We herein examine the taxonomic status of two West African forest-dwelling Leptopelis species. The small L. spiritusnoctis, described from the Upper Guinean forests of West Africa, was recently synonymized with L. aubryi, described from Gabon. The large L. macrotis, known from Ghana to Sierra Leone, was downgraded to a subspecies of L. millsoni, ranging from the Niger Delta to eastern Democratic Republic of Congo. These taxonomic decisions are in contrast to the general biogeographic pattern of African forest anurans and we consequently tested if the morphologically similar taxon pairs are indeed conspecifics by applying acoustic and molecular techniques. Both techniques confirmed that populations from West Africa differ significantly from their Central African morphological equivalents. Consequently, we herein resurrect L. spiritusnoctis as a valid species. The acoustic data indicate that L. aubryi may comprise a complex of cryptic species. We further advocate using the name L. macrotis for West African and L. millsoni for Central African populations of these larger arboreal frogs. However, we had neither genetic nor acoustic data from the type locality of L. millsoni available and could not clarify if these frogs belong to the more western or eastern taxon or even represent a Nigerian endemic. Thus, it is possible that West African populations need to be termed L. millsoni in the future. For populations east of the Cross River, Nigeria, the name L. guineensis would be available.

Advertisement calls, DNA-barcoding, biogeography, Central Africa, rainforest, taxonomy, treefrogs, West Africa

In many older and recent herpetological papers the term West Africa is used to describe a region, ranging approximately from Senegal in the northwest, along the western coast of the continent to Gabon, or even further south to northern Angola. However, recent investigations showed that the Upper and parts of the Lower Guinean forests, as defined based on plant distributions (see Udvardy 1975, White 1983), comprise very unique amphibian faunas (Penner et al. 2011). The two zoogeographic regions, West and Central Africa, are roughly separated by the Cross River in easternmost Nigeria (Schiøtz 1967, Penner et al. 2011). However, it is presumably not this river, but a tongue of the Atlantic Ocean that reached far inland from ca. 50 to 100 MA (Scotese 2001) and has acted as a zoogeographic barrier. The amphibian faunas east and west of this salt water barrier could thus independently evolve for at least that time span. Nowadays, this is visible in ecologically and most often also morphologically very similar but distinct forest species in almost all groups of anuran amphibians: e.g. Bufonidae (e.g. Perret and Amiet 1971, Rödel and Ernst 2000, Barej et al. 2011), Arthroleptidae (e.g. Blackburn et al. 2008, 2009, 2010, Ernst et al. 2008), Phrynobatrachidae (e.g. Lamotte and Xavier 1966, Zimkus et al. 2010), Hyperoliidae (e.g. Schiøtz 1967, 2007, Seymour et al. 2001, Rödel et al. 2003) and Ranidae (e.g., Perret 1977, Rödel and Bangoura 2004).

Biogeographic subdivisions are even found within West Africa (Schiøtz 1967, Penner et al. 2011). Only very few forest anuran species are still believed to be distributed in the Upper and Lower Guinea forest bioregion (Rödel and Ernst 2001, Rödel et al. 2002, Schiøtz 2007). However, in most cases there is already unpublished evidence that they actually comprise complexes of cryptic species, supporting the above sketched biogeographic borders (Rödel et al. unpubl. data). This biogeographic separation of ranges also applies for all forest-dwelling species of the predominantly arboreal arthroleptid genus Leptopelis (compare e.g. Schiøtz 1999, 2007 although there might be an overlap in southern Nigeria, see discussion). Thus, it was very surprising that in his seminal book on the treefrogs of Cameroon, Amiet (2012) synonymized a West African forest Leptopelis, L. spiritusnoctis Rödel, 2007, type locality Taï National Park, Ivory Coast (Rödel 2007), with a species known from central African forests, L. aubryi (Duméril, 1856), type locality Gabon (Duméril 1856, Guibé 1950). Amiet (2012) further downgraded another West African forest-dwelling species, L. macrotis Schiøtz, 1967, only known from the western part of the Upper Guinean forests (Schiøtz 1999, type locality: Gola Rainforest National Park, Sierra Leone), to a subspecies of the mostly central African L. millsoni (Boulenger, 1895), type locality being the Niger Delta, Nigeria (Boulenger 1895). These taxonomic conclusions were based on morphological similarities and the (visual) comparisons of published call descriptions (Amiet 2012). If the mentioned taxa are indeed conspecific, they would constitute very remarkable exceptions from the above outlined general biogeographic pattern in African forest anurans.

As detailed morphological descriptions of these species show, they are morphologically very similar to each other (Perret 1962, Schiøtz 1967, 1999, Rödel 2007; Amiet 2012, Figs 1 & 2). However, frogs of this genus are often only offering few and delicate morphological differences (Schiøtz 1967, 1999, Amiet 2012, Frétey et al. 2012). We thus tested Amiet’s (2012) conclusions, using molecular techniques and detailed comparisons of call characteristics. Both methods are well established to clarify the taxonomic status of anurans when morphology alone is not very informative (Schiøtz 1973, Vieites et al. 2009).

Figure 1. 

Life coloration and morphology in Leptopelis aubryi (a-c) and L. spiritusnoctis (d-h); a) L. aubryi, ZFMK 89562, call analyzed herein, and b) ZFMK 89560, call analyzed herein, both frogs from Big Massaka, Cameroon, large pictures color during night, inlet pictures color during daytime; c) L. aubryi (ZFMK 87728) from Nkoelon, Cameroon; d) L. spiritusnoctis (ZMB 79578) from Banambala, Guinea; e) L. spiritusnoctis (ZMB 79580) from Mt. Jideh, Liberia; f-h) L. spiritusnoctis (ZMB 79581-583) from Sapo National Park, Liberia.

Figure 2. 

Life coloration and morphology in Leptopelis macrotis (a-c) and L. millsoni (d); L. macrotis from a) Yomou, Guinea (ZMB 79585); b) Mt. Jideh, Liberia (ZMB 79587); c) Gola Rainforest National Park, Sierra Leone (type locality, not collected); d) L. millsoni from Nkoelon, Cameroon (ZFMK 87708).

Molecular data. We analyzed 454 bp of the mitochondrial 16S ribosomal RNA gene for the L. macrotis / L. millsoni comparison and 522 bp for the L. spiritusnoctis / L. aubryi comparison. The uncorrected pairwise sequence divergence between L. aubryi (N = 6) and L. spiritusnoctis (N = 6) ranged from 5.74–6.56% (6.23 ± 0.22%; N = 36). The intraspecific variation in L. aubryi ranged from 0–2.12% (0.86 ± 0.94%; N = 15). The intraspecific variation in L. spiritusnoctis ranged from 0–0.62% (0.27 ± 0.23%; N = 15). The uncorrected pairwise sequence divergence between L. macrotis (N = 4) and L. millsoni (N = 1) ranged from 11.07–11.30% (mean ± sd: 11. 24 ± 0.12%; N = 4). The intraspecific variation in L. macrotis ranged from 0–0.22% (0.11 ± 0.12%; N = 6).

Acoustic data. For L. aubryi and L. spiritusnoctis we could analyze the “chuck” (Fig. 3) and “buzz” (Fig. 4) calls (compare Amiet and Schiøtz 1974 and Grafe et al. 2000 for intraspecific call variability in Leptopelis). Both call types showed distinct differences in a variety of different call characteristics between populations of different origin (Table 2). At first sight the “chuck” call of L. aubryi (N = 22) and L. spiritusnoctis (N = 13) looked similar (Schiøtz 1967). Both calls are short single notes. However, the call of L. spiritusnoctis (Ivory Coast) was almost twice as long (18.8 ± 4.2 ms) compared to the chuck of L. aubryi (10.0 ± 2.0 ms) from Cameroon (Fig. 3). Likewise the dominant frequency differed between these populations (Table 2). The “buzz” call of these two species also showed differences. The “buzz” of L. spiritusnoctis (N = 22) was longer (334.5 ± 122.5 ms) than in L. aubryi (duration 51–105 ms; N =3) and comprised more pulses (Fig. 4; Table 2). However, the “chuck” calls of “L. aubryi” from different localities were extremely variable (Table 2). In contrast to our L. aubryi records from Cameroon “chucks” from eastern Nigeria and Equatorial Guinea (recordings A. Schiøtz and Bosch et al. 2000, respectively) were three to four times longer than in Cameroon (Table 2) and also differed in the dominant frequencies (Table 2). Some parameters of a particular L. aubryi population had similar values compared to L. spiritusnoctis calls. However, each “aubryi” population showed distinct differences to the L. spiritusnoctis calls, which exhibit only very little variation (Table 2; MOR, JP, MFB own experiences of calls heard throughout West Africa).

Figure 3. 

Oscillograms, audio spectrograms and power spectrograms of the “chuck” call of Leptopelis aubryi (ZFMK 89560) from Big Massaka, Cameroon (a) and Leptopelis spiritusnoctis from Taï National Park, Ivory Coast (b). This audio files are also available from the Tierstimmenarchiv at the Museum für Naturkunde, Berlin, through the following links: a) http://www.tierstimmenarchiv.de/recordings/Leptopelis_aubryi_DIG0174_01_short.mp3 (Suppl. material 2), b) http://www.tierstimmenarchiv.de/recordings/Leptopelis_spiritusnoctis_DIG0174_02_short.mp3 (Suppl. material 5).

Figure 4. 

Oscillograms, audio spectrograms and power spectrograms of the buzz calls of Leptopelis aubryi (ZFMK 89560) from Big Massaka, Cameroon (a) and of Leptopelis spiritusnoctis (two calls) from Taï National Park, Ivory Coast (b). This audio files are also available from the Tierstimmenarchiv at the Museum für Naturkunde, Berlin, through the following links: a) http://www.tierstimmenarchiv.de/recordings/Leptopelis_aubryi_DIG0174_03_short.mp3 (Suppl. material 1), b) http://www.tierstimmenarchiv.de/recordings/Leptopelis_spiritusnoctis_DIG0174_04_short.mp3 (Suppl. material 4).

From L. macrotis we only know of one call type (Fig. 5). Our L. macrotis calls (N = 3; recorded in Cavally Forest Reserve, Ivory Coast) had the following call characteristics (always: mean ± sd): call duration 388.0 ± 19.0 ms; note duration 92.0 ± 10.0 ms; duration of interval between notes 204.0 ± 15.0 ms; minimum frequency 1300.3 ± 18.5 Hz; maximum frequency 2432.0 ± 19.1 Hz; frequency range 1132.0 ± 19.1 Hz; dominant frequency and fundamental frequency 1729.8 ± 22.2 Hz; the dominant frequency starts at 1715.5 ± 22.2 Hz and ends with 1744.2 ± 27.2 Hz.

Figure 5. 

Oscillograms, audiospectrograms and power spectrograms of the advertisement call (“mmau”) of Leptopelis macrotis from Cavally Forest Reserve, Ivory Coast. This audio file is also available from the Tierstimmenarchiv at the Museum für Naturkunde, Berlin, through the following link http://www.tierstimmenarchiv.de/recordings/Leptopelis_macrotis_DIG0174_05_short.mp3 (Suppl. material 3).

From L. millsoni we had no own recording available. The call of L. millsoni is described as a nasal “himp”, consisting of a single tonal long note (approximately 170 ms) with a high frequency range (Amiet and Schiøtz 1974). L. macrotis calls, sounding like the meowing of a suffering cat to the human ear (“mmau”), consists of two (instead of one in L. millsoni) short, tonal notes with a smaller frequency range. In comparison to the spectra of L. millsoni the dominant frequency in L. macrotis (approximately 1500 Hz) seems to be similar, but L. millsoni may have a much lower fundamental frequency (approximately 250 Hz).

The overall morphological and presumed acoustic similarity between Leptopelis aubryi with L. spiritusnoctis and that of L. millsoni with L. macrotis (Figs 1 & 2) led Amiet (2012) to consider the West African taxa to be a synonym or subspecies, respectively, of the similar Central African taxon. In fact, this judgment is not new and in particular L. spiritusnoctis (termed L. hyloides prior to Rödel 2007, see Rödel 2007 for discussion of the taxonomic history of this taxon) was treated as a synonym of L. aubryi (e.g. Lamotte and Perret 1961, Schiøtz 1963, Perret 1966). Likewise, West African populations (e.g. from south-eastern Guinea) of the large forest Leptopelis have been assigned to L. millsoni, until Schiøtz (1967) described them as a distinct species. Thus, we can see the reasoning for the doubts on the validity of these two species. In contrast, as already outlined in the introduction, the distribution of the forest anurans, L. millsoni and L. aubryi, far into West Africa would be a remarkable exception from the almost general biogeographic pattern of western African anuran species (compare Schiøtz 1967, 2007, Penner et al. 2011).

Nevertheless, the Cross River in eastern Nigeria is not an all-or-nothing barrier, and a few Central African forest anurans indeed managed to occur further West reaching the Niger Delta or even beyond (e.g. Hymenochirus boettgeri, Amietophrynus camerunensis, Nectophryne afra, Conraua crassipes, Acanthixalus spinosus, Afrixalus paradorsalis; Akani et al. 2004, Onadeko and Rödel 2009). Also a few species of the Upper Guinean forests managed to cross the Dahomey Gap (a natural forest gap at the Gulf of Guinea coast roughly aligned with the countries Togo and Bénin; see Schiøtz 1967, Penner et al. 2011) to the east and reach the Niger River (e.g. Leptopelis cf. occidentalis, Phrynobatrachus cf. plicatus; Onadeko and Rödel 2009) or even Cameroon (e.g. Hyperolius concolor, H. guttulatus, H. fusciventris, Phlyctimantis boulengeri; Schiøtz 1967, 1999). The latter are farmbush species (species tolerating forest degradation and commonly are most abundant in the forest-savanna ecotone; Schiøtz 1967, 1999, Rödel 2000, Rödel and Ernst 2001) and thus not strictly bound to forest vegetation and respective barriers (compare Schiøtz 2007), although different subspecies of the Hyperolius species are accepted to occur in West and Central Africa (e.g. Schiøtz 1967, 2007). The four Leptopelis species treated in this paper are strictly forest-dwelling.

Our molecular and acoustic data presented herein, clearly support the view that West and Central Africa comprise distinct anuran faunas, including the Leptopelis populations in question. The genetic distances between L. aubryi and L. spiritusnoctis in the studied 16S gene fragment (5.74–6.56%) and L. millsoni and L. macrotis (11.07–11.30%), respectively, by far exceed the values which are commonly assumed to indicate distinct species (3–5% in 16S; see Vieites et al. 2009 for general statement and e.g. Blackburn et al. 2008, Zimkus 2009, Rödel et al. 2012a, Zimkus and Gvodzík 2013 for values between distinct western African anuran species). Furthermore, our detailed call analyses revealed distinct differences between these taxa in almost all examined call parameters, in particular in parameters with low intraspecific variability (i.e. fundamental frequency, dominant frequency, call duration; Gerhardt 1991, Ryan and Rand 2001). That L. spiritusnoctis and L. aubryi have very different calls was already mentioned by Schiøtz (1967, 1999) and Amiet and Schiøtz (1974). In contrast Amiet and Schiøtz (1974) regarded the calls of L. millsoni and L. macrotis as almost identical. This could not be confirmed herein.

Taxonomic conclusions. Based on the molecular and acoustic data presented we conclude that synonymizing L. spiritusnoctis with L. aubryi was not justified and thus resurrect L. spiritusnoctis for the small forest-dwelling Leptopelis species in the Upper Guinean forest zone (see Rödel 2007 for the distribution). Although our genetic samples of L. aubryi from Gabon and Cameroon were relatively similar (maximum difference in the 16S sequence being 2.12%), the large variation of call characteristics in L. aubryi might be an indication that this name currently covers a complex of cryptic species, indicating the needs for further taxonomic studies. Various names currently regarded as synonyms of L. aubryi, are available (see Frost 2013).

The downgrading of L. macrotis as a subspecies of L. millsoni, as currently understood, was likewise erroneous. However, the situation is more complex. West and Central African populations clearly represent distinct species. Currently, we cannot decide to which species the name L. millsoni (type locality is the Niger delta, geographically a part of the western species assemblage) should be applied. Unfortunately, so far we do not have genetic samples from Nigeria. Thus, we do not know which species occurs in Nigeria and we cannot even exclude that southern Nigeria harbors taxa which are morphologically similar, but distinct on the basis of genetics and acoustics from those adjacent to the east and west (compare e.g. Rödel et al. 2012b). To resolve this problem new material, tissue samples and call recordings, from the L. millsoni type locality are needed. If L. millsoni would be the valid name for the West African species, the name L. guineensis Ahl, 1929 (type: ZMB 20047; type locality: Makomo, Equatorial Guinea) would be available for the Central African species (Ahl 1929, Perret 1962). With respect to taxonomic stability we suggest to continue using the name L. macrotis and L. millsoni as defined by Schiøtz (1967, 1999) until new material becomes available.

We thank the various authorities from Cameroon, Ivory Coast, Guinea and Liberia for collection permits, and A. Schiøtz for making his call recordings available, through the sound library of the Cornell University. W. Böhme and D. Rödder (ZFMK) permitted tissue sampling from ZFMK vouchers. We thank Julia Barej for the permission to use her photos of life frogs. Jörn Köhler and Frank Glaw provided helpful critique.