Introduction

Evolutionary parallelism in bats is rather common due to constraints related to flight and echolocation (Norberg & Rayner 1987, Arita & Fenton 1997, Mayer & von Helversen 2001, Baker & Bradley 2006). True phylogenetic relationships between morphologically similar taxa may then be difficult to assess, which in turn can complicate the taxonomic arrangement of particular bat groups. Well-known examples of such a discrepancy are the Palaearctic vespertilionid bats of the Myotis mystacinus morphogroup (Ruedi & Mayer 2001, von Helversen et al. 2001, Stadelmann et al. 2004), which includes species that do not form a monophyletic assemblage and actually represent independent evolutionary units. The existence of morphological similarity without close evolutionary kinship was also revealed in the bat genus Scotophilus Leach, 1821 (Vespertilionidae), also known as house bats. This common bat group of tropical regions of Africa and Asia comprises taxa of similar morphological appearance, which makes their taxonomy a complicated affair (Robbins et al. 1985, Simmons 2005, Goodman et al. 2005, Horaček et al. 2006, Van Cakenberghe & Happold 2013a), despite recent advances achieved with the use of molecular genetic (Trujillo et al. 2009, Vallo et al. 2011, 2013, 2015). Two African species, S. dinganii (Smith, 1833) and S. viridis (Peters, 1852), originally defined just morphologically, were each shown to consist of three mitochondrial lineages in a polyphyletic arrangement (Trujillo et al. 2009), which implies species status for each of them (see Brooks & Bickham 2014).

The Robbins' — or “nutlet” — house bat Scotophilus nucella Robbins, 1983, is a rare and insufficiently known species of the African rainforest zone, separated three decades ago from the nut-coloured house bat S. nux Thomas, 1904. This separation was based on diminutive size of otherwise morphologically similar specimens of brownish colouration occurring in sympatry with the larger, traditionally distinguished species. While S. nux occurs in Guinean rainforest from Sierra Leone to Ghana and Congolean rainforest from Nigeria to Kenya, S. nucella is known from just a handful of records limited to Upper Guinean rainforest of southeastern Cote d'Ivoire and southern Ghana (Robbins et al. 1985, Simmons 2005, Van Cakenberghe & Happold 2013b, c; Fig. 1). Records from western Uganda (Robbins 1983; Fig. 1) and Tanzania (Cunneyworth 1996, Doggart et al. 1999a, b; Fig. 1) were challenged by Thorn & Kerbis Peterhans (2009) and Stanley & Goodman (2011), respectively. There exists only scarce information on natural history and taxonomic relationships of S. nucella. Most recently, it was reviewed by Decher & Fahr (2007), who succeeded in capturing ten specimens of this species in southwestern Ghana. Their catch represented half of the known specimens of S. nucella captured until then, including those upon which the species description by Robbins (1983) was based. In 2014 and 2015, six additional specimens were caught during a faunistic survey in the same locality sampled more than a decade ago by Decher & Fahr (2007). The current number of S. nucella specimens known to science thus raised to 31, including 26 from southwestern Ghana alone, of which 16 originated from a single locality.

Phylogenetic relationships of S. nucella within the genus Scotophilus have not yet been assessed, nor was its taxonomic status as a separate species supported by molecular genetic analysis. Moreover, given the polyphyly within the congeneric S. dinganii and S. viridis morphospecies (Trujillo et al. 2009), a question arises as to the actual evolutionary relationship between S. nucella and S. nux, which may have independently evolved into species adapted to life in tropical rainforest. To fill the gap in knowledge on this enigmatic species, the phylogenetic position of S. nucella within the genus Scotophilus is inferred here based on sequence analysis of maternally inherited mitochondrial cytochrome b gene (cytb) and the paternally inherited gene for the zinc finger protein on the Y-chromosome (zfy). In addition to the evaluation of the species status of S. nucella and its relationship to S. nux, the evolutionary history of both forest species is also briefly addressed.

Fig. 1.

Distribution of Scotophilus nucella shown as localities of capture of specimens analyzed in this study (white triangle) and previously published localities (black triangles) in Côte d'Ivoire, Ghana and Uganda, as summarized by Decher & Fahr (2007), and unrevised localities in Tanzania (Cunneyworth 1996, Doggart et al. 1999a, b). Localities of origin of the comparative samples of S. nux are also included (black circles), both new and previously published (Trujillo et al. 2009). The original extent of the Upper Guinean rainforest in Ghana as possible habitat for S. nucella is emphasized by shading (light grey), including the Krokosua Hills Forest Reserve (dark grey).

Material and Methods

Specimens of Scotophilus nucella included in this study originated from Krokosua Hills Forest Reserve (KHFR), southwestern Ghana (Table 1, Fig. 1). One was an earlier capture from a rapid assessment survey in 2003 (Decher & Fahr 2007). This specimen (ZFMK 2003.1044) was collected and preserved as ethanol voucher with skull extracted in the collection of the Zoological Research Museum Alexander Koenig (ZFMK) in Bonn, Germany. The other specimens were recent captures from 2014 and 2015 during a bat conservation survey, and except for one were released after measuring forearm and mass, and sampling a biopsy from the patagium. The sacrificed specimen (NMP 95434) is preserved as ethanol voucher with skull extracted in the collection of the National Museum in Prague, Czech Republic (NMP). Forearm length and cranial dimensions of this specimen were measured using a mechanical calliper as described in Vallo et al. (2011, 2013) for morphological comparison with published data on S. nucella. Seven specimens of the morphologically similar African forest species S. nux (see Robbins 1983) originating from the 2003 and 2015 surveys in Ghana, and one from a baseline study at Simandou Guinea in 2008 (Decher et al. 2015) were also included into the analysis (Table 1).

Table 1.

List of Scotophilus nucella and S. nux specimens from West Africa, included in this study. Except for S. nux ZFMK 2008.0306 from the Simandou Range, Guinea (08°32′ N, 08°54′ W), all samples originate from the Krokosua Hills Forest Reserve, Ghana (06°37′ N, 02°51′ W). acc. nr. — accession number of haplotype in Genbank database.

Fig. 2.

Bayesian cytb tree (left) and modified 50 %-majority consensus MP zfy tree (right) showing phylogenetic position of Scotophilus nucella within the genus Scotophilus. Newly obtained haplotypes of S. nux are emphasized in bold. Stars denote significant support at the respective nodes: Bayesian posterior probability of 1.00 and MP bootstrap support > 75 %.

Total genomic DNA was extracted from ethanolpreserved tissue samples (muscle or patagium) using commercial kits and the complete cytb and a fragment of zfy were PCR amplified, sequenced, and aligned as previously described (Vallo et al. 2011, 2013). Additional sequences of congeneric species published by Trujillo et al. (2009) and Vallo et al. (2011, 2013, 2015) were included into the alignment for comparison (Fig. 2). All new sequences were submitted to GenBank and can be accessed under numbers KX010973–KX010981 (Table 1). Sequences of three vespertilionids by Ruedi & Mayer (2001): Vespertilio murinus Linnaeus, 1758, Myotis nattereri (Kuhl, 1817) and Myotis schaubi Kormos, 1934, were used as outgroup for rooting phylogenetic trees.

Phylogenetic analysis of cytb sequences was carried out using Bayesian inference in MrBayes 3.1.2 (Ronquist & Huelsenbeck 2003) under GTR + I + Γ evolutionary model, as suggested by the program MrModeltest 2.3 (Nylander 2004). A consensus tree was computed from two independent runs of one million generations sampled each hundred generations with burn-in set to 25 %. Sequence divergences were expressed as pairwise Kimura two-parameter genetic distances (K2P; Kimura 1980) to allow comparison with other bat groups (Baker & Bradley 2006), and particularly with members of the genus Scotophilus (Trujillo et al. 2009, Vallo et al. 2011, 2013, 2015). Nuclear zfy sequences were analyzed using maximum parsimony (MP) in PAUP* 4.10b (Sinauer Associates, Sunderland, MA, U.S.A.) in two modes, with gaps coded either as missing characters or as the fifth state as described in Vallo et al. (2013, 2015).

Dating of evolutionary events leading to the origin of contemporary S. nucella based on cytb sequences was done in BEAST 1.8.2 (Drummond et al. 2012) under a relaxed clock model with uncorrelated lognormal rate distribution and Yule model of speciation. Four independent runs of ten million generations were sampled each thousand generations and burn-in was set to 10 %. After checking for consistent results the four runs were then combined into one. Due to lack of usable fossil evidence for the genus Scotophilus (Horaček et al. 2006), molecular clock was calibrated by the split of Myotis nattereri and M. schaubi, which has been used in previous bat studies (Hulva et al. 2004, Stadelmann et al. 2004). These two Myotis species were confirmed from early Pliocene of Central Europe, more than 5 million years ago (Mya) (Horaček & Hanak 1984). Since the timing of the actual splitting event in their common ancestor remains unknown, a time period between 7.2 and 5.3 Mya is used here as an approximate estimate. Within this period of late Miocene, the origin of these Myotis species may be plausibly placed due to geoclimatic changes resulting from desiccation of the Mediterranean sea, the socalled Messinian salinity crisis, which was a likely trigger for speciation in bats and other mammals (Hulva et al. 2004). The prior distribution for the time of the most recent common ancestor for this calibration node was thus set as normal with the mean of 6 Mya and standard deviation of 0.5 Mya, which covers the Messinian period of late Miocene and spreads up to the already known fossils.

Fig. 3.

Clock-like cytb tree showing major splitting events within Scotophilus leading to origin of contemporary African high forest species S. nucella and S. nux. Grey bands denote respective geological or geoclimatical periods that may have influenced the current phylogenetic and phylogeographic structure of the target taxa.

Results

The newly collected specimen of S. nucella (NMP 95434) was very similar in both morphological measurements and cytb sequence to the previously identified specimen from 2003 (ZFMK 2003.1044; Table 2, Fig. 2). The seven samples of Scotophilus nucella and the six of S. nux yielded each four cytb haplotypes 1007 bp long. All cytb haplotypes were coding for a protein of an essentially identical amino acid composition and contained no stop codons. In the cytb phylogenetic tree, S. nucella haplotypes were placed as a monophyletic lineage in a highly supported sister position to S. nux (Fig. 2). The haplotypes of S. nucella were closely related, differing by up to 0.6 %. The southeast Asian species S. kuhlii was the first taxon to diverge from the basal node. The clade of sister species S. nucella and S. nux branched off next from the common ancestor with the remaining African and Asian Scotophilus taxa. Genetic divergence between S. nucella and its sister species S. nux reached 7.8–9.4 %, and between S. nucella and other Scotophilus species 11.6–14.1 %. Branching pattern in S. nux was more complex than in S. nucella (Fig. 2). The haplotype of S. nux from Simandou, Guinea, significantly diverged from its conspecifics, building a lineage of its own differing 3.4–4.5 %. Two lineages could be further delimited in S. nux, comprising haplotypes from Cameroon and Ghana, and Kenya and Cameroon, respectively. Variation within these two lineages reached 0.9 % and 1.2 %, respectively, and they differed by 2.0–3.0 %. The haplotype of S. nux from Ghana, KHFR-B7, did not cluster with any of these two lineages, and differed from them by 2.6–3.0 % and 2.0–2.4 %, respectively.

Table 2.

Selected external and cranial measurements of the Scotophilus nucella specimen NMP 95434 captured in 2015 in KHFR, southwestern Ghana, and comparative published data from Decher & Fahr (2007) and Robbins (1983) for the Ghana range values. *holotype, USNM — National Museum of Natural History, Washington D.C., U.S.A.

Partial zfy sequences obtained from four male samples of S. nucella were identical. Identical to each other were also newly obtained sequences of syntopic S. nux, which corresponded to the S. nux haplotype from Cameroon and Kenya available in the Genbank database under accession number EU751017. The zfy fragment analyzed was 454 bp long with inclusion of gaps. Its overall variation was low and genetic distance between species reached 0.9 % as computed from substitutions only. Scotophilus nucella and S. nux were represented by one haplotype each (Table 1). Scotophilus nucella shared one substitution with S. nux, through which both species differed from the other Scotophilus. Moreover, it showed a 3-bp stretch identical to that in S. nux, which they both share with S. kuhlii and a yet undescribed small-sized West African species of Scotophilus (S. aff. nigritellus) and which is missing in other Scotophilus. From S. nux, S. nucella differed by three unique substitutions. MP reconstruction based on zfy sequences supported S. nucella and S. nux as sister taxa as revealed by cytb analysis (Fig. 2). In MP analysis with gaps coded as missing data, S. nucella and S. nux clustered together in all of the 13 most parsimonious trees recovered and their relationship was supported by a bootstrap value of 75 %. With the consideration of gaps coded as the fifth character, the sister relationship among the two former species was more strongly supported by a bootstrap value of 80 %. Relationships among other species were not well resolved in both analyses.

Estimates of divergence times (Fig. 3) suggest that the splitting event in Scotophilus leading to the modern group of African rainforest species S. nucella and S. nux occurred around 5.5 (3.6–8.0, 95 % C.I.) Mya. A subsequent split giving rise to these two species took place around 2.5 (1.5–3.8, 95 % C.I.) Mya. In S. nux, a rather deep split occurred around 1.2 (0.6–1.8, 95 % C.I.) Mya leading to the separation of the western Upper Guinean lineage. A subsequent split 0.6 (0.3–1.0, 95 % C.I.) Mya lead to origin of S. nux in eastern Upper Guinean, and Lower Guinean and Congolian rainforest regions, which was followed by radiation around 0.26 (0.12–0.4, 95 % C.I.) and 0.23 (0.1–0.4, 95 % C.I.) Mya within the East and West African lineages, respectively. In S. nucella, the rather shallow structure points to a recent radiation 0.13 (0.04–0.27, 95 % C.I.) Mya.

Discussion

Discordance between morphological similarity and evolutionary relationships have for a long time obscured knowledge on the bat genus Scotophilus, and particularly the African species (Robbins et al. 1985, Goodman et al. 2005, Trujillo et al. 2009). The previously revealed parallelism within the two African morphospecies, S. dinganii and S. viridis, implied that sister relationship between the rainforest species S. nucella and S. nux was not a straightforward assumption. Moreover, the two latter species, especially S. nucella, belong to the least known and rarely encountered species of the genus compared to the former common and broadly distributed savannah species. It can be considered fortunate that twelve years after the last record in Ghana (Decher & Fahr 2007), this species was confirmed to still occur in the rapidly shrinking remnants of the West African rainforest. Taxonomic affinity of the specimens of S. nucella from 2014 and 2015 was confirmed through close similarity of their cytb sequences to the sequence of the previously identified specimen of S. nucella from the same locality captured in 2003. The forearm and skull measurements of the newly collected specimen NMP 95434 also corresponded to data from the type series of S. nucella by Robbins (1983) (Table 2).

The analysis of DNA sequences clearly confirmed that S. nucella deserves a position as a separate species, distinct from S. nux, where it was once included until size variation within West African populations prompted Robbins (1983) to separate the “nutlet” from the “nut”. The two reciprocally monophyletic mitochondrial lineages occurring in sympatry, with a genetic divergence exceeding 7.8 %, are supportive of the recognition of these two size forms as separate species and of their sister relationship. This value corresponds to differences found between other Scotophilus species (Trujillo et al. 2009, Vallo et al. 2013, 2015), as well as other bat taxa (Baker & Bradley 2006). Besides the pattern in maternally inherited mtDNA, S. nucella and S. nux also share paternally inherited zfy traits unique in this pair of forest species, while attaining their own distinct characteristics. The alternative assumption to their sister relationship, namely that of possible parallel evolution of two high forest species within Scotophilus can thus be abandoned.

Inferred topologies from both markers corroborate the fact that the forest species of Scotophilus have their origin in the early evolutionary history of the genus (Horaček et al. 2006, Trujillo et al. 2009). Molecular dating using mtDNA cytb sequences placed the split of the ancestor of the contemporary African taxa approximately at the transition between Miocene and Pliocene, which was characterized by gradual cooling and aridisation resulting in the retreat of forest and the spread of savannah (deMenocal 2004, Plana 2004). Very plausibly, a part of the ancestral population remained in forest habitat while the other part, ancestral to the remaining contemporary Scotophilus species, adapted to more arid conditions of savannah. Similar phylogenetic pattern with origin of rainforest species at this time period and later diversification of savannah species was revealed e.g. in genets of the genus Genetta Cuvier, 1816 (Gaubert et al. 2004). Within the forest lineage of Scotophilus, another split likely occurred at the transition between Pliocene and Pleistocene, which again brought severe aridisation events to Africa resulting in significant shrinking of forest areas (deMenocal 2004, Plana 2004). Most probably, this split gave origin to the contemporary sister species S. nucella and S. nux through allopatric speciation in forest refugia, as has been similarly hypothesized for African forest fruitbat species of the genera Myonycteris Matschie, 1899 (Nesi et al. 2013), Casinycteris Thomas, 1910 and Scotonycteris Matschie, 1894 (Hassanin et al. 2015). Creating possible phylogeographic scenarios in S. nucella would be rather speculative due to limited sampling of just one population. On the other hand, the basal split within S. nux, which delimited the Guinean haplotype, and the subsequent split leading to the West and East African lineages roughly correspond to the two other large Pleistocene aridisation waves around 1.7 Mya and 1 Mya (deMenocal 2004). The resulting pattern after the latter split is rather confusing, given that one of the Central African haplotypes from Cameroon clusters with Kenyan haplotypes and the others together with Ghanaian haplotypes, while there remains yet one more solitary divergent Ghanaian haplotype. A plausible hypothesis explaining this pattern may be an allopatric development of these lineages within known forest refugia in West, Central and East Africa (Maley 2001, Plana 2004) as summed up by Hassanin et al. (2015), followed by recent distribution range overlap. However, the determination of the possible geographic origin of S. nux and the direction of its distribution range expansions would be rather speculative given the current geographic sampling.

Despite limited sampling, however, the inferred sequence data indicate an allopatric speciation of the two sister taxa S. nucella and S. nux, now living in sympatry in West Africa with no evidence for hybridisation. Both species represent a remnant of the rainforest lineage of Scotophilus, a genus nowadays dominated by savannah forms in sub-Saharan Africa. The East African population of presumed S. nucella should be examined regarding its taxonomic affinity and evolutionary history. Given the current lack of morphometric data on the three known Tanzanian individuals (Cunneyworth 1996, Doggart et al. 1999a, b) and the suspect identification of the Ugandan specimen (Thorn & Kerbis Peterhans 2009), it may be quite probable that these individuals will be eventually revised as S. nux. The population of S. nux in Guinea may be considered a separate subspecies given its exclusive phylogenetic relationship to other African populations. A morphological comparison, as well as more robust genetic sampling, are definitely needed here in order to substantiate such a taxonomic amendment.