On the phylogenetic position of Rhinolophus sakejiensis (Chiroptera: Rhinolophidae)

3.1 Morphology and identification

The NMP specimen is a large horseshoe bat with the forearm length (LAt) 56.6 mm, the greatest length of skull (including the praemaxilla; LCr) 25.47 mm, condylocanine length (LCc) 21.86 mm, zygomatic width (LaZ) 13.39 mm, and the length of the upper tooth-row (CM³) 9.61 mm (for other skull dimensions see Benda et al. 2022a). The size of skull of the NMP bat conforms to those mentioned by Cotterill (2002) for the type series, its dimensions fall within the ranges of the type series: LCr 24.6–25.6 mm; LCc 21.7–22.3 mm; LaZ 12.9–13.5 mm; CM³ 9.4–9.7 mm (Cotterill 2002: 169, Table 3). In two skull width dimensions, the value ranges of the type specimens lie slightly below the values of the NMP specimen (mastoidal width 11.56 mm, vs. 11.0–11.5 mm; the rostrum width across the third upper molars 9.89 mm, vs. 9.2–9.5 mm). However, these differences are very small and most likely insignificant, regarding the small sample size of the type series and a possibility of personal bias in measuring between authors. Similarly, not all dimensions of one of the paratypes that is housed in the Harrison Zoological Museum, Sevenoaks, UK (HZM), which were taken by Csorba et al. (2003), fall within the ranges of the type series given by Cotterill (2002). The forearm length of the NMP specimen exceeds the range of the type series (52.5–55.2 mm) reported by Cotterill (2002), likely due to his slightly unusual way of the dimension taking (see Cotterill 2002: 166).

The general colouration of the NMP bat is pale (considering the alcohol fixed body), the pelage is beige to very pale brown, the wing membranes are pale brown, and the noseleaf and ears are creamish. Its noseleaf is broad over the horseshoe (it was not measured due to certain deformations caused by the fixation and skull extraction), the lancet is broad and short, the connecting process of sella is blunt and medium-high, the tip of sella is low and rounded (Figure 1). Generally, the noseleaf structure corresponds with that described by Cotterill (2002) and Csorba et al. (2003), only the height of connecting process is somewhat smaller in the NMP specimen than in the type series. The skull structure of the NMP specimen corresponds with that of the HZM paratype of R. sakejiensis as depicted by Csorba et al. (2003: 132). In the NMP bat, the skull is heavily built, the sagittal crest is very high, the infraorbital bridge is narrow and short (Figure 2); the nasal swellings are very low, the posterior swellings are flattened, the anterior swellings only slightly bulged, and the frontal depression is very shallow (Figure 2). In the upper tooth-row, only one premolar (P⁴) is present (while the small premolar, P², is absent), and the canine is high and narrow, with its distal cingulum in contact with that of the premolar (P⁴). In the lower tooth-row, only two large premolars are present and in full contact, while the small (middle) premolar is absent (Figure 3).

Figure 1:

Frontal view (partial reconstruction) of the noseleaf of Rhinolophus sakejiensis (NMP 97587). Scale bar: 5 mm.

Figure 2:

Skull in lateral view of Rhinolophus sakejiensis (NMP 97587). Scale bar: 5 mm.

Figure 3:

Occlusal views of the left upper tooth-row (C–M³) and the right lower toothrow (I₁–M₃) of Rhinolophus sakejiensis (NMP 97587). Scale bar: 2 mm.

In southern Africa, where 21–23 species of the genus Rhinolophus occur, only six horseshoe bat species of similar size to the NMP specimen are known (Monadjem et al. 2020: 200; Benda et al. 2024: 241), namely R. acrotis (LAt 48–58 mm), R. alcyone Temminck, 1852 (49–58 mm), R. cervenyi Benda, Uvizl, Eiseb et Avenant, 2024 (49–56 mm), R. deckenii Peters, 1868 (48–56 mm), R. eloquens Andersen, 1905 (53–59 mm), and R. fumigatus Rüppell, 1842 (48–59 mm). However, all these species differ from the NMP specimen by the presence of small premolars in both jaws. Additionally, in R. alcyone, the upper small premolar is placed within the tooth-row and the connecting process and tip of the sella are very prominent (but low in the NMP specimen); in R. alcyone, R. eloquens, and R. fumigatus, the nasal swellings of the skull are very developed and prominent (but low and undeveloped in the NMP bat); in R. eloquens and R. fumigatus, the connecting process of the sella is undeveloped, and the sella profile is rounded (but the connecting process is blunt but developed in the NMP bat); in R. deckenii, the noseleaf (in frontal view) is of a triangular shape (but narrow oval in the NMP bat); and in R. acrotis and R. cervenyi, the tip of the sella is pointed (but blunt in the NMP bat), and the canines are more massive and relatively low (but narrow and high in the NMP bat). As a result of revision of morphological traits and of employing of the identification keys by Csorba et al. (2003) and Monadjem et al. (2020), the identification of the NMP specimen as R. sakejiensis seems to be indisputable, as it shares all essential characters mentioned by these authors.

3.2 Phylogenetic position

The newly generated cytochrome b sequence of the NMP specimen of R. sakejiensis (GenBank Accession Number OQ224762) was supplemented with two new sequences of R. fumigatus, along with additional GenBank sequences, resulting in a final dataset comprising 132 sequences. The 1,057 bp-long dataset contained 574 parsimony informative positions (54.3 % of the total length). Attempts to further examine the phylogenetic relationships of R. sakejiensis using nuclear genetic markers were unsuccessful, likely due to degradation of the nuclear DNA, which led to failed amplification. Despite this restraint, the mitochondrial marker comparison allows an evaluation of phylogenetic position. However, it is necessary to consider it with certain limits, since the incongruity between mitochondrial and nuclear genetic evidence is relatively frequent in bats and even in the horseshoe bats of southern Africa (see e.g. Benda et al. 2024; Taylor et al. 2018). However, in the case of deep divergences like that discovered here, the phylogenetic positions based on both marker types are rather congruent in the horseshoe bats (see e.g., Benda et al. 2022b, Uvizl et al. 2024).

The phylogenetic tree generated by ML analysis of the cytochrome b dataset (Figure 4) positioned the NMP specimen of R. sakejiensis within the Afro-Palaearctic clade of the genus Rhinolophus, in a sister position to R. horaceki Benda and Vallo, (2012), and outside any recognised species group. Rhinolophus horaceki is a Palaearctic African bat with a relic distribution range in the Mediterranean arboreal zone of Cyrenaica, north-eastern Libya (Benda and Vallo 2012). Together, these two species form a sister lineage to two species-rich lineages: the R. fumigatus group (comprising numerous Afrotropical species) and the Rhinolophus ferrumequinum group (which includes both Afrotropical and Palaearctic species). This phylogenetic arrangement suggests that R. sakejiensis, like R. horaceki, does not belong to any recognised species group, but rather represents a distinct evolutionary lineage. This interpretation contrasts with the view presented by Demos et al. (2019) and also with the original views by Cotterill (2002) and Csorba et al. (2003), based solely on morphological comparison.

Figure 4:

Maximum likelihood tree of reconstructed phylogenetic relationships of Rhinolophus sakejiensis (NMP 97587) with species of the ferrumequinum and fumigatus groups and other Rhinolophus groups/clades based on the cytochrome b dataset. Branch support values ≥95 % UFBoot are shown by black dots on the nodes. The numbering of lineages follows those used by Uvizl et al. (2024) (in R. ferrumequinum and R. clivosus) and Benda et al. (2024) (in R. damarensis).

The divergence time estimation tree revealed wide confidence intervals (95 % highest posterior density [HPD]), likely due to relatively high mutation rates and the short length of the mitochondrial dataset. Despite this, the divergence of R. sakejiensis from R. horaceki was estimated to have occurred around 15.3 Ma (HPD: 10.0–17.5 Ma; Figure 5). The group composed of R. sakejiensis, R. horaceki, and additionally R. eloquens, another African small-range species, was estimated to have diverged from the lineage that comprised the ferrumequinum and fumigatus groups around 20.4 Ma (HPD: 16.2–24.9 Ma; Figure 5). These periods of the Lower and Middle Miocene were characterised by an increasing prevalence of grasslands and fluctuating forest distributions in Africa (Jacobs 2004), environmental changes that may have influenced bat speciation due to the fragmentation of their populations in suitable humid versus arid habitats.

Figure 5:

Chronogram of the family Rhinolophidae based on Bayesian inference of the mitochondrial dataset (following to the model by Álvarez-Carretero et al. 2021). Numbers at nodes indicate mean divergence time estimates (Ma) and pale grey horizontal boxes indicate the 95 % highest posterior density intervals of these estimates. The asterisk (*) indicates nodes with low branch support, while the rest of the nodes were supported (PP ≥ 0.95).

Overall, the divergence estimates suggest that R. sakejiensis represents a relic of an early radiation of horseshoe bats that took place in the African continent (Demos et al. 2019). This evolutionary radiation was likely followed by the emergence of the R. landeri, R. capensis, and R. euryale groups, which subsequently gave rise to the geographically widespread R. fumigatus and R. ferrumequinum groups (the groups definition after Burgin 2019). Alongside with species such as R. damarensis and R. horaceki in southern and northern Africa, respectively, and R. xinanzhogguoensis Zhou, Guillén-Servent, Lim, Eger, Wang et Jiang, 2009 in eastern Asia, R. sakejiensis remains as a relic of these ancient radiations, indicating a complex and fragmented evolutionary history of the Afro-Palaearctic lineage of the genus Rhinolophus.