Cardioglossa manengouba, Blackburn, 2008

Blackburn, David C., 2008, A new species of Cardioglossa (Amphibia: Anura: Arthroleptidae) endemic to Mount Manengouba in the Republic of Cameroon, with an analysis of morphological diversity in the genus, Zoological Journal of the Linnean Society 154 (3), pp. 611-630 : 622-625

RELATIONSHIPS OF CARDIOGLOSSA MANENGOUBA

Both PCA and DFA reveal that Cardioglossa manengouba is morphologically similar to C. oreas , a high-altitude species restricted to Cameroon. These species share several morphological similarities to the exclusion of most other Cameroonian Cardioglossa . Both species exhibit small tympana, which are indistinct in C. oreas , a pronounced swelling in the skin immediately posterior to the tympanum that is often accompanied by a fold in the skin immediately above and behind the tympanum, and males of both species lack a hypertrophied third manual digit ( Fig. 3 ). In addition, both species lack the small inguinal spines that are present in mature males of all other Cardioglossa except C. escalerae (my unpubl. data). Currently, the call of both species remains unknown ( C. oreas: Amiet, 1973 ). Unique among Cardioglossa , these two species exhibit a general reduction or loss of typical male secondary sexual characters, which may reflect a close phylogenetic relationship. Amiet (1981) recognized C. oreas as distinctive among Cardioglossa because he observed that males lack both a hypertrophied third manual digit and spines on the manual digits. However, the two male C. oreas examined in this study (MCZ A-137922, A-137928) exhibit spines on the medial surface of the third digit and on both the lateral and the medial surfaces of the second digit ( Fig. 3B ). Similar to Amiet’s (1981) observation of C. oreas, Herrmann et al. (2004) suggested that a lack of spines on the hypertrophied manual digits of male C. alsco is a unique characteristic. Yet given the new data on the presence of manual digital spines in male C. oreas , I urge caution in interpreting the presence or absence of this character. The spines can be difficult to see without the aid of a microscope or hand lens, and this character may exhibit seasonal polyphenism in other arthroleptid taxa (e.g. Schmidt & Inger, 1959). Cardioglossa manengouba and C. oreas constitute two of the three species with the highest elevational ranges in Cameroon (the third, C. alsco from Tchabal Mbabo, was not included in this study). Whereas C. manengouba is known only from Mt Manengouba, C. oreas is documented from the Bamboutos Mountains ( Amiet, 1972a; present study), Mt Oku ( Amiet, 1987; present study) and Mt Lefo ( Böhme, 1975). In addition, Amiet (1987) states that C. oreas is present on Mt Manengouba, but no details of this record are provided; I did not find this species during fieldwork on Mt Manengouba in either 2004 or 2006. Based on morphological data, it seems likely that C. manengouba and C. oreas are sister taxa. If so, then vicariance between the forests of Mt Manengouba and other mountains to the north-east may have resulted in the allopatric speciation of these taxa. If C. oreas is present on Mt Manengouba, it may have later colonized this mountain at a time when the montane forests were continuous, or nearly so, between Mt Manengouba and the mountains to the north-east.

TADPOLE MORPHOLOGY AND ECOLOGY

The only previously described Cardioglossa tadpole is that of C. occidentalis ( Lamotte, 1961; Blackburn et al., in press). Both Amiet (1972b) and Perret (1966) provide illustration of a C. gracilis tadpole but these are not accompanied by a description; however, as figured, the tadpole morphology generally agrees with that discussed here. The tadpole of C. manengouba is similar to C. occidentalis in lacking keratinous denticles on the labia and in having a large and pronounced beak, a large suboral flap with numerous digitform projections, and an elongate, unpigmented, funnel-shaped spiracle. However, this elongate spiracle appears to be relatively longer in C. manengouba . Lamotte (1961) noted that the strong ventral pigmentation of C. occidentalis separates it from other tadpoles including those of Phrynobatrachus . In contrast, C. manengouba exhibits almost no ventral pigmentation.

Tadpoles of Cardioglossa are typically collected in leaf litter or deposits of sediment in shallow streams ( Amiet, 1972b; Rödel, Schorr & Ernst, 2001; my unpubl. data; present study). Interestingly, the morphology of C. manengouba tadpoles bears a striking resemblance to that of the South American microhylid Country abbreviations: CAR, Central African Republic; DRC, Democratic Republic of Congo (Kinshasa); RC, Republic of Congo (Brazzaville). Sources for altitudinal ranges (in metres above sea level): A72, Amiet (1972a); A81, Amiet (1981); B06, Blackburn (2006); B, Blackburn et al. (in press); Bö, Böhme & Schneider (1987); H04, Herrmann et al. (2004); H05, Herrmann et al. (2005b); I, IUCN et al. (2006); L50, Laurent (1950); L72, Laurent (1972); R, Rödel et al. (2001); S, estimated based on Schiøtz (1964); U, D. C. Blackburn, unpubl. data; *present study.

Otophryne robusta , which has unusual, fossorial tadpoles ( Wassersug & Pyburn, 1987). Cardioglossa manengouba tadpoles were collected syntopically with tadpoles of Astylosternus , Leptodactylodon and Leptopelis , all of which were collected by dragging a dip-net through leaf litter and sandy soil. However, only Cardioglossa tadpoles exhibit strong similarities to that of Otophryne . These similarities include a dorsoventrally flattened body, small eyes, a ventrally positioned mouth, a long tail and a greatly elongated spiracle. Cardioglossa is the only anuran genus with a spiracular tube that approaches the length found in Otophryne , although it is unclear whether it would function similarly to that of Otophryne . As the length of the spiracular tube relative to body length changes during O. robusta ontogeny ( Wassersug & Pyburn, 1987), it is possible that the spiracular tube of C. manengouba could be even longer relative to body length at other points in its ontogeny. The elongate branchial basket of C. manengouba is unusual and may also be related to a fossorial ecology, but its functional implications are unclear. Further examination of Cardioglossa tadpole morphology, including chondrocranial anatomy, will elucidate the amount of covergence between Cardioglossa , Otophryne and other anuran species with fossorial tadpoles.

IMPLICATIONS OF MORPHOMETRIC ANALYSES FOR CARDIOGLOSSA SYSTEMATICS

The analysis of morphometric data presented here supports a close phylogenetic relationship between C. manengouba and C. oreas . In addition, this study allows for the recognition of other Cardioglossa species groups, which are discussed below ( Table 4).

This study demonstrates that C. aureoli , from the Freetown Peninsula of Sierra Leone, is morphologically distinct from other Cardioglossa species. Based presumably only on coloration, this species was originally placed in Cardioglossa by Schiøtz (1964). Unlike other Cardioglossa , this species is typically found far from water ( Schiøtz, 1964), thus possibly exhibiting some form of development in which there is no freeliving and/or feeding tadpole (e.g. IUCN et al., 2006). DNA sequence data from the mitochondrial 16S ribosomal RNA gene indicates that this species is very divergent (> 15%) from other Cardioglossa (Blackburn et al., in press), and molecular phylogenetic analysis shows that C. aureoli probably is more closely related to other arthroleptid species than to Cardioglossa (my unpubl. data). Cardioglossa aureoli is well separated from other Cardioglossa along PC1 ( Fig. 6 ), which is largely an indication that both males and females of this species are much smaller than other Cardioglossa species. Molecular phylogenetic research and study of skeletal anatomy will probably reveal that C. aureoli should either be included in another arthroleptid genus or placed in its own genus.

Amiet (1987) suggested a close relationship between C. gratiosa and C. nigromaculata , both lowland species, and intimated the existence of a hybrid zone between these species in southern coastal Cameroon. This conclusion was based on the mixture of vocalization characteristics found in these populations ( Amiet, 1987), but there is no reason why this mixture necessitates the hybridization of these species. There are no striking morphological or colour pattern similarities between C. gratiosa and C. nigromaculata and, although the sample sizes are extremely small, no obvious pattern emerged from PCA to support the similarity of these species to the exclusion of other species. In contrast, PCA reveals that C. gratiosa is morphologically similar to C. escalerae , another lowland species that is more widespread in Central Africa. Interestingly, in the DFA, C. nigromaculata was assigned to the classification group comprising C. escalerae and C. gratiosa . Amiet (1981) proposed a close phylogenetic relationship between these three species. The results of these morphometric analyses support this hypothesis and suggest that this species grouping should continue to be recognized.

Cardioglossa melanogaster and C. schioetzi are similar morphologically, ecologically and acoustically, and both are distributed throughout the mountains of Cameroon and Nigeria ( Amiet, 1972a, 1981; Blackburn, 2006; Blackburn et al. in press). Amiet (1981) commented extensively on the similarities between C. melanogaster and C. schioetzi and hypothesized that these species may be closely related to C. leucomystax and C. gracilis ( Amiet, 1975, 1981). The calls of C. melanogaster and C. schioetzi are very similar and differentiated mostly by the former species having longer notes with greater spacing between the notes ( Amiet, 1981). In addition, the first two presacral vertebrae of C. melanogaster and C. schioetzi are fused (Blackburn et al., in press), which is the same conditition found in C. gracilis but absent in nearly all C. leucomystax examined (exceptions: USNM 563696, 563697 and 563705, all from low altitudes on Mt Nlonako in Cameroon). The plots of principal component scores reveal that both C. melanogaster and C. schioetzi are morphologically similar to C. gracilis but are easily distinguished from C. leucomystax ( Fig. 6 ). In all three of the former species the distance from the anterior margin of the eye to the tip of the rostrum is subequal to or greater than the anteroposterior diameter of the eye. The results of the DFA further emphasize the dissimilarity between C. leucomystax and the three other species because specimens of C. leucomystax were classified as belonging to every group except the one comprising C. gracilis , C. melanogaster and C. schioetzi . Based on the available evidence, I propose that C. melanogaster and C. schioetzi are sister taxa and more closely related to C. gracilis than to C. leucomystax .

The recently described C. occidentalis is very similar in both appearance and biology to C. leucomystax with which it was long recognized as conspecific (Blackburn et al., in press). If C. occidentalis and C. leucomystax are sister species, then they exhibit a biogeographical pattern similar to sister pairs of other vertebrates found in the Upper and Lower Guinean Forest Zones (Blackburn et al., in press). The results of the morphometric analyses presented here are equivocal with regard to the relationship of these two species; there is no strong evidence for either morphometric similarity or dissimilarity. In the DFA, no specimens of C. occidentalis or C. leucomystax were assigned to the classification group consisting of C. gracilis , C. melanogaster and C. schioetzi . Similarly, the three classification groups to which C. occidentalis specimens were assigned (groups 2, 4 and 5) are the same three groups to which nearly all C. leucomystax specimens were assigned. The PCA and DFA reveal a substantial amount of morphometric diversity within these two species, but the analyses also suggest that they exhibit a similar range of morphometric diversity. Phylogenetic analyses will be required to elucidate the relationship between these two species as well as their relationship to other Cardioglossa .

Based in part on both ecology and coloration, Amiet (1972a, 1975, 1981) argued that three montane taxa, C. pulchra , C. trifasciata and C. venusta , have close affinities to one another. All three species exhibit a bluish ventral coloration and dorsal skin that is covered in relatively larger tubercles than other Cardioglossa species. The scores for the first three PC axes for these three species were not significantly different from those of other species, but the species do occupy a similar region of morphospace ( Fig. 6 ). Interestingly, DFA correctly classified all specimens assigned to this classification group. This lends further support to the similarities recognized by Amiet (1972a). All three species exhibit small ranges in the mountains of the Cameroon Volcanic Line and C. trifasciata is found only on the verdant south-west slopes of Mt Manengouba. While I was unable to examine specimens of C. alsco for this study, I have followed Herrmann et al. (2004) by including this recently described species in this species group based on a blue ventral coloration, a granular dorsal skin, and a high similarity of both coloration and pattern to C. pulchra .

Most previous studies of Cardioglossa diversity have not included C. cyaneospila , a species endemic to the mountains of Rwanda and Burundi (e.g. Amiet, 1972a, 1981; Herrmann et al., 2004). The two specimens of C. cyaneospila included in this study were among the largest of all Cardioglossa males examined. Similar to C. oreas , C. pulchra , C. trifasciata and C. venusta , C. cyaneospila lacks an infratympanal line and exhibits small, scattered tubercles and like C. elegans , C. nigromaculata and C. trifasciata , the dorsal markings in C. cyaneospila are large and block-like. Based on the PCA, this species is clearly differentiated from C. gracilis , C. melanogaster and C. schioetzi but not other Cardioglossa species. Cardioglossa cyaneospila , C. pulchra , C. trifasciata and C. venusta do not form a group that can be easily differentiated from other Cardioglossa because of extreme values of the first three PC axes, but all three occupy a similar region of morphospace. In the DFA, both specimens of C. cyaneospila were assigned to the group consisting of C. pulchra , C. trifasciata and C. venusta . In light of the other similarities mentioned above, it seems reasonable to consider C. cyaneospila to be part of this species group. As all of these species are montane, this species group can be taken as further support for the historical continuity of montane faunas of different regions in Africa.

Because C. cyaneospila is known only from type specimens and the type locality is located in a conflict zone [Bururi Province, Burundi; erroneously listed by Frost (2006) as the Democratic Republic of Congo], it is unlikely that specimens for molecular phylogenetic research will be available in the near future. Another taxon from a nearby conflict zone, C. nigromaculata inornata from Fizi in the South Kivu Province in eastern Democratic Republic of Congo, probably represents a distinct species ( Frost, 2006) but specimens were unavailable for study. Our poor understanding of the relationships and taxonomy of these two Central African montane taxa emphasizes the necessary place of morphometric and morphological analyses in studies of diversity in Cardioglossa and other Central African taxa that occur in conflict zones (for a similar example, see Ohler, 1996).

Figures

Figure 3. Right lateral view of the head of the holotypes of Cardioglossa manengouba sp. nov. (A; male; MCZ A-137909) and C. oreas (B; female; MHNG 1253.87). Note the small tympanum and the swelling and fold immediately above and behind the tympanum. Ventral view of the right hand of a male paratype of C. manengouba (C; MCZ A-136933) and male C. oreas (D; MCZ A-137922). Scale bar = 1 mm.

Figure 6. Scatter plot of the first and second principal component scores. The second principal component (PC2) is plotted against the first principal component (PC1), an indicator of body size. Lines are drawn to indicate the centroid of points for each species.