limited and their localities are geographically extremely distant from the remaining analysed samples (the direct distance between the Tajikistani and central Iranian localities is some 1,800 km, across deserts and high mountains). Thus, the phylogenetic position of the easternmost populations of R. hipposideros s.str. remains to be investigated more elaborately, employing materials from all parts of Iran and West Turkestan, and from Afghanistan and Kashmir.

Fig. 9.

Portraits of Rhinolophus midas Andersen, 1905 from Oman; A, B — ♂ (NMP 93782), Misfah, Ad Dakhiliyah Province, 9 April 2011, lateral and frontal views; C — ♀ (NMP 93994), Sal Alah, Masandam Province, 13 March 2012, lateral view. Photos by A. Reiter

Although the geographic division to the western and eastern sublineages was not statistically supported by our results, it conforms to the results of previous analyses (Kůs, 2008; Dool et al., 2013), and it is additionally supported by karyological evidence. The geographical boundary between the lineages seems to be localised at the European-Asian transition between the Balkans and Anatolia, and from this location the boundary between the ranges of the 56- and 58-chromosome races is also reported (Zima et al., 1992; Zima, 2004; Arslan and Zima, 2014). Hence, the separation of the two lineages could actually be linked to the phylogenetic history of the species.

However, the morphological evidence did not contribute markedly to the reconstruction of the intraspecific relationships within R. hipposideros s.str. Two main morphological trends could be demonstrated from the data evaluated: (1) the increase of the body and skull size among the populations along the geographical gradient (latitudinal from the south to the north in the western part of the range, longitudinal from the Mediterranean to the continental climatic zone in the east) and (2) a mosaic-like distribution of characters among populations. The two population sets from the Mediterranean Basin (WMT and EMT) are the most similar to each other in the absolute and relative metric characters, although they belong to two separate sublineages. However, the most distinct population of R. hipposideros s.str. in morphometric traits is that of Central Europe. The representatives of the latter population are on average the largest in body size because they originate from the northernmost area of the species occurrence, although they represent a part of the western sublineage. The populations of the western sublineage share identical haplotypes of the mtDNA despite enormous geographical distances between them (e.g., one universal haplotype was found in Ireland, Great Britain, France, Italy, Austria, Slovenia, Slovakia, Bulgaria, and Greece). This haplotype arrangement suggests relatively recent dispersions of populations across the southern part of Europe and thus a relatively fast evolution of very distinct morphotypes.

The size differences among morphotypes of R. hipposideros may correlate with the changes of climatic conditions along a geographical gradient, in accordance with Bergmann's rule (Bergmann, 1847), that are expected to affect also bat populations (Ashton et al., 2000). On a smaller geographic scale, Salinas-Ramos et al. (2021) recently demonstrated a similar size shift in Italy along approximately 1000 km of the south-north gradient; these authors also explained that it aligned with Bergmann's rule. Other environmental influences that could be responsible for the geography-associated shift in body size, such as the character displacement (cf. Grant, 1972), do not seem to be significant in this bat species. All the evaluated populations come from regions where at least three size categories of horseshoe bats can be observed (i.e., Mediterranean, Central Europe, Central Asia, and north-eastern Africa); therefore, no effects from interspecific competition within the genus and no morphometric or other deflections in particular species were observed (see e.g., Andreas et al., 2013). If the character displacement really influenced the morphometry in R. hipposideros s.str., it would be primarily observed in the British Isles, where only two horseshoe bat species live in sympatry (the medium-sized category is missing). However, the body size of R. hipposideros on these islands is smaller than of the bats in Central Europe (Andersen, 1905; Miller, 1912) where three Rhinolophus species occur and where R. hipposideros would be much smaller if the character displacement works there. The medium body size of the British bats (in relation to the Mediterranean and Central European ones) is most likely caused by the islands' milder climate compared to Central Europe and harsher climate compared to the Mediterranean.

The Central European morphotype is the most distinct within R. hipposideros s.str. because of its extremely large skull size with a relatively narrow braincase and short rostrum, very large first upper premolar (P²), and very small and relatively wide third upper molar (M³). However, these differences seem to be a consequence of the allometric size changes of the skull, where the skull is enlarged in length (mainly the braincase), but is not enlarged to the same degree in width and in tooth-row length; the distal molars are enlarged less than the mesial ones and are relatively short (i.e., seem to be more reduced in length) but are not narrow.

The size differences along the geographical latitude from the Mediterranean to Central Europe were first discussed by Andersen (1905, 1907), who distinguished two subspecies at two edges of this gradient: R. h. hipposideros in the north and R. h. minimus in the south. However, this conception was revised when Saint Girons and Caubère (1966) and Felten et al. (1977) demonstrated a cline changes in metric traits, although Miller (1912) had already considered it to be rather dubious. Our results also give no support for such type of taxonomic division. Already Andersen (1918) demonstrated the mosaic-like distribution of morphological characters among populations of R. hipposideros s.str. in Europe and the Mediterranean and suggested the existence of six separate taxa within this species in the area between Morocco and Ireland in the west and Turkey and Cyprus in the east. This character distribution was again evaluated by Felten et al. (1977), who did not support such division and rather suggested only one, nominotypical subspecies existed in the whole area (except for Crete and Sicily). The echolocation data (another type of evidence) also showed a certain character plasticity within R. hipposideros populations in the Mediterranean area; bats living on the islands of Sardinia and Malta exhibited much higher values of call frequencies (up to 117 kHz on average — Russo et al., 2007; Mifsud and Vella, 2019) than the bats on the European continent. However, these insular bats represent an inner part of the western lineage of R. hipposideros s.str. and do not exhibit any substantial genetic differences from other populations of the lineage (Dool et al., 2013).

The documented pattern of morphological and morphometrical variability in R. hipposideros s.str. does not help when evaluating phylogenetic relationships among examined populations and the echolocation data show similar relevance when assessing the intraspecific variations in this bat species. Therefore, the results of the molecular genetic analysis remain the only evidence that support the reconstruction of the phylogenetic relationships within this species enough. Splitting the species content into two sublineages for both the nuclear and the mitochondrial genomes represents a well-detected separation event. Therefore, the sublineages could be co-identified with two subspecies. In both sublineages, similar levels of plasticity in morphological characters and similar character diversities in echolocation parameters were ascertained.

The taxonomic affiliation of the western sublineage that occurs throughout most of Europe and in the Maghreb is clear. The species is described from Germany (André, 1797) and therefore, this sublineage must to be identified with the nominotypical subspecies. The majority of the available names for R. hipposideros were proposed based on specimens from European type localities, situated in the contemporary countries of Spain, England, Germany, France, Corsica (France), Switzerland, Austria, Czech Republic, and Romania (minor Geoffroy, 1803, minutus Montagu, 1808, bihastatus Geoffroy, 1813, bifer de Blainville, 1840, alpinus Koch, 1865, pallidus Koch, 1865, typus Koch, 1865, kisnyiresiensis Daday, 1885, troglophilus Daday, 1887, helvetica Bretscher, 1904, phasma Cabrera, 1904, typicus Andersen, 1905, majori Andersen, 1918, anomalus Söderlund, 1921, intermedius Söderlund, 1921, moravicus Kostroň, 1943). Therefore, all of them should be considered junior synonyms of the nominotypical subspecies, R. hipposideros hipposideros. Two additional names were created based on bats from Morocco: escalerae Andersen, 1918 and vespa Laurent, 1937. Since the Maghrebian populations are a part of the western sublineage, these two names belong among junior synonyms of R. h. hipposideros. As summarised in Introduction, in the distribution range of the western sublineage (= R. h. hipposideros), up to six different subspecies were reported to occur (escalerae, hipposideros, majori, minutus, minimus, vespa — see Andersen, 1918; Ellerman and Morrison-Scott, 1951; Koopman, 1994; Csorba et al., 2003; Simmons, 2005; Burgin, 2019). However, this arrangement is rejected here because we did not find supporting evidence for it in our results, similar to the results by Dool et al. (2013).

The eastern sublineage of R. hipposideros s.str. is distributed in the Asian range of the species, including the Levant, Asia Minor (including adjacent islands), Crimea, and Iran (except for the Persian Gulf coastal areas). The affiliations of the populations from the eastern parts of the species distribution range (i.e., West Turkestan, Afghanistan, Kashmir) to this sublineage has not been fully resolved. Traditional taxonomic views divided this range into two parts according to the body size: the small-sized Levantine and Turkish populations were assigned to the Mediterranean taxon R. h. minimus (Ellerman and Morrison-Scott, 1951; Harrison, 1964; Koopman, 1994; Csorba et al., 2003; Burgin, 2019) or R. h. hipposideros (Felten et al., 1977; Corbet, 1978), whereas the large-sized eastern populations were assigned to R. h. midas (Andersen, 1905, 1918; Ellerman and Morrison-Scott, 1951; Harrison, 1964; Corbet, 1978; DeBlase, 1980; Harrison and Bates, 1991; Koopman, 1994; Horáček et al., 2000; Csorba et al., 2003; Benda et al., 2012; Burgin, 2019; see also Benda et al. (2006) for a more detailed review). However, our results do not support such west-east separation within the eastern sublineage (see also Dool et al., 2013).

As we demonstrated above, the name midas Andersen, 1905 is unavailable for designation of the Middle Eastern populations of R. hipposideros s.str. because this name is assigned to a different species. Interestingly, in contrast to the western part of the species range of R. hipposideros with 19 available names (see above), no synonym of this species name is currently available based on the material from Asia. However, the single Ethiopian sample examined in our analysis was shown to be a part of the eastern sublineage and it originates from the Yohannis Maikudi Church (13°51′N, 39°27′E) at Degum, Tigray State, approximately 240 km south–south-east of Keren, Eritrea, the type locality of R. minimus von Heuglin, 1861. Therefore, our Ethiopian sample could serve as a reference for topotype population of the latter name, which could be used as R. h. minimus for the eastern sublineage. This name was originally attributed to a separate species by von Heuglin (1861), but was rather early included into the species rank of R. hipposideros by Peters (1871). Andersen (1905, 1907, 1918) used this name for the small-sized Mediterranean populations of the species, but this conception was later questioned (Grulich, 1949; Saint Girons and Caubère, 1966; Felten et al., 1977; Corbet, 1978; Palmeirim, 1990; Benda et al., 2012) and is not supported by our results or the results by Dool et al. (2013). Thus, we consider the name minimus von Heuglin, 1861 to be unavailable for the European and/or Maghrebian populations of R. hipposideros s.str., although numerous recent authors applied this name in a way identical to Andersen's (1905) view (Koopman, 1994; Horáček et al., 2000; Roer and Schober, 2001; Csorba et al., 2003; Simmons, 2005; Burgin, 2019).

As already indicated, the populations of R. hipposideros s.str. that occur in the high mountains of the eastern margin of the species distribution range (Tian Shan, Pamir-Alai, Pamir, Hindu Kush, Karakoram) have an unresolved systematic position because only one specimen was examined for both types of genetic markers. These populations could be a part of the eastern sublineage (R. h. minimus), which is supported by the results of our mitochondrial marker analysis (contra Dool et al., 2013). Alternatively, they could pertain to a separate lineage of the species, as the nuclear markers show (again, contra Dool et al., 2013), and could represent a taxon of their own. In that case, no name would be available for such taxon/populations and it remains to be created (cf. Bates and Harrison, 1997; Csorba et al., 2003).

Samples of two populations of R. hipposideros, from the Caucasus region and from the southwestern part of Arabia, which are important from a biogeographical point of view, were not included in our analysis. Harrison and Bates (1991) identified R. h. minimus in the latter region; whereas from the Caucasus, two forms were reported diversely, R. h. hipposideros (Ognev, 1927; Strelkov, 1963; Kuzâkin, 1965; Koopman, 1994; Roer and Schober, 2001) or R. h. midas (Horáček et al., 2000; Csorba et al., 2003; Rahmatulina, 2005). However, based on the available data, we can estimate that both populations are affiliated with the eastern sublineage. The Caucasus region is situated in a space bordered by Crimea in the northwest and Iran in the south-east. In both of these border regions, the eastern sublineage was detected. Similarly, the south-western region of Arabia is bordered by the Levant in the north and the Ethiopian Highlands in the south, where the eastern sublineage was also detected. Therefore, it is most probable that these two populations belong to R. h. minimus.

Two additional names appeared in literature among synonyms of R. hipposideros (see e.g., Corbet, 1978, 1984; Csorba et al., 2003; Simmons, 2005) — eggenhoeffner Fitzinger, 1870 and billanjani DeBlase, 1972. However, the revisions of original sources showed both names unavailable for zoological nomenclature, being manuscript names (see Fitzinger, 1870; Miller, 1912; DeBlase, 1972, 1980; Benda et al., 2012).

To conclude, the revised taxonomic arrangement of the R. hipposideros group differs greatly from the most frequently presented views in recent years (see Csorba et al., 2003; Simmons, 2005; Burgin, 2019). The group consists of two species: R. hipposideros, which is widespread over south-western Eurasia and north-western and north-eastern Africa, and R. midas, which is distributed in a small range around the Strait of Hormuz and Gulf of Oman. These two species differ from each other in their morphological, genetic, and echolocation parameters. The extensive range of R. hipposideros s.str. is at least inhabited by two subspecies: R. h. hipposideros in the Maghreb and in Europe, west of the Dnieper River (cf. Zagorodniuk, 1999), Bosporus, and the Strait of Karpathos, and R. h. minimus east of this boundary, including the populations of Crimea, (Caucasus), the Middle East and north-eastern Africa (Sudan, Eritrea, Djibouti, Ethiopia). Besides genetic traits, these two subspecies also differ from each other in karyotype: 2n = 58 was found in R. h. minimus, and 2n = 54–56 was found in R. h. hipposideros. However, no significant morphological differences were found between the two subspecies of R. hipposideros.

Supplementary Information

Contents: Supplementary Figures:  Fig. S1 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of reconstructed phylogenetic relationships of the R. hipposideros group based on as complete as possible cytochrome-b dataset (1,103 bp). Branch support values are shown at the nodes;  Fig. S2 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of the reconstructed phylogenetic relationships of the R. hipposideros group and selected species of the genus Rhinolophus based on ACOX. Branch support values are shown above/below the branches in order SH-aLRT/UFBoot;  Fig. S3 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of the reconstructed phylogenetic relationships of the R. hipposideros group and selected species of the genus Rhinolophus based on BGN. Branch support values are shown above/below the branches in order SH-aLRT/UFBoot;  Fig. S4 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of the reconstructed phylogenetic relationships of the R. hipposideros group and selected species of the genus Rhinolophus based on COPS. Branch support values are shown above/below the branches in order SH-aLRT/UFBoot;  Fig. S5 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of the reconstructed phylogenetic relationships of the R. hipposideros group and selected species of the genus Rhinolophus based on ROGDI. Branch support values are shown above/below the branches in order SH-aLRT/UFBoot;  Fig. S6 (01-AC-24-2-p-269-298-Supplement.pdf). Maximum likelihood tree of the reconstructed phylogenetic relationships of the R. hipposideros group and selected species of the genus Rhinolophus based on STAT. Branch support values are shown above/below the branches in order SH-aLRT/UFBoot;  Fig. S7 (01-AC-24-2-p-269-298-Supplement.pdf). Chronogram of the family Rhinolophidae based on a Bayesian inference of the nuclear dataset (according to the model by Dool et al., 2016). The numbers at nodes show mean divergence time estimates (Ma) and horizontal boxes 95% highest posterior density intervals of these estimates. The asterisk (*) indicates nodes with low branch support, the rest of the nodes were supported (PP ≥ 0.95);  Fig. S8 (01-AC-24-2-p-269-298-Supplement.pdf). Bivariate plot of skull dimensions of the examined samples of the R. hipposideros group: results of the canonical discriminant analysis of selected nine plain and seven relative dimensions (see Results for details).  Supplementary Tables: Table S1. A (01-AC-24-2-p-269-298-Supplement.pdf)) Original sequences and sequences from GenBank used in the molecular genetic analysis; B) Sequences with the total length of 1,103 bp from GenBank used for the R. hipposideros tree, see  Supplementary Fig. S1; Table S2 (01-AC-24-2-p-269-298-Supplement.pdf). Names, sequences, and annealing temperatures of primers used in this study;  Table S3 (01-AC-24-2-p-269-298-Supplement.pdf). Substitution models as identified by ModelFinder for the different partitions used in MrBayes and IQTREE, respectively;  Table S4 (01-AC-24-2-p-269-298-Supplement.pdf). Summary of BPP for the nuclear dataset. Values for BPP species are posterior probabilities (PP) of delimitation from BPP runs under each of four different schemes under two different algorithms (see Table 1 in Demos et al., 2019);  Table S5 (01-AC-24-2-p-269-298-Supplement.pdf). Relative cranial dimensions of the examined sample sets of the R. hipposideros group; midas, escalerae = dimensions of the respective type specimens; for the sample set delimitations and dimension abbreviations see Materials and Methods;  Table S6 (01-AC-24-2-p-269-298-Supplement.pdf). Relative dental dimensions of the examined sample sets of the R. hipposideros group; midas = dimensions of the respective type specimen; for the sample set delimitations and dimension abbreviations see Materials and Methods;  Table S7 (01-AC-24-2-p-269-298-Supplement.pdf). Results of the one-way ANOVA test of skull dimensions between particular sample sets; for the sample set delimitations and abbreviations and for dimension abbreviations see Materials and Methods. Supplementary Information is available exclusively on BioOne.